{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9999935863106213,
  "eval_steps": 61000,
  "global_step": 311832,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.413689378609705e-05,
      "grad_norm": 8.135562896728516,
      "learning_rate": 3.206772703950744e-09,
      "loss": 0.8416,
      "step": 10
    },
    {
      "epoch": 0.0001282737875721941,
      "grad_norm": 10.716862678527832,
      "learning_rate": 6.413545407901488e-09,
      "loss": 0.9044,
      "step": 20
    },
    {
      "epoch": 0.00019241068135829114,
      "grad_norm": 11.06715202331543,
      "learning_rate": 9.620318111852233e-09,
      "loss": 0.8359,
      "step": 30
    },
    {
      "epoch": 0.0002565475751443882,
      "grad_norm": 9.63078498840332,
      "learning_rate": 1.2827090815802977e-08,
      "loss": 0.8421,
      "step": 40
    },
    {
      "epoch": 0.0003206844689304852,
      "grad_norm": 8.548972129821777,
      "learning_rate": 1.6033863519753723e-08,
      "loss": 0.8503,
      "step": 50
    },
    {
      "epoch": 0.0003848213627165823,
      "grad_norm": 10.621478080749512,
      "learning_rate": 1.9240636223704465e-08,
      "loss": 0.8608,
      "step": 60
    },
    {
      "epoch": 0.0004489582565026793,
      "grad_norm": 10.499129295349121,
      "learning_rate": 2.2447408927655208e-08,
      "loss": 0.8388,
      "step": 70
    },
    {
      "epoch": 0.0005130951502887764,
      "grad_norm": 11.931020736694336,
      "learning_rate": 2.5654181631605953e-08,
      "loss": 0.8689,
      "step": 80
    },
    {
      "epoch": 0.0005772320440748734,
      "grad_norm": 11.505675315856934,
      "learning_rate": 2.88609543355567e-08,
      "loss": 0.835,
      "step": 90
    },
    {
      "epoch": 0.0006413689378609704,
      "grad_norm": 6.3639044761657715,
      "learning_rate": 3.2067727039507445e-08,
      "loss": 0.8218,
      "step": 100
    },
    {
      "epoch": 0.0007055058316470676,
      "grad_norm": 10.279727935791016,
      "learning_rate": 3.527449974345819e-08,
      "loss": 0.8279,
      "step": 110
    },
    {
      "epoch": 0.0007696427254331646,
      "grad_norm": 9.912506103515625,
      "learning_rate": 3.848127244740893e-08,
      "loss": 0.8666,
      "step": 120
    },
    {
      "epoch": 0.0008337796192192616,
      "grad_norm": 9.041036605834961,
      "learning_rate": 4.168804515135967e-08,
      "loss": 0.8098,
      "step": 130
    },
    {
      "epoch": 0.0008979165130053586,
      "grad_norm": 10.479462623596191,
      "learning_rate": 4.4894817855310415e-08,
      "loss": 0.8174,
      "step": 140
    },
    {
      "epoch": 0.0009620534067914557,
      "grad_norm": 11.549273490905762,
      "learning_rate": 4.8101590559261164e-08,
      "loss": 0.8904,
      "step": 150
    },
    {
      "epoch": 0.0010261903005775528,
      "grad_norm": 9.424370765686035,
      "learning_rate": 5.130836326321191e-08,
      "loss": 0.8845,
      "step": 160
    },
    {
      "epoch": 0.0010903271943636498,
      "grad_norm": 10.978484153747559,
      "learning_rate": 5.451513596716265e-08,
      "loss": 0.8073,
      "step": 170
    },
    {
      "epoch": 0.0011544640881497468,
      "grad_norm": 9.741132736206055,
      "learning_rate": 5.77219086711134e-08,
      "loss": 0.8366,
      "step": 180
    },
    {
      "epoch": 0.0012186009819358439,
      "grad_norm": 6.865855693817139,
      "learning_rate": 6.092868137506415e-08,
      "loss": 0.7748,
      "step": 190
    },
    {
      "epoch": 0.0012827378757219409,
      "grad_norm": 10.588968276977539,
      "learning_rate": 6.413545407901489e-08,
      "loss": 0.8293,
      "step": 200
    },
    {
      "epoch": 0.0013468747695080379,
      "grad_norm": 10.359200477600098,
      "learning_rate": 6.734222678296562e-08,
      "loss": 0.7546,
      "step": 210
    },
    {
      "epoch": 0.001411011663294135,
      "grad_norm": 10.029732704162598,
      "learning_rate": 7.054899948691638e-08,
      "loss": 0.8056,
      "step": 220
    },
    {
      "epoch": 0.0014751485570802321,
      "grad_norm": 8.809959411621094,
      "learning_rate": 7.375577219086712e-08,
      "loss": 0.7782,
      "step": 230
    },
    {
      "epoch": 0.0015392854508663291,
      "grad_norm": 7.988217830657959,
      "learning_rate": 7.696254489481786e-08,
      "loss": 0.7392,
      "step": 240
    },
    {
      "epoch": 0.0016034223446524261,
      "grad_norm": 7.067934513092041,
      "learning_rate": 8.01693175987686e-08,
      "loss": 0.7242,
      "step": 250
    },
    {
      "epoch": 0.0016675592384385231,
      "grad_norm": 7.659463882446289,
      "learning_rate": 8.337609030271935e-08,
      "loss": 0.7337,
      "step": 260
    },
    {
      "epoch": 0.0017316961322246202,
      "grad_norm": 9.248407363891602,
      "learning_rate": 8.658286300667009e-08,
      "loss": 0.7197,
      "step": 270
    },
    {
      "epoch": 0.0017958330260107172,
      "grad_norm": 9.64366626739502,
      "learning_rate": 8.978963571062083e-08,
      "loss": 0.7122,
      "step": 280
    },
    {
      "epoch": 0.0018599699197968144,
      "grad_norm": 7.405231475830078,
      "learning_rate": 9.299640841457157e-08,
      "loss": 0.6541,
      "step": 290
    },
    {
      "epoch": 0.0019241068135829114,
      "grad_norm": 7.820592403411865,
      "learning_rate": 9.620318111852233e-08,
      "loss": 0.6595,
      "step": 300
    },
    {
      "epoch": 0.001988243707369008,
      "grad_norm": 6.586963653564453,
      "learning_rate": 9.940995382247307e-08,
      "loss": 0.629,
      "step": 310
    },
    {
      "epoch": 0.0020523806011551056,
      "grad_norm": 5.655336380004883,
      "learning_rate": 1.0261672652642381e-07,
      "loss": 0.5769,
      "step": 320
    },
    {
      "epoch": 0.0021165174949412027,
      "grad_norm": 4.75510835647583,
      "learning_rate": 1.0582349923037456e-07,
      "loss": 0.5499,
      "step": 330
    },
    {
      "epoch": 0.0021806543887272997,
      "grad_norm": 5.1090288162231445,
      "learning_rate": 1.090302719343253e-07,
      "loss": 0.5097,
      "step": 340
    },
    {
      "epoch": 0.0022447912825133967,
      "grad_norm": 5.625272274017334,
      "learning_rate": 1.1223704463827605e-07,
      "loss": 0.4836,
      "step": 350
    },
    {
      "epoch": 0.0023089281762994937,
      "grad_norm": 5.049161434173584,
      "learning_rate": 1.154438173422268e-07,
      "loss": 0.4603,
      "step": 360
    },
    {
      "epoch": 0.0023730650700855907,
      "grad_norm": 5.61590051651001,
      "learning_rate": 1.1865059004617754e-07,
      "loss": 0.4457,
      "step": 370
    },
    {
      "epoch": 0.0024372019638716877,
      "grad_norm": 4.5563435554504395,
      "learning_rate": 1.218573627501283e-07,
      "loss": 0.4312,
      "step": 380
    },
    {
      "epoch": 0.0025013388576577847,
      "grad_norm": 4.078986167907715,
      "learning_rate": 1.2506413545407904e-07,
      "loss": 0.3961,
      "step": 390
    },
    {
      "epoch": 0.0025654757514438817,
      "grad_norm": 4.474985122680664,
      "learning_rate": 1.2827090815802978e-07,
      "loss": 0.3504,
      "step": 400
    },
    {
      "epoch": 0.0026296126452299787,
      "grad_norm": 3.7105531692504883,
      "learning_rate": 1.3147768086198052e-07,
      "loss": 0.3325,
      "step": 410
    },
    {
      "epoch": 0.0026937495390160758,
      "grad_norm": 3.7654149532318115,
      "learning_rate": 1.3468445356593124e-07,
      "loss": 0.2948,
      "step": 420
    },
    {
      "epoch": 0.0027578864328021728,
      "grad_norm": 3.5739786624908447,
      "learning_rate": 1.37891226269882e-07,
      "loss": 0.2721,
      "step": 430
    },
    {
      "epoch": 0.00282202332658827,
      "grad_norm": 2.772843599319458,
      "learning_rate": 1.4109799897383275e-07,
      "loss": 0.25,
      "step": 440
    },
    {
      "epoch": 0.0028861602203743672,
      "grad_norm": 2.9764015674591064,
      "learning_rate": 1.443047716777835e-07,
      "loss": 0.2425,
      "step": 450
    },
    {
      "epoch": 0.0029502971141604642,
      "grad_norm": 2.5800020694732666,
      "learning_rate": 1.4751154438173424e-07,
      "loss": 0.2328,
      "step": 460
    },
    {
      "epoch": 0.0030144340079465612,
      "grad_norm": 2.62642765045166,
      "learning_rate": 1.5071831708568498e-07,
      "loss": 0.2177,
      "step": 470
    },
    {
      "epoch": 0.0030785709017326583,
      "grad_norm": 3.1305058002471924,
      "learning_rate": 1.5392508978963572e-07,
      "loss": 0.2358,
      "step": 480
    },
    {
      "epoch": 0.0031427077955187553,
      "grad_norm": 2.686131000518799,
      "learning_rate": 1.5713186249358646e-07,
      "loss": 0.2118,
      "step": 490
    },
    {
      "epoch": 0.0032068446893048523,
      "grad_norm": 2.276582717895508,
      "learning_rate": 1.603386351975372e-07,
      "loss": 0.2095,
      "step": 500
    },
    {
      "epoch": 0.0032709815830909493,
      "grad_norm": 2.1569366455078125,
      "learning_rate": 1.6354540790148795e-07,
      "loss": 0.2041,
      "step": 510
    },
    {
      "epoch": 0.0033351184768770463,
      "grad_norm": 2.888230800628662,
      "learning_rate": 1.667521806054387e-07,
      "loss": 0.1833,
      "step": 520
    },
    {
      "epoch": 0.0033992553706631433,
      "grad_norm": 2.5815255641937256,
      "learning_rate": 1.6995895330938946e-07,
      "loss": 0.1653,
      "step": 530
    },
    {
      "epoch": 0.0034633922644492403,
      "grad_norm": 2.9071788787841797,
      "learning_rate": 1.7316572601334018e-07,
      "loss": 0.1935,
      "step": 540
    },
    {
      "epoch": 0.0035275291582353373,
      "grad_norm": 2.7453720569610596,
      "learning_rate": 1.7637249871729094e-07,
      "loss": 0.1791,
      "step": 550
    },
    {
      "epoch": 0.0035916660520214343,
      "grad_norm": 2.607252836227417,
      "learning_rate": 1.7957927142124166e-07,
      "loss": 0.1677,
      "step": 560
    },
    {
      "epoch": 0.0036558029458075318,
      "grad_norm": 2.0972280502319336,
      "learning_rate": 1.8278604412519243e-07,
      "loss": 0.161,
      "step": 570
    },
    {
      "epoch": 0.003719939839593629,
      "grad_norm": 2.4978232383728027,
      "learning_rate": 1.8599281682914315e-07,
      "loss": 0.1847,
      "step": 580
    },
    {
      "epoch": 0.003784076733379726,
      "grad_norm": 3.0713841915130615,
      "learning_rate": 1.8919958953309391e-07,
      "loss": 0.1549,
      "step": 590
    },
    {
      "epoch": 0.003848213627165823,
      "grad_norm": 2.3856425285339355,
      "learning_rate": 1.9240636223704466e-07,
      "loss": 0.143,
      "step": 600
    },
    {
      "epoch": 0.00391235052095192,
      "grad_norm": 2.8760671615600586,
      "learning_rate": 1.956131349409954e-07,
      "loss": 0.1642,
      "step": 610
    },
    {
      "epoch": 0.003976487414738016,
      "grad_norm": 2.787055253982544,
      "learning_rate": 1.9881990764494614e-07,
      "loss": 0.1705,
      "step": 620
    },
    {
      "epoch": 0.004040624308524114,
      "grad_norm": 2.4248359203338623,
      "learning_rate": 2.020266803488969e-07,
      "loss": 0.1677,
      "step": 630
    },
    {
      "epoch": 0.004104761202310211,
      "grad_norm": 2.2203550338745117,
      "learning_rate": 2.0523345305284763e-07,
      "loss": 0.1379,
      "step": 640
    },
    {
      "epoch": 0.004168898096096308,
      "grad_norm": 2.3956847190856934,
      "learning_rate": 2.084402257567984e-07,
      "loss": 0.1413,
      "step": 650
    },
    {
      "epoch": 0.004233034989882405,
      "grad_norm": 2.7636780738830566,
      "learning_rate": 2.116469984607491e-07,
      "loss": 0.1464,
      "step": 660
    },
    {
      "epoch": 0.004297171883668502,
      "grad_norm": 2.452714204788208,
      "learning_rate": 2.1485377116469985e-07,
      "loss": 0.1429,
      "step": 670
    },
    {
      "epoch": 0.004361308777454599,
      "grad_norm": 2.292991876602173,
      "learning_rate": 2.180605438686506e-07,
      "loss": 0.1403,
      "step": 680
    },
    {
      "epoch": 0.004425445671240696,
      "grad_norm": 3.0862314701080322,
      "learning_rate": 2.2126731657260134e-07,
      "loss": 0.1445,
      "step": 690
    },
    {
      "epoch": 0.004489582565026793,
      "grad_norm": 2.497093439102173,
      "learning_rate": 2.244740892765521e-07,
      "loss": 0.1285,
      "step": 700
    },
    {
      "epoch": 0.00455371945881289,
      "grad_norm": 2.4969825744628906,
      "learning_rate": 2.2768086198050282e-07,
      "loss": 0.1384,
      "step": 710
    },
    {
      "epoch": 0.004617856352598987,
      "grad_norm": 2.2161011695861816,
      "learning_rate": 2.308876346844536e-07,
      "loss": 0.1266,
      "step": 720
    },
    {
      "epoch": 0.004681993246385084,
      "grad_norm": 2.202615737915039,
      "learning_rate": 2.340944073884043e-07,
      "loss": 0.1246,
      "step": 730
    },
    {
      "epoch": 0.004746130140171181,
      "grad_norm": 2.171642541885376,
      "learning_rate": 2.3730118009235508e-07,
      "loss": 0.1377,
      "step": 740
    },
    {
      "epoch": 0.004810267033957279,
      "grad_norm": 2.4430935382843018,
      "learning_rate": 2.405079527963058e-07,
      "loss": 0.116,
      "step": 750
    },
    {
      "epoch": 0.004874403927743375,
      "grad_norm": 2.2906432151794434,
      "learning_rate": 2.437147255002566e-07,
      "loss": 0.1166,
      "step": 760
    },
    {
      "epoch": 0.004938540821529473,
      "grad_norm": 2.0091114044189453,
      "learning_rate": 2.469214982042073e-07,
      "loss": 0.1279,
      "step": 770
    },
    {
      "epoch": 0.0050026777153155694,
      "grad_norm": 2.5473973751068115,
      "learning_rate": 2.501282709081581e-07,
      "loss": 0.1176,
      "step": 780
    },
    {
      "epoch": 0.005066814609101667,
      "grad_norm": 2.719332218170166,
      "learning_rate": 2.5333504361210877e-07,
      "loss": 0.1131,
      "step": 790
    },
    {
      "epoch": 0.0051309515028877635,
      "grad_norm": 2.3760056495666504,
      "learning_rate": 2.5654181631605956e-07,
      "loss": 0.1032,
      "step": 800
    },
    {
      "epoch": 0.005195088396673861,
      "grad_norm": 3.086043119430542,
      "learning_rate": 2.597485890200103e-07,
      "loss": 0.1031,
      "step": 810
    },
    {
      "epoch": 0.0052592252904599575,
      "grad_norm": 2.121891498565674,
      "learning_rate": 2.6295536172396105e-07,
      "loss": 0.1154,
      "step": 820
    },
    {
      "epoch": 0.005323362184246055,
      "grad_norm": 2.0646345615386963,
      "learning_rate": 2.661621344279118e-07,
      "loss": 0.0958,
      "step": 830
    },
    {
      "epoch": 0.0053874990780321515,
      "grad_norm": 2.8303093910217285,
      "learning_rate": 2.693689071318625e-07,
      "loss": 0.1153,
      "step": 840
    },
    {
      "epoch": 0.005451635971818249,
      "grad_norm": 2.213944673538208,
      "learning_rate": 2.7257567983581327e-07,
      "loss": 0.1195,
      "step": 850
    },
    {
      "epoch": 0.0055157728656043455,
      "grad_norm": 2.0894155502319336,
      "learning_rate": 2.75782452539764e-07,
      "loss": 0.0937,
      "step": 860
    },
    {
      "epoch": 0.005579909759390443,
      "grad_norm": 1.789013147354126,
      "learning_rate": 2.7898922524371476e-07,
      "loss": 0.1072,
      "step": 870
    },
    {
      "epoch": 0.00564404665317654,
      "grad_norm": 2.3917477130889893,
      "learning_rate": 2.821959979476655e-07,
      "loss": 0.114,
      "step": 880
    },
    {
      "epoch": 0.005708183546962637,
      "grad_norm": 3.153083086013794,
      "learning_rate": 2.8540277065161624e-07,
      "loss": 0.1051,
      "step": 890
    },
    {
      "epoch": 0.0057723204407487344,
      "grad_norm": 2.2773597240448,
      "learning_rate": 2.88609543355567e-07,
      "loss": 0.0967,
      "step": 900
    },
    {
      "epoch": 0.005836457334534831,
      "grad_norm": 3.3676769733428955,
      "learning_rate": 2.9181631605951773e-07,
      "loss": 0.1016,
      "step": 910
    },
    {
      "epoch": 0.0059005942283209285,
      "grad_norm": 2.569157600402832,
      "learning_rate": 2.9502308876346847e-07,
      "loss": 0.1016,
      "step": 920
    },
    {
      "epoch": 0.005964731122107025,
      "grad_norm": 2.6093432903289795,
      "learning_rate": 2.982298614674192e-07,
      "loss": 0.0815,
      "step": 930
    },
    {
      "epoch": 0.0060288680158931225,
      "grad_norm": 2.2556352615356445,
      "learning_rate": 3.0143663417136996e-07,
      "loss": 0.0999,
      "step": 940
    },
    {
      "epoch": 0.006093004909679219,
      "grad_norm": 2.0448052883148193,
      "learning_rate": 3.046434068753207e-07,
      "loss": 0.0885,
      "step": 950
    },
    {
      "epoch": 0.0061571418034653165,
      "grad_norm": 2.736361503601074,
      "learning_rate": 3.0785017957927144e-07,
      "loss": 0.0933,
      "step": 960
    },
    {
      "epoch": 0.006221278697251413,
      "grad_norm": 2.21384596824646,
      "learning_rate": 3.110569522832222e-07,
      "loss": 0.0926,
      "step": 970
    },
    {
      "epoch": 0.0062854155910375105,
      "grad_norm": 2.22341251373291,
      "learning_rate": 3.142637249871729e-07,
      "loss": 0.0842,
      "step": 980
    },
    {
      "epoch": 0.006349552484823607,
      "grad_norm": 1.8333430290222168,
      "learning_rate": 3.1747049769112367e-07,
      "loss": 0.0882,
      "step": 990
    },
    {
      "epoch": 0.0064136893786097045,
      "grad_norm": 2.1448442935943604,
      "learning_rate": 3.206772703950744e-07,
      "loss": 0.0885,
      "step": 1000
    },
    {
      "epoch": 0.006477826272395802,
      "grad_norm": 2.026063919067383,
      "learning_rate": 3.238840430990252e-07,
      "loss": 0.0797,
      "step": 1010
    },
    {
      "epoch": 0.006541963166181899,
      "grad_norm": 2.610535144805908,
      "learning_rate": 3.270908158029759e-07,
      "loss": 0.0939,
      "step": 1020
    },
    {
      "epoch": 0.006606100059967996,
      "grad_norm": 2.1977763175964355,
      "learning_rate": 3.3029758850692664e-07,
      "loss": 0.0874,
      "step": 1030
    },
    {
      "epoch": 0.006670236953754093,
      "grad_norm": 2.115877389907837,
      "learning_rate": 3.335043612108774e-07,
      "loss": 0.1032,
      "step": 1040
    },
    {
      "epoch": 0.00673437384754019,
      "grad_norm": 2.53721022605896,
      "learning_rate": 3.367111339148282e-07,
      "loss": 0.1079,
      "step": 1050
    },
    {
      "epoch": 0.006798510741326287,
      "grad_norm": 2.2047951221466064,
      "learning_rate": 3.399179066187789e-07,
      "loss": 0.0902,
      "step": 1060
    },
    {
      "epoch": 0.006862647635112384,
      "grad_norm": 1.9997676610946655,
      "learning_rate": 3.431246793227296e-07,
      "loss": 0.0878,
      "step": 1070
    },
    {
      "epoch": 0.006926784528898481,
      "grad_norm": 2.194669723510742,
      "learning_rate": 3.4633145202668035e-07,
      "loss": 0.0795,
      "step": 1080
    },
    {
      "epoch": 0.006990921422684578,
      "grad_norm": 2.847025156021118,
      "learning_rate": 3.4953822473063115e-07,
      "loss": 0.0812,
      "step": 1090
    },
    {
      "epoch": 0.007055058316470675,
      "grad_norm": 2.1623098850250244,
      "learning_rate": 3.527449974345819e-07,
      "loss": 0.0733,
      "step": 1100
    },
    {
      "epoch": 0.007119195210256772,
      "grad_norm": 2.1313631534576416,
      "learning_rate": 3.559517701385326e-07,
      "loss": 0.0887,
      "step": 1110
    },
    {
      "epoch": 0.007183332104042869,
      "grad_norm": 1.838502049446106,
      "learning_rate": 3.591585428424833e-07,
      "loss": 0.0816,
      "step": 1120
    },
    {
      "epoch": 0.007247468997828966,
      "grad_norm": 2.4724252223968506,
      "learning_rate": 3.623653155464341e-07,
      "loss": 0.0701,
      "step": 1130
    },
    {
      "epoch": 0.0073116058916150636,
      "grad_norm": 2.5112111568450928,
      "learning_rate": 3.6557208825038486e-07,
      "loss": 0.0797,
      "step": 1140
    },
    {
      "epoch": 0.00737574278540116,
      "grad_norm": 2.5839927196502686,
      "learning_rate": 3.687788609543356e-07,
      "loss": 0.0855,
      "step": 1150
    },
    {
      "epoch": 0.007439879679187258,
      "grad_norm": 3.895423650741577,
      "learning_rate": 3.719856336582863e-07,
      "loss": 0.0889,
      "step": 1160
    },
    {
      "epoch": 0.007504016572973354,
      "grad_norm": 2.3129162788391113,
      "learning_rate": 3.7519240636223703e-07,
      "loss": 0.0737,
      "step": 1170
    },
    {
      "epoch": 0.007568153466759452,
      "grad_norm": 2.2384352684020996,
      "learning_rate": 3.7839917906618783e-07,
      "loss": 0.0725,
      "step": 1180
    },
    {
      "epoch": 0.007632290360545548,
      "grad_norm": 2.1032588481903076,
      "learning_rate": 3.8160595177013857e-07,
      "loss": 0.0948,
      "step": 1190
    },
    {
      "epoch": 0.007696427254331646,
      "grad_norm": 2.255089044570923,
      "learning_rate": 3.848127244740893e-07,
      "loss": 0.0798,
      "step": 1200
    },
    {
      "epoch": 0.007760564148117742,
      "grad_norm": 1.7414604425430298,
      "learning_rate": 3.8801949717804e-07,
      "loss": 0.0772,
      "step": 1210
    },
    {
      "epoch": 0.00782470104190384,
      "grad_norm": 1.687619924545288,
      "learning_rate": 3.912262698819908e-07,
      "loss": 0.0706,
      "step": 1220
    },
    {
      "epoch": 0.007888837935689936,
      "grad_norm": 2.5747528076171875,
      "learning_rate": 3.9443304258594154e-07,
      "loss": 0.0759,
      "step": 1230
    },
    {
      "epoch": 0.007952974829476033,
      "grad_norm": 2.255894660949707,
      "learning_rate": 3.976398152898923e-07,
      "loss": 0.0845,
      "step": 1240
    },
    {
      "epoch": 0.008017111723262131,
      "grad_norm": 2.367706775665283,
      "learning_rate": 4.0084658799384303e-07,
      "loss": 0.092,
      "step": 1250
    },
    {
      "epoch": 0.008081248617048228,
      "grad_norm": 1.8855806589126587,
      "learning_rate": 4.040533606977938e-07,
      "loss": 0.0684,
      "step": 1260
    },
    {
      "epoch": 0.008145385510834324,
      "grad_norm": 2.4973232746124268,
      "learning_rate": 4.072601334017445e-07,
      "loss": 0.0626,
      "step": 1270
    },
    {
      "epoch": 0.008209522404620423,
      "grad_norm": 2.1072261333465576,
      "learning_rate": 4.1046690610569525e-07,
      "loss": 0.0799,
      "step": 1280
    },
    {
      "epoch": 0.00827365929840652,
      "grad_norm": 1.9451370239257812,
      "learning_rate": 4.13673678809646e-07,
      "loss": 0.0763,
      "step": 1290
    },
    {
      "epoch": 0.008337796192192616,
      "grad_norm": 2.526870012283325,
      "learning_rate": 4.168804515135968e-07,
      "loss": 0.0799,
      "step": 1300
    },
    {
      "epoch": 0.008401933085978712,
      "grad_norm": 1.7936248779296875,
      "learning_rate": 4.2008722421754754e-07,
      "loss": 0.0657,
      "step": 1310
    },
    {
      "epoch": 0.00846606997976481,
      "grad_norm": 1.7973401546478271,
      "learning_rate": 4.232939969214982e-07,
      "loss": 0.0768,
      "step": 1320
    },
    {
      "epoch": 0.008530206873550907,
      "grad_norm": 2.051457643508911,
      "learning_rate": 4.2650076962544897e-07,
      "loss": 0.0839,
      "step": 1330
    },
    {
      "epoch": 0.008594343767337004,
      "grad_norm": 1.8935335874557495,
      "learning_rate": 4.297075423293997e-07,
      "loss": 0.0722,
      "step": 1340
    },
    {
      "epoch": 0.0086584806611231,
      "grad_norm": 1.8838813304901123,
      "learning_rate": 4.329143150333505e-07,
      "loss": 0.0655,
      "step": 1350
    },
    {
      "epoch": 0.008722617554909199,
      "grad_norm": 1.9510728120803833,
      "learning_rate": 4.361210877373012e-07,
      "loss": 0.0616,
      "step": 1360
    },
    {
      "epoch": 0.008786754448695295,
      "grad_norm": 1.8975841999053955,
      "learning_rate": 4.3932786044125194e-07,
      "loss": 0.0611,
      "step": 1370
    },
    {
      "epoch": 0.008850891342481392,
      "grad_norm": 1.757426142692566,
      "learning_rate": 4.425346331452027e-07,
      "loss": 0.0648,
      "step": 1380
    },
    {
      "epoch": 0.00891502823626749,
      "grad_norm": 2.4363205432891846,
      "learning_rate": 4.457414058491535e-07,
      "loss": 0.0687,
      "step": 1390
    },
    {
      "epoch": 0.008979165130053587,
      "grad_norm": 2.466268301010132,
      "learning_rate": 4.489481785531042e-07,
      "loss": 0.0593,
      "step": 1400
    },
    {
      "epoch": 0.009043302023839683,
      "grad_norm": 2.0895450115203857,
      "learning_rate": 4.521549512570549e-07,
      "loss": 0.0619,
      "step": 1410
    },
    {
      "epoch": 0.00910743891762578,
      "grad_norm": 1.737841248512268,
      "learning_rate": 4.5536172396100565e-07,
      "loss": 0.066,
      "step": 1420
    },
    {
      "epoch": 0.009171575811411878,
      "grad_norm": 1.9468801021575928,
      "learning_rate": 4.5856849666495645e-07,
      "loss": 0.0703,
      "step": 1430
    },
    {
      "epoch": 0.009235712705197975,
      "grad_norm": 1.5404072999954224,
      "learning_rate": 4.617752693689072e-07,
      "loss": 0.0702,
      "step": 1440
    },
    {
      "epoch": 0.009299849598984071,
      "grad_norm": 1.623038411140442,
      "learning_rate": 4.6498204207285793e-07,
      "loss": 0.0684,
      "step": 1450
    },
    {
      "epoch": 0.009363986492770168,
      "grad_norm": 1.524778127670288,
      "learning_rate": 4.681888147768086e-07,
      "loss": 0.0503,
      "step": 1460
    },
    {
      "epoch": 0.009428123386556266,
      "grad_norm": 2.2616991996765137,
      "learning_rate": 4.713955874807594e-07,
      "loss": 0.0728,
      "step": 1470
    },
    {
      "epoch": 0.009492260280342363,
      "grad_norm": 1.9728755950927734,
      "learning_rate": 4.7460236018471016e-07,
      "loss": 0.0634,
      "step": 1480
    },
    {
      "epoch": 0.00955639717412846,
      "grad_norm": 2.3513500690460205,
      "learning_rate": 4.778091328886609e-07,
      "loss": 0.076,
      "step": 1490
    },
    {
      "epoch": 0.009620534067914558,
      "grad_norm": 2.413816452026367,
      "learning_rate": 4.810159055926116e-07,
      "loss": 0.0559,
      "step": 1500
    },
    {
      "epoch": 0.009684670961700654,
      "grad_norm": 2.188338041305542,
      "learning_rate": 4.842226782965624e-07,
      "loss": 0.0679,
      "step": 1510
    },
    {
      "epoch": 0.00974880785548675,
      "grad_norm": 1.5445910692214966,
      "learning_rate": 4.874294510005132e-07,
      "loss": 0.0616,
      "step": 1520
    },
    {
      "epoch": 0.009812944749272847,
      "grad_norm": 1.6505457162857056,
      "learning_rate": 4.906362237044639e-07,
      "loss": 0.0559,
      "step": 1530
    },
    {
      "epoch": 0.009877081643058946,
      "grad_norm": 2.2405667304992676,
      "learning_rate": 4.938429964084146e-07,
      "loss": 0.0518,
      "step": 1540
    },
    {
      "epoch": 0.009941218536845042,
      "grad_norm": 1.8405488729476929,
      "learning_rate": 4.970497691123654e-07,
      "loss": 0.0571,
      "step": 1550
    },
    {
      "epoch": 0.010005355430631139,
      "grad_norm": 1.7272032499313354,
      "learning_rate": 5.002565418163162e-07,
      "loss": 0.0643,
      "step": 1560
    },
    {
      "epoch": 0.010069492324417235,
      "grad_norm": 2.216449499130249,
      "learning_rate": 5.034633145202668e-07,
      "loss": 0.0602,
      "step": 1570
    },
    {
      "epoch": 0.010133629218203334,
      "grad_norm": 1.555444598197937,
      "learning_rate": 5.066700872242175e-07,
      "loss": 0.0601,
      "step": 1580
    },
    {
      "epoch": 0.01019776611198943,
      "grad_norm": 1.8560059070587158,
      "learning_rate": 5.098768599281683e-07,
      "loss": 0.0646,
      "step": 1590
    },
    {
      "epoch": 0.010261903005775527,
      "grad_norm": 1.8203860521316528,
      "learning_rate": 5.130836326321191e-07,
      "loss": 0.0587,
      "step": 1600
    },
    {
      "epoch": 0.010326039899561624,
      "grad_norm": 1.401741623878479,
      "learning_rate": 5.162904053360698e-07,
      "loss": 0.0666,
      "step": 1610
    },
    {
      "epoch": 0.010390176793347722,
      "grad_norm": 2.429461717605591,
      "learning_rate": 5.194971780400206e-07,
      "loss": 0.0671,
      "step": 1620
    },
    {
      "epoch": 0.010454313687133818,
      "grad_norm": 1.3045133352279663,
      "learning_rate": 5.227039507439713e-07,
      "loss": 0.0552,
      "step": 1630
    },
    {
      "epoch": 0.010518450580919915,
      "grad_norm": 1.1228832006454468,
      "learning_rate": 5.259107234479221e-07,
      "loss": 0.0569,
      "step": 1640
    },
    {
      "epoch": 0.010582587474706013,
      "grad_norm": 2.608595848083496,
      "learning_rate": 5.291174961518728e-07,
      "loss": 0.0614,
      "step": 1650
    },
    {
      "epoch": 0.01064672436849211,
      "grad_norm": 2.572727918624878,
      "learning_rate": 5.323242688558236e-07,
      "loss": 0.0639,
      "step": 1660
    },
    {
      "epoch": 0.010710861262278206,
      "grad_norm": 1.7884821891784668,
      "learning_rate": 5.355310415597743e-07,
      "loss": 0.0607,
      "step": 1670
    },
    {
      "epoch": 0.010774998156064303,
      "grad_norm": 2.0818710327148438,
      "learning_rate": 5.38737814263725e-07,
      "loss": 0.0455,
      "step": 1680
    },
    {
      "epoch": 0.010839135049850401,
      "grad_norm": 1.8627077341079712,
      "learning_rate": 5.419445869676758e-07,
      "loss": 0.0551,
      "step": 1690
    },
    {
      "epoch": 0.010903271943636498,
      "grad_norm": 1.7816085815429688,
      "learning_rate": 5.451513596716265e-07,
      "loss": 0.0463,
      "step": 1700
    },
    {
      "epoch": 0.010967408837422594,
      "grad_norm": 1.6360572576522827,
      "learning_rate": 5.483581323755772e-07,
      "loss": 0.0448,
      "step": 1710
    },
    {
      "epoch": 0.011031545731208691,
      "grad_norm": 1.8366321325302124,
      "learning_rate": 5.51564905079528e-07,
      "loss": 0.0543,
      "step": 1720
    },
    {
      "epoch": 0.01109568262499479,
      "grad_norm": 1.3522605895996094,
      "learning_rate": 5.547716777834788e-07,
      "loss": 0.0576,
      "step": 1730
    },
    {
      "epoch": 0.011159819518780886,
      "grad_norm": 1.83017897605896,
      "learning_rate": 5.579784504874295e-07,
      "loss": 0.0636,
      "step": 1740
    },
    {
      "epoch": 0.011223956412566983,
      "grad_norm": 1.6681979894638062,
      "learning_rate": 5.611852231913802e-07,
      "loss": 0.063,
      "step": 1750
    },
    {
      "epoch": 0.01128809330635308,
      "grad_norm": 1.7300772666931152,
      "learning_rate": 5.64391995895331e-07,
      "loss": 0.0581,
      "step": 1760
    },
    {
      "epoch": 0.011352230200139177,
      "grad_norm": 1.3762904405593872,
      "learning_rate": 5.675987685992818e-07,
      "loss": 0.0606,
      "step": 1770
    },
    {
      "epoch": 0.011416367093925274,
      "grad_norm": 1.9628626108169556,
      "learning_rate": 5.708055413032325e-07,
      "loss": 0.0671,
      "step": 1780
    },
    {
      "epoch": 0.01148050398771137,
      "grad_norm": 1.9082118272781372,
      "learning_rate": 5.740123140071832e-07,
      "loss": 0.0609,
      "step": 1790
    },
    {
      "epoch": 0.011544640881497469,
      "grad_norm": 2.207084894180298,
      "learning_rate": 5.77219086711134e-07,
      "loss": 0.0571,
      "step": 1800
    },
    {
      "epoch": 0.011608777775283565,
      "grad_norm": 1.3231322765350342,
      "learning_rate": 5.804258594150848e-07,
      "loss": 0.0496,
      "step": 1810
    },
    {
      "epoch": 0.011672914669069662,
      "grad_norm": 1.5998111963272095,
      "learning_rate": 5.836326321190355e-07,
      "loss": 0.0606,
      "step": 1820
    },
    {
      "epoch": 0.011737051562855759,
      "grad_norm": 1.6842598915100098,
      "learning_rate": 5.868394048229861e-07,
      "loss": 0.0557,
      "step": 1830
    },
    {
      "epoch": 0.011801188456641857,
      "grad_norm": 1.4744997024536133,
      "learning_rate": 5.900461775269369e-07,
      "loss": 0.0567,
      "step": 1840
    },
    {
      "epoch": 0.011865325350427953,
      "grad_norm": 2.192277431488037,
      "learning_rate": 5.932529502308876e-07,
      "loss": 0.0528,
      "step": 1850
    },
    {
      "epoch": 0.01192946224421405,
      "grad_norm": 1.7011133432388306,
      "learning_rate": 5.964597229348384e-07,
      "loss": 0.0459,
      "step": 1860
    },
    {
      "epoch": 0.011993599138000147,
      "grad_norm": 1.7284241914749146,
      "learning_rate": 5.996664956387892e-07,
      "loss": 0.0522,
      "step": 1870
    },
    {
      "epoch": 0.012057736031786245,
      "grad_norm": 1.853323221206665,
      "learning_rate": 6.028732683427399e-07,
      "loss": 0.0577,
      "step": 1880
    },
    {
      "epoch": 0.012121872925572342,
      "grad_norm": 1.6866708993911743,
      "learning_rate": 6.060800410466906e-07,
      "loss": 0.051,
      "step": 1890
    },
    {
      "epoch": 0.012186009819358438,
      "grad_norm": 2.820150852203369,
      "learning_rate": 6.092868137506414e-07,
      "loss": 0.0478,
      "step": 1900
    },
    {
      "epoch": 0.012250146713144536,
      "grad_norm": 1.3693703413009644,
      "learning_rate": 6.124935864545922e-07,
      "loss": 0.0417,
      "step": 1910
    },
    {
      "epoch": 0.012314283606930633,
      "grad_norm": 1.4814716577529907,
      "learning_rate": 6.157003591585429e-07,
      "loss": 0.0447,
      "step": 1920
    },
    {
      "epoch": 0.01237842050071673,
      "grad_norm": 1.7265548706054688,
      "learning_rate": 6.189071318624936e-07,
      "loss": 0.0519,
      "step": 1930
    },
    {
      "epoch": 0.012442557394502826,
      "grad_norm": 1.7359428405761719,
      "learning_rate": 6.221139045664444e-07,
      "loss": 0.0465,
      "step": 1940
    },
    {
      "epoch": 0.012506694288288924,
      "grad_norm": 1.5335360765457153,
      "learning_rate": 6.253206772703952e-07,
      "loss": 0.0476,
      "step": 1950
    },
    {
      "epoch": 0.012570831182075021,
      "grad_norm": 1.7577869892120361,
      "learning_rate": 6.285274499743459e-07,
      "loss": 0.0431,
      "step": 1960
    },
    {
      "epoch": 0.012634968075861118,
      "grad_norm": 1.253122091293335,
      "learning_rate": 6.317342226782965e-07,
      "loss": 0.0475,
      "step": 1970
    },
    {
      "epoch": 0.012699104969647214,
      "grad_norm": 1.7545239925384521,
      "learning_rate": 6.349409953822473e-07,
      "loss": 0.0523,
      "step": 1980
    },
    {
      "epoch": 0.012763241863433313,
      "grad_norm": 1.6145719289779663,
      "learning_rate": 6.38147768086198e-07,
      "loss": 0.0532,
      "step": 1990
    },
    {
      "epoch": 0.012827378757219409,
      "grad_norm": 1.6796982288360596,
      "learning_rate": 6.413545407901488e-07,
      "loss": 0.0468,
      "step": 2000
    },
    {
      "epoch": 0.012891515651005506,
      "grad_norm": 1.7424479722976685,
      "learning_rate": 6.445613134940996e-07,
      "loss": 0.0436,
      "step": 2010
    },
    {
      "epoch": 0.012955652544791604,
      "grad_norm": 2.4510207176208496,
      "learning_rate": 6.477680861980504e-07,
      "loss": 0.0602,
      "step": 2020
    },
    {
      "epoch": 0.0130197894385777,
      "grad_norm": 1.4376516342163086,
      "learning_rate": 6.509748589020011e-07,
      "loss": 0.0463,
      "step": 2030
    },
    {
      "epoch": 0.013083926332363797,
      "grad_norm": 2.1743013858795166,
      "learning_rate": 6.541816316059518e-07,
      "loss": 0.0424,
      "step": 2040
    },
    {
      "epoch": 0.013148063226149894,
      "grad_norm": 1.9213758707046509,
      "learning_rate": 6.573884043099026e-07,
      "loss": 0.0475,
      "step": 2050
    },
    {
      "epoch": 0.013212200119935992,
      "grad_norm": 2.2999606132507324,
      "learning_rate": 6.605951770138533e-07,
      "loss": 0.0678,
      "step": 2060
    },
    {
      "epoch": 0.013276337013722089,
      "grad_norm": 1.6947029829025269,
      "learning_rate": 6.63801949717804e-07,
      "loss": 0.0496,
      "step": 2070
    },
    {
      "epoch": 0.013340473907508185,
      "grad_norm": 2.068711996078491,
      "learning_rate": 6.670087224217548e-07,
      "loss": 0.0513,
      "step": 2080
    },
    {
      "epoch": 0.013404610801294282,
      "grad_norm": 2.0679399967193604,
      "learning_rate": 6.702154951257056e-07,
      "loss": 0.0608,
      "step": 2090
    },
    {
      "epoch": 0.01346874769508038,
      "grad_norm": 1.390113115310669,
      "learning_rate": 6.734222678296564e-07,
      "loss": 0.0445,
      "step": 2100
    },
    {
      "epoch": 0.013532884588866477,
      "grad_norm": 2.04164457321167,
      "learning_rate": 6.76629040533607e-07,
      "loss": 0.05,
      "step": 2110
    },
    {
      "epoch": 0.013597021482652573,
      "grad_norm": 3.0909676551818848,
      "learning_rate": 6.798358132375578e-07,
      "loss": 0.048,
      "step": 2120
    },
    {
      "epoch": 0.013661158376438672,
      "grad_norm": 1.5751962661743164,
      "learning_rate": 6.830425859415085e-07,
      "loss": 0.0506,
      "step": 2130
    },
    {
      "epoch": 0.013725295270224768,
      "grad_norm": 1.8469059467315674,
      "learning_rate": 6.862493586454592e-07,
      "loss": 0.0482,
      "step": 2140
    },
    {
      "epoch": 0.013789432164010865,
      "grad_norm": 2.0632050037384033,
      "learning_rate": 6.8945613134941e-07,
      "loss": 0.047,
      "step": 2150
    },
    {
      "epoch": 0.013853569057796961,
      "grad_norm": 1.7224044799804688,
      "learning_rate": 6.926629040533607e-07,
      "loss": 0.0413,
      "step": 2160
    },
    {
      "epoch": 0.01391770595158306,
      "grad_norm": 1.2005095481872559,
      "learning_rate": 6.958696767573114e-07,
      "loss": 0.0468,
      "step": 2170
    },
    {
      "epoch": 0.013981842845369156,
      "grad_norm": 1.6894320249557495,
      "learning_rate": 6.990764494612623e-07,
      "loss": 0.0497,
      "step": 2180
    },
    {
      "epoch": 0.014045979739155253,
      "grad_norm": 2.376833915710449,
      "learning_rate": 7.02283222165213e-07,
      "loss": 0.0418,
      "step": 2190
    },
    {
      "epoch": 0.01411011663294135,
      "grad_norm": 1.680874228477478,
      "learning_rate": 7.054899948691638e-07,
      "loss": 0.0498,
      "step": 2200
    },
    {
      "epoch": 0.014174253526727448,
      "grad_norm": 1.6105834245681763,
      "learning_rate": 7.086967675731145e-07,
      "loss": 0.0487,
      "step": 2210
    },
    {
      "epoch": 0.014238390420513544,
      "grad_norm": 1.9710890054702759,
      "learning_rate": 7.119035402770652e-07,
      "loss": 0.0497,
      "step": 2220
    },
    {
      "epoch": 0.01430252731429964,
      "grad_norm": 1.7088145017623901,
      "learning_rate": 7.15110312981016e-07,
      "loss": 0.0385,
      "step": 2230
    },
    {
      "epoch": 0.014366664208085737,
      "grad_norm": 2.4289820194244385,
      "learning_rate": 7.183170856849666e-07,
      "loss": 0.0516,
      "step": 2240
    },
    {
      "epoch": 0.014430801101871836,
      "grad_norm": 1.693321943283081,
      "learning_rate": 7.215238583889174e-07,
      "loss": 0.0435,
      "step": 2250
    },
    {
      "epoch": 0.014494937995657932,
      "grad_norm": 1.9192099571228027,
      "learning_rate": 7.247306310928682e-07,
      "loss": 0.0482,
      "step": 2260
    },
    {
      "epoch": 0.014559074889444029,
      "grad_norm": 0.8561009764671326,
      "learning_rate": 7.27937403796819e-07,
      "loss": 0.0493,
      "step": 2270
    },
    {
      "epoch": 0.014623211783230127,
      "grad_norm": 1.7811331748962402,
      "learning_rate": 7.311441765007697e-07,
      "loss": 0.049,
      "step": 2280
    },
    {
      "epoch": 0.014687348677016224,
      "grad_norm": 1.2994056940078735,
      "learning_rate": 7.343509492047204e-07,
      "loss": 0.0409,
      "step": 2290
    },
    {
      "epoch": 0.01475148557080232,
      "grad_norm": 1.252750039100647,
      "learning_rate": 7.375577219086712e-07,
      "loss": 0.042,
      "step": 2300
    },
    {
      "epoch": 0.014815622464588417,
      "grad_norm": 1.558959722518921,
      "learning_rate": 7.407644946126219e-07,
      "loss": 0.0445,
      "step": 2310
    },
    {
      "epoch": 0.014879759358374515,
      "grad_norm": 1.7213983535766602,
      "learning_rate": 7.439712673165726e-07,
      "loss": 0.0394,
      "step": 2320
    },
    {
      "epoch": 0.014943896252160612,
      "grad_norm": 1.8647485971450806,
      "learning_rate": 7.471780400205234e-07,
      "loss": 0.0441,
      "step": 2330
    },
    {
      "epoch": 0.015008033145946708,
      "grad_norm": 2.38940167427063,
      "learning_rate": 7.503848127244741e-07,
      "loss": 0.0403,
      "step": 2340
    },
    {
      "epoch": 0.015072170039732805,
      "grad_norm": 1.6554349660873413,
      "learning_rate": 7.53591585428425e-07,
      "loss": 0.0441,
      "step": 2350
    },
    {
      "epoch": 0.015136306933518903,
      "grad_norm": 1.9792137145996094,
      "learning_rate": 7.567983581323757e-07,
      "loss": 0.0478,
      "step": 2360
    },
    {
      "epoch": 0.015200443827305,
      "grad_norm": 1.3561503887176514,
      "learning_rate": 7.600051308363265e-07,
      "loss": 0.0414,
      "step": 2370
    },
    {
      "epoch": 0.015264580721091096,
      "grad_norm": 2.6571028232574463,
      "learning_rate": 7.632119035402771e-07,
      "loss": 0.0393,
      "step": 2380
    },
    {
      "epoch": 0.015328717614877195,
      "grad_norm": 1.7432042360305786,
      "learning_rate": 7.664186762442278e-07,
      "loss": 0.0366,
      "step": 2390
    },
    {
      "epoch": 0.015392854508663291,
      "grad_norm": 1.9216645956039429,
      "learning_rate": 7.696254489481786e-07,
      "loss": 0.0399,
      "step": 2400
    },
    {
      "epoch": 0.015456991402449388,
      "grad_norm": 1.4957715272903442,
      "learning_rate": 7.728322216521293e-07,
      "loss": 0.0412,
      "step": 2410
    },
    {
      "epoch": 0.015521128296235484,
      "grad_norm": 2.0664377212524414,
      "learning_rate": 7.7603899435608e-07,
      "loss": 0.0315,
      "step": 2420
    },
    {
      "epoch": 0.015585265190021583,
      "grad_norm": 1.9414721727371216,
      "learning_rate": 7.792457670600309e-07,
      "loss": 0.047,
      "step": 2430
    },
    {
      "epoch": 0.01564940208380768,
      "grad_norm": 1.556045413017273,
      "learning_rate": 7.824525397639816e-07,
      "loss": 0.0401,
      "step": 2440
    },
    {
      "epoch": 0.015713538977593776,
      "grad_norm": 1.55962336063385,
      "learning_rate": 7.856593124679324e-07,
      "loss": 0.0421,
      "step": 2450
    },
    {
      "epoch": 0.015777675871379872,
      "grad_norm": 1.3714858293533325,
      "learning_rate": 7.888660851718831e-07,
      "loss": 0.0382,
      "step": 2460
    },
    {
      "epoch": 0.01584181276516597,
      "grad_norm": 1.7481050491333008,
      "learning_rate": 7.920728578758338e-07,
      "loss": 0.0564,
      "step": 2470
    },
    {
      "epoch": 0.015905949658952066,
      "grad_norm": 1.4160561561584473,
      "learning_rate": 7.952796305797846e-07,
      "loss": 0.0358,
      "step": 2480
    },
    {
      "epoch": 0.015970086552738166,
      "grad_norm": 2.030376434326172,
      "learning_rate": 7.984864032837353e-07,
      "loss": 0.0425,
      "step": 2490
    },
    {
      "epoch": 0.016034223446524262,
      "grad_norm": 1.5082240104675293,
      "learning_rate": 8.016931759876861e-07,
      "loss": 0.0461,
      "step": 2500
    },
    {
      "epoch": 0.01609836034031036,
      "grad_norm": 1.9544262886047363,
      "learning_rate": 8.048999486916367e-07,
      "loss": 0.0383,
      "step": 2510
    },
    {
      "epoch": 0.016162497234096455,
      "grad_norm": 1.6944468021392822,
      "learning_rate": 8.081067213955876e-07,
      "loss": 0.046,
      "step": 2520
    },
    {
      "epoch": 0.016226634127882552,
      "grad_norm": 1.5525201559066772,
      "learning_rate": 8.113134940995383e-07,
      "loss": 0.0356,
      "step": 2530
    },
    {
      "epoch": 0.01629077102166865,
      "grad_norm": 2.0058481693267822,
      "learning_rate": 8.14520266803489e-07,
      "loss": 0.0449,
      "step": 2540
    },
    {
      "epoch": 0.016354907915454745,
      "grad_norm": 1.2101256847381592,
      "learning_rate": 8.177270395074398e-07,
      "loss": 0.0329,
      "step": 2550
    },
    {
      "epoch": 0.016419044809240845,
      "grad_norm": 1.590275764465332,
      "learning_rate": 8.209338122113905e-07,
      "loss": 0.0438,
      "step": 2560
    },
    {
      "epoch": 0.016483181703026942,
      "grad_norm": 1.4366024732589722,
      "learning_rate": 8.241405849153412e-07,
      "loss": 0.0358,
      "step": 2570
    },
    {
      "epoch": 0.01654731859681304,
      "grad_norm": 1.3212709426879883,
      "learning_rate": 8.27347357619292e-07,
      "loss": 0.0422,
      "step": 2580
    },
    {
      "epoch": 0.016611455490599135,
      "grad_norm": 1.3149542808532715,
      "learning_rate": 8.305541303232427e-07,
      "loss": 0.0451,
      "step": 2590
    },
    {
      "epoch": 0.01667559238438523,
      "grad_norm": 2.0486605167388916,
      "learning_rate": 8.337609030271936e-07,
      "loss": 0.0303,
      "step": 2600
    },
    {
      "epoch": 0.016739729278171328,
      "grad_norm": 2.1981372833251953,
      "learning_rate": 8.369676757311443e-07,
      "loss": 0.0503,
      "step": 2610
    },
    {
      "epoch": 0.016803866171957425,
      "grad_norm": 1.6597967147827148,
      "learning_rate": 8.401744484350951e-07,
      "loss": 0.0362,
      "step": 2620
    },
    {
      "epoch": 0.016868003065743525,
      "grad_norm": 2.1460511684417725,
      "learning_rate": 8.433812211390458e-07,
      "loss": 0.0434,
      "step": 2630
    },
    {
      "epoch": 0.01693213995952962,
      "grad_norm": 1.2565027475357056,
      "learning_rate": 8.465879938429964e-07,
      "loss": 0.0464,
      "step": 2640
    },
    {
      "epoch": 0.016996276853315718,
      "grad_norm": 1.4828057289123535,
      "learning_rate": 8.497947665469472e-07,
      "loss": 0.0373,
      "step": 2650
    },
    {
      "epoch": 0.017060413747101814,
      "grad_norm": 2.4982426166534424,
      "learning_rate": 8.530015392508979e-07,
      "loss": 0.0374,
      "step": 2660
    },
    {
      "epoch": 0.01712455064088791,
      "grad_norm": 1.8210854530334473,
      "learning_rate": 8.562083119548486e-07,
      "loss": 0.0446,
      "step": 2670
    },
    {
      "epoch": 0.017188687534674008,
      "grad_norm": 1.7910974025726318,
      "learning_rate": 8.594150846587994e-07,
      "loss": 0.0382,
      "step": 2680
    },
    {
      "epoch": 0.017252824428460104,
      "grad_norm": 1.5497280359268188,
      "learning_rate": 8.626218573627502e-07,
      "loss": 0.0439,
      "step": 2690
    },
    {
      "epoch": 0.0173169613222462,
      "grad_norm": 2.107048273086548,
      "learning_rate": 8.65828630066701e-07,
      "loss": 0.0323,
      "step": 2700
    },
    {
      "epoch": 0.0173810982160323,
      "grad_norm": 1.7939436435699463,
      "learning_rate": 8.690354027706517e-07,
      "loss": 0.0375,
      "step": 2710
    },
    {
      "epoch": 0.017445235109818397,
      "grad_norm": 1.5770498514175415,
      "learning_rate": 8.722421754746024e-07,
      "loss": 0.0365,
      "step": 2720
    },
    {
      "epoch": 0.017509372003604494,
      "grad_norm": 2.1870534420013428,
      "learning_rate": 8.754489481785532e-07,
      "loss": 0.0436,
      "step": 2730
    },
    {
      "epoch": 0.01757350889739059,
      "grad_norm": 2.0507476329803467,
      "learning_rate": 8.786557208825039e-07,
      "loss": 0.0301,
      "step": 2740
    },
    {
      "epoch": 0.017637645791176687,
      "grad_norm": 1.1799434423446655,
      "learning_rate": 8.818624935864547e-07,
      "loss": 0.0411,
      "step": 2750
    },
    {
      "epoch": 0.017701782684962784,
      "grad_norm": 1.364424467086792,
      "learning_rate": 8.850692662904054e-07,
      "loss": 0.0413,
      "step": 2760
    },
    {
      "epoch": 0.01776591957874888,
      "grad_norm": 2.17989182472229,
      "learning_rate": 8.882760389943563e-07,
      "loss": 0.0446,
      "step": 2770
    },
    {
      "epoch": 0.01783005647253498,
      "grad_norm": 1.3833954334259033,
      "learning_rate": 8.91482811698307e-07,
      "loss": 0.0448,
      "step": 2780
    },
    {
      "epoch": 0.017894193366321077,
      "grad_norm": 1.3813868761062622,
      "learning_rate": 8.946895844022576e-07,
      "loss": 0.056,
      "step": 2790
    },
    {
      "epoch": 0.017958330260107173,
      "grad_norm": 1.6611545085906982,
      "learning_rate": 8.978963571062084e-07,
      "loss": 0.0317,
      "step": 2800
    },
    {
      "epoch": 0.01802246715389327,
      "grad_norm": 2.107382297515869,
      "learning_rate": 9.011031298101591e-07,
      "loss": 0.0386,
      "step": 2810
    },
    {
      "epoch": 0.018086604047679367,
      "grad_norm": 1.9452520608901978,
      "learning_rate": 9.043099025141098e-07,
      "loss": 0.0334,
      "step": 2820
    },
    {
      "epoch": 0.018150740941465463,
      "grad_norm": 1.8797872066497803,
      "learning_rate": 9.075166752180606e-07,
      "loss": 0.0378,
      "step": 2830
    },
    {
      "epoch": 0.01821487783525156,
      "grad_norm": 1.349656105041504,
      "learning_rate": 9.107234479220113e-07,
      "loss": 0.0408,
      "step": 2840
    },
    {
      "epoch": 0.018279014729037656,
      "grad_norm": 1.657008409500122,
      "learning_rate": 9.13930220625962e-07,
      "loss": 0.0432,
      "step": 2850
    },
    {
      "epoch": 0.018343151622823756,
      "grad_norm": 1.0578302145004272,
      "learning_rate": 9.171369933299129e-07,
      "loss": 0.0468,
      "step": 2860
    },
    {
      "epoch": 0.018407288516609853,
      "grad_norm": 1.3269705772399902,
      "learning_rate": 9.203437660338637e-07,
      "loss": 0.0331,
      "step": 2870
    },
    {
      "epoch": 0.01847142541039595,
      "grad_norm": 1.4782181978225708,
      "learning_rate": 9.235505387378144e-07,
      "loss": 0.0297,
      "step": 2880
    },
    {
      "epoch": 0.018535562304182046,
      "grad_norm": 1.494398832321167,
      "learning_rate": 9.267573114417651e-07,
      "loss": 0.0348,
      "step": 2890
    },
    {
      "epoch": 0.018599699197968143,
      "grad_norm": 2.166395664215088,
      "learning_rate": 9.299640841457159e-07,
      "loss": 0.0366,
      "step": 2900
    },
    {
      "epoch": 0.01866383609175424,
      "grad_norm": 1.7287582159042358,
      "learning_rate": 9.331708568496666e-07,
      "loss": 0.036,
      "step": 2910
    },
    {
      "epoch": 0.018727972985540336,
      "grad_norm": 1.187628149986267,
      "learning_rate": 9.363776295536172e-07,
      "loss": 0.0344,
      "step": 2920
    },
    {
      "epoch": 0.018792109879326436,
      "grad_norm": 2.1952199935913086,
      "learning_rate": 9.39584402257568e-07,
      "loss": 0.0374,
      "step": 2930
    },
    {
      "epoch": 0.018856246773112532,
      "grad_norm": 1.5708807706832886,
      "learning_rate": 9.427911749615188e-07,
      "loss": 0.0244,
      "step": 2940
    },
    {
      "epoch": 0.01892038366689863,
      "grad_norm": 1.7028510570526123,
      "learning_rate": 9.459979476654696e-07,
      "loss": 0.0365,
      "step": 2950
    },
    {
      "epoch": 0.018984520560684726,
      "grad_norm": 1.1860458850860596,
      "learning_rate": 9.492047203694203e-07,
      "loss": 0.0328,
      "step": 2960
    },
    {
      "epoch": 0.019048657454470822,
      "grad_norm": 1.1099803447723389,
      "learning_rate": 9.52411493073371e-07,
      "loss": 0.0321,
      "step": 2970
    },
    {
      "epoch": 0.01911279434825692,
      "grad_norm": 1.7563613653182983,
      "learning_rate": 9.556182657773218e-07,
      "loss": 0.0387,
      "step": 2980
    },
    {
      "epoch": 0.019176931242043015,
      "grad_norm": 2.043030023574829,
      "learning_rate": 9.588250384812725e-07,
      "loss": 0.036,
      "step": 2990
    },
    {
      "epoch": 0.019241068135829115,
      "grad_norm": 1.5777499675750732,
      "learning_rate": 9.620318111852232e-07,
      "loss": 0.0326,
      "step": 3000
    },
    {
      "epoch": 0.019305205029615212,
      "grad_norm": 1.2317931652069092,
      "learning_rate": 9.652385838891739e-07,
      "loss": 0.0302,
      "step": 3010
    },
    {
      "epoch": 0.01936934192340131,
      "grad_norm": 1.5755513906478882,
      "learning_rate": 9.684453565931248e-07,
      "loss": 0.0379,
      "step": 3020
    },
    {
      "epoch": 0.019433478817187405,
      "grad_norm": 1.1730719804763794,
      "learning_rate": 9.716521292970755e-07,
      "loss": 0.0357,
      "step": 3030
    },
    {
      "epoch": 0.0194976157109735,
      "grad_norm": 1.3751786947250366,
      "learning_rate": 9.748589020010264e-07,
      "loss": 0.0379,
      "step": 3040
    },
    {
      "epoch": 0.019561752604759598,
      "grad_norm": 2.1120567321777344,
      "learning_rate": 9.78065674704977e-07,
      "loss": 0.0403,
      "step": 3050
    },
    {
      "epoch": 0.019625889498545695,
      "grad_norm": 1.4736151695251465,
      "learning_rate": 9.812724474089277e-07,
      "loss": 0.0334,
      "step": 3060
    },
    {
      "epoch": 0.01969002639233179,
      "grad_norm": 1.5400184392929077,
      "learning_rate": 9.844792201128784e-07,
      "loss": 0.0396,
      "step": 3070
    },
    {
      "epoch": 0.01975416328611789,
      "grad_norm": 1.8855392932891846,
      "learning_rate": 9.876859928168291e-07,
      "loss": 0.041,
      "step": 3080
    },
    {
      "epoch": 0.019818300179903988,
      "grad_norm": 1.3872394561767578,
      "learning_rate": 9.9089276552078e-07,
      "loss": 0.0366,
      "step": 3090
    },
    {
      "epoch": 0.019882437073690085,
      "grad_norm": 1.054284930229187,
      "learning_rate": 9.940995382247307e-07,
      "loss": 0.0335,
      "step": 3100
    },
    {
      "epoch": 0.01994657396747618,
      "grad_norm": 1.2524304389953613,
      "learning_rate": 9.973063109286816e-07,
      "loss": 0.038,
      "step": 3110
    },
    {
      "epoch": 0.020010710861262278,
      "grad_norm": 1.533583641052246,
      "learning_rate": 1.0005130836326323e-06,
      "loss": 0.0359,
      "step": 3120
    },
    {
      "epoch": 0.020074847755048374,
      "grad_norm": 2.047041177749634,
      "learning_rate": 1.003719856336583e-06,
      "loss": 0.0408,
      "step": 3130
    },
    {
      "epoch": 0.02013898464883447,
      "grad_norm": 1.6743892431259155,
      "learning_rate": 1.0069266290405337e-06,
      "loss": 0.0356,
      "step": 3140
    },
    {
      "epoch": 0.02020312154262057,
      "grad_norm": 2.2678470611572266,
      "learning_rate": 1.0101334017444844e-06,
      "loss": 0.0396,
      "step": 3150
    },
    {
      "epoch": 0.020267258436406668,
      "grad_norm": 1.4740620851516724,
      "learning_rate": 1.013340174448435e-06,
      "loss": 0.0378,
      "step": 3160
    },
    {
      "epoch": 0.020331395330192764,
      "grad_norm": 1.1527410745620728,
      "learning_rate": 1.016546947152386e-06,
      "loss": 0.0336,
      "step": 3170
    },
    {
      "epoch": 0.02039553222397886,
      "grad_norm": 2.1838512420654297,
      "learning_rate": 1.0197537198563367e-06,
      "loss": 0.0389,
      "step": 3180
    },
    {
      "epoch": 0.020459669117764957,
      "grad_norm": 1.4129743576049805,
      "learning_rate": 1.0229604925602873e-06,
      "loss": 0.0349,
      "step": 3190
    },
    {
      "epoch": 0.020523806011551054,
      "grad_norm": 2.70485782623291,
      "learning_rate": 1.0261672652642382e-06,
      "loss": 0.0437,
      "step": 3200
    },
    {
      "epoch": 0.02058794290533715,
      "grad_norm": 1.1545413732528687,
      "learning_rate": 1.029374037968189e-06,
      "loss": 0.0251,
      "step": 3210
    },
    {
      "epoch": 0.020652079799123247,
      "grad_norm": 1.4345098733901978,
      "learning_rate": 1.0325808106721396e-06,
      "loss": 0.0284,
      "step": 3220
    },
    {
      "epoch": 0.020716216692909347,
      "grad_norm": 1.5682227611541748,
      "learning_rate": 1.0357875833760903e-06,
      "loss": 0.0338,
      "step": 3230
    },
    {
      "epoch": 0.020780353586695444,
      "grad_norm": 1.8983999490737915,
      "learning_rate": 1.0389943560800412e-06,
      "loss": 0.0323,
      "step": 3240
    },
    {
      "epoch": 0.02084449048048154,
      "grad_norm": 1.4826823472976685,
      "learning_rate": 1.042201128783992e-06,
      "loss": 0.0341,
      "step": 3250
    },
    {
      "epoch": 0.020908627374267637,
      "grad_norm": 1.2763713598251343,
      "learning_rate": 1.0454079014879426e-06,
      "loss": 0.0297,
      "step": 3260
    },
    {
      "epoch": 0.020972764268053733,
      "grad_norm": 1.163649559020996,
      "learning_rate": 1.0486146741918933e-06,
      "loss": 0.0289,
      "step": 3270
    },
    {
      "epoch": 0.02103690116183983,
      "grad_norm": 1.3910245895385742,
      "learning_rate": 1.0518214468958442e-06,
      "loss": 0.0332,
      "step": 3280
    },
    {
      "epoch": 0.021101038055625927,
      "grad_norm": 1.4478743076324463,
      "learning_rate": 1.0550282195997949e-06,
      "loss": 0.0317,
      "step": 3290
    },
    {
      "epoch": 0.021165174949412027,
      "grad_norm": 1.6459060907363892,
      "learning_rate": 1.0582349923037456e-06,
      "loss": 0.0372,
      "step": 3300
    },
    {
      "epoch": 0.021229311843198123,
      "grad_norm": 1.8430840969085693,
      "learning_rate": 1.0614417650076963e-06,
      "loss": 0.0353,
      "step": 3310
    },
    {
      "epoch": 0.02129344873698422,
      "grad_norm": 1.2712984085083008,
      "learning_rate": 1.0646485377116472e-06,
      "loss": 0.031,
      "step": 3320
    },
    {
      "epoch": 0.021357585630770316,
      "grad_norm": 0.8528488278388977,
      "learning_rate": 1.0678553104155978e-06,
      "loss": 0.0284,
      "step": 3330
    },
    {
      "epoch": 0.021421722524556413,
      "grad_norm": 1.7615848779678345,
      "learning_rate": 1.0710620831195485e-06,
      "loss": 0.0329,
      "step": 3340
    },
    {
      "epoch": 0.02148585941834251,
      "grad_norm": 1.3932099342346191,
      "learning_rate": 1.0742688558234992e-06,
      "loss": 0.0307,
      "step": 3350
    },
    {
      "epoch": 0.021549996312128606,
      "grad_norm": 2.1969845294952393,
      "learning_rate": 1.07747562852745e-06,
      "loss": 0.0379,
      "step": 3360
    },
    {
      "epoch": 0.021614133205914706,
      "grad_norm": 1.2978613376617432,
      "learning_rate": 1.0806824012314008e-06,
      "loss": 0.033,
      "step": 3370
    },
    {
      "epoch": 0.021678270099700803,
      "grad_norm": 1.0141751766204834,
      "learning_rate": 1.0838891739353515e-06,
      "loss": 0.0255,
      "step": 3380
    },
    {
      "epoch": 0.0217424069934869,
      "grad_norm": 1.300687551498413,
      "learning_rate": 1.0870959466393024e-06,
      "loss": 0.0352,
      "step": 3390
    },
    {
      "epoch": 0.021806543887272996,
      "grad_norm": 1.0667314529418945,
      "learning_rate": 1.090302719343253e-06,
      "loss": 0.0275,
      "step": 3400
    },
    {
      "epoch": 0.021870680781059092,
      "grad_norm": 1.2791436910629272,
      "learning_rate": 1.0935094920472038e-06,
      "loss": 0.0253,
      "step": 3410
    },
    {
      "epoch": 0.02193481767484519,
      "grad_norm": 0.9602054953575134,
      "learning_rate": 1.0967162647511545e-06,
      "loss": 0.0305,
      "step": 3420
    },
    {
      "epoch": 0.021998954568631286,
      "grad_norm": 1.0475029945373535,
      "learning_rate": 1.0999230374551052e-06,
      "loss": 0.0272,
      "step": 3430
    },
    {
      "epoch": 0.022063091462417382,
      "grad_norm": 1.3703036308288574,
      "learning_rate": 1.103129810159056e-06,
      "loss": 0.0386,
      "step": 3440
    },
    {
      "epoch": 0.022127228356203482,
      "grad_norm": 2.4703800678253174,
      "learning_rate": 1.1063365828630068e-06,
      "loss": 0.0375,
      "step": 3450
    },
    {
      "epoch": 0.02219136524998958,
      "grad_norm": 1.3595285415649414,
      "learning_rate": 1.1095433555669577e-06,
      "loss": 0.0309,
      "step": 3460
    },
    {
      "epoch": 0.022255502143775675,
      "grad_norm": 1.59342360496521,
      "learning_rate": 1.1127501282709083e-06,
      "loss": 0.029,
      "step": 3470
    },
    {
      "epoch": 0.022319639037561772,
      "grad_norm": 1.3593720197677612,
      "learning_rate": 1.115956900974859e-06,
      "loss": 0.0272,
      "step": 3480
    },
    {
      "epoch": 0.02238377593134787,
      "grad_norm": 1.0497242212295532,
      "learning_rate": 1.1191636736788097e-06,
      "loss": 0.0307,
      "step": 3490
    },
    {
      "epoch": 0.022447912825133965,
      "grad_norm": 1.5057237148284912,
      "learning_rate": 1.1223704463827604e-06,
      "loss": 0.0308,
      "step": 3500
    },
    {
      "epoch": 0.02251204971892006,
      "grad_norm": 1.297191858291626,
      "learning_rate": 1.125577219086711e-06,
      "loss": 0.0308,
      "step": 3510
    },
    {
      "epoch": 0.02257618661270616,
      "grad_norm": 2.359473705291748,
      "learning_rate": 1.128783991790662e-06,
      "loss": 0.0302,
      "step": 3520
    },
    {
      "epoch": 0.022640323506492258,
      "grad_norm": 0.7025953531265259,
      "learning_rate": 1.1319907644946127e-06,
      "loss": 0.0259,
      "step": 3530
    },
    {
      "epoch": 0.022704460400278355,
      "grad_norm": 1.1531574726104736,
      "learning_rate": 1.1351975371985636e-06,
      "loss": 0.0314,
      "step": 3540
    },
    {
      "epoch": 0.02276859729406445,
      "grad_norm": 1.433619499206543,
      "learning_rate": 1.1384043099025143e-06,
      "loss": 0.0261,
      "step": 3550
    },
    {
      "epoch": 0.022832734187850548,
      "grad_norm": 1.3180760145187378,
      "learning_rate": 1.141611082606465e-06,
      "loss": 0.039,
      "step": 3560
    },
    {
      "epoch": 0.022896871081636645,
      "grad_norm": 1.2552984952926636,
      "learning_rate": 1.1448178553104157e-06,
      "loss": 0.0258,
      "step": 3570
    },
    {
      "epoch": 0.02296100797542274,
      "grad_norm": 1.1624325513839722,
      "learning_rate": 1.1480246280143664e-06,
      "loss": 0.0282,
      "step": 3580
    },
    {
      "epoch": 0.023025144869208838,
      "grad_norm": 1.4660444259643555,
      "learning_rate": 1.1512314007183173e-06,
      "loss": 0.0325,
      "step": 3590
    },
    {
      "epoch": 0.023089281762994938,
      "grad_norm": 1.164898157119751,
      "learning_rate": 1.154438173422268e-06,
      "loss": 0.0251,
      "step": 3600
    },
    {
      "epoch": 0.023153418656781034,
      "grad_norm": 1.5619678497314453,
      "learning_rate": 1.1576449461262186e-06,
      "loss": 0.0462,
      "step": 3610
    },
    {
      "epoch": 0.02321755555056713,
      "grad_norm": 1.7733670473098755,
      "learning_rate": 1.1608517188301695e-06,
      "loss": 0.0273,
      "step": 3620
    },
    {
      "epoch": 0.023281692444353227,
      "grad_norm": 0.8263606429100037,
      "learning_rate": 1.1640584915341202e-06,
      "loss": 0.0287,
      "step": 3630
    },
    {
      "epoch": 0.023345829338139324,
      "grad_norm": 1.2657470703125,
      "learning_rate": 1.167265264238071e-06,
      "loss": 0.0251,
      "step": 3640
    },
    {
      "epoch": 0.02340996623192542,
      "grad_norm": 1.2587162256240845,
      "learning_rate": 1.1704720369420216e-06,
      "loss": 0.0193,
      "step": 3650
    },
    {
      "epoch": 0.023474103125711517,
      "grad_norm": 1.990098237991333,
      "learning_rate": 1.1736788096459723e-06,
      "loss": 0.0254,
      "step": 3660
    },
    {
      "epoch": 0.023538240019497617,
      "grad_norm": 1.408094882965088,
      "learning_rate": 1.1768855823499232e-06,
      "loss": 0.0362,
      "step": 3670
    },
    {
      "epoch": 0.023602376913283714,
      "grad_norm": 1.239940881729126,
      "learning_rate": 1.1800923550538739e-06,
      "loss": 0.0329,
      "step": 3680
    },
    {
      "epoch": 0.02366651380706981,
      "grad_norm": 1.7103601694107056,
      "learning_rate": 1.1832991277578246e-06,
      "loss": 0.0265,
      "step": 3690
    },
    {
      "epoch": 0.023730650700855907,
      "grad_norm": 1.2067160606384277,
      "learning_rate": 1.1865059004617753e-06,
      "loss": 0.0315,
      "step": 3700
    },
    {
      "epoch": 0.023794787594642004,
      "grad_norm": 1.611244559288025,
      "learning_rate": 1.1897126731657262e-06,
      "loss": 0.0269,
      "step": 3710
    },
    {
      "epoch": 0.0238589244884281,
      "grad_norm": 1.1752269268035889,
      "learning_rate": 1.1929194458696769e-06,
      "loss": 0.031,
      "step": 3720
    },
    {
      "epoch": 0.023923061382214197,
      "grad_norm": 1.7413352727890015,
      "learning_rate": 1.1961262185736275e-06,
      "loss": 0.0338,
      "step": 3730
    },
    {
      "epoch": 0.023987198276000293,
      "grad_norm": 2.1114962100982666,
      "learning_rate": 1.1993329912775784e-06,
      "loss": 0.0289,
      "step": 3740
    },
    {
      "epoch": 0.024051335169786393,
      "grad_norm": 1.0259640216827393,
      "learning_rate": 1.2025397639815291e-06,
      "loss": 0.0278,
      "step": 3750
    },
    {
      "epoch": 0.02411547206357249,
      "grad_norm": 1.528367519378662,
      "learning_rate": 1.2057465366854798e-06,
      "loss": 0.0253,
      "step": 3760
    },
    {
      "epoch": 0.024179608957358587,
      "grad_norm": 1.4296187162399292,
      "learning_rate": 1.2089533093894305e-06,
      "loss": 0.034,
      "step": 3770
    },
    {
      "epoch": 0.024243745851144683,
      "grad_norm": 1.101138710975647,
      "learning_rate": 1.2121600820933812e-06,
      "loss": 0.0231,
      "step": 3780
    },
    {
      "epoch": 0.02430788274493078,
      "grad_norm": 1.2148758172988892,
      "learning_rate": 1.215366854797332e-06,
      "loss": 0.0276,
      "step": 3790
    },
    {
      "epoch": 0.024372019638716876,
      "grad_norm": 1.8424838781356812,
      "learning_rate": 1.2185736275012828e-06,
      "loss": 0.0307,
      "step": 3800
    },
    {
      "epoch": 0.024436156532502973,
      "grad_norm": 1.2761074304580688,
      "learning_rate": 1.2217804002052335e-06,
      "loss": 0.0237,
      "step": 3810
    },
    {
      "epoch": 0.024500293426289073,
      "grad_norm": 1.4775094985961914,
      "learning_rate": 1.2249871729091844e-06,
      "loss": 0.0278,
      "step": 3820
    },
    {
      "epoch": 0.02456443032007517,
      "grad_norm": 1.41787850856781,
      "learning_rate": 1.228193945613135e-06,
      "loss": 0.0287,
      "step": 3830
    },
    {
      "epoch": 0.024628567213861266,
      "grad_norm": 1.7441461086273193,
      "learning_rate": 1.2314007183170858e-06,
      "loss": 0.0253,
      "step": 3840
    },
    {
      "epoch": 0.024692704107647363,
      "grad_norm": 1.940899133682251,
      "learning_rate": 1.2346074910210365e-06,
      "loss": 0.0321,
      "step": 3850
    },
    {
      "epoch": 0.02475684100143346,
      "grad_norm": 1.5329341888427734,
      "learning_rate": 1.2378142637249871e-06,
      "loss": 0.0305,
      "step": 3860
    },
    {
      "epoch": 0.024820977895219556,
      "grad_norm": 1.282849907875061,
      "learning_rate": 1.241021036428938e-06,
      "loss": 0.0242,
      "step": 3870
    },
    {
      "epoch": 0.024885114789005652,
      "grad_norm": 1.5139777660369873,
      "learning_rate": 1.2442278091328887e-06,
      "loss": 0.0276,
      "step": 3880
    },
    {
      "epoch": 0.024949251682791752,
      "grad_norm": 1.3489099740982056,
      "learning_rate": 1.2474345818368396e-06,
      "loss": 0.0414,
      "step": 3890
    },
    {
      "epoch": 0.02501338857657785,
      "grad_norm": 1.1435086727142334,
      "learning_rate": 1.2506413545407903e-06,
      "loss": 0.0242,
      "step": 3900
    },
    {
      "epoch": 0.025077525470363946,
      "grad_norm": 1.0997439622879028,
      "learning_rate": 1.2538481272447408e-06,
      "loss": 0.0245,
      "step": 3910
    },
    {
      "epoch": 0.025141662364150042,
      "grad_norm": 0.8081037998199463,
      "learning_rate": 1.2570548999486917e-06,
      "loss": 0.0241,
      "step": 3920
    },
    {
      "epoch": 0.02520579925793614,
      "grad_norm": 0.8284711241722107,
      "learning_rate": 1.2602616726526426e-06,
      "loss": 0.0339,
      "step": 3930
    },
    {
      "epoch": 0.025269936151722235,
      "grad_norm": 1.4088741540908813,
      "learning_rate": 1.263468445356593e-06,
      "loss": 0.0267,
      "step": 3940
    },
    {
      "epoch": 0.025334073045508332,
      "grad_norm": 1.1950079202651978,
      "learning_rate": 1.266675218060544e-06,
      "loss": 0.0279,
      "step": 3950
    },
    {
      "epoch": 0.02539820993929443,
      "grad_norm": 1.3671220541000366,
      "learning_rate": 1.2698819907644947e-06,
      "loss": 0.0245,
      "step": 3960
    },
    {
      "epoch": 0.02546234683308053,
      "grad_norm": 2.05141282081604,
      "learning_rate": 1.2730887634684456e-06,
      "loss": 0.0278,
      "step": 3970
    },
    {
      "epoch": 0.025526483726866625,
      "grad_norm": 1.7943063974380493,
      "learning_rate": 1.276295536172396e-06,
      "loss": 0.0254,
      "step": 3980
    },
    {
      "epoch": 0.02559062062065272,
      "grad_norm": 1.4211649894714355,
      "learning_rate": 1.279502308876347e-06,
      "loss": 0.0219,
      "step": 3990
    },
    {
      "epoch": 0.025654757514438818,
      "grad_norm": 1.2778598070144653,
      "learning_rate": 1.2827090815802976e-06,
      "loss": 0.0274,
      "step": 4000
    },
    {
      "epoch": 0.025718894408224915,
      "grad_norm": 1.6716960668563843,
      "learning_rate": 1.2859158542842483e-06,
      "loss": 0.0225,
      "step": 4010
    },
    {
      "epoch": 0.02578303130201101,
      "grad_norm": 1.0186187028884888,
      "learning_rate": 1.2891226269881992e-06,
      "loss": 0.0251,
      "step": 4020
    },
    {
      "epoch": 0.025847168195797108,
      "grad_norm": 1.4226250648498535,
      "learning_rate": 1.29232939969215e-06,
      "loss": 0.0341,
      "step": 4030
    },
    {
      "epoch": 0.025911305089583208,
      "grad_norm": 1.6474844217300415,
      "learning_rate": 1.2955361723961008e-06,
      "loss": 0.0277,
      "step": 4040
    },
    {
      "epoch": 0.025975441983369305,
      "grad_norm": 1.2548375129699707,
      "learning_rate": 1.2987429451000513e-06,
      "loss": 0.0319,
      "step": 4050
    },
    {
      "epoch": 0.0260395788771554,
      "grad_norm": 1.264318585395813,
      "learning_rate": 1.3019497178040022e-06,
      "loss": 0.0255,
      "step": 4060
    },
    {
      "epoch": 0.026103715770941498,
      "grad_norm": 2.2901430130004883,
      "learning_rate": 1.305156490507953e-06,
      "loss": 0.0341,
      "step": 4070
    },
    {
      "epoch": 0.026167852664727594,
      "grad_norm": 1.6619858741760254,
      "learning_rate": 1.3083632632119036e-06,
      "loss": 0.0309,
      "step": 4080
    },
    {
      "epoch": 0.02623198955851369,
      "grad_norm": 0.9754308462142944,
      "learning_rate": 1.3115700359158545e-06,
      "loss": 0.0285,
      "step": 4090
    },
    {
      "epoch": 0.026296126452299787,
      "grad_norm": 1.2731003761291504,
      "learning_rate": 1.3147768086198052e-06,
      "loss": 0.0277,
      "step": 4100
    },
    {
      "epoch": 0.026360263346085884,
      "grad_norm": 1.1012314558029175,
      "learning_rate": 1.317983581323756e-06,
      "loss": 0.0286,
      "step": 4110
    },
    {
      "epoch": 0.026424400239871984,
      "grad_norm": 1.2308241128921509,
      "learning_rate": 1.3211903540277066e-06,
      "loss": 0.0292,
      "step": 4120
    },
    {
      "epoch": 0.02648853713365808,
      "grad_norm": 1.3766791820526123,
      "learning_rate": 1.3243971267316575e-06,
      "loss": 0.0317,
      "step": 4130
    },
    {
      "epoch": 0.026552674027444177,
      "grad_norm": 1.2170155048370361,
      "learning_rate": 1.327603899435608e-06,
      "loss": 0.0266,
      "step": 4140
    },
    {
      "epoch": 0.026616810921230274,
      "grad_norm": 1.0523362159729004,
      "learning_rate": 1.3308106721395588e-06,
      "loss": 0.033,
      "step": 4150
    },
    {
      "epoch": 0.02668094781501637,
      "grad_norm": 1.247113585472107,
      "learning_rate": 1.3340174448435095e-06,
      "loss": 0.0362,
      "step": 4160
    },
    {
      "epoch": 0.026745084708802467,
      "grad_norm": 1.7340353727340698,
      "learning_rate": 1.3372242175474604e-06,
      "loss": 0.029,
      "step": 4170
    },
    {
      "epoch": 0.026809221602588564,
      "grad_norm": 1.5098899602890015,
      "learning_rate": 1.3404309902514111e-06,
      "loss": 0.0365,
      "step": 4180
    },
    {
      "epoch": 0.026873358496374664,
      "grad_norm": 1.6181445121765137,
      "learning_rate": 1.3436377629553618e-06,
      "loss": 0.029,
      "step": 4190
    },
    {
      "epoch": 0.02693749539016076,
      "grad_norm": 0.9857799410820007,
      "learning_rate": 1.3468445356593127e-06,
      "loss": 0.0231,
      "step": 4200
    },
    {
      "epoch": 0.027001632283946857,
      "grad_norm": 1.0968294143676758,
      "learning_rate": 1.3500513083632632e-06,
      "loss": 0.0285,
      "step": 4210
    },
    {
      "epoch": 0.027065769177732953,
      "grad_norm": 0.9057127833366394,
      "learning_rate": 1.353258081067214e-06,
      "loss": 0.0182,
      "step": 4220
    },
    {
      "epoch": 0.02712990607151905,
      "grad_norm": 1.425401210784912,
      "learning_rate": 1.3564648537711648e-06,
      "loss": 0.031,
      "step": 4230
    },
    {
      "epoch": 0.027194042965305146,
      "grad_norm": 1.5106675624847412,
      "learning_rate": 1.3596716264751157e-06,
      "loss": 0.0348,
      "step": 4240
    },
    {
      "epoch": 0.027258179859091243,
      "grad_norm": 1.6456857919692993,
      "learning_rate": 1.3628783991790662e-06,
      "loss": 0.0271,
      "step": 4250
    },
    {
      "epoch": 0.027322316752877343,
      "grad_norm": 1.3953471183776855,
      "learning_rate": 1.366085171883017e-06,
      "loss": 0.0261,
      "step": 4260
    },
    {
      "epoch": 0.02738645364666344,
      "grad_norm": 1.5983757972717285,
      "learning_rate": 1.369291944586968e-06,
      "loss": 0.025,
      "step": 4270
    },
    {
      "epoch": 0.027450590540449536,
      "grad_norm": 1.4413799047470093,
      "learning_rate": 1.3724987172909184e-06,
      "loss": 0.0254,
      "step": 4280
    },
    {
      "epoch": 0.027514727434235633,
      "grad_norm": 1.8395564556121826,
      "learning_rate": 1.3757054899948693e-06,
      "loss": 0.0265,
      "step": 4290
    },
    {
      "epoch": 0.02757886432802173,
      "grad_norm": 3.691532850265503,
      "learning_rate": 1.37891226269882e-06,
      "loss": 0.0357,
      "step": 4300
    },
    {
      "epoch": 0.027643001221807826,
      "grad_norm": 2.7965006828308105,
      "learning_rate": 1.3821190354027707e-06,
      "loss": 0.0271,
      "step": 4310
    },
    {
      "epoch": 0.027707138115593923,
      "grad_norm": 0.9594042301177979,
      "learning_rate": 1.3853258081067214e-06,
      "loss": 0.0251,
      "step": 4320
    },
    {
      "epoch": 0.02777127500938002,
      "grad_norm": 0.89903724193573,
      "learning_rate": 1.3885325808106723e-06,
      "loss": 0.0211,
      "step": 4330
    },
    {
      "epoch": 0.02783541190316612,
      "grad_norm": 1.770849347114563,
      "learning_rate": 1.3917393535146228e-06,
      "loss": 0.0229,
      "step": 4340
    },
    {
      "epoch": 0.027899548796952216,
      "grad_norm": 1.7908738851547241,
      "learning_rate": 1.3949461262185737e-06,
      "loss": 0.0293,
      "step": 4350
    },
    {
      "epoch": 0.027963685690738312,
      "grad_norm": 1.2615981101989746,
      "learning_rate": 1.3981528989225246e-06,
      "loss": 0.0345,
      "step": 4360
    },
    {
      "epoch": 0.02802782258452441,
      "grad_norm": 1.5795953273773193,
      "learning_rate": 1.4013596716264753e-06,
      "loss": 0.0277,
      "step": 4370
    },
    {
      "epoch": 0.028091959478310505,
      "grad_norm": 0.9272065758705139,
      "learning_rate": 1.404566444330426e-06,
      "loss": 0.024,
      "step": 4380
    },
    {
      "epoch": 0.028156096372096602,
      "grad_norm": 1.234574794769287,
      "learning_rate": 1.4077732170343767e-06,
      "loss": 0.0179,
      "step": 4390
    },
    {
      "epoch": 0.0282202332658827,
      "grad_norm": 0.859353244304657,
      "learning_rate": 1.4109799897383276e-06,
      "loss": 0.0282,
      "step": 4400
    },
    {
      "epoch": 0.0282843701596688,
      "grad_norm": 1.1903260946273804,
      "learning_rate": 1.414186762442278e-06,
      "loss": 0.0219,
      "step": 4410
    },
    {
      "epoch": 0.028348507053454895,
      "grad_norm": 1.5210109949111938,
      "learning_rate": 1.417393535146229e-06,
      "loss": 0.0208,
      "step": 4420
    },
    {
      "epoch": 0.028412643947240992,
      "grad_norm": 1.2154312133789062,
      "learning_rate": 1.4206003078501798e-06,
      "loss": 0.0203,
      "step": 4430
    },
    {
      "epoch": 0.02847678084102709,
      "grad_norm": 1.0572129487991333,
      "learning_rate": 1.4238070805541303e-06,
      "loss": 0.0273,
      "step": 4440
    },
    {
      "epoch": 0.028540917734813185,
      "grad_norm": 2.9614217281341553,
      "learning_rate": 1.4270138532580812e-06,
      "loss": 0.0287,
      "step": 4450
    },
    {
      "epoch": 0.02860505462859928,
      "grad_norm": 0.9641210436820984,
      "learning_rate": 1.430220625962032e-06,
      "loss": 0.0226,
      "step": 4460
    },
    {
      "epoch": 0.028669191522385378,
      "grad_norm": 1.0650194883346558,
      "learning_rate": 1.4334273986659828e-06,
      "loss": 0.0249,
      "step": 4470
    },
    {
      "epoch": 0.028733328416171475,
      "grad_norm": 1.7810701131820679,
      "learning_rate": 1.4366341713699333e-06,
      "loss": 0.0241,
      "step": 4480
    },
    {
      "epoch": 0.028797465309957575,
      "grad_norm": 1.532281517982483,
      "learning_rate": 1.4398409440738842e-06,
      "loss": 0.0277,
      "step": 4490
    },
    {
      "epoch": 0.02886160220374367,
      "grad_norm": 1.1858774423599243,
      "learning_rate": 1.4430477167778349e-06,
      "loss": 0.0276,
      "step": 4500
    },
    {
      "epoch": 0.028925739097529768,
      "grad_norm": 1.3994327783584595,
      "learning_rate": 1.4462544894817856e-06,
      "loss": 0.0299,
      "step": 4510
    },
    {
      "epoch": 0.028989875991315864,
      "grad_norm": 1.0907750129699707,
      "learning_rate": 1.4494612621857365e-06,
      "loss": 0.023,
      "step": 4520
    },
    {
      "epoch": 0.02905401288510196,
      "grad_norm": 0.8516006469726562,
      "learning_rate": 1.4526680348896872e-06,
      "loss": 0.0283,
      "step": 4530
    },
    {
      "epoch": 0.029118149778888058,
      "grad_norm": 1.7701914310455322,
      "learning_rate": 1.455874807593638e-06,
      "loss": 0.0251,
      "step": 4540
    },
    {
      "epoch": 0.029182286672674154,
      "grad_norm": 2.0644941329956055,
      "learning_rate": 1.4590815802975885e-06,
      "loss": 0.0264,
      "step": 4550
    },
    {
      "epoch": 0.029246423566460254,
      "grad_norm": 1.278272271156311,
      "learning_rate": 1.4622883530015394e-06,
      "loss": 0.0317,
      "step": 4560
    },
    {
      "epoch": 0.02931056046024635,
      "grad_norm": 1.3904606103897095,
      "learning_rate": 1.46549512570549e-06,
      "loss": 0.0269,
      "step": 4570
    },
    {
      "epoch": 0.029374697354032447,
      "grad_norm": 1.9059582948684692,
      "learning_rate": 1.4687018984094408e-06,
      "loss": 0.0311,
      "step": 4580
    },
    {
      "epoch": 0.029438834247818544,
      "grad_norm": 1.3313368558883667,
      "learning_rate": 1.4719086711133915e-06,
      "loss": 0.0305,
      "step": 4590
    },
    {
      "epoch": 0.02950297114160464,
      "grad_norm": 1.8630759716033936,
      "learning_rate": 1.4751154438173424e-06,
      "loss": 0.0232,
      "step": 4600
    },
    {
      "epoch": 0.029567108035390737,
      "grad_norm": 1.6356956958770752,
      "learning_rate": 1.4783222165212933e-06,
      "loss": 0.027,
      "step": 4610
    },
    {
      "epoch": 0.029631244929176834,
      "grad_norm": 0.6639159917831421,
      "learning_rate": 1.4815289892252438e-06,
      "loss": 0.0244,
      "step": 4620
    },
    {
      "epoch": 0.029695381822962934,
      "grad_norm": 1.786681890487671,
      "learning_rate": 1.4847357619291947e-06,
      "loss": 0.0266,
      "step": 4630
    },
    {
      "epoch": 0.02975951871674903,
      "grad_norm": 1.597255825996399,
      "learning_rate": 1.4879425346331452e-06,
      "loss": 0.0262,
      "step": 4640
    },
    {
      "epoch": 0.029823655610535127,
      "grad_norm": 1.2564306259155273,
      "learning_rate": 1.491149307337096e-06,
      "loss": 0.0245,
      "step": 4650
    },
    {
      "epoch": 0.029887792504321224,
      "grad_norm": 1.2096034288406372,
      "learning_rate": 1.4943560800410468e-06,
      "loss": 0.0245,
      "step": 4660
    },
    {
      "epoch": 0.02995192939810732,
      "grad_norm": 1.2752938270568848,
      "learning_rate": 1.4975628527449977e-06,
      "loss": 0.0198,
      "step": 4670
    },
    {
      "epoch": 0.030016066291893417,
      "grad_norm": 1.5911647081375122,
      "learning_rate": 1.5007696254489481e-06,
      "loss": 0.0248,
      "step": 4680
    },
    {
      "epoch": 0.030080203185679513,
      "grad_norm": 2.105846643447876,
      "learning_rate": 1.503976398152899e-06,
      "loss": 0.0173,
      "step": 4690
    },
    {
      "epoch": 0.03014434007946561,
      "grad_norm": 0.8787049055099487,
      "learning_rate": 1.50718317085685e-06,
      "loss": 0.0312,
      "step": 4700
    },
    {
      "epoch": 0.03020847697325171,
      "grad_norm": 1.3518826961517334,
      "learning_rate": 1.5103899435608004e-06,
      "loss": 0.0251,
      "step": 4710
    },
    {
      "epoch": 0.030272613867037806,
      "grad_norm": 0.5953378677368164,
      "learning_rate": 1.5135967162647513e-06,
      "loss": 0.0268,
      "step": 4720
    },
    {
      "epoch": 0.030336750760823903,
      "grad_norm": 0.946031928062439,
      "learning_rate": 1.516803488968702e-06,
      "loss": 0.0207,
      "step": 4730
    },
    {
      "epoch": 0.03040088765461,
      "grad_norm": 1.2910767793655396,
      "learning_rate": 1.520010261672653e-06,
      "loss": 0.0288,
      "step": 4740
    },
    {
      "epoch": 0.030465024548396096,
      "grad_norm": 0.6571576595306396,
      "learning_rate": 1.5232170343766034e-06,
      "loss": 0.017,
      "step": 4750
    },
    {
      "epoch": 0.030529161442182193,
      "grad_norm": 0.7259587645530701,
      "learning_rate": 1.5264238070805543e-06,
      "loss": 0.0227,
      "step": 4760
    },
    {
      "epoch": 0.03059329833596829,
      "grad_norm": 1.0711289644241333,
      "learning_rate": 1.5296305797845052e-06,
      "loss": 0.0287,
      "step": 4770
    },
    {
      "epoch": 0.03065743522975439,
      "grad_norm": 1.5687236785888672,
      "learning_rate": 1.5328373524884557e-06,
      "loss": 0.0354,
      "step": 4780
    },
    {
      "epoch": 0.030721572123540486,
      "grad_norm": 1.2231394052505493,
      "learning_rate": 1.5360441251924066e-06,
      "loss": 0.0215,
      "step": 4790
    },
    {
      "epoch": 0.030785709017326583,
      "grad_norm": 1.2256968021392822,
      "learning_rate": 1.5392508978963573e-06,
      "loss": 0.0213,
      "step": 4800
    },
    {
      "epoch": 0.03084984591111268,
      "grad_norm": 1.2588187456130981,
      "learning_rate": 1.542457670600308e-06,
      "loss": 0.0177,
      "step": 4810
    },
    {
      "epoch": 0.030913982804898776,
      "grad_norm": 1.4336833953857422,
      "learning_rate": 1.5456644433042586e-06,
      "loss": 0.0205,
      "step": 4820
    },
    {
      "epoch": 0.030978119698684872,
      "grad_norm": 0.8296056389808655,
      "learning_rate": 1.5488712160082095e-06,
      "loss": 0.0283,
      "step": 4830
    },
    {
      "epoch": 0.03104225659247097,
      "grad_norm": 1.141822099685669,
      "learning_rate": 1.55207798871216e-06,
      "loss": 0.0304,
      "step": 4840
    },
    {
      "epoch": 0.031106393486257065,
      "grad_norm": 0.9502159357070923,
      "learning_rate": 1.555284761416111e-06,
      "loss": 0.0209,
      "step": 4850
    },
    {
      "epoch": 0.031170530380043165,
      "grad_norm": 1.7776676416397095,
      "learning_rate": 1.5584915341200618e-06,
      "loss": 0.0323,
      "step": 4860
    },
    {
      "epoch": 0.031234667273829262,
      "grad_norm": 0.9667540788650513,
      "learning_rate": 1.5616983068240125e-06,
      "loss": 0.018,
      "step": 4870
    },
    {
      "epoch": 0.03129880416761536,
      "grad_norm": 1.545588731765747,
      "learning_rate": 1.5649050795279632e-06,
      "loss": 0.0201,
      "step": 4880
    },
    {
      "epoch": 0.03136294106140146,
      "grad_norm": 0.8739742636680603,
      "learning_rate": 1.5681118522319139e-06,
      "loss": 0.0236,
      "step": 4890
    },
    {
      "epoch": 0.03142707795518755,
      "grad_norm": 1.9178887605667114,
      "learning_rate": 1.5713186249358648e-06,
      "loss": 0.0263,
      "step": 4900
    },
    {
      "epoch": 0.03149121484897365,
      "grad_norm": 1.7376662492752075,
      "learning_rate": 1.5745253976398153e-06,
      "loss": 0.0173,
      "step": 4910
    },
    {
      "epoch": 0.031555351742759745,
      "grad_norm": 1.293989658355713,
      "learning_rate": 1.5777321703437662e-06,
      "loss": 0.0283,
      "step": 4920
    },
    {
      "epoch": 0.031619488636545845,
      "grad_norm": 1.388981819152832,
      "learning_rate": 1.5809389430477169e-06,
      "loss": 0.0188,
      "step": 4930
    },
    {
      "epoch": 0.03168362553033194,
      "grad_norm": 1.5035390853881836,
      "learning_rate": 1.5841457157516675e-06,
      "loss": 0.0219,
      "step": 4940
    },
    {
      "epoch": 0.03174776242411804,
      "grad_norm": 1.5644978284835815,
      "learning_rate": 1.5873524884556184e-06,
      "loss": 0.0296,
      "step": 4950
    },
    {
      "epoch": 0.03181189931790413,
      "grad_norm": 0.9671863317489624,
      "learning_rate": 1.5905592611595691e-06,
      "loss": 0.0357,
      "step": 4960
    },
    {
      "epoch": 0.03187603621169023,
      "grad_norm": 1.728955864906311,
      "learning_rate": 1.59376603386352e-06,
      "loss": 0.0257,
      "step": 4970
    },
    {
      "epoch": 0.03194017310547633,
      "grad_norm": 0.9994457960128784,
      "learning_rate": 1.5969728065674705e-06,
      "loss": 0.0256,
      "step": 4980
    },
    {
      "epoch": 0.032004309999262424,
      "grad_norm": 1.5000872611999512,
      "learning_rate": 1.6001795792714214e-06,
      "loss": 0.0213,
      "step": 4990
    },
    {
      "epoch": 0.032068446893048524,
      "grad_norm": 1.2088123559951782,
      "learning_rate": 1.6033863519753721e-06,
      "loss": 0.0179,
      "step": 5000
    },
    {
      "epoch": 0.03213258378683462,
      "grad_norm": 1.0315769910812378,
      "learning_rate": 1.6065931246793228e-06,
      "loss": 0.0192,
      "step": 5010
    },
    {
      "epoch": 0.03219672068062072,
      "grad_norm": 1.391422986984253,
      "learning_rate": 1.6097998973832735e-06,
      "loss": 0.0285,
      "step": 5020
    },
    {
      "epoch": 0.03226085757440681,
      "grad_norm": 1.2687987089157104,
      "learning_rate": 1.6130066700872244e-06,
      "loss": 0.0255,
      "step": 5030
    },
    {
      "epoch": 0.03232499446819291,
      "grad_norm": 0.7762933373451233,
      "learning_rate": 1.6162134427911753e-06,
      "loss": 0.0217,
      "step": 5040
    },
    {
      "epoch": 0.03238913136197901,
      "grad_norm": 1.1362630128860474,
      "learning_rate": 1.6194202154951258e-06,
      "loss": 0.0219,
      "step": 5050
    },
    {
      "epoch": 0.032453268255765104,
      "grad_norm": 0.9996119141578674,
      "learning_rate": 1.6226269881990767e-06,
      "loss": 0.0227,
      "step": 5060
    },
    {
      "epoch": 0.032517405149551204,
      "grad_norm": 1.7603073120117188,
      "learning_rate": 1.6258337609030271e-06,
      "loss": 0.0244,
      "step": 5070
    },
    {
      "epoch": 0.0325815420433373,
      "grad_norm": 1.5313562154769897,
      "learning_rate": 1.629040533606978e-06,
      "loss": 0.0316,
      "step": 5080
    },
    {
      "epoch": 0.0326456789371234,
      "grad_norm": 1.6863077878952026,
      "learning_rate": 1.6322473063109287e-06,
      "loss": 0.0243,
      "step": 5090
    },
    {
      "epoch": 0.03270981583090949,
      "grad_norm": 0.6779791116714478,
      "learning_rate": 1.6354540790148796e-06,
      "loss": 0.0169,
      "step": 5100
    },
    {
      "epoch": 0.03277395272469559,
      "grad_norm": 1.0107567310333252,
      "learning_rate": 1.6386608517188305e-06,
      "loss": 0.0222,
      "step": 5110
    },
    {
      "epoch": 0.03283808961848169,
      "grad_norm": 1.782139778137207,
      "learning_rate": 1.641867624422781e-06,
      "loss": 0.0238,
      "step": 5120
    },
    {
      "epoch": 0.03290222651226778,
      "grad_norm": 0.9885055422782898,
      "learning_rate": 1.645074397126732e-06,
      "loss": 0.0219,
      "step": 5130
    },
    {
      "epoch": 0.032966363406053883,
      "grad_norm": 1.1917113065719604,
      "learning_rate": 1.6482811698306824e-06,
      "loss": 0.0196,
      "step": 5140
    },
    {
      "epoch": 0.03303050029983998,
      "grad_norm": 1.1517614126205444,
      "learning_rate": 1.6514879425346333e-06,
      "loss": 0.0287,
      "step": 5150
    },
    {
      "epoch": 0.03309463719362608,
      "grad_norm": 1.6611928939819336,
      "learning_rate": 1.654694715238584e-06,
      "loss": 0.0179,
      "step": 5160
    },
    {
      "epoch": 0.03315877408741217,
      "grad_norm": 0.7962343096733093,
      "learning_rate": 1.6579014879425349e-06,
      "loss": 0.0241,
      "step": 5170
    },
    {
      "epoch": 0.03322291098119827,
      "grad_norm": 1.4753403663635254,
      "learning_rate": 1.6611082606464854e-06,
      "loss": 0.0274,
      "step": 5180
    },
    {
      "epoch": 0.03328704787498437,
      "grad_norm": 1.1891019344329834,
      "learning_rate": 1.6643150333504363e-06,
      "loss": 0.0172,
      "step": 5190
    },
    {
      "epoch": 0.03335118476877046,
      "grad_norm": 1.0140347480773926,
      "learning_rate": 1.6675218060543872e-06,
      "loss": 0.0236,
      "step": 5200
    },
    {
      "epoch": 0.03341532166255656,
      "grad_norm": 1.0874922275543213,
      "learning_rate": 1.6707285787583376e-06,
      "loss": 0.0189,
      "step": 5210
    },
    {
      "epoch": 0.033479458556342656,
      "grad_norm": 1.6356042623519897,
      "learning_rate": 1.6739353514622885e-06,
      "loss": 0.0189,
      "step": 5220
    },
    {
      "epoch": 0.033543595450128756,
      "grad_norm": 1.688215732574463,
      "learning_rate": 1.6771421241662392e-06,
      "loss": 0.025,
      "step": 5230
    },
    {
      "epoch": 0.03360773234391485,
      "grad_norm": 1.1185798645019531,
      "learning_rate": 1.6803488968701901e-06,
      "loss": 0.0189,
      "step": 5240
    },
    {
      "epoch": 0.03367186923770095,
      "grad_norm": 1.1338945627212524,
      "learning_rate": 1.6835556695741406e-06,
      "loss": 0.0224,
      "step": 5250
    },
    {
      "epoch": 0.03373600613148705,
      "grad_norm": 1.0788078308105469,
      "learning_rate": 1.6867624422780915e-06,
      "loss": 0.023,
      "step": 5260
    },
    {
      "epoch": 0.03380014302527314,
      "grad_norm": 1.0811059474945068,
      "learning_rate": 1.689969214982042e-06,
      "loss": 0.0236,
      "step": 5270
    },
    {
      "epoch": 0.03386427991905924,
      "grad_norm": 0.7724621891975403,
      "learning_rate": 1.693175987685993e-06,
      "loss": 0.0196,
      "step": 5280
    },
    {
      "epoch": 0.033928416812845336,
      "grad_norm": 0.8252857327461243,
      "learning_rate": 1.6963827603899438e-06,
      "loss": 0.0274,
      "step": 5290
    },
    {
      "epoch": 0.033992553706631436,
      "grad_norm": 1.3175960779190063,
      "learning_rate": 1.6995895330938945e-06,
      "loss": 0.0184,
      "step": 5300
    },
    {
      "epoch": 0.03405669060041753,
      "grad_norm": 1.2741248607635498,
      "learning_rate": 1.7027963057978452e-06,
      "loss": 0.0275,
      "step": 5310
    },
    {
      "epoch": 0.03412082749420363,
      "grad_norm": 0.8108416795730591,
      "learning_rate": 1.7060030785017959e-06,
      "loss": 0.0129,
      "step": 5320
    },
    {
      "epoch": 0.03418496438798972,
      "grad_norm": 1.051501989364624,
      "learning_rate": 1.7092098512057468e-06,
      "loss": 0.0199,
      "step": 5330
    },
    {
      "epoch": 0.03424910128177582,
      "grad_norm": 1.7660843133926392,
      "learning_rate": 1.7124166239096972e-06,
      "loss": 0.0239,
      "step": 5340
    },
    {
      "epoch": 0.03431323817556192,
      "grad_norm": 1.312746524810791,
      "learning_rate": 1.7156233966136481e-06,
      "loss": 0.0257,
      "step": 5350
    },
    {
      "epoch": 0.034377375069348015,
      "grad_norm": 1.3452777862548828,
      "learning_rate": 1.7188301693175988e-06,
      "loss": 0.021,
      "step": 5360
    },
    {
      "epoch": 0.034441511963134115,
      "grad_norm": 0.9213612079620361,
      "learning_rate": 1.7220369420215497e-06,
      "loss": 0.0184,
      "step": 5370
    },
    {
      "epoch": 0.03450564885692021,
      "grad_norm": 1.207118272781372,
      "learning_rate": 1.7252437147255004e-06,
      "loss": 0.0214,
      "step": 5380
    },
    {
      "epoch": 0.03456978575070631,
      "grad_norm": 0.9979588389396667,
      "learning_rate": 1.7284504874294511e-06,
      "loss": 0.0211,
      "step": 5390
    },
    {
      "epoch": 0.0346339226444924,
      "grad_norm": 1.328615665435791,
      "learning_rate": 1.731657260133402e-06,
      "loss": 0.0302,
      "step": 5400
    },
    {
      "epoch": 0.0346980595382785,
      "grad_norm": 1.2631783485412598,
      "learning_rate": 1.7348640328373525e-06,
      "loss": 0.0198,
      "step": 5410
    },
    {
      "epoch": 0.0347621964320646,
      "grad_norm": 1.2121427059173584,
      "learning_rate": 1.7380708055413034e-06,
      "loss": 0.028,
      "step": 5420
    },
    {
      "epoch": 0.034826333325850695,
      "grad_norm": 1.0202001333236694,
      "learning_rate": 1.741277578245254e-06,
      "loss": 0.0274,
      "step": 5430
    },
    {
      "epoch": 0.034890470219636795,
      "grad_norm": 1.1604700088500977,
      "learning_rate": 1.7444843509492048e-06,
      "loss": 0.0269,
      "step": 5440
    },
    {
      "epoch": 0.03495460711342289,
      "grad_norm": 0.8548974990844727,
      "learning_rate": 1.7476911236531557e-06,
      "loss": 0.0269,
      "step": 5450
    },
    {
      "epoch": 0.03501874400720899,
      "grad_norm": 3.13950252532959,
      "learning_rate": 1.7508978963571064e-06,
      "loss": 0.016,
      "step": 5460
    },
    {
      "epoch": 0.03508288090099508,
      "grad_norm": 1.0822423696517944,
      "learning_rate": 1.7541046690610573e-06,
      "loss": 0.0175,
      "step": 5470
    },
    {
      "epoch": 0.03514701779478118,
      "grad_norm": 1.2122656106948853,
      "learning_rate": 1.7573114417650077e-06,
      "loss": 0.0189,
      "step": 5480
    },
    {
      "epoch": 0.03521115468856728,
      "grad_norm": 1.1777442693710327,
      "learning_rate": 1.7605182144689587e-06,
      "loss": 0.018,
      "step": 5490
    },
    {
      "epoch": 0.035275291582353374,
      "grad_norm": 1.0098204612731934,
      "learning_rate": 1.7637249871729093e-06,
      "loss": 0.0185,
      "step": 5500
    },
    {
      "epoch": 0.035339428476139474,
      "grad_norm": 0.7142383456230164,
      "learning_rate": 1.76693175987686e-06,
      "loss": 0.0184,
      "step": 5510
    },
    {
      "epoch": 0.03540356536992557,
      "grad_norm": 0.5789303183555603,
      "learning_rate": 1.7701385325808107e-06,
      "loss": 0.0119,
      "step": 5520
    },
    {
      "epoch": 0.03546770226371167,
      "grad_norm": 1.0011595487594604,
      "learning_rate": 1.7733453052847616e-06,
      "loss": 0.0236,
      "step": 5530
    },
    {
      "epoch": 0.03553183915749776,
      "grad_norm": 1.1528468132019043,
      "learning_rate": 1.7765520779887125e-06,
      "loss": 0.0287,
      "step": 5540
    },
    {
      "epoch": 0.03559597605128386,
      "grad_norm": 1.0052666664123535,
      "learning_rate": 1.779758850692663e-06,
      "loss": 0.0201,
      "step": 5550
    },
    {
      "epoch": 0.03566011294506996,
      "grad_norm": 1.2844178676605225,
      "learning_rate": 1.782965623396614e-06,
      "loss": 0.0259,
      "step": 5560
    },
    {
      "epoch": 0.035724249838856054,
      "grad_norm": 0.8071389198303223,
      "learning_rate": 1.7861723961005644e-06,
      "loss": 0.0261,
      "step": 5570
    },
    {
      "epoch": 0.035788386732642154,
      "grad_norm": 1.3172229528427124,
      "learning_rate": 1.7893791688045153e-06,
      "loss": 0.0193,
      "step": 5580
    },
    {
      "epoch": 0.03585252362642825,
      "grad_norm": 0.5351840257644653,
      "learning_rate": 1.792585941508466e-06,
      "loss": 0.0221,
      "step": 5590
    },
    {
      "epoch": 0.03591666052021435,
      "grad_norm": 1.4215500354766846,
      "learning_rate": 1.7957927142124169e-06,
      "loss": 0.0213,
      "step": 5600
    },
    {
      "epoch": 0.03598079741400044,
      "grad_norm": 1.0777472257614136,
      "learning_rate": 1.7989994869163673e-06,
      "loss": 0.0207,
      "step": 5610
    },
    {
      "epoch": 0.03604493430778654,
      "grad_norm": 2.632100820541382,
      "learning_rate": 1.8022062596203183e-06,
      "loss": 0.0265,
      "step": 5620
    },
    {
      "epoch": 0.03610907120157264,
      "grad_norm": 1.2422752380371094,
      "learning_rate": 1.8054130323242692e-06,
      "loss": 0.0182,
      "step": 5630
    },
    {
      "epoch": 0.03617320809535873,
      "grad_norm": 1.2604893445968628,
      "learning_rate": 1.8086198050282196e-06,
      "loss": 0.019,
      "step": 5640
    },
    {
      "epoch": 0.03623734498914483,
      "grad_norm": 0.9792515635490417,
      "learning_rate": 1.8118265777321705e-06,
      "loss": 0.0235,
      "step": 5650
    },
    {
      "epoch": 0.036301481882930926,
      "grad_norm": 0.6919666528701782,
      "learning_rate": 1.8150333504361212e-06,
      "loss": 0.0166,
      "step": 5660
    },
    {
      "epoch": 0.036365618776717026,
      "grad_norm": 0.734473705291748,
      "learning_rate": 1.8182401231400721e-06,
      "loss": 0.0206,
      "step": 5670
    },
    {
      "epoch": 0.03642975567050312,
      "grad_norm": 1.429874062538147,
      "learning_rate": 1.8214468958440226e-06,
      "loss": 0.0199,
      "step": 5680
    },
    {
      "epoch": 0.03649389256428922,
      "grad_norm": 1.1574482917785645,
      "learning_rate": 1.8246536685479735e-06,
      "loss": 0.0174,
      "step": 5690
    },
    {
      "epoch": 0.03655802945807531,
      "grad_norm": 0.7550815939903259,
      "learning_rate": 1.827860441251924e-06,
      "loss": 0.0229,
      "step": 5700
    },
    {
      "epoch": 0.03662216635186141,
      "grad_norm": 0.7385497093200684,
      "learning_rate": 1.8310672139558749e-06,
      "loss": 0.0283,
      "step": 5710
    },
    {
      "epoch": 0.03668630324564751,
      "grad_norm": 0.99532550573349,
      "learning_rate": 1.8342739866598258e-06,
      "loss": 0.0173,
      "step": 5720
    },
    {
      "epoch": 0.036750440139433606,
      "grad_norm": 0.7917321920394897,
      "learning_rate": 1.8374807593637765e-06,
      "loss": 0.0208,
      "step": 5730
    },
    {
      "epoch": 0.036814577033219706,
      "grad_norm": 1.0840858221054077,
      "learning_rate": 1.8406875320677274e-06,
      "loss": 0.0201,
      "step": 5740
    },
    {
      "epoch": 0.0368787139270058,
      "grad_norm": 0.8415246605873108,
      "learning_rate": 1.8438943047716779e-06,
      "loss": 0.025,
      "step": 5750
    },
    {
      "epoch": 0.0369428508207919,
      "grad_norm": 1.3072212934494019,
      "learning_rate": 1.8471010774756288e-06,
      "loss": 0.0176,
      "step": 5760
    },
    {
      "epoch": 0.03700698771457799,
      "grad_norm": 1.1021171808242798,
      "learning_rate": 1.8503078501795792e-06,
      "loss": 0.0181,
      "step": 5770
    },
    {
      "epoch": 0.03707112460836409,
      "grad_norm": 0.6983956098556519,
      "learning_rate": 1.8535146228835301e-06,
      "loss": 0.0235,
      "step": 5780
    },
    {
      "epoch": 0.03713526150215019,
      "grad_norm": 1.2423068284988403,
      "learning_rate": 1.856721395587481e-06,
      "loss": 0.0182,
      "step": 5790
    },
    {
      "epoch": 0.037199398395936285,
      "grad_norm": 2.747347354888916,
      "learning_rate": 1.8599281682914317e-06,
      "loss": 0.0209,
      "step": 5800
    },
    {
      "epoch": 0.037263535289722385,
      "grad_norm": 1.6623913049697876,
      "learning_rate": 1.8631349409953824e-06,
      "loss": 0.02,
      "step": 5810
    },
    {
      "epoch": 0.03732767218350848,
      "grad_norm": 0.9565622806549072,
      "learning_rate": 1.866341713699333e-06,
      "loss": 0.0154,
      "step": 5820
    },
    {
      "epoch": 0.03739180907729458,
      "grad_norm": 0.966018795967102,
      "learning_rate": 1.869548486403284e-06,
      "loss": 0.021,
      "step": 5830
    },
    {
      "epoch": 0.03745594597108067,
      "grad_norm": 1.0688713788986206,
      "learning_rate": 1.8727552591072345e-06,
      "loss": 0.0247,
      "step": 5840
    },
    {
      "epoch": 0.03752008286486677,
      "grad_norm": 1.0856199264526367,
      "learning_rate": 1.8759620318111854e-06,
      "loss": 0.0235,
      "step": 5850
    },
    {
      "epoch": 0.03758421975865287,
      "grad_norm": 1.232641339302063,
      "learning_rate": 1.879168804515136e-06,
      "loss": 0.0224,
      "step": 5860
    },
    {
      "epoch": 0.037648356652438965,
      "grad_norm": 0.9010834097862244,
      "learning_rate": 1.882375577219087e-06,
      "loss": 0.0235,
      "step": 5870
    },
    {
      "epoch": 0.037712493546225065,
      "grad_norm": 0.8574920296669006,
      "learning_rate": 1.8855823499230377e-06,
      "loss": 0.0229,
      "step": 5880
    },
    {
      "epoch": 0.03777663044001116,
      "grad_norm": 0.9682068824768066,
      "learning_rate": 1.8887891226269884e-06,
      "loss": 0.0189,
      "step": 5890
    },
    {
      "epoch": 0.03784076733379726,
      "grad_norm": 1.4444180727005005,
      "learning_rate": 1.8919958953309393e-06,
      "loss": 0.025,
      "step": 5900
    },
    {
      "epoch": 0.03790490422758335,
      "grad_norm": 0.8763965964317322,
      "learning_rate": 1.8952026680348897e-06,
      "loss": 0.0216,
      "step": 5910
    },
    {
      "epoch": 0.03796904112136945,
      "grad_norm": 1.5463361740112305,
      "learning_rate": 1.8984094407388406e-06,
      "loss": 0.0192,
      "step": 5920
    },
    {
      "epoch": 0.03803317801515555,
      "grad_norm": 0.7160257697105408,
      "learning_rate": 1.9016162134427913e-06,
      "loss": 0.0173,
      "step": 5930
    },
    {
      "epoch": 0.038097314908941644,
      "grad_norm": 2.2366437911987305,
      "learning_rate": 1.904822986146742e-06,
      "loss": 0.0241,
      "step": 5940
    },
    {
      "epoch": 0.038161451802727744,
      "grad_norm": 1.0212844610214233,
      "learning_rate": 1.9080297588506925e-06,
      "loss": 0.0186,
      "step": 5950
    },
    {
      "epoch": 0.03822558869651384,
      "grad_norm": 1.2753028869628906,
      "learning_rate": 1.9112365315546436e-06,
      "loss": 0.0189,
      "step": 5960
    },
    {
      "epoch": 0.03828972559029994,
      "grad_norm": 1.4174270629882812,
      "learning_rate": 1.9144433042585943e-06,
      "loss": 0.0208,
      "step": 5970
    },
    {
      "epoch": 0.03835386248408603,
      "grad_norm": 1.3357608318328857,
      "learning_rate": 1.917650076962545e-06,
      "loss": 0.0195,
      "step": 5980
    },
    {
      "epoch": 0.03841799937787213,
      "grad_norm": 1.2511043548583984,
      "learning_rate": 1.9208568496664957e-06,
      "loss": 0.0167,
      "step": 5990
    },
    {
      "epoch": 0.03848213627165823,
      "grad_norm": 0.9986559748649597,
      "learning_rate": 1.9240636223704464e-06,
      "loss": 0.0183,
      "step": 6000
    },
    {
      "epoch": 0.038546273165444324,
      "grad_norm": 1.2135497331619263,
      "learning_rate": 1.9272703950743975e-06,
      "loss": 0.0202,
      "step": 6010
    },
    {
      "epoch": 0.038610410059230424,
      "grad_norm": 0.6283160448074341,
      "learning_rate": 1.9304771677783477e-06,
      "loss": 0.0216,
      "step": 6020
    },
    {
      "epoch": 0.03867454695301652,
      "grad_norm": 0.8352381587028503,
      "learning_rate": 1.933683940482299e-06,
      "loss": 0.0216,
      "step": 6030
    },
    {
      "epoch": 0.03873868384680262,
      "grad_norm": 0.8876038789749146,
      "learning_rate": 1.9368907131862495e-06,
      "loss": 0.0202,
      "step": 6040
    },
    {
      "epoch": 0.03880282074058871,
      "grad_norm": 1.5208381414413452,
      "learning_rate": 1.9400974858902002e-06,
      "loss": 0.0218,
      "step": 6050
    },
    {
      "epoch": 0.03886695763437481,
      "grad_norm": 0.8314782977104187,
      "learning_rate": 1.943304258594151e-06,
      "loss": 0.0154,
      "step": 6060
    },
    {
      "epoch": 0.0389310945281609,
      "grad_norm": 0.7073200345039368,
      "learning_rate": 1.9465110312981016e-06,
      "loss": 0.02,
      "step": 6070
    },
    {
      "epoch": 0.038995231421947,
      "grad_norm": 0.969404935836792,
      "learning_rate": 1.9497178040020527e-06,
      "loss": 0.0219,
      "step": 6080
    },
    {
      "epoch": 0.0390593683157331,
      "grad_norm": 1.1272072792053223,
      "learning_rate": 1.952924576706003e-06,
      "loss": 0.0236,
      "step": 6090
    },
    {
      "epoch": 0.039123505209519197,
      "grad_norm": 1.5298181772232056,
      "learning_rate": 1.956131349409954e-06,
      "loss": 0.0248,
      "step": 6100
    },
    {
      "epoch": 0.0391876421033053,
      "grad_norm": 1.1177719831466675,
      "learning_rate": 1.959338122113905e-06,
      "loss": 0.0212,
      "step": 6110
    },
    {
      "epoch": 0.03925177899709139,
      "grad_norm": 1.3308947086334229,
      "learning_rate": 1.9625448948178555e-06,
      "loss": 0.0261,
      "step": 6120
    },
    {
      "epoch": 0.03931591589087749,
      "grad_norm": 0.9389868974685669,
      "learning_rate": 1.965751667521806e-06,
      "loss": 0.0164,
      "step": 6130
    },
    {
      "epoch": 0.03938005278466358,
      "grad_norm": 1.041259527206421,
      "learning_rate": 1.968958440225757e-06,
      "loss": 0.0268,
      "step": 6140
    },
    {
      "epoch": 0.03944418967844968,
      "grad_norm": 0.7473919987678528,
      "learning_rate": 1.972165212929708e-06,
      "loss": 0.0268,
      "step": 6150
    },
    {
      "epoch": 0.03950832657223578,
      "grad_norm": 1.006229043006897,
      "learning_rate": 1.9753719856336582e-06,
      "loss": 0.0238,
      "step": 6160
    },
    {
      "epoch": 0.039572463466021876,
      "grad_norm": 1.1355164051055908,
      "learning_rate": 1.9785787583376094e-06,
      "loss": 0.0203,
      "step": 6170
    },
    {
      "epoch": 0.039636600359807976,
      "grad_norm": 1.5693013668060303,
      "learning_rate": 1.98178553104156e-06,
      "loss": 0.0193,
      "step": 6180
    },
    {
      "epoch": 0.03970073725359407,
      "grad_norm": 1.2391225099563599,
      "learning_rate": 1.9849923037455107e-06,
      "loss": 0.0234,
      "step": 6190
    },
    {
      "epoch": 0.03976487414738017,
      "grad_norm": 1.2182879447937012,
      "learning_rate": 1.9881990764494614e-06,
      "loss": 0.0211,
      "step": 6200
    },
    {
      "epoch": 0.03982901104116626,
      "grad_norm": 1.1928573846817017,
      "learning_rate": 1.991405849153412e-06,
      "loss": 0.0235,
      "step": 6210
    },
    {
      "epoch": 0.03989314793495236,
      "grad_norm": 1.5095957517623901,
      "learning_rate": 1.9946126218573632e-06,
      "loss": 0.0201,
      "step": 6220
    },
    {
      "epoch": 0.03995728482873846,
      "grad_norm": 1.1891467571258545,
      "learning_rate": 1.9978193945613135e-06,
      "loss": 0.022,
      "step": 6230
    },
    {
      "epoch": 0.040021421722524556,
      "grad_norm": 1.2522825002670288,
      "learning_rate": 2.0010261672652646e-06,
      "loss": 0.025,
      "step": 6240
    },
    {
      "epoch": 0.040085558616310656,
      "grad_norm": 1.1242830753326416,
      "learning_rate": 2.0042329399692153e-06,
      "loss": 0.0231,
      "step": 6250
    },
    {
      "epoch": 0.04014969551009675,
      "grad_norm": 1.4165798425674438,
      "learning_rate": 2.007439712673166e-06,
      "loss": 0.0193,
      "step": 6260
    },
    {
      "epoch": 0.04021383240388285,
      "grad_norm": 1.4682064056396484,
      "learning_rate": 2.0106464853771167e-06,
      "loss": 0.0254,
      "step": 6270
    },
    {
      "epoch": 0.04027796929766894,
      "grad_norm": 0.9028723835945129,
      "learning_rate": 2.0138532580810674e-06,
      "loss": 0.0229,
      "step": 6280
    },
    {
      "epoch": 0.04034210619145504,
      "grad_norm": 0.7347221374511719,
      "learning_rate": 2.017060030785018e-06,
      "loss": 0.0204,
      "step": 6290
    },
    {
      "epoch": 0.04040624308524114,
      "grad_norm": 1.0344412326812744,
      "learning_rate": 2.0202668034889687e-06,
      "loss": 0.0249,
      "step": 6300
    },
    {
      "epoch": 0.040470379979027235,
      "grad_norm": 1.3016290664672852,
      "learning_rate": 2.02347357619292e-06,
      "loss": 0.0179,
      "step": 6310
    },
    {
      "epoch": 0.040534516872813335,
      "grad_norm": 1.060326337814331,
      "learning_rate": 2.02668034889687e-06,
      "loss": 0.0171,
      "step": 6320
    },
    {
      "epoch": 0.04059865376659943,
      "grad_norm": 1.2995046377182007,
      "learning_rate": 2.0298871216008212e-06,
      "loss": 0.0211,
      "step": 6330
    },
    {
      "epoch": 0.04066279066038553,
      "grad_norm": 1.4574737548828125,
      "learning_rate": 2.033093894304772e-06,
      "loss": 0.0196,
      "step": 6340
    },
    {
      "epoch": 0.04072692755417162,
      "grad_norm": 1.034865379333496,
      "learning_rate": 2.0363006670087226e-06,
      "loss": 0.0203,
      "step": 6350
    },
    {
      "epoch": 0.04079106444795772,
      "grad_norm": 1.1399013996124268,
      "learning_rate": 2.0395074397126733e-06,
      "loss": 0.0237,
      "step": 6360
    },
    {
      "epoch": 0.04085520134174382,
      "grad_norm": 1.309984803199768,
      "learning_rate": 2.042714212416624e-06,
      "loss": 0.0205,
      "step": 6370
    },
    {
      "epoch": 0.040919338235529915,
      "grad_norm": 0.6266277432441711,
      "learning_rate": 2.0459209851205747e-06,
      "loss": 0.0134,
      "step": 6380
    },
    {
      "epoch": 0.040983475129316015,
      "grad_norm": 0.8712389469146729,
      "learning_rate": 2.0491277578245254e-06,
      "loss": 0.021,
      "step": 6390
    },
    {
      "epoch": 0.04104761202310211,
      "grad_norm": 0.8083272576332092,
      "learning_rate": 2.0523345305284765e-06,
      "loss": 0.0185,
      "step": 6400
    },
    {
      "epoch": 0.04111174891688821,
      "grad_norm": 0.7754104733467102,
      "learning_rate": 2.055541303232427e-06,
      "loss": 0.0282,
      "step": 6410
    },
    {
      "epoch": 0.0411758858106743,
      "grad_norm": 0.6917300224304199,
      "learning_rate": 2.058748075936378e-06,
      "loss": 0.0195,
      "step": 6420
    },
    {
      "epoch": 0.0412400227044604,
      "grad_norm": 1.2205440998077393,
      "learning_rate": 2.0619548486403286e-06,
      "loss": 0.0155,
      "step": 6430
    },
    {
      "epoch": 0.041304159598246494,
      "grad_norm": 1.1273776292800903,
      "learning_rate": 2.0651616213442792e-06,
      "loss": 0.0199,
      "step": 6440
    },
    {
      "epoch": 0.041368296492032594,
      "grad_norm": 1.1309514045715332,
      "learning_rate": 2.06836839404823e-06,
      "loss": 0.0207,
      "step": 6450
    },
    {
      "epoch": 0.041432433385818694,
      "grad_norm": 1.0427113771438599,
      "learning_rate": 2.0715751667521806e-06,
      "loss": 0.0162,
      "step": 6460
    },
    {
      "epoch": 0.04149657027960479,
      "grad_norm": 1.0652873516082764,
      "learning_rate": 2.0747819394561317e-06,
      "loss": 0.0195,
      "step": 6470
    },
    {
      "epoch": 0.04156070717339089,
      "grad_norm": 1.2061506509780884,
      "learning_rate": 2.0779887121600824e-06,
      "loss": 0.0219,
      "step": 6480
    },
    {
      "epoch": 0.04162484406717698,
      "grad_norm": 1.1749380826950073,
      "learning_rate": 2.081195484864033e-06,
      "loss": 0.0235,
      "step": 6490
    },
    {
      "epoch": 0.04168898096096308,
      "grad_norm": 0.9566291570663452,
      "learning_rate": 2.084402257567984e-06,
      "loss": 0.0186,
      "step": 6500
    },
    {
      "epoch": 0.041753117854749174,
      "grad_norm": 1.0603986978530884,
      "learning_rate": 2.0876090302719345e-06,
      "loss": 0.0205,
      "step": 6510
    },
    {
      "epoch": 0.041817254748535274,
      "grad_norm": 1.050813913345337,
      "learning_rate": 2.090815802975885e-06,
      "loss": 0.0223,
      "step": 6520
    },
    {
      "epoch": 0.041881391642321374,
      "grad_norm": 0.7652897238731384,
      "learning_rate": 2.094022575679836e-06,
      "loss": 0.0172,
      "step": 6530
    },
    {
      "epoch": 0.04194552853610747,
      "grad_norm": 0.9831271171569824,
      "learning_rate": 2.0972293483837866e-06,
      "loss": 0.0229,
      "step": 6540
    },
    {
      "epoch": 0.04200966542989357,
      "grad_norm": 0.8050175309181213,
      "learning_rate": 2.1004361210877377e-06,
      "loss": 0.017,
      "step": 6550
    },
    {
      "epoch": 0.04207380232367966,
      "grad_norm": 1.143938660621643,
      "learning_rate": 2.1036428937916884e-06,
      "loss": 0.0158,
      "step": 6560
    },
    {
      "epoch": 0.04213793921746576,
      "grad_norm": 0.8601158857345581,
      "learning_rate": 2.106849666495639e-06,
      "loss": 0.0229,
      "step": 6570
    },
    {
      "epoch": 0.04220207611125185,
      "grad_norm": 0.7796095013618469,
      "learning_rate": 2.1100564391995897e-06,
      "loss": 0.0211,
      "step": 6580
    },
    {
      "epoch": 0.04226621300503795,
      "grad_norm": 0.9684072732925415,
      "learning_rate": 2.1132632119035404e-06,
      "loss": 0.0188,
      "step": 6590
    },
    {
      "epoch": 0.04233034989882405,
      "grad_norm": 0.705844521522522,
      "learning_rate": 2.116469984607491e-06,
      "loss": 0.0204,
      "step": 6600
    },
    {
      "epoch": 0.042394486792610146,
      "grad_norm": 1.1430692672729492,
      "learning_rate": 2.119676757311442e-06,
      "loss": 0.0183,
      "step": 6610
    },
    {
      "epoch": 0.042458623686396246,
      "grad_norm": 0.8286201357841492,
      "learning_rate": 2.1228835300153925e-06,
      "loss": 0.0205,
      "step": 6620
    },
    {
      "epoch": 0.04252276058018234,
      "grad_norm": 1.245947241783142,
      "learning_rate": 2.126090302719343e-06,
      "loss": 0.0303,
      "step": 6630
    },
    {
      "epoch": 0.04258689747396844,
      "grad_norm": 2.2712762355804443,
      "learning_rate": 2.1292970754232943e-06,
      "loss": 0.0269,
      "step": 6640
    },
    {
      "epoch": 0.04265103436775453,
      "grad_norm": 0.8857012987136841,
      "learning_rate": 2.132503848127245e-06,
      "loss": 0.0182,
      "step": 6650
    },
    {
      "epoch": 0.04271517126154063,
      "grad_norm": 1.162306547164917,
      "learning_rate": 2.1357106208311957e-06,
      "loss": 0.022,
      "step": 6660
    },
    {
      "epoch": 0.04277930815532673,
      "grad_norm": 0.4371137022972107,
      "learning_rate": 2.1389173935351464e-06,
      "loss": 0.0205,
      "step": 6670
    },
    {
      "epoch": 0.042843445049112826,
      "grad_norm": 0.761978805065155,
      "learning_rate": 2.142124166239097e-06,
      "loss": 0.0203,
      "step": 6680
    },
    {
      "epoch": 0.042907581942898926,
      "grad_norm": 0.7236703038215637,
      "learning_rate": 2.1453309389430478e-06,
      "loss": 0.0199,
      "step": 6690
    },
    {
      "epoch": 0.04297171883668502,
      "grad_norm": 0.8764296770095825,
      "learning_rate": 2.1485377116469984e-06,
      "loss": 0.02,
      "step": 6700
    },
    {
      "epoch": 0.04303585573047112,
      "grad_norm": 0.6730541586875916,
      "learning_rate": 2.1517444843509496e-06,
      "loss": 0.0194,
      "step": 6710
    },
    {
      "epoch": 0.04309999262425721,
      "grad_norm": 1.8022714853286743,
      "learning_rate": 2.1549512570549e-06,
      "loss": 0.019,
      "step": 6720
    },
    {
      "epoch": 0.04316412951804331,
      "grad_norm": 0.8802977800369263,
      "learning_rate": 2.158158029758851e-06,
      "loss": 0.0149,
      "step": 6730
    },
    {
      "epoch": 0.04322826641182941,
      "grad_norm": 1.2370784282684326,
      "learning_rate": 2.1613648024628016e-06,
      "loss": 0.0172,
      "step": 6740
    },
    {
      "epoch": 0.043292403305615505,
      "grad_norm": 1.1011195182800293,
      "learning_rate": 2.1645715751667523e-06,
      "loss": 0.0253,
      "step": 6750
    },
    {
      "epoch": 0.043356540199401605,
      "grad_norm": 0.8227287530899048,
      "learning_rate": 2.167778347870703e-06,
      "loss": 0.0185,
      "step": 6760
    },
    {
      "epoch": 0.0434206770931877,
      "grad_norm": 0.6026843190193176,
      "learning_rate": 2.1709851205746537e-06,
      "loss": 0.0148,
      "step": 6770
    },
    {
      "epoch": 0.0434848139869738,
      "grad_norm": 1.1184115409851074,
      "learning_rate": 2.174191893278605e-06,
      "loss": 0.0178,
      "step": 6780
    },
    {
      "epoch": 0.04354895088075989,
      "grad_norm": 0.9370375275611877,
      "learning_rate": 2.177398665982555e-06,
      "loss": 0.0209,
      "step": 6790
    },
    {
      "epoch": 0.04361308777454599,
      "grad_norm": 1.184848427772522,
      "learning_rate": 2.180605438686506e-06,
      "loss": 0.024,
      "step": 6800
    },
    {
      "epoch": 0.043677224668332085,
      "grad_norm": 0.7680546045303345,
      "learning_rate": 2.183812211390457e-06,
      "loss": 0.0112,
      "step": 6810
    },
    {
      "epoch": 0.043741361562118185,
      "grad_norm": 1.324097990989685,
      "learning_rate": 2.1870189840944076e-06,
      "loss": 0.0226,
      "step": 6820
    },
    {
      "epoch": 0.043805498455904285,
      "grad_norm": 0.9823472499847412,
      "learning_rate": 2.1902257567983583e-06,
      "loss": 0.025,
      "step": 6830
    },
    {
      "epoch": 0.04386963534969038,
      "grad_norm": 1.1413768529891968,
      "learning_rate": 2.193432529502309e-06,
      "loss": 0.0247,
      "step": 6840
    },
    {
      "epoch": 0.04393377224347648,
      "grad_norm": 1.236708402633667,
      "learning_rate": 2.19663930220626e-06,
      "loss": 0.0223,
      "step": 6850
    },
    {
      "epoch": 0.04399790913726257,
      "grad_norm": 0.8317570090293884,
      "learning_rate": 2.1998460749102103e-06,
      "loss": 0.0145,
      "step": 6860
    },
    {
      "epoch": 0.04406204603104867,
      "grad_norm": 1.6575742959976196,
      "learning_rate": 2.2030528476141614e-06,
      "loss": 0.0226,
      "step": 6870
    },
    {
      "epoch": 0.044126182924834764,
      "grad_norm": 0.9479625821113586,
      "learning_rate": 2.206259620318112e-06,
      "loss": 0.0168,
      "step": 6880
    },
    {
      "epoch": 0.044190319818620864,
      "grad_norm": 0.414814293384552,
      "learning_rate": 2.209466393022063e-06,
      "loss": 0.0202,
      "step": 6890
    },
    {
      "epoch": 0.044254456712406964,
      "grad_norm": 1.0208877325057983,
      "learning_rate": 2.2126731657260135e-06,
      "loss": 0.0172,
      "step": 6900
    },
    {
      "epoch": 0.04431859360619306,
      "grad_norm": 0.67899489402771,
      "learning_rate": 2.215879938429964e-06,
      "loss": 0.0195,
      "step": 6910
    },
    {
      "epoch": 0.04438273049997916,
      "grad_norm": 1.2453488111495972,
      "learning_rate": 2.2190867111339153e-06,
      "loss": 0.0213,
      "step": 6920
    },
    {
      "epoch": 0.04444686739376525,
      "grad_norm": 1.481083631515503,
      "learning_rate": 2.2222934838378656e-06,
      "loss": 0.0172,
      "step": 6930
    },
    {
      "epoch": 0.04451100428755135,
      "grad_norm": 0.8547831177711487,
      "learning_rate": 2.2255002565418167e-06,
      "loss": 0.0157,
      "step": 6940
    },
    {
      "epoch": 0.044575141181337444,
      "grad_norm": 1.2726802825927734,
      "learning_rate": 2.228707029245767e-06,
      "loss": 0.0189,
      "step": 6950
    },
    {
      "epoch": 0.044639278075123544,
      "grad_norm": 0.828345000743866,
      "learning_rate": 2.231913801949718e-06,
      "loss": 0.0186,
      "step": 6960
    },
    {
      "epoch": 0.044703414968909644,
      "grad_norm": 0.7094138264656067,
      "learning_rate": 2.2351205746536688e-06,
      "loss": 0.0208,
      "step": 6970
    },
    {
      "epoch": 0.04476755186269574,
      "grad_norm": 0.6570111513137817,
      "learning_rate": 2.2383273473576194e-06,
      "loss": 0.0172,
      "step": 6980
    },
    {
      "epoch": 0.04483168875648184,
      "grad_norm": 1.5432921648025513,
      "learning_rate": 2.24153412006157e-06,
      "loss": 0.0245,
      "step": 6990
    },
    {
      "epoch": 0.04489582565026793,
      "grad_norm": 0.7850481271743774,
      "learning_rate": 2.244740892765521e-06,
      "loss": 0.0207,
      "step": 7000
    },
    {
      "epoch": 0.04495996254405403,
      "grad_norm": 0.7977728247642517,
      "learning_rate": 2.247947665469472e-06,
      "loss": 0.0119,
      "step": 7010
    },
    {
      "epoch": 0.04502409943784012,
      "grad_norm": 1.1667042970657349,
      "learning_rate": 2.251154438173422e-06,
      "loss": 0.015,
      "step": 7020
    },
    {
      "epoch": 0.04508823633162622,
      "grad_norm": 1.0411245822906494,
      "learning_rate": 2.2543612108773733e-06,
      "loss": 0.0194,
      "step": 7030
    },
    {
      "epoch": 0.04515237322541232,
      "grad_norm": 1.2678431272506714,
      "learning_rate": 2.257567983581324e-06,
      "loss": 0.0217,
      "step": 7040
    },
    {
      "epoch": 0.045216510119198416,
      "grad_norm": 1.2046818733215332,
      "learning_rate": 2.2607747562852747e-06,
      "loss": 0.0197,
      "step": 7050
    },
    {
      "epoch": 0.045280647012984516,
      "grad_norm": 1.0384752750396729,
      "learning_rate": 2.2639815289892254e-06,
      "loss": 0.019,
      "step": 7060
    },
    {
      "epoch": 0.04534478390677061,
      "grad_norm": 0.8046727776527405,
      "learning_rate": 2.267188301693176e-06,
      "loss": 0.0198,
      "step": 7070
    },
    {
      "epoch": 0.04540892080055671,
      "grad_norm": 1.2377238273620605,
      "learning_rate": 2.270395074397127e-06,
      "loss": 0.0196,
      "step": 7080
    },
    {
      "epoch": 0.0454730576943428,
      "grad_norm": 1.2584959268569946,
      "learning_rate": 2.2736018471010775e-06,
      "loss": 0.0176,
      "step": 7090
    },
    {
      "epoch": 0.0455371945881289,
      "grad_norm": 0.8299155831336975,
      "learning_rate": 2.2768086198050286e-06,
      "loss": 0.0201,
      "step": 7100
    },
    {
      "epoch": 0.045601331481914996,
      "grad_norm": 0.7531854510307312,
      "learning_rate": 2.2800153925089793e-06,
      "loss": 0.0155,
      "step": 7110
    },
    {
      "epoch": 0.045665468375701096,
      "grad_norm": 0.6106268763542175,
      "learning_rate": 2.28322216521293e-06,
      "loss": 0.0162,
      "step": 7120
    },
    {
      "epoch": 0.045729605269487196,
      "grad_norm": 1.0959511995315552,
      "learning_rate": 2.2864289379168806e-06,
      "loss": 0.0154,
      "step": 7130
    },
    {
      "epoch": 0.04579374216327329,
      "grad_norm": 1.0702574253082275,
      "learning_rate": 2.2896357106208313e-06,
      "loss": 0.0205,
      "step": 7140
    },
    {
      "epoch": 0.04585787905705939,
      "grad_norm": 0.9860323667526245,
      "learning_rate": 2.2928424833247824e-06,
      "loss": 0.019,
      "step": 7150
    },
    {
      "epoch": 0.04592201595084548,
      "grad_norm": 0.8199664950370789,
      "learning_rate": 2.2960492560287327e-06,
      "loss": 0.017,
      "step": 7160
    },
    {
      "epoch": 0.04598615284463158,
      "grad_norm": 0.6780887842178345,
      "learning_rate": 2.299256028732684e-06,
      "loss": 0.0198,
      "step": 7170
    },
    {
      "epoch": 0.046050289738417675,
      "grad_norm": 0.9349753260612488,
      "learning_rate": 2.3024628014366345e-06,
      "loss": 0.0147,
      "step": 7180
    },
    {
      "epoch": 0.046114426632203775,
      "grad_norm": 0.8048481941223145,
      "learning_rate": 2.305669574140585e-06,
      "loss": 0.0195,
      "step": 7190
    },
    {
      "epoch": 0.046178563525989876,
      "grad_norm": 0.797756016254425,
      "learning_rate": 2.308876346844536e-06,
      "loss": 0.0211,
      "step": 7200
    },
    {
      "epoch": 0.04624270041977597,
      "grad_norm": 1.1557680368423462,
      "learning_rate": 2.3120831195484866e-06,
      "loss": 0.0184,
      "step": 7210
    },
    {
      "epoch": 0.04630683731356207,
      "grad_norm": 1.015023946762085,
      "learning_rate": 2.3152898922524373e-06,
      "loss": 0.02,
      "step": 7220
    },
    {
      "epoch": 0.04637097420734816,
      "grad_norm": 0.8543024063110352,
      "learning_rate": 2.318496664956388e-06,
      "loss": 0.0208,
      "step": 7230
    },
    {
      "epoch": 0.04643511110113426,
      "grad_norm": 1.2135494947433472,
      "learning_rate": 2.321703437660339e-06,
      "loss": 0.0171,
      "step": 7240
    },
    {
      "epoch": 0.046499247994920355,
      "grad_norm": 1.1175837516784668,
      "learning_rate": 2.3249102103642893e-06,
      "loss": 0.018,
      "step": 7250
    },
    {
      "epoch": 0.046563384888706455,
      "grad_norm": 0.9123100638389587,
      "learning_rate": 2.3281169830682404e-06,
      "loss": 0.0145,
      "step": 7260
    },
    {
      "epoch": 0.046627521782492555,
      "grad_norm": 0.9876992702484131,
      "learning_rate": 2.331323755772191e-06,
      "loss": 0.0151,
      "step": 7270
    },
    {
      "epoch": 0.04669165867627865,
      "grad_norm": 1.081762671470642,
      "learning_rate": 2.334530528476142e-06,
      "loss": 0.0172,
      "step": 7280
    },
    {
      "epoch": 0.04675579557006475,
      "grad_norm": 1.0402169227600098,
      "learning_rate": 2.3377373011800925e-06,
      "loss": 0.0153,
      "step": 7290
    },
    {
      "epoch": 0.04681993246385084,
      "grad_norm": 0.9691343307495117,
      "learning_rate": 2.340944073884043e-06,
      "loss": 0.0135,
      "step": 7300
    },
    {
      "epoch": 0.04688406935763694,
      "grad_norm": 0.9390423893928528,
      "learning_rate": 2.344150846587994e-06,
      "loss": 0.0135,
      "step": 7310
    },
    {
      "epoch": 0.046948206251423034,
      "grad_norm": 0.5622203946113586,
      "learning_rate": 2.3473576192919446e-06,
      "loss": 0.0112,
      "step": 7320
    },
    {
      "epoch": 0.047012343145209134,
      "grad_norm": 0.5227202773094177,
      "learning_rate": 2.3505643919958957e-06,
      "loss": 0.0112,
      "step": 7330
    },
    {
      "epoch": 0.047076480038995235,
      "grad_norm": 0.6646885275840759,
      "learning_rate": 2.3537711646998464e-06,
      "loss": 0.0171,
      "step": 7340
    },
    {
      "epoch": 0.04714061693278133,
      "grad_norm": 0.8758254051208496,
      "learning_rate": 2.356977937403797e-06,
      "loss": 0.0159,
      "step": 7350
    },
    {
      "epoch": 0.04720475382656743,
      "grad_norm": 1.2100036144256592,
      "learning_rate": 2.3601847101077478e-06,
      "loss": 0.016,
      "step": 7360
    },
    {
      "epoch": 0.04726889072035352,
      "grad_norm": 0.970321774482727,
      "learning_rate": 2.3633914828116985e-06,
      "loss": 0.0196,
      "step": 7370
    },
    {
      "epoch": 0.04733302761413962,
      "grad_norm": 0.9241225719451904,
      "learning_rate": 2.366598255515649e-06,
      "loss": 0.0155,
      "step": 7380
    },
    {
      "epoch": 0.047397164507925714,
      "grad_norm": 1.0202387571334839,
      "learning_rate": 2.3698050282196e-06,
      "loss": 0.0138,
      "step": 7390
    },
    {
      "epoch": 0.047461301401711814,
      "grad_norm": 1.1858984231948853,
      "learning_rate": 2.3730118009235505e-06,
      "loss": 0.0143,
      "step": 7400
    },
    {
      "epoch": 0.047525438295497914,
      "grad_norm": 0.733315646648407,
      "learning_rate": 2.3762185736275016e-06,
      "loss": 0.018,
      "step": 7410
    },
    {
      "epoch": 0.04758957518928401,
      "grad_norm": 1.1660308837890625,
      "learning_rate": 2.3794253463314523e-06,
      "loss": 0.0162,
      "step": 7420
    },
    {
      "epoch": 0.04765371208307011,
      "grad_norm": 0.8735769987106323,
      "learning_rate": 2.382632119035403e-06,
      "loss": 0.0196,
      "step": 7430
    },
    {
      "epoch": 0.0477178489768562,
      "grad_norm": 1.218178629875183,
      "learning_rate": 2.3858388917393537e-06,
      "loss": 0.0167,
      "step": 7440
    },
    {
      "epoch": 0.0477819858706423,
      "grad_norm": 1.3256758451461792,
      "learning_rate": 2.3890456644433044e-06,
      "loss": 0.013,
      "step": 7450
    },
    {
      "epoch": 0.04784612276442839,
      "grad_norm": 1.5111521482467651,
      "learning_rate": 2.392252437147255e-06,
      "loss": 0.0195,
      "step": 7460
    },
    {
      "epoch": 0.047910259658214494,
      "grad_norm": 0.35723578929901123,
      "learning_rate": 2.3954592098512058e-06,
      "loss": 0.0157,
      "step": 7470
    },
    {
      "epoch": 0.04797439655200059,
      "grad_norm": 0.7256841063499451,
      "learning_rate": 2.398665982555157e-06,
      "loss": 0.0187,
      "step": 7480
    },
    {
      "epoch": 0.04803853344578669,
      "grad_norm": 0.9517735242843628,
      "learning_rate": 2.4018727552591076e-06,
      "loss": 0.0141,
      "step": 7490
    },
    {
      "epoch": 0.04810267033957279,
      "grad_norm": 0.8512372970581055,
      "learning_rate": 2.4050795279630583e-06,
      "loss": 0.0224,
      "step": 7500
    },
    {
      "epoch": 0.04816680723335888,
      "grad_norm": 1.0928106307983398,
      "learning_rate": 2.408286300667009e-06,
      "loss": 0.0144,
      "step": 7510
    },
    {
      "epoch": 0.04823094412714498,
      "grad_norm": 0.7627331614494324,
      "learning_rate": 2.4114930733709596e-06,
      "loss": 0.0183,
      "step": 7520
    },
    {
      "epoch": 0.04829508102093107,
      "grad_norm": 0.8778024315834045,
      "learning_rate": 2.4146998460749103e-06,
      "loss": 0.0182,
      "step": 7530
    },
    {
      "epoch": 0.04835921791471717,
      "grad_norm": 0.993202805519104,
      "learning_rate": 2.417906618778861e-06,
      "loss": 0.0145,
      "step": 7540
    },
    {
      "epoch": 0.048423354808503266,
      "grad_norm": 0.9249763488769531,
      "learning_rate": 2.421113391482812e-06,
      "loss": 0.0183,
      "step": 7550
    },
    {
      "epoch": 0.048487491702289366,
      "grad_norm": 0.865367591381073,
      "learning_rate": 2.4243201641867624e-06,
      "loss": 0.0219,
      "step": 7560
    },
    {
      "epoch": 0.048551628596075466,
      "grad_norm": 0.7539327144622803,
      "learning_rate": 2.4275269368907135e-06,
      "loss": 0.0181,
      "step": 7570
    },
    {
      "epoch": 0.04861576548986156,
      "grad_norm": 0.7270578742027283,
      "learning_rate": 2.430733709594664e-06,
      "loss": 0.0147,
      "step": 7580
    },
    {
      "epoch": 0.04867990238364766,
      "grad_norm": 0.9653283357620239,
      "learning_rate": 2.433940482298615e-06,
      "loss": 0.021,
      "step": 7590
    },
    {
      "epoch": 0.04874403927743375,
      "grad_norm": 0.9090064764022827,
      "learning_rate": 2.4371472550025656e-06,
      "loss": 0.0199,
      "step": 7600
    },
    {
      "epoch": 0.04880817617121985,
      "grad_norm": 0.9785999655723572,
      "learning_rate": 2.4403540277065163e-06,
      "loss": 0.0127,
      "step": 7610
    },
    {
      "epoch": 0.048872313065005946,
      "grad_norm": 0.8310036063194275,
      "learning_rate": 2.443560800410467e-06,
      "loss": 0.0118,
      "step": 7620
    },
    {
      "epoch": 0.048936449958792046,
      "grad_norm": 0.9784269332885742,
      "learning_rate": 2.4467675731144177e-06,
      "loss": 0.0148,
      "step": 7630
    },
    {
      "epoch": 0.049000586852578146,
      "grad_norm": 0.8932363986968994,
      "learning_rate": 2.4499743458183688e-06,
      "loss": 0.0193,
      "step": 7640
    },
    {
      "epoch": 0.04906472374636424,
      "grad_norm": 0.9446199536323547,
      "learning_rate": 2.453181118522319e-06,
      "loss": 0.0151,
      "step": 7650
    },
    {
      "epoch": 0.04912886064015034,
      "grad_norm": 1.0221552848815918,
      "learning_rate": 2.45638789122627e-06,
      "loss": 0.0181,
      "step": 7660
    },
    {
      "epoch": 0.04919299753393643,
      "grad_norm": 0.6242429614067078,
      "learning_rate": 2.459594663930221e-06,
      "loss": 0.0144,
      "step": 7670
    },
    {
      "epoch": 0.04925713442772253,
      "grad_norm": 1.2173144817352295,
      "learning_rate": 2.4628014366341715e-06,
      "loss": 0.0122,
      "step": 7680
    },
    {
      "epoch": 0.049321271321508625,
      "grad_norm": 0.6649882197380066,
      "learning_rate": 2.4660082093381222e-06,
      "loss": 0.0148,
      "step": 7690
    },
    {
      "epoch": 0.049385408215294725,
      "grad_norm": 1.4457169771194458,
      "learning_rate": 2.469214982042073e-06,
      "loss": 0.0186,
      "step": 7700
    },
    {
      "epoch": 0.049449545109080825,
      "grad_norm": 0.8608556389808655,
      "learning_rate": 2.472421754746024e-06,
      "loss": 0.0184,
      "step": 7710
    },
    {
      "epoch": 0.04951368200286692,
      "grad_norm": 1.0127040147781372,
      "learning_rate": 2.4756285274499743e-06,
      "loss": 0.0193,
      "step": 7720
    },
    {
      "epoch": 0.04957781889665302,
      "grad_norm": 1.072650671005249,
      "learning_rate": 2.4788353001539254e-06,
      "loss": 0.0184,
      "step": 7730
    },
    {
      "epoch": 0.04964195579043911,
      "grad_norm": 1.0660983324050903,
      "learning_rate": 2.482042072857876e-06,
      "loss": 0.0184,
      "step": 7740
    },
    {
      "epoch": 0.04970609268422521,
      "grad_norm": 0.8504105806350708,
      "learning_rate": 2.4852488455618268e-06,
      "loss": 0.0156,
      "step": 7750
    },
    {
      "epoch": 0.049770229578011305,
      "grad_norm": 0.5149958729743958,
      "learning_rate": 2.4884556182657775e-06,
      "loss": 0.0155,
      "step": 7760
    },
    {
      "epoch": 0.049834366471797405,
      "grad_norm": 1.1049935817718506,
      "learning_rate": 2.491662390969728e-06,
      "loss": 0.0188,
      "step": 7770
    },
    {
      "epoch": 0.049898503365583505,
      "grad_norm": 1.0066155195236206,
      "learning_rate": 2.4948691636736793e-06,
      "loss": 0.0151,
      "step": 7780
    },
    {
      "epoch": 0.0499626402593696,
      "grad_norm": 0.8193107843399048,
      "learning_rate": 2.4980759363776295e-06,
      "loss": 0.0154,
      "step": 7790
    },
    {
      "epoch": 0.0500267771531557,
      "grad_norm": 1.3258004188537598,
      "learning_rate": 2.5012827090815806e-06,
      "loss": 0.0194,
      "step": 7800
    },
    {
      "epoch": 0.05009091404694179,
      "grad_norm": 1.3659716844558716,
      "learning_rate": 2.5044894817855313e-06,
      "loss": 0.0195,
      "step": 7810
    },
    {
      "epoch": 0.05015505094072789,
      "grad_norm": 0.598633885383606,
      "learning_rate": 2.5076962544894816e-06,
      "loss": 0.0164,
      "step": 7820
    },
    {
      "epoch": 0.050219187834513984,
      "grad_norm": 0.8737356662750244,
      "learning_rate": 2.5109030271934327e-06,
      "loss": 0.0114,
      "step": 7830
    },
    {
      "epoch": 0.050283324728300084,
      "grad_norm": 1.0397603511810303,
      "learning_rate": 2.5141097998973834e-06,
      "loss": 0.0151,
      "step": 7840
    },
    {
      "epoch": 0.05034746162208618,
      "grad_norm": 0.946444571018219,
      "learning_rate": 2.5173165726013345e-06,
      "loss": 0.02,
      "step": 7850
    },
    {
      "epoch": 0.05041159851587228,
      "grad_norm": 0.6684794425964355,
      "learning_rate": 2.520523345305285e-06,
      "loss": 0.0155,
      "step": 7860
    },
    {
      "epoch": 0.05047573540965838,
      "grad_norm": 1.3526276350021362,
      "learning_rate": 2.5237301180092355e-06,
      "loss": 0.015,
      "step": 7870
    },
    {
      "epoch": 0.05053987230344447,
      "grad_norm": 1.45792555809021,
      "learning_rate": 2.526936890713186e-06,
      "loss": 0.0116,
      "step": 7880
    },
    {
      "epoch": 0.05060400919723057,
      "grad_norm": 0.5818025469779968,
      "learning_rate": 2.5301436634171373e-06,
      "loss": 0.0152,
      "step": 7890
    },
    {
      "epoch": 0.050668146091016664,
      "grad_norm": 1.4632349014282227,
      "learning_rate": 2.533350436121088e-06,
      "loss": 0.0138,
      "step": 7900
    },
    {
      "epoch": 0.050732282984802764,
      "grad_norm": 1.1218039989471436,
      "learning_rate": 2.5365572088250382e-06,
      "loss": 0.0145,
      "step": 7910
    },
    {
      "epoch": 0.05079641987858886,
      "grad_norm": 1.6577773094177246,
      "learning_rate": 2.5397639815289893e-06,
      "loss": 0.0165,
      "step": 7920
    },
    {
      "epoch": 0.05086055677237496,
      "grad_norm": 1.0832455158233643,
      "learning_rate": 2.54297075423294e-06,
      "loss": 0.015,
      "step": 7930
    },
    {
      "epoch": 0.05092469366616106,
      "grad_norm": 0.9392962455749512,
      "learning_rate": 2.546177526936891e-06,
      "loss": 0.0275,
      "step": 7940
    },
    {
      "epoch": 0.05098883055994715,
      "grad_norm": 0.7276794910430908,
      "learning_rate": 2.549384299640842e-06,
      "loss": 0.0215,
      "step": 7950
    },
    {
      "epoch": 0.05105296745373325,
      "grad_norm": 0.7038756012916565,
      "learning_rate": 2.552591072344792e-06,
      "loss": 0.0187,
      "step": 7960
    },
    {
      "epoch": 0.05111710434751934,
      "grad_norm": 1.1616355180740356,
      "learning_rate": 2.5557978450487432e-06,
      "loss": 0.0137,
      "step": 7970
    },
    {
      "epoch": 0.05118124124130544,
      "grad_norm": 1.0356817245483398,
      "learning_rate": 2.559004617752694e-06,
      "loss": 0.0181,
      "step": 7980
    },
    {
      "epoch": 0.051245378135091536,
      "grad_norm": 1.3789634704589844,
      "learning_rate": 2.5622113904566446e-06,
      "loss": 0.019,
      "step": 7990
    },
    {
      "epoch": 0.051309515028877636,
      "grad_norm": 1.080942988395691,
      "learning_rate": 2.5654181631605953e-06,
      "loss": 0.0144,
      "step": 8000
    },
    {
      "epoch": 0.051373651922663736,
      "grad_norm": 0.87821364402771,
      "learning_rate": 2.568624935864546e-06,
      "loss": 0.014,
      "step": 8010
    },
    {
      "epoch": 0.05143778881644983,
      "grad_norm": 0.784485936164856,
      "learning_rate": 2.5718317085684967e-06,
      "loss": 0.0146,
      "step": 8020
    },
    {
      "epoch": 0.05150192571023593,
      "grad_norm": 0.8742894530296326,
      "learning_rate": 2.5750384812724478e-06,
      "loss": 0.0162,
      "step": 8030
    },
    {
      "epoch": 0.05156606260402202,
      "grad_norm": 0.8528547883033752,
      "learning_rate": 2.5782452539763985e-06,
      "loss": 0.0178,
      "step": 8040
    },
    {
      "epoch": 0.05163019949780812,
      "grad_norm": 1.3443467617034912,
      "learning_rate": 2.5814520266803487e-06,
      "loss": 0.0154,
      "step": 8050
    },
    {
      "epoch": 0.051694336391594216,
      "grad_norm": 0.8554036021232605,
      "learning_rate": 2.5846587993843e-06,
      "loss": 0.0194,
      "step": 8060
    },
    {
      "epoch": 0.051758473285380316,
      "grad_norm": 1.0235211849212646,
      "learning_rate": 2.5878655720882505e-06,
      "loss": 0.0138,
      "step": 8070
    },
    {
      "epoch": 0.051822610179166416,
      "grad_norm": 0.93379807472229,
      "learning_rate": 2.5910723447922017e-06,
      "loss": 0.0143,
      "step": 8080
    },
    {
      "epoch": 0.05188674707295251,
      "grad_norm": 0.8252114057540894,
      "learning_rate": 2.5942791174961523e-06,
      "loss": 0.018,
      "step": 8090
    },
    {
      "epoch": 0.05195088396673861,
      "grad_norm": 0.7677938342094421,
      "learning_rate": 2.5974858902001026e-06,
      "loss": 0.0117,
      "step": 8100
    },
    {
      "epoch": 0.0520150208605247,
      "grad_norm": 1.0340348482131958,
      "learning_rate": 2.6006926629040537e-06,
      "loss": 0.0165,
      "step": 8110
    },
    {
      "epoch": 0.0520791577543108,
      "grad_norm": 0.7640451192855835,
      "learning_rate": 2.6038994356080044e-06,
      "loss": 0.0188,
      "step": 8120
    },
    {
      "epoch": 0.052143294648096895,
      "grad_norm": 0.7858720421791077,
      "learning_rate": 2.607106208311955e-06,
      "loss": 0.0168,
      "step": 8130
    },
    {
      "epoch": 0.052207431541882995,
      "grad_norm": 0.6935146450996399,
      "learning_rate": 2.610312981015906e-06,
      "loss": 0.017,
      "step": 8140
    },
    {
      "epoch": 0.052271568435669095,
      "grad_norm": 0.6653669476509094,
      "learning_rate": 2.6135197537198565e-06,
      "loss": 0.018,
      "step": 8150
    },
    {
      "epoch": 0.05233570532945519,
      "grad_norm": 0.853435754776001,
      "learning_rate": 2.616726526423807e-06,
      "loss": 0.0139,
      "step": 8160
    },
    {
      "epoch": 0.05239984222324129,
      "grad_norm": 1.3172193765640259,
      "learning_rate": 2.6199332991277583e-06,
      "loss": 0.0171,
      "step": 8170
    },
    {
      "epoch": 0.05246397911702738,
      "grad_norm": 0.8625347018241882,
      "learning_rate": 2.623140071831709e-06,
      "loss": 0.0136,
      "step": 8180
    },
    {
      "epoch": 0.05252811601081348,
      "grad_norm": 1.255388617515564,
      "learning_rate": 2.6263468445356592e-06,
      "loss": 0.0157,
      "step": 8190
    },
    {
      "epoch": 0.052592252904599575,
      "grad_norm": 0.9203137159347534,
      "learning_rate": 2.6295536172396104e-06,
      "loss": 0.0158,
      "step": 8200
    },
    {
      "epoch": 0.052656389798385675,
      "grad_norm": 1.5580936670303345,
      "learning_rate": 2.632760389943561e-06,
      "loss": 0.0172,
      "step": 8210
    },
    {
      "epoch": 0.05272052669217177,
      "grad_norm": 0.9016255140304565,
      "learning_rate": 2.635967162647512e-06,
      "loss": 0.0173,
      "step": 8220
    },
    {
      "epoch": 0.05278466358595787,
      "grad_norm": 0.9950858950614929,
      "learning_rate": 2.6391739353514624e-06,
      "loss": 0.0167,
      "step": 8230
    },
    {
      "epoch": 0.05284880047974397,
      "grad_norm": 0.932772159576416,
      "learning_rate": 2.642380708055413e-06,
      "loss": 0.0156,
      "step": 8240
    },
    {
      "epoch": 0.05291293737353006,
      "grad_norm": 1.7568084001541138,
      "learning_rate": 2.645587480759364e-06,
      "loss": 0.0186,
      "step": 8250
    },
    {
      "epoch": 0.05297707426731616,
      "grad_norm": 0.7419828176498413,
      "learning_rate": 2.648794253463315e-06,
      "loss": 0.0122,
      "step": 8260
    },
    {
      "epoch": 0.053041211161102254,
      "grad_norm": 0.6230597496032715,
      "learning_rate": 2.6520010261672656e-06,
      "loss": 0.0108,
      "step": 8270
    },
    {
      "epoch": 0.053105348054888354,
      "grad_norm": 0.6508496999740601,
      "learning_rate": 2.655207798871216e-06,
      "loss": 0.0157,
      "step": 8280
    },
    {
      "epoch": 0.05316948494867445,
      "grad_norm": 0.5818604826927185,
      "learning_rate": 2.658414571575167e-06,
      "loss": 0.0222,
      "step": 8290
    },
    {
      "epoch": 0.05323362184246055,
      "grad_norm": 0.8822751641273499,
      "learning_rate": 2.6616213442791177e-06,
      "loss": 0.0178,
      "step": 8300
    },
    {
      "epoch": 0.05329775873624665,
      "grad_norm": 1.0668436288833618,
      "learning_rate": 2.6648281169830688e-06,
      "loss": 0.0123,
      "step": 8310
    },
    {
      "epoch": 0.05336189563003274,
      "grad_norm": 0.8772680759429932,
      "learning_rate": 2.668034889687019e-06,
      "loss": 0.0173,
      "step": 8320
    },
    {
      "epoch": 0.05342603252381884,
      "grad_norm": 1.3241474628448486,
      "learning_rate": 2.6712416623909697e-06,
      "loss": 0.0198,
      "step": 8330
    },
    {
      "epoch": 0.053490169417604934,
      "grad_norm": 1.102897047996521,
      "learning_rate": 2.674448435094921e-06,
      "loss": 0.0154,
      "step": 8340
    },
    {
      "epoch": 0.053554306311391034,
      "grad_norm": 0.7951586842536926,
      "learning_rate": 2.6776552077988715e-06,
      "loss": 0.0171,
      "step": 8350
    },
    {
      "epoch": 0.05361844320517713,
      "grad_norm": 1.398962378501892,
      "learning_rate": 2.6808619805028222e-06,
      "loss": 0.0138,
      "step": 8360
    },
    {
      "epoch": 0.05368258009896323,
      "grad_norm": 1.0954784154891968,
      "learning_rate": 2.684068753206773e-06,
      "loss": 0.0173,
      "step": 8370
    },
    {
      "epoch": 0.05374671699274933,
      "grad_norm": 0.9455815553665161,
      "learning_rate": 2.6872755259107236e-06,
      "loss": 0.0165,
      "step": 8380
    },
    {
      "epoch": 0.05381085388653542,
      "grad_norm": 0.6192914843559265,
      "learning_rate": 2.6904822986146743e-06,
      "loss": 0.0127,
      "step": 8390
    },
    {
      "epoch": 0.05387499078032152,
      "grad_norm": 0.47157925367355347,
      "learning_rate": 2.6936890713186254e-06,
      "loss": 0.0136,
      "step": 8400
    },
    {
      "epoch": 0.05393912767410761,
      "grad_norm": 1.3008977174758911,
      "learning_rate": 2.6968958440225757e-06,
      "loss": 0.0172,
      "step": 8410
    },
    {
      "epoch": 0.05400326456789371,
      "grad_norm": 0.40314775705337524,
      "learning_rate": 2.7001026167265264e-06,
      "loss": 0.0177,
      "step": 8420
    },
    {
      "epoch": 0.05406740146167981,
      "grad_norm": 0.7646990418434143,
      "learning_rate": 2.7033093894304775e-06,
      "loss": 0.0151,
      "step": 8430
    },
    {
      "epoch": 0.05413153835546591,
      "grad_norm": 0.6532096862792969,
      "learning_rate": 2.706516162134428e-06,
      "loss": 0.0137,
      "step": 8440
    },
    {
      "epoch": 0.05419567524925201,
      "grad_norm": 0.6426984071731567,
      "learning_rate": 2.7097229348383793e-06,
      "loss": 0.016,
      "step": 8450
    },
    {
      "epoch": 0.0542598121430381,
      "grad_norm": 0.6302536725997925,
      "learning_rate": 2.7129297075423296e-06,
      "loss": 0.0216,
      "step": 8460
    },
    {
      "epoch": 0.0543239490368242,
      "grad_norm": 0.5971202850341797,
      "learning_rate": 2.7161364802462802e-06,
      "loss": 0.0155,
      "step": 8470
    },
    {
      "epoch": 0.05438808593061029,
      "grad_norm": 0.7935372591018677,
      "learning_rate": 2.7193432529502314e-06,
      "loss": 0.0181,
      "step": 8480
    },
    {
      "epoch": 0.05445222282439639,
      "grad_norm": 0.7118707299232483,
      "learning_rate": 2.722550025654182e-06,
      "loss": 0.0173,
      "step": 8490
    },
    {
      "epoch": 0.054516359718182486,
      "grad_norm": 0.8970438241958618,
      "learning_rate": 2.7257567983581323e-06,
      "loss": 0.0162,
      "step": 8500
    },
    {
      "epoch": 0.054580496611968586,
      "grad_norm": 1.0925875902175903,
      "learning_rate": 2.728963571062083e-06,
      "loss": 0.017,
      "step": 8510
    },
    {
      "epoch": 0.054644633505754686,
      "grad_norm": 1.1510151624679565,
      "learning_rate": 2.732170343766034e-06,
      "loss": 0.0117,
      "step": 8520
    },
    {
      "epoch": 0.05470877039954078,
      "grad_norm": 0.77836674451828,
      "learning_rate": 2.735377116469985e-06,
      "loss": 0.0126,
      "step": 8530
    },
    {
      "epoch": 0.05477290729332688,
      "grad_norm": 0.48642775416374207,
      "learning_rate": 2.738583889173936e-06,
      "loss": 0.0168,
      "step": 8540
    },
    {
      "epoch": 0.05483704418711297,
      "grad_norm": 0.5595241189002991,
      "learning_rate": 2.741790661877886e-06,
      "loss": 0.0112,
      "step": 8550
    },
    {
      "epoch": 0.05490118108089907,
      "grad_norm": 0.9097594618797302,
      "learning_rate": 2.744997434581837e-06,
      "loss": 0.0111,
      "step": 8560
    },
    {
      "epoch": 0.054965317974685166,
      "grad_norm": 1.075947880744934,
      "learning_rate": 2.748204207285788e-06,
      "loss": 0.0155,
      "step": 8570
    },
    {
      "epoch": 0.055029454868471266,
      "grad_norm": 0.7200561165809631,
      "learning_rate": 2.7514109799897387e-06,
      "loss": 0.0168,
      "step": 8580
    },
    {
      "epoch": 0.05509359176225736,
      "grad_norm": 1.1899163722991943,
      "learning_rate": 2.754617752693689e-06,
      "loss": 0.0228,
      "step": 8590
    },
    {
      "epoch": 0.05515772865604346,
      "grad_norm": 0.45061996579170227,
      "learning_rate": 2.75782452539764e-06,
      "loss": 0.0154,
      "step": 8600
    },
    {
      "epoch": 0.05522186554982956,
      "grad_norm": 0.6197959184646606,
      "learning_rate": 2.7610312981015907e-06,
      "loss": 0.0134,
      "step": 8610
    },
    {
      "epoch": 0.05528600244361565,
      "grad_norm": 1.1750388145446777,
      "learning_rate": 2.7642380708055414e-06,
      "loss": 0.0208,
      "step": 8620
    },
    {
      "epoch": 0.05535013933740175,
      "grad_norm": 0.905691921710968,
      "learning_rate": 2.7674448435094925e-06,
      "loss": 0.0189,
      "step": 8630
    },
    {
      "epoch": 0.055414276231187845,
      "grad_norm": 0.9650370478630066,
      "learning_rate": 2.770651616213443e-06,
      "loss": 0.0203,
      "step": 8640
    },
    {
      "epoch": 0.055478413124973945,
      "grad_norm": 0.8335312604904175,
      "learning_rate": 2.7738583889173935e-06,
      "loss": 0.0107,
      "step": 8650
    },
    {
      "epoch": 0.05554255001876004,
      "grad_norm": 1.6039456129074097,
      "learning_rate": 2.7770651616213446e-06,
      "loss": 0.0221,
      "step": 8660
    },
    {
      "epoch": 0.05560668691254614,
      "grad_norm": 0.6446152925491333,
      "learning_rate": 2.7802719343252953e-06,
      "loss": 0.0137,
      "step": 8670
    },
    {
      "epoch": 0.05567082380633224,
      "grad_norm": 0.7407627701759338,
      "learning_rate": 2.7834787070292456e-06,
      "loss": 0.0217,
      "step": 8680
    },
    {
      "epoch": 0.05573496070011833,
      "grad_norm": 0.4696844518184662,
      "learning_rate": 2.7866854797331967e-06,
      "loss": 0.0113,
      "step": 8690
    },
    {
      "epoch": 0.05579909759390443,
      "grad_norm": 0.8112123012542725,
      "learning_rate": 2.7898922524371474e-06,
      "loss": 0.0192,
      "step": 8700
    },
    {
      "epoch": 0.055863234487690525,
      "grad_norm": 0.8872155547142029,
      "learning_rate": 2.7930990251410985e-06,
      "loss": 0.0194,
      "step": 8710
    },
    {
      "epoch": 0.055927371381476625,
      "grad_norm": 0.4063515365123749,
      "learning_rate": 2.796305797845049e-06,
      "loss": 0.0194,
      "step": 8720
    },
    {
      "epoch": 0.05599150827526272,
      "grad_norm": 0.9252082705497742,
      "learning_rate": 2.7995125705489994e-06,
      "loss": 0.0178,
      "step": 8730
    },
    {
      "epoch": 0.05605564516904882,
      "grad_norm": 1.4230228662490845,
      "learning_rate": 2.8027193432529506e-06,
      "loss": 0.015,
      "step": 8740
    },
    {
      "epoch": 0.05611978206283492,
      "grad_norm": 0.4911647439002991,
      "learning_rate": 2.8059261159569012e-06,
      "loss": 0.0159,
      "step": 8750
    },
    {
      "epoch": 0.05618391895662101,
      "grad_norm": 0.9393802881240845,
      "learning_rate": 2.809132888660852e-06,
      "loss": 0.0204,
      "step": 8760
    },
    {
      "epoch": 0.05624805585040711,
      "grad_norm": 1.24632728099823,
      "learning_rate": 2.812339661364803e-06,
      "loss": 0.0177,
      "step": 8770
    },
    {
      "epoch": 0.056312192744193204,
      "grad_norm": 1.1230436563491821,
      "learning_rate": 2.8155464340687533e-06,
      "loss": 0.0217,
      "step": 8780
    },
    {
      "epoch": 0.056376329637979304,
      "grad_norm": 0.7382339239120483,
      "learning_rate": 2.818753206772704e-06,
      "loss": 0.0141,
      "step": 8790
    },
    {
      "epoch": 0.0564404665317654,
      "grad_norm": 1.1875547170639038,
      "learning_rate": 2.821959979476655e-06,
      "loss": 0.0145,
      "step": 8800
    },
    {
      "epoch": 0.0565046034255515,
      "grad_norm": 0.7362310290336609,
      "learning_rate": 2.825166752180606e-06,
      "loss": 0.0158,
      "step": 8810
    },
    {
      "epoch": 0.0565687403193376,
      "grad_norm": 0.9873781800270081,
      "learning_rate": 2.828373524884556e-06,
      "loss": 0.01,
      "step": 8820
    },
    {
      "epoch": 0.05663287721312369,
      "grad_norm": 0.7471084594726562,
      "learning_rate": 2.831580297588507e-06,
      "loss": 0.0145,
      "step": 8830
    },
    {
      "epoch": 0.05669701410690979,
      "grad_norm": 0.8579428791999817,
      "learning_rate": 2.834787070292458e-06,
      "loss": 0.0176,
      "step": 8840
    },
    {
      "epoch": 0.056761151000695884,
      "grad_norm": 0.5471231341362,
      "learning_rate": 2.837993842996409e-06,
      "loss": 0.016,
      "step": 8850
    },
    {
      "epoch": 0.056825287894481984,
      "grad_norm": 0.9654415845870972,
      "learning_rate": 2.8412006157003597e-06,
      "loss": 0.0157,
      "step": 8860
    },
    {
      "epoch": 0.05688942478826808,
      "grad_norm": 0.9503028392791748,
      "learning_rate": 2.84440738840431e-06,
      "loss": 0.0147,
      "step": 8870
    },
    {
      "epoch": 0.05695356168205418,
      "grad_norm": 1.0068310499191284,
      "learning_rate": 2.8476141611082606e-06,
      "loss": 0.0099,
      "step": 8880
    },
    {
      "epoch": 0.05701769857584028,
      "grad_norm": 0.6317100524902344,
      "learning_rate": 2.8508209338122117e-06,
      "loss": 0.0128,
      "step": 8890
    },
    {
      "epoch": 0.05708183546962637,
      "grad_norm": 0.96009361743927,
      "learning_rate": 2.8540277065161624e-06,
      "loss": 0.018,
      "step": 8900
    },
    {
      "epoch": 0.05714597236341247,
      "grad_norm": 0.7821346521377563,
      "learning_rate": 2.8572344792201127e-06,
      "loss": 0.0136,
      "step": 8910
    },
    {
      "epoch": 0.05721010925719856,
      "grad_norm": 0.7975736260414124,
      "learning_rate": 2.860441251924064e-06,
      "loss": 0.0219,
      "step": 8920
    },
    {
      "epoch": 0.05727424615098466,
      "grad_norm": 0.9637255072593689,
      "learning_rate": 2.8636480246280145e-06,
      "loss": 0.019,
      "step": 8930
    },
    {
      "epoch": 0.057338383044770756,
      "grad_norm": 1.2077337503433228,
      "learning_rate": 2.8668547973319656e-06,
      "loss": 0.012,
      "step": 8940
    },
    {
      "epoch": 0.057402519938556856,
      "grad_norm": 0.8413091897964478,
      "learning_rate": 2.8700615700359163e-06,
      "loss": 0.0136,
      "step": 8950
    },
    {
      "epoch": 0.05746665683234295,
      "grad_norm": 0.570277750492096,
      "learning_rate": 2.8732683427398666e-06,
      "loss": 0.0122,
      "step": 8960
    },
    {
      "epoch": 0.05753079372612905,
      "grad_norm": 1.1096901893615723,
      "learning_rate": 2.8764751154438177e-06,
      "loss": 0.0133,
      "step": 8970
    },
    {
      "epoch": 0.05759493061991515,
      "grad_norm": 0.5581759214401245,
      "learning_rate": 2.8796818881477684e-06,
      "loss": 0.013,
      "step": 8980
    },
    {
      "epoch": 0.05765906751370124,
      "grad_norm": 0.8480189442634583,
      "learning_rate": 2.882888660851719e-06,
      "loss": 0.0161,
      "step": 8990
    },
    {
      "epoch": 0.05772320440748734,
      "grad_norm": 1.3270263671875,
      "learning_rate": 2.8860954335556698e-06,
      "loss": 0.0144,
      "step": 9000
    },
    {
      "epoch": 0.057787341301273436,
      "grad_norm": 1.3251687288284302,
      "learning_rate": 2.8893022062596204e-06,
      "loss": 0.0166,
      "step": 9010
    },
    {
      "epoch": 0.057851478195059536,
      "grad_norm": 0.2769591212272644,
      "learning_rate": 2.892508978963571e-06,
      "loss": 0.0167,
      "step": 9020
    },
    {
      "epoch": 0.05791561508884563,
      "grad_norm": 0.5797544121742249,
      "learning_rate": 2.8957157516675222e-06,
      "loss": 0.0174,
      "step": 9030
    },
    {
      "epoch": 0.05797975198263173,
      "grad_norm": 0.8181819319725037,
      "learning_rate": 2.898922524371473e-06,
      "loss": 0.0233,
      "step": 9040
    },
    {
      "epoch": 0.05804388887641783,
      "grad_norm": 0.547584593296051,
      "learning_rate": 2.902129297075423e-06,
      "loss": 0.0113,
      "step": 9050
    },
    {
      "epoch": 0.05810802577020392,
      "grad_norm": 1.119734525680542,
      "learning_rate": 2.9053360697793743e-06,
      "loss": 0.0111,
      "step": 9060
    },
    {
      "epoch": 0.05817216266399002,
      "grad_norm": 0.8142489790916443,
      "learning_rate": 2.908542842483325e-06,
      "loss": 0.0146,
      "step": 9070
    },
    {
      "epoch": 0.058236299557776115,
      "grad_norm": 0.8385538458824158,
      "learning_rate": 2.911749615187276e-06,
      "loss": 0.013,
      "step": 9080
    },
    {
      "epoch": 0.058300436451562215,
      "grad_norm": 0.585768461227417,
      "learning_rate": 2.9149563878912264e-06,
      "loss": 0.0129,
      "step": 9090
    },
    {
      "epoch": 0.05836457334534831,
      "grad_norm": 0.6681596636772156,
      "learning_rate": 2.918163160595177e-06,
      "loss": 0.0105,
      "step": 9100
    },
    {
      "epoch": 0.05842871023913441,
      "grad_norm": 0.8642899394035339,
      "learning_rate": 2.921369933299128e-06,
      "loss": 0.0193,
      "step": 9110
    },
    {
      "epoch": 0.05849284713292051,
      "grad_norm": 1.0386979579925537,
      "learning_rate": 2.924576706003079e-06,
      "loss": 0.0158,
      "step": 9120
    },
    {
      "epoch": 0.0585569840267066,
      "grad_norm": 0.6511367559432983,
      "learning_rate": 2.9277834787070296e-06,
      "loss": 0.0154,
      "step": 9130
    },
    {
      "epoch": 0.0586211209204927,
      "grad_norm": 0.3735872805118561,
      "learning_rate": 2.93099025141098e-06,
      "loss": 0.0094,
      "step": 9140
    },
    {
      "epoch": 0.058685257814278795,
      "grad_norm": 0.6131263375282288,
      "learning_rate": 2.934197024114931e-06,
      "loss": 0.015,
      "step": 9150
    },
    {
      "epoch": 0.058749394708064895,
      "grad_norm": 0.6784162521362305,
      "learning_rate": 2.9374037968188816e-06,
      "loss": 0.0117,
      "step": 9160
    },
    {
      "epoch": 0.05881353160185099,
      "grad_norm": 0.806260347366333,
      "learning_rate": 2.9406105695228327e-06,
      "loss": 0.0163,
      "step": 9170
    },
    {
      "epoch": 0.05887766849563709,
      "grad_norm": 1.063302993774414,
      "learning_rate": 2.943817342226783e-06,
      "loss": 0.0142,
      "step": 9180
    },
    {
      "epoch": 0.05894180538942319,
      "grad_norm": 0.3902188539505005,
      "learning_rate": 2.9470241149307337e-06,
      "loss": 0.0138,
      "step": 9190
    },
    {
      "epoch": 0.05900594228320928,
      "grad_norm": 0.5390955805778503,
      "learning_rate": 2.950230887634685e-06,
      "loss": 0.0143,
      "step": 9200
    },
    {
      "epoch": 0.05907007917699538,
      "grad_norm": 0.5154760479927063,
      "learning_rate": 2.9534376603386355e-06,
      "loss": 0.0154,
      "step": 9210
    },
    {
      "epoch": 0.059134216070781474,
      "grad_norm": 0.9843395948410034,
      "learning_rate": 2.9566444330425866e-06,
      "loss": 0.0176,
      "step": 9220
    },
    {
      "epoch": 0.059198352964567574,
      "grad_norm": 0.8421329855918884,
      "learning_rate": 2.959851205746537e-06,
      "loss": 0.0159,
      "step": 9230
    },
    {
      "epoch": 0.05926248985835367,
      "grad_norm": 0.8714247345924377,
      "learning_rate": 2.9630579784504876e-06,
      "loss": 0.0177,
      "step": 9240
    },
    {
      "epoch": 0.05932662675213977,
      "grad_norm": 0.9572415351867676,
      "learning_rate": 2.9662647511544383e-06,
      "loss": 0.0159,
      "step": 9250
    },
    {
      "epoch": 0.05939076364592587,
      "grad_norm": 0.38546404242515564,
      "learning_rate": 2.9694715238583894e-06,
      "loss": 0.0132,
      "step": 9260
    },
    {
      "epoch": 0.05945490053971196,
      "grad_norm": 0.9735164046287537,
      "learning_rate": 2.9726782965623396e-06,
      "loss": 0.011,
      "step": 9270
    },
    {
      "epoch": 0.05951903743349806,
      "grad_norm": 0.7716529965400696,
      "learning_rate": 2.9758850692662903e-06,
      "loss": 0.0162,
      "step": 9280
    },
    {
      "epoch": 0.059583174327284154,
      "grad_norm": 0.7090407013893127,
      "learning_rate": 2.9790918419702414e-06,
      "loss": 0.015,
      "step": 9290
    },
    {
      "epoch": 0.059647311221070254,
      "grad_norm": 0.7737558484077454,
      "learning_rate": 2.982298614674192e-06,
      "loss": 0.0131,
      "step": 9300
    },
    {
      "epoch": 0.05971144811485635,
      "grad_norm": 0.61954665184021,
      "learning_rate": 2.9855053873781432e-06,
      "loss": 0.0171,
      "step": 9310
    },
    {
      "epoch": 0.05977558500864245,
      "grad_norm": 0.8703621625900269,
      "learning_rate": 2.9887121600820935e-06,
      "loss": 0.0161,
      "step": 9320
    },
    {
      "epoch": 0.05983972190242854,
      "grad_norm": 0.6462575793266296,
      "learning_rate": 2.991918932786044e-06,
      "loss": 0.0083,
      "step": 9330
    },
    {
      "epoch": 0.05990385879621464,
      "grad_norm": 0.275501549243927,
      "learning_rate": 2.9951257054899953e-06,
      "loss": 0.0109,
      "step": 9340
    },
    {
      "epoch": 0.05996799569000074,
      "grad_norm": 1.0189841985702515,
      "learning_rate": 2.998332478193946e-06,
      "loss": 0.014,
      "step": 9350
    },
    {
      "epoch": 0.06003213258378683,
      "grad_norm": 1.032149076461792,
      "learning_rate": 3.0015392508978963e-06,
      "loss": 0.0169,
      "step": 9360
    },
    {
      "epoch": 0.06009626947757293,
      "grad_norm": 0.9786964654922485,
      "learning_rate": 3.0047460236018474e-06,
      "loss": 0.0165,
      "step": 9370
    },
    {
      "epoch": 0.060160406371359026,
      "grad_norm": 0.7089354991912842,
      "learning_rate": 3.007952796305798e-06,
      "loss": 0.013,
      "step": 9380
    },
    {
      "epoch": 0.060224543265145127,
      "grad_norm": 0.9810829162597656,
      "learning_rate": 3.0111595690097488e-06,
      "loss": 0.0143,
      "step": 9390
    },
    {
      "epoch": 0.06028868015893122,
      "grad_norm": 1.4317636489868164,
      "learning_rate": 3.0143663417137e-06,
      "loss": 0.0158,
      "step": 9400
    },
    {
      "epoch": 0.06035281705271732,
      "grad_norm": 1.0124961137771606,
      "learning_rate": 3.01757311441765e-06,
      "loss": 0.0169,
      "step": 9410
    },
    {
      "epoch": 0.06041695394650342,
      "grad_norm": 0.7623961567878723,
      "learning_rate": 3.020779887121601e-06,
      "loss": 0.0181,
      "step": 9420
    },
    {
      "epoch": 0.06048109084028951,
      "grad_norm": 0.6873654127120972,
      "learning_rate": 3.023986659825552e-06,
      "loss": 0.0145,
      "step": 9430
    },
    {
      "epoch": 0.06054522773407561,
      "grad_norm": 0.7793513536453247,
      "learning_rate": 3.0271934325295026e-06,
      "loss": 0.0144,
      "step": 9440
    },
    {
      "epoch": 0.060609364627861706,
      "grad_norm": 0.5543414354324341,
      "learning_rate": 3.0304002052334537e-06,
      "loss": 0.0155,
      "step": 9450
    },
    {
      "epoch": 0.060673501521647806,
      "grad_norm": 0.34898149967193604,
      "learning_rate": 3.033606977937404e-06,
      "loss": 0.0144,
      "step": 9460
    },
    {
      "epoch": 0.0607376384154339,
      "grad_norm": 0.7438449263572693,
      "learning_rate": 3.0368137506413547e-06,
      "loss": 0.0207,
      "step": 9470
    },
    {
      "epoch": 0.06080177530922,
      "grad_norm": 0.5092421770095825,
      "learning_rate": 3.040020523345306e-06,
      "loss": 0.0124,
      "step": 9480
    },
    {
      "epoch": 0.0608659122030061,
      "grad_norm": 0.7169559597969055,
      "learning_rate": 3.0432272960492565e-06,
      "loss": 0.011,
      "step": 9490
    },
    {
      "epoch": 0.06093004909679219,
      "grad_norm": 1.0686838626861572,
      "learning_rate": 3.0464340687532068e-06,
      "loss": 0.015,
      "step": 9500
    },
    {
      "epoch": 0.06099418599057829,
      "grad_norm": 1.1948643922805786,
      "learning_rate": 3.0496408414571575e-06,
      "loss": 0.019,
      "step": 9510
    },
    {
      "epoch": 0.061058322884364385,
      "grad_norm": 0.9161044359207153,
      "learning_rate": 3.0528476141611086e-06,
      "loss": 0.0152,
      "step": 9520
    },
    {
      "epoch": 0.061122459778150486,
      "grad_norm": 0.5058609843254089,
      "learning_rate": 3.0560543868650593e-06,
      "loss": 0.0132,
      "step": 9530
    },
    {
      "epoch": 0.06118659667193658,
      "grad_norm": 1.7881057262420654,
      "learning_rate": 3.0592611595690104e-06,
      "loss": 0.0128,
      "step": 9540
    },
    {
      "epoch": 0.06125073356572268,
      "grad_norm": 0.7522915005683899,
      "learning_rate": 3.0624679322729606e-06,
      "loss": 0.0193,
      "step": 9550
    },
    {
      "epoch": 0.06131487045950878,
      "grad_norm": 1.0016465187072754,
      "learning_rate": 3.0656747049769113e-06,
      "loss": 0.0142,
      "step": 9560
    },
    {
      "epoch": 0.06137900735329487,
      "grad_norm": 0.5280004739761353,
      "learning_rate": 3.0688814776808624e-06,
      "loss": 0.016,
      "step": 9570
    },
    {
      "epoch": 0.06144314424708097,
      "grad_norm": 0.850499153137207,
      "learning_rate": 3.072088250384813e-06,
      "loss": 0.0147,
      "step": 9580
    },
    {
      "epoch": 0.061507281140867065,
      "grad_norm": 0.8008180856704712,
      "learning_rate": 3.0752950230887634e-06,
      "loss": 0.0084,
      "step": 9590
    },
    {
      "epoch": 0.061571418034653165,
      "grad_norm": 0.9171088337898254,
      "learning_rate": 3.0785017957927145e-06,
      "loss": 0.0124,
      "step": 9600
    },
    {
      "epoch": 0.06163555492843926,
      "grad_norm": 0.8631775975227356,
      "learning_rate": 3.081708568496665e-06,
      "loss": 0.0124,
      "step": 9610
    },
    {
      "epoch": 0.06169969182222536,
      "grad_norm": 0.7938309907913208,
      "learning_rate": 3.084915341200616e-06,
      "loss": 0.0184,
      "step": 9620
    },
    {
      "epoch": 0.06176382871601146,
      "grad_norm": 1.385733962059021,
      "learning_rate": 3.088122113904567e-06,
      "loss": 0.0196,
      "step": 9630
    },
    {
      "epoch": 0.06182796560979755,
      "grad_norm": 1.151713252067566,
      "learning_rate": 3.0913288866085173e-06,
      "loss": 0.0177,
      "step": 9640
    },
    {
      "epoch": 0.06189210250358365,
      "grad_norm": 0.8653784394264221,
      "learning_rate": 3.094535659312468e-06,
      "loss": 0.02,
      "step": 9650
    },
    {
      "epoch": 0.061956239397369745,
      "grad_norm": 0.7082491517066956,
      "learning_rate": 3.097742432016419e-06,
      "loss": 0.0186,
      "step": 9660
    },
    {
      "epoch": 0.062020376291155845,
      "grad_norm": 1.227051854133606,
      "learning_rate": 3.1009492047203698e-06,
      "loss": 0.0119,
      "step": 9670
    },
    {
      "epoch": 0.06208451318494194,
      "grad_norm": 0.8086697459220886,
      "learning_rate": 3.10415597742432e-06,
      "loss": 0.0126,
      "step": 9680
    },
    {
      "epoch": 0.06214865007872804,
      "grad_norm": 0.7787764072418213,
      "learning_rate": 3.107362750128271e-06,
      "loss": 0.0108,
      "step": 9690
    },
    {
      "epoch": 0.06221278697251413,
      "grad_norm": 1.1162656545639038,
      "learning_rate": 3.110569522832222e-06,
      "loss": 0.0135,
      "step": 9700
    },
    {
      "epoch": 0.06227692386630023,
      "grad_norm": 0.9261019229888916,
      "learning_rate": 3.113776295536173e-06,
      "loss": 0.0122,
      "step": 9710
    },
    {
      "epoch": 0.06234106076008633,
      "grad_norm": 0.7241477370262146,
      "learning_rate": 3.1169830682401236e-06,
      "loss": 0.0154,
      "step": 9720
    },
    {
      "epoch": 0.062405197653872424,
      "grad_norm": 0.6688586473464966,
      "learning_rate": 3.120189840944074e-06,
      "loss": 0.014,
      "step": 9730
    },
    {
      "epoch": 0.062469334547658524,
      "grad_norm": 0.9395526051521301,
      "learning_rate": 3.123396613648025e-06,
      "loss": 0.0115,
      "step": 9740
    },
    {
      "epoch": 0.06253347144144462,
      "grad_norm": 1.0174471139907837,
      "learning_rate": 3.1266033863519757e-06,
      "loss": 0.0148,
      "step": 9750
    },
    {
      "epoch": 0.06259760833523072,
      "grad_norm": 0.7402105927467346,
      "learning_rate": 3.1298101590559264e-06,
      "loss": 0.0154,
      "step": 9760
    },
    {
      "epoch": 0.06266174522901681,
      "grad_norm": 0.7139943242073059,
      "learning_rate": 3.1330169317598767e-06,
      "loss": 0.0156,
      "step": 9770
    },
    {
      "epoch": 0.06272588212280292,
      "grad_norm": 0.7116230726242065,
      "learning_rate": 3.1362237044638278e-06,
      "loss": 0.0159,
      "step": 9780
    },
    {
      "epoch": 0.06279001901658901,
      "grad_norm": 1.0926690101623535,
      "learning_rate": 3.1394304771677785e-06,
      "loss": 0.0123,
      "step": 9790
    },
    {
      "epoch": 0.0628541559103751,
      "grad_norm": 0.7361863851547241,
      "learning_rate": 3.1426372498717296e-06,
      "loss": 0.0119,
      "step": 9800
    },
    {
      "epoch": 0.0629182928041612,
      "grad_norm": 1.3561779260635376,
      "learning_rate": 3.1458440225756803e-06,
      "loss": 0.0155,
      "step": 9810
    },
    {
      "epoch": 0.0629824296979473,
      "grad_norm": 1.122456431388855,
      "learning_rate": 3.1490507952796305e-06,
      "loss": 0.0176,
      "step": 9820
    },
    {
      "epoch": 0.0630465665917334,
      "grad_norm": 1.144529938697815,
      "learning_rate": 3.1522575679835816e-06,
      "loss": 0.0144,
      "step": 9830
    },
    {
      "epoch": 0.06311070348551949,
      "grad_norm": 1.3597661256790161,
      "learning_rate": 3.1554643406875323e-06,
      "loss": 0.0171,
      "step": 9840
    },
    {
      "epoch": 0.06317484037930558,
      "grad_norm": 0.7596487402915955,
      "learning_rate": 3.1586711133914834e-06,
      "loss": 0.0159,
      "step": 9850
    },
    {
      "epoch": 0.06323897727309169,
      "grad_norm": 0.6753873825073242,
      "learning_rate": 3.1618778860954337e-06,
      "loss": 0.016,
      "step": 9860
    },
    {
      "epoch": 0.06330311416687778,
      "grad_norm": 0.7201382517814636,
      "learning_rate": 3.1650846587993844e-06,
      "loss": 0.0122,
      "step": 9870
    },
    {
      "epoch": 0.06336725106066388,
      "grad_norm": 0.5358558297157288,
      "learning_rate": 3.168291431503335e-06,
      "loss": 0.0162,
      "step": 9880
    },
    {
      "epoch": 0.06343138795444998,
      "grad_norm": 0.7316409945487976,
      "learning_rate": 3.171498204207286e-06,
      "loss": 0.0127,
      "step": 9890
    },
    {
      "epoch": 0.06349552484823608,
      "grad_norm": 0.6747068166732788,
      "learning_rate": 3.174704976911237e-06,
      "loss": 0.0134,
      "step": 9900
    },
    {
      "epoch": 0.06355966174202217,
      "grad_norm": 0.730777382850647,
      "learning_rate": 3.177911749615187e-06,
      "loss": 0.0156,
      "step": 9910
    },
    {
      "epoch": 0.06362379863580826,
      "grad_norm": 0.7868751287460327,
      "learning_rate": 3.1811185223191383e-06,
      "loss": 0.0126,
      "step": 9920
    },
    {
      "epoch": 0.06368793552959437,
      "grad_norm": 1.0477957725524902,
      "learning_rate": 3.184325295023089e-06,
      "loss": 0.018,
      "step": 9930
    },
    {
      "epoch": 0.06375207242338046,
      "grad_norm": 0.6357901692390442,
      "learning_rate": 3.18753206772704e-06,
      "loss": 0.0156,
      "step": 9940
    },
    {
      "epoch": 0.06381620931716656,
      "grad_norm": 1.0937377214431763,
      "learning_rate": 3.1907388404309903e-06,
      "loss": 0.0124,
      "step": 9950
    },
    {
      "epoch": 0.06388034621095266,
      "grad_norm": 0.5854929089546204,
      "learning_rate": 3.193945613134941e-06,
      "loss": 0.0166,
      "step": 9960
    },
    {
      "epoch": 0.06394448310473876,
      "grad_norm": 0.7687313556671143,
      "learning_rate": 3.197152385838892e-06,
      "loss": 0.0194,
      "step": 9970
    },
    {
      "epoch": 0.06400861999852485,
      "grad_norm": 0.7688573002815247,
      "learning_rate": 3.200359158542843e-06,
      "loss": 0.0131,
      "step": 9980
    },
    {
      "epoch": 0.06407275689231094,
      "grad_norm": 1.1176494359970093,
      "learning_rate": 3.2035659312467935e-06,
      "loss": 0.016,
      "step": 9990
    },
    {
      "epoch": 0.06413689378609705,
      "grad_norm": 0.5380460023880005,
      "learning_rate": 3.2067727039507442e-06,
      "loss": 0.0192,
      "step": 10000
    },
    {
      "epoch": 0.06420103067988314,
      "grad_norm": 0.764922559261322,
      "learning_rate": 3.209979476654695e-06,
      "loss": 0.0149,
      "step": 10010
    },
    {
      "epoch": 0.06426516757366924,
      "grad_norm": 0.9032078981399536,
      "learning_rate": 3.2131862493586456e-06,
      "loss": 0.0104,
      "step": 10020
    },
    {
      "epoch": 0.06432930446745534,
      "grad_norm": 0.678203284740448,
      "learning_rate": 3.2163930220625967e-06,
      "loss": 0.0155,
      "step": 10030
    },
    {
      "epoch": 0.06439344136124144,
      "grad_norm": 1.1898125410079956,
      "learning_rate": 3.219599794766547e-06,
      "loss": 0.0171,
      "step": 10040
    },
    {
      "epoch": 0.06445757825502753,
      "grad_norm": 1.0876600742340088,
      "learning_rate": 3.2228065674704977e-06,
      "loss": 0.0163,
      "step": 10050
    },
    {
      "epoch": 0.06452171514881362,
      "grad_norm": 0.40626418590545654,
      "learning_rate": 3.2260133401744488e-06,
      "loss": 0.0156,
      "step": 10060
    },
    {
      "epoch": 0.06458585204259973,
      "grad_norm": 0.5721704363822937,
      "learning_rate": 3.2292201128783995e-06,
      "loss": 0.0128,
      "step": 10070
    },
    {
      "epoch": 0.06464998893638582,
      "grad_norm": 1.0086840391159058,
      "learning_rate": 3.2324268855823506e-06,
      "loss": 0.0173,
      "step": 10080
    },
    {
      "epoch": 0.06471412583017191,
      "grad_norm": 0.7133669853210449,
      "learning_rate": 3.235633658286301e-06,
      "loss": 0.0164,
      "step": 10090
    },
    {
      "epoch": 0.06477826272395802,
      "grad_norm": 0.36082619428634644,
      "learning_rate": 3.2388404309902515e-06,
      "loss": 0.0138,
      "step": 10100
    },
    {
      "epoch": 0.06484239961774411,
      "grad_norm": 0.7135457992553711,
      "learning_rate": 3.2420472036942026e-06,
      "loss": 0.0128,
      "step": 10110
    },
    {
      "epoch": 0.06490653651153021,
      "grad_norm": 0.5620806217193604,
      "learning_rate": 3.2452539763981533e-06,
      "loss": 0.0114,
      "step": 10120
    },
    {
      "epoch": 0.0649706734053163,
      "grad_norm": 0.7917492389678955,
      "learning_rate": 3.248460749102104e-06,
      "loss": 0.0137,
      "step": 10130
    },
    {
      "epoch": 0.06503481029910241,
      "grad_norm": 0.8087460398674011,
      "learning_rate": 3.2516675218060543e-06,
      "loss": 0.0079,
      "step": 10140
    },
    {
      "epoch": 0.0650989471928885,
      "grad_norm": 0.7026078104972839,
      "learning_rate": 3.2548742945100054e-06,
      "loss": 0.0139,
      "step": 10150
    },
    {
      "epoch": 0.0651630840866746,
      "grad_norm": 0.4520063102245331,
      "learning_rate": 3.258081067213956e-06,
      "loss": 0.0125,
      "step": 10160
    },
    {
      "epoch": 0.0652272209804607,
      "grad_norm": 0.7053927183151245,
      "learning_rate": 3.261287839917907e-06,
      "loss": 0.014,
      "step": 10170
    },
    {
      "epoch": 0.0652913578742468,
      "grad_norm": 0.6258370280265808,
      "learning_rate": 3.2644946126218575e-06,
      "loss": 0.0128,
      "step": 10180
    },
    {
      "epoch": 0.06535549476803289,
      "grad_norm": 0.3676292598247528,
      "learning_rate": 3.267701385325808e-06,
      "loss": 0.0114,
      "step": 10190
    },
    {
      "epoch": 0.06541963166181898,
      "grad_norm": 0.8146242499351501,
      "learning_rate": 3.2709081580297593e-06,
      "loss": 0.0117,
      "step": 10200
    },
    {
      "epoch": 0.06548376855560509,
      "grad_norm": 0.7208275198936462,
      "learning_rate": 3.27411493073371e-06,
      "loss": 0.0102,
      "step": 10210
    },
    {
      "epoch": 0.06554790544939118,
      "grad_norm": 0.8345526456832886,
      "learning_rate": 3.277321703437661e-06,
      "loss": 0.0167,
      "step": 10220
    },
    {
      "epoch": 0.06561204234317727,
      "grad_norm": 0.7537758946418762,
      "learning_rate": 3.2805284761416113e-06,
      "loss": 0.0167,
      "step": 10230
    },
    {
      "epoch": 0.06567617923696338,
      "grad_norm": 1.0677050352096558,
      "learning_rate": 3.283735248845562e-06,
      "loss": 0.0152,
      "step": 10240
    },
    {
      "epoch": 0.06574031613074947,
      "grad_norm": 0.5786572098731995,
      "learning_rate": 3.2869420215495127e-06,
      "loss": 0.0128,
      "step": 10250
    },
    {
      "epoch": 0.06580445302453557,
      "grad_norm": 0.6964079737663269,
      "learning_rate": 3.290148794253464e-06,
      "loss": 0.0168,
      "step": 10260
    },
    {
      "epoch": 0.06586858991832166,
      "grad_norm": 0.7183060646057129,
      "learning_rate": 3.293355566957414e-06,
      "loss": 0.015,
      "step": 10270
    },
    {
      "epoch": 0.06593272681210777,
      "grad_norm": 0.7558525204658508,
      "learning_rate": 3.296562339661365e-06,
      "loss": 0.013,
      "step": 10280
    },
    {
      "epoch": 0.06599686370589386,
      "grad_norm": 0.6874273419380188,
      "learning_rate": 3.299769112365316e-06,
      "loss": 0.0135,
      "step": 10290
    },
    {
      "epoch": 0.06606100059967995,
      "grad_norm": 0.6457377672195435,
      "learning_rate": 3.3029758850692666e-06,
      "loss": 0.0117,
      "step": 10300
    },
    {
      "epoch": 0.06612513749346606,
      "grad_norm": 1.20498526096344,
      "learning_rate": 3.3061826577732177e-06,
      "loss": 0.0151,
      "step": 10310
    },
    {
      "epoch": 0.06618927438725215,
      "grad_norm": 0.7991650700569153,
      "learning_rate": 3.309389430477168e-06,
      "loss": 0.0171,
      "step": 10320
    },
    {
      "epoch": 0.06625341128103825,
      "grad_norm": 0.8451363444328308,
      "learning_rate": 3.3125962031811187e-06,
      "loss": 0.0136,
      "step": 10330
    },
    {
      "epoch": 0.06631754817482434,
      "grad_norm": 0.6329699158668518,
      "learning_rate": 3.3158029758850698e-06,
      "loss": 0.0135,
      "step": 10340
    },
    {
      "epoch": 0.06638168506861045,
      "grad_norm": 1.151140570640564,
      "learning_rate": 3.3190097485890205e-06,
      "loss": 0.0127,
      "step": 10350
    },
    {
      "epoch": 0.06644582196239654,
      "grad_norm": 0.7782281637191772,
      "learning_rate": 3.3222165212929707e-06,
      "loss": 0.0175,
      "step": 10360
    },
    {
      "epoch": 0.06650995885618263,
      "grad_norm": 0.5035085082054138,
      "learning_rate": 3.325423293996922e-06,
      "loss": 0.0108,
      "step": 10370
    },
    {
      "epoch": 0.06657409574996874,
      "grad_norm": 0.6519835591316223,
      "learning_rate": 3.3286300667008725e-06,
      "loss": 0.0133,
      "step": 10380
    },
    {
      "epoch": 0.06663823264375483,
      "grad_norm": 0.8411291241645813,
      "learning_rate": 3.3318368394048232e-06,
      "loss": 0.0133,
      "step": 10390
    },
    {
      "epoch": 0.06670236953754093,
      "grad_norm": 0.943038821220398,
      "learning_rate": 3.3350436121087743e-06,
      "loss": 0.0109,
      "step": 10400
    },
    {
      "epoch": 0.06676650643132702,
      "grad_norm": 0.9379919767379761,
      "learning_rate": 3.3382503848127246e-06,
      "loss": 0.014,
      "step": 10410
    },
    {
      "epoch": 0.06683064332511313,
      "grad_norm": 0.35660219192504883,
      "learning_rate": 3.3414571575166753e-06,
      "loss": 0.0145,
      "step": 10420
    },
    {
      "epoch": 0.06689478021889922,
      "grad_norm": 0.6724391579627991,
      "learning_rate": 3.3446639302206264e-06,
      "loss": 0.0147,
      "step": 10430
    },
    {
      "epoch": 0.06695891711268531,
      "grad_norm": 0.7699723839759827,
      "learning_rate": 3.347870702924577e-06,
      "loss": 0.0142,
      "step": 10440
    },
    {
      "epoch": 0.06702305400647142,
      "grad_norm": 0.9424741268157959,
      "learning_rate": 3.3510774756285274e-06,
      "loss": 0.0152,
      "step": 10450
    },
    {
      "epoch": 0.06708719090025751,
      "grad_norm": 1.0261833667755127,
      "learning_rate": 3.3542842483324785e-06,
      "loss": 0.017,
      "step": 10460
    },
    {
      "epoch": 0.0671513277940436,
      "grad_norm": 0.460065096616745,
      "learning_rate": 3.357491021036429e-06,
      "loss": 0.011,
      "step": 10470
    },
    {
      "epoch": 0.0672154646878297,
      "grad_norm": 0.33964595198631287,
      "learning_rate": 3.3606977937403803e-06,
      "loss": 0.0178,
      "step": 10480
    },
    {
      "epoch": 0.0672796015816158,
      "grad_norm": 0.44423216581344604,
      "learning_rate": 3.363904566444331e-06,
      "loss": 0.0125,
      "step": 10490
    },
    {
      "epoch": 0.0673437384754019,
      "grad_norm": 0.8569762110710144,
      "learning_rate": 3.3671113391482812e-06,
      "loss": 0.0079,
      "step": 10500
    },
    {
      "epoch": 0.06740787536918799,
      "grad_norm": 0.7153170108795166,
      "learning_rate": 3.370318111852232e-06,
      "loss": 0.0174,
      "step": 10510
    },
    {
      "epoch": 0.0674720122629741,
      "grad_norm": 0.6602340936660767,
      "learning_rate": 3.373524884556183e-06,
      "loss": 0.0116,
      "step": 10520
    },
    {
      "epoch": 0.06753614915676019,
      "grad_norm": 1.0937711000442505,
      "learning_rate": 3.3767316572601337e-06,
      "loss": 0.0103,
      "step": 10530
    },
    {
      "epoch": 0.06760028605054628,
      "grad_norm": 0.680429220199585,
      "learning_rate": 3.379938429964084e-06,
      "loss": 0.0161,
      "step": 10540
    },
    {
      "epoch": 0.06766442294433238,
      "grad_norm": 0.8192482590675354,
      "learning_rate": 3.383145202668035e-06,
      "loss": 0.0171,
      "step": 10550
    },
    {
      "epoch": 0.06772855983811849,
      "grad_norm": 0.48834553360939026,
      "learning_rate": 3.386351975371986e-06,
      "loss": 0.019,
      "step": 10560
    },
    {
      "epoch": 0.06779269673190458,
      "grad_norm": 0.8775561451911926,
      "learning_rate": 3.389558748075937e-06,
      "loss": 0.0157,
      "step": 10570
    },
    {
      "epoch": 0.06785683362569067,
      "grad_norm": 0.7037835717201233,
      "learning_rate": 3.3927655207798876e-06,
      "loss": 0.0122,
      "step": 10580
    },
    {
      "epoch": 0.06792097051947676,
      "grad_norm": 0.9321784973144531,
      "learning_rate": 3.395972293483838e-06,
      "loss": 0.0138,
      "step": 10590
    },
    {
      "epoch": 0.06798510741326287,
      "grad_norm": 0.6849508881568909,
      "learning_rate": 3.399179066187789e-06,
      "loss": 0.0103,
      "step": 10600
    },
    {
      "epoch": 0.06804924430704896,
      "grad_norm": 0.6341637969017029,
      "learning_rate": 3.4023858388917397e-06,
      "loss": 0.0137,
      "step": 10610
    },
    {
      "epoch": 0.06811338120083506,
      "grad_norm": 0.5561732053756714,
      "learning_rate": 3.4055926115956904e-06,
      "loss": 0.0169,
      "step": 10620
    },
    {
      "epoch": 0.06817751809462116,
      "grad_norm": 0.7867043614387512,
      "learning_rate": 3.408799384299641e-06,
      "loss": 0.0142,
      "step": 10630
    },
    {
      "epoch": 0.06824165498840726,
      "grad_norm": 0.40953245759010315,
      "learning_rate": 3.4120061570035917e-06,
      "loss": 0.0149,
      "step": 10640
    },
    {
      "epoch": 0.06830579188219335,
      "grad_norm": 1.0042667388916016,
      "learning_rate": 3.4152129297075424e-06,
      "loss": 0.0178,
      "step": 10650
    },
    {
      "epoch": 0.06836992877597944,
      "grad_norm": 0.5220630764961243,
      "learning_rate": 3.4184197024114935e-06,
      "loss": 0.011,
      "step": 10660
    },
    {
      "epoch": 0.06843406566976555,
      "grad_norm": 0.8816062808036804,
      "learning_rate": 3.4216264751154442e-06,
      "loss": 0.0162,
      "step": 10670
    },
    {
      "epoch": 0.06849820256355164,
      "grad_norm": 0.3584265112876892,
      "learning_rate": 3.4248332478193945e-06,
      "loss": 0.0123,
      "step": 10680
    },
    {
      "epoch": 0.06856233945733774,
      "grad_norm": 0.9158117175102234,
      "learning_rate": 3.4280400205233456e-06,
      "loss": 0.0125,
      "step": 10690
    },
    {
      "epoch": 0.06862647635112384,
      "grad_norm": 0.6624887585639954,
      "learning_rate": 3.4312467932272963e-06,
      "loss": 0.0131,
      "step": 10700
    },
    {
      "epoch": 0.06869061324490994,
      "grad_norm": 0.5710936188697815,
      "learning_rate": 3.4344535659312474e-06,
      "loss": 0.0129,
      "step": 10710
    },
    {
      "epoch": 0.06875475013869603,
      "grad_norm": 0.5341808199882507,
      "learning_rate": 3.4376603386351977e-06,
      "loss": 0.0189,
      "step": 10720
    },
    {
      "epoch": 0.06881888703248212,
      "grad_norm": 0.9151763916015625,
      "learning_rate": 3.4408671113391484e-06,
      "loss": 0.0122,
      "step": 10730
    },
    {
      "epoch": 0.06888302392626823,
      "grad_norm": 1.0330543518066406,
      "learning_rate": 3.4440738840430995e-06,
      "loss": 0.0101,
      "step": 10740
    },
    {
      "epoch": 0.06894716082005432,
      "grad_norm": 0.3488490879535675,
      "learning_rate": 3.44728065674705e-06,
      "loss": 0.0128,
      "step": 10750
    },
    {
      "epoch": 0.06901129771384042,
      "grad_norm": 0.4824540615081787,
      "learning_rate": 3.450487429451001e-06,
      "loss": 0.0149,
      "step": 10760
    },
    {
      "epoch": 0.06907543460762652,
      "grad_norm": 0.7239876389503479,
      "learning_rate": 3.453694202154951e-06,
      "loss": 0.0111,
      "step": 10770
    },
    {
      "epoch": 0.06913957150141262,
      "grad_norm": 0.7549890875816345,
      "learning_rate": 3.4569009748589022e-06,
      "loss": 0.0149,
      "step": 10780
    },
    {
      "epoch": 0.06920370839519871,
      "grad_norm": 0.44856587052345276,
      "learning_rate": 3.460107747562853e-06,
      "loss": 0.0122,
      "step": 10790
    },
    {
      "epoch": 0.0692678452889848,
      "grad_norm": 0.9687082171440125,
      "learning_rate": 3.463314520266804e-06,
      "loss": 0.0144,
      "step": 10800
    },
    {
      "epoch": 0.06933198218277091,
      "grad_norm": 0.6999725699424744,
      "learning_rate": 3.4665212929707547e-06,
      "loss": 0.0116,
      "step": 10810
    },
    {
      "epoch": 0.069396119076557,
      "grad_norm": 0.7693178653717041,
      "learning_rate": 3.469728065674705e-06,
      "loss": 0.0151,
      "step": 10820
    },
    {
      "epoch": 0.0694602559703431,
      "grad_norm": 0.5353623032569885,
      "learning_rate": 3.472934838378656e-06,
      "loss": 0.0128,
      "step": 10830
    },
    {
      "epoch": 0.0695243928641292,
      "grad_norm": 0.7831609845161438,
      "learning_rate": 3.476141611082607e-06,
      "loss": 0.0128,
      "step": 10840
    },
    {
      "epoch": 0.0695885297579153,
      "grad_norm": 0.5129367113113403,
      "learning_rate": 3.479348383786558e-06,
      "loss": 0.0169,
      "step": 10850
    },
    {
      "epoch": 0.06965266665170139,
      "grad_norm": 0.5755190849304199,
      "learning_rate": 3.482555156490508e-06,
      "loss": 0.0134,
      "step": 10860
    },
    {
      "epoch": 0.06971680354548748,
      "grad_norm": 0.9079647660255432,
      "learning_rate": 3.485761929194459e-06,
      "loss": 0.0134,
      "step": 10870
    },
    {
      "epoch": 0.06978094043927359,
      "grad_norm": 0.889003574848175,
      "learning_rate": 3.4889687018984096e-06,
      "loss": 0.0169,
      "step": 10880
    },
    {
      "epoch": 0.06984507733305968,
      "grad_norm": 0.8207578659057617,
      "learning_rate": 3.4921754746023607e-06,
      "loss": 0.0159,
      "step": 10890
    },
    {
      "epoch": 0.06990921422684578,
      "grad_norm": 0.5410775542259216,
      "learning_rate": 3.4953822473063114e-06,
      "loss": 0.0145,
      "step": 10900
    },
    {
      "epoch": 0.06997335112063188,
      "grad_norm": 0.848727822303772,
      "learning_rate": 3.4985890200102616e-06,
      "loss": 0.0172,
      "step": 10910
    },
    {
      "epoch": 0.07003748801441798,
      "grad_norm": 0.6241754293441772,
      "learning_rate": 3.5017957927142127e-06,
      "loss": 0.0127,
      "step": 10920
    },
    {
      "epoch": 0.07010162490820407,
      "grad_norm": 0.6075505614280701,
      "learning_rate": 3.5050025654181634e-06,
      "loss": 0.011,
      "step": 10930
    },
    {
      "epoch": 0.07016576180199016,
      "grad_norm": 0.86773282289505,
      "learning_rate": 3.5082093381221145e-06,
      "loss": 0.0142,
      "step": 10940
    },
    {
      "epoch": 0.07022989869577627,
      "grad_norm": 0.6941466927528381,
      "learning_rate": 3.511416110826065e-06,
      "loss": 0.0133,
      "step": 10950
    },
    {
      "epoch": 0.07029403558956236,
      "grad_norm": 0.7797922492027283,
      "learning_rate": 3.5146228835300155e-06,
      "loss": 0.01,
      "step": 10960
    },
    {
      "epoch": 0.07035817248334846,
      "grad_norm": 0.8995891809463501,
      "learning_rate": 3.5178296562339666e-06,
      "loss": 0.0129,
      "step": 10970
    },
    {
      "epoch": 0.07042230937713456,
      "grad_norm": 0.8827101588249207,
      "learning_rate": 3.5210364289379173e-06,
      "loss": 0.0116,
      "step": 10980
    },
    {
      "epoch": 0.07048644627092066,
      "grad_norm": 0.3860144317150116,
      "learning_rate": 3.524243201641868e-06,
      "loss": 0.0109,
      "step": 10990
    },
    {
      "epoch": 0.07055058316470675,
      "grad_norm": 0.4791621267795563,
      "learning_rate": 3.5274499743458187e-06,
      "loss": 0.0109,
      "step": 11000
    },
    {
      "epoch": 0.07061472005849284,
      "grad_norm": 0.6825062036514282,
      "learning_rate": 3.5306567470497694e-06,
      "loss": 0.0121,
      "step": 11010
    },
    {
      "epoch": 0.07067885695227895,
      "grad_norm": 1.4695130586624146,
      "learning_rate": 3.53386351975372e-06,
      "loss": 0.0131,
      "step": 11020
    },
    {
      "epoch": 0.07074299384606504,
      "grad_norm": 0.32003363966941833,
      "learning_rate": 3.537070292457671e-06,
      "loss": 0.0148,
      "step": 11030
    },
    {
      "epoch": 0.07080713073985113,
      "grad_norm": 0.6566299200057983,
      "learning_rate": 3.5402770651616214e-06,
      "loss": 0.0124,
      "step": 11040
    },
    {
      "epoch": 0.07087126763363724,
      "grad_norm": 0.7709537744522095,
      "learning_rate": 3.543483837865572e-06,
      "loss": 0.0137,
      "step": 11050
    },
    {
      "epoch": 0.07093540452742333,
      "grad_norm": 0.46081745624542236,
      "learning_rate": 3.5466906105695232e-06,
      "loss": 0.0108,
      "step": 11060
    },
    {
      "epoch": 0.07099954142120943,
      "grad_norm": 0.5471722483634949,
      "learning_rate": 3.549897383273474e-06,
      "loss": 0.0115,
      "step": 11070
    },
    {
      "epoch": 0.07106367831499552,
      "grad_norm": 0.9127947092056274,
      "learning_rate": 3.553104155977425e-06,
      "loss": 0.016,
      "step": 11080
    },
    {
      "epoch": 0.07112781520878163,
      "grad_norm": 0.6626812219619751,
      "learning_rate": 3.5563109286813753e-06,
      "loss": 0.0165,
      "step": 11090
    },
    {
      "epoch": 0.07119195210256772,
      "grad_norm": 0.9916837215423584,
      "learning_rate": 3.559517701385326e-06,
      "loss": 0.0157,
      "step": 11100
    },
    {
      "epoch": 0.07125608899635381,
      "grad_norm": 0.5472110509872437,
      "learning_rate": 3.562724474089277e-06,
      "loss": 0.0111,
      "step": 11110
    },
    {
      "epoch": 0.07132022589013992,
      "grad_norm": 0.5907249450683594,
      "learning_rate": 3.565931246793228e-06,
      "loss": 0.0133,
      "step": 11120
    },
    {
      "epoch": 0.07138436278392601,
      "grad_norm": 0.7342523336410522,
      "learning_rate": 3.569138019497178e-06,
      "loss": 0.013,
      "step": 11130
    },
    {
      "epoch": 0.07144849967771211,
      "grad_norm": 0.8366237282752991,
      "learning_rate": 3.5723447922011288e-06,
      "loss": 0.0086,
      "step": 11140
    },
    {
      "epoch": 0.0715126365714982,
      "grad_norm": 0.23374402523040771,
      "learning_rate": 3.57555156490508e-06,
      "loss": 0.0106,
      "step": 11150
    },
    {
      "epoch": 0.07157677346528431,
      "grad_norm": 0.6893095970153809,
      "learning_rate": 3.5787583376090306e-06,
      "loss": 0.0168,
      "step": 11160
    },
    {
      "epoch": 0.0716409103590704,
      "grad_norm": 0.6772871017456055,
      "learning_rate": 3.5819651103129817e-06,
      "loss": 0.0111,
      "step": 11170
    },
    {
      "epoch": 0.0717050472528565,
      "grad_norm": 0.7183294296264648,
      "learning_rate": 3.585171883016932e-06,
      "loss": 0.0182,
      "step": 11180
    },
    {
      "epoch": 0.0717691841466426,
      "grad_norm": 0.7910516262054443,
      "learning_rate": 3.5883786557208826e-06,
      "loss": 0.0176,
      "step": 11190
    },
    {
      "epoch": 0.0718333210404287,
      "grad_norm": 0.6576840877532959,
      "learning_rate": 3.5915854284248337e-06,
      "loss": 0.0107,
      "step": 11200
    },
    {
      "epoch": 0.07189745793421479,
      "grad_norm": 0.470316618680954,
      "learning_rate": 3.5947922011287844e-06,
      "loss": 0.0083,
      "step": 11210
    },
    {
      "epoch": 0.07196159482800088,
      "grad_norm": 0.8724121451377869,
      "learning_rate": 3.5979989738327347e-06,
      "loss": 0.0135,
      "step": 11220
    },
    {
      "epoch": 0.07202573172178699,
      "grad_norm": 0.9997464418411255,
      "learning_rate": 3.601205746536686e-06,
      "loss": 0.0134,
      "step": 11230
    },
    {
      "epoch": 0.07208986861557308,
      "grad_norm": 0.8020079731941223,
      "learning_rate": 3.6044125192406365e-06,
      "loss": 0.0157,
      "step": 11240
    },
    {
      "epoch": 0.07215400550935917,
      "grad_norm": 0.3713897466659546,
      "learning_rate": 3.607619291944587e-06,
      "loss": 0.0083,
      "step": 11250
    },
    {
      "epoch": 0.07221814240314528,
      "grad_norm": 0.7209635972976685,
      "learning_rate": 3.6108260646485383e-06,
      "loss": 0.0122,
      "step": 11260
    },
    {
      "epoch": 0.07228227929693137,
      "grad_norm": 0.40024682879447937,
      "learning_rate": 3.6140328373524886e-06,
      "loss": 0.0126,
      "step": 11270
    },
    {
      "epoch": 0.07234641619071747,
      "grad_norm": 0.8981276154518127,
      "learning_rate": 3.6172396100564393e-06,
      "loss": 0.018,
      "step": 11280
    },
    {
      "epoch": 0.07241055308450356,
      "grad_norm": 0.822223424911499,
      "learning_rate": 3.6204463827603904e-06,
      "loss": 0.0131,
      "step": 11290
    },
    {
      "epoch": 0.07247468997828967,
      "grad_norm": 0.5042352676391602,
      "learning_rate": 3.623653155464341e-06,
      "loss": 0.0098,
      "step": 11300
    },
    {
      "epoch": 0.07253882687207576,
      "grad_norm": 0.5179206728935242,
      "learning_rate": 3.6268599281682913e-06,
      "loss": 0.0114,
      "step": 11310
    },
    {
      "epoch": 0.07260296376586185,
      "grad_norm": 0.6678454875946045,
      "learning_rate": 3.6300667008722424e-06,
      "loss": 0.0153,
      "step": 11320
    },
    {
      "epoch": 0.07266710065964795,
      "grad_norm": 0.52106773853302,
      "learning_rate": 3.633273473576193e-06,
      "loss": 0.0116,
      "step": 11330
    },
    {
      "epoch": 0.07273123755343405,
      "grad_norm": 0.9093338847160339,
      "learning_rate": 3.6364802462801442e-06,
      "loss": 0.0111,
      "step": 11340
    },
    {
      "epoch": 0.07279537444722015,
      "grad_norm": 0.8999060392379761,
      "learning_rate": 3.639687018984095e-06,
      "loss": 0.0171,
      "step": 11350
    },
    {
      "epoch": 0.07285951134100624,
      "grad_norm": 1.9983782768249512,
      "learning_rate": 3.642893791688045e-06,
      "loss": 0.0146,
      "step": 11360
    },
    {
      "epoch": 0.07292364823479235,
      "grad_norm": 0.3759327530860901,
      "learning_rate": 3.6461005643919963e-06,
      "loss": 0.0124,
      "step": 11370
    },
    {
      "epoch": 0.07298778512857844,
      "grad_norm": 0.7182804942131042,
      "learning_rate": 3.649307337095947e-06,
      "loss": 0.0125,
      "step": 11380
    },
    {
      "epoch": 0.07305192202236453,
      "grad_norm": 0.8969981670379639,
      "learning_rate": 3.6525141097998977e-06,
      "loss": 0.0152,
      "step": 11390
    },
    {
      "epoch": 0.07311605891615063,
      "grad_norm": 0.5347842574119568,
      "learning_rate": 3.655720882503848e-06,
      "loss": 0.0108,
      "step": 11400
    },
    {
      "epoch": 0.07318019580993673,
      "grad_norm": 0.6039052605628967,
      "learning_rate": 3.658927655207799e-06,
      "loss": 0.0134,
      "step": 11410
    },
    {
      "epoch": 0.07324433270372283,
      "grad_norm": 0.42864710092544556,
      "learning_rate": 3.6621344279117498e-06,
      "loss": 0.0145,
      "step": 11420
    },
    {
      "epoch": 0.07330846959750892,
      "grad_norm": 1.079379677772522,
      "learning_rate": 3.665341200615701e-06,
      "loss": 0.011,
      "step": 11430
    },
    {
      "epoch": 0.07337260649129503,
      "grad_norm": 0.4444161355495453,
      "learning_rate": 3.6685479733196516e-06,
      "loss": 0.0144,
      "step": 11440
    },
    {
      "epoch": 0.07343674338508112,
      "grad_norm": 0.4957866370677948,
      "learning_rate": 3.671754746023602e-06,
      "loss": 0.0124,
      "step": 11450
    },
    {
      "epoch": 0.07350088027886721,
      "grad_norm": 1.4191908836364746,
      "learning_rate": 3.674961518727553e-06,
      "loss": 0.0153,
      "step": 11460
    },
    {
      "epoch": 0.0735650171726533,
      "grad_norm": 0.5625303387641907,
      "learning_rate": 3.6781682914315036e-06,
      "loss": 0.0142,
      "step": 11470
    },
    {
      "epoch": 0.07362915406643941,
      "grad_norm": 1.1107635498046875,
      "learning_rate": 3.6813750641354547e-06,
      "loss": 0.0149,
      "step": 11480
    },
    {
      "epoch": 0.0736932909602255,
      "grad_norm": 0.32591041922569275,
      "learning_rate": 3.6845818368394054e-06,
      "loss": 0.0133,
      "step": 11490
    },
    {
      "epoch": 0.0737574278540116,
      "grad_norm": 0.37027707695961,
      "learning_rate": 3.6877886095433557e-06,
      "loss": 0.0137,
      "step": 11500
    },
    {
      "epoch": 0.0738215647477977,
      "grad_norm": 0.7116591334342957,
      "learning_rate": 3.6909953822473064e-06,
      "loss": 0.0098,
      "step": 11510
    },
    {
      "epoch": 0.0738857016415838,
      "grad_norm": 0.4879796802997589,
      "learning_rate": 3.6942021549512575e-06,
      "loss": 0.0121,
      "step": 11520
    },
    {
      "epoch": 0.07394983853536989,
      "grad_norm": 0.46357476711273193,
      "learning_rate": 3.697408927655208e-06,
      "loss": 0.008,
      "step": 11530
    },
    {
      "epoch": 0.07401397542915598,
      "grad_norm": 0.6622515320777893,
      "learning_rate": 3.7006157003591585e-06,
      "loss": 0.0087,
      "step": 11540
    },
    {
      "epoch": 0.07407811232294209,
      "grad_norm": 1.0612263679504395,
      "learning_rate": 3.7038224730631096e-06,
      "loss": 0.0114,
      "step": 11550
    },
    {
      "epoch": 0.07414224921672818,
      "grad_norm": 0.630497932434082,
      "learning_rate": 3.7070292457670603e-06,
      "loss": 0.0146,
      "step": 11560
    },
    {
      "epoch": 0.07420638611051428,
      "grad_norm": 0.5127402544021606,
      "learning_rate": 3.7102360184710114e-06,
      "loss": 0.0094,
      "step": 11570
    },
    {
      "epoch": 0.07427052300430038,
      "grad_norm": 0.39021891355514526,
      "learning_rate": 3.713442791174962e-06,
      "loss": 0.0112,
      "step": 11580
    },
    {
      "epoch": 0.07433465989808648,
      "grad_norm": 0.6863677501678467,
      "learning_rate": 3.7166495638789123e-06,
      "loss": 0.0153,
      "step": 11590
    },
    {
      "epoch": 0.07439879679187257,
      "grad_norm": 0.7049649953842163,
      "learning_rate": 3.7198563365828634e-06,
      "loss": 0.0145,
      "step": 11600
    },
    {
      "epoch": 0.07446293368565866,
      "grad_norm": 0.5492684841156006,
      "learning_rate": 3.723063109286814e-06,
      "loss": 0.0098,
      "step": 11610
    },
    {
      "epoch": 0.07452707057944477,
      "grad_norm": 0.5002531409263611,
      "learning_rate": 3.726269881990765e-06,
      "loss": 0.0093,
      "step": 11620
    },
    {
      "epoch": 0.07459120747323086,
      "grad_norm": 0.783821702003479,
      "learning_rate": 3.7294766546947155e-06,
      "loss": 0.0128,
      "step": 11630
    },
    {
      "epoch": 0.07465534436701696,
      "grad_norm": 0.539421021938324,
      "learning_rate": 3.732683427398666e-06,
      "loss": 0.0145,
      "step": 11640
    },
    {
      "epoch": 0.07471948126080306,
      "grad_norm": 0.740747332572937,
      "learning_rate": 3.735890200102617e-06,
      "loss": 0.0136,
      "step": 11650
    },
    {
      "epoch": 0.07478361815458916,
      "grad_norm": 0.5146269798278809,
      "learning_rate": 3.739096972806568e-06,
      "loss": 0.0139,
      "step": 11660
    },
    {
      "epoch": 0.07484775504837525,
      "grad_norm": 0.39279046654701233,
      "learning_rate": 3.7423037455105187e-06,
      "loss": 0.0117,
      "step": 11670
    },
    {
      "epoch": 0.07491189194216134,
      "grad_norm": 0.3348548114299774,
      "learning_rate": 3.745510518214469e-06,
      "loss": 0.0144,
      "step": 11680
    },
    {
      "epoch": 0.07497602883594745,
      "grad_norm": 0.9481940865516663,
      "learning_rate": 3.74871729091842e-06,
      "loss": 0.0122,
      "step": 11690
    },
    {
      "epoch": 0.07504016572973354,
      "grad_norm": 0.5905875563621521,
      "learning_rate": 3.7519240636223708e-06,
      "loss": 0.0084,
      "step": 11700
    },
    {
      "epoch": 0.07510430262351964,
      "grad_norm": 0.7137616276741028,
      "learning_rate": 3.755130836326322e-06,
      "loss": 0.0129,
      "step": 11710
    },
    {
      "epoch": 0.07516843951730574,
      "grad_norm": 0.6572869420051575,
      "learning_rate": 3.758337609030272e-06,
      "loss": 0.0177,
      "step": 11720
    },
    {
      "epoch": 0.07523257641109184,
      "grad_norm": 0.6707942485809326,
      "learning_rate": 3.761544381734223e-06,
      "loss": 0.0119,
      "step": 11730
    },
    {
      "epoch": 0.07529671330487793,
      "grad_norm": 0.3048764169216156,
      "learning_rate": 3.764751154438174e-06,
      "loss": 0.0164,
      "step": 11740
    },
    {
      "epoch": 0.07536085019866402,
      "grad_norm": 0.8465550541877747,
      "learning_rate": 3.7679579271421246e-06,
      "loss": 0.0121,
      "step": 11750
    },
    {
      "epoch": 0.07542498709245013,
      "grad_norm": 1.304787516593933,
      "learning_rate": 3.7711646998460753e-06,
      "loss": 0.0099,
      "step": 11760
    },
    {
      "epoch": 0.07548912398623622,
      "grad_norm": 0.7193863987922668,
      "learning_rate": 3.7743714725500256e-06,
      "loss": 0.0106,
      "step": 11770
    },
    {
      "epoch": 0.07555326088002232,
      "grad_norm": 0.6778226494789124,
      "learning_rate": 3.7775782452539767e-06,
      "loss": 0.01,
      "step": 11780
    },
    {
      "epoch": 0.07561739777380842,
      "grad_norm": 0.7179293036460876,
      "learning_rate": 3.7807850179579274e-06,
      "loss": 0.0138,
      "step": 11790
    },
    {
      "epoch": 0.07568153466759452,
      "grad_norm": 0.3168581426143646,
      "learning_rate": 3.7839917906618785e-06,
      "loss": 0.0111,
      "step": 11800
    },
    {
      "epoch": 0.07574567156138061,
      "grad_norm": 0.6792516708374023,
      "learning_rate": 3.7871985633658288e-06,
      "loss": 0.0124,
      "step": 11810
    },
    {
      "epoch": 0.0758098084551667,
      "grad_norm": 0.9062817096710205,
      "learning_rate": 3.7904053360697795e-06,
      "loss": 0.0097,
      "step": 11820
    },
    {
      "epoch": 0.07587394534895281,
      "grad_norm": 0.5379997491836548,
      "learning_rate": 3.7936121087737306e-06,
      "loss": 0.0115,
      "step": 11830
    },
    {
      "epoch": 0.0759380822427389,
      "grad_norm": 0.6321361064910889,
      "learning_rate": 3.7968188814776813e-06,
      "loss": 0.0112,
      "step": 11840
    },
    {
      "epoch": 0.076002219136525,
      "grad_norm": 0.6613107919692993,
      "learning_rate": 3.800025654181632e-06,
      "loss": 0.0107,
      "step": 11850
    },
    {
      "epoch": 0.0760663560303111,
      "grad_norm": 0.433931440114975,
      "learning_rate": 3.8032324268855826e-06,
      "loss": 0.013,
      "step": 11860
    },
    {
      "epoch": 0.0761304929240972,
      "grad_norm": 0.5778363347053528,
      "learning_rate": 3.8064391995895333e-06,
      "loss": 0.016,
      "step": 11870
    },
    {
      "epoch": 0.07619462981788329,
      "grad_norm": 0.47336652874946594,
      "learning_rate": 3.809645972293484e-06,
      "loss": 0.0091,
      "step": 11880
    },
    {
      "epoch": 0.07625876671166938,
      "grad_norm": 0.691404402256012,
      "learning_rate": 3.812852744997435e-06,
      "loss": 0.0092,
      "step": 11890
    },
    {
      "epoch": 0.07632290360545549,
      "grad_norm": 0.5509531497955322,
      "learning_rate": 3.816059517701385e-06,
      "loss": 0.0118,
      "step": 11900
    },
    {
      "epoch": 0.07638704049924158,
      "grad_norm": 1.3395763635635376,
      "learning_rate": 3.819266290405336e-06,
      "loss": 0.0117,
      "step": 11910
    },
    {
      "epoch": 0.07645117739302768,
      "grad_norm": 0.6070330739021301,
      "learning_rate": 3.822473063109287e-06,
      "loss": 0.0102,
      "step": 11920
    },
    {
      "epoch": 0.07651531428681378,
      "grad_norm": 9.7800931930542,
      "learning_rate": 3.825679835813238e-06,
      "loss": 0.01,
      "step": 11930
    },
    {
      "epoch": 0.07657945118059988,
      "grad_norm": 0.5915313959121704,
      "learning_rate": 3.828886608517189e-06,
      "loss": 0.016,
      "step": 11940
    },
    {
      "epoch": 0.07664358807438597,
      "grad_norm": 0.6098341345787048,
      "learning_rate": 3.832093381221139e-06,
      "loss": 0.0144,
      "step": 11950
    },
    {
      "epoch": 0.07670772496817206,
      "grad_norm": 0.6589644551277161,
      "learning_rate": 3.83530015392509e-06,
      "loss": 0.0141,
      "step": 11960
    },
    {
      "epoch": 0.07677186186195817,
      "grad_norm": 0.26208600401878357,
      "learning_rate": 3.838506926629041e-06,
      "loss": 0.01,
      "step": 11970
    },
    {
      "epoch": 0.07683599875574426,
      "grad_norm": 0.5078624486923218,
      "learning_rate": 3.841713699332991e-06,
      "loss": 0.0149,
      "step": 11980
    },
    {
      "epoch": 0.07690013564953035,
      "grad_norm": 0.7186983227729797,
      "learning_rate": 3.8449204720369425e-06,
      "loss": 0.0086,
      "step": 11990
    },
    {
      "epoch": 0.07696427254331646,
      "grad_norm": 0.6783637404441833,
      "learning_rate": 3.848127244740893e-06,
      "loss": 0.0142,
      "step": 12000
    },
    {
      "epoch": 0.07702840943710255,
      "grad_norm": 0.7996017932891846,
      "learning_rate": 3.851334017444844e-06,
      "loss": 0.0129,
      "step": 12010
    },
    {
      "epoch": 0.07709254633088865,
      "grad_norm": 1.3191332817077637,
      "learning_rate": 3.854540790148795e-06,
      "loss": 0.0099,
      "step": 12020
    },
    {
      "epoch": 0.07715668322467474,
      "grad_norm": 1.0860892534255981,
      "learning_rate": 3.857747562852745e-06,
      "loss": 0.0115,
      "step": 12030
    },
    {
      "epoch": 0.07722082011846085,
      "grad_norm": 0.4692663550376892,
      "learning_rate": 3.8609543355566955e-06,
      "loss": 0.0081,
      "step": 12040
    },
    {
      "epoch": 0.07728495701224694,
      "grad_norm": 0.9405308961868286,
      "learning_rate": 3.864161108260647e-06,
      "loss": 0.0146,
      "step": 12050
    },
    {
      "epoch": 0.07734909390603303,
      "grad_norm": 0.7099402546882629,
      "learning_rate": 3.867367880964598e-06,
      "loss": 0.012,
      "step": 12060
    },
    {
      "epoch": 0.07741323079981913,
      "grad_norm": 0.7617799639701843,
      "learning_rate": 3.870574653668549e-06,
      "loss": 0.0142,
      "step": 12070
    },
    {
      "epoch": 0.07747736769360523,
      "grad_norm": 0.8338746428489685,
      "learning_rate": 3.873781426372499e-06,
      "loss": 0.0111,
      "step": 12080
    },
    {
      "epoch": 0.07754150458739133,
      "grad_norm": 0.9001340270042419,
      "learning_rate": 3.876988199076449e-06,
      "loss": 0.0103,
      "step": 12090
    },
    {
      "epoch": 0.07760564148117742,
      "grad_norm": 0.6787715554237366,
      "learning_rate": 3.8801949717804005e-06,
      "loss": 0.008,
      "step": 12100
    },
    {
      "epoch": 0.07766977837496353,
      "grad_norm": 0.7984828948974609,
      "learning_rate": 3.883401744484352e-06,
      "loss": 0.0122,
      "step": 12110
    },
    {
      "epoch": 0.07773391526874962,
      "grad_norm": 0.8640044331550598,
      "learning_rate": 3.886608517188302e-06,
      "loss": 0.0121,
      "step": 12120
    },
    {
      "epoch": 0.07779805216253571,
      "grad_norm": 0.4400997459888458,
      "learning_rate": 3.889815289892253e-06,
      "loss": 0.0149,
      "step": 12130
    },
    {
      "epoch": 0.0778621890563218,
      "grad_norm": 1.0008609294891357,
      "learning_rate": 3.893022062596203e-06,
      "loss": 0.0121,
      "step": 12140
    },
    {
      "epoch": 0.07792632595010791,
      "grad_norm": 0.7383760809898376,
      "learning_rate": 3.896228835300154e-06,
      "loss": 0.0111,
      "step": 12150
    },
    {
      "epoch": 0.077990462843894,
      "grad_norm": 0.9170994758605957,
      "learning_rate": 3.8994356080041054e-06,
      "loss": 0.0099,
      "step": 12160
    },
    {
      "epoch": 0.0780545997376801,
      "grad_norm": 0.700387179851532,
      "learning_rate": 3.902642380708056e-06,
      "loss": 0.0127,
      "step": 12170
    },
    {
      "epoch": 0.0781187366314662,
      "grad_norm": 0.6357445120811462,
      "learning_rate": 3.905849153412006e-06,
      "loss": 0.0157,
      "step": 12180
    },
    {
      "epoch": 0.0781828735252523,
      "grad_norm": 0.8678653836250305,
      "learning_rate": 3.909055926115957e-06,
      "loss": 0.0124,
      "step": 12190
    },
    {
      "epoch": 0.07824701041903839,
      "grad_norm": 0.6583905220031738,
      "learning_rate": 3.912262698819908e-06,
      "loss": 0.0129,
      "step": 12200
    },
    {
      "epoch": 0.07831114731282449,
      "grad_norm": 0.4002678692340851,
      "learning_rate": 3.915469471523859e-06,
      "loss": 0.0091,
      "step": 12210
    },
    {
      "epoch": 0.0783752842066106,
      "grad_norm": 0.7340361475944519,
      "learning_rate": 3.91867624422781e-06,
      "loss": 0.011,
      "step": 12220
    },
    {
      "epoch": 0.07843942110039669,
      "grad_norm": 0.5798015594482422,
      "learning_rate": 3.92188301693176e-06,
      "loss": 0.0159,
      "step": 12230
    },
    {
      "epoch": 0.07850355799418278,
      "grad_norm": 0.6820700764656067,
      "learning_rate": 3.925089789635711e-06,
      "loss": 0.0111,
      "step": 12240
    },
    {
      "epoch": 0.07856769488796889,
      "grad_norm": 0.5610462427139282,
      "learning_rate": 3.928296562339662e-06,
      "loss": 0.0089,
      "step": 12250
    },
    {
      "epoch": 0.07863183178175498,
      "grad_norm": 0.39736512303352356,
      "learning_rate": 3.931503335043612e-06,
      "loss": 0.0104,
      "step": 12260
    },
    {
      "epoch": 0.07869596867554107,
      "grad_norm": 0.6950397491455078,
      "learning_rate": 3.934710107747563e-06,
      "loss": 0.0101,
      "step": 12270
    },
    {
      "epoch": 0.07876010556932717,
      "grad_norm": 0.8704506754875183,
      "learning_rate": 3.937916880451514e-06,
      "loss": 0.0131,
      "step": 12280
    },
    {
      "epoch": 0.07882424246311327,
      "grad_norm": 0.7454317808151245,
      "learning_rate": 3.941123653155465e-06,
      "loss": 0.0112,
      "step": 12290
    },
    {
      "epoch": 0.07888837935689937,
      "grad_norm": 0.33978649973869324,
      "learning_rate": 3.944330425859416e-06,
      "loss": 0.0105,
      "step": 12300
    },
    {
      "epoch": 0.07895251625068546,
      "grad_norm": 0.8149883151054382,
      "learning_rate": 3.947537198563366e-06,
      "loss": 0.0177,
      "step": 12310
    },
    {
      "epoch": 0.07901665314447157,
      "grad_norm": 0.5741153955459595,
      "learning_rate": 3.9507439712673165e-06,
      "loss": 0.0118,
      "step": 12320
    },
    {
      "epoch": 0.07908079003825766,
      "grad_norm": 0.8584007620811462,
      "learning_rate": 3.953950743971268e-06,
      "loss": 0.0122,
      "step": 12330
    },
    {
      "epoch": 0.07914492693204375,
      "grad_norm": 0.5068797469139099,
      "learning_rate": 3.957157516675219e-06,
      "loss": 0.0124,
      "step": 12340
    },
    {
      "epoch": 0.07920906382582985,
      "grad_norm": 0.36949658393859863,
      "learning_rate": 3.960364289379169e-06,
      "loss": 0.011,
      "step": 12350
    },
    {
      "epoch": 0.07927320071961595,
      "grad_norm": 0.41789624094963074,
      "learning_rate": 3.96357106208312e-06,
      "loss": 0.0119,
      "step": 12360
    },
    {
      "epoch": 0.07933733761340205,
      "grad_norm": 0.1528804749250412,
      "learning_rate": 3.96677783478707e-06,
      "loss": 0.0132,
      "step": 12370
    },
    {
      "epoch": 0.07940147450718814,
      "grad_norm": 0.3877411186695099,
      "learning_rate": 3.9699846074910215e-06,
      "loss": 0.0154,
      "step": 12380
    },
    {
      "epoch": 0.07946561140097425,
      "grad_norm": 0.5584845542907715,
      "learning_rate": 3.973191380194973e-06,
      "loss": 0.0127,
      "step": 12390
    },
    {
      "epoch": 0.07952974829476034,
      "grad_norm": 0.28894999623298645,
      "learning_rate": 3.976398152898923e-06,
      "loss": 0.0127,
      "step": 12400
    },
    {
      "epoch": 0.07959388518854643,
      "grad_norm": 0.7544544339179993,
      "learning_rate": 3.979604925602873e-06,
      "loss": 0.0134,
      "step": 12410
    },
    {
      "epoch": 0.07965802208233252,
      "grad_norm": 0.5420180559158325,
      "learning_rate": 3.982811698306824e-06,
      "loss": 0.0142,
      "step": 12420
    },
    {
      "epoch": 0.07972215897611863,
      "grad_norm": 0.4947992265224457,
      "learning_rate": 3.986018471010775e-06,
      "loss": 0.0099,
      "step": 12430
    },
    {
      "epoch": 0.07978629586990472,
      "grad_norm": 0.8791900873184204,
      "learning_rate": 3.9892252437147265e-06,
      "loss": 0.0146,
      "step": 12440
    },
    {
      "epoch": 0.07985043276369082,
      "grad_norm": 0.5872790217399597,
      "learning_rate": 3.992432016418677e-06,
      "loss": 0.0094,
      "step": 12450
    },
    {
      "epoch": 0.07991456965747692,
      "grad_norm": 1.1887428760528564,
      "learning_rate": 3.995638789122627e-06,
      "loss": 0.0113,
      "step": 12460
    },
    {
      "epoch": 0.07997870655126302,
      "grad_norm": 0.6307410001754761,
      "learning_rate": 3.998845561826578e-06,
      "loss": 0.009,
      "step": 12470
    },
    {
      "epoch": 0.08004284344504911,
      "grad_norm": 0.4119446277618408,
      "learning_rate": 4.002052334530529e-06,
      "loss": 0.0143,
      "step": 12480
    },
    {
      "epoch": 0.0801069803388352,
      "grad_norm": 0.6567076444625854,
      "learning_rate": 4.0052591072344795e-06,
      "loss": 0.013,
      "step": 12490
    },
    {
      "epoch": 0.08017111723262131,
      "grad_norm": 0.7034739851951599,
      "learning_rate": 4.008465879938431e-06,
      "loss": 0.01,
      "step": 12500
    },
    {
      "epoch": 0.0802352541264074,
      "grad_norm": 0.5242000222206116,
      "learning_rate": 4.011672652642381e-06,
      "loss": 0.013,
      "step": 12510
    },
    {
      "epoch": 0.0802993910201935,
      "grad_norm": 0.644496500492096,
      "learning_rate": 4.014879425346332e-06,
      "loss": 0.0132,
      "step": 12520
    },
    {
      "epoch": 0.0803635279139796,
      "grad_norm": 0.5115908980369568,
      "learning_rate": 4.018086198050283e-06,
      "loss": 0.0111,
      "step": 12530
    },
    {
      "epoch": 0.0804276648077657,
      "grad_norm": 0.1372889280319214,
      "learning_rate": 4.021292970754233e-06,
      "loss": 0.0139,
      "step": 12540
    },
    {
      "epoch": 0.08049180170155179,
      "grad_norm": 0.8904168009757996,
      "learning_rate": 4.024499743458184e-06,
      "loss": 0.0118,
      "step": 12550
    },
    {
      "epoch": 0.08055593859533788,
      "grad_norm": 0.7952111959457397,
      "learning_rate": 4.027706516162135e-06,
      "loss": 0.0114,
      "step": 12560
    },
    {
      "epoch": 0.08062007548912399,
      "grad_norm": 0.4509839713573456,
      "learning_rate": 4.030913288866086e-06,
      "loss": 0.0096,
      "step": 12570
    },
    {
      "epoch": 0.08068421238291008,
      "grad_norm": 0.4335967004299164,
      "learning_rate": 4.034120061570036e-06,
      "loss": 0.0089,
      "step": 12580
    },
    {
      "epoch": 0.08074834927669618,
      "grad_norm": 0.6105585694313049,
      "learning_rate": 4.037326834273987e-06,
      "loss": 0.0143,
      "step": 12590
    },
    {
      "epoch": 0.08081248617048228,
      "grad_norm": 0.5058387517929077,
      "learning_rate": 4.0405336069779375e-06,
      "loss": 0.011,
      "step": 12600
    },
    {
      "epoch": 0.08087662306426838,
      "grad_norm": 0.7142662405967712,
      "learning_rate": 4.043740379681889e-06,
      "loss": 0.0159,
      "step": 12610
    },
    {
      "epoch": 0.08094075995805447,
      "grad_norm": 0.6708266735076904,
      "learning_rate": 4.04694715238584e-06,
      "loss": 0.0107,
      "step": 12620
    },
    {
      "epoch": 0.08100489685184056,
      "grad_norm": 0.49649274349212646,
      "learning_rate": 4.05015392508979e-06,
      "loss": 0.0099,
      "step": 12630
    },
    {
      "epoch": 0.08106903374562667,
      "grad_norm": 0.737576425075531,
      "learning_rate": 4.05336069779374e-06,
      "loss": 0.0097,
      "step": 12640
    },
    {
      "epoch": 0.08113317063941276,
      "grad_norm": 0.6056515574455261,
      "learning_rate": 4.056567470497691e-06,
      "loss": 0.0156,
      "step": 12650
    },
    {
      "epoch": 0.08119730753319886,
      "grad_norm": 0.6571008563041687,
      "learning_rate": 4.0597742432016425e-06,
      "loss": 0.0085,
      "step": 12660
    },
    {
      "epoch": 0.08126144442698496,
      "grad_norm": 0.7000786662101746,
      "learning_rate": 4.062981015905593e-06,
      "loss": 0.0163,
      "step": 12670
    },
    {
      "epoch": 0.08132558132077106,
      "grad_norm": 0.9011369943618774,
      "learning_rate": 4.066187788609544e-06,
      "loss": 0.0098,
      "step": 12680
    },
    {
      "epoch": 0.08138971821455715,
      "grad_norm": 0.9465958476066589,
      "learning_rate": 4.069394561313494e-06,
      "loss": 0.0093,
      "step": 12690
    },
    {
      "epoch": 0.08145385510834324,
      "grad_norm": 0.7725071310997009,
      "learning_rate": 4.072601334017445e-06,
      "loss": 0.0149,
      "step": 12700
    },
    {
      "epoch": 0.08151799200212935,
      "grad_norm": 0.2763436734676361,
      "learning_rate": 4.075808106721396e-06,
      "loss": 0.0111,
      "step": 12710
    },
    {
      "epoch": 0.08158212889591544,
      "grad_norm": 0.7053089141845703,
      "learning_rate": 4.079014879425347e-06,
      "loss": 0.0087,
      "step": 12720
    },
    {
      "epoch": 0.08164626578970154,
      "grad_norm": 0.6975532174110413,
      "learning_rate": 4.082221652129298e-06,
      "loss": 0.0115,
      "step": 12730
    },
    {
      "epoch": 0.08171040268348764,
      "grad_norm": 0.3494133949279785,
      "learning_rate": 4.085428424833248e-06,
      "loss": 0.01,
      "step": 12740
    },
    {
      "epoch": 0.08177453957727374,
      "grad_norm": 0.6423804759979248,
      "learning_rate": 4.088635197537199e-06,
      "loss": 0.0122,
      "step": 12750
    },
    {
      "epoch": 0.08183867647105983,
      "grad_norm": 0.5009698271751404,
      "learning_rate": 4.091841970241149e-06,
      "loss": 0.0092,
      "step": 12760
    },
    {
      "epoch": 0.08190281336484592,
      "grad_norm": 0.36906102299690247,
      "learning_rate": 4.0950487429451005e-06,
      "loss": 0.0079,
      "step": 12770
    },
    {
      "epoch": 0.08196695025863203,
      "grad_norm": 0.6030924320220947,
      "learning_rate": 4.098255515649051e-06,
      "loss": 0.014,
      "step": 12780
    },
    {
      "epoch": 0.08203108715241812,
      "grad_norm": 0.6758723855018616,
      "learning_rate": 4.101462288353002e-06,
      "loss": 0.0129,
      "step": 12790
    },
    {
      "epoch": 0.08209522404620422,
      "grad_norm": 0.4944661855697632,
      "learning_rate": 4.104669061056953e-06,
      "loss": 0.0101,
      "step": 12800
    },
    {
      "epoch": 0.08215936093999031,
      "grad_norm": 0.4592941403388977,
      "learning_rate": 4.107875833760903e-06,
      "loss": 0.0112,
      "step": 12810
    },
    {
      "epoch": 0.08222349783377642,
      "grad_norm": 0.9778560996055603,
      "learning_rate": 4.111082606464854e-06,
      "loss": 0.0123,
      "step": 12820
    },
    {
      "epoch": 0.08228763472756251,
      "grad_norm": 0.9011563658714294,
      "learning_rate": 4.114289379168805e-06,
      "loss": 0.0093,
      "step": 12830
    },
    {
      "epoch": 0.0823517716213486,
      "grad_norm": 0.7265444397926331,
      "learning_rate": 4.117496151872756e-06,
      "loss": 0.0122,
      "step": 12840
    },
    {
      "epoch": 0.08241590851513471,
      "grad_norm": 0.29877620935440063,
      "learning_rate": 4.120702924576706e-06,
      "loss": 0.0109,
      "step": 12850
    },
    {
      "epoch": 0.0824800454089208,
      "grad_norm": 0.8235241770744324,
      "learning_rate": 4.123909697280657e-06,
      "loss": 0.0106,
      "step": 12860
    },
    {
      "epoch": 0.0825441823027069,
      "grad_norm": 0.7094138860702515,
      "learning_rate": 4.127116469984608e-06,
      "loss": 0.0087,
      "step": 12870
    },
    {
      "epoch": 0.08260831919649299,
      "grad_norm": 0.6995874047279358,
      "learning_rate": 4.1303232426885585e-06,
      "loss": 0.0107,
      "step": 12880
    },
    {
      "epoch": 0.0826724560902791,
      "grad_norm": 0.9243060350418091,
      "learning_rate": 4.13353001539251e-06,
      "loss": 0.0139,
      "step": 12890
    },
    {
      "epoch": 0.08273659298406519,
      "grad_norm": 0.4975259006023407,
      "learning_rate": 4.13673678809646e-06,
      "loss": 0.0127,
      "step": 12900
    },
    {
      "epoch": 0.08280072987785128,
      "grad_norm": 0.6469900012016296,
      "learning_rate": 4.139943560800411e-06,
      "loss": 0.0167,
      "step": 12910
    },
    {
      "epoch": 0.08286486677163739,
      "grad_norm": 0.7069517970085144,
      "learning_rate": 4.143150333504361e-06,
      "loss": 0.0114,
      "step": 12920
    },
    {
      "epoch": 0.08292900366542348,
      "grad_norm": 0.690664529800415,
      "learning_rate": 4.146357106208312e-06,
      "loss": 0.0157,
      "step": 12930
    },
    {
      "epoch": 0.08299314055920957,
      "grad_norm": 0.48529475927352905,
      "learning_rate": 4.1495638789122635e-06,
      "loss": 0.015,
      "step": 12940
    },
    {
      "epoch": 0.08305727745299567,
      "grad_norm": 0.3557664155960083,
      "learning_rate": 4.152770651616214e-06,
      "loss": 0.0097,
      "step": 12950
    },
    {
      "epoch": 0.08312141434678177,
      "grad_norm": 0.7546350955963135,
      "learning_rate": 4.155977424320165e-06,
      "loss": 0.0131,
      "step": 12960
    },
    {
      "epoch": 0.08318555124056787,
      "grad_norm": 0.6388615965843201,
      "learning_rate": 4.159184197024115e-06,
      "loss": 0.0113,
      "step": 12970
    },
    {
      "epoch": 0.08324968813435396,
      "grad_norm": 0.41935300827026367,
      "learning_rate": 4.162390969728066e-06,
      "loss": 0.0105,
      "step": 12980
    },
    {
      "epoch": 0.08331382502814007,
      "grad_norm": 0.8952448964118958,
      "learning_rate": 4.1655977424320165e-06,
      "loss": 0.012,
      "step": 12990
    },
    {
      "epoch": 0.08337796192192616,
      "grad_norm": 1.4644192457199097,
      "learning_rate": 4.168804515135968e-06,
      "loss": 0.0117,
      "step": 13000
    },
    {
      "epoch": 0.08344209881571225,
      "grad_norm": 0.7456547021865845,
      "learning_rate": 4.172011287839918e-06,
      "loss": 0.0084,
      "step": 13010
    },
    {
      "epoch": 0.08350623570949835,
      "grad_norm": 0.5758523344993591,
      "learning_rate": 4.175218060543869e-06,
      "loss": 0.0144,
      "step": 13020
    },
    {
      "epoch": 0.08357037260328445,
      "grad_norm": 0.4921818971633911,
      "learning_rate": 4.17842483324782e-06,
      "loss": 0.0104,
      "step": 13030
    },
    {
      "epoch": 0.08363450949707055,
      "grad_norm": 0.3467360734939575,
      "learning_rate": 4.18163160595177e-06,
      "loss": 0.012,
      "step": 13040
    },
    {
      "epoch": 0.08369864639085664,
      "grad_norm": 0.9223970174789429,
      "learning_rate": 4.1848383786557215e-06,
      "loss": 0.0113,
      "step": 13050
    },
    {
      "epoch": 0.08376278328464275,
      "grad_norm": 0.6632959246635437,
      "learning_rate": 4.188045151359672e-06,
      "loss": 0.0158,
      "step": 13060
    },
    {
      "epoch": 0.08382692017842884,
      "grad_norm": 0.754456102848053,
      "learning_rate": 4.191251924063623e-06,
      "loss": 0.0139,
      "step": 13070
    },
    {
      "epoch": 0.08389105707221493,
      "grad_norm": 0.7275996804237366,
      "learning_rate": 4.194458696767573e-06,
      "loss": 0.0101,
      "step": 13080
    },
    {
      "epoch": 0.08395519396600103,
      "grad_norm": 0.7685105204582214,
      "learning_rate": 4.197665469471524e-06,
      "loss": 0.0131,
      "step": 13090
    },
    {
      "epoch": 0.08401933085978713,
      "grad_norm": 0.4401727616786957,
      "learning_rate": 4.200872242175475e-06,
      "loss": 0.0107,
      "step": 13100
    },
    {
      "epoch": 0.08408346775357323,
      "grad_norm": 0.9203099012374878,
      "learning_rate": 4.204079014879426e-06,
      "loss": 0.0132,
      "step": 13110
    },
    {
      "epoch": 0.08414760464735932,
      "grad_norm": 0.4970671236515045,
      "learning_rate": 4.207285787583377e-06,
      "loss": 0.0128,
      "step": 13120
    },
    {
      "epoch": 0.08421174154114543,
      "grad_norm": 0.4000563621520996,
      "learning_rate": 4.210492560287327e-06,
      "loss": 0.0114,
      "step": 13130
    },
    {
      "epoch": 0.08427587843493152,
      "grad_norm": 0.5417554378509521,
      "learning_rate": 4.213699332991278e-06,
      "loss": 0.0117,
      "step": 13140
    },
    {
      "epoch": 0.08434001532871761,
      "grad_norm": 0.47524988651275635,
      "learning_rate": 4.216906105695228e-06,
      "loss": 0.0079,
      "step": 13150
    },
    {
      "epoch": 0.0844041522225037,
      "grad_norm": 0.502637505531311,
      "learning_rate": 4.2201128783991795e-06,
      "loss": 0.0124,
      "step": 13160
    },
    {
      "epoch": 0.08446828911628981,
      "grad_norm": 0.45672890543937683,
      "learning_rate": 4.22331965110313e-06,
      "loss": 0.0117,
      "step": 13170
    },
    {
      "epoch": 0.0845324260100759,
      "grad_norm": 0.5679007172584534,
      "learning_rate": 4.226526423807081e-06,
      "loss": 0.0093,
      "step": 13180
    },
    {
      "epoch": 0.084596562903862,
      "grad_norm": 0.5174263715744019,
      "learning_rate": 4.229733196511032e-06,
      "loss": 0.0087,
      "step": 13190
    },
    {
      "epoch": 0.0846606997976481,
      "grad_norm": 0.6781324148178101,
      "learning_rate": 4.232939969214982e-06,
      "loss": 0.0082,
      "step": 13200
    },
    {
      "epoch": 0.0847248366914342,
      "grad_norm": 0.8433681130409241,
      "learning_rate": 4.236146741918933e-06,
      "loss": 0.0116,
      "step": 13210
    },
    {
      "epoch": 0.08478897358522029,
      "grad_norm": 0.4254860579967499,
      "learning_rate": 4.239353514622884e-06,
      "loss": 0.0111,
      "step": 13220
    },
    {
      "epoch": 0.08485311047900639,
      "grad_norm": 0.4683290719985962,
      "learning_rate": 4.242560287326835e-06,
      "loss": 0.008,
      "step": 13230
    },
    {
      "epoch": 0.08491724737279249,
      "grad_norm": 0.9889957308769226,
      "learning_rate": 4.245767060030785e-06,
      "loss": 0.0127,
      "step": 13240
    },
    {
      "epoch": 0.08498138426657859,
      "grad_norm": 0.9414776563644409,
      "learning_rate": 4.248973832734736e-06,
      "loss": 0.0111,
      "step": 13250
    },
    {
      "epoch": 0.08504552116036468,
      "grad_norm": 0.6853829622268677,
      "learning_rate": 4.252180605438686e-06,
      "loss": 0.0146,
      "step": 13260
    },
    {
      "epoch": 0.08510965805415079,
      "grad_norm": 0.8550167679786682,
      "learning_rate": 4.2553873781426375e-06,
      "loss": 0.0153,
      "step": 13270
    },
    {
      "epoch": 0.08517379494793688,
      "grad_norm": 0.49315908551216125,
      "learning_rate": 4.258594150846589e-06,
      "loss": 0.0079,
      "step": 13280
    },
    {
      "epoch": 0.08523793184172297,
      "grad_norm": 0.842653751373291,
      "learning_rate": 4.261800923550539e-06,
      "loss": 0.0131,
      "step": 13290
    },
    {
      "epoch": 0.08530206873550907,
      "grad_norm": 0.841410219669342,
      "learning_rate": 4.26500769625449e-06,
      "loss": 0.0104,
      "step": 13300
    },
    {
      "epoch": 0.08536620562929517,
      "grad_norm": 0.7665841579437256,
      "learning_rate": 4.26821446895844e-06,
      "loss": 0.0107,
      "step": 13310
    },
    {
      "epoch": 0.08543034252308127,
      "grad_norm": 0.44842883944511414,
      "learning_rate": 4.271421241662391e-06,
      "loss": 0.0143,
      "step": 13320
    },
    {
      "epoch": 0.08549447941686736,
      "grad_norm": 0.47715622186660767,
      "learning_rate": 4.2746280143663425e-06,
      "loss": 0.0125,
      "step": 13330
    },
    {
      "epoch": 0.08555861631065347,
      "grad_norm": 0.6261366009712219,
      "learning_rate": 4.277834787070293e-06,
      "loss": 0.008,
      "step": 13340
    },
    {
      "epoch": 0.08562275320443956,
      "grad_norm": 0.6199816465377808,
      "learning_rate": 4.281041559774243e-06,
      "loss": 0.0099,
      "step": 13350
    },
    {
      "epoch": 0.08568689009822565,
      "grad_norm": 0.6077964901924133,
      "learning_rate": 4.284248332478194e-06,
      "loss": 0.0094,
      "step": 13360
    },
    {
      "epoch": 0.08575102699201174,
      "grad_norm": 0.5811752676963806,
      "learning_rate": 4.287455105182145e-06,
      "loss": 0.0094,
      "step": 13370
    },
    {
      "epoch": 0.08581516388579785,
      "grad_norm": 0.5186547636985779,
      "learning_rate": 4.2906618778860955e-06,
      "loss": 0.0089,
      "step": 13380
    },
    {
      "epoch": 0.08587930077958394,
      "grad_norm": 0.6199879050254822,
      "learning_rate": 4.293868650590047e-06,
      "loss": 0.0131,
      "step": 13390
    },
    {
      "epoch": 0.08594343767337004,
      "grad_norm": 0.5385717153549194,
      "learning_rate": 4.297075423293997e-06,
      "loss": 0.0155,
      "step": 13400
    },
    {
      "epoch": 0.08600757456715614,
      "grad_norm": 0.5268189907073975,
      "learning_rate": 4.300282195997948e-06,
      "loss": 0.0094,
      "step": 13410
    },
    {
      "epoch": 0.08607171146094224,
      "grad_norm": 1.222302794456482,
      "learning_rate": 4.303488968701899e-06,
      "loss": 0.0145,
      "step": 13420
    },
    {
      "epoch": 0.08613584835472833,
      "grad_norm": 0.36908653378486633,
      "learning_rate": 4.306695741405849e-06,
      "loss": 0.0103,
      "step": 13430
    },
    {
      "epoch": 0.08619998524851442,
      "grad_norm": 1.2218557596206665,
      "learning_rate": 4.3099025141098e-06,
      "loss": 0.0094,
      "step": 13440
    },
    {
      "epoch": 0.08626412214230053,
      "grad_norm": 1.115786075592041,
      "learning_rate": 4.313109286813751e-06,
      "loss": 0.0106,
      "step": 13450
    },
    {
      "epoch": 0.08632825903608662,
      "grad_norm": 0.48821571469306946,
      "learning_rate": 4.316316059517702e-06,
      "loss": 0.0102,
      "step": 13460
    },
    {
      "epoch": 0.08639239592987272,
      "grad_norm": 0.3381451964378357,
      "learning_rate": 4.319522832221653e-06,
      "loss": 0.0078,
      "step": 13470
    },
    {
      "epoch": 0.08645653282365882,
      "grad_norm": 0.682102382183075,
      "learning_rate": 4.322729604925603e-06,
      "loss": 0.01,
      "step": 13480
    },
    {
      "epoch": 0.08652066971744492,
      "grad_norm": 0.3813456892967224,
      "learning_rate": 4.3259363776295535e-06,
      "loss": 0.0113,
      "step": 13490
    },
    {
      "epoch": 0.08658480661123101,
      "grad_norm": 0.4198649823665619,
      "learning_rate": 4.329143150333505e-06,
      "loss": 0.0084,
      "step": 13500
    },
    {
      "epoch": 0.0866489435050171,
      "grad_norm": 0.5584994554519653,
      "learning_rate": 4.332349923037456e-06,
      "loss": 0.0132,
      "step": 13510
    },
    {
      "epoch": 0.08671308039880321,
      "grad_norm": 0.490875780582428,
      "learning_rate": 4.335556695741406e-06,
      "loss": 0.0111,
      "step": 13520
    },
    {
      "epoch": 0.0867772172925893,
      "grad_norm": 0.3352775275707245,
      "learning_rate": 4.338763468445356e-06,
      "loss": 0.011,
      "step": 13530
    },
    {
      "epoch": 0.0868413541863754,
      "grad_norm": 0.7659197449684143,
      "learning_rate": 4.341970241149307e-06,
      "loss": 0.0131,
      "step": 13540
    },
    {
      "epoch": 0.08690549108016149,
      "grad_norm": 0.5991475582122803,
      "learning_rate": 4.3451770138532585e-06,
      "loss": 0.0102,
      "step": 13550
    },
    {
      "epoch": 0.0869696279739476,
      "grad_norm": 0.6653827428817749,
      "learning_rate": 4.34838378655721e-06,
      "loss": 0.009,
      "step": 13560
    },
    {
      "epoch": 0.08703376486773369,
      "grad_norm": 0.39620816707611084,
      "learning_rate": 4.35159055926116e-06,
      "loss": 0.0076,
      "step": 13570
    },
    {
      "epoch": 0.08709790176151978,
      "grad_norm": 1.3507369756698608,
      "learning_rate": 4.35479733196511e-06,
      "loss": 0.0089,
      "step": 13580
    },
    {
      "epoch": 0.08716203865530589,
      "grad_norm": 0.7978771924972534,
      "learning_rate": 4.358004104669061e-06,
      "loss": 0.0097,
      "step": 13590
    },
    {
      "epoch": 0.08722617554909198,
      "grad_norm": 0.5015578269958496,
      "learning_rate": 4.361210877373012e-06,
      "loss": 0.0126,
      "step": 13600
    },
    {
      "epoch": 0.08729031244287808,
      "grad_norm": 0.8807877898216248,
      "learning_rate": 4.364417650076963e-06,
      "loss": 0.0109,
      "step": 13610
    },
    {
      "epoch": 0.08735444933666417,
      "grad_norm": 0.7356600761413574,
      "learning_rate": 4.367624422780914e-06,
      "loss": 0.0103,
      "step": 13620
    },
    {
      "epoch": 0.08741858623045028,
      "grad_norm": 0.839003324508667,
      "learning_rate": 4.370831195484864e-06,
      "loss": 0.012,
      "step": 13630
    },
    {
      "epoch": 0.08748272312423637,
      "grad_norm": 0.6434532403945923,
      "learning_rate": 4.374037968188815e-06,
      "loss": 0.0135,
      "step": 13640
    },
    {
      "epoch": 0.08754686001802246,
      "grad_norm": 0.8763306140899658,
      "learning_rate": 4.377244740892766e-06,
      "loss": 0.0102,
      "step": 13650
    },
    {
      "epoch": 0.08761099691180857,
      "grad_norm": 0.4847932457923889,
      "learning_rate": 4.3804515135967165e-06,
      "loss": 0.0129,
      "step": 13660
    },
    {
      "epoch": 0.08767513380559466,
      "grad_norm": 0.3382943868637085,
      "learning_rate": 4.383658286300667e-06,
      "loss": 0.0087,
      "step": 13670
    },
    {
      "epoch": 0.08773927069938076,
      "grad_norm": 0.5592327117919922,
      "learning_rate": 4.386865059004618e-06,
      "loss": 0.008,
      "step": 13680
    },
    {
      "epoch": 0.08780340759316685,
      "grad_norm": 1.143094539642334,
      "learning_rate": 4.390071831708569e-06,
      "loss": 0.0128,
      "step": 13690
    },
    {
      "epoch": 0.08786754448695296,
      "grad_norm": 0.4922727644443512,
      "learning_rate": 4.39327860441252e-06,
      "loss": 0.0093,
      "step": 13700
    },
    {
      "epoch": 0.08793168138073905,
      "grad_norm": 0.504297137260437,
      "learning_rate": 4.39648537711647e-06,
      "loss": 0.0118,
      "step": 13710
    },
    {
      "epoch": 0.08799581827452514,
      "grad_norm": 0.5197631120681763,
      "learning_rate": 4.399692149820421e-06,
      "loss": 0.0098,
      "step": 13720
    },
    {
      "epoch": 0.08805995516831125,
      "grad_norm": 0.8234639167785645,
      "learning_rate": 4.402898922524372e-06,
      "loss": 0.0112,
      "step": 13730
    },
    {
      "epoch": 0.08812409206209734,
      "grad_norm": 0.1754753142595291,
      "learning_rate": 4.406105695228323e-06,
      "loss": 0.0056,
      "step": 13740
    },
    {
      "epoch": 0.08818822895588344,
      "grad_norm": 0.4347337484359741,
      "learning_rate": 4.409312467932273e-06,
      "loss": 0.0097,
      "step": 13750
    },
    {
      "epoch": 0.08825236584966953,
      "grad_norm": 0.875468909740448,
      "learning_rate": 4.412519240636224e-06,
      "loss": 0.0118,
      "step": 13760
    },
    {
      "epoch": 0.08831650274345564,
      "grad_norm": 0.7216804623603821,
      "learning_rate": 4.4157260133401745e-06,
      "loss": 0.0107,
      "step": 13770
    },
    {
      "epoch": 0.08838063963724173,
      "grad_norm": 0.7757262587547302,
      "learning_rate": 4.418932786044126e-06,
      "loss": 0.0146,
      "step": 13780
    },
    {
      "epoch": 0.08844477653102782,
      "grad_norm": 0.23120538890361786,
      "learning_rate": 4.422139558748077e-06,
      "loss": 0.0069,
      "step": 13790
    },
    {
      "epoch": 0.08850891342481393,
      "grad_norm": 0.4751850664615631,
      "learning_rate": 4.425346331452027e-06,
      "loss": 0.0133,
      "step": 13800
    },
    {
      "epoch": 0.08857305031860002,
      "grad_norm": 0.7578360438346863,
      "learning_rate": 4.428553104155977e-06,
      "loss": 0.0221,
      "step": 13810
    },
    {
      "epoch": 0.08863718721238611,
      "grad_norm": 0.5103495121002197,
      "learning_rate": 4.431759876859928e-06,
      "loss": 0.0081,
      "step": 13820
    },
    {
      "epoch": 0.08870132410617221,
      "grad_norm": 0.7559943795204163,
      "learning_rate": 4.4349666495638795e-06,
      "loss": 0.0161,
      "step": 13830
    },
    {
      "epoch": 0.08876546099995831,
      "grad_norm": 0.6206265687942505,
      "learning_rate": 4.438173422267831e-06,
      "loss": 0.0163,
      "step": 13840
    },
    {
      "epoch": 0.08882959789374441,
      "grad_norm": 0.5653325915336609,
      "learning_rate": 4.441380194971781e-06,
      "loss": 0.0116,
      "step": 13850
    },
    {
      "epoch": 0.0888937347875305,
      "grad_norm": 0.28707340359687805,
      "learning_rate": 4.444586967675731e-06,
      "loss": 0.0078,
      "step": 13860
    },
    {
      "epoch": 0.08895787168131661,
      "grad_norm": 0.45919936895370483,
      "learning_rate": 4.447793740379682e-06,
      "loss": 0.0127,
      "step": 13870
    },
    {
      "epoch": 0.0890220085751027,
      "grad_norm": 0.4506450593471527,
      "learning_rate": 4.451000513083633e-06,
      "loss": 0.0092,
      "step": 13880
    },
    {
      "epoch": 0.0890861454688888,
      "grad_norm": 0.8949733376502991,
      "learning_rate": 4.454207285787584e-06,
      "loss": 0.0094,
      "step": 13890
    },
    {
      "epoch": 0.08915028236267489,
      "grad_norm": 0.4191751480102539,
      "learning_rate": 4.457414058491534e-06,
      "loss": 0.0109,
      "step": 13900
    },
    {
      "epoch": 0.089214419256461,
      "grad_norm": 0.8228646516799927,
      "learning_rate": 4.460620831195485e-06,
      "loss": 0.0099,
      "step": 13910
    },
    {
      "epoch": 0.08927855615024709,
      "grad_norm": 0.8701620697975159,
      "learning_rate": 4.463827603899436e-06,
      "loss": 0.0129,
      "step": 13920
    },
    {
      "epoch": 0.08934269304403318,
      "grad_norm": 0.6736899614334106,
      "learning_rate": 4.467034376603387e-06,
      "loss": 0.0072,
      "step": 13930
    },
    {
      "epoch": 0.08940682993781929,
      "grad_norm": 0.4631296396255493,
      "learning_rate": 4.4702411493073375e-06,
      "loss": 0.0123,
      "step": 13940
    },
    {
      "epoch": 0.08947096683160538,
      "grad_norm": 1.0359212160110474,
      "learning_rate": 4.473447922011288e-06,
      "loss": 0.0125,
      "step": 13950
    },
    {
      "epoch": 0.08953510372539147,
      "grad_norm": 0.6013842821121216,
      "learning_rate": 4.476654694715239e-06,
      "loss": 0.012,
      "step": 13960
    },
    {
      "epoch": 0.08959924061917757,
      "grad_norm": 0.4382389485836029,
      "learning_rate": 4.47986146741919e-06,
      "loss": 0.0132,
      "step": 13970
    },
    {
      "epoch": 0.08966337751296367,
      "grad_norm": 0.5488568544387817,
      "learning_rate": 4.48306824012314e-06,
      "loss": 0.0105,
      "step": 13980
    },
    {
      "epoch": 0.08972751440674977,
      "grad_norm": 0.7292414903640747,
      "learning_rate": 4.486275012827091e-06,
      "loss": 0.0119,
      "step": 13990
    },
    {
      "epoch": 0.08979165130053586,
      "grad_norm": 0.930444598197937,
      "learning_rate": 4.489481785531042e-06,
      "loss": 0.0086,
      "step": 14000
    },
    {
      "epoch": 0.08985578819432197,
      "grad_norm": 0.6052126884460449,
      "learning_rate": 4.492688558234993e-06,
      "loss": 0.0104,
      "step": 14010
    },
    {
      "epoch": 0.08991992508810806,
      "grad_norm": 0.46579042077064514,
      "learning_rate": 4.495895330938944e-06,
      "loss": 0.0082,
      "step": 14020
    },
    {
      "epoch": 0.08998406198189415,
      "grad_norm": 0.8902080655097961,
      "learning_rate": 4.499102103642894e-06,
      "loss": 0.0098,
      "step": 14030
    },
    {
      "epoch": 0.09004819887568025,
      "grad_norm": 0.9307649731636047,
      "learning_rate": 4.502308876346844e-06,
      "loss": 0.0131,
      "step": 14040
    },
    {
      "epoch": 0.09011233576946635,
      "grad_norm": 0.33789774775505066,
      "learning_rate": 4.5055156490507955e-06,
      "loss": 0.0101,
      "step": 14050
    },
    {
      "epoch": 0.09017647266325245,
      "grad_norm": 0.8060634136199951,
      "learning_rate": 4.508722421754747e-06,
      "loss": 0.0096,
      "step": 14060
    },
    {
      "epoch": 0.09024060955703854,
      "grad_norm": 0.5572049021720886,
      "learning_rate": 4.511929194458698e-06,
      "loss": 0.0084,
      "step": 14070
    },
    {
      "epoch": 0.09030474645082465,
      "grad_norm": 0.6087406277656555,
      "learning_rate": 4.515135967162648e-06,
      "loss": 0.0089,
      "step": 14080
    },
    {
      "epoch": 0.09036888334461074,
      "grad_norm": 0.681932270526886,
      "learning_rate": 4.518342739866598e-06,
      "loss": 0.0132,
      "step": 14090
    },
    {
      "epoch": 0.09043302023839683,
      "grad_norm": 0.38711613416671753,
      "learning_rate": 4.521549512570549e-06,
      "loss": 0.0086,
      "step": 14100
    },
    {
      "epoch": 0.09049715713218293,
      "grad_norm": 0.24457596242427826,
      "learning_rate": 4.5247562852745005e-06,
      "loss": 0.0125,
      "step": 14110
    },
    {
      "epoch": 0.09056129402596903,
      "grad_norm": 0.3963066339492798,
      "learning_rate": 4.527963057978451e-06,
      "loss": 0.0091,
      "step": 14120
    },
    {
      "epoch": 0.09062543091975513,
      "grad_norm": 0.30504074692726135,
      "learning_rate": 4.531169830682401e-06,
      "loss": 0.0109,
      "step": 14130
    },
    {
      "epoch": 0.09068956781354122,
      "grad_norm": 0.5058887004852295,
      "learning_rate": 4.534376603386352e-06,
      "loss": 0.0176,
      "step": 14140
    },
    {
      "epoch": 0.09075370470732733,
      "grad_norm": 0.29121342301368713,
      "learning_rate": 4.537583376090303e-06,
      "loss": 0.0088,
      "step": 14150
    },
    {
      "epoch": 0.09081784160111342,
      "grad_norm": 0.9375796318054199,
      "learning_rate": 4.540790148794254e-06,
      "loss": 0.0138,
      "step": 14160
    },
    {
      "epoch": 0.09088197849489951,
      "grad_norm": 1.0603824853897095,
      "learning_rate": 4.543996921498205e-06,
      "loss": 0.0111,
      "step": 14170
    },
    {
      "epoch": 0.0909461153886856,
      "grad_norm": 0.8005550503730774,
      "learning_rate": 4.547203694202155e-06,
      "loss": 0.014,
      "step": 14180
    },
    {
      "epoch": 0.09101025228247171,
      "grad_norm": 0.4290587604045868,
      "learning_rate": 4.550410466906106e-06,
      "loss": 0.0159,
      "step": 14190
    },
    {
      "epoch": 0.0910743891762578,
      "grad_norm": 0.436431884765625,
      "learning_rate": 4.553617239610057e-06,
      "loss": 0.0112,
      "step": 14200
    },
    {
      "epoch": 0.0911385260700439,
      "grad_norm": 0.5074294209480286,
      "learning_rate": 4.556824012314007e-06,
      "loss": 0.0081,
      "step": 14210
    },
    {
      "epoch": 0.09120266296382999,
      "grad_norm": 0.5892575979232788,
      "learning_rate": 4.5600307850179585e-06,
      "loss": 0.0094,
      "step": 14220
    },
    {
      "epoch": 0.0912667998576161,
      "grad_norm": 0.7842941284179688,
      "learning_rate": 4.563237557721909e-06,
      "loss": 0.0091,
      "step": 14230
    },
    {
      "epoch": 0.09133093675140219,
      "grad_norm": 0.582675039768219,
      "learning_rate": 4.56644433042586e-06,
      "loss": 0.0164,
      "step": 14240
    },
    {
      "epoch": 0.09139507364518829,
      "grad_norm": 0.7669952511787415,
      "learning_rate": 4.569651103129811e-06,
      "loss": 0.0146,
      "step": 14250
    },
    {
      "epoch": 0.09145921053897439,
      "grad_norm": 0.6250030398368835,
      "learning_rate": 4.572857875833761e-06,
      "loss": 0.0111,
      "step": 14260
    },
    {
      "epoch": 0.09152334743276049,
      "grad_norm": 0.7588163018226624,
      "learning_rate": 4.5760646485377115e-06,
      "loss": 0.0114,
      "step": 14270
    },
    {
      "epoch": 0.09158748432654658,
      "grad_norm": 0.6219152808189392,
      "learning_rate": 4.579271421241663e-06,
      "loss": 0.0107,
      "step": 14280
    },
    {
      "epoch": 0.09165162122033267,
      "grad_norm": 0.585282027721405,
      "learning_rate": 4.582478193945614e-06,
      "loss": 0.0141,
      "step": 14290
    },
    {
      "epoch": 0.09171575811411878,
      "grad_norm": 0.5579394698143005,
      "learning_rate": 4.585684966649565e-06,
      "loss": 0.0095,
      "step": 14300
    },
    {
      "epoch": 0.09177989500790487,
      "grad_norm": 0.5351959466934204,
      "learning_rate": 4.588891739353515e-06,
      "loss": 0.0121,
      "step": 14310
    },
    {
      "epoch": 0.09184403190169096,
      "grad_norm": 0.49338340759277344,
      "learning_rate": 4.592098512057465e-06,
      "loss": 0.0131,
      "step": 14320
    },
    {
      "epoch": 0.09190816879547707,
      "grad_norm": 0.5949311256408691,
      "learning_rate": 4.5953052847614165e-06,
      "loss": 0.0108,
      "step": 14330
    },
    {
      "epoch": 0.09197230568926316,
      "grad_norm": 0.5237823128700256,
      "learning_rate": 4.598512057465368e-06,
      "loss": 0.0128,
      "step": 14340
    },
    {
      "epoch": 0.09203644258304926,
      "grad_norm": 0.5870658755302429,
      "learning_rate": 4.601718830169318e-06,
      "loss": 0.0088,
      "step": 14350
    },
    {
      "epoch": 0.09210057947683535,
      "grad_norm": 0.9670231938362122,
      "learning_rate": 4.604925602873269e-06,
      "loss": 0.0159,
      "step": 14360
    },
    {
      "epoch": 0.09216471637062146,
      "grad_norm": 0.5894929766654968,
      "learning_rate": 4.608132375577219e-06,
      "loss": 0.0106,
      "step": 14370
    },
    {
      "epoch": 0.09222885326440755,
      "grad_norm": 0.38806742429733276,
      "learning_rate": 4.61133914828117e-06,
      "loss": 0.0112,
      "step": 14380
    },
    {
      "epoch": 0.09229299015819364,
      "grad_norm": 0.5953904986381531,
      "learning_rate": 4.6145459209851215e-06,
      "loss": 0.0101,
      "step": 14390
    },
    {
      "epoch": 0.09235712705197975,
      "grad_norm": 0.5008606910705566,
      "learning_rate": 4.617752693689072e-06,
      "loss": 0.012,
      "step": 14400
    },
    {
      "epoch": 0.09242126394576584,
      "grad_norm": 1.0356851816177368,
      "learning_rate": 4.620959466393022e-06,
      "loss": 0.0082,
      "step": 14410
    },
    {
      "epoch": 0.09248540083955194,
      "grad_norm": 0.5471305251121521,
      "learning_rate": 4.624166239096973e-06,
      "loss": 0.0083,
      "step": 14420
    },
    {
      "epoch": 0.09254953773333803,
      "grad_norm": 0.4250575304031372,
      "learning_rate": 4.627373011800924e-06,
      "loss": 0.018,
      "step": 14430
    },
    {
      "epoch": 0.09261367462712414,
      "grad_norm": 0.44478684663772583,
      "learning_rate": 4.6305797845048745e-06,
      "loss": 0.0105,
      "step": 14440
    },
    {
      "epoch": 0.09267781152091023,
      "grad_norm": 0.6541079878807068,
      "learning_rate": 4.633786557208826e-06,
      "loss": 0.009,
      "step": 14450
    },
    {
      "epoch": 0.09274194841469632,
      "grad_norm": 0.3483266234397888,
      "learning_rate": 4.636993329912776e-06,
      "loss": 0.0097,
      "step": 14460
    },
    {
      "epoch": 0.09280608530848243,
      "grad_norm": 0.34588584303855896,
      "learning_rate": 4.640200102616727e-06,
      "loss": 0.0065,
      "step": 14470
    },
    {
      "epoch": 0.09287022220226852,
      "grad_norm": 0.26781347393989563,
      "learning_rate": 4.643406875320678e-06,
      "loss": 0.0093,
      "step": 14480
    },
    {
      "epoch": 0.09293435909605462,
      "grad_norm": 0.6617163419723511,
      "learning_rate": 4.646613648024628e-06,
      "loss": 0.0112,
      "step": 14490
    },
    {
      "epoch": 0.09299849598984071,
      "grad_norm": 0.6089116930961609,
      "learning_rate": 4.649820420728579e-06,
      "loss": 0.0171,
      "step": 14500
    },
    {
      "epoch": 0.09306263288362682,
      "grad_norm": 0.2882973253726959,
      "learning_rate": 4.65302719343253e-06,
      "loss": 0.0104,
      "step": 14510
    },
    {
      "epoch": 0.09312676977741291,
      "grad_norm": 0.4431282877922058,
      "learning_rate": 4.656233966136481e-06,
      "loss": 0.01,
      "step": 14520
    },
    {
      "epoch": 0.093190906671199,
      "grad_norm": 0.6200217008590698,
      "learning_rate": 4.659440738840431e-06,
      "loss": 0.0099,
      "step": 14530
    },
    {
      "epoch": 0.09325504356498511,
      "grad_norm": 0.8319776654243469,
      "learning_rate": 4.662647511544382e-06,
      "loss": 0.0122,
      "step": 14540
    },
    {
      "epoch": 0.0933191804587712,
      "grad_norm": 0.5067435503005981,
      "learning_rate": 4.6658542842483325e-06,
      "loss": 0.013,
      "step": 14550
    },
    {
      "epoch": 0.0933833173525573,
      "grad_norm": 0.5280515551567078,
      "learning_rate": 4.669061056952284e-06,
      "loss": 0.0099,
      "step": 14560
    },
    {
      "epoch": 0.09344745424634339,
      "grad_norm": 0.6483435034751892,
      "learning_rate": 4.672267829656235e-06,
      "loss": 0.0134,
      "step": 14570
    },
    {
      "epoch": 0.0935115911401295,
      "grad_norm": 0.42229729890823364,
      "learning_rate": 4.675474602360185e-06,
      "loss": 0.0096,
      "step": 14580
    },
    {
      "epoch": 0.09357572803391559,
      "grad_norm": 0.7056224346160889,
      "learning_rate": 4.678681375064136e-06,
      "loss": 0.011,
      "step": 14590
    },
    {
      "epoch": 0.09363986492770168,
      "grad_norm": 0.636827826499939,
      "learning_rate": 4.681888147768086e-06,
      "loss": 0.0087,
      "step": 14600
    },
    {
      "epoch": 0.09370400182148779,
      "grad_norm": 0.660778820514679,
      "learning_rate": 4.6850949204720375e-06,
      "loss": 0.008,
      "step": 14610
    },
    {
      "epoch": 0.09376813871527388,
      "grad_norm": 0.4916780889034271,
      "learning_rate": 4.688301693175988e-06,
      "loss": 0.0078,
      "step": 14620
    },
    {
      "epoch": 0.09383227560905998,
      "grad_norm": 1.1069315671920776,
      "learning_rate": 4.691508465879939e-06,
      "loss": 0.0141,
      "step": 14630
    },
    {
      "epoch": 0.09389641250284607,
      "grad_norm": 0.9133396744728088,
      "learning_rate": 4.694715238583889e-06,
      "loss": 0.0077,
      "step": 14640
    },
    {
      "epoch": 0.09396054939663218,
      "grad_norm": 0.42861101031303406,
      "learning_rate": 4.69792201128784e-06,
      "loss": 0.0116,
      "step": 14650
    },
    {
      "epoch": 0.09402468629041827,
      "grad_norm": 0.5793526768684387,
      "learning_rate": 4.701128783991791e-06,
      "loss": 0.0138,
      "step": 14660
    },
    {
      "epoch": 0.09408882318420436,
      "grad_norm": 0.24592435359954834,
      "learning_rate": 4.704335556695742e-06,
      "loss": 0.0145,
      "step": 14670
    },
    {
      "epoch": 0.09415296007799047,
      "grad_norm": 0.4596893787384033,
      "learning_rate": 4.707542329399693e-06,
      "loss": 0.0143,
      "step": 14680
    },
    {
      "epoch": 0.09421709697177656,
      "grad_norm": 0.25949400663375854,
      "learning_rate": 4.710749102103643e-06,
      "loss": 0.0094,
      "step": 14690
    },
    {
      "epoch": 0.09428123386556266,
      "grad_norm": 0.7687262892723083,
      "learning_rate": 4.713955874807594e-06,
      "loss": 0.0148,
      "step": 14700
    },
    {
      "epoch": 0.09434537075934875,
      "grad_norm": 1.0646096467971802,
      "learning_rate": 4.717162647511544e-06,
      "loss": 0.0146,
      "step": 14710
    },
    {
      "epoch": 0.09440950765313486,
      "grad_norm": 0.5260186195373535,
      "learning_rate": 4.7203694202154955e-06,
      "loss": 0.0097,
      "step": 14720
    },
    {
      "epoch": 0.09447364454692095,
      "grad_norm": 0.5607770681381226,
      "learning_rate": 4.723576192919447e-06,
      "loss": 0.0103,
      "step": 14730
    },
    {
      "epoch": 0.09453778144070704,
      "grad_norm": 0.7789020538330078,
      "learning_rate": 4.726782965623397e-06,
      "loss": 0.0144,
      "step": 14740
    },
    {
      "epoch": 0.09460191833449315,
      "grad_norm": 0.48097801208496094,
      "learning_rate": 4.729989738327348e-06,
      "loss": 0.0083,
      "step": 14750
    },
    {
      "epoch": 0.09466605522827924,
      "grad_norm": 0.668645441532135,
      "learning_rate": 4.733196511031298e-06,
      "loss": 0.013,
      "step": 14760
    },
    {
      "epoch": 0.09473019212206533,
      "grad_norm": 0.4790743291378021,
      "learning_rate": 4.736403283735249e-06,
      "loss": 0.0097,
      "step": 14770
    },
    {
      "epoch": 0.09479432901585143,
      "grad_norm": 1.270419716835022,
      "learning_rate": 4.7396100564392e-06,
      "loss": 0.0091,
      "step": 14780
    },
    {
      "epoch": 0.09485846590963753,
      "grad_norm": 0.5246705412864685,
      "learning_rate": 4.742816829143151e-06,
      "loss": 0.016,
      "step": 14790
    },
    {
      "epoch": 0.09492260280342363,
      "grad_norm": 0.3491441607475281,
      "learning_rate": 4.746023601847101e-06,
      "loss": 0.013,
      "step": 14800
    },
    {
      "epoch": 0.09498673969720972,
      "grad_norm": 0.49751850962638855,
      "learning_rate": 4.749230374551052e-06,
      "loss": 0.0086,
      "step": 14810
    },
    {
      "epoch": 0.09505087659099583,
      "grad_norm": 0.7094282507896423,
      "learning_rate": 4.752437147255003e-06,
      "loss": 0.0166,
      "step": 14820
    },
    {
      "epoch": 0.09511501348478192,
      "grad_norm": 0.5809402465820312,
      "learning_rate": 4.7556439199589535e-06,
      "loss": 0.0067,
      "step": 14830
    },
    {
      "epoch": 0.09517915037856801,
      "grad_norm": 0.3063122034072876,
      "learning_rate": 4.758850692662905e-06,
      "loss": 0.0059,
      "step": 14840
    },
    {
      "epoch": 0.09524328727235411,
      "grad_norm": 0.8257580399513245,
      "learning_rate": 4.762057465366855e-06,
      "loss": 0.008,
      "step": 14850
    },
    {
      "epoch": 0.09530742416614021,
      "grad_norm": 0.5764210820198059,
      "learning_rate": 4.765264238070806e-06,
      "loss": 0.0099,
      "step": 14860
    },
    {
      "epoch": 0.09537156105992631,
      "grad_norm": 0.4129992723464966,
      "learning_rate": 4.768471010774756e-06,
      "loss": 0.0087,
      "step": 14870
    },
    {
      "epoch": 0.0954356979537124,
      "grad_norm": 0.4546699821949005,
      "learning_rate": 4.771677783478707e-06,
      "loss": 0.012,
      "step": 14880
    },
    {
      "epoch": 0.09549983484749851,
      "grad_norm": 0.5095817446708679,
      "learning_rate": 4.774884556182658e-06,
      "loss": 0.014,
      "step": 14890
    },
    {
      "epoch": 0.0955639717412846,
      "grad_norm": 0.6998573541641235,
      "learning_rate": 4.778091328886609e-06,
      "loss": 0.0123,
      "step": 14900
    },
    {
      "epoch": 0.0956281086350707,
      "grad_norm": 0.30628302693367004,
      "learning_rate": 4.78129810159056e-06,
      "loss": 0.0096,
      "step": 14910
    },
    {
      "epoch": 0.09569224552885679,
      "grad_norm": 0.42637142539024353,
      "learning_rate": 4.78450487429451e-06,
      "loss": 0.0073,
      "step": 14920
    },
    {
      "epoch": 0.0957563824226429,
      "grad_norm": 0.4169028401374817,
      "learning_rate": 4.787711646998461e-06,
      "loss": 0.0083,
      "step": 14930
    },
    {
      "epoch": 0.09582051931642899,
      "grad_norm": 0.6567114591598511,
      "learning_rate": 4.7909184197024116e-06,
      "loss": 0.0117,
      "step": 14940
    },
    {
      "epoch": 0.09588465621021508,
      "grad_norm": 0.3499446213245392,
      "learning_rate": 4.794125192406363e-06,
      "loss": 0.0184,
      "step": 14950
    },
    {
      "epoch": 0.09594879310400117,
      "grad_norm": 0.7126962542533875,
      "learning_rate": 4.797331965110314e-06,
      "loss": 0.0161,
      "step": 14960
    },
    {
      "epoch": 0.09601292999778728,
      "grad_norm": 0.5603629350662231,
      "learning_rate": 4.800538737814264e-06,
      "loss": 0.0094,
      "step": 14970
    },
    {
      "epoch": 0.09607706689157337,
      "grad_norm": 0.6502920389175415,
      "learning_rate": 4.803745510518215e-06,
      "loss": 0.0083,
      "step": 14980
    },
    {
      "epoch": 0.09614120378535947,
      "grad_norm": 0.5913206338882446,
      "learning_rate": 4.806952283222165e-06,
      "loss": 0.0095,
      "step": 14990
    },
    {
      "epoch": 0.09620534067914557,
      "grad_norm": 0.809626579284668,
      "learning_rate": 4.8101590559261165e-06,
      "loss": 0.0097,
      "step": 15000
    },
    {
      "epoch": 0.09626947757293167,
      "grad_norm": 0.544957160949707,
      "learning_rate": 4.813365828630067e-06,
      "loss": 0.0137,
      "step": 15010
    },
    {
      "epoch": 0.09633361446671776,
      "grad_norm": 0.7568247318267822,
      "learning_rate": 4.816572601334018e-06,
      "loss": 0.0137,
      "step": 15020
    },
    {
      "epoch": 0.09639775136050385,
      "grad_norm": 0.6684417724609375,
      "learning_rate": 4.819779374037968e-06,
      "loss": 0.0097,
      "step": 15030
    },
    {
      "epoch": 0.09646188825428996,
      "grad_norm": 0.2786485254764557,
      "learning_rate": 4.822986146741919e-06,
      "loss": 0.0089,
      "step": 15040
    },
    {
      "epoch": 0.09652602514807605,
      "grad_norm": 0.5178800225257874,
      "learning_rate": 4.82619291944587e-06,
      "loss": 0.0108,
      "step": 15050
    },
    {
      "epoch": 0.09659016204186215,
      "grad_norm": 0.27339088916778564,
      "learning_rate": 4.829399692149821e-06,
      "loss": 0.0088,
      "step": 15060
    },
    {
      "epoch": 0.09665429893564825,
      "grad_norm": 0.39263832569122314,
      "learning_rate": 4.832606464853772e-06,
      "loss": 0.0081,
      "step": 15070
    },
    {
      "epoch": 0.09671843582943435,
      "grad_norm": 0.7981494069099426,
      "learning_rate": 4.835813237557722e-06,
      "loss": 0.0115,
      "step": 15080
    },
    {
      "epoch": 0.09678257272322044,
      "grad_norm": 0.8040106296539307,
      "learning_rate": 4.839020010261673e-06,
      "loss": 0.0087,
      "step": 15090
    },
    {
      "epoch": 0.09684670961700653,
      "grad_norm": 0.39302074909210205,
      "learning_rate": 4.842226782965624e-06,
      "loss": 0.0134,
      "step": 15100
    },
    {
      "epoch": 0.09691084651079264,
      "grad_norm": 0.26070883870124817,
      "learning_rate": 4.8454335556695745e-06,
      "loss": 0.011,
      "step": 15110
    },
    {
      "epoch": 0.09697498340457873,
      "grad_norm": 0.4337320327758789,
      "learning_rate": 4.848640328373525e-06,
      "loss": 0.0092,
      "step": 15120
    },
    {
      "epoch": 0.09703912029836483,
      "grad_norm": 0.7534663677215576,
      "learning_rate": 4.851847101077476e-06,
      "loss": 0.0131,
      "step": 15130
    },
    {
      "epoch": 0.09710325719215093,
      "grad_norm": 0.891177237033844,
      "learning_rate": 4.855053873781427e-06,
      "loss": 0.0096,
      "step": 15140
    },
    {
      "epoch": 0.09716739408593703,
      "grad_norm": 0.6958470344543457,
      "learning_rate": 4.858260646485377e-06,
      "loss": 0.0123,
      "step": 15150
    },
    {
      "epoch": 0.09723153097972312,
      "grad_norm": 0.5443466305732727,
      "learning_rate": 4.861467419189328e-06,
      "loss": 0.0115,
      "step": 15160
    },
    {
      "epoch": 0.09729566787350921,
      "grad_norm": 0.3978094160556793,
      "learning_rate": 4.864674191893279e-06,
      "loss": 0.01,
      "step": 15170
    },
    {
      "epoch": 0.09735980476729532,
      "grad_norm": 0.6599048972129822,
      "learning_rate": 4.86788096459723e-06,
      "loss": 0.0092,
      "step": 15180
    },
    {
      "epoch": 0.09742394166108141,
      "grad_norm": 0.490038126707077,
      "learning_rate": 4.871087737301181e-06,
      "loss": 0.0075,
      "step": 15190
    },
    {
      "epoch": 0.0974880785548675,
      "grad_norm": 0.5247456431388855,
      "learning_rate": 4.874294510005131e-06,
      "loss": 0.0101,
      "step": 15200
    },
    {
      "epoch": 0.09755221544865361,
      "grad_norm": 0.7498074769973755,
      "learning_rate": 4.8775012827090814e-06,
      "loss": 0.0076,
      "step": 15210
    },
    {
      "epoch": 0.0976163523424397,
      "grad_norm": 0.5493730902671814,
      "learning_rate": 4.8807080554130326e-06,
      "loss": 0.0084,
      "step": 15220
    },
    {
      "epoch": 0.0976804892362258,
      "grad_norm": 0.5536153316497803,
      "learning_rate": 4.883914828116984e-06,
      "loss": 0.0116,
      "step": 15230
    },
    {
      "epoch": 0.09774462613001189,
      "grad_norm": 0.5220369100570679,
      "learning_rate": 4.887121600820934e-06,
      "loss": 0.0066,
      "step": 15240
    },
    {
      "epoch": 0.097808763023798,
      "grad_norm": 0.8064947724342346,
      "learning_rate": 4.890328373524885e-06,
      "loss": 0.0128,
      "step": 15250
    },
    {
      "epoch": 0.09787289991758409,
      "grad_norm": 0.42613735795021057,
      "learning_rate": 4.893535146228835e-06,
      "loss": 0.0115,
      "step": 15260
    },
    {
      "epoch": 0.09793703681137018,
      "grad_norm": 0.5674329996109009,
      "learning_rate": 4.8967419189327864e-06,
      "loss": 0.0127,
      "step": 15270
    },
    {
      "epoch": 0.09800117370515629,
      "grad_norm": 0.30216166377067566,
      "learning_rate": 4.8999486916367375e-06,
      "loss": 0.0109,
      "step": 15280
    },
    {
      "epoch": 0.09806531059894238,
      "grad_norm": 0.6081557273864746,
      "learning_rate": 4.903155464340688e-06,
      "loss": 0.014,
      "step": 15290
    },
    {
      "epoch": 0.09812944749272848,
      "grad_norm": 0.8605742454528809,
      "learning_rate": 4.906362237044638e-06,
      "loss": 0.0115,
      "step": 15300
    },
    {
      "epoch": 0.09819358438651457,
      "grad_norm": 0.5217424631118774,
      "learning_rate": 4.909569009748589e-06,
      "loss": 0.0076,
      "step": 15310
    },
    {
      "epoch": 0.09825772128030068,
      "grad_norm": 0.6119788885116577,
      "learning_rate": 4.91277578245254e-06,
      "loss": 0.0125,
      "step": 15320
    },
    {
      "epoch": 0.09832185817408677,
      "grad_norm": 0.6533595323562622,
      "learning_rate": 4.915982555156491e-06,
      "loss": 0.0101,
      "step": 15330
    },
    {
      "epoch": 0.09838599506787286,
      "grad_norm": 0.30561283230781555,
      "learning_rate": 4.919189327860442e-06,
      "loss": 0.0084,
      "step": 15340
    },
    {
      "epoch": 0.09845013196165897,
      "grad_norm": 0.4871523678302765,
      "learning_rate": 4.922396100564392e-06,
      "loss": 0.0095,
      "step": 15350
    },
    {
      "epoch": 0.09851426885544506,
      "grad_norm": 0.6592676043510437,
      "learning_rate": 4.925602873268343e-06,
      "loss": 0.0114,
      "step": 15360
    },
    {
      "epoch": 0.09857840574923116,
      "grad_norm": 0.9625756144523621,
      "learning_rate": 4.928809645972294e-06,
      "loss": 0.0124,
      "step": 15370
    },
    {
      "epoch": 0.09864254264301725,
      "grad_norm": 0.4479202628135681,
      "learning_rate": 4.9320164186762444e-06,
      "loss": 0.0118,
      "step": 15380
    },
    {
      "epoch": 0.09870667953680336,
      "grad_norm": 0.7261309623718262,
      "learning_rate": 4.935223191380195e-06,
      "loss": 0.0095,
      "step": 15390
    },
    {
      "epoch": 0.09877081643058945,
      "grad_norm": 0.31453683972358704,
      "learning_rate": 4.938429964084146e-06,
      "loss": 0.0104,
      "step": 15400
    },
    {
      "epoch": 0.09883495332437554,
      "grad_norm": 0.833005428314209,
      "learning_rate": 4.941636736788097e-06,
      "loss": 0.0108,
      "step": 15410
    },
    {
      "epoch": 0.09889909021816165,
      "grad_norm": 0.22341343760490417,
      "learning_rate": 4.944843509492048e-06,
      "loss": 0.0119,
      "step": 15420
    },
    {
      "epoch": 0.09896322711194774,
      "grad_norm": 0.23929423093795776,
      "learning_rate": 4.948050282195998e-06,
      "loss": 0.0095,
      "step": 15430
    },
    {
      "epoch": 0.09902736400573384,
      "grad_norm": 0.6590454578399658,
      "learning_rate": 4.9512570548999486e-06,
      "loss": 0.0076,
      "step": 15440
    },
    {
      "epoch": 0.09909150089951993,
      "grad_norm": 0.7365688681602478,
      "learning_rate": 4.9544638276039e-06,
      "loss": 0.0091,
      "step": 15450
    },
    {
      "epoch": 0.09915563779330604,
      "grad_norm": 0.48849421739578247,
      "learning_rate": 4.957670600307851e-06,
      "loss": 0.0084,
      "step": 15460
    },
    {
      "epoch": 0.09921977468709213,
      "grad_norm": 1.1581774950027466,
      "learning_rate": 4.960877373011802e-06,
      "loss": 0.0113,
      "step": 15470
    },
    {
      "epoch": 0.09928391158087822,
      "grad_norm": 0.6785595417022705,
      "learning_rate": 4.964084145715752e-06,
      "loss": 0.0077,
      "step": 15480
    },
    {
      "epoch": 0.09934804847466433,
      "grad_norm": 0.6807507276535034,
      "learning_rate": 4.9672909184197024e-06,
      "loss": 0.0095,
      "step": 15490
    },
    {
      "epoch": 0.09941218536845042,
      "grad_norm": 1.0348047018051147,
      "learning_rate": 4.9704976911236536e-06,
      "loss": 0.0127,
      "step": 15500
    },
    {
      "epoch": 0.09947632226223652,
      "grad_norm": 0.5778086185455322,
      "learning_rate": 4.973704463827605e-06,
      "loss": 0.0101,
      "step": 15510
    },
    {
      "epoch": 0.09954045915602261,
      "grad_norm": 0.6600560545921326,
      "learning_rate": 4.976911236531555e-06,
      "loss": 0.0091,
      "step": 15520
    },
    {
      "epoch": 0.09960459604980872,
      "grad_norm": 0.6662725210189819,
      "learning_rate": 4.980118009235505e-06,
      "loss": 0.0128,
      "step": 15530
    },
    {
      "epoch": 0.09966873294359481,
      "grad_norm": 0.3879687786102295,
      "learning_rate": 4.983324781939456e-06,
      "loss": 0.0094,
      "step": 15540
    },
    {
      "epoch": 0.0997328698373809,
      "grad_norm": 0.3806883990764618,
      "learning_rate": 4.9865315546434074e-06,
      "loss": 0.0082,
      "step": 15550
    },
    {
      "epoch": 0.09979700673116701,
      "grad_norm": 0.4219231903553009,
      "learning_rate": 4.9897383273473585e-06,
      "loss": 0.0081,
      "step": 15560
    },
    {
      "epoch": 0.0998611436249531,
      "grad_norm": 0.428092896938324,
      "learning_rate": 4.992945100051309e-06,
      "loss": 0.0097,
      "step": 15570
    },
    {
      "epoch": 0.0999252805187392,
      "grad_norm": 0.8046025633811951,
      "learning_rate": 4.996151872755259e-06,
      "loss": 0.0121,
      "step": 15580
    },
    {
      "epoch": 0.09998941741252529,
      "grad_norm": 0.5787171125411987,
      "learning_rate": 4.99935864545921e-06,
      "loss": 0.0104,
      "step": 15590
    },
    {
      "epoch": 0.1000535543063114,
      "grad_norm": 0.3348299264907837,
      "learning_rate": 5.002565418163161e-06,
      "loss": 0.0102,
      "step": 15600
    },
    {
      "epoch": 0.10011769120009749,
      "grad_norm": 0.3324357569217682,
      "learning_rate": 5.0057721908671116e-06,
      "loss": 0.0061,
      "step": 15610
    },
    {
      "epoch": 0.10018182809388358,
      "grad_norm": 0.9565490484237671,
      "learning_rate": 5.008978963571063e-06,
      "loss": 0.0176,
      "step": 15620
    },
    {
      "epoch": 0.10024596498766969,
      "grad_norm": 0.6187155842781067,
      "learning_rate": 5.012185736275014e-06,
      "loss": 0.0103,
      "step": 15630
    },
    {
      "epoch": 0.10031010188145578,
      "grad_norm": 0.6529510617256165,
      "learning_rate": 5.015392508978963e-06,
      "loss": 0.0067,
      "step": 15640
    },
    {
      "epoch": 0.10037423877524188,
      "grad_norm": 0.6332399845123291,
      "learning_rate": 5.018599281682914e-06,
      "loss": 0.0108,
      "step": 15650
    },
    {
      "epoch": 0.10043837566902797,
      "grad_norm": 0.5133268237113953,
      "learning_rate": 5.0218060543868654e-06,
      "loss": 0.0089,
      "step": 15660
    },
    {
      "epoch": 0.10050251256281408,
      "grad_norm": 0.34205323457717896,
      "learning_rate": 5.025012827090816e-06,
      "loss": 0.0112,
      "step": 15670
    },
    {
      "epoch": 0.10056664945660017,
      "grad_norm": 0.6201695203781128,
      "learning_rate": 5.028219599794767e-06,
      "loss": 0.0108,
      "step": 15680
    },
    {
      "epoch": 0.10063078635038626,
      "grad_norm": 0.30734115839004517,
      "learning_rate": 5.031426372498718e-06,
      "loss": 0.0176,
      "step": 15690
    },
    {
      "epoch": 0.10069492324417235,
      "grad_norm": 0.5728802680969238,
      "learning_rate": 5.034633145202669e-06,
      "loss": 0.0128,
      "step": 15700
    },
    {
      "epoch": 0.10075906013795846,
      "grad_norm": 0.5966327786445618,
      "learning_rate": 5.037839917906619e-06,
      "loss": 0.0142,
      "step": 15710
    },
    {
      "epoch": 0.10082319703174455,
      "grad_norm": 0.3857520818710327,
      "learning_rate": 5.04104669061057e-06,
      "loss": 0.007,
      "step": 15720
    },
    {
      "epoch": 0.10088733392553065,
      "grad_norm": 0.8242173790931702,
      "learning_rate": 5.04425346331452e-06,
      "loss": 0.0073,
      "step": 15730
    },
    {
      "epoch": 0.10095147081931675,
      "grad_norm": 0.3519569933414459,
      "learning_rate": 5.047460236018471e-06,
      "loss": 0.0085,
      "step": 15740
    },
    {
      "epoch": 0.10101560771310285,
      "grad_norm": 0.5322737097740173,
      "learning_rate": 5.050667008722422e-06,
      "loss": 0.0097,
      "step": 15750
    },
    {
      "epoch": 0.10107974460688894,
      "grad_norm": 0.6807852387428284,
      "learning_rate": 5.053873781426372e-06,
      "loss": 0.0127,
      "step": 15760
    },
    {
      "epoch": 0.10114388150067503,
      "grad_norm": 0.7069212198257446,
      "learning_rate": 5.0570805541303234e-06,
      "loss": 0.0091,
      "step": 15770
    },
    {
      "epoch": 0.10120801839446114,
      "grad_norm": 0.6726041436195374,
      "learning_rate": 5.0602873268342746e-06,
      "loss": 0.0104,
      "step": 15780
    },
    {
      "epoch": 0.10127215528824723,
      "grad_norm": 0.5097774267196655,
      "learning_rate": 5.063494099538226e-06,
      "loss": 0.0128,
      "step": 15790
    },
    {
      "epoch": 0.10133629218203333,
      "grad_norm": 0.5260890126228333,
      "learning_rate": 5.066700872242176e-06,
      "loss": 0.0062,
      "step": 15800
    },
    {
      "epoch": 0.10140042907581943,
      "grad_norm": 0.41690969467163086,
      "learning_rate": 5.069907644946127e-06,
      "loss": 0.0102,
      "step": 15810
    },
    {
      "epoch": 0.10146456596960553,
      "grad_norm": 0.41804683208465576,
      "learning_rate": 5.0731144176500765e-06,
      "loss": 0.0134,
      "step": 15820
    },
    {
      "epoch": 0.10152870286339162,
      "grad_norm": 0.3769904375076294,
      "learning_rate": 5.076321190354028e-06,
      "loss": 0.0101,
      "step": 15830
    },
    {
      "epoch": 0.10159283975717771,
      "grad_norm": 1.0027804374694824,
      "learning_rate": 5.079527963057979e-06,
      "loss": 0.0114,
      "step": 15840
    },
    {
      "epoch": 0.10165697665096382,
      "grad_norm": 0.556800365447998,
      "learning_rate": 5.08273473576193e-06,
      "loss": 0.0086,
      "step": 15850
    },
    {
      "epoch": 0.10172111354474991,
      "grad_norm": 0.4666617512702942,
      "learning_rate": 5.08594150846588e-06,
      "loss": 0.0112,
      "step": 15860
    },
    {
      "epoch": 0.101785250438536,
      "grad_norm": 0.542410135269165,
      "learning_rate": 5.089148281169831e-06,
      "loss": 0.007,
      "step": 15870
    },
    {
      "epoch": 0.10184938733232211,
      "grad_norm": 0.6153366565704346,
      "learning_rate": 5.092355053873782e-06,
      "loss": 0.0073,
      "step": 15880
    },
    {
      "epoch": 0.1019135242261082,
      "grad_norm": 0.44405901432037354,
      "learning_rate": 5.0955618265777326e-06,
      "loss": 0.007,
      "step": 15890
    },
    {
      "epoch": 0.1019776611198943,
      "grad_norm": 0.3870573937892914,
      "learning_rate": 5.098768599281684e-06,
      "loss": 0.0132,
      "step": 15900
    },
    {
      "epoch": 0.1020417980136804,
      "grad_norm": 0.5652061104774475,
      "learning_rate": 5.101975371985634e-06,
      "loss": 0.0105,
      "step": 15910
    },
    {
      "epoch": 0.1021059349074665,
      "grad_norm": 0.8508217930793762,
      "learning_rate": 5.105182144689584e-06,
      "loss": 0.0135,
      "step": 15920
    },
    {
      "epoch": 0.1021700718012526,
      "grad_norm": 0.3459599018096924,
      "learning_rate": 5.108388917393535e-06,
      "loss": 0.0086,
      "step": 15930
    },
    {
      "epoch": 0.10223420869503869,
      "grad_norm": 0.5796306133270264,
      "learning_rate": 5.1115956900974864e-06,
      "loss": 0.0129,
      "step": 15940
    },
    {
      "epoch": 0.1022983455888248,
      "grad_norm": 0.5738620162010193,
      "learning_rate": 5.114802462801437e-06,
      "loss": 0.0147,
      "step": 15950
    },
    {
      "epoch": 0.10236248248261089,
      "grad_norm": 0.3462698459625244,
      "learning_rate": 5.118009235505388e-06,
      "loss": 0.0112,
      "step": 15960
    },
    {
      "epoch": 0.10242661937639698,
      "grad_norm": 0.519221305847168,
      "learning_rate": 5.121216008209339e-06,
      "loss": 0.008,
      "step": 15970
    },
    {
      "epoch": 0.10249075627018307,
      "grad_norm": 0.42029598355293274,
      "learning_rate": 5.124422780913289e-06,
      "loss": 0.0092,
      "step": 15980
    },
    {
      "epoch": 0.10255489316396918,
      "grad_norm": 0.4115634262561798,
      "learning_rate": 5.12762955361724e-06,
      "loss": 0.0066,
      "step": 15990
    },
    {
      "epoch": 0.10261903005775527,
      "grad_norm": 0.9606252908706665,
      "learning_rate": 5.130836326321191e-06,
      "loss": 0.0098,
      "step": 16000
    },
    {
      "epoch": 0.10268316695154137,
      "grad_norm": 0.40013089776039124,
      "learning_rate": 5.134043099025141e-06,
      "loss": 0.0131,
      "step": 16010
    },
    {
      "epoch": 0.10274730384532747,
      "grad_norm": 0.724273681640625,
      "learning_rate": 5.137249871729092e-06,
      "loss": 0.0101,
      "step": 16020
    },
    {
      "epoch": 0.10281144073911357,
      "grad_norm": 0.4710618853569031,
      "learning_rate": 5.140456644433043e-06,
      "loss": 0.0125,
      "step": 16030
    },
    {
      "epoch": 0.10287557763289966,
      "grad_norm": 0.40870100259780884,
      "learning_rate": 5.143663417136993e-06,
      "loss": 0.0092,
      "step": 16040
    },
    {
      "epoch": 0.10293971452668575,
      "grad_norm": 0.640579342842102,
      "learning_rate": 5.1468701898409444e-06,
      "loss": 0.0112,
      "step": 16050
    },
    {
      "epoch": 0.10300385142047186,
      "grad_norm": 0.5337666273117065,
      "learning_rate": 5.1500769625448956e-06,
      "loss": 0.0107,
      "step": 16060
    },
    {
      "epoch": 0.10306798831425795,
      "grad_norm": 0.42939725518226624,
      "learning_rate": 5.153283735248847e-06,
      "loss": 0.011,
      "step": 16070
    },
    {
      "epoch": 0.10313212520804405,
      "grad_norm": 0.532259464263916,
      "learning_rate": 5.156490507952797e-06,
      "loss": 0.0129,
      "step": 16080
    },
    {
      "epoch": 0.10319626210183015,
      "grad_norm": 0.5188320279121399,
      "learning_rate": 5.159697280656748e-06,
      "loss": 0.0092,
      "step": 16090
    },
    {
      "epoch": 0.10326039899561625,
      "grad_norm": 0.5226402282714844,
      "learning_rate": 5.1629040533606975e-06,
      "loss": 0.0085,
      "step": 16100
    },
    {
      "epoch": 0.10332453588940234,
      "grad_norm": 0.45598104596138,
      "learning_rate": 5.166110826064649e-06,
      "loss": 0.0106,
      "step": 16110
    },
    {
      "epoch": 0.10338867278318843,
      "grad_norm": 0.3412771224975586,
      "learning_rate": 5.1693175987686e-06,
      "loss": 0.0109,
      "step": 16120
    },
    {
      "epoch": 0.10345280967697454,
      "grad_norm": 0.22648046910762787,
      "learning_rate": 5.17252437147255e-06,
      "loss": 0.0098,
      "step": 16130
    },
    {
      "epoch": 0.10351694657076063,
      "grad_norm": 0.5391767024993896,
      "learning_rate": 5.175731144176501e-06,
      "loss": 0.0084,
      "step": 16140
    },
    {
      "epoch": 0.10358108346454672,
      "grad_norm": 0.7075507640838623,
      "learning_rate": 5.178937916880452e-06,
      "loss": 0.0132,
      "step": 16150
    },
    {
      "epoch": 0.10364522035833283,
      "grad_norm": 0.6682233214378357,
      "learning_rate": 5.182144689584403e-06,
      "loss": 0.0115,
      "step": 16160
    },
    {
      "epoch": 0.10370935725211892,
      "grad_norm": 0.4709302484989166,
      "learning_rate": 5.1853514622883536e-06,
      "loss": 0.0093,
      "step": 16170
    },
    {
      "epoch": 0.10377349414590502,
      "grad_norm": 0.4433741867542267,
      "learning_rate": 5.188558234992305e-06,
      "loss": 0.0088,
      "step": 16180
    },
    {
      "epoch": 0.10383763103969111,
      "grad_norm": 0.6409521102905273,
      "learning_rate": 5.191765007696254e-06,
      "loss": 0.0117,
      "step": 16190
    },
    {
      "epoch": 0.10390176793347722,
      "grad_norm": 0.639723539352417,
      "learning_rate": 5.194971780400205e-06,
      "loss": 0.0115,
      "step": 16200
    },
    {
      "epoch": 0.10396590482726331,
      "grad_norm": 0.46562427282333374,
      "learning_rate": 5.198178553104156e-06,
      "loss": 0.0077,
      "step": 16210
    },
    {
      "epoch": 0.1040300417210494,
      "grad_norm": 0.44589903950691223,
      "learning_rate": 5.2013853258081074e-06,
      "loss": 0.0073,
      "step": 16220
    },
    {
      "epoch": 0.10409417861483551,
      "grad_norm": 0.5284266471862793,
      "learning_rate": 5.204592098512058e-06,
      "loss": 0.0109,
      "step": 16230
    },
    {
      "epoch": 0.1041583155086216,
      "grad_norm": 0.4344564974308014,
      "learning_rate": 5.207798871216009e-06,
      "loss": 0.0144,
      "step": 16240
    },
    {
      "epoch": 0.1042224524024077,
      "grad_norm": 0.5707141757011414,
      "learning_rate": 5.21100564391996e-06,
      "loss": 0.0086,
      "step": 16250
    },
    {
      "epoch": 0.10428658929619379,
      "grad_norm": 0.43890857696533203,
      "learning_rate": 5.21421241662391e-06,
      "loss": 0.0098,
      "step": 16260
    },
    {
      "epoch": 0.1043507261899799,
      "grad_norm": 0.2899274528026581,
      "learning_rate": 5.217419189327861e-06,
      "loss": 0.0078,
      "step": 16270
    },
    {
      "epoch": 0.10441486308376599,
      "grad_norm": 0.5155450701713562,
      "learning_rate": 5.220625962031812e-06,
      "loss": 0.0104,
      "step": 16280
    },
    {
      "epoch": 0.10447899997755208,
      "grad_norm": 0.3467402756214142,
      "learning_rate": 5.223832734735762e-06,
      "loss": 0.0078,
      "step": 16290
    },
    {
      "epoch": 0.10454313687133819,
      "grad_norm": 0.389515221118927,
      "learning_rate": 5.227039507439713e-06,
      "loss": 0.0126,
      "step": 16300
    },
    {
      "epoch": 0.10460727376512428,
      "grad_norm": 0.8770573735237122,
      "learning_rate": 5.230246280143664e-06,
      "loss": 0.0102,
      "step": 16310
    },
    {
      "epoch": 0.10467141065891038,
      "grad_norm": 0.3336693048477173,
      "learning_rate": 5.233453052847614e-06,
      "loss": 0.0102,
      "step": 16320
    },
    {
      "epoch": 0.10473554755269647,
      "grad_norm": 0.46077433228492737,
      "learning_rate": 5.2366598255515655e-06,
      "loss": 0.0101,
      "step": 16330
    },
    {
      "epoch": 0.10479968444648258,
      "grad_norm": 0.29403117299079895,
      "learning_rate": 5.2398665982555166e-06,
      "loss": 0.0131,
      "step": 16340
    },
    {
      "epoch": 0.10486382134026867,
      "grad_norm": 0.600862443447113,
      "learning_rate": 5.243073370959467e-06,
      "loss": 0.0126,
      "step": 16350
    },
    {
      "epoch": 0.10492795823405476,
      "grad_norm": 0.3670068383216858,
      "learning_rate": 5.246280143663418e-06,
      "loss": 0.0077,
      "step": 16360
    },
    {
      "epoch": 0.10499209512784087,
      "grad_norm": 0.5938912034034729,
      "learning_rate": 5.249486916367368e-06,
      "loss": 0.0082,
      "step": 16370
    },
    {
      "epoch": 0.10505623202162696,
      "grad_norm": 0.5170342326164246,
      "learning_rate": 5.2526936890713185e-06,
      "loss": 0.0091,
      "step": 16380
    },
    {
      "epoch": 0.10512036891541306,
      "grad_norm": 0.534372866153717,
      "learning_rate": 5.25590046177527e-06,
      "loss": 0.0106,
      "step": 16390
    },
    {
      "epoch": 0.10518450580919915,
      "grad_norm": 0.49109458923339844,
      "learning_rate": 5.259107234479221e-06,
      "loss": 0.0087,
      "step": 16400
    },
    {
      "epoch": 0.10524864270298526,
      "grad_norm": 0.6332600712776184,
      "learning_rate": 5.262314007183171e-06,
      "loss": 0.0104,
      "step": 16410
    },
    {
      "epoch": 0.10531277959677135,
      "grad_norm": 0.2460511326789856,
      "learning_rate": 5.265520779887122e-06,
      "loss": 0.0065,
      "step": 16420
    },
    {
      "epoch": 0.10537691649055744,
      "grad_norm": 0.3532591164112091,
      "learning_rate": 5.268727552591073e-06,
      "loss": 0.0121,
      "step": 16430
    },
    {
      "epoch": 0.10544105338434354,
      "grad_norm": 0.3050209879875183,
      "learning_rate": 5.271934325295024e-06,
      "loss": 0.0341,
      "step": 16440
    },
    {
      "epoch": 0.10550519027812964,
      "grad_norm": 0.44340020418167114,
      "learning_rate": 5.2751410979989746e-06,
      "loss": 0.0117,
      "step": 16450
    },
    {
      "epoch": 0.10556932717191574,
      "grad_norm": 0.5648112297058105,
      "learning_rate": 5.278347870702925e-06,
      "loss": 0.0082,
      "step": 16460
    },
    {
      "epoch": 0.10563346406570183,
      "grad_norm": 1.0201709270477295,
      "learning_rate": 5.281554643406875e-06,
      "loss": 0.0091,
      "step": 16470
    },
    {
      "epoch": 0.10569760095948794,
      "grad_norm": 0.5138731002807617,
      "learning_rate": 5.284761416110826e-06,
      "loss": 0.0093,
      "step": 16480
    },
    {
      "epoch": 0.10576173785327403,
      "grad_norm": 0.3440841734409332,
      "learning_rate": 5.287968188814777e-06,
      "loss": 0.0114,
      "step": 16490
    },
    {
      "epoch": 0.10582587474706012,
      "grad_norm": 0.37862738966941833,
      "learning_rate": 5.291174961518728e-06,
      "loss": 0.0092,
      "step": 16500
    },
    {
      "epoch": 0.10589001164084622,
      "grad_norm": 0.7785543203353882,
      "learning_rate": 5.294381734222679e-06,
      "loss": 0.0078,
      "step": 16510
    },
    {
      "epoch": 0.10595414853463232,
      "grad_norm": 0.5919600129127502,
      "learning_rate": 5.29758850692663e-06,
      "loss": 0.0093,
      "step": 16520
    },
    {
      "epoch": 0.10601828542841842,
      "grad_norm": 0.4332359731197357,
      "learning_rate": 5.300795279630581e-06,
      "loss": 0.0105,
      "step": 16530
    },
    {
      "epoch": 0.10608242232220451,
      "grad_norm": 0.5844279527664185,
      "learning_rate": 5.304002052334531e-06,
      "loss": 0.011,
      "step": 16540
    },
    {
      "epoch": 0.10614655921599062,
      "grad_norm": 0.7590405941009521,
      "learning_rate": 5.3072088250384815e-06,
      "loss": 0.0064,
      "step": 16550
    },
    {
      "epoch": 0.10621069610977671,
      "grad_norm": 0.4074823260307312,
      "learning_rate": 5.310415597742432e-06,
      "loss": 0.0125,
      "step": 16560
    },
    {
      "epoch": 0.1062748330035628,
      "grad_norm": 0.6540939807891846,
      "learning_rate": 5.313622370446383e-06,
      "loss": 0.0121,
      "step": 16570
    },
    {
      "epoch": 0.1063389698973489,
      "grad_norm": 0.8707407116889954,
      "learning_rate": 5.316829143150334e-06,
      "loss": 0.0082,
      "step": 16580
    },
    {
      "epoch": 0.106403106791135,
      "grad_norm": 0.6418952345848083,
      "learning_rate": 5.320035915854285e-06,
      "loss": 0.0131,
      "step": 16590
    },
    {
      "epoch": 0.1064672436849211,
      "grad_norm": 0.5803223848342896,
      "learning_rate": 5.323242688558235e-06,
      "loss": 0.0094,
      "step": 16600
    },
    {
      "epoch": 0.10653138057870719,
      "grad_norm": 0.6605579257011414,
      "learning_rate": 5.3264494612621865e-06,
      "loss": 0.0086,
      "step": 16610
    },
    {
      "epoch": 0.1065955174724933,
      "grad_norm": 0.6360081434249878,
      "learning_rate": 5.3296562339661376e-06,
      "loss": 0.012,
      "step": 16620
    },
    {
      "epoch": 0.10665965436627939,
      "grad_norm": 0.5654777884483337,
      "learning_rate": 5.332863006670088e-06,
      "loss": 0.0117,
      "step": 16630
    },
    {
      "epoch": 0.10672379126006548,
      "grad_norm": 0.6596087217330933,
      "learning_rate": 5.336069779374038e-06,
      "loss": 0.012,
      "step": 16640
    },
    {
      "epoch": 0.10678792815385157,
      "grad_norm": 0.7838721871376038,
      "learning_rate": 5.339276552077988e-06,
      "loss": 0.0157,
      "step": 16650
    },
    {
      "epoch": 0.10685206504763768,
      "grad_norm": 0.420419305562973,
      "learning_rate": 5.3424833247819395e-06,
      "loss": 0.009,
      "step": 16660
    },
    {
      "epoch": 0.10691620194142377,
      "grad_norm": 0.40264374017715454,
      "learning_rate": 5.345690097485891e-06,
      "loss": 0.0111,
      "step": 16670
    },
    {
      "epoch": 0.10698033883520987,
      "grad_norm": 0.32008737325668335,
      "learning_rate": 5.348896870189842e-06,
      "loss": 0.0062,
      "step": 16680
    },
    {
      "epoch": 0.10704447572899597,
      "grad_norm": 0.6018548011779785,
      "learning_rate": 5.352103642893792e-06,
      "loss": 0.0091,
      "step": 16690
    },
    {
      "epoch": 0.10710861262278207,
      "grad_norm": 0.3959011733531952,
      "learning_rate": 5.355310415597743e-06,
      "loss": 0.0077,
      "step": 16700
    },
    {
      "epoch": 0.10717274951656816,
      "grad_norm": 0.6982358694076538,
      "learning_rate": 5.358517188301694e-06,
      "loss": 0.0115,
      "step": 16710
    },
    {
      "epoch": 0.10723688641035425,
      "grad_norm": 0.34304675459861755,
      "learning_rate": 5.3617239610056445e-06,
      "loss": 0.0108,
      "step": 16720
    },
    {
      "epoch": 0.10730102330414036,
      "grad_norm": 0.67317795753479,
      "learning_rate": 5.364930733709595e-06,
      "loss": 0.0123,
      "step": 16730
    },
    {
      "epoch": 0.10736516019792645,
      "grad_norm": 0.4527972340583801,
      "learning_rate": 5.368137506413546e-06,
      "loss": 0.0096,
      "step": 16740
    },
    {
      "epoch": 0.10742929709171255,
      "grad_norm": 0.8204785585403442,
      "learning_rate": 5.371344279117496e-06,
      "loss": 0.01,
      "step": 16750
    },
    {
      "epoch": 0.10749343398549865,
      "grad_norm": 0.1776057481765747,
      "learning_rate": 5.374551051821447e-06,
      "loss": 0.0073,
      "step": 16760
    },
    {
      "epoch": 0.10755757087928475,
      "grad_norm": 0.36770233511924744,
      "learning_rate": 5.377757824525398e-06,
      "loss": 0.0095,
      "step": 16770
    },
    {
      "epoch": 0.10762170777307084,
      "grad_norm": 0.4649524688720703,
      "learning_rate": 5.380964597229349e-06,
      "loss": 0.0077,
      "step": 16780
    },
    {
      "epoch": 0.10768584466685693,
      "grad_norm": 0.4981425702571869,
      "learning_rate": 5.3841713699333e-06,
      "loss": 0.0089,
      "step": 16790
    },
    {
      "epoch": 0.10774998156064304,
      "grad_norm": 0.5665914416313171,
      "learning_rate": 5.387378142637251e-06,
      "loss": 0.011,
      "step": 16800
    },
    {
      "epoch": 0.10781411845442913,
      "grad_norm": 0.6016936302185059,
      "learning_rate": 5.390584915341202e-06,
      "loss": 0.0091,
      "step": 16810
    },
    {
      "epoch": 0.10787825534821523,
      "grad_norm": 0.3479364812374115,
      "learning_rate": 5.393791688045151e-06,
      "loss": 0.009,
      "step": 16820
    },
    {
      "epoch": 0.10794239224200133,
      "grad_norm": 0.40164706110954285,
      "learning_rate": 5.3969984607491025e-06,
      "loss": 0.0101,
      "step": 16830
    },
    {
      "epoch": 0.10800652913578743,
      "grad_norm": 0.24736908078193665,
      "learning_rate": 5.400205233453053e-06,
      "loss": 0.0087,
      "step": 16840
    },
    {
      "epoch": 0.10807066602957352,
      "grad_norm": 0.5277517437934875,
      "learning_rate": 5.403412006157004e-06,
      "loss": 0.0099,
      "step": 16850
    },
    {
      "epoch": 0.10813480292335961,
      "grad_norm": 0.5787237882614136,
      "learning_rate": 5.406618778860955e-06,
      "loss": 0.0118,
      "step": 16860
    },
    {
      "epoch": 0.10819893981714572,
      "grad_norm": 0.6116265654563904,
      "learning_rate": 5.409825551564905e-06,
      "loss": 0.0116,
      "step": 16870
    },
    {
      "epoch": 0.10826307671093181,
      "grad_norm": 0.5526065826416016,
      "learning_rate": 5.413032324268856e-06,
      "loss": 0.0083,
      "step": 16880
    },
    {
      "epoch": 0.1083272136047179,
      "grad_norm": 0.5182403922080994,
      "learning_rate": 5.4162390969728075e-06,
      "loss": 0.012,
      "step": 16890
    },
    {
      "epoch": 0.10839135049850401,
      "grad_norm": 0.5181537866592407,
      "learning_rate": 5.4194458696767586e-06,
      "loss": 0.0089,
      "step": 16900
    },
    {
      "epoch": 0.1084554873922901,
      "grad_norm": 0.6523613333702087,
      "learning_rate": 5.422652642380708e-06,
      "loss": 0.013,
      "step": 16910
    },
    {
      "epoch": 0.1085196242860762,
      "grad_norm": 0.12294157594442368,
      "learning_rate": 5.425859415084659e-06,
      "loss": 0.0081,
      "step": 16920
    },
    {
      "epoch": 0.10858376117986229,
      "grad_norm": 0.5861759781837463,
      "learning_rate": 5.429066187788609e-06,
      "loss": 0.0073,
      "step": 16930
    },
    {
      "epoch": 0.1086478980736484,
      "grad_norm": 0.5482618808746338,
      "learning_rate": 5.4322729604925605e-06,
      "loss": 0.0097,
      "step": 16940
    },
    {
      "epoch": 0.10871203496743449,
      "grad_norm": 0.5014786124229431,
      "learning_rate": 5.435479733196512e-06,
      "loss": 0.0126,
      "step": 16950
    },
    {
      "epoch": 0.10877617186122059,
      "grad_norm": 0.31347060203552246,
      "learning_rate": 5.438686505900463e-06,
      "loss": 0.0136,
      "step": 16960
    },
    {
      "epoch": 0.10884030875500669,
      "grad_norm": 0.5638201832771301,
      "learning_rate": 5.441893278604413e-06,
      "loss": 0.0088,
      "step": 16970
    },
    {
      "epoch": 0.10890444564879279,
      "grad_norm": 1.0175175666809082,
      "learning_rate": 5.445100051308364e-06,
      "loss": 0.0106,
      "step": 16980
    },
    {
      "epoch": 0.10896858254257888,
      "grad_norm": 0.6966453194618225,
      "learning_rate": 5.448306824012315e-06,
      "loss": 0.0131,
      "step": 16990
    },
    {
      "epoch": 0.10903271943636497,
      "grad_norm": 0.4546926021575928,
      "learning_rate": 5.451513596716265e-06,
      "loss": 0.0073,
      "step": 17000
    },
    {
      "epoch": 0.10909685633015108,
      "grad_norm": 0.4377993941307068,
      "learning_rate": 5.454720369420216e-06,
      "loss": 0.0167,
      "step": 17010
    },
    {
      "epoch": 0.10916099322393717,
      "grad_norm": 1.2318954467773438,
      "learning_rate": 5.457927142124166e-06,
      "loss": 0.0159,
      "step": 17020
    },
    {
      "epoch": 0.10922513011772327,
      "grad_norm": 0.6596897840499878,
      "learning_rate": 5.461133914828117e-06,
      "loss": 0.0073,
      "step": 17030
    },
    {
      "epoch": 0.10928926701150937,
      "grad_norm": 0.694681704044342,
      "learning_rate": 5.464340687532068e-06,
      "loss": 0.011,
      "step": 17040
    },
    {
      "epoch": 0.10935340390529547,
      "grad_norm": 0.24846361577510834,
      "learning_rate": 5.467547460236019e-06,
      "loss": 0.0084,
      "step": 17050
    },
    {
      "epoch": 0.10941754079908156,
      "grad_norm": 0.630030632019043,
      "learning_rate": 5.47075423293997e-06,
      "loss": 0.0076,
      "step": 17060
    },
    {
      "epoch": 0.10948167769286765,
      "grad_norm": 0.8209272027015686,
      "learning_rate": 5.473961005643921e-06,
      "loss": 0.0131,
      "step": 17070
    },
    {
      "epoch": 0.10954581458665376,
      "grad_norm": 0.5075275897979736,
      "learning_rate": 5.477167778347872e-06,
      "loss": 0.0107,
      "step": 17080
    },
    {
      "epoch": 0.10960995148043985,
      "grad_norm": 0.5664962530136108,
      "learning_rate": 5.480374551051821e-06,
      "loss": 0.0128,
      "step": 17090
    },
    {
      "epoch": 0.10967408837422594,
      "grad_norm": 0.5662223100662231,
      "learning_rate": 5.483581323755772e-06,
      "loss": 0.0081,
      "step": 17100
    },
    {
      "epoch": 0.10973822526801205,
      "grad_norm": 0.5928486585617065,
      "learning_rate": 5.4867880964597235e-06,
      "loss": 0.0114,
      "step": 17110
    },
    {
      "epoch": 0.10980236216179814,
      "grad_norm": 0.4811129868030548,
      "learning_rate": 5.489994869163674e-06,
      "loss": 0.0074,
      "step": 17120
    },
    {
      "epoch": 0.10986649905558424,
      "grad_norm": 0.3544856607913971,
      "learning_rate": 5.493201641867625e-06,
      "loss": 0.0088,
      "step": 17130
    },
    {
      "epoch": 0.10993063594937033,
      "grad_norm": 0.45926907658576965,
      "learning_rate": 5.496408414571576e-06,
      "loss": 0.0093,
      "step": 17140
    },
    {
      "epoch": 0.10999477284315644,
      "grad_norm": 0.6390130519866943,
      "learning_rate": 5.499615187275526e-06,
      "loss": 0.0061,
      "step": 17150
    },
    {
      "epoch": 0.11005890973694253,
      "grad_norm": 0.7102174758911133,
      "learning_rate": 5.502821959979477e-06,
      "loss": 0.0121,
      "step": 17160
    },
    {
      "epoch": 0.11012304663072862,
      "grad_norm": 0.5696839094161987,
      "learning_rate": 5.5060287326834285e-06,
      "loss": 0.0081,
      "step": 17170
    },
    {
      "epoch": 0.11018718352451472,
      "grad_norm": 0.4979040026664734,
      "learning_rate": 5.509235505387378e-06,
      "loss": 0.0139,
      "step": 17180
    },
    {
      "epoch": 0.11025132041830082,
      "grad_norm": 0.4420786201953888,
      "learning_rate": 5.512442278091329e-06,
      "loss": 0.0124,
      "step": 17190
    },
    {
      "epoch": 0.11031545731208692,
      "grad_norm": 0.3444763422012329,
      "learning_rate": 5.51564905079528e-06,
      "loss": 0.0082,
      "step": 17200
    },
    {
      "epoch": 0.11037959420587301,
      "grad_norm": 0.638555645942688,
      "learning_rate": 5.51885582349923e-06,
      "loss": 0.0126,
      "step": 17210
    },
    {
      "epoch": 0.11044373109965912,
      "grad_norm": 0.5583418607711792,
      "learning_rate": 5.5220625962031815e-06,
      "loss": 0.0094,
      "step": 17220
    },
    {
      "epoch": 0.11050786799344521,
      "grad_norm": 0.8487377166748047,
      "learning_rate": 5.525269368907133e-06,
      "loss": 0.0128,
      "step": 17230
    },
    {
      "epoch": 0.1105720048872313,
      "grad_norm": 0.3511020541191101,
      "learning_rate": 5.528476141611083e-06,
      "loss": 0.0081,
      "step": 17240
    },
    {
      "epoch": 0.1106361417810174,
      "grad_norm": 0.4752103388309479,
      "learning_rate": 5.531682914315034e-06,
      "loss": 0.0121,
      "step": 17250
    },
    {
      "epoch": 0.1107002786748035,
      "grad_norm": 0.2458493560552597,
      "learning_rate": 5.534889687018985e-06,
      "loss": 0.0089,
      "step": 17260
    },
    {
      "epoch": 0.1107644155685896,
      "grad_norm": 0.2905209958553314,
      "learning_rate": 5.5380964597229345e-06,
      "loss": 0.0097,
      "step": 17270
    },
    {
      "epoch": 0.11082855246237569,
      "grad_norm": 0.46376416087150574,
      "learning_rate": 5.541303232426886e-06,
      "loss": 0.0097,
      "step": 17280
    },
    {
      "epoch": 0.1108926893561618,
      "grad_norm": 0.3083602786064148,
      "learning_rate": 5.544510005130837e-06,
      "loss": 0.0076,
      "step": 17290
    },
    {
      "epoch": 0.11095682624994789,
      "grad_norm": 0.7128134965896606,
      "learning_rate": 5.547716777834787e-06,
      "loss": 0.0126,
      "step": 17300
    },
    {
      "epoch": 0.11102096314373398,
      "grad_norm": 0.772965133190155,
      "learning_rate": 5.550923550538738e-06,
      "loss": 0.0113,
      "step": 17310
    },
    {
      "epoch": 0.11108510003752008,
      "grad_norm": 0.5096204280853271,
      "learning_rate": 5.554130323242689e-06,
      "loss": 0.0109,
      "step": 17320
    },
    {
      "epoch": 0.11114923693130618,
      "grad_norm": 0.5388255715370178,
      "learning_rate": 5.55733709594664e-06,
      "loss": 0.0085,
      "step": 17330
    },
    {
      "epoch": 0.11121337382509228,
      "grad_norm": 0.7835232019424438,
      "learning_rate": 5.560543868650591e-06,
      "loss": 0.009,
      "step": 17340
    },
    {
      "epoch": 0.11127751071887837,
      "grad_norm": 0.38351520895957947,
      "learning_rate": 5.563750641354542e-06,
      "loss": 0.0086,
      "step": 17350
    },
    {
      "epoch": 0.11134164761266448,
      "grad_norm": 0.4709744155406952,
      "learning_rate": 5.566957414058491e-06,
      "loss": 0.009,
      "step": 17360
    },
    {
      "epoch": 0.11140578450645057,
      "grad_norm": 0.46643105149269104,
      "learning_rate": 5.570164186762442e-06,
      "loss": 0.0091,
      "step": 17370
    },
    {
      "epoch": 0.11146992140023666,
      "grad_norm": 0.4783753454685211,
      "learning_rate": 5.573370959466393e-06,
      "loss": 0.0061,
      "step": 17380
    },
    {
      "epoch": 0.11153405829402276,
      "grad_norm": 0.2818965017795563,
      "learning_rate": 5.576577732170344e-06,
      "loss": 0.0101,
      "step": 17390
    },
    {
      "epoch": 0.11159819518780886,
      "grad_norm": 0.3929568827152252,
      "learning_rate": 5.579784504874295e-06,
      "loss": 0.0058,
      "step": 17400
    },
    {
      "epoch": 0.11166233208159496,
      "grad_norm": 0.3165569007396698,
      "learning_rate": 5.582991277578246e-06,
      "loss": 0.0107,
      "step": 17410
    },
    {
      "epoch": 0.11172646897538105,
      "grad_norm": 0.37585151195526123,
      "learning_rate": 5.586198050282197e-06,
      "loss": 0.0096,
      "step": 17420
    },
    {
      "epoch": 0.11179060586916716,
      "grad_norm": 0.3941322863101959,
      "learning_rate": 5.589404822986147e-06,
      "loss": 0.01,
      "step": 17430
    },
    {
      "epoch": 0.11185474276295325,
      "grad_norm": 0.38753095269203186,
      "learning_rate": 5.592611595690098e-06,
      "loss": 0.0072,
      "step": 17440
    },
    {
      "epoch": 0.11191887965673934,
      "grad_norm": 0.6736488938331604,
      "learning_rate": 5.5958183683940495e-06,
      "loss": 0.0106,
      "step": 17450
    },
    {
      "epoch": 0.11198301655052544,
      "grad_norm": 0.45821017026901245,
      "learning_rate": 5.599025141097999e-06,
      "loss": 0.0115,
      "step": 17460
    },
    {
      "epoch": 0.11204715344431154,
      "grad_norm": 0.3283272981643677,
      "learning_rate": 5.60223191380195e-06,
      "loss": 0.0088,
      "step": 17470
    },
    {
      "epoch": 0.11211129033809764,
      "grad_norm": 0.7205197215080261,
      "learning_rate": 5.605438686505901e-06,
      "loss": 0.0086,
      "step": 17480
    },
    {
      "epoch": 0.11217542723188373,
      "grad_norm": 0.6044069528579712,
      "learning_rate": 5.608645459209851e-06,
      "loss": 0.0074,
      "step": 17490
    },
    {
      "epoch": 0.11223956412566984,
      "grad_norm": 0.39601925015449524,
      "learning_rate": 5.6118522319138025e-06,
      "loss": 0.0064,
      "step": 17500
    },
    {
      "epoch": 0.11230370101945593,
      "grad_norm": 0.21867237985134125,
      "learning_rate": 5.615059004617754e-06,
      "loss": 0.0087,
      "step": 17510
    },
    {
      "epoch": 0.11236783791324202,
      "grad_norm": 0.1866510808467865,
      "learning_rate": 5.618265777321704e-06,
      "loss": 0.0075,
      "step": 17520
    },
    {
      "epoch": 0.11243197480702811,
      "grad_norm": 0.4482560455799103,
      "learning_rate": 5.621472550025655e-06,
      "loss": 0.0118,
      "step": 17530
    },
    {
      "epoch": 0.11249611170081422,
      "grad_norm": 0.6520987153053284,
      "learning_rate": 5.624679322729606e-06,
      "loss": 0.0086,
      "step": 17540
    },
    {
      "epoch": 0.11256024859460032,
      "grad_norm": 0.4319920837879181,
      "learning_rate": 5.6278860954335555e-06,
      "loss": 0.0047,
      "step": 17550
    },
    {
      "epoch": 0.11262438548838641,
      "grad_norm": 0.8728720545768738,
      "learning_rate": 5.631092868137507e-06,
      "loss": 0.0112,
      "step": 17560
    },
    {
      "epoch": 0.11268852238217252,
      "grad_norm": 0.29425284266471863,
      "learning_rate": 5.634299640841458e-06,
      "loss": 0.0082,
      "step": 17570
    },
    {
      "epoch": 0.11275265927595861,
      "grad_norm": 0.5924003720283508,
      "learning_rate": 5.637506413545408e-06,
      "loss": 0.0089,
      "step": 17580
    },
    {
      "epoch": 0.1128167961697447,
      "grad_norm": 0.6752727031707764,
      "learning_rate": 5.640713186249359e-06,
      "loss": 0.009,
      "step": 17590
    },
    {
      "epoch": 0.1128809330635308,
      "grad_norm": 0.5387188196182251,
      "learning_rate": 5.64391995895331e-06,
      "loss": 0.0127,
      "step": 17600
    },
    {
      "epoch": 0.1129450699573169,
      "grad_norm": 0.38861316442489624,
      "learning_rate": 5.6471267316572605e-06,
      "loss": 0.0095,
      "step": 17610
    },
    {
      "epoch": 0.113009206851103,
      "grad_norm": 0.7571175694465637,
      "learning_rate": 5.650333504361212e-06,
      "loss": 0.0092,
      "step": 17620
    },
    {
      "epoch": 0.11307334374488909,
      "grad_norm": 0.10430482029914856,
      "learning_rate": 5.653540277065163e-06,
      "loss": 0.0079,
      "step": 17630
    },
    {
      "epoch": 0.1131374806386752,
      "grad_norm": 0.27501529455184937,
      "learning_rate": 5.656747049769112e-06,
      "loss": 0.0076,
      "step": 17640
    },
    {
      "epoch": 0.11320161753246129,
      "grad_norm": 0.18850356340408325,
      "learning_rate": 5.659953822473063e-06,
      "loss": 0.0071,
      "step": 17650
    },
    {
      "epoch": 0.11326575442624738,
      "grad_norm": 0.7970367074012756,
      "learning_rate": 5.663160595177014e-06,
      "loss": 0.0102,
      "step": 17660
    },
    {
      "epoch": 0.11332989132003347,
      "grad_norm": 0.5236055254936218,
      "learning_rate": 5.666367367880965e-06,
      "loss": 0.0086,
      "step": 17670
    },
    {
      "epoch": 0.11339402821381958,
      "grad_norm": 0.45878875255584717,
      "learning_rate": 5.669574140584916e-06,
      "loss": 0.0083,
      "step": 17680
    },
    {
      "epoch": 0.11345816510760567,
      "grad_norm": 0.3952488303184509,
      "learning_rate": 5.672780913288867e-06,
      "loss": 0.0059,
      "step": 17690
    },
    {
      "epoch": 0.11352230200139177,
      "grad_norm": 0.28443267941474915,
      "learning_rate": 5.675987685992818e-06,
      "loss": 0.0111,
      "step": 17700
    },
    {
      "epoch": 0.11358643889517787,
      "grad_norm": 0.5733482837677002,
      "learning_rate": 5.679194458696768e-06,
      "loss": 0.0112,
      "step": 17710
    },
    {
      "epoch": 0.11365057578896397,
      "grad_norm": 0.6616878509521484,
      "learning_rate": 5.682401231400719e-06,
      "loss": 0.0093,
      "step": 17720
    },
    {
      "epoch": 0.11371471268275006,
      "grad_norm": 0.4348270893096924,
      "learning_rate": 5.685608004104669e-06,
      "loss": 0.0063,
      "step": 17730
    },
    {
      "epoch": 0.11377884957653615,
      "grad_norm": 0.2959847152233124,
      "learning_rate": 5.68881477680862e-06,
      "loss": 0.006,
      "step": 17740
    },
    {
      "epoch": 0.11384298647032226,
      "grad_norm": 0.814254641532898,
      "learning_rate": 5.692021549512571e-06,
      "loss": 0.0112,
      "step": 17750
    },
    {
      "epoch": 0.11390712336410835,
      "grad_norm": 0.40536028146743774,
      "learning_rate": 5.695228322216521e-06,
      "loss": 0.0125,
      "step": 17760
    },
    {
      "epoch": 0.11397126025789445,
      "grad_norm": 0.3108545243740082,
      "learning_rate": 5.698435094920472e-06,
      "loss": 0.0095,
      "step": 17770
    },
    {
      "epoch": 0.11403539715168055,
      "grad_norm": 0.09603261202573776,
      "learning_rate": 5.7016418676244235e-06,
      "loss": 0.0065,
      "step": 17780
    },
    {
      "epoch": 0.11409953404546665,
      "grad_norm": 1.1272550821304321,
      "learning_rate": 5.704848640328375e-06,
      "loss": 0.0143,
      "step": 17790
    },
    {
      "epoch": 0.11416367093925274,
      "grad_norm": 0.629813551902771,
      "learning_rate": 5.708055413032325e-06,
      "loss": 0.0089,
      "step": 17800
    },
    {
      "epoch": 0.11422780783303883,
      "grad_norm": 0.2537789046764374,
      "learning_rate": 5.711262185736276e-06,
      "loss": 0.0089,
      "step": 17810
    },
    {
      "epoch": 0.11429194472682494,
      "grad_norm": 0.7877181768417358,
      "learning_rate": 5.714468958440225e-06,
      "loss": 0.0093,
      "step": 17820
    },
    {
      "epoch": 0.11435608162061103,
      "grad_norm": 0.498452365398407,
      "learning_rate": 5.7176757311441765e-06,
      "loss": 0.0138,
      "step": 17830
    },
    {
      "epoch": 0.11442021851439713,
      "grad_norm": 0.41156306862831116,
      "learning_rate": 5.720882503848128e-06,
      "loss": 0.0083,
      "step": 17840
    },
    {
      "epoch": 0.11448435540818323,
      "grad_norm": 0.23975320160388947,
      "learning_rate": 5.724089276552079e-06,
      "loss": 0.0092,
      "step": 17850
    },
    {
      "epoch": 0.11454849230196933,
      "grad_norm": 0.5480472445487976,
      "learning_rate": 5.727296049256029e-06,
      "loss": 0.0095,
      "step": 17860
    },
    {
      "epoch": 0.11461262919575542,
      "grad_norm": 0.6654664874076843,
      "learning_rate": 5.73050282195998e-06,
      "loss": 0.0119,
      "step": 17870
    },
    {
      "epoch": 0.11467676608954151,
      "grad_norm": 0.3661085367202759,
      "learning_rate": 5.733709594663931e-06,
      "loss": 0.0105,
      "step": 17880
    },
    {
      "epoch": 0.11474090298332762,
      "grad_norm": 0.34792476892471313,
      "learning_rate": 5.7369163673678815e-06,
      "loss": 0.0104,
      "step": 17890
    },
    {
      "epoch": 0.11480503987711371,
      "grad_norm": 0.8125250339508057,
      "learning_rate": 5.740123140071833e-06,
      "loss": 0.0086,
      "step": 17900
    },
    {
      "epoch": 0.1148691767708998,
      "grad_norm": 0.4975816309452057,
      "learning_rate": 5.743329912775782e-06,
      "loss": 0.0091,
      "step": 17910
    },
    {
      "epoch": 0.1149333136646859,
      "grad_norm": 0.6501782536506653,
      "learning_rate": 5.746536685479733e-06,
      "loss": 0.0095,
      "step": 17920
    },
    {
      "epoch": 0.114997450558472,
      "grad_norm": 0.38356733322143555,
      "learning_rate": 5.749743458183684e-06,
      "loss": 0.0116,
      "step": 17930
    },
    {
      "epoch": 0.1150615874522581,
      "grad_norm": 0.6213951706886292,
      "learning_rate": 5.752950230887635e-06,
      "loss": 0.0111,
      "step": 17940
    },
    {
      "epoch": 0.11512572434604419,
      "grad_norm": 0.43015363812446594,
      "learning_rate": 5.756157003591586e-06,
      "loss": 0.009,
      "step": 17950
    },
    {
      "epoch": 0.1151898612398303,
      "grad_norm": 0.1405862420797348,
      "learning_rate": 5.759363776295537e-06,
      "loss": 0.0115,
      "step": 17960
    },
    {
      "epoch": 0.11525399813361639,
      "grad_norm": 0.4237961769104004,
      "learning_rate": 5.762570548999488e-06,
      "loss": 0.0084,
      "step": 17970
    },
    {
      "epoch": 0.11531813502740249,
      "grad_norm": 0.9272621870040894,
      "learning_rate": 5.765777321703438e-06,
      "loss": 0.0079,
      "step": 17980
    },
    {
      "epoch": 0.11538227192118858,
      "grad_norm": 0.24716678261756897,
      "learning_rate": 5.768984094407389e-06,
      "loss": 0.0095,
      "step": 17990
    },
    {
      "epoch": 0.11544640881497469,
      "grad_norm": 0.4450685679912567,
      "learning_rate": 5.7721908671113395e-06,
      "loss": 0.0083,
      "step": 18000
    },
    {
      "epoch": 0.11551054570876078,
      "grad_norm": 0.5911607146263123,
      "learning_rate": 5.77539763981529e-06,
      "loss": 0.0093,
      "step": 18010
    },
    {
      "epoch": 0.11557468260254687,
      "grad_norm": 0.382109671831131,
      "learning_rate": 5.778604412519241e-06,
      "loss": 0.0062,
      "step": 18020
    },
    {
      "epoch": 0.11563881949633298,
      "grad_norm": 0.4133998453617096,
      "learning_rate": 5.781811185223192e-06,
      "loss": 0.0075,
      "step": 18030
    },
    {
      "epoch": 0.11570295639011907,
      "grad_norm": 0.38716354966163635,
      "learning_rate": 5.785017957927142e-06,
      "loss": 0.008,
      "step": 18040
    },
    {
      "epoch": 0.11576709328390516,
      "grad_norm": 0.4410455822944641,
      "learning_rate": 5.788224730631093e-06,
      "loss": 0.007,
      "step": 18050
    },
    {
      "epoch": 0.11583123017769126,
      "grad_norm": 0.7385231852531433,
      "learning_rate": 5.7914315033350445e-06,
      "loss": 0.0108,
      "step": 18060
    },
    {
      "epoch": 0.11589536707147736,
      "grad_norm": 1.0614713430404663,
      "learning_rate": 5.794638276038996e-06,
      "loss": 0.0151,
      "step": 18070
    },
    {
      "epoch": 0.11595950396526346,
      "grad_norm": 0.44553372263908386,
      "learning_rate": 5.797845048742946e-06,
      "loss": 0.0069,
      "step": 18080
    },
    {
      "epoch": 0.11602364085904955,
      "grad_norm": 0.4055965840816498,
      "learning_rate": 5.801051821446896e-06,
      "loss": 0.0118,
      "step": 18090
    },
    {
      "epoch": 0.11608777775283566,
      "grad_norm": 0.5591070652008057,
      "learning_rate": 5.804258594150846e-06,
      "loss": 0.0126,
      "step": 18100
    },
    {
      "epoch": 0.11615191464662175,
      "grad_norm": 0.6480048298835754,
      "learning_rate": 5.8074653668547975e-06,
      "loss": 0.0077,
      "step": 18110
    },
    {
      "epoch": 0.11621605154040784,
      "grad_norm": 0.34888118505477905,
      "learning_rate": 5.810672139558749e-06,
      "loss": 0.0123,
      "step": 18120
    },
    {
      "epoch": 0.11628018843419394,
      "grad_norm": 0.7395862340927124,
      "learning_rate": 5.813878912262699e-06,
      "loss": 0.0102,
      "step": 18130
    },
    {
      "epoch": 0.11634432532798004,
      "grad_norm": 0.34636107087135315,
      "learning_rate": 5.81708568496665e-06,
      "loss": 0.0075,
      "step": 18140
    },
    {
      "epoch": 0.11640846222176614,
      "grad_norm": 0.6415095329284668,
      "learning_rate": 5.820292457670601e-06,
      "loss": 0.0116,
      "step": 18150
    },
    {
      "epoch": 0.11647259911555223,
      "grad_norm": 0.2900804579257965,
      "learning_rate": 5.823499230374552e-06,
      "loss": 0.0098,
      "step": 18160
    },
    {
      "epoch": 0.11653673600933834,
      "grad_norm": 0.28599801659584045,
      "learning_rate": 5.8267060030785025e-06,
      "loss": 0.0104,
      "step": 18170
    },
    {
      "epoch": 0.11660087290312443,
      "grad_norm": 1.0962491035461426,
      "learning_rate": 5.829912775782453e-06,
      "loss": 0.0078,
      "step": 18180
    },
    {
      "epoch": 0.11666500979691052,
      "grad_norm": 0.8533971309661865,
      "learning_rate": 5.833119548486403e-06,
      "loss": 0.0096,
      "step": 18190
    },
    {
      "epoch": 0.11672914669069662,
      "grad_norm": 0.47534847259521484,
      "learning_rate": 5.836326321190354e-06,
      "loss": 0.0094,
      "step": 18200
    },
    {
      "epoch": 0.11679328358448272,
      "grad_norm": 0.3004346489906311,
      "learning_rate": 5.839533093894305e-06,
      "loss": 0.0078,
      "step": 18210
    },
    {
      "epoch": 0.11685742047826882,
      "grad_norm": 0.4630396366119385,
      "learning_rate": 5.842739866598256e-06,
      "loss": 0.0086,
      "step": 18220
    },
    {
      "epoch": 0.11692155737205491,
      "grad_norm": 0.5481510162353516,
      "learning_rate": 5.845946639302207e-06,
      "loss": 0.0122,
      "step": 18230
    },
    {
      "epoch": 0.11698569426584102,
      "grad_norm": 0.3296847641468048,
      "learning_rate": 5.849153412006158e-06,
      "loss": 0.009,
      "step": 18240
    },
    {
      "epoch": 0.11704983115962711,
      "grad_norm": 0.6358070969581604,
      "learning_rate": 5.852360184710109e-06,
      "loss": 0.0112,
      "step": 18250
    },
    {
      "epoch": 0.1171139680534132,
      "grad_norm": 0.3781026005744934,
      "learning_rate": 5.855566957414059e-06,
      "loss": 0.0095,
      "step": 18260
    },
    {
      "epoch": 0.1171781049471993,
      "grad_norm": 0.4986782968044281,
      "learning_rate": 5.858773730118009e-06,
      "loss": 0.0087,
      "step": 18270
    },
    {
      "epoch": 0.1172422418409854,
      "grad_norm": 0.21493154764175415,
      "learning_rate": 5.86198050282196e-06,
      "loss": 0.0103,
      "step": 18280
    },
    {
      "epoch": 0.1173063787347715,
      "grad_norm": 0.24460344016551971,
      "learning_rate": 5.865187275525911e-06,
      "loss": 0.009,
      "step": 18290
    },
    {
      "epoch": 0.11737051562855759,
      "grad_norm": 0.372915655374527,
      "learning_rate": 5.868394048229862e-06,
      "loss": 0.0081,
      "step": 18300
    },
    {
      "epoch": 0.1174346525223437,
      "grad_norm": 0.8102298974990845,
      "learning_rate": 5.871600820933813e-06,
      "loss": 0.0105,
      "step": 18310
    },
    {
      "epoch": 0.11749878941612979,
      "grad_norm": 0.6621906161308289,
      "learning_rate": 5.874807593637763e-06,
      "loss": 0.0131,
      "step": 18320
    },
    {
      "epoch": 0.11756292630991588,
      "grad_norm": 0.6036761403083801,
      "learning_rate": 5.878014366341714e-06,
      "loss": 0.0095,
      "step": 18330
    },
    {
      "epoch": 0.11762706320370198,
      "grad_norm": 0.5530111193656921,
      "learning_rate": 5.8812211390456655e-06,
      "loss": 0.0106,
      "step": 18340
    },
    {
      "epoch": 0.11769120009748808,
      "grad_norm": 0.40899187326431274,
      "learning_rate": 5.884427911749616e-06,
      "loss": 0.0077,
      "step": 18350
    },
    {
      "epoch": 0.11775533699127418,
      "grad_norm": 0.6335707902908325,
      "learning_rate": 5.887634684453566e-06,
      "loss": 0.0092,
      "step": 18360
    },
    {
      "epoch": 0.11781947388506027,
      "grad_norm": 0.5672249794006348,
      "learning_rate": 5.890841457157517e-06,
      "loss": 0.0116,
      "step": 18370
    },
    {
      "epoch": 0.11788361077884638,
      "grad_norm": 0.4070172905921936,
      "learning_rate": 5.894048229861467e-06,
      "loss": 0.0127,
      "step": 18380
    },
    {
      "epoch": 0.11794774767263247,
      "grad_norm": 0.4748009145259857,
      "learning_rate": 5.8972550025654185e-06,
      "loss": 0.0111,
      "step": 18390
    },
    {
      "epoch": 0.11801188456641856,
      "grad_norm": 0.6402952671051025,
      "learning_rate": 5.90046177526937e-06,
      "loss": 0.0129,
      "step": 18400
    },
    {
      "epoch": 0.11807602146020466,
      "grad_norm": 0.706387460231781,
      "learning_rate": 5.90366854797332e-06,
      "loss": 0.0122,
      "step": 18410
    },
    {
      "epoch": 0.11814015835399076,
      "grad_norm": 1.0082502365112305,
      "learning_rate": 5.906875320677271e-06,
      "loss": 0.007,
      "step": 18420
    },
    {
      "epoch": 0.11820429524777686,
      "grad_norm": 0.30910611152648926,
      "learning_rate": 5.910082093381222e-06,
      "loss": 0.0085,
      "step": 18430
    },
    {
      "epoch": 0.11826843214156295,
      "grad_norm": 0.7558024525642395,
      "learning_rate": 5.913288866085173e-06,
      "loss": 0.0082,
      "step": 18440
    },
    {
      "epoch": 0.11833256903534906,
      "grad_norm": 0.5140365958213806,
      "learning_rate": 5.916495638789123e-06,
      "loss": 0.0082,
      "step": 18450
    },
    {
      "epoch": 0.11839670592913515,
      "grad_norm": 0.46017810702323914,
      "learning_rate": 5.919702411493074e-06,
      "loss": 0.0094,
      "step": 18460
    },
    {
      "epoch": 0.11846084282292124,
      "grad_norm": 0.40746715664863586,
      "learning_rate": 5.922909184197024e-06,
      "loss": 0.0104,
      "step": 18470
    },
    {
      "epoch": 0.11852497971670733,
      "grad_norm": 0.6365993022918701,
      "learning_rate": 5.926115956900975e-06,
      "loss": 0.0124,
      "step": 18480
    },
    {
      "epoch": 0.11858911661049344,
      "grad_norm": 0.47270020842552185,
      "learning_rate": 5.929322729604926e-06,
      "loss": 0.0069,
      "step": 18490
    },
    {
      "epoch": 0.11865325350427953,
      "grad_norm": 0.4353969097137451,
      "learning_rate": 5.9325295023088765e-06,
      "loss": 0.0093,
      "step": 18500
    },
    {
      "epoch": 0.11871739039806563,
      "grad_norm": 0.7127542495727539,
      "learning_rate": 5.935736275012828e-06,
      "loss": 0.0087,
      "step": 18510
    },
    {
      "epoch": 0.11878152729185174,
      "grad_norm": 0.3726551830768585,
      "learning_rate": 5.938943047716779e-06,
      "loss": 0.0065,
      "step": 18520
    },
    {
      "epoch": 0.11884566418563783,
      "grad_norm": 0.8096532821655273,
      "learning_rate": 5.94214982042073e-06,
      "loss": 0.0101,
      "step": 18530
    },
    {
      "epoch": 0.11890980107942392,
      "grad_norm": 0.38727736473083496,
      "learning_rate": 5.945356593124679e-06,
      "loss": 0.0148,
      "step": 18540
    },
    {
      "epoch": 0.11897393797321001,
      "grad_norm": 0.3246394097805023,
      "learning_rate": 5.94856336582863e-06,
      "loss": 0.0128,
      "step": 18550
    },
    {
      "epoch": 0.11903807486699612,
      "grad_norm": 0.3038940131664276,
      "learning_rate": 5.951770138532581e-06,
      "loss": 0.0101,
      "step": 18560
    },
    {
      "epoch": 0.11910221176078221,
      "grad_norm": 0.8246645927429199,
      "learning_rate": 5.954976911236532e-06,
      "loss": 0.0117,
      "step": 18570
    },
    {
      "epoch": 0.11916634865456831,
      "grad_norm": 0.45428499579429626,
      "learning_rate": 5.958183683940483e-06,
      "loss": 0.0119,
      "step": 18580
    },
    {
      "epoch": 0.1192304855483544,
      "grad_norm": 0.7145015001296997,
      "learning_rate": 5.961390456644434e-06,
      "loss": 0.0061,
      "step": 18590
    },
    {
      "epoch": 0.11929462244214051,
      "grad_norm": 0.6768432259559631,
      "learning_rate": 5.964597229348384e-06,
      "loss": 0.0087,
      "step": 18600
    },
    {
      "epoch": 0.1193587593359266,
      "grad_norm": 0.7962307929992676,
      "learning_rate": 5.967804002052335e-06,
      "loss": 0.0094,
      "step": 18610
    },
    {
      "epoch": 0.1194228962297127,
      "grad_norm": 0.40162888169288635,
      "learning_rate": 5.9710107747562865e-06,
      "loss": 0.011,
      "step": 18620
    },
    {
      "epoch": 0.1194870331234988,
      "grad_norm": 0.6926724910736084,
      "learning_rate": 5.974217547460236e-06,
      "loss": 0.0105,
      "step": 18630
    },
    {
      "epoch": 0.1195511700172849,
      "grad_norm": 0.4717659652233124,
      "learning_rate": 5.977424320164187e-06,
      "loss": 0.0086,
      "step": 18640
    },
    {
      "epoch": 0.11961530691107099,
      "grad_norm": 0.7448294162750244,
      "learning_rate": 5.980631092868137e-06,
      "loss": 0.0092,
      "step": 18650
    },
    {
      "epoch": 0.11967944380485708,
      "grad_norm": 0.590056836605072,
      "learning_rate": 5.983837865572088e-06,
      "loss": 0.0093,
      "step": 18660
    },
    {
      "epoch": 0.11974358069864319,
      "grad_norm": 0.7150057554244995,
      "learning_rate": 5.9870446382760395e-06,
      "loss": 0.0106,
      "step": 18670
    },
    {
      "epoch": 0.11980771759242928,
      "grad_norm": 0.5308887958526611,
      "learning_rate": 5.990251410979991e-06,
      "loss": 0.0119,
      "step": 18680
    },
    {
      "epoch": 0.11987185448621537,
      "grad_norm": 0.503163754940033,
      "learning_rate": 5.993458183683941e-06,
      "loss": 0.0076,
      "step": 18690
    },
    {
      "epoch": 0.11993599138000148,
      "grad_norm": 0.6761953234672546,
      "learning_rate": 5.996664956387892e-06,
      "loss": 0.009,
      "step": 18700
    },
    {
      "epoch": 0.12000012827378757,
      "grad_norm": 0.4515625536441803,
      "learning_rate": 5.999871729091843e-06,
      "loss": 0.0169,
      "step": 18710
    },
    {
      "epoch": 0.12006426516757367,
      "grad_norm": 0.42195725440979004,
      "learning_rate": 6.0030785017957925e-06,
      "loss": 0.0081,
      "step": 18720
    },
    {
      "epoch": 0.12012840206135976,
      "grad_norm": 0.21539296209812164,
      "learning_rate": 6.006285274499744e-06,
      "loss": 0.0097,
      "step": 18730
    },
    {
      "epoch": 0.12019253895514587,
      "grad_norm": 0.7816265225410461,
      "learning_rate": 6.009492047203695e-06,
      "loss": 0.0137,
      "step": 18740
    },
    {
      "epoch": 0.12025667584893196,
      "grad_norm": 0.4108607769012451,
      "learning_rate": 6.012698819907645e-06,
      "loss": 0.008,
      "step": 18750
    },
    {
      "epoch": 0.12032081274271805,
      "grad_norm": 0.6747710704803467,
      "learning_rate": 6.015905592611596e-06,
      "loss": 0.0098,
      "step": 18760
    },
    {
      "epoch": 0.12038494963650416,
      "grad_norm": 0.4920329749584198,
      "learning_rate": 6.019112365315547e-06,
      "loss": 0.0082,
      "step": 18770
    },
    {
      "epoch": 0.12044908653029025,
      "grad_norm": 0.46777355670928955,
      "learning_rate": 6.0223191380194975e-06,
      "loss": 0.0101,
      "step": 18780
    },
    {
      "epoch": 0.12051322342407635,
      "grad_norm": 0.5914434790611267,
      "learning_rate": 6.025525910723449e-06,
      "loss": 0.0118,
      "step": 18790
    },
    {
      "epoch": 0.12057736031786244,
      "grad_norm": 0.5254697799682617,
      "learning_rate": 6.0287326834274e-06,
      "loss": 0.0072,
      "step": 18800
    },
    {
      "epoch": 0.12064149721164855,
      "grad_norm": 0.6814507246017456,
      "learning_rate": 6.031939456131351e-06,
      "loss": 0.009,
      "step": 18810
    },
    {
      "epoch": 0.12070563410543464,
      "grad_norm": 0.4084748923778534,
      "learning_rate": 6.0351462288353e-06,
      "loss": 0.0088,
      "step": 18820
    },
    {
      "epoch": 0.12076977099922073,
      "grad_norm": 0.3717684745788574,
      "learning_rate": 6.038353001539251e-06,
      "loss": 0.0063,
      "step": 18830
    },
    {
      "epoch": 0.12083390789300684,
      "grad_norm": 0.6668663024902344,
      "learning_rate": 6.041559774243202e-06,
      "loss": 0.0115,
      "step": 18840
    },
    {
      "epoch": 0.12089804478679293,
      "grad_norm": 0.49359339475631714,
      "learning_rate": 6.044766546947153e-06,
      "loss": 0.0064,
      "step": 18850
    },
    {
      "epoch": 0.12096218168057903,
      "grad_norm": 0.7852402329444885,
      "learning_rate": 6.047973319651104e-06,
      "loss": 0.0077,
      "step": 18860
    },
    {
      "epoch": 0.12102631857436512,
      "grad_norm": 0.5898550748825073,
      "learning_rate": 6.051180092355054e-06,
      "loss": 0.0087,
      "step": 18870
    },
    {
      "epoch": 0.12109045546815123,
      "grad_norm": 0.46745458245277405,
      "learning_rate": 6.054386865059005e-06,
      "loss": 0.0078,
      "step": 18880
    },
    {
      "epoch": 0.12115459236193732,
      "grad_norm": 0.5510157942771912,
      "learning_rate": 6.057593637762956e-06,
      "loss": 0.0105,
      "step": 18890
    },
    {
      "epoch": 0.12121872925572341,
      "grad_norm": 0.6177038550376892,
      "learning_rate": 6.0608004104669075e-06,
      "loss": 0.0103,
      "step": 18900
    },
    {
      "epoch": 0.12128286614950952,
      "grad_norm": 0.18779337406158447,
      "learning_rate": 6.064007183170857e-06,
      "loss": 0.0062,
      "step": 18910
    },
    {
      "epoch": 0.12134700304329561,
      "grad_norm": 0.5417159199714661,
      "learning_rate": 6.067213955874808e-06,
      "loss": 0.008,
      "step": 18920
    },
    {
      "epoch": 0.1214111399370817,
      "grad_norm": 0.25111818313598633,
      "learning_rate": 6.070420728578758e-06,
      "loss": 0.0056,
      "step": 18930
    },
    {
      "epoch": 0.1214752768308678,
      "grad_norm": 0.21491633355617523,
      "learning_rate": 6.073627501282709e-06,
      "loss": 0.0081,
      "step": 18940
    },
    {
      "epoch": 0.1215394137246539,
      "grad_norm": 0.5052981972694397,
      "learning_rate": 6.0768342739866605e-06,
      "loss": 0.0075,
      "step": 18950
    },
    {
      "epoch": 0.12160355061844,
      "grad_norm": 0.7487515211105347,
      "learning_rate": 6.080041046690612e-06,
      "loss": 0.0069,
      "step": 18960
    },
    {
      "epoch": 0.12166768751222609,
      "grad_norm": 0.7578110694885254,
      "learning_rate": 6.083247819394562e-06,
      "loss": 0.0089,
      "step": 18970
    },
    {
      "epoch": 0.1217318244060122,
      "grad_norm": 0.18499760329723358,
      "learning_rate": 6.086454592098513e-06,
      "loss": 0.0092,
      "step": 18980
    },
    {
      "epoch": 0.12179596129979829,
      "grad_norm": 0.33362337946891785,
      "learning_rate": 6.089661364802464e-06,
      "loss": 0.0133,
      "step": 18990
    },
    {
      "epoch": 0.12186009819358438,
      "grad_norm": 0.8806824088096619,
      "learning_rate": 6.0928681375064135e-06,
      "loss": 0.0083,
      "step": 19000
    },
    {
      "epoch": 0.12192423508737048,
      "grad_norm": 0.30265992879867554,
      "learning_rate": 6.096074910210365e-06,
      "loss": 0.0111,
      "step": 19010
    },
    {
      "epoch": 0.12198837198115658,
      "grad_norm": 0.4938734471797943,
      "learning_rate": 6.099281682914315e-06,
      "loss": 0.0087,
      "step": 19020
    },
    {
      "epoch": 0.12205250887494268,
      "grad_norm": 0.9325695037841797,
      "learning_rate": 6.102488455618266e-06,
      "loss": 0.0142,
      "step": 19030
    },
    {
      "epoch": 0.12211664576872877,
      "grad_norm": 0.20962482690811157,
      "learning_rate": 6.105695228322217e-06,
      "loss": 0.0085,
      "step": 19040
    },
    {
      "epoch": 0.12218078266251488,
      "grad_norm": 0.37175944447517395,
      "learning_rate": 6.108902001026168e-06,
      "loss": 0.0083,
      "step": 19050
    },
    {
      "epoch": 0.12224491955630097,
      "grad_norm": 0.3731585144996643,
      "learning_rate": 6.1121087737301185e-06,
      "loss": 0.0056,
      "step": 19060
    },
    {
      "epoch": 0.12230905645008706,
      "grad_norm": 0.9835683107376099,
      "learning_rate": 6.11531554643407e-06,
      "loss": 0.0102,
      "step": 19070
    },
    {
      "epoch": 0.12237319334387316,
      "grad_norm": 0.548466145992279,
      "learning_rate": 6.118522319138021e-06,
      "loss": 0.01,
      "step": 19080
    },
    {
      "epoch": 0.12243733023765926,
      "grad_norm": 0.1948307752609253,
      "learning_rate": 6.12172909184197e-06,
      "loss": 0.008,
      "step": 19090
    },
    {
      "epoch": 0.12250146713144536,
      "grad_norm": 0.28201523423194885,
      "learning_rate": 6.124935864545921e-06,
      "loss": 0.008,
      "step": 19100
    },
    {
      "epoch": 0.12256560402523145,
      "grad_norm": 0.35290905833244324,
      "learning_rate": 6.128142637249872e-06,
      "loss": 0.0074,
      "step": 19110
    },
    {
      "epoch": 0.12262974091901756,
      "grad_norm": 0.45489513874053955,
      "learning_rate": 6.131349409953823e-06,
      "loss": 0.0123,
      "step": 19120
    },
    {
      "epoch": 0.12269387781280365,
      "grad_norm": 0.23540635406970978,
      "learning_rate": 6.134556182657774e-06,
      "loss": 0.009,
      "step": 19130
    },
    {
      "epoch": 0.12275801470658974,
      "grad_norm": 0.33407628536224365,
      "learning_rate": 6.137762955361725e-06,
      "loss": 0.0103,
      "step": 19140
    },
    {
      "epoch": 0.12282215160037584,
      "grad_norm": 0.6046065092086792,
      "learning_rate": 6.140969728065675e-06,
      "loss": 0.0083,
      "step": 19150
    },
    {
      "epoch": 0.12288628849416194,
      "grad_norm": 0.2274067997932434,
      "learning_rate": 6.144176500769626e-06,
      "loss": 0.0065,
      "step": 19160
    },
    {
      "epoch": 0.12295042538794804,
      "grad_norm": 0.2769779562950134,
      "learning_rate": 6.147383273473577e-06,
      "loss": 0.0086,
      "step": 19170
    },
    {
      "epoch": 0.12301456228173413,
      "grad_norm": 0.7009875774383545,
      "learning_rate": 6.150590046177527e-06,
      "loss": 0.0067,
      "step": 19180
    },
    {
      "epoch": 0.12307869917552024,
      "grad_norm": 0.6626484394073486,
      "learning_rate": 6.153796818881478e-06,
      "loss": 0.023,
      "step": 19190
    },
    {
      "epoch": 0.12314283606930633,
      "grad_norm": 0.43245089054107666,
      "learning_rate": 6.157003591585429e-06,
      "loss": 0.0103,
      "step": 19200
    },
    {
      "epoch": 0.12320697296309242,
      "grad_norm": 0.43614763021469116,
      "learning_rate": 6.160210364289379e-06,
      "loss": 0.0076,
      "step": 19210
    },
    {
      "epoch": 0.12327110985687852,
      "grad_norm": 0.5906426310539246,
      "learning_rate": 6.16341713699333e-06,
      "loss": 0.0117,
      "step": 19220
    },
    {
      "epoch": 0.12333524675066462,
      "grad_norm": 0.40125492215156555,
      "learning_rate": 6.1666239096972815e-06,
      "loss": 0.0064,
      "step": 19230
    },
    {
      "epoch": 0.12339938364445072,
      "grad_norm": 0.6000845432281494,
      "learning_rate": 6.169830682401232e-06,
      "loss": 0.0071,
      "step": 19240
    },
    {
      "epoch": 0.12346352053823681,
      "grad_norm": 0.6457440257072449,
      "learning_rate": 6.173037455105183e-06,
      "loss": 0.0074,
      "step": 19250
    },
    {
      "epoch": 0.12352765743202292,
      "grad_norm": 0.29348060488700867,
      "learning_rate": 6.176244227809134e-06,
      "loss": 0.0065,
      "step": 19260
    },
    {
      "epoch": 0.12359179432580901,
      "grad_norm": 0.5351219177246094,
      "learning_rate": 6.1794510005130834e-06,
      "loss": 0.0086,
      "step": 19270
    },
    {
      "epoch": 0.1236559312195951,
      "grad_norm": 0.6515761613845825,
      "learning_rate": 6.1826577732170345e-06,
      "loss": 0.0105,
      "step": 19280
    },
    {
      "epoch": 0.1237200681133812,
      "grad_norm": 0.3923013508319855,
      "learning_rate": 6.185864545920986e-06,
      "loss": 0.0096,
      "step": 19290
    },
    {
      "epoch": 0.1237842050071673,
      "grad_norm": 0.48430967330932617,
      "learning_rate": 6.189071318624936e-06,
      "loss": 0.0058,
      "step": 19300
    },
    {
      "epoch": 0.1238483419009534,
      "grad_norm": 0.24645860493183136,
      "learning_rate": 6.192278091328887e-06,
      "loss": 0.0073,
      "step": 19310
    },
    {
      "epoch": 0.12391247879473949,
      "grad_norm": 0.27493560314178467,
      "learning_rate": 6.195484864032838e-06,
      "loss": 0.0062,
      "step": 19320
    },
    {
      "epoch": 0.12397661568852558,
      "grad_norm": 0.48881635069847107,
      "learning_rate": 6.198691636736789e-06,
      "loss": 0.008,
      "step": 19330
    },
    {
      "epoch": 0.12404075258231169,
      "grad_norm": 0.5472639799118042,
      "learning_rate": 6.2018984094407395e-06,
      "loss": 0.0108,
      "step": 19340
    },
    {
      "epoch": 0.12410488947609778,
      "grad_norm": 0.5145623683929443,
      "learning_rate": 6.205105182144691e-06,
      "loss": 0.01,
      "step": 19350
    },
    {
      "epoch": 0.12416902636988388,
      "grad_norm": 0.6084321141242981,
      "learning_rate": 6.20831195484864e-06,
      "loss": 0.0101,
      "step": 19360
    },
    {
      "epoch": 0.12423316326366998,
      "grad_norm": 0.11005605757236481,
      "learning_rate": 6.211518727552591e-06,
      "loss": 0.0054,
      "step": 19370
    },
    {
      "epoch": 0.12429730015745608,
      "grad_norm": 0.46959608793258667,
      "learning_rate": 6.214725500256542e-06,
      "loss": 0.0072,
      "step": 19380
    },
    {
      "epoch": 0.12436143705124217,
      "grad_norm": 0.7657153606414795,
      "learning_rate": 6.2179322729604926e-06,
      "loss": 0.0095,
      "step": 19390
    },
    {
      "epoch": 0.12442557394502826,
      "grad_norm": 0.552920401096344,
      "learning_rate": 6.221139045664444e-06,
      "loss": 0.0088,
      "step": 19400
    },
    {
      "epoch": 0.12448971083881437,
      "grad_norm": 0.5344424247741699,
      "learning_rate": 6.224345818368395e-06,
      "loss": 0.0085,
      "step": 19410
    },
    {
      "epoch": 0.12455384773260046,
      "grad_norm": 0.3879581689834595,
      "learning_rate": 6.227552591072346e-06,
      "loss": 0.0079,
      "step": 19420
    },
    {
      "epoch": 0.12461798462638655,
      "grad_norm": 0.521867036819458,
      "learning_rate": 6.230759363776296e-06,
      "loss": 0.0208,
      "step": 19430
    },
    {
      "epoch": 0.12468212152017266,
      "grad_norm": 0.6008889079093933,
      "learning_rate": 6.233966136480247e-06,
      "loss": 0.0084,
      "step": 19440
    },
    {
      "epoch": 0.12474625841395875,
      "grad_norm": 0.25370872020721436,
      "learning_rate": 6.237172909184197e-06,
      "loss": 0.0095,
      "step": 19450
    },
    {
      "epoch": 0.12481039530774485,
      "grad_norm": 0.3903500437736511,
      "learning_rate": 6.240379681888148e-06,
      "loss": 0.011,
      "step": 19460
    },
    {
      "epoch": 0.12487453220153094,
      "grad_norm": 0.3653107285499573,
      "learning_rate": 6.243586454592099e-06,
      "loss": 0.0052,
      "step": 19470
    },
    {
      "epoch": 0.12493866909531705,
      "grad_norm": 0.25027909874916077,
      "learning_rate": 6.24679322729605e-06,
      "loss": 0.0106,
      "step": 19480
    },
    {
      "epoch": 0.12500280598910316,
      "grad_norm": 0.2770934998989105,
      "learning_rate": 6.25e-06,
      "loss": 0.0063,
      "step": 19490
    },
    {
      "epoch": 0.12506694288288925,
      "grad_norm": 0.6216777563095093,
      "learning_rate": 6.253206772703951e-06,
      "loss": 0.0081,
      "step": 19500
    },
    {
      "epoch": 0.12513107977667534,
      "grad_norm": 0.46903955936431885,
      "learning_rate": 6.2564135454079025e-06,
      "loss": 0.0116,
      "step": 19510
    },
    {
      "epoch": 0.12519521667046143,
      "grad_norm": 0.5048975348472595,
      "learning_rate": 6.259620318111853e-06,
      "loss": 0.0094,
      "step": 19520
    },
    {
      "epoch": 0.12525935356424753,
      "grad_norm": 0.6152871251106262,
      "learning_rate": 6.262827090815804e-06,
      "loss": 0.0125,
      "step": 19530
    },
    {
      "epoch": 0.12532349045803362,
      "grad_norm": 0.5237569808959961,
      "learning_rate": 6.266033863519753e-06,
      "loss": 0.0143,
      "step": 19540
    },
    {
      "epoch": 0.1253876273518197,
      "grad_norm": 0.6239390969276428,
      "learning_rate": 6.2692406362237044e-06,
      "loss": 0.0111,
      "step": 19550
    },
    {
      "epoch": 0.12545176424560583,
      "grad_norm": 0.6508604288101196,
      "learning_rate": 6.2724474089276556e-06,
      "loss": 0.0093,
      "step": 19560
    },
    {
      "epoch": 0.12551590113939193,
      "grad_norm": 0.29125237464904785,
      "learning_rate": 6.275654181631607e-06,
      "loss": 0.0084,
      "step": 19570
    },
    {
      "epoch": 0.12558003803317802,
      "grad_norm": 0.5597888827323914,
      "learning_rate": 6.278860954335557e-06,
      "loss": 0.0094,
      "step": 19580
    },
    {
      "epoch": 0.12564417492696411,
      "grad_norm": 0.3374530076980591,
      "learning_rate": 6.282067727039508e-06,
      "loss": 0.0069,
      "step": 19590
    },
    {
      "epoch": 0.1257083118207502,
      "grad_norm": 0.6125375628471375,
      "learning_rate": 6.285274499743459e-06,
      "loss": 0.0111,
      "step": 19600
    },
    {
      "epoch": 0.1257724487145363,
      "grad_norm": 0.40054383873939514,
      "learning_rate": 6.288481272447409e-06,
      "loss": 0.0066,
      "step": 19610
    },
    {
      "epoch": 0.1258365856083224,
      "grad_norm": 0.18805621564388275,
      "learning_rate": 6.2916880451513605e-06,
      "loss": 0.0095,
      "step": 19620
    },
    {
      "epoch": 0.12590072250210851,
      "grad_norm": 0.34585040807724,
      "learning_rate": 6.294894817855311e-06,
      "loss": 0.0071,
      "step": 19630
    },
    {
      "epoch": 0.1259648593958946,
      "grad_norm": 0.5074530839920044,
      "learning_rate": 6.298101590559261e-06,
      "loss": 0.0055,
      "step": 19640
    },
    {
      "epoch": 0.1260289962896807,
      "grad_norm": 0.27524951100349426,
      "learning_rate": 6.301308363263212e-06,
      "loss": 0.007,
      "step": 19650
    },
    {
      "epoch": 0.1260931331834668,
      "grad_norm": 0.4609309434890747,
      "learning_rate": 6.304515135967163e-06,
      "loss": 0.0065,
      "step": 19660
    },
    {
      "epoch": 0.1261572700772529,
      "grad_norm": 0.46933484077453613,
      "learning_rate": 6.3077219086711136e-06,
      "loss": 0.0091,
      "step": 19670
    },
    {
      "epoch": 0.12622140697103898,
      "grad_norm": 0.542717456817627,
      "learning_rate": 6.310928681375065e-06,
      "loss": 0.0097,
      "step": 19680
    },
    {
      "epoch": 0.12628554386482507,
      "grad_norm": 0.3351193368434906,
      "learning_rate": 6.314135454079016e-06,
      "loss": 0.0067,
      "step": 19690
    },
    {
      "epoch": 0.12634968075861117,
      "grad_norm": 0.495302677154541,
      "learning_rate": 6.317342226782967e-06,
      "loss": 0.0101,
      "step": 19700
    },
    {
      "epoch": 0.1264138176523973,
      "grad_norm": 0.20892836153507233,
      "learning_rate": 6.320548999486917e-06,
      "loss": 0.0071,
      "step": 19710
    },
    {
      "epoch": 0.12647795454618338,
      "grad_norm": 0.417507529258728,
      "learning_rate": 6.3237557721908674e-06,
      "loss": 0.0106,
      "step": 19720
    },
    {
      "epoch": 0.12654209143996947,
      "grad_norm": 0.17814181745052338,
      "learning_rate": 6.326962544894818e-06,
      "loss": 0.0088,
      "step": 19730
    },
    {
      "epoch": 0.12660622833375557,
      "grad_norm": 0.524132251739502,
      "learning_rate": 6.330169317598769e-06,
      "loss": 0.0077,
      "step": 19740
    },
    {
      "epoch": 0.12667036522754166,
      "grad_norm": 1.6826726198196411,
      "learning_rate": 6.33337609030272e-06,
      "loss": 0.0126,
      "step": 19750
    },
    {
      "epoch": 0.12673450212132775,
      "grad_norm": 0.45619019865989685,
      "learning_rate": 6.33658286300667e-06,
      "loss": 0.0099,
      "step": 19760
    },
    {
      "epoch": 0.12679863901511385,
      "grad_norm": 0.6970486640930176,
      "learning_rate": 6.339789635710621e-06,
      "loss": 0.0101,
      "step": 19770
    },
    {
      "epoch": 0.12686277590889997,
      "grad_norm": 0.5018978118896484,
      "learning_rate": 6.342996408414572e-06,
      "loss": 0.0086,
      "step": 19780
    },
    {
      "epoch": 0.12692691280268606,
      "grad_norm": 0.8621292114257812,
      "learning_rate": 6.3462031811185235e-06,
      "loss": 0.0192,
      "step": 19790
    },
    {
      "epoch": 0.12699104969647215,
      "grad_norm": 0.6635532975196838,
      "learning_rate": 6.349409953822474e-06,
      "loss": 0.0082,
      "step": 19800
    },
    {
      "epoch": 0.12705518659025825,
      "grad_norm": 0.18149831891059875,
      "learning_rate": 6.352616726526424e-06,
      "loss": 0.0096,
      "step": 19810
    },
    {
      "epoch": 0.12711932348404434,
      "grad_norm": 0.22729986906051636,
      "learning_rate": 6.355823499230374e-06,
      "loss": 0.0109,
      "step": 19820
    },
    {
      "epoch": 0.12718346037783043,
      "grad_norm": 0.6454578042030334,
      "learning_rate": 6.3590302719343254e-06,
      "loss": 0.0081,
      "step": 19830
    },
    {
      "epoch": 0.12724759727161652,
      "grad_norm": 0.4148944020271301,
      "learning_rate": 6.3622370446382766e-06,
      "loss": 0.0118,
      "step": 19840
    },
    {
      "epoch": 0.12731173416540265,
      "grad_norm": 0.40859290957450867,
      "learning_rate": 6.365443817342228e-06,
      "loss": 0.0095,
      "step": 19850
    },
    {
      "epoch": 0.12737587105918874,
      "grad_norm": 0.3300305902957916,
      "learning_rate": 6.368650590046178e-06,
      "loss": 0.009,
      "step": 19860
    },
    {
      "epoch": 0.12744000795297483,
      "grad_norm": 0.5568016767501831,
      "learning_rate": 6.371857362750129e-06,
      "loss": 0.0101,
      "step": 19870
    },
    {
      "epoch": 0.12750414484676093,
      "grad_norm": 0.9923796653747559,
      "learning_rate": 6.37506413545408e-06,
      "loss": 0.0124,
      "step": 19880
    },
    {
      "epoch": 0.12756828174054702,
      "grad_norm": 0.5063279867172241,
      "learning_rate": 6.3782709081580304e-06,
      "loss": 0.0097,
      "step": 19890
    },
    {
      "epoch": 0.1276324186343331,
      "grad_norm": 0.6805953979492188,
      "learning_rate": 6.381477680861981e-06,
      "loss": 0.0103,
      "step": 19900
    },
    {
      "epoch": 0.1276965555281192,
      "grad_norm": 0.19922682642936707,
      "learning_rate": 6.384684453565931e-06,
      "loss": 0.0096,
      "step": 19910
    },
    {
      "epoch": 0.12776069242190533,
      "grad_norm": 0.5955575108528137,
      "learning_rate": 6.387891226269882e-06,
      "loss": 0.0098,
      "step": 19920
    },
    {
      "epoch": 0.12782482931569142,
      "grad_norm": 0.872285008430481,
      "learning_rate": 6.391097998973833e-06,
      "loss": 0.0091,
      "step": 19930
    },
    {
      "epoch": 0.1278889662094775,
      "grad_norm": 0.9698531627655029,
      "learning_rate": 6.394304771677784e-06,
      "loss": 0.0094,
      "step": 19940
    },
    {
      "epoch": 0.1279531031032636,
      "grad_norm": 0.3564271330833435,
      "learning_rate": 6.3975115443817346e-06,
      "loss": 0.0066,
      "step": 19950
    },
    {
      "epoch": 0.1280172399970497,
      "grad_norm": 0.6537336707115173,
      "learning_rate": 6.400718317085686e-06,
      "loss": 0.0096,
      "step": 19960
    },
    {
      "epoch": 0.1280813768908358,
      "grad_norm": 0.30640843510627747,
      "learning_rate": 6.403925089789637e-06,
      "loss": 0.0057,
      "step": 19970
    },
    {
      "epoch": 0.12814551378462188,
      "grad_norm": 0.5473002791404724,
      "learning_rate": 6.407131862493587e-06,
      "loss": 0.0079,
      "step": 19980
    },
    {
      "epoch": 0.128209650678408,
      "grad_norm": 0.5697849988937378,
      "learning_rate": 6.410338635197537e-06,
      "loss": 0.0116,
      "step": 19990
    },
    {
      "epoch": 0.1282737875721941,
      "grad_norm": 0.3163318932056427,
      "learning_rate": 6.4135454079014884e-06,
      "loss": 0.0116,
      "step": 20000
    },
    {
      "epoch": 0.1283379244659802,
      "grad_norm": 0.408660352230072,
      "learning_rate": 6.416752180605439e-06,
      "loss": 0.0094,
      "step": 20010
    },
    {
      "epoch": 0.12840206135976628,
      "grad_norm": 0.40231066942214966,
      "learning_rate": 6.41995895330939e-06,
      "loss": 0.0077,
      "step": 20020
    },
    {
      "epoch": 0.12846619825355238,
      "grad_norm": 0.38235190510749817,
      "learning_rate": 6.423165726013341e-06,
      "loss": 0.0059,
      "step": 20030
    },
    {
      "epoch": 0.12853033514733847,
      "grad_norm": 0.6513561010360718,
      "learning_rate": 6.426372498717291e-06,
      "loss": 0.0101,
      "step": 20040
    },
    {
      "epoch": 0.12859447204112456,
      "grad_norm": 0.6265757083892822,
      "learning_rate": 6.429579271421242e-06,
      "loss": 0.0067,
      "step": 20050
    },
    {
      "epoch": 0.12865860893491068,
      "grad_norm": 0.17988623678684235,
      "learning_rate": 6.432786044125193e-06,
      "loss": 0.0079,
      "step": 20060
    },
    {
      "epoch": 0.12872274582869678,
      "grad_norm": 0.347770631313324,
      "learning_rate": 6.4359928168291445e-06,
      "loss": 0.0088,
      "step": 20070
    },
    {
      "epoch": 0.12878688272248287,
      "grad_norm": 0.302829384803772,
      "learning_rate": 6.439199589533094e-06,
      "loss": 0.0099,
      "step": 20080
    },
    {
      "epoch": 0.12885101961626896,
      "grad_norm": 0.717654824256897,
      "learning_rate": 6.442406362237045e-06,
      "loss": 0.0112,
      "step": 20090
    },
    {
      "epoch": 0.12891515651005506,
      "grad_norm": 0.38352274894714355,
      "learning_rate": 6.445613134940995e-06,
      "loss": 0.0151,
      "step": 20100
    },
    {
      "epoch": 0.12897929340384115,
      "grad_norm": 0.6565334796905518,
      "learning_rate": 6.4488199076449464e-06,
      "loss": 0.0129,
      "step": 20110
    },
    {
      "epoch": 0.12904343029762724,
      "grad_norm": 0.38155457377433777,
      "learning_rate": 6.4520266803488976e-06,
      "loss": 0.0085,
      "step": 20120
    },
    {
      "epoch": 0.12910756719141336,
      "grad_norm": 0.37201428413391113,
      "learning_rate": 6.455233453052848e-06,
      "loss": 0.012,
      "step": 20130
    },
    {
      "epoch": 0.12917170408519946,
      "grad_norm": 0.2879607081413269,
      "learning_rate": 6.458440225756799e-06,
      "loss": 0.0079,
      "step": 20140
    },
    {
      "epoch": 0.12923584097898555,
      "grad_norm": 0.5509543418884277,
      "learning_rate": 6.46164699846075e-06,
      "loss": 0.0077,
      "step": 20150
    },
    {
      "epoch": 0.12929997787277164,
      "grad_norm": 0.39315474033355713,
      "learning_rate": 6.464853771164701e-06,
      "loss": 0.0096,
      "step": 20160
    },
    {
      "epoch": 0.12936411476655774,
      "grad_norm": 0.5400700569152832,
      "learning_rate": 6.468060543868651e-06,
      "loss": 0.0067,
      "step": 20170
    },
    {
      "epoch": 0.12942825166034383,
      "grad_norm": 0.5487892627716064,
      "learning_rate": 6.471267316572602e-06,
      "loss": 0.0079,
      "step": 20180
    },
    {
      "epoch": 0.12949238855412992,
      "grad_norm": 0.7011224031448364,
      "learning_rate": 6.474474089276552e-06,
      "loss": 0.0152,
      "step": 20190
    },
    {
      "epoch": 0.12955652544791604,
      "grad_norm": 0.45283573865890503,
      "learning_rate": 6.477680861980503e-06,
      "loss": 0.0059,
      "step": 20200
    },
    {
      "epoch": 0.12962066234170214,
      "grad_norm": 0.8114439845085144,
      "learning_rate": 6.480887634684454e-06,
      "loss": 0.0114,
      "step": 20210
    },
    {
      "epoch": 0.12968479923548823,
      "grad_norm": 0.42993730306625366,
      "learning_rate": 6.484094407388405e-06,
      "loss": 0.0073,
      "step": 20220
    },
    {
      "epoch": 0.12974893612927432,
      "grad_norm": 0.30626970529556274,
      "learning_rate": 6.4873011800923556e-06,
      "loss": 0.0091,
      "step": 20230
    },
    {
      "epoch": 0.12981307302306042,
      "grad_norm": 0.41342639923095703,
      "learning_rate": 6.490507952796307e-06,
      "loss": 0.006,
      "step": 20240
    },
    {
      "epoch": 0.1298772099168465,
      "grad_norm": 0.691035807132721,
      "learning_rate": 6.493714725500258e-06,
      "loss": 0.0187,
      "step": 20250
    },
    {
      "epoch": 0.1299413468106326,
      "grad_norm": 0.6163961887359619,
      "learning_rate": 6.496921498204208e-06,
      "loss": 0.008,
      "step": 20260
    },
    {
      "epoch": 0.13000548370441872,
      "grad_norm": 0.29074427485466003,
      "learning_rate": 6.500128270908158e-06,
      "loss": 0.0076,
      "step": 20270
    },
    {
      "epoch": 0.13006962059820482,
      "grad_norm": 0.38345515727996826,
      "learning_rate": 6.503335043612109e-06,
      "loss": 0.0076,
      "step": 20280
    },
    {
      "epoch": 0.1301337574919909,
      "grad_norm": 0.510400652885437,
      "learning_rate": 6.50654181631606e-06,
      "loss": 0.0078,
      "step": 20290
    },
    {
      "epoch": 0.130197894385777,
      "grad_norm": 0.3353305459022522,
      "learning_rate": 6.509748589020011e-06,
      "loss": 0.0107,
      "step": 20300
    },
    {
      "epoch": 0.1302620312795631,
      "grad_norm": 0.20535027980804443,
      "learning_rate": 6.512955361723962e-06,
      "loss": 0.0066,
      "step": 20310
    },
    {
      "epoch": 0.1303261681733492,
      "grad_norm": 0.6824355721473694,
      "learning_rate": 6.516162134427912e-06,
      "loss": 0.0091,
      "step": 20320
    },
    {
      "epoch": 0.13039030506713528,
      "grad_norm": 0.3136247992515564,
      "learning_rate": 6.519368907131863e-06,
      "loss": 0.0069,
      "step": 20330
    },
    {
      "epoch": 0.1304544419609214,
      "grad_norm": 0.4937147796154022,
      "learning_rate": 6.522575679835814e-06,
      "loss": 0.0086,
      "step": 20340
    },
    {
      "epoch": 0.1305185788547075,
      "grad_norm": 0.3378153443336487,
      "learning_rate": 6.525782452539765e-06,
      "loss": 0.0058,
      "step": 20350
    },
    {
      "epoch": 0.1305827157484936,
      "grad_norm": 0.6026080250740051,
      "learning_rate": 6.528989225243715e-06,
      "loss": 0.0125,
      "step": 20360
    },
    {
      "epoch": 0.13064685264227968,
      "grad_norm": 0.25930720567703247,
      "learning_rate": 6.532195997947666e-06,
      "loss": 0.0068,
      "step": 20370
    },
    {
      "epoch": 0.13071098953606577,
      "grad_norm": 0.46107766032218933,
      "learning_rate": 6.535402770651616e-06,
      "loss": 0.0104,
      "step": 20380
    },
    {
      "epoch": 0.13077512642985187,
      "grad_norm": 0.47708791494369507,
      "learning_rate": 6.5386095433555674e-06,
      "loss": 0.0074,
      "step": 20390
    },
    {
      "epoch": 0.13083926332363796,
      "grad_norm": 0.7677677869796753,
      "learning_rate": 6.5418163160595186e-06,
      "loss": 0.0092,
      "step": 20400
    },
    {
      "epoch": 0.13090340021742408,
      "grad_norm": 0.3615362346172333,
      "learning_rate": 6.545023088763469e-06,
      "loss": 0.0092,
      "step": 20410
    },
    {
      "epoch": 0.13096753711121017,
      "grad_norm": 0.49676114320755005,
      "learning_rate": 6.54822986146742e-06,
      "loss": 0.0054,
      "step": 20420
    },
    {
      "epoch": 0.13103167400499627,
      "grad_norm": 0.47132083773612976,
      "learning_rate": 6.551436634171371e-06,
      "loss": 0.0087,
      "step": 20430
    },
    {
      "epoch": 0.13109581089878236,
      "grad_norm": 0.4946648180484772,
      "learning_rate": 6.554643406875322e-06,
      "loss": 0.01,
      "step": 20440
    },
    {
      "epoch": 0.13115994779256845,
      "grad_norm": 0.41357100009918213,
      "learning_rate": 6.557850179579272e-06,
      "loss": 0.0117,
      "step": 20450
    },
    {
      "epoch": 0.13122408468635455,
      "grad_norm": 0.7563621401786804,
      "learning_rate": 6.561056952283223e-06,
      "loss": 0.0103,
      "step": 20460
    },
    {
      "epoch": 0.13128822158014064,
      "grad_norm": 0.39744338393211365,
      "learning_rate": 6.564263724987173e-06,
      "loss": 0.0081,
      "step": 20470
    },
    {
      "epoch": 0.13135235847392676,
      "grad_norm": 0.43926945328712463,
      "learning_rate": 6.567470497691124e-06,
      "loss": 0.0094,
      "step": 20480
    },
    {
      "epoch": 0.13141649536771285,
      "grad_norm": 0.3090771436691284,
      "learning_rate": 6.570677270395075e-06,
      "loss": 0.0086,
      "step": 20490
    },
    {
      "epoch": 0.13148063226149895,
      "grad_norm": 0.5703118443489075,
      "learning_rate": 6.5738840430990255e-06,
      "loss": 0.0087,
      "step": 20500
    },
    {
      "epoch": 0.13154476915528504,
      "grad_norm": 0.4070364534854889,
      "learning_rate": 6.5770908158029766e-06,
      "loss": 0.0081,
      "step": 20510
    },
    {
      "epoch": 0.13160890604907113,
      "grad_norm": 0.9661247730255127,
      "learning_rate": 6.580297588506928e-06,
      "loss": 0.0106,
      "step": 20520
    },
    {
      "epoch": 0.13167304294285723,
      "grad_norm": 0.43930143117904663,
      "learning_rate": 6.583504361210879e-06,
      "loss": 0.0097,
      "step": 20530
    },
    {
      "epoch": 0.13173717983664332,
      "grad_norm": 0.6114511489868164,
      "learning_rate": 6.586711133914828e-06,
      "loss": 0.0082,
      "step": 20540
    },
    {
      "epoch": 0.13180131673042944,
      "grad_norm": 0.46881040930747986,
      "learning_rate": 6.589917906618779e-06,
      "loss": 0.0083,
      "step": 20550
    },
    {
      "epoch": 0.13186545362421553,
      "grad_norm": 0.6186842322349548,
      "learning_rate": 6.59312467932273e-06,
      "loss": 0.01,
      "step": 20560
    },
    {
      "epoch": 0.13192959051800163,
      "grad_norm": 0.0744633749127388,
      "learning_rate": 6.596331452026681e-06,
      "loss": 0.0078,
      "step": 20570
    },
    {
      "epoch": 0.13199372741178772,
      "grad_norm": 0.3463693857192993,
      "learning_rate": 6.599538224730632e-06,
      "loss": 0.0055,
      "step": 20580
    },
    {
      "epoch": 0.1320578643055738,
      "grad_norm": 0.49897441267967224,
      "learning_rate": 6.602744997434583e-06,
      "loss": 0.0091,
      "step": 20590
    },
    {
      "epoch": 0.1321220011993599,
      "grad_norm": 0.4889982044696808,
      "learning_rate": 6.605951770138533e-06,
      "loss": 0.0094,
      "step": 20600
    },
    {
      "epoch": 0.132186138093146,
      "grad_norm": 0.4408458173274994,
      "learning_rate": 6.609158542842484e-06,
      "loss": 0.0093,
      "step": 20610
    },
    {
      "epoch": 0.13225027498693212,
      "grad_norm": 0.8227345943450928,
      "learning_rate": 6.612365315546435e-06,
      "loss": 0.0129,
      "step": 20620
    },
    {
      "epoch": 0.1323144118807182,
      "grad_norm": 0.33699744939804077,
      "learning_rate": 6.615572088250385e-06,
      "loss": 0.0093,
      "step": 20630
    },
    {
      "epoch": 0.1323785487745043,
      "grad_norm": 0.28413769602775574,
      "learning_rate": 6.618778860954336e-06,
      "loss": 0.0067,
      "step": 20640
    },
    {
      "epoch": 0.1324426856682904,
      "grad_norm": 1.1441147327423096,
      "learning_rate": 6.621985633658286e-06,
      "loss": 0.0107,
      "step": 20650
    },
    {
      "epoch": 0.1325068225620765,
      "grad_norm": 0.26640936732292175,
      "learning_rate": 6.625192406362237e-06,
      "loss": 0.0072,
      "step": 20660
    },
    {
      "epoch": 0.13257095945586259,
      "grad_norm": 0.322395384311676,
      "learning_rate": 6.6283991790661884e-06,
      "loss": 0.007,
      "step": 20670
    },
    {
      "epoch": 0.13263509634964868,
      "grad_norm": 0.727817177772522,
      "learning_rate": 6.6316059517701396e-06,
      "loss": 0.0095,
      "step": 20680
    },
    {
      "epoch": 0.1326992332434348,
      "grad_norm": 0.5444375872612,
      "learning_rate": 6.63481272447409e-06,
      "loss": 0.0076,
      "step": 20690
    },
    {
      "epoch": 0.1327633701372209,
      "grad_norm": 0.5482738018035889,
      "learning_rate": 6.638019497178041e-06,
      "loss": 0.0071,
      "step": 20700
    },
    {
      "epoch": 0.13282750703100699,
      "grad_norm": 0.38756558299064636,
      "learning_rate": 6.641226269881992e-06,
      "loss": 0.0057,
      "step": 20710
    },
    {
      "epoch": 0.13289164392479308,
      "grad_norm": 0.4515531063079834,
      "learning_rate": 6.6444330425859415e-06,
      "loss": 0.0098,
      "step": 20720
    },
    {
      "epoch": 0.13295578081857917,
      "grad_norm": 0.3611808717250824,
      "learning_rate": 6.647639815289893e-06,
      "loss": 0.0066,
      "step": 20730
    },
    {
      "epoch": 0.13301991771236527,
      "grad_norm": 0.49165627360343933,
      "learning_rate": 6.650846587993844e-06,
      "loss": 0.0098,
      "step": 20740
    },
    {
      "epoch": 0.13308405460615136,
      "grad_norm": 0.20099137723445892,
      "learning_rate": 6.654053360697794e-06,
      "loss": 0.0073,
      "step": 20750
    },
    {
      "epoch": 0.13314819149993748,
      "grad_norm": 0.2407764047384262,
      "learning_rate": 6.657260133401745e-06,
      "loss": 0.0099,
      "step": 20760
    },
    {
      "epoch": 0.13321232839372357,
      "grad_norm": 0.21744117140769958,
      "learning_rate": 6.660466906105696e-06,
      "loss": 0.0085,
      "step": 20770
    },
    {
      "epoch": 0.13327646528750967,
      "grad_norm": 0.5023742914199829,
      "learning_rate": 6.6636736788096465e-06,
      "loss": 0.0141,
      "step": 20780
    },
    {
      "epoch": 0.13334060218129576,
      "grad_norm": 0.2848564684391022,
      "learning_rate": 6.6668804515135976e-06,
      "loss": 0.009,
      "step": 20790
    },
    {
      "epoch": 0.13340473907508185,
      "grad_norm": 0.46269291639328003,
      "learning_rate": 6.670087224217549e-06,
      "loss": 0.0105,
      "step": 20800
    },
    {
      "epoch": 0.13346887596886794,
      "grad_norm": 0.2645798623561859,
      "learning_rate": 6.673293996921498e-06,
      "loss": 0.0048,
      "step": 20810
    },
    {
      "epoch": 0.13353301286265404,
      "grad_norm": 0.7819306254386902,
      "learning_rate": 6.676500769625449e-06,
      "loss": 0.0073,
      "step": 20820
    },
    {
      "epoch": 0.13359714975644016,
      "grad_norm": 0.5828709006309509,
      "learning_rate": 6.6797075423294e-06,
      "loss": 0.0105,
      "step": 20830
    },
    {
      "epoch": 0.13366128665022625,
      "grad_norm": 0.18396398425102234,
      "learning_rate": 6.682914315033351e-06,
      "loss": 0.0074,
      "step": 20840
    },
    {
      "epoch": 0.13372542354401235,
      "grad_norm": 0.34209463000297546,
      "learning_rate": 6.686121087737302e-06,
      "loss": 0.0058,
      "step": 20850
    },
    {
      "epoch": 0.13378956043779844,
      "grad_norm": 0.6025264859199524,
      "learning_rate": 6.689327860441253e-06,
      "loss": 0.0063,
      "step": 20860
    },
    {
      "epoch": 0.13385369733158453,
      "grad_norm": 0.7551537156105042,
      "learning_rate": 6.692534633145203e-06,
      "loss": 0.0081,
      "step": 20870
    },
    {
      "epoch": 0.13391783422537062,
      "grad_norm": 0.7358047962188721,
      "learning_rate": 6.695741405849154e-06,
      "loss": 0.0136,
      "step": 20880
    },
    {
      "epoch": 0.13398197111915672,
      "grad_norm": 0.5618805289268494,
      "learning_rate": 6.698948178553105e-06,
      "loss": 0.01,
      "step": 20890
    },
    {
      "epoch": 0.13404610801294284,
      "grad_norm": 0.47844335436820984,
      "learning_rate": 6.702154951257055e-06,
      "loss": 0.0097,
      "step": 20900
    },
    {
      "epoch": 0.13411024490672893,
      "grad_norm": 0.4649708569049835,
      "learning_rate": 6.705361723961006e-06,
      "loss": 0.0122,
      "step": 20910
    },
    {
      "epoch": 0.13417438180051502,
      "grad_norm": 0.2919611632823944,
      "learning_rate": 6.708568496664957e-06,
      "loss": 0.0081,
      "step": 20920
    },
    {
      "epoch": 0.13423851869430112,
      "grad_norm": 0.5163574814796448,
      "learning_rate": 6.711775269368907e-06,
      "loss": 0.0063,
      "step": 20930
    },
    {
      "epoch": 0.1343026555880872,
      "grad_norm": 0.38330158591270447,
      "learning_rate": 6.714982042072858e-06,
      "loss": 0.0084,
      "step": 20940
    },
    {
      "epoch": 0.1343667924818733,
      "grad_norm": 0.3098152279853821,
      "learning_rate": 6.7181888147768094e-06,
      "loss": 0.0108,
      "step": 20950
    },
    {
      "epoch": 0.1344309293756594,
      "grad_norm": 0.5839094519615173,
      "learning_rate": 6.7213955874807606e-06,
      "loss": 0.0096,
      "step": 20960
    },
    {
      "epoch": 0.13449506626944552,
      "grad_norm": 0.226688951253891,
      "learning_rate": 6.724602360184711e-06,
      "loss": 0.0122,
      "step": 20970
    },
    {
      "epoch": 0.1345592031632316,
      "grad_norm": 0.7050507068634033,
      "learning_rate": 6.727809132888662e-06,
      "loss": 0.0094,
      "step": 20980
    },
    {
      "epoch": 0.1346233400570177,
      "grad_norm": 0.34695059061050415,
      "learning_rate": 6.731015905592611e-06,
      "loss": 0.0078,
      "step": 20990
    },
    {
      "epoch": 0.1346874769508038,
      "grad_norm": 0.34674152731895447,
      "learning_rate": 6.7342226782965625e-06,
      "loss": 0.0087,
      "step": 21000
    },
    {
      "epoch": 0.1347516138445899,
      "grad_norm": 0.3843696117401123,
      "learning_rate": 6.737429451000514e-06,
      "loss": 0.0058,
      "step": 21010
    },
    {
      "epoch": 0.13481575073837598,
      "grad_norm": 0.14192050695419312,
      "learning_rate": 6.740636223704464e-06,
      "loss": 0.0113,
      "step": 21020
    },
    {
      "epoch": 0.13487988763216208,
      "grad_norm": 0.35683688521385193,
      "learning_rate": 6.743842996408415e-06,
      "loss": 0.0078,
      "step": 21030
    },
    {
      "epoch": 0.1349440245259482,
      "grad_norm": 0.1909664124250412,
      "learning_rate": 6.747049769112366e-06,
      "loss": 0.0084,
      "step": 21040
    },
    {
      "epoch": 0.1350081614197343,
      "grad_norm": 0.36528199911117554,
      "learning_rate": 6.750256541816317e-06,
      "loss": 0.01,
      "step": 21050
    },
    {
      "epoch": 0.13507229831352038,
      "grad_norm": 0.26217061281204224,
      "learning_rate": 6.7534633145202675e-06,
      "loss": 0.0058,
      "step": 21060
    },
    {
      "epoch": 0.13513643520730648,
      "grad_norm": 0.26856938004493713,
      "learning_rate": 6.7566700872242186e-06,
      "loss": 0.011,
      "step": 21070
    },
    {
      "epoch": 0.13520057210109257,
      "grad_norm": 0.5780160427093506,
      "learning_rate": 6.759876859928168e-06,
      "loss": 0.0097,
      "step": 21080
    },
    {
      "epoch": 0.13526470899487866,
      "grad_norm": 0.6351631283760071,
      "learning_rate": 6.763083632632119e-06,
      "loss": 0.0113,
      "step": 21090
    },
    {
      "epoch": 0.13532884588866476,
      "grad_norm": 0.4127849340438843,
      "learning_rate": 6.76629040533607e-06,
      "loss": 0.0118,
      "step": 21100
    },
    {
      "epoch": 0.13539298278245088,
      "grad_norm": 0.5747127532958984,
      "learning_rate": 6.769497178040021e-06,
      "loss": 0.0082,
      "step": 21110
    },
    {
      "epoch": 0.13545711967623697,
      "grad_norm": 0.36885106563568115,
      "learning_rate": 6.772703950743972e-06,
      "loss": 0.0093,
      "step": 21120
    },
    {
      "epoch": 0.13552125657002306,
      "grad_norm": 0.557569682598114,
      "learning_rate": 6.775910723447923e-06,
      "loss": 0.01,
      "step": 21130
    },
    {
      "epoch": 0.13558539346380916,
      "grad_norm": 0.4298759996891022,
      "learning_rate": 6.779117496151874e-06,
      "loss": 0.0068,
      "step": 21140
    },
    {
      "epoch": 0.13564953035759525,
      "grad_norm": 0.22811418771743774,
      "learning_rate": 6.782324268855824e-06,
      "loss": 0.0066,
      "step": 21150
    },
    {
      "epoch": 0.13571366725138134,
      "grad_norm": 0.5950798392295837,
      "learning_rate": 6.785531041559775e-06,
      "loss": 0.0069,
      "step": 21160
    },
    {
      "epoch": 0.13577780414516744,
      "grad_norm": 0.3545633554458618,
      "learning_rate": 6.788737814263725e-06,
      "loss": 0.0097,
      "step": 21170
    },
    {
      "epoch": 0.13584194103895353,
      "grad_norm": 0.40008631348609924,
      "learning_rate": 6.791944586967676e-06,
      "loss": 0.0114,
      "step": 21180
    },
    {
      "epoch": 0.13590607793273965,
      "grad_norm": 0.7489058971405029,
      "learning_rate": 6.795151359671627e-06,
      "loss": 0.0095,
      "step": 21190
    },
    {
      "epoch": 0.13597021482652574,
      "grad_norm": 0.5277951955795288,
      "learning_rate": 6.798358132375578e-06,
      "loss": 0.0085,
      "step": 21200
    },
    {
      "epoch": 0.13603435172031184,
      "grad_norm": 0.4878092408180237,
      "learning_rate": 6.801564905079528e-06,
      "loss": 0.0094,
      "step": 21210
    },
    {
      "epoch": 0.13609848861409793,
      "grad_norm": 0.5894260406494141,
      "learning_rate": 6.804771677783479e-06,
      "loss": 0.0089,
      "step": 21220
    },
    {
      "epoch": 0.13616262550788402,
      "grad_norm": 0.42668089270591736,
      "learning_rate": 6.8079784504874305e-06,
      "loss": 0.0082,
      "step": 21230
    },
    {
      "epoch": 0.13622676240167012,
      "grad_norm": 0.18119743466377258,
      "learning_rate": 6.811185223191381e-06,
      "loss": 0.0073,
      "step": 21240
    },
    {
      "epoch": 0.1362908992954562,
      "grad_norm": 0.49351224303245544,
      "learning_rate": 6.814391995895332e-06,
      "loss": 0.0093,
      "step": 21250
    },
    {
      "epoch": 0.13635503618924233,
      "grad_norm": 0.4177885353565216,
      "learning_rate": 6.817598768599282e-06,
      "loss": 0.0095,
      "step": 21260
    },
    {
      "epoch": 0.13641917308302842,
      "grad_norm": 0.5884034633636475,
      "learning_rate": 6.820805541303232e-06,
      "loss": 0.0065,
      "step": 21270
    },
    {
      "epoch": 0.13648330997681452,
      "grad_norm": 0.34864699840545654,
      "learning_rate": 6.8240123140071835e-06,
      "loss": 0.0093,
      "step": 21280
    },
    {
      "epoch": 0.1365474468706006,
      "grad_norm": 0.4722350537776947,
      "learning_rate": 6.827219086711135e-06,
      "loss": 0.0114,
      "step": 21290
    },
    {
      "epoch": 0.1366115837643867,
      "grad_norm": 0.29134905338287354,
      "learning_rate": 6.830425859415085e-06,
      "loss": 0.0178,
      "step": 21300
    },
    {
      "epoch": 0.1366757206581728,
      "grad_norm": 0.2316485494375229,
      "learning_rate": 6.833632632119036e-06,
      "loss": 0.0075,
      "step": 21310
    },
    {
      "epoch": 0.1367398575519589,
      "grad_norm": 0.3858979344367981,
      "learning_rate": 6.836839404822987e-06,
      "loss": 0.0066,
      "step": 21320
    },
    {
      "epoch": 0.136803994445745,
      "grad_norm": 0.30195894837379456,
      "learning_rate": 6.840046177526938e-06,
      "loss": 0.0086,
      "step": 21330
    },
    {
      "epoch": 0.1368681313395311,
      "grad_norm": 0.39725568890571594,
      "learning_rate": 6.8432529502308885e-06,
      "loss": 0.0064,
      "step": 21340
    },
    {
      "epoch": 0.1369322682333172,
      "grad_norm": 0.44647717475891113,
      "learning_rate": 6.846459722934839e-06,
      "loss": 0.0074,
      "step": 21350
    },
    {
      "epoch": 0.1369964051271033,
      "grad_norm": 0.19102302193641663,
      "learning_rate": 6.849666495638789e-06,
      "loss": 0.0053,
      "step": 21360
    },
    {
      "epoch": 0.13706054202088938,
      "grad_norm": 0.3509165048599243,
      "learning_rate": 6.85287326834274e-06,
      "loss": 0.0121,
      "step": 21370
    },
    {
      "epoch": 0.13712467891467547,
      "grad_norm": 0.3943997621536255,
      "learning_rate": 6.856080041046691e-06,
      "loss": 0.0073,
      "step": 21380
    },
    {
      "epoch": 0.13718881580846157,
      "grad_norm": 0.29147347807884216,
      "learning_rate": 6.8592868137506415e-06,
      "loss": 0.0088,
      "step": 21390
    },
    {
      "epoch": 0.1372529527022477,
      "grad_norm": 0.4324707090854645,
      "learning_rate": 6.862493586454593e-06,
      "loss": 0.0108,
      "step": 21400
    },
    {
      "epoch": 0.13731708959603378,
      "grad_norm": 0.5054298043251038,
      "learning_rate": 6.865700359158544e-06,
      "loss": 0.0067,
      "step": 21410
    },
    {
      "epoch": 0.13738122648981987,
      "grad_norm": 0.3810000717639923,
      "learning_rate": 6.868907131862495e-06,
      "loss": 0.0118,
      "step": 21420
    },
    {
      "epoch": 0.13744536338360597,
      "grad_norm": 0.6489289402961731,
      "learning_rate": 6.872113904566445e-06,
      "loss": 0.0089,
      "step": 21430
    },
    {
      "epoch": 0.13750950027739206,
      "grad_norm": 0.5280774831771851,
      "learning_rate": 6.875320677270395e-06,
      "loss": 0.0132,
      "step": 21440
    },
    {
      "epoch": 0.13757363717117815,
      "grad_norm": 0.5226761698722839,
      "learning_rate": 6.878527449974346e-06,
      "loss": 0.0094,
      "step": 21450
    },
    {
      "epoch": 0.13763777406496425,
      "grad_norm": 0.3569224178791046,
      "learning_rate": 6.881734222678297e-06,
      "loss": 0.0079,
      "step": 21460
    },
    {
      "epoch": 0.13770191095875037,
      "grad_norm": 0.35034656524658203,
      "learning_rate": 6.884940995382248e-06,
      "loss": 0.009,
      "step": 21470
    },
    {
      "epoch": 0.13776604785253646,
      "grad_norm": 0.4666964113712311,
      "learning_rate": 6.888147768086199e-06,
      "loss": 0.0128,
      "step": 21480
    },
    {
      "epoch": 0.13783018474632255,
      "grad_norm": 0.49302372336387634,
      "learning_rate": 6.891354540790149e-06,
      "loss": 0.0087,
      "step": 21490
    },
    {
      "epoch": 0.13789432164010865,
      "grad_norm": 0.44650712609291077,
      "learning_rate": 6.8945613134941e-06,
      "loss": 0.0075,
      "step": 21500
    },
    {
      "epoch": 0.13795845853389474,
      "grad_norm": 0.26706868410110474,
      "learning_rate": 6.8977680861980515e-06,
      "loss": 0.0086,
      "step": 21510
    },
    {
      "epoch": 0.13802259542768083,
      "grad_norm": 0.6490288972854614,
      "learning_rate": 6.900974858902002e-06,
      "loss": 0.0102,
      "step": 21520
    },
    {
      "epoch": 0.13808673232146693,
      "grad_norm": 0.48663124442100525,
      "learning_rate": 6.904181631605952e-06,
      "loss": 0.0053,
      "step": 21530
    },
    {
      "epoch": 0.13815086921525305,
      "grad_norm": 0.4921961724758148,
      "learning_rate": 6.907388404309902e-06,
      "loss": 0.0074,
      "step": 21540
    },
    {
      "epoch": 0.13821500610903914,
      "grad_norm": 0.48281329870224,
      "learning_rate": 6.910595177013853e-06,
      "loss": 0.0067,
      "step": 21550
    },
    {
      "epoch": 0.13827914300282523,
      "grad_norm": 0.4729806184768677,
      "learning_rate": 6.9138019497178045e-06,
      "loss": 0.0106,
      "step": 21560
    },
    {
      "epoch": 0.13834327989661133,
      "grad_norm": 0.5231598615646362,
      "learning_rate": 6.917008722421756e-06,
      "loss": 0.0107,
      "step": 21570
    },
    {
      "epoch": 0.13840741679039742,
      "grad_norm": 0.2413243055343628,
      "learning_rate": 6.920215495125706e-06,
      "loss": 0.0091,
      "step": 21580
    },
    {
      "epoch": 0.1384715536841835,
      "grad_norm": 0.49566343426704407,
      "learning_rate": 6.923422267829657e-06,
      "loss": 0.0066,
      "step": 21590
    },
    {
      "epoch": 0.1385356905779696,
      "grad_norm": 0.4083350896835327,
      "learning_rate": 6.926629040533608e-06,
      "loss": 0.008,
      "step": 21600
    },
    {
      "epoch": 0.13859982747175573,
      "grad_norm": 0.3106785714626312,
      "learning_rate": 6.929835813237558e-06,
      "loss": 0.0051,
      "step": 21610
    },
    {
      "epoch": 0.13866396436554182,
      "grad_norm": 0.5513443350791931,
      "learning_rate": 6.9330425859415095e-06,
      "loss": 0.0125,
      "step": 21620
    },
    {
      "epoch": 0.1387281012593279,
      "grad_norm": 0.607825756072998,
      "learning_rate": 6.93624935864546e-06,
      "loss": 0.0099,
      "step": 21630
    },
    {
      "epoch": 0.138792238153114,
      "grad_norm": 0.30102694034576416,
      "learning_rate": 6.93945613134941e-06,
      "loss": 0.0077,
      "step": 21640
    },
    {
      "epoch": 0.1388563750469001,
      "grad_norm": 0.3206464350223541,
      "learning_rate": 6.942662904053361e-06,
      "loss": 0.0072,
      "step": 21650
    },
    {
      "epoch": 0.1389205119406862,
      "grad_norm": 0.8913500308990479,
      "learning_rate": 6.945869676757312e-06,
      "loss": 0.0079,
      "step": 21660
    },
    {
      "epoch": 0.13898464883447229,
      "grad_norm": 0.4227841794490814,
      "learning_rate": 6.9490764494612625e-06,
      "loss": 0.01,
      "step": 21670
    },
    {
      "epoch": 0.1390487857282584,
      "grad_norm": 0.4441201388835907,
      "learning_rate": 6.952283222165214e-06,
      "loss": 0.0101,
      "step": 21680
    },
    {
      "epoch": 0.1391129226220445,
      "grad_norm": 0.40509462356567383,
      "learning_rate": 6.955489994869165e-06,
      "loss": 0.009,
      "step": 21690
    },
    {
      "epoch": 0.1391770595158306,
      "grad_norm": 0.35223808884620667,
      "learning_rate": 6.958696767573116e-06,
      "loss": 0.0073,
      "step": 21700
    },
    {
      "epoch": 0.13924119640961669,
      "grad_norm": 0.3851562738418579,
      "learning_rate": 6.961903540277066e-06,
      "loss": 0.0076,
      "step": 21710
    },
    {
      "epoch": 0.13930533330340278,
      "grad_norm": 0.3918079733848572,
      "learning_rate": 6.965110312981016e-06,
      "loss": 0.0078,
      "step": 21720
    },
    {
      "epoch": 0.13936947019718887,
      "grad_norm": 0.5268648862838745,
      "learning_rate": 6.968317085684967e-06,
      "loss": 0.0114,
      "step": 21730
    },
    {
      "epoch": 0.13943360709097496,
      "grad_norm": 0.38811802864074707,
      "learning_rate": 6.971523858388918e-06,
      "loss": 0.0079,
      "step": 21740
    },
    {
      "epoch": 0.13949774398476109,
      "grad_norm": 0.5009214878082275,
      "learning_rate": 6.974730631092869e-06,
      "loss": 0.0123,
      "step": 21750
    },
    {
      "epoch": 0.13956188087854718,
      "grad_norm": 0.36382487416267395,
      "learning_rate": 6.977937403796819e-06,
      "loss": 0.0095,
      "step": 21760
    },
    {
      "epoch": 0.13962601777233327,
      "grad_norm": 0.40182816982269287,
      "learning_rate": 6.98114417650077e-06,
      "loss": 0.0121,
      "step": 21770
    },
    {
      "epoch": 0.13969015466611936,
      "grad_norm": 0.23861971497535706,
      "learning_rate": 6.984350949204721e-06,
      "loss": 0.0086,
      "step": 21780
    },
    {
      "epoch": 0.13975429155990546,
      "grad_norm": 0.4996737837791443,
      "learning_rate": 6.9875577219086725e-06,
      "loss": 0.0066,
      "step": 21790
    },
    {
      "epoch": 0.13981842845369155,
      "grad_norm": 0.5609036087989807,
      "learning_rate": 6.990764494612623e-06,
      "loss": 0.0071,
      "step": 21800
    },
    {
      "epoch": 0.13988256534747764,
      "grad_norm": 0.2084120810031891,
      "learning_rate": 6.993971267316573e-06,
      "loss": 0.0059,
      "step": 21810
    },
    {
      "epoch": 0.13994670224126377,
      "grad_norm": 0.3993605673313141,
      "learning_rate": 6.997178040020523e-06,
      "loss": 0.0087,
      "step": 21820
    },
    {
      "epoch": 0.14001083913504986,
      "grad_norm": 0.315652996301651,
      "learning_rate": 7.000384812724474e-06,
      "loss": 0.0086,
      "step": 21830
    },
    {
      "epoch": 0.14007497602883595,
      "grad_norm": 0.12413597851991653,
      "learning_rate": 7.0035915854284255e-06,
      "loss": 0.0098,
      "step": 21840
    },
    {
      "epoch": 0.14013911292262204,
      "grad_norm": 0.21881459653377533,
      "learning_rate": 7.006798358132377e-06,
      "loss": 0.0084,
      "step": 21850
    },
    {
      "epoch": 0.14020324981640814,
      "grad_norm": 0.4514821469783783,
      "learning_rate": 7.010005130836327e-06,
      "loss": 0.0078,
      "step": 21860
    },
    {
      "epoch": 0.14026738671019423,
      "grad_norm": 0.4085347652435303,
      "learning_rate": 7.013211903540278e-06,
      "loss": 0.0081,
      "step": 21870
    },
    {
      "epoch": 0.14033152360398032,
      "grad_norm": 0.24071069061756134,
      "learning_rate": 7.016418676244229e-06,
      "loss": 0.0111,
      "step": 21880
    },
    {
      "epoch": 0.14039566049776644,
      "grad_norm": 0.30820876359939575,
      "learning_rate": 7.019625448948179e-06,
      "loss": 0.0094,
      "step": 21890
    },
    {
      "epoch": 0.14045979739155254,
      "grad_norm": 0.484391987323761,
      "learning_rate": 7.02283222165213e-06,
      "loss": 0.0085,
      "step": 21900
    },
    {
      "epoch": 0.14052393428533863,
      "grad_norm": 0.5085844993591309,
      "learning_rate": 7.02603899435608e-06,
      "loss": 0.0071,
      "step": 21910
    },
    {
      "epoch": 0.14058807117912472,
      "grad_norm": 0.5190669894218445,
      "learning_rate": 7.029245767060031e-06,
      "loss": 0.0102,
      "step": 21920
    },
    {
      "epoch": 0.14065220807291082,
      "grad_norm": 0.7314053773880005,
      "learning_rate": 7.032452539763982e-06,
      "loss": 0.0097,
      "step": 21930
    },
    {
      "epoch": 0.1407163449666969,
      "grad_norm": 0.15084406733512878,
      "learning_rate": 7.035659312467933e-06,
      "loss": 0.0069,
      "step": 21940
    },
    {
      "epoch": 0.140780481860483,
      "grad_norm": 0.5159353017807007,
      "learning_rate": 7.0388660851718835e-06,
      "loss": 0.0121,
      "step": 21950
    },
    {
      "epoch": 0.14084461875426912,
      "grad_norm": 0.3132694661617279,
      "learning_rate": 7.042072857875835e-06,
      "loss": 0.0112,
      "step": 21960
    },
    {
      "epoch": 0.14090875564805522,
      "grad_norm": 0.20876124501228333,
      "learning_rate": 7.045279630579786e-06,
      "loss": 0.0067,
      "step": 21970
    },
    {
      "epoch": 0.1409728925418413,
      "grad_norm": 0.3610847592353821,
      "learning_rate": 7.048486403283736e-06,
      "loss": 0.0053,
      "step": 21980
    },
    {
      "epoch": 0.1410370294356274,
      "grad_norm": 0.30084025859832764,
      "learning_rate": 7.051693175987686e-06,
      "loss": 0.0084,
      "step": 21990
    },
    {
      "epoch": 0.1411011663294135,
      "grad_norm": 1.0451889038085938,
      "learning_rate": 7.054899948691637e-06,
      "loss": 0.0101,
      "step": 22000
    },
    {
      "epoch": 0.1411653032231996,
      "grad_norm": 0.397225558757782,
      "learning_rate": 7.058106721395588e-06,
      "loss": 0.0059,
      "step": 22010
    },
    {
      "epoch": 0.14122944011698568,
      "grad_norm": 0.37977731227874756,
      "learning_rate": 7.061313494099539e-06,
      "loss": 0.0078,
      "step": 22020
    },
    {
      "epoch": 0.1412935770107718,
      "grad_norm": 0.5413793921470642,
      "learning_rate": 7.06452026680349e-06,
      "loss": 0.0058,
      "step": 22030
    },
    {
      "epoch": 0.1413577139045579,
      "grad_norm": 0.5112346410751343,
      "learning_rate": 7.06772703950744e-06,
      "loss": 0.009,
      "step": 22040
    },
    {
      "epoch": 0.141421850798344,
      "grad_norm": 0.15585067868232727,
      "learning_rate": 7.070933812211391e-06,
      "loss": 0.0078,
      "step": 22050
    },
    {
      "epoch": 0.14148598769213008,
      "grad_norm": 0.6915168166160583,
      "learning_rate": 7.074140584915342e-06,
      "loss": 0.0058,
      "step": 22060
    },
    {
      "epoch": 0.14155012458591618,
      "grad_norm": 0.38445475697517395,
      "learning_rate": 7.0773473576192935e-06,
      "loss": 0.0077,
      "step": 22070
    },
    {
      "epoch": 0.14161426147970227,
      "grad_norm": 0.2983399033546448,
      "learning_rate": 7.080554130323243e-06,
      "loss": 0.0065,
      "step": 22080
    },
    {
      "epoch": 0.14167839837348836,
      "grad_norm": 0.3959824740886688,
      "learning_rate": 7.083760903027194e-06,
      "loss": 0.0085,
      "step": 22090
    },
    {
      "epoch": 0.14174253526727448,
      "grad_norm": 0.2227490246295929,
      "learning_rate": 7.086967675731144e-06,
      "loss": 0.0083,
      "step": 22100
    },
    {
      "epoch": 0.14180667216106058,
      "grad_norm": 0.3485599756240845,
      "learning_rate": 7.090174448435095e-06,
      "loss": 0.0068,
      "step": 22110
    },
    {
      "epoch": 0.14187080905484667,
      "grad_norm": 0.5139448642730713,
      "learning_rate": 7.0933812211390465e-06,
      "loss": 0.0069,
      "step": 22120
    },
    {
      "epoch": 0.14193494594863276,
      "grad_norm": 0.16740815341472626,
      "learning_rate": 7.096587993842997e-06,
      "loss": 0.0063,
      "step": 22130
    },
    {
      "epoch": 0.14199908284241886,
      "grad_norm": 0.5977540612220764,
      "learning_rate": 7.099794766546948e-06,
      "loss": 0.0072,
      "step": 22140
    },
    {
      "epoch": 0.14206321973620495,
      "grad_norm": 0.6218705177307129,
      "learning_rate": 7.103001539250899e-06,
      "loss": 0.0075,
      "step": 22150
    },
    {
      "epoch": 0.14212735662999104,
      "grad_norm": 0.605303168296814,
      "learning_rate": 7.10620831195485e-06,
      "loss": 0.0101,
      "step": 22160
    },
    {
      "epoch": 0.14219149352377716,
      "grad_norm": 0.309969961643219,
      "learning_rate": 7.1094150846587995e-06,
      "loss": 0.0056,
      "step": 22170
    },
    {
      "epoch": 0.14225563041756326,
      "grad_norm": 0.3917008638381958,
      "learning_rate": 7.112621857362751e-06,
      "loss": 0.0096,
      "step": 22180
    },
    {
      "epoch": 0.14231976731134935,
      "grad_norm": 0.2769995331764221,
      "learning_rate": 7.115828630066701e-06,
      "loss": 0.0092,
      "step": 22190
    },
    {
      "epoch": 0.14238390420513544,
      "grad_norm": 0.28504928946495056,
      "learning_rate": 7.119035402770652e-06,
      "loss": 0.0061,
      "step": 22200
    },
    {
      "epoch": 0.14244804109892154,
      "grad_norm": 0.28232064843177795,
      "learning_rate": 7.122242175474603e-06,
      "loss": 0.0061,
      "step": 22210
    },
    {
      "epoch": 0.14251217799270763,
      "grad_norm": 1.2043057680130005,
      "learning_rate": 7.125448948178554e-06,
      "loss": 0.0068,
      "step": 22220
    },
    {
      "epoch": 0.14257631488649372,
      "grad_norm": 0.6536058783531189,
      "learning_rate": 7.1286557208825045e-06,
      "loss": 0.0076,
      "step": 22230
    },
    {
      "epoch": 0.14264045178027984,
      "grad_norm": 0.3707522451877594,
      "learning_rate": 7.131862493586456e-06,
      "loss": 0.0142,
      "step": 22240
    },
    {
      "epoch": 0.14270458867406594,
      "grad_norm": 0.480421245098114,
      "learning_rate": 7.135069266290407e-06,
      "loss": 0.0107,
      "step": 22250
    },
    {
      "epoch": 0.14276872556785203,
      "grad_norm": 0.3505784571170807,
      "learning_rate": 7.138276038994356e-06,
      "loss": 0.0094,
      "step": 22260
    },
    {
      "epoch": 0.14283286246163812,
      "grad_norm": 0.46982210874557495,
      "learning_rate": 7.141482811698307e-06,
      "loss": 0.0128,
      "step": 22270
    },
    {
      "epoch": 0.14289699935542421,
      "grad_norm": 0.2160736620426178,
      "learning_rate": 7.1446895844022575e-06,
      "loss": 0.0078,
      "step": 22280
    },
    {
      "epoch": 0.1429611362492103,
      "grad_norm": 0.08916353434324265,
      "learning_rate": 7.147896357106209e-06,
      "loss": 0.0083,
      "step": 22290
    },
    {
      "epoch": 0.1430252731429964,
      "grad_norm": 0.332522988319397,
      "learning_rate": 7.15110312981016e-06,
      "loss": 0.0082,
      "step": 22300
    },
    {
      "epoch": 0.14308941003678252,
      "grad_norm": 0.7146239876747131,
      "learning_rate": 7.154309902514111e-06,
      "loss": 0.0097,
      "step": 22310
    },
    {
      "epoch": 0.14315354693056861,
      "grad_norm": 0.5682917237281799,
      "learning_rate": 7.157516675218061e-06,
      "loss": 0.0105,
      "step": 22320
    },
    {
      "epoch": 0.1432176838243547,
      "grad_norm": 0.14967156946659088,
      "learning_rate": 7.160723447922012e-06,
      "loss": 0.008,
      "step": 22330
    },
    {
      "epoch": 0.1432818207181408,
      "grad_norm": 0.814206063747406,
      "learning_rate": 7.163930220625963e-06,
      "loss": 0.0064,
      "step": 22340
    },
    {
      "epoch": 0.1433459576119269,
      "grad_norm": 0.36975303292274475,
      "learning_rate": 7.167136993329913e-06,
      "loss": 0.0072,
      "step": 22350
    },
    {
      "epoch": 0.143410094505713,
      "grad_norm": 0.18659432232379913,
      "learning_rate": 7.170343766033864e-06,
      "loss": 0.0071,
      "step": 22360
    },
    {
      "epoch": 0.14347423139949908,
      "grad_norm": 0.4955224096775055,
      "learning_rate": 7.173550538737815e-06,
      "loss": 0.0087,
      "step": 22370
    },
    {
      "epoch": 0.1435383682932852,
      "grad_norm": 0.46361032128334045,
      "learning_rate": 7.176757311441765e-06,
      "loss": 0.0062,
      "step": 22380
    },
    {
      "epoch": 0.1436025051870713,
      "grad_norm": 0.24322505295276642,
      "learning_rate": 7.179964084145716e-06,
      "loss": 0.0076,
      "step": 22390
    },
    {
      "epoch": 0.1436666420808574,
      "grad_norm": 0.8162360787391663,
      "learning_rate": 7.1831708568496675e-06,
      "loss": 0.0098,
      "step": 22400
    },
    {
      "epoch": 0.14373077897464348,
      "grad_norm": 0.4669942259788513,
      "learning_rate": 7.186377629553618e-06,
      "loss": 0.0075,
      "step": 22410
    },
    {
      "epoch": 0.14379491586842957,
      "grad_norm": 0.2904738783836365,
      "learning_rate": 7.189584402257569e-06,
      "loss": 0.0091,
      "step": 22420
    },
    {
      "epoch": 0.14385905276221567,
      "grad_norm": 0.6553630828857422,
      "learning_rate": 7.19279117496152e-06,
      "loss": 0.0089,
      "step": 22430
    },
    {
      "epoch": 0.14392318965600176,
      "grad_norm": 0.31168097257614136,
      "learning_rate": 7.195997947665469e-06,
      "loss": 0.0077,
      "step": 22440
    },
    {
      "epoch": 0.14398732654978788,
      "grad_norm": 0.2955361604690552,
      "learning_rate": 7.1992047203694205e-06,
      "loss": 0.0064,
      "step": 22450
    },
    {
      "epoch": 0.14405146344357397,
      "grad_norm": 0.2961823642253876,
      "learning_rate": 7.202411493073372e-06,
      "loss": 0.0103,
      "step": 22460
    },
    {
      "epoch": 0.14411560033736007,
      "grad_norm": 0.4810275435447693,
      "learning_rate": 7.205618265777322e-06,
      "loss": 0.0077,
      "step": 22470
    },
    {
      "epoch": 0.14417973723114616,
      "grad_norm": 0.3276253342628479,
      "learning_rate": 7.208825038481273e-06,
      "loss": 0.0065,
      "step": 22480
    },
    {
      "epoch": 0.14424387412493225,
      "grad_norm": 0.26311808824539185,
      "learning_rate": 7.212031811185224e-06,
      "loss": 0.007,
      "step": 22490
    },
    {
      "epoch": 0.14430801101871835,
      "grad_norm": 0.3071235716342926,
      "learning_rate": 7.215238583889174e-06,
      "loss": 0.0095,
      "step": 22500
    },
    {
      "epoch": 0.14437214791250444,
      "grad_norm": 0.2706223130226135,
      "learning_rate": 7.2184453565931255e-06,
      "loss": 0.0079,
      "step": 22510
    },
    {
      "epoch": 0.14443628480629056,
      "grad_norm": 0.2890281677246094,
      "learning_rate": 7.221652129297077e-06,
      "loss": 0.0072,
      "step": 22520
    },
    {
      "epoch": 0.14450042170007665,
      "grad_norm": 0.3435359001159668,
      "learning_rate": 7.224858902001026e-06,
      "loss": 0.0065,
      "step": 22530
    },
    {
      "epoch": 0.14456455859386275,
      "grad_norm": 0.368247926235199,
      "learning_rate": 7.228065674704977e-06,
      "loss": 0.0071,
      "step": 22540
    },
    {
      "epoch": 0.14462869548764884,
      "grad_norm": 0.0901537537574768,
      "learning_rate": 7.231272447408928e-06,
      "loss": 0.0084,
      "step": 22550
    },
    {
      "epoch": 0.14469283238143493,
      "grad_norm": 0.47297295928001404,
      "learning_rate": 7.2344792201128785e-06,
      "loss": 0.0072,
      "step": 22560
    },
    {
      "epoch": 0.14475696927522103,
      "grad_norm": 0.3122113049030304,
      "learning_rate": 7.23768599281683e-06,
      "loss": 0.0071,
      "step": 22570
    },
    {
      "epoch": 0.14482110616900712,
      "grad_norm": 0.4448734223842621,
      "learning_rate": 7.240892765520781e-06,
      "loss": 0.006,
      "step": 22580
    },
    {
      "epoch": 0.14488524306279324,
      "grad_norm": 0.47366371750831604,
      "learning_rate": 7.244099538224732e-06,
      "loss": 0.0061,
      "step": 22590
    },
    {
      "epoch": 0.14494937995657933,
      "grad_norm": 0.3275550603866577,
      "learning_rate": 7.247306310928682e-06,
      "loss": 0.0083,
      "step": 22600
    },
    {
      "epoch": 0.14501351685036543,
      "grad_norm": 0.2917988896369934,
      "learning_rate": 7.250513083632633e-06,
      "loss": 0.0073,
      "step": 22610
    },
    {
      "epoch": 0.14507765374415152,
      "grad_norm": 0.5993563532829285,
      "learning_rate": 7.253719856336583e-06,
      "loss": 0.0057,
      "step": 22620
    },
    {
      "epoch": 0.1451417906379376,
      "grad_norm": 0.6318184733390808,
      "learning_rate": 7.256926629040534e-06,
      "loss": 0.0074,
      "step": 22630
    },
    {
      "epoch": 0.1452059275317237,
      "grad_norm": 0.25389841198921204,
      "learning_rate": 7.260133401744485e-06,
      "loss": 0.0064,
      "step": 22640
    },
    {
      "epoch": 0.1452700644255098,
      "grad_norm": 0.4760797619819641,
      "learning_rate": 7.263340174448435e-06,
      "loss": 0.0113,
      "step": 22650
    },
    {
      "epoch": 0.1453342013192959,
      "grad_norm": 0.5239165425300598,
      "learning_rate": 7.266546947152386e-06,
      "loss": 0.0078,
      "step": 22660
    },
    {
      "epoch": 0.145398338213082,
      "grad_norm": 0.3103371262550354,
      "learning_rate": 7.269753719856337e-06,
      "loss": 0.0082,
      "step": 22670
    },
    {
      "epoch": 0.1454624751068681,
      "grad_norm": 0.28552502393722534,
      "learning_rate": 7.2729604925602885e-06,
      "loss": 0.0065,
      "step": 22680
    },
    {
      "epoch": 0.1455266120006542,
      "grad_norm": 0.48143863677978516,
      "learning_rate": 7.276167265264239e-06,
      "loss": 0.0092,
      "step": 22690
    },
    {
      "epoch": 0.1455907488944403,
      "grad_norm": 0.44797706604003906,
      "learning_rate": 7.27937403796819e-06,
      "loss": 0.0094,
      "step": 22700
    },
    {
      "epoch": 0.14565488578822638,
      "grad_norm": 0.6186256408691406,
      "learning_rate": 7.282580810672139e-06,
      "loss": 0.0081,
      "step": 22710
    },
    {
      "epoch": 0.14571902268201248,
      "grad_norm": 0.3706546127796173,
      "learning_rate": 7.28578758337609e-06,
      "loss": 0.0091,
      "step": 22720
    },
    {
      "epoch": 0.14578315957579857,
      "grad_norm": 0.4360228180885315,
      "learning_rate": 7.2889943560800415e-06,
      "loss": 0.0085,
      "step": 22730
    },
    {
      "epoch": 0.1458472964695847,
      "grad_norm": 0.2986323833465576,
      "learning_rate": 7.292201128783993e-06,
      "loss": 0.0071,
      "step": 22740
    },
    {
      "epoch": 0.14591143336337078,
      "grad_norm": 0.35248422622680664,
      "learning_rate": 7.295407901487943e-06,
      "loss": 0.0076,
      "step": 22750
    },
    {
      "epoch": 0.14597557025715688,
      "grad_norm": 0.24654223024845123,
      "learning_rate": 7.298614674191894e-06,
      "loss": 0.008,
      "step": 22760
    },
    {
      "epoch": 0.14603970715094297,
      "grad_norm": 0.09405659139156342,
      "learning_rate": 7.301821446895845e-06,
      "loss": 0.0062,
      "step": 22770
    },
    {
      "epoch": 0.14610384404472906,
      "grad_norm": 0.3926496207714081,
      "learning_rate": 7.305028219599795e-06,
      "loss": 0.0107,
      "step": 22780
    },
    {
      "epoch": 0.14616798093851516,
      "grad_norm": 0.5553023219108582,
      "learning_rate": 7.3082349923037465e-06,
      "loss": 0.0088,
      "step": 22790
    },
    {
      "epoch": 0.14623211783230125,
      "grad_norm": 0.479950875043869,
      "learning_rate": 7.311441765007696e-06,
      "loss": 0.0062,
      "step": 22800
    },
    {
      "epoch": 0.14629625472608737,
      "grad_norm": 0.23034152388572693,
      "learning_rate": 7.314648537711647e-06,
      "loss": 0.0121,
      "step": 22810
    },
    {
      "epoch": 0.14636039161987346,
      "grad_norm": 0.4224400520324707,
      "learning_rate": 7.317855310415598e-06,
      "loss": 0.0103,
      "step": 22820
    },
    {
      "epoch": 0.14642452851365956,
      "grad_norm": 0.31793448328971863,
      "learning_rate": 7.321062083119549e-06,
      "loss": 0.0073,
      "step": 22830
    },
    {
      "epoch": 0.14648866540744565,
      "grad_norm": 0.46968749165534973,
      "learning_rate": 7.3242688558234995e-06,
      "loss": 0.0094,
      "step": 22840
    },
    {
      "epoch": 0.14655280230123174,
      "grad_norm": 0.5646508932113647,
      "learning_rate": 7.327475628527451e-06,
      "loss": 0.0094,
      "step": 22850
    },
    {
      "epoch": 0.14661693919501784,
      "grad_norm": 0.7915888428688049,
      "learning_rate": 7.330682401231402e-06,
      "loss": 0.0061,
      "step": 22860
    },
    {
      "epoch": 0.14668107608880393,
      "grad_norm": 0.410305380821228,
      "learning_rate": 7.333889173935352e-06,
      "loss": 0.0086,
      "step": 22870
    },
    {
      "epoch": 0.14674521298259005,
      "grad_norm": 0.6306430697441101,
      "learning_rate": 7.337095946639303e-06,
      "loss": 0.0085,
      "step": 22880
    },
    {
      "epoch": 0.14680934987637614,
      "grad_norm": 0.27923277020454407,
      "learning_rate": 7.340302719343253e-06,
      "loss": 0.0137,
      "step": 22890
    },
    {
      "epoch": 0.14687348677016224,
      "grad_norm": 0.384808212518692,
      "learning_rate": 7.343509492047204e-06,
      "loss": 0.0051,
      "step": 22900
    },
    {
      "epoch": 0.14693762366394833,
      "grad_norm": 0.5522543787956238,
      "learning_rate": 7.346716264751155e-06,
      "loss": 0.0067,
      "step": 22910
    },
    {
      "epoch": 0.14700176055773442,
      "grad_norm": 0.41765451431274414,
      "learning_rate": 7.349923037455106e-06,
      "loss": 0.0062,
      "step": 22920
    },
    {
      "epoch": 0.14706589745152052,
      "grad_norm": 0.2115785926580429,
      "learning_rate": 7.353129810159056e-06,
      "loss": 0.0093,
      "step": 22930
    },
    {
      "epoch": 0.1471300343453066,
      "grad_norm": 0.46930214762687683,
      "learning_rate": 7.356336582863007e-06,
      "loss": 0.0088,
      "step": 22940
    },
    {
      "epoch": 0.14719417123909273,
      "grad_norm": 0.4150533080101013,
      "learning_rate": 7.359543355566958e-06,
      "loss": 0.0083,
      "step": 22950
    },
    {
      "epoch": 0.14725830813287882,
      "grad_norm": 0.3615683913230896,
      "learning_rate": 7.3627501282709095e-06,
      "loss": 0.0086,
      "step": 22960
    },
    {
      "epoch": 0.14732244502666492,
      "grad_norm": 0.32483333349227905,
      "learning_rate": 7.36595690097486e-06,
      "loss": 0.0054,
      "step": 22970
    },
    {
      "epoch": 0.147386581920451,
      "grad_norm": 0.46817225217819214,
      "learning_rate": 7.369163673678811e-06,
      "loss": 0.0044,
      "step": 22980
    },
    {
      "epoch": 0.1474507188142371,
      "grad_norm": 0.2266695201396942,
      "learning_rate": 7.37237044638276e-06,
      "loss": 0.0061,
      "step": 22990
    },
    {
      "epoch": 0.1475148557080232,
      "grad_norm": 0.293531209230423,
      "learning_rate": 7.375577219086711e-06,
      "loss": 0.01,
      "step": 23000
    },
    {
      "epoch": 0.1475789926018093,
      "grad_norm": 0.6825623512268066,
      "learning_rate": 7.3787839917906625e-06,
      "loss": 0.008,
      "step": 23010
    },
    {
      "epoch": 0.1476431294955954,
      "grad_norm": 0.5003395080566406,
      "learning_rate": 7.381990764494613e-06,
      "loss": 0.0074,
      "step": 23020
    },
    {
      "epoch": 0.1477072663893815,
      "grad_norm": 0.17529790103435516,
      "learning_rate": 7.385197537198564e-06,
      "loss": 0.0083,
      "step": 23030
    },
    {
      "epoch": 0.1477714032831676,
      "grad_norm": 0.32682469487190247,
      "learning_rate": 7.388404309902515e-06,
      "loss": 0.0072,
      "step": 23040
    },
    {
      "epoch": 0.1478355401769537,
      "grad_norm": 0.41206035017967224,
      "learning_rate": 7.391611082606466e-06,
      "loss": 0.0094,
      "step": 23050
    },
    {
      "epoch": 0.14789967707073978,
      "grad_norm": 0.3241424262523651,
      "learning_rate": 7.394817855310416e-06,
      "loss": 0.0148,
      "step": 23060
    },
    {
      "epoch": 0.14796381396452588,
      "grad_norm": 0.7326326966285706,
      "learning_rate": 7.3980246280143675e-06,
      "loss": 0.0098,
      "step": 23070
    },
    {
      "epoch": 0.14802795085831197,
      "grad_norm": 0.5066380500793457,
      "learning_rate": 7.401231400718317e-06,
      "loss": 0.0104,
      "step": 23080
    },
    {
      "epoch": 0.1480920877520981,
      "grad_norm": 0.42016395926475525,
      "learning_rate": 7.404438173422268e-06,
      "loss": 0.007,
      "step": 23090
    },
    {
      "epoch": 0.14815622464588418,
      "grad_norm": 0.346256285905838,
      "learning_rate": 7.407644946126219e-06,
      "loss": 0.01,
      "step": 23100
    },
    {
      "epoch": 0.14822036153967028,
      "grad_norm": 0.2951318323612213,
      "learning_rate": 7.41085171883017e-06,
      "loss": 0.009,
      "step": 23110
    },
    {
      "epoch": 0.14828449843345637,
      "grad_norm": 0.4058953523635864,
      "learning_rate": 7.4140584915341205e-06,
      "loss": 0.01,
      "step": 23120
    },
    {
      "epoch": 0.14834863532724246,
      "grad_norm": 0.482075572013855,
      "learning_rate": 7.417265264238072e-06,
      "loss": 0.0098,
      "step": 23130
    },
    {
      "epoch": 0.14841277222102855,
      "grad_norm": 0.2302228808403015,
      "learning_rate": 7.420472036942023e-06,
      "loss": 0.0083,
      "step": 23140
    },
    {
      "epoch": 0.14847690911481465,
      "grad_norm": 0.46454161405563354,
      "learning_rate": 7.423678809645973e-06,
      "loss": 0.0095,
      "step": 23150
    },
    {
      "epoch": 0.14854104600860077,
      "grad_norm": 0.42806532979011536,
      "learning_rate": 7.426885582349924e-06,
      "loss": 0.0075,
      "step": 23160
    },
    {
      "epoch": 0.14860518290238686,
      "grad_norm": 0.7626487016677856,
      "learning_rate": 7.4300923550538735e-06,
      "loss": 0.0086,
      "step": 23170
    },
    {
      "epoch": 0.14866931979617296,
      "grad_norm": 0.292229026556015,
      "learning_rate": 7.433299127757825e-06,
      "loss": 0.0071,
      "step": 23180
    },
    {
      "epoch": 0.14873345668995905,
      "grad_norm": 0.24265456199645996,
      "learning_rate": 7.436505900461776e-06,
      "loss": 0.0069,
      "step": 23190
    },
    {
      "epoch": 0.14879759358374514,
      "grad_norm": 0.3216264843940735,
      "learning_rate": 7.439712673165727e-06,
      "loss": 0.0053,
      "step": 23200
    },
    {
      "epoch": 0.14886173047753123,
      "grad_norm": 0.4742765724658966,
      "learning_rate": 7.442919445869677e-06,
      "loss": 0.0083,
      "step": 23210
    },
    {
      "epoch": 0.14892586737131733,
      "grad_norm": 0.41154447197914124,
      "learning_rate": 7.446126218573628e-06,
      "loss": 0.0126,
      "step": 23220
    },
    {
      "epoch": 0.14899000426510345,
      "grad_norm": 0.43802884221076965,
      "learning_rate": 7.449332991277579e-06,
      "loss": 0.0052,
      "step": 23230
    },
    {
      "epoch": 0.14905414115888954,
      "grad_norm": 0.34096428751945496,
      "learning_rate": 7.45253976398153e-06,
      "loss": 0.0087,
      "step": 23240
    },
    {
      "epoch": 0.14911827805267563,
      "grad_norm": 0.49796128273010254,
      "learning_rate": 7.455746536685481e-06,
      "loss": 0.0061,
      "step": 23250
    },
    {
      "epoch": 0.14918241494646173,
      "grad_norm": 0.3448033928871155,
      "learning_rate": 7.458953309389431e-06,
      "loss": 0.0092,
      "step": 23260
    },
    {
      "epoch": 0.14924655184024782,
      "grad_norm": 0.4206487238407135,
      "learning_rate": 7.462160082093381e-06,
      "loss": 0.0087,
      "step": 23270
    },
    {
      "epoch": 0.14931068873403391,
      "grad_norm": 0.13565437495708466,
      "learning_rate": 7.465366854797332e-06,
      "loss": 0.0071,
      "step": 23280
    },
    {
      "epoch": 0.14937482562782,
      "grad_norm": 0.47501200437545776,
      "learning_rate": 7.4685736275012835e-06,
      "loss": 0.009,
      "step": 23290
    },
    {
      "epoch": 0.14943896252160613,
      "grad_norm": 1.0063420534133911,
      "learning_rate": 7.471780400205234e-06,
      "loss": 0.0072,
      "step": 23300
    },
    {
      "epoch": 0.14950309941539222,
      "grad_norm": 0.23026001453399658,
      "learning_rate": 7.474987172909185e-06,
      "loss": 0.006,
      "step": 23310
    },
    {
      "epoch": 0.14956723630917831,
      "grad_norm": 0.7847120761871338,
      "learning_rate": 7.478193945613136e-06,
      "loss": 0.0076,
      "step": 23320
    },
    {
      "epoch": 0.1496313732029644,
      "grad_norm": 0.36820703744888306,
      "learning_rate": 7.481400718317086e-06,
      "loss": 0.0139,
      "step": 23330
    },
    {
      "epoch": 0.1496955100967505,
      "grad_norm": 0.21741189062595367,
      "learning_rate": 7.484607491021037e-06,
      "loss": 0.0118,
      "step": 23340
    },
    {
      "epoch": 0.1497596469905366,
      "grad_norm": 0.2632524371147156,
      "learning_rate": 7.487814263724988e-06,
      "loss": 0.006,
      "step": 23350
    },
    {
      "epoch": 0.1498237838843227,
      "grad_norm": 0.5101386904716492,
      "learning_rate": 7.491021036428938e-06,
      "loss": 0.0068,
      "step": 23360
    },
    {
      "epoch": 0.1498879207781088,
      "grad_norm": 0.30875131487846375,
      "learning_rate": 7.494227809132889e-06,
      "loss": 0.0096,
      "step": 23370
    },
    {
      "epoch": 0.1499520576718949,
      "grad_norm": 0.4480379819869995,
      "learning_rate": 7.49743458183684e-06,
      "loss": 0.0082,
      "step": 23380
    },
    {
      "epoch": 0.150016194565681,
      "grad_norm": 0.514462947845459,
      "learning_rate": 7.50064135454079e-06,
      "loss": 0.013,
      "step": 23390
    },
    {
      "epoch": 0.1500803314594671,
      "grad_norm": 0.4939626455307007,
      "learning_rate": 7.5038481272447415e-06,
      "loss": 0.0106,
      "step": 23400
    },
    {
      "epoch": 0.15014446835325318,
      "grad_norm": 0.3024665415287018,
      "learning_rate": 7.507054899948693e-06,
      "loss": 0.0092,
      "step": 23410
    },
    {
      "epoch": 0.15020860524703927,
      "grad_norm": 0.6919910311698914,
      "learning_rate": 7.510261672652644e-06,
      "loss": 0.0087,
      "step": 23420
    },
    {
      "epoch": 0.15027274214082537,
      "grad_norm": 0.09423622488975525,
      "learning_rate": 7.513468445356594e-06,
      "loss": 0.0067,
      "step": 23430
    },
    {
      "epoch": 0.1503368790346115,
      "grad_norm": 0.26807406544685364,
      "learning_rate": 7.516675218060544e-06,
      "loss": 0.0058,
      "step": 23440
    },
    {
      "epoch": 0.15040101592839758,
      "grad_norm": 0.5968019962310791,
      "learning_rate": 7.5198819907644946e-06,
      "loss": 0.006,
      "step": 23450
    },
    {
      "epoch": 0.15046515282218367,
      "grad_norm": 0.42045214772224426,
      "learning_rate": 7.523088763468446e-06,
      "loss": 0.0094,
      "step": 23460
    },
    {
      "epoch": 0.15052928971596977,
      "grad_norm": 0.2310306280851364,
      "learning_rate": 7.526295536172397e-06,
      "loss": 0.0071,
      "step": 23470
    },
    {
      "epoch": 0.15059342660975586,
      "grad_norm": 0.4738796651363373,
      "learning_rate": 7.529502308876348e-06,
      "loss": 0.0077,
      "step": 23480
    },
    {
      "epoch": 0.15065756350354195,
      "grad_norm": 0.37021204829216003,
      "learning_rate": 7.532709081580298e-06,
      "loss": 0.0082,
      "step": 23490
    },
    {
      "epoch": 0.15072170039732805,
      "grad_norm": 1.2935945987701416,
      "learning_rate": 7.535915854284249e-06,
      "loss": 0.0115,
      "step": 23500
    },
    {
      "epoch": 0.15078583729111417,
      "grad_norm": 0.4377444088459015,
      "learning_rate": 7.5391226269882e-06,
      "loss": 0.0108,
      "step": 23510
    },
    {
      "epoch": 0.15084997418490026,
      "grad_norm": 0.24022506177425385,
      "learning_rate": 7.542329399692151e-06,
      "loss": 0.008,
      "step": 23520
    },
    {
      "epoch": 0.15091411107868635,
      "grad_norm": 0.38265594840049744,
      "learning_rate": 7.545536172396101e-06,
      "loss": 0.0082,
      "step": 23530
    },
    {
      "epoch": 0.15097824797247245,
      "grad_norm": 0.14455340802669525,
      "learning_rate": 7.548742945100051e-06,
      "loss": 0.0107,
      "step": 23540
    },
    {
      "epoch": 0.15104238486625854,
      "grad_norm": 0.37466180324554443,
      "learning_rate": 7.551949717804002e-06,
      "loss": 0.009,
      "step": 23550
    },
    {
      "epoch": 0.15110652176004463,
      "grad_norm": 0.5194727778434753,
      "learning_rate": 7.555156490507953e-06,
      "loss": 0.0081,
      "step": 23560
    },
    {
      "epoch": 0.15117065865383073,
      "grad_norm": 0.44248753786087036,
      "learning_rate": 7.5583632632119045e-06,
      "loss": 0.008,
      "step": 23570
    },
    {
      "epoch": 0.15123479554761685,
      "grad_norm": 0.16237616539001465,
      "learning_rate": 7.561570035915855e-06,
      "loss": 0.0048,
      "step": 23580
    },
    {
      "epoch": 0.15129893244140294,
      "grad_norm": 0.2832142114639282,
      "learning_rate": 7.564776808619806e-06,
      "loss": 0.0075,
      "step": 23590
    },
    {
      "epoch": 0.15136306933518903,
      "grad_norm": 0.2390090376138687,
      "learning_rate": 7.567983581323757e-06,
      "loss": 0.0051,
      "step": 23600
    },
    {
      "epoch": 0.15142720622897513,
      "grad_norm": 0.27855297923088074,
      "learning_rate": 7.571190354027707e-06,
      "loss": 0.0072,
      "step": 23610
    },
    {
      "epoch": 0.15149134312276122,
      "grad_norm": 0.9485284686088562,
      "learning_rate": 7.5743971267316575e-06,
      "loss": 0.0068,
      "step": 23620
    },
    {
      "epoch": 0.1515554800165473,
      "grad_norm": 0.44275709986686707,
      "learning_rate": 7.577603899435609e-06,
      "loss": 0.0063,
      "step": 23630
    },
    {
      "epoch": 0.1516196169103334,
      "grad_norm": 0.5310981273651123,
      "learning_rate": 7.580810672139559e-06,
      "loss": 0.0071,
      "step": 23640
    },
    {
      "epoch": 0.15168375380411953,
      "grad_norm": 0.5224945545196533,
      "learning_rate": 7.58401744484351e-06,
      "loss": 0.0101,
      "step": 23650
    },
    {
      "epoch": 0.15174789069790562,
      "grad_norm": 0.202161505818367,
      "learning_rate": 7.587224217547461e-06,
      "loss": 0.0061,
      "step": 23660
    },
    {
      "epoch": 0.1518120275916917,
      "grad_norm": 0.5099035501480103,
      "learning_rate": 7.590430990251411e-06,
      "loss": 0.0075,
      "step": 23670
    },
    {
      "epoch": 0.1518761644854778,
      "grad_norm": 0.6963080763816833,
      "learning_rate": 7.5936377629553625e-06,
      "loss": 0.0073,
      "step": 23680
    },
    {
      "epoch": 0.1519403013792639,
      "grad_norm": 0.5661160349845886,
      "learning_rate": 7.596844535659314e-06,
      "loss": 0.0101,
      "step": 23690
    },
    {
      "epoch": 0.15200443827305,
      "grad_norm": 0.1545230895280838,
      "learning_rate": 7.600051308363264e-06,
      "loss": 0.0094,
      "step": 23700
    },
    {
      "epoch": 0.15206857516683608,
      "grad_norm": 0.38300302624702454,
      "learning_rate": 7.603258081067214e-06,
      "loss": 0.0125,
      "step": 23710
    },
    {
      "epoch": 0.1521327120606222,
      "grad_norm": 0.3603207468986511,
      "learning_rate": 7.606464853771165e-06,
      "loss": 0.0086,
      "step": 23720
    },
    {
      "epoch": 0.1521968489544083,
      "grad_norm": 0.40290364623069763,
      "learning_rate": 7.6096716264751156e-06,
      "loss": 0.0059,
      "step": 23730
    },
    {
      "epoch": 0.1522609858481944,
      "grad_norm": 0.4015500545501709,
      "learning_rate": 7.612878399179067e-06,
      "loss": 0.0065,
      "step": 23740
    },
    {
      "epoch": 0.15232512274198048,
      "grad_norm": 0.19602881371974945,
      "learning_rate": 7.616085171883018e-06,
      "loss": 0.0063,
      "step": 23750
    },
    {
      "epoch": 0.15238925963576658,
      "grad_norm": 0.4438900947570801,
      "learning_rate": 7.619291944586968e-06,
      "loss": 0.0064,
      "step": 23760
    },
    {
      "epoch": 0.15245339652955267,
      "grad_norm": 0.30078378319740295,
      "learning_rate": 7.622498717290919e-06,
      "loss": 0.0067,
      "step": 23770
    },
    {
      "epoch": 0.15251753342333876,
      "grad_norm": 0.13329555094242096,
      "learning_rate": 7.62570548999487e-06,
      "loss": 0.0071,
      "step": 23780
    },
    {
      "epoch": 0.15258167031712488,
      "grad_norm": 0.38784217834472656,
      "learning_rate": 7.628912262698821e-06,
      "loss": 0.0067,
      "step": 23790
    },
    {
      "epoch": 0.15264580721091098,
      "grad_norm": 0.4762949049472809,
      "learning_rate": 7.63211903540277e-06,
      "loss": 0.01,
      "step": 23800
    },
    {
      "epoch": 0.15270994410469707,
      "grad_norm": 0.3756033778190613,
      "learning_rate": 7.635325808106721e-06,
      "loss": 0.0105,
      "step": 23810
    },
    {
      "epoch": 0.15277408099848316,
      "grad_norm": 0.24478082358837128,
      "learning_rate": 7.638532580810672e-06,
      "loss": 0.0071,
      "step": 23820
    },
    {
      "epoch": 0.15283821789226926,
      "grad_norm": 0.3203602433204651,
      "learning_rate": 7.641739353514623e-06,
      "loss": 0.0079,
      "step": 23830
    },
    {
      "epoch": 0.15290235478605535,
      "grad_norm": 0.22945840656757355,
      "learning_rate": 7.644946126218574e-06,
      "loss": 0.0044,
      "step": 23840
    },
    {
      "epoch": 0.15296649167984144,
      "grad_norm": 0.2977195680141449,
      "learning_rate": 7.648152898922526e-06,
      "loss": 0.0056,
      "step": 23850
    },
    {
      "epoch": 0.15303062857362756,
      "grad_norm": 0.38903388381004333,
      "learning_rate": 7.651359671626477e-06,
      "loss": 0.0085,
      "step": 23860
    },
    {
      "epoch": 0.15309476546741366,
      "grad_norm": 0.42640334367752075,
      "learning_rate": 7.654566444330426e-06,
      "loss": 0.0158,
      "step": 23870
    },
    {
      "epoch": 0.15315890236119975,
      "grad_norm": 0.41411834955215454,
      "learning_rate": 7.657773217034377e-06,
      "loss": 0.0062,
      "step": 23880
    },
    {
      "epoch": 0.15322303925498584,
      "grad_norm": 0.6199022531509399,
      "learning_rate": 7.660979989738328e-06,
      "loss": 0.0068,
      "step": 23890
    },
    {
      "epoch": 0.15328717614877194,
      "grad_norm": 0.4416908323764801,
      "learning_rate": 7.664186762442278e-06,
      "loss": 0.008,
      "step": 23900
    },
    {
      "epoch": 0.15335131304255803,
      "grad_norm": 0.4486631751060486,
      "learning_rate": 7.667393535146229e-06,
      "loss": 0.0115,
      "step": 23910
    },
    {
      "epoch": 0.15341544993634412,
      "grad_norm": 0.20878660678863525,
      "learning_rate": 7.67060030785018e-06,
      "loss": 0.0072,
      "step": 23920
    },
    {
      "epoch": 0.15347958683013024,
      "grad_norm": 0.24708037078380585,
      "learning_rate": 7.673807080554131e-06,
      "loss": 0.0076,
      "step": 23930
    },
    {
      "epoch": 0.15354372372391634,
      "grad_norm": 0.48210886120796204,
      "learning_rate": 7.677013853258082e-06,
      "loss": 0.007,
      "step": 23940
    },
    {
      "epoch": 0.15360786061770243,
      "grad_norm": 0.2741550803184509,
      "learning_rate": 7.680220625962033e-06,
      "loss": 0.0128,
      "step": 23950
    },
    {
      "epoch": 0.15367199751148852,
      "grad_norm": 0.39950141310691833,
      "learning_rate": 7.683427398665983e-06,
      "loss": 0.0064,
      "step": 23960
    },
    {
      "epoch": 0.15373613440527462,
      "grad_norm": 0.5735801458358765,
      "learning_rate": 7.686634171369934e-06,
      "loss": 0.0088,
      "step": 23970
    },
    {
      "epoch": 0.1538002712990607,
      "grad_norm": 0.512901246547699,
      "learning_rate": 7.689840944073885e-06,
      "loss": 0.0072,
      "step": 23980
    },
    {
      "epoch": 0.1538644081928468,
      "grad_norm": 0.40003034472465515,
      "learning_rate": 7.693047716777834e-06,
      "loss": 0.0063,
      "step": 23990
    },
    {
      "epoch": 0.15392854508663292,
      "grad_norm": 0.3761098384857178,
      "learning_rate": 7.696254489481785e-06,
      "loss": 0.0082,
      "step": 24000
    },
    {
      "epoch": 0.15399268198041902,
      "grad_norm": 0.18713784217834473,
      "learning_rate": 7.699461262185737e-06,
      "loss": 0.0093,
      "step": 24010
    },
    {
      "epoch": 0.1540568188742051,
      "grad_norm": 0.21950750052928925,
      "learning_rate": 7.702668034889688e-06,
      "loss": 0.007,
      "step": 24020
    },
    {
      "epoch": 0.1541209557679912,
      "grad_norm": 0.04866539686918259,
      "learning_rate": 7.705874807593639e-06,
      "loss": 0.0102,
      "step": 24030
    },
    {
      "epoch": 0.1541850926617773,
      "grad_norm": 0.4602527916431427,
      "learning_rate": 7.70908158029759e-06,
      "loss": 0.0085,
      "step": 24040
    },
    {
      "epoch": 0.1542492295555634,
      "grad_norm": 0.293916791677475,
      "learning_rate": 7.712288353001541e-06,
      "loss": 0.0079,
      "step": 24050
    },
    {
      "epoch": 0.15431336644934948,
      "grad_norm": 0.3953401446342468,
      "learning_rate": 7.71549512570549e-06,
      "loss": 0.0079,
      "step": 24060
    },
    {
      "epoch": 0.15437750334313557,
      "grad_norm": 0.5296419858932495,
      "learning_rate": 7.718701898409442e-06,
      "loss": 0.0112,
      "step": 24070
    },
    {
      "epoch": 0.1544416402369217,
      "grad_norm": 0.3289147615432739,
      "learning_rate": 7.721908671113391e-06,
      "loss": 0.0078,
      "step": 24080
    },
    {
      "epoch": 0.1545057771307078,
      "grad_norm": 0.3659955561161041,
      "learning_rate": 7.725115443817342e-06,
      "loss": 0.009,
      "step": 24090
    },
    {
      "epoch": 0.15456991402449388,
      "grad_norm": 0.6170902848243713,
      "learning_rate": 7.728322216521293e-06,
      "loss": 0.0072,
      "step": 24100
    },
    {
      "epoch": 0.15463405091827997,
      "grad_norm": 0.4635654389858246,
      "learning_rate": 7.731528989225244e-06,
      "loss": 0.0109,
      "step": 24110
    },
    {
      "epoch": 0.15469818781206607,
      "grad_norm": 0.3954100012779236,
      "learning_rate": 7.734735761929195e-06,
      "loss": 0.011,
      "step": 24120
    },
    {
      "epoch": 0.15476232470585216,
      "grad_norm": 0.4503888487815857,
      "learning_rate": 7.737942534633147e-06,
      "loss": 0.0079,
      "step": 24130
    },
    {
      "epoch": 0.15482646159963825,
      "grad_norm": 0.03226611018180847,
      "learning_rate": 7.741149307337098e-06,
      "loss": 0.0079,
      "step": 24140
    },
    {
      "epoch": 0.15489059849342438,
      "grad_norm": 0.17917682230472565,
      "learning_rate": 7.744356080041047e-06,
      "loss": 0.0117,
      "step": 24150
    },
    {
      "epoch": 0.15495473538721047,
      "grad_norm": 0.5344177484512329,
      "learning_rate": 7.747562852744998e-06,
      "loss": 0.0074,
      "step": 24160
    },
    {
      "epoch": 0.15501887228099656,
      "grad_norm": 0.42366012930870056,
      "learning_rate": 7.750769625448948e-06,
      "loss": 0.007,
      "step": 24170
    },
    {
      "epoch": 0.15508300917478265,
      "grad_norm": 0.2631521224975586,
      "learning_rate": 7.753976398152899e-06,
      "loss": 0.0077,
      "step": 24180
    },
    {
      "epoch": 0.15514714606856875,
      "grad_norm": 0.6201837062835693,
      "learning_rate": 7.75718317085685e-06,
      "loss": 0.006,
      "step": 24190
    },
    {
      "epoch": 0.15521128296235484,
      "grad_norm": 0.4369581341743469,
      "learning_rate": 7.760389943560801e-06,
      "loss": 0.0069,
      "step": 24200
    },
    {
      "epoch": 0.15527541985614093,
      "grad_norm": 0.1973450779914856,
      "learning_rate": 7.763596716264752e-06,
      "loss": 0.0084,
      "step": 24210
    },
    {
      "epoch": 0.15533955674992705,
      "grad_norm": 0.3332526981830597,
      "learning_rate": 7.766803488968703e-06,
      "loss": 0.0105,
      "step": 24220
    },
    {
      "epoch": 0.15540369364371315,
      "grad_norm": 0.23551328480243683,
      "learning_rate": 7.770010261672654e-06,
      "loss": 0.008,
      "step": 24230
    },
    {
      "epoch": 0.15546783053749924,
      "grad_norm": 0.3230539560317993,
      "learning_rate": 7.773217034376604e-06,
      "loss": 0.0064,
      "step": 24240
    },
    {
      "epoch": 0.15553196743128533,
      "grad_norm": 0.8207498788833618,
      "learning_rate": 7.776423807080555e-06,
      "loss": 0.0074,
      "step": 24250
    },
    {
      "epoch": 0.15559610432507143,
      "grad_norm": 0.457095742225647,
      "learning_rate": 7.779630579784506e-06,
      "loss": 0.0074,
      "step": 24260
    },
    {
      "epoch": 0.15566024121885752,
      "grad_norm": 0.8549022078514099,
      "learning_rate": 7.782837352488455e-06,
      "loss": 0.0081,
      "step": 24270
    },
    {
      "epoch": 0.1557243781126436,
      "grad_norm": 0.39721837639808655,
      "learning_rate": 7.786044125192406e-06,
      "loss": 0.0062,
      "step": 24280
    },
    {
      "epoch": 0.15578851500642973,
      "grad_norm": 0.2372387945652008,
      "learning_rate": 7.789250897896358e-06,
      "loss": 0.0101,
      "step": 24290
    },
    {
      "epoch": 0.15585265190021583,
      "grad_norm": 0.3453238308429718,
      "learning_rate": 7.792457670600309e-06,
      "loss": 0.0066,
      "step": 24300
    },
    {
      "epoch": 0.15591678879400192,
      "grad_norm": 0.70710289478302,
      "learning_rate": 7.79566444330426e-06,
      "loss": 0.0058,
      "step": 24310
    },
    {
      "epoch": 0.155980925687788,
      "grad_norm": 0.37608903646469116,
      "learning_rate": 7.798871216008211e-06,
      "loss": 0.0066,
      "step": 24320
    },
    {
      "epoch": 0.1560450625815741,
      "grad_norm": 0.3334524929523468,
      "learning_rate": 7.80207798871216e-06,
      "loss": 0.0073,
      "step": 24330
    },
    {
      "epoch": 0.1561091994753602,
      "grad_norm": 0.44738757610321045,
      "learning_rate": 7.805284761416111e-06,
      "loss": 0.0089,
      "step": 24340
    },
    {
      "epoch": 0.1561733363691463,
      "grad_norm": 0.5732530355453491,
      "learning_rate": 7.808491534120063e-06,
      "loss": 0.0092,
      "step": 24350
    },
    {
      "epoch": 0.1562374732629324,
      "grad_norm": 0.3681492805480957,
      "learning_rate": 7.811698306824012e-06,
      "loss": 0.0067,
      "step": 24360
    },
    {
      "epoch": 0.1563016101567185,
      "grad_norm": 0.30083394050598145,
      "learning_rate": 7.814905079527963e-06,
      "loss": 0.0073,
      "step": 24370
    },
    {
      "epoch": 0.1563657470505046,
      "grad_norm": 0.3898124098777771,
      "learning_rate": 7.818111852231914e-06,
      "loss": 0.0068,
      "step": 24380
    },
    {
      "epoch": 0.1564298839442907,
      "grad_norm": 0.4957955777645111,
      "learning_rate": 7.821318624935865e-06,
      "loss": 0.0074,
      "step": 24390
    },
    {
      "epoch": 0.15649402083807679,
      "grad_norm": 0.538674533367157,
      "learning_rate": 7.824525397639816e-06,
      "loss": 0.0076,
      "step": 24400
    },
    {
      "epoch": 0.15655815773186288,
      "grad_norm": 0.510328471660614,
      "learning_rate": 7.827732170343768e-06,
      "loss": 0.0077,
      "step": 24410
    },
    {
      "epoch": 0.15662229462564897,
      "grad_norm": 0.9383547306060791,
      "learning_rate": 7.830938943047719e-06,
      "loss": 0.0086,
      "step": 24420
    },
    {
      "epoch": 0.1566864315194351,
      "grad_norm": 0.37724044919013977,
      "learning_rate": 7.834145715751668e-06,
      "loss": 0.0064,
      "step": 24430
    },
    {
      "epoch": 0.1567505684132212,
      "grad_norm": 0.4641571342945099,
      "learning_rate": 7.83735248845562e-06,
      "loss": 0.0066,
      "step": 24440
    },
    {
      "epoch": 0.15681470530700728,
      "grad_norm": 0.3275459110736847,
      "learning_rate": 7.840559261159569e-06,
      "loss": 0.0075,
      "step": 24450
    },
    {
      "epoch": 0.15687884220079337,
      "grad_norm": 0.2929161489009857,
      "learning_rate": 7.84376603386352e-06,
      "loss": 0.0069,
      "step": 24460
    },
    {
      "epoch": 0.15694297909457947,
      "grad_norm": 0.35929611325263977,
      "learning_rate": 7.84697280656747e-06,
      "loss": 0.0054,
      "step": 24470
    },
    {
      "epoch": 0.15700711598836556,
      "grad_norm": 0.6709334254264832,
      "learning_rate": 7.850179579271422e-06,
      "loss": 0.0088,
      "step": 24480
    },
    {
      "epoch": 0.15707125288215165,
      "grad_norm": 0.2960211932659149,
      "learning_rate": 7.853386351975373e-06,
      "loss": 0.0078,
      "step": 24490
    },
    {
      "epoch": 0.15713538977593777,
      "grad_norm": 0.39208248257637024,
      "learning_rate": 7.856593124679324e-06,
      "loss": 0.0063,
      "step": 24500
    },
    {
      "epoch": 0.15719952666972387,
      "grad_norm": 0.276226282119751,
      "learning_rate": 7.859799897383275e-06,
      "loss": 0.005,
      "step": 24510
    },
    {
      "epoch": 0.15726366356350996,
      "grad_norm": 0.20240812003612518,
      "learning_rate": 7.863006670087225e-06,
      "loss": 0.0146,
      "step": 24520
    },
    {
      "epoch": 0.15732780045729605,
      "grad_norm": 0.5281372666358948,
      "learning_rate": 7.866213442791176e-06,
      "loss": 0.0092,
      "step": 24530
    },
    {
      "epoch": 0.15739193735108215,
      "grad_norm": 0.750211238861084,
      "learning_rate": 7.869420215495125e-06,
      "loss": 0.0057,
      "step": 24540
    },
    {
      "epoch": 0.15745607424486824,
      "grad_norm": 0.4016280174255371,
      "learning_rate": 7.872626988199076e-06,
      "loss": 0.0064,
      "step": 24550
    },
    {
      "epoch": 0.15752021113865433,
      "grad_norm": 0.3035252094268799,
      "learning_rate": 7.875833760903027e-06,
      "loss": 0.0068,
      "step": 24560
    },
    {
      "epoch": 0.15758434803244045,
      "grad_norm": 0.4220190942287445,
      "learning_rate": 7.879040533606979e-06,
      "loss": 0.0075,
      "step": 24570
    },
    {
      "epoch": 0.15764848492622655,
      "grad_norm": 0.29820379614830017,
      "learning_rate": 7.88224730631093e-06,
      "loss": 0.0056,
      "step": 24580
    },
    {
      "epoch": 0.15771262182001264,
      "grad_norm": 0.2260063886642456,
      "learning_rate": 7.88545407901488e-06,
      "loss": 0.0115,
      "step": 24590
    },
    {
      "epoch": 0.15777675871379873,
      "grad_norm": 0.2958219051361084,
      "learning_rate": 7.888660851718832e-06,
      "loss": 0.0064,
      "step": 24600
    },
    {
      "epoch": 0.15784089560758482,
      "grad_norm": 0.29102352261543274,
      "learning_rate": 7.891867624422781e-06,
      "loss": 0.0068,
      "step": 24610
    },
    {
      "epoch": 0.15790503250137092,
      "grad_norm": 0.32563960552215576,
      "learning_rate": 7.895074397126732e-06,
      "loss": 0.0086,
      "step": 24620
    },
    {
      "epoch": 0.157969169395157,
      "grad_norm": 0.38926222920417786,
      "learning_rate": 7.898281169830684e-06,
      "loss": 0.0086,
      "step": 24630
    },
    {
      "epoch": 0.15803330628894313,
      "grad_norm": 0.27023231983184814,
      "learning_rate": 7.901487942534633e-06,
      "loss": 0.0069,
      "step": 24640
    },
    {
      "epoch": 0.15809744318272922,
      "grad_norm": 0.23996005952358246,
      "learning_rate": 7.904694715238584e-06,
      "loss": 0.0075,
      "step": 24650
    },
    {
      "epoch": 0.15816158007651532,
      "grad_norm": 0.423515647649765,
      "learning_rate": 7.907901487942535e-06,
      "loss": 0.0064,
      "step": 24660
    },
    {
      "epoch": 0.1582257169703014,
      "grad_norm": 0.37240225076675415,
      "learning_rate": 7.911108260646486e-06,
      "loss": 0.0076,
      "step": 24670
    },
    {
      "epoch": 0.1582898538640875,
      "grad_norm": 0.41030654311180115,
      "learning_rate": 7.914315033350437e-06,
      "loss": 0.0103,
      "step": 24680
    },
    {
      "epoch": 0.1583539907578736,
      "grad_norm": 0.5382036566734314,
      "learning_rate": 7.917521806054389e-06,
      "loss": 0.0061,
      "step": 24690
    },
    {
      "epoch": 0.1584181276516597,
      "grad_norm": 0.29591771960258484,
      "learning_rate": 7.920728578758338e-06,
      "loss": 0.0078,
      "step": 24700
    },
    {
      "epoch": 0.1584822645454458,
      "grad_norm": 0.580381453037262,
      "learning_rate": 7.923935351462289e-06,
      "loss": 0.0054,
      "step": 24710
    },
    {
      "epoch": 0.1585464014392319,
      "grad_norm": 0.7036392688751221,
      "learning_rate": 7.92714212416624e-06,
      "loss": 0.0122,
      "step": 24720
    },
    {
      "epoch": 0.158610538333018,
      "grad_norm": 0.6105961799621582,
      "learning_rate": 7.93034889687019e-06,
      "loss": 0.0088,
      "step": 24730
    },
    {
      "epoch": 0.1586746752268041,
      "grad_norm": 0.22060944139957428,
      "learning_rate": 7.93355566957414e-06,
      "loss": 0.0033,
      "step": 24740
    },
    {
      "epoch": 0.15873881212059018,
      "grad_norm": 0.35518988966941833,
      "learning_rate": 7.936762442278092e-06,
      "loss": 0.0081,
      "step": 24750
    },
    {
      "epoch": 0.15880294901437628,
      "grad_norm": 0.4817931652069092,
      "learning_rate": 7.939969214982043e-06,
      "loss": 0.0064,
      "step": 24760
    },
    {
      "epoch": 0.15886708590816237,
      "grad_norm": 0.18447251617908478,
      "learning_rate": 7.943175987685994e-06,
      "loss": 0.0098,
      "step": 24770
    },
    {
      "epoch": 0.1589312228019485,
      "grad_norm": 0.6530489325523376,
      "learning_rate": 7.946382760389945e-06,
      "loss": 0.0086,
      "step": 24780
    },
    {
      "epoch": 0.15899535969573458,
      "grad_norm": 0.37190836668014526,
      "learning_rate": 7.949589533093896e-06,
      "loss": 0.0113,
      "step": 24790
    },
    {
      "epoch": 0.15905949658952068,
      "grad_norm": 0.367475688457489,
      "learning_rate": 7.952796305797846e-06,
      "loss": 0.0055,
      "step": 24800
    },
    {
      "epoch": 0.15912363348330677,
      "grad_norm": 0.3782992362976074,
      "learning_rate": 7.956003078501797e-06,
      "loss": 0.0038,
      "step": 24810
    },
    {
      "epoch": 0.15918777037709286,
      "grad_norm": 0.08628425747156143,
      "learning_rate": 7.959209851205746e-06,
      "loss": 0.0073,
      "step": 24820
    },
    {
      "epoch": 0.15925190727087896,
      "grad_norm": 0.3039507269859314,
      "learning_rate": 7.962416623909697e-06,
      "loss": 0.0071,
      "step": 24830
    },
    {
      "epoch": 0.15931604416466505,
      "grad_norm": 0.3761020004749298,
      "learning_rate": 7.965623396613648e-06,
      "loss": 0.0097,
      "step": 24840
    },
    {
      "epoch": 0.15938018105845117,
      "grad_norm": 0.2636759877204895,
      "learning_rate": 7.9688301693176e-06,
      "loss": 0.0072,
      "step": 24850
    },
    {
      "epoch": 0.15944431795223726,
      "grad_norm": 0.16530971229076385,
      "learning_rate": 7.97203694202155e-06,
      "loss": 0.007,
      "step": 24860
    },
    {
      "epoch": 0.15950845484602336,
      "grad_norm": 0.5165784955024719,
      "learning_rate": 7.975243714725502e-06,
      "loss": 0.0104,
      "step": 24870
    },
    {
      "epoch": 0.15957259173980945,
      "grad_norm": 0.22670114040374756,
      "learning_rate": 7.978450487429453e-06,
      "loss": 0.0068,
      "step": 24880
    },
    {
      "epoch": 0.15963672863359554,
      "grad_norm": 0.3174070417881012,
      "learning_rate": 7.981657260133402e-06,
      "loss": 0.0075,
      "step": 24890
    },
    {
      "epoch": 0.15970086552738164,
      "grad_norm": 0.090137779712677,
      "learning_rate": 7.984864032837353e-06,
      "loss": 0.0046,
      "step": 24900
    },
    {
      "epoch": 0.15976500242116773,
      "grad_norm": 0.4895990192890167,
      "learning_rate": 7.988070805541303e-06,
      "loss": 0.0096,
      "step": 24910
    },
    {
      "epoch": 0.15982913931495385,
      "grad_norm": 0.42553743720054626,
      "learning_rate": 7.991277578245254e-06,
      "loss": 0.0072,
      "step": 24920
    },
    {
      "epoch": 0.15989327620873994,
      "grad_norm": 0.570526123046875,
      "learning_rate": 7.994484350949205e-06,
      "loss": 0.0089,
      "step": 24930
    },
    {
      "epoch": 0.15995741310252604,
      "grad_norm": 0.15483412146568298,
      "learning_rate": 7.997691123653156e-06,
      "loss": 0.0051,
      "step": 24940
    },
    {
      "epoch": 0.16002154999631213,
      "grad_norm": 0.2970651686191559,
      "learning_rate": 8.000897896357107e-06,
      "loss": 0.0093,
      "step": 24950
    },
    {
      "epoch": 0.16008568689009822,
      "grad_norm": 0.6350005269050598,
      "learning_rate": 8.004104669061058e-06,
      "loss": 0.0055,
      "step": 24960
    },
    {
      "epoch": 0.16014982378388432,
      "grad_norm": 0.2805918753147125,
      "learning_rate": 8.00731144176501e-06,
      "loss": 0.0089,
      "step": 24970
    },
    {
      "epoch": 0.1602139606776704,
      "grad_norm": 1.4337726831436157,
      "learning_rate": 8.010518214468959e-06,
      "loss": 0.0081,
      "step": 24980
    },
    {
      "epoch": 0.16027809757145653,
      "grad_norm": 0.1360992044210434,
      "learning_rate": 8.01372498717291e-06,
      "loss": 0.0077,
      "step": 24990
    },
    {
      "epoch": 0.16034223446524262,
      "grad_norm": 0.3743704557418823,
      "learning_rate": 8.016931759876861e-06,
      "loss": 0.0069,
      "step": 25000
    },
    {
      "epoch": 0.16040637135902872,
      "grad_norm": 0.25996172428131104,
      "learning_rate": 8.02013853258081e-06,
      "loss": 0.0095,
      "step": 25010
    },
    {
      "epoch": 0.1604705082528148,
      "grad_norm": 0.3653346598148346,
      "learning_rate": 8.023345305284762e-06,
      "loss": 0.0068,
      "step": 25020
    },
    {
      "epoch": 0.1605346451466009,
      "grad_norm": 0.2275839000940323,
      "learning_rate": 8.026552077988713e-06,
      "loss": 0.0095,
      "step": 25030
    },
    {
      "epoch": 0.160598782040387,
      "grad_norm": 0.4239732027053833,
      "learning_rate": 8.029758850692664e-06,
      "loss": 0.0111,
      "step": 25040
    },
    {
      "epoch": 0.1606629189341731,
      "grad_norm": 0.3384075164794922,
      "learning_rate": 8.032965623396615e-06,
      "loss": 0.0076,
      "step": 25050
    },
    {
      "epoch": 0.1607270558279592,
      "grad_norm": 0.2895222306251526,
      "learning_rate": 8.036172396100566e-06,
      "loss": 0.0097,
      "step": 25060
    },
    {
      "epoch": 0.1607911927217453,
      "grad_norm": 0.16481779515743256,
      "learning_rate": 8.039379168804516e-06,
      "loss": 0.0097,
      "step": 25070
    },
    {
      "epoch": 0.1608553296155314,
      "grad_norm": 0.4353906512260437,
      "learning_rate": 8.042585941508467e-06,
      "loss": 0.0131,
      "step": 25080
    },
    {
      "epoch": 0.1609194665093175,
      "grad_norm": 0.4619003236293793,
      "learning_rate": 8.045792714212418e-06,
      "loss": 0.0078,
      "step": 25090
    },
    {
      "epoch": 0.16098360340310358,
      "grad_norm": 0.4785225987434387,
      "learning_rate": 8.048999486916367e-06,
      "loss": 0.0074,
      "step": 25100
    },
    {
      "epoch": 0.16104774029688967,
      "grad_norm": 0.3400071859359741,
      "learning_rate": 8.052206259620318e-06,
      "loss": 0.0099,
      "step": 25110
    },
    {
      "epoch": 0.16111187719067577,
      "grad_norm": 0.212716743350029,
      "learning_rate": 8.05541303232427e-06,
      "loss": 0.0109,
      "step": 25120
    },
    {
      "epoch": 0.1611760140844619,
      "grad_norm": 0.13593250513076782,
      "learning_rate": 8.05861980502822e-06,
      "loss": 0.0071,
      "step": 25130
    },
    {
      "epoch": 0.16124015097824798,
      "grad_norm": 0.3976585865020752,
      "learning_rate": 8.061826577732172e-06,
      "loss": 0.0098,
      "step": 25140
    },
    {
      "epoch": 0.16130428787203407,
      "grad_norm": 0.4405446946620941,
      "learning_rate": 8.065033350436123e-06,
      "loss": 0.0065,
      "step": 25150
    },
    {
      "epoch": 0.16136842476582017,
      "grad_norm": 0.4052540063858032,
      "learning_rate": 8.068240123140072e-06,
      "loss": 0.0068,
      "step": 25160
    },
    {
      "epoch": 0.16143256165960626,
      "grad_norm": 0.48169589042663574,
      "learning_rate": 8.071446895844023e-06,
      "loss": 0.0094,
      "step": 25170
    },
    {
      "epoch": 0.16149669855339235,
      "grad_norm": 0.6314558386802673,
      "learning_rate": 8.074653668547974e-06,
      "loss": 0.0096,
      "step": 25180
    },
    {
      "epoch": 0.16156083544717845,
      "grad_norm": 0.3743741810321808,
      "learning_rate": 8.077860441251924e-06,
      "loss": 0.009,
      "step": 25190
    },
    {
      "epoch": 0.16162497234096457,
      "grad_norm": 0.49911192059516907,
      "learning_rate": 8.081067213955875e-06,
      "loss": 0.0084,
      "step": 25200
    },
    {
      "epoch": 0.16168910923475066,
      "grad_norm": 0.3587149381637573,
      "learning_rate": 8.084273986659826e-06,
      "loss": 0.0078,
      "step": 25210
    },
    {
      "epoch": 0.16175324612853675,
      "grad_norm": 0.38517385721206665,
      "learning_rate": 8.087480759363777e-06,
      "loss": 0.012,
      "step": 25220
    },
    {
      "epoch": 0.16181738302232285,
      "grad_norm": 0.62278813123703,
      "learning_rate": 8.090687532067728e-06,
      "loss": 0.0114,
      "step": 25230
    },
    {
      "epoch": 0.16188151991610894,
      "grad_norm": 0.4665565490722656,
      "learning_rate": 8.09389430477168e-06,
      "loss": 0.0067,
      "step": 25240
    },
    {
      "epoch": 0.16194565680989503,
      "grad_norm": 0.2966174781322479,
      "learning_rate": 8.097101077475629e-06,
      "loss": 0.008,
      "step": 25250
    },
    {
      "epoch": 0.16200979370368113,
      "grad_norm": 0.6743813753128052,
      "learning_rate": 8.10030785017958e-06,
      "loss": 0.007,
      "step": 25260
    },
    {
      "epoch": 0.16207393059746725,
      "grad_norm": 0.5041316151618958,
      "learning_rate": 8.103514622883531e-06,
      "loss": 0.0064,
      "step": 25270
    },
    {
      "epoch": 0.16213806749125334,
      "grad_norm": 0.20831072330474854,
      "learning_rate": 8.10672139558748e-06,
      "loss": 0.0062,
      "step": 25280
    },
    {
      "epoch": 0.16220220438503943,
      "grad_norm": 0.3768036365509033,
      "learning_rate": 8.109928168291432e-06,
      "loss": 0.0061,
      "step": 25290
    },
    {
      "epoch": 0.16226634127882553,
      "grad_norm": 0.32609325647354126,
      "learning_rate": 8.113134940995383e-06,
      "loss": 0.01,
      "step": 25300
    },
    {
      "epoch": 0.16233047817261162,
      "grad_norm": 0.4458755552768707,
      "learning_rate": 8.116341713699334e-06,
      "loss": 0.0069,
      "step": 25310
    },
    {
      "epoch": 0.1623946150663977,
      "grad_norm": 0.30376890301704407,
      "learning_rate": 8.119548486403285e-06,
      "loss": 0.0055,
      "step": 25320
    },
    {
      "epoch": 0.1624587519601838,
      "grad_norm": 0.3464112877845764,
      "learning_rate": 8.122755259107236e-06,
      "loss": 0.0078,
      "step": 25330
    },
    {
      "epoch": 0.16252288885396993,
      "grad_norm": 0.2156023532152176,
      "learning_rate": 8.125962031811185e-06,
      "loss": 0.0102,
      "step": 25340
    },
    {
      "epoch": 0.16258702574775602,
      "grad_norm": 0.761702835559845,
      "learning_rate": 8.129168804515137e-06,
      "loss": 0.0064,
      "step": 25350
    },
    {
      "epoch": 0.1626511626415421,
      "grad_norm": 0.150650754570961,
      "learning_rate": 8.132375577219088e-06,
      "loss": 0.0059,
      "step": 25360
    },
    {
      "epoch": 0.1627152995353282,
      "grad_norm": 0.4692308008670807,
      "learning_rate": 8.135582349923039e-06,
      "loss": 0.01,
      "step": 25370
    },
    {
      "epoch": 0.1627794364291143,
      "grad_norm": 0.4848043918609619,
      "learning_rate": 8.138789122626988e-06,
      "loss": 0.0068,
      "step": 25380
    },
    {
      "epoch": 0.1628435733229004,
      "grad_norm": 0.23428279161453247,
      "learning_rate": 8.14199589533094e-06,
      "loss": 0.0076,
      "step": 25390
    },
    {
      "epoch": 0.16290771021668649,
      "grad_norm": 0.7011100053787231,
      "learning_rate": 8.14520266803489e-06,
      "loss": 0.0074,
      "step": 25400
    },
    {
      "epoch": 0.1629718471104726,
      "grad_norm": 0.2428915947675705,
      "learning_rate": 8.148409440738842e-06,
      "loss": 0.0054,
      "step": 25410
    },
    {
      "epoch": 0.1630359840042587,
      "grad_norm": 0.25832948088645935,
      "learning_rate": 8.151616213442793e-06,
      "loss": 0.008,
      "step": 25420
    },
    {
      "epoch": 0.1631001208980448,
      "grad_norm": 0.5887409448623657,
      "learning_rate": 8.154822986146742e-06,
      "loss": 0.006,
      "step": 25430
    },
    {
      "epoch": 0.16316425779183089,
      "grad_norm": 0.31074729561805725,
      "learning_rate": 8.158029758850693e-06,
      "loss": 0.0048,
      "step": 25440
    },
    {
      "epoch": 0.16322839468561698,
      "grad_norm": 0.21727532148361206,
      "learning_rate": 8.161236531554644e-06,
      "loss": 0.0064,
      "step": 25450
    },
    {
      "epoch": 0.16329253157940307,
      "grad_norm": 0.26646608114242554,
      "learning_rate": 8.164443304258595e-06,
      "loss": 0.0085,
      "step": 25460
    },
    {
      "epoch": 0.16335666847318916,
      "grad_norm": 0.6402791738510132,
      "learning_rate": 8.167650076962545e-06,
      "loss": 0.0065,
      "step": 25470
    },
    {
      "epoch": 0.16342080536697529,
      "grad_norm": 0.2471948117017746,
      "learning_rate": 8.170856849666496e-06,
      "loss": 0.006,
      "step": 25480
    },
    {
      "epoch": 0.16348494226076138,
      "grad_norm": 0.36383992433547974,
      "learning_rate": 8.174063622370447e-06,
      "loss": 0.007,
      "step": 25490
    },
    {
      "epoch": 0.16354907915454747,
      "grad_norm": 0.3720337450504303,
      "learning_rate": 8.177270395074398e-06,
      "loss": 0.0084,
      "step": 25500
    },
    {
      "epoch": 0.16361321604833357,
      "grad_norm": 0.42418551445007324,
      "learning_rate": 8.18047716777835e-06,
      "loss": 0.0088,
      "step": 25510
    },
    {
      "epoch": 0.16367735294211966,
      "grad_norm": 0.35232800245285034,
      "learning_rate": 8.183683940482299e-06,
      "loss": 0.0063,
      "step": 25520
    },
    {
      "epoch": 0.16374148983590575,
      "grad_norm": 0.34052902460098267,
      "learning_rate": 8.18689071318625e-06,
      "loss": 0.0083,
      "step": 25530
    },
    {
      "epoch": 0.16380562672969184,
      "grad_norm": 0.43512094020843506,
      "learning_rate": 8.190097485890201e-06,
      "loss": 0.0104,
      "step": 25540
    },
    {
      "epoch": 0.16386976362347794,
      "grad_norm": 0.548888087272644,
      "learning_rate": 8.193304258594152e-06,
      "loss": 0.0119,
      "step": 25550
    },
    {
      "epoch": 0.16393390051726406,
      "grad_norm": 0.48956605792045593,
      "learning_rate": 8.196511031298101e-06,
      "loss": 0.0064,
      "step": 25560
    },
    {
      "epoch": 0.16399803741105015,
      "grad_norm": 0.20779581367969513,
      "learning_rate": 8.199717804002053e-06,
      "loss": 0.0075,
      "step": 25570
    },
    {
      "epoch": 0.16406217430483624,
      "grad_norm": 0.3951849937438965,
      "learning_rate": 8.202924576706004e-06,
      "loss": 0.0105,
      "step": 25580
    },
    {
      "epoch": 0.16412631119862234,
      "grad_norm": 0.31702741980552673,
      "learning_rate": 8.206131349409955e-06,
      "loss": 0.0081,
      "step": 25590
    },
    {
      "epoch": 0.16419044809240843,
      "grad_norm": 0.2765990197658539,
      "learning_rate": 8.209338122113906e-06,
      "loss": 0.0114,
      "step": 25600
    },
    {
      "epoch": 0.16425458498619452,
      "grad_norm": 0.31762373447418213,
      "learning_rate": 8.212544894817855e-06,
      "loss": 0.0052,
      "step": 25610
    },
    {
      "epoch": 0.16431872187998062,
      "grad_norm": 0.3267388939857483,
      "learning_rate": 8.215751667521806e-06,
      "loss": 0.0101,
      "step": 25620
    },
    {
      "epoch": 0.16438285877376674,
      "grad_norm": 0.5397911071777344,
      "learning_rate": 8.218958440225758e-06,
      "loss": 0.0071,
      "step": 25630
    },
    {
      "epoch": 0.16444699566755283,
      "grad_norm": 0.5041503310203552,
      "learning_rate": 8.222165212929709e-06,
      "loss": 0.0094,
      "step": 25640
    },
    {
      "epoch": 0.16451113256133892,
      "grad_norm": 0.2852059602737427,
      "learning_rate": 8.225371985633658e-06,
      "loss": 0.0059,
      "step": 25650
    },
    {
      "epoch": 0.16457526945512502,
      "grad_norm": 0.3237536549568176,
      "learning_rate": 8.22857875833761e-06,
      "loss": 0.0052,
      "step": 25660
    },
    {
      "epoch": 0.1646394063489111,
      "grad_norm": 0.4522680342197418,
      "learning_rate": 8.23178553104156e-06,
      "loss": 0.0081,
      "step": 25670
    },
    {
      "epoch": 0.1647035432426972,
      "grad_norm": 0.24704231321811676,
      "learning_rate": 8.234992303745511e-06,
      "loss": 0.0079,
      "step": 25680
    },
    {
      "epoch": 0.1647676801364833,
      "grad_norm": 0.5617715120315552,
      "learning_rate": 8.238199076449463e-06,
      "loss": 0.0092,
      "step": 25690
    },
    {
      "epoch": 0.16483181703026942,
      "grad_norm": 0.32703539729118347,
      "learning_rate": 8.241405849153412e-06,
      "loss": 0.0086,
      "step": 25700
    },
    {
      "epoch": 0.1648959539240555,
      "grad_norm": 0.3462812900543213,
      "learning_rate": 8.244612621857363e-06,
      "loss": 0.0074,
      "step": 25710
    },
    {
      "epoch": 0.1649600908178416,
      "grad_norm": 0.24970732629299164,
      "learning_rate": 8.247819394561314e-06,
      "loss": 0.0091,
      "step": 25720
    },
    {
      "epoch": 0.1650242277116277,
      "grad_norm": 0.3338163495063782,
      "learning_rate": 8.251026167265265e-06,
      "loss": 0.0074,
      "step": 25730
    },
    {
      "epoch": 0.1650883646054138,
      "grad_norm": 0.20319034159183502,
      "learning_rate": 8.254232939969216e-06,
      "loss": 0.0088,
      "step": 25740
    },
    {
      "epoch": 0.16515250149919988,
      "grad_norm": 0.4668441414833069,
      "learning_rate": 8.257439712673166e-06,
      "loss": 0.0056,
      "step": 25750
    },
    {
      "epoch": 0.16521663839298598,
      "grad_norm": 0.3017856776714325,
      "learning_rate": 8.260646485377117e-06,
      "loss": 0.0099,
      "step": 25760
    },
    {
      "epoch": 0.1652807752867721,
      "grad_norm": 0.6447399258613586,
      "learning_rate": 8.263853258081068e-06,
      "loss": 0.0089,
      "step": 25770
    },
    {
      "epoch": 0.1653449121805582,
      "grad_norm": 0.5781249403953552,
      "learning_rate": 8.26706003078502e-06,
      "loss": 0.0089,
      "step": 25780
    },
    {
      "epoch": 0.16540904907434428,
      "grad_norm": 0.20946453511714935,
      "learning_rate": 8.27026680348897e-06,
      "loss": 0.0063,
      "step": 25790
    },
    {
      "epoch": 0.16547318596813038,
      "grad_norm": 0.4722101390361786,
      "learning_rate": 8.27347357619292e-06,
      "loss": 0.0086,
      "step": 25800
    },
    {
      "epoch": 0.16553732286191647,
      "grad_norm": 0.284199595451355,
      "learning_rate": 8.27668034889687e-06,
      "loss": 0.0065,
      "step": 25810
    },
    {
      "epoch": 0.16560145975570256,
      "grad_norm": 0.38956043124198914,
      "learning_rate": 8.279887121600822e-06,
      "loss": 0.0087,
      "step": 25820
    },
    {
      "epoch": 0.16566559664948866,
      "grad_norm": 0.32880067825317383,
      "learning_rate": 8.283093894304773e-06,
      "loss": 0.0071,
      "step": 25830
    },
    {
      "epoch": 0.16572973354327478,
      "grad_norm": 0.5591426491737366,
      "learning_rate": 8.286300667008722e-06,
      "loss": 0.0068,
      "step": 25840
    },
    {
      "epoch": 0.16579387043706087,
      "grad_norm": 0.2781529724597931,
      "learning_rate": 8.289507439712674e-06,
      "loss": 0.0054,
      "step": 25850
    },
    {
      "epoch": 0.16585800733084696,
      "grad_norm": 0.6771969795227051,
      "learning_rate": 8.292714212416625e-06,
      "loss": 0.0089,
      "step": 25860
    },
    {
      "epoch": 0.16592214422463306,
      "grad_norm": 0.2917311489582062,
      "learning_rate": 8.295920985120576e-06,
      "loss": 0.0068,
      "step": 25870
    },
    {
      "epoch": 0.16598628111841915,
      "grad_norm": 0.12180343270301819,
      "learning_rate": 8.299127757824527e-06,
      "loss": 0.0088,
      "step": 25880
    },
    {
      "epoch": 0.16605041801220524,
      "grad_norm": 0.2622659206390381,
      "learning_rate": 8.302334530528476e-06,
      "loss": 0.0073,
      "step": 25890
    },
    {
      "epoch": 0.16611455490599134,
      "grad_norm": 0.5130379796028137,
      "learning_rate": 8.305541303232427e-06,
      "loss": 0.0096,
      "step": 25900
    },
    {
      "epoch": 0.16617869179977746,
      "grad_norm": 0.3362933397293091,
      "learning_rate": 8.308748075936379e-06,
      "loss": 0.0059,
      "step": 25910
    },
    {
      "epoch": 0.16624282869356355,
      "grad_norm": 0.4755385220050812,
      "learning_rate": 8.31195484864033e-06,
      "loss": 0.0123,
      "step": 25920
    },
    {
      "epoch": 0.16630696558734964,
      "grad_norm": 0.35895073413848877,
      "learning_rate": 8.315161621344279e-06,
      "loss": 0.0078,
      "step": 25930
    },
    {
      "epoch": 0.16637110248113574,
      "grad_norm": 0.35405483841896057,
      "learning_rate": 8.31836839404823e-06,
      "loss": 0.0075,
      "step": 25940
    },
    {
      "epoch": 0.16643523937492183,
      "grad_norm": 0.3995170295238495,
      "learning_rate": 8.321575166752181e-06,
      "loss": 0.007,
      "step": 25950
    },
    {
      "epoch": 0.16649937626870792,
      "grad_norm": 0.3175564706325531,
      "learning_rate": 8.324781939456132e-06,
      "loss": 0.0061,
      "step": 25960
    },
    {
      "epoch": 0.16656351316249401,
      "grad_norm": 0.4694744944572449,
      "learning_rate": 8.327988712160084e-06,
      "loss": 0.0072,
      "step": 25970
    },
    {
      "epoch": 0.16662765005628014,
      "grad_norm": 0.3355288505554199,
      "learning_rate": 8.331195484864033e-06,
      "loss": 0.0064,
      "step": 25980
    },
    {
      "epoch": 0.16669178695006623,
      "grad_norm": 0.4550934135913849,
      "learning_rate": 8.334402257567984e-06,
      "loss": 0.0066,
      "step": 25990
    },
    {
      "epoch": 0.16675592384385232,
      "grad_norm": 0.47918909788131714,
      "learning_rate": 8.337609030271935e-06,
      "loss": 0.0088,
      "step": 26000
    },
    {
      "epoch": 0.16682006073763841,
      "grad_norm": 0.4235347807407379,
      "learning_rate": 8.340815802975886e-06,
      "loss": 0.0064,
      "step": 26010
    },
    {
      "epoch": 0.1668841976314245,
      "grad_norm": 0.45766931772232056,
      "learning_rate": 8.344022575679836e-06,
      "loss": 0.0066,
      "step": 26020
    },
    {
      "epoch": 0.1669483345252106,
      "grad_norm": 0.5351958274841309,
      "learning_rate": 8.347229348383787e-06,
      "loss": 0.0076,
      "step": 26030
    },
    {
      "epoch": 0.1670124714189967,
      "grad_norm": 0.6322855949401855,
      "learning_rate": 8.350436121087738e-06,
      "loss": 0.0095,
      "step": 26040
    },
    {
      "epoch": 0.16707660831278281,
      "grad_norm": 0.3424733281135559,
      "learning_rate": 8.353642893791689e-06,
      "loss": 0.0071,
      "step": 26050
    },
    {
      "epoch": 0.1671407452065689,
      "grad_norm": 0.13436731696128845,
      "learning_rate": 8.35684966649564e-06,
      "loss": 0.0064,
      "step": 26060
    },
    {
      "epoch": 0.167204882100355,
      "grad_norm": 0.4394909143447876,
      "learning_rate": 8.36005643919959e-06,
      "loss": 0.0061,
      "step": 26070
    },
    {
      "epoch": 0.1672690189941411,
      "grad_norm": 0.7153190970420837,
      "learning_rate": 8.36326321190354e-06,
      "loss": 0.0057,
      "step": 26080
    },
    {
      "epoch": 0.1673331558879272,
      "grad_norm": 0.43987026810646057,
      "learning_rate": 8.366469984607492e-06,
      "loss": 0.0113,
      "step": 26090
    },
    {
      "epoch": 0.16739729278171328,
      "grad_norm": 0.2604074776172638,
      "learning_rate": 8.369676757311443e-06,
      "loss": 0.0078,
      "step": 26100
    },
    {
      "epoch": 0.16746142967549937,
      "grad_norm": 0.41117510199546814,
      "learning_rate": 8.372883530015392e-06,
      "loss": 0.0123,
      "step": 26110
    },
    {
      "epoch": 0.1675255665692855,
      "grad_norm": 0.5456597805023193,
      "learning_rate": 8.376090302719343e-06,
      "loss": 0.006,
      "step": 26120
    },
    {
      "epoch": 0.1675897034630716,
      "grad_norm": 0.32660984992980957,
      "learning_rate": 8.379297075423295e-06,
      "loss": 0.0091,
      "step": 26130
    },
    {
      "epoch": 0.16765384035685768,
      "grad_norm": 0.3434271514415741,
      "learning_rate": 8.382503848127246e-06,
      "loss": 0.0081,
      "step": 26140
    },
    {
      "epoch": 0.16771797725064377,
      "grad_norm": 0.4495420455932617,
      "learning_rate": 8.385710620831197e-06,
      "loss": 0.0149,
      "step": 26150
    },
    {
      "epoch": 0.16778211414442987,
      "grad_norm": 0.14051438868045807,
      "learning_rate": 8.388917393535146e-06,
      "loss": 0.0075,
      "step": 26160
    },
    {
      "epoch": 0.16784625103821596,
      "grad_norm": 0.3010246455669403,
      "learning_rate": 8.392124166239097e-06,
      "loss": 0.0077,
      "step": 26170
    },
    {
      "epoch": 0.16791038793200205,
      "grad_norm": 0.4757477939128876,
      "learning_rate": 8.395330938943048e-06,
      "loss": 0.0075,
      "step": 26180
    },
    {
      "epoch": 0.16797452482578817,
      "grad_norm": 0.40908652544021606,
      "learning_rate": 8.398537711647e-06,
      "loss": 0.0111,
      "step": 26190
    },
    {
      "epoch": 0.16803866171957427,
      "grad_norm": 0.5626053214073181,
      "learning_rate": 8.40174448435095e-06,
      "loss": 0.0057,
      "step": 26200
    },
    {
      "epoch": 0.16810279861336036,
      "grad_norm": 0.5374211668968201,
      "learning_rate": 8.4049512570549e-06,
      "loss": 0.0068,
      "step": 26210
    },
    {
      "epoch": 0.16816693550714645,
      "grad_norm": 0.5747615694999695,
      "learning_rate": 8.408158029758851e-06,
      "loss": 0.0065,
      "step": 26220
    },
    {
      "epoch": 0.16823107240093255,
      "grad_norm": 0.3368001878261566,
      "learning_rate": 8.411364802462802e-06,
      "loss": 0.0073,
      "step": 26230
    },
    {
      "epoch": 0.16829520929471864,
      "grad_norm": 0.29381346702575684,
      "learning_rate": 8.414571575166753e-06,
      "loss": 0.0101,
      "step": 26240
    },
    {
      "epoch": 0.16835934618850473,
      "grad_norm": 0.1899987906217575,
      "learning_rate": 8.417778347870703e-06,
      "loss": 0.006,
      "step": 26250
    },
    {
      "epoch": 0.16842348308229085,
      "grad_norm": 0.29310908913612366,
      "learning_rate": 8.420985120574654e-06,
      "loss": 0.0079,
      "step": 26260
    },
    {
      "epoch": 0.16848761997607695,
      "grad_norm": 0.33029741048812866,
      "learning_rate": 8.424191893278605e-06,
      "loss": 0.0076,
      "step": 26270
    },
    {
      "epoch": 0.16855175686986304,
      "grad_norm": 0.4537966549396515,
      "learning_rate": 8.427398665982556e-06,
      "loss": 0.0033,
      "step": 26280
    },
    {
      "epoch": 0.16861589376364913,
      "grad_norm": 0.1576673835515976,
      "learning_rate": 8.430605438686507e-06,
      "loss": 0.0047,
      "step": 26290
    },
    {
      "epoch": 0.16868003065743523,
      "grad_norm": 0.40747445821762085,
      "learning_rate": 8.433812211390457e-06,
      "loss": 0.0053,
      "step": 26300
    },
    {
      "epoch": 0.16874416755122132,
      "grad_norm": 0.49658462405204773,
      "learning_rate": 8.437018984094408e-06,
      "loss": 0.0113,
      "step": 26310
    },
    {
      "epoch": 0.1688083044450074,
      "grad_norm": 0.4060763418674469,
      "learning_rate": 8.440225756798359e-06,
      "loss": 0.0079,
      "step": 26320
    },
    {
      "epoch": 0.16887244133879353,
      "grad_norm": 0.31417739391326904,
      "learning_rate": 8.44343252950231e-06,
      "loss": 0.0086,
      "step": 26330
    },
    {
      "epoch": 0.16893657823257963,
      "grad_norm": 0.36448004841804504,
      "learning_rate": 8.44663930220626e-06,
      "loss": 0.0073,
      "step": 26340
    },
    {
      "epoch": 0.16900071512636572,
      "grad_norm": 0.3465425968170166,
      "learning_rate": 8.44984607491021e-06,
      "loss": 0.006,
      "step": 26350
    },
    {
      "epoch": 0.1690648520201518,
      "grad_norm": 0.4072333574295044,
      "learning_rate": 8.453052847614162e-06,
      "loss": 0.0061,
      "step": 26360
    },
    {
      "epoch": 0.1691289889139379,
      "grad_norm": 0.2462216317653656,
      "learning_rate": 8.456259620318113e-06,
      "loss": 0.0056,
      "step": 26370
    },
    {
      "epoch": 0.169193125807724,
      "grad_norm": 0.46497079730033875,
      "learning_rate": 8.459466393022064e-06,
      "loss": 0.0062,
      "step": 26380
    },
    {
      "epoch": 0.1692572627015101,
      "grad_norm": 0.33011871576309204,
      "learning_rate": 8.462673165726013e-06,
      "loss": 0.0083,
      "step": 26390
    },
    {
      "epoch": 0.1693213995952962,
      "grad_norm": 0.577763020992279,
      "learning_rate": 8.465879938429964e-06,
      "loss": 0.0064,
      "step": 26400
    },
    {
      "epoch": 0.1693855364890823,
      "grad_norm": 0.34762248396873474,
      "learning_rate": 8.469086711133916e-06,
      "loss": 0.0075,
      "step": 26410
    },
    {
      "epoch": 0.1694496733828684,
      "grad_norm": 0.47082164883613586,
      "learning_rate": 8.472293483837867e-06,
      "loss": 0.0056,
      "step": 26420
    },
    {
      "epoch": 0.1695138102766545,
      "grad_norm": 0.7073124051094055,
      "learning_rate": 8.475500256541816e-06,
      "loss": 0.0085,
      "step": 26430
    },
    {
      "epoch": 0.16957794717044058,
      "grad_norm": 0.5106498003005981,
      "learning_rate": 8.478707029245767e-06,
      "loss": 0.0131,
      "step": 26440
    },
    {
      "epoch": 0.16964208406422668,
      "grad_norm": 0.6935898661613464,
      "learning_rate": 8.481913801949718e-06,
      "loss": 0.0064,
      "step": 26450
    },
    {
      "epoch": 0.16970622095801277,
      "grad_norm": 0.45264732837677,
      "learning_rate": 8.48512057465367e-06,
      "loss": 0.0056,
      "step": 26460
    },
    {
      "epoch": 0.1697703578517989,
      "grad_norm": 0.38825467228889465,
      "learning_rate": 8.48832734735762e-06,
      "loss": 0.0053,
      "step": 26470
    },
    {
      "epoch": 0.16983449474558499,
      "grad_norm": 0.7266507744789124,
      "learning_rate": 8.49153412006157e-06,
      "loss": 0.0095,
      "step": 26480
    },
    {
      "epoch": 0.16989863163937108,
      "grad_norm": 0.07236064970493317,
      "learning_rate": 8.494740892765521e-06,
      "loss": 0.0061,
      "step": 26490
    },
    {
      "epoch": 0.16996276853315717,
      "grad_norm": 0.1668139547109604,
      "learning_rate": 8.497947665469472e-06,
      "loss": 0.0064,
      "step": 26500
    },
    {
      "epoch": 0.17002690542694326,
      "grad_norm": 0.33534303307533264,
      "learning_rate": 8.501154438173423e-06,
      "loss": 0.0069,
      "step": 26510
    },
    {
      "epoch": 0.17009104232072936,
      "grad_norm": 0.2349112182855606,
      "learning_rate": 8.504361210877373e-06,
      "loss": 0.008,
      "step": 26520
    },
    {
      "epoch": 0.17015517921451545,
      "grad_norm": 0.4541345536708832,
      "learning_rate": 8.507567983581324e-06,
      "loss": 0.0078,
      "step": 26530
    },
    {
      "epoch": 0.17021931610830157,
      "grad_norm": 0.39967626333236694,
      "learning_rate": 8.510774756285275e-06,
      "loss": 0.0063,
      "step": 26540
    },
    {
      "epoch": 0.17028345300208766,
      "grad_norm": 0.7688847780227661,
      "learning_rate": 8.513981528989226e-06,
      "loss": 0.0085,
      "step": 26550
    },
    {
      "epoch": 0.17034758989587376,
      "grad_norm": 0.3346398174762726,
      "learning_rate": 8.517188301693177e-06,
      "loss": 0.006,
      "step": 26560
    },
    {
      "epoch": 0.17041172678965985,
      "grad_norm": 0.6446905732154846,
      "learning_rate": 8.520395074397128e-06,
      "loss": 0.009,
      "step": 26570
    },
    {
      "epoch": 0.17047586368344594,
      "grad_norm": 0.38544151186943054,
      "learning_rate": 8.523601847101078e-06,
      "loss": 0.0089,
      "step": 26580
    },
    {
      "epoch": 0.17054000057723204,
      "grad_norm": 0.24830207228660583,
      "learning_rate": 8.526808619805029e-06,
      "loss": 0.0113,
      "step": 26590
    },
    {
      "epoch": 0.17060413747101813,
      "grad_norm": 0.40618616342544556,
      "learning_rate": 8.53001539250898e-06,
      "loss": 0.0049,
      "step": 26600
    },
    {
      "epoch": 0.17066827436480425,
      "grad_norm": 0.2531229555606842,
      "learning_rate": 8.53322216521293e-06,
      "loss": 0.0065,
      "step": 26610
    },
    {
      "epoch": 0.17073241125859034,
      "grad_norm": 0.17107687890529633,
      "learning_rate": 8.53642893791688e-06,
      "loss": 0.0055,
      "step": 26620
    },
    {
      "epoch": 0.17079654815237644,
      "grad_norm": 0.674441933631897,
      "learning_rate": 8.539635710620832e-06,
      "loss": 0.0084,
      "step": 26630
    },
    {
      "epoch": 0.17086068504616253,
      "grad_norm": 0.5298900604248047,
      "learning_rate": 8.542842483324783e-06,
      "loss": 0.0073,
      "step": 26640
    },
    {
      "epoch": 0.17092482193994862,
      "grad_norm": 0.13486960530281067,
      "learning_rate": 8.546049256028734e-06,
      "loss": 0.0063,
      "step": 26650
    },
    {
      "epoch": 0.17098895883373472,
      "grad_norm": 0.2705175280570984,
      "learning_rate": 8.549256028732685e-06,
      "loss": 0.0089,
      "step": 26660
    },
    {
      "epoch": 0.1710530957275208,
      "grad_norm": 0.34008297324180603,
      "learning_rate": 8.552462801436634e-06,
      "loss": 0.0105,
      "step": 26670
    },
    {
      "epoch": 0.17111723262130693,
      "grad_norm": 0.20941926538944244,
      "learning_rate": 8.555669574140585e-06,
      "loss": 0.0046,
      "step": 26680
    },
    {
      "epoch": 0.17118136951509302,
      "grad_norm": 0.3021007478237152,
      "learning_rate": 8.558876346844537e-06,
      "loss": 0.007,
      "step": 26690
    },
    {
      "epoch": 0.17124550640887912,
      "grad_norm": 0.5261565446853638,
      "learning_rate": 8.562083119548486e-06,
      "loss": 0.0107,
      "step": 26700
    },
    {
      "epoch": 0.1713096433026652,
      "grad_norm": 0.34425294399261475,
      "learning_rate": 8.565289892252437e-06,
      "loss": 0.0067,
      "step": 26710
    },
    {
      "epoch": 0.1713737801964513,
      "grad_norm": 0.09715772420167923,
      "learning_rate": 8.568496664956388e-06,
      "loss": 0.005,
      "step": 26720
    },
    {
      "epoch": 0.1714379170902374,
      "grad_norm": 0.4063803553581238,
      "learning_rate": 8.57170343766034e-06,
      "loss": 0.0101,
      "step": 26730
    },
    {
      "epoch": 0.1715020539840235,
      "grad_norm": 0.21971246600151062,
      "learning_rate": 8.57491021036429e-06,
      "loss": 0.0075,
      "step": 26740
    },
    {
      "epoch": 0.1715661908778096,
      "grad_norm": 0.1712648570537567,
      "learning_rate": 8.578116983068242e-06,
      "loss": 0.005,
      "step": 26750
    },
    {
      "epoch": 0.1716303277715957,
      "grad_norm": 0.19962908327579498,
      "learning_rate": 8.581323755772191e-06,
      "loss": 0.012,
      "step": 26760
    },
    {
      "epoch": 0.1716944646653818,
      "grad_norm": 0.3044357895851135,
      "learning_rate": 8.584530528476142e-06,
      "loss": 0.0063,
      "step": 26770
    },
    {
      "epoch": 0.1717586015591679,
      "grad_norm": 0.3212033212184906,
      "learning_rate": 8.587737301180093e-06,
      "loss": 0.0094,
      "step": 26780
    },
    {
      "epoch": 0.17182273845295398,
      "grad_norm": 0.10855749994516373,
      "learning_rate": 8.590944073884043e-06,
      "loss": 0.0102,
      "step": 26790
    },
    {
      "epoch": 0.17188687534674008,
      "grad_norm": 0.18229839205741882,
      "learning_rate": 8.594150846587994e-06,
      "loss": 0.0063,
      "step": 26800
    },
    {
      "epoch": 0.17195101224052617,
      "grad_norm": 0.30547040700912476,
      "learning_rate": 8.597357619291945e-06,
      "loss": 0.0096,
      "step": 26810
    },
    {
      "epoch": 0.1720151491343123,
      "grad_norm": 0.5064559578895569,
      "learning_rate": 8.600564391995896e-06,
      "loss": 0.0121,
      "step": 26820
    },
    {
      "epoch": 0.17207928602809838,
      "grad_norm": 0.4874636232852936,
      "learning_rate": 8.603771164699847e-06,
      "loss": 0.0087,
      "step": 26830
    },
    {
      "epoch": 0.17214342292188448,
      "grad_norm": 0.4293060600757599,
      "learning_rate": 8.606977937403798e-06,
      "loss": 0.0082,
      "step": 26840
    },
    {
      "epoch": 0.17220755981567057,
      "grad_norm": 0.4626399874687195,
      "learning_rate": 8.610184710107748e-06,
      "loss": 0.0098,
      "step": 26850
    },
    {
      "epoch": 0.17227169670945666,
      "grad_norm": 0.34719598293304443,
      "learning_rate": 8.613391482811699e-06,
      "loss": 0.0091,
      "step": 26860
    },
    {
      "epoch": 0.17233583360324276,
      "grad_norm": 0.45511436462402344,
      "learning_rate": 8.61659825551565e-06,
      "loss": 0.0083,
      "step": 26870
    },
    {
      "epoch": 0.17239997049702885,
      "grad_norm": 0.3324184715747833,
      "learning_rate": 8.6198050282196e-06,
      "loss": 0.0076,
      "step": 26880
    },
    {
      "epoch": 0.17246410739081497,
      "grad_norm": 0.42451098561286926,
      "learning_rate": 8.62301180092355e-06,
      "loss": 0.0068,
      "step": 26890
    },
    {
      "epoch": 0.17252824428460106,
      "grad_norm": 0.501450777053833,
      "learning_rate": 8.626218573627502e-06,
      "loss": 0.0095,
      "step": 26900
    },
    {
      "epoch": 0.17259238117838716,
      "grad_norm": 0.2772112488746643,
      "learning_rate": 8.629425346331453e-06,
      "loss": 0.0077,
      "step": 26910
    },
    {
      "epoch": 0.17265651807217325,
      "grad_norm": 0.24453534185886383,
      "learning_rate": 8.632632119035404e-06,
      "loss": 0.0059,
      "step": 26920
    },
    {
      "epoch": 0.17272065496595934,
      "grad_norm": 0.26350727677345276,
      "learning_rate": 8.635838891739355e-06,
      "loss": 0.006,
      "step": 26930
    },
    {
      "epoch": 0.17278479185974543,
      "grad_norm": 0.20161627233028412,
      "learning_rate": 8.639045664443306e-06,
      "loss": 0.0054,
      "step": 26940
    },
    {
      "epoch": 0.17284892875353153,
      "grad_norm": 0.25609177350997925,
      "learning_rate": 8.642252437147255e-06,
      "loss": 0.0062,
      "step": 26950
    },
    {
      "epoch": 0.17291306564731765,
      "grad_norm": 0.43955254554748535,
      "learning_rate": 8.645459209851206e-06,
      "loss": 0.0081,
      "step": 26960
    },
    {
      "epoch": 0.17297720254110374,
      "grad_norm": 0.4165267050266266,
      "learning_rate": 8.648665982555156e-06,
      "loss": 0.008,
      "step": 26970
    },
    {
      "epoch": 0.17304133943488983,
      "grad_norm": 0.3899036645889282,
      "learning_rate": 8.651872755259107e-06,
      "loss": 0.0068,
      "step": 26980
    },
    {
      "epoch": 0.17310547632867593,
      "grad_norm": 0.3477489650249481,
      "learning_rate": 8.655079527963058e-06,
      "loss": 0.0043,
      "step": 26990
    },
    {
      "epoch": 0.17316961322246202,
      "grad_norm": 0.6160642504692078,
      "learning_rate": 8.65828630066701e-06,
      "loss": 0.005,
      "step": 27000
    },
    {
      "epoch": 0.17323375011624811,
      "grad_norm": 0.20395542681217194,
      "learning_rate": 8.66149307337096e-06,
      "loss": 0.005,
      "step": 27010
    },
    {
      "epoch": 0.1732978870100342,
      "grad_norm": 0.5390647649765015,
      "learning_rate": 8.664699846074911e-06,
      "loss": 0.0074,
      "step": 27020
    },
    {
      "epoch": 0.1733620239038203,
      "grad_norm": 0.2438899427652359,
      "learning_rate": 8.667906618778863e-06,
      "loss": 0.0086,
      "step": 27030
    },
    {
      "epoch": 0.17342616079760642,
      "grad_norm": 0.3125239908695221,
      "learning_rate": 8.671113391482812e-06,
      "loss": 0.0121,
      "step": 27040
    },
    {
      "epoch": 0.17349029769139251,
      "grad_norm": 0.41965559124946594,
      "learning_rate": 8.674320164186763e-06,
      "loss": 0.0091,
      "step": 27050
    },
    {
      "epoch": 0.1735544345851786,
      "grad_norm": 0.7748135924339294,
      "learning_rate": 8.677526936890713e-06,
      "loss": 0.008,
      "step": 27060
    },
    {
      "epoch": 0.1736185714789647,
      "grad_norm": 0.29431599378585815,
      "learning_rate": 8.680733709594664e-06,
      "loss": 0.0108,
      "step": 27070
    },
    {
      "epoch": 0.1736827083727508,
      "grad_norm": 0.1197415143251419,
      "learning_rate": 8.683940482298615e-06,
      "loss": 0.0047,
      "step": 27080
    },
    {
      "epoch": 0.1737468452665369,
      "grad_norm": 0.40981724858283997,
      "learning_rate": 8.687147255002566e-06,
      "loss": 0.0081,
      "step": 27090
    },
    {
      "epoch": 0.17381098216032298,
      "grad_norm": 0.08727839589118958,
      "learning_rate": 8.690354027706517e-06,
      "loss": 0.0063,
      "step": 27100
    },
    {
      "epoch": 0.1738751190541091,
      "grad_norm": 0.44716620445251465,
      "learning_rate": 8.693560800410468e-06,
      "loss": 0.0087,
      "step": 27110
    },
    {
      "epoch": 0.1739392559478952,
      "grad_norm": 0.47064927220344543,
      "learning_rate": 8.69676757311442e-06,
      "loss": 0.008,
      "step": 27120
    },
    {
      "epoch": 0.1740033928416813,
      "grad_norm": 0.4345616400241852,
      "learning_rate": 8.699974345818369e-06,
      "loss": 0.008,
      "step": 27130
    },
    {
      "epoch": 0.17406752973546738,
      "grad_norm": 0.40541645884513855,
      "learning_rate": 8.70318111852232e-06,
      "loss": 0.0055,
      "step": 27140
    },
    {
      "epoch": 0.17413166662925347,
      "grad_norm": 0.6267321705818176,
      "learning_rate": 8.706387891226271e-06,
      "loss": 0.0105,
      "step": 27150
    },
    {
      "epoch": 0.17419580352303957,
      "grad_norm": 0.2842670977115631,
      "learning_rate": 8.70959466393022e-06,
      "loss": 0.0095,
      "step": 27160
    },
    {
      "epoch": 0.17425994041682566,
      "grad_norm": 0.36302152276039124,
      "learning_rate": 8.712801436634171e-06,
      "loss": 0.0066,
      "step": 27170
    },
    {
      "epoch": 0.17432407731061178,
      "grad_norm": 0.2736494243144989,
      "learning_rate": 8.716008209338123e-06,
      "loss": 0.0109,
      "step": 27180
    },
    {
      "epoch": 0.17438821420439787,
      "grad_norm": 0.46017733216285706,
      "learning_rate": 8.719214982042074e-06,
      "loss": 0.0089,
      "step": 27190
    },
    {
      "epoch": 0.17445235109818397,
      "grad_norm": 0.4222848117351532,
      "learning_rate": 8.722421754746025e-06,
      "loss": 0.0079,
      "step": 27200
    },
    {
      "epoch": 0.17451648799197006,
      "grad_norm": 0.5578427910804749,
      "learning_rate": 8.725628527449976e-06,
      "loss": 0.0087,
      "step": 27210
    },
    {
      "epoch": 0.17458062488575615,
      "grad_norm": 0.1966858059167862,
      "learning_rate": 8.728835300153925e-06,
      "loss": 0.0113,
      "step": 27220
    },
    {
      "epoch": 0.17464476177954225,
      "grad_norm": 0.6067239046096802,
      "learning_rate": 8.732042072857876e-06,
      "loss": 0.009,
      "step": 27230
    },
    {
      "epoch": 0.17470889867332834,
      "grad_norm": 0.5752755999565125,
      "learning_rate": 8.735248845561828e-06,
      "loss": 0.007,
      "step": 27240
    },
    {
      "epoch": 0.17477303556711446,
      "grad_norm": 0.15181779861450195,
      "learning_rate": 8.738455618265777e-06,
      "loss": 0.0077,
      "step": 27250
    },
    {
      "epoch": 0.17483717246090055,
      "grad_norm": 0.3850315809249878,
      "learning_rate": 8.741662390969728e-06,
      "loss": 0.0062,
      "step": 27260
    },
    {
      "epoch": 0.17490130935468665,
      "grad_norm": 0.40511053800582886,
      "learning_rate": 8.744869163673679e-06,
      "loss": 0.0097,
      "step": 27270
    },
    {
      "epoch": 0.17496544624847274,
      "grad_norm": 0.23062027990818024,
      "learning_rate": 8.74807593637763e-06,
      "loss": 0.0052,
      "step": 27280
    },
    {
      "epoch": 0.17502958314225883,
      "grad_norm": 0.2363414466381073,
      "learning_rate": 8.751282709081581e-06,
      "loss": 0.0068,
      "step": 27290
    },
    {
      "epoch": 0.17509372003604493,
      "grad_norm": 0.4977990388870239,
      "learning_rate": 8.754489481785532e-06,
      "loss": 0.009,
      "step": 27300
    },
    {
      "epoch": 0.17515785692983102,
      "grad_norm": 0.6856558322906494,
      "learning_rate": 8.757696254489484e-06,
      "loss": 0.0074,
      "step": 27310
    },
    {
      "epoch": 0.17522199382361714,
      "grad_norm": 1.0530060529708862,
      "learning_rate": 8.760903027193433e-06,
      "loss": 0.0116,
      "step": 27320
    },
    {
      "epoch": 0.17528613071740323,
      "grad_norm": 0.6366263031959534,
      "learning_rate": 8.764109799897384e-06,
      "loss": 0.008,
      "step": 27330
    },
    {
      "epoch": 0.17535026761118933,
      "grad_norm": 0.35337623953819275,
      "learning_rate": 8.767316572601334e-06,
      "loss": 0.0059,
      "step": 27340
    },
    {
      "epoch": 0.17541440450497542,
      "grad_norm": 0.9613513946533203,
      "learning_rate": 8.770523345305285e-06,
      "loss": 0.0095,
      "step": 27350
    },
    {
      "epoch": 0.1754785413987615,
      "grad_norm": 0.2593524754047394,
      "learning_rate": 8.773730118009236e-06,
      "loss": 0.0064,
      "step": 27360
    },
    {
      "epoch": 0.1755426782925476,
      "grad_norm": 0.5186986327171326,
      "learning_rate": 8.776936890713187e-06,
      "loss": 0.007,
      "step": 27370
    },
    {
      "epoch": 0.1756068151863337,
      "grad_norm": 0.22691476345062256,
      "learning_rate": 8.780143663417138e-06,
      "loss": 0.0074,
      "step": 27380
    },
    {
      "epoch": 0.17567095208011982,
      "grad_norm": 0.4468889832496643,
      "learning_rate": 8.783350436121089e-06,
      "loss": 0.0078,
      "step": 27390
    },
    {
      "epoch": 0.1757350889739059,
      "grad_norm": 0.8645966649055481,
      "learning_rate": 8.78655720882504e-06,
      "loss": 0.006,
      "step": 27400
    },
    {
      "epoch": 0.175799225867692,
      "grad_norm": 0.5231471061706543,
      "learning_rate": 8.78976398152899e-06,
      "loss": 0.0081,
      "step": 27410
    },
    {
      "epoch": 0.1758633627614781,
      "grad_norm": 0.39743897318840027,
      "learning_rate": 8.79297075423294e-06,
      "loss": 0.0077,
      "step": 27420
    },
    {
      "epoch": 0.1759274996552642,
      "grad_norm": 0.49018046259880066,
      "learning_rate": 8.79617752693689e-06,
      "loss": 0.0065,
      "step": 27430
    },
    {
      "epoch": 0.17599163654905028,
      "grad_norm": 0.4271479845046997,
      "learning_rate": 8.799384299640841e-06,
      "loss": 0.0059,
      "step": 27440
    },
    {
      "epoch": 0.17605577344283638,
      "grad_norm": 0.3524341285228729,
      "learning_rate": 8.802591072344792e-06,
      "loss": 0.0111,
      "step": 27450
    },
    {
      "epoch": 0.1761199103366225,
      "grad_norm": 0.4612078070640564,
      "learning_rate": 8.805797845048744e-06,
      "loss": 0.0074,
      "step": 27460
    },
    {
      "epoch": 0.1761840472304086,
      "grad_norm": 0.24353353679180145,
      "learning_rate": 8.809004617752695e-06,
      "loss": 0.0104,
      "step": 27470
    },
    {
      "epoch": 0.17624818412419468,
      "grad_norm": 0.13203102350234985,
      "learning_rate": 8.812211390456646e-06,
      "loss": 0.0069,
      "step": 27480
    },
    {
      "epoch": 0.17631232101798078,
      "grad_norm": 0.09631278365850449,
      "learning_rate": 8.815418163160597e-06,
      "loss": 0.0074,
      "step": 27490
    },
    {
      "epoch": 0.17637645791176687,
      "grad_norm": 0.308401495218277,
      "learning_rate": 8.818624935864546e-06,
      "loss": 0.0062,
      "step": 27500
    },
    {
      "epoch": 0.17644059480555296,
      "grad_norm": 0.13009114563465118,
      "learning_rate": 8.821831708568497e-06,
      "loss": 0.0083,
      "step": 27510
    },
    {
      "epoch": 0.17650473169933906,
      "grad_norm": 0.1699720323085785,
      "learning_rate": 8.825038481272449e-06,
      "loss": 0.0057,
      "step": 27520
    },
    {
      "epoch": 0.17656886859312518,
      "grad_norm": 0.03352305293083191,
      "learning_rate": 8.828245253976398e-06,
      "loss": 0.0049,
      "step": 27530
    },
    {
      "epoch": 0.17663300548691127,
      "grad_norm": 0.4160129427909851,
      "learning_rate": 8.831452026680349e-06,
      "loss": 0.0064,
      "step": 27540
    },
    {
      "epoch": 0.17669714238069736,
      "grad_norm": 0.27879947423934937,
      "learning_rate": 8.8346587993843e-06,
      "loss": 0.0097,
      "step": 27550
    },
    {
      "epoch": 0.17676127927448346,
      "grad_norm": 0.19119217991828918,
      "learning_rate": 8.837865572088251e-06,
      "loss": 0.0076,
      "step": 27560
    },
    {
      "epoch": 0.17682541616826955,
      "grad_norm": 0.26554617285728455,
      "learning_rate": 8.841072344792202e-06,
      "loss": 0.0078,
      "step": 27570
    },
    {
      "epoch": 0.17688955306205564,
      "grad_norm": 0.4061427116394043,
      "learning_rate": 8.844279117496153e-06,
      "loss": 0.0051,
      "step": 27580
    },
    {
      "epoch": 0.17695368995584174,
      "grad_norm": 0.13531051576137543,
      "learning_rate": 8.847485890200103e-06,
      "loss": 0.0102,
      "step": 27590
    },
    {
      "epoch": 0.17701782684962786,
      "grad_norm": 0.31303972005844116,
      "learning_rate": 8.850692662904054e-06,
      "loss": 0.005,
      "step": 27600
    },
    {
      "epoch": 0.17708196374341395,
      "grad_norm": 0.42270204424858093,
      "learning_rate": 8.853899435608005e-06,
      "loss": 0.0103,
      "step": 27610
    },
    {
      "epoch": 0.17714610063720004,
      "grad_norm": 0.22162435948848724,
      "learning_rate": 8.857106208311955e-06,
      "loss": 0.0064,
      "step": 27620
    },
    {
      "epoch": 0.17721023753098614,
      "grad_norm": 0.16792258620262146,
      "learning_rate": 8.860312981015906e-06,
      "loss": 0.006,
      "step": 27630
    },
    {
      "epoch": 0.17727437442477223,
      "grad_norm": 0.32242026925086975,
      "learning_rate": 8.863519753719857e-06,
      "loss": 0.0054,
      "step": 27640
    },
    {
      "epoch": 0.17733851131855832,
      "grad_norm": 0.48972517251968384,
      "learning_rate": 8.866726526423808e-06,
      "loss": 0.0058,
      "step": 27650
    },
    {
      "epoch": 0.17740264821234442,
      "grad_norm": 0.36505600810050964,
      "learning_rate": 8.869933299127759e-06,
      "loss": 0.0094,
      "step": 27660
    },
    {
      "epoch": 0.17746678510613054,
      "grad_norm": 0.2752530574798584,
      "learning_rate": 8.87314007183171e-06,
      "loss": 0.007,
      "step": 27670
    },
    {
      "epoch": 0.17753092199991663,
      "grad_norm": 0.22372488677501678,
      "learning_rate": 8.876346844535661e-06,
      "loss": 0.0054,
      "step": 27680
    },
    {
      "epoch": 0.17759505889370272,
      "grad_norm": 0.6196951270103455,
      "learning_rate": 8.87955361723961e-06,
      "loss": 0.0079,
      "step": 27690
    },
    {
      "epoch": 0.17765919578748882,
      "grad_norm": 0.3792284429073334,
      "learning_rate": 8.882760389943562e-06,
      "loss": 0.0084,
      "step": 27700
    },
    {
      "epoch": 0.1777233326812749,
      "grad_norm": 0.33533328771591187,
      "learning_rate": 8.885967162647511e-06,
      "loss": 0.007,
      "step": 27710
    },
    {
      "epoch": 0.177787469575061,
      "grad_norm": 0.20695863664150238,
      "learning_rate": 8.889173935351462e-06,
      "loss": 0.0079,
      "step": 27720
    },
    {
      "epoch": 0.1778516064688471,
      "grad_norm": 0.3518396317958832,
      "learning_rate": 8.892380708055413e-06,
      "loss": 0.0072,
      "step": 27730
    },
    {
      "epoch": 0.17791574336263322,
      "grad_norm": 0.24414193630218506,
      "learning_rate": 8.895587480759365e-06,
      "loss": 0.0044,
      "step": 27740
    },
    {
      "epoch": 0.1779798802564193,
      "grad_norm": 0.4045836925506592,
      "learning_rate": 8.898794253463316e-06,
      "loss": 0.0085,
      "step": 27750
    },
    {
      "epoch": 0.1780440171502054,
      "grad_norm": 0.30575981736183167,
      "learning_rate": 8.902001026167267e-06,
      "loss": 0.0083,
      "step": 27760
    },
    {
      "epoch": 0.1781081540439915,
      "grad_norm": 0.5452970266342163,
      "learning_rate": 8.905207798871218e-06,
      "loss": 0.0082,
      "step": 27770
    },
    {
      "epoch": 0.1781722909377776,
      "grad_norm": 0.31361040472984314,
      "learning_rate": 8.908414571575167e-06,
      "loss": 0.005,
      "step": 27780
    },
    {
      "epoch": 0.17823642783156368,
      "grad_norm": 0.2590939402580261,
      "learning_rate": 8.911621344279118e-06,
      "loss": 0.0074,
      "step": 27790
    },
    {
      "epoch": 0.17830056472534977,
      "grad_norm": 0.3190052807331085,
      "learning_rate": 8.914828116983068e-06,
      "loss": 0.0064,
      "step": 27800
    },
    {
      "epoch": 0.1783647016191359,
      "grad_norm": 0.35626527667045593,
      "learning_rate": 8.918034889687019e-06,
      "loss": 0.0077,
      "step": 27810
    },
    {
      "epoch": 0.178428838512922,
      "grad_norm": 0.40019169449806213,
      "learning_rate": 8.92124166239097e-06,
      "loss": 0.0075,
      "step": 27820
    },
    {
      "epoch": 0.17849297540670808,
      "grad_norm": 0.2987542450428009,
      "learning_rate": 8.924448435094921e-06,
      "loss": 0.0095,
      "step": 27830
    },
    {
      "epoch": 0.17855711230049418,
      "grad_norm": 0.4885701835155487,
      "learning_rate": 8.927655207798872e-06,
      "loss": 0.0078,
      "step": 27840
    },
    {
      "epoch": 0.17862124919428027,
      "grad_norm": 0.5281029939651489,
      "learning_rate": 8.930861980502823e-06,
      "loss": 0.0041,
      "step": 27850
    },
    {
      "epoch": 0.17868538608806636,
      "grad_norm": 0.5410029292106628,
      "learning_rate": 8.934068753206774e-06,
      "loss": 0.0106,
      "step": 27860
    },
    {
      "epoch": 0.17874952298185245,
      "grad_norm": 0.3174389898777008,
      "learning_rate": 8.937275525910724e-06,
      "loss": 0.006,
      "step": 27870
    },
    {
      "epoch": 0.17881365987563858,
      "grad_norm": 0.3071681559085846,
      "learning_rate": 8.940482298614675e-06,
      "loss": 0.0093,
      "step": 27880
    },
    {
      "epoch": 0.17887779676942467,
      "grad_norm": 0.2326798141002655,
      "learning_rate": 8.943689071318626e-06,
      "loss": 0.0067,
      "step": 27890
    },
    {
      "epoch": 0.17894193366321076,
      "grad_norm": 0.27320408821105957,
      "learning_rate": 8.946895844022576e-06,
      "loss": 0.0111,
      "step": 27900
    },
    {
      "epoch": 0.17900607055699685,
      "grad_norm": 0.3682153522968292,
      "learning_rate": 8.950102616726527e-06,
      "loss": 0.0052,
      "step": 27910
    },
    {
      "epoch": 0.17907020745078295,
      "grad_norm": 0.20906396210193634,
      "learning_rate": 8.953309389430478e-06,
      "loss": 0.0097,
      "step": 27920
    },
    {
      "epoch": 0.17913434434456904,
      "grad_norm": 0.5595477819442749,
      "learning_rate": 8.956516162134429e-06,
      "loss": 0.0104,
      "step": 27930
    },
    {
      "epoch": 0.17919848123835513,
      "grad_norm": 0.5162923336029053,
      "learning_rate": 8.95972293483838e-06,
      "loss": 0.0073,
      "step": 27940
    },
    {
      "epoch": 0.17926261813214125,
      "grad_norm": 0.463011771440506,
      "learning_rate": 8.962929707542331e-06,
      "loss": 0.0066,
      "step": 27950
    },
    {
      "epoch": 0.17932675502592735,
      "grad_norm": 0.30121463537216187,
      "learning_rate": 8.96613648024628e-06,
      "loss": 0.0037,
      "step": 27960
    },
    {
      "epoch": 0.17939089191971344,
      "grad_norm": 0.3653218746185303,
      "learning_rate": 8.969343252950232e-06,
      "loss": 0.0098,
      "step": 27970
    },
    {
      "epoch": 0.17945502881349953,
      "grad_norm": 0.3108031451702118,
      "learning_rate": 8.972550025654183e-06,
      "loss": 0.007,
      "step": 27980
    },
    {
      "epoch": 0.17951916570728563,
      "grad_norm": 0.12764820456504822,
      "learning_rate": 8.975756798358132e-06,
      "loss": 0.0049,
      "step": 27990
    },
    {
      "epoch": 0.17958330260107172,
      "grad_norm": 0.336069792509079,
      "learning_rate": 8.978963571062083e-06,
      "loss": 0.0046,
      "step": 28000
    },
    {
      "epoch": 0.1796474394948578,
      "grad_norm": 0.522538959980011,
      "learning_rate": 8.982170343766034e-06,
      "loss": 0.0094,
      "step": 28010
    },
    {
      "epoch": 0.17971157638864393,
      "grad_norm": 0.35158011317253113,
      "learning_rate": 8.985377116469986e-06,
      "loss": 0.0061,
      "step": 28020
    },
    {
      "epoch": 0.17977571328243003,
      "grad_norm": 0.2685913145542145,
      "learning_rate": 8.988583889173937e-06,
      "loss": 0.0075,
      "step": 28030
    },
    {
      "epoch": 0.17983985017621612,
      "grad_norm": 0.21148598194122314,
      "learning_rate": 8.991790661877888e-06,
      "loss": 0.0048,
      "step": 28040
    },
    {
      "epoch": 0.1799039870700022,
      "grad_norm": 0.08513572067022324,
      "learning_rate": 8.994997434581839e-06,
      "loss": 0.0087,
      "step": 28050
    },
    {
      "epoch": 0.1799681239637883,
      "grad_norm": 0.39910271763801575,
      "learning_rate": 8.998204207285788e-06,
      "loss": 0.0066,
      "step": 28060
    },
    {
      "epoch": 0.1800322608575744,
      "grad_norm": 0.40334391593933105,
      "learning_rate": 9.00141097998974e-06,
      "loss": 0.0076,
      "step": 28070
    },
    {
      "epoch": 0.1800963977513605,
      "grad_norm": 0.20072820782661438,
      "learning_rate": 9.004617752693689e-06,
      "loss": 0.0067,
      "step": 28080
    },
    {
      "epoch": 0.1801605346451466,
      "grad_norm": 0.6156127452850342,
      "learning_rate": 9.00782452539764e-06,
      "loss": 0.008,
      "step": 28090
    },
    {
      "epoch": 0.1802246715389327,
      "grad_norm": 0.5517195463180542,
      "learning_rate": 9.011031298101591e-06,
      "loss": 0.0079,
      "step": 28100
    },
    {
      "epoch": 0.1802888084327188,
      "grad_norm": 0.3041163980960846,
      "learning_rate": 9.014238070805542e-06,
      "loss": 0.0073,
      "step": 28110
    },
    {
      "epoch": 0.1803529453265049,
      "grad_norm": 0.21832886338233948,
      "learning_rate": 9.017444843509493e-06,
      "loss": 0.0056,
      "step": 28120
    },
    {
      "epoch": 0.180417082220291,
      "grad_norm": 0.2679687440395355,
      "learning_rate": 9.020651616213444e-06,
      "loss": 0.0059,
      "step": 28130
    },
    {
      "epoch": 0.18048121911407708,
      "grad_norm": 0.12943078577518463,
      "learning_rate": 9.023858388917395e-06,
      "loss": 0.0046,
      "step": 28140
    },
    {
      "epoch": 0.18054535600786317,
      "grad_norm": 0.39300015568733215,
      "learning_rate": 9.027065161621345e-06,
      "loss": 0.0067,
      "step": 28150
    },
    {
      "epoch": 0.1806094929016493,
      "grad_norm": 0.0813639909029007,
      "learning_rate": 9.030271934325296e-06,
      "loss": 0.0081,
      "step": 28160
    },
    {
      "epoch": 0.1806736297954354,
      "grad_norm": 0.2814038097858429,
      "learning_rate": 9.033478707029245e-06,
      "loss": 0.0094,
      "step": 28170
    },
    {
      "epoch": 0.18073776668922148,
      "grad_norm": 0.2565455734729767,
      "learning_rate": 9.036685479733197e-06,
      "loss": 0.0041,
      "step": 28180
    },
    {
      "epoch": 0.18080190358300757,
      "grad_norm": 0.4624595046043396,
      "learning_rate": 9.039892252437148e-06,
      "loss": 0.0047,
      "step": 28190
    },
    {
      "epoch": 0.18086604047679367,
      "grad_norm": 0.3168759346008301,
      "learning_rate": 9.043099025141099e-06,
      "loss": 0.0066,
      "step": 28200
    },
    {
      "epoch": 0.18093017737057976,
      "grad_norm": 0.06198474392294884,
      "learning_rate": 9.04630579784505e-06,
      "loss": 0.0069,
      "step": 28210
    },
    {
      "epoch": 0.18099431426436585,
      "grad_norm": 0.14223305881023407,
      "learning_rate": 9.049512570549001e-06,
      "loss": 0.0071,
      "step": 28220
    },
    {
      "epoch": 0.18105845115815197,
      "grad_norm": 0.2489401400089264,
      "learning_rate": 9.052719343252952e-06,
      "loss": 0.0085,
      "step": 28230
    },
    {
      "epoch": 0.18112258805193807,
      "grad_norm": 0.22115850448608398,
      "learning_rate": 9.055926115956902e-06,
      "loss": 0.0069,
      "step": 28240
    },
    {
      "epoch": 0.18118672494572416,
      "grad_norm": 0.3180225193500519,
      "learning_rate": 9.059132888660853e-06,
      "loss": 0.0091,
      "step": 28250
    },
    {
      "epoch": 0.18125086183951025,
      "grad_norm": 0.2616806626319885,
      "learning_rate": 9.062339661364802e-06,
      "loss": 0.011,
      "step": 28260
    },
    {
      "epoch": 0.18131499873329635,
      "grad_norm": 0.33274537324905396,
      "learning_rate": 9.065546434068753e-06,
      "loss": 0.0087,
      "step": 28270
    },
    {
      "epoch": 0.18137913562708244,
      "grad_norm": 0.35504183173179626,
      "learning_rate": 9.068753206772704e-06,
      "loss": 0.0101,
      "step": 28280
    },
    {
      "epoch": 0.18144327252086853,
      "grad_norm": 0.5860472917556763,
      "learning_rate": 9.071959979476655e-06,
      "loss": 0.0061,
      "step": 28290
    },
    {
      "epoch": 0.18150740941465465,
      "grad_norm": 0.2541690170764923,
      "learning_rate": 9.075166752180607e-06,
      "loss": 0.0095,
      "step": 28300
    },
    {
      "epoch": 0.18157154630844075,
      "grad_norm": 0.3021218776702881,
      "learning_rate": 9.078373524884558e-06,
      "loss": 0.0059,
      "step": 28310
    },
    {
      "epoch": 0.18163568320222684,
      "grad_norm": 0.5009143948554993,
      "learning_rate": 9.081580297588509e-06,
      "loss": 0.0057,
      "step": 28320
    },
    {
      "epoch": 0.18169982009601293,
      "grad_norm": 0.13357484340667725,
      "learning_rate": 9.084787070292458e-06,
      "loss": 0.0079,
      "step": 28330
    },
    {
      "epoch": 0.18176395698979902,
      "grad_norm": 0.29337722063064575,
      "learning_rate": 9.08799384299641e-06,
      "loss": 0.0069,
      "step": 28340
    },
    {
      "epoch": 0.18182809388358512,
      "grad_norm": 0.3855084180831909,
      "learning_rate": 9.09120061570036e-06,
      "loss": 0.0061,
      "step": 28350
    },
    {
      "epoch": 0.1818922307773712,
      "grad_norm": 0.548573911190033,
      "learning_rate": 9.09440738840431e-06,
      "loss": 0.0072,
      "step": 28360
    },
    {
      "epoch": 0.18195636767115733,
      "grad_norm": 0.8141303062438965,
      "learning_rate": 9.097614161108261e-06,
      "loss": 0.0085,
      "step": 28370
    },
    {
      "epoch": 0.18202050456494343,
      "grad_norm": 0.3887251317501068,
      "learning_rate": 9.100820933812212e-06,
      "loss": 0.0118,
      "step": 28380
    },
    {
      "epoch": 0.18208464145872952,
      "grad_norm": 0.033384427428245544,
      "learning_rate": 9.104027706516163e-06,
      "loss": 0.0081,
      "step": 28390
    },
    {
      "epoch": 0.1821487783525156,
      "grad_norm": 0.17048637568950653,
      "learning_rate": 9.107234479220114e-06,
      "loss": 0.0074,
      "step": 28400
    },
    {
      "epoch": 0.1822129152463017,
      "grad_norm": 0.2683856189250946,
      "learning_rate": 9.110441251924065e-06,
      "loss": 0.0048,
      "step": 28410
    },
    {
      "epoch": 0.1822770521400878,
      "grad_norm": 0.3027239143848419,
      "learning_rate": 9.113648024628015e-06,
      "loss": 0.0069,
      "step": 28420
    },
    {
      "epoch": 0.1823411890338739,
      "grad_norm": 0.5860539078712463,
      "learning_rate": 9.116854797331966e-06,
      "loss": 0.012,
      "step": 28430
    },
    {
      "epoch": 0.18240532592765998,
      "grad_norm": 0.675926923751831,
      "learning_rate": 9.120061570035917e-06,
      "loss": 0.0078,
      "step": 28440
    },
    {
      "epoch": 0.1824694628214461,
      "grad_norm": 0.4775112271308899,
      "learning_rate": 9.123268342739866e-06,
      "loss": 0.0055,
      "step": 28450
    },
    {
      "epoch": 0.1825335997152322,
      "grad_norm": 0.41792356967926025,
      "learning_rate": 9.126475115443818e-06,
      "loss": 0.0085,
      "step": 28460
    },
    {
      "epoch": 0.1825977366090183,
      "grad_norm": 0.22479628026485443,
      "learning_rate": 9.129681888147769e-06,
      "loss": 0.0063,
      "step": 28470
    },
    {
      "epoch": 0.18266187350280438,
      "grad_norm": 0.7009286284446716,
      "learning_rate": 9.13288866085172e-06,
      "loss": 0.0077,
      "step": 28480
    },
    {
      "epoch": 0.18272601039659048,
      "grad_norm": 0.33191394805908203,
      "learning_rate": 9.136095433555671e-06,
      "loss": 0.0095,
      "step": 28490
    },
    {
      "epoch": 0.18279014729037657,
      "grad_norm": 0.655703067779541,
      "learning_rate": 9.139302206259622e-06,
      "loss": 0.019,
      "step": 28500
    },
    {
      "epoch": 0.18285428418416266,
      "grad_norm": 0.43687865138053894,
      "learning_rate": 9.142508978963571e-06,
      "loss": 0.0065,
      "step": 28510
    },
    {
      "epoch": 0.18291842107794878,
      "grad_norm": 0.1936979442834854,
      "learning_rate": 9.145715751667523e-06,
      "loss": 0.0075,
      "step": 28520
    },
    {
      "epoch": 0.18298255797173488,
      "grad_norm": 0.27804601192474365,
      "learning_rate": 9.148922524371474e-06,
      "loss": 0.0078,
      "step": 28530
    },
    {
      "epoch": 0.18304669486552097,
      "grad_norm": 0.34704890847206116,
      "learning_rate": 9.152129297075423e-06,
      "loss": 0.0062,
      "step": 28540
    },
    {
      "epoch": 0.18311083175930706,
      "grad_norm": 0.27939388155937195,
      "learning_rate": 9.155336069779374e-06,
      "loss": 0.0062,
      "step": 28550
    },
    {
      "epoch": 0.18317496865309316,
      "grad_norm": 0.222448468208313,
      "learning_rate": 9.158542842483325e-06,
      "loss": 0.0054,
      "step": 28560
    },
    {
      "epoch": 0.18323910554687925,
      "grad_norm": 0.12520845234394073,
      "learning_rate": 9.161749615187276e-06,
      "loss": 0.0112,
      "step": 28570
    },
    {
      "epoch": 0.18330324244066534,
      "grad_norm": 0.39360687136650085,
      "learning_rate": 9.164956387891228e-06,
      "loss": 0.0111,
      "step": 28580
    },
    {
      "epoch": 0.18336737933445146,
      "grad_norm": 0.32524222135543823,
      "learning_rate": 9.168163160595179e-06,
      "loss": 0.0089,
      "step": 28590
    },
    {
      "epoch": 0.18343151622823756,
      "grad_norm": 0.4135376513004303,
      "learning_rate": 9.17136993329913e-06,
      "loss": 0.005,
      "step": 28600
    },
    {
      "epoch": 0.18349565312202365,
      "grad_norm": 0.39116549491882324,
      "learning_rate": 9.17457670600308e-06,
      "loss": 0.0065,
      "step": 28610
    },
    {
      "epoch": 0.18355979001580974,
      "grad_norm": 0.25644803047180176,
      "learning_rate": 9.17778347870703e-06,
      "loss": 0.0089,
      "step": 28620
    },
    {
      "epoch": 0.18362392690959584,
      "grad_norm": 0.41352519392967224,
      "learning_rate": 9.18099025141098e-06,
      "loss": 0.0133,
      "step": 28630
    },
    {
      "epoch": 0.18368806380338193,
      "grad_norm": 0.756648600101471,
      "learning_rate": 9.18419702411493e-06,
      "loss": 0.007,
      "step": 28640
    },
    {
      "epoch": 0.18375220069716802,
      "grad_norm": 0.48206591606140137,
      "learning_rate": 9.187403796818882e-06,
      "loss": 0.0061,
      "step": 28650
    },
    {
      "epoch": 0.18381633759095414,
      "grad_norm": 0.47298529744148254,
      "learning_rate": 9.190610569522833e-06,
      "loss": 0.0055,
      "step": 28660
    },
    {
      "epoch": 0.18388047448474024,
      "grad_norm": 0.4739581346511841,
      "learning_rate": 9.193817342226784e-06,
      "loss": 0.0066,
      "step": 28670
    },
    {
      "epoch": 0.18394461137852633,
      "grad_norm": 0.39162930846214294,
      "learning_rate": 9.197024114930735e-06,
      "loss": 0.007,
      "step": 28680
    },
    {
      "epoch": 0.18400874827231242,
      "grad_norm": 0.37037205696105957,
      "learning_rate": 9.200230887634686e-06,
      "loss": 0.0064,
      "step": 28690
    },
    {
      "epoch": 0.18407288516609852,
      "grad_norm": 0.22075816988945007,
      "learning_rate": 9.203437660338636e-06,
      "loss": 0.006,
      "step": 28700
    },
    {
      "epoch": 0.1841370220598846,
      "grad_norm": 0.212757408618927,
      "learning_rate": 9.206644433042587e-06,
      "loss": 0.0087,
      "step": 28710
    },
    {
      "epoch": 0.1842011589536707,
      "grad_norm": 0.4505358040332794,
      "learning_rate": 9.209851205746538e-06,
      "loss": 0.0065,
      "step": 28720
    },
    {
      "epoch": 0.18426529584745682,
      "grad_norm": 0.6356704235076904,
      "learning_rate": 9.213057978450487e-06,
      "loss": 0.0058,
      "step": 28730
    },
    {
      "epoch": 0.18432943274124292,
      "grad_norm": 0.4008321464061737,
      "learning_rate": 9.216264751154439e-06,
      "loss": 0.0073,
      "step": 28740
    },
    {
      "epoch": 0.184393569635029,
      "grad_norm": 0.43277493119239807,
      "learning_rate": 9.21947152385839e-06,
      "loss": 0.0118,
      "step": 28750
    },
    {
      "epoch": 0.1844577065288151,
      "grad_norm": 0.3399185240268707,
      "learning_rate": 9.22267829656234e-06,
      "loss": 0.0087,
      "step": 28760
    },
    {
      "epoch": 0.1845218434226012,
      "grad_norm": 0.5668254494667053,
      "learning_rate": 9.225885069266292e-06,
      "loss": 0.0061,
      "step": 28770
    },
    {
      "epoch": 0.1845859803163873,
      "grad_norm": 0.5704218149185181,
      "learning_rate": 9.229091841970243e-06,
      "loss": 0.006,
      "step": 28780
    },
    {
      "epoch": 0.18465011721017338,
      "grad_norm": 0.6402566432952881,
      "learning_rate": 9.232298614674192e-06,
      "loss": 0.0075,
      "step": 28790
    },
    {
      "epoch": 0.1847142541039595,
      "grad_norm": 0.14951485395431519,
      "learning_rate": 9.235505387378144e-06,
      "loss": 0.0072,
      "step": 28800
    },
    {
      "epoch": 0.1847783909977456,
      "grad_norm": 0.5898628234863281,
      "learning_rate": 9.238712160082095e-06,
      "loss": 0.0099,
      "step": 28810
    },
    {
      "epoch": 0.1848425278915317,
      "grad_norm": 0.49422821402549744,
      "learning_rate": 9.241918932786044e-06,
      "loss": 0.0055,
      "step": 28820
    },
    {
      "epoch": 0.18490666478531778,
      "grad_norm": 0.2988821268081665,
      "learning_rate": 9.245125705489995e-06,
      "loss": 0.0057,
      "step": 28830
    },
    {
      "epoch": 0.18497080167910387,
      "grad_norm": 0.4461117684841156,
      "learning_rate": 9.248332478193946e-06,
      "loss": 0.0079,
      "step": 28840
    },
    {
      "epoch": 0.18503493857288997,
      "grad_norm": 0.43934503197669983,
      "learning_rate": 9.251539250897897e-06,
      "loss": 0.0114,
      "step": 28850
    },
    {
      "epoch": 0.18509907546667606,
      "grad_norm": 0.3138737976551056,
      "learning_rate": 9.254746023601849e-06,
      "loss": 0.0082,
      "step": 28860
    },
    {
      "epoch": 0.18516321236046218,
      "grad_norm": 0.6254568099975586,
      "learning_rate": 9.2579527963058e-06,
      "loss": 0.0081,
      "step": 28870
    },
    {
      "epoch": 0.18522734925424827,
      "grad_norm": 0.5110483765602112,
      "learning_rate": 9.261159569009749e-06,
      "loss": 0.006,
      "step": 28880
    },
    {
      "epoch": 0.18529148614803437,
      "grad_norm": 0.3529542088508606,
      "learning_rate": 9.2643663417137e-06,
      "loss": 0.0072,
      "step": 28890
    },
    {
      "epoch": 0.18535562304182046,
      "grad_norm": 0.5061513781547546,
      "learning_rate": 9.267573114417651e-06,
      "loss": 0.0049,
      "step": 28900
    },
    {
      "epoch": 0.18541975993560655,
      "grad_norm": 0.4316900074481964,
      "learning_rate": 9.2707798871216e-06,
      "loss": 0.0083,
      "step": 28910
    },
    {
      "epoch": 0.18548389682939265,
      "grad_norm": 0.6369470953941345,
      "learning_rate": 9.273986659825552e-06,
      "loss": 0.0096,
      "step": 28920
    },
    {
      "epoch": 0.18554803372317874,
      "grad_norm": 0.25403037667274475,
      "learning_rate": 9.277193432529503e-06,
      "loss": 0.0107,
      "step": 28930
    },
    {
      "epoch": 0.18561217061696486,
      "grad_norm": 0.278654545545578,
      "learning_rate": 9.280400205233454e-06,
      "loss": 0.01,
      "step": 28940
    },
    {
      "epoch": 0.18567630751075095,
      "grad_norm": 0.20908258855342865,
      "learning_rate": 9.283606977937405e-06,
      "loss": 0.0067,
      "step": 28950
    },
    {
      "epoch": 0.18574044440453705,
      "grad_norm": 0.23901280760765076,
      "learning_rate": 9.286813750641356e-06,
      "loss": 0.0064,
      "step": 28960
    },
    {
      "epoch": 0.18580458129832314,
      "grad_norm": 0.4733389616012573,
      "learning_rate": 9.290020523345306e-06,
      "loss": 0.0083,
      "step": 28970
    },
    {
      "epoch": 0.18586871819210923,
      "grad_norm": 0.31443294882774353,
      "learning_rate": 9.293227296049257e-06,
      "loss": 0.0071,
      "step": 28980
    },
    {
      "epoch": 0.18593285508589533,
      "grad_norm": 0.27634161710739136,
      "learning_rate": 9.296434068753208e-06,
      "loss": 0.0056,
      "step": 28990
    },
    {
      "epoch": 0.18599699197968142,
      "grad_norm": 0.25937992334365845,
      "learning_rate": 9.299640841457157e-06,
      "loss": 0.0046,
      "step": 29000
    },
    {
      "epoch": 0.18606112887346754,
      "grad_norm": 0.29781779646873474,
      "learning_rate": 9.302847614161108e-06,
      "loss": 0.0076,
      "step": 29010
    },
    {
      "epoch": 0.18612526576725363,
      "grad_norm": 0.2678763270378113,
      "learning_rate": 9.30605438686506e-06,
      "loss": 0.0068,
      "step": 29020
    },
    {
      "epoch": 0.18618940266103973,
      "grad_norm": 0.29163113236427307,
      "learning_rate": 9.30926115956901e-06,
      "loss": 0.0073,
      "step": 29030
    },
    {
      "epoch": 0.18625353955482582,
      "grad_norm": 0.27826422452926636,
      "learning_rate": 9.312467932272962e-06,
      "loss": 0.0067,
      "step": 29040
    },
    {
      "epoch": 0.1863176764486119,
      "grad_norm": 0.36537352204322815,
      "learning_rate": 9.315674704976913e-06,
      "loss": 0.0063,
      "step": 29050
    },
    {
      "epoch": 0.186381813342398,
      "grad_norm": 0.5334638357162476,
      "learning_rate": 9.318881477680862e-06,
      "loss": 0.0057,
      "step": 29060
    },
    {
      "epoch": 0.1864459502361841,
      "grad_norm": 0.453370600938797,
      "learning_rate": 9.322088250384813e-06,
      "loss": 0.0065,
      "step": 29070
    },
    {
      "epoch": 0.18651008712997022,
      "grad_norm": 0.47658535838127136,
      "learning_rate": 9.325295023088765e-06,
      "loss": 0.0057,
      "step": 29080
    },
    {
      "epoch": 0.1865742240237563,
      "grad_norm": 0.09743408113718033,
      "learning_rate": 9.328501795792716e-06,
      "loss": 0.0048,
      "step": 29090
    },
    {
      "epoch": 0.1866383609175424,
      "grad_norm": 0.4015127420425415,
      "learning_rate": 9.331708568496665e-06,
      "loss": 0.0069,
      "step": 29100
    },
    {
      "epoch": 0.1867024978113285,
      "grad_norm": 0.36822250485420227,
      "learning_rate": 9.334915341200616e-06,
      "loss": 0.0062,
      "step": 29110
    },
    {
      "epoch": 0.1867666347051146,
      "grad_norm": 0.26752158999443054,
      "learning_rate": 9.338122113904567e-06,
      "loss": 0.0039,
      "step": 29120
    },
    {
      "epoch": 0.18683077159890069,
      "grad_norm": 0.1569271683692932,
      "learning_rate": 9.341328886608518e-06,
      "loss": 0.0067,
      "step": 29130
    },
    {
      "epoch": 0.18689490849268678,
      "grad_norm": 0.11092264950275421,
      "learning_rate": 9.34453565931247e-06,
      "loss": 0.0049,
      "step": 29140
    },
    {
      "epoch": 0.1869590453864729,
      "grad_norm": 0.20999155938625336,
      "learning_rate": 9.347742432016419e-06,
      "loss": 0.0071,
      "step": 29150
    },
    {
      "epoch": 0.187023182280259,
      "grad_norm": 0.22878888249397278,
      "learning_rate": 9.35094920472037e-06,
      "loss": 0.0057,
      "step": 29160
    },
    {
      "epoch": 0.18708731917404509,
      "grad_norm": 0.5292526483535767,
      "learning_rate": 9.354155977424321e-06,
      "loss": 0.0099,
      "step": 29170
    },
    {
      "epoch": 0.18715145606783118,
      "grad_norm": 0.250640869140625,
      "learning_rate": 9.357362750128272e-06,
      "loss": 0.0047,
      "step": 29180
    },
    {
      "epoch": 0.18721559296161727,
      "grad_norm": 0.4628705680370331,
      "learning_rate": 9.360569522832222e-06,
      "loss": 0.0109,
      "step": 29190
    },
    {
      "epoch": 0.18727972985540337,
      "grad_norm": 0.08864600211381912,
      "learning_rate": 9.363776295536173e-06,
      "loss": 0.0052,
      "step": 29200
    },
    {
      "epoch": 0.18734386674918946,
      "grad_norm": 0.4074329137802124,
      "learning_rate": 9.366983068240124e-06,
      "loss": 0.0042,
      "step": 29210
    },
    {
      "epoch": 0.18740800364297558,
      "grad_norm": 0.08895924687385559,
      "learning_rate": 9.370189840944075e-06,
      "loss": 0.0045,
      "step": 29220
    },
    {
      "epoch": 0.18747214053676167,
      "grad_norm": 0.39348146319389343,
      "learning_rate": 9.373396613648026e-06,
      "loss": 0.0092,
      "step": 29230
    },
    {
      "epoch": 0.18753627743054777,
      "grad_norm": 0.7161470055580139,
      "learning_rate": 9.376603386351976e-06,
      "loss": 0.0056,
      "step": 29240
    },
    {
      "epoch": 0.18760041432433386,
      "grad_norm": 0.2930874526500702,
      "learning_rate": 9.379810159055927e-06,
      "loss": 0.0089,
      "step": 29250
    },
    {
      "epoch": 0.18766455121811995,
      "grad_norm": 0.47388607263565063,
      "learning_rate": 9.383016931759878e-06,
      "loss": 0.0089,
      "step": 29260
    },
    {
      "epoch": 0.18772868811190604,
      "grad_norm": 0.4754874110221863,
      "learning_rate": 9.386223704463829e-06,
      "loss": 0.0053,
      "step": 29270
    },
    {
      "epoch": 0.18779282500569214,
      "grad_norm": 0.026910239830613136,
      "learning_rate": 9.389430477167778e-06,
      "loss": 0.0066,
      "step": 29280
    },
    {
      "epoch": 0.18785696189947826,
      "grad_norm": 0.26451611518859863,
      "learning_rate": 9.39263724987173e-06,
      "loss": 0.0115,
      "step": 29290
    },
    {
      "epoch": 0.18792109879326435,
      "grad_norm": 0.36335819959640503,
      "learning_rate": 9.39584402257568e-06,
      "loss": 0.0112,
      "step": 29300
    },
    {
      "epoch": 0.18798523568705044,
      "grad_norm": 0.3709248900413513,
      "learning_rate": 9.399050795279632e-06,
      "loss": 0.0073,
      "step": 29310
    },
    {
      "epoch": 0.18804937258083654,
      "grad_norm": 0.6065011620521545,
      "learning_rate": 9.402257567983583e-06,
      "loss": 0.0063,
      "step": 29320
    },
    {
      "epoch": 0.18811350947462263,
      "grad_norm": 0.5511764883995056,
      "learning_rate": 9.405464340687532e-06,
      "loss": 0.0054,
      "step": 29330
    },
    {
      "epoch": 0.18817764636840872,
      "grad_norm": 0.5607746839523315,
      "learning_rate": 9.408671113391483e-06,
      "loss": 0.0048,
      "step": 29340
    },
    {
      "epoch": 0.18824178326219482,
      "grad_norm": 0.1629454493522644,
      "learning_rate": 9.411877886095434e-06,
      "loss": 0.0062,
      "step": 29350
    },
    {
      "epoch": 0.18830592015598094,
      "grad_norm": 0.114433154463768,
      "learning_rate": 9.415084658799386e-06,
      "loss": 0.0087,
      "step": 29360
    },
    {
      "epoch": 0.18837005704976703,
      "grad_norm": 0.21146710216999054,
      "learning_rate": 9.418291431503335e-06,
      "loss": 0.0073,
      "step": 29370
    },
    {
      "epoch": 0.18843419394355312,
      "grad_norm": 0.28075212240219116,
      "learning_rate": 9.421498204207286e-06,
      "loss": 0.0075,
      "step": 29380
    },
    {
      "epoch": 0.18849833083733922,
      "grad_norm": 0.6241480708122253,
      "learning_rate": 9.424704976911237e-06,
      "loss": 0.0093,
      "step": 29390
    },
    {
      "epoch": 0.1885624677311253,
      "grad_norm": 0.4124491810798645,
      "learning_rate": 9.427911749615188e-06,
      "loss": 0.0053,
      "step": 29400
    },
    {
      "epoch": 0.1886266046249114,
      "grad_norm": 0.2754786014556885,
      "learning_rate": 9.43111852231914e-06,
      "loss": 0.0062,
      "step": 29410
    },
    {
      "epoch": 0.1886907415186975,
      "grad_norm": 0.566390872001648,
      "learning_rate": 9.434325295023089e-06,
      "loss": 0.0103,
      "step": 29420
    },
    {
      "epoch": 0.18875487841248362,
      "grad_norm": 0.538544774055481,
      "learning_rate": 9.43753206772704e-06,
      "loss": 0.0083,
      "step": 29430
    },
    {
      "epoch": 0.1888190153062697,
      "grad_norm": 0.2464735358953476,
      "learning_rate": 9.440738840430991e-06,
      "loss": 0.0056,
      "step": 29440
    },
    {
      "epoch": 0.1888831522000558,
      "grad_norm": 0.17093630135059357,
      "learning_rate": 9.443945613134942e-06,
      "loss": 0.005,
      "step": 29450
    },
    {
      "epoch": 0.1889472890938419,
      "grad_norm": 0.6086510419845581,
      "learning_rate": 9.447152385838893e-06,
      "loss": 0.0076,
      "step": 29460
    },
    {
      "epoch": 0.189011425987628,
      "grad_norm": 0.3101136088371277,
      "learning_rate": 9.450359158542843e-06,
      "loss": 0.0062,
      "step": 29470
    },
    {
      "epoch": 0.18907556288141408,
      "grad_norm": 0.18571898341178894,
      "learning_rate": 9.453565931246794e-06,
      "loss": 0.005,
      "step": 29480
    },
    {
      "epoch": 0.18913969977520018,
      "grad_norm": 0.14237219095230103,
      "learning_rate": 9.456772703950745e-06,
      "loss": 0.0049,
      "step": 29490
    },
    {
      "epoch": 0.1892038366689863,
      "grad_norm": 0.22469183802604675,
      "learning_rate": 9.459979476654696e-06,
      "loss": 0.0111,
      "step": 29500
    },
    {
      "epoch": 0.1892679735627724,
      "grad_norm": 0.08391445875167847,
      "learning_rate": 9.463186249358645e-06,
      "loss": 0.0087,
      "step": 29510
    },
    {
      "epoch": 0.18933211045655848,
      "grad_norm": 0.3000529110431671,
      "learning_rate": 9.466393022062597e-06,
      "loss": 0.0062,
      "step": 29520
    },
    {
      "epoch": 0.18939624735034458,
      "grad_norm": 0.3420923352241516,
      "learning_rate": 9.469599794766548e-06,
      "loss": 0.0078,
      "step": 29530
    },
    {
      "epoch": 0.18946038424413067,
      "grad_norm": 0.38952866196632385,
      "learning_rate": 9.472806567470499e-06,
      "loss": 0.0116,
      "step": 29540
    },
    {
      "epoch": 0.18952452113791676,
      "grad_norm": 0.39586111903190613,
      "learning_rate": 9.47601334017445e-06,
      "loss": 0.0068,
      "step": 29550
    },
    {
      "epoch": 0.18958865803170286,
      "grad_norm": 0.26339074969291687,
      "learning_rate": 9.4792201128784e-06,
      "loss": 0.0049,
      "step": 29560
    },
    {
      "epoch": 0.18965279492548898,
      "grad_norm": 0.2637289762496948,
      "learning_rate": 9.48242688558235e-06,
      "loss": 0.0044,
      "step": 29570
    },
    {
      "epoch": 0.18971693181927507,
      "grad_norm": 0.28041067719459534,
      "learning_rate": 9.485633658286302e-06,
      "loss": 0.0051,
      "step": 29580
    },
    {
      "epoch": 0.18978106871306116,
      "grad_norm": 0.30262765288352966,
      "learning_rate": 9.488840430990253e-06,
      "loss": 0.0122,
      "step": 29590
    },
    {
      "epoch": 0.18984520560684726,
      "grad_norm": 0.27783453464508057,
      "learning_rate": 9.492047203694202e-06,
      "loss": 0.0102,
      "step": 29600
    },
    {
      "epoch": 0.18990934250063335,
      "grad_norm": 0.3596213161945343,
      "learning_rate": 9.495253976398153e-06,
      "loss": 0.0065,
      "step": 29610
    },
    {
      "epoch": 0.18997347939441944,
      "grad_norm": 0.26507431268692017,
      "learning_rate": 9.498460749102104e-06,
      "loss": 0.0077,
      "step": 29620
    },
    {
      "epoch": 0.19003761628820554,
      "grad_norm": 0.32735857367515564,
      "learning_rate": 9.501667521806055e-06,
      "loss": 0.0063,
      "step": 29630
    },
    {
      "epoch": 0.19010175318199166,
      "grad_norm": 0.4577884376049042,
      "learning_rate": 9.504874294510007e-06,
      "loss": 0.0086,
      "step": 29640
    },
    {
      "epoch": 0.19016589007577775,
      "grad_norm": 0.48929426074028015,
      "learning_rate": 9.508081067213956e-06,
      "loss": 0.0059,
      "step": 29650
    },
    {
      "epoch": 0.19023002696956384,
      "grad_norm": 0.3111748993396759,
      "learning_rate": 9.511287839917907e-06,
      "loss": 0.0054,
      "step": 29660
    },
    {
      "epoch": 0.19029416386334994,
      "grad_norm": 0.3823695778846741,
      "learning_rate": 9.514494612621858e-06,
      "loss": 0.0117,
      "step": 29670
    },
    {
      "epoch": 0.19035830075713603,
      "grad_norm": 0.11964625865221024,
      "learning_rate": 9.51770138532581e-06,
      "loss": 0.0058,
      "step": 29680
    },
    {
      "epoch": 0.19042243765092212,
      "grad_norm": 0.16966375708580017,
      "learning_rate": 9.520908158029759e-06,
      "loss": 0.0085,
      "step": 29690
    },
    {
      "epoch": 0.19048657454470821,
      "grad_norm": 0.5360361933708191,
      "learning_rate": 9.52411493073371e-06,
      "loss": 0.0117,
      "step": 29700
    },
    {
      "epoch": 0.19055071143849434,
      "grad_norm": 0.3585972785949707,
      "learning_rate": 9.527321703437661e-06,
      "loss": 0.0061,
      "step": 29710
    },
    {
      "epoch": 0.19061484833228043,
      "grad_norm": 0.15603461861610413,
      "learning_rate": 9.530528476141612e-06,
      "loss": 0.0041,
      "step": 29720
    },
    {
      "epoch": 0.19067898522606652,
      "grad_norm": 0.3922441601753235,
      "learning_rate": 9.533735248845563e-06,
      "loss": 0.005,
      "step": 29730
    },
    {
      "epoch": 0.19074312211985262,
      "grad_norm": 0.2955012619495392,
      "learning_rate": 9.536942021549513e-06,
      "loss": 0.0053,
      "step": 29740
    },
    {
      "epoch": 0.1908072590136387,
      "grad_norm": 0.24606548249721527,
      "learning_rate": 9.540148794253464e-06,
      "loss": 0.0049,
      "step": 29750
    },
    {
      "epoch": 0.1908713959074248,
      "grad_norm": 0.4470388889312744,
      "learning_rate": 9.543355566957415e-06,
      "loss": 0.007,
      "step": 29760
    },
    {
      "epoch": 0.1909355328012109,
      "grad_norm": 0.30352315306663513,
      "learning_rate": 9.546562339661366e-06,
      "loss": 0.0083,
      "step": 29770
    },
    {
      "epoch": 0.19099966969499702,
      "grad_norm": 0.439728707075119,
      "learning_rate": 9.549769112365315e-06,
      "loss": 0.0051,
      "step": 29780
    },
    {
      "epoch": 0.1910638065887831,
      "grad_norm": 0.4757135510444641,
      "learning_rate": 9.552975885069266e-06,
      "loss": 0.0056,
      "step": 29790
    },
    {
      "epoch": 0.1911279434825692,
      "grad_norm": 0.37213197350502014,
      "learning_rate": 9.556182657773218e-06,
      "loss": 0.0047,
      "step": 29800
    },
    {
      "epoch": 0.1911920803763553,
      "grad_norm": 0.4364951252937317,
      "learning_rate": 9.559389430477169e-06,
      "loss": 0.0091,
      "step": 29810
    },
    {
      "epoch": 0.1912562172701414,
      "grad_norm": 0.3021470606327057,
      "learning_rate": 9.56259620318112e-06,
      "loss": 0.0078,
      "step": 29820
    },
    {
      "epoch": 0.19132035416392748,
      "grad_norm": 0.25977569818496704,
      "learning_rate": 9.565802975885071e-06,
      "loss": 0.0091,
      "step": 29830
    },
    {
      "epoch": 0.19138449105771357,
      "grad_norm": 0.3275696039199829,
      "learning_rate": 9.56900974858902e-06,
      "loss": 0.0061,
      "step": 29840
    },
    {
      "epoch": 0.1914486279514997,
      "grad_norm": 0.3866559565067291,
      "learning_rate": 9.572216521292971e-06,
      "loss": 0.0061,
      "step": 29850
    },
    {
      "epoch": 0.1915127648452858,
      "grad_norm": 0.3434855341911316,
      "learning_rate": 9.575423293996923e-06,
      "loss": 0.0062,
      "step": 29860
    },
    {
      "epoch": 0.19157690173907188,
      "grad_norm": 0.3796941637992859,
      "learning_rate": 9.578630066700872e-06,
      "loss": 0.011,
      "step": 29870
    },
    {
      "epoch": 0.19164103863285797,
      "grad_norm": 0.6490646600723267,
      "learning_rate": 9.581836839404823e-06,
      "loss": 0.0081,
      "step": 29880
    },
    {
      "epoch": 0.19170517552664407,
      "grad_norm": 0.2909601330757141,
      "learning_rate": 9.585043612108774e-06,
      "loss": 0.0072,
      "step": 29890
    },
    {
      "epoch": 0.19176931242043016,
      "grad_norm": 0.44852450489997864,
      "learning_rate": 9.588250384812725e-06,
      "loss": 0.0079,
      "step": 29900
    },
    {
      "epoch": 0.19183344931421625,
      "grad_norm": 0.6037708520889282,
      "learning_rate": 9.591457157516676e-06,
      "loss": 0.007,
      "step": 29910
    },
    {
      "epoch": 0.19189758620800235,
      "grad_norm": 0.20274221897125244,
      "learning_rate": 9.594663930220628e-06,
      "loss": 0.007,
      "step": 29920
    },
    {
      "epoch": 0.19196172310178847,
      "grad_norm": 0.2396126240491867,
      "learning_rate": 9.597870702924577e-06,
      "loss": 0.0062,
      "step": 29930
    },
    {
      "epoch": 0.19202585999557456,
      "grad_norm": 0.50922691822052,
      "learning_rate": 9.601077475628528e-06,
      "loss": 0.0059,
      "step": 29940
    },
    {
      "epoch": 0.19208999688936065,
      "grad_norm": 0.5943179130554199,
      "learning_rate": 9.60428424833248e-06,
      "loss": 0.0086,
      "step": 29950
    },
    {
      "epoch": 0.19215413378314675,
      "grad_norm": 0.11651797592639923,
      "learning_rate": 9.60749102103643e-06,
      "loss": 0.0064,
      "step": 29960
    },
    {
      "epoch": 0.19221827067693284,
      "grad_norm": 0.3450041711330414,
      "learning_rate": 9.61069779374038e-06,
      "loss": 0.0077,
      "step": 29970
    },
    {
      "epoch": 0.19228240757071893,
      "grad_norm": 0.3452431261539459,
      "learning_rate": 9.61390456644433e-06,
      "loss": 0.0077,
      "step": 29980
    },
    {
      "epoch": 0.19234654446450503,
      "grad_norm": 0.5599872469902039,
      "learning_rate": 9.617111339148282e-06,
      "loss": 0.0062,
      "step": 29990
    },
    {
      "epoch": 0.19241068135829115,
      "grad_norm": 0.11537335813045502,
      "learning_rate": 9.620318111852233e-06,
      "loss": 0.0053,
      "step": 30000
    },
    {
      "epoch": 0.19247481825207724,
      "grad_norm": 0.2817453444004059,
      "learning_rate": 9.623524884556184e-06,
      "loss": 0.0092,
      "step": 30010
    },
    {
      "epoch": 0.19253895514586333,
      "grad_norm": 0.37333157658576965,
      "learning_rate": 9.626731657260134e-06,
      "loss": 0.0106,
      "step": 30020
    },
    {
      "epoch": 0.19260309203964943,
      "grad_norm": 0.45478856563568115,
      "learning_rate": 9.629938429964085e-06,
      "loss": 0.0055,
      "step": 30030
    },
    {
      "epoch": 0.19266722893343552,
      "grad_norm": 0.2714729905128479,
      "learning_rate": 9.633145202668036e-06,
      "loss": 0.0063,
      "step": 30040
    },
    {
      "epoch": 0.1927313658272216,
      "grad_norm": 0.2567165791988373,
      "learning_rate": 9.636351975371987e-06,
      "loss": 0.0088,
      "step": 30050
    },
    {
      "epoch": 0.1927955027210077,
      "grad_norm": 0.24161703884601593,
      "learning_rate": 9.639558748075936e-06,
      "loss": 0.0078,
      "step": 30060
    },
    {
      "epoch": 0.19285963961479383,
      "grad_norm": 0.5712881088256836,
      "learning_rate": 9.642765520779887e-06,
      "loss": 0.0061,
      "step": 30070
    },
    {
      "epoch": 0.19292377650857992,
      "grad_norm": 0.4481494426727295,
      "learning_rate": 9.645972293483839e-06,
      "loss": 0.0061,
      "step": 30080
    },
    {
      "epoch": 0.192987913402366,
      "grad_norm": 0.41341614723205566,
      "learning_rate": 9.64917906618779e-06,
      "loss": 0.006,
      "step": 30090
    },
    {
      "epoch": 0.1930520502961521,
      "grad_norm": 0.5246740579605103,
      "learning_rate": 9.65238583889174e-06,
      "loss": 0.0068,
      "step": 30100
    },
    {
      "epoch": 0.1931161871899382,
      "grad_norm": 0.271650493144989,
      "learning_rate": 9.65559261159569e-06,
      "loss": 0.0087,
      "step": 30110
    },
    {
      "epoch": 0.1931803240837243,
      "grad_norm": 0.32311326265335083,
      "learning_rate": 9.658799384299641e-06,
      "loss": 0.0053,
      "step": 30120
    },
    {
      "epoch": 0.19324446097751038,
      "grad_norm": 0.48358315229415894,
      "learning_rate": 9.662006157003592e-06,
      "loss": 0.0072,
      "step": 30130
    },
    {
      "epoch": 0.1933085978712965,
      "grad_norm": 0.2274123579263687,
      "learning_rate": 9.665212929707544e-06,
      "loss": 0.0061,
      "step": 30140
    },
    {
      "epoch": 0.1933727347650826,
      "grad_norm": 0.6029508709907532,
      "learning_rate": 9.668419702411493e-06,
      "loss": 0.0095,
      "step": 30150
    },
    {
      "epoch": 0.1934368716588687,
      "grad_norm": 0.6068460941314697,
      "learning_rate": 9.671626475115444e-06,
      "loss": 0.0048,
      "step": 30160
    },
    {
      "epoch": 0.19350100855265479,
      "grad_norm": 0.16866932809352875,
      "learning_rate": 9.674833247819395e-06,
      "loss": 0.0098,
      "step": 30170
    },
    {
      "epoch": 0.19356514544644088,
      "grad_norm": 0.5950133204460144,
      "learning_rate": 9.678040020523346e-06,
      "loss": 0.0102,
      "step": 30180
    },
    {
      "epoch": 0.19362928234022697,
      "grad_norm": 0.18013131618499756,
      "learning_rate": 9.681246793227297e-06,
      "loss": 0.0121,
      "step": 30190
    },
    {
      "epoch": 0.19369341923401306,
      "grad_norm": 0.5216000080108643,
      "learning_rate": 9.684453565931249e-06,
      "loss": 0.0082,
      "step": 30200
    },
    {
      "epoch": 0.19375755612779919,
      "grad_norm": 0.22303879261016846,
      "learning_rate": 9.687660338635198e-06,
      "loss": 0.0071,
      "step": 30210
    },
    {
      "epoch": 0.19382169302158528,
      "grad_norm": 0.3154642581939697,
      "learning_rate": 9.690867111339149e-06,
      "loss": 0.0093,
      "step": 30220
    },
    {
      "epoch": 0.19388582991537137,
      "grad_norm": 0.24289870262145996,
      "learning_rate": 9.6940738840431e-06,
      "loss": 0.006,
      "step": 30230
    },
    {
      "epoch": 0.19394996680915746,
      "grad_norm": 0.32565346360206604,
      "learning_rate": 9.69728065674705e-06,
      "loss": 0.0088,
      "step": 30240
    },
    {
      "epoch": 0.19401410370294356,
      "grad_norm": 0.2683010399341583,
      "learning_rate": 9.700487429451e-06,
      "loss": 0.005,
      "step": 30250
    },
    {
      "epoch": 0.19407824059672965,
      "grad_norm": 0.26952335238456726,
      "learning_rate": 9.703694202154952e-06,
      "loss": 0.0058,
      "step": 30260
    },
    {
      "epoch": 0.19414237749051574,
      "grad_norm": 0.25967660546302795,
      "learning_rate": 9.706900974858903e-06,
      "loss": 0.0062,
      "step": 30270
    },
    {
      "epoch": 0.19420651438430186,
      "grad_norm": 0.15537609159946442,
      "learning_rate": 9.710107747562854e-06,
      "loss": 0.005,
      "step": 30280
    },
    {
      "epoch": 0.19427065127808796,
      "grad_norm": 0.4608062505722046,
      "learning_rate": 9.713314520266805e-06,
      "loss": 0.0074,
      "step": 30290
    },
    {
      "epoch": 0.19433478817187405,
      "grad_norm": 0.4129866063594818,
      "learning_rate": 9.716521292970755e-06,
      "loss": 0.0102,
      "step": 30300
    },
    {
      "epoch": 0.19439892506566014,
      "grad_norm": 0.5129449367523193,
      "learning_rate": 9.719728065674706e-06,
      "loss": 0.0077,
      "step": 30310
    },
    {
      "epoch": 0.19446306195944624,
      "grad_norm": 0.33677932620048523,
      "learning_rate": 9.722934838378657e-06,
      "loss": 0.0083,
      "step": 30320
    },
    {
      "epoch": 0.19452719885323233,
      "grad_norm": 0.8424015641212463,
      "learning_rate": 9.726141611082606e-06,
      "loss": 0.0074,
      "step": 30330
    },
    {
      "epoch": 0.19459133574701842,
      "grad_norm": 0.29892176389694214,
      "learning_rate": 9.729348383786557e-06,
      "loss": 0.0074,
      "step": 30340
    },
    {
      "epoch": 0.19465547264080454,
      "grad_norm": 0.2060842663049698,
      "learning_rate": 9.732555156490508e-06,
      "loss": 0.007,
      "step": 30350
    },
    {
      "epoch": 0.19471960953459064,
      "grad_norm": 1.0599238872528076,
      "learning_rate": 9.73576192919446e-06,
      "loss": 0.0091,
      "step": 30360
    },
    {
      "epoch": 0.19478374642837673,
      "grad_norm": 0.1715250313282013,
      "learning_rate": 9.73896870189841e-06,
      "loss": 0.0114,
      "step": 30370
    },
    {
      "epoch": 0.19484788332216282,
      "grad_norm": 0.2989075481891632,
      "learning_rate": 9.742175474602362e-06,
      "loss": 0.0059,
      "step": 30380
    },
    {
      "epoch": 0.19491202021594892,
      "grad_norm": 0.34631791710853577,
      "learning_rate": 9.745382247306311e-06,
      "loss": 0.007,
      "step": 30390
    },
    {
      "epoch": 0.194976157109735,
      "grad_norm": 0.15520182251930237,
      "learning_rate": 9.748589020010262e-06,
      "loss": 0.005,
      "step": 30400
    },
    {
      "epoch": 0.1950402940035211,
      "grad_norm": 0.6479896306991577,
      "learning_rate": 9.751795792714213e-06,
      "loss": 0.0051,
      "step": 30410
    },
    {
      "epoch": 0.19510443089730722,
      "grad_norm": 0.38978007435798645,
      "learning_rate": 9.755002565418163e-06,
      "loss": 0.0062,
      "step": 30420
    },
    {
      "epoch": 0.19516856779109332,
      "grad_norm": 0.29684966802597046,
      "learning_rate": 9.758209338122114e-06,
      "loss": 0.0044,
      "step": 30430
    },
    {
      "epoch": 0.1952327046848794,
      "grad_norm": 0.61577969789505,
      "learning_rate": 9.761416110826065e-06,
      "loss": 0.008,
      "step": 30440
    },
    {
      "epoch": 0.1952968415786655,
      "grad_norm": 0.2743522822856903,
      "learning_rate": 9.764622883530016e-06,
      "loss": 0.0061,
      "step": 30450
    },
    {
      "epoch": 0.1953609784724516,
      "grad_norm": 0.10397171974182129,
      "learning_rate": 9.767829656233967e-06,
      "loss": 0.0081,
      "step": 30460
    },
    {
      "epoch": 0.1954251153662377,
      "grad_norm": 0.2834140956401825,
      "learning_rate": 9.771036428937918e-06,
      "loss": 0.01,
      "step": 30470
    },
    {
      "epoch": 0.19548925226002378,
      "grad_norm": 0.191796213388443,
      "learning_rate": 9.774243201641868e-06,
      "loss": 0.0069,
      "step": 30480
    },
    {
      "epoch": 0.1955533891538099,
      "grad_norm": 0.1656953990459442,
      "learning_rate": 9.777449974345819e-06,
      "loss": 0.0064,
      "step": 30490
    },
    {
      "epoch": 0.195617526047596,
      "grad_norm": 0.2885013222694397,
      "learning_rate": 9.78065674704977e-06,
      "loss": 0.0094,
      "step": 30500
    },
    {
      "epoch": 0.1956816629413821,
      "grad_norm": 0.41713154315948486,
      "learning_rate": 9.78386351975372e-06,
      "loss": 0.0071,
      "step": 30510
    },
    {
      "epoch": 0.19574579983516818,
      "grad_norm": 0.3589705228805542,
      "learning_rate": 9.78707029245767e-06,
      "loss": 0.0098,
      "step": 30520
    },
    {
      "epoch": 0.19580993672895428,
      "grad_norm": 0.19352507591247559,
      "learning_rate": 9.790277065161622e-06,
      "loss": 0.0074,
      "step": 30530
    },
    {
      "epoch": 0.19587407362274037,
      "grad_norm": 0.8712581396102905,
      "learning_rate": 9.793483837865573e-06,
      "loss": 0.009,
      "step": 30540
    },
    {
      "epoch": 0.19593821051652646,
      "grad_norm": 0.21297553181648254,
      "learning_rate": 9.796690610569524e-06,
      "loss": 0.0128,
      "step": 30550
    },
    {
      "epoch": 0.19600234741031258,
      "grad_norm": 0.40038996934890747,
      "learning_rate": 9.799897383273475e-06,
      "loss": 0.0078,
      "step": 30560
    },
    {
      "epoch": 0.19606648430409868,
      "grad_norm": 0.4325382113456726,
      "learning_rate": 9.803104155977426e-06,
      "loss": 0.0099,
      "step": 30570
    },
    {
      "epoch": 0.19613062119788477,
      "grad_norm": 0.19598670303821564,
      "learning_rate": 9.806310928681376e-06,
      "loss": 0.01,
      "step": 30580
    },
    {
      "epoch": 0.19619475809167086,
      "grad_norm": 0.27098730206489563,
      "learning_rate": 9.809517701385327e-06,
      "loss": 0.0053,
      "step": 30590
    },
    {
      "epoch": 0.19625889498545696,
      "grad_norm": 0.1887778639793396,
      "learning_rate": 9.812724474089276e-06,
      "loss": 0.0066,
      "step": 30600
    },
    {
      "epoch": 0.19632303187924305,
      "grad_norm": 0.4308469593524933,
      "learning_rate": 9.815931246793227e-06,
      "loss": 0.0082,
      "step": 30610
    },
    {
      "epoch": 0.19638716877302914,
      "grad_norm": 0.2181924432516098,
      "learning_rate": 9.819138019497178e-06,
      "loss": 0.0076,
      "step": 30620
    },
    {
      "epoch": 0.19645130566681526,
      "grad_norm": 0.38740473985671997,
      "learning_rate": 9.82234479220113e-06,
      "loss": 0.0061,
      "step": 30630
    },
    {
      "epoch": 0.19651544256060136,
      "grad_norm": 0.36458855867385864,
      "learning_rate": 9.82555156490508e-06,
      "loss": 0.0055,
      "step": 30640
    },
    {
      "epoch": 0.19657957945438745,
      "grad_norm": 0.21326607465744019,
      "learning_rate": 9.828758337609032e-06,
      "loss": 0.0071,
      "step": 30650
    },
    {
      "epoch": 0.19664371634817354,
      "grad_norm": 0.2693590819835663,
      "learning_rate": 9.831965110312983e-06,
      "loss": 0.0074,
      "step": 30660
    },
    {
      "epoch": 0.19670785324195963,
      "grad_norm": 0.3972683250904083,
      "learning_rate": 9.835171883016932e-06,
      "loss": 0.0054,
      "step": 30670
    },
    {
      "epoch": 0.19677199013574573,
      "grad_norm": 0.23552118241786957,
      "learning_rate": 9.838378655720883e-06,
      "loss": 0.0067,
      "step": 30680
    },
    {
      "epoch": 0.19683612702953182,
      "grad_norm": 0.18564055860042572,
      "learning_rate": 9.841585428424833e-06,
      "loss": 0.0116,
      "step": 30690
    },
    {
      "epoch": 0.19690026392331794,
      "grad_norm": 0.2473290115594864,
      "learning_rate": 9.844792201128784e-06,
      "loss": 0.0068,
      "step": 30700
    },
    {
      "epoch": 0.19696440081710404,
      "grad_norm": 0.13441959023475647,
      "learning_rate": 9.847998973832735e-06,
      "loss": 0.0057,
      "step": 30710
    },
    {
      "epoch": 0.19702853771089013,
      "grad_norm": 0.24626748263835907,
      "learning_rate": 9.851205746536686e-06,
      "loss": 0.0061,
      "step": 30720
    },
    {
      "epoch": 0.19709267460467622,
      "grad_norm": 0.2407292276620865,
      "learning_rate": 9.854412519240637e-06,
      "loss": 0.0091,
      "step": 30730
    },
    {
      "epoch": 0.19715681149846231,
      "grad_norm": 0.31166961789131165,
      "learning_rate": 9.857619291944588e-06,
      "loss": 0.0081,
      "step": 30740
    },
    {
      "epoch": 0.1972209483922484,
      "grad_norm": 0.18418951332569122,
      "learning_rate": 9.86082606464854e-06,
      "loss": 0.0063,
      "step": 30750
    },
    {
      "epoch": 0.1972850852860345,
      "grad_norm": 0.5213150382041931,
      "learning_rate": 9.864032837352489e-06,
      "loss": 0.0039,
      "step": 30760
    },
    {
      "epoch": 0.19734922217982062,
      "grad_norm": 0.657940685749054,
      "learning_rate": 9.86723961005644e-06,
      "loss": 0.0064,
      "step": 30770
    },
    {
      "epoch": 0.19741335907360671,
      "grad_norm": 0.24551035463809967,
      "learning_rate": 9.87044638276039e-06,
      "loss": 0.0088,
      "step": 30780
    },
    {
      "epoch": 0.1974774959673928,
      "grad_norm": 0.31944477558135986,
      "learning_rate": 9.87365315546434e-06,
      "loss": 0.0087,
      "step": 30790
    },
    {
      "epoch": 0.1975416328611789,
      "grad_norm": 0.6363940834999084,
      "learning_rate": 9.876859928168292e-06,
      "loss": 0.0073,
      "step": 30800
    },
    {
      "epoch": 0.197605769754965,
      "grad_norm": 0.37719208002090454,
      "learning_rate": 9.880066700872243e-06,
      "loss": 0.0061,
      "step": 30810
    },
    {
      "epoch": 0.1976699066487511,
      "grad_norm": 0.09619004279375076,
      "learning_rate": 9.883273473576194e-06,
      "loss": 0.007,
      "step": 30820
    },
    {
      "epoch": 0.19773404354253718,
      "grad_norm": 0.48498639464378357,
      "learning_rate": 9.886480246280145e-06,
      "loss": 0.0068,
      "step": 30830
    },
    {
      "epoch": 0.1977981804363233,
      "grad_norm": 0.96433025598526,
      "learning_rate": 9.889687018984096e-06,
      "loss": 0.0084,
      "step": 30840
    },
    {
      "epoch": 0.1978623173301094,
      "grad_norm": 0.48657941818237305,
      "learning_rate": 9.892893791688045e-06,
      "loss": 0.0089,
      "step": 30850
    },
    {
      "epoch": 0.1979264542238955,
      "grad_norm": 0.5844976902008057,
      "learning_rate": 9.896100564391997e-06,
      "loss": 0.007,
      "step": 30860
    },
    {
      "epoch": 0.19799059111768158,
      "grad_norm": 0.34406018257141113,
      "learning_rate": 9.899307337095948e-06,
      "loss": 0.0055,
      "step": 30870
    },
    {
      "epoch": 0.19805472801146767,
      "grad_norm": 0.481023907661438,
      "learning_rate": 9.902514109799897e-06,
      "loss": 0.01,
      "step": 30880
    },
    {
      "epoch": 0.19811886490525377,
      "grad_norm": 0.11957148462533951,
      "learning_rate": 9.905720882503848e-06,
      "loss": 0.0095,
      "step": 30890
    },
    {
      "epoch": 0.19818300179903986,
      "grad_norm": 0.5999955534934998,
      "learning_rate": 9.9089276552078e-06,
      "loss": 0.0078,
      "step": 30900
    },
    {
      "epoch": 0.19824713869282598,
      "grad_norm": 0.3113568127155304,
      "learning_rate": 9.91213442791175e-06,
      "loss": 0.0061,
      "step": 30910
    },
    {
      "epoch": 0.19831127558661207,
      "grad_norm": 0.39044150710105896,
      "learning_rate": 9.915341200615702e-06,
      "loss": 0.0054,
      "step": 30920
    },
    {
      "epoch": 0.19837541248039817,
      "grad_norm": 0.5722967386245728,
      "learning_rate": 9.918547973319653e-06,
      "loss": 0.009,
      "step": 30930
    },
    {
      "epoch": 0.19843954937418426,
      "grad_norm": 0.35687652230262756,
      "learning_rate": 9.921754746023604e-06,
      "loss": 0.0052,
      "step": 30940
    },
    {
      "epoch": 0.19850368626797035,
      "grad_norm": 0.5795350074768066,
      "learning_rate": 9.924961518727553e-06,
      "loss": 0.0059,
      "step": 30950
    },
    {
      "epoch": 0.19856782316175645,
      "grad_norm": 0.4220397174358368,
      "learning_rate": 9.928168291431504e-06,
      "loss": 0.0104,
      "step": 30960
    },
    {
      "epoch": 0.19863196005554254,
      "grad_norm": 0.1680145561695099,
      "learning_rate": 9.931375064135454e-06,
      "loss": 0.0064,
      "step": 30970
    },
    {
      "epoch": 0.19869609694932866,
      "grad_norm": 0.4890241026878357,
      "learning_rate": 9.934581836839405e-06,
      "loss": 0.0074,
      "step": 30980
    },
    {
      "epoch": 0.19876023384311475,
      "grad_norm": 0.1341424286365509,
      "learning_rate": 9.937788609543356e-06,
      "loss": 0.0068,
      "step": 30990
    },
    {
      "epoch": 0.19882437073690085,
      "grad_norm": 0.29779118299484253,
      "learning_rate": 9.940995382247307e-06,
      "loss": 0.0042,
      "step": 31000
    },
    {
      "epoch": 0.19888850763068694,
      "grad_norm": 0.1636631190776825,
      "learning_rate": 9.944202154951258e-06,
      "loss": 0.0062,
      "step": 31010
    },
    {
      "epoch": 0.19895264452447303,
      "grad_norm": 0.18600665032863617,
      "learning_rate": 9.94740892765521e-06,
      "loss": 0.0055,
      "step": 31020
    },
    {
      "epoch": 0.19901678141825913,
      "grad_norm": 0.5404052734375,
      "learning_rate": 9.95061570035916e-06,
      "loss": 0.0066,
      "step": 31030
    },
    {
      "epoch": 0.19908091831204522,
      "grad_norm": 0.12097226083278656,
      "learning_rate": 9.95382247306311e-06,
      "loss": 0.0035,
      "step": 31040
    },
    {
      "epoch": 0.19914505520583134,
      "grad_norm": 0.14039276540279388,
      "learning_rate": 9.957029245767061e-06,
      "loss": 0.0076,
      "step": 31050
    },
    {
      "epoch": 0.19920919209961743,
      "grad_norm": 0.3843585252761841,
      "learning_rate": 9.96023601847101e-06,
      "loss": 0.007,
      "step": 31060
    },
    {
      "epoch": 0.19927332899340353,
      "grad_norm": 0.3972879648208618,
      "learning_rate": 9.963442791174962e-06,
      "loss": 0.0053,
      "step": 31070
    },
    {
      "epoch": 0.19933746588718962,
      "grad_norm": 0.3833004832267761,
      "learning_rate": 9.966649563878913e-06,
      "loss": 0.0063,
      "step": 31080
    },
    {
      "epoch": 0.1994016027809757,
      "grad_norm": 0.37778857350349426,
      "learning_rate": 9.969856336582864e-06,
      "loss": 0.0059,
      "step": 31090
    },
    {
      "epoch": 0.1994657396747618,
      "grad_norm": 0.3843926191329956,
      "learning_rate": 9.973063109286815e-06,
      "loss": 0.0098,
      "step": 31100
    },
    {
      "epoch": 0.1995298765685479,
      "grad_norm": 0.2783735990524292,
      "learning_rate": 9.976269881990766e-06,
      "loss": 0.0097,
      "step": 31110
    },
    {
      "epoch": 0.19959401346233402,
      "grad_norm": 0.33093342185020447,
      "learning_rate": 9.979476654694717e-06,
      "loss": 0.0062,
      "step": 31120
    },
    {
      "epoch": 0.1996581503561201,
      "grad_norm": 0.27166324853897095,
      "learning_rate": 9.982683427398667e-06,
      "loss": 0.0059,
      "step": 31130
    },
    {
      "epoch": 0.1997222872499062,
      "grad_norm": 0.48899003863334656,
      "learning_rate": 9.985890200102618e-06,
      "loss": 0.0088,
      "step": 31140
    },
    {
      "epoch": 0.1997864241436923,
      "grad_norm": 0.4287528991699219,
      "learning_rate": 9.989096972806567e-06,
      "loss": 0.0089,
      "step": 31150
    },
    {
      "epoch": 0.1998505610374784,
      "grad_norm": 0.37305253744125366,
      "learning_rate": 9.992303745510518e-06,
      "loss": 0.0071,
      "step": 31160
    },
    {
      "epoch": 0.19991469793126448,
      "grad_norm": 0.38936832547187805,
      "learning_rate": 9.99551051821447e-06,
      "loss": 0.0046,
      "step": 31170
    },
    {
      "epoch": 0.19997883482505058,
      "grad_norm": 0.03365279734134674,
      "learning_rate": 9.99871729091842e-06,
      "loss": 0.0062,
      "step": 31180
    },
    {
      "epoch": 0.2000429717188367,
      "grad_norm": 0.13471749424934387,
      "learning_rate": 9.999999988722357e-06,
      "loss": 0.0072,
      "step": 31190
    },
    {
      "epoch": 0.2001071086126228,
      "grad_norm": 0.3005850911140442,
      "learning_rate": 9.999999919803426e-06,
      "loss": 0.0045,
      "step": 31200
    },
    {
      "epoch": 0.20017124550640888,
      "grad_norm": 0.37838831543922424,
      "learning_rate": 9.999999788230924e-06,
      "loss": 0.0068,
      "step": 31210
    },
    {
      "epoch": 0.20023538240019498,
      "grad_norm": 0.27392446994781494,
      "learning_rate": 9.99999959400485e-06,
      "loss": 0.0085,
      "step": 31220
    },
    {
      "epoch": 0.20029951929398107,
      "grad_norm": 0.22249607741832733,
      "learning_rate": 9.999999337125208e-06,
      "loss": 0.0076,
      "step": 31230
    },
    {
      "epoch": 0.20036365618776716,
      "grad_norm": 0.5627263188362122,
      "learning_rate": 9.999999017592e-06,
      "loss": 0.0081,
      "step": 31240
    },
    {
      "epoch": 0.20042779308155326,
      "grad_norm": 0.32999005913734436,
      "learning_rate": 9.999998635405232e-06,
      "loss": 0.0076,
      "step": 31250
    },
    {
      "epoch": 0.20049192997533938,
      "grad_norm": 0.21928158402442932,
      "learning_rate": 9.999998190564907e-06,
      "loss": 0.0093,
      "step": 31260
    },
    {
      "epoch": 0.20055606686912547,
      "grad_norm": 0.3306523263454437,
      "learning_rate": 9.99999768307103e-06,
      "loss": 0.0075,
      "step": 31270
    },
    {
      "epoch": 0.20062020376291156,
      "grad_norm": 0.2662363648414612,
      "learning_rate": 9.999997112923611e-06,
      "loss": 0.0052,
      "step": 31280
    },
    {
      "epoch": 0.20068434065669766,
      "grad_norm": 0.2550363838672638,
      "learning_rate": 9.999996480122654e-06,
      "loss": 0.0078,
      "step": 31290
    },
    {
      "epoch": 0.20074847755048375,
      "grad_norm": 0.530947208404541,
      "learning_rate": 9.999995784668167e-06,
      "loss": 0.0079,
      "step": 31300
    },
    {
      "epoch": 0.20081261444426984,
      "grad_norm": 0.2493947297334671,
      "learning_rate": 9.99999502656016e-06,
      "loss": 0.007,
      "step": 31310
    },
    {
      "epoch": 0.20087675133805594,
      "grad_norm": 0.3358502984046936,
      "learning_rate": 9.999994205798643e-06,
      "loss": 0.0078,
      "step": 31320
    },
    {
      "epoch": 0.20094088823184206,
      "grad_norm": 0.2657833993434906,
      "learning_rate": 9.999993322383621e-06,
      "loss": 0.0082,
      "step": 31330
    },
    {
      "epoch": 0.20100502512562815,
      "grad_norm": 0.9135074019432068,
      "learning_rate": 9.999992376315113e-06,
      "loss": 0.0065,
      "step": 31340
    },
    {
      "epoch": 0.20106916201941424,
      "grad_norm": 0.47939422726631165,
      "learning_rate": 9.999991367593127e-06,
      "loss": 0.0084,
      "step": 31350
    },
    {
      "epoch": 0.20113329891320034,
      "grad_norm": 0.21267169713974,
      "learning_rate": 9.999990296217674e-06,
      "loss": 0.0048,
      "step": 31360
    },
    {
      "epoch": 0.20119743580698643,
      "grad_norm": 0.2891674339771271,
      "learning_rate": 9.99998916218877e-06,
      "loss": 0.0112,
      "step": 31370
    },
    {
      "epoch": 0.20126157270077252,
      "grad_norm": 0.33775970339775085,
      "learning_rate": 9.999987965506428e-06,
      "loss": 0.0078,
      "step": 31380
    },
    {
      "epoch": 0.20132570959455862,
      "grad_norm": 0.2934929430484772,
      "learning_rate": 9.999986706170664e-06,
      "loss": 0.0063,
      "step": 31390
    },
    {
      "epoch": 0.2013898464883447,
      "grad_norm": 0.3903428912162781,
      "learning_rate": 9.999985384181492e-06,
      "loss": 0.0081,
      "step": 31400
    },
    {
      "epoch": 0.20145398338213083,
      "grad_norm": 0.3547932207584381,
      "learning_rate": 9.999983999538932e-06,
      "loss": 0.0086,
      "step": 31410
    },
    {
      "epoch": 0.20151812027591692,
      "grad_norm": 0.5304170846939087,
      "learning_rate": 9.999982552242998e-06,
      "loss": 0.0068,
      "step": 31420
    },
    {
      "epoch": 0.20158225716970302,
      "grad_norm": 0.6028372645378113,
      "learning_rate": 9.999981042293707e-06,
      "loss": 0.008,
      "step": 31430
    },
    {
      "epoch": 0.2016463940634891,
      "grad_norm": 0.06574447453022003,
      "learning_rate": 9.999979469691082e-06,
      "loss": 0.0045,
      "step": 31440
    },
    {
      "epoch": 0.2017105309572752,
      "grad_norm": 0.35700806975364685,
      "learning_rate": 9.999977834435141e-06,
      "loss": 0.0063,
      "step": 31450
    },
    {
      "epoch": 0.2017746678510613,
      "grad_norm": 0.1741345375776291,
      "learning_rate": 9.999976136525904e-06,
      "loss": 0.0059,
      "step": 31460
    },
    {
      "epoch": 0.2018388047448474,
      "grad_norm": 0.3376474976539612,
      "learning_rate": 9.999974375963393e-06,
      "loss": 0.0071,
      "step": 31470
    },
    {
      "epoch": 0.2019029416386335,
      "grad_norm": 0.33258095383644104,
      "learning_rate": 9.999972552747629e-06,
      "loss": 0.0098,
      "step": 31480
    },
    {
      "epoch": 0.2019670785324196,
      "grad_norm": 0.5604947805404663,
      "learning_rate": 9.999970666878635e-06,
      "loss": 0.0068,
      "step": 31490
    },
    {
      "epoch": 0.2020312154262057,
      "grad_norm": 0.2540450692176819,
      "learning_rate": 9.999968718356437e-06,
      "loss": 0.0036,
      "step": 31500
    },
    {
      "epoch": 0.2020953523199918,
      "grad_norm": 0.21161718666553497,
      "learning_rate": 9.999966707181055e-06,
      "loss": 0.0088,
      "step": 31510
    },
    {
      "epoch": 0.20215948921377788,
      "grad_norm": 0.5596034526824951,
      "learning_rate": 9.999964633352519e-06,
      "loss": 0.0082,
      "step": 31520
    },
    {
      "epoch": 0.20222362610756398,
      "grad_norm": 0.3756974935531616,
      "learning_rate": 9.999962496870852e-06,
      "loss": 0.0276,
      "step": 31530
    },
    {
      "epoch": 0.20228776300135007,
      "grad_norm": 0.3303465247154236,
      "learning_rate": 9.99996029773608e-06,
      "loss": 0.0067,
      "step": 31540
    },
    {
      "epoch": 0.2023518998951362,
      "grad_norm": 0.07992622256278992,
      "learning_rate": 9.999958035948233e-06,
      "loss": 0.0086,
      "step": 31550
    },
    {
      "epoch": 0.20241603678892228,
      "grad_norm": 0.4249248802661896,
      "learning_rate": 9.999955711507338e-06,
      "loss": 0.005,
      "step": 31560
    },
    {
      "epoch": 0.20248017368270838,
      "grad_norm": 0.20567820966243744,
      "learning_rate": 9.999953324413428e-06,
      "loss": 0.004,
      "step": 31570
    },
    {
      "epoch": 0.20254431057649447,
      "grad_norm": 0.8298195600509644,
      "learning_rate": 9.999950874666525e-06,
      "loss": 0.0086,
      "step": 31580
    },
    {
      "epoch": 0.20260844747028056,
      "grad_norm": 0.2139284908771515,
      "learning_rate": 9.999948362266666e-06,
      "loss": 0.008,
      "step": 31590
    },
    {
      "epoch": 0.20267258436406665,
      "grad_norm": 0.49156635999679565,
      "learning_rate": 9.99994578721388e-06,
      "loss": 0.006,
      "step": 31600
    },
    {
      "epoch": 0.20273672125785275,
      "grad_norm": 0.177789106965065,
      "learning_rate": 9.999943149508201e-06,
      "loss": 0.005,
      "step": 31610
    },
    {
      "epoch": 0.20280085815163887,
      "grad_norm": 0.4381600618362427,
      "learning_rate": 9.999940449149659e-06,
      "loss": 0.0076,
      "step": 31620
    },
    {
      "epoch": 0.20286499504542496,
      "grad_norm": 0.242206871509552,
      "learning_rate": 9.999937686138292e-06,
      "loss": 0.0052,
      "step": 31630
    },
    {
      "epoch": 0.20292913193921105,
      "grad_norm": 0.6824889779090881,
      "learning_rate": 9.999934860474132e-06,
      "loss": 0.0093,
      "step": 31640
    },
    {
      "epoch": 0.20299326883299715,
      "grad_norm": 0.524763286113739,
      "learning_rate": 9.999931972157214e-06,
      "loss": 0.0063,
      "step": 31650
    },
    {
      "epoch": 0.20305740572678324,
      "grad_norm": 0.39510250091552734,
      "learning_rate": 9.999929021187575e-06,
      "loss": 0.0048,
      "step": 31660
    },
    {
      "epoch": 0.20312154262056933,
      "grad_norm": 0.3214946985244751,
      "learning_rate": 9.999926007565253e-06,
      "loss": 0.0096,
      "step": 31670
    },
    {
      "epoch": 0.20318567951435543,
      "grad_norm": 0.38387608528137207,
      "learning_rate": 9.999922931290282e-06,
      "loss": 0.0054,
      "step": 31680
    },
    {
      "epoch": 0.20324981640814155,
      "grad_norm": 0.4253934323787689,
      "learning_rate": 9.999919792362708e-06,
      "loss": 0.0068,
      "step": 31690
    },
    {
      "epoch": 0.20331395330192764,
      "grad_norm": 0.2738664150238037,
      "learning_rate": 9.999916590782561e-06,
      "loss": 0.0054,
      "step": 31700
    },
    {
      "epoch": 0.20337809019571373,
      "grad_norm": 0.2658179700374603,
      "learning_rate": 9.999913326549888e-06,
      "loss": 0.0054,
      "step": 31710
    },
    {
      "epoch": 0.20344222708949983,
      "grad_norm": 0.5510397553443909,
      "learning_rate": 9.999909999664727e-06,
      "loss": 0.0082,
      "step": 31720
    },
    {
      "epoch": 0.20350636398328592,
      "grad_norm": 0.17201349139213562,
      "learning_rate": 9.999906610127119e-06,
      "loss": 0.0081,
      "step": 31730
    },
    {
      "epoch": 0.203570500877072,
      "grad_norm": 0.4930127263069153,
      "learning_rate": 9.99990315793711e-06,
      "loss": 0.0082,
      "step": 31740
    },
    {
      "epoch": 0.2036346377708581,
      "grad_norm": 0.23746931552886963,
      "learning_rate": 9.99989964309474e-06,
      "loss": 0.0068,
      "step": 31750
    },
    {
      "epoch": 0.20369877466464423,
      "grad_norm": 0.27894327044487,
      "learning_rate": 9.999896065600054e-06,
      "loss": 0.0062,
      "step": 31760
    },
    {
      "epoch": 0.20376291155843032,
      "grad_norm": 0.28237441182136536,
      "learning_rate": 9.999892425453098e-06,
      "loss": 0.0094,
      "step": 31770
    },
    {
      "epoch": 0.2038270484522164,
      "grad_norm": 0.41106197237968445,
      "learning_rate": 9.999888722653917e-06,
      "loss": 0.005,
      "step": 31780
    },
    {
      "epoch": 0.2038911853460025,
      "grad_norm": 0.2211090624332428,
      "learning_rate": 9.999884957202555e-06,
      "loss": 0.0069,
      "step": 31790
    },
    {
      "epoch": 0.2039553222397886,
      "grad_norm": 0.3440389633178711,
      "learning_rate": 9.999881129099062e-06,
      "loss": 0.0104,
      "step": 31800
    },
    {
      "epoch": 0.2040194591335747,
      "grad_norm": 0.5909982323646545,
      "learning_rate": 9.999877238343485e-06,
      "loss": 0.009,
      "step": 31810
    },
    {
      "epoch": 0.2040835960273608,
      "grad_norm": 0.4488292336463928,
      "learning_rate": 9.999873284935873e-06,
      "loss": 0.0091,
      "step": 31820
    },
    {
      "epoch": 0.2041477329211469,
      "grad_norm": 0.3296765387058258,
      "learning_rate": 9.999869268876275e-06,
      "loss": 0.0045,
      "step": 31830
    },
    {
      "epoch": 0.204211869814933,
      "grad_norm": 0.4224696457386017,
      "learning_rate": 9.999865190164741e-06,
      "loss": 0.0194,
      "step": 31840
    },
    {
      "epoch": 0.2042760067087191,
      "grad_norm": 0.1356227546930313,
      "learning_rate": 9.999861048801324e-06,
      "loss": 0.008,
      "step": 31850
    },
    {
      "epoch": 0.2043401436025052,
      "grad_norm": 0.8743406534194946,
      "learning_rate": 9.999856844786076e-06,
      "loss": 0.0065,
      "step": 31860
    },
    {
      "epoch": 0.20440428049629128,
      "grad_norm": 0.707042932510376,
      "learning_rate": 9.999852578119046e-06,
      "loss": 0.0059,
      "step": 31870
    },
    {
      "epoch": 0.20446841739007737,
      "grad_norm": 0.33693328499794006,
      "learning_rate": 9.99984824880029e-06,
      "loss": 0.0051,
      "step": 31880
    },
    {
      "epoch": 0.20453255428386347,
      "grad_norm": 0.6988204717636108,
      "learning_rate": 9.999843856829862e-06,
      "loss": 0.0065,
      "step": 31890
    },
    {
      "epoch": 0.2045966911776496,
      "grad_norm": 0.6744667291641235,
      "learning_rate": 9.999839402207819e-06,
      "loss": 0.0111,
      "step": 31900
    },
    {
      "epoch": 0.20466082807143568,
      "grad_norm": 0.658195436000824,
      "learning_rate": 9.999834884934214e-06,
      "loss": 0.0048,
      "step": 31910
    },
    {
      "epoch": 0.20472496496522177,
      "grad_norm": 0.3740374743938446,
      "learning_rate": 9.999830305009104e-06,
      "loss": 0.0065,
      "step": 31920
    },
    {
      "epoch": 0.20478910185900787,
      "grad_norm": 0.36082711815834045,
      "learning_rate": 9.999825662432547e-06,
      "loss": 0.0072,
      "step": 31930
    },
    {
      "epoch": 0.20485323875279396,
      "grad_norm": 0.40542277693748474,
      "learning_rate": 9.999820957204602e-06,
      "loss": 0.0065,
      "step": 31940
    },
    {
      "epoch": 0.20491737564658005,
      "grad_norm": 0.9706642031669617,
      "learning_rate": 9.999816189325327e-06,
      "loss": 0.0106,
      "step": 31950
    },
    {
      "epoch": 0.20498151254036615,
      "grad_norm": 0.3222975730895996,
      "learning_rate": 9.999811358794781e-06,
      "loss": 0.0083,
      "step": 31960
    },
    {
      "epoch": 0.20504564943415227,
      "grad_norm": 0.11145718395709991,
      "learning_rate": 9.999806465613027e-06,
      "loss": 0.0068,
      "step": 31970
    },
    {
      "epoch": 0.20510978632793836,
      "grad_norm": 0.4002038836479187,
      "learning_rate": 9.999801509780123e-06,
      "loss": 0.0052,
      "step": 31980
    },
    {
      "epoch": 0.20517392322172445,
      "grad_norm": 0.29977932572364807,
      "learning_rate": 9.999796491296134e-06,
      "loss": 0.0053,
      "step": 31990
    },
    {
      "epoch": 0.20523806011551055,
      "grad_norm": 0.3992215692996979,
      "learning_rate": 9.999791410161121e-06,
      "loss": 0.0051,
      "step": 32000
    },
    {
      "epoch": 0.20530219700929664,
      "grad_norm": 0.24485594034194946,
      "learning_rate": 9.999786266375147e-06,
      "loss": 0.0066,
      "step": 32010
    },
    {
      "epoch": 0.20536633390308273,
      "grad_norm": 0.36206868290901184,
      "learning_rate": 9.99978105993828e-06,
      "loss": 0.009,
      "step": 32020
    },
    {
      "epoch": 0.20543047079686882,
      "grad_norm": 0.5218985080718994,
      "learning_rate": 9.999775790850582e-06,
      "loss": 0.0096,
      "step": 32030
    },
    {
      "epoch": 0.20549460769065495,
      "grad_norm": 0.1210787445306778,
      "learning_rate": 9.99977045911212e-06,
      "loss": 0.0062,
      "step": 32040
    },
    {
      "epoch": 0.20555874458444104,
      "grad_norm": 0.18466708064079285,
      "learning_rate": 9.999765064722961e-06,
      "loss": 0.0069,
      "step": 32050
    },
    {
      "epoch": 0.20562288147822713,
      "grad_norm": 0.5108962655067444,
      "learning_rate": 9.999759607683173e-06,
      "loss": 0.0075,
      "step": 32060
    },
    {
      "epoch": 0.20568701837201323,
      "grad_norm": 0.2827768623828888,
      "learning_rate": 9.999754087992823e-06,
      "loss": 0.0082,
      "step": 32070
    },
    {
      "epoch": 0.20575115526579932,
      "grad_norm": 0.5526918172836304,
      "learning_rate": 9.999748505651981e-06,
      "loss": 0.0091,
      "step": 32080
    },
    {
      "epoch": 0.2058152921595854,
      "grad_norm": 0.19573715329170227,
      "learning_rate": 9.999742860660716e-06,
      "loss": 0.0045,
      "step": 32090
    },
    {
      "epoch": 0.2058794290533715,
      "grad_norm": 0.3106071352958679,
      "learning_rate": 9.999737153019102e-06,
      "loss": 0.0105,
      "step": 32100
    },
    {
      "epoch": 0.20594356594715763,
      "grad_norm": 0.26938295364379883,
      "learning_rate": 9.999731382727207e-06,
      "loss": 0.0082,
      "step": 32110
    },
    {
      "epoch": 0.20600770284094372,
      "grad_norm": 0.3470536470413208,
      "learning_rate": 9.999725549785107e-06,
      "loss": 0.0072,
      "step": 32120
    },
    {
      "epoch": 0.2060718397347298,
      "grad_norm": 0.2686624825000763,
      "learning_rate": 9.99971965419287e-06,
      "loss": 0.0068,
      "step": 32130
    },
    {
      "epoch": 0.2061359766285159,
      "grad_norm": 0.23761259019374847,
      "learning_rate": 9.999713695950573e-06,
      "loss": 0.01,
      "step": 32140
    },
    {
      "epoch": 0.206200113522302,
      "grad_norm": 0.188786581158638,
      "learning_rate": 9.999707675058289e-06,
      "loss": 0.0047,
      "step": 32150
    },
    {
      "epoch": 0.2062642504160881,
      "grad_norm": 0.39413657784461975,
      "learning_rate": 9.999701591516095e-06,
      "loss": 0.0058,
      "step": 32160
    },
    {
      "epoch": 0.20632838730987418,
      "grad_norm": 0.6398225426673889,
      "learning_rate": 9.999695445324069e-06,
      "loss": 0.0075,
      "step": 32170
    },
    {
      "epoch": 0.2063925242036603,
      "grad_norm": 0.6331750750541687,
      "learning_rate": 9.999689236482283e-06,
      "loss": 0.0039,
      "step": 32180
    },
    {
      "epoch": 0.2064566610974464,
      "grad_norm": 0.21261058747768402,
      "learning_rate": 9.99968296499082e-06,
      "loss": 0.0084,
      "step": 32190
    },
    {
      "epoch": 0.2065207979912325,
      "grad_norm": 0.2594454884529114,
      "learning_rate": 9.999676630849756e-06,
      "loss": 0.0091,
      "step": 32200
    },
    {
      "epoch": 0.20658493488501858,
      "grad_norm": 0.3002074956893921,
      "learning_rate": 9.999670234059167e-06,
      "loss": 0.0049,
      "step": 32210
    },
    {
      "epoch": 0.20664907177880468,
      "grad_norm": 0.40826472640037537,
      "learning_rate": 9.99966377461914e-06,
      "loss": 0.0086,
      "step": 32220
    },
    {
      "epoch": 0.20671320867259077,
      "grad_norm": 0.6430691480636597,
      "learning_rate": 9.999657252529754e-06,
      "loss": 0.0072,
      "step": 32230
    },
    {
      "epoch": 0.20677734556637686,
      "grad_norm": 0.4967564642429352,
      "learning_rate": 9.999650667791087e-06,
      "loss": 0.009,
      "step": 32240
    },
    {
      "epoch": 0.20684148246016298,
      "grad_norm": 0.3127409815788269,
      "learning_rate": 9.999644020403225e-06,
      "loss": 0.0078,
      "step": 32250
    },
    {
      "epoch": 0.20690561935394908,
      "grad_norm": 0.22701984643936157,
      "learning_rate": 9.999637310366248e-06,
      "loss": 0.0066,
      "step": 32260
    },
    {
      "epoch": 0.20696975624773517,
      "grad_norm": 0.33977678418159485,
      "learning_rate": 9.999630537680245e-06,
      "loss": 0.0067,
      "step": 32270
    },
    {
      "epoch": 0.20703389314152126,
      "grad_norm": 0.260306715965271,
      "learning_rate": 9.999623702345296e-06,
      "loss": 0.0056,
      "step": 32280
    },
    {
      "epoch": 0.20709803003530736,
      "grad_norm": 0.2731505334377289,
      "learning_rate": 9.999616804361491e-06,
      "loss": 0.0046,
      "step": 32290
    },
    {
      "epoch": 0.20716216692909345,
      "grad_norm": 0.2968423664569855,
      "learning_rate": 9.999609843728914e-06,
      "loss": 0.0114,
      "step": 32300
    },
    {
      "epoch": 0.20722630382287954,
      "grad_norm": 0.2855728566646576,
      "learning_rate": 9.999602820447651e-06,
      "loss": 0.0069,
      "step": 32310
    },
    {
      "epoch": 0.20729044071666566,
      "grad_norm": 0.15290455520153046,
      "learning_rate": 9.999595734517793e-06,
      "loss": 0.0026,
      "step": 32320
    },
    {
      "epoch": 0.20735457761045176,
      "grad_norm": 0.5125821232795715,
      "learning_rate": 9.999588585939426e-06,
      "loss": 0.0051,
      "step": 32330
    },
    {
      "epoch": 0.20741871450423785,
      "grad_norm": 0.7281843423843384,
      "learning_rate": 9.99958137471264e-06,
      "loss": 0.004,
      "step": 32340
    },
    {
      "epoch": 0.20748285139802394,
      "grad_norm": 0.3147353231906891,
      "learning_rate": 9.99957410083753e-06,
      "loss": 0.0072,
      "step": 32350
    },
    {
      "epoch": 0.20754698829181004,
      "grad_norm": 0.24015134572982788,
      "learning_rate": 9.999566764314182e-06,
      "loss": 0.0053,
      "step": 32360
    },
    {
      "epoch": 0.20761112518559613,
      "grad_norm": 0.20664353668689728,
      "learning_rate": 9.999559365142688e-06,
      "loss": 0.0062,
      "step": 32370
    },
    {
      "epoch": 0.20767526207938222,
      "grad_norm": 0.24446997046470642,
      "learning_rate": 9.999551903323141e-06,
      "loss": 0.0063,
      "step": 32380
    },
    {
      "epoch": 0.20773939897316834,
      "grad_norm": 0.14070431888103485,
      "learning_rate": 9.999544378855637e-06,
      "loss": 0.0044,
      "step": 32390
    },
    {
      "epoch": 0.20780353586695444,
      "grad_norm": 0.8572035431861877,
      "learning_rate": 9.99953679174027e-06,
      "loss": 0.0085,
      "step": 32400
    },
    {
      "epoch": 0.20786767276074053,
      "grad_norm": 0.28308725357055664,
      "learning_rate": 9.999529141977132e-06,
      "loss": 0.0053,
      "step": 32410
    },
    {
      "epoch": 0.20793180965452662,
      "grad_norm": 0.06351161003112793,
      "learning_rate": 9.99952142956632e-06,
      "loss": 0.0081,
      "step": 32420
    },
    {
      "epoch": 0.20799594654831272,
      "grad_norm": 0.36633729934692383,
      "learning_rate": 9.999513654507934e-06,
      "loss": 0.0043,
      "step": 32430
    },
    {
      "epoch": 0.2080600834420988,
      "grad_norm": 0.4253617525100708,
      "learning_rate": 9.999505816802069e-06,
      "loss": 0.0048,
      "step": 32440
    },
    {
      "epoch": 0.2081242203358849,
      "grad_norm": 0.06172889843583107,
      "learning_rate": 9.99949791644882e-06,
      "loss": 0.0077,
      "step": 32450
    },
    {
      "epoch": 0.20818835722967102,
      "grad_norm": 0.18378232419490814,
      "learning_rate": 9.999489953448291e-06,
      "loss": 0.0063,
      "step": 32460
    },
    {
      "epoch": 0.20825249412345712,
      "grad_norm": 0.22072811424732208,
      "learning_rate": 9.999481927800579e-06,
      "loss": 0.0072,
      "step": 32470
    },
    {
      "epoch": 0.2083166310172432,
      "grad_norm": 0.5811721086502075,
      "learning_rate": 9.999473839505786e-06,
      "loss": 0.0079,
      "step": 32480
    },
    {
      "epoch": 0.2083807679110293,
      "grad_norm": 0.32655444741249084,
      "learning_rate": 9.999465688564014e-06,
      "loss": 0.0094,
      "step": 32490
    },
    {
      "epoch": 0.2084449048048154,
      "grad_norm": 0.08320006728172302,
      "learning_rate": 9.999457474975362e-06,
      "loss": 0.0069,
      "step": 32500
    },
    {
      "epoch": 0.2085090416986015,
      "grad_norm": 0.29463374614715576,
      "learning_rate": 9.999449198739935e-06,
      "loss": 0.0061,
      "step": 32510
    },
    {
      "epoch": 0.20857317859238758,
      "grad_norm": 0.19419896602630615,
      "learning_rate": 9.999440859857836e-06,
      "loss": 0.0049,
      "step": 32520
    },
    {
      "epoch": 0.2086373154861737,
      "grad_norm": 0.2826046645641327,
      "learning_rate": 9.99943245832917e-06,
      "loss": 0.0064,
      "step": 32530
    },
    {
      "epoch": 0.2087014523799598,
      "grad_norm": 0.45521748065948486,
      "learning_rate": 9.999423994154043e-06,
      "loss": 0.0071,
      "step": 32540
    },
    {
      "epoch": 0.2087655892737459,
      "grad_norm": 0.34207192063331604,
      "learning_rate": 9.99941546733256e-06,
      "loss": 0.0073,
      "step": 32550
    },
    {
      "epoch": 0.20882972616753198,
      "grad_norm": 0.1273985207080841,
      "learning_rate": 9.99940687786483e-06,
      "loss": 0.0059,
      "step": 32560
    },
    {
      "epoch": 0.20889386306131807,
      "grad_norm": 0.19721442461013794,
      "learning_rate": 9.999398225750956e-06,
      "loss": 0.0055,
      "step": 32570
    },
    {
      "epoch": 0.20895799995510417,
      "grad_norm": 0.3487097918987274,
      "learning_rate": 9.999389510991051e-06,
      "loss": 0.0061,
      "step": 32580
    },
    {
      "epoch": 0.20902213684889026,
      "grad_norm": 0.29255926609039307,
      "learning_rate": 9.999380733585222e-06,
      "loss": 0.0087,
      "step": 32590
    },
    {
      "epoch": 0.20908627374267638,
      "grad_norm": 0.28362610936164856,
      "learning_rate": 9.999371893533579e-06,
      "loss": 0.0052,
      "step": 32600
    },
    {
      "epoch": 0.20915041063646247,
      "grad_norm": 0.45663630962371826,
      "learning_rate": 9.999362990836233e-06,
      "loss": 0.006,
      "step": 32610
    },
    {
      "epoch": 0.20921454753024857,
      "grad_norm": 0.2847580015659332,
      "learning_rate": 9.999354025493297e-06,
      "loss": 0.0056,
      "step": 32620
    },
    {
      "epoch": 0.20927868442403466,
      "grad_norm": 0.24016176164150238,
      "learning_rate": 9.99934499750488e-06,
      "loss": 0.0071,
      "step": 32630
    },
    {
      "epoch": 0.20934282131782075,
      "grad_norm": 0.0941351130604744,
      "learning_rate": 9.999335906871099e-06,
      "loss": 0.0073,
      "step": 32640
    },
    {
      "epoch": 0.20940695821160685,
      "grad_norm": 0.4059191346168518,
      "learning_rate": 9.999326753592066e-06,
      "loss": 0.0057,
      "step": 32650
    },
    {
      "epoch": 0.20947109510539294,
      "grad_norm": 0.32373708486557007,
      "learning_rate": 9.999317537667894e-06,
      "loss": 0.0077,
      "step": 32660
    },
    {
      "epoch": 0.20953523199917906,
      "grad_norm": 0.30504995584487915,
      "learning_rate": 9.999308259098703e-06,
      "loss": 0.0077,
      "step": 32670
    },
    {
      "epoch": 0.20959936889296515,
      "grad_norm": 0.24574518203735352,
      "learning_rate": 9.999298917884606e-06,
      "loss": 0.0059,
      "step": 32680
    },
    {
      "epoch": 0.20966350578675125,
      "grad_norm": 0.23023582994937897,
      "learning_rate": 9.999289514025718e-06,
      "loss": 0.0076,
      "step": 32690
    },
    {
      "epoch": 0.20972764268053734,
      "grad_norm": 0.27018046379089355,
      "learning_rate": 9.999280047522161e-06,
      "loss": 0.0065,
      "step": 32700
    },
    {
      "epoch": 0.20979177957432343,
      "grad_norm": 0.2876282036304474,
      "learning_rate": 9.999270518374054e-06,
      "loss": 0.0117,
      "step": 32710
    },
    {
      "epoch": 0.20985591646810953,
      "grad_norm": 0.16353587806224823,
      "learning_rate": 9.999260926581513e-06,
      "loss": 0.0042,
      "step": 32720
    },
    {
      "epoch": 0.20992005336189562,
      "grad_norm": 0.35183534026145935,
      "learning_rate": 9.99925127214466e-06,
      "loss": 0.0063,
      "step": 32730
    },
    {
      "epoch": 0.20998419025568174,
      "grad_norm": 0.4123291075229645,
      "learning_rate": 9.999241555063614e-06,
      "loss": 0.008,
      "step": 32740
    },
    {
      "epoch": 0.21004832714946783,
      "grad_norm": 0.37130826711654663,
      "learning_rate": 9.999231775338499e-06,
      "loss": 0.0072,
      "step": 32750
    },
    {
      "epoch": 0.21011246404325393,
      "grad_norm": 0.16660301387310028,
      "learning_rate": 9.999221932969439e-06,
      "loss": 0.0054,
      "step": 32760
    },
    {
      "epoch": 0.21017660093704002,
      "grad_norm": 0.35891416668891907,
      "learning_rate": 9.999212027956553e-06,
      "loss": 0.0039,
      "step": 32770
    },
    {
      "epoch": 0.2102407378308261,
      "grad_norm": 0.44732221961021423,
      "learning_rate": 9.999202060299968e-06,
      "loss": 0.0138,
      "step": 32780
    },
    {
      "epoch": 0.2103048747246122,
      "grad_norm": 0.41310566663742065,
      "learning_rate": 9.999192029999809e-06,
      "loss": 0.0073,
      "step": 32790
    },
    {
      "epoch": 0.2103690116183983,
      "grad_norm": 0.33074861764907837,
      "learning_rate": 9.999181937056199e-06,
      "loss": 0.006,
      "step": 32800
    },
    {
      "epoch": 0.2104331485121844,
      "grad_norm": 0.14894649386405945,
      "learning_rate": 9.999171781469268e-06,
      "loss": 0.008,
      "step": 32810
    },
    {
      "epoch": 0.2104972854059705,
      "grad_norm": 0.1162961795926094,
      "learning_rate": 9.999161563239143e-06,
      "loss": 0.0059,
      "step": 32820
    },
    {
      "epoch": 0.2105614222997566,
      "grad_norm": 0.25399044156074524,
      "learning_rate": 9.999151282365948e-06,
      "loss": 0.0046,
      "step": 32830
    },
    {
      "epoch": 0.2106255591935427,
      "grad_norm": 0.5153788924217224,
      "learning_rate": 9.999140938849816e-06,
      "loss": 0.0093,
      "step": 32840
    },
    {
      "epoch": 0.2106896960873288,
      "grad_norm": 0.8990302681922913,
      "learning_rate": 9.999130532690876e-06,
      "loss": 0.0093,
      "step": 32850
    },
    {
      "epoch": 0.21075383298111489,
      "grad_norm": 0.33208703994750977,
      "learning_rate": 9.999120063889258e-06,
      "loss": 0.0085,
      "step": 32860
    },
    {
      "epoch": 0.21081796987490098,
      "grad_norm": 0.3936298191547394,
      "learning_rate": 9.99910953244509e-06,
      "loss": 0.0061,
      "step": 32870
    },
    {
      "epoch": 0.21088210676868707,
      "grad_norm": 0.3910579979419708,
      "learning_rate": 9.999098938358508e-06,
      "loss": 0.0081,
      "step": 32880
    },
    {
      "epoch": 0.2109462436624732,
      "grad_norm": 0.1852560192346573,
      "learning_rate": 9.999088281629645e-06,
      "loss": 0.0054,
      "step": 32890
    },
    {
      "epoch": 0.21101038055625929,
      "grad_norm": 0.3421950936317444,
      "learning_rate": 9.999077562258632e-06,
      "loss": 0.0069,
      "step": 32900
    },
    {
      "epoch": 0.21107451745004538,
      "grad_norm": 0.0992131233215332,
      "learning_rate": 9.999066780245605e-06,
      "loss": 0.0069,
      "step": 32910
    },
    {
      "epoch": 0.21113865434383147,
      "grad_norm": 0.08349548280239105,
      "learning_rate": 9.999055935590697e-06,
      "loss": 0.0054,
      "step": 32920
    },
    {
      "epoch": 0.21120279123761757,
      "grad_norm": 0.14864757657051086,
      "learning_rate": 9.999045028294045e-06,
      "loss": 0.0084,
      "step": 32930
    },
    {
      "epoch": 0.21126692813140366,
      "grad_norm": 0.19409166276454926,
      "learning_rate": 9.999034058355788e-06,
      "loss": 0.0042,
      "step": 32940
    },
    {
      "epoch": 0.21133106502518975,
      "grad_norm": 0.46400272846221924,
      "learning_rate": 9.99902302577606e-06,
      "loss": 0.0071,
      "step": 32950
    },
    {
      "epoch": 0.21139520191897587,
      "grad_norm": 0.3607134222984314,
      "learning_rate": 9.999011930555002e-06,
      "loss": 0.0061,
      "step": 32960
    },
    {
      "epoch": 0.21145933881276197,
      "grad_norm": 0.3315497934818268,
      "learning_rate": 9.99900077269275e-06,
      "loss": 0.0052,
      "step": 32970
    },
    {
      "epoch": 0.21152347570654806,
      "grad_norm": 0.21923942863941193,
      "learning_rate": 9.998989552189449e-06,
      "loss": 0.0062,
      "step": 32980
    },
    {
      "epoch": 0.21158761260033415,
      "grad_norm": 0.5289713740348816,
      "learning_rate": 9.998978269045233e-06,
      "loss": 0.0089,
      "step": 32990
    },
    {
      "epoch": 0.21165174949412024,
      "grad_norm": 0.3155423104763031,
      "learning_rate": 9.998966923260247e-06,
      "loss": 0.0095,
      "step": 33000
    },
    {
      "epoch": 0.21171588638790634,
      "grad_norm": 0.365531325340271,
      "learning_rate": 9.998955514834634e-06,
      "loss": 0.0055,
      "step": 33010
    },
    {
      "epoch": 0.21178002328169243,
      "grad_norm": 0.30708417296409607,
      "learning_rate": 9.998944043768534e-06,
      "loss": 0.0064,
      "step": 33020
    },
    {
      "epoch": 0.21184416017547855,
      "grad_norm": 0.3104497790336609,
      "learning_rate": 9.998932510062093e-06,
      "loss": 0.0062,
      "step": 33030
    },
    {
      "epoch": 0.21190829706926465,
      "grad_norm": 0.5184586644172668,
      "learning_rate": 9.998920913715457e-06,
      "loss": 0.0055,
      "step": 33040
    },
    {
      "epoch": 0.21197243396305074,
      "grad_norm": 0.2403479516506195,
      "learning_rate": 9.998909254728767e-06,
      "loss": 0.0052,
      "step": 33050
    },
    {
      "epoch": 0.21203657085683683,
      "grad_norm": 0.47688665986061096,
      "learning_rate": 9.998897533102173e-06,
      "loss": 0.0062,
      "step": 33060
    },
    {
      "epoch": 0.21210070775062292,
      "grad_norm": 0.4534326493740082,
      "learning_rate": 9.998885748835819e-06,
      "loss": 0.0065,
      "step": 33070
    },
    {
      "epoch": 0.21216484464440902,
      "grad_norm": 0.3673833906650543,
      "learning_rate": 9.998873901929853e-06,
      "loss": 0.007,
      "step": 33080
    },
    {
      "epoch": 0.2122289815381951,
      "grad_norm": 0.16532112658023834,
      "learning_rate": 9.998861992384426e-06,
      "loss": 0.007,
      "step": 33090
    },
    {
      "epoch": 0.21229311843198123,
      "grad_norm": 0.368022084236145,
      "learning_rate": 9.998850020199686e-06,
      "loss": 0.0064,
      "step": 33100
    },
    {
      "epoch": 0.21235725532576732,
      "grad_norm": 0.21564172208309174,
      "learning_rate": 9.99883798537578e-06,
      "loss": 0.0064,
      "step": 33110
    },
    {
      "epoch": 0.21242139221955342,
      "grad_norm": 0.29615518450737,
      "learning_rate": 9.998825887912865e-06,
      "loss": 0.0091,
      "step": 33120
    },
    {
      "epoch": 0.2124855291133395,
      "grad_norm": 0.07216177135705948,
      "learning_rate": 9.998813727811085e-06,
      "loss": 0.0059,
      "step": 33130
    },
    {
      "epoch": 0.2125496660071256,
      "grad_norm": 0.17017757892608643,
      "learning_rate": 9.998801505070599e-06,
      "loss": 0.0062,
      "step": 33140
    },
    {
      "epoch": 0.2126138029009117,
      "grad_norm": 0.26409536600112915,
      "learning_rate": 9.998789219691557e-06,
      "loss": 0.0058,
      "step": 33150
    },
    {
      "epoch": 0.2126779397946978,
      "grad_norm": 0.1319057047367096,
      "learning_rate": 9.998776871674114e-06,
      "loss": 0.0043,
      "step": 33160
    },
    {
      "epoch": 0.2127420766884839,
      "grad_norm": 0.4250667989253998,
      "learning_rate": 9.998764461018423e-06,
      "loss": 0.0059,
      "step": 33170
    },
    {
      "epoch": 0.21280621358227,
      "grad_norm": 0.2796914577484131,
      "learning_rate": 9.998751987724643e-06,
      "loss": 0.0062,
      "step": 33180
    },
    {
      "epoch": 0.2128703504760561,
      "grad_norm": 0.20251691341400146,
      "learning_rate": 9.998739451792927e-06,
      "loss": 0.0063,
      "step": 33190
    },
    {
      "epoch": 0.2129344873698422,
      "grad_norm": 0.2854151129722595,
      "learning_rate": 9.998726853223432e-06,
      "loss": 0.0045,
      "step": 33200
    },
    {
      "epoch": 0.21299862426362828,
      "grad_norm": 0.17091292142868042,
      "learning_rate": 9.998714192016318e-06,
      "loss": 0.007,
      "step": 33210
    },
    {
      "epoch": 0.21306276115741438,
      "grad_norm": 0.12076331675052643,
      "learning_rate": 9.998701468171743e-06,
      "loss": 0.0048,
      "step": 33220
    },
    {
      "epoch": 0.21312689805120047,
      "grad_norm": 0.3442078232765198,
      "learning_rate": 9.998688681689865e-06,
      "loss": 0.0058,
      "step": 33230
    },
    {
      "epoch": 0.2131910349449866,
      "grad_norm": 0.35692065954208374,
      "learning_rate": 9.998675832570845e-06,
      "loss": 0.0053,
      "step": 33240
    },
    {
      "epoch": 0.21325517183877268,
      "grad_norm": 0.31049343943595886,
      "learning_rate": 9.998662920814846e-06,
      "loss": 0.0063,
      "step": 33250
    },
    {
      "epoch": 0.21331930873255878,
      "grad_norm": 0.4583049416542053,
      "learning_rate": 9.998649946422028e-06,
      "loss": 0.008,
      "step": 33260
    },
    {
      "epoch": 0.21338344562634487,
      "grad_norm": 0.22530090808868408,
      "learning_rate": 9.998636909392551e-06,
      "loss": 0.006,
      "step": 33270
    },
    {
      "epoch": 0.21344758252013096,
      "grad_norm": 0.3867437541484833,
      "learning_rate": 9.998623809726585e-06,
      "loss": 0.0053,
      "step": 33280
    },
    {
      "epoch": 0.21351171941391706,
      "grad_norm": 0.5634160041809082,
      "learning_rate": 9.998610647424287e-06,
      "loss": 0.0091,
      "step": 33290
    },
    {
      "epoch": 0.21357585630770315,
      "grad_norm": 0.16131380200386047,
      "learning_rate": 9.998597422485826e-06,
      "loss": 0.0112,
      "step": 33300
    },
    {
      "epoch": 0.21363999320148927,
      "grad_norm": 0.8143040537834167,
      "learning_rate": 9.998584134911368e-06,
      "loss": 0.0063,
      "step": 33310
    },
    {
      "epoch": 0.21370413009527536,
      "grad_norm": 0.1807277947664261,
      "learning_rate": 9.998570784701077e-06,
      "loss": 0.0061,
      "step": 33320
    },
    {
      "epoch": 0.21376826698906146,
      "grad_norm": 0.36162883043289185,
      "learning_rate": 9.998557371855123e-06,
      "loss": 0.0064,
      "step": 33330
    },
    {
      "epoch": 0.21383240388284755,
      "grad_norm": 0.25059476494789124,
      "learning_rate": 9.998543896373672e-06,
      "loss": 0.0086,
      "step": 33340
    },
    {
      "epoch": 0.21389654077663364,
      "grad_norm": 0.4012435972690582,
      "learning_rate": 9.998530358256893e-06,
      "loss": 0.0071,
      "step": 33350
    },
    {
      "epoch": 0.21396067767041974,
      "grad_norm": 0.2773272395133972,
      "learning_rate": 9.998516757504958e-06,
      "loss": 0.0071,
      "step": 33360
    },
    {
      "epoch": 0.21402481456420583,
      "grad_norm": 0.2923520803451538,
      "learning_rate": 9.998503094118033e-06,
      "loss": 0.0085,
      "step": 33370
    },
    {
      "epoch": 0.21408895145799195,
      "grad_norm": 0.3366430103778839,
      "learning_rate": 9.998489368096293e-06,
      "loss": 0.0058,
      "step": 33380
    },
    {
      "epoch": 0.21415308835177804,
      "grad_norm": 0.1886427104473114,
      "learning_rate": 9.998475579439909e-06,
      "loss": 0.0058,
      "step": 33390
    },
    {
      "epoch": 0.21421722524556414,
      "grad_norm": 0.4118272364139557,
      "learning_rate": 9.998461728149053e-06,
      "loss": 0.0075,
      "step": 33400
    },
    {
      "epoch": 0.21428136213935023,
      "grad_norm": 0.32713785767555237,
      "learning_rate": 9.9984478142239e-06,
      "loss": 0.0049,
      "step": 33410
    },
    {
      "epoch": 0.21434549903313632,
      "grad_norm": 0.21947899460792542,
      "learning_rate": 9.998433837664623e-06,
      "loss": 0.0064,
      "step": 33420
    },
    {
      "epoch": 0.21440963592692242,
      "grad_norm": 0.6944822669029236,
      "learning_rate": 9.998419798471399e-06,
      "loss": 0.0084,
      "step": 33430
    },
    {
      "epoch": 0.2144737728207085,
      "grad_norm": 0.05396656319499016,
      "learning_rate": 9.998405696644402e-06,
      "loss": 0.0043,
      "step": 33440
    },
    {
      "epoch": 0.21453790971449463,
      "grad_norm": 0.37956616282463074,
      "learning_rate": 9.998391532183809e-06,
      "loss": 0.0075,
      "step": 33450
    },
    {
      "epoch": 0.21460204660828072,
      "grad_norm": 0.3165275454521179,
      "learning_rate": 9.998377305089797e-06,
      "loss": 0.0045,
      "step": 33460
    },
    {
      "epoch": 0.21466618350206682,
      "grad_norm": 0.3055875301361084,
      "learning_rate": 9.998363015362546e-06,
      "loss": 0.0034,
      "step": 33470
    },
    {
      "epoch": 0.2147303203958529,
      "grad_norm": 0.30824825167655945,
      "learning_rate": 9.998348663002234e-06,
      "loss": 0.0073,
      "step": 33480
    },
    {
      "epoch": 0.214794457289639,
      "grad_norm": 0.08666949719190598,
      "learning_rate": 9.998334248009041e-06,
      "loss": 0.0096,
      "step": 33490
    },
    {
      "epoch": 0.2148585941834251,
      "grad_norm": 0.08656428754329681,
      "learning_rate": 9.998319770383148e-06,
      "loss": 0.0076,
      "step": 33500
    },
    {
      "epoch": 0.2149227310772112,
      "grad_norm": 0.3311026096343994,
      "learning_rate": 9.998305230124736e-06,
      "loss": 0.0071,
      "step": 33510
    },
    {
      "epoch": 0.2149868679709973,
      "grad_norm": 0.1963089257478714,
      "learning_rate": 9.998290627233986e-06,
      "loss": 0.0041,
      "step": 33520
    },
    {
      "epoch": 0.2150510048647834,
      "grad_norm": 0.23175717890262604,
      "learning_rate": 9.998275961711083e-06,
      "loss": 0.0047,
      "step": 33530
    },
    {
      "epoch": 0.2151151417585695,
      "grad_norm": 0.14738452434539795,
      "learning_rate": 9.99826123355621e-06,
      "loss": 0.0103,
      "step": 33540
    },
    {
      "epoch": 0.2151792786523556,
      "grad_norm": 0.3874845802783966,
      "learning_rate": 9.998246442769552e-06,
      "loss": 0.007,
      "step": 33550
    },
    {
      "epoch": 0.21524341554614168,
      "grad_norm": 0.20483845472335815,
      "learning_rate": 9.998231589351295e-06,
      "loss": 0.0066,
      "step": 33560
    },
    {
      "epoch": 0.21530755243992777,
      "grad_norm": 0.5111588835716248,
      "learning_rate": 9.998216673301622e-06,
      "loss": 0.0093,
      "step": 33570
    },
    {
      "epoch": 0.21537168933371387,
      "grad_norm": 0.48205098509788513,
      "learning_rate": 9.998201694620722e-06,
      "loss": 0.0093,
      "step": 33580
    },
    {
      "epoch": 0.2154358262275,
      "grad_norm": 0.3506639301776886,
      "learning_rate": 9.998186653308784e-06,
      "loss": 0.0045,
      "step": 33590
    },
    {
      "epoch": 0.21549996312128608,
      "grad_norm": 0.15712834894657135,
      "learning_rate": 9.998171549365992e-06,
      "loss": 0.0093,
      "step": 33600
    },
    {
      "epoch": 0.21556410001507217,
      "grad_norm": 0.46175265312194824,
      "learning_rate": 9.99815638279254e-06,
      "loss": 0.0104,
      "step": 33610
    },
    {
      "epoch": 0.21562823690885827,
      "grad_norm": 0.2752005159854889,
      "learning_rate": 9.998141153588619e-06,
      "loss": 0.0055,
      "step": 33620
    },
    {
      "epoch": 0.21569237380264436,
      "grad_norm": 0.17104874551296234,
      "learning_rate": 9.998125861754414e-06,
      "loss": 0.0095,
      "step": 33630
    },
    {
      "epoch": 0.21575651069643045,
      "grad_norm": 0.1689721643924713,
      "learning_rate": 9.998110507290122e-06,
      "loss": 0.006,
      "step": 33640
    },
    {
      "epoch": 0.21582064759021655,
      "grad_norm": 0.3183901906013489,
      "learning_rate": 9.998095090195932e-06,
      "loss": 0.006,
      "step": 33650
    },
    {
      "epoch": 0.21588478448400267,
      "grad_norm": 0.47291427850723267,
      "learning_rate": 9.998079610472038e-06,
      "loss": 0.0067,
      "step": 33660
    },
    {
      "epoch": 0.21594892137778876,
      "grad_norm": 0.4215930700302124,
      "learning_rate": 9.998064068118635e-06,
      "loss": 0.0043,
      "step": 33670
    },
    {
      "epoch": 0.21601305827157485,
      "grad_norm": 0.2335570752620697,
      "learning_rate": 9.998048463135916e-06,
      "loss": 0.0052,
      "step": 33680
    },
    {
      "epoch": 0.21607719516536095,
      "grad_norm": 0.3899843692779541,
      "learning_rate": 9.99803279552408e-06,
      "loss": 0.0063,
      "step": 33690
    },
    {
      "epoch": 0.21614133205914704,
      "grad_norm": 0.4035715162754059,
      "learning_rate": 9.998017065283321e-06,
      "loss": 0.0071,
      "step": 33700
    },
    {
      "epoch": 0.21620546895293313,
      "grad_norm": 0.18261416256427765,
      "learning_rate": 9.998001272413833e-06,
      "loss": 0.0062,
      "step": 33710
    },
    {
      "epoch": 0.21626960584671923,
      "grad_norm": 0.5986291170120239,
      "learning_rate": 9.99798541691582e-06,
      "loss": 0.0048,
      "step": 33720
    },
    {
      "epoch": 0.21633374274050535,
      "grad_norm": 0.3167121112346649,
      "learning_rate": 9.997969498789478e-06,
      "loss": 0.0053,
      "step": 33730
    },
    {
      "epoch": 0.21639787963429144,
      "grad_norm": 0.1511198729276657,
      "learning_rate": 9.997953518035004e-06,
      "loss": 0.0079,
      "step": 33740
    },
    {
      "epoch": 0.21646201652807753,
      "grad_norm": 1.4042974710464478,
      "learning_rate": 9.997937474652602e-06,
      "loss": 0.0063,
      "step": 33750
    },
    {
      "epoch": 0.21652615342186363,
      "grad_norm": 0.27419793605804443,
      "learning_rate": 9.997921368642471e-06,
      "loss": 0.0115,
      "step": 33760
    },
    {
      "epoch": 0.21659029031564972,
      "grad_norm": 0.2841130495071411,
      "learning_rate": 9.997905200004813e-06,
      "loss": 0.007,
      "step": 33770
    },
    {
      "epoch": 0.2166544272094358,
      "grad_norm": 0.13329987227916718,
      "learning_rate": 9.99788896873983e-06,
      "loss": 0.0065,
      "step": 33780
    },
    {
      "epoch": 0.2167185641032219,
      "grad_norm": 0.33807572722435,
      "learning_rate": 9.997872674847729e-06,
      "loss": 0.0115,
      "step": 33790
    },
    {
      "epoch": 0.21678270099700803,
      "grad_norm": 0.27996155619621277,
      "learning_rate": 9.99785631832871e-06,
      "loss": 0.0069,
      "step": 33800
    },
    {
      "epoch": 0.21684683789079412,
      "grad_norm": 0.13712485134601593,
      "learning_rate": 9.997839899182978e-06,
      "loss": 0.0076,
      "step": 33810
    },
    {
      "epoch": 0.2169109747845802,
      "grad_norm": 0.23094645142555237,
      "learning_rate": 9.99782341741074e-06,
      "loss": 0.0069,
      "step": 33820
    },
    {
      "epoch": 0.2169751116783663,
      "grad_norm": 0.18428556621074677,
      "learning_rate": 9.997806873012206e-06,
      "loss": 0.0077,
      "step": 33830
    },
    {
      "epoch": 0.2170392485721524,
      "grad_norm": 0.4980505704879761,
      "learning_rate": 9.997790265987579e-06,
      "loss": 0.0072,
      "step": 33840
    },
    {
      "epoch": 0.2171033854659385,
      "grad_norm": 0.22734621167182922,
      "learning_rate": 9.997773596337065e-06,
      "loss": 0.0079,
      "step": 33850
    },
    {
      "epoch": 0.21716752235972459,
      "grad_norm": 0.479688435792923,
      "learning_rate": 9.997756864060878e-06,
      "loss": 0.0066,
      "step": 33860
    },
    {
      "epoch": 0.2172316592535107,
      "grad_norm": 0.38074034452438354,
      "learning_rate": 9.997740069159227e-06,
      "loss": 0.0063,
      "step": 33870
    },
    {
      "epoch": 0.2172957961472968,
      "grad_norm": 0.2710610330104828,
      "learning_rate": 9.997723211632319e-06,
      "loss": 0.0047,
      "step": 33880
    },
    {
      "epoch": 0.2173599330410829,
      "grad_norm": 0.2434622347354889,
      "learning_rate": 9.997706291480369e-06,
      "loss": 0.0085,
      "step": 33890
    },
    {
      "epoch": 0.21742406993486899,
      "grad_norm": 0.35984596610069275,
      "learning_rate": 9.997689308703586e-06,
      "loss": 0.0106,
      "step": 33900
    },
    {
      "epoch": 0.21748820682865508,
      "grad_norm": 0.14613239467144012,
      "learning_rate": 9.997672263302187e-06,
      "loss": 0.0044,
      "step": 33910
    },
    {
      "epoch": 0.21755234372244117,
      "grad_norm": 0.2217494696378708,
      "learning_rate": 9.99765515527638e-06,
      "loss": 0.0081,
      "step": 33920
    },
    {
      "epoch": 0.21761648061622726,
      "grad_norm": 0.39456450939178467,
      "learning_rate": 9.997637984626382e-06,
      "loss": 0.0044,
      "step": 33930
    },
    {
      "epoch": 0.21768061751001339,
      "grad_norm": 0.16979649662971497,
      "learning_rate": 9.997620751352409e-06,
      "loss": 0.0058,
      "step": 33940
    },
    {
      "epoch": 0.21774475440379948,
      "grad_norm": 0.5155903697013855,
      "learning_rate": 9.997603455454676e-06,
      "loss": 0.0075,
      "step": 33950
    },
    {
      "epoch": 0.21780889129758557,
      "grad_norm": 0.26639315485954285,
      "learning_rate": 9.9975860969334e-06,
      "loss": 0.006,
      "step": 33960
    },
    {
      "epoch": 0.21787302819137166,
      "grad_norm": 0.46729981899261475,
      "learning_rate": 9.997568675788801e-06,
      "loss": 0.0071,
      "step": 33970
    },
    {
      "epoch": 0.21793716508515776,
      "grad_norm": 0.26663681864738464,
      "learning_rate": 9.997551192021092e-06,
      "loss": 0.0041,
      "step": 33980
    },
    {
      "epoch": 0.21800130197894385,
      "grad_norm": 0.3270251154899597,
      "learning_rate": 9.997533645630495e-06,
      "loss": 0.0054,
      "step": 33990
    },
    {
      "epoch": 0.21806543887272994,
      "grad_norm": 0.23128148913383484,
      "learning_rate": 9.99751603661723e-06,
      "loss": 0.0053,
      "step": 34000
    },
    {
      "epoch": 0.21812957576651607,
      "grad_norm": 0.3948480188846588,
      "learning_rate": 9.997498364981516e-06,
      "loss": 0.0077,
      "step": 34010
    },
    {
      "epoch": 0.21819371266030216,
      "grad_norm": 0.2914036810398102,
      "learning_rate": 9.997480630723578e-06,
      "loss": 0.0063,
      "step": 34020
    },
    {
      "epoch": 0.21825784955408825,
      "grad_norm": 0.2660025656223297,
      "learning_rate": 9.997462833843636e-06,
      "loss": 0.0054,
      "step": 34030
    },
    {
      "epoch": 0.21832198644787434,
      "grad_norm": 0.26283159852027893,
      "learning_rate": 9.997444974341912e-06,
      "loss": 0.0063,
      "step": 34040
    },
    {
      "epoch": 0.21838612334166044,
      "grad_norm": 0.4147302806377411,
      "learning_rate": 9.997427052218632e-06,
      "loss": 0.0101,
      "step": 34050
    },
    {
      "epoch": 0.21845026023544653,
      "grad_norm": 0.21125684678554535,
      "learning_rate": 9.997409067474018e-06,
      "loss": 0.0046,
      "step": 34060
    },
    {
      "epoch": 0.21851439712923262,
      "grad_norm": 0.1434185951948166,
      "learning_rate": 9.997391020108298e-06,
      "loss": 0.0049,
      "step": 34070
    },
    {
      "epoch": 0.21857853402301874,
      "grad_norm": 0.20431987941265106,
      "learning_rate": 9.997372910121696e-06,
      "loss": 0.005,
      "step": 34080
    },
    {
      "epoch": 0.21864267091680484,
      "grad_norm": 0.3831230700016022,
      "learning_rate": 9.99735473751444e-06,
      "loss": 0.0045,
      "step": 34090
    },
    {
      "epoch": 0.21870680781059093,
      "grad_norm": 0.3873409330844879,
      "learning_rate": 9.997336502286758e-06,
      "loss": 0.0055,
      "step": 34100
    },
    {
      "epoch": 0.21877094470437702,
      "grad_norm": 0.47922199964523315,
      "learning_rate": 9.997318204438879e-06,
      "loss": 0.0086,
      "step": 34110
    },
    {
      "epoch": 0.21883508159816312,
      "grad_norm": 0.19254538416862488,
      "learning_rate": 9.997299843971029e-06,
      "loss": 0.005,
      "step": 34120
    },
    {
      "epoch": 0.2188992184919492,
      "grad_norm": 0.28071871399879456,
      "learning_rate": 9.997281420883443e-06,
      "loss": 0.0048,
      "step": 34130
    },
    {
      "epoch": 0.2189633553857353,
      "grad_norm": 0.46463486552238464,
      "learning_rate": 9.997262935176347e-06,
      "loss": 0.0056,
      "step": 34140
    },
    {
      "epoch": 0.21902749227952142,
      "grad_norm": 0.30839383602142334,
      "learning_rate": 9.997244386849976e-06,
      "loss": 0.0056,
      "step": 34150
    },
    {
      "epoch": 0.21909162917330752,
      "grad_norm": 0.3349360227584839,
      "learning_rate": 9.997225775904561e-06,
      "loss": 0.005,
      "step": 34160
    },
    {
      "epoch": 0.2191557660670936,
      "grad_norm": 0.24762451648712158,
      "learning_rate": 9.997207102340336e-06,
      "loss": 0.0065,
      "step": 34170
    },
    {
      "epoch": 0.2192199029608797,
      "grad_norm": 0.26319772005081177,
      "learning_rate": 9.997188366157536e-06,
      "loss": 0.006,
      "step": 34180
    },
    {
      "epoch": 0.2192840398546658,
      "grad_norm": 0.4329639673233032,
      "learning_rate": 9.997169567356392e-06,
      "loss": 0.0053,
      "step": 34190
    },
    {
      "epoch": 0.2193481767484519,
      "grad_norm": 0.23011185228824615,
      "learning_rate": 9.997150705937141e-06,
      "loss": 0.0046,
      "step": 34200
    },
    {
      "epoch": 0.21941231364223798,
      "grad_norm": 0.19329330325126648,
      "learning_rate": 9.997131781900023e-06,
      "loss": 0.0125,
      "step": 34210
    },
    {
      "epoch": 0.2194764505360241,
      "grad_norm": 0.07114830613136292,
      "learning_rate": 9.997112795245272e-06,
      "loss": 0.0054,
      "step": 34220
    },
    {
      "epoch": 0.2195405874298102,
      "grad_norm": 0.40303030610084534,
      "learning_rate": 9.997093745973125e-06,
      "loss": 0.0082,
      "step": 34230
    },
    {
      "epoch": 0.2196047243235963,
      "grad_norm": 0.6453074812889099,
      "learning_rate": 9.997074634083823e-06,
      "loss": 0.009,
      "step": 34240
    },
    {
      "epoch": 0.21966886121738238,
      "grad_norm": 0.4790237545967102,
      "learning_rate": 9.997055459577605e-06,
      "loss": 0.0057,
      "step": 34250
    },
    {
      "epoch": 0.21973299811116848,
      "grad_norm": 0.36487022042274475,
      "learning_rate": 9.997036222454709e-06,
      "loss": 0.0069,
      "step": 34260
    },
    {
      "epoch": 0.21979713500495457,
      "grad_norm": 0.3129631578922272,
      "learning_rate": 9.99701692271538e-06,
      "loss": 0.0088,
      "step": 34270
    },
    {
      "epoch": 0.21986127189874066,
      "grad_norm": 0.3710615634918213,
      "learning_rate": 9.996997560359855e-06,
      "loss": 0.0067,
      "step": 34280
    },
    {
      "epoch": 0.21992540879252676,
      "grad_norm": 0.2760692238807678,
      "learning_rate": 9.99697813538838e-06,
      "loss": 0.0092,
      "step": 34290
    },
    {
      "epoch": 0.21998954568631288,
      "grad_norm": 0.3143475651741028,
      "learning_rate": 9.996958647801199e-06,
      "loss": 0.0065,
      "step": 34300
    },
    {
      "epoch": 0.22005368258009897,
      "grad_norm": 0.4855823814868927,
      "learning_rate": 9.996939097598554e-06,
      "loss": 0.0083,
      "step": 34310
    },
    {
      "epoch": 0.22011781947388506,
      "grad_norm": 0.19354212284088135,
      "learning_rate": 9.996919484780692e-06,
      "loss": 0.0047,
      "step": 34320
    },
    {
      "epoch": 0.22018195636767116,
      "grad_norm": 0.4365869164466858,
      "learning_rate": 9.996899809347855e-06,
      "loss": 0.0064,
      "step": 34330
    },
    {
      "epoch": 0.22024609326145725,
      "grad_norm": 0.3569786250591278,
      "learning_rate": 9.996880071300294e-06,
      "loss": 0.006,
      "step": 34340
    },
    {
      "epoch": 0.22031023015524334,
      "grad_norm": 0.5439212918281555,
      "learning_rate": 9.996860270638255e-06,
      "loss": 0.0091,
      "step": 34350
    },
    {
      "epoch": 0.22037436704902943,
      "grad_norm": 0.22508126497268677,
      "learning_rate": 9.996840407361983e-06,
      "loss": 0.0086,
      "step": 34360
    },
    {
      "epoch": 0.22043850394281556,
      "grad_norm": 0.40013641119003296,
      "learning_rate": 9.996820481471734e-06,
      "loss": 0.0051,
      "step": 34370
    },
    {
      "epoch": 0.22050264083660165,
      "grad_norm": 0.10357312113046646,
      "learning_rate": 9.996800492967748e-06,
      "loss": 0.0048,
      "step": 34380
    },
    {
      "epoch": 0.22056677773038774,
      "grad_norm": 0.212563619017601,
      "learning_rate": 9.996780441850284e-06,
      "loss": 0.0057,
      "step": 34390
    },
    {
      "epoch": 0.22063091462417384,
      "grad_norm": 0.4786578416824341,
      "learning_rate": 9.99676032811959e-06,
      "loss": 0.0074,
      "step": 34400
    },
    {
      "epoch": 0.22069505151795993,
      "grad_norm": 0.14567528665065765,
      "learning_rate": 9.996740151775917e-06,
      "loss": 0.0058,
      "step": 34410
    },
    {
      "epoch": 0.22075918841174602,
      "grad_norm": 0.6325238347053528,
      "learning_rate": 9.996719912819519e-06,
      "loss": 0.0072,
      "step": 34420
    },
    {
      "epoch": 0.22082332530553211,
      "grad_norm": 0.22808292508125305,
      "learning_rate": 9.996699611250648e-06,
      "loss": 0.0084,
      "step": 34430
    },
    {
      "epoch": 0.22088746219931824,
      "grad_norm": 0.2220643162727356,
      "learning_rate": 9.996679247069561e-06,
      "loss": 0.0063,
      "step": 34440
    },
    {
      "epoch": 0.22095159909310433,
      "grad_norm": 0.20701898634433746,
      "learning_rate": 9.996658820276512e-06,
      "loss": 0.0069,
      "step": 34450
    },
    {
      "epoch": 0.22101573598689042,
      "grad_norm": 0.16193971037864685,
      "learning_rate": 9.996638330871758e-06,
      "loss": 0.0073,
      "step": 34460
    },
    {
      "epoch": 0.22107987288067651,
      "grad_norm": 0.31429579854011536,
      "learning_rate": 9.996617778855553e-06,
      "loss": 0.0043,
      "step": 34470
    },
    {
      "epoch": 0.2211440097744626,
      "grad_norm": 0.17158426344394684,
      "learning_rate": 9.996597164228157e-06,
      "loss": 0.0077,
      "step": 34480
    },
    {
      "epoch": 0.2212081466682487,
      "grad_norm": 0.29266834259033203,
      "learning_rate": 9.996576486989827e-06,
      "loss": 0.0063,
      "step": 34490
    },
    {
      "epoch": 0.2212722835620348,
      "grad_norm": 0.5624863505363464,
      "learning_rate": 9.996555747140824e-06,
      "loss": 0.0069,
      "step": 34500
    },
    {
      "epoch": 0.22133642045582091,
      "grad_norm": 0.26331159472465515,
      "learning_rate": 9.996534944681404e-06,
      "loss": 0.0052,
      "step": 34510
    },
    {
      "epoch": 0.221400557349607,
      "grad_norm": 0.17117565870285034,
      "learning_rate": 9.996514079611832e-06,
      "loss": 0.0061,
      "step": 34520
    },
    {
      "epoch": 0.2214646942433931,
      "grad_norm": 0.20332291722297668,
      "learning_rate": 9.996493151932367e-06,
      "loss": 0.0054,
      "step": 34530
    },
    {
      "epoch": 0.2215288311371792,
      "grad_norm": 0.2634843587875366,
      "learning_rate": 9.996472161643273e-06,
      "loss": 0.0068,
      "step": 34540
    },
    {
      "epoch": 0.2215929680309653,
      "grad_norm": 0.4386470317840576,
      "learning_rate": 9.996451108744811e-06,
      "loss": 0.0041,
      "step": 34550
    },
    {
      "epoch": 0.22165710492475138,
      "grad_norm": 0.4105231761932373,
      "learning_rate": 9.996429993237245e-06,
      "loss": 0.0049,
      "step": 34560
    },
    {
      "epoch": 0.22172124181853747,
      "grad_norm": 0.2908138334751129,
      "learning_rate": 9.996408815120842e-06,
      "loss": 0.0061,
      "step": 34570
    },
    {
      "epoch": 0.2217853787123236,
      "grad_norm": 0.29927143454551697,
      "learning_rate": 9.996387574395863e-06,
      "loss": 0.005,
      "step": 34580
    },
    {
      "epoch": 0.2218495156061097,
      "grad_norm": 0.29443830251693726,
      "learning_rate": 9.996366271062578e-06,
      "loss": 0.0051,
      "step": 34590
    },
    {
      "epoch": 0.22191365249989578,
      "grad_norm": 0.13422074913978577,
      "learning_rate": 9.996344905121255e-06,
      "loss": 0.0057,
      "step": 34600
    },
    {
      "epoch": 0.22197778939368187,
      "grad_norm": 0.39134156703948975,
      "learning_rate": 9.996323476572157e-06,
      "loss": 0.0081,
      "step": 34610
    },
    {
      "epoch": 0.22204192628746797,
      "grad_norm": 0.22255438566207886,
      "learning_rate": 9.996301985415556e-06,
      "loss": 0.0118,
      "step": 34620
    },
    {
      "epoch": 0.22210606318125406,
      "grad_norm": 0.35640034079551697,
      "learning_rate": 9.99628043165172e-06,
      "loss": 0.0052,
      "step": 34630
    },
    {
      "epoch": 0.22217020007504015,
      "grad_norm": 0.5733597874641418,
      "learning_rate": 9.996258815280919e-06,
      "loss": 0.0058,
      "step": 34640
    },
    {
      "epoch": 0.22223433696882627,
      "grad_norm": 0.32103994488716125,
      "learning_rate": 9.996237136303426e-06,
      "loss": 0.0076,
      "step": 34650
    },
    {
      "epoch": 0.22229847386261237,
      "grad_norm": 0.5610170960426331,
      "learning_rate": 9.996215394719509e-06,
      "loss": 0.0083,
      "step": 34660
    },
    {
      "epoch": 0.22236261075639846,
      "grad_norm": 0.28157588839530945,
      "learning_rate": 9.996193590529445e-06,
      "loss": 0.009,
      "step": 34670
    },
    {
      "epoch": 0.22242674765018455,
      "grad_norm": 0.379728764295578,
      "learning_rate": 9.9961717237335e-06,
      "loss": 0.0071,
      "step": 34680
    },
    {
      "epoch": 0.22249088454397065,
      "grad_norm": 0.3714889585971832,
      "learning_rate": 9.996149794331955e-06,
      "loss": 0.0083,
      "step": 34690
    },
    {
      "epoch": 0.22255502143775674,
      "grad_norm": 0.1807042509317398,
      "learning_rate": 9.996127802325084e-06,
      "loss": 0.0042,
      "step": 34700
    },
    {
      "epoch": 0.22261915833154283,
      "grad_norm": 0.5503183007240295,
      "learning_rate": 9.996105747713158e-06,
      "loss": 0.01,
      "step": 34710
    },
    {
      "epoch": 0.22268329522532895,
      "grad_norm": 0.39598900079727173,
      "learning_rate": 9.99608363049646e-06,
      "loss": 0.0087,
      "step": 34720
    },
    {
      "epoch": 0.22274743211911505,
      "grad_norm": 0.4356114864349365,
      "learning_rate": 9.996061450675261e-06,
      "loss": 0.006,
      "step": 34730
    },
    {
      "epoch": 0.22281156901290114,
      "grad_norm": 0.15179894864559174,
      "learning_rate": 9.996039208249842e-06,
      "loss": 0.0079,
      "step": 34740
    },
    {
      "epoch": 0.22287570590668723,
      "grad_norm": 0.25440022349357605,
      "learning_rate": 9.99601690322048e-06,
      "loss": 0.0071,
      "step": 34750
    },
    {
      "epoch": 0.22293984280047333,
      "grad_norm": 0.18919359147548676,
      "learning_rate": 9.995994535587456e-06,
      "loss": 0.008,
      "step": 34760
    },
    {
      "epoch": 0.22300397969425942,
      "grad_norm": 0.9403988122940063,
      "learning_rate": 9.99597210535105e-06,
      "loss": 0.0104,
      "step": 34770
    },
    {
      "epoch": 0.2230681165880455,
      "grad_norm": 0.4549175500869751,
      "learning_rate": 9.995949612511545e-06,
      "loss": 0.0047,
      "step": 34780
    },
    {
      "epoch": 0.22313225348183163,
      "grad_norm": 0.4007059335708618,
      "learning_rate": 9.995927057069218e-06,
      "loss": 0.0054,
      "step": 34790
    },
    {
      "epoch": 0.22319639037561773,
      "grad_norm": 0.31075727939605713,
      "learning_rate": 9.995904439024356e-06,
      "loss": 0.0041,
      "step": 34800
    },
    {
      "epoch": 0.22326052726940382,
      "grad_norm": 0.3774067163467407,
      "learning_rate": 9.99588175837724e-06,
      "loss": 0.0065,
      "step": 34810
    },
    {
      "epoch": 0.2233246641631899,
      "grad_norm": 0.21462136507034302,
      "learning_rate": 9.995859015128156e-06,
      "loss": 0.0093,
      "step": 34820
    },
    {
      "epoch": 0.223388801056976,
      "grad_norm": 0.2967641055583954,
      "learning_rate": 9.995836209277388e-06,
      "loss": 0.0073,
      "step": 34830
    },
    {
      "epoch": 0.2234529379507621,
      "grad_norm": 0.3427407145500183,
      "learning_rate": 9.99581334082522e-06,
      "loss": 0.0047,
      "step": 34840
    },
    {
      "epoch": 0.2235170748445482,
      "grad_norm": 0.2846042513847351,
      "learning_rate": 9.995790409771942e-06,
      "loss": 0.0053,
      "step": 34850
    },
    {
      "epoch": 0.2235812117383343,
      "grad_norm": 0.23080459237098694,
      "learning_rate": 9.99576741611784e-06,
      "loss": 0.0066,
      "step": 34860
    },
    {
      "epoch": 0.2236453486321204,
      "grad_norm": 0.41316619515419006,
      "learning_rate": 9.995744359863201e-06,
      "loss": 0.0075,
      "step": 34870
    },
    {
      "epoch": 0.2237094855259065,
      "grad_norm": 0.22799259424209595,
      "learning_rate": 9.995721241008315e-06,
      "loss": 0.0048,
      "step": 34880
    },
    {
      "epoch": 0.2237736224196926,
      "grad_norm": 0.3459303081035614,
      "learning_rate": 9.995698059553471e-06,
      "loss": 0.0037,
      "step": 34890
    },
    {
      "epoch": 0.22383775931347868,
      "grad_norm": 0.19496068358421326,
      "learning_rate": 9.995674815498961e-06,
      "loss": 0.0072,
      "step": 34900
    },
    {
      "epoch": 0.22390189620726478,
      "grad_norm": 0.162460595369339,
      "learning_rate": 9.995651508845074e-06,
      "loss": 0.005,
      "step": 34910
    },
    {
      "epoch": 0.22396603310105087,
      "grad_norm": 0.3905678391456604,
      "learning_rate": 9.995628139592103e-06,
      "loss": 0.0067,
      "step": 34920
    },
    {
      "epoch": 0.224030169994837,
      "grad_norm": 0.3376942574977875,
      "learning_rate": 9.995604707740343e-06,
      "loss": 0.0046,
      "step": 34930
    },
    {
      "epoch": 0.22409430688862308,
      "grad_norm": 0.21789802610874176,
      "learning_rate": 9.995581213290085e-06,
      "loss": 0.0054,
      "step": 34940
    },
    {
      "epoch": 0.22415844378240918,
      "grad_norm": 0.45708635449409485,
      "learning_rate": 9.995557656241624e-06,
      "loss": 0.01,
      "step": 34950
    },
    {
      "epoch": 0.22422258067619527,
      "grad_norm": 0.30806267261505127,
      "learning_rate": 9.995534036595256e-06,
      "loss": 0.0046,
      "step": 34960
    },
    {
      "epoch": 0.22428671756998136,
      "grad_norm": 0.40519043803215027,
      "learning_rate": 9.995510354351275e-06,
      "loss": 0.0075,
      "step": 34970
    },
    {
      "epoch": 0.22435085446376746,
      "grad_norm": 0.3858379125595093,
      "learning_rate": 9.995486609509978e-06,
      "loss": 0.0074,
      "step": 34980
    },
    {
      "epoch": 0.22441499135755355,
      "grad_norm": 0.1513359248638153,
      "learning_rate": 9.995462802071666e-06,
      "loss": 0.0048,
      "step": 34990
    },
    {
      "epoch": 0.22447912825133967,
      "grad_norm": 0.1579079031944275,
      "learning_rate": 9.995438932036634e-06,
      "loss": 0.0045,
      "step": 35000
    },
    {
      "epoch": 0.22454326514512576,
      "grad_norm": 0.38252583146095276,
      "learning_rate": 9.995414999405182e-06,
      "loss": 0.0082,
      "step": 35010
    },
    {
      "epoch": 0.22460740203891186,
      "grad_norm": 0.17620162665843964,
      "learning_rate": 9.99539100417761e-06,
      "loss": 0.0071,
      "step": 35020
    },
    {
      "epoch": 0.22467153893269795,
      "grad_norm": 0.4304888844490051,
      "learning_rate": 9.995366946354218e-06,
      "loss": 0.006,
      "step": 35030
    },
    {
      "epoch": 0.22473567582648404,
      "grad_norm": 0.17609496414661407,
      "learning_rate": 9.995342825935309e-06,
      "loss": 0.0052,
      "step": 35040
    },
    {
      "epoch": 0.22479981272027014,
      "grad_norm": 0.294415146112442,
      "learning_rate": 9.995318642921183e-06,
      "loss": 0.0104,
      "step": 35050
    },
    {
      "epoch": 0.22486394961405623,
      "grad_norm": 0.3411390483379364,
      "learning_rate": 9.995294397312145e-06,
      "loss": 0.0052,
      "step": 35060
    },
    {
      "epoch": 0.22492808650784235,
      "grad_norm": 0.45513299107551575,
      "learning_rate": 9.995270089108498e-06,
      "loss": 0.0044,
      "step": 35070
    },
    {
      "epoch": 0.22499222340162844,
      "grad_norm": 0.15879599750041962,
      "learning_rate": 9.995245718310546e-06,
      "loss": 0.0041,
      "step": 35080
    },
    {
      "epoch": 0.22505636029541454,
      "grad_norm": 0.10694967210292816,
      "learning_rate": 9.995221284918596e-06,
      "loss": 0.0056,
      "step": 35090
    },
    {
      "epoch": 0.22512049718920063,
      "grad_norm": 0.2496778517961502,
      "learning_rate": 9.995196788932955e-06,
      "loss": 0.0086,
      "step": 35100
    },
    {
      "epoch": 0.22518463408298672,
      "grad_norm": 0.2647297978401184,
      "learning_rate": 9.995172230353924e-06,
      "loss": 0.0045,
      "step": 35110
    },
    {
      "epoch": 0.22524877097677282,
      "grad_norm": 0.5168192982673645,
      "learning_rate": 9.995147609181819e-06,
      "loss": 0.0061,
      "step": 35120
    },
    {
      "epoch": 0.2253129078705589,
      "grad_norm": 0.2683510184288025,
      "learning_rate": 9.995122925416943e-06,
      "loss": 0.0046,
      "step": 35130
    },
    {
      "epoch": 0.22537704476434503,
      "grad_norm": 0.2050730139017105,
      "learning_rate": 9.995098179059605e-06,
      "loss": 0.0069,
      "step": 35140
    },
    {
      "epoch": 0.22544118165813112,
      "grad_norm": 0.27218306064605713,
      "learning_rate": 9.99507337011012e-06,
      "loss": 0.0062,
      "step": 35150
    },
    {
      "epoch": 0.22550531855191722,
      "grad_norm": 0.08528287708759308,
      "learning_rate": 9.995048498568793e-06,
      "loss": 0.0041,
      "step": 35160
    },
    {
      "epoch": 0.2255694554457033,
      "grad_norm": 0.20639532804489136,
      "learning_rate": 9.99502356443594e-06,
      "loss": 0.0062,
      "step": 35170
    },
    {
      "epoch": 0.2256335923394894,
      "grad_norm": 0.15732760727405548,
      "learning_rate": 9.99499856771187e-06,
      "loss": 0.0036,
      "step": 35180
    },
    {
      "epoch": 0.2256977292332755,
      "grad_norm": 0.23593586683273315,
      "learning_rate": 9.994973508396901e-06,
      "loss": 0.0045,
      "step": 35190
    },
    {
      "epoch": 0.2257618661270616,
      "grad_norm": 0.3014291524887085,
      "learning_rate": 9.994948386491343e-06,
      "loss": 0.0069,
      "step": 35200
    },
    {
      "epoch": 0.2258260030208477,
      "grad_norm": 0.37358778715133667,
      "learning_rate": 9.99492320199551e-06,
      "loss": 0.0077,
      "step": 35210
    },
    {
      "epoch": 0.2258901399146338,
      "grad_norm": 0.09573258459568024,
      "learning_rate": 9.994897954909722e-06,
      "loss": 0.0079,
      "step": 35220
    },
    {
      "epoch": 0.2259542768084199,
      "grad_norm": 0.22441881895065308,
      "learning_rate": 9.994872645234293e-06,
      "loss": 0.0046,
      "step": 35230
    },
    {
      "epoch": 0.226018413702206,
      "grad_norm": 0.3367602527141571,
      "learning_rate": 9.994847272969538e-06,
      "loss": 0.0052,
      "step": 35240
    },
    {
      "epoch": 0.22608255059599208,
      "grad_norm": 0.4609188735485077,
      "learning_rate": 9.994821838115778e-06,
      "loss": 0.0058,
      "step": 35250
    },
    {
      "epoch": 0.22614668748977818,
      "grad_norm": 0.36705097556114197,
      "learning_rate": 9.99479634067333e-06,
      "loss": 0.0054,
      "step": 35260
    },
    {
      "epoch": 0.22621082438356427,
      "grad_norm": 0.07663752138614655,
      "learning_rate": 9.994770780642514e-06,
      "loss": 0.0053,
      "step": 35270
    },
    {
      "epoch": 0.2262749612773504,
      "grad_norm": 0.40321534872055054,
      "learning_rate": 9.994745158023652e-06,
      "loss": 0.0081,
      "step": 35280
    },
    {
      "epoch": 0.22633909817113648,
      "grad_norm": 0.47762489318847656,
      "learning_rate": 9.99471947281706e-06,
      "loss": 0.0102,
      "step": 35290
    },
    {
      "epoch": 0.22640323506492258,
      "grad_norm": 0.37097352743148804,
      "learning_rate": 9.994693725023067e-06,
      "loss": 0.0042,
      "step": 35300
    },
    {
      "epoch": 0.22646737195870867,
      "grad_norm": 0.33247286081314087,
      "learning_rate": 9.994667914641991e-06,
      "loss": 0.0053,
      "step": 35310
    },
    {
      "epoch": 0.22653150885249476,
      "grad_norm": 0.18313194811344147,
      "learning_rate": 9.994642041674156e-06,
      "loss": 0.009,
      "step": 35320
    },
    {
      "epoch": 0.22659564574628085,
      "grad_norm": 0.12064996361732483,
      "learning_rate": 9.994616106119885e-06,
      "loss": 0.0034,
      "step": 35330
    },
    {
      "epoch": 0.22665978264006695,
      "grad_norm": 0.09952157735824585,
      "learning_rate": 9.994590107979508e-06,
      "loss": 0.0063,
      "step": 35340
    },
    {
      "epoch": 0.22672391953385307,
      "grad_norm": 0.3346708416938782,
      "learning_rate": 9.994564047253346e-06,
      "loss": 0.0068,
      "step": 35350
    },
    {
      "epoch": 0.22678805642763916,
      "grad_norm": 0.16705653071403503,
      "learning_rate": 9.994537923941725e-06,
      "loss": 0.0065,
      "step": 35360
    },
    {
      "epoch": 0.22685219332142526,
      "grad_norm": 0.24314428865909576,
      "learning_rate": 9.994511738044976e-06,
      "loss": 0.0067,
      "step": 35370
    },
    {
      "epoch": 0.22691633021521135,
      "grad_norm": 0.28826290369033813,
      "learning_rate": 9.994485489563425e-06,
      "loss": 0.0043,
      "step": 35380
    },
    {
      "epoch": 0.22698046710899744,
      "grad_norm": 0.2939811050891876,
      "learning_rate": 9.994459178497403e-06,
      "loss": 0.0053,
      "step": 35390
    },
    {
      "epoch": 0.22704460400278353,
      "grad_norm": 0.21459046006202698,
      "learning_rate": 9.994432804847235e-06,
      "loss": 0.0077,
      "step": 35400
    },
    {
      "epoch": 0.22710874089656963,
      "grad_norm": 0.3848009407520294,
      "learning_rate": 9.994406368613257e-06,
      "loss": 0.0071,
      "step": 35410
    },
    {
      "epoch": 0.22717287779035575,
      "grad_norm": 0.31907474994659424,
      "learning_rate": 9.994379869795797e-06,
      "loss": 0.0056,
      "step": 35420
    },
    {
      "epoch": 0.22723701468414184,
      "grad_norm": 0.6475202441215515,
      "learning_rate": 9.994353308395186e-06,
      "loss": 0.0059,
      "step": 35430
    },
    {
      "epoch": 0.22730115157792793,
      "grad_norm": 0.23448844254016876,
      "learning_rate": 9.99432668441176e-06,
      "loss": 0.0062,
      "step": 35440
    },
    {
      "epoch": 0.22736528847171403,
      "grad_norm": 0.18885894119739532,
      "learning_rate": 9.994299997845851e-06,
      "loss": 0.007,
      "step": 35450
    },
    {
      "epoch": 0.22742942536550012,
      "grad_norm": 0.30449241399765015,
      "learning_rate": 9.994273248697794e-06,
      "loss": 0.0067,
      "step": 35460
    },
    {
      "epoch": 0.22749356225928621,
      "grad_norm": 0.2622857987880707,
      "learning_rate": 9.994246436967925e-06,
      "loss": 0.0097,
      "step": 35470
    },
    {
      "epoch": 0.2275576991530723,
      "grad_norm": 0.2114802598953247,
      "learning_rate": 9.994219562656577e-06,
      "loss": 0.0071,
      "step": 35480
    },
    {
      "epoch": 0.22762183604685843,
      "grad_norm": 0.2560284435749054,
      "learning_rate": 9.99419262576409e-06,
      "loss": 0.0071,
      "step": 35490
    },
    {
      "epoch": 0.22768597294064452,
      "grad_norm": 0.2646476626396179,
      "learning_rate": 9.994165626290799e-06,
      "loss": 0.008,
      "step": 35500
    },
    {
      "epoch": 0.22775010983443061,
      "grad_norm": 0.1781512349843979,
      "learning_rate": 9.994138564237044e-06,
      "loss": 0.0055,
      "step": 35510
    },
    {
      "epoch": 0.2278142467282167,
      "grad_norm": 0.18580837547779083,
      "learning_rate": 9.994111439603164e-06,
      "loss": 0.0045,
      "step": 35520
    },
    {
      "epoch": 0.2278783836220028,
      "grad_norm": 0.32868054509162903,
      "learning_rate": 9.994084252389497e-06,
      "loss": 0.0034,
      "step": 35530
    },
    {
      "epoch": 0.2279425205157889,
      "grad_norm": 0.3121737539768219,
      "learning_rate": 9.994057002596387e-06,
      "loss": 0.0075,
      "step": 35540
    },
    {
      "epoch": 0.228006657409575,
      "grad_norm": 0.4381645917892456,
      "learning_rate": 9.994029690224173e-06,
      "loss": 0.0097,
      "step": 35550
    },
    {
      "epoch": 0.2280707943033611,
      "grad_norm": 0.26858723163604736,
      "learning_rate": 9.994002315273196e-06,
      "loss": 0.0071,
      "step": 35560
    },
    {
      "epoch": 0.2281349311971472,
      "grad_norm": 0.19299057126045227,
      "learning_rate": 9.993974877743802e-06,
      "loss": 0.0079,
      "step": 35570
    },
    {
      "epoch": 0.2281990680909333,
      "grad_norm": 0.39440590143203735,
      "learning_rate": 9.993947377636332e-06,
      "loss": 0.0081,
      "step": 35580
    },
    {
      "epoch": 0.2282632049847194,
      "grad_norm": 0.11068882048130035,
      "learning_rate": 9.993919814951135e-06,
      "loss": 0.0061,
      "step": 35590
    },
    {
      "epoch": 0.22832734187850548,
      "grad_norm": 0.27898338437080383,
      "learning_rate": 9.993892189688552e-06,
      "loss": 0.0069,
      "step": 35600
    },
    {
      "epoch": 0.22839147877229157,
      "grad_norm": 0.28347986936569214,
      "learning_rate": 9.993864501848933e-06,
      "loss": 0.004,
      "step": 35610
    },
    {
      "epoch": 0.22845561566607767,
      "grad_norm": 0.16570374369621277,
      "learning_rate": 9.99383675143262e-06,
      "loss": 0.0055,
      "step": 35620
    },
    {
      "epoch": 0.2285197525598638,
      "grad_norm": 0.6783402562141418,
      "learning_rate": 9.993808938439965e-06,
      "loss": 0.0062,
      "step": 35630
    },
    {
      "epoch": 0.22858388945364988,
      "grad_norm": 0.3398296535015106,
      "learning_rate": 9.993781062871316e-06,
      "loss": 0.0069,
      "step": 35640
    },
    {
      "epoch": 0.22864802634743597,
      "grad_norm": 0.3182373046875,
      "learning_rate": 9.993753124727018e-06,
      "loss": 0.0054,
      "step": 35650
    },
    {
      "epoch": 0.22871216324122207,
      "grad_norm": 0.0939110517501831,
      "learning_rate": 9.993725124007427e-06,
      "loss": 0.0059,
      "step": 35660
    },
    {
      "epoch": 0.22877630013500816,
      "grad_norm": 0.33174213767051697,
      "learning_rate": 9.99369706071289e-06,
      "loss": 0.0042,
      "step": 35670
    },
    {
      "epoch": 0.22884043702879425,
      "grad_norm": 0.08618547767400742,
      "learning_rate": 9.99366893484376e-06,
      "loss": 0.0038,
      "step": 35680
    },
    {
      "epoch": 0.22890457392258035,
      "grad_norm": 0.34387779235839844,
      "learning_rate": 9.99364074640039e-06,
      "loss": 0.0084,
      "step": 35690
    },
    {
      "epoch": 0.22896871081636647,
      "grad_norm": 0.22006486356258392,
      "learning_rate": 9.993612495383131e-06,
      "loss": 0.0053,
      "step": 35700
    },
    {
      "epoch": 0.22903284771015256,
      "grad_norm": 0.15135833621025085,
      "learning_rate": 9.99358418179234e-06,
      "loss": 0.0064,
      "step": 35710
    },
    {
      "epoch": 0.22909698460393865,
      "grad_norm": 0.27555614709854126,
      "learning_rate": 9.99355580562837e-06,
      "loss": 0.0063,
      "step": 35720
    },
    {
      "epoch": 0.22916112149772475,
      "grad_norm": 0.3538702130317688,
      "learning_rate": 9.993527366891578e-06,
      "loss": 0.0088,
      "step": 35730
    },
    {
      "epoch": 0.22922525839151084,
      "grad_norm": 0.19812867045402527,
      "learning_rate": 9.993498865582318e-06,
      "loss": 0.0054,
      "step": 35740
    },
    {
      "epoch": 0.22928939528529693,
      "grad_norm": 0.23928295075893402,
      "learning_rate": 9.993470301700948e-06,
      "loss": 0.0064,
      "step": 35750
    },
    {
      "epoch": 0.22935353217908303,
      "grad_norm": 0.2500852346420288,
      "learning_rate": 9.993441675247829e-06,
      "loss": 0.0071,
      "step": 35760
    },
    {
      "epoch": 0.22941766907286912,
      "grad_norm": 0.2701261043548584,
      "learning_rate": 9.993412986223313e-06,
      "loss": 0.0048,
      "step": 35770
    },
    {
      "epoch": 0.22948180596665524,
      "grad_norm": 0.23142360150814056,
      "learning_rate": 9.993384234627765e-06,
      "loss": 0.0068,
      "step": 35780
    },
    {
      "epoch": 0.22954594286044133,
      "grad_norm": 0.543775200843811,
      "learning_rate": 9.993355420461546e-06,
      "loss": 0.0053,
      "step": 35790
    },
    {
      "epoch": 0.22961007975422743,
      "grad_norm": 0.17761576175689697,
      "learning_rate": 9.993326543725012e-06,
      "loss": 0.0041,
      "step": 35800
    },
    {
      "epoch": 0.22967421664801352,
      "grad_norm": 0.06380312889814377,
      "learning_rate": 9.99329760441853e-06,
      "loss": 0.0048,
      "step": 35810
    },
    {
      "epoch": 0.2297383535417996,
      "grad_norm": 0.23243442177772522,
      "learning_rate": 9.99326860254246e-06,
      "loss": 0.0064,
      "step": 35820
    },
    {
      "epoch": 0.2298024904355857,
      "grad_norm": 0.33521875739097595,
      "learning_rate": 9.993239538097164e-06,
      "loss": 0.0059,
      "step": 35830
    },
    {
      "epoch": 0.2298666273293718,
      "grad_norm": 0.14889921247959137,
      "learning_rate": 9.99321041108301e-06,
      "loss": 0.0028,
      "step": 35840
    },
    {
      "epoch": 0.22993076422315792,
      "grad_norm": 0.49306491017341614,
      "learning_rate": 9.993181221500358e-06,
      "loss": 0.0065,
      "step": 35850
    },
    {
      "epoch": 0.229994901116944,
      "grad_norm": 0.231283500790596,
      "learning_rate": 9.99315196934958e-06,
      "loss": 0.0048,
      "step": 35860
    },
    {
      "epoch": 0.2300590380107301,
      "grad_norm": 0.10951454937458038,
      "learning_rate": 9.993122654631037e-06,
      "loss": 0.0086,
      "step": 35870
    },
    {
      "epoch": 0.2301231749045162,
      "grad_norm": 0.21918267011642456,
      "learning_rate": 9.993093277345101e-06,
      "loss": 0.0035,
      "step": 35880
    },
    {
      "epoch": 0.2301873117983023,
      "grad_norm": 0.26090681552886963,
      "learning_rate": 9.993063837492136e-06,
      "loss": 0.0068,
      "step": 35890
    },
    {
      "epoch": 0.23025144869208838,
      "grad_norm": 0.17361082136631012,
      "learning_rate": 9.993034335072514e-06,
      "loss": 0.0065,
      "step": 35900
    },
    {
      "epoch": 0.23031558558587448,
      "grad_norm": 0.20613285899162292,
      "learning_rate": 9.993004770086602e-06,
      "loss": 0.0073,
      "step": 35910
    },
    {
      "epoch": 0.2303797224796606,
      "grad_norm": 0.2803209722042084,
      "learning_rate": 9.992975142534771e-06,
      "loss": 0.0053,
      "step": 35920
    },
    {
      "epoch": 0.2304438593734467,
      "grad_norm": 0.16461430490016937,
      "learning_rate": 9.992945452417396e-06,
      "loss": 0.0068,
      "step": 35930
    },
    {
      "epoch": 0.23050799626723278,
      "grad_norm": 0.3616570830345154,
      "learning_rate": 9.992915699734843e-06,
      "loss": 0.0068,
      "step": 35940
    },
    {
      "epoch": 0.23057213316101888,
      "grad_norm": 1.1020573377609253,
      "learning_rate": 9.992885884487488e-06,
      "loss": 0.0089,
      "step": 35950
    },
    {
      "epoch": 0.23063627005480497,
      "grad_norm": 0.2968498170375824,
      "learning_rate": 9.992856006675706e-06,
      "loss": 0.0054,
      "step": 35960
    },
    {
      "epoch": 0.23070040694859106,
      "grad_norm": 0.8094943761825562,
      "learning_rate": 9.99282606629987e-06,
      "loss": 0.0075,
      "step": 35970
    },
    {
      "epoch": 0.23076454384237716,
      "grad_norm": 0.528534471988678,
      "learning_rate": 9.992796063360354e-06,
      "loss": 0.0042,
      "step": 35980
    },
    {
      "epoch": 0.23082868073616328,
      "grad_norm": 0.27213770151138306,
      "learning_rate": 9.992765997857536e-06,
      "loss": 0.0111,
      "step": 35990
    },
    {
      "epoch": 0.23089281762994937,
      "grad_norm": 0.3508973717689514,
      "learning_rate": 9.992735869791792e-06,
      "loss": 0.0102,
      "step": 36000
    },
    {
      "epoch": 0.23095695452373546,
      "grad_norm": 0.2607177495956421,
      "learning_rate": 9.9927056791635e-06,
      "loss": 0.0056,
      "step": 36010
    },
    {
      "epoch": 0.23102109141752156,
      "grad_norm": 0.2572328746318817,
      "learning_rate": 9.992675425973035e-06,
      "loss": 0.0059,
      "step": 36020
    },
    {
      "epoch": 0.23108522831130765,
      "grad_norm": 0.10305281728506088,
      "learning_rate": 9.99264511022078e-06,
      "loss": 0.0068,
      "step": 36030
    },
    {
      "epoch": 0.23114936520509374,
      "grad_norm": 0.29231107234954834,
      "learning_rate": 9.992614731907114e-06,
      "loss": 0.0048,
      "step": 36040
    },
    {
      "epoch": 0.23121350209887984,
      "grad_norm": 0.4050886332988739,
      "learning_rate": 9.992584291032417e-06,
      "loss": 0.0056,
      "step": 36050
    },
    {
      "epoch": 0.23127763899266596,
      "grad_norm": 0.7118836641311646,
      "learning_rate": 9.99255378759707e-06,
      "loss": 0.0133,
      "step": 36060
    },
    {
      "epoch": 0.23134177588645205,
      "grad_norm": 0.22194357216358185,
      "learning_rate": 9.992523221601457e-06,
      "loss": 0.0057,
      "step": 36070
    },
    {
      "epoch": 0.23140591278023814,
      "grad_norm": 0.14988495409488678,
      "learning_rate": 9.992492593045958e-06,
      "loss": 0.0079,
      "step": 36080
    },
    {
      "epoch": 0.23147004967402424,
      "grad_norm": 0.24507054686546326,
      "learning_rate": 9.992461901930961e-06,
      "loss": 0.0064,
      "step": 36090
    },
    {
      "epoch": 0.23153418656781033,
      "grad_norm": 0.14381228387355804,
      "learning_rate": 9.992431148256847e-06,
      "loss": 0.007,
      "step": 36100
    },
    {
      "epoch": 0.23159832346159642,
      "grad_norm": 0.5826447606086731,
      "learning_rate": 9.992400332024004e-06,
      "loss": 0.0114,
      "step": 36110
    },
    {
      "epoch": 0.23166246035538252,
      "grad_norm": 0.30199071764945984,
      "learning_rate": 9.992369453232815e-06,
      "loss": 0.0066,
      "step": 36120
    },
    {
      "epoch": 0.23172659724916864,
      "grad_norm": 0.41684022545814514,
      "learning_rate": 9.99233851188367e-06,
      "loss": 0.0038,
      "step": 36130
    },
    {
      "epoch": 0.23179073414295473,
      "grad_norm": 0.10940320044755936,
      "learning_rate": 9.992307507976955e-06,
      "loss": 0.0084,
      "step": 36140
    },
    {
      "epoch": 0.23185487103674082,
      "grad_norm": 0.41128668189048767,
      "learning_rate": 9.992276441513061e-06,
      "loss": 0.0064,
      "step": 36150
    },
    {
      "epoch": 0.23191900793052692,
      "grad_norm": 0.6398462653160095,
      "learning_rate": 9.992245312492372e-06,
      "loss": 0.009,
      "step": 36160
    },
    {
      "epoch": 0.231983144824313,
      "grad_norm": 0.5770807266235352,
      "learning_rate": 9.992214120915282e-06,
      "loss": 0.0055,
      "step": 36170
    },
    {
      "epoch": 0.2320472817180991,
      "grad_norm": 0.46604737639427185,
      "learning_rate": 9.992182866782183e-06,
      "loss": 0.008,
      "step": 36180
    },
    {
      "epoch": 0.2321114186118852,
      "grad_norm": 0.25087466835975647,
      "learning_rate": 9.992151550093464e-06,
      "loss": 0.0047,
      "step": 36190
    },
    {
      "epoch": 0.23217555550567132,
      "grad_norm": 0.11956658959388733,
      "learning_rate": 9.992120170849517e-06,
      "loss": 0.0062,
      "step": 36200
    },
    {
      "epoch": 0.2322396923994574,
      "grad_norm": 0.2919653058052063,
      "learning_rate": 9.992088729050737e-06,
      "loss": 0.0073,
      "step": 36210
    },
    {
      "epoch": 0.2323038292932435,
      "grad_norm": 0.26989543437957764,
      "learning_rate": 9.992057224697518e-06,
      "loss": 0.0054,
      "step": 36220
    },
    {
      "epoch": 0.2323679661870296,
      "grad_norm": 0.1296006292104721,
      "learning_rate": 9.992025657790254e-06,
      "loss": 0.0044,
      "step": 36230
    },
    {
      "epoch": 0.2324321030808157,
      "grad_norm": 0.10228066891431808,
      "learning_rate": 9.991994028329341e-06,
      "loss": 0.0059,
      "step": 36240
    },
    {
      "epoch": 0.23249623997460178,
      "grad_norm": 0.23078081011772156,
      "learning_rate": 9.991962336315175e-06,
      "loss": 0.0042,
      "step": 36250
    },
    {
      "epoch": 0.23256037686838787,
      "grad_norm": 0.27274227142333984,
      "learning_rate": 9.991930581748154e-06,
      "loss": 0.0079,
      "step": 36260
    },
    {
      "epoch": 0.232624513762174,
      "grad_norm": 0.17164266109466553,
      "learning_rate": 9.991898764628674e-06,
      "loss": 0.0056,
      "step": 36270
    },
    {
      "epoch": 0.2326886506559601,
      "grad_norm": 0.1612294614315033,
      "learning_rate": 9.991866884957135e-06,
      "loss": 0.0055,
      "step": 36280
    },
    {
      "epoch": 0.23275278754974618,
      "grad_norm": 0.31363794207572937,
      "learning_rate": 9.991834942733936e-06,
      "loss": 0.0064,
      "step": 36290
    },
    {
      "epoch": 0.23281692444353227,
      "grad_norm": 0.3872830867767334,
      "learning_rate": 9.991802937959478e-06,
      "loss": 0.0058,
      "step": 36300
    },
    {
      "epoch": 0.23288106133731837,
      "grad_norm": 0.43687549233436584,
      "learning_rate": 9.991770870634163e-06,
      "loss": 0.0066,
      "step": 36310
    },
    {
      "epoch": 0.23294519823110446,
      "grad_norm": 0.5026020407676697,
      "learning_rate": 9.991738740758388e-06,
      "loss": 0.0066,
      "step": 36320
    },
    {
      "epoch": 0.23300933512489055,
      "grad_norm": 0.66605144739151,
      "learning_rate": 9.991706548332563e-06,
      "loss": 0.0043,
      "step": 36330
    },
    {
      "epoch": 0.23307347201867668,
      "grad_norm": 0.2193024456501007,
      "learning_rate": 9.991674293357085e-06,
      "loss": 0.0046,
      "step": 36340
    },
    {
      "epoch": 0.23313760891246277,
      "grad_norm": 0.2898307144641876,
      "learning_rate": 9.99164197583236e-06,
      "loss": 0.0041,
      "step": 36350
    },
    {
      "epoch": 0.23320174580624886,
      "grad_norm": 0.3485690653324127,
      "learning_rate": 9.991609595758796e-06,
      "loss": 0.0066,
      "step": 36360
    },
    {
      "epoch": 0.23326588270003495,
      "grad_norm": 0.4199317991733551,
      "learning_rate": 9.991577153136794e-06,
      "loss": 0.0057,
      "step": 36370
    },
    {
      "epoch": 0.23333001959382105,
      "grad_norm": 0.634312093257904,
      "learning_rate": 9.991544647966765e-06,
      "loss": 0.0038,
      "step": 36380
    },
    {
      "epoch": 0.23339415648760714,
      "grad_norm": 0.32351168990135193,
      "learning_rate": 9.991512080249111e-06,
      "loss": 0.0055,
      "step": 36390
    },
    {
      "epoch": 0.23345829338139323,
      "grad_norm": 0.16520749032497406,
      "learning_rate": 9.991479449984246e-06,
      "loss": 0.0037,
      "step": 36400
    },
    {
      "epoch": 0.23352243027517935,
      "grad_norm": 0.4310011863708496,
      "learning_rate": 9.991446757172575e-06,
      "loss": 0.0052,
      "step": 36410
    },
    {
      "epoch": 0.23358656716896545,
      "grad_norm": 0.3078899085521698,
      "learning_rate": 9.99141400181451e-06,
      "loss": 0.0061,
      "step": 36420
    },
    {
      "epoch": 0.23365070406275154,
      "grad_norm": 0.49376317858695984,
      "learning_rate": 9.991381183910459e-06,
      "loss": 0.0059,
      "step": 36430
    },
    {
      "epoch": 0.23371484095653763,
      "grad_norm": 0.3923306167125702,
      "learning_rate": 9.991348303460834e-06,
      "loss": 0.0077,
      "step": 36440
    },
    {
      "epoch": 0.23377897785032373,
      "grad_norm": 0.25772348046302795,
      "learning_rate": 9.99131536046605e-06,
      "loss": 0.0044,
      "step": 36450
    },
    {
      "epoch": 0.23384311474410982,
      "grad_norm": 0.244273841381073,
      "learning_rate": 9.991282354926515e-06,
      "loss": 0.0056,
      "step": 36460
    },
    {
      "epoch": 0.2339072516378959,
      "grad_norm": 0.27689307928085327,
      "learning_rate": 9.991249286842645e-06,
      "loss": 0.005,
      "step": 36470
    },
    {
      "epoch": 0.23397138853168203,
      "grad_norm": 0.14308667182922363,
      "learning_rate": 9.991216156214855e-06,
      "loss": 0.0068,
      "step": 36480
    },
    {
      "epoch": 0.23403552542546813,
      "grad_norm": 0.2195776104927063,
      "learning_rate": 9.991182963043558e-06,
      "loss": 0.0045,
      "step": 36490
    },
    {
      "epoch": 0.23409966231925422,
      "grad_norm": 0.20184378325939178,
      "learning_rate": 9.991149707329173e-06,
      "loss": 0.0088,
      "step": 36500
    },
    {
      "epoch": 0.2341637992130403,
      "grad_norm": 0.13293959200382233,
      "learning_rate": 9.991116389072115e-06,
      "loss": 0.0048,
      "step": 36510
    },
    {
      "epoch": 0.2342279361068264,
      "grad_norm": 0.10092591494321823,
      "learning_rate": 9.991083008272801e-06,
      "loss": 0.0081,
      "step": 36520
    },
    {
      "epoch": 0.2342920730006125,
      "grad_norm": 0.3733716309070587,
      "learning_rate": 9.99104956493165e-06,
      "loss": 0.0082,
      "step": 36530
    },
    {
      "epoch": 0.2343562098943986,
      "grad_norm": 0.5411534309387207,
      "learning_rate": 9.991016059049081e-06,
      "loss": 0.0125,
      "step": 36540
    },
    {
      "epoch": 0.2344203467881847,
      "grad_norm": 0.4143863022327423,
      "learning_rate": 9.990982490625514e-06,
      "loss": 0.0065,
      "step": 36550
    },
    {
      "epoch": 0.2344844836819708,
      "grad_norm": 0.15255770087242126,
      "learning_rate": 9.990948859661367e-06,
      "loss": 0.0066,
      "step": 36560
    },
    {
      "epoch": 0.2345486205757569,
      "grad_norm": 0.1986321359872818,
      "learning_rate": 9.990915166157067e-06,
      "loss": 0.0075,
      "step": 36570
    },
    {
      "epoch": 0.234612757469543,
      "grad_norm": 0.26919758319854736,
      "learning_rate": 9.990881410113032e-06,
      "loss": 0.0057,
      "step": 36580
    },
    {
      "epoch": 0.23467689436332909,
      "grad_norm": 0.24907876551151276,
      "learning_rate": 9.990847591529685e-06,
      "loss": 0.0037,
      "step": 36590
    },
    {
      "epoch": 0.23474103125711518,
      "grad_norm": 0.13349837064743042,
      "learning_rate": 9.990813710407453e-06,
      "loss": 0.0054,
      "step": 36600
    },
    {
      "epoch": 0.23480516815090127,
      "grad_norm": 0.1132342666387558,
      "learning_rate": 9.990779766746757e-06,
      "loss": 0.0044,
      "step": 36610
    },
    {
      "epoch": 0.2348693050446874,
      "grad_norm": 0.446804016828537,
      "learning_rate": 9.990745760548023e-06,
      "loss": 0.0064,
      "step": 36620
    },
    {
      "epoch": 0.2349334419384735,
      "grad_norm": 0.40333107113838196,
      "learning_rate": 9.990711691811679e-06,
      "loss": 0.0064,
      "step": 36630
    },
    {
      "epoch": 0.23499757883225958,
      "grad_norm": 0.23259688913822174,
      "learning_rate": 9.990677560538149e-06,
      "loss": 0.0059,
      "step": 36640
    },
    {
      "epoch": 0.23506171572604567,
      "grad_norm": 0.3403295874595642,
      "learning_rate": 9.990643366727863e-06,
      "loss": 0.0042,
      "step": 36650
    },
    {
      "epoch": 0.23512585261983177,
      "grad_norm": 0.21020762622356415,
      "learning_rate": 9.99060911038125e-06,
      "loss": 0.0078,
      "step": 36660
    },
    {
      "epoch": 0.23518998951361786,
      "grad_norm": 0.170172318816185,
      "learning_rate": 9.990574791498737e-06,
      "loss": 0.0073,
      "step": 36670
    },
    {
      "epoch": 0.23525412640740395,
      "grad_norm": 0.17274317145347595,
      "learning_rate": 9.990540410080755e-06,
      "loss": 0.0051,
      "step": 36680
    },
    {
      "epoch": 0.23531826330119007,
      "grad_norm": 0.1390146017074585,
      "learning_rate": 9.990505966127735e-06,
      "loss": 0.0067,
      "step": 36690
    },
    {
      "epoch": 0.23538240019497617,
      "grad_norm": 0.34374529123306274,
      "learning_rate": 9.990471459640109e-06,
      "loss": 0.0078,
      "step": 36700
    },
    {
      "epoch": 0.23544653708876226,
      "grad_norm": 0.4592181146144867,
      "learning_rate": 9.99043689061831e-06,
      "loss": 0.0063,
      "step": 36710
    },
    {
      "epoch": 0.23551067398254835,
      "grad_norm": 0.3880027234554291,
      "learning_rate": 9.99040225906277e-06,
      "loss": 0.0054,
      "step": 36720
    },
    {
      "epoch": 0.23557481087633445,
      "grad_norm": 0.11227881163358688,
      "learning_rate": 9.990367564973922e-06,
      "loss": 0.0058,
      "step": 36730
    },
    {
      "epoch": 0.23563894777012054,
      "grad_norm": 0.3381422460079193,
      "learning_rate": 9.990332808352203e-06,
      "loss": 0.0056,
      "step": 36740
    },
    {
      "epoch": 0.23570308466390663,
      "grad_norm": 0.18786120414733887,
      "learning_rate": 9.990297989198046e-06,
      "loss": 0.0043,
      "step": 36750
    },
    {
      "epoch": 0.23576722155769275,
      "grad_norm": 0.2107134759426117,
      "learning_rate": 9.990263107511888e-06,
      "loss": 0.007,
      "step": 36760
    },
    {
      "epoch": 0.23583135845147885,
      "grad_norm": 0.22882787883281708,
      "learning_rate": 9.990228163294169e-06,
      "loss": 0.0057,
      "step": 36770
    },
    {
      "epoch": 0.23589549534526494,
      "grad_norm": 0.28724777698516846,
      "learning_rate": 9.990193156545323e-06,
      "loss": 0.0055,
      "step": 36780
    },
    {
      "epoch": 0.23595963223905103,
      "grad_norm": 0.45361918210983276,
      "learning_rate": 9.990158087265791e-06,
      "loss": 0.006,
      "step": 36790
    },
    {
      "epoch": 0.23602376913283712,
      "grad_norm": 0.23359541594982147,
      "learning_rate": 9.990122955456012e-06,
      "loss": 0.0086,
      "step": 36800
    },
    {
      "epoch": 0.23608790602662322,
      "grad_norm": 0.1540917605161667,
      "learning_rate": 9.990087761116426e-06,
      "loss": 0.0054,
      "step": 36810
    },
    {
      "epoch": 0.2361520429204093,
      "grad_norm": 0.37155643105506897,
      "learning_rate": 9.990052504247473e-06,
      "loss": 0.0054,
      "step": 36820
    },
    {
      "epoch": 0.23621617981419543,
      "grad_norm": 0.07868891954421997,
      "learning_rate": 9.990017184849597e-06,
      "loss": 0.0045,
      "step": 36830
    },
    {
      "epoch": 0.23628031670798152,
      "grad_norm": 0.1814006119966507,
      "learning_rate": 9.989981802923238e-06,
      "loss": 0.0055,
      "step": 36840
    },
    {
      "epoch": 0.23634445360176762,
      "grad_norm": 0.3288971483707428,
      "learning_rate": 9.989946358468843e-06,
      "loss": 0.0071,
      "step": 36850
    },
    {
      "epoch": 0.2364085904955537,
      "grad_norm": 0.25540879368782043,
      "learning_rate": 9.989910851486853e-06,
      "loss": 0.0047,
      "step": 36860
    },
    {
      "epoch": 0.2364727273893398,
      "grad_norm": 0.18015730381011963,
      "learning_rate": 9.989875281977713e-06,
      "loss": 0.0057,
      "step": 36870
    },
    {
      "epoch": 0.2365368642831259,
      "grad_norm": 0.26143142580986023,
      "learning_rate": 9.989839649941867e-06,
      "loss": 0.0071,
      "step": 36880
    },
    {
      "epoch": 0.236601001176912,
      "grad_norm": 0.16240955889225006,
      "learning_rate": 9.98980395537977e-06,
      "loss": 0.0087,
      "step": 36890
    },
    {
      "epoch": 0.2366651380706981,
      "grad_norm": 0.3834659159183502,
      "learning_rate": 9.989768198291858e-06,
      "loss": 0.0082,
      "step": 36900
    },
    {
      "epoch": 0.2367292749644842,
      "grad_norm": 0.16273871064186096,
      "learning_rate": 9.989732378678587e-06,
      "loss": 0.0035,
      "step": 36910
    },
    {
      "epoch": 0.2367934118582703,
      "grad_norm": 0.2915313243865967,
      "learning_rate": 9.989696496540401e-06,
      "loss": 0.0065,
      "step": 36920
    },
    {
      "epoch": 0.2368575487520564,
      "grad_norm": 0.2651917338371277,
      "learning_rate": 9.989660551877755e-06,
      "loss": 0.0044,
      "step": 36930
    },
    {
      "epoch": 0.23692168564584248,
      "grad_norm": 0.312575101852417,
      "learning_rate": 9.989624544691095e-06,
      "loss": 0.007,
      "step": 36940
    },
    {
      "epoch": 0.23698582253962858,
      "grad_norm": 0.049576129764318466,
      "learning_rate": 9.989588474980873e-06,
      "loss": 0.0068,
      "step": 36950
    },
    {
      "epoch": 0.23704995943341467,
      "grad_norm": 0.13811209797859192,
      "learning_rate": 9.989552342747542e-06,
      "loss": 0.0035,
      "step": 36960
    },
    {
      "epoch": 0.2371140963272008,
      "grad_norm": 0.1147325187921524,
      "learning_rate": 9.989516147991552e-06,
      "loss": 0.0187,
      "step": 36970
    },
    {
      "epoch": 0.23717823322098688,
      "grad_norm": 0.4486019015312195,
      "learning_rate": 9.989479890713361e-06,
      "loss": 0.0066,
      "step": 36980
    },
    {
      "epoch": 0.23724237011477298,
      "grad_norm": 0.13639159500598907,
      "learning_rate": 9.989443570913419e-06,
      "loss": 0.0043,
      "step": 36990
    },
    {
      "epoch": 0.23730650700855907,
      "grad_norm": 0.531525731086731,
      "learning_rate": 9.989407188592186e-06,
      "loss": 0.0057,
      "step": 37000
    },
    {
      "epoch": 0.23737064390234516,
      "grad_norm": 0.34439024329185486,
      "learning_rate": 9.989370743750113e-06,
      "loss": 0.0063,
      "step": 37010
    },
    {
      "epoch": 0.23743478079613126,
      "grad_norm": 0.36032649874687195,
      "learning_rate": 9.98933423638766e-06,
      "loss": 0.0045,
      "step": 37020
    },
    {
      "epoch": 0.23749891768991735,
      "grad_norm": 0.3823663294315338,
      "learning_rate": 9.989297666505282e-06,
      "loss": 0.0056,
      "step": 37030
    },
    {
      "epoch": 0.23756305458370347,
      "grad_norm": 0.2125890702009201,
      "learning_rate": 9.98926103410344e-06,
      "loss": 0.0049,
      "step": 37040
    },
    {
      "epoch": 0.23762719147748956,
      "grad_norm": 0.32678574323654175,
      "learning_rate": 9.98922433918259e-06,
      "loss": 0.0053,
      "step": 37050
    },
    {
      "epoch": 0.23769132837127566,
      "grad_norm": 0.20189277827739716,
      "learning_rate": 9.989187581743194e-06,
      "loss": 0.0038,
      "step": 37060
    },
    {
      "epoch": 0.23775546526506175,
      "grad_norm": 0.25536254048347473,
      "learning_rate": 9.98915076178571e-06,
      "loss": 0.0069,
      "step": 37070
    },
    {
      "epoch": 0.23781960215884784,
      "grad_norm": 0.441403865814209,
      "learning_rate": 9.989113879310604e-06,
      "loss": 0.0056,
      "step": 37080
    },
    {
      "epoch": 0.23788373905263394,
      "grad_norm": 0.31718453764915466,
      "learning_rate": 9.989076934318334e-06,
      "loss": 0.004,
      "step": 37090
    },
    {
      "epoch": 0.23794787594642003,
      "grad_norm": 0.4428984224796295,
      "learning_rate": 9.989039926809363e-06,
      "loss": 0.0036,
      "step": 37100
    },
    {
      "epoch": 0.23801201284020615,
      "grad_norm": 0.05789393559098244,
      "learning_rate": 9.989002856784159e-06,
      "loss": 0.0023,
      "step": 37110
    },
    {
      "epoch": 0.23807614973399224,
      "grad_norm": 0.49261847138404846,
      "learning_rate": 9.988965724243183e-06,
      "loss": 0.0056,
      "step": 37120
    },
    {
      "epoch": 0.23814028662777834,
      "grad_norm": 0.36797863245010376,
      "learning_rate": 9.9889285291869e-06,
      "loss": 0.0059,
      "step": 37130
    },
    {
      "epoch": 0.23820442352156443,
      "grad_norm": 0.2522566020488739,
      "learning_rate": 9.988891271615777e-06,
      "loss": 0.0062,
      "step": 37140
    },
    {
      "epoch": 0.23826856041535052,
      "grad_norm": 0.43269291520118713,
      "learning_rate": 9.98885395153028e-06,
      "loss": 0.0061,
      "step": 37150
    },
    {
      "epoch": 0.23833269730913662,
      "grad_norm": 0.3380693197250366,
      "learning_rate": 9.988816568930879e-06,
      "loss": 0.0083,
      "step": 37160
    },
    {
      "epoch": 0.2383968342029227,
      "grad_norm": 0.1102917343378067,
      "learning_rate": 9.98877912381804e-06,
      "loss": 0.0096,
      "step": 37170
    },
    {
      "epoch": 0.2384609710967088,
      "grad_norm": 0.1320537030696869,
      "learning_rate": 9.988741616192234e-06,
      "loss": 0.0099,
      "step": 37180
    },
    {
      "epoch": 0.23852510799049492,
      "grad_norm": 0.24336595833301544,
      "learning_rate": 9.98870404605393e-06,
      "loss": 0.006,
      "step": 37190
    },
    {
      "epoch": 0.23858924488428102,
      "grad_norm": 0.2764676511287689,
      "learning_rate": 9.9886664134036e-06,
      "loss": 0.0082,
      "step": 37200
    },
    {
      "epoch": 0.2386533817780671,
      "grad_norm": 0.24236895143985748,
      "learning_rate": 9.988628718241712e-06,
      "loss": 0.0063,
      "step": 37210
    },
    {
      "epoch": 0.2387175186718532,
      "grad_norm": 0.2600942552089691,
      "learning_rate": 9.988590960568743e-06,
      "loss": 0.0059,
      "step": 37220
    },
    {
      "epoch": 0.2387816555656393,
      "grad_norm": 0.1426173597574234,
      "learning_rate": 9.988553140385163e-06,
      "loss": 0.0048,
      "step": 37230
    },
    {
      "epoch": 0.2388457924594254,
      "grad_norm": 0.31119289994239807,
      "learning_rate": 9.988515257691446e-06,
      "loss": 0.009,
      "step": 37240
    },
    {
      "epoch": 0.23890992935321148,
      "grad_norm": 0.26968684792518616,
      "learning_rate": 9.988477312488069e-06,
      "loss": 0.004,
      "step": 37250
    },
    {
      "epoch": 0.2389740662469976,
      "grad_norm": 0.3957492411136627,
      "learning_rate": 9.988439304775506e-06,
      "loss": 0.0045,
      "step": 37260
    },
    {
      "epoch": 0.2390382031407837,
      "grad_norm": 0.10733895003795624,
      "learning_rate": 9.988401234554233e-06,
      "loss": 0.0042,
      "step": 37270
    },
    {
      "epoch": 0.2391023400345698,
      "grad_norm": 0.3841085135936737,
      "learning_rate": 9.988363101824727e-06,
      "loss": 0.0053,
      "step": 37280
    },
    {
      "epoch": 0.23916647692835588,
      "grad_norm": 0.33194515109062195,
      "learning_rate": 9.988324906587467e-06,
      "loss": 0.0049,
      "step": 37290
    },
    {
      "epoch": 0.23923061382214197,
      "grad_norm": 0.4566282033920288,
      "learning_rate": 9.98828664884293e-06,
      "loss": 0.0054,
      "step": 37300
    },
    {
      "epoch": 0.23929475071592807,
      "grad_norm": 0.17479823529720306,
      "learning_rate": 9.988248328591597e-06,
      "loss": 0.0058,
      "step": 37310
    },
    {
      "epoch": 0.23935888760971416,
      "grad_norm": 0.2093168944120407,
      "learning_rate": 9.988209945833947e-06,
      "loss": 0.0079,
      "step": 37320
    },
    {
      "epoch": 0.23942302450350028,
      "grad_norm": 0.511039137840271,
      "learning_rate": 9.98817150057046e-06,
      "loss": 0.0047,
      "step": 37330
    },
    {
      "epoch": 0.23948716139728637,
      "grad_norm": 0.2602264881134033,
      "learning_rate": 9.988132992801623e-06,
      "loss": 0.0076,
      "step": 37340
    },
    {
      "epoch": 0.23955129829107247,
      "grad_norm": 0.2345946729183197,
      "learning_rate": 9.98809442252791e-06,
      "loss": 0.0039,
      "step": 37350
    },
    {
      "epoch": 0.23961543518485856,
      "grad_norm": 0.3673745095729828,
      "learning_rate": 9.98805578974981e-06,
      "loss": 0.0059,
      "step": 37360
    },
    {
      "epoch": 0.23967957207864465,
      "grad_norm": 0.424146831035614,
      "learning_rate": 9.988017094467809e-06,
      "loss": 0.0044,
      "step": 37370
    },
    {
      "epoch": 0.23974370897243075,
      "grad_norm": 0.13633568584918976,
      "learning_rate": 9.987978336682388e-06,
      "loss": 0.007,
      "step": 37380
    },
    {
      "epoch": 0.23980784586621684,
      "grad_norm": 0.20849905908107758,
      "learning_rate": 9.987939516394032e-06,
      "loss": 0.0048,
      "step": 37390
    },
    {
      "epoch": 0.23987198276000296,
      "grad_norm": 0.3548513352870941,
      "learning_rate": 9.98790063360323e-06,
      "loss": 0.0059,
      "step": 37400
    },
    {
      "epoch": 0.23993611965378905,
      "grad_norm": 0.5116329193115234,
      "learning_rate": 9.987861688310467e-06,
      "loss": 0.008,
      "step": 37410
    },
    {
      "epoch": 0.24000025654757515,
      "grad_norm": 0.28186020255088806,
      "learning_rate": 9.987822680516232e-06,
      "loss": 0.0035,
      "step": 37420
    },
    {
      "epoch": 0.24006439344136124,
      "grad_norm": 0.24905213713645935,
      "learning_rate": 9.987783610221015e-06,
      "loss": 0.0054,
      "step": 37430
    },
    {
      "epoch": 0.24012853033514733,
      "grad_norm": 0.15364207327365875,
      "learning_rate": 9.987744477425306e-06,
      "loss": 0.0055,
      "step": 37440
    },
    {
      "epoch": 0.24019266722893343,
      "grad_norm": 0.24339143931865692,
      "learning_rate": 9.987705282129593e-06,
      "loss": 0.0077,
      "step": 37450
    },
    {
      "epoch": 0.24025680412271952,
      "grad_norm": 0.07584648579359055,
      "learning_rate": 9.987666024334368e-06,
      "loss": 0.008,
      "step": 37460
    },
    {
      "epoch": 0.24032094101650564,
      "grad_norm": 0.24445678293704987,
      "learning_rate": 9.987626704040123e-06,
      "loss": 0.0044,
      "step": 37470
    },
    {
      "epoch": 0.24038507791029173,
      "grad_norm": 0.1876668930053711,
      "learning_rate": 9.98758732124735e-06,
      "loss": 0.0048,
      "step": 37480
    },
    {
      "epoch": 0.24044921480407783,
      "grad_norm": 0.3630131185054779,
      "learning_rate": 9.987547875956545e-06,
      "loss": 0.0053,
      "step": 37490
    },
    {
      "epoch": 0.24051335169786392,
      "grad_norm": 0.2916017174720764,
      "learning_rate": 9.987508368168199e-06,
      "loss": 0.0075,
      "step": 37500
    },
    {
      "epoch": 0.24057748859165,
      "grad_norm": 0.5088058710098267,
      "learning_rate": 9.98746879788281e-06,
      "loss": 0.0088,
      "step": 37510
    },
    {
      "epoch": 0.2406416254854361,
      "grad_norm": 0.5135015249252319,
      "learning_rate": 9.987429165100873e-06,
      "loss": 0.0065,
      "step": 37520
    },
    {
      "epoch": 0.2407057623792222,
      "grad_norm": 0.36910855770111084,
      "learning_rate": 9.98738946982288e-06,
      "loss": 0.0084,
      "step": 37530
    },
    {
      "epoch": 0.24076989927300832,
      "grad_norm": 0.20661070942878723,
      "learning_rate": 9.987349712049337e-06,
      "loss": 0.006,
      "step": 37540
    },
    {
      "epoch": 0.2408340361667944,
      "grad_norm": 0.21455512940883636,
      "learning_rate": 9.987309891780737e-06,
      "loss": 0.0055,
      "step": 37550
    },
    {
      "epoch": 0.2408981730605805,
      "grad_norm": 0.24813826382160187,
      "learning_rate": 9.987270009017578e-06,
      "loss": 0.0039,
      "step": 37560
    },
    {
      "epoch": 0.2409623099543666,
      "grad_norm": 0.1844363510608673,
      "learning_rate": 9.987230063760363e-06,
      "loss": 0.0058,
      "step": 37570
    },
    {
      "epoch": 0.2410264468481527,
      "grad_norm": 0.5201581120491028,
      "learning_rate": 9.98719005600959e-06,
      "loss": 0.0052,
      "step": 37580
    },
    {
      "epoch": 0.24109058374193879,
      "grad_norm": 0.33404937386512756,
      "learning_rate": 9.987149985765762e-06,
      "loss": 0.006,
      "step": 37590
    },
    {
      "epoch": 0.24115472063572488,
      "grad_norm": 0.18592841923236847,
      "learning_rate": 9.987109853029379e-06,
      "loss": 0.0107,
      "step": 37600
    },
    {
      "epoch": 0.241218857529511,
      "grad_norm": 0.3620583117008209,
      "learning_rate": 9.987069657800947e-06,
      "loss": 0.0051,
      "step": 37610
    },
    {
      "epoch": 0.2412829944232971,
      "grad_norm": 0.4495629668235779,
      "learning_rate": 9.987029400080968e-06,
      "loss": 0.0055,
      "step": 37620
    },
    {
      "epoch": 0.24134713131708319,
      "grad_norm": 0.17936894297599792,
      "learning_rate": 9.986989079869944e-06,
      "loss": 0.0066,
      "step": 37630
    },
    {
      "epoch": 0.24141126821086928,
      "grad_norm": 0.0979299545288086,
      "learning_rate": 9.986948697168386e-06,
      "loss": 0.0048,
      "step": 37640
    },
    {
      "epoch": 0.24147540510465537,
      "grad_norm": 0.2822265625,
      "learning_rate": 9.986908251976794e-06,
      "loss": 0.0063,
      "step": 37650
    },
    {
      "epoch": 0.24153954199844146,
      "grad_norm": 0.360325425863266,
      "learning_rate": 9.986867744295678e-06,
      "loss": 0.0076,
      "step": 37660
    },
    {
      "epoch": 0.24160367889222756,
      "grad_norm": 0.3158207833766937,
      "learning_rate": 9.986827174125547e-06,
      "loss": 0.0041,
      "step": 37670
    },
    {
      "epoch": 0.24166781578601368,
      "grad_norm": 0.203022301197052,
      "learning_rate": 9.986786541466907e-06,
      "loss": 0.0039,
      "step": 37680
    },
    {
      "epoch": 0.24173195267979977,
      "grad_norm": 0.09906020760536194,
      "learning_rate": 9.986745846320266e-06,
      "loss": 0.0051,
      "step": 37690
    },
    {
      "epoch": 0.24179608957358587,
      "grad_norm": 0.29850175976753235,
      "learning_rate": 9.986705088686136e-06,
      "loss": 0.0055,
      "step": 37700
    },
    {
      "epoch": 0.24186022646737196,
      "grad_norm": 0.2408210039138794,
      "learning_rate": 9.986664268565027e-06,
      "loss": 0.0064,
      "step": 37710
    },
    {
      "epoch": 0.24192436336115805,
      "grad_norm": 0.21141502261161804,
      "learning_rate": 9.986623385957452e-06,
      "loss": 0.0045,
      "step": 37720
    },
    {
      "epoch": 0.24198850025494414,
      "grad_norm": 0.26400238275527954,
      "learning_rate": 9.986582440863923e-06,
      "loss": 0.0048,
      "step": 37730
    },
    {
      "epoch": 0.24205263714873024,
      "grad_norm": 0.33857569098472595,
      "learning_rate": 9.98654143328495e-06,
      "loss": 0.0039,
      "step": 37740
    },
    {
      "epoch": 0.24211677404251636,
      "grad_norm": 0.25933921337127686,
      "learning_rate": 9.98650036322105e-06,
      "loss": 0.0041,
      "step": 37750
    },
    {
      "epoch": 0.24218091093630245,
      "grad_norm": 0.16440115869045258,
      "learning_rate": 9.986459230672736e-06,
      "loss": 0.0039,
      "step": 37760
    },
    {
      "epoch": 0.24224504783008854,
      "grad_norm": 0.11378602683544159,
      "learning_rate": 9.986418035640523e-06,
      "loss": 0.0072,
      "step": 37770
    },
    {
      "epoch": 0.24230918472387464,
      "grad_norm": 0.1978156566619873,
      "learning_rate": 9.98637677812493e-06,
      "loss": 0.0037,
      "step": 37780
    },
    {
      "epoch": 0.24237332161766073,
      "grad_norm": 0.24849757552146912,
      "learning_rate": 9.986335458126472e-06,
      "loss": 0.006,
      "step": 37790
    },
    {
      "epoch": 0.24243745851144682,
      "grad_norm": 0.25591734051704407,
      "learning_rate": 9.986294075645668e-06,
      "loss": 0.0057,
      "step": 37800
    },
    {
      "epoch": 0.24250159540523292,
      "grad_norm": 0.10914437472820282,
      "learning_rate": 9.986252630683035e-06,
      "loss": 0.005,
      "step": 37810
    },
    {
      "epoch": 0.24256573229901904,
      "grad_norm": 0.33318716287612915,
      "learning_rate": 9.986211123239092e-06,
      "loss": 0.0077,
      "step": 37820
    },
    {
      "epoch": 0.24262986919280513,
      "grad_norm": 0.20896196365356445,
      "learning_rate": 9.98616955331436e-06,
      "loss": 0.004,
      "step": 37830
    },
    {
      "epoch": 0.24269400608659122,
      "grad_norm": 0.525875449180603,
      "learning_rate": 9.986127920909361e-06,
      "loss": 0.0115,
      "step": 37840
    },
    {
      "epoch": 0.24275814298037732,
      "grad_norm": 0.25195562839508057,
      "learning_rate": 9.986086226024615e-06,
      "loss": 0.0113,
      "step": 37850
    },
    {
      "epoch": 0.2428222798741634,
      "grad_norm": 0.48433178663253784,
      "learning_rate": 9.986044468660646e-06,
      "loss": 0.007,
      "step": 37860
    },
    {
      "epoch": 0.2428864167679495,
      "grad_norm": 0.19795604050159454,
      "learning_rate": 9.986002648817976e-06,
      "loss": 0.0052,
      "step": 37870
    },
    {
      "epoch": 0.2429505536617356,
      "grad_norm": 0.1475193351507187,
      "learning_rate": 9.985960766497129e-06,
      "loss": 0.006,
      "step": 37880
    },
    {
      "epoch": 0.24301469055552172,
      "grad_norm": 0.1122908964753151,
      "learning_rate": 9.98591882169863e-06,
      "loss": 0.0049,
      "step": 37890
    },
    {
      "epoch": 0.2430788274493078,
      "grad_norm": 0.22580060362815857,
      "learning_rate": 9.985876814423005e-06,
      "loss": 0.0068,
      "step": 37900
    },
    {
      "epoch": 0.2431429643430939,
      "grad_norm": 0.17058011889457703,
      "learning_rate": 9.985834744670779e-06,
      "loss": 0.0056,
      "step": 37910
    },
    {
      "epoch": 0.24320710123688,
      "grad_norm": 0.46447932720184326,
      "learning_rate": 9.985792612442482e-06,
      "loss": 0.0087,
      "step": 37920
    },
    {
      "epoch": 0.2432712381306661,
      "grad_norm": 0.07212353497743607,
      "learning_rate": 9.98575041773864e-06,
      "loss": 0.0051,
      "step": 37930
    },
    {
      "epoch": 0.24333537502445218,
      "grad_norm": 0.2533189356327057,
      "learning_rate": 9.985708160559781e-06,
      "loss": 0.0063,
      "step": 37940
    },
    {
      "epoch": 0.24339951191823828,
      "grad_norm": 0.2570257782936096,
      "learning_rate": 9.985665840906437e-06,
      "loss": 0.0099,
      "step": 37950
    },
    {
      "epoch": 0.2434636488120244,
      "grad_norm": 0.2666739821434021,
      "learning_rate": 9.985623458779136e-06,
      "loss": 0.0043,
      "step": 37960
    },
    {
      "epoch": 0.2435277857058105,
      "grad_norm": 0.31614717841148376,
      "learning_rate": 9.98558101417841e-06,
      "loss": 0.0043,
      "step": 37970
    },
    {
      "epoch": 0.24359192259959658,
      "grad_norm": 0.4189772605895996,
      "learning_rate": 9.98553850710479e-06,
      "loss": 0.0085,
      "step": 37980
    },
    {
      "epoch": 0.24365605949338268,
      "grad_norm": 0.07213608175516129,
      "learning_rate": 9.98549593755881e-06,
      "loss": 0.0079,
      "step": 37990
    },
    {
      "epoch": 0.24372019638716877,
      "grad_norm": 0.45046406984329224,
      "learning_rate": 9.985453305541002e-06,
      "loss": 0.0059,
      "step": 38000
    },
    {
      "epoch": 0.24378433328095486,
      "grad_norm": 0.2924140393733978,
      "learning_rate": 9.985410611051903e-06,
      "loss": 0.0068,
      "step": 38010
    },
    {
      "epoch": 0.24384847017474096,
      "grad_norm": 0.13637924194335938,
      "learning_rate": 9.985367854092044e-06,
      "loss": 0.0043,
      "step": 38020
    },
    {
      "epoch": 0.24391260706852708,
      "grad_norm": 0.07379632443189621,
      "learning_rate": 9.985325034661965e-06,
      "loss": 0.0041,
      "step": 38030
    },
    {
      "epoch": 0.24397674396231317,
      "grad_norm": 0.14553886651992798,
      "learning_rate": 9.985282152762199e-06,
      "loss": 0.0055,
      "step": 38040
    },
    {
      "epoch": 0.24404088085609926,
      "grad_norm": 0.12040738761425018,
      "learning_rate": 9.985239208393285e-06,
      "loss": 0.0041,
      "step": 38050
    },
    {
      "epoch": 0.24410501774988536,
      "grad_norm": 0.3329266905784607,
      "learning_rate": 9.98519620155576e-06,
      "loss": 0.0095,
      "step": 38060
    },
    {
      "epoch": 0.24416915464367145,
      "grad_norm": 0.2862496078014374,
      "learning_rate": 9.985153132250165e-06,
      "loss": 0.0057,
      "step": 38070
    },
    {
      "epoch": 0.24423329153745754,
      "grad_norm": 0.16985680162906647,
      "learning_rate": 9.985110000477036e-06,
      "loss": 0.0046,
      "step": 38080
    },
    {
      "epoch": 0.24429742843124364,
      "grad_norm": 0.532107949256897,
      "learning_rate": 9.985066806236919e-06,
      "loss": 0.0056,
      "step": 38090
    },
    {
      "epoch": 0.24436156532502976,
      "grad_norm": 0.28466662764549255,
      "learning_rate": 9.985023549530351e-06,
      "loss": 0.0072,
      "step": 38100
    },
    {
      "epoch": 0.24442570221881585,
      "grad_norm": 0.1789170801639557,
      "learning_rate": 9.984980230357875e-06,
      "loss": 0.0043,
      "step": 38110
    },
    {
      "epoch": 0.24448983911260194,
      "grad_norm": 0.18516568839550018,
      "learning_rate": 9.984936848720032e-06,
      "loss": 0.0087,
      "step": 38120
    },
    {
      "epoch": 0.24455397600638804,
      "grad_norm": 0.29143643379211426,
      "learning_rate": 9.98489340461737e-06,
      "loss": 0.0065,
      "step": 38130
    },
    {
      "epoch": 0.24461811290017413,
      "grad_norm": 0.2915891706943512,
      "learning_rate": 9.98484989805043e-06,
      "loss": 0.0092,
      "step": 38140
    },
    {
      "epoch": 0.24468224979396022,
      "grad_norm": 0.16607245802879333,
      "learning_rate": 9.984806329019761e-06,
      "loss": 0.006,
      "step": 38150
    },
    {
      "epoch": 0.24474638668774631,
      "grad_norm": 0.21408230066299438,
      "learning_rate": 9.984762697525903e-06,
      "loss": 0.0027,
      "step": 38160
    },
    {
      "epoch": 0.24481052358153244,
      "grad_norm": 0.28983208537101746,
      "learning_rate": 9.984719003569407e-06,
      "loss": 0.0064,
      "step": 38170
    },
    {
      "epoch": 0.24487466047531853,
      "grad_norm": 0.3070462942123413,
      "learning_rate": 9.984675247150818e-06,
      "loss": 0.0099,
      "step": 38180
    },
    {
      "epoch": 0.24493879736910462,
      "grad_norm": 0.1305485963821411,
      "learning_rate": 9.984631428270687e-06,
      "loss": 0.0051,
      "step": 38190
    },
    {
      "epoch": 0.24500293426289071,
      "grad_norm": 0.21113575994968414,
      "learning_rate": 9.984587546929562e-06,
      "loss": 0.0072,
      "step": 38200
    },
    {
      "epoch": 0.2450670711566768,
      "grad_norm": 0.19226740300655365,
      "learning_rate": 9.984543603127992e-06,
      "loss": 0.0042,
      "step": 38210
    },
    {
      "epoch": 0.2451312080504629,
      "grad_norm": 0.36949214339256287,
      "learning_rate": 9.984499596866528e-06,
      "loss": 0.0062,
      "step": 38220
    },
    {
      "epoch": 0.245195344944249,
      "grad_norm": 0.5497113466262817,
      "learning_rate": 9.984455528145723e-06,
      "loss": 0.0061,
      "step": 38230
    },
    {
      "epoch": 0.24525948183803511,
      "grad_norm": 0.21563158929347992,
      "learning_rate": 9.984411396966125e-06,
      "loss": 0.0055,
      "step": 38240
    },
    {
      "epoch": 0.2453236187318212,
      "grad_norm": 0.19525288045406342,
      "learning_rate": 9.984367203328292e-06,
      "loss": 0.0064,
      "step": 38250
    },
    {
      "epoch": 0.2453877556256073,
      "grad_norm": 0.32135501503944397,
      "learning_rate": 9.984322947232776e-06,
      "loss": 0.0063,
      "step": 38260
    },
    {
      "epoch": 0.2454518925193934,
      "grad_norm": 0.43099191784858704,
      "learning_rate": 9.98427862868013e-06,
      "loss": 0.0087,
      "step": 38270
    },
    {
      "epoch": 0.2455160294131795,
      "grad_norm": 0.5807035565376282,
      "learning_rate": 9.984234247670912e-06,
      "loss": 0.0058,
      "step": 38280
    },
    {
      "epoch": 0.24558016630696558,
      "grad_norm": 0.24664615094661713,
      "learning_rate": 9.984189804205676e-06,
      "loss": 0.0037,
      "step": 38290
    },
    {
      "epoch": 0.24564430320075167,
      "grad_norm": 0.4822005331516266,
      "learning_rate": 9.98414529828498e-06,
      "loss": 0.0117,
      "step": 38300
    },
    {
      "epoch": 0.2457084400945378,
      "grad_norm": 0.22811010479927063,
      "learning_rate": 9.98410072990938e-06,
      "loss": 0.0057,
      "step": 38310
    },
    {
      "epoch": 0.2457725769883239,
      "grad_norm": 0.2250605970621109,
      "learning_rate": 9.984056099079437e-06,
      "loss": 0.011,
      "step": 38320
    },
    {
      "epoch": 0.24583671388210998,
      "grad_norm": 0.48564302921295166,
      "learning_rate": 9.984011405795709e-06,
      "loss": 0.0077,
      "step": 38330
    },
    {
      "epoch": 0.24590085077589607,
      "grad_norm": 0.5844722986221313,
      "learning_rate": 9.983966650058756e-06,
      "loss": 0.009,
      "step": 38340
    },
    {
      "epoch": 0.24596498766968217,
      "grad_norm": 0.20418652892112732,
      "learning_rate": 9.983921831869137e-06,
      "loss": 0.0058,
      "step": 38350
    },
    {
      "epoch": 0.24602912456346826,
      "grad_norm": 0.39494210481643677,
      "learning_rate": 9.983876951227418e-06,
      "loss": 0.0057,
      "step": 38360
    },
    {
      "epoch": 0.24609326145725435,
      "grad_norm": 0.13403719663619995,
      "learning_rate": 9.983832008134157e-06,
      "loss": 0.0058,
      "step": 38370
    },
    {
      "epoch": 0.24615739835104047,
      "grad_norm": 0.2099473774433136,
      "learning_rate": 9.98378700258992e-06,
      "loss": 0.0042,
      "step": 38380
    },
    {
      "epoch": 0.24622153524482657,
      "grad_norm": 0.26247361302375793,
      "learning_rate": 9.98374193459527e-06,
      "loss": 0.0084,
      "step": 38390
    },
    {
      "epoch": 0.24628567213861266,
      "grad_norm": 0.36011576652526855,
      "learning_rate": 9.983696804150772e-06,
      "loss": 0.0056,
      "step": 38400
    },
    {
      "epoch": 0.24634980903239875,
      "grad_norm": 0.2732531726360321,
      "learning_rate": 9.98365161125699e-06,
      "loss": 0.0051,
      "step": 38410
    },
    {
      "epoch": 0.24641394592618485,
      "grad_norm": 0.0891374871134758,
      "learning_rate": 9.983606355914493e-06,
      "loss": 0.0055,
      "step": 38420
    },
    {
      "epoch": 0.24647808281997094,
      "grad_norm": 0.092893585562706,
      "learning_rate": 9.983561038123847e-06,
      "loss": 0.0033,
      "step": 38430
    },
    {
      "epoch": 0.24654221971375703,
      "grad_norm": 0.21625855565071106,
      "learning_rate": 9.98351565788562e-06,
      "loss": 0.0072,
      "step": 38440
    },
    {
      "epoch": 0.24660635660754315,
      "grad_norm": 0.12428835779428482,
      "learning_rate": 9.983470215200377e-06,
      "loss": 0.0063,
      "step": 38450
    },
    {
      "epoch": 0.24667049350132925,
      "grad_norm": 0.11295032501220703,
      "learning_rate": 9.983424710068693e-06,
      "loss": 0.0079,
      "step": 38460
    },
    {
      "epoch": 0.24673463039511534,
      "grad_norm": 0.20468109846115112,
      "learning_rate": 9.983379142491134e-06,
      "loss": 0.0051,
      "step": 38470
    },
    {
      "epoch": 0.24679876728890143,
      "grad_norm": 0.3960379958152771,
      "learning_rate": 9.983333512468274e-06,
      "loss": 0.0052,
      "step": 38480
    },
    {
      "epoch": 0.24686290418268753,
      "grad_norm": 0.20143310725688934,
      "learning_rate": 9.983287820000684e-06,
      "loss": 0.0062,
      "step": 38490
    },
    {
      "epoch": 0.24692704107647362,
      "grad_norm": 0.2517850399017334,
      "learning_rate": 9.983242065088937e-06,
      "loss": 0.0071,
      "step": 38500
    },
    {
      "epoch": 0.2469911779702597,
      "grad_norm": 0.710372269153595,
      "learning_rate": 9.983196247733602e-06,
      "loss": 0.005,
      "step": 38510
    },
    {
      "epoch": 0.24705531486404583,
      "grad_norm": 0.2682972848415375,
      "learning_rate": 9.983150367935259e-06,
      "loss": 0.0038,
      "step": 38520
    },
    {
      "epoch": 0.24711945175783193,
      "grad_norm": 0.058020081371068954,
      "learning_rate": 9.98310442569448e-06,
      "loss": 0.0049,
      "step": 38530
    },
    {
      "epoch": 0.24718358865161802,
      "grad_norm": 0.47107598185539246,
      "learning_rate": 9.983058421011842e-06,
      "loss": 0.0052,
      "step": 38540
    },
    {
      "epoch": 0.2472477255454041,
      "grad_norm": 0.2273666262626648,
      "learning_rate": 9.98301235388792e-06,
      "loss": 0.0052,
      "step": 38550
    },
    {
      "epoch": 0.2473118624391902,
      "grad_norm": 0.16423803567886353,
      "learning_rate": 9.982966224323292e-06,
      "loss": 0.0068,
      "step": 38560
    },
    {
      "epoch": 0.2473759993329763,
      "grad_norm": 0.33906319737434387,
      "learning_rate": 9.982920032318536e-06,
      "loss": 0.0057,
      "step": 38570
    },
    {
      "epoch": 0.2474401362267624,
      "grad_norm": 0.3201685845851898,
      "learning_rate": 9.982873777874228e-06,
      "loss": 0.0056,
      "step": 38580
    },
    {
      "epoch": 0.2475042731205485,
      "grad_norm": 0.18730387091636658,
      "learning_rate": 9.982827460990955e-06,
      "loss": 0.0034,
      "step": 38590
    },
    {
      "epoch": 0.2475684100143346,
      "grad_norm": 0.4003932774066925,
      "learning_rate": 9.98278108166929e-06,
      "loss": 0.0086,
      "step": 38600
    },
    {
      "epoch": 0.2476325469081207,
      "grad_norm": 0.11151555180549622,
      "learning_rate": 9.982734639909819e-06,
      "loss": 0.0044,
      "step": 38610
    },
    {
      "epoch": 0.2476966838019068,
      "grad_norm": 0.438439279794693,
      "learning_rate": 9.982688135713119e-06,
      "loss": 0.0048,
      "step": 38620
    },
    {
      "epoch": 0.24776082069569288,
      "grad_norm": 0.3096461594104767,
      "learning_rate": 9.982641569079777e-06,
      "loss": 0.0088,
      "step": 38630
    },
    {
      "epoch": 0.24782495758947898,
      "grad_norm": 0.23310524225234985,
      "learning_rate": 9.982594940010375e-06,
      "loss": 0.0075,
      "step": 38640
    },
    {
      "epoch": 0.24788909448326507,
      "grad_norm": 0.1498975306749344,
      "learning_rate": 9.982548248505497e-06,
      "loss": 0.0067,
      "step": 38650
    },
    {
      "epoch": 0.24795323137705116,
      "grad_norm": 0.2816608250141144,
      "learning_rate": 9.98250149456573e-06,
      "loss": 0.0062,
      "step": 38660
    },
    {
      "epoch": 0.24801736827083729,
      "grad_norm": 0.05591500177979469,
      "learning_rate": 9.982454678191658e-06,
      "loss": 0.0045,
      "step": 38670
    },
    {
      "epoch": 0.24808150516462338,
      "grad_norm": 0.5578827857971191,
      "learning_rate": 9.982407799383866e-06,
      "loss": 0.005,
      "step": 38680
    },
    {
      "epoch": 0.24814564205840947,
      "grad_norm": 0.3391058146953583,
      "learning_rate": 9.982360858142945e-06,
      "loss": 0.0055,
      "step": 38690
    },
    {
      "epoch": 0.24820977895219556,
      "grad_norm": 0.19329701364040375,
      "learning_rate": 9.982313854469481e-06,
      "loss": 0.0053,
      "step": 38700
    },
    {
      "epoch": 0.24827391584598166,
      "grad_norm": 0.10662292689085007,
      "learning_rate": 9.982266788364064e-06,
      "loss": 0.0054,
      "step": 38710
    },
    {
      "epoch": 0.24833805273976775,
      "grad_norm": 0.18978509306907654,
      "learning_rate": 9.982219659827283e-06,
      "loss": 0.0073,
      "step": 38720
    },
    {
      "epoch": 0.24840218963355384,
      "grad_norm": 0.35354286432266235,
      "learning_rate": 9.98217246885973e-06,
      "loss": 0.0067,
      "step": 38730
    },
    {
      "epoch": 0.24846632652733996,
      "grad_norm": 0.16298361122608185,
      "learning_rate": 9.982125215461992e-06,
      "loss": 0.0049,
      "step": 38740
    },
    {
      "epoch": 0.24853046342112606,
      "grad_norm": 0.15791991353034973,
      "learning_rate": 9.982077899634668e-06,
      "loss": 0.0078,
      "step": 38750
    },
    {
      "epoch": 0.24859460031491215,
      "grad_norm": 0.21328802406787872,
      "learning_rate": 9.982030521378346e-06,
      "loss": 0.0054,
      "step": 38760
    },
    {
      "epoch": 0.24865873720869824,
      "grad_norm": 0.12775328755378723,
      "learning_rate": 9.98198308069362e-06,
      "loss": 0.0084,
      "step": 38770
    },
    {
      "epoch": 0.24872287410248434,
      "grad_norm": 0.2946627140045166,
      "learning_rate": 9.981935577581088e-06,
      "loss": 0.0097,
      "step": 38780
    },
    {
      "epoch": 0.24878701099627043,
      "grad_norm": 0.20051661133766174,
      "learning_rate": 9.981888012041341e-06,
      "loss": 0.0041,
      "step": 38790
    },
    {
      "epoch": 0.24885114789005652,
      "grad_norm": 0.2328706979751587,
      "learning_rate": 9.981840384074977e-06,
      "loss": 0.0061,
      "step": 38800
    },
    {
      "epoch": 0.24891528478384264,
      "grad_norm": 0.18127407133579254,
      "learning_rate": 9.981792693682592e-06,
      "loss": 0.0116,
      "step": 38810
    },
    {
      "epoch": 0.24897942167762874,
      "grad_norm": 0.31884151697158813,
      "learning_rate": 9.981744940864786e-06,
      "loss": 0.0053,
      "step": 38820
    },
    {
      "epoch": 0.24904355857141483,
      "grad_norm": 0.31905224919319153,
      "learning_rate": 9.981697125622153e-06,
      "loss": 0.0084,
      "step": 38830
    },
    {
      "epoch": 0.24910769546520092,
      "grad_norm": 0.30726176500320435,
      "learning_rate": 9.981649247955296e-06,
      "loss": 0.0073,
      "step": 38840
    },
    {
      "epoch": 0.24917183235898702,
      "grad_norm": 0.3163350224494934,
      "learning_rate": 9.981601307864811e-06,
      "loss": 0.0054,
      "step": 38850
    },
    {
      "epoch": 0.2492359692527731,
      "grad_norm": 0.10571905225515366,
      "learning_rate": 9.981553305351306e-06,
      "loss": 0.004,
      "step": 38860
    },
    {
      "epoch": 0.2493001061465592,
      "grad_norm": 0.20794761180877686,
      "learning_rate": 9.981505240415375e-06,
      "loss": 0.0066,
      "step": 38870
    },
    {
      "epoch": 0.24936424304034532,
      "grad_norm": 0.29438677430152893,
      "learning_rate": 9.981457113057623e-06,
      "loss": 0.0061,
      "step": 38880
    },
    {
      "epoch": 0.24942837993413142,
      "grad_norm": 0.16410395503044128,
      "learning_rate": 9.981408923278654e-06,
      "loss": 0.0034,
      "step": 38890
    },
    {
      "epoch": 0.2494925168279175,
      "grad_norm": 0.12223443388938904,
      "learning_rate": 9.98136067107907e-06,
      "loss": 0.0075,
      "step": 38900
    },
    {
      "epoch": 0.2495566537217036,
      "grad_norm": 0.16954652965068817,
      "learning_rate": 9.981312356459479e-06,
      "loss": 0.0058,
      "step": 38910
    },
    {
      "epoch": 0.2496207906154897,
      "grad_norm": 0.40928614139556885,
      "learning_rate": 9.981263979420481e-06,
      "loss": 0.0056,
      "step": 38920
    },
    {
      "epoch": 0.2496849275092758,
      "grad_norm": 0.1617717742919922,
      "learning_rate": 9.981215539962687e-06,
      "loss": 0.0039,
      "step": 38930
    },
    {
      "epoch": 0.24974906440306188,
      "grad_norm": 0.18253344297409058,
      "learning_rate": 9.981167038086701e-06,
      "loss": 0.0048,
      "step": 38940
    },
    {
      "epoch": 0.249813201296848,
      "grad_norm": 0.3788161277770996,
      "learning_rate": 9.981118473793135e-06,
      "loss": 0.0061,
      "step": 38950
    },
    {
      "epoch": 0.2498773381906341,
      "grad_norm": 0.5014322996139526,
      "learning_rate": 9.981069847082592e-06,
      "loss": 0.0074,
      "step": 38960
    },
    {
      "epoch": 0.2499414750844202,
      "grad_norm": 0.11842537671327591,
      "learning_rate": 9.981021157955684e-06,
      "loss": 0.0048,
      "step": 38970
    },
    {
      "epoch": 0.2500056119782063,
      "grad_norm": 0.3788881301879883,
      "learning_rate": 9.980972406413022e-06,
      "loss": 0.0069,
      "step": 38980
    },
    {
      "epoch": 0.2500697488719924,
      "grad_norm": 0.21051029860973358,
      "learning_rate": 9.980923592455214e-06,
      "loss": 0.0045,
      "step": 38990
    },
    {
      "epoch": 0.2501338857657785,
      "grad_norm": 0.1375262290239334,
      "learning_rate": 9.980874716082875e-06,
      "loss": 0.0039,
      "step": 39000
    },
    {
      "epoch": 0.2501980226595646,
      "grad_norm": 0.31067126989364624,
      "learning_rate": 9.980825777296617e-06,
      "loss": 0.0079,
      "step": 39010
    },
    {
      "epoch": 0.2502621595533507,
      "grad_norm": 0.5094921588897705,
      "learning_rate": 9.98077677609705e-06,
      "loss": 0.0071,
      "step": 39020
    },
    {
      "epoch": 0.2503262964471368,
      "grad_norm": 0.13262316584587097,
      "learning_rate": 9.980727712484792e-06,
      "loss": 0.0066,
      "step": 39030
    },
    {
      "epoch": 0.25039043334092287,
      "grad_norm": 0.32757246494293213,
      "learning_rate": 9.980678586460455e-06,
      "loss": 0.0082,
      "step": 39040
    },
    {
      "epoch": 0.25045457023470896,
      "grad_norm": 0.29297035932540894,
      "learning_rate": 9.980629398024658e-06,
      "loss": 0.0079,
      "step": 39050
    },
    {
      "epoch": 0.25051870712849506,
      "grad_norm": 0.18163636326789856,
      "learning_rate": 9.980580147178011e-06,
      "loss": 0.0047,
      "step": 39060
    },
    {
      "epoch": 0.25058284402228115,
      "grad_norm": 0.21941807866096497,
      "learning_rate": 9.980530833921139e-06,
      "loss": 0.0058,
      "step": 39070
    },
    {
      "epoch": 0.25064698091606724,
      "grad_norm": 0.37371599674224854,
      "learning_rate": 9.980481458254654e-06,
      "loss": 0.0085,
      "step": 39080
    },
    {
      "epoch": 0.25071111780985333,
      "grad_norm": 0.027556492015719414,
      "learning_rate": 9.980432020179179e-06,
      "loss": 0.0066,
      "step": 39090
    },
    {
      "epoch": 0.2507752547036394,
      "grad_norm": 0.20441341400146484,
      "learning_rate": 9.98038251969533e-06,
      "loss": 0.0079,
      "step": 39100
    },
    {
      "epoch": 0.2508393915974255,
      "grad_norm": 0.25886884331703186,
      "learning_rate": 9.980332956803727e-06,
      "loss": 0.0073,
      "step": 39110
    },
    {
      "epoch": 0.25090352849121167,
      "grad_norm": 0.3851901888847351,
      "learning_rate": 9.980283331504994e-06,
      "loss": 0.0043,
      "step": 39120
    },
    {
      "epoch": 0.25096766538499776,
      "grad_norm": 0.15157650411128998,
      "learning_rate": 9.980233643799751e-06,
      "loss": 0.0058,
      "step": 39130
    },
    {
      "epoch": 0.25103180227878386,
      "grad_norm": 0.3190149962902069,
      "learning_rate": 9.980183893688621e-06,
      "loss": 0.0069,
      "step": 39140
    },
    {
      "epoch": 0.25109593917256995,
      "grad_norm": 0.39699846506118774,
      "learning_rate": 9.98013408117223e-06,
      "loss": 0.0054,
      "step": 39150
    },
    {
      "epoch": 0.25116007606635604,
      "grad_norm": 0.09573323279619217,
      "learning_rate": 9.980084206251197e-06,
      "loss": 0.0044,
      "step": 39160
    },
    {
      "epoch": 0.25122421296014213,
      "grad_norm": 0.29710081219673157,
      "learning_rate": 9.980034268926151e-06,
      "loss": 0.0055,
      "step": 39170
    },
    {
      "epoch": 0.25128834985392823,
      "grad_norm": 0.07126680016517639,
      "learning_rate": 9.979984269197715e-06,
      "loss": 0.0032,
      "step": 39180
    },
    {
      "epoch": 0.2513524867477143,
      "grad_norm": 0.11775899678468704,
      "learning_rate": 9.97993420706652e-06,
      "loss": 0.0047,
      "step": 39190
    },
    {
      "epoch": 0.2514166236415004,
      "grad_norm": 0.4710961580276489,
      "learning_rate": 9.979884082533188e-06,
      "loss": 0.0034,
      "step": 39200
    },
    {
      "epoch": 0.2514807605352865,
      "grad_norm": 0.22415786981582642,
      "learning_rate": 9.97983389559835e-06,
      "loss": 0.0051,
      "step": 39210
    },
    {
      "epoch": 0.2515448974290726,
      "grad_norm": 0.30657926201820374,
      "learning_rate": 9.979783646262633e-06,
      "loss": 0.0069,
      "step": 39220
    },
    {
      "epoch": 0.2516090343228587,
      "grad_norm": 0.14880214631557465,
      "learning_rate": 9.97973333452667e-06,
      "loss": 0.0059,
      "step": 39230
    },
    {
      "epoch": 0.2516731712166448,
      "grad_norm": 0.194294273853302,
      "learning_rate": 9.979682960391086e-06,
      "loss": 0.0059,
      "step": 39240
    },
    {
      "epoch": 0.2517373081104309,
      "grad_norm": 0.2686821520328522,
      "learning_rate": 9.97963252385652e-06,
      "loss": 0.0055,
      "step": 39250
    },
    {
      "epoch": 0.25180144500421703,
      "grad_norm": 0.5156296491622925,
      "learning_rate": 9.979582024923596e-06,
      "loss": 0.0072,
      "step": 39260
    },
    {
      "epoch": 0.2518655818980031,
      "grad_norm": 0.3677895963191986,
      "learning_rate": 9.979531463592951e-06,
      "loss": 0.0036,
      "step": 39270
    },
    {
      "epoch": 0.2519297187917892,
      "grad_norm": 0.1801011562347412,
      "learning_rate": 9.979480839865219e-06,
      "loss": 0.0054,
      "step": 39280
    },
    {
      "epoch": 0.2519938556855753,
      "grad_norm": 0.26837414503097534,
      "learning_rate": 9.979430153741034e-06,
      "loss": 0.0035,
      "step": 39290
    },
    {
      "epoch": 0.2520579925793614,
      "grad_norm": 0.21830004453659058,
      "learning_rate": 9.979379405221028e-06,
      "loss": 0.0047,
      "step": 39300
    },
    {
      "epoch": 0.2521221294731475,
      "grad_norm": 0.5536841750144958,
      "learning_rate": 9.979328594305842e-06,
      "loss": 0.004,
      "step": 39310
    },
    {
      "epoch": 0.2521862663669336,
      "grad_norm": 0.29669874906539917,
      "learning_rate": 9.979277720996108e-06,
      "loss": 0.0075,
      "step": 39320
    },
    {
      "epoch": 0.2522504032607197,
      "grad_norm": 0.163956880569458,
      "learning_rate": 9.979226785292465e-06,
      "loss": 0.0052,
      "step": 39330
    },
    {
      "epoch": 0.2523145401545058,
      "grad_norm": 0.3207421898841858,
      "learning_rate": 9.979175787195556e-06,
      "loss": 0.0067,
      "step": 39340
    },
    {
      "epoch": 0.25237867704829187,
      "grad_norm": 0.3795740604400635,
      "learning_rate": 9.979124726706011e-06,
      "loss": 0.0052,
      "step": 39350
    },
    {
      "epoch": 0.25244281394207796,
      "grad_norm": 0.22112007439136505,
      "learning_rate": 9.979073603824477e-06,
      "loss": 0.0054,
      "step": 39360
    },
    {
      "epoch": 0.25250695083586405,
      "grad_norm": 0.2207915037870407,
      "learning_rate": 9.979022418551592e-06,
      "loss": 0.0075,
      "step": 39370
    },
    {
      "epoch": 0.25257108772965015,
      "grad_norm": 0.2956860363483429,
      "learning_rate": 9.978971170887997e-06,
      "loss": 0.0047,
      "step": 39380
    },
    {
      "epoch": 0.25263522462343624,
      "grad_norm": 0.24269051849842072,
      "learning_rate": 9.978919860834337e-06,
      "loss": 0.007,
      "step": 39390
    },
    {
      "epoch": 0.25269936151722233,
      "grad_norm": 0.2062823325395584,
      "learning_rate": 9.97886848839125e-06,
      "loss": 0.0044,
      "step": 39400
    },
    {
      "epoch": 0.2527634984110085,
      "grad_norm": 0.299936443567276,
      "learning_rate": 9.978817053559384e-06,
      "loss": 0.0057,
      "step": 39410
    },
    {
      "epoch": 0.2528276353047946,
      "grad_norm": 0.17424818873405457,
      "learning_rate": 9.97876555633938e-06,
      "loss": 0.006,
      "step": 39420
    },
    {
      "epoch": 0.25289177219858067,
      "grad_norm": 0.526239812374115,
      "learning_rate": 9.978713996731888e-06,
      "loss": 0.0072,
      "step": 39430
    },
    {
      "epoch": 0.25295590909236676,
      "grad_norm": 0.2941557466983795,
      "learning_rate": 9.978662374737551e-06,
      "loss": 0.006,
      "step": 39440
    },
    {
      "epoch": 0.25302004598615285,
      "grad_norm": 0.57313472032547,
      "learning_rate": 9.978610690357014e-06,
      "loss": 0.0055,
      "step": 39450
    },
    {
      "epoch": 0.25308418287993895,
      "grad_norm": 0.36666339635849,
      "learning_rate": 9.978558943590929e-06,
      "loss": 0.0067,
      "step": 39460
    },
    {
      "epoch": 0.25314831977372504,
      "grad_norm": 0.24211323261260986,
      "learning_rate": 9.978507134439941e-06,
      "loss": 0.0037,
      "step": 39470
    },
    {
      "epoch": 0.25321245666751113,
      "grad_norm": 0.36444950103759766,
      "learning_rate": 9.978455262904703e-06,
      "loss": 0.0066,
      "step": 39480
    },
    {
      "epoch": 0.2532765935612972,
      "grad_norm": 0.293956458568573,
      "learning_rate": 9.97840332898586e-06,
      "loss": 0.0088,
      "step": 39490
    },
    {
      "epoch": 0.2533407304550833,
      "grad_norm": 0.27882516384124756,
      "learning_rate": 9.978351332684066e-06,
      "loss": 0.0066,
      "step": 39500
    },
    {
      "epoch": 0.2534048673488694,
      "grad_norm": 0.1138320341706276,
      "learning_rate": 9.978299273999972e-06,
      "loss": 0.0071,
      "step": 39510
    },
    {
      "epoch": 0.2534690042426555,
      "grad_norm": 0.2853226065635681,
      "learning_rate": 9.97824715293423e-06,
      "loss": 0.0056,
      "step": 39520
    },
    {
      "epoch": 0.2535331411364416,
      "grad_norm": 0.05086997523903847,
      "learning_rate": 9.978194969487493e-06,
      "loss": 0.0057,
      "step": 39530
    },
    {
      "epoch": 0.2535972780302277,
      "grad_norm": 0.17800335586071014,
      "learning_rate": 9.978142723660415e-06,
      "loss": 0.0045,
      "step": 39540
    },
    {
      "epoch": 0.25366141492401384,
      "grad_norm": 0.337404727935791,
      "learning_rate": 9.978090415453651e-06,
      "loss": 0.0056,
      "step": 39550
    },
    {
      "epoch": 0.25372555181779993,
      "grad_norm": 0.2746542692184448,
      "learning_rate": 9.978038044867858e-06,
      "loss": 0.0059,
      "step": 39560
    },
    {
      "epoch": 0.253789688711586,
      "grad_norm": 0.8087136745452881,
      "learning_rate": 9.977985611903688e-06,
      "loss": 0.0056,
      "step": 39570
    },
    {
      "epoch": 0.2538538256053721,
      "grad_norm": 0.07329525798559189,
      "learning_rate": 9.977933116561801e-06,
      "loss": 0.0051,
      "step": 39580
    },
    {
      "epoch": 0.2539179624991582,
      "grad_norm": 0.16835616528987885,
      "learning_rate": 9.977880558842857e-06,
      "loss": 0.005,
      "step": 39590
    },
    {
      "epoch": 0.2539820993929443,
      "grad_norm": 0.22268901765346527,
      "learning_rate": 9.977827938747508e-06,
      "loss": 0.0055,
      "step": 39600
    },
    {
      "epoch": 0.2540462362867304,
      "grad_norm": 0.18357685208320618,
      "learning_rate": 9.977775256276421e-06,
      "loss": 0.0067,
      "step": 39610
    },
    {
      "epoch": 0.2541103731805165,
      "grad_norm": 0.24646100401878357,
      "learning_rate": 9.97772251143025e-06,
      "loss": 0.0064,
      "step": 39620
    },
    {
      "epoch": 0.2541745100743026,
      "grad_norm": 0.249356210231781,
      "learning_rate": 9.97766970420966e-06,
      "loss": 0.0047,
      "step": 39630
    },
    {
      "epoch": 0.2542386469680887,
      "grad_norm": 0.25969377160072327,
      "learning_rate": 9.977616834615308e-06,
      "loss": 0.0042,
      "step": 39640
    },
    {
      "epoch": 0.25430278386187477,
      "grad_norm": 0.15693524479866028,
      "learning_rate": 9.977563902647865e-06,
      "loss": 0.0052,
      "step": 39650
    },
    {
      "epoch": 0.25436692075566086,
      "grad_norm": 0.3180239200592041,
      "learning_rate": 9.977510908307985e-06,
      "loss": 0.0072,
      "step": 39660
    },
    {
      "epoch": 0.25443105764944696,
      "grad_norm": 0.3399795591831207,
      "learning_rate": 9.977457851596337e-06,
      "loss": 0.004,
      "step": 39670
    },
    {
      "epoch": 0.25449519454323305,
      "grad_norm": 0.07352469861507416,
      "learning_rate": 9.977404732513584e-06,
      "loss": 0.0061,
      "step": 39680
    },
    {
      "epoch": 0.2545593314370192,
      "grad_norm": 0.3775775134563446,
      "learning_rate": 9.977351551060392e-06,
      "loss": 0.0076,
      "step": 39690
    },
    {
      "epoch": 0.2546234683308053,
      "grad_norm": 0.15693242847919464,
      "learning_rate": 9.97729830723743e-06,
      "loss": 0.0086,
      "step": 39700
    },
    {
      "epoch": 0.2546876052245914,
      "grad_norm": 0.29596182703971863,
      "learning_rate": 9.977245001045362e-06,
      "loss": 0.0053,
      "step": 39710
    },
    {
      "epoch": 0.2547517421183775,
      "grad_norm": 0.12882474064826965,
      "learning_rate": 9.977191632484857e-06,
      "loss": 0.0067,
      "step": 39720
    },
    {
      "epoch": 0.25481587901216357,
      "grad_norm": 0.3904571831226349,
      "learning_rate": 9.977138201556583e-06,
      "loss": 0.0048,
      "step": 39730
    },
    {
      "epoch": 0.25488001590594966,
      "grad_norm": 0.2068895399570465,
      "learning_rate": 9.977084708261211e-06,
      "loss": 0.0072,
      "step": 39740
    },
    {
      "epoch": 0.25494415279973576,
      "grad_norm": 0.10844743996858597,
      "learning_rate": 9.97703115259941e-06,
      "loss": 0.0064,
      "step": 39750
    },
    {
      "epoch": 0.25500828969352185,
      "grad_norm": 0.19294798374176025,
      "learning_rate": 9.976977534571853e-06,
      "loss": 0.005,
      "step": 39760
    },
    {
      "epoch": 0.25507242658730794,
      "grad_norm": 0.2761460840702057,
      "learning_rate": 9.976923854179207e-06,
      "loss": 0.0062,
      "step": 39770
    },
    {
      "epoch": 0.25513656348109404,
      "grad_norm": 0.1745172142982483,
      "learning_rate": 9.976870111422152e-06,
      "loss": 0.0052,
      "step": 39780
    },
    {
      "epoch": 0.25520070037488013,
      "grad_norm": 0.11423773318529129,
      "learning_rate": 9.976816306301355e-06,
      "loss": 0.0075,
      "step": 39790
    },
    {
      "epoch": 0.2552648372686662,
      "grad_norm": 0.12186460942029953,
      "learning_rate": 9.976762438817494e-06,
      "loss": 0.0053,
      "step": 39800
    },
    {
      "epoch": 0.2553289741624523,
      "grad_norm": 0.15649254620075226,
      "learning_rate": 9.976708508971242e-06,
      "loss": 0.0047,
      "step": 39810
    },
    {
      "epoch": 0.2553931110562384,
      "grad_norm": 0.13888505101203918,
      "learning_rate": 9.976654516763278e-06,
      "loss": 0.0075,
      "step": 39820
    },
    {
      "epoch": 0.25545724795002456,
      "grad_norm": 0.18757475912570953,
      "learning_rate": 9.976600462194273e-06,
      "loss": 0.0057,
      "step": 39830
    },
    {
      "epoch": 0.25552138484381065,
      "grad_norm": 0.34580785036087036,
      "learning_rate": 9.97654634526491e-06,
      "loss": 0.0058,
      "step": 39840
    },
    {
      "epoch": 0.25558552173759674,
      "grad_norm": 0.5235243439674377,
      "learning_rate": 9.976492165975864e-06,
      "loss": 0.007,
      "step": 39850
    },
    {
      "epoch": 0.25564965863138284,
      "grad_norm": 0.3960067927837372,
      "learning_rate": 9.976437924327813e-06,
      "loss": 0.0049,
      "step": 39860
    },
    {
      "epoch": 0.25571379552516893,
      "grad_norm": 0.33573824167251587,
      "learning_rate": 9.97638362032144e-06,
      "loss": 0.0059,
      "step": 39870
    },
    {
      "epoch": 0.255777932418955,
      "grad_norm": 0.22803282737731934,
      "learning_rate": 9.976329253957422e-06,
      "loss": 0.0057,
      "step": 39880
    },
    {
      "epoch": 0.2558420693127411,
      "grad_norm": 0.03270663693547249,
      "learning_rate": 9.976274825236443e-06,
      "loss": 0.0032,
      "step": 39890
    },
    {
      "epoch": 0.2559062062065272,
      "grad_norm": 0.25057414174079895,
      "learning_rate": 9.976220334159183e-06,
      "loss": 0.0058,
      "step": 39900
    },
    {
      "epoch": 0.2559703431003133,
      "grad_norm": 0.2685162425041199,
      "learning_rate": 9.976165780726328e-06,
      "loss": 0.0039,
      "step": 39910
    },
    {
      "epoch": 0.2560344799940994,
      "grad_norm": 0.3304634988307953,
      "learning_rate": 9.976111164938558e-06,
      "loss": 0.0044,
      "step": 39920
    },
    {
      "epoch": 0.2560986168878855,
      "grad_norm": 0.2324581891298294,
      "learning_rate": 9.976056486796557e-06,
      "loss": 0.0084,
      "step": 39930
    },
    {
      "epoch": 0.2561627537816716,
      "grad_norm": 0.4089649021625519,
      "learning_rate": 9.976001746301014e-06,
      "loss": 0.0047,
      "step": 39940
    },
    {
      "epoch": 0.2562268906754577,
      "grad_norm": 0.23112742602825165,
      "learning_rate": 9.975946943452613e-06,
      "loss": 0.0056,
      "step": 39950
    },
    {
      "epoch": 0.25629102756924377,
      "grad_norm": 0.3333764374256134,
      "learning_rate": 9.97589207825204e-06,
      "loss": 0.0081,
      "step": 39960
    },
    {
      "epoch": 0.2563551644630299,
      "grad_norm": 0.30956459045410156,
      "learning_rate": 9.975837150699984e-06,
      "loss": 0.0052,
      "step": 39970
    },
    {
      "epoch": 0.256419301356816,
      "grad_norm": 0.41459202766418457,
      "learning_rate": 9.975782160797129e-06,
      "loss": 0.0056,
      "step": 39980
    },
    {
      "epoch": 0.2564834382506021,
      "grad_norm": 0.1197529062628746,
      "learning_rate": 9.975727108544171e-06,
      "loss": 0.0038,
      "step": 39990
    },
    {
      "epoch": 0.2565475751443882,
      "grad_norm": 0.09705403447151184,
      "learning_rate": 9.975671993941795e-06,
      "loss": 0.0044,
      "step": 40000
    },
    {
      "epoch": 0.2566117120381743,
      "grad_norm": 0.448026180267334,
      "learning_rate": 9.975616816990691e-06,
      "loss": 0.004,
      "step": 40010
    },
    {
      "epoch": 0.2566758489319604,
      "grad_norm": 0.1669142097234726,
      "learning_rate": 9.975561577691555e-06,
      "loss": 0.0087,
      "step": 40020
    },
    {
      "epoch": 0.2567399858257465,
      "grad_norm": 0.33060795068740845,
      "learning_rate": 9.975506276045076e-06,
      "loss": 0.0078,
      "step": 40030
    },
    {
      "epoch": 0.25680412271953257,
      "grad_norm": 0.10030943900346756,
      "learning_rate": 9.975450912051946e-06,
      "loss": 0.0041,
      "step": 40040
    },
    {
      "epoch": 0.25686825961331866,
      "grad_norm": 0.35985592007637024,
      "learning_rate": 9.97539548571286e-06,
      "loss": 0.0069,
      "step": 40050
    },
    {
      "epoch": 0.25693239650710475,
      "grad_norm": 0.148360937833786,
      "learning_rate": 9.975339997028516e-06,
      "loss": 0.0061,
      "step": 40060
    },
    {
      "epoch": 0.25699653340089085,
      "grad_norm": 0.2986155152320862,
      "learning_rate": 9.975284445999602e-06,
      "loss": 0.0056,
      "step": 40070
    },
    {
      "epoch": 0.25706067029467694,
      "grad_norm": 0.20368216931819916,
      "learning_rate": 9.97522883262682e-06,
      "loss": 0.0069,
      "step": 40080
    },
    {
      "epoch": 0.25712480718846303,
      "grad_norm": 0.4418974220752716,
      "learning_rate": 9.975173156910865e-06,
      "loss": 0.0067,
      "step": 40090
    },
    {
      "epoch": 0.2571889440822491,
      "grad_norm": 0.45510780811309814,
      "learning_rate": 9.975117418852434e-06,
      "loss": 0.0048,
      "step": 40100
    },
    {
      "epoch": 0.2572530809760353,
      "grad_norm": 0.3036324977874756,
      "learning_rate": 9.975061618452228e-06,
      "loss": 0.0051,
      "step": 40110
    },
    {
      "epoch": 0.25731721786982137,
      "grad_norm": 0.28871700167655945,
      "learning_rate": 9.975005755710943e-06,
      "loss": 0.004,
      "step": 40120
    },
    {
      "epoch": 0.25738135476360746,
      "grad_norm": 0.11629597097635269,
      "learning_rate": 9.974949830629279e-06,
      "loss": 0.0067,
      "step": 40130
    },
    {
      "epoch": 0.25744549165739355,
      "grad_norm": 0.19242048263549805,
      "learning_rate": 9.97489384320794e-06,
      "loss": 0.0068,
      "step": 40140
    },
    {
      "epoch": 0.25750962855117965,
      "grad_norm": 0.07171380519866943,
      "learning_rate": 9.974837793447625e-06,
      "loss": 0.0032,
      "step": 40150
    },
    {
      "epoch": 0.25757376544496574,
      "grad_norm": 0.3329668343067169,
      "learning_rate": 9.974781681349037e-06,
      "loss": 0.0065,
      "step": 40160
    },
    {
      "epoch": 0.25763790233875183,
      "grad_norm": 0.3116569221019745,
      "learning_rate": 9.974725506912879e-06,
      "loss": 0.0085,
      "step": 40170
    },
    {
      "epoch": 0.2577020392325379,
      "grad_norm": 0.21919970214366913,
      "learning_rate": 9.974669270139854e-06,
      "loss": 0.0084,
      "step": 40180
    },
    {
      "epoch": 0.257766176126324,
      "grad_norm": 0.1700431853532791,
      "learning_rate": 9.974612971030668e-06,
      "loss": 0.0051,
      "step": 40190
    },
    {
      "epoch": 0.2578303130201101,
      "grad_norm": 0.09541097283363342,
      "learning_rate": 9.974556609586028e-06,
      "loss": 0.0037,
      "step": 40200
    },
    {
      "epoch": 0.2578944499138962,
      "grad_norm": 0.12062127888202667,
      "learning_rate": 9.974500185806637e-06,
      "loss": 0.0035,
      "step": 40210
    },
    {
      "epoch": 0.2579585868076823,
      "grad_norm": 0.3263625502586365,
      "learning_rate": 9.974443699693204e-06,
      "loss": 0.0068,
      "step": 40220
    },
    {
      "epoch": 0.2580227237014684,
      "grad_norm": 0.3020590841770172,
      "learning_rate": 9.974387151246437e-06,
      "loss": 0.0109,
      "step": 40230
    },
    {
      "epoch": 0.2580868605952545,
      "grad_norm": 0.46705231070518494,
      "learning_rate": 9.97433054046704e-06,
      "loss": 0.0103,
      "step": 40240
    },
    {
      "epoch": 0.25815099748904063,
      "grad_norm": 0.1899949014186859,
      "learning_rate": 9.97427386735573e-06,
      "loss": 0.0058,
      "step": 40250
    },
    {
      "epoch": 0.2582151343828267,
      "grad_norm": 0.18206621706485748,
      "learning_rate": 9.974217131913213e-06,
      "loss": 0.0028,
      "step": 40260
    },
    {
      "epoch": 0.2582792712766128,
      "grad_norm": 0.3200024664402008,
      "learning_rate": 9.974160334140199e-06,
      "loss": 0.0083,
      "step": 40270
    },
    {
      "epoch": 0.2583434081703989,
      "grad_norm": 0.4450424611568451,
      "learning_rate": 9.974103474037403e-06,
      "loss": 0.0075,
      "step": 40280
    },
    {
      "epoch": 0.258407545064185,
      "grad_norm": 0.13085024058818817,
      "learning_rate": 9.974046551605533e-06,
      "loss": 0.0051,
      "step": 40290
    },
    {
      "epoch": 0.2584716819579711,
      "grad_norm": 0.27274298667907715,
      "learning_rate": 9.973989566845307e-06,
      "loss": 0.0052,
      "step": 40300
    },
    {
      "epoch": 0.2585358188517572,
      "grad_norm": 0.36792632937431335,
      "learning_rate": 9.973932519757437e-06,
      "loss": 0.0064,
      "step": 40310
    },
    {
      "epoch": 0.2585999557455433,
      "grad_norm": 0.20151731371879578,
      "learning_rate": 9.973875410342636e-06,
      "loss": 0.0071,
      "step": 40320
    },
    {
      "epoch": 0.2586640926393294,
      "grad_norm": 0.2218293994665146,
      "learning_rate": 9.973818238601623e-06,
      "loss": 0.0041,
      "step": 40330
    },
    {
      "epoch": 0.2587282295331155,
      "grad_norm": 0.19607120752334595,
      "learning_rate": 9.97376100453511e-06,
      "loss": 0.0055,
      "step": 40340
    },
    {
      "epoch": 0.25879236642690157,
      "grad_norm": 0.44621092081069946,
      "learning_rate": 9.973703708143819e-06,
      "loss": 0.0038,
      "step": 40350
    },
    {
      "epoch": 0.25885650332068766,
      "grad_norm": 0.7802656292915344,
      "learning_rate": 9.973646349428465e-06,
      "loss": 0.0064,
      "step": 40360
    },
    {
      "epoch": 0.25892064021447375,
      "grad_norm": 0.13441035151481628,
      "learning_rate": 9.973588928389768e-06,
      "loss": 0.005,
      "step": 40370
    },
    {
      "epoch": 0.25898477710825984,
      "grad_norm": 0.19954079389572144,
      "learning_rate": 9.973531445028448e-06,
      "loss": 0.0034,
      "step": 40380
    },
    {
      "epoch": 0.259048914002046,
      "grad_norm": 0.17247618734836578,
      "learning_rate": 9.973473899345226e-06,
      "loss": 0.0067,
      "step": 40390
    },
    {
      "epoch": 0.2591130508958321,
      "grad_norm": 0.3464730381965637,
      "learning_rate": 9.973416291340818e-06,
      "loss": 0.0103,
      "step": 40400
    },
    {
      "epoch": 0.2591771877896182,
      "grad_norm": 0.14043326675891876,
      "learning_rate": 9.973358621015951e-06,
      "loss": 0.006,
      "step": 40410
    },
    {
      "epoch": 0.2592413246834043,
      "grad_norm": 0.20075811445713043,
      "learning_rate": 9.973300888371345e-06,
      "loss": 0.0069,
      "step": 40420
    },
    {
      "epoch": 0.25930546157719037,
      "grad_norm": 0.4112997055053711,
      "learning_rate": 9.973243093407726e-06,
      "loss": 0.0074,
      "step": 40430
    },
    {
      "epoch": 0.25936959847097646,
      "grad_norm": 0.5253247618675232,
      "learning_rate": 9.973185236125815e-06,
      "loss": 0.0074,
      "step": 40440
    },
    {
      "epoch": 0.25943373536476255,
      "grad_norm": 0.11537769436836243,
      "learning_rate": 9.973127316526341e-06,
      "loss": 0.0079,
      "step": 40450
    },
    {
      "epoch": 0.25949787225854865,
      "grad_norm": 0.3097275495529175,
      "learning_rate": 9.973069334610027e-06,
      "loss": 0.0094,
      "step": 40460
    },
    {
      "epoch": 0.25956200915233474,
      "grad_norm": 0.26016682386398315,
      "learning_rate": 9.973011290377598e-06,
      "loss": 0.0083,
      "step": 40470
    },
    {
      "epoch": 0.25962614604612083,
      "grad_norm": 0.34955739974975586,
      "learning_rate": 9.972953183829786e-06,
      "loss": 0.0053,
      "step": 40480
    },
    {
      "epoch": 0.2596902829399069,
      "grad_norm": 0.1267748326063156,
      "learning_rate": 9.972895014967315e-06,
      "loss": 0.0054,
      "step": 40490
    },
    {
      "epoch": 0.259754419833693,
      "grad_norm": 0.21058782935142517,
      "learning_rate": 9.972836783790915e-06,
      "loss": 0.0046,
      "step": 40500
    },
    {
      "epoch": 0.2598185567274791,
      "grad_norm": 0.17960603535175323,
      "learning_rate": 9.972778490301317e-06,
      "loss": 0.0065,
      "step": 40510
    },
    {
      "epoch": 0.2598826936212652,
      "grad_norm": 0.14524775743484497,
      "learning_rate": 9.97272013449925e-06,
      "loss": 0.0067,
      "step": 40520
    },
    {
      "epoch": 0.25994683051505135,
      "grad_norm": 0.3339650332927704,
      "learning_rate": 9.972661716385446e-06,
      "loss": 0.0033,
      "step": 40530
    },
    {
      "epoch": 0.26001096740883745,
      "grad_norm": 0.2901301383972168,
      "learning_rate": 9.972603235960636e-06,
      "loss": 0.0066,
      "step": 40540
    },
    {
      "epoch": 0.26007510430262354,
      "grad_norm": 0.25822460651397705,
      "learning_rate": 9.972544693225554e-06,
      "loss": 0.0052,
      "step": 40550
    },
    {
      "epoch": 0.26013924119640963,
      "grad_norm": 0.47759416699409485,
      "learning_rate": 9.972486088180936e-06,
      "loss": 0.0054,
      "step": 40560
    },
    {
      "epoch": 0.2602033780901957,
      "grad_norm": 0.05683809518814087,
      "learning_rate": 9.97242742082751e-06,
      "loss": 0.0047,
      "step": 40570
    },
    {
      "epoch": 0.2602675149839818,
      "grad_norm": 0.597055971622467,
      "learning_rate": 9.972368691166014e-06,
      "loss": 0.0157,
      "step": 40580
    },
    {
      "epoch": 0.2603316518777679,
      "grad_norm": 0.2670753002166748,
      "learning_rate": 9.972309899197187e-06,
      "loss": 0.0055,
      "step": 40590
    },
    {
      "epoch": 0.260395788771554,
      "grad_norm": 0.0818215012550354,
      "learning_rate": 9.972251044921764e-06,
      "loss": 0.0065,
      "step": 40600
    },
    {
      "epoch": 0.2604599256653401,
      "grad_norm": 0.229543998837471,
      "learning_rate": 9.97219212834048e-06,
      "loss": 0.0044,
      "step": 40610
    },
    {
      "epoch": 0.2605240625591262,
      "grad_norm": 0.21901388466358185,
      "learning_rate": 9.972133149454075e-06,
      "loss": 0.0068,
      "step": 40620
    },
    {
      "epoch": 0.2605881994529123,
      "grad_norm": 0.3773210048675537,
      "learning_rate": 9.972074108263289e-06,
      "loss": 0.0072,
      "step": 40630
    },
    {
      "epoch": 0.2606523363466984,
      "grad_norm": 0.2899719774723053,
      "learning_rate": 9.97201500476886e-06,
      "loss": 0.0058,
      "step": 40640
    },
    {
      "epoch": 0.26071647324048447,
      "grad_norm": 0.14465318620204926,
      "learning_rate": 9.97195583897153e-06,
      "loss": 0.0043,
      "step": 40650
    },
    {
      "epoch": 0.26078061013427056,
      "grad_norm": 0.2485213279724121,
      "learning_rate": 9.971896610872041e-06,
      "loss": 0.0047,
      "step": 40660
    },
    {
      "epoch": 0.2608447470280567,
      "grad_norm": 0.27826058864593506,
      "learning_rate": 9.971837320471132e-06,
      "loss": 0.005,
      "step": 40670
    },
    {
      "epoch": 0.2609088839218428,
      "grad_norm": 0.21804702281951904,
      "learning_rate": 9.971777967769549e-06,
      "loss": 0.0047,
      "step": 40680
    },
    {
      "epoch": 0.2609730208156289,
      "grad_norm": 0.22855201363563538,
      "learning_rate": 9.971718552768035e-06,
      "loss": 0.0057,
      "step": 40690
    },
    {
      "epoch": 0.261037157709415,
      "grad_norm": 0.298373818397522,
      "learning_rate": 9.971659075467335e-06,
      "loss": 0.0039,
      "step": 40700
    },
    {
      "epoch": 0.2611012946032011,
      "grad_norm": 0.15485908091068268,
      "learning_rate": 9.971599535868193e-06,
      "loss": 0.0059,
      "step": 40710
    },
    {
      "epoch": 0.2611654314969872,
      "grad_norm": 0.24103794991970062,
      "learning_rate": 9.971539933971355e-06,
      "loss": 0.005,
      "step": 40720
    },
    {
      "epoch": 0.26122956839077327,
      "grad_norm": 0.199654221534729,
      "learning_rate": 9.971480269777569e-06,
      "loss": 0.0051,
      "step": 40730
    },
    {
      "epoch": 0.26129370528455936,
      "grad_norm": 0.11665575951337814,
      "learning_rate": 9.971420543287582e-06,
      "loss": 0.0071,
      "step": 40740
    },
    {
      "epoch": 0.26135784217834546,
      "grad_norm": 0.18692216277122498,
      "learning_rate": 9.971360754502143e-06,
      "loss": 0.0045,
      "step": 40750
    },
    {
      "epoch": 0.26142197907213155,
      "grad_norm": 0.27884986996650696,
      "learning_rate": 9.971300903422001e-06,
      "loss": 0.0036,
      "step": 40760
    },
    {
      "epoch": 0.26148611596591764,
      "grad_norm": 0.19081412255764008,
      "learning_rate": 9.971240990047904e-06,
      "loss": 0.0075,
      "step": 40770
    },
    {
      "epoch": 0.26155025285970374,
      "grad_norm": 0.24674800038337708,
      "learning_rate": 9.971181014380607e-06,
      "loss": 0.006,
      "step": 40780
    },
    {
      "epoch": 0.26161438975348983,
      "grad_norm": 0.10528982430696487,
      "learning_rate": 9.971120976420857e-06,
      "loss": 0.0049,
      "step": 40790
    },
    {
      "epoch": 0.2616785266472759,
      "grad_norm": 0.2989997863769531,
      "learning_rate": 9.97106087616941e-06,
      "loss": 0.0077,
      "step": 40800
    },
    {
      "epoch": 0.26174266354106207,
      "grad_norm": 0.10974735766649246,
      "learning_rate": 9.971000713627016e-06,
      "loss": 0.0102,
      "step": 40810
    },
    {
      "epoch": 0.26180680043484816,
      "grad_norm": 0.2013663798570633,
      "learning_rate": 9.970940488794432e-06,
      "loss": 0.0062,
      "step": 40820
    },
    {
      "epoch": 0.26187093732863426,
      "grad_norm": 0.04378150776028633,
      "learning_rate": 9.970880201672409e-06,
      "loss": 0.0044,
      "step": 40830
    },
    {
      "epoch": 0.26193507422242035,
      "grad_norm": 0.2523619532585144,
      "learning_rate": 9.970819852261707e-06,
      "loss": 0.0055,
      "step": 40840
    },
    {
      "epoch": 0.26199921111620644,
      "grad_norm": 0.36605334281921387,
      "learning_rate": 9.970759440563075e-06,
      "loss": 0.0042,
      "step": 40850
    },
    {
      "epoch": 0.26206334800999254,
      "grad_norm": 0.3867529034614563,
      "learning_rate": 9.97069896657728e-06,
      "loss": 0.0064,
      "step": 40860
    },
    {
      "epoch": 0.26212748490377863,
      "grad_norm": 0.3114843964576721,
      "learning_rate": 9.970638430305071e-06,
      "loss": 0.0054,
      "step": 40870
    },
    {
      "epoch": 0.2621916217975647,
      "grad_norm": 0.12710262835025787,
      "learning_rate": 9.97057783174721e-06,
      "loss": 0.0045,
      "step": 40880
    },
    {
      "epoch": 0.2622557586913508,
      "grad_norm": 0.21920526027679443,
      "learning_rate": 9.970517170904459e-06,
      "loss": 0.0049,
      "step": 40890
    },
    {
      "epoch": 0.2623198955851369,
      "grad_norm": 0.08163820952177048,
      "learning_rate": 9.970456447777574e-06,
      "loss": 0.0074,
      "step": 40900
    },
    {
      "epoch": 0.262384032478923,
      "grad_norm": 0.31037309765815735,
      "learning_rate": 9.970395662367318e-06,
      "loss": 0.0053,
      "step": 40910
    },
    {
      "epoch": 0.2624481693727091,
      "grad_norm": 0.24946099519729614,
      "learning_rate": 9.970334814674452e-06,
      "loss": 0.0053,
      "step": 40920
    },
    {
      "epoch": 0.2625123062664952,
      "grad_norm": 0.12234389036893845,
      "learning_rate": 9.970273904699736e-06,
      "loss": 0.005,
      "step": 40930
    },
    {
      "epoch": 0.2625764431602813,
      "grad_norm": 0.7745259404182434,
      "learning_rate": 9.970212932443939e-06,
      "loss": 0.0053,
      "step": 40940
    },
    {
      "epoch": 0.2626405800540674,
      "grad_norm": 0.1711163967847824,
      "learning_rate": 9.97015189790782e-06,
      "loss": 0.0053,
      "step": 40950
    },
    {
      "epoch": 0.2627047169478535,
      "grad_norm": 0.10916420817375183,
      "learning_rate": 9.970090801092146e-06,
      "loss": 0.0039,
      "step": 40960
    },
    {
      "epoch": 0.2627688538416396,
      "grad_norm": 0.19058328866958618,
      "learning_rate": 9.970029641997683e-06,
      "loss": 0.0065,
      "step": 40970
    },
    {
      "epoch": 0.2628329907354257,
      "grad_norm": 0.19633597135543823,
      "learning_rate": 9.969968420625196e-06,
      "loss": 0.0048,
      "step": 40980
    },
    {
      "epoch": 0.2628971276292118,
      "grad_norm": 0.056958526372909546,
      "learning_rate": 9.969907136975451e-06,
      "loss": 0.0066,
      "step": 40990
    },
    {
      "epoch": 0.2629612645229979,
      "grad_norm": 0.13696123659610748,
      "learning_rate": 9.96984579104922e-06,
      "loss": 0.0044,
      "step": 41000
    },
    {
      "epoch": 0.263025401416784,
      "grad_norm": 0.47774285078048706,
      "learning_rate": 9.96978438284727e-06,
      "loss": 0.0073,
      "step": 41010
    },
    {
      "epoch": 0.2630895383105701,
      "grad_norm": 0.30265384912490845,
      "learning_rate": 9.969722912370367e-06,
      "loss": 0.0079,
      "step": 41020
    },
    {
      "epoch": 0.2631536752043562,
      "grad_norm": 0.3104373514652252,
      "learning_rate": 9.969661379619286e-06,
      "loss": 0.006,
      "step": 41030
    },
    {
      "epoch": 0.26321781209814227,
      "grad_norm": 0.5163923501968384,
      "learning_rate": 9.969599784594795e-06,
      "loss": 0.0064,
      "step": 41040
    },
    {
      "epoch": 0.26328194899192836,
      "grad_norm": 0.21843431890010834,
      "learning_rate": 9.969538127297668e-06,
      "loss": 0.0105,
      "step": 41050
    },
    {
      "epoch": 0.26334608588571445,
      "grad_norm": 0.26374852657318115,
      "learning_rate": 9.969476407728675e-06,
      "loss": 0.0072,
      "step": 41060
    },
    {
      "epoch": 0.26341022277950055,
      "grad_norm": 0.1182999536395073,
      "learning_rate": 9.969414625888594e-06,
      "loss": 0.006,
      "step": 41070
    },
    {
      "epoch": 0.26347435967328664,
      "grad_norm": 0.2888597846031189,
      "learning_rate": 9.969352781778195e-06,
      "loss": 0.0049,
      "step": 41080
    },
    {
      "epoch": 0.26353849656707273,
      "grad_norm": 0.09545408189296722,
      "learning_rate": 9.969290875398252e-06,
      "loss": 0.0059,
      "step": 41090
    },
    {
      "epoch": 0.2636026334608589,
      "grad_norm": 0.17045533657073975,
      "learning_rate": 9.969228906749544e-06,
      "loss": 0.0056,
      "step": 41100
    },
    {
      "epoch": 0.263666770354645,
      "grad_norm": 0.24985957145690918,
      "learning_rate": 9.969166875832848e-06,
      "loss": 0.004,
      "step": 41110
    },
    {
      "epoch": 0.26373090724843107,
      "grad_norm": 0.4133603870868683,
      "learning_rate": 9.969104782648938e-06,
      "loss": 0.0053,
      "step": 41120
    },
    {
      "epoch": 0.26379504414221716,
      "grad_norm": 0.30793261528015137,
      "learning_rate": 9.969042627198592e-06,
      "loss": 0.0061,
      "step": 41130
    },
    {
      "epoch": 0.26385918103600325,
      "grad_norm": 0.3550465703010559,
      "learning_rate": 9.968980409482594e-06,
      "loss": 0.0094,
      "step": 41140
    },
    {
      "epoch": 0.26392331792978935,
      "grad_norm": 0.13868685066699982,
      "learning_rate": 9.968918129501718e-06,
      "loss": 0.0047,
      "step": 41150
    },
    {
      "epoch": 0.26398745482357544,
      "grad_norm": 0.10272249579429626,
      "learning_rate": 9.968855787256749e-06,
      "loss": 0.0052,
      "step": 41160
    },
    {
      "epoch": 0.26405159171736153,
      "grad_norm": 0.0827287957072258,
      "learning_rate": 9.968793382748462e-06,
      "loss": 0.0046,
      "step": 41170
    },
    {
      "epoch": 0.2641157286111476,
      "grad_norm": 0.2657158374786377,
      "learning_rate": 9.968730915977647e-06,
      "loss": 0.0042,
      "step": 41180
    },
    {
      "epoch": 0.2641798655049337,
      "grad_norm": 0.12274724245071411,
      "learning_rate": 9.96866838694508e-06,
      "loss": 0.0058,
      "step": 41190
    },
    {
      "epoch": 0.2642440023987198,
      "grad_norm": 0.09837764501571655,
      "learning_rate": 9.968605795651548e-06,
      "loss": 0.0056,
      "step": 41200
    },
    {
      "epoch": 0.2643081392925059,
      "grad_norm": 0.3094668686389923,
      "learning_rate": 9.968543142097833e-06,
      "loss": 0.0089,
      "step": 41210
    },
    {
      "epoch": 0.264372276186292,
      "grad_norm": 0.30115365982055664,
      "learning_rate": 9.968480426284723e-06,
      "loss": 0.0054,
      "step": 41220
    },
    {
      "epoch": 0.2644364130800781,
      "grad_norm": 0.6357099413871765,
      "learning_rate": 9.968417648213e-06,
      "loss": 0.005,
      "step": 41230
    },
    {
      "epoch": 0.26450054997386424,
      "grad_norm": 0.40032604336738586,
      "learning_rate": 9.968354807883456e-06,
      "loss": 0.0051,
      "step": 41240
    },
    {
      "epoch": 0.26456468686765033,
      "grad_norm": 0.4655933678150177,
      "learning_rate": 9.968291905296874e-06,
      "loss": 0.0031,
      "step": 41250
    },
    {
      "epoch": 0.2646288237614364,
      "grad_norm": 0.3356774151325226,
      "learning_rate": 9.968228940454043e-06,
      "loss": 0.0057,
      "step": 41260
    },
    {
      "epoch": 0.2646929606552225,
      "grad_norm": 0.04913124814629555,
      "learning_rate": 9.968165913355754e-06,
      "loss": 0.0044,
      "step": 41270
    },
    {
      "epoch": 0.2647570975490086,
      "grad_norm": 0.1629490703344345,
      "learning_rate": 9.968102824002795e-06,
      "loss": 0.0044,
      "step": 41280
    },
    {
      "epoch": 0.2648212344427947,
      "grad_norm": 0.17841202020645142,
      "learning_rate": 9.968039672395957e-06,
      "loss": 0.0057,
      "step": 41290
    },
    {
      "epoch": 0.2648853713365808,
      "grad_norm": 0.11928600072860718,
      "learning_rate": 9.96797645853603e-06,
      "loss": 0.0074,
      "step": 41300
    },
    {
      "epoch": 0.2649495082303669,
      "grad_norm": 0.39552077651023865,
      "learning_rate": 9.96791318242381e-06,
      "loss": 0.007,
      "step": 41310
    },
    {
      "epoch": 0.265013645124153,
      "grad_norm": 0.14500652253627777,
      "learning_rate": 9.967849844060084e-06,
      "loss": 0.004,
      "step": 41320
    },
    {
      "epoch": 0.2650777820179391,
      "grad_norm": 0.24915923178195953,
      "learning_rate": 9.967786443445651e-06,
      "loss": 0.0056,
      "step": 41330
    },
    {
      "epoch": 0.26514191891172517,
      "grad_norm": 0.19904428720474243,
      "learning_rate": 9.967722980581303e-06,
      "loss": 0.0066,
      "step": 41340
    },
    {
      "epoch": 0.26520605580551126,
      "grad_norm": 0.22887201607227325,
      "learning_rate": 9.967659455467836e-06,
      "loss": 0.0035,
      "step": 41350
    },
    {
      "epoch": 0.26527019269929736,
      "grad_norm": 0.2342652678489685,
      "learning_rate": 9.967595868106045e-06,
      "loss": 0.0064,
      "step": 41360
    },
    {
      "epoch": 0.26533432959308345,
      "grad_norm": 0.17694617807865143,
      "learning_rate": 9.96753221849673e-06,
      "loss": 0.0061,
      "step": 41370
    },
    {
      "epoch": 0.2653984664868696,
      "grad_norm": 0.25880196690559387,
      "learning_rate": 9.967468506640684e-06,
      "loss": 0.0051,
      "step": 41380
    },
    {
      "epoch": 0.2654626033806557,
      "grad_norm": 0.10445227473974228,
      "learning_rate": 9.967404732538706e-06,
      "loss": 0.0033,
      "step": 41390
    },
    {
      "epoch": 0.2655267402744418,
      "grad_norm": 0.26488828659057617,
      "learning_rate": 9.967340896191598e-06,
      "loss": 0.0068,
      "step": 41400
    },
    {
      "epoch": 0.2655908771682279,
      "grad_norm": 0.13136650621891022,
      "learning_rate": 9.96727699760016e-06,
      "loss": 0.0057,
      "step": 41410
    },
    {
      "epoch": 0.26565501406201397,
      "grad_norm": 0.4326325058937073,
      "learning_rate": 9.967213036765188e-06,
      "loss": 0.0042,
      "step": 41420
    },
    {
      "epoch": 0.26571915095580007,
      "grad_norm": 0.3093279302120209,
      "learning_rate": 9.967149013687489e-06,
      "loss": 0.0033,
      "step": 41430
    },
    {
      "epoch": 0.26578328784958616,
      "grad_norm": 0.2260848581790924,
      "learning_rate": 9.967084928367862e-06,
      "loss": 0.0051,
      "step": 41440
    },
    {
      "epoch": 0.26584742474337225,
      "grad_norm": 0.23080188035964966,
      "learning_rate": 9.967020780807111e-06,
      "loss": 0.0069,
      "step": 41450
    },
    {
      "epoch": 0.26591156163715834,
      "grad_norm": 0.4775310754776001,
      "learning_rate": 9.96695657100604e-06,
      "loss": 0.0084,
      "step": 41460
    },
    {
      "epoch": 0.26597569853094444,
      "grad_norm": 0.1611340492963791,
      "learning_rate": 9.966892298965453e-06,
      "loss": 0.0074,
      "step": 41470
    },
    {
      "epoch": 0.26603983542473053,
      "grad_norm": 0.3432773947715759,
      "learning_rate": 9.966827964686157e-06,
      "loss": 0.0033,
      "step": 41480
    },
    {
      "epoch": 0.2661039723185166,
      "grad_norm": 0.09466341137886047,
      "learning_rate": 9.966763568168955e-06,
      "loss": 0.0077,
      "step": 41490
    },
    {
      "epoch": 0.2661681092123027,
      "grad_norm": 0.22389864921569824,
      "learning_rate": 9.966699109414657e-06,
      "loss": 0.0044,
      "step": 41500
    },
    {
      "epoch": 0.2662322461060888,
      "grad_norm": 0.1356392204761505,
      "learning_rate": 9.96663458842407e-06,
      "loss": 0.0055,
      "step": 41510
    },
    {
      "epoch": 0.26629638299987496,
      "grad_norm": 0.11065568774938583,
      "learning_rate": 9.966570005197999e-06,
      "loss": 0.0028,
      "step": 41520
    },
    {
      "epoch": 0.26636051989366105,
      "grad_norm": 0.5627606511116028,
      "learning_rate": 9.96650535973726e-06,
      "loss": 0.0069,
      "step": 41530
    },
    {
      "epoch": 0.26642465678744714,
      "grad_norm": 0.2600993514060974,
      "learning_rate": 9.966440652042657e-06,
      "loss": 0.0043,
      "step": 41540
    },
    {
      "epoch": 0.26648879368123324,
      "grad_norm": 0.14146417379379272,
      "learning_rate": 9.966375882115005e-06,
      "loss": 0.0113,
      "step": 41550
    },
    {
      "epoch": 0.26655293057501933,
      "grad_norm": 0.3202704191207886,
      "learning_rate": 9.966311049955111e-06,
      "loss": 0.0051,
      "step": 41560
    },
    {
      "epoch": 0.2666170674688054,
      "grad_norm": 0.18367043137550354,
      "learning_rate": 9.966246155563793e-06,
      "loss": 0.0066,
      "step": 41570
    },
    {
      "epoch": 0.2666812043625915,
      "grad_norm": 0.07598559558391571,
      "learning_rate": 9.96618119894186e-06,
      "loss": 0.0055,
      "step": 41580
    },
    {
      "epoch": 0.2667453412563776,
      "grad_norm": 0.3397144675254822,
      "learning_rate": 9.966116180090128e-06,
      "loss": 0.0072,
      "step": 41590
    },
    {
      "epoch": 0.2668094781501637,
      "grad_norm": 0.32975801825523376,
      "learning_rate": 9.96605109900941e-06,
      "loss": 0.0052,
      "step": 41600
    },
    {
      "epoch": 0.2668736150439498,
      "grad_norm": 0.26922792196273804,
      "learning_rate": 9.965985955700523e-06,
      "loss": 0.0068,
      "step": 41610
    },
    {
      "epoch": 0.2669377519377359,
      "grad_norm": 0.15959490835666656,
      "learning_rate": 9.965920750164283e-06,
      "loss": 0.0091,
      "step": 41620
    },
    {
      "epoch": 0.267001888831522,
      "grad_norm": 0.3165683150291443,
      "learning_rate": 9.965855482401507e-06,
      "loss": 0.0059,
      "step": 41630
    },
    {
      "epoch": 0.2670660257253081,
      "grad_norm": 0.1645023375749588,
      "learning_rate": 9.965790152413013e-06,
      "loss": 0.0039,
      "step": 41640
    },
    {
      "epoch": 0.26713016261909417,
      "grad_norm": 0.3173259198665619,
      "learning_rate": 9.965724760199618e-06,
      "loss": 0.0053,
      "step": 41650
    },
    {
      "epoch": 0.2671942995128803,
      "grad_norm": 0.2163485437631607,
      "learning_rate": 9.965659305762144e-06,
      "loss": 0.0071,
      "step": 41660
    },
    {
      "epoch": 0.2672584364066664,
      "grad_norm": 0.09772640466690063,
      "learning_rate": 9.96559378910141e-06,
      "loss": 0.0067,
      "step": 41670
    },
    {
      "epoch": 0.2673225733004525,
      "grad_norm": 0.16423320770263672,
      "learning_rate": 9.965528210218236e-06,
      "loss": 0.0082,
      "step": 41680
    },
    {
      "epoch": 0.2673867101942386,
      "grad_norm": 0.23379464447498322,
      "learning_rate": 9.965462569113447e-06,
      "loss": 0.0058,
      "step": 41690
    },
    {
      "epoch": 0.2674508470880247,
      "grad_norm": 0.619549036026001,
      "learning_rate": 9.965396865787861e-06,
      "loss": 0.0035,
      "step": 41700
    },
    {
      "epoch": 0.2675149839818108,
      "grad_norm": 0.21900302171707153,
      "learning_rate": 9.965331100242303e-06,
      "loss": 0.0063,
      "step": 41710
    },
    {
      "epoch": 0.2675791208755969,
      "grad_norm": 0.30182841420173645,
      "learning_rate": 9.9652652724776e-06,
      "loss": 0.004,
      "step": 41720
    },
    {
      "epoch": 0.26764325776938297,
      "grad_norm": 0.334244966506958,
      "learning_rate": 9.965199382494574e-06,
      "loss": 0.0055,
      "step": 41730
    },
    {
      "epoch": 0.26770739466316906,
      "grad_norm": 0.787428081035614,
      "learning_rate": 9.96513343029405e-06,
      "loss": 0.0069,
      "step": 41740
    },
    {
      "epoch": 0.26777153155695516,
      "grad_norm": 0.2864348590373993,
      "learning_rate": 9.965067415876857e-06,
      "loss": 0.006,
      "step": 41750
    },
    {
      "epoch": 0.26783566845074125,
      "grad_norm": 0.40290093421936035,
      "learning_rate": 9.965001339243819e-06,
      "loss": 0.0069,
      "step": 41760
    },
    {
      "epoch": 0.26789980534452734,
      "grad_norm": 0.2561686336994171,
      "learning_rate": 9.964935200395767e-06,
      "loss": 0.006,
      "step": 41770
    },
    {
      "epoch": 0.26796394223831344,
      "grad_norm": 0.19381536543369293,
      "learning_rate": 9.964868999333528e-06,
      "loss": 0.0055,
      "step": 41780
    },
    {
      "epoch": 0.26802807913209953,
      "grad_norm": 0.19894808530807495,
      "learning_rate": 9.964802736057933e-06,
      "loss": 0.0059,
      "step": 41790
    },
    {
      "epoch": 0.2680922160258857,
      "grad_norm": 0.16865375638008118,
      "learning_rate": 9.96473641056981e-06,
      "loss": 0.0023,
      "step": 41800
    },
    {
      "epoch": 0.26815635291967177,
      "grad_norm": 0.337950199842453,
      "learning_rate": 9.964670022869994e-06,
      "loss": 0.0083,
      "step": 41810
    },
    {
      "epoch": 0.26822048981345786,
      "grad_norm": 0.08005267381668091,
      "learning_rate": 9.964603572959312e-06,
      "loss": 0.0041,
      "step": 41820
    },
    {
      "epoch": 0.26828462670724396,
      "grad_norm": 0.25674504041671753,
      "learning_rate": 9.9645370608386e-06,
      "loss": 0.0079,
      "step": 41830
    },
    {
      "epoch": 0.26834876360103005,
      "grad_norm": 0.2807120680809021,
      "learning_rate": 9.96447048650869e-06,
      "loss": 0.0046,
      "step": 41840
    },
    {
      "epoch": 0.26841290049481614,
      "grad_norm": 0.5037015080451965,
      "learning_rate": 9.964403849970416e-06,
      "loss": 0.0056,
      "step": 41850
    },
    {
      "epoch": 0.26847703738860224,
      "grad_norm": 0.2273329496383667,
      "learning_rate": 9.964337151224617e-06,
      "loss": 0.0039,
      "step": 41860
    },
    {
      "epoch": 0.26854117428238833,
      "grad_norm": 0.15029305219650269,
      "learning_rate": 9.964270390272123e-06,
      "loss": 0.0056,
      "step": 41870
    },
    {
      "epoch": 0.2686053111761744,
      "grad_norm": 0.6880108714103699,
      "learning_rate": 9.964203567113773e-06,
      "loss": 0.0084,
      "step": 41880
    },
    {
      "epoch": 0.2686694480699605,
      "grad_norm": 0.1664203554391861,
      "learning_rate": 9.964136681750406e-06,
      "loss": 0.005,
      "step": 41890
    },
    {
      "epoch": 0.2687335849637466,
      "grad_norm": 0.445311963558197,
      "learning_rate": 9.964069734182858e-06,
      "loss": 0.007,
      "step": 41900
    },
    {
      "epoch": 0.2687977218575327,
      "grad_norm": 0.14195266366004944,
      "learning_rate": 9.964002724411967e-06,
      "loss": 0.0031,
      "step": 41910
    },
    {
      "epoch": 0.2688618587513188,
      "grad_norm": 0.0973130464553833,
      "learning_rate": 9.963935652438575e-06,
      "loss": 0.0052,
      "step": 41920
    },
    {
      "epoch": 0.2689259956451049,
      "grad_norm": 0.21904417872428894,
      "learning_rate": 9.963868518263521e-06,
      "loss": 0.0048,
      "step": 41930
    },
    {
      "epoch": 0.26899013253889104,
      "grad_norm": 0.39342308044433594,
      "learning_rate": 9.963801321887648e-06,
      "loss": 0.0055,
      "step": 41940
    },
    {
      "epoch": 0.26905426943267713,
      "grad_norm": 0.236109659075737,
      "learning_rate": 9.963734063311797e-06,
      "loss": 0.0048,
      "step": 41950
    },
    {
      "epoch": 0.2691184063264632,
      "grad_norm": 0.40211576223373413,
      "learning_rate": 9.963666742536811e-06,
      "loss": 0.0111,
      "step": 41960
    },
    {
      "epoch": 0.2691825432202493,
      "grad_norm": 0.07679940015077591,
      "learning_rate": 9.963599359563532e-06,
      "loss": 0.0052,
      "step": 41970
    },
    {
      "epoch": 0.2692466801140354,
      "grad_norm": 0.3705988824367523,
      "learning_rate": 9.963531914392806e-06,
      "loss": 0.005,
      "step": 41980
    },
    {
      "epoch": 0.2693108170078215,
      "grad_norm": 0.36707741022109985,
      "learning_rate": 9.963464407025478e-06,
      "loss": 0.0047,
      "step": 41990
    },
    {
      "epoch": 0.2693749539016076,
      "grad_norm": 0.31048306822776794,
      "learning_rate": 9.963396837462392e-06,
      "loss": 0.0063,
      "step": 42000
    },
    {
      "epoch": 0.2694390907953937,
      "grad_norm": 0.24470870196819305,
      "learning_rate": 9.963329205704397e-06,
      "loss": 0.006,
      "step": 42010
    },
    {
      "epoch": 0.2695032276891798,
      "grad_norm": 0.37310677766799927,
      "learning_rate": 9.963261511752341e-06,
      "loss": 0.0077,
      "step": 42020
    },
    {
      "epoch": 0.2695673645829659,
      "grad_norm": 0.23146171867847443,
      "learning_rate": 9.96319375560707e-06,
      "loss": 0.0063,
      "step": 42030
    },
    {
      "epoch": 0.26963150147675197,
      "grad_norm": 0.4126221835613251,
      "learning_rate": 9.963125937269435e-06,
      "loss": 0.0085,
      "step": 42040
    },
    {
      "epoch": 0.26969563837053806,
      "grad_norm": 0.37276023626327515,
      "learning_rate": 9.963058056740284e-06,
      "loss": 0.0113,
      "step": 42050
    },
    {
      "epoch": 0.26975977526432415,
      "grad_norm": 0.16797538101673126,
      "learning_rate": 9.962990114020469e-06,
      "loss": 0.0044,
      "step": 42060
    },
    {
      "epoch": 0.26982391215811025,
      "grad_norm": 0.26113346219062805,
      "learning_rate": 9.962922109110841e-06,
      "loss": 0.006,
      "step": 42070
    },
    {
      "epoch": 0.2698880490518964,
      "grad_norm": 0.3887992203235626,
      "learning_rate": 9.962854042012253e-06,
      "loss": 0.007,
      "step": 42080
    },
    {
      "epoch": 0.2699521859456825,
      "grad_norm": 0.14516444504261017,
      "learning_rate": 9.962785912725556e-06,
      "loss": 0.003,
      "step": 42090
    },
    {
      "epoch": 0.2700163228394686,
      "grad_norm": 0.12583817541599274,
      "learning_rate": 9.962717721251604e-06,
      "loss": 0.0049,
      "step": 42100
    },
    {
      "epoch": 0.2700804597332547,
      "grad_norm": 0.18013928830623627,
      "learning_rate": 9.962649467591253e-06,
      "loss": 0.005,
      "step": 42110
    },
    {
      "epoch": 0.27014459662704077,
      "grad_norm": 0.6400836706161499,
      "learning_rate": 9.962581151745358e-06,
      "loss": 0.0049,
      "step": 42120
    },
    {
      "epoch": 0.27020873352082686,
      "grad_norm": 0.6360874772071838,
      "learning_rate": 9.962512773714773e-06,
      "loss": 0.0096,
      "step": 42130
    },
    {
      "epoch": 0.27027287041461295,
      "grad_norm": 0.3784855604171753,
      "learning_rate": 9.962444333500358e-06,
      "loss": 0.0113,
      "step": 42140
    },
    {
      "epoch": 0.27033700730839905,
      "grad_norm": 0.32857540249824524,
      "learning_rate": 9.962375831102968e-06,
      "loss": 0.0102,
      "step": 42150
    },
    {
      "epoch": 0.27040114420218514,
      "grad_norm": 0.24528630077838898,
      "learning_rate": 9.96230726652346e-06,
      "loss": 0.0062,
      "step": 42160
    },
    {
      "epoch": 0.27046528109597123,
      "grad_norm": 0.4703960120677948,
      "learning_rate": 9.962238639762697e-06,
      "loss": 0.0057,
      "step": 42170
    },
    {
      "epoch": 0.2705294179897573,
      "grad_norm": 0.2784826159477234,
      "learning_rate": 9.96216995082154e-06,
      "loss": 0.0102,
      "step": 42180
    },
    {
      "epoch": 0.2705935548835434,
      "grad_norm": 0.34280872344970703,
      "learning_rate": 9.962101199700845e-06,
      "loss": 0.0069,
      "step": 42190
    },
    {
      "epoch": 0.2706576917773295,
      "grad_norm": 0.34265586733818054,
      "learning_rate": 9.962032386401475e-06,
      "loss": 0.0043,
      "step": 42200
    },
    {
      "epoch": 0.2707218286711156,
      "grad_norm": 0.14476878941059113,
      "learning_rate": 9.961963510924295e-06,
      "loss": 0.0051,
      "step": 42210
    },
    {
      "epoch": 0.27078596556490175,
      "grad_norm": 0.028610268607735634,
      "learning_rate": 9.961894573270163e-06,
      "loss": 0.0033,
      "step": 42220
    },
    {
      "epoch": 0.27085010245868785,
      "grad_norm": 0.3419274687767029,
      "learning_rate": 9.961825573439947e-06,
      "loss": 0.0047,
      "step": 42230
    },
    {
      "epoch": 0.27091423935247394,
      "grad_norm": 0.11877810209989548,
      "learning_rate": 9.96175651143451e-06,
      "loss": 0.0054,
      "step": 42240
    },
    {
      "epoch": 0.27097837624626003,
      "grad_norm": 0.11344487965106964,
      "learning_rate": 9.96168738725472e-06,
      "loss": 0.0051,
      "step": 42250
    },
    {
      "epoch": 0.2710425131400461,
      "grad_norm": 0.2172202169895172,
      "learning_rate": 9.96161820090144e-06,
      "loss": 0.0043,
      "step": 42260
    },
    {
      "epoch": 0.2711066500338322,
      "grad_norm": 0.1751595288515091,
      "learning_rate": 9.961548952375537e-06,
      "loss": 0.0066,
      "step": 42270
    },
    {
      "epoch": 0.2711707869276183,
      "grad_norm": 0.20869243144989014,
      "learning_rate": 9.96147964167788e-06,
      "loss": 0.0052,
      "step": 42280
    },
    {
      "epoch": 0.2712349238214044,
      "grad_norm": 0.35894539952278137,
      "learning_rate": 9.961410268809338e-06,
      "loss": 0.0058,
      "step": 42290
    },
    {
      "epoch": 0.2712990607151905,
      "grad_norm": 0.21298572421073914,
      "learning_rate": 9.961340833770778e-06,
      "loss": 0.0054,
      "step": 42300
    },
    {
      "epoch": 0.2713631976089766,
      "grad_norm": 0.10361744463443756,
      "learning_rate": 9.961271336563073e-06,
      "loss": 0.0081,
      "step": 42310
    },
    {
      "epoch": 0.2714273345027627,
      "grad_norm": 0.12398401647806168,
      "learning_rate": 9.961201777187091e-06,
      "loss": 0.0029,
      "step": 42320
    },
    {
      "epoch": 0.2714914713965488,
      "grad_norm": 0.08253161609172821,
      "learning_rate": 9.961132155643704e-06,
      "loss": 0.006,
      "step": 42330
    },
    {
      "epoch": 0.27155560829033487,
      "grad_norm": 0.12968796491622925,
      "learning_rate": 9.961062471933788e-06,
      "loss": 0.0036,
      "step": 42340
    },
    {
      "epoch": 0.27161974518412096,
      "grad_norm": 0.03698797523975372,
      "learning_rate": 9.960992726058212e-06,
      "loss": 0.0043,
      "step": 42350
    },
    {
      "epoch": 0.27168388207790706,
      "grad_norm": 0.3756403923034668,
      "learning_rate": 9.960922918017852e-06,
      "loss": 0.0144,
      "step": 42360
    },
    {
      "epoch": 0.2717480189716932,
      "grad_norm": 0.776396632194519,
      "learning_rate": 9.960853047813583e-06,
      "loss": 0.0065,
      "step": 42370
    },
    {
      "epoch": 0.2718121558654793,
      "grad_norm": 0.20592395961284637,
      "learning_rate": 9.960783115446279e-06,
      "loss": 0.0029,
      "step": 42380
    },
    {
      "epoch": 0.2718762927592654,
      "grad_norm": 0.5793088674545288,
      "learning_rate": 9.960713120916818e-06,
      "loss": 0.0063,
      "step": 42390
    },
    {
      "epoch": 0.2719404296530515,
      "grad_norm": 0.2557103633880615,
      "learning_rate": 9.960643064226077e-06,
      "loss": 0.0038,
      "step": 42400
    },
    {
      "epoch": 0.2720045665468376,
      "grad_norm": 0.2395615428686142,
      "learning_rate": 9.960572945374932e-06,
      "loss": 0.0091,
      "step": 42410
    },
    {
      "epoch": 0.27206870344062367,
      "grad_norm": 0.4039780795574188,
      "learning_rate": 9.960502764364262e-06,
      "loss": 0.0062,
      "step": 42420
    },
    {
      "epoch": 0.27213284033440976,
      "grad_norm": 0.18832018971443176,
      "learning_rate": 9.960432521194947e-06,
      "loss": 0.0052,
      "step": 42430
    },
    {
      "epoch": 0.27219697722819586,
      "grad_norm": 0.2135220468044281,
      "learning_rate": 9.960362215867868e-06,
      "loss": 0.0044,
      "step": 42440
    },
    {
      "epoch": 0.27226111412198195,
      "grad_norm": 0.28974127769470215,
      "learning_rate": 9.960291848383904e-06,
      "loss": 0.0056,
      "step": 42450
    },
    {
      "epoch": 0.27232525101576804,
      "grad_norm": 0.17635077238082886,
      "learning_rate": 9.96022141874394e-06,
      "loss": 0.0067,
      "step": 42460
    },
    {
      "epoch": 0.27238938790955414,
      "grad_norm": 0.12280075252056122,
      "learning_rate": 9.960150926948857e-06,
      "loss": 0.0053,
      "step": 42470
    },
    {
      "epoch": 0.27245352480334023,
      "grad_norm": 0.1510230302810669,
      "learning_rate": 9.960080372999537e-06,
      "loss": 0.0073,
      "step": 42480
    },
    {
      "epoch": 0.2725176616971263,
      "grad_norm": 0.21415026485919952,
      "learning_rate": 9.960009756896865e-06,
      "loss": 0.0039,
      "step": 42490
    },
    {
      "epoch": 0.2725817985909124,
      "grad_norm": 0.20693327486515045,
      "learning_rate": 9.959939078641725e-06,
      "loss": 0.0031,
      "step": 42500
    },
    {
      "epoch": 0.27264593548469856,
      "grad_norm": 0.192563995718956,
      "learning_rate": 9.959868338235004e-06,
      "loss": 0.0049,
      "step": 42510
    },
    {
      "epoch": 0.27271007237848466,
      "grad_norm": 0.1595400720834732,
      "learning_rate": 9.959797535677589e-06,
      "loss": 0.004,
      "step": 42520
    },
    {
      "epoch": 0.27277420927227075,
      "grad_norm": 0.3150067925453186,
      "learning_rate": 9.959726670970366e-06,
      "loss": 0.0085,
      "step": 42530
    },
    {
      "epoch": 0.27283834616605684,
      "grad_norm": 0.08167749643325806,
      "learning_rate": 9.959655744114223e-06,
      "loss": 0.0046,
      "step": 42540
    },
    {
      "epoch": 0.27290248305984294,
      "grad_norm": 0.4045458734035492,
      "learning_rate": 9.959584755110048e-06,
      "loss": 0.0074,
      "step": 42550
    },
    {
      "epoch": 0.27296661995362903,
      "grad_norm": 0.30020734667778015,
      "learning_rate": 9.959513703958732e-06,
      "loss": 0.004,
      "step": 42560
    },
    {
      "epoch": 0.2730307568474151,
      "grad_norm": 0.28891295194625854,
      "learning_rate": 9.959442590661165e-06,
      "loss": 0.0036,
      "step": 42570
    },
    {
      "epoch": 0.2730948937412012,
      "grad_norm": 0.10582451522350311,
      "learning_rate": 9.959371415218238e-06,
      "loss": 0.0051,
      "step": 42580
    },
    {
      "epoch": 0.2731590306349873,
      "grad_norm": 0.3654150068759918,
      "learning_rate": 9.959300177630842e-06,
      "loss": 0.0053,
      "step": 42590
    },
    {
      "epoch": 0.2732231675287734,
      "grad_norm": 0.43862292170524597,
      "learning_rate": 9.95922887789987e-06,
      "loss": 0.0043,
      "step": 42600
    },
    {
      "epoch": 0.2732873044225595,
      "grad_norm": 0.5285431146621704,
      "learning_rate": 9.959157516026217e-06,
      "loss": 0.0057,
      "step": 42610
    },
    {
      "epoch": 0.2733514413163456,
      "grad_norm": 0.17073945701122284,
      "learning_rate": 9.959086092010776e-06,
      "loss": 0.0077,
      "step": 42620
    },
    {
      "epoch": 0.2734155782101317,
      "grad_norm": 0.1495482623577118,
      "learning_rate": 9.959014605854443e-06,
      "loss": 0.0035,
      "step": 42630
    },
    {
      "epoch": 0.2734797151039178,
      "grad_norm": 0.40101027488708496,
      "learning_rate": 9.958943057558111e-06,
      "loss": 0.0076,
      "step": 42640
    },
    {
      "epoch": 0.2735438519977039,
      "grad_norm": 0.2977857291698456,
      "learning_rate": 9.958871447122678e-06,
      "loss": 0.0071,
      "step": 42650
    },
    {
      "epoch": 0.27360798889149,
      "grad_norm": 0.2916841506958008,
      "learning_rate": 9.958799774549044e-06,
      "loss": 0.0042,
      "step": 42660
    },
    {
      "epoch": 0.2736721257852761,
      "grad_norm": 0.15789587795734406,
      "learning_rate": 9.958728039838104e-06,
      "loss": 0.0048,
      "step": 42670
    },
    {
      "epoch": 0.2737362626790622,
      "grad_norm": 0.11915109306573868,
      "learning_rate": 9.958656242990757e-06,
      "loss": 0.0068,
      "step": 42680
    },
    {
      "epoch": 0.2738003995728483,
      "grad_norm": 0.29966673254966736,
      "learning_rate": 9.958584384007904e-06,
      "loss": 0.0041,
      "step": 42690
    },
    {
      "epoch": 0.2738645364666344,
      "grad_norm": 0.0362437441945076,
      "learning_rate": 9.958512462890444e-06,
      "loss": 0.0051,
      "step": 42700
    },
    {
      "epoch": 0.2739286733604205,
      "grad_norm": 0.1600048989057541,
      "learning_rate": 9.95844047963928e-06,
      "loss": 0.0058,
      "step": 42710
    },
    {
      "epoch": 0.2739928102542066,
      "grad_norm": 0.24165469408035278,
      "learning_rate": 9.958368434255312e-06,
      "loss": 0.0051,
      "step": 42720
    },
    {
      "epoch": 0.27405694714799267,
      "grad_norm": 0.32953357696533203,
      "learning_rate": 9.958296326739444e-06,
      "loss": 0.0053,
      "step": 42730
    },
    {
      "epoch": 0.27412108404177876,
      "grad_norm": 0.47327950596809387,
      "learning_rate": 9.95822415709258e-06,
      "loss": 0.0071,
      "step": 42740
    },
    {
      "epoch": 0.27418522093556486,
      "grad_norm": 0.2136109620332718,
      "learning_rate": 9.958151925315624e-06,
      "loss": 0.0066,
      "step": 42750
    },
    {
      "epoch": 0.27424935782935095,
      "grad_norm": 0.28177398443222046,
      "learning_rate": 9.95807963140948e-06,
      "loss": 0.0071,
      "step": 42760
    },
    {
      "epoch": 0.27431349472313704,
      "grad_norm": 0.25047868490219116,
      "learning_rate": 9.958007275375054e-06,
      "loss": 0.0057,
      "step": 42770
    },
    {
      "epoch": 0.27437763161692313,
      "grad_norm": 0.6044594049453735,
      "learning_rate": 9.957934857213256e-06,
      "loss": 0.0034,
      "step": 42780
    },
    {
      "epoch": 0.2744417685107093,
      "grad_norm": 0.37561270594596863,
      "learning_rate": 9.957862376924989e-06,
      "loss": 0.0051,
      "step": 42790
    },
    {
      "epoch": 0.2745059054044954,
      "grad_norm": 0.5243725180625916,
      "learning_rate": 9.957789834511164e-06,
      "loss": 0.0075,
      "step": 42800
    },
    {
      "epoch": 0.27457004229828147,
      "grad_norm": 0.35506904125213623,
      "learning_rate": 9.957717229972687e-06,
      "loss": 0.0046,
      "step": 42810
    },
    {
      "epoch": 0.27463417919206756,
      "grad_norm": 0.2856937646865845,
      "learning_rate": 9.957644563310472e-06,
      "loss": 0.0085,
      "step": 42820
    },
    {
      "epoch": 0.27469831608585366,
      "grad_norm": 0.29843416810035706,
      "learning_rate": 9.957571834525427e-06,
      "loss": 0.0069,
      "step": 42830
    },
    {
      "epoch": 0.27476245297963975,
      "grad_norm": 0.12337861210107803,
      "learning_rate": 9.957499043618464e-06,
      "loss": 0.0068,
      "step": 42840
    },
    {
      "epoch": 0.27482658987342584,
      "grad_norm": 0.12973666191101074,
      "learning_rate": 9.957426190590494e-06,
      "loss": 0.0044,
      "step": 42850
    },
    {
      "epoch": 0.27489072676721193,
      "grad_norm": 0.74559485912323,
      "learning_rate": 9.957353275442431e-06,
      "loss": 0.0092,
      "step": 42860
    },
    {
      "epoch": 0.27495486366099803,
      "grad_norm": 0.12368279695510864,
      "learning_rate": 9.95728029817519e-06,
      "loss": 0.0036,
      "step": 42870
    },
    {
      "epoch": 0.2750190005547841,
      "grad_norm": 0.20460005104541779,
      "learning_rate": 9.957207258789683e-06,
      "loss": 0.0044,
      "step": 42880
    },
    {
      "epoch": 0.2750831374485702,
      "grad_norm": 0.3100970983505249,
      "learning_rate": 9.957134157286825e-06,
      "loss": 0.0061,
      "step": 42890
    },
    {
      "epoch": 0.2751472743423563,
      "grad_norm": 0.10070610046386719,
      "learning_rate": 9.957060993667534e-06,
      "loss": 0.0051,
      "step": 42900
    },
    {
      "epoch": 0.2752114112361424,
      "grad_norm": 0.2688789665699005,
      "learning_rate": 9.956987767932727e-06,
      "loss": 0.0042,
      "step": 42910
    },
    {
      "epoch": 0.2752755481299285,
      "grad_norm": 0.2600405812263489,
      "learning_rate": 9.956914480083319e-06,
      "loss": 0.0079,
      "step": 42920
    },
    {
      "epoch": 0.27533968502371464,
      "grad_norm": 0.07776429504156113,
      "learning_rate": 9.956841130120232e-06,
      "loss": 0.0048,
      "step": 42930
    },
    {
      "epoch": 0.27540382191750074,
      "grad_norm": 0.42027747631073,
      "learning_rate": 9.95676771804438e-06,
      "loss": 0.0058,
      "step": 42940
    },
    {
      "epoch": 0.27546795881128683,
      "grad_norm": 0.11998993158340454,
      "learning_rate": 9.956694243856689e-06,
      "loss": 0.0047,
      "step": 42950
    },
    {
      "epoch": 0.2755320957050729,
      "grad_norm": 0.26443660259246826,
      "learning_rate": 9.956620707558076e-06,
      "loss": 0.0073,
      "step": 42960
    },
    {
      "epoch": 0.275596232598859,
      "grad_norm": 0.408774197101593,
      "learning_rate": 9.95654710914946e-06,
      "loss": 0.0055,
      "step": 42970
    },
    {
      "epoch": 0.2756603694926451,
      "grad_norm": 0.13061800599098206,
      "learning_rate": 9.95647344863177e-06,
      "loss": 0.0035,
      "step": 42980
    },
    {
      "epoch": 0.2757245063864312,
      "grad_norm": 0.29762259125709534,
      "learning_rate": 9.956399726005924e-06,
      "loss": 0.0067,
      "step": 42990
    },
    {
      "epoch": 0.2757886432802173,
      "grad_norm": 0.1276562511920929,
      "learning_rate": 9.956325941272847e-06,
      "loss": 0.0072,
      "step": 43000
    },
    {
      "epoch": 0.2758527801740034,
      "grad_norm": 0.39894869923591614,
      "learning_rate": 9.956252094433464e-06,
      "loss": 0.0068,
      "step": 43010
    },
    {
      "epoch": 0.2759169170677895,
      "grad_norm": 0.27307799458503723,
      "learning_rate": 9.9561781854887e-06,
      "loss": 0.0051,
      "step": 43020
    },
    {
      "epoch": 0.2759810539615756,
      "grad_norm": 0.38455525040626526,
      "learning_rate": 9.956104214439481e-06,
      "loss": 0.0098,
      "step": 43030
    },
    {
      "epoch": 0.27604519085536167,
      "grad_norm": 0.16045977175235748,
      "learning_rate": 9.956030181286736e-06,
      "loss": 0.0038,
      "step": 43040
    },
    {
      "epoch": 0.27610932774914776,
      "grad_norm": 0.3692812919616699,
      "learning_rate": 9.955956086031387e-06,
      "loss": 0.0032,
      "step": 43050
    },
    {
      "epoch": 0.27617346464293385,
      "grad_norm": 0.059148553758859634,
      "learning_rate": 9.955881928674369e-06,
      "loss": 0.0062,
      "step": 43060
    },
    {
      "epoch": 0.27623760153672,
      "grad_norm": 0.2269824594259262,
      "learning_rate": 9.955807709216609e-06,
      "loss": 0.005,
      "step": 43070
    },
    {
      "epoch": 0.2763017384305061,
      "grad_norm": 0.26325589418411255,
      "learning_rate": 9.955733427659034e-06,
      "loss": 0.0066,
      "step": 43080
    },
    {
      "epoch": 0.2763658753242922,
      "grad_norm": 0.22607797384262085,
      "learning_rate": 9.95565908400258e-06,
      "loss": 0.0062,
      "step": 43090
    },
    {
      "epoch": 0.2764300122180783,
      "grad_norm": 0.24247755110263824,
      "learning_rate": 9.955584678248173e-06,
      "loss": 0.0061,
      "step": 43100
    },
    {
      "epoch": 0.2764941491118644,
      "grad_norm": 0.2428087443113327,
      "learning_rate": 9.95551021039675e-06,
      "loss": 0.0044,
      "step": 43110
    },
    {
      "epoch": 0.27655828600565047,
      "grad_norm": 0.4491789937019348,
      "learning_rate": 9.955435680449243e-06,
      "loss": 0.0072,
      "step": 43120
    },
    {
      "epoch": 0.27662242289943656,
      "grad_norm": 0.28530851006507874,
      "learning_rate": 9.955361088406585e-06,
      "loss": 0.0073,
      "step": 43130
    },
    {
      "epoch": 0.27668655979322265,
      "grad_norm": 0.3485097885131836,
      "learning_rate": 9.95528643426971e-06,
      "loss": 0.0074,
      "step": 43140
    },
    {
      "epoch": 0.27675069668700875,
      "grad_norm": 0.2821292281150818,
      "learning_rate": 9.955211718039558e-06,
      "loss": 0.0052,
      "step": 43150
    },
    {
      "epoch": 0.27681483358079484,
      "grad_norm": 0.09028469026088715,
      "learning_rate": 9.955136939717057e-06,
      "loss": 0.006,
      "step": 43160
    },
    {
      "epoch": 0.27687897047458093,
      "grad_norm": 0.24043463170528412,
      "learning_rate": 9.955062099303151e-06,
      "loss": 0.0036,
      "step": 43170
    },
    {
      "epoch": 0.276943107368367,
      "grad_norm": 0.0799112319946289,
      "learning_rate": 9.954987196798776e-06,
      "loss": 0.0041,
      "step": 43180
    },
    {
      "epoch": 0.2770072442621531,
      "grad_norm": 0.28518977761268616,
      "learning_rate": 9.95491223220487e-06,
      "loss": 0.0057,
      "step": 43190
    },
    {
      "epoch": 0.2770713811559392,
      "grad_norm": 0.1286603808403015,
      "learning_rate": 9.954837205522371e-06,
      "loss": 0.005,
      "step": 43200
    },
    {
      "epoch": 0.27713551804972536,
      "grad_norm": 0.3737489879131317,
      "learning_rate": 9.95476211675222e-06,
      "loss": 0.0036,
      "step": 43210
    },
    {
      "epoch": 0.27719965494351145,
      "grad_norm": 0.401049941778183,
      "learning_rate": 9.954686965895361e-06,
      "loss": 0.0037,
      "step": 43220
    },
    {
      "epoch": 0.27726379183729755,
      "grad_norm": 0.2654677927494049,
      "learning_rate": 9.954611752952733e-06,
      "loss": 0.0064,
      "step": 43230
    },
    {
      "epoch": 0.27732792873108364,
      "grad_norm": 0.4903867244720459,
      "learning_rate": 9.954536477925279e-06,
      "loss": 0.0071,
      "step": 43240
    },
    {
      "epoch": 0.27739206562486973,
      "grad_norm": 0.26085662841796875,
      "learning_rate": 9.95446114081394e-06,
      "loss": 0.0042,
      "step": 43250
    },
    {
      "epoch": 0.2774562025186558,
      "grad_norm": 0.18222586810588837,
      "learning_rate": 9.954385741619663e-06,
      "loss": 0.0037,
      "step": 43260
    },
    {
      "epoch": 0.2775203394124419,
      "grad_norm": 0.0768163800239563,
      "learning_rate": 9.954310280343394e-06,
      "loss": 0.007,
      "step": 43270
    },
    {
      "epoch": 0.277584476306228,
      "grad_norm": 0.11794343590736389,
      "learning_rate": 9.954234756986072e-06,
      "loss": 0.0076,
      "step": 43280
    },
    {
      "epoch": 0.2776486132000141,
      "grad_norm": 0.07045776396989822,
      "learning_rate": 9.954159171548653e-06,
      "loss": 0.0054,
      "step": 43290
    },
    {
      "epoch": 0.2777127500938002,
      "grad_norm": 0.3573853373527527,
      "learning_rate": 9.954083524032075e-06,
      "loss": 0.0074,
      "step": 43300
    },
    {
      "epoch": 0.2777768869875863,
      "grad_norm": 0.0045158397406339645,
      "learning_rate": 9.95400781443729e-06,
      "loss": 0.0089,
      "step": 43310
    },
    {
      "epoch": 0.2778410238813724,
      "grad_norm": 0.1320253610610962,
      "learning_rate": 9.953932042765247e-06,
      "loss": 0.0047,
      "step": 43320
    },
    {
      "epoch": 0.2779051607751585,
      "grad_norm": 0.2115936279296875,
      "learning_rate": 9.953856209016895e-06,
      "loss": 0.0045,
      "step": 43330
    },
    {
      "epoch": 0.27796929766894457,
      "grad_norm": 0.13962361216545105,
      "learning_rate": 9.953780313193185e-06,
      "loss": 0.0082,
      "step": 43340
    },
    {
      "epoch": 0.2780334345627307,
      "grad_norm": 0.206785187125206,
      "learning_rate": 9.953704355295066e-06,
      "loss": 0.0066,
      "step": 43350
    },
    {
      "epoch": 0.2780975714565168,
      "grad_norm": 0.15403756499290466,
      "learning_rate": 9.953628335323494e-06,
      "loss": 0.0054,
      "step": 43360
    },
    {
      "epoch": 0.2781617083503029,
      "grad_norm": 0.124571792781353,
      "learning_rate": 9.953552253279415e-06,
      "loss": 0.0035,
      "step": 43370
    },
    {
      "epoch": 0.278225845244089,
      "grad_norm": 0.12376170605421066,
      "learning_rate": 9.953476109163788e-06,
      "loss": 0.0041,
      "step": 43380
    },
    {
      "epoch": 0.2782899821378751,
      "grad_norm": 0.2316630780696869,
      "learning_rate": 9.953399902977565e-06,
      "loss": 0.0064,
      "step": 43390
    },
    {
      "epoch": 0.2783541190316612,
      "grad_norm": 0.24657541513442993,
      "learning_rate": 9.953323634721701e-06,
      "loss": 0.004,
      "step": 43400
    },
    {
      "epoch": 0.2784182559254473,
      "grad_norm": 0.21009762585163116,
      "learning_rate": 9.953247304397151e-06,
      "loss": 0.0032,
      "step": 43410
    },
    {
      "epoch": 0.27848239281923337,
      "grad_norm": 0.17280659079551697,
      "learning_rate": 9.953170912004873e-06,
      "loss": 0.0057,
      "step": 43420
    },
    {
      "epoch": 0.27854652971301946,
      "grad_norm": 0.20576423406600952,
      "learning_rate": 9.953094457545824e-06,
      "loss": 0.0038,
      "step": 43430
    },
    {
      "epoch": 0.27861066660680556,
      "grad_norm": 0.1083400622010231,
      "learning_rate": 9.953017941020959e-06,
      "loss": 0.0038,
      "step": 43440
    },
    {
      "epoch": 0.27867480350059165,
      "grad_norm": 0.16818398237228394,
      "learning_rate": 9.952941362431242e-06,
      "loss": 0.0047,
      "step": 43450
    },
    {
      "epoch": 0.27873894039437774,
      "grad_norm": 0.1237788051366806,
      "learning_rate": 9.952864721777629e-06,
      "loss": 0.0033,
      "step": 43460
    },
    {
      "epoch": 0.27880307728816384,
      "grad_norm": 0.42831873893737793,
      "learning_rate": 9.952788019061082e-06,
      "loss": 0.0061,
      "step": 43470
    },
    {
      "epoch": 0.27886721418194993,
      "grad_norm": 0.20468097925186157,
      "learning_rate": 9.95271125428256e-06,
      "loss": 0.0037,
      "step": 43480
    },
    {
      "epoch": 0.2789313510757361,
      "grad_norm": 0.3154074549674988,
      "learning_rate": 9.952634427443027e-06,
      "loss": 0.0094,
      "step": 43490
    },
    {
      "epoch": 0.27899548796952217,
      "grad_norm": 0.3426404893398285,
      "learning_rate": 9.952557538543445e-06,
      "loss": 0.0046,
      "step": 43500
    },
    {
      "epoch": 0.27905962486330826,
      "grad_norm": 0.11730002611875534,
      "learning_rate": 9.952480587584779e-06,
      "loss": 0.0043,
      "step": 43510
    },
    {
      "epoch": 0.27912376175709436,
      "grad_norm": 0.22600385546684265,
      "learning_rate": 9.952403574567991e-06,
      "loss": 0.0055,
      "step": 43520
    },
    {
      "epoch": 0.27918789865088045,
      "grad_norm": 0.2184022217988968,
      "learning_rate": 9.952326499494046e-06,
      "loss": 0.0046,
      "step": 43530
    },
    {
      "epoch": 0.27925203554466654,
      "grad_norm": 0.15621612966060638,
      "learning_rate": 9.952249362363909e-06,
      "loss": 0.0053,
      "step": 43540
    },
    {
      "epoch": 0.27931617243845264,
      "grad_norm": 0.2692352533340454,
      "learning_rate": 9.95217216317855e-06,
      "loss": 0.0034,
      "step": 43550
    },
    {
      "epoch": 0.27938030933223873,
      "grad_norm": 0.2867933511734009,
      "learning_rate": 9.952094901938935e-06,
      "loss": 0.006,
      "step": 43560
    },
    {
      "epoch": 0.2794444462260248,
      "grad_norm": 0.2318105250597,
      "learning_rate": 9.952017578646032e-06,
      "loss": 0.013,
      "step": 43570
    },
    {
      "epoch": 0.2795085831198109,
      "grad_norm": 0.15158458054065704,
      "learning_rate": 9.951940193300808e-06,
      "loss": 0.0043,
      "step": 43580
    },
    {
      "epoch": 0.279572720013597,
      "grad_norm": 0.19293011724948883,
      "learning_rate": 9.951862745904235e-06,
      "loss": 0.0074,
      "step": 43590
    },
    {
      "epoch": 0.2796368569073831,
      "grad_norm": 0.23179151117801666,
      "learning_rate": 9.951785236457283e-06,
      "loss": 0.0051,
      "step": 43600
    },
    {
      "epoch": 0.2797009938011692,
      "grad_norm": 0.25460487604141235,
      "learning_rate": 9.951707664960922e-06,
      "loss": 0.0068,
      "step": 43610
    },
    {
      "epoch": 0.2797651306949553,
      "grad_norm": 0.16452178359031677,
      "learning_rate": 9.951630031416127e-06,
      "loss": 0.0045,
      "step": 43620
    },
    {
      "epoch": 0.27982926758874144,
      "grad_norm": 0.1425665020942688,
      "learning_rate": 9.951552335823866e-06,
      "loss": 0.004,
      "step": 43630
    },
    {
      "epoch": 0.27989340448252753,
      "grad_norm": 0.2122945785522461,
      "learning_rate": 9.951474578185117e-06,
      "loss": 0.0033,
      "step": 43640
    },
    {
      "epoch": 0.2799575413763136,
      "grad_norm": 0.08789437264204025,
      "learning_rate": 9.951396758500854e-06,
      "loss": 0.0046,
      "step": 43650
    },
    {
      "epoch": 0.2800216782700997,
      "grad_norm": 0.25604885816574097,
      "learning_rate": 9.951318876772049e-06,
      "loss": 0.0076,
      "step": 43660
    },
    {
      "epoch": 0.2800858151638858,
      "grad_norm": 0.07172763347625732,
      "learning_rate": 9.951240932999681e-06,
      "loss": 0.0054,
      "step": 43670
    },
    {
      "epoch": 0.2801499520576719,
      "grad_norm": 0.3497158885002136,
      "learning_rate": 9.951162927184724e-06,
      "loss": 0.0074,
      "step": 43680
    },
    {
      "epoch": 0.280214088951458,
      "grad_norm": 0.21498006582260132,
      "learning_rate": 9.951084859328159e-06,
      "loss": 0.0031,
      "step": 43690
    },
    {
      "epoch": 0.2802782258452441,
      "grad_norm": 0.35907694697380066,
      "learning_rate": 9.95100672943096e-06,
      "loss": 0.0035,
      "step": 43700
    },
    {
      "epoch": 0.2803423627390302,
      "grad_norm": 0.2003171443939209,
      "learning_rate": 9.95092853749411e-06,
      "loss": 0.0043,
      "step": 43710
    },
    {
      "epoch": 0.2804064996328163,
      "grad_norm": 0.33922964334487915,
      "learning_rate": 9.950850283518587e-06,
      "loss": 0.0043,
      "step": 43720
    },
    {
      "epoch": 0.28047063652660237,
      "grad_norm": 0.5845425724983215,
      "learning_rate": 9.95077196750537e-06,
      "loss": 0.0121,
      "step": 43730
    },
    {
      "epoch": 0.28053477342038846,
      "grad_norm": 0.16597582399845123,
      "learning_rate": 9.950693589455444e-06,
      "loss": 0.0094,
      "step": 43740
    },
    {
      "epoch": 0.28059891031417455,
      "grad_norm": 0.2852199077606201,
      "learning_rate": 9.950615149369788e-06,
      "loss": 0.0093,
      "step": 43750
    },
    {
      "epoch": 0.28066304720796065,
      "grad_norm": 0.21773932874202728,
      "learning_rate": 9.950536647249387e-06,
      "loss": 0.0042,
      "step": 43760
    },
    {
      "epoch": 0.28072718410174674,
      "grad_norm": 0.09281706809997559,
      "learning_rate": 9.950458083095222e-06,
      "loss": 0.0062,
      "step": 43770
    },
    {
      "epoch": 0.2807913209955329,
      "grad_norm": 0.07119549065828323,
      "learning_rate": 9.95037945690828e-06,
      "loss": 0.0046,
      "step": 43780
    },
    {
      "epoch": 0.280855457889319,
      "grad_norm": 0.06815053522586823,
      "learning_rate": 9.950300768689547e-06,
      "loss": 0.0063,
      "step": 43790
    },
    {
      "epoch": 0.2809195947831051,
      "grad_norm": 0.24725092947483063,
      "learning_rate": 9.950222018440006e-06,
      "loss": 0.0043,
      "step": 43800
    },
    {
      "epoch": 0.28098373167689117,
      "grad_norm": 0.15198677778244019,
      "learning_rate": 9.950143206160646e-06,
      "loss": 0.0045,
      "step": 43810
    },
    {
      "epoch": 0.28104786857067726,
      "grad_norm": 0.26507192850112915,
      "learning_rate": 9.950064331852452e-06,
      "loss": 0.0078,
      "step": 43820
    },
    {
      "epoch": 0.28111200546446335,
      "grad_norm": 0.34943291544914246,
      "learning_rate": 9.949985395516416e-06,
      "loss": 0.0073,
      "step": 43830
    },
    {
      "epoch": 0.28117614235824945,
      "grad_norm": 0.11543948948383331,
      "learning_rate": 9.949906397153524e-06,
      "loss": 0.0054,
      "step": 43840
    },
    {
      "epoch": 0.28124027925203554,
      "grad_norm": 0.24528025090694427,
      "learning_rate": 9.949827336764767e-06,
      "loss": 0.0066,
      "step": 43850
    },
    {
      "epoch": 0.28130441614582163,
      "grad_norm": 0.2551104724407196,
      "learning_rate": 9.949748214351135e-06,
      "loss": 0.0042,
      "step": 43860
    },
    {
      "epoch": 0.2813685530396077,
      "grad_norm": 0.2274218201637268,
      "learning_rate": 9.949669029913625e-06,
      "loss": 0.0056,
      "step": 43870
    },
    {
      "epoch": 0.2814326899333938,
      "grad_norm": 0.07005950063467026,
      "learning_rate": 9.94958978345322e-06,
      "loss": 0.0077,
      "step": 43880
    },
    {
      "epoch": 0.2814968268271799,
      "grad_norm": 0.27010342478752136,
      "learning_rate": 9.949510474970919e-06,
      "loss": 0.0047,
      "step": 43890
    },
    {
      "epoch": 0.281560963720966,
      "grad_norm": 0.3861885666847229,
      "learning_rate": 9.949431104467716e-06,
      "loss": 0.0066,
      "step": 43900
    },
    {
      "epoch": 0.2816251006147521,
      "grad_norm": 0.07936596125364304,
      "learning_rate": 9.9493516719446e-06,
      "loss": 0.0035,
      "step": 43910
    },
    {
      "epoch": 0.28168923750853825,
      "grad_norm": 0.03691532835364342,
      "learning_rate": 9.949272177402574e-06,
      "loss": 0.005,
      "step": 43920
    },
    {
      "epoch": 0.28175337440232434,
      "grad_norm": 0.19261986017227173,
      "learning_rate": 9.949192620842629e-06,
      "loss": 0.005,
      "step": 43930
    },
    {
      "epoch": 0.28181751129611043,
      "grad_norm": 0.3985759913921356,
      "learning_rate": 9.949113002265764e-06,
      "loss": 0.0046,
      "step": 43940
    },
    {
      "epoch": 0.2818816481898965,
      "grad_norm": 0.38030776381492615,
      "learning_rate": 9.949033321672977e-06,
      "loss": 0.0055,
      "step": 43950
    },
    {
      "epoch": 0.2819457850836826,
      "grad_norm": 0.31178340315818787,
      "learning_rate": 9.948953579065262e-06,
      "loss": 0.0057,
      "step": 43960
    },
    {
      "epoch": 0.2820099219774687,
      "grad_norm": 0.1481586992740631,
      "learning_rate": 9.948873774443623e-06,
      "loss": 0.0042,
      "step": 43970
    },
    {
      "epoch": 0.2820740588712548,
      "grad_norm": 0.4430583715438843,
      "learning_rate": 9.94879390780906e-06,
      "loss": 0.0067,
      "step": 43980
    },
    {
      "epoch": 0.2821381957650409,
      "grad_norm": 0.06289871037006378,
      "learning_rate": 9.948713979162571e-06,
      "loss": 0.0072,
      "step": 43990
    },
    {
      "epoch": 0.282202332658827,
      "grad_norm": 0.14244233071804047,
      "learning_rate": 9.948633988505161e-06,
      "loss": 0.0064,
      "step": 44000
    },
    {
      "epoch": 0.2822664695526131,
      "grad_norm": 0.3439823389053345,
      "learning_rate": 9.948553935837831e-06,
      "loss": 0.0069,
      "step": 44010
    },
    {
      "epoch": 0.2823306064463992,
      "grad_norm": 0.20238210260868073,
      "learning_rate": 9.94847382116158e-06,
      "loss": 0.0045,
      "step": 44020
    },
    {
      "epoch": 0.2823947433401853,
      "grad_norm": 0.4001085162162781,
      "learning_rate": 9.948393644477417e-06,
      "loss": 0.0043,
      "step": 44030
    },
    {
      "epoch": 0.28245888023397137,
      "grad_norm": 0.2548503875732422,
      "learning_rate": 9.948313405786346e-06,
      "loss": 0.0035,
      "step": 44040
    },
    {
      "epoch": 0.28252301712775746,
      "grad_norm": 0.3001789152622223,
      "learning_rate": 9.948233105089371e-06,
      "loss": 0.0039,
      "step": 44050
    },
    {
      "epoch": 0.2825871540215436,
      "grad_norm": 0.10993659496307373,
      "learning_rate": 9.948152742387498e-06,
      "loss": 0.004,
      "step": 44060
    },
    {
      "epoch": 0.2826512909153297,
      "grad_norm": 0.2728452682495117,
      "learning_rate": 9.948072317681737e-06,
      "loss": 0.0038,
      "step": 44070
    },
    {
      "epoch": 0.2827154278091158,
      "grad_norm": 0.1629365235567093,
      "learning_rate": 9.94799183097309e-06,
      "loss": 0.0037,
      "step": 44080
    },
    {
      "epoch": 0.2827795647029019,
      "grad_norm": 0.4628024995326996,
      "learning_rate": 9.947911282262571e-06,
      "loss": 0.0064,
      "step": 44090
    },
    {
      "epoch": 0.282843701596688,
      "grad_norm": 0.08697532117366791,
      "learning_rate": 9.947830671551187e-06,
      "loss": 0.0044,
      "step": 44100
    },
    {
      "epoch": 0.2829078384904741,
      "grad_norm": 0.37861886620521545,
      "learning_rate": 9.94774999883995e-06,
      "loss": 0.004,
      "step": 44110
    },
    {
      "epoch": 0.28297197538426017,
      "grad_norm": 0.37521013617515564,
      "learning_rate": 9.947669264129867e-06,
      "loss": 0.0042,
      "step": 44120
    },
    {
      "epoch": 0.28303611227804626,
      "grad_norm": 0.12726260721683502,
      "learning_rate": 9.947588467421954e-06,
      "loss": 0.0036,
      "step": 44130
    },
    {
      "epoch": 0.28310024917183235,
      "grad_norm": 0.09171196818351746,
      "learning_rate": 9.94750760871722e-06,
      "loss": 0.0073,
      "step": 44140
    },
    {
      "epoch": 0.28316438606561845,
      "grad_norm": 0.21508584916591644,
      "learning_rate": 9.94742668801668e-06,
      "loss": 0.005,
      "step": 44150
    },
    {
      "epoch": 0.28322852295940454,
      "grad_norm": 0.2806234657764435,
      "learning_rate": 9.947345705321349e-06,
      "loss": 0.0072,
      "step": 44160
    },
    {
      "epoch": 0.28329265985319063,
      "grad_norm": 0.27383875846862793,
      "learning_rate": 9.947264660632241e-06,
      "loss": 0.0051,
      "step": 44170
    },
    {
      "epoch": 0.2833567967469767,
      "grad_norm": 0.10557805001735687,
      "learning_rate": 9.94718355395037e-06,
      "loss": 0.0059,
      "step": 44180
    },
    {
      "epoch": 0.2834209336407628,
      "grad_norm": 0.2006809264421463,
      "learning_rate": 9.947102385276752e-06,
      "loss": 0.006,
      "step": 44190
    },
    {
      "epoch": 0.28348507053454897,
      "grad_norm": 0.3478815257549286,
      "learning_rate": 9.947021154612407e-06,
      "loss": 0.0078,
      "step": 44200
    },
    {
      "epoch": 0.28354920742833506,
      "grad_norm": 0.4356499910354614,
      "learning_rate": 9.946939861958352e-06,
      "loss": 0.0057,
      "step": 44210
    },
    {
      "epoch": 0.28361334432212115,
      "grad_norm": 0.1853245198726654,
      "learning_rate": 9.946858507315603e-06,
      "loss": 0.0034,
      "step": 44220
    },
    {
      "epoch": 0.28367748121590725,
      "grad_norm": 0.06792756915092468,
      "learning_rate": 9.946777090685182e-06,
      "loss": 0.0051,
      "step": 44230
    },
    {
      "epoch": 0.28374161810969334,
      "grad_norm": 0.16064020991325378,
      "learning_rate": 9.94669561206811e-06,
      "loss": 0.0047,
      "step": 44240
    },
    {
      "epoch": 0.28380575500347943,
      "grad_norm": 0.11461981385946274,
      "learning_rate": 9.946614071465405e-06,
      "loss": 0.0051,
      "step": 44250
    },
    {
      "epoch": 0.2838698918972655,
      "grad_norm": 0.4316380023956299,
      "learning_rate": 9.946532468878091e-06,
      "loss": 0.0077,
      "step": 44260
    },
    {
      "epoch": 0.2839340287910516,
      "grad_norm": 0.1925811767578125,
      "learning_rate": 9.946450804307191e-06,
      "loss": 0.0062,
      "step": 44270
    },
    {
      "epoch": 0.2839981656848377,
      "grad_norm": 0.40506237745285034,
      "learning_rate": 9.946369077753725e-06,
      "loss": 0.0064,
      "step": 44280
    },
    {
      "epoch": 0.2840623025786238,
      "grad_norm": 0.12813763320446014,
      "learning_rate": 9.946287289218722e-06,
      "loss": 0.0086,
      "step": 44290
    },
    {
      "epoch": 0.2841264394724099,
      "grad_norm": 0.22591009736061096,
      "learning_rate": 9.946205438703202e-06,
      "loss": 0.0043,
      "step": 44300
    },
    {
      "epoch": 0.284190576366196,
      "grad_norm": 0.15613387525081635,
      "learning_rate": 9.946123526208194e-06,
      "loss": 0.0053,
      "step": 44310
    },
    {
      "epoch": 0.2842547132599821,
      "grad_norm": 0.46575087308883667,
      "learning_rate": 9.946041551734724e-06,
      "loss": 0.005,
      "step": 44320
    },
    {
      "epoch": 0.2843188501537682,
      "grad_norm": 0.24979087710380554,
      "learning_rate": 9.945959515283817e-06,
      "loss": 0.0044,
      "step": 44330
    },
    {
      "epoch": 0.2843829870475543,
      "grad_norm": 0.27205783128738403,
      "learning_rate": 9.945877416856504e-06,
      "loss": 0.0065,
      "step": 44340
    },
    {
      "epoch": 0.2844471239413404,
      "grad_norm": 0.16671212017536163,
      "learning_rate": 9.94579525645381e-06,
      "loss": 0.003,
      "step": 44350
    },
    {
      "epoch": 0.2845112608351265,
      "grad_norm": 0.5268667936325073,
      "learning_rate": 9.945713034076767e-06,
      "loss": 0.007,
      "step": 44360
    },
    {
      "epoch": 0.2845753977289126,
      "grad_norm": 0.2374076396226883,
      "learning_rate": 9.945630749726408e-06,
      "loss": 0.0042,
      "step": 44370
    },
    {
      "epoch": 0.2846395346226987,
      "grad_norm": 0.2763907015323639,
      "learning_rate": 9.945548403403757e-06,
      "loss": 0.0102,
      "step": 44380
    },
    {
      "epoch": 0.2847036715164848,
      "grad_norm": 0.1680525243282318,
      "learning_rate": 9.945465995109854e-06,
      "loss": 0.0052,
      "step": 44390
    },
    {
      "epoch": 0.2847678084102709,
      "grad_norm": 0.21754974126815796,
      "learning_rate": 9.945383524845724e-06,
      "loss": 0.0058,
      "step": 44400
    },
    {
      "epoch": 0.284831945304057,
      "grad_norm": 0.39914047718048096,
      "learning_rate": 9.945300992612406e-06,
      "loss": 0.0055,
      "step": 44410
    },
    {
      "epoch": 0.28489608219784307,
      "grad_norm": 0.030866149812936783,
      "learning_rate": 9.945218398410932e-06,
      "loss": 0.004,
      "step": 44420
    },
    {
      "epoch": 0.28496021909162916,
      "grad_norm": 0.10678695142269135,
      "learning_rate": 9.945135742242337e-06,
      "loss": 0.0056,
      "step": 44430
    },
    {
      "epoch": 0.28502435598541526,
      "grad_norm": 0.22009611129760742,
      "learning_rate": 9.945053024107656e-06,
      "loss": 0.005,
      "step": 44440
    },
    {
      "epoch": 0.28508849287920135,
      "grad_norm": 0.27552640438079834,
      "learning_rate": 9.944970244007927e-06,
      "loss": 0.0046,
      "step": 44450
    },
    {
      "epoch": 0.28515262977298744,
      "grad_norm": 0.5489904284477234,
      "learning_rate": 9.944887401944187e-06,
      "loss": 0.0072,
      "step": 44460
    },
    {
      "epoch": 0.28521676666677354,
      "grad_norm": 0.2531215250492096,
      "learning_rate": 9.944804497917475e-06,
      "loss": 0.0052,
      "step": 44470
    },
    {
      "epoch": 0.2852809035605597,
      "grad_norm": 0.3310990631580353,
      "learning_rate": 9.944721531928828e-06,
      "loss": 0.0061,
      "step": 44480
    },
    {
      "epoch": 0.2853450404543458,
      "grad_norm": 0.5441588163375854,
      "learning_rate": 9.944638503979284e-06,
      "loss": 0.0047,
      "step": 44490
    },
    {
      "epoch": 0.28540917734813187,
      "grad_norm": 0.16787196695804596,
      "learning_rate": 9.944555414069888e-06,
      "loss": 0.0036,
      "step": 44500
    },
    {
      "epoch": 0.28547331424191796,
      "grad_norm": 0.20527492463588715,
      "learning_rate": 9.94447226220168e-06,
      "loss": 0.0061,
      "step": 44510
    },
    {
      "epoch": 0.28553745113570406,
      "grad_norm": 0.38600656390190125,
      "learning_rate": 9.944389048375697e-06,
      "loss": 0.0036,
      "step": 44520
    },
    {
      "epoch": 0.28560158802949015,
      "grad_norm": 0.15580099821090698,
      "learning_rate": 9.944305772592987e-06,
      "loss": 0.0047,
      "step": 44530
    },
    {
      "epoch": 0.28566572492327624,
      "grad_norm": 0.3504888117313385,
      "learning_rate": 9.944222434854595e-06,
      "loss": 0.0078,
      "step": 44540
    },
    {
      "epoch": 0.28572986181706234,
      "grad_norm": 0.16987036168575287,
      "learning_rate": 9.94413903516156e-06,
      "loss": 0.0065,
      "step": 44550
    },
    {
      "epoch": 0.28579399871084843,
      "grad_norm": 0.5898092985153198,
      "learning_rate": 9.944055573514928e-06,
      "loss": 0.0025,
      "step": 44560
    },
    {
      "epoch": 0.2858581356046345,
      "grad_norm": 0.10460730642080307,
      "learning_rate": 9.943972049915748e-06,
      "loss": 0.0031,
      "step": 44570
    },
    {
      "epoch": 0.2859222724984206,
      "grad_norm": 0.3808610439300537,
      "learning_rate": 9.943888464365065e-06,
      "loss": 0.0056,
      "step": 44580
    },
    {
      "epoch": 0.2859864093922067,
      "grad_norm": 0.13328585028648376,
      "learning_rate": 9.943804816863925e-06,
      "loss": 0.0068,
      "step": 44590
    },
    {
      "epoch": 0.2860505462859928,
      "grad_norm": 0.06399616599082947,
      "learning_rate": 9.943721107413378e-06,
      "loss": 0.0036,
      "step": 44600
    },
    {
      "epoch": 0.2861146831797789,
      "grad_norm": 0.23969586193561554,
      "learning_rate": 9.943637336014472e-06,
      "loss": 0.0043,
      "step": 44610
    },
    {
      "epoch": 0.28617882007356504,
      "grad_norm": 0.34828394651412964,
      "learning_rate": 9.943553502668257e-06,
      "loss": 0.0115,
      "step": 44620
    },
    {
      "epoch": 0.28624295696735114,
      "grad_norm": 0.15816433727741241,
      "learning_rate": 9.943469607375784e-06,
      "loss": 0.0087,
      "step": 44630
    },
    {
      "epoch": 0.28630709386113723,
      "grad_norm": 0.2197151482105255,
      "learning_rate": 9.943385650138103e-06,
      "loss": 0.0066,
      "step": 44640
    },
    {
      "epoch": 0.2863712307549233,
      "grad_norm": 0.08302151411771774,
      "learning_rate": 9.943301630956268e-06,
      "loss": 0.0051,
      "step": 44650
    },
    {
      "epoch": 0.2864353676487094,
      "grad_norm": 0.35876697301864624,
      "learning_rate": 9.94321754983133e-06,
      "loss": 0.0079,
      "step": 44660
    },
    {
      "epoch": 0.2864995045424955,
      "grad_norm": 0.17010320723056793,
      "learning_rate": 9.943133406764342e-06,
      "loss": 0.0035,
      "step": 44670
    },
    {
      "epoch": 0.2865636414362816,
      "grad_norm": 0.18060402572155,
      "learning_rate": 9.94304920175636e-06,
      "loss": 0.0054,
      "step": 44680
    },
    {
      "epoch": 0.2866277783300677,
      "grad_norm": 0.03345949575304985,
      "learning_rate": 9.942964934808442e-06,
      "loss": 0.0057,
      "step": 44690
    },
    {
      "epoch": 0.2866919152238538,
      "grad_norm": 0.23722924292087555,
      "learning_rate": 9.942880605921637e-06,
      "loss": 0.0067,
      "step": 44700
    },
    {
      "epoch": 0.2867560521176399,
      "grad_norm": 0.37032759189605713,
      "learning_rate": 9.942796215097007e-06,
      "loss": 0.0051,
      "step": 44710
    },
    {
      "epoch": 0.286820189011426,
      "grad_norm": 0.2218584567308426,
      "learning_rate": 9.942711762335608e-06,
      "loss": 0.0081,
      "step": 44720
    },
    {
      "epoch": 0.28688432590521207,
      "grad_norm": 0.08566952496767044,
      "learning_rate": 9.942627247638497e-06,
      "loss": 0.0059,
      "step": 44730
    },
    {
      "epoch": 0.28694846279899816,
      "grad_norm": 0.18911112844944,
      "learning_rate": 9.942542671006734e-06,
      "loss": 0.0052,
      "step": 44740
    },
    {
      "epoch": 0.28701259969278425,
      "grad_norm": 0.3110847771167755,
      "learning_rate": 9.94245803244138e-06,
      "loss": 0.0053,
      "step": 44750
    },
    {
      "epoch": 0.2870767365865704,
      "grad_norm": 0.2588106691837311,
      "learning_rate": 9.942373331943494e-06,
      "loss": 0.0042,
      "step": 44760
    },
    {
      "epoch": 0.2871408734803565,
      "grad_norm": 0.1012803167104721,
      "learning_rate": 9.942288569514139e-06,
      "loss": 0.0048,
      "step": 44770
    },
    {
      "epoch": 0.2872050103741426,
      "grad_norm": 0.24208807945251465,
      "learning_rate": 9.942203745154375e-06,
      "loss": 0.0065,
      "step": 44780
    },
    {
      "epoch": 0.2872691472679287,
      "grad_norm": 0.1405334770679474,
      "learning_rate": 9.942118858865266e-06,
      "loss": 0.003,
      "step": 44790
    },
    {
      "epoch": 0.2873332841617148,
      "grad_norm": 0.15158362686634064,
      "learning_rate": 9.942033910647875e-06,
      "loss": 0.0035,
      "step": 44800
    },
    {
      "epoch": 0.28739742105550087,
      "grad_norm": 0.2788327932357788,
      "learning_rate": 9.94194890050327e-06,
      "loss": 0.0048,
      "step": 44810
    },
    {
      "epoch": 0.28746155794928696,
      "grad_norm": 0.18016715347766876,
      "learning_rate": 9.94186382843251e-06,
      "loss": 0.0043,
      "step": 44820
    },
    {
      "epoch": 0.28752569484307305,
      "grad_norm": 0.1494971513748169,
      "learning_rate": 9.941778694436665e-06,
      "loss": 0.0054,
      "step": 44830
    },
    {
      "epoch": 0.28758983173685915,
      "grad_norm": 0.4434851109981537,
      "learning_rate": 9.941693498516802e-06,
      "loss": 0.0091,
      "step": 44840
    },
    {
      "epoch": 0.28765396863064524,
      "grad_norm": 0.14780807495117188,
      "learning_rate": 9.941608240673985e-06,
      "loss": 0.0031,
      "step": 44850
    },
    {
      "epoch": 0.28771810552443133,
      "grad_norm": 0.13415762782096863,
      "learning_rate": 9.941522920909287e-06,
      "loss": 0.0091,
      "step": 44860
    },
    {
      "epoch": 0.2877822424182174,
      "grad_norm": 0.3207700252532959,
      "learning_rate": 9.941437539223777e-06,
      "loss": 0.0041,
      "step": 44870
    },
    {
      "epoch": 0.2878463793120035,
      "grad_norm": 0.2459515631198883,
      "learning_rate": 9.941352095618522e-06,
      "loss": 0.0044,
      "step": 44880
    },
    {
      "epoch": 0.2879105162057896,
      "grad_norm": 0.2570980191230774,
      "learning_rate": 9.941266590094593e-06,
      "loss": 0.0057,
      "step": 44890
    },
    {
      "epoch": 0.28797465309957576,
      "grad_norm": 0.2937203347682953,
      "learning_rate": 9.941181022653061e-06,
      "loss": 0.0071,
      "step": 44900
    },
    {
      "epoch": 0.28803878999336185,
      "grad_norm": 0.2644745409488678,
      "learning_rate": 9.941095393295002e-06,
      "loss": 0.0034,
      "step": 44910
    },
    {
      "epoch": 0.28810292688714795,
      "grad_norm": 0.13715563714504242,
      "learning_rate": 9.941009702021484e-06,
      "loss": 0.0041,
      "step": 44920
    },
    {
      "epoch": 0.28816706378093404,
      "grad_norm": 0.13606588542461395,
      "learning_rate": 9.940923948833585e-06,
      "loss": 0.0048,
      "step": 44930
    },
    {
      "epoch": 0.28823120067472013,
      "grad_norm": 0.3833830952644348,
      "learning_rate": 9.940838133732376e-06,
      "loss": 0.0082,
      "step": 44940
    },
    {
      "epoch": 0.2882953375685062,
      "grad_norm": 0.2200787365436554,
      "learning_rate": 9.940752256718936e-06,
      "loss": 0.005,
      "step": 44950
    },
    {
      "epoch": 0.2883594744622923,
      "grad_norm": 0.19088926911354065,
      "learning_rate": 9.940666317794337e-06,
      "loss": 0.0073,
      "step": 44960
    },
    {
      "epoch": 0.2884236113560784,
      "grad_norm": 0.3547663986682892,
      "learning_rate": 9.94058031695966e-06,
      "loss": 0.005,
      "step": 44970
    },
    {
      "epoch": 0.2884877482498645,
      "grad_norm": 0.38969674706459045,
      "learning_rate": 9.94049425421598e-06,
      "loss": 0.0061,
      "step": 44980
    },
    {
      "epoch": 0.2885518851436506,
      "grad_norm": 0.25331956148147583,
      "learning_rate": 9.940408129564375e-06,
      "loss": 0.0054,
      "step": 44990
    },
    {
      "epoch": 0.2886160220374367,
      "grad_norm": 0.22191715240478516,
      "learning_rate": 9.940321943005927e-06,
      "loss": 0.0066,
      "step": 45000
    },
    {
      "epoch": 0.2886801589312228,
      "grad_norm": 0.264688640832901,
      "learning_rate": 9.940235694541712e-06,
      "loss": 0.0068,
      "step": 45010
    },
    {
      "epoch": 0.2887442958250089,
      "grad_norm": 0.3253937363624573,
      "learning_rate": 9.940149384172815e-06,
      "loss": 0.0056,
      "step": 45020
    },
    {
      "epoch": 0.28880843271879497,
      "grad_norm": 0.6022482514381409,
      "learning_rate": 9.940063011900314e-06,
      "loss": 0.0065,
      "step": 45030
    },
    {
      "epoch": 0.2888725696125811,
      "grad_norm": 0.21916916966438293,
      "learning_rate": 9.939976577725294e-06,
      "loss": 0.0043,
      "step": 45040
    },
    {
      "epoch": 0.2889367065063672,
      "grad_norm": 0.13018429279327393,
      "learning_rate": 9.939890081648837e-06,
      "loss": 0.0047,
      "step": 45050
    },
    {
      "epoch": 0.2890008434001533,
      "grad_norm": 0.2375756800174713,
      "learning_rate": 9.939803523672027e-06,
      "loss": 0.0067,
      "step": 45060
    },
    {
      "epoch": 0.2890649802939394,
      "grad_norm": 0.049471210688352585,
      "learning_rate": 9.939716903795947e-06,
      "loss": 0.0047,
      "step": 45070
    },
    {
      "epoch": 0.2891291171877255,
      "grad_norm": 0.09005885571241379,
      "learning_rate": 9.939630222021685e-06,
      "loss": 0.0045,
      "step": 45080
    },
    {
      "epoch": 0.2891932540815116,
      "grad_norm": 0.23219124972820282,
      "learning_rate": 9.939543478350327e-06,
      "loss": 0.0037,
      "step": 45090
    },
    {
      "epoch": 0.2892573909752977,
      "grad_norm": 0.2335500717163086,
      "learning_rate": 9.939456672782957e-06,
      "loss": 0.0038,
      "step": 45100
    },
    {
      "epoch": 0.28932152786908377,
      "grad_norm": 0.20651297271251678,
      "learning_rate": 9.939369805320664e-06,
      "loss": 0.0043,
      "step": 45110
    },
    {
      "epoch": 0.28938566476286987,
      "grad_norm": 0.30232393741607666,
      "learning_rate": 9.93928287596454e-06,
      "loss": 0.0057,
      "step": 45120
    },
    {
      "epoch": 0.28944980165665596,
      "grad_norm": 0.32386380434036255,
      "learning_rate": 9.939195884715669e-06,
      "loss": 0.0055,
      "step": 45130
    },
    {
      "epoch": 0.28951393855044205,
      "grad_norm": 0.1387457400560379,
      "learning_rate": 9.939108831575144e-06,
      "loss": 0.005,
      "step": 45140
    },
    {
      "epoch": 0.28957807544422814,
      "grad_norm": 0.15047401189804077,
      "learning_rate": 9.939021716544057e-06,
      "loss": 0.0038,
      "step": 45150
    },
    {
      "epoch": 0.28964221233801424,
      "grad_norm": 0.2504582107067108,
      "learning_rate": 9.938934539623497e-06,
      "loss": 0.006,
      "step": 45160
    },
    {
      "epoch": 0.28970634923180033,
      "grad_norm": 0.0815175250172615,
      "learning_rate": 9.938847300814558e-06,
      "loss": 0.0078,
      "step": 45170
    },
    {
      "epoch": 0.2897704861255865,
      "grad_norm": 0.11432450264692307,
      "learning_rate": 9.938760000118333e-06,
      "loss": 0.0037,
      "step": 45180
    },
    {
      "epoch": 0.2898346230193726,
      "grad_norm": 0.4249918460845947,
      "learning_rate": 9.938672637535913e-06,
      "loss": 0.0045,
      "step": 45190
    },
    {
      "epoch": 0.28989875991315867,
      "grad_norm": 0.2470531463623047,
      "learning_rate": 9.938585213068398e-06,
      "loss": 0.0034,
      "step": 45200
    },
    {
      "epoch": 0.28996289680694476,
      "grad_norm": 0.15708892047405243,
      "learning_rate": 9.938497726716879e-06,
      "loss": 0.0035,
      "step": 45210
    },
    {
      "epoch": 0.29002703370073085,
      "grad_norm": 0.2380772829055786,
      "learning_rate": 9.938410178482455e-06,
      "loss": 0.0047,
      "step": 45220
    },
    {
      "epoch": 0.29009117059451694,
      "grad_norm": 0.21689799427986145,
      "learning_rate": 9.93832256836622e-06,
      "loss": 0.0069,
      "step": 45230
    },
    {
      "epoch": 0.29015530748830304,
      "grad_norm": 0.15360818803310394,
      "learning_rate": 9.938234896369276e-06,
      "loss": 0.0048,
      "step": 45240
    },
    {
      "epoch": 0.29021944438208913,
      "grad_norm": 0.1445130854845047,
      "learning_rate": 9.93814716249272e-06,
      "loss": 0.0046,
      "step": 45250
    },
    {
      "epoch": 0.2902835812758752,
      "grad_norm": 0.21403834223747253,
      "learning_rate": 9.93805936673765e-06,
      "loss": 0.0041,
      "step": 45260
    },
    {
      "epoch": 0.2903477181696613,
      "grad_norm": 0.29812654852867126,
      "learning_rate": 9.937971509105166e-06,
      "loss": 0.0068,
      "step": 45270
    },
    {
      "epoch": 0.2904118550634474,
      "grad_norm": 0.08340287953615189,
      "learning_rate": 9.93788358959637e-06,
      "loss": 0.0063,
      "step": 45280
    },
    {
      "epoch": 0.2904759919572335,
      "grad_norm": 0.21106629073619843,
      "learning_rate": 9.937795608212367e-06,
      "loss": 0.0092,
      "step": 45290
    },
    {
      "epoch": 0.2905401288510196,
      "grad_norm": 0.06347090005874634,
      "learning_rate": 9.937707564954251e-06,
      "loss": 0.0082,
      "step": 45300
    },
    {
      "epoch": 0.2906042657448057,
      "grad_norm": 0.2607266306877136,
      "learning_rate": 9.937619459823133e-06,
      "loss": 0.0048,
      "step": 45310
    },
    {
      "epoch": 0.2906684026385918,
      "grad_norm": 0.13297231495380402,
      "learning_rate": 9.937531292820114e-06,
      "loss": 0.0063,
      "step": 45320
    },
    {
      "epoch": 0.29073253953237793,
      "grad_norm": 0.3041931092739105,
      "learning_rate": 9.9374430639463e-06,
      "loss": 0.0048,
      "step": 45330
    },
    {
      "epoch": 0.290796676426164,
      "grad_norm": 0.2948283553123474,
      "learning_rate": 9.937354773202792e-06,
      "loss": 0.005,
      "step": 45340
    },
    {
      "epoch": 0.2908608133199501,
      "grad_norm": 0.4528610110282898,
      "learning_rate": 9.937266420590702e-06,
      "loss": 0.0056,
      "step": 45350
    },
    {
      "epoch": 0.2909249502137362,
      "grad_norm": 0.2601458728313446,
      "learning_rate": 9.937178006111138e-06,
      "loss": 0.005,
      "step": 45360
    },
    {
      "epoch": 0.2909890871075223,
      "grad_norm": 0.15300409495830536,
      "learning_rate": 9.937089529765203e-06,
      "loss": 0.0036,
      "step": 45370
    },
    {
      "epoch": 0.2910532240013084,
      "grad_norm": 0.24543161690235138,
      "learning_rate": 9.937000991554007e-06,
      "loss": 0.0064,
      "step": 45380
    },
    {
      "epoch": 0.2911173608950945,
      "grad_norm": 0.18194890022277832,
      "learning_rate": 9.93691239147866e-06,
      "loss": 0.0047,
      "step": 45390
    },
    {
      "epoch": 0.2911814977888806,
      "grad_norm": 0.3864964246749878,
      "learning_rate": 9.936823729540274e-06,
      "loss": 0.008,
      "step": 45400
    },
    {
      "epoch": 0.2912456346826667,
      "grad_norm": 0.014018191024661064,
      "learning_rate": 9.936735005739958e-06,
      "loss": 0.0046,
      "step": 45410
    },
    {
      "epoch": 0.29130977157645277,
      "grad_norm": 0.29989856481552124,
      "learning_rate": 9.936646220078823e-06,
      "loss": 0.0059,
      "step": 45420
    },
    {
      "epoch": 0.29137390847023886,
      "grad_norm": 0.4319022297859192,
      "learning_rate": 9.936557372557982e-06,
      "loss": 0.0051,
      "step": 45430
    },
    {
      "epoch": 0.29143804536402496,
      "grad_norm": 0.25548070669174194,
      "learning_rate": 9.93646846317855e-06,
      "loss": 0.0041,
      "step": 45440
    },
    {
      "epoch": 0.29150218225781105,
      "grad_norm": 0.1536119133234024,
      "learning_rate": 9.936379491941641e-06,
      "loss": 0.005,
      "step": 45450
    },
    {
      "epoch": 0.29156631915159714,
      "grad_norm": 0.17052310705184937,
      "learning_rate": 9.936290458848367e-06,
      "loss": 0.005,
      "step": 45460
    },
    {
      "epoch": 0.2916304560453833,
      "grad_norm": 0.7218708992004395,
      "learning_rate": 9.93620136389985e-06,
      "loss": 0.0033,
      "step": 45470
    },
    {
      "epoch": 0.2916945929391694,
      "grad_norm": 0.2695336639881134,
      "learning_rate": 9.936112207097197e-06,
      "loss": 0.0031,
      "step": 45480
    },
    {
      "epoch": 0.2917587298329555,
      "grad_norm": 0.26749128103256226,
      "learning_rate": 9.936022988441533e-06,
      "loss": 0.0056,
      "step": 45490
    },
    {
      "epoch": 0.29182286672674157,
      "grad_norm": 0.3378068208694458,
      "learning_rate": 9.935933707933972e-06,
      "loss": 0.0059,
      "step": 45500
    },
    {
      "epoch": 0.29188700362052766,
      "grad_norm": 0.2979477345943451,
      "learning_rate": 9.935844365575635e-06,
      "loss": 0.0055,
      "step": 45510
    },
    {
      "epoch": 0.29195114051431376,
      "grad_norm": 0.22918939590454102,
      "learning_rate": 9.93575496136764e-06,
      "loss": 0.0087,
      "step": 45520
    },
    {
      "epoch": 0.29201527740809985,
      "grad_norm": 0.6446087956428528,
      "learning_rate": 9.935665495311108e-06,
      "loss": 0.0098,
      "step": 45530
    },
    {
      "epoch": 0.29207941430188594,
      "grad_norm": 0.2776643633842468,
      "learning_rate": 9.93557596740716e-06,
      "loss": 0.0055,
      "step": 45540
    },
    {
      "epoch": 0.29214355119567204,
      "grad_norm": 0.18786029517650604,
      "learning_rate": 9.935486377656917e-06,
      "loss": 0.0047,
      "step": 45550
    },
    {
      "epoch": 0.29220768808945813,
      "grad_norm": 0.45293381810188293,
      "learning_rate": 9.935396726061503e-06,
      "loss": 0.0076,
      "step": 45560
    },
    {
      "epoch": 0.2922718249832442,
      "grad_norm": 0.17026039958000183,
      "learning_rate": 9.93530701262204e-06,
      "loss": 0.0037,
      "step": 45570
    },
    {
      "epoch": 0.2923359618770303,
      "grad_norm": 0.19190262258052826,
      "learning_rate": 9.935217237339654e-06,
      "loss": 0.004,
      "step": 45580
    },
    {
      "epoch": 0.2924000987708164,
      "grad_norm": 0.08137141168117523,
      "learning_rate": 9.935127400215468e-06,
      "loss": 0.0073,
      "step": 45590
    },
    {
      "epoch": 0.2924642356646025,
      "grad_norm": 0.105504609644413,
      "learning_rate": 9.935037501250608e-06,
      "loss": 0.004,
      "step": 45600
    },
    {
      "epoch": 0.29252837255838865,
      "grad_norm": 0.18592245876789093,
      "learning_rate": 9.934947540446203e-06,
      "loss": 0.0057,
      "step": 45610
    },
    {
      "epoch": 0.29259250945217474,
      "grad_norm": 0.3530644476413727,
      "learning_rate": 9.934857517803376e-06,
      "loss": 0.0069,
      "step": 45620
    },
    {
      "epoch": 0.29265664634596084,
      "grad_norm": 0.2561638355255127,
      "learning_rate": 9.93476743332326e-06,
      "loss": 0.0043,
      "step": 45630
    },
    {
      "epoch": 0.29272078323974693,
      "grad_norm": 0.08145124465227127,
      "learning_rate": 9.934677287006979e-06,
      "loss": 0.0049,
      "step": 45640
    },
    {
      "epoch": 0.292784920133533,
      "grad_norm": 0.26823779940605164,
      "learning_rate": 9.934587078855666e-06,
      "loss": 0.0052,
      "step": 45650
    },
    {
      "epoch": 0.2928490570273191,
      "grad_norm": 0.21624739468097687,
      "learning_rate": 9.93449680887045e-06,
      "loss": 0.0056,
      "step": 45660
    },
    {
      "epoch": 0.2929131939211052,
      "grad_norm": 0.1595057100057602,
      "learning_rate": 9.934406477052463e-06,
      "loss": 0.0056,
      "step": 45670
    },
    {
      "epoch": 0.2929773308148913,
      "grad_norm": 0.09059811383485794,
      "learning_rate": 9.934316083402834e-06,
      "loss": 0.005,
      "step": 45680
    },
    {
      "epoch": 0.2930414677086774,
      "grad_norm": 0.24553117156028748,
      "learning_rate": 9.9342256279227e-06,
      "loss": 0.0051,
      "step": 45690
    },
    {
      "epoch": 0.2931056046024635,
      "grad_norm": 0.2377704530954361,
      "learning_rate": 9.934135110613193e-06,
      "loss": 0.0055,
      "step": 45700
    },
    {
      "epoch": 0.2931697414962496,
      "grad_norm": 0.16822881996631622,
      "learning_rate": 9.934044531475446e-06,
      "loss": 0.0066,
      "step": 45710
    },
    {
      "epoch": 0.2932338783900357,
      "grad_norm": 0.049672652035951614,
      "learning_rate": 9.933953890510594e-06,
      "loss": 0.0026,
      "step": 45720
    },
    {
      "epoch": 0.29329801528382177,
      "grad_norm": 0.22833536565303802,
      "learning_rate": 9.933863187719774e-06,
      "loss": 0.0029,
      "step": 45730
    },
    {
      "epoch": 0.29336215217760786,
      "grad_norm": 0.2805958688259125,
      "learning_rate": 9.933772423104122e-06,
      "loss": 0.0038,
      "step": 45740
    },
    {
      "epoch": 0.293426289071394,
      "grad_norm": 0.14065445959568024,
      "learning_rate": 9.933681596664778e-06,
      "loss": 0.0047,
      "step": 45750
    },
    {
      "epoch": 0.2934904259651801,
      "grad_norm": 0.3495698571205139,
      "learning_rate": 9.933590708402873e-06,
      "loss": 0.0055,
      "step": 45760
    },
    {
      "epoch": 0.2935545628589662,
      "grad_norm": 0.21391554176807404,
      "learning_rate": 9.933499758319554e-06,
      "loss": 0.004,
      "step": 45770
    },
    {
      "epoch": 0.2936186997527523,
      "grad_norm": 0.17593559622764587,
      "learning_rate": 9.933408746415954e-06,
      "loss": 0.0048,
      "step": 45780
    },
    {
      "epoch": 0.2936828366465384,
      "grad_norm": 0.20354965329170227,
      "learning_rate": 9.93331767269322e-06,
      "loss": 0.0051,
      "step": 45790
    },
    {
      "epoch": 0.2937469735403245,
      "grad_norm": 0.2857542335987091,
      "learning_rate": 9.933226537152487e-06,
      "loss": 0.0045,
      "step": 45800
    },
    {
      "epoch": 0.29381111043411057,
      "grad_norm": 0.2183580994606018,
      "learning_rate": 9.9331353397949e-06,
      "loss": 0.0033,
      "step": 45810
    },
    {
      "epoch": 0.29387524732789666,
      "grad_norm": 0.25809770822525024,
      "learning_rate": 9.933044080621602e-06,
      "loss": 0.0054,
      "step": 45820
    },
    {
      "epoch": 0.29393938422168275,
      "grad_norm": 0.136774942278862,
      "learning_rate": 9.932952759633736e-06,
      "loss": 0.0035,
      "step": 45830
    },
    {
      "epoch": 0.29400352111546885,
      "grad_norm": 0.2631630599498749,
      "learning_rate": 9.932861376832449e-06,
      "loss": 0.0041,
      "step": 45840
    },
    {
      "epoch": 0.29406765800925494,
      "grad_norm": 0.39657431840896606,
      "learning_rate": 9.932769932218879e-06,
      "loss": 0.0068,
      "step": 45850
    },
    {
      "epoch": 0.29413179490304103,
      "grad_norm": 0.23736067116260529,
      "learning_rate": 9.93267842579418e-06,
      "loss": 0.0037,
      "step": 45860
    },
    {
      "epoch": 0.2941959317968271,
      "grad_norm": 0.1151050254702568,
      "learning_rate": 9.932586857559492e-06,
      "loss": 0.0035,
      "step": 45870
    },
    {
      "epoch": 0.2942600686906132,
      "grad_norm": 0.1694098711013794,
      "learning_rate": 9.932495227515968e-06,
      "loss": 0.005,
      "step": 45880
    },
    {
      "epoch": 0.29432420558439937,
      "grad_norm": 0.2936653792858124,
      "learning_rate": 9.932403535664752e-06,
      "loss": 0.0045,
      "step": 45890
    },
    {
      "epoch": 0.29438834247818546,
      "grad_norm": 0.05786910280585289,
      "learning_rate": 9.932311782006995e-06,
      "loss": 0.0059,
      "step": 45900
    },
    {
      "epoch": 0.29445247937197155,
      "grad_norm": 0.43607258796691895,
      "learning_rate": 9.932219966543846e-06,
      "loss": 0.0059,
      "step": 45910
    },
    {
      "epoch": 0.29451661626575765,
      "grad_norm": 0.3279356360435486,
      "learning_rate": 9.932128089276455e-06,
      "loss": 0.0041,
      "step": 45920
    },
    {
      "epoch": 0.29458075315954374,
      "grad_norm": 0.22652366757392883,
      "learning_rate": 9.932036150205976e-06,
      "loss": 0.0049,
      "step": 45930
    },
    {
      "epoch": 0.29464489005332983,
      "grad_norm": 0.4583686888217926,
      "learning_rate": 9.93194414933356e-06,
      "loss": 0.0063,
      "step": 45940
    },
    {
      "epoch": 0.2947090269471159,
      "grad_norm": 0.36138418316841125,
      "learning_rate": 9.931852086660357e-06,
      "loss": 0.005,
      "step": 45950
    },
    {
      "epoch": 0.294773163840902,
      "grad_norm": 0.27291011810302734,
      "learning_rate": 9.931759962187524e-06,
      "loss": 0.0057,
      "step": 45960
    },
    {
      "epoch": 0.2948373007346881,
      "grad_norm": 0.14691214263439178,
      "learning_rate": 9.931667775916212e-06,
      "loss": 0.0055,
      "step": 45970
    },
    {
      "epoch": 0.2949014376284742,
      "grad_norm": 0.2556776702404022,
      "learning_rate": 9.931575527847578e-06,
      "loss": 0.0064,
      "step": 45980
    },
    {
      "epoch": 0.2949655745222603,
      "grad_norm": 0.17870935797691345,
      "learning_rate": 9.93148321798278e-06,
      "loss": 0.0034,
      "step": 45990
    },
    {
      "epoch": 0.2950297114160464,
      "grad_norm": 0.17593887448310852,
      "learning_rate": 9.931390846322973e-06,
      "loss": 0.0063,
      "step": 46000
    },
    {
      "epoch": 0.2950938483098325,
      "grad_norm": 0.1608942449092865,
      "learning_rate": 9.931298412869314e-06,
      "loss": 0.0054,
      "step": 46010
    },
    {
      "epoch": 0.2951579852036186,
      "grad_norm": 0.1781020164489746,
      "learning_rate": 9.93120591762296e-06,
      "loss": 0.0046,
      "step": 46020
    },
    {
      "epoch": 0.2952221220974047,
      "grad_norm": 0.19401803612709045,
      "learning_rate": 9.931113360585073e-06,
      "loss": 0.0043,
      "step": 46030
    },
    {
      "epoch": 0.2952862589911908,
      "grad_norm": 0.24899917840957642,
      "learning_rate": 9.931020741756811e-06,
      "loss": 0.0038,
      "step": 46040
    },
    {
      "epoch": 0.2953503958849769,
      "grad_norm": 0.31500309705734253,
      "learning_rate": 9.930928061139334e-06,
      "loss": 0.0046,
      "step": 46050
    },
    {
      "epoch": 0.295414532778763,
      "grad_norm": 0.12831856310367584,
      "learning_rate": 9.930835318733806e-06,
      "loss": 0.0032,
      "step": 46060
    },
    {
      "epoch": 0.2954786696725491,
      "grad_norm": 0.32607731223106384,
      "learning_rate": 9.930742514541387e-06,
      "loss": 0.0077,
      "step": 46070
    },
    {
      "epoch": 0.2955428065663352,
      "grad_norm": 0.42799004912376404,
      "learning_rate": 9.93064964856324e-06,
      "loss": 0.0067,
      "step": 46080
    },
    {
      "epoch": 0.2956069434601213,
      "grad_norm": 0.20071062445640564,
      "learning_rate": 9.930556720800527e-06,
      "loss": 0.0049,
      "step": 46090
    },
    {
      "epoch": 0.2956710803539074,
      "grad_norm": 0.1952655017375946,
      "learning_rate": 9.930463731254419e-06,
      "loss": 0.0063,
      "step": 46100
    },
    {
      "epoch": 0.29573521724769347,
      "grad_norm": 0.1607760787010193,
      "learning_rate": 9.930370679926073e-06,
      "loss": 0.0036,
      "step": 46110
    },
    {
      "epoch": 0.29579935414147956,
      "grad_norm": 0.3157738149166107,
      "learning_rate": 9.93027756681666e-06,
      "loss": 0.0051,
      "step": 46120
    },
    {
      "epoch": 0.29586349103526566,
      "grad_norm": 0.06405570358037949,
      "learning_rate": 9.930184391927344e-06,
      "loss": 0.0025,
      "step": 46130
    },
    {
      "epoch": 0.29592762792905175,
      "grad_norm": 0.2306731790304184,
      "learning_rate": 9.930091155259296e-06,
      "loss": 0.0053,
      "step": 46140
    },
    {
      "epoch": 0.29599176482283784,
      "grad_norm": 0.410203754901886,
      "learning_rate": 9.929997856813682e-06,
      "loss": 0.0061,
      "step": 46150
    },
    {
      "epoch": 0.29605590171662394,
      "grad_norm": 0.2868666350841522,
      "learning_rate": 9.929904496591672e-06,
      "loss": 0.0087,
      "step": 46160
    },
    {
      "epoch": 0.2961200386104101,
      "grad_norm": 0.367072731256485,
      "learning_rate": 9.929811074594434e-06,
      "loss": 0.0046,
      "step": 46170
    },
    {
      "epoch": 0.2961841755041962,
      "grad_norm": 0.19694176316261292,
      "learning_rate": 9.92971759082314e-06,
      "loss": 0.0042,
      "step": 46180
    },
    {
      "epoch": 0.29624831239798227,
      "grad_norm": 0.17494481801986694,
      "learning_rate": 9.929624045278962e-06,
      "loss": 0.0043,
      "step": 46190
    },
    {
      "epoch": 0.29631244929176837,
      "grad_norm": 0.3938837945461273,
      "learning_rate": 9.92953043796307e-06,
      "loss": 0.0029,
      "step": 46200
    },
    {
      "epoch": 0.29637658618555446,
      "grad_norm": 0.39776164293289185,
      "learning_rate": 9.929436768876642e-06,
      "loss": 0.0039,
      "step": 46210
    },
    {
      "epoch": 0.29644072307934055,
      "grad_norm": 0.2527408301830292,
      "learning_rate": 9.929343038020845e-06,
      "loss": 0.0038,
      "step": 46220
    },
    {
      "epoch": 0.29650485997312664,
      "grad_norm": 0.25211331248283386,
      "learning_rate": 9.929249245396858e-06,
      "loss": 0.005,
      "step": 46230
    },
    {
      "epoch": 0.29656899686691274,
      "grad_norm": 0.17463168501853943,
      "learning_rate": 9.929155391005857e-06,
      "loss": 0.0052,
      "step": 46240
    },
    {
      "epoch": 0.29663313376069883,
      "grad_norm": 0.26096269488334656,
      "learning_rate": 9.929061474849013e-06,
      "loss": 0.0045,
      "step": 46250
    },
    {
      "epoch": 0.2966972706544849,
      "grad_norm": 0.05219439044594765,
      "learning_rate": 9.928967496927507e-06,
      "loss": 0.0036,
      "step": 46260
    },
    {
      "epoch": 0.296761407548271,
      "grad_norm": 0.17849355936050415,
      "learning_rate": 9.928873457242515e-06,
      "loss": 0.0069,
      "step": 46270
    },
    {
      "epoch": 0.2968255444420571,
      "grad_norm": 0.2509777247905731,
      "learning_rate": 9.928779355795217e-06,
      "loss": 0.0052,
      "step": 46280
    },
    {
      "epoch": 0.2968896813358432,
      "grad_norm": 0.7019671201705933,
      "learning_rate": 9.92868519258679e-06,
      "loss": 0.0044,
      "step": 46290
    },
    {
      "epoch": 0.2969538182296293,
      "grad_norm": 0.2720184922218323,
      "learning_rate": 9.928590967618417e-06,
      "loss": 0.0103,
      "step": 46300
    },
    {
      "epoch": 0.29701795512341544,
      "grad_norm": 0.17571432888507843,
      "learning_rate": 9.928496680891276e-06,
      "loss": 0.0037,
      "step": 46310
    },
    {
      "epoch": 0.29708209201720154,
      "grad_norm": 0.28019189834594727,
      "learning_rate": 9.928402332406549e-06,
      "loss": 0.0108,
      "step": 46320
    },
    {
      "epoch": 0.29714622891098763,
      "grad_norm": 0.24343018233776093,
      "learning_rate": 9.928307922165417e-06,
      "loss": 0.0067,
      "step": 46330
    },
    {
      "epoch": 0.2972103658047737,
      "grad_norm": 0.40319544076919556,
      "learning_rate": 9.928213450169066e-06,
      "loss": 0.0045,
      "step": 46340
    },
    {
      "epoch": 0.2972745026985598,
      "grad_norm": 0.23438304662704468,
      "learning_rate": 9.92811891641868e-06,
      "loss": 0.0085,
      "step": 46350
    },
    {
      "epoch": 0.2973386395923459,
      "grad_norm": 0.23929394781589508,
      "learning_rate": 9.928024320915438e-06,
      "loss": 0.0087,
      "step": 46360
    },
    {
      "epoch": 0.297402776486132,
      "grad_norm": 0.1507410705089569,
      "learning_rate": 9.927929663660532e-06,
      "loss": 0.003,
      "step": 46370
    },
    {
      "epoch": 0.2974669133799181,
      "grad_norm": 0.38207119703292847,
      "learning_rate": 9.927834944655144e-06,
      "loss": 0.0036,
      "step": 46380
    },
    {
      "epoch": 0.2975310502737042,
      "grad_norm": 0.24152018129825592,
      "learning_rate": 9.927740163900463e-06,
      "loss": 0.0072,
      "step": 46390
    },
    {
      "epoch": 0.2975951871674903,
      "grad_norm": 0.43097245693206787,
      "learning_rate": 9.927645321397676e-06,
      "loss": 0.0073,
      "step": 46400
    },
    {
      "epoch": 0.2976593240612764,
      "grad_norm": 0.25575006008148193,
      "learning_rate": 9.927550417147971e-06,
      "loss": 0.006,
      "step": 46410
    },
    {
      "epoch": 0.29772346095506247,
      "grad_norm": 0.26496264338493347,
      "learning_rate": 9.92745545115254e-06,
      "loss": 0.0083,
      "step": 46420
    },
    {
      "epoch": 0.29778759784884856,
      "grad_norm": 0.18561311066150665,
      "learning_rate": 9.927360423412566e-06,
      "loss": 0.0056,
      "step": 46430
    },
    {
      "epoch": 0.29785173474263466,
      "grad_norm": 0.3656439781188965,
      "learning_rate": 9.927265333929248e-06,
      "loss": 0.0033,
      "step": 46440
    },
    {
      "epoch": 0.2979158716364208,
      "grad_norm": 0.17973309755325317,
      "learning_rate": 9.927170182703772e-06,
      "loss": 0.0109,
      "step": 46450
    },
    {
      "epoch": 0.2979800085302069,
      "grad_norm": 0.12009453773498535,
      "learning_rate": 9.927074969737334e-06,
      "loss": 0.0045,
      "step": 46460
    },
    {
      "epoch": 0.298044145423993,
      "grad_norm": 0.886565089225769,
      "learning_rate": 9.926979695031126e-06,
      "loss": 0.0054,
      "step": 46470
    },
    {
      "epoch": 0.2981082823177791,
      "grad_norm": 0.07676417380571365,
      "learning_rate": 9.926884358586337e-06,
      "loss": 0.0037,
      "step": 46480
    },
    {
      "epoch": 0.2981724192115652,
      "grad_norm": 0.23020076751708984,
      "learning_rate": 9.926788960404169e-06,
      "loss": 0.0042,
      "step": 46490
    },
    {
      "epoch": 0.29823655610535127,
      "grad_norm": 0.2501325309276581,
      "learning_rate": 9.926693500485814e-06,
      "loss": 0.0075,
      "step": 46500
    },
    {
      "epoch": 0.29830069299913736,
      "grad_norm": 0.2389826774597168,
      "learning_rate": 9.926597978832467e-06,
      "loss": 0.0053,
      "step": 46510
    },
    {
      "epoch": 0.29836482989292346,
      "grad_norm": 0.0756034404039383,
      "learning_rate": 9.926502395445328e-06,
      "loss": 0.0044,
      "step": 46520
    },
    {
      "epoch": 0.29842896678670955,
      "grad_norm": 0.3056093752384186,
      "learning_rate": 9.926406750325591e-06,
      "loss": 0.004,
      "step": 46530
    },
    {
      "epoch": 0.29849310368049564,
      "grad_norm": 0.15931715071201324,
      "learning_rate": 9.92631104347446e-06,
      "loss": 0.0086,
      "step": 46540
    },
    {
      "epoch": 0.29855724057428173,
      "grad_norm": 0.2678433954715729,
      "learning_rate": 9.926215274893128e-06,
      "loss": 0.0063,
      "step": 46550
    },
    {
      "epoch": 0.29862137746806783,
      "grad_norm": 0.0851227343082428,
      "learning_rate": 9.926119444582798e-06,
      "loss": 0.0084,
      "step": 46560
    },
    {
      "epoch": 0.2986855143618539,
      "grad_norm": 0.17011059820652008,
      "learning_rate": 9.92602355254467e-06,
      "loss": 0.006,
      "step": 46570
    },
    {
      "epoch": 0.29874965125564,
      "grad_norm": 0.3575952351093292,
      "learning_rate": 9.925927598779948e-06,
      "loss": 0.0053,
      "step": 46580
    },
    {
      "epoch": 0.29881378814942616,
      "grad_norm": 0.1385606825351715,
      "learning_rate": 9.925831583289834e-06,
      "loss": 0.0056,
      "step": 46590
    },
    {
      "epoch": 0.29887792504321226,
      "grad_norm": 0.2562883198261261,
      "learning_rate": 9.925735506075526e-06,
      "loss": 0.0045,
      "step": 46600
    },
    {
      "epoch": 0.29894206193699835,
      "grad_norm": 0.1414983868598938,
      "learning_rate": 9.925639367138235e-06,
      "loss": 0.0042,
      "step": 46610
    },
    {
      "epoch": 0.29900619883078444,
      "grad_norm": 0.5601732730865479,
      "learning_rate": 9.925543166479162e-06,
      "loss": 0.0057,
      "step": 46620
    },
    {
      "epoch": 0.29907033572457054,
      "grad_norm": 0.2273009717464447,
      "learning_rate": 9.925446904099511e-06,
      "loss": 0.006,
      "step": 46630
    },
    {
      "epoch": 0.29913447261835663,
      "grad_norm": 0.3177121579647064,
      "learning_rate": 9.925350580000493e-06,
      "loss": 0.0045,
      "step": 46640
    },
    {
      "epoch": 0.2991986095121427,
      "grad_norm": 0.8014094233512878,
      "learning_rate": 9.92525419418331e-06,
      "loss": 0.0049,
      "step": 46650
    },
    {
      "epoch": 0.2992627464059288,
      "grad_norm": 0.26341408491134644,
      "learning_rate": 9.92515774664917e-06,
      "loss": 0.0076,
      "step": 46660
    },
    {
      "epoch": 0.2993268832997149,
      "grad_norm": 0.43867725133895874,
      "learning_rate": 9.925061237399287e-06,
      "loss": 0.0052,
      "step": 46670
    },
    {
      "epoch": 0.299391020193501,
      "grad_norm": 0.1924295425415039,
      "learning_rate": 9.924964666434866e-06,
      "loss": 0.0054,
      "step": 46680
    },
    {
      "epoch": 0.2994551570872871,
      "grad_norm": 0.5385870337486267,
      "learning_rate": 9.924868033757119e-06,
      "loss": 0.0079,
      "step": 46690
    },
    {
      "epoch": 0.2995192939810732,
      "grad_norm": 0.28352195024490356,
      "learning_rate": 9.924771339367253e-06,
      "loss": 0.005,
      "step": 46700
    },
    {
      "epoch": 0.2995834308748593,
      "grad_norm": 0.2408374845981598,
      "learning_rate": 9.924674583266483e-06,
      "loss": 0.0045,
      "step": 46710
    },
    {
      "epoch": 0.2996475677686454,
      "grad_norm": 0.12736321985721588,
      "learning_rate": 9.924577765456023e-06,
      "loss": 0.0051,
      "step": 46720
    },
    {
      "epoch": 0.29971170466243147,
      "grad_norm": 0.12144535034894943,
      "learning_rate": 9.924480885937082e-06,
      "loss": 0.0059,
      "step": 46730
    },
    {
      "epoch": 0.2997758415562176,
      "grad_norm": 0.09012685716152191,
      "learning_rate": 9.924383944710875e-06,
      "loss": 0.0039,
      "step": 46740
    },
    {
      "epoch": 0.2998399784500037,
      "grad_norm": 0.14406102895736694,
      "learning_rate": 9.92428694177862e-06,
      "loss": 0.0042,
      "step": 46750
    },
    {
      "epoch": 0.2999041153437898,
      "grad_norm": 0.05325210839509964,
      "learning_rate": 9.92418987714153e-06,
      "loss": 0.0039,
      "step": 46760
    },
    {
      "epoch": 0.2999682522375759,
      "grad_norm": 0.19741320610046387,
      "learning_rate": 9.924092750800823e-06,
      "loss": 0.0049,
      "step": 46770
    },
    {
      "epoch": 0.300032389131362,
      "grad_norm": 0.15774212777614594,
      "learning_rate": 9.92399556275771e-06,
      "loss": 0.005,
      "step": 46780
    },
    {
      "epoch": 0.3000965260251481,
      "grad_norm": 0.273052453994751,
      "learning_rate": 9.923898313013419e-06,
      "loss": 0.0058,
      "step": 46790
    },
    {
      "epoch": 0.3001606629189342,
      "grad_norm": 0.15874101221561432,
      "learning_rate": 9.92380100156916e-06,
      "loss": 0.0044,
      "step": 46800
    },
    {
      "epoch": 0.30022479981272027,
      "grad_norm": 0.11979794502258301,
      "learning_rate": 9.923703628426155e-06,
      "loss": 0.0041,
      "step": 46810
    },
    {
      "epoch": 0.30028893670650636,
      "grad_norm": 0.1807662546634674,
      "learning_rate": 9.923606193585627e-06,
      "loss": 0.005,
      "step": 46820
    },
    {
      "epoch": 0.30035307360029245,
      "grad_norm": 0.3019179701805115,
      "learning_rate": 9.923508697048792e-06,
      "loss": 0.0092,
      "step": 46830
    },
    {
      "epoch": 0.30041721049407855,
      "grad_norm": 0.2532137930393219,
      "learning_rate": 9.923411138816876e-06,
      "loss": 0.0072,
      "step": 46840
    },
    {
      "epoch": 0.30048134738786464,
      "grad_norm": 0.2682390511035919,
      "learning_rate": 9.923313518891099e-06,
      "loss": 0.0069,
      "step": 46850
    },
    {
      "epoch": 0.30054548428165073,
      "grad_norm": 0.2450132519006729,
      "learning_rate": 9.923215837272684e-06,
      "loss": 0.0052,
      "step": 46860
    },
    {
      "epoch": 0.3006096211754368,
      "grad_norm": 0.24705521762371063,
      "learning_rate": 9.923118093962858e-06,
      "loss": 0.0047,
      "step": 46870
    },
    {
      "epoch": 0.300673758069223,
      "grad_norm": 0.08048294484615326,
      "learning_rate": 9.923020288962843e-06,
      "loss": 0.0068,
      "step": 46880
    },
    {
      "epoch": 0.30073789496300907,
      "grad_norm": 0.13034740090370178,
      "learning_rate": 9.922922422273866e-06,
      "loss": 0.0052,
      "step": 46890
    },
    {
      "epoch": 0.30080203185679516,
      "grad_norm": 0.20540949702262878,
      "learning_rate": 9.922824493897153e-06,
      "loss": 0.0056,
      "step": 46900
    },
    {
      "epoch": 0.30086616875058125,
      "grad_norm": 0.28085941076278687,
      "learning_rate": 9.922726503833928e-06,
      "loss": 0.0048,
      "step": 46910
    },
    {
      "epoch": 0.30093030564436735,
      "grad_norm": 0.25790196657180786,
      "learning_rate": 9.922628452085423e-06,
      "loss": 0.0042,
      "step": 46920
    },
    {
      "epoch": 0.30099444253815344,
      "grad_norm": 0.21742090582847595,
      "learning_rate": 9.922530338652867e-06,
      "loss": 0.007,
      "step": 46930
    },
    {
      "epoch": 0.30105857943193953,
      "grad_norm": 0.19471445679664612,
      "learning_rate": 9.922432163537486e-06,
      "loss": 0.0051,
      "step": 46940
    },
    {
      "epoch": 0.3011227163257256,
      "grad_norm": 0.109901562333107,
      "learning_rate": 9.922333926740513e-06,
      "loss": 0.0036,
      "step": 46950
    },
    {
      "epoch": 0.3011868532195117,
      "grad_norm": 0.24060367047786713,
      "learning_rate": 9.922235628263177e-06,
      "loss": 0.0051,
      "step": 46960
    },
    {
      "epoch": 0.3012509901132978,
      "grad_norm": 0.09304339438676834,
      "learning_rate": 9.922137268106711e-06,
      "loss": 0.0045,
      "step": 46970
    },
    {
      "epoch": 0.3013151270070839,
      "grad_norm": 0.3272383511066437,
      "learning_rate": 9.922038846272347e-06,
      "loss": 0.0069,
      "step": 46980
    },
    {
      "epoch": 0.30137926390087,
      "grad_norm": 0.2325884848833084,
      "learning_rate": 9.92194036276132e-06,
      "loss": 0.0043,
      "step": 46990
    },
    {
      "epoch": 0.3014434007946561,
      "grad_norm": 0.17553608119487762,
      "learning_rate": 9.92184181757486e-06,
      "loss": 0.0059,
      "step": 47000
    },
    {
      "epoch": 0.3015075376884422,
      "grad_norm": 0.15212492644786835,
      "learning_rate": 9.921743210714207e-06,
      "loss": 0.0052,
      "step": 47010
    },
    {
      "epoch": 0.30157167458222833,
      "grad_norm": 0.5152463316917419,
      "learning_rate": 9.921644542180593e-06,
      "loss": 0.0074,
      "step": 47020
    },
    {
      "epoch": 0.3016358114760144,
      "grad_norm": 0.1545516699552536,
      "learning_rate": 9.921545811975254e-06,
      "loss": 0.0046,
      "step": 47030
    },
    {
      "epoch": 0.3016999483698005,
      "grad_norm": 0.25146082043647766,
      "learning_rate": 9.921447020099431e-06,
      "loss": 0.0056,
      "step": 47040
    },
    {
      "epoch": 0.3017640852635866,
      "grad_norm": 0.15383067727088928,
      "learning_rate": 9.921348166554359e-06,
      "loss": 0.0065,
      "step": 47050
    },
    {
      "epoch": 0.3018282221573727,
      "grad_norm": 0.361632764339447,
      "learning_rate": 9.921249251341276e-06,
      "loss": 0.005,
      "step": 47060
    },
    {
      "epoch": 0.3018923590511588,
      "grad_norm": 0.09196867793798447,
      "learning_rate": 9.921150274461424e-06,
      "loss": 0.0088,
      "step": 47070
    },
    {
      "epoch": 0.3019564959449449,
      "grad_norm": 0.29238617420196533,
      "learning_rate": 9.921051235916042e-06,
      "loss": 0.0038,
      "step": 47080
    },
    {
      "epoch": 0.302020632838731,
      "grad_norm": 0.2663821578025818,
      "learning_rate": 9.92095213570637e-06,
      "loss": 0.0089,
      "step": 47090
    },
    {
      "epoch": 0.3020847697325171,
      "grad_norm": 0.46504130959510803,
      "learning_rate": 9.920852973833651e-06,
      "loss": 0.0062,
      "step": 47100
    },
    {
      "epoch": 0.30214890662630317,
      "grad_norm": 0.2274145931005478,
      "learning_rate": 9.920753750299128e-06,
      "loss": 0.0058,
      "step": 47110
    },
    {
      "epoch": 0.30221304352008926,
      "grad_norm": 0.10128103941679001,
      "learning_rate": 9.920654465104042e-06,
      "loss": 0.0057,
      "step": 47120
    },
    {
      "epoch": 0.30227718041387536,
      "grad_norm": 0.1892748475074768,
      "learning_rate": 9.920555118249641e-06,
      "loss": 0.0031,
      "step": 47130
    },
    {
      "epoch": 0.30234131730766145,
      "grad_norm": 0.30800819396972656,
      "learning_rate": 9.920455709737166e-06,
      "loss": 0.0045,
      "step": 47140
    },
    {
      "epoch": 0.30240545420144754,
      "grad_norm": 0.3005007803440094,
      "learning_rate": 9.920356239567867e-06,
      "loss": 0.0048,
      "step": 47150
    },
    {
      "epoch": 0.3024695910952337,
      "grad_norm": 0.30953630805015564,
      "learning_rate": 9.920256707742985e-06,
      "loss": 0.0048,
      "step": 47160
    },
    {
      "epoch": 0.3025337279890198,
      "grad_norm": 0.26445797085762024,
      "learning_rate": 9.920157114263772e-06,
      "loss": 0.0074,
      "step": 47170
    },
    {
      "epoch": 0.3025978648828059,
      "grad_norm": 0.2116151750087738,
      "learning_rate": 9.920057459131472e-06,
      "loss": 0.0055,
      "step": 47180
    },
    {
      "epoch": 0.30266200177659197,
      "grad_norm": 0.14083783328533173,
      "learning_rate": 9.919957742347336e-06,
      "loss": 0.0051,
      "step": 47190
    },
    {
      "epoch": 0.30272613867037806,
      "grad_norm": 0.456478089094162,
      "learning_rate": 9.919857963912615e-06,
      "loss": 0.0069,
      "step": 47200
    },
    {
      "epoch": 0.30279027556416416,
      "grad_norm": 0.10006704926490784,
      "learning_rate": 9.919758123828558e-06,
      "loss": 0.0037,
      "step": 47210
    },
    {
      "epoch": 0.30285441245795025,
      "grad_norm": 0.05251394584774971,
      "learning_rate": 9.919658222096414e-06,
      "loss": 0.0026,
      "step": 47220
    },
    {
      "epoch": 0.30291854935173634,
      "grad_norm": 0.17398954927921295,
      "learning_rate": 9.919558258717436e-06,
      "loss": 0.0052,
      "step": 47230
    },
    {
      "epoch": 0.30298268624552244,
      "grad_norm": 0.07287753373384476,
      "learning_rate": 9.919458233692877e-06,
      "loss": 0.0023,
      "step": 47240
    },
    {
      "epoch": 0.30304682313930853,
      "grad_norm": 0.17693035304546356,
      "learning_rate": 9.919358147023992e-06,
      "loss": 0.0065,
      "step": 47250
    },
    {
      "epoch": 0.3031109600330946,
      "grad_norm": 0.06258239597082138,
      "learning_rate": 9.919257998712032e-06,
      "loss": 0.0048,
      "step": 47260
    },
    {
      "epoch": 0.3031750969268807,
      "grad_norm": 0.1813821941614151,
      "learning_rate": 9.919157788758253e-06,
      "loss": 0.0064,
      "step": 47270
    },
    {
      "epoch": 0.3032392338206668,
      "grad_norm": 0.12876521050930023,
      "learning_rate": 9.919057517163913e-06,
      "loss": 0.0058,
      "step": 47280
    },
    {
      "epoch": 0.3033033707144529,
      "grad_norm": 0.28710272908210754,
      "learning_rate": 9.918957183930267e-06,
      "loss": 0.0034,
      "step": 47290
    },
    {
      "epoch": 0.30336750760823905,
      "grad_norm": 0.22267496585845947,
      "learning_rate": 9.91885678905857e-06,
      "loss": 0.0053,
      "step": 47300
    },
    {
      "epoch": 0.30343164450202514,
      "grad_norm": 0.24553196132183075,
      "learning_rate": 9.918756332550084e-06,
      "loss": 0.0052,
      "step": 47310
    },
    {
      "epoch": 0.30349578139581124,
      "grad_norm": 0.13842783868312836,
      "learning_rate": 9.918655814406064e-06,
      "loss": 0.0057,
      "step": 47320
    },
    {
      "epoch": 0.30355991828959733,
      "grad_norm": 0.25684648752212524,
      "learning_rate": 9.918555234627773e-06,
      "loss": 0.0039,
      "step": 47330
    },
    {
      "epoch": 0.3036240551833834,
      "grad_norm": 0.11633238941431046,
      "learning_rate": 9.91845459321647e-06,
      "loss": 0.0045,
      "step": 47340
    },
    {
      "epoch": 0.3036881920771695,
      "grad_norm": 0.18775558471679688,
      "learning_rate": 9.918353890173414e-06,
      "loss": 0.0034,
      "step": 47350
    },
    {
      "epoch": 0.3037523289709556,
      "grad_norm": 0.14034156501293182,
      "learning_rate": 9.918253125499869e-06,
      "loss": 0.0046,
      "step": 47360
    },
    {
      "epoch": 0.3038164658647417,
      "grad_norm": 0.05788616091012955,
      "learning_rate": 9.918152299197097e-06,
      "loss": 0.0029,
      "step": 47370
    },
    {
      "epoch": 0.3038806027585278,
      "grad_norm": 0.1960693895816803,
      "learning_rate": 9.918051411266363e-06,
      "loss": 0.004,
      "step": 47380
    },
    {
      "epoch": 0.3039447396523139,
      "grad_norm": 0.30320316553115845,
      "learning_rate": 9.917950461708929e-06,
      "loss": 0.0038,
      "step": 47390
    },
    {
      "epoch": 0.3040088765461,
      "grad_norm": 0.29490649700164795,
      "learning_rate": 9.917849450526061e-06,
      "loss": 0.0048,
      "step": 47400
    },
    {
      "epoch": 0.3040730134398861,
      "grad_norm": 0.20866785943508148,
      "learning_rate": 9.917748377719025e-06,
      "loss": 0.0056,
      "step": 47410
    },
    {
      "epoch": 0.30413715033367217,
      "grad_norm": 0.12553279101848602,
      "learning_rate": 9.917647243289087e-06,
      "loss": 0.0062,
      "step": 47420
    },
    {
      "epoch": 0.30420128722745826,
      "grad_norm": 0.48011791706085205,
      "learning_rate": 9.917546047237513e-06,
      "loss": 0.0049,
      "step": 47430
    },
    {
      "epoch": 0.3042654241212444,
      "grad_norm": 0.16010598838329315,
      "learning_rate": 9.917444789565576e-06,
      "loss": 0.0045,
      "step": 47440
    },
    {
      "epoch": 0.3043295610150305,
      "grad_norm": 0.05319499596953392,
      "learning_rate": 9.917343470274539e-06,
      "loss": 0.005,
      "step": 47450
    },
    {
      "epoch": 0.3043936979088166,
      "grad_norm": 0.3981022834777832,
      "learning_rate": 9.917242089365674e-06,
      "loss": 0.0045,
      "step": 47460
    },
    {
      "epoch": 0.3044578348026027,
      "grad_norm": 0.4476795196533203,
      "learning_rate": 9.917140646840252e-06,
      "loss": 0.0075,
      "step": 47470
    },
    {
      "epoch": 0.3045219716963888,
      "grad_norm": 0.2710070013999939,
      "learning_rate": 9.917039142699542e-06,
      "loss": 0.0056,
      "step": 47480
    },
    {
      "epoch": 0.3045861085901749,
      "grad_norm": 0.14953048527240753,
      "learning_rate": 9.91693757694482e-06,
      "loss": 0.0052,
      "step": 47490
    },
    {
      "epoch": 0.30465024548396097,
      "grad_norm": 0.221146821975708,
      "learning_rate": 9.916835949577355e-06,
      "loss": 0.0061,
      "step": 47500
    },
    {
      "epoch": 0.30471438237774706,
      "grad_norm": 0.22094900906085968,
      "learning_rate": 9.91673426059842e-06,
      "loss": 0.0035,
      "step": 47510
    },
    {
      "epoch": 0.30477851927153315,
      "grad_norm": 0.16071777045726776,
      "learning_rate": 9.916632510009292e-06,
      "loss": 0.0048,
      "step": 47520
    },
    {
      "epoch": 0.30484265616531925,
      "grad_norm": 0.07116349786520004,
      "learning_rate": 9.916530697811244e-06,
      "loss": 0.0051,
      "step": 47530
    },
    {
      "epoch": 0.30490679305910534,
      "grad_norm": 0.17552363872528076,
      "learning_rate": 9.916428824005554e-06,
      "loss": 0.0053,
      "step": 47540
    },
    {
      "epoch": 0.30497092995289143,
      "grad_norm": 0.15659548342227936,
      "learning_rate": 9.916326888593498e-06,
      "loss": 0.0051,
      "step": 47550
    },
    {
      "epoch": 0.3050350668466775,
      "grad_norm": 0.15439796447753906,
      "learning_rate": 9.916224891576349e-06,
      "loss": 0.0034,
      "step": 47560
    },
    {
      "epoch": 0.3050992037404636,
      "grad_norm": 0.2635513246059418,
      "learning_rate": 9.91612283295539e-06,
      "loss": 0.0057,
      "step": 47570
    },
    {
      "epoch": 0.30516334063424977,
      "grad_norm": 0.42883941531181335,
      "learning_rate": 9.9160207127319e-06,
      "loss": 0.0099,
      "step": 47580
    },
    {
      "epoch": 0.30522747752803586,
      "grad_norm": 0.19080328941345215,
      "learning_rate": 9.915918530907155e-06,
      "loss": 0.0046,
      "step": 47590
    },
    {
      "epoch": 0.30529161442182196,
      "grad_norm": 0.12923285365104675,
      "learning_rate": 9.915816287482438e-06,
      "loss": 0.0041,
      "step": 47600
    },
    {
      "epoch": 0.30535575131560805,
      "grad_norm": 0.3636232018470764,
      "learning_rate": 9.915713982459029e-06,
      "loss": 0.004,
      "step": 47610
    },
    {
      "epoch": 0.30541988820939414,
      "grad_norm": 0.09407957643270493,
      "learning_rate": 9.915611615838212e-06,
      "loss": 0.0054,
      "step": 47620
    },
    {
      "epoch": 0.30548402510318023,
      "grad_norm": 0.25751787424087524,
      "learning_rate": 9.915509187621269e-06,
      "loss": 0.0038,
      "step": 47630
    },
    {
      "epoch": 0.3055481619969663,
      "grad_norm": 0.17965631186962128,
      "learning_rate": 9.91540669780948e-06,
      "loss": 0.0037,
      "step": 47640
    },
    {
      "epoch": 0.3056122988907524,
      "grad_norm": 0.24733319878578186,
      "learning_rate": 9.915304146404135e-06,
      "loss": 0.0068,
      "step": 47650
    },
    {
      "epoch": 0.3056764357845385,
      "grad_norm": 0.030508002266287804,
      "learning_rate": 9.915201533406514e-06,
      "loss": 0.0046,
      "step": 47660
    },
    {
      "epoch": 0.3057405726783246,
      "grad_norm": 0.1589413732290268,
      "learning_rate": 9.915098858817907e-06,
      "loss": 0.003,
      "step": 47670
    },
    {
      "epoch": 0.3058047095721107,
      "grad_norm": 0.4408036768436432,
      "learning_rate": 9.914996122639596e-06,
      "loss": 0.0061,
      "step": 47680
    },
    {
      "epoch": 0.3058688464658968,
      "grad_norm": 0.18685586750507355,
      "learning_rate": 9.914893324872871e-06,
      "loss": 0.0028,
      "step": 47690
    },
    {
      "epoch": 0.3059329833596829,
      "grad_norm": 0.282780259847641,
      "learning_rate": 9.914790465519021e-06,
      "loss": 0.0043,
      "step": 47700
    },
    {
      "epoch": 0.305997120253469,
      "grad_norm": 0.27768674492836,
      "learning_rate": 9.914687544579335e-06,
      "loss": 0.0053,
      "step": 47710
    },
    {
      "epoch": 0.30606125714725513,
      "grad_norm": 0.40035441517829895,
      "learning_rate": 9.9145845620551e-06,
      "loss": 0.0049,
      "step": 47720
    },
    {
      "epoch": 0.3061253940410412,
      "grad_norm": 0.014787815511226654,
      "learning_rate": 9.914481517947609e-06,
      "loss": 0.0048,
      "step": 47730
    },
    {
      "epoch": 0.3061895309348273,
      "grad_norm": 0.5116310715675354,
      "learning_rate": 9.914378412258151e-06,
      "loss": 0.0061,
      "step": 47740
    },
    {
      "epoch": 0.3062536678286134,
      "grad_norm": 0.18756724894046783,
      "learning_rate": 9.914275244988021e-06,
      "loss": 0.003,
      "step": 47750
    },
    {
      "epoch": 0.3063178047223995,
      "grad_norm": 0.2650206387042999,
      "learning_rate": 9.91417201613851e-06,
      "loss": 0.0059,
      "step": 47760
    },
    {
      "epoch": 0.3063819416161856,
      "grad_norm": 0.24275191128253937,
      "learning_rate": 9.914068725710912e-06,
      "loss": 0.0057,
      "step": 47770
    },
    {
      "epoch": 0.3064460785099717,
      "grad_norm": 0.36543434858322144,
      "learning_rate": 9.91396537370652e-06,
      "loss": 0.0052,
      "step": 47780
    },
    {
      "epoch": 0.3065102154037578,
      "grad_norm": 0.18792115151882172,
      "learning_rate": 9.913861960126628e-06,
      "loss": 0.0038,
      "step": 47790
    },
    {
      "epoch": 0.3065743522975439,
      "grad_norm": 0.14406542479991913,
      "learning_rate": 9.913758484972536e-06,
      "loss": 0.0039,
      "step": 47800
    },
    {
      "epoch": 0.30663848919132997,
      "grad_norm": 0.12238479405641556,
      "learning_rate": 9.913654948245536e-06,
      "loss": 0.0034,
      "step": 47810
    },
    {
      "epoch": 0.30670262608511606,
      "grad_norm": 0.2577309012413025,
      "learning_rate": 9.913551349946931e-06,
      "loss": 0.0028,
      "step": 47820
    },
    {
      "epoch": 0.30676676297890215,
      "grad_norm": 0.18023546040058136,
      "learning_rate": 9.913447690078012e-06,
      "loss": 0.0056,
      "step": 47830
    },
    {
      "epoch": 0.30683089987268825,
      "grad_norm": 0.105728380382061,
      "learning_rate": 9.913343968640085e-06,
      "loss": 0.0029,
      "step": 47840
    },
    {
      "epoch": 0.30689503676647434,
      "grad_norm": 0.0686994418501854,
      "learning_rate": 9.913240185634448e-06,
      "loss": 0.0042,
      "step": 47850
    },
    {
      "epoch": 0.3069591736602605,
      "grad_norm": 0.20035584270954132,
      "learning_rate": 9.913136341062397e-06,
      "loss": 0.0045,
      "step": 47860
    },
    {
      "epoch": 0.3070233105540466,
      "grad_norm": 0.23853257298469543,
      "learning_rate": 9.913032434925236e-06,
      "loss": 0.0051,
      "step": 47870
    },
    {
      "epoch": 0.3070874474478327,
      "grad_norm": 0.18853001296520233,
      "learning_rate": 9.91292846722427e-06,
      "loss": 0.0103,
      "step": 47880
    },
    {
      "epoch": 0.30715158434161877,
      "grad_norm": 0.3605303466320038,
      "learning_rate": 9.912824437960796e-06,
      "loss": 0.0051,
      "step": 47890
    },
    {
      "epoch": 0.30721572123540486,
      "grad_norm": 0.06613648682832718,
      "learning_rate": 9.912720347136122e-06,
      "loss": 0.0043,
      "step": 47900
    },
    {
      "epoch": 0.30727985812919095,
      "grad_norm": 0.1732809990644455,
      "learning_rate": 9.912616194751553e-06,
      "loss": 0.0051,
      "step": 47910
    },
    {
      "epoch": 0.30734399502297705,
      "grad_norm": 0.13078118860721588,
      "learning_rate": 9.912511980808388e-06,
      "loss": 0.0061,
      "step": 47920
    },
    {
      "epoch": 0.30740813191676314,
      "grad_norm": 0.25918957591056824,
      "learning_rate": 9.912407705307941e-06,
      "loss": 0.0038,
      "step": 47930
    },
    {
      "epoch": 0.30747226881054923,
      "grad_norm": 0.36120375990867615,
      "learning_rate": 9.91230336825151e-06,
      "loss": 0.005,
      "step": 47940
    },
    {
      "epoch": 0.3075364057043353,
      "grad_norm": 0.0602896548807621,
      "learning_rate": 9.912198969640412e-06,
      "loss": 0.0026,
      "step": 47950
    },
    {
      "epoch": 0.3076005425981214,
      "grad_norm": 0.12653125822544098,
      "learning_rate": 9.912094509475947e-06,
      "loss": 0.0051,
      "step": 47960
    },
    {
      "epoch": 0.3076646794919075,
      "grad_norm": 0.5669960975646973,
      "learning_rate": 9.911989987759429e-06,
      "loss": 0.0128,
      "step": 47970
    },
    {
      "epoch": 0.3077288163856936,
      "grad_norm": 0.168401300907135,
      "learning_rate": 9.911885404492164e-06,
      "loss": 0.0037,
      "step": 47980
    },
    {
      "epoch": 0.3077929532794797,
      "grad_norm": 0.2946607768535614,
      "learning_rate": 9.911780759675465e-06,
      "loss": 0.006,
      "step": 47990
    },
    {
      "epoch": 0.30785709017326585,
      "grad_norm": 0.2680601179599762,
      "learning_rate": 9.911676053310644e-06,
      "loss": 0.0038,
      "step": 48000
    },
    {
      "epoch": 0.30792122706705194,
      "grad_norm": 0.22053630650043488,
      "learning_rate": 9.911571285399011e-06,
      "loss": 0.005,
      "step": 48010
    },
    {
      "epoch": 0.30798536396083803,
      "grad_norm": 0.3797345757484436,
      "learning_rate": 9.911466455941878e-06,
      "loss": 0.0051,
      "step": 48020
    },
    {
      "epoch": 0.3080495008546241,
      "grad_norm": 0.09887450188398361,
      "learning_rate": 9.911361564940562e-06,
      "loss": 0.0037,
      "step": 48030
    },
    {
      "epoch": 0.3081136377484102,
      "grad_norm": 0.054559383541345596,
      "learning_rate": 9.911256612396377e-06,
      "loss": 0.0046,
      "step": 48040
    },
    {
      "epoch": 0.3081777746421963,
      "grad_norm": 0.21160346269607544,
      "learning_rate": 9.911151598310633e-06,
      "loss": 0.0087,
      "step": 48050
    },
    {
      "epoch": 0.3082419115359824,
      "grad_norm": 0.23671574890613556,
      "learning_rate": 9.911046522684651e-06,
      "loss": 0.0035,
      "step": 48060
    },
    {
      "epoch": 0.3083060484297685,
      "grad_norm": 0.3388223350048065,
      "learning_rate": 9.910941385519747e-06,
      "loss": 0.0051,
      "step": 48070
    },
    {
      "epoch": 0.3083701853235546,
      "grad_norm": 0.1838870793581009,
      "learning_rate": 9.910836186817237e-06,
      "loss": 0.004,
      "step": 48080
    },
    {
      "epoch": 0.3084343222173407,
      "grad_norm": 0.19456282258033752,
      "learning_rate": 9.91073092657844e-06,
      "loss": 0.0057,
      "step": 48090
    },
    {
      "epoch": 0.3084984591111268,
      "grad_norm": 0.36534351110458374,
      "learning_rate": 9.910625604804675e-06,
      "loss": 0.0047,
      "step": 48100
    },
    {
      "epoch": 0.30856259600491287,
      "grad_norm": 0.20905707776546478,
      "learning_rate": 9.910520221497261e-06,
      "loss": 0.0053,
      "step": 48110
    },
    {
      "epoch": 0.30862673289869896,
      "grad_norm": 0.1807992160320282,
      "learning_rate": 9.91041477665752e-06,
      "loss": 0.0071,
      "step": 48120
    },
    {
      "epoch": 0.30869086979248506,
      "grad_norm": 0.3418310880661011,
      "learning_rate": 9.91030927028677e-06,
      "loss": 0.0055,
      "step": 48130
    },
    {
      "epoch": 0.30875500668627115,
      "grad_norm": 0.18091419339179993,
      "learning_rate": 9.910203702386339e-06,
      "loss": 0.0054,
      "step": 48140
    },
    {
      "epoch": 0.3088191435800573,
      "grad_norm": 0.1711590737104416,
      "learning_rate": 9.910098072957543e-06,
      "loss": 0.0046,
      "step": 48150
    },
    {
      "epoch": 0.3088832804738434,
      "grad_norm": 0.3131524324417114,
      "learning_rate": 9.909992382001712e-06,
      "loss": 0.005,
      "step": 48160
    },
    {
      "epoch": 0.3089474173676295,
      "grad_norm": 0.3353725075721741,
      "learning_rate": 9.909886629520165e-06,
      "loss": 0.0044,
      "step": 48170
    },
    {
      "epoch": 0.3090115542614156,
      "grad_norm": 0.09161921590566635,
      "learning_rate": 9.909780815514231e-06,
      "loss": 0.005,
      "step": 48180
    },
    {
      "epoch": 0.30907569115520167,
      "grad_norm": 0.14977139234542847,
      "learning_rate": 9.909674939985235e-06,
      "loss": 0.005,
      "step": 48190
    },
    {
      "epoch": 0.30913982804898776,
      "grad_norm": 0.49496692419052124,
      "learning_rate": 9.909569002934501e-06,
      "loss": 0.0043,
      "step": 48200
    },
    {
      "epoch": 0.30920396494277386,
      "grad_norm": 0.08514609932899475,
      "learning_rate": 9.909463004363359e-06,
      "loss": 0.0083,
      "step": 48210
    },
    {
      "epoch": 0.30926810183655995,
      "grad_norm": 0.17172439396381378,
      "learning_rate": 9.90935694427314e-06,
      "loss": 0.009,
      "step": 48220
    },
    {
      "epoch": 0.30933223873034604,
      "grad_norm": 0.3003605604171753,
      "learning_rate": 9.909250822665166e-06,
      "loss": 0.0061,
      "step": 48230
    },
    {
      "epoch": 0.30939637562413214,
      "grad_norm": 0.20939604938030243,
      "learning_rate": 9.909144639540772e-06,
      "loss": 0.0068,
      "step": 48240
    },
    {
      "epoch": 0.30946051251791823,
      "grad_norm": 0.3731301426887512,
      "learning_rate": 9.909038394901287e-06,
      "loss": 0.0036,
      "step": 48250
    },
    {
      "epoch": 0.3095246494117043,
      "grad_norm": 0.025112763047218323,
      "learning_rate": 9.908932088748042e-06,
      "loss": 0.0049,
      "step": 48260
    },
    {
      "epoch": 0.3095887863054904,
      "grad_norm": 0.12015421688556671,
      "learning_rate": 9.908825721082371e-06,
      "loss": 0.0061,
      "step": 48270
    },
    {
      "epoch": 0.3096529231992765,
      "grad_norm": 0.2233283519744873,
      "learning_rate": 9.908719291905605e-06,
      "loss": 0.0049,
      "step": 48280
    },
    {
      "epoch": 0.30971706009306266,
      "grad_norm": 0.12509781122207642,
      "learning_rate": 9.908612801219075e-06,
      "loss": 0.0063,
      "step": 48290
    },
    {
      "epoch": 0.30978119698684875,
      "grad_norm": 0.35053691267967224,
      "learning_rate": 9.908506249024122e-06,
      "loss": 0.0079,
      "step": 48300
    },
    {
      "epoch": 0.30984533388063484,
      "grad_norm": 0.15429310500621796,
      "learning_rate": 9.908399635322077e-06,
      "loss": 0.0052,
      "step": 48310
    },
    {
      "epoch": 0.30990947077442094,
      "grad_norm": 0.40210241079330444,
      "learning_rate": 9.908292960114276e-06,
      "loss": 0.008,
      "step": 48320
    },
    {
      "epoch": 0.30997360766820703,
      "grad_norm": 0.07661717385053635,
      "learning_rate": 9.908186223402056e-06,
      "loss": 0.0055,
      "step": 48330
    },
    {
      "epoch": 0.3100377445619931,
      "grad_norm": 0.2834452688694,
      "learning_rate": 9.908079425186757e-06,
      "loss": 0.0039,
      "step": 48340
    },
    {
      "epoch": 0.3101018814557792,
      "grad_norm": 0.04436261206865311,
      "learning_rate": 9.907972565469713e-06,
      "loss": 0.0046,
      "step": 48350
    },
    {
      "epoch": 0.3101660183495653,
      "grad_norm": 0.30914679169654846,
      "learning_rate": 9.907865644252265e-06,
      "loss": 0.0053,
      "step": 48360
    },
    {
      "epoch": 0.3102301552433514,
      "grad_norm": 0.06785024702548981,
      "learning_rate": 9.907758661535753e-06,
      "loss": 0.0027,
      "step": 48370
    },
    {
      "epoch": 0.3102942921371375,
      "grad_norm": 0.24976181983947754,
      "learning_rate": 9.907651617321518e-06,
      "loss": 0.0054,
      "step": 48380
    },
    {
      "epoch": 0.3103584290309236,
      "grad_norm": 0.152425155043602,
      "learning_rate": 9.9075445116109e-06,
      "loss": 0.0031,
      "step": 48390
    },
    {
      "epoch": 0.3104225659247097,
      "grad_norm": 0.14601565897464752,
      "learning_rate": 9.907437344405244e-06,
      "loss": 0.0064,
      "step": 48400
    },
    {
      "epoch": 0.3104867028184958,
      "grad_norm": 0.18691451847553253,
      "learning_rate": 9.90733011570589e-06,
      "loss": 0.0054,
      "step": 48410
    },
    {
      "epoch": 0.31055083971228187,
      "grad_norm": 0.03606029227375984,
      "learning_rate": 9.90722282551418e-06,
      "loss": 0.0046,
      "step": 48420
    },
    {
      "epoch": 0.310614976606068,
      "grad_norm": 0.23096883296966553,
      "learning_rate": 9.907115473831462e-06,
      "loss": 0.0029,
      "step": 48430
    },
    {
      "epoch": 0.3106791134998541,
      "grad_norm": 0.3457227647304535,
      "learning_rate": 9.90700806065908e-06,
      "loss": 0.0032,
      "step": 48440
    },
    {
      "epoch": 0.3107432503936402,
      "grad_norm": 0.17288430035114288,
      "learning_rate": 9.90690058599838e-06,
      "loss": 0.0038,
      "step": 48450
    },
    {
      "epoch": 0.3108073872874263,
      "grad_norm": 0.2500647008419037,
      "learning_rate": 9.906793049850712e-06,
      "loss": 0.0045,
      "step": 48460
    },
    {
      "epoch": 0.3108715241812124,
      "grad_norm": 0.22884602844715118,
      "learning_rate": 9.906685452217418e-06,
      "loss": 0.0054,
      "step": 48470
    },
    {
      "epoch": 0.3109356610749985,
      "grad_norm": 0.29955244064331055,
      "learning_rate": 9.906577793099847e-06,
      "loss": 0.004,
      "step": 48480
    },
    {
      "epoch": 0.3109997979687846,
      "grad_norm": 0.3578546345233917,
      "learning_rate": 9.90647007249935e-06,
      "loss": 0.006,
      "step": 48490
    },
    {
      "epoch": 0.31106393486257067,
      "grad_norm": 0.289204865694046,
      "learning_rate": 9.906362290417277e-06,
      "loss": 0.0042,
      "step": 48500
    },
    {
      "epoch": 0.31112807175635676,
      "grad_norm": 0.2857840359210968,
      "learning_rate": 9.906254446854979e-06,
      "loss": 0.0037,
      "step": 48510
    },
    {
      "epoch": 0.31119220865014285,
      "grad_norm": 0.18643750250339508,
      "learning_rate": 9.906146541813804e-06,
      "loss": 0.0104,
      "step": 48520
    },
    {
      "epoch": 0.31125634554392895,
      "grad_norm": 0.1863306611776352,
      "learning_rate": 9.90603857529511e-06,
      "loss": 0.0064,
      "step": 48530
    },
    {
      "epoch": 0.31132048243771504,
      "grad_norm": 0.1412154883146286,
      "learning_rate": 9.905930547300244e-06,
      "loss": 0.0048,
      "step": 48540
    },
    {
      "epoch": 0.31138461933150113,
      "grad_norm": 0.27536478638648987,
      "learning_rate": 9.90582245783056e-06,
      "loss": 0.0063,
      "step": 48550
    },
    {
      "epoch": 0.3114487562252872,
      "grad_norm": 0.476123183965683,
      "learning_rate": 9.905714306887417e-06,
      "loss": 0.0053,
      "step": 48560
    },
    {
      "epoch": 0.3115128931190734,
      "grad_norm": 0.1640210747718811,
      "learning_rate": 9.905606094472168e-06,
      "loss": 0.0032,
      "step": 48570
    },
    {
      "epoch": 0.31157703001285947,
      "grad_norm": 0.23330853879451752,
      "learning_rate": 9.905497820586168e-06,
      "loss": 0.0037,
      "step": 48580
    },
    {
      "epoch": 0.31164116690664556,
      "grad_norm": 0.20885784924030304,
      "learning_rate": 9.905389485230776e-06,
      "loss": 0.0036,
      "step": 48590
    },
    {
      "epoch": 0.31170530380043165,
      "grad_norm": 0.3812311887741089,
      "learning_rate": 9.905281088407346e-06,
      "loss": 0.0057,
      "step": 48600
    },
    {
      "epoch": 0.31176944069421775,
      "grad_norm": 0.2555144727230072,
      "learning_rate": 9.905172630117238e-06,
      "loss": 0.0038,
      "step": 48610
    },
    {
      "epoch": 0.31183357758800384,
      "grad_norm": 0.3070421516895294,
      "learning_rate": 9.905064110361813e-06,
      "loss": 0.0033,
      "step": 48620
    },
    {
      "epoch": 0.31189771448178993,
      "grad_norm": 0.2051517218351364,
      "learning_rate": 9.904955529142427e-06,
      "loss": 0.0043,
      "step": 48630
    },
    {
      "epoch": 0.311961851375576,
      "grad_norm": 0.39771977066993713,
      "learning_rate": 9.904846886460443e-06,
      "loss": 0.0054,
      "step": 48640
    },
    {
      "epoch": 0.3120259882693621,
      "grad_norm": 0.39983877539634705,
      "learning_rate": 9.904738182317224e-06,
      "loss": 0.0035,
      "step": 48650
    },
    {
      "epoch": 0.3120901251631482,
      "grad_norm": 0.19409961998462677,
      "learning_rate": 9.90462941671413e-06,
      "loss": 0.0031,
      "step": 48660
    },
    {
      "epoch": 0.3121542620569343,
      "grad_norm": 0.42691096663475037,
      "learning_rate": 9.904520589652523e-06,
      "loss": 0.0056,
      "step": 48670
    },
    {
      "epoch": 0.3122183989507204,
      "grad_norm": 0.1716923862695694,
      "learning_rate": 9.904411701133766e-06,
      "loss": 0.0034,
      "step": 48680
    },
    {
      "epoch": 0.3122825358445065,
      "grad_norm": 0.1936800628900528,
      "learning_rate": 9.904302751159227e-06,
      "loss": 0.0049,
      "step": 48690
    },
    {
      "epoch": 0.3123466727382926,
      "grad_norm": 0.25935399532318115,
      "learning_rate": 9.90419373973027e-06,
      "loss": 0.0042,
      "step": 48700
    },
    {
      "epoch": 0.31241080963207873,
      "grad_norm": 0.10071436315774918,
      "learning_rate": 9.90408466684826e-06,
      "loss": 0.0032,
      "step": 48710
    },
    {
      "epoch": 0.3124749465258648,
      "grad_norm": 0.23191823065280914,
      "learning_rate": 9.903975532514564e-06,
      "loss": 0.0046,
      "step": 48720
    },
    {
      "epoch": 0.3125390834196509,
      "grad_norm": 0.2393505722284317,
      "learning_rate": 9.90386633673055e-06,
      "loss": 0.0044,
      "step": 48730
    },
    {
      "epoch": 0.312603220313437,
      "grad_norm": 0.32246753573417664,
      "learning_rate": 9.903757079497587e-06,
      "loss": 0.0052,
      "step": 48740
    },
    {
      "epoch": 0.3126673572072231,
      "grad_norm": 0.30864331126213074,
      "learning_rate": 9.903647760817041e-06,
      "loss": 0.0078,
      "step": 48750
    },
    {
      "epoch": 0.3127314941010092,
      "grad_norm": 0.4452064037322998,
      "learning_rate": 9.903538380690284e-06,
      "loss": 0.0056,
      "step": 48760
    },
    {
      "epoch": 0.3127956309947953,
      "grad_norm": 0.1841566562652588,
      "learning_rate": 9.90342893911869e-06,
      "loss": 0.0046,
      "step": 48770
    },
    {
      "epoch": 0.3128597678885814,
      "grad_norm": 0.3984428942203522,
      "learning_rate": 9.903319436103625e-06,
      "loss": 0.0084,
      "step": 48780
    },
    {
      "epoch": 0.3129239047823675,
      "grad_norm": 0.4029768109321594,
      "learning_rate": 9.903209871646463e-06,
      "loss": 0.0058,
      "step": 48790
    },
    {
      "epoch": 0.31298804167615357,
      "grad_norm": 0.24533171951770782,
      "learning_rate": 9.903100245748576e-06,
      "loss": 0.0059,
      "step": 48800
    },
    {
      "epoch": 0.31305217856993967,
      "grad_norm": 0.13123102486133575,
      "learning_rate": 9.90299055841134e-06,
      "loss": 0.0045,
      "step": 48810
    },
    {
      "epoch": 0.31311631546372576,
      "grad_norm": 0.39378902316093445,
      "learning_rate": 9.90288080963613e-06,
      "loss": 0.0058,
      "step": 48820
    },
    {
      "epoch": 0.31318045235751185,
      "grad_norm": 0.28770095109939575,
      "learning_rate": 9.902770999424314e-06,
      "loss": 0.0049,
      "step": 48830
    },
    {
      "epoch": 0.31324458925129794,
      "grad_norm": 0.055193688720464706,
      "learning_rate": 9.902661127777277e-06,
      "loss": 0.0033,
      "step": 48840
    },
    {
      "epoch": 0.3133087261450841,
      "grad_norm": 0.056696776300668716,
      "learning_rate": 9.902551194696393e-06,
      "loss": 0.0056,
      "step": 48850
    },
    {
      "epoch": 0.3133728630388702,
      "grad_norm": 0.4966791868209839,
      "learning_rate": 9.902441200183038e-06,
      "loss": 0.0068,
      "step": 48860
    },
    {
      "epoch": 0.3134369999326563,
      "grad_norm": 0.31629717350006104,
      "learning_rate": 9.902331144238591e-06,
      "loss": 0.0109,
      "step": 48870
    },
    {
      "epoch": 0.3135011368264424,
      "grad_norm": 0.16882503032684326,
      "learning_rate": 9.90222102686443e-06,
      "loss": 0.0056,
      "step": 48880
    },
    {
      "epoch": 0.31356527372022847,
      "grad_norm": 0.07945968955755234,
      "learning_rate": 9.902110848061937e-06,
      "loss": 0.0037,
      "step": 48890
    },
    {
      "epoch": 0.31362941061401456,
      "grad_norm": 0.1186680793762207,
      "learning_rate": 9.90200060783249e-06,
      "loss": 0.0033,
      "step": 48900
    },
    {
      "epoch": 0.31369354750780065,
      "grad_norm": 0.13710758090019226,
      "learning_rate": 9.901890306177474e-06,
      "loss": 0.0045,
      "step": 48910
    },
    {
      "epoch": 0.31375768440158674,
      "grad_norm": 0.1552942842245102,
      "learning_rate": 9.901779943098269e-06,
      "loss": 0.005,
      "step": 48920
    },
    {
      "epoch": 0.31382182129537284,
      "grad_norm": 0.25293827056884766,
      "learning_rate": 9.901669518596258e-06,
      "loss": 0.004,
      "step": 48930
    },
    {
      "epoch": 0.31388595818915893,
      "grad_norm": 0.11367998272180557,
      "learning_rate": 9.901559032672824e-06,
      "loss": 0.0034,
      "step": 48940
    },
    {
      "epoch": 0.313950095082945,
      "grad_norm": 0.06295833736658096,
      "learning_rate": 9.901448485329353e-06,
      "loss": 0.0083,
      "step": 48950
    },
    {
      "epoch": 0.3140142319767311,
      "grad_norm": 0.2326330542564392,
      "learning_rate": 9.901337876567227e-06,
      "loss": 0.005,
      "step": 48960
    },
    {
      "epoch": 0.3140783688705172,
      "grad_norm": 0.2274763286113739,
      "learning_rate": 9.901227206387838e-06,
      "loss": 0.0048,
      "step": 48970
    },
    {
      "epoch": 0.3141425057643033,
      "grad_norm": 0.2659567594528198,
      "learning_rate": 9.901116474792566e-06,
      "loss": 0.0041,
      "step": 48980
    },
    {
      "epoch": 0.31420664265808945,
      "grad_norm": 0.14676564931869507,
      "learning_rate": 9.901005681782805e-06,
      "loss": 0.0039,
      "step": 48990
    },
    {
      "epoch": 0.31427077955187555,
      "grad_norm": 0.06894460320472717,
      "learning_rate": 9.900894827359937e-06,
      "loss": 0.0045,
      "step": 49000
    },
    {
      "epoch": 0.31433491644566164,
      "grad_norm": 0.13913805782794952,
      "learning_rate": 9.900783911525357e-06,
      "loss": 0.0038,
      "step": 49010
    },
    {
      "epoch": 0.31439905333944773,
      "grad_norm": 0.22315600514411926,
      "learning_rate": 9.900672934280451e-06,
      "loss": 0.0048,
      "step": 49020
    },
    {
      "epoch": 0.3144631902332338,
      "grad_norm": 0.10821523517370224,
      "learning_rate": 9.90056189562661e-06,
      "loss": 0.0038,
      "step": 49030
    },
    {
      "epoch": 0.3145273271270199,
      "grad_norm": 0.2065705507993698,
      "learning_rate": 9.900450795565226e-06,
      "loss": 0.004,
      "step": 49040
    },
    {
      "epoch": 0.314591464020806,
      "grad_norm": 0.24347898364067078,
      "learning_rate": 9.900339634097693e-06,
      "loss": 0.0029,
      "step": 49050
    },
    {
      "epoch": 0.3146556009145921,
      "grad_norm": 0.200932577252388,
      "learning_rate": 9.900228411225401e-06,
      "loss": 0.0033,
      "step": 49060
    },
    {
      "epoch": 0.3147197378083782,
      "grad_norm": 0.21611665189266205,
      "learning_rate": 9.900117126949744e-06,
      "loss": 0.0037,
      "step": 49070
    },
    {
      "epoch": 0.3147838747021643,
      "grad_norm": 0.31797540187835693,
      "learning_rate": 9.900005781272118e-06,
      "loss": 0.004,
      "step": 49080
    },
    {
      "epoch": 0.3148480115959504,
      "grad_norm": 0.22289103269577026,
      "learning_rate": 9.899894374193918e-06,
      "loss": 0.0051,
      "step": 49090
    },
    {
      "epoch": 0.3149121484897365,
      "grad_norm": 0.35321786999702454,
      "learning_rate": 9.899782905716539e-06,
      "loss": 0.0053,
      "step": 49100
    },
    {
      "epoch": 0.31497628538352257,
      "grad_norm": 0.242935910820961,
      "learning_rate": 9.89967137584138e-06,
      "loss": 0.0042,
      "step": 49110
    },
    {
      "epoch": 0.31504042227730866,
      "grad_norm": 0.0777844786643982,
      "learning_rate": 9.899559784569834e-06,
      "loss": 0.0027,
      "step": 49120
    },
    {
      "epoch": 0.3151045591710948,
      "grad_norm": 0.07172335684299469,
      "learning_rate": 9.899448131903305e-06,
      "loss": 0.0044,
      "step": 49130
    },
    {
      "epoch": 0.3151686960648809,
      "grad_norm": 0.1554524302482605,
      "learning_rate": 9.899336417843188e-06,
      "loss": 0.0086,
      "step": 49140
    },
    {
      "epoch": 0.315232832958667,
      "grad_norm": 0.2136203497648239,
      "learning_rate": 9.899224642390884e-06,
      "loss": 0.0077,
      "step": 49150
    },
    {
      "epoch": 0.3152969698524531,
      "grad_norm": 0.1316751092672348,
      "learning_rate": 9.899112805547795e-06,
      "loss": 0.0062,
      "step": 49160
    },
    {
      "epoch": 0.3153611067462392,
      "grad_norm": 0.2350343018770218,
      "learning_rate": 9.89900090731532e-06,
      "loss": 0.0072,
      "step": 49170
    },
    {
      "epoch": 0.3154252436400253,
      "grad_norm": 0.2078297734260559,
      "learning_rate": 9.898888947694863e-06,
      "loss": 0.0045,
      "step": 49180
    },
    {
      "epoch": 0.31548938053381137,
      "grad_norm": 0.1405486911535263,
      "learning_rate": 9.898776926687826e-06,
      "loss": 0.0039,
      "step": 49190
    },
    {
      "epoch": 0.31555351742759746,
      "grad_norm": 0.1346013844013214,
      "learning_rate": 9.898664844295612e-06,
      "loss": 0.0046,
      "step": 49200
    },
    {
      "epoch": 0.31561765432138356,
      "grad_norm": 0.23130005598068237,
      "learning_rate": 9.898552700519628e-06,
      "loss": 0.0036,
      "step": 49210
    },
    {
      "epoch": 0.31568179121516965,
      "grad_norm": 0.0856684073805809,
      "learning_rate": 9.898440495361277e-06,
      "loss": 0.0057,
      "step": 49220
    },
    {
      "epoch": 0.31574592810895574,
      "grad_norm": 0.19981451332569122,
      "learning_rate": 9.898328228821966e-06,
      "loss": 0.0052,
      "step": 49230
    },
    {
      "epoch": 0.31581006500274184,
      "grad_norm": 0.31087830662727356,
      "learning_rate": 9.898215900903101e-06,
      "loss": 0.0067,
      "step": 49240
    },
    {
      "epoch": 0.31587420189652793,
      "grad_norm": 0.27627718448638916,
      "learning_rate": 9.898103511606092e-06,
      "loss": 0.0021,
      "step": 49250
    },
    {
      "epoch": 0.315938338790314,
      "grad_norm": 0.30112534761428833,
      "learning_rate": 9.897991060932345e-06,
      "loss": 0.0074,
      "step": 49260
    },
    {
      "epoch": 0.31600247568410017,
      "grad_norm": 0.20653463900089264,
      "learning_rate": 9.897878548883266e-06,
      "loss": 0.0037,
      "step": 49270
    },
    {
      "epoch": 0.31606661257788626,
      "grad_norm": 0.4281090497970581,
      "learning_rate": 9.897765975460272e-06,
      "loss": 0.0046,
      "step": 49280
    },
    {
      "epoch": 0.31613074947167236,
      "grad_norm": 0.18075962364673615,
      "learning_rate": 9.897653340664767e-06,
      "loss": 0.005,
      "step": 49290
    },
    {
      "epoch": 0.31619488636545845,
      "grad_norm": 0.19884683191776276,
      "learning_rate": 9.897540644498168e-06,
      "loss": 0.0043,
      "step": 49300
    },
    {
      "epoch": 0.31625902325924454,
      "grad_norm": 0.10651399940252304,
      "learning_rate": 9.897427886961884e-06,
      "loss": 0.0031,
      "step": 49310
    },
    {
      "epoch": 0.31632316015303064,
      "grad_norm": 0.38676655292510986,
      "learning_rate": 9.897315068057327e-06,
      "loss": 0.0042,
      "step": 49320
    },
    {
      "epoch": 0.31638729704681673,
      "grad_norm": 0.012912292033433914,
      "learning_rate": 9.897202187785913e-06,
      "loss": 0.0042,
      "step": 49330
    },
    {
      "epoch": 0.3164514339406028,
      "grad_norm": 0.08300850540399551,
      "learning_rate": 9.897089246149056e-06,
      "loss": 0.0034,
      "step": 49340
    },
    {
      "epoch": 0.3165155708343889,
      "grad_norm": 0.14932957291603088,
      "learning_rate": 9.896976243148171e-06,
      "loss": 0.0065,
      "step": 49350
    },
    {
      "epoch": 0.316579707728175,
      "grad_norm": 0.17913511395454407,
      "learning_rate": 9.896863178784673e-06,
      "loss": 0.0054,
      "step": 49360
    },
    {
      "epoch": 0.3166438446219611,
      "grad_norm": 0.32844895124435425,
      "learning_rate": 9.89675005305998e-06,
      "loss": 0.0046,
      "step": 49370
    },
    {
      "epoch": 0.3167079815157472,
      "grad_norm": 0.3574899137020111,
      "learning_rate": 9.896636865975509e-06,
      "loss": 0.007,
      "step": 49380
    },
    {
      "epoch": 0.3167721184095333,
      "grad_norm": 0.5497488975524902,
      "learning_rate": 9.89652361753268e-06,
      "loss": 0.0049,
      "step": 49390
    },
    {
      "epoch": 0.3168362553033194,
      "grad_norm": 0.3612213730812073,
      "learning_rate": 9.896410307732909e-06,
      "loss": 0.0048,
      "step": 49400
    },
    {
      "epoch": 0.31690039219710553,
      "grad_norm": 0.1442115157842636,
      "learning_rate": 9.896296936577618e-06,
      "loss": 0.0043,
      "step": 49410
    },
    {
      "epoch": 0.3169645290908916,
      "grad_norm": 0.18238170444965363,
      "learning_rate": 9.896183504068226e-06,
      "loss": 0.0029,
      "step": 49420
    },
    {
      "epoch": 0.3170286659846777,
      "grad_norm": 0.18228575587272644,
      "learning_rate": 9.896070010206155e-06,
      "loss": 0.0055,
      "step": 49430
    },
    {
      "epoch": 0.3170928028784638,
      "grad_norm": 0.3545880913734436,
      "learning_rate": 9.89595645499283e-06,
      "loss": 0.0084,
      "step": 49440
    },
    {
      "epoch": 0.3171569397722499,
      "grad_norm": 0.10810332745313644,
      "learning_rate": 9.89584283842967e-06,
      "loss": 0.0069,
      "step": 49450
    },
    {
      "epoch": 0.317221076666036,
      "grad_norm": 0.2051643431186676,
      "learning_rate": 9.895729160518102e-06,
      "loss": 0.0051,
      "step": 49460
    },
    {
      "epoch": 0.3172852135598221,
      "grad_norm": 0.22339393198490143,
      "learning_rate": 9.895615421259548e-06,
      "loss": 0.0054,
      "step": 49470
    },
    {
      "epoch": 0.3173493504536082,
      "grad_norm": 0.026365969330072403,
      "learning_rate": 9.895501620655431e-06,
      "loss": 0.0039,
      "step": 49480
    },
    {
      "epoch": 0.3174134873473943,
      "grad_norm": 0.2481347918510437,
      "learning_rate": 9.895387758707182e-06,
      "loss": 0.0043,
      "step": 49490
    },
    {
      "epoch": 0.31747762424118037,
      "grad_norm": 0.3272586166858673,
      "learning_rate": 9.895273835416227e-06,
      "loss": 0.0044,
      "step": 49500
    },
    {
      "epoch": 0.31754176113496646,
      "grad_norm": 0.13250267505645752,
      "learning_rate": 9.895159850783991e-06,
      "loss": 0.0033,
      "step": 49510
    },
    {
      "epoch": 0.31760589802875255,
      "grad_norm": 0.266339510679245,
      "learning_rate": 9.895045804811903e-06,
      "loss": 0.0061,
      "step": 49520
    },
    {
      "epoch": 0.31767003492253865,
      "grad_norm": 0.30501192808151245,
      "learning_rate": 9.894931697501394e-06,
      "loss": 0.003,
      "step": 49530
    },
    {
      "epoch": 0.31773417181632474,
      "grad_norm": 0.22355207800865173,
      "learning_rate": 9.894817528853891e-06,
      "loss": 0.0039,
      "step": 49540
    },
    {
      "epoch": 0.3177983087101109,
      "grad_norm": 0.23635293543338776,
      "learning_rate": 9.894703298870827e-06,
      "loss": 0.0046,
      "step": 49550
    },
    {
      "epoch": 0.317862445603897,
      "grad_norm": 0.269288569688797,
      "learning_rate": 9.894589007553631e-06,
      "loss": 0.0066,
      "step": 49560
    },
    {
      "epoch": 0.3179265824976831,
      "grad_norm": 0.24839213490486145,
      "learning_rate": 9.894474654903739e-06,
      "loss": 0.0028,
      "step": 49570
    },
    {
      "epoch": 0.31799071939146917,
      "grad_norm": 0.07492738962173462,
      "learning_rate": 9.894360240922581e-06,
      "loss": 0.0038,
      "step": 49580
    },
    {
      "epoch": 0.31805485628525526,
      "grad_norm": 0.1257239133119583,
      "learning_rate": 9.89424576561159e-06,
      "loss": 0.0055,
      "step": 49590
    },
    {
      "epoch": 0.31811899317904135,
      "grad_norm": 0.12412641942501068,
      "learning_rate": 9.894131228972203e-06,
      "loss": 0.0052,
      "step": 49600
    },
    {
      "epoch": 0.31818313007282745,
      "grad_norm": 0.19532410800457,
      "learning_rate": 9.894016631005853e-06,
      "loss": 0.0041,
      "step": 49610
    },
    {
      "epoch": 0.31824726696661354,
      "grad_norm": 0.27602460980415344,
      "learning_rate": 9.893901971713978e-06,
      "loss": 0.0064,
      "step": 49620
    },
    {
      "epoch": 0.31831140386039963,
      "grad_norm": 0.5332382917404175,
      "learning_rate": 9.893787251098012e-06,
      "loss": 0.0065,
      "step": 49630
    },
    {
      "epoch": 0.3183755407541857,
      "grad_norm": 0.32323774695396423,
      "learning_rate": 9.893672469159396e-06,
      "loss": 0.0051,
      "step": 49640
    },
    {
      "epoch": 0.3184396776479718,
      "grad_norm": 0.14860421419143677,
      "learning_rate": 9.893557625899565e-06,
      "loss": 0.0037,
      "step": 49650
    },
    {
      "epoch": 0.3185038145417579,
      "grad_norm": 0.043334994465112686,
      "learning_rate": 9.893442721319962e-06,
      "loss": 0.0055,
      "step": 49660
    },
    {
      "epoch": 0.318567951435544,
      "grad_norm": 0.4149910509586334,
      "learning_rate": 9.893327755422022e-06,
      "loss": 0.0069,
      "step": 49670
    },
    {
      "epoch": 0.3186320883293301,
      "grad_norm": 0.20254351198673248,
      "learning_rate": 9.89321272820719e-06,
      "loss": 0.008,
      "step": 49680
    },
    {
      "epoch": 0.3186962252231162,
      "grad_norm": 0.2773299813270569,
      "learning_rate": 9.893097639676902e-06,
      "loss": 0.0062,
      "step": 49690
    },
    {
      "epoch": 0.31876036211690234,
      "grad_norm": 0.23761241137981415,
      "learning_rate": 9.892982489832605e-06,
      "loss": 0.0052,
      "step": 49700
    },
    {
      "epoch": 0.31882449901068843,
      "grad_norm": 0.2280692160129547,
      "learning_rate": 9.892867278675742e-06,
      "loss": 0.006,
      "step": 49710
    },
    {
      "epoch": 0.3188886359044745,
      "grad_norm": 0.39356282353401184,
      "learning_rate": 9.892752006207754e-06,
      "loss": 0.006,
      "step": 49720
    },
    {
      "epoch": 0.3189527727982606,
      "grad_norm": 0.11281643807888031,
      "learning_rate": 9.892636672430088e-06,
      "loss": 0.0034,
      "step": 49730
    },
    {
      "epoch": 0.3190169096920467,
      "grad_norm": 0.1185554713010788,
      "learning_rate": 9.892521277344184e-06,
      "loss": 0.0047,
      "step": 49740
    },
    {
      "epoch": 0.3190810465858328,
      "grad_norm": 0.09055802971124649,
      "learning_rate": 9.892405820951496e-06,
      "loss": 0.0053,
      "step": 49750
    },
    {
      "epoch": 0.3191451834796189,
      "grad_norm": 0.19612687826156616,
      "learning_rate": 9.892290303253465e-06,
      "loss": 0.0025,
      "step": 49760
    },
    {
      "epoch": 0.319209320373405,
      "grad_norm": 0.3884764611721039,
      "learning_rate": 9.89217472425154e-06,
      "loss": 0.0046,
      "step": 49770
    },
    {
      "epoch": 0.3192734572671911,
      "grad_norm": 0.2126941829919815,
      "learning_rate": 9.892059083947168e-06,
      "loss": 0.0038,
      "step": 49780
    },
    {
      "epoch": 0.3193375941609772,
      "grad_norm": 0.31118959188461304,
      "learning_rate": 9.891943382341799e-06,
      "loss": 0.0067,
      "step": 49790
    },
    {
      "epoch": 0.31940173105476327,
      "grad_norm": 0.13976700603961945,
      "learning_rate": 9.891827619436886e-06,
      "loss": 0.0047,
      "step": 49800
    },
    {
      "epoch": 0.31946586794854936,
      "grad_norm": 0.15904361009597778,
      "learning_rate": 9.891711795233876e-06,
      "loss": 0.0044,
      "step": 49810
    },
    {
      "epoch": 0.31953000484233546,
      "grad_norm": 0.3219791650772095,
      "learning_rate": 9.891595909734219e-06,
      "loss": 0.0063,
      "step": 49820
    },
    {
      "epoch": 0.31959414173612155,
      "grad_norm": 0.2676171362400055,
      "learning_rate": 9.891479962939371e-06,
      "loss": 0.0032,
      "step": 49830
    },
    {
      "epoch": 0.3196582786299077,
      "grad_norm": 0.16709725558757782,
      "learning_rate": 9.891363954850784e-06,
      "loss": 0.0054,
      "step": 49840
    },
    {
      "epoch": 0.3197224155236938,
      "grad_norm": 0.21087214350700378,
      "learning_rate": 9.89124788546991e-06,
      "loss": 0.0049,
      "step": 49850
    },
    {
      "epoch": 0.3197865524174799,
      "grad_norm": 0.13865779340267181,
      "learning_rate": 9.891131754798204e-06,
      "loss": 0.0042,
      "step": 49860
    },
    {
      "epoch": 0.319850689311266,
      "grad_norm": 0.30375921726226807,
      "learning_rate": 9.891015562837122e-06,
      "loss": 0.0048,
      "step": 49870
    },
    {
      "epoch": 0.31991482620505207,
      "grad_norm": 0.19928953051567078,
      "learning_rate": 9.890899309588119e-06,
      "loss": 0.0042,
      "step": 49880
    },
    {
      "epoch": 0.31997896309883817,
      "grad_norm": 0.23441387712955475,
      "learning_rate": 9.890782995052652e-06,
      "loss": 0.002,
      "step": 49890
    },
    {
      "epoch": 0.32004309999262426,
      "grad_norm": 0.18868665397167206,
      "learning_rate": 9.89066661923218e-06,
      "loss": 0.0041,
      "step": 49900
    },
    {
      "epoch": 0.32010723688641035,
      "grad_norm": 0.3113063871860504,
      "learning_rate": 9.890550182128158e-06,
      "loss": 0.0036,
      "step": 49910
    },
    {
      "epoch": 0.32017137378019644,
      "grad_norm": 0.1571812927722931,
      "learning_rate": 9.890433683742047e-06,
      "loss": 0.0029,
      "step": 49920
    },
    {
      "epoch": 0.32023551067398254,
      "grad_norm": 0.3203980624675751,
      "learning_rate": 9.890317124075308e-06,
      "loss": 0.0061,
      "step": 49930
    },
    {
      "epoch": 0.32029964756776863,
      "grad_norm": 0.21253187954425812,
      "learning_rate": 9.8902005031294e-06,
      "loss": 0.0091,
      "step": 49940
    },
    {
      "epoch": 0.3203637844615547,
      "grad_norm": 0.09002522379159927,
      "learning_rate": 9.890083820905785e-06,
      "loss": 0.0084,
      "step": 49950
    },
    {
      "epoch": 0.3204279213553408,
      "grad_norm": 0.17045079171657562,
      "learning_rate": 9.889967077405925e-06,
      "loss": 0.0075,
      "step": 49960
    },
    {
      "epoch": 0.3204920582491269,
      "grad_norm": 0.10132025927305222,
      "learning_rate": 9.889850272631283e-06,
      "loss": 0.0038,
      "step": 49970
    },
    {
      "epoch": 0.32055619514291306,
      "grad_norm": 0.11106637120246887,
      "learning_rate": 9.889733406583321e-06,
      "loss": 0.004,
      "step": 49980
    },
    {
      "epoch": 0.32062033203669915,
      "grad_norm": 0.07718731462955475,
      "learning_rate": 9.889616479263506e-06,
      "loss": 0.007,
      "step": 49990
    },
    {
      "epoch": 0.32068446893048524,
      "grad_norm": 0.34893959760665894,
      "learning_rate": 9.889499490673301e-06,
      "loss": 0.0037,
      "step": 50000
    },
    {
      "epoch": 0.32074860582427134,
      "grad_norm": 0.13644975423812866,
      "learning_rate": 9.889382440814172e-06,
      "loss": 0.0068,
      "step": 50010
    },
    {
      "epoch": 0.32081274271805743,
      "grad_norm": 0.2971213757991791,
      "learning_rate": 9.88926532968759e-06,
      "loss": 0.0044,
      "step": 50020
    },
    {
      "epoch": 0.3208768796118435,
      "grad_norm": 0.07066557556390762,
      "learning_rate": 9.889148157295017e-06,
      "loss": 0.0031,
      "step": 50030
    },
    {
      "epoch": 0.3209410165056296,
      "grad_norm": 0.9618569016456604,
      "learning_rate": 9.889030923637923e-06,
      "loss": 0.0105,
      "step": 50040
    },
    {
      "epoch": 0.3210051533994157,
      "grad_norm": 0.5154543519020081,
      "learning_rate": 9.888913628717778e-06,
      "loss": 0.0042,
      "step": 50050
    },
    {
      "epoch": 0.3210692902932018,
      "grad_norm": 0.22388550639152527,
      "learning_rate": 9.88879627253605e-06,
      "loss": 0.0051,
      "step": 50060
    },
    {
      "epoch": 0.3211334271869879,
      "grad_norm": 0.17283904552459717,
      "learning_rate": 9.888678855094213e-06,
      "loss": 0.0048,
      "step": 50070
    },
    {
      "epoch": 0.321197564080774,
      "grad_norm": 0.21307899057865143,
      "learning_rate": 9.888561376393734e-06,
      "loss": 0.0063,
      "step": 50080
    },
    {
      "epoch": 0.3212617009745601,
      "grad_norm": 0.4096912741661072,
      "learning_rate": 9.888443836436088e-06,
      "loss": 0.0054,
      "step": 50090
    },
    {
      "epoch": 0.3213258378683462,
      "grad_norm": 0.28107506036758423,
      "learning_rate": 9.888326235222748e-06,
      "loss": 0.0056,
      "step": 50100
    },
    {
      "epoch": 0.32138997476213227,
      "grad_norm": 0.08909334242343903,
      "learning_rate": 9.888208572755185e-06,
      "loss": 0.0049,
      "step": 50110
    },
    {
      "epoch": 0.3214541116559184,
      "grad_norm": 0.13259750604629517,
      "learning_rate": 9.888090849034876e-06,
      "loss": 0.0033,
      "step": 50120
    },
    {
      "epoch": 0.3215182485497045,
      "grad_norm": 0.2727391719818115,
      "learning_rate": 9.887973064063294e-06,
      "loss": 0.0036,
      "step": 50130
    },
    {
      "epoch": 0.3215823854434906,
      "grad_norm": 0.06979379802942276,
      "learning_rate": 9.887855217841918e-06,
      "loss": 0.0036,
      "step": 50140
    },
    {
      "epoch": 0.3216465223372767,
      "grad_norm": 0.5669009685516357,
      "learning_rate": 9.887737310372221e-06,
      "loss": 0.0059,
      "step": 50150
    },
    {
      "epoch": 0.3217106592310628,
      "grad_norm": 0.178040012717247,
      "learning_rate": 9.887619341655683e-06,
      "loss": 0.0037,
      "step": 50160
    },
    {
      "epoch": 0.3217747961248489,
      "grad_norm": 0.4985739290714264,
      "learning_rate": 9.887501311693782e-06,
      "loss": 0.0058,
      "step": 50170
    },
    {
      "epoch": 0.321838933018635,
      "grad_norm": 0.054809264838695526,
      "learning_rate": 9.887383220487995e-06,
      "loss": 0.004,
      "step": 50180
    },
    {
      "epoch": 0.32190306991242107,
      "grad_norm": 0.29745447635650635,
      "learning_rate": 9.887265068039806e-06,
      "loss": 0.0056,
      "step": 50190
    },
    {
      "epoch": 0.32196720680620716,
      "grad_norm": 0.2449481040239334,
      "learning_rate": 9.887146854350692e-06,
      "loss": 0.0046,
      "step": 50200
    },
    {
      "epoch": 0.32203134369999326,
      "grad_norm": 0.07773438841104507,
      "learning_rate": 9.887028579422132e-06,
      "loss": 0.0038,
      "step": 50210
    },
    {
      "epoch": 0.32209548059377935,
      "grad_norm": 0.08787532150745392,
      "learning_rate": 9.886910243255614e-06,
      "loss": 0.0034,
      "step": 50220
    },
    {
      "epoch": 0.32215961748756544,
      "grad_norm": 0.10316630452871323,
      "learning_rate": 9.886791845852619e-06,
      "loss": 0.0037,
      "step": 50230
    },
    {
      "epoch": 0.32222375438135153,
      "grad_norm": 0.24471049010753632,
      "learning_rate": 9.886673387214628e-06,
      "loss": 0.0084,
      "step": 50240
    },
    {
      "epoch": 0.32228789127513763,
      "grad_norm": 0.13041454553604126,
      "learning_rate": 9.886554867343127e-06,
      "loss": 0.0025,
      "step": 50250
    },
    {
      "epoch": 0.3223520281689238,
      "grad_norm": 0.4005851447582245,
      "learning_rate": 9.886436286239602e-06,
      "loss": 0.0044,
      "step": 50260
    },
    {
      "epoch": 0.32241616506270987,
      "grad_norm": 0.19282877445220947,
      "learning_rate": 9.886317643905536e-06,
      "loss": 0.0026,
      "step": 50270
    },
    {
      "epoch": 0.32248030195649596,
      "grad_norm": 0.2606230676174164,
      "learning_rate": 9.88619894034242e-06,
      "loss": 0.0058,
      "step": 50280
    },
    {
      "epoch": 0.32254443885028206,
      "grad_norm": 0.13031746447086334,
      "learning_rate": 9.886080175551738e-06,
      "loss": 0.0046,
      "step": 50290
    },
    {
      "epoch": 0.32260857574406815,
      "grad_norm": 0.2334151417016983,
      "learning_rate": 9.885961349534981e-06,
      "loss": 0.0039,
      "step": 50300
    },
    {
      "epoch": 0.32267271263785424,
      "grad_norm": 0.17930831015110016,
      "learning_rate": 9.885842462293633e-06,
      "loss": 0.009,
      "step": 50310
    },
    {
      "epoch": 0.32273684953164034,
      "grad_norm": 0.2496260404586792,
      "learning_rate": 9.88572351382919e-06,
      "loss": 0.0049,
      "step": 50320
    },
    {
      "epoch": 0.32280098642542643,
      "grad_norm": 0.22151890397071838,
      "learning_rate": 9.885604504143138e-06,
      "loss": 0.009,
      "step": 50330
    },
    {
      "epoch": 0.3228651233192125,
      "grad_norm": 0.09284746646881104,
      "learning_rate": 9.88548543323697e-06,
      "loss": 0.0061,
      "step": 50340
    },
    {
      "epoch": 0.3229292602129986,
      "grad_norm": 0.19882896542549133,
      "learning_rate": 9.88536630111218e-06,
      "loss": 0.0084,
      "step": 50350
    },
    {
      "epoch": 0.3229933971067847,
      "grad_norm": 0.11681967228651047,
      "learning_rate": 9.885247107770257e-06,
      "loss": 0.0052,
      "step": 50360
    },
    {
      "epoch": 0.3230575340005708,
      "grad_norm": 0.10497705638408661,
      "learning_rate": 9.885127853212696e-06,
      "loss": 0.0045,
      "step": 50370
    },
    {
      "epoch": 0.3231216708943569,
      "grad_norm": 0.17021536827087402,
      "learning_rate": 9.885008537440992e-06,
      "loss": 0.0056,
      "step": 50380
    },
    {
      "epoch": 0.323185807788143,
      "grad_norm": 0.3915821313858032,
      "learning_rate": 9.884889160456642e-06,
      "loss": 0.0046,
      "step": 50390
    },
    {
      "epoch": 0.32324994468192914,
      "grad_norm": 0.26966342329978943,
      "learning_rate": 9.884769722261136e-06,
      "loss": 0.0053,
      "step": 50400
    },
    {
      "epoch": 0.32331408157571523,
      "grad_norm": 0.3279496133327484,
      "learning_rate": 9.884650222855976e-06,
      "loss": 0.0038,
      "step": 50410
    },
    {
      "epoch": 0.3233782184695013,
      "grad_norm": 0.1679442822933197,
      "learning_rate": 9.884530662242658e-06,
      "loss": 0.0043,
      "step": 50420
    },
    {
      "epoch": 0.3234423553632874,
      "grad_norm": 0.07241009920835495,
      "learning_rate": 9.884411040422679e-06,
      "loss": 0.0037,
      "step": 50430
    },
    {
      "epoch": 0.3235064922570735,
      "grad_norm": 0.20607556402683258,
      "learning_rate": 9.884291357397539e-06,
      "loss": 0.0045,
      "step": 50440
    },
    {
      "epoch": 0.3235706291508596,
      "grad_norm": 0.24917933344841003,
      "learning_rate": 9.884171613168738e-06,
      "loss": 0.0034,
      "step": 50450
    },
    {
      "epoch": 0.3236347660446457,
      "grad_norm": 0.2530268728733063,
      "learning_rate": 9.884051807737776e-06,
      "loss": 0.0028,
      "step": 50460
    },
    {
      "epoch": 0.3236989029384318,
      "grad_norm": 0.1264554262161255,
      "learning_rate": 9.883931941106154e-06,
      "loss": 0.0071,
      "step": 50470
    },
    {
      "epoch": 0.3237630398322179,
      "grad_norm": 0.32021564245224,
      "learning_rate": 9.883812013275374e-06,
      "loss": 0.0052,
      "step": 50480
    },
    {
      "epoch": 0.323827176726004,
      "grad_norm": 0.03336715325713158,
      "learning_rate": 9.88369202424694e-06,
      "loss": 0.0028,
      "step": 50490
    },
    {
      "epoch": 0.32389131361979007,
      "grad_norm": 0.1920616328716278,
      "learning_rate": 9.883571974022355e-06,
      "loss": 0.008,
      "step": 50500
    },
    {
      "epoch": 0.32395545051357616,
      "grad_norm": 0.1073715016245842,
      "learning_rate": 9.883451862603121e-06,
      "loss": 0.0047,
      "step": 50510
    },
    {
      "epoch": 0.32401958740736225,
      "grad_norm": 0.17998799681663513,
      "learning_rate": 9.883331689990746e-06,
      "loss": 0.0053,
      "step": 50520
    },
    {
      "epoch": 0.32408372430114835,
      "grad_norm": 0.4126385748386383,
      "learning_rate": 9.883211456186735e-06,
      "loss": 0.0074,
      "step": 50530
    },
    {
      "epoch": 0.3241478611949345,
      "grad_norm": 0.2791917026042938,
      "learning_rate": 9.883091161192595e-06,
      "loss": 0.0054,
      "step": 50540
    },
    {
      "epoch": 0.3242119980887206,
      "grad_norm": 0.1871456801891327,
      "learning_rate": 9.882970805009831e-06,
      "loss": 0.0048,
      "step": 50550
    },
    {
      "epoch": 0.3242761349825067,
      "grad_norm": 0.1261935979127884,
      "learning_rate": 9.882850387639956e-06,
      "loss": 0.006,
      "step": 50560
    },
    {
      "epoch": 0.3243402718762928,
      "grad_norm": 0.25968706607818604,
      "learning_rate": 9.882729909084472e-06,
      "loss": 0.0055,
      "step": 50570
    },
    {
      "epoch": 0.32440440877007887,
      "grad_norm": 0.12232595682144165,
      "learning_rate": 9.882609369344895e-06,
      "loss": 0.0038,
      "step": 50580
    },
    {
      "epoch": 0.32446854566386496,
      "grad_norm": 0.25618013739585876,
      "learning_rate": 9.882488768422733e-06,
      "loss": 0.0071,
      "step": 50590
    },
    {
      "epoch": 0.32453268255765105,
      "grad_norm": 0.0873841866850853,
      "learning_rate": 9.882368106319497e-06,
      "loss": 0.005,
      "step": 50600
    },
    {
      "epoch": 0.32459681945143715,
      "grad_norm": 0.28166040778160095,
      "learning_rate": 9.8822473830367e-06,
      "loss": 0.0034,
      "step": 50610
    },
    {
      "epoch": 0.32466095634522324,
      "grad_norm": 0.19365577399730682,
      "learning_rate": 9.882126598575854e-06,
      "loss": 0.0044,
      "step": 50620
    },
    {
      "epoch": 0.32472509323900933,
      "grad_norm": 0.14441215991973877,
      "learning_rate": 9.882005752938471e-06,
      "loss": 0.004,
      "step": 50630
    },
    {
      "epoch": 0.3247892301327954,
      "grad_norm": 0.2671840786933899,
      "learning_rate": 9.881884846126069e-06,
      "loss": 0.0055,
      "step": 50640
    },
    {
      "epoch": 0.3248533670265815,
      "grad_norm": 0.5790324211120605,
      "learning_rate": 9.881763878140157e-06,
      "loss": 0.0052,
      "step": 50650
    },
    {
      "epoch": 0.3249175039203676,
      "grad_norm": 0.03954846411943436,
      "learning_rate": 9.881642848982258e-06,
      "loss": 0.0028,
      "step": 50660
    },
    {
      "epoch": 0.3249816408141537,
      "grad_norm": 0.10793976485729218,
      "learning_rate": 9.881521758653885e-06,
      "loss": 0.0056,
      "step": 50670
    },
    {
      "epoch": 0.32504577770793985,
      "grad_norm": 0.06970980763435364,
      "learning_rate": 9.881400607156556e-06,
      "loss": 0.0071,
      "step": 50680
    },
    {
      "epoch": 0.32510991460172595,
      "grad_norm": 0.29071202874183655,
      "learning_rate": 9.881279394491787e-06,
      "loss": 0.0074,
      "step": 50690
    },
    {
      "epoch": 0.32517405149551204,
      "grad_norm": 0.30976402759552,
      "learning_rate": 9.881158120661098e-06,
      "loss": 0.0059,
      "step": 50700
    },
    {
      "epoch": 0.32523818838929813,
      "grad_norm": 0.27598875761032104,
      "learning_rate": 9.88103678566601e-06,
      "loss": 0.0036,
      "step": 50710
    },
    {
      "epoch": 0.3253023252830842,
      "grad_norm": 0.4413817524909973,
      "learning_rate": 9.880915389508045e-06,
      "loss": 0.0045,
      "step": 50720
    },
    {
      "epoch": 0.3253664621768703,
      "grad_norm": 0.11329472064971924,
      "learning_rate": 9.88079393218872e-06,
      "loss": 0.0054,
      "step": 50730
    },
    {
      "epoch": 0.3254305990706564,
      "grad_norm": 0.22125688195228577,
      "learning_rate": 9.880672413709559e-06,
      "loss": 0.0067,
      "step": 50740
    },
    {
      "epoch": 0.3254947359644425,
      "grad_norm": 0.4026035964488983,
      "learning_rate": 9.880550834072084e-06,
      "loss": 0.0061,
      "step": 50750
    },
    {
      "epoch": 0.3255588728582286,
      "grad_norm": 0.11307325959205627,
      "learning_rate": 9.88042919327782e-06,
      "loss": 0.0045,
      "step": 50760
    },
    {
      "epoch": 0.3256230097520147,
      "grad_norm": 0.19422641396522522,
      "learning_rate": 9.88030749132829e-06,
      "loss": 0.0052,
      "step": 50770
    },
    {
      "epoch": 0.3256871466458008,
      "grad_norm": 0.34293127059936523,
      "learning_rate": 9.880185728225017e-06,
      "loss": 0.0056,
      "step": 50780
    },
    {
      "epoch": 0.3257512835395869,
      "grad_norm": 0.20531544089317322,
      "learning_rate": 9.88006390396953e-06,
      "loss": 0.0053,
      "step": 50790
    },
    {
      "epoch": 0.32581542043337297,
      "grad_norm": 0.2079891413450241,
      "learning_rate": 9.879942018563357e-06,
      "loss": 0.0032,
      "step": 50800
    },
    {
      "epoch": 0.32587955732715906,
      "grad_norm": 0.5488170981407166,
      "learning_rate": 9.87982007200802e-06,
      "loss": 0.0056,
      "step": 50810
    },
    {
      "epoch": 0.3259436942209452,
      "grad_norm": 0.0822836309671402,
      "learning_rate": 9.879698064305052e-06,
      "loss": 0.0027,
      "step": 50820
    },
    {
      "epoch": 0.3260078311147313,
      "grad_norm": 0.12823008000850677,
      "learning_rate": 9.879575995455979e-06,
      "loss": 0.0036,
      "step": 50830
    },
    {
      "epoch": 0.3260719680085174,
      "grad_norm": 0.16132020950317383,
      "learning_rate": 9.87945386546233e-06,
      "loss": 0.0046,
      "step": 50840
    },
    {
      "epoch": 0.3261361049023035,
      "grad_norm": 0.14670330286026,
      "learning_rate": 9.879331674325638e-06,
      "loss": 0.0046,
      "step": 50850
    },
    {
      "epoch": 0.3262002417960896,
      "grad_norm": 0.10751719772815704,
      "learning_rate": 9.879209422047434e-06,
      "loss": 0.0056,
      "step": 50860
    },
    {
      "epoch": 0.3262643786898757,
      "grad_norm": 0.3112606108188629,
      "learning_rate": 9.87908710862925e-06,
      "loss": 0.0047,
      "step": 50870
    },
    {
      "epoch": 0.32632851558366177,
      "grad_norm": 0.2563963234424591,
      "learning_rate": 9.878964734072614e-06,
      "loss": 0.0056,
      "step": 50880
    },
    {
      "epoch": 0.32639265247744786,
      "grad_norm": 0.4979173243045807,
      "learning_rate": 9.878842298379063e-06,
      "loss": 0.0068,
      "step": 50890
    },
    {
      "epoch": 0.32645678937123396,
      "grad_norm": 0.08930937200784683,
      "learning_rate": 9.878719801550132e-06,
      "loss": 0.0034,
      "step": 50900
    },
    {
      "epoch": 0.32652092626502005,
      "grad_norm": 0.14936961233615875,
      "learning_rate": 9.878597243587356e-06,
      "loss": 0.0059,
      "step": 50910
    },
    {
      "epoch": 0.32658506315880614,
      "grad_norm": 0.1063743308186531,
      "learning_rate": 9.87847462449227e-06,
      "loss": 0.0041,
      "step": 50920
    },
    {
      "epoch": 0.32664920005259224,
      "grad_norm": 0.2172681838274002,
      "learning_rate": 9.878351944266408e-06,
      "loss": 0.0063,
      "step": 50930
    },
    {
      "epoch": 0.32671333694637833,
      "grad_norm": 0.27539241313934326,
      "learning_rate": 9.878229202911312e-06,
      "loss": 0.0055,
      "step": 50940
    },
    {
      "epoch": 0.3267774738401644,
      "grad_norm": 0.10550977289676666,
      "learning_rate": 9.878106400428518e-06,
      "loss": 0.0045,
      "step": 50950
    },
    {
      "epoch": 0.32684161073395057,
      "grad_norm": 0.1692620813846588,
      "learning_rate": 9.877983536819563e-06,
      "loss": 0.0036,
      "step": 50960
    },
    {
      "epoch": 0.32690574762773666,
      "grad_norm": 0.18884654343128204,
      "learning_rate": 9.877860612085987e-06,
      "loss": 0.0053,
      "step": 50970
    },
    {
      "epoch": 0.32696988452152276,
      "grad_norm": 0.225121408700943,
      "learning_rate": 9.877737626229334e-06,
      "loss": 0.0042,
      "step": 50980
    },
    {
      "epoch": 0.32703402141530885,
      "grad_norm": 0.0583873875439167,
      "learning_rate": 9.877614579251142e-06,
      "loss": 0.0029,
      "step": 50990
    },
    {
      "epoch": 0.32709815830909494,
      "grad_norm": 0.26959750056266785,
      "learning_rate": 9.877491471152953e-06,
      "loss": 0.0034,
      "step": 51000
    },
    {
      "epoch": 0.32716229520288104,
      "grad_norm": 0.31422290205955505,
      "learning_rate": 9.877368301936309e-06,
      "loss": 0.0054,
      "step": 51010
    },
    {
      "epoch": 0.32722643209666713,
      "grad_norm": 0.4650602340698242,
      "learning_rate": 9.877245071602755e-06,
      "loss": 0.0073,
      "step": 51020
    },
    {
      "epoch": 0.3272905689904532,
      "grad_norm": 0.1505156308412552,
      "learning_rate": 9.877121780153834e-06,
      "loss": 0.0056,
      "step": 51030
    },
    {
      "epoch": 0.3273547058842393,
      "grad_norm": 0.0990140438079834,
      "learning_rate": 9.876998427591092e-06,
      "loss": 0.0048,
      "step": 51040
    },
    {
      "epoch": 0.3274188427780254,
      "grad_norm": 0.14473867416381836,
      "learning_rate": 9.876875013916076e-06,
      "loss": 0.0054,
      "step": 51050
    },
    {
      "epoch": 0.3274829796718115,
      "grad_norm": 0.042040273547172546,
      "learning_rate": 9.876751539130328e-06,
      "loss": 0.0073,
      "step": 51060
    },
    {
      "epoch": 0.3275471165655976,
      "grad_norm": 0.3856167197227478,
      "learning_rate": 9.876628003235398e-06,
      "loss": 0.006,
      "step": 51070
    },
    {
      "epoch": 0.3276112534593837,
      "grad_norm": 0.1594998985528946,
      "learning_rate": 9.876504406232834e-06,
      "loss": 0.0048,
      "step": 51080
    },
    {
      "epoch": 0.3276753903531698,
      "grad_norm": 0.11787860840559006,
      "learning_rate": 9.876380748124186e-06,
      "loss": 0.0048,
      "step": 51090
    },
    {
      "epoch": 0.3277395272469559,
      "grad_norm": 0.3186417520046234,
      "learning_rate": 9.876257028911001e-06,
      "loss": 0.0042,
      "step": 51100
    },
    {
      "epoch": 0.327803664140742,
      "grad_norm": 0.15282252430915833,
      "learning_rate": 9.876133248594831e-06,
      "loss": 0.0071,
      "step": 51110
    },
    {
      "epoch": 0.3278678010345281,
      "grad_norm": 0.7064988017082214,
      "learning_rate": 9.876009407177226e-06,
      "loss": 0.0033,
      "step": 51120
    },
    {
      "epoch": 0.3279319379283142,
      "grad_norm": 0.4531448185443878,
      "learning_rate": 9.875885504659738e-06,
      "loss": 0.0055,
      "step": 51130
    },
    {
      "epoch": 0.3279960748221003,
      "grad_norm": 0.2045564502477646,
      "learning_rate": 9.875761541043921e-06,
      "loss": 0.0033,
      "step": 51140
    },
    {
      "epoch": 0.3280602117158864,
      "grad_norm": 0.43037521839141846,
      "learning_rate": 9.875637516331327e-06,
      "loss": 0.0064,
      "step": 51150
    },
    {
      "epoch": 0.3281243486096725,
      "grad_norm": 0.3135991096496582,
      "learning_rate": 9.875513430523509e-06,
      "loss": 0.0047,
      "step": 51160
    },
    {
      "epoch": 0.3281884855034586,
      "grad_norm": 0.19914768636226654,
      "learning_rate": 9.875389283622025e-06,
      "loss": 0.0097,
      "step": 51170
    },
    {
      "epoch": 0.3282526223972447,
      "grad_norm": 0.23053988814353943,
      "learning_rate": 9.875265075628429e-06,
      "loss": 0.0036,
      "step": 51180
    },
    {
      "epoch": 0.32831675929103077,
      "grad_norm": 0.15195836126804352,
      "learning_rate": 9.875140806544275e-06,
      "loss": 0.0036,
      "step": 51190
    },
    {
      "epoch": 0.32838089618481686,
      "grad_norm": 0.059653397649526596,
      "learning_rate": 9.875016476371124e-06,
      "loss": 0.0054,
      "step": 51200
    },
    {
      "epoch": 0.32844503307860295,
      "grad_norm": 0.062237270176410675,
      "learning_rate": 9.874892085110532e-06,
      "loss": 0.0056,
      "step": 51210
    },
    {
      "epoch": 0.32850916997238905,
      "grad_norm": 0.549043595790863,
      "learning_rate": 9.87476763276406e-06,
      "loss": 0.0041,
      "step": 51220
    },
    {
      "epoch": 0.32857330686617514,
      "grad_norm": 0.13534654676914215,
      "learning_rate": 9.874643119333262e-06,
      "loss": 0.0047,
      "step": 51230
    },
    {
      "epoch": 0.32863744375996123,
      "grad_norm": 0.31206753849983215,
      "learning_rate": 9.874518544819704e-06,
      "loss": 0.0081,
      "step": 51240
    },
    {
      "epoch": 0.3287015806537474,
      "grad_norm": 0.3598429560661316,
      "learning_rate": 9.874393909224946e-06,
      "loss": 0.0043,
      "step": 51250
    },
    {
      "epoch": 0.3287657175475335,
      "grad_norm": 0.2721366286277771,
      "learning_rate": 9.874269212550546e-06,
      "loss": 0.0044,
      "step": 51260
    },
    {
      "epoch": 0.32882985444131957,
      "grad_norm": 0.0643984004855156,
      "learning_rate": 9.87414445479807e-06,
      "loss": 0.0034,
      "step": 51270
    },
    {
      "epoch": 0.32889399133510566,
      "grad_norm": 0.4401140511035919,
      "learning_rate": 9.874019635969084e-06,
      "loss": 0.0106,
      "step": 51280
    },
    {
      "epoch": 0.32895812822889176,
      "grad_norm": 0.3248900771141052,
      "learning_rate": 9.873894756065145e-06,
      "loss": 0.005,
      "step": 51290
    },
    {
      "epoch": 0.32902226512267785,
      "grad_norm": 0.19265004992485046,
      "learning_rate": 9.873769815087822e-06,
      "loss": 0.0027,
      "step": 51300
    },
    {
      "epoch": 0.32908640201646394,
      "grad_norm": 0.14860068261623383,
      "learning_rate": 9.87364481303868e-06,
      "loss": 0.0039,
      "step": 51310
    },
    {
      "epoch": 0.32915053891025003,
      "grad_norm": 0.46626561880111694,
      "learning_rate": 9.873519749919285e-06,
      "loss": 0.01,
      "step": 51320
    },
    {
      "epoch": 0.3292146758040361,
      "grad_norm": 0.11976033449172974,
      "learning_rate": 9.873394625731205e-06,
      "loss": 0.0044,
      "step": 51330
    },
    {
      "epoch": 0.3292788126978222,
      "grad_norm": 0.09584904462099075,
      "learning_rate": 9.873269440476007e-06,
      "loss": 0.0049,
      "step": 51340
    },
    {
      "epoch": 0.3293429495916083,
      "grad_norm": 0.19121158123016357,
      "learning_rate": 9.87314419415526e-06,
      "loss": 0.0164,
      "step": 51350
    },
    {
      "epoch": 0.3294070864853944,
      "grad_norm": 0.16358615458011627,
      "learning_rate": 9.873018886770533e-06,
      "loss": 0.0049,
      "step": 51360
    },
    {
      "epoch": 0.3294712233791805,
      "grad_norm": 0.2830831706523895,
      "learning_rate": 9.872893518323398e-06,
      "loss": 0.0058,
      "step": 51370
    },
    {
      "epoch": 0.3295353602729666,
      "grad_norm": 0.07893689721822739,
      "learning_rate": 9.872768088815425e-06,
      "loss": 0.0042,
      "step": 51380
    },
    {
      "epoch": 0.32959949716675274,
      "grad_norm": 0.2660653591156006,
      "learning_rate": 9.872642598248184e-06,
      "loss": 0.0047,
      "step": 51390
    },
    {
      "epoch": 0.32966363406053883,
      "grad_norm": 0.08735953271389008,
      "learning_rate": 9.872517046623249e-06,
      "loss": 0.0031,
      "step": 51400
    },
    {
      "epoch": 0.32972777095432493,
      "grad_norm": 0.13710825145244598,
      "learning_rate": 9.872391433942191e-06,
      "loss": 0.0045,
      "step": 51410
    },
    {
      "epoch": 0.329791907848111,
      "grad_norm": 0.42409124970436096,
      "learning_rate": 9.87226576020659e-06,
      "loss": 0.0042,
      "step": 51420
    },
    {
      "epoch": 0.3298560447418971,
      "grad_norm": 0.18673092126846313,
      "learning_rate": 9.872140025418015e-06,
      "loss": 0.0041,
      "step": 51430
    },
    {
      "epoch": 0.3299201816356832,
      "grad_norm": 0.2237669676542282,
      "learning_rate": 9.872014229578044e-06,
      "loss": 0.0043,
      "step": 51440
    },
    {
      "epoch": 0.3299843185294693,
      "grad_norm": 0.2903503477573395,
      "learning_rate": 9.871888372688252e-06,
      "loss": 0.0029,
      "step": 51450
    },
    {
      "epoch": 0.3300484554232554,
      "grad_norm": 0.35877344012260437,
      "learning_rate": 9.871762454750218e-06,
      "loss": 0.004,
      "step": 51460
    },
    {
      "epoch": 0.3301125923170415,
      "grad_norm": 0.20692265033721924,
      "learning_rate": 9.871636475765517e-06,
      "loss": 0.0055,
      "step": 51470
    },
    {
      "epoch": 0.3301767292108276,
      "grad_norm": 0.19338805973529816,
      "learning_rate": 9.87151043573573e-06,
      "loss": 0.006,
      "step": 51480
    },
    {
      "epoch": 0.3302408661046137,
      "grad_norm": 0.2326938956975937,
      "learning_rate": 9.871384334662436e-06,
      "loss": 0.0042,
      "step": 51490
    },
    {
      "epoch": 0.33030500299839977,
      "grad_norm": 0.23102594912052155,
      "learning_rate": 9.871258172547215e-06,
      "loss": 0.0038,
      "step": 51500
    },
    {
      "epoch": 0.33036913989218586,
      "grad_norm": 0.2446339726448059,
      "learning_rate": 9.871131949391645e-06,
      "loss": 0.0041,
      "step": 51510
    },
    {
      "epoch": 0.33043327678597195,
      "grad_norm": 0.2786913812160492,
      "learning_rate": 9.871005665197313e-06,
      "loss": 0.0044,
      "step": 51520
    },
    {
      "epoch": 0.3304974136797581,
      "grad_norm": 0.12254165858030319,
      "learning_rate": 9.870879319965798e-06,
      "loss": 0.0046,
      "step": 51530
    },
    {
      "epoch": 0.3305615505735442,
      "grad_norm": 0.25559279322624207,
      "learning_rate": 9.870752913698685e-06,
      "loss": 0.0079,
      "step": 51540
    },
    {
      "epoch": 0.3306256874673303,
      "grad_norm": 0.3081286549568176,
      "learning_rate": 9.870626446397555e-06,
      "loss": 0.0055,
      "step": 51550
    },
    {
      "epoch": 0.3306898243611164,
      "grad_norm": 0.17108890414237976,
      "learning_rate": 9.870499918063996e-06,
      "loss": 0.0057,
      "step": 51560
    },
    {
      "epoch": 0.3307539612549025,
      "grad_norm": 0.45415928959846497,
      "learning_rate": 9.870373328699591e-06,
      "loss": 0.0044,
      "step": 51570
    },
    {
      "epoch": 0.33081809814868857,
      "grad_norm": 0.49789750576019287,
      "learning_rate": 9.87024667830593e-06,
      "loss": 0.0039,
      "step": 51580
    },
    {
      "epoch": 0.33088223504247466,
      "grad_norm": 0.2867346704006195,
      "learning_rate": 9.870119966884595e-06,
      "loss": 0.0053,
      "step": 51590
    },
    {
      "epoch": 0.33094637193626075,
      "grad_norm": 0.22337113320827484,
      "learning_rate": 9.869993194437177e-06,
      "loss": 0.0051,
      "step": 51600
    },
    {
      "epoch": 0.33101050883004685,
      "grad_norm": 0.18967677652835846,
      "learning_rate": 9.869866360965263e-06,
      "loss": 0.0032,
      "step": 51610
    },
    {
      "epoch": 0.33107464572383294,
      "grad_norm": 0.30326980352401733,
      "learning_rate": 9.869739466470444e-06,
      "loss": 0.0037,
      "step": 51620
    },
    {
      "epoch": 0.33113878261761903,
      "grad_norm": 0.13057635724544525,
      "learning_rate": 9.86961251095431e-06,
      "loss": 0.0045,
      "step": 51630
    },
    {
      "epoch": 0.3312029195114051,
      "grad_norm": 0.23110616207122803,
      "learning_rate": 9.869485494418449e-06,
      "loss": 0.0062,
      "step": 51640
    },
    {
      "epoch": 0.3312670564051912,
      "grad_norm": 0.22782118618488312,
      "learning_rate": 9.869358416864454e-06,
      "loss": 0.0052,
      "step": 51650
    },
    {
      "epoch": 0.3313311932989773,
      "grad_norm": 0.21024642884731293,
      "learning_rate": 9.86923127829392e-06,
      "loss": 0.0049,
      "step": 51660
    },
    {
      "epoch": 0.33139533019276346,
      "grad_norm": 0.13623197376728058,
      "learning_rate": 9.869104078708439e-06,
      "loss": 0.0055,
      "step": 51670
    },
    {
      "epoch": 0.33145946708654955,
      "grad_norm": 0.16198650002479553,
      "learning_rate": 9.868976818109602e-06,
      "loss": 0.0074,
      "step": 51680
    },
    {
      "epoch": 0.33152360398033565,
      "grad_norm": 0.24429954588413239,
      "learning_rate": 9.868849496499004e-06,
      "loss": 0.0048,
      "step": 51690
    },
    {
      "epoch": 0.33158774087412174,
      "grad_norm": 0.24430863559246063,
      "learning_rate": 9.868722113878245e-06,
      "loss": 0.0048,
      "step": 51700
    },
    {
      "epoch": 0.33165187776790783,
      "grad_norm": 0.1951116919517517,
      "learning_rate": 9.868594670248917e-06,
      "loss": 0.0048,
      "step": 51710
    },
    {
      "epoch": 0.3317160146616939,
      "grad_norm": 0.03462325781583786,
      "learning_rate": 9.868467165612619e-06,
      "loss": 0.0044,
      "step": 51720
    },
    {
      "epoch": 0.33178015155548,
      "grad_norm": 0.18571272492408752,
      "learning_rate": 9.868339599970948e-06,
      "loss": 0.0052,
      "step": 51730
    },
    {
      "epoch": 0.3318442884492661,
      "grad_norm": 0.2798001766204834,
      "learning_rate": 9.8682119733255e-06,
      "loss": 0.0042,
      "step": 51740
    },
    {
      "epoch": 0.3319084253430522,
      "grad_norm": 0.34794336557388306,
      "learning_rate": 9.86808428567788e-06,
      "loss": 0.0066,
      "step": 51750
    },
    {
      "epoch": 0.3319725622368383,
      "grad_norm": 0.06256379932165146,
      "learning_rate": 9.867956537029684e-06,
      "loss": 0.0054,
      "step": 51760
    },
    {
      "epoch": 0.3320366991306244,
      "grad_norm": 0.17388401925563812,
      "learning_rate": 9.867828727382514e-06,
      "loss": 0.0047,
      "step": 51770
    },
    {
      "epoch": 0.3321008360244105,
      "grad_norm": 0.15907210111618042,
      "learning_rate": 9.86770085673797e-06,
      "loss": 0.0034,
      "step": 51780
    },
    {
      "epoch": 0.3321649729181966,
      "grad_norm": 0.12382688373327255,
      "learning_rate": 9.867572925097655e-06,
      "loss": 0.0046,
      "step": 51790
    },
    {
      "epoch": 0.33222910981198267,
      "grad_norm": 0.37066325545310974,
      "learning_rate": 9.867444932463173e-06,
      "loss": 0.0057,
      "step": 51800
    },
    {
      "epoch": 0.3322932467057688,
      "grad_norm": 0.07459404319524765,
      "learning_rate": 9.867316878836126e-06,
      "loss": 0.0042,
      "step": 51810
    },
    {
      "epoch": 0.3323573835995549,
      "grad_norm": 0.2496977597475052,
      "learning_rate": 9.86718876421812e-06,
      "loss": 0.0068,
      "step": 51820
    },
    {
      "epoch": 0.332421520493341,
      "grad_norm": 0.1255044937133789,
      "learning_rate": 9.867060588610762e-06,
      "loss": 0.0058,
      "step": 51830
    },
    {
      "epoch": 0.3324856573871271,
      "grad_norm": 0.16789741814136505,
      "learning_rate": 9.866932352015657e-06,
      "loss": 0.004,
      "step": 51840
    },
    {
      "epoch": 0.3325497942809132,
      "grad_norm": 0.125722274184227,
      "learning_rate": 9.866804054434409e-06,
      "loss": 0.0037,
      "step": 51850
    },
    {
      "epoch": 0.3326139311746993,
      "grad_norm": 0.11827981472015381,
      "learning_rate": 9.86667569586863e-06,
      "loss": 0.0063,
      "step": 51860
    },
    {
      "epoch": 0.3326780680684854,
      "grad_norm": 0.29303354024887085,
      "learning_rate": 9.866547276319925e-06,
      "loss": 0.0072,
      "step": 51870
    },
    {
      "epoch": 0.33274220496227147,
      "grad_norm": 0.35147103667259216,
      "learning_rate": 9.866418795789905e-06,
      "loss": 0.004,
      "step": 51880
    },
    {
      "epoch": 0.33280634185605756,
      "grad_norm": 0.3894512355327606,
      "learning_rate": 9.86629025428018e-06,
      "loss": 0.0031,
      "step": 51890
    },
    {
      "epoch": 0.33287047874984366,
      "grad_norm": 0.17828503251075745,
      "learning_rate": 9.866161651792358e-06,
      "loss": 0.0042,
      "step": 51900
    },
    {
      "epoch": 0.33293461564362975,
      "grad_norm": 0.13376423716545105,
      "learning_rate": 9.866032988328056e-06,
      "loss": 0.004,
      "step": 51910
    },
    {
      "epoch": 0.33299875253741584,
      "grad_norm": 0.29704684019088745,
      "learning_rate": 9.865904263888882e-06,
      "loss": 0.0059,
      "step": 51920
    },
    {
      "epoch": 0.33306288943120194,
      "grad_norm": 0.16345566511154175,
      "learning_rate": 9.865775478476447e-06,
      "loss": 0.0054,
      "step": 51930
    },
    {
      "epoch": 0.33312702632498803,
      "grad_norm": 0.17470292747020721,
      "learning_rate": 9.86564663209237e-06,
      "loss": 0.0056,
      "step": 51940
    },
    {
      "epoch": 0.3331911632187742,
      "grad_norm": 0.14169193804264069,
      "learning_rate": 9.865517724738263e-06,
      "loss": 0.0044,
      "step": 51950
    },
    {
      "epoch": 0.33325530011256027,
      "grad_norm": 0.1908574253320694,
      "learning_rate": 9.865388756415744e-06,
      "loss": 0.0037,
      "step": 51960
    },
    {
      "epoch": 0.33331943700634636,
      "grad_norm": 0.13204747438430786,
      "learning_rate": 9.865259727126424e-06,
      "loss": 0.0023,
      "step": 51970
    },
    {
      "epoch": 0.33338357390013246,
      "grad_norm": 0.13022635877132416,
      "learning_rate": 9.865130636871922e-06,
      "loss": 0.0046,
      "step": 51980
    },
    {
      "epoch": 0.33344771079391855,
      "grad_norm": 0.30808025598526,
      "learning_rate": 9.865001485653858e-06,
      "loss": 0.0043,
      "step": 51990
    },
    {
      "epoch": 0.33351184768770464,
      "grad_norm": 0.26063185930252075,
      "learning_rate": 9.864872273473846e-06,
      "loss": 0.0045,
      "step": 52000
    },
    {
      "epoch": 0.33357598458149074,
      "grad_norm": 0.1141660287976265,
      "learning_rate": 9.86474300033351e-06,
      "loss": 0.0044,
      "step": 52010
    },
    {
      "epoch": 0.33364012147527683,
      "grad_norm": 0.07287078350782394,
      "learning_rate": 9.864613666234465e-06,
      "loss": 0.0041,
      "step": 52020
    },
    {
      "epoch": 0.3337042583690629,
      "grad_norm": 0.0985855832695961,
      "learning_rate": 9.864484271178336e-06,
      "loss": 0.0055,
      "step": 52030
    },
    {
      "epoch": 0.333768395262849,
      "grad_norm": 0.09198424965143204,
      "learning_rate": 9.864354815166744e-06,
      "loss": 0.0048,
      "step": 52040
    },
    {
      "epoch": 0.3338325321566351,
      "grad_norm": 0.12575598061084747,
      "learning_rate": 9.864225298201306e-06,
      "loss": 0.0041,
      "step": 52050
    },
    {
      "epoch": 0.3338966690504212,
      "grad_norm": 0.4537132978439331,
      "learning_rate": 9.864095720283651e-06,
      "loss": 0.0071,
      "step": 52060
    },
    {
      "epoch": 0.3339608059442073,
      "grad_norm": 0.21206165850162506,
      "learning_rate": 9.863966081415399e-06,
      "loss": 0.0023,
      "step": 52070
    },
    {
      "epoch": 0.3340249428379934,
      "grad_norm": 0.19575679302215576,
      "learning_rate": 9.863836381598176e-06,
      "loss": 0.0029,
      "step": 52080
    },
    {
      "epoch": 0.33408907973177954,
      "grad_norm": 0.06109379231929779,
      "learning_rate": 9.863706620833608e-06,
      "loss": 0.0037,
      "step": 52090
    },
    {
      "epoch": 0.33415321662556563,
      "grad_norm": 0.18857845664024353,
      "learning_rate": 9.86357679912332e-06,
      "loss": 0.0039,
      "step": 52100
    },
    {
      "epoch": 0.3342173535193517,
      "grad_norm": 0.2904978096485138,
      "learning_rate": 9.863446916468938e-06,
      "loss": 0.0064,
      "step": 52110
    },
    {
      "epoch": 0.3342814904131378,
      "grad_norm": 0.3255593180656433,
      "learning_rate": 9.86331697287209e-06,
      "loss": 0.005,
      "step": 52120
    },
    {
      "epoch": 0.3343456273069239,
      "grad_norm": 0.19768422842025757,
      "learning_rate": 9.863186968334404e-06,
      "loss": 0.0037,
      "step": 52130
    },
    {
      "epoch": 0.33440976420071,
      "grad_norm": 0.06513660401105881,
      "learning_rate": 9.863056902857511e-06,
      "loss": 0.0038,
      "step": 52140
    },
    {
      "epoch": 0.3344739010944961,
      "grad_norm": 0.408059298992157,
      "learning_rate": 9.862926776443039e-06,
      "loss": 0.0077,
      "step": 52150
    },
    {
      "epoch": 0.3345380379882822,
      "grad_norm": 0.17330044507980347,
      "learning_rate": 9.86279658909262e-06,
      "loss": 0.0032,
      "step": 52160
    },
    {
      "epoch": 0.3346021748820683,
      "grad_norm": 0.17973282933235168,
      "learning_rate": 9.862666340807882e-06,
      "loss": 0.0065,
      "step": 52170
    },
    {
      "epoch": 0.3346663117758544,
      "grad_norm": 0.17612335085868835,
      "learning_rate": 9.862536031590462e-06,
      "loss": 0.0059,
      "step": 52180
    },
    {
      "epoch": 0.33473044866964047,
      "grad_norm": 0.19796328246593475,
      "learning_rate": 9.862405661441988e-06,
      "loss": 0.0043,
      "step": 52190
    },
    {
      "epoch": 0.33479458556342656,
      "grad_norm": 0.1836937516927719,
      "learning_rate": 9.862275230364099e-06,
      "loss": 0.0049,
      "step": 52200
    },
    {
      "epoch": 0.33485872245721265,
      "grad_norm": 0.35114187002182007,
      "learning_rate": 9.862144738358424e-06,
      "loss": 0.0059,
      "step": 52210
    },
    {
      "epoch": 0.33492285935099875,
      "grad_norm": 0.3235044777393341,
      "learning_rate": 9.862014185426601e-06,
      "loss": 0.004,
      "step": 52220
    },
    {
      "epoch": 0.3349869962447849,
      "grad_norm": 0.25618669390678406,
      "learning_rate": 9.861883571570265e-06,
      "loss": 0.0039,
      "step": 52230
    },
    {
      "epoch": 0.335051133138571,
      "grad_norm": 0.16893281042575836,
      "learning_rate": 9.861752896791052e-06,
      "loss": 0.0056,
      "step": 52240
    },
    {
      "epoch": 0.3351152700323571,
      "grad_norm": 0.05286022275686264,
      "learning_rate": 9.861622161090602e-06,
      "loss": 0.007,
      "step": 52250
    },
    {
      "epoch": 0.3351794069261432,
      "grad_norm": 0.3475116491317749,
      "learning_rate": 9.861491364470553e-06,
      "loss": 0.0046,
      "step": 52260
    },
    {
      "epoch": 0.33524354381992927,
      "grad_norm": 0.021251840516924858,
      "learning_rate": 9.861360506932542e-06,
      "loss": 0.0031,
      "step": 52270
    },
    {
      "epoch": 0.33530768071371536,
      "grad_norm": 0.05894012749195099,
      "learning_rate": 9.861229588478208e-06,
      "loss": 0.0027,
      "step": 52280
    },
    {
      "epoch": 0.33537181760750145,
      "grad_norm": 0.32902398705482483,
      "learning_rate": 9.861098609109194e-06,
      "loss": 0.0033,
      "step": 52290
    },
    {
      "epoch": 0.33543595450128755,
      "grad_norm": 0.14335854351520538,
      "learning_rate": 9.86096756882714e-06,
      "loss": 0.0036,
      "step": 52300
    },
    {
      "epoch": 0.33550009139507364,
      "grad_norm": 0.13611845672130585,
      "learning_rate": 9.86083646763369e-06,
      "loss": 0.0036,
      "step": 52310
    },
    {
      "epoch": 0.33556422828885973,
      "grad_norm": 0.13622123003005981,
      "learning_rate": 9.860705305530482e-06,
      "loss": 0.0039,
      "step": 52320
    },
    {
      "epoch": 0.3356283651826458,
      "grad_norm": 0.13387630879878998,
      "learning_rate": 9.860574082519165e-06,
      "loss": 0.0043,
      "step": 52330
    },
    {
      "epoch": 0.3356925020764319,
      "grad_norm": 0.7947002649307251,
      "learning_rate": 9.86044279860138e-06,
      "loss": 0.0046,
      "step": 52340
    },
    {
      "epoch": 0.335756638970218,
      "grad_norm": 0.23654408752918243,
      "learning_rate": 9.860311453778773e-06,
      "loss": 0.0042,
      "step": 52350
    },
    {
      "epoch": 0.3358207758640041,
      "grad_norm": 0.1772829294204712,
      "learning_rate": 9.86018004805299e-06,
      "loss": 0.0043,
      "step": 52360
    },
    {
      "epoch": 0.33588491275779025,
      "grad_norm": 0.5392045378684998,
      "learning_rate": 9.860048581425679e-06,
      "loss": 0.0036,
      "step": 52370
    },
    {
      "epoch": 0.33594904965157635,
      "grad_norm": 0.13509030640125275,
      "learning_rate": 9.859917053898485e-06,
      "loss": 0.0061,
      "step": 52380
    },
    {
      "epoch": 0.33601318654536244,
      "grad_norm": 0.11849337071180344,
      "learning_rate": 9.859785465473055e-06,
      "loss": 0.0073,
      "step": 52390
    },
    {
      "epoch": 0.33607732343914853,
      "grad_norm": 0.42279133200645447,
      "learning_rate": 9.859653816151042e-06,
      "loss": 0.0054,
      "step": 52400
    },
    {
      "epoch": 0.3361414603329346,
      "grad_norm": 0.10719570517539978,
      "learning_rate": 9.859522105934092e-06,
      "loss": 0.004,
      "step": 52410
    },
    {
      "epoch": 0.3362055972267207,
      "grad_norm": 0.20735955238342285,
      "learning_rate": 9.859390334823859e-06,
      "loss": 0.004,
      "step": 52420
    },
    {
      "epoch": 0.3362697341205068,
      "grad_norm": 0.19455531239509583,
      "learning_rate": 9.85925850282199e-06,
      "loss": 0.0048,
      "step": 52430
    },
    {
      "epoch": 0.3363338710142929,
      "grad_norm": 0.36882632970809937,
      "learning_rate": 9.859126609930141e-06,
      "loss": 0.0046,
      "step": 52440
    },
    {
      "epoch": 0.336398007908079,
      "grad_norm": 0.0744689404964447,
      "learning_rate": 9.858994656149961e-06,
      "loss": 0.0059,
      "step": 52450
    },
    {
      "epoch": 0.3364621448018651,
      "grad_norm": 0.18131551146507263,
      "learning_rate": 9.858862641483107e-06,
      "loss": 0.0043,
      "step": 52460
    },
    {
      "epoch": 0.3365262816956512,
      "grad_norm": 0.12779557704925537,
      "learning_rate": 9.85873056593123e-06,
      "loss": 0.0019,
      "step": 52470
    },
    {
      "epoch": 0.3365904185894373,
      "grad_norm": 0.21120983362197876,
      "learning_rate": 9.858598429495986e-06,
      "loss": 0.0047,
      "step": 52480
    },
    {
      "epoch": 0.33665455548322337,
      "grad_norm": 0.36974161863327026,
      "learning_rate": 9.858466232179033e-06,
      "loss": 0.0028,
      "step": 52490
    },
    {
      "epoch": 0.33671869237700947,
      "grad_norm": 0.10227707773447037,
      "learning_rate": 9.858333973982026e-06,
      "loss": 0.0032,
      "step": 52500
    },
    {
      "epoch": 0.33678282927079556,
      "grad_norm": 0.10719114542007446,
      "learning_rate": 9.858201654906621e-06,
      "loss": 0.0063,
      "step": 52510
    },
    {
      "epoch": 0.3368469661645817,
      "grad_norm": 0.22539356350898743,
      "learning_rate": 9.858069274954479e-06,
      "loss": 0.0049,
      "step": 52520
    },
    {
      "epoch": 0.3369111030583678,
      "grad_norm": 0.28657108545303345,
      "learning_rate": 9.857936834127255e-06,
      "loss": 0.0043,
      "step": 52530
    },
    {
      "epoch": 0.3369752399521539,
      "grad_norm": 0.14741523563861847,
      "learning_rate": 9.85780433242661e-06,
      "loss": 0.0054,
      "step": 52540
    },
    {
      "epoch": 0.33703937684594,
      "grad_norm": 0.18382100760936737,
      "learning_rate": 9.857671769854207e-06,
      "loss": 0.0058,
      "step": 52550
    },
    {
      "epoch": 0.3371035137397261,
      "grad_norm": 0.1844129115343094,
      "learning_rate": 9.857539146411703e-06,
      "loss": 0.004,
      "step": 52560
    },
    {
      "epoch": 0.3371676506335122,
      "grad_norm": 0.3120049238204956,
      "learning_rate": 9.857406462100764e-06,
      "loss": 0.0053,
      "step": 52570
    },
    {
      "epoch": 0.33723178752729827,
      "grad_norm": 0.13739828765392303,
      "learning_rate": 9.85727371692305e-06,
      "loss": 0.0063,
      "step": 52580
    },
    {
      "epoch": 0.33729592442108436,
      "grad_norm": 0.13995586335659027,
      "learning_rate": 9.857140910880223e-06,
      "loss": 0.004,
      "step": 52590
    },
    {
      "epoch": 0.33736006131487045,
      "grad_norm": 0.27522650361061096,
      "learning_rate": 9.85700804397395e-06,
      "loss": 0.0054,
      "step": 52600
    },
    {
      "epoch": 0.33742419820865655,
      "grad_norm": 0.2531697750091553,
      "learning_rate": 9.856875116205895e-06,
      "loss": 0.0052,
      "step": 52610
    },
    {
      "epoch": 0.33748833510244264,
      "grad_norm": 0.08868460357189178,
      "learning_rate": 9.856742127577726e-06,
      "loss": 0.0052,
      "step": 52620
    },
    {
      "epoch": 0.33755247199622873,
      "grad_norm": 0.13962849974632263,
      "learning_rate": 9.856609078091104e-06,
      "loss": 0.0047,
      "step": 52630
    },
    {
      "epoch": 0.3376166088900148,
      "grad_norm": 0.2872208058834076,
      "learning_rate": 9.8564759677477e-06,
      "loss": 0.0059,
      "step": 52640
    },
    {
      "epoch": 0.3376807457838009,
      "grad_norm": 0.22213566303253174,
      "learning_rate": 9.856342796549181e-06,
      "loss": 0.0042,
      "step": 52650
    },
    {
      "epoch": 0.33774488267758707,
      "grad_norm": 0.3229414224624634,
      "learning_rate": 9.856209564497217e-06,
      "loss": 0.0048,
      "step": 52660
    },
    {
      "epoch": 0.33780901957137316,
      "grad_norm": 0.17573681473731995,
      "learning_rate": 9.856076271593476e-06,
      "loss": 0.0036,
      "step": 52670
    },
    {
      "epoch": 0.33787315646515925,
      "grad_norm": 0.09097541868686676,
      "learning_rate": 9.85594291783963e-06,
      "loss": 0.0038,
      "step": 52680
    },
    {
      "epoch": 0.33793729335894535,
      "grad_norm": 0.07667984813451767,
      "learning_rate": 9.855809503237345e-06,
      "loss": 0.0027,
      "step": 52690
    },
    {
      "epoch": 0.33800143025273144,
      "grad_norm": 0.14468225836753845,
      "learning_rate": 9.8556760277883e-06,
      "loss": 0.0033,
      "step": 52700
    },
    {
      "epoch": 0.33806556714651753,
      "grad_norm": 0.1518298238515854,
      "learning_rate": 9.855542491494163e-06,
      "loss": 0.0081,
      "step": 52710
    },
    {
      "epoch": 0.3381297040403036,
      "grad_norm": 0.13446344435214996,
      "learning_rate": 9.855408894356608e-06,
      "loss": 0.0035,
      "step": 52720
    },
    {
      "epoch": 0.3381938409340897,
      "grad_norm": 0.279994934797287,
      "learning_rate": 9.85527523637731e-06,
      "loss": 0.007,
      "step": 52730
    },
    {
      "epoch": 0.3382579778278758,
      "grad_norm": 0.12073291093111038,
      "learning_rate": 9.855141517557944e-06,
      "loss": 0.0037,
      "step": 52740
    },
    {
      "epoch": 0.3383221147216619,
      "grad_norm": 0.11250226199626923,
      "learning_rate": 9.855007737900184e-06,
      "loss": 0.0035,
      "step": 52750
    },
    {
      "epoch": 0.338386251615448,
      "grad_norm": 0.38413527607917786,
      "learning_rate": 9.854873897405708e-06,
      "loss": 0.0067,
      "step": 52760
    },
    {
      "epoch": 0.3384503885092341,
      "grad_norm": 0.14131279289722443,
      "learning_rate": 9.85473999607619e-06,
      "loss": 0.0046,
      "step": 52770
    },
    {
      "epoch": 0.3385145254030202,
      "grad_norm": 0.23225539922714233,
      "learning_rate": 9.854606033913312e-06,
      "loss": 0.0036,
      "step": 52780
    },
    {
      "epoch": 0.3385786622968063,
      "grad_norm": 0.15460790693759918,
      "learning_rate": 9.854472010918751e-06,
      "loss": 0.0131,
      "step": 52790
    },
    {
      "epoch": 0.3386427991905924,
      "grad_norm": 0.1357378214597702,
      "learning_rate": 9.854337927094186e-06,
      "loss": 0.004,
      "step": 52800
    },
    {
      "epoch": 0.3387069360843785,
      "grad_norm": 0.11922430992126465,
      "learning_rate": 9.854203782441299e-06,
      "loss": 0.0097,
      "step": 52810
    },
    {
      "epoch": 0.3387710729781646,
      "grad_norm": 0.35543474555015564,
      "learning_rate": 9.854069576961767e-06,
      "loss": 0.0051,
      "step": 52820
    },
    {
      "epoch": 0.3388352098719507,
      "grad_norm": 0.125699982047081,
      "learning_rate": 9.853935310657275e-06,
      "loss": 0.0048,
      "step": 52830
    },
    {
      "epoch": 0.3388993467657368,
      "grad_norm": 0.0511971153318882,
      "learning_rate": 9.853800983529503e-06,
      "loss": 0.0032,
      "step": 52840
    },
    {
      "epoch": 0.3389634836595229,
      "grad_norm": 0.20448780059814453,
      "learning_rate": 9.853666595580138e-06,
      "loss": 0.0057,
      "step": 52850
    },
    {
      "epoch": 0.339027620553309,
      "grad_norm": 0.15460821986198425,
      "learning_rate": 9.85353214681086e-06,
      "loss": 0.0034,
      "step": 52860
    },
    {
      "epoch": 0.3390917574470951,
      "grad_norm": 0.26127907633781433,
      "learning_rate": 9.853397637223357e-06,
      "loss": 0.0036,
      "step": 52870
    },
    {
      "epoch": 0.33915589434088117,
      "grad_norm": 0.16217119991779327,
      "learning_rate": 9.853263066819312e-06,
      "loss": 0.0051,
      "step": 52880
    },
    {
      "epoch": 0.33922003123466726,
      "grad_norm": 0.1911216527223587,
      "learning_rate": 9.853128435600412e-06,
      "loss": 0.0048,
      "step": 52890
    },
    {
      "epoch": 0.33928416812845336,
      "grad_norm": 0.21904310584068298,
      "learning_rate": 9.852993743568345e-06,
      "loss": 0.0081,
      "step": 52900
    },
    {
      "epoch": 0.33934830502223945,
      "grad_norm": 0.21412821114063263,
      "learning_rate": 9.852858990724796e-06,
      "loss": 0.0038,
      "step": 52910
    },
    {
      "epoch": 0.33941244191602554,
      "grad_norm": 0.198672816157341,
      "learning_rate": 9.852724177071457e-06,
      "loss": 0.003,
      "step": 52920
    },
    {
      "epoch": 0.33947657880981164,
      "grad_norm": 0.3219127953052521,
      "learning_rate": 9.852589302610016e-06,
      "loss": 0.0059,
      "step": 52930
    },
    {
      "epoch": 0.3395407157035978,
      "grad_norm": 0.15994399785995483,
      "learning_rate": 9.852454367342163e-06,
      "loss": 0.0049,
      "step": 52940
    },
    {
      "epoch": 0.3396048525973839,
      "grad_norm": 0.5959430932998657,
      "learning_rate": 9.85231937126959e-06,
      "loss": 0.0039,
      "step": 52950
    },
    {
      "epoch": 0.33966898949116997,
      "grad_norm": 0.225297749042511,
      "learning_rate": 9.852184314393985e-06,
      "loss": 0.0041,
      "step": 52960
    },
    {
      "epoch": 0.33973312638495606,
      "grad_norm": 0.23923683166503906,
      "learning_rate": 9.852049196717045e-06,
      "loss": 0.0057,
      "step": 52970
    },
    {
      "epoch": 0.33979726327874216,
      "grad_norm": 0.41412973403930664,
      "learning_rate": 9.851914018240458e-06,
      "loss": 0.0058,
      "step": 52980
    },
    {
      "epoch": 0.33986140017252825,
      "grad_norm": 0.1495286226272583,
      "learning_rate": 9.851778778965923e-06,
      "loss": 0.0044,
      "step": 52990
    },
    {
      "epoch": 0.33992553706631434,
      "grad_norm": 0.24865113198757172,
      "learning_rate": 9.851643478895132e-06,
      "loss": 0.0073,
      "step": 53000
    },
    {
      "epoch": 0.33998967396010044,
      "grad_norm": 0.06629683077335358,
      "learning_rate": 9.85150811802978e-06,
      "loss": 0.0025,
      "step": 53010
    },
    {
      "epoch": 0.34005381085388653,
      "grad_norm": 0.16571450233459473,
      "learning_rate": 9.851372696371563e-06,
      "loss": 0.0055,
      "step": 53020
    },
    {
      "epoch": 0.3401179477476726,
      "grad_norm": 0.11346513032913208,
      "learning_rate": 9.851237213922182e-06,
      "loss": 0.0038,
      "step": 53030
    },
    {
      "epoch": 0.3401820846414587,
      "grad_norm": 0.17872054874897003,
      "learning_rate": 9.85110167068333e-06,
      "loss": 0.0059,
      "step": 53040
    },
    {
      "epoch": 0.3402462215352448,
      "grad_norm": 0.17010082304477692,
      "learning_rate": 9.850966066656707e-06,
      "loss": 0.0029,
      "step": 53050
    },
    {
      "epoch": 0.3403103584290309,
      "grad_norm": 0.249822735786438,
      "learning_rate": 9.850830401844012e-06,
      "loss": 0.0037,
      "step": 53060
    },
    {
      "epoch": 0.340374495322817,
      "grad_norm": 0.5932525992393494,
      "learning_rate": 9.850694676246945e-06,
      "loss": 0.0063,
      "step": 53070
    },
    {
      "epoch": 0.34043863221660314,
      "grad_norm": 0.24662402272224426,
      "learning_rate": 9.850558889867207e-06,
      "loss": 0.0033,
      "step": 53080
    },
    {
      "epoch": 0.34050276911038924,
      "grad_norm": 0.2894830107688904,
      "learning_rate": 9.850423042706501e-06,
      "loss": 0.0052,
      "step": 53090
    },
    {
      "epoch": 0.34056690600417533,
      "grad_norm": 0.11228039860725403,
      "learning_rate": 9.850287134766527e-06,
      "loss": 0.0061,
      "step": 53100
    },
    {
      "epoch": 0.3406310428979614,
      "grad_norm": 0.23333176970481873,
      "learning_rate": 9.850151166048988e-06,
      "loss": 0.0039,
      "step": 53110
    },
    {
      "epoch": 0.3406951797917475,
      "grad_norm": 0.12623150646686554,
      "learning_rate": 9.850015136555589e-06,
      "loss": 0.0047,
      "step": 53120
    },
    {
      "epoch": 0.3407593166855336,
      "grad_norm": 0.31074294447898865,
      "learning_rate": 9.849879046288033e-06,
      "loss": 0.0039,
      "step": 53130
    },
    {
      "epoch": 0.3408234535793197,
      "grad_norm": 0.1798170506954193,
      "learning_rate": 9.84974289524803e-06,
      "loss": 0.0029,
      "step": 53140
    },
    {
      "epoch": 0.3408875904731058,
      "grad_norm": 0.08633461594581604,
      "learning_rate": 9.849606683437278e-06,
      "loss": 0.0047,
      "step": 53150
    },
    {
      "epoch": 0.3409517273668919,
      "grad_norm": 0.28256675601005554,
      "learning_rate": 9.84947041085749e-06,
      "loss": 0.0031,
      "step": 53160
    },
    {
      "epoch": 0.341015864260678,
      "grad_norm": 0.2820548713207245,
      "learning_rate": 9.849334077510373e-06,
      "loss": 0.0037,
      "step": 53170
    },
    {
      "epoch": 0.3410800011544641,
      "grad_norm": 0.1605130136013031,
      "learning_rate": 9.849197683397633e-06,
      "loss": 0.004,
      "step": 53180
    },
    {
      "epoch": 0.34114413804825017,
      "grad_norm": 0.1496482938528061,
      "learning_rate": 9.84906122852098e-06,
      "loss": 0.0058,
      "step": 53190
    },
    {
      "epoch": 0.34120827494203626,
      "grad_norm": 0.14850284159183502,
      "learning_rate": 9.848924712882126e-06,
      "loss": 0.0047,
      "step": 53200
    },
    {
      "epoch": 0.34127241183582235,
      "grad_norm": 0.3548075258731842,
      "learning_rate": 9.848788136482778e-06,
      "loss": 0.0032,
      "step": 53210
    },
    {
      "epoch": 0.3413365487296085,
      "grad_norm": 0.456371545791626,
      "learning_rate": 9.84865149932465e-06,
      "loss": 0.0036,
      "step": 53220
    },
    {
      "epoch": 0.3414006856233946,
      "grad_norm": 0.1298443228006363,
      "learning_rate": 9.848514801409454e-06,
      "loss": 0.0063,
      "step": 53230
    },
    {
      "epoch": 0.3414648225171807,
      "grad_norm": 0.21353577077388763,
      "learning_rate": 9.848378042738903e-06,
      "loss": 0.0024,
      "step": 53240
    },
    {
      "epoch": 0.3415289594109668,
      "grad_norm": 0.18850825726985931,
      "learning_rate": 9.848241223314708e-06,
      "loss": 0.0033,
      "step": 53250
    },
    {
      "epoch": 0.3415930963047529,
      "grad_norm": 0.312389612197876,
      "learning_rate": 9.848104343138586e-06,
      "loss": 0.0051,
      "step": 53260
    },
    {
      "epoch": 0.34165723319853897,
      "grad_norm": 0.19968365132808685,
      "learning_rate": 9.847967402212253e-06,
      "loss": 0.0073,
      "step": 53270
    },
    {
      "epoch": 0.34172137009232506,
      "grad_norm": 0.4982510209083557,
      "learning_rate": 9.847830400537421e-06,
      "loss": 0.0045,
      "step": 53280
    },
    {
      "epoch": 0.34178550698611115,
      "grad_norm": 0.3132604658603668,
      "learning_rate": 9.847693338115811e-06,
      "loss": 0.0036,
      "step": 53290
    },
    {
      "epoch": 0.34184964387989725,
      "grad_norm": 0.07934847474098206,
      "learning_rate": 9.84755621494914e-06,
      "loss": 0.0048,
      "step": 53300
    },
    {
      "epoch": 0.34191378077368334,
      "grad_norm": 0.03799527883529663,
      "learning_rate": 9.847419031039125e-06,
      "loss": 0.0044,
      "step": 53310
    },
    {
      "epoch": 0.34197791766746943,
      "grad_norm": 0.14706788957118988,
      "learning_rate": 9.847281786387483e-06,
      "loss": 0.0051,
      "step": 53320
    },
    {
      "epoch": 0.3420420545612555,
      "grad_norm": 0.2135099470615387,
      "learning_rate": 9.847144480995938e-06,
      "loss": 0.0038,
      "step": 53330
    },
    {
      "epoch": 0.3421061914550416,
      "grad_norm": 0.17504918575286865,
      "learning_rate": 9.847007114866207e-06,
      "loss": 0.0045,
      "step": 53340
    },
    {
      "epoch": 0.3421703283488277,
      "grad_norm": 0.16393157839775085,
      "learning_rate": 9.846869688000013e-06,
      "loss": 0.0036,
      "step": 53350
    },
    {
      "epoch": 0.34223446524261386,
      "grad_norm": 0.24033233523368835,
      "learning_rate": 9.846732200399079e-06,
      "loss": 0.0071,
      "step": 53360
    },
    {
      "epoch": 0.34229860213639995,
      "grad_norm": 0.32101425528526306,
      "learning_rate": 9.846594652065126e-06,
      "loss": 0.0051,
      "step": 53370
    },
    {
      "epoch": 0.34236273903018605,
      "grad_norm": 0.1657356470823288,
      "learning_rate": 9.846457042999877e-06,
      "loss": 0.0066,
      "step": 53380
    },
    {
      "epoch": 0.34242687592397214,
      "grad_norm": 0.2051573246717453,
      "learning_rate": 9.846319373205059e-06,
      "loss": 0.0037,
      "step": 53390
    },
    {
      "epoch": 0.34249101281775823,
      "grad_norm": 0.38923710584640503,
      "learning_rate": 9.846181642682395e-06,
      "loss": 0.0056,
      "step": 53400
    },
    {
      "epoch": 0.3425551497115443,
      "grad_norm": 0.20260879397392273,
      "learning_rate": 9.846043851433612e-06,
      "loss": 0.0049,
      "step": 53410
    },
    {
      "epoch": 0.3426192866053304,
      "grad_norm": 0.13483817875385284,
      "learning_rate": 9.845905999460436e-06,
      "loss": 0.0046,
      "step": 53420
    },
    {
      "epoch": 0.3426834234991165,
      "grad_norm": 0.49794188141822815,
      "learning_rate": 9.845768086764594e-06,
      "loss": 0.0038,
      "step": 53430
    },
    {
      "epoch": 0.3427475603929026,
      "grad_norm": 0.5910177826881409,
      "learning_rate": 9.845630113347814e-06,
      "loss": 0.007,
      "step": 53440
    },
    {
      "epoch": 0.3428116972866887,
      "grad_norm": 0.14928394556045532,
      "learning_rate": 9.845492079211827e-06,
      "loss": 0.0044,
      "step": 53450
    },
    {
      "epoch": 0.3428758341804748,
      "grad_norm": 0.18685732781887054,
      "learning_rate": 9.84535398435836e-06,
      "loss": 0.0031,
      "step": 53460
    },
    {
      "epoch": 0.3429399710742609,
      "grad_norm": 0.20124071836471558,
      "learning_rate": 9.845215828789148e-06,
      "loss": 0.004,
      "step": 53470
    },
    {
      "epoch": 0.343004107968047,
      "grad_norm": 0.22984161972999573,
      "learning_rate": 9.845077612505916e-06,
      "loss": 0.0034,
      "step": 53480
    },
    {
      "epoch": 0.34306824486183307,
      "grad_norm": 0.35153457522392273,
      "learning_rate": 9.844939335510397e-06,
      "loss": 0.0046,
      "step": 53490
    },
    {
      "epoch": 0.3431323817556192,
      "grad_norm": 0.16015851497650146,
      "learning_rate": 9.844800997804328e-06,
      "loss": 0.0112,
      "step": 53500
    },
    {
      "epoch": 0.3431965186494053,
      "grad_norm": 0.09807637333869934,
      "learning_rate": 9.844662599389442e-06,
      "loss": 0.0051,
      "step": 53510
    },
    {
      "epoch": 0.3432606555431914,
      "grad_norm": 0.2192123383283615,
      "learning_rate": 9.844524140267466e-06,
      "loss": 0.0038,
      "step": 53520
    },
    {
      "epoch": 0.3433247924369775,
      "grad_norm": 0.3703880310058594,
      "learning_rate": 9.844385620440144e-06,
      "loss": 0.0041,
      "step": 53530
    },
    {
      "epoch": 0.3433889293307636,
      "grad_norm": 0.29409804940223694,
      "learning_rate": 9.844247039909207e-06,
      "loss": 0.0041,
      "step": 53540
    },
    {
      "epoch": 0.3434530662245497,
      "grad_norm": 0.18779346346855164,
      "learning_rate": 9.844108398676392e-06,
      "loss": 0.0066,
      "step": 53550
    },
    {
      "epoch": 0.3435172031183358,
      "grad_norm": 0.16231729090213776,
      "learning_rate": 9.843969696743437e-06,
      "loss": 0.008,
      "step": 53560
    },
    {
      "epoch": 0.34358134001212187,
      "grad_norm": 0.16454057395458221,
      "learning_rate": 9.843830934112079e-06,
      "loss": 0.0076,
      "step": 53570
    },
    {
      "epoch": 0.34364547690590797,
      "grad_norm": 0.24449777603149414,
      "learning_rate": 9.84369211078406e-06,
      "loss": 0.0051,
      "step": 53580
    },
    {
      "epoch": 0.34370961379969406,
      "grad_norm": 0.14895787835121155,
      "learning_rate": 9.843553226761115e-06,
      "loss": 0.004,
      "step": 53590
    },
    {
      "epoch": 0.34377375069348015,
      "grad_norm": 0.025796731933951378,
      "learning_rate": 9.843414282044988e-06,
      "loss": 0.0024,
      "step": 53600
    },
    {
      "epoch": 0.34383788758726624,
      "grad_norm": 0.6771606206893921,
      "learning_rate": 9.843275276637416e-06,
      "loss": 0.0025,
      "step": 53610
    },
    {
      "epoch": 0.34390202448105234,
      "grad_norm": 0.20201508700847626,
      "learning_rate": 9.843136210540145e-06,
      "loss": 0.0033,
      "step": 53620
    },
    {
      "epoch": 0.34396616137483843,
      "grad_norm": 0.2704976499080658,
      "learning_rate": 9.842997083754915e-06,
      "loss": 0.0045,
      "step": 53630
    },
    {
      "epoch": 0.3440302982686246,
      "grad_norm": 0.18581216037273407,
      "learning_rate": 9.842857896283472e-06,
      "loss": 0.0041,
      "step": 53640
    },
    {
      "epoch": 0.3440944351624107,
      "grad_norm": 0.19817693531513214,
      "learning_rate": 9.842718648127557e-06,
      "loss": 0.0039,
      "step": 53650
    },
    {
      "epoch": 0.34415857205619677,
      "grad_norm": 0.0884748250246048,
      "learning_rate": 9.842579339288917e-06,
      "loss": 0.0074,
      "step": 53660
    },
    {
      "epoch": 0.34422270894998286,
      "grad_norm": 0.18690407276153564,
      "learning_rate": 9.842439969769298e-06,
      "loss": 0.0071,
      "step": 53670
    },
    {
      "epoch": 0.34428684584376895,
      "grad_norm": 0.4473888874053955,
      "learning_rate": 9.842300539570442e-06,
      "loss": 0.005,
      "step": 53680
    },
    {
      "epoch": 0.34435098273755504,
      "grad_norm": 0.14846082031726837,
      "learning_rate": 9.842161048694102e-06,
      "loss": 0.006,
      "step": 53690
    },
    {
      "epoch": 0.34441511963134114,
      "grad_norm": 0.20833083987236023,
      "learning_rate": 9.842021497142023e-06,
      "loss": 0.0042,
      "step": 53700
    },
    {
      "epoch": 0.34447925652512723,
      "grad_norm": 0.18778206408023834,
      "learning_rate": 9.841881884915953e-06,
      "loss": 0.0036,
      "step": 53710
    },
    {
      "epoch": 0.3445433934189133,
      "grad_norm": 0.4774073660373688,
      "learning_rate": 9.841742212017645e-06,
      "loss": 0.0031,
      "step": 53720
    },
    {
      "epoch": 0.3446075303126994,
      "grad_norm": 0.6549052596092224,
      "learning_rate": 9.841602478448845e-06,
      "loss": 0.0035,
      "step": 53730
    },
    {
      "epoch": 0.3446716672064855,
      "grad_norm": 0.1550072878599167,
      "learning_rate": 9.841462684211307e-06,
      "loss": 0.0039,
      "step": 53740
    },
    {
      "epoch": 0.3447358041002716,
      "grad_norm": 0.2029644101858139,
      "learning_rate": 9.841322829306779e-06,
      "loss": 0.0036,
      "step": 53750
    },
    {
      "epoch": 0.3447999409940577,
      "grad_norm": 1.0198986530303955,
      "learning_rate": 9.841182913737018e-06,
      "loss": 0.0072,
      "step": 53760
    },
    {
      "epoch": 0.3448640778878438,
      "grad_norm": 0.46527236700057983,
      "learning_rate": 9.841042937503775e-06,
      "loss": 0.0052,
      "step": 53770
    },
    {
      "epoch": 0.34492821478162994,
      "grad_norm": 0.30712541937828064,
      "learning_rate": 9.840902900608802e-06,
      "loss": 0.0049,
      "step": 53780
    },
    {
      "epoch": 0.34499235167541603,
      "grad_norm": 0.09056705236434937,
      "learning_rate": 9.840762803053858e-06,
      "loss": 0.0149,
      "step": 53790
    },
    {
      "epoch": 0.3450564885692021,
      "grad_norm": 0.2204216718673706,
      "learning_rate": 9.840622644840698e-06,
      "loss": 0.0062,
      "step": 53800
    },
    {
      "epoch": 0.3451206254629882,
      "grad_norm": 0.11173199862241745,
      "learning_rate": 9.840482425971074e-06,
      "loss": 0.0031,
      "step": 53810
    },
    {
      "epoch": 0.3451847623567743,
      "grad_norm": 0.23205724358558655,
      "learning_rate": 9.840342146446746e-06,
      "loss": 0.0074,
      "step": 53820
    },
    {
      "epoch": 0.3452488992505604,
      "grad_norm": 0.12278829514980316,
      "learning_rate": 9.840201806269472e-06,
      "loss": 0.0031,
      "step": 53830
    },
    {
      "epoch": 0.3453130361443465,
      "grad_norm": 0.2744913697242737,
      "learning_rate": 9.84006140544101e-06,
      "loss": 0.0038,
      "step": 53840
    },
    {
      "epoch": 0.3453771730381326,
      "grad_norm": 0.30894720554351807,
      "learning_rate": 9.839920943963119e-06,
      "loss": 0.0063,
      "step": 53850
    },
    {
      "epoch": 0.3454413099319187,
      "grad_norm": 0.06933877617120743,
      "learning_rate": 9.839780421837562e-06,
      "loss": 0.0049,
      "step": 53860
    },
    {
      "epoch": 0.3455054468257048,
      "grad_norm": 0.1928311139345169,
      "learning_rate": 9.839639839066096e-06,
      "loss": 0.0038,
      "step": 53870
    },
    {
      "epoch": 0.34556958371949087,
      "grad_norm": 0.12387371063232422,
      "learning_rate": 9.839499195650482e-06,
      "loss": 0.0032,
      "step": 53880
    },
    {
      "epoch": 0.34563372061327696,
      "grad_norm": 0.12215980142354965,
      "learning_rate": 9.839358491592486e-06,
      "loss": 0.0049,
      "step": 53890
    },
    {
      "epoch": 0.34569785750706306,
      "grad_norm": 0.22633928060531616,
      "learning_rate": 9.83921772689387e-06,
      "loss": 0.0042,
      "step": 53900
    },
    {
      "epoch": 0.34576199440084915,
      "grad_norm": 0.22962556779384613,
      "learning_rate": 9.839076901556397e-06,
      "loss": 0.0088,
      "step": 53910
    },
    {
      "epoch": 0.3458261312946353,
      "grad_norm": 0.12060944736003876,
      "learning_rate": 9.83893601558183e-06,
      "loss": 0.0038,
      "step": 53920
    },
    {
      "epoch": 0.3458902681884214,
      "grad_norm": 0.027461474761366844,
      "learning_rate": 9.838795068971939e-06,
      "loss": 0.0042,
      "step": 53930
    },
    {
      "epoch": 0.3459544050822075,
      "grad_norm": 0.17180879414081573,
      "learning_rate": 9.838654061728487e-06,
      "loss": 0.006,
      "step": 53940
    },
    {
      "epoch": 0.3460185419759936,
      "grad_norm": 0.04157579317688942,
      "learning_rate": 9.83851299385324e-06,
      "loss": 0.0041,
      "step": 53950
    },
    {
      "epoch": 0.34608267886977967,
      "grad_norm": 0.15837259590625763,
      "learning_rate": 9.83837186534797e-06,
      "loss": 0.0032,
      "step": 53960
    },
    {
      "epoch": 0.34614681576356576,
      "grad_norm": 0.1311531513929367,
      "learning_rate": 9.838230676214439e-06,
      "loss": 0.0035,
      "step": 53970
    },
    {
      "epoch": 0.34621095265735186,
      "grad_norm": 0.24102309346199036,
      "learning_rate": 9.838089426454422e-06,
      "loss": 0.0044,
      "step": 53980
    },
    {
      "epoch": 0.34627508955113795,
      "grad_norm": 0.12493369728326797,
      "learning_rate": 9.837948116069686e-06,
      "loss": 0.0057,
      "step": 53990
    },
    {
      "epoch": 0.34633922644492404,
      "grad_norm": 0.04451517388224602,
      "learning_rate": 9.837806745062004e-06,
      "loss": 0.0047,
      "step": 54000
    },
    {
      "epoch": 0.34640336333871014,
      "grad_norm": 0.13316604495048523,
      "learning_rate": 9.837665313433144e-06,
      "loss": 0.0055,
      "step": 54010
    },
    {
      "epoch": 0.34646750023249623,
      "grad_norm": 0.15386483073234558,
      "learning_rate": 9.837523821184883e-06,
      "loss": 0.0046,
      "step": 54020
    },
    {
      "epoch": 0.3465316371262823,
      "grad_norm": 0.0709504559636116,
      "learning_rate": 9.837382268318988e-06,
      "loss": 0.0039,
      "step": 54030
    },
    {
      "epoch": 0.3465957740200684,
      "grad_norm": 0.4904066324234009,
      "learning_rate": 9.837240654837237e-06,
      "loss": 0.0044,
      "step": 54040
    },
    {
      "epoch": 0.3466599109138545,
      "grad_norm": 0.1809903234243393,
      "learning_rate": 9.837098980741405e-06,
      "loss": 0.0033,
      "step": 54050
    },
    {
      "epoch": 0.3467240478076406,
      "grad_norm": 0.12470248341560364,
      "learning_rate": 9.836957246033262e-06,
      "loss": 0.0031,
      "step": 54060
    },
    {
      "epoch": 0.34678818470142675,
      "grad_norm": 0.32245323061943054,
      "learning_rate": 9.836815450714591e-06,
      "loss": 0.0047,
      "step": 54070
    },
    {
      "epoch": 0.34685232159521284,
      "grad_norm": 0.11289974302053452,
      "learning_rate": 9.836673594787165e-06,
      "loss": 0.0036,
      "step": 54080
    },
    {
      "epoch": 0.34691645848899894,
      "grad_norm": 0.02664255164563656,
      "learning_rate": 9.836531678252762e-06,
      "loss": 0.0067,
      "step": 54090
    },
    {
      "epoch": 0.34698059538278503,
      "grad_norm": 0.26872023940086365,
      "learning_rate": 9.83638970111316e-06,
      "loss": 0.0035,
      "step": 54100
    },
    {
      "epoch": 0.3470447322765711,
      "grad_norm": 0.12747934460639954,
      "learning_rate": 9.83624766337014e-06,
      "loss": 0.0041,
      "step": 54110
    },
    {
      "epoch": 0.3471088691703572,
      "grad_norm": 0.07346111536026001,
      "learning_rate": 9.836105565025478e-06,
      "loss": 0.0041,
      "step": 54120
    },
    {
      "epoch": 0.3471730060641433,
      "grad_norm": 0.09227632731199265,
      "learning_rate": 9.83596340608096e-06,
      "loss": 0.0049,
      "step": 54130
    },
    {
      "epoch": 0.3472371429579294,
      "grad_norm": 0.07450766861438751,
      "learning_rate": 9.835821186538362e-06,
      "loss": 0.0043,
      "step": 54140
    },
    {
      "epoch": 0.3473012798517155,
      "grad_norm": 0.35511043667793274,
      "learning_rate": 9.835678906399468e-06,
      "loss": 0.0088,
      "step": 54150
    },
    {
      "epoch": 0.3473654167455016,
      "grad_norm": 0.24837952852249146,
      "learning_rate": 9.835536565666062e-06,
      "loss": 0.0036,
      "step": 54160
    },
    {
      "epoch": 0.3474295536392877,
      "grad_norm": 0.0934607982635498,
      "learning_rate": 9.835394164339927e-06,
      "loss": 0.0039,
      "step": 54170
    },
    {
      "epoch": 0.3474936905330738,
      "grad_norm": 0.2472117394208908,
      "learning_rate": 9.835251702422847e-06,
      "loss": 0.0053,
      "step": 54180
    },
    {
      "epoch": 0.34755782742685987,
      "grad_norm": 0.28953588008880615,
      "learning_rate": 9.83510917991661e-06,
      "loss": 0.0042,
      "step": 54190
    },
    {
      "epoch": 0.34762196432064596,
      "grad_norm": 0.36132022738456726,
      "learning_rate": 9.834966596822996e-06,
      "loss": 0.0067,
      "step": 54200
    },
    {
      "epoch": 0.3476861012144321,
      "grad_norm": 0.1938018649816513,
      "learning_rate": 9.834823953143798e-06,
      "loss": 0.0025,
      "step": 54210
    },
    {
      "epoch": 0.3477502381082182,
      "grad_norm": 0.23491692543029785,
      "learning_rate": 9.834681248880798e-06,
      "loss": 0.0079,
      "step": 54220
    },
    {
      "epoch": 0.3478143750020043,
      "grad_norm": 0.24288392066955566,
      "learning_rate": 9.834538484035788e-06,
      "loss": 0.0056,
      "step": 54230
    },
    {
      "epoch": 0.3478785118957904,
      "grad_norm": 0.14312800765037537,
      "learning_rate": 9.834395658610554e-06,
      "loss": 0.0046,
      "step": 54240
    },
    {
      "epoch": 0.3479426487895765,
      "grad_norm": 0.3459871709346771,
      "learning_rate": 9.834252772606888e-06,
      "loss": 0.0044,
      "step": 54250
    },
    {
      "epoch": 0.3480067856833626,
      "grad_norm": 0.466331422328949,
      "learning_rate": 9.834109826026582e-06,
      "loss": 0.0028,
      "step": 54260
    },
    {
      "epoch": 0.34807092257714867,
      "grad_norm": 0.07888225466012955,
      "learning_rate": 9.83396681887142e-06,
      "loss": 0.0057,
      "step": 54270
    },
    {
      "epoch": 0.34813505947093476,
      "grad_norm": 0.3045555055141449,
      "learning_rate": 9.833823751143204e-06,
      "loss": 0.0056,
      "step": 54280
    },
    {
      "epoch": 0.34819919636472085,
      "grad_norm": 0.14433254301548004,
      "learning_rate": 9.83368062284372e-06,
      "loss": 0.0037,
      "step": 54290
    },
    {
      "epoch": 0.34826333325850695,
      "grad_norm": 0.270381897687912,
      "learning_rate": 9.833537433974762e-06,
      "loss": 0.0065,
      "step": 54300
    },
    {
      "epoch": 0.34832747015229304,
      "grad_norm": 0.028089692816138268,
      "learning_rate": 9.833394184538128e-06,
      "loss": 0.0052,
      "step": 54310
    },
    {
      "epoch": 0.34839160704607913,
      "grad_norm": 0.04288167878985405,
      "learning_rate": 9.833250874535608e-06,
      "loss": 0.0029,
      "step": 54320
    },
    {
      "epoch": 0.3484557439398652,
      "grad_norm": 0.18451502919197083,
      "learning_rate": 9.833107503969001e-06,
      "loss": 0.007,
      "step": 54330
    },
    {
      "epoch": 0.3485198808336513,
      "grad_norm": 0.30367568135261536,
      "learning_rate": 9.832964072840104e-06,
      "loss": 0.0042,
      "step": 54340
    },
    {
      "epoch": 0.34858401772743747,
      "grad_norm": 0.08944468945264816,
      "learning_rate": 9.832820581150713e-06,
      "loss": 0.0045,
      "step": 54350
    },
    {
      "epoch": 0.34864815462122356,
      "grad_norm": 0.130207359790802,
      "learning_rate": 9.832677028902625e-06,
      "loss": 0.0065,
      "step": 54360
    },
    {
      "epoch": 0.34871229151500965,
      "grad_norm": 0.08129074424505234,
      "learning_rate": 9.83253341609764e-06,
      "loss": 0.0049,
      "step": 54370
    },
    {
      "epoch": 0.34877642840879575,
      "grad_norm": 0.059726521372795105,
      "learning_rate": 9.832389742737558e-06,
      "loss": 0.0044,
      "step": 54380
    },
    {
      "epoch": 0.34884056530258184,
      "grad_norm": 0.287548303604126,
      "learning_rate": 9.832246008824177e-06,
      "loss": 0.0035,
      "step": 54390
    },
    {
      "epoch": 0.34890470219636793,
      "grad_norm": 0.280642032623291,
      "learning_rate": 9.832102214359303e-06,
      "loss": 0.0076,
      "step": 54400
    },
    {
      "epoch": 0.348968839090154,
      "grad_norm": 0.39101871848106384,
      "learning_rate": 9.831958359344733e-06,
      "loss": 0.0049,
      "step": 54410
    },
    {
      "epoch": 0.3490329759839401,
      "grad_norm": 0.16636443138122559,
      "learning_rate": 9.831814443782273e-06,
      "loss": 0.003,
      "step": 54420
    },
    {
      "epoch": 0.3490971128777262,
      "grad_norm": 0.026397155597805977,
      "learning_rate": 9.831670467673723e-06,
      "loss": 0.0045,
      "step": 54430
    },
    {
      "epoch": 0.3491612497715123,
      "grad_norm": 0.11856728792190552,
      "learning_rate": 9.831526431020891e-06,
      "loss": 0.0045,
      "step": 54440
    },
    {
      "epoch": 0.3492253866652984,
      "grad_norm": 0.2409236580133438,
      "learning_rate": 9.831382333825576e-06,
      "loss": 0.0036,
      "step": 54450
    },
    {
      "epoch": 0.3492895235590845,
      "grad_norm": 0.06977161765098572,
      "learning_rate": 9.83123817608959e-06,
      "loss": 0.0042,
      "step": 54460
    },
    {
      "epoch": 0.3493536604528706,
      "grad_norm": 0.3645520806312561,
      "learning_rate": 9.831093957814737e-06,
      "loss": 0.0037,
      "step": 54470
    },
    {
      "epoch": 0.3494177973466567,
      "grad_norm": 0.38606497645378113,
      "learning_rate": 9.830949679002824e-06,
      "loss": 0.0047,
      "step": 54480
    },
    {
      "epoch": 0.3494819342404428,
      "grad_norm": 0.11708054691553116,
      "learning_rate": 9.830805339655658e-06,
      "loss": 0.0053,
      "step": 54490
    },
    {
      "epoch": 0.3495460711342289,
      "grad_norm": 0.11722180992364883,
      "learning_rate": 9.83066093977505e-06,
      "loss": 0.0057,
      "step": 54500
    },
    {
      "epoch": 0.349610208028015,
      "grad_norm": 0.16923189163208008,
      "learning_rate": 9.830516479362807e-06,
      "loss": 0.0053,
      "step": 54510
    },
    {
      "epoch": 0.3496743449218011,
      "grad_norm": 0.060270559042692184,
      "learning_rate": 9.830371958420738e-06,
      "loss": 0.0037,
      "step": 54520
    },
    {
      "epoch": 0.3497384818155872,
      "grad_norm": 0.20828652381896973,
      "learning_rate": 9.830227376950661e-06,
      "loss": 0.0038,
      "step": 54530
    },
    {
      "epoch": 0.3498026187093733,
      "grad_norm": 0.2548787295818329,
      "learning_rate": 9.83008273495438e-06,
      "loss": 0.01,
      "step": 54540
    },
    {
      "epoch": 0.3498667556031594,
      "grad_norm": 0.11429157108068466,
      "learning_rate": 9.829938032433711e-06,
      "loss": 0.0049,
      "step": 54550
    },
    {
      "epoch": 0.3499308924969455,
      "grad_norm": 0.20134198665618896,
      "learning_rate": 9.829793269390465e-06,
      "loss": 0.0036,
      "step": 54560
    },
    {
      "epoch": 0.34999502939073157,
      "grad_norm": 0.22732007503509521,
      "learning_rate": 9.829648445826459e-06,
      "loss": 0.0033,
      "step": 54570
    },
    {
      "epoch": 0.35005916628451766,
      "grad_norm": 0.66534024477005,
      "learning_rate": 9.829503561743505e-06,
      "loss": 0.0031,
      "step": 54580
    },
    {
      "epoch": 0.35012330317830376,
      "grad_norm": 0.2801349461078644,
      "learning_rate": 9.829358617143421e-06,
      "loss": 0.0031,
      "step": 54590
    },
    {
      "epoch": 0.35018744007208985,
      "grad_norm": 0.06797728687524796,
      "learning_rate": 9.829213612028021e-06,
      "loss": 0.0066,
      "step": 54600
    },
    {
      "epoch": 0.35025157696587594,
      "grad_norm": 0.31216877698898315,
      "learning_rate": 9.829068546399126e-06,
      "loss": 0.0038,
      "step": 54610
    },
    {
      "epoch": 0.35031571385966204,
      "grad_norm": 0.24418389797210693,
      "learning_rate": 9.828923420258547e-06,
      "loss": 0.004,
      "step": 54620
    },
    {
      "epoch": 0.3503798507534482,
      "grad_norm": 0.15022139251232147,
      "learning_rate": 9.828778233608109e-06,
      "loss": 0.0036,
      "step": 54630
    },
    {
      "epoch": 0.3504439876472343,
      "grad_norm": 0.10417608171701431,
      "learning_rate": 9.828632986449627e-06,
      "loss": 0.0061,
      "step": 54640
    },
    {
      "epoch": 0.35050812454102037,
      "grad_norm": 0.2197953164577484,
      "learning_rate": 9.828487678784923e-06,
      "loss": 0.0043,
      "step": 54650
    },
    {
      "epoch": 0.35057226143480646,
      "grad_norm": 0.5072107911109924,
      "learning_rate": 9.828342310615818e-06,
      "loss": 0.0056,
      "step": 54660
    },
    {
      "epoch": 0.35063639832859256,
      "grad_norm": 0.07112255692481995,
      "learning_rate": 9.828196881944133e-06,
      "loss": 0.0039,
      "step": 54670
    },
    {
      "epoch": 0.35070053522237865,
      "grad_norm": 0.5350751280784607,
      "learning_rate": 9.828051392771691e-06,
      "loss": 0.0048,
      "step": 54680
    },
    {
      "epoch": 0.35076467211616474,
      "grad_norm": 0.2553097605705261,
      "learning_rate": 9.827905843100314e-06,
      "loss": 0.0033,
      "step": 54690
    },
    {
      "epoch": 0.35082880900995084,
      "grad_norm": 0.26859140396118164,
      "learning_rate": 9.827760232931826e-06,
      "loss": 0.0041,
      "step": 54700
    },
    {
      "epoch": 0.35089294590373693,
      "grad_norm": 0.5209015607833862,
      "learning_rate": 9.827614562268052e-06,
      "loss": 0.005,
      "step": 54710
    },
    {
      "epoch": 0.350957082797523,
      "grad_norm": 0.1657070517539978,
      "learning_rate": 9.827468831110818e-06,
      "loss": 0.0052,
      "step": 54720
    },
    {
      "epoch": 0.3510212196913091,
      "grad_norm": 0.24692420661449432,
      "learning_rate": 9.827323039461948e-06,
      "loss": 0.0044,
      "step": 54730
    },
    {
      "epoch": 0.3510853565850952,
      "grad_norm": 0.28886139392852783,
      "learning_rate": 9.827177187323271e-06,
      "loss": 0.004,
      "step": 54740
    },
    {
      "epoch": 0.3511494934788813,
      "grad_norm": 1.9326666593551636,
      "learning_rate": 9.827031274696614e-06,
      "loss": 0.0064,
      "step": 54750
    },
    {
      "epoch": 0.3512136303726674,
      "grad_norm": 0.2602798640727997,
      "learning_rate": 9.826885301583805e-06,
      "loss": 0.0061,
      "step": 54760
    },
    {
      "epoch": 0.35127776726645354,
      "grad_norm": 0.37088316679000854,
      "learning_rate": 9.826739267986673e-06,
      "loss": 0.0091,
      "step": 54770
    },
    {
      "epoch": 0.35134190416023964,
      "grad_norm": 0.29487156867980957,
      "learning_rate": 9.826593173907051e-06,
      "loss": 0.0054,
      "step": 54780
    },
    {
      "epoch": 0.35140604105402573,
      "grad_norm": 0.24256150424480438,
      "learning_rate": 9.826447019346764e-06,
      "loss": 0.0028,
      "step": 54790
    },
    {
      "epoch": 0.3514701779478118,
      "grad_norm": 0.24807745218276978,
      "learning_rate": 9.826300804307648e-06,
      "loss": 0.0047,
      "step": 54800
    },
    {
      "epoch": 0.3515343148415979,
      "grad_norm": 0.1723044365644455,
      "learning_rate": 9.826154528791534e-06,
      "loss": 0.004,
      "step": 54810
    },
    {
      "epoch": 0.351598451735384,
      "grad_norm": 0.018190717324614525,
      "learning_rate": 9.826008192800253e-06,
      "loss": 0.0036,
      "step": 54820
    },
    {
      "epoch": 0.3516625886291701,
      "grad_norm": 0.1684645712375641,
      "learning_rate": 9.82586179633564e-06,
      "loss": 0.0038,
      "step": 54830
    },
    {
      "epoch": 0.3517267255229562,
      "grad_norm": 0.19667181372642517,
      "learning_rate": 9.825715339399532e-06,
      "loss": 0.0057,
      "step": 54840
    },
    {
      "epoch": 0.3517908624167423,
      "grad_norm": 0.2179258018732071,
      "learning_rate": 9.825568821993761e-06,
      "loss": 0.0061,
      "step": 54850
    },
    {
      "epoch": 0.3518549993105284,
      "grad_norm": 0.18799977004528046,
      "learning_rate": 9.825422244120162e-06,
      "loss": 0.0028,
      "step": 54860
    },
    {
      "epoch": 0.3519191362043145,
      "grad_norm": 0.3359462320804596,
      "learning_rate": 9.825275605780575e-06,
      "loss": 0.0049,
      "step": 54870
    },
    {
      "epoch": 0.35198327309810057,
      "grad_norm": 0.1551109403371811,
      "learning_rate": 9.825128906976837e-06,
      "loss": 0.0071,
      "step": 54880
    },
    {
      "epoch": 0.35204740999188666,
      "grad_norm": 0.03857658803462982,
      "learning_rate": 9.824982147710785e-06,
      "loss": 0.0027,
      "step": 54890
    },
    {
      "epoch": 0.35211154688567275,
      "grad_norm": 0.0902140811085701,
      "learning_rate": 9.824835327984256e-06,
      "loss": 0.004,
      "step": 54900
    },
    {
      "epoch": 0.3521756837794589,
      "grad_norm": 0.28138551115989685,
      "learning_rate": 9.824688447799095e-06,
      "loss": 0.0039,
      "step": 54910
    },
    {
      "epoch": 0.352239820673245,
      "grad_norm": 0.07199631631374359,
      "learning_rate": 9.824541507157138e-06,
      "loss": 0.0048,
      "step": 54920
    },
    {
      "epoch": 0.3523039575670311,
      "grad_norm": 0.3760847747325897,
      "learning_rate": 9.824394506060228e-06,
      "loss": 0.0068,
      "step": 54930
    },
    {
      "epoch": 0.3523680944608172,
      "grad_norm": 0.2786344885826111,
      "learning_rate": 9.824247444510209e-06,
      "loss": 0.0043,
      "step": 54940
    },
    {
      "epoch": 0.3524322313546033,
      "grad_norm": 0.21483300626277924,
      "learning_rate": 9.824100322508918e-06,
      "loss": 0.0052,
      "step": 54950
    },
    {
      "epoch": 0.35249636824838937,
      "grad_norm": 0.24060052633285522,
      "learning_rate": 9.823953140058205e-06,
      "loss": 0.008,
      "step": 54960
    },
    {
      "epoch": 0.35256050514217546,
      "grad_norm": 0.288587749004364,
      "learning_rate": 9.823805897159911e-06,
      "loss": 0.006,
      "step": 54970
    },
    {
      "epoch": 0.35262464203596156,
      "grad_norm": 0.22966255247592926,
      "learning_rate": 9.82365859381588e-06,
      "loss": 0.0051,
      "step": 54980
    },
    {
      "epoch": 0.35268877892974765,
      "grad_norm": 0.16944147646427155,
      "learning_rate": 9.823511230027962e-06,
      "loss": 0.0178,
      "step": 54990
    },
    {
      "epoch": 0.35275291582353374,
      "grad_norm": 0.1313783973455429,
      "learning_rate": 9.823363805798e-06,
      "loss": 0.0049,
      "step": 55000
    },
    {
      "epoch": 0.35281705271731983,
      "grad_norm": 0.07974176853895187,
      "learning_rate": 9.823216321127842e-06,
      "loss": 0.0029,
      "step": 55010
    },
    {
      "epoch": 0.3528811896111059,
      "grad_norm": 0.21326179802417755,
      "learning_rate": 9.823068776019336e-06,
      "loss": 0.0064,
      "step": 55020
    },
    {
      "epoch": 0.352945326504892,
      "grad_norm": 0.21994377672672272,
      "learning_rate": 9.822921170474332e-06,
      "loss": 0.0033,
      "step": 55030
    },
    {
      "epoch": 0.3530094633986781,
      "grad_norm": 0.0296862181276083,
      "learning_rate": 9.822773504494678e-06,
      "loss": 0.0031,
      "step": 55040
    },
    {
      "epoch": 0.35307360029246426,
      "grad_norm": 0.21492598950862885,
      "learning_rate": 9.822625778082226e-06,
      "loss": 0.0042,
      "step": 55050
    },
    {
      "epoch": 0.35313773718625036,
      "grad_norm": 0.36219337582588196,
      "learning_rate": 9.822477991238826e-06,
      "loss": 0.0055,
      "step": 55060
    },
    {
      "epoch": 0.35320187408003645,
      "grad_norm": 0.40082961320877075,
      "learning_rate": 9.82233014396633e-06,
      "loss": 0.0033,
      "step": 55070
    },
    {
      "epoch": 0.35326601097382254,
      "grad_norm": 0.36822453141212463,
      "learning_rate": 9.822182236266591e-06,
      "loss": 0.0047,
      "step": 55080
    },
    {
      "epoch": 0.35333014786760863,
      "grad_norm": 0.22897125780582428,
      "learning_rate": 9.822034268141462e-06,
      "loss": 0.0035,
      "step": 55090
    },
    {
      "epoch": 0.35339428476139473,
      "grad_norm": 0.09583339095115662,
      "learning_rate": 9.821886239592797e-06,
      "loss": 0.0041,
      "step": 55100
    },
    {
      "epoch": 0.3534584216551808,
      "grad_norm": 0.057896021753549576,
      "learning_rate": 9.821738150622453e-06,
      "loss": 0.0043,
      "step": 55110
    },
    {
      "epoch": 0.3535225585489669,
      "grad_norm": 0.26879703998565674,
      "learning_rate": 9.821590001232282e-06,
      "loss": 0.0067,
      "step": 55120
    },
    {
      "epoch": 0.353586695442753,
      "grad_norm": 0.48501795530319214,
      "learning_rate": 9.821441791424143e-06,
      "loss": 0.0039,
      "step": 55130
    },
    {
      "epoch": 0.3536508323365391,
      "grad_norm": 0.23591944575309753,
      "learning_rate": 9.821293521199892e-06,
      "loss": 0.004,
      "step": 55140
    },
    {
      "epoch": 0.3537149692303252,
      "grad_norm": 0.2928210198879242,
      "learning_rate": 9.82114519056139e-06,
      "loss": 0.0051,
      "step": 55150
    },
    {
      "epoch": 0.3537791061241113,
      "grad_norm": 0.0741635262966156,
      "learning_rate": 9.820996799510491e-06,
      "loss": 0.0044,
      "step": 55160
    },
    {
      "epoch": 0.3538432430178974,
      "grad_norm": 0.16287928819656372,
      "learning_rate": 9.820848348049057e-06,
      "loss": 0.0077,
      "step": 55170
    },
    {
      "epoch": 0.3539073799116835,
      "grad_norm": 0.1516534388065338,
      "learning_rate": 9.820699836178946e-06,
      "loss": 0.0052,
      "step": 55180
    },
    {
      "epoch": 0.3539715168054696,
      "grad_norm": 0.315433144569397,
      "learning_rate": 9.820551263902024e-06,
      "loss": 0.0044,
      "step": 55190
    },
    {
      "epoch": 0.3540356536992557,
      "grad_norm": 0.09990456700325012,
      "learning_rate": 9.820402631220147e-06,
      "loss": 0.0031,
      "step": 55200
    },
    {
      "epoch": 0.3540997905930418,
      "grad_norm": 0.17717120051383972,
      "learning_rate": 9.82025393813518e-06,
      "loss": 0.0044,
      "step": 55210
    },
    {
      "epoch": 0.3541639274868279,
      "grad_norm": 0.12074436247348785,
      "learning_rate": 9.820105184648988e-06,
      "loss": 0.004,
      "step": 55220
    },
    {
      "epoch": 0.354228064380614,
      "grad_norm": 0.12373752892017365,
      "learning_rate": 9.819956370763432e-06,
      "loss": 0.0039,
      "step": 55230
    },
    {
      "epoch": 0.3542922012744001,
      "grad_norm": 0.17480479180812836,
      "learning_rate": 9.819807496480377e-06,
      "loss": 0.0063,
      "step": 55240
    },
    {
      "epoch": 0.3543563381681862,
      "grad_norm": 0.12161833792924881,
      "learning_rate": 9.81965856180169e-06,
      "loss": 0.0026,
      "step": 55250
    },
    {
      "epoch": 0.3544204750619723,
      "grad_norm": 0.19953209161758423,
      "learning_rate": 9.819509566729238e-06,
      "loss": 0.0032,
      "step": 55260
    },
    {
      "epoch": 0.35448461195575837,
      "grad_norm": 0.259747177362442,
      "learning_rate": 9.819360511264886e-06,
      "loss": 0.0035,
      "step": 55270
    },
    {
      "epoch": 0.35454874884954446,
      "grad_norm": 0.11968179792165756,
      "learning_rate": 9.819211395410502e-06,
      "loss": 0.0069,
      "step": 55280
    },
    {
      "epoch": 0.35461288574333055,
      "grad_norm": 0.175838440656662,
      "learning_rate": 9.819062219167956e-06,
      "loss": 0.0057,
      "step": 55290
    },
    {
      "epoch": 0.35467702263711665,
      "grad_norm": 0.10187926888465881,
      "learning_rate": 9.818912982539114e-06,
      "loss": 0.0047,
      "step": 55300
    },
    {
      "epoch": 0.35474115953090274,
      "grad_norm": 0.2792898416519165,
      "learning_rate": 9.81876368552585e-06,
      "loss": 0.0027,
      "step": 55310
    },
    {
      "epoch": 0.35480529642468883,
      "grad_norm": 0.11766412854194641,
      "learning_rate": 9.818614328130035e-06,
      "loss": 0.0044,
      "step": 55320
    },
    {
      "epoch": 0.354869433318475,
      "grad_norm": 0.2481955587863922,
      "learning_rate": 9.818464910353536e-06,
      "loss": 0.0048,
      "step": 55330
    },
    {
      "epoch": 0.3549335702122611,
      "grad_norm": 0.1017189770936966,
      "learning_rate": 9.818315432198226e-06,
      "loss": 0.0039,
      "step": 55340
    },
    {
      "epoch": 0.35499770710604717,
      "grad_norm": 0.18881593644618988,
      "learning_rate": 9.818165893665985e-06,
      "loss": 0.0029,
      "step": 55350
    },
    {
      "epoch": 0.35506184399983326,
      "grad_norm": 0.30533653497695923,
      "learning_rate": 9.818016294758679e-06,
      "loss": 0.0051,
      "step": 55360
    },
    {
      "epoch": 0.35512598089361935,
      "grad_norm": 0.14908923208713531,
      "learning_rate": 9.817866635478185e-06,
      "loss": 0.0047,
      "step": 55370
    },
    {
      "epoch": 0.35519011778740545,
      "grad_norm": 0.29038989543914795,
      "learning_rate": 9.81771691582638e-06,
      "loss": 0.0044,
      "step": 55380
    },
    {
      "epoch": 0.35525425468119154,
      "grad_norm": 0.10095025599002838,
      "learning_rate": 9.817567135805138e-06,
      "loss": 0.0023,
      "step": 55390
    },
    {
      "epoch": 0.35531839157497763,
      "grad_norm": 0.11912602931261063,
      "learning_rate": 9.817417295416337e-06,
      "loss": 0.0035,
      "step": 55400
    },
    {
      "epoch": 0.3553825284687637,
      "grad_norm": 0.46975573897361755,
      "learning_rate": 9.817267394661854e-06,
      "loss": 0.0031,
      "step": 55410
    },
    {
      "epoch": 0.3554466653625498,
      "grad_norm": 0.4472891688346863,
      "learning_rate": 9.81711743354357e-06,
      "loss": 0.0054,
      "step": 55420
    },
    {
      "epoch": 0.3555108022563359,
      "grad_norm": 0.15583384037017822,
      "learning_rate": 9.816967412063359e-06,
      "loss": 0.0064,
      "step": 55430
    },
    {
      "epoch": 0.355574939150122,
      "grad_norm": 0.1715603917837143,
      "learning_rate": 9.816817330223105e-06,
      "loss": 0.0059,
      "step": 55440
    },
    {
      "epoch": 0.3556390760439081,
      "grad_norm": 0.2636367976665497,
      "learning_rate": 9.816667188024687e-06,
      "loss": 0.0039,
      "step": 55450
    },
    {
      "epoch": 0.3557032129376942,
      "grad_norm": 0.03399881720542908,
      "learning_rate": 9.816516985469986e-06,
      "loss": 0.0027,
      "step": 55460
    },
    {
      "epoch": 0.3557673498314803,
      "grad_norm": 0.3632083833217621,
      "learning_rate": 9.816366722560887e-06,
      "loss": 0.0072,
      "step": 55470
    },
    {
      "epoch": 0.35583148672526643,
      "grad_norm": 0.13824380934238434,
      "learning_rate": 9.81621639929927e-06,
      "loss": 0.0049,
      "step": 55480
    },
    {
      "epoch": 0.3558956236190525,
      "grad_norm": 0.4382361173629761,
      "learning_rate": 9.816066015687017e-06,
      "loss": 0.0048,
      "step": 55490
    },
    {
      "epoch": 0.3559597605128386,
      "grad_norm": 0.13956326246261597,
      "learning_rate": 9.815915571726018e-06,
      "loss": 0.0035,
      "step": 55500
    },
    {
      "epoch": 0.3560238974066247,
      "grad_norm": 0.14240862429141998,
      "learning_rate": 9.815765067418152e-06,
      "loss": 0.0037,
      "step": 55510
    },
    {
      "epoch": 0.3560880343004108,
      "grad_norm": 0.20180271565914154,
      "learning_rate": 9.815614502765311e-06,
      "loss": 0.011,
      "step": 55520
    },
    {
      "epoch": 0.3561521711941969,
      "grad_norm": 0.13986170291900635,
      "learning_rate": 9.815463877769377e-06,
      "loss": 0.0083,
      "step": 55530
    },
    {
      "epoch": 0.356216308087983,
      "grad_norm": 0.4541982114315033,
      "learning_rate": 9.81531319243224e-06,
      "loss": 0.0048,
      "step": 55540
    },
    {
      "epoch": 0.3562804449817691,
      "grad_norm": 0.44397735595703125,
      "learning_rate": 9.815162446755786e-06,
      "loss": 0.0062,
      "step": 55550
    },
    {
      "epoch": 0.3563445818755552,
      "grad_norm": 0.20005185902118683,
      "learning_rate": 9.815011640741905e-06,
      "loss": 0.0066,
      "step": 55560
    },
    {
      "epoch": 0.35640871876934127,
      "grad_norm": 0.4921051561832428,
      "learning_rate": 9.814860774392488e-06,
      "loss": 0.005,
      "step": 55570
    },
    {
      "epoch": 0.35647285566312736,
      "grad_norm": 0.10769011080265045,
      "learning_rate": 9.814709847709424e-06,
      "loss": 0.0049,
      "step": 55580
    },
    {
      "epoch": 0.35653699255691346,
      "grad_norm": 0.4561218023300171,
      "learning_rate": 9.814558860694604e-06,
      "loss": 0.0072,
      "step": 55590
    },
    {
      "epoch": 0.35660112945069955,
      "grad_norm": 0.0678478330373764,
      "learning_rate": 9.814407813349921e-06,
      "loss": 0.0048,
      "step": 55600
    },
    {
      "epoch": 0.35666526634448564,
      "grad_norm": 0.19913871586322784,
      "learning_rate": 9.814256705677268e-06,
      "loss": 0.005,
      "step": 55610
    },
    {
      "epoch": 0.3567294032382718,
      "grad_norm": 0.2567561864852905,
      "learning_rate": 9.814105537678536e-06,
      "loss": 0.0039,
      "step": 55620
    },
    {
      "epoch": 0.3567935401320579,
      "grad_norm": 0.13420292735099792,
      "learning_rate": 9.813954309355621e-06,
      "loss": 0.0075,
      "step": 55630
    },
    {
      "epoch": 0.356857677025844,
      "grad_norm": 0.19008968770503998,
      "learning_rate": 9.813803020710422e-06,
      "loss": 0.0055,
      "step": 55640
    },
    {
      "epoch": 0.35692181391963007,
      "grad_norm": 0.12023936957120895,
      "learning_rate": 9.813651671744827e-06,
      "loss": 0.0038,
      "step": 55650
    },
    {
      "epoch": 0.35698595081341616,
      "grad_norm": 0.3929827809333801,
      "learning_rate": 9.813500262460738e-06,
      "loss": 0.0038,
      "step": 55660
    },
    {
      "epoch": 0.35705008770720226,
      "grad_norm": 0.14300161600112915,
      "learning_rate": 9.81334879286005e-06,
      "loss": 0.0038,
      "step": 55670
    },
    {
      "epoch": 0.35711422460098835,
      "grad_norm": 0.11527574062347412,
      "learning_rate": 9.813197262944661e-06,
      "loss": 0.0035,
      "step": 55680
    },
    {
      "epoch": 0.35717836149477444,
      "grad_norm": 0.22679197788238525,
      "learning_rate": 9.81304567271647e-06,
      "loss": 0.0037,
      "step": 55690
    },
    {
      "epoch": 0.35724249838856054,
      "grad_norm": 0.12605354189872742,
      "learning_rate": 9.812894022177378e-06,
      "loss": 0.0031,
      "step": 55700
    },
    {
      "epoch": 0.35730663528234663,
      "grad_norm": 0.18351490795612335,
      "learning_rate": 9.812742311329284e-06,
      "loss": 0.002,
      "step": 55710
    },
    {
      "epoch": 0.3573707721761327,
      "grad_norm": 0.1666477471590042,
      "learning_rate": 9.812590540174089e-06,
      "loss": 0.0036,
      "step": 55720
    },
    {
      "epoch": 0.3574349090699188,
      "grad_norm": 0.2192985564470291,
      "learning_rate": 9.812438708713695e-06,
      "loss": 0.0042,
      "step": 55730
    },
    {
      "epoch": 0.3574990459637049,
      "grad_norm": 0.3767503798007965,
      "learning_rate": 9.812286816950006e-06,
      "loss": 0.0048,
      "step": 55740
    },
    {
      "epoch": 0.357563182857491,
      "grad_norm": 0.09017223119735718,
      "learning_rate": 9.81213486488492e-06,
      "loss": 0.0048,
      "step": 55750
    },
    {
      "epoch": 0.35762731975127715,
      "grad_norm": 0.336397647857666,
      "learning_rate": 9.81198285252035e-06,
      "loss": 0.0044,
      "step": 55760
    },
    {
      "epoch": 0.35769145664506324,
      "grad_norm": 0.1848268359899521,
      "learning_rate": 9.811830779858193e-06,
      "loss": 0.0033,
      "step": 55770
    },
    {
      "epoch": 0.35775559353884934,
      "grad_norm": 0.2532404661178589,
      "learning_rate": 9.811678646900357e-06,
      "loss": 0.0048,
      "step": 55780
    },
    {
      "epoch": 0.35781973043263543,
      "grad_norm": 0.13859908282756805,
      "learning_rate": 9.81152645364875e-06,
      "loss": 0.0027,
      "step": 55790
    },
    {
      "epoch": 0.3578838673264215,
      "grad_norm": 0.0673801600933075,
      "learning_rate": 9.811374200105277e-06,
      "loss": 0.0042,
      "step": 55800
    },
    {
      "epoch": 0.3579480042202076,
      "grad_norm": 0.28130680322647095,
      "learning_rate": 9.811221886271846e-06,
      "loss": 0.0042,
      "step": 55810
    },
    {
      "epoch": 0.3580121411139937,
      "grad_norm": 0.11280748248100281,
      "learning_rate": 9.811069512150367e-06,
      "loss": 0.0031,
      "step": 55820
    },
    {
      "epoch": 0.3580762780077798,
      "grad_norm": 0.46585577726364136,
      "learning_rate": 9.810917077742748e-06,
      "loss": 0.0082,
      "step": 55830
    },
    {
      "epoch": 0.3581404149015659,
      "grad_norm": 0.08036241680383682,
      "learning_rate": 9.810764583050902e-06,
      "loss": 0.0044,
      "step": 55840
    },
    {
      "epoch": 0.358204551795352,
      "grad_norm": 0.08849810808897018,
      "learning_rate": 9.810612028076737e-06,
      "loss": 0.0037,
      "step": 55850
    },
    {
      "epoch": 0.3582686886891381,
      "grad_norm": 0.1307210773229599,
      "learning_rate": 9.810459412822162e-06,
      "loss": 0.0031,
      "step": 55860
    },
    {
      "epoch": 0.3583328255829242,
      "grad_norm": 0.2808101773262024,
      "learning_rate": 9.810306737289095e-06,
      "loss": 0.0048,
      "step": 55870
    },
    {
      "epoch": 0.35839696247671027,
      "grad_norm": 0.09498907625675201,
      "learning_rate": 9.810154001479447e-06,
      "loss": 0.0027,
      "step": 55880
    },
    {
      "epoch": 0.35846109937049636,
      "grad_norm": 0.13824990391731262,
      "learning_rate": 9.810001205395129e-06,
      "loss": 0.0028,
      "step": 55890
    },
    {
      "epoch": 0.3585252362642825,
      "grad_norm": 0.20933681726455688,
      "learning_rate": 9.809848349038063e-06,
      "loss": 0.0046,
      "step": 55900
    },
    {
      "epoch": 0.3585893731580686,
      "grad_norm": 0.06892430037260056,
      "learning_rate": 9.809695432410155e-06,
      "loss": 0.004,
      "step": 55910
    },
    {
      "epoch": 0.3586535100518547,
      "grad_norm": 0.1586724817752838,
      "learning_rate": 9.809542455513328e-06,
      "loss": 0.0074,
      "step": 55920
    },
    {
      "epoch": 0.3587176469456408,
      "grad_norm": 0.25260305404663086,
      "learning_rate": 9.809389418349496e-06,
      "loss": 0.0037,
      "step": 55930
    },
    {
      "epoch": 0.3587817838394269,
      "grad_norm": 0.37511733174324036,
      "learning_rate": 9.809236320920578e-06,
      "loss": 0.0034,
      "step": 55940
    },
    {
      "epoch": 0.358845920733213,
      "grad_norm": 0.18205967545509338,
      "learning_rate": 9.809083163228493e-06,
      "loss": 0.0075,
      "step": 55950
    },
    {
      "epoch": 0.35891005762699907,
      "grad_norm": 0.12275759875774384,
      "learning_rate": 9.808929945275157e-06,
      "loss": 0.0024,
      "step": 55960
    },
    {
      "epoch": 0.35897419452078516,
      "grad_norm": 0.27325794100761414,
      "learning_rate": 9.808776667062493e-06,
      "loss": 0.0023,
      "step": 55970
    },
    {
      "epoch": 0.35903833141457125,
      "grad_norm": 0.09226607531309128,
      "learning_rate": 9.80862332859242e-06,
      "loss": 0.0026,
      "step": 55980
    },
    {
      "epoch": 0.35910246830835735,
      "grad_norm": 0.19567179679870605,
      "learning_rate": 9.808469929866858e-06,
      "loss": 0.0073,
      "step": 55990
    },
    {
      "epoch": 0.35916660520214344,
      "grad_norm": 0.19236637651920319,
      "learning_rate": 9.808316470887732e-06,
      "loss": 0.0066,
      "step": 56000
    },
    {
      "epoch": 0.35923074209592953,
      "grad_norm": 0.28044867515563965,
      "learning_rate": 9.808162951656968e-06,
      "loss": 0.0041,
      "step": 56010
    },
    {
      "epoch": 0.3592948789897156,
      "grad_norm": 0.13495536148548126,
      "learning_rate": 9.808009372176483e-06,
      "loss": 0.0091,
      "step": 56020
    },
    {
      "epoch": 0.3593590158835017,
      "grad_norm": 0.14301171898841858,
      "learning_rate": 9.807855732448204e-06,
      "loss": 0.0041,
      "step": 56030
    },
    {
      "epoch": 0.35942315277728787,
      "grad_norm": 0.17056189477443695,
      "learning_rate": 9.807702032474057e-06,
      "loss": 0.0071,
      "step": 56040
    },
    {
      "epoch": 0.35948728967107396,
      "grad_norm": 0.19335652887821198,
      "learning_rate": 9.807548272255968e-06,
      "loss": 0.0056,
      "step": 56050
    },
    {
      "epoch": 0.35955142656486005,
      "grad_norm": 0.19761638343334198,
      "learning_rate": 9.807394451795863e-06,
      "loss": 0.0045,
      "step": 56060
    },
    {
      "epoch": 0.35961556345864615,
      "grad_norm": 0.22033339738845825,
      "learning_rate": 9.807240571095669e-06,
      "loss": 0.0029,
      "step": 56070
    },
    {
      "epoch": 0.35967970035243224,
      "grad_norm": 0.3078736960887909,
      "learning_rate": 9.807086630157317e-06,
      "loss": 0.0068,
      "step": 56080
    },
    {
      "epoch": 0.35974383724621833,
      "grad_norm": 0.2283318042755127,
      "learning_rate": 9.806932628982731e-06,
      "loss": 0.0034,
      "step": 56090
    },
    {
      "epoch": 0.3598079741400044,
      "grad_norm": 0.25131726264953613,
      "learning_rate": 9.806778567573846e-06,
      "loss": 0.0036,
      "step": 56100
    },
    {
      "epoch": 0.3598721110337905,
      "grad_norm": 0.11536833643913269,
      "learning_rate": 9.806624445932588e-06,
      "loss": 0.0037,
      "step": 56110
    },
    {
      "epoch": 0.3599362479275766,
      "grad_norm": 0.23182439804077148,
      "learning_rate": 9.806470264060893e-06,
      "loss": 0.0048,
      "step": 56120
    },
    {
      "epoch": 0.3600003848213627,
      "grad_norm": 0.2506871521472931,
      "learning_rate": 9.80631602196069e-06,
      "loss": 0.0042,
      "step": 56130
    },
    {
      "epoch": 0.3600645217151488,
      "grad_norm": 0.19607239961624146,
      "learning_rate": 9.806161719633911e-06,
      "loss": 0.0045,
      "step": 56140
    },
    {
      "epoch": 0.3601286586089349,
      "grad_norm": 0.22224532067775726,
      "learning_rate": 9.80600735708249e-06,
      "loss": 0.0088,
      "step": 56150
    },
    {
      "epoch": 0.360192795502721,
      "grad_norm": 0.20805025100708008,
      "learning_rate": 9.805852934308363e-06,
      "loss": 0.0051,
      "step": 56160
    },
    {
      "epoch": 0.3602569323965071,
      "grad_norm": 0.08118850737810135,
      "learning_rate": 9.805698451313465e-06,
      "loss": 0.0044,
      "step": 56170
    },
    {
      "epoch": 0.3603210692902932,
      "grad_norm": 0.07110437750816345,
      "learning_rate": 9.80554390809973e-06,
      "loss": 0.005,
      "step": 56180
    },
    {
      "epoch": 0.3603852061840793,
      "grad_norm": 0.173774853348732,
      "learning_rate": 9.805389304669097e-06,
      "loss": 0.0041,
      "step": 56190
    },
    {
      "epoch": 0.3604493430778654,
      "grad_norm": 0.15812437236309052,
      "learning_rate": 9.8052346410235e-06,
      "loss": 0.0053,
      "step": 56200
    },
    {
      "epoch": 0.3605134799716515,
      "grad_norm": 0.3079059422016144,
      "learning_rate": 9.80507991716488e-06,
      "loss": 0.0047,
      "step": 56210
    },
    {
      "epoch": 0.3605776168654376,
      "grad_norm": 0.27946552634239197,
      "learning_rate": 9.804925133095173e-06,
      "loss": 0.005,
      "step": 56220
    },
    {
      "epoch": 0.3606417537592237,
      "grad_norm": 0.19736504554748535,
      "learning_rate": 9.804770288816318e-06,
      "loss": 0.0062,
      "step": 56230
    },
    {
      "epoch": 0.3607058906530098,
      "grad_norm": 0.19446900486946106,
      "learning_rate": 9.804615384330262e-06,
      "loss": 0.0053,
      "step": 56240
    },
    {
      "epoch": 0.3607700275467959,
      "grad_norm": 0.07679780572652817,
      "learning_rate": 9.80446041963894e-06,
      "loss": 0.0057,
      "step": 56250
    },
    {
      "epoch": 0.360834164440582,
      "grad_norm": 0.056387584656476974,
      "learning_rate": 9.804305394744293e-06,
      "loss": 0.0032,
      "step": 56260
    },
    {
      "epoch": 0.36089830133436807,
      "grad_norm": 0.126143217086792,
      "learning_rate": 9.804150309648267e-06,
      "loss": 0.0029,
      "step": 56270
    },
    {
      "epoch": 0.36096243822815416,
      "grad_norm": 0.16625581681728363,
      "learning_rate": 9.803995164352804e-06,
      "loss": 0.0044,
      "step": 56280
    },
    {
      "epoch": 0.36102657512194025,
      "grad_norm": 0.3765086233615875,
      "learning_rate": 9.803839958859849e-06,
      "loss": 0.004,
      "step": 56290
    },
    {
      "epoch": 0.36109071201572635,
      "grad_norm": 0.21009819209575653,
      "learning_rate": 9.803684693171347e-06,
      "loss": 0.0047,
      "step": 56300
    },
    {
      "epoch": 0.36115484890951244,
      "grad_norm": 0.06138169765472412,
      "learning_rate": 9.80352936728924e-06,
      "loss": 0.0059,
      "step": 56310
    },
    {
      "epoch": 0.3612189858032986,
      "grad_norm": 0.2305053174495697,
      "learning_rate": 9.803373981215478e-06,
      "loss": 0.0052,
      "step": 56320
    },
    {
      "epoch": 0.3612831226970847,
      "grad_norm": 0.25863513350486755,
      "learning_rate": 9.803218534952008e-06,
      "loss": 0.004,
      "step": 56330
    },
    {
      "epoch": 0.3613472595908708,
      "grad_norm": 0.41510552167892456,
      "learning_rate": 9.803063028500776e-06,
      "loss": 0.0034,
      "step": 56340
    },
    {
      "epoch": 0.36141139648465687,
      "grad_norm": 0.18759652972221375,
      "learning_rate": 9.802907461863734e-06,
      "loss": 0.0032,
      "step": 56350
    },
    {
      "epoch": 0.36147553337844296,
      "grad_norm": 0.3013518452644348,
      "learning_rate": 9.802751835042825e-06,
      "loss": 0.0031,
      "step": 56360
    },
    {
      "epoch": 0.36153967027222905,
      "grad_norm": 0.26576077938079834,
      "learning_rate": 9.802596148040006e-06,
      "loss": 0.0103,
      "step": 56370
    },
    {
      "epoch": 0.36160380716601515,
      "grad_norm": 0.2088194191455841,
      "learning_rate": 9.802440400857223e-06,
      "loss": 0.0025,
      "step": 56380
    },
    {
      "epoch": 0.36166794405980124,
      "grad_norm": 0.06073303148150444,
      "learning_rate": 9.80228459349643e-06,
      "loss": 0.0039,
      "step": 56390
    },
    {
      "epoch": 0.36173208095358733,
      "grad_norm": 0.04010922834277153,
      "learning_rate": 9.80212872595958e-06,
      "loss": 0.004,
      "step": 56400
    },
    {
      "epoch": 0.3617962178473734,
      "grad_norm": 0.19746612012386322,
      "learning_rate": 9.801972798248624e-06,
      "loss": 0.0037,
      "step": 56410
    },
    {
      "epoch": 0.3618603547411595,
      "grad_norm": 0.19448116421699524,
      "learning_rate": 9.801816810365518e-06,
      "loss": 0.0035,
      "step": 56420
    },
    {
      "epoch": 0.3619244916349456,
      "grad_norm": 0.17412406206130981,
      "learning_rate": 9.801660762312216e-06,
      "loss": 0.0033,
      "step": 56430
    },
    {
      "epoch": 0.3619886285287317,
      "grad_norm": 0.27388522028923035,
      "learning_rate": 9.80150465409067e-06,
      "loss": 0.0058,
      "step": 56440
    },
    {
      "epoch": 0.3620527654225178,
      "grad_norm": 0.23078088462352753,
      "learning_rate": 9.801348485702843e-06,
      "loss": 0.004,
      "step": 56450
    },
    {
      "epoch": 0.36211690231630395,
      "grad_norm": 0.19716985523700714,
      "learning_rate": 9.801192257150685e-06,
      "loss": 0.0033,
      "step": 56460
    },
    {
      "epoch": 0.36218103921009004,
      "grad_norm": 0.08864498883485794,
      "learning_rate": 9.801035968436157e-06,
      "loss": 0.0038,
      "step": 56470
    },
    {
      "epoch": 0.36224517610387613,
      "grad_norm": 0.19277538359165192,
      "learning_rate": 9.800879619561216e-06,
      "loss": 0.0042,
      "step": 56480
    },
    {
      "epoch": 0.3623093129976622,
      "grad_norm": 0.36649471521377563,
      "learning_rate": 9.800723210527825e-06,
      "loss": 0.0045,
      "step": 56490
    },
    {
      "epoch": 0.3623734498914483,
      "grad_norm": 0.1863778978586197,
      "learning_rate": 9.800566741337941e-06,
      "loss": 0.0148,
      "step": 56500
    },
    {
      "epoch": 0.3624375867852344,
      "grad_norm": 0.09909823536872864,
      "learning_rate": 9.800410211993523e-06,
      "loss": 0.0045,
      "step": 56510
    },
    {
      "epoch": 0.3625017236790205,
      "grad_norm": 0.14066021144390106,
      "learning_rate": 9.800253622496534e-06,
      "loss": 0.0032,
      "step": 56520
    },
    {
      "epoch": 0.3625658605728066,
      "grad_norm": 0.2353345900774002,
      "learning_rate": 9.800096972848938e-06,
      "loss": 0.005,
      "step": 56530
    },
    {
      "epoch": 0.3626299974665927,
      "grad_norm": 0.30205559730529785,
      "learning_rate": 9.799940263052696e-06,
      "loss": 0.0042,
      "step": 56540
    },
    {
      "epoch": 0.3626941343603788,
      "grad_norm": 0.18081988394260406,
      "learning_rate": 9.799783493109772e-06,
      "loss": 0.0043,
      "step": 56550
    },
    {
      "epoch": 0.3627582712541649,
      "grad_norm": 0.3089219331741333,
      "learning_rate": 9.799626663022129e-06,
      "loss": 0.005,
      "step": 56560
    },
    {
      "epoch": 0.36282240814795097,
      "grad_norm": 0.21994343400001526,
      "learning_rate": 9.799469772791736e-06,
      "loss": 0.0051,
      "step": 56570
    },
    {
      "epoch": 0.36288654504173706,
      "grad_norm": 0.19096291065216064,
      "learning_rate": 9.799312822420554e-06,
      "loss": 0.0029,
      "step": 56580
    },
    {
      "epoch": 0.36295068193552316,
      "grad_norm": 0.11202064901590347,
      "learning_rate": 9.799155811910555e-06,
      "loss": 0.0026,
      "step": 56590
    },
    {
      "epoch": 0.3630148188293093,
      "grad_norm": 0.21112869679927826,
      "learning_rate": 9.798998741263703e-06,
      "loss": 0.0034,
      "step": 56600
    },
    {
      "epoch": 0.3630789557230954,
      "grad_norm": 0.17619720101356506,
      "learning_rate": 9.798841610481967e-06,
      "loss": 0.0047,
      "step": 56610
    },
    {
      "epoch": 0.3631430926168815,
      "grad_norm": 0.18076972663402557,
      "learning_rate": 9.798684419567314e-06,
      "loss": 0.0042,
      "step": 56620
    },
    {
      "epoch": 0.3632072295106676,
      "grad_norm": 0.19960550963878632,
      "learning_rate": 9.79852716852172e-06,
      "loss": 0.0036,
      "step": 56630
    },
    {
      "epoch": 0.3632713664044537,
      "grad_norm": 0.2491637021303177,
      "learning_rate": 9.798369857347147e-06,
      "loss": 0.0053,
      "step": 56640
    },
    {
      "epoch": 0.36333550329823977,
      "grad_norm": 0.12342242896556854,
      "learning_rate": 9.79821248604557e-06,
      "loss": 0.0059,
      "step": 56650
    },
    {
      "epoch": 0.36339964019202586,
      "grad_norm": 0.0697750374674797,
      "learning_rate": 9.798055054618965e-06,
      "loss": 0.0032,
      "step": 56660
    },
    {
      "epoch": 0.36346377708581196,
      "grad_norm": 0.19258704781532288,
      "learning_rate": 9.797897563069299e-06,
      "loss": 0.0044,
      "step": 56670
    },
    {
      "epoch": 0.36352791397959805,
      "grad_norm": 0.14493019878864288,
      "learning_rate": 9.797740011398547e-06,
      "loss": 0.003,
      "step": 56680
    },
    {
      "epoch": 0.36359205087338414,
      "grad_norm": 0.261229544878006,
      "learning_rate": 9.797582399608686e-06,
      "loss": 0.0036,
      "step": 56690
    },
    {
      "epoch": 0.36365618776717024,
      "grad_norm": 0.08727063238620758,
      "learning_rate": 9.797424727701685e-06,
      "loss": 0.0038,
      "step": 56700
    },
    {
      "epoch": 0.36372032466095633,
      "grad_norm": 0.17275263369083405,
      "learning_rate": 9.797266995679527e-06,
      "loss": 0.0052,
      "step": 56710
    },
    {
      "epoch": 0.3637844615547424,
      "grad_norm": 0.6614558696746826,
      "learning_rate": 9.797109203544183e-06,
      "loss": 0.0043,
      "step": 56720
    },
    {
      "epoch": 0.3638485984485285,
      "grad_norm": 0.20420671999454498,
      "learning_rate": 9.796951351297632e-06,
      "loss": 0.0069,
      "step": 56730
    },
    {
      "epoch": 0.36391273534231466,
      "grad_norm": 0.14583484828472137,
      "learning_rate": 9.796793438941853e-06,
      "loss": 0.0039,
      "step": 56740
    },
    {
      "epoch": 0.36397687223610076,
      "grad_norm": 0.3670613467693329,
      "learning_rate": 9.796635466478824e-06,
      "loss": 0.0042,
      "step": 56750
    },
    {
      "epoch": 0.36404100912988685,
      "grad_norm": 0.2378796488046646,
      "learning_rate": 9.796477433910526e-06,
      "loss": 0.0049,
      "step": 56760
    },
    {
      "epoch": 0.36410514602367294,
      "grad_norm": 0.24503864347934723,
      "learning_rate": 9.796319341238936e-06,
      "loss": 0.0028,
      "step": 56770
    },
    {
      "epoch": 0.36416928291745904,
      "grad_norm": 0.1846599280834198,
      "learning_rate": 9.796161188466036e-06,
      "loss": 0.0026,
      "step": 56780
    },
    {
      "epoch": 0.36423341981124513,
      "grad_norm": 0.28431859612464905,
      "learning_rate": 9.79600297559381e-06,
      "loss": 0.0045,
      "step": 56790
    },
    {
      "epoch": 0.3642975567050312,
      "grad_norm": 0.14934979379177094,
      "learning_rate": 9.795844702624239e-06,
      "loss": 0.0049,
      "step": 56800
    },
    {
      "epoch": 0.3643616935988173,
      "grad_norm": 0.03336900845170021,
      "learning_rate": 9.795686369559304e-06,
      "loss": 0.0045,
      "step": 56810
    },
    {
      "epoch": 0.3644258304926034,
      "grad_norm": 0.054301392287015915,
      "learning_rate": 9.795527976400996e-06,
      "loss": 0.0026,
      "step": 56820
    },
    {
      "epoch": 0.3644899673863895,
      "grad_norm": 0.08391859382390976,
      "learning_rate": 9.795369523151292e-06,
      "loss": 0.0047,
      "step": 56830
    },
    {
      "epoch": 0.3645541042801756,
      "grad_norm": 0.16880223155021667,
      "learning_rate": 9.795211009812181e-06,
      "loss": 0.0046,
      "step": 56840
    },
    {
      "epoch": 0.3646182411739617,
      "grad_norm": 0.16764570772647858,
      "learning_rate": 9.79505243638565e-06,
      "loss": 0.0053,
      "step": 56850
    },
    {
      "epoch": 0.3646823780677478,
      "grad_norm": 0.17726831138134003,
      "learning_rate": 9.794893802873683e-06,
      "loss": 0.0038,
      "step": 56860
    },
    {
      "epoch": 0.3647465149615339,
      "grad_norm": 0.09660816192626953,
      "learning_rate": 9.794735109278272e-06,
      "loss": 0.0048,
      "step": 56870
    },
    {
      "epoch": 0.36481065185531997,
      "grad_norm": 0.14028716087341309,
      "learning_rate": 9.794576355601401e-06,
      "loss": 0.0028,
      "step": 56880
    },
    {
      "epoch": 0.3648747887491061,
      "grad_norm": 0.21155264973640442,
      "learning_rate": 9.794417541845064e-06,
      "loss": 0.0054,
      "step": 56890
    },
    {
      "epoch": 0.3649389256428922,
      "grad_norm": 0.1422126740217209,
      "learning_rate": 9.794258668011247e-06,
      "loss": 0.0045,
      "step": 56900
    },
    {
      "epoch": 0.3650030625366783,
      "grad_norm": 0.11498111486434937,
      "learning_rate": 9.794099734101943e-06,
      "loss": 0.0036,
      "step": 56910
    },
    {
      "epoch": 0.3650671994304644,
      "grad_norm": 0.32154351472854614,
      "learning_rate": 9.793940740119143e-06,
      "loss": 0.0067,
      "step": 56920
    },
    {
      "epoch": 0.3651313363242505,
      "grad_norm": 0.16818641126155853,
      "learning_rate": 9.793781686064841e-06,
      "loss": 0.0036,
      "step": 56930
    },
    {
      "epoch": 0.3651954732180366,
      "grad_norm": 0.1259140521287918,
      "learning_rate": 9.793622571941026e-06,
      "loss": 0.0053,
      "step": 56940
    },
    {
      "epoch": 0.3652596101118227,
      "grad_norm": 0.20569679141044617,
      "learning_rate": 9.793463397749695e-06,
      "loss": 0.0057,
      "step": 56950
    },
    {
      "epoch": 0.36532374700560877,
      "grad_norm": 0.09647531062364578,
      "learning_rate": 9.793304163492843e-06,
      "loss": 0.0035,
      "step": 56960
    },
    {
      "epoch": 0.36538788389939486,
      "grad_norm": 0.2747953236103058,
      "learning_rate": 9.793144869172462e-06,
      "loss": 0.0041,
      "step": 56970
    },
    {
      "epoch": 0.36545202079318095,
      "grad_norm": 0.26019948720932007,
      "learning_rate": 9.792985514790552e-06,
      "loss": 0.0041,
      "step": 56980
    },
    {
      "epoch": 0.36551615768696705,
      "grad_norm": 0.08870185166597366,
      "learning_rate": 9.792826100349106e-06,
      "loss": 0.0019,
      "step": 56990
    },
    {
      "epoch": 0.36558029458075314,
      "grad_norm": 0.16642318665981293,
      "learning_rate": 9.792666625850126e-06,
      "loss": 0.0043,
      "step": 57000
    },
    {
      "epoch": 0.36564443147453923,
      "grad_norm": 0.36684736609458923,
      "learning_rate": 9.792507091295607e-06,
      "loss": 0.0028,
      "step": 57010
    },
    {
      "epoch": 0.3657085683683253,
      "grad_norm": 0.21510466933250427,
      "learning_rate": 9.792347496687548e-06,
      "loss": 0.0071,
      "step": 57020
    },
    {
      "epoch": 0.3657727052621115,
      "grad_norm": 0.274148166179657,
      "learning_rate": 9.79218784202795e-06,
      "loss": 0.005,
      "step": 57030
    },
    {
      "epoch": 0.36583684215589757,
      "grad_norm": 0.10369301587343216,
      "learning_rate": 9.792028127318815e-06,
      "loss": 0.0035,
      "step": 57040
    },
    {
      "epoch": 0.36590097904968366,
      "grad_norm": 0.2976493537425995,
      "learning_rate": 9.79186835256214e-06,
      "loss": 0.0029,
      "step": 57050
    },
    {
      "epoch": 0.36596511594346975,
      "grad_norm": 0.05087069049477577,
      "learning_rate": 9.791708517759933e-06,
      "loss": 0.0044,
      "step": 57060
    },
    {
      "epoch": 0.36602925283725585,
      "grad_norm": 0.12944692373275757,
      "learning_rate": 9.791548622914191e-06,
      "loss": 0.0067,
      "step": 57070
    },
    {
      "epoch": 0.36609338973104194,
      "grad_norm": 0.13219457864761353,
      "learning_rate": 9.791388668026923e-06,
      "loss": 0.0027,
      "step": 57080
    },
    {
      "epoch": 0.36615752662482803,
      "grad_norm": 0.15380585193634033,
      "learning_rate": 9.79122865310013e-06,
      "loss": 0.0056,
      "step": 57090
    },
    {
      "epoch": 0.3662216635186141,
      "grad_norm": 0.10875172913074493,
      "learning_rate": 9.791068578135817e-06,
      "loss": 0.0029,
      "step": 57100
    },
    {
      "epoch": 0.3662858004124002,
      "grad_norm": 0.28335443139076233,
      "learning_rate": 9.79090844313599e-06,
      "loss": 0.0049,
      "step": 57110
    },
    {
      "epoch": 0.3663499373061863,
      "grad_norm": 0.4130868911743164,
      "learning_rate": 9.790748248102654e-06,
      "loss": 0.005,
      "step": 57120
    },
    {
      "epoch": 0.3664140741999724,
      "grad_norm": 0.05378401651978493,
      "learning_rate": 9.790587993037824e-06,
      "loss": 0.0034,
      "step": 57130
    },
    {
      "epoch": 0.3664782110937585,
      "grad_norm": 0.14871534705162048,
      "learning_rate": 9.7904276779435e-06,
      "loss": 0.0056,
      "step": 57140
    },
    {
      "epoch": 0.3665423479875446,
      "grad_norm": 0.19948628544807434,
      "learning_rate": 9.790267302821692e-06,
      "loss": 0.0038,
      "step": 57150
    },
    {
      "epoch": 0.3666064848813307,
      "grad_norm": 0.21524202823638916,
      "learning_rate": 9.790106867674415e-06,
      "loss": 0.0052,
      "step": 57160
    },
    {
      "epoch": 0.36667062177511683,
      "grad_norm": 0.2842295467853546,
      "learning_rate": 9.789946372503672e-06,
      "loss": 0.0041,
      "step": 57170
    },
    {
      "epoch": 0.3667347586689029,
      "grad_norm": 0.1311556100845337,
      "learning_rate": 9.789785817311477e-06,
      "loss": 0.0065,
      "step": 57180
    },
    {
      "epoch": 0.366798895562689,
      "grad_norm": 0.5092558860778809,
      "learning_rate": 9.789625202099846e-06,
      "loss": 0.0046,
      "step": 57190
    },
    {
      "epoch": 0.3668630324564751,
      "grad_norm": 0.13105471432209015,
      "learning_rate": 9.789464526870785e-06,
      "loss": 0.0053,
      "step": 57200
    },
    {
      "epoch": 0.3669271693502612,
      "grad_norm": 0.17737802863121033,
      "learning_rate": 9.789303791626313e-06,
      "loss": 0.0037,
      "step": 57210
    },
    {
      "epoch": 0.3669913062440473,
      "grad_norm": 0.2169865518808365,
      "learning_rate": 9.78914299636844e-06,
      "loss": 0.0039,
      "step": 57220
    },
    {
      "epoch": 0.3670554431378334,
      "grad_norm": 0.5936599969863892,
      "learning_rate": 9.788982141099182e-06,
      "loss": 0.0076,
      "step": 57230
    },
    {
      "epoch": 0.3671195800316195,
      "grad_norm": 0.2433180958032608,
      "learning_rate": 9.788821225820557e-06,
      "loss": 0.0031,
      "step": 57240
    },
    {
      "epoch": 0.3671837169254056,
      "grad_norm": 0.11742263287305832,
      "learning_rate": 9.788660250534579e-06,
      "loss": 0.0044,
      "step": 57250
    },
    {
      "epoch": 0.36724785381919167,
      "grad_norm": 0.11817143857479095,
      "learning_rate": 9.788499215243264e-06,
      "loss": 0.0061,
      "step": 57260
    },
    {
      "epoch": 0.36731199071297777,
      "grad_norm": 0.08936230838298798,
      "learning_rate": 9.788338119948634e-06,
      "loss": 0.005,
      "step": 57270
    },
    {
      "epoch": 0.36737612760676386,
      "grad_norm": 0.25392237305641174,
      "learning_rate": 9.788176964652703e-06,
      "loss": 0.009,
      "step": 57280
    },
    {
      "epoch": 0.36744026450054995,
      "grad_norm": 0.14953184127807617,
      "learning_rate": 9.788015749357493e-06,
      "loss": 0.0032,
      "step": 57290
    },
    {
      "epoch": 0.36750440139433604,
      "grad_norm": 1.2418384552001953,
      "learning_rate": 9.787854474065026e-06,
      "loss": 0.0023,
      "step": 57300
    },
    {
      "epoch": 0.3675685382881222,
      "grad_norm": 0.039404936134815216,
      "learning_rate": 9.78769313877732e-06,
      "loss": 0.0069,
      "step": 57310
    },
    {
      "epoch": 0.3676326751819083,
      "grad_norm": 0.16414855420589447,
      "learning_rate": 9.787531743496394e-06,
      "loss": 0.0085,
      "step": 57320
    },
    {
      "epoch": 0.3676968120756944,
      "grad_norm": 0.20494332909584045,
      "learning_rate": 9.787370288224277e-06,
      "loss": 0.0107,
      "step": 57330
    },
    {
      "epoch": 0.3677609489694805,
      "grad_norm": 0.07614117115736008,
      "learning_rate": 9.787208772962988e-06,
      "loss": 0.0047,
      "step": 57340
    },
    {
      "epoch": 0.36782508586326657,
      "grad_norm": 0.03945811837911606,
      "learning_rate": 9.787047197714549e-06,
      "loss": 0.0083,
      "step": 57350
    },
    {
      "epoch": 0.36788922275705266,
      "grad_norm": 0.04837341606616974,
      "learning_rate": 9.78688556248099e-06,
      "loss": 0.0029,
      "step": 57360
    },
    {
      "epoch": 0.36795335965083875,
      "grad_norm": 0.19262497127056122,
      "learning_rate": 9.786723867264335e-06,
      "loss": 0.0054,
      "step": 57370
    },
    {
      "epoch": 0.36801749654462484,
      "grad_norm": 0.7808809280395508,
      "learning_rate": 9.786562112066606e-06,
      "loss": 0.0063,
      "step": 57380
    },
    {
      "epoch": 0.36808163343841094,
      "grad_norm": 0.21556198596954346,
      "learning_rate": 9.786400296889835e-06,
      "loss": 0.0043,
      "step": 57390
    },
    {
      "epoch": 0.36814577033219703,
      "grad_norm": 0.12813107669353485,
      "learning_rate": 9.786238421736048e-06,
      "loss": 0.0044,
      "step": 57400
    },
    {
      "epoch": 0.3682099072259831,
      "grad_norm": 0.14205877482891083,
      "learning_rate": 9.78607648660727e-06,
      "loss": 0.0051,
      "step": 57410
    },
    {
      "epoch": 0.3682740441197692,
      "grad_norm": 0.12445349991321564,
      "learning_rate": 9.785914491505535e-06,
      "loss": 0.0047,
      "step": 57420
    },
    {
      "epoch": 0.3683381810135553,
      "grad_norm": 0.22566372156143188,
      "learning_rate": 9.785752436432871e-06,
      "loss": 0.0045,
      "step": 57430
    },
    {
      "epoch": 0.3684023179073414,
      "grad_norm": 0.1090468168258667,
      "learning_rate": 9.785590321391308e-06,
      "loss": 0.0045,
      "step": 57440
    },
    {
      "epoch": 0.36846645480112755,
      "grad_norm": 0.22308801114559174,
      "learning_rate": 9.785428146382878e-06,
      "loss": 0.0058,
      "step": 57450
    },
    {
      "epoch": 0.36853059169491365,
      "grad_norm": 0.2822139859199524,
      "learning_rate": 9.785265911409613e-06,
      "loss": 0.0028,
      "step": 57460
    },
    {
      "epoch": 0.36859472858869974,
      "grad_norm": 0.11054620891809464,
      "learning_rate": 9.785103616473547e-06,
      "loss": 0.0039,
      "step": 57470
    },
    {
      "epoch": 0.36865886548248583,
      "grad_norm": 0.17774257063865662,
      "learning_rate": 9.784941261576714e-06,
      "loss": 0.0057,
      "step": 57480
    },
    {
      "epoch": 0.3687230023762719,
      "grad_norm": 0.09684696048498154,
      "learning_rate": 9.784778846721146e-06,
      "loss": 0.0036,
      "step": 57490
    },
    {
      "epoch": 0.368787139270058,
      "grad_norm": 0.15395021438598633,
      "learning_rate": 9.784616371908879e-06,
      "loss": 0.0019,
      "step": 57500
    },
    {
      "epoch": 0.3688512761638441,
      "grad_norm": 0.05425999313592911,
      "learning_rate": 9.78445383714195e-06,
      "loss": 0.0041,
      "step": 57510
    },
    {
      "epoch": 0.3689154130576302,
      "grad_norm": 0.24702796339988708,
      "learning_rate": 9.784291242422394e-06,
      "loss": 0.0051,
      "step": 57520
    },
    {
      "epoch": 0.3689795499514163,
      "grad_norm": 0.13330377638339996,
      "learning_rate": 9.78412858775225e-06,
      "loss": 0.0024,
      "step": 57530
    },
    {
      "epoch": 0.3690436868452024,
      "grad_norm": 0.22147420048713684,
      "learning_rate": 9.783965873133557e-06,
      "loss": 0.006,
      "step": 57540
    },
    {
      "epoch": 0.3691078237389885,
      "grad_norm": 0.25603482127189636,
      "learning_rate": 9.78380309856835e-06,
      "loss": 0.0021,
      "step": 57550
    },
    {
      "epoch": 0.3691719606327746,
      "grad_norm": 0.14216257631778717,
      "learning_rate": 9.783640264058674e-06,
      "loss": 0.0043,
      "step": 57560
    },
    {
      "epoch": 0.36923609752656067,
      "grad_norm": 0.16568593680858612,
      "learning_rate": 9.783477369606565e-06,
      "loss": 0.0032,
      "step": 57570
    },
    {
      "epoch": 0.36930023442034676,
      "grad_norm": 0.09964045882225037,
      "learning_rate": 9.783314415214065e-06,
      "loss": 0.0049,
      "step": 57580
    },
    {
      "epoch": 0.3693643713141329,
      "grad_norm": 0.2552603781223297,
      "learning_rate": 9.783151400883217e-06,
      "loss": 0.0062,
      "step": 57590
    },
    {
      "epoch": 0.369428508207919,
      "grad_norm": 0.04480994865298271,
      "learning_rate": 9.782988326616065e-06,
      "loss": 0.0032,
      "step": 57600
    },
    {
      "epoch": 0.3694926451017051,
      "grad_norm": 0.11393241584300995,
      "learning_rate": 9.782825192414651e-06,
      "loss": 0.0023,
      "step": 57610
    },
    {
      "epoch": 0.3695567819954912,
      "grad_norm": 0.3635132610797882,
      "learning_rate": 9.782661998281018e-06,
      "loss": 0.0062,
      "step": 57620
    },
    {
      "epoch": 0.3696209188892773,
      "grad_norm": 0.3223370909690857,
      "learning_rate": 9.782498744217213e-06,
      "loss": 0.0057,
      "step": 57630
    },
    {
      "epoch": 0.3696850557830634,
      "grad_norm": 0.27565741539001465,
      "learning_rate": 9.78233543022528e-06,
      "loss": 0.0043,
      "step": 57640
    },
    {
      "epoch": 0.36974919267684947,
      "grad_norm": 0.055533505976200104,
      "learning_rate": 9.782172056307267e-06,
      "loss": 0.0035,
      "step": 57650
    },
    {
      "epoch": 0.36981332957063556,
      "grad_norm": 0.09045581519603729,
      "learning_rate": 9.782008622465218e-06,
      "loss": 0.0048,
      "step": 57660
    },
    {
      "epoch": 0.36987746646442166,
      "grad_norm": 0.04754919931292534,
      "learning_rate": 9.781845128701186e-06,
      "loss": 0.0038,
      "step": 57670
    },
    {
      "epoch": 0.36994160335820775,
      "grad_norm": 0.14638595283031464,
      "learning_rate": 9.781681575017216e-06,
      "loss": 0.0041,
      "step": 57680
    },
    {
      "epoch": 0.37000574025199384,
      "grad_norm": 0.1129072830080986,
      "learning_rate": 9.781517961415358e-06,
      "loss": 0.0045,
      "step": 57690
    },
    {
      "epoch": 0.37006987714577994,
      "grad_norm": 0.07457966357469559,
      "learning_rate": 9.781354287897663e-06,
      "loss": 0.005,
      "step": 57700
    },
    {
      "epoch": 0.37013401403956603,
      "grad_norm": 0.2532108426094055,
      "learning_rate": 9.781190554466183e-06,
      "loss": 0.0053,
      "step": 57710
    },
    {
      "epoch": 0.3701981509333521,
      "grad_norm": 0.07001078128814697,
      "learning_rate": 9.781026761122965e-06,
      "loss": 0.005,
      "step": 57720
    },
    {
      "epoch": 0.37026228782713827,
      "grad_norm": 0.04986302927136421,
      "learning_rate": 9.780862907870068e-06,
      "loss": 0.0037,
      "step": 57730
    },
    {
      "epoch": 0.37032642472092436,
      "grad_norm": 0.04155554622411728,
      "learning_rate": 9.78069899470954e-06,
      "loss": 0.0058,
      "step": 57740
    },
    {
      "epoch": 0.37039056161471046,
      "grad_norm": 0.10352253913879395,
      "learning_rate": 9.780535021643438e-06,
      "loss": 0.0038,
      "step": 57750
    },
    {
      "epoch": 0.37045469850849655,
      "grad_norm": 0.5587517023086548,
      "learning_rate": 9.780370988673815e-06,
      "loss": 0.0051,
      "step": 57760
    },
    {
      "epoch": 0.37051883540228264,
      "grad_norm": 0.24137362837791443,
      "learning_rate": 9.780206895802726e-06,
      "loss": 0.006,
      "step": 57770
    },
    {
      "epoch": 0.37058297229606874,
      "grad_norm": 0.16731369495391846,
      "learning_rate": 9.78004274303223e-06,
      "loss": 0.004,
      "step": 57780
    },
    {
      "epoch": 0.37064710918985483,
      "grad_norm": 0.19096648693084717,
      "learning_rate": 9.779878530364382e-06,
      "loss": 0.0029,
      "step": 57790
    },
    {
      "epoch": 0.3707112460836409,
      "grad_norm": 0.19307313859462738,
      "learning_rate": 9.779714257801239e-06,
      "loss": 0.0208,
      "step": 57800
    },
    {
      "epoch": 0.370775382977427,
      "grad_norm": 0.19411031901836395,
      "learning_rate": 9.779549925344861e-06,
      "loss": 0.0047,
      "step": 57810
    },
    {
      "epoch": 0.3708395198712131,
      "grad_norm": 0.3477558493614197,
      "learning_rate": 9.779385532997305e-06,
      "loss": 0.0031,
      "step": 57820
    },
    {
      "epoch": 0.3709036567649992,
      "grad_norm": 0.18182215094566345,
      "learning_rate": 9.779221080760634e-06,
      "loss": 0.0047,
      "step": 57830
    },
    {
      "epoch": 0.3709677936587853,
      "grad_norm": 0.25277140736579895,
      "learning_rate": 9.779056568636908e-06,
      "loss": 0.0103,
      "step": 57840
    },
    {
      "epoch": 0.3710319305525714,
      "grad_norm": 0.09088609367609024,
      "learning_rate": 9.778891996628186e-06,
      "loss": 0.0056,
      "step": 57850
    },
    {
      "epoch": 0.3710960674463575,
      "grad_norm": 0.2450553923845291,
      "learning_rate": 9.778727364736533e-06,
      "loss": 0.0051,
      "step": 57860
    },
    {
      "epoch": 0.37116020434014363,
      "grad_norm": 0.23536743223667145,
      "learning_rate": 9.77856267296401e-06,
      "loss": 0.0034,
      "step": 57870
    },
    {
      "epoch": 0.3712243412339297,
      "grad_norm": 0.12124660611152649,
      "learning_rate": 9.778397921312683e-06,
      "loss": 0.0055,
      "step": 57880
    },
    {
      "epoch": 0.3712884781277158,
      "grad_norm": 0.08905904740095139,
      "learning_rate": 9.778233109784614e-06,
      "loss": 0.0028,
      "step": 57890
    },
    {
      "epoch": 0.3713526150215019,
      "grad_norm": 0.1720379889011383,
      "learning_rate": 9.778068238381868e-06,
      "loss": 0.004,
      "step": 57900
    },
    {
      "epoch": 0.371416751915288,
      "grad_norm": 0.295754611492157,
      "learning_rate": 9.777903307106513e-06,
      "loss": 0.0046,
      "step": 57910
    },
    {
      "epoch": 0.3714808888090741,
      "grad_norm": 0.1361151933670044,
      "learning_rate": 9.777738315960615e-06,
      "loss": 0.0034,
      "step": 57920
    },
    {
      "epoch": 0.3715450257028602,
      "grad_norm": 0.14360782504081726,
      "learning_rate": 9.777573264946241e-06,
      "loss": 0.0029,
      "step": 57930
    },
    {
      "epoch": 0.3716091625966463,
      "grad_norm": 0.0977223590016365,
      "learning_rate": 9.77740815406546e-06,
      "loss": 0.0038,
      "step": 57940
    },
    {
      "epoch": 0.3716732994904324,
      "grad_norm": 0.05808310583233833,
      "learning_rate": 9.77724298332034e-06,
      "loss": 0.0024,
      "step": 57950
    },
    {
      "epoch": 0.37173743638421847,
      "grad_norm": 0.04208715260028839,
      "learning_rate": 9.777077752712952e-06,
      "loss": 0.0031,
      "step": 57960
    },
    {
      "epoch": 0.37180157327800456,
      "grad_norm": 0.4183732867240906,
      "learning_rate": 9.776912462245365e-06,
      "loss": 0.0071,
      "step": 57970
    },
    {
      "epoch": 0.37186571017179065,
      "grad_norm": 0.19289630651474,
      "learning_rate": 9.776747111919652e-06,
      "loss": 0.0062,
      "step": 57980
    },
    {
      "epoch": 0.37192984706557675,
      "grad_norm": 0.11601032316684723,
      "learning_rate": 9.776581701737881e-06,
      "loss": 0.014,
      "step": 57990
    },
    {
      "epoch": 0.37199398395936284,
      "grad_norm": 0.09035704284906387,
      "learning_rate": 9.776416231702131e-06,
      "loss": 0.0055,
      "step": 58000
    },
    {
      "epoch": 0.372058120853149,
      "grad_norm": 0.12546135485172272,
      "learning_rate": 9.77625070181447e-06,
      "loss": 0.0062,
      "step": 58010
    },
    {
      "epoch": 0.3721222577469351,
      "grad_norm": 0.2818664014339447,
      "learning_rate": 9.776085112076974e-06,
      "loss": 0.0038,
      "step": 58020
    },
    {
      "epoch": 0.3721863946407212,
      "grad_norm": 0.11678871512413025,
      "learning_rate": 9.775919462491718e-06,
      "loss": 0.0074,
      "step": 58030
    },
    {
      "epoch": 0.37225053153450727,
      "grad_norm": 0.2236698418855667,
      "learning_rate": 9.77575375306078e-06,
      "loss": 0.0058,
      "step": 58040
    },
    {
      "epoch": 0.37231466842829336,
      "grad_norm": 0.20366476476192474,
      "learning_rate": 9.775587983786231e-06,
      "loss": 0.0033,
      "step": 58050
    },
    {
      "epoch": 0.37237880532207945,
      "grad_norm": 0.15319477021694183,
      "learning_rate": 9.775422154670154e-06,
      "loss": 0.0043,
      "step": 58060
    },
    {
      "epoch": 0.37244294221586555,
      "grad_norm": 0.09600375592708588,
      "learning_rate": 9.775256265714623e-06,
      "loss": 0.0035,
      "step": 58070
    },
    {
      "epoch": 0.37250707910965164,
      "grad_norm": 0.17463268339633942,
      "learning_rate": 9.775090316921717e-06,
      "loss": 0.0041,
      "step": 58080
    },
    {
      "epoch": 0.37257121600343773,
      "grad_norm": 0.12594033777713776,
      "learning_rate": 9.77492430829352e-06,
      "loss": 0.0053,
      "step": 58090
    },
    {
      "epoch": 0.3726353528972238,
      "grad_norm": 0.18876934051513672,
      "learning_rate": 9.774758239832107e-06,
      "loss": 0.0042,
      "step": 58100
    },
    {
      "epoch": 0.3726994897910099,
      "grad_norm": 0.31719061732292175,
      "learning_rate": 9.774592111539561e-06,
      "loss": 0.0042,
      "step": 58110
    },
    {
      "epoch": 0.372763626684796,
      "grad_norm": 0.21381841599941254,
      "learning_rate": 9.774425923417964e-06,
      "loss": 0.0039,
      "step": 58120
    },
    {
      "epoch": 0.3728277635785821,
      "grad_norm": 0.05288473144173622,
      "learning_rate": 9.774259675469397e-06,
      "loss": 0.0036,
      "step": 58130
    },
    {
      "epoch": 0.3728919004723682,
      "grad_norm": 0.1426960974931717,
      "learning_rate": 9.774093367695945e-06,
      "loss": 0.0025,
      "step": 58140
    },
    {
      "epoch": 0.37295603736615435,
      "grad_norm": 0.2808662950992584,
      "learning_rate": 9.77392700009969e-06,
      "loss": 0.0046,
      "step": 58150
    },
    {
      "epoch": 0.37302017425994044,
      "grad_norm": 0.19471795856952667,
      "learning_rate": 9.77376057268272e-06,
      "loss": 0.0038,
      "step": 58160
    },
    {
      "epoch": 0.37308431115372653,
      "grad_norm": 0.11051013320684433,
      "learning_rate": 9.773594085447116e-06,
      "loss": 0.005,
      "step": 58170
    },
    {
      "epoch": 0.3731484480475126,
      "grad_norm": 0.08199208229780197,
      "learning_rate": 9.773427538394967e-06,
      "loss": 0.0045,
      "step": 58180
    },
    {
      "epoch": 0.3732125849412987,
      "grad_norm": 0.0872143879532814,
      "learning_rate": 9.77326093152836e-06,
      "loss": 0.0043,
      "step": 58190
    },
    {
      "epoch": 0.3732767218350848,
      "grad_norm": 0.2271701693534851,
      "learning_rate": 9.773094264849381e-06,
      "loss": 0.004,
      "step": 58200
    },
    {
      "epoch": 0.3733408587288709,
      "grad_norm": 0.07518661767244339,
      "learning_rate": 9.772927538360121e-06,
      "loss": 0.0083,
      "step": 58210
    },
    {
      "epoch": 0.373404995622657,
      "grad_norm": 0.3071039319038391,
      "learning_rate": 9.772760752062666e-06,
      "loss": 0.0035,
      "step": 58220
    },
    {
      "epoch": 0.3734691325164431,
      "grad_norm": 0.2126374989748001,
      "learning_rate": 9.77259390595911e-06,
      "loss": 0.0029,
      "step": 58230
    },
    {
      "epoch": 0.3735332694102292,
      "grad_norm": 0.3216870427131653,
      "learning_rate": 9.77242700005154e-06,
      "loss": 0.0044,
      "step": 58240
    },
    {
      "epoch": 0.3735974063040153,
      "grad_norm": 0.2805640995502472,
      "learning_rate": 9.77226003434205e-06,
      "loss": 0.0042,
      "step": 58250
    },
    {
      "epoch": 0.37366154319780137,
      "grad_norm": 0.18705041706562042,
      "learning_rate": 9.77209300883273e-06,
      "loss": 0.004,
      "step": 58260
    },
    {
      "epoch": 0.37372568009158746,
      "grad_norm": 0.4507295489311218,
      "learning_rate": 9.771925923525674e-06,
      "loss": 0.0069,
      "step": 58270
    },
    {
      "epoch": 0.37378981698537356,
      "grad_norm": 0.3732795715332031,
      "learning_rate": 9.771758778422977e-06,
      "loss": 0.0039,
      "step": 58280
    },
    {
      "epoch": 0.3738539538791597,
      "grad_norm": 0.25247877836227417,
      "learning_rate": 9.77159157352673e-06,
      "loss": 0.0051,
      "step": 58290
    },
    {
      "epoch": 0.3739180907729458,
      "grad_norm": 0.01693500392138958,
      "learning_rate": 9.771424308839033e-06,
      "loss": 0.003,
      "step": 58300
    },
    {
      "epoch": 0.3739822276667319,
      "grad_norm": 0.028985632583498955,
      "learning_rate": 9.771256984361978e-06,
      "loss": 0.0041,
      "step": 58310
    },
    {
      "epoch": 0.374046364560518,
      "grad_norm": 0.3382354974746704,
      "learning_rate": 9.771089600097663e-06,
      "loss": 0.0068,
      "step": 58320
    },
    {
      "epoch": 0.3741105014543041,
      "grad_norm": 0.17548677325248718,
      "learning_rate": 9.770922156048185e-06,
      "loss": 0.0058,
      "step": 58330
    },
    {
      "epoch": 0.37417463834809017,
      "grad_norm": 0.4082449972629547,
      "learning_rate": 9.770754652215644e-06,
      "loss": 0.0078,
      "step": 58340
    },
    {
      "epoch": 0.37423877524187626,
      "grad_norm": 0.14011430740356445,
      "learning_rate": 9.770587088602135e-06,
      "loss": 0.0047,
      "step": 58350
    },
    {
      "epoch": 0.37430291213566236,
      "grad_norm": 0.07062096893787384,
      "learning_rate": 9.770419465209762e-06,
      "loss": 0.0037,
      "step": 58360
    },
    {
      "epoch": 0.37436704902944845,
      "grad_norm": 0.1349411904811859,
      "learning_rate": 9.770251782040625e-06,
      "loss": 0.004,
      "step": 58370
    },
    {
      "epoch": 0.37443118592323454,
      "grad_norm": 0.16179393231868744,
      "learning_rate": 9.770084039096823e-06,
      "loss": 0.0093,
      "step": 58380
    },
    {
      "epoch": 0.37449532281702064,
      "grad_norm": 0.3049332797527313,
      "learning_rate": 9.769916236380458e-06,
      "loss": 0.0033,
      "step": 58390
    },
    {
      "epoch": 0.37455945971080673,
      "grad_norm": 0.22818659245967865,
      "learning_rate": 9.769748373893633e-06,
      "loss": 0.0035,
      "step": 58400
    },
    {
      "epoch": 0.3746235966045928,
      "grad_norm": 0.09892981499433517,
      "learning_rate": 9.769580451638454e-06,
      "loss": 0.0037,
      "step": 58410
    },
    {
      "epoch": 0.3746877334983789,
      "grad_norm": 0.15319883823394775,
      "learning_rate": 9.769412469617022e-06,
      "loss": 0.0033,
      "step": 58420
    },
    {
      "epoch": 0.374751870392165,
      "grad_norm": 0.14722667634487152,
      "learning_rate": 9.769244427831444e-06,
      "loss": 0.0032,
      "step": 58430
    },
    {
      "epoch": 0.37481600728595116,
      "grad_norm": 0.6439252495765686,
      "learning_rate": 9.769076326283824e-06,
      "loss": 0.0034,
      "step": 58440
    },
    {
      "epoch": 0.37488014417973725,
      "grad_norm": 0.16144661605358124,
      "learning_rate": 9.768908164976269e-06,
      "loss": 0.0037,
      "step": 58450
    },
    {
      "epoch": 0.37494428107352334,
      "grad_norm": 0.10690446943044662,
      "learning_rate": 9.768739943910887e-06,
      "loss": 0.0047,
      "step": 58460
    },
    {
      "epoch": 0.37500841796730944,
      "grad_norm": 0.19510377943515778,
      "learning_rate": 9.768571663089786e-06,
      "loss": 0.0055,
      "step": 58470
    },
    {
      "epoch": 0.37507255486109553,
      "grad_norm": 0.5048791170120239,
      "learning_rate": 9.768403322515072e-06,
      "loss": 0.005,
      "step": 58480
    },
    {
      "epoch": 0.3751366917548816,
      "grad_norm": 0.20897120237350464,
      "learning_rate": 9.768234922188858e-06,
      "loss": 0.0061,
      "step": 58490
    },
    {
      "epoch": 0.3752008286486677,
      "grad_norm": 0.14457884430885315,
      "learning_rate": 9.768066462113252e-06,
      "loss": 0.0051,
      "step": 58500
    },
    {
      "epoch": 0.3752649655424538,
      "grad_norm": 0.15007531642913818,
      "learning_rate": 9.767897942290366e-06,
      "loss": 0.003,
      "step": 58510
    },
    {
      "epoch": 0.3753291024362399,
      "grad_norm": 0.07479389756917953,
      "learning_rate": 9.76772936272231e-06,
      "loss": 0.0024,
      "step": 58520
    },
    {
      "epoch": 0.375393239330026,
      "grad_norm": 0.20837247371673584,
      "learning_rate": 9.767560723411199e-06,
      "loss": 0.0067,
      "step": 58530
    },
    {
      "epoch": 0.3754573762238121,
      "grad_norm": 0.22486932575702667,
      "learning_rate": 9.767392024359145e-06,
      "loss": 0.0036,
      "step": 58540
    },
    {
      "epoch": 0.3755215131175982,
      "grad_norm": 0.03745689615607262,
      "learning_rate": 9.767223265568261e-06,
      "loss": 0.0024,
      "step": 58550
    },
    {
      "epoch": 0.3755856500113843,
      "grad_norm": 0.308506041765213,
      "learning_rate": 9.767054447040663e-06,
      "loss": 0.0042,
      "step": 58560
    },
    {
      "epoch": 0.37564978690517037,
      "grad_norm": 0.06810969114303589,
      "learning_rate": 9.766885568778465e-06,
      "loss": 0.0048,
      "step": 58570
    },
    {
      "epoch": 0.3757139237989565,
      "grad_norm": 0.13696783781051636,
      "learning_rate": 9.766716630783784e-06,
      "loss": 0.0053,
      "step": 58580
    },
    {
      "epoch": 0.3757780606927426,
      "grad_norm": 0.11493898183107376,
      "learning_rate": 9.766547633058737e-06,
      "loss": 0.0062,
      "step": 58590
    },
    {
      "epoch": 0.3758421975865287,
      "grad_norm": 0.16757753491401672,
      "learning_rate": 9.766378575605442e-06,
      "loss": 0.0054,
      "step": 58600
    },
    {
      "epoch": 0.3759063344803148,
      "grad_norm": 0.4030192494392395,
      "learning_rate": 9.766209458426018e-06,
      "loss": 0.0033,
      "step": 58610
    },
    {
      "epoch": 0.3759704713741009,
      "grad_norm": 0.20232348144054413,
      "learning_rate": 9.766040281522581e-06,
      "loss": 0.0038,
      "step": 58620
    },
    {
      "epoch": 0.376034608267887,
      "grad_norm": 0.2894548773765564,
      "learning_rate": 9.765871044897254e-06,
      "loss": 0.0032,
      "step": 58630
    },
    {
      "epoch": 0.3760987451616731,
      "grad_norm": 0.29514172673225403,
      "learning_rate": 9.765701748552157e-06,
      "loss": 0.0044,
      "step": 58640
    },
    {
      "epoch": 0.37616288205545917,
      "grad_norm": 0.11390168219804764,
      "learning_rate": 9.76553239248941e-06,
      "loss": 0.0042,
      "step": 58650
    },
    {
      "epoch": 0.37622701894924526,
      "grad_norm": 0.17661869525909424,
      "learning_rate": 9.765362976711138e-06,
      "loss": 0.0034,
      "step": 58660
    },
    {
      "epoch": 0.37629115584303136,
      "grad_norm": 0.5084266066551208,
      "learning_rate": 9.76519350121946e-06,
      "loss": 0.0034,
      "step": 58670
    },
    {
      "epoch": 0.37635529273681745,
      "grad_norm": 0.28248316049575806,
      "learning_rate": 9.765023966016502e-06,
      "loss": 0.0031,
      "step": 58680
    },
    {
      "epoch": 0.37641942963060354,
      "grad_norm": 0.3346003592014313,
      "learning_rate": 9.76485437110439e-06,
      "loss": 0.0038,
      "step": 58690
    },
    {
      "epoch": 0.37648356652438963,
      "grad_norm": 0.137682244181633,
      "learning_rate": 9.764684716485247e-06,
      "loss": 0.0058,
      "step": 58700
    },
    {
      "epoch": 0.3765477034181757,
      "grad_norm": 0.15332180261611938,
      "learning_rate": 9.764515002161198e-06,
      "loss": 0.0036,
      "step": 58710
    },
    {
      "epoch": 0.3766118403119619,
      "grad_norm": 0.1447349637746811,
      "learning_rate": 9.76434522813437e-06,
      "loss": 0.0043,
      "step": 58720
    },
    {
      "epoch": 0.37667597720574797,
      "grad_norm": 0.15648894011974335,
      "learning_rate": 9.764175394406894e-06,
      "loss": 0.0042,
      "step": 58730
    },
    {
      "epoch": 0.37674011409953406,
      "grad_norm": 0.08089536428451538,
      "learning_rate": 9.764005500980895e-06,
      "loss": 0.0078,
      "step": 58740
    },
    {
      "epoch": 0.37680425099332016,
      "grad_norm": 0.17732544243335724,
      "learning_rate": 9.763835547858502e-06,
      "loss": 0.0054,
      "step": 58750
    },
    {
      "epoch": 0.37686838788710625,
      "grad_norm": 0.24036003649234772,
      "learning_rate": 9.763665535041845e-06,
      "loss": 0.0048,
      "step": 58760
    },
    {
      "epoch": 0.37693252478089234,
      "grad_norm": 0.01988343335688114,
      "learning_rate": 9.763495462533053e-06,
      "loss": 0.0048,
      "step": 58770
    },
    {
      "epoch": 0.37699666167467843,
      "grad_norm": 0.42720386385917664,
      "learning_rate": 9.763325330334259e-06,
      "loss": 0.0052,
      "step": 58780
    },
    {
      "epoch": 0.37706079856846453,
      "grad_norm": 0.24175818264484406,
      "learning_rate": 9.763155138447593e-06,
      "loss": 0.0032,
      "step": 58790
    },
    {
      "epoch": 0.3771249354622506,
      "grad_norm": 0.16139830648899078,
      "learning_rate": 9.76298488687519e-06,
      "loss": 0.0072,
      "step": 58800
    },
    {
      "epoch": 0.3771890723560367,
      "grad_norm": 0.11897553503513336,
      "learning_rate": 9.762814575619184e-06,
      "loss": 0.0036,
      "step": 58810
    },
    {
      "epoch": 0.3772532092498228,
      "grad_norm": 0.06003080680966377,
      "learning_rate": 9.762644204681706e-06,
      "loss": 0.0053,
      "step": 58820
    },
    {
      "epoch": 0.3773173461436089,
      "grad_norm": 0.21815739572048187,
      "learning_rate": 9.762473774064894e-06,
      "loss": 0.0045,
      "step": 58830
    },
    {
      "epoch": 0.377381483037395,
      "grad_norm": 0.28276723623275757,
      "learning_rate": 9.76230328377088e-06,
      "loss": 0.0056,
      "step": 58840
    },
    {
      "epoch": 0.3774456199311811,
      "grad_norm": 0.0559268556535244,
      "learning_rate": 9.762132733801802e-06,
      "loss": 0.0028,
      "step": 58850
    },
    {
      "epoch": 0.37750975682496724,
      "grad_norm": 0.07177907228469849,
      "learning_rate": 9.761962124159799e-06,
      "loss": 0.0044,
      "step": 58860
    },
    {
      "epoch": 0.37757389371875333,
      "grad_norm": 0.4044466018676758,
      "learning_rate": 9.761791454847005e-06,
      "loss": 0.0064,
      "step": 58870
    },
    {
      "epoch": 0.3776380306125394,
      "grad_norm": 0.15512405335903168,
      "learning_rate": 9.761620725865563e-06,
      "loss": 0.0041,
      "step": 58880
    },
    {
      "epoch": 0.3777021675063255,
      "grad_norm": 0.10939635336399078,
      "learning_rate": 9.761449937217609e-06,
      "loss": 0.0042,
      "step": 58890
    },
    {
      "epoch": 0.3777663044001116,
      "grad_norm": 0.08103182166814804,
      "learning_rate": 9.761279088905285e-06,
      "loss": 0.004,
      "step": 58900
    },
    {
      "epoch": 0.3778304412938977,
      "grad_norm": 0.2021099030971527,
      "learning_rate": 9.761108180930731e-06,
      "loss": 0.0048,
      "step": 58910
    },
    {
      "epoch": 0.3778945781876838,
      "grad_norm": 0.15852302312850952,
      "learning_rate": 9.760937213296089e-06,
      "loss": 0.0021,
      "step": 58920
    },
    {
      "epoch": 0.3779587150814699,
      "grad_norm": 0.2298976182937622,
      "learning_rate": 9.7607661860035e-06,
      "loss": 0.003,
      "step": 58930
    },
    {
      "epoch": 0.378022851975256,
      "grad_norm": 0.16082763671875,
      "learning_rate": 9.760595099055108e-06,
      "loss": 0.0039,
      "step": 58940
    },
    {
      "epoch": 0.3780869888690421,
      "grad_norm": 0.28142333030700684,
      "learning_rate": 9.76042395245306e-06,
      "loss": 0.0045,
      "step": 58950
    },
    {
      "epoch": 0.37815112576282817,
      "grad_norm": 0.27675023674964905,
      "learning_rate": 9.760252746199495e-06,
      "loss": 0.0044,
      "step": 58960
    },
    {
      "epoch": 0.37821526265661426,
      "grad_norm": 0.2622521221637726,
      "learning_rate": 9.760081480296561e-06,
      "loss": 0.0028,
      "step": 58970
    },
    {
      "epoch": 0.37827939955040035,
      "grad_norm": 0.2026662826538086,
      "learning_rate": 9.759910154746406e-06,
      "loss": 0.0058,
      "step": 58980
    },
    {
      "epoch": 0.37834353644418645,
      "grad_norm": 0.14158673584461212,
      "learning_rate": 9.759738769551174e-06,
      "loss": 0.0047,
      "step": 58990
    },
    {
      "epoch": 0.3784076733379726,
      "grad_norm": 0.21854153275489807,
      "learning_rate": 9.759567324713013e-06,
      "loss": 0.0097,
      "step": 59000
    },
    {
      "epoch": 0.3784718102317587,
      "grad_norm": 0.20481084287166595,
      "learning_rate": 9.759395820234071e-06,
      "loss": 0.0051,
      "step": 59010
    },
    {
      "epoch": 0.3785359471255448,
      "grad_norm": 0.12422089278697968,
      "learning_rate": 9.759224256116499e-06,
      "loss": 0.0049,
      "step": 59020
    },
    {
      "epoch": 0.3786000840193309,
      "grad_norm": 0.07579615712165833,
      "learning_rate": 9.759052632362447e-06,
      "loss": 0.0055,
      "step": 59030
    },
    {
      "epoch": 0.37866422091311697,
      "grad_norm": 0.2826728820800781,
      "learning_rate": 9.758880948974061e-06,
      "loss": 0.0068,
      "step": 59040
    },
    {
      "epoch": 0.37872835780690306,
      "grad_norm": 0.23323026299476624,
      "learning_rate": 9.758709205953498e-06,
      "loss": 0.0047,
      "step": 59050
    },
    {
      "epoch": 0.37879249470068915,
      "grad_norm": 0.12119197100400925,
      "learning_rate": 9.758537403302908e-06,
      "loss": 0.0046,
      "step": 59060
    },
    {
      "epoch": 0.37885663159447525,
      "grad_norm": 0.26631271839141846,
      "learning_rate": 9.758365541024442e-06,
      "loss": 0.0027,
      "step": 59070
    },
    {
      "epoch": 0.37892076848826134,
      "grad_norm": 0.09156379848718643,
      "learning_rate": 9.758193619120256e-06,
      "loss": 0.0056,
      "step": 59080
    },
    {
      "epoch": 0.37898490538204743,
      "grad_norm": 0.2119128257036209,
      "learning_rate": 9.758021637592504e-06,
      "loss": 0.0044,
      "step": 59090
    },
    {
      "epoch": 0.3790490422758335,
      "grad_norm": 0.09771595895290375,
      "learning_rate": 9.757849596443338e-06,
      "loss": 0.0047,
      "step": 59100
    },
    {
      "epoch": 0.3791131791696196,
      "grad_norm": 0.1576254814863205,
      "learning_rate": 9.757677495674918e-06,
      "loss": 0.0041,
      "step": 59110
    },
    {
      "epoch": 0.3791773160634057,
      "grad_norm": 0.1455041766166687,
      "learning_rate": 9.7575053352894e-06,
      "loss": 0.0048,
      "step": 59120
    },
    {
      "epoch": 0.3792414529571918,
      "grad_norm": 0.19070494174957275,
      "learning_rate": 9.757333115288935e-06,
      "loss": 0.0035,
      "step": 59130
    },
    {
      "epoch": 0.37930558985097795,
      "grad_norm": 0.28895506262779236,
      "learning_rate": 9.757160835675692e-06,
      "loss": 0.0033,
      "step": 59140
    },
    {
      "epoch": 0.37936972674476405,
      "grad_norm": 0.15067146718502045,
      "learning_rate": 9.75698849645182e-06,
      "loss": 0.0021,
      "step": 59150
    },
    {
      "epoch": 0.37943386363855014,
      "grad_norm": 0.15311887860298157,
      "learning_rate": 9.756816097619483e-06,
      "loss": 0.0067,
      "step": 59160
    },
    {
      "epoch": 0.37949800053233623,
      "grad_norm": 0.17993053793907166,
      "learning_rate": 9.756643639180842e-06,
      "loss": 0.0045,
      "step": 59170
    },
    {
      "epoch": 0.3795621374261223,
      "grad_norm": 0.09240105003118515,
      "learning_rate": 9.756471121138056e-06,
      "loss": 0.0048,
      "step": 59180
    },
    {
      "epoch": 0.3796262743199084,
      "grad_norm": 0.46541252732276917,
      "learning_rate": 9.756298543493286e-06,
      "loss": 0.003,
      "step": 59190
    },
    {
      "epoch": 0.3796904112136945,
      "grad_norm": 0.1715819388628006,
      "learning_rate": 9.756125906248698e-06,
      "loss": 0.0034,
      "step": 59200
    },
    {
      "epoch": 0.3797545481074806,
      "grad_norm": 0.4843077063560486,
      "learning_rate": 9.755953209406452e-06,
      "loss": 0.0063,
      "step": 59210
    },
    {
      "epoch": 0.3798186850012667,
      "grad_norm": 0.06416106224060059,
      "learning_rate": 9.755780452968715e-06,
      "loss": 0.0028,
      "step": 59220
    },
    {
      "epoch": 0.3798828218950528,
      "grad_norm": 0.12089081108570099,
      "learning_rate": 9.755607636937647e-06,
      "loss": 0.0037,
      "step": 59230
    },
    {
      "epoch": 0.3799469587888389,
      "grad_norm": 0.2047852724790573,
      "learning_rate": 9.755434761315418e-06,
      "loss": 0.0043,
      "step": 59240
    },
    {
      "epoch": 0.380011095682625,
      "grad_norm": 0.04684087261557579,
      "learning_rate": 9.755261826104194e-06,
      "loss": 0.006,
      "step": 59250
    },
    {
      "epoch": 0.38007523257641107,
      "grad_norm": 0.1112871989607811,
      "learning_rate": 9.75508883130614e-06,
      "loss": 0.0046,
      "step": 59260
    },
    {
      "epoch": 0.38013936947019716,
      "grad_norm": 0.12849275767803192,
      "learning_rate": 9.754915776923425e-06,
      "loss": 0.0032,
      "step": 59270
    },
    {
      "epoch": 0.3802035063639833,
      "grad_norm": 0.05961214005947113,
      "learning_rate": 9.754742662958217e-06,
      "loss": 0.0065,
      "step": 59280
    },
    {
      "epoch": 0.3802676432577694,
      "grad_norm": 0.26884549856185913,
      "learning_rate": 9.754569489412684e-06,
      "loss": 0.0047,
      "step": 59290
    },
    {
      "epoch": 0.3803317801515555,
      "grad_norm": 0.10179663449525833,
      "learning_rate": 9.754396256288998e-06,
      "loss": 0.0043,
      "step": 59300
    },
    {
      "epoch": 0.3803959170453416,
      "grad_norm": 0.13700559735298157,
      "learning_rate": 9.754222963589328e-06,
      "loss": 0.0042,
      "step": 59310
    },
    {
      "epoch": 0.3804600539391277,
      "grad_norm": 0.11233152449131012,
      "learning_rate": 9.754049611315847e-06,
      "loss": 0.003,
      "step": 59320
    },
    {
      "epoch": 0.3805241908329138,
      "grad_norm": 0.3106655776500702,
      "learning_rate": 9.753876199470729e-06,
      "loss": 0.0067,
      "step": 59330
    },
    {
      "epoch": 0.38058832772669987,
      "grad_norm": 0.05697939917445183,
      "learning_rate": 9.753702728056143e-06,
      "loss": 0.0059,
      "step": 59340
    },
    {
      "epoch": 0.38065246462048596,
      "grad_norm": 0.1379857361316681,
      "learning_rate": 9.753529197074263e-06,
      "loss": 0.0051,
      "step": 59350
    },
    {
      "epoch": 0.38071660151427206,
      "grad_norm": 0.30130454897880554,
      "learning_rate": 9.753355606527267e-06,
      "loss": 0.0045,
      "step": 59360
    },
    {
      "epoch": 0.38078073840805815,
      "grad_norm": 0.10157103836536407,
      "learning_rate": 9.753181956417327e-06,
      "loss": 0.0026,
      "step": 59370
    },
    {
      "epoch": 0.38084487530184424,
      "grad_norm": 0.16066139936447144,
      "learning_rate": 9.75300824674662e-06,
      "loss": 0.0034,
      "step": 59380
    },
    {
      "epoch": 0.38090901219563034,
      "grad_norm": 0.3818877935409546,
      "learning_rate": 9.752834477517322e-06,
      "loss": 0.0065,
      "step": 59390
    },
    {
      "epoch": 0.38097314908941643,
      "grad_norm": 0.11675024032592773,
      "learning_rate": 9.752660648731614e-06,
      "loss": 0.0065,
      "step": 59400
    },
    {
      "epoch": 0.3810372859832025,
      "grad_norm": 0.20888476073741913,
      "learning_rate": 9.752486760391669e-06,
      "loss": 0.0029,
      "step": 59410
    },
    {
      "epoch": 0.38110142287698867,
      "grad_norm": 0.20465844869613647,
      "learning_rate": 9.752312812499667e-06,
      "loss": 0.0049,
      "step": 59420
    },
    {
      "epoch": 0.38116555977077476,
      "grad_norm": 0.16326738893985748,
      "learning_rate": 9.75213880505779e-06,
      "loss": 0.0064,
      "step": 59430
    },
    {
      "epoch": 0.38122969666456086,
      "grad_norm": 0.1689678281545639,
      "learning_rate": 9.75196473806822e-06,
      "loss": 0.0018,
      "step": 59440
    },
    {
      "epoch": 0.38129383355834695,
      "grad_norm": 0.32500091195106506,
      "learning_rate": 9.751790611533134e-06,
      "loss": 0.0028,
      "step": 59450
    },
    {
      "epoch": 0.38135797045213304,
      "grad_norm": 0.28971317410469055,
      "learning_rate": 9.751616425454716e-06,
      "loss": 0.0031,
      "step": 59460
    },
    {
      "epoch": 0.38142210734591914,
      "grad_norm": 0.1515723168849945,
      "learning_rate": 9.751442179835148e-06,
      "loss": 0.0031,
      "step": 59470
    },
    {
      "epoch": 0.38148624423970523,
      "grad_norm": 0.16772037744522095,
      "learning_rate": 9.751267874676614e-06,
      "loss": 0.0056,
      "step": 59480
    },
    {
      "epoch": 0.3815503811334913,
      "grad_norm": 0.21892644464969635,
      "learning_rate": 9.751093509981297e-06,
      "loss": 0.0037,
      "step": 59490
    },
    {
      "epoch": 0.3816145180272774,
      "grad_norm": 0.033471908420324326,
      "learning_rate": 9.750919085751383e-06,
      "loss": 0.0039,
      "step": 59500
    },
    {
      "epoch": 0.3816786549210635,
      "grad_norm": 0.8038234114646912,
      "learning_rate": 9.750744601989057e-06,
      "loss": 0.0056,
      "step": 59510
    },
    {
      "epoch": 0.3817427918148496,
      "grad_norm": 0.09216465801000595,
      "learning_rate": 9.750570058696507e-06,
      "loss": 0.0093,
      "step": 59520
    },
    {
      "epoch": 0.3818069287086357,
      "grad_norm": 0.18910400569438934,
      "learning_rate": 9.750395455875918e-06,
      "loss": 0.0047,
      "step": 59530
    },
    {
      "epoch": 0.3818710656024218,
      "grad_norm": 0.5617573857307434,
      "learning_rate": 9.75022079352948e-06,
      "loss": 0.0056,
      "step": 59540
    },
    {
      "epoch": 0.3819352024962079,
      "grad_norm": 0.3218870759010315,
      "learning_rate": 9.75004607165938e-06,
      "loss": 0.0044,
      "step": 59550
    },
    {
      "epoch": 0.38199933938999403,
      "grad_norm": 0.21268276870250702,
      "learning_rate": 9.749871290267807e-06,
      "loss": 0.0052,
      "step": 59560
    },
    {
      "epoch": 0.3820634762837801,
      "grad_norm": 0.17883072793483734,
      "learning_rate": 9.749696449356953e-06,
      "loss": 0.0036,
      "step": 59570
    },
    {
      "epoch": 0.3821276131775662,
      "grad_norm": 0.24500377476215363,
      "learning_rate": 9.749521548929008e-06,
      "loss": 0.0054,
      "step": 59580
    },
    {
      "epoch": 0.3821917500713523,
      "grad_norm": 0.34320369362831116,
      "learning_rate": 9.749346588986163e-06,
      "loss": 0.005,
      "step": 59590
    },
    {
      "epoch": 0.3822558869651384,
      "grad_norm": 0.2913901209831238,
      "learning_rate": 9.74917156953061e-06,
      "loss": 0.0019,
      "step": 59600
    },
    {
      "epoch": 0.3823200238589245,
      "grad_norm": 0.2491830289363861,
      "learning_rate": 9.748996490564546e-06,
      "loss": 0.0044,
      "step": 59610
    },
    {
      "epoch": 0.3823841607527106,
      "grad_norm": 0.0809091329574585,
      "learning_rate": 9.748821352090159e-06,
      "loss": 0.0031,
      "step": 59620
    },
    {
      "epoch": 0.3824482976464967,
      "grad_norm": 0.16306327283382416,
      "learning_rate": 9.748646154109648e-06,
      "loss": 0.0051,
      "step": 59630
    },
    {
      "epoch": 0.3825124345402828,
      "grad_norm": 0.09098932892084122,
      "learning_rate": 9.748470896625206e-06,
      "loss": 0.0025,
      "step": 59640
    },
    {
      "epoch": 0.38257657143406887,
      "grad_norm": 0.16054843366146088,
      "learning_rate": 9.748295579639031e-06,
      "loss": 0.0053,
      "step": 59650
    },
    {
      "epoch": 0.38264070832785496,
      "grad_norm": 0.3167816698551178,
      "learning_rate": 9.748120203153318e-06,
      "loss": 0.0046,
      "step": 59660
    },
    {
      "epoch": 0.38270484522164105,
      "grad_norm": 0.08863595128059387,
      "learning_rate": 9.747944767170267e-06,
      "loss": 0.0047,
      "step": 59670
    },
    {
      "epoch": 0.38276898211542715,
      "grad_norm": 0.12066241353750229,
      "learning_rate": 9.747769271692074e-06,
      "loss": 0.0043,
      "step": 59680
    },
    {
      "epoch": 0.38283311900921324,
      "grad_norm": 0.16344675421714783,
      "learning_rate": 9.747593716720937e-06,
      "loss": 0.0042,
      "step": 59690
    },
    {
      "epoch": 0.3828972559029994,
      "grad_norm": 0.1318945288658142,
      "learning_rate": 9.747418102259059e-06,
      "loss": 0.0064,
      "step": 59700
    },
    {
      "epoch": 0.3829613927967855,
      "grad_norm": 0.2592341899871826,
      "learning_rate": 9.747242428308639e-06,
      "loss": 0.0037,
      "step": 59710
    },
    {
      "epoch": 0.3830255296905716,
      "grad_norm": 0.7534883618354797,
      "learning_rate": 9.747066694871881e-06,
      "loss": 0.0079,
      "step": 59720
    },
    {
      "epoch": 0.38308966658435767,
      "grad_norm": 0.18560761213302612,
      "learning_rate": 9.74689090195098e-06,
      "loss": 0.0028,
      "step": 59730
    },
    {
      "epoch": 0.38315380347814376,
      "grad_norm": 0.207354336977005,
      "learning_rate": 9.746715049548148e-06,
      "loss": 0.0034,
      "step": 59740
    },
    {
      "epoch": 0.38321794037192985,
      "grad_norm": 0.1272207796573639,
      "learning_rate": 9.746539137665582e-06,
      "loss": 0.0056,
      "step": 59750
    },
    {
      "epoch": 0.38328207726571595,
      "grad_norm": 0.23403020203113556,
      "learning_rate": 9.746363166305487e-06,
      "loss": 0.0123,
      "step": 59760
    },
    {
      "epoch": 0.38334621415950204,
      "grad_norm": 0.26903918385505676,
      "learning_rate": 9.746187135470072e-06,
      "loss": 0.0041,
      "step": 59770
    },
    {
      "epoch": 0.38341035105328813,
      "grad_norm": 3.419872999191284,
      "learning_rate": 9.746011045161538e-06,
      "loss": 0.0069,
      "step": 59780
    },
    {
      "epoch": 0.3834744879470742,
      "grad_norm": 0.24554139375686646,
      "learning_rate": 9.745834895382095e-06,
      "loss": 0.0038,
      "step": 59790
    },
    {
      "epoch": 0.3835386248408603,
      "grad_norm": 0.05302513390779495,
      "learning_rate": 9.745658686133947e-06,
      "loss": 0.0035,
      "step": 59800
    },
    {
      "epoch": 0.3836027617346464,
      "grad_norm": 0.19432343542575836,
      "learning_rate": 9.745482417419306e-06,
      "loss": 0.0035,
      "step": 59810
    },
    {
      "epoch": 0.3836668986284325,
      "grad_norm": 0.17563576996326447,
      "learning_rate": 9.745306089240379e-06,
      "loss": 0.01,
      "step": 59820
    },
    {
      "epoch": 0.3837310355222186,
      "grad_norm": 0.16629278659820557,
      "learning_rate": 9.745129701599374e-06,
      "loss": 0.0052,
      "step": 59830
    },
    {
      "epoch": 0.3837951724160047,
      "grad_norm": 0.239976704120636,
      "learning_rate": 9.744953254498503e-06,
      "loss": 0.006,
      "step": 59840
    },
    {
      "epoch": 0.38385930930979084,
      "grad_norm": 0.19879131019115448,
      "learning_rate": 9.744776747939977e-06,
      "loss": 0.0042,
      "step": 59850
    },
    {
      "epoch": 0.38392344620357693,
      "grad_norm": 0.0788252130150795,
      "learning_rate": 9.744600181926007e-06,
      "loss": 0.0034,
      "step": 59860
    },
    {
      "epoch": 0.383987583097363,
      "grad_norm": 0.3260972201824188,
      "learning_rate": 9.744423556458805e-06,
      "loss": 0.0039,
      "step": 59870
    },
    {
      "epoch": 0.3840517199911491,
      "grad_norm": 0.16560789942741394,
      "learning_rate": 9.744246871540585e-06,
      "loss": 0.0029,
      "step": 59880
    },
    {
      "epoch": 0.3841158568849352,
      "grad_norm": 0.10938803106546402,
      "learning_rate": 9.744070127173562e-06,
      "loss": 0.0047,
      "step": 59890
    },
    {
      "epoch": 0.3841799937787213,
      "grad_norm": 0.28586849570274353,
      "learning_rate": 9.74389332335995e-06,
      "loss": 0.0038,
      "step": 59900
    },
    {
      "epoch": 0.3842441306725074,
      "grad_norm": 0.19201096892356873,
      "learning_rate": 9.743716460101965e-06,
      "loss": 0.0058,
      "step": 59910
    },
    {
      "epoch": 0.3843082675662935,
      "grad_norm": 0.1080024465918541,
      "learning_rate": 9.743539537401821e-06,
      "loss": 0.0037,
      "step": 59920
    },
    {
      "epoch": 0.3843724044600796,
      "grad_norm": 0.3251629173755646,
      "learning_rate": 9.743362555261737e-06,
      "loss": 0.0059,
      "step": 59930
    },
    {
      "epoch": 0.3844365413538657,
      "grad_norm": 0.33186784386634827,
      "learning_rate": 9.74318551368393e-06,
      "loss": 0.006,
      "step": 59940
    },
    {
      "epoch": 0.3845006782476518,
      "grad_norm": 0.20968593657016754,
      "learning_rate": 9.743008412670618e-06,
      "loss": 0.0035,
      "step": 59950
    },
    {
      "epoch": 0.38456481514143787,
      "grad_norm": 0.1586632877588272,
      "learning_rate": 9.74283125222402e-06,
      "loss": 0.0036,
      "step": 59960
    },
    {
      "epoch": 0.38462895203522396,
      "grad_norm": 0.012320125475525856,
      "learning_rate": 9.742654032346359e-06,
      "loss": 0.0031,
      "step": 59970
    },
    {
      "epoch": 0.38469308892901005,
      "grad_norm": 0.21408711373806,
      "learning_rate": 9.742476753039852e-06,
      "loss": 0.0035,
      "step": 59980
    },
    {
      "epoch": 0.3847572258227962,
      "grad_norm": 0.16455549001693726,
      "learning_rate": 9.742299414306722e-06,
      "loss": 0.0037,
      "step": 59990
    },
    {
      "epoch": 0.3848213627165823,
      "grad_norm": 0.20752200484275818,
      "learning_rate": 9.742122016149193e-06,
      "loss": 0.0041,
      "step": 60000
    },
    {
      "epoch": 0.3848854996103684,
      "grad_norm": 0.27880942821502686,
      "learning_rate": 9.741944558569483e-06,
      "loss": 0.0043,
      "step": 60010
    },
    {
      "epoch": 0.3849496365041545,
      "grad_norm": 0.9511381983757019,
      "learning_rate": 9.741767041569821e-06,
      "loss": 0.0052,
      "step": 60020
    },
    {
      "epoch": 0.3850137733979406,
      "grad_norm": 0.42910051345825195,
      "learning_rate": 9.741589465152427e-06,
      "loss": 0.0051,
      "step": 60030
    },
    {
      "epoch": 0.38507791029172667,
      "grad_norm": 0.17123812437057495,
      "learning_rate": 9.741411829319531e-06,
      "loss": 0.0052,
      "step": 60040
    },
    {
      "epoch": 0.38514204718551276,
      "grad_norm": 0.10715224593877792,
      "learning_rate": 9.741234134073354e-06,
      "loss": 0.0018,
      "step": 60050
    },
    {
      "epoch": 0.38520618407929885,
      "grad_norm": 0.21001029014587402,
      "learning_rate": 9.741056379416125e-06,
      "loss": 0.0037,
      "step": 60060
    },
    {
      "epoch": 0.38527032097308495,
      "grad_norm": 0.18050143122673035,
      "learning_rate": 9.740878565350072e-06,
      "loss": 0.0036,
      "step": 60070
    },
    {
      "epoch": 0.38533445786687104,
      "grad_norm": 0.012618201784789562,
      "learning_rate": 9.740700691877423e-06,
      "loss": 0.003,
      "step": 60080
    },
    {
      "epoch": 0.38539859476065713,
      "grad_norm": 0.25209400057792664,
      "learning_rate": 9.740522759000404e-06,
      "loss": 0.0035,
      "step": 60090
    },
    {
      "epoch": 0.3854627316544432,
      "grad_norm": 0.49384912848472595,
      "learning_rate": 9.740344766721248e-06,
      "loss": 0.0071,
      "step": 60100
    },
    {
      "epoch": 0.3855268685482293,
      "grad_norm": 0.14024218916893005,
      "learning_rate": 9.740166715042184e-06,
      "loss": 0.0056,
      "step": 60110
    },
    {
      "epoch": 0.3855910054420154,
      "grad_norm": 0.026863152161240578,
      "learning_rate": 9.739988603965444e-06,
      "loss": 0.0039,
      "step": 60120
    },
    {
      "epoch": 0.38565514233580156,
      "grad_norm": 0.3183062970638275,
      "learning_rate": 9.739810433493258e-06,
      "loss": 0.0041,
      "step": 60130
    },
    {
      "epoch": 0.38571927922958765,
      "grad_norm": 0.20068994164466858,
      "learning_rate": 9.73963220362786e-06,
      "loss": 0.0041,
      "step": 60140
    },
    {
      "epoch": 0.38578341612337375,
      "grad_norm": 0.09155667573213577,
      "learning_rate": 9.739453914371484e-06,
      "loss": 0.0043,
      "step": 60150
    },
    {
      "epoch": 0.38584755301715984,
      "grad_norm": 0.055506039410829544,
      "learning_rate": 9.739275565726363e-06,
      "loss": 0.0037,
      "step": 60160
    },
    {
      "epoch": 0.38591168991094593,
      "grad_norm": 0.11468420922756195,
      "learning_rate": 9.73909715769473e-06,
      "loss": 0.0033,
      "step": 60170
    },
    {
      "epoch": 0.385975826804732,
      "grad_norm": 0.2966194748878479,
      "learning_rate": 9.738918690278826e-06,
      "loss": 0.0048,
      "step": 60180
    },
    {
      "epoch": 0.3860399636985181,
      "grad_norm": 0.022596042603254318,
      "learning_rate": 9.738740163480882e-06,
      "loss": 0.0027,
      "step": 60190
    },
    {
      "epoch": 0.3861041005923042,
      "grad_norm": 0.3111661374568939,
      "learning_rate": 9.738561577303139e-06,
      "loss": 0.0039,
      "step": 60200
    },
    {
      "epoch": 0.3861682374860903,
      "grad_norm": 0.15545833110809326,
      "learning_rate": 9.73838293174783e-06,
      "loss": 0.0034,
      "step": 60210
    },
    {
      "epoch": 0.3862323743798764,
      "grad_norm": 0.1548236757516861,
      "learning_rate": 9.738204226817197e-06,
      "loss": 0.0033,
      "step": 60220
    },
    {
      "epoch": 0.3862965112736625,
      "grad_norm": 0.16571511328220367,
      "learning_rate": 9.73802546251348e-06,
      "loss": 0.0044,
      "step": 60230
    },
    {
      "epoch": 0.3863606481674486,
      "grad_norm": 0.302288293838501,
      "learning_rate": 9.737846638838918e-06,
      "loss": 0.0095,
      "step": 60240
    },
    {
      "epoch": 0.3864247850612347,
      "grad_norm": 0.2831825613975525,
      "learning_rate": 9.73766775579575e-06,
      "loss": 0.0039,
      "step": 60250
    },
    {
      "epoch": 0.38648892195502077,
      "grad_norm": 0.04732539504766464,
      "learning_rate": 9.737488813386219e-06,
      "loss": 0.0026,
      "step": 60260
    },
    {
      "epoch": 0.3865530588488069,
      "grad_norm": 0.1678842306137085,
      "learning_rate": 9.737309811612567e-06,
      "loss": 0.022,
      "step": 60270
    },
    {
      "epoch": 0.386617195742593,
      "grad_norm": 0.4333482086658478,
      "learning_rate": 9.73713075047704e-06,
      "loss": 0.0056,
      "step": 60280
    },
    {
      "epoch": 0.3866813326363791,
      "grad_norm": 0.1695176213979721,
      "learning_rate": 9.736951629981877e-06,
      "loss": 0.0036,
      "step": 60290
    },
    {
      "epoch": 0.3867454695301652,
      "grad_norm": 0.2635030746459961,
      "learning_rate": 9.736772450129325e-06,
      "loss": 0.0065,
      "step": 60300
    },
    {
      "epoch": 0.3868096064239513,
      "grad_norm": 0.06427385658025742,
      "learning_rate": 9.73659321092163e-06,
      "loss": 0.0034,
      "step": 60310
    },
    {
      "epoch": 0.3868737433177374,
      "grad_norm": 0.25503623485565186,
      "learning_rate": 9.736413912361035e-06,
      "loss": 0.0055,
      "step": 60320
    },
    {
      "epoch": 0.3869378802115235,
      "grad_norm": 0.13860364258289337,
      "learning_rate": 9.736234554449788e-06,
      "loss": 0.0039,
      "step": 60330
    },
    {
      "epoch": 0.38700201710530957,
      "grad_norm": 0.13788312673568726,
      "learning_rate": 9.736055137190139e-06,
      "loss": 0.0046,
      "step": 60340
    },
    {
      "epoch": 0.38706615399909566,
      "grad_norm": 0.15529346466064453,
      "learning_rate": 9.735875660584335e-06,
      "loss": 0.0043,
      "step": 60350
    },
    {
      "epoch": 0.38713029089288176,
      "grad_norm": 0.15514250099658966,
      "learning_rate": 9.735696124634623e-06,
      "loss": 0.0046,
      "step": 60360
    },
    {
      "epoch": 0.38719442778666785,
      "grad_norm": 0.1338747888803482,
      "learning_rate": 9.735516529343255e-06,
      "loss": 0.0031,
      "step": 60370
    },
    {
      "epoch": 0.38725856468045394,
      "grad_norm": 0.17875945568084717,
      "learning_rate": 9.735336874712478e-06,
      "loss": 0.0043,
      "step": 60380
    },
    {
      "epoch": 0.38732270157424004,
      "grad_norm": 0.06392395496368408,
      "learning_rate": 9.735157160744548e-06,
      "loss": 0.0062,
      "step": 60390
    },
    {
      "epoch": 0.38738683846802613,
      "grad_norm": 0.11207230389118195,
      "learning_rate": 9.734977387441713e-06,
      "loss": 0.0056,
      "step": 60400
    },
    {
      "epoch": 0.3874509753618123,
      "grad_norm": 0.15044747292995453,
      "learning_rate": 9.734797554806229e-06,
      "loss": 0.0031,
      "step": 60410
    },
    {
      "epoch": 0.38751511225559837,
      "grad_norm": 0.24024957418441772,
      "learning_rate": 9.734617662840347e-06,
      "loss": 0.0047,
      "step": 60420
    },
    {
      "epoch": 0.38757924914938446,
      "grad_norm": 0.13344290852546692,
      "learning_rate": 9.734437711546321e-06,
      "loss": 0.0046,
      "step": 60430
    },
    {
      "epoch": 0.38764338604317056,
      "grad_norm": 0.2346503883600235,
      "learning_rate": 9.734257700926408e-06,
      "loss": 0.0044,
      "step": 60440
    },
    {
      "epoch": 0.38770752293695665,
      "grad_norm": 0.14192451536655426,
      "learning_rate": 9.734077630982863e-06,
      "loss": 0.0041,
      "step": 60450
    },
    {
      "epoch": 0.38777165983074274,
      "grad_norm": 0.23871690034866333,
      "learning_rate": 9.73389750171794e-06,
      "loss": 0.007,
      "step": 60460
    },
    {
      "epoch": 0.38783579672452884,
      "grad_norm": 0.16708698868751526,
      "learning_rate": 9.733717313133899e-06,
      "loss": 0.003,
      "step": 60470
    },
    {
      "epoch": 0.38789993361831493,
      "grad_norm": 0.09737008064985275,
      "learning_rate": 9.733537065232998e-06,
      "loss": 0.0025,
      "step": 60480
    },
    {
      "epoch": 0.387964070512101,
      "grad_norm": 0.13287779688835144,
      "learning_rate": 9.733356758017495e-06,
      "loss": 0.0042,
      "step": 60490
    },
    {
      "epoch": 0.3880282074058871,
      "grad_norm": 0.06711868941783905,
      "learning_rate": 9.733176391489644e-06,
      "loss": 0.0047,
      "step": 60500
    },
    {
      "epoch": 0.3880923442996732,
      "grad_norm": 0.08290399610996246,
      "learning_rate": 9.732995965651715e-06,
      "loss": 0.0058,
      "step": 60510
    },
    {
      "epoch": 0.3881564811934593,
      "grad_norm": 0.2034006267786026,
      "learning_rate": 9.732815480505962e-06,
      "loss": 0.0036,
      "step": 60520
    },
    {
      "epoch": 0.3882206180872454,
      "grad_norm": 0.3656480014324188,
      "learning_rate": 9.73263493605465e-06,
      "loss": 0.0037,
      "step": 60530
    },
    {
      "epoch": 0.3882847549810315,
      "grad_norm": 0.11950068920850754,
      "learning_rate": 9.732454332300039e-06,
      "loss": 0.0038,
      "step": 60540
    },
    {
      "epoch": 0.38834889187481764,
      "grad_norm": 0.5113601684570312,
      "learning_rate": 9.732273669244392e-06,
      "loss": 0.0047,
      "step": 60550
    },
    {
      "epoch": 0.38841302876860373,
      "grad_norm": 0.2951570153236389,
      "learning_rate": 9.732092946889974e-06,
      "loss": 0.0034,
      "step": 60560
    },
    {
      "epoch": 0.3884771656623898,
      "grad_norm": 0.12037892639636993,
      "learning_rate": 9.731912165239052e-06,
      "loss": 0.0036,
      "step": 60570
    },
    {
      "epoch": 0.3885413025561759,
      "grad_norm": 0.14643444120883942,
      "learning_rate": 9.731731324293887e-06,
      "loss": 0.0023,
      "step": 60580
    },
    {
      "epoch": 0.388605439449962,
      "grad_norm": 0.2314303070306778,
      "learning_rate": 9.731550424056745e-06,
      "loss": 0.004,
      "step": 60590
    },
    {
      "epoch": 0.3886695763437481,
      "grad_norm": 0.13713759183883667,
      "learning_rate": 9.731369464529897e-06,
      "loss": 0.0034,
      "step": 60600
    },
    {
      "epoch": 0.3887337132375342,
      "grad_norm": 0.22650521993637085,
      "learning_rate": 9.731188445715608e-06,
      "loss": 0.0052,
      "step": 60610
    },
    {
      "epoch": 0.3887978501313203,
      "grad_norm": 0.15206363797187805,
      "learning_rate": 9.731007367616146e-06,
      "loss": 0.0045,
      "step": 60620
    },
    {
      "epoch": 0.3888619870251064,
      "grad_norm": 0.1030779778957367,
      "learning_rate": 9.73082623023378e-06,
      "loss": 0.0033,
      "step": 60630
    },
    {
      "epoch": 0.3889261239188925,
      "grad_norm": 0.2049216777086258,
      "learning_rate": 9.73064503357078e-06,
      "loss": 0.0052,
      "step": 60640
    },
    {
      "epoch": 0.38899026081267857,
      "grad_norm": 0.11878073960542679,
      "learning_rate": 9.730463777629417e-06,
      "loss": 0.0028,
      "step": 60650
    },
    {
      "epoch": 0.38905439770646466,
      "grad_norm": 0.06839001923799515,
      "learning_rate": 9.730282462411964e-06,
      "loss": 0.0026,
      "step": 60660
    },
    {
      "epoch": 0.38911853460025075,
      "grad_norm": 0.023172229528427124,
      "learning_rate": 9.73010108792069e-06,
      "loss": 0.0043,
      "step": 60670
    },
    {
      "epoch": 0.38918267149403685,
      "grad_norm": 0.0986587256193161,
      "learning_rate": 9.729919654157867e-06,
      "loss": 0.0041,
      "step": 60680
    },
    {
      "epoch": 0.389246808387823,
      "grad_norm": 0.1795051544904709,
      "learning_rate": 9.729738161125772e-06,
      "loss": 0.0041,
      "step": 60690
    },
    {
      "epoch": 0.3893109452816091,
      "grad_norm": 0.22513192892074585,
      "learning_rate": 9.729556608826676e-06,
      "loss": 0.0048,
      "step": 60700
    },
    {
      "epoch": 0.3893750821753952,
      "grad_norm": 0.496263712644577,
      "learning_rate": 9.729374997262857e-06,
      "loss": 0.0047,
      "step": 60710
    },
    {
      "epoch": 0.3894392190691813,
      "grad_norm": 0.22901271283626556,
      "learning_rate": 9.72919332643659e-06,
      "loss": 0.0037,
      "step": 60720
    },
    {
      "epoch": 0.38950335596296737,
      "grad_norm": 0.15260516107082367,
      "learning_rate": 9.729011596350148e-06,
      "loss": 0.0058,
      "step": 60730
    },
    {
      "epoch": 0.38956749285675346,
      "grad_norm": 0.3243672251701355,
      "learning_rate": 9.728829807005812e-06,
      "loss": 0.0045,
      "step": 60740
    },
    {
      "epoch": 0.38963162975053955,
      "grad_norm": 0.3474341034889221,
      "learning_rate": 9.728647958405861e-06,
      "loss": 0.0079,
      "step": 60750
    },
    {
      "epoch": 0.38969576664432565,
      "grad_norm": 0.25854116678237915,
      "learning_rate": 9.72846605055257e-06,
      "loss": 0.0074,
      "step": 60760
    },
    {
      "epoch": 0.38975990353811174,
      "grad_norm": 0.1458887904882431,
      "learning_rate": 9.72828408344822e-06,
      "loss": 0.0033,
      "step": 60770
    },
    {
      "epoch": 0.38982404043189783,
      "grad_norm": 0.12434220314025879,
      "learning_rate": 9.72810205709509e-06,
      "loss": 0.0049,
      "step": 60780
    },
    {
      "epoch": 0.3898881773256839,
      "grad_norm": 0.04019024968147278,
      "learning_rate": 9.727919971495464e-06,
      "loss": 0.0038,
      "step": 60790
    },
    {
      "epoch": 0.38995231421947,
      "grad_norm": 0.1241234689950943,
      "learning_rate": 9.727737826651621e-06,
      "loss": 0.0071,
      "step": 60800
    },
    {
      "epoch": 0.3900164511132561,
      "grad_norm": 0.2500709295272827,
      "learning_rate": 9.727555622565845e-06,
      "loss": 0.0042,
      "step": 60810
    },
    {
      "epoch": 0.3900805880070422,
      "grad_norm": 0.17337116599082947,
      "learning_rate": 9.727373359240417e-06,
      "loss": 0.0052,
      "step": 60820
    },
    {
      "epoch": 0.39014472490082835,
      "grad_norm": 0.10613390803337097,
      "learning_rate": 9.727191036677624e-06,
      "loss": 0.002,
      "step": 60830
    },
    {
      "epoch": 0.39020886179461445,
      "grad_norm": 0.13982334733009338,
      "learning_rate": 9.727008654879748e-06,
      "loss": 0.0042,
      "step": 60840
    },
    {
      "epoch": 0.39027299868840054,
      "grad_norm": 0.17445537447929382,
      "learning_rate": 9.726826213849074e-06,
      "loss": 0.0045,
      "step": 60850
    },
    {
      "epoch": 0.39033713558218663,
      "grad_norm": 0.3702257573604584,
      "learning_rate": 9.72664371358789e-06,
      "loss": 0.0077,
      "step": 60860
    },
    {
      "epoch": 0.3904012724759727,
      "grad_norm": 0.11560417711734772,
      "learning_rate": 9.726461154098482e-06,
      "loss": 0.004,
      "step": 60870
    },
    {
      "epoch": 0.3904654093697588,
      "grad_norm": 0.11796073615550995,
      "learning_rate": 9.726278535383138e-06,
      "loss": 0.0036,
      "step": 60880
    },
    {
      "epoch": 0.3905295462635449,
      "grad_norm": 0.1939150094985962,
      "learning_rate": 9.726095857444147e-06,
      "loss": 0.0063,
      "step": 60890
    },
    {
      "epoch": 0.390593683157331,
      "grad_norm": 0.23682385683059692,
      "learning_rate": 9.725913120283796e-06,
      "loss": 0.0021,
      "step": 60900
    },
    {
      "epoch": 0.3906578200511171,
      "grad_norm": 0.12155026942491531,
      "learning_rate": 9.725730323904377e-06,
      "loss": 0.0025,
      "step": 60910
    },
    {
      "epoch": 0.3907219569449032,
      "grad_norm": 0.2544938921928406,
      "learning_rate": 9.725547468308178e-06,
      "loss": 0.0059,
      "step": 60920
    },
    {
      "epoch": 0.3907860938386893,
      "grad_norm": 0.08578786253929138,
      "learning_rate": 9.725364553497494e-06,
      "loss": 0.0038,
      "step": 60930
    },
    {
      "epoch": 0.3908502307324754,
      "grad_norm": 0.15020376443862915,
      "learning_rate": 9.725181579474614e-06,
      "loss": 0.0048,
      "step": 60940
    },
    {
      "epoch": 0.39091436762626147,
      "grad_norm": 0.15422581136226654,
      "learning_rate": 9.724998546241831e-06,
      "loss": 0.0046,
      "step": 60950
    },
    {
      "epoch": 0.39097850452004757,
      "grad_norm": 0.24887314438819885,
      "learning_rate": 9.72481545380144e-06,
      "loss": 0.0048,
      "step": 60960
    },
    {
      "epoch": 0.3910426414138337,
      "grad_norm": 0.1745447963476181,
      "learning_rate": 9.724632302155735e-06,
      "loss": 0.0033,
      "step": 60970
    },
    {
      "epoch": 0.3911067783076198,
      "grad_norm": 0.21266213059425354,
      "learning_rate": 9.724449091307011e-06,
      "loss": 0.0057,
      "step": 60980
    },
    {
      "epoch": 0.3911709152014059,
      "grad_norm": 0.16322961449623108,
      "learning_rate": 9.724265821257562e-06,
      "loss": 0.0038,
      "step": 60990
    },
    {
      "epoch": 0.391235052095192,
      "grad_norm": 0.2518809735774994,
      "learning_rate": 9.724082492009687e-06,
      "loss": 0.0064,
      "step": 61000
    },
    {
      "epoch": 0.391235052095192,
      "eval_loss": 0.004069021437317133,
      "eval_runtime": 3.3277,
      "eval_samples_per_second": 60.101,
      "eval_steps_per_second": 15.025,
      "step": 61000
    },
    {
      "epoch": 0.3912991889889781,
      "grad_norm": 0.11623580008745193,
      "learning_rate": 9.72389910356568e-06,
      "loss": 0.006,
      "step": 61010
    },
    {
      "epoch": 0.3913633258827642,
      "grad_norm": 0.0640629306435585,
      "learning_rate": 9.723715655927844e-06,
      "loss": 0.0036,
      "step": 61020
    },
    {
      "epoch": 0.3914274627765503,
      "grad_norm": 0.11713147163391113,
      "learning_rate": 9.723532149098473e-06,
      "loss": 0.0031,
      "step": 61030
    },
    {
      "epoch": 0.39149159967033637,
      "grad_norm": 0.1606680452823639,
      "learning_rate": 9.72334858307987e-06,
      "loss": 0.0039,
      "step": 61040
    },
    {
      "epoch": 0.39155573656412246,
      "grad_norm": 0.2158287763595581,
      "learning_rate": 9.723164957874332e-06,
      "loss": 0.007,
      "step": 61050
    },
    {
      "epoch": 0.39161987345790855,
      "grad_norm": 0.2687753438949585,
      "learning_rate": 9.722981273484162e-06,
      "loss": 0.0058,
      "step": 61060
    },
    {
      "epoch": 0.39168401035169464,
      "grad_norm": 0.048382628709077835,
      "learning_rate": 9.722797529911662e-06,
      "loss": 0.0036,
      "step": 61070
    },
    {
      "epoch": 0.39174814724548074,
      "grad_norm": 0.3799852430820465,
      "learning_rate": 9.722613727159133e-06,
      "loss": 0.0034,
      "step": 61080
    },
    {
      "epoch": 0.39181228413926683,
      "grad_norm": 0.1671069860458374,
      "learning_rate": 9.722429865228878e-06,
      "loss": 0.0045,
      "step": 61090
    },
    {
      "epoch": 0.3918764210330529,
      "grad_norm": 0.2643856704235077,
      "learning_rate": 9.722245944123202e-06,
      "loss": 0.0048,
      "step": 61100
    },
    {
      "epoch": 0.3919405579268391,
      "grad_norm": 0.1426529586315155,
      "learning_rate": 9.72206196384441e-06,
      "loss": 0.0045,
      "step": 61110
    },
    {
      "epoch": 0.39200469482062517,
      "grad_norm": 0.07896137982606888,
      "learning_rate": 9.721877924394807e-06,
      "loss": 0.0046,
      "step": 61120
    },
    {
      "epoch": 0.39206883171441126,
      "grad_norm": 0.22783268988132477,
      "learning_rate": 9.721693825776697e-06,
      "loss": 0.0039,
      "step": 61130
    },
    {
      "epoch": 0.39213296860819735,
      "grad_norm": 0.11826110631227493,
      "learning_rate": 9.721509667992391e-06,
      "loss": 0.0048,
      "step": 61140
    },
    {
      "epoch": 0.39219710550198345,
      "grad_norm": 0.36788371205329895,
      "learning_rate": 9.721325451044194e-06,
      "loss": 0.0168,
      "step": 61150
    },
    {
      "epoch": 0.39226124239576954,
      "grad_norm": 0.2228899598121643,
      "learning_rate": 9.721141174934414e-06,
      "loss": 0.0049,
      "step": 61160
    },
    {
      "epoch": 0.39232537928955563,
      "grad_norm": 0.11477508395910263,
      "learning_rate": 9.72095683966536e-06,
      "loss": 0.0054,
      "step": 61170
    },
    {
      "epoch": 0.3923895161833417,
      "grad_norm": 0.12067042291164398,
      "learning_rate": 9.720772445239345e-06,
      "loss": 0.0029,
      "step": 61180
    },
    {
      "epoch": 0.3924536530771278,
      "grad_norm": 0.22580377757549286,
      "learning_rate": 9.720587991658677e-06,
      "loss": 0.0042,
      "step": 61190
    },
    {
      "epoch": 0.3925177899709139,
      "grad_norm": 0.09509395062923431,
      "learning_rate": 9.720403478925667e-06,
      "loss": 0.0033,
      "step": 61200
    },
    {
      "epoch": 0.3925819268647,
      "grad_norm": 0.14699745178222656,
      "learning_rate": 9.72021890704263e-06,
      "loss": 0.0034,
      "step": 61210
    },
    {
      "epoch": 0.3926460637584861,
      "grad_norm": 0.1703781932592392,
      "learning_rate": 9.720034276011874e-06,
      "loss": 0.0038,
      "step": 61220
    },
    {
      "epoch": 0.3927102006522722,
      "grad_norm": 0.2723254859447479,
      "learning_rate": 9.719849585835715e-06,
      "loss": 0.0044,
      "step": 61230
    },
    {
      "epoch": 0.3927743375460583,
      "grad_norm": 0.09000599384307861,
      "learning_rate": 9.719664836516468e-06,
      "loss": 0.0029,
      "step": 61240
    },
    {
      "epoch": 0.3928384744398444,
      "grad_norm": 0.16326692700386047,
      "learning_rate": 9.719480028056448e-06,
      "loss": 0.0036,
      "step": 61250
    },
    {
      "epoch": 0.3929026113336305,
      "grad_norm": 0.29439839720726013,
      "learning_rate": 9.71929516045797e-06,
      "loss": 0.0041,
      "step": 61260
    },
    {
      "epoch": 0.3929667482274166,
      "grad_norm": 0.1782740354537964,
      "learning_rate": 9.719110233723352e-06,
      "loss": 0.0033,
      "step": 61270
    },
    {
      "epoch": 0.3930308851212027,
      "grad_norm": 0.19152049720287323,
      "learning_rate": 9.718925247854908e-06,
      "loss": 0.0052,
      "step": 61280
    },
    {
      "epoch": 0.3930950220149888,
      "grad_norm": 0.25347113609313965,
      "learning_rate": 9.71874020285496e-06,
      "loss": 0.0039,
      "step": 61290
    },
    {
      "epoch": 0.3931591589087749,
      "grad_norm": 0.10487768054008484,
      "learning_rate": 9.718555098725824e-06,
      "loss": 0.0016,
      "step": 61300
    },
    {
      "epoch": 0.393223295802561,
      "grad_norm": 0.1674564778804779,
      "learning_rate": 9.71836993546982e-06,
      "loss": 0.005,
      "step": 61310
    },
    {
      "epoch": 0.3932874326963471,
      "grad_norm": 0.1391586810350418,
      "learning_rate": 9.718184713089268e-06,
      "loss": 0.0037,
      "step": 61320
    },
    {
      "epoch": 0.3933515695901332,
      "grad_norm": 0.14953207969665527,
      "learning_rate": 9.71799943158649e-06,
      "loss": 0.003,
      "step": 61330
    },
    {
      "epoch": 0.39341570648391927,
      "grad_norm": 0.32289430499076843,
      "learning_rate": 9.717814090963807e-06,
      "loss": 0.0074,
      "step": 61340
    },
    {
      "epoch": 0.39347984337770536,
      "grad_norm": 0.14276909828186035,
      "learning_rate": 9.717628691223542e-06,
      "loss": 0.0036,
      "step": 61350
    },
    {
      "epoch": 0.39354398027149146,
      "grad_norm": 0.11924324929714203,
      "learning_rate": 9.717443232368018e-06,
      "loss": 0.0052,
      "step": 61360
    },
    {
      "epoch": 0.39360811716527755,
      "grad_norm": 0.20032940804958344,
      "learning_rate": 9.71725771439956e-06,
      "loss": 0.0036,
      "step": 61370
    },
    {
      "epoch": 0.39367225405906364,
      "grad_norm": 0.16514188051223755,
      "learning_rate": 9.717072137320489e-06,
      "loss": 0.0031,
      "step": 61380
    },
    {
      "epoch": 0.39373639095284974,
      "grad_norm": 0.2525750398635864,
      "learning_rate": 9.716886501133135e-06,
      "loss": 0.0044,
      "step": 61390
    },
    {
      "epoch": 0.3938005278466359,
      "grad_norm": 0.08110995590686798,
      "learning_rate": 9.71670080583982e-06,
      "loss": 0.0036,
      "step": 61400
    },
    {
      "epoch": 0.393864664740422,
      "grad_norm": 0.13291466236114502,
      "learning_rate": 9.716515051442876e-06,
      "loss": 0.0039,
      "step": 61410
    },
    {
      "epoch": 0.39392880163420807,
      "grad_norm": 0.4126153588294983,
      "learning_rate": 9.716329237944624e-06,
      "loss": 0.0044,
      "step": 61420
    },
    {
      "epoch": 0.39399293852799416,
      "grad_norm": 0.19799847900867462,
      "learning_rate": 9.716143365347398e-06,
      "loss": 0.0023,
      "step": 61430
    },
    {
      "epoch": 0.39405707542178026,
      "grad_norm": 0.20207390189170837,
      "learning_rate": 9.715957433653524e-06,
      "loss": 0.0049,
      "step": 61440
    },
    {
      "epoch": 0.39412121231556635,
      "grad_norm": 0.10306062549352646,
      "learning_rate": 9.715771442865334e-06,
      "loss": 0.0051,
      "step": 61450
    },
    {
      "epoch": 0.39418534920935244,
      "grad_norm": 0.47320789098739624,
      "learning_rate": 9.715585392985157e-06,
      "loss": 0.0054,
      "step": 61460
    },
    {
      "epoch": 0.39424948610313854,
      "grad_norm": 0.2486879825592041,
      "learning_rate": 9.715399284015325e-06,
      "loss": 0.0047,
      "step": 61470
    },
    {
      "epoch": 0.39431362299692463,
      "grad_norm": 0.17979192733764648,
      "learning_rate": 9.71521311595817e-06,
      "loss": 0.0063,
      "step": 61480
    },
    {
      "epoch": 0.3943777598907107,
      "grad_norm": 0.23293660581111908,
      "learning_rate": 9.715026888816024e-06,
      "loss": 0.0079,
      "step": 61490
    },
    {
      "epoch": 0.3944418967844968,
      "grad_norm": 0.35949429869651794,
      "learning_rate": 9.714840602591222e-06,
      "loss": 0.0052,
      "step": 61500
    },
    {
      "epoch": 0.3945060336782829,
      "grad_norm": 0.26240459084510803,
      "learning_rate": 9.714654257286098e-06,
      "loss": 0.0029,
      "step": 61510
    },
    {
      "epoch": 0.394570170572069,
      "grad_norm": 0.0906815305352211,
      "learning_rate": 9.714467852902987e-06,
      "loss": 0.006,
      "step": 61520
    },
    {
      "epoch": 0.3946343074658551,
      "grad_norm": 0.1907978355884552,
      "learning_rate": 9.714281389444222e-06,
      "loss": 0.0055,
      "step": 61530
    },
    {
      "epoch": 0.39469844435964124,
      "grad_norm": 0.2204655110836029,
      "learning_rate": 9.714094866912144e-06,
      "loss": 0.0036,
      "step": 61540
    },
    {
      "epoch": 0.39476258125342734,
      "grad_norm": 0.05758964642882347,
      "learning_rate": 9.713908285309088e-06,
      "loss": 0.0033,
      "step": 61550
    },
    {
      "epoch": 0.39482671814721343,
      "grad_norm": 0.1178441122174263,
      "learning_rate": 9.713721644637391e-06,
      "loss": 0.0053,
      "step": 61560
    },
    {
      "epoch": 0.3948908550409995,
      "grad_norm": 0.1403207629919052,
      "learning_rate": 9.713534944899394e-06,
      "loss": 0.0039,
      "step": 61570
    },
    {
      "epoch": 0.3949549919347856,
      "grad_norm": 0.15063434839248657,
      "learning_rate": 9.713348186097435e-06,
      "loss": 0.0051,
      "step": 61580
    },
    {
      "epoch": 0.3950191288285717,
      "grad_norm": 0.04773133993148804,
      "learning_rate": 9.713161368233857e-06,
      "loss": 0.0055,
      "step": 61590
    },
    {
      "epoch": 0.3950832657223578,
      "grad_norm": 0.35290268063545227,
      "learning_rate": 9.712974491310996e-06,
      "loss": 0.0042,
      "step": 61600
    },
    {
      "epoch": 0.3951474026161439,
      "grad_norm": 0.15310856699943542,
      "learning_rate": 9.712787555331196e-06,
      "loss": 0.0032,
      "step": 61610
    },
    {
      "epoch": 0.39521153950993,
      "grad_norm": 0.09890574961900711,
      "learning_rate": 9.712600560296802e-06,
      "loss": 0.0054,
      "step": 61620
    },
    {
      "epoch": 0.3952756764037161,
      "grad_norm": 0.17375418543815613,
      "learning_rate": 9.712413506210151e-06,
      "loss": 0.0029,
      "step": 61630
    },
    {
      "epoch": 0.3953398132975022,
      "grad_norm": 0.19882069528102875,
      "learning_rate": 9.712226393073596e-06,
      "loss": 0.0044,
      "step": 61640
    },
    {
      "epoch": 0.39540395019128827,
      "grad_norm": 0.07727350294589996,
      "learning_rate": 9.712039220889473e-06,
      "loss": 0.0047,
      "step": 61650
    },
    {
      "epoch": 0.39546808708507436,
      "grad_norm": 0.336527556180954,
      "learning_rate": 9.711851989660131e-06,
      "loss": 0.0039,
      "step": 61660
    },
    {
      "epoch": 0.39553222397886045,
      "grad_norm": 0.4891013205051422,
      "learning_rate": 9.711664699387917e-06,
      "loss": 0.0072,
      "step": 61670
    },
    {
      "epoch": 0.3955963608726466,
      "grad_norm": 0.16443385183811188,
      "learning_rate": 9.711477350075179e-06,
      "loss": 0.0041,
      "step": 61680
    },
    {
      "epoch": 0.3956604977664327,
      "grad_norm": 0.1870088130235672,
      "learning_rate": 9.71128994172426e-06,
      "loss": 0.0033,
      "step": 61690
    },
    {
      "epoch": 0.3957246346602188,
      "grad_norm": 0.4158104360103607,
      "learning_rate": 9.71110247433751e-06,
      "loss": 0.0037,
      "step": 61700
    },
    {
      "epoch": 0.3957887715540049,
      "grad_norm": 0.30746588110923767,
      "learning_rate": 9.71091494791728e-06,
      "loss": 0.0048,
      "step": 61710
    },
    {
      "epoch": 0.395852908447791,
      "grad_norm": 0.22402305901050568,
      "learning_rate": 9.710727362465919e-06,
      "loss": 0.0045,
      "step": 61720
    },
    {
      "epoch": 0.39591704534157707,
      "grad_norm": 0.26094120740890503,
      "learning_rate": 9.710539717985778e-06,
      "loss": 0.004,
      "step": 61730
    },
    {
      "epoch": 0.39598118223536316,
      "grad_norm": 0.0821923241019249,
      "learning_rate": 9.710352014479208e-06,
      "loss": 0.0051,
      "step": 61740
    },
    {
      "epoch": 0.39604531912914925,
      "grad_norm": 0.08290160447359085,
      "learning_rate": 9.71016425194856e-06,
      "loss": 0.0057,
      "step": 61750
    },
    {
      "epoch": 0.39610945602293535,
      "grad_norm": 0.19509842991828918,
      "learning_rate": 9.709976430396188e-06,
      "loss": 0.0047,
      "step": 61760
    },
    {
      "epoch": 0.39617359291672144,
      "grad_norm": 0.13025161623954773,
      "learning_rate": 9.709788549824445e-06,
      "loss": 0.003,
      "step": 61770
    },
    {
      "epoch": 0.39623772981050753,
      "grad_norm": 0.36803415417671204,
      "learning_rate": 9.709600610235686e-06,
      "loss": 0.0034,
      "step": 61780
    },
    {
      "epoch": 0.3963018667042936,
      "grad_norm": 0.11032091081142426,
      "learning_rate": 9.709412611632264e-06,
      "loss": 0.005,
      "step": 61790
    },
    {
      "epoch": 0.3963660035980797,
      "grad_norm": 0.10881927609443665,
      "learning_rate": 9.709224554016537e-06,
      "loss": 0.0038,
      "step": 61800
    },
    {
      "epoch": 0.3964301404918658,
      "grad_norm": 0.05322081968188286,
      "learning_rate": 9.709036437390861e-06,
      "loss": 0.005,
      "step": 61810
    },
    {
      "epoch": 0.39649427738565196,
      "grad_norm": 0.06323673576116562,
      "learning_rate": 9.708848261757593e-06,
      "loss": 0.0019,
      "step": 61820
    },
    {
      "epoch": 0.39655841427943805,
      "grad_norm": 0.47415101528167725,
      "learning_rate": 9.70866002711909e-06,
      "loss": 0.0079,
      "step": 61830
    },
    {
      "epoch": 0.39662255117322415,
      "grad_norm": 0.23023667931556702,
      "learning_rate": 9.708471733477714e-06,
      "loss": 0.0048,
      "step": 61840
    },
    {
      "epoch": 0.39668668806701024,
      "grad_norm": 0.21648605167865753,
      "learning_rate": 9.708283380835818e-06,
      "loss": 0.0086,
      "step": 61850
    },
    {
      "epoch": 0.39675082496079633,
      "grad_norm": 0.2983563244342804,
      "learning_rate": 9.708094969195769e-06,
      "loss": 0.0038,
      "step": 61860
    },
    {
      "epoch": 0.3968149618545824,
      "grad_norm": 0.15146301686763763,
      "learning_rate": 9.707906498559926e-06,
      "loss": 0.0043,
      "step": 61870
    },
    {
      "epoch": 0.3968790987483685,
      "grad_norm": 0.153154194355011,
      "learning_rate": 9.707717968930648e-06,
      "loss": 0.0026,
      "step": 61880
    },
    {
      "epoch": 0.3969432356421546,
      "grad_norm": 0.13013337552547455,
      "learning_rate": 9.707529380310301e-06,
      "loss": 0.006,
      "step": 61890
    },
    {
      "epoch": 0.3970073725359407,
      "grad_norm": 0.17496158182621002,
      "learning_rate": 9.707340732701245e-06,
      "loss": 0.0045,
      "step": 61900
    },
    {
      "epoch": 0.3970715094297268,
      "grad_norm": 0.4708886742591858,
      "learning_rate": 9.707152026105846e-06,
      "loss": 0.0065,
      "step": 61910
    },
    {
      "epoch": 0.3971356463235129,
      "grad_norm": 0.11030770093202591,
      "learning_rate": 9.706963260526468e-06,
      "loss": 0.0053,
      "step": 61920
    },
    {
      "epoch": 0.397199783217299,
      "grad_norm": 0.16902785003185272,
      "learning_rate": 9.706774435965477e-06,
      "loss": 0.0046,
      "step": 61930
    },
    {
      "epoch": 0.3972639201110851,
      "grad_norm": 0.2294456958770752,
      "learning_rate": 9.706585552425236e-06,
      "loss": 0.0056,
      "step": 61940
    },
    {
      "epoch": 0.39732805700487117,
      "grad_norm": 0.2020728588104248,
      "learning_rate": 9.706396609908115e-06,
      "loss": 0.0045,
      "step": 61950
    },
    {
      "epoch": 0.3973921938986573,
      "grad_norm": 0.061888452619314194,
      "learning_rate": 9.706207608416483e-06,
      "loss": 0.0049,
      "step": 61960
    },
    {
      "epoch": 0.3974563307924434,
      "grad_norm": 0.08395054936408997,
      "learning_rate": 9.706018547952704e-06,
      "loss": 0.0028,
      "step": 61970
    },
    {
      "epoch": 0.3975204676862295,
      "grad_norm": 0.1081097349524498,
      "learning_rate": 9.70582942851915e-06,
      "loss": 0.0042,
      "step": 61980
    },
    {
      "epoch": 0.3975846045800156,
      "grad_norm": 0.16980428993701935,
      "learning_rate": 9.70564025011819e-06,
      "loss": 0.004,
      "step": 61990
    },
    {
      "epoch": 0.3976487414738017,
      "grad_norm": 0.22608888149261475,
      "learning_rate": 9.705451012752196e-06,
      "loss": 0.0042,
      "step": 62000
    },
    {
      "epoch": 0.3977128783675878,
      "grad_norm": 0.2874959409236908,
      "learning_rate": 9.705261716423536e-06,
      "loss": 0.0037,
      "step": 62010
    },
    {
      "epoch": 0.3977770152613739,
      "grad_norm": 0.1783289909362793,
      "learning_rate": 9.705072361134584e-06,
      "loss": 0.0041,
      "step": 62020
    },
    {
      "epoch": 0.39784115215515997,
      "grad_norm": 0.11932146549224854,
      "learning_rate": 9.704882946887715e-06,
      "loss": 0.0174,
      "step": 62030
    },
    {
      "epoch": 0.39790528904894606,
      "grad_norm": 0.22151726484298706,
      "learning_rate": 9.704693473685298e-06,
      "loss": 0.0027,
      "step": 62040
    },
    {
      "epoch": 0.39796942594273216,
      "grad_norm": 0.17277520895004272,
      "learning_rate": 9.70450394152971e-06,
      "loss": 0.003,
      "step": 62050
    },
    {
      "epoch": 0.39803356283651825,
      "grad_norm": 0.2572478950023651,
      "learning_rate": 9.704314350423325e-06,
      "loss": 0.0051,
      "step": 62060
    },
    {
      "epoch": 0.39809769973030434,
      "grad_norm": 0.21472808718681335,
      "learning_rate": 9.704124700368518e-06,
      "loss": 0.0041,
      "step": 62070
    },
    {
      "epoch": 0.39816183662409044,
      "grad_norm": 0.053281769156455994,
      "learning_rate": 9.703934991367668e-06,
      "loss": 0.0063,
      "step": 62080
    },
    {
      "epoch": 0.39822597351787653,
      "grad_norm": 0.09664373099803925,
      "learning_rate": 9.70374522342315e-06,
      "loss": 0.0029,
      "step": 62090
    },
    {
      "epoch": 0.3982901104116627,
      "grad_norm": 0.10475917160511017,
      "learning_rate": 9.703555396537343e-06,
      "loss": 0.0042,
      "step": 62100
    },
    {
      "epoch": 0.39835424730544877,
      "grad_norm": 0.16188351809978485,
      "learning_rate": 9.703365510712626e-06,
      "loss": 0.0038,
      "step": 62110
    },
    {
      "epoch": 0.39841838419923487,
      "grad_norm": 0.2163073718547821,
      "learning_rate": 9.703175565951376e-06,
      "loss": 0.0023,
      "step": 62120
    },
    {
      "epoch": 0.39848252109302096,
      "grad_norm": 0.1845230907201767,
      "learning_rate": 9.702985562255978e-06,
      "loss": 0.0045,
      "step": 62130
    },
    {
      "epoch": 0.39854665798680705,
      "grad_norm": 0.22202342748641968,
      "learning_rate": 9.702795499628807e-06,
      "loss": 0.0037,
      "step": 62140
    },
    {
      "epoch": 0.39861079488059314,
      "grad_norm": 0.21595120429992676,
      "learning_rate": 9.702605378072248e-06,
      "loss": 0.0053,
      "step": 62150
    },
    {
      "epoch": 0.39867493177437924,
      "grad_norm": 0.45596906542778015,
      "learning_rate": 9.702415197588682e-06,
      "loss": 0.0072,
      "step": 62160
    },
    {
      "epoch": 0.39873906866816533,
      "grad_norm": 0.5580806732177734,
      "learning_rate": 9.702224958180494e-06,
      "loss": 0.0051,
      "step": 62170
    },
    {
      "epoch": 0.3988032055619514,
      "grad_norm": 0.2247030884027481,
      "learning_rate": 9.702034659850066e-06,
      "loss": 0.004,
      "step": 62180
    },
    {
      "epoch": 0.3988673424557375,
      "grad_norm": 0.18748068809509277,
      "learning_rate": 9.701844302599784e-06,
      "loss": 0.0042,
      "step": 62190
    },
    {
      "epoch": 0.3989314793495236,
      "grad_norm": 0.42575669288635254,
      "learning_rate": 9.701653886432031e-06,
      "loss": 0.0048,
      "step": 62200
    },
    {
      "epoch": 0.3989956162433097,
      "grad_norm": 0.1619352400302887,
      "learning_rate": 9.701463411349195e-06,
      "loss": 0.0036,
      "step": 62210
    },
    {
      "epoch": 0.3990597531370958,
      "grad_norm": 0.12561991810798645,
      "learning_rate": 9.701272877353664e-06,
      "loss": 0.0049,
      "step": 62220
    },
    {
      "epoch": 0.3991238900308819,
      "grad_norm": 0.23943594098091125,
      "learning_rate": 9.701082284447823e-06,
      "loss": 0.0042,
      "step": 62230
    },
    {
      "epoch": 0.39918802692466804,
      "grad_norm": 0.1591866910457611,
      "learning_rate": 9.700891632634059e-06,
      "loss": 0.0097,
      "step": 62240
    },
    {
      "epoch": 0.39925216381845413,
      "grad_norm": 0.09385634958744049,
      "learning_rate": 9.700700921914765e-06,
      "loss": 0.0063,
      "step": 62250
    },
    {
      "epoch": 0.3993163007122402,
      "grad_norm": 0.12799791991710663,
      "learning_rate": 9.70051015229233e-06,
      "loss": 0.0026,
      "step": 62260
    },
    {
      "epoch": 0.3993804376060263,
      "grad_norm": 0.06453348696231842,
      "learning_rate": 9.700319323769143e-06,
      "loss": 0.0032,
      "step": 62270
    },
    {
      "epoch": 0.3994445744998124,
      "grad_norm": 0.2493322491645813,
      "learning_rate": 9.700128436347595e-06,
      "loss": 0.0041,
      "step": 62280
    },
    {
      "epoch": 0.3995087113935985,
      "grad_norm": 0.13772499561309814,
      "learning_rate": 9.699937490030078e-06,
      "loss": 0.003,
      "step": 62290
    },
    {
      "epoch": 0.3995728482873846,
      "grad_norm": 0.3720828592777252,
      "learning_rate": 9.699746484818986e-06,
      "loss": 0.0028,
      "step": 62300
    },
    {
      "epoch": 0.3996369851811707,
      "grad_norm": 0.19337330758571625,
      "learning_rate": 9.699555420716711e-06,
      "loss": 0.0032,
      "step": 62310
    },
    {
      "epoch": 0.3997011220749568,
      "grad_norm": 0.1157507449388504,
      "learning_rate": 9.699364297725649e-06,
      "loss": 0.0043,
      "step": 62320
    },
    {
      "epoch": 0.3997652589687429,
      "grad_norm": 0.14450615644454956,
      "learning_rate": 9.699173115848193e-06,
      "loss": 0.0051,
      "step": 62330
    },
    {
      "epoch": 0.39982939586252897,
      "grad_norm": 0.24375490844249725,
      "learning_rate": 9.698981875086739e-06,
      "loss": 0.0034,
      "step": 62340
    },
    {
      "epoch": 0.39989353275631506,
      "grad_norm": 0.3356948792934418,
      "learning_rate": 9.698790575443685e-06,
      "loss": 0.0043,
      "step": 62350
    },
    {
      "epoch": 0.39995766965010116,
      "grad_norm": 0.11101289093494415,
      "learning_rate": 9.698599216921426e-06,
      "loss": 0.0038,
      "step": 62360
    },
    {
      "epoch": 0.40002180654388725,
      "grad_norm": 0.1932825893163681,
      "learning_rate": 9.69840779952236e-06,
      "loss": 0.0053,
      "step": 62370
    },
    {
      "epoch": 0.4000859434376734,
      "grad_norm": 0.2052309662103653,
      "learning_rate": 9.698216323248888e-06,
      "loss": 0.0027,
      "step": 62380
    },
    {
      "epoch": 0.4001500803314595,
      "grad_norm": 0.16054415702819824,
      "learning_rate": 9.698024788103406e-06,
      "loss": 0.0054,
      "step": 62390
    },
    {
      "epoch": 0.4002142172252456,
      "grad_norm": 0.22121907770633698,
      "learning_rate": 9.697833194088317e-06,
      "loss": 0.004,
      "step": 62400
    },
    {
      "epoch": 0.4002783541190317,
      "grad_norm": 0.35723811388015747,
      "learning_rate": 9.69764154120602e-06,
      "loss": 0.0039,
      "step": 62410
    },
    {
      "epoch": 0.40034249101281777,
      "grad_norm": 0.17922592163085938,
      "learning_rate": 9.697449829458918e-06,
      "loss": 0.0048,
      "step": 62420
    },
    {
      "epoch": 0.40040662790660386,
      "grad_norm": 0.3429785966873169,
      "learning_rate": 9.697258058849413e-06,
      "loss": 0.0036,
      "step": 62430
    },
    {
      "epoch": 0.40047076480038996,
      "grad_norm": 0.23051443696022034,
      "learning_rate": 9.697066229379906e-06,
      "loss": 0.0068,
      "step": 62440
    },
    {
      "epoch": 0.40053490169417605,
      "grad_norm": 0.12451818585395813,
      "learning_rate": 9.696874341052803e-06,
      "loss": 0.0049,
      "step": 62450
    },
    {
      "epoch": 0.40059903858796214,
      "grad_norm": 0.11693712323904037,
      "learning_rate": 9.696682393870507e-06,
      "loss": 0.004,
      "step": 62460
    },
    {
      "epoch": 0.40066317548174823,
      "grad_norm": 0.4547955095767975,
      "learning_rate": 9.696490387835425e-06,
      "loss": 0.0033,
      "step": 62470
    },
    {
      "epoch": 0.40072731237553433,
      "grad_norm": 0.0886857807636261,
      "learning_rate": 9.69629832294996e-06,
      "loss": 0.0024,
      "step": 62480
    },
    {
      "epoch": 0.4007914492693204,
      "grad_norm": 0.10484019666910172,
      "learning_rate": 9.696106199216521e-06,
      "loss": 0.0036,
      "step": 62490
    },
    {
      "epoch": 0.4008555861631065,
      "grad_norm": 0.3736126720905304,
      "learning_rate": 9.695914016637517e-06,
      "loss": 0.0078,
      "step": 62500
    },
    {
      "epoch": 0.4009197230568926,
      "grad_norm": 0.21714961528778076,
      "learning_rate": 9.695721775215353e-06,
      "loss": 0.0029,
      "step": 62510
    },
    {
      "epoch": 0.40098385995067876,
      "grad_norm": 0.1946527063846588,
      "learning_rate": 9.695529474952439e-06,
      "loss": 0.0044,
      "step": 62520
    },
    {
      "epoch": 0.40104799684446485,
      "grad_norm": 0.14760996401309967,
      "learning_rate": 9.695337115851186e-06,
      "loss": 0.0032,
      "step": 62530
    },
    {
      "epoch": 0.40111213373825094,
      "grad_norm": 0.5645552277565002,
      "learning_rate": 9.695144697914001e-06,
      "loss": 0.0111,
      "step": 62540
    },
    {
      "epoch": 0.40117627063203704,
      "grad_norm": 0.32809337973594666,
      "learning_rate": 9.694952221143299e-06,
      "loss": 0.0035,
      "step": 62550
    },
    {
      "epoch": 0.40124040752582313,
      "grad_norm": 0.1572769433259964,
      "learning_rate": 9.694759685541491e-06,
      "loss": 0.0061,
      "step": 62560
    },
    {
      "epoch": 0.4013045444196092,
      "grad_norm": 0.07307353615760803,
      "learning_rate": 9.694567091110987e-06,
      "loss": 0.0067,
      "step": 62570
    },
    {
      "epoch": 0.4013686813133953,
      "grad_norm": 0.2429163157939911,
      "learning_rate": 9.694374437854204e-06,
      "loss": 0.0039,
      "step": 62580
    },
    {
      "epoch": 0.4014328182071814,
      "grad_norm": 0.17218413949012756,
      "learning_rate": 9.694181725773553e-06,
      "loss": 0.0059,
      "step": 62590
    },
    {
      "epoch": 0.4014969551009675,
      "grad_norm": 0.11023490130901337,
      "learning_rate": 9.693988954871449e-06,
      "loss": 0.0057,
      "step": 62600
    },
    {
      "epoch": 0.4015610919947536,
      "grad_norm": 0.008556004613637924,
      "learning_rate": 9.693796125150312e-06,
      "loss": 0.0044,
      "step": 62610
    },
    {
      "epoch": 0.4016252288885397,
      "grad_norm": 0.35116973519325256,
      "learning_rate": 9.693603236612551e-06,
      "loss": 0.0041,
      "step": 62620
    },
    {
      "epoch": 0.4016893657823258,
      "grad_norm": 0.2881322503089905,
      "learning_rate": 9.69341028926059e-06,
      "loss": 0.0047,
      "step": 62630
    },
    {
      "epoch": 0.4017535026761119,
      "grad_norm": 0.1554815024137497,
      "learning_rate": 9.693217283096843e-06,
      "loss": 0.0055,
      "step": 62640
    },
    {
      "epoch": 0.40181763956989797,
      "grad_norm": 0.2013331651687622,
      "learning_rate": 9.693024218123728e-06,
      "loss": 0.0047,
      "step": 62650
    },
    {
      "epoch": 0.4018817764636841,
      "grad_norm": 0.19363749027252197,
      "learning_rate": 9.692831094343666e-06,
      "loss": 0.0052,
      "step": 62660
    },
    {
      "epoch": 0.4019459133574702,
      "grad_norm": 0.10971488803625107,
      "learning_rate": 9.692637911759077e-06,
      "loss": 0.0037,
      "step": 62670
    },
    {
      "epoch": 0.4020100502512563,
      "grad_norm": 0.24738819897174835,
      "learning_rate": 9.692444670372382e-06,
      "loss": 0.0044,
      "step": 62680
    },
    {
      "epoch": 0.4020741871450424,
      "grad_norm": 0.05652572959661484,
      "learning_rate": 9.692251370186002e-06,
      "loss": 0.0029,
      "step": 62690
    },
    {
      "epoch": 0.4021383240388285,
      "grad_norm": 0.2002425640821457,
      "learning_rate": 9.692058011202356e-06,
      "loss": 0.004,
      "step": 62700
    },
    {
      "epoch": 0.4022024609326146,
      "grad_norm": 0.10862979292869568,
      "learning_rate": 9.691864593423872e-06,
      "loss": 0.0037,
      "step": 62710
    },
    {
      "epoch": 0.4022665978264007,
      "grad_norm": 0.15046946704387665,
      "learning_rate": 9.69167111685297e-06,
      "loss": 0.0024,
      "step": 62720
    },
    {
      "epoch": 0.40233073472018677,
      "grad_norm": 0.046050697565078735,
      "learning_rate": 9.691477581492076e-06,
      "loss": 0.005,
      "step": 62730
    },
    {
      "epoch": 0.40239487161397286,
      "grad_norm": 0.20435963571071625,
      "learning_rate": 9.691283987343616e-06,
      "loss": 0.0045,
      "step": 62740
    },
    {
      "epoch": 0.40245900850775895,
      "grad_norm": 0.32580456137657166,
      "learning_rate": 9.691090334410014e-06,
      "loss": 0.0031,
      "step": 62750
    },
    {
      "epoch": 0.40252314540154505,
      "grad_norm": 0.1473287045955658,
      "learning_rate": 9.6908966226937e-06,
      "loss": 0.004,
      "step": 62760
    },
    {
      "epoch": 0.40258728229533114,
      "grad_norm": 0.1179991215467453,
      "learning_rate": 9.690702852197094e-06,
      "loss": 0.0044,
      "step": 62770
    },
    {
      "epoch": 0.40265141918911723,
      "grad_norm": 0.33118948340415955,
      "learning_rate": 9.690509022922632e-06,
      "loss": 0.0088,
      "step": 62780
    },
    {
      "epoch": 0.4027155560829033,
      "grad_norm": 0.36124172806739807,
      "learning_rate": 9.690315134872738e-06,
      "loss": 0.0043,
      "step": 62790
    },
    {
      "epoch": 0.4027796929766894,
      "grad_norm": 0.2574802339076996,
      "learning_rate": 9.690121188049845e-06,
      "loss": 0.0036,
      "step": 62800
    },
    {
      "epoch": 0.40284382987047557,
      "grad_norm": 0.2140289694070816,
      "learning_rate": 9.689927182456383e-06,
      "loss": 0.0038,
      "step": 62810
    },
    {
      "epoch": 0.40290796676426166,
      "grad_norm": 0.041685204952955246,
      "learning_rate": 9.689733118094777e-06,
      "loss": 0.0042,
      "step": 62820
    },
    {
      "epoch": 0.40297210365804775,
      "grad_norm": 0.2141823172569275,
      "learning_rate": 9.689538994967467e-06,
      "loss": 0.0032,
      "step": 62830
    },
    {
      "epoch": 0.40303624055183385,
      "grad_norm": 0.17390379309654236,
      "learning_rate": 9.68934481307688e-06,
      "loss": 0.008,
      "step": 62840
    },
    {
      "epoch": 0.40310037744561994,
      "grad_norm": 0.23098264634609222,
      "learning_rate": 9.689150572425452e-06,
      "loss": 0.0052,
      "step": 62850
    },
    {
      "epoch": 0.40316451433940603,
      "grad_norm": 0.18938297033309937,
      "learning_rate": 9.688956273015617e-06,
      "loss": 0.005,
      "step": 62860
    },
    {
      "epoch": 0.4032286512331921,
      "grad_norm": 0.14759260416030884,
      "learning_rate": 9.688761914849806e-06,
      "loss": 0.0046,
      "step": 62870
    },
    {
      "epoch": 0.4032927881269782,
      "grad_norm": 0.11562065780162811,
      "learning_rate": 9.688567497930458e-06,
      "loss": 0.0041,
      "step": 62880
    },
    {
      "epoch": 0.4033569250207643,
      "grad_norm": 0.20137348771095276,
      "learning_rate": 9.68837302226001e-06,
      "loss": 0.0053,
      "step": 62890
    },
    {
      "epoch": 0.4034210619145504,
      "grad_norm": 0.08316779136657715,
      "learning_rate": 9.688178487840896e-06,
      "loss": 0.0032,
      "step": 62900
    },
    {
      "epoch": 0.4034851988083365,
      "grad_norm": 0.1608303189277649,
      "learning_rate": 9.687983894675555e-06,
      "loss": 0.0044,
      "step": 62910
    },
    {
      "epoch": 0.4035493357021226,
      "grad_norm": 0.19574500620365143,
      "learning_rate": 9.687789242766426e-06,
      "loss": 0.0026,
      "step": 62920
    },
    {
      "epoch": 0.4036134725959087,
      "grad_norm": 0.4132160246372223,
      "learning_rate": 9.687594532115947e-06,
      "loss": 0.0044,
      "step": 62930
    },
    {
      "epoch": 0.4036776094896948,
      "grad_norm": 0.2931022346019745,
      "learning_rate": 9.687399762726559e-06,
      "loss": 0.0069,
      "step": 62940
    },
    {
      "epoch": 0.4037417463834809,
      "grad_norm": 0.15385998785495758,
      "learning_rate": 9.687204934600701e-06,
      "loss": 0.0033,
      "step": 62950
    },
    {
      "epoch": 0.403805883277267,
      "grad_norm": 0.1795760542154312,
      "learning_rate": 9.687010047740816e-06,
      "loss": 0.0043,
      "step": 62960
    },
    {
      "epoch": 0.4038700201710531,
      "grad_norm": 0.33173665404319763,
      "learning_rate": 9.686815102149346e-06,
      "loss": 0.0044,
      "step": 62970
    },
    {
      "epoch": 0.4039341570648392,
      "grad_norm": 0.3033214211463928,
      "learning_rate": 9.686620097828732e-06,
      "loss": 0.0029,
      "step": 62980
    },
    {
      "epoch": 0.4039982939586253,
      "grad_norm": 0.1678515523672104,
      "learning_rate": 9.686425034781418e-06,
      "loss": 0.0033,
      "step": 62990
    },
    {
      "epoch": 0.4040624308524114,
      "grad_norm": 0.07023715227842331,
      "learning_rate": 9.686229913009851e-06,
      "loss": 0.0058,
      "step": 63000
    },
    {
      "epoch": 0.4041265677461975,
      "grad_norm": 0.06690052151679993,
      "learning_rate": 9.686034732516471e-06,
      "loss": 0.0031,
      "step": 63010
    },
    {
      "epoch": 0.4041907046399836,
      "grad_norm": 0.1076996698975563,
      "learning_rate": 9.68583949330373e-06,
      "loss": 0.0052,
      "step": 63020
    },
    {
      "epoch": 0.40425484153376967,
      "grad_norm": 0.13543549180030823,
      "learning_rate": 9.685644195374068e-06,
      "loss": 0.0039,
      "step": 63030
    },
    {
      "epoch": 0.40431897842755576,
      "grad_norm": 0.07360082864761353,
      "learning_rate": 9.685448838729939e-06,
      "loss": 0.0038,
      "step": 63040
    },
    {
      "epoch": 0.40438311532134186,
      "grad_norm": 0.19905249774456024,
      "learning_rate": 9.685253423373786e-06,
      "loss": 0.0034,
      "step": 63050
    },
    {
      "epoch": 0.40444725221512795,
      "grad_norm": 0.08411014080047607,
      "learning_rate": 9.68505794930806e-06,
      "loss": 0.0029,
      "step": 63060
    },
    {
      "epoch": 0.40451138910891404,
      "grad_norm": 0.103714220225811,
      "learning_rate": 9.684862416535207e-06,
      "loss": 0.0034,
      "step": 63070
    },
    {
      "epoch": 0.40457552600270014,
      "grad_norm": 0.2893984317779541,
      "learning_rate": 9.68466682505768e-06,
      "loss": 0.0043,
      "step": 63080
    },
    {
      "epoch": 0.4046396628964863,
      "grad_norm": 0.11062141507863998,
      "learning_rate": 9.684471174877932e-06,
      "loss": 0.004,
      "step": 63090
    },
    {
      "epoch": 0.4047037997902724,
      "grad_norm": 0.20066706836223602,
      "learning_rate": 9.68427546599841e-06,
      "loss": 0.003,
      "step": 63100
    },
    {
      "epoch": 0.40476793668405847,
      "grad_norm": 0.29416951537132263,
      "learning_rate": 9.684079698421569e-06,
      "loss": 0.0063,
      "step": 63110
    },
    {
      "epoch": 0.40483207357784456,
      "grad_norm": 0.06612326204776764,
      "learning_rate": 9.683883872149864e-06,
      "loss": 0.0031,
      "step": 63120
    },
    {
      "epoch": 0.40489621047163066,
      "grad_norm": 0.03334604203701019,
      "learning_rate": 9.683687987185744e-06,
      "loss": 0.0036,
      "step": 63130
    },
    {
      "epoch": 0.40496034736541675,
      "grad_norm": 0.6492279767990112,
      "learning_rate": 9.683492043531665e-06,
      "loss": 0.0058,
      "step": 63140
    },
    {
      "epoch": 0.40502448425920284,
      "grad_norm": 0.2623600661754608,
      "learning_rate": 9.683296041190087e-06,
      "loss": 0.0039,
      "step": 63150
    },
    {
      "epoch": 0.40508862115298894,
      "grad_norm": 0.13972336053848267,
      "learning_rate": 9.683099980163461e-06,
      "loss": 0.0037,
      "step": 63160
    },
    {
      "epoch": 0.40515275804677503,
      "grad_norm": 0.35072267055511475,
      "learning_rate": 9.682903860454245e-06,
      "loss": 0.0031,
      "step": 63170
    },
    {
      "epoch": 0.4052168949405611,
      "grad_norm": 0.1438094824552536,
      "learning_rate": 9.682707682064897e-06,
      "loss": 0.0034,
      "step": 63180
    },
    {
      "epoch": 0.4052810318343472,
      "grad_norm": 0.4629147946834564,
      "learning_rate": 9.682511444997876e-06,
      "loss": 0.0057,
      "step": 63190
    },
    {
      "epoch": 0.4053451687281333,
      "grad_norm": 0.135417640209198,
      "learning_rate": 9.682315149255638e-06,
      "loss": 0.0035,
      "step": 63200
    },
    {
      "epoch": 0.4054093056219194,
      "grad_norm": 0.30805763602256775,
      "learning_rate": 9.682118794840647e-06,
      "loss": 0.0044,
      "step": 63210
    },
    {
      "epoch": 0.4054734425157055,
      "grad_norm": 0.12094740569591522,
      "learning_rate": 9.681922381755359e-06,
      "loss": 0.0032,
      "step": 63220
    },
    {
      "epoch": 0.40553757940949164,
      "grad_norm": 0.1189800575375557,
      "learning_rate": 9.681725910002238e-06,
      "loss": 0.0054,
      "step": 63230
    },
    {
      "epoch": 0.40560171630327774,
      "grad_norm": 0.2504672706127167,
      "learning_rate": 9.681529379583746e-06,
      "loss": 0.0031,
      "step": 63240
    },
    {
      "epoch": 0.40566585319706383,
      "grad_norm": 0.12307994812726974,
      "learning_rate": 9.681332790502345e-06,
      "loss": 0.0034,
      "step": 63250
    },
    {
      "epoch": 0.4057299900908499,
      "grad_norm": 0.26728805899620056,
      "learning_rate": 9.6811361427605e-06,
      "loss": 0.0035,
      "step": 63260
    },
    {
      "epoch": 0.405794126984636,
      "grad_norm": 0.23914340138435364,
      "learning_rate": 9.68093943636067e-06,
      "loss": 0.0033,
      "step": 63270
    },
    {
      "epoch": 0.4058582638784221,
      "grad_norm": 0.058856040239334106,
      "learning_rate": 9.680742671305326e-06,
      "loss": 0.0041,
      "step": 63280
    },
    {
      "epoch": 0.4059224007722082,
      "grad_norm": 0.05040334537625313,
      "learning_rate": 9.680545847596929e-06,
      "loss": 0.0039,
      "step": 63290
    },
    {
      "epoch": 0.4059865376659943,
      "grad_norm": 0.11939733475446701,
      "learning_rate": 9.68034896523795e-06,
      "loss": 0.0039,
      "step": 63300
    },
    {
      "epoch": 0.4060506745597804,
      "grad_norm": 0.24056608974933624,
      "learning_rate": 9.680152024230853e-06,
      "loss": 0.0041,
      "step": 63310
    },
    {
      "epoch": 0.4061148114535665,
      "grad_norm": 0.06175704672932625,
      "learning_rate": 9.679955024578105e-06,
      "loss": 0.0022,
      "step": 63320
    },
    {
      "epoch": 0.4061789483473526,
      "grad_norm": 0.10102628171443939,
      "learning_rate": 9.679757966282177e-06,
      "loss": 0.0043,
      "step": 63330
    },
    {
      "epoch": 0.40624308524113867,
      "grad_norm": 0.050519414246082306,
      "learning_rate": 9.679560849345535e-06,
      "loss": 0.0035,
      "step": 63340
    },
    {
      "epoch": 0.40630722213492476,
      "grad_norm": 0.37368524074554443,
      "learning_rate": 9.679363673770655e-06,
      "loss": 0.0046,
      "step": 63350
    },
    {
      "epoch": 0.40637135902871085,
      "grad_norm": 0.10394777357578278,
      "learning_rate": 9.67916643956e-06,
      "loss": 0.0014,
      "step": 63360
    },
    {
      "epoch": 0.406435495922497,
      "grad_norm": 0.1660034954547882,
      "learning_rate": 9.678969146716046e-06,
      "loss": 0.0037,
      "step": 63370
    },
    {
      "epoch": 0.4064996328162831,
      "grad_norm": 0.22908726334571838,
      "learning_rate": 9.678771795241267e-06,
      "loss": 0.0044,
      "step": 63380
    },
    {
      "epoch": 0.4065637697100692,
      "grad_norm": 0.11898116022348404,
      "learning_rate": 9.67857438513813e-06,
      "loss": 0.0058,
      "step": 63390
    },
    {
      "epoch": 0.4066279066038553,
      "grad_norm": 0.267651230096817,
      "learning_rate": 9.678376916409116e-06,
      "loss": 0.0047,
      "step": 63400
    },
    {
      "epoch": 0.4066920434976414,
      "grad_norm": 0.3415893614292145,
      "learning_rate": 9.678179389056694e-06,
      "loss": 0.0033,
      "step": 63410
    },
    {
      "epoch": 0.40675618039142747,
      "grad_norm": 0.3092597723007202,
      "learning_rate": 9.677981803083341e-06,
      "loss": 0.0051,
      "step": 63420
    },
    {
      "epoch": 0.40682031728521356,
      "grad_norm": 0.2247481346130371,
      "learning_rate": 9.677784158491532e-06,
      "loss": 0.0143,
      "step": 63430
    },
    {
      "epoch": 0.40688445417899965,
      "grad_norm": 0.3895609378814697,
      "learning_rate": 9.677586455283745e-06,
      "loss": 0.0065,
      "step": 63440
    },
    {
      "epoch": 0.40694859107278575,
      "grad_norm": 0.22449402511119843,
      "learning_rate": 9.677388693462456e-06,
      "loss": 0.004,
      "step": 63450
    },
    {
      "epoch": 0.40701272796657184,
      "grad_norm": 0.2221873253583908,
      "learning_rate": 9.677190873030144e-06,
      "loss": 0.0045,
      "step": 63460
    },
    {
      "epoch": 0.40707686486035793,
      "grad_norm": 0.19022652506828308,
      "learning_rate": 9.676992993989286e-06,
      "loss": 0.0037,
      "step": 63470
    },
    {
      "epoch": 0.407141001754144,
      "grad_norm": 0.18011286854743958,
      "learning_rate": 9.676795056342367e-06,
      "loss": 0.003,
      "step": 63480
    },
    {
      "epoch": 0.4072051386479301,
      "grad_norm": 0.35013285279273987,
      "learning_rate": 9.676597060091861e-06,
      "loss": 0.0036,
      "step": 63490
    },
    {
      "epoch": 0.4072692755417162,
      "grad_norm": 0.1795942634344101,
      "learning_rate": 9.676399005240252e-06,
      "loss": 0.0031,
      "step": 63500
    },
    {
      "epoch": 0.40733341243550236,
      "grad_norm": 0.27482670545578003,
      "learning_rate": 9.676200891790021e-06,
      "loss": 0.004,
      "step": 63510
    },
    {
      "epoch": 0.40739754932928846,
      "grad_norm": 0.21835391223430634,
      "learning_rate": 9.67600271974365e-06,
      "loss": 0.0045,
      "step": 63520
    },
    {
      "epoch": 0.40746168622307455,
      "grad_norm": 0.20426027476787567,
      "learning_rate": 9.675804489103626e-06,
      "loss": 0.0048,
      "step": 63530
    },
    {
      "epoch": 0.40752582311686064,
      "grad_norm": 0.3146619200706482,
      "learning_rate": 9.675606199872429e-06,
      "loss": 0.0058,
      "step": 63540
    },
    {
      "epoch": 0.40758996001064673,
      "grad_norm": 0.18899372220039368,
      "learning_rate": 9.675407852052542e-06,
      "loss": 0.0051,
      "step": 63550
    },
    {
      "epoch": 0.4076540969044328,
      "grad_norm": 0.24110999703407288,
      "learning_rate": 9.675209445646456e-06,
      "loss": 0.0039,
      "step": 63560
    },
    {
      "epoch": 0.4077182337982189,
      "grad_norm": 0.16395263373851776,
      "learning_rate": 9.675010980656654e-06,
      "loss": 0.0035,
      "step": 63570
    },
    {
      "epoch": 0.407782370692005,
      "grad_norm": 0.10138542205095291,
      "learning_rate": 9.674812457085623e-06,
      "loss": 0.0018,
      "step": 63580
    },
    {
      "epoch": 0.4078465075857911,
      "grad_norm": 0.10391603410243988,
      "learning_rate": 9.67461387493585e-06,
      "loss": 0.0037,
      "step": 63590
    },
    {
      "epoch": 0.4079106444795772,
      "grad_norm": 0.10009389370679855,
      "learning_rate": 9.674415234209824e-06,
      "loss": 0.0057,
      "step": 63600
    },
    {
      "epoch": 0.4079747813733633,
      "grad_norm": 0.16913220286369324,
      "learning_rate": 9.674216534910034e-06,
      "loss": 0.0023,
      "step": 63610
    },
    {
      "epoch": 0.4080389182671494,
      "grad_norm": 0.22187593579292297,
      "learning_rate": 9.674017777038973e-06,
      "loss": 0.004,
      "step": 63620
    },
    {
      "epoch": 0.4081030551609355,
      "grad_norm": 0.21074946224689484,
      "learning_rate": 9.673818960599127e-06,
      "loss": 0.0023,
      "step": 63630
    },
    {
      "epoch": 0.4081671920547216,
      "grad_norm": 0.3142641484737396,
      "learning_rate": 9.673620085592989e-06,
      "loss": 0.003,
      "step": 63640
    },
    {
      "epoch": 0.4082313289485077,
      "grad_norm": 0.1447477489709854,
      "learning_rate": 9.67342115202305e-06,
      "loss": 0.0043,
      "step": 63650
    },
    {
      "epoch": 0.4082954658422938,
      "grad_norm": 0.24877741932868958,
      "learning_rate": 9.673222159891807e-06,
      "loss": 0.0038,
      "step": 63660
    },
    {
      "epoch": 0.4083596027360799,
      "grad_norm": 0.27425238490104675,
      "learning_rate": 9.673023109201747e-06,
      "loss": 0.0027,
      "step": 63670
    },
    {
      "epoch": 0.408423739629866,
      "grad_norm": 0.16497421264648438,
      "learning_rate": 9.67282399995537e-06,
      "loss": 0.0051,
      "step": 63680
    },
    {
      "epoch": 0.4084878765236521,
      "grad_norm": 0.33673834800720215,
      "learning_rate": 9.672624832155168e-06,
      "loss": 0.0038,
      "step": 63690
    },
    {
      "epoch": 0.4085520134174382,
      "grad_norm": 0.2985970377922058,
      "learning_rate": 9.672425605803636e-06,
      "loss": 0.0049,
      "step": 63700
    },
    {
      "epoch": 0.4086161503112243,
      "grad_norm": 0.15841737389564514,
      "learning_rate": 9.672226320903275e-06,
      "loss": 0.004,
      "step": 63710
    },
    {
      "epoch": 0.4086802872050104,
      "grad_norm": 0.18191616237163544,
      "learning_rate": 9.672026977456576e-06,
      "loss": 0.005,
      "step": 63720
    },
    {
      "epoch": 0.40874442409879647,
      "grad_norm": 0.5124845504760742,
      "learning_rate": 9.671827575466041e-06,
      "loss": 0.0037,
      "step": 63730
    },
    {
      "epoch": 0.40880856099258256,
      "grad_norm": 0.06640222668647766,
      "learning_rate": 9.671628114934169e-06,
      "loss": 0.0058,
      "step": 63740
    },
    {
      "epoch": 0.40887269788636865,
      "grad_norm": 0.44437193870544434,
      "learning_rate": 9.671428595863457e-06,
      "loss": 0.0053,
      "step": 63750
    },
    {
      "epoch": 0.40893683478015475,
      "grad_norm": 0.20514225959777832,
      "learning_rate": 9.671229018256405e-06,
      "loss": 0.005,
      "step": 63760
    },
    {
      "epoch": 0.40900097167394084,
      "grad_norm": 0.1101483628153801,
      "learning_rate": 9.671029382115516e-06,
      "loss": 0.0055,
      "step": 63770
    },
    {
      "epoch": 0.40906510856772693,
      "grad_norm": 0.30075669288635254,
      "learning_rate": 9.67082968744329e-06,
      "loss": 0.0041,
      "step": 63780
    },
    {
      "epoch": 0.4091292454615131,
      "grad_norm": 0.38170281052589417,
      "learning_rate": 9.67062993424223e-06,
      "loss": 0.0044,
      "step": 63790
    },
    {
      "epoch": 0.4091933823552992,
      "grad_norm": 0.13880175352096558,
      "learning_rate": 9.670430122514839e-06,
      "loss": 0.004,
      "step": 63800
    },
    {
      "epoch": 0.40925751924908527,
      "grad_norm": 0.3685900568962097,
      "learning_rate": 9.670230252263621e-06,
      "loss": 0.0075,
      "step": 63810
    },
    {
      "epoch": 0.40932165614287136,
      "grad_norm": 0.16778992116451263,
      "learning_rate": 9.670030323491079e-06,
      "loss": 0.004,
      "step": 63820
    },
    {
      "epoch": 0.40938579303665745,
      "grad_norm": 0.09747684001922607,
      "learning_rate": 9.66983033619972e-06,
      "loss": 0.0074,
      "step": 63830
    },
    {
      "epoch": 0.40944992993044355,
      "grad_norm": 0.1732402741909027,
      "learning_rate": 9.669630290392051e-06,
      "loss": 0.0055,
      "step": 63840
    },
    {
      "epoch": 0.40951406682422964,
      "grad_norm": 0.030933083966374397,
      "learning_rate": 9.669430186070575e-06,
      "loss": 0.0046,
      "step": 63850
    },
    {
      "epoch": 0.40957820371801573,
      "grad_norm": 0.010907845571637154,
      "learning_rate": 9.669230023237803e-06,
      "loss": 0.0027,
      "step": 63860
    },
    {
      "epoch": 0.4096423406118018,
      "grad_norm": 0.20749413967132568,
      "learning_rate": 9.66902980189624e-06,
      "loss": 0.0034,
      "step": 63870
    },
    {
      "epoch": 0.4097064775055879,
      "grad_norm": 0.36841881275177,
      "learning_rate": 9.668829522048397e-06,
      "loss": 0.0044,
      "step": 63880
    },
    {
      "epoch": 0.409770614399374,
      "grad_norm": 0.26955240964889526,
      "learning_rate": 9.668629183696784e-06,
      "loss": 0.005,
      "step": 63890
    },
    {
      "epoch": 0.4098347512931601,
      "grad_norm": 0.060388293117284775,
      "learning_rate": 9.668428786843911e-06,
      "loss": 0.0045,
      "step": 63900
    },
    {
      "epoch": 0.4098988881869462,
      "grad_norm": 0.23592258989810944,
      "learning_rate": 9.668228331492288e-06,
      "loss": 0.0025,
      "step": 63910
    },
    {
      "epoch": 0.4099630250807323,
      "grad_norm": 0.3113408386707306,
      "learning_rate": 9.66802781764443e-06,
      "loss": 0.0046,
      "step": 63920
    },
    {
      "epoch": 0.41002716197451844,
      "grad_norm": 0.04270043224096298,
      "learning_rate": 9.667827245302845e-06,
      "loss": 0.0027,
      "step": 63930
    },
    {
      "epoch": 0.41009129886830453,
      "grad_norm": 0.0638560876250267,
      "learning_rate": 9.66762661447005e-06,
      "loss": 0.0041,
      "step": 63940
    },
    {
      "epoch": 0.4101554357620906,
      "grad_norm": 0.11248234659433365,
      "learning_rate": 9.667425925148555e-06,
      "loss": 0.0033,
      "step": 63950
    },
    {
      "epoch": 0.4102195726558767,
      "grad_norm": 0.03995837643742561,
      "learning_rate": 9.66722517734088e-06,
      "loss": 0.004,
      "step": 63960
    },
    {
      "epoch": 0.4102837095496628,
      "grad_norm": 0.24541008472442627,
      "learning_rate": 9.667024371049537e-06,
      "loss": 0.005,
      "step": 63970
    },
    {
      "epoch": 0.4103478464434489,
      "grad_norm": 0.1571703851222992,
      "learning_rate": 9.666823506277044e-06,
      "loss": 0.0046,
      "step": 63980
    },
    {
      "epoch": 0.410411983337235,
      "grad_norm": 0.18189582228660583,
      "learning_rate": 9.666622583025915e-06,
      "loss": 0.0047,
      "step": 63990
    },
    {
      "epoch": 0.4104761202310211,
      "grad_norm": 0.274532288312912,
      "learning_rate": 9.66642160129867e-06,
      "loss": 0.0039,
      "step": 64000
    },
    {
      "epoch": 0.4105402571248072,
      "grad_norm": 0.1679452359676361,
      "learning_rate": 9.66622056109783e-06,
      "loss": 0.005,
      "step": 64010
    },
    {
      "epoch": 0.4106043940185933,
      "grad_norm": 0.1124291941523552,
      "learning_rate": 9.66601946242591e-06,
      "loss": 0.003,
      "step": 64020
    },
    {
      "epoch": 0.41066853091237937,
      "grad_norm": 0.0889439508318901,
      "learning_rate": 9.66581830528543e-06,
      "loss": 0.0034,
      "step": 64030
    },
    {
      "epoch": 0.41073266780616546,
      "grad_norm": 0.12128795683383942,
      "learning_rate": 9.665617089678913e-06,
      "loss": 0.0025,
      "step": 64040
    },
    {
      "epoch": 0.41079680469995156,
      "grad_norm": 0.18444238603115082,
      "learning_rate": 9.66541581560888e-06,
      "loss": 0.004,
      "step": 64050
    },
    {
      "epoch": 0.41086094159373765,
      "grad_norm": 0.08144691586494446,
      "learning_rate": 9.665214483077852e-06,
      "loss": 0.004,
      "step": 64060
    },
    {
      "epoch": 0.4109250784875238,
      "grad_norm": 0.25433894991874695,
      "learning_rate": 9.665013092088353e-06,
      "loss": 0.0038,
      "step": 64070
    },
    {
      "epoch": 0.4109892153813099,
      "grad_norm": 0.08502595871686935,
      "learning_rate": 9.664811642642905e-06,
      "loss": 0.0036,
      "step": 64080
    },
    {
      "epoch": 0.411053352275096,
      "grad_norm": 0.08852414786815643,
      "learning_rate": 9.664610134744034e-06,
      "loss": 0.0043,
      "step": 64090
    },
    {
      "epoch": 0.4111174891688821,
      "grad_norm": 0.11151493340730667,
      "learning_rate": 9.664408568394265e-06,
      "loss": 0.0033,
      "step": 64100
    },
    {
      "epoch": 0.41118162606266817,
      "grad_norm": 0.18578369915485382,
      "learning_rate": 9.664206943596122e-06,
      "loss": 0.0033,
      "step": 64110
    },
    {
      "epoch": 0.41124576295645426,
      "grad_norm": 0.09410499036312103,
      "learning_rate": 9.664005260352132e-06,
      "loss": 0.004,
      "step": 64120
    },
    {
      "epoch": 0.41130989985024036,
      "grad_norm": 0.24096010625362396,
      "learning_rate": 9.663803518664823e-06,
      "loss": 0.0042,
      "step": 64130
    },
    {
      "epoch": 0.41137403674402645,
      "grad_norm": 0.24409767985343933,
      "learning_rate": 9.663601718536724e-06,
      "loss": 0.0054,
      "step": 64140
    },
    {
      "epoch": 0.41143817363781254,
      "grad_norm": 0.33792802691459656,
      "learning_rate": 9.663399859970361e-06,
      "loss": 0.004,
      "step": 64150
    },
    {
      "epoch": 0.41150231053159864,
      "grad_norm": 0.18288464844226837,
      "learning_rate": 9.663197942968266e-06,
      "loss": 0.0044,
      "step": 64160
    },
    {
      "epoch": 0.41156644742538473,
      "grad_norm": 0.24157993495464325,
      "learning_rate": 9.662995967532966e-06,
      "loss": 0.0067,
      "step": 64170
    },
    {
      "epoch": 0.4116305843191708,
      "grad_norm": 0.24919168651103973,
      "learning_rate": 9.662793933666997e-06,
      "loss": 0.0028,
      "step": 64180
    },
    {
      "epoch": 0.4116947212129569,
      "grad_norm": 0.12063928693532944,
      "learning_rate": 9.662591841372885e-06,
      "loss": 0.0042,
      "step": 64190
    },
    {
      "epoch": 0.411758858106743,
      "grad_norm": 0.05848376452922821,
      "learning_rate": 9.662389690653166e-06,
      "loss": 0.0039,
      "step": 64200
    },
    {
      "epoch": 0.4118229950005291,
      "grad_norm": 0.09494820982217789,
      "learning_rate": 9.662187481510371e-06,
      "loss": 0.0035,
      "step": 64210
    },
    {
      "epoch": 0.41188713189431525,
      "grad_norm": 0.09575354307889938,
      "learning_rate": 9.661985213947037e-06,
      "loss": 0.0032,
      "step": 64220
    },
    {
      "epoch": 0.41195126878810134,
      "grad_norm": 0.36199530959129333,
      "learning_rate": 9.661782887965694e-06,
      "loss": 0.0095,
      "step": 64230
    },
    {
      "epoch": 0.41201540568188744,
      "grad_norm": 0.007769663352519274,
      "learning_rate": 9.661580503568881e-06,
      "loss": 0.0042,
      "step": 64240
    },
    {
      "epoch": 0.41207954257567353,
      "grad_norm": 0.2087806910276413,
      "learning_rate": 9.661378060759132e-06,
      "loss": 0.0036,
      "step": 64250
    },
    {
      "epoch": 0.4121436794694596,
      "grad_norm": 0.5809451937675476,
      "learning_rate": 9.661175559538987e-06,
      "loss": 0.0059,
      "step": 64260
    },
    {
      "epoch": 0.4122078163632457,
      "grad_norm": 0.29702523350715637,
      "learning_rate": 9.660972999910977e-06,
      "loss": 0.0032,
      "step": 64270
    },
    {
      "epoch": 0.4122719532570318,
      "grad_norm": 0.12000985443592072,
      "learning_rate": 9.660770381877647e-06,
      "loss": 0.0032,
      "step": 64280
    },
    {
      "epoch": 0.4123360901508179,
      "grad_norm": 0.26523804664611816,
      "learning_rate": 9.660567705441532e-06,
      "loss": 0.0035,
      "step": 64290
    },
    {
      "epoch": 0.412400227044604,
      "grad_norm": 0.20523472130298615,
      "learning_rate": 9.660364970605174e-06,
      "loss": 0.0037,
      "step": 64300
    },
    {
      "epoch": 0.4124643639383901,
      "grad_norm": 1.1858102083206177,
      "learning_rate": 9.66016217737111e-06,
      "loss": 0.0059,
      "step": 64310
    },
    {
      "epoch": 0.4125285008321762,
      "grad_norm": 0.09031245857477188,
      "learning_rate": 9.659959325741885e-06,
      "loss": 0.0035,
      "step": 64320
    },
    {
      "epoch": 0.4125926377259623,
      "grad_norm": 0.13391923904418945,
      "learning_rate": 9.659756415720038e-06,
      "loss": 0.0049,
      "step": 64330
    },
    {
      "epoch": 0.41265677461974837,
      "grad_norm": 0.21507328748703003,
      "learning_rate": 9.659553447308114e-06,
      "loss": 0.0041,
      "step": 64340
    },
    {
      "epoch": 0.41272091151353446,
      "grad_norm": 0.22505846619606018,
      "learning_rate": 9.659350420508656e-06,
      "loss": 0.005,
      "step": 64350
    },
    {
      "epoch": 0.4127850484073206,
      "grad_norm": 0.22557596862316132,
      "learning_rate": 9.659147335324205e-06,
      "loss": 0.0043,
      "step": 64360
    },
    {
      "epoch": 0.4128491853011067,
      "grad_norm": 0.16148000955581665,
      "learning_rate": 9.658944191757309e-06,
      "loss": 0.0044,
      "step": 64370
    },
    {
      "epoch": 0.4129133221948928,
      "grad_norm": 0.14673484861850739,
      "learning_rate": 9.658740989810512e-06,
      "loss": 0.0055,
      "step": 64380
    },
    {
      "epoch": 0.4129774590886789,
      "grad_norm": 0.06320397555828094,
      "learning_rate": 9.658537729486362e-06,
      "loss": 0.0029,
      "step": 64390
    },
    {
      "epoch": 0.413041595982465,
      "grad_norm": 0.15201781690120697,
      "learning_rate": 9.658334410787405e-06,
      "loss": 0.0035,
      "step": 64400
    },
    {
      "epoch": 0.4131057328762511,
      "grad_norm": 0.3072662353515625,
      "learning_rate": 9.658131033716187e-06,
      "loss": 0.0047,
      "step": 64410
    },
    {
      "epoch": 0.41316986977003717,
      "grad_norm": 0.10264178365468979,
      "learning_rate": 9.65792759827526e-06,
      "loss": 0.0031,
      "step": 64420
    },
    {
      "epoch": 0.41323400666382326,
      "grad_norm": 0.17404109239578247,
      "learning_rate": 9.65772410446717e-06,
      "loss": 0.0097,
      "step": 64430
    },
    {
      "epoch": 0.41329814355760935,
      "grad_norm": 0.35060179233551025,
      "learning_rate": 9.657520552294469e-06,
      "loss": 0.0041,
      "step": 64440
    },
    {
      "epoch": 0.41336228045139545,
      "grad_norm": 0.2839616537094116,
      "learning_rate": 9.657316941759705e-06,
      "loss": 0.0062,
      "step": 64450
    },
    {
      "epoch": 0.41342641734518154,
      "grad_norm": 0.21618971228599548,
      "learning_rate": 9.657113272865433e-06,
      "loss": 0.0026,
      "step": 64460
    },
    {
      "epoch": 0.41349055423896763,
      "grad_norm": 0.37534889578819275,
      "learning_rate": 9.656909545614201e-06,
      "loss": 0.0081,
      "step": 64470
    },
    {
      "epoch": 0.4135546911327537,
      "grad_norm": 0.176219180226326,
      "learning_rate": 9.656705760008566e-06,
      "loss": 0.0045,
      "step": 64480
    },
    {
      "epoch": 0.4136188280265398,
      "grad_norm": 0.3384178876876831,
      "learning_rate": 9.65650191605108e-06,
      "loss": 0.0034,
      "step": 64490
    },
    {
      "epoch": 0.41368296492032597,
      "grad_norm": 0.06688294559717178,
      "learning_rate": 9.656298013744296e-06,
      "loss": 0.0021,
      "step": 64500
    },
    {
      "epoch": 0.41374710181411206,
      "grad_norm": 0.17957651615142822,
      "learning_rate": 9.65609405309077e-06,
      "loss": 0.0032,
      "step": 64510
    },
    {
      "epoch": 0.41381123870789815,
      "grad_norm": 0.42804670333862305,
      "learning_rate": 9.655890034093059e-06,
      "loss": 0.0043,
      "step": 64520
    },
    {
      "epoch": 0.41387537560168425,
      "grad_norm": 0.29949918389320374,
      "learning_rate": 9.655685956753718e-06,
      "loss": 0.0039,
      "step": 64530
    },
    {
      "epoch": 0.41393951249547034,
      "grad_norm": 0.22192387282848358,
      "learning_rate": 9.655481821075305e-06,
      "loss": 0.0028,
      "step": 64540
    },
    {
      "epoch": 0.41400364938925643,
      "grad_norm": 0.17723120748996735,
      "learning_rate": 9.655277627060376e-06,
      "loss": 0.0071,
      "step": 64550
    },
    {
      "epoch": 0.4140677862830425,
      "grad_norm": 0.15555155277252197,
      "learning_rate": 9.655073374711492e-06,
      "loss": 0.0026,
      "step": 64560
    },
    {
      "epoch": 0.4141319231768286,
      "grad_norm": 0.2917649745941162,
      "learning_rate": 9.654869064031212e-06,
      "loss": 0.0042,
      "step": 64570
    },
    {
      "epoch": 0.4141960600706147,
      "grad_norm": 0.21928732097148895,
      "learning_rate": 9.654664695022096e-06,
      "loss": 0.0032,
      "step": 64580
    },
    {
      "epoch": 0.4142601969644008,
      "grad_norm": 0.3652102053165436,
      "learning_rate": 9.654460267686705e-06,
      "loss": 0.0046,
      "step": 64590
    },
    {
      "epoch": 0.4143243338581869,
      "grad_norm": 0.1732681393623352,
      "learning_rate": 9.654255782027599e-06,
      "loss": 0.0044,
      "step": 64600
    },
    {
      "epoch": 0.414388470751973,
      "grad_norm": 0.0742977187037468,
      "learning_rate": 9.654051238047343e-06,
      "loss": 0.0026,
      "step": 64610
    },
    {
      "epoch": 0.4144526076457591,
      "grad_norm": 0.2823348045349121,
      "learning_rate": 9.653846635748498e-06,
      "loss": 0.0085,
      "step": 64620
    },
    {
      "epoch": 0.4145167445395452,
      "grad_norm": 0.4840265214443207,
      "learning_rate": 9.65364197513363e-06,
      "loss": 0.0055,
      "step": 64630
    },
    {
      "epoch": 0.4145808814333313,
      "grad_norm": 0.2717534303665161,
      "learning_rate": 9.6534372562053e-06,
      "loss": 0.0057,
      "step": 64640
    },
    {
      "epoch": 0.4146450183271174,
      "grad_norm": 0.3104613423347473,
      "learning_rate": 9.653232478966076e-06,
      "loss": 0.0063,
      "step": 64650
    },
    {
      "epoch": 0.4147091552209035,
      "grad_norm": 0.25344496965408325,
      "learning_rate": 9.653027643418525e-06,
      "loss": 0.0042,
      "step": 64660
    },
    {
      "epoch": 0.4147732921146896,
      "grad_norm": 0.059659067541360855,
      "learning_rate": 9.65282274956521e-06,
      "loss": 0.005,
      "step": 64670
    },
    {
      "epoch": 0.4148374290084757,
      "grad_norm": 0.10441508889198303,
      "learning_rate": 9.652617797408702e-06,
      "loss": 0.0034,
      "step": 64680
    },
    {
      "epoch": 0.4149015659022618,
      "grad_norm": 0.054133035242557526,
      "learning_rate": 9.652412786951568e-06,
      "loss": 0.0031,
      "step": 64690
    },
    {
      "epoch": 0.4149657027960479,
      "grad_norm": 1.1095691919326782,
      "learning_rate": 9.652207718196376e-06,
      "loss": 0.0043,
      "step": 64700
    },
    {
      "epoch": 0.415029839689834,
      "grad_norm": 0.10577990114688873,
      "learning_rate": 9.652002591145697e-06,
      "loss": 0.0048,
      "step": 64710
    },
    {
      "epoch": 0.4150939765836201,
      "grad_norm": 0.19246532022953033,
      "learning_rate": 9.651797405802099e-06,
      "loss": 0.0029,
      "step": 64720
    },
    {
      "epoch": 0.41515811347740617,
      "grad_norm": 0.1980627179145813,
      "learning_rate": 9.651592162168157e-06,
      "loss": 0.0046,
      "step": 64730
    },
    {
      "epoch": 0.41522225037119226,
      "grad_norm": 0.11953790485858917,
      "learning_rate": 9.651386860246442e-06,
      "loss": 0.0042,
      "step": 64740
    },
    {
      "epoch": 0.41528638726497835,
      "grad_norm": 0.12883557379245758,
      "learning_rate": 9.651181500039522e-06,
      "loss": 0.0054,
      "step": 64750
    },
    {
      "epoch": 0.41535052415876444,
      "grad_norm": 0.23231658339500427,
      "learning_rate": 9.650976081549975e-06,
      "loss": 0.0027,
      "step": 64760
    },
    {
      "epoch": 0.41541466105255054,
      "grad_norm": 0.11305467039346695,
      "learning_rate": 9.650770604780375e-06,
      "loss": 0.0051,
      "step": 64770
    },
    {
      "epoch": 0.4154787979463367,
      "grad_norm": 0.10067601501941681,
      "learning_rate": 9.650565069733293e-06,
      "loss": 0.0049,
      "step": 64780
    },
    {
      "epoch": 0.4155429348401228,
      "grad_norm": 0.23430085182189941,
      "learning_rate": 9.650359476411309e-06,
      "loss": 0.0059,
      "step": 64790
    },
    {
      "epoch": 0.4156070717339089,
      "grad_norm": 0.1708226501941681,
      "learning_rate": 9.650153824816995e-06,
      "loss": 0.0033,
      "step": 64800
    },
    {
      "epoch": 0.41567120862769497,
      "grad_norm": 0.1529286801815033,
      "learning_rate": 9.649948114952932e-06,
      "loss": 0.0055,
      "step": 64810
    },
    {
      "epoch": 0.41573534552148106,
      "grad_norm": 0.01400473341345787,
      "learning_rate": 9.649742346821696e-06,
      "loss": 0.0049,
      "step": 64820
    },
    {
      "epoch": 0.41579948241526715,
      "grad_norm": 0.08111272007226944,
      "learning_rate": 9.649536520425864e-06,
      "loss": 0.0041,
      "step": 64830
    },
    {
      "epoch": 0.41586361930905325,
      "grad_norm": 0.1852743923664093,
      "learning_rate": 9.649330635768017e-06,
      "loss": 0.0034,
      "step": 64840
    },
    {
      "epoch": 0.41592775620283934,
      "grad_norm": 0.20524005591869354,
      "learning_rate": 9.649124692850736e-06,
      "loss": 0.0038,
      "step": 64850
    },
    {
      "epoch": 0.41599189309662543,
      "grad_norm": 0.18945936858654022,
      "learning_rate": 9.648918691676597e-06,
      "loss": 0.006,
      "step": 64860
    },
    {
      "epoch": 0.4160560299904115,
      "grad_norm": 0.5125710964202881,
      "learning_rate": 9.648712632248188e-06,
      "loss": 0.0038,
      "step": 64870
    },
    {
      "epoch": 0.4161201668841976,
      "grad_norm": 0.06856992095708847,
      "learning_rate": 9.648506514568083e-06,
      "loss": 0.0048,
      "step": 64880
    },
    {
      "epoch": 0.4161843037779837,
      "grad_norm": 0.31569772958755493,
      "learning_rate": 9.648300338638872e-06,
      "loss": 0.0048,
      "step": 64890
    },
    {
      "epoch": 0.4162484406717698,
      "grad_norm": 0.07696610689163208,
      "learning_rate": 9.648094104463135e-06,
      "loss": 0.0042,
      "step": 64900
    },
    {
      "epoch": 0.4163125775655559,
      "grad_norm": 0.15830492973327637,
      "learning_rate": 9.647887812043457e-06,
      "loss": 0.0025,
      "step": 64910
    },
    {
      "epoch": 0.41637671445934205,
      "grad_norm": 0.22576509416103363,
      "learning_rate": 9.647681461382421e-06,
      "loss": 0.0074,
      "step": 64920
    },
    {
      "epoch": 0.41644085135312814,
      "grad_norm": 0.21804648637771606,
      "learning_rate": 9.647475052482617e-06,
      "loss": 0.0055,
      "step": 64930
    },
    {
      "epoch": 0.41650498824691423,
      "grad_norm": 0.18545271456241608,
      "learning_rate": 9.647268585346627e-06,
      "loss": 0.0037,
      "step": 64940
    },
    {
      "epoch": 0.4165691251407003,
      "grad_norm": 0.16034992039203644,
      "learning_rate": 9.647062059977043e-06,
      "loss": 0.0027,
      "step": 64950
    },
    {
      "epoch": 0.4166332620344864,
      "grad_norm": 0.18599581718444824,
      "learning_rate": 9.646855476376448e-06,
      "loss": 0.0042,
      "step": 64960
    },
    {
      "epoch": 0.4166973989282725,
      "grad_norm": 0.21841683983802795,
      "learning_rate": 9.646648834547434e-06,
      "loss": 0.004,
      "step": 64970
    },
    {
      "epoch": 0.4167615358220586,
      "grad_norm": 0.06854478269815445,
      "learning_rate": 9.646442134492589e-06,
      "loss": 0.0045,
      "step": 64980
    },
    {
      "epoch": 0.4168256727158447,
      "grad_norm": 0.100202277302742,
      "learning_rate": 9.646235376214503e-06,
      "loss": 0.0036,
      "step": 64990
    },
    {
      "epoch": 0.4168898096096308,
      "grad_norm": 0.5943800210952759,
      "learning_rate": 9.646028559715767e-06,
      "loss": 0.0036,
      "step": 65000
    },
    {
      "epoch": 0.4169539465034169,
      "grad_norm": 0.21876904368400574,
      "learning_rate": 9.645821684998972e-06,
      "loss": 0.0032,
      "step": 65010
    },
    {
      "epoch": 0.417018083397203,
      "grad_norm": 0.1890275627374649,
      "learning_rate": 9.645614752066715e-06,
      "loss": 0.0036,
      "step": 65020
    },
    {
      "epoch": 0.41708222029098907,
      "grad_norm": 0.3648800849914551,
      "learning_rate": 9.64540776092158e-06,
      "loss": 0.006,
      "step": 65030
    },
    {
      "epoch": 0.41714635718477516,
      "grad_norm": 0.2900066077709198,
      "learning_rate": 9.645200711566169e-06,
      "loss": 0.0037,
      "step": 65040
    },
    {
      "epoch": 0.41721049407856126,
      "grad_norm": 0.20067039132118225,
      "learning_rate": 9.644993604003071e-06,
      "loss": 0.0045,
      "step": 65050
    },
    {
      "epoch": 0.4172746309723474,
      "grad_norm": 0.19734036922454834,
      "learning_rate": 9.644786438234884e-06,
      "loss": 0.0041,
      "step": 65060
    },
    {
      "epoch": 0.4173387678661335,
      "grad_norm": 0.1580866575241089,
      "learning_rate": 9.644579214264204e-06,
      "loss": 0.0029,
      "step": 65070
    },
    {
      "epoch": 0.4174029047599196,
      "grad_norm": 0.17017881572246552,
      "learning_rate": 9.644371932093627e-06,
      "loss": 0.0036,
      "step": 65080
    },
    {
      "epoch": 0.4174670416537057,
      "grad_norm": 0.13641498982906342,
      "learning_rate": 9.644164591725748e-06,
      "loss": 0.0056,
      "step": 65090
    },
    {
      "epoch": 0.4175311785474918,
      "grad_norm": 0.3891378939151764,
      "learning_rate": 9.643957193163172e-06,
      "loss": 0.0053,
      "step": 65100
    },
    {
      "epoch": 0.41759531544127787,
      "grad_norm": 0.3308427035808563,
      "learning_rate": 9.643749736408489e-06,
      "loss": 0.0034,
      "step": 65110
    },
    {
      "epoch": 0.41765945233506396,
      "grad_norm": 0.2427821010351181,
      "learning_rate": 9.643542221464306e-06,
      "loss": 0.0035,
      "step": 65120
    },
    {
      "epoch": 0.41772358922885006,
      "grad_norm": 0.04089587926864624,
      "learning_rate": 9.643334648333218e-06,
      "loss": 0.0044,
      "step": 65130
    },
    {
      "epoch": 0.41778772612263615,
      "grad_norm": 0.2266939878463745,
      "learning_rate": 9.64312701701783e-06,
      "loss": 0.0033,
      "step": 65140
    },
    {
      "epoch": 0.41785186301642224,
      "grad_norm": 0.17413191497325897,
      "learning_rate": 9.642919327520741e-06,
      "loss": 0.0044,
      "step": 65150
    },
    {
      "epoch": 0.41791599991020834,
      "grad_norm": 0.15931077301502228,
      "learning_rate": 9.642711579844558e-06,
      "loss": 0.0035,
      "step": 65160
    },
    {
      "epoch": 0.41798013680399443,
      "grad_norm": 0.1775781512260437,
      "learning_rate": 9.642503773991876e-06,
      "loss": 0.0028,
      "step": 65170
    },
    {
      "epoch": 0.4180442736977805,
      "grad_norm": 0.17191636562347412,
      "learning_rate": 9.642295909965305e-06,
      "loss": 0.0048,
      "step": 65180
    },
    {
      "epoch": 0.4181084105915666,
      "grad_norm": 0.29047107696533203,
      "learning_rate": 9.64208798776745e-06,
      "loss": 0.0045,
      "step": 65190
    },
    {
      "epoch": 0.41817254748535276,
      "grad_norm": 0.18347904086112976,
      "learning_rate": 9.641880007400915e-06,
      "loss": 0.0047,
      "step": 65200
    },
    {
      "epoch": 0.41823668437913886,
      "grad_norm": 0.11868282407522202,
      "learning_rate": 9.641671968868305e-06,
      "loss": 0.0048,
      "step": 65210
    },
    {
      "epoch": 0.41830082127292495,
      "grad_norm": 0.11434541642665863,
      "learning_rate": 9.641463872172227e-06,
      "loss": 0.003,
      "step": 65220
    },
    {
      "epoch": 0.41836495816671104,
      "grad_norm": 0.30812957882881165,
      "learning_rate": 9.64125571731529e-06,
      "loss": 0.0066,
      "step": 65230
    },
    {
      "epoch": 0.41842909506049714,
      "grad_norm": 0.11301054805517197,
      "learning_rate": 9.641047504300103e-06,
      "loss": 0.0064,
      "step": 65240
    },
    {
      "epoch": 0.41849323195428323,
      "grad_norm": 0.24813212454319,
      "learning_rate": 9.640839233129274e-06,
      "loss": 0.0043,
      "step": 65250
    },
    {
      "epoch": 0.4185573688480693,
      "grad_norm": 0.18483519554138184,
      "learning_rate": 9.640630903805411e-06,
      "loss": 0.0033,
      "step": 65260
    },
    {
      "epoch": 0.4186215057418554,
      "grad_norm": 0.11620765924453735,
      "learning_rate": 9.640422516331127e-06,
      "loss": 0.0031,
      "step": 65270
    },
    {
      "epoch": 0.4186856426356415,
      "grad_norm": 0.21796372532844543,
      "learning_rate": 9.640214070709033e-06,
      "loss": 0.0035,
      "step": 65280
    },
    {
      "epoch": 0.4187497795294276,
      "grad_norm": 0.18819208443164825,
      "learning_rate": 9.64000556694174e-06,
      "loss": 0.0035,
      "step": 65290
    },
    {
      "epoch": 0.4188139164232137,
      "grad_norm": 0.18838195502758026,
      "learning_rate": 9.639797005031859e-06,
      "loss": 0.0053,
      "step": 65300
    },
    {
      "epoch": 0.4188780533169998,
      "grad_norm": 0.22085854411125183,
      "learning_rate": 9.639588384982008e-06,
      "loss": 0.0109,
      "step": 65310
    },
    {
      "epoch": 0.4189421902107859,
      "grad_norm": 0.14810186624526978,
      "learning_rate": 9.639379706794798e-06,
      "loss": 0.0028,
      "step": 65320
    },
    {
      "epoch": 0.419006327104572,
      "grad_norm": 0.24572016298770905,
      "learning_rate": 9.639170970472845e-06,
      "loss": 0.0037,
      "step": 65330
    },
    {
      "epoch": 0.4190704639983581,
      "grad_norm": 0.28217583894729614,
      "learning_rate": 9.638962176018762e-06,
      "loss": 0.003,
      "step": 65340
    },
    {
      "epoch": 0.4191346008921442,
      "grad_norm": 0.06116315349936485,
      "learning_rate": 9.63875332343517e-06,
      "loss": 0.0045,
      "step": 65350
    },
    {
      "epoch": 0.4191987377859303,
      "grad_norm": 0.033075958490371704,
      "learning_rate": 9.638544412724682e-06,
      "loss": 0.0037,
      "step": 65360
    },
    {
      "epoch": 0.4192628746797164,
      "grad_norm": 0.10180090367794037,
      "learning_rate": 9.63833544388992e-06,
      "loss": 0.0043,
      "step": 65370
    },
    {
      "epoch": 0.4193270115735025,
      "grad_norm": 0.22493982315063477,
      "learning_rate": 9.638126416933497e-06,
      "loss": 0.003,
      "step": 65380
    },
    {
      "epoch": 0.4193911484672886,
      "grad_norm": 0.1564272940158844,
      "learning_rate": 9.637917331858037e-06,
      "loss": 0.0034,
      "step": 65390
    },
    {
      "epoch": 0.4194552853610747,
      "grad_norm": 0.1368493139743805,
      "learning_rate": 9.637708188666156e-06,
      "loss": 0.0039,
      "step": 65400
    },
    {
      "epoch": 0.4195194222548608,
      "grad_norm": 0.16612175107002258,
      "learning_rate": 9.637498987360479e-06,
      "loss": 0.0046,
      "step": 65410
    },
    {
      "epoch": 0.41958355914864687,
      "grad_norm": 0.17109805345535278,
      "learning_rate": 9.637289727943625e-06,
      "loss": 0.0036,
      "step": 65420
    },
    {
      "epoch": 0.41964769604243296,
      "grad_norm": 0.08433777093887329,
      "learning_rate": 9.637080410418215e-06,
      "loss": 0.0061,
      "step": 65430
    },
    {
      "epoch": 0.41971183293621905,
      "grad_norm": 0.30764898657798767,
      "learning_rate": 9.636871034786876e-06,
      "loss": 0.0061,
      "step": 65440
    },
    {
      "epoch": 0.41977596983000515,
      "grad_norm": 0.08212987333536148,
      "learning_rate": 9.636661601052227e-06,
      "loss": 0.005,
      "step": 65450
    },
    {
      "epoch": 0.41984010672379124,
      "grad_norm": 0.32802894711494446,
      "learning_rate": 9.636452109216894e-06,
      "loss": 0.0032,
      "step": 65460
    },
    {
      "epoch": 0.41990424361757733,
      "grad_norm": 0.1307886391878128,
      "learning_rate": 9.636242559283505e-06,
      "loss": 0.0055,
      "step": 65470
    },
    {
      "epoch": 0.4199683805113635,
      "grad_norm": 0.2157278209924698,
      "learning_rate": 9.636032951254681e-06,
      "loss": 0.0039,
      "step": 65480
    },
    {
      "epoch": 0.4200325174051496,
      "grad_norm": 0.2503538429737091,
      "learning_rate": 9.635823285133052e-06,
      "loss": 0.0049,
      "step": 65490
    },
    {
      "epoch": 0.42009665429893567,
      "grad_norm": 0.34493184089660645,
      "learning_rate": 9.635613560921242e-06,
      "loss": 0.0041,
      "step": 65500
    },
    {
      "epoch": 0.42016079119272176,
      "grad_norm": 0.2957816421985626,
      "learning_rate": 9.635403778621883e-06,
      "loss": 0.0061,
      "step": 65510
    },
    {
      "epoch": 0.42022492808650785,
      "grad_norm": 0.22828510403633118,
      "learning_rate": 9.6351939382376e-06,
      "loss": 0.0033,
      "step": 65520
    },
    {
      "epoch": 0.42028906498029395,
      "grad_norm": 0.31681329011917114,
      "learning_rate": 9.634984039771027e-06,
      "loss": 0.0045,
      "step": 65530
    },
    {
      "epoch": 0.42035320187408004,
      "grad_norm": 0.1677008718252182,
      "learning_rate": 9.634774083224789e-06,
      "loss": 0.0042,
      "step": 65540
    },
    {
      "epoch": 0.42041733876786613,
      "grad_norm": 0.0993567481637001,
      "learning_rate": 9.634564068601521e-06,
      "loss": 0.0022,
      "step": 65550
    },
    {
      "epoch": 0.4204814756616522,
      "grad_norm": 0.1264619082212448,
      "learning_rate": 9.634353995903852e-06,
      "loss": 0.0065,
      "step": 65560
    },
    {
      "epoch": 0.4205456125554383,
      "grad_norm": 0.04126043617725372,
      "learning_rate": 9.634143865134415e-06,
      "loss": 0.0027,
      "step": 65570
    },
    {
      "epoch": 0.4206097494492244,
      "grad_norm": 0.20881688594818115,
      "learning_rate": 9.633933676295847e-06,
      "loss": 0.003,
      "step": 65580
    },
    {
      "epoch": 0.4206738863430105,
      "grad_norm": 0.20779581367969513,
      "learning_rate": 9.633723429390773e-06,
      "loss": 0.0045,
      "step": 65590
    },
    {
      "epoch": 0.4207380232367966,
      "grad_norm": 0.18272095918655396,
      "learning_rate": 9.633513124421834e-06,
      "loss": 0.0043,
      "step": 65600
    },
    {
      "epoch": 0.4208021601305827,
      "grad_norm": 0.1345992386341095,
      "learning_rate": 9.633302761391665e-06,
      "loss": 0.0039,
      "step": 65610
    },
    {
      "epoch": 0.4208662970243688,
      "grad_norm": 0.275342732667923,
      "learning_rate": 9.633092340302902e-06,
      "loss": 0.003,
      "step": 65620
    },
    {
      "epoch": 0.42093043391815493,
      "grad_norm": 0.27025195956230164,
      "learning_rate": 9.632881861158179e-06,
      "loss": 0.0051,
      "step": 65630
    },
    {
      "epoch": 0.420994570811941,
      "grad_norm": 0.06833307445049286,
      "learning_rate": 9.632671323960137e-06,
      "loss": 0.0034,
      "step": 65640
    },
    {
      "epoch": 0.4210587077057271,
      "grad_norm": 0.20985299348831177,
      "learning_rate": 9.63246072871141e-06,
      "loss": 0.0033,
      "step": 65650
    },
    {
      "epoch": 0.4211228445995132,
      "grad_norm": 0.09783653914928436,
      "learning_rate": 9.632250075414642e-06,
      "loss": 0.0026,
      "step": 65660
    },
    {
      "epoch": 0.4211869814932993,
      "grad_norm": 0.27587318420410156,
      "learning_rate": 9.632039364072467e-06,
      "loss": 0.0043,
      "step": 65670
    },
    {
      "epoch": 0.4212511183870854,
      "grad_norm": 0.14510606229305267,
      "learning_rate": 9.63182859468753e-06,
      "loss": 0.0033,
      "step": 65680
    },
    {
      "epoch": 0.4213152552808715,
      "grad_norm": 0.2852429449558258,
      "learning_rate": 9.631617767262472e-06,
      "loss": 0.0032,
      "step": 65690
    },
    {
      "epoch": 0.4213793921746576,
      "grad_norm": 0.23456509411334991,
      "learning_rate": 9.63140688179993e-06,
      "loss": 0.0071,
      "step": 65700
    },
    {
      "epoch": 0.4214435290684437,
      "grad_norm": 0.6252493262290955,
      "learning_rate": 9.631195938302551e-06,
      "loss": 0.0053,
      "step": 65710
    },
    {
      "epoch": 0.42150766596222977,
      "grad_norm": 0.3259626030921936,
      "learning_rate": 9.630984936772979e-06,
      "loss": 0.0049,
      "step": 65720
    },
    {
      "epoch": 0.42157180285601586,
      "grad_norm": 0.26559481024742126,
      "learning_rate": 9.630773877213853e-06,
      "loss": 0.0042,
      "step": 65730
    },
    {
      "epoch": 0.42163593974980196,
      "grad_norm": 0.1573127657175064,
      "learning_rate": 9.630562759627822e-06,
      "loss": 0.0043,
      "step": 65740
    },
    {
      "epoch": 0.42170007664358805,
      "grad_norm": 0.18279732763767242,
      "learning_rate": 9.63035158401753e-06,
      "loss": 0.0024,
      "step": 65750
    },
    {
      "epoch": 0.42176421353737414,
      "grad_norm": 0.14254005253314972,
      "learning_rate": 9.630140350385623e-06,
      "loss": 0.0045,
      "step": 65760
    },
    {
      "epoch": 0.4218283504311603,
      "grad_norm": 0.16114702820777893,
      "learning_rate": 9.62992905873475e-06,
      "loss": 0.003,
      "step": 65770
    },
    {
      "epoch": 0.4218924873249464,
      "grad_norm": 0.4747255742549896,
      "learning_rate": 9.629717709067557e-06,
      "loss": 0.0072,
      "step": 65780
    },
    {
      "epoch": 0.4219566242187325,
      "grad_norm": 0.1535458266735077,
      "learning_rate": 9.629506301386689e-06,
      "loss": 0.0047,
      "step": 65790
    },
    {
      "epoch": 0.42202076111251857,
      "grad_norm": 0.15451142191886902,
      "learning_rate": 9.629294835694801e-06,
      "loss": 0.0057,
      "step": 65800
    },
    {
      "epoch": 0.42208489800630467,
      "grad_norm": 0.1924429088830948,
      "learning_rate": 9.629083311994541e-06,
      "loss": 0.0047,
      "step": 65810
    },
    {
      "epoch": 0.42214903490009076,
      "grad_norm": 0.05497288703918457,
      "learning_rate": 9.628871730288555e-06,
      "loss": 0.0042,
      "step": 65820
    },
    {
      "epoch": 0.42221317179387685,
      "grad_norm": 0.2067718654870987,
      "learning_rate": 9.628660090579503e-06,
      "loss": 0.005,
      "step": 65830
    },
    {
      "epoch": 0.42227730868766294,
      "grad_norm": 0.16331195831298828,
      "learning_rate": 9.62844839287003e-06,
      "loss": 0.0044,
      "step": 65840
    },
    {
      "epoch": 0.42234144558144904,
      "grad_norm": 0.14620499312877655,
      "learning_rate": 9.628236637162788e-06,
      "loss": 0.0032,
      "step": 65850
    },
    {
      "epoch": 0.42240558247523513,
      "grad_norm": 0.0732194036245346,
      "learning_rate": 9.628024823460436e-06,
      "loss": 0.0023,
      "step": 65860
    },
    {
      "epoch": 0.4224697193690212,
      "grad_norm": 0.5353759527206421,
      "learning_rate": 9.627812951765624e-06,
      "loss": 0.0053,
      "step": 65870
    },
    {
      "epoch": 0.4225338562628073,
      "grad_norm": 0.18316306173801422,
      "learning_rate": 9.62760102208101e-06,
      "loss": 0.0036,
      "step": 65880
    },
    {
      "epoch": 0.4225979931565934,
      "grad_norm": 0.1757930964231491,
      "learning_rate": 9.627389034409247e-06,
      "loss": 0.0036,
      "step": 65890
    },
    {
      "epoch": 0.4226621300503795,
      "grad_norm": 0.1982915848493576,
      "learning_rate": 9.627176988752992e-06,
      "loss": 0.0043,
      "step": 65900
    },
    {
      "epoch": 0.42272626694416565,
      "grad_norm": 0.1785518229007721,
      "learning_rate": 9.626964885114902e-06,
      "loss": 0.0061,
      "step": 65910
    },
    {
      "epoch": 0.42279040383795174,
      "grad_norm": 0.19323821365833282,
      "learning_rate": 9.626752723497637e-06,
      "loss": 0.0064,
      "step": 65920
    },
    {
      "epoch": 0.42285454073173784,
      "grad_norm": 0.15979072451591492,
      "learning_rate": 9.626540503903852e-06,
      "loss": 0.004,
      "step": 65930
    },
    {
      "epoch": 0.42291867762552393,
      "grad_norm": 0.10884411633014679,
      "learning_rate": 9.626328226336209e-06,
      "loss": 0.0066,
      "step": 65940
    },
    {
      "epoch": 0.42298281451931,
      "grad_norm": 0.21978871524333954,
      "learning_rate": 9.626115890797365e-06,
      "loss": 0.004,
      "step": 65950
    },
    {
      "epoch": 0.4230469514130961,
      "grad_norm": 0.10031398385763168,
      "learning_rate": 9.625903497289984e-06,
      "loss": 0.0029,
      "step": 65960
    },
    {
      "epoch": 0.4231110883068822,
      "grad_norm": 0.10546796768903732,
      "learning_rate": 9.625691045816726e-06,
      "loss": 0.0042,
      "step": 65970
    },
    {
      "epoch": 0.4231752252006683,
      "grad_norm": 0.11720538139343262,
      "learning_rate": 9.625478536380252e-06,
      "loss": 0.0058,
      "step": 65980
    },
    {
      "epoch": 0.4232393620944544,
      "grad_norm": 0.044053900986909866,
      "learning_rate": 9.625265968983228e-06,
      "loss": 0.0031,
      "step": 65990
    },
    {
      "epoch": 0.4233034989882405,
      "grad_norm": 0.29401299357414246,
      "learning_rate": 9.625053343628316e-06,
      "loss": 0.0053,
      "step": 66000
    },
    {
      "epoch": 0.4233676358820266,
      "grad_norm": 0.11990197002887726,
      "learning_rate": 9.62484066031818e-06,
      "loss": 0.0048,
      "step": 66010
    },
    {
      "epoch": 0.4234317727758127,
      "grad_norm": 0.23408041894435883,
      "learning_rate": 9.624627919055484e-06,
      "loss": 0.0041,
      "step": 66020
    },
    {
      "epoch": 0.42349590966959877,
      "grad_norm": 0.16822808980941772,
      "learning_rate": 9.624415119842895e-06,
      "loss": 0.0032,
      "step": 66030
    },
    {
      "epoch": 0.42356004656338486,
      "grad_norm": 0.3313102722167969,
      "learning_rate": 9.62420226268308e-06,
      "loss": 0.005,
      "step": 66040
    },
    {
      "epoch": 0.423624183457171,
      "grad_norm": 0.21190787851810455,
      "learning_rate": 9.623989347578706e-06,
      "loss": 0.0029,
      "step": 66050
    },
    {
      "epoch": 0.4236883203509571,
      "grad_norm": 0.1187615618109703,
      "learning_rate": 9.623776374532442e-06,
      "loss": 0.0078,
      "step": 66060
    },
    {
      "epoch": 0.4237524572447432,
      "grad_norm": 0.20854294300079346,
      "learning_rate": 9.623563343546954e-06,
      "loss": 0.0039,
      "step": 66070
    },
    {
      "epoch": 0.4238165941385293,
      "grad_norm": 0.23553693294525146,
      "learning_rate": 9.62335025462491e-06,
      "loss": 0.0047,
      "step": 66080
    },
    {
      "epoch": 0.4238807310323154,
      "grad_norm": 0.042550235986709595,
      "learning_rate": 9.623137107768987e-06,
      "loss": 0.0044,
      "step": 66090
    },
    {
      "epoch": 0.4239448679261015,
      "grad_norm": 0.12277571856975555,
      "learning_rate": 9.62292390298185e-06,
      "loss": 0.0039,
      "step": 66100
    },
    {
      "epoch": 0.42400900481988757,
      "grad_norm": 0.24997268617153168,
      "learning_rate": 9.622710640266175e-06,
      "loss": 0.0053,
      "step": 66110
    },
    {
      "epoch": 0.42407314171367366,
      "grad_norm": 0.11305372416973114,
      "learning_rate": 9.62249731962463e-06,
      "loss": 0.0045,
      "step": 66120
    },
    {
      "epoch": 0.42413727860745976,
      "grad_norm": 0.3188675045967102,
      "learning_rate": 9.62228394105989e-06,
      "loss": 0.0035,
      "step": 66130
    },
    {
      "epoch": 0.42420141550124585,
      "grad_norm": 0.15568873286247253,
      "learning_rate": 9.62207050457463e-06,
      "loss": 0.0042,
      "step": 66140
    },
    {
      "epoch": 0.42426555239503194,
      "grad_norm": 0.3959549069404602,
      "learning_rate": 9.621857010171523e-06,
      "loss": 0.0045,
      "step": 66150
    },
    {
      "epoch": 0.42432968928881803,
      "grad_norm": 0.11153994500637054,
      "learning_rate": 9.621643457853243e-06,
      "loss": 0.0048,
      "step": 66160
    },
    {
      "epoch": 0.42439382618260413,
      "grad_norm": 0.08319900184869766,
      "learning_rate": 9.621429847622467e-06,
      "loss": 0.0032,
      "step": 66170
    },
    {
      "epoch": 0.4244579630763902,
      "grad_norm": 0.2318873405456543,
      "learning_rate": 9.621216179481874e-06,
      "loss": 0.0062,
      "step": 66180
    },
    {
      "epoch": 0.42452209997017637,
      "grad_norm": 0.16006936132907867,
      "learning_rate": 9.621002453434138e-06,
      "loss": 0.0054,
      "step": 66190
    },
    {
      "epoch": 0.42458623686396246,
      "grad_norm": 0.24659541249275208,
      "learning_rate": 9.62078866948194e-06,
      "loss": 0.0031,
      "step": 66200
    },
    {
      "epoch": 0.42465037375774856,
      "grad_norm": 0.0987880527973175,
      "learning_rate": 9.620574827627957e-06,
      "loss": 0.0038,
      "step": 66210
    },
    {
      "epoch": 0.42471451065153465,
      "grad_norm": 0.18199047446250916,
      "learning_rate": 9.62036092787487e-06,
      "loss": 0.0044,
      "step": 66220
    },
    {
      "epoch": 0.42477864754532074,
      "grad_norm": 0.16193823516368866,
      "learning_rate": 9.620146970225357e-06,
      "loss": 0.0042,
      "step": 66230
    },
    {
      "epoch": 0.42484278443910684,
      "grad_norm": 0.32942768931388855,
      "learning_rate": 9.6199329546821e-06,
      "loss": 0.0044,
      "step": 66240
    },
    {
      "epoch": 0.42490692133289293,
      "grad_norm": 0.30845940113067627,
      "learning_rate": 9.619718881247784e-06,
      "loss": 0.006,
      "step": 66250
    },
    {
      "epoch": 0.424971058226679,
      "grad_norm": 0.1413591355085373,
      "learning_rate": 9.619504749925086e-06,
      "loss": 0.0048,
      "step": 66260
    },
    {
      "epoch": 0.4250351951204651,
      "grad_norm": 0.019584035500884056,
      "learning_rate": 9.619290560716694e-06,
      "loss": 0.0022,
      "step": 66270
    },
    {
      "epoch": 0.4250993320142512,
      "grad_norm": 0.02616703324019909,
      "learning_rate": 9.619076313625289e-06,
      "loss": 0.004,
      "step": 66280
    },
    {
      "epoch": 0.4251634689080373,
      "grad_norm": 0.3452214300632477,
      "learning_rate": 9.618862008653556e-06,
      "loss": 0.0038,
      "step": 66290
    },
    {
      "epoch": 0.4252276058018234,
      "grad_norm": 0.06459818035364151,
      "learning_rate": 9.618647645804182e-06,
      "loss": 0.0047,
      "step": 66300
    },
    {
      "epoch": 0.4252917426956095,
      "grad_norm": 0.15330830216407776,
      "learning_rate": 9.618433225079852e-06,
      "loss": 0.0035,
      "step": 66310
    },
    {
      "epoch": 0.4253558795893956,
      "grad_norm": 0.17286977171897888,
      "learning_rate": 9.618218746483251e-06,
      "loss": 0.0037,
      "step": 66320
    },
    {
      "epoch": 0.42542001648318173,
      "grad_norm": 0.498504102230072,
      "learning_rate": 9.618004210017072e-06,
      "loss": 0.0046,
      "step": 66330
    },
    {
      "epoch": 0.4254841533769678,
      "grad_norm": 0.043882135301828384,
      "learning_rate": 9.617789615683998e-06,
      "loss": 0.0043,
      "step": 66340
    },
    {
      "epoch": 0.4255482902707539,
      "grad_norm": 0.15280118584632874,
      "learning_rate": 9.61757496348672e-06,
      "loss": 0.0039,
      "step": 66350
    },
    {
      "epoch": 0.42561242716454,
      "grad_norm": 0.1655089557170868,
      "learning_rate": 9.617360253427927e-06,
      "loss": 0.0027,
      "step": 66360
    },
    {
      "epoch": 0.4256765640583261,
      "grad_norm": 0.11018262803554535,
      "learning_rate": 9.617145485510311e-06,
      "loss": 0.0031,
      "step": 66370
    },
    {
      "epoch": 0.4257407009521122,
      "grad_norm": 0.16910359263420105,
      "learning_rate": 9.616930659736561e-06,
      "loss": 0.0051,
      "step": 66380
    },
    {
      "epoch": 0.4258048378458983,
      "grad_norm": 0.38406896591186523,
      "learning_rate": 9.61671577610937e-06,
      "loss": 0.0055,
      "step": 66390
    },
    {
      "epoch": 0.4258689747396844,
      "grad_norm": 0.2939324378967285,
      "learning_rate": 9.616500834631433e-06,
      "loss": 0.0027,
      "step": 66400
    },
    {
      "epoch": 0.4259331116334705,
      "grad_norm": 0.24132554233074188,
      "learning_rate": 9.61628583530544e-06,
      "loss": 0.003,
      "step": 66410
    },
    {
      "epoch": 0.42599724852725657,
      "grad_norm": 0.3862217664718628,
      "learning_rate": 9.616070778134085e-06,
      "loss": 0.0046,
      "step": 66420
    },
    {
      "epoch": 0.42606138542104266,
      "grad_norm": 0.2583540081977844,
      "learning_rate": 9.615855663120066e-06,
      "loss": 0.0038,
      "step": 66430
    },
    {
      "epoch": 0.42612552231482875,
      "grad_norm": 0.07580406218767166,
      "learning_rate": 9.615640490266074e-06,
      "loss": 0.0042,
      "step": 66440
    },
    {
      "epoch": 0.42618965920861485,
      "grad_norm": 0.22733516991138458,
      "learning_rate": 9.615425259574812e-06,
      "loss": 0.0039,
      "step": 66450
    },
    {
      "epoch": 0.42625379610240094,
      "grad_norm": 0.25381848216056824,
      "learning_rate": 9.61520997104897e-06,
      "loss": 0.0045,
      "step": 66460
    },
    {
      "epoch": 0.4263179329961871,
      "grad_norm": 0.107947438955307,
      "learning_rate": 9.614994624691248e-06,
      "loss": 0.0029,
      "step": 66470
    },
    {
      "epoch": 0.4263820698899732,
      "grad_norm": 0.11911238729953766,
      "learning_rate": 9.614779220504347e-06,
      "loss": 0.0034,
      "step": 66480
    },
    {
      "epoch": 0.4264462067837593,
      "grad_norm": 0.26806512475013733,
      "learning_rate": 9.614563758490963e-06,
      "loss": 0.0039,
      "step": 66490
    },
    {
      "epoch": 0.42651034367754537,
      "grad_norm": 0.25328999757766724,
      "learning_rate": 9.6143482386538e-06,
      "loss": 0.0038,
      "step": 66500
    },
    {
      "epoch": 0.42657448057133146,
      "grad_norm": 0.1805124580860138,
      "learning_rate": 9.614132660995553e-06,
      "loss": 0.0036,
      "step": 66510
    },
    {
      "epoch": 0.42663861746511755,
      "grad_norm": 0.32967978715896606,
      "learning_rate": 9.613917025518925e-06,
      "loss": 0.0045,
      "step": 66520
    },
    {
      "epoch": 0.42670275435890365,
      "grad_norm": 0.12399055808782578,
      "learning_rate": 9.613701332226622e-06,
      "loss": 0.0048,
      "step": 66530
    },
    {
      "epoch": 0.42676689125268974,
      "grad_norm": 0.17118918895721436,
      "learning_rate": 9.61348558112134e-06,
      "loss": 0.0038,
      "step": 66540
    },
    {
      "epoch": 0.42683102814647583,
      "grad_norm": 0.15596435964107513,
      "learning_rate": 9.613269772205791e-06,
      "loss": 0.0083,
      "step": 66550
    },
    {
      "epoch": 0.4268951650402619,
      "grad_norm": 0.11981067806482315,
      "learning_rate": 9.613053905482672e-06,
      "loss": 0.0028,
      "step": 66560
    },
    {
      "epoch": 0.426959301934048,
      "grad_norm": 0.2925792932510376,
      "learning_rate": 9.612837980954692e-06,
      "loss": 0.0056,
      "step": 66570
    },
    {
      "epoch": 0.4270234388278341,
      "grad_norm": 0.07776892930269241,
      "learning_rate": 9.612621998624554e-06,
      "loss": 0.005,
      "step": 66580
    },
    {
      "epoch": 0.4270875757216202,
      "grad_norm": 0.20200687646865845,
      "learning_rate": 9.612405958494967e-06,
      "loss": 0.0029,
      "step": 66590
    },
    {
      "epoch": 0.4271517126154063,
      "grad_norm": 0.1193302720785141,
      "learning_rate": 9.612189860568636e-06,
      "loss": 0.0036,
      "step": 66600
    },
    {
      "epoch": 0.42721584950919245,
      "grad_norm": 0.19039000570774078,
      "learning_rate": 9.61197370484827e-06,
      "loss": 0.0042,
      "step": 66610
    },
    {
      "epoch": 0.42727998640297854,
      "grad_norm": 0.21106891334056854,
      "learning_rate": 9.611757491336578e-06,
      "loss": 0.0055,
      "step": 66620
    },
    {
      "epoch": 0.42734412329676463,
      "grad_norm": 0.11191011965274811,
      "learning_rate": 9.61154122003627e-06,
      "loss": 0.0037,
      "step": 66630
    },
    {
      "epoch": 0.4274082601905507,
      "grad_norm": 0.15065521001815796,
      "learning_rate": 9.611324890950052e-06,
      "loss": 0.0037,
      "step": 66640
    },
    {
      "epoch": 0.4274723970843368,
      "grad_norm": 0.21732980012893677,
      "learning_rate": 9.611108504080637e-06,
      "loss": 0.0054,
      "step": 66650
    },
    {
      "epoch": 0.4275365339781229,
      "grad_norm": 0.15280568599700928,
      "learning_rate": 9.610892059430738e-06,
      "loss": 0.0032,
      "step": 66660
    },
    {
      "epoch": 0.427600670871909,
      "grad_norm": 0.14303740859031677,
      "learning_rate": 9.610675557003067e-06,
      "loss": 0.0106,
      "step": 66670
    },
    {
      "epoch": 0.4276648077656951,
      "grad_norm": 0.30987024307250977,
      "learning_rate": 9.610458996800336e-06,
      "loss": 0.0051,
      "step": 66680
    },
    {
      "epoch": 0.4277289446594812,
      "grad_norm": 0.16005373001098633,
      "learning_rate": 9.610242378825258e-06,
      "loss": 0.0045,
      "step": 66690
    },
    {
      "epoch": 0.4277930815532673,
      "grad_norm": 0.23138290643692017,
      "learning_rate": 9.61002570308055e-06,
      "loss": 0.0036,
      "step": 66700
    },
    {
      "epoch": 0.4278572184470534,
      "grad_norm": 0.29233992099761963,
      "learning_rate": 9.609808969568922e-06,
      "loss": 0.0039,
      "step": 66710
    },
    {
      "epoch": 0.42792135534083947,
      "grad_norm": 0.256077378988266,
      "learning_rate": 9.609592178293095e-06,
      "loss": 0.0046,
      "step": 66720
    },
    {
      "epoch": 0.42798549223462556,
      "grad_norm": 0.24460352957248688,
      "learning_rate": 9.609375329255784e-06,
      "loss": 0.0058,
      "step": 66730
    },
    {
      "epoch": 0.42804962912841166,
      "grad_norm": 0.5737667083740234,
      "learning_rate": 9.609158422459703e-06,
      "loss": 0.0045,
      "step": 66740
    },
    {
      "epoch": 0.4281137660221978,
      "grad_norm": 0.09366708993911743,
      "learning_rate": 9.608941457907576e-06,
      "loss": 0.0023,
      "step": 66750
    },
    {
      "epoch": 0.4281779029159839,
      "grad_norm": 0.12595345079898834,
      "learning_rate": 9.608724435602117e-06,
      "loss": 0.0044,
      "step": 66760
    },
    {
      "epoch": 0.42824203980977,
      "grad_norm": 0.27345553040504456,
      "learning_rate": 9.608507355546048e-06,
      "loss": 0.0031,
      "step": 66770
    },
    {
      "epoch": 0.4283061767035561,
      "grad_norm": 0.261642187833786,
      "learning_rate": 9.608290217742086e-06,
      "loss": 0.0047,
      "step": 66780
    },
    {
      "epoch": 0.4283703135973422,
      "grad_norm": 0.16929571330547333,
      "learning_rate": 9.608073022192956e-06,
      "loss": 0.0035,
      "step": 66790
    },
    {
      "epoch": 0.42843445049112827,
      "grad_norm": 0.07420452684164047,
      "learning_rate": 9.607855768901378e-06,
      "loss": 0.0048,
      "step": 66800
    },
    {
      "epoch": 0.42849858738491436,
      "grad_norm": 0.04938085377216339,
      "learning_rate": 9.607638457870075e-06,
      "loss": 0.0025,
      "step": 66810
    },
    {
      "epoch": 0.42856272427870046,
      "grad_norm": 0.0963282585144043,
      "learning_rate": 9.607421089101767e-06,
      "loss": 0.0033,
      "step": 66820
    },
    {
      "epoch": 0.42862686117248655,
      "grad_norm": 0.1891966015100479,
      "learning_rate": 9.60720366259918e-06,
      "loss": 0.0021,
      "step": 66830
    },
    {
      "epoch": 0.42869099806627264,
      "grad_norm": 0.15832215547561646,
      "learning_rate": 9.606986178365038e-06,
      "loss": 0.0037,
      "step": 66840
    },
    {
      "epoch": 0.42875513496005874,
      "grad_norm": 0.39101964235305786,
      "learning_rate": 9.606768636402069e-06,
      "loss": 0.0054,
      "step": 66850
    },
    {
      "epoch": 0.42881927185384483,
      "grad_norm": 0.17650574445724487,
      "learning_rate": 9.606551036712995e-06,
      "loss": 0.0077,
      "step": 66860
    },
    {
      "epoch": 0.4288834087476309,
      "grad_norm": 0.15891918540000916,
      "learning_rate": 9.606333379300542e-06,
      "loss": 0.0047,
      "step": 66870
    },
    {
      "epoch": 0.428947545641417,
      "grad_norm": 0.18898074328899384,
      "learning_rate": 9.606115664167443e-06,
      "loss": 0.0026,
      "step": 66880
    },
    {
      "epoch": 0.42901168253520316,
      "grad_norm": 0.05703195556998253,
      "learning_rate": 9.605897891316422e-06,
      "loss": 0.0038,
      "step": 66890
    },
    {
      "epoch": 0.42907581942898926,
      "grad_norm": 0.24253630638122559,
      "learning_rate": 9.605680060750208e-06,
      "loss": 0.004,
      "step": 66900
    },
    {
      "epoch": 0.42913995632277535,
      "grad_norm": 0.3023951053619385,
      "learning_rate": 9.605462172471532e-06,
      "loss": 0.0038,
      "step": 66910
    },
    {
      "epoch": 0.42920409321656144,
      "grad_norm": 0.13640546798706055,
      "learning_rate": 9.605244226483123e-06,
      "loss": 0.005,
      "step": 66920
    },
    {
      "epoch": 0.42926823011034754,
      "grad_norm": 0.2519367039203644,
      "learning_rate": 9.605026222787712e-06,
      "loss": 0.0038,
      "step": 66930
    },
    {
      "epoch": 0.42933236700413363,
      "grad_norm": 0.26465660333633423,
      "learning_rate": 9.604808161388034e-06,
      "loss": 0.0032,
      "step": 66940
    },
    {
      "epoch": 0.4293965038979197,
      "grad_norm": 0.16541948914527893,
      "learning_rate": 9.604590042286814e-06,
      "loss": 0.0039,
      "step": 66950
    },
    {
      "epoch": 0.4294606407917058,
      "grad_norm": 0.20352678000926971,
      "learning_rate": 9.604371865486793e-06,
      "loss": 0.0037,
      "step": 66960
    },
    {
      "epoch": 0.4295247776854919,
      "grad_norm": 0.323935866355896,
      "learning_rate": 9.6041536309907e-06,
      "loss": 0.0042,
      "step": 66970
    },
    {
      "epoch": 0.429588914579278,
      "grad_norm": 0.22220510244369507,
      "learning_rate": 9.603935338801275e-06,
      "loss": 0.0039,
      "step": 66980
    },
    {
      "epoch": 0.4296530514730641,
      "grad_norm": 0.20077809691429138,
      "learning_rate": 9.603716988921249e-06,
      "loss": 0.0046,
      "step": 66990
    },
    {
      "epoch": 0.4297171883668502,
      "grad_norm": 0.05188808962702751,
      "learning_rate": 9.603498581353355e-06,
      "loss": 0.0047,
      "step": 67000
    },
    {
      "epoch": 0.4297813252606363,
      "grad_norm": 0.1651322990655899,
      "learning_rate": 9.603280116100336e-06,
      "loss": 0.007,
      "step": 67010
    },
    {
      "epoch": 0.4298454621544224,
      "grad_norm": 0.19030052423477173,
      "learning_rate": 9.603061593164928e-06,
      "loss": 0.0094,
      "step": 67020
    },
    {
      "epoch": 0.4299095990482085,
      "grad_norm": 0.316009521484375,
      "learning_rate": 9.602843012549867e-06,
      "loss": 0.0039,
      "step": 67030
    },
    {
      "epoch": 0.4299737359419946,
      "grad_norm": 0.05603104084730148,
      "learning_rate": 9.602624374257897e-06,
      "loss": 0.0031,
      "step": 67040
    },
    {
      "epoch": 0.4300378728357807,
      "grad_norm": 0.1649898737668991,
      "learning_rate": 9.602405678291751e-06,
      "loss": 0.003,
      "step": 67050
    },
    {
      "epoch": 0.4301020097295668,
      "grad_norm": 0.1882244497537613,
      "learning_rate": 9.602186924654172e-06,
      "loss": 0.0057,
      "step": 67060
    },
    {
      "epoch": 0.4301661466233529,
      "grad_norm": 0.2369825690984726,
      "learning_rate": 9.601968113347904e-06,
      "loss": 0.0065,
      "step": 67070
    },
    {
      "epoch": 0.430230283517139,
      "grad_norm": 0.24803709983825684,
      "learning_rate": 9.601749244375684e-06,
      "loss": 0.0038,
      "step": 67080
    },
    {
      "epoch": 0.4302944204109251,
      "grad_norm": 0.2131635546684265,
      "learning_rate": 9.601530317740258e-06,
      "loss": 0.0044,
      "step": 67090
    },
    {
      "epoch": 0.4303585573047112,
      "grad_norm": 0.05162657052278519,
      "learning_rate": 9.60131133344437e-06,
      "loss": 0.0046,
      "step": 67100
    },
    {
      "epoch": 0.43042269419849727,
      "grad_norm": 0.11941119283437729,
      "learning_rate": 9.601092291490761e-06,
      "loss": 0.0047,
      "step": 67110
    },
    {
      "epoch": 0.43048683109228336,
      "grad_norm": 0.3676694929599762,
      "learning_rate": 9.600873191882178e-06,
      "loss": 0.0037,
      "step": 67120
    },
    {
      "epoch": 0.43055096798606945,
      "grad_norm": 0.1545587182044983,
      "learning_rate": 9.600654034621366e-06,
      "loss": 0.0046,
      "step": 67130
    },
    {
      "epoch": 0.43061510487985555,
      "grad_norm": 0.21092180907726288,
      "learning_rate": 9.600434819711068e-06,
      "loss": 0.004,
      "step": 67140
    },
    {
      "epoch": 0.43067924177364164,
      "grad_norm": 0.24385568499565125,
      "learning_rate": 9.600215547154037e-06,
      "loss": 0.0041,
      "step": 67150
    },
    {
      "epoch": 0.43074337866742773,
      "grad_norm": 0.06852970272302628,
      "learning_rate": 9.599996216953017e-06,
      "loss": 0.0056,
      "step": 67160
    },
    {
      "epoch": 0.4308075155612138,
      "grad_norm": 0.14821739494800568,
      "learning_rate": 9.599776829110757e-06,
      "loss": 0.0034,
      "step": 67170
    },
    {
      "epoch": 0.430871652455,
      "grad_norm": 0.11472298949956894,
      "learning_rate": 9.599557383630005e-06,
      "loss": 0.0078,
      "step": 67180
    },
    {
      "epoch": 0.43093578934878607,
      "grad_norm": 0.08596605807542801,
      "learning_rate": 9.599337880513511e-06,
      "loss": 0.0043,
      "step": 67190
    },
    {
      "epoch": 0.43099992624257216,
      "grad_norm": 0.23955640196800232,
      "learning_rate": 9.599118319764028e-06,
      "loss": 0.0047,
      "step": 67200
    },
    {
      "epoch": 0.43106406313635826,
      "grad_norm": 0.1155315563082695,
      "learning_rate": 9.598898701384306e-06,
      "loss": 0.0032,
      "step": 67210
    },
    {
      "epoch": 0.43112820003014435,
      "grad_norm": 0.0761498361825943,
      "learning_rate": 9.598679025377095e-06,
      "loss": 0.0038,
      "step": 67220
    },
    {
      "epoch": 0.43119233692393044,
      "grad_norm": 0.41123166680336,
      "learning_rate": 9.59845929174515e-06,
      "loss": 0.0028,
      "step": 67230
    },
    {
      "epoch": 0.43125647381771653,
      "grad_norm": 0.05414069443941116,
      "learning_rate": 9.598239500491222e-06,
      "loss": 0.0039,
      "step": 67240
    },
    {
      "epoch": 0.43132061071150263,
      "grad_norm": 0.10585710406303406,
      "learning_rate": 9.598019651618068e-06,
      "loss": 0.0042,
      "step": 67250
    },
    {
      "epoch": 0.4313847476052887,
      "grad_norm": 0.09694528579711914,
      "learning_rate": 9.597799745128442e-06,
      "loss": 0.0043,
      "step": 67260
    },
    {
      "epoch": 0.4314488844990748,
      "grad_norm": 0.09088045358657837,
      "learning_rate": 9.597579781025098e-06,
      "loss": 0.003,
      "step": 67270
    },
    {
      "epoch": 0.4315130213928609,
      "grad_norm": 0.07393519580364227,
      "learning_rate": 9.597359759310793e-06,
      "loss": 0.0053,
      "step": 67280
    },
    {
      "epoch": 0.431577158286647,
      "grad_norm": 0.23588484525680542,
      "learning_rate": 9.597139679988287e-06,
      "loss": 0.0023,
      "step": 67290
    },
    {
      "epoch": 0.4316412951804331,
      "grad_norm": 0.13709813356399536,
      "learning_rate": 9.596919543060334e-06,
      "loss": 0.004,
      "step": 67300
    },
    {
      "epoch": 0.4317054320742192,
      "grad_norm": 0.27046436071395874,
      "learning_rate": 9.596699348529695e-06,
      "loss": 0.0054,
      "step": 67310
    },
    {
      "epoch": 0.43176956896800534,
      "grad_norm": 0.1580420732498169,
      "learning_rate": 9.596479096399125e-06,
      "loss": 0.0063,
      "step": 67320
    },
    {
      "epoch": 0.43183370586179143,
      "grad_norm": 0.4832228124141693,
      "learning_rate": 9.596258786671389e-06,
      "loss": 0.0039,
      "step": 67330
    },
    {
      "epoch": 0.4318978427555775,
      "grad_norm": 0.11211079359054565,
      "learning_rate": 9.596038419349246e-06,
      "loss": 0.0031,
      "step": 67340
    },
    {
      "epoch": 0.4319619796493636,
      "grad_norm": 0.22395701706409454,
      "learning_rate": 9.595817994435454e-06,
      "loss": 0.0065,
      "step": 67350
    },
    {
      "epoch": 0.4320261165431497,
      "grad_norm": 0.1439417004585266,
      "learning_rate": 9.595597511932782e-06,
      "loss": 0.003,
      "step": 67360
    },
    {
      "epoch": 0.4320902534369358,
      "grad_norm": 0.06668273359537125,
      "learning_rate": 9.595376971843986e-06,
      "loss": 0.0043,
      "step": 67370
    },
    {
      "epoch": 0.4321543903307219,
      "grad_norm": 0.11449944972991943,
      "learning_rate": 9.595156374171833e-06,
      "loss": 0.0031,
      "step": 67380
    },
    {
      "epoch": 0.432218527224508,
      "grad_norm": 0.13027827441692352,
      "learning_rate": 9.594935718919086e-06,
      "loss": 0.0026,
      "step": 67390
    },
    {
      "epoch": 0.4322826641182941,
      "grad_norm": 0.3017429709434509,
      "learning_rate": 9.594715006088511e-06,
      "loss": 0.0033,
      "step": 67400
    },
    {
      "epoch": 0.4323468010120802,
      "grad_norm": 0.15744541585445404,
      "learning_rate": 9.594494235682873e-06,
      "loss": 0.004,
      "step": 67410
    },
    {
      "epoch": 0.43241093790586627,
      "grad_norm": 0.162835031747818,
      "learning_rate": 9.594273407704938e-06,
      "loss": 0.0049,
      "step": 67420
    },
    {
      "epoch": 0.43247507479965236,
      "grad_norm": 0.1257157027721405,
      "learning_rate": 9.594052522157476e-06,
      "loss": 0.0026,
      "step": 67430
    },
    {
      "epoch": 0.43253921169343845,
      "grad_norm": 0.14831848442554474,
      "learning_rate": 9.59383157904325e-06,
      "loss": 0.0048,
      "step": 67440
    },
    {
      "epoch": 0.43260334858722455,
      "grad_norm": 0.1549537628889084,
      "learning_rate": 9.593610578365033e-06,
      "loss": 0.0039,
      "step": 67450
    },
    {
      "epoch": 0.4326674854810107,
      "grad_norm": 0.07446286827325821,
      "learning_rate": 9.593389520125591e-06,
      "loss": 0.0077,
      "step": 67460
    },
    {
      "epoch": 0.4327316223747968,
      "grad_norm": 0.33825135231018066,
      "learning_rate": 9.593168404327697e-06,
      "loss": 0.0031,
      "step": 67470
    },
    {
      "epoch": 0.4327957592685829,
      "grad_norm": 0.18623273074626923,
      "learning_rate": 9.592947230974119e-06,
      "loss": 0.0042,
      "step": 67480
    },
    {
      "epoch": 0.432859896162369,
      "grad_norm": 0.0629829540848732,
      "learning_rate": 9.592726000067629e-06,
      "loss": 0.0027,
      "step": 67490
    },
    {
      "epoch": 0.43292403305615507,
      "grad_norm": 0.07424663752317429,
      "learning_rate": 9.592504711611001e-06,
      "loss": 0.0043,
      "step": 67500
    },
    {
      "epoch": 0.43298816994994116,
      "grad_norm": 0.1887056678533554,
      "learning_rate": 9.592283365607008e-06,
      "loss": 0.0039,
      "step": 67510
    },
    {
      "epoch": 0.43305230684372725,
      "grad_norm": 0.11508423835039139,
      "learning_rate": 9.592061962058418e-06,
      "loss": 0.0032,
      "step": 67520
    },
    {
      "epoch": 0.43311644373751335,
      "grad_norm": 0.24281243979930878,
      "learning_rate": 9.591840500968014e-06,
      "loss": 0.0035,
      "step": 67530
    },
    {
      "epoch": 0.43318058063129944,
      "grad_norm": 0.11407119780778885,
      "learning_rate": 9.591618982338565e-06,
      "loss": 0.0015,
      "step": 67540
    },
    {
      "epoch": 0.43324471752508553,
      "grad_norm": 0.16879330575466156,
      "learning_rate": 9.591397406172848e-06,
      "loss": 0.0033,
      "step": 67550
    },
    {
      "epoch": 0.4333088544188716,
      "grad_norm": 0.23465494811534882,
      "learning_rate": 9.591175772473642e-06,
      "loss": 0.0034,
      "step": 67560
    },
    {
      "epoch": 0.4333729913126577,
      "grad_norm": 0.13142143189907074,
      "learning_rate": 9.590954081243722e-06,
      "loss": 0.0044,
      "step": 67570
    },
    {
      "epoch": 0.4334371282064438,
      "grad_norm": 0.15959471464157104,
      "learning_rate": 9.590732332485865e-06,
      "loss": 0.0036,
      "step": 67580
    },
    {
      "epoch": 0.4335012651002299,
      "grad_norm": 0.26229506731033325,
      "learning_rate": 9.590510526202852e-06,
      "loss": 0.0048,
      "step": 67590
    },
    {
      "epoch": 0.43356540199401605,
      "grad_norm": 0.39532706141471863,
      "learning_rate": 9.590288662397462e-06,
      "loss": 0.0033,
      "step": 67600
    },
    {
      "epoch": 0.43362953888780215,
      "grad_norm": 0.1433018445968628,
      "learning_rate": 9.590066741072472e-06,
      "loss": 0.0032,
      "step": 67610
    },
    {
      "epoch": 0.43369367578158824,
      "grad_norm": 0.771902859210968,
      "learning_rate": 9.589844762230666e-06,
      "loss": 0.0045,
      "step": 67620
    },
    {
      "epoch": 0.43375781267537433,
      "grad_norm": 0.3603050112724304,
      "learning_rate": 9.589622725874827e-06,
      "loss": 0.0036,
      "step": 67630
    },
    {
      "epoch": 0.4338219495691604,
      "grad_norm": 0.04795801639556885,
      "learning_rate": 9.589400632007733e-06,
      "loss": 0.0023,
      "step": 67640
    },
    {
      "epoch": 0.4338860864629465,
      "grad_norm": 0.2554285228252411,
      "learning_rate": 9.589178480632169e-06,
      "loss": 0.0032,
      "step": 67650
    },
    {
      "epoch": 0.4339502233567326,
      "grad_norm": 0.08371347934007645,
      "learning_rate": 9.58895627175092e-06,
      "loss": 0.0014,
      "step": 67660
    },
    {
      "epoch": 0.4340143602505187,
      "grad_norm": 0.3796055018901825,
      "learning_rate": 9.588734005366768e-06,
      "loss": 0.0032,
      "step": 67670
    },
    {
      "epoch": 0.4340784971443048,
      "grad_norm": 0.20644047856330872,
      "learning_rate": 9.588511681482499e-06,
      "loss": 0.0047,
      "step": 67680
    },
    {
      "epoch": 0.4341426340380909,
      "grad_norm": 0.23971392214298248,
      "learning_rate": 9.588289300100901e-06,
      "loss": 0.0063,
      "step": 67690
    },
    {
      "epoch": 0.434206770931877,
      "grad_norm": 0.15959949791431427,
      "learning_rate": 9.588066861224758e-06,
      "loss": 0.003,
      "step": 67700
    },
    {
      "epoch": 0.4342709078256631,
      "grad_norm": 0.3129645884037018,
      "learning_rate": 9.58784436485686e-06,
      "loss": 0.0075,
      "step": 67710
    },
    {
      "epoch": 0.43433504471944917,
      "grad_norm": 0.18099181354045868,
      "learning_rate": 9.58762181099999e-06,
      "loss": 0.0032,
      "step": 67720
    },
    {
      "epoch": 0.43439918161323526,
      "grad_norm": 0.3097049295902252,
      "learning_rate": 9.587399199656941e-06,
      "loss": 0.0118,
      "step": 67730
    },
    {
      "epoch": 0.4344633185070214,
      "grad_norm": 0.2578749656677246,
      "learning_rate": 9.587176530830503e-06,
      "loss": 0.0046,
      "step": 67740
    },
    {
      "epoch": 0.4345274554008075,
      "grad_norm": 0.19821570813655853,
      "learning_rate": 9.586953804523465e-06,
      "loss": 0.0029,
      "step": 67750
    },
    {
      "epoch": 0.4345915922945936,
      "grad_norm": 0.2031194269657135,
      "learning_rate": 9.586731020738615e-06,
      "loss": 0.0039,
      "step": 67760
    },
    {
      "epoch": 0.4346557291883797,
      "grad_norm": 0.16306443512439728,
      "learning_rate": 9.586508179478749e-06,
      "loss": 0.0029,
      "step": 67770
    },
    {
      "epoch": 0.4347198660821658,
      "grad_norm": 0.12638349831104279,
      "learning_rate": 9.586285280746657e-06,
      "loss": 0.003,
      "step": 67780
    },
    {
      "epoch": 0.4347840029759519,
      "grad_norm": 0.23570360243320465,
      "learning_rate": 9.586062324545131e-06,
      "loss": 0.0038,
      "step": 67790
    },
    {
      "epoch": 0.43484813986973797,
      "grad_norm": 0.17586968839168549,
      "learning_rate": 9.585839310876969e-06,
      "loss": 0.0027,
      "step": 67800
    },
    {
      "epoch": 0.43491227676352406,
      "grad_norm": 0.1429784595966339,
      "learning_rate": 9.585616239744963e-06,
      "loss": 0.0035,
      "step": 67810
    },
    {
      "epoch": 0.43497641365731016,
      "grad_norm": 0.17840762436389923,
      "learning_rate": 9.585393111151908e-06,
      "loss": 0.0034,
      "step": 67820
    },
    {
      "epoch": 0.43504055055109625,
      "grad_norm": 0.04029030352830887,
      "learning_rate": 9.585169925100599e-06,
      "loss": 0.0044,
      "step": 67830
    },
    {
      "epoch": 0.43510468744488234,
      "grad_norm": 0.2338280975818634,
      "learning_rate": 9.584946681593834e-06,
      "loss": 0.0038,
      "step": 67840
    },
    {
      "epoch": 0.43516882433866844,
      "grad_norm": 0.36239415407180786,
      "learning_rate": 9.58472338063441e-06,
      "loss": 0.0021,
      "step": 67850
    },
    {
      "epoch": 0.43523296123245453,
      "grad_norm": 0.2979362905025482,
      "learning_rate": 9.584500022225128e-06,
      "loss": 0.0057,
      "step": 67860
    },
    {
      "epoch": 0.4352970981262406,
      "grad_norm": 0.2265724241733551,
      "learning_rate": 9.584276606368781e-06,
      "loss": 0.0047,
      "step": 67870
    },
    {
      "epoch": 0.43536123502002677,
      "grad_norm": 0.09479842334985733,
      "learning_rate": 9.584053133068173e-06,
      "loss": 0.0028,
      "step": 67880
    },
    {
      "epoch": 0.43542537191381286,
      "grad_norm": 0.08840498328208923,
      "learning_rate": 9.583829602326104e-06,
      "loss": 0.0027,
      "step": 67890
    },
    {
      "epoch": 0.43548950880759896,
      "grad_norm": 0.10118495672941208,
      "learning_rate": 9.583606014145373e-06,
      "loss": 0.0054,
      "step": 67900
    },
    {
      "epoch": 0.43555364570138505,
      "grad_norm": 0.3275286555290222,
      "learning_rate": 9.583382368528781e-06,
      "loss": 0.0027,
      "step": 67910
    },
    {
      "epoch": 0.43561778259517114,
      "grad_norm": 0.10471635311841965,
      "learning_rate": 9.583158665479135e-06,
      "loss": 0.0033,
      "step": 67920
    },
    {
      "epoch": 0.43568191948895724,
      "grad_norm": 0.17559470236301422,
      "learning_rate": 9.582934904999234e-06,
      "loss": 0.0038,
      "step": 67930
    },
    {
      "epoch": 0.43574605638274333,
      "grad_norm": 0.09734024107456207,
      "learning_rate": 9.582711087091884e-06,
      "loss": 0.0036,
      "step": 67940
    },
    {
      "epoch": 0.4358101932765294,
      "grad_norm": 0.12025992572307587,
      "learning_rate": 9.58248721175989e-06,
      "loss": 0.0043,
      "step": 67950
    },
    {
      "epoch": 0.4358743301703155,
      "grad_norm": 0.38899630308151245,
      "learning_rate": 9.582263279006055e-06,
      "loss": 0.0033,
      "step": 67960
    },
    {
      "epoch": 0.4359384670641016,
      "grad_norm": 0.24911749362945557,
      "learning_rate": 9.582039288833187e-06,
      "loss": 0.0056,
      "step": 67970
    },
    {
      "epoch": 0.4360026039578877,
      "grad_norm": 0.14373457431793213,
      "learning_rate": 9.58181524124409e-06,
      "loss": 0.0041,
      "step": 67980
    },
    {
      "epoch": 0.4360667408516738,
      "grad_norm": 0.49666643142700195,
      "learning_rate": 9.581591136241575e-06,
      "loss": 0.0036,
      "step": 67990
    },
    {
      "epoch": 0.4361308777454599,
      "grad_norm": 0.2246396541595459,
      "learning_rate": 9.581366973828448e-06,
      "loss": 0.0023,
      "step": 68000
    },
    {
      "epoch": 0.436195014639246,
      "grad_norm": 0.23568378388881683,
      "learning_rate": 9.58114275400752e-06,
      "loss": 0.0033,
      "step": 68010
    },
    {
      "epoch": 0.43625915153303213,
      "grad_norm": 0.10890336334705353,
      "learning_rate": 9.580918476781598e-06,
      "loss": 0.0035,
      "step": 68020
    },
    {
      "epoch": 0.4363232884268182,
      "grad_norm": 0.045626670122146606,
      "learning_rate": 9.580694142153492e-06,
      "loss": 0.0036,
      "step": 68030
    },
    {
      "epoch": 0.4363874253206043,
      "grad_norm": 0.18727976083755493,
      "learning_rate": 9.580469750126018e-06,
      "loss": 0.0027,
      "step": 68040
    },
    {
      "epoch": 0.4364515622143904,
      "grad_norm": 0.23157556354999542,
      "learning_rate": 9.580245300701982e-06,
      "loss": 0.0058,
      "step": 68050
    },
    {
      "epoch": 0.4365156991081765,
      "grad_norm": 0.13583151996135712,
      "learning_rate": 9.5800207938842e-06,
      "loss": 0.0045,
      "step": 68060
    },
    {
      "epoch": 0.4365798360019626,
      "grad_norm": 0.14473675191402435,
      "learning_rate": 9.579796229675482e-06,
      "loss": 0.0028,
      "step": 68070
    },
    {
      "epoch": 0.4366439728957487,
      "grad_norm": 0.295076459646225,
      "learning_rate": 9.579571608078647e-06,
      "loss": 0.0045,
      "step": 68080
    },
    {
      "epoch": 0.4367081097895348,
      "grad_norm": 0.04209938272833824,
      "learning_rate": 9.579346929096505e-06,
      "loss": 0.0038,
      "step": 68090
    },
    {
      "epoch": 0.4367722466833209,
      "grad_norm": 0.22433242201805115,
      "learning_rate": 9.579122192731874e-06,
      "loss": 0.004,
      "step": 68100
    },
    {
      "epoch": 0.43683638357710697,
      "grad_norm": 0.08067035675048828,
      "learning_rate": 9.57889739898757e-06,
      "loss": 0.0042,
      "step": 68110
    },
    {
      "epoch": 0.43690052047089306,
      "grad_norm": 0.10899264365434647,
      "learning_rate": 9.578672547866408e-06,
      "loss": 0.0103,
      "step": 68120
    },
    {
      "epoch": 0.43696465736467915,
      "grad_norm": 0.4292784631252289,
      "learning_rate": 9.578447639371207e-06,
      "loss": 0.0044,
      "step": 68130
    },
    {
      "epoch": 0.43702879425846525,
      "grad_norm": 0.5548121333122253,
      "learning_rate": 9.578222673504785e-06,
      "loss": 0.0048,
      "step": 68140
    },
    {
      "epoch": 0.43709293115225134,
      "grad_norm": 0.23005898296833038,
      "learning_rate": 9.577997650269959e-06,
      "loss": 0.0054,
      "step": 68150
    },
    {
      "epoch": 0.4371570680460375,
      "grad_norm": 0.11167255789041519,
      "learning_rate": 9.577772569669552e-06,
      "loss": 0.0052,
      "step": 68160
    },
    {
      "epoch": 0.4372212049398236,
      "grad_norm": 0.056567851454019547,
      "learning_rate": 9.577547431706384e-06,
      "loss": 0.0053,
      "step": 68170
    },
    {
      "epoch": 0.4372853418336097,
      "grad_norm": 0.29500824213027954,
      "learning_rate": 9.577322236383276e-06,
      "loss": 0.0045,
      "step": 68180
    },
    {
      "epoch": 0.43734947872739577,
      "grad_norm": 0.14376701414585114,
      "learning_rate": 9.577096983703046e-06,
      "loss": 0.0047,
      "step": 68190
    },
    {
      "epoch": 0.43741361562118186,
      "grad_norm": 0.15609560906887054,
      "learning_rate": 9.57687167366852e-06,
      "loss": 0.002,
      "step": 68200
    },
    {
      "epoch": 0.43747775251496795,
      "grad_norm": 0.09901092201471329,
      "learning_rate": 9.576646306282523e-06,
      "loss": 0.0047,
      "step": 68210
    },
    {
      "epoch": 0.43754188940875405,
      "grad_norm": 0.24435746669769287,
      "learning_rate": 9.576420881547875e-06,
      "loss": 0.0045,
      "step": 68220
    },
    {
      "epoch": 0.43760602630254014,
      "grad_norm": 0.15269044041633606,
      "learning_rate": 9.576195399467404e-06,
      "loss": 0.0032,
      "step": 68230
    },
    {
      "epoch": 0.43767016319632623,
      "grad_norm": 0.15219302475452423,
      "learning_rate": 9.575969860043934e-06,
      "loss": 0.0058,
      "step": 68240
    },
    {
      "epoch": 0.4377343000901123,
      "grad_norm": 0.2758413255214691,
      "learning_rate": 9.57574426328029e-06,
      "loss": 0.0027,
      "step": 68250
    },
    {
      "epoch": 0.4377984369838984,
      "grad_norm": 0.35995030403137207,
      "learning_rate": 9.575518609179302e-06,
      "loss": 0.003,
      "step": 68260
    },
    {
      "epoch": 0.4378625738776845,
      "grad_norm": 0.13239847123622894,
      "learning_rate": 9.575292897743793e-06,
      "loss": 0.0035,
      "step": 68270
    },
    {
      "epoch": 0.4379267107714706,
      "grad_norm": 0.07212992012500763,
      "learning_rate": 9.575067128976596e-06,
      "loss": 0.0038,
      "step": 68280
    },
    {
      "epoch": 0.4379908476652567,
      "grad_norm": 0.2119167298078537,
      "learning_rate": 9.574841302880538e-06,
      "loss": 0.0032,
      "step": 68290
    },
    {
      "epoch": 0.43805498455904285,
      "grad_norm": 0.24278458952903748,
      "learning_rate": 9.574615419458448e-06,
      "loss": 0.0035,
      "step": 68300
    },
    {
      "epoch": 0.43811912145282894,
      "grad_norm": 0.17022500932216644,
      "learning_rate": 9.57438947871316e-06,
      "loss": 0.0048,
      "step": 68310
    },
    {
      "epoch": 0.43818325834661503,
      "grad_norm": 0.1462307572364807,
      "learning_rate": 9.5741634806475e-06,
      "loss": 0.0054,
      "step": 68320
    },
    {
      "epoch": 0.4382473952404011,
      "grad_norm": 0.16263353824615479,
      "learning_rate": 9.573937425264304e-06,
      "loss": 0.0029,
      "step": 68330
    },
    {
      "epoch": 0.4383115321341872,
      "grad_norm": 0.1209472119808197,
      "learning_rate": 9.5737113125664e-06,
      "loss": 0.0031,
      "step": 68340
    },
    {
      "epoch": 0.4383756690279733,
      "grad_norm": 0.14892970025539398,
      "learning_rate": 9.573485142556629e-06,
      "loss": 0.0025,
      "step": 68350
    },
    {
      "epoch": 0.4384398059217594,
      "grad_norm": 0.06195222958922386,
      "learning_rate": 9.573258915237818e-06,
      "loss": 0.0039,
      "step": 68360
    },
    {
      "epoch": 0.4385039428155455,
      "grad_norm": 0.35005781054496765,
      "learning_rate": 9.573032630612804e-06,
      "loss": 0.0041,
      "step": 68370
    },
    {
      "epoch": 0.4385680797093316,
      "grad_norm": 0.15759289264678955,
      "learning_rate": 9.572806288684425e-06,
      "loss": 0.0045,
      "step": 68380
    },
    {
      "epoch": 0.4386322166031177,
      "grad_norm": 0.07934365421533585,
      "learning_rate": 9.572579889455513e-06,
      "loss": 0.0047,
      "step": 68390
    },
    {
      "epoch": 0.4386963534969038,
      "grad_norm": 0.2578205168247223,
      "learning_rate": 9.572353432928907e-06,
      "loss": 0.0044,
      "step": 68400
    },
    {
      "epoch": 0.4387604903906899,
      "grad_norm": 0.1401975154876709,
      "learning_rate": 9.572126919107445e-06,
      "loss": 0.0056,
      "step": 68410
    },
    {
      "epoch": 0.43882462728447597,
      "grad_norm": 0.29037439823150635,
      "learning_rate": 9.571900347993965e-06,
      "loss": 0.0031,
      "step": 68420
    },
    {
      "epoch": 0.43888876417826206,
      "grad_norm": 0.22732864320278168,
      "learning_rate": 9.571673719591307e-06,
      "loss": 0.0032,
      "step": 68430
    },
    {
      "epoch": 0.4389529010720482,
      "grad_norm": 0.1276884377002716,
      "learning_rate": 9.571447033902309e-06,
      "loss": 0.0029,
      "step": 68440
    },
    {
      "epoch": 0.4390170379658343,
      "grad_norm": 0.21823787689208984,
      "learning_rate": 9.571220290929812e-06,
      "loss": 0.0037,
      "step": 68450
    },
    {
      "epoch": 0.4390811748596204,
      "grad_norm": 0.10427073389291763,
      "learning_rate": 9.570993490676658e-06,
      "loss": 0.0027,
      "step": 68460
    },
    {
      "epoch": 0.4391453117534065,
      "grad_norm": 0.4085180163383484,
      "learning_rate": 9.57076663314569e-06,
      "loss": 0.0047,
      "step": 68470
    },
    {
      "epoch": 0.4392094486471926,
      "grad_norm": 0.12851525843143463,
      "learning_rate": 9.57053971833975e-06,
      "loss": 0.0096,
      "step": 68480
    },
    {
      "epoch": 0.4392735855409787,
      "grad_norm": 0.15416646003723145,
      "learning_rate": 9.570312746261678e-06,
      "loss": 0.0046,
      "step": 68490
    },
    {
      "epoch": 0.43933772243476477,
      "grad_norm": 0.2808114290237427,
      "learning_rate": 9.570085716914323e-06,
      "loss": 0.0034,
      "step": 68500
    },
    {
      "epoch": 0.43940185932855086,
      "grad_norm": 0.13371069729328156,
      "learning_rate": 9.569858630300528e-06,
      "loss": 0.0041,
      "step": 68510
    },
    {
      "epoch": 0.43946599622233695,
      "grad_norm": 0.05387774109840393,
      "learning_rate": 9.569631486423138e-06,
      "loss": 0.0035,
      "step": 68520
    },
    {
      "epoch": 0.43953013311612305,
      "grad_norm": 0.24315620958805084,
      "learning_rate": 9.569404285284999e-06,
      "loss": 0.0068,
      "step": 68530
    },
    {
      "epoch": 0.43959427000990914,
      "grad_norm": 0.044655054807662964,
      "learning_rate": 9.569177026888958e-06,
      "loss": 0.0033,
      "step": 68540
    },
    {
      "epoch": 0.43965840690369523,
      "grad_norm": 0.04376442730426788,
      "learning_rate": 9.568949711237865e-06,
      "loss": 0.0031,
      "step": 68550
    },
    {
      "epoch": 0.4397225437974813,
      "grad_norm": 0.24665457010269165,
      "learning_rate": 9.568722338334567e-06,
      "loss": 0.0053,
      "step": 68560
    },
    {
      "epoch": 0.4397866806912674,
      "grad_norm": 0.08923192322254181,
      "learning_rate": 9.568494908181911e-06,
      "loss": 0.0065,
      "step": 68570
    },
    {
      "epoch": 0.4398508175850535,
      "grad_norm": 0.20345203578472137,
      "learning_rate": 9.568267420782749e-06,
      "loss": 0.0035,
      "step": 68580
    },
    {
      "epoch": 0.43991495447883966,
      "grad_norm": 0.22325508296489716,
      "learning_rate": 9.568039876139932e-06,
      "loss": 0.0041,
      "step": 68590
    },
    {
      "epoch": 0.43997909137262575,
      "grad_norm": 0.3755197823047638,
      "learning_rate": 9.56781227425631e-06,
      "loss": 0.0032,
      "step": 68600
    },
    {
      "epoch": 0.44004322826641185,
      "grad_norm": 0.23552502691745758,
      "learning_rate": 9.567584615134738e-06,
      "loss": 0.0051,
      "step": 68610
    },
    {
      "epoch": 0.44010736516019794,
      "grad_norm": 0.30500590801239014,
      "learning_rate": 9.567356898778064e-06,
      "loss": 0.0061,
      "step": 68620
    },
    {
      "epoch": 0.44017150205398403,
      "grad_norm": 0.28967979550361633,
      "learning_rate": 9.567129125189143e-06,
      "loss": 0.0048,
      "step": 68630
    },
    {
      "epoch": 0.4402356389477701,
      "grad_norm": 0.13767173886299133,
      "learning_rate": 9.566901294370832e-06,
      "loss": 0.004,
      "step": 68640
    },
    {
      "epoch": 0.4402997758415562,
      "grad_norm": 0.1392887532711029,
      "learning_rate": 9.566673406325983e-06,
      "loss": 0.002,
      "step": 68650
    },
    {
      "epoch": 0.4403639127353423,
      "grad_norm": 0.12102367728948593,
      "learning_rate": 9.566445461057452e-06,
      "loss": 0.004,
      "step": 68660
    },
    {
      "epoch": 0.4404280496291284,
      "grad_norm": 0.17383573949337006,
      "learning_rate": 9.566217458568096e-06,
      "loss": 0.0036,
      "step": 68670
    },
    {
      "epoch": 0.4404921865229145,
      "grad_norm": 0.103236123919487,
      "learning_rate": 9.565989398860774e-06,
      "loss": 0.0042,
      "step": 68680
    },
    {
      "epoch": 0.4405563234167006,
      "grad_norm": 0.06880658864974976,
      "learning_rate": 9.56576128193834e-06,
      "loss": 0.0035,
      "step": 68690
    },
    {
      "epoch": 0.4406204603104867,
      "grad_norm": 0.11575897783041,
      "learning_rate": 9.565533107803652e-06,
      "loss": 0.0029,
      "step": 68700
    },
    {
      "epoch": 0.4406845972042728,
      "grad_norm": 0.06390535086393356,
      "learning_rate": 9.565304876459574e-06,
      "loss": 0.0057,
      "step": 68710
    },
    {
      "epoch": 0.44074873409805887,
      "grad_norm": 0.22714078426361084,
      "learning_rate": 9.565076587908962e-06,
      "loss": 0.0037,
      "step": 68720
    },
    {
      "epoch": 0.440812870991845,
      "grad_norm": 0.09389592707157135,
      "learning_rate": 9.564848242154678e-06,
      "loss": 0.0097,
      "step": 68730
    },
    {
      "epoch": 0.4408770078856311,
      "grad_norm": 0.12381519377231598,
      "learning_rate": 9.564619839199583e-06,
      "loss": 0.007,
      "step": 68740
    },
    {
      "epoch": 0.4409411447794172,
      "grad_norm": 0.14362052083015442,
      "learning_rate": 9.564391379046539e-06,
      "loss": 0.0037,
      "step": 68750
    },
    {
      "epoch": 0.4410052816732033,
      "grad_norm": 0.09533475339412689,
      "learning_rate": 9.56416286169841e-06,
      "loss": 0.0044,
      "step": 68760
    },
    {
      "epoch": 0.4410694185669894,
      "grad_norm": 0.06411031633615494,
      "learning_rate": 9.563934287158057e-06,
      "loss": 0.0021,
      "step": 68770
    },
    {
      "epoch": 0.4411335554607755,
      "grad_norm": 0.5412489175796509,
      "learning_rate": 9.563705655428347e-06,
      "loss": 0.0049,
      "step": 68780
    },
    {
      "epoch": 0.4411976923545616,
      "grad_norm": 0.2374279946088791,
      "learning_rate": 9.56347696651214e-06,
      "loss": 0.0025,
      "step": 68790
    },
    {
      "epoch": 0.44126182924834767,
      "grad_norm": 0.31077682971954346,
      "learning_rate": 9.56324822041231e-06,
      "loss": 0.0067,
      "step": 68800
    },
    {
      "epoch": 0.44132596614213376,
      "grad_norm": 0.07084587961435318,
      "learning_rate": 9.563019417131716e-06,
      "loss": 0.0056,
      "step": 68810
    },
    {
      "epoch": 0.44139010303591986,
      "grad_norm": 0.097480908036232,
      "learning_rate": 9.562790556673228e-06,
      "loss": 0.0034,
      "step": 68820
    },
    {
      "epoch": 0.44145423992970595,
      "grad_norm": 0.02128240466117859,
      "learning_rate": 9.562561639039711e-06,
      "loss": 0.0018,
      "step": 68830
    },
    {
      "epoch": 0.44151837682349204,
      "grad_norm": 0.14919476211071014,
      "learning_rate": 9.56233266423404e-06,
      "loss": 0.0021,
      "step": 68840
    },
    {
      "epoch": 0.44158251371727814,
      "grad_norm": 0.2957935631275177,
      "learning_rate": 9.562103632259076e-06,
      "loss": 0.004,
      "step": 68850
    },
    {
      "epoch": 0.44164665061106423,
      "grad_norm": 0.14767687022686005,
      "learning_rate": 9.561874543117695e-06,
      "loss": 0.0036,
      "step": 68860
    },
    {
      "epoch": 0.4417107875048504,
      "grad_norm": 0.007925967685878277,
      "learning_rate": 9.561645396812767e-06,
      "loss": 0.0073,
      "step": 68870
    },
    {
      "epoch": 0.44177492439863647,
      "grad_norm": 0.2985318899154663,
      "learning_rate": 9.56141619334716e-06,
      "loss": 0.0029,
      "step": 68880
    },
    {
      "epoch": 0.44183906129242256,
      "grad_norm": 0.247298926115036,
      "learning_rate": 9.561186932723748e-06,
      "loss": 0.0031,
      "step": 68890
    },
    {
      "epoch": 0.44190319818620866,
      "grad_norm": 0.40475961565971375,
      "learning_rate": 9.560957614945406e-06,
      "loss": 0.0047,
      "step": 68900
    },
    {
      "epoch": 0.44196733507999475,
      "grad_norm": 0.100443035364151,
      "learning_rate": 9.560728240015003e-06,
      "loss": 0.004,
      "step": 68910
    },
    {
      "epoch": 0.44203147197378084,
      "grad_norm": 0.18375732004642487,
      "learning_rate": 9.560498807935416e-06,
      "loss": 0.0049,
      "step": 68920
    },
    {
      "epoch": 0.44209560886756694,
      "grad_norm": 0.06509524583816528,
      "learning_rate": 9.56026931870952e-06,
      "loss": 0.0035,
      "step": 68930
    },
    {
      "epoch": 0.44215974576135303,
      "grad_norm": 0.10219797492027283,
      "learning_rate": 9.560039772340191e-06,
      "loss": 0.0039,
      "step": 68940
    },
    {
      "epoch": 0.4422238826551391,
      "grad_norm": 0.2416619211435318,
      "learning_rate": 9.559810168830304e-06,
      "loss": 0.0036,
      "step": 68950
    },
    {
      "epoch": 0.4422880195489252,
      "grad_norm": 0.14863581955432892,
      "learning_rate": 9.559580508182737e-06,
      "loss": 0.005,
      "step": 68960
    },
    {
      "epoch": 0.4423521564427113,
      "grad_norm": 0.23987102508544922,
      "learning_rate": 9.559350790400369e-06,
      "loss": 0.0039,
      "step": 68970
    },
    {
      "epoch": 0.4424162933364974,
      "grad_norm": 0.042374927550554276,
      "learning_rate": 9.559121015486075e-06,
      "loss": 0.0036,
      "step": 68980
    },
    {
      "epoch": 0.4424804302302835,
      "grad_norm": 0.11949405819177628,
      "learning_rate": 9.558891183442736e-06,
      "loss": 0.0044,
      "step": 68990
    },
    {
      "epoch": 0.4425445671240696,
      "grad_norm": 0.2754271328449249,
      "learning_rate": 9.558661294273234e-06,
      "loss": 0.0047,
      "step": 69000
    },
    {
      "epoch": 0.44260870401785574,
      "grad_norm": 0.19826200604438782,
      "learning_rate": 9.558431347980447e-06,
      "loss": 0.0044,
      "step": 69010
    },
    {
      "epoch": 0.44267284091164183,
      "grad_norm": 0.5750806927680969,
      "learning_rate": 9.558201344567257e-06,
      "loss": 0.0052,
      "step": 69020
    },
    {
      "epoch": 0.4427369778054279,
      "grad_norm": 0.1713956892490387,
      "learning_rate": 9.557971284036547e-06,
      "loss": 0.0022,
      "step": 69030
    },
    {
      "epoch": 0.442801114699214,
      "grad_norm": 0.22112691402435303,
      "learning_rate": 9.557741166391198e-06,
      "loss": 0.0045,
      "step": 69040
    },
    {
      "epoch": 0.4428652515930001,
      "grad_norm": 0.2011345624923706,
      "learning_rate": 9.557510991634097e-06,
      "loss": 0.0046,
      "step": 69050
    },
    {
      "epoch": 0.4429293884867862,
      "grad_norm": 0.1980188637971878,
      "learning_rate": 9.557280759768126e-06,
      "loss": 0.0064,
      "step": 69060
    },
    {
      "epoch": 0.4429935253805723,
      "grad_norm": 0.1128661185503006,
      "learning_rate": 9.557050470796169e-06,
      "loss": 0.0054,
      "step": 69070
    },
    {
      "epoch": 0.4430576622743584,
      "grad_norm": 0.134240061044693,
      "learning_rate": 9.556820124721113e-06,
      "loss": 0.0051,
      "step": 69080
    },
    {
      "epoch": 0.4431217991681445,
      "grad_norm": 0.047759849578142166,
      "learning_rate": 9.556589721545844e-06,
      "loss": 0.0046,
      "step": 69090
    },
    {
      "epoch": 0.4431859360619306,
      "grad_norm": 0.15535880625247955,
      "learning_rate": 9.556359261273249e-06,
      "loss": 0.0029,
      "step": 69100
    },
    {
      "epoch": 0.44325007295571667,
      "grad_norm": 0.15249252319335938,
      "learning_rate": 9.556128743906216e-06,
      "loss": 0.0022,
      "step": 69110
    },
    {
      "epoch": 0.44331420984950276,
      "grad_norm": 0.3138236999511719,
      "learning_rate": 9.555898169447636e-06,
      "loss": 0.0095,
      "step": 69120
    },
    {
      "epoch": 0.44337834674328885,
      "grad_norm": 0.0569743849337101,
      "learning_rate": 9.555667537900393e-06,
      "loss": 0.0034,
      "step": 69130
    },
    {
      "epoch": 0.44344248363707495,
      "grad_norm": 0.03304928541183472,
      "learning_rate": 9.55543684926738e-06,
      "loss": 0.0022,
      "step": 69140
    },
    {
      "epoch": 0.4435066205308611,
      "grad_norm": 0.16622063517570496,
      "learning_rate": 9.555206103551488e-06,
      "loss": 0.0052,
      "step": 69150
    },
    {
      "epoch": 0.4435707574246472,
      "grad_norm": 0.1225697472691536,
      "learning_rate": 9.554975300755608e-06,
      "loss": 0.0037,
      "step": 69160
    },
    {
      "epoch": 0.4436348943184333,
      "grad_norm": 0.1855577975511551,
      "learning_rate": 9.554744440882633e-06,
      "loss": 0.0063,
      "step": 69170
    },
    {
      "epoch": 0.4436990312122194,
      "grad_norm": 0.46637067198753357,
      "learning_rate": 9.554513523935454e-06,
      "loss": 0.0054,
      "step": 69180
    },
    {
      "epoch": 0.44376316810600547,
      "grad_norm": 0.11185431480407715,
      "learning_rate": 9.554282549916966e-06,
      "loss": 0.0032,
      "step": 69190
    },
    {
      "epoch": 0.44382730499979156,
      "grad_norm": 0.14200040698051453,
      "learning_rate": 9.554051518830062e-06,
      "loss": 0.0032,
      "step": 69200
    },
    {
      "epoch": 0.44389144189357765,
      "grad_norm": 0.2912452518939972,
      "learning_rate": 9.553820430677639e-06,
      "loss": 0.0038,
      "step": 69210
    },
    {
      "epoch": 0.44395557878736375,
      "grad_norm": 0.24687382578849792,
      "learning_rate": 9.55358928546259e-06,
      "loss": 0.0026,
      "step": 69220
    },
    {
      "epoch": 0.44401971568114984,
      "grad_norm": 0.13388454914093018,
      "learning_rate": 9.553358083187813e-06,
      "loss": 0.003,
      "step": 69230
    },
    {
      "epoch": 0.44408385257493593,
      "grad_norm": 0.12316068261861801,
      "learning_rate": 9.553126823856204e-06,
      "loss": 0.0028,
      "step": 69240
    },
    {
      "epoch": 0.444147989468722,
      "grad_norm": 0.15901079773902893,
      "learning_rate": 9.552895507470665e-06,
      "loss": 0.0033,
      "step": 69250
    },
    {
      "epoch": 0.4442121263625081,
      "grad_norm": 0.32043325901031494,
      "learning_rate": 9.55266413403409e-06,
      "loss": 0.0033,
      "step": 69260
    },
    {
      "epoch": 0.4442762632562942,
      "grad_norm": 0.4112547039985657,
      "learning_rate": 9.552432703549379e-06,
      "loss": 0.0033,
      "step": 69270
    },
    {
      "epoch": 0.4443404001500803,
      "grad_norm": 0.12032436579465866,
      "learning_rate": 9.552201216019432e-06,
      "loss": 0.0046,
      "step": 69280
    },
    {
      "epoch": 0.44440453704386645,
      "grad_norm": 0.03089107573032379,
      "learning_rate": 9.551969671447152e-06,
      "loss": 0.0015,
      "step": 69290
    },
    {
      "epoch": 0.44446867393765255,
      "grad_norm": 0.23203957080841064,
      "learning_rate": 9.551738069835438e-06,
      "loss": 0.004,
      "step": 69300
    },
    {
      "epoch": 0.44453281083143864,
      "grad_norm": 0.15184760093688965,
      "learning_rate": 9.551506411187194e-06,
      "loss": 0.0041,
      "step": 69310
    },
    {
      "epoch": 0.44459694772522473,
      "grad_norm": 0.07851779460906982,
      "learning_rate": 9.551274695505321e-06,
      "loss": 0.0022,
      "step": 69320
    },
    {
      "epoch": 0.4446610846190108,
      "grad_norm": 0.2914571762084961,
      "learning_rate": 9.551042922792721e-06,
      "loss": 0.0041,
      "step": 69330
    },
    {
      "epoch": 0.4447252215127969,
      "grad_norm": 0.10128697007894516,
      "learning_rate": 9.550811093052304e-06,
      "loss": 0.0056,
      "step": 69340
    },
    {
      "epoch": 0.444789358406583,
      "grad_norm": 0.1086357980966568,
      "learning_rate": 9.55057920628697e-06,
      "loss": 0.0026,
      "step": 69350
    },
    {
      "epoch": 0.4448534953003691,
      "grad_norm": 0.20596520602703094,
      "learning_rate": 9.550347262499626e-06,
      "loss": 0.0042,
      "step": 69360
    },
    {
      "epoch": 0.4449176321941552,
      "grad_norm": 0.15991929173469543,
      "learning_rate": 9.55011526169318e-06,
      "loss": 0.0044,
      "step": 69370
    },
    {
      "epoch": 0.4449817690879413,
      "grad_norm": 0.463979035615921,
      "learning_rate": 9.549883203870537e-06,
      "loss": 0.0044,
      "step": 69380
    },
    {
      "epoch": 0.4450459059817274,
      "grad_norm": 0.056873906403779984,
      "learning_rate": 9.549651089034607e-06,
      "loss": 0.0042,
      "step": 69390
    },
    {
      "epoch": 0.4451100428755135,
      "grad_norm": 0.07245271652936935,
      "learning_rate": 9.549418917188295e-06,
      "loss": 0.0094,
      "step": 69400
    },
    {
      "epoch": 0.44517417976929957,
      "grad_norm": 0.1018892154097557,
      "learning_rate": 9.549186688334514e-06,
      "loss": 0.0028,
      "step": 69410
    },
    {
      "epoch": 0.44523831666308566,
      "grad_norm": 0.14666366577148438,
      "learning_rate": 9.548954402476172e-06,
      "loss": 0.0039,
      "step": 69420
    },
    {
      "epoch": 0.4453024535568718,
      "grad_norm": 0.17333391308784485,
      "learning_rate": 9.54872205961618e-06,
      "loss": 0.004,
      "step": 69430
    },
    {
      "epoch": 0.4453665904506579,
      "grad_norm": 0.2100229263305664,
      "learning_rate": 9.54848965975745e-06,
      "loss": 0.0035,
      "step": 69440
    },
    {
      "epoch": 0.445430727344444,
      "grad_norm": 0.34473252296447754,
      "learning_rate": 9.548257202902896e-06,
      "loss": 0.0056,
      "step": 69450
    },
    {
      "epoch": 0.4454948642382301,
      "grad_norm": 0.15477558970451355,
      "learning_rate": 9.548024689055426e-06,
      "loss": 0.0046,
      "step": 69460
    },
    {
      "epoch": 0.4455590011320162,
      "grad_norm": 0.17489725351333618,
      "learning_rate": 9.547792118217956e-06,
      "loss": 0.003,
      "step": 69470
    },
    {
      "epoch": 0.4456231380258023,
      "grad_norm": 0.1335802674293518,
      "learning_rate": 9.5475594903934e-06,
      "loss": 0.003,
      "step": 69480
    },
    {
      "epoch": 0.44568727491958837,
      "grad_norm": 0.22918473184108734,
      "learning_rate": 9.547326805584676e-06,
      "loss": 0.0032,
      "step": 69490
    },
    {
      "epoch": 0.44575141181337447,
      "grad_norm": 0.14939549565315247,
      "learning_rate": 9.547094063794697e-06,
      "loss": 0.0052,
      "step": 69500
    },
    {
      "epoch": 0.44581554870716056,
      "grad_norm": 0.051198314875364304,
      "learning_rate": 9.546861265026379e-06,
      "loss": 0.003,
      "step": 69510
    },
    {
      "epoch": 0.44587968560094665,
      "grad_norm": 0.11114027351140976,
      "learning_rate": 9.546628409282638e-06,
      "loss": 0.0031,
      "step": 69520
    },
    {
      "epoch": 0.44594382249473274,
      "grad_norm": 0.4134508967399597,
      "learning_rate": 9.546395496566394e-06,
      "loss": 0.0032,
      "step": 69530
    },
    {
      "epoch": 0.44600795938851884,
      "grad_norm": 0.34015557169914246,
      "learning_rate": 9.546162526880566e-06,
      "loss": 0.003,
      "step": 69540
    },
    {
      "epoch": 0.44607209628230493,
      "grad_norm": 0.1437218338251114,
      "learning_rate": 9.545929500228074e-06,
      "loss": 0.0045,
      "step": 69550
    },
    {
      "epoch": 0.446136233176091,
      "grad_norm": 0.26935333013534546,
      "learning_rate": 9.545696416611835e-06,
      "loss": 0.0043,
      "step": 69560
    },
    {
      "epoch": 0.4462003700698772,
      "grad_norm": 0.2559076249599457,
      "learning_rate": 9.545463276034772e-06,
      "loss": 0.0037,
      "step": 69570
    },
    {
      "epoch": 0.44626450696366327,
      "grad_norm": 0.2649496793746948,
      "learning_rate": 9.545230078499803e-06,
      "loss": 0.0034,
      "step": 69580
    },
    {
      "epoch": 0.44632864385744936,
      "grad_norm": 0.16384756565093994,
      "learning_rate": 9.544996824009855e-06,
      "loss": 0.0033,
      "step": 69590
    },
    {
      "epoch": 0.44639278075123545,
      "grad_norm": 0.23684534430503845,
      "learning_rate": 9.544763512567849e-06,
      "loss": 0.0048,
      "step": 69600
    },
    {
      "epoch": 0.44645691764502154,
      "grad_norm": 0.07358350604772568,
      "learning_rate": 9.544530144176707e-06,
      "loss": 0.0026,
      "step": 69610
    },
    {
      "epoch": 0.44652105453880764,
      "grad_norm": 0.2703312635421753,
      "learning_rate": 9.544296718839354e-06,
      "loss": 0.0047,
      "step": 69620
    },
    {
      "epoch": 0.44658519143259373,
      "grad_norm": 0.1530401110649109,
      "learning_rate": 9.544063236558715e-06,
      "loss": 0.0042,
      "step": 69630
    },
    {
      "epoch": 0.4466493283263798,
      "grad_norm": 0.238456130027771,
      "learning_rate": 9.543829697337717e-06,
      "loss": 0.0028,
      "step": 69640
    },
    {
      "epoch": 0.4467134652201659,
      "grad_norm": 0.343877911567688,
      "learning_rate": 9.543596101179285e-06,
      "loss": 0.005,
      "step": 69650
    },
    {
      "epoch": 0.446777602113952,
      "grad_norm": 0.08465104550123215,
      "learning_rate": 9.543362448086347e-06,
      "loss": 0.0043,
      "step": 69660
    },
    {
      "epoch": 0.4468417390077381,
      "grad_norm": 0.08829823136329651,
      "learning_rate": 9.543128738061832e-06,
      "loss": 0.0019,
      "step": 69670
    },
    {
      "epoch": 0.4469058759015242,
      "grad_norm": 0.04483136162161827,
      "learning_rate": 9.542894971108664e-06,
      "loss": 0.0044,
      "step": 69680
    },
    {
      "epoch": 0.4469700127953103,
      "grad_norm": 0.4056674838066101,
      "learning_rate": 9.542661147229777e-06,
      "loss": 0.0048,
      "step": 69690
    },
    {
      "epoch": 0.4470341496890964,
      "grad_norm": 0.1599629819393158,
      "learning_rate": 9.542427266428099e-06,
      "loss": 0.0032,
      "step": 69700
    },
    {
      "epoch": 0.44709828658288253,
      "grad_norm": 0.10350514948368073,
      "learning_rate": 9.54219332870656e-06,
      "loss": 0.0048,
      "step": 69710
    },
    {
      "epoch": 0.4471624234766686,
      "grad_norm": 0.15455134212970734,
      "learning_rate": 9.541959334068094e-06,
      "loss": 0.0038,
      "step": 69720
    },
    {
      "epoch": 0.4472265603704547,
      "grad_norm": 0.26008278131484985,
      "learning_rate": 9.541725282515631e-06,
      "loss": 0.0041,
      "step": 69730
    },
    {
      "epoch": 0.4472906972642408,
      "grad_norm": 0.24542184174060822,
      "learning_rate": 9.541491174052101e-06,
      "loss": 0.0043,
      "step": 69740
    },
    {
      "epoch": 0.4473548341580269,
      "grad_norm": 0.07911483943462372,
      "learning_rate": 9.541257008680445e-06,
      "loss": 0.002,
      "step": 69750
    },
    {
      "epoch": 0.447418971051813,
      "grad_norm": 0.2870321571826935,
      "learning_rate": 9.541022786403592e-06,
      "loss": 0.0036,
      "step": 69760
    },
    {
      "epoch": 0.4474831079455991,
      "grad_norm": 0.3157743811607361,
      "learning_rate": 9.540788507224478e-06,
      "loss": 0.0051,
      "step": 69770
    },
    {
      "epoch": 0.4475472448393852,
      "grad_norm": 0.0588856041431427,
      "learning_rate": 9.54055417114604e-06,
      "loss": 0.0037,
      "step": 69780
    },
    {
      "epoch": 0.4476113817331713,
      "grad_norm": 0.29160788655281067,
      "learning_rate": 9.54031977817121e-06,
      "loss": 0.0018,
      "step": 69790
    },
    {
      "epoch": 0.44767551862695737,
      "grad_norm": 0.5045191049575806,
      "learning_rate": 9.54008532830293e-06,
      "loss": 0.0055,
      "step": 69800
    },
    {
      "epoch": 0.44773965552074346,
      "grad_norm": 0.11215846240520477,
      "learning_rate": 9.539850821544137e-06,
      "loss": 0.0051,
      "step": 69810
    },
    {
      "epoch": 0.44780379241452956,
      "grad_norm": 0.24262496829032898,
      "learning_rate": 9.539616257897766e-06,
      "loss": 0.0059,
      "step": 69820
    },
    {
      "epoch": 0.44786792930831565,
      "grad_norm": 0.08639699220657349,
      "learning_rate": 9.539381637366762e-06,
      "loss": 0.0205,
      "step": 69830
    },
    {
      "epoch": 0.44793206620210174,
      "grad_norm": 0.3160640299320221,
      "learning_rate": 9.53914695995406e-06,
      "loss": 0.004,
      "step": 69840
    },
    {
      "epoch": 0.4479962030958879,
      "grad_norm": 0.1780281960964203,
      "learning_rate": 9.538912225662602e-06,
      "loss": 0.0038,
      "step": 69850
    },
    {
      "epoch": 0.448060339989674,
      "grad_norm": 0.07597067207098007,
      "learning_rate": 9.538677434495331e-06,
      "loss": 0.0065,
      "step": 69860
    },
    {
      "epoch": 0.4481244768834601,
      "grad_norm": 0.24969497323036194,
      "learning_rate": 9.538442586455187e-06,
      "loss": 0.0048,
      "step": 69870
    },
    {
      "epoch": 0.44818861377724617,
      "grad_norm": 0.5276156663894653,
      "learning_rate": 9.538207681545115e-06,
      "loss": 0.0044,
      "step": 69880
    },
    {
      "epoch": 0.44825275067103226,
      "grad_norm": 0.05741346254944801,
      "learning_rate": 9.537972719768059e-06,
      "loss": 0.0047,
      "step": 69890
    },
    {
      "epoch": 0.44831688756481836,
      "grad_norm": 0.1554454267024994,
      "learning_rate": 9.53773770112696e-06,
      "loss": 0.0021,
      "step": 69900
    },
    {
      "epoch": 0.44838102445860445,
      "grad_norm": 0.27648258209228516,
      "learning_rate": 9.53750262562476e-06,
      "loss": 0.0044,
      "step": 69910
    },
    {
      "epoch": 0.44844516135239054,
      "grad_norm": 0.08419109880924225,
      "learning_rate": 9.537267493264415e-06,
      "loss": 0.0057,
      "step": 69920
    },
    {
      "epoch": 0.44850929824617664,
      "grad_norm": 0.13251163065433502,
      "learning_rate": 9.537032304048864e-06,
      "loss": 0.0043,
      "step": 69930
    },
    {
      "epoch": 0.44857343513996273,
      "grad_norm": 0.12250209599733353,
      "learning_rate": 9.536797057981055e-06,
      "loss": 0.0025,
      "step": 69940
    },
    {
      "epoch": 0.4486375720337488,
      "grad_norm": 0.37543269991874695,
      "learning_rate": 9.536561755063937e-06,
      "loss": 0.0029,
      "step": 69950
    },
    {
      "epoch": 0.4487017089275349,
      "grad_norm": 0.14071187376976013,
      "learning_rate": 9.53632639530046e-06,
      "loss": 0.0033,
      "step": 69960
    },
    {
      "epoch": 0.448765845821321,
      "grad_norm": 0.1607028692960739,
      "learning_rate": 9.536090978693568e-06,
      "loss": 0.0021,
      "step": 69970
    },
    {
      "epoch": 0.4488299827151071,
      "grad_norm": 0.19182954728603363,
      "learning_rate": 9.535855505246215e-06,
      "loss": 0.0054,
      "step": 69980
    },
    {
      "epoch": 0.4488941196088932,
      "grad_norm": 0.11695774644613266,
      "learning_rate": 9.535619974961352e-06,
      "loss": 0.0065,
      "step": 69990
    },
    {
      "epoch": 0.44895825650267934,
      "grad_norm": 0.1232537105679512,
      "learning_rate": 9.535384387841927e-06,
      "loss": 0.0036,
      "step": 70000
    },
    {
      "epoch": 0.44902239339646544,
      "grad_norm": 0.08189796656370163,
      "learning_rate": 9.535148743890896e-06,
      "loss": 0.0032,
      "step": 70010
    },
    {
      "epoch": 0.44908653029025153,
      "grad_norm": 0.20595207810401917,
      "learning_rate": 9.534913043111209e-06,
      "loss": 0.0029,
      "step": 70020
    },
    {
      "epoch": 0.4491506671840376,
      "grad_norm": 0.31134435534477234,
      "learning_rate": 9.534677285505822e-06,
      "loss": 0.0034,
      "step": 70030
    },
    {
      "epoch": 0.4492148040778237,
      "grad_norm": 0.26455387473106384,
      "learning_rate": 9.534441471077687e-06,
      "loss": 0.0027,
      "step": 70040
    },
    {
      "epoch": 0.4492789409716098,
      "grad_norm": 0.34831151366233826,
      "learning_rate": 9.534205599829758e-06,
      "loss": 0.0037,
      "step": 70050
    },
    {
      "epoch": 0.4493430778653959,
      "grad_norm": 0.11280174553394318,
      "learning_rate": 9.533969671764994e-06,
      "loss": 0.0022,
      "step": 70060
    },
    {
      "epoch": 0.449407214759182,
      "grad_norm": 0.4784580171108246,
      "learning_rate": 9.533733686886349e-06,
      "loss": 0.0055,
      "step": 70070
    },
    {
      "epoch": 0.4494713516529681,
      "grad_norm": 0.33552631735801697,
      "learning_rate": 9.533497645196781e-06,
      "loss": 0.0047,
      "step": 70080
    },
    {
      "epoch": 0.4495354885467542,
      "grad_norm": 0.03899163007736206,
      "learning_rate": 9.533261546699248e-06,
      "loss": 0.0037,
      "step": 70090
    },
    {
      "epoch": 0.4495996254405403,
      "grad_norm": 0.1701935976743698,
      "learning_rate": 9.533025391396708e-06,
      "loss": 0.0048,
      "step": 70100
    },
    {
      "epoch": 0.44966376233432637,
      "grad_norm": 0.4440838694572449,
      "learning_rate": 9.532789179292119e-06,
      "loss": 0.0028,
      "step": 70110
    },
    {
      "epoch": 0.44972789922811246,
      "grad_norm": 0.34770143032073975,
      "learning_rate": 9.532552910388441e-06,
      "loss": 0.0058,
      "step": 70120
    },
    {
      "epoch": 0.44979203612189855,
      "grad_norm": 0.289668470621109,
      "learning_rate": 9.53231658468864e-06,
      "loss": 0.005,
      "step": 70130
    },
    {
      "epoch": 0.4498561730156847,
      "grad_norm": 0.18488343060016632,
      "learning_rate": 9.532080202195669e-06,
      "loss": 0.0031,
      "step": 70140
    },
    {
      "epoch": 0.4499203099094708,
      "grad_norm": 0.06246650218963623,
      "learning_rate": 9.531843762912496e-06,
      "loss": 0.0045,
      "step": 70150
    },
    {
      "epoch": 0.4499844468032569,
      "grad_norm": 0.1957128643989563,
      "learning_rate": 9.53160726684208e-06,
      "loss": 0.0038,
      "step": 70160
    },
    {
      "epoch": 0.450048583697043,
      "grad_norm": 0.10655216872692108,
      "learning_rate": 9.531370713987388e-06,
      "loss": 0.0041,
      "step": 70170
    },
    {
      "epoch": 0.4501127205908291,
      "grad_norm": 0.23367515206336975,
      "learning_rate": 9.531134104351382e-06,
      "loss": 0.0061,
      "step": 70180
    },
    {
      "epoch": 0.45017685748461517,
      "grad_norm": 0.08084133267402649,
      "learning_rate": 9.530897437937027e-06,
      "loss": 0.0034,
      "step": 70190
    },
    {
      "epoch": 0.45024099437840126,
      "grad_norm": 0.17834603786468506,
      "learning_rate": 9.530660714747288e-06,
      "loss": 0.0042,
      "step": 70200
    },
    {
      "epoch": 0.45030513127218735,
      "grad_norm": 0.17205336689949036,
      "learning_rate": 9.530423934785133e-06,
      "loss": 0.0035,
      "step": 70210
    },
    {
      "epoch": 0.45036926816597345,
      "grad_norm": 0.1258828490972519,
      "learning_rate": 9.53018709805353e-06,
      "loss": 0.0026,
      "step": 70220
    },
    {
      "epoch": 0.45043340505975954,
      "grad_norm": 0.18292279541492462,
      "learning_rate": 9.529950204555443e-06,
      "loss": 0.0066,
      "step": 70230
    },
    {
      "epoch": 0.45049754195354563,
      "grad_norm": 0.1315743774175644,
      "learning_rate": 9.529713254293841e-06,
      "loss": 0.004,
      "step": 70240
    },
    {
      "epoch": 0.4505616788473317,
      "grad_norm": 0.11516522616147995,
      "learning_rate": 9.529476247271698e-06,
      "loss": 0.0047,
      "step": 70250
    },
    {
      "epoch": 0.4506258157411178,
      "grad_norm": 0.2952657639980316,
      "learning_rate": 9.529239183491978e-06,
      "loss": 0.0054,
      "step": 70260
    },
    {
      "epoch": 0.4506899526349039,
      "grad_norm": 0.12432711571455002,
      "learning_rate": 9.529002062957657e-06,
      "loss": 0.0042,
      "step": 70270
    },
    {
      "epoch": 0.45075408952869006,
      "grad_norm": 0.7580963969230652,
      "learning_rate": 9.5287648856717e-06,
      "loss": 0.002,
      "step": 70280
    },
    {
      "epoch": 0.45081822642247615,
      "grad_norm": 0.13588938117027283,
      "learning_rate": 9.528527651637084e-06,
      "loss": 0.0052,
      "step": 70290
    },
    {
      "epoch": 0.45088236331626225,
      "grad_norm": 0.0825934186577797,
      "learning_rate": 9.52829036085678e-06,
      "loss": 0.0044,
      "step": 70300
    },
    {
      "epoch": 0.45094650021004834,
      "grad_norm": 0.14910919964313507,
      "learning_rate": 9.52805301333376e-06,
      "loss": 0.0068,
      "step": 70310
    },
    {
      "epoch": 0.45101063710383443,
      "grad_norm": 0.06526334583759308,
      "learning_rate": 9.527815609071002e-06,
      "loss": 0.0046,
      "step": 70320
    },
    {
      "epoch": 0.4510747739976205,
      "grad_norm": 0.21731288731098175,
      "learning_rate": 9.527578148071476e-06,
      "loss": 0.0043,
      "step": 70330
    },
    {
      "epoch": 0.4511389108914066,
      "grad_norm": 0.16899237036705017,
      "learning_rate": 9.527340630338162e-06,
      "loss": 0.0045,
      "step": 70340
    },
    {
      "epoch": 0.4512030477851927,
      "grad_norm": 0.11214538663625717,
      "learning_rate": 9.527103055874034e-06,
      "loss": 0.0041,
      "step": 70350
    },
    {
      "epoch": 0.4512671846789788,
      "grad_norm": 0.19497740268707275,
      "learning_rate": 9.526865424682068e-06,
      "loss": 0.0034,
      "step": 70360
    },
    {
      "epoch": 0.4513313215727649,
      "grad_norm": 0.40791797637939453,
      "learning_rate": 9.526627736765245e-06,
      "loss": 0.0029,
      "step": 70370
    },
    {
      "epoch": 0.451395458466551,
      "grad_norm": 0.052078232169151306,
      "learning_rate": 9.52638999212654e-06,
      "loss": 0.0073,
      "step": 70380
    },
    {
      "epoch": 0.4514595953603371,
      "grad_norm": 0.18052902817726135,
      "learning_rate": 9.526152190768932e-06,
      "loss": 0.0032,
      "step": 70390
    },
    {
      "epoch": 0.4515237322541232,
      "grad_norm": 0.23692411184310913,
      "learning_rate": 9.525914332695403e-06,
      "loss": 0.0041,
      "step": 70400
    },
    {
      "epoch": 0.45158786914790927,
      "grad_norm": 0.1767560839653015,
      "learning_rate": 9.525676417908933e-06,
      "loss": 0.0056,
      "step": 70410
    },
    {
      "epoch": 0.4516520060416954,
      "grad_norm": 0.1576530635356903,
      "learning_rate": 9.525438446412504e-06,
      "loss": 0.0045,
      "step": 70420
    },
    {
      "epoch": 0.4517161429354815,
      "grad_norm": 0.18247544765472412,
      "learning_rate": 9.525200418209095e-06,
      "loss": 0.0065,
      "step": 70430
    },
    {
      "epoch": 0.4517802798292676,
      "grad_norm": 0.05818657577037811,
      "learning_rate": 9.524962333301694e-06,
      "loss": 0.0049,
      "step": 70440
    },
    {
      "epoch": 0.4518444167230537,
      "grad_norm": 0.2344546765089035,
      "learning_rate": 9.52472419169328e-06,
      "loss": 0.0045,
      "step": 70450
    },
    {
      "epoch": 0.4519085536168398,
      "grad_norm": 0.19400371611118317,
      "learning_rate": 9.524485993386836e-06,
      "loss": 0.002,
      "step": 70460
    },
    {
      "epoch": 0.4519726905106259,
      "grad_norm": 0.07352828234434128,
      "learning_rate": 9.52424773838535e-06,
      "loss": 0.003,
      "step": 70470
    },
    {
      "epoch": 0.452036827404412,
      "grad_norm": 0.030290966853499413,
      "learning_rate": 9.524009426691806e-06,
      "loss": 0.0021,
      "step": 70480
    },
    {
      "epoch": 0.45210096429819807,
      "grad_norm": 0.17533548176288605,
      "learning_rate": 9.523771058309192e-06,
      "loss": 0.0037,
      "step": 70490
    },
    {
      "epoch": 0.45216510119198416,
      "grad_norm": 0.16610278189182281,
      "learning_rate": 9.523532633240492e-06,
      "loss": 0.0036,
      "step": 70500
    },
    {
      "epoch": 0.45222923808577026,
      "grad_norm": 0.4312562346458435,
      "learning_rate": 9.523294151488696e-06,
      "loss": 0.0047,
      "step": 70510
    },
    {
      "epoch": 0.45229337497955635,
      "grad_norm": 0.15876057744026184,
      "learning_rate": 9.523055613056791e-06,
      "loss": 0.004,
      "step": 70520
    },
    {
      "epoch": 0.45235751187334244,
      "grad_norm": 0.06935004144906998,
      "learning_rate": 9.522817017947767e-06,
      "loss": 0.0037,
      "step": 70530
    },
    {
      "epoch": 0.45242164876712854,
      "grad_norm": 0.07293408364057541,
      "learning_rate": 9.522578366164614e-06,
      "loss": 0.0069,
      "step": 70540
    },
    {
      "epoch": 0.45248578566091463,
      "grad_norm": 0.06884395331144333,
      "learning_rate": 9.52233965771032e-06,
      "loss": 0.0028,
      "step": 70550
    },
    {
      "epoch": 0.4525499225547008,
      "grad_norm": 0.13001060485839844,
      "learning_rate": 9.52210089258788e-06,
      "loss": 0.0026,
      "step": 70560
    },
    {
      "epoch": 0.45261405944848687,
      "grad_norm": 0.38709232211112976,
      "learning_rate": 9.521862070800283e-06,
      "loss": 0.0046,
      "step": 70570
    },
    {
      "epoch": 0.45267819634227296,
      "grad_norm": 0.26731571555137634,
      "learning_rate": 9.521623192350522e-06,
      "loss": 0.004,
      "step": 70580
    },
    {
      "epoch": 0.45274233323605906,
      "grad_norm": 0.20551343262195587,
      "learning_rate": 9.521384257241592e-06,
      "loss": 0.0043,
      "step": 70590
    },
    {
      "epoch": 0.45280647012984515,
      "grad_norm": 1.012932300567627,
      "learning_rate": 9.521145265476486e-06,
      "loss": 0.0132,
      "step": 70600
    },
    {
      "epoch": 0.45287060702363124,
      "grad_norm": 0.15066519379615784,
      "learning_rate": 9.520906217058197e-06,
      "loss": 0.0027,
      "step": 70610
    },
    {
      "epoch": 0.45293474391741734,
      "grad_norm": 0.11236685514450073,
      "learning_rate": 9.520667111989722e-06,
      "loss": 0.0036,
      "step": 70620
    },
    {
      "epoch": 0.45299888081120343,
      "grad_norm": 0.3001996576786041,
      "learning_rate": 9.52042795027406e-06,
      "loss": 0.0051,
      "step": 70630
    },
    {
      "epoch": 0.4530630177049895,
      "grad_norm": 0.05467584356665611,
      "learning_rate": 9.520188731914203e-06,
      "loss": 0.0071,
      "step": 70640
    },
    {
      "epoch": 0.4531271545987756,
      "grad_norm": 0.06468317657709122,
      "learning_rate": 9.51994945691315e-06,
      "loss": 0.004,
      "step": 70650
    },
    {
      "epoch": 0.4531912914925617,
      "grad_norm": 0.1886305809020996,
      "learning_rate": 9.519710125273902e-06,
      "loss": 0.0184,
      "step": 70660
    },
    {
      "epoch": 0.4532554283863478,
      "grad_norm": 0.0639183297753334,
      "learning_rate": 9.519470736999455e-06,
      "loss": 0.0037,
      "step": 70670
    },
    {
      "epoch": 0.4533195652801339,
      "grad_norm": 0.0735633373260498,
      "learning_rate": 9.519231292092808e-06,
      "loss": 0.0045,
      "step": 70680
    },
    {
      "epoch": 0.45338370217392,
      "grad_norm": 0.34701141715049744,
      "learning_rate": 9.518991790556965e-06,
      "loss": 0.0043,
      "step": 70690
    },
    {
      "epoch": 0.45344783906770614,
      "grad_norm": 0.18114013969898224,
      "learning_rate": 9.518752232394925e-06,
      "loss": 0.0036,
      "step": 70700
    },
    {
      "epoch": 0.45351197596149223,
      "grad_norm": 0.32887938618659973,
      "learning_rate": 9.51851261760969e-06,
      "loss": 0.0059,
      "step": 70710
    },
    {
      "epoch": 0.4535761128552783,
      "grad_norm": 0.2900322675704956,
      "learning_rate": 9.518272946204263e-06,
      "loss": 0.0052,
      "step": 70720
    },
    {
      "epoch": 0.4536402497490644,
      "grad_norm": 0.28895503282546997,
      "learning_rate": 9.518033218181646e-06,
      "loss": 0.0061,
      "step": 70730
    },
    {
      "epoch": 0.4537043866428505,
      "grad_norm": 0.11479459702968597,
      "learning_rate": 9.517793433544844e-06,
      "loss": 0.0027,
      "step": 70740
    },
    {
      "epoch": 0.4537685235366366,
      "grad_norm": 0.07610035687685013,
      "learning_rate": 9.51755359229686e-06,
      "loss": 0.0036,
      "step": 70750
    },
    {
      "epoch": 0.4538326604304227,
      "grad_norm": 0.10561048239469528,
      "learning_rate": 9.517313694440702e-06,
      "loss": 0.0024,
      "step": 70760
    },
    {
      "epoch": 0.4538967973242088,
      "grad_norm": 0.2110764980316162,
      "learning_rate": 9.517073739979377e-06,
      "loss": 0.0033,
      "step": 70770
    },
    {
      "epoch": 0.4539609342179949,
      "grad_norm": 0.19672437012195587,
      "learning_rate": 9.516833728915887e-06,
      "loss": 0.0042,
      "step": 70780
    },
    {
      "epoch": 0.454025071111781,
      "grad_norm": 0.1142696738243103,
      "learning_rate": 9.516593661253244e-06,
      "loss": 0.0031,
      "step": 70790
    },
    {
      "epoch": 0.45408920800556707,
      "grad_norm": 0.17315009236335754,
      "learning_rate": 9.516353536994452e-06,
      "loss": 0.0107,
      "step": 70800
    },
    {
      "epoch": 0.45415334489935316,
      "grad_norm": 0.10731589049100876,
      "learning_rate": 9.516113356142525e-06,
      "loss": 0.0054,
      "step": 70810
    },
    {
      "epoch": 0.45421748179313925,
      "grad_norm": 0.10094081610441208,
      "learning_rate": 9.515873118700469e-06,
      "loss": 0.0021,
      "step": 70820
    },
    {
      "epoch": 0.45428161868692535,
      "grad_norm": 0.12132128328084946,
      "learning_rate": 9.515632824671294e-06,
      "loss": 0.0016,
      "step": 70830
    },
    {
      "epoch": 0.4543457555807115,
      "grad_norm": 0.21277032792568207,
      "learning_rate": 9.515392474058015e-06,
      "loss": 0.0031,
      "step": 70840
    },
    {
      "epoch": 0.4544098924744976,
      "grad_norm": 0.04919726029038429,
      "learning_rate": 9.51515206686364e-06,
      "loss": 0.0037,
      "step": 70850
    },
    {
      "epoch": 0.4544740293682837,
      "grad_norm": 0.031588517129421234,
      "learning_rate": 9.514911603091183e-06,
      "loss": 0.008,
      "step": 70860
    },
    {
      "epoch": 0.4545381662620698,
      "grad_norm": 0.2616173028945923,
      "learning_rate": 9.514671082743656e-06,
      "loss": 0.0028,
      "step": 70870
    },
    {
      "epoch": 0.45460230315585587,
      "grad_norm": 0.19474613666534424,
      "learning_rate": 9.514430505824075e-06,
      "loss": 0.0043,
      "step": 70880
    },
    {
      "epoch": 0.45466644004964196,
      "grad_norm": 0.09102250635623932,
      "learning_rate": 9.514189872335454e-06,
      "loss": 0.0033,
      "step": 70890
    },
    {
      "epoch": 0.45473057694342806,
      "grad_norm": 0.1772601306438446,
      "learning_rate": 9.513949182280804e-06,
      "loss": 0.0047,
      "step": 70900
    },
    {
      "epoch": 0.45479471383721415,
      "grad_norm": 0.254000186920166,
      "learning_rate": 9.513708435663147e-06,
      "loss": 0.0038,
      "step": 70910
    },
    {
      "epoch": 0.45485885073100024,
      "grad_norm": 0.15698865056037903,
      "learning_rate": 9.513467632485498e-06,
      "loss": 0.0044,
      "step": 70920
    },
    {
      "epoch": 0.45492298762478633,
      "grad_norm": 0.43258821964263916,
      "learning_rate": 9.513226772750873e-06,
      "loss": 0.0049,
      "step": 70930
    },
    {
      "epoch": 0.45498712451857243,
      "grad_norm": 0.02216915786266327,
      "learning_rate": 9.512985856462292e-06,
      "loss": 0.0036,
      "step": 70940
    },
    {
      "epoch": 0.4550512614123585,
      "grad_norm": 0.19537481665611267,
      "learning_rate": 9.512744883622772e-06,
      "loss": 0.0045,
      "step": 70950
    },
    {
      "epoch": 0.4551153983061446,
      "grad_norm": 0.1020500659942627,
      "learning_rate": 9.512503854235333e-06,
      "loss": 0.0038,
      "step": 70960
    },
    {
      "epoch": 0.4551795351999307,
      "grad_norm": 0.23862917721271515,
      "learning_rate": 9.512262768302996e-06,
      "loss": 0.0043,
      "step": 70970
    },
    {
      "epoch": 0.45524367209371686,
      "grad_norm": 0.1556473970413208,
      "learning_rate": 9.512021625828782e-06,
      "loss": 0.0059,
      "step": 70980
    },
    {
      "epoch": 0.45530780898750295,
      "grad_norm": 0.09845450520515442,
      "learning_rate": 9.511780426815712e-06,
      "loss": 0.0039,
      "step": 70990
    },
    {
      "epoch": 0.45537194588128904,
      "grad_norm": 0.22103340923786163,
      "learning_rate": 9.511539171266808e-06,
      "loss": 0.0031,
      "step": 71000
    },
    {
      "epoch": 0.45543608277507514,
      "grad_norm": 0.20481856167316437,
      "learning_rate": 9.511297859185095e-06,
      "loss": 0.003,
      "step": 71010
    },
    {
      "epoch": 0.45550021966886123,
      "grad_norm": 0.12524117529392242,
      "learning_rate": 9.511056490573596e-06,
      "loss": 0.006,
      "step": 71020
    },
    {
      "epoch": 0.4555643565626473,
      "grad_norm": 0.2552565634250641,
      "learning_rate": 9.510815065435335e-06,
      "loss": 0.0031,
      "step": 71030
    },
    {
      "epoch": 0.4556284934564334,
      "grad_norm": 0.10889584571123123,
      "learning_rate": 9.510573583773336e-06,
      "loss": 0.004,
      "step": 71040
    },
    {
      "epoch": 0.4556926303502195,
      "grad_norm": 0.06692889332771301,
      "learning_rate": 9.510332045590627e-06,
      "loss": 0.0048,
      "step": 71050
    },
    {
      "epoch": 0.4557567672440056,
      "grad_norm": 0.3315446078777313,
      "learning_rate": 9.510090450890236e-06,
      "loss": 0.0045,
      "step": 71060
    },
    {
      "epoch": 0.4558209041377917,
      "grad_norm": 0.20021110773086548,
      "learning_rate": 9.509848799675186e-06,
      "loss": 0.0069,
      "step": 71070
    },
    {
      "epoch": 0.4558850410315778,
      "grad_norm": 0.33070963621139526,
      "learning_rate": 9.509607091948507e-06,
      "loss": 0.0038,
      "step": 71080
    },
    {
      "epoch": 0.4559491779253639,
      "grad_norm": 0.05399494618177414,
      "learning_rate": 9.509365327713229e-06,
      "loss": 0.002,
      "step": 71090
    },
    {
      "epoch": 0.45601331481915,
      "grad_norm": 0.0842948704957962,
      "learning_rate": 9.509123506972382e-06,
      "loss": 0.0051,
      "step": 71100
    },
    {
      "epoch": 0.45607745171293607,
      "grad_norm": 0.0796792134642601,
      "learning_rate": 9.508881629728992e-06,
      "loss": 0.0037,
      "step": 71110
    },
    {
      "epoch": 0.4561415886067222,
      "grad_norm": 0.32939353585243225,
      "learning_rate": 9.508639695986094e-06,
      "loss": 0.0038,
      "step": 71120
    },
    {
      "epoch": 0.4562057255005083,
      "grad_norm": 0.24936258792877197,
      "learning_rate": 9.508397705746719e-06,
      "loss": 0.003,
      "step": 71130
    },
    {
      "epoch": 0.4562698623942944,
      "grad_norm": 0.23419713973999023,
      "learning_rate": 9.5081556590139e-06,
      "loss": 0.0035,
      "step": 71140
    },
    {
      "epoch": 0.4563339992880805,
      "grad_norm": 0.5287534594535828,
      "learning_rate": 9.507913555790666e-06,
      "loss": 0.0066,
      "step": 71150
    },
    {
      "epoch": 0.4563981361818666,
      "grad_norm": 0.2167656570672989,
      "learning_rate": 9.507671396080054e-06,
      "loss": 0.0051,
      "step": 71160
    },
    {
      "epoch": 0.4564622730756527,
      "grad_norm": 0.034404855221509933,
      "learning_rate": 9.5074291798851e-06,
      "loss": 0.0056,
      "step": 71170
    },
    {
      "epoch": 0.4565264099694388,
      "grad_norm": 0.46575120091438293,
      "learning_rate": 9.507186907208834e-06,
      "loss": 0.0059,
      "step": 71180
    },
    {
      "epoch": 0.45659054686322487,
      "grad_norm": 0.11899673938751221,
      "learning_rate": 9.506944578054295e-06,
      "loss": 0.0038,
      "step": 71190
    },
    {
      "epoch": 0.45665468375701096,
      "grad_norm": 0.48142024874687195,
      "learning_rate": 9.506702192424522e-06,
      "loss": 0.0041,
      "step": 71200
    },
    {
      "epoch": 0.45671882065079705,
      "grad_norm": 0.15543551743030548,
      "learning_rate": 9.506459750322548e-06,
      "loss": 0.0052,
      "step": 71210
    },
    {
      "epoch": 0.45678295754458315,
      "grad_norm": 0.1856965869665146,
      "learning_rate": 9.506217251751411e-06,
      "loss": 0.0046,
      "step": 71220
    },
    {
      "epoch": 0.45684709443836924,
      "grad_norm": 0.23232631385326385,
      "learning_rate": 9.505974696714153e-06,
      "loss": 0.0043,
      "step": 71230
    },
    {
      "epoch": 0.45691123133215533,
      "grad_norm": 0.23498240113258362,
      "learning_rate": 9.505732085213812e-06,
      "loss": 0.0029,
      "step": 71240
    },
    {
      "epoch": 0.4569753682259414,
      "grad_norm": 0.10573287308216095,
      "learning_rate": 9.505489417253428e-06,
      "loss": 0.0042,
      "step": 71250
    },
    {
      "epoch": 0.4570395051197276,
      "grad_norm": 0.13049164414405823,
      "learning_rate": 9.505246692836041e-06,
      "loss": 0.0021,
      "step": 71260
    },
    {
      "epoch": 0.45710364201351367,
      "grad_norm": 0.24431076645851135,
      "learning_rate": 9.505003911964692e-06,
      "loss": 0.0035,
      "step": 71270
    },
    {
      "epoch": 0.45716777890729976,
      "grad_norm": 0.14189817011356354,
      "learning_rate": 9.504761074642426e-06,
      "loss": 0.0035,
      "step": 71280
    },
    {
      "epoch": 0.45723191580108585,
      "grad_norm": 0.08879505097866058,
      "learning_rate": 9.504518180872283e-06,
      "loss": 0.0041,
      "step": 71290
    },
    {
      "epoch": 0.45729605269487195,
      "grad_norm": 0.177837073802948,
      "learning_rate": 9.504275230657309e-06,
      "loss": 0.0033,
      "step": 71300
    },
    {
      "epoch": 0.45736018958865804,
      "grad_norm": 0.182501420378685,
      "learning_rate": 9.504032224000546e-06,
      "loss": 0.0035,
      "step": 71310
    },
    {
      "epoch": 0.45742432648244413,
      "grad_norm": 0.1433417797088623,
      "learning_rate": 9.503789160905042e-06,
      "loss": 0.0064,
      "step": 71320
    },
    {
      "epoch": 0.4574884633762302,
      "grad_norm": 0.009760375134646893,
      "learning_rate": 9.503546041373838e-06,
      "loss": 0.0046,
      "step": 71330
    },
    {
      "epoch": 0.4575526002700163,
      "grad_norm": 0.040973514318466187,
      "learning_rate": 9.503302865409987e-06,
      "loss": 0.0029,
      "step": 71340
    },
    {
      "epoch": 0.4576167371638024,
      "grad_norm": 0.057639192789793015,
      "learning_rate": 9.503059633016529e-06,
      "loss": 0.0026,
      "step": 71350
    },
    {
      "epoch": 0.4576808740575885,
      "grad_norm": 0.10568902641534805,
      "learning_rate": 9.502816344196517e-06,
      "loss": 0.0032,
      "step": 71360
    },
    {
      "epoch": 0.4577450109513746,
      "grad_norm": 0.2689015865325928,
      "learning_rate": 9.502572998953e-06,
      "loss": 0.0022,
      "step": 71370
    },
    {
      "epoch": 0.4578091478451607,
      "grad_norm": 0.049804773181676865,
      "learning_rate": 9.502329597289025e-06,
      "loss": 0.0024,
      "step": 71380
    },
    {
      "epoch": 0.4578732847389468,
      "grad_norm": 0.1805211454629898,
      "learning_rate": 9.50208613920764e-06,
      "loss": 0.004,
      "step": 71390
    },
    {
      "epoch": 0.45793742163273293,
      "grad_norm": 0.1536896973848343,
      "learning_rate": 9.501842624711899e-06,
      "loss": 0.0053,
      "step": 71400
    },
    {
      "epoch": 0.458001558526519,
      "grad_norm": 0.14408615231513977,
      "learning_rate": 9.501599053804854e-06,
      "loss": 0.0049,
      "step": 71410
    },
    {
      "epoch": 0.4580656954203051,
      "grad_norm": 0.568912923336029,
      "learning_rate": 9.501355426489553e-06,
      "loss": 0.0025,
      "step": 71420
    },
    {
      "epoch": 0.4581298323140912,
      "grad_norm": 0.1006404310464859,
      "learning_rate": 9.501111742769054e-06,
      "loss": 0.0027,
      "step": 71430
    },
    {
      "epoch": 0.4581939692078773,
      "grad_norm": 0.010831396095454693,
      "learning_rate": 9.500868002646407e-06,
      "loss": 0.0052,
      "step": 71440
    },
    {
      "epoch": 0.4582581061016634,
      "grad_norm": 0.3710175156593323,
      "learning_rate": 9.500624206124667e-06,
      "loss": 0.0062,
      "step": 71450
    },
    {
      "epoch": 0.4583222429954495,
      "grad_norm": 0.15602272748947144,
      "learning_rate": 9.50038035320689e-06,
      "loss": 0.0079,
      "step": 71460
    },
    {
      "epoch": 0.4583863798892356,
      "grad_norm": 0.11349541693925858,
      "learning_rate": 9.50013644389613e-06,
      "loss": 0.0045,
      "step": 71470
    },
    {
      "epoch": 0.4584505167830217,
      "grad_norm": 0.13774773478507996,
      "learning_rate": 9.499892478195444e-06,
      "loss": 0.0029,
      "step": 71480
    },
    {
      "epoch": 0.45851465367680777,
      "grad_norm": 0.10175567120313644,
      "learning_rate": 9.49964845610789e-06,
      "loss": 0.0029,
      "step": 71490
    },
    {
      "epoch": 0.45857879057059386,
      "grad_norm": 0.14116734266281128,
      "learning_rate": 9.499404377636523e-06,
      "loss": 0.0051,
      "step": 71500
    },
    {
      "epoch": 0.45864292746437996,
      "grad_norm": 0.13040600717067719,
      "learning_rate": 9.499160242784406e-06,
      "loss": 0.0026,
      "step": 71510
    },
    {
      "epoch": 0.45870706435816605,
      "grad_norm": 0.21801449358463287,
      "learning_rate": 9.498916051554595e-06,
      "loss": 0.0035,
      "step": 71520
    },
    {
      "epoch": 0.45877120125195214,
      "grad_norm": 0.34369441866874695,
      "learning_rate": 9.49867180395015e-06,
      "loss": 0.0026,
      "step": 71530
    },
    {
      "epoch": 0.45883533814573824,
      "grad_norm": 0.18488413095474243,
      "learning_rate": 9.49842749997413e-06,
      "loss": 0.0035,
      "step": 71540
    },
    {
      "epoch": 0.4588994750395244,
      "grad_norm": 0.14039179682731628,
      "learning_rate": 9.498183139629602e-06,
      "loss": 0.0031,
      "step": 71550
    },
    {
      "epoch": 0.4589636119333105,
      "grad_norm": 0.18675455451011658,
      "learning_rate": 9.497938722919623e-06,
      "loss": 0.0049,
      "step": 71560
    },
    {
      "epoch": 0.45902774882709657,
      "grad_norm": 0.1299862116575241,
      "learning_rate": 9.497694249847258e-06,
      "loss": 0.0025,
      "step": 71570
    },
    {
      "epoch": 0.45909188572088266,
      "grad_norm": 0.16409076750278473,
      "learning_rate": 9.497449720415568e-06,
      "loss": 0.0033,
      "step": 71580
    },
    {
      "epoch": 0.45915602261466876,
      "grad_norm": 0.19340041279792786,
      "learning_rate": 9.497205134627621e-06,
      "loss": 0.0023,
      "step": 71590
    },
    {
      "epoch": 0.45922015950845485,
      "grad_norm": 0.11618074029684067,
      "learning_rate": 9.496960492486478e-06,
      "loss": 0.0033,
      "step": 71600
    },
    {
      "epoch": 0.45928429640224094,
      "grad_norm": 0.3043084740638733,
      "learning_rate": 9.496715793995206e-06,
      "loss": 0.0064,
      "step": 71610
    },
    {
      "epoch": 0.45934843329602704,
      "grad_norm": 0.10752145200967789,
      "learning_rate": 9.496471039156871e-06,
      "loss": 0.0021,
      "step": 71620
    },
    {
      "epoch": 0.45941257018981313,
      "grad_norm": 0.08993718028068542,
      "learning_rate": 9.49622622797454e-06,
      "loss": 0.003,
      "step": 71630
    },
    {
      "epoch": 0.4594767070835992,
      "grad_norm": 0.14545069634914398,
      "learning_rate": 9.495981360451283e-06,
      "loss": 0.0052,
      "step": 71640
    },
    {
      "epoch": 0.4595408439773853,
      "grad_norm": 0.06524930894374847,
      "learning_rate": 9.495736436590166e-06,
      "loss": 0.0029,
      "step": 71650
    },
    {
      "epoch": 0.4596049808711714,
      "grad_norm": 0.09713975340127945,
      "learning_rate": 9.495491456394257e-06,
      "loss": 0.0032,
      "step": 71660
    },
    {
      "epoch": 0.4596691177649575,
      "grad_norm": 0.09874634444713593,
      "learning_rate": 9.495246419866628e-06,
      "loss": 0.0029,
      "step": 71670
    },
    {
      "epoch": 0.4597332546587436,
      "grad_norm": 0.04133143648505211,
      "learning_rate": 9.49500132701035e-06,
      "loss": 0.0032,
      "step": 71680
    },
    {
      "epoch": 0.45979739155252974,
      "grad_norm": 0.48549845814704895,
      "learning_rate": 9.49475617782849e-06,
      "loss": 0.0041,
      "step": 71690
    },
    {
      "epoch": 0.45986152844631584,
      "grad_norm": 0.08148328214883804,
      "learning_rate": 9.494510972324124e-06,
      "loss": 0.004,
      "step": 71700
    },
    {
      "epoch": 0.45992566534010193,
      "grad_norm": 0.2185431867837906,
      "learning_rate": 9.494265710500324e-06,
      "loss": 0.0033,
      "step": 71710
    },
    {
      "epoch": 0.459989802233888,
      "grad_norm": 0.13162735104560852,
      "learning_rate": 9.494020392360161e-06,
      "loss": 0.0021,
      "step": 71720
    },
    {
      "epoch": 0.4600539391276741,
      "grad_norm": 0.17970408499240875,
      "learning_rate": 9.493775017906712e-06,
      "loss": 0.0034,
      "step": 71730
    },
    {
      "epoch": 0.4601180760214602,
      "grad_norm": 0.19356241822242737,
      "learning_rate": 9.49352958714305e-06,
      "loss": 0.0089,
      "step": 71740
    },
    {
      "epoch": 0.4601822129152463,
      "grad_norm": 0.1996283084154129,
      "learning_rate": 9.49328410007225e-06,
      "loss": 0.0024,
      "step": 71750
    },
    {
      "epoch": 0.4602463498090324,
      "grad_norm": 0.28187283873558044,
      "learning_rate": 9.493038556697391e-06,
      "loss": 0.0053,
      "step": 71760
    },
    {
      "epoch": 0.4603104867028185,
      "grad_norm": 0.06501268595457077,
      "learning_rate": 9.492792957021546e-06,
      "loss": 0.0051,
      "step": 71770
    },
    {
      "epoch": 0.4603746235966046,
      "grad_norm": 0.3292664885520935,
      "learning_rate": 9.492547301047794e-06,
      "loss": 0.0038,
      "step": 71780
    },
    {
      "epoch": 0.4604387604903907,
      "grad_norm": 0.15045075118541718,
      "learning_rate": 9.492301588779215e-06,
      "loss": 0.003,
      "step": 71790
    },
    {
      "epoch": 0.46050289738417677,
      "grad_norm": 0.07394632697105408,
      "learning_rate": 9.492055820218886e-06,
      "loss": 0.0035,
      "step": 71800
    },
    {
      "epoch": 0.46056703427796286,
      "grad_norm": 0.09739391505718231,
      "learning_rate": 9.491809995369888e-06,
      "loss": 0.0034,
      "step": 71810
    },
    {
      "epoch": 0.46063117117174895,
      "grad_norm": 0.11213880777359009,
      "learning_rate": 9.491564114235299e-06,
      "loss": 0.0037,
      "step": 71820
    },
    {
      "epoch": 0.4606953080655351,
      "grad_norm": 0.10225711017847061,
      "learning_rate": 9.491318176818203e-06,
      "loss": 0.005,
      "step": 71830
    },
    {
      "epoch": 0.4607594449593212,
      "grad_norm": 0.1804584562778473,
      "learning_rate": 9.491072183121679e-06,
      "loss": 0.0029,
      "step": 71840
    },
    {
      "epoch": 0.4608235818531073,
      "grad_norm": 0.21691973507404327,
      "learning_rate": 9.490826133148812e-06,
      "loss": 0.007,
      "step": 71850
    },
    {
      "epoch": 0.4608877187468934,
      "grad_norm": 0.40952345728874207,
      "learning_rate": 9.490580026902683e-06,
      "loss": 0.003,
      "step": 71860
    },
    {
      "epoch": 0.4609518556406795,
      "grad_norm": 0.07483315467834473,
      "learning_rate": 9.490333864386377e-06,
      "loss": 0.0048,
      "step": 71870
    },
    {
      "epoch": 0.46101599253446557,
      "grad_norm": 0.06613636016845703,
      "learning_rate": 9.490087645602978e-06,
      "loss": 0.0039,
      "step": 71880
    },
    {
      "epoch": 0.46108012942825166,
      "grad_norm": 0.20446856319904327,
      "learning_rate": 9.489841370555573e-06,
      "loss": 0.0033,
      "step": 71890
    },
    {
      "epoch": 0.46114426632203775,
      "grad_norm": 0.23011180758476257,
      "learning_rate": 9.489595039247246e-06,
      "loss": 0.0054,
      "step": 71900
    },
    {
      "epoch": 0.46120840321582385,
      "grad_norm": 0.16415435075759888,
      "learning_rate": 9.489348651681085e-06,
      "loss": 0.0053,
      "step": 71910
    },
    {
      "epoch": 0.46127254010960994,
      "grad_norm": 0.2752489149570465,
      "learning_rate": 9.489102207860175e-06,
      "loss": 0.0056,
      "step": 71920
    },
    {
      "epoch": 0.46133667700339603,
      "grad_norm": 0.19223229587078094,
      "learning_rate": 9.488855707787609e-06,
      "loss": 0.0034,
      "step": 71930
    },
    {
      "epoch": 0.4614008138971821,
      "grad_norm": 0.05069417878985405,
      "learning_rate": 9.488609151466471e-06,
      "loss": 0.0032,
      "step": 71940
    },
    {
      "epoch": 0.4614649507909682,
      "grad_norm": 0.11428840458393097,
      "learning_rate": 9.488362538899854e-06,
      "loss": 0.0044,
      "step": 71950
    },
    {
      "epoch": 0.4615290876847543,
      "grad_norm": 0.1510365903377533,
      "learning_rate": 9.488115870090843e-06,
      "loss": 0.0041,
      "step": 71960
    },
    {
      "epoch": 0.46159322457854046,
      "grad_norm": 0.2194390743970871,
      "learning_rate": 9.487869145042537e-06,
      "loss": 0.0043,
      "step": 71970
    },
    {
      "epoch": 0.46165736147232656,
      "grad_norm": 0.22613966464996338,
      "learning_rate": 9.48762236375802e-06,
      "loss": 0.0048,
      "step": 71980
    },
    {
      "epoch": 0.46172149836611265,
      "grad_norm": 0.16205669939517975,
      "learning_rate": 9.48737552624039e-06,
      "loss": 0.0054,
      "step": 71990
    },
    {
      "epoch": 0.46178563525989874,
      "grad_norm": 0.24870528280735016,
      "learning_rate": 9.487128632492735e-06,
      "loss": 0.0039,
      "step": 72000
    },
    {
      "epoch": 0.46184977215368483,
      "grad_norm": 0.16427086293697357,
      "learning_rate": 9.486881682518155e-06,
      "loss": 0.0024,
      "step": 72010
    },
    {
      "epoch": 0.4619139090474709,
      "grad_norm": 0.14098943769931793,
      "learning_rate": 9.486634676319736e-06,
      "loss": 0.0026,
      "step": 72020
    },
    {
      "epoch": 0.461978045941257,
      "grad_norm": 0.13580282032489777,
      "learning_rate": 9.48638761390058e-06,
      "loss": 0.0052,
      "step": 72030
    },
    {
      "epoch": 0.4620421828350431,
      "grad_norm": 0.14998015761375427,
      "learning_rate": 9.486140495263783e-06,
      "loss": 0.0026,
      "step": 72040
    },
    {
      "epoch": 0.4621063197288292,
      "grad_norm": 0.11328137665987015,
      "learning_rate": 9.485893320412438e-06,
      "loss": 0.0027,
      "step": 72050
    },
    {
      "epoch": 0.4621704566226153,
      "grad_norm": 0.15475858747959137,
      "learning_rate": 9.485646089349643e-06,
      "loss": 0.0054,
      "step": 72060
    },
    {
      "epoch": 0.4622345935164014,
      "grad_norm": 0.1065763458609581,
      "learning_rate": 9.485398802078497e-06,
      "loss": 0.006,
      "step": 72070
    },
    {
      "epoch": 0.4622987304101875,
      "grad_norm": 0.19568683207035065,
      "learning_rate": 9.485151458602097e-06,
      "loss": 0.0043,
      "step": 72080
    },
    {
      "epoch": 0.4623628673039736,
      "grad_norm": 0.2813143730163574,
      "learning_rate": 9.484904058923546e-06,
      "loss": 0.0067,
      "step": 72090
    },
    {
      "epoch": 0.4624270041977597,
      "grad_norm": 0.22191986441612244,
      "learning_rate": 9.48465660304594e-06,
      "loss": 0.0046,
      "step": 72100
    },
    {
      "epoch": 0.4624911410915458,
      "grad_norm": 0.17017143964767456,
      "learning_rate": 9.484409090972384e-06,
      "loss": 0.0028,
      "step": 72110
    },
    {
      "epoch": 0.4625552779853319,
      "grad_norm": 0.11975177377462387,
      "learning_rate": 9.484161522705975e-06,
      "loss": 0.0036,
      "step": 72120
    },
    {
      "epoch": 0.462619414879118,
      "grad_norm": 0.04267343133687973,
      "learning_rate": 9.483913898249817e-06,
      "loss": 0.003,
      "step": 72130
    },
    {
      "epoch": 0.4626835517729041,
      "grad_norm": 0.035006728023290634,
      "learning_rate": 9.483666217607015e-06,
      "loss": 0.0039,
      "step": 72140
    },
    {
      "epoch": 0.4627476886666902,
      "grad_norm": 0.09996423125267029,
      "learning_rate": 9.48341848078067e-06,
      "loss": 0.0024,
      "step": 72150
    },
    {
      "epoch": 0.4628118255604763,
      "grad_norm": 0.44871264696121216,
      "learning_rate": 9.483170687773888e-06,
      "loss": 0.0049,
      "step": 72160
    },
    {
      "epoch": 0.4628759624542624,
      "grad_norm": 0.10117613524198532,
      "learning_rate": 9.482922838589772e-06,
      "loss": 0.0034,
      "step": 72170
    },
    {
      "epoch": 0.4629400993480485,
      "grad_norm": 0.1156907007098198,
      "learning_rate": 9.482674933231428e-06,
      "loss": 0.0025,
      "step": 72180
    },
    {
      "epoch": 0.46300423624183457,
      "grad_norm": 0.05373195931315422,
      "learning_rate": 9.482426971701966e-06,
      "loss": 0.0032,
      "step": 72190
    },
    {
      "epoch": 0.46306837313562066,
      "grad_norm": 0.24135103821754456,
      "learning_rate": 9.482178954004488e-06,
      "loss": 0.0033,
      "step": 72200
    },
    {
      "epoch": 0.46313251002940675,
      "grad_norm": 0.2547978162765503,
      "learning_rate": 9.481930880142107e-06,
      "loss": 0.0041,
      "step": 72210
    },
    {
      "epoch": 0.46319664692319285,
      "grad_norm": 0.27852529287338257,
      "learning_rate": 9.481682750117926e-06,
      "loss": 0.0038,
      "step": 72220
    },
    {
      "epoch": 0.46326078381697894,
      "grad_norm": 0.12560653686523438,
      "learning_rate": 9.48143456393506e-06,
      "loss": 0.006,
      "step": 72230
    },
    {
      "epoch": 0.46332492071076503,
      "grad_norm": 0.15211427211761475,
      "learning_rate": 9.481186321596614e-06,
      "loss": 0.005,
      "step": 72240
    },
    {
      "epoch": 0.4633890576045512,
      "grad_norm": 0.16033463180065155,
      "learning_rate": 9.480938023105702e-06,
      "loss": 0.0031,
      "step": 72250
    },
    {
      "epoch": 0.4634531944983373,
      "grad_norm": 0.1595131754875183,
      "learning_rate": 9.480689668465433e-06,
      "loss": 0.0039,
      "step": 72260
    },
    {
      "epoch": 0.46351733139212337,
      "grad_norm": 0.27040454745292664,
      "learning_rate": 9.48044125767892e-06,
      "loss": 0.0038,
      "step": 72270
    },
    {
      "epoch": 0.46358146828590946,
      "grad_norm": 0.37454044818878174,
      "learning_rate": 9.480192790749277e-06,
      "loss": 0.0038,
      "step": 72280
    },
    {
      "epoch": 0.46364560517969555,
      "grad_norm": 0.16823004186153412,
      "learning_rate": 9.479944267679617e-06,
      "loss": 0.0041,
      "step": 72290
    },
    {
      "epoch": 0.46370974207348165,
      "grad_norm": 0.07666993141174316,
      "learning_rate": 9.479695688473051e-06,
      "loss": 0.0029,
      "step": 72300
    },
    {
      "epoch": 0.46377387896726774,
      "grad_norm": 0.2639802098274231,
      "learning_rate": 9.4794470531327e-06,
      "loss": 0.0052,
      "step": 72310
    },
    {
      "epoch": 0.46383801586105383,
      "grad_norm": 0.05207554250955582,
      "learning_rate": 9.479198361661673e-06,
      "loss": 0.0041,
      "step": 72320
    },
    {
      "epoch": 0.4639021527548399,
      "grad_norm": 0.11141420155763626,
      "learning_rate": 9.47894961406309e-06,
      "loss": 0.003,
      "step": 72330
    },
    {
      "epoch": 0.463966289648626,
      "grad_norm": 0.16782517731189728,
      "learning_rate": 9.478700810340067e-06,
      "loss": 0.0045,
      "step": 72340
    },
    {
      "epoch": 0.4640304265424121,
      "grad_norm": 0.10479758679866791,
      "learning_rate": 9.478451950495725e-06,
      "loss": 0.0034,
      "step": 72350
    },
    {
      "epoch": 0.4640945634361982,
      "grad_norm": 0.1993977129459381,
      "learning_rate": 9.478203034533176e-06,
      "loss": 0.0031,
      "step": 72360
    },
    {
      "epoch": 0.4641587003299843,
      "grad_norm": 0.12023179978132248,
      "learning_rate": 9.477954062455543e-06,
      "loss": 0.0032,
      "step": 72370
    },
    {
      "epoch": 0.4642228372237704,
      "grad_norm": 0.1690799742937088,
      "learning_rate": 9.477705034265945e-06,
      "loss": 0.0033,
      "step": 72380
    },
    {
      "epoch": 0.46428697411755654,
      "grad_norm": 0.15061965584754944,
      "learning_rate": 9.477455949967504e-06,
      "loss": 0.0025,
      "step": 72390
    },
    {
      "epoch": 0.46435111101134263,
      "grad_norm": 0.1160641685128212,
      "learning_rate": 9.47720680956334e-06,
      "loss": 0.0028,
      "step": 72400
    },
    {
      "epoch": 0.4644152479051287,
      "grad_norm": 0.4018762409687042,
      "learning_rate": 9.476957613056574e-06,
      "loss": 0.0042,
      "step": 72410
    },
    {
      "epoch": 0.4644793847989148,
      "grad_norm": 0.22380676865577698,
      "learning_rate": 9.476708360450328e-06,
      "loss": 0.0041,
      "step": 72420
    },
    {
      "epoch": 0.4645435216927009,
      "grad_norm": 0.19054381549358368,
      "learning_rate": 9.476459051747729e-06,
      "loss": 0.004,
      "step": 72430
    },
    {
      "epoch": 0.464607658586487,
      "grad_norm": 0.1574215590953827,
      "learning_rate": 9.476209686951898e-06,
      "loss": 0.0042,
      "step": 72440
    },
    {
      "epoch": 0.4646717954802731,
      "grad_norm": 0.20319758355617523,
      "learning_rate": 9.475960266065962e-06,
      "loss": 0.0053,
      "step": 72450
    },
    {
      "epoch": 0.4647359323740592,
      "grad_norm": 0.25375211238861084,
      "learning_rate": 9.475710789093043e-06,
      "loss": 0.0024,
      "step": 72460
    },
    {
      "epoch": 0.4648000692678453,
      "grad_norm": 0.2655993402004242,
      "learning_rate": 9.475461256036268e-06,
      "loss": 0.006,
      "step": 72470
    },
    {
      "epoch": 0.4648642061616314,
      "grad_norm": 0.05914941430091858,
      "learning_rate": 9.475211666898769e-06,
      "loss": 0.0029,
      "step": 72480
    },
    {
      "epoch": 0.46492834305541747,
      "grad_norm": 0.18190039694309235,
      "learning_rate": 9.474962021683667e-06,
      "loss": 0.0034,
      "step": 72490
    },
    {
      "epoch": 0.46499247994920356,
      "grad_norm": 0.1939253956079483,
      "learning_rate": 9.474712320394092e-06,
      "loss": 0.004,
      "step": 72500
    },
    {
      "epoch": 0.46505661684298966,
      "grad_norm": 0.0900583490729332,
      "learning_rate": 9.474462563033174e-06,
      "loss": 0.0034,
      "step": 72510
    },
    {
      "epoch": 0.46512075373677575,
      "grad_norm": 0.15111422538757324,
      "learning_rate": 9.474212749604044e-06,
      "loss": 0.004,
      "step": 72520
    },
    {
      "epoch": 0.4651848906305619,
      "grad_norm": 0.05144800618290901,
      "learning_rate": 9.473962880109828e-06,
      "loss": 0.0026,
      "step": 72530
    },
    {
      "epoch": 0.465249027524348,
      "grad_norm": 0.0993695855140686,
      "learning_rate": 9.473712954553661e-06,
      "loss": 0.0028,
      "step": 72540
    },
    {
      "epoch": 0.4653131644181341,
      "grad_norm": 0.2637014389038086,
      "learning_rate": 9.473462972938673e-06,
      "loss": 0.0049,
      "step": 72550
    },
    {
      "epoch": 0.4653773013119202,
      "grad_norm": 0.031208178028464317,
      "learning_rate": 9.473212935267997e-06,
      "loss": 0.003,
      "step": 72560
    },
    {
      "epoch": 0.46544143820570627,
      "grad_norm": 0.18791769444942474,
      "learning_rate": 9.472962841544767e-06,
      "loss": 0.0028,
      "step": 72570
    },
    {
      "epoch": 0.46550557509949236,
      "grad_norm": 0.25973379611968994,
      "learning_rate": 9.472712691772114e-06,
      "loss": 0.0031,
      "step": 72580
    },
    {
      "epoch": 0.46556971199327846,
      "grad_norm": 0.4147563874721527,
      "learning_rate": 9.472462485953175e-06,
      "loss": 0.0032,
      "step": 72590
    },
    {
      "epoch": 0.46563384888706455,
      "grad_norm": 0.1889244019985199,
      "learning_rate": 9.472212224091084e-06,
      "loss": 0.0052,
      "step": 72600
    },
    {
      "epoch": 0.46569798578085064,
      "grad_norm": 0.14517726004123688,
      "learning_rate": 9.47196190618898e-06,
      "loss": 0.0022,
      "step": 72610
    },
    {
      "epoch": 0.46576212267463674,
      "grad_norm": 0.18792518973350525,
      "learning_rate": 9.471711532249994e-06,
      "loss": 0.0037,
      "step": 72620
    },
    {
      "epoch": 0.46582625956842283,
      "grad_norm": 0.10088556259870529,
      "learning_rate": 9.471461102277269e-06,
      "loss": 0.0047,
      "step": 72630
    },
    {
      "epoch": 0.4658903964622089,
      "grad_norm": 0.055122584104537964,
      "learning_rate": 9.471210616273941e-06,
      "loss": 0.0042,
      "step": 72640
    },
    {
      "epoch": 0.465954533355995,
      "grad_norm": 0.24400022625923157,
      "learning_rate": 9.470960074243146e-06,
      "loss": 0.0039,
      "step": 72650
    },
    {
      "epoch": 0.4660186702497811,
      "grad_norm": 0.22660301625728607,
      "learning_rate": 9.470709476188027e-06,
      "loss": 0.0022,
      "step": 72660
    },
    {
      "epoch": 0.46608280714356726,
      "grad_norm": 0.2594574987888336,
      "learning_rate": 9.470458822111724e-06,
      "loss": 0.005,
      "step": 72670
    },
    {
      "epoch": 0.46614694403735335,
      "grad_norm": 0.21476195752620697,
      "learning_rate": 9.470208112017376e-06,
      "loss": 0.0092,
      "step": 72680
    },
    {
      "epoch": 0.46621108093113944,
      "grad_norm": 0.1171347126364708,
      "learning_rate": 9.469957345908125e-06,
      "loss": 0.0039,
      "step": 72690
    },
    {
      "epoch": 0.46627521782492554,
      "grad_norm": 0.2394641488790512,
      "learning_rate": 9.469706523787116e-06,
      "loss": 0.0039,
      "step": 72700
    },
    {
      "epoch": 0.46633935471871163,
      "grad_norm": 0.1809658259153366,
      "learning_rate": 9.469455645657488e-06,
      "loss": 0.0035,
      "step": 72710
    },
    {
      "epoch": 0.4664034916124977,
      "grad_norm": 0.0744139775633812,
      "learning_rate": 9.469204711522387e-06,
      "loss": 0.0036,
      "step": 72720
    },
    {
      "epoch": 0.4664676285062838,
      "grad_norm": 0.14177826046943665,
      "learning_rate": 9.468953721384957e-06,
      "loss": 0.0029,
      "step": 72730
    },
    {
      "epoch": 0.4665317654000699,
      "grad_norm": 0.1294241100549698,
      "learning_rate": 9.468702675248342e-06,
      "loss": 0.0029,
      "step": 72740
    },
    {
      "epoch": 0.466595902293856,
      "grad_norm": 0.07392989099025726,
      "learning_rate": 9.46845157311569e-06,
      "loss": 0.0034,
      "step": 72750
    },
    {
      "epoch": 0.4666600391876421,
      "grad_norm": 0.1666719615459442,
      "learning_rate": 9.468200414990147e-06,
      "loss": 0.0026,
      "step": 72760
    },
    {
      "epoch": 0.4667241760814282,
      "grad_norm": 0.22215905785560608,
      "learning_rate": 9.467949200874858e-06,
      "loss": 0.0053,
      "step": 72770
    },
    {
      "epoch": 0.4667883129752143,
      "grad_norm": 0.11765000224113464,
      "learning_rate": 9.467697930772972e-06,
      "loss": 0.0035,
      "step": 72780
    },
    {
      "epoch": 0.4668524498690004,
      "grad_norm": 0.5070384740829468,
      "learning_rate": 9.467446604687639e-06,
      "loss": 0.0048,
      "step": 72790
    },
    {
      "epoch": 0.46691658676278647,
      "grad_norm": 0.20242191851139069,
      "learning_rate": 9.467195222622005e-06,
      "loss": 0.0021,
      "step": 72800
    },
    {
      "epoch": 0.4669807236565726,
      "grad_norm": 0.09090352803468704,
      "learning_rate": 9.466943784579226e-06,
      "loss": 0.0032,
      "step": 72810
    },
    {
      "epoch": 0.4670448605503587,
      "grad_norm": 0.02958475984632969,
      "learning_rate": 9.466692290562445e-06,
      "loss": 0.003,
      "step": 72820
    },
    {
      "epoch": 0.4671089974441448,
      "grad_norm": 0.20340225100517273,
      "learning_rate": 9.46644074057482e-06,
      "loss": 0.0037,
      "step": 72830
    },
    {
      "epoch": 0.4671731343379309,
      "grad_norm": 0.22365257143974304,
      "learning_rate": 9.4661891346195e-06,
      "loss": 0.0046,
      "step": 72840
    },
    {
      "epoch": 0.467237271231717,
      "grad_norm": 0.18572106957435608,
      "learning_rate": 9.465937472699638e-06,
      "loss": 0.0042,
      "step": 72850
    },
    {
      "epoch": 0.4673014081255031,
      "grad_norm": 0.2335018366575241,
      "learning_rate": 9.465685754818387e-06,
      "loss": 0.0038,
      "step": 72860
    },
    {
      "epoch": 0.4673655450192892,
      "grad_norm": 0.15035131573677063,
      "learning_rate": 9.465433980978902e-06,
      "loss": 0.0032,
      "step": 72870
    },
    {
      "epoch": 0.46742968191307527,
      "grad_norm": 0.43644216656684875,
      "learning_rate": 9.465182151184337e-06,
      "loss": 0.0041,
      "step": 72880
    },
    {
      "epoch": 0.46749381880686136,
      "grad_norm": 0.15509265661239624,
      "learning_rate": 9.46493026543785e-06,
      "loss": 0.0032,
      "step": 72890
    },
    {
      "epoch": 0.46755795570064745,
      "grad_norm": 0.16943609714508057,
      "learning_rate": 9.464678323742595e-06,
      "loss": 0.0026,
      "step": 72900
    },
    {
      "epoch": 0.46762209259443355,
      "grad_norm": 0.05730758234858513,
      "learning_rate": 9.46442632610173e-06,
      "loss": 0.0045,
      "step": 72910
    },
    {
      "epoch": 0.46768622948821964,
      "grad_norm": 0.1519549936056137,
      "learning_rate": 9.464174272518414e-06,
      "loss": 0.0036,
      "step": 72920
    },
    {
      "epoch": 0.46775036638200573,
      "grad_norm": 0.2133433073759079,
      "learning_rate": 9.463922162995801e-06,
      "loss": 0.0047,
      "step": 72930
    },
    {
      "epoch": 0.4678145032757918,
      "grad_norm": 0.11244549602270126,
      "learning_rate": 9.463669997537055e-06,
      "loss": 0.004,
      "step": 72940
    },
    {
      "epoch": 0.4678786401695779,
      "grad_norm": 0.2576900124549866,
      "learning_rate": 9.463417776145334e-06,
      "loss": 0.0038,
      "step": 72950
    },
    {
      "epoch": 0.46794277706336407,
      "grad_norm": 0.079642154276371,
      "learning_rate": 9.463165498823797e-06,
      "loss": 0.0071,
      "step": 72960
    },
    {
      "epoch": 0.46800691395715016,
      "grad_norm": 0.24542100727558136,
      "learning_rate": 9.462913165575606e-06,
      "loss": 0.0048,
      "step": 72970
    },
    {
      "epoch": 0.46807105085093625,
      "grad_norm": 0.14651556313037872,
      "learning_rate": 9.462660776403924e-06,
      "loss": 0.0042,
      "step": 72980
    },
    {
      "epoch": 0.46813518774472235,
      "grad_norm": 0.00939899031072855,
      "learning_rate": 9.462408331311914e-06,
      "loss": 0.004,
      "step": 72990
    },
    {
      "epoch": 0.46819932463850844,
      "grad_norm": 0.31788161396980286,
      "learning_rate": 9.462155830302738e-06,
      "loss": 0.003,
      "step": 73000
    },
    {
      "epoch": 0.46826346153229453,
      "grad_norm": 0.15321850776672363,
      "learning_rate": 9.46190327337956e-06,
      "loss": 0.0031,
      "step": 73010
    },
    {
      "epoch": 0.4683275984260806,
      "grad_norm": 0.4166426658630371,
      "learning_rate": 9.461650660545547e-06,
      "loss": 0.0029,
      "step": 73020
    },
    {
      "epoch": 0.4683917353198667,
      "grad_norm": 0.06505458801984787,
      "learning_rate": 9.46139799180386e-06,
      "loss": 0.003,
      "step": 73030
    },
    {
      "epoch": 0.4684558722136528,
      "grad_norm": 0.46712160110473633,
      "learning_rate": 9.46114526715767e-06,
      "loss": 0.0048,
      "step": 73040
    },
    {
      "epoch": 0.4685200091074389,
      "grad_norm": 0.09600020200014114,
      "learning_rate": 9.460892486610138e-06,
      "loss": 0.0046,
      "step": 73050
    },
    {
      "epoch": 0.468584146001225,
      "grad_norm": 0.1537150740623474,
      "learning_rate": 9.460639650164439e-06,
      "loss": 0.0032,
      "step": 73060
    },
    {
      "epoch": 0.4686482828950111,
      "grad_norm": 0.3263317346572876,
      "learning_rate": 9.460386757823734e-06,
      "loss": 0.0053,
      "step": 73070
    },
    {
      "epoch": 0.4687124197887972,
      "grad_norm": 0.17880694568157196,
      "learning_rate": 9.460133809591197e-06,
      "loss": 0.0035,
      "step": 73080
    },
    {
      "epoch": 0.4687765566825833,
      "grad_norm": 0.0824461504817009,
      "learning_rate": 9.459880805469994e-06,
      "loss": 0.0036,
      "step": 73090
    },
    {
      "epoch": 0.4688406935763694,
      "grad_norm": 0.17056573927402496,
      "learning_rate": 9.459627745463298e-06,
      "loss": 0.0053,
      "step": 73100
    },
    {
      "epoch": 0.4689048304701555,
      "grad_norm": 0.6064552664756775,
      "learning_rate": 9.459374629574279e-06,
      "loss": 0.0034,
      "step": 73110
    },
    {
      "epoch": 0.4689689673639416,
      "grad_norm": 0.06469376385211945,
      "learning_rate": 9.45912145780611e-06,
      "loss": 0.0037,
      "step": 73120
    },
    {
      "epoch": 0.4690331042577277,
      "grad_norm": 0.09984765201807022,
      "learning_rate": 9.458868230161962e-06,
      "loss": 0.0059,
      "step": 73130
    },
    {
      "epoch": 0.4690972411515138,
      "grad_norm": 0.1453811526298523,
      "learning_rate": 9.458614946645006e-06,
      "loss": 0.0041,
      "step": 73140
    },
    {
      "epoch": 0.4691613780452999,
      "grad_norm": 0.07419601827859879,
      "learning_rate": 9.45836160725842e-06,
      "loss": 0.0043,
      "step": 73150
    },
    {
      "epoch": 0.469225514939086,
      "grad_norm": 0.2962948679924011,
      "learning_rate": 9.458108212005378e-06,
      "loss": 0.0033,
      "step": 73160
    },
    {
      "epoch": 0.4692896518328721,
      "grad_norm": 0.3422074615955353,
      "learning_rate": 9.457854760889052e-06,
      "loss": 0.0047,
      "step": 73170
    },
    {
      "epoch": 0.46935378872665817,
      "grad_norm": 0.13518819212913513,
      "learning_rate": 9.45760125391262e-06,
      "loss": 0.004,
      "step": 73180
    },
    {
      "epoch": 0.46941792562044427,
      "grad_norm": 0.053319867700338364,
      "learning_rate": 9.457347691079259e-06,
      "loss": 0.002,
      "step": 73190
    },
    {
      "epoch": 0.46948206251423036,
      "grad_norm": 0.11041786521673203,
      "learning_rate": 9.457094072392145e-06,
      "loss": 0.0039,
      "step": 73200
    },
    {
      "epoch": 0.46954619940801645,
      "grad_norm": 0.16768397390842438,
      "learning_rate": 9.456840397854459e-06,
      "loss": 0.0059,
      "step": 73210
    },
    {
      "epoch": 0.46961033630180254,
      "grad_norm": 0.15237963199615479,
      "learning_rate": 9.456586667469376e-06,
      "loss": 0.0045,
      "step": 73220
    },
    {
      "epoch": 0.46967447319558864,
      "grad_norm": 0.12915267050266266,
      "learning_rate": 9.456332881240077e-06,
      "loss": 0.0047,
      "step": 73230
    },
    {
      "epoch": 0.4697386100893748,
      "grad_norm": 0.1947353184223175,
      "learning_rate": 9.456079039169743e-06,
      "loss": 0.0083,
      "step": 73240
    },
    {
      "epoch": 0.4698027469831609,
      "grad_norm": 0.1648906022310257,
      "learning_rate": 9.455825141261552e-06,
      "loss": 0.004,
      "step": 73250
    },
    {
      "epoch": 0.469866883876947,
      "grad_norm": 0.15711960196495056,
      "learning_rate": 9.455571187518689e-06,
      "loss": 0.0019,
      "step": 73260
    },
    {
      "epoch": 0.46993102077073307,
      "grad_norm": 0.12882910668849945,
      "learning_rate": 9.455317177944335e-06,
      "loss": 0.0039,
      "step": 73270
    },
    {
      "epoch": 0.46999515766451916,
      "grad_norm": 0.21856532990932465,
      "learning_rate": 9.455063112541672e-06,
      "loss": 0.0052,
      "step": 73280
    },
    {
      "epoch": 0.47005929455830525,
      "grad_norm": 0.1816015988588333,
      "learning_rate": 9.454808991313883e-06,
      "loss": 0.0029,
      "step": 73290
    },
    {
      "epoch": 0.47012343145209134,
      "grad_norm": 0.05643494427204132,
      "learning_rate": 9.454554814264155e-06,
      "loss": 0.0041,
      "step": 73300
    },
    {
      "epoch": 0.47018756834587744,
      "grad_norm": 0.05270713195204735,
      "learning_rate": 9.45430058139567e-06,
      "loss": 0.0022,
      "step": 73310
    },
    {
      "epoch": 0.47025170523966353,
      "grad_norm": 0.11443763971328735,
      "learning_rate": 9.454046292711617e-06,
      "loss": 0.0027,
      "step": 73320
    },
    {
      "epoch": 0.4703158421334496,
      "grad_norm": 0.40134239196777344,
      "learning_rate": 9.453791948215181e-06,
      "loss": 0.0033,
      "step": 73330
    },
    {
      "epoch": 0.4703799790272357,
      "grad_norm": 0.05452294647693634,
      "learning_rate": 9.453537547909547e-06,
      "loss": 0.0037,
      "step": 73340
    },
    {
      "epoch": 0.4704441159210218,
      "grad_norm": 0.22592099010944366,
      "learning_rate": 9.453283091797905e-06,
      "loss": 0.0029,
      "step": 73350
    },
    {
      "epoch": 0.4705082528148079,
      "grad_norm": 0.12825137376785278,
      "learning_rate": 9.453028579883446e-06,
      "loss": 0.0042,
      "step": 73360
    },
    {
      "epoch": 0.470572389708594,
      "grad_norm": 0.323760062456131,
      "learning_rate": 9.452774012169352e-06,
      "loss": 0.0038,
      "step": 73370
    },
    {
      "epoch": 0.47063652660238015,
      "grad_norm": 0.07360520958900452,
      "learning_rate": 9.45251938865882e-06,
      "loss": 0.0024,
      "step": 73380
    },
    {
      "epoch": 0.47070066349616624,
      "grad_norm": 0.19476427137851715,
      "learning_rate": 9.452264709355037e-06,
      "loss": 0.0031,
      "step": 73390
    },
    {
      "epoch": 0.47076480038995233,
      "grad_norm": 0.11744065582752228,
      "learning_rate": 9.452009974261196e-06,
      "loss": 0.0023,
      "step": 73400
    },
    {
      "epoch": 0.4708289372837384,
      "grad_norm": 0.5189984440803528,
      "learning_rate": 9.451755183380487e-06,
      "loss": 0.0033,
      "step": 73410
    },
    {
      "epoch": 0.4708930741775245,
      "grad_norm": 0.20697681605815887,
      "learning_rate": 9.451500336716106e-06,
      "loss": 0.0029,
      "step": 73420
    },
    {
      "epoch": 0.4709572110713106,
      "grad_norm": 0.1401924192905426,
      "learning_rate": 9.45124543427124e-06,
      "loss": 0.0034,
      "step": 73430
    },
    {
      "epoch": 0.4710213479650967,
      "grad_norm": 0.16114145517349243,
      "learning_rate": 9.450990476049092e-06,
      "loss": 0.0029,
      "step": 73440
    },
    {
      "epoch": 0.4710854848588828,
      "grad_norm": 0.43921709060668945,
      "learning_rate": 9.45073546205285e-06,
      "loss": 0.0028,
      "step": 73450
    },
    {
      "epoch": 0.4711496217526689,
      "grad_norm": 0.09842345863580704,
      "learning_rate": 9.450480392285714e-06,
      "loss": 0.0026,
      "step": 73460
    },
    {
      "epoch": 0.471213758646455,
      "grad_norm": 0.0862642303109169,
      "learning_rate": 9.450225266750877e-06,
      "loss": 0.0023,
      "step": 73470
    },
    {
      "epoch": 0.4712778955402411,
      "grad_norm": 0.09110996127128601,
      "learning_rate": 9.449970085451535e-06,
      "loss": 0.0126,
      "step": 73480
    },
    {
      "epoch": 0.47134203243402717,
      "grad_norm": 0.08711467683315277,
      "learning_rate": 9.449714848390889e-06,
      "loss": 0.0029,
      "step": 73490
    },
    {
      "epoch": 0.47140616932781326,
      "grad_norm": 0.14931470155715942,
      "learning_rate": 9.449459555572135e-06,
      "loss": 0.0018,
      "step": 73500
    },
    {
      "epoch": 0.47147030622159936,
      "grad_norm": 0.3300265967845917,
      "learning_rate": 9.449204206998474e-06,
      "loss": 0.0043,
      "step": 73510
    },
    {
      "epoch": 0.4715344431153855,
      "grad_norm": 0.19805078208446503,
      "learning_rate": 9.448948802673103e-06,
      "loss": 0.0037,
      "step": 73520
    },
    {
      "epoch": 0.4715985800091716,
      "grad_norm": 0.07131168246269226,
      "learning_rate": 9.448693342599225e-06,
      "loss": 0.0028,
      "step": 73530
    },
    {
      "epoch": 0.4716627169029577,
      "grad_norm": 0.14850689470767975,
      "learning_rate": 9.448437826780041e-06,
      "loss": 0.0022,
      "step": 73540
    },
    {
      "epoch": 0.4717268537967438,
      "grad_norm": 0.1711905300617218,
      "learning_rate": 9.44818225521875e-06,
      "loss": 0.0043,
      "step": 73550
    },
    {
      "epoch": 0.4717909906905299,
      "grad_norm": 1.425281286239624,
      "learning_rate": 9.447926627918557e-06,
      "loss": 0.0063,
      "step": 73560
    },
    {
      "epoch": 0.47185512758431597,
      "grad_norm": 0.1310764104127884,
      "learning_rate": 9.447670944882663e-06,
      "loss": 0.0033,
      "step": 73570
    },
    {
      "epoch": 0.47191926447810206,
      "grad_norm": 0.24520087242126465,
      "learning_rate": 9.447415206114275e-06,
      "loss": 0.0046,
      "step": 73580
    },
    {
      "epoch": 0.47198340137188816,
      "grad_norm": 0.2960132956504822,
      "learning_rate": 9.447159411616595e-06,
      "loss": 0.0069,
      "step": 73590
    },
    {
      "epoch": 0.47204753826567425,
      "grad_norm": 0.12461918592453003,
      "learning_rate": 9.44690356139283e-06,
      "loss": 0.0038,
      "step": 73600
    },
    {
      "epoch": 0.47211167515946034,
      "grad_norm": 0.08798740059137344,
      "learning_rate": 9.446647655446186e-06,
      "loss": 0.0036,
      "step": 73610
    },
    {
      "epoch": 0.47217581205324644,
      "grad_norm": 0.05591832101345062,
      "learning_rate": 9.446391693779868e-06,
      "loss": 0.0087,
      "step": 73620
    },
    {
      "epoch": 0.47223994894703253,
      "grad_norm": 0.13037869334220886,
      "learning_rate": 9.446135676397084e-06,
      "loss": 0.0032,
      "step": 73630
    },
    {
      "epoch": 0.4723040858408186,
      "grad_norm": 0.20414581894874573,
      "learning_rate": 9.445879603301043e-06,
      "loss": 0.0044,
      "step": 73640
    },
    {
      "epoch": 0.4723682227346047,
      "grad_norm": 0.2094263732433319,
      "learning_rate": 9.445623474494951e-06,
      "loss": 0.0052,
      "step": 73650
    },
    {
      "epoch": 0.47243235962839086,
      "grad_norm": 0.08255743235349655,
      "learning_rate": 9.445367289982022e-06,
      "loss": 0.0033,
      "step": 73660
    },
    {
      "epoch": 0.47249649652217696,
      "grad_norm": 0.024607084691524506,
      "learning_rate": 9.445111049765463e-06,
      "loss": 0.0033,
      "step": 73670
    },
    {
      "epoch": 0.47256063341596305,
      "grad_norm": 0.029220154508948326,
      "learning_rate": 9.444854753848485e-06,
      "loss": 0.0034,
      "step": 73680
    },
    {
      "epoch": 0.47262477030974914,
      "grad_norm": 0.13529035449028015,
      "learning_rate": 9.444598402234302e-06,
      "loss": 0.0028,
      "step": 73690
    },
    {
      "epoch": 0.47268890720353524,
      "grad_norm": 0.03135405853390694,
      "learning_rate": 9.444341994926122e-06,
      "loss": 0.0059,
      "step": 73700
    },
    {
      "epoch": 0.47275304409732133,
      "grad_norm": 0.2813445031642914,
      "learning_rate": 9.444085531927162e-06,
      "loss": 0.0031,
      "step": 73710
    },
    {
      "epoch": 0.4728171809911074,
      "grad_norm": 0.2533155679702759,
      "learning_rate": 9.443829013240635e-06,
      "loss": 0.0041,
      "step": 73720
    },
    {
      "epoch": 0.4728813178848935,
      "grad_norm": 0.5630262494087219,
      "learning_rate": 9.443572438869754e-06,
      "loss": 0.0036,
      "step": 73730
    },
    {
      "epoch": 0.4729454547786796,
      "grad_norm": 0.27068030834198,
      "learning_rate": 9.443315808817735e-06,
      "loss": 0.0047,
      "step": 73740
    },
    {
      "epoch": 0.4730095916724657,
      "grad_norm": 0.2189641296863556,
      "learning_rate": 9.443059123087793e-06,
      "loss": 0.0023,
      "step": 73750
    },
    {
      "epoch": 0.4730737285662518,
      "grad_norm": 0.09532841295003891,
      "learning_rate": 9.442802381683144e-06,
      "loss": 0.0029,
      "step": 73760
    },
    {
      "epoch": 0.4731378654600379,
      "grad_norm": 0.10535168647766113,
      "learning_rate": 9.442545584607005e-06,
      "loss": 0.0033,
      "step": 73770
    },
    {
      "epoch": 0.473202002353824,
      "grad_norm": 0.11029849946498871,
      "learning_rate": 9.442288731862597e-06,
      "loss": 0.0037,
      "step": 73780
    },
    {
      "epoch": 0.4732661392476101,
      "grad_norm": 0.0792105495929718,
      "learning_rate": 9.442031823453134e-06,
      "loss": 0.0025,
      "step": 73790
    },
    {
      "epoch": 0.4733302761413962,
      "grad_norm": 0.10897326469421387,
      "learning_rate": 9.441774859381841e-06,
      "loss": 0.003,
      "step": 73800
    },
    {
      "epoch": 0.4733944130351823,
      "grad_norm": 0.05002043768763542,
      "learning_rate": 9.441517839651932e-06,
      "loss": 0.0034,
      "step": 73810
    },
    {
      "epoch": 0.4734585499289684,
      "grad_norm": 0.039975326508283615,
      "learning_rate": 9.441260764266632e-06,
      "loss": 0.0036,
      "step": 73820
    },
    {
      "epoch": 0.4735226868227545,
      "grad_norm": 0.14692874252796173,
      "learning_rate": 9.441003633229158e-06,
      "loss": 0.0061,
      "step": 73830
    },
    {
      "epoch": 0.4735868237165406,
      "grad_norm": 0.31991007924079895,
      "learning_rate": 9.440746446542736e-06,
      "loss": 0.0031,
      "step": 73840
    },
    {
      "epoch": 0.4736509606103267,
      "grad_norm": 0.1329358071088791,
      "learning_rate": 9.440489204210588e-06,
      "loss": 0.0038,
      "step": 73850
    },
    {
      "epoch": 0.4737150975041128,
      "grad_norm": 0.16461928188800812,
      "learning_rate": 9.440231906235935e-06,
      "loss": 0.0035,
      "step": 73860
    },
    {
      "epoch": 0.4737792343978989,
      "grad_norm": 0.13529208302497864,
      "learning_rate": 9.439974552622003e-06,
      "loss": 0.0042,
      "step": 73870
    },
    {
      "epoch": 0.47384337129168497,
      "grad_norm": 0.4698379337787628,
      "learning_rate": 9.439717143372017e-06,
      "loss": 0.0074,
      "step": 73880
    },
    {
      "epoch": 0.47390750818547106,
      "grad_norm": 0.2502039968967438,
      "learning_rate": 9.439459678489203e-06,
      "loss": 0.0037,
      "step": 73890
    },
    {
      "epoch": 0.47397164507925715,
      "grad_norm": 0.06690411269664764,
      "learning_rate": 9.439202157976786e-06,
      "loss": 0.0031,
      "step": 73900
    },
    {
      "epoch": 0.47403578197304325,
      "grad_norm": 0.27009642124176025,
      "learning_rate": 9.438944581837993e-06,
      "loss": 0.0027,
      "step": 73910
    },
    {
      "epoch": 0.47409991886682934,
      "grad_norm": 0.12579452991485596,
      "learning_rate": 9.438686950076052e-06,
      "loss": 0.0061,
      "step": 73920
    },
    {
      "epoch": 0.47416405576061543,
      "grad_norm": 0.2248860001564026,
      "learning_rate": 9.43842926269419e-06,
      "loss": 0.0099,
      "step": 73930
    },
    {
      "epoch": 0.4742281926544016,
      "grad_norm": 0.23569662868976593,
      "learning_rate": 9.43817151969564e-06,
      "loss": 0.0067,
      "step": 73940
    },
    {
      "epoch": 0.4742923295481877,
      "grad_norm": 0.055896393954753876,
      "learning_rate": 9.437913721083628e-06,
      "loss": 0.0024,
      "step": 73950
    },
    {
      "epoch": 0.47435646644197377,
      "grad_norm": 0.043094929307699203,
      "learning_rate": 9.437655866861383e-06,
      "loss": 0.0037,
      "step": 73960
    },
    {
      "epoch": 0.47442060333575986,
      "grad_norm": 0.07713795453310013,
      "learning_rate": 9.437397957032141e-06,
      "loss": 0.0027,
      "step": 73970
    },
    {
      "epoch": 0.47448474022954595,
      "grad_norm": 0.1753903180360794,
      "learning_rate": 9.43713999159913e-06,
      "loss": 0.0033,
      "step": 73980
    },
    {
      "epoch": 0.47454887712333205,
      "grad_norm": 0.0889492928981781,
      "learning_rate": 9.436881970565583e-06,
      "loss": 0.0033,
      "step": 73990
    },
    {
      "epoch": 0.47461301401711814,
      "grad_norm": 0.14199770987033844,
      "learning_rate": 9.436623893934735e-06,
      "loss": 0.0031,
      "step": 74000
    },
    {
      "epoch": 0.47467715091090423,
      "grad_norm": 0.39867103099823,
      "learning_rate": 9.43636576170982e-06,
      "loss": 0.0052,
      "step": 74010
    },
    {
      "epoch": 0.4747412878046903,
      "grad_norm": 0.11389205604791641,
      "learning_rate": 9.43610757389407e-06,
      "loss": 0.002,
      "step": 74020
    },
    {
      "epoch": 0.4748054246984764,
      "grad_norm": 0.2963626980781555,
      "learning_rate": 9.435849330490722e-06,
      "loss": 0.0049,
      "step": 74030
    },
    {
      "epoch": 0.4748695615922625,
      "grad_norm": 0.2825103998184204,
      "learning_rate": 9.43559103150301e-06,
      "loss": 0.0045,
      "step": 74040
    },
    {
      "epoch": 0.4749336984860486,
      "grad_norm": 0.17604900896549225,
      "learning_rate": 9.435332676934174e-06,
      "loss": 0.003,
      "step": 74050
    },
    {
      "epoch": 0.4749978353798347,
      "grad_norm": 0.16739438474178314,
      "learning_rate": 9.435074266787451e-06,
      "loss": 0.0032,
      "step": 74060
    },
    {
      "epoch": 0.4750619722736208,
      "grad_norm": 0.27554914355278015,
      "learning_rate": 9.434815801066076e-06,
      "loss": 0.0031,
      "step": 74070
    },
    {
      "epoch": 0.47512610916740694,
      "grad_norm": 0.09619660675525665,
      "learning_rate": 9.43455727977329e-06,
      "loss": 0.0045,
      "step": 74080
    },
    {
      "epoch": 0.47519024606119303,
      "grad_norm": 0.29939550161361694,
      "learning_rate": 9.434298702912333e-06,
      "loss": 0.0042,
      "step": 74090
    },
    {
      "epoch": 0.4752543829549791,
      "grad_norm": 0.09388145804405212,
      "learning_rate": 9.43404007048644e-06,
      "loss": 0.0039,
      "step": 74100
    },
    {
      "epoch": 0.4753185198487652,
      "grad_norm": 0.16488589346408844,
      "learning_rate": 9.433781382498862e-06,
      "loss": 0.0045,
      "step": 74110
    },
    {
      "epoch": 0.4753826567425513,
      "grad_norm": 0.0874987542629242,
      "learning_rate": 9.43352263895283e-06,
      "loss": 0.0036,
      "step": 74120
    },
    {
      "epoch": 0.4754467936363374,
      "grad_norm": 0.35496601462364197,
      "learning_rate": 9.433263839851592e-06,
      "loss": 0.0032,
      "step": 74130
    },
    {
      "epoch": 0.4755109305301235,
      "grad_norm": 0.25578826665878296,
      "learning_rate": 9.43300498519839e-06,
      "loss": 0.0023,
      "step": 74140
    },
    {
      "epoch": 0.4755750674239096,
      "grad_norm": 0.13040082156658173,
      "learning_rate": 9.432746074996466e-06,
      "loss": 0.003,
      "step": 74150
    },
    {
      "epoch": 0.4756392043176957,
      "grad_norm": 0.10859714448451996,
      "learning_rate": 9.432487109249067e-06,
      "loss": 0.0028,
      "step": 74160
    },
    {
      "epoch": 0.4757033412114818,
      "grad_norm": 0.1338357925415039,
      "learning_rate": 9.432228087959436e-06,
      "loss": 0.0026,
      "step": 74170
    },
    {
      "epoch": 0.47576747810526787,
      "grad_norm": 0.2157885879278183,
      "learning_rate": 9.43196901113082e-06,
      "loss": 0.0032,
      "step": 74180
    },
    {
      "epoch": 0.47583161499905396,
      "grad_norm": 1.2301640510559082,
      "learning_rate": 9.431709878766465e-06,
      "loss": 0.0056,
      "step": 74190
    },
    {
      "epoch": 0.47589575189284006,
      "grad_norm": 0.0766885057091713,
      "learning_rate": 9.431450690869617e-06,
      "loss": 0.0068,
      "step": 74200
    },
    {
      "epoch": 0.47595988878662615,
      "grad_norm": 0.26054835319519043,
      "learning_rate": 9.431191447443526e-06,
      "loss": 0.0028,
      "step": 74210
    },
    {
      "epoch": 0.4760240256804123,
      "grad_norm": 0.10278470069169998,
      "learning_rate": 9.430932148491439e-06,
      "loss": 0.0044,
      "step": 74220
    },
    {
      "epoch": 0.4760881625741984,
      "grad_norm": 0.16061940789222717,
      "learning_rate": 9.430672794016604e-06,
      "loss": 0.0051,
      "step": 74230
    },
    {
      "epoch": 0.4761522994679845,
      "grad_norm": 0.16396574676036835,
      "learning_rate": 9.430413384022273e-06,
      "loss": 0.0049,
      "step": 74240
    },
    {
      "epoch": 0.4762164363617706,
      "grad_norm": 0.07798841595649719,
      "learning_rate": 9.430153918511696e-06,
      "loss": 0.0044,
      "step": 74250
    },
    {
      "epoch": 0.47628057325555667,
      "grad_norm": 0.07246286422014236,
      "learning_rate": 9.429894397488125e-06,
      "loss": 0.0029,
      "step": 74260
    },
    {
      "epoch": 0.47634471014934276,
      "grad_norm": 0.3123680055141449,
      "learning_rate": 9.42963482095481e-06,
      "loss": 0.0024,
      "step": 74270
    },
    {
      "epoch": 0.47640884704312886,
      "grad_norm": 0.27817103266716003,
      "learning_rate": 9.429375188915007e-06,
      "loss": 0.0035,
      "step": 74280
    },
    {
      "epoch": 0.47647298393691495,
      "grad_norm": 0.179836243391037,
      "learning_rate": 9.429115501371963e-06,
      "loss": 0.0029,
      "step": 74290
    },
    {
      "epoch": 0.47653712083070104,
      "grad_norm": 0.04475254565477371,
      "learning_rate": 9.42885575832894e-06,
      "loss": 0.0033,
      "step": 74300
    },
    {
      "epoch": 0.47660125772448714,
      "grad_norm": 0.1160324290394783,
      "learning_rate": 9.428595959789188e-06,
      "loss": 0.0029,
      "step": 74310
    },
    {
      "epoch": 0.47666539461827323,
      "grad_norm": 0.11190337687730789,
      "learning_rate": 9.428336105755964e-06,
      "loss": 0.0053,
      "step": 74320
    },
    {
      "epoch": 0.4767295315120593,
      "grad_norm": 0.16089123487472534,
      "learning_rate": 9.428076196232522e-06,
      "loss": 0.0029,
      "step": 74330
    },
    {
      "epoch": 0.4767936684058454,
      "grad_norm": 0.25040191411972046,
      "learning_rate": 9.427816231222124e-06,
      "loss": 0.0042,
      "step": 74340
    },
    {
      "epoch": 0.4768578052996315,
      "grad_norm": 0.07773885130882263,
      "learning_rate": 9.427556210728022e-06,
      "loss": 0.0028,
      "step": 74350
    },
    {
      "epoch": 0.4769219421934176,
      "grad_norm": 0.13476145267486572,
      "learning_rate": 9.427296134753477e-06,
      "loss": 0.0021,
      "step": 74360
    },
    {
      "epoch": 0.47698607908720375,
      "grad_norm": 0.008924542926251888,
      "learning_rate": 9.427036003301746e-06,
      "loss": 0.004,
      "step": 74370
    },
    {
      "epoch": 0.47705021598098984,
      "grad_norm": 0.09451120346784592,
      "learning_rate": 9.42677581637609e-06,
      "loss": 0.0037,
      "step": 74380
    },
    {
      "epoch": 0.47711435287477594,
      "grad_norm": 0.14821767807006836,
      "learning_rate": 9.42651557397977e-06,
      "loss": 0.003,
      "step": 74390
    },
    {
      "epoch": 0.47717848976856203,
      "grad_norm": 0.1648290902376175,
      "learning_rate": 9.426255276116046e-06,
      "loss": 0.0042,
      "step": 74400
    },
    {
      "epoch": 0.4772426266623481,
      "grad_norm": 0.16127754747867584,
      "learning_rate": 9.42599492278818e-06,
      "loss": 0.0045,
      "step": 74410
    },
    {
      "epoch": 0.4773067635561342,
      "grad_norm": 0.03292185813188553,
      "learning_rate": 9.425734513999435e-06,
      "loss": 0.0033,
      "step": 74420
    },
    {
      "epoch": 0.4773709004499203,
      "grad_norm": 0.13805095851421356,
      "learning_rate": 9.425474049753074e-06,
      "loss": 0.0029,
      "step": 74430
    },
    {
      "epoch": 0.4774350373437064,
      "grad_norm": 0.06245690584182739,
      "learning_rate": 9.425213530052359e-06,
      "loss": 0.0032,
      "step": 74440
    },
    {
      "epoch": 0.4774991742374925,
      "grad_norm": 0.08769892156124115,
      "learning_rate": 9.424952954900556e-06,
      "loss": 0.0025,
      "step": 74450
    },
    {
      "epoch": 0.4775633111312786,
      "grad_norm": 0.08083859831094742,
      "learning_rate": 9.42469232430093e-06,
      "loss": 0.0042,
      "step": 74460
    },
    {
      "epoch": 0.4776274480250647,
      "grad_norm": 0.16766510903835297,
      "learning_rate": 9.424431638256746e-06,
      "loss": 0.0025,
      "step": 74470
    },
    {
      "epoch": 0.4776915849188508,
      "grad_norm": 0.27961698174476624,
      "learning_rate": 9.424170896771274e-06,
      "loss": 0.0027,
      "step": 74480
    },
    {
      "epoch": 0.47775572181263687,
      "grad_norm": 0.06486645340919495,
      "learning_rate": 9.423910099847777e-06,
      "loss": 0.0031,
      "step": 74490
    },
    {
      "epoch": 0.47781985870642296,
      "grad_norm": 0.20511706173419952,
      "learning_rate": 9.423649247489525e-06,
      "loss": 0.0045,
      "step": 74500
    },
    {
      "epoch": 0.4778839956002091,
      "grad_norm": 0.21036206185817719,
      "learning_rate": 9.423388339699786e-06,
      "loss": 0.0029,
      "step": 74510
    },
    {
      "epoch": 0.4779481324939952,
      "grad_norm": 0.41878727078437805,
      "learning_rate": 9.42312737648183e-06,
      "loss": 0.0033,
      "step": 74520
    },
    {
      "epoch": 0.4780122693877813,
      "grad_norm": 0.17183734476566315,
      "learning_rate": 9.422866357838928e-06,
      "loss": 0.0027,
      "step": 74530
    },
    {
      "epoch": 0.4780764062815674,
      "grad_norm": 0.0930318757891655,
      "learning_rate": 9.422605283774349e-06,
      "loss": 0.0028,
      "step": 74540
    },
    {
      "epoch": 0.4781405431753535,
      "grad_norm": 0.23828619718551636,
      "learning_rate": 9.422344154291364e-06,
      "loss": 0.0047,
      "step": 74550
    },
    {
      "epoch": 0.4782046800691396,
      "grad_norm": 0.1259741336107254,
      "learning_rate": 9.422082969393246e-06,
      "loss": 0.0027,
      "step": 74560
    },
    {
      "epoch": 0.47826881696292567,
      "grad_norm": 0.07219063490629196,
      "learning_rate": 9.42182172908327e-06,
      "loss": 0.0028,
      "step": 74570
    },
    {
      "epoch": 0.47833295385671176,
      "grad_norm": 0.33875951170921326,
      "learning_rate": 9.421560433364706e-06,
      "loss": 0.0039,
      "step": 74580
    },
    {
      "epoch": 0.47839709075049786,
      "grad_norm": 0.03250084072351456,
      "learning_rate": 9.42129908224083e-06,
      "loss": 0.0018,
      "step": 74590
    },
    {
      "epoch": 0.47846122764428395,
      "grad_norm": 0.10010303556919098,
      "learning_rate": 9.421037675714917e-06,
      "loss": 0.003,
      "step": 74600
    },
    {
      "epoch": 0.47852536453807004,
      "grad_norm": 0.2187255173921585,
      "learning_rate": 9.42077621379024e-06,
      "loss": 0.0033,
      "step": 74610
    },
    {
      "epoch": 0.47858950143185613,
      "grad_norm": 0.007869267836213112,
      "learning_rate": 9.42051469647008e-06,
      "loss": 0.0048,
      "step": 74620
    },
    {
      "epoch": 0.47865363832564223,
      "grad_norm": 0.25898319482803345,
      "learning_rate": 9.420253123757712e-06,
      "loss": 0.0024,
      "step": 74630
    },
    {
      "epoch": 0.4787177752194283,
      "grad_norm": 0.1980488896369934,
      "learning_rate": 9.419991495656411e-06,
      "loss": 0.0048,
      "step": 74640
    },
    {
      "epoch": 0.47878191211321447,
      "grad_norm": 0.15262740850448608,
      "learning_rate": 9.41972981216946e-06,
      "loss": 0.0039,
      "step": 74650
    },
    {
      "epoch": 0.47884604900700056,
      "grad_norm": 0.1269340068101883,
      "learning_rate": 9.419468073300135e-06,
      "loss": 0.0037,
      "step": 74660
    },
    {
      "epoch": 0.47891018590078666,
      "grad_norm": 0.0766548290848732,
      "learning_rate": 9.419206279051716e-06,
      "loss": 0.0038,
      "step": 74670
    },
    {
      "epoch": 0.47897432279457275,
      "grad_norm": 0.11249018460512161,
      "learning_rate": 9.418944429427484e-06,
      "loss": 0.0031,
      "step": 74680
    },
    {
      "epoch": 0.47903845968835884,
      "grad_norm": 0.05062698945403099,
      "learning_rate": 9.41868252443072e-06,
      "loss": 0.0034,
      "step": 74690
    },
    {
      "epoch": 0.47910259658214494,
      "grad_norm": 0.15498970448970795,
      "learning_rate": 9.418420564064706e-06,
      "loss": 0.0026,
      "step": 74700
    },
    {
      "epoch": 0.47916673347593103,
      "grad_norm": 0.5996679067611694,
      "learning_rate": 9.418158548332725e-06,
      "loss": 0.0061,
      "step": 74710
    },
    {
      "epoch": 0.4792308703697171,
      "grad_norm": 0.2909507155418396,
      "learning_rate": 9.41789647723806e-06,
      "loss": 0.0026,
      "step": 74720
    },
    {
      "epoch": 0.4792950072635032,
      "grad_norm": 0.05826156586408615,
      "learning_rate": 9.417634350783993e-06,
      "loss": 0.0029,
      "step": 74730
    },
    {
      "epoch": 0.4793591441572893,
      "grad_norm": 0.20538924634456635,
      "learning_rate": 9.417372168973812e-06,
      "loss": 0.0032,
      "step": 74740
    },
    {
      "epoch": 0.4794232810510754,
      "grad_norm": 0.3443412482738495,
      "learning_rate": 9.417109931810799e-06,
      "loss": 0.0035,
      "step": 74750
    },
    {
      "epoch": 0.4794874179448615,
      "grad_norm": 0.08114752173423767,
      "learning_rate": 9.416847639298244e-06,
      "loss": 0.007,
      "step": 74760
    },
    {
      "epoch": 0.4795515548386476,
      "grad_norm": 0.031641677021980286,
      "learning_rate": 9.41658529143943e-06,
      "loss": 0.0033,
      "step": 74770
    },
    {
      "epoch": 0.4796156917324337,
      "grad_norm": 0.19229751825332642,
      "learning_rate": 9.416322888237646e-06,
      "loss": 0.0041,
      "step": 74780
    },
    {
      "epoch": 0.47967982862621983,
      "grad_norm": 0.18572895228862762,
      "learning_rate": 9.41606042969618e-06,
      "loss": 0.005,
      "step": 74790
    },
    {
      "epoch": 0.4797439655200059,
      "grad_norm": 0.07986016571521759,
      "learning_rate": 9.415797915818322e-06,
      "loss": 0.0077,
      "step": 74800
    },
    {
      "epoch": 0.479808102413792,
      "grad_norm": 0.07696294039487839,
      "learning_rate": 9.415535346607358e-06,
      "loss": 0.006,
      "step": 74810
    },
    {
      "epoch": 0.4798722393075781,
      "grad_norm": 0.15090428292751312,
      "learning_rate": 9.415272722066581e-06,
      "loss": 0.0034,
      "step": 74820
    },
    {
      "epoch": 0.4799363762013642,
      "grad_norm": 0.12085235863924026,
      "learning_rate": 9.415010042199283e-06,
      "loss": 0.0036,
      "step": 74830
    },
    {
      "epoch": 0.4800005130951503,
      "grad_norm": 0.5605502724647522,
      "learning_rate": 9.414747307008752e-06,
      "loss": 0.0059,
      "step": 74840
    },
    {
      "epoch": 0.4800646499889364,
      "grad_norm": 0.16201697289943695,
      "learning_rate": 9.414484516498281e-06,
      "loss": 0.0053,
      "step": 74850
    },
    {
      "epoch": 0.4801287868827225,
      "grad_norm": 0.14369170367717743,
      "learning_rate": 9.414221670671167e-06,
      "loss": 0.0021,
      "step": 74860
    },
    {
      "epoch": 0.4801929237765086,
      "grad_norm": 0.16418346762657166,
      "learning_rate": 9.413958769530698e-06,
      "loss": 0.0034,
      "step": 74870
    },
    {
      "epoch": 0.48025706067029467,
      "grad_norm": 0.46877381205558777,
      "learning_rate": 9.413695813080173e-06,
      "loss": 0.0034,
      "step": 74880
    },
    {
      "epoch": 0.48032119756408076,
      "grad_norm": 0.16352395713329315,
      "learning_rate": 9.413432801322883e-06,
      "loss": 0.0035,
      "step": 74890
    },
    {
      "epoch": 0.48038533445786685,
      "grad_norm": 0.25804394483566284,
      "learning_rate": 9.413169734262128e-06,
      "loss": 0.0029,
      "step": 74900
    },
    {
      "epoch": 0.48044947135165295,
      "grad_norm": 0.3428088426589966,
      "learning_rate": 9.4129066119012e-06,
      "loss": 0.0034,
      "step": 74910
    },
    {
      "epoch": 0.48051360824543904,
      "grad_norm": 0.2453552633523941,
      "learning_rate": 9.4126434342434e-06,
      "loss": 0.0027,
      "step": 74920
    },
    {
      "epoch": 0.4805777451392252,
      "grad_norm": 0.5071871876716614,
      "learning_rate": 9.412380201292023e-06,
      "loss": 0.0028,
      "step": 74930
    },
    {
      "epoch": 0.4806418820330113,
      "grad_norm": 0.40177619457244873,
      "learning_rate": 9.412116913050371e-06,
      "loss": 0.0055,
      "step": 74940
    },
    {
      "epoch": 0.4807060189267974,
      "grad_norm": 0.3960021734237671,
      "learning_rate": 9.411853569521738e-06,
      "loss": 0.0039,
      "step": 74950
    },
    {
      "epoch": 0.48077015582058347,
      "grad_norm": 0.22092705965042114,
      "learning_rate": 9.411590170709429e-06,
      "loss": 0.0065,
      "step": 74960
    },
    {
      "epoch": 0.48083429271436956,
      "grad_norm": 0.0896230936050415,
      "learning_rate": 9.41132671661674e-06,
      "loss": 0.0038,
      "step": 74970
    },
    {
      "epoch": 0.48089842960815565,
      "grad_norm": 0.21776001155376434,
      "learning_rate": 9.411063207246976e-06,
      "loss": 0.0044,
      "step": 74980
    },
    {
      "epoch": 0.48096256650194175,
      "grad_norm": 0.06775982677936554,
      "learning_rate": 9.410799642603435e-06,
      "loss": 0.0034,
      "step": 74990
    },
    {
      "epoch": 0.48102670339572784,
      "grad_norm": 0.13864344358444214,
      "learning_rate": 9.410536022689425e-06,
      "loss": 0.0019,
      "step": 75000
    },
    {
      "epoch": 0.48109084028951393,
      "grad_norm": 0.1059882789850235,
      "learning_rate": 9.410272347508245e-06,
      "loss": 0.0035,
      "step": 75010
    },
    {
      "epoch": 0.4811549771833,
      "grad_norm": 0.11833040416240692,
      "learning_rate": 9.4100086170632e-06,
      "loss": 0.0044,
      "step": 75020
    },
    {
      "epoch": 0.4812191140770861,
      "grad_norm": 0.13698428869247437,
      "learning_rate": 9.409744831357597e-06,
      "loss": 0.0041,
      "step": 75030
    },
    {
      "epoch": 0.4812832509708722,
      "grad_norm": 0.05028630048036575,
      "learning_rate": 9.40948099039474e-06,
      "loss": 0.0035,
      "step": 75040
    },
    {
      "epoch": 0.4813473878646583,
      "grad_norm": 0.10551158338785172,
      "learning_rate": 9.409217094177932e-06,
      "loss": 0.0038,
      "step": 75050
    },
    {
      "epoch": 0.4814115247584444,
      "grad_norm": 0.08216790854930878,
      "learning_rate": 9.408953142710483e-06,
      "loss": 0.0029,
      "step": 75060
    },
    {
      "epoch": 0.48147566165223055,
      "grad_norm": 0.2770686447620392,
      "learning_rate": 9.408689135995704e-06,
      "loss": 0.0057,
      "step": 75070
    },
    {
      "epoch": 0.48153979854601664,
      "grad_norm": 0.07415860891342163,
      "learning_rate": 9.408425074036895e-06,
      "loss": 0.0029,
      "step": 75080
    },
    {
      "epoch": 0.48160393543980273,
      "grad_norm": 0.24299880862236023,
      "learning_rate": 9.408160956837372e-06,
      "loss": 0.0026,
      "step": 75090
    },
    {
      "epoch": 0.4816680723335888,
      "grad_norm": 0.17356914281845093,
      "learning_rate": 9.40789678440044e-06,
      "loss": 0.0026,
      "step": 75100
    },
    {
      "epoch": 0.4817322092273749,
      "grad_norm": 0.23093180358409882,
      "learning_rate": 9.407632556729412e-06,
      "loss": 0.0066,
      "step": 75110
    },
    {
      "epoch": 0.481796346121161,
      "grad_norm": 0.07763124257326126,
      "learning_rate": 9.407368273827598e-06,
      "loss": 0.0042,
      "step": 75120
    },
    {
      "epoch": 0.4818604830149471,
      "grad_norm": 0.12229952216148376,
      "learning_rate": 9.407103935698308e-06,
      "loss": 0.0051,
      "step": 75130
    },
    {
      "epoch": 0.4819246199087332,
      "grad_norm": 0.033002957701683044,
      "learning_rate": 9.406839542344857e-06,
      "loss": 0.0037,
      "step": 75140
    },
    {
      "epoch": 0.4819887568025193,
      "grad_norm": 0.0705639198422432,
      "learning_rate": 9.406575093770558e-06,
      "loss": 0.0025,
      "step": 75150
    },
    {
      "epoch": 0.4820528936963054,
      "grad_norm": 0.06316683441400528,
      "learning_rate": 9.406310589978721e-06,
      "loss": 0.0027,
      "step": 75160
    },
    {
      "epoch": 0.4821170305900915,
      "grad_norm": 0.1080327183008194,
      "learning_rate": 9.406046030972666e-06,
      "loss": 0.0038,
      "step": 75170
    },
    {
      "epoch": 0.48218116748387757,
      "grad_norm": 0.18605861067771912,
      "learning_rate": 9.405781416755704e-06,
      "loss": 0.0031,
      "step": 75180
    },
    {
      "epoch": 0.48224530437766366,
      "grad_norm": 0.14519374072551727,
      "learning_rate": 9.405516747331152e-06,
      "loss": 0.0022,
      "step": 75190
    },
    {
      "epoch": 0.48230944127144976,
      "grad_norm": 0.22531086206436157,
      "learning_rate": 9.405252022702328e-06,
      "loss": 0.0036,
      "step": 75200
    },
    {
      "epoch": 0.4823735781652359,
      "grad_norm": 0.13491190969944,
      "learning_rate": 9.404987242872547e-06,
      "loss": 0.0045,
      "step": 75210
    },
    {
      "epoch": 0.482437715059022,
      "grad_norm": 0.2310151308774948,
      "learning_rate": 9.404722407845128e-06,
      "loss": 0.0052,
      "step": 75220
    },
    {
      "epoch": 0.4825018519528081,
      "grad_norm": 0.0630815178155899,
      "learning_rate": 9.404457517623388e-06,
      "loss": 0.0038,
      "step": 75230
    },
    {
      "epoch": 0.4825659888465942,
      "grad_norm": 0.033768199384212494,
      "learning_rate": 9.404192572210648e-06,
      "loss": 0.0041,
      "step": 75240
    },
    {
      "epoch": 0.4826301257403803,
      "grad_norm": 0.07099565863609314,
      "learning_rate": 9.403927571610227e-06,
      "loss": 0.0037,
      "step": 75250
    },
    {
      "epoch": 0.48269426263416637,
      "grad_norm": 0.17977134883403778,
      "learning_rate": 9.403662515825447e-06,
      "loss": 0.003,
      "step": 75260
    },
    {
      "epoch": 0.48275839952795246,
      "grad_norm": 0.05882582440972328,
      "learning_rate": 9.403397404859628e-06,
      "loss": 0.0053,
      "step": 75270
    },
    {
      "epoch": 0.48282253642173856,
      "grad_norm": 0.29024794697761536,
      "learning_rate": 9.403132238716093e-06,
      "loss": 0.0035,
      "step": 75280
    },
    {
      "epoch": 0.48288667331552465,
      "grad_norm": 0.23191240429878235,
      "learning_rate": 9.402867017398163e-06,
      "loss": 0.0035,
      "step": 75290
    },
    {
      "epoch": 0.48295081020931074,
      "grad_norm": 0.24628372490406036,
      "learning_rate": 9.402601740909163e-06,
      "loss": 0.0036,
      "step": 75300
    },
    {
      "epoch": 0.48301494710309684,
      "grad_norm": 0.3608076572418213,
      "learning_rate": 9.402336409252417e-06,
      "loss": 0.0066,
      "step": 75310
    },
    {
      "epoch": 0.48307908399688293,
      "grad_norm": 0.0643908828496933,
      "learning_rate": 9.40207102243125e-06,
      "loss": 0.0038,
      "step": 75320
    },
    {
      "epoch": 0.483143220890669,
      "grad_norm": 0.21337231993675232,
      "learning_rate": 9.401805580448986e-06,
      "loss": 0.0031,
      "step": 75330
    },
    {
      "epoch": 0.4832073577844551,
      "grad_norm": 0.15173020958900452,
      "learning_rate": 9.401540083308954e-06,
      "loss": 0.0024,
      "step": 75340
    },
    {
      "epoch": 0.48327149467824126,
      "grad_norm": 0.10597799718379974,
      "learning_rate": 9.401274531014477e-06,
      "loss": 0.0035,
      "step": 75350
    },
    {
      "epoch": 0.48333563157202736,
      "grad_norm": 0.09346621483564377,
      "learning_rate": 9.401008923568883e-06,
      "loss": 0.0026,
      "step": 75360
    },
    {
      "epoch": 0.48339976846581345,
      "grad_norm": 0.1949937343597412,
      "learning_rate": 9.400743260975505e-06,
      "loss": 0.0031,
      "step": 75370
    },
    {
      "epoch": 0.48346390535959954,
      "grad_norm": 0.2209199219942093,
      "learning_rate": 9.400477543237669e-06,
      "loss": 0.0038,
      "step": 75380
    },
    {
      "epoch": 0.48352804225338564,
      "grad_norm": 0.0370480976998806,
      "learning_rate": 9.400211770358702e-06,
      "loss": 0.0069,
      "step": 75390
    },
    {
      "epoch": 0.48359217914717173,
      "grad_norm": 0.1484469175338745,
      "learning_rate": 9.399945942341939e-06,
      "loss": 0.0033,
      "step": 75400
    },
    {
      "epoch": 0.4836563160409578,
      "grad_norm": 0.20110897719860077,
      "learning_rate": 9.399680059190708e-06,
      "loss": 0.0028,
      "step": 75410
    },
    {
      "epoch": 0.4837204529347439,
      "grad_norm": 0.2016342133283615,
      "learning_rate": 9.39941412090834e-06,
      "loss": 0.0046,
      "step": 75420
    },
    {
      "epoch": 0.48378458982853,
      "grad_norm": 0.12587103247642517,
      "learning_rate": 9.39914812749817e-06,
      "loss": 0.0118,
      "step": 75430
    },
    {
      "epoch": 0.4838487267223161,
      "grad_norm": 0.04521534591913223,
      "learning_rate": 9.39888207896353e-06,
      "loss": 0.0041,
      "step": 75440
    },
    {
      "epoch": 0.4839128636161022,
      "grad_norm": 0.128385528922081,
      "learning_rate": 9.398615975307755e-06,
      "loss": 0.0041,
      "step": 75450
    },
    {
      "epoch": 0.4839770005098883,
      "grad_norm": 0.1759091317653656,
      "learning_rate": 9.398349816534176e-06,
      "loss": 0.0028,
      "step": 75460
    },
    {
      "epoch": 0.4840411374036744,
      "grad_norm": 0.16885876655578613,
      "learning_rate": 9.398083602646131e-06,
      "loss": 0.009,
      "step": 75470
    },
    {
      "epoch": 0.4841052742974605,
      "grad_norm": 0.20741593837738037,
      "learning_rate": 9.397817333646955e-06,
      "loss": 0.0045,
      "step": 75480
    },
    {
      "epoch": 0.4841694111912466,
      "grad_norm": 0.18445414304733276,
      "learning_rate": 9.397551009539985e-06,
      "loss": 0.0025,
      "step": 75490
    },
    {
      "epoch": 0.4842335480850327,
      "grad_norm": 0.23914507031440735,
      "learning_rate": 9.397284630328558e-06,
      "loss": 0.0029,
      "step": 75500
    },
    {
      "epoch": 0.4842976849788188,
      "grad_norm": 0.11996269971132278,
      "learning_rate": 9.397018196016012e-06,
      "loss": 0.0025,
      "step": 75510
    },
    {
      "epoch": 0.4843618218726049,
      "grad_norm": 0.169418066740036,
      "learning_rate": 9.396751706605686e-06,
      "loss": 0.0041,
      "step": 75520
    },
    {
      "epoch": 0.484425958766391,
      "grad_norm": 0.08760794252157211,
      "learning_rate": 9.39648516210092e-06,
      "loss": 0.0024,
      "step": 75530
    },
    {
      "epoch": 0.4844900956601771,
      "grad_norm": 0.12806841731071472,
      "learning_rate": 9.39621856250505e-06,
      "loss": 0.003,
      "step": 75540
    },
    {
      "epoch": 0.4845542325539632,
      "grad_norm": 0.2134033590555191,
      "learning_rate": 9.39595190782142e-06,
      "loss": 0.0025,
      "step": 75550
    },
    {
      "epoch": 0.4846183694477493,
      "grad_norm": 0.152133971452713,
      "learning_rate": 9.395685198053372e-06,
      "loss": 0.0039,
      "step": 75560
    },
    {
      "epoch": 0.48468250634153537,
      "grad_norm": 0.07009343057870865,
      "learning_rate": 9.395418433204246e-06,
      "loss": 0.0055,
      "step": 75570
    },
    {
      "epoch": 0.48474664323532146,
      "grad_norm": 0.19351892173290253,
      "learning_rate": 9.395151613277385e-06,
      "loss": 0.0027,
      "step": 75580
    },
    {
      "epoch": 0.48481078012910755,
      "grad_norm": 0.09136834740638733,
      "learning_rate": 9.394884738276136e-06,
      "loss": 0.0026,
      "step": 75590
    },
    {
      "epoch": 0.48487491702289365,
      "grad_norm": 0.15920791029930115,
      "learning_rate": 9.394617808203837e-06,
      "loss": 0.0025,
      "step": 75600
    },
    {
      "epoch": 0.48493905391667974,
      "grad_norm": 0.12583747506141663,
      "learning_rate": 9.394350823063839e-06,
      "loss": 0.0051,
      "step": 75610
    },
    {
      "epoch": 0.48500319081046583,
      "grad_norm": 0.13267534971237183,
      "learning_rate": 9.394083782859482e-06,
      "loss": 0.0042,
      "step": 75620
    },
    {
      "epoch": 0.485067327704252,
      "grad_norm": 0.4726957380771637,
      "learning_rate": 9.393816687594115e-06,
      "loss": 0.0044,
      "step": 75630
    },
    {
      "epoch": 0.4851314645980381,
      "grad_norm": 0.16043677926063538,
      "learning_rate": 9.393549537271086e-06,
      "loss": 0.0045,
      "step": 75640
    },
    {
      "epoch": 0.48519560149182417,
      "grad_norm": 0.06730765104293823,
      "learning_rate": 9.39328233189374e-06,
      "loss": 0.0031,
      "step": 75650
    },
    {
      "epoch": 0.48525973838561026,
      "grad_norm": 0.23805025219917297,
      "learning_rate": 9.393015071465427e-06,
      "loss": 0.0044,
      "step": 75660
    },
    {
      "epoch": 0.48532387527939636,
      "grad_norm": 0.12560462951660156,
      "learning_rate": 9.392747755989497e-06,
      "loss": 0.0041,
      "step": 75670
    },
    {
      "epoch": 0.48538801217318245,
      "grad_norm": 0.2656175196170807,
      "learning_rate": 9.392480385469298e-06,
      "loss": 0.0028,
      "step": 75680
    },
    {
      "epoch": 0.48545214906696854,
      "grad_norm": 0.09987766295671463,
      "learning_rate": 9.392212959908179e-06,
      "loss": 0.0032,
      "step": 75690
    },
    {
      "epoch": 0.48551628596075463,
      "grad_norm": 0.07306644320487976,
      "learning_rate": 9.391945479309494e-06,
      "loss": 0.0076,
      "step": 75700
    },
    {
      "epoch": 0.4855804228545407,
      "grad_norm": 0.05576948821544647,
      "learning_rate": 9.391677943676592e-06,
      "loss": 0.0028,
      "step": 75710
    },
    {
      "epoch": 0.4856445597483268,
      "grad_norm": 0.22854074835777283,
      "learning_rate": 9.391410353012826e-06,
      "loss": 0.0032,
      "step": 75720
    },
    {
      "epoch": 0.4857086966421129,
      "grad_norm": 0.24014638364315033,
      "learning_rate": 9.391142707321553e-06,
      "loss": 0.0035,
      "step": 75730
    },
    {
      "epoch": 0.485772833535899,
      "grad_norm": 0.25563791394233704,
      "learning_rate": 9.39087500660612e-06,
      "loss": 0.0033,
      "step": 75740
    },
    {
      "epoch": 0.4858369704296851,
      "grad_norm": 0.2181861698627472,
      "learning_rate": 9.390607250869885e-06,
      "loss": 0.0022,
      "step": 75750
    },
    {
      "epoch": 0.4859011073234712,
      "grad_norm": 0.7732840776443481,
      "learning_rate": 9.390339440116206e-06,
      "loss": 0.0144,
      "step": 75760
    },
    {
      "epoch": 0.48596524421725734,
      "grad_norm": 0.04870379716157913,
      "learning_rate": 9.390071574348434e-06,
      "loss": 0.004,
      "step": 75770
    },
    {
      "epoch": 0.48602938111104343,
      "grad_norm": 0.06384976208209991,
      "learning_rate": 9.389803653569929e-06,
      "loss": 0.0036,
      "step": 75780
    },
    {
      "epoch": 0.48609351800482953,
      "grad_norm": 0.1405668556690216,
      "learning_rate": 9.389535677784046e-06,
      "loss": 0.0032,
      "step": 75790
    },
    {
      "epoch": 0.4861576548986156,
      "grad_norm": 0.29477834701538086,
      "learning_rate": 9.389267646994145e-06,
      "loss": 0.0027,
      "step": 75800
    },
    {
      "epoch": 0.4862217917924017,
      "grad_norm": 0.25403454899787903,
      "learning_rate": 9.388999561203582e-06,
      "loss": 0.0068,
      "step": 75810
    },
    {
      "epoch": 0.4862859286861878,
      "grad_norm": 0.08830229938030243,
      "learning_rate": 9.388731420415718e-06,
      "loss": 0.005,
      "step": 75820
    },
    {
      "epoch": 0.4863500655799739,
      "grad_norm": 0.1264958530664444,
      "learning_rate": 9.388463224633912e-06,
      "loss": 0.003,
      "step": 75830
    },
    {
      "epoch": 0.48641420247376,
      "grad_norm": 0.49424171447753906,
      "learning_rate": 9.388194973861528e-06,
      "loss": 0.0041,
      "step": 75840
    },
    {
      "epoch": 0.4864783393675461,
      "grad_norm": 0.04571045562624931,
      "learning_rate": 9.387926668101923e-06,
      "loss": 0.0035,
      "step": 75850
    },
    {
      "epoch": 0.4865424762613322,
      "grad_norm": 0.07948622107505798,
      "learning_rate": 9.38765830735846e-06,
      "loss": 0.0023,
      "step": 75860
    },
    {
      "epoch": 0.4866066131551183,
      "grad_norm": 0.07302649319171906,
      "learning_rate": 9.387389891634505e-06,
      "loss": 0.0041,
      "step": 75870
    },
    {
      "epoch": 0.48667075004890437,
      "grad_norm": 0.15792128443717957,
      "learning_rate": 9.387121420933417e-06,
      "loss": 0.0046,
      "step": 75880
    },
    {
      "epoch": 0.48673488694269046,
      "grad_norm": 0.15861865878105164,
      "learning_rate": 9.386852895258562e-06,
      "loss": 0.0055,
      "step": 75890
    },
    {
      "epoch": 0.48679902383647655,
      "grad_norm": 0.1404324769973755,
      "learning_rate": 9.386584314613307e-06,
      "loss": 0.0028,
      "step": 75900
    },
    {
      "epoch": 0.48686316073026265,
      "grad_norm": 0.06636282056570053,
      "learning_rate": 9.386315679001014e-06,
      "loss": 0.0025,
      "step": 75910
    },
    {
      "epoch": 0.4869272976240488,
      "grad_norm": 0.26765817403793335,
      "learning_rate": 9.38604698842505e-06,
      "loss": 0.0037,
      "step": 75920
    },
    {
      "epoch": 0.4869914345178349,
      "grad_norm": 0.17180879414081573,
      "learning_rate": 9.385778242888784e-06,
      "loss": 0.0032,
      "step": 75930
    },
    {
      "epoch": 0.487055571411621,
      "grad_norm": 0.06586155295372009,
      "learning_rate": 9.385509442395583e-06,
      "loss": 0.0032,
      "step": 75940
    },
    {
      "epoch": 0.4871197083054071,
      "grad_norm": 0.14726197719573975,
      "learning_rate": 9.385240586948812e-06,
      "loss": 0.0043,
      "step": 75950
    },
    {
      "epoch": 0.48718384519919317,
      "grad_norm": 0.2654615342617035,
      "learning_rate": 9.384971676551844e-06,
      "loss": 0.0043,
      "step": 75960
    },
    {
      "epoch": 0.48724798209297926,
      "grad_norm": 0.2780192494392395,
      "learning_rate": 9.384702711208047e-06,
      "loss": 0.0037,
      "step": 75970
    },
    {
      "epoch": 0.48731211898676535,
      "grad_norm": 0.1765231043100357,
      "learning_rate": 9.38443369092079e-06,
      "loss": 0.0028,
      "step": 75980
    },
    {
      "epoch": 0.48737625588055145,
      "grad_norm": 0.17971616983413696,
      "learning_rate": 9.384164615693445e-06,
      "loss": 0.0023,
      "step": 75990
    },
    {
      "epoch": 0.48744039277433754,
      "grad_norm": 0.17881891131401062,
      "learning_rate": 9.383895485529387e-06,
      "loss": 0.0025,
      "step": 76000
    },
    {
      "epoch": 0.48750452966812363,
      "grad_norm": 0.03880459442734718,
      "learning_rate": 9.383626300431982e-06,
      "loss": 0.0084,
      "step": 76010
    },
    {
      "epoch": 0.4875686665619097,
      "grad_norm": 0.20588567852973938,
      "learning_rate": 9.38335706040461e-06,
      "loss": 0.003,
      "step": 76020
    },
    {
      "epoch": 0.4876328034556958,
      "grad_norm": 0.14494794607162476,
      "learning_rate": 9.383087765450638e-06,
      "loss": 0.0037,
      "step": 76030
    },
    {
      "epoch": 0.4876969403494819,
      "grad_norm": 0.2012653797864914,
      "learning_rate": 9.382818415573446e-06,
      "loss": 0.0059,
      "step": 76040
    },
    {
      "epoch": 0.487761077243268,
      "grad_norm": 0.06287068873643875,
      "learning_rate": 9.382549010776407e-06,
      "loss": 0.0046,
      "step": 76050
    },
    {
      "epoch": 0.48782521413705415,
      "grad_norm": 0.13338372111320496,
      "learning_rate": 9.382279551062894e-06,
      "loss": 0.0049,
      "step": 76060
    },
    {
      "epoch": 0.48788935103084025,
      "grad_norm": 0.4038761258125305,
      "learning_rate": 9.38201003643629e-06,
      "loss": 0.0032,
      "step": 76070
    },
    {
      "epoch": 0.48795348792462634,
      "grad_norm": 0.16861388087272644,
      "learning_rate": 9.381740466899966e-06,
      "loss": 0.0048,
      "step": 76080
    },
    {
      "epoch": 0.48801762481841243,
      "grad_norm": 0.11070922762155533,
      "learning_rate": 9.381470842457304e-06,
      "loss": 0.0038,
      "step": 76090
    },
    {
      "epoch": 0.4880817617121985,
      "grad_norm": 0.11704026162624359,
      "learning_rate": 9.381201163111678e-06,
      "loss": 0.0063,
      "step": 76100
    },
    {
      "epoch": 0.4881458986059846,
      "grad_norm": 0.23959235846996307,
      "learning_rate": 9.380931428866473e-06,
      "loss": 0.0039,
      "step": 76110
    },
    {
      "epoch": 0.4882100354997707,
      "grad_norm": 0.1530313491821289,
      "learning_rate": 9.380661639725065e-06,
      "loss": 0.0038,
      "step": 76120
    },
    {
      "epoch": 0.4882741723935568,
      "grad_norm": 0.24976380169391632,
      "learning_rate": 9.380391795690836e-06,
      "loss": 0.0047,
      "step": 76130
    },
    {
      "epoch": 0.4883383092873429,
      "grad_norm": 0.28350764513015747,
      "learning_rate": 9.380121896767167e-06,
      "loss": 0.0037,
      "step": 76140
    },
    {
      "epoch": 0.488402446181129,
      "grad_norm": 0.06945674866437912,
      "learning_rate": 9.37985194295744e-06,
      "loss": 0.002,
      "step": 76150
    },
    {
      "epoch": 0.4884665830749151,
      "grad_norm": 0.4729636311531067,
      "learning_rate": 9.379581934265039e-06,
      "loss": 0.0039,
      "step": 76160
    },
    {
      "epoch": 0.4885307199687012,
      "grad_norm": 0.04570367559790611,
      "learning_rate": 9.379311870693346e-06,
      "loss": 0.0027,
      "step": 76170
    },
    {
      "epoch": 0.48859485686248727,
      "grad_norm": 0.112314872443676,
      "learning_rate": 9.379041752245746e-06,
      "loss": 0.0027,
      "step": 76180
    },
    {
      "epoch": 0.48865899375627336,
      "grad_norm": 0.11408799141645432,
      "learning_rate": 9.378771578925621e-06,
      "loss": 0.0032,
      "step": 76190
    },
    {
      "epoch": 0.4887231306500595,
      "grad_norm": 0.2096250206232071,
      "learning_rate": 9.37850135073636e-06,
      "loss": 0.0045,
      "step": 76200
    },
    {
      "epoch": 0.4887872675438456,
      "grad_norm": 0.06932587921619415,
      "learning_rate": 9.378231067681349e-06,
      "loss": 0.0036,
      "step": 76210
    },
    {
      "epoch": 0.4888514044376317,
      "grad_norm": 0.04841315373778343,
      "learning_rate": 9.377960729763971e-06,
      "loss": 0.0059,
      "step": 76220
    },
    {
      "epoch": 0.4889155413314178,
      "grad_norm": 0.14756424725055695,
      "learning_rate": 9.377690336987619e-06,
      "loss": 0.0038,
      "step": 76230
    },
    {
      "epoch": 0.4889796782252039,
      "grad_norm": 0.1798335611820221,
      "learning_rate": 9.377419889355677e-06,
      "loss": 0.0045,
      "step": 76240
    },
    {
      "epoch": 0.48904381511899,
      "grad_norm": 0.10555068403482437,
      "learning_rate": 9.377149386871536e-06,
      "loss": 0.0027,
      "step": 76250
    },
    {
      "epoch": 0.48910795201277607,
      "grad_norm": 0.0862729549407959,
      "learning_rate": 9.376878829538583e-06,
      "loss": 0.0022,
      "step": 76260
    },
    {
      "epoch": 0.48917208890656216,
      "grad_norm": 0.1376761496067047,
      "learning_rate": 9.376608217360212e-06,
      "loss": 0.0039,
      "step": 76270
    },
    {
      "epoch": 0.48923622580034826,
      "grad_norm": 0.2567867934703827,
      "learning_rate": 9.376337550339811e-06,
      "loss": 0.0056,
      "step": 76280
    },
    {
      "epoch": 0.48930036269413435,
      "grad_norm": 0.05342590808868408,
      "learning_rate": 9.376066828480774e-06,
      "loss": 0.0018,
      "step": 76290
    },
    {
      "epoch": 0.48936449958792044,
      "grad_norm": 0.22637926042079926,
      "learning_rate": 9.375796051786492e-06,
      "loss": 0.0027,
      "step": 76300
    },
    {
      "epoch": 0.48942863648170654,
      "grad_norm": 0.24293087422847748,
      "learning_rate": 9.375525220260356e-06,
      "loss": 0.004,
      "step": 76310
    },
    {
      "epoch": 0.48949277337549263,
      "grad_norm": 0.2584352195262909,
      "learning_rate": 9.375254333905764e-06,
      "loss": 0.0047,
      "step": 76320
    },
    {
      "epoch": 0.4895569102692787,
      "grad_norm": 0.045442432165145874,
      "learning_rate": 9.374983392726107e-06,
      "loss": 0.0052,
      "step": 76330
    },
    {
      "epoch": 0.48962104716306487,
      "grad_norm": 0.1458451747894287,
      "learning_rate": 9.374712396724782e-06,
      "loss": 0.0028,
      "step": 76340
    },
    {
      "epoch": 0.48968518405685096,
      "grad_norm": 0.07689321041107178,
      "learning_rate": 9.374441345905184e-06,
      "loss": 0.0032,
      "step": 76350
    },
    {
      "epoch": 0.48974932095063706,
      "grad_norm": 0.4120532274246216,
      "learning_rate": 9.37417024027071e-06,
      "loss": 0.007,
      "step": 76360
    },
    {
      "epoch": 0.48981345784442315,
      "grad_norm": 0.03383118659257889,
      "learning_rate": 9.373899079824757e-06,
      "loss": 0.0024,
      "step": 76370
    },
    {
      "epoch": 0.48987759473820924,
      "grad_norm": 0.2416459619998932,
      "learning_rate": 9.373627864570722e-06,
      "loss": 0.0041,
      "step": 76380
    },
    {
      "epoch": 0.48994173163199534,
      "grad_norm": 0.12423533946275711,
      "learning_rate": 9.373356594512004e-06,
      "loss": 0.0024,
      "step": 76390
    },
    {
      "epoch": 0.49000586852578143,
      "grad_norm": 0.14251717925071716,
      "learning_rate": 9.373085269652003e-06,
      "loss": 0.0042,
      "step": 76400
    },
    {
      "epoch": 0.4900700054195675,
      "grad_norm": 0.23361019790172577,
      "learning_rate": 9.372813889994116e-06,
      "loss": 0.003,
      "step": 76410
    },
    {
      "epoch": 0.4901341423133536,
      "grad_norm": 0.15831750631332397,
      "learning_rate": 9.372542455541748e-06,
      "loss": 0.0039,
      "step": 76420
    },
    {
      "epoch": 0.4901982792071397,
      "grad_norm": 0.10762511938810349,
      "learning_rate": 9.372270966298296e-06,
      "loss": 0.0037,
      "step": 76430
    },
    {
      "epoch": 0.4902624161009258,
      "grad_norm": 0.0905207172036171,
      "learning_rate": 9.371999422267166e-06,
      "loss": 0.0054,
      "step": 76440
    },
    {
      "epoch": 0.4903265529947119,
      "grad_norm": 0.025468701496720314,
      "learning_rate": 9.371727823451758e-06,
      "loss": 0.0036,
      "step": 76450
    },
    {
      "epoch": 0.490390689888498,
      "grad_norm": 0.08132292330265045,
      "learning_rate": 9.371456169855476e-06,
      "loss": 0.0037,
      "step": 76460
    },
    {
      "epoch": 0.4904548267822841,
      "grad_norm": 0.2487182915210724,
      "learning_rate": 9.371184461481724e-06,
      "loss": 0.0033,
      "step": 76470
    },
    {
      "epoch": 0.49051896367607023,
      "grad_norm": 0.14117854833602905,
      "learning_rate": 9.370912698333906e-06,
      "loss": 0.0039,
      "step": 76480
    },
    {
      "epoch": 0.4905831005698563,
      "grad_norm": 0.2164766639471054,
      "learning_rate": 9.370640880415428e-06,
      "loss": 0.0042,
      "step": 76490
    },
    {
      "epoch": 0.4906472374636424,
      "grad_norm": 0.1885480433702469,
      "learning_rate": 9.370369007729697e-06,
      "loss": 0.0025,
      "step": 76500
    },
    {
      "epoch": 0.4907113743574285,
      "grad_norm": 0.016325635835528374,
      "learning_rate": 9.370097080280118e-06,
      "loss": 0.0045,
      "step": 76510
    },
    {
      "epoch": 0.4907755112512146,
      "grad_norm": 0.21638783812522888,
      "learning_rate": 9.3698250980701e-06,
      "loss": 0.0038,
      "step": 76520
    },
    {
      "epoch": 0.4908396481450007,
      "grad_norm": 0.13634824752807617,
      "learning_rate": 9.36955306110305e-06,
      "loss": 0.0066,
      "step": 76530
    },
    {
      "epoch": 0.4909037850387868,
      "grad_norm": 0.012613404542207718,
      "learning_rate": 9.369280969382378e-06,
      "loss": 0.006,
      "step": 76540
    },
    {
      "epoch": 0.4909679219325729,
      "grad_norm": 0.09728101640939713,
      "learning_rate": 9.369008822911492e-06,
      "loss": 0.0037,
      "step": 76550
    },
    {
      "epoch": 0.491032058826359,
      "grad_norm": 0.06714236736297607,
      "learning_rate": 9.368736621693803e-06,
      "loss": 0.004,
      "step": 76560
    },
    {
      "epoch": 0.49109619572014507,
      "grad_norm": 0.08015460520982742,
      "learning_rate": 9.368464365732721e-06,
      "loss": 0.0041,
      "step": 76570
    },
    {
      "epoch": 0.49116033261393116,
      "grad_norm": 0.3189891278743744,
      "learning_rate": 9.36819205503166e-06,
      "loss": 0.0028,
      "step": 76580
    },
    {
      "epoch": 0.49122446950771725,
      "grad_norm": 0.06811632215976715,
      "learning_rate": 9.367919689594031e-06,
      "loss": 0.0028,
      "step": 76590
    },
    {
      "epoch": 0.49128860640150335,
      "grad_norm": 0.36249682307243347,
      "learning_rate": 9.367647269423246e-06,
      "loss": 0.003,
      "step": 76600
    },
    {
      "epoch": 0.49135274329528944,
      "grad_norm": 0.2812708616256714,
      "learning_rate": 9.367374794522719e-06,
      "loss": 0.004,
      "step": 76610
    },
    {
      "epoch": 0.4914168801890756,
      "grad_norm": 0.19123730063438416,
      "learning_rate": 9.367102264895864e-06,
      "loss": 0.0042,
      "step": 76620
    },
    {
      "epoch": 0.4914810170828617,
      "grad_norm": 0.10681630671024323,
      "learning_rate": 9.366829680546096e-06,
      "loss": 0.0028,
      "step": 76630
    },
    {
      "epoch": 0.4915451539766478,
      "grad_norm": 0.11751936376094818,
      "learning_rate": 9.366557041476832e-06,
      "loss": 0.0023,
      "step": 76640
    },
    {
      "epoch": 0.49160929087043387,
      "grad_norm": 0.2586619555950165,
      "learning_rate": 9.366284347691489e-06,
      "loss": 0.0028,
      "step": 76650
    },
    {
      "epoch": 0.49167342776421996,
      "grad_norm": 0.09923292696475983,
      "learning_rate": 9.36601159919348e-06,
      "loss": 0.005,
      "step": 76660
    },
    {
      "epoch": 0.49173756465800605,
      "grad_norm": 0.07947537302970886,
      "learning_rate": 9.365738795986227e-06,
      "loss": 0.0036,
      "step": 76670
    },
    {
      "epoch": 0.49180170155179215,
      "grad_norm": 0.2412528395652771,
      "learning_rate": 9.365465938073146e-06,
      "loss": 0.0043,
      "step": 76680
    },
    {
      "epoch": 0.49186583844557824,
      "grad_norm": 0.03592299297451973,
      "learning_rate": 9.365193025457657e-06,
      "loss": 0.0037,
      "step": 76690
    },
    {
      "epoch": 0.49192997533936433,
      "grad_norm": 0.3043968677520752,
      "learning_rate": 9.364920058143181e-06,
      "loss": 0.0064,
      "step": 76700
    },
    {
      "epoch": 0.4919941122331504,
      "grad_norm": 0.10949330776929855,
      "learning_rate": 9.364647036133135e-06,
      "loss": 0.0043,
      "step": 76710
    },
    {
      "epoch": 0.4920582491269365,
      "grad_norm": 0.07797112315893173,
      "learning_rate": 9.364373959430944e-06,
      "loss": 0.0053,
      "step": 76720
    },
    {
      "epoch": 0.4921223860207226,
      "grad_norm": 0.28760960698127747,
      "learning_rate": 9.364100828040026e-06,
      "loss": 0.0062,
      "step": 76730
    },
    {
      "epoch": 0.4921865229145087,
      "grad_norm": 0.20983117818832397,
      "learning_rate": 9.363827641963808e-06,
      "loss": 0.0032,
      "step": 76740
    },
    {
      "epoch": 0.4922506598082948,
      "grad_norm": 0.06841584295034409,
      "learning_rate": 9.36355440120571e-06,
      "loss": 0.0048,
      "step": 76750
    },
    {
      "epoch": 0.49231479670208095,
      "grad_norm": 0.1825391948223114,
      "learning_rate": 9.363281105769155e-06,
      "loss": 0.003,
      "step": 76760
    },
    {
      "epoch": 0.49237893359586704,
      "grad_norm": 0.19833020865917206,
      "learning_rate": 9.363007755657571e-06,
      "loss": 0.0048,
      "step": 76770
    },
    {
      "epoch": 0.49244307048965313,
      "grad_norm": 0.1410582810640335,
      "learning_rate": 9.362734350874382e-06,
      "loss": 0.0018,
      "step": 76780
    },
    {
      "epoch": 0.4925072073834392,
      "grad_norm": 0.1286795288324356,
      "learning_rate": 9.362460891423013e-06,
      "loss": 0.0031,
      "step": 76790
    },
    {
      "epoch": 0.4925713442772253,
      "grad_norm": 0.17199252545833588,
      "learning_rate": 9.362187377306892e-06,
      "loss": 0.0035,
      "step": 76800
    },
    {
      "epoch": 0.4926354811710114,
      "grad_norm": 0.2555850148200989,
      "learning_rate": 9.361913808529443e-06,
      "loss": 0.0046,
      "step": 76810
    },
    {
      "epoch": 0.4926996180647975,
      "grad_norm": 0.1507069617509842,
      "learning_rate": 9.3616401850941e-06,
      "loss": 0.0037,
      "step": 76820
    },
    {
      "epoch": 0.4927637549585836,
      "grad_norm": 0.20394714176654816,
      "learning_rate": 9.361366507004286e-06,
      "loss": 0.0026,
      "step": 76830
    },
    {
      "epoch": 0.4928278918523697,
      "grad_norm": 0.26713767647743225,
      "learning_rate": 9.361092774263434e-06,
      "loss": 0.0076,
      "step": 76840
    },
    {
      "epoch": 0.4928920287461558,
      "grad_norm": 0.07934827357530594,
      "learning_rate": 9.360818986874971e-06,
      "loss": 0.0044,
      "step": 76850
    },
    {
      "epoch": 0.4929561656399419,
      "grad_norm": 0.13926687836647034,
      "learning_rate": 9.360545144842332e-06,
      "loss": 0.0036,
      "step": 76860
    },
    {
      "epoch": 0.49302030253372797,
      "grad_norm": 0.09511993080377579,
      "learning_rate": 9.360271248168944e-06,
      "loss": 0.0032,
      "step": 76870
    },
    {
      "epoch": 0.49308443942751407,
      "grad_norm": 0.28910213708877563,
      "learning_rate": 9.359997296858241e-06,
      "loss": 0.0044,
      "step": 76880
    },
    {
      "epoch": 0.49314857632130016,
      "grad_norm": 0.10169561207294464,
      "learning_rate": 9.359723290913656e-06,
      "loss": 0.0045,
      "step": 76890
    },
    {
      "epoch": 0.4932127132150863,
      "grad_norm": 0.1778484284877777,
      "learning_rate": 9.359449230338622e-06,
      "loss": 0.0031,
      "step": 76900
    },
    {
      "epoch": 0.4932768501088724,
      "grad_norm": 0.44378116726875305,
      "learning_rate": 9.359175115136575e-06,
      "loss": 0.0032,
      "step": 76910
    },
    {
      "epoch": 0.4933409870026585,
      "grad_norm": 0.31982800364494324,
      "learning_rate": 9.358900945310947e-06,
      "loss": 0.0047,
      "step": 76920
    },
    {
      "epoch": 0.4934051238964446,
      "grad_norm": 0.11058761179447174,
      "learning_rate": 9.358626720865176e-06,
      "loss": 0.0049,
      "step": 76930
    },
    {
      "epoch": 0.4934692607902307,
      "grad_norm": 0.28282925486564636,
      "learning_rate": 9.358352441802696e-06,
      "loss": 0.0087,
      "step": 76940
    },
    {
      "epoch": 0.4935333976840168,
      "grad_norm": 0.07016579061746597,
      "learning_rate": 9.358078108126947e-06,
      "loss": 0.0032,
      "step": 76950
    },
    {
      "epoch": 0.49359753457780287,
      "grad_norm": 0.05021855607628822,
      "learning_rate": 9.357803719841362e-06,
      "loss": 0.002,
      "step": 76960
    },
    {
      "epoch": 0.49366167147158896,
      "grad_norm": 0.24600309133529663,
      "learning_rate": 9.357529276949383e-06,
      "loss": 0.0036,
      "step": 76970
    },
    {
      "epoch": 0.49372580836537505,
      "grad_norm": 0.21833530068397522,
      "learning_rate": 9.357254779454448e-06,
      "loss": 0.005,
      "step": 76980
    },
    {
      "epoch": 0.49378994525916114,
      "grad_norm": 0.19952186942100525,
      "learning_rate": 9.356980227359998e-06,
      "loss": 0.007,
      "step": 76990
    },
    {
      "epoch": 0.49385408215294724,
      "grad_norm": 0.11434321850538254,
      "learning_rate": 9.356705620669469e-06,
      "loss": 0.0049,
      "step": 77000
    },
    {
      "epoch": 0.49391821904673333,
      "grad_norm": 0.1786748468875885,
      "learning_rate": 9.356430959386307e-06,
      "loss": 0.0046,
      "step": 77010
    },
    {
      "epoch": 0.4939823559405194,
      "grad_norm": 0.364108145236969,
      "learning_rate": 9.35615624351395e-06,
      "loss": 0.0062,
      "step": 77020
    },
    {
      "epoch": 0.4940464928343055,
      "grad_norm": 0.3673844337463379,
      "learning_rate": 9.355881473055844e-06,
      "loss": 0.005,
      "step": 77030
    },
    {
      "epoch": 0.49411062972809167,
      "grad_norm": 0.11770845949649811,
      "learning_rate": 9.355606648015428e-06,
      "loss": 0.0039,
      "step": 77040
    },
    {
      "epoch": 0.49417476662187776,
      "grad_norm": 0.022180555388331413,
      "learning_rate": 9.355331768396148e-06,
      "loss": 0.0039,
      "step": 77050
    },
    {
      "epoch": 0.49423890351566385,
      "grad_norm": 0.1450425386428833,
      "learning_rate": 9.35505683420145e-06,
      "loss": 0.0024,
      "step": 77060
    },
    {
      "epoch": 0.49430304040944995,
      "grad_norm": 0.0696171298623085,
      "learning_rate": 9.354781845434774e-06,
      "loss": 0.0028,
      "step": 77070
    },
    {
      "epoch": 0.49436717730323604,
      "grad_norm": 0.12352907657623291,
      "learning_rate": 9.354506802099572e-06,
      "loss": 0.003,
      "step": 77080
    },
    {
      "epoch": 0.49443131419702213,
      "grad_norm": 0.050051745027303696,
      "learning_rate": 9.354231704199288e-06,
      "loss": 0.0025,
      "step": 77090
    },
    {
      "epoch": 0.4944954510908082,
      "grad_norm": 0.22080251574516296,
      "learning_rate": 9.353956551737367e-06,
      "loss": 0.0028,
      "step": 77100
    },
    {
      "epoch": 0.4945595879845943,
      "grad_norm": 0.07721296697854996,
      "learning_rate": 9.35368134471726e-06,
      "loss": 0.0028,
      "step": 77110
    },
    {
      "epoch": 0.4946237248783804,
      "grad_norm": 0.009079434908926487,
      "learning_rate": 9.353406083142414e-06,
      "loss": 0.0038,
      "step": 77120
    },
    {
      "epoch": 0.4946878617721665,
      "grad_norm": 0.23502038419246674,
      "learning_rate": 9.353130767016278e-06,
      "loss": 0.0037,
      "step": 77130
    },
    {
      "epoch": 0.4947519986659526,
      "grad_norm": 0.0968853309750557,
      "learning_rate": 9.352855396342302e-06,
      "loss": 0.004,
      "step": 77140
    },
    {
      "epoch": 0.4948161355597387,
      "grad_norm": 0.16828656196594238,
      "learning_rate": 9.352579971123938e-06,
      "loss": 0.0026,
      "step": 77150
    },
    {
      "epoch": 0.4948802724535248,
      "grad_norm": 0.1029152199625969,
      "learning_rate": 9.352304491364636e-06,
      "loss": 0.0062,
      "step": 77160
    },
    {
      "epoch": 0.4949444093473109,
      "grad_norm": 0.12365434318780899,
      "learning_rate": 9.352028957067848e-06,
      "loss": 0.0036,
      "step": 77170
    },
    {
      "epoch": 0.495008546241097,
      "grad_norm": 0.1805618852376938,
      "learning_rate": 9.351753368237027e-06,
      "loss": 0.0028,
      "step": 77180
    },
    {
      "epoch": 0.4950726831348831,
      "grad_norm": 0.16249337792396545,
      "learning_rate": 9.351477724875623e-06,
      "loss": 0.0026,
      "step": 77190
    },
    {
      "epoch": 0.4951368200286692,
      "grad_norm": 0.07809522747993469,
      "learning_rate": 9.351202026987098e-06,
      "loss": 0.0026,
      "step": 77200
    },
    {
      "epoch": 0.4952009569224553,
      "grad_norm": 0.06574942171573639,
      "learning_rate": 9.3509262745749e-06,
      "loss": 0.0046,
      "step": 77210
    },
    {
      "epoch": 0.4952650938162414,
      "grad_norm": 0.05719909444451332,
      "learning_rate": 9.350650467642486e-06,
      "loss": 0.0032,
      "step": 77220
    },
    {
      "epoch": 0.4953292307100275,
      "grad_norm": 0.016466276720166206,
      "learning_rate": 9.350374606193311e-06,
      "loss": 0.0032,
      "step": 77230
    },
    {
      "epoch": 0.4953933676038136,
      "grad_norm": 0.1982014775276184,
      "learning_rate": 9.350098690230835e-06,
      "loss": 0.0014,
      "step": 77240
    },
    {
      "epoch": 0.4954575044975997,
      "grad_norm": 0.3372277021408081,
      "learning_rate": 9.349822719758514e-06,
      "loss": 0.0041,
      "step": 77250
    },
    {
      "epoch": 0.49552164139138577,
      "grad_norm": 0.07121748477220535,
      "learning_rate": 9.349546694779803e-06,
      "loss": 0.0026,
      "step": 77260
    },
    {
      "epoch": 0.49558577828517186,
      "grad_norm": 0.18666061758995056,
      "learning_rate": 9.349270615298165e-06,
      "loss": 0.0032,
      "step": 77270
    },
    {
      "epoch": 0.49564991517895796,
      "grad_norm": 0.4052756428718567,
      "learning_rate": 9.348994481317057e-06,
      "loss": 0.0041,
      "step": 77280
    },
    {
      "epoch": 0.49571405207274405,
      "grad_norm": 0.32904452085494995,
      "learning_rate": 9.34871829283994e-06,
      "loss": 0.0069,
      "step": 77290
    },
    {
      "epoch": 0.49577818896653014,
      "grad_norm": 0.052296873182058334,
      "learning_rate": 9.348442049870276e-06,
      "loss": 0.0026,
      "step": 77300
    },
    {
      "epoch": 0.49584232586031624,
      "grad_norm": 0.27496102452278137,
      "learning_rate": 9.348165752411524e-06,
      "loss": 0.0044,
      "step": 77310
    },
    {
      "epoch": 0.49590646275410233,
      "grad_norm": 0.24938185513019562,
      "learning_rate": 9.34788940046715e-06,
      "loss": 0.0024,
      "step": 77320
    },
    {
      "epoch": 0.4959705996478885,
      "grad_norm": 0.1310003399848938,
      "learning_rate": 9.34761299404061e-06,
      "loss": 0.0045,
      "step": 77330
    },
    {
      "epoch": 0.49603473654167457,
      "grad_norm": 0.18255215883255005,
      "learning_rate": 9.347336533135376e-06,
      "loss": 0.0033,
      "step": 77340
    },
    {
      "epoch": 0.49609887343546066,
      "grad_norm": 0.12947233021259308,
      "learning_rate": 9.347060017754908e-06,
      "loss": 0.0058,
      "step": 77350
    },
    {
      "epoch": 0.49616301032924676,
      "grad_norm": 0.0997142493724823,
      "learning_rate": 9.34678344790267e-06,
      "loss": 0.0017,
      "step": 77360
    },
    {
      "epoch": 0.49622714722303285,
      "grad_norm": 0.0739133283495903,
      "learning_rate": 9.346506823582128e-06,
      "loss": 0.0025,
      "step": 77370
    },
    {
      "epoch": 0.49629128411681894,
      "grad_norm": 0.14539377391338348,
      "learning_rate": 9.34623014479675e-06,
      "loss": 0.0032,
      "step": 77380
    },
    {
      "epoch": 0.49635542101060504,
      "grad_norm": 0.1064482033252716,
      "learning_rate": 9.345953411550002e-06,
      "loss": 0.004,
      "step": 77390
    },
    {
      "epoch": 0.49641955790439113,
      "grad_norm": 0.19258669018745422,
      "learning_rate": 9.345676623845351e-06,
      "loss": 0.0018,
      "step": 77400
    },
    {
      "epoch": 0.4964836947981772,
      "grad_norm": 0.1373535841703415,
      "learning_rate": 9.345399781686267e-06,
      "loss": 0.0043,
      "step": 77410
    },
    {
      "epoch": 0.4965478316919633,
      "grad_norm": 0.2516781687736511,
      "learning_rate": 9.345122885076219e-06,
      "loss": 0.0044,
      "step": 77420
    },
    {
      "epoch": 0.4966119685857494,
      "grad_norm": 0.1292618066072464,
      "learning_rate": 9.344845934018674e-06,
      "loss": 0.0023,
      "step": 77430
    },
    {
      "epoch": 0.4966761054795355,
      "grad_norm": 0.2453163117170334,
      "learning_rate": 9.344568928517105e-06,
      "loss": 0.0071,
      "step": 77440
    },
    {
      "epoch": 0.4967402423733216,
      "grad_norm": 0.07375278323888779,
      "learning_rate": 9.344291868574982e-06,
      "loss": 0.0041,
      "step": 77450
    },
    {
      "epoch": 0.4968043792671077,
      "grad_norm": 0.1308964341878891,
      "learning_rate": 9.344014754195779e-06,
      "loss": 0.0023,
      "step": 77460
    },
    {
      "epoch": 0.49686851616089384,
      "grad_norm": 0.10841777175664902,
      "learning_rate": 9.343737585382963e-06,
      "loss": 0.0035,
      "step": 77470
    },
    {
      "epoch": 0.49693265305467993,
      "grad_norm": 0.16187961399555206,
      "learning_rate": 9.343460362140014e-06,
      "loss": 0.0029,
      "step": 77480
    },
    {
      "epoch": 0.496996789948466,
      "grad_norm": 0.14524035155773163,
      "learning_rate": 9.3431830844704e-06,
      "loss": 0.0065,
      "step": 77490
    },
    {
      "epoch": 0.4970609268422521,
      "grad_norm": 0.23628103733062744,
      "learning_rate": 9.342905752377598e-06,
      "loss": 0.0075,
      "step": 77500
    },
    {
      "epoch": 0.4971250637360382,
      "grad_norm": 0.2749999761581421,
      "learning_rate": 9.342628365865084e-06,
      "loss": 0.0027,
      "step": 77510
    },
    {
      "epoch": 0.4971892006298243,
      "grad_norm": 0.2397809773683548,
      "learning_rate": 9.342350924936335e-06,
      "loss": 0.0052,
      "step": 77520
    },
    {
      "epoch": 0.4972533375236104,
      "grad_norm": 0.2140444964170456,
      "learning_rate": 9.342073429594822e-06,
      "loss": 0.0051,
      "step": 77530
    },
    {
      "epoch": 0.4973174744173965,
      "grad_norm": 0.19114813208580017,
      "learning_rate": 9.341795879844026e-06,
      "loss": 0.0068,
      "step": 77540
    },
    {
      "epoch": 0.4973816113111826,
      "grad_norm": 0.09587656706571579,
      "learning_rate": 9.341518275687426e-06,
      "loss": 0.0027,
      "step": 77550
    },
    {
      "epoch": 0.4974457482049687,
      "grad_norm": 0.10873784869909286,
      "learning_rate": 9.341240617128499e-06,
      "loss": 0.003,
      "step": 77560
    },
    {
      "epoch": 0.49750988509875477,
      "grad_norm": 0.290728360414505,
      "learning_rate": 9.340962904170726e-06,
      "loss": 0.0034,
      "step": 77570
    },
    {
      "epoch": 0.49757402199254086,
      "grad_norm": 0.0844135656952858,
      "learning_rate": 9.340685136817582e-06,
      "loss": 0.003,
      "step": 77580
    },
    {
      "epoch": 0.49763815888632695,
      "grad_norm": 0.3078419268131256,
      "learning_rate": 9.340407315072553e-06,
      "loss": 0.0036,
      "step": 77590
    },
    {
      "epoch": 0.49770229578011305,
      "grad_norm": 0.08919071406126022,
      "learning_rate": 9.340129438939119e-06,
      "loss": 0.003,
      "step": 77600
    },
    {
      "epoch": 0.4977664326738992,
      "grad_norm": 0.18456967175006866,
      "learning_rate": 9.33985150842076e-06,
      "loss": 0.0025,
      "step": 77610
    },
    {
      "epoch": 0.4978305695676853,
      "grad_norm": 0.24292078614234924,
      "learning_rate": 9.33957352352096e-06,
      "loss": 0.0036,
      "step": 77620
    },
    {
      "epoch": 0.4978947064614714,
      "grad_norm": 0.09577593952417374,
      "learning_rate": 9.339295484243203e-06,
      "loss": 0.0022,
      "step": 77630
    },
    {
      "epoch": 0.4979588433552575,
      "grad_norm": 0.08392419666051865,
      "learning_rate": 9.339017390590971e-06,
      "loss": 0.0024,
      "step": 77640
    },
    {
      "epoch": 0.49802298024904357,
      "grad_norm": 0.20557403564453125,
      "learning_rate": 9.338739242567752e-06,
      "loss": 0.0048,
      "step": 77650
    },
    {
      "epoch": 0.49808711714282966,
      "grad_norm": 0.06604457646608353,
      "learning_rate": 9.338461040177026e-06,
      "loss": 0.0032,
      "step": 77660
    },
    {
      "epoch": 0.49815125403661575,
      "grad_norm": 0.055695466697216034,
      "learning_rate": 9.338182783422286e-06,
      "loss": 0.0026,
      "step": 77670
    },
    {
      "epoch": 0.49821539093040185,
      "grad_norm": 0.2561768591403961,
      "learning_rate": 9.337904472307013e-06,
      "loss": 0.0034,
      "step": 77680
    },
    {
      "epoch": 0.49827952782418794,
      "grad_norm": 0.07979772239923477,
      "learning_rate": 9.337626106834698e-06,
      "loss": 0.0041,
      "step": 77690
    },
    {
      "epoch": 0.49834366471797403,
      "grad_norm": 0.2786787748336792,
      "learning_rate": 9.337347687008828e-06,
      "loss": 0.0046,
      "step": 77700
    },
    {
      "epoch": 0.4984078016117601,
      "grad_norm": 0.09194999188184738,
      "learning_rate": 9.337069212832892e-06,
      "loss": 0.004,
      "step": 77710
    },
    {
      "epoch": 0.4984719385055462,
      "grad_norm": 0.1742803454399109,
      "learning_rate": 9.336790684310377e-06,
      "loss": 0.0035,
      "step": 77720
    },
    {
      "epoch": 0.4985360753993323,
      "grad_norm": 0.11242389678955078,
      "learning_rate": 9.336512101444776e-06,
      "loss": 0.0046,
      "step": 77730
    },
    {
      "epoch": 0.4986002122931184,
      "grad_norm": 0.1943175494670868,
      "learning_rate": 9.33623346423958e-06,
      "loss": 0.0036,
      "step": 77740
    },
    {
      "epoch": 0.49866434918690455,
      "grad_norm": 0.07592809945344925,
      "learning_rate": 9.335954772698282e-06,
      "loss": 0.002,
      "step": 77750
    },
    {
      "epoch": 0.49872848608069065,
      "grad_norm": 0.09480135142803192,
      "learning_rate": 9.335676026824367e-06,
      "loss": 0.0034,
      "step": 77760
    },
    {
      "epoch": 0.49879262297447674,
      "grad_norm": 0.041317373514175415,
      "learning_rate": 9.335397226621336e-06,
      "loss": 0.0033,
      "step": 77770
    },
    {
      "epoch": 0.49885675986826283,
      "grad_norm": 0.09598638862371445,
      "learning_rate": 9.335118372092679e-06,
      "loss": 0.0027,
      "step": 77780
    },
    {
      "epoch": 0.4989208967620489,
      "grad_norm": 0.18568679690361023,
      "learning_rate": 9.33483946324189e-06,
      "loss": 0.0036,
      "step": 77790
    },
    {
      "epoch": 0.498985033655835,
      "grad_norm": 0.04725318402051926,
      "learning_rate": 9.334560500072463e-06,
      "loss": 0.0032,
      "step": 77800
    },
    {
      "epoch": 0.4990491705496211,
      "grad_norm": 0.08631976693868637,
      "learning_rate": 9.334281482587897e-06,
      "loss": 0.0095,
      "step": 77810
    },
    {
      "epoch": 0.4991133074434072,
      "grad_norm": 0.13214997947216034,
      "learning_rate": 9.334002410791685e-06,
      "loss": 0.0038,
      "step": 77820
    },
    {
      "epoch": 0.4991774443371933,
      "grad_norm": 0.10650306940078735,
      "learning_rate": 9.333723284687326e-06,
      "loss": 0.0052,
      "step": 77830
    },
    {
      "epoch": 0.4992415812309794,
      "grad_norm": 0.2304917722940445,
      "learning_rate": 9.333444104278317e-06,
      "loss": 0.0055,
      "step": 77840
    },
    {
      "epoch": 0.4993057181247655,
      "grad_norm": 0.3047102987766266,
      "learning_rate": 9.333164869568156e-06,
      "loss": 0.0041,
      "step": 77850
    },
    {
      "epoch": 0.4993698550185516,
      "grad_norm": 0.2160777449607849,
      "learning_rate": 9.332885580560342e-06,
      "loss": 0.0046,
      "step": 77860
    },
    {
      "epoch": 0.49943399191233767,
      "grad_norm": 0.17151705920696259,
      "learning_rate": 9.332606237258376e-06,
      "loss": 0.0071,
      "step": 77870
    },
    {
      "epoch": 0.49949812880612376,
      "grad_norm": 0.12222845107316971,
      "learning_rate": 9.332326839665758e-06,
      "loss": 0.0034,
      "step": 77880
    },
    {
      "epoch": 0.4995622656999099,
      "grad_norm": 0.07964403182268143,
      "learning_rate": 9.332047387785988e-06,
      "loss": 0.004,
      "step": 77890
    },
    {
      "epoch": 0.499626402593696,
      "grad_norm": 0.10945140570402145,
      "learning_rate": 9.331767881622567e-06,
      "loss": 0.0039,
      "step": 77900
    },
    {
      "epoch": 0.4996905394874821,
      "grad_norm": 0.023508165031671524,
      "learning_rate": 9.331488321178999e-06,
      "loss": 0.0034,
      "step": 77910
    },
    {
      "epoch": 0.4997546763812682,
      "grad_norm": 0.11984086781740189,
      "learning_rate": 9.331208706458787e-06,
      "loss": 0.005,
      "step": 77920
    },
    {
      "epoch": 0.4998188132750543,
      "grad_norm": 0.11363513022661209,
      "learning_rate": 9.330929037465435e-06,
      "loss": 0.0029,
      "step": 77930
    },
    {
      "epoch": 0.4998829501688404,
      "grad_norm": 0.060195907950401306,
      "learning_rate": 9.330649314202444e-06,
      "loss": 0.003,
      "step": 77940
    },
    {
      "epoch": 0.49994708706262647,
      "grad_norm": 0.05659855902194977,
      "learning_rate": 9.330369536673324e-06,
      "loss": 0.003,
      "step": 77950
    },
    {
      "epoch": 0.5000112239564126,
      "grad_norm": 0.48760223388671875,
      "learning_rate": 9.33008970488158e-06,
      "loss": 0.0055,
      "step": 77960
    },
    {
      "epoch": 0.5000753608501987,
      "grad_norm": 0.12812398374080658,
      "learning_rate": 9.329809818830717e-06,
      "loss": 0.0027,
      "step": 77970
    },
    {
      "epoch": 0.5001394977439848,
      "grad_norm": 1.885194182395935,
      "learning_rate": 9.329529878524242e-06,
      "loss": 0.0035,
      "step": 77980
    },
    {
      "epoch": 0.5002036346377708,
      "grad_norm": 0.2891188859939575,
      "learning_rate": 9.329249883965663e-06,
      "loss": 0.0052,
      "step": 77990
    },
    {
      "epoch": 0.500267771531557,
      "grad_norm": 0.15436619520187378,
      "learning_rate": 9.328969835158489e-06,
      "loss": 0.004,
      "step": 78000
    },
    {
      "epoch": 0.500331908425343,
      "grad_norm": 0.32490304112434387,
      "learning_rate": 9.328689732106229e-06,
      "loss": 0.0038,
      "step": 78010
    },
    {
      "epoch": 0.5003960453191292,
      "grad_norm": 0.1172533929347992,
      "learning_rate": 9.328409574812394e-06,
      "loss": 0.0028,
      "step": 78020
    },
    {
      "epoch": 0.5004601822129152,
      "grad_norm": 0.1780451238155365,
      "learning_rate": 9.328129363280492e-06,
      "loss": 0.0029,
      "step": 78030
    },
    {
      "epoch": 0.5005243191067014,
      "grad_norm": 0.13680961728096008,
      "learning_rate": 9.327849097514038e-06,
      "loss": 0.0041,
      "step": 78040
    },
    {
      "epoch": 0.5005884560004874,
      "grad_norm": 0.04050092399120331,
      "learning_rate": 9.327568777516538e-06,
      "loss": 0.0049,
      "step": 78050
    },
    {
      "epoch": 0.5006525928942736,
      "grad_norm": 0.14011479914188385,
      "learning_rate": 9.32728840329151e-06,
      "loss": 0.004,
      "step": 78060
    },
    {
      "epoch": 0.5007167297880596,
      "grad_norm": 0.06274769455194473,
      "learning_rate": 9.327007974842468e-06,
      "loss": 0.0023,
      "step": 78070
    },
    {
      "epoch": 0.5007808666818457,
      "grad_norm": 0.1820869743824005,
      "learning_rate": 9.326727492172921e-06,
      "loss": 0.003,
      "step": 78080
    },
    {
      "epoch": 0.5008450035756318,
      "grad_norm": 0.08584950864315033,
      "learning_rate": 9.326446955286387e-06,
      "loss": 0.0045,
      "step": 78090
    },
    {
      "epoch": 0.5009091404694179,
      "grad_norm": 0.1383316069841385,
      "learning_rate": 9.32616636418638e-06,
      "loss": 0.0029,
      "step": 78100
    },
    {
      "epoch": 0.5009732773632041,
      "grad_norm": 0.23082928359508514,
      "learning_rate": 9.325885718876419e-06,
      "loss": 0.0041,
      "step": 78110
    },
    {
      "epoch": 0.5010374142569901,
      "grad_norm": 0.16334521770477295,
      "learning_rate": 9.325605019360015e-06,
      "loss": 0.0033,
      "step": 78120
    },
    {
      "epoch": 0.5011015511507763,
      "grad_norm": 0.08774948865175247,
      "learning_rate": 9.325324265640692e-06,
      "loss": 0.0035,
      "step": 78130
    },
    {
      "epoch": 0.5011656880445623,
      "grad_norm": 0.1896505355834961,
      "learning_rate": 9.325043457721964e-06,
      "loss": 0.0027,
      "step": 78140
    },
    {
      "epoch": 0.5012298249383484,
      "grad_norm": 0.07469774037599564,
      "learning_rate": 9.324762595607348e-06,
      "loss": 0.0026,
      "step": 78150
    },
    {
      "epoch": 0.5012939618321345,
      "grad_norm": 0.13461455702781677,
      "learning_rate": 9.324481679300366e-06,
      "loss": 0.0036,
      "step": 78160
    },
    {
      "epoch": 0.5013580987259206,
      "grad_norm": 0.12720540165901184,
      "learning_rate": 9.32420070880454e-06,
      "loss": 0.0058,
      "step": 78170
    },
    {
      "epoch": 0.5014222356197067,
      "grad_norm": 0.128628209233284,
      "learning_rate": 9.323919684123388e-06,
      "loss": 0.004,
      "step": 78180
    },
    {
      "epoch": 0.5014863725134928,
      "grad_norm": 0.22741807997226715,
      "learning_rate": 9.323638605260432e-06,
      "loss": 0.0019,
      "step": 78190
    },
    {
      "epoch": 0.5015505094072789,
      "grad_norm": 0.16844815015792847,
      "learning_rate": 9.323357472219195e-06,
      "loss": 0.0042,
      "step": 78200
    },
    {
      "epoch": 0.501614646301065,
      "grad_norm": 0.11421308666467667,
      "learning_rate": 9.323076285003197e-06,
      "loss": 0.0021,
      "step": 78210
    },
    {
      "epoch": 0.501678783194851,
      "grad_norm": 0.12311401963233948,
      "learning_rate": 9.322795043615964e-06,
      "loss": 0.0026,
      "step": 78220
    },
    {
      "epoch": 0.5017429200886372,
      "grad_norm": 0.18353450298309326,
      "learning_rate": 9.32251374806102e-06,
      "loss": 0.0024,
      "step": 78230
    },
    {
      "epoch": 0.5018070569824233,
      "grad_norm": 0.24150238931179047,
      "learning_rate": 9.32223239834189e-06,
      "loss": 0.0039,
      "step": 78240
    },
    {
      "epoch": 0.5018711938762094,
      "grad_norm": 0.1451845020055771,
      "learning_rate": 9.3219509944621e-06,
      "loss": 0.0032,
      "step": 78250
    },
    {
      "epoch": 0.5019353307699955,
      "grad_norm": 0.1299157738685608,
      "learning_rate": 9.321669536425172e-06,
      "loss": 0.0024,
      "step": 78260
    },
    {
      "epoch": 0.5019994676637816,
      "grad_norm": 0.10516326129436493,
      "learning_rate": 9.321388024234638e-06,
      "loss": 0.0033,
      "step": 78270
    },
    {
      "epoch": 0.5020636045575677,
      "grad_norm": 0.26224756240844727,
      "learning_rate": 9.321106457894023e-06,
      "loss": 0.0031,
      "step": 78280
    },
    {
      "epoch": 0.5021277414513537,
      "grad_norm": 0.16212397813796997,
      "learning_rate": 9.320824837406856e-06,
      "loss": 0.0017,
      "step": 78290
    },
    {
      "epoch": 0.5021918783451399,
      "grad_norm": 0.1332699954509735,
      "learning_rate": 9.320543162776667e-06,
      "loss": 0.0041,
      "step": 78300
    },
    {
      "epoch": 0.5022560152389259,
      "grad_norm": 0.06167587637901306,
      "learning_rate": 9.320261434006983e-06,
      "loss": 0.0032,
      "step": 78310
    },
    {
      "epoch": 0.5023201521327121,
      "grad_norm": 0.03406332805752754,
      "learning_rate": 9.319979651101336e-06,
      "loss": 0.0027,
      "step": 78320
    },
    {
      "epoch": 0.5023842890264981,
      "grad_norm": 0.24660973250865936,
      "learning_rate": 9.319697814063257e-06,
      "loss": 0.0038,
      "step": 78330
    },
    {
      "epoch": 0.5024484259202843,
      "grad_norm": 0.0480784997344017,
      "learning_rate": 9.319415922896278e-06,
      "loss": 0.0028,
      "step": 78340
    },
    {
      "epoch": 0.5025125628140703,
      "grad_norm": 0.07655072212219238,
      "learning_rate": 9.319133977603929e-06,
      "loss": 0.0044,
      "step": 78350
    },
    {
      "epoch": 0.5025766997078565,
      "grad_norm": 0.15556862950325012,
      "learning_rate": 9.318851978189744e-06,
      "loss": 0.0043,
      "step": 78360
    },
    {
      "epoch": 0.5026408366016425,
      "grad_norm": 0.19853724539279938,
      "learning_rate": 9.318569924657257e-06,
      "loss": 0.004,
      "step": 78370
    },
    {
      "epoch": 0.5027049734954286,
      "grad_norm": 0.12763874232769012,
      "learning_rate": 9.318287817010003e-06,
      "loss": 0.003,
      "step": 78380
    },
    {
      "epoch": 0.5027691103892148,
      "grad_norm": 0.16955415904521942,
      "learning_rate": 9.318005655251517e-06,
      "loss": 0.0072,
      "step": 78390
    },
    {
      "epoch": 0.5028332472830008,
      "grad_norm": 0.20948928594589233,
      "learning_rate": 9.317723439385333e-06,
      "loss": 0.0033,
      "step": 78400
    },
    {
      "epoch": 0.502897384176787,
      "grad_norm": 0.205460324883461,
      "learning_rate": 9.317441169414989e-06,
      "loss": 0.004,
      "step": 78410
    },
    {
      "epoch": 0.502961521070573,
      "grad_norm": 0.10899683088064194,
      "learning_rate": 9.317158845344022e-06,
      "loss": 0.004,
      "step": 78420
    },
    {
      "epoch": 0.5030256579643592,
      "grad_norm": 0.1819201558828354,
      "learning_rate": 9.316876467175969e-06,
      "loss": 0.009,
      "step": 78430
    },
    {
      "epoch": 0.5030897948581452,
      "grad_norm": 0.1324969232082367,
      "learning_rate": 9.316594034914368e-06,
      "loss": 0.003,
      "step": 78440
    },
    {
      "epoch": 0.5031539317519313,
      "grad_norm": 0.11342664808034897,
      "learning_rate": 9.31631154856276e-06,
      "loss": 0.006,
      "step": 78450
    },
    {
      "epoch": 0.5032180686457174,
      "grad_norm": 0.09595182538032532,
      "learning_rate": 9.316029008124682e-06,
      "loss": 0.0038,
      "step": 78460
    },
    {
      "epoch": 0.5032822055395035,
      "grad_norm": 0.26570847630500793,
      "learning_rate": 9.315746413603676e-06,
      "loss": 0.0034,
      "step": 78470
    },
    {
      "epoch": 0.5033463424332896,
      "grad_norm": 0.14782577753067017,
      "learning_rate": 9.315463765003283e-06,
      "loss": 0.003,
      "step": 78480
    },
    {
      "epoch": 0.5034104793270757,
      "grad_norm": 0.20060312747955322,
      "learning_rate": 9.315181062327046e-06,
      "loss": 0.003,
      "step": 78490
    },
    {
      "epoch": 0.5034746162208618,
      "grad_norm": 0.15654371678829193,
      "learning_rate": 9.314898305578505e-06,
      "loss": 0.0037,
      "step": 78500
    },
    {
      "epoch": 0.5035387531146479,
      "grad_norm": 0.05840962380170822,
      "learning_rate": 9.314615494761206e-06,
      "loss": 0.0029,
      "step": 78510
    },
    {
      "epoch": 0.5036028900084341,
      "grad_norm": 0.1299424022436142,
      "learning_rate": 9.314332629878691e-06,
      "loss": 0.0042,
      "step": 78520
    },
    {
      "epoch": 0.5036670269022201,
      "grad_norm": 0.2413955181837082,
      "learning_rate": 9.314049710934504e-06,
      "loss": 0.0053,
      "step": 78530
    },
    {
      "epoch": 0.5037311637960062,
      "grad_norm": 0.05778886750340462,
      "learning_rate": 9.31376673793219e-06,
      "loss": 0.0028,
      "step": 78540
    },
    {
      "epoch": 0.5037953006897923,
      "grad_norm": 0.015478234738111496,
      "learning_rate": 9.313483710875298e-06,
      "loss": 0.0019,
      "step": 78550
    },
    {
      "epoch": 0.5038594375835784,
      "grad_norm": 0.07730729877948761,
      "learning_rate": 9.313200629767371e-06,
      "loss": 0.0046,
      "step": 78560
    },
    {
      "epoch": 0.5039235744773645,
      "grad_norm": 0.0688692033290863,
      "learning_rate": 9.312917494611957e-06,
      "loss": 0.0039,
      "step": 78570
    },
    {
      "epoch": 0.5039877113711506,
      "grad_norm": 0.29817402362823486,
      "learning_rate": 9.312634305412604e-06,
      "loss": 0.0118,
      "step": 78580
    },
    {
      "epoch": 0.5040518482649367,
      "grad_norm": 0.09822961688041687,
      "learning_rate": 9.312351062172862e-06,
      "loss": 0.0036,
      "step": 78590
    },
    {
      "epoch": 0.5041159851587228,
      "grad_norm": 0.2513461410999298,
      "learning_rate": 9.312067764896279e-06,
      "loss": 0.0049,
      "step": 78600
    },
    {
      "epoch": 0.5041801220525088,
      "grad_norm": 0.13050639629364014,
      "learning_rate": 9.311784413586407e-06,
      "loss": 0.0034,
      "step": 78610
    },
    {
      "epoch": 0.504244258946295,
      "grad_norm": 0.44473275542259216,
      "learning_rate": 9.311501008246792e-06,
      "loss": 0.0056,
      "step": 78620
    },
    {
      "epoch": 0.504308395840081,
      "grad_norm": 0.07539486885070801,
      "learning_rate": 9.311217548880988e-06,
      "loss": 0.0031,
      "step": 78630
    },
    {
      "epoch": 0.5043725327338672,
      "grad_norm": 0.0902412086725235,
      "learning_rate": 9.31093403549255e-06,
      "loss": 0.0031,
      "step": 78640
    },
    {
      "epoch": 0.5044366696276532,
      "grad_norm": 0.15105240046977997,
      "learning_rate": 9.310650468085023e-06,
      "loss": 0.0075,
      "step": 78650
    },
    {
      "epoch": 0.5045008065214394,
      "grad_norm": 0.15025536715984344,
      "learning_rate": 9.310366846661969e-06,
      "loss": 0.0035,
      "step": 78660
    },
    {
      "epoch": 0.5045649434152255,
      "grad_norm": 0.18562453985214233,
      "learning_rate": 9.310083171226935e-06,
      "loss": 0.0054,
      "step": 78670
    },
    {
      "epoch": 0.5046290803090115,
      "grad_norm": 0.2005065530538559,
      "learning_rate": 9.30979944178348e-06,
      "loss": 0.0048,
      "step": 78680
    },
    {
      "epoch": 0.5046932172027977,
      "grad_norm": 0.08502445369958878,
      "learning_rate": 9.309515658335158e-06,
      "loss": 0.0038,
      "step": 78690
    },
    {
      "epoch": 0.5047573540965837,
      "grad_norm": 0.06351223587989807,
      "learning_rate": 9.309231820885523e-06,
      "loss": 0.0024,
      "step": 78700
    },
    {
      "epoch": 0.5048214909903699,
      "grad_norm": 0.13348788022994995,
      "learning_rate": 9.308947929438135e-06,
      "loss": 0.003,
      "step": 78710
    },
    {
      "epoch": 0.5048856278841559,
      "grad_norm": 0.08967173844575882,
      "learning_rate": 9.30866398399655e-06,
      "loss": 0.0032,
      "step": 78720
    },
    {
      "epoch": 0.5049497647779421,
      "grad_norm": 0.20979301631450653,
      "learning_rate": 9.308379984564326e-06,
      "loss": 0.0043,
      "step": 78730
    },
    {
      "epoch": 0.5050139016717281,
      "grad_norm": 0.18976452946662903,
      "learning_rate": 9.308095931145021e-06,
      "loss": 0.0068,
      "step": 78740
    },
    {
      "epoch": 0.5050780385655143,
      "grad_norm": 0.048686880618333817,
      "learning_rate": 9.307811823742195e-06,
      "loss": 0.0039,
      "step": 78750
    },
    {
      "epoch": 0.5051421754593003,
      "grad_norm": 0.2524215579032898,
      "learning_rate": 9.307527662359408e-06,
      "loss": 0.0054,
      "step": 78760
    },
    {
      "epoch": 0.5052063123530864,
      "grad_norm": 0.2951013445854187,
      "learning_rate": 9.307243447000222e-06,
      "loss": 0.0044,
      "step": 78770
    },
    {
      "epoch": 0.5052704492468725,
      "grad_norm": 0.28178176283836365,
      "learning_rate": 9.306959177668195e-06,
      "loss": 0.0052,
      "step": 78780
    },
    {
      "epoch": 0.5053345861406586,
      "grad_norm": 0.1739809662103653,
      "learning_rate": 9.306674854366895e-06,
      "loss": 0.0027,
      "step": 78790
    },
    {
      "epoch": 0.5053987230344447,
      "grad_norm": 0.18351706862449646,
      "learning_rate": 9.306390477099878e-06,
      "loss": 0.0059,
      "step": 78800
    },
    {
      "epoch": 0.5054628599282308,
      "grad_norm": 0.15412580966949463,
      "learning_rate": 9.30610604587071e-06,
      "loss": 0.004,
      "step": 78810
    },
    {
      "epoch": 0.505526996822017,
      "grad_norm": 0.08657550066709518,
      "learning_rate": 9.305821560682959e-06,
      "loss": 0.006,
      "step": 78820
    },
    {
      "epoch": 0.505591133715803,
      "grad_norm": 4.012050151824951,
      "learning_rate": 9.305537021540186e-06,
      "loss": 0.0118,
      "step": 78830
    },
    {
      "epoch": 0.5056552706095891,
      "grad_norm": 0.16037659347057343,
      "learning_rate": 9.305252428445954e-06,
      "loss": 0.0055,
      "step": 78840
    },
    {
      "epoch": 0.5057194075033752,
      "grad_norm": 0.0963616594672203,
      "learning_rate": 9.304967781403835e-06,
      "loss": 0.0057,
      "step": 78850
    },
    {
      "epoch": 0.5057835443971613,
      "grad_norm": 0.11402077227830887,
      "learning_rate": 9.304683080417392e-06,
      "loss": 0.0045,
      "step": 78860
    },
    {
      "epoch": 0.5058476812909474,
      "grad_norm": 0.13955287635326385,
      "learning_rate": 9.304398325490194e-06,
      "loss": 0.0032,
      "step": 78870
    },
    {
      "epoch": 0.5059118181847335,
      "grad_norm": 0.3319750428199768,
      "learning_rate": 9.304113516625808e-06,
      "loss": 0.0073,
      "step": 78880
    },
    {
      "epoch": 0.5059759550785196,
      "grad_norm": 0.09229553490877151,
      "learning_rate": 9.303828653827802e-06,
      "loss": 0.0074,
      "step": 78890
    },
    {
      "epoch": 0.5060400919723057,
      "grad_norm": 0.21243679523468018,
      "learning_rate": 9.303543737099749e-06,
      "loss": 0.004,
      "step": 78900
    },
    {
      "epoch": 0.5061042288660917,
      "grad_norm": 0.20914272964000702,
      "learning_rate": 9.303258766445218e-06,
      "loss": 0.0051,
      "step": 78910
    },
    {
      "epoch": 0.5061683657598779,
      "grad_norm": 0.10656081140041351,
      "learning_rate": 9.302973741867778e-06,
      "loss": 0.0029,
      "step": 78920
    },
    {
      "epoch": 0.5062325026536639,
      "grad_norm": 0.08997952193021774,
      "learning_rate": 9.302688663371002e-06,
      "loss": 0.0058,
      "step": 78930
    },
    {
      "epoch": 0.5062966395474501,
      "grad_norm": 0.1475543975830078,
      "learning_rate": 9.302403530958462e-06,
      "loss": 0.0037,
      "step": 78940
    },
    {
      "epoch": 0.5063607764412362,
      "grad_norm": 0.1437944918870926,
      "learning_rate": 9.30211834463373e-06,
      "loss": 0.0034,
      "step": 78950
    },
    {
      "epoch": 0.5064249133350223,
      "grad_norm": 0.17114025354385376,
      "learning_rate": 9.301833104400382e-06,
      "loss": 0.0041,
      "step": 78960
    },
    {
      "epoch": 0.5064890502288084,
      "grad_norm": 0.11799921095371246,
      "learning_rate": 9.30154781026199e-06,
      "loss": 0.0031,
      "step": 78970
    },
    {
      "epoch": 0.5065531871225945,
      "grad_norm": 0.28136688470840454,
      "learning_rate": 9.30126246222213e-06,
      "loss": 0.0058,
      "step": 78980
    },
    {
      "epoch": 0.5066173240163806,
      "grad_norm": 0.10884694010019302,
      "learning_rate": 9.300977060284377e-06,
      "loss": 0.0027,
      "step": 78990
    },
    {
      "epoch": 0.5066814609101666,
      "grad_norm": 0.3540794849395752,
      "learning_rate": 9.300691604452306e-06,
      "loss": 0.0052,
      "step": 79000
    },
    {
      "epoch": 0.5067455978039528,
      "grad_norm": 0.24561040103435516,
      "learning_rate": 9.300406094729496e-06,
      "loss": 0.0041,
      "step": 79010
    },
    {
      "epoch": 0.5068097346977388,
      "grad_norm": 0.09596218168735504,
      "learning_rate": 9.300120531119527e-06,
      "loss": 0.0046,
      "step": 79020
    },
    {
      "epoch": 0.506873871591525,
      "grad_norm": 0.09170177578926086,
      "learning_rate": 9.299834913625971e-06,
      "loss": 0.0043,
      "step": 79030
    },
    {
      "epoch": 0.506938008485311,
      "grad_norm": 0.06456504762172699,
      "learning_rate": 9.299549242252414e-06,
      "loss": 0.0046,
      "step": 79040
    },
    {
      "epoch": 0.5070021453790972,
      "grad_norm": 0.24337686598300934,
      "learning_rate": 9.29926351700243e-06,
      "loss": 0.0029,
      "step": 79050
    },
    {
      "epoch": 0.5070662822728832,
      "grad_norm": 0.2393103688955307,
      "learning_rate": 9.298977737879602e-06,
      "loss": 0.0053,
      "step": 79060
    },
    {
      "epoch": 0.5071304191666693,
      "grad_norm": 0.2494836449623108,
      "learning_rate": 9.298691904887508e-06,
      "loss": 0.0025,
      "step": 79070
    },
    {
      "epoch": 0.5071945560604554,
      "grad_norm": 0.42681920528411865,
      "learning_rate": 9.298406018029737e-06,
      "loss": 0.0084,
      "step": 79080
    },
    {
      "epoch": 0.5072586929542415,
      "grad_norm": 0.13219456374645233,
      "learning_rate": 9.298120077309864e-06,
      "loss": 0.0051,
      "step": 79090
    },
    {
      "epoch": 0.5073228298480277,
      "grad_norm": 0.13975024223327637,
      "learning_rate": 9.297834082731474e-06,
      "loss": 0.0024,
      "step": 79100
    },
    {
      "epoch": 0.5073869667418137,
      "grad_norm": 0.2604387402534485,
      "learning_rate": 9.297548034298151e-06,
      "loss": 0.0063,
      "step": 79110
    },
    {
      "epoch": 0.5074511036355999,
      "grad_norm": 0.1672421097755432,
      "learning_rate": 9.29726193201348e-06,
      "loss": 0.0029,
      "step": 79120
    },
    {
      "epoch": 0.5075152405293859,
      "grad_norm": 0.11025706678628922,
      "learning_rate": 9.296975775881049e-06,
      "loss": 0.0029,
      "step": 79130
    },
    {
      "epoch": 0.507579377423172,
      "grad_norm": 0.19144493341445923,
      "learning_rate": 9.296689565904437e-06,
      "loss": 0.0033,
      "step": 79140
    },
    {
      "epoch": 0.5076435143169581,
      "grad_norm": 0.06302493065595627,
      "learning_rate": 9.296403302087236e-06,
      "loss": 0.0029,
      "step": 79150
    },
    {
      "epoch": 0.5077076512107442,
      "grad_norm": 0.12776464223861694,
      "learning_rate": 9.29611698443303e-06,
      "loss": 0.0036,
      "step": 79160
    },
    {
      "epoch": 0.5077717881045303,
      "grad_norm": 0.10405514389276505,
      "learning_rate": 9.295830612945406e-06,
      "loss": 0.0022,
      "step": 79170
    },
    {
      "epoch": 0.5078359249983164,
      "grad_norm": 0.12541966140270233,
      "learning_rate": 9.295544187627957e-06,
      "loss": 0.0044,
      "step": 79180
    },
    {
      "epoch": 0.5079000618921025,
      "grad_norm": 0.2913911044597626,
      "learning_rate": 9.295257708484269e-06,
      "loss": 0.0058,
      "step": 79190
    },
    {
      "epoch": 0.5079641987858886,
      "grad_norm": 0.11696656048297882,
      "learning_rate": 9.294971175517931e-06,
      "loss": 0.0033,
      "step": 79200
    },
    {
      "epoch": 0.5080283356796746,
      "grad_norm": 0.18840357661247253,
      "learning_rate": 9.294684588732536e-06,
      "loss": 0.0021,
      "step": 79210
    },
    {
      "epoch": 0.5080924725734608,
      "grad_norm": 0.17457997798919678,
      "learning_rate": 9.294397948131673e-06,
      "loss": 0.0061,
      "step": 79220
    },
    {
      "epoch": 0.508156609467247,
      "grad_norm": 0.14138942956924438,
      "learning_rate": 9.294111253718934e-06,
      "loss": 0.0028,
      "step": 79230
    },
    {
      "epoch": 0.508220746361033,
      "grad_norm": 0.10878881067037582,
      "learning_rate": 9.293824505497912e-06,
      "loss": 0.0028,
      "step": 79240
    },
    {
      "epoch": 0.5082848832548191,
      "grad_norm": 0.14829964935779572,
      "learning_rate": 9.2935377034722e-06,
      "loss": 0.0052,
      "step": 79250
    },
    {
      "epoch": 0.5083490201486052,
      "grad_norm": 0.11220408231019974,
      "learning_rate": 9.293250847645394e-06,
      "loss": 0.0057,
      "step": 79260
    },
    {
      "epoch": 0.5084131570423913,
      "grad_norm": 0.11413662135601044,
      "learning_rate": 9.292963938021085e-06,
      "loss": 0.0035,
      "step": 79270
    },
    {
      "epoch": 0.5084772939361774,
      "grad_norm": 0.20025412738323212,
      "learning_rate": 9.29267697460287e-06,
      "loss": 0.0027,
      "step": 79280
    },
    {
      "epoch": 0.5085414308299635,
      "grad_norm": 0.1414087861776352,
      "learning_rate": 9.292389957394345e-06,
      "loss": 0.0083,
      "step": 79290
    },
    {
      "epoch": 0.5086055677237495,
      "grad_norm": 0.5542187094688416,
      "learning_rate": 9.292102886399104e-06,
      "loss": 0.0052,
      "step": 79300
    },
    {
      "epoch": 0.5086697046175357,
      "grad_norm": 0.1466957926750183,
      "learning_rate": 9.291815761620748e-06,
      "loss": 0.0033,
      "step": 79310
    },
    {
      "epoch": 0.5087338415113217,
      "grad_norm": 0.051578179001808167,
      "learning_rate": 9.291528583062873e-06,
      "loss": 0.0026,
      "step": 79320
    },
    {
      "epoch": 0.5087979784051079,
      "grad_norm": 0.07024681568145752,
      "learning_rate": 9.291241350729078e-06,
      "loss": 0.0044,
      "step": 79330
    },
    {
      "epoch": 0.5088621152988939,
      "grad_norm": 0.083675317466259,
      "learning_rate": 9.290954064622963e-06,
      "loss": 0.0033,
      "step": 79340
    },
    {
      "epoch": 0.5089262521926801,
      "grad_norm": 0.06051542982459068,
      "learning_rate": 9.290666724748125e-06,
      "loss": 0.0018,
      "step": 79350
    },
    {
      "epoch": 0.5089903890864661,
      "grad_norm": 0.11042298376560211,
      "learning_rate": 9.290379331108168e-06,
      "loss": 0.0029,
      "step": 79360
    },
    {
      "epoch": 0.5090545259802522,
      "grad_norm": 0.3282707631587982,
      "learning_rate": 9.290091883706692e-06,
      "loss": 0.0031,
      "step": 79370
    },
    {
      "epoch": 0.5091186628740384,
      "grad_norm": 0.25048360228538513,
      "learning_rate": 9.289804382547296e-06,
      "loss": 0.0031,
      "step": 79380
    },
    {
      "epoch": 0.5091827997678244,
      "grad_norm": 0.12474612891674042,
      "learning_rate": 9.289516827633587e-06,
      "loss": 0.0027,
      "step": 79390
    },
    {
      "epoch": 0.5092469366616106,
      "grad_norm": 0.0450366735458374,
      "learning_rate": 9.289229218969166e-06,
      "loss": 0.002,
      "step": 79400
    },
    {
      "epoch": 0.5093110735553966,
      "grad_norm": 0.20078837871551514,
      "learning_rate": 9.288941556557639e-06,
      "loss": 0.0036,
      "step": 79410
    },
    {
      "epoch": 0.5093752104491828,
      "grad_norm": 0.04043214023113251,
      "learning_rate": 9.288653840402607e-06,
      "loss": 0.0022,
      "step": 79420
    },
    {
      "epoch": 0.5094393473429688,
      "grad_norm": 0.16537418961524963,
      "learning_rate": 9.288366070507677e-06,
      "loss": 0.0038,
      "step": 79430
    },
    {
      "epoch": 0.509503484236755,
      "grad_norm": 0.08483941853046417,
      "learning_rate": 9.288078246876456e-06,
      "loss": 0.0118,
      "step": 79440
    },
    {
      "epoch": 0.509567621130541,
      "grad_norm": 0.10477130115032196,
      "learning_rate": 9.28779036951255e-06,
      "loss": 0.0042,
      "step": 79450
    },
    {
      "epoch": 0.5096317580243271,
      "grad_norm": 0.10791230201721191,
      "learning_rate": 9.287502438419567e-06,
      "loss": 0.0041,
      "step": 79460
    },
    {
      "epoch": 0.5096958949181132,
      "grad_norm": 0.17048099637031555,
      "learning_rate": 9.287214453601115e-06,
      "loss": 0.0027,
      "step": 79470
    },
    {
      "epoch": 0.5097600318118993,
      "grad_norm": 0.20219752192497253,
      "learning_rate": 9.2869264150608e-06,
      "loss": 0.005,
      "step": 79480
    },
    {
      "epoch": 0.5098241687056854,
      "grad_norm": 0.08709075301885605,
      "learning_rate": 9.286638322802233e-06,
      "loss": 0.003,
      "step": 79490
    },
    {
      "epoch": 0.5098883055994715,
      "grad_norm": 0.23793639242649078,
      "learning_rate": 9.286350176829024e-06,
      "loss": 0.0026,
      "step": 79500
    },
    {
      "epoch": 0.5099524424932577,
      "grad_norm": 0.122853584587574,
      "learning_rate": 9.286061977144786e-06,
      "loss": 0.0042,
      "step": 79510
    },
    {
      "epoch": 0.5100165793870437,
      "grad_norm": 0.21153560280799866,
      "learning_rate": 9.285773723753127e-06,
      "loss": 0.0054,
      "step": 79520
    },
    {
      "epoch": 0.5100807162808298,
      "grad_norm": 0.1601635068655014,
      "learning_rate": 9.28548541665766e-06,
      "loss": 0.0036,
      "step": 79530
    },
    {
      "epoch": 0.5101448531746159,
      "grad_norm": 0.10221725702285767,
      "learning_rate": 9.285197055861998e-06,
      "loss": 0.0037,
      "step": 79540
    },
    {
      "epoch": 0.510208990068402,
      "grad_norm": 0.10448180884122849,
      "learning_rate": 9.284908641369755e-06,
      "loss": 0.0049,
      "step": 79550
    },
    {
      "epoch": 0.5102731269621881,
      "grad_norm": 0.31749647855758667,
      "learning_rate": 9.284620173184545e-06,
      "loss": 0.0026,
      "step": 79560
    },
    {
      "epoch": 0.5103372638559742,
      "grad_norm": 0.21306177973747253,
      "learning_rate": 9.28433165130998e-06,
      "loss": 0.0039,
      "step": 79570
    },
    {
      "epoch": 0.5104014007497603,
      "grad_norm": 0.03203563764691353,
      "learning_rate": 9.284043075749678e-06,
      "loss": 0.004,
      "step": 79580
    },
    {
      "epoch": 0.5104655376435464,
      "grad_norm": 0.05935882776975632,
      "learning_rate": 9.283754446507253e-06,
      "loss": 0.0032,
      "step": 79590
    },
    {
      "epoch": 0.5105296745373324,
      "grad_norm": 0.05813802778720856,
      "learning_rate": 9.283465763586325e-06,
      "loss": 0.0025,
      "step": 79600
    },
    {
      "epoch": 0.5105938114311186,
      "grad_norm": 0.08461745083332062,
      "learning_rate": 9.283177026990512e-06,
      "loss": 0.0031,
      "step": 79610
    },
    {
      "epoch": 0.5106579483249046,
      "grad_norm": 0.07877620309591293,
      "learning_rate": 9.282888236723426e-06,
      "loss": 0.0053,
      "step": 79620
    },
    {
      "epoch": 0.5107220852186908,
      "grad_norm": 0.15824204683303833,
      "learning_rate": 9.282599392788692e-06,
      "loss": 0.003,
      "step": 79630
    },
    {
      "epoch": 0.5107862221124768,
      "grad_norm": 0.05316735431551933,
      "learning_rate": 9.282310495189926e-06,
      "loss": 0.0025,
      "step": 79640
    },
    {
      "epoch": 0.510850359006263,
      "grad_norm": 0.18523530662059784,
      "learning_rate": 9.282021543930748e-06,
      "loss": 0.0023,
      "step": 79650
    },
    {
      "epoch": 0.5109144959000491,
      "grad_norm": 0.37745577096939087,
      "learning_rate": 9.28173253901478e-06,
      "loss": 0.0052,
      "step": 79660
    },
    {
      "epoch": 0.5109786327938352,
      "grad_norm": 0.1288391649723053,
      "learning_rate": 9.281443480445644e-06,
      "loss": 0.0024,
      "step": 79670
    },
    {
      "epoch": 0.5110427696876213,
      "grad_norm": 0.20558273792266846,
      "learning_rate": 9.281154368226961e-06,
      "loss": 0.0057,
      "step": 79680
    },
    {
      "epoch": 0.5111069065814073,
      "grad_norm": 0.0038497569039463997,
      "learning_rate": 9.280865202362355e-06,
      "loss": 0.003,
      "step": 79690
    },
    {
      "epoch": 0.5111710434751935,
      "grad_norm": 0.11807820200920105,
      "learning_rate": 9.280575982855447e-06,
      "loss": 0.0036,
      "step": 79700
    },
    {
      "epoch": 0.5112351803689795,
      "grad_norm": 0.057972654700279236,
      "learning_rate": 9.280286709709862e-06,
      "loss": 0.0058,
      "step": 79710
    },
    {
      "epoch": 0.5112993172627657,
      "grad_norm": 0.3088960647583008,
      "learning_rate": 9.279997382929227e-06,
      "loss": 0.0075,
      "step": 79720
    },
    {
      "epoch": 0.5113634541565517,
      "grad_norm": 0.10937351733446121,
      "learning_rate": 9.279708002517166e-06,
      "loss": 0.004,
      "step": 79730
    },
    {
      "epoch": 0.5114275910503379,
      "grad_norm": 0.09601975977420807,
      "learning_rate": 9.279418568477305e-06,
      "loss": 0.0036,
      "step": 79740
    },
    {
      "epoch": 0.5114917279441239,
      "grad_norm": 0.11423894762992859,
      "learning_rate": 9.279129080813269e-06,
      "loss": 0.0025,
      "step": 79750
    },
    {
      "epoch": 0.51155586483791,
      "grad_norm": 0.16922509670257568,
      "learning_rate": 9.27883953952869e-06,
      "loss": 0.0029,
      "step": 79760
    },
    {
      "epoch": 0.5116200017316961,
      "grad_norm": 0.1398465782403946,
      "learning_rate": 9.278549944627192e-06,
      "loss": 0.0022,
      "step": 79770
    },
    {
      "epoch": 0.5116841386254822,
      "grad_norm": 0.2333763986825943,
      "learning_rate": 9.278260296112406e-06,
      "loss": 0.0045,
      "step": 79780
    },
    {
      "epoch": 0.5117482755192684,
      "grad_norm": 0.07165589928627014,
      "learning_rate": 9.277970593987961e-06,
      "loss": 0.0045,
      "step": 79790
    },
    {
      "epoch": 0.5118124124130544,
      "grad_norm": 0.1786176860332489,
      "learning_rate": 9.277680838257486e-06,
      "loss": 0.0036,
      "step": 79800
    },
    {
      "epoch": 0.5118765493068406,
      "grad_norm": 0.1743316501379013,
      "learning_rate": 9.277391028924614e-06,
      "loss": 0.0049,
      "step": 79810
    },
    {
      "epoch": 0.5119406862006266,
      "grad_norm": 0.1622972935438156,
      "learning_rate": 9.277101165992975e-06,
      "loss": 0.0024,
      "step": 79820
    },
    {
      "epoch": 0.5120048230944128,
      "grad_norm": 0.02936510369181633,
      "learning_rate": 9.276811249466201e-06,
      "loss": 0.0037,
      "step": 79830
    },
    {
      "epoch": 0.5120689599881988,
      "grad_norm": 0.13769282400608063,
      "learning_rate": 9.276521279347926e-06,
      "loss": 0.0026,
      "step": 79840
    },
    {
      "epoch": 0.5121330968819849,
      "grad_norm": 0.03349493816494942,
      "learning_rate": 9.276231255641783e-06,
      "loss": 0.0061,
      "step": 79850
    },
    {
      "epoch": 0.512197233775771,
      "grad_norm": 0.23810327053070068,
      "learning_rate": 9.275941178351406e-06,
      "loss": 0.003,
      "step": 79860
    },
    {
      "epoch": 0.5122613706695571,
      "grad_norm": 0.06807134300470352,
      "learning_rate": 9.275651047480431e-06,
      "loss": 0.0028,
      "step": 79870
    },
    {
      "epoch": 0.5123255075633432,
      "grad_norm": 0.11463180184364319,
      "learning_rate": 9.275360863032492e-06,
      "loss": 0.003,
      "step": 79880
    },
    {
      "epoch": 0.5123896444571293,
      "grad_norm": 0.2272919863462448,
      "learning_rate": 9.275070625011226e-06,
      "loss": 0.0045,
      "step": 79890
    },
    {
      "epoch": 0.5124537813509153,
      "grad_norm": 0.060492340475320816,
      "learning_rate": 9.27478033342027e-06,
      "loss": 0.002,
      "step": 79900
    },
    {
      "epoch": 0.5125179182447015,
      "grad_norm": 0.1360568255186081,
      "learning_rate": 9.27448998826326e-06,
      "loss": 0.0068,
      "step": 79910
    },
    {
      "epoch": 0.5125820551384875,
      "grad_norm": 0.5712762475013733,
      "learning_rate": 9.274199589543836e-06,
      "loss": 0.0082,
      "step": 79920
    },
    {
      "epoch": 0.5126461920322737,
      "grad_norm": 0.13371999561786652,
      "learning_rate": 9.273909137265637e-06,
      "loss": 0.004,
      "step": 79930
    },
    {
      "epoch": 0.5127103289260598,
      "grad_norm": 0.14738905429840088,
      "learning_rate": 9.273618631432301e-06,
      "loss": 0.003,
      "step": 79940
    },
    {
      "epoch": 0.5127744658198459,
      "grad_norm": 0.08841729909181595,
      "learning_rate": 9.27332807204747e-06,
      "loss": 0.0035,
      "step": 79950
    },
    {
      "epoch": 0.512838602713632,
      "grad_norm": 0.19606488943099976,
      "learning_rate": 9.273037459114784e-06,
      "loss": 0.0041,
      "step": 79960
    },
    {
      "epoch": 0.5129027396074181,
      "grad_norm": 0.05790529400110245,
      "learning_rate": 9.272746792637885e-06,
      "loss": 0.0047,
      "step": 79970
    },
    {
      "epoch": 0.5129668765012042,
      "grad_norm": 0.22945904731750488,
      "learning_rate": 9.272456072620413e-06,
      "loss": 0.0049,
      "step": 79980
    },
    {
      "epoch": 0.5130310133949902,
      "grad_norm": 0.05827128514647484,
      "learning_rate": 9.272165299066016e-06,
      "loss": 0.0038,
      "step": 79990
    },
    {
      "epoch": 0.5130951502887764,
      "grad_norm": 0.10648482292890549,
      "learning_rate": 9.271874471978333e-06,
      "loss": 0.0018,
      "step": 80000
    },
    {
      "epoch": 0.5131592871825624,
      "grad_norm": 0.10429185628890991,
      "learning_rate": 9.27158359136101e-06,
      "loss": 0.0032,
      "step": 80010
    },
    {
      "epoch": 0.5132234240763486,
      "grad_norm": 0.12299264967441559,
      "learning_rate": 9.271292657217692e-06,
      "loss": 0.0034,
      "step": 80020
    },
    {
      "epoch": 0.5132875609701346,
      "grad_norm": 0.16474628448486328,
      "learning_rate": 9.271001669552024e-06,
      "loss": 0.003,
      "step": 80030
    },
    {
      "epoch": 0.5133516978639208,
      "grad_norm": 0.08531386405229568,
      "learning_rate": 9.270710628367653e-06,
      "loss": 0.0034,
      "step": 80040
    },
    {
      "epoch": 0.5134158347577068,
      "grad_norm": 0.18470297753810883,
      "learning_rate": 9.270419533668225e-06,
      "loss": 0.0025,
      "step": 80050
    },
    {
      "epoch": 0.513479971651493,
      "grad_norm": 0.09929078817367554,
      "learning_rate": 9.270128385457389e-06,
      "loss": 0.0053,
      "step": 80060
    },
    {
      "epoch": 0.5135441085452791,
      "grad_norm": 0.1945580393075943,
      "learning_rate": 9.269837183738792e-06,
      "loss": 0.0038,
      "step": 80070
    },
    {
      "epoch": 0.5136082454390651,
      "grad_norm": 0.07254920899868011,
      "learning_rate": 9.269545928516083e-06,
      "loss": 0.0035,
      "step": 80080
    },
    {
      "epoch": 0.5136723823328513,
      "grad_norm": 0.1123448982834816,
      "learning_rate": 9.269254619792914e-06,
      "loss": 0.0019,
      "step": 80090
    },
    {
      "epoch": 0.5137365192266373,
      "grad_norm": 0.5760213136672974,
      "learning_rate": 9.268963257572932e-06,
      "loss": 0.0043,
      "step": 80100
    },
    {
      "epoch": 0.5138006561204235,
      "grad_norm": 0.11048474907875061,
      "learning_rate": 9.268671841859789e-06,
      "loss": 0.0022,
      "step": 80110
    },
    {
      "epoch": 0.5138647930142095,
      "grad_norm": 0.34163668751716614,
      "learning_rate": 9.268380372657137e-06,
      "loss": 0.0031,
      "step": 80120
    },
    {
      "epoch": 0.5139289299079957,
      "grad_norm": 0.13767167925834656,
      "learning_rate": 9.268088849968629e-06,
      "loss": 0.0035,
      "step": 80130
    },
    {
      "epoch": 0.5139930668017817,
      "grad_norm": 0.15491144359111786,
      "learning_rate": 9.267797273797918e-06,
      "loss": 0.003,
      "step": 80140
    },
    {
      "epoch": 0.5140572036955678,
      "grad_norm": 0.0027098332066088915,
      "learning_rate": 9.267505644148655e-06,
      "loss": 0.0028,
      "step": 80150
    },
    {
      "epoch": 0.5141213405893539,
      "grad_norm": 0.391956627368927,
      "learning_rate": 9.267213961024499e-06,
      "loss": 0.0038,
      "step": 80160
    },
    {
      "epoch": 0.51418547748314,
      "grad_norm": 0.3579157292842865,
      "learning_rate": 9.2669222244291e-06,
      "loss": 0.0031,
      "step": 80170
    },
    {
      "epoch": 0.5142496143769261,
      "grad_norm": 0.3213461935520172,
      "learning_rate": 9.266630434366118e-06,
      "loss": 0.0036,
      "step": 80180
    },
    {
      "epoch": 0.5143137512707122,
      "grad_norm": 0.12931661307811737,
      "learning_rate": 9.266338590839205e-06,
      "loss": 0.0039,
      "step": 80190
    },
    {
      "epoch": 0.5143778881644983,
      "grad_norm": 0.08285976946353912,
      "learning_rate": 9.266046693852023e-06,
      "loss": 0.0022,
      "step": 80200
    },
    {
      "epoch": 0.5144420250582844,
      "grad_norm": 0.24085961282253265,
      "learning_rate": 9.265754743408225e-06,
      "loss": 0.004,
      "step": 80210
    },
    {
      "epoch": 0.5145061619520706,
      "grad_norm": 0.25267624855041504,
      "learning_rate": 9.265462739511473e-06,
      "loss": 0.0053,
      "step": 80220
    },
    {
      "epoch": 0.5145702988458566,
      "grad_norm": 0.12650226056575775,
      "learning_rate": 9.265170682165423e-06,
      "loss": 0.0023,
      "step": 80230
    },
    {
      "epoch": 0.5146344357396427,
      "grad_norm": 0.09412693977355957,
      "learning_rate": 9.264878571373737e-06,
      "loss": 0.0038,
      "step": 80240
    },
    {
      "epoch": 0.5146985726334288,
      "grad_norm": 0.1114264577627182,
      "learning_rate": 9.264586407140074e-06,
      "loss": 0.0034,
      "step": 80250
    },
    {
      "epoch": 0.5147627095272149,
      "grad_norm": 0.22017377614974976,
      "learning_rate": 9.264294189468095e-06,
      "loss": 0.0026,
      "step": 80260
    },
    {
      "epoch": 0.514826846421001,
      "grad_norm": 0.11743912845849991,
      "learning_rate": 9.264001918361462e-06,
      "loss": 0.0029,
      "step": 80270
    },
    {
      "epoch": 0.5148909833147871,
      "grad_norm": 0.2741183638572693,
      "learning_rate": 9.263709593823839e-06,
      "loss": 0.0059,
      "step": 80280
    },
    {
      "epoch": 0.5149551202085731,
      "grad_norm": 0.12666159868240356,
      "learning_rate": 9.263417215858886e-06,
      "loss": 0.0028,
      "step": 80290
    },
    {
      "epoch": 0.5150192571023593,
      "grad_norm": 1.0686538219451904,
      "learning_rate": 9.26312478447027e-06,
      "loss": 0.0044,
      "step": 80300
    },
    {
      "epoch": 0.5150833939961453,
      "grad_norm": 0.42256465554237366,
      "learning_rate": 9.262832299661652e-06,
      "loss": 0.0057,
      "step": 80310
    },
    {
      "epoch": 0.5151475308899315,
      "grad_norm": 0.09631174802780151,
      "learning_rate": 9.2625397614367e-06,
      "loss": 0.0033,
      "step": 80320
    },
    {
      "epoch": 0.5152116677837175,
      "grad_norm": 0.17305755615234375,
      "learning_rate": 9.262247169799077e-06,
      "loss": 0.0045,
      "step": 80330
    },
    {
      "epoch": 0.5152758046775037,
      "grad_norm": 0.2103009968996048,
      "learning_rate": 9.261954524752452e-06,
      "loss": 0.0034,
      "step": 80340
    },
    {
      "epoch": 0.5153399415712897,
      "grad_norm": 0.19020456075668335,
      "learning_rate": 9.26166182630049e-06,
      "loss": 0.0036,
      "step": 80350
    },
    {
      "epoch": 0.5154040784650759,
      "grad_norm": 0.24571269750595093,
      "learning_rate": 9.26136907444686e-06,
      "loss": 0.0048,
      "step": 80360
    },
    {
      "epoch": 0.515468215358862,
      "grad_norm": 0.5711245536804199,
      "learning_rate": 9.261076269195229e-06,
      "loss": 0.0046,
      "step": 80370
    },
    {
      "epoch": 0.515532352252648,
      "grad_norm": 0.1847742795944214,
      "learning_rate": 9.260783410549268e-06,
      "loss": 0.0047,
      "step": 80380
    },
    {
      "epoch": 0.5155964891464342,
      "grad_norm": 0.2800198197364807,
      "learning_rate": 9.260490498512647e-06,
      "loss": 0.0054,
      "step": 80390
    },
    {
      "epoch": 0.5156606260402202,
      "grad_norm": 0.18284857273101807,
      "learning_rate": 9.260197533089032e-06,
      "loss": 0.0036,
      "step": 80400
    },
    {
      "epoch": 0.5157247629340064,
      "grad_norm": 0.06317662447690964,
      "learning_rate": 9.259904514282099e-06,
      "loss": 0.0052,
      "step": 80410
    },
    {
      "epoch": 0.5157888998277924,
      "grad_norm": 0.08233999460935593,
      "learning_rate": 9.259611442095518e-06,
      "loss": 0.0024,
      "step": 80420
    },
    {
      "epoch": 0.5158530367215786,
      "grad_norm": 0.18084761500358582,
      "learning_rate": 9.259318316532962e-06,
      "loss": 0.0049,
      "step": 80430
    },
    {
      "epoch": 0.5159171736153646,
      "grad_norm": 0.3981788456439972,
      "learning_rate": 9.2590251375981e-06,
      "loss": 0.0039,
      "step": 80440
    },
    {
      "epoch": 0.5159813105091507,
      "grad_norm": 0.3020228147506714,
      "learning_rate": 9.258731905294614e-06,
      "loss": 0.0049,
      "step": 80450
    },
    {
      "epoch": 0.5160454474029368,
      "grad_norm": 0.23093923926353455,
      "learning_rate": 9.258438619626171e-06,
      "loss": 0.0022,
      "step": 80460
    },
    {
      "epoch": 0.5161095842967229,
      "grad_norm": 0.25309401750564575,
      "learning_rate": 9.258145280596449e-06,
      "loss": 0.0044,
      "step": 80470
    },
    {
      "epoch": 0.516173721190509,
      "grad_norm": 0.13883396983146667,
      "learning_rate": 9.257851888209122e-06,
      "loss": 0.0046,
      "step": 80480
    },
    {
      "epoch": 0.5162378580842951,
      "grad_norm": 0.15709243714809418,
      "learning_rate": 9.25755844246787e-06,
      "loss": 0.0041,
      "step": 80490
    },
    {
      "epoch": 0.5163019949780813,
      "grad_norm": 0.2982932925224304,
      "learning_rate": 9.257264943376367e-06,
      "loss": 0.003,
      "step": 80500
    },
    {
      "epoch": 0.5163661318718673,
      "grad_norm": 0.06413408368825912,
      "learning_rate": 9.256971390938293e-06,
      "loss": 0.0036,
      "step": 80510
    },
    {
      "epoch": 0.5164302687656535,
      "grad_norm": 0.17831914126873016,
      "learning_rate": 9.256677785157322e-06,
      "loss": 0.0034,
      "step": 80520
    },
    {
      "epoch": 0.5164944056594395,
      "grad_norm": 0.08256985247135162,
      "learning_rate": 9.256384126037138e-06,
      "loss": 0.0028,
      "step": 80530
    },
    {
      "epoch": 0.5165585425532256,
      "grad_norm": 0.2097683995962143,
      "learning_rate": 9.256090413581418e-06,
      "loss": 0.0024,
      "step": 80540
    },
    {
      "epoch": 0.5166226794470117,
      "grad_norm": 0.1839880794286728,
      "learning_rate": 9.255796647793845e-06,
      "loss": 0.0041,
      "step": 80550
    },
    {
      "epoch": 0.5166868163407978,
      "grad_norm": 0.16421499848365784,
      "learning_rate": 9.255502828678098e-06,
      "loss": 0.0024,
      "step": 80560
    },
    {
      "epoch": 0.5167509532345839,
      "grad_norm": 0.06348922848701477,
      "learning_rate": 9.255208956237859e-06,
      "loss": 0.0027,
      "step": 80570
    },
    {
      "epoch": 0.51681509012837,
      "grad_norm": 0.04008301720023155,
      "learning_rate": 9.254915030476809e-06,
      "loss": 0.0021,
      "step": 80580
    },
    {
      "epoch": 0.516879227022156,
      "grad_norm": 0.163941890001297,
      "learning_rate": 9.254621051398634e-06,
      "loss": 0.0035,
      "step": 80590
    },
    {
      "epoch": 0.5169433639159422,
      "grad_norm": 0.21079358458518982,
      "learning_rate": 9.254327019007017e-06,
      "loss": 0.003,
      "step": 80600
    },
    {
      "epoch": 0.5170075008097282,
      "grad_norm": 0.11273708194494247,
      "learning_rate": 9.25403293330564e-06,
      "loss": 0.0032,
      "step": 80610
    },
    {
      "epoch": 0.5170716377035144,
      "grad_norm": 0.23483584821224213,
      "learning_rate": 9.253738794298192e-06,
      "loss": 0.0048,
      "step": 80620
    },
    {
      "epoch": 0.5171357745973004,
      "grad_norm": 0.16384592652320862,
      "learning_rate": 9.253444601988358e-06,
      "loss": 0.0053,
      "step": 80630
    },
    {
      "epoch": 0.5171999114910866,
      "grad_norm": 0.14626704156398773,
      "learning_rate": 9.25315035637982e-06,
      "loss": 0.0042,
      "step": 80640
    },
    {
      "epoch": 0.5172640483848727,
      "grad_norm": 0.2711910009384155,
      "learning_rate": 9.25285605747627e-06,
      "loss": 0.0033,
      "step": 80650
    },
    {
      "epoch": 0.5173281852786588,
      "grad_norm": 0.23008853197097778,
      "learning_rate": 9.252561705281392e-06,
      "loss": 0.0032,
      "step": 80660
    },
    {
      "epoch": 0.5173923221724449,
      "grad_norm": 0.020175419747829437,
      "learning_rate": 9.252267299798881e-06,
      "loss": 0.0015,
      "step": 80670
    },
    {
      "epoch": 0.517456459066231,
      "grad_norm": 0.2377467155456543,
      "learning_rate": 9.251972841032419e-06,
      "loss": 0.004,
      "step": 80680
    },
    {
      "epoch": 0.5175205959600171,
      "grad_norm": 0.16429409384727478,
      "learning_rate": 9.251678328985697e-06,
      "loss": 0.0036,
      "step": 80690
    },
    {
      "epoch": 0.5175847328538031,
      "grad_norm": 0.08902248740196228,
      "learning_rate": 9.25138376366241e-06,
      "loss": 0.0029,
      "step": 80700
    },
    {
      "epoch": 0.5176488697475893,
      "grad_norm": 0.597663938999176,
      "learning_rate": 9.251089145066246e-06,
      "loss": 0.0032,
      "step": 80710
    },
    {
      "epoch": 0.5177130066413753,
      "grad_norm": 0.5348776578903198,
      "learning_rate": 9.250794473200895e-06,
      "loss": 0.0035,
      "step": 80720
    },
    {
      "epoch": 0.5177771435351615,
      "grad_norm": 0.09424172341823578,
      "learning_rate": 9.25049974807005e-06,
      "loss": 0.0033,
      "step": 80730
    },
    {
      "epoch": 0.5178412804289475,
      "grad_norm": 0.19990532100200653,
      "learning_rate": 9.250204969677408e-06,
      "loss": 0.0031,
      "step": 80740
    },
    {
      "epoch": 0.5179054173227337,
      "grad_norm": 0.06266714632511139,
      "learning_rate": 9.249910138026658e-06,
      "loss": 0.0029,
      "step": 80750
    },
    {
      "epoch": 0.5179695542165197,
      "grad_norm": 0.42398256063461304,
      "learning_rate": 9.249615253121498e-06,
      "loss": 0.004,
      "step": 80760
    },
    {
      "epoch": 0.5180336911103058,
      "grad_norm": 0.12699414789676666,
      "learning_rate": 9.249320314965621e-06,
      "loss": 0.0044,
      "step": 80770
    },
    {
      "epoch": 0.518097828004092,
      "grad_norm": 0.04501233622431755,
      "learning_rate": 9.249025323562726e-06,
      "loss": 0.0048,
      "step": 80780
    },
    {
      "epoch": 0.518161964897878,
      "grad_norm": 0.434689998626709,
      "learning_rate": 9.248730278916505e-06,
      "loss": 0.0036,
      "step": 80790
    },
    {
      "epoch": 0.5182261017916642,
      "grad_norm": 0.14772498607635498,
      "learning_rate": 9.248435181030657e-06,
      "loss": 0.0029,
      "step": 80800
    },
    {
      "epoch": 0.5182902386854502,
      "grad_norm": 0.05148780718445778,
      "learning_rate": 9.248140029908879e-06,
      "loss": 0.0047,
      "step": 80810
    },
    {
      "epoch": 0.5183543755792364,
      "grad_norm": 0.42253732681274414,
      "learning_rate": 9.247844825554872e-06,
      "loss": 0.0039,
      "step": 80820
    },
    {
      "epoch": 0.5184185124730224,
      "grad_norm": 0.2593359351158142,
      "learning_rate": 9.247549567972332e-06,
      "loss": 0.0035,
      "step": 80830
    },
    {
      "epoch": 0.5184826493668085,
      "grad_norm": 0.3482550382614136,
      "learning_rate": 9.247254257164962e-06,
      "loss": 0.0038,
      "step": 80840
    },
    {
      "epoch": 0.5185467862605946,
      "grad_norm": 0.29686230421066284,
      "learning_rate": 9.246958893136459e-06,
      "loss": 0.0056,
      "step": 80850
    },
    {
      "epoch": 0.5186109231543807,
      "grad_norm": 0.04358689486980438,
      "learning_rate": 9.246663475890528e-06,
      "loss": 0.0043,
      "step": 80860
    },
    {
      "epoch": 0.5186750600481668,
      "grad_norm": 0.18454205989837646,
      "learning_rate": 9.246368005430866e-06,
      "loss": 0.0026,
      "step": 80870
    },
    {
      "epoch": 0.5187391969419529,
      "grad_norm": 0.17590831220149994,
      "learning_rate": 9.24607248176118e-06,
      "loss": 0.0031,
      "step": 80880
    },
    {
      "epoch": 0.518803333835739,
      "grad_norm": 0.0882411077618599,
      "learning_rate": 9.24577690488517e-06,
      "loss": 0.0044,
      "step": 80890
    },
    {
      "epoch": 0.5188674707295251,
      "grad_norm": 0.15208327770233154,
      "learning_rate": 9.24548127480654e-06,
      "loss": 0.0038,
      "step": 80900
    },
    {
      "epoch": 0.5189316076233111,
      "grad_norm": 0.1888751983642578,
      "learning_rate": 9.245185591528997e-06,
      "loss": 0.0039,
      "step": 80910
    },
    {
      "epoch": 0.5189957445170973,
      "grad_norm": 0.05959264189004898,
      "learning_rate": 9.244889855056245e-06,
      "loss": 0.0024,
      "step": 80920
    },
    {
      "epoch": 0.5190598814108834,
      "grad_norm": 0.18062947690486908,
      "learning_rate": 9.244594065391989e-06,
      "loss": 0.0025,
      "step": 80930
    },
    {
      "epoch": 0.5191240183046695,
      "grad_norm": 0.07806473970413208,
      "learning_rate": 9.244298222539936e-06,
      "loss": 0.0024,
      "step": 80940
    },
    {
      "epoch": 0.5191881551984556,
      "grad_norm": 0.1433127522468567,
      "learning_rate": 9.244002326503792e-06,
      "loss": 0.0046,
      "step": 80950
    },
    {
      "epoch": 0.5192522920922417,
      "grad_norm": 0.6837491989135742,
      "learning_rate": 9.243706377287264e-06,
      "loss": 0.0031,
      "step": 80960
    },
    {
      "epoch": 0.5193164289860278,
      "grad_norm": 0.2809620201587677,
      "learning_rate": 9.243410374894066e-06,
      "loss": 0.0029,
      "step": 80970
    },
    {
      "epoch": 0.5193805658798138,
      "grad_norm": 0.15400190651416779,
      "learning_rate": 9.243114319327902e-06,
      "loss": 0.0054,
      "step": 80980
    },
    {
      "epoch": 0.5194447027736,
      "grad_norm": 0.12674783170223236,
      "learning_rate": 9.24281821059248e-06,
      "loss": 0.0035,
      "step": 80990
    },
    {
      "epoch": 0.519508839667386,
      "grad_norm": 0.3740582764148712,
      "learning_rate": 9.242522048691518e-06,
      "loss": 0.0042,
      "step": 81000
    },
    {
      "epoch": 0.5195729765611722,
      "grad_norm": 0.1282520294189453,
      "learning_rate": 9.242225833628721e-06,
      "loss": 0.0031,
      "step": 81010
    },
    {
      "epoch": 0.5196371134549582,
      "grad_norm": 0.48991167545318604,
      "learning_rate": 9.2419295654078e-06,
      "loss": 0.0022,
      "step": 81020
    },
    {
      "epoch": 0.5197012503487444,
      "grad_norm": 0.15189383924007416,
      "learning_rate": 9.241633244032474e-06,
      "loss": 0.0032,
      "step": 81030
    },
    {
      "epoch": 0.5197653872425304,
      "grad_norm": 0.17220990359783173,
      "learning_rate": 9.24133686950645e-06,
      "loss": 0.0084,
      "step": 81040
    },
    {
      "epoch": 0.5198295241363166,
      "grad_norm": 0.2771851122379303,
      "learning_rate": 9.241040441833444e-06,
      "loss": 0.0044,
      "step": 81050
    },
    {
      "epoch": 0.5198936610301027,
      "grad_norm": 0.11452841013669968,
      "learning_rate": 9.24074396101717e-06,
      "loss": 0.0038,
      "step": 81060
    },
    {
      "epoch": 0.5199577979238887,
      "grad_norm": 0.12710310518741608,
      "learning_rate": 9.240447427061343e-06,
      "loss": 0.003,
      "step": 81070
    },
    {
      "epoch": 0.5200219348176749,
      "grad_norm": 0.08416347950696945,
      "learning_rate": 9.24015083996968e-06,
      "loss": 0.0025,
      "step": 81080
    },
    {
      "epoch": 0.5200860717114609,
      "grad_norm": 0.2778503894805908,
      "learning_rate": 9.239854199745897e-06,
      "loss": 0.0024,
      "step": 81090
    },
    {
      "epoch": 0.5201502086052471,
      "grad_norm": 0.25798657536506653,
      "learning_rate": 9.239557506393709e-06,
      "loss": 0.0035,
      "step": 81100
    },
    {
      "epoch": 0.5202143454990331,
      "grad_norm": 0.11615985631942749,
      "learning_rate": 9.239260759916836e-06,
      "loss": 0.0039,
      "step": 81110
    },
    {
      "epoch": 0.5202784823928193,
      "grad_norm": 0.11622320860624313,
      "learning_rate": 9.238963960318996e-06,
      "loss": 0.0042,
      "step": 81120
    },
    {
      "epoch": 0.5203426192866053,
      "grad_norm": 0.43766507506370544,
      "learning_rate": 9.23866710760391e-06,
      "loss": 0.0038,
      "step": 81130
    },
    {
      "epoch": 0.5204067561803914,
      "grad_norm": 0.21479366719722748,
      "learning_rate": 9.238370201775294e-06,
      "loss": 0.0037,
      "step": 81140
    },
    {
      "epoch": 0.5204708930741775,
      "grad_norm": 0.37137898802757263,
      "learning_rate": 9.238073242836868e-06,
      "loss": 0.0022,
      "step": 81150
    },
    {
      "epoch": 0.5205350299679636,
      "grad_norm": 0.20445744693279266,
      "learning_rate": 9.237776230792359e-06,
      "loss": 0.0034,
      "step": 81160
    },
    {
      "epoch": 0.5205991668617497,
      "grad_norm": 0.23320919275283813,
      "learning_rate": 9.237479165645484e-06,
      "loss": 0.0031,
      "step": 81170
    },
    {
      "epoch": 0.5206633037555358,
      "grad_norm": 0.13351120054721832,
      "learning_rate": 9.237182047399966e-06,
      "loss": 0.0035,
      "step": 81180
    },
    {
      "epoch": 0.5207274406493219,
      "grad_norm": 0.1444133073091507,
      "learning_rate": 9.236884876059529e-06,
      "loss": 0.0037,
      "step": 81190
    },
    {
      "epoch": 0.520791577543108,
      "grad_norm": 0.47807008028030396,
      "learning_rate": 9.236587651627898e-06,
      "loss": 0.0037,
      "step": 81200
    },
    {
      "epoch": 0.5208557144368942,
      "grad_norm": 0.17233484983444214,
      "learning_rate": 9.236290374108794e-06,
      "loss": 0.004,
      "step": 81210
    },
    {
      "epoch": 0.5209198513306802,
      "grad_norm": 0.23898974061012268,
      "learning_rate": 9.235993043505943e-06,
      "loss": 0.0037,
      "step": 81220
    },
    {
      "epoch": 0.5209839882244663,
      "grad_norm": 0.063786581158638,
      "learning_rate": 9.235695659823074e-06,
      "loss": 0.0026,
      "step": 81230
    },
    {
      "epoch": 0.5210481251182524,
      "grad_norm": 0.17210139334201813,
      "learning_rate": 9.23539822306391e-06,
      "loss": 0.0025,
      "step": 81240
    },
    {
      "epoch": 0.5211122620120385,
      "grad_norm": 0.27867937088012695,
      "learning_rate": 9.235100733232181e-06,
      "loss": 0.0049,
      "step": 81250
    },
    {
      "epoch": 0.5211763989058246,
      "grad_norm": 0.0784267783164978,
      "learning_rate": 9.23480319033161e-06,
      "loss": 0.004,
      "step": 81260
    },
    {
      "epoch": 0.5212405357996107,
      "grad_norm": 0.10288307815790176,
      "learning_rate": 9.23450559436593e-06,
      "loss": 0.0034,
      "step": 81270
    },
    {
      "epoch": 0.5213046726933968,
      "grad_norm": 0.03400523215532303,
      "learning_rate": 9.234207945338869e-06,
      "loss": 0.0027,
      "step": 81280
    },
    {
      "epoch": 0.5213688095871829,
      "grad_norm": 0.2448701560497284,
      "learning_rate": 9.233910243254156e-06,
      "loss": 0.0032,
      "step": 81290
    },
    {
      "epoch": 0.5214329464809689,
      "grad_norm": 0.33407291769981384,
      "learning_rate": 9.233612488115521e-06,
      "loss": 0.0064,
      "step": 81300
    },
    {
      "epoch": 0.5214970833747551,
      "grad_norm": 0.2003421038389206,
      "learning_rate": 9.233314679926698e-06,
      "loss": 0.0032,
      "step": 81310
    },
    {
      "epoch": 0.5215612202685411,
      "grad_norm": 0.12202927470207214,
      "learning_rate": 9.233016818691415e-06,
      "loss": 0.0038,
      "step": 81320
    },
    {
      "epoch": 0.5216253571623273,
      "grad_norm": 0.2097141444683075,
      "learning_rate": 9.232718904413405e-06,
      "loss": 0.0021,
      "step": 81330
    },
    {
      "epoch": 0.5216894940561134,
      "grad_norm": 0.17498457431793213,
      "learning_rate": 9.232420937096403e-06,
      "loss": 0.0052,
      "step": 81340
    },
    {
      "epoch": 0.5217536309498995,
      "grad_norm": 0.21181584894657135,
      "learning_rate": 9.232122916744141e-06,
      "loss": 0.002,
      "step": 81350
    },
    {
      "epoch": 0.5218177678436856,
      "grad_norm": 0.22371451556682587,
      "learning_rate": 9.231824843360356e-06,
      "loss": 0.0022,
      "step": 81360
    },
    {
      "epoch": 0.5218819047374716,
      "grad_norm": 0.11706782877445221,
      "learning_rate": 9.23152671694878e-06,
      "loss": 0.0031,
      "step": 81370
    },
    {
      "epoch": 0.5219460416312578,
      "grad_norm": 0.19505396485328674,
      "learning_rate": 9.23122853751315e-06,
      "loss": 0.0041,
      "step": 81380
    },
    {
      "epoch": 0.5220101785250438,
      "grad_norm": 0.2122873216867447,
      "learning_rate": 9.230930305057204e-06,
      "loss": 0.0028,
      "step": 81390
    },
    {
      "epoch": 0.52207431541883,
      "grad_norm": 0.17082518339157104,
      "learning_rate": 9.230632019584676e-06,
      "loss": 0.0023,
      "step": 81400
    },
    {
      "epoch": 0.522138452312616,
      "grad_norm": 0.1508171409368515,
      "learning_rate": 9.230333681099305e-06,
      "loss": 0.0028,
      "step": 81410
    },
    {
      "epoch": 0.5222025892064022,
      "grad_norm": 0.07956456393003464,
      "learning_rate": 9.23003528960483e-06,
      "loss": 0.0023,
      "step": 81420
    },
    {
      "epoch": 0.5222667261001882,
      "grad_norm": 0.05028301477432251,
      "learning_rate": 9.229736845104991e-06,
      "loss": 0.0037,
      "step": 81430
    },
    {
      "epoch": 0.5223308629939744,
      "grad_norm": 0.09844990819692612,
      "learning_rate": 9.229438347603525e-06,
      "loss": 0.0043,
      "step": 81440
    },
    {
      "epoch": 0.5223949998877604,
      "grad_norm": 0.11157780140638351,
      "learning_rate": 9.229139797104173e-06,
      "loss": 0.002,
      "step": 81450
    },
    {
      "epoch": 0.5224591367815465,
      "grad_norm": 0.22767135500907898,
      "learning_rate": 9.228841193610679e-06,
      "loss": 0.0035,
      "step": 81460
    },
    {
      "epoch": 0.5225232736753326,
      "grad_norm": 0.09955685585737228,
      "learning_rate": 9.22854253712678e-06,
      "loss": 0.0034,
      "step": 81470
    },
    {
      "epoch": 0.5225874105691187,
      "grad_norm": 0.19802632927894592,
      "learning_rate": 9.228243827656222e-06,
      "loss": 0.0036,
      "step": 81480
    },
    {
      "epoch": 0.5226515474629049,
      "grad_norm": 0.0587104968726635,
      "learning_rate": 9.227945065202746e-06,
      "loss": 0.0032,
      "step": 81490
    },
    {
      "epoch": 0.5227156843566909,
      "grad_norm": 0.29555052518844604,
      "learning_rate": 9.2276462497701e-06,
      "loss": 0.0059,
      "step": 81500
    },
    {
      "epoch": 0.5227798212504771,
      "grad_norm": 0.14725208282470703,
      "learning_rate": 9.227347381362021e-06,
      "loss": 0.0035,
      "step": 81510
    },
    {
      "epoch": 0.5228439581442631,
      "grad_norm": 0.15954144299030304,
      "learning_rate": 9.227048459982261e-06,
      "loss": 0.0024,
      "step": 81520
    },
    {
      "epoch": 0.5229080950380492,
      "grad_norm": 0.1816718876361847,
      "learning_rate": 9.226749485634561e-06,
      "loss": 0.0044,
      "step": 81530
    },
    {
      "epoch": 0.5229722319318353,
      "grad_norm": 0.017237460240721703,
      "learning_rate": 9.22645045832267e-06,
      "loss": 0.0036,
      "step": 81540
    },
    {
      "epoch": 0.5230363688256214,
      "grad_norm": 0.1542568802833557,
      "learning_rate": 9.226151378050334e-06,
      "loss": 0.0015,
      "step": 81550
    },
    {
      "epoch": 0.5231005057194075,
      "grad_norm": 0.25452756881713867,
      "learning_rate": 9.2258522448213e-06,
      "loss": 0.0039,
      "step": 81560
    },
    {
      "epoch": 0.5231646426131936,
      "grad_norm": 0.20770305395126343,
      "learning_rate": 9.225553058639318e-06,
      "loss": 0.002,
      "step": 81570
    },
    {
      "epoch": 0.5232287795069797,
      "grad_norm": 0.25020110607147217,
      "learning_rate": 9.225253819508137e-06,
      "loss": 0.0045,
      "step": 81580
    },
    {
      "epoch": 0.5232929164007658,
      "grad_norm": 0.06710415333509445,
      "learning_rate": 9.224954527431504e-06,
      "loss": 0.0023,
      "step": 81590
    },
    {
      "epoch": 0.5233570532945518,
      "grad_norm": 0.1380765736103058,
      "learning_rate": 9.224655182413174e-06,
      "loss": 0.0041,
      "step": 81600
    },
    {
      "epoch": 0.523421190188338,
      "grad_norm": 0.06237601116299629,
      "learning_rate": 9.224355784456894e-06,
      "loss": 0.0022,
      "step": 81610
    },
    {
      "epoch": 0.5234853270821241,
      "grad_norm": 0.15989889204502106,
      "learning_rate": 9.224056333566416e-06,
      "loss": 0.0035,
      "step": 81620
    },
    {
      "epoch": 0.5235494639759102,
      "grad_norm": 0.06872167438268661,
      "learning_rate": 9.223756829745494e-06,
      "loss": 0.0028,
      "step": 81630
    },
    {
      "epoch": 0.5236136008696963,
      "grad_norm": 0.06739316880702972,
      "learning_rate": 9.223457272997878e-06,
      "loss": 0.006,
      "step": 81640
    },
    {
      "epoch": 0.5236777377634824,
      "grad_norm": 0.10684234648942947,
      "learning_rate": 9.223157663327326e-06,
      "loss": 0.0065,
      "step": 81650
    },
    {
      "epoch": 0.5237418746572685,
      "grad_norm": 0.2152058184146881,
      "learning_rate": 9.22285800073759e-06,
      "loss": 0.0036,
      "step": 81660
    },
    {
      "epoch": 0.5238060115510546,
      "grad_norm": 0.100911445915699,
      "learning_rate": 9.222558285232426e-06,
      "loss": 0.0027,
      "step": 81670
    },
    {
      "epoch": 0.5238701484448407,
      "grad_norm": 0.1213904395699501,
      "learning_rate": 9.222258516815589e-06,
      "loss": 0.0025,
      "step": 81680
    },
    {
      "epoch": 0.5239342853386267,
      "grad_norm": 0.25971806049346924,
      "learning_rate": 9.221958695490834e-06,
      "loss": 0.004,
      "step": 81690
    },
    {
      "epoch": 0.5239984222324129,
      "grad_norm": 0.17598684132099152,
      "learning_rate": 9.22165882126192e-06,
      "loss": 0.0034,
      "step": 81700
    },
    {
      "epoch": 0.5240625591261989,
      "grad_norm": 0.2927113473415375,
      "learning_rate": 9.221358894132604e-06,
      "loss": 0.0023,
      "step": 81710
    },
    {
      "epoch": 0.5241266960199851,
      "grad_norm": 0.1922122985124588,
      "learning_rate": 9.221058914106643e-06,
      "loss": 0.0039,
      "step": 81720
    },
    {
      "epoch": 0.5241908329137711,
      "grad_norm": 0.060337141156196594,
      "learning_rate": 9.220758881187797e-06,
      "loss": 0.0058,
      "step": 81730
    },
    {
      "epoch": 0.5242549698075573,
      "grad_norm": 0.4425851106643677,
      "learning_rate": 9.220458795379825e-06,
      "loss": 0.0034,
      "step": 81740
    },
    {
      "epoch": 0.5243191067013433,
      "grad_norm": 0.15111497044563293,
      "learning_rate": 9.22015865668649e-06,
      "loss": 0.0018,
      "step": 81750
    },
    {
      "epoch": 0.5243832435951294,
      "grad_norm": 0.30985045433044434,
      "learning_rate": 9.219858465111551e-06,
      "loss": 0.0031,
      "step": 81760
    },
    {
      "epoch": 0.5244473804889156,
      "grad_norm": 0.17037814855575562,
      "learning_rate": 9.219558220658768e-06,
      "loss": 0.0054,
      "step": 81770
    },
    {
      "epoch": 0.5245115173827016,
      "grad_norm": 0.19051095843315125,
      "learning_rate": 9.219257923331906e-06,
      "loss": 0.0033,
      "step": 81780
    },
    {
      "epoch": 0.5245756542764878,
      "grad_norm": 0.13113057613372803,
      "learning_rate": 9.218957573134725e-06,
      "loss": 0.004,
      "step": 81790
    },
    {
      "epoch": 0.5246397911702738,
      "grad_norm": 0.14635784924030304,
      "learning_rate": 9.218657170070993e-06,
      "loss": 0.0031,
      "step": 81800
    },
    {
      "epoch": 0.52470392806406,
      "grad_norm": 0.07720329612493515,
      "learning_rate": 9.21835671414447e-06,
      "loss": 0.0021,
      "step": 81810
    },
    {
      "epoch": 0.524768064957846,
      "grad_norm": 0.046794239431619644,
      "learning_rate": 9.218056205358924e-06,
      "loss": 0.0039,
      "step": 81820
    },
    {
      "epoch": 0.5248322018516322,
      "grad_norm": 0.0852995216846466,
      "learning_rate": 9.217755643718116e-06,
      "loss": 0.0045,
      "step": 81830
    },
    {
      "epoch": 0.5248963387454182,
      "grad_norm": 0.16625656187534332,
      "learning_rate": 9.217455029225818e-06,
      "loss": 0.0027,
      "step": 81840
    },
    {
      "epoch": 0.5249604756392043,
      "grad_norm": 0.01320594735443592,
      "learning_rate": 9.217154361885794e-06,
      "loss": 0.0031,
      "step": 81850
    },
    {
      "epoch": 0.5250246125329904,
      "grad_norm": 0.4065563976764679,
      "learning_rate": 9.216853641701811e-06,
      "loss": 0.0026,
      "step": 81860
    },
    {
      "epoch": 0.5250887494267765,
      "grad_norm": 0.07078109681606293,
      "learning_rate": 9.216552868677639e-06,
      "loss": 0.003,
      "step": 81870
    },
    {
      "epoch": 0.5251528863205626,
      "grad_norm": 0.2689833641052246,
      "learning_rate": 9.216252042817045e-06,
      "loss": 0.0032,
      "step": 81880
    },
    {
      "epoch": 0.5252170232143487,
      "grad_norm": 0.4440864026546478,
      "learning_rate": 9.2159511641238e-06,
      "loss": 0.0038,
      "step": 81890
    },
    {
      "epoch": 0.5252811601081347,
      "grad_norm": 0.09967883676290512,
      "learning_rate": 9.215650232601673e-06,
      "loss": 0.0043,
      "step": 81900
    },
    {
      "epoch": 0.5253452970019209,
      "grad_norm": 0.35125958919525146,
      "learning_rate": 9.215349248254439e-06,
      "loss": 0.0046,
      "step": 81910
    },
    {
      "epoch": 0.525409433895707,
      "grad_norm": 0.034344110637903214,
      "learning_rate": 9.215048211085864e-06,
      "loss": 0.0017,
      "step": 81920
    },
    {
      "epoch": 0.5254735707894931,
      "grad_norm": 0.20028561353683472,
      "learning_rate": 9.214747121099721e-06,
      "loss": 0.0032,
      "step": 81930
    },
    {
      "epoch": 0.5255377076832792,
      "grad_norm": 0.16437016427516937,
      "learning_rate": 9.214445978299787e-06,
      "loss": 0.004,
      "step": 81940
    },
    {
      "epoch": 0.5256018445770653,
      "grad_norm": 0.20637601613998413,
      "learning_rate": 9.214144782689832e-06,
      "loss": 0.0025,
      "step": 81950
    },
    {
      "epoch": 0.5256659814708514,
      "grad_norm": 0.09163656085729599,
      "learning_rate": 9.21384353427363e-06,
      "loss": 0.0023,
      "step": 81960
    },
    {
      "epoch": 0.5257301183646375,
      "grad_norm": 0.22363221645355225,
      "learning_rate": 9.213542233054958e-06,
      "loss": 0.0039,
      "step": 81970
    },
    {
      "epoch": 0.5257942552584236,
      "grad_norm": 0.174314945936203,
      "learning_rate": 9.21324087903759e-06,
      "loss": 0.0036,
      "step": 81980
    },
    {
      "epoch": 0.5258583921522096,
      "grad_norm": 0.30793848633766174,
      "learning_rate": 9.212939472225304e-06,
      "loss": 0.0028,
      "step": 81990
    },
    {
      "epoch": 0.5259225290459958,
      "grad_norm": 0.38619428873062134,
      "learning_rate": 9.212638012621875e-06,
      "loss": 0.003,
      "step": 82000
    },
    {
      "epoch": 0.5259866659397818,
      "grad_norm": 0.43908780813217163,
      "learning_rate": 9.212336500231079e-06,
      "loss": 0.0065,
      "step": 82010
    },
    {
      "epoch": 0.526050802833568,
      "grad_norm": 0.08649495989084244,
      "learning_rate": 9.212034935056698e-06,
      "loss": 0.0017,
      "step": 82020
    },
    {
      "epoch": 0.526114939727354,
      "grad_norm": 0.10458250343799591,
      "learning_rate": 9.211733317102509e-06,
      "loss": 0.0045,
      "step": 82030
    },
    {
      "epoch": 0.5261790766211402,
      "grad_norm": 0.14031673967838287,
      "learning_rate": 9.21143164637229e-06,
      "loss": 0.0029,
      "step": 82040
    },
    {
      "epoch": 0.5262432135149263,
      "grad_norm": 0.2695571780204773,
      "learning_rate": 9.211129922869823e-06,
      "loss": 0.0033,
      "step": 82050
    },
    {
      "epoch": 0.5263073504087123,
      "grad_norm": 0.05847835913300514,
      "learning_rate": 9.210828146598889e-06,
      "loss": 0.0028,
      "step": 82060
    },
    {
      "epoch": 0.5263714873024985,
      "grad_norm": 0.12969672679901123,
      "learning_rate": 9.210526317563269e-06,
      "loss": 0.0033,
      "step": 82070
    },
    {
      "epoch": 0.5264356241962845,
      "grad_norm": 0.24454118311405182,
      "learning_rate": 9.210224435766743e-06,
      "loss": 0.0045,
      "step": 82080
    },
    {
      "epoch": 0.5264997610900707,
      "grad_norm": 0.1197141483426094,
      "learning_rate": 9.209922501213098e-06,
      "loss": 0.0041,
      "step": 82090
    },
    {
      "epoch": 0.5265638979838567,
      "grad_norm": 0.3682247996330261,
      "learning_rate": 9.209620513906116e-06,
      "loss": 0.0052,
      "step": 82100
    },
    {
      "epoch": 0.5266280348776429,
      "grad_norm": 0.2559485137462616,
      "learning_rate": 9.20931847384958e-06,
      "loss": 0.0022,
      "step": 82110
    },
    {
      "epoch": 0.5266921717714289,
      "grad_norm": 0.07267298549413681,
      "learning_rate": 9.209016381047273e-06,
      "loss": 0.0033,
      "step": 82120
    },
    {
      "epoch": 0.5267563086652151,
      "grad_norm": 0.155486598610878,
      "learning_rate": 9.208714235502985e-06,
      "loss": 0.0052,
      "step": 82130
    },
    {
      "epoch": 0.5268204455590011,
      "grad_norm": 0.14211998879909515,
      "learning_rate": 9.208412037220497e-06,
      "loss": 0.0046,
      "step": 82140
    },
    {
      "epoch": 0.5268845824527872,
      "grad_norm": 0.16526386141777039,
      "learning_rate": 9.208109786203602e-06,
      "loss": 0.0023,
      "step": 82150
    },
    {
      "epoch": 0.5269487193465733,
      "grad_norm": 0.40999835729599,
      "learning_rate": 9.207807482456081e-06,
      "loss": 0.0061,
      "step": 82160
    },
    {
      "epoch": 0.5270128562403594,
      "grad_norm": 0.3757804036140442,
      "learning_rate": 9.207505125981726e-06,
      "loss": 0.0073,
      "step": 82170
    },
    {
      "epoch": 0.5270769931341455,
      "grad_norm": 0.1633116602897644,
      "learning_rate": 9.207202716784326e-06,
      "loss": 0.0051,
      "step": 82180
    },
    {
      "epoch": 0.5271411300279316,
      "grad_norm": 0.3592239320278168,
      "learning_rate": 9.206900254867669e-06,
      "loss": 0.007,
      "step": 82190
    },
    {
      "epoch": 0.5272052669217178,
      "grad_norm": 0.27414438128471375,
      "learning_rate": 9.206597740235543e-06,
      "loss": 0.0071,
      "step": 82200
    },
    {
      "epoch": 0.5272694038155038,
      "grad_norm": 0.07317782193422318,
      "learning_rate": 9.206295172891742e-06,
      "loss": 0.0036,
      "step": 82210
    },
    {
      "epoch": 0.52733354070929,
      "grad_norm": 0.11561580747365952,
      "learning_rate": 9.205992552840056e-06,
      "loss": 0.0043,
      "step": 82220
    },
    {
      "epoch": 0.527397677603076,
      "grad_norm": 0.14711986482143402,
      "learning_rate": 9.205689880084277e-06,
      "loss": 0.0034,
      "step": 82230
    },
    {
      "epoch": 0.5274618144968621,
      "grad_norm": 0.09318558871746063,
      "learning_rate": 9.205387154628198e-06,
      "loss": 0.0058,
      "step": 82240
    },
    {
      "epoch": 0.5275259513906482,
      "grad_norm": 0.14137667417526245,
      "learning_rate": 9.205084376475615e-06,
      "loss": 0.0031,
      "step": 82250
    },
    {
      "epoch": 0.5275900882844343,
      "grad_norm": 0.22283531725406647,
      "learning_rate": 9.204781545630317e-06,
      "loss": 0.0044,
      "step": 82260
    },
    {
      "epoch": 0.5276542251782204,
      "grad_norm": 0.15352214872837067,
      "learning_rate": 9.204478662096101e-06,
      "loss": 0.0036,
      "step": 82270
    },
    {
      "epoch": 0.5277183620720065,
      "grad_norm": 0.18237556517124176,
      "learning_rate": 9.204175725876762e-06,
      "loss": 0.0018,
      "step": 82280
    },
    {
      "epoch": 0.5277824989657925,
      "grad_norm": 0.07068685442209244,
      "learning_rate": 9.203872736976098e-06,
      "loss": 0.0046,
      "step": 82290
    },
    {
      "epoch": 0.5278466358595787,
      "grad_norm": 0.1488722711801529,
      "learning_rate": 9.203569695397905e-06,
      "loss": 0.0034,
      "step": 82300
    },
    {
      "epoch": 0.5279107727533647,
      "grad_norm": 0.10469914972782135,
      "learning_rate": 9.203266601145977e-06,
      "loss": 0.0037,
      "step": 82310
    },
    {
      "epoch": 0.5279749096471509,
      "grad_norm": 0.17077229917049408,
      "learning_rate": 9.202963454224117e-06,
      "loss": 0.0028,
      "step": 82320
    },
    {
      "epoch": 0.528039046540937,
      "grad_norm": 0.34882014989852905,
      "learning_rate": 9.202660254636118e-06,
      "loss": 0.0058,
      "step": 82330
    },
    {
      "epoch": 0.5281031834347231,
      "grad_norm": 0.2677260935306549,
      "learning_rate": 9.202357002385784e-06,
      "loss": 0.0047,
      "step": 82340
    },
    {
      "epoch": 0.5281673203285092,
      "grad_norm": 0.19920547306537628,
      "learning_rate": 9.202053697476915e-06,
      "loss": 0.0042,
      "step": 82350
    },
    {
      "epoch": 0.5282314572222953,
      "grad_norm": 0.03182956576347351,
      "learning_rate": 9.201750339913309e-06,
      "loss": 0.0025,
      "step": 82360
    },
    {
      "epoch": 0.5282955941160814,
      "grad_norm": 0.08129862695932388,
      "learning_rate": 9.201446929698767e-06,
      "loss": 0.0028,
      "step": 82370
    },
    {
      "epoch": 0.5283597310098674,
      "grad_norm": 0.20995286107063293,
      "learning_rate": 9.201143466837093e-06,
      "loss": 0.0017,
      "step": 82380
    },
    {
      "epoch": 0.5284238679036536,
      "grad_norm": 0.0474662110209465,
      "learning_rate": 9.200839951332088e-06,
      "loss": 0.0048,
      "step": 82390
    },
    {
      "epoch": 0.5284880047974396,
      "grad_norm": 0.14652827382087708,
      "learning_rate": 9.200536383187557e-06,
      "loss": 0.0024,
      "step": 82400
    },
    {
      "epoch": 0.5285521416912258,
      "grad_norm": 0.07190393656492233,
      "learning_rate": 9.200232762407305e-06,
      "loss": 0.0025,
      "step": 82410
    },
    {
      "epoch": 0.5286162785850118,
      "grad_norm": 0.043861404061317444,
      "learning_rate": 9.199929088995132e-06,
      "loss": 0.0041,
      "step": 82420
    },
    {
      "epoch": 0.528680415478798,
      "grad_norm": 0.08191066980361938,
      "learning_rate": 9.199625362954847e-06,
      "loss": 0.0034,
      "step": 82430
    },
    {
      "epoch": 0.528744552372584,
      "grad_norm": 0.14318427443504333,
      "learning_rate": 9.199321584290253e-06,
      "loss": 0.0044,
      "step": 82440
    },
    {
      "epoch": 0.5288086892663701,
      "grad_norm": 0.10137626528739929,
      "learning_rate": 9.199017753005159e-06,
      "loss": 0.0047,
      "step": 82450
    },
    {
      "epoch": 0.5288728261601562,
      "grad_norm": 0.29027968645095825,
      "learning_rate": 9.198713869103373e-06,
      "loss": 0.0034,
      "step": 82460
    },
    {
      "epoch": 0.5289369630539423,
      "grad_norm": 0.17395056784152985,
      "learning_rate": 9.1984099325887e-06,
      "loss": 0.0049,
      "step": 82470
    },
    {
      "epoch": 0.5290010999477285,
      "grad_norm": 0.1001816838979721,
      "learning_rate": 9.198105943464953e-06,
      "loss": 0.0054,
      "step": 82480
    },
    {
      "epoch": 0.5290652368415145,
      "grad_norm": 0.08453565835952759,
      "learning_rate": 9.197801901735934e-06,
      "loss": 0.003,
      "step": 82490
    },
    {
      "epoch": 0.5291293737353007,
      "grad_norm": 0.17507454752922058,
      "learning_rate": 9.19749780740546e-06,
      "loss": 0.0024,
      "step": 82500
    },
    {
      "epoch": 0.5291935106290867,
      "grad_norm": 0.3217713534832001,
      "learning_rate": 9.197193660477337e-06,
      "loss": 0.0023,
      "step": 82510
    },
    {
      "epoch": 0.5292576475228729,
      "grad_norm": 0.45715153217315674,
      "learning_rate": 9.196889460955379e-06,
      "loss": 0.0044,
      "step": 82520
    },
    {
      "epoch": 0.5293217844166589,
      "grad_norm": 0.07727304846048355,
      "learning_rate": 9.196585208843397e-06,
      "loss": 0.003,
      "step": 82530
    },
    {
      "epoch": 0.529385921310445,
      "grad_norm": 0.0397333949804306,
      "learning_rate": 9.1962809041452e-06,
      "loss": 0.0036,
      "step": 82540
    },
    {
      "epoch": 0.5294500582042311,
      "grad_norm": 0.03212263435125351,
      "learning_rate": 9.195976546864607e-06,
      "loss": 0.0041,
      "step": 82550
    },
    {
      "epoch": 0.5295141950980172,
      "grad_norm": 0.46381548047065735,
      "learning_rate": 9.195672137005427e-06,
      "loss": 0.0023,
      "step": 82560
    },
    {
      "epoch": 0.5295783319918033,
      "grad_norm": 0.31793341040611267,
      "learning_rate": 9.195367674571477e-06,
      "loss": 0.0031,
      "step": 82570
    },
    {
      "epoch": 0.5296424688855894,
      "grad_norm": 0.05849529430270195,
      "learning_rate": 9.195063159566574e-06,
      "loss": 0.0024,
      "step": 82580
    },
    {
      "epoch": 0.5297066057793755,
      "grad_norm": 0.2585103511810303,
      "learning_rate": 9.19475859199453e-06,
      "loss": 0.0024,
      "step": 82590
    },
    {
      "epoch": 0.5297707426731616,
      "grad_norm": 0.1477556675672531,
      "learning_rate": 9.194453971859162e-06,
      "loss": 0.0027,
      "step": 82600
    },
    {
      "epoch": 0.5298348795669477,
      "grad_norm": 0.12497889250516891,
      "learning_rate": 9.19414929916429e-06,
      "loss": 0.0038,
      "step": 82610
    },
    {
      "epoch": 0.5298990164607338,
      "grad_norm": 0.2745285630226135,
      "learning_rate": 9.193844573913726e-06,
      "loss": 0.0022,
      "step": 82620
    },
    {
      "epoch": 0.5299631533545199,
      "grad_norm": 0.029844246804714203,
      "learning_rate": 9.193539796111296e-06,
      "loss": 0.0039,
      "step": 82630
    },
    {
      "epoch": 0.530027290248306,
      "grad_norm": 0.24429070949554443,
      "learning_rate": 9.193234965760813e-06,
      "loss": 0.0046,
      "step": 82640
    },
    {
      "epoch": 0.5300914271420921,
      "grad_norm": 0.14826074242591858,
      "learning_rate": 9.1929300828661e-06,
      "loss": 0.0026,
      "step": 82650
    },
    {
      "epoch": 0.5301555640358782,
      "grad_norm": 0.15577656030654907,
      "learning_rate": 9.192625147430978e-06,
      "loss": 0.003,
      "step": 82660
    },
    {
      "epoch": 0.5302197009296643,
      "grad_norm": 0.0887686088681221,
      "learning_rate": 9.192320159459263e-06,
      "loss": 0.0035,
      "step": 82670
    },
    {
      "epoch": 0.5302838378234503,
      "grad_norm": 0.058395545929670334,
      "learning_rate": 9.192015118954782e-06,
      "loss": 0.0025,
      "step": 82680
    },
    {
      "epoch": 0.5303479747172365,
      "grad_norm": 0.023756619542837143,
      "learning_rate": 9.191710025921356e-06,
      "loss": 0.0037,
      "step": 82690
    },
    {
      "epoch": 0.5304121116110225,
      "grad_norm": 0.3501819670200348,
      "learning_rate": 9.191404880362807e-06,
      "loss": 0.0038,
      "step": 82700
    },
    {
      "epoch": 0.5304762485048087,
      "grad_norm": 0.2135632187128067,
      "learning_rate": 9.19109968228296e-06,
      "loss": 0.0019,
      "step": 82710
    },
    {
      "epoch": 0.5305403853985947,
      "grad_norm": 0.11029896140098572,
      "learning_rate": 9.19079443168564e-06,
      "loss": 0.0025,
      "step": 82720
    },
    {
      "epoch": 0.5306045222923809,
      "grad_norm": 0.23255330324172974,
      "learning_rate": 9.190489128574666e-06,
      "loss": 0.0047,
      "step": 82730
    },
    {
      "epoch": 0.5306686591861669,
      "grad_norm": 0.08605362474918365,
      "learning_rate": 9.190183772953872e-06,
      "loss": 0.0039,
      "step": 82740
    },
    {
      "epoch": 0.530732796079953,
      "grad_norm": 0.1807021051645279,
      "learning_rate": 9.18987836482708e-06,
      "loss": 0.003,
      "step": 82750
    },
    {
      "epoch": 0.5307969329737392,
      "grad_norm": 0.12389001250267029,
      "learning_rate": 9.189572904198118e-06,
      "loss": 0.0029,
      "step": 82760
    },
    {
      "epoch": 0.5308610698675252,
      "grad_norm": 0.045452624559402466,
      "learning_rate": 9.189267391070812e-06,
      "loss": 0.004,
      "step": 82770
    },
    {
      "epoch": 0.5309252067613114,
      "grad_norm": 0.05268412083387375,
      "learning_rate": 9.188961825448992e-06,
      "loss": 0.003,
      "step": 82780
    },
    {
      "epoch": 0.5309893436550974,
      "grad_norm": 0.2025286704301834,
      "learning_rate": 9.188656207336488e-06,
      "loss": 0.0031,
      "step": 82790
    },
    {
      "epoch": 0.5310534805488836,
      "grad_norm": 0.0918094664812088,
      "learning_rate": 9.188350536737126e-06,
      "loss": 0.005,
      "step": 82800
    },
    {
      "epoch": 0.5311176174426696,
      "grad_norm": 0.11574849486351013,
      "learning_rate": 9.18804481365474e-06,
      "loss": 0.0038,
      "step": 82810
    },
    {
      "epoch": 0.5311817543364558,
      "grad_norm": 0.2847580015659332,
      "learning_rate": 9.187739038093157e-06,
      "loss": 0.0061,
      "step": 82820
    },
    {
      "epoch": 0.5312458912302418,
      "grad_norm": 0.3507842719554901,
      "learning_rate": 9.187433210056214e-06,
      "loss": 0.0027,
      "step": 82830
    },
    {
      "epoch": 0.5313100281240279,
      "grad_norm": 0.196197047829628,
      "learning_rate": 9.187127329547739e-06,
      "loss": 0.0017,
      "step": 82840
    },
    {
      "epoch": 0.531374165017814,
      "grad_norm": 0.1247221902012825,
      "learning_rate": 9.186821396571565e-06,
      "loss": 0.0054,
      "step": 82850
    },
    {
      "epoch": 0.5314383019116001,
      "grad_norm": 0.16086354851722717,
      "learning_rate": 9.186515411131527e-06,
      "loss": 0.0019,
      "step": 82860
    },
    {
      "epoch": 0.5315024388053862,
      "grad_norm": 0.11792601644992828,
      "learning_rate": 9.186209373231461e-06,
      "loss": 0.0042,
      "step": 82870
    },
    {
      "epoch": 0.5315665756991723,
      "grad_norm": 0.23305004835128784,
      "learning_rate": 9.185903282875197e-06,
      "loss": 0.0029,
      "step": 82880
    },
    {
      "epoch": 0.5316307125929585,
      "grad_norm": 0.23246848583221436,
      "learning_rate": 9.185597140066573e-06,
      "loss": 0.0035,
      "step": 82890
    },
    {
      "epoch": 0.5316948494867445,
      "grad_norm": 0.2900582253932953,
      "learning_rate": 9.185290944809429e-06,
      "loss": 0.0037,
      "step": 82900
    },
    {
      "epoch": 0.5317589863805307,
      "grad_norm": 0.11810173094272614,
      "learning_rate": 9.184984697107594e-06,
      "loss": 0.0026,
      "step": 82910
    },
    {
      "epoch": 0.5318231232743167,
      "grad_norm": 0.18069452047348022,
      "learning_rate": 9.184678396964911e-06,
      "loss": 0.0032,
      "step": 82920
    },
    {
      "epoch": 0.5318872601681028,
      "grad_norm": 0.06355767697095871,
      "learning_rate": 9.184372044385219e-06,
      "loss": 0.0033,
      "step": 82930
    },
    {
      "epoch": 0.5319513970618889,
      "grad_norm": 0.018011651933193207,
      "learning_rate": 9.184065639372353e-06,
      "loss": 0.0027,
      "step": 82940
    },
    {
      "epoch": 0.532015533955675,
      "grad_norm": 0.11940117180347443,
      "learning_rate": 9.183759181930154e-06,
      "loss": 0.0019,
      "step": 82950
    },
    {
      "epoch": 0.5320796708494611,
      "grad_norm": 0.14173869788646698,
      "learning_rate": 9.183452672062464e-06,
      "loss": 0.0027,
      "step": 82960
    },
    {
      "epoch": 0.5321438077432472,
      "grad_norm": 0.014904815703630447,
      "learning_rate": 9.183146109773121e-06,
      "loss": 0.0031,
      "step": 82970
    },
    {
      "epoch": 0.5322079446370332,
      "grad_norm": 0.132496178150177,
      "learning_rate": 9.182839495065968e-06,
      "loss": 0.0025,
      "step": 82980
    },
    {
      "epoch": 0.5322720815308194,
      "grad_norm": 0.12323196232318878,
      "learning_rate": 9.182532827944846e-06,
      "loss": 0.004,
      "step": 82990
    },
    {
      "epoch": 0.5323362184246054,
      "grad_norm": 0.2556149661540985,
      "learning_rate": 9.1822261084136e-06,
      "loss": 0.008,
      "step": 83000
    },
    {
      "epoch": 0.5324003553183916,
      "grad_norm": 0.11383537948131561,
      "learning_rate": 9.18191933647607e-06,
      "loss": 0.0032,
      "step": 83010
    },
    {
      "epoch": 0.5324644922121776,
      "grad_norm": 0.24259330332279205,
      "learning_rate": 9.181612512136103e-06,
      "loss": 0.004,
      "step": 83020
    },
    {
      "epoch": 0.5325286291059638,
      "grad_norm": 0.296653151512146,
      "learning_rate": 9.181305635397542e-06,
      "loss": 0.0027,
      "step": 83030
    },
    {
      "epoch": 0.5325927659997499,
      "grad_norm": 0.07351360470056534,
      "learning_rate": 9.180998706264234e-06,
      "loss": 0.0024,
      "step": 83040
    },
    {
      "epoch": 0.532656902893536,
      "grad_norm": 0.11533034592866898,
      "learning_rate": 9.180691724740023e-06,
      "loss": 0.0031,
      "step": 83050
    },
    {
      "epoch": 0.5327210397873221,
      "grad_norm": 0.07853560894727707,
      "learning_rate": 9.180384690828758e-06,
      "loss": 0.0027,
      "step": 83060
    },
    {
      "epoch": 0.5327851766811081,
      "grad_norm": 0.08176155388355255,
      "learning_rate": 9.180077604534283e-06,
      "loss": 0.0027,
      "step": 83070
    },
    {
      "epoch": 0.5328493135748943,
      "grad_norm": 0.3176024854183197,
      "learning_rate": 9.17977046586045e-06,
      "loss": 0.0028,
      "step": 83080
    },
    {
      "epoch": 0.5329134504686803,
      "grad_norm": 0.28645059466362,
      "learning_rate": 9.179463274811106e-06,
      "loss": 0.0028,
      "step": 83090
    },
    {
      "epoch": 0.5329775873624665,
      "grad_norm": 0.16513247787952423,
      "learning_rate": 9.179156031390099e-06,
      "loss": 0.0051,
      "step": 83100
    },
    {
      "epoch": 0.5330417242562525,
      "grad_norm": 0.1360175758600235,
      "learning_rate": 9.178848735601281e-06,
      "loss": 0.0043,
      "step": 83110
    },
    {
      "epoch": 0.5331058611500387,
      "grad_norm": 0.07889141887426376,
      "learning_rate": 9.178541387448502e-06,
      "loss": 0.0043,
      "step": 83120
    },
    {
      "epoch": 0.5331699980438247,
      "grad_norm": 0.11900272965431213,
      "learning_rate": 9.178233986935612e-06,
      "loss": 0.0043,
      "step": 83130
    },
    {
      "epoch": 0.5332341349376108,
      "grad_norm": 0.09275635331869125,
      "learning_rate": 9.177926534066466e-06,
      "loss": 0.0023,
      "step": 83140
    },
    {
      "epoch": 0.5332982718313969,
      "grad_norm": 0.014314381405711174,
      "learning_rate": 9.177619028844911e-06,
      "loss": 0.0029,
      "step": 83150
    },
    {
      "epoch": 0.533362408725183,
      "grad_norm": 0.10282004624605179,
      "learning_rate": 9.177311471274808e-06,
      "loss": 0.0044,
      "step": 83160
    },
    {
      "epoch": 0.5334265456189691,
      "grad_norm": 0.1841009259223938,
      "learning_rate": 9.177003861360003e-06,
      "loss": 0.0063,
      "step": 83170
    },
    {
      "epoch": 0.5334906825127552,
      "grad_norm": 0.1191323846578598,
      "learning_rate": 9.176696199104358e-06,
      "loss": 0.0032,
      "step": 83180
    },
    {
      "epoch": 0.5335548194065414,
      "grad_norm": 0.047740787267684937,
      "learning_rate": 9.176388484511722e-06,
      "loss": 0.003,
      "step": 83190
    },
    {
      "epoch": 0.5336189563003274,
      "grad_norm": 0.27387723326683044,
      "learning_rate": 9.176080717585954e-06,
      "loss": 0.0043,
      "step": 83200
    },
    {
      "epoch": 0.5336830931941136,
      "grad_norm": 0.2824685871601105,
      "learning_rate": 9.17577289833091e-06,
      "loss": 0.0038,
      "step": 83210
    },
    {
      "epoch": 0.5337472300878996,
      "grad_norm": 0.16182850301265717,
      "learning_rate": 9.175465026750447e-06,
      "loss": 0.0049,
      "step": 83220
    },
    {
      "epoch": 0.5338113669816857,
      "grad_norm": 0.07123143970966339,
      "learning_rate": 9.175157102848425e-06,
      "loss": 0.0043,
      "step": 83230
    },
    {
      "epoch": 0.5338755038754718,
      "grad_norm": 0.22581571340560913,
      "learning_rate": 9.174849126628698e-06,
      "loss": 0.0025,
      "step": 83240
    },
    {
      "epoch": 0.5339396407692579,
      "grad_norm": 0.11887729167938232,
      "learning_rate": 9.174541098095128e-06,
      "loss": 0.0042,
      "step": 83250
    },
    {
      "epoch": 0.534003777663044,
      "grad_norm": 0.09506013989448547,
      "learning_rate": 9.174233017251577e-06,
      "loss": 0.0025,
      "step": 83260
    },
    {
      "epoch": 0.5340679145568301,
      "grad_norm": 0.3059485852718353,
      "learning_rate": 9.173924884101902e-06,
      "loss": 0.0035,
      "step": 83270
    },
    {
      "epoch": 0.5341320514506162,
      "grad_norm": 0.2697980999946594,
      "learning_rate": 9.173616698649963e-06,
      "loss": 0.0034,
      "step": 83280
    },
    {
      "epoch": 0.5341961883444023,
      "grad_norm": 0.06674520671367645,
      "learning_rate": 9.173308460899627e-06,
      "loss": 0.0032,
      "step": 83290
    },
    {
      "epoch": 0.5342603252381883,
      "grad_norm": 0.33980584144592285,
      "learning_rate": 9.173000170854752e-06,
      "loss": 0.003,
      "step": 83300
    },
    {
      "epoch": 0.5343244621319745,
      "grad_norm": 0.1164994016289711,
      "learning_rate": 9.1726918285192e-06,
      "loss": 0.0047,
      "step": 83310
    },
    {
      "epoch": 0.5343885990257606,
      "grad_norm": 0.21461822092533112,
      "learning_rate": 9.172383433896841e-06,
      "loss": 0.0042,
      "step": 83320
    },
    {
      "epoch": 0.5344527359195467,
      "grad_norm": 0.1043817549943924,
      "learning_rate": 9.172074986991532e-06,
      "loss": 0.0029,
      "step": 83330
    },
    {
      "epoch": 0.5345168728133328,
      "grad_norm": 0.22230912744998932,
      "learning_rate": 9.171766487807146e-06,
      "loss": 0.0032,
      "step": 83340
    },
    {
      "epoch": 0.5345810097071189,
      "grad_norm": 0.1703413873910904,
      "learning_rate": 9.171457936347541e-06,
      "loss": 0.0035,
      "step": 83350
    },
    {
      "epoch": 0.534645146600905,
      "grad_norm": 0.20179855823516846,
      "learning_rate": 9.171149332616589e-06,
      "loss": 0.0054,
      "step": 83360
    },
    {
      "epoch": 0.534709283494691,
      "grad_norm": 0.12678289413452148,
      "learning_rate": 9.170840676618153e-06,
      "loss": 0.0053,
      "step": 83370
    },
    {
      "epoch": 0.5347734203884772,
      "grad_norm": 0.1558840423822403,
      "learning_rate": 9.170531968356103e-06,
      "loss": 0.0033,
      "step": 83380
    },
    {
      "epoch": 0.5348375572822632,
      "grad_norm": 0.06316942721605301,
      "learning_rate": 9.170223207834308e-06,
      "loss": 0.0022,
      "step": 83390
    },
    {
      "epoch": 0.5349016941760494,
      "grad_norm": 0.16530416905879974,
      "learning_rate": 9.169914395056634e-06,
      "loss": 0.0029,
      "step": 83400
    },
    {
      "epoch": 0.5349658310698354,
      "grad_norm": 0.11882420629262924,
      "learning_rate": 9.169605530026953e-06,
      "loss": 0.0014,
      "step": 83410
    },
    {
      "epoch": 0.5350299679636216,
      "grad_norm": 0.1492297351360321,
      "learning_rate": 9.169296612749134e-06,
      "loss": 0.0031,
      "step": 83420
    },
    {
      "epoch": 0.5350941048574076,
      "grad_norm": 0.05903792381286621,
      "learning_rate": 9.16898764322705e-06,
      "loss": 0.0027,
      "step": 83430
    },
    {
      "epoch": 0.5351582417511938,
      "grad_norm": 0.14036016166210175,
      "learning_rate": 9.16867862146457e-06,
      "loss": 0.0029,
      "step": 83440
    },
    {
      "epoch": 0.5352223786449798,
      "grad_norm": 0.14284096658229828,
      "learning_rate": 9.16836954746557e-06,
      "loss": 0.0033,
      "step": 83450
    },
    {
      "epoch": 0.5352865155387659,
      "grad_norm": 0.17533113062381744,
      "learning_rate": 9.168060421233918e-06,
      "loss": 0.0017,
      "step": 83460
    },
    {
      "epoch": 0.5353506524325521,
      "grad_norm": 0.11281425505876541,
      "learning_rate": 9.167751242773492e-06,
      "loss": 0.0032,
      "step": 83470
    },
    {
      "epoch": 0.5354147893263381,
      "grad_norm": 0.09841414541006088,
      "learning_rate": 9.167442012088164e-06,
      "loss": 0.0023,
      "step": 83480
    },
    {
      "epoch": 0.5354789262201243,
      "grad_norm": 0.14227591454982758,
      "learning_rate": 9.167132729181807e-06,
      "loss": 0.0035,
      "step": 83490
    },
    {
      "epoch": 0.5355430631139103,
      "grad_norm": 0.1604137420654297,
      "learning_rate": 9.1668233940583e-06,
      "loss": 0.0047,
      "step": 83500
    },
    {
      "epoch": 0.5356072000076965,
      "grad_norm": 0.3364444375038147,
      "learning_rate": 9.16651400672152e-06,
      "loss": 0.0053,
      "step": 83510
    },
    {
      "epoch": 0.5356713369014825,
      "grad_norm": 0.22063446044921875,
      "learning_rate": 9.166204567175338e-06,
      "loss": 0.0035,
      "step": 83520
    },
    {
      "epoch": 0.5357354737952686,
      "grad_norm": 0.08615615218877792,
      "learning_rate": 9.165895075423638e-06,
      "loss": 0.0027,
      "step": 83530
    },
    {
      "epoch": 0.5357996106890547,
      "grad_norm": 0.14934763312339783,
      "learning_rate": 9.165585531470294e-06,
      "loss": 0.0041,
      "step": 83540
    },
    {
      "epoch": 0.5358637475828408,
      "grad_norm": 0.12722347676753998,
      "learning_rate": 9.165275935319186e-06,
      "loss": 0.0019,
      "step": 83550
    },
    {
      "epoch": 0.5359278844766269,
      "grad_norm": 0.15359367430210114,
      "learning_rate": 9.164966286974195e-06,
      "loss": 0.0037,
      "step": 83560
    },
    {
      "epoch": 0.535992021370413,
      "grad_norm": 0.055601347237825394,
      "learning_rate": 9.164656586439199e-06,
      "loss": 0.0025,
      "step": 83570
    },
    {
      "epoch": 0.5360561582641991,
      "grad_norm": 0.1848033219575882,
      "learning_rate": 9.16434683371808e-06,
      "loss": 0.0028,
      "step": 83580
    },
    {
      "epoch": 0.5361202951579852,
      "grad_norm": 0.16610196232795715,
      "learning_rate": 9.164037028814718e-06,
      "loss": 0.0039,
      "step": 83590
    },
    {
      "epoch": 0.5361844320517714,
      "grad_norm": 0.012549067847430706,
      "learning_rate": 9.163727171732997e-06,
      "loss": 0.0021,
      "step": 83600
    },
    {
      "epoch": 0.5362485689455574,
      "grad_norm": 0.14107391238212585,
      "learning_rate": 9.163417262476797e-06,
      "loss": 0.003,
      "step": 83610
    },
    {
      "epoch": 0.5363127058393435,
      "grad_norm": 0.21862909197807312,
      "learning_rate": 9.163107301050005e-06,
      "loss": 0.0041,
      "step": 83620
    },
    {
      "epoch": 0.5363768427331296,
      "grad_norm": 0.1434558480978012,
      "learning_rate": 9.162797287456502e-06,
      "loss": 0.0025,
      "step": 83630
    },
    {
      "epoch": 0.5364409796269157,
      "grad_norm": 0.07252000272274017,
      "learning_rate": 9.162487221700176e-06,
      "loss": 0.0029,
      "step": 83640
    },
    {
      "epoch": 0.5365051165207018,
      "grad_norm": 0.2747831642627716,
      "learning_rate": 9.162177103784908e-06,
      "loss": 0.0028,
      "step": 83650
    },
    {
      "epoch": 0.5365692534144879,
      "grad_norm": 0.07841059565544128,
      "learning_rate": 9.161866933714587e-06,
      "loss": 0.0035,
      "step": 83660
    },
    {
      "epoch": 0.536633390308274,
      "grad_norm": 0.12187693268060684,
      "learning_rate": 9.161556711493098e-06,
      "loss": 0.0021,
      "step": 83670
    },
    {
      "epoch": 0.5366975272020601,
      "grad_norm": 0.27229297161102295,
      "learning_rate": 9.161246437124331e-06,
      "loss": 0.004,
      "step": 83680
    },
    {
      "epoch": 0.5367616640958461,
      "grad_norm": 0.06621730327606201,
      "learning_rate": 9.160936110612172e-06,
      "loss": 0.0045,
      "step": 83690
    },
    {
      "epoch": 0.5368258009896323,
      "grad_norm": 0.09571243077516556,
      "learning_rate": 9.160625731960508e-06,
      "loss": 0.0036,
      "step": 83700
    },
    {
      "epoch": 0.5368899378834183,
      "grad_norm": 0.16894841194152832,
      "learning_rate": 9.16031530117323e-06,
      "loss": 0.003,
      "step": 83710
    },
    {
      "epoch": 0.5369540747772045,
      "grad_norm": 0.08017881959676743,
      "learning_rate": 9.16000481825423e-06,
      "loss": 0.0035,
      "step": 83720
    },
    {
      "epoch": 0.5370182116709905,
      "grad_norm": 0.1632446050643921,
      "learning_rate": 9.159694283207395e-06,
      "loss": 0.0031,
      "step": 83730
    },
    {
      "epoch": 0.5370823485647767,
      "grad_norm": 0.005749837029725313,
      "learning_rate": 9.159383696036618e-06,
      "loss": 0.0026,
      "step": 83740
    },
    {
      "epoch": 0.5371464854585628,
      "grad_norm": 0.05540066584944725,
      "learning_rate": 9.159073056745791e-06,
      "loss": 0.0026,
      "step": 83750
    },
    {
      "epoch": 0.5372106223523488,
      "grad_norm": 0.09091489017009735,
      "learning_rate": 9.158762365338807e-06,
      "loss": 0.0029,
      "step": 83760
    },
    {
      "epoch": 0.537274759246135,
      "grad_norm": 0.11926014721393585,
      "learning_rate": 9.158451621819558e-06,
      "loss": 0.0039,
      "step": 83770
    },
    {
      "epoch": 0.537338896139921,
      "grad_norm": 0.2170051485300064,
      "learning_rate": 9.158140826191936e-06,
      "loss": 0.0037,
      "step": 83780
    },
    {
      "epoch": 0.5374030330337072,
      "grad_norm": 0.04482047259807587,
      "learning_rate": 9.15782997845984e-06,
      "loss": 0.0027,
      "step": 83790
    },
    {
      "epoch": 0.5374671699274932,
      "grad_norm": 0.07231690734624863,
      "learning_rate": 9.157519078627162e-06,
      "loss": 0.0027,
      "step": 83800
    },
    {
      "epoch": 0.5375313068212794,
      "grad_norm": 0.15165331959724426,
      "learning_rate": 9.157208126697797e-06,
      "loss": 0.0047,
      "step": 83810
    },
    {
      "epoch": 0.5375954437150654,
      "grad_norm": 0.21467633545398712,
      "learning_rate": 9.156897122675645e-06,
      "loss": 0.0051,
      "step": 83820
    },
    {
      "epoch": 0.5376595806088516,
      "grad_norm": 0.30737295746803284,
      "learning_rate": 9.1565860665646e-06,
      "loss": 0.0049,
      "step": 83830
    },
    {
      "epoch": 0.5377237175026376,
      "grad_norm": 0.2601054310798645,
      "learning_rate": 9.156274958368563e-06,
      "loss": 0.0034,
      "step": 83840
    },
    {
      "epoch": 0.5377878543964237,
      "grad_norm": 0.08841709047555923,
      "learning_rate": 9.15596379809143e-06,
      "loss": 0.0034,
      "step": 83850
    },
    {
      "epoch": 0.5378519912902098,
      "grad_norm": 0.1914767175912857,
      "learning_rate": 9.155652585737098e-06,
      "loss": 0.0028,
      "step": 83860
    },
    {
      "epoch": 0.5379161281839959,
      "grad_norm": 0.6179146766662598,
      "learning_rate": 9.15534132130947e-06,
      "loss": 0.0037,
      "step": 83870
    },
    {
      "epoch": 0.5379802650777821,
      "grad_norm": 0.2201271653175354,
      "learning_rate": 9.155030004812446e-06,
      "loss": 0.004,
      "step": 83880
    },
    {
      "epoch": 0.5380444019715681,
      "grad_norm": 0.3279046416282654,
      "learning_rate": 9.154718636249928e-06,
      "loss": 0.0043,
      "step": 83890
    },
    {
      "epoch": 0.5381085388653543,
      "grad_norm": 0.10568535327911377,
      "learning_rate": 9.154407215625814e-06,
      "loss": 0.0022,
      "step": 83900
    },
    {
      "epoch": 0.5381726757591403,
      "grad_norm": 0.12581396102905273,
      "learning_rate": 9.15409574294401e-06,
      "loss": 0.0029,
      "step": 83910
    },
    {
      "epoch": 0.5382368126529264,
      "grad_norm": 0.2874618470668793,
      "learning_rate": 9.153784218208416e-06,
      "loss": 0.0021,
      "step": 83920
    },
    {
      "epoch": 0.5383009495467125,
      "grad_norm": 0.16133472323417664,
      "learning_rate": 9.153472641422938e-06,
      "loss": 0.0048,
      "step": 83930
    },
    {
      "epoch": 0.5383650864404986,
      "grad_norm": 0.17416003346443176,
      "learning_rate": 9.153161012591478e-06,
      "loss": 0.0029,
      "step": 83940
    },
    {
      "epoch": 0.5384292233342847,
      "grad_norm": 0.15187788009643555,
      "learning_rate": 9.152849331717944e-06,
      "loss": 0.0023,
      "step": 83950
    },
    {
      "epoch": 0.5384933602280708,
      "grad_norm": 0.2774655222892761,
      "learning_rate": 9.15253759880624e-06,
      "loss": 0.0038,
      "step": 83960
    },
    {
      "epoch": 0.5385574971218569,
      "grad_norm": 0.2558445632457733,
      "learning_rate": 9.15222581386027e-06,
      "loss": 0.0029,
      "step": 83970
    },
    {
      "epoch": 0.538621634015643,
      "grad_norm": 0.2403038740158081,
      "learning_rate": 9.151913976883944e-06,
      "loss": 0.0031,
      "step": 83980
    },
    {
      "epoch": 0.538685770909429,
      "grad_norm": 0.18249399960041046,
      "learning_rate": 9.151602087881169e-06,
      "loss": 0.0043,
      "step": 83990
    },
    {
      "epoch": 0.5387499078032152,
      "grad_norm": 0.2081063836812973,
      "learning_rate": 9.151290146855853e-06,
      "loss": 0.0054,
      "step": 84000
    },
    {
      "epoch": 0.5388140446970012,
      "grad_norm": 0.17994354665279388,
      "learning_rate": 9.150978153811904e-06,
      "loss": 0.0038,
      "step": 84010
    },
    {
      "epoch": 0.5388781815907874,
      "grad_norm": 0.07765493541955948,
      "learning_rate": 9.150666108753232e-06,
      "loss": 0.0038,
      "step": 84020
    },
    {
      "epoch": 0.5389423184845735,
      "grad_norm": 0.140521839261055,
      "learning_rate": 9.150354011683748e-06,
      "loss": 0.0049,
      "step": 84030
    },
    {
      "epoch": 0.5390064553783596,
      "grad_norm": 0.10827630013227463,
      "learning_rate": 9.150041862607362e-06,
      "loss": 0.0023,
      "step": 84040
    },
    {
      "epoch": 0.5390705922721457,
      "grad_norm": 0.14978931844234467,
      "learning_rate": 9.149729661527984e-06,
      "loss": 0.0035,
      "step": 84050
    },
    {
      "epoch": 0.5391347291659317,
      "grad_norm": 0.12695223093032837,
      "learning_rate": 9.149417408449528e-06,
      "loss": 0.0029,
      "step": 84060
    },
    {
      "epoch": 0.5391988660597179,
      "grad_norm": 0.20081543922424316,
      "learning_rate": 9.149105103375908e-06,
      "loss": 0.0034,
      "step": 84070
    },
    {
      "epoch": 0.5392630029535039,
      "grad_norm": 0.3733041286468506,
      "learning_rate": 9.148792746311034e-06,
      "loss": 0.0044,
      "step": 84080
    },
    {
      "epoch": 0.5393271398472901,
      "grad_norm": 0.10281600058078766,
      "learning_rate": 9.148480337258824e-06,
      "loss": 0.0024,
      "step": 84090
    },
    {
      "epoch": 0.5393912767410761,
      "grad_norm": 0.12156189978122711,
      "learning_rate": 9.148167876223188e-06,
      "loss": 0.0035,
      "step": 84100
    },
    {
      "epoch": 0.5394554136348623,
      "grad_norm": 0.06447888165712357,
      "learning_rate": 9.147855363208044e-06,
      "loss": 0.0031,
      "step": 84110
    },
    {
      "epoch": 0.5395195505286483,
      "grad_norm": 0.14546574652194977,
      "learning_rate": 9.147542798217309e-06,
      "loss": 0.0029,
      "step": 84120
    },
    {
      "epoch": 0.5395836874224345,
      "grad_norm": 0.4374399781227112,
      "learning_rate": 9.147230181254898e-06,
      "loss": 0.0051,
      "step": 84130
    },
    {
      "epoch": 0.5396478243162205,
      "grad_norm": 0.08889298886060715,
      "learning_rate": 9.14691751232473e-06,
      "loss": 0.0038,
      "step": 84140
    },
    {
      "epoch": 0.5397119612100066,
      "grad_norm": 0.013681549578905106,
      "learning_rate": 9.14660479143072e-06,
      "loss": 0.003,
      "step": 84150
    },
    {
      "epoch": 0.5397760981037928,
      "grad_norm": 0.1032249853014946,
      "learning_rate": 9.146292018576788e-06,
      "loss": 0.0032,
      "step": 84160
    },
    {
      "epoch": 0.5398402349975788,
      "grad_norm": 0.16689357161521912,
      "learning_rate": 9.145979193766855e-06,
      "loss": 0.0023,
      "step": 84170
    },
    {
      "epoch": 0.539904371891365,
      "grad_norm": 0.38058799505233765,
      "learning_rate": 9.14566631700484e-06,
      "loss": 0.0043,
      "step": 84180
    },
    {
      "epoch": 0.539968508785151,
      "grad_norm": 0.2105763554573059,
      "learning_rate": 9.145353388294662e-06,
      "loss": 0.0037,
      "step": 84190
    },
    {
      "epoch": 0.5400326456789372,
      "grad_norm": 0.09349032491445541,
      "learning_rate": 9.145040407640245e-06,
      "loss": 0.0044,
      "step": 84200
    },
    {
      "epoch": 0.5400967825727232,
      "grad_norm": 0.13553950190544128,
      "learning_rate": 9.144727375045507e-06,
      "loss": 0.0043,
      "step": 84210
    },
    {
      "epoch": 0.5401609194665093,
      "grad_norm": 0.07749900966882706,
      "learning_rate": 9.144414290514374e-06,
      "loss": 0.0017,
      "step": 84220
    },
    {
      "epoch": 0.5402250563602954,
      "grad_norm": 0.2512030005455017,
      "learning_rate": 9.144101154050769e-06,
      "loss": 0.0048,
      "step": 84230
    },
    {
      "epoch": 0.5402891932540815,
      "grad_norm": 0.031024569645524025,
      "learning_rate": 9.143787965658615e-06,
      "loss": 0.005,
      "step": 84240
    },
    {
      "epoch": 0.5403533301478676,
      "grad_norm": 0.1312863677740097,
      "learning_rate": 9.143474725341835e-06,
      "loss": 0.0019,
      "step": 84250
    },
    {
      "epoch": 0.5404174670416537,
      "grad_norm": 0.5630732178688049,
      "learning_rate": 9.143161433104355e-06,
      "loss": 0.0033,
      "step": 84260
    },
    {
      "epoch": 0.5404816039354398,
      "grad_norm": 0.17601896822452545,
      "learning_rate": 9.142848088950102e-06,
      "loss": 0.0045,
      "step": 84270
    },
    {
      "epoch": 0.5405457408292259,
      "grad_norm": 0.2565477192401886,
      "learning_rate": 9.142534692883002e-06,
      "loss": 0.0025,
      "step": 84280
    },
    {
      "epoch": 0.540609877723012,
      "grad_norm": 0.21444521844387054,
      "learning_rate": 9.14222124490698e-06,
      "loss": 0.0034,
      "step": 84290
    },
    {
      "epoch": 0.5406740146167981,
      "grad_norm": 0.07533738762140274,
      "learning_rate": 9.141907745025966e-06,
      "loss": 0.002,
      "step": 84300
    },
    {
      "epoch": 0.5407381515105842,
      "grad_norm": 0.0890752300620079,
      "learning_rate": 9.141594193243888e-06,
      "loss": 0.004,
      "step": 84310
    },
    {
      "epoch": 0.5408022884043703,
      "grad_norm": 0.2074747085571289,
      "learning_rate": 9.141280589564676e-06,
      "loss": 0.004,
      "step": 84320
    },
    {
      "epoch": 0.5408664252981564,
      "grad_norm": 0.16621294617652893,
      "learning_rate": 9.140966933992256e-06,
      "loss": 0.0041,
      "step": 84330
    },
    {
      "epoch": 0.5409305621919425,
      "grad_norm": 0.04957873001694679,
      "learning_rate": 9.140653226530564e-06,
      "loss": 0.0026,
      "step": 84340
    },
    {
      "epoch": 0.5409946990857286,
      "grad_norm": 0.3442468047142029,
      "learning_rate": 9.140339467183525e-06,
      "loss": 0.0041,
      "step": 84350
    },
    {
      "epoch": 0.5410588359795147,
      "grad_norm": 0.07848238199949265,
      "learning_rate": 9.140025655955076e-06,
      "loss": 0.004,
      "step": 84360
    },
    {
      "epoch": 0.5411229728733008,
      "grad_norm": 0.044110093265771866,
      "learning_rate": 9.139711792849144e-06,
      "loss": 0.0032,
      "step": 84370
    },
    {
      "epoch": 0.5411871097670868,
      "grad_norm": 0.2451447993516922,
      "learning_rate": 9.139397877869667e-06,
      "loss": 0.0033,
      "step": 84380
    },
    {
      "epoch": 0.541251246660873,
      "grad_norm": 0.0834902822971344,
      "learning_rate": 9.139083911020574e-06,
      "loss": 0.0054,
      "step": 84390
    },
    {
      "epoch": 0.541315383554659,
      "grad_norm": 0.12984836101531982,
      "learning_rate": 9.138769892305803e-06,
      "loss": 0.0022,
      "step": 84400
    },
    {
      "epoch": 0.5413795204484452,
      "grad_norm": 0.14569932222366333,
      "learning_rate": 9.138455821729287e-06,
      "loss": 0.0043,
      "step": 84410
    },
    {
      "epoch": 0.5414436573422312,
      "grad_norm": 0.30914196372032166,
      "learning_rate": 9.138141699294962e-06,
      "loss": 0.0087,
      "step": 84420
    },
    {
      "epoch": 0.5415077942360174,
      "grad_norm": 0.27151378989219666,
      "learning_rate": 9.137827525006763e-06,
      "loss": 0.0033,
      "step": 84430
    },
    {
      "epoch": 0.5415719311298035,
      "grad_norm": 0.14409923553466797,
      "learning_rate": 9.13751329886863e-06,
      "loss": 0.0029,
      "step": 84440
    },
    {
      "epoch": 0.5416360680235895,
      "grad_norm": 0.1737694889307022,
      "learning_rate": 9.137199020884497e-06,
      "loss": 0.0025,
      "step": 84450
    },
    {
      "epoch": 0.5417002049173757,
      "grad_norm": 0.019738130271434784,
      "learning_rate": 9.136884691058303e-06,
      "loss": 0.0026,
      "step": 84460
    },
    {
      "epoch": 0.5417643418111617,
      "grad_norm": 0.0797569677233696,
      "learning_rate": 9.136570309393988e-06,
      "loss": 0.0034,
      "step": 84470
    },
    {
      "epoch": 0.5418284787049479,
      "grad_norm": 0.21312910318374634,
      "learning_rate": 9.13625587589549e-06,
      "loss": 0.0034,
      "step": 84480
    },
    {
      "epoch": 0.5418926155987339,
      "grad_norm": 0.151839017868042,
      "learning_rate": 9.135941390566749e-06,
      "loss": 0.0021,
      "step": 84490
    },
    {
      "epoch": 0.5419567524925201,
      "grad_norm": 0.28048601746559143,
      "learning_rate": 9.135626853411707e-06,
      "loss": 0.0054,
      "step": 84500
    },
    {
      "epoch": 0.5420208893863061,
      "grad_norm": 0.11831089854240417,
      "learning_rate": 9.135312264434306e-06,
      "loss": 0.0027,
      "step": 84510
    },
    {
      "epoch": 0.5420850262800923,
      "grad_norm": 0.0896664634346962,
      "learning_rate": 9.134997623638487e-06,
      "loss": 0.0038,
      "step": 84520
    },
    {
      "epoch": 0.5421491631738783,
      "grad_norm": 0.09071475267410278,
      "learning_rate": 9.134682931028192e-06,
      "loss": 0.0044,
      "step": 84530
    },
    {
      "epoch": 0.5422133000676644,
      "grad_norm": 0.08034510165452957,
      "learning_rate": 9.134368186607363e-06,
      "loss": 0.0049,
      "step": 84540
    },
    {
      "epoch": 0.5422774369614505,
      "grad_norm": 0.30859583616256714,
      "learning_rate": 9.134053390379948e-06,
      "loss": 0.0042,
      "step": 84550
    },
    {
      "epoch": 0.5423415738552366,
      "grad_norm": 0.24405072629451752,
      "learning_rate": 9.133738542349888e-06,
      "loss": 0.0033,
      "step": 84560
    },
    {
      "epoch": 0.5424057107490227,
      "grad_norm": 0.14305658638477325,
      "learning_rate": 9.13342364252113e-06,
      "loss": 0.0044,
      "step": 84570
    },
    {
      "epoch": 0.5424698476428088,
      "grad_norm": 0.2623572051525116,
      "learning_rate": 9.13310869089762e-06,
      "loss": 0.0069,
      "step": 84580
    },
    {
      "epoch": 0.542533984536595,
      "grad_norm": 0.35897016525268555,
      "learning_rate": 9.132793687483305e-06,
      "loss": 0.0034,
      "step": 84590
    },
    {
      "epoch": 0.542598121430381,
      "grad_norm": 0.1529109925031662,
      "learning_rate": 9.132478632282132e-06,
      "loss": 0.0039,
      "step": 84600
    },
    {
      "epoch": 0.5426622583241671,
      "grad_norm": 0.20279178023338318,
      "learning_rate": 9.132163525298047e-06,
      "loss": 0.0051,
      "step": 84610
    },
    {
      "epoch": 0.5427263952179532,
      "grad_norm": 0.09714586287736893,
      "learning_rate": 9.131848366535e-06,
      "loss": 0.0052,
      "step": 84620
    },
    {
      "epoch": 0.5427905321117393,
      "grad_norm": 0.21548357605934143,
      "learning_rate": 9.131533155996939e-06,
      "loss": 0.0028,
      "step": 84630
    },
    {
      "epoch": 0.5428546690055254,
      "grad_norm": 0.12653973698616028,
      "learning_rate": 9.131217893687817e-06,
      "loss": 0.0044,
      "step": 84640
    },
    {
      "epoch": 0.5429188058993115,
      "grad_norm": 0.0521419532597065,
      "learning_rate": 9.130902579611581e-06,
      "loss": 0.0027,
      "step": 84650
    },
    {
      "epoch": 0.5429829427930976,
      "grad_norm": 0.2173508256673813,
      "learning_rate": 9.130587213772181e-06,
      "loss": 0.0028,
      "step": 84660
    },
    {
      "epoch": 0.5430470796868837,
      "grad_norm": 0.3113865256309509,
      "learning_rate": 9.130271796173576e-06,
      "loss": 0.0036,
      "step": 84670
    },
    {
      "epoch": 0.5431112165806697,
      "grad_norm": 0.24822306632995605,
      "learning_rate": 9.12995632681971e-06,
      "loss": 0.006,
      "step": 84680
    },
    {
      "epoch": 0.5431753534744559,
      "grad_norm": 0.22321628034114838,
      "learning_rate": 9.129640805714542e-06,
      "loss": 0.0031,
      "step": 84690
    },
    {
      "epoch": 0.5432394903682419,
      "grad_norm": 0.1794038861989975,
      "learning_rate": 9.129325232862021e-06,
      "loss": 0.0027,
      "step": 84700
    },
    {
      "epoch": 0.5433036272620281,
      "grad_norm": 0.21157395839691162,
      "learning_rate": 9.129009608266104e-06,
      "loss": 0.0026,
      "step": 84710
    },
    {
      "epoch": 0.5433677641558141,
      "grad_norm": 0.19182851910591125,
      "learning_rate": 9.128693931930747e-06,
      "loss": 0.0033,
      "step": 84720
    },
    {
      "epoch": 0.5434319010496003,
      "grad_norm": 0.17911753058433533,
      "learning_rate": 9.128378203859902e-06,
      "loss": 0.0039,
      "step": 84730
    },
    {
      "epoch": 0.5434960379433864,
      "grad_norm": 0.22276780009269714,
      "learning_rate": 9.12806242405753e-06,
      "loss": 0.0025,
      "step": 84740
    },
    {
      "epoch": 0.5435601748371724,
      "grad_norm": 0.08404522389173508,
      "learning_rate": 9.127746592527585e-06,
      "loss": 0.0039,
      "step": 84750
    },
    {
      "epoch": 0.5436243117309586,
      "grad_norm": 0.1111041009426117,
      "learning_rate": 9.127430709274025e-06,
      "loss": 0.0024,
      "step": 84760
    },
    {
      "epoch": 0.5436884486247446,
      "grad_norm": 0.4655708372592926,
      "learning_rate": 9.127114774300807e-06,
      "loss": 0.0058,
      "step": 84770
    },
    {
      "epoch": 0.5437525855185308,
      "grad_norm": 0.11205190420150757,
      "learning_rate": 9.126798787611892e-06,
      "loss": 0.0058,
      "step": 84780
    },
    {
      "epoch": 0.5438167224123168,
      "grad_norm": 0.04771514981985092,
      "learning_rate": 9.12648274921124e-06,
      "loss": 0.0021,
      "step": 84790
    },
    {
      "epoch": 0.543880859306103,
      "grad_norm": 0.27232682704925537,
      "learning_rate": 9.12616665910281e-06,
      "loss": 0.0034,
      "step": 84800
    },
    {
      "epoch": 0.543944996199889,
      "grad_norm": 0.17340819537639618,
      "learning_rate": 9.125850517290561e-06,
      "loss": 0.0035,
      "step": 84810
    },
    {
      "epoch": 0.5440091330936752,
      "grad_norm": 0.1362583041191101,
      "learning_rate": 9.125534323778458e-06,
      "loss": 0.0037,
      "step": 84820
    },
    {
      "epoch": 0.5440732699874612,
      "grad_norm": 0.15964961051940918,
      "learning_rate": 9.12521807857046e-06,
      "loss": 0.0039,
      "step": 84830
    },
    {
      "epoch": 0.5441374068812473,
      "grad_norm": 0.29718807339668274,
      "learning_rate": 9.124901781670533e-06,
      "loss": 0.0033,
      "step": 84840
    },
    {
      "epoch": 0.5442015437750334,
      "grad_norm": 0.1734071522951126,
      "learning_rate": 9.124585433082638e-06,
      "loss": 0.0024,
      "step": 84850
    },
    {
      "epoch": 0.5442656806688195,
      "grad_norm": 0.11542264372110367,
      "learning_rate": 9.12426903281074e-06,
      "loss": 0.0031,
      "step": 84860
    },
    {
      "epoch": 0.5443298175626057,
      "grad_norm": 0.06102282553911209,
      "learning_rate": 9.123952580858801e-06,
      "loss": 0.0038,
      "step": 84870
    },
    {
      "epoch": 0.5443939544563917,
      "grad_norm": 0.1172720268368721,
      "learning_rate": 9.123636077230792e-06,
      "loss": 0.002,
      "step": 84880
    },
    {
      "epoch": 0.5444580913501779,
      "grad_norm": 0.29675355553627014,
      "learning_rate": 9.123319521930674e-06,
      "loss": 0.0022,
      "step": 84890
    },
    {
      "epoch": 0.5445222282439639,
      "grad_norm": 0.037560850381851196,
      "learning_rate": 9.123002914962417e-06,
      "loss": 0.0032,
      "step": 84900
    },
    {
      "epoch": 0.54458636513775,
      "grad_norm": 0.15551374852657318,
      "learning_rate": 9.122686256329985e-06,
      "loss": 0.0039,
      "step": 84910
    },
    {
      "epoch": 0.5446505020315361,
      "grad_norm": 0.23275664448738098,
      "learning_rate": 9.122369546037348e-06,
      "loss": 0.0046,
      "step": 84920
    },
    {
      "epoch": 0.5447146389253222,
      "grad_norm": 0.30361834168434143,
      "learning_rate": 9.122052784088476e-06,
      "loss": 0.0046,
      "step": 84930
    },
    {
      "epoch": 0.5447787758191083,
      "grad_norm": 0.216138556599617,
      "learning_rate": 9.121735970487335e-06,
      "loss": 0.0048,
      "step": 84940
    },
    {
      "epoch": 0.5448429127128944,
      "grad_norm": 0.08227672427892685,
      "learning_rate": 9.121419105237898e-06,
      "loss": 0.0029,
      "step": 84950
    },
    {
      "epoch": 0.5449070496066805,
      "grad_norm": 0.4803912937641144,
      "learning_rate": 9.121102188344132e-06,
      "loss": 0.0021,
      "step": 84960
    },
    {
      "epoch": 0.5449711865004666,
      "grad_norm": 0.0508844256401062,
      "learning_rate": 9.120785219810012e-06,
      "loss": 0.0038,
      "step": 84970
    },
    {
      "epoch": 0.5450353233942526,
      "grad_norm": 0.1788831651210785,
      "learning_rate": 9.120468199639507e-06,
      "loss": 0.0042,
      "step": 84980
    },
    {
      "epoch": 0.5450994602880388,
      "grad_norm": 0.38108617067337036,
      "learning_rate": 9.120151127836591e-06,
      "loss": 0.0053,
      "step": 84990
    },
    {
      "epoch": 0.5451635971818248,
      "grad_norm": 0.0781199187040329,
      "learning_rate": 9.119834004405239e-06,
      "loss": 0.0062,
      "step": 85000
    },
    {
      "epoch": 0.545227734075611,
      "grad_norm": 0.08282937109470367,
      "learning_rate": 9.11951682934942e-06,
      "loss": 0.0022,
      "step": 85010
    },
    {
      "epoch": 0.5452918709693971,
      "grad_norm": 0.11112157255411148,
      "learning_rate": 9.119199602673112e-06,
      "loss": 0.0021,
      "step": 85020
    },
    {
      "epoch": 0.5453560078631832,
      "grad_norm": 0.09030723571777344,
      "learning_rate": 9.118882324380287e-06,
      "loss": 0.0054,
      "step": 85030
    },
    {
      "epoch": 0.5454201447569693,
      "grad_norm": 0.07033137232065201,
      "learning_rate": 9.118564994474925e-06,
      "loss": 0.004,
      "step": 85040
    },
    {
      "epoch": 0.5454842816507554,
      "grad_norm": 0.18817007541656494,
      "learning_rate": 9.118247612961e-06,
      "loss": 0.0041,
      "step": 85050
    },
    {
      "epoch": 0.5455484185445415,
      "grad_norm": 0.26980462670326233,
      "learning_rate": 9.117930179842488e-06,
      "loss": 0.0027,
      "step": 85060
    },
    {
      "epoch": 0.5456125554383275,
      "grad_norm": 0.10838302224874496,
      "learning_rate": 9.117612695123368e-06,
      "loss": 0.0047,
      "step": 85070
    },
    {
      "epoch": 0.5456766923321137,
      "grad_norm": 0.16677549481391907,
      "learning_rate": 9.117295158807618e-06,
      "loss": 0.0021,
      "step": 85080
    },
    {
      "epoch": 0.5457408292258997,
      "grad_norm": 0.1712741255760193,
      "learning_rate": 9.116977570899219e-06,
      "loss": 0.0061,
      "step": 85090
    },
    {
      "epoch": 0.5458049661196859,
      "grad_norm": 0.05557706952095032,
      "learning_rate": 9.116659931402147e-06,
      "loss": 0.0065,
      "step": 85100
    },
    {
      "epoch": 0.5458691030134719,
      "grad_norm": 0.009485971182584763,
      "learning_rate": 9.116342240320384e-06,
      "loss": 0.0028,
      "step": 85110
    },
    {
      "epoch": 0.5459332399072581,
      "grad_norm": 0.20825040340423584,
      "learning_rate": 9.11602449765791e-06,
      "loss": 0.005,
      "step": 85120
    },
    {
      "epoch": 0.5459973768010441,
      "grad_norm": 0.12041359394788742,
      "learning_rate": 9.115706703418709e-06,
      "loss": 0.0025,
      "step": 85130
    },
    {
      "epoch": 0.5460615136948302,
      "grad_norm": 0.2229558229446411,
      "learning_rate": 9.11538885760676e-06,
      "loss": 0.0039,
      "step": 85140
    },
    {
      "epoch": 0.5461256505886164,
      "grad_norm": 0.23140066862106323,
      "learning_rate": 9.11507096022605e-06,
      "loss": 0.0035,
      "step": 85150
    },
    {
      "epoch": 0.5461897874824024,
      "grad_norm": 0.25123244524002075,
      "learning_rate": 9.114753011280557e-06,
      "loss": 0.0039,
      "step": 85160
    },
    {
      "epoch": 0.5462539243761886,
      "grad_norm": 0.1634216457605362,
      "learning_rate": 9.11443501077427e-06,
      "loss": 0.0028,
      "step": 85170
    },
    {
      "epoch": 0.5463180612699746,
      "grad_norm": 0.2737909257411957,
      "learning_rate": 9.11411695871117e-06,
      "loss": 0.0037,
      "step": 85180
    },
    {
      "epoch": 0.5463821981637608,
      "grad_norm": 0.08838663250207901,
      "learning_rate": 9.113798855095244e-06,
      "loss": 0.0022,
      "step": 85190
    },
    {
      "epoch": 0.5464463350575468,
      "grad_norm": 0.050015341490507126,
      "learning_rate": 9.113480699930479e-06,
      "loss": 0.0018,
      "step": 85200
    },
    {
      "epoch": 0.546510471951333,
      "grad_norm": 0.2101522982120514,
      "learning_rate": 9.113162493220862e-06,
      "loss": 0.0049,
      "step": 85210
    },
    {
      "epoch": 0.546574608845119,
      "grad_norm": 0.10002795606851578,
      "learning_rate": 9.112844234970379e-06,
      "loss": 0.0047,
      "step": 85220
    },
    {
      "epoch": 0.5466387457389051,
      "grad_norm": 0.17288456857204437,
      "learning_rate": 9.112525925183017e-06,
      "loss": 0.0036,
      "step": 85230
    },
    {
      "epoch": 0.5467028826326912,
      "grad_norm": 0.34862926602363586,
      "learning_rate": 9.112207563862767e-06,
      "loss": 0.0035,
      "step": 85240
    },
    {
      "epoch": 0.5467670195264773,
      "grad_norm": 0.08521570265293121,
      "learning_rate": 9.111889151013618e-06,
      "loss": 0.0038,
      "step": 85250
    },
    {
      "epoch": 0.5468311564202634,
      "grad_norm": 0.14284466207027435,
      "learning_rate": 9.11157068663956e-06,
      "loss": 0.0033,
      "step": 85260
    },
    {
      "epoch": 0.5468952933140495,
      "grad_norm": 0.24869143962860107,
      "learning_rate": 9.11125217074458e-06,
      "loss": 0.0052,
      "step": 85270
    },
    {
      "epoch": 0.5469594302078356,
      "grad_norm": 0.07540538907051086,
      "learning_rate": 9.110933603332674e-06,
      "loss": 0.0039,
      "step": 85280
    },
    {
      "epoch": 0.5470235671016217,
      "grad_norm": 0.18324674665927887,
      "learning_rate": 9.110614984407831e-06,
      "loss": 0.0037,
      "step": 85290
    },
    {
      "epoch": 0.5470877039954078,
      "grad_norm": 0.009447806514799595,
      "learning_rate": 9.110296313974043e-06,
      "loss": 0.0037,
      "step": 85300
    },
    {
      "epoch": 0.5471518408891939,
      "grad_norm": 0.4537900388240814,
      "learning_rate": 9.109977592035308e-06,
      "loss": 0.0034,
      "step": 85310
    },
    {
      "epoch": 0.54721597778298,
      "grad_norm": 0.16369648277759552,
      "learning_rate": 9.109658818595614e-06,
      "loss": 0.0039,
      "step": 85320
    },
    {
      "epoch": 0.5472801146767661,
      "grad_norm": 0.13680054247379303,
      "learning_rate": 9.10933999365896e-06,
      "loss": 0.003,
      "step": 85330
    },
    {
      "epoch": 0.5473442515705522,
      "grad_norm": 0.104710653424263,
      "learning_rate": 9.109021117229336e-06,
      "loss": 0.0023,
      "step": 85340
    },
    {
      "epoch": 0.5474083884643383,
      "grad_norm": 0.1867726445198059,
      "learning_rate": 9.108702189310742e-06,
      "loss": 0.0033,
      "step": 85350
    },
    {
      "epoch": 0.5474725253581244,
      "grad_norm": 0.07051552832126617,
      "learning_rate": 9.108383209907173e-06,
      "loss": 0.009,
      "step": 85360
    },
    {
      "epoch": 0.5475366622519104,
      "grad_norm": 0.06497001647949219,
      "learning_rate": 9.108064179022626e-06,
      "loss": 0.0028,
      "step": 85370
    },
    {
      "epoch": 0.5476007991456966,
      "grad_norm": 0.08399824798107147,
      "learning_rate": 9.1077450966611e-06,
      "loss": 0.0023,
      "step": 85380
    },
    {
      "epoch": 0.5476649360394826,
      "grad_norm": 0.19393685460090637,
      "learning_rate": 9.107425962826592e-06,
      "loss": 0.004,
      "step": 85390
    },
    {
      "epoch": 0.5477290729332688,
      "grad_norm": 0.2073168158531189,
      "learning_rate": 9.1071067775231e-06,
      "loss": 0.0052,
      "step": 85400
    },
    {
      "epoch": 0.5477932098270548,
      "grad_norm": 0.06114586070179939,
      "learning_rate": 9.106787540754626e-06,
      "loss": 0.0028,
      "step": 85410
    },
    {
      "epoch": 0.547857346720841,
      "grad_norm": 0.07383716851472855,
      "learning_rate": 9.106468252525168e-06,
      "loss": 0.0039,
      "step": 85420
    },
    {
      "epoch": 0.5479214836146271,
      "grad_norm": 0.14534394443035126,
      "learning_rate": 9.106148912838729e-06,
      "loss": 0.0037,
      "step": 85430
    },
    {
      "epoch": 0.5479856205084132,
      "grad_norm": 0.2565113604068756,
      "learning_rate": 9.10582952169931e-06,
      "loss": 0.0081,
      "step": 85440
    },
    {
      "epoch": 0.5480497574021993,
      "grad_norm": 0.2744908630847931,
      "learning_rate": 9.10551007911091e-06,
      "loss": 0.0044,
      "step": 85450
    },
    {
      "epoch": 0.5481138942959853,
      "grad_norm": 0.1304347813129425,
      "learning_rate": 9.105190585077537e-06,
      "loss": 0.0026,
      "step": 85460
    },
    {
      "epoch": 0.5481780311897715,
      "grad_norm": 0.18746119737625122,
      "learning_rate": 9.104871039603192e-06,
      "loss": 0.0041,
      "step": 85470
    },
    {
      "epoch": 0.5482421680835575,
      "grad_norm": 0.06296703964471817,
      "learning_rate": 9.104551442691878e-06,
      "loss": 0.005,
      "step": 85480
    },
    {
      "epoch": 0.5483063049773437,
      "grad_norm": 0.22583769261837006,
      "learning_rate": 9.1042317943476e-06,
      "loss": 0.0029,
      "step": 85490
    },
    {
      "epoch": 0.5483704418711297,
      "grad_norm": 0.21536467969417572,
      "learning_rate": 9.103912094574365e-06,
      "loss": 0.0043,
      "step": 85500
    },
    {
      "epoch": 0.5484345787649159,
      "grad_norm": 0.03523802012205124,
      "learning_rate": 9.10359234337618e-06,
      "loss": 0.0038,
      "step": 85510
    },
    {
      "epoch": 0.5484987156587019,
      "grad_norm": 0.08168116956949234,
      "learning_rate": 9.103272540757048e-06,
      "loss": 0.0022,
      "step": 85520
    },
    {
      "epoch": 0.548562852552488,
      "grad_norm": 0.19365718960762024,
      "learning_rate": 9.10295268672098e-06,
      "loss": 0.0052,
      "step": 85530
    },
    {
      "epoch": 0.5486269894462741,
      "grad_norm": 0.0632062554359436,
      "learning_rate": 9.102632781271983e-06,
      "loss": 0.0029,
      "step": 85540
    },
    {
      "epoch": 0.5486911263400602,
      "grad_norm": 0.0994727835059166,
      "learning_rate": 9.102312824414064e-06,
      "loss": 0.0036,
      "step": 85550
    },
    {
      "epoch": 0.5487552632338463,
      "grad_norm": 0.11295632272958755,
      "learning_rate": 9.101992816151236e-06,
      "loss": 0.0033,
      "step": 85560
    },
    {
      "epoch": 0.5488194001276324,
      "grad_norm": 0.15867988765239716,
      "learning_rate": 9.101672756487503e-06,
      "loss": 0.0072,
      "step": 85570
    },
    {
      "epoch": 0.5488835370214186,
      "grad_norm": 0.161295548081398,
      "learning_rate": 9.101352645426879e-06,
      "loss": 0.0028,
      "step": 85580
    },
    {
      "epoch": 0.5489476739152046,
      "grad_norm": 0.07710425555706024,
      "learning_rate": 9.101032482973378e-06,
      "loss": 0.0029,
      "step": 85590
    },
    {
      "epoch": 0.5490118108089908,
      "grad_norm": 0.07768481969833374,
      "learning_rate": 9.100712269131008e-06,
      "loss": 0.0036,
      "step": 85600
    },
    {
      "epoch": 0.5490759477027768,
      "grad_norm": 0.22926555573940277,
      "learning_rate": 9.10039200390378e-06,
      "loss": 0.0043,
      "step": 85610
    },
    {
      "epoch": 0.5491400845965629,
      "grad_norm": 0.10633791983127594,
      "learning_rate": 9.100071687295712e-06,
      "loss": 0.0026,
      "step": 85620
    },
    {
      "epoch": 0.549204221490349,
      "grad_norm": 0.16838468611240387,
      "learning_rate": 9.099751319310814e-06,
      "loss": 0.0033,
      "step": 85630
    },
    {
      "epoch": 0.5492683583841351,
      "grad_norm": 0.16128893196582794,
      "learning_rate": 9.099430899953103e-06,
      "loss": 0.0034,
      "step": 85640
    },
    {
      "epoch": 0.5493324952779212,
      "grad_norm": 0.18620869517326355,
      "learning_rate": 9.099110429226593e-06,
      "loss": 0.0055,
      "step": 85650
    },
    {
      "epoch": 0.5493966321717073,
      "grad_norm": 0.11803747713565826,
      "learning_rate": 9.098789907135302e-06,
      "loss": 0.0024,
      "step": 85660
    },
    {
      "epoch": 0.5494607690654933,
      "grad_norm": 0.19250069558620453,
      "learning_rate": 9.09846933368324e-06,
      "loss": 0.0032,
      "step": 85670
    },
    {
      "epoch": 0.5495249059592795,
      "grad_norm": 0.15483903884887695,
      "learning_rate": 9.09814870887443e-06,
      "loss": 0.004,
      "step": 85680
    },
    {
      "epoch": 0.5495890428530655,
      "grad_norm": 0.030415749177336693,
      "learning_rate": 9.097828032712888e-06,
      "loss": 0.0053,
      "step": 85690
    },
    {
      "epoch": 0.5496531797468517,
      "grad_norm": 0.1431485414505005,
      "learning_rate": 9.097507305202632e-06,
      "loss": 0.0025,
      "step": 85700
    },
    {
      "epoch": 0.5497173166406378,
      "grad_norm": 0.3654477000236511,
      "learning_rate": 9.09718652634768e-06,
      "loss": 0.005,
      "step": 85710
    },
    {
      "epoch": 0.5497814535344239,
      "grad_norm": 0.06691776216030121,
      "learning_rate": 9.096865696152053e-06,
      "loss": 0.0026,
      "step": 85720
    },
    {
      "epoch": 0.54984559042821,
      "grad_norm": 0.043519288301467896,
      "learning_rate": 9.096544814619772e-06,
      "loss": 0.0029,
      "step": 85730
    },
    {
      "epoch": 0.5499097273219961,
      "grad_norm": 0.2356233447790146,
      "learning_rate": 9.096223881754855e-06,
      "loss": 0.0039,
      "step": 85740
    },
    {
      "epoch": 0.5499738642157822,
      "grad_norm": 1.9528368711471558,
      "learning_rate": 9.095902897561326e-06,
      "loss": 0.0072,
      "step": 85750
    },
    {
      "epoch": 0.5500380011095682,
      "grad_norm": 0.13375617563724518,
      "learning_rate": 9.095581862043207e-06,
      "loss": 0.0044,
      "step": 85760
    },
    {
      "epoch": 0.5501021380033544,
      "grad_norm": 0.13341820240020752,
      "learning_rate": 9.09526077520452e-06,
      "loss": 0.0025,
      "step": 85770
    },
    {
      "epoch": 0.5501662748971404,
      "grad_norm": 0.2529837489128113,
      "learning_rate": 9.094939637049288e-06,
      "loss": 0.0028,
      "step": 85780
    },
    {
      "epoch": 0.5502304117909266,
      "grad_norm": 0.14536824822425842,
      "learning_rate": 9.094618447581537e-06,
      "loss": 0.0032,
      "step": 85790
    },
    {
      "epoch": 0.5502945486847126,
      "grad_norm": 0.13131175935268402,
      "learning_rate": 9.09429720680529e-06,
      "loss": 0.0058,
      "step": 85800
    },
    {
      "epoch": 0.5503586855784988,
      "grad_norm": 0.0458420030772686,
      "learning_rate": 9.093975914724572e-06,
      "loss": 0.0023,
      "step": 85810
    },
    {
      "epoch": 0.5504228224722848,
      "grad_norm": 0.137325718998909,
      "learning_rate": 9.093654571343411e-06,
      "loss": 0.0037,
      "step": 85820
    },
    {
      "epoch": 0.550486959366071,
      "grad_norm": 0.07254275679588318,
      "learning_rate": 9.093333176665831e-06,
      "loss": 0.0084,
      "step": 85830
    },
    {
      "epoch": 0.550551096259857,
      "grad_norm": 0.21434733271598816,
      "learning_rate": 9.093011730695861e-06,
      "loss": 0.0032,
      "step": 85840
    },
    {
      "epoch": 0.5506152331536431,
      "grad_norm": 0.11275189369916916,
      "learning_rate": 9.092690233437531e-06,
      "loss": 0.0024,
      "step": 85850
    },
    {
      "epoch": 0.5506793700474293,
      "grad_norm": 0.25155019760131836,
      "learning_rate": 9.092368684894865e-06,
      "loss": 0.003,
      "step": 85860
    },
    {
      "epoch": 0.5507435069412153,
      "grad_norm": 0.14386850595474243,
      "learning_rate": 9.092047085071895e-06,
      "loss": 0.0035,
      "step": 85870
    },
    {
      "epoch": 0.5508076438350015,
      "grad_norm": 0.08390353620052338,
      "learning_rate": 9.091725433972651e-06,
      "loss": 0.0041,
      "step": 85880
    },
    {
      "epoch": 0.5508717807287875,
      "grad_norm": 0.09561295807361603,
      "learning_rate": 9.091403731601165e-06,
      "loss": 0.0017,
      "step": 85890
    },
    {
      "epoch": 0.5509359176225737,
      "grad_norm": 0.14668424427509308,
      "learning_rate": 9.091081977961462e-06,
      "loss": 0.0064,
      "step": 85900
    },
    {
      "epoch": 0.5510000545163597,
      "grad_norm": 0.26864367723464966,
      "learning_rate": 9.090760173057581e-06,
      "loss": 0.0021,
      "step": 85910
    },
    {
      "epoch": 0.5510641914101458,
      "grad_norm": 0.2498502880334854,
      "learning_rate": 9.090438316893551e-06,
      "loss": 0.002,
      "step": 85920
    },
    {
      "epoch": 0.5511283283039319,
      "grad_norm": 0.2963554859161377,
      "learning_rate": 9.090116409473404e-06,
      "loss": 0.0048,
      "step": 85930
    },
    {
      "epoch": 0.551192465197718,
      "grad_norm": 0.18236757814884186,
      "learning_rate": 9.089794450801178e-06,
      "loss": 0.0043,
      "step": 85940
    },
    {
      "epoch": 0.5512566020915041,
      "grad_norm": 0.09088752418756485,
      "learning_rate": 9.089472440880901e-06,
      "loss": 0.0035,
      "step": 85950
    },
    {
      "epoch": 0.5513207389852902,
      "grad_norm": 0.14687426388263702,
      "learning_rate": 9.089150379716614e-06,
      "loss": 0.0028,
      "step": 85960
    },
    {
      "epoch": 0.5513848758790763,
      "grad_norm": 0.07180049270391464,
      "learning_rate": 9.08882826731235e-06,
      "loss": 0.0057,
      "step": 85970
    },
    {
      "epoch": 0.5514490127728624,
      "grad_norm": 0.06035429984331131,
      "learning_rate": 9.088506103672146e-06,
      "loss": 0.0025,
      "step": 85980
    },
    {
      "epoch": 0.5515131496666485,
      "grad_norm": 0.2776930034160614,
      "learning_rate": 9.088183888800038e-06,
      "loss": 0.0034,
      "step": 85990
    },
    {
      "epoch": 0.5515772865604346,
      "grad_norm": 0.11867208778858185,
      "learning_rate": 9.087861622700062e-06,
      "loss": 0.0031,
      "step": 86000
    },
    {
      "epoch": 0.5516414234542207,
      "grad_norm": 0.3291597068309784,
      "learning_rate": 9.087539305376261e-06,
      "loss": 0.0045,
      "step": 86010
    },
    {
      "epoch": 0.5517055603480068,
      "grad_norm": 0.19839000701904297,
      "learning_rate": 9.08721693683267e-06,
      "loss": 0.0026,
      "step": 86020
    },
    {
      "epoch": 0.5517696972417929,
      "grad_norm": 0.05654887109994888,
      "learning_rate": 9.08689451707333e-06,
      "loss": 0.0033,
      "step": 86030
    },
    {
      "epoch": 0.551833834135579,
      "grad_norm": 0.0997452586889267,
      "learning_rate": 9.086572046102281e-06,
      "loss": 0.0045,
      "step": 86040
    },
    {
      "epoch": 0.5518979710293651,
      "grad_norm": 0.28388577699661255,
      "learning_rate": 9.086249523923563e-06,
      "loss": 0.003,
      "step": 86050
    },
    {
      "epoch": 0.5519621079231511,
      "grad_norm": 0.34730029106140137,
      "learning_rate": 9.085926950541219e-06,
      "loss": 0.0045,
      "step": 86060
    },
    {
      "epoch": 0.5520262448169373,
      "grad_norm": 0.10112051665782928,
      "learning_rate": 9.08560432595929e-06,
      "loss": 0.0038,
      "step": 86070
    },
    {
      "epoch": 0.5520903817107233,
      "grad_norm": 0.2704380452632904,
      "learning_rate": 9.085281650181818e-06,
      "loss": 0.0039,
      "step": 86080
    },
    {
      "epoch": 0.5521545186045095,
      "grad_norm": 0.023337703198194504,
      "learning_rate": 9.084958923212848e-06,
      "loss": 0.0033,
      "step": 86090
    },
    {
      "epoch": 0.5522186554982955,
      "grad_norm": 0.12786415219306946,
      "learning_rate": 9.084636145056423e-06,
      "loss": 0.0031,
      "step": 86100
    },
    {
      "epoch": 0.5522827923920817,
      "grad_norm": 0.29230159521102905,
      "learning_rate": 9.084313315716589e-06,
      "loss": 0.0029,
      "step": 86110
    },
    {
      "epoch": 0.5523469292858677,
      "grad_norm": 0.160085529088974,
      "learning_rate": 9.083990435197389e-06,
      "loss": 0.0036,
      "step": 86120
    },
    {
      "epoch": 0.5524110661796539,
      "grad_norm": 0.14612051844596863,
      "learning_rate": 9.08366750350287e-06,
      "loss": 0.005,
      "step": 86130
    },
    {
      "epoch": 0.55247520307344,
      "grad_norm": 0.17617826163768768,
      "learning_rate": 9.083344520637079e-06,
      "loss": 0.0034,
      "step": 86140
    },
    {
      "epoch": 0.552539339967226,
      "grad_norm": 0.055370114743709564,
      "learning_rate": 9.083021486604062e-06,
      "loss": 0.0048,
      "step": 86150
    },
    {
      "epoch": 0.5526034768610122,
      "grad_norm": 0.05733392387628555,
      "learning_rate": 9.082698401407868e-06,
      "loss": 0.0032,
      "step": 86160
    },
    {
      "epoch": 0.5526676137547982,
      "grad_norm": 0.04752342775464058,
      "learning_rate": 9.082375265052545e-06,
      "loss": 0.0046,
      "step": 86170
    },
    {
      "epoch": 0.5527317506485844,
      "grad_norm": 0.13037846982479095,
      "learning_rate": 9.082052077542142e-06,
      "loss": 0.0029,
      "step": 86180
    },
    {
      "epoch": 0.5527958875423704,
      "grad_norm": 0.06878279894590378,
      "learning_rate": 9.08172883888071e-06,
      "loss": 0.0017,
      "step": 86190
    },
    {
      "epoch": 0.5528600244361566,
      "grad_norm": 0.10957800596952438,
      "learning_rate": 9.081405549072299e-06,
      "loss": 0.0042,
      "step": 86200
    },
    {
      "epoch": 0.5529241613299426,
      "grad_norm": 0.05714105814695358,
      "learning_rate": 9.08108220812096e-06,
      "loss": 0.0067,
      "step": 86210
    },
    {
      "epoch": 0.5529882982237287,
      "grad_norm": 0.16612716019153595,
      "learning_rate": 9.080758816030741e-06,
      "loss": 0.0025,
      "step": 86220
    },
    {
      "epoch": 0.5530524351175148,
      "grad_norm": 0.12985143065452576,
      "learning_rate": 9.0804353728057e-06,
      "loss": 0.0043,
      "step": 86230
    },
    {
      "epoch": 0.5531165720113009,
      "grad_norm": 0.4922974109649658,
      "learning_rate": 9.080111878449887e-06,
      "loss": 0.0034,
      "step": 86240
    },
    {
      "epoch": 0.553180708905087,
      "grad_norm": 0.17751812934875488,
      "learning_rate": 9.079788332967357e-06,
      "loss": 0.0058,
      "step": 86250
    },
    {
      "epoch": 0.5532448457988731,
      "grad_norm": 0.1458687037229538,
      "learning_rate": 9.079464736362162e-06,
      "loss": 0.0031,
      "step": 86260
    },
    {
      "epoch": 0.5533089826926592,
      "grad_norm": 0.13435406982898712,
      "learning_rate": 9.07914108863836e-06,
      "loss": 0.0021,
      "step": 86270
    },
    {
      "epoch": 0.5533731195864453,
      "grad_norm": 0.18028303980827332,
      "learning_rate": 9.078817389800004e-06,
      "loss": 0.005,
      "step": 86280
    },
    {
      "epoch": 0.5534372564802315,
      "grad_norm": 0.059606894850730896,
      "learning_rate": 9.07849363985115e-06,
      "loss": 0.0057,
      "step": 86290
    },
    {
      "epoch": 0.5535013933740175,
      "grad_norm": 0.408214271068573,
      "learning_rate": 9.078169838795858e-06,
      "loss": 0.0023,
      "step": 86300
    },
    {
      "epoch": 0.5535655302678036,
      "grad_norm": 0.15087488293647766,
      "learning_rate": 9.077845986638181e-06,
      "loss": 0.0025,
      "step": 86310
    },
    {
      "epoch": 0.5536296671615897,
      "grad_norm": 0.10014703124761581,
      "learning_rate": 9.07752208338218e-06,
      "loss": 0.0018,
      "step": 86320
    },
    {
      "epoch": 0.5536938040553758,
      "grad_norm": 0.14538487792015076,
      "learning_rate": 9.077198129031915e-06,
      "loss": 0.0036,
      "step": 86330
    },
    {
      "epoch": 0.5537579409491619,
      "grad_norm": 0.02291569486260414,
      "learning_rate": 9.076874123591441e-06,
      "loss": 0.0031,
      "step": 86340
    },
    {
      "epoch": 0.553822077842948,
      "grad_norm": 0.34157443046569824,
      "learning_rate": 9.076550067064822e-06,
      "loss": 0.0035,
      "step": 86350
    },
    {
      "epoch": 0.553886214736734,
      "grad_norm": 0.059290993958711624,
      "learning_rate": 9.076225959456117e-06,
      "loss": 0.0043,
      "step": 86360
    },
    {
      "epoch": 0.5539503516305202,
      "grad_norm": 0.1451360583305359,
      "learning_rate": 9.075901800769387e-06,
      "loss": 0.0014,
      "step": 86370
    },
    {
      "epoch": 0.5540144885243062,
      "grad_norm": 0.08115583658218384,
      "learning_rate": 9.075577591008694e-06,
      "loss": 0.0012,
      "step": 86380
    },
    {
      "epoch": 0.5540786254180924,
      "grad_norm": 0.11323795467615128,
      "learning_rate": 9.075253330178102e-06,
      "loss": 0.004,
      "step": 86390
    },
    {
      "epoch": 0.5541427623118784,
      "grad_norm": 0.1205209419131279,
      "learning_rate": 9.074929018281672e-06,
      "loss": 0.002,
      "step": 86400
    },
    {
      "epoch": 0.5542068992056646,
      "grad_norm": 0.12457386404275894,
      "learning_rate": 9.07460465532347e-06,
      "loss": 0.0026,
      "step": 86410
    },
    {
      "epoch": 0.5542710360994507,
      "grad_norm": 0.20637083053588867,
      "learning_rate": 9.07428024130756e-06,
      "loss": 0.0055,
      "step": 86420
    },
    {
      "epoch": 0.5543351729932368,
      "grad_norm": 0.32814687490463257,
      "learning_rate": 9.073955776238006e-06,
      "loss": 0.0039,
      "step": 86430
    },
    {
      "epoch": 0.5543993098870229,
      "grad_norm": 0.24031226336956024,
      "learning_rate": 9.073631260118875e-06,
      "loss": 0.0033,
      "step": 86440
    },
    {
      "epoch": 0.5544634467808089,
      "grad_norm": 0.1292991042137146,
      "learning_rate": 9.073306692954234e-06,
      "loss": 0.0029,
      "step": 86450
    },
    {
      "epoch": 0.5545275836745951,
      "grad_norm": 0.044169165194034576,
      "learning_rate": 9.072982074748147e-06,
      "loss": 0.0032,
      "step": 86460
    },
    {
      "epoch": 0.5545917205683811,
      "grad_norm": 0.06301355361938477,
      "learning_rate": 9.072657405504685e-06,
      "loss": 0.0031,
      "step": 86470
    },
    {
      "epoch": 0.5546558574621673,
      "grad_norm": 0.04632085561752319,
      "learning_rate": 9.072332685227913e-06,
      "loss": 0.0035,
      "step": 86480
    },
    {
      "epoch": 0.5547199943559533,
      "grad_norm": 0.13859742879867554,
      "learning_rate": 9.072007913921906e-06,
      "loss": 0.0028,
      "step": 86490
    },
    {
      "epoch": 0.5547841312497395,
      "grad_norm": 0.11913814395666122,
      "learning_rate": 9.071683091590726e-06,
      "loss": 0.0037,
      "step": 86500
    },
    {
      "epoch": 0.5548482681435255,
      "grad_norm": 0.26309624314308167,
      "learning_rate": 9.071358218238448e-06,
      "loss": 0.0049,
      "step": 86510
    },
    {
      "epoch": 0.5549124050373117,
      "grad_norm": 0.0424310527741909,
      "learning_rate": 9.071033293869142e-06,
      "loss": 0.0034,
      "step": 86520
    },
    {
      "epoch": 0.5549765419310977,
      "grad_norm": 0.1482691913843155,
      "learning_rate": 9.070708318486878e-06,
      "loss": 0.0027,
      "step": 86530
    },
    {
      "epoch": 0.5550406788248838,
      "grad_norm": 0.15104617178440094,
      "learning_rate": 9.070383292095731e-06,
      "loss": 0.0024,
      "step": 86540
    },
    {
      "epoch": 0.5551048157186699,
      "grad_norm": 0.17228814959526062,
      "learning_rate": 9.070058214699772e-06,
      "loss": 0.0031,
      "step": 86550
    },
    {
      "epoch": 0.555168952612456,
      "grad_norm": 0.11480726301670074,
      "learning_rate": 9.069733086303075e-06,
      "loss": 0.002,
      "step": 86560
    },
    {
      "epoch": 0.5552330895062422,
      "grad_norm": 0.0560038685798645,
      "learning_rate": 9.069407906909712e-06,
      "loss": 0.0035,
      "step": 86570
    },
    {
      "epoch": 0.5552972264000282,
      "grad_norm": 0.14466722309589386,
      "learning_rate": 9.069082676523762e-06,
      "loss": 0.004,
      "step": 86580
    },
    {
      "epoch": 0.5553613632938144,
      "grad_norm": 0.028515275567770004,
      "learning_rate": 9.068757395149297e-06,
      "loss": 0.0049,
      "step": 86590
    },
    {
      "epoch": 0.5554255001876004,
      "grad_norm": 0.07086604088544846,
      "learning_rate": 9.068432062790393e-06,
      "loss": 0.0025,
      "step": 86600
    },
    {
      "epoch": 0.5554896370813865,
      "grad_norm": 0.09528445452451706,
      "learning_rate": 9.06810667945113e-06,
      "loss": 0.0036,
      "step": 86610
    },
    {
      "epoch": 0.5555537739751726,
      "grad_norm": 0.13632024824619293,
      "learning_rate": 9.067781245135581e-06,
      "loss": 0.0036,
      "step": 86620
    },
    {
      "epoch": 0.5556179108689587,
      "grad_norm": 0.16784769296646118,
      "learning_rate": 9.067455759847825e-06,
      "loss": 0.004,
      "step": 86630
    },
    {
      "epoch": 0.5556820477627448,
      "grad_norm": 0.0556255541741848,
      "learning_rate": 9.067130223591943e-06,
      "loss": 0.0026,
      "step": 86640
    },
    {
      "epoch": 0.5557461846565309,
      "grad_norm": 0.2098006159067154,
      "learning_rate": 9.066804636372011e-06,
      "loss": 0.0031,
      "step": 86650
    },
    {
      "epoch": 0.555810321550317,
      "grad_norm": 0.21030020713806152,
      "learning_rate": 9.066478998192113e-06,
      "loss": 0.003,
      "step": 86660
    },
    {
      "epoch": 0.5558744584441031,
      "grad_norm": 0.2861911356449127,
      "learning_rate": 9.066153309056325e-06,
      "loss": 0.0048,
      "step": 86670
    },
    {
      "epoch": 0.5559385953378891,
      "grad_norm": 0.0764365941286087,
      "learning_rate": 9.06582756896873e-06,
      "loss": 0.0029,
      "step": 86680
    },
    {
      "epoch": 0.5560027322316753,
      "grad_norm": 0.10217791795730591,
      "learning_rate": 9.06550177793341e-06,
      "loss": 0.0026,
      "step": 86690
    },
    {
      "epoch": 0.5560668691254614,
      "grad_norm": 0.2130669355392456,
      "learning_rate": 9.065175935954447e-06,
      "loss": 0.0037,
      "step": 86700
    },
    {
      "epoch": 0.5561310060192475,
      "grad_norm": 0.12732025980949402,
      "learning_rate": 9.064850043035925e-06,
      "loss": 0.0029,
      "step": 86710
    },
    {
      "epoch": 0.5561951429130336,
      "grad_norm": 0.2850833237171173,
      "learning_rate": 9.064524099181925e-06,
      "loss": 0.0046,
      "step": 86720
    },
    {
      "epoch": 0.5562592798068197,
      "grad_norm": 0.1203378364443779,
      "learning_rate": 9.064198104396536e-06,
      "loss": 0.0037,
      "step": 86730
    },
    {
      "epoch": 0.5563234167006058,
      "grad_norm": 0.3390064835548401,
      "learning_rate": 9.063872058683838e-06,
      "loss": 0.0033,
      "step": 86740
    },
    {
      "epoch": 0.5563875535943918,
      "grad_norm": 0.09684395045042038,
      "learning_rate": 9.06354596204792e-06,
      "loss": 0.0017,
      "step": 86750
    },
    {
      "epoch": 0.556451690488178,
      "grad_norm": 0.12574532628059387,
      "learning_rate": 9.063219814492866e-06,
      "loss": 0.0025,
      "step": 86760
    },
    {
      "epoch": 0.556515827381964,
      "grad_norm": 0.12113095074892044,
      "learning_rate": 9.062893616022763e-06,
      "loss": 0.0034,
      "step": 86770
    },
    {
      "epoch": 0.5565799642757502,
      "grad_norm": 0.04478220269083977,
      "learning_rate": 9.062567366641701e-06,
      "loss": 0.004,
      "step": 86780
    },
    {
      "epoch": 0.5566441011695362,
      "grad_norm": 0.2971895635128021,
      "learning_rate": 9.062241066353766e-06,
      "loss": 0.0033,
      "step": 86790
    },
    {
      "epoch": 0.5567082380633224,
      "grad_norm": 0.2617679834365845,
      "learning_rate": 9.061914715163047e-06,
      "loss": 0.0036,
      "step": 86800
    },
    {
      "epoch": 0.5567723749571084,
      "grad_norm": 0.09581983834505081,
      "learning_rate": 9.061588313073633e-06,
      "loss": 0.0035,
      "step": 86810
    },
    {
      "epoch": 0.5568365118508946,
      "grad_norm": 0.2524416148662567,
      "learning_rate": 9.061261860089614e-06,
      "loss": 0.0025,
      "step": 86820
    },
    {
      "epoch": 0.5569006487446806,
      "grad_norm": 0.29080846905708313,
      "learning_rate": 9.060935356215084e-06,
      "loss": 0.0054,
      "step": 86830
    },
    {
      "epoch": 0.5569647856384667,
      "grad_norm": 0.3046024441719055,
      "learning_rate": 9.06060880145413e-06,
      "loss": 0.0042,
      "step": 86840
    },
    {
      "epoch": 0.5570289225322529,
      "grad_norm": 0.14829081296920776,
      "learning_rate": 9.060282195810845e-06,
      "loss": 0.0026,
      "step": 86850
    },
    {
      "epoch": 0.5570930594260389,
      "grad_norm": 0.18649496138095856,
      "learning_rate": 9.059955539289323e-06,
      "loss": 0.0015,
      "step": 86860
    },
    {
      "epoch": 0.5571571963198251,
      "grad_norm": 0.23416541516780853,
      "learning_rate": 9.059628831893657e-06,
      "loss": 0.0039,
      "step": 86870
    },
    {
      "epoch": 0.5572213332136111,
      "grad_norm": 0.24305544793605804,
      "learning_rate": 9.05930207362794e-06,
      "loss": 0.0051,
      "step": 86880
    },
    {
      "epoch": 0.5572854701073973,
      "grad_norm": 0.3699105381965637,
      "learning_rate": 9.058975264496266e-06,
      "loss": 0.006,
      "step": 86890
    },
    {
      "epoch": 0.5573496070011833,
      "grad_norm": 0.07164635509252548,
      "learning_rate": 9.058648404502732e-06,
      "loss": 0.0039,
      "step": 86900
    },
    {
      "epoch": 0.5574137438949694,
      "grad_norm": 0.1825900375843048,
      "learning_rate": 9.058321493651432e-06,
      "loss": 0.0018,
      "step": 86910
    },
    {
      "epoch": 0.5574778807887555,
      "grad_norm": 0.292298823595047,
      "learning_rate": 9.057994531946463e-06,
      "loss": 0.004,
      "step": 86920
    },
    {
      "epoch": 0.5575420176825416,
      "grad_norm": 0.14147339761257172,
      "learning_rate": 9.057667519391924e-06,
      "loss": 0.0025,
      "step": 86930
    },
    {
      "epoch": 0.5576061545763277,
      "grad_norm": 0.3002302646636963,
      "learning_rate": 9.057340455991912e-06,
      "loss": 0.0028,
      "step": 86940
    },
    {
      "epoch": 0.5576702914701138,
      "grad_norm": 0.23513545095920563,
      "learning_rate": 9.057013341750522e-06,
      "loss": 0.0045,
      "step": 86950
    },
    {
      "epoch": 0.5577344283638999,
      "grad_norm": 0.10269004851579666,
      "learning_rate": 9.056686176671855e-06,
      "loss": 0.0054,
      "step": 86960
    },
    {
      "epoch": 0.557798565257686,
      "grad_norm": 0.13707150518894196,
      "learning_rate": 9.056358960760013e-06,
      "loss": 0.0047,
      "step": 86970
    },
    {
      "epoch": 0.5578627021514722,
      "grad_norm": 0.18656758964061737,
      "learning_rate": 9.056031694019092e-06,
      "loss": 0.0151,
      "step": 86980
    },
    {
      "epoch": 0.5579268390452582,
      "grad_norm": 0.29598280787467957,
      "learning_rate": 9.055704376453197e-06,
      "loss": 0.0039,
      "step": 86990
    },
    {
      "epoch": 0.5579909759390443,
      "grad_norm": 0.07861374318599701,
      "learning_rate": 9.055377008066425e-06,
      "loss": 0.0036,
      "step": 87000
    },
    {
      "epoch": 0.5580551128328304,
      "grad_norm": 0.24993683397769928,
      "learning_rate": 9.055049588862883e-06,
      "loss": 0.004,
      "step": 87010
    },
    {
      "epoch": 0.5581192497266165,
      "grad_norm": 0.06908705830574036,
      "learning_rate": 9.054722118846671e-06,
      "loss": 0.0022,
      "step": 87020
    },
    {
      "epoch": 0.5581833866204026,
      "grad_norm": 0.062278080731630325,
      "learning_rate": 9.054394598021894e-06,
      "loss": 0.003,
      "step": 87030
    },
    {
      "epoch": 0.5582475235141887,
      "grad_norm": 0.13321839272975922,
      "learning_rate": 9.054067026392654e-06,
      "loss": 0.0026,
      "step": 87040
    },
    {
      "epoch": 0.5583116604079748,
      "grad_norm": 0.12728707492351532,
      "learning_rate": 9.053739403963056e-06,
      "loss": 0.0029,
      "step": 87050
    },
    {
      "epoch": 0.5583757973017609,
      "grad_norm": 0.12646393477916718,
      "learning_rate": 9.053411730737208e-06,
      "loss": 0.0027,
      "step": 87060
    },
    {
      "epoch": 0.5584399341955469,
      "grad_norm": 0.0910581424832344,
      "learning_rate": 9.053084006719212e-06,
      "loss": 0.0054,
      "step": 87070
    },
    {
      "epoch": 0.5585040710893331,
      "grad_norm": 0.0338672511279583,
      "learning_rate": 9.052756231913178e-06,
      "loss": 0.002,
      "step": 87080
    },
    {
      "epoch": 0.5585682079831191,
      "grad_norm": 0.013740995898842812,
      "learning_rate": 9.05242840632321e-06,
      "loss": 0.0026,
      "step": 87090
    },
    {
      "epoch": 0.5586323448769053,
      "grad_norm": 0.05321336165070534,
      "learning_rate": 9.05210052995342e-06,
      "loss": 0.0045,
      "step": 87100
    },
    {
      "epoch": 0.5586964817706913,
      "grad_norm": 0.16705931723117828,
      "learning_rate": 9.051772602807913e-06,
      "loss": 0.0028,
      "step": 87110
    },
    {
      "epoch": 0.5587606186644775,
      "grad_norm": 0.1733585000038147,
      "learning_rate": 9.051444624890801e-06,
      "loss": 0.003,
      "step": 87120
    },
    {
      "epoch": 0.5588247555582636,
      "grad_norm": 0.20235127210617065,
      "learning_rate": 9.051116596206192e-06,
      "loss": 0.0026,
      "step": 87130
    },
    {
      "epoch": 0.5588888924520496,
      "grad_norm": 0.07634595036506653,
      "learning_rate": 9.050788516758196e-06,
      "loss": 0.002,
      "step": 87140
    },
    {
      "epoch": 0.5589530293458358,
      "grad_norm": 0.20534726977348328,
      "learning_rate": 9.050460386550926e-06,
      "loss": 0.0034,
      "step": 87150
    },
    {
      "epoch": 0.5590171662396218,
      "grad_norm": 0.14344623684883118,
      "learning_rate": 9.050132205588492e-06,
      "loss": 0.0027,
      "step": 87160
    },
    {
      "epoch": 0.559081303133408,
      "grad_norm": 0.11710309237241745,
      "learning_rate": 9.049803973875007e-06,
      "loss": 0.0034,
      "step": 87170
    },
    {
      "epoch": 0.559145440027194,
      "grad_norm": 0.1879531890153885,
      "learning_rate": 9.049475691414582e-06,
      "loss": 0.0035,
      "step": 87180
    },
    {
      "epoch": 0.5592095769209802,
      "grad_norm": 0.1346030980348587,
      "learning_rate": 9.049147358211336e-06,
      "loss": 0.0041,
      "step": 87190
    },
    {
      "epoch": 0.5592737138147662,
      "grad_norm": 0.1864037811756134,
      "learning_rate": 9.048818974269378e-06,
      "loss": 0.0035,
      "step": 87200
    },
    {
      "epoch": 0.5593378507085524,
      "grad_norm": 0.08595872670412064,
      "learning_rate": 9.048490539592824e-06,
      "loss": 0.004,
      "step": 87210
    },
    {
      "epoch": 0.5594019876023384,
      "grad_norm": 0.3378288447856903,
      "learning_rate": 9.048162054185789e-06,
      "loss": 0.0057,
      "step": 87220
    },
    {
      "epoch": 0.5594661244961245,
      "grad_norm": 0.017530549317598343,
      "learning_rate": 9.047833518052393e-06,
      "loss": 0.0031,
      "step": 87230
    },
    {
      "epoch": 0.5595302613899106,
      "grad_norm": 0.11569032818078995,
      "learning_rate": 9.047504931196748e-06,
      "loss": 0.0034,
      "step": 87240
    },
    {
      "epoch": 0.5595943982836967,
      "grad_norm": 0.34084203839302063,
      "learning_rate": 9.047176293622975e-06,
      "loss": 0.0052,
      "step": 87250
    },
    {
      "epoch": 0.5596585351774829,
      "grad_norm": 0.12566979229450226,
      "learning_rate": 9.046847605335189e-06,
      "loss": 0.0034,
      "step": 87260
    },
    {
      "epoch": 0.5597226720712689,
      "grad_norm": 0.21158519387245178,
      "learning_rate": 9.04651886633751e-06,
      "loss": 0.0038,
      "step": 87270
    },
    {
      "epoch": 0.5597868089650551,
      "grad_norm": 0.009799067862331867,
      "learning_rate": 9.04619007663406e-06,
      "loss": 0.0045,
      "step": 87280
    },
    {
      "epoch": 0.5598509458588411,
      "grad_norm": 0.20503848791122437,
      "learning_rate": 9.045861236228955e-06,
      "loss": 0.0027,
      "step": 87290
    },
    {
      "epoch": 0.5599150827526272,
      "grad_norm": 0.05585271492600441,
      "learning_rate": 9.045532345126318e-06,
      "loss": 0.0026,
      "step": 87300
    },
    {
      "epoch": 0.5599792196464133,
      "grad_norm": 0.16156123578548431,
      "learning_rate": 9.04520340333027e-06,
      "loss": 0.0029,
      "step": 87310
    },
    {
      "epoch": 0.5600433565401994,
      "grad_norm": 0.07933583110570908,
      "learning_rate": 9.044874410844931e-06,
      "loss": 0.0027,
      "step": 87320
    },
    {
      "epoch": 0.5601074934339855,
      "grad_norm": 0.2687130570411682,
      "learning_rate": 9.044545367674425e-06,
      "loss": 0.0046,
      "step": 87330
    },
    {
      "epoch": 0.5601716303277716,
      "grad_norm": 0.24467165768146515,
      "learning_rate": 9.044216273822875e-06,
      "loss": 0.0032,
      "step": 87340
    },
    {
      "epoch": 0.5602357672215577,
      "grad_norm": 0.1341111958026886,
      "learning_rate": 9.043887129294403e-06,
      "loss": 0.0037,
      "step": 87350
    },
    {
      "epoch": 0.5602999041153438,
      "grad_norm": 0.1187986359000206,
      "learning_rate": 9.043557934093138e-06,
      "loss": 0.0024,
      "step": 87360
    },
    {
      "epoch": 0.5603640410091298,
      "grad_norm": 0.14248231053352356,
      "learning_rate": 9.0432286882232e-06,
      "loss": 0.0029,
      "step": 87370
    },
    {
      "epoch": 0.560428177902916,
      "grad_norm": 0.2501228153705597,
      "learning_rate": 9.04289939168872e-06,
      "loss": 0.0026,
      "step": 87380
    },
    {
      "epoch": 0.560492314796702,
      "grad_norm": 0.20401206612586975,
      "learning_rate": 9.042570044493817e-06,
      "loss": 0.005,
      "step": 87390
    },
    {
      "epoch": 0.5605564516904882,
      "grad_norm": 0.19732318818569183,
      "learning_rate": 9.042240646642627e-06,
      "loss": 0.0035,
      "step": 87400
    },
    {
      "epoch": 0.5606205885842743,
      "grad_norm": 0.17643578350543976,
      "learning_rate": 9.041911198139267e-06,
      "loss": 0.0034,
      "step": 87410
    },
    {
      "epoch": 0.5606847254780604,
      "grad_norm": 0.12217200547456741,
      "learning_rate": 9.041581698987876e-06,
      "loss": 0.0023,
      "step": 87420
    },
    {
      "epoch": 0.5607488623718465,
      "grad_norm": 0.16141095757484436,
      "learning_rate": 9.041252149192575e-06,
      "loss": 0.0039,
      "step": 87430
    },
    {
      "epoch": 0.5608129992656326,
      "grad_norm": 0.2334379255771637,
      "learning_rate": 9.040922548757497e-06,
      "loss": 0.0057,
      "step": 87440
    },
    {
      "epoch": 0.5608771361594187,
      "grad_norm": 0.20374825596809387,
      "learning_rate": 9.040592897686772e-06,
      "loss": 0.0031,
      "step": 87450
    },
    {
      "epoch": 0.5609412730532047,
      "grad_norm": 0.22686593234539032,
      "learning_rate": 9.04026319598453e-06,
      "loss": 0.0053,
      "step": 87460
    },
    {
      "epoch": 0.5610054099469909,
      "grad_norm": 0.02577713504433632,
      "learning_rate": 9.039933443654902e-06,
      "loss": 0.0033,
      "step": 87470
    },
    {
      "epoch": 0.5610695468407769,
      "grad_norm": 0.0653962716460228,
      "learning_rate": 9.039603640702022e-06,
      "loss": 0.0039,
      "step": 87480
    },
    {
      "epoch": 0.5611336837345631,
      "grad_norm": 0.09157300740480423,
      "learning_rate": 9.039273787130021e-06,
      "loss": 0.0055,
      "step": 87490
    },
    {
      "epoch": 0.5611978206283491,
      "grad_norm": 0.23918955028057098,
      "learning_rate": 9.038943882943031e-06,
      "loss": 0.0048,
      "step": 87500
    },
    {
      "epoch": 0.5612619575221353,
      "grad_norm": 0.134276881814003,
      "learning_rate": 9.038613928145187e-06,
      "loss": 0.0033,
      "step": 87510
    },
    {
      "epoch": 0.5613260944159213,
      "grad_norm": 0.044359344989061356,
      "learning_rate": 9.038283922740626e-06,
      "loss": 0.0042,
      "step": 87520
    },
    {
      "epoch": 0.5613902313097074,
      "grad_norm": 0.045472290366888046,
      "learning_rate": 9.037953866733482e-06,
      "loss": 0.0031,
      "step": 87530
    },
    {
      "epoch": 0.5614543682034935,
      "grad_norm": 0.1305588036775589,
      "learning_rate": 9.03762376012789e-06,
      "loss": 0.0053,
      "step": 87540
    },
    {
      "epoch": 0.5615185050972796,
      "grad_norm": 0.142921581864357,
      "learning_rate": 9.037293602927986e-06,
      "loss": 0.0038,
      "step": 87550
    },
    {
      "epoch": 0.5615826419910658,
      "grad_norm": 0.25706934928894043,
      "learning_rate": 9.036963395137907e-06,
      "loss": 0.004,
      "step": 87560
    },
    {
      "epoch": 0.5616467788848518,
      "grad_norm": 0.13787363469600677,
      "learning_rate": 9.036633136761792e-06,
      "loss": 0.0029,
      "step": 87570
    },
    {
      "epoch": 0.561710915778638,
      "grad_norm": 0.21606731414794922,
      "learning_rate": 9.03630282780378e-06,
      "loss": 0.0025,
      "step": 87580
    },
    {
      "epoch": 0.561775052672424,
      "grad_norm": 0.26017582416534424,
      "learning_rate": 9.035972468268007e-06,
      "loss": 0.0051,
      "step": 87590
    },
    {
      "epoch": 0.5618391895662102,
      "grad_norm": 0.32015421986579895,
      "learning_rate": 9.035642058158616e-06,
      "loss": 0.0042,
      "step": 87600
    },
    {
      "epoch": 0.5619033264599962,
      "grad_norm": 0.10264544934034348,
      "learning_rate": 9.035311597479746e-06,
      "loss": 0.0026,
      "step": 87610
    },
    {
      "epoch": 0.5619674633537823,
      "grad_norm": 0.12089379131793976,
      "learning_rate": 9.034981086235535e-06,
      "loss": 0.0026,
      "step": 87620
    },
    {
      "epoch": 0.5620316002475684,
      "grad_norm": 0.06433777511119843,
      "learning_rate": 9.034650524430129e-06,
      "loss": 0.0035,
      "step": 87630
    },
    {
      "epoch": 0.5620957371413545,
      "grad_norm": 0.13378623127937317,
      "learning_rate": 9.034319912067669e-06,
      "loss": 0.0037,
      "step": 87640
    },
    {
      "epoch": 0.5621598740351406,
      "grad_norm": 0.13473980128765106,
      "learning_rate": 9.033989249152297e-06,
      "loss": 0.0022,
      "step": 87650
    },
    {
      "epoch": 0.5622240109289267,
      "grad_norm": 0.13181231915950775,
      "learning_rate": 9.033658535688157e-06,
      "loss": 0.0028,
      "step": 87660
    },
    {
      "epoch": 0.5622881478227127,
      "grad_norm": 0.05946122109889984,
      "learning_rate": 9.033327771679393e-06,
      "loss": 0.0028,
      "step": 87670
    },
    {
      "epoch": 0.5623522847164989,
      "grad_norm": 0.25498464703559875,
      "learning_rate": 9.032996957130146e-06,
      "loss": 0.0062,
      "step": 87680
    },
    {
      "epoch": 0.562416421610285,
      "grad_norm": 0.1765357255935669,
      "learning_rate": 9.03266609204457e-06,
      "loss": 0.0033,
      "step": 87690
    },
    {
      "epoch": 0.5624805585040711,
      "grad_norm": 0.11471286416053772,
      "learning_rate": 9.032335176426801e-06,
      "loss": 0.0026,
      "step": 87700
    },
    {
      "epoch": 0.5625446953978572,
      "grad_norm": 0.436392605304718,
      "learning_rate": 9.032004210280993e-06,
      "loss": 0.0043,
      "step": 87710
    },
    {
      "epoch": 0.5626088322916433,
      "grad_norm": 0.09560250490903854,
      "learning_rate": 9.03167319361129e-06,
      "loss": 0.0018,
      "step": 87720
    },
    {
      "epoch": 0.5626729691854294,
      "grad_norm": 0.1273191273212433,
      "learning_rate": 9.03134212642184e-06,
      "loss": 0.0071,
      "step": 87730
    },
    {
      "epoch": 0.5627371060792155,
      "grad_norm": 0.09865515679121017,
      "learning_rate": 9.031011008716793e-06,
      "loss": 0.0038,
      "step": 87740
    },
    {
      "epoch": 0.5628012429730016,
      "grad_norm": 0.2741306722164154,
      "learning_rate": 9.030679840500298e-06,
      "loss": 0.0044,
      "step": 87750
    },
    {
      "epoch": 0.5628653798667876,
      "grad_norm": 0.10659819841384888,
      "learning_rate": 9.030348621776502e-06,
      "loss": 0.0019,
      "step": 87760
    },
    {
      "epoch": 0.5629295167605738,
      "grad_norm": 0.280533105134964,
      "learning_rate": 9.030017352549558e-06,
      "loss": 0.0046,
      "step": 87770
    },
    {
      "epoch": 0.5629936536543598,
      "grad_norm": 0.06319417804479599,
      "learning_rate": 9.029686032823615e-06,
      "loss": 0.0083,
      "step": 87780
    },
    {
      "epoch": 0.563057790548146,
      "grad_norm": 0.13399365544319153,
      "learning_rate": 9.029354662602829e-06,
      "loss": 0.0029,
      "step": 87790
    },
    {
      "epoch": 0.563121927441932,
      "grad_norm": 0.2020081877708435,
      "learning_rate": 9.029023241891347e-06,
      "loss": 0.0028,
      "step": 87800
    },
    {
      "epoch": 0.5631860643357182,
      "grad_norm": 0.4207175672054291,
      "learning_rate": 9.028691770693324e-06,
      "loss": 0.0039,
      "step": 87810
    },
    {
      "epoch": 0.5632502012295042,
      "grad_norm": 0.39699438214302063,
      "learning_rate": 9.028360249012915e-06,
      "loss": 0.0041,
      "step": 87820
    },
    {
      "epoch": 0.5633143381232903,
      "grad_norm": 0.02185790054500103,
      "learning_rate": 9.028028676854274e-06,
      "loss": 0.0044,
      "step": 87830
    },
    {
      "epoch": 0.5633784750170765,
      "grad_norm": 0.0680369883775711,
      "learning_rate": 9.027697054221554e-06,
      "loss": 0.0038,
      "step": 87840
    },
    {
      "epoch": 0.5634426119108625,
      "grad_norm": 0.1581745594739914,
      "learning_rate": 9.02736538111891e-06,
      "loss": 0.0036,
      "step": 87850
    },
    {
      "epoch": 0.5635067488046487,
      "grad_norm": 0.4550938606262207,
      "learning_rate": 9.027033657550502e-06,
      "loss": 0.0087,
      "step": 87860
    },
    {
      "epoch": 0.5635708856984347,
      "grad_norm": 0.055353473871946335,
      "learning_rate": 9.026701883520481e-06,
      "loss": 0.0031,
      "step": 87870
    },
    {
      "epoch": 0.5636350225922209,
      "grad_norm": 0.16755340993404388,
      "learning_rate": 9.02637005903301e-06,
      "loss": 0.0035,
      "step": 87880
    },
    {
      "epoch": 0.5636991594860069,
      "grad_norm": 0.16791850328445435,
      "learning_rate": 9.026038184092244e-06,
      "loss": 0.0024,
      "step": 87890
    },
    {
      "epoch": 0.563763296379793,
      "grad_norm": 0.13935136795043945,
      "learning_rate": 9.025706258702343e-06,
      "loss": 0.0029,
      "step": 87900
    },
    {
      "epoch": 0.5638274332735791,
      "grad_norm": 0.22306275367736816,
      "learning_rate": 9.025374282867465e-06,
      "loss": 0.0035,
      "step": 87910
    },
    {
      "epoch": 0.5638915701673652,
      "grad_norm": 0.09611920267343521,
      "learning_rate": 9.02504225659177e-06,
      "loss": 0.0017,
      "step": 87920
    },
    {
      "epoch": 0.5639557070611513,
      "grad_norm": 0.2897023856639862,
      "learning_rate": 9.024710179879417e-06,
      "loss": 0.0036,
      "step": 87930
    },
    {
      "epoch": 0.5640198439549374,
      "grad_norm": 0.21228285133838654,
      "learning_rate": 9.02437805273457e-06,
      "loss": 0.0033,
      "step": 87940
    },
    {
      "epoch": 0.5640839808487235,
      "grad_norm": 0.01964620314538479,
      "learning_rate": 9.024045875161392e-06,
      "loss": 0.0023,
      "step": 87950
    },
    {
      "epoch": 0.5641481177425096,
      "grad_norm": 0.8125170469284058,
      "learning_rate": 9.023713647164041e-06,
      "loss": 0.0038,
      "step": 87960
    },
    {
      "epoch": 0.5642122546362958,
      "grad_norm": 0.09887786954641342,
      "learning_rate": 9.023381368746685e-06,
      "loss": 0.0053,
      "step": 87970
    },
    {
      "epoch": 0.5642763915300818,
      "grad_norm": 0.05003981292247772,
      "learning_rate": 9.023049039913482e-06,
      "loss": 0.0014,
      "step": 87980
    },
    {
      "epoch": 0.564340528423868,
      "grad_norm": 0.2274905890226364,
      "learning_rate": 9.0227166606686e-06,
      "loss": 0.0042,
      "step": 87990
    },
    {
      "epoch": 0.564404665317654,
      "grad_norm": 0.1140158474445343,
      "learning_rate": 9.022384231016203e-06,
      "loss": 0.0033,
      "step": 88000
    },
    {
      "epoch": 0.5644688022114401,
      "grad_norm": 0.16795748472213745,
      "learning_rate": 9.022051750960458e-06,
      "loss": 0.0029,
      "step": 88010
    },
    {
      "epoch": 0.5645329391052262,
      "grad_norm": 0.18950389325618744,
      "learning_rate": 9.02171922050553e-06,
      "loss": 0.0043,
      "step": 88020
    },
    {
      "epoch": 0.5645970759990123,
      "grad_norm": 0.14368632435798645,
      "learning_rate": 9.021386639655585e-06,
      "loss": 0.0053,
      "step": 88030
    },
    {
      "epoch": 0.5646612128927984,
      "grad_norm": 0.18149404227733612,
      "learning_rate": 9.021054008414791e-06,
      "loss": 0.0037,
      "step": 88040
    },
    {
      "epoch": 0.5647253497865845,
      "grad_norm": 0.0776793360710144,
      "learning_rate": 9.02072132678732e-06,
      "loss": 0.0035,
      "step": 88050
    },
    {
      "epoch": 0.5647894866803705,
      "grad_norm": 0.08136197924613953,
      "learning_rate": 9.020388594777333e-06,
      "loss": 0.0036,
      "step": 88060
    },
    {
      "epoch": 0.5648536235741567,
      "grad_norm": 0.17837202548980713,
      "learning_rate": 9.020055812389005e-06,
      "loss": 0.0066,
      "step": 88070
    },
    {
      "epoch": 0.5649177604679427,
      "grad_norm": 0.13673412799835205,
      "learning_rate": 9.019722979626504e-06,
      "loss": 0.0026,
      "step": 88080
    },
    {
      "epoch": 0.5649818973617289,
      "grad_norm": 0.13098108768463135,
      "learning_rate": 9.019390096494003e-06,
      "loss": 0.003,
      "step": 88090
    },
    {
      "epoch": 0.5650460342555149,
      "grad_norm": 0.1096213161945343,
      "learning_rate": 9.01905716299567e-06,
      "loss": 0.0026,
      "step": 88100
    },
    {
      "epoch": 0.5651101711493011,
      "grad_norm": 0.06392141431570053,
      "learning_rate": 9.018724179135679e-06,
      "loss": 0.0038,
      "step": 88110
    },
    {
      "epoch": 0.5651743080430872,
      "grad_norm": 0.20007091760635376,
      "learning_rate": 9.018391144918201e-06,
      "loss": 0.005,
      "step": 88120
    },
    {
      "epoch": 0.5652384449368733,
      "grad_norm": 0.10015270113945007,
      "learning_rate": 9.01805806034741e-06,
      "loss": 0.0023,
      "step": 88130
    },
    {
      "epoch": 0.5653025818306594,
      "grad_norm": 0.12199140340089798,
      "learning_rate": 9.017724925427482e-06,
      "loss": 0.0021,
      "step": 88140
    },
    {
      "epoch": 0.5653667187244454,
      "grad_norm": 0.12377575039863586,
      "learning_rate": 9.017391740162587e-06,
      "loss": 0.0018,
      "step": 88150
    },
    {
      "epoch": 0.5654308556182316,
      "grad_norm": 0.35765019059181213,
      "learning_rate": 9.017058504556904e-06,
      "loss": 0.0026,
      "step": 88160
    },
    {
      "epoch": 0.5654949925120176,
      "grad_norm": 0.12522195279598236,
      "learning_rate": 9.016725218614606e-06,
      "loss": 0.0028,
      "step": 88170
    },
    {
      "epoch": 0.5655591294058038,
      "grad_norm": 0.12536902725696564,
      "learning_rate": 9.01639188233987e-06,
      "loss": 0.0065,
      "step": 88180
    },
    {
      "epoch": 0.5656232662995898,
      "grad_norm": 0.187701016664505,
      "learning_rate": 9.016058495736873e-06,
      "loss": 0.0038,
      "step": 88190
    },
    {
      "epoch": 0.565687403193376,
      "grad_norm": 0.09773750603199005,
      "learning_rate": 9.015725058809793e-06,
      "loss": 0.0019,
      "step": 88200
    },
    {
      "epoch": 0.565751540087162,
      "grad_norm": 0.011167613789439201,
      "learning_rate": 9.015391571562809e-06,
      "loss": 0.0023,
      "step": 88210
    },
    {
      "epoch": 0.5658156769809481,
      "grad_norm": 0.16488158702850342,
      "learning_rate": 9.015058034000097e-06,
      "loss": 0.0024,
      "step": 88220
    },
    {
      "epoch": 0.5658798138747342,
      "grad_norm": 0.19378674030303955,
      "learning_rate": 9.01472444612584e-06,
      "loss": 0.003,
      "step": 88230
    },
    {
      "epoch": 0.5659439507685203,
      "grad_norm": 0.11837329715490341,
      "learning_rate": 9.014390807944217e-06,
      "loss": 0.0038,
      "step": 88240
    },
    {
      "epoch": 0.5660080876623065,
      "grad_norm": 0.05158904567360878,
      "learning_rate": 9.014057119459408e-06,
      "loss": 0.0036,
      "step": 88250
    },
    {
      "epoch": 0.5660722245560925,
      "grad_norm": 0.08860959112644196,
      "learning_rate": 9.013723380675591e-06,
      "loss": 0.0021,
      "step": 88260
    },
    {
      "epoch": 0.5661363614498787,
      "grad_norm": 0.1362515538930893,
      "learning_rate": 9.013389591596954e-06,
      "loss": 0.0034,
      "step": 88270
    },
    {
      "epoch": 0.5662004983436647,
      "grad_norm": 0.21222984790802002,
      "learning_rate": 9.013055752227679e-06,
      "loss": 0.0039,
      "step": 88280
    },
    {
      "epoch": 0.5662646352374509,
      "grad_norm": 0.10679440200328827,
      "learning_rate": 9.012721862571945e-06,
      "loss": 0.0024,
      "step": 88290
    },
    {
      "epoch": 0.5663287721312369,
      "grad_norm": 0.2545750141143799,
      "learning_rate": 9.012387922633938e-06,
      "loss": 0.0038,
      "step": 88300
    },
    {
      "epoch": 0.566392909025023,
      "grad_norm": 0.2821141183376312,
      "learning_rate": 9.012053932417844e-06,
      "loss": 0.0029,
      "step": 88310
    },
    {
      "epoch": 0.5664570459188091,
      "grad_norm": 0.16089150309562683,
      "learning_rate": 9.011719891927846e-06,
      "loss": 0.0035,
      "step": 88320
    },
    {
      "epoch": 0.5665211828125952,
      "grad_norm": 0.10448617488145828,
      "learning_rate": 9.011385801168131e-06,
      "loss": 0.0026,
      "step": 88330
    },
    {
      "epoch": 0.5665853197063813,
      "grad_norm": 0.13759052753448486,
      "learning_rate": 9.011051660142885e-06,
      "loss": 0.0022,
      "step": 88340
    },
    {
      "epoch": 0.5666494566001674,
      "grad_norm": 0.3571251928806305,
      "learning_rate": 9.010717468856295e-06,
      "loss": 0.0029,
      "step": 88350
    },
    {
      "epoch": 0.5667135934939534,
      "grad_norm": 0.08513778448104858,
      "learning_rate": 9.01038322731255e-06,
      "loss": 0.0032,
      "step": 88360
    },
    {
      "epoch": 0.5667777303877396,
      "grad_norm": 0.06159251555800438,
      "learning_rate": 9.010048935515835e-06,
      "loss": 0.0036,
      "step": 88370
    },
    {
      "epoch": 0.5668418672815256,
      "grad_norm": 0.06133711710572243,
      "learning_rate": 9.00971459347034e-06,
      "loss": 0.0026,
      "step": 88380
    },
    {
      "epoch": 0.5669060041753118,
      "grad_norm": 0.3495323061943054,
      "learning_rate": 9.009380201180258e-06,
      "loss": 0.0045,
      "step": 88390
    },
    {
      "epoch": 0.5669701410690979,
      "grad_norm": 0.13650336861610413,
      "learning_rate": 9.009045758649777e-06,
      "loss": 0.0037,
      "step": 88400
    },
    {
      "epoch": 0.567034277962884,
      "grad_norm": 0.2663722038269043,
      "learning_rate": 9.008711265883086e-06,
      "loss": 0.0042,
      "step": 88410
    },
    {
      "epoch": 0.5670984148566701,
      "grad_norm": 0.30290576815605164,
      "learning_rate": 9.008376722884376e-06,
      "loss": 0.0052,
      "step": 88420
    },
    {
      "epoch": 0.5671625517504562,
      "grad_norm": 0.07226832211017609,
      "learning_rate": 9.008042129657843e-06,
      "loss": 0.0026,
      "step": 88430
    },
    {
      "epoch": 0.5672266886442423,
      "grad_norm": 0.3313843905925751,
      "learning_rate": 9.007707486207677e-06,
      "loss": 0.0032,
      "step": 88440
    },
    {
      "epoch": 0.5672908255380283,
      "grad_norm": 0.1292109340429306,
      "learning_rate": 9.007372792538074e-06,
      "loss": 0.0033,
      "step": 88450
    },
    {
      "epoch": 0.5673549624318145,
      "grad_norm": 0.08601272851228714,
      "learning_rate": 9.007038048653224e-06,
      "loss": 0.0029,
      "step": 88460
    },
    {
      "epoch": 0.5674190993256005,
      "grad_norm": 0.09319724887609482,
      "learning_rate": 9.006703254557324e-06,
      "loss": 0.0014,
      "step": 88470
    },
    {
      "epoch": 0.5674832362193867,
      "grad_norm": 0.04385272040963173,
      "learning_rate": 9.006368410254568e-06,
      "loss": 0.0031,
      "step": 88480
    },
    {
      "epoch": 0.5675473731131727,
      "grad_norm": 0.16889996826648712,
      "learning_rate": 9.006033515749153e-06,
      "loss": 0.0026,
      "step": 88490
    },
    {
      "epoch": 0.5676115100069589,
      "grad_norm": 0.02448371797800064,
      "learning_rate": 9.005698571045273e-06,
      "loss": 0.0031,
      "step": 88500
    },
    {
      "epoch": 0.5676756469007449,
      "grad_norm": 0.15226224064826965,
      "learning_rate": 9.00536357614713e-06,
      "loss": 0.0036,
      "step": 88510
    },
    {
      "epoch": 0.567739783794531,
      "grad_norm": 0.14777974784374237,
      "learning_rate": 9.005028531058918e-06,
      "loss": 0.0039,
      "step": 88520
    },
    {
      "epoch": 0.5678039206883172,
      "grad_norm": 0.209447979927063,
      "learning_rate": 9.004693435784837e-06,
      "loss": 0.0034,
      "step": 88530
    },
    {
      "epoch": 0.5678680575821032,
      "grad_norm": 0.14773207902908325,
      "learning_rate": 9.004358290329082e-06,
      "loss": 0.0036,
      "step": 88540
    },
    {
      "epoch": 0.5679321944758894,
      "grad_norm": 0.0850907489657402,
      "learning_rate": 9.00402309469586e-06,
      "loss": 0.004,
      "step": 88550
    },
    {
      "epoch": 0.5679963313696754,
      "grad_norm": 0.10000663995742798,
      "learning_rate": 9.003687848889365e-06,
      "loss": 0.0033,
      "step": 88560
    },
    {
      "epoch": 0.5680604682634616,
      "grad_norm": 0.131527841091156,
      "learning_rate": 9.003352552913799e-06,
      "loss": 0.0026,
      "step": 88570
    },
    {
      "epoch": 0.5681246051572476,
      "grad_norm": 0.27508410811424255,
      "learning_rate": 9.003017206773365e-06,
      "loss": 0.004,
      "step": 88580
    },
    {
      "epoch": 0.5681887420510338,
      "grad_norm": 0.09187949448823929,
      "learning_rate": 9.002681810472265e-06,
      "loss": 0.003,
      "step": 88590
    },
    {
      "epoch": 0.5682528789448198,
      "grad_norm": 0.2346142679452896,
      "learning_rate": 9.0023463640147e-06,
      "loss": 0.0067,
      "step": 88600
    },
    {
      "epoch": 0.5683170158386059,
      "grad_norm": 0.09984955191612244,
      "learning_rate": 9.002010867404876e-06,
      "loss": 0.0019,
      "step": 88610
    },
    {
      "epoch": 0.568381152732392,
      "grad_norm": 0.10223357379436493,
      "learning_rate": 9.001675320646994e-06,
      "loss": 0.002,
      "step": 88620
    },
    {
      "epoch": 0.5684452896261781,
      "grad_norm": 0.14836329221725464,
      "learning_rate": 9.001339723745262e-06,
      "loss": 0.0029,
      "step": 88630
    },
    {
      "epoch": 0.5685094265199642,
      "grad_norm": 0.11841240525245667,
      "learning_rate": 9.001004076703883e-06,
      "loss": 0.0033,
      "step": 88640
    },
    {
      "epoch": 0.5685735634137503,
      "grad_norm": 0.24895945191383362,
      "learning_rate": 9.000668379527062e-06,
      "loss": 0.0026,
      "step": 88650
    },
    {
      "epoch": 0.5686377003075364,
      "grad_norm": 0.05293124541640282,
      "learning_rate": 9.000332632219009e-06,
      "loss": 0.0022,
      "step": 88660
    },
    {
      "epoch": 0.5687018372013225,
      "grad_norm": 0.16243022680282593,
      "learning_rate": 8.999996834783929e-06,
      "loss": 0.0037,
      "step": 88670
    },
    {
      "epoch": 0.5687659740951087,
      "grad_norm": 0.06476463377475739,
      "learning_rate": 8.99966098722603e-06,
      "loss": 0.0035,
      "step": 88680
    },
    {
      "epoch": 0.5688301109888947,
      "grad_norm": 0.12459227442741394,
      "learning_rate": 8.999325089549518e-06,
      "loss": 0.0036,
      "step": 88690
    },
    {
      "epoch": 0.5688942478826808,
      "grad_norm": 0.1006166934967041,
      "learning_rate": 8.998989141758607e-06,
      "loss": 0.0079,
      "step": 88700
    },
    {
      "epoch": 0.5689583847764669,
      "grad_norm": 0.07273389399051666,
      "learning_rate": 8.998653143857501e-06,
      "loss": 0.0029,
      "step": 88710
    },
    {
      "epoch": 0.569022521670253,
      "grad_norm": 0.0876203402876854,
      "learning_rate": 8.998317095850416e-06,
      "loss": 0.0027,
      "step": 88720
    },
    {
      "epoch": 0.5690866585640391,
      "grad_norm": 0.06741435080766678,
      "learning_rate": 8.99798099774156e-06,
      "loss": 0.0041,
      "step": 88730
    },
    {
      "epoch": 0.5691507954578252,
      "grad_norm": 0.12473282963037491,
      "learning_rate": 8.997644849535144e-06,
      "loss": 0.0022,
      "step": 88740
    },
    {
      "epoch": 0.5692149323516112,
      "grad_norm": 0.13442359864711761,
      "learning_rate": 8.99730865123538e-06,
      "loss": 0.0058,
      "step": 88750
    },
    {
      "epoch": 0.5692790692453974,
      "grad_norm": 0.1543642282485962,
      "learning_rate": 8.996972402846485e-06,
      "loss": 0.0032,
      "step": 88760
    },
    {
      "epoch": 0.5693432061391834,
      "grad_norm": 0.07570601254701614,
      "learning_rate": 8.996636104372667e-06,
      "loss": 0.0027,
      "step": 88770
    },
    {
      "epoch": 0.5694073430329696,
      "grad_norm": 0.20600873231887817,
      "learning_rate": 8.996299755818142e-06,
      "loss": 0.0036,
      "step": 88780
    },
    {
      "epoch": 0.5694714799267556,
      "grad_norm": 0.09505226463079453,
      "learning_rate": 8.995963357187125e-06,
      "loss": 0.0028,
      "step": 88790
    },
    {
      "epoch": 0.5695356168205418,
      "grad_norm": 0.10055418312549591,
      "learning_rate": 8.995626908483833e-06,
      "loss": 0.003,
      "step": 88800
    },
    {
      "epoch": 0.5695997537143279,
      "grad_norm": 0.08040245622396469,
      "learning_rate": 8.995290409712479e-06,
      "loss": 0.006,
      "step": 88810
    },
    {
      "epoch": 0.569663890608114,
      "grad_norm": 0.3970910608768463,
      "learning_rate": 8.99495386087728e-06,
      "loss": 0.0023,
      "step": 88820
    },
    {
      "epoch": 0.5697280275019001,
      "grad_norm": 0.0972469300031662,
      "learning_rate": 8.994617261982454e-06,
      "loss": 0.0026,
      "step": 88830
    },
    {
      "epoch": 0.5697921643956861,
      "grad_norm": 0.17260612547397614,
      "learning_rate": 8.99428061303222e-06,
      "loss": 0.0042,
      "step": 88840
    },
    {
      "epoch": 0.5698563012894723,
      "grad_norm": 0.0984884649515152,
      "learning_rate": 8.993943914030793e-06,
      "loss": 0.0034,
      "step": 88850
    },
    {
      "epoch": 0.5699204381832583,
      "grad_norm": 0.2577366232872009,
      "learning_rate": 8.993607164982398e-06,
      "loss": 0.0031,
      "step": 88860
    },
    {
      "epoch": 0.5699845750770445,
      "grad_norm": 0.0792558565735817,
      "learning_rate": 8.993270365891249e-06,
      "loss": 0.002,
      "step": 88870
    },
    {
      "epoch": 0.5700487119708305,
      "grad_norm": 0.05996266379952431,
      "learning_rate": 8.992933516761569e-06,
      "loss": 0.004,
      "step": 88880
    },
    {
      "epoch": 0.5701128488646167,
      "grad_norm": 0.21832101047039032,
      "learning_rate": 8.992596617597577e-06,
      "loss": 0.0038,
      "step": 88890
    },
    {
      "epoch": 0.5701769857584027,
      "grad_norm": 0.18120752274990082,
      "learning_rate": 8.992259668403495e-06,
      "loss": 0.0027,
      "step": 88900
    },
    {
      "epoch": 0.5702411226521888,
      "grad_norm": 0.01756487786769867,
      "learning_rate": 8.99192266918355e-06,
      "loss": 0.0032,
      "step": 88910
    },
    {
      "epoch": 0.5703052595459749,
      "grad_norm": 0.2363017350435257,
      "learning_rate": 8.991585619941958e-06,
      "loss": 0.0067,
      "step": 88920
    },
    {
      "epoch": 0.570369396439761,
      "grad_norm": 0.27013689279556274,
      "learning_rate": 8.991248520682948e-06,
      "loss": 0.0029,
      "step": 88930
    },
    {
      "epoch": 0.5704335333335471,
      "grad_norm": 0.031348928809165955,
      "learning_rate": 8.990911371410738e-06,
      "loss": 0.003,
      "step": 88940
    },
    {
      "epoch": 0.5704976702273332,
      "grad_norm": 0.07888729870319366,
      "learning_rate": 8.990574172129559e-06,
      "loss": 0.0029,
      "step": 88950
    },
    {
      "epoch": 0.5705618071211194,
      "grad_norm": 0.16150203347206116,
      "learning_rate": 8.990236922843632e-06,
      "loss": 0.0031,
      "step": 88960
    },
    {
      "epoch": 0.5706259440149054,
      "grad_norm": 0.26639118790626526,
      "learning_rate": 8.989899623557184e-06,
      "loss": 0.0025,
      "step": 88970
    },
    {
      "epoch": 0.5706900809086916,
      "grad_norm": 0.13787950575351715,
      "learning_rate": 8.989562274274444e-06,
      "loss": 0.0039,
      "step": 88980
    },
    {
      "epoch": 0.5707542178024776,
      "grad_norm": 0.11798261106014252,
      "learning_rate": 8.989224874999637e-06,
      "loss": 0.0056,
      "step": 88990
    },
    {
      "epoch": 0.5708183546962637,
      "grad_norm": 0.20718160271644592,
      "learning_rate": 8.98888742573699e-06,
      "loss": 0.005,
      "step": 89000
    },
    {
      "epoch": 0.5708824915900498,
      "grad_norm": 0.23089097440242767,
      "learning_rate": 8.988549926490736e-06,
      "loss": 0.0046,
      "step": 89010
    },
    {
      "epoch": 0.5709466284838359,
      "grad_norm": 0.02872941456735134,
      "learning_rate": 8.988212377265098e-06,
      "loss": 0.0023,
      "step": 89020
    },
    {
      "epoch": 0.571010765377622,
      "grad_norm": 0.10600224137306213,
      "learning_rate": 8.987874778064309e-06,
      "loss": 0.0041,
      "step": 89030
    },
    {
      "epoch": 0.5710749022714081,
      "grad_norm": 0.32066240906715393,
      "learning_rate": 8.987537128892598e-06,
      "loss": 0.0051,
      "step": 89040
    },
    {
      "epoch": 0.5711390391651942,
      "grad_norm": 0.05352598428726196,
      "learning_rate": 8.987199429754199e-06,
      "loss": 0.0053,
      "step": 89050
    },
    {
      "epoch": 0.5712031760589803,
      "grad_norm": 0.05201287567615509,
      "learning_rate": 8.98686168065334e-06,
      "loss": 0.0047,
      "step": 89060
    },
    {
      "epoch": 0.5712673129527663,
      "grad_norm": 0.18884700536727905,
      "learning_rate": 8.986523881594255e-06,
      "loss": 0.0023,
      "step": 89070
    },
    {
      "epoch": 0.5713314498465525,
      "grad_norm": 0.2029830664396286,
      "learning_rate": 8.986186032581177e-06,
      "loss": 0.0045,
      "step": 89080
    },
    {
      "epoch": 0.5713955867403385,
      "grad_norm": 0.043966054916381836,
      "learning_rate": 8.98584813361834e-06,
      "loss": 0.0051,
      "step": 89090
    },
    {
      "epoch": 0.5714597236341247,
      "grad_norm": 0.053840771317481995,
      "learning_rate": 8.985510184709976e-06,
      "loss": 0.003,
      "step": 89100
    },
    {
      "epoch": 0.5715238605279108,
      "grad_norm": 0.08549485355615616,
      "learning_rate": 8.985172185860321e-06,
      "loss": 0.0051,
      "step": 89110
    },
    {
      "epoch": 0.5715879974216969,
      "grad_norm": 0.15409742295742035,
      "learning_rate": 8.984834137073611e-06,
      "loss": 0.003,
      "step": 89120
    },
    {
      "epoch": 0.571652134315483,
      "grad_norm": 0.09277409315109253,
      "learning_rate": 8.984496038354081e-06,
      "loss": 0.003,
      "step": 89130
    },
    {
      "epoch": 0.571716271209269,
      "grad_norm": 0.16708652675151825,
      "learning_rate": 8.984157889705968e-06,
      "loss": 0.003,
      "step": 89140
    },
    {
      "epoch": 0.5717804081030552,
      "grad_norm": 0.09447616338729858,
      "learning_rate": 8.983819691133508e-06,
      "loss": 0.003,
      "step": 89150
    },
    {
      "epoch": 0.5718445449968412,
      "grad_norm": 0.03978119418025017,
      "learning_rate": 8.983481442640942e-06,
      "loss": 0.0033,
      "step": 89160
    },
    {
      "epoch": 0.5719086818906274,
      "grad_norm": 0.03759925439953804,
      "learning_rate": 8.983143144232507e-06,
      "loss": 0.0029,
      "step": 89170
    },
    {
      "epoch": 0.5719728187844134,
      "grad_norm": 0.06725975126028061,
      "learning_rate": 8.98280479591244e-06,
      "loss": 0.0037,
      "step": 89180
    },
    {
      "epoch": 0.5720369556781996,
      "grad_norm": 0.15373003482818604,
      "learning_rate": 8.982466397684984e-06,
      "loss": 0.0021,
      "step": 89190
    },
    {
      "epoch": 0.5721010925719856,
      "grad_norm": 0.18197908997535706,
      "learning_rate": 8.982127949554375e-06,
      "loss": 0.0021,
      "step": 89200
    },
    {
      "epoch": 0.5721652294657718,
      "grad_norm": 0.004371248185634613,
      "learning_rate": 8.981789451524859e-06,
      "loss": 0.0022,
      "step": 89210
    },
    {
      "epoch": 0.5722293663595578,
      "grad_norm": 0.10552152991294861,
      "learning_rate": 8.981450903600674e-06,
      "loss": 0.0023,
      "step": 89220
    },
    {
      "epoch": 0.5722935032533439,
      "grad_norm": 0.23676060140132904,
      "learning_rate": 8.981112305786066e-06,
      "loss": 0.0044,
      "step": 89230
    },
    {
      "epoch": 0.5723576401471301,
      "grad_norm": 0.21808232367038727,
      "learning_rate": 8.980773658085274e-06,
      "loss": 0.0042,
      "step": 89240
    },
    {
      "epoch": 0.5724217770409161,
      "grad_norm": 0.11256436258554459,
      "learning_rate": 8.980434960502544e-06,
      "loss": 0.0023,
      "step": 89250
    },
    {
      "epoch": 0.5724859139347023,
      "grad_norm": 0.18191315233707428,
      "learning_rate": 8.980096213042118e-06,
      "loss": 0.0027,
      "step": 89260
    },
    {
      "epoch": 0.5725500508284883,
      "grad_norm": 0.24376623332500458,
      "learning_rate": 8.979757415708243e-06,
      "loss": 0.0036,
      "step": 89270
    },
    {
      "epoch": 0.5726141877222745,
      "grad_norm": 0.17087090015411377,
      "learning_rate": 8.979418568505163e-06,
      "loss": 0.0028,
      "step": 89280
    },
    {
      "epoch": 0.5726783246160605,
      "grad_norm": 0.6307005286216736,
      "learning_rate": 8.979079671437123e-06,
      "loss": 0.0163,
      "step": 89290
    },
    {
      "epoch": 0.5727424615098466,
      "grad_norm": 0.005229447968304157,
      "learning_rate": 8.978740724508373e-06,
      "loss": 0.0024,
      "step": 89300
    },
    {
      "epoch": 0.5728065984036327,
      "grad_norm": 0.17592014372348785,
      "learning_rate": 8.978401727723156e-06,
      "loss": 0.0036,
      "step": 89310
    },
    {
      "epoch": 0.5728707352974188,
      "grad_norm": 0.24231776595115662,
      "learning_rate": 8.978062681085724e-06,
      "loss": 0.003,
      "step": 89320
    },
    {
      "epoch": 0.5729348721912049,
      "grad_norm": 0.18802590668201447,
      "learning_rate": 8.977723584600322e-06,
      "loss": 0.0038,
      "step": 89330
    },
    {
      "epoch": 0.572999009084991,
      "grad_norm": 0.12922777235507965,
      "learning_rate": 8.977384438271202e-06,
      "loss": 0.0034,
      "step": 89340
    },
    {
      "epoch": 0.573063145978777,
      "grad_norm": 0.07981414347887039,
      "learning_rate": 8.977045242102611e-06,
      "loss": 0.0025,
      "step": 89350
    },
    {
      "epoch": 0.5731272828725632,
      "grad_norm": 0.04034131392836571,
      "learning_rate": 8.976705996098801e-06,
      "loss": 0.0025,
      "step": 89360
    },
    {
      "epoch": 0.5731914197663492,
      "grad_norm": 0.07132384926080704,
      "learning_rate": 8.976366700264023e-06,
      "loss": 0.0025,
      "step": 89370
    },
    {
      "epoch": 0.5732555566601354,
      "grad_norm": 0.04554812237620354,
      "learning_rate": 8.97602735460253e-06,
      "loss": 0.002,
      "step": 89380
    },
    {
      "epoch": 0.5733196935539215,
      "grad_norm": 0.15574468672275543,
      "learning_rate": 8.975687959118571e-06,
      "loss": 0.0033,
      "step": 89390
    },
    {
      "epoch": 0.5733838304477076,
      "grad_norm": 0.05257434397935867,
      "learning_rate": 8.9753485138164e-06,
      "loss": 0.0057,
      "step": 89400
    },
    {
      "epoch": 0.5734479673414937,
      "grad_norm": 0.356705904006958,
      "learning_rate": 8.975009018700271e-06,
      "loss": 0.004,
      "step": 89410
    },
    {
      "epoch": 0.5735121042352798,
      "grad_norm": 0.1239226907491684,
      "learning_rate": 8.97466947377444e-06,
      "loss": 0.0029,
      "step": 89420
    },
    {
      "epoch": 0.5735762411290659,
      "grad_norm": 0.08364835381507874,
      "learning_rate": 8.974329879043158e-06,
      "loss": 0.0031,
      "step": 89430
    },
    {
      "epoch": 0.573640378022852,
      "grad_norm": 0.24030348658561707,
      "learning_rate": 8.973990234510684e-06,
      "loss": 0.0031,
      "step": 89440
    },
    {
      "epoch": 0.5737045149166381,
      "grad_norm": 0.1440877467393875,
      "learning_rate": 8.973650540181271e-06,
      "loss": 0.0027,
      "step": 89450
    },
    {
      "epoch": 0.5737686518104241,
      "grad_norm": 0.07367430627346039,
      "learning_rate": 8.973310796059175e-06,
      "loss": 0.0047,
      "step": 89460
    },
    {
      "epoch": 0.5738327887042103,
      "grad_norm": 0.1588653326034546,
      "learning_rate": 8.972971002148658e-06,
      "loss": 0.0034,
      "step": 89470
    },
    {
      "epoch": 0.5738969255979963,
      "grad_norm": 0.2074684500694275,
      "learning_rate": 8.972631158453973e-06,
      "loss": 0.0032,
      "step": 89480
    },
    {
      "epoch": 0.5739610624917825,
      "grad_norm": 0.2064570039510727,
      "learning_rate": 8.972291264979383e-06,
      "loss": 0.0029,
      "step": 89490
    },
    {
      "epoch": 0.5740251993855685,
      "grad_norm": 0.13895684480667114,
      "learning_rate": 8.971951321729142e-06,
      "loss": 0.0035,
      "step": 89500
    },
    {
      "epoch": 0.5740893362793547,
      "grad_norm": 0.18333616852760315,
      "learning_rate": 8.971611328707512e-06,
      "loss": 0.0029,
      "step": 89510
    },
    {
      "epoch": 0.5741534731731408,
      "grad_norm": 0.08863773941993713,
      "learning_rate": 8.971271285918755e-06,
      "loss": 0.0037,
      "step": 89520
    },
    {
      "epoch": 0.5742176100669268,
      "grad_norm": 0.1741122305393219,
      "learning_rate": 8.97093119336713e-06,
      "loss": 0.005,
      "step": 89530
    },
    {
      "epoch": 0.574281746960713,
      "grad_norm": 0.28641340136528015,
      "learning_rate": 8.970591051056899e-06,
      "loss": 0.0052,
      "step": 89540
    },
    {
      "epoch": 0.574345883854499,
      "grad_norm": 0.1354037970304489,
      "learning_rate": 8.970250858992325e-06,
      "loss": 0.0023,
      "step": 89550
    },
    {
      "epoch": 0.5744100207482852,
      "grad_norm": 0.10131777077913284,
      "learning_rate": 8.969910617177668e-06,
      "loss": 0.0033,
      "step": 89560
    },
    {
      "epoch": 0.5744741576420712,
      "grad_norm": 0.19914120435714722,
      "learning_rate": 8.969570325617196e-06,
      "loss": 0.0034,
      "step": 89570
    },
    {
      "epoch": 0.5745382945358574,
      "grad_norm": 0.18307963013648987,
      "learning_rate": 8.969229984315172e-06,
      "loss": 0.003,
      "step": 89580
    },
    {
      "epoch": 0.5746024314296434,
      "grad_norm": 0.054778408259153366,
      "learning_rate": 8.968889593275857e-06,
      "loss": 0.0039,
      "step": 89590
    },
    {
      "epoch": 0.5746665683234295,
      "grad_norm": 0.09347351640462875,
      "learning_rate": 8.96854915250352e-06,
      "loss": 0.0031,
      "step": 89600
    },
    {
      "epoch": 0.5747307052172156,
      "grad_norm": 0.22702820599079132,
      "learning_rate": 8.968208662002425e-06,
      "loss": 0.0041,
      "step": 89610
    },
    {
      "epoch": 0.5747948421110017,
      "grad_norm": 0.10594911873340607,
      "learning_rate": 8.967868121776841e-06,
      "loss": 0.0032,
      "step": 89620
    },
    {
      "epoch": 0.5748589790047878,
      "grad_norm": 0.10526852309703827,
      "learning_rate": 8.967527531831033e-06,
      "loss": 0.0036,
      "step": 89630
    },
    {
      "epoch": 0.5749231158985739,
      "grad_norm": 0.14717718958854675,
      "learning_rate": 8.967186892169269e-06,
      "loss": 0.002,
      "step": 89640
    },
    {
      "epoch": 0.57498725279236,
      "grad_norm": 0.14983873069286346,
      "learning_rate": 8.966846202795818e-06,
      "loss": 0.0046,
      "step": 89650
    },
    {
      "epoch": 0.5750513896861461,
      "grad_norm": 0.19545316696166992,
      "learning_rate": 8.966505463714948e-06,
      "loss": 0.0057,
      "step": 89660
    },
    {
      "epoch": 0.5751155265799323,
      "grad_norm": 0.25773897767066956,
      "learning_rate": 8.96616467493093e-06,
      "loss": 0.0026,
      "step": 89670
    },
    {
      "epoch": 0.5751796634737183,
      "grad_norm": 0.2134181708097458,
      "learning_rate": 8.965823836448035e-06,
      "loss": 0.0037,
      "step": 89680
    },
    {
      "epoch": 0.5752438003675044,
      "grad_norm": 0.024075880646705627,
      "learning_rate": 8.965482948270533e-06,
      "loss": 0.0099,
      "step": 89690
    },
    {
      "epoch": 0.5753079372612905,
      "grad_norm": 0.18777407705783844,
      "learning_rate": 8.965142010402696e-06,
      "loss": 0.0037,
      "step": 89700
    },
    {
      "epoch": 0.5753720741550766,
      "grad_norm": 0.2322371006011963,
      "learning_rate": 8.964801022848795e-06,
      "loss": 0.0054,
      "step": 89710
    },
    {
      "epoch": 0.5754362110488627,
      "grad_norm": 0.1455300748348236,
      "learning_rate": 8.964459985613104e-06,
      "loss": 0.0022,
      "step": 89720
    },
    {
      "epoch": 0.5755003479426488,
      "grad_norm": 0.1370176523923874,
      "learning_rate": 8.964118898699896e-06,
      "loss": 0.0028,
      "step": 89730
    },
    {
      "epoch": 0.5755644848364349,
      "grad_norm": 0.3160856068134308,
      "learning_rate": 8.963777762113445e-06,
      "loss": 0.0053,
      "step": 89740
    },
    {
      "epoch": 0.575628621730221,
      "grad_norm": 0.12635572254657745,
      "learning_rate": 8.963436575858026e-06,
      "loss": 0.0062,
      "step": 89750
    },
    {
      "epoch": 0.575692758624007,
      "grad_norm": 0.29390159249305725,
      "learning_rate": 8.963095339937914e-06,
      "loss": 0.0038,
      "step": 89760
    },
    {
      "epoch": 0.5757568955177932,
      "grad_norm": 0.09332332015037537,
      "learning_rate": 8.962754054357385e-06,
      "loss": 0.0036,
      "step": 89770
    },
    {
      "epoch": 0.5758210324115792,
      "grad_norm": 0.4807676672935486,
      "learning_rate": 8.962412719120715e-06,
      "loss": 0.0044,
      "step": 89780
    },
    {
      "epoch": 0.5758851693053654,
      "grad_norm": 0.14331871271133423,
      "learning_rate": 8.962071334232182e-06,
      "loss": 0.0029,
      "step": 89790
    },
    {
      "epoch": 0.5759493061991515,
      "grad_norm": 0.3488805294036865,
      "learning_rate": 8.961729899696064e-06,
      "loss": 0.0067,
      "step": 89800
    },
    {
      "epoch": 0.5760134430929376,
      "grad_norm": 1.0194358825683594,
      "learning_rate": 8.961388415516638e-06,
      "loss": 0.0025,
      "step": 89810
    },
    {
      "epoch": 0.5760775799867237,
      "grad_norm": 0.2737617790699005,
      "learning_rate": 8.961046881698184e-06,
      "loss": 0.0038,
      "step": 89820
    },
    {
      "epoch": 0.5761417168805097,
      "grad_norm": 0.15384909510612488,
      "learning_rate": 8.960705298244982e-06,
      "loss": 0.0024,
      "step": 89830
    },
    {
      "epoch": 0.5762058537742959,
      "grad_norm": 0.11745814234018326,
      "learning_rate": 8.960363665161313e-06,
      "loss": 0.0054,
      "step": 89840
    },
    {
      "epoch": 0.5762699906680819,
      "grad_norm": 0.1430215686559677,
      "learning_rate": 8.960021982451455e-06,
      "loss": 0.0023,
      "step": 89850
    },
    {
      "epoch": 0.5763341275618681,
      "grad_norm": 0.13121038675308228,
      "learning_rate": 8.959680250119693e-06,
      "loss": 0.0043,
      "step": 89860
    },
    {
      "epoch": 0.5763982644556541,
      "grad_norm": 0.24054284393787384,
      "learning_rate": 8.959338468170307e-06,
      "loss": 0.003,
      "step": 89870
    },
    {
      "epoch": 0.5764624013494403,
      "grad_norm": 0.12309806048870087,
      "learning_rate": 8.95899663660758e-06,
      "loss": 0.0036,
      "step": 89880
    },
    {
      "epoch": 0.5765265382432263,
      "grad_norm": 0.24552305042743683,
      "learning_rate": 8.958654755435796e-06,
      "loss": 0.0038,
      "step": 89890
    },
    {
      "epoch": 0.5765906751370125,
      "grad_norm": 0.17345677316188812,
      "learning_rate": 8.95831282465924e-06,
      "loss": 0.0064,
      "step": 89900
    },
    {
      "epoch": 0.5766548120307985,
      "grad_norm": 0.16772368550300598,
      "learning_rate": 8.957970844282192e-06,
      "loss": 0.0036,
      "step": 89910
    },
    {
      "epoch": 0.5767189489245846,
      "grad_norm": 0.14221370220184326,
      "learning_rate": 8.957628814308943e-06,
      "loss": 0.0022,
      "step": 89920
    },
    {
      "epoch": 0.5767830858183707,
      "grad_norm": 0.12244567275047302,
      "learning_rate": 8.957286734743775e-06,
      "loss": 0.0046,
      "step": 89930
    },
    {
      "epoch": 0.5768472227121568,
      "grad_norm": 0.4485814869403839,
      "learning_rate": 8.956944605590979e-06,
      "loss": 0.0043,
      "step": 89940
    },
    {
      "epoch": 0.576911359605943,
      "grad_norm": 0.04381153732538223,
      "learning_rate": 8.956602426854836e-06,
      "loss": 0.0025,
      "step": 89950
    },
    {
      "epoch": 0.576975496499729,
      "grad_norm": 0.12222953140735626,
      "learning_rate": 8.956260198539637e-06,
      "loss": 0.005,
      "step": 89960
    },
    {
      "epoch": 0.5770396333935152,
      "grad_norm": 0.011206181719899178,
      "learning_rate": 8.955917920649672e-06,
      "loss": 0.0019,
      "step": 89970
    },
    {
      "epoch": 0.5771037702873012,
      "grad_norm": 0.22010645270347595,
      "learning_rate": 8.955575593189227e-06,
      "loss": 0.0032,
      "step": 89980
    },
    {
      "epoch": 0.5771679071810873,
      "grad_norm": 0.1770000010728836,
      "learning_rate": 8.955233216162594e-06,
      "loss": 0.0028,
      "step": 89990
    },
    {
      "epoch": 0.5772320440748734,
      "grad_norm": 0.18463435769081116,
      "learning_rate": 8.95489078957406e-06,
      "loss": 0.0037,
      "step": 90000
    },
    {
      "epoch": 0.5772961809686595,
      "grad_norm": 0.11717001348733902,
      "learning_rate": 8.95454831342792e-06,
      "loss": 0.0019,
      "step": 90010
    },
    {
      "epoch": 0.5773603178624456,
      "grad_norm": 0.06984009593725204,
      "learning_rate": 8.954205787728462e-06,
      "loss": 0.0019,
      "step": 90020
    },
    {
      "epoch": 0.5774244547562317,
      "grad_norm": 0.38813281059265137,
      "learning_rate": 8.95386321247998e-06,
      "loss": 0.0047,
      "step": 90030
    },
    {
      "epoch": 0.5774885916500178,
      "grad_norm": 0.15131741762161255,
      "learning_rate": 8.953520587686766e-06,
      "loss": 0.0045,
      "step": 90040
    },
    {
      "epoch": 0.5775527285438039,
      "grad_norm": 0.05685955658555031,
      "learning_rate": 8.953177913353113e-06,
      "loss": 0.0025,
      "step": 90050
    },
    {
      "epoch": 0.5776168654375899,
      "grad_norm": 0.3332849144935608,
      "learning_rate": 8.952835189483316e-06,
      "loss": 0.0033,
      "step": 90060
    },
    {
      "epoch": 0.5776810023313761,
      "grad_norm": 0.15496273338794708,
      "learning_rate": 8.95249241608167e-06,
      "loss": 0.0024,
      "step": 90070
    },
    {
      "epoch": 0.5777451392251622,
      "grad_norm": 0.20683197677135468,
      "learning_rate": 8.952149593152468e-06,
      "loss": 0.0025,
      "step": 90080
    },
    {
      "epoch": 0.5778092761189483,
      "grad_norm": 0.1645357310771942,
      "learning_rate": 8.951806720700007e-06,
      "loss": 0.004,
      "step": 90090
    },
    {
      "epoch": 0.5778734130127344,
      "grad_norm": 0.2908009886741638,
      "learning_rate": 8.951463798728584e-06,
      "loss": 0.0045,
      "step": 90100
    },
    {
      "epoch": 0.5779375499065205,
      "grad_norm": 0.13992683589458466,
      "learning_rate": 8.951120827242495e-06,
      "loss": 0.0031,
      "step": 90110
    },
    {
      "epoch": 0.5780016868003066,
      "grad_norm": 0.06215394660830498,
      "learning_rate": 8.950777806246039e-06,
      "loss": 0.0014,
      "step": 90120
    },
    {
      "epoch": 0.5780658236940927,
      "grad_norm": 0.04137173667550087,
      "learning_rate": 8.950434735743512e-06,
      "loss": 0.0029,
      "step": 90130
    },
    {
      "epoch": 0.5781299605878788,
      "grad_norm": 0.7353517413139343,
      "learning_rate": 8.950091615739217e-06,
      "loss": 0.0036,
      "step": 90140
    },
    {
      "epoch": 0.5781940974816648,
      "grad_norm": 0.17141081392765045,
      "learning_rate": 8.949748446237448e-06,
      "loss": 0.0041,
      "step": 90150
    },
    {
      "epoch": 0.578258234375451,
      "grad_norm": 0.2574852406978607,
      "learning_rate": 8.949405227242509e-06,
      "loss": 0.0045,
      "step": 90160
    },
    {
      "epoch": 0.578322371269237,
      "grad_norm": 0.10717305541038513,
      "learning_rate": 8.9490619587587e-06,
      "loss": 0.003,
      "step": 90170
    },
    {
      "epoch": 0.5783865081630232,
      "grad_norm": 0.14781375229358673,
      "learning_rate": 8.948718640790323e-06,
      "loss": 0.0037,
      "step": 90180
    },
    {
      "epoch": 0.5784506450568092,
      "grad_norm": 0.21729065477848053,
      "learning_rate": 8.948375273341681e-06,
      "loss": 0.0032,
      "step": 90190
    },
    {
      "epoch": 0.5785147819505954,
      "grad_norm": 0.3163394629955292,
      "learning_rate": 8.948031856417072e-06,
      "loss": 0.0025,
      "step": 90200
    },
    {
      "epoch": 0.5785789188443814,
      "grad_norm": 0.3094060719013214,
      "learning_rate": 8.947688390020803e-06,
      "loss": 0.0148,
      "step": 90210
    },
    {
      "epoch": 0.5786430557381675,
      "grad_norm": 0.12238597124814987,
      "learning_rate": 8.947344874157179e-06,
      "loss": 0.0059,
      "step": 90220
    },
    {
      "epoch": 0.5787071926319537,
      "grad_norm": 0.12434793263673782,
      "learning_rate": 8.9470013088305e-06,
      "loss": 0.0033,
      "step": 90230
    },
    {
      "epoch": 0.5787713295257397,
      "grad_norm": 0.07099161297082901,
      "learning_rate": 8.946657694045074e-06,
      "loss": 0.003,
      "step": 90240
    },
    {
      "epoch": 0.5788354664195259,
      "grad_norm": 0.336298406124115,
      "learning_rate": 8.946314029805208e-06,
      "loss": 0.0028,
      "step": 90250
    },
    {
      "epoch": 0.5788996033133119,
      "grad_norm": 0.6149856448173523,
      "learning_rate": 8.945970316115205e-06,
      "loss": 0.004,
      "step": 90260
    },
    {
      "epoch": 0.5789637402070981,
      "grad_norm": 0.02784036658704281,
      "learning_rate": 8.945626552979377e-06,
      "loss": 0.0032,
      "step": 90270
    },
    {
      "epoch": 0.5790278771008841,
      "grad_norm": 0.12195788323879242,
      "learning_rate": 8.945282740402024e-06,
      "loss": 0.0042,
      "step": 90280
    },
    {
      "epoch": 0.5790920139946703,
      "grad_norm": 0.16925176978111267,
      "learning_rate": 8.944938878387461e-06,
      "loss": 0.0042,
      "step": 90290
    },
    {
      "epoch": 0.5791561508884563,
      "grad_norm": 0.16154712438583374,
      "learning_rate": 8.944594966939994e-06,
      "loss": 0.0053,
      "step": 90300
    },
    {
      "epoch": 0.5792202877822424,
      "grad_norm": 0.07512946426868439,
      "learning_rate": 8.944251006063934e-06,
      "loss": 0.0022,
      "step": 90310
    },
    {
      "epoch": 0.5792844246760285,
      "grad_norm": 0.18358130753040314,
      "learning_rate": 8.94390699576359e-06,
      "loss": 0.0047,
      "step": 90320
    },
    {
      "epoch": 0.5793485615698146,
      "grad_norm": 0.010401815176010132,
      "learning_rate": 8.94356293604327e-06,
      "loss": 0.0022,
      "step": 90330
    },
    {
      "epoch": 0.5794126984636007,
      "grad_norm": 0.1601855605840683,
      "learning_rate": 8.94321882690729e-06,
      "loss": 0.0028,
      "step": 90340
    },
    {
      "epoch": 0.5794768353573868,
      "grad_norm": 0.12499598413705826,
      "learning_rate": 8.94287466835996e-06,
      "loss": 0.0026,
      "step": 90350
    },
    {
      "epoch": 0.579540972251173,
      "grad_norm": 0.0920039489865303,
      "learning_rate": 8.942530460405592e-06,
      "loss": 0.0033,
      "step": 90360
    },
    {
      "epoch": 0.579605109144959,
      "grad_norm": 0.0942494347691536,
      "learning_rate": 8.942186203048499e-06,
      "loss": 0.0033,
      "step": 90370
    },
    {
      "epoch": 0.5796692460387451,
      "grad_norm": 0.19644950330257416,
      "learning_rate": 8.941841896292997e-06,
      "loss": 0.0051,
      "step": 90380
    },
    {
      "epoch": 0.5797333829325312,
      "grad_norm": 0.15651078522205353,
      "learning_rate": 8.941497540143397e-06,
      "loss": 0.0031,
      "step": 90390
    },
    {
      "epoch": 0.5797975198263173,
      "grad_norm": 0.2172597348690033,
      "learning_rate": 8.941153134604018e-06,
      "loss": 0.0025,
      "step": 90400
    },
    {
      "epoch": 0.5798616567201034,
      "grad_norm": 0.17114651203155518,
      "learning_rate": 8.940808679679172e-06,
      "loss": 0.0039,
      "step": 90410
    },
    {
      "epoch": 0.5799257936138895,
      "grad_norm": 0.15437696874141693,
      "learning_rate": 8.940464175373178e-06,
      "loss": 0.0041,
      "step": 90420
    },
    {
      "epoch": 0.5799899305076756,
      "grad_norm": 0.39087653160095215,
      "learning_rate": 8.940119621690351e-06,
      "loss": 0.0044,
      "step": 90430
    },
    {
      "epoch": 0.5800540674014617,
      "grad_norm": 0.1864572912454605,
      "learning_rate": 8.939775018635008e-06,
      "loss": 0.0024,
      "step": 90440
    },
    {
      "epoch": 0.5801182042952477,
      "grad_norm": 0.10369669646024704,
      "learning_rate": 8.93943036621147e-06,
      "loss": 0.005,
      "step": 90450
    },
    {
      "epoch": 0.5801823411890339,
      "grad_norm": 0.09310199320316315,
      "learning_rate": 8.939085664424055e-06,
      "loss": 0.0024,
      "step": 90460
    },
    {
      "epoch": 0.5802464780828199,
      "grad_norm": 0.046526405960321426,
      "learning_rate": 8.938740913277079e-06,
      "loss": 0.0027,
      "step": 90470
    },
    {
      "epoch": 0.5803106149766061,
      "grad_norm": 0.14390631020069122,
      "learning_rate": 8.938396112774866e-06,
      "loss": 0.0028,
      "step": 90480
    },
    {
      "epoch": 0.5803747518703921,
      "grad_norm": 0.13936559855937958,
      "learning_rate": 8.938051262921735e-06,
      "loss": 0.0033,
      "step": 90490
    },
    {
      "epoch": 0.5804388887641783,
      "grad_norm": 0.05774078145623207,
      "learning_rate": 8.937706363722004e-06,
      "loss": 0.0033,
      "step": 90500
    },
    {
      "epoch": 0.5805030256579644,
      "grad_norm": 0.05919841304421425,
      "learning_rate": 8.937361415180001e-06,
      "loss": 0.0047,
      "step": 90510
    },
    {
      "epoch": 0.5805671625517504,
      "grad_norm": 0.3022819757461548,
      "learning_rate": 8.937016417300046e-06,
      "loss": 0.0021,
      "step": 90520
    },
    {
      "epoch": 0.5806312994455366,
      "grad_norm": 0.2363024353981018,
      "learning_rate": 8.93667137008646e-06,
      "loss": 0.0028,
      "step": 90530
    },
    {
      "epoch": 0.5806954363393226,
      "grad_norm": 0.3508833348751068,
      "learning_rate": 8.93632627354357e-06,
      "loss": 0.0029,
      "step": 90540
    },
    {
      "epoch": 0.5807595732331088,
      "grad_norm": 0.1320483237504959,
      "learning_rate": 8.935981127675695e-06,
      "loss": 0.0022,
      "step": 90550
    },
    {
      "epoch": 0.5808237101268948,
      "grad_norm": 0.2474808692932129,
      "learning_rate": 8.935635932487166e-06,
      "loss": 0.0037,
      "step": 90560
    },
    {
      "epoch": 0.580887847020681,
      "grad_norm": 0.08279784768819809,
      "learning_rate": 8.935290687982306e-06,
      "loss": 0.0039,
      "step": 90570
    },
    {
      "epoch": 0.580951983914467,
      "grad_norm": 0.11578167974948883,
      "learning_rate": 8.934945394165442e-06,
      "loss": 0.0028,
      "step": 90580
    },
    {
      "epoch": 0.5810161208082532,
      "grad_norm": 0.5153977870941162,
      "learning_rate": 8.934600051040898e-06,
      "loss": 0.0047,
      "step": 90590
    },
    {
      "epoch": 0.5810802577020392,
      "grad_norm": 0.15942715108394623,
      "learning_rate": 8.934254658613003e-06,
      "loss": 0.0066,
      "step": 90600
    },
    {
      "epoch": 0.5811443945958253,
      "grad_norm": 0.10940330475568771,
      "learning_rate": 8.933909216886087e-06,
      "loss": 0.0028,
      "step": 90610
    },
    {
      "epoch": 0.5812085314896114,
      "grad_norm": 0.3347488045692444,
      "learning_rate": 8.933563725864478e-06,
      "loss": 0.004,
      "step": 90620
    },
    {
      "epoch": 0.5812726683833975,
      "grad_norm": 0.15730610489845276,
      "learning_rate": 8.933218185552503e-06,
      "loss": 0.003,
      "step": 90630
    },
    {
      "epoch": 0.5813368052771836,
      "grad_norm": 0.18061256408691406,
      "learning_rate": 8.932872595954493e-06,
      "loss": 0.0024,
      "step": 90640
    },
    {
      "epoch": 0.5814009421709697,
      "grad_norm": 0.09252581745386124,
      "learning_rate": 8.93252695707478e-06,
      "loss": 0.0025,
      "step": 90650
    },
    {
      "epoch": 0.5814650790647559,
      "grad_norm": 0.24369707703590393,
      "learning_rate": 8.93218126891769e-06,
      "loss": 0.0031,
      "step": 90660
    },
    {
      "epoch": 0.5815292159585419,
      "grad_norm": 0.2268710881471634,
      "learning_rate": 8.93183553148756e-06,
      "loss": 0.0036,
      "step": 90670
    },
    {
      "epoch": 0.581593352852328,
      "grad_norm": 0.1491227000951767,
      "learning_rate": 8.931489744788722e-06,
      "loss": 0.0018,
      "step": 90680
    },
    {
      "epoch": 0.5816574897461141,
      "grad_norm": 0.18808454275131226,
      "learning_rate": 8.931143908825508e-06,
      "loss": 0.0043,
      "step": 90690
    },
    {
      "epoch": 0.5817216266399002,
      "grad_norm": 0.10794474184513092,
      "learning_rate": 8.93079802360225e-06,
      "loss": 0.004,
      "step": 90700
    },
    {
      "epoch": 0.5817857635336863,
      "grad_norm": 0.14503294229507446,
      "learning_rate": 8.930452089123283e-06,
      "loss": 0.0026,
      "step": 90710
    },
    {
      "epoch": 0.5818499004274724,
      "grad_norm": 0.10027197748422623,
      "learning_rate": 8.930106105392944e-06,
      "loss": 0.0011,
      "step": 90720
    },
    {
      "epoch": 0.5819140373212585,
      "grad_norm": 0.057823918759822845,
      "learning_rate": 8.929760072415565e-06,
      "loss": 0.0047,
      "step": 90730
    },
    {
      "epoch": 0.5819781742150446,
      "grad_norm": 0.19504736363887787,
      "learning_rate": 8.929413990195485e-06,
      "loss": 0.0031,
      "step": 90740
    },
    {
      "epoch": 0.5820423111088306,
      "grad_norm": 0.17649537324905396,
      "learning_rate": 8.929067858737039e-06,
      "loss": 0.0029,
      "step": 90750
    },
    {
      "epoch": 0.5821064480026168,
      "grad_norm": 0.05666594207286835,
      "learning_rate": 8.928721678044564e-06,
      "loss": 0.0032,
      "step": 90760
    },
    {
      "epoch": 0.5821705848964028,
      "grad_norm": 0.14649075269699097,
      "learning_rate": 8.928375448122399e-06,
      "loss": 0.0024,
      "step": 90770
    },
    {
      "epoch": 0.582234721790189,
      "grad_norm": 0.26468437910079956,
      "learning_rate": 8.928029168974881e-06,
      "loss": 0.0029,
      "step": 90780
    },
    {
      "epoch": 0.5822988586839751,
      "grad_norm": 0.07763198018074036,
      "learning_rate": 8.927682840606352e-06,
      "loss": 0.0028,
      "step": 90790
    },
    {
      "epoch": 0.5823629955777612,
      "grad_norm": 0.014735487289726734,
      "learning_rate": 8.92733646302115e-06,
      "loss": 0.0032,
      "step": 90800
    },
    {
      "epoch": 0.5824271324715473,
      "grad_norm": 0.26133033633232117,
      "learning_rate": 8.926990036223615e-06,
      "loss": 0.0033,
      "step": 90810
    },
    {
      "epoch": 0.5824912693653334,
      "grad_norm": 0.15196596086025238,
      "learning_rate": 8.926643560218087e-06,
      "loss": 0.0029,
      "step": 90820
    },
    {
      "epoch": 0.5825554062591195,
      "grad_norm": 0.14441141486167908,
      "learning_rate": 8.92629703500891e-06,
      "loss": 0.002,
      "step": 90830
    },
    {
      "epoch": 0.5826195431529055,
      "grad_norm": 0.2370399534702301,
      "learning_rate": 8.925950460600425e-06,
      "loss": 0.0029,
      "step": 90840
    },
    {
      "epoch": 0.5826836800466917,
      "grad_norm": 0.5152098536491394,
      "learning_rate": 8.925603836996975e-06,
      "loss": 0.0022,
      "step": 90850
    },
    {
      "epoch": 0.5827478169404777,
      "grad_norm": 0.29338768124580383,
      "learning_rate": 8.925257164202903e-06,
      "loss": 0.002,
      "step": 90860
    },
    {
      "epoch": 0.5828119538342639,
      "grad_norm": 0.05307412147521973,
      "learning_rate": 8.924910442222555e-06,
      "loss": 0.0031,
      "step": 90870
    },
    {
      "epoch": 0.5828760907280499,
      "grad_norm": 0.16998408734798431,
      "learning_rate": 8.924563671060272e-06,
      "loss": 0.0034,
      "step": 90880
    },
    {
      "epoch": 0.5829402276218361,
      "grad_norm": 0.1047939583659172,
      "learning_rate": 8.9242168507204e-06,
      "loss": 0.0013,
      "step": 90890
    },
    {
      "epoch": 0.5830043645156221,
      "grad_norm": 0.09160842001438141,
      "learning_rate": 8.923869981207289e-06,
      "loss": 0.0037,
      "step": 90900
    },
    {
      "epoch": 0.5830685014094082,
      "grad_norm": 0.10264736413955688,
      "learning_rate": 8.923523062525282e-06,
      "loss": 0.0039,
      "step": 90910
    },
    {
      "epoch": 0.5831326383031943,
      "grad_norm": 0.05445673316717148,
      "learning_rate": 8.923176094678726e-06,
      "loss": 0.0043,
      "step": 90920
    },
    {
      "epoch": 0.5831967751969804,
      "grad_norm": 0.06704435497522354,
      "learning_rate": 8.92282907767197e-06,
      "loss": 0.0055,
      "step": 90930
    },
    {
      "epoch": 0.5832609120907666,
      "grad_norm": 0.03559799864888191,
      "learning_rate": 8.922482011509364e-06,
      "loss": 0.006,
      "step": 90940
    },
    {
      "epoch": 0.5833250489845526,
      "grad_norm": 0.11612100899219513,
      "learning_rate": 8.922134896195253e-06,
      "loss": 0.0044,
      "step": 90950
    },
    {
      "epoch": 0.5833891858783388,
      "grad_norm": 0.013056891039013863,
      "learning_rate": 8.92178773173399e-06,
      "loss": 0.0027,
      "step": 90960
    },
    {
      "epoch": 0.5834533227721248,
      "grad_norm": 0.06981601566076279,
      "learning_rate": 8.921440518129922e-06,
      "loss": 0.0033,
      "step": 90970
    },
    {
      "epoch": 0.583517459665911,
      "grad_norm": 0.11842454969882965,
      "learning_rate": 8.921093255387402e-06,
      "loss": 0.0028,
      "step": 90980
    },
    {
      "epoch": 0.583581596559697,
      "grad_norm": 0.35216161608695984,
      "learning_rate": 8.920745943510783e-06,
      "loss": 0.0026,
      "step": 90990
    },
    {
      "epoch": 0.5836457334534831,
      "grad_norm": 0.05675550177693367,
      "learning_rate": 8.920398582504415e-06,
      "loss": 0.0057,
      "step": 91000
    },
    {
      "epoch": 0.5837098703472692,
      "grad_norm": 0.09108414500951767,
      "learning_rate": 8.92005117237265e-06,
      "loss": 0.0025,
      "step": 91010
    },
    {
      "epoch": 0.5837740072410553,
      "grad_norm": 0.15295842289924622,
      "learning_rate": 8.919703713119842e-06,
      "loss": 0.0018,
      "step": 91020
    },
    {
      "epoch": 0.5838381441348414,
      "grad_norm": 0.16002964973449707,
      "learning_rate": 8.919356204750346e-06,
      "loss": 0.0039,
      "step": 91030
    },
    {
      "epoch": 0.5839022810286275,
      "grad_norm": 0.3844493329524994,
      "learning_rate": 8.919008647268515e-06,
      "loss": 0.0032,
      "step": 91040
    },
    {
      "epoch": 0.5839664179224135,
      "grad_norm": 0.11783535778522491,
      "learning_rate": 8.918661040678705e-06,
      "loss": 0.002,
      "step": 91050
    },
    {
      "epoch": 0.5840305548161997,
      "grad_norm": 0.3455294072628021,
      "learning_rate": 8.918313384985271e-06,
      "loss": 0.0041,
      "step": 91060
    },
    {
      "epoch": 0.5840946917099858,
      "grad_norm": 0.05667266622185707,
      "learning_rate": 8.91796568019257e-06,
      "loss": 0.0031,
      "step": 91070
    },
    {
      "epoch": 0.5841588286037719,
      "grad_norm": 0.05197868496179581,
      "learning_rate": 8.917617926304957e-06,
      "loss": 0.0061,
      "step": 91080
    },
    {
      "epoch": 0.584222965497558,
      "grad_norm": 0.16658474504947662,
      "learning_rate": 8.917270123326796e-06,
      "loss": 0.0029,
      "step": 91090
    },
    {
      "epoch": 0.5842871023913441,
      "grad_norm": 0.4391426146030426,
      "learning_rate": 8.916922271262438e-06,
      "loss": 0.0029,
      "step": 91100
    },
    {
      "epoch": 0.5843512392851302,
      "grad_norm": 0.16311413049697876,
      "learning_rate": 8.916574370116245e-06,
      "loss": 0.0031,
      "step": 91110
    },
    {
      "epoch": 0.5844153761789163,
      "grad_norm": 0.2503029704093933,
      "learning_rate": 8.916226419892576e-06,
      "loss": 0.0023,
      "step": 91120
    },
    {
      "epoch": 0.5844795130727024,
      "grad_norm": 0.040789201855659485,
      "learning_rate": 8.91587842059579e-06,
      "loss": 0.004,
      "step": 91130
    },
    {
      "epoch": 0.5845436499664884,
      "grad_norm": 0.5060019493103027,
      "learning_rate": 8.91553037223025e-06,
      "loss": 0.0086,
      "step": 91140
    },
    {
      "epoch": 0.5846077868602746,
      "grad_norm": 0.13935501873493195,
      "learning_rate": 8.915182274800315e-06,
      "loss": 0.0026,
      "step": 91150
    },
    {
      "epoch": 0.5846719237540606,
      "grad_norm": 0.2682758867740631,
      "learning_rate": 8.91483412831035e-06,
      "loss": 0.0029,
      "step": 91160
    },
    {
      "epoch": 0.5847360606478468,
      "grad_norm": 0.2726670205593109,
      "learning_rate": 8.914485932764714e-06,
      "loss": 0.0034,
      "step": 91170
    },
    {
      "epoch": 0.5848001975416328,
      "grad_norm": 0.1472742110490799,
      "learning_rate": 8.914137688167772e-06,
      "loss": 0.0041,
      "step": 91180
    },
    {
      "epoch": 0.584864334435419,
      "grad_norm": 0.21696236729621887,
      "learning_rate": 8.913789394523887e-06,
      "loss": 0.0037,
      "step": 91190
    },
    {
      "epoch": 0.584928471329205,
      "grad_norm": 0.16352948546409607,
      "learning_rate": 8.913441051837424e-06,
      "loss": 0.0038,
      "step": 91200
    },
    {
      "epoch": 0.5849926082229912,
      "grad_norm": 0.1005765050649643,
      "learning_rate": 8.913092660112748e-06,
      "loss": 0.0048,
      "step": 91210
    },
    {
      "epoch": 0.5850567451167773,
      "grad_norm": 0.1923978626728058,
      "learning_rate": 8.912744219354224e-06,
      "loss": 0.004,
      "step": 91220
    },
    {
      "epoch": 0.5851208820105633,
      "grad_norm": 0.054473694413900375,
      "learning_rate": 8.912395729566219e-06,
      "loss": 0.0029,
      "step": 91230
    },
    {
      "epoch": 0.5851850189043495,
      "grad_norm": 0.10405640304088593,
      "learning_rate": 8.912047190753098e-06,
      "loss": 0.0032,
      "step": 91240
    },
    {
      "epoch": 0.5852491557981355,
      "grad_norm": 0.1484929919242859,
      "learning_rate": 8.91169860291923e-06,
      "loss": 0.0041,
      "step": 91250
    },
    {
      "epoch": 0.5853132926919217,
      "grad_norm": 0.22620069980621338,
      "learning_rate": 8.911349966068986e-06,
      "loss": 0.0049,
      "step": 91260
    },
    {
      "epoch": 0.5853774295857077,
      "grad_norm": 0.18138428032398224,
      "learning_rate": 8.911001280206728e-06,
      "loss": 0.0035,
      "step": 91270
    },
    {
      "epoch": 0.5854415664794939,
      "grad_norm": 0.09323873370885849,
      "learning_rate": 8.910652545336828e-06,
      "loss": 0.0024,
      "step": 91280
    },
    {
      "epoch": 0.5855057033732799,
      "grad_norm": 0.21446284651756287,
      "learning_rate": 8.91030376146366e-06,
      "loss": 0.0041,
      "step": 91290
    },
    {
      "epoch": 0.585569840267066,
      "grad_norm": 0.07625725865364075,
      "learning_rate": 8.90995492859159e-06,
      "loss": 0.0023,
      "step": 91300
    },
    {
      "epoch": 0.5856339771608521,
      "grad_norm": 0.15408417582511902,
      "learning_rate": 8.90960604672499e-06,
      "loss": 0.0026,
      "step": 91310
    },
    {
      "epoch": 0.5856981140546382,
      "grad_norm": 0.14646489918231964,
      "learning_rate": 8.909257115868232e-06,
      "loss": 0.0021,
      "step": 91320
    },
    {
      "epoch": 0.5857622509484243,
      "grad_norm": 0.08991856873035431,
      "learning_rate": 8.908908136025689e-06,
      "loss": 0.0018,
      "step": 91330
    },
    {
      "epoch": 0.5858263878422104,
      "grad_norm": 0.39072686433792114,
      "learning_rate": 8.908559107201732e-06,
      "loss": 0.0033,
      "step": 91340
    },
    {
      "epoch": 0.5858905247359966,
      "grad_norm": 0.08406838774681091,
      "learning_rate": 8.908210029400738e-06,
      "loss": 0.0033,
      "step": 91350
    },
    {
      "epoch": 0.5859546616297826,
      "grad_norm": 0.07078109681606293,
      "learning_rate": 8.90786090262708e-06,
      "loss": 0.0032,
      "step": 91360
    },
    {
      "epoch": 0.5860187985235688,
      "grad_norm": 0.20980200171470642,
      "learning_rate": 8.90751172688513e-06,
      "loss": 0.0044,
      "step": 91370
    },
    {
      "epoch": 0.5860829354173548,
      "grad_norm": 0.13432657718658447,
      "learning_rate": 8.907162502179266e-06,
      "loss": 0.0028,
      "step": 91380
    },
    {
      "epoch": 0.5861470723111409,
      "grad_norm": 0.1478530764579773,
      "learning_rate": 8.906813228513863e-06,
      "loss": 0.0033,
      "step": 91390
    },
    {
      "epoch": 0.586211209204927,
      "grad_norm": 0.3381941020488739,
      "learning_rate": 8.906463905893296e-06,
      "loss": 0.0037,
      "step": 91400
    },
    {
      "epoch": 0.5862753460987131,
      "grad_norm": 0.22154425084590912,
      "learning_rate": 8.906114534321948e-06,
      "loss": 0.0034,
      "step": 91410
    },
    {
      "epoch": 0.5863394829924992,
      "grad_norm": 0.04053284227848053,
      "learning_rate": 8.90576511380419e-06,
      "loss": 0.0048,
      "step": 91420
    },
    {
      "epoch": 0.5864036198862853,
      "grad_norm": 0.11504218727350235,
      "learning_rate": 8.905415644344406e-06,
      "loss": 0.0025,
      "step": 91430
    },
    {
      "epoch": 0.5864677567800713,
      "grad_norm": 0.2638290226459503,
      "learning_rate": 8.905066125946973e-06,
      "loss": 0.0033,
      "step": 91440
    },
    {
      "epoch": 0.5865318936738575,
      "grad_norm": 0.12543456256389618,
      "learning_rate": 8.904716558616269e-06,
      "loss": 0.0031,
      "step": 91450
    },
    {
      "epoch": 0.5865960305676435,
      "grad_norm": 0.1517900973558426,
      "learning_rate": 8.904366942356677e-06,
      "loss": 0.0029,
      "step": 91460
    },
    {
      "epoch": 0.5866601674614297,
      "grad_norm": 0.2582097351551056,
      "learning_rate": 8.904017277172577e-06,
      "loss": 0.0036,
      "step": 91470
    },
    {
      "epoch": 0.5867243043552157,
      "grad_norm": 0.12383686751127243,
      "learning_rate": 8.90366756306835e-06,
      "loss": 0.0031,
      "step": 91480
    },
    {
      "epoch": 0.5867884412490019,
      "grad_norm": 0.08443358540534973,
      "learning_rate": 8.903317800048378e-06,
      "loss": 0.0058,
      "step": 91490
    },
    {
      "epoch": 0.586852578142788,
      "grad_norm": 0.23556144535541534,
      "learning_rate": 8.902967988117044e-06,
      "loss": 0.003,
      "step": 91500
    },
    {
      "epoch": 0.586916715036574,
      "grad_norm": 0.09631785750389099,
      "learning_rate": 8.902618127278733e-06,
      "loss": 0.0031,
      "step": 91510
    },
    {
      "epoch": 0.5869808519303602,
      "grad_norm": 0.08299261331558228,
      "learning_rate": 8.902268217537827e-06,
      "loss": 0.0019,
      "step": 91520
    },
    {
      "epoch": 0.5870449888241462,
      "grad_norm": 0.0696060061454773,
      "learning_rate": 8.901918258898711e-06,
      "loss": 0.0038,
      "step": 91530
    },
    {
      "epoch": 0.5871091257179324,
      "grad_norm": 0.22105728089809418,
      "learning_rate": 8.90156825136577e-06,
      "loss": 0.0049,
      "step": 91540
    },
    {
      "epoch": 0.5871732626117184,
      "grad_norm": 0.1963120400905609,
      "learning_rate": 8.901218194943392e-06,
      "loss": 0.002,
      "step": 91550
    },
    {
      "epoch": 0.5872373995055046,
      "grad_norm": 0.1863541156053543,
      "learning_rate": 8.900868089635963e-06,
      "loss": 0.0033,
      "step": 91560
    },
    {
      "epoch": 0.5873015363992906,
      "grad_norm": 0.16078346967697144,
      "learning_rate": 8.900517935447866e-06,
      "loss": 0.0028,
      "step": 91570
    },
    {
      "epoch": 0.5873656732930768,
      "grad_norm": 0.13000799715518951,
      "learning_rate": 8.900167732383494e-06,
      "loss": 0.0032,
      "step": 91580
    },
    {
      "epoch": 0.5874298101868628,
      "grad_norm": 0.3557916581630707,
      "learning_rate": 8.89981748044723e-06,
      "loss": 0.005,
      "step": 91590
    },
    {
      "epoch": 0.587493947080649,
      "grad_norm": 0.1393110156059265,
      "learning_rate": 8.899467179643469e-06,
      "loss": 0.0046,
      "step": 91600
    },
    {
      "epoch": 0.587558083974435,
      "grad_norm": 0.15164828300476074,
      "learning_rate": 8.899116829976595e-06,
      "loss": 0.0045,
      "step": 91610
    },
    {
      "epoch": 0.5876222208682211,
      "grad_norm": 0.14568741619586945,
      "learning_rate": 8.898766431451001e-06,
      "loss": 0.0043,
      "step": 91620
    },
    {
      "epoch": 0.5876863577620073,
      "grad_norm": 0.058165088295936584,
      "learning_rate": 8.898415984071078e-06,
      "loss": 0.0026,
      "step": 91630
    },
    {
      "epoch": 0.5877504946557933,
      "grad_norm": 0.08098440617322922,
      "learning_rate": 8.898065487841216e-06,
      "loss": 0.002,
      "step": 91640
    },
    {
      "epoch": 0.5878146315495795,
      "grad_norm": 0.021908633410930634,
      "learning_rate": 8.897714942765806e-06,
      "loss": 0.0019,
      "step": 91650
    },
    {
      "epoch": 0.5878787684433655,
      "grad_norm": 0.07424472272396088,
      "learning_rate": 8.897364348849244e-06,
      "loss": 0.0057,
      "step": 91660
    },
    {
      "epoch": 0.5879429053371517,
      "grad_norm": 0.39349886775016785,
      "learning_rate": 8.897013706095921e-06,
      "loss": 0.004,
      "step": 91670
    },
    {
      "epoch": 0.5880070422309377,
      "grad_norm": 0.14584460854530334,
      "learning_rate": 8.896663014510231e-06,
      "loss": 0.0022,
      "step": 91680
    },
    {
      "epoch": 0.5880711791247238,
      "grad_norm": 0.18785099685192108,
      "learning_rate": 8.89631227409657e-06,
      "loss": 0.004,
      "step": 91690
    },
    {
      "epoch": 0.5881353160185099,
      "grad_norm": 0.1897226721048355,
      "learning_rate": 8.89596148485933e-06,
      "loss": 0.004,
      "step": 91700
    },
    {
      "epoch": 0.588199452912296,
      "grad_norm": 0.0959388017654419,
      "learning_rate": 8.895610646802907e-06,
      "loss": 0.0049,
      "step": 91710
    },
    {
      "epoch": 0.5882635898060821,
      "grad_norm": 0.04107668995857239,
      "learning_rate": 8.895259759931701e-06,
      "loss": 0.0038,
      "step": 91720
    },
    {
      "epoch": 0.5883277266998682,
      "grad_norm": 0.30913951992988586,
      "learning_rate": 8.894908824250106e-06,
      "loss": 0.0046,
      "step": 91730
    },
    {
      "epoch": 0.5883918635936543,
      "grad_norm": 0.13641460239887238,
      "learning_rate": 8.894557839762518e-06,
      "loss": 0.0043,
      "step": 91740
    },
    {
      "epoch": 0.5884560004874404,
      "grad_norm": 0.291328489780426,
      "learning_rate": 8.894206806473337e-06,
      "loss": 0.0027,
      "step": 91750
    },
    {
      "epoch": 0.5885201373812264,
      "grad_norm": 0.14278985559940338,
      "learning_rate": 8.893855724386964e-06,
      "loss": 0.0038,
      "step": 91760
    },
    {
      "epoch": 0.5885842742750126,
      "grad_norm": 0.17391382157802582,
      "learning_rate": 8.893504593507793e-06,
      "loss": 0.0025,
      "step": 91770
    },
    {
      "epoch": 0.5886484111687987,
      "grad_norm": 0.10660814493894577,
      "learning_rate": 8.893153413840228e-06,
      "loss": 0.003,
      "step": 91780
    },
    {
      "epoch": 0.5887125480625848,
      "grad_norm": 0.15488052368164062,
      "learning_rate": 8.892802185388669e-06,
      "loss": 0.0062,
      "step": 91790
    },
    {
      "epoch": 0.5887766849563709,
      "grad_norm": 0.02289946936070919,
      "learning_rate": 8.892450908157514e-06,
      "loss": 0.0062,
      "step": 91800
    },
    {
      "epoch": 0.588840821850157,
      "grad_norm": 0.03142036497592926,
      "learning_rate": 8.89209958215117e-06,
      "loss": 0.0041,
      "step": 91810
    },
    {
      "epoch": 0.5889049587439431,
      "grad_norm": 0.042830560356378555,
      "learning_rate": 8.891748207374036e-06,
      "loss": 0.0032,
      "step": 91820
    },
    {
      "epoch": 0.5889690956377291,
      "grad_norm": 0.1880921572446823,
      "learning_rate": 8.891396783830515e-06,
      "loss": 0.0053,
      "step": 91830
    },
    {
      "epoch": 0.5890332325315153,
      "grad_norm": 0.33972400426864624,
      "learning_rate": 8.891045311525011e-06,
      "loss": 0.0017,
      "step": 91840
    },
    {
      "epoch": 0.5890973694253013,
      "grad_norm": 0.2004566192626953,
      "learning_rate": 8.89069379046193e-06,
      "loss": 0.0024,
      "step": 91850
    },
    {
      "epoch": 0.5891615063190875,
      "grad_norm": 0.12435439974069595,
      "learning_rate": 8.890342220645674e-06,
      "loss": 0.0032,
      "step": 91860
    },
    {
      "epoch": 0.5892256432128735,
      "grad_norm": 0.22347719967365265,
      "learning_rate": 8.889990602080649e-06,
      "loss": 0.0048,
      "step": 91870
    },
    {
      "epoch": 0.5892897801066597,
      "grad_norm": 0.15794594585895538,
      "learning_rate": 8.889638934771262e-06,
      "loss": 0.0028,
      "step": 91880
    },
    {
      "epoch": 0.5893539170004457,
      "grad_norm": 0.16316534578800201,
      "learning_rate": 8.889287218721921e-06,
      "loss": 0.0037,
      "step": 91890
    },
    {
      "epoch": 0.5894180538942319,
      "grad_norm": 0.23359239101409912,
      "learning_rate": 8.888935453937028e-06,
      "loss": 0.0043,
      "step": 91900
    },
    {
      "epoch": 0.5894821907880179,
      "grad_norm": 0.25535106658935547,
      "learning_rate": 8.888583640420996e-06,
      "loss": 0.0039,
      "step": 91910
    },
    {
      "epoch": 0.589546327681804,
      "grad_norm": 0.3038575351238251,
      "learning_rate": 8.888231778178234e-06,
      "loss": 0.0034,
      "step": 91920
    },
    {
      "epoch": 0.5896104645755902,
      "grad_norm": 0.21951676905155182,
      "learning_rate": 8.887879867213146e-06,
      "loss": 0.0034,
      "step": 91930
    },
    {
      "epoch": 0.5896746014693762,
      "grad_norm": 0.1366313397884369,
      "learning_rate": 8.887527907530146e-06,
      "loss": 0.0056,
      "step": 91940
    },
    {
      "epoch": 0.5897387383631624,
      "grad_norm": 0.12945735454559326,
      "learning_rate": 8.887175899133642e-06,
      "loss": 0.0023,
      "step": 91950
    },
    {
      "epoch": 0.5898028752569484,
      "grad_norm": 0.21439526975154877,
      "learning_rate": 8.886823842028047e-06,
      "loss": 0.0059,
      "step": 91960
    },
    {
      "epoch": 0.5898670121507346,
      "grad_norm": 0.2983899712562561,
      "learning_rate": 8.88647173621777e-06,
      "loss": 0.0026,
      "step": 91970
    },
    {
      "epoch": 0.5899311490445206,
      "grad_norm": 0.05508747324347496,
      "learning_rate": 8.886119581707227e-06,
      "loss": 0.0022,
      "step": 91980
    },
    {
      "epoch": 0.5899952859383067,
      "grad_norm": 0.45103803277015686,
      "learning_rate": 8.885767378500827e-06,
      "loss": 0.0069,
      "step": 91990
    },
    {
      "epoch": 0.5900594228320928,
      "grad_norm": 0.07513292878866196,
      "learning_rate": 8.885415126602983e-06,
      "loss": 0.0045,
      "step": 92000
    },
    {
      "epoch": 0.5901235597258789,
      "grad_norm": 0.2919937074184418,
      "learning_rate": 8.88506282601811e-06,
      "loss": 0.0027,
      "step": 92010
    },
    {
      "epoch": 0.590187696619665,
      "grad_norm": 0.13594791293144226,
      "learning_rate": 8.884710476750628e-06,
      "loss": 0.0015,
      "step": 92020
    },
    {
      "epoch": 0.5902518335134511,
      "grad_norm": 0.10354795306921005,
      "learning_rate": 8.884358078804944e-06,
      "loss": 0.0038,
      "step": 92030
    },
    {
      "epoch": 0.5903159704072372,
      "grad_norm": 0.20591247081756592,
      "learning_rate": 8.884005632185477e-06,
      "loss": 0.0019,
      "step": 92040
    },
    {
      "epoch": 0.5903801073010233,
      "grad_norm": 0.19913780689239502,
      "learning_rate": 8.883653136896644e-06,
      "loss": 0.0037,
      "step": 92050
    },
    {
      "epoch": 0.5904442441948095,
      "grad_norm": 0.15183718502521515,
      "learning_rate": 8.883300592942863e-06,
      "loss": 0.0027,
      "step": 92060
    },
    {
      "epoch": 0.5905083810885955,
      "grad_norm": 0.1707528531551361,
      "learning_rate": 8.882948000328548e-06,
      "loss": 0.0029,
      "step": 92070
    },
    {
      "epoch": 0.5905725179823816,
      "grad_norm": 0.21265354752540588,
      "learning_rate": 8.882595359058122e-06,
      "loss": 0.0044,
      "step": 92080
    },
    {
      "epoch": 0.5906366548761677,
      "grad_norm": 0.11014064401388168,
      "learning_rate": 8.882242669135999e-06,
      "loss": 0.0038,
      "step": 92090
    },
    {
      "epoch": 0.5907007917699538,
      "grad_norm": 0.31167688965797424,
      "learning_rate": 8.8818899305666e-06,
      "loss": 0.0025,
      "step": 92100
    },
    {
      "epoch": 0.5907649286637399,
      "grad_norm": 0.2547362148761749,
      "learning_rate": 8.881537143354349e-06,
      "loss": 0.004,
      "step": 92110
    },
    {
      "epoch": 0.590829065557526,
      "grad_norm": 0.09043922275304794,
      "learning_rate": 8.881184307503662e-06,
      "loss": 0.0029,
      "step": 92120
    },
    {
      "epoch": 0.590893202451312,
      "grad_norm": 0.14289937913417816,
      "learning_rate": 8.88083142301896e-06,
      "loss": 0.0022,
      "step": 92130
    },
    {
      "epoch": 0.5909573393450982,
      "grad_norm": 0.1801345944404602,
      "learning_rate": 8.880478489904669e-06,
      "loss": 0.0058,
      "step": 92140
    },
    {
      "epoch": 0.5910214762388842,
      "grad_norm": 0.07819484919309616,
      "learning_rate": 8.88012550816521e-06,
      "loss": 0.0028,
      "step": 92150
    },
    {
      "epoch": 0.5910856131326704,
      "grad_norm": 0.12236767262220383,
      "learning_rate": 8.879772477805003e-06,
      "loss": 0.0034,
      "step": 92160
    },
    {
      "epoch": 0.5911497500264564,
      "grad_norm": 0.14268234372138977,
      "learning_rate": 8.879419398828476e-06,
      "loss": 0.0042,
      "step": 92170
    },
    {
      "epoch": 0.5912138869202426,
      "grad_norm": 0.15880000591278076,
      "learning_rate": 8.87906627124005e-06,
      "loss": 0.0018,
      "step": 92180
    },
    {
      "epoch": 0.5912780238140286,
      "grad_norm": 0.1848147213459015,
      "learning_rate": 8.878713095044152e-06,
      "loss": 0.0032,
      "step": 92190
    },
    {
      "epoch": 0.5913421607078148,
      "grad_norm": 0.3286655843257904,
      "learning_rate": 8.878359870245205e-06,
      "loss": 0.0043,
      "step": 92200
    },
    {
      "epoch": 0.5914062976016009,
      "grad_norm": 0.3613602817058563,
      "learning_rate": 8.878006596847638e-06,
      "loss": 0.0039,
      "step": 92210
    },
    {
      "epoch": 0.5914704344953869,
      "grad_norm": 0.1943521499633789,
      "learning_rate": 8.877653274855877e-06,
      "loss": 0.0033,
      "step": 92220
    },
    {
      "epoch": 0.5915345713891731,
      "grad_norm": 0.28001415729522705,
      "learning_rate": 8.87729990427435e-06,
      "loss": 0.0032,
      "step": 92230
    },
    {
      "epoch": 0.5915987082829591,
      "grad_norm": 0.03330661356449127,
      "learning_rate": 8.876946485107482e-06,
      "loss": 0.0025,
      "step": 92240
    },
    {
      "epoch": 0.5916628451767453,
      "grad_norm": 0.16898031532764435,
      "learning_rate": 8.876593017359706e-06,
      "loss": 0.0088,
      "step": 92250
    },
    {
      "epoch": 0.5917269820705313,
      "grad_norm": 0.18300198018550873,
      "learning_rate": 8.876239501035448e-06,
      "loss": 0.0049,
      "step": 92260
    },
    {
      "epoch": 0.5917911189643175,
      "grad_norm": 0.15687091648578644,
      "learning_rate": 8.87588593613914e-06,
      "loss": 0.0027,
      "step": 92270
    },
    {
      "epoch": 0.5918552558581035,
      "grad_norm": 0.500644326210022,
      "learning_rate": 8.875532322675208e-06,
      "loss": 0.0064,
      "step": 92280
    },
    {
      "epoch": 0.5919193927518897,
      "grad_norm": 0.06440155953168869,
      "learning_rate": 8.87517866064809e-06,
      "loss": 0.0052,
      "step": 92290
    },
    {
      "epoch": 0.5919835296456757,
      "grad_norm": 0.05679481849074364,
      "learning_rate": 8.87482495006221e-06,
      "loss": 0.0037,
      "step": 92300
    },
    {
      "epoch": 0.5920476665394618,
      "grad_norm": 0.13847221434116364,
      "learning_rate": 8.874471190922007e-06,
      "loss": 0.0031,
      "step": 92310
    },
    {
      "epoch": 0.5921118034332479,
      "grad_norm": 0.07948891073465347,
      "learning_rate": 8.87411738323191e-06,
      "loss": 0.0023,
      "step": 92320
    },
    {
      "epoch": 0.592175940327034,
      "grad_norm": 0.21524807810783386,
      "learning_rate": 8.873763526996353e-06,
      "loss": 0.0025,
      "step": 92330
    },
    {
      "epoch": 0.5922400772208202,
      "grad_norm": 0.050780776888132095,
      "learning_rate": 8.873409622219771e-06,
      "loss": 0.0031,
      "step": 92340
    },
    {
      "epoch": 0.5923042141146062,
      "grad_norm": 0.1852717250585556,
      "learning_rate": 8.873055668906597e-06,
      "loss": 0.0028,
      "step": 92350
    },
    {
      "epoch": 0.5923683510083924,
      "grad_norm": 0.15096399188041687,
      "learning_rate": 8.87270166706127e-06,
      "loss": 0.0031,
      "step": 92360
    },
    {
      "epoch": 0.5924324879021784,
      "grad_norm": 0.16869080066680908,
      "learning_rate": 8.872347616688222e-06,
      "loss": 0.0032,
      "step": 92370
    },
    {
      "epoch": 0.5924966247959645,
      "grad_norm": 0.31817638874053955,
      "learning_rate": 8.871993517791891e-06,
      "loss": 0.0071,
      "step": 92380
    },
    {
      "epoch": 0.5925607616897506,
      "grad_norm": 0.11699803173542023,
      "learning_rate": 8.871639370376713e-06,
      "loss": 0.0033,
      "step": 92390
    },
    {
      "epoch": 0.5926248985835367,
      "grad_norm": 0.1312762051820755,
      "learning_rate": 8.871285174447127e-06,
      "loss": 0.0038,
      "step": 92400
    },
    {
      "epoch": 0.5926890354773228,
      "grad_norm": 0.12681037187576294,
      "learning_rate": 8.87093093000757e-06,
      "loss": 0.0034,
      "step": 92410
    },
    {
      "epoch": 0.5927531723711089,
      "grad_norm": 0.2736920416355133,
      "learning_rate": 8.870576637062484e-06,
      "loss": 0.0045,
      "step": 92420
    },
    {
      "epoch": 0.592817309264895,
      "grad_norm": 0.15666793286800385,
      "learning_rate": 8.870222295616307e-06,
      "loss": 0.0052,
      "step": 92430
    },
    {
      "epoch": 0.5928814461586811,
      "grad_norm": 0.06761337071657181,
      "learning_rate": 8.869867905673478e-06,
      "loss": 0.0031,
      "step": 92440
    },
    {
      "epoch": 0.5929455830524671,
      "grad_norm": 0.09654685854911804,
      "learning_rate": 8.869513467238437e-06,
      "loss": 0.0037,
      "step": 92450
    },
    {
      "epoch": 0.5930097199462533,
      "grad_norm": 0.14291098713874817,
      "learning_rate": 8.869158980315626e-06,
      "loss": 0.0023,
      "step": 92460
    },
    {
      "epoch": 0.5930738568400393,
      "grad_norm": 0.14262396097183228,
      "learning_rate": 8.868804444909488e-06,
      "loss": 0.0029,
      "step": 92470
    },
    {
      "epoch": 0.5931379937338255,
      "grad_norm": 0.36440786719322205,
      "learning_rate": 8.868449861024468e-06,
      "loss": 0.0028,
      "step": 92480
    },
    {
      "epoch": 0.5932021306276116,
      "grad_norm": 0.10059228539466858,
      "learning_rate": 8.868095228665006e-06,
      "loss": 0.0025,
      "step": 92490
    },
    {
      "epoch": 0.5932662675213977,
      "grad_norm": 0.25650525093078613,
      "learning_rate": 8.867740547835544e-06,
      "loss": 0.0039,
      "step": 92500
    },
    {
      "epoch": 0.5933304044151838,
      "grad_norm": 0.08289292454719543,
      "learning_rate": 8.86738581854053e-06,
      "loss": 0.0026,
      "step": 92510
    },
    {
      "epoch": 0.5933945413089698,
      "grad_norm": 0.13036350905895233,
      "learning_rate": 8.86703104078441e-06,
      "loss": 0.0031,
      "step": 92520
    },
    {
      "epoch": 0.593458678202756,
      "grad_norm": 0.038229234516620636,
      "learning_rate": 8.866676214571623e-06,
      "loss": 0.0049,
      "step": 92530
    },
    {
      "epoch": 0.593522815096542,
      "grad_norm": 0.2246652990579605,
      "learning_rate": 8.866321339906623e-06,
      "loss": 0.0028,
      "step": 92540
    },
    {
      "epoch": 0.5935869519903282,
      "grad_norm": 0.2030942291021347,
      "learning_rate": 8.86596641679385e-06,
      "loss": 0.003,
      "step": 92550
    },
    {
      "epoch": 0.5936510888841142,
      "grad_norm": 0.10550177097320557,
      "learning_rate": 8.865611445237758e-06,
      "loss": 0.0046,
      "step": 92560
    },
    {
      "epoch": 0.5937152257779004,
      "grad_norm": 0.18679168820381165,
      "learning_rate": 8.86525642524279e-06,
      "loss": 0.003,
      "step": 92570
    },
    {
      "epoch": 0.5937793626716864,
      "grad_norm": 0.11270534247159958,
      "learning_rate": 8.864901356813398e-06,
      "loss": 0.0055,
      "step": 92580
    },
    {
      "epoch": 0.5938434995654726,
      "grad_norm": 0.02422017604112625,
      "learning_rate": 8.864546239954028e-06,
      "loss": 0.0026,
      "step": 92590
    },
    {
      "epoch": 0.5939076364592586,
      "grad_norm": 0.16464656591415405,
      "learning_rate": 8.864191074669133e-06,
      "loss": 0.0025,
      "step": 92600
    },
    {
      "epoch": 0.5939717733530447,
      "grad_norm": 0.13741368055343628,
      "learning_rate": 8.863835860963162e-06,
      "loss": 0.0079,
      "step": 92610
    },
    {
      "epoch": 0.5940359102468309,
      "grad_norm": 0.02941173128783703,
      "learning_rate": 8.863480598840565e-06,
      "loss": 0.0016,
      "step": 92620
    },
    {
      "epoch": 0.5941000471406169,
      "grad_norm": 0.048090964555740356,
      "learning_rate": 8.863125288305797e-06,
      "loss": 0.0033,
      "step": 92630
    },
    {
      "epoch": 0.5941641840344031,
      "grad_norm": 0.07795095443725586,
      "learning_rate": 8.862769929363307e-06,
      "loss": 0.0029,
      "step": 92640
    },
    {
      "epoch": 0.5942283209281891,
      "grad_norm": 0.13922512531280518,
      "learning_rate": 8.862414522017549e-06,
      "loss": 0.0019,
      "step": 92650
    },
    {
      "epoch": 0.5942924578219753,
      "grad_norm": 0.23520700633525848,
      "learning_rate": 8.862059066272978e-06,
      "loss": 0.0042,
      "step": 92660
    },
    {
      "epoch": 0.5943565947157613,
      "grad_norm": 0.22586287558078766,
      "learning_rate": 8.861703562134046e-06,
      "loss": 0.0031,
      "step": 92670
    },
    {
      "epoch": 0.5944207316095474,
      "grad_norm": 0.28761547803878784,
      "learning_rate": 8.861348009605207e-06,
      "loss": 0.0025,
      "step": 92680
    },
    {
      "epoch": 0.5944848685033335,
      "grad_norm": 0.05647381395101547,
      "learning_rate": 8.860992408690919e-06,
      "loss": 0.0033,
      "step": 92690
    },
    {
      "epoch": 0.5945490053971196,
      "grad_norm": 0.22699567675590515,
      "learning_rate": 8.860636759395637e-06,
      "loss": 0.0044,
      "step": 92700
    },
    {
      "epoch": 0.5946131422909057,
      "grad_norm": 0.03414135426282883,
      "learning_rate": 8.860281061723816e-06,
      "loss": 0.0037,
      "step": 92710
    },
    {
      "epoch": 0.5946772791846918,
      "grad_norm": 0.08566999435424805,
      "learning_rate": 8.859925315679916e-06,
      "loss": 0.0038,
      "step": 92720
    },
    {
      "epoch": 0.5947414160784779,
      "grad_norm": 0.0905214250087738,
      "learning_rate": 8.859569521268391e-06,
      "loss": 0.0017,
      "step": 92730
    },
    {
      "epoch": 0.594805552972264,
      "grad_norm": 0.142642080783844,
      "learning_rate": 8.859213678493703e-06,
      "loss": 0.0037,
      "step": 92740
    },
    {
      "epoch": 0.59486968986605,
      "grad_norm": 0.05979500338435173,
      "learning_rate": 8.858857787360311e-06,
      "loss": 0.002,
      "step": 92750
    },
    {
      "epoch": 0.5949338267598362,
      "grad_norm": 0.10995723307132721,
      "learning_rate": 8.858501847872671e-06,
      "loss": 0.0032,
      "step": 92760
    },
    {
      "epoch": 0.5949979636536223,
      "grad_norm": 0.18088792264461517,
      "learning_rate": 8.858145860035246e-06,
      "loss": 0.0035,
      "step": 92770
    },
    {
      "epoch": 0.5950621005474084,
      "grad_norm": 0.16171729564666748,
      "learning_rate": 8.857789823852495e-06,
      "loss": 0.0044,
      "step": 92780
    },
    {
      "epoch": 0.5951262374411945,
      "grad_norm": 0.20499688386917114,
      "learning_rate": 8.85743373932888e-06,
      "loss": 0.0031,
      "step": 92790
    },
    {
      "epoch": 0.5951903743349806,
      "grad_norm": 0.3269047141075134,
      "learning_rate": 8.857077606468864e-06,
      "loss": 0.0042,
      "step": 92800
    },
    {
      "epoch": 0.5952545112287667,
      "grad_norm": 0.10731284320354462,
      "learning_rate": 8.856721425276912e-06,
      "loss": 0.0025,
      "step": 92810
    },
    {
      "epoch": 0.5953186481225528,
      "grad_norm": 0.13672778010368347,
      "learning_rate": 8.85636519575748e-06,
      "loss": 0.0024,
      "step": 92820
    },
    {
      "epoch": 0.5953827850163389,
      "grad_norm": 0.07215884327888489,
      "learning_rate": 8.856008917915037e-06,
      "loss": 0.003,
      "step": 92830
    },
    {
      "epoch": 0.5954469219101249,
      "grad_norm": 0.07725408673286438,
      "learning_rate": 8.855652591754047e-06,
      "loss": 0.0037,
      "step": 92840
    },
    {
      "epoch": 0.5955110588039111,
      "grad_norm": 0.10322929918766022,
      "learning_rate": 8.855296217278974e-06,
      "loss": 0.0029,
      "step": 92850
    },
    {
      "epoch": 0.5955751956976971,
      "grad_norm": 0.04164673015475273,
      "learning_rate": 8.854939794494284e-06,
      "loss": 0.0022,
      "step": 92860
    },
    {
      "epoch": 0.5956393325914833,
      "grad_norm": 0.13567589223384857,
      "learning_rate": 8.854583323404443e-06,
      "loss": 0.003,
      "step": 92870
    },
    {
      "epoch": 0.5957034694852693,
      "grad_norm": 0.07293649017810822,
      "learning_rate": 8.854226804013921e-06,
      "loss": 0.0017,
      "step": 92880
    },
    {
      "epoch": 0.5957676063790555,
      "grad_norm": 0.11784832179546356,
      "learning_rate": 8.85387023632718e-06,
      "loss": 0.0052,
      "step": 92890
    },
    {
      "epoch": 0.5958317432728416,
      "grad_norm": 0.3238033056259155,
      "learning_rate": 8.85351362034869e-06,
      "loss": 0.0038,
      "step": 92900
    },
    {
      "epoch": 0.5958958801666276,
      "grad_norm": 0.038372065871953964,
      "learning_rate": 8.853156956082921e-06,
      "loss": 0.0065,
      "step": 92910
    },
    {
      "epoch": 0.5959600170604138,
      "grad_norm": 0.09675122052431107,
      "learning_rate": 8.852800243534343e-06,
      "loss": 0.004,
      "step": 92920
    },
    {
      "epoch": 0.5960241539541998,
      "grad_norm": 0.15376578271389008,
      "learning_rate": 8.852443482707423e-06,
      "loss": 0.0031,
      "step": 92930
    },
    {
      "epoch": 0.596088290847986,
      "grad_norm": 0.1450980007648468,
      "learning_rate": 8.852086673606634e-06,
      "loss": 0.0034,
      "step": 92940
    },
    {
      "epoch": 0.596152427741772,
      "grad_norm": 0.1521768420934677,
      "learning_rate": 8.851729816236445e-06,
      "loss": 0.0038,
      "step": 92950
    },
    {
      "epoch": 0.5962165646355582,
      "grad_norm": 0.1504070907831192,
      "learning_rate": 8.851372910601328e-06,
      "loss": 0.0043,
      "step": 92960
    },
    {
      "epoch": 0.5962807015293442,
      "grad_norm": 0.09424560517072678,
      "learning_rate": 8.851015956705757e-06,
      "loss": 0.0036,
      "step": 92970
    },
    {
      "epoch": 0.5963448384231304,
      "grad_norm": 0.09183598309755325,
      "learning_rate": 8.850658954554203e-06,
      "loss": 0.0024,
      "step": 92980
    },
    {
      "epoch": 0.5964089753169164,
      "grad_norm": 0.08142650872468948,
      "learning_rate": 8.85030190415114e-06,
      "loss": 0.0041,
      "step": 92990
    },
    {
      "epoch": 0.5964731122107025,
      "grad_norm": 0.05937165766954422,
      "learning_rate": 8.849944805501043e-06,
      "loss": 0.0021,
      "step": 93000
    },
    {
      "epoch": 0.5965372491044886,
      "grad_norm": 0.4259531497955322,
      "learning_rate": 8.849587658608386e-06,
      "loss": 0.003,
      "step": 93010
    },
    {
      "epoch": 0.5966013859982747,
      "grad_norm": 0.11123054474592209,
      "learning_rate": 8.849230463477645e-06,
      "loss": 0.0033,
      "step": 93020
    },
    {
      "epoch": 0.5966655228920608,
      "grad_norm": 0.21839596331119537,
      "learning_rate": 8.848873220113294e-06,
      "loss": 0.0018,
      "step": 93030
    },
    {
      "epoch": 0.5967296597858469,
      "grad_norm": 0.053809747099876404,
      "learning_rate": 8.848515928519812e-06,
      "loss": 0.0019,
      "step": 93040
    },
    {
      "epoch": 0.5967937966796331,
      "grad_norm": 0.11655854433774948,
      "learning_rate": 8.848158588701674e-06,
      "loss": 0.0037,
      "step": 93050
    },
    {
      "epoch": 0.5968579335734191,
      "grad_norm": 0.22731737792491913,
      "learning_rate": 8.84780120066336e-06,
      "loss": 0.0033,
      "step": 93060
    },
    {
      "epoch": 0.5969220704672052,
      "grad_norm": 0.07772057503461838,
      "learning_rate": 8.847443764409344e-06,
      "loss": 0.0052,
      "step": 93070
    },
    {
      "epoch": 0.5969862073609913,
      "grad_norm": 0.11905039846897125,
      "learning_rate": 8.847086279944112e-06,
      "loss": 0.0026,
      "step": 93080
    },
    {
      "epoch": 0.5970503442547774,
      "grad_norm": 0.23587697744369507,
      "learning_rate": 8.846728747272137e-06,
      "loss": 0.0024,
      "step": 93090
    },
    {
      "epoch": 0.5971144811485635,
      "grad_norm": 0.02773495949804783,
      "learning_rate": 8.846371166397903e-06,
      "loss": 0.0044,
      "step": 93100
    },
    {
      "epoch": 0.5971786180423496,
      "grad_norm": 0.4221494197845459,
      "learning_rate": 8.846013537325887e-06,
      "loss": 0.0048,
      "step": 93110
    },
    {
      "epoch": 0.5972427549361357,
      "grad_norm": 0.11520206928253174,
      "learning_rate": 8.845655860060574e-06,
      "loss": 0.0031,
      "step": 93120
    },
    {
      "epoch": 0.5973068918299218,
      "grad_norm": 0.22873573005199432,
      "learning_rate": 8.845298134606445e-06,
      "loss": 0.0016,
      "step": 93130
    },
    {
      "epoch": 0.5973710287237078,
      "grad_norm": 0.3438142240047455,
      "learning_rate": 8.844940360967981e-06,
      "loss": 0.0033,
      "step": 93140
    },
    {
      "epoch": 0.597435165617494,
      "grad_norm": 0.10318133234977722,
      "learning_rate": 8.844582539149667e-06,
      "loss": 0.0025,
      "step": 93150
    },
    {
      "epoch": 0.59749930251128,
      "grad_norm": 0.15780027210712433,
      "learning_rate": 8.844224669155986e-06,
      "loss": 0.004,
      "step": 93160
    },
    {
      "epoch": 0.5975634394050662,
      "grad_norm": 0.09695665538311005,
      "learning_rate": 8.843866750991424e-06,
      "loss": 0.0043,
      "step": 93170
    },
    {
      "epoch": 0.5976275762988523,
      "grad_norm": 0.03867492824792862,
      "learning_rate": 8.843508784660461e-06,
      "loss": 0.0021,
      "step": 93180
    },
    {
      "epoch": 0.5976917131926384,
      "grad_norm": 0.07050751894712448,
      "learning_rate": 8.843150770167589e-06,
      "loss": 0.0034,
      "step": 93190
    },
    {
      "epoch": 0.5977558500864245,
      "grad_norm": 0.09821964800357819,
      "learning_rate": 8.84279270751729e-06,
      "loss": 0.0024,
      "step": 93200
    },
    {
      "epoch": 0.5978199869802105,
      "grad_norm": 0.05456443503499031,
      "learning_rate": 8.842434596714054e-06,
      "loss": 0.0023,
      "step": 93210
    },
    {
      "epoch": 0.5978841238739967,
      "grad_norm": 0.3301343619823456,
      "learning_rate": 8.842076437762364e-06,
      "loss": 0.0044,
      "step": 93220
    },
    {
      "epoch": 0.5979482607677827,
      "grad_norm": 0.11671298742294312,
      "learning_rate": 8.84171823066671e-06,
      "loss": 0.0039,
      "step": 93230
    },
    {
      "epoch": 0.5980123976615689,
      "grad_norm": 0.05689849331974983,
      "learning_rate": 8.841359975431583e-06,
      "loss": 0.0028,
      "step": 93240
    },
    {
      "epoch": 0.5980765345553549,
      "grad_norm": 0.02990148402750492,
      "learning_rate": 8.841001672061468e-06,
      "loss": 0.0031,
      "step": 93250
    },
    {
      "epoch": 0.5981406714491411,
      "grad_norm": 0.05530443415045738,
      "learning_rate": 8.84064332056086e-06,
      "loss": 0.0029,
      "step": 93260
    },
    {
      "epoch": 0.5982048083429271,
      "grad_norm": 0.10282295942306519,
      "learning_rate": 8.840284920934243e-06,
      "loss": 0.0052,
      "step": 93270
    },
    {
      "epoch": 0.5982689452367133,
      "grad_norm": 0.07693000882863998,
      "learning_rate": 8.839926473186114e-06,
      "loss": 0.0041,
      "step": 93280
    },
    {
      "epoch": 0.5983330821304993,
      "grad_norm": 0.11836773157119751,
      "learning_rate": 8.83956797732096e-06,
      "loss": 0.0033,
      "step": 93290
    },
    {
      "epoch": 0.5983972190242854,
      "grad_norm": 0.11034104973077774,
      "learning_rate": 8.839209433343273e-06,
      "loss": 0.0036,
      "step": 93300
    },
    {
      "epoch": 0.5984613559180715,
      "grad_norm": 0.11941079050302505,
      "learning_rate": 8.838850841257552e-06,
      "loss": 0.0018,
      "step": 93310
    },
    {
      "epoch": 0.5985254928118576,
      "grad_norm": 0.17596879601478577,
      "learning_rate": 8.838492201068285e-06,
      "loss": 0.0038,
      "step": 93320
    },
    {
      "epoch": 0.5985896297056438,
      "grad_norm": 0.004393393639475107,
      "learning_rate": 8.838133512779968e-06,
      "loss": 0.0042,
      "step": 93330
    },
    {
      "epoch": 0.5986537665994298,
      "grad_norm": 0.3173533082008362,
      "learning_rate": 8.837774776397095e-06,
      "loss": 0.0031,
      "step": 93340
    },
    {
      "epoch": 0.598717903493216,
      "grad_norm": 0.11862670630216599,
      "learning_rate": 8.83741599192416e-06,
      "loss": 0.0034,
      "step": 93350
    },
    {
      "epoch": 0.598782040387002,
      "grad_norm": 0.10155455023050308,
      "learning_rate": 8.83705715936566e-06,
      "loss": 0.0021,
      "step": 93360
    },
    {
      "epoch": 0.5988461772807882,
      "grad_norm": 0.15258212387561798,
      "learning_rate": 8.836698278726092e-06,
      "loss": 0.0029,
      "step": 93370
    },
    {
      "epoch": 0.5989103141745742,
      "grad_norm": 0.11390714347362518,
      "learning_rate": 8.836339350009954e-06,
      "loss": 0.0024,
      "step": 93380
    },
    {
      "epoch": 0.5989744510683603,
      "grad_norm": 0.0799383893609047,
      "learning_rate": 8.835980373221741e-06,
      "loss": 0.002,
      "step": 93390
    },
    {
      "epoch": 0.5990385879621464,
      "grad_norm": 0.10548446327447891,
      "learning_rate": 8.835621348365952e-06,
      "loss": 0.0022,
      "step": 93400
    },
    {
      "epoch": 0.5991027248559325,
      "grad_norm": 0.26774832606315613,
      "learning_rate": 8.835262275447087e-06,
      "loss": 0.0049,
      "step": 93410
    },
    {
      "epoch": 0.5991668617497186,
      "grad_norm": 0.18672409653663635,
      "learning_rate": 8.834903154469643e-06,
      "loss": 0.0019,
      "step": 93420
    },
    {
      "epoch": 0.5992309986435047,
      "grad_norm": 0.16137036681175232,
      "learning_rate": 8.834543985438125e-06,
      "loss": 0.0043,
      "step": 93430
    },
    {
      "epoch": 0.5992951355372907,
      "grad_norm": 0.08167710155248642,
      "learning_rate": 8.834184768357028e-06,
      "loss": 0.0033,
      "step": 93440
    },
    {
      "epoch": 0.5993592724310769,
      "grad_norm": 0.14367537200450897,
      "learning_rate": 8.833825503230854e-06,
      "loss": 0.0043,
      "step": 93450
    },
    {
      "epoch": 0.5994234093248629,
      "grad_norm": 0.0750400498509407,
      "learning_rate": 8.83346619006411e-06,
      "loss": 0.0022,
      "step": 93460
    },
    {
      "epoch": 0.5994875462186491,
      "grad_norm": 0.6188009977340698,
      "learning_rate": 8.833106828861294e-06,
      "loss": 0.0034,
      "step": 93470
    },
    {
      "epoch": 0.5995516831124352,
      "grad_norm": 0.3307935893535614,
      "learning_rate": 8.832747419626908e-06,
      "loss": 0.0049,
      "step": 93480
    },
    {
      "epoch": 0.5996158200062213,
      "grad_norm": 0.10426060855388641,
      "learning_rate": 8.832387962365458e-06,
      "loss": 0.0031,
      "step": 93490
    },
    {
      "epoch": 0.5996799569000074,
      "grad_norm": 0.298533171415329,
      "learning_rate": 8.83202845708145e-06,
      "loss": 0.0052,
      "step": 93500
    },
    {
      "epoch": 0.5997440937937935,
      "grad_norm": 0.13313566148281097,
      "learning_rate": 8.831668903779384e-06,
      "loss": 0.0025,
      "step": 93510
    },
    {
      "epoch": 0.5998082306875796,
      "grad_norm": 0.08680471777915955,
      "learning_rate": 8.83130930246377e-06,
      "loss": 0.0048,
      "step": 93520
    },
    {
      "epoch": 0.5998723675813656,
      "grad_norm": 0.07953489571809769,
      "learning_rate": 8.83094965313911e-06,
      "loss": 0.0033,
      "step": 93530
    },
    {
      "epoch": 0.5999365044751518,
      "grad_norm": 0.1352090984582901,
      "learning_rate": 8.830589955809916e-06,
      "loss": 0.0029,
      "step": 93540
    },
    {
      "epoch": 0.6000006413689378,
      "grad_norm": 0.1552029699087143,
      "learning_rate": 8.830230210480692e-06,
      "loss": 0.0024,
      "step": 93550
    },
    {
      "epoch": 0.600064778262724,
      "grad_norm": 0.2718900442123413,
      "learning_rate": 8.829870417155944e-06,
      "loss": 0.0027,
      "step": 93560
    },
    {
      "epoch": 0.60012891515651,
      "grad_norm": 0.14161957800388336,
      "learning_rate": 8.829510575840184e-06,
      "loss": 0.0027,
      "step": 93570
    },
    {
      "epoch": 0.6001930520502962,
      "grad_norm": 0.22129161655902863,
      "learning_rate": 8.829150686537919e-06,
      "loss": 0.0022,
      "step": 93580
    },
    {
      "epoch": 0.6002571889440822,
      "grad_norm": 0.1630723923444748,
      "learning_rate": 8.82879074925366e-06,
      "loss": 0.0035,
      "step": 93590
    },
    {
      "epoch": 0.6003213258378683,
      "grad_norm": 0.06927991658449173,
      "learning_rate": 8.828430763991916e-06,
      "loss": 0.0018,
      "step": 93600
    },
    {
      "epoch": 0.6003854627316545,
      "grad_norm": 0.14366739988327026,
      "learning_rate": 8.828070730757196e-06,
      "loss": 0.0018,
      "step": 93610
    },
    {
      "epoch": 0.6004495996254405,
      "grad_norm": 0.12288960069417953,
      "learning_rate": 8.827710649554018e-06,
      "loss": 0.0022,
      "step": 93620
    },
    {
      "epoch": 0.6005137365192267,
      "grad_norm": 0.13222576677799225,
      "learning_rate": 8.827350520386886e-06,
      "loss": 0.0037,
      "step": 93630
    },
    {
      "epoch": 0.6005778734130127,
      "grad_norm": 0.0924672782421112,
      "learning_rate": 8.82699034326032e-06,
      "loss": 0.0043,
      "step": 93640
    },
    {
      "epoch": 0.6006420103067989,
      "grad_norm": 0.11825573444366455,
      "learning_rate": 8.826630118178828e-06,
      "loss": 0.003,
      "step": 93650
    },
    {
      "epoch": 0.6007061472005849,
      "grad_norm": 0.11818478256464005,
      "learning_rate": 8.826269845146926e-06,
      "loss": 0.0018,
      "step": 93660
    },
    {
      "epoch": 0.600770284094371,
      "grad_norm": 0.12195321172475815,
      "learning_rate": 8.825909524169129e-06,
      "loss": 0.003,
      "step": 93670
    },
    {
      "epoch": 0.6008344209881571,
      "grad_norm": 0.081720270216465,
      "learning_rate": 8.825549155249951e-06,
      "loss": 0.0026,
      "step": 93680
    },
    {
      "epoch": 0.6008985578819432,
      "grad_norm": 0.08222363889217377,
      "learning_rate": 8.825188738393908e-06,
      "loss": 0.0022,
      "step": 93690
    },
    {
      "epoch": 0.6009626947757293,
      "grad_norm": 0.1412651687860489,
      "learning_rate": 8.824828273605515e-06,
      "loss": 0.0057,
      "step": 93700
    },
    {
      "epoch": 0.6010268316695154,
      "grad_norm": 0.09899317473173141,
      "learning_rate": 8.824467760889291e-06,
      "loss": 0.0043,
      "step": 93710
    },
    {
      "epoch": 0.6010909685633015,
      "grad_norm": 0.23038393259048462,
      "learning_rate": 8.824107200249754e-06,
      "loss": 0.002,
      "step": 93720
    },
    {
      "epoch": 0.6011551054570876,
      "grad_norm": 0.26497548818588257,
      "learning_rate": 8.82374659169142e-06,
      "loss": 0.0046,
      "step": 93730
    },
    {
      "epoch": 0.6012192423508737,
      "grad_norm": 0.36057302355766296,
      "learning_rate": 8.823385935218806e-06,
      "loss": 0.0032,
      "step": 93740
    },
    {
      "epoch": 0.6012833792446598,
      "grad_norm": 0.16276083886623383,
      "learning_rate": 8.823025230836438e-06,
      "loss": 0.0034,
      "step": 93750
    },
    {
      "epoch": 0.601347516138446,
      "grad_norm": 0.12810267508029938,
      "learning_rate": 8.822664478548829e-06,
      "loss": 0.0044,
      "step": 93760
    },
    {
      "epoch": 0.601411653032232,
      "grad_norm": 0.08066945523023605,
      "learning_rate": 8.822303678360502e-06,
      "loss": 0.0029,
      "step": 93770
    },
    {
      "epoch": 0.6014757899260181,
      "grad_norm": 0.3909737169742584,
      "learning_rate": 8.821942830275978e-06,
      "loss": 0.0021,
      "step": 93780
    },
    {
      "epoch": 0.6015399268198042,
      "grad_norm": 0.33293431997299194,
      "learning_rate": 8.82158193429978e-06,
      "loss": 0.0043,
      "step": 93790
    },
    {
      "epoch": 0.6016040637135903,
      "grad_norm": 0.16712966561317444,
      "learning_rate": 8.821220990436427e-06,
      "loss": 0.0035,
      "step": 93800
    },
    {
      "epoch": 0.6016682006073764,
      "grad_norm": 0.20190496742725372,
      "learning_rate": 8.820859998690448e-06,
      "loss": 0.0027,
      "step": 93810
    },
    {
      "epoch": 0.6017323375011625,
      "grad_norm": 0.25855904817581177,
      "learning_rate": 8.820498959066359e-06,
      "loss": 0.0118,
      "step": 93820
    },
    {
      "epoch": 0.6017964743949485,
      "grad_norm": 0.06966379284858704,
      "learning_rate": 8.820137871568688e-06,
      "loss": 0.005,
      "step": 93830
    },
    {
      "epoch": 0.6018606112887347,
      "grad_norm": 0.34347179532051086,
      "learning_rate": 8.81977673620196e-06,
      "loss": 0.0019,
      "step": 93840
    },
    {
      "epoch": 0.6019247481825207,
      "grad_norm": 0.5182726979255676,
      "learning_rate": 8.8194155529707e-06,
      "loss": 0.0027,
      "step": 93850
    },
    {
      "epoch": 0.6019888850763069,
      "grad_norm": 0.09111955761909485,
      "learning_rate": 8.819054321879433e-06,
      "loss": 0.0046,
      "step": 93860
    },
    {
      "epoch": 0.6020530219700929,
      "grad_norm": 0.16851943731307983,
      "learning_rate": 8.818693042932685e-06,
      "loss": 0.0019,
      "step": 93870
    },
    {
      "epoch": 0.6021171588638791,
      "grad_norm": 0.07118560373783112,
      "learning_rate": 8.818331716134984e-06,
      "loss": 0.0021,
      "step": 93880
    },
    {
      "epoch": 0.6021812957576652,
      "grad_norm": 0.10934966057538986,
      "learning_rate": 8.817970341490859e-06,
      "loss": 0.0029,
      "step": 93890
    },
    {
      "epoch": 0.6022454326514513,
      "grad_norm": 0.05146744102239609,
      "learning_rate": 8.817608919004836e-06,
      "loss": 0.0028,
      "step": 93900
    },
    {
      "epoch": 0.6023095695452374,
      "grad_norm": 0.05124472454190254,
      "learning_rate": 8.817247448681446e-06,
      "loss": 0.0025,
      "step": 93910
    },
    {
      "epoch": 0.6023737064390234,
      "grad_norm": 0.16353295743465424,
      "learning_rate": 8.816885930525216e-06,
      "loss": 0.0054,
      "step": 93920
    },
    {
      "epoch": 0.6024378433328096,
      "grad_norm": 0.1566086709499359,
      "learning_rate": 8.816524364540678e-06,
      "loss": 0.0065,
      "step": 93930
    },
    {
      "epoch": 0.6025019802265956,
      "grad_norm": 0.21201664209365845,
      "learning_rate": 8.816162750732362e-06,
      "loss": 0.0037,
      "step": 93940
    },
    {
      "epoch": 0.6025661171203818,
      "grad_norm": 0.3268226683139801,
      "learning_rate": 8.815801089104799e-06,
      "loss": 0.0013,
      "step": 93950
    },
    {
      "epoch": 0.6026302540141678,
      "grad_norm": 0.042904358357191086,
      "learning_rate": 8.815439379662522e-06,
      "loss": 0.0049,
      "step": 93960
    },
    {
      "epoch": 0.602694390907954,
      "grad_norm": 0.08511478453874588,
      "learning_rate": 8.815077622410062e-06,
      "loss": 0.0024,
      "step": 93970
    },
    {
      "epoch": 0.60275852780174,
      "grad_norm": 0.07031531631946564,
      "learning_rate": 8.814715817351954e-06,
      "loss": 0.005,
      "step": 93980
    },
    {
      "epoch": 0.6028226646955261,
      "grad_norm": 0.22578291594982147,
      "learning_rate": 8.814353964492729e-06,
      "loss": 0.0028,
      "step": 93990
    },
    {
      "epoch": 0.6028868015893122,
      "grad_norm": 0.07537861168384552,
      "learning_rate": 8.813992063836923e-06,
      "loss": 0.0037,
      "step": 94000
    },
    {
      "epoch": 0.6029509384830983,
      "grad_norm": 0.3766474723815918,
      "learning_rate": 8.813630115389071e-06,
      "loss": 0.0022,
      "step": 94010
    },
    {
      "epoch": 0.6030150753768844,
      "grad_norm": 0.09774509072303772,
      "learning_rate": 8.813268119153709e-06,
      "loss": 0.0027,
      "step": 94020
    },
    {
      "epoch": 0.6030792122706705,
      "grad_norm": 0.1362939476966858,
      "learning_rate": 8.81290607513537e-06,
      "loss": 0.0029,
      "step": 94030
    },
    {
      "epoch": 0.6031433491644567,
      "grad_norm": 0.14567211270332336,
      "learning_rate": 8.812543983338595e-06,
      "loss": 0.0033,
      "step": 94040
    },
    {
      "epoch": 0.6032074860582427,
      "grad_norm": 0.3329886496067047,
      "learning_rate": 8.812181843767918e-06,
      "loss": 0.0042,
      "step": 94050
    },
    {
      "epoch": 0.6032716229520289,
      "grad_norm": 0.1347803920507431,
      "learning_rate": 8.811819656427877e-06,
      "loss": 0.0019,
      "step": 94060
    },
    {
      "epoch": 0.6033357598458149,
      "grad_norm": 0.058834757655858994,
      "learning_rate": 8.811457421323013e-06,
      "loss": 0.003,
      "step": 94070
    },
    {
      "epoch": 0.603399896739601,
      "grad_norm": 0.1033516675233841,
      "learning_rate": 8.811095138457863e-06,
      "loss": 0.0032,
      "step": 94080
    },
    {
      "epoch": 0.6034640336333871,
      "grad_norm": 0.1391102820634842,
      "learning_rate": 8.810732807836968e-06,
      "loss": 0.0043,
      "step": 94090
    },
    {
      "epoch": 0.6035281705271732,
      "grad_norm": 0.14105257391929626,
      "learning_rate": 8.810370429464867e-06,
      "loss": 0.0034,
      "step": 94100
    },
    {
      "epoch": 0.6035923074209593,
      "grad_norm": 0.1752769649028778,
      "learning_rate": 8.8100080033461e-06,
      "loss": 0.0022,
      "step": 94110
    },
    {
      "epoch": 0.6036564443147454,
      "grad_norm": 0.07758922874927521,
      "learning_rate": 8.80964552948521e-06,
      "loss": 0.003,
      "step": 94120
    },
    {
      "epoch": 0.6037205812085314,
      "grad_norm": 0.17690975964069366,
      "learning_rate": 8.80928300788674e-06,
      "loss": 0.0024,
      "step": 94130
    },
    {
      "epoch": 0.6037847181023176,
      "grad_norm": 0.444504976272583,
      "learning_rate": 8.808920438555231e-06,
      "loss": 0.0025,
      "step": 94140
    },
    {
      "epoch": 0.6038488549961036,
      "grad_norm": 0.20921805500984192,
      "learning_rate": 8.808557821495227e-06,
      "loss": 0.0041,
      "step": 94150
    },
    {
      "epoch": 0.6039129918898898,
      "grad_norm": 0.03435481712222099,
      "learning_rate": 8.808195156711273e-06,
      "loss": 0.002,
      "step": 94160
    },
    {
      "epoch": 0.6039771287836759,
      "grad_norm": 0.24317345023155212,
      "learning_rate": 8.80783244420791e-06,
      "loss": 0.0031,
      "step": 94170
    },
    {
      "epoch": 0.604041265677462,
      "grad_norm": 0.07621653378009796,
      "learning_rate": 8.807469683989685e-06,
      "loss": 0.0053,
      "step": 94180
    },
    {
      "epoch": 0.6041054025712481,
      "grad_norm": 0.04928126186132431,
      "learning_rate": 8.807106876061143e-06,
      "loss": 0.0032,
      "step": 94190
    },
    {
      "epoch": 0.6041695394650342,
      "grad_norm": 0.0837152749300003,
      "learning_rate": 8.806744020426832e-06,
      "loss": 0.0026,
      "step": 94200
    },
    {
      "epoch": 0.6042336763588203,
      "grad_norm": 0.04882385581731796,
      "learning_rate": 8.806381117091298e-06,
      "loss": 0.0034,
      "step": 94210
    },
    {
      "epoch": 0.6042978132526063,
      "grad_norm": 0.1601816862821579,
      "learning_rate": 8.806018166059087e-06,
      "loss": 0.0052,
      "step": 94220
    },
    {
      "epoch": 0.6043619501463925,
      "grad_norm": 0.655596911907196,
      "learning_rate": 8.80565516733475e-06,
      "loss": 0.0037,
      "step": 94230
    },
    {
      "epoch": 0.6044260870401785,
      "grad_norm": 0.25453537702560425,
      "learning_rate": 8.805292120922832e-06,
      "loss": 0.0034,
      "step": 94240
    },
    {
      "epoch": 0.6044902239339647,
      "grad_norm": 0.19453126192092896,
      "learning_rate": 8.804929026827887e-06,
      "loss": 0.0021,
      "step": 94250
    },
    {
      "epoch": 0.6045543608277507,
      "grad_norm": 0.1085091158747673,
      "learning_rate": 8.804565885054458e-06,
      "loss": 0.0021,
      "step": 94260
    },
    {
      "epoch": 0.6046184977215369,
      "grad_norm": 0.22164227068424225,
      "learning_rate": 8.804202695607102e-06,
      "loss": 0.0035,
      "step": 94270
    },
    {
      "epoch": 0.6046826346153229,
      "grad_norm": 0.13246764242649078,
      "learning_rate": 8.803839458490368e-06,
      "loss": 0.005,
      "step": 94280
    },
    {
      "epoch": 0.604746771509109,
      "grad_norm": 0.23474083840847015,
      "learning_rate": 8.803476173708806e-06,
      "loss": 0.0026,
      "step": 94290
    },
    {
      "epoch": 0.6048109084028951,
      "grad_norm": 0.01400019135326147,
      "learning_rate": 8.80311284126697e-06,
      "loss": 0.0025,
      "step": 94300
    },
    {
      "epoch": 0.6048750452966812,
      "grad_norm": 0.11559020727872849,
      "learning_rate": 8.802749461169411e-06,
      "loss": 0.0021,
      "step": 94310
    },
    {
      "epoch": 0.6049391821904674,
      "grad_norm": 0.08287182450294495,
      "learning_rate": 8.802386033420682e-06,
      "loss": 0.0028,
      "step": 94320
    },
    {
      "epoch": 0.6050033190842534,
      "grad_norm": 0.10514973849058151,
      "learning_rate": 8.802022558025341e-06,
      "loss": 0.0022,
      "step": 94330
    },
    {
      "epoch": 0.6050674559780396,
      "grad_norm": 0.3619040846824646,
      "learning_rate": 8.80165903498794e-06,
      "loss": 0.0036,
      "step": 94340
    },
    {
      "epoch": 0.6051315928718256,
      "grad_norm": 0.09895238280296326,
      "learning_rate": 8.801295464313032e-06,
      "loss": 0.0032,
      "step": 94350
    },
    {
      "epoch": 0.6051957297656118,
      "grad_norm": 0.035954151302576065,
      "learning_rate": 8.800931846005178e-06,
      "loss": 0.0038,
      "step": 94360
    },
    {
      "epoch": 0.6052598666593978,
      "grad_norm": 0.08849111944437027,
      "learning_rate": 8.80056818006893e-06,
      "loss": 0.0021,
      "step": 94370
    },
    {
      "epoch": 0.6053240035531839,
      "grad_norm": 0.05876295641064644,
      "learning_rate": 8.800204466508846e-06,
      "loss": 0.0021,
      "step": 94380
    },
    {
      "epoch": 0.60538814044697,
      "grad_norm": 0.07266619056463242,
      "learning_rate": 8.799840705329486e-06,
      "loss": 0.0042,
      "step": 94390
    },
    {
      "epoch": 0.6054522773407561,
      "grad_norm": 0.1223178282380104,
      "learning_rate": 8.799476896535403e-06,
      "loss": 0.0021,
      "step": 94400
    },
    {
      "epoch": 0.6055164142345422,
      "grad_norm": 0.10297498852014542,
      "learning_rate": 8.799113040131161e-06,
      "loss": 0.004,
      "step": 94410
    },
    {
      "epoch": 0.6055805511283283,
      "grad_norm": 0.1572066694498062,
      "learning_rate": 8.798749136121318e-06,
      "loss": 0.0046,
      "step": 94420
    },
    {
      "epoch": 0.6056446880221144,
      "grad_norm": 0.10034254193305969,
      "learning_rate": 8.798385184510435e-06,
      "loss": 0.0038,
      "step": 94430
    },
    {
      "epoch": 0.6057088249159005,
      "grad_norm": 0.3818894624710083,
      "learning_rate": 8.798021185303067e-06,
      "loss": 0.0043,
      "step": 94440
    },
    {
      "epoch": 0.6057729618096866,
      "grad_norm": 0.12883812189102173,
      "learning_rate": 8.797657138503781e-06,
      "loss": 0.0015,
      "step": 94450
    },
    {
      "epoch": 0.6058370987034727,
      "grad_norm": 0.18104223906993866,
      "learning_rate": 8.797293044117137e-06,
      "loss": 0.0026,
      "step": 94460
    },
    {
      "epoch": 0.6059012355972588,
      "grad_norm": 0.05865728482604027,
      "learning_rate": 8.796928902147698e-06,
      "loss": 0.0022,
      "step": 94470
    },
    {
      "epoch": 0.6059653724910449,
      "grad_norm": 0.146976038813591,
      "learning_rate": 8.796564712600024e-06,
      "loss": 0.0052,
      "step": 94480
    },
    {
      "epoch": 0.606029509384831,
      "grad_norm": 0.1646391898393631,
      "learning_rate": 8.796200475478683e-06,
      "loss": 0.0033,
      "step": 94490
    },
    {
      "epoch": 0.6060936462786171,
      "grad_norm": 0.1387776881456375,
      "learning_rate": 8.795836190788237e-06,
      "loss": 0.0026,
      "step": 94500
    },
    {
      "epoch": 0.6061577831724032,
      "grad_norm": 0.18178357183933258,
      "learning_rate": 8.795471858533249e-06,
      "loss": 0.0032,
      "step": 94510
    },
    {
      "epoch": 0.6062219200661892,
      "grad_norm": 0.27202317118644714,
      "learning_rate": 8.795107478718287e-06,
      "loss": 0.0034,
      "step": 94520
    },
    {
      "epoch": 0.6062860569599754,
      "grad_norm": 0.19006876647472382,
      "learning_rate": 8.794743051347916e-06,
      "loss": 0.0031,
      "step": 94530
    },
    {
      "epoch": 0.6063501938537614,
      "grad_norm": 0.07087966054677963,
      "learning_rate": 8.794378576426702e-06,
      "loss": 0.003,
      "step": 94540
    },
    {
      "epoch": 0.6064143307475476,
      "grad_norm": 0.008801287040114403,
      "learning_rate": 8.794014053959211e-06,
      "loss": 0.0024,
      "step": 94550
    },
    {
      "epoch": 0.6064784676413336,
      "grad_norm": 0.11481478065252304,
      "learning_rate": 8.793649483950015e-06,
      "loss": 0.0055,
      "step": 94560
    },
    {
      "epoch": 0.6065426045351198,
      "grad_norm": 0.1139615997672081,
      "learning_rate": 8.79328486640368e-06,
      "loss": 0.0018,
      "step": 94570
    },
    {
      "epoch": 0.6066067414289058,
      "grad_norm": 0.09423598647117615,
      "learning_rate": 8.792920201324773e-06,
      "loss": 0.0047,
      "step": 94580
    },
    {
      "epoch": 0.606670878322692,
      "grad_norm": 0.12620431184768677,
      "learning_rate": 8.792555488717866e-06,
      "loss": 0.0114,
      "step": 94590
    },
    {
      "epoch": 0.6067350152164781,
      "grad_norm": 0.12396939843893051,
      "learning_rate": 8.792190728587529e-06,
      "loss": 0.0026,
      "step": 94600
    },
    {
      "epoch": 0.6067991521102641,
      "grad_norm": 0.4281218349933624,
      "learning_rate": 8.79182592093833e-06,
      "loss": 0.0061,
      "step": 94610
    },
    {
      "epoch": 0.6068632890040503,
      "grad_norm": 0.07776124775409698,
      "learning_rate": 8.791461065774843e-06,
      "loss": 0.0019,
      "step": 94620
    },
    {
      "epoch": 0.6069274258978363,
      "grad_norm": 0.07456205040216446,
      "learning_rate": 8.791096163101639e-06,
      "loss": 0.004,
      "step": 94630
    },
    {
      "epoch": 0.6069915627916225,
      "grad_norm": 0.12895146012306213,
      "learning_rate": 8.790731212923292e-06,
      "loss": 0.0021,
      "step": 94640
    },
    {
      "epoch": 0.6070556996854085,
      "grad_norm": 0.13422219455242157,
      "learning_rate": 8.790366215244372e-06,
      "loss": 0.0028,
      "step": 94650
    },
    {
      "epoch": 0.6071198365791947,
      "grad_norm": 0.15804046392440796,
      "learning_rate": 8.790001170069454e-06,
      "loss": 0.0033,
      "step": 94660
    },
    {
      "epoch": 0.6071839734729807,
      "grad_norm": 0.04166124016046524,
      "learning_rate": 8.789636077403114e-06,
      "loss": 0.0042,
      "step": 94670
    },
    {
      "epoch": 0.6072481103667668,
      "grad_norm": 0.06918887048959732,
      "learning_rate": 8.789270937249925e-06,
      "loss": 0.0025,
      "step": 94680
    },
    {
      "epoch": 0.6073122472605529,
      "grad_norm": 0.0580136813223362,
      "learning_rate": 8.788905749614463e-06,
      "loss": 0.0026,
      "step": 94690
    },
    {
      "epoch": 0.607376384154339,
      "grad_norm": 0.1450560837984085,
      "learning_rate": 8.788540514501305e-06,
      "loss": 0.0037,
      "step": 94700
    },
    {
      "epoch": 0.6074405210481251,
      "grad_norm": 0.1259308159351349,
      "learning_rate": 8.788175231915026e-06,
      "loss": 0.0035,
      "step": 94710
    },
    {
      "epoch": 0.6075046579419112,
      "grad_norm": 0.10257034003734589,
      "learning_rate": 8.787809901860203e-06,
      "loss": 0.0022,
      "step": 94720
    },
    {
      "epoch": 0.6075687948356974,
      "grad_norm": 0.3289741277694702,
      "learning_rate": 8.787444524341414e-06,
      "loss": 0.0034,
      "step": 94730
    },
    {
      "epoch": 0.6076329317294834,
      "grad_norm": 0.13260440528392792,
      "learning_rate": 8.787079099363241e-06,
      "loss": 0.0027,
      "step": 94740
    },
    {
      "epoch": 0.6076970686232696,
      "grad_norm": 0.176558718085289,
      "learning_rate": 8.78671362693026e-06,
      "loss": 0.0046,
      "step": 94750
    },
    {
      "epoch": 0.6077612055170556,
      "grad_norm": 0.023784136399626732,
      "learning_rate": 8.78634810704705e-06,
      "loss": 0.0038,
      "step": 94760
    },
    {
      "epoch": 0.6078253424108417,
      "grad_norm": 0.09431969374418259,
      "learning_rate": 8.785982539718194e-06,
      "loss": 0.0045,
      "step": 94770
    },
    {
      "epoch": 0.6078894793046278,
      "grad_norm": 0.10205917805433273,
      "learning_rate": 8.785616924948269e-06,
      "loss": 0.0038,
      "step": 94780
    },
    {
      "epoch": 0.6079536161984139,
      "grad_norm": 0.2259370982646942,
      "learning_rate": 8.785251262741858e-06,
      "loss": 0.0039,
      "step": 94790
    },
    {
      "epoch": 0.6080177530922,
      "grad_norm": 0.021746670827269554,
      "learning_rate": 8.784885553103543e-06,
      "loss": 0.0021,
      "step": 94800
    },
    {
      "epoch": 0.6080818899859861,
      "grad_norm": 0.06631411612033844,
      "learning_rate": 8.784519796037909e-06,
      "loss": 0.0044,
      "step": 94810
    },
    {
      "epoch": 0.6081460268797722,
      "grad_norm": 1.4281866550445557,
      "learning_rate": 8.784153991549537e-06,
      "loss": 0.0032,
      "step": 94820
    },
    {
      "epoch": 0.6082101637735583,
      "grad_norm": 0.13056783378124237,
      "learning_rate": 8.78378813964301e-06,
      "loss": 0.0053,
      "step": 94830
    },
    {
      "epoch": 0.6082743006673443,
      "grad_norm": 0.13347940146923065,
      "learning_rate": 8.783422240322913e-06,
      "loss": 0.0028,
      "step": 94840
    },
    {
      "epoch": 0.6083384375611305,
      "grad_norm": 0.16992448270320892,
      "learning_rate": 8.783056293593832e-06,
      "loss": 0.0027,
      "step": 94850
    },
    {
      "epoch": 0.6084025744549165,
      "grad_norm": 0.1951434165239334,
      "learning_rate": 8.782690299460353e-06,
      "loss": 0.0028,
      "step": 94860
    },
    {
      "epoch": 0.6084667113487027,
      "grad_norm": 0.05179301276803017,
      "learning_rate": 8.78232425792706e-06,
      "loss": 0.0032,
      "step": 94870
    },
    {
      "epoch": 0.6085308482424888,
      "grad_norm": 0.10104519128799438,
      "learning_rate": 8.781958168998542e-06,
      "loss": 0.0032,
      "step": 94880
    },
    {
      "epoch": 0.6085949851362749,
      "grad_norm": 0.23673437535762787,
      "learning_rate": 8.781592032679382e-06,
      "loss": 0.0046,
      "step": 94890
    },
    {
      "epoch": 0.608659122030061,
      "grad_norm": 0.3082229197025299,
      "learning_rate": 8.781225848974174e-06,
      "loss": 0.0066,
      "step": 94900
    },
    {
      "epoch": 0.608723258923847,
      "grad_norm": 0.2882019281387329,
      "learning_rate": 8.780859617887503e-06,
      "loss": 0.0031,
      "step": 94910
    },
    {
      "epoch": 0.6087873958176332,
      "grad_norm": 0.09817387163639069,
      "learning_rate": 8.780493339423958e-06,
      "loss": 0.0036,
      "step": 94920
    },
    {
      "epoch": 0.6088515327114192,
      "grad_norm": 0.12644025683403015,
      "learning_rate": 8.780127013588131e-06,
      "loss": 0.0018,
      "step": 94930
    },
    {
      "epoch": 0.6089156696052054,
      "grad_norm": 0.04789069667458534,
      "learning_rate": 8.779760640384612e-06,
      "loss": 0.003,
      "step": 94940
    },
    {
      "epoch": 0.6089798064989914,
      "grad_norm": 0.06322155147790909,
      "learning_rate": 8.779394219817989e-06,
      "loss": 0.0022,
      "step": 94950
    },
    {
      "epoch": 0.6090439433927776,
      "grad_norm": 0.15940053761005402,
      "learning_rate": 8.779027751892855e-06,
      "loss": 0.0024,
      "step": 94960
    },
    {
      "epoch": 0.6091080802865636,
      "grad_norm": 0.18789884448051453,
      "learning_rate": 8.778661236613803e-06,
      "loss": 0.0027,
      "step": 94970
    },
    {
      "epoch": 0.6091722171803498,
      "grad_norm": 0.06264634430408478,
      "learning_rate": 8.778294673985426e-06,
      "loss": 0.0043,
      "step": 94980
    },
    {
      "epoch": 0.6092363540741358,
      "grad_norm": 0.05772439390420914,
      "learning_rate": 8.777928064012315e-06,
      "loss": 0.0023,
      "step": 94990
    },
    {
      "epoch": 0.6093004909679219,
      "grad_norm": 0.17368662357330322,
      "learning_rate": 8.777561406699065e-06,
      "loss": 0.0031,
      "step": 95000
    },
    {
      "epoch": 0.609364627861708,
      "grad_norm": 0.062287699431180954,
      "learning_rate": 8.777194702050273e-06,
      "loss": 0.0032,
      "step": 95010
    },
    {
      "epoch": 0.6094287647554941,
      "grad_norm": 0.04523130878806114,
      "learning_rate": 8.77682795007053e-06,
      "loss": 0.0027,
      "step": 95020
    },
    {
      "epoch": 0.6094929016492803,
      "grad_norm": 0.3858431875705719,
      "learning_rate": 8.776461150764434e-06,
      "loss": 0.0038,
      "step": 95030
    },
    {
      "epoch": 0.6095570385430663,
      "grad_norm": 0.11962731927633286,
      "learning_rate": 8.776094304136581e-06,
      "loss": 0.0033,
      "step": 95040
    },
    {
      "epoch": 0.6096211754368525,
      "grad_norm": 0.3771175444126129,
      "learning_rate": 8.775727410191566e-06,
      "loss": 0.0028,
      "step": 95050
    },
    {
      "epoch": 0.6096853123306385,
      "grad_norm": 0.3290991485118866,
      "learning_rate": 8.77536046893399e-06,
      "loss": 0.0044,
      "step": 95060
    },
    {
      "epoch": 0.6097494492244246,
      "grad_norm": 0.05832473561167717,
      "learning_rate": 8.77499348036845e-06,
      "loss": 0.0029,
      "step": 95070
    },
    {
      "epoch": 0.6098135861182107,
      "grad_norm": 0.11864414066076279,
      "learning_rate": 8.774626444499542e-06,
      "loss": 0.0044,
      "step": 95080
    },
    {
      "epoch": 0.6098777230119968,
      "grad_norm": 0.17416812479496002,
      "learning_rate": 8.774259361331868e-06,
      "loss": 0.0031,
      "step": 95090
    },
    {
      "epoch": 0.6099418599057829,
      "grad_norm": 0.06899726390838623,
      "learning_rate": 8.773892230870025e-06,
      "loss": 0.0026,
      "step": 95100
    },
    {
      "epoch": 0.610005996799569,
      "grad_norm": 0.2982441782951355,
      "learning_rate": 8.773525053118617e-06,
      "loss": 0.004,
      "step": 95110
    },
    {
      "epoch": 0.610070133693355,
      "grad_norm": 0.07267230749130249,
      "learning_rate": 8.773157828082242e-06,
      "loss": 0.0061,
      "step": 95120
    },
    {
      "epoch": 0.6101342705871412,
      "grad_norm": 0.20534218847751617,
      "learning_rate": 8.772790555765504e-06,
      "loss": 0.0026,
      "step": 95130
    },
    {
      "epoch": 0.6101984074809272,
      "grad_norm": 0.0956711694598198,
      "learning_rate": 8.772423236173001e-06,
      "loss": 0.0028,
      "step": 95140
    },
    {
      "epoch": 0.6102625443747134,
      "grad_norm": 0.3632100224494934,
      "learning_rate": 8.772055869309343e-06,
      "loss": 0.003,
      "step": 95150
    },
    {
      "epoch": 0.6103266812684995,
      "grad_norm": 0.3185942769050598,
      "learning_rate": 8.771688455179128e-06,
      "loss": 0.0043,
      "step": 95160
    },
    {
      "epoch": 0.6103908181622856,
      "grad_norm": 0.21670421957969666,
      "learning_rate": 8.771320993786958e-06,
      "loss": 0.0031,
      "step": 95170
    },
    {
      "epoch": 0.6104549550560717,
      "grad_norm": 0.2991900146007538,
      "learning_rate": 8.770953485137444e-06,
      "loss": 0.0034,
      "step": 95180
    },
    {
      "epoch": 0.6105190919498578,
      "grad_norm": 0.12183010578155518,
      "learning_rate": 8.770585929235187e-06,
      "loss": 0.0031,
      "step": 95190
    },
    {
      "epoch": 0.6105832288436439,
      "grad_norm": 0.23525065183639526,
      "learning_rate": 8.770218326084793e-06,
      "loss": 0.003,
      "step": 95200
    },
    {
      "epoch": 0.61064736573743,
      "grad_norm": 0.23921328783035278,
      "learning_rate": 8.76985067569087e-06,
      "loss": 0.005,
      "step": 95210
    },
    {
      "epoch": 0.6107115026312161,
      "grad_norm": 0.22330203652381897,
      "learning_rate": 8.769482978058021e-06,
      "loss": 0.0025,
      "step": 95220
    },
    {
      "epoch": 0.6107756395250021,
      "grad_norm": 0.27113255858421326,
      "learning_rate": 8.769115233190859e-06,
      "loss": 0.0034,
      "step": 95230
    },
    {
      "epoch": 0.6108397764187883,
      "grad_norm": 0.07810204476118088,
      "learning_rate": 8.768747441093988e-06,
      "loss": 0.0033,
      "step": 95240
    },
    {
      "epoch": 0.6109039133125743,
      "grad_norm": 0.15932008624076843,
      "learning_rate": 8.768379601772018e-06,
      "loss": 0.0042,
      "step": 95250
    },
    {
      "epoch": 0.6109680502063605,
      "grad_norm": 0.07703088223934174,
      "learning_rate": 8.768011715229559e-06,
      "loss": 0.0047,
      "step": 95260
    },
    {
      "epoch": 0.6110321871001465,
      "grad_norm": 0.4311964511871338,
      "learning_rate": 8.76764378147122e-06,
      "loss": 0.0048,
      "step": 95270
    },
    {
      "epoch": 0.6110963239939327,
      "grad_norm": 0.07001207023859024,
      "learning_rate": 8.76727580050161e-06,
      "loss": 0.0025,
      "step": 95280
    },
    {
      "epoch": 0.6111604608877187,
      "grad_norm": 0.12717720866203308,
      "learning_rate": 8.766907772325344e-06,
      "loss": 0.0023,
      "step": 95290
    },
    {
      "epoch": 0.6112245977815048,
      "grad_norm": 0.10921081900596619,
      "learning_rate": 8.766539696947032e-06,
      "loss": 0.003,
      "step": 95300
    },
    {
      "epoch": 0.611288734675291,
      "grad_norm": 0.06188970059156418,
      "learning_rate": 8.766171574371285e-06,
      "loss": 0.0035,
      "step": 95310
    },
    {
      "epoch": 0.611352871569077,
      "grad_norm": 0.08212490379810333,
      "learning_rate": 8.765803404602716e-06,
      "loss": 0.0049,
      "step": 95320
    },
    {
      "epoch": 0.6114170084628632,
      "grad_norm": 0.11185979098081589,
      "learning_rate": 8.76543518764594e-06,
      "loss": 0.0028,
      "step": 95330
    },
    {
      "epoch": 0.6114811453566492,
      "grad_norm": 0.2138770967721939,
      "learning_rate": 8.765066923505569e-06,
      "loss": 0.0025,
      "step": 95340
    },
    {
      "epoch": 0.6115452822504354,
      "grad_norm": 0.12622995674610138,
      "learning_rate": 8.764698612186217e-06,
      "loss": 0.0025,
      "step": 95350
    },
    {
      "epoch": 0.6116094191442214,
      "grad_norm": 0.11631227284669876,
      "learning_rate": 8.764330253692504e-06,
      "loss": 0.0063,
      "step": 95360
    },
    {
      "epoch": 0.6116735560380075,
      "grad_norm": 0.10350869596004486,
      "learning_rate": 8.76396184802904e-06,
      "loss": 0.0026,
      "step": 95370
    },
    {
      "epoch": 0.6117376929317936,
      "grad_norm": 0.18187642097473145,
      "learning_rate": 8.763593395200446e-06,
      "loss": 0.0034,
      "step": 95380
    },
    {
      "epoch": 0.6118018298255797,
      "grad_norm": 0.17044878005981445,
      "learning_rate": 8.763224895211336e-06,
      "loss": 0.0036,
      "step": 95390
    },
    {
      "epoch": 0.6118659667193658,
      "grad_norm": 0.08452159911394119,
      "learning_rate": 8.76285634806633e-06,
      "loss": 0.0037,
      "step": 95400
    },
    {
      "epoch": 0.6119301036131519,
      "grad_norm": 0.12100055813789368,
      "learning_rate": 8.762487753770041e-06,
      "loss": 0.0047,
      "step": 95410
    },
    {
      "epoch": 0.611994240506938,
      "grad_norm": 0.12239952385425568,
      "learning_rate": 8.762119112327095e-06,
      "loss": 0.0045,
      "step": 95420
    },
    {
      "epoch": 0.6120583774007241,
      "grad_norm": 0.9005413055419922,
      "learning_rate": 8.761750423742106e-06,
      "loss": 0.0035,
      "step": 95430
    },
    {
      "epoch": 0.6121225142945103,
      "grad_norm": 0.48149168491363525,
      "learning_rate": 8.761381688019696e-06,
      "loss": 0.0037,
      "step": 95440
    },
    {
      "epoch": 0.6121866511882963,
      "grad_norm": 0.435469388961792,
      "learning_rate": 8.761012905164487e-06,
      "loss": 0.0035,
      "step": 95450
    },
    {
      "epoch": 0.6122507880820824,
      "grad_norm": 0.24214886128902435,
      "learning_rate": 8.760644075181097e-06,
      "loss": 0.0033,
      "step": 95460
    },
    {
      "epoch": 0.6123149249758685,
      "grad_norm": 0.09286938607692719,
      "learning_rate": 8.760275198074148e-06,
      "loss": 0.0038,
      "step": 95470
    },
    {
      "epoch": 0.6123790618696546,
      "grad_norm": 0.06739024817943573,
      "learning_rate": 8.759906273848265e-06,
      "loss": 0.0038,
      "step": 95480
    },
    {
      "epoch": 0.6124431987634407,
      "grad_norm": 0.24788588285446167,
      "learning_rate": 8.759537302508067e-06,
      "loss": 0.0044,
      "step": 95490
    },
    {
      "epoch": 0.6125073356572268,
      "grad_norm": 0.09615003317594528,
      "learning_rate": 8.759168284058182e-06,
      "loss": 0.0034,
      "step": 95500
    },
    {
      "epoch": 0.6125714725510129,
      "grad_norm": 0.22313320636749268,
      "learning_rate": 8.758799218503232e-06,
      "loss": 0.0032,
      "step": 95510
    },
    {
      "epoch": 0.612635609444799,
      "grad_norm": 0.20353081822395325,
      "learning_rate": 8.758430105847839e-06,
      "loss": 0.0043,
      "step": 95520
    },
    {
      "epoch": 0.612699746338585,
      "grad_norm": 0.38399964570999146,
      "learning_rate": 8.75806094609663e-06,
      "loss": 0.0085,
      "step": 95530
    },
    {
      "epoch": 0.6127638832323712,
      "grad_norm": 0.1326301544904709,
      "learning_rate": 8.757691739254235e-06,
      "loss": 0.0092,
      "step": 95540
    },
    {
      "epoch": 0.6128280201261572,
      "grad_norm": 0.112067811191082,
      "learning_rate": 8.757322485325276e-06,
      "loss": 0.0019,
      "step": 95550
    },
    {
      "epoch": 0.6128921570199434,
      "grad_norm": 0.4138074517250061,
      "learning_rate": 8.75695318431438e-06,
      "loss": 0.0051,
      "step": 95560
    },
    {
      "epoch": 0.6129562939137294,
      "grad_norm": 0.24018257856369019,
      "learning_rate": 8.756583836226176e-06,
      "loss": 0.0028,
      "step": 95570
    },
    {
      "epoch": 0.6130204308075156,
      "grad_norm": 0.20457017421722412,
      "learning_rate": 8.756214441065291e-06,
      "loss": 0.0035,
      "step": 95580
    },
    {
      "epoch": 0.6130845677013017,
      "grad_norm": 0.11519509553909302,
      "learning_rate": 8.755844998836356e-06,
      "loss": 0.0043,
      "step": 95590
    },
    {
      "epoch": 0.6131487045950877,
      "grad_norm": 0.21396492421627045,
      "learning_rate": 8.755475509543999e-06,
      "loss": 0.0053,
      "step": 95600
    },
    {
      "epoch": 0.6132128414888739,
      "grad_norm": 0.09090571105480194,
      "learning_rate": 8.755105973192847e-06,
      "loss": 0.0063,
      "step": 95610
    },
    {
      "epoch": 0.6132769783826599,
      "grad_norm": 0.21443317830562592,
      "learning_rate": 8.754736389787536e-06,
      "loss": 0.0032,
      "step": 95620
    },
    {
      "epoch": 0.6133411152764461,
      "grad_norm": 0.08013524115085602,
      "learning_rate": 8.754366759332695e-06,
      "loss": 0.0028,
      "step": 95630
    },
    {
      "epoch": 0.6134052521702321,
      "grad_norm": 0.31743213534355164,
      "learning_rate": 8.753997081832954e-06,
      "loss": 0.0051,
      "step": 95640
    },
    {
      "epoch": 0.6134693890640183,
      "grad_norm": 0.14229175448417664,
      "learning_rate": 8.753627357292947e-06,
      "loss": 0.0036,
      "step": 95650
    },
    {
      "epoch": 0.6135335259578043,
      "grad_norm": 0.07585162669420242,
      "learning_rate": 8.753257585717305e-06,
      "loss": 0.0023,
      "step": 95660
    },
    {
      "epoch": 0.6135976628515905,
      "grad_norm": 0.17819391191005707,
      "learning_rate": 8.752887767110666e-06,
      "loss": 0.0051,
      "step": 95670
    },
    {
      "epoch": 0.6136617997453765,
      "grad_norm": 0.0805431678891182,
      "learning_rate": 8.752517901477658e-06,
      "loss": 0.0014,
      "step": 95680
    },
    {
      "epoch": 0.6137259366391626,
      "grad_norm": 0.18486371636390686,
      "learning_rate": 8.752147988822921e-06,
      "loss": 0.0023,
      "step": 95690
    },
    {
      "epoch": 0.6137900735329487,
      "grad_norm": 0.2803860306739807,
      "learning_rate": 8.751778029151087e-06,
      "loss": 0.0039,
      "step": 95700
    },
    {
      "epoch": 0.6138542104267348,
      "grad_norm": 0.1302492767572403,
      "learning_rate": 8.751408022466794e-06,
      "loss": 0.0033,
      "step": 95710
    },
    {
      "epoch": 0.613918347320521,
      "grad_norm": 0.11332884430885315,
      "learning_rate": 8.751037968774676e-06,
      "loss": 0.002,
      "step": 95720
    },
    {
      "epoch": 0.613982484214307,
      "grad_norm": 0.08457744121551514,
      "learning_rate": 8.750667868079372e-06,
      "loss": 0.0028,
      "step": 95730
    },
    {
      "epoch": 0.6140466211080932,
      "grad_norm": 0.08075898140668869,
      "learning_rate": 8.75029772038552e-06,
      "loss": 0.0065,
      "step": 95740
    },
    {
      "epoch": 0.6141107580018792,
      "grad_norm": 0.06535733491182327,
      "learning_rate": 8.749927525697757e-06,
      "loss": 0.0035,
      "step": 95750
    },
    {
      "epoch": 0.6141748948956653,
      "grad_norm": 0.1477227509021759,
      "learning_rate": 8.74955728402072e-06,
      "loss": 0.0056,
      "step": 95760
    },
    {
      "epoch": 0.6142390317894514,
      "grad_norm": 0.10869273543357849,
      "learning_rate": 8.749186995359054e-06,
      "loss": 0.0024,
      "step": 95770
    },
    {
      "epoch": 0.6143031686832375,
      "grad_norm": 0.2968710958957672,
      "learning_rate": 8.748816659717392e-06,
      "loss": 0.0057,
      "step": 95780
    },
    {
      "epoch": 0.6143673055770236,
      "grad_norm": 0.07847687602043152,
      "learning_rate": 8.74844627710038e-06,
      "loss": 0.0032,
      "step": 95790
    },
    {
      "epoch": 0.6144314424708097,
      "grad_norm": 0.14137952029705048,
      "learning_rate": 8.748075847512656e-06,
      "loss": 0.0035,
      "step": 95800
    },
    {
      "epoch": 0.6144955793645958,
      "grad_norm": 0.22979572415351868,
      "learning_rate": 8.747705370958865e-06,
      "loss": 0.0031,
      "step": 95810
    },
    {
      "epoch": 0.6145597162583819,
      "grad_norm": 0.05405260995030403,
      "learning_rate": 8.747334847443644e-06,
      "loss": 0.0011,
      "step": 95820
    },
    {
      "epoch": 0.6146238531521679,
      "grad_norm": 0.2479952871799469,
      "learning_rate": 8.746964276971642e-06,
      "loss": 0.004,
      "step": 95830
    },
    {
      "epoch": 0.6146879900459541,
      "grad_norm": 0.08774492889642715,
      "learning_rate": 8.7465936595475e-06,
      "loss": 0.0053,
      "step": 95840
    },
    {
      "epoch": 0.6147521269397401,
      "grad_norm": 0.16067706048488617,
      "learning_rate": 8.74622299517586e-06,
      "loss": 0.004,
      "step": 95850
    },
    {
      "epoch": 0.6148162638335263,
      "grad_norm": 0.22178137302398682,
      "learning_rate": 8.745852283861368e-06,
      "loss": 0.0034,
      "step": 95860
    },
    {
      "epoch": 0.6148804007273124,
      "grad_norm": 0.1768014281988144,
      "learning_rate": 8.745481525608671e-06,
      "loss": 0.0016,
      "step": 95870
    },
    {
      "epoch": 0.6149445376210985,
      "grad_norm": 0.04831257462501526,
      "learning_rate": 8.745110720422414e-06,
      "loss": 0.0027,
      "step": 95880
    },
    {
      "epoch": 0.6150086745148846,
      "grad_norm": 0.3030824363231659,
      "learning_rate": 8.744739868307241e-06,
      "loss": 0.0041,
      "step": 95890
    },
    {
      "epoch": 0.6150728114086706,
      "grad_norm": 0.15502512454986572,
      "learning_rate": 8.744368969267804e-06,
      "loss": 0.0044,
      "step": 95900
    },
    {
      "epoch": 0.6151369483024568,
      "grad_norm": 0.052240338176488876,
      "learning_rate": 8.743998023308747e-06,
      "loss": 0.0017,
      "step": 95910
    },
    {
      "epoch": 0.6152010851962428,
      "grad_norm": 0.2639593482017517,
      "learning_rate": 8.743627030434718e-06,
      "loss": 0.0031,
      "step": 95920
    },
    {
      "epoch": 0.615265222090029,
      "grad_norm": 0.21429011225700378,
      "learning_rate": 8.743255990650365e-06,
      "loss": 0.0029,
      "step": 95930
    },
    {
      "epoch": 0.615329358983815,
      "grad_norm": 0.28203192353248596,
      "learning_rate": 8.742884903960343e-06,
      "loss": 0.0023,
      "step": 95940
    },
    {
      "epoch": 0.6153934958776012,
      "grad_norm": 0.0851309522986412,
      "learning_rate": 8.742513770369297e-06,
      "loss": 0.0018,
      "step": 95950
    },
    {
      "epoch": 0.6154576327713872,
      "grad_norm": 0.09846670180559158,
      "learning_rate": 8.742142589881876e-06,
      "loss": 0.0022,
      "step": 95960
    },
    {
      "epoch": 0.6155217696651734,
      "grad_norm": 0.059684813022613525,
      "learning_rate": 8.741771362502738e-06,
      "loss": 0.0039,
      "step": 95970
    },
    {
      "epoch": 0.6155859065589594,
      "grad_norm": 0.04888633266091347,
      "learning_rate": 8.741400088236527e-06,
      "loss": 0.0025,
      "step": 95980
    },
    {
      "epoch": 0.6156500434527455,
      "grad_norm": 0.22315432131290436,
      "learning_rate": 8.741028767087902e-06,
      "loss": 0.0033,
      "step": 95990
    },
    {
      "epoch": 0.6157141803465317,
      "grad_norm": 0.08538077026605606,
      "learning_rate": 8.740657399061513e-06,
      "loss": 0.0025,
      "step": 96000
    },
    {
      "epoch": 0.6157783172403177,
      "grad_norm": 0.320173442363739,
      "learning_rate": 8.74028598416201e-06,
      "loss": 0.0038,
      "step": 96010
    },
    {
      "epoch": 0.6158424541341039,
      "grad_norm": 0.15246106684207916,
      "learning_rate": 8.739914522394052e-06,
      "loss": 0.0046,
      "step": 96020
    },
    {
      "epoch": 0.6159065910278899,
      "grad_norm": 0.12204425036907196,
      "learning_rate": 8.739543013762292e-06,
      "loss": 0.005,
      "step": 96030
    },
    {
      "epoch": 0.6159707279216761,
      "grad_norm": 0.16110706329345703,
      "learning_rate": 8.739171458271387e-06,
      "loss": 0.0031,
      "step": 96040
    },
    {
      "epoch": 0.6160348648154621,
      "grad_norm": 0.8272030353546143,
      "learning_rate": 8.738799855925991e-06,
      "loss": 0.0026,
      "step": 96050
    },
    {
      "epoch": 0.6160990017092483,
      "grad_norm": 0.2025410681962967,
      "learning_rate": 8.738428206730758e-06,
      "loss": 0.0049,
      "step": 96060
    },
    {
      "epoch": 0.6161631386030343,
      "grad_norm": 0.1599372774362564,
      "learning_rate": 8.73805651069035e-06,
      "loss": 0.0036,
      "step": 96070
    },
    {
      "epoch": 0.6162272754968204,
      "grad_norm": 0.10138299316167831,
      "learning_rate": 8.737684767809423e-06,
      "loss": 0.0027,
      "step": 96080
    },
    {
      "epoch": 0.6162914123906065,
      "grad_norm": 0.16237998008728027,
      "learning_rate": 8.737312978092634e-06,
      "loss": 0.0029,
      "step": 96090
    },
    {
      "epoch": 0.6163555492843926,
      "grad_norm": 0.035956889390945435,
      "learning_rate": 8.736941141544642e-06,
      "loss": 0.0037,
      "step": 96100
    },
    {
      "epoch": 0.6164196861781787,
      "grad_norm": 0.09827184677124023,
      "learning_rate": 8.736569258170107e-06,
      "loss": 0.004,
      "step": 96110
    },
    {
      "epoch": 0.6164838230719648,
      "grad_norm": 0.25321224331855774,
      "learning_rate": 8.736197327973688e-06,
      "loss": 0.0025,
      "step": 96120
    },
    {
      "epoch": 0.6165479599657508,
      "grad_norm": 0.08422735333442688,
      "learning_rate": 8.735825350960045e-06,
      "loss": 0.0041,
      "step": 96130
    },
    {
      "epoch": 0.616612096859537,
      "grad_norm": 0.12227088958024979,
      "learning_rate": 8.735453327133842e-06,
      "loss": 0.0024,
      "step": 96140
    },
    {
      "epoch": 0.6166762337533231,
      "grad_norm": 0.06718363612890244,
      "learning_rate": 8.73508125649974e-06,
      "loss": 0.0029,
      "step": 96150
    },
    {
      "epoch": 0.6167403706471092,
      "grad_norm": 0.038292769342660904,
      "learning_rate": 8.7347091390624e-06,
      "loss": 0.0016,
      "step": 96160
    },
    {
      "epoch": 0.6168045075408953,
      "grad_norm": 0.08811763674020767,
      "learning_rate": 8.734336974826486e-06,
      "loss": 0.0028,
      "step": 96170
    },
    {
      "epoch": 0.6168686444346814,
      "grad_norm": 0.1309226006269455,
      "learning_rate": 8.733964763796659e-06,
      "loss": 0.0032,
      "step": 96180
    },
    {
      "epoch": 0.6169327813284675,
      "grad_norm": 0.13165056705474854,
      "learning_rate": 8.733592505977586e-06,
      "loss": 0.0039,
      "step": 96190
    },
    {
      "epoch": 0.6169969182222536,
      "grad_norm": 0.06533897668123245,
      "learning_rate": 8.73322020137393e-06,
      "loss": 0.0032,
      "step": 96200
    },
    {
      "epoch": 0.6170610551160397,
      "grad_norm": 0.20202882587909698,
      "learning_rate": 8.732847849990358e-06,
      "loss": 0.0028,
      "step": 96210
    },
    {
      "epoch": 0.6171251920098257,
      "grad_norm": 0.1397777944803238,
      "learning_rate": 8.732475451831533e-06,
      "loss": 0.0039,
      "step": 96220
    },
    {
      "epoch": 0.6171893289036119,
      "grad_norm": 0.1611679047346115,
      "learning_rate": 8.732103006902125e-06,
      "loss": 0.0037,
      "step": 96230
    },
    {
      "epoch": 0.6172534657973979,
      "grad_norm": 0.18416641652584076,
      "learning_rate": 8.731730515206796e-06,
      "loss": 0.0033,
      "step": 96240
    },
    {
      "epoch": 0.6173176026911841,
      "grad_norm": 0.04296709969639778,
      "learning_rate": 8.731357976750219e-06,
      "loss": 0.0032,
      "step": 96250
    },
    {
      "epoch": 0.6173817395849701,
      "grad_norm": 0.11918571591377258,
      "learning_rate": 8.730985391537059e-06,
      "loss": 0.0022,
      "step": 96260
    },
    {
      "epoch": 0.6174458764787563,
      "grad_norm": 0.4370180666446686,
      "learning_rate": 8.730612759571986e-06,
      "loss": 0.0044,
      "step": 96270
    },
    {
      "epoch": 0.6175100133725423,
      "grad_norm": 0.05504896864295006,
      "learning_rate": 8.730240080859667e-06,
      "loss": 0.0018,
      "step": 96280
    },
    {
      "epoch": 0.6175741502663284,
      "grad_norm": 0.2740112245082855,
      "learning_rate": 8.729867355404776e-06,
      "loss": 0.0033,
      "step": 96290
    },
    {
      "epoch": 0.6176382871601146,
      "grad_norm": 0.1244712620973587,
      "learning_rate": 8.729494583211981e-06,
      "loss": 0.0025,
      "step": 96300
    },
    {
      "epoch": 0.6177024240539006,
      "grad_norm": 0.04363995045423508,
      "learning_rate": 8.729121764285953e-06,
      "loss": 0.0025,
      "step": 96310
    },
    {
      "epoch": 0.6177665609476868,
      "grad_norm": 0.1507520228624344,
      "learning_rate": 8.728748898631365e-06,
      "loss": 0.0031,
      "step": 96320
    },
    {
      "epoch": 0.6178306978414728,
      "grad_norm": 0.08464416861534119,
      "learning_rate": 8.728375986252888e-06,
      "loss": 0.0046,
      "step": 96330
    },
    {
      "epoch": 0.617894834735259,
      "grad_norm": 0.1334334909915924,
      "learning_rate": 8.728003027155194e-06,
      "loss": 0.0038,
      "step": 96340
    },
    {
      "epoch": 0.617958971629045,
      "grad_norm": 0.2237366884946823,
      "learning_rate": 8.727630021342958e-06,
      "loss": 0.0046,
      "step": 96350
    },
    {
      "epoch": 0.6180231085228312,
      "grad_norm": 0.1561523973941803,
      "learning_rate": 8.727256968820855e-06,
      "loss": 0.0032,
      "step": 96360
    },
    {
      "epoch": 0.6180872454166172,
      "grad_norm": 0.13996556401252747,
      "learning_rate": 8.72688386959356e-06,
      "loss": 0.0071,
      "step": 96370
    },
    {
      "epoch": 0.6181513823104033,
      "grad_norm": 0.06766559183597565,
      "learning_rate": 8.726510723665742e-06,
      "loss": 0.0022,
      "step": 96380
    },
    {
      "epoch": 0.6182155192041894,
      "grad_norm": 0.3309437334537506,
      "learning_rate": 8.726137531042084e-06,
      "loss": 0.004,
      "step": 96390
    },
    {
      "epoch": 0.6182796560979755,
      "grad_norm": 0.08681367337703705,
      "learning_rate": 8.72576429172726e-06,
      "loss": 0.003,
      "step": 96400
    },
    {
      "epoch": 0.6183437929917616,
      "grad_norm": 0.13890603184700012,
      "learning_rate": 8.725391005725944e-06,
      "loss": 0.0049,
      "step": 96410
    },
    {
      "epoch": 0.6184079298855477,
      "grad_norm": 0.0536317341029644,
      "learning_rate": 8.725017673042819e-06,
      "loss": 0.0026,
      "step": 96420
    },
    {
      "epoch": 0.6184720667793339,
      "grad_norm": 0.1660311073064804,
      "learning_rate": 8.72464429368256e-06,
      "loss": 0.0033,
      "step": 96430
    },
    {
      "epoch": 0.6185362036731199,
      "grad_norm": 0.07175929844379425,
      "learning_rate": 8.724270867649846e-06,
      "loss": 0.0018,
      "step": 96440
    },
    {
      "epoch": 0.618600340566906,
      "grad_norm": 0.08966167271137238,
      "learning_rate": 8.723897394949357e-06,
      "loss": 0.0045,
      "step": 96450
    },
    {
      "epoch": 0.6186644774606921,
      "grad_norm": 0.1288372278213501,
      "learning_rate": 8.72352387558577e-06,
      "loss": 0.0019,
      "step": 96460
    },
    {
      "epoch": 0.6187286143544782,
      "grad_norm": 0.10833292454481125,
      "learning_rate": 8.723150309563769e-06,
      "loss": 0.0029,
      "step": 96470
    },
    {
      "epoch": 0.6187927512482643,
      "grad_norm": 0.011052173562347889,
      "learning_rate": 8.722776696888033e-06,
      "loss": 0.0031,
      "step": 96480
    },
    {
      "epoch": 0.6188568881420504,
      "grad_norm": 0.11594557017087936,
      "learning_rate": 8.722403037563244e-06,
      "loss": 0.0024,
      "step": 96490
    },
    {
      "epoch": 0.6189210250358365,
      "grad_norm": 0.1987140029668808,
      "learning_rate": 8.722029331594086e-06,
      "loss": 0.0033,
      "step": 96500
    },
    {
      "epoch": 0.6189851619296226,
      "grad_norm": 0.05850119888782501,
      "learning_rate": 8.72165557898524e-06,
      "loss": 0.0028,
      "step": 96510
    },
    {
      "epoch": 0.6190492988234086,
      "grad_norm": 0.10307785868644714,
      "learning_rate": 8.721281779741391e-06,
      "loss": 0.003,
      "step": 96520
    },
    {
      "epoch": 0.6191134357171948,
      "grad_norm": 0.046492837369441986,
      "learning_rate": 8.72090793386722e-06,
      "loss": 0.0067,
      "step": 96530
    },
    {
      "epoch": 0.6191775726109808,
      "grad_norm": 0.3101746439933777,
      "learning_rate": 8.720534041367414e-06,
      "loss": 0.0026,
      "step": 96540
    },
    {
      "epoch": 0.619241709504767,
      "grad_norm": 0.11407823860645294,
      "learning_rate": 8.720160102246657e-06,
      "loss": 0.003,
      "step": 96550
    },
    {
      "epoch": 0.619305846398553,
      "grad_norm": 0.2032564878463745,
      "learning_rate": 8.719786116509633e-06,
      "loss": 0.0052,
      "step": 96560
    },
    {
      "epoch": 0.6193699832923392,
      "grad_norm": 0.2516917586326599,
      "learning_rate": 8.719412084161034e-06,
      "loss": 0.0031,
      "step": 96570
    },
    {
      "epoch": 0.6194341201861253,
      "grad_norm": 0.2087072730064392,
      "learning_rate": 8.719038005205542e-06,
      "loss": 0.0028,
      "step": 96580
    },
    {
      "epoch": 0.6194982570799114,
      "grad_norm": 0.06007470190525055,
      "learning_rate": 8.718663879647846e-06,
      "loss": 0.006,
      "step": 96590
    },
    {
      "epoch": 0.6195623939736975,
      "grad_norm": 0.05991493910551071,
      "learning_rate": 8.718289707492634e-06,
      "loss": 0.0053,
      "step": 96600
    },
    {
      "epoch": 0.6196265308674835,
      "grad_norm": 0.2905253767967224,
      "learning_rate": 8.717915488744595e-06,
      "loss": 0.0043,
      "step": 96610
    },
    {
      "epoch": 0.6196906677612697,
      "grad_norm": 0.16421377658843994,
      "learning_rate": 8.717541223408417e-06,
      "loss": 0.0042,
      "step": 96620
    },
    {
      "epoch": 0.6197548046550557,
      "grad_norm": 0.14254648983478546,
      "learning_rate": 8.71716691148879e-06,
      "loss": 0.0034,
      "step": 96630
    },
    {
      "epoch": 0.6198189415488419,
      "grad_norm": 0.1617589294910431,
      "learning_rate": 8.716792552990405e-06,
      "loss": 0.004,
      "step": 96640
    },
    {
      "epoch": 0.6198830784426279,
      "grad_norm": 0.2060152143239975,
      "learning_rate": 8.716418147917954e-06,
      "loss": 0.0028,
      "step": 96650
    },
    {
      "epoch": 0.6199472153364141,
      "grad_norm": 0.14975200593471527,
      "learning_rate": 8.716043696276128e-06,
      "loss": 0.0029,
      "step": 96660
    },
    {
      "epoch": 0.6200113522302001,
      "grad_norm": 0.06567097455263138,
      "learning_rate": 8.715669198069617e-06,
      "loss": 0.0035,
      "step": 96670
    },
    {
      "epoch": 0.6200754891239862,
      "grad_norm": 0.11589567363262177,
      "learning_rate": 8.715294653303117e-06,
      "loss": 0.0022,
      "step": 96680
    },
    {
      "epoch": 0.6201396260177723,
      "grad_norm": 0.22879017889499664,
      "learning_rate": 8.714920061981317e-06,
      "loss": 0.0026,
      "step": 96690
    },
    {
      "epoch": 0.6202037629115584,
      "grad_norm": 0.2359607070684433,
      "learning_rate": 8.714545424108918e-06,
      "loss": 0.0032,
      "step": 96700
    },
    {
      "epoch": 0.6202678998053446,
      "grad_norm": 0.10882429778575897,
      "learning_rate": 8.714170739690606e-06,
      "loss": 0.002,
      "step": 96710
    },
    {
      "epoch": 0.6203320366991306,
      "grad_norm": 0.48844268918037415,
      "learning_rate": 8.713796008731084e-06,
      "loss": 0.0058,
      "step": 96720
    },
    {
      "epoch": 0.6203961735929168,
      "grad_norm": 0.2939786911010742,
      "learning_rate": 8.713421231235039e-06,
      "loss": 0.0023,
      "step": 96730
    },
    {
      "epoch": 0.6204603104867028,
      "grad_norm": 0.03793249651789665,
      "learning_rate": 8.713046407207176e-06,
      "loss": 0.0029,
      "step": 96740
    },
    {
      "epoch": 0.620524447380489,
      "grad_norm": 0.0540674589574337,
      "learning_rate": 8.712671536652187e-06,
      "loss": 0.0061,
      "step": 96750
    },
    {
      "epoch": 0.620588584274275,
      "grad_norm": 0.11687786132097244,
      "learning_rate": 8.712296619574769e-06,
      "loss": 0.0026,
      "step": 96760
    },
    {
      "epoch": 0.6206527211680611,
      "grad_norm": 0.14689037203788757,
      "learning_rate": 8.711921655979622e-06,
      "loss": 0.0026,
      "step": 96770
    },
    {
      "epoch": 0.6207168580618472,
      "grad_norm": 0.15522897243499756,
      "learning_rate": 8.711546645871444e-06,
      "loss": 0.0032,
      "step": 96780
    },
    {
      "epoch": 0.6207809949556333,
      "grad_norm": 0.2766576111316681,
      "learning_rate": 8.711171589254934e-06,
      "loss": 0.0025,
      "step": 96790
    },
    {
      "epoch": 0.6208451318494194,
      "grad_norm": 0.15739668905735016,
      "learning_rate": 8.71079648613479e-06,
      "loss": 0.0047,
      "step": 96800
    },
    {
      "epoch": 0.6209092687432055,
      "grad_norm": 0.14163656532764435,
      "learning_rate": 8.710421336515715e-06,
      "loss": 0.0037,
      "step": 96810
    },
    {
      "epoch": 0.6209734056369915,
      "grad_norm": 0.30086272954940796,
      "learning_rate": 8.71004614040241e-06,
      "loss": 0.0042,
      "step": 96820
    },
    {
      "epoch": 0.6210375425307777,
      "grad_norm": 0.029013173654675484,
      "learning_rate": 8.709670897799574e-06,
      "loss": 0.0018,
      "step": 96830
    },
    {
      "epoch": 0.6211016794245637,
      "grad_norm": 0.11647813767194748,
      "learning_rate": 8.709295608711912e-06,
      "loss": 0.0055,
      "step": 96840
    },
    {
      "epoch": 0.6211658163183499,
      "grad_norm": 0.14268749952316284,
      "learning_rate": 8.708920273144124e-06,
      "loss": 0.0021,
      "step": 96850
    },
    {
      "epoch": 0.621229953212136,
      "grad_norm": 0.37514549493789673,
      "learning_rate": 8.708544891100914e-06,
      "loss": 0.0041,
      "step": 96860
    },
    {
      "epoch": 0.6212940901059221,
      "grad_norm": 0.2563324570655823,
      "learning_rate": 8.708169462586986e-06,
      "loss": 0.0035,
      "step": 96870
    },
    {
      "epoch": 0.6213582269997082,
      "grad_norm": 0.5971559882164001,
      "learning_rate": 8.707793987607044e-06,
      "loss": 0.0059,
      "step": 96880
    },
    {
      "epoch": 0.6214223638934943,
      "grad_norm": 0.17791274189949036,
      "learning_rate": 8.707418466165795e-06,
      "loss": 0.0029,
      "step": 96890
    },
    {
      "epoch": 0.6214865007872804,
      "grad_norm": 0.08535439521074295,
      "learning_rate": 8.70704289826794e-06,
      "loss": 0.0031,
      "step": 96900
    },
    {
      "epoch": 0.6215506376810664,
      "grad_norm": 0.28252682089805603,
      "learning_rate": 8.70666728391819e-06,
      "loss": 0.0031,
      "step": 96910
    },
    {
      "epoch": 0.6216147745748526,
      "grad_norm": 0.07908795028924942,
      "learning_rate": 8.706291623121252e-06,
      "loss": 0.0027,
      "step": 96920
    },
    {
      "epoch": 0.6216789114686386,
      "grad_norm": 0.06172311678528786,
      "learning_rate": 8.705915915881828e-06,
      "loss": 0.0027,
      "step": 96930
    },
    {
      "epoch": 0.6217430483624248,
      "grad_norm": 0.0566222220659256,
      "learning_rate": 8.70554016220463e-06,
      "loss": 0.0024,
      "step": 96940
    },
    {
      "epoch": 0.6218071852562108,
      "grad_norm": 0.3074969947338104,
      "learning_rate": 8.705164362094366e-06,
      "loss": 0.0025,
      "step": 96950
    },
    {
      "epoch": 0.621871322149997,
      "grad_norm": 0.007374696433544159,
      "learning_rate": 8.704788515555745e-06,
      "loss": 0.0025,
      "step": 96960
    },
    {
      "epoch": 0.621935459043783,
      "grad_norm": 0.13997507095336914,
      "learning_rate": 8.704412622593474e-06,
      "loss": 0.0022,
      "step": 96970
    },
    {
      "epoch": 0.6219995959375691,
      "grad_norm": 0.16851702332496643,
      "learning_rate": 8.704036683212268e-06,
      "loss": 0.0033,
      "step": 96980
    },
    {
      "epoch": 0.6220637328313553,
      "grad_norm": 0.3087330460548401,
      "learning_rate": 8.703660697416832e-06,
      "loss": 0.0028,
      "step": 96990
    },
    {
      "epoch": 0.6221278697251413,
      "grad_norm": 0.6273163557052612,
      "learning_rate": 8.703284665211882e-06,
      "loss": 0.0024,
      "step": 97000
    },
    {
      "epoch": 0.6221920066189275,
      "grad_norm": 0.3395378887653351,
      "learning_rate": 8.702908586602128e-06,
      "loss": 0.0053,
      "step": 97010
    },
    {
      "epoch": 0.6222561435127135,
      "grad_norm": 0.026514574885368347,
      "learning_rate": 8.702532461592283e-06,
      "loss": 0.0027,
      "step": 97020
    },
    {
      "epoch": 0.6223202804064997,
      "grad_norm": 0.051178060472011566,
      "learning_rate": 8.702156290187061e-06,
      "loss": 0.0027,
      "step": 97030
    },
    {
      "epoch": 0.6223844173002857,
      "grad_norm": 0.24971944093704224,
      "learning_rate": 8.701780072391175e-06,
      "loss": 0.0025,
      "step": 97040
    },
    {
      "epoch": 0.6224485541940719,
      "grad_norm": 0.11027555167675018,
      "learning_rate": 8.701403808209339e-06,
      "loss": 0.0021,
      "step": 97050
    },
    {
      "epoch": 0.6225126910878579,
      "grad_norm": 0.15564948320388794,
      "learning_rate": 8.701027497646267e-06,
      "loss": 0.0018,
      "step": 97060
    },
    {
      "epoch": 0.622576827981644,
      "grad_norm": 0.21082234382629395,
      "learning_rate": 8.700651140706676e-06,
      "loss": 0.0019,
      "step": 97070
    },
    {
      "epoch": 0.6226409648754301,
      "grad_norm": 0.09402766078710556,
      "learning_rate": 8.700274737395282e-06,
      "loss": 0.0035,
      "step": 97080
    },
    {
      "epoch": 0.6227051017692162,
      "grad_norm": 0.09840180724859238,
      "learning_rate": 8.6998982877168e-06,
      "loss": 0.0018,
      "step": 97090
    },
    {
      "epoch": 0.6227692386630023,
      "grad_norm": 0.06667862832546234,
      "learning_rate": 8.699521791675947e-06,
      "loss": 0.0019,
      "step": 97100
    },
    {
      "epoch": 0.6228333755567884,
      "grad_norm": 0.11117648333311081,
      "learning_rate": 8.699145249277444e-06,
      "loss": 0.004,
      "step": 97110
    },
    {
      "epoch": 0.6228975124505745,
      "grad_norm": 0.2256828397512436,
      "learning_rate": 8.698768660526007e-06,
      "loss": 0.0045,
      "step": 97120
    },
    {
      "epoch": 0.6229616493443606,
      "grad_norm": 0.07285215705633163,
      "learning_rate": 8.698392025426355e-06,
      "loss": 0.0032,
      "step": 97130
    },
    {
      "epoch": 0.6230257862381468,
      "grad_norm": 0.3315751254558563,
      "learning_rate": 8.698015343983205e-06,
      "loss": 0.0026,
      "step": 97140
    },
    {
      "epoch": 0.6230899231319328,
      "grad_norm": 0.15173321962356567,
      "learning_rate": 8.697638616201284e-06,
      "loss": 0.0013,
      "step": 97150
    },
    {
      "epoch": 0.6231540600257189,
      "grad_norm": 0.1298132687807083,
      "learning_rate": 8.697261842085306e-06,
      "loss": 0.0039,
      "step": 97160
    },
    {
      "epoch": 0.623218196919505,
      "grad_norm": 0.19186744093894958,
      "learning_rate": 8.696885021639996e-06,
      "loss": 0.003,
      "step": 97170
    },
    {
      "epoch": 0.6232823338132911,
      "grad_norm": 0.3175899386405945,
      "learning_rate": 8.696508154870073e-06,
      "loss": 0.0036,
      "step": 97180
    },
    {
      "epoch": 0.6233464707070772,
      "grad_norm": 0.2632112205028534,
      "learning_rate": 8.696131241780264e-06,
      "loss": 0.0036,
      "step": 97190
    },
    {
      "epoch": 0.6234106076008633,
      "grad_norm": 0.21239538490772247,
      "learning_rate": 8.695754282375285e-06,
      "loss": 0.0036,
      "step": 97200
    },
    {
      "epoch": 0.6234747444946493,
      "grad_norm": 0.23316553235054016,
      "learning_rate": 8.695377276659867e-06,
      "loss": 0.0037,
      "step": 97210
    },
    {
      "epoch": 0.6235388813884355,
      "grad_norm": 0.0754329115152359,
      "learning_rate": 8.695000224638729e-06,
      "loss": 0.0048,
      "step": 97220
    },
    {
      "epoch": 0.6236030182822215,
      "grad_norm": 0.0366726852953434,
      "learning_rate": 8.694623126316596e-06,
      "loss": 0.0026,
      "step": 97230
    },
    {
      "epoch": 0.6236671551760077,
      "grad_norm": 0.2261466383934021,
      "learning_rate": 8.694245981698198e-06,
      "loss": 0.0031,
      "step": 97240
    },
    {
      "epoch": 0.6237312920697937,
      "grad_norm": 0.14822643995285034,
      "learning_rate": 8.693868790788256e-06,
      "loss": 0.002,
      "step": 97250
    },
    {
      "epoch": 0.6237954289635799,
      "grad_norm": 0.035959940403699875,
      "learning_rate": 8.693491553591498e-06,
      "loss": 0.0046,
      "step": 97260
    },
    {
      "epoch": 0.623859565857366,
      "grad_norm": 0.12748566269874573,
      "learning_rate": 8.69311427011265e-06,
      "loss": 0.0024,
      "step": 97270
    },
    {
      "epoch": 0.623923702751152,
      "grad_norm": 0.08464737236499786,
      "learning_rate": 8.692736940356443e-06,
      "loss": 0.0021,
      "step": 97280
    },
    {
      "epoch": 0.6239878396449382,
      "grad_norm": 0.028743699193000793,
      "learning_rate": 8.692359564327601e-06,
      "loss": 0.0044,
      "step": 97290
    },
    {
      "epoch": 0.6240519765387242,
      "grad_norm": 0.07888203114271164,
      "learning_rate": 8.691982142030857e-06,
      "loss": 0.0031,
      "step": 97300
    },
    {
      "epoch": 0.6241161134325104,
      "grad_norm": 0.08890816569328308,
      "learning_rate": 8.691604673470935e-06,
      "loss": 0.0025,
      "step": 97310
    },
    {
      "epoch": 0.6241802503262964,
      "grad_norm": 0.28032347559928894,
      "learning_rate": 8.69122715865257e-06,
      "loss": 0.0018,
      "step": 97320
    },
    {
      "epoch": 0.6242443872200826,
      "grad_norm": 0.3093320429325104,
      "learning_rate": 8.690849597580491e-06,
      "loss": 0.0029,
      "step": 97330
    },
    {
      "epoch": 0.6243085241138686,
      "grad_norm": 0.48800718784332275,
      "learning_rate": 8.690471990259429e-06,
      "loss": 0.0025,
      "step": 97340
    },
    {
      "epoch": 0.6243726610076548,
      "grad_norm": 0.15695936977863312,
      "learning_rate": 8.690094336694115e-06,
      "loss": 0.0038,
      "step": 97350
    },
    {
      "epoch": 0.6244367979014408,
      "grad_norm": 0.09537257999181747,
      "learning_rate": 8.68971663688928e-06,
      "loss": 0.0018,
      "step": 97360
    },
    {
      "epoch": 0.624500934795227,
      "grad_norm": 0.007126423995941877,
      "learning_rate": 8.68933889084966e-06,
      "loss": 0.0036,
      "step": 97370
    },
    {
      "epoch": 0.624565071689013,
      "grad_norm": 0.20154450833797455,
      "learning_rate": 8.688961098579987e-06,
      "loss": 0.0055,
      "step": 97380
    },
    {
      "epoch": 0.6246292085827991,
      "grad_norm": 0.012513830326497555,
      "learning_rate": 8.688583260084995e-06,
      "loss": 0.0069,
      "step": 97390
    },
    {
      "epoch": 0.6246933454765852,
      "grad_norm": 0.17888513207435608,
      "learning_rate": 8.688205375369418e-06,
      "loss": 0.0017,
      "step": 97400
    },
    {
      "epoch": 0.6247574823703713,
      "grad_norm": 0.1948905736207962,
      "learning_rate": 8.687827444437993e-06,
      "loss": 0.0016,
      "step": 97410
    },
    {
      "epoch": 0.6248216192641575,
      "grad_norm": 0.12320704758167267,
      "learning_rate": 8.687449467295453e-06,
      "loss": 0.0023,
      "step": 97420
    },
    {
      "epoch": 0.6248857561579435,
      "grad_norm": 0.06823863089084625,
      "learning_rate": 8.687071443946538e-06,
      "loss": 0.0021,
      "step": 97430
    },
    {
      "epoch": 0.6249498930517297,
      "grad_norm": 0.14087337255477905,
      "learning_rate": 8.68669337439598e-06,
      "loss": 0.003,
      "step": 97440
    },
    {
      "epoch": 0.6250140299455157,
      "grad_norm": 0.08630002290010452,
      "learning_rate": 8.68631525864852e-06,
      "loss": 0.0026,
      "step": 97450
    },
    {
      "epoch": 0.6250781668393018,
      "grad_norm": 0.05025516450405121,
      "learning_rate": 8.685937096708896e-06,
      "loss": 0.0038,
      "step": 97460
    },
    {
      "epoch": 0.6251423037330879,
      "grad_norm": 0.09566188603639603,
      "learning_rate": 8.685558888581845e-06,
      "loss": 0.0028,
      "step": 97470
    },
    {
      "epoch": 0.625206440626874,
      "grad_norm": 0.11923102289438248,
      "learning_rate": 8.685180634272108e-06,
      "loss": 0.0023,
      "step": 97480
    },
    {
      "epoch": 0.6252705775206601,
      "grad_norm": 0.09973512589931488,
      "learning_rate": 8.684802333784423e-06,
      "loss": 0.0033,
      "step": 97490
    },
    {
      "epoch": 0.6253347144144462,
      "grad_norm": 0.29155007004737854,
      "learning_rate": 8.684423987123532e-06,
      "loss": 0.0042,
      "step": 97500
    },
    {
      "epoch": 0.6253988513082323,
      "grad_norm": 0.1789742112159729,
      "learning_rate": 8.684045594294176e-06,
      "loss": 0.0039,
      "step": 97510
    },
    {
      "epoch": 0.6254629882020184,
      "grad_norm": 0.15239335596561432,
      "learning_rate": 8.683667155301093e-06,
      "loss": 0.0044,
      "step": 97520
    },
    {
      "epoch": 0.6255271250958044,
      "grad_norm": 0.15978781878948212,
      "learning_rate": 8.68328867014903e-06,
      "loss": 0.0024,
      "step": 97530
    },
    {
      "epoch": 0.6255912619895906,
      "grad_norm": 0.07719306647777557,
      "learning_rate": 8.682910138842725e-06,
      "loss": 0.0027,
      "step": 97540
    },
    {
      "epoch": 0.6256553988833767,
      "grad_norm": 0.0992724671959877,
      "learning_rate": 8.682531561386927e-06,
      "loss": 0.0038,
      "step": 97550
    },
    {
      "epoch": 0.6257195357771628,
      "grad_norm": 0.026066681370139122,
      "learning_rate": 8.682152937786376e-06,
      "loss": 0.0042,
      "step": 97560
    },
    {
      "epoch": 0.6257836726709489,
      "grad_norm": 0.04518354311585426,
      "learning_rate": 8.681774268045817e-06,
      "loss": 0.0032,
      "step": 97570
    },
    {
      "epoch": 0.625847809564735,
      "grad_norm": 0.06145719811320305,
      "learning_rate": 8.681395552169993e-06,
      "loss": 0.002,
      "step": 97580
    },
    {
      "epoch": 0.6259119464585211,
      "grad_norm": 0.2337571531534195,
      "learning_rate": 8.681016790163654e-06,
      "loss": 0.0027,
      "step": 97590
    },
    {
      "epoch": 0.6259760833523071,
      "grad_norm": 0.10886605829000473,
      "learning_rate": 8.680637982031543e-06,
      "loss": 0.007,
      "step": 97600
    },
    {
      "epoch": 0.6260402202460933,
      "grad_norm": 0.13759419322013855,
      "learning_rate": 8.680259127778408e-06,
      "loss": 0.0037,
      "step": 97610
    },
    {
      "epoch": 0.6261043571398793,
      "grad_norm": 0.11640704423189163,
      "learning_rate": 8.679880227408997e-06,
      "loss": 0.0075,
      "step": 97620
    },
    {
      "epoch": 0.6261684940336655,
      "grad_norm": 0.08335297554731369,
      "learning_rate": 8.679501280928055e-06,
      "loss": 0.0031,
      "step": 97630
    },
    {
      "epoch": 0.6262326309274515,
      "grad_norm": 0.7324811816215515,
      "learning_rate": 8.679122288340332e-06,
      "loss": 0.0041,
      "step": 97640
    },
    {
      "epoch": 0.6262967678212377,
      "grad_norm": 0.2840883433818817,
      "learning_rate": 8.678743249650579e-06,
      "loss": 0.0025,
      "step": 97650
    },
    {
      "epoch": 0.6263609047150237,
      "grad_norm": 0.08255748450756073,
      "learning_rate": 8.678364164863541e-06,
      "loss": 0.0021,
      "step": 97660
    },
    {
      "epoch": 0.6264250416088099,
      "grad_norm": 0.13584935665130615,
      "learning_rate": 8.677985033983974e-06,
      "loss": 0.0023,
      "step": 97670
    },
    {
      "epoch": 0.6264891785025959,
      "grad_norm": 0.2118493914604187,
      "learning_rate": 8.677605857016625e-06,
      "loss": 0.0028,
      "step": 97680
    },
    {
      "epoch": 0.626553315396382,
      "grad_norm": 0.16938568651676178,
      "learning_rate": 8.677226633966248e-06,
      "loss": 0.0031,
      "step": 97690
    },
    {
      "epoch": 0.6266174522901682,
      "grad_norm": 0.2965881824493408,
      "learning_rate": 8.67684736483759e-06,
      "loss": 0.0034,
      "step": 97700
    },
    {
      "epoch": 0.6266815891839542,
      "grad_norm": 0.1577543169260025,
      "learning_rate": 8.676468049635409e-06,
      "loss": 0.0034,
      "step": 97710
    },
    {
      "epoch": 0.6267457260777404,
      "grad_norm": 0.15394441783428192,
      "learning_rate": 8.676088688364454e-06,
      "loss": 0.0032,
      "step": 97720
    },
    {
      "epoch": 0.6268098629715264,
      "grad_norm": 0.10031639784574509,
      "learning_rate": 8.675709281029483e-06,
      "loss": 0.0021,
      "step": 97730
    },
    {
      "epoch": 0.6268739998653126,
      "grad_norm": 0.29562628269195557,
      "learning_rate": 8.675329827635246e-06,
      "loss": 0.0023,
      "step": 97740
    },
    {
      "epoch": 0.6269381367590986,
      "grad_norm": 0.06728474795818329,
      "learning_rate": 8.674950328186499e-06,
      "loss": 0.0028,
      "step": 97750
    },
    {
      "epoch": 0.6270022736528847,
      "grad_norm": 0.10657131671905518,
      "learning_rate": 8.674570782687999e-06,
      "loss": 0.003,
      "step": 97760
    },
    {
      "epoch": 0.6270664105466708,
      "grad_norm": 0.026148566976189613,
      "learning_rate": 8.6741911911445e-06,
      "loss": 0.0029,
      "step": 97770
    },
    {
      "epoch": 0.6271305474404569,
      "grad_norm": 0.36615893244743347,
      "learning_rate": 8.673811553560761e-06,
      "loss": 0.0033,
      "step": 97780
    },
    {
      "epoch": 0.627194684334243,
      "grad_norm": 0.20040257275104523,
      "learning_rate": 8.673431869941535e-06,
      "loss": 0.0028,
      "step": 97790
    },
    {
      "epoch": 0.6272588212280291,
      "grad_norm": 0.034277066588401794,
      "learning_rate": 8.673052140291584e-06,
      "loss": 0.0033,
      "step": 97800
    },
    {
      "epoch": 0.6273229581218152,
      "grad_norm": 0.06177086755633354,
      "learning_rate": 8.672672364615665e-06,
      "loss": 0.0018,
      "step": 97810
    },
    {
      "epoch": 0.6273870950156013,
      "grad_norm": 0.16230791807174683,
      "learning_rate": 8.672292542918537e-06,
      "loss": 0.0032,
      "step": 97820
    },
    {
      "epoch": 0.6274512319093873,
      "grad_norm": 0.16837529838085175,
      "learning_rate": 8.671912675204957e-06,
      "loss": 0.0036,
      "step": 97830
    },
    {
      "epoch": 0.6275153688031735,
      "grad_norm": 0.06041654571890831,
      "learning_rate": 8.671532761479688e-06,
      "loss": 0.0044,
      "step": 97840
    },
    {
      "epoch": 0.6275795056969596,
      "grad_norm": 0.10664664208889008,
      "learning_rate": 8.671152801747489e-06,
      "loss": 0.0031,
      "step": 97850
    },
    {
      "epoch": 0.6276436425907457,
      "grad_norm": 0.08125737309455872,
      "learning_rate": 8.670772796013122e-06,
      "loss": 0.0024,
      "step": 97860
    },
    {
      "epoch": 0.6277077794845318,
      "grad_norm": 0.22814540565013885,
      "learning_rate": 8.670392744281348e-06,
      "loss": 0.0057,
      "step": 97870
    },
    {
      "epoch": 0.6277719163783179,
      "grad_norm": 0.14703914523124695,
      "learning_rate": 8.670012646556931e-06,
      "loss": 0.0026,
      "step": 97880
    },
    {
      "epoch": 0.627836053272104,
      "grad_norm": 0.0672401413321495,
      "learning_rate": 8.669632502844631e-06,
      "loss": 0.0017,
      "step": 97890
    },
    {
      "epoch": 0.62790019016589,
      "grad_norm": 0.09100368618965149,
      "learning_rate": 8.669252313149214e-06,
      "loss": 0.0039,
      "step": 97900
    },
    {
      "epoch": 0.6279643270596762,
      "grad_norm": 0.05269211530685425,
      "learning_rate": 8.668872077475443e-06,
      "loss": 0.0021,
      "step": 97910
    },
    {
      "epoch": 0.6280284639534622,
      "grad_norm": 0.12907226383686066,
      "learning_rate": 8.668491795828082e-06,
      "loss": 0.0037,
      "step": 97920
    },
    {
      "epoch": 0.6280926008472484,
      "grad_norm": 0.41889700293540955,
      "learning_rate": 8.668111468211898e-06,
      "loss": 0.0015,
      "step": 97930
    },
    {
      "epoch": 0.6281567377410344,
      "grad_norm": 0.09742482751607895,
      "learning_rate": 8.667731094631656e-06,
      "loss": 0.0027,
      "step": 97940
    },
    {
      "epoch": 0.6282208746348206,
      "grad_norm": 0.15996062755584717,
      "learning_rate": 8.667350675092121e-06,
      "loss": 0.0041,
      "step": 97950
    },
    {
      "epoch": 0.6282850115286066,
      "grad_norm": 0.06355822086334229,
      "learning_rate": 8.666970209598062e-06,
      "loss": 0.0018,
      "step": 97960
    },
    {
      "epoch": 0.6283491484223928,
      "grad_norm": 0.11760903149843216,
      "learning_rate": 8.666589698154245e-06,
      "loss": 0.0033,
      "step": 97970
    },
    {
      "epoch": 0.6284132853161789,
      "grad_norm": 0.05329267680644989,
      "learning_rate": 8.666209140765437e-06,
      "loss": 0.0022,
      "step": 97980
    },
    {
      "epoch": 0.6284774222099649,
      "grad_norm": 0.2392531931400299,
      "learning_rate": 8.665828537436412e-06,
      "loss": 0.0041,
      "step": 97990
    },
    {
      "epoch": 0.6285415591037511,
      "grad_norm": 0.09396185725927353,
      "learning_rate": 8.665447888171933e-06,
      "loss": 0.0059,
      "step": 98000
    },
    {
      "epoch": 0.6286056959975371,
      "grad_norm": 0.24304385483264923,
      "learning_rate": 8.665067192976773e-06,
      "loss": 0.0023,
      "step": 98010
    },
    {
      "epoch": 0.6286698328913233,
      "grad_norm": 0.15212446451187134,
      "learning_rate": 8.664686451855701e-06,
      "loss": 0.0027,
      "step": 98020
    },
    {
      "epoch": 0.6287339697851093,
      "grad_norm": 0.20396752655506134,
      "learning_rate": 8.664305664813488e-06,
      "loss": 0.0037,
      "step": 98030
    },
    {
      "epoch": 0.6287981066788955,
      "grad_norm": 0.12427476048469543,
      "learning_rate": 8.663924831854909e-06,
      "loss": 0.003,
      "step": 98040
    },
    {
      "epoch": 0.6288622435726815,
      "grad_norm": 0.3079765737056732,
      "learning_rate": 8.66354395298473e-06,
      "loss": 0.003,
      "step": 98050
    },
    {
      "epoch": 0.6289263804664676,
      "grad_norm": 0.15311117470264435,
      "learning_rate": 8.663163028207728e-06,
      "loss": 0.0045,
      "step": 98060
    },
    {
      "epoch": 0.6289905173602537,
      "grad_norm": 0.12783585488796234,
      "learning_rate": 8.662782057528677e-06,
      "loss": 0.0037,
      "step": 98070
    },
    {
      "epoch": 0.6290546542540398,
      "grad_norm": 0.19281822443008423,
      "learning_rate": 8.662401040952349e-06,
      "loss": 0.0023,
      "step": 98080
    },
    {
      "epoch": 0.6291187911478259,
      "grad_norm": 0.09296832233667374,
      "learning_rate": 8.662019978483516e-06,
      "loss": 0.0023,
      "step": 98090
    },
    {
      "epoch": 0.629182928041612,
      "grad_norm": 0.29387709498405457,
      "learning_rate": 8.661638870126956e-06,
      "loss": 0.0059,
      "step": 98100
    },
    {
      "epoch": 0.6292470649353981,
      "grad_norm": 0.09903071820735931,
      "learning_rate": 8.661257715887446e-06,
      "loss": 0.0023,
      "step": 98110
    },
    {
      "epoch": 0.6293112018291842,
      "grad_norm": 0.11103134602308273,
      "learning_rate": 8.660876515769759e-06,
      "loss": 0.0018,
      "step": 98120
    },
    {
      "epoch": 0.6293753387229704,
      "grad_norm": 0.08121154457330704,
      "learning_rate": 8.660495269778673e-06,
      "loss": 0.0027,
      "step": 98130
    },
    {
      "epoch": 0.6294394756167564,
      "grad_norm": 0.12221727520227432,
      "learning_rate": 8.660113977918964e-06,
      "loss": 0.0026,
      "step": 98140
    },
    {
      "epoch": 0.6295036125105425,
      "grad_norm": 0.36664316058158875,
      "learning_rate": 8.659732640195411e-06,
      "loss": 0.0041,
      "step": 98150
    },
    {
      "epoch": 0.6295677494043286,
      "grad_norm": 0.16236406564712524,
      "learning_rate": 8.659351256612795e-06,
      "loss": 0.0039,
      "step": 98160
    },
    {
      "epoch": 0.6296318862981147,
      "grad_norm": 0.20031431317329407,
      "learning_rate": 8.658969827175891e-06,
      "loss": 0.004,
      "step": 98170
    },
    {
      "epoch": 0.6296960231919008,
      "grad_norm": 0.181797593832016,
      "learning_rate": 8.658588351889478e-06,
      "loss": 0.003,
      "step": 98180
    },
    {
      "epoch": 0.6297601600856869,
      "grad_norm": 0.07122781127691269,
      "learning_rate": 8.658206830758342e-06,
      "loss": 0.0033,
      "step": 98190
    },
    {
      "epoch": 0.629824296979473,
      "grad_norm": 0.15625609457492828,
      "learning_rate": 8.657825263787258e-06,
      "loss": 0.002,
      "step": 98200
    },
    {
      "epoch": 0.6298884338732591,
      "grad_norm": 0.15279321372509003,
      "learning_rate": 8.657443650981007e-06,
      "loss": 0.0037,
      "step": 98210
    },
    {
      "epoch": 0.6299525707670451,
      "grad_norm": 0.11406347155570984,
      "learning_rate": 8.657061992344377e-06,
      "loss": 0.0025,
      "step": 98220
    },
    {
      "epoch": 0.6300167076608313,
      "grad_norm": 0.13814681768417358,
      "learning_rate": 8.656680287882145e-06,
      "loss": 0.003,
      "step": 98230
    },
    {
      "epoch": 0.6300808445546173,
      "grad_norm": 0.4896417558193207,
      "learning_rate": 8.656298537599094e-06,
      "loss": 0.0031,
      "step": 98240
    },
    {
      "epoch": 0.6301449814484035,
      "grad_norm": 0.2060568630695343,
      "learning_rate": 8.655916741500013e-06,
      "loss": 0.0039,
      "step": 98250
    },
    {
      "epoch": 0.6302091183421896,
      "grad_norm": 0.11104012280702591,
      "learning_rate": 8.65553489958968e-06,
      "loss": 0.0026,
      "step": 98260
    },
    {
      "epoch": 0.6302732552359757,
      "grad_norm": 0.07269799709320068,
      "learning_rate": 8.65515301187288e-06,
      "loss": 0.0041,
      "step": 98270
    },
    {
      "epoch": 0.6303373921297618,
      "grad_norm": 0.17827273905277252,
      "learning_rate": 8.654771078354405e-06,
      "loss": 0.0028,
      "step": 98280
    },
    {
      "epoch": 0.6304015290235478,
      "grad_norm": 0.049873966723680496,
      "learning_rate": 8.654389099039034e-06,
      "loss": 0.0017,
      "step": 98290
    },
    {
      "epoch": 0.630465665917334,
      "grad_norm": 0.3912801146507263,
      "learning_rate": 8.654007073931556e-06,
      "loss": 0.0033,
      "step": 98300
    },
    {
      "epoch": 0.63052980281112,
      "grad_norm": 0.13663004338741302,
      "learning_rate": 8.653625003036757e-06,
      "loss": 0.0033,
      "step": 98310
    },
    {
      "epoch": 0.6305939397049062,
      "grad_norm": 0.07535296678543091,
      "learning_rate": 8.653242886359427e-06,
      "loss": 0.0021,
      "step": 98320
    },
    {
      "epoch": 0.6306580765986922,
      "grad_norm": 0.20357947051525116,
      "learning_rate": 8.652860723904352e-06,
      "loss": 0.0021,
      "step": 98330
    },
    {
      "epoch": 0.6307222134924784,
      "grad_norm": 0.1544470340013504,
      "learning_rate": 8.652478515676322e-06,
      "loss": 0.0024,
      "step": 98340
    },
    {
      "epoch": 0.6307863503862644,
      "grad_norm": 0.3345314562320709,
      "learning_rate": 8.652096261680125e-06,
      "loss": 0.0042,
      "step": 98350
    },
    {
      "epoch": 0.6308504872800506,
      "grad_norm": 0.18697094917297363,
      "learning_rate": 8.651713961920552e-06,
      "loss": 0.0044,
      "step": 98360
    },
    {
      "epoch": 0.6309146241738366,
      "grad_norm": 0.03998371213674545,
      "learning_rate": 8.651331616402392e-06,
      "loss": 0.0024,
      "step": 98370
    },
    {
      "epoch": 0.6309787610676227,
      "grad_norm": 0.18124331533908844,
      "learning_rate": 8.650949225130439e-06,
      "loss": 0.0036,
      "step": 98380
    },
    {
      "epoch": 0.6310428979614088,
      "grad_norm": 0.12990468740463257,
      "learning_rate": 8.650566788109482e-06,
      "loss": 0.0019,
      "step": 98390
    },
    {
      "epoch": 0.6311070348551949,
      "grad_norm": 0.0784650593996048,
      "learning_rate": 8.650184305344315e-06,
      "loss": 0.0041,
      "step": 98400
    },
    {
      "epoch": 0.6311711717489811,
      "grad_norm": 0.11469721794128418,
      "learning_rate": 8.649801776839731e-06,
      "loss": 0.0029,
      "step": 98410
    },
    {
      "epoch": 0.6312353086427671,
      "grad_norm": 0.14325492084026337,
      "learning_rate": 8.649419202600519e-06,
      "loss": 0.0029,
      "step": 98420
    },
    {
      "epoch": 0.6312994455365533,
      "grad_norm": 0.07019853591918945,
      "learning_rate": 8.649036582631479e-06,
      "loss": 0.0042,
      "step": 98430
    },
    {
      "epoch": 0.6313635824303393,
      "grad_norm": 0.1867142617702484,
      "learning_rate": 8.648653916937401e-06,
      "loss": 0.003,
      "step": 98440
    },
    {
      "epoch": 0.6314277193241254,
      "grad_norm": 0.24688571691513062,
      "learning_rate": 8.648271205523082e-06,
      "loss": 0.0041,
      "step": 98450
    },
    {
      "epoch": 0.6314918562179115,
      "grad_norm": 0.008389686234295368,
      "learning_rate": 8.647888448393317e-06,
      "loss": 0.0018,
      "step": 98460
    },
    {
      "epoch": 0.6315559931116976,
      "grad_norm": 0.24063777923583984,
      "learning_rate": 8.647505645552905e-06,
      "loss": 0.0042,
      "step": 98470
    },
    {
      "epoch": 0.6316201300054837,
      "grad_norm": 0.04026223346590996,
      "learning_rate": 8.64712279700664e-06,
      "loss": 0.0027,
      "step": 98480
    },
    {
      "epoch": 0.6316842668992698,
      "grad_norm": 0.49291589856147766,
      "learning_rate": 8.646739902759317e-06,
      "loss": 0.0048,
      "step": 98490
    },
    {
      "epoch": 0.6317484037930559,
      "grad_norm": 0.2567426562309265,
      "learning_rate": 8.646356962815738e-06,
      "loss": 0.003,
      "step": 98500
    },
    {
      "epoch": 0.631812540686842,
      "grad_norm": 0.1662866771221161,
      "learning_rate": 8.645973977180702e-06,
      "loss": 0.0019,
      "step": 98510
    },
    {
      "epoch": 0.631876677580628,
      "grad_norm": 0.0940394327044487,
      "learning_rate": 8.645590945859005e-06,
      "loss": 0.0037,
      "step": 98520
    },
    {
      "epoch": 0.6319408144744142,
      "grad_norm": 0.052625905722379684,
      "learning_rate": 8.645207868855447e-06,
      "loss": 0.0021,
      "step": 98530
    },
    {
      "epoch": 0.6320049513682003,
      "grad_norm": 0.09014574438333511,
      "learning_rate": 8.64482474617483e-06,
      "loss": 0.0015,
      "step": 98540
    },
    {
      "epoch": 0.6320690882619864,
      "grad_norm": 0.18140502274036407,
      "learning_rate": 8.644441577821955e-06,
      "loss": 0.0024,
      "step": 98550
    },
    {
      "epoch": 0.6321332251557725,
      "grad_norm": 0.047012731432914734,
      "learning_rate": 8.644058363801622e-06,
      "loss": 0.0065,
      "step": 98560
    },
    {
      "epoch": 0.6321973620495586,
      "grad_norm": 0.24054725468158722,
      "learning_rate": 8.643675104118631e-06,
      "loss": 0.0033,
      "step": 98570
    },
    {
      "epoch": 0.6322614989433447,
      "grad_norm": 0.12858544290065765,
      "learning_rate": 8.64329179877779e-06,
      "loss": 0.0027,
      "step": 98580
    },
    {
      "epoch": 0.6323256358371308,
      "grad_norm": 0.06197696551680565,
      "learning_rate": 8.642908447783898e-06,
      "loss": 0.0033,
      "step": 98590
    },
    {
      "epoch": 0.6323897727309169,
      "grad_norm": 0.08043265342712402,
      "learning_rate": 8.64252505114176e-06,
      "loss": 0.0023,
      "step": 98600
    },
    {
      "epoch": 0.6324539096247029,
      "grad_norm": 0.14281512796878815,
      "learning_rate": 8.642141608856178e-06,
      "loss": 0.003,
      "step": 98610
    },
    {
      "epoch": 0.6325180465184891,
      "grad_norm": 0.11095147579908371,
      "learning_rate": 8.64175812093196e-06,
      "loss": 0.0019,
      "step": 98620
    },
    {
      "epoch": 0.6325821834122751,
      "grad_norm": 0.14786294102668762,
      "learning_rate": 8.64137458737391e-06,
      "loss": 0.0038,
      "step": 98630
    },
    {
      "epoch": 0.6326463203060613,
      "grad_norm": 0.12939073145389557,
      "learning_rate": 8.640991008186834e-06,
      "loss": 0.0021,
      "step": 98640
    },
    {
      "epoch": 0.6327104571998473,
      "grad_norm": 0.12580779194831848,
      "learning_rate": 8.640607383375539e-06,
      "loss": 0.0012,
      "step": 98650
    },
    {
      "epoch": 0.6327745940936335,
      "grad_norm": 0.13374793529510498,
      "learning_rate": 8.64022371294483e-06,
      "loss": 0.0043,
      "step": 98660
    },
    {
      "epoch": 0.6328387309874195,
      "grad_norm": 0.14486871659755707,
      "learning_rate": 8.639839996899516e-06,
      "loss": 0.0029,
      "step": 98670
    },
    {
      "epoch": 0.6329028678812056,
      "grad_norm": 0.153135284781456,
      "learning_rate": 8.639456235244408e-06,
      "loss": 0.002,
      "step": 98680
    },
    {
      "epoch": 0.6329670047749918,
      "grad_norm": 0.171351820230484,
      "learning_rate": 8.63907242798431e-06,
      "loss": 0.0026,
      "step": 98690
    },
    {
      "epoch": 0.6330311416687778,
      "grad_norm": 0.17365366220474243,
      "learning_rate": 8.638688575124034e-06,
      "loss": 0.0034,
      "step": 98700
    },
    {
      "epoch": 0.633095278562564,
      "grad_norm": 0.28094935417175293,
      "learning_rate": 8.638304676668392e-06,
      "loss": 0.0035,
      "step": 98710
    },
    {
      "epoch": 0.63315941545635,
      "grad_norm": 0.23114296793937683,
      "learning_rate": 8.63792073262219e-06,
      "loss": 0.0027,
      "step": 98720
    },
    {
      "epoch": 0.6332235523501362,
      "grad_norm": 0.09477975219488144,
      "learning_rate": 8.63753674299024e-06,
      "loss": 0.0039,
      "step": 98730
    },
    {
      "epoch": 0.6332876892439222,
      "grad_norm": 0.26978856325149536,
      "learning_rate": 8.637152707777356e-06,
      "loss": 0.0023,
      "step": 98740
    },
    {
      "epoch": 0.6333518261377084,
      "grad_norm": 0.09891938418149948,
      "learning_rate": 8.63676862698835e-06,
      "loss": 0.0025,
      "step": 98750
    },
    {
      "epoch": 0.6334159630314944,
      "grad_norm": 0.012529808096587658,
      "learning_rate": 8.636384500628034e-06,
      "loss": 0.0027,
      "step": 98760
    },
    {
      "epoch": 0.6334800999252805,
      "grad_norm": 0.07490794360637665,
      "learning_rate": 8.636000328701222e-06,
      "loss": 0.0017,
      "step": 98770
    },
    {
      "epoch": 0.6335442368190666,
      "grad_norm": 0.11942527443170547,
      "learning_rate": 8.635616111212725e-06,
      "loss": 0.0043,
      "step": 98780
    },
    {
      "epoch": 0.6336083737128527,
      "grad_norm": 0.18825405836105347,
      "learning_rate": 8.635231848167361e-06,
      "loss": 0.0034,
      "step": 98790
    },
    {
      "epoch": 0.6336725106066388,
      "grad_norm": 0.08807805180549622,
      "learning_rate": 8.634847539569944e-06,
      "loss": 0.0047,
      "step": 98800
    },
    {
      "epoch": 0.6337366475004249,
      "grad_norm": 0.12823423743247986,
      "learning_rate": 8.63446318542529e-06,
      "loss": 0.0026,
      "step": 98810
    },
    {
      "epoch": 0.6338007843942111,
      "grad_norm": 0.08802173286676407,
      "learning_rate": 8.634078785738213e-06,
      "loss": 0.0017,
      "step": 98820
    },
    {
      "epoch": 0.6338649212879971,
      "grad_norm": 0.024244409054517746,
      "learning_rate": 8.633694340513533e-06,
      "loss": 0.0031,
      "step": 98830
    },
    {
      "epoch": 0.6339290581817832,
      "grad_norm": 0.023430783301591873,
      "learning_rate": 8.633309849756067e-06,
      "loss": 0.0036,
      "step": 98840
    },
    {
      "epoch": 0.6339931950755693,
      "grad_norm": 0.0960763692855835,
      "learning_rate": 8.63292531347063e-06,
      "loss": 0.0032,
      "step": 98850
    },
    {
      "epoch": 0.6340573319693554,
      "grad_norm": 0.20833489298820496,
      "learning_rate": 8.632540731662042e-06,
      "loss": 0.004,
      "step": 98860
    },
    {
      "epoch": 0.6341214688631415,
      "grad_norm": 0.044883791357278824,
      "learning_rate": 8.632156104335123e-06,
      "loss": 0.0027,
      "step": 98870
    },
    {
      "epoch": 0.6341856057569276,
      "grad_norm": 0.09762940555810928,
      "learning_rate": 8.631771431494694e-06,
      "loss": 0.0024,
      "step": 98880
    },
    {
      "epoch": 0.6342497426507137,
      "grad_norm": 0.18341349065303802,
      "learning_rate": 8.631386713145572e-06,
      "loss": 0.0027,
      "step": 98890
    },
    {
      "epoch": 0.6343138795444998,
      "grad_norm": 0.0660662055015564,
      "learning_rate": 8.631001949292579e-06,
      "loss": 0.003,
      "step": 98900
    },
    {
      "epoch": 0.6343780164382858,
      "grad_norm": 0.1865323781967163,
      "learning_rate": 8.630617139940536e-06,
      "loss": 0.0034,
      "step": 98910
    },
    {
      "epoch": 0.634442153332072,
      "grad_norm": 0.04117047041654587,
      "learning_rate": 8.630232285094266e-06,
      "loss": 0.0025,
      "step": 98920
    },
    {
      "epoch": 0.634506290225858,
      "grad_norm": 0.19933748245239258,
      "learning_rate": 8.629847384758592e-06,
      "loss": 0.0033,
      "step": 98930
    },
    {
      "epoch": 0.6345704271196442,
      "grad_norm": 0.2053903043270111,
      "learning_rate": 8.629462438938333e-06,
      "loss": 0.0018,
      "step": 98940
    },
    {
      "epoch": 0.6346345640134302,
      "grad_norm": 0.20126311480998993,
      "learning_rate": 8.629077447638319e-06,
      "loss": 0.0038,
      "step": 98950
    },
    {
      "epoch": 0.6346987009072164,
      "grad_norm": 0.1050531417131424,
      "learning_rate": 8.62869241086337e-06,
      "loss": 0.0028,
      "step": 98960
    },
    {
      "epoch": 0.6347628378010025,
      "grad_norm": 0.35862889885902405,
      "learning_rate": 8.628307328618312e-06,
      "loss": 0.0049,
      "step": 98970
    },
    {
      "epoch": 0.6348269746947885,
      "grad_norm": 0.14179842174053192,
      "learning_rate": 8.62792220090797e-06,
      "loss": 0.003,
      "step": 98980
    },
    {
      "epoch": 0.6348911115885747,
      "grad_norm": 0.13463851809501648,
      "learning_rate": 8.627537027737169e-06,
      "loss": 0.003,
      "step": 98990
    },
    {
      "epoch": 0.6349552484823607,
      "grad_norm": 0.08901742845773697,
      "learning_rate": 8.627151809110737e-06,
      "loss": 0.0027,
      "step": 99000
    },
    {
      "epoch": 0.6350193853761469,
      "grad_norm": 0.1018950343132019,
      "learning_rate": 8.6267665450335e-06,
      "loss": 0.003,
      "step": 99010
    },
    {
      "epoch": 0.6350835222699329,
      "grad_norm": 0.13207198679447174,
      "learning_rate": 8.626381235510286e-06,
      "loss": 0.0016,
      "step": 99020
    },
    {
      "epoch": 0.6351476591637191,
      "grad_norm": 0.15876226127147675,
      "learning_rate": 8.625995880545926e-06,
      "loss": 0.0027,
      "step": 99030
    },
    {
      "epoch": 0.6352117960575051,
      "grad_norm": 0.05085441470146179,
      "learning_rate": 8.625610480145244e-06,
      "loss": 0.002,
      "step": 99040
    },
    {
      "epoch": 0.6352759329512913,
      "grad_norm": 0.04116792231798172,
      "learning_rate": 8.625225034313071e-06,
      "loss": 0.0037,
      "step": 99050
    },
    {
      "epoch": 0.6353400698450773,
      "grad_norm": 0.18012458086013794,
      "learning_rate": 8.624839543054238e-06,
      "loss": 0.003,
      "step": 99060
    },
    {
      "epoch": 0.6354042067388634,
      "grad_norm": 0.16937007009983063,
      "learning_rate": 8.624454006373577e-06,
      "loss": 0.0022,
      "step": 99070
    },
    {
      "epoch": 0.6354683436326495,
      "grad_norm": 0.33522242307662964,
      "learning_rate": 8.624068424275913e-06,
      "loss": 0.004,
      "step": 99080
    },
    {
      "epoch": 0.6355324805264356,
      "grad_norm": 0.07128586620092392,
      "learning_rate": 8.623682796766085e-06,
      "loss": 0.0025,
      "step": 99090
    },
    {
      "epoch": 0.6355966174202218,
      "grad_norm": 0.1760418862104416,
      "learning_rate": 8.62329712384892e-06,
      "loss": 0.0024,
      "step": 99100
    },
    {
      "epoch": 0.6356607543140078,
      "grad_norm": 0.19041508436203003,
      "learning_rate": 8.622911405529253e-06,
      "loss": 0.0048,
      "step": 99110
    },
    {
      "epoch": 0.635724891207794,
      "grad_norm": 0.16518241167068481,
      "learning_rate": 8.622525641811917e-06,
      "loss": 0.0034,
      "step": 99120
    },
    {
      "epoch": 0.63578902810158,
      "grad_norm": 0.034229397773742676,
      "learning_rate": 8.622139832701744e-06,
      "loss": 0.0029,
      "step": 99130
    },
    {
      "epoch": 0.6358531649953661,
      "grad_norm": 0.19417065382003784,
      "learning_rate": 8.621753978203572e-06,
      "loss": 0.0035,
      "step": 99140
    },
    {
      "epoch": 0.6359173018891522,
      "grad_norm": 0.0759410634636879,
      "learning_rate": 8.621368078322234e-06,
      "loss": 0.0025,
      "step": 99150
    },
    {
      "epoch": 0.6359814387829383,
      "grad_norm": 0.07072167843580246,
      "learning_rate": 8.620982133062566e-06,
      "loss": 0.0026,
      "step": 99160
    },
    {
      "epoch": 0.6360455756767244,
      "grad_norm": 0.16522380709648132,
      "learning_rate": 8.620596142429402e-06,
      "loss": 0.0047,
      "step": 99170
    },
    {
      "epoch": 0.6361097125705105,
      "grad_norm": 0.22960606217384338,
      "learning_rate": 8.620210106427584e-06,
      "loss": 0.0031,
      "step": 99180
    },
    {
      "epoch": 0.6361738494642966,
      "grad_norm": 0.0563676580786705,
      "learning_rate": 8.619824025061945e-06,
      "loss": 0.0026,
      "step": 99190
    },
    {
      "epoch": 0.6362379863580827,
      "grad_norm": 0.1583438366651535,
      "learning_rate": 8.619437898337323e-06,
      "loss": 0.0063,
      "step": 99200
    },
    {
      "epoch": 0.6363021232518687,
      "grad_norm": 0.5747680068016052,
      "learning_rate": 8.619051726258557e-06,
      "loss": 0.0021,
      "step": 99210
    },
    {
      "epoch": 0.6363662601456549,
      "grad_norm": 0.1903034895658493,
      "learning_rate": 8.618665508830488e-06,
      "loss": 0.0053,
      "step": 99220
    },
    {
      "epoch": 0.6364303970394409,
      "grad_norm": 0.15191338956356049,
      "learning_rate": 8.618279246057953e-06,
      "loss": 0.003,
      "step": 99230
    },
    {
      "epoch": 0.6364945339332271,
      "grad_norm": 0.18410761654376984,
      "learning_rate": 8.617892937945794e-06,
      "loss": 0.004,
      "step": 99240
    },
    {
      "epoch": 0.6365586708270132,
      "grad_norm": 0.16988497972488403,
      "learning_rate": 8.61750658449885e-06,
      "loss": 0.0036,
      "step": 99250
    },
    {
      "epoch": 0.6366228077207993,
      "grad_norm": 0.07580164074897766,
      "learning_rate": 8.61712018572196e-06,
      "loss": 0.0022,
      "step": 99260
    },
    {
      "epoch": 0.6366869446145854,
      "grad_norm": 0.11030968278646469,
      "learning_rate": 8.616733741619973e-06,
      "loss": 0.0048,
      "step": 99270
    },
    {
      "epoch": 0.6367510815083715,
      "grad_norm": 0.07485426962375641,
      "learning_rate": 8.616347252197727e-06,
      "loss": 0.0025,
      "step": 99280
    },
    {
      "epoch": 0.6368152184021576,
      "grad_norm": 0.16262781620025635,
      "learning_rate": 8.615960717460065e-06,
      "loss": 0.0024,
      "step": 99290
    },
    {
      "epoch": 0.6368793552959436,
      "grad_norm": 0.12966498732566833,
      "learning_rate": 8.61557413741183e-06,
      "loss": 0.0025,
      "step": 99300
    },
    {
      "epoch": 0.6369434921897298,
      "grad_norm": 0.41973453760147095,
      "learning_rate": 8.615187512057867e-06,
      "loss": 0.0021,
      "step": 99310
    },
    {
      "epoch": 0.6370076290835158,
      "grad_norm": 0.1630977839231491,
      "learning_rate": 8.614800841403023e-06,
      "loss": 0.0026,
      "step": 99320
    },
    {
      "epoch": 0.637071765977302,
      "grad_norm": 0.1655716449022293,
      "learning_rate": 8.614414125452139e-06,
      "loss": 0.0019,
      "step": 99330
    },
    {
      "epoch": 0.637135902871088,
      "grad_norm": 0.06753179430961609,
      "learning_rate": 8.614027364210064e-06,
      "loss": 0.0031,
      "step": 99340
    },
    {
      "epoch": 0.6372000397648742,
      "grad_norm": 0.08307766169309616,
      "learning_rate": 8.613640557681642e-06,
      "loss": 0.0067,
      "step": 99350
    },
    {
      "epoch": 0.6372641766586602,
      "grad_norm": 0.10214082151651382,
      "learning_rate": 8.613253705871723e-06,
      "loss": 0.0027,
      "step": 99360
    },
    {
      "epoch": 0.6373283135524463,
      "grad_norm": 0.021376170217990875,
      "learning_rate": 8.61286680878515e-06,
      "loss": 0.0032,
      "step": 99370
    },
    {
      "epoch": 0.6373924504462324,
      "grad_norm": 0.16414782404899597,
      "learning_rate": 8.612479866426775e-06,
      "loss": 0.0042,
      "step": 99380
    },
    {
      "epoch": 0.6374565873400185,
      "grad_norm": 0.21608483791351318,
      "learning_rate": 8.612092878801446e-06,
      "loss": 0.0029,
      "step": 99390
    },
    {
      "epoch": 0.6375207242338047,
      "grad_norm": 0.5427488684654236,
      "learning_rate": 8.611705845914011e-06,
      "loss": 0.0079,
      "step": 99400
    },
    {
      "epoch": 0.6375848611275907,
      "grad_norm": 0.10978326946496964,
      "learning_rate": 8.611318767769321e-06,
      "loss": 0.0036,
      "step": 99410
    },
    {
      "epoch": 0.6376489980213769,
      "grad_norm": 0.23859313130378723,
      "learning_rate": 8.610931644372226e-06,
      "loss": 0.0026,
      "step": 99420
    },
    {
      "epoch": 0.6377131349151629,
      "grad_norm": 0.1963741034269333,
      "learning_rate": 8.610544475727576e-06,
      "loss": 0.0019,
      "step": 99430
    },
    {
      "epoch": 0.637777271808949,
      "grad_norm": 0.31492742896080017,
      "learning_rate": 8.610157261840224e-06,
      "loss": 0.004,
      "step": 99440
    },
    {
      "epoch": 0.6378414087027351,
      "grad_norm": 0.15649573504924774,
      "learning_rate": 8.609770002715022e-06,
      "loss": 0.003,
      "step": 99450
    },
    {
      "epoch": 0.6379055455965212,
      "grad_norm": 0.10112041980028152,
      "learning_rate": 8.60938269835682e-06,
      "loss": 0.0032,
      "step": 99460
    },
    {
      "epoch": 0.6379696824903073,
      "grad_norm": 0.18804387748241425,
      "learning_rate": 8.608995348770474e-06,
      "loss": 0.0037,
      "step": 99470
    },
    {
      "epoch": 0.6380338193840934,
      "grad_norm": 0.07315003871917725,
      "learning_rate": 8.60860795396084e-06,
      "loss": 0.0056,
      "step": 99480
    },
    {
      "epoch": 0.6380979562778795,
      "grad_norm": 0.08535873889923096,
      "learning_rate": 8.608220513932766e-06,
      "loss": 0.003,
      "step": 99490
    },
    {
      "epoch": 0.6381620931716656,
      "grad_norm": 0.28759923577308655,
      "learning_rate": 8.607833028691111e-06,
      "loss": 0.0026,
      "step": 99500
    },
    {
      "epoch": 0.6382262300654516,
      "grad_norm": 0.2628440856933594,
      "learning_rate": 8.607445498240729e-06,
      "loss": 0.004,
      "step": 99510
    },
    {
      "epoch": 0.6382903669592378,
      "grad_norm": 0.18443447351455688,
      "learning_rate": 8.607057922586477e-06,
      "loss": 0.0033,
      "step": 99520
    },
    {
      "epoch": 0.638354503853024,
      "grad_norm": 0.08273148536682129,
      "learning_rate": 8.606670301733212e-06,
      "loss": 0.003,
      "step": 99530
    },
    {
      "epoch": 0.63841864074681,
      "grad_norm": 0.25067704916000366,
      "learning_rate": 8.606282635685789e-06,
      "loss": 0.0048,
      "step": 99540
    },
    {
      "epoch": 0.6384827776405961,
      "grad_norm": 0.17173099517822266,
      "learning_rate": 8.60589492444907e-06,
      "loss": 0.006,
      "step": 99550
    },
    {
      "epoch": 0.6385469145343822,
      "grad_norm": 0.1476057916879654,
      "learning_rate": 8.60550716802791e-06,
      "loss": 0.0019,
      "step": 99560
    },
    {
      "epoch": 0.6386110514281683,
      "grad_norm": 0.04123391956090927,
      "learning_rate": 8.605119366427165e-06,
      "loss": 0.002,
      "step": 99570
    },
    {
      "epoch": 0.6386751883219544,
      "grad_norm": 0.1771586537361145,
      "learning_rate": 8.6047315196517e-06,
      "loss": 0.005,
      "step": 99580
    },
    {
      "epoch": 0.6387393252157405,
      "grad_norm": 0.21558478474617004,
      "learning_rate": 8.604343627706374e-06,
      "loss": 0.0038,
      "step": 99590
    },
    {
      "epoch": 0.6388034621095265,
      "grad_norm": 0.1837262660264969,
      "learning_rate": 8.603955690596044e-06,
      "loss": 0.0047,
      "step": 99600
    },
    {
      "epoch": 0.6388675990033127,
      "grad_norm": 0.07837624102830887,
      "learning_rate": 8.603567708325576e-06,
      "loss": 0.0025,
      "step": 99610
    },
    {
      "epoch": 0.6389317358970987,
      "grad_norm": 0.008626329712569714,
      "learning_rate": 8.603179680899827e-06,
      "loss": 0.0034,
      "step": 99620
    },
    {
      "epoch": 0.6389958727908849,
      "grad_norm": 0.05696270242333412,
      "learning_rate": 8.602791608323662e-06,
      "loss": 0.005,
      "step": 99630
    },
    {
      "epoch": 0.6390600096846709,
      "grad_norm": 0.07516892999410629,
      "learning_rate": 8.602403490601943e-06,
      "loss": 0.0035,
      "step": 99640
    },
    {
      "epoch": 0.6391241465784571,
      "grad_norm": 0.15452761948108673,
      "learning_rate": 8.602015327739535e-06,
      "loss": 0.0038,
      "step": 99650
    },
    {
      "epoch": 0.6391882834722431,
      "grad_norm": 0.09730653464794159,
      "learning_rate": 8.6016271197413e-06,
      "loss": 0.0039,
      "step": 99660
    },
    {
      "epoch": 0.6392524203660293,
      "grad_norm": 0.23667676746845245,
      "learning_rate": 8.601238866612103e-06,
      "loss": 0.0031,
      "step": 99670
    },
    {
      "epoch": 0.6393165572598154,
      "grad_norm": 0.17529785633087158,
      "learning_rate": 8.60085056835681e-06,
      "loss": 0.0025,
      "step": 99680
    },
    {
      "epoch": 0.6393806941536014,
      "grad_norm": 0.08143072575330734,
      "learning_rate": 8.600462224980283e-06,
      "loss": 0.0044,
      "step": 99690
    },
    {
      "epoch": 0.6394448310473876,
      "grad_norm": 0.07394933700561523,
      "learning_rate": 8.600073836487392e-06,
      "loss": 0.005,
      "step": 99700
    },
    {
      "epoch": 0.6395089679411736,
      "grad_norm": 0.058153364807367325,
      "learning_rate": 8.599685402883004e-06,
      "loss": 0.001,
      "step": 99710
    },
    {
      "epoch": 0.6395731048349598,
      "grad_norm": 0.4131733775138855,
      "learning_rate": 8.599296924171987e-06,
      "loss": 0.0034,
      "step": 99720
    },
    {
      "epoch": 0.6396372417287458,
      "grad_norm": 0.31498780846595764,
      "learning_rate": 8.598908400359205e-06,
      "loss": 0.0031,
      "step": 99730
    },
    {
      "epoch": 0.639701378622532,
      "grad_norm": 0.24626566469669342,
      "learning_rate": 8.59851983144953e-06,
      "loss": 0.003,
      "step": 99740
    },
    {
      "epoch": 0.639765515516318,
      "grad_norm": 0.06797293573617935,
      "learning_rate": 8.598131217447826e-06,
      "loss": 0.0026,
      "step": 99750
    },
    {
      "epoch": 0.6398296524101041,
      "grad_norm": 0.090883269906044,
      "learning_rate": 8.59774255835897e-06,
      "loss": 0.0051,
      "step": 99760
    },
    {
      "epoch": 0.6398937893038902,
      "grad_norm": 0.08502686023712158,
      "learning_rate": 8.59735385418783e-06,
      "loss": 0.0038,
      "step": 99770
    },
    {
      "epoch": 0.6399579261976763,
      "grad_norm": 0.26698365807533264,
      "learning_rate": 8.596965104939271e-06,
      "loss": 0.0039,
      "step": 99780
    },
    {
      "epoch": 0.6400220630914624,
      "grad_norm": 0.15803366899490356,
      "learning_rate": 8.59657631061817e-06,
      "loss": 0.0013,
      "step": 99790
    },
    {
      "epoch": 0.6400861999852485,
      "grad_norm": 0.0447712242603302,
      "learning_rate": 8.596187471229398e-06,
      "loss": 0.0037,
      "step": 99800
    },
    {
      "epoch": 0.6401503368790347,
      "grad_norm": 0.14180706441402435,
      "learning_rate": 8.595798586777827e-06,
      "loss": 0.0035,
      "step": 99810
    },
    {
      "epoch": 0.6402144737728207,
      "grad_norm": 0.556554913520813,
      "learning_rate": 8.59540965726833e-06,
      "loss": 0.0069,
      "step": 99820
    },
    {
      "epoch": 0.6402786106666069,
      "grad_norm": 0.0933576375246048,
      "learning_rate": 8.595020682705778e-06,
      "loss": 0.0032,
      "step": 99830
    },
    {
      "epoch": 0.6403427475603929,
      "grad_norm": 0.09699434041976929,
      "learning_rate": 8.59463166309505e-06,
      "loss": 0.0049,
      "step": 99840
    },
    {
      "epoch": 0.640406884454179,
      "grad_norm": 0.21019862592220306,
      "learning_rate": 8.594242598441018e-06,
      "loss": 0.0048,
      "step": 99850
    },
    {
      "epoch": 0.6404710213479651,
      "grad_norm": 0.08167964965105057,
      "learning_rate": 8.593853488748557e-06,
      "loss": 0.0025,
      "step": 99860
    },
    {
      "epoch": 0.6405351582417512,
      "grad_norm": 0.06507817655801773,
      "learning_rate": 8.593464334022543e-06,
      "loss": 0.0019,
      "step": 99870
    },
    {
      "epoch": 0.6405992951355373,
      "grad_norm": 0.09788227826356888,
      "learning_rate": 8.593075134267852e-06,
      "loss": 0.0015,
      "step": 99880
    },
    {
      "epoch": 0.6406634320293234,
      "grad_norm": 0.12715961039066315,
      "learning_rate": 8.592685889489363e-06,
      "loss": 0.0028,
      "step": 99890
    },
    {
      "epoch": 0.6407275689231094,
      "grad_norm": 0.12133368849754333,
      "learning_rate": 8.592296599691952e-06,
      "loss": 0.0042,
      "step": 99900
    },
    {
      "epoch": 0.6407917058168956,
      "grad_norm": 0.1371624916791916,
      "learning_rate": 8.591907264880497e-06,
      "loss": 0.0028,
      "step": 99910
    },
    {
      "epoch": 0.6408558427106816,
      "grad_norm": 0.14875027537345886,
      "learning_rate": 8.591517885059877e-06,
      "loss": 0.0069,
      "step": 99920
    },
    {
      "epoch": 0.6409199796044678,
      "grad_norm": 0.1025613322854042,
      "learning_rate": 8.59112846023497e-06,
      "loss": 0.0025,
      "step": 99930
    },
    {
      "epoch": 0.6409841164982538,
      "grad_norm": 0.28941577672958374,
      "learning_rate": 8.59073899041066e-06,
      "loss": 0.003,
      "step": 99940
    },
    {
      "epoch": 0.64104825339204,
      "grad_norm": 0.08539040386676788,
      "learning_rate": 8.59034947559182e-06,
      "loss": 0.0025,
      "step": 99950
    },
    {
      "epoch": 0.6411123902858261,
      "grad_norm": 0.1081579178571701,
      "learning_rate": 8.589959915783337e-06,
      "loss": 0.0019,
      "step": 99960
    },
    {
      "epoch": 0.6411765271796122,
      "grad_norm": 0.13859598338603973,
      "learning_rate": 8.589570310990093e-06,
      "loss": 0.0031,
      "step": 99970
    },
    {
      "epoch": 0.6412406640733983,
      "grad_norm": 0.08517938107252121,
      "learning_rate": 8.589180661216962e-06,
      "loss": 0.0027,
      "step": 99980
    },
    {
      "epoch": 0.6413048009671843,
      "grad_norm": 0.23588401079177856,
      "learning_rate": 8.588790966468836e-06,
      "loss": 0.0035,
      "step": 99990
    },
    {
      "epoch": 0.6413689378609705,
      "grad_norm": 0.34334316849708557,
      "learning_rate": 8.588401226750595e-06,
      "loss": 0.0031,
      "step": 100000
    },
    {
      "epoch": 0.6414330747547565,
      "grad_norm": 0.2593560814857483,
      "learning_rate": 8.58801144206712e-06,
      "loss": 0.0038,
      "step": 100010
    },
    {
      "epoch": 0.6414972116485427,
      "grad_norm": 0.1017594113945961,
      "learning_rate": 8.587621612423298e-06,
      "loss": 0.0023,
      "step": 100020
    },
    {
      "epoch": 0.6415613485423287,
      "grad_norm": 0.09830541163682938,
      "learning_rate": 8.587231737824013e-06,
      "loss": 0.003,
      "step": 100030
    },
    {
      "epoch": 0.6416254854361149,
      "grad_norm": 0.07964377105236053,
      "learning_rate": 8.586841818274152e-06,
      "loss": 0.0013,
      "step": 100040
    },
    {
      "epoch": 0.6416896223299009,
      "grad_norm": 0.049259621649980545,
      "learning_rate": 8.586451853778597e-06,
      "loss": 0.0025,
      "step": 100050
    },
    {
      "epoch": 0.641753759223687,
      "grad_norm": 0.10759458690881729,
      "learning_rate": 8.58606184434224e-06,
      "loss": 0.0029,
      "step": 100060
    },
    {
      "epoch": 0.6418178961174731,
      "grad_norm": 0.06379517912864685,
      "learning_rate": 8.585671789969963e-06,
      "loss": 0.0042,
      "step": 100070
    },
    {
      "epoch": 0.6418820330112592,
      "grad_norm": 0.20491823554039001,
      "learning_rate": 8.585281690666656e-06,
      "loss": 0.0028,
      "step": 100080
    },
    {
      "epoch": 0.6419461699050454,
      "grad_norm": 0.4074902832508087,
      "learning_rate": 8.584891546437206e-06,
      "loss": 0.0049,
      "step": 100090
    },
    {
      "epoch": 0.6420103067988314,
      "grad_norm": 0.18549945950508118,
      "learning_rate": 8.584501357286505e-06,
      "loss": 0.0023,
      "step": 100100
    },
    {
      "epoch": 0.6420744436926176,
      "grad_norm": 0.1402589976787567,
      "learning_rate": 8.584111123219438e-06,
      "loss": 0.0025,
      "step": 100110
    },
    {
      "epoch": 0.6421385805864036,
      "grad_norm": 0.10942061245441437,
      "learning_rate": 8.5837208442409e-06,
      "loss": 0.0028,
      "step": 100120
    },
    {
      "epoch": 0.6422027174801898,
      "grad_norm": 0.16121609508991241,
      "learning_rate": 8.583330520355777e-06,
      "loss": 0.0024,
      "step": 100130
    },
    {
      "epoch": 0.6422668543739758,
      "grad_norm": 0.1497829556465149,
      "learning_rate": 8.582940151568961e-06,
      "loss": 0.0023,
      "step": 100140
    },
    {
      "epoch": 0.6423309912677619,
      "grad_norm": 0.10273135453462601,
      "learning_rate": 8.582549737885346e-06,
      "loss": 0.0048,
      "step": 100150
    },
    {
      "epoch": 0.642395128161548,
      "grad_norm": 0.172627791762352,
      "learning_rate": 8.582159279309821e-06,
      "loss": 0.0028,
      "step": 100160
    },
    {
      "epoch": 0.6424592650553341,
      "grad_norm": 0.18407230079174042,
      "learning_rate": 8.58176877584728e-06,
      "loss": 0.0051,
      "step": 100170
    },
    {
      "epoch": 0.6425234019491202,
      "grad_norm": 0.14890094101428986,
      "learning_rate": 8.581378227502616e-06,
      "loss": 0.0064,
      "step": 100180
    },
    {
      "epoch": 0.6425875388429063,
      "grad_norm": 0.11829569190740585,
      "learning_rate": 8.580987634280727e-06,
      "loss": 0.0028,
      "step": 100190
    },
    {
      "epoch": 0.6426516757366924,
      "grad_norm": 0.05861892178654671,
      "learning_rate": 8.580596996186499e-06,
      "loss": 0.0015,
      "step": 100200
    },
    {
      "epoch": 0.6427158126304785,
      "grad_norm": 0.09609236568212509,
      "learning_rate": 8.580206313224833e-06,
      "loss": 0.0053,
      "step": 100210
    },
    {
      "epoch": 0.6427799495242645,
      "grad_norm": 0.2850412130355835,
      "learning_rate": 8.579815585400624e-06,
      "loss": 0.0021,
      "step": 100220
    },
    {
      "epoch": 0.6428440864180507,
      "grad_norm": 0.030681833624839783,
      "learning_rate": 8.579424812718767e-06,
      "loss": 0.0019,
      "step": 100230
    },
    {
      "epoch": 0.6429082233118368,
      "grad_norm": 0.07587326318025589,
      "learning_rate": 8.579033995184159e-06,
      "loss": 0.0039,
      "step": 100240
    },
    {
      "epoch": 0.6429723602056229,
      "grad_norm": 0.03634805977344513,
      "learning_rate": 8.578643132801697e-06,
      "loss": 0.0016,
      "step": 100250
    },
    {
      "epoch": 0.643036497099409,
      "grad_norm": 0.272158145904541,
      "learning_rate": 8.578252225576278e-06,
      "loss": 0.0047,
      "step": 100260
    },
    {
      "epoch": 0.6431006339931951,
      "grad_norm": 0.20765464007854462,
      "learning_rate": 8.577861273512801e-06,
      "loss": 0.0028,
      "step": 100270
    },
    {
      "epoch": 0.6431647708869812,
      "grad_norm": 0.1623903512954712,
      "learning_rate": 8.577470276616166e-06,
      "loss": 0.0027,
      "step": 100280
    },
    {
      "epoch": 0.6432289077807672,
      "grad_norm": 0.24394509196281433,
      "learning_rate": 8.577079234891273e-06,
      "loss": 0.0029,
      "step": 100290
    },
    {
      "epoch": 0.6432930446745534,
      "grad_norm": 0.09946839511394501,
      "learning_rate": 8.57668814834302e-06,
      "loss": 0.0015,
      "step": 100300
    },
    {
      "epoch": 0.6433571815683394,
      "grad_norm": 0.01027774065732956,
      "learning_rate": 8.576297016976307e-06,
      "loss": 0.005,
      "step": 100310
    },
    {
      "epoch": 0.6434213184621256,
      "grad_norm": 0.140450119972229,
      "learning_rate": 8.575905840796038e-06,
      "loss": 0.0045,
      "step": 100320
    },
    {
      "epoch": 0.6434854553559116,
      "grad_norm": 0.08717735856771469,
      "learning_rate": 8.575514619807112e-06,
      "loss": 0.0061,
      "step": 100330
    },
    {
      "epoch": 0.6435495922496978,
      "grad_norm": 0.21483832597732544,
      "learning_rate": 8.575123354014432e-06,
      "loss": 0.0033,
      "step": 100340
    },
    {
      "epoch": 0.6436137291434838,
      "grad_norm": 0.1500617414712906,
      "learning_rate": 8.574732043422902e-06,
      "loss": 0.0024,
      "step": 100350
    },
    {
      "epoch": 0.64367786603727,
      "grad_norm": 0.11223375052213669,
      "learning_rate": 8.574340688037426e-06,
      "loss": 0.0025,
      "step": 100360
    },
    {
      "epoch": 0.6437420029310561,
      "grad_norm": 0.02819480001926422,
      "learning_rate": 8.573949287862905e-06,
      "loss": 0.0016,
      "step": 100370
    },
    {
      "epoch": 0.6438061398248421,
      "grad_norm": 0.0947762280702591,
      "learning_rate": 8.573557842904245e-06,
      "loss": 0.0016,
      "step": 100380
    },
    {
      "epoch": 0.6438702767186283,
      "grad_norm": 0.2611057758331299,
      "learning_rate": 8.573166353166352e-06,
      "loss": 0.0034,
      "step": 100390
    },
    {
      "epoch": 0.6439344136124143,
      "grad_norm": 0.1276160478591919,
      "learning_rate": 8.57277481865413e-06,
      "loss": 0.003,
      "step": 100400
    },
    {
      "epoch": 0.6439985505062005,
      "grad_norm": 0.10863440483808517,
      "learning_rate": 8.572383239372488e-06,
      "loss": 0.0024,
      "step": 100410
    },
    {
      "epoch": 0.6440626873999865,
      "grad_norm": 0.10743433982133865,
      "learning_rate": 8.57199161532633e-06,
      "loss": 0.0026,
      "step": 100420
    },
    {
      "epoch": 0.6441268242937727,
      "grad_norm": 0.2762799561023712,
      "learning_rate": 8.571599946520563e-06,
      "loss": 0.004,
      "step": 100430
    },
    {
      "epoch": 0.6441909611875587,
      "grad_norm": 0.050531577318906784,
      "learning_rate": 8.571208232960097e-06,
      "loss": 0.0044,
      "step": 100440
    },
    {
      "epoch": 0.6442550980813448,
      "grad_norm": 0.33933791518211365,
      "learning_rate": 8.570816474649838e-06,
      "loss": 0.007,
      "step": 100450
    },
    {
      "epoch": 0.6443192349751309,
      "grad_norm": 0.13272233307361603,
      "learning_rate": 8.5704246715947e-06,
      "loss": 0.0034,
      "step": 100460
    },
    {
      "epoch": 0.644383371868917,
      "grad_norm": 0.05244365707039833,
      "learning_rate": 8.570032823799588e-06,
      "loss": 0.0015,
      "step": 100470
    },
    {
      "epoch": 0.6444475087627031,
      "grad_norm": 0.08542118966579437,
      "learning_rate": 8.569640931269411e-06,
      "loss": 0.0031,
      "step": 100480
    },
    {
      "epoch": 0.6445116456564892,
      "grad_norm": 0.4075598418712616,
      "learning_rate": 8.569248994009083e-06,
      "loss": 0.0049,
      "step": 100490
    },
    {
      "epoch": 0.6445757825502753,
      "grad_norm": 0.055834926664829254,
      "learning_rate": 8.568857012023515e-06,
      "loss": 0.002,
      "step": 100500
    },
    {
      "epoch": 0.6446399194440614,
      "grad_norm": 0.11826679110527039,
      "learning_rate": 8.568464985317618e-06,
      "loss": 0.0028,
      "step": 100510
    },
    {
      "epoch": 0.6447040563378476,
      "grad_norm": 0.5990362763404846,
      "learning_rate": 8.568072913896304e-06,
      "loss": 0.0029,
      "step": 100520
    },
    {
      "epoch": 0.6447681932316336,
      "grad_norm": 0.03626729175448418,
      "learning_rate": 8.567680797764486e-06,
      "loss": 0.0026,
      "step": 100530
    },
    {
      "epoch": 0.6448323301254197,
      "grad_norm": 0.0920369103550911,
      "learning_rate": 8.567288636927078e-06,
      "loss": 0.0018,
      "step": 100540
    },
    {
      "epoch": 0.6448964670192058,
      "grad_norm": 0.3881969749927521,
      "learning_rate": 8.566896431388994e-06,
      "loss": 0.0051,
      "step": 100550
    },
    {
      "epoch": 0.6449606039129919,
      "grad_norm": 0.1653403490781784,
      "learning_rate": 8.56650418115515e-06,
      "loss": 0.0023,
      "step": 100560
    },
    {
      "epoch": 0.645024740806778,
      "grad_norm": 0.271144300699234,
      "learning_rate": 8.566111886230457e-06,
      "loss": 0.0028,
      "step": 100570
    },
    {
      "epoch": 0.6450888777005641,
      "grad_norm": 0.15923184156417847,
      "learning_rate": 8.565719546619835e-06,
      "loss": 0.002,
      "step": 100580
    },
    {
      "epoch": 0.6451530145943501,
      "grad_norm": 0.09394088387489319,
      "learning_rate": 8.565327162328199e-06,
      "loss": 0.0027,
      "step": 100590
    },
    {
      "epoch": 0.6452171514881363,
      "grad_norm": 0.23985207080841064,
      "learning_rate": 8.564934733360463e-06,
      "loss": 0.0031,
      "step": 100600
    },
    {
      "epoch": 0.6452812883819223,
      "grad_norm": 0.2356472611427307,
      "learning_rate": 8.56454225972155e-06,
      "loss": 0.0023,
      "step": 100610
    },
    {
      "epoch": 0.6453454252757085,
      "grad_norm": 0.25987300276756287,
      "learning_rate": 8.564149741416372e-06,
      "loss": 0.0032,
      "step": 100620
    },
    {
      "epoch": 0.6454095621694945,
      "grad_norm": 0.18951743841171265,
      "learning_rate": 8.563757178449854e-06,
      "loss": 0.0038,
      "step": 100630
    },
    {
      "epoch": 0.6454736990632807,
      "grad_norm": 0.14653554558753967,
      "learning_rate": 8.56336457082691e-06,
      "loss": 0.0023,
      "step": 100640
    },
    {
      "epoch": 0.6455378359570667,
      "grad_norm": 0.10666285455226898,
      "learning_rate": 8.56297191855246e-06,
      "loss": 0.0024,
      "step": 100650
    },
    {
      "epoch": 0.6456019728508529,
      "grad_norm": 0.4330354332923889,
      "learning_rate": 8.562579221631427e-06,
      "loss": 0.0062,
      "step": 100660
    },
    {
      "epoch": 0.645666109744639,
      "grad_norm": 0.011996464803814888,
      "learning_rate": 8.562186480068727e-06,
      "loss": 0.002,
      "step": 100670
    },
    {
      "epoch": 0.645730246638425,
      "grad_norm": 0.10304580628871918,
      "learning_rate": 8.56179369386929e-06,
      "loss": 0.0034,
      "step": 100680
    },
    {
      "epoch": 0.6457943835322112,
      "grad_norm": 0.22657014429569244,
      "learning_rate": 8.561400863038029e-06,
      "loss": 0.0027,
      "step": 100690
    },
    {
      "epoch": 0.6458585204259972,
      "grad_norm": 0.057219889014959335,
      "learning_rate": 8.56100798757987e-06,
      "loss": 0.0015,
      "step": 100700
    },
    {
      "epoch": 0.6459226573197834,
      "grad_norm": 0.11843618005514145,
      "learning_rate": 8.560615067499737e-06,
      "loss": 0.0037,
      "step": 100710
    },
    {
      "epoch": 0.6459867942135694,
      "grad_norm": 0.10030394047498703,
      "learning_rate": 8.56022210280255e-06,
      "loss": 0.0036,
      "step": 100720
    },
    {
      "epoch": 0.6460509311073556,
      "grad_norm": 0.13362765312194824,
      "learning_rate": 8.55982909349324e-06,
      "loss": 0.0033,
      "step": 100730
    },
    {
      "epoch": 0.6461150680011416,
      "grad_norm": 0.029844725504517555,
      "learning_rate": 8.559436039576723e-06,
      "loss": 0.0035,
      "step": 100740
    },
    {
      "epoch": 0.6461792048949278,
      "grad_norm": 0.1949865221977234,
      "learning_rate": 8.559042941057931e-06,
      "loss": 0.003,
      "step": 100750
    },
    {
      "epoch": 0.6462433417887138,
      "grad_norm": 0.0806402787566185,
      "learning_rate": 8.558649797941788e-06,
      "loss": 0.0032,
      "step": 100760
    },
    {
      "epoch": 0.6463074786824999,
      "grad_norm": 0.13549336791038513,
      "learning_rate": 8.558256610233218e-06,
      "loss": 0.0019,
      "step": 100770
    },
    {
      "epoch": 0.646371615576286,
      "grad_norm": 0.287882924079895,
      "learning_rate": 8.55786337793715e-06,
      "loss": 0.003,
      "step": 100780
    },
    {
      "epoch": 0.6464357524700721,
      "grad_norm": 0.08676916360855103,
      "learning_rate": 8.55747010105851e-06,
      "loss": 0.0025,
      "step": 100790
    },
    {
      "epoch": 0.6464998893638583,
      "grad_norm": 0.06810786575078964,
      "learning_rate": 8.557076779602229e-06,
      "loss": 0.0044,
      "step": 100800
    },
    {
      "epoch": 0.6465640262576443,
      "grad_norm": 0.14305730164051056,
      "learning_rate": 8.556683413573233e-06,
      "loss": 0.0044,
      "step": 100810
    },
    {
      "epoch": 0.6466281631514305,
      "grad_norm": 0.1948164701461792,
      "learning_rate": 8.556290002976452e-06,
      "loss": 0.0023,
      "step": 100820
    },
    {
      "epoch": 0.6466923000452165,
      "grad_norm": 0.16761861741542816,
      "learning_rate": 8.555896547816815e-06,
      "loss": 0.0023,
      "step": 100830
    },
    {
      "epoch": 0.6467564369390026,
      "grad_norm": 0.10438597947359085,
      "learning_rate": 8.555503048099253e-06,
      "loss": 0.0038,
      "step": 100840
    },
    {
      "epoch": 0.6468205738327887,
      "grad_norm": 0.07459600269794464,
      "learning_rate": 8.555109503828699e-06,
      "loss": 0.004,
      "step": 100850
    },
    {
      "epoch": 0.6468847107265748,
      "grad_norm": 0.05684248358011246,
      "learning_rate": 8.55471591501008e-06,
      "loss": 0.0015,
      "step": 100860
    },
    {
      "epoch": 0.6469488476203609,
      "grad_norm": 0.14277034997940063,
      "learning_rate": 8.55432228164833e-06,
      "loss": 0.0036,
      "step": 100870
    },
    {
      "epoch": 0.647012984514147,
      "grad_norm": 0.28751543164253235,
      "learning_rate": 8.55392860374838e-06,
      "loss": 0.0035,
      "step": 100880
    },
    {
      "epoch": 0.647077121407933,
      "grad_norm": 0.1565263569355011,
      "learning_rate": 8.55353488131517e-06,
      "loss": 0.0033,
      "step": 100890
    },
    {
      "epoch": 0.6471412583017192,
      "grad_norm": 0.2728831171989441,
      "learning_rate": 8.553141114353622e-06,
      "loss": 0.0038,
      "step": 100900
    },
    {
      "epoch": 0.6472053951955052,
      "grad_norm": 0.11188970506191254,
      "learning_rate": 8.55274730286868e-06,
      "loss": 0.0028,
      "step": 100910
    },
    {
      "epoch": 0.6472695320892914,
      "grad_norm": 0.13675358891487122,
      "learning_rate": 8.552353446865275e-06,
      "loss": 0.0044,
      "step": 100920
    },
    {
      "epoch": 0.6473336689830774,
      "grad_norm": 0.16897651553153992,
      "learning_rate": 8.551959546348341e-06,
      "loss": 0.0019,
      "step": 100930
    },
    {
      "epoch": 0.6473978058768636,
      "grad_norm": 0.04027769714593887,
      "learning_rate": 8.551565601322818e-06,
      "loss": 0.0023,
      "step": 100940
    },
    {
      "epoch": 0.6474619427706497,
      "grad_norm": 0.1581668108701706,
      "learning_rate": 8.551171611793637e-06,
      "loss": 0.0017,
      "step": 100950
    },
    {
      "epoch": 0.6475260796644358,
      "grad_norm": 0.12060859799385071,
      "learning_rate": 8.550777577765739e-06,
      "loss": 0.0024,
      "step": 100960
    },
    {
      "epoch": 0.6475902165582219,
      "grad_norm": 0.06283990293741226,
      "learning_rate": 8.550383499244059e-06,
      "loss": 0.003,
      "step": 100970
    },
    {
      "epoch": 0.647654353452008,
      "grad_norm": 0.14902283251285553,
      "learning_rate": 8.549989376233539e-06,
      "loss": 0.0027,
      "step": 100980
    },
    {
      "epoch": 0.6477184903457941,
      "grad_norm": 0.07595626264810562,
      "learning_rate": 8.549595208739111e-06,
      "loss": 0.0025,
      "step": 100990
    },
    {
      "epoch": 0.6477826272395801,
      "grad_norm": 0.11632867902517319,
      "learning_rate": 8.549200996765722e-06,
      "loss": 0.0026,
      "step": 101000
    },
    {
      "epoch": 0.6478467641333663,
      "grad_norm": 0.03205646947026253,
      "learning_rate": 8.548806740318305e-06,
      "loss": 0.0019,
      "step": 101010
    },
    {
      "epoch": 0.6479109010271523,
      "grad_norm": 0.11372930556535721,
      "learning_rate": 8.548412439401805e-06,
      "loss": 0.0013,
      "step": 101020
    },
    {
      "epoch": 0.6479750379209385,
      "grad_norm": 0.2693907916545868,
      "learning_rate": 8.548018094021161e-06,
      "loss": 0.0031,
      "step": 101030
    },
    {
      "epoch": 0.6480391748147245,
      "grad_norm": 0.021016070619225502,
      "learning_rate": 8.547623704181316e-06,
      "loss": 0.0027,
      "step": 101040
    },
    {
      "epoch": 0.6481033117085107,
      "grad_norm": 0.04376423358917236,
      "learning_rate": 8.547229269887207e-06,
      "loss": 0.002,
      "step": 101050
    },
    {
      "epoch": 0.6481674486022967,
      "grad_norm": 0.06742499768733978,
      "learning_rate": 8.546834791143783e-06,
      "loss": 0.0023,
      "step": 101060
    },
    {
      "epoch": 0.6482315854960828,
      "grad_norm": 0.19039209187030792,
      "learning_rate": 8.546440267955982e-06,
      "loss": 0.0034,
      "step": 101070
    },
    {
      "epoch": 0.648295722389869,
      "grad_norm": 0.17136657238006592,
      "learning_rate": 8.54604570032875e-06,
      "loss": 0.0028,
      "step": 101080
    },
    {
      "epoch": 0.648359859283655,
      "grad_norm": 0.10204244405031204,
      "learning_rate": 8.545651088267035e-06,
      "loss": 0.004,
      "step": 101090
    },
    {
      "epoch": 0.6484239961774412,
      "grad_norm": 0.04198237136006355,
      "learning_rate": 8.545256431775774e-06,
      "loss": 0.0035,
      "step": 101100
    },
    {
      "epoch": 0.6484881330712272,
      "grad_norm": 0.15968136489391327,
      "learning_rate": 8.544861730859917e-06,
      "loss": 0.0038,
      "step": 101110
    },
    {
      "epoch": 0.6485522699650134,
      "grad_norm": 0.16758500039577484,
      "learning_rate": 8.54446698552441e-06,
      "loss": 0.0052,
      "step": 101120
    },
    {
      "epoch": 0.6486164068587994,
      "grad_norm": 0.0753367617726326,
      "learning_rate": 8.544072195774195e-06,
      "loss": 0.0038,
      "step": 101130
    },
    {
      "epoch": 0.6486805437525855,
      "grad_norm": 0.06411454826593399,
      "learning_rate": 8.543677361614226e-06,
      "loss": 0.002,
      "step": 101140
    },
    {
      "epoch": 0.6487446806463716,
      "grad_norm": 0.5838292241096497,
      "learning_rate": 8.543282483049446e-06,
      "loss": 0.0029,
      "step": 101150
    },
    {
      "epoch": 0.6488088175401577,
      "grad_norm": 0.039258599281311035,
      "learning_rate": 8.542887560084806e-06,
      "loss": 0.0036,
      "step": 101160
    },
    {
      "epoch": 0.6488729544339438,
      "grad_norm": 0.11288158595561981,
      "learning_rate": 8.542492592725248e-06,
      "loss": 0.0046,
      "step": 101170
    },
    {
      "epoch": 0.6489370913277299,
      "grad_norm": 0.14804130792617798,
      "learning_rate": 8.54209758097573e-06,
      "loss": 0.0049,
      "step": 101180
    },
    {
      "epoch": 0.649001228221516,
      "grad_norm": 0.16995002329349518,
      "learning_rate": 8.541702524841198e-06,
      "loss": 0.0029,
      "step": 101190
    },
    {
      "epoch": 0.6490653651153021,
      "grad_norm": 0.1034122109413147,
      "learning_rate": 8.5413074243266e-06,
      "loss": 0.0042,
      "step": 101200
    },
    {
      "epoch": 0.6491295020090881,
      "grad_norm": 0.25434547662734985,
      "learning_rate": 8.54091227943689e-06,
      "loss": 0.0019,
      "step": 101210
    },
    {
      "epoch": 0.6491936389028743,
      "grad_norm": 0.13532720506191254,
      "learning_rate": 8.540517090177019e-06,
      "loss": 0.003,
      "step": 101220
    },
    {
      "epoch": 0.6492577757966604,
      "grad_norm": 0.03820735588669777,
      "learning_rate": 8.540121856551938e-06,
      "loss": 0.0036,
      "step": 101230
    },
    {
      "epoch": 0.6493219126904465,
      "grad_norm": 0.09814277291297913,
      "learning_rate": 8.5397265785666e-06,
      "loss": 0.0033,
      "step": 101240
    },
    {
      "epoch": 0.6493860495842326,
      "grad_norm": 0.32696372270584106,
      "learning_rate": 8.539331256225958e-06,
      "loss": 0.0046,
      "step": 101250
    },
    {
      "epoch": 0.6494501864780187,
      "grad_norm": 0.23048560321331024,
      "learning_rate": 8.538935889534966e-06,
      "loss": 0.0056,
      "step": 101260
    },
    {
      "epoch": 0.6495143233718048,
      "grad_norm": 0.17196024954319,
      "learning_rate": 8.538540478498577e-06,
      "loss": 0.0055,
      "step": 101270
    },
    {
      "epoch": 0.6495784602655909,
      "grad_norm": 0.09881290793418884,
      "learning_rate": 8.538145023121748e-06,
      "loss": 0.0032,
      "step": 101280
    },
    {
      "epoch": 0.649642597159377,
      "grad_norm": 0.10250058025121689,
      "learning_rate": 8.537749523409434e-06,
      "loss": 0.0022,
      "step": 101290
    },
    {
      "epoch": 0.649706734053163,
      "grad_norm": 0.07143467664718628,
      "learning_rate": 8.537353979366589e-06,
      "loss": 0.0044,
      "step": 101300
    },
    {
      "epoch": 0.6497708709469492,
      "grad_norm": 0.19280396401882172,
      "learning_rate": 8.53695839099817e-06,
      "loss": 0.0031,
      "step": 101310
    },
    {
      "epoch": 0.6498350078407352,
      "grad_norm": 0.14994238317012787,
      "learning_rate": 8.536562758309138e-06,
      "loss": 0.003,
      "step": 101320
    },
    {
      "epoch": 0.6498991447345214,
      "grad_norm": 0.16695557534694672,
      "learning_rate": 8.536167081304442e-06,
      "loss": 0.0038,
      "step": 101330
    },
    {
      "epoch": 0.6499632816283074,
      "grad_norm": 0.08642906695604324,
      "learning_rate": 8.53577135998905e-06,
      "loss": 0.0031,
      "step": 101340
    },
    {
      "epoch": 0.6500274185220936,
      "grad_norm": 0.21799863874912262,
      "learning_rate": 8.535375594367911e-06,
      "loss": 0.0037,
      "step": 101350
    },
    {
      "epoch": 0.6500915554158797,
      "grad_norm": 0.09290055185556412,
      "learning_rate": 8.534979784445992e-06,
      "loss": 0.0039,
      "step": 101360
    },
    {
      "epoch": 0.6501556923096657,
      "grad_norm": 0.11244393140077591,
      "learning_rate": 8.53458393022825e-06,
      "loss": 0.0027,
      "step": 101370
    },
    {
      "epoch": 0.6502198292034519,
      "grad_norm": 0.4753751754760742,
      "learning_rate": 8.534188031719645e-06,
      "loss": 0.0031,
      "step": 101380
    },
    {
      "epoch": 0.6502839660972379,
      "grad_norm": 0.04352414980530739,
      "learning_rate": 8.533792088925137e-06,
      "loss": 0.0021,
      "step": 101390
    },
    {
      "epoch": 0.6503481029910241,
      "grad_norm": 0.2539691925048828,
      "learning_rate": 8.533396101849689e-06,
      "loss": 0.0021,
      "step": 101400
    },
    {
      "epoch": 0.6504122398848101,
      "grad_norm": 0.08388808369636536,
      "learning_rate": 8.533000070498264e-06,
      "loss": 0.0022,
      "step": 101410
    },
    {
      "epoch": 0.6504763767785963,
      "grad_norm": 0.037809524685144424,
      "learning_rate": 8.532603994875823e-06,
      "loss": 0.0011,
      "step": 101420
    },
    {
      "epoch": 0.6505405136723823,
      "grad_norm": 0.31021374464035034,
      "learning_rate": 8.532207874987327e-06,
      "loss": 0.0052,
      "step": 101430
    },
    {
      "epoch": 0.6506046505661685,
      "grad_norm": 0.10550630837678909,
      "learning_rate": 8.531811710837742e-06,
      "loss": 0.0037,
      "step": 101440
    },
    {
      "epoch": 0.6506687874599545,
      "grad_norm": 0.1269540637731552,
      "learning_rate": 8.531415502432035e-06,
      "loss": 0.0033,
      "step": 101450
    },
    {
      "epoch": 0.6507329243537406,
      "grad_norm": 0.19207847118377686,
      "learning_rate": 8.531019249775166e-06,
      "loss": 0.003,
      "step": 101460
    },
    {
      "epoch": 0.6507970612475267,
      "grad_norm": 0.017273563891649246,
      "learning_rate": 8.530622952872104e-06,
      "loss": 0.0017,
      "step": 101470
    },
    {
      "epoch": 0.6508611981413128,
      "grad_norm": 0.14974822103977203,
      "learning_rate": 8.53022661172781e-06,
      "loss": 0.0025,
      "step": 101480
    },
    {
      "epoch": 0.6509253350350989,
      "grad_norm": 0.06803373247385025,
      "learning_rate": 8.529830226347256e-06,
      "loss": 0.0023,
      "step": 101490
    },
    {
      "epoch": 0.650989471928885,
      "grad_norm": 0.11219818890094757,
      "learning_rate": 8.529433796735404e-06,
      "loss": 0.0025,
      "step": 101500
    },
    {
      "epoch": 0.6510536088226712,
      "grad_norm": 0.24247153103351593,
      "learning_rate": 8.529037322897227e-06,
      "loss": 0.0048,
      "step": 101510
    },
    {
      "epoch": 0.6511177457164572,
      "grad_norm": 0.14471827447414398,
      "learning_rate": 8.528640804837689e-06,
      "loss": 0.004,
      "step": 101520
    },
    {
      "epoch": 0.6511818826102433,
      "grad_norm": 0.17130020260810852,
      "learning_rate": 8.528244242561758e-06,
      "loss": 0.004,
      "step": 101530
    },
    {
      "epoch": 0.6512460195040294,
      "grad_norm": 0.1778392642736435,
      "learning_rate": 8.527847636074408e-06,
      "loss": 0.0022,
      "step": 101540
    },
    {
      "epoch": 0.6513101563978155,
      "grad_norm": 0.2364269196987152,
      "learning_rate": 8.527450985380605e-06,
      "loss": 0.0024,
      "step": 101550
    },
    {
      "epoch": 0.6513742932916016,
      "grad_norm": 0.16268400847911835,
      "learning_rate": 8.527054290485319e-06,
      "loss": 0.0023,
      "step": 101560
    },
    {
      "epoch": 0.6514384301853877,
      "grad_norm": 0.35830166935920715,
      "learning_rate": 8.526657551393522e-06,
      "loss": 0.0042,
      "step": 101570
    },
    {
      "epoch": 0.6515025670791738,
      "grad_norm": 0.12425675988197327,
      "learning_rate": 8.526260768110185e-06,
      "loss": 0.0042,
      "step": 101580
    },
    {
      "epoch": 0.6515667039729599,
      "grad_norm": 0.16214703023433685,
      "learning_rate": 8.525863940640282e-06,
      "loss": 0.0055,
      "step": 101590
    },
    {
      "epoch": 0.6516308408667459,
      "grad_norm": 0.41682323813438416,
      "learning_rate": 8.525467068988783e-06,
      "loss": 0.0024,
      "step": 101600
    },
    {
      "epoch": 0.6516949777605321,
      "grad_norm": 0.37435153126716614,
      "learning_rate": 8.525070153160664e-06,
      "loss": 0.0032,
      "step": 101610
    },
    {
      "epoch": 0.6517591146543181,
      "grad_norm": 0.40198203921318054,
      "learning_rate": 8.524673193160894e-06,
      "loss": 0.0033,
      "step": 101620
    },
    {
      "epoch": 0.6518232515481043,
      "grad_norm": 0.007691803388297558,
      "learning_rate": 8.524276188994452e-06,
      "loss": 0.0039,
      "step": 101630
    },
    {
      "epoch": 0.6518873884418904,
      "grad_norm": 0.19811321794986725,
      "learning_rate": 8.523879140666308e-06,
      "loss": 0.0028,
      "step": 101640
    },
    {
      "epoch": 0.6519515253356765,
      "grad_norm": 0.4874114990234375,
      "learning_rate": 8.523482048181443e-06,
      "loss": 0.0042,
      "step": 101650
    },
    {
      "epoch": 0.6520156622294626,
      "grad_norm": 0.27024710178375244,
      "learning_rate": 8.523084911544827e-06,
      "loss": 0.0043,
      "step": 101660
    },
    {
      "epoch": 0.6520797991232486,
      "grad_norm": 0.12862282991409302,
      "learning_rate": 8.522687730761441e-06,
      "loss": 0.0025,
      "step": 101670
    },
    {
      "epoch": 0.6521439360170348,
      "grad_norm": 0.04814813286066055,
      "learning_rate": 8.522290505836259e-06,
      "loss": 0.0029,
      "step": 101680
    },
    {
      "epoch": 0.6522080729108208,
      "grad_norm": 0.1256941705942154,
      "learning_rate": 8.521893236774262e-06,
      "loss": 0.0029,
      "step": 101690
    },
    {
      "epoch": 0.652272209804607,
      "grad_norm": 0.021735558286309242,
      "learning_rate": 8.52149592358042e-06,
      "loss": 0.0037,
      "step": 101700
    },
    {
      "epoch": 0.652336346698393,
      "grad_norm": 0.21016880869865417,
      "learning_rate": 8.521098566259723e-06,
      "loss": 0.0035,
      "step": 101710
    },
    {
      "epoch": 0.6524004835921792,
      "grad_norm": 0.11760730296373367,
      "learning_rate": 8.520701164817142e-06,
      "loss": 0.0026,
      "step": 101720
    },
    {
      "epoch": 0.6524646204859652,
      "grad_norm": 0.2804056406021118,
      "learning_rate": 8.52030371925766e-06,
      "loss": 0.004,
      "step": 101730
    },
    {
      "epoch": 0.6525287573797514,
      "grad_norm": 0.16390132904052734,
      "learning_rate": 8.519906229586255e-06,
      "loss": 0.0017,
      "step": 101740
    },
    {
      "epoch": 0.6525928942735374,
      "grad_norm": 0.10632698237895966,
      "learning_rate": 8.519508695807912e-06,
      "loss": 0.0024,
      "step": 101750
    },
    {
      "epoch": 0.6526570311673235,
      "grad_norm": 0.05945868045091629,
      "learning_rate": 8.519111117927608e-06,
      "loss": 0.0018,
      "step": 101760
    },
    {
      "epoch": 0.6527211680611096,
      "grad_norm": 0.14257098734378815,
      "learning_rate": 8.518713495950326e-06,
      "loss": 0.0024,
      "step": 101770
    },
    {
      "epoch": 0.6527853049548957,
      "grad_norm": 0.1678561270236969,
      "learning_rate": 8.51831582988105e-06,
      "loss": 0.0033,
      "step": 101780
    },
    {
      "epoch": 0.6528494418486819,
      "grad_norm": 0.14132031798362732,
      "learning_rate": 8.517918119724762e-06,
      "loss": 0.0059,
      "step": 101790
    },
    {
      "epoch": 0.6529135787424679,
      "grad_norm": 0.16761355102062225,
      "learning_rate": 8.517520365486445e-06,
      "loss": 0.0027,
      "step": 101800
    },
    {
      "epoch": 0.6529777156362541,
      "grad_norm": 0.06852803379297256,
      "learning_rate": 8.517122567171085e-06,
      "loss": 0.0021,
      "step": 101810
    },
    {
      "epoch": 0.6530418525300401,
      "grad_norm": 0.1299738734960556,
      "learning_rate": 8.516724724783665e-06,
      "loss": 0.0035,
      "step": 101820
    },
    {
      "epoch": 0.6531059894238262,
      "grad_norm": 0.0322495736181736,
      "learning_rate": 8.516326838329172e-06,
      "loss": 0.0033,
      "step": 101830
    },
    {
      "epoch": 0.6531701263176123,
      "grad_norm": 0.11352141946554184,
      "learning_rate": 8.51592890781259e-06,
      "loss": 0.0021,
      "step": 101840
    },
    {
      "epoch": 0.6532342632113984,
      "grad_norm": 0.14519372582435608,
      "learning_rate": 8.515530933238904e-06,
      "loss": 0.0023,
      "step": 101850
    },
    {
      "epoch": 0.6532984001051845,
      "grad_norm": 0.13392499089241028,
      "learning_rate": 8.515132914613105e-06,
      "loss": 0.0021,
      "step": 101860
    },
    {
      "epoch": 0.6533625369989706,
      "grad_norm": 0.039920829236507416,
      "learning_rate": 8.514734851940178e-06,
      "loss": 0.0024,
      "step": 101870
    },
    {
      "epoch": 0.6534266738927567,
      "grad_norm": 0.07632692903280258,
      "learning_rate": 8.51433674522511e-06,
      "loss": 0.0143,
      "step": 101880
    },
    {
      "epoch": 0.6534908107865428,
      "grad_norm": 0.10139046609401703,
      "learning_rate": 8.513938594472895e-06,
      "loss": 0.0027,
      "step": 101890
    },
    {
      "epoch": 0.6535549476803288,
      "grad_norm": 0.06001872569322586,
      "learning_rate": 8.513540399688514e-06,
      "loss": 0.0023,
      "step": 101900
    },
    {
      "epoch": 0.653619084574115,
      "grad_norm": 0.15371845662593842,
      "learning_rate": 8.513142160876962e-06,
      "loss": 0.0018,
      "step": 101910
    },
    {
      "epoch": 0.6536832214679011,
      "grad_norm": 0.08152685314416885,
      "learning_rate": 8.512743878043228e-06,
      "loss": 0.0043,
      "step": 101920
    },
    {
      "epoch": 0.6537473583616872,
      "grad_norm": 0.17926916480064392,
      "learning_rate": 8.512345551192302e-06,
      "loss": 0.0035,
      "step": 101930
    },
    {
      "epoch": 0.6538114952554733,
      "grad_norm": 0.29780837893486023,
      "learning_rate": 8.511947180329177e-06,
      "loss": 0.0029,
      "step": 101940
    },
    {
      "epoch": 0.6538756321492594,
      "grad_norm": 0.07611296325922012,
      "learning_rate": 8.511548765458844e-06,
      "loss": 0.0052,
      "step": 101950
    },
    {
      "epoch": 0.6539397690430455,
      "grad_norm": 0.15462587773799896,
      "learning_rate": 8.511150306586295e-06,
      "loss": 0.0041,
      "step": 101960
    },
    {
      "epoch": 0.6540039059368316,
      "grad_norm": 0.17087388038635254,
      "learning_rate": 8.510751803716523e-06,
      "loss": 0.0026,
      "step": 101970
    },
    {
      "epoch": 0.6540680428306177,
      "grad_norm": 0.011938631534576416,
      "learning_rate": 8.51035325685452e-06,
      "loss": 0.0014,
      "step": 101980
    },
    {
      "epoch": 0.6541321797244037,
      "grad_norm": 0.03938727453351021,
      "learning_rate": 8.509954666005285e-06,
      "loss": 0.0027,
      "step": 101990
    },
    {
      "epoch": 0.6541963166181899,
      "grad_norm": 0.1165597215294838,
      "learning_rate": 8.509556031173808e-06,
      "loss": 0.0023,
      "step": 102000
    },
    {
      "epoch": 0.6542604535119759,
      "grad_norm": 0.41937723755836487,
      "learning_rate": 8.509157352365087e-06,
      "loss": 0.0075,
      "step": 102010
    },
    {
      "epoch": 0.6543245904057621,
      "grad_norm": 0.07766172289848328,
      "learning_rate": 8.508758629584113e-06,
      "loss": 0.0024,
      "step": 102020
    },
    {
      "epoch": 0.6543887272995481,
      "grad_norm": 0.1976241171360016,
      "learning_rate": 8.50835986283589e-06,
      "loss": 0.0035,
      "step": 102030
    },
    {
      "epoch": 0.6544528641933343,
      "grad_norm": 0.05941390618681908,
      "learning_rate": 8.507961052125409e-06,
      "loss": 0.0027,
      "step": 102040
    },
    {
      "epoch": 0.6545170010871203,
      "grad_norm": 0.00836424995213747,
      "learning_rate": 8.507562197457667e-06,
      "loss": 0.003,
      "step": 102050
    },
    {
      "epoch": 0.6545811379809064,
      "grad_norm": 0.0073838010430336,
      "learning_rate": 8.507163298837667e-06,
      "loss": 0.0045,
      "step": 102060
    },
    {
      "epoch": 0.6546452748746926,
      "grad_norm": 0.08924354612827301,
      "learning_rate": 8.5067643562704e-06,
      "loss": 0.0048,
      "step": 102070
    },
    {
      "epoch": 0.6547094117684786,
      "grad_norm": 0.16871915757656097,
      "learning_rate": 8.506365369760872e-06,
      "loss": 0.0038,
      "step": 102080
    },
    {
      "epoch": 0.6547735486622648,
      "grad_norm": 0.7401137351989746,
      "learning_rate": 8.50596633931408e-06,
      "loss": 0.0029,
      "step": 102090
    },
    {
      "epoch": 0.6548376855560508,
      "grad_norm": 0.11338161677122116,
      "learning_rate": 8.505567264935025e-06,
      "loss": 0.0021,
      "step": 102100
    },
    {
      "epoch": 0.654901822449837,
      "grad_norm": 0.3115667700767517,
      "learning_rate": 8.505168146628707e-06,
      "loss": 0.0035,
      "step": 102110
    },
    {
      "epoch": 0.654965959343623,
      "grad_norm": 0.14822430908679962,
      "learning_rate": 8.504768984400125e-06,
      "loss": 0.0029,
      "step": 102120
    },
    {
      "epoch": 0.6550300962374092,
      "grad_norm": 0.25849664211273193,
      "learning_rate": 8.504369778254283e-06,
      "loss": 0.0069,
      "step": 102130
    },
    {
      "epoch": 0.6550942331311952,
      "grad_norm": 0.10550545156002045,
      "learning_rate": 8.503970528196183e-06,
      "loss": 0.0033,
      "step": 102140
    },
    {
      "epoch": 0.6551583700249813,
      "grad_norm": 0.22024445235729218,
      "learning_rate": 8.503571234230829e-06,
      "loss": 0.0035,
      "step": 102150
    },
    {
      "epoch": 0.6552225069187674,
      "grad_norm": 0.11373266577720642,
      "learning_rate": 8.503171896363225e-06,
      "loss": 0.0027,
      "step": 102160
    },
    {
      "epoch": 0.6552866438125535,
      "grad_norm": 0.08975688368082047,
      "learning_rate": 8.502772514598371e-06,
      "loss": 0.0037,
      "step": 102170
    },
    {
      "epoch": 0.6553507807063396,
      "grad_norm": 0.10338490456342697,
      "learning_rate": 8.502373088941273e-06,
      "loss": 0.0028,
      "step": 102180
    },
    {
      "epoch": 0.6554149176001257,
      "grad_norm": 0.0868222787976265,
      "learning_rate": 8.501973619396941e-06,
      "loss": 0.0056,
      "step": 102190
    },
    {
      "epoch": 0.6554790544939118,
      "grad_norm": 0.0775335505604744,
      "learning_rate": 8.501574105970373e-06,
      "loss": 0.0028,
      "step": 102200
    },
    {
      "epoch": 0.6555431913876979,
      "grad_norm": 0.2917906641960144,
      "learning_rate": 8.501174548666582e-06,
      "loss": 0.0041,
      "step": 102210
    },
    {
      "epoch": 0.655607328281484,
      "grad_norm": 0.10409858822822571,
      "learning_rate": 8.500774947490569e-06,
      "loss": 0.0024,
      "step": 102220
    },
    {
      "epoch": 0.6556714651752701,
      "grad_norm": 0.08417538553476334,
      "learning_rate": 8.500375302447346e-06,
      "loss": 0.0026,
      "step": 102230
    },
    {
      "epoch": 0.6557356020690562,
      "grad_norm": 0.03060046024620533,
      "learning_rate": 8.499975613541917e-06,
      "loss": 0.0064,
      "step": 102240
    },
    {
      "epoch": 0.6557997389628423,
      "grad_norm": 0.21709373593330383,
      "learning_rate": 8.499575880779295e-06,
      "loss": 0.0047,
      "step": 102250
    },
    {
      "epoch": 0.6558638758566284,
      "grad_norm": 0.13628128170967102,
      "learning_rate": 8.499176104164482e-06,
      "loss": 0.0029,
      "step": 102260
    },
    {
      "epoch": 0.6559280127504145,
      "grad_norm": 0.09415052086114883,
      "learning_rate": 8.498776283702495e-06,
      "loss": 0.0034,
      "step": 102270
    },
    {
      "epoch": 0.6559921496442006,
      "grad_norm": 0.14981704950332642,
      "learning_rate": 8.49837641939834e-06,
      "loss": 0.0041,
      "step": 102280
    },
    {
      "epoch": 0.6560562865379866,
      "grad_norm": 0.09459282457828522,
      "learning_rate": 8.497976511257027e-06,
      "loss": 0.0026,
      "step": 102290
    },
    {
      "epoch": 0.6561204234317728,
      "grad_norm": 0.23267565667629242,
      "learning_rate": 8.497576559283569e-06,
      "loss": 0.006,
      "step": 102300
    },
    {
      "epoch": 0.6561845603255588,
      "grad_norm": 0.2411142885684967,
      "learning_rate": 8.497176563482976e-06,
      "loss": 0.004,
      "step": 102310
    },
    {
      "epoch": 0.656248697219345,
      "grad_norm": 0.05634717270731926,
      "learning_rate": 8.496776523860263e-06,
      "loss": 0.0026,
      "step": 102320
    },
    {
      "epoch": 0.656312834113131,
      "grad_norm": 0.08043432235717773,
      "learning_rate": 8.496376440420439e-06,
      "loss": 0.0026,
      "step": 102330
    },
    {
      "epoch": 0.6563769710069172,
      "grad_norm": 0.38449281454086304,
      "learning_rate": 8.495976313168522e-06,
      "loss": 0.0029,
      "step": 102340
    },
    {
      "epoch": 0.6564411079007033,
      "grad_norm": 0.06370898336172104,
      "learning_rate": 8.495576142109521e-06,
      "loss": 0.003,
      "step": 102350
    },
    {
      "epoch": 0.6565052447944894,
      "grad_norm": 0.039983682334423065,
      "learning_rate": 8.495175927248454e-06,
      "loss": 0.0033,
      "step": 102360
    },
    {
      "epoch": 0.6565693816882755,
      "grad_norm": 0.16125039756298065,
      "learning_rate": 8.494775668590334e-06,
      "loss": 0.0036,
      "step": 102370
    },
    {
      "epoch": 0.6566335185820615,
      "grad_norm": 0.11923573166131973,
      "learning_rate": 8.494375366140177e-06,
      "loss": 0.0046,
      "step": 102380
    },
    {
      "epoch": 0.6566976554758477,
      "grad_norm": 0.037645936012268066,
      "learning_rate": 8.493975019903e-06,
      "loss": 0.0028,
      "step": 102390
    },
    {
      "epoch": 0.6567617923696337,
      "grad_norm": 0.16430379450321198,
      "learning_rate": 8.493574629883818e-06,
      "loss": 0.0035,
      "step": 102400
    },
    {
      "epoch": 0.6568259292634199,
      "grad_norm": 0.056253910064697266,
      "learning_rate": 8.49317419608765e-06,
      "loss": 0.0027,
      "step": 102410
    },
    {
      "epoch": 0.6568900661572059,
      "grad_norm": 0.1980583667755127,
      "learning_rate": 8.492773718519513e-06,
      "loss": 0.0031,
      "step": 102420
    },
    {
      "epoch": 0.6569542030509921,
      "grad_norm": 0.15534088015556335,
      "learning_rate": 8.492373197184425e-06,
      "loss": 0.002,
      "step": 102430
    },
    {
      "epoch": 0.6570183399447781,
      "grad_norm": 0.056002579629421234,
      "learning_rate": 8.491972632087404e-06,
      "loss": 0.002,
      "step": 102440
    },
    {
      "epoch": 0.6570824768385642,
      "grad_norm": 0.18479229509830475,
      "learning_rate": 8.49157202323347e-06,
      "loss": 0.0039,
      "step": 102450
    },
    {
      "epoch": 0.6571466137323503,
      "grad_norm": 0.15330883860588074,
      "learning_rate": 8.491171370627645e-06,
      "loss": 0.0021,
      "step": 102460
    },
    {
      "epoch": 0.6572107506261364,
      "grad_norm": 0.036105621606111526,
      "learning_rate": 8.490770674274945e-06,
      "loss": 0.0036,
      "step": 102470
    },
    {
      "epoch": 0.6572748875199225,
      "grad_norm": 0.33400094509124756,
      "learning_rate": 8.490369934180396e-06,
      "loss": 0.0031,
      "step": 102480
    },
    {
      "epoch": 0.6573390244137086,
      "grad_norm": 0.04812569171190262,
      "learning_rate": 8.489969150349016e-06,
      "loss": 0.003,
      "step": 102490
    },
    {
      "epoch": 0.6574031613074948,
      "grad_norm": 0.1350105255842209,
      "learning_rate": 8.489568322785828e-06,
      "loss": 0.0036,
      "step": 102500
    },
    {
      "epoch": 0.6574672982012808,
      "grad_norm": 0.11404775083065033,
      "learning_rate": 8.489167451495857e-06,
      "loss": 0.0024,
      "step": 102510
    },
    {
      "epoch": 0.657531435095067,
      "grad_norm": 0.0941624641418457,
      "learning_rate": 8.48876653648412e-06,
      "loss": 0.0024,
      "step": 102520
    },
    {
      "epoch": 0.657595571988853,
      "grad_norm": 0.1571592390537262,
      "learning_rate": 8.48836557775565e-06,
      "loss": 0.002,
      "step": 102530
    },
    {
      "epoch": 0.6576597088826391,
      "grad_norm": 0.05018900707364082,
      "learning_rate": 8.487964575315463e-06,
      "loss": 0.0021,
      "step": 102540
    },
    {
      "epoch": 0.6577238457764252,
      "grad_norm": 0.2466675490140915,
      "learning_rate": 8.487563529168588e-06,
      "loss": 0.0036,
      "step": 102550
    },
    {
      "epoch": 0.6577879826702113,
      "grad_norm": 0.13947294652462006,
      "learning_rate": 8.487162439320048e-06,
      "loss": 0.0022,
      "step": 102560
    },
    {
      "epoch": 0.6578521195639974,
      "grad_norm": 0.15182632207870483,
      "learning_rate": 8.486761305774874e-06,
      "loss": 0.0025,
      "step": 102570
    },
    {
      "epoch": 0.6579162564577835,
      "grad_norm": 0.09285365790128708,
      "learning_rate": 8.486360128538084e-06,
      "loss": 0.0048,
      "step": 102580
    },
    {
      "epoch": 0.6579803933515695,
      "grad_norm": 0.06940281391143799,
      "learning_rate": 8.485958907614714e-06,
      "loss": 0.0022,
      "step": 102590
    },
    {
      "epoch": 0.6580445302453557,
      "grad_norm": 0.09990634024143219,
      "learning_rate": 8.485557643009786e-06,
      "loss": 0.0021,
      "step": 102600
    },
    {
      "epoch": 0.6581086671391417,
      "grad_norm": 0.15104538202285767,
      "learning_rate": 8.485156334728331e-06,
      "loss": 0.0085,
      "step": 102610
    },
    {
      "epoch": 0.6581728040329279,
      "grad_norm": 0.1043042317032814,
      "learning_rate": 8.484754982775374e-06,
      "loss": 0.0029,
      "step": 102620
    },
    {
      "epoch": 0.658236940926714,
      "grad_norm": 0.4126298129558563,
      "learning_rate": 8.484353587155947e-06,
      "loss": 0.0016,
      "step": 102630
    },
    {
      "epoch": 0.6583010778205001,
      "grad_norm": 0.06483572721481323,
      "learning_rate": 8.48395214787508e-06,
      "loss": 0.0023,
      "step": 102640
    },
    {
      "epoch": 0.6583652147142862,
      "grad_norm": 0.16036419570446014,
      "learning_rate": 8.483550664937804e-06,
      "loss": 0.0019,
      "step": 102650
    },
    {
      "epoch": 0.6584293516080723,
      "grad_norm": 0.10775353759527206,
      "learning_rate": 8.483149138349146e-06,
      "loss": 0.0023,
      "step": 102660
    },
    {
      "epoch": 0.6584934885018584,
      "grad_norm": 0.31090420484542847,
      "learning_rate": 8.48274756811414e-06,
      "loss": 0.0038,
      "step": 102670
    },
    {
      "epoch": 0.6585576253956444,
      "grad_norm": 0.17591483891010284,
      "learning_rate": 8.48234595423782e-06,
      "loss": 0.0014,
      "step": 102680
    },
    {
      "epoch": 0.6586217622894306,
      "grad_norm": 0.2681688666343689,
      "learning_rate": 8.481944296725216e-06,
      "loss": 0.0051,
      "step": 102690
    },
    {
      "epoch": 0.6586858991832166,
      "grad_norm": 0.3551349639892578,
      "learning_rate": 8.48154259558136e-06,
      "loss": 0.0053,
      "step": 102700
    },
    {
      "epoch": 0.6587500360770028,
      "grad_norm": 0.04370328038930893,
      "learning_rate": 8.481140850811287e-06,
      "loss": 0.0023,
      "step": 102710
    },
    {
      "epoch": 0.6588141729707888,
      "grad_norm": 0.06229441985487938,
      "learning_rate": 8.480739062420033e-06,
      "loss": 0.0035,
      "step": 102720
    },
    {
      "epoch": 0.658878309864575,
      "grad_norm": 0.018355660140514374,
      "learning_rate": 8.48033723041263e-06,
      "loss": 0.0023,
      "step": 102730
    },
    {
      "epoch": 0.658942446758361,
      "grad_norm": 0.15766875445842743,
      "learning_rate": 8.479935354794113e-06,
      "loss": 0.0031,
      "step": 102740
    },
    {
      "epoch": 0.6590065836521471,
      "grad_norm": 0.21535296738147736,
      "learning_rate": 8.47953343556952e-06,
      "loss": 0.0021,
      "step": 102750
    },
    {
      "epoch": 0.6590707205459332,
      "grad_norm": 0.058433424681425095,
      "learning_rate": 8.479131472743887e-06,
      "loss": 0.0022,
      "step": 102760
    },
    {
      "epoch": 0.6591348574397193,
      "grad_norm": 0.0867202877998352,
      "learning_rate": 8.47872946632225e-06,
      "loss": 0.0047,
      "step": 102770
    },
    {
      "epoch": 0.6591989943335055,
      "grad_norm": 0.15360093116760254,
      "learning_rate": 8.478327416309645e-06,
      "loss": 0.003,
      "step": 102780
    },
    {
      "epoch": 0.6592631312272915,
      "grad_norm": 0.1224280372262001,
      "learning_rate": 8.477925322711112e-06,
      "loss": 0.0036,
      "step": 102790
    },
    {
      "epoch": 0.6593272681210777,
      "grad_norm": 0.13371972739696503,
      "learning_rate": 8.47752318553169e-06,
      "loss": 0.0015,
      "step": 102800
    },
    {
      "epoch": 0.6593914050148637,
      "grad_norm": 0.0641275942325592,
      "learning_rate": 8.477121004776417e-06,
      "loss": 0.0018,
      "step": 102810
    },
    {
      "epoch": 0.6594555419086499,
      "grad_norm": 0.14218440651893616,
      "learning_rate": 8.476718780450332e-06,
      "loss": 0.0026,
      "step": 102820
    },
    {
      "epoch": 0.6595196788024359,
      "grad_norm": 0.17348818480968475,
      "learning_rate": 8.476316512558475e-06,
      "loss": 0.0018,
      "step": 102830
    },
    {
      "epoch": 0.659583815696222,
      "grad_norm": 0.10095454752445221,
      "learning_rate": 8.47591420110589e-06,
      "loss": 0.0016,
      "step": 102840
    },
    {
      "epoch": 0.6596479525900081,
      "grad_norm": 0.048792753368616104,
      "learning_rate": 8.475511846097615e-06,
      "loss": 0.0044,
      "step": 102850
    },
    {
      "epoch": 0.6597120894837942,
      "grad_norm": 0.17745624482631683,
      "learning_rate": 8.475109447538691e-06,
      "loss": 0.0031,
      "step": 102860
    },
    {
      "epoch": 0.6597762263775803,
      "grad_norm": 0.12286870181560516,
      "learning_rate": 8.474707005434165e-06,
      "loss": 0.0028,
      "step": 102870
    },
    {
      "epoch": 0.6598403632713664,
      "grad_norm": 0.22015081346035004,
      "learning_rate": 8.474304519789077e-06,
      "loss": 0.0028,
      "step": 102880
    },
    {
      "epoch": 0.6599045001651525,
      "grad_norm": 0.20826350152492523,
      "learning_rate": 8.473901990608467e-06,
      "loss": 0.0021,
      "step": 102890
    },
    {
      "epoch": 0.6599686370589386,
      "grad_norm": 0.19223351776599884,
      "learning_rate": 8.473499417897384e-06,
      "loss": 0.002,
      "step": 102900
    },
    {
      "epoch": 0.6600327739527247,
      "grad_norm": 0.07205517590045929,
      "learning_rate": 8.473096801660871e-06,
      "loss": 0.0029,
      "step": 102910
    },
    {
      "epoch": 0.6600969108465108,
      "grad_norm": 0.1505216509103775,
      "learning_rate": 8.472694141903971e-06,
      "loss": 0.0028,
      "step": 102920
    },
    {
      "epoch": 0.6601610477402969,
      "grad_norm": 0.1652790606021881,
      "learning_rate": 8.472291438631735e-06,
      "loss": 0.0036,
      "step": 102930
    },
    {
      "epoch": 0.660225184634083,
      "grad_norm": 0.10180681943893433,
      "learning_rate": 8.471888691849205e-06,
      "loss": 0.0018,
      "step": 102940
    },
    {
      "epoch": 0.6602893215278691,
      "grad_norm": 0.06222664564847946,
      "learning_rate": 8.471485901561426e-06,
      "loss": 0.0029,
      "step": 102950
    },
    {
      "epoch": 0.6603534584216552,
      "grad_norm": 0.1303701251745224,
      "learning_rate": 8.471083067773448e-06,
      "loss": 0.0029,
      "step": 102960
    },
    {
      "epoch": 0.6604175953154413,
      "grad_norm": 0.23145528137683868,
      "learning_rate": 8.470680190490319e-06,
      "loss": 0.0036,
      "step": 102970
    },
    {
      "epoch": 0.6604817322092273,
      "grad_norm": 0.13066375255584717,
      "learning_rate": 8.47027726971709e-06,
      "loss": 0.0024,
      "step": 102980
    },
    {
      "epoch": 0.6605458691030135,
      "grad_norm": 0.18564635515213013,
      "learning_rate": 8.469874305458802e-06,
      "loss": 0.0038,
      "step": 102990
    },
    {
      "epoch": 0.6606100059967995,
      "grad_norm": 0.1200670599937439,
      "learning_rate": 8.46947129772051e-06,
      "loss": 0.0023,
      "step": 103000
    },
    {
      "epoch": 0.6606741428905857,
      "grad_norm": 0.05317830666899681,
      "learning_rate": 8.469068246507264e-06,
      "loss": 0.002,
      "step": 103010
    },
    {
      "epoch": 0.6607382797843717,
      "grad_norm": 0.33798936009407043,
      "learning_rate": 8.468665151824113e-06,
      "loss": 0.0032,
      "step": 103020
    },
    {
      "epoch": 0.6608024166781579,
      "grad_norm": 0.17617559432983398,
      "learning_rate": 8.46826201367611e-06,
      "loss": 0.0023,
      "step": 103030
    },
    {
      "epoch": 0.6608665535719439,
      "grad_norm": 0.2390466332435608,
      "learning_rate": 8.467858832068303e-06,
      "loss": 0.0029,
      "step": 103040
    },
    {
      "epoch": 0.66093069046573,
      "grad_norm": 0.11099249869585037,
      "learning_rate": 8.467455607005748e-06,
      "loss": 0.0021,
      "step": 103050
    },
    {
      "epoch": 0.6609948273595162,
      "grad_norm": 0.06324342638254166,
      "learning_rate": 8.467052338493496e-06,
      "loss": 0.0019,
      "step": 103060
    },
    {
      "epoch": 0.6610589642533022,
      "grad_norm": 0.11985218524932861,
      "learning_rate": 8.466649026536599e-06,
      "loss": 0.0034,
      "step": 103070
    },
    {
      "epoch": 0.6611231011470884,
      "grad_norm": 0.22718679904937744,
      "learning_rate": 8.466245671140115e-06,
      "loss": 0.0025,
      "step": 103080
    },
    {
      "epoch": 0.6611872380408744,
      "grad_norm": 0.07870079576969147,
      "learning_rate": 8.465842272309093e-06,
      "loss": 0.0021,
      "step": 103090
    },
    {
      "epoch": 0.6612513749346606,
      "grad_norm": 0.3526822030544281,
      "learning_rate": 8.46543883004859e-06,
      "loss": 0.0031,
      "step": 103100
    },
    {
      "epoch": 0.6613155118284466,
      "grad_norm": 0.08089350908994675,
      "learning_rate": 8.465035344363664e-06,
      "loss": 0.0069,
      "step": 103110
    },
    {
      "epoch": 0.6613796487222328,
      "grad_norm": 0.676216185092926,
      "learning_rate": 8.464631815259368e-06,
      "loss": 0.0056,
      "step": 103120
    },
    {
      "epoch": 0.6614437856160188,
      "grad_norm": 0.07288841158151627,
      "learning_rate": 8.464228242740758e-06,
      "loss": 0.003,
      "step": 103130
    },
    {
      "epoch": 0.661507922509805,
      "grad_norm": 0.17153863608837128,
      "learning_rate": 8.463824626812893e-06,
      "loss": 0.0018,
      "step": 103140
    },
    {
      "epoch": 0.661572059403591,
      "grad_norm": 0.06785765290260315,
      "learning_rate": 8.463420967480828e-06,
      "loss": 0.0031,
      "step": 103150
    },
    {
      "epoch": 0.6616361962973771,
      "grad_norm": 0.26055482029914856,
      "learning_rate": 8.463017264749626e-06,
      "loss": 0.003,
      "step": 103160
    },
    {
      "epoch": 0.6617003331911632,
      "grad_norm": 0.12863045930862427,
      "learning_rate": 8.462613518624341e-06,
      "loss": 0.0033,
      "step": 103170
    },
    {
      "epoch": 0.6617644700849493,
      "grad_norm": 0.1738891750574112,
      "learning_rate": 8.462209729110035e-06,
      "loss": 0.0024,
      "step": 103180
    },
    {
      "epoch": 0.6618286069787355,
      "grad_norm": 0.06312116235494614,
      "learning_rate": 8.461805896211766e-06,
      "loss": 0.0025,
      "step": 103190
    },
    {
      "epoch": 0.6618927438725215,
      "grad_norm": 0.008281880989670753,
      "learning_rate": 8.461402019934594e-06,
      "loss": 0.0027,
      "step": 103200
    },
    {
      "epoch": 0.6619568807663077,
      "grad_norm": 0.1504332423210144,
      "learning_rate": 8.460998100283584e-06,
      "loss": 0.0025,
      "step": 103210
    },
    {
      "epoch": 0.6620210176600937,
      "grad_norm": 0.10833275318145752,
      "learning_rate": 8.460594137263792e-06,
      "loss": 0.0033,
      "step": 103220
    },
    {
      "epoch": 0.6620851545538798,
      "grad_norm": 0.15487238764762878,
      "learning_rate": 8.460190130880282e-06,
      "loss": 0.0046,
      "step": 103230
    },
    {
      "epoch": 0.6621492914476659,
      "grad_norm": 0.08042314648628235,
      "learning_rate": 8.459786081138116e-06,
      "loss": 0.0018,
      "step": 103240
    },
    {
      "epoch": 0.662213428341452,
      "grad_norm": 0.09010224789381027,
      "learning_rate": 8.45938198804236e-06,
      "loss": 0.0034,
      "step": 103250
    },
    {
      "epoch": 0.6622775652352381,
      "grad_norm": 0.16834987699985504,
      "learning_rate": 8.458977851598074e-06,
      "loss": 0.0031,
      "step": 103260
    },
    {
      "epoch": 0.6623417021290242,
      "grad_norm": 0.11550785601139069,
      "learning_rate": 8.458573671810323e-06,
      "loss": 0.0027,
      "step": 103270
    },
    {
      "epoch": 0.6624058390228102,
      "grad_norm": 0.14988340437412262,
      "learning_rate": 8.458169448684172e-06,
      "loss": 0.0045,
      "step": 103280
    },
    {
      "epoch": 0.6624699759165964,
      "grad_norm": 0.03008304163813591,
      "learning_rate": 8.457765182224688e-06,
      "loss": 0.0028,
      "step": 103290
    },
    {
      "epoch": 0.6625341128103824,
      "grad_norm": 0.20641933381557465,
      "learning_rate": 8.457360872436933e-06,
      "loss": 0.0019,
      "step": 103300
    },
    {
      "epoch": 0.6625982497041686,
      "grad_norm": 0.09332001209259033,
      "learning_rate": 8.456956519325976e-06,
      "loss": 0.0045,
      "step": 103310
    },
    {
      "epoch": 0.6626623865979546,
      "grad_norm": 0.16936729848384857,
      "learning_rate": 8.456552122896884e-06,
      "loss": 0.0037,
      "step": 103320
    },
    {
      "epoch": 0.6627265234917408,
      "grad_norm": 0.03418375924229622,
      "learning_rate": 8.456147683154722e-06,
      "loss": 0.0039,
      "step": 103330
    },
    {
      "epoch": 0.6627906603855269,
      "grad_norm": 0.10110632330179214,
      "learning_rate": 8.455743200104563e-06,
      "loss": 0.0022,
      "step": 103340
    },
    {
      "epoch": 0.662854797279313,
      "grad_norm": 0.07637202739715576,
      "learning_rate": 8.455338673751467e-06,
      "loss": 0.0045,
      "step": 103350
    },
    {
      "epoch": 0.6629189341730991,
      "grad_norm": 0.02998146414756775,
      "learning_rate": 8.454934104100513e-06,
      "loss": 0.0041,
      "step": 103360
    },
    {
      "epoch": 0.6629830710668851,
      "grad_norm": 0.14679434895515442,
      "learning_rate": 8.454529491156762e-06,
      "loss": 0.0031,
      "step": 103370
    },
    {
      "epoch": 0.6630472079606713,
      "grad_norm": 0.1804307997226715,
      "learning_rate": 8.454124834925289e-06,
      "loss": 0.005,
      "step": 103380
    },
    {
      "epoch": 0.6631113448544573,
      "grad_norm": 0.044211965054273605,
      "learning_rate": 8.45372013541116e-06,
      "loss": 0.0091,
      "step": 103390
    },
    {
      "epoch": 0.6631754817482435,
      "grad_norm": 0.2491360604763031,
      "learning_rate": 8.453315392619453e-06,
      "loss": 0.0031,
      "step": 103400
    },
    {
      "epoch": 0.6632396186420295,
      "grad_norm": 0.08462058752775192,
      "learning_rate": 8.452910606555236e-06,
      "loss": 0.0022,
      "step": 103410
    },
    {
      "epoch": 0.6633037555358157,
      "grad_norm": 0.22145047783851624,
      "learning_rate": 8.45250577722358e-06,
      "loss": 0.0062,
      "step": 103420
    },
    {
      "epoch": 0.6633678924296017,
      "grad_norm": 0.0859028697013855,
      "learning_rate": 8.45210090462956e-06,
      "loss": 0.0022,
      "step": 103430
    },
    {
      "epoch": 0.6634320293233879,
      "grad_norm": 0.08140894770622253,
      "learning_rate": 8.451695988778246e-06,
      "loss": 0.003,
      "step": 103440
    },
    {
      "epoch": 0.6634961662171739,
      "grad_norm": 0.16863515973091125,
      "learning_rate": 8.451291029674717e-06,
      "loss": 0.0059,
      "step": 103450
    },
    {
      "epoch": 0.66356030311096,
      "grad_norm": 0.08412231504917145,
      "learning_rate": 8.450886027324045e-06,
      "loss": 0.0045,
      "step": 103460
    },
    {
      "epoch": 0.6636244400047462,
      "grad_norm": 0.2956025004386902,
      "learning_rate": 8.450480981731303e-06,
      "loss": 0.0036,
      "step": 103470
    },
    {
      "epoch": 0.6636885768985322,
      "grad_norm": 0.18763288855552673,
      "learning_rate": 8.450075892901571e-06,
      "loss": 0.0029,
      "step": 103480
    },
    {
      "epoch": 0.6637527137923184,
      "grad_norm": 0.35756680369377136,
      "learning_rate": 8.449670760839919e-06,
      "loss": 0.0031,
      "step": 103490
    },
    {
      "epoch": 0.6638168506861044,
      "grad_norm": 0.12096337974071503,
      "learning_rate": 8.44926558555143e-06,
      "loss": 0.0033,
      "step": 103500
    },
    {
      "epoch": 0.6638809875798906,
      "grad_norm": 0.2119651436805725,
      "learning_rate": 8.448860367041176e-06,
      "loss": 0.0033,
      "step": 103510
    },
    {
      "epoch": 0.6639451244736766,
      "grad_norm": 0.37487390637397766,
      "learning_rate": 8.448455105314238e-06,
      "loss": 0.0027,
      "step": 103520
    },
    {
      "epoch": 0.6640092613674627,
      "grad_norm": 0.049838390201330185,
      "learning_rate": 8.448049800375691e-06,
      "loss": 0.003,
      "step": 103530
    },
    {
      "epoch": 0.6640733982612488,
      "grad_norm": 0.08873938024044037,
      "learning_rate": 8.447644452230617e-06,
      "loss": 0.0024,
      "step": 103540
    },
    {
      "epoch": 0.6641375351550349,
      "grad_norm": 0.529564380645752,
      "learning_rate": 8.447239060884094e-06,
      "loss": 0.0024,
      "step": 103550
    },
    {
      "epoch": 0.664201672048821,
      "grad_norm": 0.11816108971834183,
      "learning_rate": 8.446833626341202e-06,
      "loss": 0.0031,
      "step": 103560
    },
    {
      "epoch": 0.6642658089426071,
      "grad_norm": 0.08333977311849594,
      "learning_rate": 8.44642814860702e-06,
      "loss": 0.0023,
      "step": 103570
    },
    {
      "epoch": 0.6643299458363932,
      "grad_norm": 0.11790352314710617,
      "learning_rate": 8.446022627686632e-06,
      "loss": 0.0035,
      "step": 103580
    },
    {
      "epoch": 0.6643940827301793,
      "grad_norm": 0.10611840337514877,
      "learning_rate": 8.445617063585116e-06,
      "loss": 0.003,
      "step": 103590
    },
    {
      "epoch": 0.6644582196239653,
      "grad_norm": 0.19753526151180267,
      "learning_rate": 8.445211456307557e-06,
      "loss": 0.0038,
      "step": 103600
    },
    {
      "epoch": 0.6645223565177515,
      "grad_norm": 0.04932020232081413,
      "learning_rate": 8.444805805859036e-06,
      "loss": 0.0058,
      "step": 103610
    },
    {
      "epoch": 0.6645864934115376,
      "grad_norm": 0.17282375693321228,
      "learning_rate": 8.444400112244635e-06,
      "loss": 0.0031,
      "step": 103620
    },
    {
      "epoch": 0.6646506303053237,
      "grad_norm": 0.10250164568424225,
      "learning_rate": 8.44399437546944e-06,
      "loss": 0.003,
      "step": 103630
    },
    {
      "epoch": 0.6647147671991098,
      "grad_norm": 0.29081571102142334,
      "learning_rate": 8.443588595538534e-06,
      "loss": 0.0053,
      "step": 103640
    },
    {
      "epoch": 0.6647789040928959,
      "grad_norm": 0.28330540657043457,
      "learning_rate": 8.443182772457002e-06,
      "loss": 0.0023,
      "step": 103650
    },
    {
      "epoch": 0.664843040986682,
      "grad_norm": 0.24046741425991058,
      "learning_rate": 8.442776906229928e-06,
      "loss": 0.0053,
      "step": 103660
    },
    {
      "epoch": 0.664907177880468,
      "grad_norm": 0.0589718371629715,
      "learning_rate": 8.442370996862402e-06,
      "loss": 0.0012,
      "step": 103670
    },
    {
      "epoch": 0.6649713147742542,
      "grad_norm": 0.14298389852046967,
      "learning_rate": 8.441965044359504e-06,
      "loss": 0.0016,
      "step": 103680
    },
    {
      "epoch": 0.6650354516680402,
      "grad_norm": 0.1267651468515396,
      "learning_rate": 8.441559048726324e-06,
      "loss": 0.0025,
      "step": 103690
    },
    {
      "epoch": 0.6650995885618264,
      "grad_norm": 0.014233440160751343,
      "learning_rate": 8.441153009967951e-06,
      "loss": 0.0016,
      "step": 103700
    },
    {
      "epoch": 0.6651637254556124,
      "grad_norm": 0.18992894887924194,
      "learning_rate": 8.44074692808947e-06,
      "loss": 0.0032,
      "step": 103710
    },
    {
      "epoch": 0.6652278623493986,
      "grad_norm": 0.06404879689216614,
      "learning_rate": 8.440340803095972e-06,
      "loss": 0.0023,
      "step": 103720
    },
    {
      "epoch": 0.6652919992431846,
      "grad_norm": 0.03764347359538078,
      "learning_rate": 8.439934634992547e-06,
      "loss": 0.002,
      "step": 103730
    },
    {
      "epoch": 0.6653561361369708,
      "grad_norm": 0.42796599864959717,
      "learning_rate": 8.439528423784278e-06,
      "loss": 0.0044,
      "step": 103740
    },
    {
      "epoch": 0.6654202730307568,
      "grad_norm": 0.07582218945026398,
      "learning_rate": 8.439122169476262e-06,
      "loss": 0.0028,
      "step": 103750
    },
    {
      "epoch": 0.6654844099245429,
      "grad_norm": 0.30783307552337646,
      "learning_rate": 8.438715872073588e-06,
      "loss": 0.0028,
      "step": 103760
    },
    {
      "epoch": 0.6655485468183291,
      "grad_norm": 0.10885994136333466,
      "learning_rate": 8.438309531581345e-06,
      "loss": 0.0026,
      "step": 103770
    },
    {
      "epoch": 0.6656126837121151,
      "grad_norm": 0.1359860599040985,
      "learning_rate": 8.437903148004627e-06,
      "loss": 0.0048,
      "step": 103780
    },
    {
      "epoch": 0.6656768206059013,
      "grad_norm": 0.107600137591362,
      "learning_rate": 8.437496721348526e-06,
      "loss": 0.0038,
      "step": 103790
    },
    {
      "epoch": 0.6657409574996873,
      "grad_norm": 0.10635979473590851,
      "learning_rate": 8.437090251618134e-06,
      "loss": 0.0028,
      "step": 103800
    },
    {
      "epoch": 0.6658050943934735,
      "grad_norm": 0.11203435063362122,
      "learning_rate": 8.436683738818546e-06,
      "loss": 0.005,
      "step": 103810
    },
    {
      "epoch": 0.6658692312872595,
      "grad_norm": 0.07321670651435852,
      "learning_rate": 8.436277182954852e-06,
      "loss": 0.0028,
      "step": 103820
    },
    {
      "epoch": 0.6659333681810456,
      "grad_norm": 0.1538209766149521,
      "learning_rate": 8.435870584032149e-06,
      "loss": 0.0024,
      "step": 103830
    },
    {
      "epoch": 0.6659975050748317,
      "grad_norm": 0.15127509832382202,
      "learning_rate": 8.435463942055534e-06,
      "loss": 0.0035,
      "step": 103840
    },
    {
      "epoch": 0.6660616419686178,
      "grad_norm": 0.07844885438680649,
      "learning_rate": 8.435057257030099e-06,
      "loss": 0.0027,
      "step": 103850
    },
    {
      "epoch": 0.6661257788624039,
      "grad_norm": 0.13272012770175934,
      "learning_rate": 8.434650528960944e-06,
      "loss": 0.0026,
      "step": 103860
    },
    {
      "epoch": 0.66618991575619,
      "grad_norm": 0.07353035360574722,
      "learning_rate": 8.43424375785316e-06,
      "loss": 0.0027,
      "step": 103870
    },
    {
      "epoch": 0.6662540526499761,
      "grad_norm": 0.0674884021282196,
      "learning_rate": 8.433836943711849e-06,
      "loss": 0.0033,
      "step": 103880
    },
    {
      "epoch": 0.6663181895437622,
      "grad_norm": 0.054299406707286835,
      "learning_rate": 8.433430086542107e-06,
      "loss": 0.0026,
      "step": 103890
    },
    {
      "epoch": 0.6663823264375484,
      "grad_norm": 0.24317777156829834,
      "learning_rate": 8.433023186349032e-06,
      "loss": 0.0022,
      "step": 103900
    },
    {
      "epoch": 0.6664464633313344,
      "grad_norm": 0.1725510209798813,
      "learning_rate": 8.432616243137723e-06,
      "loss": 0.0078,
      "step": 103910
    },
    {
      "epoch": 0.6665106002251205,
      "grad_norm": 0.18429318070411682,
      "learning_rate": 8.432209256913279e-06,
      "loss": 0.0035,
      "step": 103920
    },
    {
      "epoch": 0.6665747371189066,
      "grad_norm": 0.11312185972929001,
      "learning_rate": 8.4318022276808e-06,
      "loss": 0.0034,
      "step": 103930
    },
    {
      "epoch": 0.6666388740126927,
      "grad_norm": 0.07917297631502151,
      "learning_rate": 8.431395155445386e-06,
      "loss": 0.0036,
      "step": 103940
    },
    {
      "epoch": 0.6667030109064788,
      "grad_norm": 0.1329788863658905,
      "learning_rate": 8.430988040212139e-06,
      "loss": 0.0037,
      "step": 103950
    },
    {
      "epoch": 0.6667671478002649,
      "grad_norm": 0.08584097772836685,
      "learning_rate": 8.430580881986159e-06,
      "loss": 0.0036,
      "step": 103960
    },
    {
      "epoch": 0.666831284694051,
      "grad_norm": 0.13817085325717926,
      "learning_rate": 8.43017368077255e-06,
      "loss": 0.0037,
      "step": 103970
    },
    {
      "epoch": 0.6668954215878371,
      "grad_norm": 0.20671610534191132,
      "learning_rate": 8.429766436576413e-06,
      "loss": 0.0025,
      "step": 103980
    },
    {
      "epoch": 0.6669595584816231,
      "grad_norm": 0.2456267923116684,
      "learning_rate": 8.42935914940285e-06,
      "loss": 0.0022,
      "step": 103990
    },
    {
      "epoch": 0.6670236953754093,
      "grad_norm": 0.07276738435029984,
      "learning_rate": 8.428951819256968e-06,
      "loss": 0.0034,
      "step": 104000
    },
    {
      "epoch": 0.6670878322691953,
      "grad_norm": 0.04010990262031555,
      "learning_rate": 8.428544446143867e-06,
      "loss": 0.0035,
      "step": 104010
    },
    {
      "epoch": 0.6671519691629815,
      "grad_norm": 0.10394640266895294,
      "learning_rate": 8.428137030068654e-06,
      "loss": 0.0028,
      "step": 104020
    },
    {
      "epoch": 0.6672161060567675,
      "grad_norm": 0.06314918398857117,
      "learning_rate": 8.427729571036435e-06,
      "loss": 0.0012,
      "step": 104030
    },
    {
      "epoch": 0.6672802429505537,
      "grad_norm": 0.00908320490270853,
      "learning_rate": 8.427322069052315e-06,
      "loss": 0.0033,
      "step": 104040
    },
    {
      "epoch": 0.6673443798443398,
      "grad_norm": 0.1839200258255005,
      "learning_rate": 8.4269145241214e-06,
      "loss": 0.0026,
      "step": 104050
    },
    {
      "epoch": 0.6674085167381258,
      "grad_norm": 0.11998382210731506,
      "learning_rate": 8.426506936248794e-06,
      "loss": 0.002,
      "step": 104060
    },
    {
      "epoch": 0.667472653631912,
      "grad_norm": 0.30450543761253357,
      "learning_rate": 8.42609930543961e-06,
      "loss": 0.0024,
      "step": 104070
    },
    {
      "epoch": 0.667536790525698,
      "grad_norm": 0.12889540195465088,
      "learning_rate": 8.425691631698954e-06,
      "loss": 0.0017,
      "step": 104080
    },
    {
      "epoch": 0.6676009274194842,
      "grad_norm": 0.0068665496073663235,
      "learning_rate": 8.425283915031931e-06,
      "loss": 0.0021,
      "step": 104090
    },
    {
      "epoch": 0.6676650643132702,
      "grad_norm": 0.08512814342975616,
      "learning_rate": 8.424876155443653e-06,
      "loss": 0.0038,
      "step": 104100
    },
    {
      "epoch": 0.6677292012070564,
      "grad_norm": 0.026818742975592613,
      "learning_rate": 8.42446835293923e-06,
      "loss": 0.002,
      "step": 104110
    },
    {
      "epoch": 0.6677933381008424,
      "grad_norm": 0.08405889570713043,
      "learning_rate": 8.42406050752377e-06,
      "loss": 0.0051,
      "step": 104120
    },
    {
      "epoch": 0.6678574749946286,
      "grad_norm": 0.25920823216438293,
      "learning_rate": 8.423652619202386e-06,
      "loss": 0.0036,
      "step": 104130
    },
    {
      "epoch": 0.6679216118884146,
      "grad_norm": 0.1724155843257904,
      "learning_rate": 8.423244687980188e-06,
      "loss": 0.005,
      "step": 104140
    },
    {
      "epoch": 0.6679857487822007,
      "grad_norm": 0.06032963842153549,
      "learning_rate": 8.422836713862285e-06,
      "loss": 0.0026,
      "step": 104150
    },
    {
      "epoch": 0.6680498856759868,
      "grad_norm": 0.03222297877073288,
      "learning_rate": 8.422428696853795e-06,
      "loss": 0.0043,
      "step": 104160
    },
    {
      "epoch": 0.6681140225697729,
      "grad_norm": 0.03689512237906456,
      "learning_rate": 8.422020636959826e-06,
      "loss": 0.0019,
      "step": 104170
    },
    {
      "epoch": 0.6681781594635591,
      "grad_norm": 0.12515857815742493,
      "learning_rate": 8.421612534185493e-06,
      "loss": 0.0025,
      "step": 104180
    },
    {
      "epoch": 0.6682422963573451,
      "grad_norm": 0.09245526045560837,
      "learning_rate": 8.421204388535908e-06,
      "loss": 0.0034,
      "step": 104190
    },
    {
      "epoch": 0.6683064332511313,
      "grad_norm": 0.1860002726316452,
      "learning_rate": 8.42079620001619e-06,
      "loss": 0.0028,
      "step": 104200
    },
    {
      "epoch": 0.6683705701449173,
      "grad_norm": 0.08889582753181458,
      "learning_rate": 8.420387968631448e-06,
      "loss": 0.003,
      "step": 104210
    },
    {
      "epoch": 0.6684347070387034,
      "grad_norm": 0.20289276540279388,
      "learning_rate": 8.419979694386802e-06,
      "loss": 0.0046,
      "step": 104220
    },
    {
      "epoch": 0.6684988439324895,
      "grad_norm": 0.16857382655143738,
      "learning_rate": 8.419571377287366e-06,
      "loss": 0.0028,
      "step": 104230
    },
    {
      "epoch": 0.6685629808262756,
      "grad_norm": 0.012377630919218063,
      "learning_rate": 8.419163017338254e-06,
      "loss": 0.0034,
      "step": 104240
    },
    {
      "epoch": 0.6686271177200617,
      "grad_norm": 0.15620973706245422,
      "learning_rate": 8.41875461454459e-06,
      "loss": 0.0028,
      "step": 104250
    },
    {
      "epoch": 0.6686912546138478,
      "grad_norm": 0.08912920951843262,
      "learning_rate": 8.418346168911485e-06,
      "loss": 0.0016,
      "step": 104260
    },
    {
      "epoch": 0.6687553915076339,
      "grad_norm": 0.25533559918403625,
      "learning_rate": 8.41793768044406e-06,
      "loss": 0.0031,
      "step": 104270
    },
    {
      "epoch": 0.66881952840142,
      "grad_norm": 0.0831306129693985,
      "learning_rate": 8.417529149147431e-06,
      "loss": 0.0021,
      "step": 104280
    },
    {
      "epoch": 0.668883665295206,
      "grad_norm": 0.14187638461589813,
      "learning_rate": 8.417120575026721e-06,
      "loss": 0.0024,
      "step": 104290
    },
    {
      "epoch": 0.6689478021889922,
      "grad_norm": 0.14810237288475037,
      "learning_rate": 8.41671195808705e-06,
      "loss": 0.0027,
      "step": 104300
    },
    {
      "epoch": 0.6690119390827782,
      "grad_norm": 0.22693736851215363,
      "learning_rate": 8.416303298333533e-06,
      "loss": 0.0038,
      "step": 104310
    },
    {
      "epoch": 0.6690760759765644,
      "grad_norm": 0.07919272780418396,
      "learning_rate": 8.415894595771295e-06,
      "loss": 0.0025,
      "step": 104320
    },
    {
      "epoch": 0.6691402128703505,
      "grad_norm": 0.09138061106204987,
      "learning_rate": 8.415485850405456e-06,
      "loss": 0.0027,
      "step": 104330
    },
    {
      "epoch": 0.6692043497641366,
      "grad_norm": 0.21584628522396088,
      "learning_rate": 8.41507706224114e-06,
      "loss": 0.0021,
      "step": 104340
    },
    {
      "epoch": 0.6692684866579227,
      "grad_norm": 0.04365155100822449,
      "learning_rate": 8.414668231283468e-06,
      "loss": 0.0022,
      "step": 104350
    },
    {
      "epoch": 0.6693326235517087,
      "grad_norm": 0.2859312891960144,
      "learning_rate": 8.41425935753756e-06,
      "loss": 0.0035,
      "step": 104360
    },
    {
      "epoch": 0.6693967604454949,
      "grad_norm": 0.09819292277097702,
      "learning_rate": 8.413850441008545e-06,
      "loss": 0.003,
      "step": 104370
    },
    {
      "epoch": 0.6694608973392809,
      "grad_norm": 0.02959253638982773,
      "learning_rate": 8.41344148170154e-06,
      "loss": 0.0017,
      "step": 104380
    },
    {
      "epoch": 0.6695250342330671,
      "grad_norm": 0.26345840096473694,
      "learning_rate": 8.413032479621678e-06,
      "loss": 0.0044,
      "step": 104390
    },
    {
      "epoch": 0.6695891711268531,
      "grad_norm": 0.15691974759101868,
      "learning_rate": 8.412623434774078e-06,
      "loss": 0.0033,
      "step": 104400
    },
    {
      "epoch": 0.6696533080206393,
      "grad_norm": 0.156972736120224,
      "learning_rate": 8.412214347163867e-06,
      "loss": 0.003,
      "step": 104410
    },
    {
      "epoch": 0.6697174449144253,
      "grad_norm": 0.084920734167099,
      "learning_rate": 8.411805216796172e-06,
      "loss": 0.0027,
      "step": 104420
    },
    {
      "epoch": 0.6697815818082115,
      "grad_norm": 0.12769630551338196,
      "learning_rate": 8.41139604367612e-06,
      "loss": 0.0041,
      "step": 104430
    },
    {
      "epoch": 0.6698457187019975,
      "grad_norm": 0.3678871989250183,
      "learning_rate": 8.410986827808836e-06,
      "loss": 0.0047,
      "step": 104440
    },
    {
      "epoch": 0.6699098555957836,
      "grad_norm": 0.06114371865987778,
      "learning_rate": 8.41057756919945e-06,
      "loss": 0.0022,
      "step": 104450
    },
    {
      "epoch": 0.6699739924895698,
      "grad_norm": 0.08031799644231796,
      "learning_rate": 8.41016826785309e-06,
      "loss": 0.0056,
      "step": 104460
    },
    {
      "epoch": 0.6700381293833558,
      "grad_norm": 0.2285887897014618,
      "learning_rate": 8.409758923774885e-06,
      "loss": 0.0015,
      "step": 104470
    },
    {
      "epoch": 0.670102266277142,
      "grad_norm": 0.009799333289265633,
      "learning_rate": 8.409349536969962e-06,
      "loss": 0.0025,
      "step": 104480
    },
    {
      "epoch": 0.670166403170928,
      "grad_norm": 0.16017772257328033,
      "learning_rate": 8.408940107443452e-06,
      "loss": 0.0043,
      "step": 104490
    },
    {
      "epoch": 0.6702305400647142,
      "grad_norm": 0.043988995254039764,
      "learning_rate": 8.40853063520049e-06,
      "loss": 0.0021,
      "step": 104500
    },
    {
      "epoch": 0.6702946769585002,
      "grad_norm": 0.1321130394935608,
      "learning_rate": 8.4081211202462e-06,
      "loss": 0.0031,
      "step": 104510
    },
    {
      "epoch": 0.6703588138522864,
      "grad_norm": 0.23636986315250397,
      "learning_rate": 8.407711562585717e-06,
      "loss": 0.0023,
      "step": 104520
    },
    {
      "epoch": 0.6704229507460724,
      "grad_norm": 0.06349032372236252,
      "learning_rate": 8.407301962224174e-06,
      "loss": 0.0017,
      "step": 104530
    },
    {
      "epoch": 0.6704870876398585,
      "grad_norm": 0.10937687754631042,
      "learning_rate": 8.406892319166701e-06,
      "loss": 0.0031,
      "step": 104540
    },
    {
      "epoch": 0.6705512245336446,
      "grad_norm": 0.3712023198604584,
      "learning_rate": 8.406482633418432e-06,
      "loss": 0.0052,
      "step": 104550
    },
    {
      "epoch": 0.6706153614274307,
      "grad_norm": 0.09075483679771423,
      "learning_rate": 8.406072904984501e-06,
      "loss": 0.0032,
      "step": 104560
    },
    {
      "epoch": 0.6706794983212168,
      "grad_norm": 0.13852781057357788,
      "learning_rate": 8.405663133870044e-06,
      "loss": 0.0035,
      "step": 104570
    },
    {
      "epoch": 0.6707436352150029,
      "grad_norm": 0.3938485085964203,
      "learning_rate": 8.405253320080194e-06,
      "loss": 0.0041,
      "step": 104580
    },
    {
      "epoch": 0.670807772108789,
      "grad_norm": 0.20136603713035583,
      "learning_rate": 8.404843463620083e-06,
      "loss": 0.0049,
      "step": 104590
    },
    {
      "epoch": 0.6708719090025751,
      "grad_norm": 0.13110202550888062,
      "learning_rate": 8.404433564494852e-06,
      "loss": 0.003,
      "step": 104600
    },
    {
      "epoch": 0.6709360458963612,
      "grad_norm": 0.26542943716049194,
      "learning_rate": 8.404023622709636e-06,
      "loss": 0.0027,
      "step": 104610
    },
    {
      "epoch": 0.6710001827901473,
      "grad_norm": 0.1232231929898262,
      "learning_rate": 8.403613638269569e-06,
      "loss": 0.0032,
      "step": 104620
    },
    {
      "epoch": 0.6710643196839334,
      "grad_norm": 0.15485745668411255,
      "learning_rate": 8.403203611179794e-06,
      "loss": 0.0056,
      "step": 104630
    },
    {
      "epoch": 0.6711284565777195,
      "grad_norm": 0.06909587234258652,
      "learning_rate": 8.402793541445443e-06,
      "loss": 0.004,
      "step": 104640
    },
    {
      "epoch": 0.6711925934715056,
      "grad_norm": 0.011975661851465702,
      "learning_rate": 8.402383429071657e-06,
      "loss": 0.0046,
      "step": 104650
    },
    {
      "epoch": 0.6712567303652917,
      "grad_norm": 0.16399651765823364,
      "learning_rate": 8.401973274063576e-06,
      "loss": 0.0029,
      "step": 104660
    },
    {
      "epoch": 0.6713208672590778,
      "grad_norm": 0.04149583727121353,
      "learning_rate": 8.401563076426338e-06,
      "loss": 0.0042,
      "step": 104670
    },
    {
      "epoch": 0.6713850041528638,
      "grad_norm": 0.15252408385276794,
      "learning_rate": 8.401152836165085e-06,
      "loss": 0.0054,
      "step": 104680
    },
    {
      "epoch": 0.67144914104665,
      "grad_norm": 0.051730990409851074,
      "learning_rate": 8.400742553284954e-06,
      "loss": 0.0033,
      "step": 104690
    },
    {
      "epoch": 0.671513277940436,
      "grad_norm": 0.3143163323402405,
      "learning_rate": 8.400332227791089e-06,
      "loss": 0.0034,
      "step": 104700
    },
    {
      "epoch": 0.6715774148342222,
      "grad_norm": 0.11804922670125961,
      "learning_rate": 8.39992185968863e-06,
      "loss": 0.0017,
      "step": 104710
    },
    {
      "epoch": 0.6716415517280082,
      "grad_norm": 0.12357106059789658,
      "learning_rate": 8.399511448982724e-06,
      "loss": 0.0053,
      "step": 104720
    },
    {
      "epoch": 0.6717056886217944,
      "grad_norm": 0.0229622982442379,
      "learning_rate": 8.399100995678506e-06,
      "loss": 0.0027,
      "step": 104730
    },
    {
      "epoch": 0.6717698255155805,
      "grad_norm": 0.2615741789340973,
      "learning_rate": 8.398690499781125e-06,
      "loss": 0.0016,
      "step": 104740
    },
    {
      "epoch": 0.6718339624093665,
      "grad_norm": 0.012564479373395443,
      "learning_rate": 8.398279961295723e-06,
      "loss": 0.0023,
      "step": 104750
    },
    {
      "epoch": 0.6718980993031527,
      "grad_norm": 0.09483960270881653,
      "learning_rate": 8.397869380227444e-06,
      "loss": 0.003,
      "step": 104760
    },
    {
      "epoch": 0.6719622361969387,
      "grad_norm": 0.15407006442546844,
      "learning_rate": 8.397458756581432e-06,
      "loss": 0.0036,
      "step": 104770
    },
    {
      "epoch": 0.6720263730907249,
      "grad_norm": 0.0985838919878006,
      "learning_rate": 8.397048090362836e-06,
      "loss": 0.003,
      "step": 104780
    },
    {
      "epoch": 0.6720905099845109,
      "grad_norm": 0.06903088092803955,
      "learning_rate": 8.396637381576797e-06,
      "loss": 0.0015,
      "step": 104790
    },
    {
      "epoch": 0.6721546468782971,
      "grad_norm": 0.15456879138946533,
      "learning_rate": 8.396226630228466e-06,
      "loss": 0.0033,
      "step": 104800
    },
    {
      "epoch": 0.6722187837720831,
      "grad_norm": 0.14249612390995026,
      "learning_rate": 8.395815836322988e-06,
      "loss": 0.004,
      "step": 104810
    },
    {
      "epoch": 0.6722829206658693,
      "grad_norm": 0.08765123039484024,
      "learning_rate": 8.395404999865511e-06,
      "loss": 0.0019,
      "step": 104820
    },
    {
      "epoch": 0.6723470575596553,
      "grad_norm": 0.09199008345603943,
      "learning_rate": 8.394994120861182e-06,
      "loss": 0.0013,
      "step": 104830
    },
    {
      "epoch": 0.6724111944534414,
      "grad_norm": 0.09696024656295776,
      "learning_rate": 8.39458319931515e-06,
      "loss": 0.0023,
      "step": 104840
    },
    {
      "epoch": 0.6724753313472275,
      "grad_norm": 0.26852133870124817,
      "learning_rate": 8.394172235232564e-06,
      "loss": 0.0034,
      "step": 104850
    },
    {
      "epoch": 0.6725394682410136,
      "grad_norm": 0.21888944506645203,
      "learning_rate": 8.393761228618576e-06,
      "loss": 0.0027,
      "step": 104860
    },
    {
      "epoch": 0.6726036051347997,
      "grad_norm": 0.14674261212348938,
      "learning_rate": 8.393350179478333e-06,
      "loss": 0.0029,
      "step": 104870
    },
    {
      "epoch": 0.6726677420285858,
      "grad_norm": 0.0999063029885292,
      "learning_rate": 8.392939087816987e-06,
      "loss": 0.0038,
      "step": 104880
    },
    {
      "epoch": 0.672731878922372,
      "grad_norm": 0.18848982453346252,
      "learning_rate": 8.39252795363969e-06,
      "loss": 0.0019,
      "step": 104890
    },
    {
      "epoch": 0.672796015816158,
      "grad_norm": 0.11278213560581207,
      "learning_rate": 8.392116776951592e-06,
      "loss": 0.003,
      "step": 104900
    },
    {
      "epoch": 0.6728601527099441,
      "grad_norm": 0.043803002685308456,
      "learning_rate": 8.391705557757848e-06,
      "loss": 0.0025,
      "step": 104910
    },
    {
      "epoch": 0.6729242896037302,
      "grad_norm": 0.09668834507465363,
      "learning_rate": 8.391294296063608e-06,
      "loss": 0.004,
      "step": 104920
    },
    {
      "epoch": 0.6729884264975163,
      "grad_norm": 0.0840938612818718,
      "learning_rate": 8.390882991874028e-06,
      "loss": 0.0025,
      "step": 104930
    },
    {
      "epoch": 0.6730525633913024,
      "grad_norm": 0.22783967852592468,
      "learning_rate": 8.390471645194258e-06,
      "loss": 0.0025,
      "step": 104940
    },
    {
      "epoch": 0.6731167002850885,
      "grad_norm": 0.1316479593515396,
      "learning_rate": 8.390060256029457e-06,
      "loss": 0.0033,
      "step": 104950
    },
    {
      "epoch": 0.6731808371788746,
      "grad_norm": 0.04257906228303909,
      "learning_rate": 8.38964882438478e-06,
      "loss": 0.004,
      "step": 104960
    },
    {
      "epoch": 0.6732449740726607,
      "grad_norm": 0.16650620102882385,
      "learning_rate": 8.389237350265378e-06,
      "loss": 0.0049,
      "step": 104970
    },
    {
      "epoch": 0.6733091109664467,
      "grad_norm": 0.04592067003250122,
      "learning_rate": 8.38882583367641e-06,
      "loss": 0.003,
      "step": 104980
    },
    {
      "epoch": 0.6733732478602329,
      "grad_norm": 0.040006231516599655,
      "learning_rate": 8.388414274623034e-06,
      "loss": 0.0022,
      "step": 104990
    },
    {
      "epoch": 0.6734373847540189,
      "grad_norm": 0.07911369949579239,
      "learning_rate": 8.388002673110406e-06,
      "loss": 0.0026,
      "step": 105000
    },
    {
      "epoch": 0.6735015216478051,
      "grad_norm": 0.15251368284225464,
      "learning_rate": 8.387591029143681e-06,
      "loss": 0.0027,
      "step": 105010
    },
    {
      "epoch": 0.6735656585415911,
      "grad_norm": 0.05519964545965195,
      "learning_rate": 8.387179342728019e-06,
      "loss": 0.0025,
      "step": 105020
    },
    {
      "epoch": 0.6736297954353773,
      "grad_norm": 0.16007065773010254,
      "learning_rate": 8.38676761386858e-06,
      "loss": 0.0026,
      "step": 105030
    },
    {
      "epoch": 0.6736939323291634,
      "grad_norm": 0.09965284168720245,
      "learning_rate": 8.386355842570522e-06,
      "loss": 0.0024,
      "step": 105040
    },
    {
      "epoch": 0.6737580692229495,
      "grad_norm": 0.09603425860404968,
      "learning_rate": 8.385944028839006e-06,
      "loss": 0.0026,
      "step": 105050
    },
    {
      "epoch": 0.6738222061167356,
      "grad_norm": 0.06844042986631393,
      "learning_rate": 8.38553217267919e-06,
      "loss": 0.0026,
      "step": 105060
    },
    {
      "epoch": 0.6738863430105216,
      "grad_norm": 0.05518994480371475,
      "learning_rate": 8.385120274096238e-06,
      "loss": 0.0014,
      "step": 105070
    },
    {
      "epoch": 0.6739504799043078,
      "grad_norm": 0.2585274875164032,
      "learning_rate": 8.384708333095308e-06,
      "loss": 0.003,
      "step": 105080
    },
    {
      "epoch": 0.6740146167980938,
      "grad_norm": 0.08640757948160172,
      "learning_rate": 8.384296349681565e-06,
      "loss": 0.003,
      "step": 105090
    },
    {
      "epoch": 0.67407875369188,
      "grad_norm": 0.2759339213371277,
      "learning_rate": 8.383884323860168e-06,
      "loss": 0.0045,
      "step": 105100
    },
    {
      "epoch": 0.674142890585666,
      "grad_norm": 0.10095436871051788,
      "learning_rate": 8.383472255636285e-06,
      "loss": 0.0022,
      "step": 105110
    },
    {
      "epoch": 0.6742070274794522,
      "grad_norm": 0.2735072374343872,
      "learning_rate": 8.383060145015075e-06,
      "loss": 0.0028,
      "step": 105120
    },
    {
      "epoch": 0.6742711643732382,
      "grad_norm": 0.03803719952702522,
      "learning_rate": 8.382647992001703e-06,
      "loss": 0.0022,
      "step": 105130
    },
    {
      "epoch": 0.6743353012670243,
      "grad_norm": 0.14733968675136566,
      "learning_rate": 8.382235796601334e-06,
      "loss": 0.0028,
      "step": 105140
    },
    {
      "epoch": 0.6743994381608104,
      "grad_norm": 0.20038215816020966,
      "learning_rate": 8.381823558819133e-06,
      "loss": 0.0029,
      "step": 105150
    },
    {
      "epoch": 0.6744635750545965,
      "grad_norm": 0.059229981154203415,
      "learning_rate": 8.381411278660268e-06,
      "loss": 0.0037,
      "step": 105160
    },
    {
      "epoch": 0.6745277119483827,
      "grad_norm": 0.1375369429588318,
      "learning_rate": 8.3809989561299e-06,
      "loss": 0.0025,
      "step": 105170
    },
    {
      "epoch": 0.6745918488421687,
      "grad_norm": 0.055774934589862823,
      "learning_rate": 8.380586591233201e-06,
      "loss": 0.0024,
      "step": 105180
    },
    {
      "epoch": 0.6746559857359549,
      "grad_norm": 0.11341521143913269,
      "learning_rate": 8.380174183975336e-06,
      "loss": 0.0026,
      "step": 105190
    },
    {
      "epoch": 0.6747201226297409,
      "grad_norm": 0.20973770320415497,
      "learning_rate": 8.37976173436147e-06,
      "loss": 0.0048,
      "step": 105200
    },
    {
      "epoch": 0.674784259523527,
      "grad_norm": 0.15058109164237976,
      "learning_rate": 8.379349242396777e-06,
      "loss": 0.0063,
      "step": 105210
    },
    {
      "epoch": 0.6748483964173131,
      "grad_norm": 0.09401237219572067,
      "learning_rate": 8.378936708086422e-06,
      "loss": 0.0039,
      "step": 105220
    },
    {
      "epoch": 0.6749125333110992,
      "grad_norm": 0.16512343287467957,
      "learning_rate": 8.378524131435575e-06,
      "loss": 0.0021,
      "step": 105230
    },
    {
      "epoch": 0.6749766702048853,
      "grad_norm": 0.08649599552154541,
      "learning_rate": 8.378111512449406e-06,
      "loss": 0.0025,
      "step": 105240
    },
    {
      "epoch": 0.6750408070986714,
      "grad_norm": 0.05820036306977272,
      "learning_rate": 8.377698851133085e-06,
      "loss": 0.0022,
      "step": 105250
    },
    {
      "epoch": 0.6751049439924575,
      "grad_norm": 0.12218081206083298,
      "learning_rate": 8.377286147491784e-06,
      "loss": 0.003,
      "step": 105260
    },
    {
      "epoch": 0.6751690808862436,
      "grad_norm": 0.5234624147415161,
      "learning_rate": 8.376873401530674e-06,
      "loss": 0.0043,
      "step": 105270
    },
    {
      "epoch": 0.6752332177800296,
      "grad_norm": 0.20814195275306702,
      "learning_rate": 8.376460613254926e-06,
      "loss": 0.0019,
      "step": 105280
    },
    {
      "epoch": 0.6752973546738158,
      "grad_norm": 0.10753771662712097,
      "learning_rate": 8.376047782669713e-06,
      "loss": 0.0042,
      "step": 105290
    },
    {
      "epoch": 0.6753614915676018,
      "grad_norm": 0.12748068571090698,
      "learning_rate": 8.37563490978021e-06,
      "loss": 0.0031,
      "step": 105300
    },
    {
      "epoch": 0.675425628461388,
      "grad_norm": 0.21075694262981415,
      "learning_rate": 8.375221994591589e-06,
      "loss": 0.0032,
      "step": 105310
    },
    {
      "epoch": 0.6754897653551741,
      "grad_norm": 0.0806947872042656,
      "learning_rate": 8.374809037109024e-06,
      "loss": 0.0034,
      "step": 105320
    },
    {
      "epoch": 0.6755539022489602,
      "grad_norm": 0.21661756932735443,
      "learning_rate": 8.374396037337688e-06,
      "loss": 0.0024,
      "step": 105330
    },
    {
      "epoch": 0.6756180391427463,
      "grad_norm": 0.07489059120416641,
      "learning_rate": 8.373982995282762e-06,
      "loss": 0.0021,
      "step": 105340
    },
    {
      "epoch": 0.6756821760365324,
      "grad_norm": 0.11895561218261719,
      "learning_rate": 8.373569910949414e-06,
      "loss": 0.0021,
      "step": 105350
    },
    {
      "epoch": 0.6757463129303185,
      "grad_norm": 0.04192807152867317,
      "learning_rate": 8.373156784342825e-06,
      "loss": 0.0026,
      "step": 105360
    },
    {
      "epoch": 0.6758104498241045,
      "grad_norm": 0.03970417380332947,
      "learning_rate": 8.372743615468171e-06,
      "loss": 0.0021,
      "step": 105370
    },
    {
      "epoch": 0.6758745867178907,
      "grad_norm": 0.11419545859098434,
      "learning_rate": 8.372330404330628e-06,
      "loss": 0.0056,
      "step": 105380
    },
    {
      "epoch": 0.6759387236116767,
      "grad_norm": 0.31705695390701294,
      "learning_rate": 8.371917150935378e-06,
      "loss": 0.0032,
      "step": 105390
    },
    {
      "epoch": 0.6760028605054629,
      "grad_norm": 0.03702298924326897,
      "learning_rate": 8.371503855287593e-06,
      "loss": 0.0019,
      "step": 105400
    },
    {
      "epoch": 0.6760669973992489,
      "grad_norm": 0.12052658200263977,
      "learning_rate": 8.371090517392455e-06,
      "loss": 0.0049,
      "step": 105410
    },
    {
      "epoch": 0.6761311342930351,
      "grad_norm": 0.012704034335911274,
      "learning_rate": 8.370677137255145e-06,
      "loss": 0.0028,
      "step": 105420
    },
    {
      "epoch": 0.6761952711868211,
      "grad_norm": 0.11155135929584503,
      "learning_rate": 8.370263714880843e-06,
      "loss": 0.0034,
      "step": 105430
    },
    {
      "epoch": 0.6762594080806072,
      "grad_norm": 0.14455078542232513,
      "learning_rate": 8.369850250274725e-06,
      "loss": 0.004,
      "step": 105440
    },
    {
      "epoch": 0.6763235449743934,
      "grad_norm": 0.15305106341838837,
      "learning_rate": 8.369436743441977e-06,
      "loss": 0.0026,
      "step": 105450
    },
    {
      "epoch": 0.6763876818681794,
      "grad_norm": 0.2909550070762634,
      "learning_rate": 8.369023194387777e-06,
      "loss": 0.0043,
      "step": 105460
    },
    {
      "epoch": 0.6764518187619656,
      "grad_norm": 0.2682478427886963,
      "learning_rate": 8.368609603117307e-06,
      "loss": 0.0028,
      "step": 105470
    },
    {
      "epoch": 0.6765159556557516,
      "grad_norm": 0.13300544023513794,
      "learning_rate": 8.368195969635756e-06,
      "loss": 0.0023,
      "step": 105480
    },
    {
      "epoch": 0.6765800925495378,
      "grad_norm": 0.1423490047454834,
      "learning_rate": 8.367782293948299e-06,
      "loss": 0.0029,
      "step": 105490
    },
    {
      "epoch": 0.6766442294433238,
      "grad_norm": 0.10811354219913483,
      "learning_rate": 8.367368576060122e-06,
      "loss": 0.0032,
      "step": 105500
    },
    {
      "epoch": 0.67670836633711,
      "grad_norm": 0.15842215716838837,
      "learning_rate": 8.366954815976412e-06,
      "loss": 0.0031,
      "step": 105510
    },
    {
      "epoch": 0.676772503230896,
      "grad_norm": 0.03228527307510376,
      "learning_rate": 8.366541013702351e-06,
      "loss": 0.0021,
      "step": 105520
    },
    {
      "epoch": 0.6768366401246821,
      "grad_norm": 0.23487135767936707,
      "learning_rate": 8.366127169243126e-06,
      "loss": 0.0022,
      "step": 105530
    },
    {
      "epoch": 0.6769007770184682,
      "grad_norm": 0.10690513998270035,
      "learning_rate": 8.365713282603923e-06,
      "loss": 0.0018,
      "step": 105540
    },
    {
      "epoch": 0.6769649139122543,
      "grad_norm": 0.23628225922584534,
      "learning_rate": 8.365299353789924e-06,
      "loss": 0.003,
      "step": 105550
    },
    {
      "epoch": 0.6770290508060404,
      "grad_norm": 0.14227193593978882,
      "learning_rate": 8.364885382806321e-06,
      "loss": 0.01,
      "step": 105560
    },
    {
      "epoch": 0.6770931876998265,
      "grad_norm": 0.09515100717544556,
      "learning_rate": 8.364471369658299e-06,
      "loss": 0.0022,
      "step": 105570
    },
    {
      "epoch": 0.6771573245936126,
      "grad_norm": 0.3159456253051758,
      "learning_rate": 8.364057314351044e-06,
      "loss": 0.0026,
      "step": 105580
    },
    {
      "epoch": 0.6772214614873987,
      "grad_norm": 0.12352485954761505,
      "learning_rate": 8.36364321688975e-06,
      "loss": 0.0028,
      "step": 105590
    },
    {
      "epoch": 0.6772855983811849,
      "grad_norm": 0.06309013813734055,
      "learning_rate": 8.3632290772796e-06,
      "loss": 0.002,
      "step": 105600
    },
    {
      "epoch": 0.6773497352749709,
      "grad_norm": 0.08990509808063507,
      "learning_rate": 8.362814895525787e-06,
      "loss": 0.0024,
      "step": 105610
    },
    {
      "epoch": 0.677413872168757,
      "grad_norm": 0.2627856433391571,
      "learning_rate": 8.362400671633501e-06,
      "loss": 0.0028,
      "step": 105620
    },
    {
      "epoch": 0.6774780090625431,
      "grad_norm": 0.16149525344371796,
      "learning_rate": 8.361986405607931e-06,
      "loss": 0.0022,
      "step": 105630
    },
    {
      "epoch": 0.6775421459563292,
      "grad_norm": 0.20449747145175934,
      "learning_rate": 8.361572097454268e-06,
      "loss": 0.0037,
      "step": 105640
    },
    {
      "epoch": 0.6776062828501153,
      "grad_norm": 0.48451846837997437,
      "learning_rate": 8.361157747177703e-06,
      "loss": 0.0081,
      "step": 105650
    },
    {
      "epoch": 0.6776704197439014,
      "grad_norm": 0.09318164736032486,
      "learning_rate": 8.360743354783432e-06,
      "loss": 0.0016,
      "step": 105660
    },
    {
      "epoch": 0.6777345566376874,
      "grad_norm": 0.1288747489452362,
      "learning_rate": 8.360328920276644e-06,
      "loss": 0.0029,
      "step": 105670
    },
    {
      "epoch": 0.6777986935314736,
      "grad_norm": 0.05139090493321419,
      "learning_rate": 8.359914443662532e-06,
      "loss": 0.0033,
      "step": 105680
    },
    {
      "epoch": 0.6778628304252596,
      "grad_norm": 0.11765347421169281,
      "learning_rate": 8.359499924946291e-06,
      "loss": 0.0027,
      "step": 105690
    },
    {
      "epoch": 0.6779269673190458,
      "grad_norm": 0.22567494213581085,
      "learning_rate": 8.359085364133115e-06,
      "loss": 0.0028,
      "step": 105700
    },
    {
      "epoch": 0.6779911042128318,
      "grad_norm": 0.1894441545009613,
      "learning_rate": 8.358670761228198e-06,
      "loss": 0.0045,
      "step": 105710
    },
    {
      "epoch": 0.678055241106618,
      "grad_norm": 0.09808771312236786,
      "learning_rate": 8.358256116236738e-06,
      "loss": 0.0025,
      "step": 105720
    },
    {
      "epoch": 0.6781193780004041,
      "grad_norm": 0.1675807237625122,
      "learning_rate": 8.357841429163927e-06,
      "loss": 0.0018,
      "step": 105730
    },
    {
      "epoch": 0.6781835148941902,
      "grad_norm": 0.2744348645210266,
      "learning_rate": 8.357426700014963e-06,
      "loss": 0.0015,
      "step": 105740
    },
    {
      "epoch": 0.6782476517879763,
      "grad_norm": 0.16625988483428955,
      "learning_rate": 8.357011928795045e-06,
      "loss": 0.0032,
      "step": 105750
    },
    {
      "epoch": 0.6783117886817623,
      "grad_norm": 0.04192342236638069,
      "learning_rate": 8.356597115509365e-06,
      "loss": 0.0041,
      "step": 105760
    },
    {
      "epoch": 0.6783759255755485,
      "grad_norm": 0.02599765732884407,
      "learning_rate": 8.356182260163128e-06,
      "loss": 0.0026,
      "step": 105770
    },
    {
      "epoch": 0.6784400624693345,
      "grad_norm": 0.18142566084861755,
      "learning_rate": 8.355767362761526e-06,
      "loss": 0.002,
      "step": 105780
    },
    {
      "epoch": 0.6785041993631207,
      "grad_norm": 0.12043923884630203,
      "learning_rate": 8.355352423309762e-06,
      "loss": 0.0055,
      "step": 105790
    },
    {
      "epoch": 0.6785683362569067,
      "grad_norm": 0.05678357183933258,
      "learning_rate": 8.354937441813032e-06,
      "loss": 0.0047,
      "step": 105800
    },
    {
      "epoch": 0.6786324731506929,
      "grad_norm": 0.04629009962081909,
      "learning_rate": 8.354522418276541e-06,
      "loss": 0.0045,
      "step": 105810
    },
    {
      "epoch": 0.6786966100444789,
      "grad_norm": 0.19256654381752014,
      "learning_rate": 8.354107352705484e-06,
      "loss": 0.0023,
      "step": 105820
    },
    {
      "epoch": 0.678760746938265,
      "grad_norm": 0.015678269788622856,
      "learning_rate": 8.353692245105066e-06,
      "loss": 0.0044,
      "step": 105830
    },
    {
      "epoch": 0.6788248838320511,
      "grad_norm": 0.11686255782842636,
      "learning_rate": 8.353277095480487e-06,
      "loss": 0.0101,
      "step": 105840
    },
    {
      "epoch": 0.6788890207258372,
      "grad_norm": 0.20758382976055145,
      "learning_rate": 8.352861903836951e-06,
      "loss": 0.0034,
      "step": 105850
    },
    {
      "epoch": 0.6789531576196233,
      "grad_norm": 0.11493109166622162,
      "learning_rate": 8.352446670179656e-06,
      "loss": 0.0027,
      "step": 105860
    },
    {
      "epoch": 0.6790172945134094,
      "grad_norm": 0.17362689971923828,
      "learning_rate": 8.35203139451381e-06,
      "loss": 0.0031,
      "step": 105870
    },
    {
      "epoch": 0.6790814314071956,
      "grad_norm": 0.09448014944791794,
      "learning_rate": 8.351616076844615e-06,
      "loss": 0.0032,
      "step": 105880
    },
    {
      "epoch": 0.6791455683009816,
      "grad_norm": 0.07692663371562958,
      "learning_rate": 8.351200717177276e-06,
      "loss": 0.0033,
      "step": 105890
    },
    {
      "epoch": 0.6792097051947678,
      "grad_norm": 0.2039974331855774,
      "learning_rate": 8.350785315516997e-06,
      "loss": 0.0036,
      "step": 105900
    },
    {
      "epoch": 0.6792738420885538,
      "grad_norm": 0.4582759737968445,
      "learning_rate": 8.350369871868982e-06,
      "loss": 0.0033,
      "step": 105910
    },
    {
      "epoch": 0.6793379789823399,
      "grad_norm": 0.28730398416519165,
      "learning_rate": 8.349954386238437e-06,
      "loss": 0.0025,
      "step": 105920
    },
    {
      "epoch": 0.679402115876126,
      "grad_norm": 0.11131205409765244,
      "learning_rate": 8.34953885863057e-06,
      "loss": 0.0011,
      "step": 105930
    },
    {
      "epoch": 0.6794662527699121,
      "grad_norm": 0.3950798213481903,
      "learning_rate": 8.349123289050589e-06,
      "loss": 0.0023,
      "step": 105940
    },
    {
      "epoch": 0.6795303896636982,
      "grad_norm": 0.13830557465553284,
      "learning_rate": 8.348707677503698e-06,
      "loss": 0.0033,
      "step": 105950
    },
    {
      "epoch": 0.6795945265574843,
      "grad_norm": 0.20884209871292114,
      "learning_rate": 8.348292023995108e-06,
      "loss": 0.0031,
      "step": 105960
    },
    {
      "epoch": 0.6796586634512704,
      "grad_norm": 0.22090451419353485,
      "learning_rate": 8.347876328530024e-06,
      "loss": 0.0024,
      "step": 105970
    },
    {
      "epoch": 0.6797228003450565,
      "grad_norm": 0.29745471477508545,
      "learning_rate": 8.347460591113658e-06,
      "loss": 0.005,
      "step": 105980
    },
    {
      "epoch": 0.6797869372388425,
      "grad_norm": 0.09324681013822556,
      "learning_rate": 8.34704481175122e-06,
      "loss": 0.0036,
      "step": 105990
    },
    {
      "epoch": 0.6798510741326287,
      "grad_norm": 0.28559282422065735,
      "learning_rate": 8.346628990447916e-06,
      "loss": 0.003,
      "step": 106000
    },
    {
      "epoch": 0.6799152110264148,
      "grad_norm": 0.1490248143672943,
      "learning_rate": 8.34621312720896e-06,
      "loss": 0.0045,
      "step": 106010
    },
    {
      "epoch": 0.6799793479202009,
      "grad_norm": 0.04657105728983879,
      "learning_rate": 8.345797222039562e-06,
      "loss": 0.0033,
      "step": 106020
    },
    {
      "epoch": 0.680043484813987,
      "grad_norm": 0.09144826978445053,
      "learning_rate": 8.345381274944932e-06,
      "loss": 0.007,
      "step": 106030
    },
    {
      "epoch": 0.6801076217077731,
      "grad_norm": 0.11840663850307465,
      "learning_rate": 8.344965285930286e-06,
      "loss": 0.0028,
      "step": 106040
    },
    {
      "epoch": 0.6801717586015592,
      "grad_norm": 0.18734104931354523,
      "learning_rate": 8.344549255000833e-06,
      "loss": 0.0026,
      "step": 106050
    },
    {
      "epoch": 0.6802358954953452,
      "grad_norm": 0.1605960726737976,
      "learning_rate": 8.344133182161788e-06,
      "loss": 0.0021,
      "step": 106060
    },
    {
      "epoch": 0.6803000323891314,
      "grad_norm": 0.18541651964187622,
      "learning_rate": 8.343717067418364e-06,
      "loss": 0.0032,
      "step": 106070
    },
    {
      "epoch": 0.6803641692829174,
      "grad_norm": 0.14585164189338684,
      "learning_rate": 8.343300910775777e-06,
      "loss": 0.0025,
      "step": 106080
    },
    {
      "epoch": 0.6804283061767036,
      "grad_norm": 0.27248677611351013,
      "learning_rate": 8.342884712239238e-06,
      "loss": 0.0021,
      "step": 106090
    },
    {
      "epoch": 0.6804924430704896,
      "grad_norm": 0.3925865590572357,
      "learning_rate": 8.342468471813965e-06,
      "loss": 0.0065,
      "step": 106100
    },
    {
      "epoch": 0.6805565799642758,
      "grad_norm": 0.08288202434778214,
      "learning_rate": 8.342052189505175e-06,
      "loss": 0.0036,
      "step": 106110
    },
    {
      "epoch": 0.6806207168580618,
      "grad_norm": 0.20843671262264252,
      "learning_rate": 8.34163586531808e-06,
      "loss": 0.0015,
      "step": 106120
    },
    {
      "epoch": 0.680684853751848,
      "grad_norm": 0.10642247647047043,
      "learning_rate": 8.3412194992579e-06,
      "loss": 0.0024,
      "step": 106130
    },
    {
      "epoch": 0.680748990645634,
      "grad_norm": 0.07317498326301575,
      "learning_rate": 8.340803091329852e-06,
      "loss": 0.0031,
      "step": 106140
    },
    {
      "epoch": 0.6808131275394201,
      "grad_norm": 0.09721281379461288,
      "learning_rate": 8.340386641539153e-06,
      "loss": 0.0045,
      "step": 106150
    },
    {
      "epoch": 0.6808772644332063,
      "grad_norm": 0.08040506392717361,
      "learning_rate": 8.339970149891024e-06,
      "loss": 0.003,
      "step": 106160
    },
    {
      "epoch": 0.6809414013269923,
      "grad_norm": 0.2040429264307022,
      "learning_rate": 8.33955361639068e-06,
      "loss": 0.0035,
      "step": 106170
    },
    {
      "epoch": 0.6810055382207785,
      "grad_norm": 0.2061181366443634,
      "learning_rate": 8.339137041043343e-06,
      "loss": 0.0025,
      "step": 106180
    },
    {
      "epoch": 0.6810696751145645,
      "grad_norm": 0.135468527674675,
      "learning_rate": 8.338720423854233e-06,
      "loss": 0.0032,
      "step": 106190
    },
    {
      "epoch": 0.6811338120083507,
      "grad_norm": 0.09755928814411163,
      "learning_rate": 8.33830376482857e-06,
      "loss": 0.009,
      "step": 106200
    },
    {
      "epoch": 0.6811979489021367,
      "grad_norm": 0.3986220955848694,
      "learning_rate": 8.337887063971575e-06,
      "loss": 0.0035,
      "step": 106210
    },
    {
      "epoch": 0.6812620857959228,
      "grad_norm": 0.22247038781642914,
      "learning_rate": 8.337470321288468e-06,
      "loss": 0.0035,
      "step": 106220
    },
    {
      "epoch": 0.6813262226897089,
      "grad_norm": 0.16053420305252075,
      "learning_rate": 8.337053536784474e-06,
      "loss": 0.0051,
      "step": 106230
    },
    {
      "epoch": 0.681390359583495,
      "grad_norm": 0.08029992133378983,
      "learning_rate": 8.336636710464813e-06,
      "loss": 0.0033,
      "step": 106240
    },
    {
      "epoch": 0.6814544964772811,
      "grad_norm": 0.19211438298225403,
      "learning_rate": 8.336219842334712e-06,
      "loss": 0.0026,
      "step": 106250
    },
    {
      "epoch": 0.6815186333710672,
      "grad_norm": 0.21166279911994934,
      "learning_rate": 8.335802932399389e-06,
      "loss": 0.0022,
      "step": 106260
    },
    {
      "epoch": 0.6815827702648533,
      "grad_norm": 0.11894948780536652,
      "learning_rate": 8.335385980664072e-06,
      "loss": 0.0019,
      "step": 106270
    },
    {
      "epoch": 0.6816469071586394,
      "grad_norm": 0.11670980602502823,
      "learning_rate": 8.334968987133987e-06,
      "loss": 0.0025,
      "step": 106280
    },
    {
      "epoch": 0.6817110440524256,
      "grad_norm": 0.23702040314674377,
      "learning_rate": 8.334551951814354e-06,
      "loss": 0.0016,
      "step": 106290
    },
    {
      "epoch": 0.6817751809462116,
      "grad_norm": 0.13341011106967926,
      "learning_rate": 8.334134874710403e-06,
      "loss": 0.0033,
      "step": 106300
    },
    {
      "epoch": 0.6818393178399977,
      "grad_norm": 0.0981462150812149,
      "learning_rate": 8.33371775582736e-06,
      "loss": 0.0045,
      "step": 106310
    },
    {
      "epoch": 0.6819034547337838,
      "grad_norm": 0.0713023766875267,
      "learning_rate": 8.33330059517045e-06,
      "loss": 0.0016,
      "step": 106320
    },
    {
      "epoch": 0.6819675916275699,
      "grad_norm": 0.0074376314878463745,
      "learning_rate": 8.332883392744903e-06,
      "loss": 0.0026,
      "step": 106330
    },
    {
      "epoch": 0.682031728521356,
      "grad_norm": 0.11056984215974808,
      "learning_rate": 8.332466148555944e-06,
      "loss": 0.0023,
      "step": 106340
    },
    {
      "epoch": 0.6820958654151421,
      "grad_norm": 0.2786552608013153,
      "learning_rate": 8.332048862608801e-06,
      "loss": 0.0031,
      "step": 106350
    },
    {
      "epoch": 0.6821600023089281,
      "grad_norm": 0.10690515488386154,
      "learning_rate": 8.331631534908706e-06,
      "loss": 0.0014,
      "step": 106360
    },
    {
      "epoch": 0.6822241392027143,
      "grad_norm": 0.1778476983308792,
      "learning_rate": 8.331214165460885e-06,
      "loss": 0.005,
      "step": 106370
    },
    {
      "epoch": 0.6822882760965003,
      "grad_norm": 0.3530474603176117,
      "learning_rate": 8.330796754270572e-06,
      "loss": 0.005,
      "step": 106380
    },
    {
      "epoch": 0.6823524129902865,
      "grad_norm": 0.15935179591178894,
      "learning_rate": 8.330379301342994e-06,
      "loss": 0.0027,
      "step": 106390
    },
    {
      "epoch": 0.6824165498840725,
      "grad_norm": 0.21754521131515503,
      "learning_rate": 8.329961806683384e-06,
      "loss": 0.0016,
      "step": 106400
    },
    {
      "epoch": 0.6824806867778587,
      "grad_norm": 0.5963869690895081,
      "learning_rate": 8.329544270296973e-06,
      "loss": 0.0104,
      "step": 106410
    },
    {
      "epoch": 0.6825448236716447,
      "grad_norm": 0.14186711609363556,
      "learning_rate": 8.329126692188991e-06,
      "loss": 0.0041,
      "step": 106420
    },
    {
      "epoch": 0.6826089605654309,
      "grad_norm": 0.10958068817853928,
      "learning_rate": 8.328709072364674e-06,
      "loss": 0.0019,
      "step": 106430
    },
    {
      "epoch": 0.682673097459217,
      "grad_norm": 0.1777481883764267,
      "learning_rate": 8.328291410829252e-06,
      "loss": 0.0051,
      "step": 106440
    },
    {
      "epoch": 0.682737234353003,
      "grad_norm": 0.4342871904373169,
      "learning_rate": 8.327873707587961e-06,
      "loss": 0.0014,
      "step": 106450
    },
    {
      "epoch": 0.6828013712467892,
      "grad_norm": 0.20745062828063965,
      "learning_rate": 8.327455962646034e-06,
      "loss": 0.0018,
      "step": 106460
    },
    {
      "epoch": 0.6828655081405752,
      "grad_norm": 0.18711547553539276,
      "learning_rate": 8.327038176008705e-06,
      "loss": 0.0032,
      "step": 106470
    },
    {
      "epoch": 0.6829296450343614,
      "grad_norm": 0.1682271659374237,
      "learning_rate": 8.32662034768121e-06,
      "loss": 0.001,
      "step": 106480
    },
    {
      "epoch": 0.6829937819281474,
      "grad_norm": 0.06280054152011871,
      "learning_rate": 8.326202477668786e-06,
      "loss": 0.0031,
      "step": 106490
    },
    {
      "epoch": 0.6830579188219336,
      "grad_norm": 0.4445003867149353,
      "learning_rate": 8.325784565976665e-06,
      "loss": 0.0032,
      "step": 106500
    },
    {
      "epoch": 0.6831220557157196,
      "grad_norm": 0.27170541882514954,
      "learning_rate": 8.325366612610088e-06,
      "loss": 0.0043,
      "step": 106510
    },
    {
      "epoch": 0.6831861926095057,
      "grad_norm": 0.034090470522642136,
      "learning_rate": 8.324948617574292e-06,
      "loss": 0.0022,
      "step": 106520
    },
    {
      "epoch": 0.6832503295032918,
      "grad_norm": 0.08589209616184235,
      "learning_rate": 8.32453058087451e-06,
      "loss": 0.0023,
      "step": 106530
    },
    {
      "epoch": 0.6833144663970779,
      "grad_norm": 0.030689751729369164,
      "learning_rate": 8.324112502515986e-06,
      "loss": 0.0017,
      "step": 106540
    },
    {
      "epoch": 0.683378603290864,
      "grad_norm": 0.23931913077831268,
      "learning_rate": 8.323694382503958e-06,
      "loss": 0.0032,
      "step": 106550
    },
    {
      "epoch": 0.6834427401846501,
      "grad_norm": 0.11819394677877426,
      "learning_rate": 8.323276220843661e-06,
      "loss": 0.0017,
      "step": 106560
    },
    {
      "epoch": 0.6835068770784362,
      "grad_norm": 0.15443769097328186,
      "learning_rate": 8.322858017540341e-06,
      "loss": 0.0036,
      "step": 106570
    },
    {
      "epoch": 0.6835710139722223,
      "grad_norm": 0.02909116819500923,
      "learning_rate": 8.322439772599233e-06,
      "loss": 0.003,
      "step": 106580
    },
    {
      "epoch": 0.6836351508660085,
      "grad_norm": 0.1778852790594101,
      "learning_rate": 8.32202148602558e-06,
      "loss": 0.0023,
      "step": 106590
    },
    {
      "epoch": 0.6836992877597945,
      "grad_norm": 0.06952908635139465,
      "learning_rate": 8.321603157824625e-06,
      "loss": 0.0035,
      "step": 106600
    },
    {
      "epoch": 0.6837634246535806,
      "grad_norm": 0.069687619805336,
      "learning_rate": 8.321184788001609e-06,
      "loss": 0.0059,
      "step": 106610
    },
    {
      "epoch": 0.6838275615473667,
      "grad_norm": 0.07610204070806503,
      "learning_rate": 8.32076637656177e-06,
      "loss": 0.002,
      "step": 106620
    },
    {
      "epoch": 0.6838916984411528,
      "grad_norm": 0.1031593456864357,
      "learning_rate": 8.320347923510359e-06,
      "loss": 0.0035,
      "step": 106630
    },
    {
      "epoch": 0.6839558353349389,
      "grad_norm": 0.14455530047416687,
      "learning_rate": 8.319929428852613e-06,
      "loss": 0.0034,
      "step": 106640
    },
    {
      "epoch": 0.684019972228725,
      "grad_norm": 0.11688445508480072,
      "learning_rate": 8.31951089259378e-06,
      "loss": 0.0082,
      "step": 106650
    },
    {
      "epoch": 0.684084109122511,
      "grad_norm": 0.26938173174858093,
      "learning_rate": 8.319092314739103e-06,
      "loss": 0.0043,
      "step": 106660
    },
    {
      "epoch": 0.6841482460162972,
      "grad_norm": 0.08609388768672943,
      "learning_rate": 8.318673695293829e-06,
      "loss": 0.0047,
      "step": 106670
    },
    {
      "epoch": 0.6842123829100832,
      "grad_norm": 0.22282618284225464,
      "learning_rate": 8.3182550342632e-06,
      "loss": 0.0035,
      "step": 106680
    },
    {
      "epoch": 0.6842765198038694,
      "grad_norm": 0.14885510504245758,
      "learning_rate": 8.317836331652463e-06,
      "loss": 0.0018,
      "step": 106690
    },
    {
      "epoch": 0.6843406566976554,
      "grad_norm": 0.12032639980316162,
      "learning_rate": 8.317417587466866e-06,
      "loss": 0.0015,
      "step": 106700
    },
    {
      "epoch": 0.6844047935914416,
      "grad_norm": 0.1598871946334839,
      "learning_rate": 8.316998801711655e-06,
      "loss": 0.0024,
      "step": 106710
    },
    {
      "epoch": 0.6844689304852277,
      "grad_norm": 0.10817835479974747,
      "learning_rate": 8.316579974392081e-06,
      "loss": 0.0048,
      "step": 106720
    },
    {
      "epoch": 0.6845330673790138,
      "grad_norm": 0.16026908159255981,
      "learning_rate": 8.316161105513387e-06,
      "loss": 0.0026,
      "step": 106730
    },
    {
      "epoch": 0.6845972042727999,
      "grad_norm": 0.17550593614578247,
      "learning_rate": 8.315742195080824e-06,
      "loss": 0.0031,
      "step": 106740
    },
    {
      "epoch": 0.684661341166586,
      "grad_norm": 0.08394819498062134,
      "learning_rate": 8.315323243099644e-06,
      "loss": 0.0053,
      "step": 106750
    },
    {
      "epoch": 0.6847254780603721,
      "grad_norm": 0.17244599759578705,
      "learning_rate": 8.314904249575093e-06,
      "loss": 0.0032,
      "step": 106760
    },
    {
      "epoch": 0.6847896149541581,
      "grad_norm": 0.06605623662471771,
      "learning_rate": 8.314485214512424e-06,
      "loss": 0.0024,
      "step": 106770
    },
    {
      "epoch": 0.6848537518479443,
      "grad_norm": 0.2065979242324829,
      "learning_rate": 8.314066137916885e-06,
      "loss": 0.0034,
      "step": 106780
    },
    {
      "epoch": 0.6849178887417303,
      "grad_norm": 0.22780033946037292,
      "learning_rate": 8.31364701979373e-06,
      "loss": 0.0026,
      "step": 106790
    },
    {
      "epoch": 0.6849820256355165,
      "grad_norm": 0.2013179212808609,
      "learning_rate": 8.313227860148208e-06,
      "loss": 0.0021,
      "step": 106800
    },
    {
      "epoch": 0.6850461625293025,
      "grad_norm": 0.08906995505094528,
      "learning_rate": 8.312808658985575e-06,
      "loss": 0.0025,
      "step": 106810
    },
    {
      "epoch": 0.6851102994230887,
      "grad_norm": 0.049765344709157944,
      "learning_rate": 8.31238941631108e-06,
      "loss": 0.0028,
      "step": 106820
    },
    {
      "epoch": 0.6851744363168747,
      "grad_norm": 0.017044998705387115,
      "learning_rate": 8.311970132129978e-06,
      "loss": 0.0031,
      "step": 106830
    },
    {
      "epoch": 0.6852385732106608,
      "grad_norm": 0.23861342668533325,
      "learning_rate": 8.311550806447523e-06,
      "loss": 0.0019,
      "step": 106840
    },
    {
      "epoch": 0.6853027101044469,
      "grad_norm": 0.14938272535800934,
      "learning_rate": 8.311131439268971e-06,
      "loss": 0.0028,
      "step": 106850
    },
    {
      "epoch": 0.685366846998233,
      "grad_norm": 0.11288265883922577,
      "learning_rate": 8.310712030599576e-06,
      "loss": 0.0028,
      "step": 106860
    },
    {
      "epoch": 0.6854309838920192,
      "grad_norm": 0.02465154230594635,
      "learning_rate": 8.310292580444593e-06,
      "loss": 0.002,
      "step": 106870
    },
    {
      "epoch": 0.6854951207858052,
      "grad_norm": 0.05949242785573006,
      "learning_rate": 8.309873088809278e-06,
      "loss": 0.0039,
      "step": 106880
    },
    {
      "epoch": 0.6855592576795914,
      "grad_norm": 0.20034784078598022,
      "learning_rate": 8.309453555698886e-06,
      "loss": 0.0029,
      "step": 106890
    },
    {
      "epoch": 0.6856233945733774,
      "grad_norm": 0.19417355954647064,
      "learning_rate": 8.309033981118678e-06,
      "loss": 0.0026,
      "step": 106900
    },
    {
      "epoch": 0.6856875314671635,
      "grad_norm": 0.022811679169535637,
      "learning_rate": 8.308614365073908e-06,
      "loss": 0.0017,
      "step": 106910
    },
    {
      "epoch": 0.6857516683609496,
      "grad_norm": 0.10058672726154327,
      "learning_rate": 8.308194707569836e-06,
      "loss": 0.0034,
      "step": 106920
    },
    {
      "epoch": 0.6858158052547357,
      "grad_norm": 0.1317497044801712,
      "learning_rate": 8.30777500861172e-06,
      "loss": 0.0016,
      "step": 106930
    },
    {
      "epoch": 0.6858799421485218,
      "grad_norm": 0.1980799287557602,
      "learning_rate": 8.307355268204817e-06,
      "loss": 0.0018,
      "step": 106940
    },
    {
      "epoch": 0.6859440790423079,
      "grad_norm": 0.1505780667066574,
      "learning_rate": 8.306935486354391e-06,
      "loss": 0.0036,
      "step": 106950
    },
    {
      "epoch": 0.686008215936094,
      "grad_norm": 0.17634980380535126,
      "learning_rate": 8.306515663065699e-06,
      "loss": 0.0032,
      "step": 106960
    },
    {
      "epoch": 0.6860723528298801,
      "grad_norm": 0.08681733906269073,
      "learning_rate": 8.306095798344004e-06,
      "loss": 0.0036,
      "step": 106970
    },
    {
      "epoch": 0.6861364897236661,
      "grad_norm": 0.06697957962751389,
      "learning_rate": 8.305675892194564e-06,
      "loss": 0.0028,
      "step": 106980
    },
    {
      "epoch": 0.6862006266174523,
      "grad_norm": 0.038370467722415924,
      "learning_rate": 8.305255944622644e-06,
      "loss": 0.0031,
      "step": 106990
    },
    {
      "epoch": 0.6862647635112384,
      "grad_norm": 0.12728312611579895,
      "learning_rate": 8.304835955633504e-06,
      "loss": 0.0024,
      "step": 107000
    },
    {
      "epoch": 0.6863289004050245,
      "grad_norm": 0.25424623489379883,
      "learning_rate": 8.304415925232407e-06,
      "loss": 0.0026,
      "step": 107010
    },
    {
      "epoch": 0.6863930372988106,
      "grad_norm": 0.37205880880355835,
      "learning_rate": 8.303995853424617e-06,
      "loss": 0.0023,
      "step": 107020
    },
    {
      "epoch": 0.6864571741925967,
      "grad_norm": 0.07315251231193542,
      "learning_rate": 8.303575740215399e-06,
      "loss": 0.0018,
      "step": 107030
    },
    {
      "epoch": 0.6865213110863828,
      "grad_norm": 0.08466996252536774,
      "learning_rate": 8.303155585610017e-06,
      "loss": 0.0035,
      "step": 107040
    },
    {
      "epoch": 0.6865854479801689,
      "grad_norm": 0.08715147525072098,
      "learning_rate": 8.302735389613733e-06,
      "loss": 0.004,
      "step": 107050
    },
    {
      "epoch": 0.686649584873955,
      "grad_norm": 0.21168333292007446,
      "learning_rate": 8.302315152231812e-06,
      "loss": 0.0026,
      "step": 107060
    },
    {
      "epoch": 0.686713721767741,
      "grad_norm": 0.3521987199783325,
      "learning_rate": 8.301894873469524e-06,
      "loss": 0.0043,
      "step": 107070
    },
    {
      "epoch": 0.6867778586615272,
      "grad_norm": 0.06002356857061386,
      "learning_rate": 8.301474553332134e-06,
      "loss": 0.0036,
      "step": 107080
    },
    {
      "epoch": 0.6868419955553132,
      "grad_norm": 0.22865359485149384,
      "learning_rate": 8.301054191824907e-06,
      "loss": 0.0031,
      "step": 107090
    },
    {
      "epoch": 0.6869061324490994,
      "grad_norm": 0.08421870321035385,
      "learning_rate": 8.300633788953113e-06,
      "loss": 0.0034,
      "step": 107100
    },
    {
      "epoch": 0.6869702693428854,
      "grad_norm": 0.27708613872528076,
      "learning_rate": 8.30021334472202e-06,
      "loss": 0.0052,
      "step": 107110
    },
    {
      "epoch": 0.6870344062366716,
      "grad_norm": 0.06739004701375961,
      "learning_rate": 8.299792859136892e-06,
      "loss": 0.0032,
      "step": 107120
    },
    {
      "epoch": 0.6870985431304576,
      "grad_norm": 0.03162672743201256,
      "learning_rate": 8.299372332203002e-06,
      "loss": 0.0026,
      "step": 107130
    },
    {
      "epoch": 0.6871626800242437,
      "grad_norm": 0.11178423464298248,
      "learning_rate": 8.29895176392562e-06,
      "loss": 0.0044,
      "step": 107140
    },
    {
      "epoch": 0.6872268169180299,
      "grad_norm": 0.04783308506011963,
      "learning_rate": 8.298531154310013e-06,
      "loss": 0.002,
      "step": 107150
    },
    {
      "epoch": 0.6872909538118159,
      "grad_norm": 0.08391523361206055,
      "learning_rate": 8.298110503361453e-06,
      "loss": 0.0035,
      "step": 107160
    },
    {
      "epoch": 0.6873550907056021,
      "grad_norm": 0.066973976790905,
      "learning_rate": 8.297689811085215e-06,
      "loss": 0.0023,
      "step": 107170
    },
    {
      "epoch": 0.6874192275993881,
      "grad_norm": 0.47545182704925537,
      "learning_rate": 8.297269077486563e-06,
      "loss": 0.0068,
      "step": 107180
    },
    {
      "epoch": 0.6874833644931743,
      "grad_norm": 0.22450487315654755,
      "learning_rate": 8.296848302570775e-06,
      "loss": 0.0024,
      "step": 107190
    },
    {
      "epoch": 0.6875475013869603,
      "grad_norm": 0.25167715549468994,
      "learning_rate": 8.29642748634312e-06,
      "loss": 0.0035,
      "step": 107200
    },
    {
      "epoch": 0.6876116382807465,
      "grad_norm": 0.32105982303619385,
      "learning_rate": 8.296006628808872e-06,
      "loss": 0.0038,
      "step": 107210
    },
    {
      "epoch": 0.6876757751745325,
      "grad_norm": 0.2346927523612976,
      "learning_rate": 8.295585729973307e-06,
      "loss": 0.0024,
      "step": 107220
    },
    {
      "epoch": 0.6877399120683186,
      "grad_norm": 0.06694585829973221,
      "learning_rate": 8.295164789841699e-06,
      "loss": 0.0037,
      "step": 107230
    },
    {
      "epoch": 0.6878040489621047,
      "grad_norm": 0.4164973199367523,
      "learning_rate": 8.294743808419319e-06,
      "loss": 0.0042,
      "step": 107240
    },
    {
      "epoch": 0.6878681858558908,
      "grad_norm": 0.14734579622745514,
      "learning_rate": 8.294322785711444e-06,
      "loss": 0.003,
      "step": 107250
    },
    {
      "epoch": 0.6879323227496769,
      "grad_norm": 0.33766835927963257,
      "learning_rate": 8.29390172172335e-06,
      "loss": 0.0033,
      "step": 107260
    },
    {
      "epoch": 0.687996459643463,
      "grad_norm": 0.16679903864860535,
      "learning_rate": 8.293480616460314e-06,
      "loss": 0.0027,
      "step": 107270
    },
    {
      "epoch": 0.6880605965372492,
      "grad_norm": 0.10429362207651138,
      "learning_rate": 8.293059469927614e-06,
      "loss": 0.0043,
      "step": 107280
    },
    {
      "epoch": 0.6881247334310352,
      "grad_norm": 0.1312631070613861,
      "learning_rate": 8.292638282130523e-06,
      "loss": 0.0021,
      "step": 107290
    },
    {
      "epoch": 0.6881888703248213,
      "grad_norm": 0.198140487074852,
      "learning_rate": 8.29221705307432e-06,
      "loss": 0.0028,
      "step": 107300
    },
    {
      "epoch": 0.6882530072186074,
      "grad_norm": 0.1893017292022705,
      "learning_rate": 8.291795782764287e-06,
      "loss": 0.0018,
      "step": 107310
    },
    {
      "epoch": 0.6883171441123935,
      "grad_norm": 0.17619863152503967,
      "learning_rate": 8.2913744712057e-06,
      "loss": 0.002,
      "step": 107320
    },
    {
      "epoch": 0.6883812810061796,
      "grad_norm": 0.15994544327259064,
      "learning_rate": 8.290953118403838e-06,
      "loss": 0.0014,
      "step": 107330
    },
    {
      "epoch": 0.6884454178999657,
      "grad_norm": 0.13528279960155487,
      "learning_rate": 8.29053172436398e-06,
      "loss": 0.0014,
      "step": 107340
    },
    {
      "epoch": 0.6885095547937518,
      "grad_norm": 0.15273387730121613,
      "learning_rate": 8.29011028909141e-06,
      "loss": 0.0033,
      "step": 107350
    },
    {
      "epoch": 0.6885736916875379,
      "grad_norm": 0.11530592292547226,
      "learning_rate": 8.289688812591405e-06,
      "loss": 0.0024,
      "step": 107360
    },
    {
      "epoch": 0.6886378285813239,
      "grad_norm": 0.02847360447049141,
      "learning_rate": 8.289267294869249e-06,
      "loss": 0.0023,
      "step": 107370
    },
    {
      "epoch": 0.6887019654751101,
      "grad_norm": 0.0729435458779335,
      "learning_rate": 8.288845735930221e-06,
      "loss": 0.0026,
      "step": 107380
    },
    {
      "epoch": 0.6887661023688961,
      "grad_norm": 0.10790400207042694,
      "learning_rate": 8.28842413577961e-06,
      "loss": 0.003,
      "step": 107390
    },
    {
      "epoch": 0.6888302392626823,
      "grad_norm": 0.07635890692472458,
      "learning_rate": 8.28800249442269e-06,
      "loss": 0.0043,
      "step": 107400
    },
    {
      "epoch": 0.6888943761564683,
      "grad_norm": 0.12413554638624191,
      "learning_rate": 8.287580811864751e-06,
      "loss": 0.0019,
      "step": 107410
    },
    {
      "epoch": 0.6889585130502545,
      "grad_norm": 0.2052922397851944,
      "learning_rate": 8.287159088111075e-06,
      "loss": 0.0026,
      "step": 107420
    },
    {
      "epoch": 0.6890226499440406,
      "grad_norm": 0.23809462785720825,
      "learning_rate": 8.286737323166945e-06,
      "loss": 0.0024,
      "step": 107430
    },
    {
      "epoch": 0.6890867868378266,
      "grad_norm": 0.11305812001228333,
      "learning_rate": 8.286315517037646e-06,
      "loss": 0.0051,
      "step": 107440
    },
    {
      "epoch": 0.6891509237316128,
      "grad_norm": 0.03867397457361221,
      "learning_rate": 8.285893669728467e-06,
      "loss": 0.0038,
      "step": 107450
    },
    {
      "epoch": 0.6892150606253988,
      "grad_norm": 0.17843380570411682,
      "learning_rate": 8.285471781244692e-06,
      "loss": 0.0042,
      "step": 107460
    },
    {
      "epoch": 0.689279197519185,
      "grad_norm": 0.31200113892555237,
      "learning_rate": 8.285049851591606e-06,
      "loss": 0.0057,
      "step": 107470
    },
    {
      "epoch": 0.689343334412971,
      "grad_norm": 0.15629765391349792,
      "learning_rate": 8.2846278807745e-06,
      "loss": 0.0023,
      "step": 107480
    },
    {
      "epoch": 0.6894074713067572,
      "grad_norm": 0.12026771157979965,
      "learning_rate": 8.284205868798658e-06,
      "loss": 0.0025,
      "step": 107490
    },
    {
      "epoch": 0.6894716082005432,
      "grad_norm": 0.0701836422085762,
      "learning_rate": 8.283783815669369e-06,
      "loss": 0.0024,
      "step": 107500
    },
    {
      "epoch": 0.6895357450943294,
      "grad_norm": 0.3383144438266754,
      "learning_rate": 8.283361721391923e-06,
      "loss": 0.004,
      "step": 107510
    },
    {
      "epoch": 0.6895998819881154,
      "grad_norm": 0.15997891128063202,
      "learning_rate": 8.282939585971606e-06,
      "loss": 0.0028,
      "step": 107520
    },
    {
      "epoch": 0.6896640188819015,
      "grad_norm": 0.1713678538799286,
      "learning_rate": 8.282517409413711e-06,
      "loss": 0.0028,
      "step": 107530
    },
    {
      "epoch": 0.6897281557756876,
      "grad_norm": 0.21673579514026642,
      "learning_rate": 8.282095191723527e-06,
      "loss": 0.002,
      "step": 107540
    },
    {
      "epoch": 0.6897922926694737,
      "grad_norm": 0.17311428487300873,
      "learning_rate": 8.281672932906344e-06,
      "loss": 0.0037,
      "step": 107550
    },
    {
      "epoch": 0.6898564295632599,
      "grad_norm": 0.24376513063907623,
      "learning_rate": 8.281250632967455e-06,
      "loss": 0.0034,
      "step": 107560
    },
    {
      "epoch": 0.6899205664570459,
      "grad_norm": 0.0667029544711113,
      "learning_rate": 8.280828291912148e-06,
      "loss": 0.002,
      "step": 107570
    },
    {
      "epoch": 0.6899847033508321,
      "grad_norm": 0.1129802018404007,
      "learning_rate": 8.28040590974572e-06,
      "loss": 0.0042,
      "step": 107580
    },
    {
      "epoch": 0.6900488402446181,
      "grad_norm": 0.15882444381713867,
      "learning_rate": 8.27998348647346e-06,
      "loss": 0.0031,
      "step": 107590
    },
    {
      "epoch": 0.6901129771384042,
      "grad_norm": 0.03456937521696091,
      "learning_rate": 8.279561022100665e-06,
      "loss": 0.0013,
      "step": 107600
    },
    {
      "epoch": 0.6901771140321903,
      "grad_norm": 0.1546746790409088,
      "learning_rate": 8.279138516632624e-06,
      "loss": 0.007,
      "step": 107610
    },
    {
      "epoch": 0.6902412509259764,
      "grad_norm": 0.04763857275247574,
      "learning_rate": 8.278715970074636e-06,
      "loss": 0.0027,
      "step": 107620
    },
    {
      "epoch": 0.6903053878197625,
      "grad_norm": 0.13871750235557556,
      "learning_rate": 8.278293382431992e-06,
      "loss": 0.003,
      "step": 107630
    },
    {
      "epoch": 0.6903695247135486,
      "grad_norm": 0.07775358855724335,
      "learning_rate": 8.277870753709988e-06,
      "loss": 0.002,
      "step": 107640
    },
    {
      "epoch": 0.6904336616073347,
      "grad_norm": 0.13904011249542236,
      "learning_rate": 8.277448083913924e-06,
      "loss": 0.0038,
      "step": 107650
    },
    {
      "epoch": 0.6904977985011208,
      "grad_norm": 0.06254595518112183,
      "learning_rate": 8.27702537304909e-06,
      "loss": 0.0023,
      "step": 107660
    },
    {
      "epoch": 0.6905619353949068,
      "grad_norm": 0.1485593467950821,
      "learning_rate": 8.276602621120788e-06,
      "loss": 0.0052,
      "step": 107670
    },
    {
      "epoch": 0.690626072288693,
      "grad_norm": 0.1240374967455864,
      "learning_rate": 8.276179828134314e-06,
      "loss": 0.0042,
      "step": 107680
    },
    {
      "epoch": 0.690690209182479,
      "grad_norm": 0.1361633539199829,
      "learning_rate": 8.275756994094963e-06,
      "loss": 0.0025,
      "step": 107690
    },
    {
      "epoch": 0.6907543460762652,
      "grad_norm": 0.0703364908695221,
      "learning_rate": 8.275334119008037e-06,
      "loss": 0.0037,
      "step": 107700
    },
    {
      "epoch": 0.6908184829700513,
      "grad_norm": 0.09072870761156082,
      "learning_rate": 8.274911202878834e-06,
      "loss": 0.0011,
      "step": 107710
    },
    {
      "epoch": 0.6908826198638374,
      "grad_norm": 0.043299779295921326,
      "learning_rate": 8.274488245712653e-06,
      "loss": 0.0021,
      "step": 107720
    },
    {
      "epoch": 0.6909467567576235,
      "grad_norm": 0.08427557349205017,
      "learning_rate": 8.274065247514793e-06,
      "loss": 0.0062,
      "step": 107730
    },
    {
      "epoch": 0.6910108936514096,
      "grad_norm": 0.13290773332118988,
      "learning_rate": 8.273642208290555e-06,
      "loss": 0.0023,
      "step": 107740
    },
    {
      "epoch": 0.6910750305451957,
      "grad_norm": 0.05695051699876785,
      "learning_rate": 8.273219128045241e-06,
      "loss": 0.0045,
      "step": 107750
    },
    {
      "epoch": 0.6911391674389817,
      "grad_norm": 0.068753182888031,
      "learning_rate": 8.272796006784153e-06,
      "loss": 0.0023,
      "step": 107760
    },
    {
      "epoch": 0.6912033043327679,
      "grad_norm": 0.05978182330727577,
      "learning_rate": 8.272372844512593e-06,
      "loss": 0.0052,
      "step": 107770
    },
    {
      "epoch": 0.6912674412265539,
      "grad_norm": 0.17988887429237366,
      "learning_rate": 8.271949641235861e-06,
      "loss": 0.0034,
      "step": 107780
    },
    {
      "epoch": 0.6913315781203401,
      "grad_norm": 0.14107699692249298,
      "learning_rate": 8.271526396959261e-06,
      "loss": 0.0026,
      "step": 107790
    },
    {
      "epoch": 0.6913957150141261,
      "grad_norm": 0.3287390470504761,
      "learning_rate": 8.271103111688098e-06,
      "loss": 0.0021,
      "step": 107800
    },
    {
      "epoch": 0.6914598519079123,
      "grad_norm": 0.5077758431434631,
      "learning_rate": 8.270679785427675e-06,
      "loss": 0.0041,
      "step": 107810
    },
    {
      "epoch": 0.6915239888016983,
      "grad_norm": 0.3003099858760834,
      "learning_rate": 8.270256418183297e-06,
      "loss": 0.004,
      "step": 107820
    },
    {
      "epoch": 0.6915881256954844,
      "grad_norm": 0.10210637748241425,
      "learning_rate": 8.26983300996027e-06,
      "loss": 0.0068,
      "step": 107830
    },
    {
      "epoch": 0.6916522625892706,
      "grad_norm": 0.11000536382198334,
      "learning_rate": 8.269409560763896e-06,
      "loss": 0.0032,
      "step": 107840
    },
    {
      "epoch": 0.6917163994830566,
      "grad_norm": 0.12499922513961792,
      "learning_rate": 8.268986070599485e-06,
      "loss": 0.0023,
      "step": 107850
    },
    {
      "epoch": 0.6917805363768428,
      "grad_norm": 0.19037102162837982,
      "learning_rate": 8.268562539472342e-06,
      "loss": 0.0043,
      "step": 107860
    },
    {
      "epoch": 0.6918446732706288,
      "grad_norm": 0.06971777230501175,
      "learning_rate": 8.268138967387773e-06,
      "loss": 0.0024,
      "step": 107870
    },
    {
      "epoch": 0.691908810164415,
      "grad_norm": 0.07947616279125214,
      "learning_rate": 8.267715354351088e-06,
      "loss": 0.0041,
      "step": 107880
    },
    {
      "epoch": 0.691972947058201,
      "grad_norm": 0.18607938289642334,
      "learning_rate": 8.267291700367595e-06,
      "loss": 0.0031,
      "step": 107890
    },
    {
      "epoch": 0.6920370839519872,
      "grad_norm": 0.12689931690692902,
      "learning_rate": 8.266868005442603e-06,
      "loss": 0.0025,
      "step": 107900
    },
    {
      "epoch": 0.6921012208457732,
      "grad_norm": 0.09177198261022568,
      "learning_rate": 8.266444269581417e-06,
      "loss": 0.0024,
      "step": 107910
    },
    {
      "epoch": 0.6921653577395593,
      "grad_norm": 0.13121455907821655,
      "learning_rate": 8.266020492789352e-06,
      "loss": 0.0048,
      "step": 107920
    },
    {
      "epoch": 0.6922294946333454,
      "grad_norm": 0.35130009055137634,
      "learning_rate": 8.265596675071715e-06,
      "loss": 0.0028,
      "step": 107930
    },
    {
      "epoch": 0.6922936315271315,
      "grad_norm": 0.18090005218982697,
      "learning_rate": 8.265172816433818e-06,
      "loss": 0.0042,
      "step": 107940
    },
    {
      "epoch": 0.6923577684209176,
      "grad_norm": 0.14326050877571106,
      "learning_rate": 8.264748916880973e-06,
      "loss": 0.0028,
      "step": 107950
    },
    {
      "epoch": 0.6924219053147037,
      "grad_norm": 0.09501796215772629,
      "learning_rate": 8.264324976418489e-06,
      "loss": 0.0027,
      "step": 107960
    },
    {
      "epoch": 0.6924860422084897,
      "grad_norm": 0.1570703536272049,
      "learning_rate": 8.263900995051681e-06,
      "loss": 0.0029,
      "step": 107970
    },
    {
      "epoch": 0.6925501791022759,
      "grad_norm": 0.03497795760631561,
      "learning_rate": 8.263476972785862e-06,
      "loss": 0.0025,
      "step": 107980
    },
    {
      "epoch": 0.692614315996062,
      "grad_norm": 0.040670089423656464,
      "learning_rate": 8.263052909626343e-06,
      "loss": 0.0023,
      "step": 107990
    },
    {
      "epoch": 0.6926784528898481,
      "grad_norm": 0.14497928321361542,
      "learning_rate": 8.26262880557844e-06,
      "loss": 0.0028,
      "step": 108000
    },
    {
      "epoch": 0.6927425897836342,
      "grad_norm": 0.17702153325080872,
      "learning_rate": 8.262204660647463e-06,
      "loss": 0.0022,
      "step": 108010
    },
    {
      "epoch": 0.6928067266774203,
      "grad_norm": 0.07231775671243668,
      "learning_rate": 8.261780474838735e-06,
      "loss": 0.0027,
      "step": 108020
    },
    {
      "epoch": 0.6928708635712064,
      "grad_norm": 0.12166418880224228,
      "learning_rate": 8.261356248157563e-06,
      "loss": 0.0025,
      "step": 108030
    },
    {
      "epoch": 0.6929350004649925,
      "grad_norm": 0.017073825001716614,
      "learning_rate": 8.260931980609268e-06,
      "loss": 0.0024,
      "step": 108040
    },
    {
      "epoch": 0.6929991373587786,
      "grad_norm": 0.023968270048499107,
      "learning_rate": 8.260507672199163e-06,
      "loss": 0.0025,
      "step": 108050
    },
    {
      "epoch": 0.6930632742525646,
      "grad_norm": 0.17493650317192078,
      "learning_rate": 8.260083322932569e-06,
      "loss": 0.004,
      "step": 108060
    },
    {
      "epoch": 0.6931274111463508,
      "grad_norm": 0.3527976870536804,
      "learning_rate": 8.259658932814798e-06,
      "loss": 0.0029,
      "step": 108070
    },
    {
      "epoch": 0.6931915480401368,
      "grad_norm": 0.28017330169677734,
      "learning_rate": 8.259234501851173e-06,
      "loss": 0.0025,
      "step": 108080
    },
    {
      "epoch": 0.693255684933923,
      "grad_norm": 0.14413754642009735,
      "learning_rate": 8.25881003004701e-06,
      "loss": 0.0022,
      "step": 108090
    },
    {
      "epoch": 0.693319821827709,
      "grad_norm": 0.21677450835704803,
      "learning_rate": 8.258385517407627e-06,
      "loss": 0.0022,
      "step": 108100
    },
    {
      "epoch": 0.6933839587214952,
      "grad_norm": 0.11794870346784592,
      "learning_rate": 8.257960963938348e-06,
      "loss": 0.0042,
      "step": 108110
    },
    {
      "epoch": 0.6934480956152812,
      "grad_norm": 0.12071501463651657,
      "learning_rate": 8.257536369644487e-06,
      "loss": 0.0026,
      "step": 108120
    },
    {
      "epoch": 0.6935122325090674,
      "grad_norm": 0.12205830961465836,
      "learning_rate": 8.257111734531367e-06,
      "loss": 0.0055,
      "step": 108130
    },
    {
      "epoch": 0.6935763694028535,
      "grad_norm": 0.08712588995695114,
      "learning_rate": 8.256687058604312e-06,
      "loss": 0.0037,
      "step": 108140
    },
    {
      "epoch": 0.6936405062966395,
      "grad_norm": 0.060235004872083664,
      "learning_rate": 8.256262341868637e-06,
      "loss": 0.0034,
      "step": 108150
    },
    {
      "epoch": 0.6937046431904257,
      "grad_norm": 0.12908978760242462,
      "learning_rate": 8.255837584329671e-06,
      "loss": 0.0039,
      "step": 108160
    },
    {
      "epoch": 0.6937687800842117,
      "grad_norm": 0.06922983378171921,
      "learning_rate": 8.255412785992731e-06,
      "loss": 0.0022,
      "step": 108170
    },
    {
      "epoch": 0.6938329169779979,
      "grad_norm": 0.07774290442466736,
      "learning_rate": 8.254987946863142e-06,
      "loss": 0.0042,
      "step": 108180
    },
    {
      "epoch": 0.6938970538717839,
      "grad_norm": 0.06928818672895432,
      "learning_rate": 8.25456306694623e-06,
      "loss": 0.0036,
      "step": 108190
    },
    {
      "epoch": 0.6939611907655701,
      "grad_norm": 0.31480279564857483,
      "learning_rate": 8.254138146247313e-06,
      "loss": 0.0025,
      "step": 108200
    },
    {
      "epoch": 0.6940253276593561,
      "grad_norm": 0.22580046951770782,
      "learning_rate": 8.25371318477172e-06,
      "loss": 0.0026,
      "step": 108210
    },
    {
      "epoch": 0.6940894645531422,
      "grad_norm": 0.25423887372016907,
      "learning_rate": 8.253288182524778e-06,
      "loss": 0.003,
      "step": 108220
    },
    {
      "epoch": 0.6941536014469283,
      "grad_norm": 0.1763753741979599,
      "learning_rate": 8.252863139511809e-06,
      "loss": 0.0027,
      "step": 108230
    },
    {
      "epoch": 0.6942177383407144,
      "grad_norm": 0.09514214098453522,
      "learning_rate": 8.25243805573814e-06,
      "loss": 0.0032,
      "step": 108240
    },
    {
      "epoch": 0.6942818752345005,
      "grad_norm": 0.06112831085920334,
      "learning_rate": 8.252012931209097e-06,
      "loss": 0.0064,
      "step": 108250
    },
    {
      "epoch": 0.6943460121282866,
      "grad_norm": 0.02216607704758644,
      "learning_rate": 8.251587765930009e-06,
      "loss": 0.0022,
      "step": 108260
    },
    {
      "epoch": 0.6944101490220728,
      "grad_norm": 0.09686776995658875,
      "learning_rate": 8.251162559906201e-06,
      "loss": 0.0028,
      "step": 108270
    },
    {
      "epoch": 0.6944742859158588,
      "grad_norm": 0.0727643296122551,
      "learning_rate": 8.250737313143003e-06,
      "loss": 0.0024,
      "step": 108280
    },
    {
      "epoch": 0.694538422809645,
      "grad_norm": 0.17950348556041718,
      "learning_rate": 8.250312025645744e-06,
      "loss": 0.0032,
      "step": 108290
    },
    {
      "epoch": 0.694602559703431,
      "grad_norm": 0.11519443243741989,
      "learning_rate": 8.249886697419753e-06,
      "loss": 0.0026,
      "step": 108300
    },
    {
      "epoch": 0.6946666965972171,
      "grad_norm": 0.1178741529583931,
      "learning_rate": 8.249461328470357e-06,
      "loss": 0.0072,
      "step": 108310
    },
    {
      "epoch": 0.6947308334910032,
      "grad_norm": 0.21187977492809296,
      "learning_rate": 8.24903591880289e-06,
      "loss": 0.0031,
      "step": 108320
    },
    {
      "epoch": 0.6947949703847893,
      "grad_norm": 0.1189272403717041,
      "learning_rate": 8.248610468422679e-06,
      "loss": 0.0026,
      "step": 108330
    },
    {
      "epoch": 0.6948591072785754,
      "grad_norm": 0.2139284908771515,
      "learning_rate": 8.24818497733506e-06,
      "loss": 0.0041,
      "step": 108340
    },
    {
      "epoch": 0.6949232441723615,
      "grad_norm": 0.2531370520591736,
      "learning_rate": 8.247759445545358e-06,
      "loss": 0.002,
      "step": 108350
    },
    {
      "epoch": 0.6949873810661475,
      "grad_norm": 0.11315297335386276,
      "learning_rate": 8.247333873058912e-06,
      "loss": 0.003,
      "step": 108360
    },
    {
      "epoch": 0.6950515179599337,
      "grad_norm": 0.05931203439831734,
      "learning_rate": 8.24690825988105e-06,
      "loss": 0.0025,
      "step": 108370
    },
    {
      "epoch": 0.6951156548537197,
      "grad_norm": 0.08369827270507812,
      "learning_rate": 8.246482606017107e-06,
      "loss": 0.0026,
      "step": 108380
    },
    {
      "epoch": 0.6951797917475059,
      "grad_norm": 0.0766642689704895,
      "learning_rate": 8.246056911472417e-06,
      "loss": 0.0022,
      "step": 108390
    },
    {
      "epoch": 0.6952439286412919,
      "grad_norm": 0.16475103795528412,
      "learning_rate": 8.245631176252316e-06,
      "loss": 0.0022,
      "step": 108400
    },
    {
      "epoch": 0.6953080655350781,
      "grad_norm": 0.14208433032035828,
      "learning_rate": 8.245205400362134e-06,
      "loss": 0.0025,
      "step": 108410
    },
    {
      "epoch": 0.6953722024288642,
      "grad_norm": 0.2255280315876007,
      "learning_rate": 8.24477958380721e-06,
      "loss": 0.0079,
      "step": 108420
    },
    {
      "epoch": 0.6954363393226503,
      "grad_norm": 0.11754395067691803,
      "learning_rate": 8.24435372659288e-06,
      "loss": 0.0023,
      "step": 108430
    },
    {
      "epoch": 0.6955004762164364,
      "grad_norm": 0.12908974289894104,
      "learning_rate": 8.243927828724475e-06,
      "loss": 0.0043,
      "step": 108440
    },
    {
      "epoch": 0.6955646131102224,
      "grad_norm": 0.08824348449707031,
      "learning_rate": 8.24350189020734e-06,
      "loss": 0.0027,
      "step": 108450
    },
    {
      "epoch": 0.6956287500040086,
      "grad_norm": 0.22994863986968994,
      "learning_rate": 8.243075911046807e-06,
      "loss": 0.0025,
      "step": 108460
    },
    {
      "epoch": 0.6956928868977946,
      "grad_norm": 0.17440223693847656,
      "learning_rate": 8.242649891248215e-06,
      "loss": 0.0026,
      "step": 108470
    },
    {
      "epoch": 0.6957570237915808,
      "grad_norm": 0.17761161923408508,
      "learning_rate": 8.242223830816905e-06,
      "loss": 0.0029,
      "step": 108480
    },
    {
      "epoch": 0.6958211606853668,
      "grad_norm": 0.21196191012859344,
      "learning_rate": 8.24179772975821e-06,
      "loss": 0.0023,
      "step": 108490
    },
    {
      "epoch": 0.695885297579153,
      "grad_norm": 0.15021458268165588,
      "learning_rate": 8.241371588077473e-06,
      "loss": 0.002,
      "step": 108500
    },
    {
      "epoch": 0.695949434472939,
      "grad_norm": 0.058959461748600006,
      "learning_rate": 8.240945405780035e-06,
      "loss": 0.003,
      "step": 108510
    },
    {
      "epoch": 0.6960135713667251,
      "grad_norm": 0.22570465505123138,
      "learning_rate": 8.240519182871232e-06,
      "loss": 0.0042,
      "step": 108520
    },
    {
      "epoch": 0.6960777082605112,
      "grad_norm": 0.22025471925735474,
      "learning_rate": 8.240092919356409e-06,
      "loss": 0.0043,
      "step": 108530
    },
    {
      "epoch": 0.6961418451542973,
      "grad_norm": 0.19205790758132935,
      "learning_rate": 8.239666615240906e-06,
      "loss": 0.0023,
      "step": 108540
    },
    {
      "epoch": 0.6962059820480835,
      "grad_norm": 0.16703635454177856,
      "learning_rate": 8.239240270530065e-06,
      "loss": 0.0025,
      "step": 108550
    },
    {
      "epoch": 0.6962701189418695,
      "grad_norm": 0.1193094551563263,
      "learning_rate": 8.23881388522923e-06,
      "loss": 0.0019,
      "step": 108560
    },
    {
      "epoch": 0.6963342558356557,
      "grad_norm": 0.052605997771024704,
      "learning_rate": 8.23838745934374e-06,
      "loss": 0.0033,
      "step": 108570
    },
    {
      "epoch": 0.6963983927294417,
      "grad_norm": 0.046040553599596024,
      "learning_rate": 8.237960992878941e-06,
      "loss": 0.002,
      "step": 108580
    },
    {
      "epoch": 0.6964625296232279,
      "grad_norm": 0.06242847442626953,
      "learning_rate": 8.237534485840177e-06,
      "loss": 0.0022,
      "step": 108590
    },
    {
      "epoch": 0.6965266665170139,
      "grad_norm": 0.15265408158302307,
      "learning_rate": 8.237107938232792e-06,
      "loss": 0.005,
      "step": 108600
    },
    {
      "epoch": 0.6965908034108,
      "grad_norm": 0.2434060275554657,
      "learning_rate": 8.23668135006213e-06,
      "loss": 0.0047,
      "step": 108610
    },
    {
      "epoch": 0.6966549403045861,
      "grad_norm": 0.1061597615480423,
      "learning_rate": 8.236254721333541e-06,
      "loss": 0.0028,
      "step": 108620
    },
    {
      "epoch": 0.6967190771983722,
      "grad_norm": 0.2710915803909302,
      "learning_rate": 8.235828052052363e-06,
      "loss": 0.0022,
      "step": 108630
    },
    {
      "epoch": 0.6967832140921583,
      "grad_norm": 0.22070899605751038,
      "learning_rate": 8.23540134222395e-06,
      "loss": 0.002,
      "step": 108640
    },
    {
      "epoch": 0.6968473509859444,
      "grad_norm": 0.022739490494132042,
      "learning_rate": 8.234974591853646e-06,
      "loss": 0.0026,
      "step": 108650
    },
    {
      "epoch": 0.6969114878797305,
      "grad_norm": 0.10844630002975464,
      "learning_rate": 8.234547800946797e-06,
      "loss": 0.004,
      "step": 108660
    },
    {
      "epoch": 0.6969756247735166,
      "grad_norm": 0.29461532831192017,
      "learning_rate": 8.234120969508752e-06,
      "loss": 0.0029,
      "step": 108670
    },
    {
      "epoch": 0.6970397616673026,
      "grad_norm": 0.5747034549713135,
      "learning_rate": 8.23369409754486e-06,
      "loss": 0.0033,
      "step": 108680
    },
    {
      "epoch": 0.6971038985610888,
      "grad_norm": 0.19798482954502106,
      "learning_rate": 8.233267185060473e-06,
      "loss": 0.0022,
      "step": 108690
    },
    {
      "epoch": 0.6971680354548749,
      "grad_norm": 0.1314244568347931,
      "learning_rate": 8.232840232060932e-06,
      "loss": 0.0014,
      "step": 108700
    },
    {
      "epoch": 0.697232172348661,
      "grad_norm": 0.11207201331853867,
      "learning_rate": 8.232413238551595e-06,
      "loss": 0.0034,
      "step": 108710
    },
    {
      "epoch": 0.6972963092424471,
      "grad_norm": 0.10562792420387268,
      "learning_rate": 8.231986204537811e-06,
      "loss": 0.0015,
      "step": 108720
    },
    {
      "epoch": 0.6973604461362332,
      "grad_norm": 0.18661898374557495,
      "learning_rate": 8.231559130024928e-06,
      "loss": 0.0031,
      "step": 108730
    },
    {
      "epoch": 0.6974245830300193,
      "grad_norm": 0.3229295313358307,
      "learning_rate": 8.231132015018301e-06,
      "loss": 0.0038,
      "step": 108740
    },
    {
      "epoch": 0.6974887199238053,
      "grad_norm": 0.05444495379924774,
      "learning_rate": 8.230704859523281e-06,
      "loss": 0.0018,
      "step": 108750
    },
    {
      "epoch": 0.6975528568175915,
      "grad_norm": 0.22647680342197418,
      "learning_rate": 8.230277663545218e-06,
      "loss": 0.005,
      "step": 108760
    },
    {
      "epoch": 0.6976169937113775,
      "grad_norm": 0.09481486678123474,
      "learning_rate": 8.22985042708947e-06,
      "loss": 0.0023,
      "step": 108770
    },
    {
      "epoch": 0.6976811306051637,
      "grad_norm": 0.06593775004148483,
      "learning_rate": 8.229423150161387e-06,
      "loss": 0.0045,
      "step": 108780
    },
    {
      "epoch": 0.6977452674989497,
      "grad_norm": 0.19967499375343323,
      "learning_rate": 8.228995832766322e-06,
      "loss": 0.0018,
      "step": 108790
    },
    {
      "epoch": 0.6978094043927359,
      "grad_norm": 0.11123993247747421,
      "learning_rate": 8.228568474909632e-06,
      "loss": 0.0029,
      "step": 108800
    },
    {
      "epoch": 0.6978735412865219,
      "grad_norm": 0.039687998592853546,
      "learning_rate": 8.228141076596673e-06,
      "loss": 0.0018,
      "step": 108810
    },
    {
      "epoch": 0.697937678180308,
      "grad_norm": 0.08039312809705734,
      "learning_rate": 8.227713637832799e-06,
      "loss": 0.0025,
      "step": 108820
    },
    {
      "epoch": 0.6980018150740942,
      "grad_norm": 0.1507989466190338,
      "learning_rate": 8.227286158623367e-06,
      "loss": 0.0024,
      "step": 108830
    },
    {
      "epoch": 0.6980659519678802,
      "grad_norm": 0.18190570175647736,
      "learning_rate": 8.226858638973731e-06,
      "loss": 0.0015,
      "step": 108840
    },
    {
      "epoch": 0.6981300888616664,
      "grad_norm": 0.08606167882680893,
      "learning_rate": 8.226431078889252e-06,
      "loss": 0.0024,
      "step": 108850
    },
    {
      "epoch": 0.6981942257554524,
      "grad_norm": 0.11085907369852066,
      "learning_rate": 8.226003478375285e-06,
      "loss": 0.0029,
      "step": 108860
    },
    {
      "epoch": 0.6982583626492386,
      "grad_norm": 0.02125435322523117,
      "learning_rate": 8.225575837437187e-06,
      "loss": 0.003,
      "step": 108870
    },
    {
      "epoch": 0.6983224995430246,
      "grad_norm": 0.1312132328748703,
      "learning_rate": 8.22514815608032e-06,
      "loss": 0.0034,
      "step": 108880
    },
    {
      "epoch": 0.6983866364368108,
      "grad_norm": 0.18251964449882507,
      "learning_rate": 8.224720434310042e-06,
      "loss": 0.0018,
      "step": 108890
    },
    {
      "epoch": 0.6984507733305968,
      "grad_norm": 0.09210048615932465,
      "learning_rate": 8.224292672131711e-06,
      "loss": 0.0034,
      "step": 108900
    },
    {
      "epoch": 0.698514910224383,
      "grad_norm": 0.11854143440723419,
      "learning_rate": 8.223864869550689e-06,
      "loss": 0.0021,
      "step": 108910
    },
    {
      "epoch": 0.698579047118169,
      "grad_norm": 0.1463458389043808,
      "learning_rate": 8.223437026572336e-06,
      "loss": 0.004,
      "step": 108920
    },
    {
      "epoch": 0.6986431840119551,
      "grad_norm": 0.28385433554649353,
      "learning_rate": 8.223009143202015e-06,
      "loss": 0.0028,
      "step": 108930
    },
    {
      "epoch": 0.6987073209057412,
      "grad_norm": 0.1049434095621109,
      "learning_rate": 8.222581219445084e-06,
      "loss": 0.0028,
      "step": 108940
    },
    {
      "epoch": 0.6987714577995273,
      "grad_norm": 0.3270774185657501,
      "learning_rate": 8.222153255306908e-06,
      "loss": 0.0051,
      "step": 108950
    },
    {
      "epoch": 0.6988355946933134,
      "grad_norm": 0.048887260258197784,
      "learning_rate": 8.221725250792849e-06,
      "loss": 0.0019,
      "step": 108960
    },
    {
      "epoch": 0.6988997315870995,
      "grad_norm": 0.1261194795370102,
      "learning_rate": 8.221297205908269e-06,
      "loss": 0.0057,
      "step": 108970
    },
    {
      "epoch": 0.6989638684808857,
      "grad_norm": 0.05719394236803055,
      "learning_rate": 8.220869120658534e-06,
      "loss": 0.0018,
      "step": 108980
    },
    {
      "epoch": 0.6990280053746717,
      "grad_norm": 0.09047127515077591,
      "learning_rate": 8.220440995049007e-06,
      "loss": 0.0035,
      "step": 108990
    },
    {
      "epoch": 0.6990921422684578,
      "grad_norm": 0.058118436485528946,
      "learning_rate": 8.220012829085051e-06,
      "loss": 0.0022,
      "step": 109000
    },
    {
      "epoch": 0.6991562791622439,
      "grad_norm": 0.09301625192165375,
      "learning_rate": 8.219584622772034e-06,
      "loss": 0.003,
      "step": 109010
    },
    {
      "epoch": 0.69922041605603,
      "grad_norm": 0.03872682526707649,
      "learning_rate": 8.219156376115321e-06,
      "loss": 0.0035,
      "step": 109020
    },
    {
      "epoch": 0.6992845529498161,
      "grad_norm": 0.12496228516101837,
      "learning_rate": 8.218728089120278e-06,
      "loss": 0.0025,
      "step": 109030
    },
    {
      "epoch": 0.6993486898436022,
      "grad_norm": 0.2176547795534134,
      "learning_rate": 8.218299761792272e-06,
      "loss": 0.004,
      "step": 109040
    },
    {
      "epoch": 0.6994128267373882,
      "grad_norm": 0.06557410210371017,
      "learning_rate": 8.217871394136669e-06,
      "loss": 0.0033,
      "step": 109050
    },
    {
      "epoch": 0.6994769636311744,
      "grad_norm": 0.23859727382659912,
      "learning_rate": 8.217442986158837e-06,
      "loss": 0.0024,
      "step": 109060
    },
    {
      "epoch": 0.6995411005249604,
      "grad_norm": 0.07227912545204163,
      "learning_rate": 8.217014537864147e-06,
      "loss": 0.0033,
      "step": 109070
    },
    {
      "epoch": 0.6996052374187466,
      "grad_norm": 0.11460322886705399,
      "learning_rate": 8.216586049257965e-06,
      "loss": 0.0027,
      "step": 109080
    },
    {
      "epoch": 0.6996693743125326,
      "grad_norm": 0.03642609342932701,
      "learning_rate": 8.21615752034566e-06,
      "loss": 0.0022,
      "step": 109090
    },
    {
      "epoch": 0.6997335112063188,
      "grad_norm": 0.1255207508802414,
      "learning_rate": 8.215728951132603e-06,
      "loss": 0.0038,
      "step": 109100
    },
    {
      "epoch": 0.6997976481001049,
      "grad_norm": 0.15166237950325012,
      "learning_rate": 8.215300341624164e-06,
      "loss": 0.0042,
      "step": 109110
    },
    {
      "epoch": 0.699861784993891,
      "grad_norm": 0.15849529206752777,
      "learning_rate": 8.214871691825714e-06,
      "loss": 0.0027,
      "step": 109120
    },
    {
      "epoch": 0.6999259218876771,
      "grad_norm": 0.17884567379951477,
      "learning_rate": 8.214443001742624e-06,
      "loss": 0.0028,
      "step": 109130
    },
    {
      "epoch": 0.6999900587814631,
      "grad_norm": 0.10719947516918182,
      "learning_rate": 8.214014271380266e-06,
      "loss": 0.0029,
      "step": 109140
    },
    {
      "epoch": 0.7000541956752493,
      "grad_norm": 0.12732060253620148,
      "learning_rate": 8.213585500744012e-06,
      "loss": 0.0049,
      "step": 109150
    },
    {
      "epoch": 0.7001183325690353,
      "grad_norm": 0.03269217908382416,
      "learning_rate": 8.213156689839235e-06,
      "loss": 0.004,
      "step": 109160
    },
    {
      "epoch": 0.7001824694628215,
      "grad_norm": 0.21076838672161102,
      "learning_rate": 8.212727838671308e-06,
      "loss": 0.0049,
      "step": 109170
    },
    {
      "epoch": 0.7002466063566075,
      "grad_norm": 0.19034221768379211,
      "learning_rate": 8.212298947245605e-06,
      "loss": 0.0018,
      "step": 109180
    },
    {
      "epoch": 0.7003107432503937,
      "grad_norm": 0.17826853692531586,
      "learning_rate": 8.211870015567503e-06,
      "loss": 0.0034,
      "step": 109190
    },
    {
      "epoch": 0.7003748801441797,
      "grad_norm": 0.08181103318929672,
      "learning_rate": 8.211441043642371e-06,
      "loss": 0.0044,
      "step": 109200
    },
    {
      "epoch": 0.7004390170379658,
      "grad_norm": 0.17758990824222565,
      "learning_rate": 8.21101203147559e-06,
      "loss": 0.0026,
      "step": 109210
    },
    {
      "epoch": 0.7005031539317519,
      "grad_norm": 0.14397963881492615,
      "learning_rate": 8.21058297907253e-06,
      "loss": 0.0022,
      "step": 109220
    },
    {
      "epoch": 0.700567290825538,
      "grad_norm": 0.09560272097587585,
      "learning_rate": 8.210153886438573e-06,
      "loss": 0.0024,
      "step": 109230
    },
    {
      "epoch": 0.7006314277193241,
      "grad_norm": 0.18891851603984833,
      "learning_rate": 8.209724753579093e-06,
      "loss": 0.0042,
      "step": 109240
    },
    {
      "epoch": 0.7006955646131102,
      "grad_norm": 0.11252248287200928,
      "learning_rate": 8.209295580499467e-06,
      "loss": 0.0046,
      "step": 109250
    },
    {
      "epoch": 0.7007597015068964,
      "grad_norm": 0.03766665980219841,
      "learning_rate": 8.208866367205077e-06,
      "loss": 0.0029,
      "step": 109260
    },
    {
      "epoch": 0.7008238384006824,
      "grad_norm": 0.2850850224494934,
      "learning_rate": 8.208437113701295e-06,
      "loss": 0.0034,
      "step": 109270
    },
    {
      "epoch": 0.7008879752944686,
      "grad_norm": 0.043535564094781876,
      "learning_rate": 8.208007819993505e-06,
      "loss": 0.0024,
      "step": 109280
    },
    {
      "epoch": 0.7009521121882546,
      "grad_norm": 0.4656340479850769,
      "learning_rate": 8.207578486087083e-06,
      "loss": 0.0022,
      "step": 109290
    },
    {
      "epoch": 0.7010162490820407,
      "grad_norm": 0.02832138165831566,
      "learning_rate": 8.207149111987409e-06,
      "loss": 0.0024,
      "step": 109300
    },
    {
      "epoch": 0.7010803859758268,
      "grad_norm": 0.3228578567504883,
      "learning_rate": 8.206719697699866e-06,
      "loss": 0.0113,
      "step": 109310
    },
    {
      "epoch": 0.7011445228696129,
      "grad_norm": 0.3972824215888977,
      "learning_rate": 8.206290243229833e-06,
      "loss": 0.0046,
      "step": 109320
    },
    {
      "epoch": 0.701208659763399,
      "grad_norm": 0.13330219686031342,
      "learning_rate": 8.205860748582692e-06,
      "loss": 0.0051,
      "step": 109330
    },
    {
      "epoch": 0.7012727966571851,
      "grad_norm": 0.0320163257420063,
      "learning_rate": 8.205431213763824e-06,
      "loss": 0.0039,
      "step": 109340
    },
    {
      "epoch": 0.7013369335509712,
      "grad_norm": 0.08879612386226654,
      "learning_rate": 8.205001638778614e-06,
      "loss": 0.003,
      "step": 109350
    },
    {
      "epoch": 0.7014010704447573,
      "grad_norm": 0.1196327656507492,
      "learning_rate": 8.20457202363244e-06,
      "loss": 0.0031,
      "step": 109360
    },
    {
      "epoch": 0.7014652073385433,
      "grad_norm": 0.1541898101568222,
      "learning_rate": 8.20414236833069e-06,
      "loss": 0.0034,
      "step": 109370
    },
    {
      "epoch": 0.7015293442323295,
      "grad_norm": 0.18361224234104156,
      "learning_rate": 8.203712672878745e-06,
      "loss": 0.0022,
      "step": 109380
    },
    {
      "epoch": 0.7015934811261155,
      "grad_norm": 0.0930710956454277,
      "learning_rate": 8.203282937281991e-06,
      "loss": 0.0022,
      "step": 109390
    },
    {
      "epoch": 0.7016576180199017,
      "grad_norm": 0.0693768560886383,
      "learning_rate": 8.202853161545814e-06,
      "loss": 0.0027,
      "step": 109400
    },
    {
      "epoch": 0.7017217549136878,
      "grad_norm": 0.11731088906526566,
      "learning_rate": 8.202423345675597e-06,
      "loss": 0.0022,
      "step": 109410
    },
    {
      "epoch": 0.7017858918074739,
      "grad_norm": 0.12514188885688782,
      "learning_rate": 8.201993489676724e-06,
      "loss": 0.0025,
      "step": 109420
    },
    {
      "epoch": 0.70185002870126,
      "grad_norm": 0.1392947882413864,
      "learning_rate": 8.201563593554587e-06,
      "loss": 0.0025,
      "step": 109430
    },
    {
      "epoch": 0.701914165595046,
      "grad_norm": 0.1546277105808258,
      "learning_rate": 8.20113365731457e-06,
      "loss": 0.0041,
      "step": 109440
    },
    {
      "epoch": 0.7019783024888322,
      "grad_norm": 0.09571681916713715,
      "learning_rate": 8.20070368096206e-06,
      "loss": 0.0035,
      "step": 109450
    },
    {
      "epoch": 0.7020424393826182,
      "grad_norm": 0.20175130665302277,
      "learning_rate": 8.200273664502446e-06,
      "loss": 0.0025,
      "step": 109460
    },
    {
      "epoch": 0.7021065762764044,
      "grad_norm": 0.16607627272605896,
      "learning_rate": 8.199843607941115e-06,
      "loss": 0.0037,
      "step": 109470
    },
    {
      "epoch": 0.7021707131701904,
      "grad_norm": 0.21978610754013062,
      "learning_rate": 8.199413511283456e-06,
      "loss": 0.003,
      "step": 109480
    },
    {
      "epoch": 0.7022348500639766,
      "grad_norm": 0.1604081243276596,
      "learning_rate": 8.198983374534861e-06,
      "loss": 0.0029,
      "step": 109490
    },
    {
      "epoch": 0.7022989869577626,
      "grad_norm": 0.12393929809331894,
      "learning_rate": 8.198553197700717e-06,
      "loss": 0.002,
      "step": 109500
    },
    {
      "epoch": 0.7023631238515488,
      "grad_norm": 0.21514670550823212,
      "learning_rate": 8.198122980786416e-06,
      "loss": 0.003,
      "step": 109510
    },
    {
      "epoch": 0.7024272607453348,
      "grad_norm": 0.03178085386753082,
      "learning_rate": 8.197692723797349e-06,
      "loss": 0.0055,
      "step": 109520
    },
    {
      "epoch": 0.7024913976391209,
      "grad_norm": 0.2929263114929199,
      "learning_rate": 8.197262426738903e-06,
      "loss": 0.0018,
      "step": 109530
    },
    {
      "epoch": 0.7025555345329071,
      "grad_norm": 0.09754671901464462,
      "learning_rate": 8.196832089616477e-06,
      "loss": 0.0033,
      "step": 109540
    },
    {
      "epoch": 0.7026196714266931,
      "grad_norm": 0.13949570059776306,
      "learning_rate": 8.19640171243546e-06,
      "loss": 0.0023,
      "step": 109550
    },
    {
      "epoch": 0.7026838083204793,
      "grad_norm": 0.2180301547050476,
      "learning_rate": 8.195971295201245e-06,
      "loss": 0.0038,
      "step": 109560
    },
    {
      "epoch": 0.7027479452142653,
      "grad_norm": 0.059337735176086426,
      "learning_rate": 8.195540837919224e-06,
      "loss": 0.0025,
      "step": 109570
    },
    {
      "epoch": 0.7028120821080515,
      "grad_norm": 0.07762123644351959,
      "learning_rate": 8.195110340594795e-06,
      "loss": 0.004,
      "step": 109580
    },
    {
      "epoch": 0.7028762190018375,
      "grad_norm": 0.5540107488632202,
      "learning_rate": 8.19467980323335e-06,
      "loss": 0.0018,
      "step": 109590
    },
    {
      "epoch": 0.7029403558956236,
      "grad_norm": 0.24830619990825653,
      "learning_rate": 8.194249225840283e-06,
      "loss": 0.0039,
      "step": 109600
    },
    {
      "epoch": 0.7030044927894097,
      "grad_norm": 0.16720372438430786,
      "learning_rate": 8.193818608420988e-06,
      "loss": 0.0026,
      "step": 109610
    },
    {
      "epoch": 0.7030686296831958,
      "grad_norm": 0.12153538316488266,
      "learning_rate": 8.193387950980864e-06,
      "loss": 0.0018,
      "step": 109620
    },
    {
      "epoch": 0.7031327665769819,
      "grad_norm": 0.1623486578464508,
      "learning_rate": 8.19295725352531e-06,
      "loss": 0.0022,
      "step": 109630
    },
    {
      "epoch": 0.703196903470768,
      "grad_norm": 0.1736610233783722,
      "learning_rate": 8.192526516059715e-06,
      "loss": 0.0018,
      "step": 109640
    },
    {
      "epoch": 0.7032610403645541,
      "grad_norm": 0.31248173117637634,
      "learning_rate": 8.192095738589484e-06,
      "loss": 0.0017,
      "step": 109650
    },
    {
      "epoch": 0.7033251772583402,
      "grad_norm": 0.13344676792621613,
      "learning_rate": 8.191664921120014e-06,
      "loss": 0.0026,
      "step": 109660
    },
    {
      "epoch": 0.7033893141521262,
      "grad_norm": 0.13243715465068817,
      "learning_rate": 8.191234063656698e-06,
      "loss": 0.002,
      "step": 109670
    },
    {
      "epoch": 0.7034534510459124,
      "grad_norm": 0.17147253453731537,
      "learning_rate": 8.19080316620494e-06,
      "loss": 0.0027,
      "step": 109680
    },
    {
      "epoch": 0.7035175879396985,
      "grad_norm": 0.010082487016916275,
      "learning_rate": 8.190372228770137e-06,
      "loss": 0.0027,
      "step": 109690
    },
    {
      "epoch": 0.7035817248334846,
      "grad_norm": 0.3268413543701172,
      "learning_rate": 8.18994125135769e-06,
      "loss": 0.0028,
      "step": 109700
    },
    {
      "epoch": 0.7036458617272707,
      "grad_norm": 0.06293269246816635,
      "learning_rate": 8.189510233973e-06,
      "loss": 0.004,
      "step": 109710
    },
    {
      "epoch": 0.7037099986210568,
      "grad_norm": 0.16118904948234558,
      "learning_rate": 8.189079176621465e-06,
      "loss": 0.0035,
      "step": 109720
    },
    {
      "epoch": 0.7037741355148429,
      "grad_norm": 0.13254894316196442,
      "learning_rate": 8.188648079308492e-06,
      "loss": 0.0024,
      "step": 109730
    },
    {
      "epoch": 0.703838272408629,
      "grad_norm": 0.1677081286907196,
      "learning_rate": 8.188216942039477e-06,
      "loss": 0.0034,
      "step": 109740
    },
    {
      "epoch": 0.7039024093024151,
      "grad_norm": 0.2145616114139557,
      "learning_rate": 8.187785764819826e-06,
      "loss": 0.0023,
      "step": 109750
    },
    {
      "epoch": 0.7039665461962011,
      "grad_norm": 0.12412948906421661,
      "learning_rate": 8.187354547654942e-06,
      "loss": 0.0035,
      "step": 109760
    },
    {
      "epoch": 0.7040306830899873,
      "grad_norm": 0.43917155265808105,
      "learning_rate": 8.186923290550227e-06,
      "loss": 0.0041,
      "step": 109770
    },
    {
      "epoch": 0.7040948199837733,
      "grad_norm": 0.14502884447574615,
      "learning_rate": 8.186491993511086e-06,
      "loss": 0.0029,
      "step": 109780
    },
    {
      "epoch": 0.7041589568775595,
      "grad_norm": 0.052646420896053314,
      "learning_rate": 8.186060656542922e-06,
      "loss": 0.0022,
      "step": 109790
    },
    {
      "epoch": 0.7042230937713455,
      "grad_norm": 0.12846732139587402,
      "learning_rate": 8.185629279651142e-06,
      "loss": 0.002,
      "step": 109800
    },
    {
      "epoch": 0.7042872306651317,
      "grad_norm": 0.3380899429321289,
      "learning_rate": 8.18519786284115e-06,
      "loss": 0.0025,
      "step": 109810
    },
    {
      "epoch": 0.7043513675589178,
      "grad_norm": 0.09375136345624924,
      "learning_rate": 8.184766406118351e-06,
      "loss": 0.0032,
      "step": 109820
    },
    {
      "epoch": 0.7044155044527038,
      "grad_norm": 0.03884616494178772,
      "learning_rate": 8.184334909488154e-06,
      "loss": 0.0028,
      "step": 109830
    },
    {
      "epoch": 0.70447964134649,
      "grad_norm": 0.10116954147815704,
      "learning_rate": 8.183903372955965e-06,
      "loss": 0.0042,
      "step": 109840
    },
    {
      "epoch": 0.704543778240276,
      "grad_norm": 0.23590809106826782,
      "learning_rate": 8.183471796527191e-06,
      "loss": 0.0023,
      "step": 109850
    },
    {
      "epoch": 0.7046079151340622,
      "grad_norm": 0.2649398148059845,
      "learning_rate": 8.183040180207241e-06,
      "loss": 0.0026,
      "step": 109860
    },
    {
      "epoch": 0.7046720520278482,
      "grad_norm": 0.0337352491915226,
      "learning_rate": 8.18260852400152e-06,
      "loss": 0.003,
      "step": 109870
    },
    {
      "epoch": 0.7047361889216344,
      "grad_norm": 0.22226230800151825,
      "learning_rate": 8.182176827915443e-06,
      "loss": 0.0042,
      "step": 109880
    },
    {
      "epoch": 0.7048003258154204,
      "grad_norm": 0.0936257615685463,
      "learning_rate": 8.181745091954413e-06,
      "loss": 0.0049,
      "step": 109890
    },
    {
      "epoch": 0.7048644627092066,
      "grad_norm": 0.1624756008386612,
      "learning_rate": 8.181313316123844e-06,
      "loss": 0.0025,
      "step": 109900
    },
    {
      "epoch": 0.7049285996029926,
      "grad_norm": 0.022759467363357544,
      "learning_rate": 8.180881500429147e-06,
      "loss": 0.0051,
      "step": 109910
    },
    {
      "epoch": 0.7049927364967787,
      "grad_norm": 0.07997841387987137,
      "learning_rate": 8.18044964487573e-06,
      "loss": 0.0024,
      "step": 109920
    },
    {
      "epoch": 0.7050568733905648,
      "grad_norm": 0.10638929158449173,
      "learning_rate": 8.180017749469007e-06,
      "loss": 0.004,
      "step": 109930
    },
    {
      "epoch": 0.7051210102843509,
      "grad_norm": 0.11960377544164658,
      "learning_rate": 8.179585814214387e-06,
      "loss": 0.0011,
      "step": 109940
    },
    {
      "epoch": 0.705185147178137,
      "grad_norm": 0.27875372767448425,
      "learning_rate": 8.179153839117286e-06,
      "loss": 0.0029,
      "step": 109950
    },
    {
      "epoch": 0.7052492840719231,
      "grad_norm": 0.1304447054862976,
      "learning_rate": 8.178721824183116e-06,
      "loss": 0.0037,
      "step": 109960
    },
    {
      "epoch": 0.7053134209657093,
      "grad_norm": 0.14808189868927002,
      "learning_rate": 8.17828976941729e-06,
      "loss": 0.0013,
      "step": 109970
    },
    {
      "epoch": 0.7053775578594953,
      "grad_norm": 0.1582537591457367,
      "learning_rate": 8.177857674825217e-06,
      "loss": 0.0031,
      "step": 109980
    },
    {
      "epoch": 0.7054416947532814,
      "grad_norm": 0.1076810210943222,
      "learning_rate": 8.177425540412322e-06,
      "loss": 0.0046,
      "step": 109990
    },
    {
      "epoch": 0.7055058316470675,
      "grad_norm": 0.04820389673113823,
      "learning_rate": 8.176993366184012e-06,
      "loss": 0.0037,
      "step": 110000
    },
    {
      "epoch": 0.7055699685408536,
      "grad_norm": 0.1892995983362198,
      "learning_rate": 8.176561152145704e-06,
      "loss": 0.0018,
      "step": 110010
    },
    {
      "epoch": 0.7056341054346397,
      "grad_norm": 0.1477370262145996,
      "learning_rate": 8.176128898302813e-06,
      "loss": 0.0028,
      "step": 110020
    },
    {
      "epoch": 0.7056982423284258,
      "grad_norm": 0.06276816874742508,
      "learning_rate": 8.175696604660759e-06,
      "loss": 0.0038,
      "step": 110030
    },
    {
      "epoch": 0.7057623792222119,
      "grad_norm": 0.15604333579540253,
      "learning_rate": 8.175264271224957e-06,
      "loss": 0.0047,
      "step": 110040
    },
    {
      "epoch": 0.705826516115998,
      "grad_norm": 0.1175050437450409,
      "learning_rate": 8.174831898000824e-06,
      "loss": 0.003,
      "step": 110050
    },
    {
      "epoch": 0.705890653009784,
      "grad_norm": 0.21799638867378235,
      "learning_rate": 8.174399484993777e-06,
      "loss": 0.0039,
      "step": 110060
    },
    {
      "epoch": 0.7059547899035702,
      "grad_norm": 0.06989800930023193,
      "learning_rate": 8.173967032209237e-06,
      "loss": 0.0049,
      "step": 110070
    },
    {
      "epoch": 0.7060189267973562,
      "grad_norm": 0.055953364819288254,
      "learning_rate": 8.173534539652623e-06,
      "loss": 0.0018,
      "step": 110080
    },
    {
      "epoch": 0.7060830636911424,
      "grad_norm": 0.14363153278827667,
      "learning_rate": 8.173102007329353e-06,
      "loss": 0.0032,
      "step": 110090
    },
    {
      "epoch": 0.7061472005849285,
      "grad_norm": 0.05500480532646179,
      "learning_rate": 8.172669435244845e-06,
      "loss": 0.0031,
      "step": 110100
    },
    {
      "epoch": 0.7062113374787146,
      "grad_norm": 0.06739882379770279,
      "learning_rate": 8.17223682340452e-06,
      "loss": 0.0026,
      "step": 110110
    },
    {
      "epoch": 0.7062754743725007,
      "grad_norm": 0.05695938691496849,
      "learning_rate": 8.171804171813804e-06,
      "loss": 0.0018,
      "step": 110120
    },
    {
      "epoch": 0.7063396112662867,
      "grad_norm": 0.07226129621267319,
      "learning_rate": 8.171371480478115e-06,
      "loss": 0.0021,
      "step": 110130
    },
    {
      "epoch": 0.7064037481600729,
      "grad_norm": 0.20298391580581665,
      "learning_rate": 8.170938749402873e-06,
      "loss": 0.0027,
      "step": 110140
    },
    {
      "epoch": 0.7064678850538589,
      "grad_norm": 0.2183033972978592,
      "learning_rate": 8.170505978593501e-06,
      "loss": 0.0029,
      "step": 110150
    },
    {
      "epoch": 0.7065320219476451,
      "grad_norm": 0.11818025261163712,
      "learning_rate": 8.170073168055426e-06,
      "loss": 0.002,
      "step": 110160
    },
    {
      "epoch": 0.7065961588414311,
      "grad_norm": 0.06451547145843506,
      "learning_rate": 8.169640317794066e-06,
      "loss": 0.0021,
      "step": 110170
    },
    {
      "epoch": 0.7066602957352173,
      "grad_norm": 0.09229018539190292,
      "learning_rate": 8.16920742781485e-06,
      "loss": 0.0058,
      "step": 110180
    },
    {
      "epoch": 0.7067244326290033,
      "grad_norm": 0.10356393456459045,
      "learning_rate": 8.1687744981232e-06,
      "loss": 0.0037,
      "step": 110190
    },
    {
      "epoch": 0.7067885695227895,
      "grad_norm": 0.2616499960422516,
      "learning_rate": 8.168341528724539e-06,
      "loss": 0.0025,
      "step": 110200
    },
    {
      "epoch": 0.7068527064165755,
      "grad_norm": 0.07241521030664444,
      "learning_rate": 8.167908519624295e-06,
      "loss": 0.0034,
      "step": 110210
    },
    {
      "epoch": 0.7069168433103616,
      "grad_norm": 0.2841012179851532,
      "learning_rate": 8.167475470827893e-06,
      "loss": 0.0028,
      "step": 110220
    },
    {
      "epoch": 0.7069809802041477,
      "grad_norm": 0.14491187036037445,
      "learning_rate": 8.16704238234076e-06,
      "loss": 0.0025,
      "step": 110230
    },
    {
      "epoch": 0.7070451170979338,
      "grad_norm": 0.1743507832288742,
      "learning_rate": 8.166609254168321e-06,
      "loss": 0.0019,
      "step": 110240
    },
    {
      "epoch": 0.70710925399172,
      "grad_norm": 0.07084076851606369,
      "learning_rate": 8.166176086316007e-06,
      "loss": 0.003,
      "step": 110250
    },
    {
      "epoch": 0.707173390885506,
      "grad_norm": 0.12673059105873108,
      "learning_rate": 8.165742878789243e-06,
      "loss": 0.004,
      "step": 110260
    },
    {
      "epoch": 0.7072375277792922,
      "grad_norm": 0.23712843656539917,
      "learning_rate": 8.165309631593457e-06,
      "loss": 0.0026,
      "step": 110270
    },
    {
      "epoch": 0.7073016646730782,
      "grad_norm": 0.0420183502137661,
      "learning_rate": 8.164876344734081e-06,
      "loss": 0.0037,
      "step": 110280
    },
    {
      "epoch": 0.7073658015668643,
      "grad_norm": 0.2065224051475525,
      "learning_rate": 8.164443018216542e-06,
      "loss": 0.0019,
      "step": 110290
    },
    {
      "epoch": 0.7074299384606504,
      "grad_norm": 0.10177796334028244,
      "learning_rate": 8.164009652046269e-06,
      "loss": 0.0038,
      "step": 110300
    },
    {
      "epoch": 0.7074940753544365,
      "grad_norm": 0.06011974439024925,
      "learning_rate": 8.163576246228697e-06,
      "loss": 0.0021,
      "step": 110310
    },
    {
      "epoch": 0.7075582122482226,
      "grad_norm": 0.15856629610061646,
      "learning_rate": 8.16314280076925e-06,
      "loss": 0.0032,
      "step": 110320
    },
    {
      "epoch": 0.7076223491420087,
      "grad_norm": 0.2633246183395386,
      "learning_rate": 8.162709315673366e-06,
      "loss": 0.0034,
      "step": 110330
    },
    {
      "epoch": 0.7076864860357948,
      "grad_norm": 0.11729785054922104,
      "learning_rate": 8.162275790946472e-06,
      "loss": 0.0031,
      "step": 110340
    },
    {
      "epoch": 0.7077506229295809,
      "grad_norm": 0.10076826810836792,
      "learning_rate": 8.161842226594002e-06,
      "loss": 0.0042,
      "step": 110350
    },
    {
      "epoch": 0.707814759823367,
      "grad_norm": 0.12335386872291565,
      "learning_rate": 8.161408622621391e-06,
      "loss": 0.0023,
      "step": 110360
    },
    {
      "epoch": 0.7078788967171531,
      "grad_norm": 0.05566006898880005,
      "learning_rate": 8.160974979034068e-06,
      "loss": 0.0022,
      "step": 110370
    },
    {
      "epoch": 0.7079430336109392,
      "grad_norm": 0.13265535235404968,
      "learning_rate": 8.160541295837471e-06,
      "loss": 0.0034,
      "step": 110380
    },
    {
      "epoch": 0.7080071705047253,
      "grad_norm": 0.23299631476402283,
      "learning_rate": 8.160107573037034e-06,
      "loss": 0.0035,
      "step": 110390
    },
    {
      "epoch": 0.7080713073985114,
      "grad_norm": 0.040941670536994934,
      "learning_rate": 8.159673810638188e-06,
      "loss": 0.0032,
      "step": 110400
    },
    {
      "epoch": 0.7081354442922975,
      "grad_norm": 0.2033630758523941,
      "learning_rate": 8.159240008646374e-06,
      "loss": 0.003,
      "step": 110410
    },
    {
      "epoch": 0.7081995811860836,
      "grad_norm": 0.2089751809835434,
      "learning_rate": 8.158806167067022e-06,
      "loss": 0.0013,
      "step": 110420
    },
    {
      "epoch": 0.7082637180798697,
      "grad_norm": 0.07643994688987732,
      "learning_rate": 8.158372285905573e-06,
      "loss": 0.0015,
      "step": 110430
    },
    {
      "epoch": 0.7083278549736558,
      "grad_norm": 0.08906198292970657,
      "learning_rate": 8.157938365167461e-06,
      "loss": 0.0036,
      "step": 110440
    },
    {
      "epoch": 0.7083919918674418,
      "grad_norm": 0.32955411076545715,
      "learning_rate": 8.157504404858125e-06,
      "loss": 0.0024,
      "step": 110450
    },
    {
      "epoch": 0.708456128761228,
      "grad_norm": 0.18114842474460602,
      "learning_rate": 8.157070404983001e-06,
      "loss": 0.0053,
      "step": 110460
    },
    {
      "epoch": 0.708520265655014,
      "grad_norm": 0.20772533118724823,
      "learning_rate": 8.15663636554753e-06,
      "loss": 0.0031,
      "step": 110470
    },
    {
      "epoch": 0.7085844025488002,
      "grad_norm": 0.12291453778743744,
      "learning_rate": 8.156202286557149e-06,
      "loss": 0.003,
      "step": 110480
    },
    {
      "epoch": 0.7086485394425862,
      "grad_norm": 0.23288308084011078,
      "learning_rate": 8.155768168017298e-06,
      "loss": 0.0057,
      "step": 110490
    },
    {
      "epoch": 0.7087126763363724,
      "grad_norm": 0.12418215721845627,
      "learning_rate": 8.155334009933414e-06,
      "loss": 0.0045,
      "step": 110500
    },
    {
      "epoch": 0.7087768132301584,
      "grad_norm": 0.2861568033695221,
      "learning_rate": 8.154899812310942e-06,
      "loss": 0.0029,
      "step": 110510
    },
    {
      "epoch": 0.7088409501239445,
      "grad_norm": 0.11753934621810913,
      "learning_rate": 8.15446557515532e-06,
      "loss": 0.0025,
      "step": 110520
    },
    {
      "epoch": 0.7089050870177307,
      "grad_norm": 0.18481867015361786,
      "learning_rate": 8.15403129847199e-06,
      "loss": 0.0041,
      "step": 110530
    },
    {
      "epoch": 0.7089692239115167,
      "grad_norm": 0.47326895594596863,
      "learning_rate": 8.153596982266392e-06,
      "loss": 0.005,
      "step": 110540
    },
    {
      "epoch": 0.7090333608053029,
      "grad_norm": 0.16735628247261047,
      "learning_rate": 8.153162626543972e-06,
      "loss": 0.0048,
      "step": 110550
    },
    {
      "epoch": 0.7090974976990889,
      "grad_norm": 0.1563166379928589,
      "learning_rate": 8.15272823131017e-06,
      "loss": 0.0025,
      "step": 110560
    },
    {
      "epoch": 0.7091616345928751,
      "grad_norm": 0.04397542029619217,
      "learning_rate": 8.152293796570432e-06,
      "loss": 0.0031,
      "step": 110570
    },
    {
      "epoch": 0.7092257714866611,
      "grad_norm": 0.5582118034362793,
      "learning_rate": 8.151859322330197e-06,
      "loss": 0.0049,
      "step": 110580
    },
    {
      "epoch": 0.7092899083804473,
      "grad_norm": 0.1282324641942978,
      "learning_rate": 8.151424808594914e-06,
      "loss": 0.0037,
      "step": 110590
    },
    {
      "epoch": 0.7093540452742333,
      "grad_norm": 0.27893584966659546,
      "learning_rate": 8.150990255370025e-06,
      "loss": 0.0031,
      "step": 110600
    },
    {
      "epoch": 0.7094181821680194,
      "grad_norm": 0.05734465271234512,
      "learning_rate": 8.150555662660976e-06,
      "loss": 0.0029,
      "step": 110610
    },
    {
      "epoch": 0.7094823190618055,
      "grad_norm": 0.06638863682746887,
      "learning_rate": 8.150121030473214e-06,
      "loss": 0.002,
      "step": 110620
    },
    {
      "epoch": 0.7095464559555916,
      "grad_norm": 0.13825614750385284,
      "learning_rate": 8.149686358812183e-06,
      "loss": 0.0041,
      "step": 110630
    },
    {
      "epoch": 0.7096105928493777,
      "grad_norm": 0.10324165970087051,
      "learning_rate": 8.149251647683332e-06,
      "loss": 0.0067,
      "step": 110640
    },
    {
      "epoch": 0.7096747297431638,
      "grad_norm": 0.04771149903535843,
      "learning_rate": 8.148816897092106e-06,
      "loss": 0.0024,
      "step": 110650
    },
    {
      "epoch": 0.70973886663695,
      "grad_norm": 0.05605524405837059,
      "learning_rate": 8.148382107043954e-06,
      "loss": 0.005,
      "step": 110660
    },
    {
      "epoch": 0.709803003530736,
      "grad_norm": 0.0790896937251091,
      "learning_rate": 8.147947277544324e-06,
      "loss": 0.0026,
      "step": 110670
    },
    {
      "epoch": 0.7098671404245221,
      "grad_norm": 0.18265533447265625,
      "learning_rate": 8.147512408598664e-06,
      "loss": 0.0028,
      "step": 110680
    },
    {
      "epoch": 0.7099312773183082,
      "grad_norm": 0.06397712975740433,
      "learning_rate": 8.147077500212426e-06,
      "loss": 0.0025,
      "step": 110690
    },
    {
      "epoch": 0.7099954142120943,
      "grad_norm": 0.20617574453353882,
      "learning_rate": 8.146642552391057e-06,
      "loss": 0.0031,
      "step": 110700
    },
    {
      "epoch": 0.7100595511058804,
      "grad_norm": 0.09721191972494125,
      "learning_rate": 8.146207565140007e-06,
      "loss": 0.0022,
      "step": 110710
    },
    {
      "epoch": 0.7101236879996665,
      "grad_norm": 0.05267834663391113,
      "learning_rate": 8.145772538464729e-06,
      "loss": 0.0024,
      "step": 110720
    },
    {
      "epoch": 0.7101878248934526,
      "grad_norm": 0.12130790948867798,
      "learning_rate": 8.145337472370672e-06,
      "loss": 0.002,
      "step": 110730
    },
    {
      "epoch": 0.7102519617872387,
      "grad_norm": 0.23529313504695892,
      "learning_rate": 8.14490236686329e-06,
      "loss": 0.0026,
      "step": 110740
    },
    {
      "epoch": 0.7103160986810247,
      "grad_norm": 0.06817247718572617,
      "learning_rate": 8.14446722194803e-06,
      "loss": 0.0018,
      "step": 110750
    },
    {
      "epoch": 0.7103802355748109,
      "grad_norm": 0.15603575110435486,
      "learning_rate": 8.144032037630351e-06,
      "loss": 0.0025,
      "step": 110760
    },
    {
      "epoch": 0.7104443724685969,
      "grad_norm": 0.061270855367183685,
      "learning_rate": 8.143596813915704e-06,
      "loss": 0.0027,
      "step": 110770
    },
    {
      "epoch": 0.7105085093623831,
      "grad_norm": 0.1100795716047287,
      "learning_rate": 8.143161550809542e-06,
      "loss": 0.0033,
      "step": 110780
    },
    {
      "epoch": 0.7105726462561691,
      "grad_norm": 0.6656703352928162,
      "learning_rate": 8.14272624831732e-06,
      "loss": 0.0098,
      "step": 110790
    },
    {
      "epoch": 0.7106367831499553,
      "grad_norm": 0.1621638387441635,
      "learning_rate": 8.14229090644449e-06,
      "loss": 0.0032,
      "step": 110800
    },
    {
      "epoch": 0.7107009200437414,
      "grad_norm": 0.03211973235011101,
      "learning_rate": 8.14185552519651e-06,
      "loss": 0.0024,
      "step": 110810
    },
    {
      "epoch": 0.7107650569375275,
      "grad_norm": 0.12337400019168854,
      "learning_rate": 8.141420104578836e-06,
      "loss": 0.002,
      "step": 110820
    },
    {
      "epoch": 0.7108291938313136,
      "grad_norm": 0.027781283482909203,
      "learning_rate": 8.140984644596921e-06,
      "loss": 0.001,
      "step": 110830
    },
    {
      "epoch": 0.7108933307250996,
      "grad_norm": 0.19105949997901917,
      "learning_rate": 8.140549145256225e-06,
      "loss": 0.0032,
      "step": 110840
    },
    {
      "epoch": 0.7109574676188858,
      "grad_norm": 0.15865445137023926,
      "learning_rate": 8.140113606562204e-06,
      "loss": 0.0025,
      "step": 110850
    },
    {
      "epoch": 0.7110216045126718,
      "grad_norm": 0.16358639299869537,
      "learning_rate": 8.139678028520315e-06,
      "loss": 0.0026,
      "step": 110860
    },
    {
      "epoch": 0.711085741406458,
      "grad_norm": 0.3070227801799774,
      "learning_rate": 8.139242411136015e-06,
      "loss": 0.0028,
      "step": 110870
    },
    {
      "epoch": 0.711149878300244,
      "grad_norm": 0.20219583809375763,
      "learning_rate": 8.138806754414765e-06,
      "loss": 0.0041,
      "step": 110880
    },
    {
      "epoch": 0.7112140151940302,
      "grad_norm": 0.34467488527297974,
      "learning_rate": 8.138371058362022e-06,
      "loss": 0.0024,
      "step": 110890
    },
    {
      "epoch": 0.7112781520878162,
      "grad_norm": 0.10602174699306488,
      "learning_rate": 8.137935322983247e-06,
      "loss": 0.0061,
      "step": 110900
    },
    {
      "epoch": 0.7113422889816023,
      "grad_norm": 0.0546005479991436,
      "learning_rate": 8.137499548283902e-06,
      "loss": 0.003,
      "step": 110910
    },
    {
      "epoch": 0.7114064258753884,
      "grad_norm": 0.11193834245204926,
      "learning_rate": 8.137063734269444e-06,
      "loss": 0.0043,
      "step": 110920
    },
    {
      "epoch": 0.7114705627691745,
      "grad_norm": 0.24336637556552887,
      "learning_rate": 8.136627880945336e-06,
      "loss": 0.0021,
      "step": 110930
    },
    {
      "epoch": 0.7115346996629606,
      "grad_norm": 0.14344747364521027,
      "learning_rate": 8.136191988317037e-06,
      "loss": 0.0035,
      "step": 110940
    },
    {
      "epoch": 0.7115988365567467,
      "grad_norm": 0.04977329447865486,
      "learning_rate": 8.135756056390013e-06,
      "loss": 0.0038,
      "step": 110950
    },
    {
      "epoch": 0.7116629734505329,
      "grad_norm": 0.18715287744998932,
      "learning_rate": 8.135320085169723e-06,
      "loss": 0.0031,
      "step": 110960
    },
    {
      "epoch": 0.7117271103443189,
      "grad_norm": 0.010879084467887878,
      "learning_rate": 8.134884074661633e-06,
      "loss": 0.0021,
      "step": 110970
    },
    {
      "epoch": 0.711791247238105,
      "grad_norm": 0.15092532336711884,
      "learning_rate": 8.134448024871204e-06,
      "loss": 0.0042,
      "step": 110980
    },
    {
      "epoch": 0.7118553841318911,
      "grad_norm": 0.11990223824977875,
      "learning_rate": 8.134011935803901e-06,
      "loss": 0.0026,
      "step": 110990
    },
    {
      "epoch": 0.7119195210256772,
      "grad_norm": 0.0894118919968605,
      "learning_rate": 8.13357580746519e-06,
      "loss": 0.003,
      "step": 111000
    },
    {
      "epoch": 0.7119836579194633,
      "grad_norm": 0.1143919974565506,
      "learning_rate": 8.133139639860533e-06,
      "loss": 0.0021,
      "step": 111010
    },
    {
      "epoch": 0.7120477948132494,
      "grad_norm": 0.16754014790058136,
      "learning_rate": 8.132703432995398e-06,
      "loss": 0.0042,
      "step": 111020
    },
    {
      "epoch": 0.7121119317070355,
      "grad_norm": 0.11708588898181915,
      "learning_rate": 8.13226718687525e-06,
      "loss": 0.002,
      "step": 111030
    },
    {
      "epoch": 0.7121760686008216,
      "grad_norm": 0.13375858962535858,
      "learning_rate": 8.131830901505556e-06,
      "loss": 0.0032,
      "step": 111040
    },
    {
      "epoch": 0.7122402054946076,
      "grad_norm": 0.054817404597997665,
      "learning_rate": 8.13139457689178e-06,
      "loss": 0.0027,
      "step": 111050
    },
    {
      "epoch": 0.7123043423883938,
      "grad_norm": 0.2613508701324463,
      "learning_rate": 8.130958213039395e-06,
      "loss": 0.0023,
      "step": 111060
    },
    {
      "epoch": 0.7123684792821798,
      "grad_norm": 0.04319367557764053,
      "learning_rate": 8.130521809953863e-06,
      "loss": 0.0027,
      "step": 111070
    },
    {
      "epoch": 0.712432616175966,
      "grad_norm": 0.15703964233398438,
      "learning_rate": 8.130085367640659e-06,
      "loss": 0.0021,
      "step": 111080
    },
    {
      "epoch": 0.7124967530697521,
      "grad_norm": 0.21569177508354187,
      "learning_rate": 8.129648886105246e-06,
      "loss": 0.0017,
      "step": 111090
    },
    {
      "epoch": 0.7125608899635382,
      "grad_norm": 0.08998652547597885,
      "learning_rate": 8.129212365353096e-06,
      "loss": 0.0032,
      "step": 111100
    },
    {
      "epoch": 0.7126250268573243,
      "grad_norm": 0.07066009193658829,
      "learning_rate": 8.12877580538968e-06,
      "loss": 0.0046,
      "step": 111110
    },
    {
      "epoch": 0.7126891637511104,
      "grad_norm": 0.13526740670204163,
      "learning_rate": 8.128339206220466e-06,
      "loss": 0.0021,
      "step": 111120
    },
    {
      "epoch": 0.7127533006448965,
      "grad_norm": 0.2631956934928894,
      "learning_rate": 8.127902567850924e-06,
      "loss": 0.0038,
      "step": 111130
    },
    {
      "epoch": 0.7128174375386825,
      "grad_norm": 0.14094075560569763,
      "learning_rate": 8.12746589028653e-06,
      "loss": 0.0024,
      "step": 111140
    },
    {
      "epoch": 0.7128815744324687,
      "grad_norm": 0.08425044268369675,
      "learning_rate": 8.127029173532753e-06,
      "loss": 0.0027,
      "step": 111150
    },
    {
      "epoch": 0.7129457113262547,
      "grad_norm": 0.14520998299121857,
      "learning_rate": 8.126592417595065e-06,
      "loss": 0.0039,
      "step": 111160
    },
    {
      "epoch": 0.7130098482200409,
      "grad_norm": 0.07398927211761475,
      "learning_rate": 8.12615562247894e-06,
      "loss": 0.0015,
      "step": 111170
    },
    {
      "epoch": 0.7130739851138269,
      "grad_norm": 0.036775704473257065,
      "learning_rate": 8.12571878818985e-06,
      "loss": 0.0029,
      "step": 111180
    },
    {
      "epoch": 0.7131381220076131,
      "grad_norm": 0.34897202253341675,
      "learning_rate": 8.12528191473327e-06,
      "loss": 0.0046,
      "step": 111190
    },
    {
      "epoch": 0.7132022589013991,
      "grad_norm": 0.2502409517765045,
      "learning_rate": 8.124845002114674e-06,
      "loss": 0.0036,
      "step": 111200
    },
    {
      "epoch": 0.7132663957951852,
      "grad_norm": 0.07636435329914093,
      "learning_rate": 8.124408050339536e-06,
      "loss": 0.0124,
      "step": 111210
    },
    {
      "epoch": 0.7133305326889713,
      "grad_norm": 0.1467924267053604,
      "learning_rate": 8.123971059413333e-06,
      "loss": 0.0032,
      "step": 111220
    },
    {
      "epoch": 0.7133946695827574,
      "grad_norm": 0.12310754507780075,
      "learning_rate": 8.12353402934154e-06,
      "loss": 0.0033,
      "step": 111230
    },
    {
      "epoch": 0.7134588064765436,
      "grad_norm": 0.2299281358718872,
      "learning_rate": 8.123096960129633e-06,
      "loss": 0.0037,
      "step": 111240
    },
    {
      "epoch": 0.7135229433703296,
      "grad_norm": 0.1423099786043167,
      "learning_rate": 8.12265985178309e-06,
      "loss": 0.0048,
      "step": 111250
    },
    {
      "epoch": 0.7135870802641158,
      "grad_norm": 0.12548957765102386,
      "learning_rate": 8.122222704307386e-06,
      "loss": 0.0019,
      "step": 111260
    },
    {
      "epoch": 0.7136512171579018,
      "grad_norm": 0.10565450042486191,
      "learning_rate": 8.121785517708e-06,
      "loss": 0.0034,
      "step": 111270
    },
    {
      "epoch": 0.713715354051688,
      "grad_norm": 0.11274101585149765,
      "learning_rate": 8.121348291990411e-06,
      "loss": 0.0025,
      "step": 111280
    },
    {
      "epoch": 0.713779490945474,
      "grad_norm": 0.08379390090703964,
      "learning_rate": 8.120911027160097e-06,
      "loss": 0.0027,
      "step": 111290
    },
    {
      "epoch": 0.7138436278392601,
      "grad_norm": 0.20944179594516754,
      "learning_rate": 8.120473723222537e-06,
      "loss": 0.0042,
      "step": 111300
    },
    {
      "epoch": 0.7139077647330462,
      "grad_norm": 0.3111606240272522,
      "learning_rate": 8.120036380183212e-06,
      "loss": 0.0039,
      "step": 111310
    },
    {
      "epoch": 0.7139719016268323,
      "grad_norm": 0.22834256291389465,
      "learning_rate": 8.1195989980476e-06,
      "loss": 0.0025,
      "step": 111320
    },
    {
      "epoch": 0.7140360385206184,
      "grad_norm": 0.186443492770195,
      "learning_rate": 8.119161576821185e-06,
      "loss": 0.002,
      "step": 111330
    },
    {
      "epoch": 0.7141001754144045,
      "grad_norm": 0.056203048676252365,
      "learning_rate": 8.118724116509444e-06,
      "loss": 0.0031,
      "step": 111340
    },
    {
      "epoch": 0.7141643123081906,
      "grad_norm": 0.17152699828147888,
      "learning_rate": 8.118286617117863e-06,
      "loss": 0.0034,
      "step": 111350
    },
    {
      "epoch": 0.7142284492019767,
      "grad_norm": 0.09365913271903992,
      "learning_rate": 8.11784907865192e-06,
      "loss": 0.0034,
      "step": 111360
    },
    {
      "epoch": 0.7142925860957628,
      "grad_norm": 0.23707567155361176,
      "learning_rate": 8.1174115011171e-06,
      "loss": 0.0052,
      "step": 111370
    },
    {
      "epoch": 0.7143567229895489,
      "grad_norm": 0.0613962784409523,
      "learning_rate": 8.116973884518888e-06,
      "loss": 0.0055,
      "step": 111380
    },
    {
      "epoch": 0.714420859883335,
      "grad_norm": 0.1931670755147934,
      "learning_rate": 8.116536228862764e-06,
      "loss": 0.0033,
      "step": 111390
    },
    {
      "epoch": 0.7144849967771211,
      "grad_norm": 0.39843279123306274,
      "learning_rate": 8.116098534154214e-06,
      "loss": 0.0043,
      "step": 111400
    },
    {
      "epoch": 0.7145491336709072,
      "grad_norm": 0.10760731995105743,
      "learning_rate": 8.115660800398723e-06,
      "loss": 0.003,
      "step": 111410
    },
    {
      "epoch": 0.7146132705646933,
      "grad_norm": 0.23782630264759064,
      "learning_rate": 8.115223027601776e-06,
      "loss": 0.0037,
      "step": 111420
    },
    {
      "epoch": 0.7146774074584794,
      "grad_norm": 0.1823507845401764,
      "learning_rate": 8.114785215768854e-06,
      "loss": 0.0015,
      "step": 111430
    },
    {
      "epoch": 0.7147415443522654,
      "grad_norm": 0.1340901404619217,
      "learning_rate": 8.114347364905451e-06,
      "loss": 0.0035,
      "step": 111440
    },
    {
      "epoch": 0.7148056812460516,
      "grad_norm": 0.12971192598342896,
      "learning_rate": 8.11390947501705e-06,
      "loss": 0.0033,
      "step": 111450
    },
    {
      "epoch": 0.7148698181398376,
      "grad_norm": 0.20929332077503204,
      "learning_rate": 8.113471546109135e-06,
      "loss": 0.0021,
      "step": 111460
    },
    {
      "epoch": 0.7149339550336238,
      "grad_norm": 0.13546518981456757,
      "learning_rate": 8.113033578187199e-06,
      "loss": 0.0028,
      "step": 111470
    },
    {
      "epoch": 0.7149980919274098,
      "grad_norm": 0.15590383112430573,
      "learning_rate": 8.112595571256725e-06,
      "loss": 0.0038,
      "step": 111480
    },
    {
      "epoch": 0.715062228821196,
      "grad_norm": 0.08301430940628052,
      "learning_rate": 8.112157525323206e-06,
      "loss": 0.0022,
      "step": 111490
    },
    {
      "epoch": 0.715126365714982,
      "grad_norm": 0.04288941249251366,
      "learning_rate": 8.111719440392127e-06,
      "loss": 0.0015,
      "step": 111500
    },
    {
      "epoch": 0.7151905026087682,
      "grad_norm": 0.21083122491836548,
      "learning_rate": 8.111281316468981e-06,
      "loss": 0.002,
      "step": 111510
    },
    {
      "epoch": 0.7152546395025543,
      "grad_norm": 0.06173493340611458,
      "learning_rate": 8.110843153559257e-06,
      "loss": 0.0048,
      "step": 111520
    },
    {
      "epoch": 0.7153187763963403,
      "grad_norm": 0.17022399604320526,
      "learning_rate": 8.110404951668444e-06,
      "loss": 0.0168,
      "step": 111530
    },
    {
      "epoch": 0.7153829132901265,
      "grad_norm": 0.03184816986322403,
      "learning_rate": 8.109966710802033e-06,
      "loss": 0.0021,
      "step": 111540
    },
    {
      "epoch": 0.7154470501839125,
      "grad_norm": 0.1753406822681427,
      "learning_rate": 8.10952843096552e-06,
      "loss": 0.0031,
      "step": 111550
    },
    {
      "epoch": 0.7155111870776987,
      "grad_norm": 0.12510491907596588,
      "learning_rate": 8.10909011216439e-06,
      "loss": 0.0036,
      "step": 111560
    },
    {
      "epoch": 0.7155753239714847,
      "grad_norm": 0.07739254832267761,
      "learning_rate": 8.10865175440414e-06,
      "loss": 0.0027,
      "step": 111570
    },
    {
      "epoch": 0.7156394608652709,
      "grad_norm": 0.034791167825460434,
      "learning_rate": 8.10821335769026e-06,
      "loss": 0.0021,
      "step": 111580
    },
    {
      "epoch": 0.7157035977590569,
      "grad_norm": 0.19611859321594238,
      "learning_rate": 8.107774922028248e-06,
      "loss": 0.0028,
      "step": 111590
    },
    {
      "epoch": 0.715767734652843,
      "grad_norm": 0.410812646150589,
      "learning_rate": 8.107336447423594e-06,
      "loss": 0.0026,
      "step": 111600
    },
    {
      "epoch": 0.7158318715466291,
      "grad_norm": 0.1915685534477234,
      "learning_rate": 8.106897933881794e-06,
      "loss": 0.0038,
      "step": 111610
    },
    {
      "epoch": 0.7158960084404152,
      "grad_norm": 0.31799638271331787,
      "learning_rate": 8.106459381408342e-06,
      "loss": 0.0029,
      "step": 111620
    },
    {
      "epoch": 0.7159601453342013,
      "grad_norm": 0.061092738062143326,
      "learning_rate": 8.106020790008732e-06,
      "loss": 0.0037,
      "step": 111630
    },
    {
      "epoch": 0.7160242822279874,
      "grad_norm": 0.1637531816959381,
      "learning_rate": 8.105582159688465e-06,
      "loss": 0.0091,
      "step": 111640
    },
    {
      "epoch": 0.7160884191217736,
      "grad_norm": 0.1069796234369278,
      "learning_rate": 8.105143490453031e-06,
      "loss": 0.0021,
      "step": 111650
    },
    {
      "epoch": 0.7161525560155596,
      "grad_norm": 0.009329847991466522,
      "learning_rate": 8.104704782307932e-06,
      "loss": 0.0038,
      "step": 111660
    },
    {
      "epoch": 0.7162166929093458,
      "grad_norm": 0.04874225705862045,
      "learning_rate": 8.104266035258663e-06,
      "loss": 0.0027,
      "step": 111670
    },
    {
      "epoch": 0.7162808298031318,
      "grad_norm": 0.04451783001422882,
      "learning_rate": 8.10382724931072e-06,
      "loss": 0.0025,
      "step": 111680
    },
    {
      "epoch": 0.7163449666969179,
      "grad_norm": 0.0949191078543663,
      "learning_rate": 8.103388424469604e-06,
      "loss": 0.0015,
      "step": 111690
    },
    {
      "epoch": 0.716409103590704,
      "grad_norm": 0.16580696403980255,
      "learning_rate": 8.102949560740816e-06,
      "loss": 0.002,
      "step": 111700
    },
    {
      "epoch": 0.7164732404844901,
      "grad_norm": 0.08396851271390915,
      "learning_rate": 8.10251065812985e-06,
      "loss": 0.002,
      "step": 111710
    },
    {
      "epoch": 0.7165373773782762,
      "grad_norm": 0.14391309022903442,
      "learning_rate": 8.102071716642209e-06,
      "loss": 0.0032,
      "step": 111720
    },
    {
      "epoch": 0.7166015142720623,
      "grad_norm": 0.1950514167547226,
      "learning_rate": 8.10163273628339e-06,
      "loss": 0.004,
      "step": 111730
    },
    {
      "epoch": 0.7166656511658483,
      "grad_norm": 0.09286870807409286,
      "learning_rate": 8.101193717058898e-06,
      "loss": 0.0039,
      "step": 111740
    },
    {
      "epoch": 0.7167297880596345,
      "grad_norm": 0.2724185585975647,
      "learning_rate": 8.100754658974233e-06,
      "loss": 0.0032,
      "step": 111750
    },
    {
      "epoch": 0.7167939249534205,
      "grad_norm": 0.24928061664104462,
      "learning_rate": 8.100315562034896e-06,
      "loss": 0.004,
      "step": 111760
    },
    {
      "epoch": 0.7168580618472067,
      "grad_norm": 0.0693969577550888,
      "learning_rate": 8.099876426246387e-06,
      "loss": 0.0032,
      "step": 111770
    },
    {
      "epoch": 0.7169221987409927,
      "grad_norm": 0.06157182157039642,
      "learning_rate": 8.099437251614212e-06,
      "loss": 0.002,
      "step": 111780
    },
    {
      "epoch": 0.7169863356347789,
      "grad_norm": 0.0845608115196228,
      "learning_rate": 8.098998038143873e-06,
      "loss": 0.0017,
      "step": 111790
    },
    {
      "epoch": 0.717050472528565,
      "grad_norm": 0.23274195194244385,
      "learning_rate": 8.098558785840876e-06,
      "loss": 0.0021,
      "step": 111800
    },
    {
      "epoch": 0.7171146094223511,
      "grad_norm": 0.16222944855690002,
      "learning_rate": 8.098119494710721e-06,
      "loss": 0.0035,
      "step": 111810
    },
    {
      "epoch": 0.7171787463161372,
      "grad_norm": 0.20074084401130676,
      "learning_rate": 8.097680164758914e-06,
      "loss": 0.0028,
      "step": 111820
    },
    {
      "epoch": 0.7172428832099232,
      "grad_norm": 0.07163853198289871,
      "learning_rate": 8.09724079599096e-06,
      "loss": 0.0022,
      "step": 111830
    },
    {
      "epoch": 0.7173070201037094,
      "grad_norm": 0.15048721432685852,
      "learning_rate": 8.096801388412368e-06,
      "loss": 0.0042,
      "step": 111840
    },
    {
      "epoch": 0.7173711569974954,
      "grad_norm": 0.054579176008701324,
      "learning_rate": 8.09636194202864e-06,
      "loss": 0.0018,
      "step": 111850
    },
    {
      "epoch": 0.7174352938912816,
      "grad_norm": 0.2996588945388794,
      "learning_rate": 8.095922456845284e-06,
      "loss": 0.0032,
      "step": 111860
    },
    {
      "epoch": 0.7174994307850676,
      "grad_norm": 0.11844748258590698,
      "learning_rate": 8.095482932867807e-06,
      "loss": 0.0027,
      "step": 111870
    },
    {
      "epoch": 0.7175635676788538,
      "grad_norm": 0.03289315477013588,
      "learning_rate": 8.095043370101716e-06,
      "loss": 0.0019,
      "step": 111880
    },
    {
      "epoch": 0.7176277045726398,
      "grad_norm": 0.1594281643629074,
      "learning_rate": 8.094603768552521e-06,
      "loss": 0.0033,
      "step": 111890
    },
    {
      "epoch": 0.717691841466426,
      "grad_norm": 0.054819926619529724,
      "learning_rate": 8.094164128225727e-06,
      "loss": 0.0039,
      "step": 111900
    },
    {
      "epoch": 0.717755978360212,
      "grad_norm": 0.058008257299661636,
      "learning_rate": 8.093724449126846e-06,
      "loss": 0.005,
      "step": 111910
    },
    {
      "epoch": 0.7178201152539981,
      "grad_norm": 0.041709840297698975,
      "learning_rate": 8.093284731261387e-06,
      "loss": 0.0018,
      "step": 111920
    },
    {
      "epoch": 0.7178842521477843,
      "grad_norm": 0.07255569100379944,
      "learning_rate": 8.09284497463486e-06,
      "loss": 0.0026,
      "step": 111930
    },
    {
      "epoch": 0.7179483890415703,
      "grad_norm": 0.19830240309238434,
      "learning_rate": 8.092405179252774e-06,
      "loss": 0.0047,
      "step": 111940
    },
    {
      "epoch": 0.7180125259353565,
      "grad_norm": 0.12875518202781677,
      "learning_rate": 8.091965345120641e-06,
      "loss": 0.0029,
      "step": 111950
    },
    {
      "epoch": 0.7180766628291425,
      "grad_norm": 0.27468159794807434,
      "learning_rate": 8.091525472243972e-06,
      "loss": 0.0023,
      "step": 111960
    },
    {
      "epoch": 0.7181407997229287,
      "grad_norm": 0.1361653357744217,
      "learning_rate": 8.091085560628282e-06,
      "loss": 0.0026,
      "step": 111970
    },
    {
      "epoch": 0.7182049366167147,
      "grad_norm": 0.17453205585479736,
      "learning_rate": 8.090645610279078e-06,
      "loss": 0.0039,
      "step": 111980
    },
    {
      "epoch": 0.7182690735105008,
      "grad_norm": 0.053696081042289734,
      "learning_rate": 8.090205621201878e-06,
      "loss": 0.0025,
      "step": 111990
    },
    {
      "epoch": 0.7183332104042869,
      "grad_norm": 0.30712389945983887,
      "learning_rate": 8.089765593402192e-06,
      "loss": 0.0024,
      "step": 112000
    },
    {
      "epoch": 0.718397347298073,
      "grad_norm": 0.3034997284412384,
      "learning_rate": 8.089325526885534e-06,
      "loss": 0.006,
      "step": 112010
    },
    {
      "epoch": 0.7184614841918591,
      "grad_norm": 0.2254328578710556,
      "learning_rate": 8.08888542165742e-06,
      "loss": 0.0028,
      "step": 112020
    },
    {
      "epoch": 0.7185256210856452,
      "grad_norm": 0.16973623633384705,
      "learning_rate": 8.088445277723362e-06,
      "loss": 0.0026,
      "step": 112030
    },
    {
      "epoch": 0.7185897579794313,
      "grad_norm": 0.13380980491638184,
      "learning_rate": 8.08800509508888e-06,
      "loss": 0.0021,
      "step": 112040
    },
    {
      "epoch": 0.7186538948732174,
      "grad_norm": 0.139179527759552,
      "learning_rate": 8.087564873759486e-06,
      "loss": 0.0025,
      "step": 112050
    },
    {
      "epoch": 0.7187180317670034,
      "grad_norm": 0.06634630262851715,
      "learning_rate": 8.087124613740698e-06,
      "loss": 0.0041,
      "step": 112060
    },
    {
      "epoch": 0.7187821686607896,
      "grad_norm": 0.30491846799850464,
      "learning_rate": 8.086684315038033e-06,
      "loss": 0.0032,
      "step": 112070
    },
    {
      "epoch": 0.7188463055545757,
      "grad_norm": 0.026031427085399628,
      "learning_rate": 8.086243977657005e-06,
      "loss": 0.0028,
      "step": 112080
    },
    {
      "epoch": 0.7189104424483618,
      "grad_norm": 0.19404006004333496,
      "learning_rate": 8.085803601603138e-06,
      "loss": 0.0027,
      "step": 112090
    },
    {
      "epoch": 0.7189745793421479,
      "grad_norm": 0.06454899907112122,
      "learning_rate": 8.085363186881942e-06,
      "loss": 0.002,
      "step": 112100
    },
    {
      "epoch": 0.719038716235934,
      "grad_norm": 0.15043289959430695,
      "learning_rate": 8.084922733498943e-06,
      "loss": 0.0029,
      "step": 112110
    },
    {
      "epoch": 0.7191028531297201,
      "grad_norm": 0.09538527578115463,
      "learning_rate": 8.084482241459658e-06,
      "loss": 0.0027,
      "step": 112120
    },
    {
      "epoch": 0.7191669900235061,
      "grad_norm": 0.3106303811073303,
      "learning_rate": 8.084041710769606e-06,
      "loss": 0.0043,
      "step": 112130
    },
    {
      "epoch": 0.7192311269172923,
      "grad_norm": 0.08692137897014618,
      "learning_rate": 8.083601141434305e-06,
      "loss": 0.0021,
      "step": 112140
    },
    {
      "epoch": 0.7192952638110783,
      "grad_norm": 0.12955030798912048,
      "learning_rate": 8.08316053345928e-06,
      "loss": 0.0018,
      "step": 112150
    },
    {
      "epoch": 0.7193594007048645,
      "grad_norm": 0.24629080295562744,
      "learning_rate": 8.082719886850048e-06,
      "loss": 0.0037,
      "step": 112160
    },
    {
      "epoch": 0.7194235375986505,
      "grad_norm": 0.035296481102705,
      "learning_rate": 8.082279201612135e-06,
      "loss": 0.0033,
      "step": 112170
    },
    {
      "epoch": 0.7194876744924367,
      "grad_norm": 0.13728514313697815,
      "learning_rate": 8.08183847775106e-06,
      "loss": 0.0037,
      "step": 112180
    },
    {
      "epoch": 0.7195518113862227,
      "grad_norm": 0.2915211021900177,
      "learning_rate": 8.081397715272346e-06,
      "loss": 0.0028,
      "step": 112190
    },
    {
      "epoch": 0.7196159482800089,
      "grad_norm": 0.13166002929210663,
      "learning_rate": 8.080956914181515e-06,
      "loss": 0.0044,
      "step": 112200
    },
    {
      "epoch": 0.719680085173795,
      "grad_norm": 0.24949447810649872,
      "learning_rate": 8.080516074484093e-06,
      "loss": 0.0035,
      "step": 112210
    },
    {
      "epoch": 0.719744222067581,
      "grad_norm": 0.2722225487232208,
      "learning_rate": 8.080075196185605e-06,
      "loss": 0.003,
      "step": 112220
    },
    {
      "epoch": 0.7198083589613672,
      "grad_norm": 0.03342172130942345,
      "learning_rate": 8.07963427929157e-06,
      "loss": 0.0033,
      "step": 112230
    },
    {
      "epoch": 0.7198724958551532,
      "grad_norm": 0.0899524912238121,
      "learning_rate": 8.079193323807519e-06,
      "loss": 0.0039,
      "step": 112240
    },
    {
      "epoch": 0.7199366327489394,
      "grad_norm": 0.016616996377706528,
      "learning_rate": 8.078752329738974e-06,
      "loss": 0.0016,
      "step": 112250
    },
    {
      "epoch": 0.7200007696427254,
      "grad_norm": 0.10823524743318558,
      "learning_rate": 8.078311297091462e-06,
      "loss": 0.0034,
      "step": 112260
    },
    {
      "epoch": 0.7200649065365116,
      "grad_norm": 0.28435730934143066,
      "learning_rate": 8.07787022587051e-06,
      "loss": 0.0041,
      "step": 112270
    },
    {
      "epoch": 0.7201290434302976,
      "grad_norm": 0.09790398925542831,
      "learning_rate": 8.077429116081643e-06,
      "loss": 0.0059,
      "step": 112280
    },
    {
      "epoch": 0.7201931803240837,
      "grad_norm": 0.20814689993858337,
      "learning_rate": 8.07698796773039e-06,
      "loss": 0.0034,
      "step": 112290
    },
    {
      "epoch": 0.7202573172178698,
      "grad_norm": 0.2655559182167053,
      "learning_rate": 8.076546780822281e-06,
      "loss": 0.0024,
      "step": 112300
    },
    {
      "epoch": 0.7203214541116559,
      "grad_norm": 0.25793513655662537,
      "learning_rate": 8.07610555536284e-06,
      "loss": 0.0045,
      "step": 112310
    },
    {
      "epoch": 0.720385591005442,
      "grad_norm": 0.1332065910100937,
      "learning_rate": 8.075664291357598e-06,
      "loss": 0.004,
      "step": 112320
    },
    {
      "epoch": 0.7204497278992281,
      "grad_norm": 0.11241687834262848,
      "learning_rate": 8.075222988812085e-06,
      "loss": 0.0024,
      "step": 112330
    },
    {
      "epoch": 0.7205138647930142,
      "grad_norm": 0.1313232034444809,
      "learning_rate": 8.07478164773183e-06,
      "loss": 0.0023,
      "step": 112340
    },
    {
      "epoch": 0.7205780016868003,
      "grad_norm": 0.046554699540138245,
      "learning_rate": 8.074340268122363e-06,
      "loss": 0.002,
      "step": 112350
    },
    {
      "epoch": 0.7206421385805865,
      "grad_norm": 0.10994312912225723,
      "learning_rate": 8.073898849989214e-06,
      "loss": 0.0019,
      "step": 112360
    },
    {
      "epoch": 0.7207062754743725,
      "grad_norm": 0.09446525573730469,
      "learning_rate": 8.073457393337918e-06,
      "loss": 0.0024,
      "step": 112370
    },
    {
      "epoch": 0.7207704123681586,
      "grad_norm": 0.5231614708900452,
      "learning_rate": 8.073015898174003e-06,
      "loss": 0.0024,
      "step": 112380
    },
    {
      "epoch": 0.7208345492619447,
      "grad_norm": 0.1255865842103958,
      "learning_rate": 8.072574364503002e-06,
      "loss": 0.0023,
      "step": 112390
    },
    {
      "epoch": 0.7208986861557308,
      "grad_norm": 0.1481911838054657,
      "learning_rate": 8.07213279233045e-06,
      "loss": 0.002,
      "step": 112400
    },
    {
      "epoch": 0.7209628230495169,
      "grad_norm": 0.25029078125953674,
      "learning_rate": 8.071691181661879e-06,
      "loss": 0.0025,
      "step": 112410
    },
    {
      "epoch": 0.721026959943303,
      "grad_norm": 0.18313568830490112,
      "learning_rate": 8.071249532502818e-06,
      "loss": 0.0017,
      "step": 112420
    },
    {
      "epoch": 0.721091096837089,
      "grad_norm": 0.3427973687648773,
      "learning_rate": 8.070807844858808e-06,
      "loss": 0.0046,
      "step": 112430
    },
    {
      "epoch": 0.7211552337308752,
      "grad_norm": 0.10546506941318512,
      "learning_rate": 8.070366118735381e-06,
      "loss": 0.0018,
      "step": 112440
    },
    {
      "epoch": 0.7212193706246612,
      "grad_norm": 0.12715408205986023,
      "learning_rate": 8.069924354138073e-06,
      "loss": 0.0026,
      "step": 112450
    },
    {
      "epoch": 0.7212835075184474,
      "grad_norm": 0.1146603673696518,
      "learning_rate": 8.069482551072419e-06,
      "loss": 0.0025,
      "step": 112460
    },
    {
      "epoch": 0.7213476444122334,
      "grad_norm": 0.07918599247932434,
      "learning_rate": 8.069040709543953e-06,
      "loss": 0.0016,
      "step": 112470
    },
    {
      "epoch": 0.7214117813060196,
      "grad_norm": 0.24079379439353943,
      "learning_rate": 8.068598829558216e-06,
      "loss": 0.0017,
      "step": 112480
    },
    {
      "epoch": 0.7214759181998056,
      "grad_norm": 0.041503965854644775,
      "learning_rate": 8.06815691112074e-06,
      "loss": 0.0017,
      "step": 112490
    },
    {
      "epoch": 0.7215400550935918,
      "grad_norm": 0.04950868710875511,
      "learning_rate": 8.067714954237066e-06,
      "loss": 0.0029,
      "step": 112500
    },
    {
      "epoch": 0.7216041919873779,
      "grad_norm": 0.2447650134563446,
      "learning_rate": 8.067272958912732e-06,
      "loss": 0.0034,
      "step": 112510
    },
    {
      "epoch": 0.721668328881164,
      "grad_norm": 0.07529481500387192,
      "learning_rate": 8.066830925153276e-06,
      "loss": 0.0026,
      "step": 112520
    },
    {
      "epoch": 0.7217324657749501,
      "grad_norm": 0.17679598927497864,
      "learning_rate": 8.066388852964235e-06,
      "loss": 0.003,
      "step": 112530
    },
    {
      "epoch": 0.7217966026687361,
      "grad_norm": 0.05174412950873375,
      "learning_rate": 8.06594674235115e-06,
      "loss": 0.002,
      "step": 112540
    },
    {
      "epoch": 0.7218607395625223,
      "grad_norm": 0.09445449709892273,
      "learning_rate": 8.065504593319561e-06,
      "loss": 0.0034,
      "step": 112550
    },
    {
      "epoch": 0.7219248764563083,
      "grad_norm": 0.1533842384815216,
      "learning_rate": 8.065062405875011e-06,
      "loss": 0.0056,
      "step": 112560
    },
    {
      "epoch": 0.7219890133500945,
      "grad_norm": 0.07002757489681244,
      "learning_rate": 8.064620180023037e-06,
      "loss": 0.002,
      "step": 112570
    },
    {
      "epoch": 0.7220531502438805,
      "grad_norm": 0.05978749319911003,
      "learning_rate": 8.064177915769182e-06,
      "loss": 0.0019,
      "step": 112580
    },
    {
      "epoch": 0.7221172871376667,
      "grad_norm": 0.14716807007789612,
      "learning_rate": 8.063735613118988e-06,
      "loss": 0.0026,
      "step": 112590
    },
    {
      "epoch": 0.7221814240314527,
      "grad_norm": 0.015161024406552315,
      "learning_rate": 8.063293272077996e-06,
      "loss": 0.003,
      "step": 112600
    },
    {
      "epoch": 0.7222455609252388,
      "grad_norm": 0.23610588908195496,
      "learning_rate": 8.062850892651752e-06,
      "loss": 0.0028,
      "step": 112610
    },
    {
      "epoch": 0.7223096978190249,
      "grad_norm": 0.1815919280052185,
      "learning_rate": 8.062408474845796e-06,
      "loss": 0.0025,
      "step": 112620
    },
    {
      "epoch": 0.722373834712811,
      "grad_norm": 0.19692429900169373,
      "learning_rate": 8.061966018665672e-06,
      "loss": 0.0037,
      "step": 112630
    },
    {
      "epoch": 0.7224379716065972,
      "grad_norm": 0.16444410383701324,
      "learning_rate": 8.061523524116927e-06,
      "loss": 0.005,
      "step": 112640
    },
    {
      "epoch": 0.7225021085003832,
      "grad_norm": 0.11123505234718323,
      "learning_rate": 8.061080991205104e-06,
      "loss": 0.0019,
      "step": 112650
    },
    {
      "epoch": 0.7225662453941694,
      "grad_norm": 0.11508255451917648,
      "learning_rate": 8.06063841993575e-06,
      "loss": 0.0027,
      "step": 112660
    },
    {
      "epoch": 0.7226303822879554,
      "grad_norm": 0.06001582741737366,
      "learning_rate": 8.060195810314407e-06,
      "loss": 0.0047,
      "step": 112670
    },
    {
      "epoch": 0.7226945191817415,
      "grad_norm": 0.763580322265625,
      "learning_rate": 8.059753162346624e-06,
      "loss": 0.0057,
      "step": 112680
    },
    {
      "epoch": 0.7227586560755276,
      "grad_norm": 0.34137997031211853,
      "learning_rate": 8.059310476037947e-06,
      "loss": 0.0038,
      "step": 112690
    },
    {
      "epoch": 0.7228227929693137,
      "grad_norm": 0.21813026070594788,
      "learning_rate": 8.058867751393923e-06,
      "loss": 0.0043,
      "step": 112700
    },
    {
      "epoch": 0.7228869298630998,
      "grad_norm": 0.23461997509002686,
      "learning_rate": 8.058424988420101e-06,
      "loss": 0.0034,
      "step": 112710
    },
    {
      "epoch": 0.7229510667568859,
      "grad_norm": 0.16151976585388184,
      "learning_rate": 8.057982187122027e-06,
      "loss": 0.004,
      "step": 112720
    },
    {
      "epoch": 0.723015203650672,
      "grad_norm": 0.09632989764213562,
      "learning_rate": 8.057539347505252e-06,
      "loss": 0.0033,
      "step": 112730
    },
    {
      "epoch": 0.7230793405444581,
      "grad_norm": 0.2594684958457947,
      "learning_rate": 8.057096469575324e-06,
      "loss": 0.0023,
      "step": 112740
    },
    {
      "epoch": 0.7231434774382441,
      "grad_norm": 0.1471938192844391,
      "learning_rate": 8.05665355333779e-06,
      "loss": 0.0052,
      "step": 112750
    },
    {
      "epoch": 0.7232076143320303,
      "grad_norm": 0.1528429388999939,
      "learning_rate": 8.056210598798204e-06,
      "loss": 0.0022,
      "step": 112760
    },
    {
      "epoch": 0.7232717512258163,
      "grad_norm": 0.14047491550445557,
      "learning_rate": 8.055767605962115e-06,
      "loss": 0.003,
      "step": 112770
    },
    {
      "epoch": 0.7233358881196025,
      "grad_norm": 0.10037308186292648,
      "learning_rate": 8.055324574835072e-06,
      "loss": 0.0025,
      "step": 112780
    },
    {
      "epoch": 0.7234000250133886,
      "grad_norm": 0.05584083870053291,
      "learning_rate": 8.054881505422631e-06,
      "loss": 0.0032,
      "step": 112790
    },
    {
      "epoch": 0.7234641619071747,
      "grad_norm": 0.18206900358200073,
      "learning_rate": 8.05443839773034e-06,
      "loss": 0.0024,
      "step": 112800
    },
    {
      "epoch": 0.7235282988009608,
      "grad_norm": 0.01469647977501154,
      "learning_rate": 8.053995251763753e-06,
      "loss": 0.0038,
      "step": 112810
    },
    {
      "epoch": 0.7235924356947468,
      "grad_norm": 0.021825265139341354,
      "learning_rate": 8.053552067528423e-06,
      "loss": 0.0036,
      "step": 112820
    },
    {
      "epoch": 0.723656572588533,
      "grad_norm": 0.15762962400913239,
      "learning_rate": 8.053108845029905e-06,
      "loss": 0.0019,
      "step": 112830
    },
    {
      "epoch": 0.723720709482319,
      "grad_norm": 0.08991697430610657,
      "learning_rate": 8.05266558427375e-06,
      "loss": 0.0032,
      "step": 112840
    },
    {
      "epoch": 0.7237848463761052,
      "grad_norm": 0.09897933900356293,
      "learning_rate": 8.052222285265512e-06,
      "loss": 0.0024,
      "step": 112850
    },
    {
      "epoch": 0.7238489832698912,
      "grad_norm": 0.03703915700316429,
      "learning_rate": 8.05177894801075e-06,
      "loss": 0.0047,
      "step": 112860
    },
    {
      "epoch": 0.7239131201636774,
      "grad_norm": 0.20504747331142426,
      "learning_rate": 8.051335572515013e-06,
      "loss": 0.0022,
      "step": 112870
    },
    {
      "epoch": 0.7239772570574634,
      "grad_norm": 0.22929313778877258,
      "learning_rate": 8.050892158783862e-06,
      "loss": 0.0031,
      "step": 112880
    },
    {
      "epoch": 0.7240413939512496,
      "grad_norm": 0.21101884543895721,
      "learning_rate": 8.050448706822855e-06,
      "loss": 0.0045,
      "step": 112890
    },
    {
      "epoch": 0.7241055308450356,
      "grad_norm": 0.062430642545223236,
      "learning_rate": 8.050005216637543e-06,
      "loss": 0.0027,
      "step": 112900
    },
    {
      "epoch": 0.7241696677388217,
      "grad_norm": 0.035764604806900024,
      "learning_rate": 8.049561688233485e-06,
      "loss": 0.0026,
      "step": 112910
    },
    {
      "epoch": 0.7242338046326079,
      "grad_norm": 0.052385181188583374,
      "learning_rate": 8.049118121616242e-06,
      "loss": 0.0023,
      "step": 112920
    },
    {
      "epoch": 0.7242979415263939,
      "grad_norm": 0.02150088921189308,
      "learning_rate": 8.048674516791368e-06,
      "loss": 0.0021,
      "step": 112930
    },
    {
      "epoch": 0.7243620784201801,
      "grad_norm": 0.026390841230750084,
      "learning_rate": 8.048230873764422e-06,
      "loss": 0.0018,
      "step": 112940
    },
    {
      "epoch": 0.7244262153139661,
      "grad_norm": 0.19471149146556854,
      "learning_rate": 8.047787192540967e-06,
      "loss": 0.0041,
      "step": 112950
    },
    {
      "epoch": 0.7244903522077523,
      "grad_norm": 0.12749211490154266,
      "learning_rate": 8.047343473126559e-06,
      "loss": 0.0037,
      "step": 112960
    },
    {
      "epoch": 0.7245544891015383,
      "grad_norm": 0.07260242104530334,
      "learning_rate": 8.046899715526762e-06,
      "loss": 0.0027,
      "step": 112970
    },
    {
      "epoch": 0.7246186259953245,
      "grad_norm": 0.13378237187862396,
      "learning_rate": 8.046455919747131e-06,
      "loss": 0.0017,
      "step": 112980
    },
    {
      "epoch": 0.7246827628891105,
      "grad_norm": 0.15227557718753815,
      "learning_rate": 8.046012085793231e-06,
      "loss": 0.0022,
      "step": 112990
    },
    {
      "epoch": 0.7247468997828966,
      "grad_norm": 0.11496797949075699,
      "learning_rate": 8.045568213670623e-06,
      "loss": 0.0024,
      "step": 113000
    },
    {
      "epoch": 0.7248110366766827,
      "grad_norm": 0.05050666630268097,
      "learning_rate": 8.045124303384868e-06,
      "loss": 0.0017,
      "step": 113010
    },
    {
      "epoch": 0.7248751735704688,
      "grad_norm": 0.042164478451013565,
      "learning_rate": 8.04468035494153e-06,
      "loss": 0.0018,
      "step": 113020
    },
    {
      "epoch": 0.7249393104642549,
      "grad_norm": 0.09713324904441833,
      "learning_rate": 8.04423636834617e-06,
      "loss": 0.0047,
      "step": 113030
    },
    {
      "epoch": 0.725003447358041,
      "grad_norm": 0.05479344353079796,
      "learning_rate": 8.043792343604354e-06,
      "loss": 0.0027,
      "step": 113040
    },
    {
      "epoch": 0.725067584251827,
      "grad_norm": 0.029082629829645157,
      "learning_rate": 8.043348280721643e-06,
      "loss": 0.0036,
      "step": 113050
    },
    {
      "epoch": 0.7251317211456132,
      "grad_norm": 0.09260375797748566,
      "learning_rate": 8.042904179703605e-06,
      "loss": 0.002,
      "step": 113060
    },
    {
      "epoch": 0.7251958580393993,
      "grad_norm": 0.1256377398967743,
      "learning_rate": 8.042460040555802e-06,
      "loss": 0.0034,
      "step": 113070
    },
    {
      "epoch": 0.7252599949331854,
      "grad_norm": 0.07848475128412247,
      "learning_rate": 8.042015863283799e-06,
      "loss": 0.0025,
      "step": 113080
    },
    {
      "epoch": 0.7253241318269715,
      "grad_norm": 0.0060674287378787994,
      "learning_rate": 8.041571647893165e-06,
      "loss": 0.002,
      "step": 113090
    },
    {
      "epoch": 0.7253882687207576,
      "grad_norm": 0.12943817675113678,
      "learning_rate": 8.041127394389462e-06,
      "loss": 0.0026,
      "step": 113100
    },
    {
      "epoch": 0.7254524056145437,
      "grad_norm": 0.019311005249619484,
      "learning_rate": 8.040683102778262e-06,
      "loss": 0.0019,
      "step": 113110
    },
    {
      "epoch": 0.7255165425083298,
      "grad_norm": 0.23863111436367035,
      "learning_rate": 8.040238773065128e-06,
      "loss": 0.0032,
      "step": 113120
    },
    {
      "epoch": 0.7255806794021159,
      "grad_norm": 0.17583878338336945,
      "learning_rate": 8.03979440525563e-06,
      "loss": 0.0029,
      "step": 113130
    },
    {
      "epoch": 0.7256448162959019,
      "grad_norm": 0.2913358211517334,
      "learning_rate": 8.039349999355336e-06,
      "loss": 0.0042,
      "step": 113140
    },
    {
      "epoch": 0.7257089531896881,
      "grad_norm": 0.022446373477578163,
      "learning_rate": 8.038905555369814e-06,
      "loss": 0.0024,
      "step": 113150
    },
    {
      "epoch": 0.7257730900834741,
      "grad_norm": 0.1001579761505127,
      "learning_rate": 8.038461073304633e-06,
      "loss": 0.0024,
      "step": 113160
    },
    {
      "epoch": 0.7258372269772603,
      "grad_norm": 0.09300924092531204,
      "learning_rate": 8.038016553165363e-06,
      "loss": 0.0048,
      "step": 113170
    },
    {
      "epoch": 0.7259013638710463,
      "grad_norm": 0.09971919655799866,
      "learning_rate": 8.037571994957576e-06,
      "loss": 0.0013,
      "step": 113180
    },
    {
      "epoch": 0.7259655007648325,
      "grad_norm": 0.1383122205734253,
      "learning_rate": 8.037127398686838e-06,
      "loss": 0.0025,
      "step": 113190
    },
    {
      "epoch": 0.7260296376586186,
      "grad_norm": 0.179524227976799,
      "learning_rate": 8.036682764358725e-06,
      "loss": 0.002,
      "step": 113200
    },
    {
      "epoch": 0.7260937745524046,
      "grad_norm": 0.11412809044122696,
      "learning_rate": 8.036238091978808e-06,
      "loss": 0.0043,
      "step": 113210
    },
    {
      "epoch": 0.7261579114461908,
      "grad_norm": 0.08796469867229462,
      "learning_rate": 8.035793381552655e-06,
      "loss": 0.0017,
      "step": 113220
    },
    {
      "epoch": 0.7262220483399768,
      "grad_norm": 0.05354146286845207,
      "learning_rate": 8.035348633085842e-06,
      "loss": 0.002,
      "step": 113230
    },
    {
      "epoch": 0.726286185233763,
      "grad_norm": 0.13761520385742188,
      "learning_rate": 8.034903846583942e-06,
      "loss": 0.0146,
      "step": 113240
    },
    {
      "epoch": 0.726350322127549,
      "grad_norm": 3.3249149322509766,
      "learning_rate": 8.034459022052527e-06,
      "loss": 0.0147,
      "step": 113250
    },
    {
      "epoch": 0.7264144590213352,
      "grad_norm": 0.0731549933552742,
      "learning_rate": 8.034014159497174e-06,
      "loss": 0.0027,
      "step": 113260
    },
    {
      "epoch": 0.7264785959151212,
      "grad_norm": 0.32172027230262756,
      "learning_rate": 8.033569258923453e-06,
      "loss": 0.0031,
      "step": 113270
    },
    {
      "epoch": 0.7265427328089074,
      "grad_norm": 0.18017232418060303,
      "learning_rate": 8.033124320336942e-06,
      "loss": 0.0066,
      "step": 113280
    },
    {
      "epoch": 0.7266068697026934,
      "grad_norm": 0.08678385615348816,
      "learning_rate": 8.032679343743215e-06,
      "loss": 0.0018,
      "step": 113290
    },
    {
      "epoch": 0.7266710065964795,
      "grad_norm": 0.17082479596138,
      "learning_rate": 8.032234329147849e-06,
      "loss": 0.0031,
      "step": 113300
    },
    {
      "epoch": 0.7267351434902656,
      "grad_norm": 0.017062757164239883,
      "learning_rate": 8.031789276556418e-06,
      "loss": 0.0027,
      "step": 113310
    },
    {
      "epoch": 0.7267992803840517,
      "grad_norm": 0.38885724544525146,
      "learning_rate": 8.031344185974504e-06,
      "loss": 0.0017,
      "step": 113320
    },
    {
      "epoch": 0.7268634172778378,
      "grad_norm": 0.08904733508825302,
      "learning_rate": 8.030899057407679e-06,
      "loss": 0.003,
      "step": 113330
    },
    {
      "epoch": 0.7269275541716239,
      "grad_norm": 0.07714398205280304,
      "learning_rate": 8.030453890861522e-06,
      "loss": 0.0022,
      "step": 113340
    },
    {
      "epoch": 0.7269916910654101,
      "grad_norm": 0.206559419631958,
      "learning_rate": 8.030008686341613e-06,
      "loss": 0.0018,
      "step": 113350
    },
    {
      "epoch": 0.7270558279591961,
      "grad_norm": 0.21958573162555695,
      "learning_rate": 8.029563443853529e-06,
      "loss": 0.0032,
      "step": 113360
    },
    {
      "epoch": 0.7271199648529822,
      "grad_norm": 0.12396115809679031,
      "learning_rate": 8.029118163402849e-06,
      "loss": 0.0028,
      "step": 113370
    },
    {
      "epoch": 0.7271841017467683,
      "grad_norm": 0.1274358332157135,
      "learning_rate": 8.028672844995155e-06,
      "loss": 0.003,
      "step": 113380
    },
    {
      "epoch": 0.7272482386405544,
      "grad_norm": 0.13518588244915009,
      "learning_rate": 8.028227488636024e-06,
      "loss": 0.0015,
      "step": 113390
    },
    {
      "epoch": 0.7273123755343405,
      "grad_norm": 0.11445048451423645,
      "learning_rate": 8.027782094331041e-06,
      "loss": 0.0042,
      "step": 113400
    },
    {
      "epoch": 0.7273765124281266,
      "grad_norm": 0.10959281027317047,
      "learning_rate": 8.027336662085784e-06,
      "loss": 0.0059,
      "step": 113410
    },
    {
      "epoch": 0.7274406493219127,
      "grad_norm": 0.06854724884033203,
      "learning_rate": 8.026891191905832e-06,
      "loss": 0.003,
      "step": 113420
    },
    {
      "epoch": 0.7275047862156988,
      "grad_norm": 0.15871262550354004,
      "learning_rate": 8.026445683796772e-06,
      "loss": 0.0026,
      "step": 113430
    },
    {
      "epoch": 0.7275689231094848,
      "grad_norm": 0.058151207864284515,
      "learning_rate": 8.026000137764185e-06,
      "loss": 0.0024,
      "step": 113440
    },
    {
      "epoch": 0.727633060003271,
      "grad_norm": 0.19693368673324585,
      "learning_rate": 8.025554553813654e-06,
      "loss": 0.0028,
      "step": 113450
    },
    {
      "epoch": 0.727697196897057,
      "grad_norm": 0.15146328508853912,
      "learning_rate": 8.02510893195076e-06,
      "loss": 0.0017,
      "step": 113460
    },
    {
      "epoch": 0.7277613337908432,
      "grad_norm": 0.023606721311807632,
      "learning_rate": 8.024663272181092e-06,
      "loss": 0.0025,
      "step": 113470
    },
    {
      "epoch": 0.7278254706846293,
      "grad_norm": 0.15048794448375702,
      "learning_rate": 8.024217574510229e-06,
      "loss": 0.003,
      "step": 113480
    },
    {
      "epoch": 0.7278896075784154,
      "grad_norm": 0.10539175570011139,
      "learning_rate": 8.02377183894376e-06,
      "loss": 0.003,
      "step": 113490
    },
    {
      "epoch": 0.7279537444722015,
      "grad_norm": 0.2851220667362213,
      "learning_rate": 8.023326065487267e-06,
      "loss": 0.0032,
      "step": 113500
    },
    {
      "epoch": 0.7280178813659876,
      "grad_norm": 0.11118152737617493,
      "learning_rate": 8.02288025414634e-06,
      "loss": 0.0023,
      "step": 113510
    },
    {
      "epoch": 0.7280820182597737,
      "grad_norm": 0.08770406246185303,
      "learning_rate": 8.022434404926563e-06,
      "loss": 0.0021,
      "step": 113520
    },
    {
      "epoch": 0.7281461551535597,
      "grad_norm": 0.1460132896900177,
      "learning_rate": 8.02198851783352e-06,
      "loss": 0.0024,
      "step": 113530
    },
    {
      "epoch": 0.7282102920473459,
      "grad_norm": 0.1536615639925003,
      "learning_rate": 8.021542592872804e-06,
      "loss": 0.0023,
      "step": 113540
    },
    {
      "epoch": 0.7282744289411319,
      "grad_norm": 0.07052811980247498,
      "learning_rate": 8.021096630049999e-06,
      "loss": 0.0021,
      "step": 113550
    },
    {
      "epoch": 0.7283385658349181,
      "grad_norm": 0.07707761973142624,
      "learning_rate": 8.020650629370692e-06,
      "loss": 0.0033,
      "step": 113560
    },
    {
      "epoch": 0.7284027027287041,
      "grad_norm": 0.19061174988746643,
      "learning_rate": 8.020204590840474e-06,
      "loss": 0.0025,
      "step": 113570
    },
    {
      "epoch": 0.7284668396224903,
      "grad_norm": 0.023339703679084778,
      "learning_rate": 8.019758514464936e-06,
      "loss": 0.0043,
      "step": 113580
    },
    {
      "epoch": 0.7285309765162763,
      "grad_norm": 0.1458740234375,
      "learning_rate": 8.019312400249665e-06,
      "loss": 0.0026,
      "step": 113590
    },
    {
      "epoch": 0.7285951134100624,
      "grad_norm": 0.1815797984600067,
      "learning_rate": 8.018866248200253e-06,
      "loss": 0.0042,
      "step": 113600
    },
    {
      "epoch": 0.7286592503038485,
      "grad_norm": 0.09067380428314209,
      "learning_rate": 8.018420058322288e-06,
      "loss": 0.0026,
      "step": 113610
    },
    {
      "epoch": 0.7287233871976346,
      "grad_norm": 0.03138352185487747,
      "learning_rate": 8.017973830621363e-06,
      "loss": 0.0031,
      "step": 113620
    },
    {
      "epoch": 0.7287875240914208,
      "grad_norm": 0.11717572808265686,
      "learning_rate": 8.017527565103068e-06,
      "loss": 0.0025,
      "step": 113630
    },
    {
      "epoch": 0.7288516609852068,
      "grad_norm": 0.12940505146980286,
      "learning_rate": 8.017081261772996e-06,
      "loss": 0.0051,
      "step": 113640
    },
    {
      "epoch": 0.728915797878993,
      "grad_norm": 0.1723572462797165,
      "learning_rate": 8.016634920636741e-06,
      "loss": 0.0037,
      "step": 113650
    },
    {
      "epoch": 0.728979934772779,
      "grad_norm": 0.07741840928792953,
      "learning_rate": 8.016188541699894e-06,
      "loss": 0.0044,
      "step": 113660
    },
    {
      "epoch": 0.7290440716665652,
      "grad_norm": 0.018096577376127243,
      "learning_rate": 8.015742124968048e-06,
      "loss": 0.0023,
      "step": 113670
    },
    {
      "epoch": 0.7291082085603512,
      "grad_norm": 0.10388552397489548,
      "learning_rate": 8.015295670446799e-06,
      "loss": 0.0058,
      "step": 113680
    },
    {
      "epoch": 0.7291723454541373,
      "grad_norm": 0.1819618195295334,
      "learning_rate": 8.01484917814174e-06,
      "loss": 0.0025,
      "step": 113690
    },
    {
      "epoch": 0.7292364823479234,
      "grad_norm": 0.003572376910597086,
      "learning_rate": 8.014402648058465e-06,
      "loss": 0.0016,
      "step": 113700
    },
    {
      "epoch": 0.7293006192417095,
      "grad_norm": 0.054669469594955444,
      "learning_rate": 8.013956080202571e-06,
      "loss": 0.0023,
      "step": 113710
    },
    {
      "epoch": 0.7293647561354956,
      "grad_norm": 0.09258153289556503,
      "learning_rate": 8.013509474579653e-06,
      "loss": 0.0068,
      "step": 113720
    },
    {
      "epoch": 0.7294288930292817,
      "grad_norm": 0.17361032962799072,
      "learning_rate": 8.013062831195309e-06,
      "loss": 0.0027,
      "step": 113730
    },
    {
      "epoch": 0.7294930299230677,
      "grad_norm": 0.165913388133049,
      "learning_rate": 8.012616150055133e-06,
      "loss": 0.0028,
      "step": 113740
    },
    {
      "epoch": 0.7295571668168539,
      "grad_norm": 0.33771705627441406,
      "learning_rate": 8.012169431164725e-06,
      "loss": 0.006,
      "step": 113750
    },
    {
      "epoch": 0.7296213037106399,
      "grad_norm": 0.06362960487604141,
      "learning_rate": 8.01172267452968e-06,
      "loss": 0.0036,
      "step": 113760
    },
    {
      "epoch": 0.7296854406044261,
      "grad_norm": 0.06443426012992859,
      "learning_rate": 8.011275880155599e-06,
      "loss": 0.002,
      "step": 113770
    },
    {
      "epoch": 0.7297495774982122,
      "grad_norm": 0.14011427760124207,
      "learning_rate": 8.010829048048075e-06,
      "loss": 0.0024,
      "step": 113780
    },
    {
      "epoch": 0.7298137143919983,
      "grad_norm": 0.165731742978096,
      "learning_rate": 8.010382178212714e-06,
      "loss": 0.0013,
      "step": 113790
    },
    {
      "epoch": 0.7298778512857844,
      "grad_norm": 0.04552525281906128,
      "learning_rate": 8.009935270655114e-06,
      "loss": 0.002,
      "step": 113800
    },
    {
      "epoch": 0.7299419881795705,
      "grad_norm": 0.1588314175605774,
      "learning_rate": 8.009488325380872e-06,
      "loss": 0.0055,
      "step": 113810
    },
    {
      "epoch": 0.7300061250733566,
      "grad_norm": 0.045454684644937515,
      "learning_rate": 8.00904134239559e-06,
      "loss": 0.0016,
      "step": 113820
    },
    {
      "epoch": 0.7300702619671426,
      "grad_norm": 0.25208306312561035,
      "learning_rate": 8.008594321704872e-06,
      "loss": 0.0024,
      "step": 113830
    },
    {
      "epoch": 0.7301343988609288,
      "grad_norm": 0.08057577162981033,
      "learning_rate": 8.008147263314316e-06,
      "loss": 0.002,
      "step": 113840
    },
    {
      "epoch": 0.7301985357547148,
      "grad_norm": 0.11513067781925201,
      "learning_rate": 8.007700167229525e-06,
      "loss": 0.0016,
      "step": 113850
    },
    {
      "epoch": 0.730262672648501,
      "grad_norm": 0.16625617444515228,
      "learning_rate": 8.007253033456099e-06,
      "loss": 0.0022,
      "step": 113860
    },
    {
      "epoch": 0.730326809542287,
      "grad_norm": 0.11144071072340012,
      "learning_rate": 8.006805861999645e-06,
      "loss": 0.0025,
      "step": 113870
    },
    {
      "epoch": 0.7303909464360732,
      "grad_norm": 0.12607555091381073,
      "learning_rate": 8.006358652865764e-06,
      "loss": 0.0046,
      "step": 113880
    },
    {
      "epoch": 0.7304550833298592,
      "grad_norm": 0.08143970370292664,
      "learning_rate": 8.005911406060062e-06,
      "loss": 0.002,
      "step": 113890
    },
    {
      "epoch": 0.7305192202236453,
      "grad_norm": 0.18797878921031952,
      "learning_rate": 8.005464121588142e-06,
      "loss": 0.0022,
      "step": 113900
    },
    {
      "epoch": 0.7305833571174315,
      "grad_norm": 0.13115228712558746,
      "learning_rate": 8.005016799455607e-06,
      "loss": 0.0029,
      "step": 113910
    },
    {
      "epoch": 0.7306474940112175,
      "grad_norm": 0.09957197308540344,
      "learning_rate": 8.004569439668064e-06,
      "loss": 0.0022,
      "step": 113920
    },
    {
      "epoch": 0.7307116309050037,
      "grad_norm": 0.055061738938093185,
      "learning_rate": 8.004122042231118e-06,
      "loss": 0.0036,
      "step": 113930
    },
    {
      "epoch": 0.7307757677987897,
      "grad_norm": 0.2100890427827835,
      "learning_rate": 8.003674607150377e-06,
      "loss": 0.0025,
      "step": 113940
    },
    {
      "epoch": 0.7308399046925759,
      "grad_norm": 0.1487421691417694,
      "learning_rate": 8.003227134431446e-06,
      "loss": 0.0041,
      "step": 113950
    },
    {
      "epoch": 0.7309040415863619,
      "grad_norm": 0.09251925349235535,
      "learning_rate": 8.002779624079933e-06,
      "loss": 0.0022,
      "step": 113960
    },
    {
      "epoch": 0.7309681784801481,
      "grad_norm": 0.08976790308952332,
      "learning_rate": 8.002332076101445e-06,
      "loss": 0.0017,
      "step": 113970
    },
    {
      "epoch": 0.7310323153739341,
      "grad_norm": 0.11529594659805298,
      "learning_rate": 8.00188449050159e-06,
      "loss": 0.0028,
      "step": 113980
    },
    {
      "epoch": 0.7310964522677202,
      "grad_norm": 0.09520257264375687,
      "learning_rate": 8.001436867285977e-06,
      "loss": 0.0019,
      "step": 113990
    },
    {
      "epoch": 0.7311605891615063,
      "grad_norm": 0.02042161114513874,
      "learning_rate": 8.000989206460215e-06,
      "loss": 0.003,
      "step": 114000
    },
    {
      "epoch": 0.7312247260552924,
      "grad_norm": 0.06751138716936111,
      "learning_rate": 8.000541508029914e-06,
      "loss": 0.0016,
      "step": 114010
    },
    {
      "epoch": 0.7312888629490785,
      "grad_norm": 0.17682015895843506,
      "learning_rate": 8.000093772000682e-06,
      "loss": 0.003,
      "step": 114020
    },
    {
      "epoch": 0.7313529998428646,
      "grad_norm": 0.12699271738529205,
      "learning_rate": 7.999645998378133e-06,
      "loss": 0.0037,
      "step": 114030
    },
    {
      "epoch": 0.7314171367366507,
      "grad_norm": 0.10249500721693039,
      "learning_rate": 7.999198187167875e-06,
      "loss": 0.0094,
      "step": 114040
    },
    {
      "epoch": 0.7314812736304368,
      "grad_norm": 0.019974878057837486,
      "learning_rate": 7.99875033837552e-06,
      "loss": 0.0018,
      "step": 114050
    },
    {
      "epoch": 0.731545410524223,
      "grad_norm": 0.12484804540872574,
      "learning_rate": 7.99830245200668e-06,
      "loss": 0.0034,
      "step": 114060
    },
    {
      "epoch": 0.731609547418009,
      "grad_norm": 0.27671071887016296,
      "learning_rate": 7.997854528066968e-06,
      "loss": 0.0058,
      "step": 114070
    },
    {
      "epoch": 0.7316736843117951,
      "grad_norm": 0.08566673845052719,
      "learning_rate": 7.997406566561996e-06,
      "loss": 0.0032,
      "step": 114080
    },
    {
      "epoch": 0.7317378212055812,
      "grad_norm": 0.03367238491773605,
      "learning_rate": 7.996958567497377e-06,
      "loss": 0.0013,
      "step": 114090
    },
    {
      "epoch": 0.7318019580993673,
      "grad_norm": 0.11499220132827759,
      "learning_rate": 7.996510530878726e-06,
      "loss": 0.0014,
      "step": 114100
    },
    {
      "epoch": 0.7318660949931534,
      "grad_norm": 0.10906586796045303,
      "learning_rate": 7.996062456711656e-06,
      "loss": 0.0025,
      "step": 114110
    },
    {
      "epoch": 0.7319302318869395,
      "grad_norm": 0.10528113692998886,
      "learning_rate": 7.995614345001783e-06,
      "loss": 0.0021,
      "step": 114120
    },
    {
      "epoch": 0.7319943687807255,
      "grad_norm": 0.014887388795614243,
      "learning_rate": 7.99516619575472e-06,
      "loss": 0.0018,
      "step": 114130
    },
    {
      "epoch": 0.7320585056745117,
      "grad_norm": 0.171325221657753,
      "learning_rate": 7.994718008976085e-06,
      "loss": 0.003,
      "step": 114140
    },
    {
      "epoch": 0.7321226425682977,
      "grad_norm": 0.05207236483693123,
      "learning_rate": 7.99426978467149e-06,
      "loss": 0.0017,
      "step": 114150
    },
    {
      "epoch": 0.7321867794620839,
      "grad_norm": 0.4033132791519165,
      "learning_rate": 7.993821522846559e-06,
      "loss": 0.0026,
      "step": 114160
    },
    {
      "epoch": 0.7322509163558699,
      "grad_norm": 0.43490123748779297,
      "learning_rate": 7.993373223506904e-06,
      "loss": 0.003,
      "step": 114170
    },
    {
      "epoch": 0.7323150532496561,
      "grad_norm": 0.24397335946559906,
      "learning_rate": 7.992924886658141e-06,
      "loss": 0.002,
      "step": 114180
    },
    {
      "epoch": 0.7323791901434422,
      "grad_norm": 0.04103788733482361,
      "learning_rate": 7.99247651230589e-06,
      "loss": 0.001,
      "step": 114190
    },
    {
      "epoch": 0.7324433270372283,
      "grad_norm": 0.1666671484708786,
      "learning_rate": 7.992028100455774e-06,
      "loss": 0.0035,
      "step": 114200
    },
    {
      "epoch": 0.7325074639310144,
      "grad_norm": 0.09332185238599777,
      "learning_rate": 7.991579651113404e-06,
      "loss": 0.0036,
      "step": 114210
    },
    {
      "epoch": 0.7325716008248004,
      "grad_norm": 0.04582349583506584,
      "learning_rate": 7.991131164284402e-06,
      "loss": 0.0015,
      "step": 114220
    },
    {
      "epoch": 0.7326357377185866,
      "grad_norm": 0.1251188963651657,
      "learning_rate": 7.99068263997439e-06,
      "loss": 0.0022,
      "step": 114230
    },
    {
      "epoch": 0.7326998746123726,
      "grad_norm": 0.07623518258333206,
      "learning_rate": 7.990234078188988e-06,
      "loss": 0.003,
      "step": 114240
    },
    {
      "epoch": 0.7327640115061588,
      "grad_norm": 0.019086243584752083,
      "learning_rate": 7.989785478933814e-06,
      "loss": 0.0018,
      "step": 114250
    },
    {
      "epoch": 0.7328281483999448,
      "grad_norm": 0.11019234359264374,
      "learning_rate": 7.989336842214492e-06,
      "loss": 0.0025,
      "step": 114260
    },
    {
      "epoch": 0.732892285293731,
      "grad_norm": 0.1383860558271408,
      "learning_rate": 7.988888168036644e-06,
      "loss": 0.002,
      "step": 114270
    },
    {
      "epoch": 0.732956422187517,
      "grad_norm": 0.2850213646888733,
      "learning_rate": 7.98843945640589e-06,
      "loss": 0.0018,
      "step": 114280
    },
    {
      "epoch": 0.7330205590813031,
      "grad_norm": 0.06595045328140259,
      "learning_rate": 7.987990707327854e-06,
      "loss": 0.0028,
      "step": 114290
    },
    {
      "epoch": 0.7330846959750892,
      "grad_norm": 0.15467871725559235,
      "learning_rate": 7.98754192080816e-06,
      "loss": 0.002,
      "step": 114300
    },
    {
      "epoch": 0.7331488328688753,
      "grad_norm": 0.05410248786211014,
      "learning_rate": 7.98709309685243e-06,
      "loss": 0.0027,
      "step": 114310
    },
    {
      "epoch": 0.7332129697626614,
      "grad_norm": 0.002955447882413864,
      "learning_rate": 7.986644235466287e-06,
      "loss": 0.0016,
      "step": 114320
    },
    {
      "epoch": 0.7332771066564475,
      "grad_norm": 0.16507507860660553,
      "learning_rate": 7.986195336655359e-06,
      "loss": 0.0019,
      "step": 114330
    },
    {
      "epoch": 0.7333412435502337,
      "grad_norm": 0.14873072504997253,
      "learning_rate": 7.985746400425268e-06,
      "loss": 0.0029,
      "step": 114340
    },
    {
      "epoch": 0.7334053804440197,
      "grad_norm": 0.046916212886571884,
      "learning_rate": 7.985297426781641e-06,
      "loss": 0.0055,
      "step": 114350
    },
    {
      "epoch": 0.7334695173378059,
      "grad_norm": 0.09673117846250534,
      "learning_rate": 7.984848415730104e-06,
      "loss": 0.0071,
      "step": 114360
    },
    {
      "epoch": 0.7335336542315919,
      "grad_norm": 0.15769807994365692,
      "learning_rate": 7.984399367276283e-06,
      "loss": 0.0041,
      "step": 114370
    },
    {
      "epoch": 0.733597791125378,
      "grad_norm": 0.023746896535158157,
      "learning_rate": 7.983950281425806e-06,
      "loss": 0.0033,
      "step": 114380
    },
    {
      "epoch": 0.7336619280191641,
      "grad_norm": 0.09823162108659744,
      "learning_rate": 7.9835011581843e-06,
      "loss": 0.0023,
      "step": 114390
    },
    {
      "epoch": 0.7337260649129502,
      "grad_norm": 0.1715027391910553,
      "learning_rate": 7.98305199755739e-06,
      "loss": 0.0041,
      "step": 114400
    },
    {
      "epoch": 0.7337902018067363,
      "grad_norm": 0.11333558708429337,
      "learning_rate": 7.982602799550707e-06,
      "loss": 0.0021,
      "step": 114410
    },
    {
      "epoch": 0.7338543387005224,
      "grad_norm": 0.10286201536655426,
      "learning_rate": 7.98215356416988e-06,
      "loss": 0.0027,
      "step": 114420
    },
    {
      "epoch": 0.7339184755943085,
      "grad_norm": 0.11100887507200241,
      "learning_rate": 7.981704291420536e-06,
      "loss": 0.0032,
      "step": 114430
    },
    {
      "epoch": 0.7339826124880946,
      "grad_norm": 0.12641538679599762,
      "learning_rate": 7.981254981308308e-06,
      "loss": 0.0058,
      "step": 114440
    },
    {
      "epoch": 0.7340467493818806,
      "grad_norm": 0.09120380133390427,
      "learning_rate": 7.980805633838824e-06,
      "loss": 0.0028,
      "step": 114450
    },
    {
      "epoch": 0.7341108862756668,
      "grad_norm": 0.06803891807794571,
      "learning_rate": 7.980356249017716e-06,
      "loss": 0.0019,
      "step": 114460
    },
    {
      "epoch": 0.7341750231694529,
      "grad_norm": 0.06600677222013474,
      "learning_rate": 7.979906826850611e-06,
      "loss": 0.0181,
      "step": 114470
    },
    {
      "epoch": 0.734239160063239,
      "grad_norm": 0.29142773151397705,
      "learning_rate": 7.979457367343147e-06,
      "loss": 0.0025,
      "step": 114480
    },
    {
      "epoch": 0.7343032969570251,
      "grad_norm": 0.13607315719127655,
      "learning_rate": 7.979007870500952e-06,
      "loss": 0.0036,
      "step": 114490
    },
    {
      "epoch": 0.7343674338508112,
      "grad_norm": 0.09334375709295273,
      "learning_rate": 7.978558336329658e-06,
      "loss": 0.0023,
      "step": 114500
    },
    {
      "epoch": 0.7344315707445973,
      "grad_norm": 0.011306799948215485,
      "learning_rate": 7.978108764834899e-06,
      "loss": 0.0019,
      "step": 114510
    },
    {
      "epoch": 0.7344957076383833,
      "grad_norm": 0.03890885412693024,
      "learning_rate": 7.977659156022311e-06,
      "loss": 0.0024,
      "step": 114520
    },
    {
      "epoch": 0.7345598445321695,
      "grad_norm": 0.022687526419758797,
      "learning_rate": 7.977209509897526e-06,
      "loss": 0.003,
      "step": 114530
    },
    {
      "epoch": 0.7346239814259555,
      "grad_norm": 0.23136533796787262,
      "learning_rate": 7.976759826466177e-06,
      "loss": 0.0025,
      "step": 114540
    },
    {
      "epoch": 0.7346881183197417,
      "grad_norm": 0.09535142779350281,
      "learning_rate": 7.976310105733899e-06,
      "loss": 0.0029,
      "step": 114550
    },
    {
      "epoch": 0.7347522552135277,
      "grad_norm": 0.29435694217681885,
      "learning_rate": 7.975860347706328e-06,
      "loss": 0.005,
      "step": 114560
    },
    {
      "epoch": 0.7348163921073139,
      "grad_norm": 0.0760713517665863,
      "learning_rate": 7.975410552389102e-06,
      "loss": 0.002,
      "step": 114570
    },
    {
      "epoch": 0.7348805290010999,
      "grad_norm": 0.14145128428936005,
      "learning_rate": 7.974960719787854e-06,
      "loss": 0.0037,
      "step": 114580
    },
    {
      "epoch": 0.734944665894886,
      "grad_norm": 0.17666327953338623,
      "learning_rate": 7.974510849908221e-06,
      "loss": 0.0073,
      "step": 114590
    },
    {
      "epoch": 0.7350088027886721,
      "grad_norm": 0.10045598447322845,
      "learning_rate": 7.974060942755844e-06,
      "loss": 0.0028,
      "step": 114600
    },
    {
      "epoch": 0.7350729396824582,
      "grad_norm": 0.1442408561706543,
      "learning_rate": 7.973610998336356e-06,
      "loss": 0.0043,
      "step": 114610
    },
    {
      "epoch": 0.7351370765762444,
      "grad_norm": 0.18976947665214539,
      "learning_rate": 7.973161016655397e-06,
      "loss": 0.0019,
      "step": 114620
    },
    {
      "epoch": 0.7352012134700304,
      "grad_norm": 0.09054972976446152,
      "learning_rate": 7.972710997718606e-06,
      "loss": 0.0025,
      "step": 114630
    },
    {
      "epoch": 0.7352653503638166,
      "grad_norm": 0.27474135160446167,
      "learning_rate": 7.972260941531621e-06,
      "loss": 0.0031,
      "step": 114640
    },
    {
      "epoch": 0.7353294872576026,
      "grad_norm": 0.19614854454994202,
      "learning_rate": 7.971810848100083e-06,
      "loss": 0.0031,
      "step": 114650
    },
    {
      "epoch": 0.7353936241513888,
      "grad_norm": 0.06520947068929672,
      "learning_rate": 7.97136071742963e-06,
      "loss": 0.0016,
      "step": 114660
    },
    {
      "epoch": 0.7354577610451748,
      "grad_norm": 0.142005056142807,
      "learning_rate": 7.970910549525905e-06,
      "loss": 0.0048,
      "step": 114670
    },
    {
      "epoch": 0.735521897938961,
      "grad_norm": 0.10520284622907639,
      "learning_rate": 7.970460344394547e-06,
      "loss": 0.0051,
      "step": 114680
    },
    {
      "epoch": 0.735586034832747,
      "grad_norm": 0.12937726080417633,
      "learning_rate": 7.970010102041198e-06,
      "loss": 0.0024,
      "step": 114690
    },
    {
      "epoch": 0.7356501717265331,
      "grad_norm": 0.06415631622076035,
      "learning_rate": 7.9695598224715e-06,
      "loss": 0.0015,
      "step": 114700
    },
    {
      "epoch": 0.7357143086203192,
      "grad_norm": 0.23039346933364868,
      "learning_rate": 7.969109505691095e-06,
      "loss": 0.0055,
      "step": 114710
    },
    {
      "epoch": 0.7357784455141053,
      "grad_norm": 0.20240607857704163,
      "learning_rate": 7.968659151705625e-06,
      "loss": 0.0018,
      "step": 114720
    },
    {
      "epoch": 0.7358425824078914,
      "grad_norm": 0.18377630412578583,
      "learning_rate": 7.968208760520734e-06,
      "loss": 0.0014,
      "step": 114730
    },
    {
      "epoch": 0.7359067193016775,
      "grad_norm": 0.05094422027468681,
      "learning_rate": 7.967758332142066e-06,
      "loss": 0.0026,
      "step": 114740
    },
    {
      "epoch": 0.7359708561954637,
      "grad_norm": 0.3360855281352997,
      "learning_rate": 7.967307866575266e-06,
      "loss": 0.0039,
      "step": 114750
    },
    {
      "epoch": 0.7360349930892497,
      "grad_norm": 0.15867236256599426,
      "learning_rate": 7.966857363825978e-06,
      "loss": 0.0016,
      "step": 114760
    },
    {
      "epoch": 0.7360991299830358,
      "grad_norm": 0.2619006335735321,
      "learning_rate": 7.966406823899846e-06,
      "loss": 0.0035,
      "step": 114770
    },
    {
      "epoch": 0.7361632668768219,
      "grad_norm": 0.15900486707687378,
      "learning_rate": 7.965956246802518e-06,
      "loss": 0.0025,
      "step": 114780
    },
    {
      "epoch": 0.736227403770608,
      "grad_norm": 0.012450681068003178,
      "learning_rate": 7.965505632539637e-06,
      "loss": 0.002,
      "step": 114790
    },
    {
      "epoch": 0.7362915406643941,
      "grad_norm": 0.11295517534017563,
      "learning_rate": 7.96505498111685e-06,
      "loss": 0.0025,
      "step": 114800
    },
    {
      "epoch": 0.7363556775581802,
      "grad_norm": 0.121597059071064,
      "learning_rate": 7.964604292539807e-06,
      "loss": 0.004,
      "step": 114810
    },
    {
      "epoch": 0.7364198144519662,
      "grad_norm": 0.18834513425827026,
      "learning_rate": 7.964153566814152e-06,
      "loss": 0.0025,
      "step": 114820
    },
    {
      "epoch": 0.7364839513457524,
      "grad_norm": 0.066258504986763,
      "learning_rate": 7.963702803945537e-06,
      "loss": 0.0019,
      "step": 114830
    },
    {
      "epoch": 0.7365480882395384,
      "grad_norm": 0.12791961431503296,
      "learning_rate": 7.963252003939604e-06,
      "loss": 0.0017,
      "step": 114840
    },
    {
      "epoch": 0.7366122251333246,
      "grad_norm": 0.13025297224521637,
      "learning_rate": 7.962801166802008e-06,
      "loss": 0.0037,
      "step": 114850
    },
    {
      "epoch": 0.7366763620271106,
      "grad_norm": 0.07951223105192184,
      "learning_rate": 7.962350292538395e-06,
      "loss": 0.0019,
      "step": 114860
    },
    {
      "epoch": 0.7367404989208968,
      "grad_norm": 0.09436211735010147,
      "learning_rate": 7.961899381154416e-06,
      "loss": 0.0062,
      "step": 114870
    },
    {
      "epoch": 0.7368046358146828,
      "grad_norm": 0.017899997532367706,
      "learning_rate": 7.961448432655721e-06,
      "loss": 0.0036,
      "step": 114880
    },
    {
      "epoch": 0.736868772708469,
      "grad_norm": 0.07869095355272293,
      "learning_rate": 7.96099744704796e-06,
      "loss": 0.0038,
      "step": 114890
    },
    {
      "epoch": 0.7369329096022551,
      "grad_norm": 0.034353915601968765,
      "learning_rate": 7.960546424336786e-06,
      "loss": 0.0015,
      "step": 114900
    },
    {
      "epoch": 0.7369970464960411,
      "grad_norm": 0.022068442776799202,
      "learning_rate": 7.960095364527849e-06,
      "loss": 0.002,
      "step": 114910
    },
    {
      "epoch": 0.7370611833898273,
      "grad_norm": 0.05181076377630234,
      "learning_rate": 7.9596442676268e-06,
      "loss": 0.0035,
      "step": 114920
    },
    {
      "epoch": 0.7371253202836133,
      "grad_norm": 0.014081995002925396,
      "learning_rate": 7.959193133639296e-06,
      "loss": 0.0023,
      "step": 114930
    },
    {
      "epoch": 0.7371894571773995,
      "grad_norm": 0.03727160021662712,
      "learning_rate": 7.958741962570985e-06,
      "loss": 0.0032,
      "step": 114940
    },
    {
      "epoch": 0.7372535940711855,
      "grad_norm": 0.1262127161026001,
      "learning_rate": 7.958290754427524e-06,
      "loss": 0.0019,
      "step": 114950
    },
    {
      "epoch": 0.7373177309649717,
      "grad_norm": 0.07796236872673035,
      "learning_rate": 7.957839509214565e-06,
      "loss": 0.0034,
      "step": 114960
    },
    {
      "epoch": 0.7373818678587577,
      "grad_norm": 0.1312754899263382,
      "learning_rate": 7.957388226937762e-06,
      "loss": 0.0021,
      "step": 114970
    },
    {
      "epoch": 0.7374460047525438,
      "grad_norm": 0.22057336568832397,
      "learning_rate": 7.95693690760277e-06,
      "loss": 0.0071,
      "step": 114980
    },
    {
      "epoch": 0.7375101416463299,
      "grad_norm": 0.1529206484556198,
      "learning_rate": 7.956485551215247e-06,
      "loss": 0.0028,
      "step": 114990
    },
    {
      "epoch": 0.737574278540116,
      "grad_norm": 0.17639851570129395,
      "learning_rate": 7.956034157780848e-06,
      "loss": 0.0029,
      "step": 115000
    },
    {
      "epoch": 0.7376384154339021,
      "grad_norm": 0.18121221661567688,
      "learning_rate": 7.955582727305226e-06,
      "loss": 0.0026,
      "step": 115010
    },
    {
      "epoch": 0.7377025523276882,
      "grad_norm": 0.15257194638252258,
      "learning_rate": 7.955131259794042e-06,
      "loss": 0.0034,
      "step": 115020
    },
    {
      "epoch": 0.7377666892214744,
      "grad_norm": 0.2247915416955948,
      "learning_rate": 7.954679755252953e-06,
      "loss": 0.0033,
      "step": 115030
    },
    {
      "epoch": 0.7378308261152604,
      "grad_norm": 0.1564386785030365,
      "learning_rate": 7.954228213687611e-06,
      "loss": 0.0024,
      "step": 115040
    },
    {
      "epoch": 0.7378949630090466,
      "grad_norm": 0.04837026074528694,
      "learning_rate": 7.95377663510368e-06,
      "loss": 0.0029,
      "step": 115050
    },
    {
      "epoch": 0.7379590999028326,
      "grad_norm": 0.26205313205718994,
      "learning_rate": 7.953325019506817e-06,
      "loss": 0.0041,
      "step": 115060
    },
    {
      "epoch": 0.7380232367966187,
      "grad_norm": 0.15560157597064972,
      "learning_rate": 7.95287336690268e-06,
      "loss": 0.0019,
      "step": 115070
    },
    {
      "epoch": 0.7380873736904048,
      "grad_norm": 0.06288137286901474,
      "learning_rate": 7.952421677296929e-06,
      "loss": 0.0041,
      "step": 115080
    },
    {
      "epoch": 0.7381515105841909,
      "grad_norm": 0.05975082889199257,
      "learning_rate": 7.951969950695226e-06,
      "loss": 0.0022,
      "step": 115090
    },
    {
      "epoch": 0.738215647477977,
      "grad_norm": 0.06963266432285309,
      "learning_rate": 7.951518187103228e-06,
      "loss": 0.0028,
      "step": 115100
    },
    {
      "epoch": 0.7382797843717631,
      "grad_norm": 0.04305969551205635,
      "learning_rate": 7.951066386526599e-06,
      "loss": 0.0029,
      "step": 115110
    },
    {
      "epoch": 0.7383439212655492,
      "grad_norm": 0.1279025375843048,
      "learning_rate": 7.950614548971e-06,
      "loss": 0.0019,
      "step": 115120
    },
    {
      "epoch": 0.7384080581593353,
      "grad_norm": 0.13910166919231415,
      "learning_rate": 7.95016267444209e-06,
      "loss": 0.0034,
      "step": 115130
    },
    {
      "epoch": 0.7384721950531213,
      "grad_norm": 0.13004015386104584,
      "learning_rate": 7.949710762945532e-06,
      "loss": 0.0024,
      "step": 115140
    },
    {
      "epoch": 0.7385363319469075,
      "grad_norm": 0.11320220679044724,
      "learning_rate": 7.949258814486992e-06,
      "loss": 0.0019,
      "step": 115150
    },
    {
      "epoch": 0.7386004688406935,
      "grad_norm": 0.08462710678577423,
      "learning_rate": 7.948806829072131e-06,
      "loss": 0.0019,
      "step": 115160
    },
    {
      "epoch": 0.7386646057344797,
      "grad_norm": 0.012864558957517147,
      "learning_rate": 7.948354806706612e-06,
      "loss": 0.0016,
      "step": 115170
    },
    {
      "epoch": 0.7387287426282658,
      "grad_norm": 0.14978507161140442,
      "learning_rate": 7.947902747396104e-06,
      "loss": 0.0023,
      "step": 115180
    },
    {
      "epoch": 0.7387928795220519,
      "grad_norm": 0.15812519192695618,
      "learning_rate": 7.947450651146263e-06,
      "loss": 0.0025,
      "step": 115190
    },
    {
      "epoch": 0.738857016415838,
      "grad_norm": 0.08912328630685806,
      "learning_rate": 7.946998517962761e-06,
      "loss": 0.0027,
      "step": 115200
    },
    {
      "epoch": 0.738921153309624,
      "grad_norm": 0.1578788161277771,
      "learning_rate": 7.946546347851261e-06,
      "loss": 0.0025,
      "step": 115210
    },
    {
      "epoch": 0.7389852902034102,
      "grad_norm": 0.021888367831707,
      "learning_rate": 7.946094140817429e-06,
      "loss": 0.0047,
      "step": 115220
    },
    {
      "epoch": 0.7390494270971962,
      "grad_norm": 0.10366988927125931,
      "learning_rate": 7.945641896866932e-06,
      "loss": 0.0033,
      "step": 115230
    },
    {
      "epoch": 0.7391135639909824,
      "grad_norm": 0.17833055555820465,
      "learning_rate": 7.945189616005437e-06,
      "loss": 0.0029,
      "step": 115240
    },
    {
      "epoch": 0.7391777008847684,
      "grad_norm": 0.167525514960289,
      "learning_rate": 7.944737298238612e-06,
      "loss": 0.002,
      "step": 115250
    },
    {
      "epoch": 0.7392418377785546,
      "grad_norm": 0.1168316900730133,
      "learning_rate": 7.944284943572124e-06,
      "loss": 0.0026,
      "step": 115260
    },
    {
      "epoch": 0.7393059746723406,
      "grad_norm": 0.08253283053636551,
      "learning_rate": 7.94383255201164e-06,
      "loss": 0.0022,
      "step": 115270
    },
    {
      "epoch": 0.7393701115661268,
      "grad_norm": 0.20252768695354462,
      "learning_rate": 7.943380123562831e-06,
      "loss": 0.0035,
      "step": 115280
    },
    {
      "epoch": 0.7394342484599128,
      "grad_norm": 0.07709493488073349,
      "learning_rate": 7.942927658231367e-06,
      "loss": 0.0033,
      "step": 115290
    },
    {
      "epoch": 0.7394983853536989,
      "grad_norm": 0.06704801321029663,
      "learning_rate": 7.942475156022914e-06,
      "loss": 0.0015,
      "step": 115300
    },
    {
      "epoch": 0.739562522247485,
      "grad_norm": 0.04719102755188942,
      "learning_rate": 7.942022616943145e-06,
      "loss": 0.0014,
      "step": 115310
    },
    {
      "epoch": 0.7396266591412711,
      "grad_norm": 0.009879032149910927,
      "learning_rate": 7.94157004099773e-06,
      "loss": 0.004,
      "step": 115320
    },
    {
      "epoch": 0.7396907960350573,
      "grad_norm": 0.17882804572582245,
      "learning_rate": 7.94111742819234e-06,
      "loss": 0.0025,
      "step": 115330
    },
    {
      "epoch": 0.7397549329288433,
      "grad_norm": 0.2576644718647003,
      "learning_rate": 7.940664778532646e-06,
      "loss": 0.0022,
      "step": 115340
    },
    {
      "epoch": 0.7398190698226295,
      "grad_norm": 0.07398247718811035,
      "learning_rate": 7.940212092024323e-06,
      "loss": 0.0033,
      "step": 115350
    },
    {
      "epoch": 0.7398832067164155,
      "grad_norm": 0.055791664868593216,
      "learning_rate": 7.93975936867304e-06,
      "loss": 0.0036,
      "step": 115360
    },
    {
      "epoch": 0.7399473436102016,
      "grad_norm": 0.16191360354423523,
      "learning_rate": 7.93930660848447e-06,
      "loss": 0.0027,
      "step": 115370
    },
    {
      "epoch": 0.7400114805039877,
      "grad_norm": 0.06203453615307808,
      "learning_rate": 7.938853811464286e-06,
      "loss": 0.0024,
      "step": 115380
    },
    {
      "epoch": 0.7400756173977738,
      "grad_norm": 0.03755979612469673,
      "learning_rate": 7.938400977618165e-06,
      "loss": 0.0012,
      "step": 115390
    },
    {
      "epoch": 0.7401397542915599,
      "grad_norm": 0.2009858340024948,
      "learning_rate": 7.937948106951781e-06,
      "loss": 0.0031,
      "step": 115400
    },
    {
      "epoch": 0.740203891185346,
      "grad_norm": 0.004805264063179493,
      "learning_rate": 7.937495199470807e-06,
      "loss": 0.0022,
      "step": 115410
    },
    {
      "epoch": 0.7402680280791321,
      "grad_norm": 0.06541886925697327,
      "learning_rate": 7.937042255180919e-06,
      "loss": 0.0032,
      "step": 115420
    },
    {
      "epoch": 0.7403321649729182,
      "grad_norm": 0.11845489591360092,
      "learning_rate": 7.936589274087791e-06,
      "loss": 0.0036,
      "step": 115430
    },
    {
      "epoch": 0.7403963018667042,
      "grad_norm": 0.07816637307405472,
      "learning_rate": 7.936136256197102e-06,
      "loss": 0.0055,
      "step": 115440
    },
    {
      "epoch": 0.7404604387604904,
      "grad_norm": 0.03445274010300636,
      "learning_rate": 7.935683201514528e-06,
      "loss": 0.0029,
      "step": 115450
    },
    {
      "epoch": 0.7405245756542765,
      "grad_norm": 0.05244622379541397,
      "learning_rate": 7.935230110045745e-06,
      "loss": 0.0016,
      "step": 115460
    },
    {
      "epoch": 0.7405887125480626,
      "grad_norm": 0.0628136396408081,
      "learning_rate": 7.934776981796428e-06,
      "loss": 0.003,
      "step": 115470
    },
    {
      "epoch": 0.7406528494418487,
      "grad_norm": 0.11316493898630142,
      "learning_rate": 7.93432381677226e-06,
      "loss": 0.0021,
      "step": 115480
    },
    {
      "epoch": 0.7407169863356348,
      "grad_norm": 0.10986774414777756,
      "learning_rate": 7.933870614978918e-06,
      "loss": 0.0032,
      "step": 115490
    },
    {
      "epoch": 0.7407811232294209,
      "grad_norm": 0.1046801283955574,
      "learning_rate": 7.93341737642208e-06,
      "loss": 0.0024,
      "step": 115500
    },
    {
      "epoch": 0.740845260123207,
      "grad_norm": 0.08572613447904587,
      "learning_rate": 7.932964101107426e-06,
      "loss": 0.0018,
      "step": 115510
    },
    {
      "epoch": 0.7409093970169931,
      "grad_norm": 0.10648373514413834,
      "learning_rate": 7.932510789040635e-06,
      "loss": 0.0033,
      "step": 115520
    },
    {
      "epoch": 0.7409735339107791,
      "grad_norm": 0.11053439974784851,
      "learning_rate": 7.932057440227387e-06,
      "loss": 0.0023,
      "step": 115530
    },
    {
      "epoch": 0.7410376708045653,
      "grad_norm": 0.045133188366889954,
      "learning_rate": 7.931604054673366e-06,
      "loss": 0.0049,
      "step": 115540
    },
    {
      "epoch": 0.7411018076983513,
      "grad_norm": 0.15804682672023773,
      "learning_rate": 7.93115063238425e-06,
      "loss": 0.0023,
      "step": 115550
    },
    {
      "epoch": 0.7411659445921375,
      "grad_norm": 0.09051597863435745,
      "learning_rate": 7.930697173365722e-06,
      "loss": 0.0024,
      "step": 115560
    },
    {
      "epoch": 0.7412300814859235,
      "grad_norm": 0.11107442528009415,
      "learning_rate": 7.930243677623464e-06,
      "loss": 0.0023,
      "step": 115570
    },
    {
      "epoch": 0.7412942183797097,
      "grad_norm": 0.14368917047977448,
      "learning_rate": 7.929790145163159e-06,
      "loss": 0.0028,
      "step": 115580
    },
    {
      "epoch": 0.7413583552734957,
      "grad_norm": 0.29199692606925964,
      "learning_rate": 7.929336575990489e-06,
      "loss": 0.0029,
      "step": 115590
    },
    {
      "epoch": 0.7414224921672818,
      "grad_norm": 0.2326393723487854,
      "learning_rate": 7.928882970111138e-06,
      "loss": 0.0045,
      "step": 115600
    },
    {
      "epoch": 0.741486629061068,
      "grad_norm": 0.09193108975887299,
      "learning_rate": 7.92842932753079e-06,
      "loss": 0.0024,
      "step": 115610
    },
    {
      "epoch": 0.741550765954854,
      "grad_norm": 0.06306244432926178,
      "learning_rate": 7.927975648255129e-06,
      "loss": 0.0046,
      "step": 115620
    },
    {
      "epoch": 0.7416149028486402,
      "grad_norm": 0.09674990922212601,
      "learning_rate": 7.927521932289841e-06,
      "loss": 0.0024,
      "step": 115630
    },
    {
      "epoch": 0.7416790397424262,
      "grad_norm": 0.15148349106311798,
      "learning_rate": 7.92706817964061e-06,
      "loss": 0.0025,
      "step": 115640
    },
    {
      "epoch": 0.7417431766362124,
      "grad_norm": 0.10181494802236557,
      "learning_rate": 7.926614390313126e-06,
      "loss": 0.0023,
      "step": 115650
    },
    {
      "epoch": 0.7418073135299984,
      "grad_norm": 0.239225372672081,
      "learning_rate": 7.92616056431307e-06,
      "loss": 0.002,
      "step": 115660
    },
    {
      "epoch": 0.7418714504237846,
      "grad_norm": 0.1631385236978531,
      "learning_rate": 7.925706701646131e-06,
      "loss": 0.002,
      "step": 115670
    },
    {
      "epoch": 0.7419355873175706,
      "grad_norm": 0.11287748068571091,
      "learning_rate": 7.925252802317995e-06,
      "loss": 0.0059,
      "step": 115680
    },
    {
      "epoch": 0.7419997242113567,
      "grad_norm": 0.17991571128368378,
      "learning_rate": 7.924798866334352e-06,
      "loss": 0.0028,
      "step": 115690
    },
    {
      "epoch": 0.7420638611051428,
      "grad_norm": 0.13380445539951324,
      "learning_rate": 7.924344893700888e-06,
      "loss": 0.0029,
      "step": 115700
    },
    {
      "epoch": 0.7421279979989289,
      "grad_norm": 0.15858301520347595,
      "learning_rate": 7.923890884423294e-06,
      "loss": 0.0029,
      "step": 115710
    },
    {
      "epoch": 0.742192134892715,
      "grad_norm": 0.16018827259540558,
      "learning_rate": 7.923436838507257e-06,
      "loss": 0.0022,
      "step": 115720
    },
    {
      "epoch": 0.7422562717865011,
      "grad_norm": 0.09589401632547379,
      "learning_rate": 7.922982755958466e-06,
      "loss": 0.0017,
      "step": 115730
    },
    {
      "epoch": 0.7423204086802873,
      "grad_norm": 0.12657934427261353,
      "learning_rate": 7.922528636782613e-06,
      "loss": 0.0022,
      "step": 115740
    },
    {
      "epoch": 0.7423845455740733,
      "grad_norm": 0.1356390118598938,
      "learning_rate": 7.922074480985386e-06,
      "loss": 0.0042,
      "step": 115750
    },
    {
      "epoch": 0.7424486824678594,
      "grad_norm": 0.08364280313253403,
      "learning_rate": 7.921620288572479e-06,
      "loss": 0.003,
      "step": 115760
    },
    {
      "epoch": 0.7425128193616455,
      "grad_norm": 0.05459647253155708,
      "learning_rate": 7.92116605954958e-06,
      "loss": 0.0024,
      "step": 115770
    },
    {
      "epoch": 0.7425769562554316,
      "grad_norm": 0.01840919628739357,
      "learning_rate": 7.920711793922386e-06,
      "loss": 0.0039,
      "step": 115780
    },
    {
      "epoch": 0.7426410931492177,
      "grad_norm": 0.15585069358348846,
      "learning_rate": 7.920257491696583e-06,
      "loss": 0.0032,
      "step": 115790
    },
    {
      "epoch": 0.7427052300430038,
      "grad_norm": 0.22318945825099945,
      "learning_rate": 7.919803152877868e-06,
      "loss": 0.0021,
      "step": 115800
    },
    {
      "epoch": 0.7427693669367899,
      "grad_norm": 0.2111048698425293,
      "learning_rate": 7.919348777471932e-06,
      "loss": 0.0024,
      "step": 115810
    },
    {
      "epoch": 0.742833503830576,
      "grad_norm": 0.04281293600797653,
      "learning_rate": 7.91889436548447e-06,
      "loss": 0.0019,
      "step": 115820
    },
    {
      "epoch": 0.742897640724362,
      "grad_norm": 0.09011639654636383,
      "learning_rate": 7.918439916921174e-06,
      "loss": 0.0021,
      "step": 115830
    },
    {
      "epoch": 0.7429617776181482,
      "grad_norm": 0.6694950461387634,
      "learning_rate": 7.91798543178774e-06,
      "loss": 0.0028,
      "step": 115840
    },
    {
      "epoch": 0.7430259145119342,
      "grad_norm": 0.08501161634922028,
      "learning_rate": 7.917530910089863e-06,
      "loss": 0.0021,
      "step": 115850
    },
    {
      "epoch": 0.7430900514057204,
      "grad_norm": 0.11211014539003372,
      "learning_rate": 7.917076351833241e-06,
      "loss": 0.0029,
      "step": 115860
    },
    {
      "epoch": 0.7431541882995064,
      "grad_norm": 0.3071892559528351,
      "learning_rate": 7.916621757023566e-06,
      "loss": 0.0041,
      "step": 115870
    },
    {
      "epoch": 0.7432183251932926,
      "grad_norm": 0.08402853459119797,
      "learning_rate": 7.916167125666535e-06,
      "loss": 0.0031,
      "step": 115880
    },
    {
      "epoch": 0.7432824620870787,
      "grad_norm": 0.1922350823879242,
      "learning_rate": 7.915712457767847e-06,
      "loss": 0.0049,
      "step": 115890
    },
    {
      "epoch": 0.7433465989808647,
      "grad_norm": 0.08401606231927872,
      "learning_rate": 7.915257753333198e-06,
      "loss": 0.0018,
      "step": 115900
    },
    {
      "epoch": 0.7434107358746509,
      "grad_norm": 0.06123369559645653,
      "learning_rate": 7.914803012368284e-06,
      "loss": 0.0015,
      "step": 115910
    },
    {
      "epoch": 0.7434748727684369,
      "grad_norm": 0.1281592696905136,
      "learning_rate": 7.914348234878809e-06,
      "loss": 0.0022,
      "step": 115920
    },
    {
      "epoch": 0.7435390096622231,
      "grad_norm": 0.47530466318130493,
      "learning_rate": 7.913893420870464e-06,
      "loss": 0.0022,
      "step": 115930
    },
    {
      "epoch": 0.7436031465560091,
      "grad_norm": 0.10956531018018723,
      "learning_rate": 7.913438570348954e-06,
      "loss": 0.003,
      "step": 115940
    },
    {
      "epoch": 0.7436672834497953,
      "grad_norm": 0.11079762876033783,
      "learning_rate": 7.912983683319977e-06,
      "loss": 0.0037,
      "step": 115950
    },
    {
      "epoch": 0.7437314203435813,
      "grad_norm": 0.12005963921546936,
      "learning_rate": 7.91252875978923e-06,
      "loss": 0.0047,
      "step": 115960
    },
    {
      "epoch": 0.7437955572373675,
      "grad_norm": 0.14262211322784424,
      "learning_rate": 7.912073799762418e-06,
      "loss": 0.0043,
      "step": 115970
    },
    {
      "epoch": 0.7438596941311535,
      "grad_norm": 0.19480326771736145,
      "learning_rate": 7.91161880324524e-06,
      "loss": 0.0014,
      "step": 115980
    },
    {
      "epoch": 0.7439238310249396,
      "grad_norm": 0.24568434059619904,
      "learning_rate": 7.911163770243397e-06,
      "loss": 0.0053,
      "step": 115990
    },
    {
      "epoch": 0.7439879679187257,
      "grad_norm": 0.0683426558971405,
      "learning_rate": 7.910708700762592e-06,
      "loss": 0.0016,
      "step": 116000
    },
    {
      "epoch": 0.7440521048125118,
      "grad_norm": 0.20744992792606354,
      "learning_rate": 7.910253594808525e-06,
      "loss": 0.0029,
      "step": 116010
    },
    {
      "epoch": 0.744116241706298,
      "grad_norm": 0.17102357745170593,
      "learning_rate": 7.909798452386903e-06,
      "loss": 0.0048,
      "step": 116020
    },
    {
      "epoch": 0.744180378600084,
      "grad_norm": 0.1635199785232544,
      "learning_rate": 7.909343273503425e-06,
      "loss": 0.0028,
      "step": 116030
    },
    {
      "epoch": 0.7442445154938702,
      "grad_norm": 0.05233636498451233,
      "learning_rate": 7.908888058163798e-06,
      "loss": 0.0025,
      "step": 116040
    },
    {
      "epoch": 0.7443086523876562,
      "grad_norm": 0.09148632735013962,
      "learning_rate": 7.908432806373722e-06,
      "loss": 0.0017,
      "step": 116050
    },
    {
      "epoch": 0.7443727892814423,
      "grad_norm": 0.10150976479053497,
      "learning_rate": 7.907977518138907e-06,
      "loss": 0.0046,
      "step": 116060
    },
    {
      "epoch": 0.7444369261752284,
      "grad_norm": 0.14146283268928528,
      "learning_rate": 7.907522193465053e-06,
      "loss": 0.0027,
      "step": 116070
    },
    {
      "epoch": 0.7445010630690145,
      "grad_norm": 0.033894527703523636,
      "learning_rate": 7.90706683235787e-06,
      "loss": 0.003,
      "step": 116080
    },
    {
      "epoch": 0.7445651999628006,
      "grad_norm": 0.11578021943569183,
      "learning_rate": 7.906611434823062e-06,
      "loss": 0.002,
      "step": 116090
    },
    {
      "epoch": 0.7446293368565867,
      "grad_norm": 0.13569103181362152,
      "learning_rate": 7.906156000866334e-06,
      "loss": 0.0047,
      "step": 116100
    },
    {
      "epoch": 0.7446934737503728,
      "grad_norm": 0.10451752692461014,
      "learning_rate": 7.905700530493395e-06,
      "loss": 0.0034,
      "step": 116110
    },
    {
      "epoch": 0.7447576106441589,
      "grad_norm": 0.13221339881420135,
      "learning_rate": 7.905245023709953e-06,
      "loss": 0.0032,
      "step": 116120
    },
    {
      "epoch": 0.744821747537945,
      "grad_norm": 0.17002162337303162,
      "learning_rate": 7.904789480521712e-06,
      "loss": 0.0021,
      "step": 116130
    },
    {
      "epoch": 0.7448858844317311,
      "grad_norm": 0.09643540531396866,
      "learning_rate": 7.904333900934384e-06,
      "loss": 0.0019,
      "step": 116140
    },
    {
      "epoch": 0.7449500213255171,
      "grad_norm": 0.18107008934020996,
      "learning_rate": 7.903878284953676e-06,
      "loss": 0.003,
      "step": 116150
    },
    {
      "epoch": 0.7450141582193033,
      "grad_norm": 0.10610051453113556,
      "learning_rate": 7.903422632585301e-06,
      "loss": 0.0023,
      "step": 116160
    },
    {
      "epoch": 0.7450782951130894,
      "grad_norm": 0.08352890610694885,
      "learning_rate": 7.902966943834961e-06,
      "loss": 0.0025,
      "step": 116170
    },
    {
      "epoch": 0.7451424320068755,
      "grad_norm": 0.13513338565826416,
      "learning_rate": 7.902511218708374e-06,
      "loss": 0.0027,
      "step": 116180
    },
    {
      "epoch": 0.7452065689006616,
      "grad_norm": 0.13289561867713928,
      "learning_rate": 7.902055457211243e-06,
      "loss": 0.0039,
      "step": 116190
    },
    {
      "epoch": 0.7452707057944477,
      "grad_norm": 0.007371044717729092,
      "learning_rate": 7.901599659349285e-06,
      "loss": 0.0028,
      "step": 116200
    },
    {
      "epoch": 0.7453348426882338,
      "grad_norm": 0.10732848942279816,
      "learning_rate": 7.90114382512821e-06,
      "loss": 0.0021,
      "step": 116210
    },
    {
      "epoch": 0.7453989795820198,
      "grad_norm": 0.03691869601607323,
      "learning_rate": 7.900687954553729e-06,
      "loss": 0.0048,
      "step": 116220
    },
    {
      "epoch": 0.745463116475806,
      "grad_norm": 0.10616063326597214,
      "learning_rate": 7.900232047631555e-06,
      "loss": 0.0013,
      "step": 116230
    },
    {
      "epoch": 0.745527253369592,
      "grad_norm": 0.14096885919570923,
      "learning_rate": 7.8997761043674e-06,
      "loss": 0.002,
      "step": 116240
    },
    {
      "epoch": 0.7455913902633782,
      "grad_norm": 0.10583070665597916,
      "learning_rate": 7.899320124766978e-06,
      "loss": 0.0025,
      "step": 116250
    },
    {
      "epoch": 0.7456555271571642,
      "grad_norm": 0.07014064490795135,
      "learning_rate": 7.898864108836003e-06,
      "loss": 0.0025,
      "step": 116260
    },
    {
      "epoch": 0.7457196640509504,
      "grad_norm": 0.054637208580970764,
      "learning_rate": 7.89840805658019e-06,
      "loss": 0.001,
      "step": 116270
    },
    {
      "epoch": 0.7457838009447364,
      "grad_norm": 0.00924642663449049,
      "learning_rate": 7.897951968005253e-06,
      "loss": 0.0016,
      "step": 116280
    },
    {
      "epoch": 0.7458479378385225,
      "grad_norm": 0.19948308169841766,
      "learning_rate": 7.897495843116905e-06,
      "loss": 0.0035,
      "step": 116290
    },
    {
      "epoch": 0.7459120747323087,
      "grad_norm": 0.09530455619096756,
      "learning_rate": 7.897039681920863e-06,
      "loss": 0.0016,
      "step": 116300
    },
    {
      "epoch": 0.7459762116260947,
      "grad_norm": 0.09297530353069305,
      "learning_rate": 7.896583484422845e-06,
      "loss": 0.0019,
      "step": 116310
    },
    {
      "epoch": 0.7460403485198809,
      "grad_norm": 0.22910071909427643,
      "learning_rate": 7.896127250628565e-06,
      "loss": 0.0034,
      "step": 116320
    },
    {
      "epoch": 0.7461044854136669,
      "grad_norm": 0.14804109930992126,
      "learning_rate": 7.895670980543742e-06,
      "loss": 0.0018,
      "step": 116330
    },
    {
      "epoch": 0.7461686223074531,
      "grad_norm": 0.07542126625776291,
      "learning_rate": 7.89521467417409e-06,
      "loss": 0.0026,
      "step": 116340
    },
    {
      "epoch": 0.7462327592012391,
      "grad_norm": 0.22744214534759521,
      "learning_rate": 7.894758331525329e-06,
      "loss": 0.004,
      "step": 116350
    },
    {
      "epoch": 0.7462968960950253,
      "grad_norm": 0.11813952028751373,
      "learning_rate": 7.894301952603178e-06,
      "loss": 0.0021,
      "step": 116360
    },
    {
      "epoch": 0.7463610329888113,
      "grad_norm": 0.0634949803352356,
      "learning_rate": 7.893845537413356e-06,
      "loss": 0.0025,
      "step": 116370
    },
    {
      "epoch": 0.7464251698825974,
      "grad_norm": 0.08142178505659103,
      "learning_rate": 7.89338908596158e-06,
      "loss": 0.003,
      "step": 116380
    },
    {
      "epoch": 0.7464893067763835,
      "grad_norm": 0.08885622769594193,
      "learning_rate": 7.892932598253571e-06,
      "loss": 0.0029,
      "step": 116390
    },
    {
      "epoch": 0.7465534436701696,
      "grad_norm": 0.04796062782406807,
      "learning_rate": 7.89247607429505e-06,
      "loss": 0.0099,
      "step": 116400
    },
    {
      "epoch": 0.7466175805639557,
      "grad_norm": 0.1358189582824707,
      "learning_rate": 7.892019514091735e-06,
      "loss": 0.0029,
      "step": 116410
    },
    {
      "epoch": 0.7466817174577418,
      "grad_norm": 0.16324840486049652,
      "learning_rate": 7.891562917649349e-06,
      "loss": 0.0032,
      "step": 116420
    },
    {
      "epoch": 0.7467458543515278,
      "grad_norm": 0.3833000659942627,
      "learning_rate": 7.891106284973613e-06,
      "loss": 0.0039,
      "step": 116430
    },
    {
      "epoch": 0.746809991245314,
      "grad_norm": 0.10605830699205399,
      "learning_rate": 7.89064961607025e-06,
      "loss": 0.0042,
      "step": 116440
    },
    {
      "epoch": 0.7468741281391001,
      "grad_norm": 0.07625432312488556,
      "learning_rate": 7.89019291094498e-06,
      "loss": 0.0022,
      "step": 116450
    },
    {
      "epoch": 0.7469382650328862,
      "grad_norm": 0.10014986991882324,
      "learning_rate": 7.889736169603528e-06,
      "loss": 0.0024,
      "step": 116460
    },
    {
      "epoch": 0.7470024019266723,
      "grad_norm": 0.09487932175397873,
      "learning_rate": 7.889279392051617e-06,
      "loss": 0.0034,
      "step": 116470
    },
    {
      "epoch": 0.7470665388204584,
      "grad_norm": 0.18687134981155396,
      "learning_rate": 7.88882257829497e-06,
      "loss": 0.0035,
      "step": 116480
    },
    {
      "epoch": 0.7471306757142445,
      "grad_norm": 0.2099267542362213,
      "learning_rate": 7.888365728339311e-06,
      "loss": 0.0052,
      "step": 116490
    },
    {
      "epoch": 0.7471948126080306,
      "grad_norm": 0.024278149008750916,
      "learning_rate": 7.887908842190366e-06,
      "loss": 0.0035,
      "step": 116500
    },
    {
      "epoch": 0.7472589495018167,
      "grad_norm": 0.07662907987833023,
      "learning_rate": 7.887451919853858e-06,
      "loss": 0.0034,
      "step": 116510
    },
    {
      "epoch": 0.7473230863956027,
      "grad_norm": 0.009092072956264019,
      "learning_rate": 7.886994961335515e-06,
      "loss": 0.0022,
      "step": 116520
    },
    {
      "epoch": 0.7473872232893889,
      "grad_norm": 0.050177864730358124,
      "learning_rate": 7.886537966641061e-06,
      "loss": 0.0034,
      "step": 116530
    },
    {
      "epoch": 0.7474513601831749,
      "grad_norm": 0.025178076699376106,
      "learning_rate": 7.886080935776224e-06,
      "loss": 0.0031,
      "step": 116540
    },
    {
      "epoch": 0.7475154970769611,
      "grad_norm": 0.09660907834768295,
      "learning_rate": 7.885623868746729e-06,
      "loss": 0.0038,
      "step": 116550
    },
    {
      "epoch": 0.7475796339707471,
      "grad_norm": 0.008361510001122952,
      "learning_rate": 7.885166765558306e-06,
      "loss": 0.0019,
      "step": 116560
    },
    {
      "epoch": 0.7476437708645333,
      "grad_norm": 0.1537180244922638,
      "learning_rate": 7.88470962621668e-06,
      "loss": 0.0027,
      "step": 116570
    },
    {
      "epoch": 0.7477079077583194,
      "grad_norm": 0.6607270836830139,
      "learning_rate": 7.884252450727582e-06,
      "loss": 0.0027,
      "step": 116580
    },
    {
      "epoch": 0.7477720446521054,
      "grad_norm": 0.17144745588302612,
      "learning_rate": 7.883795239096739e-06,
      "loss": 0.003,
      "step": 116590
    },
    {
      "epoch": 0.7478361815458916,
      "grad_norm": 0.3937118351459503,
      "learning_rate": 7.883337991329881e-06,
      "loss": 0.0032,
      "step": 116600
    },
    {
      "epoch": 0.7479003184396776,
      "grad_norm": 0.15119680762290955,
      "learning_rate": 7.882880707432736e-06,
      "loss": 0.0037,
      "step": 116610
    },
    {
      "epoch": 0.7479644553334638,
      "grad_norm": 0.35869231820106506,
      "learning_rate": 7.882423387411037e-06,
      "loss": 0.0034,
      "step": 116620
    },
    {
      "epoch": 0.7480285922272498,
      "grad_norm": 0.055572181940078735,
      "learning_rate": 7.881966031270512e-06,
      "loss": 0.0015,
      "step": 116630
    },
    {
      "epoch": 0.748092729121036,
      "grad_norm": 0.1298820823431015,
      "learning_rate": 7.881508639016893e-06,
      "loss": 0.003,
      "step": 116640
    },
    {
      "epoch": 0.748156866014822,
      "grad_norm": 0.13784027099609375,
      "learning_rate": 7.88105121065591e-06,
      "loss": 0.0015,
      "step": 116650
    },
    {
      "epoch": 0.7482210029086082,
      "grad_norm": 0.25377780199050903,
      "learning_rate": 7.880593746193298e-06,
      "loss": 0.01,
      "step": 116660
    },
    {
      "epoch": 0.7482851398023942,
      "grad_norm": 0.06298059225082397,
      "learning_rate": 7.880136245634789e-06,
      "loss": 0.0027,
      "step": 116670
    },
    {
      "epoch": 0.7483492766961803,
      "grad_norm": 0.05065106227993965,
      "learning_rate": 7.879678708986113e-06,
      "loss": 0.0025,
      "step": 116680
    },
    {
      "epoch": 0.7484134135899664,
      "grad_norm": 0.02653634175658226,
      "learning_rate": 7.879221136253003e-06,
      "loss": 0.0023,
      "step": 116690
    },
    {
      "epoch": 0.7484775504837525,
      "grad_norm": 0.08247742801904678,
      "learning_rate": 7.878763527441198e-06,
      "loss": 0.0022,
      "step": 116700
    },
    {
      "epoch": 0.7485416873775386,
      "grad_norm": 0.10867869108915329,
      "learning_rate": 7.878305882556426e-06,
      "loss": 0.0041,
      "step": 116710
    },
    {
      "epoch": 0.7486058242713247,
      "grad_norm": 0.1672622412443161,
      "learning_rate": 7.877848201604425e-06,
      "loss": 0.0027,
      "step": 116720
    },
    {
      "epoch": 0.7486699611651109,
      "grad_norm": 0.25147178769111633,
      "learning_rate": 7.877390484590928e-06,
      "loss": 0.0025,
      "step": 116730
    },
    {
      "epoch": 0.7487340980588969,
      "grad_norm": 0.04914018139243126,
      "learning_rate": 7.876932731521673e-06,
      "loss": 0.0016,
      "step": 116740
    },
    {
      "epoch": 0.748798234952683,
      "grad_norm": 0.07496075332164764,
      "learning_rate": 7.876474942402395e-06,
      "loss": 0.0018,
      "step": 116750
    },
    {
      "epoch": 0.7488623718464691,
      "grad_norm": 0.19975468516349792,
      "learning_rate": 7.87601711723883e-06,
      "loss": 0.0038,
      "step": 116760
    },
    {
      "epoch": 0.7489265087402552,
      "grad_norm": 0.05046171694993973,
      "learning_rate": 7.875559256036714e-06,
      "loss": 0.0034,
      "step": 116770
    },
    {
      "epoch": 0.7489906456340413,
      "grad_norm": 0.07045719027519226,
      "learning_rate": 7.875101358801787e-06,
      "loss": 0.0051,
      "step": 116780
    },
    {
      "epoch": 0.7490547825278274,
      "grad_norm": 0.15396460890769958,
      "learning_rate": 7.874643425539785e-06,
      "loss": 0.0018,
      "step": 116790
    },
    {
      "epoch": 0.7491189194216135,
      "grad_norm": 0.23495283722877502,
      "learning_rate": 7.874185456256444e-06,
      "loss": 0.0038,
      "step": 116800
    },
    {
      "epoch": 0.7491830563153996,
      "grad_norm": 0.1421741098165512,
      "learning_rate": 7.873727450957506e-06,
      "loss": 0.0031,
      "step": 116810
    },
    {
      "epoch": 0.7492471932091856,
      "grad_norm": 0.06015840545296669,
      "learning_rate": 7.873269409648711e-06,
      "loss": 0.0035,
      "step": 116820
    },
    {
      "epoch": 0.7493113301029718,
      "grad_norm": 0.20923273265361786,
      "learning_rate": 7.872811332335796e-06,
      "loss": 0.0027,
      "step": 116830
    },
    {
      "epoch": 0.7493754669967578,
      "grad_norm": 0.06418105959892273,
      "learning_rate": 7.872353219024503e-06,
      "loss": 0.0024,
      "step": 116840
    },
    {
      "epoch": 0.749439603890544,
      "grad_norm": 0.06075701117515564,
      "learning_rate": 7.871895069720569e-06,
      "loss": 0.0018,
      "step": 116850
    },
    {
      "epoch": 0.74950374078433,
      "grad_norm": 0.060914766043424606,
      "learning_rate": 7.871436884429739e-06,
      "loss": 0.002,
      "step": 116860
    },
    {
      "epoch": 0.7495678776781162,
      "grad_norm": 0.003072801511734724,
      "learning_rate": 7.87097866315775e-06,
      "loss": 0.0042,
      "step": 116870
    },
    {
      "epoch": 0.7496320145719023,
      "grad_norm": 0.11973962932825089,
      "learning_rate": 7.87052040591035e-06,
      "loss": 0.0013,
      "step": 116880
    },
    {
      "epoch": 0.7496961514656884,
      "grad_norm": 0.0684128925204277,
      "learning_rate": 7.870062112693277e-06,
      "loss": 0.0042,
      "step": 116890
    },
    {
      "epoch": 0.7497602883594745,
      "grad_norm": 0.25988301634788513,
      "learning_rate": 7.869603783512273e-06,
      "loss": 0.0029,
      "step": 116900
    },
    {
      "epoch": 0.7498244252532605,
      "grad_norm": 0.14829394221305847,
      "learning_rate": 7.869145418373083e-06,
      "loss": 0.0022,
      "step": 116910
    },
    {
      "epoch": 0.7498885621470467,
      "grad_norm": 0.09994681924581528,
      "learning_rate": 7.868687017281452e-06,
      "loss": 0.0031,
      "step": 116920
    },
    {
      "epoch": 0.7499526990408327,
      "grad_norm": 0.09394358843564987,
      "learning_rate": 7.86822858024312e-06,
      "loss": 0.0042,
      "step": 116930
    },
    {
      "epoch": 0.7500168359346189,
      "grad_norm": 0.11060336232185364,
      "learning_rate": 7.867770107263837e-06,
      "loss": 0.0042,
      "step": 116940
    },
    {
      "epoch": 0.7500809728284049,
      "grad_norm": 0.25255483388900757,
      "learning_rate": 7.867311598349343e-06,
      "loss": 0.0029,
      "step": 116950
    },
    {
      "epoch": 0.7501451097221911,
      "grad_norm": 0.13214614987373352,
      "learning_rate": 7.866853053505386e-06,
      "loss": 0.0021,
      "step": 116960
    },
    {
      "epoch": 0.7502092466159771,
      "grad_norm": 0.1233336552977562,
      "learning_rate": 7.86639447273771e-06,
      "loss": 0.0019,
      "step": 116970
    },
    {
      "epoch": 0.7502733835097632,
      "grad_norm": 0.334183931350708,
      "learning_rate": 7.865935856052064e-06,
      "loss": 0.0038,
      "step": 116980
    },
    {
      "epoch": 0.7503375204035493,
      "grad_norm": 0.0827973261475563,
      "learning_rate": 7.865477203454193e-06,
      "loss": 0.0025,
      "step": 116990
    },
    {
      "epoch": 0.7504016572973354,
      "grad_norm": 0.004071071743965149,
      "learning_rate": 7.865018514949844e-06,
      "loss": 0.0028,
      "step": 117000
    },
    {
      "epoch": 0.7504657941911216,
      "grad_norm": 0.11148324608802795,
      "learning_rate": 7.864559790544768e-06,
      "loss": 0.002,
      "step": 117010
    },
    {
      "epoch": 0.7505299310849076,
      "grad_norm": 0.16427700221538544,
      "learning_rate": 7.864101030244708e-06,
      "loss": 0.0028,
      "step": 117020
    },
    {
      "epoch": 0.7505940679786938,
      "grad_norm": 0.04392922297120094,
      "learning_rate": 7.863642234055416e-06,
      "loss": 0.0009,
      "step": 117030
    },
    {
      "epoch": 0.7506582048724798,
      "grad_norm": 0.08696134388446808,
      "learning_rate": 7.86318340198264e-06,
      "loss": 0.0027,
      "step": 117040
    },
    {
      "epoch": 0.750722341766266,
      "grad_norm": 0.07006500661373138,
      "learning_rate": 7.862724534032131e-06,
      "loss": 0.0024,
      "step": 117050
    },
    {
      "epoch": 0.750786478660052,
      "grad_norm": 0.10490524768829346,
      "learning_rate": 7.862265630209635e-06,
      "loss": 0.0044,
      "step": 117060
    },
    {
      "epoch": 0.7508506155538381,
      "grad_norm": 0.0829753503203392,
      "learning_rate": 7.861806690520908e-06,
      "loss": 0.0021,
      "step": 117070
    },
    {
      "epoch": 0.7509147524476242,
      "grad_norm": 0.2763497829437256,
      "learning_rate": 7.861347714971696e-06,
      "loss": 0.002,
      "step": 117080
    },
    {
      "epoch": 0.7509788893414103,
      "grad_norm": 0.1821936070919037,
      "learning_rate": 7.860888703567753e-06,
      "loss": 0.0023,
      "step": 117090
    },
    {
      "epoch": 0.7510430262351964,
      "grad_norm": 0.1808260977268219,
      "learning_rate": 7.860429656314827e-06,
      "loss": 0.004,
      "step": 117100
    },
    {
      "epoch": 0.7511071631289825,
      "grad_norm": 0.08387022465467453,
      "learning_rate": 7.859970573218675e-06,
      "loss": 0.0029,
      "step": 117110
    },
    {
      "epoch": 0.7511713000227686,
      "grad_norm": 0.10462870448827744,
      "learning_rate": 7.85951145428505e-06,
      "loss": 0.0034,
      "step": 117120
    },
    {
      "epoch": 0.7512354369165547,
      "grad_norm": 0.06143558397889137,
      "learning_rate": 7.859052299519701e-06,
      "loss": 0.0022,
      "step": 117130
    },
    {
      "epoch": 0.7512995738103407,
      "grad_norm": 0.09308692067861557,
      "learning_rate": 7.858593108928383e-06,
      "loss": 0.0024,
      "step": 117140
    },
    {
      "epoch": 0.7513637107041269,
      "grad_norm": 0.038160715252161026,
      "learning_rate": 7.858133882516852e-06,
      "loss": 0.0021,
      "step": 117150
    },
    {
      "epoch": 0.751427847597913,
      "grad_norm": 0.13036786019802094,
      "learning_rate": 7.85767462029086e-06,
      "loss": 0.0016,
      "step": 117160
    },
    {
      "epoch": 0.7514919844916991,
      "grad_norm": 0.2630171775817871,
      "learning_rate": 7.857215322256162e-06,
      "loss": 0.0038,
      "step": 117170
    },
    {
      "epoch": 0.7515561213854852,
      "grad_norm": 0.04008711874485016,
      "learning_rate": 7.856755988418514e-06,
      "loss": 0.0049,
      "step": 117180
    },
    {
      "epoch": 0.7516202582792713,
      "grad_norm": 0.13282530009746552,
      "learning_rate": 7.856296618783672e-06,
      "loss": 0.0036,
      "step": 117190
    },
    {
      "epoch": 0.7516843951730574,
      "grad_norm": 0.13415652513504028,
      "learning_rate": 7.85583721335739e-06,
      "loss": 0.003,
      "step": 117200
    },
    {
      "epoch": 0.7517485320668434,
      "grad_norm": 0.04852619394659996,
      "learning_rate": 7.855377772145431e-06,
      "loss": 0.0023,
      "step": 117210
    },
    {
      "epoch": 0.7518126689606296,
      "grad_norm": 0.10932836681604385,
      "learning_rate": 7.854918295153546e-06,
      "loss": 0.0051,
      "step": 117220
    },
    {
      "epoch": 0.7518768058544156,
      "grad_norm": 0.0693749263882637,
      "learning_rate": 7.854458782387494e-06,
      "loss": 0.0028,
      "step": 117230
    },
    {
      "epoch": 0.7519409427482018,
      "grad_norm": 0.15589742362499237,
      "learning_rate": 7.853999233853032e-06,
      "loss": 0.0016,
      "step": 117240
    },
    {
      "epoch": 0.7520050796419878,
      "grad_norm": 0.2942682206630707,
      "learning_rate": 7.853539649555922e-06,
      "loss": 0.0041,
      "step": 117250
    },
    {
      "epoch": 0.752069216535774,
      "grad_norm": 0.10200204700231552,
      "learning_rate": 7.853080029501918e-06,
      "loss": 0.0042,
      "step": 117260
    },
    {
      "epoch": 0.75213335342956,
      "grad_norm": 0.12070576101541519,
      "learning_rate": 7.852620373696786e-06,
      "loss": 0.0034,
      "step": 117270
    },
    {
      "epoch": 0.7521974903233462,
      "grad_norm": 0.10774677991867065,
      "learning_rate": 7.85216068214628e-06,
      "loss": 0.0023,
      "step": 117280
    },
    {
      "epoch": 0.7522616272171323,
      "grad_norm": 0.15308523178100586,
      "learning_rate": 7.851700954856162e-06,
      "loss": 0.0033,
      "step": 117290
    },
    {
      "epoch": 0.7523257641109183,
      "grad_norm": 0.2831006646156311,
      "learning_rate": 7.851241191832192e-06,
      "loss": 0.0035,
      "step": 117300
    },
    {
      "epoch": 0.7523899010047045,
      "grad_norm": 0.06799483299255371,
      "learning_rate": 7.850781393080134e-06,
      "loss": 0.0011,
      "step": 117310
    },
    {
      "epoch": 0.7524540378984905,
      "grad_norm": 0.16940274834632874,
      "learning_rate": 7.850321558605748e-06,
      "loss": 0.0024,
      "step": 117320
    },
    {
      "epoch": 0.7525181747922767,
      "grad_norm": 0.0938853770494461,
      "learning_rate": 7.849861688414795e-06,
      "loss": 0.0027,
      "step": 117330
    },
    {
      "epoch": 0.7525823116860627,
      "grad_norm": 0.18646354973316193,
      "learning_rate": 7.849401782513037e-06,
      "loss": 0.0056,
      "step": 117340
    },
    {
      "epoch": 0.7526464485798489,
      "grad_norm": 0.16712817549705505,
      "learning_rate": 7.84894184090624e-06,
      "loss": 0.0034,
      "step": 117350
    },
    {
      "epoch": 0.7527105854736349,
      "grad_norm": 0.2718506157398224,
      "learning_rate": 7.848481863600165e-06,
      "loss": 0.0035,
      "step": 117360
    },
    {
      "epoch": 0.752774722367421,
      "grad_norm": 0.16131126880645752,
      "learning_rate": 7.848021850600578e-06,
      "loss": 0.0028,
      "step": 117370
    },
    {
      "epoch": 0.7528388592612071,
      "grad_norm": 0.10022114962339401,
      "learning_rate": 7.84756180191324e-06,
      "loss": 0.002,
      "step": 117380
    },
    {
      "epoch": 0.7529029961549932,
      "grad_norm": 0.18879581987857819,
      "learning_rate": 7.847101717543916e-06,
      "loss": 0.0048,
      "step": 117390
    },
    {
      "epoch": 0.7529671330487793,
      "grad_norm": 0.02280520647764206,
      "learning_rate": 7.846641597498375e-06,
      "loss": 0.0017,
      "step": 117400
    },
    {
      "epoch": 0.7530312699425654,
      "grad_norm": 0.22137390077114105,
      "learning_rate": 7.84618144178238e-06,
      "loss": 0.0031,
      "step": 117410
    },
    {
      "epoch": 0.7530954068363515,
      "grad_norm": 0.01127664279192686,
      "learning_rate": 7.845721250401697e-06,
      "loss": 0.0019,
      "step": 117420
    },
    {
      "epoch": 0.7531595437301376,
      "grad_norm": 0.15001355111598969,
      "learning_rate": 7.845261023362093e-06,
      "loss": 0.0022,
      "step": 117430
    },
    {
      "epoch": 0.7532236806239238,
      "grad_norm": 0.3733644485473633,
      "learning_rate": 7.844800760669336e-06,
      "loss": 0.0028,
      "step": 117440
    },
    {
      "epoch": 0.7532878175177098,
      "grad_norm": 0.06113684922456741,
      "learning_rate": 7.84434046232919e-06,
      "loss": 0.0021,
      "step": 117450
    },
    {
      "epoch": 0.7533519544114959,
      "grad_norm": 0.17942652106285095,
      "learning_rate": 7.843880128347426e-06,
      "loss": 0.0025,
      "step": 117460
    },
    {
      "epoch": 0.753416091305282,
      "grad_norm": 0.14272238314151764,
      "learning_rate": 7.843419758729814e-06,
      "loss": 0.0022,
      "step": 117470
    },
    {
      "epoch": 0.7534802281990681,
      "grad_norm": 0.05989169329404831,
      "learning_rate": 7.842959353482116e-06,
      "loss": 0.0014,
      "step": 117480
    },
    {
      "epoch": 0.7535443650928542,
      "grad_norm": 0.029922956600785255,
      "learning_rate": 7.842498912610109e-06,
      "loss": 0.0031,
      "step": 117490
    },
    {
      "epoch": 0.7536085019866403,
      "grad_norm": 0.16737212240695953,
      "learning_rate": 7.842038436119558e-06,
      "loss": 0.0034,
      "step": 117500
    },
    {
      "epoch": 0.7536726388804263,
      "grad_norm": 0.28119874000549316,
      "learning_rate": 7.841577924016233e-06,
      "loss": 0.0034,
      "step": 117510
    },
    {
      "epoch": 0.7537367757742125,
      "grad_norm": 0.2850058972835541,
      "learning_rate": 7.841117376305906e-06,
      "loss": 0.0018,
      "step": 117520
    },
    {
      "epoch": 0.7538009126679985,
      "grad_norm": 0.08792038261890411,
      "learning_rate": 7.84065679299435e-06,
      "loss": 0.0036,
      "step": 117530
    },
    {
      "epoch": 0.7538650495617847,
      "grad_norm": 0.032527461647987366,
      "learning_rate": 7.840196174087333e-06,
      "loss": 0.0038,
      "step": 117540
    },
    {
      "epoch": 0.7539291864555707,
      "grad_norm": 0.08409000188112259,
      "learning_rate": 7.839735519590628e-06,
      "loss": 0.0036,
      "step": 117550
    },
    {
      "epoch": 0.7539933233493569,
      "grad_norm": 0.0372193269431591,
      "learning_rate": 7.839274829510008e-06,
      "loss": 0.0024,
      "step": 117560
    },
    {
      "epoch": 0.754057460243143,
      "grad_norm": 0.17293556034564972,
      "learning_rate": 7.838814103851244e-06,
      "loss": 0.003,
      "step": 117570
    },
    {
      "epoch": 0.7541215971369291,
      "grad_norm": 0.09905791282653809,
      "learning_rate": 7.83835334262011e-06,
      "loss": 0.0022,
      "step": 117580
    },
    {
      "epoch": 0.7541857340307152,
      "grad_norm": 0.2138233780860901,
      "learning_rate": 7.837892545822381e-06,
      "loss": 0.0024,
      "step": 117590
    },
    {
      "epoch": 0.7542498709245012,
      "grad_norm": 0.35840851068496704,
      "learning_rate": 7.837431713463831e-06,
      "loss": 0.0023,
      "step": 117600
    },
    {
      "epoch": 0.7543140078182874,
      "grad_norm": 0.18870952725410461,
      "learning_rate": 7.836970845550232e-06,
      "loss": 0.0028,
      "step": 117610
    },
    {
      "epoch": 0.7543781447120734,
      "grad_norm": 0.010556691326200962,
      "learning_rate": 7.836509942087362e-06,
      "loss": 0.0041,
      "step": 117620
    },
    {
      "epoch": 0.7544422816058596,
      "grad_norm": 0.029404442757368088,
      "learning_rate": 7.836049003080994e-06,
      "loss": 0.002,
      "step": 117630
    },
    {
      "epoch": 0.7545064184996456,
      "grad_norm": 0.05506880208849907,
      "learning_rate": 7.835588028536906e-06,
      "loss": 0.0021,
      "step": 117640
    },
    {
      "epoch": 0.7545705553934318,
      "grad_norm": 0.20457029342651367,
      "learning_rate": 7.835127018460876e-06,
      "loss": 0.0039,
      "step": 117650
    },
    {
      "epoch": 0.7546346922872178,
      "grad_norm": 0.1909932941198349,
      "learning_rate": 7.834665972858674e-06,
      "loss": 0.0045,
      "step": 117660
    },
    {
      "epoch": 0.754698829181004,
      "grad_norm": 0.045761752873659134,
      "learning_rate": 7.834204891736083e-06,
      "loss": 0.0023,
      "step": 117670
    },
    {
      "epoch": 0.75476296607479,
      "grad_norm": 0.23617449402809143,
      "learning_rate": 7.83374377509888e-06,
      "loss": 0.0034,
      "step": 117680
    },
    {
      "epoch": 0.7548271029685761,
      "grad_norm": 0.018489936366677284,
      "learning_rate": 7.833282622952842e-06,
      "loss": 0.0017,
      "step": 117690
    },
    {
      "epoch": 0.7548912398623622,
      "grad_norm": 0.16313032805919647,
      "learning_rate": 7.832821435303745e-06,
      "loss": 0.0028,
      "step": 117700
    },
    {
      "epoch": 0.7549553767561483,
      "grad_norm": 0.1480388343334198,
      "learning_rate": 7.832360212157374e-06,
      "loss": 0.0028,
      "step": 117710
    },
    {
      "epoch": 0.7550195136499345,
      "grad_norm": 0.00466338824480772,
      "learning_rate": 7.831898953519505e-06,
      "loss": 0.0023,
      "step": 117720
    },
    {
      "epoch": 0.7550836505437205,
      "grad_norm": 0.14430424571037292,
      "learning_rate": 7.831437659395917e-06,
      "loss": 0.0021,
      "step": 117730
    },
    {
      "epoch": 0.7551477874375067,
      "grad_norm": 0.24340131878852844,
      "learning_rate": 7.830976329792393e-06,
      "loss": 0.002,
      "step": 117740
    },
    {
      "epoch": 0.7552119243312927,
      "grad_norm": 0.025799771770834923,
      "learning_rate": 7.83051496471471e-06,
      "loss": 0.0023,
      "step": 117750
    },
    {
      "epoch": 0.7552760612250788,
      "grad_norm": 0.12621502578258514,
      "learning_rate": 7.830053564168654e-06,
      "loss": 0.0026,
      "step": 117760
    },
    {
      "epoch": 0.7553401981188649,
      "grad_norm": 0.20064160227775574,
      "learning_rate": 7.829592128160003e-06,
      "loss": 0.0029,
      "step": 117770
    },
    {
      "epoch": 0.755404335012651,
      "grad_norm": 0.1149071529507637,
      "learning_rate": 7.82913065669454e-06,
      "loss": 0.0014,
      "step": 117780
    },
    {
      "epoch": 0.7554684719064371,
      "grad_norm": 0.23584522306919098,
      "learning_rate": 7.828669149778048e-06,
      "loss": 0.0032,
      "step": 117790
    },
    {
      "epoch": 0.7555326088002232,
      "grad_norm": 0.020709898322820663,
      "learning_rate": 7.828207607416312e-06,
      "loss": 0.0029,
      "step": 117800
    },
    {
      "epoch": 0.7555967456940093,
      "grad_norm": 0.08159718662500381,
      "learning_rate": 7.827746029615112e-06,
      "loss": 0.0033,
      "step": 117810
    },
    {
      "epoch": 0.7556608825877954,
      "grad_norm": 0.05778425559401512,
      "learning_rate": 7.827284416380231e-06,
      "loss": 0.0017,
      "step": 117820
    },
    {
      "epoch": 0.7557250194815814,
      "grad_norm": 0.08645426481962204,
      "learning_rate": 7.82682276771746e-06,
      "loss": 0.0018,
      "step": 117830
    },
    {
      "epoch": 0.7557891563753676,
      "grad_norm": 0.1733463555574417,
      "learning_rate": 7.826361083632576e-06,
      "loss": 0.0026,
      "step": 117840
    },
    {
      "epoch": 0.7558532932691537,
      "grad_norm": 0.12011563777923584,
      "learning_rate": 7.825899364131368e-06,
      "loss": 0.003,
      "step": 117850
    },
    {
      "epoch": 0.7559174301629398,
      "grad_norm": 0.07007107138633728,
      "learning_rate": 7.825437609219622e-06,
      "loss": 0.0066,
      "step": 117860
    },
    {
      "epoch": 0.7559815670567259,
      "grad_norm": 0.06176275387406349,
      "learning_rate": 7.824975818903124e-06,
      "loss": 0.0028,
      "step": 117870
    },
    {
      "epoch": 0.756045703950512,
      "grad_norm": 0.07692543417215347,
      "learning_rate": 7.82451399318766e-06,
      "loss": 0.0015,
      "step": 117880
    },
    {
      "epoch": 0.7561098408442981,
      "grad_norm": 0.33473703265190125,
      "learning_rate": 7.824052132079017e-06,
      "loss": 0.0037,
      "step": 117890
    },
    {
      "epoch": 0.7561739777380841,
      "grad_norm": 0.10095279663801193,
      "learning_rate": 7.823590235582982e-06,
      "loss": 0.0028,
      "step": 117900
    },
    {
      "epoch": 0.7562381146318703,
      "grad_norm": 0.10655076056718826,
      "learning_rate": 7.823128303705343e-06,
      "loss": 0.002,
      "step": 117910
    },
    {
      "epoch": 0.7563022515256563,
      "grad_norm": 0.10879048705101013,
      "learning_rate": 7.822666336451889e-06,
      "loss": 0.0047,
      "step": 117920
    },
    {
      "epoch": 0.7563663884194425,
      "grad_norm": 0.16622315347194672,
      "learning_rate": 7.822204333828409e-06,
      "loss": 0.0025,
      "step": 117930
    },
    {
      "epoch": 0.7564305253132285,
      "grad_norm": 0.08693701028823853,
      "learning_rate": 7.821742295840692e-06,
      "loss": 0.0021,
      "step": 117940
    },
    {
      "epoch": 0.7564946622070147,
      "grad_norm": 0.18205344676971436,
      "learning_rate": 7.821280222494526e-06,
      "loss": 0.0016,
      "step": 117950
    },
    {
      "epoch": 0.7565587991008007,
      "grad_norm": 0.16920578479766846,
      "learning_rate": 7.820818113795702e-06,
      "loss": 0.0057,
      "step": 117960
    },
    {
      "epoch": 0.7566229359945869,
      "grad_norm": 0.05380775406956673,
      "learning_rate": 7.820355969750012e-06,
      "loss": 0.0021,
      "step": 117970
    },
    {
      "epoch": 0.7566870728883729,
      "grad_norm": 0.13284482061862946,
      "learning_rate": 7.819893790363248e-06,
      "loss": 0.0019,
      "step": 117980
    },
    {
      "epoch": 0.756751209782159,
      "grad_norm": 0.3372058570384979,
      "learning_rate": 7.819431575641197e-06,
      "loss": 0.0044,
      "step": 117990
    },
    {
      "epoch": 0.7568153466759452,
      "grad_norm": 0.08662164211273193,
      "learning_rate": 7.818969325589654e-06,
      "loss": 0.0028,
      "step": 118000
    },
    {
      "epoch": 0.7568794835697312,
      "grad_norm": 0.026164045557379723,
      "learning_rate": 7.818507040214411e-06,
      "loss": 0.004,
      "step": 118010
    },
    {
      "epoch": 0.7569436204635174,
      "grad_norm": 0.14952978491783142,
      "learning_rate": 7.81804471952126e-06,
      "loss": 0.0022,
      "step": 118020
    },
    {
      "epoch": 0.7570077573573034,
      "grad_norm": 0.126239612698555,
      "learning_rate": 7.817582363515994e-06,
      "loss": 0.0045,
      "step": 118030
    },
    {
      "epoch": 0.7570718942510896,
      "grad_norm": 0.13074319064617157,
      "learning_rate": 7.817119972204409e-06,
      "loss": 0.0032,
      "step": 118040
    },
    {
      "epoch": 0.7571360311448756,
      "grad_norm": 0.09498288482427597,
      "learning_rate": 7.816657545592297e-06,
      "loss": 0.0024,
      "step": 118050
    },
    {
      "epoch": 0.7572001680386617,
      "grad_norm": 0.09087800979614258,
      "learning_rate": 7.816195083685452e-06,
      "loss": 0.0034,
      "step": 118060
    },
    {
      "epoch": 0.7572643049324478,
      "grad_norm": 0.04143861308693886,
      "learning_rate": 7.815732586489671e-06,
      "loss": 0.0051,
      "step": 118070
    },
    {
      "epoch": 0.7573284418262339,
      "grad_norm": 0.10549326241016388,
      "learning_rate": 7.815270054010747e-06,
      "loss": 0.0047,
      "step": 118080
    },
    {
      "epoch": 0.75739257872002,
      "grad_norm": 0.07025926560163498,
      "learning_rate": 7.814807486254477e-06,
      "loss": 0.0029,
      "step": 118090
    },
    {
      "epoch": 0.7574567156138061,
      "grad_norm": 0.07964323461055756,
      "learning_rate": 7.81434488322666e-06,
      "loss": 0.0025,
      "step": 118100
    },
    {
      "epoch": 0.7575208525075922,
      "grad_norm": 0.06105535849928856,
      "learning_rate": 7.813882244933086e-06,
      "loss": 0.0023,
      "step": 118110
    },
    {
      "epoch": 0.7575849894013783,
      "grad_norm": 0.0810265839099884,
      "learning_rate": 7.813419571379558e-06,
      "loss": 0.0015,
      "step": 118120
    },
    {
      "epoch": 0.7576491262951643,
      "grad_norm": 0.04925093799829483,
      "learning_rate": 7.812956862571874e-06,
      "loss": 0.0021,
      "step": 118130
    },
    {
      "epoch": 0.7577132631889505,
      "grad_norm": 0.02250676043331623,
      "learning_rate": 7.81249411851583e-06,
      "loss": 0.0015,
      "step": 118140
    },
    {
      "epoch": 0.7577774000827366,
      "grad_norm": 0.08988325297832489,
      "learning_rate": 7.812031339217223e-06,
      "loss": 0.0014,
      "step": 118150
    },
    {
      "epoch": 0.7578415369765227,
      "grad_norm": 0.02714879997074604,
      "learning_rate": 7.811568524681854e-06,
      "loss": 0.0018,
      "step": 118160
    },
    {
      "epoch": 0.7579056738703088,
      "grad_norm": 0.20410077273845673,
      "learning_rate": 7.811105674915523e-06,
      "loss": 0.0056,
      "step": 118170
    },
    {
      "epoch": 0.7579698107640949,
      "grad_norm": 0.07401904463768005,
      "learning_rate": 7.810642789924027e-06,
      "loss": 0.0018,
      "step": 118180
    },
    {
      "epoch": 0.758033947657881,
      "grad_norm": 0.1179809644818306,
      "learning_rate": 7.810179869713169e-06,
      "loss": 0.0026,
      "step": 118190
    },
    {
      "epoch": 0.758098084551667,
      "grad_norm": 0.2100778967142105,
      "learning_rate": 7.80971691428875e-06,
      "loss": 0.0032,
      "step": 118200
    },
    {
      "epoch": 0.7581622214454532,
      "grad_norm": 0.3232751190662384,
      "learning_rate": 7.809253923656567e-06,
      "loss": 0.0024,
      "step": 118210
    },
    {
      "epoch": 0.7582263583392392,
      "grad_norm": 0.09631489962339401,
      "learning_rate": 7.808790897822427e-06,
      "loss": 0.0045,
      "step": 118220
    },
    {
      "epoch": 0.7582904952330254,
      "grad_norm": 0.10500827431678772,
      "learning_rate": 7.80832783679213e-06,
      "loss": 0.0039,
      "step": 118230
    },
    {
      "epoch": 0.7583546321268114,
      "grad_norm": 0.09026511758565903,
      "learning_rate": 7.807864740571479e-06,
      "loss": 0.0019,
      "step": 118240
    },
    {
      "epoch": 0.7584187690205976,
      "grad_norm": 0.2540445625782013,
      "learning_rate": 7.807401609166274e-06,
      "loss": 0.0039,
      "step": 118250
    },
    {
      "epoch": 0.7584829059143836,
      "grad_norm": 0.21617233753204346,
      "learning_rate": 7.80693844258232e-06,
      "loss": 0.0038,
      "step": 118260
    },
    {
      "epoch": 0.7585470428081698,
      "grad_norm": 0.030382253229618073,
      "learning_rate": 7.806475240825421e-06,
      "loss": 0.0028,
      "step": 118270
    },
    {
      "epoch": 0.7586111797019559,
      "grad_norm": 0.09669791907072067,
      "learning_rate": 7.806012003901384e-06,
      "loss": 0.0016,
      "step": 118280
    },
    {
      "epoch": 0.7586753165957419,
      "grad_norm": 0.07175273448228836,
      "learning_rate": 7.805548731816009e-06,
      "loss": 0.0039,
      "step": 118290
    },
    {
      "epoch": 0.7587394534895281,
      "grad_norm": 0.1716214120388031,
      "learning_rate": 7.805085424575104e-06,
      "loss": 0.0027,
      "step": 118300
    },
    {
      "epoch": 0.7588035903833141,
      "grad_norm": 0.06944598257541656,
      "learning_rate": 7.804622082184473e-06,
      "loss": 0.0019,
      "step": 118310
    },
    {
      "epoch": 0.7588677272771003,
      "grad_norm": 0.1075916513800621,
      "learning_rate": 7.804158704649925e-06,
      "loss": 0.0028,
      "step": 118320
    },
    {
      "epoch": 0.7589318641708863,
      "grad_norm": 0.1181074008345604,
      "learning_rate": 7.803695291977262e-06,
      "loss": 0.0033,
      "step": 118330
    },
    {
      "epoch": 0.7589960010646725,
      "grad_norm": 0.13426217436790466,
      "learning_rate": 7.803231844172295e-06,
      "loss": 0.0044,
      "step": 118340
    },
    {
      "epoch": 0.7590601379584585,
      "grad_norm": 0.30457431077957153,
      "learning_rate": 7.80276836124083e-06,
      "loss": 0.004,
      "step": 118350
    },
    {
      "epoch": 0.7591242748522447,
      "grad_norm": 0.3851284980773926,
      "learning_rate": 7.802304843188672e-06,
      "loss": 0.0013,
      "step": 118360
    },
    {
      "epoch": 0.7591884117460307,
      "grad_norm": 0.15478985011577606,
      "learning_rate": 7.801841290021632e-06,
      "loss": 0.0041,
      "step": 118370
    },
    {
      "epoch": 0.7592525486398168,
      "grad_norm": 0.1672656387090683,
      "learning_rate": 7.801377701745518e-06,
      "loss": 0.003,
      "step": 118380
    },
    {
      "epoch": 0.7593166855336029,
      "grad_norm": 0.10904452949762344,
      "learning_rate": 7.800914078366142e-06,
      "loss": 0.0039,
      "step": 118390
    },
    {
      "epoch": 0.759380822427389,
      "grad_norm": 0.12326830625534058,
      "learning_rate": 7.800450419889308e-06,
      "loss": 0.0031,
      "step": 118400
    },
    {
      "epoch": 0.7594449593211751,
      "grad_norm": 0.04942172020673752,
      "learning_rate": 7.79998672632083e-06,
      "loss": 0.0023,
      "step": 118410
    },
    {
      "epoch": 0.7595090962149612,
      "grad_norm": 0.17837822437286377,
      "learning_rate": 7.799522997666517e-06,
      "loss": 0.004,
      "step": 118420
    },
    {
      "epoch": 0.7595732331087474,
      "grad_norm": 0.37312379479408264,
      "learning_rate": 7.79905923393218e-06,
      "loss": 0.0016,
      "step": 118430
    },
    {
      "epoch": 0.7596373700025334,
      "grad_norm": 0.0728418305516243,
      "learning_rate": 7.79859543512363e-06,
      "loss": 0.0013,
      "step": 118440
    },
    {
      "epoch": 0.7597015068963195,
      "grad_norm": 0.6216109991073608,
      "learning_rate": 7.798131601246679e-06,
      "loss": 0.0037,
      "step": 118450
    },
    {
      "epoch": 0.7597656437901056,
      "grad_norm": 0.283000111579895,
      "learning_rate": 7.79766773230714e-06,
      "loss": 0.0054,
      "step": 118460
    },
    {
      "epoch": 0.7598297806838917,
      "grad_norm": 0.10428867489099503,
      "learning_rate": 7.797203828310824e-06,
      "loss": 0.004,
      "step": 118470
    },
    {
      "epoch": 0.7598939175776778,
      "grad_norm": 0.10073235630989075,
      "learning_rate": 7.796739889263546e-06,
      "loss": 0.002,
      "step": 118480
    },
    {
      "epoch": 0.7599580544714639,
      "grad_norm": 0.31705033779144287,
      "learning_rate": 7.796275915171119e-06,
      "loss": 0.0051,
      "step": 118490
    },
    {
      "epoch": 0.76002219136525,
      "grad_norm": 0.13120688498020172,
      "learning_rate": 7.795811906039354e-06,
      "loss": 0.0046,
      "step": 118500
    },
    {
      "epoch": 0.7600863282590361,
      "grad_norm": 0.07548979669809341,
      "learning_rate": 7.795347861874069e-06,
      "loss": 0.0027,
      "step": 118510
    },
    {
      "epoch": 0.7601504651528221,
      "grad_norm": 0.1204274445772171,
      "learning_rate": 7.794883782681077e-06,
      "loss": 0.0051,
      "step": 118520
    },
    {
      "epoch": 0.7602146020466083,
      "grad_norm": 0.1520080715417862,
      "learning_rate": 7.794419668466194e-06,
      "loss": 0.0033,
      "step": 118530
    },
    {
      "epoch": 0.7602787389403943,
      "grad_norm": 0.24985525012016296,
      "learning_rate": 7.793955519235236e-06,
      "loss": 0.0028,
      "step": 118540
    },
    {
      "epoch": 0.7603428758341805,
      "grad_norm": 0.06198232248425484,
      "learning_rate": 7.793491334994017e-06,
      "loss": 0.0023,
      "step": 118550
    },
    {
      "epoch": 0.7604070127279666,
      "grad_norm": 0.11053550988435745,
      "learning_rate": 7.793027115748357e-06,
      "loss": 0.0036,
      "step": 118560
    },
    {
      "epoch": 0.7604711496217527,
      "grad_norm": 0.07890421152114868,
      "learning_rate": 7.79256286150407e-06,
      "loss": 0.0018,
      "step": 118570
    },
    {
      "epoch": 0.7605352865155388,
      "grad_norm": 0.3829288184642792,
      "learning_rate": 7.792098572266974e-06,
      "loss": 0.0068,
      "step": 118580
    },
    {
      "epoch": 0.7605994234093248,
      "grad_norm": 0.10420060157775879,
      "learning_rate": 7.791634248042887e-06,
      "loss": 0.0015,
      "step": 118590
    },
    {
      "epoch": 0.760663560303111,
      "grad_norm": 0.036105263978242874,
      "learning_rate": 7.79116988883763e-06,
      "loss": 0.0023,
      "step": 118600
    },
    {
      "epoch": 0.760727697196897,
      "grad_norm": 0.1294431984424591,
      "learning_rate": 7.790705494657018e-06,
      "loss": 0.0017,
      "step": 118610
    },
    {
      "epoch": 0.7607918340906832,
      "grad_norm": 0.18665571510791779,
      "learning_rate": 7.790241065506871e-06,
      "loss": 0.0018,
      "step": 118620
    },
    {
      "epoch": 0.7608559709844692,
      "grad_norm": 0.0662047490477562,
      "learning_rate": 7.78977660139301e-06,
      "loss": 0.0013,
      "step": 118630
    },
    {
      "epoch": 0.7609201078782554,
      "grad_norm": 0.1589532196521759,
      "learning_rate": 7.789312102321256e-06,
      "loss": 0.0016,
      "step": 118640
    },
    {
      "epoch": 0.7609842447720414,
      "grad_norm": 0.11786510795354843,
      "learning_rate": 7.788847568297426e-06,
      "loss": 0.0027,
      "step": 118650
    },
    {
      "epoch": 0.7610483816658276,
      "grad_norm": 0.16013629734516144,
      "learning_rate": 7.788382999327342e-06,
      "loss": 0.0016,
      "step": 118660
    },
    {
      "epoch": 0.7611125185596136,
      "grad_norm": 0.08604548871517181,
      "learning_rate": 7.78791839541683e-06,
      "loss": 0.0016,
      "step": 118670
    },
    {
      "epoch": 0.7611766554533997,
      "grad_norm": 0.09176530689001083,
      "learning_rate": 7.787453756571703e-06,
      "loss": 0.0026,
      "step": 118680
    },
    {
      "epoch": 0.7612407923471858,
      "grad_norm": 0.04124186933040619,
      "learning_rate": 7.78698908279779e-06,
      "loss": 0.0032,
      "step": 118690
    },
    {
      "epoch": 0.7613049292409719,
      "grad_norm": 0.08384433388710022,
      "learning_rate": 7.786524374100915e-06,
      "loss": 0.0047,
      "step": 118700
    },
    {
      "epoch": 0.7613690661347581,
      "grad_norm": 0.5120087265968323,
      "learning_rate": 7.786059630486895e-06,
      "loss": 0.0032,
      "step": 118710
    },
    {
      "epoch": 0.7614332030285441,
      "grad_norm": 0.364688515663147,
      "learning_rate": 7.785594851961558e-06,
      "loss": 0.005,
      "step": 118720
    },
    {
      "epoch": 0.7614973399223303,
      "grad_norm": 0.20993518829345703,
      "learning_rate": 7.785130038530726e-06,
      "loss": 0.0041,
      "step": 118730
    },
    {
      "epoch": 0.7615614768161163,
      "grad_norm": 0.1628425121307373,
      "learning_rate": 7.784665190200225e-06,
      "loss": 0.0026,
      "step": 118740
    },
    {
      "epoch": 0.7616256137099024,
      "grad_norm": 0.18203139305114746,
      "learning_rate": 7.784200306975878e-06,
      "loss": 0.0023,
      "step": 118750
    },
    {
      "epoch": 0.7616897506036885,
      "grad_norm": 0.10203225165605545,
      "learning_rate": 7.783735388863511e-06,
      "loss": 0.0013,
      "step": 118760
    },
    {
      "epoch": 0.7617538874974746,
      "grad_norm": 0.06325043737888336,
      "learning_rate": 7.78327043586895e-06,
      "loss": 0.0027,
      "step": 118770
    },
    {
      "epoch": 0.7618180243912607,
      "grad_norm": 0.05929301679134369,
      "learning_rate": 7.782805447998023e-06,
      "loss": 0.002,
      "step": 118780
    },
    {
      "epoch": 0.7618821612850468,
      "grad_norm": 0.15617991983890533,
      "learning_rate": 7.782340425256553e-06,
      "loss": 0.0042,
      "step": 118790
    },
    {
      "epoch": 0.7619462981788329,
      "grad_norm": 0.054896991699934006,
      "learning_rate": 7.78187536765037e-06,
      "loss": 0.0028,
      "step": 118800
    },
    {
      "epoch": 0.762010435072619,
      "grad_norm": 0.0946490690112114,
      "learning_rate": 7.781410275185301e-06,
      "loss": 0.0045,
      "step": 118810
    },
    {
      "epoch": 0.762074571966405,
      "grad_norm": 0.12389269471168518,
      "learning_rate": 7.780945147867172e-06,
      "loss": 0.003,
      "step": 118820
    },
    {
      "epoch": 0.7621387088601912,
      "grad_norm": 0.13792073726654053,
      "learning_rate": 7.780479985701813e-06,
      "loss": 0.0032,
      "step": 118830
    },
    {
      "epoch": 0.7622028457539773,
      "grad_norm": 0.17356400191783905,
      "learning_rate": 7.780014788695054e-06,
      "loss": 0.0022,
      "step": 118840
    },
    {
      "epoch": 0.7622669826477634,
      "grad_norm": 0.25488826632499695,
      "learning_rate": 7.779549556852722e-06,
      "loss": 0.0034,
      "step": 118850
    },
    {
      "epoch": 0.7623311195415495,
      "grad_norm": 0.14415204524993896,
      "learning_rate": 7.779084290180648e-06,
      "loss": 0.0031,
      "step": 118860
    },
    {
      "epoch": 0.7623952564353356,
      "grad_norm": 0.18693728744983673,
      "learning_rate": 7.77861898868466e-06,
      "loss": 0.0023,
      "step": 118870
    },
    {
      "epoch": 0.7624593933291217,
      "grad_norm": 0.32552337646484375,
      "learning_rate": 7.77815365237059e-06,
      "loss": 0.003,
      "step": 118880
    },
    {
      "epoch": 0.7625235302229078,
      "grad_norm": 0.21315798163414001,
      "learning_rate": 7.777688281244272e-06,
      "loss": 0.0029,
      "step": 118890
    },
    {
      "epoch": 0.7625876671166939,
      "grad_norm": 0.3089481294155121,
      "learning_rate": 7.777222875311533e-06,
      "loss": 0.0021,
      "step": 118900
    },
    {
      "epoch": 0.7626518040104799,
      "grad_norm": 0.07197723537683487,
      "learning_rate": 7.776757434578206e-06,
      "loss": 0.0023,
      "step": 118910
    },
    {
      "epoch": 0.7627159409042661,
      "grad_norm": 0.10376438498497009,
      "learning_rate": 7.776291959050125e-06,
      "loss": 0.0018,
      "step": 118920
    },
    {
      "epoch": 0.7627800777980521,
      "grad_norm": 0.16951358318328857,
      "learning_rate": 7.775826448733121e-06,
      "loss": 0.0024,
      "step": 118930
    },
    {
      "epoch": 0.7628442146918383,
      "grad_norm": 0.20929601788520813,
      "learning_rate": 7.775360903633026e-06,
      "loss": 0.0049,
      "step": 118940
    },
    {
      "epoch": 0.7629083515856243,
      "grad_norm": 0.09313658624887466,
      "learning_rate": 7.774895323755678e-06,
      "loss": 0.0032,
      "step": 118950
    },
    {
      "epoch": 0.7629724884794105,
      "grad_norm": 0.2288755476474762,
      "learning_rate": 7.774429709106907e-06,
      "loss": 0.0049,
      "step": 118960
    },
    {
      "epoch": 0.7630366253731965,
      "grad_norm": 0.13903319835662842,
      "learning_rate": 7.77396405969255e-06,
      "loss": 0.0028,
      "step": 118970
    },
    {
      "epoch": 0.7631007622669826,
      "grad_norm": 0.056591082364320755,
      "learning_rate": 7.77349837551844e-06,
      "loss": 0.002,
      "step": 118980
    },
    {
      "epoch": 0.7631648991607688,
      "grad_norm": 0.10128097981214523,
      "learning_rate": 7.773032656590414e-06,
      "loss": 0.0037,
      "step": 118990
    },
    {
      "epoch": 0.7632290360545548,
      "grad_norm": 0.2593241333961487,
      "learning_rate": 7.772566902914307e-06,
      "loss": 0.0035,
      "step": 119000
    },
    {
      "epoch": 0.763293172948341,
      "grad_norm": 0.40354588627815247,
      "learning_rate": 7.772101114495953e-06,
      "loss": 0.0041,
      "step": 119010
    },
    {
      "epoch": 0.763357309842127,
      "grad_norm": 0.10311109572649002,
      "learning_rate": 7.771635291341193e-06,
      "loss": 0.003,
      "step": 119020
    },
    {
      "epoch": 0.7634214467359132,
      "grad_norm": 0.09408441185951233,
      "learning_rate": 7.771169433455861e-06,
      "loss": 0.0018,
      "step": 119030
    },
    {
      "epoch": 0.7634855836296992,
      "grad_norm": 0.11243224889039993,
      "learning_rate": 7.770703540845797e-06,
      "loss": 0.0024,
      "step": 119040
    },
    {
      "epoch": 0.7635497205234854,
      "grad_norm": 0.07582049071788788,
      "learning_rate": 7.770237613516836e-06,
      "loss": 0.0019,
      "step": 119050
    },
    {
      "epoch": 0.7636138574172714,
      "grad_norm": 0.17291785776615143,
      "learning_rate": 7.769771651474817e-06,
      "loss": 0.0021,
      "step": 119060
    },
    {
      "epoch": 0.7636779943110575,
      "grad_norm": 0.04297932982444763,
      "learning_rate": 7.769305654725583e-06,
      "loss": 0.0018,
      "step": 119070
    },
    {
      "epoch": 0.7637421312048436,
      "grad_norm": 0.18742135167121887,
      "learning_rate": 7.768839623274967e-06,
      "loss": 0.002,
      "step": 119080
    },
    {
      "epoch": 0.7638062680986297,
      "grad_norm": 0.05562547594308853,
      "learning_rate": 7.768373557128812e-06,
      "loss": 0.0032,
      "step": 119090
    },
    {
      "epoch": 0.7638704049924158,
      "grad_norm": 0.44265061616897583,
      "learning_rate": 7.767907456292959e-06,
      "loss": 0.0018,
      "step": 119100
    },
    {
      "epoch": 0.7639345418862019,
      "grad_norm": 0.09151113778352737,
      "learning_rate": 7.767441320773246e-06,
      "loss": 0.0025,
      "step": 119110
    },
    {
      "epoch": 0.7639986787799881,
      "grad_norm": 0.13599392771720886,
      "learning_rate": 7.766975150575516e-06,
      "loss": 0.0037,
      "step": 119120
    },
    {
      "epoch": 0.7640628156737741,
      "grad_norm": 0.1208871528506279,
      "learning_rate": 7.76650894570561e-06,
      "loss": 0.0023,
      "step": 119130
    },
    {
      "epoch": 0.7641269525675602,
      "grad_norm": 0.1041627898812294,
      "learning_rate": 7.76604270616937e-06,
      "loss": 0.0058,
      "step": 119140
    },
    {
      "epoch": 0.7641910894613463,
      "grad_norm": 0.19829201698303223,
      "learning_rate": 7.765576431972637e-06,
      "loss": 0.0021,
      "step": 119150
    },
    {
      "epoch": 0.7642552263551324,
      "grad_norm": 0.19033165276050568,
      "learning_rate": 7.765110123121255e-06,
      "loss": 0.0042,
      "step": 119160
    },
    {
      "epoch": 0.7643193632489185,
      "grad_norm": 0.0724247470498085,
      "learning_rate": 7.764643779621069e-06,
      "loss": 0.0021,
      "step": 119170
    },
    {
      "epoch": 0.7643835001427046,
      "grad_norm": 0.1512024700641632,
      "learning_rate": 7.764177401477918e-06,
      "loss": 0.0017,
      "step": 119180
    },
    {
      "epoch": 0.7644476370364907,
      "grad_norm": 0.038425736129283905,
      "learning_rate": 7.76371098869765e-06,
      "loss": 0.0029,
      "step": 119190
    },
    {
      "epoch": 0.7645117739302768,
      "grad_norm": 0.06435366719961166,
      "learning_rate": 7.763244541286108e-06,
      "loss": 0.0031,
      "step": 119200
    },
    {
      "epoch": 0.7645759108240628,
      "grad_norm": 0.07712484896183014,
      "learning_rate": 7.762778059249136e-06,
      "loss": 0.0039,
      "step": 119210
    },
    {
      "epoch": 0.764640047717849,
      "grad_norm": 0.0393238440155983,
      "learning_rate": 7.762311542592579e-06,
      "loss": 0.0014,
      "step": 119220
    },
    {
      "epoch": 0.764704184611635,
      "grad_norm": 0.04551048204302788,
      "learning_rate": 7.761844991322287e-06,
      "loss": 0.0017,
      "step": 119230
    },
    {
      "epoch": 0.7647683215054212,
      "grad_norm": 0.01426377147436142,
      "learning_rate": 7.761378405444104e-06,
      "loss": 0.0026,
      "step": 119240
    },
    {
      "epoch": 0.7648324583992072,
      "grad_norm": 0.04326840117573738,
      "learning_rate": 7.760911784963874e-06,
      "loss": 0.0017,
      "step": 119250
    },
    {
      "epoch": 0.7648965952929934,
      "grad_norm": 0.16590778529644012,
      "learning_rate": 7.760445129887447e-06,
      "loss": 0.0045,
      "step": 119260
    },
    {
      "epoch": 0.7649607321867795,
      "grad_norm": 0.0777582973241806,
      "learning_rate": 7.75997844022067e-06,
      "loss": 0.0018,
      "step": 119270
    },
    {
      "epoch": 0.7650248690805656,
      "grad_norm": 0.08656816184520721,
      "learning_rate": 7.75951171596939e-06,
      "loss": 0.0024,
      "step": 119280
    },
    {
      "epoch": 0.7650890059743517,
      "grad_norm": 0.13881435990333557,
      "learning_rate": 7.759044957139456e-06,
      "loss": 0.0036,
      "step": 119290
    },
    {
      "epoch": 0.7651531428681377,
      "grad_norm": 0.19447311758995056,
      "learning_rate": 7.758578163736716e-06,
      "loss": 0.0016,
      "step": 119300
    },
    {
      "epoch": 0.7652172797619239,
      "grad_norm": 0.021220847964286804,
      "learning_rate": 7.758111335767021e-06,
      "loss": 0.0021,
      "step": 119310
    },
    {
      "epoch": 0.7652814166557099,
      "grad_norm": 0.04552415758371353,
      "learning_rate": 7.75764447323622e-06,
      "loss": 0.0019,
      "step": 119320
    },
    {
      "epoch": 0.7653455535494961,
      "grad_norm": 0.07058366388082504,
      "learning_rate": 7.757177576150164e-06,
      "loss": 0.0025,
      "step": 119330
    },
    {
      "epoch": 0.7654096904432821,
      "grad_norm": 0.04425279423594475,
      "learning_rate": 7.7567106445147e-06,
      "loss": 0.0021,
      "step": 119340
    },
    {
      "epoch": 0.7654738273370683,
      "grad_norm": 0.04987210035324097,
      "learning_rate": 7.756243678335681e-06,
      "loss": 0.0017,
      "step": 119350
    },
    {
      "epoch": 0.7655379642308543,
      "grad_norm": 0.07812074571847916,
      "learning_rate": 7.755776677618962e-06,
      "loss": 0.0032,
      "step": 119360
    },
    {
      "epoch": 0.7656021011246404,
      "grad_norm": 0.062241747975349426,
      "learning_rate": 7.75530964237039e-06,
      "loss": 0.0031,
      "step": 119370
    },
    {
      "epoch": 0.7656662380184265,
      "grad_norm": 0.5053013563156128,
      "learning_rate": 7.754842572595818e-06,
      "loss": 0.0034,
      "step": 119380
    },
    {
      "epoch": 0.7657303749122126,
      "grad_norm": 0.19220857322216034,
      "learning_rate": 7.754375468301103e-06,
      "loss": 0.0024,
      "step": 119390
    },
    {
      "epoch": 0.7657945118059988,
      "grad_norm": 0.1082582101225853,
      "learning_rate": 7.753908329492092e-06,
      "loss": 0.0031,
      "step": 119400
    },
    {
      "epoch": 0.7658586486997848,
      "grad_norm": 0.13472908735275269,
      "learning_rate": 7.753441156174642e-06,
      "loss": 0.002,
      "step": 119410
    },
    {
      "epoch": 0.765922785593571,
      "grad_norm": 0.08649571239948273,
      "learning_rate": 7.752973948354606e-06,
      "loss": 0.0025,
      "step": 119420
    },
    {
      "epoch": 0.765986922487357,
      "grad_norm": 0.1184421181678772,
      "learning_rate": 7.752506706037839e-06,
      "loss": 0.0019,
      "step": 119430
    },
    {
      "epoch": 0.7660510593811432,
      "grad_norm": 0.15916390717029572,
      "learning_rate": 7.752039429230197e-06,
      "loss": 0.0044,
      "step": 119440
    },
    {
      "epoch": 0.7661151962749292,
      "grad_norm": 0.051754746586084366,
      "learning_rate": 7.751572117937534e-06,
      "loss": 0.0031,
      "step": 119450
    },
    {
      "epoch": 0.7661793331687153,
      "grad_norm": 0.04972861707210541,
      "learning_rate": 7.751104772165704e-06,
      "loss": 0.0017,
      "step": 119460
    },
    {
      "epoch": 0.7662434700625014,
      "grad_norm": 0.08772239834070206,
      "learning_rate": 7.750637391920567e-06,
      "loss": 0.0015,
      "step": 119470
    },
    {
      "epoch": 0.7663076069562875,
      "grad_norm": 0.18049634993076324,
      "learning_rate": 7.750169977207977e-06,
      "loss": 0.0022,
      "step": 119480
    },
    {
      "epoch": 0.7663717438500736,
      "grad_norm": 0.46327391266822815,
      "learning_rate": 7.749702528033792e-06,
      "loss": 0.0018,
      "step": 119490
    },
    {
      "epoch": 0.7664358807438597,
      "grad_norm": 0.031109677627682686,
      "learning_rate": 7.749235044403869e-06,
      "loss": 0.0032,
      "step": 119500
    },
    {
      "epoch": 0.7665000176376457,
      "grad_norm": 0.052154701203107834,
      "learning_rate": 7.748767526324065e-06,
      "loss": 0.0027,
      "step": 119510
    },
    {
      "epoch": 0.7665641545314319,
      "grad_norm": 0.06638770550489426,
      "learning_rate": 7.74829997380024e-06,
      "loss": 0.0032,
      "step": 119520
    },
    {
      "epoch": 0.7666282914252179,
      "grad_norm": 0.3107570707798004,
      "learning_rate": 7.747832386838253e-06,
      "loss": 0.0023,
      "step": 119530
    },
    {
      "epoch": 0.7666924283190041,
      "grad_norm": 0.10106756538152695,
      "learning_rate": 7.74736476544396e-06,
      "loss": 0.0034,
      "step": 119540
    },
    {
      "epoch": 0.7667565652127902,
      "grad_norm": 0.10745019465684891,
      "learning_rate": 7.746897109623226e-06,
      "loss": 0.0031,
      "step": 119550
    },
    {
      "epoch": 0.7668207021065763,
      "grad_norm": 0.21665680408477783,
      "learning_rate": 7.746429419381906e-06,
      "loss": 0.0032,
      "step": 119560
    },
    {
      "epoch": 0.7668848390003624,
      "grad_norm": 0.3388836681842804,
      "learning_rate": 7.745961694725863e-06,
      "loss": 0.0025,
      "step": 119570
    },
    {
      "epoch": 0.7669489758941485,
      "grad_norm": 0.1119726300239563,
      "learning_rate": 7.745493935660957e-06,
      "loss": 0.002,
      "step": 119580
    },
    {
      "epoch": 0.7670131127879346,
      "grad_norm": 0.17020221054553986,
      "learning_rate": 7.745026142193051e-06,
      "loss": 0.0038,
      "step": 119590
    },
    {
      "epoch": 0.7670772496817206,
      "grad_norm": 0.07487424463033676,
      "learning_rate": 7.744558314328006e-06,
      "loss": 0.0013,
      "step": 119600
    },
    {
      "epoch": 0.7671413865755068,
      "grad_norm": 0.044883064925670624,
      "learning_rate": 7.744090452071682e-06,
      "loss": 0.0016,
      "step": 119610
    },
    {
      "epoch": 0.7672055234692928,
      "grad_norm": 0.2747569680213928,
      "learning_rate": 7.743622555429944e-06,
      "loss": 0.0033,
      "step": 119620
    },
    {
      "epoch": 0.767269660363079,
      "grad_norm": 0.22496001422405243,
      "learning_rate": 7.743154624408655e-06,
      "loss": 0.0055,
      "step": 119630
    },
    {
      "epoch": 0.767333797256865,
      "grad_norm": 0.08470924943685532,
      "learning_rate": 7.742686659013679e-06,
      "loss": 0.0013,
      "step": 119640
    },
    {
      "epoch": 0.7673979341506512,
      "grad_norm": 0.3732026517391205,
      "learning_rate": 7.742218659250877e-06,
      "loss": 0.0043,
      "step": 119650
    },
    {
      "epoch": 0.7674620710444372,
      "grad_norm": 0.08522894233465195,
      "learning_rate": 7.741750625126117e-06,
      "loss": 0.003,
      "step": 119660
    },
    {
      "epoch": 0.7675262079382233,
      "grad_norm": 0.038193341344594955,
      "learning_rate": 7.74128255664526e-06,
      "loss": 0.0017,
      "step": 119670
    },
    {
      "epoch": 0.7675903448320094,
      "grad_norm": 0.09714947640895844,
      "learning_rate": 7.740814453814177e-06,
      "loss": 0.0033,
      "step": 119680
    },
    {
      "epoch": 0.7676544817257955,
      "grad_norm": 0.13144101202487946,
      "learning_rate": 7.740346316638727e-06,
      "loss": 0.0034,
      "step": 119690
    },
    {
      "epoch": 0.7677186186195817,
      "grad_norm": 0.06684640794992447,
      "learning_rate": 7.739878145124779e-06,
      "loss": 0.0024,
      "step": 119700
    },
    {
      "epoch": 0.7677827555133677,
      "grad_norm": 0.0352996401488781,
      "learning_rate": 7.7394099392782e-06,
      "loss": 0.0027,
      "step": 119710
    },
    {
      "epoch": 0.7678468924071539,
      "grad_norm": 0.1148243248462677,
      "learning_rate": 7.738941699104859e-06,
      "loss": 0.0018,
      "step": 119720
    },
    {
      "epoch": 0.7679110293009399,
      "grad_norm": 0.10295584797859192,
      "learning_rate": 7.738473424610618e-06,
      "loss": 0.0033,
      "step": 119730
    },
    {
      "epoch": 0.767975166194726,
      "grad_norm": 0.26023539900779724,
      "learning_rate": 7.738005115801349e-06,
      "loss": 0.0022,
      "step": 119740
    },
    {
      "epoch": 0.7680393030885121,
      "grad_norm": 0.08188511431217194,
      "learning_rate": 7.73753677268292e-06,
      "loss": 0.0038,
      "step": 119750
    },
    {
      "epoch": 0.7681034399822982,
      "grad_norm": 0.10123003274202347,
      "learning_rate": 7.737068395261198e-06,
      "loss": 0.004,
      "step": 119760
    },
    {
      "epoch": 0.7681675768760843,
      "grad_norm": 0.14842653274536133,
      "learning_rate": 7.736599983542053e-06,
      "loss": 0.0032,
      "step": 119770
    },
    {
      "epoch": 0.7682317137698704,
      "grad_norm": 0.03348490595817566,
      "learning_rate": 7.736131537531354e-06,
      "loss": 0.0029,
      "step": 119780
    },
    {
      "epoch": 0.7682958506636565,
      "grad_norm": 0.22534605860710144,
      "learning_rate": 7.735663057234972e-06,
      "loss": 0.0027,
      "step": 119790
    },
    {
      "epoch": 0.7683599875574426,
      "grad_norm": 0.1607537865638733,
      "learning_rate": 7.735194542658776e-06,
      "loss": 0.0019,
      "step": 119800
    },
    {
      "epoch": 0.7684241244512287,
      "grad_norm": 0.1520402729511261,
      "learning_rate": 7.734725993808638e-06,
      "loss": 0.0053,
      "step": 119810
    },
    {
      "epoch": 0.7684882613450148,
      "grad_norm": 0.18805626034736633,
      "learning_rate": 7.734257410690429e-06,
      "loss": 0.003,
      "step": 119820
    },
    {
      "epoch": 0.768552398238801,
      "grad_norm": 0.2867084741592407,
      "learning_rate": 7.733788793310019e-06,
      "loss": 0.0022,
      "step": 119830
    },
    {
      "epoch": 0.768616535132587,
      "grad_norm": 0.07489742338657379,
      "learning_rate": 7.733320141673285e-06,
      "loss": 0.0034,
      "step": 119840
    },
    {
      "epoch": 0.7686806720263731,
      "grad_norm": 0.19532790780067444,
      "learning_rate": 7.732851455786092e-06,
      "loss": 0.003,
      "step": 119850
    },
    {
      "epoch": 0.7687448089201592,
      "grad_norm": 0.06110703945159912,
      "learning_rate": 7.73238273565432e-06,
      "loss": 0.0047,
      "step": 119860
    },
    {
      "epoch": 0.7688089458139453,
      "grad_norm": 0.057539068162441254,
      "learning_rate": 7.731913981283838e-06,
      "loss": 0.0035,
      "step": 119870
    },
    {
      "epoch": 0.7688730827077314,
      "grad_norm": 0.12065283209085464,
      "learning_rate": 7.731445192680522e-06,
      "loss": 0.0025,
      "step": 119880
    },
    {
      "epoch": 0.7689372196015175,
      "grad_norm": 0.20543505251407623,
      "learning_rate": 7.730976369850245e-06,
      "loss": 0.002,
      "step": 119890
    },
    {
      "epoch": 0.7690013564953035,
      "grad_norm": 0.03451468423008919,
      "learning_rate": 7.730507512798883e-06,
      "loss": 0.0024,
      "step": 119900
    },
    {
      "epoch": 0.7690654933890897,
      "grad_norm": 0.1505160927772522,
      "learning_rate": 7.730038621532312e-06,
      "loss": 0.0026,
      "step": 119910
    },
    {
      "epoch": 0.7691296302828757,
      "grad_norm": 0.020788371562957764,
      "learning_rate": 7.729569696056404e-06,
      "loss": 0.0023,
      "step": 119920
    },
    {
      "epoch": 0.7691937671766619,
      "grad_norm": 0.05163104459643364,
      "learning_rate": 7.729100736377036e-06,
      "loss": 0.0026,
      "step": 119930
    },
    {
      "epoch": 0.7692579040704479,
      "grad_norm": 0.15215818583965302,
      "learning_rate": 7.728631742500088e-06,
      "loss": 0.0027,
      "step": 119940
    },
    {
      "epoch": 0.7693220409642341,
      "grad_norm": 0.4038577377796173,
      "learning_rate": 7.728162714431431e-06,
      "loss": 0.0017,
      "step": 119950
    },
    {
      "epoch": 0.7693861778580201,
      "grad_norm": 0.11586163938045502,
      "learning_rate": 7.727693652176948e-06,
      "loss": 0.0037,
      "step": 119960
    },
    {
      "epoch": 0.7694503147518063,
      "grad_norm": 0.13993075489997864,
      "learning_rate": 7.727224555742513e-06,
      "loss": 0.0026,
      "step": 119970
    },
    {
      "epoch": 0.7695144516455924,
      "grad_norm": 0.08906079828739166,
      "learning_rate": 7.726755425134006e-06,
      "loss": 0.0027,
      "step": 119980
    },
    {
      "epoch": 0.7695785885393784,
      "grad_norm": 0.09948492795228958,
      "learning_rate": 7.726286260357304e-06,
      "loss": 0.0027,
      "step": 119990
    },
    {
      "epoch": 0.7696427254331646,
      "grad_norm": 0.05504319444298744,
      "learning_rate": 7.725817061418286e-06,
      "loss": 0.0047,
      "step": 120000
    },
    {
      "epoch": 0.7697068623269506,
      "grad_norm": 0.15614096820354462,
      "learning_rate": 7.725347828322834e-06,
      "loss": 0.0031,
      "step": 120010
    },
    {
      "epoch": 0.7697709992207368,
      "grad_norm": 0.04980633407831192,
      "learning_rate": 7.724878561076822e-06,
      "loss": 0.0015,
      "step": 120020
    },
    {
      "epoch": 0.7698351361145228,
      "grad_norm": 0.2578064799308777,
      "learning_rate": 7.724409259686139e-06,
      "loss": 0.0027,
      "step": 120030
    },
    {
      "epoch": 0.769899273008309,
      "grad_norm": 0.08008996397256851,
      "learning_rate": 7.723939924156658e-06,
      "loss": 0.002,
      "step": 120040
    },
    {
      "epoch": 0.769963409902095,
      "grad_norm": 0.2172783762216568,
      "learning_rate": 7.723470554494264e-06,
      "loss": 0.0043,
      "step": 120050
    },
    {
      "epoch": 0.7700275467958811,
      "grad_norm": 0.09386609494686127,
      "learning_rate": 7.723001150704837e-06,
      "loss": 0.0046,
      "step": 120060
    },
    {
      "epoch": 0.7700916836896672,
      "grad_norm": 0.1730341613292694,
      "learning_rate": 7.722531712794262e-06,
      "loss": 0.002,
      "step": 120070
    },
    {
      "epoch": 0.7701558205834533,
      "grad_norm": 0.055997759103775024,
      "learning_rate": 7.722062240768413e-06,
      "loss": 0.0058,
      "step": 120080
    },
    {
      "epoch": 0.7702199574772394,
      "grad_norm": 0.0635523796081543,
      "learning_rate": 7.721592734633183e-06,
      "loss": 0.0021,
      "step": 120090
    },
    {
      "epoch": 0.7702840943710255,
      "grad_norm": 0.1634179949760437,
      "learning_rate": 7.72112319439445e-06,
      "loss": 0.0024,
      "step": 120100
    },
    {
      "epoch": 0.7703482312648117,
      "grad_norm": 0.08533147722482681,
      "learning_rate": 7.720653620058101e-06,
      "loss": 0.002,
      "step": 120110
    },
    {
      "epoch": 0.7704123681585977,
      "grad_norm": 0.2712162137031555,
      "learning_rate": 7.720184011630014e-06,
      "loss": 0.0045,
      "step": 120120
    },
    {
      "epoch": 0.7704765050523839,
      "grad_norm": 0.08651968836784363,
      "learning_rate": 7.71971436911608e-06,
      "loss": 0.0033,
      "step": 120130
    },
    {
      "epoch": 0.7705406419461699,
      "grad_norm": 0.07410996407270432,
      "learning_rate": 7.71924469252218e-06,
      "loss": 0.0038,
      "step": 120140
    },
    {
      "epoch": 0.770604778839956,
      "grad_norm": 0.07435926795005798,
      "learning_rate": 7.7187749818542e-06,
      "loss": 0.0032,
      "step": 120150
    },
    {
      "epoch": 0.7706689157337421,
      "grad_norm": 0.33352407813072205,
      "learning_rate": 7.718305237118028e-06,
      "loss": 0.0034,
      "step": 120160
    },
    {
      "epoch": 0.7707330526275282,
      "grad_norm": 0.1434878557920456,
      "learning_rate": 7.717835458319546e-06,
      "loss": 0.0031,
      "step": 120170
    },
    {
      "epoch": 0.7707971895213143,
      "grad_norm": 0.19026531279087067,
      "learning_rate": 7.717365645464646e-06,
      "loss": 0.0021,
      "step": 120180
    },
    {
      "epoch": 0.7708613264151004,
      "grad_norm": 0.1479296088218689,
      "learning_rate": 7.716895798559212e-06,
      "loss": 0.0019,
      "step": 120190
    },
    {
      "epoch": 0.7709254633088864,
      "grad_norm": 0.1592334359884262,
      "learning_rate": 7.716425917609131e-06,
      "loss": 0.0021,
      "step": 120200
    },
    {
      "epoch": 0.7709896002026726,
      "grad_norm": 0.24233010411262512,
      "learning_rate": 7.715956002620293e-06,
      "loss": 0.0028,
      "step": 120210
    },
    {
      "epoch": 0.7710537370964586,
      "grad_norm": 0.11257760226726532,
      "learning_rate": 7.715486053598584e-06,
      "loss": 0.0022,
      "step": 120220
    },
    {
      "epoch": 0.7711178739902448,
      "grad_norm": 0.11522355675697327,
      "learning_rate": 7.715016070549895e-06,
      "loss": 0.0032,
      "step": 120230
    },
    {
      "epoch": 0.7711820108840308,
      "grad_norm": 0.1420866847038269,
      "learning_rate": 7.714546053480111e-06,
      "loss": 0.0022,
      "step": 120240
    },
    {
      "epoch": 0.771246147777817,
      "grad_norm": 0.1342962086200714,
      "learning_rate": 7.71407600239513e-06,
      "loss": 0.0015,
      "step": 120250
    },
    {
      "epoch": 0.7713102846716031,
      "grad_norm": 0.06081683188676834,
      "learning_rate": 7.713605917300834e-06,
      "loss": 0.0026,
      "step": 120260
    },
    {
      "epoch": 0.7713744215653892,
      "grad_norm": 0.1337926685810089,
      "learning_rate": 7.713135798203117e-06,
      "loss": 0.0065,
      "step": 120270
    },
    {
      "epoch": 0.7714385584591753,
      "grad_norm": 0.23072844743728638,
      "learning_rate": 7.71266564510787e-06,
      "loss": 0.003,
      "step": 120280
    },
    {
      "epoch": 0.7715026953529613,
      "grad_norm": 0.12262122333049774,
      "learning_rate": 7.712195458020983e-06,
      "loss": 0.0022,
      "step": 120290
    },
    {
      "epoch": 0.7715668322467475,
      "grad_norm": 0.1738017052412033,
      "learning_rate": 7.711725236948349e-06,
      "loss": 0.002,
      "step": 120300
    },
    {
      "epoch": 0.7716309691405335,
      "grad_norm": 0.31155627965927124,
      "learning_rate": 7.711254981895859e-06,
      "loss": 0.0018,
      "step": 120310
    },
    {
      "epoch": 0.7716951060343197,
      "grad_norm": 0.05566750839352608,
      "learning_rate": 7.710784692869406e-06,
      "loss": 0.003,
      "step": 120320
    },
    {
      "epoch": 0.7717592429281057,
      "grad_norm": 0.20950689911842346,
      "learning_rate": 7.710314369874885e-06,
      "loss": 0.0028,
      "step": 120330
    },
    {
      "epoch": 0.7718233798218919,
      "grad_norm": 0.15361624956130981,
      "learning_rate": 7.709844012918187e-06,
      "loss": 0.0014,
      "step": 120340
    },
    {
      "epoch": 0.7718875167156779,
      "grad_norm": 0.05247703939676285,
      "learning_rate": 7.709373622005208e-06,
      "loss": 0.0021,
      "step": 120350
    },
    {
      "epoch": 0.771951653609464,
      "grad_norm": 0.09840486198663712,
      "learning_rate": 7.70890319714184e-06,
      "loss": 0.0023,
      "step": 120360
    },
    {
      "epoch": 0.7720157905032501,
      "grad_norm": 0.10478629171848297,
      "learning_rate": 7.708432738333978e-06,
      "loss": 0.0023,
      "step": 120370
    },
    {
      "epoch": 0.7720799273970362,
      "grad_norm": 0.14441901445388794,
      "learning_rate": 7.707962245587519e-06,
      "loss": 0.0024,
      "step": 120380
    },
    {
      "epoch": 0.7721440642908224,
      "grad_norm": 0.0552758127450943,
      "learning_rate": 7.707491718908357e-06,
      "loss": 0.0024,
      "step": 120390
    },
    {
      "epoch": 0.7722082011846084,
      "grad_norm": 0.11776135116815567,
      "learning_rate": 7.70702115830239e-06,
      "loss": 0.0032,
      "step": 120400
    },
    {
      "epoch": 0.7722723380783946,
      "grad_norm": 0.08400656282901764,
      "learning_rate": 7.706550563775514e-06,
      "loss": 0.0023,
      "step": 120410
    },
    {
      "epoch": 0.7723364749721806,
      "grad_norm": 0.23327843844890594,
      "learning_rate": 7.706079935333623e-06,
      "loss": 0.0035,
      "step": 120420
    },
    {
      "epoch": 0.7724006118659668,
      "grad_norm": 0.13265545666217804,
      "learning_rate": 7.705609272982618e-06,
      "loss": 0.0024,
      "step": 120430
    },
    {
      "epoch": 0.7724647487597528,
      "grad_norm": 0.2691594362258911,
      "learning_rate": 7.705138576728395e-06,
      "loss": 0.0029,
      "step": 120440
    },
    {
      "epoch": 0.7725288856535389,
      "grad_norm": 0.12558777630329132,
      "learning_rate": 7.704667846576851e-06,
      "loss": 0.0043,
      "step": 120450
    },
    {
      "epoch": 0.772593022547325,
      "grad_norm": 0.1107628121972084,
      "learning_rate": 7.704197082533886e-06,
      "loss": 0.0028,
      "step": 120460
    },
    {
      "epoch": 0.7726571594411111,
      "grad_norm": 0.125325545668602,
      "learning_rate": 7.7037262846054e-06,
      "loss": 0.0029,
      "step": 120470
    },
    {
      "epoch": 0.7727212963348972,
      "grad_norm": 0.3176846504211426,
      "learning_rate": 7.70325545279729e-06,
      "loss": 0.004,
      "step": 120480
    },
    {
      "epoch": 0.7727854332286833,
      "grad_norm": 0.02290859818458557,
      "learning_rate": 7.702784587115458e-06,
      "loss": 0.0029,
      "step": 120490
    },
    {
      "epoch": 0.7728495701224694,
      "grad_norm": 0.14008145034313202,
      "learning_rate": 7.702313687565803e-06,
      "loss": 0.0023,
      "step": 120500
    },
    {
      "epoch": 0.7729137070162555,
      "grad_norm": 0.04517088830471039,
      "learning_rate": 7.701842754154227e-06,
      "loss": 0.0017,
      "step": 120510
    },
    {
      "epoch": 0.7729778439100415,
      "grad_norm": 0.004795001354068518,
      "learning_rate": 7.701371786886631e-06,
      "loss": 0.0025,
      "step": 120520
    },
    {
      "epoch": 0.7730419808038277,
      "grad_norm": 0.11474525183439255,
      "learning_rate": 7.700900785768914e-06,
      "loss": 0.0018,
      "step": 120530
    },
    {
      "epoch": 0.7731061176976138,
      "grad_norm": 0.18473678827285767,
      "learning_rate": 7.700429750806982e-06,
      "loss": 0.0026,
      "step": 120540
    },
    {
      "epoch": 0.7731702545913999,
      "grad_norm": 0.1098489761352539,
      "learning_rate": 7.699958682006734e-06,
      "loss": 0.0016,
      "step": 120550
    },
    {
      "epoch": 0.773234391485186,
      "grad_norm": 0.12431825697422028,
      "learning_rate": 7.699487579374074e-06,
      "loss": 0.0031,
      "step": 120560
    },
    {
      "epoch": 0.7732985283789721,
      "grad_norm": 0.14041979610919952,
      "learning_rate": 7.699016442914904e-06,
      "loss": 0.0017,
      "step": 120570
    },
    {
      "epoch": 0.7733626652727582,
      "grad_norm": 0.12669505178928375,
      "learning_rate": 7.698545272635132e-06,
      "loss": 0.0035,
      "step": 120580
    },
    {
      "epoch": 0.7734268021665442,
      "grad_norm": 0.0775560587644577,
      "learning_rate": 7.698074068540656e-06,
      "loss": 0.0013,
      "step": 120590
    },
    {
      "epoch": 0.7734909390603304,
      "grad_norm": 0.0368109866976738,
      "learning_rate": 7.697602830637386e-06,
      "loss": 0.0015,
      "step": 120600
    },
    {
      "epoch": 0.7735550759541164,
      "grad_norm": 0.15224739909172058,
      "learning_rate": 7.697131558931224e-06,
      "loss": 0.0033,
      "step": 120610
    },
    {
      "epoch": 0.7736192128479026,
      "grad_norm": 0.0786733403801918,
      "learning_rate": 7.696660253428076e-06,
      "loss": 0.0028,
      "step": 120620
    },
    {
      "epoch": 0.7736833497416886,
      "grad_norm": 0.16595987975597382,
      "learning_rate": 7.696188914133847e-06,
      "loss": 0.0028,
      "step": 120630
    },
    {
      "epoch": 0.7737474866354748,
      "grad_norm": 0.04862990230321884,
      "learning_rate": 7.695717541054445e-06,
      "loss": 0.0029,
      "step": 120640
    },
    {
      "epoch": 0.7738116235292608,
      "grad_norm": 0.09698370099067688,
      "learning_rate": 7.695246134195773e-06,
      "loss": 0.0017,
      "step": 120650
    },
    {
      "epoch": 0.773875760423047,
      "grad_norm": 0.10276313871145248,
      "learning_rate": 7.694774693563744e-06,
      "loss": 0.0022,
      "step": 120660
    },
    {
      "epoch": 0.7739398973168331,
      "grad_norm": 0.256753534078598,
      "learning_rate": 7.694303219164261e-06,
      "loss": 0.0034,
      "step": 120670
    },
    {
      "epoch": 0.7740040342106191,
      "grad_norm": 0.09404375404119492,
      "learning_rate": 7.693831711003233e-06,
      "loss": 0.0028,
      "step": 120680
    },
    {
      "epoch": 0.7740681711044053,
      "grad_norm": 0.3127039074897766,
      "learning_rate": 7.693360169086567e-06,
      "loss": 0.0037,
      "step": 120690
    },
    {
      "epoch": 0.7741323079981913,
      "grad_norm": 0.2723565101623535,
      "learning_rate": 7.692888593420176e-06,
      "loss": 0.004,
      "step": 120700
    },
    {
      "epoch": 0.7741964448919775,
      "grad_norm": 0.1416441649198532,
      "learning_rate": 7.692416984009965e-06,
      "loss": 0.0027,
      "step": 120710
    },
    {
      "epoch": 0.7742605817857635,
      "grad_norm": 0.048688746988773346,
      "learning_rate": 7.691945340861843e-06,
      "loss": 0.0037,
      "step": 120720
    },
    {
      "epoch": 0.7743247186795497,
      "grad_norm": 0.24894799292087555,
      "learning_rate": 7.691473663981726e-06,
      "loss": 0.0019,
      "step": 120730
    },
    {
      "epoch": 0.7743888555733357,
      "grad_norm": 0.07757014036178589,
      "learning_rate": 7.691001953375517e-06,
      "loss": 0.0033,
      "step": 120740
    },
    {
      "epoch": 0.7744529924671218,
      "grad_norm": 0.22163419425487518,
      "learning_rate": 7.690530209049131e-06,
      "loss": 0.003,
      "step": 120750
    },
    {
      "epoch": 0.7745171293609079,
      "grad_norm": 0.09050728380680084,
      "learning_rate": 7.69005843100848e-06,
      "loss": 0.0068,
      "step": 120760
    },
    {
      "epoch": 0.774581266254694,
      "grad_norm": 0.2518390417098999,
      "learning_rate": 7.689586619259474e-06,
      "loss": 0.0031,
      "step": 120770
    },
    {
      "epoch": 0.7746454031484801,
      "grad_norm": 0.0962381511926651,
      "learning_rate": 7.689114773808024e-06,
      "loss": 0.0023,
      "step": 120780
    },
    {
      "epoch": 0.7747095400422662,
      "grad_norm": 0.08857965469360352,
      "learning_rate": 7.688642894660044e-06,
      "loss": 0.0046,
      "step": 120790
    },
    {
      "epoch": 0.7747736769360523,
      "grad_norm": 0.1404399424791336,
      "learning_rate": 7.68817098182145e-06,
      "loss": 0.0019,
      "step": 120800
    },
    {
      "epoch": 0.7748378138298384,
      "grad_norm": 0.13727347552776337,
      "learning_rate": 7.687699035298148e-06,
      "loss": 0.0021,
      "step": 120810
    },
    {
      "epoch": 0.7749019507236246,
      "grad_norm": 0.036177147179841995,
      "learning_rate": 7.68722705509606e-06,
      "loss": 0.0042,
      "step": 120820
    },
    {
      "epoch": 0.7749660876174106,
      "grad_norm": 0.06073131412267685,
      "learning_rate": 7.686755041221095e-06,
      "loss": 0.0026,
      "step": 120830
    },
    {
      "epoch": 0.7750302245111967,
      "grad_norm": 0.07920609414577484,
      "learning_rate": 7.686282993679169e-06,
      "loss": 0.0019,
      "step": 120840
    },
    {
      "epoch": 0.7750943614049828,
      "grad_norm": 0.18040022253990173,
      "learning_rate": 7.685810912476194e-06,
      "loss": 0.0026,
      "step": 120850
    },
    {
      "epoch": 0.7751584982987689,
      "grad_norm": 0.19039228558540344,
      "learning_rate": 7.685338797618093e-06,
      "loss": 0.0018,
      "step": 120860
    },
    {
      "epoch": 0.775222635192555,
      "grad_norm": 0.04722949117422104,
      "learning_rate": 7.684866649110779e-06,
      "loss": 0.0019,
      "step": 120870
    },
    {
      "epoch": 0.7752867720863411,
      "grad_norm": 0.13327927887439728,
      "learning_rate": 7.684394466960164e-06,
      "loss": 0.0032,
      "step": 120880
    },
    {
      "epoch": 0.7753509089801272,
      "grad_norm": 0.10673610866069794,
      "learning_rate": 7.683922251172169e-06,
      "loss": 0.0024,
      "step": 120890
    },
    {
      "epoch": 0.7754150458739133,
      "grad_norm": 0.18592707812786102,
      "learning_rate": 7.683450001752708e-06,
      "loss": 0.0033,
      "step": 120900
    },
    {
      "epoch": 0.7754791827676993,
      "grad_norm": 0.1468019187450409,
      "learning_rate": 7.682977718707703e-06,
      "loss": 0.0018,
      "step": 120910
    },
    {
      "epoch": 0.7755433196614855,
      "grad_norm": 0.11070624738931656,
      "learning_rate": 7.682505402043069e-06,
      "loss": 0.0026,
      "step": 120920
    },
    {
      "epoch": 0.7756074565552715,
      "grad_norm": 0.014107703231275082,
      "learning_rate": 7.682033051764725e-06,
      "loss": 0.0017,
      "step": 120930
    },
    {
      "epoch": 0.7756715934490577,
      "grad_norm": 0.24959653615951538,
      "learning_rate": 7.681560667878591e-06,
      "loss": 0.0025,
      "step": 120940
    },
    {
      "epoch": 0.7757357303428438,
      "grad_norm": 0.061088040471076965,
      "learning_rate": 7.681088250390583e-06,
      "loss": 0.0016,
      "step": 120950
    },
    {
      "epoch": 0.7757998672366299,
      "grad_norm": 0.2523617148399353,
      "learning_rate": 7.680615799306625e-06,
      "loss": 0.0034,
      "step": 120960
    },
    {
      "epoch": 0.775864004130416,
      "grad_norm": 0.10077609121799469,
      "learning_rate": 7.680143314632635e-06,
      "loss": 0.0027,
      "step": 120970
    },
    {
      "epoch": 0.775928141024202,
      "grad_norm": 0.23623411357402802,
      "learning_rate": 7.679670796374534e-06,
      "loss": 0.002,
      "step": 120980
    },
    {
      "epoch": 0.7759922779179882,
      "grad_norm": 0.20702892541885376,
      "learning_rate": 7.679198244538241e-06,
      "loss": 0.0021,
      "step": 120990
    },
    {
      "epoch": 0.7760564148117742,
      "grad_norm": 0.1016428992152214,
      "learning_rate": 7.67872565912968e-06,
      "loss": 0.003,
      "step": 121000
    },
    {
      "epoch": 0.7761205517055604,
      "grad_norm": 0.23183010518550873,
      "learning_rate": 7.678253040154775e-06,
      "loss": 0.0042,
      "step": 121010
    },
    {
      "epoch": 0.7761846885993464,
      "grad_norm": 0.11860893666744232,
      "learning_rate": 7.677780387619443e-06,
      "loss": 0.0026,
      "step": 121020
    },
    {
      "epoch": 0.7762488254931326,
      "grad_norm": 0.10921378433704376,
      "learning_rate": 7.677307701529608e-06,
      "loss": 0.0023,
      "step": 121030
    },
    {
      "epoch": 0.7763129623869186,
      "grad_norm": 0.018968788906931877,
      "learning_rate": 7.676834981891194e-06,
      "loss": 0.001,
      "step": 121040
    },
    {
      "epoch": 0.7763770992807048,
      "grad_norm": 0.14218148589134216,
      "learning_rate": 7.676362228710125e-06,
      "loss": 0.0027,
      "step": 121050
    },
    {
      "epoch": 0.7764412361744908,
      "grad_norm": 0.11908269673585892,
      "learning_rate": 7.675889441992326e-06,
      "loss": 0.0032,
      "step": 121060
    },
    {
      "epoch": 0.7765053730682769,
      "grad_norm": 0.18243275582790375,
      "learning_rate": 7.675416621743718e-06,
      "loss": 0.0056,
      "step": 121070
    },
    {
      "epoch": 0.776569509962063,
      "grad_norm": 0.10393267124891281,
      "learning_rate": 7.674943767970229e-06,
      "loss": 0.0018,
      "step": 121080
    },
    {
      "epoch": 0.7766336468558491,
      "grad_norm": 0.2885073125362396,
      "learning_rate": 7.674470880677784e-06,
      "loss": 0.0058,
      "step": 121090
    },
    {
      "epoch": 0.7766977837496353,
      "grad_norm": 0.21264208853244781,
      "learning_rate": 7.673997959872305e-06,
      "loss": 0.0016,
      "step": 121100
    },
    {
      "epoch": 0.7767619206434213,
      "grad_norm": 0.1224047988653183,
      "learning_rate": 7.673525005559721e-06,
      "loss": 0.0028,
      "step": 121110
    },
    {
      "epoch": 0.7768260575372075,
      "grad_norm": 0.19190648198127747,
      "learning_rate": 7.673052017745958e-06,
      "loss": 0.0073,
      "step": 121120
    },
    {
      "epoch": 0.7768901944309935,
      "grad_norm": 0.08077745884656906,
      "learning_rate": 7.672578996436943e-06,
      "loss": 0.0035,
      "step": 121130
    },
    {
      "epoch": 0.7769543313247796,
      "grad_norm": 0.2599153518676758,
      "learning_rate": 7.672105941638604e-06,
      "loss": 0.0052,
      "step": 121140
    },
    {
      "epoch": 0.7770184682185657,
      "grad_norm": 0.10700134187936783,
      "learning_rate": 7.671632853356865e-06,
      "loss": 0.0023,
      "step": 121150
    },
    {
      "epoch": 0.7770826051123518,
      "grad_norm": 0.04627247527241707,
      "learning_rate": 7.67115973159766e-06,
      "loss": 0.004,
      "step": 121160
    },
    {
      "epoch": 0.7771467420061379,
      "grad_norm": 0.11000658571720123,
      "learning_rate": 7.670686576366912e-06,
      "loss": 0.002,
      "step": 121170
    },
    {
      "epoch": 0.777210878899924,
      "grad_norm": 0.2015426605939865,
      "learning_rate": 7.670213387670555e-06,
      "loss": 0.0028,
      "step": 121180
    },
    {
      "epoch": 0.7772750157937101,
      "grad_norm": 0.07996021211147308,
      "learning_rate": 7.669740165514514e-06,
      "loss": 0.0015,
      "step": 121190
    },
    {
      "epoch": 0.7773391526874962,
      "grad_norm": 0.15272463858127594,
      "learning_rate": 7.669266909904722e-06,
      "loss": 0.0016,
      "step": 121200
    },
    {
      "epoch": 0.7774032895812822,
      "grad_norm": 0.19971288740634918,
      "learning_rate": 7.668793620847108e-06,
      "loss": 0.0018,
      "step": 121210
    },
    {
      "epoch": 0.7774674264750684,
      "grad_norm": 0.28225573897361755,
      "learning_rate": 7.668320298347602e-06,
      "loss": 0.0016,
      "step": 121220
    },
    {
      "epoch": 0.7775315633688544,
      "grad_norm": 0.03997465595602989,
      "learning_rate": 7.667846942412136e-06,
      "loss": 0.0036,
      "step": 121230
    },
    {
      "epoch": 0.7775957002626406,
      "grad_norm": 0.1938021183013916,
      "learning_rate": 7.667373553046639e-06,
      "loss": 0.0039,
      "step": 121240
    },
    {
      "epoch": 0.7776598371564267,
      "grad_norm": 0.18448276817798615,
      "learning_rate": 7.666900130257046e-06,
      "loss": 0.0036,
      "step": 121250
    },
    {
      "epoch": 0.7777239740502128,
      "grad_norm": 0.02420881949365139,
      "learning_rate": 7.666426674049291e-06,
      "loss": 0.003,
      "step": 121260
    },
    {
      "epoch": 0.7777881109439989,
      "grad_norm": 0.04524382948875427,
      "learning_rate": 7.665953184429302e-06,
      "loss": 0.0016,
      "step": 121270
    },
    {
      "epoch": 0.777852247837785,
      "grad_norm": 0.08701635897159576,
      "learning_rate": 7.665479661403014e-06,
      "loss": 0.0026,
      "step": 121280
    },
    {
      "epoch": 0.7779163847315711,
      "grad_norm": 0.14590567350387573,
      "learning_rate": 7.665006104976363e-06,
      "loss": 0.0034,
      "step": 121290
    },
    {
      "epoch": 0.7779805216253571,
      "grad_norm": 0.21900488436222076,
      "learning_rate": 7.66453251515528e-06,
      "loss": 0.0024,
      "step": 121300
    },
    {
      "epoch": 0.7780446585191433,
      "grad_norm": 0.07465245574712753,
      "learning_rate": 7.664058891945699e-06,
      "loss": 0.0039,
      "step": 121310
    },
    {
      "epoch": 0.7781087954129293,
      "grad_norm": 0.20187869668006897,
      "learning_rate": 7.663585235353555e-06,
      "loss": 0.0025,
      "step": 121320
    },
    {
      "epoch": 0.7781729323067155,
      "grad_norm": 0.06113561615347862,
      "learning_rate": 7.663111545384787e-06,
      "loss": 0.0035,
      "step": 121330
    },
    {
      "epoch": 0.7782370692005015,
      "grad_norm": 0.2926420271396637,
      "learning_rate": 7.662637822045326e-06,
      "loss": 0.0032,
      "step": 121340
    },
    {
      "epoch": 0.7783012060942877,
      "grad_norm": 0.10521459579467773,
      "learning_rate": 7.662164065341112e-06,
      "loss": 0.003,
      "step": 121350
    },
    {
      "epoch": 0.7783653429880737,
      "grad_norm": 0.037617068737745285,
      "learning_rate": 7.661690275278077e-06,
      "loss": 0.0023,
      "step": 121360
    },
    {
      "epoch": 0.7784294798818598,
      "grad_norm": 0.10023058205842972,
      "learning_rate": 7.661216451862163e-06,
      "loss": 0.0018,
      "step": 121370
    },
    {
      "epoch": 0.778493616775646,
      "grad_norm": 0.1474991738796234,
      "learning_rate": 7.660742595099303e-06,
      "loss": 0.0016,
      "step": 121380
    },
    {
      "epoch": 0.778557753669432,
      "grad_norm": 0.0758986547589302,
      "learning_rate": 7.66026870499544e-06,
      "loss": 0.0048,
      "step": 121390
    },
    {
      "epoch": 0.7786218905632182,
      "grad_norm": 0.199615478515625,
      "learning_rate": 7.659794781556507e-06,
      "loss": 0.0022,
      "step": 121400
    },
    {
      "epoch": 0.7786860274570042,
      "grad_norm": 0.19785688817501068,
      "learning_rate": 7.659320824788443e-06,
      "loss": 0.0033,
      "step": 121410
    },
    {
      "epoch": 0.7787501643507904,
      "grad_norm": 0.17966535687446594,
      "learning_rate": 7.65884683469719e-06,
      "loss": 0.0034,
      "step": 121420
    },
    {
      "epoch": 0.7788143012445764,
      "grad_norm": 0.1133250966668129,
      "learning_rate": 7.658372811288687e-06,
      "loss": 0.0027,
      "step": 121430
    },
    {
      "epoch": 0.7788784381383625,
      "grad_norm": 0.13527756929397583,
      "learning_rate": 7.65789875456887e-06,
      "loss": 0.0019,
      "step": 121440
    },
    {
      "epoch": 0.7789425750321486,
      "grad_norm": 0.033486563712358475,
      "learning_rate": 7.657424664543684e-06,
      "loss": 0.0026,
      "step": 121450
    },
    {
      "epoch": 0.7790067119259347,
      "grad_norm": 0.3420703113079071,
      "learning_rate": 7.656950541219069e-06,
      "loss": 0.0023,
      "step": 121460
    },
    {
      "epoch": 0.7790708488197208,
      "grad_norm": 0.16449497640132904,
      "learning_rate": 7.65647638460096e-06,
      "loss": 0.0029,
      "step": 121470
    },
    {
      "epoch": 0.7791349857135069,
      "grad_norm": 0.07003653794527054,
      "learning_rate": 7.656002194695308e-06,
      "loss": 0.0026,
      "step": 121480
    },
    {
      "epoch": 0.779199122607293,
      "grad_norm": 0.11345074325799942,
      "learning_rate": 7.655527971508048e-06,
      "loss": 0.0037,
      "step": 121490
    },
    {
      "epoch": 0.7792632595010791,
      "grad_norm": 0.1076662465929985,
      "learning_rate": 7.655053715045126e-06,
      "loss": 0.0024,
      "step": 121500
    },
    {
      "epoch": 0.7793273963948651,
      "grad_norm": 0.07580321282148361,
      "learning_rate": 7.654579425312482e-06,
      "loss": 0.0014,
      "step": 121510
    },
    {
      "epoch": 0.7793915332886513,
      "grad_norm": 0.04696830362081528,
      "learning_rate": 7.654105102316063e-06,
      "loss": 0.0055,
      "step": 121520
    },
    {
      "epoch": 0.7794556701824374,
      "grad_norm": 0.060965411365032196,
      "learning_rate": 7.653630746061807e-06,
      "loss": 0.003,
      "step": 121530
    },
    {
      "epoch": 0.7795198070762235,
      "grad_norm": 0.03723333030939102,
      "learning_rate": 7.653156356555662e-06,
      "loss": 0.0028,
      "step": 121540
    },
    {
      "epoch": 0.7795839439700096,
      "grad_norm": 0.13844603300094604,
      "learning_rate": 7.652681933803573e-06,
      "loss": 0.0031,
      "step": 121550
    },
    {
      "epoch": 0.7796480808637957,
      "grad_norm": 0.1367502212524414,
      "learning_rate": 7.652207477811484e-06,
      "loss": 0.0019,
      "step": 121560
    },
    {
      "epoch": 0.7797122177575818,
      "grad_norm": 0.13187959790229797,
      "learning_rate": 7.651732988585338e-06,
      "loss": 0.0022,
      "step": 121570
    },
    {
      "epoch": 0.7797763546513679,
      "grad_norm": 0.13078394532203674,
      "learning_rate": 7.651258466131083e-06,
      "loss": 0.0024,
      "step": 121580
    },
    {
      "epoch": 0.779840491545154,
      "grad_norm": 0.1649218499660492,
      "learning_rate": 7.650783910454666e-06,
      "loss": 0.0022,
      "step": 121590
    },
    {
      "epoch": 0.77990462843894,
      "grad_norm": 0.09971890598535538,
      "learning_rate": 7.65030932156203e-06,
      "loss": 0.0026,
      "step": 121600
    },
    {
      "epoch": 0.7799687653327262,
      "grad_norm": 0.03387978672981262,
      "learning_rate": 7.649834699459124e-06,
      "loss": 0.002,
      "step": 121610
    },
    {
      "epoch": 0.7800329022265122,
      "grad_norm": 0.011331469751894474,
      "learning_rate": 7.649360044151896e-06,
      "loss": 0.0023,
      "step": 121620
    },
    {
      "epoch": 0.7800970391202984,
      "grad_norm": 0.3146549165248871,
      "learning_rate": 7.648885355646295e-06,
      "loss": 0.0028,
      "step": 121630
    },
    {
      "epoch": 0.7801611760140844,
      "grad_norm": 0.06297134608030319,
      "learning_rate": 7.648410633948265e-06,
      "loss": 0.0017,
      "step": 121640
    },
    {
      "epoch": 0.7802253129078706,
      "grad_norm": 0.3688918948173523,
      "learning_rate": 7.647935879063758e-06,
      "loss": 0.0021,
      "step": 121650
    },
    {
      "epoch": 0.7802894498016567,
      "grad_norm": 0.07675735652446747,
      "learning_rate": 7.647461090998722e-06,
      "loss": 0.0049,
      "step": 121660
    },
    {
      "epoch": 0.7803535866954427,
      "grad_norm": 0.1179722249507904,
      "learning_rate": 7.646986269759107e-06,
      "loss": 0.0032,
      "step": 121670
    },
    {
      "epoch": 0.7804177235892289,
      "grad_norm": 0.0887451022863388,
      "learning_rate": 7.646511415350861e-06,
      "loss": 0.0041,
      "step": 121680
    },
    {
      "epoch": 0.7804818604830149,
      "grad_norm": 0.12331262975931168,
      "learning_rate": 7.646036527779937e-06,
      "loss": 0.0022,
      "step": 121690
    },
    {
      "epoch": 0.7805459973768011,
      "grad_norm": 0.10870742052793503,
      "learning_rate": 7.645561607052283e-06,
      "loss": 0.003,
      "step": 121700
    },
    {
      "epoch": 0.7806101342705871,
      "grad_norm": 0.10599672049283981,
      "learning_rate": 7.64508665317385e-06,
      "loss": 0.0027,
      "step": 121710
    },
    {
      "epoch": 0.7806742711643733,
      "grad_norm": 0.12671251595020294,
      "learning_rate": 7.644611666150593e-06,
      "loss": 0.0045,
      "step": 121720
    },
    {
      "epoch": 0.7807384080581593,
      "grad_norm": 0.18857617676258087,
      "learning_rate": 7.644136645988463e-06,
      "loss": 0.0022,
      "step": 121730
    },
    {
      "epoch": 0.7808025449519455,
      "grad_norm": 0.08521877974271774,
      "learning_rate": 7.643661592693408e-06,
      "loss": 0.0025,
      "step": 121740
    },
    {
      "epoch": 0.7808666818457315,
      "grad_norm": 0.10013040155172348,
      "learning_rate": 7.643186506271386e-06,
      "loss": 0.0015,
      "step": 121750
    },
    {
      "epoch": 0.7809308187395176,
      "grad_norm": 0.14211426675319672,
      "learning_rate": 7.642711386728346e-06,
      "loss": 0.0015,
      "step": 121760
    },
    {
      "epoch": 0.7809949556333037,
      "grad_norm": 0.05439943075180054,
      "learning_rate": 7.642236234070246e-06,
      "loss": 0.0058,
      "step": 121770
    },
    {
      "epoch": 0.7810590925270898,
      "grad_norm": 0.11683088541030884,
      "learning_rate": 7.641761048303037e-06,
      "loss": 0.0027,
      "step": 121780
    },
    {
      "epoch": 0.7811232294208759,
      "grad_norm": 0.023444602265954018,
      "learning_rate": 7.641285829432671e-06,
      "loss": 0.0057,
      "step": 121790
    },
    {
      "epoch": 0.781187366314662,
      "grad_norm": 0.1272461712360382,
      "learning_rate": 7.64081057746511e-06,
      "loss": 0.0028,
      "step": 121800
    },
    {
      "epoch": 0.7812515032084482,
      "grad_norm": 0.04274454340338707,
      "learning_rate": 7.640335292406303e-06,
      "loss": 0.0043,
      "step": 121810
    },
    {
      "epoch": 0.7813156401022342,
      "grad_norm": 0.1526411920785904,
      "learning_rate": 7.639859974262208e-06,
      "loss": 0.0028,
      "step": 121820
    },
    {
      "epoch": 0.7813797769960203,
      "grad_norm": 0.2035612016916275,
      "learning_rate": 7.639384623038782e-06,
      "loss": 0.0097,
      "step": 121830
    },
    {
      "epoch": 0.7814439138898064,
      "grad_norm": 0.07125157862901688,
      "learning_rate": 7.638909238741978e-06,
      "loss": 0.003,
      "step": 121840
    },
    {
      "epoch": 0.7815080507835925,
      "grad_norm": 0.07473345100879669,
      "learning_rate": 7.638433821377756e-06,
      "loss": 0.0015,
      "step": 121850
    },
    {
      "epoch": 0.7815721876773786,
      "grad_norm": 0.12324604392051697,
      "learning_rate": 7.637958370952074e-06,
      "loss": 0.0018,
      "step": 121860
    },
    {
      "epoch": 0.7816363245711647,
      "grad_norm": 0.08533834666013718,
      "learning_rate": 7.637482887470886e-06,
      "loss": 0.0026,
      "step": 121870
    },
    {
      "epoch": 0.7817004614649508,
      "grad_norm": 0.12150117009878159,
      "learning_rate": 7.637007370940155e-06,
      "loss": 0.0029,
      "step": 121880
    },
    {
      "epoch": 0.7817645983587369,
      "grad_norm": 0.09350825101137161,
      "learning_rate": 7.636531821365835e-06,
      "loss": 0.0025,
      "step": 121890
    },
    {
      "epoch": 0.7818287352525229,
      "grad_norm": 0.024046439677476883,
      "learning_rate": 7.636056238753888e-06,
      "loss": 0.0012,
      "step": 121900
    },
    {
      "epoch": 0.7818928721463091,
      "grad_norm": 0.23442338407039642,
      "learning_rate": 7.635580623110273e-06,
      "loss": 0.0039,
      "step": 121910
    },
    {
      "epoch": 0.7819570090400951,
      "grad_norm": 0.08703475445508957,
      "learning_rate": 7.635104974440948e-06,
      "loss": 0.0032,
      "step": 121920
    },
    {
      "epoch": 0.7820211459338813,
      "grad_norm": 0.19465945661067963,
      "learning_rate": 7.634629292751874e-06,
      "loss": 0.0021,
      "step": 121930
    },
    {
      "epoch": 0.7820852828276674,
      "grad_norm": 0.09618143737316132,
      "learning_rate": 7.634153578049014e-06,
      "loss": 0.0045,
      "step": 121940
    },
    {
      "epoch": 0.7821494197214535,
      "grad_norm": 0.004982649814337492,
      "learning_rate": 7.633677830338326e-06,
      "loss": 0.0036,
      "step": 121950
    },
    {
      "epoch": 0.7822135566152396,
      "grad_norm": 0.15258103609085083,
      "learning_rate": 7.633202049625772e-06,
      "loss": 0.0037,
      "step": 121960
    },
    {
      "epoch": 0.7822776935090257,
      "grad_norm": 0.03397071361541748,
      "learning_rate": 7.632726235917314e-06,
      "loss": 0.0035,
      "step": 121970
    },
    {
      "epoch": 0.7823418304028118,
      "grad_norm": 0.14839141070842743,
      "learning_rate": 7.632250389218917e-06,
      "loss": 0.0031,
      "step": 121980
    },
    {
      "epoch": 0.7824059672965978,
      "grad_norm": 0.15496550500392914,
      "learning_rate": 7.631774509536538e-06,
      "loss": 0.0021,
      "step": 121990
    },
    {
      "epoch": 0.782470104190384,
      "grad_norm": 0.0802803486585617,
      "learning_rate": 7.631298596876146e-06,
      "loss": 0.0029,
      "step": 122000
    },
    {
      "epoch": 0.782470104190384,
      "eval_loss": 0.0031755194067955017,
      "eval_runtime": 3.3154,
      "eval_samples_per_second": 60.325,
      "eval_steps_per_second": 15.081,
      "step": 122000
    },
    {
      "epoch": 0.78253424108417,
      "grad_norm": 0.07263299077749252,
      "learning_rate": 7.6308226512437e-06,
      "loss": 0.002,
      "step": 122010
    },
    {
      "epoch": 0.7825983779779562,
      "grad_norm": 0.40417736768722534,
      "learning_rate": 7.630346672645168e-06,
      "loss": 0.0026,
      "step": 122020
    },
    {
      "epoch": 0.7826625148717422,
      "grad_norm": 0.0935332328081131,
      "learning_rate": 7.62987066108651e-06,
      "loss": 0.0023,
      "step": 122030
    },
    {
      "epoch": 0.7827266517655284,
      "grad_norm": 0.10363059490919113,
      "learning_rate": 7.629394616573697e-06,
      "loss": 0.0038,
      "step": 122040
    },
    {
      "epoch": 0.7827907886593144,
      "grad_norm": 0.07930910587310791,
      "learning_rate": 7.628918539112686e-06,
      "loss": 0.002,
      "step": 122050
    },
    {
      "epoch": 0.7828549255531005,
      "grad_norm": 0.06223560497164726,
      "learning_rate": 7.628442428709449e-06,
      "loss": 0.0018,
      "step": 122060
    },
    {
      "epoch": 0.7829190624468866,
      "grad_norm": 0.08662588149309158,
      "learning_rate": 7.62796628536995e-06,
      "loss": 0.0016,
      "step": 122070
    },
    {
      "epoch": 0.7829831993406727,
      "grad_norm": 0.07027466595172882,
      "learning_rate": 7.6274901091001555e-06,
      "loss": 0.0033,
      "step": 122080
    },
    {
      "epoch": 0.7830473362344589,
      "grad_norm": 0.07510018348693848,
      "learning_rate": 7.627013899906032e-06,
      "loss": 0.0028,
      "step": 122090
    },
    {
      "epoch": 0.7831114731282449,
      "grad_norm": 0.16565001010894775,
      "learning_rate": 7.626537657793545e-06,
      "loss": 0.003,
      "step": 122100
    },
    {
      "epoch": 0.7831756100220311,
      "grad_norm": 0.09801848232746124,
      "learning_rate": 7.626061382768666e-06,
      "loss": 0.003,
      "step": 122110
    },
    {
      "epoch": 0.7832397469158171,
      "grad_norm": 0.05887960270047188,
      "learning_rate": 7.625585074837361e-06,
      "loss": 0.0022,
      "step": 122120
    },
    {
      "epoch": 0.7833038838096033,
      "grad_norm": 0.17415302991867065,
      "learning_rate": 7.625108734005597e-06,
      "loss": 0.0024,
      "step": 122130
    },
    {
      "epoch": 0.7833680207033893,
      "grad_norm": 0.05905602499842644,
      "learning_rate": 7.624632360279345e-06,
      "loss": 0.0018,
      "step": 122140
    },
    {
      "epoch": 0.7834321575971754,
      "grad_norm": 0.18371962010860443,
      "learning_rate": 7.624155953664575e-06,
      "loss": 0.0026,
      "step": 122150
    },
    {
      "epoch": 0.7834962944909615,
      "grad_norm": 0.016537338495254517,
      "learning_rate": 7.623679514167254e-06,
      "loss": 0.0028,
      "step": 122160
    },
    {
      "epoch": 0.7835604313847476,
      "grad_norm": 0.04831685498356819,
      "learning_rate": 7.623203041793354e-06,
      "loss": 0.0035,
      "step": 122170
    },
    {
      "epoch": 0.7836245682785337,
      "grad_norm": 0.04566079005599022,
      "learning_rate": 7.622726536548846e-06,
      "loss": 0.0013,
      "step": 122180
    },
    {
      "epoch": 0.7836887051723198,
      "grad_norm": 0.08743573725223541,
      "learning_rate": 7.622249998439698e-06,
      "loss": 0.0021,
      "step": 122190
    },
    {
      "epoch": 0.7837528420661058,
      "grad_norm": 0.009777238592505455,
      "learning_rate": 7.621773427471886e-06,
      "loss": 0.0016,
      "step": 122200
    },
    {
      "epoch": 0.783816978959892,
      "grad_norm": 0.16177983582019806,
      "learning_rate": 7.621296823651376e-06,
      "loss": 0.0017,
      "step": 122210
    },
    {
      "epoch": 0.7838811158536781,
      "grad_norm": 0.17031234502792358,
      "learning_rate": 7.620820186984146e-06,
      "loss": 0.0035,
      "step": 122220
    },
    {
      "epoch": 0.7839452527474642,
      "grad_norm": 0.07148251682519913,
      "learning_rate": 7.620343517476165e-06,
      "loss": 0.006,
      "step": 122230
    },
    {
      "epoch": 0.7840093896412503,
      "grad_norm": 0.11304371803998947,
      "learning_rate": 7.619866815133408e-06,
      "loss": 0.0058,
      "step": 122240
    },
    {
      "epoch": 0.7840735265350364,
      "grad_norm": 0.0622345469892025,
      "learning_rate": 7.619390079961846e-06,
      "loss": 0.002,
      "step": 122250
    },
    {
      "epoch": 0.7841376634288225,
      "grad_norm": 0.0038027805276215076,
      "learning_rate": 7.618913311967455e-06,
      "loss": 0.0045,
      "step": 122260
    },
    {
      "epoch": 0.7842018003226086,
      "grad_norm": 0.23910140991210938,
      "learning_rate": 7.618436511156209e-06,
      "loss": 0.0024,
      "step": 122270
    },
    {
      "epoch": 0.7842659372163947,
      "grad_norm": 0.04277581349015236,
      "learning_rate": 7.617959677534081e-06,
      "loss": 0.0026,
      "step": 122280
    },
    {
      "epoch": 0.7843300741101807,
      "grad_norm": 0.10653826594352722,
      "learning_rate": 7.617482811107049e-06,
      "loss": 0.0025,
      "step": 122290
    },
    {
      "epoch": 0.7843942110039669,
      "grad_norm": 0.16174641251564026,
      "learning_rate": 7.617005911881085e-06,
      "loss": 0.0053,
      "step": 122300
    },
    {
      "epoch": 0.7844583478977529,
      "grad_norm": 0.14989493787288666,
      "learning_rate": 7.616528979862167e-06,
      "loss": 0.0033,
      "step": 122310
    },
    {
      "epoch": 0.7845224847915391,
      "grad_norm": 0.06515716761350632,
      "learning_rate": 7.616052015056271e-06,
      "loss": 0.0044,
      "step": 122320
    },
    {
      "epoch": 0.7845866216853251,
      "grad_norm": 0.12088882178068161,
      "learning_rate": 7.615575017469372e-06,
      "loss": 0.0027,
      "step": 122330
    },
    {
      "epoch": 0.7846507585791113,
      "grad_norm": 0.3991908133029938,
      "learning_rate": 7.615097987107452e-06,
      "loss": 0.0024,
      "step": 122340
    },
    {
      "epoch": 0.7847148954728973,
      "grad_norm": 0.06363681703805923,
      "learning_rate": 7.614620923976484e-06,
      "loss": 0.0015,
      "step": 122350
    },
    {
      "epoch": 0.7847790323666834,
      "grad_norm": 0.05574396252632141,
      "learning_rate": 7.614143828082445e-06,
      "loss": 0.0028,
      "step": 122360
    },
    {
      "epoch": 0.7848431692604696,
      "grad_norm": 0.0889599472284317,
      "learning_rate": 7.613666699431317e-06,
      "loss": 0.0027,
      "step": 122370
    },
    {
      "epoch": 0.7849073061542556,
      "grad_norm": 0.04563826695084572,
      "learning_rate": 7.613189538029078e-06,
      "loss": 0.0014,
      "step": 122380
    },
    {
      "epoch": 0.7849714430480418,
      "grad_norm": 0.027129748836159706,
      "learning_rate": 7.612712343881705e-06,
      "loss": 0.0026,
      "step": 122390
    },
    {
      "epoch": 0.7850355799418278,
      "grad_norm": 0.035565756261348724,
      "learning_rate": 7.6122351169951795e-06,
      "loss": 0.0023,
      "step": 122400
    },
    {
      "epoch": 0.785099716835614,
      "grad_norm": 0.15562231838703156,
      "learning_rate": 7.611757857375482e-06,
      "loss": 0.0037,
      "step": 122410
    },
    {
      "epoch": 0.7851638537294,
      "grad_norm": 0.05388416349887848,
      "learning_rate": 7.611280565028592e-06,
      "loss": 0.0024,
      "step": 122420
    },
    {
      "epoch": 0.7852279906231862,
      "grad_norm": 0.11233717948198318,
      "learning_rate": 7.610803239960489e-06,
      "loss": 0.0021,
      "step": 122430
    },
    {
      "epoch": 0.7852921275169722,
      "grad_norm": 0.08933355659246445,
      "learning_rate": 7.610325882177156e-06,
      "loss": 0.0024,
      "step": 122440
    },
    {
      "epoch": 0.7853562644107583,
      "grad_norm": 0.06173817813396454,
      "learning_rate": 7.609848491684575e-06,
      "loss": 0.0029,
      "step": 122450
    },
    {
      "epoch": 0.7854204013045444,
      "grad_norm": 0.21297572553157806,
      "learning_rate": 7.609371068488727e-06,
      "loss": 0.0019,
      "step": 122460
    },
    {
      "epoch": 0.7854845381983305,
      "grad_norm": 0.043749645352363586,
      "learning_rate": 7.608893612595594e-06,
      "loss": 0.0035,
      "step": 122470
    },
    {
      "epoch": 0.7855486750921166,
      "grad_norm": 0.16853967308998108,
      "learning_rate": 7.608416124011158e-06,
      "loss": 0.0042,
      "step": 122480
    },
    {
      "epoch": 0.7856128119859027,
      "grad_norm": 0.0530732087790966,
      "learning_rate": 7.607938602741407e-06,
      "loss": 0.0022,
      "step": 122490
    },
    {
      "epoch": 0.7856769488796888,
      "grad_norm": 0.12097512930631638,
      "learning_rate": 7.6074610487923194e-06,
      "loss": 0.0044,
      "step": 122500
    },
    {
      "epoch": 0.7857410857734749,
      "grad_norm": 0.10567058622837067,
      "learning_rate": 7.6069834621698815e-06,
      "loss": 0.0031,
      "step": 122510
    },
    {
      "epoch": 0.785805222667261,
      "grad_norm": 0.09319041669368744,
      "learning_rate": 7.606505842880079e-06,
      "loss": 0.0012,
      "step": 122520
    },
    {
      "epoch": 0.7858693595610471,
      "grad_norm": 0.0469089075922966,
      "learning_rate": 7.606028190928893e-06,
      "loss": 0.0034,
      "step": 122530
    },
    {
      "epoch": 0.7859334964548332,
      "grad_norm": 0.1647704839706421,
      "learning_rate": 7.6055505063223125e-06,
      "loss": 0.0026,
      "step": 122540
    },
    {
      "epoch": 0.7859976333486193,
      "grad_norm": 0.1056341826915741,
      "learning_rate": 7.605072789066322e-06,
      "loss": 0.0029,
      "step": 122550
    },
    {
      "epoch": 0.7860617702424054,
      "grad_norm": 0.07093048095703125,
      "learning_rate": 7.604595039166909e-06,
      "loss": 0.0017,
      "step": 122560
    },
    {
      "epoch": 0.7861259071361915,
      "grad_norm": 0.034791190177202225,
      "learning_rate": 7.604117256630057e-06,
      "loss": 0.0016,
      "step": 122570
    },
    {
      "epoch": 0.7861900440299776,
      "grad_norm": 0.046128518879413605,
      "learning_rate": 7.603639441461755e-06,
      "loss": 0.0043,
      "step": 122580
    },
    {
      "epoch": 0.7862541809237636,
      "grad_norm": 0.0922919511795044,
      "learning_rate": 7.603161593667989e-06,
      "loss": 0.0031,
      "step": 122590
    },
    {
      "epoch": 0.7863183178175498,
      "grad_norm": 0.25014355778694153,
      "learning_rate": 7.60268371325475e-06,
      "loss": 0.003,
      "step": 122600
    },
    {
      "epoch": 0.7863824547113358,
      "grad_norm": 0.010387763381004333,
      "learning_rate": 7.602205800228022e-06,
      "loss": 0.0016,
      "step": 122610
    },
    {
      "epoch": 0.786446591605122,
      "grad_norm": 0.03959250822663307,
      "learning_rate": 7.601727854593796e-06,
      "loss": 0.0021,
      "step": 122620
    },
    {
      "epoch": 0.786510728498908,
      "grad_norm": 0.218663290143013,
      "learning_rate": 7.601249876358061e-06,
      "loss": 0.0048,
      "step": 122630
    },
    {
      "epoch": 0.7865748653926942,
      "grad_norm": 0.07799277454614639,
      "learning_rate": 7.600771865526807e-06,
      "loss": 0.0039,
      "step": 122640
    },
    {
      "epoch": 0.7866390022864803,
      "grad_norm": 0.11367930471897125,
      "learning_rate": 7.600293822106022e-06,
      "loss": 0.0024,
      "step": 122650
    },
    {
      "epoch": 0.7867031391802664,
      "grad_norm": 0.29839062690734863,
      "learning_rate": 7.599815746101696e-06,
      "loss": 0.0023,
      "step": 122660
    },
    {
      "epoch": 0.7867672760740525,
      "grad_norm": 0.07987088710069656,
      "learning_rate": 7.599337637519821e-06,
      "loss": 0.0014,
      "step": 122670
    },
    {
      "epoch": 0.7868314129678385,
      "grad_norm": 0.05106490105390549,
      "learning_rate": 7.598859496366389e-06,
      "loss": 0.0028,
      "step": 122680
    },
    {
      "epoch": 0.7868955498616247,
      "grad_norm": 0.2657345235347748,
      "learning_rate": 7.598381322647388e-06,
      "loss": 0.0031,
      "step": 122690
    },
    {
      "epoch": 0.7869596867554107,
      "grad_norm": 0.09356044232845306,
      "learning_rate": 7.597903116368813e-06,
      "loss": 0.0027,
      "step": 122700
    },
    {
      "epoch": 0.7870238236491969,
      "grad_norm": 0.06518931686878204,
      "learning_rate": 7.597424877536656e-06,
      "loss": 0.0014,
      "step": 122710
    },
    {
      "epoch": 0.7870879605429829,
      "grad_norm": 0.1209230124950409,
      "learning_rate": 7.596946606156908e-06,
      "loss": 0.0035,
      "step": 122720
    },
    {
      "epoch": 0.7871520974367691,
      "grad_norm": 0.10231737047433853,
      "learning_rate": 7.5964683022355646e-06,
      "loss": 0.0016,
      "step": 122730
    },
    {
      "epoch": 0.7872162343305551,
      "grad_norm": 0.04111761599779129,
      "learning_rate": 7.595989965778615e-06,
      "loss": 0.0027,
      "step": 122740
    },
    {
      "epoch": 0.7872803712243412,
      "grad_norm": 0.2692529857158661,
      "learning_rate": 7.595511596792058e-06,
      "loss": 0.0026,
      "step": 122750
    },
    {
      "epoch": 0.7873445081181273,
      "grad_norm": 0.07208271324634552,
      "learning_rate": 7.595033195281884e-06,
      "loss": 0.0044,
      "step": 122760
    },
    {
      "epoch": 0.7874086450119134,
      "grad_norm": 0.04921235144138336,
      "learning_rate": 7.59455476125409e-06,
      "loss": 0.0033,
      "step": 122770
    },
    {
      "epoch": 0.7874727819056995,
      "grad_norm": 0.14421400427818298,
      "learning_rate": 7.594076294714671e-06,
      "loss": 0.003,
      "step": 122780
    },
    {
      "epoch": 0.7875369187994856,
      "grad_norm": 0.012242033146321774,
      "learning_rate": 7.593597795669623e-06,
      "loss": 0.0025,
      "step": 122790
    },
    {
      "epoch": 0.7876010556932718,
      "grad_norm": 0.04912176355719566,
      "learning_rate": 7.59311926412494e-06,
      "loss": 0.0049,
      "step": 122800
    },
    {
      "epoch": 0.7876651925870578,
      "grad_norm": 0.0413503423333168,
      "learning_rate": 7.592640700086619e-06,
      "loss": 0.0008,
      "step": 122810
    },
    {
      "epoch": 0.787729329480844,
      "grad_norm": 0.1055869534611702,
      "learning_rate": 7.592162103560656e-06,
      "loss": 0.0021,
      "step": 122820
    },
    {
      "epoch": 0.78779346637463,
      "grad_norm": 0.05369802191853523,
      "learning_rate": 7.591683474553052e-06,
      "loss": 0.003,
      "step": 122830
    },
    {
      "epoch": 0.7878576032684161,
      "grad_norm": 0.08172406256198883,
      "learning_rate": 7.5912048130698004e-06,
      "loss": 0.0022,
      "step": 122840
    },
    {
      "epoch": 0.7879217401622022,
      "grad_norm": 0.20662006735801697,
      "learning_rate": 7.5907261191169e-06,
      "loss": 0.0037,
      "step": 122850
    },
    {
      "epoch": 0.7879858770559883,
      "grad_norm": 0.1541481614112854,
      "learning_rate": 7.59024739270035e-06,
      "loss": 0.0023,
      "step": 122860
    },
    {
      "epoch": 0.7880500139497744,
      "grad_norm": 0.17998526990413666,
      "learning_rate": 7.589768633826151e-06,
      "loss": 0.0023,
      "step": 122870
    },
    {
      "epoch": 0.7881141508435605,
      "grad_norm": 0.06872714310884476,
      "learning_rate": 7.589289842500298e-06,
      "loss": 0.0023,
      "step": 122880
    },
    {
      "epoch": 0.7881782877373466,
      "grad_norm": 0.12394532561302185,
      "learning_rate": 7.588811018728793e-06,
      "loss": 0.0053,
      "step": 122890
    },
    {
      "epoch": 0.7882424246311327,
      "grad_norm": 0.060991834849119186,
      "learning_rate": 7.588332162517636e-06,
      "loss": 0.0017,
      "step": 122900
    },
    {
      "epoch": 0.7883065615249187,
      "grad_norm": 0.018962649628520012,
      "learning_rate": 7.587853273872827e-06,
      "loss": 0.0022,
      "step": 122910
    },
    {
      "epoch": 0.7883706984187049,
      "grad_norm": 0.18350961804389954,
      "learning_rate": 7.587374352800367e-06,
      "loss": 0.002,
      "step": 122920
    },
    {
      "epoch": 0.788434835312491,
      "grad_norm": 0.014333275146782398,
      "learning_rate": 7.5868953993062576e-06,
      "loss": 0.0031,
      "step": 122930
    },
    {
      "epoch": 0.7884989722062771,
      "grad_norm": 0.13420072197914124,
      "learning_rate": 7.5864164133965e-06,
      "loss": 0.0013,
      "step": 122940
    },
    {
      "epoch": 0.7885631091000632,
      "grad_norm": 0.03388292342424393,
      "learning_rate": 7.585937395077095e-06,
      "loss": 0.0017,
      "step": 122950
    },
    {
      "epoch": 0.7886272459938493,
      "grad_norm": 0.04059087857604027,
      "learning_rate": 7.585458344354049e-06,
      "loss": 0.003,
      "step": 122960
    },
    {
      "epoch": 0.7886913828876354,
      "grad_norm": 0.03942510858178139,
      "learning_rate": 7.5849792612333595e-06,
      "loss": 0.0029,
      "step": 122970
    },
    {
      "epoch": 0.7887555197814214,
      "grad_norm": 0.047274697571992874,
      "learning_rate": 7.584500145721034e-06,
      "loss": 0.0031,
      "step": 122980
    },
    {
      "epoch": 0.7888196566752076,
      "grad_norm": 0.0786990150809288,
      "learning_rate": 7.584020997823074e-06,
      "loss": 0.0032,
      "step": 122990
    },
    {
      "epoch": 0.7888837935689936,
      "grad_norm": 0.1254752278327942,
      "learning_rate": 7.583541817545483e-06,
      "loss": 0.0013,
      "step": 123000
    },
    {
      "epoch": 0.7889479304627798,
      "grad_norm": 0.19860519468784332,
      "learning_rate": 7.583062604894268e-06,
      "loss": 0.0017,
      "step": 123010
    },
    {
      "epoch": 0.7890120673565658,
      "grad_norm": 0.3162236213684082,
      "learning_rate": 7.582583359875433e-06,
      "loss": 0.0027,
      "step": 123020
    },
    {
      "epoch": 0.789076204250352,
      "grad_norm": 0.12642884254455566,
      "learning_rate": 7.582104082494981e-06,
      "loss": 0.0033,
      "step": 123030
    },
    {
      "epoch": 0.789140341144138,
      "grad_norm": 0.2137831300497055,
      "learning_rate": 7.5816247727589195e-06,
      "loss": 0.0025,
      "step": 123040
    },
    {
      "epoch": 0.7892044780379242,
      "grad_norm": 0.06308590620756149,
      "learning_rate": 7.581145430673256e-06,
      "loss": 0.003,
      "step": 123050
    },
    {
      "epoch": 0.7892686149317102,
      "grad_norm": 0.1454891562461853,
      "learning_rate": 7.580666056243995e-06,
      "loss": 0.002,
      "step": 123060
    },
    {
      "epoch": 0.7893327518254963,
      "grad_norm": 0.1343536227941513,
      "learning_rate": 7.580186649477144e-06,
      "loss": 0.0038,
      "step": 123070
    },
    {
      "epoch": 0.7893968887192825,
      "grad_norm": 0.20746780931949615,
      "learning_rate": 7.579707210378709e-06,
      "loss": 0.0024,
      "step": 123080
    },
    {
      "epoch": 0.7894610256130685,
      "grad_norm": 0.08619903028011322,
      "learning_rate": 7.579227738954701e-06,
      "loss": 0.0029,
      "step": 123090
    },
    {
      "epoch": 0.7895251625068547,
      "grad_norm": 0.024186434224247932,
      "learning_rate": 7.578748235211124e-06,
      "loss": 0.0009,
      "step": 123100
    },
    {
      "epoch": 0.7895892994006407,
      "grad_norm": 0.049256592988967896,
      "learning_rate": 7.5782686991539914e-06,
      "loss": 0.0028,
      "step": 123110
    },
    {
      "epoch": 0.7896534362944269,
      "grad_norm": 0.3305787742137909,
      "learning_rate": 7.577789130789306e-06,
      "loss": 0.0028,
      "step": 123120
    },
    {
      "epoch": 0.7897175731882129,
      "grad_norm": 0.12885724008083344,
      "learning_rate": 7.577309530123082e-06,
      "loss": 0.0042,
      "step": 123130
    },
    {
      "epoch": 0.789781710081999,
      "grad_norm": 0.10026602447032928,
      "learning_rate": 7.576829897161327e-06,
      "loss": 0.0032,
      "step": 123140
    },
    {
      "epoch": 0.7898458469757851,
      "grad_norm": 0.03776301443576813,
      "learning_rate": 7.5763502319100535e-06,
      "loss": 0.0039,
      "step": 123150
    },
    {
      "epoch": 0.7899099838695712,
      "grad_norm": 0.10966672003269196,
      "learning_rate": 7.575870534375269e-06,
      "loss": 0.0025,
      "step": 123160
    },
    {
      "epoch": 0.7899741207633573,
      "grad_norm": 0.1907765418291092,
      "learning_rate": 7.575390804562987e-06,
      "loss": 0.0042,
      "step": 123170
    },
    {
      "epoch": 0.7900382576571434,
      "grad_norm": 0.1846044957637787,
      "learning_rate": 7.574911042479216e-06,
      "loss": 0.0017,
      "step": 123180
    },
    {
      "epoch": 0.7901023945509295,
      "grad_norm": 0.09555988758802414,
      "learning_rate": 7.57443124812997e-06,
      "loss": 0.0032,
      "step": 123190
    },
    {
      "epoch": 0.7901665314447156,
      "grad_norm": 0.37532809376716614,
      "learning_rate": 7.57395142152126e-06,
      "loss": 0.0043,
      "step": 123200
    },
    {
      "epoch": 0.7902306683385018,
      "grad_norm": 0.09359215199947357,
      "learning_rate": 7.5734715626591004e-06,
      "loss": 0.0021,
      "step": 123210
    },
    {
      "epoch": 0.7902948052322878,
      "grad_norm": 0.013770547695457935,
      "learning_rate": 7.572991671549503e-06,
      "loss": 0.0018,
      "step": 123220
    },
    {
      "epoch": 0.7903589421260739,
      "grad_norm": 0.06928499788045883,
      "learning_rate": 7.57251174819848e-06,
      "loss": 0.0051,
      "step": 123230
    },
    {
      "epoch": 0.79042307901986,
      "grad_norm": 0.18177980184555054,
      "learning_rate": 7.5720317926120455e-06,
      "loss": 0.0044,
      "step": 123240
    },
    {
      "epoch": 0.7904872159136461,
      "grad_norm": 0.16453734040260315,
      "learning_rate": 7.571551804796216e-06,
      "loss": 0.0028,
      "step": 123250
    },
    {
      "epoch": 0.7905513528074322,
      "grad_norm": 0.15786118805408478,
      "learning_rate": 7.571071784757004e-06,
      "loss": 0.005,
      "step": 123260
    },
    {
      "epoch": 0.7906154897012183,
      "grad_norm": 0.0929371789097786,
      "learning_rate": 7.5705917325004254e-06,
      "loss": 0.0027,
      "step": 123270
    },
    {
      "epoch": 0.7906796265950043,
      "grad_norm": 0.08085958659648895,
      "learning_rate": 7.570111648032494e-06,
      "loss": 0.002,
      "step": 123280
    },
    {
      "epoch": 0.7907437634887905,
      "grad_norm": 0.23638342320919037,
      "learning_rate": 7.569631531359227e-06,
      "loss": 0.0031,
      "step": 123290
    },
    {
      "epoch": 0.7908079003825765,
      "grad_norm": 0.020316675305366516,
      "learning_rate": 7.569151382486641e-06,
      "loss": 0.0039,
      "step": 123300
    },
    {
      "epoch": 0.7908720372763627,
      "grad_norm": 0.05101149156689644,
      "learning_rate": 7.568671201420752e-06,
      "loss": 0.0025,
      "step": 123310
    },
    {
      "epoch": 0.7909361741701487,
      "grad_norm": 0.09833734482526779,
      "learning_rate": 7.568190988167578e-06,
      "loss": 0.0025,
      "step": 123320
    },
    {
      "epoch": 0.7910003110639349,
      "grad_norm": 0.11523966491222382,
      "learning_rate": 7.567710742733134e-06,
      "loss": 0.002,
      "step": 123330
    },
    {
      "epoch": 0.7910644479577209,
      "grad_norm": 0.04753712937235832,
      "learning_rate": 7.567230465123441e-06,
      "loss": 0.0036,
      "step": 123340
    },
    {
      "epoch": 0.791128584851507,
      "grad_norm": 0.2613712251186371,
      "learning_rate": 7.5667501553445135e-06,
      "loss": 0.004,
      "step": 123350
    },
    {
      "epoch": 0.7911927217452932,
      "grad_norm": 0.028088044375181198,
      "learning_rate": 7.566269813402374e-06,
      "loss": 0.0021,
      "step": 123360
    },
    {
      "epoch": 0.7912568586390792,
      "grad_norm": 0.12382286041975021,
      "learning_rate": 7.565789439303037e-06,
      "loss": 0.0028,
      "step": 123370
    },
    {
      "epoch": 0.7913209955328654,
      "grad_norm": 0.02080100029706955,
      "learning_rate": 7.565309033052528e-06,
      "loss": 0.0019,
      "step": 123380
    },
    {
      "epoch": 0.7913851324266514,
      "grad_norm": 0.13737565279006958,
      "learning_rate": 7.5648285946568615e-06,
      "loss": 0.0045,
      "step": 123390
    },
    {
      "epoch": 0.7914492693204376,
      "grad_norm": 0.020041542127728462,
      "learning_rate": 7.5643481241220585e-06,
      "loss": 0.0024,
      "step": 123400
    },
    {
      "epoch": 0.7915134062142236,
      "grad_norm": 0.23024889826774597,
      "learning_rate": 7.5638676214541425e-06,
      "loss": 0.0033,
      "step": 123410
    },
    {
      "epoch": 0.7915775431080098,
      "grad_norm": 0.22192806005477905,
      "learning_rate": 7.563387086659133e-06,
      "loss": 0.0029,
      "step": 123420
    },
    {
      "epoch": 0.7916416800017958,
      "grad_norm": 0.08314814418554306,
      "learning_rate": 7.562906519743051e-06,
      "loss": 0.0018,
      "step": 123430
    },
    {
      "epoch": 0.791705816895582,
      "grad_norm": 0.15509885549545288,
      "learning_rate": 7.5624259207119174e-06,
      "loss": 0.0054,
      "step": 123440
    },
    {
      "epoch": 0.791769953789368,
      "grad_norm": 0.13001513481140137,
      "learning_rate": 7.561945289571757e-06,
      "loss": 0.0027,
      "step": 123450
    },
    {
      "epoch": 0.7918340906831541,
      "grad_norm": 0.0911867767572403,
      "learning_rate": 7.561464626328591e-06,
      "loss": 0.0037,
      "step": 123460
    },
    {
      "epoch": 0.7918982275769402,
      "grad_norm": 0.12942135334014893,
      "learning_rate": 7.560983930988443e-06,
      "loss": 0.0023,
      "step": 123470
    },
    {
      "epoch": 0.7919623644707263,
      "grad_norm": 0.09583374112844467,
      "learning_rate": 7.560503203557335e-06,
      "loss": 0.0026,
      "step": 123480
    },
    {
      "epoch": 0.7920265013645125,
      "grad_norm": 0.1424064189195633,
      "learning_rate": 7.560022444041291e-06,
      "loss": 0.0029,
      "step": 123490
    },
    {
      "epoch": 0.7920906382582985,
      "grad_norm": 0.0760183185338974,
      "learning_rate": 7.559541652446338e-06,
      "loss": 0.0018,
      "step": 123500
    },
    {
      "epoch": 0.7921547751520847,
      "grad_norm": 0.13475407660007477,
      "learning_rate": 7.5590608287784995e-06,
      "loss": 0.0027,
      "step": 123510
    },
    {
      "epoch": 0.7922189120458707,
      "grad_norm": 0.12201106548309326,
      "learning_rate": 7.558579973043798e-06,
      "loss": 0.002,
      "step": 123520
    },
    {
      "epoch": 0.7922830489396568,
      "grad_norm": 0.08769958466291428,
      "learning_rate": 7.558099085248261e-06,
      "loss": 0.0024,
      "step": 123530
    },
    {
      "epoch": 0.7923471858334429,
      "grad_norm": 0.10830560326576233,
      "learning_rate": 7.557618165397913e-06,
      "loss": 0.0021,
      "step": 123540
    },
    {
      "epoch": 0.792411322727229,
      "grad_norm": 0.06387370824813843,
      "learning_rate": 7.557137213498784e-06,
      "loss": 0.0015,
      "step": 123550
    },
    {
      "epoch": 0.7924754596210151,
      "grad_norm": 0.03704097867012024,
      "learning_rate": 7.5566562295568966e-06,
      "loss": 0.0068,
      "step": 123560
    },
    {
      "epoch": 0.7925395965148012,
      "grad_norm": 0.057072702795267105,
      "learning_rate": 7.556175213578281e-06,
      "loss": 0.0034,
      "step": 123570
    },
    {
      "epoch": 0.7926037334085873,
      "grad_norm": 0.10741273313760757,
      "learning_rate": 7.555694165568962e-06,
      "loss": 0.0017,
      "step": 123580
    },
    {
      "epoch": 0.7926678703023734,
      "grad_norm": 0.06526832282543182,
      "learning_rate": 7.555213085534969e-06,
      "loss": 0.0026,
      "step": 123590
    },
    {
      "epoch": 0.7927320071961594,
      "grad_norm": 0.11733478307723999,
      "learning_rate": 7.55473197348233e-06,
      "loss": 0.0061,
      "step": 123600
    },
    {
      "epoch": 0.7927961440899456,
      "grad_norm": 0.163051575422287,
      "learning_rate": 7.554250829417072e-06,
      "loss": 0.0049,
      "step": 123610
    },
    {
      "epoch": 0.7928602809837316,
      "grad_norm": 0.11553701758384705,
      "learning_rate": 7.553769653345227e-06,
      "loss": 0.0043,
      "step": 123620
    },
    {
      "epoch": 0.7929244178775178,
      "grad_norm": 0.1378750205039978,
      "learning_rate": 7.553288445272823e-06,
      "loss": 0.005,
      "step": 123630
    },
    {
      "epoch": 0.7929885547713039,
      "grad_norm": 0.1761714071035385,
      "learning_rate": 7.55280720520589e-06,
      "loss": 0.0037,
      "step": 123640
    },
    {
      "epoch": 0.79305269166509,
      "grad_norm": 0.13274933397769928,
      "learning_rate": 7.552325933150458e-06,
      "loss": 0.0014,
      "step": 123650
    },
    {
      "epoch": 0.7931168285588761,
      "grad_norm": 0.25353050231933594,
      "learning_rate": 7.551844629112559e-06,
      "loss": 0.005,
      "step": 123660
    },
    {
      "epoch": 0.7931809654526621,
      "grad_norm": 0.042722560465335846,
      "learning_rate": 7.551363293098222e-06,
      "loss": 0.0022,
      "step": 123670
    },
    {
      "epoch": 0.7932451023464483,
      "grad_norm": 0.37281107902526855,
      "learning_rate": 7.55088192511348e-06,
      "loss": 0.0027,
      "step": 123680
    },
    {
      "epoch": 0.7933092392402343,
      "grad_norm": 0.043687593191862106,
      "learning_rate": 7.550400525164363e-06,
      "loss": 0.0064,
      "step": 123690
    },
    {
      "epoch": 0.7933733761340205,
      "grad_norm": 0.23913761973381042,
      "learning_rate": 7.549919093256905e-06,
      "loss": 0.0021,
      "step": 123700
    },
    {
      "epoch": 0.7934375130278065,
      "grad_norm": 0.1854463666677475,
      "learning_rate": 7.549437629397141e-06,
      "loss": 0.0034,
      "step": 123710
    },
    {
      "epoch": 0.7935016499215927,
      "grad_norm": 0.40040287375450134,
      "learning_rate": 7.548956133591099e-06,
      "loss": 0.0044,
      "step": 123720
    },
    {
      "epoch": 0.7935657868153787,
      "grad_norm": 0.09918133914470673,
      "learning_rate": 7.548474605844815e-06,
      "loss": 0.0028,
      "step": 123730
    },
    {
      "epoch": 0.7936299237091649,
      "grad_norm": 0.24481476843357086,
      "learning_rate": 7.5479930461643235e-06,
      "loss": 0.0019,
      "step": 123740
    },
    {
      "epoch": 0.7936940606029509,
      "grad_norm": 0.14661909639835358,
      "learning_rate": 7.547511454555657e-06,
      "loss": 0.0022,
      "step": 123750
    },
    {
      "epoch": 0.793758197496737,
      "grad_norm": 0.11050709336996078,
      "learning_rate": 7.547029831024852e-06,
      "loss": 0.0024,
      "step": 123760
    },
    {
      "epoch": 0.7938223343905232,
      "grad_norm": 0.02962980791926384,
      "learning_rate": 7.546548175577944e-06,
      "loss": 0.0031,
      "step": 123770
    },
    {
      "epoch": 0.7938864712843092,
      "grad_norm": 0.07772235572338104,
      "learning_rate": 7.5460664882209655e-06,
      "loss": 0.0024,
      "step": 123780
    },
    {
      "epoch": 0.7939506081780954,
      "grad_norm": 0.12356381863355637,
      "learning_rate": 7.545584768959956e-06,
      "loss": 0.0043,
      "step": 123790
    },
    {
      "epoch": 0.7940147450718814,
      "grad_norm": 0.1655791848897934,
      "learning_rate": 7.545103017800948e-06,
      "loss": 0.001,
      "step": 123800
    },
    {
      "epoch": 0.7940788819656676,
      "grad_norm": 0.11550326645374298,
      "learning_rate": 7.544621234749983e-06,
      "loss": 0.0024,
      "step": 123810
    },
    {
      "epoch": 0.7941430188594536,
      "grad_norm": 0.18776899576187134,
      "learning_rate": 7.544139419813093e-06,
      "loss": 0.0024,
      "step": 123820
    },
    {
      "epoch": 0.7942071557532397,
      "grad_norm": 0.021684233099222183,
      "learning_rate": 7.543657572996319e-06,
      "loss": 0.0027,
      "step": 123830
    },
    {
      "epoch": 0.7942712926470258,
      "grad_norm": 0.11533031612634659,
      "learning_rate": 7.543175694305697e-06,
      "loss": 0.0012,
      "step": 123840
    },
    {
      "epoch": 0.7943354295408119,
      "grad_norm": 0.08878498524427414,
      "learning_rate": 7.542693783747266e-06,
      "loss": 0.0035,
      "step": 123850
    },
    {
      "epoch": 0.794399566434598,
      "grad_norm": 0.06881114095449448,
      "learning_rate": 7.542211841327065e-06,
      "loss": 0.0023,
      "step": 123860
    },
    {
      "epoch": 0.7944637033283841,
      "grad_norm": 0.34978440403938293,
      "learning_rate": 7.541729867051133e-06,
      "loss": 0.0022,
      "step": 123870
    },
    {
      "epoch": 0.7945278402221702,
      "grad_norm": 0.30584773421287537,
      "learning_rate": 7.541247860925508e-06,
      "loss": 0.0052,
      "step": 123880
    },
    {
      "epoch": 0.7945919771159563,
      "grad_norm": 0.17105741798877716,
      "learning_rate": 7.5407658229562305e-06,
      "loss": 0.0021,
      "step": 123890
    },
    {
      "epoch": 0.7946561140097423,
      "grad_norm": 0.1544884592294693,
      "learning_rate": 7.540283753149344e-06,
      "loss": 0.0028,
      "step": 123900
    },
    {
      "epoch": 0.7947202509035285,
      "grad_norm": 0.08491586148738861,
      "learning_rate": 7.539801651510885e-06,
      "loss": 0.0022,
      "step": 123910
    },
    {
      "epoch": 0.7947843877973146,
      "grad_norm": 0.12208464741706848,
      "learning_rate": 7.539319518046897e-06,
      "loss": 0.0027,
      "step": 123920
    },
    {
      "epoch": 0.7948485246911007,
      "grad_norm": 0.2068665474653244,
      "learning_rate": 7.5388373527634195e-06,
      "loss": 0.0025,
      "step": 123930
    },
    {
      "epoch": 0.7949126615848868,
      "grad_norm": 0.12112890928983688,
      "learning_rate": 7.538355155666496e-06,
      "loss": 0.0035,
      "step": 123940
    },
    {
      "epoch": 0.7949767984786729,
      "grad_norm": 0.11285851150751114,
      "learning_rate": 7.537872926762168e-06,
      "loss": 0.004,
      "step": 123950
    },
    {
      "epoch": 0.795040935372459,
      "grad_norm": 0.1206316277384758,
      "learning_rate": 7.537390666056479e-06,
      "loss": 0.0027,
      "step": 123960
    },
    {
      "epoch": 0.795105072266245,
      "grad_norm": 0.24223995208740234,
      "learning_rate": 7.536908373555472e-06,
      "loss": 0.0021,
      "step": 123970
    },
    {
      "epoch": 0.7951692091600312,
      "grad_norm": 0.029918193817138672,
      "learning_rate": 7.5364260492651886e-06,
      "loss": 0.0019,
      "step": 123980
    },
    {
      "epoch": 0.7952333460538172,
      "grad_norm": 0.23039722442626953,
      "learning_rate": 7.535943693191674e-06,
      "loss": 0.0013,
      "step": 123990
    },
    {
      "epoch": 0.7952974829476034,
      "grad_norm": 0.1353999227285385,
      "learning_rate": 7.535461305340974e-06,
      "loss": 0.003,
      "step": 124000
    },
    {
      "epoch": 0.7953616198413894,
      "grad_norm": 0.3367917537689209,
      "learning_rate": 7.534978885719131e-06,
      "loss": 0.0028,
      "step": 124010
    },
    {
      "epoch": 0.7954257567351756,
      "grad_norm": 0.12652994692325592,
      "learning_rate": 7.534496434332191e-06,
      "loss": 0.0034,
      "step": 124020
    },
    {
      "epoch": 0.7954898936289616,
      "grad_norm": 0.008631177246570587,
      "learning_rate": 7.534013951186199e-06,
      "loss": 0.0014,
      "step": 124030
    },
    {
      "epoch": 0.7955540305227478,
      "grad_norm": 0.1561666578054428,
      "learning_rate": 7.533531436287203e-06,
      "loss": 0.0014,
      "step": 124040
    },
    {
      "epoch": 0.7956181674165338,
      "grad_norm": 0.3468468189239502,
      "learning_rate": 7.533048889641243e-06,
      "loss": 0.0022,
      "step": 124050
    },
    {
      "epoch": 0.7956823043103199,
      "grad_norm": 0.037443943321704865,
      "learning_rate": 7.532566311254374e-06,
      "loss": 0.001,
      "step": 124060
    },
    {
      "epoch": 0.7957464412041061,
      "grad_norm": 0.04735523462295532,
      "learning_rate": 7.532083701132637e-06,
      "loss": 0.0024,
      "step": 124070
    },
    {
      "epoch": 0.7958105780978921,
      "grad_norm": 0.12967167794704437,
      "learning_rate": 7.531601059282083e-06,
      "loss": 0.0034,
      "step": 124080
    },
    {
      "epoch": 0.7958747149916783,
      "grad_norm": 0.17066501080989838,
      "learning_rate": 7.531118385708758e-06,
      "loss": 0.0028,
      "step": 124090
    },
    {
      "epoch": 0.7959388518854643,
      "grad_norm": 0.06425528973340988,
      "learning_rate": 7.53063568041871e-06,
      "loss": 0.0023,
      "step": 124100
    },
    {
      "epoch": 0.7960029887792505,
      "grad_norm": 0.09032629430294037,
      "learning_rate": 7.53015294341799e-06,
      "loss": 0.0011,
      "step": 124110
    },
    {
      "epoch": 0.7960671256730365,
      "grad_norm": 0.2610722482204437,
      "learning_rate": 7.529670174712643e-06,
      "loss": 0.0057,
      "step": 124120
    },
    {
      "epoch": 0.7961312625668227,
      "grad_norm": 0.23317117989063263,
      "learning_rate": 7.529187374308723e-06,
      "loss": 0.0024,
      "step": 124130
    },
    {
      "epoch": 0.7961953994606087,
      "grad_norm": 0.08279764652252197,
      "learning_rate": 7.528704542212276e-06,
      "loss": 0.002,
      "step": 124140
    },
    {
      "epoch": 0.7962595363543948,
      "grad_norm": 0.02375340461730957,
      "learning_rate": 7.528221678429355e-06,
      "loss": 0.0043,
      "step": 124150
    },
    {
      "epoch": 0.7963236732481809,
      "grad_norm": 0.03122488595545292,
      "learning_rate": 7.527738782966008e-06,
      "loss": 0.0016,
      "step": 124160
    },
    {
      "epoch": 0.796387810141967,
      "grad_norm": 0.10670062899589539,
      "learning_rate": 7.52725585582829e-06,
      "loss": 0.0015,
      "step": 124170
    },
    {
      "epoch": 0.7964519470357531,
      "grad_norm": 0.05864892154932022,
      "learning_rate": 7.526772897022247e-06,
      "loss": 0.0022,
      "step": 124180
    },
    {
      "epoch": 0.7965160839295392,
      "grad_norm": 0.06157020851969719,
      "learning_rate": 7.5262899065539365e-06,
      "loss": 0.0031,
      "step": 124190
    },
    {
      "epoch": 0.7965802208233254,
      "grad_norm": 0.23451292514801025,
      "learning_rate": 7.525806884429405e-06,
      "loss": 0.0033,
      "step": 124200
    },
    {
      "epoch": 0.7966443577171114,
      "grad_norm": 0.22916685044765472,
      "learning_rate": 7.525323830654712e-06,
      "loss": 0.0064,
      "step": 124210
    },
    {
      "epoch": 0.7967084946108975,
      "grad_norm": 0.24594850838184357,
      "learning_rate": 7.524840745235903e-06,
      "loss": 0.0012,
      "step": 124220
    },
    {
      "epoch": 0.7967726315046836,
      "grad_norm": 0.1346345990896225,
      "learning_rate": 7.524357628179037e-06,
      "loss": 0.0028,
      "step": 124230
    },
    {
      "epoch": 0.7968367683984697,
      "grad_norm": 0.08220899105072021,
      "learning_rate": 7.523874479490164e-06,
      "loss": 0.002,
      "step": 124240
    },
    {
      "epoch": 0.7969009052922558,
      "grad_norm": 0.017328398302197456,
      "learning_rate": 7.52339129917534e-06,
      "loss": 0.0025,
      "step": 124250
    },
    {
      "epoch": 0.7969650421860419,
      "grad_norm": 0.0716826543211937,
      "learning_rate": 7.5229080872406215e-06,
      "loss": 0.002,
      "step": 124260
    },
    {
      "epoch": 0.797029179079828,
      "grad_norm": 0.1872122883796692,
      "learning_rate": 7.5224248436920596e-06,
      "loss": 0.0027,
      "step": 124270
    },
    {
      "epoch": 0.7970933159736141,
      "grad_norm": 0.29147228598594666,
      "learning_rate": 7.5219415685357136e-06,
      "loss": 0.0024,
      "step": 124280
    },
    {
      "epoch": 0.7971574528674001,
      "grad_norm": 0.07466059178113937,
      "learning_rate": 7.521458261777636e-06,
      "loss": 0.0049,
      "step": 124290
    },
    {
      "epoch": 0.7972215897611863,
      "grad_norm": 0.057647235691547394,
      "learning_rate": 7.520974923423885e-06,
      "loss": 0.0025,
      "step": 124300
    },
    {
      "epoch": 0.7972857266549723,
      "grad_norm": 0.0419759601354599,
      "learning_rate": 7.5204915534805154e-06,
      "loss": 0.0009,
      "step": 124310
    },
    {
      "epoch": 0.7973498635487585,
      "grad_norm": 0.19892175495624542,
      "learning_rate": 7.520008151953586e-06,
      "loss": 0.0026,
      "step": 124320
    },
    {
      "epoch": 0.7974140004425445,
      "grad_norm": 0.10667607188224792,
      "learning_rate": 7.519524718849154e-06,
      "loss": 0.0012,
      "step": 124330
    },
    {
      "epoch": 0.7974781373363307,
      "grad_norm": 0.9881979823112488,
      "learning_rate": 7.519041254173276e-06,
      "loss": 0.0028,
      "step": 124340
    },
    {
      "epoch": 0.7975422742301168,
      "grad_norm": 0.2386818826198578,
      "learning_rate": 7.518557757932011e-06,
      "loss": 0.0051,
      "step": 124350
    },
    {
      "epoch": 0.7976064111239028,
      "grad_norm": 0.12765030562877655,
      "learning_rate": 7.518074230131418e-06,
      "loss": 0.0024,
      "step": 124360
    },
    {
      "epoch": 0.797670548017689,
      "grad_norm": 0.013037353754043579,
      "learning_rate": 7.5175906707775534e-06,
      "loss": 0.0015,
      "step": 124370
    },
    {
      "epoch": 0.797734684911475,
      "grad_norm": 0.04132802411913872,
      "learning_rate": 7.51710707987648e-06,
      "loss": 0.0027,
      "step": 124380
    },
    {
      "epoch": 0.7977988218052612,
      "grad_norm": 0.030632011592388153,
      "learning_rate": 7.516623457434255e-06,
      "loss": 0.0029,
      "step": 124390
    },
    {
      "epoch": 0.7978629586990472,
      "grad_norm": 0.17068198323249817,
      "learning_rate": 7.516139803456941e-06,
      "loss": 0.0033,
      "step": 124400
    },
    {
      "epoch": 0.7979270955928334,
      "grad_norm": 0.022175561636686325,
      "learning_rate": 7.515656117950595e-06,
      "loss": 0.0016,
      "step": 124410
    },
    {
      "epoch": 0.7979912324866194,
      "grad_norm": 0.1440616101026535,
      "learning_rate": 7.515172400921281e-06,
      "loss": 0.0029,
      "step": 124420
    },
    {
      "epoch": 0.7980553693804056,
      "grad_norm": 0.015570059418678284,
      "learning_rate": 7.5146886523750596e-06,
      "loss": 0.0036,
      "step": 124430
    },
    {
      "epoch": 0.7981195062741916,
      "grad_norm": 0.05740491673350334,
      "learning_rate": 7.514204872317991e-06,
      "loss": 0.0016,
      "step": 124440
    },
    {
      "epoch": 0.7981836431679777,
      "grad_norm": 0.07615818828344345,
      "learning_rate": 7.51372106075614e-06,
      "loss": 0.0026,
      "step": 124450
    },
    {
      "epoch": 0.7982477800617638,
      "grad_norm": 0.11525259166955948,
      "learning_rate": 7.513237217695566e-06,
      "loss": 0.0028,
      "step": 124460
    },
    {
      "epoch": 0.7983119169555499,
      "grad_norm": 0.01986979879438877,
      "learning_rate": 7.512753343142334e-06,
      "loss": 0.0025,
      "step": 124470
    },
    {
      "epoch": 0.7983760538493361,
      "grad_norm": 0.1199314221739769,
      "learning_rate": 7.512269437102506e-06,
      "loss": 0.0023,
      "step": 124480
    },
    {
      "epoch": 0.7984401907431221,
      "grad_norm": 0.05994654819369316,
      "learning_rate": 7.5117854995821495e-06,
      "loss": 0.0038,
      "step": 124490
    },
    {
      "epoch": 0.7985043276369083,
      "grad_norm": 0.27077120542526245,
      "learning_rate": 7.511301530587322e-06,
      "loss": 0.0031,
      "step": 124500
    },
    {
      "epoch": 0.7985684645306943,
      "grad_norm": 0.17333291471004486,
      "learning_rate": 7.510817530124094e-06,
      "loss": 0.0022,
      "step": 124510
    },
    {
      "epoch": 0.7986326014244804,
      "grad_norm": 0.020280921831727028,
      "learning_rate": 7.510333498198525e-06,
      "loss": 0.0024,
      "step": 124520
    },
    {
      "epoch": 0.7986967383182665,
      "grad_norm": 0.17286615073680878,
      "learning_rate": 7.509849434816687e-06,
      "loss": 0.0017,
      "step": 124530
    },
    {
      "epoch": 0.7987608752120526,
      "grad_norm": 0.09009113162755966,
      "learning_rate": 7.509365339984639e-06,
      "loss": 0.003,
      "step": 124540
    },
    {
      "epoch": 0.7988250121058387,
      "grad_norm": 0.034896817058324814,
      "learning_rate": 7.508881213708451e-06,
      "loss": 0.0019,
      "step": 124550
    },
    {
      "epoch": 0.7988891489996248,
      "grad_norm": 0.15198229253292084,
      "learning_rate": 7.508397055994188e-06,
      "loss": 0.0032,
      "step": 124560
    },
    {
      "epoch": 0.7989532858934109,
      "grad_norm": 0.1587715595960617,
      "learning_rate": 7.507912866847918e-06,
      "loss": 0.0031,
      "step": 124570
    },
    {
      "epoch": 0.799017422787197,
      "grad_norm": 0.07122717797756195,
      "learning_rate": 7.507428646275705e-06,
      "loss": 0.0024,
      "step": 124580
    },
    {
      "epoch": 0.799081559680983,
      "grad_norm": 0.3986320197582245,
      "learning_rate": 7.506944394283622e-06,
      "loss": 0.0031,
      "step": 124590
    },
    {
      "epoch": 0.7991456965747692,
      "grad_norm": 0.1260489672422409,
      "learning_rate": 7.5064601108777315e-06,
      "loss": 0.0028,
      "step": 124600
    },
    {
      "epoch": 0.7992098334685552,
      "grad_norm": 0.10430359840393066,
      "learning_rate": 7.505975796064106e-06,
      "loss": 0.0025,
      "step": 124610
    },
    {
      "epoch": 0.7992739703623414,
      "grad_norm": 0.18059343099594116,
      "learning_rate": 7.505491449848812e-06,
      "loss": 0.0026,
      "step": 124620
    },
    {
      "epoch": 0.7993381072561275,
      "grad_norm": 0.1164829358458519,
      "learning_rate": 7.50500707223792e-06,
      "loss": 0.0038,
      "step": 124630
    },
    {
      "epoch": 0.7994022441499136,
      "grad_norm": 0.13949273526668549,
      "learning_rate": 7.504522663237499e-06,
      "loss": 0.0028,
      "step": 124640
    },
    {
      "epoch": 0.7994663810436997,
      "grad_norm": 0.14682637155056,
      "learning_rate": 7.504038222853619e-06,
      "loss": 0.002,
      "step": 124650
    },
    {
      "epoch": 0.7995305179374858,
      "grad_norm": 0.3267804682254791,
      "learning_rate": 7.503553751092352e-06,
      "loss": 0.0012,
      "step": 124660
    },
    {
      "epoch": 0.7995946548312719,
      "grad_norm": 0.0706501454114914,
      "learning_rate": 7.503069247959765e-06,
      "loss": 0.0032,
      "step": 124670
    },
    {
      "epoch": 0.7996587917250579,
      "grad_norm": 0.3418445289134979,
      "learning_rate": 7.502584713461931e-06,
      "loss": 0.005,
      "step": 124680
    },
    {
      "epoch": 0.7997229286188441,
      "grad_norm": 0.1267576962709427,
      "learning_rate": 7.5021001476049225e-06,
      "loss": 0.0031,
      "step": 124690
    },
    {
      "epoch": 0.7997870655126301,
      "grad_norm": 0.11980558186769485,
      "learning_rate": 7.5016155503948116e-06,
      "loss": 0.003,
      "step": 124700
    },
    {
      "epoch": 0.7998512024064163,
      "grad_norm": 0.10078209638595581,
      "learning_rate": 7.501130921837671e-06,
      "loss": 0.0037,
      "step": 124710
    },
    {
      "epoch": 0.7999153393002023,
      "grad_norm": 0.15275952219963074,
      "learning_rate": 7.500646261939571e-06,
      "loss": 0.0024,
      "step": 124720
    },
    {
      "epoch": 0.7999794761939885,
      "grad_norm": 0.3661182224750519,
      "learning_rate": 7.500161570706586e-06,
      "loss": 0.0031,
      "step": 124730
    },
    {
      "epoch": 0.8000436130877745,
      "grad_norm": 0.0601467490196228,
      "learning_rate": 7.499676848144791e-06,
      "loss": 0.0022,
      "step": 124740
    },
    {
      "epoch": 0.8001077499815606,
      "grad_norm": 0.06747540086507797,
      "learning_rate": 7.499192094260257e-06,
      "loss": 0.0022,
      "step": 124750
    },
    {
      "epoch": 0.8001718868753468,
      "grad_norm": 0.19947946071624756,
      "learning_rate": 7.498707309059061e-06,
      "loss": 0.0017,
      "step": 124760
    },
    {
      "epoch": 0.8002360237691328,
      "grad_norm": 0.15963557362556458,
      "learning_rate": 7.498222492547277e-06,
      "loss": 0.0026,
      "step": 124770
    },
    {
      "epoch": 0.800300160662919,
      "grad_norm": 0.1204855665564537,
      "learning_rate": 7.497737644730979e-06,
      "loss": 0.0015,
      "step": 124780
    },
    {
      "epoch": 0.800364297556705,
      "grad_norm": 0.03258391097187996,
      "learning_rate": 7.497252765616243e-06,
      "loss": 0.0021,
      "step": 124790
    },
    {
      "epoch": 0.8004284344504912,
      "grad_norm": 0.13139325380325317,
      "learning_rate": 7.496767855209146e-06,
      "loss": 0.0027,
      "step": 124800
    },
    {
      "epoch": 0.8004925713442772,
      "grad_norm": 0.10871946066617966,
      "learning_rate": 7.4962829135157625e-06,
      "loss": 0.0024,
      "step": 124810
    },
    {
      "epoch": 0.8005567082380634,
      "grad_norm": 0.1199418231844902,
      "learning_rate": 7.495797940542169e-06,
      "loss": 0.0014,
      "step": 124820
    },
    {
      "epoch": 0.8006208451318494,
      "grad_norm": 0.12236741185188293,
      "learning_rate": 7.495312936294445e-06,
      "loss": 0.0034,
      "step": 124830
    },
    {
      "epoch": 0.8006849820256355,
      "grad_norm": 0.15686768293380737,
      "learning_rate": 7.494827900778667e-06,
      "loss": 0.0028,
      "step": 124840
    },
    {
      "epoch": 0.8007491189194216,
      "grad_norm": 0.203244149684906,
      "learning_rate": 7.494342834000912e-06,
      "loss": 0.0058,
      "step": 124850
    },
    {
      "epoch": 0.8008132558132077,
      "grad_norm": 0.055326201021671295,
      "learning_rate": 7.493857735967258e-06,
      "loss": 0.003,
      "step": 124860
    },
    {
      "epoch": 0.8008773927069938,
      "grad_norm": 0.1566278487443924,
      "learning_rate": 7.493372606683784e-06,
      "loss": 0.0028,
      "step": 124870
    },
    {
      "epoch": 0.8009415296007799,
      "grad_norm": 0.02740250527858734,
      "learning_rate": 7.49288744615657e-06,
      "loss": 0.0022,
      "step": 124880
    },
    {
      "epoch": 0.801005666494566,
      "grad_norm": 0.12024752795696259,
      "learning_rate": 7.4924022543916955e-06,
      "loss": 0.0019,
      "step": 124890
    },
    {
      "epoch": 0.8010698033883521,
      "grad_norm": 0.02664109319448471,
      "learning_rate": 7.491917031395237e-06,
      "loss": 0.0033,
      "step": 124900
    },
    {
      "epoch": 0.8011339402821382,
      "grad_norm": 0.12732449173927307,
      "learning_rate": 7.49143177717328e-06,
      "loss": 0.0025,
      "step": 124910
    },
    {
      "epoch": 0.8011980771759243,
      "grad_norm": 0.06920778006315231,
      "learning_rate": 7.490946491731901e-06,
      "loss": 0.002,
      "step": 124920
    },
    {
      "epoch": 0.8012622140697104,
      "grad_norm": 0.14909569919109344,
      "learning_rate": 7.490461175077182e-06,
      "loss": 0.0013,
      "step": 124930
    },
    {
      "epoch": 0.8013263509634965,
      "grad_norm": 0.06066959351301193,
      "learning_rate": 7.489975827215203e-06,
      "loss": 0.0018,
      "step": 124940
    },
    {
      "epoch": 0.8013904878572826,
      "grad_norm": 0.14901049435138702,
      "learning_rate": 7.489490448152049e-06,
      "loss": 0.0025,
      "step": 124950
    },
    {
      "epoch": 0.8014546247510687,
      "grad_norm": 0.08036942780017853,
      "learning_rate": 7.489005037893802e-06,
      "loss": 0.004,
      "step": 124960
    },
    {
      "epoch": 0.8015187616448548,
      "grad_norm": 0.4628899097442627,
      "learning_rate": 7.4885195964465396e-06,
      "loss": 0.0079,
      "step": 124970
    },
    {
      "epoch": 0.8015828985386408,
      "grad_norm": 0.11064311861991882,
      "learning_rate": 7.488034123816351e-06,
      "loss": 0.0023,
      "step": 124980
    },
    {
      "epoch": 0.801647035432427,
      "grad_norm": 0.20454680919647217,
      "learning_rate": 7.487548620009315e-06,
      "loss": 0.0027,
      "step": 124990
    },
    {
      "epoch": 0.801711172326213,
      "grad_norm": 0.09361718595027924,
      "learning_rate": 7.487063085031519e-06,
      "loss": 0.0015,
      "step": 125000
    },
    {
      "epoch": 0.8017753092199992,
      "grad_norm": 0.08985812216997147,
      "learning_rate": 7.486577518889042e-06,
      "loss": 0.0038,
      "step": 125010
    },
    {
      "epoch": 0.8018394461137852,
      "grad_norm": 0.008312336169183254,
      "learning_rate": 7.486091921587975e-06,
      "loss": 0.003,
      "step": 125020
    },
    {
      "epoch": 0.8019035830075714,
      "grad_norm": 0.11346910893917084,
      "learning_rate": 7.485606293134397e-06,
      "loss": 0.0019,
      "step": 125030
    },
    {
      "epoch": 0.8019677199013575,
      "grad_norm": 0.11304417997598648,
      "learning_rate": 7.485120633534396e-06,
      "loss": 0.0028,
      "step": 125040
    },
    {
      "epoch": 0.8020318567951435,
      "grad_norm": 0.33306020498275757,
      "learning_rate": 7.484634942794058e-06,
      "loss": 0.0026,
      "step": 125050
    },
    {
      "epoch": 0.8020959936889297,
      "grad_norm": 0.10592376440763474,
      "learning_rate": 7.484149220919468e-06,
      "loss": 0.0014,
      "step": 125060
    },
    {
      "epoch": 0.8021601305827157,
      "grad_norm": 0.11216168850660324,
      "learning_rate": 7.483663467916712e-06,
      "loss": 0.0015,
      "step": 125070
    },
    {
      "epoch": 0.8022242674765019,
      "grad_norm": 0.1404658704996109,
      "learning_rate": 7.483177683791879e-06,
      "loss": 0.002,
      "step": 125080
    },
    {
      "epoch": 0.8022884043702879,
      "grad_norm": 0.34358495473861694,
      "learning_rate": 7.482691868551054e-06,
      "loss": 0.002,
      "step": 125090
    },
    {
      "epoch": 0.8023525412640741,
      "grad_norm": 0.008700758218765259,
      "learning_rate": 7.482206022200327e-06,
      "loss": 0.0022,
      "step": 125100
    },
    {
      "epoch": 0.8024166781578601,
      "grad_norm": 0.0656890794634819,
      "learning_rate": 7.481720144745783e-06,
      "loss": 0.0028,
      "step": 125110
    },
    {
      "epoch": 0.8024808150516463,
      "grad_norm": 0.06679105758666992,
      "learning_rate": 7.481234236193513e-06,
      "loss": 0.004,
      "step": 125120
    },
    {
      "epoch": 0.8025449519454323,
      "grad_norm": 0.09332876652479172,
      "learning_rate": 7.480748296549605e-06,
      "loss": 0.0035,
      "step": 125130
    },
    {
      "epoch": 0.8026090888392184,
      "grad_norm": 0.12459799647331238,
      "learning_rate": 7.480262325820147e-06,
      "loss": 0.0024,
      "step": 125140
    },
    {
      "epoch": 0.8026732257330045,
      "grad_norm": 0.17064888775348663,
      "learning_rate": 7.47977632401123e-06,
      "loss": 0.0026,
      "step": 125150
    },
    {
      "epoch": 0.8027373626267906,
      "grad_norm": 0.060531724244356155,
      "learning_rate": 7.479290291128942e-06,
      "loss": 0.0013,
      "step": 125160
    },
    {
      "epoch": 0.8028014995205767,
      "grad_norm": 0.08411803096532822,
      "learning_rate": 7.478804227179376e-06,
      "loss": 0.0048,
      "step": 125170
    },
    {
      "epoch": 0.8028656364143628,
      "grad_norm": 0.17802155017852783,
      "learning_rate": 7.478318132168621e-06,
      "loss": 0.0024,
      "step": 125180
    },
    {
      "epoch": 0.802929773308149,
      "grad_norm": 0.12030982971191406,
      "learning_rate": 7.47783200610277e-06,
      "loss": 0.0029,
      "step": 125190
    },
    {
      "epoch": 0.802993910201935,
      "grad_norm": 0.08102941513061523,
      "learning_rate": 7.477345848987911e-06,
      "loss": 0.0035,
      "step": 125200
    },
    {
      "epoch": 0.8030580470957212,
      "grad_norm": 0.06955002248287201,
      "learning_rate": 7.476859660830139e-06,
      "loss": 0.0022,
      "step": 125210
    },
    {
      "epoch": 0.8031221839895072,
      "grad_norm": 0.8118174076080322,
      "learning_rate": 7.476373441635545e-06,
      "loss": 0.0046,
      "step": 125220
    },
    {
      "epoch": 0.8031863208832933,
      "grad_norm": 0.18293973803520203,
      "learning_rate": 7.475887191410223e-06,
      "loss": 0.0012,
      "step": 125230
    },
    {
      "epoch": 0.8032504577770794,
      "grad_norm": 0.14403167366981506,
      "learning_rate": 7.4754009101602635e-06,
      "loss": 0.0018,
      "step": 125240
    },
    {
      "epoch": 0.8033145946708655,
      "grad_norm": 0.29412204027175903,
      "learning_rate": 7.474914597891763e-06,
      "loss": 0.003,
      "step": 125250
    },
    {
      "epoch": 0.8033787315646516,
      "grad_norm": 0.11173876374959946,
      "learning_rate": 7.474428254610812e-06,
      "loss": 0.0017,
      "step": 125260
    },
    {
      "epoch": 0.8034428684584377,
      "grad_norm": 0.08756504207849503,
      "learning_rate": 7.473941880323507e-06,
      "loss": 0.0023,
      "step": 125270
    },
    {
      "epoch": 0.8035070053522237,
      "grad_norm": 0.04347633942961693,
      "learning_rate": 7.473455475035942e-06,
      "loss": 0.004,
      "step": 125280
    },
    {
      "epoch": 0.8035711422460099,
      "grad_norm": 0.029598882421851158,
      "learning_rate": 7.472969038754214e-06,
      "loss": 0.0013,
      "step": 125290
    },
    {
      "epoch": 0.8036352791397959,
      "grad_norm": 0.09162858873605728,
      "learning_rate": 7.472482571484414e-06,
      "loss": 0.0033,
      "step": 125300
    },
    {
      "epoch": 0.8036994160335821,
      "grad_norm": 0.08080438524484634,
      "learning_rate": 7.471996073232641e-06,
      "loss": 0.0038,
      "step": 125310
    },
    {
      "epoch": 0.8037635529273682,
      "grad_norm": 0.1008782610297203,
      "learning_rate": 7.471509544004993e-06,
      "loss": 0.0031,
      "step": 125320
    },
    {
      "epoch": 0.8038276898211543,
      "grad_norm": 0.12188203632831573,
      "learning_rate": 7.471022983807561e-06,
      "loss": 0.0015,
      "step": 125330
    },
    {
      "epoch": 0.8038918267149404,
      "grad_norm": 0.056016705930233,
      "learning_rate": 7.470536392646446e-06,
      "loss": 0.0017,
      "step": 125340
    },
    {
      "epoch": 0.8039559636087265,
      "grad_norm": 0.17744384706020355,
      "learning_rate": 7.470049770527744e-06,
      "loss": 0.0016,
      "step": 125350
    },
    {
      "epoch": 0.8040201005025126,
      "grad_norm": 0.06114094331860542,
      "learning_rate": 7.469563117457555e-06,
      "loss": 0.0024,
      "step": 125360
    },
    {
      "epoch": 0.8040842373962986,
      "grad_norm": 0.031616564840078354,
      "learning_rate": 7.469076433441972e-06,
      "loss": 0.0034,
      "step": 125370
    },
    {
      "epoch": 0.8041483742900848,
      "grad_norm": 0.11400750279426575,
      "learning_rate": 7.468589718487099e-06,
      "loss": 0.0021,
      "step": 125380
    },
    {
      "epoch": 0.8042125111838708,
      "grad_norm": 0.11139538139104843,
      "learning_rate": 7.468102972599032e-06,
      "loss": 0.0022,
      "step": 125390
    },
    {
      "epoch": 0.804276648077657,
      "grad_norm": 0.011781953275203705,
      "learning_rate": 7.467616195783871e-06,
      "loss": 0.0041,
      "step": 125400
    },
    {
      "epoch": 0.804340784971443,
      "grad_norm": 0.08708330988883972,
      "learning_rate": 7.467129388047714e-06,
      "loss": 0.0023,
      "step": 125410
    },
    {
      "epoch": 0.8044049218652292,
      "grad_norm": 0.11689251661300659,
      "learning_rate": 7.4666425493966646e-06,
      "loss": 0.0022,
      "step": 125420
    },
    {
      "epoch": 0.8044690587590152,
      "grad_norm": 0.16960015892982483,
      "learning_rate": 7.46615567983682e-06,
      "loss": 0.0034,
      "step": 125430
    },
    {
      "epoch": 0.8045331956528013,
      "grad_norm": 0.13801692426204681,
      "learning_rate": 7.4656687793742826e-06,
      "loss": 0.0024,
      "step": 125440
    },
    {
      "epoch": 0.8045973325465874,
      "grad_norm": 0.12104618549346924,
      "learning_rate": 7.465181848015152e-06,
      "loss": 0.0029,
      "step": 125450
    },
    {
      "epoch": 0.8046614694403735,
      "grad_norm": 0.14934170246124268,
      "learning_rate": 7.464694885765532e-06,
      "loss": 0.0032,
      "step": 125460
    },
    {
      "epoch": 0.8047256063341597,
      "grad_norm": 0.08867646753787994,
      "learning_rate": 7.464207892631523e-06,
      "loss": 0.0041,
      "step": 125470
    },
    {
      "epoch": 0.8047897432279457,
      "grad_norm": 0.029540695250034332,
      "learning_rate": 7.4637208686192295e-06,
      "loss": 0.0027,
      "step": 125480
    },
    {
      "epoch": 0.8048538801217319,
      "grad_norm": 0.042014963924884796,
      "learning_rate": 7.463233813734752e-06,
      "loss": 0.0021,
      "step": 125490
    },
    {
      "epoch": 0.8049180170155179,
      "grad_norm": 0.07194870710372925,
      "learning_rate": 7.462746727984193e-06,
      "loss": 0.0023,
      "step": 125500
    },
    {
      "epoch": 0.804982153909304,
      "grad_norm": 0.11041359603404999,
      "learning_rate": 7.462259611373659e-06,
      "loss": 0.0025,
      "step": 125510
    },
    {
      "epoch": 0.8050462908030901,
      "grad_norm": 0.08371560275554657,
      "learning_rate": 7.461772463909252e-06,
      "loss": 0.0029,
      "step": 125520
    },
    {
      "epoch": 0.8051104276968762,
      "grad_norm": 0.1286308616399765,
      "learning_rate": 7.461285285597078e-06,
      "loss": 0.0046,
      "step": 125530
    },
    {
      "epoch": 0.8051745645906623,
      "grad_norm": 0.09027030318975449,
      "learning_rate": 7.460798076443237e-06,
      "loss": 0.0042,
      "step": 125540
    },
    {
      "epoch": 0.8052387014844484,
      "grad_norm": 0.020330199971795082,
      "learning_rate": 7.46031083645384e-06,
      "loss": 0.0015,
      "step": 125550
    },
    {
      "epoch": 0.8053028383782345,
      "grad_norm": 0.028711901977658272,
      "learning_rate": 7.459823565634989e-06,
      "loss": 0.0025,
      "step": 125560
    },
    {
      "epoch": 0.8053669752720206,
      "grad_norm": 0.04168624058365822,
      "learning_rate": 7.459336263992791e-06,
      "loss": 0.0028,
      "step": 125570
    },
    {
      "epoch": 0.8054311121658067,
      "grad_norm": 0.13903804123401642,
      "learning_rate": 7.4588489315333514e-06,
      "loss": 0.0028,
      "step": 125580
    },
    {
      "epoch": 0.8054952490595928,
      "grad_norm": 0.07040738314390182,
      "learning_rate": 7.458361568262778e-06,
      "loss": 0.0016,
      "step": 125590
    },
    {
      "epoch": 0.8055593859533788,
      "grad_norm": 0.03716859221458435,
      "learning_rate": 7.457874174187176e-06,
      "loss": 0.0037,
      "step": 125600
    },
    {
      "epoch": 0.805623522847165,
      "grad_norm": 0.07053796201944351,
      "learning_rate": 7.457386749312656e-06,
      "loss": 0.0024,
      "step": 125610
    },
    {
      "epoch": 0.8056876597409511,
      "grad_norm": 0.055847663432359695,
      "learning_rate": 7.456899293645322e-06,
      "loss": 0.0021,
      "step": 125620
    },
    {
      "epoch": 0.8057517966347372,
      "grad_norm": 0.1960245817899704,
      "learning_rate": 7.456411807191286e-06,
      "loss": 0.0023,
      "step": 125630
    },
    {
      "epoch": 0.8058159335285233,
      "grad_norm": 0.08720597624778748,
      "learning_rate": 7.455924289956653e-06,
      "loss": 0.0031,
      "step": 125640
    },
    {
      "epoch": 0.8058800704223094,
      "grad_norm": 0.11513277888298035,
      "learning_rate": 7.455436741947533e-06,
      "loss": 0.0023,
      "step": 125650
    },
    {
      "epoch": 0.8059442073160955,
      "grad_norm": 0.18705707788467407,
      "learning_rate": 7.454949163170037e-06,
      "loss": 0.0016,
      "step": 125660
    },
    {
      "epoch": 0.8060083442098815,
      "grad_norm": 0.16182366013526917,
      "learning_rate": 7.454461553630272e-06,
      "loss": 0.0029,
      "step": 125670
    },
    {
      "epoch": 0.8060724811036677,
      "grad_norm": 0.05336945131421089,
      "learning_rate": 7.45397391333435e-06,
      "loss": 0.0026,
      "step": 125680
    },
    {
      "epoch": 0.8061366179974537,
      "grad_norm": 0.04091660678386688,
      "learning_rate": 7.453486242288381e-06,
      "loss": 0.0036,
      "step": 125690
    },
    {
      "epoch": 0.8062007548912399,
      "grad_norm": 0.17214728891849518,
      "learning_rate": 7.452998540498477e-06,
      "loss": 0.0017,
      "step": 125700
    },
    {
      "epoch": 0.8062648917850259,
      "grad_norm": 0.20068220794200897,
      "learning_rate": 7.452510807970747e-06,
      "loss": 0.0024,
      "step": 125710
    },
    {
      "epoch": 0.8063290286788121,
      "grad_norm": 0.026574088260531425,
      "learning_rate": 7.452023044711305e-06,
      "loss": 0.0016,
      "step": 125720
    },
    {
      "epoch": 0.8063931655725981,
      "grad_norm": 0.09764420241117477,
      "learning_rate": 7.451535250726261e-06,
      "loss": 0.0043,
      "step": 125730
    },
    {
      "epoch": 0.8064573024663843,
      "grad_norm": 0.09181403368711472,
      "learning_rate": 7.451047426021728e-06,
      "loss": 0.0029,
      "step": 125740
    },
    {
      "epoch": 0.8065214393601704,
      "grad_norm": 0.0713183730840683,
      "learning_rate": 7.450559570603819e-06,
      "loss": 0.002,
      "step": 125750
    },
    {
      "epoch": 0.8065855762539564,
      "grad_norm": 0.1376122534275055,
      "learning_rate": 7.450071684478648e-06,
      "loss": 0.0015,
      "step": 125760
    },
    {
      "epoch": 0.8066497131477426,
      "grad_norm": 0.08532290905714035,
      "learning_rate": 7.449583767652326e-06,
      "loss": 0.0018,
      "step": 125770
    },
    {
      "epoch": 0.8067138500415286,
      "grad_norm": 0.09605459123849869,
      "learning_rate": 7.449095820130972e-06,
      "loss": 0.005,
      "step": 125780
    },
    {
      "epoch": 0.8067779869353148,
      "grad_norm": 0.0583382323384285,
      "learning_rate": 7.448607841920693e-06,
      "loss": 0.0028,
      "step": 125790
    },
    {
      "epoch": 0.8068421238291008,
      "grad_norm": 0.10395050793886185,
      "learning_rate": 7.448119833027609e-06,
      "loss": 0.003,
      "step": 125800
    },
    {
      "epoch": 0.806906260722887,
      "grad_norm": 0.13060331344604492,
      "learning_rate": 7.447631793457836e-06,
      "loss": 0.0027,
      "step": 125810
    },
    {
      "epoch": 0.806970397616673,
      "grad_norm": 0.09807195514440536,
      "learning_rate": 7.447143723217485e-06,
      "loss": 0.003,
      "step": 125820
    },
    {
      "epoch": 0.8070345345104591,
      "grad_norm": 0.0894220769405365,
      "learning_rate": 7.446655622312675e-06,
      "loss": 0.0023,
      "step": 125830
    },
    {
      "epoch": 0.8070986714042452,
      "grad_norm": 0.4230436682701111,
      "learning_rate": 7.446167490749522e-06,
      "loss": 0.0063,
      "step": 125840
    },
    {
      "epoch": 0.8071628082980313,
      "grad_norm": 0.20460151135921478,
      "learning_rate": 7.445679328534141e-06,
      "loss": 0.0021,
      "step": 125850
    },
    {
      "epoch": 0.8072269451918174,
      "grad_norm": 0.07890148460865021,
      "learning_rate": 7.445191135672652e-06,
      "loss": 0.0021,
      "step": 125860
    },
    {
      "epoch": 0.8072910820856035,
      "grad_norm": 0.01286834292113781,
      "learning_rate": 7.444702912171168e-06,
      "loss": 0.0019,
      "step": 125870
    },
    {
      "epoch": 0.8073552189793896,
      "grad_norm": 0.030550939962267876,
      "learning_rate": 7.44421465803581e-06,
      "loss": 0.0033,
      "step": 125880
    },
    {
      "epoch": 0.8074193558731757,
      "grad_norm": 0.1814279854297638,
      "learning_rate": 7.443726373272696e-06,
      "loss": 0.0048,
      "step": 125890
    },
    {
      "epoch": 0.8074834927669619,
      "grad_norm": 0.11485601961612701,
      "learning_rate": 7.443238057887943e-06,
      "loss": 0.0021,
      "step": 125900
    },
    {
      "epoch": 0.8075476296607479,
      "grad_norm": 0.09258340299129486,
      "learning_rate": 7.442749711887673e-06,
      "loss": 0.0016,
      "step": 125910
    },
    {
      "epoch": 0.807611766554534,
      "grad_norm": 0.10012565553188324,
      "learning_rate": 7.442261335278003e-06,
      "loss": 0.002,
      "step": 125920
    },
    {
      "epoch": 0.8076759034483201,
      "grad_norm": 0.18450987339019775,
      "learning_rate": 7.441772928065052e-06,
      "loss": 0.0016,
      "step": 125930
    },
    {
      "epoch": 0.8077400403421062,
      "grad_norm": 0.24779313802719116,
      "learning_rate": 7.4412844902549405e-06,
      "loss": 0.0046,
      "step": 125940
    },
    {
      "epoch": 0.8078041772358923,
      "grad_norm": 0.2482410967350006,
      "learning_rate": 7.440796021853792e-06,
      "loss": 0.0042,
      "step": 125950
    },
    {
      "epoch": 0.8078683141296784,
      "grad_norm": 0.1761637181043625,
      "learning_rate": 7.440307522867722e-06,
      "loss": 0.0031,
      "step": 125960
    },
    {
      "epoch": 0.8079324510234644,
      "grad_norm": 0.21982888877391815,
      "learning_rate": 7.4398189933028576e-06,
      "loss": 0.004,
      "step": 125970
    },
    {
      "epoch": 0.8079965879172506,
      "grad_norm": 0.022976882755756378,
      "learning_rate": 7.439330433165315e-06,
      "loss": 0.0025,
      "step": 125980
    },
    {
      "epoch": 0.8080607248110366,
      "grad_norm": 0.06587019562721252,
      "learning_rate": 7.438841842461221e-06,
      "loss": 0.0018,
      "step": 125990
    },
    {
      "epoch": 0.8081248617048228,
      "grad_norm": 0.09989341348409653,
      "learning_rate": 7.438353221196693e-06,
      "loss": 0.0023,
      "step": 126000
    },
    {
      "epoch": 0.8081889985986088,
      "grad_norm": 0.15177001059055328,
      "learning_rate": 7.437864569377858e-06,
      "loss": 0.0022,
      "step": 126010
    },
    {
      "epoch": 0.808253135492395,
      "grad_norm": 0.28216689825057983,
      "learning_rate": 7.437375887010838e-06,
      "loss": 0.0027,
      "step": 126020
    },
    {
      "epoch": 0.8083172723861811,
      "grad_norm": 0.05767377093434334,
      "learning_rate": 7.436887174101756e-06,
      "loss": 0.0023,
      "step": 126030
    },
    {
      "epoch": 0.8083814092799672,
      "grad_norm": 0.1304662525653839,
      "learning_rate": 7.436398430656736e-06,
      "loss": 0.0025,
      "step": 126040
    },
    {
      "epoch": 0.8084455461737533,
      "grad_norm": 0.013885071501135826,
      "learning_rate": 7.435909656681901e-06,
      "loss": 0.0023,
      "step": 126050
    },
    {
      "epoch": 0.8085096830675393,
      "grad_norm": 0.3168763816356659,
      "learning_rate": 7.435420852183379e-06,
      "loss": 0.0033,
      "step": 126060
    },
    {
      "epoch": 0.8085738199613255,
      "grad_norm": 0.17479267716407776,
      "learning_rate": 7.4349320171672926e-06,
      "loss": 0.002,
      "step": 126070
    },
    {
      "epoch": 0.8086379568551115,
      "grad_norm": 0.1837267279624939,
      "learning_rate": 7.434443151639767e-06,
      "loss": 0.0023,
      "step": 126080
    },
    {
      "epoch": 0.8087020937488977,
      "grad_norm": 0.09188154339790344,
      "learning_rate": 7.43395425560693e-06,
      "loss": 0.0024,
      "step": 126090
    },
    {
      "epoch": 0.8087662306426837,
      "grad_norm": 0.0633850023150444,
      "learning_rate": 7.433465329074907e-06,
      "loss": 0.0038,
      "step": 126100
    },
    {
      "epoch": 0.8088303675364699,
      "grad_norm": 0.07520250976085663,
      "learning_rate": 7.4329763720498226e-06,
      "loss": 0.0019,
      "step": 126110
    },
    {
      "epoch": 0.8088945044302559,
      "grad_norm": 0.09829988330602646,
      "learning_rate": 7.432487384537806e-06,
      "loss": 0.001,
      "step": 126120
    },
    {
      "epoch": 0.808958641324042,
      "grad_norm": 0.09438778460025787,
      "learning_rate": 7.431998366544985e-06,
      "loss": 0.0026,
      "step": 126130
    },
    {
      "epoch": 0.8090227782178281,
      "grad_norm": 0.085836261510849,
      "learning_rate": 7.431509318077486e-06,
      "loss": 0.0028,
      "step": 126140
    },
    {
      "epoch": 0.8090869151116142,
      "grad_norm": 0.10999659448862076,
      "learning_rate": 7.431020239141437e-06,
      "loss": 0.0036,
      "step": 126150
    },
    {
      "epoch": 0.8091510520054003,
      "grad_norm": 0.14788612723350525,
      "learning_rate": 7.4305311297429665e-06,
      "loss": 0.003,
      "step": 126160
    },
    {
      "epoch": 0.8092151888991864,
      "grad_norm": 0.12257693707942963,
      "learning_rate": 7.430041989888204e-06,
      "loss": 0.0038,
      "step": 126170
    },
    {
      "epoch": 0.8092793257929726,
      "grad_norm": 0.028312664479017258,
      "learning_rate": 7.429552819583279e-06,
      "loss": 0.0033,
      "step": 126180
    },
    {
      "epoch": 0.8093434626867586,
      "grad_norm": 0.16652943193912506,
      "learning_rate": 7.429063618834321e-06,
      "loss": 0.002,
      "step": 126190
    },
    {
      "epoch": 0.8094075995805448,
      "grad_norm": 0.12218911200761795,
      "learning_rate": 7.428574387647459e-06,
      "loss": 0.0032,
      "step": 126200
    },
    {
      "epoch": 0.8094717364743308,
      "grad_norm": 0.14190293848514557,
      "learning_rate": 7.428085126028826e-06,
      "loss": 0.002,
      "step": 126210
    },
    {
      "epoch": 0.8095358733681169,
      "grad_norm": 0.0556914247572422,
      "learning_rate": 7.427595833984549e-06,
      "loss": 0.0013,
      "step": 126220
    },
    {
      "epoch": 0.809600010261903,
      "grad_norm": 0.10317368060350418,
      "learning_rate": 7.427106511520762e-06,
      "loss": 0.003,
      "step": 126230
    },
    {
      "epoch": 0.8096641471556891,
      "grad_norm": 0.02198263444006443,
      "learning_rate": 7.426617158643595e-06,
      "loss": 0.0017,
      "step": 126240
    },
    {
      "epoch": 0.8097282840494752,
      "grad_norm": 0.0942700058221817,
      "learning_rate": 7.426127775359183e-06,
      "loss": 0.0028,
      "step": 126250
    },
    {
      "epoch": 0.8097924209432613,
      "grad_norm": 0.11407773941755295,
      "learning_rate": 7.425638361673655e-06,
      "loss": 0.0016,
      "step": 126260
    },
    {
      "epoch": 0.8098565578370474,
      "grad_norm": 0.08928665518760681,
      "learning_rate": 7.425148917593145e-06,
      "loss": 0.0025,
      "step": 126270
    },
    {
      "epoch": 0.8099206947308335,
      "grad_norm": 0.05590301752090454,
      "learning_rate": 7.424659443123784e-06,
      "loss": 0.0034,
      "step": 126280
    },
    {
      "epoch": 0.8099848316246195,
      "grad_norm": 0.08511009812355042,
      "learning_rate": 7.4241699382717095e-06,
      "loss": 0.0033,
      "step": 126290
    },
    {
      "epoch": 0.8100489685184057,
      "grad_norm": 0.10576985031366348,
      "learning_rate": 7.423680403043052e-06,
      "loss": 0.0021,
      "step": 126300
    },
    {
      "epoch": 0.8101131054121918,
      "grad_norm": 0.09998160600662231,
      "learning_rate": 7.423190837443948e-06,
      "loss": 0.0025,
      "step": 126310
    },
    {
      "epoch": 0.8101772423059779,
      "grad_norm": 0.3955870568752289,
      "learning_rate": 7.422701241480529e-06,
      "loss": 0.0019,
      "step": 126320
    },
    {
      "epoch": 0.810241379199764,
      "grad_norm": 0.06922601163387299,
      "learning_rate": 7.422211615158934e-06,
      "loss": 0.0037,
      "step": 126330
    },
    {
      "epoch": 0.8103055160935501,
      "grad_norm": 0.252462774515152,
      "learning_rate": 7.421721958485295e-06,
      "loss": 0.003,
      "step": 126340
    },
    {
      "epoch": 0.8103696529873362,
      "grad_norm": 0.10331380367279053,
      "learning_rate": 7.42123227146575e-06,
      "loss": 0.0022,
      "step": 126350
    },
    {
      "epoch": 0.8104337898811222,
      "grad_norm": 0.10928361862897873,
      "learning_rate": 7.420742554106433e-06,
      "loss": 0.003,
      "step": 126360
    },
    {
      "epoch": 0.8104979267749084,
      "grad_norm": 0.5361838936805725,
      "learning_rate": 7.420252806413482e-06,
      "loss": 0.0019,
      "step": 126370
    },
    {
      "epoch": 0.8105620636686944,
      "grad_norm": 0.1608307957649231,
      "learning_rate": 7.419763028393034e-06,
      "loss": 0.0026,
      "step": 126380
    },
    {
      "epoch": 0.8106262005624806,
      "grad_norm": 0.36987513303756714,
      "learning_rate": 7.419273220051226e-06,
      "loss": 0.0085,
      "step": 126390
    },
    {
      "epoch": 0.8106903374562666,
      "grad_norm": 0.0979725793004036,
      "learning_rate": 7.418783381394195e-06,
      "loss": 0.0018,
      "step": 126400
    },
    {
      "epoch": 0.8107544743500528,
      "grad_norm": 0.15433906018733978,
      "learning_rate": 7.418293512428081e-06,
      "loss": 0.0026,
      "step": 126410
    },
    {
      "epoch": 0.8108186112438388,
      "grad_norm": 0.05753166601061821,
      "learning_rate": 7.41780361315902e-06,
      "loss": 0.0022,
      "step": 126420
    },
    {
      "epoch": 0.810882748137625,
      "grad_norm": 0.160835862159729,
      "learning_rate": 7.417313683593152e-06,
      "loss": 0.0023,
      "step": 126430
    },
    {
      "epoch": 0.810946885031411,
      "grad_norm": 0.10924722999334335,
      "learning_rate": 7.416823723736615e-06,
      "loss": 0.0026,
      "step": 126440
    },
    {
      "epoch": 0.8110110219251971,
      "grad_norm": 0.12125886976718903,
      "learning_rate": 7.41633373359555e-06,
      "loss": 0.0015,
      "step": 126450
    },
    {
      "epoch": 0.8110751588189833,
      "grad_norm": 0.028079597279429436,
      "learning_rate": 7.415843713176096e-06,
      "loss": 0.0013,
      "step": 126460
    },
    {
      "epoch": 0.8111392957127693,
      "grad_norm": 0.086025170981884,
      "learning_rate": 7.4153536624843944e-06,
      "loss": 0.0036,
      "step": 126470
    },
    {
      "epoch": 0.8112034326065555,
      "grad_norm": 0.08843878656625748,
      "learning_rate": 7.414863581526586e-06,
      "loss": 0.0017,
      "step": 126480
    },
    {
      "epoch": 0.8112675695003415,
      "grad_norm": 0.059826839715242386,
      "learning_rate": 7.414373470308809e-06,
      "loss": 0.0027,
      "step": 126490
    },
    {
      "epoch": 0.8113317063941277,
      "grad_norm": 0.12478183209896088,
      "learning_rate": 7.413883328837208e-06,
      "loss": 0.0025,
      "step": 126500
    },
    {
      "epoch": 0.8113958432879137,
      "grad_norm": 0.09462691098451614,
      "learning_rate": 7.4133931571179244e-06,
      "loss": 0.0027,
      "step": 126510
    },
    {
      "epoch": 0.8114599801816998,
      "grad_norm": 0.26414474844932556,
      "learning_rate": 7.412902955157097e-06,
      "loss": 0.0028,
      "step": 126520
    },
    {
      "epoch": 0.8115241170754859,
      "grad_norm": 0.1463708132505417,
      "learning_rate": 7.412412722960875e-06,
      "loss": 0.0028,
      "step": 126530
    },
    {
      "epoch": 0.811588253969272,
      "grad_norm": 0.04612165316939354,
      "learning_rate": 7.411922460535394e-06,
      "loss": 0.0017,
      "step": 126540
    },
    {
      "epoch": 0.8116523908630581,
      "grad_norm": 0.3414027690887451,
      "learning_rate": 7.411432167886803e-06,
      "loss": 0.0018,
      "step": 126550
    },
    {
      "epoch": 0.8117165277568442,
      "grad_norm": 0.1312377154827118,
      "learning_rate": 7.4109418450212424e-06,
      "loss": 0.0032,
      "step": 126560
    },
    {
      "epoch": 0.8117806646506303,
      "grad_norm": 0.1252509504556656,
      "learning_rate": 7.410451491944859e-06,
      "loss": 0.0023,
      "step": 126570
    },
    {
      "epoch": 0.8118448015444164,
      "grad_norm": 0.09422841668128967,
      "learning_rate": 7.409961108663794e-06,
      "loss": 0.0011,
      "step": 126580
    },
    {
      "epoch": 0.8119089384382026,
      "grad_norm": 0.17276807129383087,
      "learning_rate": 7.409470695184193e-06,
      "loss": 0.0035,
      "step": 126590
    },
    {
      "epoch": 0.8119730753319886,
      "grad_norm": 0.1141679510474205,
      "learning_rate": 7.4089802515122035e-06,
      "loss": 0.0034,
      "step": 126600
    },
    {
      "epoch": 0.8120372122257747,
      "grad_norm": 0.0753118246793747,
      "learning_rate": 7.4084897776539695e-06,
      "loss": 0.0022,
      "step": 126610
    },
    {
      "epoch": 0.8121013491195608,
      "grad_norm": 0.25150066614151,
      "learning_rate": 7.407999273615638e-06,
      "loss": 0.0039,
      "step": 126620
    },
    {
      "epoch": 0.8121654860133469,
      "grad_norm": 0.14544636011123657,
      "learning_rate": 7.407508739403355e-06,
      "loss": 0.0027,
      "step": 126630
    },
    {
      "epoch": 0.812229622907133,
      "grad_norm": 0.08018797636032104,
      "learning_rate": 7.407018175023265e-06,
      "loss": 0.0027,
      "step": 126640
    },
    {
      "epoch": 0.8122937598009191,
      "grad_norm": 0.2047436684370041,
      "learning_rate": 7.406527580481519e-06,
      "loss": 0.0055,
      "step": 126650
    },
    {
      "epoch": 0.8123578966947052,
      "grad_norm": 0.10463303327560425,
      "learning_rate": 7.406036955784261e-06,
      "loss": 0.0035,
      "step": 126660
    },
    {
      "epoch": 0.8124220335884913,
      "grad_norm": 0.1510915458202362,
      "learning_rate": 7.405546300937641e-06,
      "loss": 0.0033,
      "step": 126670
    },
    {
      "epoch": 0.8124861704822773,
      "grad_norm": 0.10969695448875427,
      "learning_rate": 7.405055615947807e-06,
      "loss": 0.0017,
      "step": 126680
    },
    {
      "epoch": 0.8125503073760635,
      "grad_norm": 0.37942472100257874,
      "learning_rate": 7.4045649008209074e-06,
      "loss": 0.003,
      "step": 126690
    },
    {
      "epoch": 0.8126144442698495,
      "grad_norm": 0.08156800270080566,
      "learning_rate": 7.40407415556309e-06,
      "loss": 0.0035,
      "step": 126700
    },
    {
      "epoch": 0.8126785811636357,
      "grad_norm": 0.2511783242225647,
      "learning_rate": 7.403583380180506e-06,
      "loss": 0.002,
      "step": 126710
    },
    {
      "epoch": 0.8127427180574217,
      "grad_norm": 0.04892609268426895,
      "learning_rate": 7.403092574679303e-06,
      "loss": 0.0032,
      "step": 126720
    },
    {
      "epoch": 0.8128068549512079,
      "grad_norm": 0.15845321118831635,
      "learning_rate": 7.402601739065635e-06,
      "loss": 0.0043,
      "step": 126730
    },
    {
      "epoch": 0.812870991844994,
      "grad_norm": 0.11157383024692535,
      "learning_rate": 7.402110873345649e-06,
      "loss": 0.0026,
      "step": 126740
    },
    {
      "epoch": 0.81293512873878,
      "grad_norm": 0.25845885276794434,
      "learning_rate": 7.401619977525496e-06,
      "loss": 0.0048,
      "step": 126750
    },
    {
      "epoch": 0.8129992656325662,
      "grad_norm": 0.18509311974048615,
      "learning_rate": 7.40112905161133e-06,
      "loss": 0.0018,
      "step": 126760
    },
    {
      "epoch": 0.8130634025263522,
      "grad_norm": 0.21958787739276886,
      "learning_rate": 7.400638095609299e-06,
      "loss": 0.0023,
      "step": 126770
    },
    {
      "epoch": 0.8131275394201384,
      "grad_norm": 0.323677122592926,
      "learning_rate": 7.40014710952556e-06,
      "loss": 0.0032,
      "step": 126780
    },
    {
      "epoch": 0.8131916763139244,
      "grad_norm": 0.19918932020664215,
      "learning_rate": 7.399656093366259e-06,
      "loss": 0.0015,
      "step": 126790
    },
    {
      "epoch": 0.8132558132077106,
      "grad_norm": 0.2212468683719635,
      "learning_rate": 7.399165047137554e-06,
      "loss": 0.0034,
      "step": 126800
    },
    {
      "epoch": 0.8133199501014966,
      "grad_norm": 0.22588036954402924,
      "learning_rate": 7.398673970845596e-06,
      "loss": 0.0019,
      "step": 126810
    },
    {
      "epoch": 0.8133840869952828,
      "grad_norm": 0.13337135314941406,
      "learning_rate": 7.3981828644965394e-06,
      "loss": 0.0018,
      "step": 126820
    },
    {
      "epoch": 0.8134482238890688,
      "grad_norm": 0.06398195773363113,
      "learning_rate": 7.397691728096536e-06,
      "loss": 0.0048,
      "step": 126830
    },
    {
      "epoch": 0.8135123607828549,
      "grad_norm": 0.06610044091939926,
      "learning_rate": 7.397200561651743e-06,
      "loss": 0.0016,
      "step": 126840
    },
    {
      "epoch": 0.813576497676641,
      "grad_norm": 0.05419051647186279,
      "learning_rate": 7.396709365168313e-06,
      "loss": 0.002,
      "step": 126850
    },
    {
      "epoch": 0.8136406345704271,
      "grad_norm": 0.027123944833874702,
      "learning_rate": 7.396218138652401e-06,
      "loss": 0.0043,
      "step": 126860
    },
    {
      "epoch": 0.8137047714642132,
      "grad_norm": 0.06095758080482483,
      "learning_rate": 7.395726882110164e-06,
      "loss": 0.0021,
      "step": 126870
    },
    {
      "epoch": 0.8137689083579993,
      "grad_norm": 0.12250997871160507,
      "learning_rate": 7.395235595547757e-06,
      "loss": 0.0022,
      "step": 126880
    },
    {
      "epoch": 0.8138330452517855,
      "grad_norm": 0.18670979142189026,
      "learning_rate": 7.394744278971337e-06,
      "loss": 0.0027,
      "step": 126890
    },
    {
      "epoch": 0.8138971821455715,
      "grad_norm": 0.11035393923521042,
      "learning_rate": 7.394252932387058e-06,
      "loss": 0.0025,
      "step": 126900
    },
    {
      "epoch": 0.8139613190393576,
      "grad_norm": 0.06221034377813339,
      "learning_rate": 7.393761555801081e-06,
      "loss": 0.0017,
      "step": 126910
    },
    {
      "epoch": 0.8140254559331437,
      "grad_norm": 0.26314976811408997,
      "learning_rate": 7.3932701492195594e-06,
      "loss": 0.0026,
      "step": 126920
    },
    {
      "epoch": 0.8140895928269298,
      "grad_norm": 0.11423099040985107,
      "learning_rate": 7.3927787126486515e-06,
      "loss": 0.0039,
      "step": 126930
    },
    {
      "epoch": 0.8141537297207159,
      "grad_norm": 0.21977105736732483,
      "learning_rate": 7.392287246094517e-06,
      "loss": 0.0033,
      "step": 126940
    },
    {
      "epoch": 0.814217866614502,
      "grad_norm": 0.3997001647949219,
      "learning_rate": 7.391795749563315e-06,
      "loss": 0.0025,
      "step": 126950
    },
    {
      "epoch": 0.814282003508288,
      "grad_norm": 0.06910167634487152,
      "learning_rate": 7.391304223061201e-06,
      "loss": 0.0026,
      "step": 126960
    },
    {
      "epoch": 0.8143461404020742,
      "grad_norm": 0.04946955293416977,
      "learning_rate": 7.3908126665943355e-06,
      "loss": 0.0022,
      "step": 126970
    },
    {
      "epoch": 0.8144102772958602,
      "grad_norm": 0.05174172297120094,
      "learning_rate": 7.390321080168879e-06,
      "loss": 0.0016,
      "step": 126980
    },
    {
      "epoch": 0.8144744141896464,
      "grad_norm": 0.19454753398895264,
      "learning_rate": 7.389829463790993e-06,
      "loss": 0.0032,
      "step": 126990
    },
    {
      "epoch": 0.8145385510834324,
      "grad_norm": 0.03614850714802742,
      "learning_rate": 7.389337817466834e-06,
      "loss": 0.0015,
      "step": 127000
    },
    {
      "epoch": 0.8146026879772186,
      "grad_norm": 0.15189678966999054,
      "learning_rate": 7.388846141202566e-06,
      "loss": 0.002,
      "step": 127010
    },
    {
      "epoch": 0.8146668248710047,
      "grad_norm": 0.180029958486557,
      "learning_rate": 7.388354435004346e-06,
      "loss": 0.0027,
      "step": 127020
    },
    {
      "epoch": 0.8147309617647908,
      "grad_norm": 0.07591405510902405,
      "learning_rate": 7.387862698878341e-06,
      "loss": 0.0014,
      "step": 127030
    },
    {
      "epoch": 0.8147950986585769,
      "grad_norm": 0.24453487992286682,
      "learning_rate": 7.387370932830709e-06,
      "loss": 0.0034,
      "step": 127040
    },
    {
      "epoch": 0.814859235552363,
      "grad_norm": 0.04842061176896095,
      "learning_rate": 7.3868791368676125e-06,
      "loss": 0.0042,
      "step": 127050
    },
    {
      "epoch": 0.8149233724461491,
      "grad_norm": 0.1432727575302124,
      "learning_rate": 7.386387310995213e-06,
      "loss": 0.0025,
      "step": 127060
    },
    {
      "epoch": 0.8149875093399351,
      "grad_norm": 0.005806133151054382,
      "learning_rate": 7.385895455219677e-06,
      "loss": 0.002,
      "step": 127070
    },
    {
      "epoch": 0.8150516462337213,
      "grad_norm": 0.1719507873058319,
      "learning_rate": 7.385403569547166e-06,
      "loss": 0.0032,
      "step": 127080
    },
    {
      "epoch": 0.8151157831275073,
      "grad_norm": 0.018636295571923256,
      "learning_rate": 7.3849116539838415e-06,
      "loss": 0.0074,
      "step": 127090
    },
    {
      "epoch": 0.8151799200212935,
      "grad_norm": 0.09164224565029144,
      "learning_rate": 7.384419708535872e-06,
      "loss": 0.0027,
      "step": 127100
    },
    {
      "epoch": 0.8152440569150795,
      "grad_norm": 0.1219768151640892,
      "learning_rate": 7.383927733209417e-06,
      "loss": 0.0015,
      "step": 127110
    },
    {
      "epoch": 0.8153081938088657,
      "grad_norm": 0.02932930551469326,
      "learning_rate": 7.383435728010646e-06,
      "loss": 0.0019,
      "step": 127120
    },
    {
      "epoch": 0.8153723307026517,
      "grad_norm": 0.056617170572280884,
      "learning_rate": 7.38294369294572e-06,
      "loss": 0.0012,
      "step": 127130
    },
    {
      "epoch": 0.8154364675964378,
      "grad_norm": 0.11760145425796509,
      "learning_rate": 7.382451628020808e-06,
      "loss": 0.0015,
      "step": 127140
    },
    {
      "epoch": 0.8155006044902239,
      "grad_norm": 0.19532443583011627,
      "learning_rate": 7.381959533242074e-06,
      "loss": 0.0028,
      "step": 127150
    },
    {
      "epoch": 0.81556474138401,
      "grad_norm": 0.16045066714286804,
      "learning_rate": 7.381467408615684e-06,
      "loss": 0.0027,
      "step": 127160
    },
    {
      "epoch": 0.8156288782777962,
      "grad_norm": 0.23277214169502258,
      "learning_rate": 7.380975254147805e-06,
      "loss": 0.0036,
      "step": 127170
    },
    {
      "epoch": 0.8156930151715822,
      "grad_norm": 0.16956904530525208,
      "learning_rate": 7.380483069844606e-06,
      "loss": 0.003,
      "step": 127180
    },
    {
      "epoch": 0.8157571520653684,
      "grad_norm": 0.0400191992521286,
      "learning_rate": 7.379990855712251e-06,
      "loss": 0.0015,
      "step": 127190
    },
    {
      "epoch": 0.8158212889591544,
      "grad_norm": 0.2659646272659302,
      "learning_rate": 7.3794986117569116e-06,
      "loss": 0.0022,
      "step": 127200
    },
    {
      "epoch": 0.8158854258529405,
      "grad_norm": 0.05728033557534218,
      "learning_rate": 7.3790063379847525e-06,
      "loss": 0.0048,
      "step": 127210
    },
    {
      "epoch": 0.8159495627467266,
      "grad_norm": 0.09088273346424103,
      "learning_rate": 7.378514034401944e-06,
      "loss": 0.003,
      "step": 127220
    },
    {
      "epoch": 0.8160136996405127,
      "grad_norm": 0.1122676208615303,
      "learning_rate": 7.378021701014655e-06,
      "loss": 0.0075,
      "step": 127230
    },
    {
      "epoch": 0.8160778365342988,
      "grad_norm": 0.09521377086639404,
      "learning_rate": 7.377529337829054e-06,
      "loss": 0.0013,
      "step": 127240
    },
    {
      "epoch": 0.8161419734280849,
      "grad_norm": 0.0657171756029129,
      "learning_rate": 7.377036944851312e-06,
      "loss": 0.0024,
      "step": 127250
    },
    {
      "epoch": 0.816206110321871,
      "grad_norm": 0.08039627969264984,
      "learning_rate": 7.3765445220875976e-06,
      "loss": 0.0019,
      "step": 127260
    },
    {
      "epoch": 0.8162702472156571,
      "grad_norm": 0.05933055281639099,
      "learning_rate": 7.376052069544083e-06,
      "loss": 0.0037,
      "step": 127270
    },
    {
      "epoch": 0.8163343841094431,
      "grad_norm": 0.011957366019487381,
      "learning_rate": 7.375559587226936e-06,
      "loss": 0.0028,
      "step": 127280
    },
    {
      "epoch": 0.8163985210032293,
      "grad_norm": 0.06866823136806488,
      "learning_rate": 7.375067075142331e-06,
      "loss": 0.0025,
      "step": 127290
    },
    {
      "epoch": 0.8164626578970154,
      "grad_norm": 0.03869573399424553,
      "learning_rate": 7.374574533296437e-06,
      "loss": 0.0022,
      "step": 127300
    },
    {
      "epoch": 0.8165267947908015,
      "grad_norm": 0.10332153737545013,
      "learning_rate": 7.374081961695429e-06,
      "loss": 0.0034,
      "step": 127310
    },
    {
      "epoch": 0.8165909316845876,
      "grad_norm": 0.05911581963300705,
      "learning_rate": 7.3735893603454755e-06,
      "loss": 0.0024,
      "step": 127320
    },
    {
      "epoch": 0.8166550685783737,
      "grad_norm": 0.056003425270318985,
      "learning_rate": 7.373096729252751e-06,
      "loss": 0.0061,
      "step": 127330
    },
    {
      "epoch": 0.8167192054721598,
      "grad_norm": 0.07668650895357132,
      "learning_rate": 7.372604068423429e-06,
      "loss": 0.0019,
      "step": 127340
    },
    {
      "epoch": 0.8167833423659459,
      "grad_norm": 0.11102576553821564,
      "learning_rate": 7.372111377863682e-06,
      "loss": 0.0033,
      "step": 127350
    },
    {
      "epoch": 0.816847479259732,
      "grad_norm": 0.009732525795698166,
      "learning_rate": 7.371618657579683e-06,
      "loss": 0.0036,
      "step": 127360
    },
    {
      "epoch": 0.816911616153518,
      "grad_norm": 0.05169150233268738,
      "learning_rate": 7.37112590757761e-06,
      "loss": 0.0027,
      "step": 127370
    },
    {
      "epoch": 0.8169757530473042,
      "grad_norm": 0.14333856105804443,
      "learning_rate": 7.370633127863634e-06,
      "loss": 0.0022,
      "step": 127380
    },
    {
      "epoch": 0.8170398899410902,
      "grad_norm": 0.09235555678606033,
      "learning_rate": 7.370140318443929e-06,
      "loss": 0.004,
      "step": 127390
    },
    {
      "epoch": 0.8171040268348764,
      "grad_norm": 0.1903359591960907,
      "learning_rate": 7.369647479324673e-06,
      "loss": 0.0025,
      "step": 127400
    },
    {
      "epoch": 0.8171681637286624,
      "grad_norm": 0.1300700604915619,
      "learning_rate": 7.369154610512041e-06,
      "loss": 0.0028,
      "step": 127410
    },
    {
      "epoch": 0.8172323006224486,
      "grad_norm": 0.3448699712753296,
      "learning_rate": 7.368661712012207e-06,
      "loss": 0.0028,
      "step": 127420
    },
    {
      "epoch": 0.8172964375162346,
      "grad_norm": 0.5382606387138367,
      "learning_rate": 7.36816878383135e-06,
      "loss": 0.0023,
      "step": 127430
    },
    {
      "epoch": 0.8173605744100207,
      "grad_norm": 0.02058287337422371,
      "learning_rate": 7.367675825975645e-06,
      "loss": 0.0021,
      "step": 127440
    },
    {
      "epoch": 0.8174247113038069,
      "grad_norm": 0.047119539231061935,
      "learning_rate": 7.36718283845127e-06,
      "loss": 0.0045,
      "step": 127450
    },
    {
      "epoch": 0.8174888481975929,
      "grad_norm": 0.12040512263774872,
      "learning_rate": 7.366689821264403e-06,
      "loss": 0.0016,
      "step": 127460
    },
    {
      "epoch": 0.8175529850913791,
      "grad_norm": 0.0925847664475441,
      "learning_rate": 7.36619677442122e-06,
      "loss": 0.0033,
      "step": 127470
    },
    {
      "epoch": 0.8176171219851651,
      "grad_norm": 0.06187743693590164,
      "learning_rate": 7.3657036979279015e-06,
      "loss": 0.0023,
      "step": 127480
    },
    {
      "epoch": 0.8176812588789513,
      "grad_norm": 0.2978494167327881,
      "learning_rate": 7.365210591790623e-06,
      "loss": 0.0039,
      "step": 127490
    },
    {
      "epoch": 0.8177453957727373,
      "grad_norm": 0.04251876845955849,
      "learning_rate": 7.364717456015569e-06,
      "loss": 0.0023,
      "step": 127500
    },
    {
      "epoch": 0.8178095326665235,
      "grad_norm": 0.06055247038602829,
      "learning_rate": 7.364224290608911e-06,
      "loss": 0.0026,
      "step": 127510
    },
    {
      "epoch": 0.8178736695603095,
      "grad_norm": 0.0591123029589653,
      "learning_rate": 7.363731095576836e-06,
      "loss": 0.0022,
      "step": 127520
    },
    {
      "epoch": 0.8179378064540956,
      "grad_norm": 0.29559198021888733,
      "learning_rate": 7.3632378709255195e-06,
      "loss": 0.0019,
      "step": 127530
    },
    {
      "epoch": 0.8180019433478817,
      "grad_norm": 0.12792937457561493,
      "learning_rate": 7.362744616661144e-06,
      "loss": 0.0013,
      "step": 127540
    },
    {
      "epoch": 0.8180660802416678,
      "grad_norm": 0.24491946399211884,
      "learning_rate": 7.3622513327898894e-06,
      "loss": 0.0019,
      "step": 127550
    },
    {
      "epoch": 0.8181302171354539,
      "grad_norm": 0.2049376666545868,
      "learning_rate": 7.361758019317939e-06,
      "loss": 0.0023,
      "step": 127560
    },
    {
      "epoch": 0.81819435402924,
      "grad_norm": 0.07258431613445282,
      "learning_rate": 7.361264676251471e-06,
      "loss": 0.0041,
      "step": 127570
    },
    {
      "epoch": 0.8182584909230262,
      "grad_norm": 0.10781293362379074,
      "learning_rate": 7.3607713035966675e-06,
      "loss": 0.0028,
      "step": 127580
    },
    {
      "epoch": 0.8183226278168122,
      "grad_norm": 0.05559421330690384,
      "learning_rate": 7.360277901359716e-06,
      "loss": 0.0024,
      "step": 127590
    },
    {
      "epoch": 0.8183867647105983,
      "grad_norm": 0.1653282791376114,
      "learning_rate": 7.3597844695467925e-06,
      "loss": 0.0025,
      "step": 127600
    },
    {
      "epoch": 0.8184509016043844,
      "grad_norm": 0.05223669484257698,
      "learning_rate": 7.359291008164084e-06,
      "loss": 0.0026,
      "step": 127610
    },
    {
      "epoch": 0.8185150384981705,
      "grad_norm": 0.027781745418906212,
      "learning_rate": 7.358797517217774e-06,
      "loss": 0.0025,
      "step": 127620
    },
    {
      "epoch": 0.8185791753919566,
      "grad_norm": 0.07809358090162277,
      "learning_rate": 7.3583039967140446e-06,
      "loss": 0.0025,
      "step": 127630
    },
    {
      "epoch": 0.8186433122857427,
      "grad_norm": 0.22004097700119019,
      "learning_rate": 7.357810446659081e-06,
      "loss": 0.0034,
      "step": 127640
    },
    {
      "epoch": 0.8187074491795288,
      "grad_norm": 0.3288693428039551,
      "learning_rate": 7.357316867059068e-06,
      "loss": 0.0034,
      "step": 127650
    },
    {
      "epoch": 0.8187715860733149,
      "grad_norm": 0.13195857405662537,
      "learning_rate": 7.356823257920188e-06,
      "loss": 0.0017,
      "step": 127660
    },
    {
      "epoch": 0.8188357229671009,
      "grad_norm": 0.1732707917690277,
      "learning_rate": 7.35632961924863e-06,
      "loss": 0.0034,
      "step": 127670
    },
    {
      "epoch": 0.8188998598608871,
      "grad_norm": 0.057275883853435516,
      "learning_rate": 7.355835951050576e-06,
      "loss": 0.0011,
      "step": 127680
    },
    {
      "epoch": 0.8189639967546731,
      "grad_norm": 0.05436362698674202,
      "learning_rate": 7.355342253332216e-06,
      "loss": 0.0037,
      "step": 127690
    },
    {
      "epoch": 0.8190281336484593,
      "grad_norm": 0.17967228591442108,
      "learning_rate": 7.3548485260997325e-06,
      "loss": 0.0046,
      "step": 127700
    },
    {
      "epoch": 0.8190922705422453,
      "grad_norm": 0.06257455050945282,
      "learning_rate": 7.354354769359314e-06,
      "loss": 0.0019,
      "step": 127710
    },
    {
      "epoch": 0.8191564074360315,
      "grad_norm": 0.13455961644649506,
      "learning_rate": 7.353860983117149e-06,
      "loss": 0.0032,
      "step": 127720
    },
    {
      "epoch": 0.8192205443298176,
      "grad_norm": 0.0575551763176918,
      "learning_rate": 7.353367167379421e-06,
      "loss": 0.0017,
      "step": 127730
    },
    {
      "epoch": 0.8192846812236037,
      "grad_norm": 0.08193744719028473,
      "learning_rate": 7.352873322152323e-06,
      "loss": 0.0021,
      "step": 127740
    },
    {
      "epoch": 0.8193488181173898,
      "grad_norm": 0.13074032962322235,
      "learning_rate": 7.352379447442041e-06,
      "loss": 0.0022,
      "step": 127750
    },
    {
      "epoch": 0.8194129550111758,
      "grad_norm": 0.05399656295776367,
      "learning_rate": 7.351885543254761e-06,
      "loss": 0.0032,
      "step": 127760
    },
    {
      "epoch": 0.819477091904962,
      "grad_norm": 0.13823533058166504,
      "learning_rate": 7.351391609596675e-06,
      "loss": 0.0027,
      "step": 127770
    },
    {
      "epoch": 0.819541228798748,
      "grad_norm": 0.07404447346925735,
      "learning_rate": 7.350897646473974e-06,
      "loss": 0.0033,
      "step": 127780
    },
    {
      "epoch": 0.8196053656925342,
      "grad_norm": 0.012039989233016968,
      "learning_rate": 7.350403653892842e-06,
      "loss": 0.0025,
      "step": 127790
    },
    {
      "epoch": 0.8196695025863202,
      "grad_norm": 0.09605922549962997,
      "learning_rate": 7.349909631859475e-06,
      "loss": 0.0018,
      "step": 127800
    },
    {
      "epoch": 0.8197336394801064,
      "grad_norm": 0.34952735900878906,
      "learning_rate": 7.349415580380059e-06,
      "loss": 0.0022,
      "step": 127810
    },
    {
      "epoch": 0.8197977763738924,
      "grad_norm": 0.20634576678276062,
      "learning_rate": 7.348921499460788e-06,
      "loss": 0.0027,
      "step": 127820
    },
    {
      "epoch": 0.8198619132676785,
      "grad_norm": 0.022601787000894547,
      "learning_rate": 7.348427389107851e-06,
      "loss": 0.0028,
      "step": 127830
    },
    {
      "epoch": 0.8199260501614646,
      "grad_norm": 0.09635855257511139,
      "learning_rate": 7.347933249327441e-06,
      "loss": 0.0029,
      "step": 127840
    },
    {
      "epoch": 0.8199901870552507,
      "grad_norm": 0.1902410238981247,
      "learning_rate": 7.347439080125749e-06,
      "loss": 0.003,
      "step": 127850
    },
    {
      "epoch": 0.8200543239490369,
      "grad_norm": 0.19818753004074097,
      "learning_rate": 7.346944881508967e-06,
      "loss": 0.0023,
      "step": 127860
    },
    {
      "epoch": 0.8201184608428229,
      "grad_norm": 0.10266011953353882,
      "learning_rate": 7.346450653483289e-06,
      "loss": 0.0013,
      "step": 127870
    },
    {
      "epoch": 0.8201825977366091,
      "grad_norm": 0.18944767117500305,
      "learning_rate": 7.3459563960549075e-06,
      "loss": 0.0024,
      "step": 127880
    },
    {
      "epoch": 0.8202467346303951,
      "grad_norm": 0.036455295979976654,
      "learning_rate": 7.345462109230015e-06,
      "loss": 0.0019,
      "step": 127890
    },
    {
      "epoch": 0.8203108715241813,
      "grad_norm": 0.1035575121641159,
      "learning_rate": 7.344967793014808e-06,
      "loss": 0.0015,
      "step": 127900
    },
    {
      "epoch": 0.8203750084179673,
      "grad_norm": 0.03167593851685524,
      "learning_rate": 7.344473447415477e-06,
      "loss": 0.002,
      "step": 127910
    },
    {
      "epoch": 0.8204391453117534,
      "grad_norm": 0.05191560462117195,
      "learning_rate": 7.3439790724382165e-06,
      "loss": 0.0043,
      "step": 127920
    },
    {
      "epoch": 0.8205032822055395,
      "grad_norm": 0.49309152364730835,
      "learning_rate": 7.3434846680892255e-06,
      "loss": 0.0038,
      "step": 127930
    },
    {
      "epoch": 0.8205674190993256,
      "grad_norm": 0.17252881824970245,
      "learning_rate": 7.342990234374696e-06,
      "loss": 0.0018,
      "step": 127940
    },
    {
      "epoch": 0.8206315559931117,
      "grad_norm": 0.07217154651880264,
      "learning_rate": 7.342495771300825e-06,
      "loss": 0.0028,
      "step": 127950
    },
    {
      "epoch": 0.8206956928868978,
      "grad_norm": 0.03394459933042526,
      "learning_rate": 7.342001278873807e-06,
      "loss": 0.0026,
      "step": 127960
    },
    {
      "epoch": 0.8207598297806838,
      "grad_norm": 0.09284265339374542,
      "learning_rate": 7.3415067570998395e-06,
      "loss": 0.0022,
      "step": 127970
    },
    {
      "epoch": 0.82082396667447,
      "grad_norm": 0.10128198564052582,
      "learning_rate": 7.341012205985119e-06,
      "loss": 0.0034,
      "step": 127980
    },
    {
      "epoch": 0.820888103568256,
      "grad_norm": 0.10883468389511108,
      "learning_rate": 7.3405176255358425e-06,
      "loss": 0.0023,
      "step": 127990
    },
    {
      "epoch": 0.8209522404620422,
      "grad_norm": 0.07447699457406998,
      "learning_rate": 7.3400230157582065e-06,
      "loss": 0.0017,
      "step": 128000
    },
    {
      "epoch": 0.8210163773558283,
      "grad_norm": 0.05366537719964981,
      "learning_rate": 7.339528376658412e-06,
      "loss": 0.0015,
      "step": 128010
    },
    {
      "epoch": 0.8210805142496144,
      "grad_norm": 0.06155402213335037,
      "learning_rate": 7.339033708242652e-06,
      "loss": 0.0029,
      "step": 128020
    },
    {
      "epoch": 0.8211446511434005,
      "grad_norm": 0.15945905447006226,
      "learning_rate": 7.33853901051713e-06,
      "loss": 0.0021,
      "step": 128030
    },
    {
      "epoch": 0.8212087880371866,
      "grad_norm": 0.06520810723304749,
      "learning_rate": 7.3380442834880414e-06,
      "loss": 0.0041,
      "step": 128040
    },
    {
      "epoch": 0.8212729249309727,
      "grad_norm": 0.2926897704601288,
      "learning_rate": 7.337549527161588e-06,
      "loss": 0.0019,
      "step": 128050
    },
    {
      "epoch": 0.8213370618247587,
      "grad_norm": 0.04401979595422745,
      "learning_rate": 7.337054741543968e-06,
      "loss": 0.0014,
      "step": 128060
    },
    {
      "epoch": 0.8214011987185449,
      "grad_norm": 0.07479792088270187,
      "learning_rate": 7.336559926641381e-06,
      "loss": 0.0031,
      "step": 128070
    },
    {
      "epoch": 0.8214653356123309,
      "grad_norm": 0.13457804918289185,
      "learning_rate": 7.33606508246003e-06,
      "loss": 0.0026,
      "step": 128080
    },
    {
      "epoch": 0.8215294725061171,
      "grad_norm": 0.08685542643070221,
      "learning_rate": 7.335570209006112e-06,
      "loss": 0.0029,
      "step": 128090
    },
    {
      "epoch": 0.8215936093999031,
      "grad_norm": 0.09798700362443924,
      "learning_rate": 7.335075306285832e-06,
      "loss": 0.002,
      "step": 128100
    },
    {
      "epoch": 0.8216577462936893,
      "grad_norm": 0.19563564658164978,
      "learning_rate": 7.334580374305389e-06,
      "loss": 0.0031,
      "step": 128110
    },
    {
      "epoch": 0.8217218831874753,
      "grad_norm": 0.1523904949426651,
      "learning_rate": 7.334085413070984e-06,
      "loss": 0.0023,
      "step": 128120
    },
    {
      "epoch": 0.8217860200812614,
      "grad_norm": 0.07425139844417572,
      "learning_rate": 7.333590422588821e-06,
      "loss": 0.002,
      "step": 128130
    },
    {
      "epoch": 0.8218501569750476,
      "grad_norm": 0.06021131947636604,
      "learning_rate": 7.333095402865104e-06,
      "loss": 0.0014,
      "step": 128140
    },
    {
      "epoch": 0.8219142938688336,
      "grad_norm": 0.21700631082057953,
      "learning_rate": 7.332600353906032e-06,
      "loss": 0.0022,
      "step": 128150
    },
    {
      "epoch": 0.8219784307626198,
      "grad_norm": 0.049357131123542786,
      "learning_rate": 7.332105275717812e-06,
      "loss": 0.0019,
      "step": 128160
    },
    {
      "epoch": 0.8220425676564058,
      "grad_norm": 0.320541113615036,
      "learning_rate": 7.331610168306645e-06,
      "loss": 0.0035,
      "step": 128170
    },
    {
      "epoch": 0.822106704550192,
      "grad_norm": 0.23093421757221222,
      "learning_rate": 7.3311150316787385e-06,
      "loss": 0.0028,
      "step": 128180
    },
    {
      "epoch": 0.822170841443978,
      "grad_norm": 0.12479311972856522,
      "learning_rate": 7.330619865840292e-06,
      "loss": 0.003,
      "step": 128190
    },
    {
      "epoch": 0.8222349783377642,
      "grad_norm": 0.13449817895889282,
      "learning_rate": 7.330124670797515e-06,
      "loss": 0.0023,
      "step": 128200
    },
    {
      "epoch": 0.8222991152315502,
      "grad_norm": 0.15000073611736298,
      "learning_rate": 7.329629446556609e-06,
      "loss": 0.0034,
      "step": 128210
    },
    {
      "epoch": 0.8223632521253363,
      "grad_norm": 0.04748809337615967,
      "learning_rate": 7.3291341931237825e-06,
      "loss": 0.0083,
      "step": 128220
    },
    {
      "epoch": 0.8224273890191224,
      "grad_norm": 0.13411863148212433,
      "learning_rate": 7.328638910505238e-06,
      "loss": 0.0025,
      "step": 128230
    },
    {
      "epoch": 0.8224915259129085,
      "grad_norm": 0.03303762525320053,
      "learning_rate": 7.3281435987071854e-06,
      "loss": 0.0016,
      "step": 128240
    },
    {
      "epoch": 0.8225556628066946,
      "grad_norm": 0.3325529396533966,
      "learning_rate": 7.3276482577358285e-06,
      "loss": 0.0024,
      "step": 128250
    },
    {
      "epoch": 0.8226197997004807,
      "grad_norm": 0.02374950423836708,
      "learning_rate": 7.327152887597377e-06,
      "loss": 0.0027,
      "step": 128260
    },
    {
      "epoch": 0.8226839365942668,
      "grad_norm": 0.04302360117435455,
      "learning_rate": 7.326657488298035e-06,
      "loss": 0.0013,
      "step": 128270
    },
    {
      "epoch": 0.8227480734880529,
      "grad_norm": 0.0613829642534256,
      "learning_rate": 7.326162059844012e-06,
      "loss": 0.0032,
      "step": 128280
    },
    {
      "epoch": 0.822812210381839,
      "grad_norm": 0.04095643386244774,
      "learning_rate": 7.325666602241517e-06,
      "loss": 0.0041,
      "step": 128290
    },
    {
      "epoch": 0.8228763472756251,
      "grad_norm": 0.0421566404402256,
      "learning_rate": 7.325171115496755e-06,
      "loss": 0.0014,
      "step": 128300
    },
    {
      "epoch": 0.8229404841694112,
      "grad_norm": 0.03124876134097576,
      "learning_rate": 7.32467559961594e-06,
      "loss": 0.0026,
      "step": 128310
    },
    {
      "epoch": 0.8230046210631973,
      "grad_norm": 0.1574496179819107,
      "learning_rate": 7.324180054605277e-06,
      "loss": 0.0029,
      "step": 128320
    },
    {
      "epoch": 0.8230687579569834,
      "grad_norm": 0.19296900928020477,
      "learning_rate": 7.323684480470977e-06,
      "loss": 0.002,
      "step": 128330
    },
    {
      "epoch": 0.8231328948507695,
      "grad_norm": 0.10460526496171951,
      "learning_rate": 7.323188877219249e-06,
      "loss": 0.0017,
      "step": 128340
    },
    {
      "epoch": 0.8231970317445556,
      "grad_norm": 0.0408439114689827,
      "learning_rate": 7.322693244856305e-06,
      "loss": 0.0021,
      "step": 128350
    },
    {
      "epoch": 0.8232611686383416,
      "grad_norm": 0.2912435531616211,
      "learning_rate": 7.322197583388354e-06,
      "loss": 0.002,
      "step": 128360
    },
    {
      "epoch": 0.8233253055321278,
      "grad_norm": 0.04117957875132561,
      "learning_rate": 7.321701892821609e-06,
      "loss": 0.0025,
      "step": 128370
    },
    {
      "epoch": 0.8233894424259138,
      "grad_norm": 0.036341726779937744,
      "learning_rate": 7.3212061731622785e-06,
      "loss": 0.0017,
      "step": 128380
    },
    {
      "epoch": 0.8234535793197,
      "grad_norm": 0.19118447601795197,
      "learning_rate": 7.3207104244165765e-06,
      "loss": 0.0018,
      "step": 128390
    },
    {
      "epoch": 0.823517716213486,
      "grad_norm": 0.19198226928710938,
      "learning_rate": 7.320214646590713e-06,
      "loss": 0.0019,
      "step": 128400
    },
    {
      "epoch": 0.8235818531072722,
      "grad_norm": 0.16182208061218262,
      "learning_rate": 7.319718839690903e-06,
      "loss": 0.0013,
      "step": 128410
    },
    {
      "epoch": 0.8236459900010582,
      "grad_norm": 0.06703642755746841,
      "learning_rate": 7.319223003723358e-06,
      "loss": 0.0022,
      "step": 128420
    },
    {
      "epoch": 0.8237101268948444,
      "grad_norm": 0.15633465349674225,
      "learning_rate": 7.318727138694291e-06,
      "loss": 0.0028,
      "step": 128430
    },
    {
      "epoch": 0.8237742637886305,
      "grad_norm": 0.18169882893562317,
      "learning_rate": 7.318231244609916e-06,
      "loss": 0.0022,
      "step": 128440
    },
    {
      "epoch": 0.8238384006824165,
      "grad_norm": 0.060424067080020905,
      "learning_rate": 7.317735321476446e-06,
      "loss": 0.0025,
      "step": 128450
    },
    {
      "epoch": 0.8239025375762027,
      "grad_norm": 0.11214757710695267,
      "learning_rate": 7.317239369300096e-06,
      "loss": 0.002,
      "step": 128460
    },
    {
      "epoch": 0.8239666744699887,
      "grad_norm": 0.19199307262897491,
      "learning_rate": 7.316743388087081e-06,
      "loss": 0.0018,
      "step": 128470
    },
    {
      "epoch": 0.8240308113637749,
      "grad_norm": 0.22552119195461273,
      "learning_rate": 7.316247377843616e-06,
      "loss": 0.0021,
      "step": 128480
    },
    {
      "epoch": 0.8240949482575609,
      "grad_norm": 0.1364929974079132,
      "learning_rate": 7.315751338575916e-06,
      "loss": 0.0027,
      "step": 128490
    },
    {
      "epoch": 0.8241590851513471,
      "grad_norm": 0.26622894406318665,
      "learning_rate": 7.315255270290196e-06,
      "loss": 0.0045,
      "step": 128500
    },
    {
      "epoch": 0.8242232220451331,
      "grad_norm": 0.09583581984043121,
      "learning_rate": 7.314759172992671e-06,
      "loss": 0.0055,
      "step": 128510
    },
    {
      "epoch": 0.8242873589389192,
      "grad_norm": 0.1296004056930542,
      "learning_rate": 7.314263046689562e-06,
      "loss": 0.002,
      "step": 128520
    },
    {
      "epoch": 0.8243514958327053,
      "grad_norm": 0.06543765962123871,
      "learning_rate": 7.313766891387082e-06,
      "loss": 0.0024,
      "step": 128530
    },
    {
      "epoch": 0.8244156327264914,
      "grad_norm": 0.07670734822750092,
      "learning_rate": 7.313270707091451e-06,
      "loss": 0.0026,
      "step": 128540
    },
    {
      "epoch": 0.8244797696202775,
      "grad_norm": 0.17300914227962494,
      "learning_rate": 7.312774493808883e-06,
      "loss": 0.0029,
      "step": 128550
    },
    {
      "epoch": 0.8245439065140636,
      "grad_norm": 0.19316570460796356,
      "learning_rate": 7.312278251545598e-06,
      "loss": 0.0021,
      "step": 128560
    },
    {
      "epoch": 0.8246080434078498,
      "grad_norm": 0.031232358887791634,
      "learning_rate": 7.311781980307813e-06,
      "loss": 0.0038,
      "step": 128570
    },
    {
      "epoch": 0.8246721803016358,
      "grad_norm": 0.04309896379709244,
      "learning_rate": 7.311285680101749e-06,
      "loss": 0.0029,
      "step": 128580
    },
    {
      "epoch": 0.824736317195422,
      "grad_norm": 0.4938768446445465,
      "learning_rate": 7.310789350933623e-06,
      "loss": 0.0044,
      "step": 128590
    },
    {
      "epoch": 0.824800454089208,
      "grad_norm": 0.06435202807188034,
      "learning_rate": 7.310292992809655e-06,
      "loss": 0.0021,
      "step": 128600
    },
    {
      "epoch": 0.8248645909829941,
      "grad_norm": 0.21440468728542328,
      "learning_rate": 7.309796605736065e-06,
      "loss": 0.0028,
      "step": 128610
    },
    {
      "epoch": 0.8249287278767802,
      "grad_norm": 0.10133481025695801,
      "learning_rate": 7.309300189719072e-06,
      "loss": 0.0018,
      "step": 128620
    },
    {
      "epoch": 0.8249928647705663,
      "grad_norm": 0.04468727484345436,
      "learning_rate": 7.308803744764898e-06,
      "loss": 0.004,
      "step": 128630
    },
    {
      "epoch": 0.8250570016643524,
      "grad_norm": 0.06408118456602097,
      "learning_rate": 7.308307270879762e-06,
      "loss": 0.0015,
      "step": 128640
    },
    {
      "epoch": 0.8251211385581385,
      "grad_norm": 0.1306520700454712,
      "learning_rate": 7.307810768069888e-06,
      "loss": 0.003,
      "step": 128650
    },
    {
      "epoch": 0.8251852754519245,
      "grad_norm": 0.012615872547030449,
      "learning_rate": 7.307314236341494e-06,
      "loss": 0.0019,
      "step": 128660
    },
    {
      "epoch": 0.8252494123457107,
      "grad_norm": 0.046122532337903976,
      "learning_rate": 7.306817675700804e-06,
      "loss": 0.0015,
      "step": 128670
    },
    {
      "epoch": 0.8253135492394967,
      "grad_norm": 0.07508425414562225,
      "learning_rate": 7.30632108615404e-06,
      "loss": 0.0024,
      "step": 128680
    },
    {
      "epoch": 0.8253776861332829,
      "grad_norm": 0.26308661699295044,
      "learning_rate": 7.3058244677074245e-06,
      "loss": 0.0016,
      "step": 128690
    },
    {
      "epoch": 0.8254418230270689,
      "grad_norm": 0.15130212903022766,
      "learning_rate": 7.305327820367179e-06,
      "loss": 0.0014,
      "step": 128700
    },
    {
      "epoch": 0.8255059599208551,
      "grad_norm": 0.06789848208427429,
      "learning_rate": 7.304831144139531e-06,
      "loss": 0.0019,
      "step": 128710
    },
    {
      "epoch": 0.8255700968146412,
      "grad_norm": 0.009132793173193932,
      "learning_rate": 7.3043344390307005e-06,
      "loss": 0.0019,
      "step": 128720
    },
    {
      "epoch": 0.8256342337084273,
      "grad_norm": 0.0689903199672699,
      "learning_rate": 7.303837705046912e-06,
      "loss": 0.002,
      "step": 128730
    },
    {
      "epoch": 0.8256983706022134,
      "grad_norm": 0.028618192300200462,
      "learning_rate": 7.303340942194391e-06,
      "loss": 0.0038,
      "step": 128740
    },
    {
      "epoch": 0.8257625074959994,
      "grad_norm": 0.2864423990249634,
      "learning_rate": 7.302844150479362e-06,
      "loss": 0.0044,
      "step": 128750
    },
    {
      "epoch": 0.8258266443897856,
      "grad_norm": 0.165174663066864,
      "learning_rate": 7.302347329908049e-06,
      "loss": 0.0038,
      "step": 128760
    },
    {
      "epoch": 0.8258907812835716,
      "grad_norm": 0.09707002341747284,
      "learning_rate": 7.301850480486678e-06,
      "loss": 0.0018,
      "step": 128770
    },
    {
      "epoch": 0.8259549181773578,
      "grad_norm": 0.16221855580806732,
      "learning_rate": 7.301353602221478e-06,
      "loss": 0.0036,
      "step": 128780
    },
    {
      "epoch": 0.8260190550711438,
      "grad_norm": 0.21278564631938934,
      "learning_rate": 7.300856695118671e-06,
      "loss": 0.0017,
      "step": 128790
    },
    {
      "epoch": 0.82608319196493,
      "grad_norm": 0.0670095831155777,
      "learning_rate": 7.300359759184485e-06,
      "loss": 0.0033,
      "step": 128800
    },
    {
      "epoch": 0.826147328858716,
      "grad_norm": 0.07428467273712158,
      "learning_rate": 7.2998627944251476e-06,
      "loss": 0.0034,
      "step": 128810
    },
    {
      "epoch": 0.8262114657525021,
      "grad_norm": 0.2314709573984146,
      "learning_rate": 7.299365800846887e-06,
      "loss": 0.0022,
      "step": 128820
    },
    {
      "epoch": 0.8262756026462882,
      "grad_norm": 0.07954928278923035,
      "learning_rate": 7.298868778455928e-06,
      "loss": 0.0026,
      "step": 128830
    },
    {
      "epoch": 0.8263397395400743,
      "grad_norm": 0.06688231974840164,
      "learning_rate": 7.2983717272585e-06,
      "loss": 0.0015,
      "step": 128840
    },
    {
      "epoch": 0.8264038764338605,
      "grad_norm": 0.1042524129152298,
      "learning_rate": 7.297874647260832e-06,
      "loss": 0.003,
      "step": 128850
    },
    {
      "epoch": 0.8264680133276465,
      "grad_norm": 0.0997534990310669,
      "learning_rate": 7.297377538469152e-06,
      "loss": 0.0052,
      "step": 128860
    },
    {
      "epoch": 0.8265321502214327,
      "grad_norm": 0.14298345148563385,
      "learning_rate": 7.296880400889691e-06,
      "loss": 0.0027,
      "step": 128870
    },
    {
      "epoch": 0.8265962871152187,
      "grad_norm": 0.05371713638305664,
      "learning_rate": 7.2963832345286766e-06,
      "loss": 0.0058,
      "step": 128880
    },
    {
      "epoch": 0.8266604240090049,
      "grad_norm": 0.14610141515731812,
      "learning_rate": 7.295886039392337e-06,
      "loss": 0.0021,
      "step": 128890
    },
    {
      "epoch": 0.8267245609027909,
      "grad_norm": 0.19027216732501984,
      "learning_rate": 7.295388815486905e-06,
      "loss": 0.0029,
      "step": 128900
    },
    {
      "epoch": 0.826788697796577,
      "grad_norm": 0.24345919489860535,
      "learning_rate": 7.294891562818612e-06,
      "loss": 0.004,
      "step": 128910
    },
    {
      "epoch": 0.8268528346903631,
      "grad_norm": 0.19334959983825684,
      "learning_rate": 7.294394281393689e-06,
      "loss": 0.0031,
      "step": 128920
    },
    {
      "epoch": 0.8269169715841492,
      "grad_norm": 0.18153344094753265,
      "learning_rate": 7.293896971218362e-06,
      "loss": 0.0029,
      "step": 128930
    },
    {
      "epoch": 0.8269811084779353,
      "grad_norm": 0.14814141392707825,
      "learning_rate": 7.2933996322988695e-06,
      "loss": 0.0025,
      "step": 128940
    },
    {
      "epoch": 0.8270452453717214,
      "grad_norm": 1.0534054040908813,
      "learning_rate": 7.292902264641439e-06,
      "loss": 0.0065,
      "step": 128950
    },
    {
      "epoch": 0.8271093822655075,
      "grad_norm": 0.11299543082714081,
      "learning_rate": 7.292404868252307e-06,
      "loss": 0.0028,
      "step": 128960
    },
    {
      "epoch": 0.8271735191592936,
      "grad_norm": 0.513168215751648,
      "learning_rate": 7.2919074431377e-06,
      "loss": 0.0018,
      "step": 128970
    },
    {
      "epoch": 0.8272376560530796,
      "grad_norm": 0.08805724233388901,
      "learning_rate": 7.291409989303857e-06,
      "loss": 0.0021,
      "step": 128980
    },
    {
      "epoch": 0.8273017929468658,
      "grad_norm": 0.028845224529504776,
      "learning_rate": 7.290912506757009e-06,
      "loss": 0.0021,
      "step": 128990
    },
    {
      "epoch": 0.8273659298406519,
      "grad_norm": 0.13066190481185913,
      "learning_rate": 7.29041499550339e-06,
      "loss": 0.0024,
      "step": 129000
    },
    {
      "epoch": 0.827430066734438,
      "grad_norm": 0.08103853464126587,
      "learning_rate": 7.289917455549234e-06,
      "loss": 0.002,
      "step": 129010
    },
    {
      "epoch": 0.8274942036282241,
      "grad_norm": 0.07092414796352386,
      "learning_rate": 7.289419886900775e-06,
      "loss": 0.0026,
      "step": 129020
    },
    {
      "epoch": 0.8275583405220102,
      "grad_norm": 0.0720643550157547,
      "learning_rate": 7.288922289564249e-06,
      "loss": 0.0022,
      "step": 129030
    },
    {
      "epoch": 0.8276224774157963,
      "grad_norm": 0.0965786948800087,
      "learning_rate": 7.288424663545891e-06,
      "loss": 0.0027,
      "step": 129040
    },
    {
      "epoch": 0.8276866143095823,
      "grad_norm": 0.027247527614235878,
      "learning_rate": 7.287927008851937e-06,
      "loss": 0.0031,
      "step": 129050
    },
    {
      "epoch": 0.8277507512033685,
      "grad_norm": 0.041653770953416824,
      "learning_rate": 7.2874293254886206e-06,
      "loss": 0.0024,
      "step": 129060
    },
    {
      "epoch": 0.8278148880971545,
      "grad_norm": 0.09017782658338547,
      "learning_rate": 7.2869316134621825e-06,
      "loss": 0.002,
      "step": 129070
    },
    {
      "epoch": 0.8278790249909407,
      "grad_norm": 0.24532608687877655,
      "learning_rate": 7.2864338727788556e-06,
      "loss": 0.0016,
      "step": 129080
    },
    {
      "epoch": 0.8279431618847267,
      "grad_norm": 0.00983749981969595,
      "learning_rate": 7.285936103444878e-06,
      "loss": 0.0021,
      "step": 129090
    },
    {
      "epoch": 0.8280072987785129,
      "grad_norm": 0.20129895210266113,
      "learning_rate": 7.2854383054664865e-06,
      "loss": 0.0015,
      "step": 129100
    },
    {
      "epoch": 0.8280714356722989,
      "grad_norm": 0.14603929221630096,
      "learning_rate": 7.284940478849921e-06,
      "loss": 0.0033,
      "step": 129110
    },
    {
      "epoch": 0.828135572566085,
      "grad_norm": 0.14177480340003967,
      "learning_rate": 7.284442623601417e-06,
      "loss": 0.0038,
      "step": 129120
    },
    {
      "epoch": 0.8281997094598712,
      "grad_norm": 0.024781549349427223,
      "learning_rate": 7.2839447397272155e-06,
      "loss": 0.0064,
      "step": 129130
    },
    {
      "epoch": 0.8282638463536572,
      "grad_norm": 0.0348445363342762,
      "learning_rate": 7.283446827233554e-06,
      "loss": 0.0033,
      "step": 129140
    },
    {
      "epoch": 0.8283279832474434,
      "grad_norm": 0.06163106858730316,
      "learning_rate": 7.282948886126672e-06,
      "loss": 0.0028,
      "step": 129150
    },
    {
      "epoch": 0.8283921201412294,
      "grad_norm": 0.18024040758609772,
      "learning_rate": 7.282450916412809e-06,
      "loss": 0.0021,
      "step": 129160
    },
    {
      "epoch": 0.8284562570350156,
      "grad_norm": 0.012435585260391235,
      "learning_rate": 7.281952918098203e-06,
      "loss": 0.0025,
      "step": 129170
    },
    {
      "epoch": 0.8285203939288016,
      "grad_norm": 0.18005426228046417,
      "learning_rate": 7.281454891189098e-06,
      "loss": 0.0023,
      "step": 129180
    },
    {
      "epoch": 0.8285845308225878,
      "grad_norm": 0.10598894953727722,
      "learning_rate": 7.280956835691732e-06,
      "loss": 0.0025,
      "step": 129190
    },
    {
      "epoch": 0.8286486677163738,
      "grad_norm": 0.13215792179107666,
      "learning_rate": 7.280458751612346e-06,
      "loss": 0.0016,
      "step": 129200
    },
    {
      "epoch": 0.82871280461016,
      "grad_norm": 0.12053846567869186,
      "learning_rate": 7.279960638957182e-06,
      "loss": 0.0041,
      "step": 129210
    },
    {
      "epoch": 0.828776941503946,
      "grad_norm": 0.08754531294107437,
      "learning_rate": 7.279462497732484e-06,
      "loss": 0.0035,
      "step": 129220
    },
    {
      "epoch": 0.8288410783977321,
      "grad_norm": 0.08480066806077957,
      "learning_rate": 7.278964327944489e-06,
      "loss": 0.0014,
      "step": 129230
    },
    {
      "epoch": 0.8289052152915182,
      "grad_norm": 0.1272183209657669,
      "learning_rate": 7.2784661295994445e-06,
      "loss": 0.0029,
      "step": 129240
    },
    {
      "epoch": 0.8289693521853043,
      "grad_norm": 0.03505934029817581,
      "learning_rate": 7.277967902703589e-06,
      "loss": 0.0019,
      "step": 129250
    },
    {
      "epoch": 0.8290334890790904,
      "grad_norm": 0.08902698010206223,
      "learning_rate": 7.277469647263169e-06,
      "loss": 0.0015,
      "step": 129260
    },
    {
      "epoch": 0.8290976259728765,
      "grad_norm": 0.053147412836551666,
      "learning_rate": 7.276971363284424e-06,
      "loss": 0.0027,
      "step": 129270
    },
    {
      "epoch": 0.8291617628666627,
      "grad_norm": 0.07472405582666397,
      "learning_rate": 7.276473050773604e-06,
      "loss": 0.0026,
      "step": 129280
    },
    {
      "epoch": 0.8292258997604487,
      "grad_norm": 0.01741631329059601,
      "learning_rate": 7.275974709736947e-06,
      "loss": 0.0023,
      "step": 129290
    },
    {
      "epoch": 0.8292900366542348,
      "grad_norm": 0.12162137031555176,
      "learning_rate": 7.275476340180703e-06,
      "loss": 0.0027,
      "step": 129300
    },
    {
      "epoch": 0.8293541735480209,
      "grad_norm": 0.10206171870231628,
      "learning_rate": 7.2749779421111126e-06,
      "loss": 0.002,
      "step": 129310
    },
    {
      "epoch": 0.829418310441807,
      "grad_norm": 0.08548274636268616,
      "learning_rate": 7.274479515534423e-06,
      "loss": 0.0019,
      "step": 129320
    },
    {
      "epoch": 0.8294824473355931,
      "grad_norm": 0.3585454523563385,
      "learning_rate": 7.273981060456878e-06,
      "loss": 0.0032,
      "step": 129330
    },
    {
      "epoch": 0.8295465842293792,
      "grad_norm": 0.13993075489997864,
      "learning_rate": 7.273482576884726e-06,
      "loss": 0.0049,
      "step": 129340
    },
    {
      "epoch": 0.8296107211231653,
      "grad_norm": 0.09641119092702866,
      "learning_rate": 7.272984064824213e-06,
      "loss": 0.0017,
      "step": 129350
    },
    {
      "epoch": 0.8296748580169514,
      "grad_norm": 0.08146881312131882,
      "learning_rate": 7.272485524281585e-06,
      "loss": 0.0014,
      "step": 129360
    },
    {
      "epoch": 0.8297389949107374,
      "grad_norm": 0.17681117355823517,
      "learning_rate": 7.27198695526309e-06,
      "loss": 0.0024,
      "step": 129370
    },
    {
      "epoch": 0.8298031318045236,
      "grad_norm": 0.08128388226032257,
      "learning_rate": 7.271488357774974e-06,
      "loss": 0.0022,
      "step": 129380
    },
    {
      "epoch": 0.8298672686983096,
      "grad_norm": 0.2925407886505127,
      "learning_rate": 7.270989731823486e-06,
      "loss": 0.0036,
      "step": 129390
    },
    {
      "epoch": 0.8299314055920958,
      "grad_norm": 0.07133324444293976,
      "learning_rate": 7.270491077414873e-06,
      "loss": 0.0033,
      "step": 129400
    },
    {
      "epoch": 0.8299955424858819,
      "grad_norm": 0.11704239249229431,
      "learning_rate": 7.269992394555385e-06,
      "loss": 0.0015,
      "step": 129410
    },
    {
      "epoch": 0.830059679379668,
      "grad_norm": 0.0500614158809185,
      "learning_rate": 7.269493683251269e-06,
      "loss": 0.0029,
      "step": 129420
    },
    {
      "epoch": 0.8301238162734541,
      "grad_norm": 0.2708382308483124,
      "learning_rate": 7.268994943508776e-06,
      "loss": 0.0036,
      "step": 129430
    },
    {
      "epoch": 0.8301879531672401,
      "grad_norm": 0.1924009919166565,
      "learning_rate": 7.268496175334153e-06,
      "loss": 0.0024,
      "step": 129440
    },
    {
      "epoch": 0.8302520900610263,
      "grad_norm": 0.1046387329697609,
      "learning_rate": 7.2679973787336535e-06,
      "loss": 0.0028,
      "step": 129450
    },
    {
      "epoch": 0.8303162269548123,
      "grad_norm": 0.12897948920726776,
      "learning_rate": 7.267498553713525e-06,
      "loss": 0.0035,
      "step": 129460
    },
    {
      "epoch": 0.8303803638485985,
      "grad_norm": 0.11709784716367722,
      "learning_rate": 7.26699970028002e-06,
      "loss": 0.0025,
      "step": 129470
    },
    {
      "epoch": 0.8304445007423845,
      "grad_norm": 0.04053686186671257,
      "learning_rate": 7.266500818439387e-06,
      "loss": 0.0014,
      "step": 129480
    },
    {
      "epoch": 0.8305086376361707,
      "grad_norm": 0.2003374546766281,
      "learning_rate": 7.26600190819788e-06,
      "loss": 0.0021,
      "step": 129490
    },
    {
      "epoch": 0.8305727745299567,
      "grad_norm": 0.12915728986263275,
      "learning_rate": 7.265502969561748e-06,
      "loss": 0.0021,
      "step": 129500
    },
    {
      "epoch": 0.8306369114237429,
      "grad_norm": 0.0786924734711647,
      "learning_rate": 7.265004002537247e-06,
      "loss": 0.002,
      "step": 129510
    },
    {
      "epoch": 0.8307010483175289,
      "grad_norm": 0.2431574910879135,
      "learning_rate": 7.264505007130627e-06,
      "loss": 0.0023,
      "step": 129520
    },
    {
      "epoch": 0.830765185211315,
      "grad_norm": 0.09418448060750961,
      "learning_rate": 7.2640059833481395e-06,
      "loss": 0.0034,
      "step": 129530
    },
    {
      "epoch": 0.8308293221051011,
      "grad_norm": 0.09966879338026047,
      "learning_rate": 7.26350693119604e-06,
      "loss": 0.0031,
      "step": 129540
    },
    {
      "epoch": 0.8308934589988872,
      "grad_norm": 0.020565232262015343,
      "learning_rate": 7.263007850680579e-06,
      "loss": 0.0018,
      "step": 129550
    },
    {
      "epoch": 0.8309575958926734,
      "grad_norm": 0.04436400160193443,
      "learning_rate": 7.262508741808014e-06,
      "loss": 0.0019,
      "step": 129560
    },
    {
      "epoch": 0.8310217327864594,
      "grad_norm": 0.12050460278987885,
      "learning_rate": 7.2620096045845965e-06,
      "loss": 0.003,
      "step": 129570
    },
    {
      "epoch": 0.8310858696802456,
      "grad_norm": 0.10381973534822464,
      "learning_rate": 7.261510439016583e-06,
      "loss": 0.0032,
      "step": 129580
    },
    {
      "epoch": 0.8311500065740316,
      "grad_norm": 0.16438117623329163,
      "learning_rate": 7.261011245110227e-06,
      "loss": 0.0025,
      "step": 129590
    },
    {
      "epoch": 0.8312141434678177,
      "grad_norm": 0.2219746708869934,
      "learning_rate": 7.260512022871785e-06,
      "loss": 0.0047,
      "step": 129600
    },
    {
      "epoch": 0.8312782803616038,
      "grad_norm": 0.14013536274433136,
      "learning_rate": 7.260012772307511e-06,
      "loss": 0.0017,
      "step": 129610
    },
    {
      "epoch": 0.8313424172553899,
      "grad_norm": 0.16215287148952484,
      "learning_rate": 7.259513493423661e-06,
      "loss": 0.002,
      "step": 129620
    },
    {
      "epoch": 0.831406554149176,
      "grad_norm": 0.27494677901268005,
      "learning_rate": 7.259014186226493e-06,
      "loss": 0.0025,
      "step": 129630
    },
    {
      "epoch": 0.8314706910429621,
      "grad_norm": 0.03116893582046032,
      "learning_rate": 7.258514850722263e-06,
      "loss": 0.002,
      "step": 129640
    },
    {
      "epoch": 0.8315348279367482,
      "grad_norm": 0.23137922585010529,
      "learning_rate": 7.258015486917226e-06,
      "loss": 0.0027,
      "step": 129650
    },
    {
      "epoch": 0.8315989648305343,
      "grad_norm": 0.11415555328130722,
      "learning_rate": 7.257516094817644e-06,
      "loss": 0.0017,
      "step": 129660
    },
    {
      "epoch": 0.8316631017243203,
      "grad_norm": 0.2785172164440155,
      "learning_rate": 7.257016674429769e-06,
      "loss": 0.0017,
      "step": 129670
    },
    {
      "epoch": 0.8317272386181065,
      "grad_norm": 0.06461936235427856,
      "learning_rate": 7.256517225759864e-06,
      "loss": 0.0013,
      "step": 129680
    },
    {
      "epoch": 0.8317913755118926,
      "grad_norm": 0.039516519755125046,
      "learning_rate": 7.256017748814185e-06,
      "loss": 0.0041,
      "step": 129690
    },
    {
      "epoch": 0.8318555124056787,
      "grad_norm": 0.18391795456409454,
      "learning_rate": 7.255518243598992e-06,
      "loss": 0.0034,
      "step": 129700
    },
    {
      "epoch": 0.8319196492994648,
      "grad_norm": 0.14168499410152435,
      "learning_rate": 7.2550187101205416e-06,
      "loss": 0.0019,
      "step": 129710
    },
    {
      "epoch": 0.8319837861932509,
      "grad_norm": 0.16390486061573029,
      "learning_rate": 7.254519148385095e-06,
      "loss": 0.0024,
      "step": 129720
    },
    {
      "epoch": 0.832047923087037,
      "grad_norm": 0.15905983746051788,
      "learning_rate": 7.254019558398915e-06,
      "loss": 0.0078,
      "step": 129730
    },
    {
      "epoch": 0.832112059980823,
      "grad_norm": 0.046173542737960815,
      "learning_rate": 7.253519940168256e-06,
      "loss": 0.0023,
      "step": 129740
    },
    {
      "epoch": 0.8321761968746092,
      "grad_norm": 0.14757530391216278,
      "learning_rate": 7.253020293699385e-06,
      "loss": 0.0023,
      "step": 129750
    },
    {
      "epoch": 0.8322403337683952,
      "grad_norm": 0.07440400868654251,
      "learning_rate": 7.252520618998555e-06,
      "loss": 0.0029,
      "step": 129760
    },
    {
      "epoch": 0.8323044706621814,
      "grad_norm": 0.14738357067108154,
      "learning_rate": 7.252020916072035e-06,
      "loss": 0.0014,
      "step": 129770
    },
    {
      "epoch": 0.8323686075559674,
      "grad_norm": 0.19934849441051483,
      "learning_rate": 7.251521184926083e-06,
      "loss": 0.0059,
      "step": 129780
    },
    {
      "epoch": 0.8324327444497536,
      "grad_norm": 0.07529287040233612,
      "learning_rate": 7.251021425566962e-06,
      "loss": 0.0021,
      "step": 129790
    },
    {
      "epoch": 0.8324968813435396,
      "grad_norm": 0.06696192920207977,
      "learning_rate": 7.250521638000931e-06,
      "loss": 0.0016,
      "step": 129800
    },
    {
      "epoch": 0.8325610182373258,
      "grad_norm": 0.17724058032035828,
      "learning_rate": 7.250021822234259e-06,
      "loss": 0.0031,
      "step": 129810
    },
    {
      "epoch": 0.8326251551311118,
      "grad_norm": 0.12434009462594986,
      "learning_rate": 7.2495219782732025e-06,
      "loss": 0.0033,
      "step": 129820
    },
    {
      "epoch": 0.8326892920248979,
      "grad_norm": 0.03853723779320717,
      "learning_rate": 7.249022106124028e-06,
      "loss": 0.002,
      "step": 129830
    },
    {
      "epoch": 0.8327534289186841,
      "grad_norm": 0.09288733452558517,
      "learning_rate": 7.248522205793002e-06,
      "loss": 0.0034,
      "step": 129840
    },
    {
      "epoch": 0.8328175658124701,
      "grad_norm": 0.3627587854862213,
      "learning_rate": 7.248022277286384e-06,
      "loss": 0.0031,
      "step": 129850
    },
    {
      "epoch": 0.8328817027062563,
      "grad_norm": 0.06937572360038757,
      "learning_rate": 7.247522320610441e-06,
      "loss": 0.0025,
      "step": 129860
    },
    {
      "epoch": 0.8329458396000423,
      "grad_norm": 0.062247905880212784,
      "learning_rate": 7.247022335771436e-06,
      "loss": 0.0036,
      "step": 129870
    },
    {
      "epoch": 0.8330099764938285,
      "grad_norm": 0.06953172385692596,
      "learning_rate": 7.246522322775635e-06,
      "loss": 0.002,
      "step": 129880
    },
    {
      "epoch": 0.8330741133876145,
      "grad_norm": 0.141593337059021,
      "learning_rate": 7.2460222816293035e-06,
      "loss": 0.0023,
      "step": 129890
    },
    {
      "epoch": 0.8331382502814006,
      "grad_norm": 0.03468858078122139,
      "learning_rate": 7.245522212338709e-06,
      "loss": 0.0035,
      "step": 129900
    },
    {
      "epoch": 0.8332023871751867,
      "grad_norm": 0.3736875057220459,
      "learning_rate": 7.245022114910116e-06,
      "loss": 0.0055,
      "step": 129910
    },
    {
      "epoch": 0.8332665240689728,
      "grad_norm": 0.04381706193089485,
      "learning_rate": 7.244521989349791e-06,
      "loss": 0.0017,
      "step": 129920
    },
    {
      "epoch": 0.8333306609627589,
      "grad_norm": 0.10753746330738068,
      "learning_rate": 7.244021835664001e-06,
      "loss": 0.0022,
      "step": 129930
    },
    {
      "epoch": 0.833394797856545,
      "grad_norm": 0.03241729736328125,
      "learning_rate": 7.243521653859015e-06,
      "loss": 0.0033,
      "step": 129940
    },
    {
      "epoch": 0.8334589347503311,
      "grad_norm": 0.06612688302993774,
      "learning_rate": 7.243021443941098e-06,
      "loss": 0.003,
      "step": 129950
    },
    {
      "epoch": 0.8335230716441172,
      "grad_norm": 0.274962842464447,
      "learning_rate": 7.2425212059165196e-06,
      "loss": 0.0025,
      "step": 129960
    },
    {
      "epoch": 0.8335872085379032,
      "grad_norm": 0.20958292484283447,
      "learning_rate": 7.242020939791547e-06,
      "loss": 0.003,
      "step": 129970
    },
    {
      "epoch": 0.8336513454316894,
      "grad_norm": 0.11247469484806061,
      "learning_rate": 7.2415206455724506e-06,
      "loss": 0.002,
      "step": 129980
    },
    {
      "epoch": 0.8337154823254755,
      "grad_norm": 0.06534750759601593,
      "learning_rate": 7.2410203232655e-06,
      "loss": 0.0017,
      "step": 129990
    },
    {
      "epoch": 0.8337796192192616,
      "grad_norm": 0.17713600397109985,
      "learning_rate": 7.240519972876961e-06,
      "loss": 0.0058,
      "step": 130000
    },
    {
      "epoch": 0.8338437561130477,
      "grad_norm": 0.06117572262883186,
      "learning_rate": 7.2400195944131054e-06,
      "loss": 0.0028,
      "step": 130010
    },
    {
      "epoch": 0.8339078930068338,
      "grad_norm": 0.1474536657333374,
      "learning_rate": 7.239519187880204e-06,
      "loss": 0.0019,
      "step": 130020
    },
    {
      "epoch": 0.8339720299006199,
      "grad_norm": 0.17380358278751373,
      "learning_rate": 7.239018753284527e-06,
      "loss": 0.0025,
      "step": 130030
    },
    {
      "epoch": 0.834036166794406,
      "grad_norm": 0.03862884268164635,
      "learning_rate": 7.238518290632343e-06,
      "loss": 0.0013,
      "step": 130040
    },
    {
      "epoch": 0.8341003036881921,
      "grad_norm": 0.11746586114168167,
      "learning_rate": 7.238017799929926e-06,
      "loss": 0.002,
      "step": 130050
    },
    {
      "epoch": 0.8341644405819781,
      "grad_norm": 0.048239219933748245,
      "learning_rate": 7.237517281183547e-06,
      "loss": 0.004,
      "step": 130060
    },
    {
      "epoch": 0.8342285774757643,
      "grad_norm": 0.14999274909496307,
      "learning_rate": 7.237016734399478e-06,
      "loss": 0.0025,
      "step": 130070
    },
    {
      "epoch": 0.8342927143695503,
      "grad_norm": 0.037866171449422836,
      "learning_rate": 7.2365161595839885e-06,
      "loss": 0.0027,
      "step": 130080
    },
    {
      "epoch": 0.8343568512633365,
      "grad_norm": 0.13778047263622284,
      "learning_rate": 7.236015556743355e-06,
      "loss": 0.0105,
      "step": 130090
    },
    {
      "epoch": 0.8344209881571225,
      "grad_norm": 0.06160407513380051,
      "learning_rate": 7.235514925883848e-06,
      "loss": 0.002,
      "step": 130100
    },
    {
      "epoch": 0.8344851250509087,
      "grad_norm": 0.10367187112569809,
      "learning_rate": 7.235014267011742e-06,
      "loss": 0.0052,
      "step": 130110
    },
    {
      "epoch": 0.8345492619446948,
      "grad_norm": 0.11204732954502106,
      "learning_rate": 7.234513580133307e-06,
      "loss": 0.0022,
      "step": 130120
    },
    {
      "epoch": 0.8346133988384808,
      "grad_norm": 0.24648122489452362,
      "learning_rate": 7.234012865254822e-06,
      "loss": 0.0027,
      "step": 130130
    },
    {
      "epoch": 0.834677535732267,
      "grad_norm": 0.1494959592819214,
      "learning_rate": 7.23351212238256e-06,
      "loss": 0.0016,
      "step": 130140
    },
    {
      "epoch": 0.834741672626053,
      "grad_norm": 0.16294920444488525,
      "learning_rate": 7.233011351522794e-06,
      "loss": 0.0026,
      "step": 130150
    },
    {
      "epoch": 0.8348058095198392,
      "grad_norm": 0.07746896147727966,
      "learning_rate": 7.2325105526818e-06,
      "loss": 0.0034,
      "step": 130160
    },
    {
      "epoch": 0.8348699464136252,
      "grad_norm": 0.1173417940735817,
      "learning_rate": 7.232009725865853e-06,
      "loss": 0.0018,
      "step": 130170
    },
    {
      "epoch": 0.8349340833074114,
      "grad_norm": 0.3902769684791565,
      "learning_rate": 7.231508871081228e-06,
      "loss": 0.0021,
      "step": 130180
    },
    {
      "epoch": 0.8349982202011974,
      "grad_norm": 0.05905061960220337,
      "learning_rate": 7.231007988334202e-06,
      "loss": 0.0025,
      "step": 130190
    },
    {
      "epoch": 0.8350623570949836,
      "grad_norm": 0.17392592132091522,
      "learning_rate": 7.230507077631054e-06,
      "loss": 0.0028,
      "step": 130200
    },
    {
      "epoch": 0.8351264939887696,
      "grad_norm": 0.09936745464801788,
      "learning_rate": 7.230006138978055e-06,
      "loss": 0.0033,
      "step": 130210
    },
    {
      "epoch": 0.8351906308825557,
      "grad_norm": 0.2895844578742981,
      "learning_rate": 7.229505172381488e-06,
      "loss": 0.0046,
      "step": 130220
    },
    {
      "epoch": 0.8352547677763418,
      "grad_norm": 0.0891617015004158,
      "learning_rate": 7.229004177847626e-06,
      "loss": 0.004,
      "step": 130230
    },
    {
      "epoch": 0.8353189046701279,
      "grad_norm": 0.041115667670965195,
      "learning_rate": 7.22850315538275e-06,
      "loss": 0.0025,
      "step": 130240
    },
    {
      "epoch": 0.835383041563914,
      "grad_norm": 0.10365470498800278,
      "learning_rate": 7.228002104993135e-06,
      "loss": 0.005,
      "step": 130250
    },
    {
      "epoch": 0.8354471784577001,
      "grad_norm": 0.11892421543598175,
      "learning_rate": 7.227501026685063e-06,
      "loss": 0.0026,
      "step": 130260
    },
    {
      "epoch": 0.8355113153514863,
      "grad_norm": 0.053219012916088104,
      "learning_rate": 7.22699992046481e-06,
      "loss": 0.0028,
      "step": 130270
    },
    {
      "epoch": 0.8355754522452723,
      "grad_norm": 0.1299111396074295,
      "learning_rate": 7.226498786338658e-06,
      "loss": 0.003,
      "step": 130280
    },
    {
      "epoch": 0.8356395891390584,
      "grad_norm": 0.04380191117525101,
      "learning_rate": 7.225997624312883e-06,
      "loss": 0.0015,
      "step": 130290
    },
    {
      "epoch": 0.8357037260328445,
      "grad_norm": 0.1472407877445221,
      "learning_rate": 7.225496434393769e-06,
      "loss": 0.0036,
      "step": 130300
    },
    {
      "epoch": 0.8357678629266306,
      "grad_norm": 0.13299348950386047,
      "learning_rate": 7.224995216587592e-06,
      "loss": 0.0023,
      "step": 130310
    },
    {
      "epoch": 0.8358319998204167,
      "grad_norm": 0.0585072822868824,
      "learning_rate": 7.224493970900636e-06,
      "loss": 0.0038,
      "step": 130320
    },
    {
      "epoch": 0.8358961367142028,
      "grad_norm": 0.078093983232975,
      "learning_rate": 7.22399269733918e-06,
      "loss": 0.0036,
      "step": 130330
    },
    {
      "epoch": 0.8359602736079889,
      "grad_norm": 0.16029320657253265,
      "learning_rate": 7.223491395909506e-06,
      "loss": 0.0023,
      "step": 130340
    },
    {
      "epoch": 0.836024410501775,
      "grad_norm": 0.08965877443552017,
      "learning_rate": 7.2229900666178964e-06,
      "loss": 0.0037,
      "step": 130350
    },
    {
      "epoch": 0.836088547395561,
      "grad_norm": 0.12641486525535583,
      "learning_rate": 7.2224887094706305e-06,
      "loss": 0.0015,
      "step": 130360
    },
    {
      "epoch": 0.8361526842893472,
      "grad_norm": 0.27190524339675903,
      "learning_rate": 7.221987324473996e-06,
      "loss": 0.0025,
      "step": 130370
    },
    {
      "epoch": 0.8362168211831332,
      "grad_norm": 0.09276844561100006,
      "learning_rate": 7.22148591163427e-06,
      "loss": 0.0015,
      "step": 130380
    },
    {
      "epoch": 0.8362809580769194,
      "grad_norm": 0.11088048666715622,
      "learning_rate": 7.220984470957739e-06,
      "loss": 0.0022,
      "step": 130390
    },
    {
      "epoch": 0.8363450949707055,
      "grad_norm": 0.050899092108011246,
      "learning_rate": 7.220483002450685e-06,
      "loss": 0.0027,
      "step": 130400
    },
    {
      "epoch": 0.8364092318644916,
      "grad_norm": 0.12364206463098526,
      "learning_rate": 7.219981506119391e-06,
      "loss": 0.0021,
      "step": 130410
    },
    {
      "epoch": 0.8364733687582777,
      "grad_norm": 0.051249511539936066,
      "learning_rate": 7.219479981970141e-06,
      "loss": 0.0031,
      "step": 130420
    },
    {
      "epoch": 0.8365375056520638,
      "grad_norm": 0.0800173357129097,
      "learning_rate": 7.218978430009224e-06,
      "loss": 0.0019,
      "step": 130430
    },
    {
      "epoch": 0.8366016425458499,
      "grad_norm": 0.05738505348563194,
      "learning_rate": 7.218476850242919e-06,
      "loss": 0.0015,
      "step": 130440
    },
    {
      "epoch": 0.8366657794396359,
      "grad_norm": 0.10963942855596542,
      "learning_rate": 7.217975242677515e-06,
      "loss": 0.0018,
      "step": 130450
    },
    {
      "epoch": 0.8367299163334221,
      "grad_norm": 0.15522795915603638,
      "learning_rate": 7.2174736073192954e-06,
      "loss": 0.0027,
      "step": 130460
    },
    {
      "epoch": 0.8367940532272081,
      "grad_norm": 0.11064627766609192,
      "learning_rate": 7.216971944174547e-06,
      "loss": 0.0031,
      "step": 130470
    },
    {
      "epoch": 0.8368581901209943,
      "grad_norm": 0.2598922848701477,
      "learning_rate": 7.216470253249554e-06,
      "loss": 0.0022,
      "step": 130480
    },
    {
      "epoch": 0.8369223270147803,
      "grad_norm": 0.15451756119728088,
      "learning_rate": 7.215968534550608e-06,
      "loss": 0.0032,
      "step": 130490
    },
    {
      "epoch": 0.8369864639085665,
      "grad_norm": 0.12410331517457962,
      "learning_rate": 7.21546678808399e-06,
      "loss": 0.0022,
      "step": 130500
    },
    {
      "epoch": 0.8370506008023525,
      "grad_norm": 0.3346932828426361,
      "learning_rate": 7.214965013855992e-06,
      "loss": 0.0046,
      "step": 130510
    },
    {
      "epoch": 0.8371147376961386,
      "grad_norm": 0.03758867457509041,
      "learning_rate": 7.214463211872896e-06,
      "loss": 0.0025,
      "step": 130520
    },
    {
      "epoch": 0.8371788745899247,
      "grad_norm": 0.18766102194786072,
      "learning_rate": 7.213961382140995e-06,
      "loss": 0.0034,
      "step": 130530
    },
    {
      "epoch": 0.8372430114837108,
      "grad_norm": 0.404788613319397,
      "learning_rate": 7.2134595246665766e-06,
      "loss": 0.0031,
      "step": 130540
    },
    {
      "epoch": 0.837307148377497,
      "grad_norm": 0.10620911419391632,
      "learning_rate": 7.212957639455926e-06,
      "loss": 0.0018,
      "step": 130550
    },
    {
      "epoch": 0.837371285271283,
      "grad_norm": 0.1350063532590866,
      "learning_rate": 7.212455726515337e-06,
      "loss": 0.0041,
      "step": 130560
    },
    {
      "epoch": 0.8374354221650692,
      "grad_norm": 0.20413848757743835,
      "learning_rate": 7.2119537858510954e-06,
      "loss": 0.0025,
      "step": 130570
    },
    {
      "epoch": 0.8374995590588552,
      "grad_norm": 0.2801940143108368,
      "learning_rate": 7.211451817469493e-06,
      "loss": 0.0014,
      "step": 130580
    },
    {
      "epoch": 0.8375636959526414,
      "grad_norm": 0.24905692040920258,
      "learning_rate": 7.210949821376817e-06,
      "loss": 0.0023,
      "step": 130590
    },
    {
      "epoch": 0.8376278328464274,
      "grad_norm": 0.07550957798957825,
      "learning_rate": 7.2104477975793605e-06,
      "loss": 0.0017,
      "step": 130600
    },
    {
      "epoch": 0.8376919697402135,
      "grad_norm": 0.04287779703736305,
      "learning_rate": 7.209945746083413e-06,
      "loss": 0.0014,
      "step": 130610
    },
    {
      "epoch": 0.8377561066339996,
      "grad_norm": 0.07093918323516846,
      "learning_rate": 7.209443666895269e-06,
      "loss": 0.003,
      "step": 130620
    },
    {
      "epoch": 0.8378202435277857,
      "grad_norm": 0.05075628682971001,
      "learning_rate": 7.208941560021212e-06,
      "loss": 0.0018,
      "step": 130630
    },
    {
      "epoch": 0.8378843804215718,
      "grad_norm": 0.1476941704750061,
      "learning_rate": 7.2084394254675415e-06,
      "loss": 0.0033,
      "step": 130640
    },
    {
      "epoch": 0.8379485173153579,
      "grad_norm": 0.14929373562335968,
      "learning_rate": 7.207937263240546e-06,
      "loss": 0.0025,
      "step": 130650
    },
    {
      "epoch": 0.838012654209144,
      "grad_norm": 0.07781266421079636,
      "learning_rate": 7.207435073346518e-06,
      "loss": 0.0035,
      "step": 130660
    },
    {
      "epoch": 0.8380767911029301,
      "grad_norm": 0.09879343956708908,
      "learning_rate": 7.20693285579175e-06,
      "loss": 0.0032,
      "step": 130670
    },
    {
      "epoch": 0.8381409279967162,
      "grad_norm": 0.040806982666254044,
      "learning_rate": 7.206430610582538e-06,
      "loss": 0.0047,
      "step": 130680
    },
    {
      "epoch": 0.8382050648905023,
      "grad_norm": 0.05319861322641373,
      "learning_rate": 7.205928337725173e-06,
      "loss": 0.0039,
      "step": 130690
    },
    {
      "epoch": 0.8382692017842884,
      "grad_norm": 0.055010631680488586,
      "learning_rate": 7.2054260372259486e-06,
      "loss": 0.0036,
      "step": 130700
    },
    {
      "epoch": 0.8383333386780745,
      "grad_norm": 0.05272675305604935,
      "learning_rate": 7.204923709091162e-06,
      "loss": 0.0037,
      "step": 130710
    },
    {
      "epoch": 0.8383974755718606,
      "grad_norm": 0.11825592070817947,
      "learning_rate": 7.204421353327104e-06,
      "loss": 0.0026,
      "step": 130720
    },
    {
      "epoch": 0.8384616124656467,
      "grad_norm": 0.07398737967014313,
      "learning_rate": 7.2039189699400705e-06,
      "loss": 0.0056,
      "step": 130730
    },
    {
      "epoch": 0.8385257493594328,
      "grad_norm": 0.20622889697551727,
      "learning_rate": 7.203416558936357e-06,
      "loss": 0.0024,
      "step": 130740
    },
    {
      "epoch": 0.8385898862532188,
      "grad_norm": 0.13918302953243256,
      "learning_rate": 7.202914120322261e-06,
      "loss": 0.0025,
      "step": 130750
    },
    {
      "epoch": 0.838654023147005,
      "grad_norm": 0.2289324551820755,
      "learning_rate": 7.202411654104074e-06,
      "loss": 0.0015,
      "step": 130760
    },
    {
      "epoch": 0.838718160040791,
      "grad_norm": 0.18719933927059174,
      "learning_rate": 7.201909160288098e-06,
      "loss": 0.0024,
      "step": 130770
    },
    {
      "epoch": 0.8387822969345772,
      "grad_norm": 0.12550872564315796,
      "learning_rate": 7.201406638880624e-06,
      "loss": 0.0013,
      "step": 130780
    },
    {
      "epoch": 0.8388464338283632,
      "grad_norm": 0.1475445032119751,
      "learning_rate": 7.200904089887954e-06,
      "loss": 0.0042,
      "step": 130790
    },
    {
      "epoch": 0.8389105707221494,
      "grad_norm": 0.0035977119114249945,
      "learning_rate": 7.20040151331638e-06,
      "loss": 0.0016,
      "step": 130800
    },
    {
      "epoch": 0.8389747076159354,
      "grad_norm": 0.16882912814617157,
      "learning_rate": 7.199898909172202e-06,
      "loss": 0.001,
      "step": 130810
    },
    {
      "epoch": 0.8390388445097215,
      "grad_norm": 0.07918764650821686,
      "learning_rate": 7.199396277461721e-06,
      "loss": 0.0019,
      "step": 130820
    },
    {
      "epoch": 0.8391029814035077,
      "grad_norm": 0.17589236795902252,
      "learning_rate": 7.198893618191231e-06,
      "loss": 0.0032,
      "step": 130830
    },
    {
      "epoch": 0.8391671182972937,
      "grad_norm": 0.15713556110858917,
      "learning_rate": 7.198390931367034e-06,
      "loss": 0.004,
      "step": 130840
    },
    {
      "epoch": 0.8392312551910799,
      "grad_norm": 0.12668083608150482,
      "learning_rate": 7.197888216995428e-06,
      "loss": 0.0027,
      "step": 130850
    },
    {
      "epoch": 0.8392953920848659,
      "grad_norm": 0.05177538841962814,
      "learning_rate": 7.19738547508271e-06,
      "loss": 0.0023,
      "step": 130860
    },
    {
      "epoch": 0.8393595289786521,
      "grad_norm": 0.045602746307849884,
      "learning_rate": 7.196882705635182e-06,
      "loss": 0.0014,
      "step": 130870
    },
    {
      "epoch": 0.8394236658724381,
      "grad_norm": 0.09825082868337631,
      "learning_rate": 7.196379908659144e-06,
      "loss": 0.0027,
      "step": 130880
    },
    {
      "epoch": 0.8394878027662243,
      "grad_norm": 0.0659918487071991,
      "learning_rate": 7.195877084160895e-06,
      "loss": 0.0016,
      "step": 130890
    },
    {
      "epoch": 0.8395519396600103,
      "grad_norm": 0.42159461975097656,
      "learning_rate": 7.195374232146738e-06,
      "loss": 0.0078,
      "step": 130900
    },
    {
      "epoch": 0.8396160765537964,
      "grad_norm": 0.08648031204938889,
      "learning_rate": 7.194871352622972e-06,
      "loss": 0.0044,
      "step": 130910
    },
    {
      "epoch": 0.8396802134475825,
      "grad_norm": 0.12601043283939362,
      "learning_rate": 7.1943684455959e-06,
      "loss": 0.0025,
      "step": 130920
    },
    {
      "epoch": 0.8397443503413686,
      "grad_norm": 0.14105676114559174,
      "learning_rate": 7.193865511071822e-06,
      "loss": 0.0029,
      "step": 130930
    },
    {
      "epoch": 0.8398084872351547,
      "grad_norm": 0.12207061052322388,
      "learning_rate": 7.1933625490570434e-06,
      "loss": 0.0015,
      "step": 130940
    },
    {
      "epoch": 0.8398726241289408,
      "grad_norm": 0.10712200403213501,
      "learning_rate": 7.192859559557863e-06,
      "loss": 0.0018,
      "step": 130950
    },
    {
      "epoch": 0.839936761022727,
      "grad_norm": 0.2787437438964844,
      "learning_rate": 7.192356542580585e-06,
      "loss": 0.0032,
      "step": 130960
    },
    {
      "epoch": 0.840000897916513,
      "grad_norm": 0.11498334258794785,
      "learning_rate": 7.191853498131512e-06,
      "loss": 0.0017,
      "step": 130970
    },
    {
      "epoch": 0.8400650348102991,
      "grad_norm": 0.19720852375030518,
      "learning_rate": 7.1913504262169495e-06,
      "loss": 0.0036,
      "step": 130980
    },
    {
      "epoch": 0.8401291717040852,
      "grad_norm": 0.04595879465341568,
      "learning_rate": 7.1908473268432e-06,
      "loss": 0.0025,
      "step": 130990
    },
    {
      "epoch": 0.8401933085978713,
      "grad_norm": 0.19816502928733826,
      "learning_rate": 7.190344200016568e-06,
      "loss": 0.0016,
      "step": 131000
    },
    {
      "epoch": 0.8402574454916574,
      "grad_norm": 0.08880461752414703,
      "learning_rate": 7.189841045743357e-06,
      "loss": 0.0019,
      "step": 131010
    },
    {
      "epoch": 0.8403215823854435,
      "grad_norm": 0.05284334719181061,
      "learning_rate": 7.189337864029872e-06,
      "loss": 0.0023,
      "step": 131020
    },
    {
      "epoch": 0.8403857192792296,
      "grad_norm": 0.05432449281215668,
      "learning_rate": 7.18883465488242e-06,
      "loss": 0.0035,
      "step": 131030
    },
    {
      "epoch": 0.8404498561730157,
      "grad_norm": 0.1101267859339714,
      "learning_rate": 7.188331418307304e-06,
      "loss": 0.0011,
      "step": 131040
    },
    {
      "epoch": 0.8405139930668017,
      "grad_norm": 0.18215373158454895,
      "learning_rate": 7.1878281543108334e-06,
      "loss": 0.0017,
      "step": 131050
    },
    {
      "epoch": 0.8405781299605879,
      "grad_norm": 0.12219468504190445,
      "learning_rate": 7.1873248628993105e-06,
      "loss": 0.0057,
      "step": 131060
    },
    {
      "epoch": 0.8406422668543739,
      "grad_norm": 0.10829061269760132,
      "learning_rate": 7.186821544079046e-06,
      "loss": 0.0026,
      "step": 131070
    },
    {
      "epoch": 0.8407064037481601,
      "grad_norm": 0.12962056696414948,
      "learning_rate": 7.186318197856343e-06,
      "loss": 0.002,
      "step": 131080
    },
    {
      "epoch": 0.8407705406419461,
      "grad_norm": 0.06855259090662003,
      "learning_rate": 7.185814824237512e-06,
      "loss": 0.0015,
      "step": 131090
    },
    {
      "epoch": 0.8408346775357323,
      "grad_norm": 0.08255480229854584,
      "learning_rate": 7.185311423228856e-06,
      "loss": 0.0033,
      "step": 131100
    },
    {
      "epoch": 0.8408988144295184,
      "grad_norm": 0.08055885881185532,
      "learning_rate": 7.184807994836689e-06,
      "loss": 0.0018,
      "step": 131110
    },
    {
      "epoch": 0.8409629513233045,
      "grad_norm": 0.08228079974651337,
      "learning_rate": 7.184304539067315e-06,
      "loss": 0.0036,
      "step": 131120
    },
    {
      "epoch": 0.8410270882170906,
      "grad_norm": 0.09898220002651215,
      "learning_rate": 7.183801055927044e-06,
      "loss": 0.004,
      "step": 131130
    },
    {
      "epoch": 0.8410912251108766,
      "grad_norm": 0.13123354315757751,
      "learning_rate": 7.183297545422185e-06,
      "loss": 0.002,
      "step": 131140
    },
    {
      "epoch": 0.8411553620046628,
      "grad_norm": 0.11321281641721725,
      "learning_rate": 7.1827940075590486e-06,
      "loss": 0.0037,
      "step": 131150
    },
    {
      "epoch": 0.8412194988984488,
      "grad_norm": 0.046443551778793335,
      "learning_rate": 7.1822904423439425e-06,
      "loss": 0.0018,
      "step": 131160
    },
    {
      "epoch": 0.841283635792235,
      "grad_norm": 0.12373586744070053,
      "learning_rate": 7.1817868497831765e-06,
      "loss": 0.002,
      "step": 131170
    },
    {
      "epoch": 0.841347772686021,
      "grad_norm": 0.11009549349546432,
      "learning_rate": 7.1812832298830626e-06,
      "loss": 0.002,
      "step": 131180
    },
    {
      "epoch": 0.8414119095798072,
      "grad_norm": 0.09203784167766571,
      "learning_rate": 7.18077958264991e-06,
      "loss": 0.0021,
      "step": 131190
    },
    {
      "epoch": 0.8414760464735932,
      "grad_norm": 0.07105744630098343,
      "learning_rate": 7.180275908090033e-06,
      "loss": 0.0021,
      "step": 131200
    },
    {
      "epoch": 0.8415401833673793,
      "grad_norm": 0.10325989872217178,
      "learning_rate": 7.179772206209739e-06,
      "loss": 0.0044,
      "step": 131210
    },
    {
      "epoch": 0.8416043202611654,
      "grad_norm": 0.14352548122406006,
      "learning_rate": 7.179268477015342e-06,
      "loss": 0.0031,
      "step": 131220
    },
    {
      "epoch": 0.8416684571549515,
      "grad_norm": 0.14424268901348114,
      "learning_rate": 7.178764720513154e-06,
      "loss": 0.0017,
      "step": 131230
    },
    {
      "epoch": 0.8417325940487376,
      "grad_norm": 0.10106702893972397,
      "learning_rate": 7.178260936709484e-06,
      "loss": 0.003,
      "step": 131240
    },
    {
      "epoch": 0.8417967309425237,
      "grad_norm": 0.20695577561855316,
      "learning_rate": 7.17775712561065e-06,
      "loss": 0.0017,
      "step": 131250
    },
    {
      "epoch": 0.8418608678363099,
      "grad_norm": 0.19014497101306915,
      "learning_rate": 7.177253287222964e-06,
      "loss": 0.0033,
      "step": 131260
    },
    {
      "epoch": 0.8419250047300959,
      "grad_norm": 0.14594174921512604,
      "learning_rate": 7.176749421552736e-06,
      "loss": 0.0046,
      "step": 131270
    },
    {
      "epoch": 0.841989141623882,
      "grad_norm": 0.14901790022850037,
      "learning_rate": 7.176245528606284e-06,
      "loss": 0.0024,
      "step": 131280
    },
    {
      "epoch": 0.8420532785176681,
      "grad_norm": 0.13513055443763733,
      "learning_rate": 7.175741608389919e-06,
      "loss": 0.0024,
      "step": 131290
    },
    {
      "epoch": 0.8421174154114542,
      "grad_norm": 0.027130547910928726,
      "learning_rate": 7.1752376609099575e-06,
      "loss": 0.0017,
      "step": 131300
    },
    {
      "epoch": 0.8421815523052403,
      "grad_norm": 0.12011363357305527,
      "learning_rate": 7.174733686172712e-06,
      "loss": 0.0052,
      "step": 131310
    },
    {
      "epoch": 0.8422456891990264,
      "grad_norm": 0.05899891257286072,
      "learning_rate": 7.1742296841845e-06,
      "loss": 0.0048,
      "step": 131320
    },
    {
      "epoch": 0.8423098260928125,
      "grad_norm": 0.06901489943265915,
      "learning_rate": 7.173725654951636e-06,
      "loss": 0.002,
      "step": 131330
    },
    {
      "epoch": 0.8423739629865986,
      "grad_norm": 0.17242315411567688,
      "learning_rate": 7.173221598480438e-06,
      "loss": 0.0024,
      "step": 131340
    },
    {
      "epoch": 0.8424380998803846,
      "grad_norm": 0.20778584480285645,
      "learning_rate": 7.172717514777217e-06,
      "loss": 0.0026,
      "step": 131350
    },
    {
      "epoch": 0.8425022367741708,
      "grad_norm": 0.13705037534236908,
      "learning_rate": 7.172213403848295e-06,
      "loss": 0.0037,
      "step": 131360
    },
    {
      "epoch": 0.8425663736679568,
      "grad_norm": 0.041558653116226196,
      "learning_rate": 7.171709265699984e-06,
      "loss": 0.0022,
      "step": 131370
    },
    {
      "epoch": 0.842630510561743,
      "grad_norm": 0.2619161605834961,
      "learning_rate": 7.171205100338605e-06,
      "loss": 0.0024,
      "step": 131380
    },
    {
      "epoch": 0.8426946474555291,
      "grad_norm": 0.004847593605518341,
      "learning_rate": 7.170700907770476e-06,
      "loss": 0.0025,
      "step": 131390
    },
    {
      "epoch": 0.8427587843493152,
      "grad_norm": 0.06404387950897217,
      "learning_rate": 7.170196688001911e-06,
      "loss": 0.0021,
      "step": 131400
    },
    {
      "epoch": 0.8428229212431013,
      "grad_norm": 0.09147094935178757,
      "learning_rate": 7.169692441039233e-06,
      "loss": 0.0031,
      "step": 131410
    },
    {
      "epoch": 0.8428870581368874,
      "grad_norm": 0.07531020790338516,
      "learning_rate": 7.1691881668887565e-06,
      "loss": 0.0014,
      "step": 131420
    },
    {
      "epoch": 0.8429511950306735,
      "grad_norm": 0.07780922949314117,
      "learning_rate": 7.168683865556803e-06,
      "loss": 0.0016,
      "step": 131430
    },
    {
      "epoch": 0.8430153319244595,
      "grad_norm": 0.0524565689265728,
      "learning_rate": 7.168179537049689e-06,
      "loss": 0.0027,
      "step": 131440
    },
    {
      "epoch": 0.8430794688182457,
      "grad_norm": 0.040026117116212845,
      "learning_rate": 7.167675181373737e-06,
      "loss": 0.0013,
      "step": 131450
    },
    {
      "epoch": 0.8431436057120317,
      "grad_norm": 0.004006249364465475,
      "learning_rate": 7.167170798535265e-06,
      "loss": 0.0011,
      "step": 131460
    },
    {
      "epoch": 0.8432077426058179,
      "grad_norm": 0.1320967972278595,
      "learning_rate": 7.166666388540595e-06,
      "loss": 0.0021,
      "step": 131470
    },
    {
      "epoch": 0.8432718794996039,
      "grad_norm": 0.09947043657302856,
      "learning_rate": 7.166161951396045e-06,
      "loss": 0.0022,
      "step": 131480
    },
    {
      "epoch": 0.8433360163933901,
      "grad_norm": 0.15966539084911346,
      "learning_rate": 7.1656574871079394e-06,
      "loss": 0.0022,
      "step": 131490
    },
    {
      "epoch": 0.8434001532871761,
      "grad_norm": 0.1235571950674057,
      "learning_rate": 7.165152995682597e-06,
      "loss": 0.0023,
      "step": 131500
    },
    {
      "epoch": 0.8434642901809623,
      "grad_norm": 0.020679375156760216,
      "learning_rate": 7.1646484771263404e-06,
      "loss": 0.0022,
      "step": 131510
    },
    {
      "epoch": 0.8435284270747483,
      "grad_norm": 0.13467086851596832,
      "learning_rate": 7.164143931445491e-06,
      "loss": 0.0034,
      "step": 131520
    },
    {
      "epoch": 0.8435925639685344,
      "grad_norm": 0.08703139424324036,
      "learning_rate": 7.16363935864637e-06,
      "loss": 0.0049,
      "step": 131530
    },
    {
      "epoch": 0.8436567008623206,
      "grad_norm": 0.09485015273094177,
      "learning_rate": 7.1631347587353035e-06,
      "loss": 0.0033,
      "step": 131540
    },
    {
      "epoch": 0.8437208377561066,
      "grad_norm": 0.024869795888662338,
      "learning_rate": 7.162630131718613e-06,
      "loss": 0.0016,
      "step": 131550
    },
    {
      "epoch": 0.8437849746498928,
      "grad_norm": 0.1564255952835083,
      "learning_rate": 7.16212547760262e-06,
      "loss": 0.0036,
      "step": 131560
    },
    {
      "epoch": 0.8438491115436788,
      "grad_norm": 0.44133636355400085,
      "learning_rate": 7.16162079639365e-06,
      "loss": 0.003,
      "step": 131570
    },
    {
      "epoch": 0.843913248437465,
      "grad_norm": 0.07069187611341476,
      "learning_rate": 7.161116088098026e-06,
      "loss": 0.0036,
      "step": 131580
    },
    {
      "epoch": 0.843977385331251,
      "grad_norm": 0.12002238631248474,
      "learning_rate": 7.160611352722073e-06,
      "loss": 0.0035,
      "step": 131590
    },
    {
      "epoch": 0.8440415222250371,
      "grad_norm": 0.10347267240285873,
      "learning_rate": 7.160106590272117e-06,
      "loss": 0.0019,
      "step": 131600
    },
    {
      "epoch": 0.8441056591188232,
      "grad_norm": 0.38686108589172363,
      "learning_rate": 7.15960180075448e-06,
      "loss": 0.0019,
      "step": 131610
    },
    {
      "epoch": 0.8441697960126093,
      "grad_norm": 0.1088900938630104,
      "learning_rate": 7.15909698417549e-06,
      "loss": 0.0017,
      "step": 131620
    },
    {
      "epoch": 0.8442339329063954,
      "grad_norm": 0.25590530037879944,
      "learning_rate": 7.15859214054147e-06,
      "loss": 0.003,
      "step": 131630
    },
    {
      "epoch": 0.8442980698001815,
      "grad_norm": 0.22568881511688232,
      "learning_rate": 7.15808726985875e-06,
      "loss": 0.0026,
      "step": 131640
    },
    {
      "epoch": 0.8443622066939676,
      "grad_norm": 0.09165618568658829,
      "learning_rate": 7.157582372133653e-06,
      "loss": 0.0027,
      "step": 131650
    },
    {
      "epoch": 0.8444263435877537,
      "grad_norm": 0.23040997982025146,
      "learning_rate": 7.157077447372507e-06,
      "loss": 0.0025,
      "step": 131660
    },
    {
      "epoch": 0.8444904804815399,
      "grad_norm": 0.0696287453174591,
      "learning_rate": 7.15657249558164e-06,
      "loss": 0.0023,
      "step": 131670
    },
    {
      "epoch": 0.8445546173753259,
      "grad_norm": 0.11746630817651749,
      "learning_rate": 7.1560675167673766e-06,
      "loss": 0.0022,
      "step": 131680
    },
    {
      "epoch": 0.844618754269112,
      "grad_norm": 0.13166771829128265,
      "learning_rate": 7.155562510936047e-06,
      "loss": 0.0026,
      "step": 131690
    },
    {
      "epoch": 0.8446828911628981,
      "grad_norm": 0.16737747192382812,
      "learning_rate": 7.155057478093979e-06,
      "loss": 0.0033,
      "step": 131700
    },
    {
      "epoch": 0.8447470280566842,
      "grad_norm": 0.0332891121506691,
      "learning_rate": 7.1545524182475005e-06,
      "loss": 0.0011,
      "step": 131710
    },
    {
      "epoch": 0.8448111649504703,
      "grad_norm": 0.15007895231246948,
      "learning_rate": 7.15404733140294e-06,
      "loss": 0.0017,
      "step": 131720
    },
    {
      "epoch": 0.8448753018442564,
      "grad_norm": 0.22700099647045135,
      "learning_rate": 7.153542217566627e-06,
      "loss": 0.0052,
      "step": 131730
    },
    {
      "epoch": 0.8449394387380424,
      "grad_norm": 0.0855378583073616,
      "learning_rate": 7.1530370767448894e-06,
      "loss": 0.0029,
      "step": 131740
    },
    {
      "epoch": 0.8450035756318286,
      "grad_norm": 0.3931421935558319,
      "learning_rate": 7.152531908944061e-06,
      "loss": 0.0035,
      "step": 131750
    },
    {
      "epoch": 0.8450677125256146,
      "grad_norm": 0.04366033524274826,
      "learning_rate": 7.152026714170468e-06,
      "loss": 0.0021,
      "step": 131760
    },
    {
      "epoch": 0.8451318494194008,
      "grad_norm": 0.07778988778591156,
      "learning_rate": 7.151521492430443e-06,
      "loss": 0.0025,
      "step": 131770
    },
    {
      "epoch": 0.8451959863131868,
      "grad_norm": 0.26300501823425293,
      "learning_rate": 7.151016243730316e-06,
      "loss": 0.0025,
      "step": 131780
    },
    {
      "epoch": 0.845260123206973,
      "grad_norm": 0.1211204007267952,
      "learning_rate": 7.150510968076419e-06,
      "loss": 0.005,
      "step": 131790
    },
    {
      "epoch": 0.845324260100759,
      "grad_norm": 0.12193798273801804,
      "learning_rate": 7.15000566547508e-06,
      "loss": 0.0034,
      "step": 131800
    },
    {
      "epoch": 0.8453883969945452,
      "grad_norm": 0.06321345269680023,
      "learning_rate": 7.149500335932636e-06,
      "loss": 0.0022,
      "step": 131810
    },
    {
      "epoch": 0.8454525338883313,
      "grad_norm": 0.48327797651290894,
      "learning_rate": 7.148994979455415e-06,
      "loss": 0.0026,
      "step": 131820
    },
    {
      "epoch": 0.8455166707821173,
      "grad_norm": 0.09827014058828354,
      "learning_rate": 7.1484895960497515e-06,
      "loss": 0.0018,
      "step": 131830
    },
    {
      "epoch": 0.8455808076759035,
      "grad_norm": 0.045801226049661636,
      "learning_rate": 7.1479841857219776e-06,
      "loss": 0.0018,
      "step": 131840
    },
    {
      "epoch": 0.8456449445696895,
      "grad_norm": 0.1387709081172943,
      "learning_rate": 7.147478748478427e-06,
      "loss": 0.0025,
      "step": 131850
    },
    {
      "epoch": 0.8457090814634757,
      "grad_norm": 0.09701228141784668,
      "learning_rate": 7.146973284325432e-06,
      "loss": 0.0034,
      "step": 131860
    },
    {
      "epoch": 0.8457732183572617,
      "grad_norm": 0.05086098238825798,
      "learning_rate": 7.146467793269329e-06,
      "loss": 0.0035,
      "step": 131870
    },
    {
      "epoch": 0.8458373552510479,
      "grad_norm": 0.07943065464496613,
      "learning_rate": 7.145962275316449e-06,
      "loss": 0.0031,
      "step": 131880
    },
    {
      "epoch": 0.8459014921448339,
      "grad_norm": 0.07321260869503021,
      "learning_rate": 7.145456730473129e-06,
      "loss": 0.003,
      "step": 131890
    },
    {
      "epoch": 0.84596562903862,
      "grad_norm": 0.1799236238002777,
      "learning_rate": 7.1449511587457035e-06,
      "loss": 0.0035,
      "step": 131900
    },
    {
      "epoch": 0.8460297659324061,
      "grad_norm": 0.13050727546215057,
      "learning_rate": 7.144445560140505e-06,
      "loss": 0.0036,
      "step": 131910
    },
    {
      "epoch": 0.8460939028261922,
      "grad_norm": 0.11861727386713028,
      "learning_rate": 7.143939934663873e-06,
      "loss": 0.0036,
      "step": 131920
    },
    {
      "epoch": 0.8461580397199783,
      "grad_norm": 0.04367021098732948,
      "learning_rate": 7.143434282322139e-06,
      "loss": 0.0017,
      "step": 131930
    },
    {
      "epoch": 0.8462221766137644,
      "grad_norm": 0.12094981223344803,
      "learning_rate": 7.142928603121644e-06,
      "loss": 0.0052,
      "step": 131940
    },
    {
      "epoch": 0.8462863135075506,
      "grad_norm": 0.16046714782714844,
      "learning_rate": 7.142422897068719e-06,
      "loss": 0.0026,
      "step": 131950
    },
    {
      "epoch": 0.8463504504013366,
      "grad_norm": 0.045945905148983,
      "learning_rate": 7.1419171641697075e-06,
      "loss": 0.0021,
      "step": 131960
    },
    {
      "epoch": 0.8464145872951228,
      "grad_norm": 0.13319644331932068,
      "learning_rate": 7.141411404430941e-06,
      "loss": 0.0016,
      "step": 131970
    },
    {
      "epoch": 0.8464787241889088,
      "grad_norm": 0.12061929702758789,
      "learning_rate": 7.14090561785876e-06,
      "loss": 0.0033,
      "step": 131980
    },
    {
      "epoch": 0.8465428610826949,
      "grad_norm": 0.09734722971916199,
      "learning_rate": 7.140399804459501e-06,
      "loss": 0.0021,
      "step": 131990
    },
    {
      "epoch": 0.846606997976481,
      "grad_norm": 0.06309890747070312,
      "learning_rate": 7.139893964239503e-06,
      "loss": 0.0027,
      "step": 132000
    },
    {
      "epoch": 0.8466711348702671,
      "grad_norm": 0.12926048040390015,
      "learning_rate": 7.139388097205104e-06,
      "loss": 0.002,
      "step": 132010
    },
    {
      "epoch": 0.8467352717640532,
      "grad_norm": 0.10321187227964401,
      "learning_rate": 7.138882203362645e-06,
      "loss": 0.0028,
      "step": 132020
    },
    {
      "epoch": 0.8467994086578393,
      "grad_norm": 0.15635032951831818,
      "learning_rate": 7.138376282718461e-06,
      "loss": 0.0016,
      "step": 132030
    },
    {
      "epoch": 0.8468635455516254,
      "grad_norm": 0.13198229670524597,
      "learning_rate": 7.137870335278896e-06,
      "loss": 0.003,
      "step": 132040
    },
    {
      "epoch": 0.8469276824454115,
      "grad_norm": 0.146623894572258,
      "learning_rate": 7.137364361050285e-06,
      "loss": 0.0032,
      "step": 132050
    },
    {
      "epoch": 0.8469918193391975,
      "grad_norm": 0.14922136068344116,
      "learning_rate": 7.136858360038973e-06,
      "loss": 0.0026,
      "step": 132060
    },
    {
      "epoch": 0.8470559562329837,
      "grad_norm": 0.14886027574539185,
      "learning_rate": 7.136352332251297e-06,
      "loss": 0.0017,
      "step": 132070
    },
    {
      "epoch": 0.8471200931267697,
      "grad_norm": 0.10785438120365143,
      "learning_rate": 7.135846277693602e-06,
      "loss": 0.0011,
      "step": 132080
    },
    {
      "epoch": 0.8471842300205559,
      "grad_norm": 0.13098682463169098,
      "learning_rate": 7.135340196372225e-06,
      "loss": 0.0027,
      "step": 132090
    },
    {
      "epoch": 0.847248366914342,
      "grad_norm": 0.11557871848344803,
      "learning_rate": 7.134834088293509e-06,
      "loss": 0.0034,
      "step": 132100
    },
    {
      "epoch": 0.8473125038081281,
      "grad_norm": 0.05154212936758995,
      "learning_rate": 7.134327953463797e-06,
      "loss": 0.0034,
      "step": 132110
    },
    {
      "epoch": 0.8473766407019142,
      "grad_norm": 0.3776550889015198,
      "learning_rate": 7.13382179188943e-06,
      "loss": 0.0021,
      "step": 132120
    },
    {
      "epoch": 0.8474407775957002,
      "grad_norm": 0.11731040477752686,
      "learning_rate": 7.1333156035767506e-06,
      "loss": 0.0032,
      "step": 132130
    },
    {
      "epoch": 0.8475049144894864,
      "grad_norm": 0.024094602093100548,
      "learning_rate": 7.132809388532101e-06,
      "loss": 0.0034,
      "step": 132140
    },
    {
      "epoch": 0.8475690513832724,
      "grad_norm": 0.15600042045116425,
      "learning_rate": 7.132303146761828e-06,
      "loss": 0.0016,
      "step": 132150
    },
    {
      "epoch": 0.8476331882770586,
      "grad_norm": 0.07214893400669098,
      "learning_rate": 7.13179687827227e-06,
      "loss": 0.0025,
      "step": 132160
    },
    {
      "epoch": 0.8476973251708446,
      "grad_norm": 0.18649162352085114,
      "learning_rate": 7.131290583069776e-06,
      "loss": 0.0021,
      "step": 132170
    },
    {
      "epoch": 0.8477614620646308,
      "grad_norm": 0.01331428810954094,
      "learning_rate": 7.1307842611606855e-06,
      "loss": 0.0015,
      "step": 132180
    },
    {
      "epoch": 0.8478255989584168,
      "grad_norm": 0.21441741287708282,
      "learning_rate": 7.130277912551348e-06,
      "loss": 0.0015,
      "step": 132190
    },
    {
      "epoch": 0.847889735852203,
      "grad_norm": 0.04190351814031601,
      "learning_rate": 7.129771537248104e-06,
      "loss": 0.0056,
      "step": 132200
    },
    {
      "epoch": 0.847953872745989,
      "grad_norm": 0.012922914698719978,
      "learning_rate": 7.1292651352573014e-06,
      "loss": 0.0022,
      "step": 132210
    },
    {
      "epoch": 0.8480180096397751,
      "grad_norm": 0.12478888034820557,
      "learning_rate": 7.128758706585284e-06,
      "loss": 0.0018,
      "step": 132220
    },
    {
      "epoch": 0.8480821465335613,
      "grad_norm": 0.10566743463277817,
      "learning_rate": 7.128252251238399e-06,
      "loss": 0.0045,
      "step": 132230
    },
    {
      "epoch": 0.8481462834273473,
      "grad_norm": 0.1309993416070938,
      "learning_rate": 7.127745769222992e-06,
      "loss": 0.0026,
      "step": 132240
    },
    {
      "epoch": 0.8482104203211335,
      "grad_norm": 0.07731198519468307,
      "learning_rate": 7.12723926054541e-06,
      "loss": 0.0043,
      "step": 132250
    },
    {
      "epoch": 0.8482745572149195,
      "grad_norm": 0.020769502967596054,
      "learning_rate": 7.126732725212e-06,
      "loss": 0.003,
      "step": 132260
    },
    {
      "epoch": 0.8483386941087057,
      "grad_norm": 0.09295207262039185,
      "learning_rate": 7.126226163229109e-06,
      "loss": 0.002,
      "step": 132270
    },
    {
      "epoch": 0.8484028310024917,
      "grad_norm": 0.0527043417096138,
      "learning_rate": 7.1257195746030835e-06,
      "loss": 0.0019,
      "step": 132280
    },
    {
      "epoch": 0.8484669678962778,
      "grad_norm": 0.09591345489025116,
      "learning_rate": 7.125212959340273e-06,
      "loss": 0.0039,
      "step": 132290
    },
    {
      "epoch": 0.8485311047900639,
      "grad_norm": 0.24664068222045898,
      "learning_rate": 7.124706317447026e-06,
      "loss": 0.0024,
      "step": 132300
    },
    {
      "epoch": 0.84859524168385,
      "grad_norm": 0.015033972449600697,
      "learning_rate": 7.1241996489296906e-06,
      "loss": 0.002,
      "step": 132310
    },
    {
      "epoch": 0.8486593785776361,
      "grad_norm": 0.04619878903031349,
      "learning_rate": 7.1236929537946146e-06,
      "loss": 0.0015,
      "step": 132320
    },
    {
      "epoch": 0.8487235154714222,
      "grad_norm": 0.09617872536182404,
      "learning_rate": 7.123186232048147e-06,
      "loss": 0.0019,
      "step": 132330
    },
    {
      "epoch": 0.8487876523652083,
      "grad_norm": 0.09924177825450897,
      "learning_rate": 7.12267948369664e-06,
      "loss": 0.0014,
      "step": 132340
    },
    {
      "epoch": 0.8488517892589944,
      "grad_norm": 0.09445368498563766,
      "learning_rate": 7.122172708746442e-06,
      "loss": 0.0013,
      "step": 132350
    },
    {
      "epoch": 0.8489159261527804,
      "grad_norm": 0.4284706115722656,
      "learning_rate": 7.121665907203903e-06,
      "loss": 0.0063,
      "step": 132360
    },
    {
      "epoch": 0.8489800630465666,
      "grad_norm": 0.2050025314092636,
      "learning_rate": 7.121159079075374e-06,
      "loss": 0.0033,
      "step": 132370
    },
    {
      "epoch": 0.8490441999403527,
      "grad_norm": 0.15669779479503632,
      "learning_rate": 7.120652224367206e-06,
      "loss": 0.0032,
      "step": 132380
    },
    {
      "epoch": 0.8491083368341388,
      "grad_norm": 0.015262565575540066,
      "learning_rate": 7.120145343085749e-06,
      "loss": 0.001,
      "step": 132390
    },
    {
      "epoch": 0.8491724737279249,
      "grad_norm": 0.18097631633281708,
      "learning_rate": 7.1196384352373574e-06,
      "loss": 0.002,
      "step": 132400
    },
    {
      "epoch": 0.849236610621711,
      "grad_norm": 0.1737731248140335,
      "learning_rate": 7.11913150082838e-06,
      "loss": 0.0028,
      "step": 132410
    },
    {
      "epoch": 0.8493007475154971,
      "grad_norm": 0.1534372866153717,
      "learning_rate": 7.118624539865171e-06,
      "loss": 0.0021,
      "step": 132420
    },
    {
      "epoch": 0.8493648844092831,
      "grad_norm": 0.1141214445233345,
      "learning_rate": 7.118117552354082e-06,
      "loss": 0.0018,
      "step": 132430
    },
    {
      "epoch": 0.8494290213030693,
      "grad_norm": 0.07327088713645935,
      "learning_rate": 7.117610538301465e-06,
      "loss": 0.0031,
      "step": 132440
    },
    {
      "epoch": 0.8494931581968553,
      "grad_norm": 0.0789964497089386,
      "learning_rate": 7.117103497713676e-06,
      "loss": 0.0018,
      "step": 132450
    },
    {
      "epoch": 0.8495572950906415,
      "grad_norm": 0.11921041458845139,
      "learning_rate": 7.116596430597067e-06,
      "loss": 0.003,
      "step": 132460
    },
    {
      "epoch": 0.8496214319844275,
      "grad_norm": 0.04953978583216667,
      "learning_rate": 7.116089336957992e-06,
      "loss": 0.0017,
      "step": 132470
    },
    {
      "epoch": 0.8496855688782137,
      "grad_norm": 0.2878396213054657,
      "learning_rate": 7.115582216802805e-06,
      "loss": 0.0016,
      "step": 132480
    },
    {
      "epoch": 0.8497497057719997,
      "grad_norm": 0.05667930841445923,
      "learning_rate": 7.115075070137862e-06,
      "loss": 0.0032,
      "step": 132490
    },
    {
      "epoch": 0.8498138426657859,
      "grad_norm": 0.0869360864162445,
      "learning_rate": 7.114567896969516e-06,
      "loss": 0.0049,
      "step": 132500
    },
    {
      "epoch": 0.849877979559572,
      "grad_norm": 0.40160295367240906,
      "learning_rate": 7.1140606973041215e-06,
      "loss": 0.0031,
      "step": 132510
    },
    {
      "epoch": 0.849942116453358,
      "grad_norm": 0.09188877791166306,
      "learning_rate": 7.113553471148037e-06,
      "loss": 0.0017,
      "step": 132520
    },
    {
      "epoch": 0.8500062533471442,
      "grad_norm": 0.10770701617002487,
      "learning_rate": 7.113046218507618e-06,
      "loss": 0.0027,
      "step": 132530
    },
    {
      "epoch": 0.8500703902409302,
      "grad_norm": 0.06739377230405807,
      "learning_rate": 7.1125389393892176e-06,
      "loss": 0.003,
      "step": 132540
    },
    {
      "epoch": 0.8501345271347164,
      "grad_norm": 0.1430237740278244,
      "learning_rate": 7.112031633799196e-06,
      "loss": 0.0027,
      "step": 132550
    },
    {
      "epoch": 0.8501986640285024,
      "grad_norm": 0.07702884823083878,
      "learning_rate": 7.111524301743907e-06,
      "loss": 0.0032,
      "step": 132560
    },
    {
      "epoch": 0.8502628009222886,
      "grad_norm": 0.01654994674026966,
      "learning_rate": 7.1110169432297114e-06,
      "loss": 0.002,
      "step": 132570
    },
    {
      "epoch": 0.8503269378160746,
      "grad_norm": 0.12832903861999512,
      "learning_rate": 7.110509558262963e-06,
      "loss": 0.002,
      "step": 132580
    },
    {
      "epoch": 0.8503910747098608,
      "grad_norm": 0.0472089983522892,
      "learning_rate": 7.110002146850021e-06,
      "loss": 0.0032,
      "step": 132590
    },
    {
      "epoch": 0.8504552116036468,
      "grad_norm": 0.3265191912651062,
      "learning_rate": 7.109494708997247e-06,
      "loss": 0.0039,
      "step": 132600
    },
    {
      "epoch": 0.8505193484974329,
      "grad_norm": 0.12164904177188873,
      "learning_rate": 7.108987244710994e-06,
      "loss": 0.0031,
      "step": 132610
    },
    {
      "epoch": 0.850583485391219,
      "grad_norm": 0.09468712657690048,
      "learning_rate": 7.108479753997626e-06,
      "loss": 0.0025,
      "step": 132620
    },
    {
      "epoch": 0.8506476222850051,
      "grad_norm": 0.1038513034582138,
      "learning_rate": 7.107972236863498e-06,
      "loss": 0.002,
      "step": 132630
    },
    {
      "epoch": 0.8507117591787912,
      "grad_norm": 0.09218443930149078,
      "learning_rate": 7.107464693314972e-06,
      "loss": 0.0017,
      "step": 132640
    },
    {
      "epoch": 0.8507758960725773,
      "grad_norm": 0.07426737248897552,
      "learning_rate": 7.106957123358405e-06,
      "loss": 0.0029,
      "step": 132650
    },
    {
      "epoch": 0.8508400329663635,
      "grad_norm": 0.057585205882787704,
      "learning_rate": 7.106449527000162e-06,
      "loss": 0.0043,
      "step": 132660
    },
    {
      "epoch": 0.8509041698601495,
      "grad_norm": 0.12777933478355408,
      "learning_rate": 7.1059419042466005e-06,
      "loss": 0.0016,
      "step": 132670
    },
    {
      "epoch": 0.8509683067539356,
      "grad_norm": 0.11784519255161285,
      "learning_rate": 7.105434255104083e-06,
      "loss": 0.0027,
      "step": 132680
    },
    {
      "epoch": 0.8510324436477217,
      "grad_norm": 0.13504132628440857,
      "learning_rate": 7.104926579578967e-06,
      "loss": 0.0027,
      "step": 132690
    },
    {
      "epoch": 0.8510965805415078,
      "grad_norm": 0.0821489617228508,
      "learning_rate": 7.104418877677618e-06,
      "loss": 0.0029,
      "step": 132700
    },
    {
      "epoch": 0.8511607174352939,
      "grad_norm": 0.19582541286945343,
      "learning_rate": 7.103911149406395e-06,
      "loss": 0.0018,
      "step": 132710
    },
    {
      "epoch": 0.85122485432908,
      "grad_norm": 0.10627574473619461,
      "learning_rate": 7.103403394771663e-06,
      "loss": 0.002,
      "step": 132720
    },
    {
      "epoch": 0.851288991222866,
      "grad_norm": 0.06380230188369751,
      "learning_rate": 7.102895613779782e-06,
      "loss": 0.0026,
      "step": 132730
    },
    {
      "epoch": 0.8513531281166522,
      "grad_norm": 0.2278783768415451,
      "learning_rate": 7.102387806437119e-06,
      "loss": 0.006,
      "step": 132740
    },
    {
      "epoch": 0.8514172650104382,
      "grad_norm": 1.431296706199646,
      "learning_rate": 7.101879972750031e-06,
      "loss": 0.0019,
      "step": 132750
    },
    {
      "epoch": 0.8514814019042244,
      "grad_norm": 0.09681957960128784,
      "learning_rate": 7.1013721127248865e-06,
      "loss": 0.0024,
      "step": 132760
    },
    {
      "epoch": 0.8515455387980104,
      "grad_norm": 0.14589335024356842,
      "learning_rate": 7.100864226368047e-06,
      "loss": 0.0021,
      "step": 132770
    },
    {
      "epoch": 0.8516096756917966,
      "grad_norm": 0.021403413265943527,
      "learning_rate": 7.100356313685877e-06,
      "loss": 0.0016,
      "step": 132780
    },
    {
      "epoch": 0.8516738125855826,
      "grad_norm": 0.047523729503154755,
      "learning_rate": 7.099848374684743e-06,
      "loss": 0.004,
      "step": 132790
    },
    {
      "epoch": 0.8517379494793688,
      "grad_norm": 0.12544597685337067,
      "learning_rate": 7.099340409371005e-06,
      "loss": 0.0022,
      "step": 132800
    },
    {
      "epoch": 0.8518020863731549,
      "grad_norm": 0.07675661146640778,
      "learning_rate": 7.0988324177510335e-06,
      "loss": 0.0017,
      "step": 132810
    },
    {
      "epoch": 0.851866223266941,
      "grad_norm": 0.049117445945739746,
      "learning_rate": 7.098324399831191e-06,
      "loss": 0.0023,
      "step": 132820
    },
    {
      "epoch": 0.8519303601607271,
      "grad_norm": 0.1694405972957611,
      "learning_rate": 7.0978163556178455e-06,
      "loss": 0.0024,
      "step": 132830
    },
    {
      "epoch": 0.8519944970545131,
      "grad_norm": 0.043329883366823196,
      "learning_rate": 7.09730828511736e-06,
      "loss": 0.0017,
      "step": 132840
    },
    {
      "epoch": 0.8520586339482993,
      "grad_norm": 0.12720754742622375,
      "learning_rate": 7.096800188336105e-06,
      "loss": 0.0016,
      "step": 132850
    },
    {
      "epoch": 0.8521227708420853,
      "grad_norm": 0.25648581981658936,
      "learning_rate": 7.096292065280444e-06,
      "loss": 0.005,
      "step": 132860
    },
    {
      "epoch": 0.8521869077358715,
      "grad_norm": 0.11244427412748337,
      "learning_rate": 7.095783915956744e-06,
      "loss": 0.0015,
      "step": 132870
    },
    {
      "epoch": 0.8522510446296575,
      "grad_norm": 0.09432416409254074,
      "learning_rate": 7.095275740371375e-06,
      "loss": 0.0015,
      "step": 132880
    },
    {
      "epoch": 0.8523151815234437,
      "grad_norm": 0.1023477166891098,
      "learning_rate": 7.094767538530703e-06,
      "loss": 0.0026,
      "step": 132890
    },
    {
      "epoch": 0.8523793184172297,
      "grad_norm": 0.08102133870124817,
      "learning_rate": 7.094259310441096e-06,
      "loss": 0.0033,
      "step": 132900
    },
    {
      "epoch": 0.8524434553110158,
      "grad_norm": 0.011946534737944603,
      "learning_rate": 7.093751056108925e-06,
      "loss": 0.0045,
      "step": 132910
    },
    {
      "epoch": 0.8525075922048019,
      "grad_norm": 0.14268122613430023,
      "learning_rate": 7.093242775540555e-06,
      "loss": 0.0022,
      "step": 132920
    },
    {
      "epoch": 0.852571729098588,
      "grad_norm": 0.045782607048749924,
      "learning_rate": 7.092734468742358e-06,
      "loss": 0.0026,
      "step": 132930
    },
    {
      "epoch": 0.8526358659923742,
      "grad_norm": 0.21428555250167847,
      "learning_rate": 7.092226135720702e-06,
      "loss": 0.002,
      "step": 132940
    },
    {
      "epoch": 0.8527000028861602,
      "grad_norm": 0.054481763392686844,
      "learning_rate": 7.091717776481957e-06,
      "loss": 0.0009,
      "step": 132950
    },
    {
      "epoch": 0.8527641397799464,
      "grad_norm": 0.06863247603178024,
      "learning_rate": 7.0912093910324946e-06,
      "loss": 0.0028,
      "step": 132960
    },
    {
      "epoch": 0.8528282766737324,
      "grad_norm": 0.01814761385321617,
      "learning_rate": 7.090700979378682e-06,
      "loss": 0.0027,
      "step": 132970
    },
    {
      "epoch": 0.8528924135675185,
      "grad_norm": 0.04927373677492142,
      "learning_rate": 7.0901925415268946e-06,
      "loss": 0.0021,
      "step": 132980
    },
    {
      "epoch": 0.8529565504613046,
      "grad_norm": 0.2550349831581116,
      "learning_rate": 7.089684077483499e-06,
      "loss": 0.0023,
      "step": 132990
    },
    {
      "epoch": 0.8530206873550907,
      "grad_norm": 0.12768761813640594,
      "learning_rate": 7.089175587254868e-06,
      "loss": 0.0038,
      "step": 133000
    },
    {
      "epoch": 0.8530848242488768,
      "grad_norm": 0.06169166415929794,
      "learning_rate": 7.088667070847375e-06,
      "loss": 0.0021,
      "step": 133010
    },
    {
      "epoch": 0.8531489611426629,
      "grad_norm": 0.09823588281869888,
      "learning_rate": 7.08815852826739e-06,
      "loss": 0.0046,
      "step": 133020
    },
    {
      "epoch": 0.853213098036449,
      "grad_norm": 0.025031019002199173,
      "learning_rate": 7.087649959521286e-06,
      "loss": 0.004,
      "step": 133030
    },
    {
      "epoch": 0.8532772349302351,
      "grad_norm": 0.02685212716460228,
      "learning_rate": 7.087141364615437e-06,
      "loss": 0.0026,
      "step": 133040
    },
    {
      "epoch": 0.8533413718240211,
      "grad_norm": 0.2141588181257248,
      "learning_rate": 7.086632743556214e-06,
      "loss": 0.0019,
      "step": 133050
    },
    {
      "epoch": 0.8534055087178073,
      "grad_norm": 0.23564761877059937,
      "learning_rate": 7.086124096349993e-06,
      "loss": 0.0025,
      "step": 133060
    },
    {
      "epoch": 0.8534696456115933,
      "grad_norm": 0.23499944806098938,
      "learning_rate": 7.085615423003145e-06,
      "loss": 0.0023,
      "step": 133070
    },
    {
      "epoch": 0.8535337825053795,
      "grad_norm": 0.12187331169843674,
      "learning_rate": 7.085106723522046e-06,
      "loss": 0.0046,
      "step": 133080
    },
    {
      "epoch": 0.8535979193991656,
      "grad_norm": 0.0823151245713234,
      "learning_rate": 7.084597997913069e-06,
      "loss": 0.0044,
      "step": 133090
    },
    {
      "epoch": 0.8536620562929517,
      "grad_norm": 0.09182952344417572,
      "learning_rate": 7.084089246182588e-06,
      "loss": 0.0012,
      "step": 133100
    },
    {
      "epoch": 0.8537261931867378,
      "grad_norm": 0.021208835765719414,
      "learning_rate": 7.08358046833698e-06,
      "loss": 0.0015,
      "step": 133110
    },
    {
      "epoch": 0.8537903300805239,
      "grad_norm": 0.2756306231021881,
      "learning_rate": 7.0830716643826206e-06,
      "loss": 0.0023,
      "step": 133120
    },
    {
      "epoch": 0.85385446697431,
      "grad_norm": 0.16646911203861237,
      "learning_rate": 7.0825628343258835e-06,
      "loss": 0.005,
      "step": 133130
    },
    {
      "epoch": 0.853918603868096,
      "grad_norm": 0.07808910310268402,
      "learning_rate": 7.082053978173147e-06,
      "loss": 0.0015,
      "step": 133140
    },
    {
      "epoch": 0.8539827407618822,
      "grad_norm": 0.10818828642368317,
      "learning_rate": 7.081545095930784e-06,
      "loss": 0.0014,
      "step": 133150
    },
    {
      "epoch": 0.8540468776556682,
      "grad_norm": 0.05248650163412094,
      "learning_rate": 7.081036187605175e-06,
      "loss": 0.0013,
      "step": 133160
    },
    {
      "epoch": 0.8541110145494544,
      "grad_norm": 0.046363357454538345,
      "learning_rate": 7.080527253202695e-06,
      "loss": 0.003,
      "step": 133170
    },
    {
      "epoch": 0.8541751514432404,
      "grad_norm": 0.22602427005767822,
      "learning_rate": 7.080018292729721e-06,
      "loss": 0.0014,
      "step": 133180
    },
    {
      "epoch": 0.8542392883370266,
      "grad_norm": 0.12549887597560883,
      "learning_rate": 7.0795093061926325e-06,
      "loss": 0.0015,
      "step": 133190
    },
    {
      "epoch": 0.8543034252308126,
      "grad_norm": 0.15486212074756622,
      "learning_rate": 7.079000293597804e-06,
      "loss": 0.0025,
      "step": 133200
    },
    {
      "epoch": 0.8543675621245987,
      "grad_norm": 0.11368501931428909,
      "learning_rate": 7.0784912549516185e-06,
      "loss": 0.003,
      "step": 133210
    },
    {
      "epoch": 0.8544316990183849,
      "grad_norm": 0.12644527852535248,
      "learning_rate": 7.077982190260451e-06,
      "loss": 0.0048,
      "step": 133220
    },
    {
      "epoch": 0.8544958359121709,
      "grad_norm": 0.10898713022470474,
      "learning_rate": 7.077473099530681e-06,
      "loss": 0.0024,
      "step": 133230
    },
    {
      "epoch": 0.8545599728059571,
      "grad_norm": 0.10378258675336838,
      "learning_rate": 7.0769639827686885e-06,
      "loss": 0.0028,
      "step": 133240
    },
    {
      "epoch": 0.8546241096997431,
      "grad_norm": 0.11458373069763184,
      "learning_rate": 7.076454839980854e-06,
      "loss": 0.0015,
      "step": 133250
    },
    {
      "epoch": 0.8546882465935293,
      "grad_norm": 0.14008843898773193,
      "learning_rate": 7.075945671173555e-06,
      "loss": 0.0039,
      "step": 133260
    },
    {
      "epoch": 0.8547523834873153,
      "grad_norm": 0.14388789236545563,
      "learning_rate": 7.0754364763531744e-06,
      "loss": 0.0018,
      "step": 133270
    },
    {
      "epoch": 0.8548165203811015,
      "grad_norm": 0.030500268563628197,
      "learning_rate": 7.07492725552609e-06,
      "loss": 0.003,
      "step": 133280
    },
    {
      "epoch": 0.8548806572748875,
      "grad_norm": 0.13390643894672394,
      "learning_rate": 7.074418008698685e-06,
      "loss": 0.003,
      "step": 133290
    },
    {
      "epoch": 0.8549447941686736,
      "grad_norm": 0.06771231442689896,
      "learning_rate": 7.073908735877339e-06,
      "loss": 0.0013,
      "step": 133300
    },
    {
      "epoch": 0.8550089310624597,
      "grad_norm": 0.11549648642539978,
      "learning_rate": 7.0733994370684355e-06,
      "loss": 0.0021,
      "step": 133310
    },
    {
      "epoch": 0.8550730679562458,
      "grad_norm": 0.11787834763526917,
      "learning_rate": 7.072890112278355e-06,
      "loss": 0.004,
      "step": 133320
    },
    {
      "epoch": 0.8551372048500319,
      "grad_norm": 0.13039426505565643,
      "learning_rate": 7.072380761513478e-06,
      "loss": 0.0018,
      "step": 133330
    },
    {
      "epoch": 0.855201341743818,
      "grad_norm": 0.04777880758047104,
      "learning_rate": 7.071871384780191e-06,
      "loss": 0.0028,
      "step": 133340
    },
    {
      "epoch": 0.855265478637604,
      "grad_norm": 0.033508896827697754,
      "learning_rate": 7.0713619820848745e-06,
      "loss": 0.0009,
      "step": 133350
    },
    {
      "epoch": 0.8553296155313902,
      "grad_norm": 0.33922964334487915,
      "learning_rate": 7.070852553433913e-06,
      "loss": 0.0025,
      "step": 133360
    },
    {
      "epoch": 0.8553937524251763,
      "grad_norm": 0.0977553054690361,
      "learning_rate": 7.070343098833687e-06,
      "loss": 0.0014,
      "step": 133370
    },
    {
      "epoch": 0.8554578893189624,
      "grad_norm": 0.1850329041481018,
      "learning_rate": 7.069833618290583e-06,
      "loss": 0.0023,
      "step": 133380
    },
    {
      "epoch": 0.8555220262127485,
      "grad_norm": 0.07582248747348785,
      "learning_rate": 7.069324111810984e-06,
      "loss": 0.0024,
      "step": 133390
    },
    {
      "epoch": 0.8555861631065346,
      "grad_norm": 0.06357874721288681,
      "learning_rate": 7.068814579401277e-06,
      "loss": 0.0025,
      "step": 133400
    },
    {
      "epoch": 0.8556503000003207,
      "grad_norm": 0.1265944540500641,
      "learning_rate": 7.068305021067843e-06,
      "loss": 0.0024,
      "step": 133410
    },
    {
      "epoch": 0.8557144368941068,
      "grad_norm": 0.04861431568861008,
      "learning_rate": 7.0677954368170694e-06,
      "loss": 0.0019,
      "step": 133420
    },
    {
      "epoch": 0.8557785737878929,
      "grad_norm": 0.2114597111940384,
      "learning_rate": 7.067285826655341e-06,
      "loss": 0.003,
      "step": 133430
    },
    {
      "epoch": 0.8558427106816789,
      "grad_norm": 0.20950160920619965,
      "learning_rate": 7.066776190589043e-06,
      "loss": 0.0032,
      "step": 133440
    },
    {
      "epoch": 0.8559068475754651,
      "grad_norm": 0.0692938044667244,
      "learning_rate": 7.066266528624563e-06,
      "loss": 0.0013,
      "step": 133450
    },
    {
      "epoch": 0.8559709844692511,
      "grad_norm": 0.037902288138866425,
      "learning_rate": 7.065756840768286e-06,
      "loss": 0.0022,
      "step": 133460
    },
    {
      "epoch": 0.8560351213630373,
      "grad_norm": 0.13756364583969116,
      "learning_rate": 7.065247127026601e-06,
      "loss": 0.0061,
      "step": 133470
    },
    {
      "epoch": 0.8560992582568233,
      "grad_norm": 0.056961458176374435,
      "learning_rate": 7.064737387405892e-06,
      "loss": 0.0021,
      "step": 133480
    },
    {
      "epoch": 0.8561633951506095,
      "grad_norm": 0.07422365248203278,
      "learning_rate": 7.064227621912549e-06,
      "loss": 0.0037,
      "step": 133490
    },
    {
      "epoch": 0.8562275320443956,
      "grad_norm": 0.21851889789104462,
      "learning_rate": 7.063717830552956e-06,
      "loss": 0.003,
      "step": 133500
    },
    {
      "epoch": 0.8562916689381816,
      "grad_norm": 0.09047169238328934,
      "learning_rate": 7.063208013333507e-06,
      "loss": 0.0039,
      "step": 133510
    },
    {
      "epoch": 0.8563558058319678,
      "grad_norm": 0.13710708916187286,
      "learning_rate": 7.062698170260585e-06,
      "loss": 0.0022,
      "step": 133520
    },
    {
      "epoch": 0.8564199427257538,
      "grad_norm": 0.15213985741138458,
      "learning_rate": 7.062188301340582e-06,
      "loss": 0.0108,
      "step": 133530
    },
    {
      "epoch": 0.85648407961954,
      "grad_norm": 0.2374606877565384,
      "learning_rate": 7.061678406579885e-06,
      "loss": 0.0046,
      "step": 133540
    },
    {
      "epoch": 0.856548216513326,
      "grad_norm": 0.08374125510454178,
      "learning_rate": 7.061168485984885e-06,
      "loss": 0.0031,
      "step": 133550
    },
    {
      "epoch": 0.8566123534071122,
      "grad_norm": 0.13061614334583282,
      "learning_rate": 7.060658539561969e-06,
      "loss": 0.0018,
      "step": 133560
    },
    {
      "epoch": 0.8566764903008982,
      "grad_norm": 0.030887693166732788,
      "learning_rate": 7.060148567317531e-06,
      "loss": 0.0029,
      "step": 133570
    },
    {
      "epoch": 0.8567406271946844,
      "grad_norm": 0.19017326831817627,
      "learning_rate": 7.059638569257957e-06,
      "loss": 0.0034,
      "step": 133580
    },
    {
      "epoch": 0.8568047640884704,
      "grad_norm": 0.23443512618541718,
      "learning_rate": 7.0591285453896415e-06,
      "loss": 0.0029,
      "step": 133590
    },
    {
      "epoch": 0.8568689009822565,
      "grad_norm": 0.14593902230262756,
      "learning_rate": 7.058618495718972e-06,
      "loss": 0.0014,
      "step": 133600
    },
    {
      "epoch": 0.8569330378760426,
      "grad_norm": 0.11629689484834671,
      "learning_rate": 7.058108420252343e-06,
      "loss": 0.0026,
      "step": 133610
    },
    {
      "epoch": 0.8569971747698287,
      "grad_norm": 0.17862918972969055,
      "learning_rate": 7.057598318996144e-06,
      "loss": 0.0022,
      "step": 133620
    },
    {
      "epoch": 0.8570613116636148,
      "grad_norm": 0.07450482994318008,
      "learning_rate": 7.057088191956767e-06,
      "loss": 0.0024,
      "step": 133630
    },
    {
      "epoch": 0.8571254485574009,
      "grad_norm": 0.02601449377834797,
      "learning_rate": 7.056578039140605e-06,
      "loss": 0.0039,
      "step": 133640
    },
    {
      "epoch": 0.8571895854511871,
      "grad_norm": 0.3861764371395111,
      "learning_rate": 7.05606786055405e-06,
      "loss": 0.003,
      "step": 133650
    },
    {
      "epoch": 0.8572537223449731,
      "grad_norm": 0.042525045573711395,
      "learning_rate": 7.055557656203497e-06,
      "loss": 0.0022,
      "step": 133660
    },
    {
      "epoch": 0.8573178592387593,
      "grad_norm": 0.12494376301765442,
      "learning_rate": 7.055047426095336e-06,
      "loss": 0.0031,
      "step": 133670
    },
    {
      "epoch": 0.8573819961325453,
      "grad_norm": 0.06158117204904556,
      "learning_rate": 7.054537170235962e-06,
      "loss": 0.0017,
      "step": 133680
    },
    {
      "epoch": 0.8574461330263314,
      "grad_norm": 0.12451702356338501,
      "learning_rate": 7.054026888631769e-06,
      "loss": 0.0029,
      "step": 133690
    },
    {
      "epoch": 0.8575102699201175,
      "grad_norm": 0.15168841183185577,
      "learning_rate": 7.053516581289153e-06,
      "loss": 0.004,
      "step": 133700
    },
    {
      "epoch": 0.8575744068139036,
      "grad_norm": 0.10787881910800934,
      "learning_rate": 7.053006248214503e-06,
      "loss": 0.0043,
      "step": 133710
    },
    {
      "epoch": 0.8576385437076897,
      "grad_norm": 0.1075955405831337,
      "learning_rate": 7.052495889414221e-06,
      "loss": 0.0045,
      "step": 133720
    },
    {
      "epoch": 0.8577026806014758,
      "grad_norm": 0.05712176486849785,
      "learning_rate": 7.051985504894696e-06,
      "loss": 0.0032,
      "step": 133730
    },
    {
      "epoch": 0.8577668174952618,
      "grad_norm": 0.15225821733474731,
      "learning_rate": 7.051475094662328e-06,
      "loss": 0.0021,
      "step": 133740
    },
    {
      "epoch": 0.857830954389048,
      "grad_norm": 0.37607520818710327,
      "learning_rate": 7.0509646587235095e-06,
      "loss": 0.0029,
      "step": 133750
    },
    {
      "epoch": 0.857895091282834,
      "grad_norm": 0.07383402436971664,
      "learning_rate": 7.050454197084638e-06,
      "loss": 0.0058,
      "step": 133760
    },
    {
      "epoch": 0.8579592281766202,
      "grad_norm": 0.0801803320646286,
      "learning_rate": 7.04994370975211e-06,
      "loss": 0.0023,
      "step": 133770
    },
    {
      "epoch": 0.8580233650704063,
      "grad_norm": 0.06811663508415222,
      "learning_rate": 7.049433196732324e-06,
      "loss": 0.0027,
      "step": 133780
    },
    {
      "epoch": 0.8580875019641924,
      "grad_norm": 0.06166665256023407,
      "learning_rate": 7.048922658031674e-06,
      "loss": 0.0018,
      "step": 133790
    },
    {
      "epoch": 0.8581516388579785,
      "grad_norm": 0.045509107410907745,
      "learning_rate": 7.048412093656558e-06,
      "loss": 0.0022,
      "step": 133800
    },
    {
      "epoch": 0.8582157757517646,
      "grad_norm": 0.2717222273349762,
      "learning_rate": 7.0479015036133755e-06,
      "loss": 0.0037,
      "step": 133810
    },
    {
      "epoch": 0.8582799126455507,
      "grad_norm": 0.23772364854812622,
      "learning_rate": 7.047390887908523e-06,
      "loss": 0.0029,
      "step": 133820
    },
    {
      "epoch": 0.8583440495393367,
      "grad_norm": 0.1597808301448822,
      "learning_rate": 7.046880246548401e-06,
      "loss": 0.0047,
      "step": 133830
    },
    {
      "epoch": 0.8584081864331229,
      "grad_norm": 0.05817929282784462,
      "learning_rate": 7.046369579539405e-06,
      "loss": 0.0032,
      "step": 133840
    },
    {
      "epoch": 0.8584723233269089,
      "grad_norm": 0.08475963771343231,
      "learning_rate": 7.045858886887936e-06,
      "loss": 0.0034,
      "step": 133850
    },
    {
      "epoch": 0.8585364602206951,
      "grad_norm": 0.011824116110801697,
      "learning_rate": 7.0453481686003926e-06,
      "loss": 0.0024,
      "step": 133860
    },
    {
      "epoch": 0.8586005971144811,
      "grad_norm": 0.14034906029701233,
      "learning_rate": 7.044837424683175e-06,
      "loss": 0.0023,
      "step": 133870
    },
    {
      "epoch": 0.8586647340082673,
      "grad_norm": 0.1342957615852356,
      "learning_rate": 7.044326655142682e-06,
      "loss": 0.0044,
      "step": 133880
    },
    {
      "epoch": 0.8587288709020533,
      "grad_norm": 0.2175799161195755,
      "learning_rate": 7.043815859985318e-06,
      "loss": 0.0034,
      "step": 133890
    },
    {
      "epoch": 0.8587930077958394,
      "grad_norm": 0.2881358563899994,
      "learning_rate": 7.043305039217478e-06,
      "loss": 0.0015,
      "step": 133900
    },
    {
      "epoch": 0.8588571446896255,
      "grad_norm": 0.14161698520183563,
      "learning_rate": 7.0427941928455666e-06,
      "loss": 0.0019,
      "step": 133910
    },
    {
      "epoch": 0.8589212815834116,
      "grad_norm": 0.11514313519001007,
      "learning_rate": 7.0422833208759845e-06,
      "loss": 0.0016,
      "step": 133920
    },
    {
      "epoch": 0.8589854184771978,
      "grad_norm": 0.026508232578635216,
      "learning_rate": 7.0417724233151315e-06,
      "loss": 0.0035,
      "step": 133930
    },
    {
      "epoch": 0.8590495553709838,
      "grad_norm": 0.052615657448768616,
      "learning_rate": 7.041261500169412e-06,
      "loss": 0.0026,
      "step": 133940
    },
    {
      "epoch": 0.85911369226477,
      "grad_norm": 0.295786052942276,
      "learning_rate": 7.040750551445227e-06,
      "loss": 0.0036,
      "step": 133950
    },
    {
      "epoch": 0.859177829158556,
      "grad_norm": 0.09007281064987183,
      "learning_rate": 7.040239577148978e-06,
      "loss": 0.0019,
      "step": 133960
    },
    {
      "epoch": 0.8592419660523422,
      "grad_norm": 0.06715760380029678,
      "learning_rate": 7.039728577287069e-06,
      "loss": 0.0035,
      "step": 133970
    },
    {
      "epoch": 0.8593061029461282,
      "grad_norm": 0.2409805804491043,
      "learning_rate": 7.039217551865904e-06,
      "loss": 0.0021,
      "step": 133980
    },
    {
      "epoch": 0.8593702398399143,
      "grad_norm": 0.06879192590713501,
      "learning_rate": 7.038706500891885e-06,
      "loss": 0.0027,
      "step": 133990
    },
    {
      "epoch": 0.8594343767337004,
      "grad_norm": 0.18605777621269226,
      "learning_rate": 7.0381954243714165e-06,
      "loss": 0.0026,
      "step": 134000
    },
    {
      "epoch": 0.8594985136274865,
      "grad_norm": 0.14286191761493683,
      "learning_rate": 7.037684322310903e-06,
      "loss": 0.002,
      "step": 134010
    },
    {
      "epoch": 0.8595626505212726,
      "grad_norm": 0.036439698189496994,
      "learning_rate": 7.037173194716748e-06,
      "loss": 0.0025,
      "step": 134020
    },
    {
      "epoch": 0.8596267874150587,
      "grad_norm": 0.11563778668642044,
      "learning_rate": 7.036662041595358e-06,
      "loss": 0.0018,
      "step": 134030
    },
    {
      "epoch": 0.8596909243088448,
      "grad_norm": 0.13866159319877625,
      "learning_rate": 7.036150862953137e-06,
      "loss": 0.0023,
      "step": 134040
    },
    {
      "epoch": 0.8597550612026309,
      "grad_norm": 0.27767258882522583,
      "learning_rate": 7.03563965879649e-06,
      "loss": 0.0035,
      "step": 134050
    },
    {
      "epoch": 0.859819198096417,
      "grad_norm": 0.062224142253398895,
      "learning_rate": 7.035128429131823e-06,
      "loss": 0.0016,
      "step": 134060
    },
    {
      "epoch": 0.8598833349902031,
      "grad_norm": 0.08909322321414948,
      "learning_rate": 7.034617173965544e-06,
      "loss": 0.0027,
      "step": 134070
    },
    {
      "epoch": 0.8599474718839892,
      "grad_norm": 0.017653265967965126,
      "learning_rate": 7.034105893304055e-06,
      "loss": 0.0037,
      "step": 134080
    },
    {
      "epoch": 0.8600116087777753,
      "grad_norm": 0.08256793767213821,
      "learning_rate": 7.033594587153767e-06,
      "loss": 0.0022,
      "step": 134090
    },
    {
      "epoch": 0.8600757456715614,
      "grad_norm": 0.042854174971580505,
      "learning_rate": 7.033083255521086e-06,
      "loss": 0.0036,
      "step": 134100
    },
    {
      "epoch": 0.8601398825653475,
      "grad_norm": 0.035144440829753876,
      "learning_rate": 7.032571898412417e-06,
      "loss": 0.0013,
      "step": 134110
    },
    {
      "epoch": 0.8602040194591336,
      "grad_norm": 0.09312219172716141,
      "learning_rate": 7.032060515834172e-06,
      "loss": 0.0123,
      "step": 134120
    },
    {
      "epoch": 0.8602681563529196,
      "grad_norm": 0.041414786130189896,
      "learning_rate": 7.031549107792753e-06,
      "loss": 0.0027,
      "step": 134130
    },
    {
      "epoch": 0.8603322932467058,
      "grad_norm": 0.10964328050613403,
      "learning_rate": 7.031037674294573e-06,
      "loss": 0.0035,
      "step": 134140
    },
    {
      "epoch": 0.8603964301404918,
      "grad_norm": 0.03819069638848305,
      "learning_rate": 7.030526215346041e-06,
      "loss": 0.0027,
      "step": 134150
    },
    {
      "epoch": 0.860460567034278,
      "grad_norm": 0.19429278373718262,
      "learning_rate": 7.030014730953563e-06,
      "loss": 0.0033,
      "step": 134160
    },
    {
      "epoch": 0.860524703928064,
      "grad_norm": 0.08986981213092804,
      "learning_rate": 7.029503221123551e-06,
      "loss": 0.0034,
      "step": 134170
    },
    {
      "epoch": 0.8605888408218502,
      "grad_norm": 0.17730683088302612,
      "learning_rate": 7.028991685862411e-06,
      "loss": 0.0019,
      "step": 134180
    },
    {
      "epoch": 0.8606529777156362,
      "grad_norm": 0.11101886630058289,
      "learning_rate": 7.028480125176556e-06,
      "loss": 0.0033,
      "step": 134190
    },
    {
      "epoch": 0.8607171146094224,
      "grad_norm": 0.21876883506774902,
      "learning_rate": 7.027968539072395e-06,
      "loss": 0.0019,
      "step": 134200
    },
    {
      "epoch": 0.8607812515032085,
      "grad_norm": 0.005220834631472826,
      "learning_rate": 7.02745692755634e-06,
      "loss": 0.0011,
      "step": 134210
    },
    {
      "epoch": 0.8608453883969945,
      "grad_norm": 0.2566666901111603,
      "learning_rate": 7.026945290634799e-06,
      "loss": 0.0018,
      "step": 134220
    },
    {
      "epoch": 0.8609095252907807,
      "grad_norm": 0.19004929065704346,
      "learning_rate": 7.026433628314186e-06,
      "loss": 0.0017,
      "step": 134230
    },
    {
      "epoch": 0.8609736621845667,
      "grad_norm": 0.13197307288646698,
      "learning_rate": 7.025921940600912e-06,
      "loss": 0.0021,
      "step": 134240
    },
    {
      "epoch": 0.8610377990783529,
      "grad_norm": 0.04679390415549278,
      "learning_rate": 7.0254102275013855e-06,
      "loss": 0.0033,
      "step": 134250
    },
    {
      "epoch": 0.8611019359721389,
      "grad_norm": 0.21860431134700775,
      "learning_rate": 7.024898489022023e-06,
      "loss": 0.0034,
      "step": 134260
    },
    {
      "epoch": 0.8611660728659251,
      "grad_norm": 0.10462875664234161,
      "learning_rate": 7.024386725169236e-06,
      "loss": 0.0033,
      "step": 134270
    },
    {
      "epoch": 0.8612302097597111,
      "grad_norm": 0.09628218412399292,
      "learning_rate": 7.023874935949435e-06,
      "loss": 0.0021,
      "step": 134280
    },
    {
      "epoch": 0.8612943466534972,
      "grad_norm": 0.08594681322574615,
      "learning_rate": 7.023363121369037e-06,
      "loss": 0.0015,
      "step": 134290
    },
    {
      "epoch": 0.8613584835472833,
      "grad_norm": 0.05004475638270378,
      "learning_rate": 7.022851281434451e-06,
      "loss": 0.0026,
      "step": 134300
    },
    {
      "epoch": 0.8614226204410694,
      "grad_norm": 0.06274831295013428,
      "learning_rate": 7.0223394161520944e-06,
      "loss": 0.002,
      "step": 134310
    },
    {
      "epoch": 0.8614867573348555,
      "grad_norm": 0.06289125978946686,
      "learning_rate": 7.0218275255283775e-06,
      "loss": 0.0024,
      "step": 134320
    },
    {
      "epoch": 0.8615508942286416,
      "grad_norm": 0.16910091042518616,
      "learning_rate": 7.021315609569719e-06,
      "loss": 0.0034,
      "step": 134330
    },
    {
      "epoch": 0.8616150311224277,
      "grad_norm": 0.19005230069160461,
      "learning_rate": 7.020803668282529e-06,
      "loss": 0.002,
      "step": 134340
    },
    {
      "epoch": 0.8616791680162138,
      "grad_norm": 0.06614792346954346,
      "learning_rate": 7.020291701673225e-06,
      "loss": 0.0028,
      "step": 134350
    },
    {
      "epoch": 0.86174330491,
      "grad_norm": 0.04979289695620537,
      "learning_rate": 7.019779709748223e-06,
      "loss": 0.0013,
      "step": 134360
    },
    {
      "epoch": 0.861807441803786,
      "grad_norm": 0.16481654345989227,
      "learning_rate": 7.019267692513938e-06,
      "loss": 0.0039,
      "step": 134370
    },
    {
      "epoch": 0.8618715786975721,
      "grad_norm": 0.0695338174700737,
      "learning_rate": 7.018755649976785e-06,
      "loss": 0.0035,
      "step": 134380
    },
    {
      "epoch": 0.8619357155913582,
      "grad_norm": 0.20109935104846954,
      "learning_rate": 7.0182435821431815e-06,
      "loss": 0.003,
      "step": 134390
    },
    {
      "epoch": 0.8619998524851443,
      "grad_norm": 0.04521843045949936,
      "learning_rate": 7.0177314890195435e-06,
      "loss": 0.0025,
      "step": 134400
    },
    {
      "epoch": 0.8620639893789304,
      "grad_norm": 0.024238646030426025,
      "learning_rate": 7.017219370612287e-06,
      "loss": 0.002,
      "step": 134410
    },
    {
      "epoch": 0.8621281262727165,
      "grad_norm": 0.09983240067958832,
      "learning_rate": 7.016707226927831e-06,
      "loss": 0.0054,
      "step": 134420
    },
    {
      "epoch": 0.8621922631665025,
      "grad_norm": 0.06918484717607498,
      "learning_rate": 7.016195057972591e-06,
      "loss": 0.003,
      "step": 134430
    },
    {
      "epoch": 0.8622564000602887,
      "grad_norm": 0.08813716471195221,
      "learning_rate": 7.015682863752988e-06,
      "loss": 0.002,
      "step": 134440
    },
    {
      "epoch": 0.8623205369540747,
      "grad_norm": 0.06289267539978027,
      "learning_rate": 7.0151706442754365e-06,
      "loss": 0.001,
      "step": 134450
    },
    {
      "epoch": 0.8623846738478609,
      "grad_norm": 0.09562243521213531,
      "learning_rate": 7.014658399546357e-06,
      "loss": 0.0025,
      "step": 134460
    },
    {
      "epoch": 0.8624488107416469,
      "grad_norm": 0.18738164007663727,
      "learning_rate": 7.014146129572168e-06,
      "loss": 0.0034,
      "step": 134470
    },
    {
      "epoch": 0.8625129476354331,
      "grad_norm": 0.16660718619823456,
      "learning_rate": 7.013633834359289e-06,
      "loss": 0.0072,
      "step": 134480
    },
    {
      "epoch": 0.8625770845292192,
      "grad_norm": 0.17574219405651093,
      "learning_rate": 7.0131215139141385e-06,
      "loss": 0.0038,
      "step": 134490
    },
    {
      "epoch": 0.8626412214230053,
      "grad_norm": 0.1404060572385788,
      "learning_rate": 7.0126091682431355e-06,
      "loss": 0.0069,
      "step": 134500
    },
    {
      "epoch": 0.8627053583167914,
      "grad_norm": 0.21174687147140503,
      "learning_rate": 7.012096797352703e-06,
      "loss": 0.0017,
      "step": 134510
    },
    {
      "epoch": 0.8627694952105774,
      "grad_norm": 0.13424913585186005,
      "learning_rate": 7.0115844012492585e-06,
      "loss": 0.0027,
      "step": 134520
    },
    {
      "epoch": 0.8628336321043636,
      "grad_norm": 0.08720983564853668,
      "learning_rate": 7.011071979939225e-06,
      "loss": 0.0024,
      "step": 134530
    },
    {
      "epoch": 0.8628977689981496,
      "grad_norm": 0.02763362228870392,
      "learning_rate": 7.0105595334290196e-06,
      "loss": 0.0052,
      "step": 134540
    },
    {
      "epoch": 0.8629619058919358,
      "grad_norm": 0.2591906189918518,
      "learning_rate": 7.01004706172507e-06,
      "loss": 0.0062,
      "step": 134550
    },
    {
      "epoch": 0.8630260427857218,
      "grad_norm": 0.07465941458940506,
      "learning_rate": 7.009534564833791e-06,
      "loss": 0.0018,
      "step": 134560
    },
    {
      "epoch": 0.863090179679508,
      "grad_norm": 0.2604738771915436,
      "learning_rate": 7.00902204276161e-06,
      "loss": 0.0027,
      "step": 134570
    },
    {
      "epoch": 0.863154316573294,
      "grad_norm": 0.07500201463699341,
      "learning_rate": 7.008509495514945e-06,
      "loss": 0.0029,
      "step": 134580
    },
    {
      "epoch": 0.8632184534670801,
      "grad_norm": 0.3114131689071655,
      "learning_rate": 7.007996923100222e-06,
      "loss": 0.0096,
      "step": 134590
    },
    {
      "epoch": 0.8632825903608662,
      "grad_norm": 0.04153985530138016,
      "learning_rate": 7.007484325523862e-06,
      "loss": 0.0018,
      "step": 134600
    },
    {
      "epoch": 0.8633467272546523,
      "grad_norm": 0.21154899895191193,
      "learning_rate": 7.006971702792289e-06,
      "loss": 0.0038,
      "step": 134610
    },
    {
      "epoch": 0.8634108641484384,
      "grad_norm": 0.14301951229572296,
      "learning_rate": 7.006459054911926e-06,
      "loss": 0.0028,
      "step": 134620
    },
    {
      "epoch": 0.8634750010422245,
      "grad_norm": 0.03243682533502579,
      "learning_rate": 7.005946381889197e-06,
      "loss": 0.0025,
      "step": 134630
    },
    {
      "epoch": 0.8635391379360107,
      "grad_norm": 0.5242324471473694,
      "learning_rate": 7.005433683730525e-06,
      "loss": 0.0039,
      "step": 134640
    },
    {
      "epoch": 0.8636032748297967,
      "grad_norm": 0.08665811270475388,
      "learning_rate": 7.004920960442337e-06,
      "loss": 0.0015,
      "step": 134650
    },
    {
      "epoch": 0.8636674117235829,
      "grad_norm": 0.06900236010551453,
      "learning_rate": 7.004408212031056e-06,
      "loss": 0.0017,
      "step": 134660
    },
    {
      "epoch": 0.8637315486173689,
      "grad_norm": 0.011264686472713947,
      "learning_rate": 7.0038954385031085e-06,
      "loss": 0.0029,
      "step": 134670
    },
    {
      "epoch": 0.863795685511155,
      "grad_norm": 0.1056748479604721,
      "learning_rate": 7.003382639864919e-06,
      "loss": 0.0024,
      "step": 134680
    },
    {
      "epoch": 0.8638598224049411,
      "grad_norm": 0.07375384867191315,
      "learning_rate": 7.002869816122912e-06,
      "loss": 0.0022,
      "step": 134690
    },
    {
      "epoch": 0.8639239592987272,
      "grad_norm": 0.11620330810546875,
      "learning_rate": 7.002356967283516e-06,
      "loss": 0.0049,
      "step": 134700
    },
    {
      "epoch": 0.8639880961925133,
      "grad_norm": 0.16854000091552734,
      "learning_rate": 7.001844093353154e-06,
      "loss": 0.0019,
      "step": 134710
    },
    {
      "epoch": 0.8640522330862994,
      "grad_norm": 0.09560932219028473,
      "learning_rate": 7.001331194338258e-06,
      "loss": 0.001,
      "step": 134720
    },
    {
      "epoch": 0.8641163699800855,
      "grad_norm": 0.12318872660398483,
      "learning_rate": 7.000818270245249e-06,
      "loss": 0.0032,
      "step": 134730
    },
    {
      "epoch": 0.8641805068738716,
      "grad_norm": 0.15796954929828644,
      "learning_rate": 7.000305321080559e-06,
      "loss": 0.0043,
      "step": 134740
    },
    {
      "epoch": 0.8642446437676576,
      "grad_norm": 0.049662332981824875,
      "learning_rate": 6.999792346850613e-06,
      "loss": 0.0034,
      "step": 134750
    },
    {
      "epoch": 0.8643087806614438,
      "grad_norm": 0.1907750964164734,
      "learning_rate": 6.99927934756184e-06,
      "loss": 0.0028,
      "step": 134760
    },
    {
      "epoch": 0.8643729175552299,
      "grad_norm": 0.1717313677072525,
      "learning_rate": 6.998766323220667e-06,
      "loss": 0.0023,
      "step": 134770
    },
    {
      "epoch": 0.864437054449016,
      "grad_norm": 0.2066045105457306,
      "learning_rate": 6.998253273833524e-06,
      "loss": 0.002,
      "step": 134780
    },
    {
      "epoch": 0.8645011913428021,
      "grad_norm": 0.1488698422908783,
      "learning_rate": 6.99774019940684e-06,
      "loss": 0.0035,
      "step": 134790
    },
    {
      "epoch": 0.8645653282365882,
      "grad_norm": 0.16711297631263733,
      "learning_rate": 6.997227099947043e-06,
      "loss": 0.0034,
      "step": 134800
    },
    {
      "epoch": 0.8646294651303743,
      "grad_norm": 0.05773639306426048,
      "learning_rate": 6.996713975460563e-06,
      "loss": 0.0025,
      "step": 134810
    },
    {
      "epoch": 0.8646936020241603,
      "grad_norm": 0.06513752043247223,
      "learning_rate": 6.996200825953829e-06,
      "loss": 0.0017,
      "step": 134820
    },
    {
      "epoch": 0.8647577389179465,
      "grad_norm": 0.22357122600078583,
      "learning_rate": 6.995687651433273e-06,
      "loss": 0.002,
      "step": 134830
    },
    {
      "epoch": 0.8648218758117325,
      "grad_norm": 0.07101402431726456,
      "learning_rate": 6.995174451905324e-06,
      "loss": 0.0018,
      "step": 134840
    },
    {
      "epoch": 0.8648860127055187,
      "grad_norm": 0.06312518566846848,
      "learning_rate": 6.994661227376414e-06,
      "loss": 0.002,
      "step": 134850
    },
    {
      "epoch": 0.8649501495993047,
      "grad_norm": 0.07518059015274048,
      "learning_rate": 6.994147977852972e-06,
      "loss": 0.0033,
      "step": 134860
    },
    {
      "epoch": 0.8650142864930909,
      "grad_norm": 0.0721321851015091,
      "learning_rate": 6.993634703341432e-06,
      "loss": 0.0015,
      "step": 134870
    },
    {
      "epoch": 0.8650784233868769,
      "grad_norm": 0.07936231046915054,
      "learning_rate": 6.993121403848223e-06,
      "loss": 0.0032,
      "step": 134880
    },
    {
      "epoch": 0.865142560280663,
      "grad_norm": 0.15310147404670715,
      "learning_rate": 6.99260807937978e-06,
      "loss": 0.0032,
      "step": 134890
    },
    {
      "epoch": 0.8652066971744491,
      "grad_norm": 0.07506764680147171,
      "learning_rate": 6.992094729942533e-06,
      "loss": 0.002,
      "step": 134900
    },
    {
      "epoch": 0.8652708340682352,
      "grad_norm": 0.05334783345460892,
      "learning_rate": 6.991581355542915e-06,
      "loss": 0.0016,
      "step": 134910
    },
    {
      "epoch": 0.8653349709620214,
      "grad_norm": 0.17147868871688843,
      "learning_rate": 6.991067956187359e-06,
      "loss": 0.0033,
      "step": 134920
    },
    {
      "epoch": 0.8653991078558074,
      "grad_norm": 0.16561748087406158,
      "learning_rate": 6.990554531882299e-06,
      "loss": 0.0015,
      "step": 134930
    },
    {
      "epoch": 0.8654632447495936,
      "grad_norm": 0.08491192013025284,
      "learning_rate": 6.9900410826341665e-06,
      "loss": 0.0015,
      "step": 134940
    },
    {
      "epoch": 0.8655273816433796,
      "grad_norm": 0.14289340376853943,
      "learning_rate": 6.989527608449399e-06,
      "loss": 0.0024,
      "step": 134950
    },
    {
      "epoch": 0.8655915185371658,
      "grad_norm": 0.13716906309127808,
      "learning_rate": 6.989014109334428e-06,
      "loss": 0.0042,
      "step": 134960
    },
    {
      "epoch": 0.8656556554309518,
      "grad_norm": 0.2592507004737854,
      "learning_rate": 6.988500585295689e-06,
      "loss": 0.0025,
      "step": 134970
    },
    {
      "epoch": 0.865719792324738,
      "grad_norm": 0.09000234305858612,
      "learning_rate": 6.987987036339616e-06,
      "loss": 0.002,
      "step": 134980
    },
    {
      "epoch": 0.865783929218524,
      "grad_norm": 0.13545222580432892,
      "learning_rate": 6.9874734624726445e-06,
      "loss": 0.0016,
      "step": 134990
    },
    {
      "epoch": 0.8658480661123101,
      "grad_norm": 0.12626180052757263,
      "learning_rate": 6.98695986370121e-06,
      "loss": 0.0022,
      "step": 135000
    },
    {
      "epoch": 0.8659122030060962,
      "grad_norm": 0.48018166422843933,
      "learning_rate": 6.986446240031749e-06,
      "loss": 0.0042,
      "step": 135010
    },
    {
      "epoch": 0.8659763398998823,
      "grad_norm": 0.18012456595897675,
      "learning_rate": 6.985932591470697e-06,
      "loss": 0.0039,
      "step": 135020
    },
    {
      "epoch": 0.8660404767936684,
      "grad_norm": 0.1186421662569046,
      "learning_rate": 6.985418918024489e-06,
      "loss": 0.0016,
      "step": 135030
    },
    {
      "epoch": 0.8661046136874545,
      "grad_norm": 0.0692361518740654,
      "learning_rate": 6.984905219699565e-06,
      "loss": 0.0044,
      "step": 135040
    },
    {
      "epoch": 0.8661687505812407,
      "grad_norm": 0.2168842852115631,
      "learning_rate": 6.984391496502358e-06,
      "loss": 0.0028,
      "step": 135050
    },
    {
      "epoch": 0.8662328874750267,
      "grad_norm": 0.15261414647102356,
      "learning_rate": 6.98387774843931e-06,
      "loss": 0.0022,
      "step": 135060
    },
    {
      "epoch": 0.8662970243688128,
      "grad_norm": 0.1448751986026764,
      "learning_rate": 6.983363975516853e-06,
      "loss": 0.0021,
      "step": 135070
    },
    {
      "epoch": 0.8663611612625989,
      "grad_norm": 0.0858541801571846,
      "learning_rate": 6.98285017774143e-06,
      "loss": 0.003,
      "step": 135080
    },
    {
      "epoch": 0.866425298156385,
      "grad_norm": 0.010609383694827557,
      "learning_rate": 6.982336355119475e-06,
      "loss": 0.0032,
      "step": 135090
    },
    {
      "epoch": 0.8664894350501711,
      "grad_norm": 0.14291512966156006,
      "learning_rate": 6.981822507657431e-06,
      "loss": 0.0014,
      "step": 135100
    },
    {
      "epoch": 0.8665535719439572,
      "grad_norm": 0.23925663530826569,
      "learning_rate": 6.9813086353617335e-06,
      "loss": 0.0012,
      "step": 135110
    },
    {
      "epoch": 0.8666177088377433,
      "grad_norm": 0.27118679881095886,
      "learning_rate": 6.980794738238823e-06,
      "loss": 0.0025,
      "step": 135120
    },
    {
      "epoch": 0.8666818457315294,
      "grad_norm": 0.15810494124889374,
      "learning_rate": 6.980280816295138e-06,
      "loss": 0.0016,
      "step": 135130
    },
    {
      "epoch": 0.8667459826253154,
      "grad_norm": 0.22343648970127106,
      "learning_rate": 6.97976686953712e-06,
      "loss": 0.0059,
      "step": 135140
    },
    {
      "epoch": 0.8668101195191016,
      "grad_norm": 0.09998579323291779,
      "learning_rate": 6.979252897971208e-06,
      "loss": 0.0018,
      "step": 135150
    },
    {
      "epoch": 0.8668742564128876,
      "grad_norm": 0.18671497702598572,
      "learning_rate": 6.978738901603843e-06,
      "loss": 0.0037,
      "step": 135160
    },
    {
      "epoch": 0.8669383933066738,
      "grad_norm": 0.021066153421998024,
      "learning_rate": 6.978224880441464e-06,
      "loss": 0.0031,
      "step": 135170
    },
    {
      "epoch": 0.8670025302004598,
      "grad_norm": 0.11940399557352066,
      "learning_rate": 6.977710834490515e-06,
      "loss": 0.004,
      "step": 135180
    },
    {
      "epoch": 0.867066667094246,
      "grad_norm": 0.046428464353084564,
      "learning_rate": 6.977196763757436e-06,
      "loss": 0.0015,
      "step": 135190
    },
    {
      "epoch": 0.8671308039880321,
      "grad_norm": 0.07993770390748978,
      "learning_rate": 6.976682668248667e-06,
      "loss": 0.0015,
      "step": 135200
    },
    {
      "epoch": 0.8671949408818181,
      "grad_norm": 0.07815032452344894,
      "learning_rate": 6.976168547970652e-06,
      "loss": 0.0034,
      "step": 135210
    },
    {
      "epoch": 0.8672590777756043,
      "grad_norm": 0.12343888729810715,
      "learning_rate": 6.9756544029298325e-06,
      "loss": 0.0026,
      "step": 135220
    },
    {
      "epoch": 0.8673232146693903,
      "grad_norm": 0.37997809052467346,
      "learning_rate": 6.975140233132652e-06,
      "loss": 0.0026,
      "step": 135230
    },
    {
      "epoch": 0.8673873515631765,
      "grad_norm": 0.047730203717947006,
      "learning_rate": 6.974626038585552e-06,
      "loss": 0.0012,
      "step": 135240
    },
    {
      "epoch": 0.8674514884569625,
      "grad_norm": 0.11115345358848572,
      "learning_rate": 6.974111819294979e-06,
      "loss": 0.0026,
      "step": 135250
    },
    {
      "epoch": 0.8675156253507487,
      "grad_norm": 0.076470285654068,
      "learning_rate": 6.973597575267371e-06,
      "loss": 0.002,
      "step": 135260
    },
    {
      "epoch": 0.8675797622445347,
      "grad_norm": 0.18768024444580078,
      "learning_rate": 6.9730833065091765e-06,
      "loss": 0.0037,
      "step": 135270
    },
    {
      "epoch": 0.8676438991383209,
      "grad_norm": 0.06109945848584175,
      "learning_rate": 6.972569013026837e-06,
      "loss": 0.0025,
      "step": 135280
    },
    {
      "epoch": 0.8677080360321069,
      "grad_norm": 0.10006918013095856,
      "learning_rate": 6.972054694826799e-06,
      "loss": 0.0037,
      "step": 135290
    },
    {
      "epoch": 0.867772172925893,
      "grad_norm": 0.21546000242233276,
      "learning_rate": 6.971540351915504e-06,
      "loss": 0.0016,
      "step": 135300
    },
    {
      "epoch": 0.8678363098196791,
      "grad_norm": 0.08633749186992645,
      "learning_rate": 6.9710259842994025e-06,
      "loss": 0.0045,
      "step": 135310
    },
    {
      "epoch": 0.8679004467134652,
      "grad_norm": 0.05979597941040993,
      "learning_rate": 6.970511591984936e-06,
      "loss": 0.0018,
      "step": 135320
    },
    {
      "epoch": 0.8679645836072514,
      "grad_norm": 0.11502660065889359,
      "learning_rate": 6.96999717497855e-06,
      "loss": 0.0038,
      "step": 135330
    },
    {
      "epoch": 0.8680287205010374,
      "grad_norm": 0.20135074853897095,
      "learning_rate": 6.969482733286691e-06,
      "loss": 0.0047,
      "step": 135340
    },
    {
      "epoch": 0.8680928573948236,
      "grad_norm": 0.297893226146698,
      "learning_rate": 6.968968266915806e-06,
      "loss": 0.0022,
      "step": 135350
    },
    {
      "epoch": 0.8681569942886096,
      "grad_norm": 0.014258438721299171,
      "learning_rate": 6.968453775872342e-06,
      "loss": 0.0026,
      "step": 135360
    },
    {
      "epoch": 0.8682211311823957,
      "grad_norm": 0.057328782975673676,
      "learning_rate": 6.967939260162746e-06,
      "loss": 0.0023,
      "step": 135370
    },
    {
      "epoch": 0.8682852680761818,
      "grad_norm": 0.16010943055152893,
      "learning_rate": 6.967424719793464e-06,
      "loss": 0.0077,
      "step": 135380
    },
    {
      "epoch": 0.8683494049699679,
      "grad_norm": 0.03984547406435013,
      "learning_rate": 6.966910154770943e-06,
      "loss": 0.0031,
      "step": 135390
    },
    {
      "epoch": 0.868413541863754,
      "grad_norm": 0.0993046835064888,
      "learning_rate": 6.966395565101634e-06,
      "loss": 0.001,
      "step": 135400
    },
    {
      "epoch": 0.8684776787575401,
      "grad_norm": 0.024639731273055077,
      "learning_rate": 6.965880950791981e-06,
      "loss": 0.0018,
      "step": 135410
    },
    {
      "epoch": 0.8685418156513262,
      "grad_norm": 0.02593991719186306,
      "learning_rate": 6.965366311848436e-06,
      "loss": 0.002,
      "step": 135420
    },
    {
      "epoch": 0.8686059525451123,
      "grad_norm": 0.1336241215467453,
      "learning_rate": 6.9648516482774464e-06,
      "loss": 0.0021,
      "step": 135430
    },
    {
      "epoch": 0.8686700894388983,
      "grad_norm": 0.04180555418133736,
      "learning_rate": 6.964336960085461e-06,
      "loss": 0.0021,
      "step": 135440
    },
    {
      "epoch": 0.8687342263326845,
      "grad_norm": 0.027477435767650604,
      "learning_rate": 6.9638222472789305e-06,
      "loss": 0.0053,
      "step": 135450
    },
    {
      "epoch": 0.8687983632264705,
      "grad_norm": 0.13674120604991913,
      "learning_rate": 6.963307509864303e-06,
      "loss": 0.0023,
      "step": 135460
    },
    {
      "epoch": 0.8688625001202567,
      "grad_norm": 0.08610444515943527,
      "learning_rate": 6.96279274784803e-06,
      "loss": 0.0024,
      "step": 135470
    },
    {
      "epoch": 0.8689266370140428,
      "grad_norm": 0.10824882239103317,
      "learning_rate": 6.962277961236561e-06,
      "loss": 0.0012,
      "step": 135480
    },
    {
      "epoch": 0.8689907739078289,
      "grad_norm": 0.12073953449726105,
      "learning_rate": 6.961763150036346e-06,
      "loss": 0.0038,
      "step": 135490
    },
    {
      "epoch": 0.869054910801615,
      "grad_norm": 0.1749655157327652,
      "learning_rate": 6.961248314253836e-06,
      "loss": 0.0027,
      "step": 135500
    },
    {
      "epoch": 0.869119047695401,
      "grad_norm": 0.30429336428642273,
      "learning_rate": 6.960733453895485e-06,
      "loss": 0.0033,
      "step": 135510
    },
    {
      "epoch": 0.8691831845891872,
      "grad_norm": 0.09175138920545578,
      "learning_rate": 6.960218568967741e-06,
      "loss": 0.0015,
      "step": 135520
    },
    {
      "epoch": 0.8692473214829732,
      "grad_norm": 0.1776876002550125,
      "learning_rate": 6.9597036594770586e-06,
      "loss": 0.003,
      "step": 135530
    },
    {
      "epoch": 0.8693114583767594,
      "grad_norm": 0.026637470349669456,
      "learning_rate": 6.9591887254298886e-06,
      "loss": 0.0044,
      "step": 135540
    },
    {
      "epoch": 0.8693755952705454,
      "grad_norm": 0.12641720473766327,
      "learning_rate": 6.958673766832682e-06,
      "loss": 0.0023,
      "step": 135550
    },
    {
      "epoch": 0.8694397321643316,
      "grad_norm": 0.10737831890583038,
      "learning_rate": 6.958158783691894e-06,
      "loss": 0.0025,
      "step": 135560
    },
    {
      "epoch": 0.8695038690581176,
      "grad_norm": 0.07200788706541061,
      "learning_rate": 6.957643776013978e-06,
      "loss": 0.0015,
      "step": 135570
    },
    {
      "epoch": 0.8695680059519038,
      "grad_norm": 0.08781659603118896,
      "learning_rate": 6.957128743805385e-06,
      "loss": 0.0019,
      "step": 135580
    },
    {
      "epoch": 0.8696321428456898,
      "grad_norm": 0.035964686423540115,
      "learning_rate": 6.956613687072571e-06,
      "loss": 0.0038,
      "step": 135590
    },
    {
      "epoch": 0.8696962797394759,
      "grad_norm": 0.1329997181892395,
      "learning_rate": 6.956098605821988e-06,
      "loss": 0.0043,
      "step": 135600
    },
    {
      "epoch": 0.869760416633262,
      "grad_norm": 0.07175031304359436,
      "learning_rate": 6.955583500060093e-06,
      "loss": 0.0028,
      "step": 135610
    },
    {
      "epoch": 0.8698245535270481,
      "grad_norm": 0.055772650986909866,
      "learning_rate": 6.955068369793338e-06,
      "loss": 0.0023,
      "step": 135620
    },
    {
      "epoch": 0.8698886904208343,
      "grad_norm": 0.08265043795108795,
      "learning_rate": 6.954553215028181e-06,
      "loss": 0.0033,
      "step": 135630
    },
    {
      "epoch": 0.8699528273146203,
      "grad_norm": 0.020286694169044495,
      "learning_rate": 6.954038035771073e-06,
      "loss": 0.0015,
      "step": 135640
    },
    {
      "epoch": 0.8700169642084065,
      "grad_norm": 0.16302058100700378,
      "learning_rate": 6.953522832028473e-06,
      "loss": 0.0018,
      "step": 135650
    },
    {
      "epoch": 0.8700811011021925,
      "grad_norm": 0.0685119554400444,
      "learning_rate": 6.953007603806835e-06,
      "loss": 0.003,
      "step": 135660
    },
    {
      "epoch": 0.8701452379959786,
      "grad_norm": 0.1629972606897354,
      "learning_rate": 6.952492351112617e-06,
      "loss": 0.0036,
      "step": 135670
    },
    {
      "epoch": 0.8702093748897647,
      "grad_norm": 0.0794086903333664,
      "learning_rate": 6.951977073952274e-06,
      "loss": 0.0018,
      "step": 135680
    },
    {
      "epoch": 0.8702735117835508,
      "grad_norm": 0.0831390917301178,
      "learning_rate": 6.951461772332263e-06,
      "loss": 0.0036,
      "step": 135690
    },
    {
      "epoch": 0.8703376486773369,
      "grad_norm": 0.18447096645832062,
      "learning_rate": 6.950946446259041e-06,
      "loss": 0.0029,
      "step": 135700
    },
    {
      "epoch": 0.870401785571123,
      "grad_norm": 0.07849343866109848,
      "learning_rate": 6.950431095739065e-06,
      "loss": 0.0016,
      "step": 135710
    },
    {
      "epoch": 0.8704659224649091,
      "grad_norm": 0.046604666858911514,
      "learning_rate": 6.9499157207787956e-06,
      "loss": 0.0018,
      "step": 135720
    },
    {
      "epoch": 0.8705300593586952,
      "grad_norm": 0.14687475562095642,
      "learning_rate": 6.949400321384687e-06,
      "loss": 0.0022,
      "step": 135730
    },
    {
      "epoch": 0.8705941962524812,
      "grad_norm": 0.12612827122211456,
      "learning_rate": 6.948884897563201e-06,
      "loss": 0.0042,
      "step": 135740
    },
    {
      "epoch": 0.8706583331462674,
      "grad_norm": 0.12573637068271637,
      "learning_rate": 6.948369449320792e-06,
      "loss": 0.0015,
      "step": 135750
    },
    {
      "epoch": 0.8707224700400535,
      "grad_norm": 0.1647786647081375,
      "learning_rate": 6.947853976663923e-06,
      "loss": 0.003,
      "step": 135760
    },
    {
      "epoch": 0.8707866069338396,
      "grad_norm": 0.10565929859876633,
      "learning_rate": 6.94733847959905e-06,
      "loss": 0.0014,
      "step": 135770
    },
    {
      "epoch": 0.8708507438276257,
      "grad_norm": 0.06949920207262039,
      "learning_rate": 6.946822958132635e-06,
      "loss": 0.0019,
      "step": 135780
    },
    {
      "epoch": 0.8709148807214118,
      "grad_norm": 0.27971917390823364,
      "learning_rate": 6.946307412271136e-06,
      "loss": 0.003,
      "step": 135790
    },
    {
      "epoch": 0.8709790176151979,
      "grad_norm": 0.004446999169886112,
      "learning_rate": 6.945791842021016e-06,
      "loss": 0.002,
      "step": 135800
    },
    {
      "epoch": 0.871043154508984,
      "grad_norm": 0.1649855375289917,
      "learning_rate": 6.945276247388732e-06,
      "loss": 0.0033,
      "step": 135810
    },
    {
      "epoch": 0.8711072914027701,
      "grad_norm": 0.12075648456811905,
      "learning_rate": 6.944760628380748e-06,
      "loss": 0.0034,
      "step": 135820
    },
    {
      "epoch": 0.8711714282965561,
      "grad_norm": 0.13726001977920532,
      "learning_rate": 6.944244985003522e-06,
      "loss": 0.0022,
      "step": 135830
    },
    {
      "epoch": 0.8712355651903423,
      "grad_norm": 0.2607235908508301,
      "learning_rate": 6.9437293172635175e-06,
      "loss": 0.0013,
      "step": 135840
    },
    {
      "epoch": 0.8712997020841283,
      "grad_norm": 0.15931189060211182,
      "learning_rate": 6.9432136251671955e-06,
      "loss": 0.0051,
      "step": 135850
    },
    {
      "epoch": 0.8713638389779145,
      "grad_norm": 0.11268272995948792,
      "learning_rate": 6.942697908721017e-06,
      "loss": 0.0024,
      "step": 135860
    },
    {
      "epoch": 0.8714279758717005,
      "grad_norm": 0.10024124383926392,
      "learning_rate": 6.942182167931446e-06,
      "loss": 0.0018,
      "step": 135870
    },
    {
      "epoch": 0.8714921127654867,
      "grad_norm": 0.023353857919573784,
      "learning_rate": 6.941666402804945e-06,
      "loss": 0.0018,
      "step": 135880
    },
    {
      "epoch": 0.8715562496592727,
      "grad_norm": 0.06861988455057144,
      "learning_rate": 6.9411506133479756e-06,
      "loss": 0.0028,
      "step": 135890
    },
    {
      "epoch": 0.8716203865530588,
      "grad_norm": 0.012388293631374836,
      "learning_rate": 6.940634799567002e-06,
      "loss": 0.0023,
      "step": 135900
    },
    {
      "epoch": 0.871684523446845,
      "grad_norm": 0.11425565928220749,
      "learning_rate": 6.9401189614684875e-06,
      "loss": 0.0045,
      "step": 135910
    },
    {
      "epoch": 0.871748660340631,
      "grad_norm": 0.07764580100774765,
      "learning_rate": 6.939603099058895e-06,
      "loss": 0.0021,
      "step": 135920
    },
    {
      "epoch": 0.8718127972344172,
      "grad_norm": 0.02401791699230671,
      "learning_rate": 6.939087212344691e-06,
      "loss": 0.0018,
      "step": 135930
    },
    {
      "epoch": 0.8718769341282032,
      "grad_norm": 0.01430745143443346,
      "learning_rate": 6.938571301332337e-06,
      "loss": 0.0021,
      "step": 135940
    },
    {
      "epoch": 0.8719410710219894,
      "grad_norm": 0.31946444511413574,
      "learning_rate": 6.938055366028299e-06,
      "loss": 0.0023,
      "step": 135950
    },
    {
      "epoch": 0.8720052079157754,
      "grad_norm": 0.05516481027007103,
      "learning_rate": 6.937539406439042e-06,
      "loss": 0.001,
      "step": 135960
    },
    {
      "epoch": 0.8720693448095616,
      "grad_norm": 0.020058229565620422,
      "learning_rate": 6.9370234225710335e-06,
      "loss": 0.0015,
      "step": 135970
    },
    {
      "epoch": 0.8721334817033476,
      "grad_norm": 0.24442732334136963,
      "learning_rate": 6.936507414430735e-06,
      "loss": 0.0021,
      "step": 135980
    },
    {
      "epoch": 0.8721976185971337,
      "grad_norm": 0.042491428554058075,
      "learning_rate": 6.935991382024616e-06,
      "loss": 0.0018,
      "step": 135990
    },
    {
      "epoch": 0.8722617554909198,
      "grad_norm": 0.07440150529146194,
      "learning_rate": 6.93547532535914e-06,
      "loss": 0.0015,
      "step": 136000
    },
    {
      "epoch": 0.8723258923847059,
      "grad_norm": 0.18436209857463837,
      "learning_rate": 6.934959244440776e-06,
      "loss": 0.004,
      "step": 136010
    },
    {
      "epoch": 0.872390029278492,
      "grad_norm": 0.08648974448442459,
      "learning_rate": 6.9344431392759895e-06,
      "loss": 0.0042,
      "step": 136020
    },
    {
      "epoch": 0.8724541661722781,
      "grad_norm": 0.030176298692822456,
      "learning_rate": 6.933927009871249e-06,
      "loss": 0.0023,
      "step": 136030
    },
    {
      "epoch": 0.8725183030660643,
      "grad_norm": 0.1476580649614334,
      "learning_rate": 6.933410856233018e-06,
      "loss": 0.0026,
      "step": 136040
    },
    {
      "epoch": 0.8725824399598503,
      "grad_norm": 0.09100686013698578,
      "learning_rate": 6.932894678367769e-06,
      "loss": 0.002,
      "step": 136050
    },
    {
      "epoch": 0.8726465768536364,
      "grad_norm": 0.002788944635540247,
      "learning_rate": 6.932378476281969e-06,
      "loss": 0.0019,
      "step": 136060
    },
    {
      "epoch": 0.8727107137474225,
      "grad_norm": 0.22135302424430847,
      "learning_rate": 6.931862249982084e-06,
      "loss": 0.0022,
      "step": 136070
    },
    {
      "epoch": 0.8727748506412086,
      "grad_norm": 0.4503108561038971,
      "learning_rate": 6.9313459994745855e-06,
      "loss": 0.0031,
      "step": 136080
    },
    {
      "epoch": 0.8728389875349947,
      "grad_norm": 0.11171503365039825,
      "learning_rate": 6.930829724765941e-06,
      "loss": 0.0017,
      "step": 136090
    },
    {
      "epoch": 0.8729031244287808,
      "grad_norm": 0.12623871862888336,
      "learning_rate": 6.93031342586262e-06,
      "loss": 0.0022,
      "step": 136100
    },
    {
      "epoch": 0.8729672613225669,
      "grad_norm": 0.3261496424674988,
      "learning_rate": 6.929797102771092e-06,
      "loss": 0.0037,
      "step": 136110
    },
    {
      "epoch": 0.873031398216353,
      "grad_norm": 0.09023922681808472,
      "learning_rate": 6.929280755497828e-06,
      "loss": 0.0018,
      "step": 136120
    },
    {
      "epoch": 0.873095535110139,
      "grad_norm": 0.0945887565612793,
      "learning_rate": 6.9287643840492965e-06,
      "loss": 0.0015,
      "step": 136130
    },
    {
      "epoch": 0.8731596720039252,
      "grad_norm": 0.1977291852235794,
      "learning_rate": 6.92824798843197e-06,
      "loss": 0.0018,
      "step": 136140
    },
    {
      "epoch": 0.8732238088977112,
      "grad_norm": 0.11037852615118027,
      "learning_rate": 6.927731568652316e-06,
      "loss": 0.0021,
      "step": 136150
    },
    {
      "epoch": 0.8732879457914974,
      "grad_norm": 0.04763404279947281,
      "learning_rate": 6.927215124716808e-06,
      "loss": 0.0011,
      "step": 136160
    },
    {
      "epoch": 0.8733520826852834,
      "grad_norm": 0.3576797544956207,
      "learning_rate": 6.926698656631918e-06,
      "loss": 0.0034,
      "step": 136170
    },
    {
      "epoch": 0.8734162195790696,
      "grad_norm": 0.03536294028162956,
      "learning_rate": 6.9261821644041185e-06,
      "loss": 0.0021,
      "step": 136180
    },
    {
      "epoch": 0.8734803564728557,
      "grad_norm": 0.051673565059900284,
      "learning_rate": 6.925665648039876e-06,
      "loss": 0.0022,
      "step": 136190
    },
    {
      "epoch": 0.8735444933666417,
      "grad_norm": 0.0459916815161705,
      "learning_rate": 6.92514910754567e-06,
      "loss": 0.0033,
      "step": 136200
    },
    {
      "epoch": 0.8736086302604279,
      "grad_norm": 0.10190891474485397,
      "learning_rate": 6.924632542927968e-06,
      "loss": 0.0009,
      "step": 136210
    },
    {
      "epoch": 0.8736727671542139,
      "grad_norm": 0.02483881264925003,
      "learning_rate": 6.924115954193247e-06,
      "loss": 0.0019,
      "step": 136220
    },
    {
      "epoch": 0.8737369040480001,
      "grad_norm": 0.081370048224926,
      "learning_rate": 6.923599341347975e-06,
      "loss": 0.0024,
      "step": 136230
    },
    {
      "epoch": 0.8738010409417861,
      "grad_norm": 0.041823867708444595,
      "learning_rate": 6.92308270439863e-06,
      "loss": 0.0025,
      "step": 136240
    },
    {
      "epoch": 0.8738651778355723,
      "grad_norm": 0.03406078368425369,
      "learning_rate": 6.922566043351684e-06,
      "loss": 0.0018,
      "step": 136250
    },
    {
      "epoch": 0.8739293147293583,
      "grad_norm": 0.10117413848638535,
      "learning_rate": 6.922049358213612e-06,
      "loss": 0.0022,
      "step": 136260
    },
    {
      "epoch": 0.8739934516231445,
      "grad_norm": 0.3346622586250305,
      "learning_rate": 6.9215326489908875e-06,
      "loss": 0.0023,
      "step": 136270
    },
    {
      "epoch": 0.8740575885169305,
      "grad_norm": 0.05216212570667267,
      "learning_rate": 6.921015915689985e-06,
      "loss": 0.0012,
      "step": 136280
    },
    {
      "epoch": 0.8741217254107166,
      "grad_norm": 0.062205005437135696,
      "learning_rate": 6.9204991583173805e-06,
      "loss": 0.0012,
      "step": 136290
    },
    {
      "epoch": 0.8741858623045027,
      "grad_norm": 0.029633566737174988,
      "learning_rate": 6.919982376879549e-06,
      "loss": 0.0054,
      "step": 136300
    },
    {
      "epoch": 0.8742499991982888,
      "grad_norm": 0.04072760045528412,
      "learning_rate": 6.919465571382966e-06,
      "loss": 0.0025,
      "step": 136310
    },
    {
      "epoch": 0.874314136092075,
      "grad_norm": 0.14085076749324799,
      "learning_rate": 6.9189487418341085e-06,
      "loss": 0.0025,
      "step": 136320
    },
    {
      "epoch": 0.874378272985861,
      "grad_norm": 0.21778567135334015,
      "learning_rate": 6.918431888239452e-06,
      "loss": 0.0041,
      "step": 136330
    },
    {
      "epoch": 0.8744424098796472,
      "grad_norm": 0.06518618762493134,
      "learning_rate": 6.917915010605471e-06,
      "loss": 0.0036,
      "step": 136340
    },
    {
      "epoch": 0.8745065467734332,
      "grad_norm": 0.10041414946317673,
      "learning_rate": 6.917398108938646e-06,
      "loss": 0.0025,
      "step": 136350
    },
    {
      "epoch": 0.8745706836672194,
      "grad_norm": 0.4006623327732086,
      "learning_rate": 6.9168811832454505e-06,
      "loss": 0.0042,
      "step": 136360
    },
    {
      "epoch": 0.8746348205610054,
      "grad_norm": 0.2387170046567917,
      "learning_rate": 6.916364233532366e-06,
      "loss": 0.0022,
      "step": 136370
    },
    {
      "epoch": 0.8746989574547915,
      "grad_norm": 0.04596169665455818,
      "learning_rate": 6.915847259805866e-06,
      "loss": 0.0041,
      "step": 136380
    },
    {
      "epoch": 0.8747630943485776,
      "grad_norm": 0.0908060297369957,
      "learning_rate": 6.915330262072433e-06,
      "loss": 0.0075,
      "step": 136390
    },
    {
      "epoch": 0.8748272312423637,
      "grad_norm": 0.11420466750860214,
      "learning_rate": 6.9148132403385405e-06,
      "loss": 0.0024,
      "step": 136400
    },
    {
      "epoch": 0.8748913681361498,
      "grad_norm": 0.09050630033016205,
      "learning_rate": 6.9142961946106705e-06,
      "loss": 0.0017,
      "step": 136410
    },
    {
      "epoch": 0.8749555050299359,
      "grad_norm": 0.1539943814277649,
      "learning_rate": 6.913779124895301e-06,
      "loss": 0.0039,
      "step": 136420
    },
    {
      "epoch": 0.875019641923722,
      "grad_norm": 0.18496167659759521,
      "learning_rate": 6.913262031198911e-06,
      "loss": 0.0029,
      "step": 136430
    },
    {
      "epoch": 0.8750837788175081,
      "grad_norm": 0.08429907262325287,
      "learning_rate": 6.9127449135279816e-06,
      "loss": 0.0017,
      "step": 136440
    },
    {
      "epoch": 0.8751479157112941,
      "grad_norm": 0.062099575996398926,
      "learning_rate": 6.91222777188899e-06,
      "loss": 0.0016,
      "step": 136450
    },
    {
      "epoch": 0.8752120526050803,
      "grad_norm": 0.075572669506073,
      "learning_rate": 6.911710606288419e-06,
      "loss": 0.0013,
      "step": 136460
    },
    {
      "epoch": 0.8752761894988664,
      "grad_norm": 0.11704915761947632,
      "learning_rate": 6.911193416732747e-06,
      "loss": 0.0013,
      "step": 136470
    },
    {
      "epoch": 0.8753403263926525,
      "grad_norm": 0.2985520660877228,
      "learning_rate": 6.910676203228456e-06,
      "loss": 0.0023,
      "step": 136480
    },
    {
      "epoch": 0.8754044632864386,
      "grad_norm": 0.1371159851551056,
      "learning_rate": 6.910158965782025e-06,
      "loss": 0.002,
      "step": 136490
    },
    {
      "epoch": 0.8754686001802247,
      "grad_norm": 0.07423175871372223,
      "learning_rate": 6.90964170439994e-06,
      "loss": 0.003,
      "step": 136500
    },
    {
      "epoch": 0.8755327370740108,
      "grad_norm": 0.05937831476330757,
      "learning_rate": 6.909124419088678e-06,
      "loss": 0.005,
      "step": 136510
    },
    {
      "epoch": 0.8755968739677968,
      "grad_norm": 0.40971502661705017,
      "learning_rate": 6.908607109854723e-06,
      "loss": 0.0028,
      "step": 136520
    },
    {
      "epoch": 0.875661010861583,
      "grad_norm": 0.1507779359817505,
      "learning_rate": 6.908089776704555e-06,
      "loss": 0.0016,
      "step": 136530
    },
    {
      "epoch": 0.875725147755369,
      "grad_norm": 0.1629064679145813,
      "learning_rate": 6.907572419644661e-06,
      "loss": 0.0015,
      "step": 136540
    },
    {
      "epoch": 0.8757892846491552,
      "grad_norm": 0.09428049623966217,
      "learning_rate": 6.9070550386815185e-06,
      "loss": 0.0026,
      "step": 136550
    },
    {
      "epoch": 0.8758534215429412,
      "grad_norm": 0.20758113265037537,
      "learning_rate": 6.906537633821616e-06,
      "loss": 0.0041,
      "step": 136560
    },
    {
      "epoch": 0.8759175584367274,
      "grad_norm": 0.028349295258522034,
      "learning_rate": 6.906020205071433e-06,
      "loss": 0.0036,
      "step": 136570
    },
    {
      "epoch": 0.8759816953305134,
      "grad_norm": 0.16029034554958344,
      "learning_rate": 6.905502752437455e-06,
      "loss": 0.0028,
      "step": 136580
    },
    {
      "epoch": 0.8760458322242995,
      "grad_norm": 0.10964485257863998,
      "learning_rate": 6.904985275926166e-06,
      "loss": 0.0032,
      "step": 136590
    },
    {
      "epoch": 0.8761099691180857,
      "grad_norm": 0.1821531057357788,
      "learning_rate": 6.90446777554405e-06,
      "loss": 0.004,
      "step": 136600
    },
    {
      "epoch": 0.8761741060118717,
      "grad_norm": 0.11746109277009964,
      "learning_rate": 6.903950251297591e-06,
      "loss": 0.0018,
      "step": 136610
    },
    {
      "epoch": 0.8762382429056579,
      "grad_norm": 0.017772624269127846,
      "learning_rate": 6.903432703193275e-06,
      "loss": 0.0015,
      "step": 136620
    },
    {
      "epoch": 0.8763023797994439,
      "grad_norm": 0.24109432101249695,
      "learning_rate": 6.902915131237586e-06,
      "loss": 0.0014,
      "step": 136630
    },
    {
      "epoch": 0.8763665166932301,
      "grad_norm": 0.0824224203824997,
      "learning_rate": 6.9023975354370125e-06,
      "loss": 0.0015,
      "step": 136640
    },
    {
      "epoch": 0.8764306535870161,
      "grad_norm": 0.15448006987571716,
      "learning_rate": 6.901879915798036e-06,
      "loss": 0.0023,
      "step": 136650
    },
    {
      "epoch": 0.8764947904808023,
      "grad_norm": 0.16880998015403748,
      "learning_rate": 6.901362272327147e-06,
      "loss": 0.0029,
      "step": 136660
    },
    {
      "epoch": 0.8765589273745883,
      "grad_norm": 0.32766395807266235,
      "learning_rate": 6.900844605030829e-06,
      "loss": 0.0043,
      "step": 136670
    },
    {
      "epoch": 0.8766230642683744,
      "grad_norm": 0.07112014293670654,
      "learning_rate": 6.900326913915569e-06,
      "loss": 0.0031,
      "step": 136680
    },
    {
      "epoch": 0.8766872011621605,
      "grad_norm": 0.15258969366550446,
      "learning_rate": 6.899809198987855e-06,
      "loss": 0.0023,
      "step": 136690
    },
    {
      "epoch": 0.8767513380559466,
      "grad_norm": 0.16571097075939178,
      "learning_rate": 6.8992914602541735e-06,
      "loss": 0.0019,
      "step": 136700
    },
    {
      "epoch": 0.8768154749497327,
      "grad_norm": 0.14944887161254883,
      "learning_rate": 6.898773697721014e-06,
      "loss": 0.0014,
      "step": 136710
    },
    {
      "epoch": 0.8768796118435188,
      "grad_norm": 0.08200392127037048,
      "learning_rate": 6.8982559113948625e-06,
      "loss": 0.0011,
      "step": 136720
    },
    {
      "epoch": 0.8769437487373049,
      "grad_norm": 0.12015662342309952,
      "learning_rate": 6.897738101282208e-06,
      "loss": 0.0037,
      "step": 136730
    },
    {
      "epoch": 0.877007885631091,
      "grad_norm": 0.24516578018665314,
      "learning_rate": 6.8972202673895375e-06,
      "loss": 0.0032,
      "step": 136740
    },
    {
      "epoch": 0.8770720225248771,
      "grad_norm": 0.07292426377534866,
      "learning_rate": 6.896702409723342e-06,
      "loss": 0.003,
      "step": 136750
    },
    {
      "epoch": 0.8771361594186632,
      "grad_norm": 0.008474384434521198,
      "learning_rate": 6.896184528290109e-06,
      "loss": 0.0013,
      "step": 136760
    },
    {
      "epoch": 0.8772002963124493,
      "grad_norm": 0.1347125917673111,
      "learning_rate": 6.8956666230963284e-06,
      "loss": 0.0035,
      "step": 136770
    },
    {
      "epoch": 0.8772644332062354,
      "grad_norm": 0.151192769408226,
      "learning_rate": 6.895148694148493e-06,
      "loss": 0.0026,
      "step": 136780
    },
    {
      "epoch": 0.8773285701000215,
      "grad_norm": 0.10084628313779831,
      "learning_rate": 6.894630741453087e-06,
      "loss": 0.003,
      "step": 136790
    },
    {
      "epoch": 0.8773927069938076,
      "grad_norm": 0.017331156879663467,
      "learning_rate": 6.8941127650166055e-06,
      "loss": 0.002,
      "step": 136800
    },
    {
      "epoch": 0.8774568438875937,
      "grad_norm": 0.045249033719301224,
      "learning_rate": 6.893594764845535e-06,
      "loss": 0.0021,
      "step": 136810
    },
    {
      "epoch": 0.8775209807813797,
      "grad_norm": 0.08385881036520004,
      "learning_rate": 6.893076740946371e-06,
      "loss": 0.0053,
      "step": 136820
    },
    {
      "epoch": 0.8775851176751659,
      "grad_norm": 0.11956585943698883,
      "learning_rate": 6.892558693325602e-06,
      "loss": 0.0015,
      "step": 136830
    },
    {
      "epoch": 0.8776492545689519,
      "grad_norm": 0.19696132838726044,
      "learning_rate": 6.892040621989721e-06,
      "loss": 0.0017,
      "step": 136840
    },
    {
      "epoch": 0.8777133914627381,
      "grad_norm": 0.04887941852211952,
      "learning_rate": 6.8915225269452165e-06,
      "loss": 0.0013,
      "step": 136850
    },
    {
      "epoch": 0.8777775283565241,
      "grad_norm": 0.12483085691928864,
      "learning_rate": 6.891004408198585e-06,
      "loss": 0.0034,
      "step": 136860
    },
    {
      "epoch": 0.8778416652503103,
      "grad_norm": 0.14218521118164062,
      "learning_rate": 6.890486265756316e-06,
      "loss": 0.0025,
      "step": 136870
    },
    {
      "epoch": 0.8779058021440964,
      "grad_norm": 0.2279711663722992,
      "learning_rate": 6.889968099624902e-06,
      "loss": 0.0028,
      "step": 136880
    },
    {
      "epoch": 0.8779699390378825,
      "grad_norm": 0.02050189860165119,
      "learning_rate": 6.889449909810838e-06,
      "loss": 0.002,
      "step": 136890
    },
    {
      "epoch": 0.8780340759316686,
      "grad_norm": 0.08747979253530502,
      "learning_rate": 6.8889316963206155e-06,
      "loss": 0.0022,
      "step": 136900
    },
    {
      "epoch": 0.8780982128254546,
      "grad_norm": 0.2141716182231903,
      "learning_rate": 6.88841345916073e-06,
      "loss": 0.0028,
      "step": 136910
    },
    {
      "epoch": 0.8781623497192408,
      "grad_norm": 0.04946558550000191,
      "learning_rate": 6.8878951983376725e-06,
      "loss": 0.0016,
      "step": 136920
    },
    {
      "epoch": 0.8782264866130268,
      "grad_norm": 0.24861359596252441,
      "learning_rate": 6.887376913857939e-06,
      "loss": 0.0022,
      "step": 136930
    },
    {
      "epoch": 0.878290623506813,
      "grad_norm": 0.0350680835545063,
      "learning_rate": 6.886858605728026e-06,
      "loss": 0.0071,
      "step": 136940
    },
    {
      "epoch": 0.878354760400599,
      "grad_norm": 0.02502746321260929,
      "learning_rate": 6.886340273954425e-06,
      "loss": 0.0016,
      "step": 136950
    },
    {
      "epoch": 0.8784188972943852,
      "grad_norm": 0.072413869202137,
      "learning_rate": 6.885821918543633e-06,
      "loss": 0.0017,
      "step": 136960
    },
    {
      "epoch": 0.8784830341881712,
      "grad_norm": 0.03558233380317688,
      "learning_rate": 6.885303539502144e-06,
      "loss": 0.0036,
      "step": 136970
    },
    {
      "epoch": 0.8785471710819573,
      "grad_norm": 0.05169002711772919,
      "learning_rate": 6.884785136836453e-06,
      "loss": 0.0008,
      "step": 136980
    },
    {
      "epoch": 0.8786113079757434,
      "grad_norm": 0.02297242172062397,
      "learning_rate": 6.884266710553059e-06,
      "loss": 0.0018,
      "step": 136990
    },
    {
      "epoch": 0.8786754448695295,
      "grad_norm": 0.18363507091999054,
      "learning_rate": 6.883748260658455e-06,
      "loss": 0.0024,
      "step": 137000
    },
    {
      "epoch": 0.8787395817633156,
      "grad_norm": 0.09243588894605637,
      "learning_rate": 6.88322978715914e-06,
      "loss": 0.0039,
      "step": 137010
    },
    {
      "epoch": 0.8788037186571017,
      "grad_norm": 0.1842324137687683,
      "learning_rate": 6.88271129006161e-06,
      "loss": 0.0044,
      "step": 137020
    },
    {
      "epoch": 0.8788678555508879,
      "grad_norm": 0.09502124041318893,
      "learning_rate": 6.882192769372362e-06,
      "loss": 0.0023,
      "step": 137030
    },
    {
      "epoch": 0.8789319924446739,
      "grad_norm": 0.08689358085393906,
      "learning_rate": 6.881674225097892e-06,
      "loss": 0.0019,
      "step": 137040
    },
    {
      "epoch": 0.87899612933846,
      "grad_norm": 0.03809691593050957,
      "learning_rate": 6.881155657244702e-06,
      "loss": 0.0015,
      "step": 137050
    },
    {
      "epoch": 0.8790602662322461,
      "grad_norm": 0.13645468652248383,
      "learning_rate": 6.880637065819284e-06,
      "loss": 0.0043,
      "step": 137060
    },
    {
      "epoch": 0.8791244031260322,
      "grad_norm": 0.14224575459957123,
      "learning_rate": 6.880118450828142e-06,
      "loss": 0.003,
      "step": 137070
    },
    {
      "epoch": 0.8791885400198183,
      "grad_norm": 0.11373443156480789,
      "learning_rate": 6.87959981227777e-06,
      "loss": 0.0014,
      "step": 137080
    },
    {
      "epoch": 0.8792526769136044,
      "grad_norm": 0.06638626754283905,
      "learning_rate": 6.87908115017467e-06,
      "loss": 0.003,
      "step": 137090
    },
    {
      "epoch": 0.8793168138073905,
      "grad_norm": 0.0888357162475586,
      "learning_rate": 6.87856246452534e-06,
      "loss": 0.0025,
      "step": 137100
    },
    {
      "epoch": 0.8793809507011766,
      "grad_norm": 0.26807594299316406,
      "learning_rate": 6.8780437553362785e-06,
      "loss": 0.0037,
      "step": 137110
    },
    {
      "epoch": 0.8794450875949626,
      "grad_norm": 0.04233105480670929,
      "learning_rate": 6.877525022613989e-06,
      "loss": 0.0029,
      "step": 137120
    },
    {
      "epoch": 0.8795092244887488,
      "grad_norm": 0.1305970698595047,
      "learning_rate": 6.877006266364967e-06,
      "loss": 0.0047,
      "step": 137130
    },
    {
      "epoch": 0.8795733613825348,
      "grad_norm": 0.15733136236667633,
      "learning_rate": 6.876487486595717e-06,
      "loss": 0.0031,
      "step": 137140
    },
    {
      "epoch": 0.879637498276321,
      "grad_norm": 0.023431070148944855,
      "learning_rate": 6.875968683312737e-06,
      "loss": 0.0015,
      "step": 137150
    },
    {
      "epoch": 0.879701635170107,
      "grad_norm": 0.34440213441848755,
      "learning_rate": 6.875449856522529e-06,
      "loss": 0.0016,
      "step": 137160
    },
    {
      "epoch": 0.8797657720638932,
      "grad_norm": 0.09305664151906967,
      "learning_rate": 6.874931006231593e-06,
      "loss": 0.0029,
      "step": 137170
    },
    {
      "epoch": 0.8798299089576793,
      "grad_norm": 0.07814795523881912,
      "learning_rate": 6.874412132446432e-06,
      "loss": 0.0018,
      "step": 137180
    },
    {
      "epoch": 0.8798940458514654,
      "grad_norm": 0.018024472519755363,
      "learning_rate": 6.8738932351735465e-06,
      "loss": 0.0066,
      "step": 137190
    },
    {
      "epoch": 0.8799581827452515,
      "grad_norm": 0.06308527290821075,
      "learning_rate": 6.873374314419441e-06,
      "loss": 0.0024,
      "step": 137200
    },
    {
      "epoch": 0.8800223196390375,
      "grad_norm": 0.06728116422891617,
      "learning_rate": 6.872855370190615e-06,
      "loss": 0.0018,
      "step": 137210
    },
    {
      "epoch": 0.8800864565328237,
      "grad_norm": 0.12315388023853302,
      "learning_rate": 6.8723364024935745e-06,
      "loss": 0.0035,
      "step": 137220
    },
    {
      "epoch": 0.8801505934266097,
      "grad_norm": 0.15933430194854736,
      "learning_rate": 6.871817411334819e-06,
      "loss": 0.0036,
      "step": 137230
    },
    {
      "epoch": 0.8802147303203959,
      "grad_norm": 0.09797076880931854,
      "learning_rate": 6.871298396720855e-06,
      "loss": 0.0011,
      "step": 137240
    },
    {
      "epoch": 0.8802788672141819,
      "grad_norm": 0.0730673298239708,
      "learning_rate": 6.870779358658183e-06,
      "loss": 0.0021,
      "step": 137250
    },
    {
      "epoch": 0.8803430041079681,
      "grad_norm": 0.15262871980667114,
      "learning_rate": 6.870260297153309e-06,
      "loss": 0.0021,
      "step": 137260
    },
    {
      "epoch": 0.8804071410017541,
      "grad_norm": 0.11767885833978653,
      "learning_rate": 6.869741212212738e-06,
      "loss": 0.0029,
      "step": 137270
    },
    {
      "epoch": 0.8804712778955402,
      "grad_norm": 0.2819925844669342,
      "learning_rate": 6.869222103842972e-06,
      "loss": 0.0037,
      "step": 137280
    },
    {
      "epoch": 0.8805354147893263,
      "grad_norm": 0.02453608252108097,
      "learning_rate": 6.868702972050518e-06,
      "loss": 0.0021,
      "step": 137290
    },
    {
      "epoch": 0.8805995516831124,
      "grad_norm": 0.127029150724411,
      "learning_rate": 6.86818381684188e-06,
      "loss": 0.0013,
      "step": 137300
    },
    {
      "epoch": 0.8806636885768986,
      "grad_norm": 0.04599674418568611,
      "learning_rate": 6.8676646382235635e-06,
      "loss": 0.002,
      "step": 137310
    },
    {
      "epoch": 0.8807278254706846,
      "grad_norm": 0.07534373551607132,
      "learning_rate": 6.867145436202074e-06,
      "loss": 0.0037,
      "step": 137320
    },
    {
      "epoch": 0.8807919623644708,
      "grad_norm": 0.2671842575073242,
      "learning_rate": 6.866626210783918e-06,
      "loss": 0.0034,
      "step": 137330
    },
    {
      "epoch": 0.8808560992582568,
      "grad_norm": 0.16933603584766388,
      "learning_rate": 6.866106961975602e-06,
      "loss": 0.0061,
      "step": 137340
    },
    {
      "epoch": 0.880920236152043,
      "grad_norm": 0.05069068819284439,
      "learning_rate": 6.865587689783631e-06,
      "loss": 0.0038,
      "step": 137350
    },
    {
      "epoch": 0.880984373045829,
      "grad_norm": 0.0438336543738842,
      "learning_rate": 6.865068394214514e-06,
      "loss": 0.0012,
      "step": 137360
    },
    {
      "epoch": 0.8810485099396151,
      "grad_norm": 0.10937488824129105,
      "learning_rate": 6.8645490752747575e-06,
      "loss": 0.0014,
      "step": 137370
    },
    {
      "epoch": 0.8811126468334012,
      "grad_norm": 0.20446348190307617,
      "learning_rate": 6.864029732970867e-06,
      "loss": 0.0015,
      "step": 137380
    },
    {
      "epoch": 0.8811767837271873,
      "grad_norm": 0.02750535123050213,
      "learning_rate": 6.863510367309353e-06,
      "loss": 0.002,
      "step": 137390
    },
    {
      "epoch": 0.8812409206209734,
      "grad_norm": 0.10042654722929001,
      "learning_rate": 6.862990978296722e-06,
      "loss": 0.0021,
      "step": 137400
    },
    {
      "epoch": 0.8813050575147595,
      "grad_norm": 0.01718439906835556,
      "learning_rate": 6.862471565939482e-06,
      "loss": 0.0025,
      "step": 137410
    },
    {
      "epoch": 0.8813691944085456,
      "grad_norm": 0.33396032452583313,
      "learning_rate": 6.861952130244143e-06,
      "loss": 0.0027,
      "step": 137420
    },
    {
      "epoch": 0.8814333313023317,
      "grad_norm": 0.12118028104305267,
      "learning_rate": 6.861432671217212e-06,
      "loss": 0.0016,
      "step": 137430
    },
    {
      "epoch": 0.8814974681961177,
      "grad_norm": 0.20148040354251862,
      "learning_rate": 6.8609131888652e-06,
      "loss": 0.0034,
      "step": 137440
    },
    {
      "epoch": 0.8815616050899039,
      "grad_norm": 0.08073599636554718,
      "learning_rate": 6.8603936831946165e-06,
      "loss": 0.0022,
      "step": 137450
    },
    {
      "epoch": 0.88162574198369,
      "grad_norm": 0.22954554855823517,
      "learning_rate": 6.859874154211969e-06,
      "loss": 0.0032,
      "step": 137460
    },
    {
      "epoch": 0.8816898788774761,
      "grad_norm": 0.07966429740190506,
      "learning_rate": 6.859354601923769e-06,
      "loss": 0.0022,
      "step": 137470
    },
    {
      "epoch": 0.8817540157712622,
      "grad_norm": 0.19060322642326355,
      "learning_rate": 6.858835026336529e-06,
      "loss": 0.002,
      "step": 137480
    },
    {
      "epoch": 0.8818181526650483,
      "grad_norm": 0.05918464809656143,
      "learning_rate": 6.858315427456755e-06,
      "loss": 0.0024,
      "step": 137490
    },
    {
      "epoch": 0.8818822895588344,
      "grad_norm": 0.14439141750335693,
      "learning_rate": 6.857795805290963e-06,
      "loss": 0.0025,
      "step": 137500
    },
    {
      "epoch": 0.8819464264526204,
      "grad_norm": 0.042854391038417816,
      "learning_rate": 6.857276159845661e-06,
      "loss": 0.0023,
      "step": 137510
    },
    {
      "epoch": 0.8820105633464066,
      "grad_norm": 0.12424889206886292,
      "learning_rate": 6.856756491127361e-06,
      "loss": 0.0035,
      "step": 137520
    },
    {
      "epoch": 0.8820747002401926,
      "grad_norm": 0.2014615684747696,
      "learning_rate": 6.856236799142575e-06,
      "loss": 0.0019,
      "step": 137530
    },
    {
      "epoch": 0.8821388371339788,
      "grad_norm": 0.06214706599712372,
      "learning_rate": 6.855717083897817e-06,
      "loss": 0.0027,
      "step": 137540
    },
    {
      "epoch": 0.8822029740277648,
      "grad_norm": 0.09805779159069061,
      "learning_rate": 6.8551973453995954e-06,
      "loss": 0.0013,
      "step": 137550
    },
    {
      "epoch": 0.882267110921551,
      "grad_norm": 0.025579053908586502,
      "learning_rate": 6.8546775836544264e-06,
      "loss": 0.0014,
      "step": 137560
    },
    {
      "epoch": 0.882331247815337,
      "grad_norm": 0.15716120600700378,
      "learning_rate": 6.854157798668821e-06,
      "loss": 0.0051,
      "step": 137570
    },
    {
      "epoch": 0.8823953847091232,
      "grad_norm": 0.11696451902389526,
      "learning_rate": 6.853637990449294e-06,
      "loss": 0.0024,
      "step": 137580
    },
    {
      "epoch": 0.8824595216029093,
      "grad_norm": 0.06991618126630783,
      "learning_rate": 6.853118159002357e-06,
      "loss": 0.003,
      "step": 137590
    },
    {
      "epoch": 0.8825236584966953,
      "grad_norm": 0.18459399044513702,
      "learning_rate": 6.852598304334528e-06,
      "loss": 0.0023,
      "step": 137600
    },
    {
      "epoch": 0.8825877953904815,
      "grad_norm": 0.07764824479818344,
      "learning_rate": 6.852078426452315e-06,
      "loss": 0.0016,
      "step": 137610
    },
    {
      "epoch": 0.8826519322842675,
      "grad_norm": 0.04978380724787712,
      "learning_rate": 6.851558525362236e-06,
      "loss": 0.0023,
      "step": 137620
    },
    {
      "epoch": 0.8827160691780537,
      "grad_norm": 0.08767449855804443,
      "learning_rate": 6.851038601070808e-06,
      "loss": 0.0009,
      "step": 137630
    },
    {
      "epoch": 0.8827802060718397,
      "grad_norm": 0.009009142406284809,
      "learning_rate": 6.8505186535845405e-06,
      "loss": 0.002,
      "step": 137640
    },
    {
      "epoch": 0.8828443429656259,
      "grad_norm": 0.2602931261062622,
      "learning_rate": 6.849998682909953e-06,
      "loss": 0.0027,
      "step": 137650
    },
    {
      "epoch": 0.8829084798594119,
      "grad_norm": 0.2864200174808502,
      "learning_rate": 6.849478689053559e-06,
      "loss": 0.0057,
      "step": 137660
    },
    {
      "epoch": 0.882972616753198,
      "grad_norm": 0.08893916755914688,
      "learning_rate": 6.848958672021877e-06,
      "loss": 0.0071,
      "step": 137670
    },
    {
      "epoch": 0.8830367536469841,
      "grad_norm": 0.1800779104232788,
      "learning_rate": 6.848438631821419e-06,
      "loss": 0.0037,
      "step": 137680
    },
    {
      "epoch": 0.8831008905407702,
      "grad_norm": 0.06310734897851944,
      "learning_rate": 6.847918568458707e-06,
      "loss": 0.0032,
      "step": 137690
    },
    {
      "epoch": 0.8831650274345563,
      "grad_norm": 0.13706275820732117,
      "learning_rate": 6.84739848194025e-06,
      "loss": 0.003,
      "step": 137700
    },
    {
      "epoch": 0.8832291643283424,
      "grad_norm": 0.031421709805727005,
      "learning_rate": 6.846878372272574e-06,
      "loss": 0.0024,
      "step": 137710
    },
    {
      "epoch": 0.8832933012221285,
      "grad_norm": 0.10083822906017303,
      "learning_rate": 6.8463582394621895e-06,
      "loss": 0.0024,
      "step": 137720
    },
    {
      "epoch": 0.8833574381159146,
      "grad_norm": 0.32716020941734314,
      "learning_rate": 6.845838083515619e-06,
      "loss": 0.0048,
      "step": 137730
    },
    {
      "epoch": 0.8834215750097008,
      "grad_norm": 0.07156256586313248,
      "learning_rate": 6.8453179044393755e-06,
      "loss": 0.0017,
      "step": 137740
    },
    {
      "epoch": 0.8834857119034868,
      "grad_norm": 0.6506041884422302,
      "learning_rate": 6.844797702239983e-06,
      "loss": 0.002,
      "step": 137750
    },
    {
      "epoch": 0.8835498487972729,
      "grad_norm": 0.12119151651859283,
      "learning_rate": 6.844277476923954e-06,
      "loss": 0.0036,
      "step": 137760
    },
    {
      "epoch": 0.883613985691059,
      "grad_norm": 0.008514746092259884,
      "learning_rate": 6.843757228497811e-06,
      "loss": 0.0012,
      "step": 137770
    },
    {
      "epoch": 0.8836781225848451,
      "grad_norm": 0.13601401448249817,
      "learning_rate": 6.843236956968072e-06,
      "loss": 0.0032,
      "step": 137780
    },
    {
      "epoch": 0.8837422594786312,
      "grad_norm": 0.320917546749115,
      "learning_rate": 6.842716662341258e-06,
      "loss": 0.003,
      "step": 137790
    },
    {
      "epoch": 0.8838063963724173,
      "grad_norm": 0.052433934062719345,
      "learning_rate": 6.842196344623886e-06,
      "loss": 0.0021,
      "step": 137800
    },
    {
      "epoch": 0.8838705332662034,
      "grad_norm": 0.12566226720809937,
      "learning_rate": 6.841676003822477e-06,
      "loss": 0.0019,
      "step": 137810
    },
    {
      "epoch": 0.8839346701599895,
      "grad_norm": 0.10092577338218689,
      "learning_rate": 6.841155639943552e-06,
      "loss": 0.0063,
      "step": 137820
    },
    {
      "epoch": 0.8839988070537755,
      "grad_norm": 0.044915057718753815,
      "learning_rate": 6.84063525299363e-06,
      "loss": 0.0025,
      "step": 137830
    },
    {
      "epoch": 0.8840629439475617,
      "grad_norm": 0.11170393228530884,
      "learning_rate": 6.8401148429792355e-06,
      "loss": 0.0022,
      "step": 137840
    },
    {
      "epoch": 0.8841270808413477,
      "grad_norm": 0.08210773766040802,
      "learning_rate": 6.8395944099068835e-06,
      "loss": 0.0016,
      "step": 137850
    },
    {
      "epoch": 0.8841912177351339,
      "grad_norm": 0.12872573733329773,
      "learning_rate": 6.839073953783101e-06,
      "loss": 0.002,
      "step": 137860
    },
    {
      "epoch": 0.88425535462892,
      "grad_norm": 0.1600671112537384,
      "learning_rate": 6.838553474614407e-06,
      "loss": 0.0035,
      "step": 137870
    },
    {
      "epoch": 0.8843194915227061,
      "grad_norm": 0.06768742203712463,
      "learning_rate": 6.838032972407324e-06,
      "loss": 0.0025,
      "step": 137880
    },
    {
      "epoch": 0.8843836284164922,
      "grad_norm": 0.04691407456994057,
      "learning_rate": 6.837512447168373e-06,
      "loss": 0.0023,
      "step": 137890
    },
    {
      "epoch": 0.8844477653102782,
      "grad_norm": 0.06426296383142471,
      "learning_rate": 6.83699189890408e-06,
      "loss": 0.0026,
      "step": 137900
    },
    {
      "epoch": 0.8845119022040644,
      "grad_norm": 0.24463340640068054,
      "learning_rate": 6.836471327620964e-06,
      "loss": 0.0025,
      "step": 137910
    },
    {
      "epoch": 0.8845760390978504,
      "grad_norm": 0.07521167397499084,
      "learning_rate": 6.8359507333255505e-06,
      "loss": 0.0029,
      "step": 137920
    },
    {
      "epoch": 0.8846401759916366,
      "grad_norm": 0.19357682764530182,
      "learning_rate": 6.835430116024362e-06,
      "loss": 0.0022,
      "step": 137930
    },
    {
      "epoch": 0.8847043128854226,
      "grad_norm": 0.11128903925418854,
      "learning_rate": 6.834909475723923e-06,
      "loss": 0.0074,
      "step": 137940
    },
    {
      "epoch": 0.8847684497792088,
      "grad_norm": 0.09577079862356186,
      "learning_rate": 6.834388812430756e-06,
      "loss": 0.0024,
      "step": 137950
    },
    {
      "epoch": 0.8848325866729948,
      "grad_norm": 0.07646466046571732,
      "learning_rate": 6.833868126151385e-06,
      "loss": 0.002,
      "step": 137960
    },
    {
      "epoch": 0.884896723566781,
      "grad_norm": 0.020462574437260628,
      "learning_rate": 6.833347416892338e-06,
      "loss": 0.0027,
      "step": 137970
    },
    {
      "epoch": 0.884960860460567,
      "grad_norm": 0.16638454794883728,
      "learning_rate": 6.832826684660137e-06,
      "loss": 0.0035,
      "step": 137980
    },
    {
      "epoch": 0.8850249973543531,
      "grad_norm": 0.14669720828533173,
      "learning_rate": 6.832305929461307e-06,
      "loss": 0.0025,
      "step": 137990
    },
    {
      "epoch": 0.8850891342481392,
      "grad_norm": 0.09566580504179001,
      "learning_rate": 6.831785151302373e-06,
      "loss": 0.0032,
      "step": 138000
    },
    {
      "epoch": 0.8851532711419253,
      "grad_norm": 0.04054740443825722,
      "learning_rate": 6.831264350189866e-06,
      "loss": 0.0031,
      "step": 138010
    },
    {
      "epoch": 0.8852174080357115,
      "grad_norm": 0.15328174829483032,
      "learning_rate": 6.8307435261303035e-06,
      "loss": 0.0029,
      "step": 138020
    },
    {
      "epoch": 0.8852815449294975,
      "grad_norm": 0.11700726300477982,
      "learning_rate": 6.830222679130219e-06,
      "loss": 0.0023,
      "step": 138030
    },
    {
      "epoch": 0.8853456818232837,
      "grad_norm": 0.10539217293262482,
      "learning_rate": 6.829701809196136e-06,
      "loss": 0.0027,
      "step": 138040
    },
    {
      "epoch": 0.8854098187170697,
      "grad_norm": 0.2931744158267975,
      "learning_rate": 6.82918091633458e-06,
      "loss": 0.0031,
      "step": 138050
    },
    {
      "epoch": 0.8854739556108558,
      "grad_norm": 0.07683542370796204,
      "learning_rate": 6.828660000552081e-06,
      "loss": 0.0023,
      "step": 138060
    },
    {
      "epoch": 0.8855380925046419,
      "grad_norm": 0.557079017162323,
      "learning_rate": 6.828139061855165e-06,
      "loss": 0.002,
      "step": 138070
    },
    {
      "epoch": 0.885602229398428,
      "grad_norm": 0.08445441722869873,
      "learning_rate": 6.82761810025036e-06,
      "loss": 0.0065,
      "step": 138080
    },
    {
      "epoch": 0.8856663662922141,
      "grad_norm": 0.10118082910776138,
      "learning_rate": 6.827097115744195e-06,
      "loss": 0.0017,
      "step": 138090
    },
    {
      "epoch": 0.8857305031860002,
      "grad_norm": 0.09960382431745529,
      "learning_rate": 6.826576108343195e-06,
      "loss": 0.0028,
      "step": 138100
    },
    {
      "epoch": 0.8857946400797863,
      "grad_norm": 0.12642377614974976,
      "learning_rate": 6.826055078053893e-06,
      "loss": 0.0026,
      "step": 138110
    },
    {
      "epoch": 0.8858587769735724,
      "grad_norm": 0.20992155373096466,
      "learning_rate": 6.825534024882815e-06,
      "loss": 0.0026,
      "step": 138120
    },
    {
      "epoch": 0.8859229138673584,
      "grad_norm": 0.1381525695323944,
      "learning_rate": 6.8250129488364915e-06,
      "loss": 0.0017,
      "step": 138130
    },
    {
      "epoch": 0.8859870507611446,
      "grad_norm": 0.056542061269283295,
      "learning_rate": 6.824491849921451e-06,
      "loss": 0.0022,
      "step": 138140
    },
    {
      "epoch": 0.8860511876549307,
      "grad_norm": 0.06584298610687256,
      "learning_rate": 6.823970728144225e-06,
      "loss": 0.0021,
      "step": 138150
    },
    {
      "epoch": 0.8861153245487168,
      "grad_norm": 0.13849897682666779,
      "learning_rate": 6.823449583511339e-06,
      "loss": 0.0022,
      "step": 138160
    },
    {
      "epoch": 0.8861794614425029,
      "grad_norm": 0.09438132494688034,
      "learning_rate": 6.822928416029329e-06,
      "loss": 0.0037,
      "step": 138170
    },
    {
      "epoch": 0.886243598336289,
      "grad_norm": 0.08102027326822281,
      "learning_rate": 6.8224072257047225e-06,
      "loss": 0.0022,
      "step": 138180
    },
    {
      "epoch": 0.8863077352300751,
      "grad_norm": 0.1010395735502243,
      "learning_rate": 6.821886012544051e-06,
      "loss": 0.0023,
      "step": 138190
    },
    {
      "epoch": 0.8863718721238611,
      "grad_norm": 0.18256144225597382,
      "learning_rate": 6.8213647765538475e-06,
      "loss": 0.0019,
      "step": 138200
    },
    {
      "epoch": 0.8864360090176473,
      "grad_norm": 0.08884212374687195,
      "learning_rate": 6.820843517740638e-06,
      "loss": 0.0017,
      "step": 138210
    },
    {
      "epoch": 0.8865001459114333,
      "grad_norm": 0.20112930238246918,
      "learning_rate": 6.820322236110961e-06,
      "loss": 0.0021,
      "step": 138220
    },
    {
      "epoch": 0.8865642828052195,
      "grad_norm": 0.09102907031774521,
      "learning_rate": 6.8198009316713435e-06,
      "loss": 0.0019,
      "step": 138230
    },
    {
      "epoch": 0.8866284196990055,
      "grad_norm": 0.05903204157948494,
      "learning_rate": 6.819279604428322e-06,
      "loss": 0.0019,
      "step": 138240
    },
    {
      "epoch": 0.8866925565927917,
      "grad_norm": 0.2465900033712387,
      "learning_rate": 6.818758254388424e-06,
      "loss": 0.0047,
      "step": 138250
    },
    {
      "epoch": 0.8867566934865777,
      "grad_norm": 0.10835077613592148,
      "learning_rate": 6.818236881558187e-06,
      "loss": 0.0023,
      "step": 138260
    },
    {
      "epoch": 0.8868208303803639,
      "grad_norm": 0.03426966816186905,
      "learning_rate": 6.817715485944142e-06,
      "loss": 0.002,
      "step": 138270
    },
    {
      "epoch": 0.8868849672741499,
      "grad_norm": 0.049255553632974625,
      "learning_rate": 6.817194067552824e-06,
      "loss": 0.0028,
      "step": 138280
    },
    {
      "epoch": 0.886949104167936,
      "grad_norm": 0.06245843693614006,
      "learning_rate": 6.816672626390763e-06,
      "loss": 0.0017,
      "step": 138290
    },
    {
      "epoch": 0.8870132410617222,
      "grad_norm": 0.14906224608421326,
      "learning_rate": 6.816151162464498e-06,
      "loss": 0.0026,
      "step": 138300
    },
    {
      "epoch": 0.8870773779555082,
      "grad_norm": 0.10730446875095367,
      "learning_rate": 6.815629675780559e-06,
      "loss": 0.0021,
      "step": 138310
    },
    {
      "epoch": 0.8871415148492944,
      "grad_norm": 0.12750211358070374,
      "learning_rate": 6.815108166345483e-06,
      "loss": 0.0029,
      "step": 138320
    },
    {
      "epoch": 0.8872056517430804,
      "grad_norm": 0.0802149772644043,
      "learning_rate": 6.814586634165806e-06,
      "loss": 0.0017,
      "step": 138330
    },
    {
      "epoch": 0.8872697886368666,
      "grad_norm": 0.3117047846317291,
      "learning_rate": 6.81406507924806e-06,
      "loss": 0.0034,
      "step": 138340
    },
    {
      "epoch": 0.8873339255306526,
      "grad_norm": 0.03203052654862404,
      "learning_rate": 6.813543501598784e-06,
      "loss": 0.0025,
      "step": 138350
    },
    {
      "epoch": 0.8873980624244387,
      "grad_norm": 0.2867416739463806,
      "learning_rate": 6.81302190122451e-06,
      "loss": 0.004,
      "step": 138360
    },
    {
      "epoch": 0.8874621993182248,
      "grad_norm": 0.1543174386024475,
      "learning_rate": 6.812500278131776e-06,
      "loss": 0.0024,
      "step": 138370
    },
    {
      "epoch": 0.8875263362120109,
      "grad_norm": 0.07901715487241745,
      "learning_rate": 6.811978632327119e-06,
      "loss": 0.0023,
      "step": 138380
    },
    {
      "epoch": 0.887590473105797,
      "grad_norm": 0.06686260551214218,
      "learning_rate": 6.811456963817075e-06,
      "loss": 0.0023,
      "step": 138390
    },
    {
      "epoch": 0.8876546099995831,
      "grad_norm": 0.12487843632698059,
      "learning_rate": 6.810935272608179e-06,
      "loss": 0.0059,
      "step": 138400
    },
    {
      "epoch": 0.8877187468933692,
      "grad_norm": 0.06749287992715836,
      "learning_rate": 6.8104135587069704e-06,
      "loss": 0.0014,
      "step": 138410
    },
    {
      "epoch": 0.8877828837871553,
      "grad_norm": 0.05625376105308533,
      "learning_rate": 6.8098918221199864e-06,
      "loss": 0.0024,
      "step": 138420
    },
    {
      "epoch": 0.8878470206809415,
      "grad_norm": 0.044884275645017624,
      "learning_rate": 6.809370062853764e-06,
      "loss": 0.0008,
      "step": 138430
    },
    {
      "epoch": 0.8879111575747275,
      "grad_norm": 0.23536013066768646,
      "learning_rate": 6.808848280914842e-06,
      "loss": 0.003,
      "step": 138440
    },
    {
      "epoch": 0.8879752944685136,
      "grad_norm": 0.06617274135351181,
      "learning_rate": 6.808326476309759e-06,
      "loss": 0.0022,
      "step": 138450
    },
    {
      "epoch": 0.8880394313622997,
      "grad_norm": 0.05690762773156166,
      "learning_rate": 6.80780464904505e-06,
      "loss": 0.0013,
      "step": 138460
    },
    {
      "epoch": 0.8881035682560858,
      "grad_norm": 0.13122613728046417,
      "learning_rate": 6.807282799127259e-06,
      "loss": 0.0034,
      "step": 138470
    },
    {
      "epoch": 0.8881677051498719,
      "grad_norm": 0.00365577545017004,
      "learning_rate": 6.806760926562922e-06,
      "loss": 0.0015,
      "step": 138480
    },
    {
      "epoch": 0.888231842043658,
      "grad_norm": 0.05660828575491905,
      "learning_rate": 6.80623903135858e-06,
      "loss": 0.0019,
      "step": 138490
    },
    {
      "epoch": 0.888295978937444,
      "grad_norm": 0.29124340415000916,
      "learning_rate": 6.805717113520771e-06,
      "loss": 0.0024,
      "step": 138500
    },
    {
      "epoch": 0.8883601158312302,
      "grad_norm": 0.06867077946662903,
      "learning_rate": 6.805195173056039e-06,
      "loss": 0.0028,
      "step": 138510
    },
    {
      "epoch": 0.8884242527250162,
      "grad_norm": 0.18916812539100647,
      "learning_rate": 6.804673209970918e-06,
      "loss": 0.0032,
      "step": 138520
    },
    {
      "epoch": 0.8884883896188024,
      "grad_norm": 0.06216156482696533,
      "learning_rate": 6.804151224271954e-06,
      "loss": 0.0039,
      "step": 138530
    },
    {
      "epoch": 0.8885525265125884,
      "grad_norm": 0.24915002286434174,
      "learning_rate": 6.803629215965684e-06,
      "loss": 0.0026,
      "step": 138540
    },
    {
      "epoch": 0.8886166634063746,
      "grad_norm": 0.09341257065534592,
      "learning_rate": 6.803107185058651e-06,
      "loss": 0.0022,
      "step": 138550
    },
    {
      "epoch": 0.8886808003001606,
      "grad_norm": 0.05511720851063728,
      "learning_rate": 6.802585131557398e-06,
      "loss": 0.0009,
      "step": 138560
    },
    {
      "epoch": 0.8887449371939468,
      "grad_norm": 0.13346849381923676,
      "learning_rate": 6.802063055468464e-06,
      "loss": 0.0026,
      "step": 138570
    },
    {
      "epoch": 0.8888090740877329,
      "grad_norm": 0.1426878720521927,
      "learning_rate": 6.801540956798393e-06,
      "loss": 0.0039,
      "step": 138580
    },
    {
      "epoch": 0.888873210981519,
      "grad_norm": 0.02670944854617119,
      "learning_rate": 6.8010188355537245e-06,
      "loss": 0.0012,
      "step": 138590
    },
    {
      "epoch": 0.8889373478753051,
      "grad_norm": 0.12257665395736694,
      "learning_rate": 6.800496691741004e-06,
      "loss": 0.0042,
      "step": 138600
    },
    {
      "epoch": 0.8890014847690911,
      "grad_norm": 0.12118265777826309,
      "learning_rate": 6.799974525366773e-06,
      "loss": 0.0026,
      "step": 138610
    },
    {
      "epoch": 0.8890656216628773,
      "grad_norm": 0.058597929775714874,
      "learning_rate": 6.7994523364375744e-06,
      "loss": 0.0019,
      "step": 138620
    },
    {
      "epoch": 0.8891297585566633,
      "grad_norm": 0.09288131445646286,
      "learning_rate": 6.798930124959952e-06,
      "loss": 0.0043,
      "step": 138630
    },
    {
      "epoch": 0.8891938954504495,
      "grad_norm": 0.18174026906490326,
      "learning_rate": 6.79840789094045e-06,
      "loss": 0.002,
      "step": 138640
    },
    {
      "epoch": 0.8892580323442355,
      "grad_norm": 0.049296993762254715,
      "learning_rate": 6.797885634385612e-06,
      "loss": 0.0033,
      "step": 138650
    },
    {
      "epoch": 0.8893221692380217,
      "grad_norm": 0.07419116795063019,
      "learning_rate": 6.797363355301981e-06,
      "loss": 0.0035,
      "step": 138660
    },
    {
      "epoch": 0.8893863061318077,
      "grad_norm": 0.17123816907405853,
      "learning_rate": 6.796841053696102e-06,
      "loss": 0.0036,
      "step": 138670
    },
    {
      "epoch": 0.8894504430255938,
      "grad_norm": 0.08635471761226654,
      "learning_rate": 6.796318729574522e-06,
      "loss": 0.0026,
      "step": 138680
    },
    {
      "epoch": 0.8895145799193799,
      "grad_norm": 0.15406401455402374,
      "learning_rate": 6.7957963829437845e-06,
      "loss": 0.0035,
      "step": 138690
    },
    {
      "epoch": 0.889578716813166,
      "grad_norm": 0.08034338802099228,
      "learning_rate": 6.795274013810435e-06,
      "loss": 0.0023,
      "step": 138700
    },
    {
      "epoch": 0.8896428537069521,
      "grad_norm": 0.11541954427957535,
      "learning_rate": 6.794751622181018e-06,
      "loss": 0.0024,
      "step": 138710
    },
    {
      "epoch": 0.8897069906007382,
      "grad_norm": 0.15283158421516418,
      "learning_rate": 6.794229208062081e-06,
      "loss": 0.0023,
      "step": 138720
    },
    {
      "epoch": 0.8897711274945244,
      "grad_norm": 0.044942475855350494,
      "learning_rate": 6.79370677146017e-06,
      "loss": 0.0034,
      "step": 138730
    },
    {
      "epoch": 0.8898352643883104,
      "grad_norm": 0.08287149667739868,
      "learning_rate": 6.793184312381831e-06,
      "loss": 0.0031,
      "step": 138740
    },
    {
      "epoch": 0.8898994012820965,
      "grad_norm": 0.07617488503456116,
      "learning_rate": 6.792661830833611e-06,
      "loss": 0.0016,
      "step": 138750
    },
    {
      "epoch": 0.8899635381758826,
      "grad_norm": 0.13483263552188873,
      "learning_rate": 6.792139326822056e-06,
      "loss": 0.0044,
      "step": 138760
    },
    {
      "epoch": 0.8900276750696687,
      "grad_norm": 0.16076132655143738,
      "learning_rate": 6.791616800353716e-06,
      "loss": 0.0019,
      "step": 138770
    },
    {
      "epoch": 0.8900918119634548,
      "grad_norm": 0.10660827159881592,
      "learning_rate": 6.791094251435137e-06,
      "loss": 0.0021,
      "step": 138780
    },
    {
      "epoch": 0.8901559488572409,
      "grad_norm": 0.1787632703781128,
      "learning_rate": 6.7905716800728664e-06,
      "loss": 0.0016,
      "step": 138790
    },
    {
      "epoch": 0.890220085751027,
      "grad_norm": 0.04764917492866516,
      "learning_rate": 6.7900490862734525e-06,
      "loss": 0.0026,
      "step": 138800
    },
    {
      "epoch": 0.8902842226448131,
      "grad_norm": 0.04433861002326012,
      "learning_rate": 6.789526470043444e-06,
      "loss": 0.0027,
      "step": 138810
    },
    {
      "epoch": 0.8903483595385991,
      "grad_norm": 0.13696281611919403,
      "learning_rate": 6.789003831389391e-06,
      "loss": 0.0034,
      "step": 138820
    },
    {
      "epoch": 0.8904124964323853,
      "grad_norm": 0.08294308930635452,
      "learning_rate": 6.78848117031784e-06,
      "loss": 0.0035,
      "step": 138830
    },
    {
      "epoch": 0.8904766333261713,
      "grad_norm": 0.07750234752893448,
      "learning_rate": 6.787958486835342e-06,
      "loss": 0.005,
      "step": 138840
    },
    {
      "epoch": 0.8905407702199575,
      "grad_norm": 0.1333630234003067,
      "learning_rate": 6.787435780948448e-06,
      "loss": 0.0021,
      "step": 138850
    },
    {
      "epoch": 0.8906049071137436,
      "grad_norm": 0.3120582699775696,
      "learning_rate": 6.786913052663705e-06,
      "loss": 0.0024,
      "step": 138860
    },
    {
      "epoch": 0.8906690440075297,
      "grad_norm": 0.21088089048862457,
      "learning_rate": 6.786390301987664e-06,
      "loss": 0.0021,
      "step": 138870
    },
    {
      "epoch": 0.8907331809013158,
      "grad_norm": 0.14093145728111267,
      "learning_rate": 6.785867528926877e-06,
      "loss": 0.0044,
      "step": 138880
    },
    {
      "epoch": 0.8907973177951019,
      "grad_norm": 0.12831830978393555,
      "learning_rate": 6.785344733487892e-06,
      "loss": 0.0045,
      "step": 138890
    },
    {
      "epoch": 0.890861454688888,
      "grad_norm": 0.13694891333580017,
      "learning_rate": 6.784821915677264e-06,
      "loss": 0.0013,
      "step": 138900
    },
    {
      "epoch": 0.890925591582674,
      "grad_norm": 0.09416618198156357,
      "learning_rate": 6.784299075501539e-06,
      "loss": 0.0016,
      "step": 138910
    },
    {
      "epoch": 0.8909897284764602,
      "grad_norm": 0.2192876935005188,
      "learning_rate": 6.7837762129672725e-06,
      "loss": 0.0017,
      "step": 138920
    },
    {
      "epoch": 0.8910538653702462,
      "grad_norm": 0.12382154911756516,
      "learning_rate": 6.783253328081015e-06,
      "loss": 0.0019,
      "step": 138930
    },
    {
      "epoch": 0.8911180022640324,
      "grad_norm": 0.15769800543785095,
      "learning_rate": 6.782730420849319e-06,
      "loss": 0.002,
      "step": 138940
    },
    {
      "epoch": 0.8911821391578184,
      "grad_norm": 0.017192568629980087,
      "learning_rate": 6.782207491278738e-06,
      "loss": 0.0024,
      "step": 138950
    },
    {
      "epoch": 0.8912462760516046,
      "grad_norm": 0.07215219736099243,
      "learning_rate": 6.781684539375822e-06,
      "loss": 0.002,
      "step": 138960
    },
    {
      "epoch": 0.8913104129453906,
      "grad_norm": 0.14700429141521454,
      "learning_rate": 6.781161565147125e-06,
      "loss": 0.0025,
      "step": 138970
    },
    {
      "epoch": 0.8913745498391767,
      "grad_norm": 0.0878363698720932,
      "learning_rate": 6.780638568599203e-06,
      "loss": 0.0023,
      "step": 138980
    },
    {
      "epoch": 0.8914386867329628,
      "grad_norm": 0.1149219200015068,
      "learning_rate": 6.780115549738604e-06,
      "loss": 0.003,
      "step": 138990
    },
    {
      "epoch": 0.8915028236267489,
      "grad_norm": 0.16724520921707153,
      "learning_rate": 6.779592508571886e-06,
      "loss": 0.0014,
      "step": 139000
    },
    {
      "epoch": 0.8915669605205351,
      "grad_norm": 0.2597067952156067,
      "learning_rate": 6.779069445105603e-06,
      "loss": 0.0025,
      "step": 139010
    },
    {
      "epoch": 0.8916310974143211,
      "grad_norm": 0.08588556200265884,
      "learning_rate": 6.7785463593463066e-06,
      "loss": 0.0017,
      "step": 139020
    },
    {
      "epoch": 0.8916952343081073,
      "grad_norm": 0.10413733124732971,
      "learning_rate": 6.778023251300555e-06,
      "loss": 0.0013,
      "step": 139030
    },
    {
      "epoch": 0.8917593712018933,
      "grad_norm": 0.052861861884593964,
      "learning_rate": 6.7775001209749005e-06,
      "loss": 0.0018,
      "step": 139040
    },
    {
      "epoch": 0.8918235080956795,
      "grad_norm": 0.2284519225358963,
      "learning_rate": 6.776976968375899e-06,
      "loss": 0.0041,
      "step": 139050
    },
    {
      "epoch": 0.8918876449894655,
      "grad_norm": 0.21057339012622833,
      "learning_rate": 6.776453793510106e-06,
      "loss": 0.002,
      "step": 139060
    },
    {
      "epoch": 0.8919517818832516,
      "grad_norm": 0.0584498755633831,
      "learning_rate": 6.775930596384078e-06,
      "loss": 0.0035,
      "step": 139070
    },
    {
      "epoch": 0.8920159187770377,
      "grad_norm": 0.16429851949214935,
      "learning_rate": 6.77540737700437e-06,
      "loss": 0.0029,
      "step": 139080
    },
    {
      "epoch": 0.8920800556708238,
      "grad_norm": 0.0653160810470581,
      "learning_rate": 6.774884135377539e-06,
      "loss": 0.0017,
      "step": 139090
    },
    {
      "epoch": 0.8921441925646099,
      "grad_norm": 0.060079559683799744,
      "learning_rate": 6.774360871510142e-06,
      "loss": 0.0017,
      "step": 139100
    },
    {
      "epoch": 0.892208329458396,
      "grad_norm": 0.049049098044633865,
      "learning_rate": 6.773837585408734e-06,
      "loss": 0.0015,
      "step": 139110
    },
    {
      "epoch": 0.892272466352182,
      "grad_norm": 0.18999944627285004,
      "learning_rate": 6.7733142770798735e-06,
      "loss": 0.0031,
      "step": 139120
    },
    {
      "epoch": 0.8923366032459682,
      "grad_norm": 0.15021562576293945,
      "learning_rate": 6.772790946530118e-06,
      "loss": 0.0018,
      "step": 139130
    },
    {
      "epoch": 0.8924007401397543,
      "grad_norm": 0.2060428410768509,
      "learning_rate": 6.7722675937660244e-06,
      "loss": 0.002,
      "step": 139140
    },
    {
      "epoch": 0.8924648770335404,
      "grad_norm": 0.12639163434505463,
      "learning_rate": 6.771744218794152e-06,
      "loss": 0.0016,
      "step": 139150
    },
    {
      "epoch": 0.8925290139273265,
      "grad_norm": 0.27944356203079224,
      "learning_rate": 6.771220821621057e-06,
      "loss": 0.0028,
      "step": 139160
    },
    {
      "epoch": 0.8925931508211126,
      "grad_norm": 0.1983378380537033,
      "learning_rate": 6.770697402253299e-06,
      "loss": 0.0028,
      "step": 139170
    },
    {
      "epoch": 0.8926572877148987,
      "grad_norm": 0.16844449937343597,
      "learning_rate": 6.770173960697439e-06,
      "loss": 0.0026,
      "step": 139180
    },
    {
      "epoch": 0.8927214246086848,
      "grad_norm": 0.1559169888496399,
      "learning_rate": 6.769650496960033e-06,
      "loss": 0.0015,
      "step": 139190
    },
    {
      "epoch": 0.8927855615024709,
      "grad_norm": 0.01990179345011711,
      "learning_rate": 6.769127011047642e-06,
      "loss": 0.0013,
      "step": 139200
    },
    {
      "epoch": 0.8928496983962569,
      "grad_norm": 0.07438834756612778,
      "learning_rate": 6.7686035029668255e-06,
      "loss": 0.0024,
      "step": 139210
    },
    {
      "epoch": 0.8929138352900431,
      "grad_norm": 0.1177377924323082,
      "learning_rate": 6.768079972724142e-06,
      "loss": 0.0033,
      "step": 139220
    },
    {
      "epoch": 0.8929779721838291,
      "grad_norm": 0.12049730867147446,
      "learning_rate": 6.767556420326153e-06,
      "loss": 0.0031,
      "step": 139230
    },
    {
      "epoch": 0.8930421090776153,
      "grad_norm": 0.055463168770074844,
      "learning_rate": 6.76703284577942e-06,
      "loss": 0.0029,
      "step": 139240
    },
    {
      "epoch": 0.8931062459714013,
      "grad_norm": 0.07392285764217377,
      "learning_rate": 6.766509249090501e-06,
      "loss": 0.0047,
      "step": 139250
    },
    {
      "epoch": 0.8931703828651875,
      "grad_norm": 0.006894891150295734,
      "learning_rate": 6.765985630265959e-06,
      "loss": 0.0033,
      "step": 139260
    },
    {
      "epoch": 0.8932345197589735,
      "grad_norm": 0.1036565899848938,
      "learning_rate": 6.765461989312355e-06,
      "loss": 0.0024,
      "step": 139270
    },
    {
      "epoch": 0.8932986566527596,
      "grad_norm": 0.21890641748905182,
      "learning_rate": 6.7649383262362524e-06,
      "loss": 0.0027,
      "step": 139280
    },
    {
      "epoch": 0.8933627935465458,
      "grad_norm": 0.0598384328186512,
      "learning_rate": 6.764414641044208e-06,
      "loss": 0.0028,
      "step": 139290
    },
    {
      "epoch": 0.8934269304403318,
      "grad_norm": 0.10389411449432373,
      "learning_rate": 6.763890933742789e-06,
      "loss": 0.0029,
      "step": 139300
    },
    {
      "epoch": 0.893491067334118,
      "grad_norm": 0.19174174964427948,
      "learning_rate": 6.763367204338556e-06,
      "loss": 0.0058,
      "step": 139310
    },
    {
      "epoch": 0.893555204227904,
      "grad_norm": 0.054285917431116104,
      "learning_rate": 6.762843452838072e-06,
      "loss": 0.0016,
      "step": 139320
    },
    {
      "epoch": 0.8936193411216902,
      "grad_norm": 0.08225361257791519,
      "learning_rate": 6.762319679247898e-06,
      "loss": 0.0039,
      "step": 139330
    },
    {
      "epoch": 0.8936834780154762,
      "grad_norm": 0.17472833395004272,
      "learning_rate": 6.7617958835746015e-06,
      "loss": 0.0039,
      "step": 139340
    },
    {
      "epoch": 0.8937476149092624,
      "grad_norm": 0.0652933418750763,
      "learning_rate": 6.761272065824741e-06,
      "loss": 0.0015,
      "step": 139350
    },
    {
      "epoch": 0.8938117518030484,
      "grad_norm": 0.1431877613067627,
      "learning_rate": 6.760748226004884e-06,
      "loss": 0.0025,
      "step": 139360
    },
    {
      "epoch": 0.8938758886968345,
      "grad_norm": 0.2542361915111542,
      "learning_rate": 6.760224364121592e-06,
      "loss": 0.0022,
      "step": 139370
    },
    {
      "epoch": 0.8939400255906206,
      "grad_norm": 0.11593648046255112,
      "learning_rate": 6.759700480181432e-06,
      "loss": 0.0022,
      "step": 139380
    },
    {
      "epoch": 0.8940041624844067,
      "grad_norm": 0.12163574993610382,
      "learning_rate": 6.7591765741909665e-06,
      "loss": 0.0023,
      "step": 139390
    },
    {
      "epoch": 0.8940682993781928,
      "grad_norm": 0.14666952192783356,
      "learning_rate": 6.758652646156763e-06,
      "loss": 0.0038,
      "step": 139400
    },
    {
      "epoch": 0.8941324362719789,
      "grad_norm": 0.14081643521785736,
      "learning_rate": 6.758128696085383e-06,
      "loss": 0.0016,
      "step": 139410
    },
    {
      "epoch": 0.8941965731657651,
      "grad_norm": 0.10789891332387924,
      "learning_rate": 6.757604723983394e-06,
      "loss": 0.0015,
      "step": 139420
    },
    {
      "epoch": 0.8942607100595511,
      "grad_norm": 0.07813584059476852,
      "learning_rate": 6.757080729857364e-06,
      "loss": 0.0021,
      "step": 139430
    },
    {
      "epoch": 0.8943248469533372,
      "grad_norm": 0.10379879176616669,
      "learning_rate": 6.756556713713853e-06,
      "loss": 0.0019,
      "step": 139440
    },
    {
      "epoch": 0.8943889838471233,
      "grad_norm": 0.12771467864513397,
      "learning_rate": 6.756032675559434e-06,
      "loss": 0.0014,
      "step": 139450
    },
    {
      "epoch": 0.8944531207409094,
      "grad_norm": 0.10882225632667542,
      "learning_rate": 6.75550861540067e-06,
      "loss": 0.0021,
      "step": 139460
    },
    {
      "epoch": 0.8945172576346955,
      "grad_norm": 0.08284783363342285,
      "learning_rate": 6.754984533244128e-06,
      "loss": 0.0061,
      "step": 139470
    },
    {
      "epoch": 0.8945813945284816,
      "grad_norm": 0.11675601452589035,
      "learning_rate": 6.754460429096374e-06,
      "loss": 0.0021,
      "step": 139480
    },
    {
      "epoch": 0.8946455314222677,
      "grad_norm": 0.1882113367319107,
      "learning_rate": 6.753936302963979e-06,
      "loss": 0.0025,
      "step": 139490
    },
    {
      "epoch": 0.8947096683160538,
      "grad_norm": 0.4749677777290344,
      "learning_rate": 6.753412154853508e-06,
      "loss": 0.0023,
      "step": 139500
    },
    {
      "epoch": 0.8947738052098398,
      "grad_norm": 0.16558413207530975,
      "learning_rate": 6.75288798477153e-06,
      "loss": 0.002,
      "step": 139510
    },
    {
      "epoch": 0.894837942103626,
      "grad_norm": 0.13556015491485596,
      "learning_rate": 6.752363792724612e-06,
      "loss": 0.0021,
      "step": 139520
    },
    {
      "epoch": 0.894902078997412,
      "grad_norm": 0.12252263724803925,
      "learning_rate": 6.751839578719324e-06,
      "loss": 0.0028,
      "step": 139530
    },
    {
      "epoch": 0.8949662158911982,
      "grad_norm": 0.06946799159049988,
      "learning_rate": 6.751315342762234e-06,
      "loss": 0.002,
      "step": 139540
    },
    {
      "epoch": 0.8950303527849842,
      "grad_norm": 0.0039226580411195755,
      "learning_rate": 6.7507910848599105e-06,
      "loss": 0.0019,
      "step": 139550
    },
    {
      "epoch": 0.8950944896787704,
      "grad_norm": 0.1051669493317604,
      "learning_rate": 6.750266805018924e-06,
      "loss": 0.002,
      "step": 139560
    },
    {
      "epoch": 0.8951586265725565,
      "grad_norm": 0.13654948770999908,
      "learning_rate": 6.749742503245843e-06,
      "loss": 0.0018,
      "step": 139570
    },
    {
      "epoch": 0.8952227634663426,
      "grad_norm": 0.08528295904397964,
      "learning_rate": 6.749218179547239e-06,
      "loss": 0.0011,
      "step": 139580
    },
    {
      "epoch": 0.8952869003601287,
      "grad_norm": 0.07897262275218964,
      "learning_rate": 6.748693833929679e-06,
      "loss": 0.0053,
      "step": 139590
    },
    {
      "epoch": 0.8953510372539147,
      "grad_norm": 0.05403195694088936,
      "learning_rate": 6.748169466399738e-06,
      "loss": 0.0027,
      "step": 139600
    },
    {
      "epoch": 0.8954151741477009,
      "grad_norm": 0.13913075625896454,
      "learning_rate": 6.747645076963981e-06,
      "loss": 0.0023,
      "step": 139610
    },
    {
      "epoch": 0.8954793110414869,
      "grad_norm": 0.050196729600429535,
      "learning_rate": 6.747120665628985e-06,
      "loss": 0.0024,
      "step": 139620
    },
    {
      "epoch": 0.8955434479352731,
      "grad_norm": 0.12445234507322311,
      "learning_rate": 6.746596232401316e-06,
      "loss": 0.0025,
      "step": 139630
    },
    {
      "epoch": 0.8956075848290591,
      "grad_norm": 0.06994674354791641,
      "learning_rate": 6.746071777287551e-06,
      "loss": 0.0017,
      "step": 139640
    },
    {
      "epoch": 0.8956717217228453,
      "grad_norm": 0.014794007875025272,
      "learning_rate": 6.745547300294256e-06,
      "loss": 0.0021,
      "step": 139650
    },
    {
      "epoch": 0.8957358586166313,
      "grad_norm": 0.1464836597442627,
      "learning_rate": 6.745022801428007e-06,
      "loss": 0.0028,
      "step": 139660
    },
    {
      "epoch": 0.8957999955104174,
      "grad_norm": 0.22069483995437622,
      "learning_rate": 6.744498280695375e-06,
      "loss": 0.0035,
      "step": 139670
    },
    {
      "epoch": 0.8958641324042035,
      "grad_norm": 0.04780168458819389,
      "learning_rate": 6.7439737381029335e-06,
      "loss": 0.0035,
      "step": 139680
    },
    {
      "epoch": 0.8959282692979896,
      "grad_norm": 0.14168903231620789,
      "learning_rate": 6.743449173657254e-06,
      "loss": 0.0065,
      "step": 139690
    },
    {
      "epoch": 0.8959924061917758,
      "grad_norm": 0.04651365801692009,
      "learning_rate": 6.742924587364911e-06,
      "loss": 0.0021,
      "step": 139700
    },
    {
      "epoch": 0.8960565430855618,
      "grad_norm": 0.05780575051903725,
      "learning_rate": 6.742399979232477e-06,
      "loss": 0.0026,
      "step": 139710
    },
    {
      "epoch": 0.896120679979348,
      "grad_norm": 0.08799566328525543,
      "learning_rate": 6.741875349266525e-06,
      "loss": 0.0025,
      "step": 139720
    },
    {
      "epoch": 0.896184816873134,
      "grad_norm": 0.1450275331735611,
      "learning_rate": 6.74135069747363e-06,
      "loss": 0.0019,
      "step": 139730
    },
    {
      "epoch": 0.8962489537669202,
      "grad_norm": 0.11965475976467133,
      "learning_rate": 6.740826023860368e-06,
      "loss": 0.006,
      "step": 139740
    },
    {
      "epoch": 0.8963130906607062,
      "grad_norm": 0.13536153733730316,
      "learning_rate": 6.740301328433309e-06,
      "loss": 0.0022,
      "step": 139750
    },
    {
      "epoch": 0.8963772275544923,
      "grad_norm": 0.06510933488607407,
      "learning_rate": 6.739776611199033e-06,
      "loss": 0.0022,
      "step": 139760
    },
    {
      "epoch": 0.8964413644482784,
      "grad_norm": 0.08963591605424881,
      "learning_rate": 6.739251872164112e-06,
      "loss": 0.002,
      "step": 139770
    },
    {
      "epoch": 0.8965055013420645,
      "grad_norm": 0.1946440488100052,
      "learning_rate": 6.738727111335122e-06,
      "loss": 0.0032,
      "step": 139780
    },
    {
      "epoch": 0.8965696382358506,
      "grad_norm": 0.06276450306177139,
      "learning_rate": 6.738202328718639e-06,
      "loss": 0.0018,
      "step": 139790
    },
    {
      "epoch": 0.8966337751296367,
      "grad_norm": 0.15514005720615387,
      "learning_rate": 6.737677524321238e-06,
      "loss": 0.002,
      "step": 139800
    },
    {
      "epoch": 0.8966979120234227,
      "grad_norm": 0.16424056887626648,
      "learning_rate": 6.737152698149496e-06,
      "loss": 0.0056,
      "step": 139810
    },
    {
      "epoch": 0.8967620489172089,
      "grad_norm": 0.0827999860048294,
      "learning_rate": 6.736627850209988e-06,
      "loss": 0.0015,
      "step": 139820
    },
    {
      "epoch": 0.8968261858109949,
      "grad_norm": 0.03634626790881157,
      "learning_rate": 6.736102980509294e-06,
      "loss": 0.0013,
      "step": 139830
    },
    {
      "epoch": 0.8968903227047811,
      "grad_norm": 0.06929469108581543,
      "learning_rate": 6.7355780890539865e-06,
      "loss": 0.0024,
      "step": 139840
    },
    {
      "epoch": 0.8969544595985672,
      "grad_norm": 0.1740657240152359,
      "learning_rate": 6.7350531758506474e-06,
      "loss": 0.0017,
      "step": 139850
    },
    {
      "epoch": 0.8970185964923533,
      "grad_norm": 0.10940054059028625,
      "learning_rate": 6.73452824090585e-06,
      "loss": 0.0026,
      "step": 139860
    },
    {
      "epoch": 0.8970827333861394,
      "grad_norm": 0.13584397733211517,
      "learning_rate": 6.734003284226175e-06,
      "loss": 0.0013,
      "step": 139870
    },
    {
      "epoch": 0.8971468702799255,
      "grad_norm": 0.1612374633550644,
      "learning_rate": 6.7334783058181995e-06,
      "loss": 0.0056,
      "step": 139880
    },
    {
      "epoch": 0.8972110071737116,
      "grad_norm": 0.07454238831996918,
      "learning_rate": 6.732953305688502e-06,
      "loss": 0.0022,
      "step": 139890
    },
    {
      "epoch": 0.8972751440674976,
      "grad_norm": 0.056154754012823105,
      "learning_rate": 6.732428283843661e-06,
      "loss": 0.0016,
      "step": 139900
    },
    {
      "epoch": 0.8973392809612838,
      "grad_norm": 0.09653551876544952,
      "learning_rate": 6.731903240290256e-06,
      "loss": 0.0011,
      "step": 139910
    },
    {
      "epoch": 0.8974034178550698,
      "grad_norm": 0.12849965691566467,
      "learning_rate": 6.731378175034866e-06,
      "loss": 0.0018,
      "step": 139920
    },
    {
      "epoch": 0.897467554748856,
      "grad_norm": 0.0481133833527565,
      "learning_rate": 6.730853088084068e-06,
      "loss": 0.002,
      "step": 139930
    },
    {
      "epoch": 0.897531691642642,
      "grad_norm": 0.18197129666805267,
      "learning_rate": 6.730327979444446e-06,
      "loss": 0.0025,
      "step": 139940
    },
    {
      "epoch": 0.8975958285364282,
      "grad_norm": 0.12247568368911743,
      "learning_rate": 6.729802849122577e-06,
      "loss": 0.0025,
      "step": 139950
    },
    {
      "epoch": 0.8976599654302142,
      "grad_norm": 0.051559291779994965,
      "learning_rate": 6.7292776971250415e-06,
      "loss": 0.0019,
      "step": 139960
    },
    {
      "epoch": 0.8977241023240004,
      "grad_norm": 0.10260986536741257,
      "learning_rate": 6.728752523458421e-06,
      "loss": 0.0019,
      "step": 139970
    },
    {
      "epoch": 0.8977882392177864,
      "grad_norm": 0.16498993337154388,
      "learning_rate": 6.728227328129296e-06,
      "loss": 0.0042,
      "step": 139980
    },
    {
      "epoch": 0.8978523761115725,
      "grad_norm": 0.1725826859474182,
      "learning_rate": 6.727702111144245e-06,
      "loss": 0.0038,
      "step": 139990
    },
    {
      "epoch": 0.8979165130053587,
      "grad_norm": 0.028135206550359726,
      "learning_rate": 6.727176872509855e-06,
      "loss": 0.0015,
      "step": 140000
    },
    {
      "epoch": 0.8979806498991447,
      "grad_norm": 0.058798953890800476,
      "learning_rate": 6.726651612232703e-06,
      "loss": 0.0034,
      "step": 140010
    },
    {
      "epoch": 0.8980447867929309,
      "grad_norm": 0.08650743216276169,
      "learning_rate": 6.726126330319373e-06,
      "loss": 0.0026,
      "step": 140020
    },
    {
      "epoch": 0.8981089236867169,
      "grad_norm": 0.030648980289697647,
      "learning_rate": 6.725601026776446e-06,
      "loss": 0.0019,
      "step": 140030
    },
    {
      "epoch": 0.8981730605805031,
      "grad_norm": 0.08183928579092026,
      "learning_rate": 6.725075701610505e-06,
      "loss": 0.0024,
      "step": 140040
    },
    {
      "epoch": 0.8982371974742891,
      "grad_norm": 0.06568136811256409,
      "learning_rate": 6.724550354828132e-06,
      "loss": 0.0021,
      "step": 140050
    },
    {
      "epoch": 0.8983013343680752,
      "grad_norm": 0.2455543428659439,
      "learning_rate": 6.724024986435912e-06,
      "loss": 0.003,
      "step": 140060
    },
    {
      "epoch": 0.8983654712618613,
      "grad_norm": 0.12305790930986404,
      "learning_rate": 6.723499596440426e-06,
      "loss": 0.0011,
      "step": 140070
    },
    {
      "epoch": 0.8984296081556474,
      "grad_norm": 0.1043807864189148,
      "learning_rate": 6.722974184848256e-06,
      "loss": 0.0018,
      "step": 140080
    },
    {
      "epoch": 0.8984937450494335,
      "grad_norm": 0.029284963384270668,
      "learning_rate": 6.722448751665992e-06,
      "loss": 0.0018,
      "step": 140090
    },
    {
      "epoch": 0.8985578819432196,
      "grad_norm": 0.2810690104961395,
      "learning_rate": 6.721923296900211e-06,
      "loss": 0.0018,
      "step": 140100
    },
    {
      "epoch": 0.8986220188370057,
      "grad_norm": 0.1602506786584854,
      "learning_rate": 6.7213978205575025e-06,
      "loss": 0.0018,
      "step": 140110
    },
    {
      "epoch": 0.8986861557307918,
      "grad_norm": 0.04824730008840561,
      "learning_rate": 6.7208723226444486e-06,
      "loss": 0.0011,
      "step": 140120
    },
    {
      "epoch": 0.898750292624578,
      "grad_norm": 0.20356017351150513,
      "learning_rate": 6.720346803167634e-06,
      "loss": 0.0029,
      "step": 140130
    },
    {
      "epoch": 0.898814429518364,
      "grad_norm": 0.2512374520301819,
      "learning_rate": 6.719821262133645e-06,
      "loss": 0.0033,
      "step": 140140
    },
    {
      "epoch": 0.8988785664121501,
      "grad_norm": 0.07396150380373001,
      "learning_rate": 6.719295699549066e-06,
      "loss": 0.0075,
      "step": 140150
    },
    {
      "epoch": 0.8989427033059362,
      "grad_norm": 0.045413654297590256,
      "learning_rate": 6.718770115420483e-06,
      "loss": 0.0028,
      "step": 140160
    },
    {
      "epoch": 0.8990068401997223,
      "grad_norm": 0.07554330676794052,
      "learning_rate": 6.7182445097544835e-06,
      "loss": 0.0025,
      "step": 140170
    },
    {
      "epoch": 0.8990709770935084,
      "grad_norm": 0.10082416236400604,
      "learning_rate": 6.7177188825576515e-06,
      "loss": 0.0019,
      "step": 140180
    },
    {
      "epoch": 0.8991351139872945,
      "grad_norm": 0.0690072625875473,
      "learning_rate": 6.717193233836574e-06,
      "loss": 0.0047,
      "step": 140190
    },
    {
      "epoch": 0.8991992508810805,
      "grad_norm": 0.056680645793676376,
      "learning_rate": 6.716667563597837e-06,
      "loss": 0.0019,
      "step": 140200
    },
    {
      "epoch": 0.8992633877748667,
      "grad_norm": 0.06069463863968849,
      "learning_rate": 6.71614187184803e-06,
      "loss": 0.0019,
      "step": 140210
    },
    {
      "epoch": 0.8993275246686527,
      "grad_norm": 0.006808622274547815,
      "learning_rate": 6.715616158593739e-06,
      "loss": 0.0025,
      "step": 140220
    },
    {
      "epoch": 0.8993916615624389,
      "grad_norm": 0.06863482296466827,
      "learning_rate": 6.715090423841549e-06,
      "loss": 0.0017,
      "step": 140230
    },
    {
      "epoch": 0.8994557984562249,
      "grad_norm": 0.3106861412525177,
      "learning_rate": 6.714564667598053e-06,
      "loss": 0.0037,
      "step": 140240
    },
    {
      "epoch": 0.8995199353500111,
      "grad_norm": 0.35957279801368713,
      "learning_rate": 6.714038889869835e-06,
      "loss": 0.0019,
      "step": 140250
    },
    {
      "epoch": 0.8995840722437971,
      "grad_norm": 0.07534538209438324,
      "learning_rate": 6.713513090663486e-06,
      "loss": 0.0019,
      "step": 140260
    },
    {
      "epoch": 0.8996482091375833,
      "grad_norm": 0.0364227332174778,
      "learning_rate": 6.712987269985592e-06,
      "loss": 0.0017,
      "step": 140270
    },
    {
      "epoch": 0.8997123460313694,
      "grad_norm": 0.06626851856708527,
      "learning_rate": 6.712461427842743e-06,
      "loss": 0.0028,
      "step": 140280
    },
    {
      "epoch": 0.8997764829251554,
      "grad_norm": 0.3194088637828827,
      "learning_rate": 6.71193556424153e-06,
      "loss": 0.0042,
      "step": 140290
    },
    {
      "epoch": 0.8998406198189416,
      "grad_norm": 0.14776144921779633,
      "learning_rate": 6.71140967918854e-06,
      "loss": 0.0013,
      "step": 140300
    },
    {
      "epoch": 0.8999047567127276,
      "grad_norm": 0.1562497466802597,
      "learning_rate": 6.710883772690362e-06,
      "loss": 0.0027,
      "step": 140310
    },
    {
      "epoch": 0.8999688936065138,
      "grad_norm": 0.16863062977790833,
      "learning_rate": 6.71035784475359e-06,
      "loss": 0.0023,
      "step": 140320
    },
    {
      "epoch": 0.9000330305002998,
      "grad_norm": 0.020131120458245277,
      "learning_rate": 6.70983189538481e-06,
      "loss": 0.0017,
      "step": 140330
    },
    {
      "epoch": 0.900097167394086,
      "grad_norm": 0.236953467130661,
      "learning_rate": 6.709305924590617e-06,
      "loss": 0.0023,
      "step": 140340
    },
    {
      "epoch": 0.900161304287872,
      "grad_norm": 0.14059460163116455,
      "learning_rate": 6.708779932377596e-06,
      "loss": 0.003,
      "step": 140350
    },
    {
      "epoch": 0.9002254411816581,
      "grad_norm": 0.1643529087305069,
      "learning_rate": 6.708253918752343e-06,
      "loss": 0.0024,
      "step": 140360
    },
    {
      "epoch": 0.9002895780754442,
      "grad_norm": 0.13108673691749573,
      "learning_rate": 6.707727883721447e-06,
      "loss": 0.0024,
      "step": 140370
    },
    {
      "epoch": 0.9003537149692303,
      "grad_norm": 0.07430348545312881,
      "learning_rate": 6.7072018272915e-06,
      "loss": 0.0021,
      "step": 140380
    },
    {
      "epoch": 0.9004178518630164,
      "grad_norm": 0.12285272777080536,
      "learning_rate": 6.706675749469093e-06,
      "loss": 0.0031,
      "step": 140390
    },
    {
      "epoch": 0.9004819887568025,
      "grad_norm": 0.08578559756278992,
      "learning_rate": 6.706149650260821e-06,
      "loss": 0.005,
      "step": 140400
    },
    {
      "epoch": 0.9005461256505887,
      "grad_norm": 0.011579250916838646,
      "learning_rate": 6.705623529673274e-06,
      "loss": 0.0012,
      "step": 140410
    },
    {
      "epoch": 0.9006102625443747,
      "grad_norm": 0.09027321636676788,
      "learning_rate": 6.705097387713046e-06,
      "loss": 0.0016,
      "step": 140420
    },
    {
      "epoch": 0.9006743994381609,
      "grad_norm": 0.045796554535627365,
      "learning_rate": 6.704571224386728e-06,
      "loss": 0.0016,
      "step": 140430
    },
    {
      "epoch": 0.9007385363319469,
      "grad_norm": 0.07548578828573227,
      "learning_rate": 6.704045039700914e-06,
      "loss": 0.0016,
      "step": 140440
    },
    {
      "epoch": 0.900802673225733,
      "grad_norm": 0.23142126202583313,
      "learning_rate": 6.703518833662198e-06,
      "loss": 0.0024,
      "step": 140450
    },
    {
      "epoch": 0.9008668101195191,
      "grad_norm": 0.11072611808776855,
      "learning_rate": 6.702992606277174e-06,
      "loss": 0.0022,
      "step": 140460
    },
    {
      "epoch": 0.9009309470133052,
      "grad_norm": 0.2234678566455841,
      "learning_rate": 6.702466357552435e-06,
      "loss": 0.0024,
      "step": 140470
    },
    {
      "epoch": 0.9009950839070913,
      "grad_norm": 0.05514800548553467,
      "learning_rate": 6.701940087494576e-06,
      "loss": 0.0016,
      "step": 140480
    },
    {
      "epoch": 0.9010592208008774,
      "grad_norm": 0.1039884090423584,
      "learning_rate": 6.701413796110192e-06,
      "loss": 0.0021,
      "step": 140490
    },
    {
      "epoch": 0.9011233576946635,
      "grad_norm": 0.013716175220906734,
      "learning_rate": 6.700887483405877e-06,
      "loss": 0.0017,
      "step": 140500
    },
    {
      "epoch": 0.9011874945884496,
      "grad_norm": 0.16248738765716553,
      "learning_rate": 6.7003611493882255e-06,
      "loss": 0.0032,
      "step": 140510
    },
    {
      "epoch": 0.9012516314822356,
      "grad_norm": 0.207486093044281,
      "learning_rate": 6.6998347940638345e-06,
      "loss": 0.0019,
      "step": 140520
    },
    {
      "epoch": 0.9013157683760218,
      "grad_norm": 0.0907140001654625,
      "learning_rate": 6.699308417439298e-06,
      "loss": 0.0013,
      "step": 140530
    },
    {
      "epoch": 0.9013799052698078,
      "grad_norm": 0.028752660378813744,
      "learning_rate": 6.698782019521213e-06,
      "loss": 0.0012,
      "step": 140540
    },
    {
      "epoch": 0.901444042163594,
      "grad_norm": 0.05771041288971901,
      "learning_rate": 6.698255600316175e-06,
      "loss": 0.0018,
      "step": 140550
    },
    {
      "epoch": 0.9015081790573801,
      "grad_norm": 0.04292020574212074,
      "learning_rate": 6.69772915983078e-06,
      "loss": 0.004,
      "step": 140560
    },
    {
      "epoch": 0.9015723159511662,
      "grad_norm": 0.16082461178302765,
      "learning_rate": 6.697202698071626e-06,
      "loss": 0.0022,
      "step": 140570
    },
    {
      "epoch": 0.9016364528449523,
      "grad_norm": 0.0972842127084732,
      "learning_rate": 6.696676215045309e-06,
      "loss": 0.0027,
      "step": 140580
    },
    {
      "epoch": 0.9017005897387383,
      "grad_norm": 0.00951351784169674,
      "learning_rate": 6.696149710758426e-06,
      "loss": 0.0029,
      "step": 140590
    },
    {
      "epoch": 0.9017647266325245,
      "grad_norm": 0.09503789246082306,
      "learning_rate": 6.695623185217576e-06,
      "loss": 0.0042,
      "step": 140600
    },
    {
      "epoch": 0.9018288635263105,
      "grad_norm": 0.13512814044952393,
      "learning_rate": 6.695096638429355e-06,
      "loss": 0.0029,
      "step": 140610
    },
    {
      "epoch": 0.9018930004200967,
      "grad_norm": 0.24733853340148926,
      "learning_rate": 6.6945700704003614e-06,
      "loss": 0.0026,
      "step": 140620
    },
    {
      "epoch": 0.9019571373138827,
      "grad_norm": 0.09181825071573257,
      "learning_rate": 6.694043481137193e-06,
      "loss": 0.0016,
      "step": 140630
    },
    {
      "epoch": 0.9020212742076689,
      "grad_norm": 0.19485434889793396,
      "learning_rate": 6.69351687064645e-06,
      "loss": 0.004,
      "step": 140640
    },
    {
      "epoch": 0.9020854111014549,
      "grad_norm": 0.029666827991604805,
      "learning_rate": 6.692990238934731e-06,
      "loss": 0.0016,
      "step": 140650
    },
    {
      "epoch": 0.902149547995241,
      "grad_norm": 0.21938619017601013,
      "learning_rate": 6.692463586008634e-06,
      "loss": 0.003,
      "step": 140660
    },
    {
      "epoch": 0.9022136848890271,
      "grad_norm": 0.21265892684459686,
      "learning_rate": 6.691936911874758e-06,
      "loss": 0.0036,
      "step": 140670
    },
    {
      "epoch": 0.9022778217828132,
      "grad_norm": 0.14057812094688416,
      "learning_rate": 6.691410216539705e-06,
      "loss": 0.0027,
      "step": 140680
    },
    {
      "epoch": 0.9023419586765994,
      "grad_norm": 0.19511573016643524,
      "learning_rate": 6.69088350001007e-06,
      "loss": 0.0015,
      "step": 140690
    },
    {
      "epoch": 0.9024060955703854,
      "grad_norm": 0.031265996396541595,
      "learning_rate": 6.690356762292459e-06,
      "loss": 0.002,
      "step": 140700
    },
    {
      "epoch": 0.9024702324641716,
      "grad_norm": 0.04314252734184265,
      "learning_rate": 6.689830003393468e-06,
      "loss": 0.0018,
      "step": 140710
    },
    {
      "epoch": 0.9025343693579576,
      "grad_norm": 0.5250045657157898,
      "learning_rate": 6.6893032233197e-06,
      "loss": 0.0039,
      "step": 140720
    },
    {
      "epoch": 0.9025985062517438,
      "grad_norm": 0.10032887011766434,
      "learning_rate": 6.688776422077756e-06,
      "loss": 0.0027,
      "step": 140730
    },
    {
      "epoch": 0.9026626431455298,
      "grad_norm": 0.15504832565784454,
      "learning_rate": 6.688249599674235e-06,
      "loss": 0.0028,
      "step": 140740
    },
    {
      "epoch": 0.902726780039316,
      "grad_norm": 0.1699640303850174,
      "learning_rate": 6.687722756115742e-06,
      "loss": 0.0018,
      "step": 140750
    },
    {
      "epoch": 0.902790916933102,
      "grad_norm": 0.15671475231647491,
      "learning_rate": 6.687195891408874e-06,
      "loss": 0.0017,
      "step": 140760
    },
    {
      "epoch": 0.9028550538268881,
      "grad_norm": 0.35086822509765625,
      "learning_rate": 6.686669005560237e-06,
      "loss": 0.0073,
      "step": 140770
    },
    {
      "epoch": 0.9029191907206742,
      "grad_norm": 0.1215127483010292,
      "learning_rate": 6.686142098576432e-06,
      "loss": 0.0024,
      "step": 140780
    },
    {
      "epoch": 0.9029833276144603,
      "grad_norm": 0.08526911586523056,
      "learning_rate": 6.685615170464061e-06,
      "loss": 0.0026,
      "step": 140790
    },
    {
      "epoch": 0.9030474645082464,
      "grad_norm": 0.07515113055706024,
      "learning_rate": 6.685088221229727e-06,
      "loss": 0.0031,
      "step": 140800
    },
    {
      "epoch": 0.9031116014020325,
      "grad_norm": 0.1216416209936142,
      "learning_rate": 6.684561250880035e-06,
      "loss": 0.0029,
      "step": 140810
    },
    {
      "epoch": 0.9031757382958185,
      "grad_norm": 0.08766988664865494,
      "learning_rate": 6.684034259421586e-06,
      "loss": 0.001,
      "step": 140820
    },
    {
      "epoch": 0.9032398751896047,
      "grad_norm": 0.10048586875200272,
      "learning_rate": 6.683507246860984e-06,
      "loss": 0.0026,
      "step": 140830
    },
    {
      "epoch": 0.9033040120833908,
      "grad_norm": 0.10893585532903671,
      "learning_rate": 6.682980213204832e-06,
      "loss": 0.0019,
      "step": 140840
    },
    {
      "epoch": 0.9033681489771769,
      "grad_norm": 0.06966837495565414,
      "learning_rate": 6.682453158459736e-06,
      "loss": 0.0021,
      "step": 140850
    },
    {
      "epoch": 0.903432285870963,
      "grad_norm": 0.050560686737298965,
      "learning_rate": 6.6819260826323e-06,
      "loss": 0.0024,
      "step": 140860
    },
    {
      "epoch": 0.9034964227647491,
      "grad_norm": 0.08276427537202835,
      "learning_rate": 6.681398985729127e-06,
      "loss": 0.0018,
      "step": 140870
    },
    {
      "epoch": 0.9035605596585352,
      "grad_norm": 0.16475242376327515,
      "learning_rate": 6.680871867756824e-06,
      "loss": 0.0011,
      "step": 140880
    },
    {
      "epoch": 0.9036246965523212,
      "grad_norm": 0.040519170463085175,
      "learning_rate": 6.680344728721995e-06,
      "loss": 0.0068,
      "step": 140890
    },
    {
      "epoch": 0.9036888334461074,
      "grad_norm": 0.08458121120929718,
      "learning_rate": 6.679817568631245e-06,
      "loss": 0.0016,
      "step": 140900
    },
    {
      "epoch": 0.9037529703398934,
      "grad_norm": 0.0888509452342987,
      "learning_rate": 6.6792903874911805e-06,
      "loss": 0.0015,
      "step": 140910
    },
    {
      "epoch": 0.9038171072336796,
      "grad_norm": 0.1488405019044876,
      "learning_rate": 6.678763185308408e-06,
      "loss": 0.0027,
      "step": 140920
    },
    {
      "epoch": 0.9038812441274656,
      "grad_norm": 0.04099796339869499,
      "learning_rate": 6.678235962089531e-06,
      "loss": 0.0017,
      "step": 140930
    },
    {
      "epoch": 0.9039453810212518,
      "grad_norm": 0.09509284049272537,
      "learning_rate": 6.67770871784116e-06,
      "loss": 0.002,
      "step": 140940
    },
    {
      "epoch": 0.9040095179150378,
      "grad_norm": 0.061781175434589386,
      "learning_rate": 6.6771814525698984e-06,
      "loss": 0.0044,
      "step": 140950
    },
    {
      "epoch": 0.904073654808824,
      "grad_norm": 0.0422157421708107,
      "learning_rate": 6.676654166282356e-06,
      "loss": 0.0034,
      "step": 140960
    },
    {
      "epoch": 0.9041377917026101,
      "grad_norm": 0.10214755684137344,
      "learning_rate": 6.6761268589851384e-06,
      "loss": 0.0017,
      "step": 140970
    },
    {
      "epoch": 0.9042019285963961,
      "grad_norm": 0.15138445794582367,
      "learning_rate": 6.675599530684853e-06,
      "loss": 0.0019,
      "step": 140980
    },
    {
      "epoch": 0.9042660654901823,
      "grad_norm": 0.06442516297101974,
      "learning_rate": 6.675072181388107e-06,
      "loss": 0.0025,
      "step": 140990
    },
    {
      "epoch": 0.9043302023839683,
      "grad_norm": 0.19241341948509216,
      "learning_rate": 6.674544811101511e-06,
      "loss": 0.0014,
      "step": 141000
    },
    {
      "epoch": 0.9043943392777545,
      "grad_norm": 0.09069796651601791,
      "learning_rate": 6.67401741983167e-06,
      "loss": 0.0019,
      "step": 141010
    },
    {
      "epoch": 0.9044584761715405,
      "grad_norm": 0.26262545585632324,
      "learning_rate": 6.673490007585196e-06,
      "loss": 0.0026,
      "step": 141020
    },
    {
      "epoch": 0.9045226130653267,
      "grad_norm": 0.32308170199394226,
      "learning_rate": 6.672962574368695e-06,
      "loss": 0.0018,
      "step": 141030
    },
    {
      "epoch": 0.9045867499591127,
      "grad_norm": 0.09142335504293442,
      "learning_rate": 6.672435120188778e-06,
      "loss": 0.0023,
      "step": 141040
    },
    {
      "epoch": 0.9046508868528989,
      "grad_norm": 0.1618300825357437,
      "learning_rate": 6.671907645052054e-06,
      "loss": 0.001,
      "step": 141050
    },
    {
      "epoch": 0.9047150237466849,
      "grad_norm": 0.07678396999835968,
      "learning_rate": 6.671380148965132e-06,
      "loss": 0.0027,
      "step": 141060
    },
    {
      "epoch": 0.904779160640471,
      "grad_norm": 0.04675278440117836,
      "learning_rate": 6.670852631934621e-06,
      "loss": 0.0014,
      "step": 141070
    },
    {
      "epoch": 0.9048432975342571,
      "grad_norm": 0.15445196628570557,
      "learning_rate": 6.670325093967133e-06,
      "loss": 0.0022,
      "step": 141080
    },
    {
      "epoch": 0.9049074344280432,
      "grad_norm": 0.009850629605352879,
      "learning_rate": 6.669797535069278e-06,
      "loss": 0.0011,
      "step": 141090
    },
    {
      "epoch": 0.9049715713218293,
      "grad_norm": 0.1014852374792099,
      "learning_rate": 6.669269955247666e-06,
      "loss": 0.0015,
      "step": 141100
    },
    {
      "epoch": 0.9050357082156154,
      "grad_norm": 0.15192469954490662,
      "learning_rate": 6.66874235450891e-06,
      "loss": 0.0011,
      "step": 141110
    },
    {
      "epoch": 0.9050998451094016,
      "grad_norm": 0.08786418288946152,
      "learning_rate": 6.668214732859618e-06,
      "loss": 0.0023,
      "step": 141120
    },
    {
      "epoch": 0.9051639820031876,
      "grad_norm": 0.02861233986914158,
      "learning_rate": 6.667687090306405e-06,
      "loss": 0.0025,
      "step": 141130
    },
    {
      "epoch": 0.9052281188969737,
      "grad_norm": 0.08352043479681015,
      "learning_rate": 6.667159426855878e-06,
      "loss": 0.0026,
      "step": 141140
    },
    {
      "epoch": 0.9052922557907598,
      "grad_norm": 0.0435749851167202,
      "learning_rate": 6.666631742514655e-06,
      "loss": 0.0021,
      "step": 141150
    },
    {
      "epoch": 0.9053563926845459,
      "grad_norm": 0.06818929314613342,
      "learning_rate": 6.666104037289343e-06,
      "loss": 0.0023,
      "step": 141160
    },
    {
      "epoch": 0.905420529578332,
      "grad_norm": 0.05088305473327637,
      "learning_rate": 6.665576311186557e-06,
      "loss": 0.0027,
      "step": 141170
    },
    {
      "epoch": 0.9054846664721181,
      "grad_norm": 0.09777114540338516,
      "learning_rate": 6.66504856421291e-06,
      "loss": 0.004,
      "step": 141180
    },
    {
      "epoch": 0.9055488033659042,
      "grad_norm": 0.1329837143421173,
      "learning_rate": 6.6645207963750145e-06,
      "loss": 0.002,
      "step": 141190
    },
    {
      "epoch": 0.9056129402596903,
      "grad_norm": 0.08360269665718079,
      "learning_rate": 6.663993007679484e-06,
      "loss": 0.0022,
      "step": 141200
    },
    {
      "epoch": 0.9056770771534763,
      "grad_norm": 0.15117284655570984,
      "learning_rate": 6.663465198132932e-06,
      "loss": 0.0011,
      "step": 141210
    },
    {
      "epoch": 0.9057412140472625,
      "grad_norm": 0.04185052961111069,
      "learning_rate": 6.66293736774197e-06,
      "loss": 0.0022,
      "step": 141220
    },
    {
      "epoch": 0.9058053509410485,
      "grad_norm": 0.27626991271972656,
      "learning_rate": 6.662409516513217e-06,
      "loss": 0.0027,
      "step": 141230
    },
    {
      "epoch": 0.9058694878348347,
      "grad_norm": 0.05853215232491493,
      "learning_rate": 6.661881644453284e-06,
      "loss": 0.0032,
      "step": 141240
    },
    {
      "epoch": 0.9059336247286208,
      "grad_norm": 0.005517345387488604,
      "learning_rate": 6.661353751568787e-06,
      "loss": 0.0036,
      "step": 141250
    },
    {
      "epoch": 0.9059977616224069,
      "grad_norm": 0.07829311490058899,
      "learning_rate": 6.660825837866338e-06,
      "loss": 0.002,
      "step": 141260
    },
    {
      "epoch": 0.906061898516193,
      "grad_norm": 0.13096913695335388,
      "learning_rate": 6.660297903352556e-06,
      "loss": 0.003,
      "step": 141270
    },
    {
      "epoch": 0.906126035409979,
      "grad_norm": 0.17294195294380188,
      "learning_rate": 6.659769948034054e-06,
      "loss": 0.0027,
      "step": 141280
    },
    {
      "epoch": 0.9061901723037652,
      "grad_norm": 0.0880667194724083,
      "learning_rate": 6.659241971917447e-06,
      "loss": 0.0024,
      "step": 141290
    },
    {
      "epoch": 0.9062543091975512,
      "grad_norm": 0.13531440496444702,
      "learning_rate": 6.6587139750093545e-06,
      "loss": 0.0026,
      "step": 141300
    },
    {
      "epoch": 0.9063184460913374,
      "grad_norm": 0.10861565917730331,
      "learning_rate": 6.6581859573163895e-06,
      "loss": 0.0021,
      "step": 141310
    },
    {
      "epoch": 0.9063825829851234,
      "grad_norm": 0.14992816746234894,
      "learning_rate": 6.657657918845169e-06,
      "loss": 0.0038,
      "step": 141320
    },
    {
      "epoch": 0.9064467198789096,
      "grad_norm": 0.19455134868621826,
      "learning_rate": 6.657129859602312e-06,
      "loss": 0.0018,
      "step": 141330
    },
    {
      "epoch": 0.9065108567726956,
      "grad_norm": 0.1379806250333786,
      "learning_rate": 6.656601779594431e-06,
      "loss": 0.0017,
      "step": 141340
    },
    {
      "epoch": 0.9065749936664818,
      "grad_norm": 0.08714434504508972,
      "learning_rate": 6.656073678828147e-06,
      "loss": 0.0017,
      "step": 141350
    },
    {
      "epoch": 0.9066391305602678,
      "grad_norm": 0.05364006757736206,
      "learning_rate": 6.655545557310077e-06,
      "loss": 0.0078,
      "step": 141360
    },
    {
      "epoch": 0.9067032674540539,
      "grad_norm": 0.019650449976325035,
      "learning_rate": 6.655017415046836e-06,
      "loss": 0.0037,
      "step": 141370
    },
    {
      "epoch": 0.90676740434784,
      "grad_norm": 0.08904875814914703,
      "learning_rate": 6.654489252045045e-06,
      "loss": 0.002,
      "step": 141380
    },
    {
      "epoch": 0.9068315412416261,
      "grad_norm": 0.1390581876039505,
      "learning_rate": 6.653961068311321e-06,
      "loss": 0.0024,
      "step": 141390
    },
    {
      "epoch": 0.9068956781354123,
      "grad_norm": 0.06195981428027153,
      "learning_rate": 6.653432863852284e-06,
      "loss": 0.0018,
      "step": 141400
    },
    {
      "epoch": 0.9069598150291983,
      "grad_norm": 0.019686702638864517,
      "learning_rate": 6.65290463867455e-06,
      "loss": 0.0013,
      "step": 141410
    },
    {
      "epoch": 0.9070239519229845,
      "grad_norm": 0.09837204217910767,
      "learning_rate": 6.65237639278474e-06,
      "loss": 0.0017,
      "step": 141420
    },
    {
      "epoch": 0.9070880888167705,
      "grad_norm": 0.08003837615251541,
      "learning_rate": 6.651848126189473e-06,
      "loss": 0.0017,
      "step": 141430
    },
    {
      "epoch": 0.9071522257105566,
      "grad_norm": 0.03738339990377426,
      "learning_rate": 6.651319838895367e-06,
      "loss": 0.0009,
      "step": 141440
    },
    {
      "epoch": 0.9072163626043427,
      "grad_norm": 0.07136432081460953,
      "learning_rate": 6.650791530909045e-06,
      "loss": 0.0016,
      "step": 141450
    },
    {
      "epoch": 0.9072804994981288,
      "grad_norm": 0.05406304448843002,
      "learning_rate": 6.650263202237125e-06,
      "loss": 0.0017,
      "step": 141460
    },
    {
      "epoch": 0.9073446363919149,
      "grad_norm": 0.30447080731391907,
      "learning_rate": 6.649734852886228e-06,
      "loss": 0.004,
      "step": 141470
    },
    {
      "epoch": 0.907408773285701,
      "grad_norm": 0.07497084140777588,
      "learning_rate": 6.6492064828629735e-06,
      "loss": 0.0027,
      "step": 141480
    },
    {
      "epoch": 0.9074729101794871,
      "grad_norm": 0.07167311012744904,
      "learning_rate": 6.648678092173983e-06,
      "loss": 0.0024,
      "step": 141490
    },
    {
      "epoch": 0.9075370470732732,
      "grad_norm": 0.13172192871570587,
      "learning_rate": 6.648149680825878e-06,
      "loss": 0.0027,
      "step": 141500
    },
    {
      "epoch": 0.9076011839670592,
      "grad_norm": 0.27396607398986816,
      "learning_rate": 6.64762124882528e-06,
      "loss": 0.0031,
      "step": 141510
    },
    {
      "epoch": 0.9076653208608454,
      "grad_norm": 0.16255521774291992,
      "learning_rate": 6.64709279617881e-06,
      "loss": 0.0024,
      "step": 141520
    },
    {
      "epoch": 0.9077294577546314,
      "grad_norm": 0.04003090411424637,
      "learning_rate": 6.6465643228930895e-06,
      "loss": 0.0018,
      "step": 141530
    },
    {
      "epoch": 0.9077935946484176,
      "grad_norm": 0.11404585838317871,
      "learning_rate": 6.646035828974742e-06,
      "loss": 0.0025,
      "step": 141540
    },
    {
      "epoch": 0.9078577315422037,
      "grad_norm": 0.5322808623313904,
      "learning_rate": 6.645507314430389e-06,
      "loss": 0.0052,
      "step": 141550
    },
    {
      "epoch": 0.9079218684359898,
      "grad_norm": 0.15718133747577667,
      "learning_rate": 6.644978779266652e-06,
      "loss": 0.0027,
      "step": 141560
    },
    {
      "epoch": 0.9079860053297759,
      "grad_norm": 0.031880151480436325,
      "learning_rate": 6.644450223490158e-06,
      "loss": 0.0024,
      "step": 141570
    },
    {
      "epoch": 0.908050142223562,
      "grad_norm": 0.0900476947426796,
      "learning_rate": 6.643921647107525e-06,
      "loss": 0.0031,
      "step": 141580
    },
    {
      "epoch": 0.9081142791173481,
      "grad_norm": 0.05382583290338516,
      "learning_rate": 6.643393050125379e-06,
      "loss": 0.0014,
      "step": 141590
    },
    {
      "epoch": 0.9081784160111341,
      "grad_norm": 0.08408217877149582,
      "learning_rate": 6.6428644325503445e-06,
      "loss": 0.005,
      "step": 141600
    },
    {
      "epoch": 0.9082425529049203,
      "grad_norm": 0.0819145068526268,
      "learning_rate": 6.6423357943890456e-06,
      "loss": 0.0018,
      "step": 141610
    },
    {
      "epoch": 0.9083066897987063,
      "grad_norm": 0.05600690841674805,
      "learning_rate": 6.641807135648104e-06,
      "loss": 0.0025,
      "step": 141620
    },
    {
      "epoch": 0.9083708266924925,
      "grad_norm": 0.09864833950996399,
      "learning_rate": 6.641278456334145e-06,
      "loss": 0.0017,
      "step": 141630
    },
    {
      "epoch": 0.9084349635862785,
      "grad_norm": 0.08638419210910797,
      "learning_rate": 6.640749756453795e-06,
      "loss": 0.0022,
      "step": 141640
    },
    {
      "epoch": 0.9084991004800647,
      "grad_norm": 0.10510845482349396,
      "learning_rate": 6.640221036013678e-06,
      "loss": 0.0017,
      "step": 141650
    },
    {
      "epoch": 0.9085632373738507,
      "grad_norm": 0.3093767762184143,
      "learning_rate": 6.639692295020419e-06,
      "loss": 0.0021,
      "step": 141660
    },
    {
      "epoch": 0.9086273742676368,
      "grad_norm": 0.08151751011610031,
      "learning_rate": 6.6391635334806426e-06,
      "loss": 0.0021,
      "step": 141670
    },
    {
      "epoch": 0.908691511161423,
      "grad_norm": 0.02787480689585209,
      "learning_rate": 6.638634751400978e-06,
      "loss": 0.0017,
      "step": 141680
    },
    {
      "epoch": 0.908755648055209,
      "grad_norm": 0.17475752532482147,
      "learning_rate": 6.638105948788046e-06,
      "loss": 0.0025,
      "step": 141690
    },
    {
      "epoch": 0.9088197849489952,
      "grad_norm": 0.08409693837165833,
      "learning_rate": 6.637577125648478e-06,
      "loss": 0.002,
      "step": 141700
    },
    {
      "epoch": 0.9088839218427812,
      "grad_norm": 0.2646850347518921,
      "learning_rate": 6.637048281988896e-06,
      "loss": 0.0036,
      "step": 141710
    },
    {
      "epoch": 0.9089480587365674,
      "grad_norm": 0.10334274172782898,
      "learning_rate": 6.636519417815932e-06,
      "loss": 0.0068,
      "step": 141720
    },
    {
      "epoch": 0.9090121956303534,
      "grad_norm": 0.07118162512779236,
      "learning_rate": 6.635990533136207e-06,
      "loss": 0.0018,
      "step": 141730
    },
    {
      "epoch": 0.9090763325241396,
      "grad_norm": 0.10584995150566101,
      "learning_rate": 6.6354616279563545e-06,
      "loss": 0.003,
      "step": 141740
    },
    {
      "epoch": 0.9091404694179256,
      "grad_norm": 0.01536076795309782,
      "learning_rate": 6.634932702282997e-06,
      "loss": 0.0024,
      "step": 141750
    },
    {
      "epoch": 0.9092046063117117,
      "grad_norm": 0.10196884721517563,
      "learning_rate": 6.634403756122765e-06,
      "loss": 0.0033,
      "step": 141760
    },
    {
      "epoch": 0.9092687432054978,
      "grad_norm": 0.014073599129915237,
      "learning_rate": 6.6338747894822845e-06,
      "loss": 0.001,
      "step": 141770
    },
    {
      "epoch": 0.9093328800992839,
      "grad_norm": 0.24571390450000763,
      "learning_rate": 6.633345802368185e-06,
      "loss": 0.0022,
      "step": 141780
    },
    {
      "epoch": 0.90939701699307,
      "grad_norm": 0.09062595665454865,
      "learning_rate": 6.632816794787098e-06,
      "loss": 0.0016,
      "step": 141790
    },
    {
      "epoch": 0.9094611538868561,
      "grad_norm": 0.20013825595378876,
      "learning_rate": 6.632287766745647e-06,
      "loss": 0.0022,
      "step": 141800
    },
    {
      "epoch": 0.9095252907806421,
      "grad_norm": 0.0434277281165123,
      "learning_rate": 6.631758718250465e-06,
      "loss": 0.0015,
      "step": 141810
    },
    {
      "epoch": 0.9095894276744283,
      "grad_norm": 0.03250228241086006,
      "learning_rate": 6.63122964930818e-06,
      "loss": 0.0018,
      "step": 141820
    },
    {
      "epoch": 0.9096535645682144,
      "grad_norm": 0.009286737069487572,
      "learning_rate": 6.6307005599254215e-06,
      "loss": 0.0021,
      "step": 141830
    },
    {
      "epoch": 0.9097177014620005,
      "grad_norm": 0.07002612203359604,
      "learning_rate": 6.63017145010882e-06,
      "loss": 0.0017,
      "step": 141840
    },
    {
      "epoch": 0.9097818383557866,
      "grad_norm": 0.1045675277709961,
      "learning_rate": 6.629642319865004e-06,
      "loss": 0.002,
      "step": 141850
    },
    {
      "epoch": 0.9098459752495727,
      "grad_norm": 0.09787771105766296,
      "learning_rate": 6.629113169200606e-06,
      "loss": 0.0014,
      "step": 141860
    },
    {
      "epoch": 0.9099101121433588,
      "grad_norm": 0.06352692097425461,
      "learning_rate": 6.628583998122256e-06,
      "loss": 0.0011,
      "step": 141870
    },
    {
      "epoch": 0.9099742490371449,
      "grad_norm": 0.2079721987247467,
      "learning_rate": 6.628054806636583e-06,
      "loss": 0.0017,
      "step": 141880
    },
    {
      "epoch": 0.910038385930931,
      "grad_norm": 0.11491197347640991,
      "learning_rate": 6.627525594750221e-06,
      "loss": 0.0022,
      "step": 141890
    },
    {
      "epoch": 0.910102522824717,
      "grad_norm": 0.03413490578532219,
      "learning_rate": 6.6269963624698e-06,
      "loss": 0.0012,
      "step": 141900
    },
    {
      "epoch": 0.9101666597185032,
      "grad_norm": 0.09412901848554611,
      "learning_rate": 6.6264671098019505e-06,
      "loss": 0.004,
      "step": 141910
    },
    {
      "epoch": 0.9102307966122892,
      "grad_norm": 0.03136226907372475,
      "learning_rate": 6.625937836753307e-06,
      "loss": 0.0015,
      "step": 141920
    },
    {
      "epoch": 0.9102949335060754,
      "grad_norm": 0.1041698008775711,
      "learning_rate": 6.625408543330501e-06,
      "loss": 0.0034,
      "step": 141930
    },
    {
      "epoch": 0.9103590703998614,
      "grad_norm": 0.238687202334404,
      "learning_rate": 6.624879229540162e-06,
      "loss": 0.0069,
      "step": 141940
    },
    {
      "epoch": 0.9104232072936476,
      "grad_norm": 0.14690439403057098,
      "learning_rate": 6.6243498953889276e-06,
      "loss": 0.0035,
      "step": 141950
    },
    {
      "epoch": 0.9104873441874337,
      "grad_norm": 0.1875958889722824,
      "learning_rate": 6.623820540883425e-06,
      "loss": 0.0088,
      "step": 141960
    },
    {
      "epoch": 0.9105514810812197,
      "grad_norm": 0.11991121619939804,
      "learning_rate": 6.623291166030294e-06,
      "loss": 0.0034,
      "step": 141970
    },
    {
      "epoch": 0.9106156179750059,
      "grad_norm": 0.10912397503852844,
      "learning_rate": 6.6227617708361625e-06,
      "loss": 0.0024,
      "step": 141980
    },
    {
      "epoch": 0.9106797548687919,
      "grad_norm": 0.009573899209499359,
      "learning_rate": 6.6222323553076666e-06,
      "loss": 0.0026,
      "step": 141990
    },
    {
      "epoch": 0.9107438917625781,
      "grad_norm": 0.024858275428414345,
      "learning_rate": 6.62170291945144e-06,
      "loss": 0.0014,
      "step": 142000
    },
    {
      "epoch": 0.9108080286563641,
      "grad_norm": 0.05668169632554054,
      "learning_rate": 6.621173463274116e-06,
      "loss": 0.0017,
      "step": 142010
    },
    {
      "epoch": 0.9108721655501503,
      "grad_norm": 0.018332339823246002,
      "learning_rate": 6.620643986782331e-06,
      "loss": 0.0018,
      "step": 142020
    },
    {
      "epoch": 0.9109363024439363,
      "grad_norm": 0.2745995819568634,
      "learning_rate": 6.620114489982718e-06,
      "loss": 0.0041,
      "step": 142030
    },
    {
      "epoch": 0.9110004393377225,
      "grad_norm": 0.1281847506761551,
      "learning_rate": 6.619584972881914e-06,
      "loss": 0.0023,
      "step": 142040
    },
    {
      "epoch": 0.9110645762315085,
      "grad_norm": 0.0568624772131443,
      "learning_rate": 6.61905543548655e-06,
      "loss": 0.001,
      "step": 142050
    },
    {
      "epoch": 0.9111287131252946,
      "grad_norm": 0.03093951940536499,
      "learning_rate": 6.618525877803267e-06,
      "loss": 0.0021,
      "step": 142060
    },
    {
      "epoch": 0.9111928500190807,
      "grad_norm": 0.18486446142196655,
      "learning_rate": 6.617996299838696e-06,
      "loss": 0.0016,
      "step": 142070
    },
    {
      "epoch": 0.9112569869128668,
      "grad_norm": 0.11845473200082779,
      "learning_rate": 6.6174667015994765e-06,
      "loss": 0.0021,
      "step": 142080
    },
    {
      "epoch": 0.9113211238066529,
      "grad_norm": 0.2680804133415222,
      "learning_rate": 6.616937083092243e-06,
      "loss": 0.0038,
      "step": 142090
    },
    {
      "epoch": 0.911385260700439,
      "grad_norm": 0.1053524985909462,
      "learning_rate": 6.616407444323631e-06,
      "loss": 0.0013,
      "step": 142100
    },
    {
      "epoch": 0.9114493975942252,
      "grad_norm": 0.18218301236629486,
      "learning_rate": 6.61587778530028e-06,
      "loss": 0.0019,
      "step": 142110
    },
    {
      "epoch": 0.9115135344880112,
      "grad_norm": 0.2066538780927658,
      "learning_rate": 6.615348106028825e-06,
      "loss": 0.0026,
      "step": 142120
    },
    {
      "epoch": 0.9115776713817973,
      "grad_norm": 0.05156905576586723,
      "learning_rate": 6.614818406515904e-06,
      "loss": 0.002,
      "step": 142130
    },
    {
      "epoch": 0.9116418082755834,
      "grad_norm": 0.1713191121816635,
      "learning_rate": 6.614288686768153e-06,
      "loss": 0.0023,
      "step": 142140
    },
    {
      "epoch": 0.9117059451693695,
      "grad_norm": 0.15974517166614532,
      "learning_rate": 6.613758946792212e-06,
      "loss": 0.0024,
      "step": 142150
    },
    {
      "epoch": 0.9117700820631556,
      "grad_norm": 0.23358303308486938,
      "learning_rate": 6.613229186594717e-06,
      "loss": 0.0027,
      "step": 142160
    },
    {
      "epoch": 0.9118342189569417,
      "grad_norm": 0.03668862581253052,
      "learning_rate": 6.6126994061823094e-06,
      "loss": 0.0014,
      "step": 142170
    },
    {
      "epoch": 0.9118983558507278,
      "grad_norm": 0.0947398766875267,
      "learning_rate": 6.612169605561625e-06,
      "loss": 0.0041,
      "step": 142180
    },
    {
      "epoch": 0.9119624927445139,
      "grad_norm": 0.10485409945249557,
      "learning_rate": 6.611639784739303e-06,
      "loss": 0.005,
      "step": 142190
    },
    {
      "epoch": 0.9120266296383,
      "grad_norm": 0.4552534222602844,
      "learning_rate": 6.6111099437219815e-06,
      "loss": 0.0019,
      "step": 142200
    },
    {
      "epoch": 0.9120907665320861,
      "grad_norm": 0.03338133916258812,
      "learning_rate": 6.610580082516302e-06,
      "loss": 0.0024,
      "step": 142210
    },
    {
      "epoch": 0.9121549034258721,
      "grad_norm": 0.20968693494796753,
      "learning_rate": 6.610050201128904e-06,
      "loss": 0.0023,
      "step": 142220
    },
    {
      "epoch": 0.9122190403196583,
      "grad_norm": 0.11558353900909424,
      "learning_rate": 6.609520299566425e-06,
      "loss": 0.003,
      "step": 142230
    },
    {
      "epoch": 0.9122831772134444,
      "grad_norm": 0.22716671228408813,
      "learning_rate": 6.6089903778355056e-06,
      "loss": 0.0025,
      "step": 142240
    },
    {
      "epoch": 0.9123473141072305,
      "grad_norm": 0.09540443867444992,
      "learning_rate": 6.608460435942788e-06,
      "loss": 0.0015,
      "step": 142250
    },
    {
      "epoch": 0.9124114510010166,
      "grad_norm": 0.15700772404670715,
      "learning_rate": 6.607930473894912e-06,
      "loss": 0.0021,
      "step": 142260
    },
    {
      "epoch": 0.9124755878948027,
      "grad_norm": 0.02807980217039585,
      "learning_rate": 6.607400491698518e-06,
      "loss": 0.002,
      "step": 142270
    },
    {
      "epoch": 0.9125397247885888,
      "grad_norm": 0.1129683256149292,
      "learning_rate": 6.606870489360245e-06,
      "loss": 0.003,
      "step": 142280
    },
    {
      "epoch": 0.9126038616823748,
      "grad_norm": 0.2775423228740692,
      "learning_rate": 6.606340466886738e-06,
      "loss": 0.004,
      "step": 142290
    },
    {
      "epoch": 0.912667998576161,
      "grad_norm": 0.16915129125118256,
      "learning_rate": 6.605810424284637e-06,
      "loss": 0.0023,
      "step": 142300
    },
    {
      "epoch": 0.912732135469947,
      "grad_norm": 0.06936042010784149,
      "learning_rate": 6.605280361560583e-06,
      "loss": 0.0017,
      "step": 142310
    },
    {
      "epoch": 0.9127962723637332,
      "grad_norm": 0.3886728286743164,
      "learning_rate": 6.604750278721219e-06,
      "loss": 0.0033,
      "step": 142320
    },
    {
      "epoch": 0.9128604092575192,
      "grad_norm": 0.18234775960445404,
      "learning_rate": 6.604220175773188e-06,
      "loss": 0.0015,
      "step": 142330
    },
    {
      "epoch": 0.9129245461513054,
      "grad_norm": 0.03708178177475929,
      "learning_rate": 6.6036900527231306e-06,
      "loss": 0.0036,
      "step": 142340
    },
    {
      "epoch": 0.9129886830450914,
      "grad_norm": 0.11459732055664062,
      "learning_rate": 6.60315990957769e-06,
      "loss": 0.0076,
      "step": 142350
    },
    {
      "epoch": 0.9130528199388775,
      "grad_norm": 0.12645256519317627,
      "learning_rate": 6.602629746343512e-06,
      "loss": 0.0021,
      "step": 142360
    },
    {
      "epoch": 0.9131169568326636,
      "grad_norm": 0.12674662470817566,
      "learning_rate": 6.602099563027236e-06,
      "loss": 0.0034,
      "step": 142370
    },
    {
      "epoch": 0.9131810937264497,
      "grad_norm": 0.03884998336434364,
      "learning_rate": 6.601569359635509e-06,
      "loss": 0.0024,
      "step": 142380
    },
    {
      "epoch": 0.9132452306202359,
      "grad_norm": 0.24977393448352814,
      "learning_rate": 6.601039136174973e-06,
      "loss": 0.004,
      "step": 142390
    },
    {
      "epoch": 0.9133093675140219,
      "grad_norm": 0.0631488561630249,
      "learning_rate": 6.600508892652273e-06,
      "loss": 0.0034,
      "step": 142400
    },
    {
      "epoch": 0.9133735044078081,
      "grad_norm": 0.206329807639122,
      "learning_rate": 6.599978629074051e-06,
      "loss": 0.0017,
      "step": 142410
    },
    {
      "epoch": 0.9134376413015941,
      "grad_norm": 0.052349966019392014,
      "learning_rate": 6.599448345446955e-06,
      "loss": 0.0029,
      "step": 142420
    },
    {
      "epoch": 0.9135017781953803,
      "grad_norm": 2.474010705947876,
      "learning_rate": 6.598918041777626e-06,
      "loss": 0.016,
      "step": 142430
    },
    {
      "epoch": 0.9135659150891663,
      "grad_norm": 0.05902921408414841,
      "learning_rate": 6.598387718072714e-06,
      "loss": 0.0024,
      "step": 142440
    },
    {
      "epoch": 0.9136300519829524,
      "grad_norm": 0.09081000089645386,
      "learning_rate": 6.597857374338859e-06,
      "loss": 0.0018,
      "step": 142450
    },
    {
      "epoch": 0.9136941888767385,
      "grad_norm": 0.017076801508665085,
      "learning_rate": 6.59732701058271e-06,
      "loss": 0.0016,
      "step": 142460
    },
    {
      "epoch": 0.9137583257705246,
      "grad_norm": 0.09036340564489365,
      "learning_rate": 6.596796626810913e-06,
      "loss": 0.0021,
      "step": 142470
    },
    {
      "epoch": 0.9138224626643107,
      "grad_norm": 0.01195220835506916,
      "learning_rate": 6.596266223030111e-06,
      "loss": 0.0023,
      "step": 142480
    },
    {
      "epoch": 0.9138865995580968,
      "grad_norm": 0.1422252207994461,
      "learning_rate": 6.595735799246954e-06,
      "loss": 0.0027,
      "step": 142490
    },
    {
      "epoch": 0.9139507364518829,
      "grad_norm": 0.07921761274337769,
      "learning_rate": 6.5952053554680865e-06,
      "loss": 0.0024,
      "step": 142500
    },
    {
      "epoch": 0.914014873345669,
      "grad_norm": 0.07937291264533997,
      "learning_rate": 6.594674891700157e-06,
      "loss": 0.0016,
      "step": 142510
    },
    {
      "epoch": 0.9140790102394551,
      "grad_norm": 0.027815645560622215,
      "learning_rate": 6.594144407949809e-06,
      "loss": 0.0012,
      "step": 142520
    },
    {
      "epoch": 0.9141431471332412,
      "grad_norm": 0.19019463658332825,
      "learning_rate": 6.593613904223695e-06,
      "loss": 0.0024,
      "step": 142530
    },
    {
      "epoch": 0.9142072840270273,
      "grad_norm": 0.2717309594154358,
      "learning_rate": 6.593083380528457e-06,
      "loss": 0.0012,
      "step": 142540
    },
    {
      "epoch": 0.9142714209208134,
      "grad_norm": 0.07188116759061813,
      "learning_rate": 6.592552836870747e-06,
      "loss": 0.0034,
      "step": 142550
    },
    {
      "epoch": 0.9143355578145995,
      "grad_norm": 0.06552331894636154,
      "learning_rate": 6.592022273257213e-06,
      "loss": 0.0014,
      "step": 142560
    },
    {
      "epoch": 0.9143996947083856,
      "grad_norm": 0.055768199265003204,
      "learning_rate": 6.5914916896945e-06,
      "loss": 0.0048,
      "step": 142570
    },
    {
      "epoch": 0.9144638316021717,
      "grad_norm": 0.15143147110939026,
      "learning_rate": 6.590961086189259e-06,
      "loss": 0.0023,
      "step": 142580
    },
    {
      "epoch": 0.9145279684959577,
      "grad_norm": 0.16667847335338593,
      "learning_rate": 6.590430462748141e-06,
      "loss": 0.0023,
      "step": 142590
    },
    {
      "epoch": 0.9145921053897439,
      "grad_norm": 0.11879737675189972,
      "learning_rate": 6.58989981937779e-06,
      "loss": 0.0014,
      "step": 142600
    },
    {
      "epoch": 0.9146562422835299,
      "grad_norm": 0.2997582256793976,
      "learning_rate": 6.5893691560848595e-06,
      "loss": 0.004,
      "step": 142610
    },
    {
      "epoch": 0.9147203791773161,
      "grad_norm": 0.017078684642910957,
      "learning_rate": 6.588838472875996e-06,
      "loss": 0.0019,
      "step": 142620
    },
    {
      "epoch": 0.9147845160711021,
      "grad_norm": 0.1263207495212555,
      "learning_rate": 6.588307769757851e-06,
      "loss": 0.0021,
      "step": 142630
    },
    {
      "epoch": 0.9148486529648883,
      "grad_norm": 0.1354454904794693,
      "learning_rate": 6.587777046737076e-06,
      "loss": 0.0018,
      "step": 142640
    },
    {
      "epoch": 0.9149127898586743,
      "grad_norm": 0.12904682755470276,
      "learning_rate": 6.5872463038203185e-06,
      "loss": 0.0045,
      "step": 142650
    },
    {
      "epoch": 0.9149769267524605,
      "grad_norm": 0.12708112597465515,
      "learning_rate": 6.586715541014232e-06,
      "loss": 0.0016,
      "step": 142660
    },
    {
      "epoch": 0.9150410636462466,
      "grad_norm": 0.12494788318872452,
      "learning_rate": 6.5861847583254645e-06,
      "loss": 0.0014,
      "step": 142670
    },
    {
      "epoch": 0.9151052005400326,
      "grad_norm": 0.2556142210960388,
      "learning_rate": 6.585653955760668e-06,
      "loss": 0.0015,
      "step": 142680
    },
    {
      "epoch": 0.9151693374338188,
      "grad_norm": 0.13654863834381104,
      "learning_rate": 6.585123133326495e-06,
      "loss": 0.0022,
      "step": 142690
    },
    {
      "epoch": 0.9152334743276048,
      "grad_norm": 0.08078952133655548,
      "learning_rate": 6.584592291029596e-06,
      "loss": 0.002,
      "step": 142700
    },
    {
      "epoch": 0.915297611221391,
      "grad_norm": 0.09141595661640167,
      "learning_rate": 6.584061428876623e-06,
      "loss": 0.0013,
      "step": 142710
    },
    {
      "epoch": 0.915361748115177,
      "grad_norm": 0.1005011573433876,
      "learning_rate": 6.583530546874228e-06,
      "loss": 0.0028,
      "step": 142720
    },
    {
      "epoch": 0.9154258850089632,
      "grad_norm": 0.11311080306768417,
      "learning_rate": 6.582999645029065e-06,
      "loss": 0.0019,
      "step": 142730
    },
    {
      "epoch": 0.9154900219027492,
      "grad_norm": 0.2210063636302948,
      "learning_rate": 6.5824687233477835e-06,
      "loss": 0.0022,
      "step": 142740
    },
    {
      "epoch": 0.9155541587965353,
      "grad_norm": 0.08203542232513428,
      "learning_rate": 6.581937781837038e-06,
      "loss": 0.0011,
      "step": 142750
    },
    {
      "epoch": 0.9156182956903214,
      "grad_norm": 0.08673785626888275,
      "learning_rate": 6.581406820503482e-06,
      "loss": 0.0017,
      "step": 142760
    },
    {
      "epoch": 0.9156824325841075,
      "grad_norm": 0.2789863646030426,
      "learning_rate": 6.5808758393537685e-06,
      "loss": 0.0026,
      "step": 142770
    },
    {
      "epoch": 0.9157465694778936,
      "grad_norm": 0.05158243700861931,
      "learning_rate": 6.580344838394551e-06,
      "loss": 0.0023,
      "step": 142780
    },
    {
      "epoch": 0.9158107063716797,
      "grad_norm": 0.13372482359409332,
      "learning_rate": 6.579813817632482e-06,
      "loss": 0.0035,
      "step": 142790
    },
    {
      "epoch": 0.9158748432654659,
      "grad_norm": 0.17898721992969513,
      "learning_rate": 6.579282777074218e-06,
      "loss": 0.0023,
      "step": 142800
    },
    {
      "epoch": 0.9159389801592519,
      "grad_norm": 0.07201813161373138,
      "learning_rate": 6.578751716726411e-06,
      "loss": 0.0056,
      "step": 142810
    },
    {
      "epoch": 0.916003117053038,
      "grad_norm": 0.1174902692437172,
      "learning_rate": 6.578220636595719e-06,
      "loss": 0.0032,
      "step": 142820
    },
    {
      "epoch": 0.9160672539468241,
      "grad_norm": 0.18898825347423553,
      "learning_rate": 6.577689536688792e-06,
      "loss": 0.0016,
      "step": 142830
    },
    {
      "epoch": 0.9161313908406102,
      "grad_norm": 0.17814204096794128,
      "learning_rate": 6.5771584170122885e-06,
      "loss": 0.0022,
      "step": 142840
    },
    {
      "epoch": 0.9161955277343963,
      "grad_norm": 0.19003327190876007,
      "learning_rate": 6.576627277572863e-06,
      "loss": 0.0039,
      "step": 142850
    },
    {
      "epoch": 0.9162596646281824,
      "grad_norm": 0.16185325384140015,
      "learning_rate": 6.576096118377171e-06,
      "loss": 0.0044,
      "step": 142860
    },
    {
      "epoch": 0.9163238015219685,
      "grad_norm": 0.16603900492191315,
      "learning_rate": 6.5755649394318675e-06,
      "loss": 0.0032,
      "step": 142870
    },
    {
      "epoch": 0.9163879384157546,
      "grad_norm": 0.1000480130314827,
      "learning_rate": 6.575033740743609e-06,
      "loss": 0.0024,
      "step": 142880
    },
    {
      "epoch": 0.9164520753095406,
      "grad_norm": 0.045742470771074295,
      "learning_rate": 6.5745025223190535e-06,
      "loss": 0.0041,
      "step": 142890
    },
    {
      "epoch": 0.9165162122033268,
      "grad_norm": 0.1930939108133316,
      "learning_rate": 6.573971284164855e-06,
      "loss": 0.0022,
      "step": 142900
    },
    {
      "epoch": 0.9165803490971128,
      "grad_norm": 0.27457255125045776,
      "learning_rate": 6.573440026287673e-06,
      "loss": 0.0027,
      "step": 142910
    },
    {
      "epoch": 0.916644485990899,
      "grad_norm": 0.21804606914520264,
      "learning_rate": 6.5729087486941615e-06,
      "loss": 0.0028,
      "step": 142920
    },
    {
      "epoch": 0.916708622884685,
      "grad_norm": 0.23076127469539642,
      "learning_rate": 6.57237745139098e-06,
      "loss": 0.0022,
      "step": 142930
    },
    {
      "epoch": 0.9167727597784712,
      "grad_norm": 0.29028287529945374,
      "learning_rate": 6.5718461343847835e-06,
      "loss": 0.0034,
      "step": 142940
    },
    {
      "epoch": 0.9168368966722573,
      "grad_norm": 0.15482687950134277,
      "learning_rate": 6.571314797682234e-06,
      "loss": 0.0019,
      "step": 142950
    },
    {
      "epoch": 0.9169010335660434,
      "grad_norm": 0.13907399773597717,
      "learning_rate": 6.570783441289985e-06,
      "loss": 0.0023,
      "step": 142960
    },
    {
      "epoch": 0.9169651704598295,
      "grad_norm": 0.089432492852211,
      "learning_rate": 6.570252065214699e-06,
      "loss": 0.0015,
      "step": 142970
    },
    {
      "epoch": 0.9170293073536155,
      "grad_norm": 0.097746342420578,
      "learning_rate": 6.56972066946303e-06,
      "loss": 0.0013,
      "step": 142980
    },
    {
      "epoch": 0.9170934442474017,
      "grad_norm": 0.11432447284460068,
      "learning_rate": 6.56918925404164e-06,
      "loss": 0.0023,
      "step": 142990
    },
    {
      "epoch": 0.9171575811411877,
      "grad_norm": 0.06727922707796097,
      "learning_rate": 6.568657818957188e-06,
      "loss": 0.0027,
      "step": 143000
    },
    {
      "epoch": 0.9172217180349739,
      "grad_norm": 0.07588520646095276,
      "learning_rate": 6.5681263642163316e-06,
      "loss": 0.002,
      "step": 143010
    },
    {
      "epoch": 0.9172858549287599,
      "grad_norm": 0.03804763779044151,
      "learning_rate": 6.567594889825733e-06,
      "loss": 0.002,
      "step": 143020
    },
    {
      "epoch": 0.9173499918225461,
      "grad_norm": 0.01098368689417839,
      "learning_rate": 6.5670633957920475e-06,
      "loss": 0.0029,
      "step": 143030
    },
    {
      "epoch": 0.9174141287163321,
      "grad_norm": 0.13806407153606415,
      "learning_rate": 6.566531882121938e-06,
      "loss": 0.0016,
      "step": 143040
    },
    {
      "epoch": 0.9174782656101182,
      "grad_norm": 0.10076095908880234,
      "learning_rate": 6.566000348822066e-06,
      "loss": 0.0033,
      "step": 143050
    },
    {
      "epoch": 0.9175424025039043,
      "grad_norm": 0.03555780649185181,
      "learning_rate": 6.565468795899088e-06,
      "loss": 0.0013,
      "step": 143060
    },
    {
      "epoch": 0.9176065393976904,
      "grad_norm": 0.3157411515712738,
      "learning_rate": 6.5649372233596685e-06,
      "loss": 0.0026,
      "step": 143070
    },
    {
      "epoch": 0.9176706762914765,
      "grad_norm": 0.20569480955600739,
      "learning_rate": 6.564405631210469e-06,
      "loss": 0.0024,
      "step": 143080
    },
    {
      "epoch": 0.9177348131852626,
      "grad_norm": 0.2229042947292328,
      "learning_rate": 6.563874019458146e-06,
      "loss": 0.0027,
      "step": 143090
    },
    {
      "epoch": 0.9177989500790488,
      "grad_norm": 0.14811183512210846,
      "learning_rate": 6.563342388109366e-06,
      "loss": 0.003,
      "step": 143100
    },
    {
      "epoch": 0.9178630869728348,
      "grad_norm": 0.10701605677604675,
      "learning_rate": 6.562810737170787e-06,
      "loss": 0.0017,
      "step": 143110
    },
    {
      "epoch": 0.917927223866621,
      "grad_norm": 0.1293468326330185,
      "learning_rate": 6.562279066649073e-06,
      "loss": 0.0019,
      "step": 143120
    },
    {
      "epoch": 0.917991360760407,
      "grad_norm": 0.08432118594646454,
      "learning_rate": 6.561747376550887e-06,
      "loss": 0.004,
      "step": 143130
    },
    {
      "epoch": 0.9180554976541931,
      "grad_norm": 0.16910623013973236,
      "learning_rate": 6.5612156668828885e-06,
      "loss": 0.0019,
      "step": 143140
    },
    {
      "epoch": 0.9181196345479792,
      "grad_norm": 0.07571445405483246,
      "learning_rate": 6.560683937651743e-06,
      "loss": 0.0016,
      "step": 143150
    },
    {
      "epoch": 0.9181837714417653,
      "grad_norm": 0.07605468481779099,
      "learning_rate": 6.560152188864112e-06,
      "loss": 0.0026,
      "step": 143160
    },
    {
      "epoch": 0.9182479083355514,
      "grad_norm": 0.061501987278461456,
      "learning_rate": 6.559620420526659e-06,
      "loss": 0.0028,
      "step": 143170
    },
    {
      "epoch": 0.9183120452293375,
      "grad_norm": 0.03861086070537567,
      "learning_rate": 6.559088632646048e-06,
      "loss": 0.0053,
      "step": 143180
    },
    {
      "epoch": 0.9183761821231236,
      "grad_norm": 0.4427987337112427,
      "learning_rate": 6.558556825228941e-06,
      "loss": 0.0013,
      "step": 143190
    },
    {
      "epoch": 0.9184403190169097,
      "grad_norm": 0.11889459192752838,
      "learning_rate": 6.5580249982820035e-06,
      "loss": 0.0051,
      "step": 143200
    },
    {
      "epoch": 0.9185044559106957,
      "grad_norm": 0.029812904074788094,
      "learning_rate": 6.5574931518119e-06,
      "loss": 0.0014,
      "step": 143210
    },
    {
      "epoch": 0.9185685928044819,
      "grad_norm": 0.05113794654607773,
      "learning_rate": 6.556961285825293e-06,
      "loss": 0.0018,
      "step": 143220
    },
    {
      "epoch": 0.918632729698268,
      "grad_norm": 0.06058286875486374,
      "learning_rate": 6.55642940032885e-06,
      "loss": 0.0017,
      "step": 143230
    },
    {
      "epoch": 0.9186968665920541,
      "grad_norm": 0.070572130382061,
      "learning_rate": 6.555897495329232e-06,
      "loss": 0.0023,
      "step": 143240
    },
    {
      "epoch": 0.9187610034858402,
      "grad_norm": 0.0866912454366684,
      "learning_rate": 6.555365570833109e-06,
      "loss": 0.0033,
      "step": 143250
    },
    {
      "epoch": 0.9188251403796263,
      "grad_norm": 0.079118512570858,
      "learning_rate": 6.5548336268471415e-06,
      "loss": 0.0012,
      "step": 143260
    },
    {
      "epoch": 0.9188892772734124,
      "grad_norm": 0.1792449653148651,
      "learning_rate": 6.5543016633779985e-06,
      "loss": 0.0021,
      "step": 143270
    },
    {
      "epoch": 0.9189534141671984,
      "grad_norm": 0.1889599859714508,
      "learning_rate": 6.553769680432345e-06,
      "loss": 0.0026,
      "step": 143280
    },
    {
      "epoch": 0.9190175510609846,
      "grad_norm": 0.1904626339673996,
      "learning_rate": 6.553237678016847e-06,
      "loss": 0.0022,
      "step": 143290
    },
    {
      "epoch": 0.9190816879547706,
      "grad_norm": 0.08636914938688278,
      "learning_rate": 6.55270565613817e-06,
      "loss": 0.0031,
      "step": 143300
    },
    {
      "epoch": 0.9191458248485568,
      "grad_norm": 0.07031205296516418,
      "learning_rate": 6.552173614802982e-06,
      "loss": 0.0017,
      "step": 143310
    },
    {
      "epoch": 0.9192099617423428,
      "grad_norm": 0.08173853904008865,
      "learning_rate": 6.551641554017949e-06,
      "loss": 0.0017,
      "step": 143320
    },
    {
      "epoch": 0.919274098636129,
      "grad_norm": 0.008939284831285477,
      "learning_rate": 6.551109473789739e-06,
      "loss": 0.0026,
      "step": 143330
    },
    {
      "epoch": 0.919338235529915,
      "grad_norm": 0.14659515023231506,
      "learning_rate": 6.550577374125018e-06,
      "loss": 0.0021,
      "step": 143340
    },
    {
      "epoch": 0.9194023724237012,
      "grad_norm": 0.0790574699640274,
      "learning_rate": 6.5500452550304546e-06,
      "loss": 0.0015,
      "step": 143350
    },
    {
      "epoch": 0.9194665093174872,
      "grad_norm": 0.03465283662080765,
      "learning_rate": 6.549513116512717e-06,
      "loss": 0.0019,
      "step": 143360
    },
    {
      "epoch": 0.9195306462112733,
      "grad_norm": 0.10260052233934402,
      "learning_rate": 6.548980958578471e-06,
      "loss": 0.0045,
      "step": 143370
    },
    {
      "epoch": 0.9195947831050595,
      "grad_norm": 0.07524410635232925,
      "learning_rate": 6.548448781234389e-06,
      "loss": 0.0024,
      "step": 143380
    },
    {
      "epoch": 0.9196589199988455,
      "grad_norm": 0.1725485920906067,
      "learning_rate": 6.547916584487135e-06,
      "loss": 0.0026,
      "step": 143390
    },
    {
      "epoch": 0.9197230568926317,
      "grad_norm": 0.03351062163710594,
      "learning_rate": 6.5473843683433815e-06,
      "loss": 0.0019,
      "step": 143400
    },
    {
      "epoch": 0.9197871937864177,
      "grad_norm": 0.17824223637580872,
      "learning_rate": 6.5468521328097935e-06,
      "loss": 0.0052,
      "step": 143410
    },
    {
      "epoch": 0.9198513306802039,
      "grad_norm": 0.46944063901901245,
      "learning_rate": 6.546319877893045e-06,
      "loss": 0.0031,
      "step": 143420
    },
    {
      "epoch": 0.9199154675739899,
      "grad_norm": 0.1069873496890068,
      "learning_rate": 6.545787603599803e-06,
      "loss": 0.0015,
      "step": 143430
    },
    {
      "epoch": 0.919979604467776,
      "grad_norm": 0.01805989444255829,
      "learning_rate": 6.545255309936736e-06,
      "loss": 0.003,
      "step": 143440
    },
    {
      "epoch": 0.9200437413615621,
      "grad_norm": 0.059725284576416016,
      "learning_rate": 6.5447229969105166e-06,
      "loss": 0.0025,
      "step": 143450
    },
    {
      "epoch": 0.9201078782553482,
      "grad_norm": 0.07746428996324539,
      "learning_rate": 6.544190664527814e-06,
      "loss": 0.002,
      "step": 143460
    },
    {
      "epoch": 0.9201720151491343,
      "grad_norm": 0.022890908643603325,
      "learning_rate": 6.543658312795299e-06,
      "loss": 0.0044,
      "step": 143470
    },
    {
      "epoch": 0.9202361520429204,
      "grad_norm": 0.11122360080480576,
      "learning_rate": 6.543125941719643e-06,
      "loss": 0.0059,
      "step": 143480
    },
    {
      "epoch": 0.9203002889367065,
      "grad_norm": 0.02852690778672695,
      "learning_rate": 6.542593551307514e-06,
      "loss": 0.0031,
      "step": 143490
    },
    {
      "epoch": 0.9203644258304926,
      "grad_norm": 0.21097946166992188,
      "learning_rate": 6.542061141565588e-06,
      "loss": 0.0026,
      "step": 143500
    },
    {
      "epoch": 0.9204285627242788,
      "grad_norm": 0.04008246585726738,
      "learning_rate": 6.541528712500531e-06,
      "loss": 0.002,
      "step": 143510
    },
    {
      "epoch": 0.9204926996180648,
      "grad_norm": 0.2406572848558426,
      "learning_rate": 6.54099626411902e-06,
      "loss": 0.0026,
      "step": 143520
    },
    {
      "epoch": 0.9205568365118509,
      "grad_norm": 0.21601951122283936,
      "learning_rate": 6.540463796427723e-06,
      "loss": 0.0016,
      "step": 143530
    },
    {
      "epoch": 0.920620973405637,
      "grad_norm": 0.1107027605175972,
      "learning_rate": 6.539931309433313e-06,
      "loss": 0.0052,
      "step": 143540
    },
    {
      "epoch": 0.9206851102994231,
      "grad_norm": 0.27741020917892456,
      "learning_rate": 6.539398803142465e-06,
      "loss": 0.0025,
      "step": 143550
    },
    {
      "epoch": 0.9207492471932092,
      "grad_norm": 0.11819496750831604,
      "learning_rate": 6.5388662775618485e-06,
      "loss": 0.0011,
      "step": 143560
    },
    {
      "epoch": 0.9208133840869953,
      "grad_norm": 0.10231605172157288,
      "learning_rate": 6.538333732698139e-06,
      "loss": 0.0022,
      "step": 143570
    },
    {
      "epoch": 0.9208775209807813,
      "grad_norm": 0.2636953890323639,
      "learning_rate": 6.537801168558007e-06,
      "loss": 0.0036,
      "step": 143580
    },
    {
      "epoch": 0.9209416578745675,
      "grad_norm": 0.2232527881860733,
      "learning_rate": 6.537268585148127e-06,
      "loss": 0.0029,
      "step": 143590
    },
    {
      "epoch": 0.9210057947683535,
      "grad_norm": 0.1344190239906311,
      "learning_rate": 6.536735982475174e-06,
      "loss": 0.0017,
      "step": 143600
    },
    {
      "epoch": 0.9210699316621397,
      "grad_norm": 0.08762157708406448,
      "learning_rate": 6.536203360545821e-06,
      "loss": 0.0019,
      "step": 143610
    },
    {
      "epoch": 0.9211340685559257,
      "grad_norm": 0.14623139798641205,
      "learning_rate": 6.535670719366741e-06,
      "loss": 0.0033,
      "step": 143620
    },
    {
      "epoch": 0.9211982054497119,
      "grad_norm": 0.04193584993481636,
      "learning_rate": 6.535138058944611e-06,
      "loss": 0.0028,
      "step": 143630
    },
    {
      "epoch": 0.9212623423434979,
      "grad_norm": 0.1585153192281723,
      "learning_rate": 6.534605379286103e-06,
      "loss": 0.0092,
      "step": 143640
    },
    {
      "epoch": 0.9213264792372841,
      "grad_norm": 0.2073761224746704,
      "learning_rate": 6.534072680397893e-06,
      "loss": 0.0019,
      "step": 143650
    },
    {
      "epoch": 0.9213906161310702,
      "grad_norm": 0.11256733536720276,
      "learning_rate": 6.533539962286655e-06,
      "loss": 0.0023,
      "step": 143660
    },
    {
      "epoch": 0.9214547530248562,
      "grad_norm": 0.03633716329932213,
      "learning_rate": 6.533007224959067e-06,
      "loss": 0.002,
      "step": 143670
    },
    {
      "epoch": 0.9215188899186424,
      "grad_norm": 0.01778295449912548,
      "learning_rate": 6.532474468421801e-06,
      "loss": 0.0022,
      "step": 143680
    },
    {
      "epoch": 0.9215830268124284,
      "grad_norm": 0.11002130806446075,
      "learning_rate": 6.531941692681535e-06,
      "loss": 0.0016,
      "step": 143690
    },
    {
      "epoch": 0.9216471637062146,
      "grad_norm": 0.09549775719642639,
      "learning_rate": 6.531408897744946e-06,
      "loss": 0.0028,
      "step": 143700
    },
    {
      "epoch": 0.9217113006000006,
      "grad_norm": 0.10814757645130157,
      "learning_rate": 6.530876083618708e-06,
      "loss": 0.0021,
      "step": 143710
    },
    {
      "epoch": 0.9217754374937868,
      "grad_norm": 0.12686097621917725,
      "learning_rate": 6.530343250309499e-06,
      "loss": 0.0035,
      "step": 143720
    },
    {
      "epoch": 0.9218395743875728,
      "grad_norm": 0.22388538718223572,
      "learning_rate": 6.5298103978239946e-06,
      "loss": 0.003,
      "step": 143730
    },
    {
      "epoch": 0.921903711281359,
      "grad_norm": 0.12337753176689148,
      "learning_rate": 6.529277526168873e-06,
      "loss": 0.0096,
      "step": 143740
    },
    {
      "epoch": 0.921967848175145,
      "grad_norm": 0.07037145644426346,
      "learning_rate": 6.528744635350812e-06,
      "loss": 0.0011,
      "step": 143750
    },
    {
      "epoch": 0.9220319850689311,
      "grad_norm": 0.0782921090722084,
      "learning_rate": 6.528211725376488e-06,
      "loss": 0.0027,
      "step": 143760
    },
    {
      "epoch": 0.9220961219627172,
      "grad_norm": 0.13416391611099243,
      "learning_rate": 6.5276787962525775e-06,
      "loss": 0.0019,
      "step": 143770
    },
    {
      "epoch": 0.9221602588565033,
      "grad_norm": 0.08078598231077194,
      "learning_rate": 6.5271458479857606e-06,
      "loss": 0.0026,
      "step": 143780
    },
    {
      "epoch": 0.9222243957502895,
      "grad_norm": 0.06023133918642998,
      "learning_rate": 6.526612880582714e-06,
      "loss": 0.0019,
      "step": 143790
    },
    {
      "epoch": 0.9222885326440755,
      "grad_norm": 0.06747840344905853,
      "learning_rate": 6.526079894050119e-06,
      "loss": 0.002,
      "step": 143800
    },
    {
      "epoch": 0.9223526695378617,
      "grad_norm": 0.020860394462943077,
      "learning_rate": 6.525546888394651e-06,
      "loss": 0.0025,
      "step": 143810
    },
    {
      "epoch": 0.9224168064316477,
      "grad_norm": 0.1317310333251953,
      "learning_rate": 6.5250138636229895e-06,
      "loss": 0.0021,
      "step": 143820
    },
    {
      "epoch": 0.9224809433254338,
      "grad_norm": 0.17135535180568695,
      "learning_rate": 6.5244808197418145e-06,
      "loss": 0.0024,
      "step": 143830
    },
    {
      "epoch": 0.9225450802192199,
      "grad_norm": 0.1718551367521286,
      "learning_rate": 6.523947756757807e-06,
      "loss": 0.0021,
      "step": 143840
    },
    {
      "epoch": 0.922609217113006,
      "grad_norm": 0.17481783032417297,
      "learning_rate": 6.523414674677643e-06,
      "loss": 0.0028,
      "step": 143850
    },
    {
      "epoch": 0.9226733540067921,
      "grad_norm": 0.06103501841425896,
      "learning_rate": 6.522881573508005e-06,
      "loss": 0.0031,
      "step": 143860
    },
    {
      "epoch": 0.9227374909005782,
      "grad_norm": 0.14446958899497986,
      "learning_rate": 6.522348453255572e-06,
      "loss": 0.0025,
      "step": 143870
    },
    {
      "epoch": 0.9228016277943643,
      "grad_norm": 0.17649446427822113,
      "learning_rate": 6.521815313927027e-06,
      "loss": 0.0018,
      "step": 143880
    },
    {
      "epoch": 0.9228657646881504,
      "grad_norm": 0.1549309492111206,
      "learning_rate": 6.521282155529047e-06,
      "loss": 0.0034,
      "step": 143890
    },
    {
      "epoch": 0.9229299015819364,
      "grad_norm": 0.08576612919569016,
      "learning_rate": 6.5207489780683146e-06,
      "loss": 0.0022,
      "step": 143900
    },
    {
      "epoch": 0.9229940384757226,
      "grad_norm": 0.07953940331935883,
      "learning_rate": 6.52021578155151e-06,
      "loss": 0.0019,
      "step": 143910
    },
    {
      "epoch": 0.9230581753695086,
      "grad_norm": 0.08306507021188736,
      "learning_rate": 6.519682565985315e-06,
      "loss": 0.0023,
      "step": 143920
    },
    {
      "epoch": 0.9231223122632948,
      "grad_norm": 0.034287504851818085,
      "learning_rate": 6.519149331376413e-06,
      "loss": 0.0019,
      "step": 143930
    },
    {
      "epoch": 0.9231864491570809,
      "grad_norm": 0.027775052934885025,
      "learning_rate": 6.518616077731482e-06,
      "loss": 0.0037,
      "step": 143940
    },
    {
      "epoch": 0.923250586050867,
      "grad_norm": 0.08457817882299423,
      "learning_rate": 6.518082805057208e-06,
      "loss": 0.003,
      "step": 143950
    },
    {
      "epoch": 0.9233147229446531,
      "grad_norm": 0.19895368814468384,
      "learning_rate": 6.5175495133602704e-06,
      "loss": 0.002,
      "step": 143960
    },
    {
      "epoch": 0.9233788598384391,
      "grad_norm": 0.06647168844938278,
      "learning_rate": 6.517016202647354e-06,
      "loss": 0.0029,
      "step": 143970
    },
    {
      "epoch": 0.9234429967322253,
      "grad_norm": 0.08054918050765991,
      "learning_rate": 6.5164828729251395e-06,
      "loss": 0.0024,
      "step": 143980
    },
    {
      "epoch": 0.9235071336260113,
      "grad_norm": 0.05321136489510536,
      "learning_rate": 6.515949524200311e-06,
      "loss": 0.0025,
      "step": 143990
    },
    {
      "epoch": 0.9235712705197975,
      "grad_norm": 0.35143348574638367,
      "learning_rate": 6.515416156479551e-06,
      "loss": 0.0032,
      "step": 144000
    },
    {
      "epoch": 0.9236354074135835,
      "grad_norm": 0.048732027411460876,
      "learning_rate": 6.514882769769545e-06,
      "loss": 0.0019,
      "step": 144010
    },
    {
      "epoch": 0.9236995443073697,
      "grad_norm": 0.056389711797237396,
      "learning_rate": 6.514349364076973e-06,
      "loss": 0.0023,
      "step": 144020
    },
    {
      "epoch": 0.9237636812011557,
      "grad_norm": 0.03896265849471092,
      "learning_rate": 6.513815939408523e-06,
      "loss": 0.0018,
      "step": 144030
    },
    {
      "epoch": 0.9238278180949419,
      "grad_norm": 0.15420575439929962,
      "learning_rate": 6.513282495770876e-06,
      "loss": 0.0032,
      "step": 144040
    },
    {
      "epoch": 0.9238919549887279,
      "grad_norm": 0.14683672785758972,
      "learning_rate": 6.5127490331707184e-06,
      "loss": 0.0017,
      "step": 144050
    },
    {
      "epoch": 0.923956091882514,
      "grad_norm": 0.024806208908557892,
      "learning_rate": 6.512215551614735e-06,
      "loss": 0.0032,
      "step": 144060
    },
    {
      "epoch": 0.9240202287763002,
      "grad_norm": 0.11284174770116806,
      "learning_rate": 6.511682051109609e-06,
      "loss": 0.0029,
      "step": 144070
    },
    {
      "epoch": 0.9240843656700862,
      "grad_norm": 0.04989278316497803,
      "learning_rate": 6.5111485316620284e-06,
      "loss": 0.0011,
      "step": 144080
    },
    {
      "epoch": 0.9241485025638724,
      "grad_norm": 0.14955037832260132,
      "learning_rate": 6.510614993278674e-06,
      "loss": 0.0031,
      "step": 144090
    },
    {
      "epoch": 0.9242126394576584,
      "grad_norm": 0.09255171567201614,
      "learning_rate": 6.510081435966235e-06,
      "loss": 0.0015,
      "step": 144100
    },
    {
      "epoch": 0.9242767763514446,
      "grad_norm": 0.1275353878736496,
      "learning_rate": 6.509547859731398e-06,
      "loss": 0.0024,
      "step": 144110
    },
    {
      "epoch": 0.9243409132452306,
      "grad_norm": 0.07732610404491425,
      "learning_rate": 6.509014264580846e-06,
      "loss": 0.0023,
      "step": 144120
    },
    {
      "epoch": 0.9244050501390167,
      "grad_norm": 0.05314570665359497,
      "learning_rate": 6.508480650521266e-06,
      "loss": 0.0018,
      "step": 144130
    },
    {
      "epoch": 0.9244691870328028,
      "grad_norm": 0.13815511763095856,
      "learning_rate": 6.507947017559347e-06,
      "loss": 0.0022,
      "step": 144140
    },
    {
      "epoch": 0.9245333239265889,
      "grad_norm": 0.07134740799665451,
      "learning_rate": 6.507413365701773e-06,
      "loss": 0.0017,
      "step": 144150
    },
    {
      "epoch": 0.924597460820375,
      "grad_norm": 0.08940762281417847,
      "learning_rate": 6.506879694955233e-06,
      "loss": 0.0031,
      "step": 144160
    },
    {
      "epoch": 0.9246615977141611,
      "grad_norm": 0.14411704242229462,
      "learning_rate": 6.506346005326412e-06,
      "loss": 0.0037,
      "step": 144170
    },
    {
      "epoch": 0.9247257346079472,
      "grad_norm": 0.08996514976024628,
      "learning_rate": 6.505812296822002e-06,
      "loss": 0.0025,
      "step": 144180
    },
    {
      "epoch": 0.9247898715017333,
      "grad_norm": 0.11479675769805908,
      "learning_rate": 6.505278569448685e-06,
      "loss": 0.0079,
      "step": 144190
    },
    {
      "epoch": 0.9248540083955193,
      "grad_norm": 0.14187529683113098,
      "learning_rate": 6.504744823213152e-06,
      "loss": 0.0031,
      "step": 144200
    },
    {
      "epoch": 0.9249181452893055,
      "grad_norm": 0.1020754873752594,
      "learning_rate": 6.504211058122091e-06,
      "loss": 0.0021,
      "step": 144210
    },
    {
      "epoch": 0.9249822821830916,
      "grad_norm": 0.07498777657747269,
      "learning_rate": 6.503677274182192e-06,
      "loss": 0.003,
      "step": 144220
    },
    {
      "epoch": 0.9250464190768777,
      "grad_norm": 0.05531124398112297,
      "learning_rate": 6.5031434714001395e-06,
      "loss": 0.0027,
      "step": 144230
    },
    {
      "epoch": 0.9251105559706638,
      "grad_norm": 0.15193775296211243,
      "learning_rate": 6.502609649782627e-06,
      "loss": 0.0019,
      "step": 144240
    },
    {
      "epoch": 0.9251746928644499,
      "grad_norm": 0.26516246795654297,
      "learning_rate": 6.502075809336341e-06,
      "loss": 0.0025,
      "step": 144250
    },
    {
      "epoch": 0.925238829758236,
      "grad_norm": 0.23687143623828888,
      "learning_rate": 6.501541950067971e-06,
      "loss": 0.0045,
      "step": 144260
    },
    {
      "epoch": 0.925302966652022,
      "grad_norm": 0.06759777665138245,
      "learning_rate": 6.501008071984209e-06,
      "loss": 0.0012,
      "step": 144270
    },
    {
      "epoch": 0.9253671035458082,
      "grad_norm": 0.31802356243133545,
      "learning_rate": 6.500474175091742e-06,
      "loss": 0.0036,
      "step": 144280
    },
    {
      "epoch": 0.9254312404395942,
      "grad_norm": 0.06832915544509888,
      "learning_rate": 6.499940259397262e-06,
      "loss": 0.002,
      "step": 144290
    },
    {
      "epoch": 0.9254953773333804,
      "grad_norm": 0.04168454185128212,
      "learning_rate": 6.4994063249074565e-06,
      "loss": 0.0033,
      "step": 144300
    },
    {
      "epoch": 0.9255595142271664,
      "grad_norm": 0.11130115389823914,
      "learning_rate": 6.498872371629021e-06,
      "loss": 0.0021,
      "step": 144310
    },
    {
      "epoch": 0.9256236511209526,
      "grad_norm": 0.08546898514032364,
      "learning_rate": 6.498338399568641e-06,
      "loss": 0.0016,
      "step": 144320
    },
    {
      "epoch": 0.9256877880147386,
      "grad_norm": 0.13469122350215912,
      "learning_rate": 6.497804408733012e-06,
      "loss": 0.0014,
      "step": 144330
    },
    {
      "epoch": 0.9257519249085248,
      "grad_norm": 0.07961229234933853,
      "learning_rate": 6.497270399128821e-06,
      "loss": 0.0019,
      "step": 144340
    },
    {
      "epoch": 0.9258160618023108,
      "grad_norm": 0.19842271506786346,
      "learning_rate": 6.496736370762764e-06,
      "loss": 0.0035,
      "step": 144350
    },
    {
      "epoch": 0.925880198696097,
      "grad_norm": 0.347851037979126,
      "learning_rate": 6.49620232364153e-06,
      "loss": 0.0024,
      "step": 144360
    },
    {
      "epoch": 0.9259443355898831,
      "grad_norm": 0.10434811562299728,
      "learning_rate": 6.495668257771811e-06,
      "loss": 0.0025,
      "step": 144370
    },
    {
      "epoch": 0.9260084724836691,
      "grad_norm": 0.06839905679225922,
      "learning_rate": 6.4951341731603e-06,
      "loss": 0.0018,
      "step": 144380
    },
    {
      "epoch": 0.9260726093774553,
      "grad_norm": 0.15542304515838623,
      "learning_rate": 6.4946000698136876e-06,
      "loss": 0.0025,
      "step": 144390
    },
    {
      "epoch": 0.9261367462712413,
      "grad_norm": 0.03447660058736801,
      "learning_rate": 6.494065947738672e-06,
      "loss": 0.0029,
      "step": 144400
    },
    {
      "epoch": 0.9262008831650275,
      "grad_norm": 0.1372140794992447,
      "learning_rate": 6.49353180694194e-06,
      "loss": 0.0021,
      "step": 144410
    },
    {
      "epoch": 0.9262650200588135,
      "grad_norm": 0.10015080869197845,
      "learning_rate": 6.492997647430186e-06,
      "loss": 0.0041,
      "step": 144420
    },
    {
      "epoch": 0.9263291569525997,
      "grad_norm": 0.19411984086036682,
      "learning_rate": 6.492463469210106e-06,
      "loss": 0.0023,
      "step": 144430
    },
    {
      "epoch": 0.9263932938463857,
      "grad_norm": 0.028752420097589493,
      "learning_rate": 6.491929272288392e-06,
      "loss": 0.0025,
      "step": 144440
    },
    {
      "epoch": 0.9264574307401718,
      "grad_norm": 0.08016210049390793,
      "learning_rate": 6.491395056671736e-06,
      "loss": 0.0031,
      "step": 144450
    },
    {
      "epoch": 0.9265215676339579,
      "grad_norm": 0.25969594717025757,
      "learning_rate": 6.490860822366838e-06,
      "loss": 0.002,
      "step": 144460
    },
    {
      "epoch": 0.926585704527744,
      "grad_norm": 0.09981193393468857,
      "learning_rate": 6.490326569380385e-06,
      "loss": 0.0026,
      "step": 144470
    },
    {
      "epoch": 0.9266498414215301,
      "grad_norm": 0.12202033400535583,
      "learning_rate": 6.489792297719076e-06,
      "loss": 0.0025,
      "step": 144480
    },
    {
      "epoch": 0.9267139783153162,
      "grad_norm": 0.16252553462982178,
      "learning_rate": 6.489258007389605e-06,
      "loss": 0.0018,
      "step": 144490
    },
    {
      "epoch": 0.9267781152091024,
      "grad_norm": 0.19570359587669373,
      "learning_rate": 6.488723698398667e-06,
      "loss": 0.0023,
      "step": 144500
    },
    {
      "epoch": 0.9268422521028884,
      "grad_norm": 0.08531318604946136,
      "learning_rate": 6.4881893707529566e-06,
      "loss": 0.0028,
      "step": 144510
    },
    {
      "epoch": 0.9269063889966745,
      "grad_norm": 0.2799188196659088,
      "learning_rate": 6.487655024459171e-06,
      "loss": 0.0023,
      "step": 144520
    },
    {
      "epoch": 0.9269705258904606,
      "grad_norm": 0.054953236132860184,
      "learning_rate": 6.487120659524002e-06,
      "loss": 0.0014,
      "step": 144530
    },
    {
      "epoch": 0.9270346627842467,
      "grad_norm": 0.14100810885429382,
      "learning_rate": 6.48658627595415e-06,
      "loss": 0.0025,
      "step": 144540
    },
    {
      "epoch": 0.9270987996780328,
      "grad_norm": 0.050611674785614014,
      "learning_rate": 6.48605187375631e-06,
      "loss": 0.0029,
      "step": 144550
    },
    {
      "epoch": 0.9271629365718189,
      "grad_norm": 0.13280843198299408,
      "learning_rate": 6.485517452937177e-06,
      "loss": 0.0017,
      "step": 144560
    },
    {
      "epoch": 0.927227073465605,
      "grad_norm": 0.06337030977010727,
      "learning_rate": 6.48498301350345e-06,
      "loss": 0.0026,
      "step": 144570
    },
    {
      "epoch": 0.9272912103593911,
      "grad_norm": 0.08777425438165665,
      "learning_rate": 6.484448555461823e-06,
      "loss": 0.0011,
      "step": 144580
    },
    {
      "epoch": 0.9273553472531771,
      "grad_norm": 0.06910335272550583,
      "learning_rate": 6.483914078818995e-06,
      "loss": 0.002,
      "step": 144590
    },
    {
      "epoch": 0.9274194841469633,
      "grad_norm": 0.09499827027320862,
      "learning_rate": 6.483379583581662e-06,
      "loss": 0.0031,
      "step": 144600
    },
    {
      "epoch": 0.9274836210407493,
      "grad_norm": 0.05050915479660034,
      "learning_rate": 6.482845069756525e-06,
      "loss": 0.0014,
      "step": 144610
    },
    {
      "epoch": 0.9275477579345355,
      "grad_norm": 0.23042018711566925,
      "learning_rate": 6.482310537350278e-06,
      "loss": 0.0028,
      "step": 144620
    },
    {
      "epoch": 0.9276118948283215,
      "grad_norm": 0.15545450150966644,
      "learning_rate": 6.481775986369622e-06,
      "loss": 0.0019,
      "step": 144630
    },
    {
      "epoch": 0.9276760317221077,
      "grad_norm": 0.15038074553012848,
      "learning_rate": 6.481241416821252e-06,
      "loss": 0.0019,
      "step": 144640
    },
    {
      "epoch": 0.9277401686158938,
      "grad_norm": 0.04699162021279335,
      "learning_rate": 6.4807068287118705e-06,
      "loss": 0.0022,
      "step": 144650
    },
    {
      "epoch": 0.9278043055096798,
      "grad_norm": 0.1671448051929474,
      "learning_rate": 6.480172222048172e-06,
      "loss": 0.0021,
      "step": 144660
    },
    {
      "epoch": 0.927868442403466,
      "grad_norm": 0.06620363146066666,
      "learning_rate": 6.4796375968368594e-06,
      "loss": 0.0033,
      "step": 144670
    },
    {
      "epoch": 0.927932579297252,
      "grad_norm": 0.2066885381937027,
      "learning_rate": 6.479102953084629e-06,
      "loss": 0.0034,
      "step": 144680
    },
    {
      "epoch": 0.9279967161910382,
      "grad_norm": 0.0637272372841835,
      "learning_rate": 6.478568290798183e-06,
      "loss": 0.0018,
      "step": 144690
    },
    {
      "epoch": 0.9280608530848242,
      "grad_norm": 0.09442780911922455,
      "learning_rate": 6.4780336099842175e-06,
      "loss": 0.0035,
      "step": 144700
    },
    {
      "epoch": 0.9281249899786104,
      "grad_norm": 0.25165337324142456,
      "learning_rate": 6.477498910649437e-06,
      "loss": 0.002,
      "step": 144710
    },
    {
      "epoch": 0.9281891268723964,
      "grad_norm": 0.3011432886123657,
      "learning_rate": 6.4769641928005365e-06,
      "loss": 0.0053,
      "step": 144720
    },
    {
      "epoch": 0.9282532637661826,
      "grad_norm": 0.14117687940597534,
      "learning_rate": 6.476429456444222e-06,
      "loss": 0.0051,
      "step": 144730
    },
    {
      "epoch": 0.9283174006599686,
      "grad_norm": 0.09507281333208084,
      "learning_rate": 6.475894701587189e-06,
      "loss": 0.0023,
      "step": 144740
    },
    {
      "epoch": 0.9283815375537547,
      "grad_norm": 0.08003487437963486,
      "learning_rate": 6.475359928236141e-06,
      "loss": 0.0029,
      "step": 144750
    },
    {
      "epoch": 0.9284456744475408,
      "grad_norm": 0.13146911561489105,
      "learning_rate": 6.47482513639778e-06,
      "loss": 0.0018,
      "step": 144760
    },
    {
      "epoch": 0.9285098113413269,
      "grad_norm": 0.559633731842041,
      "learning_rate": 6.4742903260788036e-06,
      "loss": 0.0019,
      "step": 144770
    },
    {
      "epoch": 0.9285739482351131,
      "grad_norm": 0.1207059770822525,
      "learning_rate": 6.473755497285918e-06,
      "loss": 0.0013,
      "step": 144780
    },
    {
      "epoch": 0.9286380851288991,
      "grad_norm": 0.03859927877783775,
      "learning_rate": 6.473220650025822e-06,
      "loss": 0.002,
      "step": 144790
    },
    {
      "epoch": 0.9287022220226853,
      "grad_norm": 0.18498145043849945,
      "learning_rate": 6.472685784305218e-06,
      "loss": 0.0033,
      "step": 144800
    },
    {
      "epoch": 0.9287663589164713,
      "grad_norm": 0.07351011037826538,
      "learning_rate": 6.472150900130809e-06,
      "loss": 0.003,
      "step": 144810
    },
    {
      "epoch": 0.9288304958102575,
      "grad_norm": 0.08685048669576645,
      "learning_rate": 6.4716159975092975e-06,
      "loss": 0.002,
      "step": 144820
    },
    {
      "epoch": 0.9288946327040435,
      "grad_norm": 0.16433115303516388,
      "learning_rate": 6.471081076447385e-06,
      "loss": 0.0012,
      "step": 144830
    },
    {
      "epoch": 0.9289587695978296,
      "grad_norm": 0.14450232684612274,
      "learning_rate": 6.470546136951776e-06,
      "loss": 0.0029,
      "step": 144840
    },
    {
      "epoch": 0.9290229064916157,
      "grad_norm": 0.05751143768429756,
      "learning_rate": 6.470011179029172e-06,
      "loss": 0.0011,
      "step": 144850
    },
    {
      "epoch": 0.9290870433854018,
      "grad_norm": 0.0785200223326683,
      "learning_rate": 6.4694762026862774e-06,
      "loss": 0.0015,
      "step": 144860
    },
    {
      "epoch": 0.9291511802791879,
      "grad_norm": 0.09653989970684052,
      "learning_rate": 6.468941207929797e-06,
      "loss": 0.0026,
      "step": 144870
    },
    {
      "epoch": 0.929215317172974,
      "grad_norm": 0.12254025787115097,
      "learning_rate": 6.468406194766433e-06,
      "loss": 0.0021,
      "step": 144880
    },
    {
      "epoch": 0.92927945406676,
      "grad_norm": 0.11895779520273209,
      "learning_rate": 6.467871163202888e-06,
      "loss": 0.0014,
      "step": 144890
    },
    {
      "epoch": 0.9293435909605462,
      "grad_norm": 0.17474906146526337,
      "learning_rate": 6.46733611324587e-06,
      "loss": 0.0019,
      "step": 144900
    },
    {
      "epoch": 0.9294077278543322,
      "grad_norm": 0.02767779491841793,
      "learning_rate": 6.466801044902081e-06,
      "loss": 0.0012,
      "step": 144910
    },
    {
      "epoch": 0.9294718647481184,
      "grad_norm": 0.19321908056735992,
      "learning_rate": 6.466265958178227e-06,
      "loss": 0.0023,
      "step": 144920
    },
    {
      "epoch": 0.9295360016419045,
      "grad_norm": 0.21164913475513458,
      "learning_rate": 6.465730853081013e-06,
      "loss": 0.0042,
      "step": 144930
    },
    {
      "epoch": 0.9296001385356906,
      "grad_norm": 0.192432701587677,
      "learning_rate": 6.465195729617144e-06,
      "loss": 0.002,
      "step": 144940
    },
    {
      "epoch": 0.9296642754294767,
      "grad_norm": 0.09866433590650558,
      "learning_rate": 6.4646605877933255e-06,
      "loss": 0.0014,
      "step": 144950
    },
    {
      "epoch": 0.9297284123232628,
      "grad_norm": 0.15407894551753998,
      "learning_rate": 6.464125427616261e-06,
      "loss": 0.0017,
      "step": 144960
    },
    {
      "epoch": 0.9297925492170489,
      "grad_norm": 0.05903768539428711,
      "learning_rate": 6.46359024909266e-06,
      "loss": 0.0026,
      "step": 144970
    },
    {
      "epoch": 0.9298566861108349,
      "grad_norm": 0.1296510547399521,
      "learning_rate": 6.463055052229227e-06,
      "loss": 0.0027,
      "step": 144980
    },
    {
      "epoch": 0.9299208230046211,
      "grad_norm": 0.22783802449703217,
      "learning_rate": 6.4625198370326695e-06,
      "loss": 0.0033,
      "step": 144990
    },
    {
      "epoch": 0.9299849598984071,
      "grad_norm": 0.1219167411327362,
      "learning_rate": 6.461984603509692e-06,
      "loss": 0.0023,
      "step": 145000
    },
    {
      "epoch": 0.9300490967921933,
      "grad_norm": 0.06050629913806915,
      "learning_rate": 6.461449351667004e-06,
      "loss": 0.0018,
      "step": 145010
    },
    {
      "epoch": 0.9301132336859793,
      "grad_norm": 0.2846671938896179,
      "learning_rate": 6.460914081511309e-06,
      "loss": 0.0025,
      "step": 145020
    },
    {
      "epoch": 0.9301773705797655,
      "grad_norm": 0.10349977016448975,
      "learning_rate": 6.460378793049318e-06,
      "loss": 0.0019,
      "step": 145030
    },
    {
      "epoch": 0.9302415074735515,
      "grad_norm": 0.14169473946094513,
      "learning_rate": 6.459843486287735e-06,
      "loss": 0.002,
      "step": 145040
    },
    {
      "epoch": 0.9303056443673376,
      "grad_norm": 0.2114722728729248,
      "learning_rate": 6.459308161233273e-06,
      "loss": 0.0018,
      "step": 145050
    },
    {
      "epoch": 0.9303697812611238,
      "grad_norm": 0.13831239938735962,
      "learning_rate": 6.458772817892635e-06,
      "loss": 0.0019,
      "step": 145060
    },
    {
      "epoch": 0.9304339181549098,
      "grad_norm": 0.1724303513765335,
      "learning_rate": 6.458237456272532e-06,
      "loss": 0.0024,
      "step": 145070
    },
    {
      "epoch": 0.930498055048696,
      "grad_norm": 0.050732627511024475,
      "learning_rate": 6.45770207637967e-06,
      "loss": 0.0028,
      "step": 145080
    },
    {
      "epoch": 0.930562191942482,
      "grad_norm": 0.27567073702812195,
      "learning_rate": 6.457166678220761e-06,
      "loss": 0.0018,
      "step": 145090
    },
    {
      "epoch": 0.9306263288362682,
      "grad_norm": 0.01489699725061655,
      "learning_rate": 6.4566312618025094e-06,
      "loss": 0.0019,
      "step": 145100
    },
    {
      "epoch": 0.9306904657300542,
      "grad_norm": 0.04095577076077461,
      "learning_rate": 6.456095827131629e-06,
      "loss": 0.0019,
      "step": 145110
    },
    {
      "epoch": 0.9307546026238404,
      "grad_norm": 0.18893811106681824,
      "learning_rate": 6.455560374214826e-06,
      "loss": 0.0035,
      "step": 145120
    },
    {
      "epoch": 0.9308187395176264,
      "grad_norm": 0.1257791519165039,
      "learning_rate": 6.455024903058813e-06,
      "loss": 0.0019,
      "step": 145130
    },
    {
      "epoch": 0.9308828764114125,
      "grad_norm": 0.05663027986884117,
      "learning_rate": 6.454489413670297e-06,
      "loss": 0.0017,
      "step": 145140
    },
    {
      "epoch": 0.9309470133051986,
      "grad_norm": 0.10859649628400803,
      "learning_rate": 6.4539539060559896e-06,
      "loss": 0.0017,
      "step": 145150
    },
    {
      "epoch": 0.9310111501989847,
      "grad_norm": 0.04232072830200195,
      "learning_rate": 6.4534183802226e-06,
      "loss": 0.0026,
      "step": 145160
    },
    {
      "epoch": 0.9310752870927708,
      "grad_norm": 0.07066509127616882,
      "learning_rate": 6.452882836176839e-06,
      "loss": 0.0019,
      "step": 145170
    },
    {
      "epoch": 0.9311394239865569,
      "grad_norm": 0.267238587141037,
      "learning_rate": 6.45234727392542e-06,
      "loss": 0.0023,
      "step": 145180
    },
    {
      "epoch": 0.931203560880343,
      "grad_norm": 0.264752596616745,
      "learning_rate": 6.45181169347505e-06,
      "loss": 0.0022,
      "step": 145190
    },
    {
      "epoch": 0.9312676977741291,
      "grad_norm": 0.09979977458715439,
      "learning_rate": 6.451276094832441e-06,
      "loss": 0.0019,
      "step": 145200
    },
    {
      "epoch": 0.9313318346679152,
      "grad_norm": 0.02484172396361828,
      "learning_rate": 6.450740478004307e-06,
      "loss": 0.0016,
      "step": 145210
    },
    {
      "epoch": 0.9313959715617013,
      "grad_norm": 0.13323238492012024,
      "learning_rate": 6.450204842997358e-06,
      "loss": 0.0021,
      "step": 145220
    },
    {
      "epoch": 0.9314601084554874,
      "grad_norm": 0.12850530445575714,
      "learning_rate": 6.449669189818304e-06,
      "loss": 0.0015,
      "step": 145230
    },
    {
      "epoch": 0.9315242453492735,
      "grad_norm": 0.0978880450129509,
      "learning_rate": 6.44913351847386e-06,
      "loss": 0.0012,
      "step": 145240
    },
    {
      "epoch": 0.9315883822430596,
      "grad_norm": 0.05493517220020294,
      "learning_rate": 6.448597828970738e-06,
      "loss": 0.0021,
      "step": 145250
    },
    {
      "epoch": 0.9316525191368457,
      "grad_norm": 0.057626523077487946,
      "learning_rate": 6.448062121315648e-06,
      "loss": 0.0016,
      "step": 145260
    },
    {
      "epoch": 0.9317166560306318,
      "grad_norm": 0.19279751181602478,
      "learning_rate": 6.447526395515307e-06,
      "loss": 0.0026,
      "step": 145270
    },
    {
      "epoch": 0.9317807929244178,
      "grad_norm": 0.08471494168043137,
      "learning_rate": 6.446990651576425e-06,
      "loss": 0.0018,
      "step": 145280
    },
    {
      "epoch": 0.931844929818204,
      "grad_norm": 0.16393303871154785,
      "learning_rate": 6.446454889505715e-06,
      "loss": 0.0031,
      "step": 145290
    },
    {
      "epoch": 0.93190906671199,
      "grad_norm": 0.15938109159469604,
      "learning_rate": 6.445919109309893e-06,
      "loss": 0.0021,
      "step": 145300
    },
    {
      "epoch": 0.9319732036057762,
      "grad_norm": 0.08110189437866211,
      "learning_rate": 6.445383310995671e-06,
      "loss": 0.0022,
      "step": 145310
    },
    {
      "epoch": 0.9320373404995622,
      "grad_norm": 0.028752269223332405,
      "learning_rate": 6.444847494569761e-06,
      "loss": 0.0018,
      "step": 145320
    },
    {
      "epoch": 0.9321014773933484,
      "grad_norm": 0.45185551047325134,
      "learning_rate": 6.444311660038882e-06,
      "loss": 0.0033,
      "step": 145330
    },
    {
      "epoch": 0.9321656142871345,
      "grad_norm": 0.07926983386278152,
      "learning_rate": 6.443775807409745e-06,
      "loss": 0.0014,
      "step": 145340
    },
    {
      "epoch": 0.9322297511809206,
      "grad_norm": 0.07370294630527496,
      "learning_rate": 6.443239936689064e-06,
      "loss": 0.0017,
      "step": 145350
    },
    {
      "epoch": 0.9322938880747067,
      "grad_norm": 0.2275826334953308,
      "learning_rate": 6.442704047883555e-06,
      "loss": 0.0018,
      "step": 145360
    },
    {
      "epoch": 0.9323580249684927,
      "grad_norm": 0.05165311321616173,
      "learning_rate": 6.442168140999935e-06,
      "loss": 0.0016,
      "step": 145370
    },
    {
      "epoch": 0.9324221618622789,
      "grad_norm": 0.10057036578655243,
      "learning_rate": 6.441632216044915e-06,
      "loss": 0.0024,
      "step": 145380
    },
    {
      "epoch": 0.9324862987560649,
      "grad_norm": 0.10764387249946594,
      "learning_rate": 6.441096273025216e-06,
      "loss": 0.0012,
      "step": 145390
    },
    {
      "epoch": 0.9325504356498511,
      "grad_norm": 0.05163717269897461,
      "learning_rate": 6.440560311947549e-06,
      "loss": 0.0042,
      "step": 145400
    },
    {
      "epoch": 0.9326145725436371,
      "grad_norm": 0.04539763927459717,
      "learning_rate": 6.440024332818633e-06,
      "loss": 0.0014,
      "step": 145410
    },
    {
      "epoch": 0.9326787094374233,
      "grad_norm": 0.1427016705274582,
      "learning_rate": 6.439488335645181e-06,
      "loss": 0.0014,
      "step": 145420
    },
    {
      "epoch": 0.9327428463312093,
      "grad_norm": 0.2075621485710144,
      "learning_rate": 6.438952320433913e-06,
      "loss": 0.0024,
      "step": 145430
    },
    {
      "epoch": 0.9328069832249954,
      "grad_norm": 0.20911450684070587,
      "learning_rate": 6.4384162871915425e-06,
      "loss": 0.002,
      "step": 145440
    },
    {
      "epoch": 0.9328711201187815,
      "grad_norm": 0.09059228003025055,
      "learning_rate": 6.437880235924788e-06,
      "loss": 0.0034,
      "step": 145450
    },
    {
      "epoch": 0.9329352570125676,
      "grad_norm": 0.07383474707603455,
      "learning_rate": 6.437344166640369e-06,
      "loss": 0.0026,
      "step": 145460
    },
    {
      "epoch": 0.9329993939063537,
      "grad_norm": 0.0879674181342125,
      "learning_rate": 6.436808079344998e-06,
      "loss": 0.0034,
      "step": 145470
    },
    {
      "epoch": 0.9330635308001398,
      "grad_norm": 0.1926436424255371,
      "learning_rate": 6.436271974045396e-06,
      "loss": 0.0037,
      "step": 145480
    },
    {
      "epoch": 0.933127667693926,
      "grad_norm": 0.06891998648643494,
      "learning_rate": 6.435735850748279e-06,
      "loss": 0.0022,
      "step": 145490
    },
    {
      "epoch": 0.933191804587712,
      "grad_norm": 0.11858411133289337,
      "learning_rate": 6.435199709460366e-06,
      "loss": 0.0038,
      "step": 145500
    },
    {
      "epoch": 0.9332559414814982,
      "grad_norm": 0.08976802974939346,
      "learning_rate": 6.434663550188375e-06,
      "loss": 0.0024,
      "step": 145510
    },
    {
      "epoch": 0.9333200783752842,
      "grad_norm": 0.11608156561851501,
      "learning_rate": 6.434127372939024e-06,
      "loss": 0.0023,
      "step": 145520
    },
    {
      "epoch": 0.9333842152690703,
      "grad_norm": 0.0822378545999527,
      "learning_rate": 6.433591177719032e-06,
      "loss": 0.0024,
      "step": 145530
    },
    {
      "epoch": 0.9334483521628564,
      "grad_norm": 0.16821332275867462,
      "learning_rate": 6.433054964535119e-06,
      "loss": 0.0026,
      "step": 145540
    },
    {
      "epoch": 0.9335124890566425,
      "grad_norm": 0.04819389805197716,
      "learning_rate": 6.432518733394002e-06,
      "loss": 0.0037,
      "step": 145550
    },
    {
      "epoch": 0.9335766259504286,
      "grad_norm": 0.006087715271860361,
      "learning_rate": 6.431982484302402e-06,
      "loss": 0.0024,
      "step": 145560
    },
    {
      "epoch": 0.9336407628442147,
      "grad_norm": 0.08751381188631058,
      "learning_rate": 6.4314462172670375e-06,
      "loss": 0.0018,
      "step": 145570
    },
    {
      "epoch": 0.9337048997380007,
      "grad_norm": 0.060294367372989655,
      "learning_rate": 6.43090993229463e-06,
      "loss": 0.0021,
      "step": 145580
    },
    {
      "epoch": 0.9337690366317869,
      "grad_norm": 0.06259045004844666,
      "learning_rate": 6.430373629391897e-06,
      "loss": 0.0028,
      "step": 145590
    },
    {
      "epoch": 0.9338331735255729,
      "grad_norm": 0.16633011400699615,
      "learning_rate": 6.4298373085655606e-06,
      "loss": 0.0025,
      "step": 145600
    },
    {
      "epoch": 0.9338973104193591,
      "grad_norm": 0.31002166867256165,
      "learning_rate": 6.429300969822341e-06,
      "loss": 0.0018,
      "step": 145610
    },
    {
      "epoch": 0.9339614473131452,
      "grad_norm": 0.09039387851953506,
      "learning_rate": 6.428764613168958e-06,
      "loss": 0.0017,
      "step": 145620
    },
    {
      "epoch": 0.9340255842069313,
      "grad_norm": 0.11676201969385147,
      "learning_rate": 6.428228238612135e-06,
      "loss": 0.0027,
      "step": 145630
    },
    {
      "epoch": 0.9340897211007174,
      "grad_norm": 0.18348877131938934,
      "learning_rate": 6.4276918461585906e-06,
      "loss": 0.0013,
      "step": 145640
    },
    {
      "epoch": 0.9341538579945035,
      "grad_norm": 0.1378244161605835,
      "learning_rate": 6.427155435815047e-06,
      "loss": 0.0038,
      "step": 145650
    },
    {
      "epoch": 0.9342179948882896,
      "grad_norm": 0.027310635894536972,
      "learning_rate": 6.426619007588225e-06,
      "loss": 0.0014,
      "step": 145660
    },
    {
      "epoch": 0.9342821317820756,
      "grad_norm": 0.10352227091789246,
      "learning_rate": 6.426082561484848e-06,
      "loss": 0.0022,
      "step": 145670
    },
    {
      "epoch": 0.9343462686758618,
      "grad_norm": 0.11045675724744797,
      "learning_rate": 6.425546097511637e-06,
      "loss": 0.0018,
      "step": 145680
    },
    {
      "epoch": 0.9344104055696478,
      "grad_norm": 0.055172890424728394,
      "learning_rate": 6.425009615675316e-06,
      "loss": 0.0016,
      "step": 145690
    },
    {
      "epoch": 0.934474542463434,
      "grad_norm": 0.007064759731292725,
      "learning_rate": 6.424473115982603e-06,
      "loss": 0.0016,
      "step": 145700
    },
    {
      "epoch": 0.93453867935722,
      "grad_norm": 0.18204259872436523,
      "learning_rate": 6.423936598440228e-06,
      "loss": 0.0035,
      "step": 145710
    },
    {
      "epoch": 0.9346028162510062,
      "grad_norm": 0.07678377628326416,
      "learning_rate": 6.4234000630549065e-06,
      "loss": 0.0014,
      "step": 145720
    },
    {
      "epoch": 0.9346669531447922,
      "grad_norm": 0.1188676655292511,
      "learning_rate": 6.422863509833366e-06,
      "loss": 0.0014,
      "step": 145730
    },
    {
      "epoch": 0.9347310900385783,
      "grad_norm": 0.16479258239269257,
      "learning_rate": 6.422326938782328e-06,
      "loss": 0.0017,
      "step": 145740
    },
    {
      "epoch": 0.9347952269323644,
      "grad_norm": 0.09554725140333176,
      "learning_rate": 6.421790349908518e-06,
      "loss": 0.0026,
      "step": 145750
    },
    {
      "epoch": 0.9348593638261505,
      "grad_norm": 0.16004672646522522,
      "learning_rate": 6.421253743218658e-06,
      "loss": 0.0016,
      "step": 145760
    },
    {
      "epoch": 0.9349235007199367,
      "grad_norm": 0.0543198436498642,
      "learning_rate": 6.420717118719473e-06,
      "loss": 0.0055,
      "step": 145770
    },
    {
      "epoch": 0.9349876376137227,
      "grad_norm": 0.08797811716794968,
      "learning_rate": 6.420180476417688e-06,
      "loss": 0.0014,
      "step": 145780
    },
    {
      "epoch": 0.9350517745075089,
      "grad_norm": 0.005532593000680208,
      "learning_rate": 6.419643816320026e-06,
      "loss": 0.002,
      "step": 145790
    },
    {
      "epoch": 0.9351159114012949,
      "grad_norm": 0.14849628508090973,
      "learning_rate": 6.419107138433211e-06,
      "loss": 0.0023,
      "step": 145800
    },
    {
      "epoch": 0.9351800482950811,
      "grad_norm": 0.07060623168945312,
      "learning_rate": 6.41857044276397e-06,
      "loss": 0.0018,
      "step": 145810
    },
    {
      "epoch": 0.9352441851888671,
      "grad_norm": 0.031063968315720558,
      "learning_rate": 6.41803372931903e-06,
      "loss": 0.0018,
      "step": 145820
    },
    {
      "epoch": 0.9353083220826532,
      "grad_norm": 0.13810567557811737,
      "learning_rate": 6.417496998105112e-06,
      "loss": 0.0023,
      "step": 145830
    },
    {
      "epoch": 0.9353724589764393,
      "grad_norm": 0.17485027015209198,
      "learning_rate": 6.416960249128943e-06,
      "loss": 0.002,
      "step": 145840
    },
    {
      "epoch": 0.9354365958702254,
      "grad_norm": 0.14516933262348175,
      "learning_rate": 6.41642348239725e-06,
      "loss": 0.0024,
      "step": 145850
    },
    {
      "epoch": 0.9355007327640115,
      "grad_norm": 0.07958931475877762,
      "learning_rate": 6.415886697916759e-06,
      "loss": 0.0033,
      "step": 145860
    },
    {
      "epoch": 0.9355648696577976,
      "grad_norm": 0.03135542571544647,
      "learning_rate": 6.415349895694195e-06,
      "loss": 0.0021,
      "step": 145870
    },
    {
      "epoch": 0.9356290065515837,
      "grad_norm": 0.11623740941286087,
      "learning_rate": 6.414813075736286e-06,
      "loss": 0.002,
      "step": 145880
    },
    {
      "epoch": 0.9356931434453698,
      "grad_norm": 0.11329323798418045,
      "learning_rate": 6.4142762380497565e-06,
      "loss": 0.0043,
      "step": 145890
    },
    {
      "epoch": 0.9357572803391558,
      "grad_norm": 0.03688264265656471,
      "learning_rate": 6.413739382641338e-06,
      "loss": 0.0028,
      "step": 145900
    },
    {
      "epoch": 0.935821417232942,
      "grad_norm": 0.06331092119216919,
      "learning_rate": 6.413202509517752e-06,
      "loss": 0.0031,
      "step": 145910
    },
    {
      "epoch": 0.9358855541267281,
      "grad_norm": 0.1076519787311554,
      "learning_rate": 6.412665618685729e-06,
      "loss": 0.003,
      "step": 145920
    },
    {
      "epoch": 0.9359496910205142,
      "grad_norm": 0.06783204525709152,
      "learning_rate": 6.412128710151997e-06,
      "loss": 0.0017,
      "step": 145930
    },
    {
      "epoch": 0.9360138279143003,
      "grad_norm": 0.12745817005634308,
      "learning_rate": 6.411591783923282e-06,
      "loss": 0.0046,
      "step": 145940
    },
    {
      "epoch": 0.9360779648080864,
      "grad_norm": 0.26193997263908386,
      "learning_rate": 6.411054840006313e-06,
      "loss": 0.0052,
      "step": 145950
    },
    {
      "epoch": 0.9361421017018725,
      "grad_norm": 0.07764274626970291,
      "learning_rate": 6.410517878407819e-06,
      "loss": 0.0014,
      "step": 145960
    },
    {
      "epoch": 0.9362062385956585,
      "grad_norm": 0.06901423633098602,
      "learning_rate": 6.409980899134529e-06,
      "loss": 0.0017,
      "step": 145970
    },
    {
      "epoch": 0.9362703754894447,
      "grad_norm": 0.039977312088012695,
      "learning_rate": 6.409443902193169e-06,
      "loss": 0.0025,
      "step": 145980
    },
    {
      "epoch": 0.9363345123832307,
      "grad_norm": 0.14563529193401337,
      "learning_rate": 6.40890688759047e-06,
      "loss": 0.0029,
      "step": 145990
    },
    {
      "epoch": 0.9363986492770169,
      "grad_norm": 0.18063455820083618,
      "learning_rate": 6.408369855333161e-06,
      "loss": 0.0017,
      "step": 146000
    },
    {
      "epoch": 0.9364627861708029,
      "grad_norm": 0.04741799831390381,
      "learning_rate": 6.407832805427971e-06,
      "loss": 0.0018,
      "step": 146010
    },
    {
      "epoch": 0.9365269230645891,
      "grad_norm": 0.02267642319202423,
      "learning_rate": 6.407295737881629e-06,
      "loss": 0.0014,
      "step": 146020
    },
    {
      "epoch": 0.9365910599583751,
      "grad_norm": 0.014480840414762497,
      "learning_rate": 6.406758652700867e-06,
      "loss": 0.0023,
      "step": 146030
    },
    {
      "epoch": 0.9366551968521613,
      "grad_norm": 0.08123641461133957,
      "learning_rate": 6.406221549892413e-06,
      "loss": 0.0035,
      "step": 146040
    },
    {
      "epoch": 0.9367193337459474,
      "grad_norm": 0.08571852743625641,
      "learning_rate": 6.405684429463e-06,
      "loss": 0.0026,
      "step": 146050
    },
    {
      "epoch": 0.9367834706397334,
      "grad_norm": 0.11626120656728745,
      "learning_rate": 6.405147291419352e-06,
      "loss": 0.0026,
      "step": 146060
    },
    {
      "epoch": 0.9368476075335196,
      "grad_norm": 0.02037181705236435,
      "learning_rate": 6.404610135768208e-06,
      "loss": 0.0034,
      "step": 146070
    },
    {
      "epoch": 0.9369117444273056,
      "grad_norm": 0.24351300299167633,
      "learning_rate": 6.404072962516293e-06,
      "loss": 0.0036,
      "step": 146080
    },
    {
      "epoch": 0.9369758813210918,
      "grad_norm": 0.12650911509990692,
      "learning_rate": 6.403535771670342e-06,
      "loss": 0.0031,
      "step": 146090
    },
    {
      "epoch": 0.9370400182148778,
      "grad_norm": 0.11917576938867569,
      "learning_rate": 6.402998563237084e-06,
      "loss": 0.0019,
      "step": 146100
    },
    {
      "epoch": 0.937104155108664,
      "grad_norm": 0.052274610847234726,
      "learning_rate": 6.40246133722325e-06,
      "loss": 0.0018,
      "step": 146110
    },
    {
      "epoch": 0.93716829200245,
      "grad_norm": 0.14794179797172546,
      "learning_rate": 6.401924093635574e-06,
      "loss": 0.002,
      "step": 146120
    },
    {
      "epoch": 0.9372324288962361,
      "grad_norm": 0.14314155280590057,
      "learning_rate": 6.4013868324807885e-06,
      "loss": 0.0018,
      "step": 146130
    },
    {
      "epoch": 0.9372965657900222,
      "grad_norm": 0.13320375978946686,
      "learning_rate": 6.400849553765622e-06,
      "loss": 0.0035,
      "step": 146140
    },
    {
      "epoch": 0.9373607026838083,
      "grad_norm": 0.019696349278092384,
      "learning_rate": 6.400312257496812e-06,
      "loss": 0.001,
      "step": 146150
    },
    {
      "epoch": 0.9374248395775944,
      "grad_norm": 0.008471256121993065,
      "learning_rate": 6.399774943681088e-06,
      "loss": 0.0014,
      "step": 146160
    },
    {
      "epoch": 0.9374889764713805,
      "grad_norm": 0.052590277045965195,
      "learning_rate": 6.399237612325182e-06,
      "loss": 0.0022,
      "step": 146170
    },
    {
      "epoch": 0.9375531133651666,
      "grad_norm": 0.056718725711107254,
      "learning_rate": 6.39870026343583e-06,
      "loss": 0.0026,
      "step": 146180
    },
    {
      "epoch": 0.9376172502589527,
      "grad_norm": 0.1991223394870758,
      "learning_rate": 6.3981628970197625e-06,
      "loss": 0.003,
      "step": 146190
    },
    {
      "epoch": 0.9376813871527389,
      "grad_norm": 0.06018494814634323,
      "learning_rate": 6.397625513083717e-06,
      "loss": 0.0012,
      "step": 146200
    },
    {
      "epoch": 0.9377455240465249,
      "grad_norm": 0.14340412616729736,
      "learning_rate": 6.397088111634423e-06,
      "loss": 0.0016,
      "step": 146210
    },
    {
      "epoch": 0.937809660940311,
      "grad_norm": 0.05904007703065872,
      "learning_rate": 6.396550692678618e-06,
      "loss": 0.0037,
      "step": 146220
    },
    {
      "epoch": 0.9378737978340971,
      "grad_norm": 0.13746248185634613,
      "learning_rate": 6.396013256223034e-06,
      "loss": 0.0032,
      "step": 146230
    },
    {
      "epoch": 0.9379379347278832,
      "grad_norm": 0.05195393040776253,
      "learning_rate": 6.395475802274407e-06,
      "loss": 0.0034,
      "step": 146240
    },
    {
      "epoch": 0.9380020716216693,
      "grad_norm": 0.040026769042015076,
      "learning_rate": 6.394938330839468e-06,
      "loss": 0.0045,
      "step": 146250
    },
    {
      "epoch": 0.9380662085154554,
      "grad_norm": 0.11002197116613388,
      "learning_rate": 6.394400841924959e-06,
      "loss": 0.0012,
      "step": 146260
    },
    {
      "epoch": 0.9381303454092415,
      "grad_norm": 0.1482343077659607,
      "learning_rate": 6.393863335537608e-06,
      "loss": 0.0017,
      "step": 146270
    },
    {
      "epoch": 0.9381944823030276,
      "grad_norm": 0.0892859548330307,
      "learning_rate": 6.393325811684154e-06,
      "loss": 0.0018,
      "step": 146280
    },
    {
      "epoch": 0.9382586191968136,
      "grad_norm": 0.07135730981826782,
      "learning_rate": 6.392788270371332e-06,
      "loss": 0.0018,
      "step": 146290
    },
    {
      "epoch": 0.9383227560905998,
      "grad_norm": 0.12789849936962128,
      "learning_rate": 6.392250711605876e-06,
      "loss": 0.0033,
      "step": 146300
    },
    {
      "epoch": 0.9383868929843858,
      "grad_norm": 0.11650566756725311,
      "learning_rate": 6.391713135394526e-06,
      "loss": 0.0032,
      "step": 146310
    },
    {
      "epoch": 0.938451029878172,
      "grad_norm": 0.04727376624941826,
      "learning_rate": 6.391175541744014e-06,
      "loss": 0.0014,
      "step": 146320
    },
    {
      "epoch": 0.9385151667719581,
      "grad_norm": 0.03639853373169899,
      "learning_rate": 6.39063793066108e-06,
      "loss": 0.0013,
      "step": 146330
    },
    {
      "epoch": 0.9385793036657442,
      "grad_norm": 0.07490959763526917,
      "learning_rate": 6.390100302152456e-06,
      "loss": 0.0024,
      "step": 146340
    },
    {
      "epoch": 0.9386434405595303,
      "grad_norm": 0.05077169090509415,
      "learning_rate": 6.3895626562248845e-06,
      "loss": 0.0025,
      "step": 146350
    },
    {
      "epoch": 0.9387075774533163,
      "grad_norm": 0.04926026239991188,
      "learning_rate": 6.389024992885099e-06,
      "loss": 0.0024,
      "step": 146360
    },
    {
      "epoch": 0.9387717143471025,
      "grad_norm": 0.15370850265026093,
      "learning_rate": 6.388487312139837e-06,
      "loss": 0.0024,
      "step": 146370
    },
    {
      "epoch": 0.9388358512408885,
      "grad_norm": 0.0960734486579895,
      "learning_rate": 6.387949613995838e-06,
      "loss": 0.0016,
      "step": 146380
    },
    {
      "epoch": 0.9388999881346747,
      "grad_norm": 0.11249937862157822,
      "learning_rate": 6.387411898459836e-06,
      "loss": 0.0022,
      "step": 146390
    },
    {
      "epoch": 0.9389641250284607,
      "grad_norm": 0.16878405213356018,
      "learning_rate": 6.386874165538573e-06,
      "loss": 0.003,
      "step": 146400
    },
    {
      "epoch": 0.9390282619222469,
      "grad_norm": 0.21584518253803253,
      "learning_rate": 6.386336415238786e-06,
      "loss": 0.002,
      "step": 146410
    },
    {
      "epoch": 0.9390923988160329,
      "grad_norm": 0.22585085034370422,
      "learning_rate": 6.385798647567212e-06,
      "loss": 0.0031,
      "step": 146420
    },
    {
      "epoch": 0.939156535709819,
      "grad_norm": 0.07451896369457245,
      "learning_rate": 6.385260862530591e-06,
      "loss": 0.0027,
      "step": 146430
    },
    {
      "epoch": 0.9392206726036051,
      "grad_norm": 0.11135633289813995,
      "learning_rate": 6.38472306013566e-06,
      "loss": 0.0016,
      "step": 146440
    },
    {
      "epoch": 0.9392848094973912,
      "grad_norm": 0.29068341851234436,
      "learning_rate": 6.3841852403891604e-06,
      "loss": 0.0042,
      "step": 146450
    },
    {
      "epoch": 0.9393489463911773,
      "grad_norm": 0.03432978317141533,
      "learning_rate": 6.3836474032978315e-06,
      "loss": 0.0014,
      "step": 146460
    },
    {
      "epoch": 0.9394130832849634,
      "grad_norm": 0.0034370392095297575,
      "learning_rate": 6.383109548868411e-06,
      "loss": 0.0009,
      "step": 146470
    },
    {
      "epoch": 0.9394772201787496,
      "grad_norm": 0.14675849676132202,
      "learning_rate": 6.3825716771076386e-06,
      "loss": 0.0021,
      "step": 146480
    },
    {
      "epoch": 0.9395413570725356,
      "grad_norm": 0.028965268284082413,
      "learning_rate": 6.382033788022255e-06,
      "loss": 0.0021,
      "step": 146490
    },
    {
      "epoch": 0.9396054939663218,
      "grad_norm": 0.051502782851457596,
      "learning_rate": 6.381495881619001e-06,
      "loss": 0.0024,
      "step": 146500
    },
    {
      "epoch": 0.9396696308601078,
      "grad_norm": 0.03157828748226166,
      "learning_rate": 6.380957957904615e-06,
      "loss": 0.0016,
      "step": 146510
    },
    {
      "epoch": 0.939733767753894,
      "grad_norm": 0.18266567587852478,
      "learning_rate": 6.380420016885841e-06,
      "loss": 0.0024,
      "step": 146520
    },
    {
      "epoch": 0.93979790464768,
      "grad_norm": 0.16004875302314758,
      "learning_rate": 6.379882058569417e-06,
      "loss": 0.0034,
      "step": 146530
    },
    {
      "epoch": 0.9398620415414661,
      "grad_norm": 0.17029421031475067,
      "learning_rate": 6.379344082962084e-06,
      "loss": 0.0029,
      "step": 146540
    },
    {
      "epoch": 0.9399261784352522,
      "grad_norm": 0.13152047991752625,
      "learning_rate": 6.378806090070584e-06,
      "loss": 0.0015,
      "step": 146550
    },
    {
      "epoch": 0.9399903153290383,
      "grad_norm": 0.07515915483236313,
      "learning_rate": 6.3782680799016584e-06,
      "loss": 0.0032,
      "step": 146560
    },
    {
      "epoch": 0.9400544522228244,
      "grad_norm": 0.1369394063949585,
      "learning_rate": 6.377730052462048e-06,
      "loss": 0.0033,
      "step": 146570
    },
    {
      "epoch": 0.9401185891166105,
      "grad_norm": 0.020615696907043457,
      "learning_rate": 6.377192007758497e-06,
      "loss": 0.0026,
      "step": 146580
    },
    {
      "epoch": 0.9401827260103965,
      "grad_norm": 0.009614650160074234,
      "learning_rate": 6.376653945797744e-06,
      "loss": 0.0018,
      "step": 146590
    },
    {
      "epoch": 0.9402468629041827,
      "grad_norm": 0.1355360448360443,
      "learning_rate": 6.376115866586534e-06,
      "loss": 0.0015,
      "step": 146600
    },
    {
      "epoch": 0.9403109997979688,
      "grad_norm": 0.2663950026035309,
      "learning_rate": 6.3755777701316095e-06,
      "loss": 0.0022,
      "step": 146610
    },
    {
      "epoch": 0.9403751366917549,
      "grad_norm": 0.010578290559351444,
      "learning_rate": 6.375039656439712e-06,
      "loss": 0.0016,
      "step": 146620
    },
    {
      "epoch": 0.940439273585541,
      "grad_norm": 0.32353025674819946,
      "learning_rate": 6.374501525517585e-06,
      "loss": 0.002,
      "step": 146630
    },
    {
      "epoch": 0.9405034104793271,
      "grad_norm": 0.06088297441601753,
      "learning_rate": 6.373963377371971e-06,
      "loss": 0.0029,
      "step": 146640
    },
    {
      "epoch": 0.9405675473731132,
      "grad_norm": 0.1083764135837555,
      "learning_rate": 6.373425212009613e-06,
      "loss": 0.0023,
      "step": 146650
    },
    {
      "epoch": 0.9406316842668992,
      "grad_norm": 0.0899743065237999,
      "learning_rate": 6.372887029437256e-06,
      "loss": 0.0023,
      "step": 146660
    },
    {
      "epoch": 0.9406958211606854,
      "grad_norm": 0.11933674663305283,
      "learning_rate": 6.372348829661645e-06,
      "loss": 0.0032,
      "step": 146670
    },
    {
      "epoch": 0.9407599580544714,
      "grad_norm": 0.044247616082429886,
      "learning_rate": 6.371810612689521e-06,
      "loss": 0.0035,
      "step": 146680
    },
    {
      "epoch": 0.9408240949482576,
      "grad_norm": 0.06650304794311523,
      "learning_rate": 6.37127237852763e-06,
      "loss": 0.0022,
      "step": 146690
    },
    {
      "epoch": 0.9408882318420436,
      "grad_norm": 0.2597753703594208,
      "learning_rate": 6.3707341271827165e-06,
      "loss": 0.0021,
      "step": 146700
    },
    {
      "epoch": 0.9409523687358298,
      "grad_norm": 0.050154637545347214,
      "learning_rate": 6.370195858661523e-06,
      "loss": 0.0017,
      "step": 146710
    },
    {
      "epoch": 0.9410165056296158,
      "grad_norm": 0.0485159307718277,
      "learning_rate": 6.369657572970798e-06,
      "loss": 0.0019,
      "step": 146720
    },
    {
      "epoch": 0.941080642523402,
      "grad_norm": 0.06361440569162369,
      "learning_rate": 6.369119270117285e-06,
      "loss": 0.0017,
      "step": 146730
    },
    {
      "epoch": 0.941144779417188,
      "grad_norm": 0.24076053500175476,
      "learning_rate": 6.3685809501077265e-06,
      "loss": 0.002,
      "step": 146740
    },
    {
      "epoch": 0.9412089163109741,
      "grad_norm": 0.4782833158969879,
      "learning_rate": 6.368042612948872e-06,
      "loss": 0.0045,
      "step": 146750
    },
    {
      "epoch": 0.9412730532047603,
      "grad_norm": 0.09444776177406311,
      "learning_rate": 6.3675042586474665e-06,
      "loss": 0.0023,
      "step": 146760
    },
    {
      "epoch": 0.9413371900985463,
      "grad_norm": 0.11751112341880798,
      "learning_rate": 6.366965887210255e-06,
      "loss": 0.0016,
      "step": 146770
    },
    {
      "epoch": 0.9414013269923325,
      "grad_norm": 0.1627361923456192,
      "learning_rate": 6.366427498643983e-06,
      "loss": 0.0014,
      "step": 146780
    },
    {
      "epoch": 0.9414654638861185,
      "grad_norm": 0.050041310489177704,
      "learning_rate": 6.3658890929554e-06,
      "loss": 0.0029,
      "step": 146790
    },
    {
      "epoch": 0.9415296007799047,
      "grad_norm": 0.040631361305713654,
      "learning_rate": 6.365350670151249e-06,
      "loss": 0.0029,
      "step": 146800
    },
    {
      "epoch": 0.9415937376736907,
      "grad_norm": 0.10871771723031998,
      "learning_rate": 6.364812230238277e-06,
      "loss": 0.0017,
      "step": 146810
    },
    {
      "epoch": 0.9416578745674768,
      "grad_norm": 0.09430640190839767,
      "learning_rate": 6.364273773223235e-06,
      "loss": 0.0039,
      "step": 146820
    },
    {
      "epoch": 0.9417220114612629,
      "grad_norm": 0.0646919384598732,
      "learning_rate": 6.3637352991128654e-06,
      "loss": 0.0021,
      "step": 146830
    },
    {
      "epoch": 0.941786148355049,
      "grad_norm": 0.27819910645484924,
      "learning_rate": 6.363196807913919e-06,
      "loss": 0.0032,
      "step": 146840
    },
    {
      "epoch": 0.9418502852488351,
      "grad_norm": 0.29656410217285156,
      "learning_rate": 6.362658299633142e-06,
      "loss": 0.0032,
      "step": 146850
    },
    {
      "epoch": 0.9419144221426212,
      "grad_norm": 0.15691252052783966,
      "learning_rate": 6.362119774277284e-06,
      "loss": 0.0021,
      "step": 146860
    },
    {
      "epoch": 0.9419785590364073,
      "grad_norm": 0.02931254915893078,
      "learning_rate": 6.36158123185309e-06,
      "loss": 0.0026,
      "step": 146870
    },
    {
      "epoch": 0.9420426959301934,
      "grad_norm": 0.3193877637386322,
      "learning_rate": 6.361042672367311e-06,
      "loss": 0.0019,
      "step": 146880
    },
    {
      "epoch": 0.9421068328239796,
      "grad_norm": 0.10397673398256302,
      "learning_rate": 6.360504095826693e-06,
      "loss": 0.0013,
      "step": 146890
    },
    {
      "epoch": 0.9421709697177656,
      "grad_norm": 0.10111892223358154,
      "learning_rate": 6.359965502237988e-06,
      "loss": 0.0025,
      "step": 146900
    },
    {
      "epoch": 0.9422351066115517,
      "grad_norm": 0.07842773199081421,
      "learning_rate": 6.3594268916079425e-06,
      "loss": 0.0018,
      "step": 146910
    },
    {
      "epoch": 0.9422992435053378,
      "grad_norm": 0.016308825463056564,
      "learning_rate": 6.358888263943307e-06,
      "loss": 0.0015,
      "step": 146920
    },
    {
      "epoch": 0.9423633803991239,
      "grad_norm": 0.12908826768398285,
      "learning_rate": 6.35834961925083e-06,
      "loss": 0.0022,
      "step": 146930
    },
    {
      "epoch": 0.94242751729291,
      "grad_norm": 0.058444876223802567,
      "learning_rate": 6.357810957537261e-06,
      "loss": 0.0016,
      "step": 146940
    },
    {
      "epoch": 0.9424916541866961,
      "grad_norm": 0.20450295507907867,
      "learning_rate": 6.357272278809351e-06,
      "loss": 0.0025,
      "step": 146950
    },
    {
      "epoch": 0.9425557910804822,
      "grad_norm": 0.0351119339466095,
      "learning_rate": 6.3567335830738494e-06,
      "loss": 0.0021,
      "step": 146960
    },
    {
      "epoch": 0.9426199279742683,
      "grad_norm": 0.003150160191580653,
      "learning_rate": 6.356194870337507e-06,
      "loss": 0.0026,
      "step": 146970
    },
    {
      "epoch": 0.9426840648680543,
      "grad_norm": 0.08454978466033936,
      "learning_rate": 6.3556561406070724e-06,
      "loss": 0.0029,
      "step": 146980
    },
    {
      "epoch": 0.9427482017618405,
      "grad_norm": 0.004040864296257496,
      "learning_rate": 6.3551173938892984e-06,
      "loss": 0.0012,
      "step": 146990
    },
    {
      "epoch": 0.9428123386556265,
      "grad_norm": 0.08898429572582245,
      "learning_rate": 6.3545786301909355e-06,
      "loss": 0.0012,
      "step": 147000
    },
    {
      "epoch": 0.9428764755494127,
      "grad_norm": 0.1371237337589264,
      "learning_rate": 6.354039849518732e-06,
      "loss": 0.002,
      "step": 147010
    },
    {
      "epoch": 0.9429406124431987,
      "grad_norm": 0.07550517469644547,
      "learning_rate": 6.353501051879441e-06,
      "loss": 0.0018,
      "step": 147020
    },
    {
      "epoch": 0.9430047493369849,
      "grad_norm": 0.137441948056221,
      "learning_rate": 6.352962237279818e-06,
      "loss": 0.0031,
      "step": 147030
    },
    {
      "epoch": 0.943068886230771,
      "grad_norm": 0.1083943098783493,
      "learning_rate": 6.352423405726609e-06,
      "loss": 0.0028,
      "step": 147040
    },
    {
      "epoch": 0.943133023124557,
      "grad_norm": 0.056081295013427734,
      "learning_rate": 6.3518845572265685e-06,
      "loss": 0.0015,
      "step": 147050
    },
    {
      "epoch": 0.9431971600183432,
      "grad_norm": 0.37494710087776184,
      "learning_rate": 6.351345691786448e-06,
      "loss": 0.0016,
      "step": 147060
    },
    {
      "epoch": 0.9432612969121292,
      "grad_norm": 0.03048224374651909,
      "learning_rate": 6.350806809413001e-06,
      "loss": 0.0014,
      "step": 147070
    },
    {
      "epoch": 0.9433254338059154,
      "grad_norm": 0.09839826822280884,
      "learning_rate": 6.35026791011298e-06,
      "loss": 0.0033,
      "step": 147080
    },
    {
      "epoch": 0.9433895706997014,
      "grad_norm": 0.08801209181547165,
      "learning_rate": 6.349728993893135e-06,
      "loss": 0.003,
      "step": 147090
    },
    {
      "epoch": 0.9434537075934876,
      "grad_norm": 0.030154164880514145,
      "learning_rate": 6.349190060760222e-06,
      "loss": 0.0021,
      "step": 147100
    },
    {
      "epoch": 0.9435178444872736,
      "grad_norm": 0.14104928076267242,
      "learning_rate": 6.348651110720993e-06,
      "loss": 0.003,
      "step": 147110
    },
    {
      "epoch": 0.9435819813810598,
      "grad_norm": 0.02902240678668022,
      "learning_rate": 6.348112143782203e-06,
      "loss": 0.0028,
      "step": 147120
    },
    {
      "epoch": 0.9436461182748458,
      "grad_norm": 0.007148304954171181,
      "learning_rate": 6.347573159950603e-06,
      "loss": 0.0017,
      "step": 147130
    },
    {
      "epoch": 0.9437102551686319,
      "grad_norm": 0.12616923451423645,
      "learning_rate": 6.347034159232948e-06,
      "loss": 0.0014,
      "step": 147140
    },
    {
      "epoch": 0.943774392062418,
      "grad_norm": 0.05462987348437309,
      "learning_rate": 6.346495141635992e-06,
      "loss": 0.0015,
      "step": 147150
    },
    {
      "epoch": 0.9438385289562041,
      "grad_norm": 0.12261845171451569,
      "learning_rate": 6.345956107166491e-06,
      "loss": 0.0016,
      "step": 147160
    },
    {
      "epoch": 0.9439026658499903,
      "grad_norm": 0.18668530881404877,
      "learning_rate": 6.345417055831198e-06,
      "loss": 0.0018,
      "step": 147170
    },
    {
      "epoch": 0.9439668027437763,
      "grad_norm": 0.07512176036834717,
      "learning_rate": 6.344877987636867e-06,
      "loss": 0.0014,
      "step": 147180
    },
    {
      "epoch": 0.9440309396375625,
      "grad_norm": 0.17846761643886566,
      "learning_rate": 6.3443389025902535e-06,
      "loss": 0.0021,
      "step": 147190
    },
    {
      "epoch": 0.9440950765313485,
      "grad_norm": 0.03234937787055969,
      "learning_rate": 6.343799800698114e-06,
      "loss": 0.0016,
      "step": 147200
    },
    {
      "epoch": 0.9441592134251346,
      "grad_norm": 0.095039963722229,
      "learning_rate": 6.3432606819672006e-06,
      "loss": 0.0044,
      "step": 147210
    },
    {
      "epoch": 0.9442233503189207,
      "grad_norm": 0.09685095399618149,
      "learning_rate": 6.342721546404272e-06,
      "loss": 0.0017,
      "step": 147220
    },
    {
      "epoch": 0.9442874872127068,
      "grad_norm": 0.12833555042743683,
      "learning_rate": 6.342182394016083e-06,
      "loss": 0.0024,
      "step": 147230
    },
    {
      "epoch": 0.9443516241064929,
      "grad_norm": 0.06482169032096863,
      "learning_rate": 6.34164322480939e-06,
      "loss": 0.0021,
      "step": 147240
    },
    {
      "epoch": 0.944415761000279,
      "grad_norm": 0.10324189066886902,
      "learning_rate": 6.3411040387909464e-06,
      "loss": 0.003,
      "step": 147250
    },
    {
      "epoch": 0.9444798978940651,
      "grad_norm": 0.046588193625211716,
      "learning_rate": 6.3405648359675136e-06,
      "loss": 0.002,
      "step": 147260
    },
    {
      "epoch": 0.9445440347878512,
      "grad_norm": 0.30959293246269226,
      "learning_rate": 6.340025616345842e-06,
      "loss": 0.0023,
      "step": 147270
    },
    {
      "epoch": 0.9446081716816372,
      "grad_norm": 0.02668837085366249,
      "learning_rate": 6.339486379932693e-06,
      "loss": 0.001,
      "step": 147280
    },
    {
      "epoch": 0.9446723085754234,
      "grad_norm": 0.187612384557724,
      "learning_rate": 6.338947126734823e-06,
      "loss": 0.003,
      "step": 147290
    },
    {
      "epoch": 0.9447364454692094,
      "grad_norm": 0.14825215935707092,
      "learning_rate": 6.338407856758988e-06,
      "loss": 0.0027,
      "step": 147300
    },
    {
      "epoch": 0.9448005823629956,
      "grad_norm": 0.1354726254940033,
      "learning_rate": 6.337868570011946e-06,
      "loss": 0.0034,
      "step": 147310
    },
    {
      "epoch": 0.9448647192567817,
      "grad_norm": 0.04497535526752472,
      "learning_rate": 6.337329266500456e-06,
      "loss": 0.0025,
      "step": 147320
    },
    {
      "epoch": 0.9449288561505678,
      "grad_norm": 0.1626058667898178,
      "learning_rate": 6.336789946231272e-06,
      "loss": 0.0012,
      "step": 147330
    },
    {
      "epoch": 0.9449929930443539,
      "grad_norm": 0.5120843052864075,
      "learning_rate": 6.3362506092111565e-06,
      "loss": 0.0033,
      "step": 147340
    },
    {
      "epoch": 0.94505712993814,
      "grad_norm": 0.08398311585187912,
      "learning_rate": 6.335711255446866e-06,
      "loss": 0.0026,
      "step": 147350
    },
    {
      "epoch": 0.9451212668319261,
      "grad_norm": 0.24006327986717224,
      "learning_rate": 6.3351718849451575e-06,
      "loss": 0.0021,
      "step": 147360
    },
    {
      "epoch": 0.9451854037257121,
      "grad_norm": 0.038649216294288635,
      "learning_rate": 6.334632497712792e-06,
      "loss": 0.0022,
      "step": 147370
    },
    {
      "epoch": 0.9452495406194983,
      "grad_norm": 0.2920440435409546,
      "learning_rate": 6.334093093756527e-06,
      "loss": 0.0036,
      "step": 147380
    },
    {
      "epoch": 0.9453136775132843,
      "grad_norm": 0.36874157190322876,
      "learning_rate": 6.3335536730831225e-06,
      "loss": 0.0045,
      "step": 147390
    },
    {
      "epoch": 0.9453778144070705,
      "grad_norm": 0.13857769966125488,
      "learning_rate": 6.333014235699338e-06,
      "loss": 0.0016,
      "step": 147400
    },
    {
      "epoch": 0.9454419513008565,
      "grad_norm": 0.11178428679704666,
      "learning_rate": 6.332474781611931e-06,
      "loss": 0.0021,
      "step": 147410
    },
    {
      "epoch": 0.9455060881946427,
      "grad_norm": 0.349579781293869,
      "learning_rate": 6.331935310827664e-06,
      "loss": 0.0031,
      "step": 147420
    },
    {
      "epoch": 0.9455702250884287,
      "grad_norm": 0.03589482605457306,
      "learning_rate": 6.331395823353295e-06,
      "loss": 0.0008,
      "step": 147430
    },
    {
      "epoch": 0.9456343619822148,
      "grad_norm": 0.011520350351929665,
      "learning_rate": 6.3308563191955844e-06,
      "loss": 0.0012,
      "step": 147440
    },
    {
      "epoch": 0.9456984988760009,
      "grad_norm": 0.33133986592292786,
      "learning_rate": 6.330316798361294e-06,
      "loss": 0.0037,
      "step": 147450
    },
    {
      "epoch": 0.945762635769787,
      "grad_norm": 0.08037800341844559,
      "learning_rate": 6.329777260857181e-06,
      "loss": 0.0018,
      "step": 147460
    },
    {
      "epoch": 0.9458267726635732,
      "grad_norm": 0.05772693455219269,
      "learning_rate": 6.329237706690011e-06,
      "loss": 0.0024,
      "step": 147470
    },
    {
      "epoch": 0.9458909095573592,
      "grad_norm": 0.07055383920669556,
      "learning_rate": 6.328698135866542e-06,
      "loss": 0.002,
      "step": 147480
    },
    {
      "epoch": 0.9459550464511454,
      "grad_norm": 0.0548737533390522,
      "learning_rate": 6.3281585483935355e-06,
      "loss": 0.002,
      "step": 147490
    },
    {
      "epoch": 0.9460191833449314,
      "grad_norm": 0.1246550902724266,
      "learning_rate": 6.3276189442777534e-06,
      "loss": 0.0016,
      "step": 147500
    },
    {
      "epoch": 0.9460833202387176,
      "grad_norm": 0.12380314618349075,
      "learning_rate": 6.327079323525956e-06,
      "loss": 0.0021,
      "step": 147510
    },
    {
      "epoch": 0.9461474571325036,
      "grad_norm": 0.04626064375042915,
      "learning_rate": 6.326539686144908e-06,
      "loss": 0.0021,
      "step": 147520
    },
    {
      "epoch": 0.9462115940262897,
      "grad_norm": 0.1103246882557869,
      "learning_rate": 6.326000032141368e-06,
      "loss": 0.0027,
      "step": 147530
    },
    {
      "epoch": 0.9462757309200758,
      "grad_norm": 0.08117210865020752,
      "learning_rate": 6.325460361522102e-06,
      "loss": 0.002,
      "step": 147540
    },
    {
      "epoch": 0.9463398678138619,
      "grad_norm": 0.09314190596342087,
      "learning_rate": 6.324920674293868e-06,
      "loss": 0.0017,
      "step": 147550
    },
    {
      "epoch": 0.946404004707648,
      "grad_norm": 0.03479532524943352,
      "learning_rate": 6.324380970463433e-06,
      "loss": 0.0023,
      "step": 147560
    },
    {
      "epoch": 0.9464681416014341,
      "grad_norm": 0.12663571536540985,
      "learning_rate": 6.3238412500375564e-06,
      "loss": 0.0016,
      "step": 147570
    },
    {
      "epoch": 0.9465322784952201,
      "grad_norm": 0.04988948628306389,
      "learning_rate": 6.323301513023004e-06,
      "loss": 0.004,
      "step": 147580
    },
    {
      "epoch": 0.9465964153890063,
      "grad_norm": 0.010820391587913036,
      "learning_rate": 6.322761759426537e-06,
      "loss": 0.0021,
      "step": 147590
    },
    {
      "epoch": 0.9466605522827924,
      "grad_norm": 0.10937017947435379,
      "learning_rate": 6.322221989254921e-06,
      "loss": 0.0019,
      "step": 147600
    },
    {
      "epoch": 0.9467246891765785,
      "grad_norm": 0.08107168227434158,
      "learning_rate": 6.321682202514917e-06,
      "loss": 0.0025,
      "step": 147610
    },
    {
      "epoch": 0.9467888260703646,
      "grad_norm": 0.03284059092402458,
      "learning_rate": 6.321142399213291e-06,
      "loss": 0.0027,
      "step": 147620
    },
    {
      "epoch": 0.9468529629641507,
      "grad_norm": 0.04868131875991821,
      "learning_rate": 6.320602579356807e-06,
      "loss": 0.001,
      "step": 147630
    },
    {
      "epoch": 0.9469170998579368,
      "grad_norm": 0.19157037138938904,
      "learning_rate": 6.320062742952229e-06,
      "loss": 0.0033,
      "step": 147640
    },
    {
      "epoch": 0.9469812367517229,
      "grad_norm": 0.1776953786611557,
      "learning_rate": 6.319522890006319e-06,
      "loss": 0.0016,
      "step": 147650
    },
    {
      "epoch": 0.947045373645509,
      "grad_norm": 0.2312333732843399,
      "learning_rate": 6.318983020525846e-06,
      "loss": 0.0038,
      "step": 147660
    },
    {
      "epoch": 0.947109510539295,
      "grad_norm": 0.036973197013139725,
      "learning_rate": 6.318443134517573e-06,
      "loss": 0.003,
      "step": 147670
    },
    {
      "epoch": 0.9471736474330812,
      "grad_norm": 0.20453284680843353,
      "learning_rate": 6.317903231988266e-06,
      "loss": 0.0031,
      "step": 147680
    },
    {
      "epoch": 0.9472377843268672,
      "grad_norm": 0.12957876920700073,
      "learning_rate": 6.317363312944689e-06,
      "loss": 0.0033,
      "step": 147690
    },
    {
      "epoch": 0.9473019212206534,
      "grad_norm": 0.07207232713699341,
      "learning_rate": 6.316823377393608e-06,
      "loss": 0.0014,
      "step": 147700
    },
    {
      "epoch": 0.9473660581144394,
      "grad_norm": 0.05391666665673256,
      "learning_rate": 6.316283425341789e-06,
      "loss": 0.0038,
      "step": 147710
    },
    {
      "epoch": 0.9474301950082256,
      "grad_norm": 0.06373003125190735,
      "learning_rate": 6.315743456795997e-06,
      "loss": 0.002,
      "step": 147720
    },
    {
      "epoch": 0.9474943319020116,
      "grad_norm": 0.03036843053996563,
      "learning_rate": 6.315203471763001e-06,
      "loss": 0.003,
      "step": 147730
    },
    {
      "epoch": 0.9475584687957977,
      "grad_norm": 0.041159600019454956,
      "learning_rate": 6.314663470249564e-06,
      "loss": 0.0029,
      "step": 147740
    },
    {
      "epoch": 0.9476226056895839,
      "grad_norm": 0.04843321442604065,
      "learning_rate": 6.314123452262455e-06,
      "loss": 0.0021,
      "step": 147750
    },
    {
      "epoch": 0.9476867425833699,
      "grad_norm": 0.15022115409374237,
      "learning_rate": 6.31358341780844e-06,
      "loss": 0.0019,
      "step": 147760
    },
    {
      "epoch": 0.9477508794771561,
      "grad_norm": 0.009687594138085842,
      "learning_rate": 6.313043366894287e-06,
      "loss": 0.0024,
      "step": 147770
    },
    {
      "epoch": 0.9478150163709421,
      "grad_norm": 0.08602087944746017,
      "learning_rate": 6.31250329952676e-06,
      "loss": 0.0022,
      "step": 147780
    },
    {
      "epoch": 0.9478791532647283,
      "grad_norm": 0.24058887362480164,
      "learning_rate": 6.31196321571263e-06,
      "loss": 0.0055,
      "step": 147790
    },
    {
      "epoch": 0.9479432901585143,
      "grad_norm": 0.04920877516269684,
      "learning_rate": 6.311423115458663e-06,
      "loss": 0.0022,
      "step": 147800
    },
    {
      "epoch": 0.9480074270523005,
      "grad_norm": 0.10023945569992065,
      "learning_rate": 6.310882998771627e-06,
      "loss": 0.002,
      "step": 147810
    },
    {
      "epoch": 0.9480715639460865,
      "grad_norm": 0.19308824837207794,
      "learning_rate": 6.31034286565829e-06,
      "loss": 0.0022,
      "step": 147820
    },
    {
      "epoch": 0.9481357008398726,
      "grad_norm": 0.05049790441989899,
      "learning_rate": 6.30980271612542e-06,
      "loss": 0.0013,
      "step": 147830
    },
    {
      "epoch": 0.9481998377336587,
      "grad_norm": 0.1057187020778656,
      "learning_rate": 6.309262550179787e-06,
      "loss": 0.0028,
      "step": 147840
    },
    {
      "epoch": 0.9482639746274448,
      "grad_norm": 0.14519499242305756,
      "learning_rate": 6.308722367828158e-06,
      "loss": 0.0029,
      "step": 147850
    },
    {
      "epoch": 0.9483281115212309,
      "grad_norm": 0.047008804976940155,
      "learning_rate": 6.308182169077301e-06,
      "loss": 0.0027,
      "step": 147860
    },
    {
      "epoch": 0.948392248415017,
      "grad_norm": 0.06405046582221985,
      "learning_rate": 6.307641953933988e-06,
      "loss": 0.002,
      "step": 147870
    },
    {
      "epoch": 0.9484563853088032,
      "grad_norm": 0.06086205318570137,
      "learning_rate": 6.307101722404987e-06,
      "loss": 0.0009,
      "step": 147880
    },
    {
      "epoch": 0.9485205222025892,
      "grad_norm": 0.26466992497444153,
      "learning_rate": 6.306561474497066e-06,
      "loss": 0.0038,
      "step": 147890
    },
    {
      "epoch": 0.9485846590963753,
      "grad_norm": 0.045898932963609695,
      "learning_rate": 6.306021210216998e-06,
      "loss": 0.0018,
      "step": 147900
    },
    {
      "epoch": 0.9486487959901614,
      "grad_norm": 0.0372375026345253,
      "learning_rate": 6.305480929571549e-06,
      "loss": 0.0025,
      "step": 147910
    },
    {
      "epoch": 0.9487129328839475,
      "grad_norm": 0.009371262043714523,
      "learning_rate": 6.304940632567492e-06,
      "loss": 0.0012,
      "step": 147920
    },
    {
      "epoch": 0.9487770697777336,
      "grad_norm": 0.19069337844848633,
      "learning_rate": 6.304400319211596e-06,
      "loss": 0.003,
      "step": 147930
    },
    {
      "epoch": 0.9488412066715197,
      "grad_norm": 0.027903838083148003,
      "learning_rate": 6.303859989510631e-06,
      "loss": 0.0022,
      "step": 147940
    },
    {
      "epoch": 0.9489053435653058,
      "grad_norm": 0.007853977382183075,
      "learning_rate": 6.303319643471368e-06,
      "loss": 0.0013,
      "step": 147950
    },
    {
      "epoch": 0.9489694804590919,
      "grad_norm": 0.101915642619133,
      "learning_rate": 6.302779281100581e-06,
      "loss": 0.0025,
      "step": 147960
    },
    {
      "epoch": 0.949033617352878,
      "grad_norm": 0.26874613761901855,
      "learning_rate": 6.302238902405035e-06,
      "loss": 0.0069,
      "step": 147970
    },
    {
      "epoch": 0.9490977542466641,
      "grad_norm": 0.23425744473934174,
      "learning_rate": 6.301698507391508e-06,
      "loss": 0.0038,
      "step": 147980
    },
    {
      "epoch": 0.9491618911404501,
      "grad_norm": 0.08065503090620041,
      "learning_rate": 6.301158096066767e-06,
      "loss": 0.0014,
      "step": 147990
    },
    {
      "epoch": 0.9492260280342363,
      "grad_norm": 0.010573102161288261,
      "learning_rate": 6.300617668437585e-06,
      "loss": 0.0026,
      "step": 148000
    },
    {
      "epoch": 0.9492901649280223,
      "grad_norm": 0.166752889752388,
      "learning_rate": 6.300077224510734e-06,
      "loss": 0.0016,
      "step": 148010
    },
    {
      "epoch": 0.9493543018218085,
      "grad_norm": 0.02941727079451084,
      "learning_rate": 6.299536764292987e-06,
      "loss": 0.0039,
      "step": 148020
    },
    {
      "epoch": 0.9494184387155946,
      "grad_norm": 0.012016337364912033,
      "learning_rate": 6.298996287791116e-06,
      "loss": 0.0021,
      "step": 148030
    },
    {
      "epoch": 0.9494825756093807,
      "grad_norm": 0.2681630849838257,
      "learning_rate": 6.298455795011892e-06,
      "loss": 0.0024,
      "step": 148040
    },
    {
      "epoch": 0.9495467125031668,
      "grad_norm": 0.06486742198467255,
      "learning_rate": 6.2979152859620886e-06,
      "loss": 0.0027,
      "step": 148050
    },
    {
      "epoch": 0.9496108493969528,
      "grad_norm": 0.2404012829065323,
      "learning_rate": 6.297374760648479e-06,
      "loss": 0.0018,
      "step": 148060
    },
    {
      "epoch": 0.949674986290739,
      "grad_norm": 0.21434836089611053,
      "learning_rate": 6.296834219077837e-06,
      "loss": 0.003,
      "step": 148070
    },
    {
      "epoch": 0.949739123184525,
      "grad_norm": 0.17221471667289734,
      "learning_rate": 6.2962936612569355e-06,
      "loss": 0.0039,
      "step": 148080
    },
    {
      "epoch": 0.9498032600783112,
      "grad_norm": 0.0725397989153862,
      "learning_rate": 6.295753087192549e-06,
      "loss": 0.0011,
      "step": 148090
    },
    {
      "epoch": 0.9498673969720972,
      "grad_norm": 0.09452808648347855,
      "learning_rate": 6.295212496891449e-06,
      "loss": 0.0023,
      "step": 148100
    },
    {
      "epoch": 0.9499315338658834,
      "grad_norm": 0.06127987802028656,
      "learning_rate": 6.294671890360411e-06,
      "loss": 0.0034,
      "step": 148110
    },
    {
      "epoch": 0.9499956707596694,
      "grad_norm": 0.09638182073831558,
      "learning_rate": 6.294131267606208e-06,
      "loss": 0.002,
      "step": 148120
    },
    {
      "epoch": 0.9500598076534555,
      "grad_norm": 0.0705321803689003,
      "learning_rate": 6.293590628635617e-06,
      "loss": 0.0031,
      "step": 148130
    },
    {
      "epoch": 0.9501239445472416,
      "grad_norm": 0.1473604440689087,
      "learning_rate": 6.2930499734554094e-06,
      "loss": 0.0013,
      "step": 148140
    },
    {
      "epoch": 0.9501880814410277,
      "grad_norm": 0.21525196731090546,
      "learning_rate": 6.292509302072364e-06,
      "loss": 0.005,
      "step": 148150
    },
    {
      "epoch": 0.9502522183348139,
      "grad_norm": 0.03943829610943794,
      "learning_rate": 6.29196861449325e-06,
      "loss": 0.0018,
      "step": 148160
    },
    {
      "epoch": 0.9503163552285999,
      "grad_norm": 0.2596535086631775,
      "learning_rate": 6.291427910724849e-06,
      "loss": 0.0011,
      "step": 148170
    },
    {
      "epoch": 0.9503804921223861,
      "grad_norm": 0.1013018786907196,
      "learning_rate": 6.290887190773931e-06,
      "loss": 0.0029,
      "step": 148180
    },
    {
      "epoch": 0.9504446290161721,
      "grad_norm": 0.5317591428756714,
      "learning_rate": 6.2903464546472745e-06,
      "loss": 0.0021,
      "step": 148190
    },
    {
      "epoch": 0.9505087659099583,
      "grad_norm": 0.15050390362739563,
      "learning_rate": 6.289805702351654e-06,
      "loss": 0.0039,
      "step": 148200
    },
    {
      "epoch": 0.9505729028037443,
      "grad_norm": 0.08444549143314362,
      "learning_rate": 6.2892649338938474e-06,
      "loss": 0.0019,
      "step": 148210
    },
    {
      "epoch": 0.9506370396975304,
      "grad_norm": 0.15187714993953705,
      "learning_rate": 6.2887241492806295e-06,
      "loss": 0.0028,
      "step": 148220
    },
    {
      "epoch": 0.9507011765913165,
      "grad_norm": 0.04425651952624321,
      "learning_rate": 6.288183348518777e-06,
      "loss": 0.0016,
      "step": 148230
    },
    {
      "epoch": 0.9507653134851026,
      "grad_norm": 0.20383568108081818,
      "learning_rate": 6.287642531615067e-06,
      "loss": 0.0014,
      "step": 148240
    },
    {
      "epoch": 0.9508294503788887,
      "grad_norm": 0.19574086368083954,
      "learning_rate": 6.287101698576274e-06,
      "loss": 0.0028,
      "step": 148250
    },
    {
      "epoch": 0.9508935872726748,
      "grad_norm": 0.03341750428080559,
      "learning_rate": 6.28656084940918e-06,
      "loss": 0.0011,
      "step": 148260
    },
    {
      "epoch": 0.9509577241664608,
      "grad_norm": 0.17480899393558502,
      "learning_rate": 6.286019984120556e-06,
      "loss": 0.0022,
      "step": 148270
    },
    {
      "epoch": 0.951021861060247,
      "grad_norm": 0.14411622285842896,
      "learning_rate": 6.2854791027171845e-06,
      "loss": 0.0015,
      "step": 148280
    },
    {
      "epoch": 0.951085997954033,
      "grad_norm": 0.048610031604766846,
      "learning_rate": 6.284938205205839e-06,
      "loss": 0.0019,
      "step": 148290
    },
    {
      "epoch": 0.9511501348478192,
      "grad_norm": 0.1088913083076477,
      "learning_rate": 6.2843972915933025e-06,
      "loss": 0.0037,
      "step": 148300
    },
    {
      "epoch": 0.9512142717416053,
      "grad_norm": 0.020047571510076523,
      "learning_rate": 6.283856361886347e-06,
      "loss": 0.0016,
      "step": 148310
    },
    {
      "epoch": 0.9512784086353914,
      "grad_norm": 0.1922512650489807,
      "learning_rate": 6.283315416091755e-06,
      "loss": 0.0017,
      "step": 148320
    },
    {
      "epoch": 0.9513425455291775,
      "grad_norm": 0.2446134388446808,
      "learning_rate": 6.2827744542163035e-06,
      "loss": 0.0029,
      "step": 148330
    },
    {
      "epoch": 0.9514066824229636,
      "grad_norm": 0.22289535403251648,
      "learning_rate": 6.282233476266773e-06,
      "loss": 0.0024,
      "step": 148340
    },
    {
      "epoch": 0.9514708193167497,
      "grad_norm": 0.25141197443008423,
      "learning_rate": 6.281692482249938e-06,
      "loss": 0.0034,
      "step": 148350
    },
    {
      "epoch": 0.9515349562105357,
      "grad_norm": 0.0894373208284378,
      "learning_rate": 6.281151472172581e-06,
      "loss": 0.0022,
      "step": 148360
    },
    {
      "epoch": 0.9515990931043219,
      "grad_norm": 0.151262566447258,
      "learning_rate": 6.2806104460414805e-06,
      "loss": 0.0013,
      "step": 148370
    },
    {
      "epoch": 0.9516632299981079,
      "grad_norm": 0.15166862308979034,
      "learning_rate": 6.280069403863416e-06,
      "loss": 0.0015,
      "step": 148380
    },
    {
      "epoch": 0.9517273668918941,
      "grad_norm": 0.11685135215520859,
      "learning_rate": 6.279528345645168e-06,
      "loss": 0.0018,
      "step": 148390
    },
    {
      "epoch": 0.9517915037856801,
      "grad_norm": 0.06985295563936234,
      "learning_rate": 6.278987271393514e-06,
      "loss": 0.0024,
      "step": 148400
    },
    {
      "epoch": 0.9518556406794663,
      "grad_norm": 0.2967289090156555,
      "learning_rate": 6.278446181115237e-06,
      "loss": 0.0038,
      "step": 148410
    },
    {
      "epoch": 0.9519197775732523,
      "grad_norm": 0.14818093180656433,
      "learning_rate": 6.277905074817112e-06,
      "loss": 0.0024,
      "step": 148420
    },
    {
      "epoch": 0.9519839144670385,
      "grad_norm": 0.1534287929534912,
      "learning_rate": 6.277363952505926e-06,
      "loss": 0.0036,
      "step": 148430
    },
    {
      "epoch": 0.9520480513608246,
      "grad_norm": 0.06887183338403702,
      "learning_rate": 6.276822814188457e-06,
      "loss": 0.0026,
      "step": 148440
    },
    {
      "epoch": 0.9521121882546106,
      "grad_norm": 0.08078733831644058,
      "learning_rate": 6.276281659871485e-06,
      "loss": 0.0028,
      "step": 148450
    },
    {
      "epoch": 0.9521763251483968,
      "grad_norm": 0.4600955545902252,
      "learning_rate": 6.275740489561791e-06,
      "loss": 0.0048,
      "step": 148460
    },
    {
      "epoch": 0.9522404620421828,
      "grad_norm": 0.06801516562700272,
      "learning_rate": 6.275199303266158e-06,
      "loss": 0.0022,
      "step": 148470
    },
    {
      "epoch": 0.952304598935969,
      "grad_norm": 0.10994677245616913,
      "learning_rate": 6.274658100991365e-06,
      "loss": 0.0013,
      "step": 148480
    },
    {
      "epoch": 0.952368735829755,
      "grad_norm": 0.05557885766029358,
      "learning_rate": 6.274116882744197e-06,
      "loss": 0.0014,
      "step": 148490
    },
    {
      "epoch": 0.9524328727235412,
      "grad_norm": 0.07244043052196503,
      "learning_rate": 6.273575648531433e-06,
      "loss": 0.0016,
      "step": 148500
    },
    {
      "epoch": 0.9524970096173272,
      "grad_norm": 0.11104308813810349,
      "learning_rate": 6.2730343983598556e-06,
      "loss": 0.0039,
      "step": 148510
    },
    {
      "epoch": 0.9525611465111133,
      "grad_norm": 0.1490146517753601,
      "learning_rate": 6.272493132236247e-06,
      "loss": 0.0014,
      "step": 148520
    },
    {
      "epoch": 0.9526252834048994,
      "grad_norm": 0.09591233730316162,
      "learning_rate": 6.2719518501673905e-06,
      "loss": 0.0018,
      "step": 148530
    },
    {
      "epoch": 0.9526894202986855,
      "grad_norm": 0.03019757568836212,
      "learning_rate": 6.271410552160069e-06,
      "loss": 0.0015,
      "step": 148540
    },
    {
      "epoch": 0.9527535571924716,
      "grad_norm": 0.06000197306275368,
      "learning_rate": 6.270869238221064e-06,
      "loss": 0.0019,
      "step": 148550
    },
    {
      "epoch": 0.9528176940862577,
      "grad_norm": 0.11967896670103073,
      "learning_rate": 6.270327908357159e-06,
      "loss": 0.0021,
      "step": 148560
    },
    {
      "epoch": 0.9528818309800438,
      "grad_norm": 0.03409275785088539,
      "learning_rate": 6.269786562575136e-06,
      "loss": 0.0026,
      "step": 148570
    },
    {
      "epoch": 0.9529459678738299,
      "grad_norm": 0.42724716663360596,
      "learning_rate": 6.269245200881781e-06,
      "loss": 0.0026,
      "step": 148580
    },
    {
      "epoch": 0.953010104767616,
      "grad_norm": 0.046714745461940765,
      "learning_rate": 6.268703823283877e-06,
      "loss": 0.0024,
      "step": 148590
    },
    {
      "epoch": 0.9530742416614021,
      "grad_norm": 0.028787871822714806,
      "learning_rate": 6.268162429788209e-06,
      "loss": 0.0013,
      "step": 148600
    },
    {
      "epoch": 0.9531383785551882,
      "grad_norm": 0.10607559978961945,
      "learning_rate": 6.267621020401557e-06,
      "loss": 0.0025,
      "step": 148610
    },
    {
      "epoch": 0.9532025154489743,
      "grad_norm": 0.2284134030342102,
      "learning_rate": 6.2670795951307085e-06,
      "loss": 0.0036,
      "step": 148620
    },
    {
      "epoch": 0.9532666523427604,
      "grad_norm": 0.027532169595360756,
      "learning_rate": 6.266538153982446e-06,
      "loss": 0.0018,
      "step": 148630
    },
    {
      "epoch": 0.9533307892365465,
      "grad_norm": 0.093606136739254,
      "learning_rate": 6.265996696963556e-06,
      "loss": 0.0035,
      "step": 148640
    },
    {
      "epoch": 0.9533949261303326,
      "grad_norm": 0.10961989313364029,
      "learning_rate": 6.265455224080823e-06,
      "loss": 0.0068,
      "step": 148650
    },
    {
      "epoch": 0.9534590630241186,
      "grad_norm": 0.0573560893535614,
      "learning_rate": 6.264913735341032e-06,
      "loss": 0.0024,
      "step": 148660
    },
    {
      "epoch": 0.9535231999179048,
      "grad_norm": 0.05544622987508774,
      "learning_rate": 6.264372230750967e-06,
      "loss": 0.0027,
      "step": 148670
    },
    {
      "epoch": 0.9535873368116908,
      "grad_norm": 0.028547391295433044,
      "learning_rate": 6.2638307103174145e-06,
      "loss": 0.0013,
      "step": 148680
    },
    {
      "epoch": 0.953651473705477,
      "grad_norm": 0.06154303252696991,
      "learning_rate": 6.26328917404716e-06,
      "loss": 0.0015,
      "step": 148690
    },
    {
      "epoch": 0.953715610599263,
      "grad_norm": 0.12723928689956665,
      "learning_rate": 6.26274762194699e-06,
      "loss": 0.0022,
      "step": 148700
    },
    {
      "epoch": 0.9537797474930492,
      "grad_norm": 0.060413897037506104,
      "learning_rate": 6.262206054023688e-06,
      "loss": 0.002,
      "step": 148710
    },
    {
      "epoch": 0.9538438843868352,
      "grad_norm": 0.0428997203707695,
      "learning_rate": 6.261664470284044e-06,
      "loss": 0.0043,
      "step": 148720
    },
    {
      "epoch": 0.9539080212806214,
      "grad_norm": 0.03595926612615585,
      "learning_rate": 6.261122870734841e-06,
      "loss": 0.0033,
      "step": 148730
    },
    {
      "epoch": 0.9539721581744075,
      "grad_norm": 0.03506409004330635,
      "learning_rate": 6.2605812553828675e-06,
      "loss": 0.0018,
      "step": 148740
    },
    {
      "epoch": 0.9540362950681935,
      "grad_norm": 0.034021202474832535,
      "learning_rate": 6.26003962423491e-06,
      "loss": 0.0021,
      "step": 148750
    },
    {
      "epoch": 0.9541004319619797,
      "grad_norm": 0.004513194784522057,
      "learning_rate": 6.259497977297756e-06,
      "loss": 0.0015,
      "step": 148760
    },
    {
      "epoch": 0.9541645688557657,
      "grad_norm": 0.05352545529603958,
      "learning_rate": 6.258956314578193e-06,
      "loss": 0.0029,
      "step": 148770
    },
    {
      "epoch": 0.9542287057495519,
      "grad_norm": 0.04517129808664322,
      "learning_rate": 6.258414636083005e-06,
      "loss": 0.0021,
      "step": 148780
    },
    {
      "epoch": 0.9542928426433379,
      "grad_norm": 0.026336653158068657,
      "learning_rate": 6.257872941818984e-06,
      "loss": 0.0023,
      "step": 148790
    },
    {
      "epoch": 0.9543569795371241,
      "grad_norm": 0.09534820914268494,
      "learning_rate": 6.257331231792915e-06,
      "loss": 0.0018,
      "step": 148800
    },
    {
      "epoch": 0.9544211164309101,
      "grad_norm": 0.06527793407440186,
      "learning_rate": 6.256789506011588e-06,
      "loss": 0.0013,
      "step": 148810
    },
    {
      "epoch": 0.9544852533246962,
      "grad_norm": 0.16163085401058197,
      "learning_rate": 6.25624776448179e-06,
      "loss": 0.0031,
      "step": 148820
    },
    {
      "epoch": 0.9545493902184823,
      "grad_norm": 0.0022701562847942114,
      "learning_rate": 6.25570600721031e-06,
      "loss": 0.0012,
      "step": 148830
    },
    {
      "epoch": 0.9546135271122684,
      "grad_norm": 0.4319455325603485,
      "learning_rate": 6.255164234203936e-06,
      "loss": 0.0023,
      "step": 148840
    },
    {
      "epoch": 0.9546776640060545,
      "grad_norm": 0.159915953874588,
      "learning_rate": 6.254622445469458e-06,
      "loss": 0.0025,
      "step": 148850
    },
    {
      "epoch": 0.9547418008998406,
      "grad_norm": 0.24642303586006165,
      "learning_rate": 6.254080641013662e-06,
      "loss": 0.0043,
      "step": 148860
    },
    {
      "epoch": 0.9548059377936268,
      "grad_norm": 0.09090591967105865,
      "learning_rate": 6.253538820843341e-06,
      "loss": 0.0029,
      "step": 148870
    },
    {
      "epoch": 0.9548700746874128,
      "grad_norm": 0.16093572974205017,
      "learning_rate": 6.252996984965283e-06,
      "loss": 0.0024,
      "step": 148880
    },
    {
      "epoch": 0.954934211581199,
      "grad_norm": 0.06859154254198074,
      "learning_rate": 6.252455133386277e-06,
      "loss": 0.0023,
      "step": 148890
    },
    {
      "epoch": 0.954998348474985,
      "grad_norm": 0.09639650583267212,
      "learning_rate": 6.251913266113112e-06,
      "loss": 0.0029,
      "step": 148900
    },
    {
      "epoch": 0.9550624853687711,
      "grad_norm": 0.09566006064414978,
      "learning_rate": 6.25137138315258e-06,
      "loss": 0.0028,
      "step": 148910
    },
    {
      "epoch": 0.9551266222625572,
      "grad_norm": 0.08538588881492615,
      "learning_rate": 6.250829484511469e-06,
      "loss": 0.0026,
      "step": 148920
    },
    {
      "epoch": 0.9551907591563433,
      "grad_norm": 0.08481060713529587,
      "learning_rate": 6.2502875701965715e-06,
      "loss": 0.0027,
      "step": 148930
    },
    {
      "epoch": 0.9552548960501294,
      "grad_norm": 0.11255362629890442,
      "learning_rate": 6.249745640214677e-06,
      "loss": 0.0032,
      "step": 148940
    },
    {
      "epoch": 0.9553190329439155,
      "grad_norm": 0.11876288056373596,
      "learning_rate": 6.249203694572577e-06,
      "loss": 0.0026,
      "step": 148950
    },
    {
      "epoch": 0.9553831698377016,
      "grad_norm": 0.07284615933895111,
      "learning_rate": 6.248661733277062e-06,
      "loss": 0.0016,
      "step": 148960
    },
    {
      "epoch": 0.9554473067314877,
      "grad_norm": 0.12033560872077942,
      "learning_rate": 6.2481197563349215e-06,
      "loss": 0.0034,
      "step": 148970
    },
    {
      "epoch": 0.9555114436252737,
      "grad_norm": 0.15866336226463318,
      "learning_rate": 6.24757776375295e-06,
      "loss": 0.0019,
      "step": 148980
    },
    {
      "epoch": 0.9555755805190599,
      "grad_norm": 0.024077093228697777,
      "learning_rate": 6.247035755537937e-06,
      "loss": 0.0027,
      "step": 148990
    },
    {
      "epoch": 0.9556397174128459,
      "grad_norm": 0.10702551156282425,
      "learning_rate": 6.246493731696676e-06,
      "loss": 0.0043,
      "step": 149000
    },
    {
      "epoch": 0.9557038543066321,
      "grad_norm": 0.13556323945522308,
      "learning_rate": 6.245951692235955e-06,
      "loss": 0.0028,
      "step": 149010
    },
    {
      "epoch": 0.9557679912004182,
      "grad_norm": 0.08915980160236359,
      "learning_rate": 6.2454096371625715e-06,
      "loss": 0.0035,
      "step": 149020
    },
    {
      "epoch": 0.9558321280942043,
      "grad_norm": 0.08045701682567596,
      "learning_rate": 6.244867566483313e-06,
      "loss": 0.0016,
      "step": 149030
    },
    {
      "epoch": 0.9558962649879904,
      "grad_norm": 0.03717053681612015,
      "learning_rate": 6.244325480204976e-06,
      "loss": 0.002,
      "step": 149040
    },
    {
      "epoch": 0.9559604018817764,
      "grad_norm": 0.0456254780292511,
      "learning_rate": 6.243783378334349e-06,
      "loss": 0.003,
      "step": 149050
    },
    {
      "epoch": 0.9560245387755626,
      "grad_norm": 0.0967431589961052,
      "learning_rate": 6.243241260878229e-06,
      "loss": 0.0013,
      "step": 149060
    },
    {
      "epoch": 0.9560886756693486,
      "grad_norm": 0.10826282948255539,
      "learning_rate": 6.242699127843408e-06,
      "loss": 0.002,
      "step": 149070
    },
    {
      "epoch": 0.9561528125631348,
      "grad_norm": 0.04903615638613701,
      "learning_rate": 6.242156979236678e-06,
      "loss": 0.0014,
      "step": 149080
    },
    {
      "epoch": 0.9562169494569208,
      "grad_norm": 0.0974036231637001,
      "learning_rate": 6.241614815064833e-06,
      "loss": 0.0015,
      "step": 149090
    },
    {
      "epoch": 0.956281086350707,
      "grad_norm": 0.029251504689455032,
      "learning_rate": 6.241072635334669e-06,
      "loss": 0.0029,
      "step": 149100
    },
    {
      "epoch": 0.956345223244493,
      "grad_norm": 0.0733976736664772,
      "learning_rate": 6.240530440052976e-06,
      "loss": 0.0017,
      "step": 149110
    },
    {
      "epoch": 0.9564093601382792,
      "grad_norm": 0.04980620741844177,
      "learning_rate": 6.23998822922655e-06,
      "loss": 0.0012,
      "step": 149120
    },
    {
      "epoch": 0.9564734970320652,
      "grad_norm": 0.004516866523772478,
      "learning_rate": 6.239446002862186e-06,
      "loss": 0.0016,
      "step": 149130
    },
    {
      "epoch": 0.9565376339258513,
      "grad_norm": 0.0453466959297657,
      "learning_rate": 6.238903760966677e-06,
      "loss": 0.0012,
      "step": 149140
    },
    {
      "epoch": 0.9566017708196375,
      "grad_norm": 0.23368796706199646,
      "learning_rate": 6.238361503546819e-06,
      "loss": 0.0037,
      "step": 149150
    },
    {
      "epoch": 0.9566659077134235,
      "grad_norm": 0.1779107004404068,
      "learning_rate": 6.237819230609407e-06,
      "loss": 0.0029,
      "step": 149160
    },
    {
      "epoch": 0.9567300446072097,
      "grad_norm": 0.10037069767713547,
      "learning_rate": 6.237276942161234e-06,
      "loss": 0.001,
      "step": 149170
    },
    {
      "epoch": 0.9567941815009957,
      "grad_norm": 0.1792505979537964,
      "learning_rate": 6.2367346382090964e-06,
      "loss": 0.002,
      "step": 149180
    },
    {
      "epoch": 0.9568583183947819,
      "grad_norm": 0.04580867290496826,
      "learning_rate": 6.23619231875979e-06,
      "loss": 0.0018,
      "step": 149190
    },
    {
      "epoch": 0.9569224552885679,
      "grad_norm": 0.02684248611330986,
      "learning_rate": 6.2356499838201115e-06,
      "loss": 0.0026,
      "step": 149200
    },
    {
      "epoch": 0.956986592182354,
      "grad_norm": 0.004145835526287556,
      "learning_rate": 6.235107633396855e-06,
      "loss": 0.0012,
      "step": 149210
    },
    {
      "epoch": 0.9570507290761401,
      "grad_norm": 0.06484013795852661,
      "learning_rate": 6.234565267496817e-06,
      "loss": 0.0023,
      "step": 149220
    },
    {
      "epoch": 0.9571148659699262,
      "grad_norm": 0.12458470463752747,
      "learning_rate": 6.234022886126795e-06,
      "loss": 0.002,
      "step": 149230
    },
    {
      "epoch": 0.9571790028637123,
      "grad_norm": 0.021018436178565025,
      "learning_rate": 6.233480489293583e-06,
      "loss": 0.0017,
      "step": 149240
    },
    {
      "epoch": 0.9572431397574984,
      "grad_norm": 0.20648528635501862,
      "learning_rate": 6.23293807700398e-06,
      "loss": 0.0019,
      "step": 149250
    },
    {
      "epoch": 0.9573072766512845,
      "grad_norm": 0.17190244793891907,
      "learning_rate": 6.23239564926478e-06,
      "loss": 0.006,
      "step": 149260
    },
    {
      "epoch": 0.9573714135450706,
      "grad_norm": 0.16801367700099945,
      "learning_rate": 6.231853206082783e-06,
      "loss": 0.0014,
      "step": 149270
    },
    {
      "epoch": 0.9574355504388566,
      "grad_norm": 0.05028393119573593,
      "learning_rate": 6.231310747464785e-06,
      "loss": 0.0017,
      "step": 149280
    },
    {
      "epoch": 0.9574996873326428,
      "grad_norm": 0.06627675890922546,
      "learning_rate": 6.230768273417582e-06,
      "loss": 0.002,
      "step": 149290
    },
    {
      "epoch": 0.9575638242264289,
      "grad_norm": 0.17414219677448273,
      "learning_rate": 6.230225783947975e-06,
      "loss": 0.0035,
      "step": 149300
    },
    {
      "epoch": 0.957627961120215,
      "grad_norm": 0.06667561084032059,
      "learning_rate": 6.229683279062758e-06,
      "loss": 0.0013,
      "step": 149310
    },
    {
      "epoch": 0.9576920980140011,
      "grad_norm": 0.048860788345336914,
      "learning_rate": 6.229140758768732e-06,
      "loss": 0.0016,
      "step": 149320
    },
    {
      "epoch": 0.9577562349077872,
      "grad_norm": 0.11049015820026398,
      "learning_rate": 6.228598223072692e-06,
      "loss": 0.0031,
      "step": 149330
    },
    {
      "epoch": 0.9578203718015733,
      "grad_norm": 0.07773560285568237,
      "learning_rate": 6.228055671981441e-06,
      "loss": 0.0023,
      "step": 149340
    },
    {
      "epoch": 0.9578845086953593,
      "grad_norm": 0.09300699084997177,
      "learning_rate": 6.227513105501773e-06,
      "loss": 0.0017,
      "step": 149350
    },
    {
      "epoch": 0.9579486455891455,
      "grad_norm": 0.11326320469379425,
      "learning_rate": 6.226970523640489e-06,
      "loss": 0.0009,
      "step": 149360
    },
    {
      "epoch": 0.9580127824829315,
      "grad_norm": 0.09000983834266663,
      "learning_rate": 6.226427926404387e-06,
      "loss": 0.0024,
      "step": 149370
    },
    {
      "epoch": 0.9580769193767177,
      "grad_norm": 0.09515196084976196,
      "learning_rate": 6.225885313800267e-06,
      "loss": 0.0019,
      "step": 149380
    },
    {
      "epoch": 0.9581410562705037,
      "grad_norm": 0.04229377210140228,
      "learning_rate": 6.225342685834927e-06,
      "loss": 0.0017,
      "step": 149390
    },
    {
      "epoch": 0.9582051931642899,
      "grad_norm": 0.03817709535360336,
      "learning_rate": 6.224800042515169e-06,
      "loss": 0.0015,
      "step": 149400
    },
    {
      "epoch": 0.9582693300580759,
      "grad_norm": 0.19991615414619446,
      "learning_rate": 6.224257383847789e-06,
      "loss": 0.0017,
      "step": 149410
    },
    {
      "epoch": 0.9583334669518621,
      "grad_norm": 0.030445698648691177,
      "learning_rate": 6.22371470983959e-06,
      "loss": 0.002,
      "step": 149420
    },
    {
      "epoch": 0.9583976038456482,
      "grad_norm": 0.08594322949647903,
      "learning_rate": 6.223172020497372e-06,
      "loss": 0.0017,
      "step": 149430
    },
    {
      "epoch": 0.9584617407394342,
      "grad_norm": 0.047370102256536484,
      "learning_rate": 6.2226293158279324e-06,
      "loss": 0.0016,
      "step": 149440
    },
    {
      "epoch": 0.9585258776332204,
      "grad_norm": 0.4207480847835541,
      "learning_rate": 6.222086595838076e-06,
      "loss": 0.0039,
      "step": 149450
    },
    {
      "epoch": 0.9585900145270064,
      "grad_norm": 0.04545342177152634,
      "learning_rate": 6.221543860534599e-06,
      "loss": 0.0018,
      "step": 149460
    },
    {
      "epoch": 0.9586541514207926,
      "grad_norm": 0.07949335128068924,
      "learning_rate": 6.221001109924306e-06,
      "loss": 0.0017,
      "step": 149470
    },
    {
      "epoch": 0.9587182883145786,
      "grad_norm": 0.11357042193412781,
      "learning_rate": 6.220458344013995e-06,
      "loss": 0.002,
      "step": 149480
    },
    {
      "epoch": 0.9587824252083648,
      "grad_norm": 0.0045299651101231575,
      "learning_rate": 6.2199155628104715e-06,
      "loss": 0.0012,
      "step": 149490
    },
    {
      "epoch": 0.9588465621021508,
      "grad_norm": 0.037436194717884064,
      "learning_rate": 6.219372766320531e-06,
      "loss": 0.0014,
      "step": 149500
    },
    {
      "epoch": 0.958910698995937,
      "grad_norm": 0.11348027735948563,
      "learning_rate": 6.21882995455098e-06,
      "loss": 0.0017,
      "step": 149510
    },
    {
      "epoch": 0.958974835889723,
      "grad_norm": 0.08938112109899521,
      "learning_rate": 6.218287127508618e-06,
      "loss": 0.0014,
      "step": 149520
    },
    {
      "epoch": 0.9590389727835091,
      "grad_norm": 0.008515220135450363,
      "learning_rate": 6.217744285200248e-06,
      "loss": 0.0048,
      "step": 149530
    },
    {
      "epoch": 0.9591031096772952,
      "grad_norm": 0.20820820331573486,
      "learning_rate": 6.217201427632671e-06,
      "loss": 0.0072,
      "step": 149540
    },
    {
      "epoch": 0.9591672465710813,
      "grad_norm": 0.0789012610912323,
      "learning_rate": 6.216658554812691e-06,
      "loss": 0.0016,
      "step": 149550
    },
    {
      "epoch": 0.9592313834648674,
      "grad_norm": 0.07566607743501663,
      "learning_rate": 6.216115666747109e-06,
      "loss": 0.0053,
      "step": 149560
    },
    {
      "epoch": 0.9592955203586535,
      "grad_norm": 0.1387602835893631,
      "learning_rate": 6.215572763442729e-06,
      "loss": 0.0043,
      "step": 149570
    },
    {
      "epoch": 0.9593596572524397,
      "grad_norm": 0.133415088057518,
      "learning_rate": 6.215029844906353e-06,
      "loss": 0.0022,
      "step": 149580
    },
    {
      "epoch": 0.9594237941462257,
      "grad_norm": 0.1288202852010727,
      "learning_rate": 6.214486911144786e-06,
      "loss": 0.0025,
      "step": 149590
    },
    {
      "epoch": 0.9594879310400118,
      "grad_norm": 0.1541908234357834,
      "learning_rate": 6.2139439621648275e-06,
      "loss": 0.0017,
      "step": 149600
    },
    {
      "epoch": 0.9595520679337979,
      "grad_norm": 0.3520669937133789,
      "learning_rate": 6.213400997973286e-06,
      "loss": 0.003,
      "step": 149610
    },
    {
      "epoch": 0.959616204827584,
      "grad_norm": 0.040001530200242996,
      "learning_rate": 6.212858018576962e-06,
      "loss": 0.0012,
      "step": 149620
    },
    {
      "epoch": 0.9596803417213701,
      "grad_norm": 0.17043310403823853,
      "learning_rate": 6.2123150239826605e-06,
      "loss": 0.0016,
      "step": 149630
    },
    {
      "epoch": 0.9597444786151562,
      "grad_norm": 0.11625032871961594,
      "learning_rate": 6.211772014197185e-06,
      "loss": 0.0037,
      "step": 149640
    },
    {
      "epoch": 0.9598086155089423,
      "grad_norm": 0.14066296815872192,
      "learning_rate": 6.211228989227339e-06,
      "loss": 0.0032,
      "step": 149650
    },
    {
      "epoch": 0.9598727524027284,
      "grad_norm": 0.10146692395210266,
      "learning_rate": 6.2106859490799305e-06,
      "loss": 0.002,
      "step": 149660
    },
    {
      "epoch": 0.9599368892965144,
      "grad_norm": 0.09130629897117615,
      "learning_rate": 6.21014289376176e-06,
      "loss": 0.0018,
      "step": 149670
    },
    {
      "epoch": 0.9600010261903006,
      "grad_norm": 0.10161637514829636,
      "learning_rate": 6.209599823279635e-06,
      "loss": 0.002,
      "step": 149680
    },
    {
      "epoch": 0.9600651630840866,
      "grad_norm": 0.07758409529924393,
      "learning_rate": 6.20905673764036e-06,
      "loss": 0.0021,
      "step": 149690
    },
    {
      "epoch": 0.9601292999778728,
      "grad_norm": 0.06893595308065414,
      "learning_rate": 6.208513636850739e-06,
      "loss": 0.0021,
      "step": 149700
    },
    {
      "epoch": 0.9601934368716589,
      "grad_norm": 0.11017578840255737,
      "learning_rate": 6.207970520917579e-06,
      "loss": 0.0015,
      "step": 149710
    },
    {
      "epoch": 0.960257573765445,
      "grad_norm": 0.035975776612758636,
      "learning_rate": 6.207427389847685e-06,
      "loss": 0.0018,
      "step": 149720
    },
    {
      "epoch": 0.9603217106592311,
      "grad_norm": 0.2288781851530075,
      "learning_rate": 6.206884243647863e-06,
      "loss": 0.0019,
      "step": 149730
    },
    {
      "epoch": 0.9603858475530171,
      "grad_norm": 0.3251713216304779,
      "learning_rate": 6.206341082324919e-06,
      "loss": 0.0026,
      "step": 149740
    },
    {
      "epoch": 0.9604499844468033,
      "grad_norm": 0.062379252165555954,
      "learning_rate": 6.205797905885658e-06,
      "loss": 0.0017,
      "step": 149750
    },
    {
      "epoch": 0.9605141213405893,
      "grad_norm": 0.130185067653656,
      "learning_rate": 6.205254714336889e-06,
      "loss": 0.0017,
      "step": 149760
    },
    {
      "epoch": 0.9605782582343755,
      "grad_norm": 0.04989314824342728,
      "learning_rate": 6.204711507685416e-06,
      "loss": 0.0035,
      "step": 149770
    },
    {
      "epoch": 0.9606423951281615,
      "grad_norm": 0.09429958462715149,
      "learning_rate": 6.204168285938046e-06,
      "loss": 0.0027,
      "step": 149780
    },
    {
      "epoch": 0.9607065320219477,
      "grad_norm": 0.13752683997154236,
      "learning_rate": 6.20362504910159e-06,
      "loss": 0.0022,
      "step": 149790
    },
    {
      "epoch": 0.9607706689157337,
      "grad_norm": 0.05318170785903931,
      "learning_rate": 6.203081797182848e-06,
      "loss": 0.0009,
      "step": 149800
    },
    {
      "epoch": 0.9608348058095199,
      "grad_norm": 0.0863172858953476,
      "learning_rate": 6.2025385301886335e-06,
      "loss": 0.0028,
      "step": 149810
    },
    {
      "epoch": 0.9608989427033059,
      "grad_norm": 0.4169495105743408,
      "learning_rate": 6.201995248125752e-06,
      "loss": 0.0032,
      "step": 149820
    },
    {
      "epoch": 0.960963079597092,
      "grad_norm": 0.08999016135931015,
      "learning_rate": 6.201451951001009e-06,
      "loss": 0.0033,
      "step": 149830
    },
    {
      "epoch": 0.9610272164908781,
      "grad_norm": 0.09195482730865479,
      "learning_rate": 6.200908638821216e-06,
      "loss": 0.0019,
      "step": 149840
    },
    {
      "epoch": 0.9610913533846642,
      "grad_norm": 0.09480104595422745,
      "learning_rate": 6.20036531159318e-06,
      "loss": 0.0011,
      "step": 149850
    },
    {
      "epoch": 0.9611554902784504,
      "grad_norm": 0.19265435636043549,
      "learning_rate": 6.199821969323707e-06,
      "loss": 0.002,
      "step": 149860
    },
    {
      "epoch": 0.9612196271722364,
      "grad_norm": 0.04201951250433922,
      "learning_rate": 6.199278612019609e-06,
      "loss": 0.0019,
      "step": 149870
    },
    {
      "epoch": 0.9612837640660226,
      "grad_norm": 0.13683591783046722,
      "learning_rate": 6.198735239687692e-06,
      "loss": 0.0014,
      "step": 149880
    },
    {
      "epoch": 0.9613479009598086,
      "grad_norm": 0.07254856079816818,
      "learning_rate": 6.198191852334766e-06,
      "loss": 0.0027,
      "step": 149890
    },
    {
      "epoch": 0.9614120378535947,
      "grad_norm": 0.013781868852674961,
      "learning_rate": 6.197648449967639e-06,
      "loss": 0.0027,
      "step": 149900
    },
    {
      "epoch": 0.9614761747473808,
      "grad_norm": 0.059696830809116364,
      "learning_rate": 6.197105032593121e-06,
      "loss": 0.0025,
      "step": 149910
    },
    {
      "epoch": 0.9615403116411669,
      "grad_norm": 0.1655527651309967,
      "learning_rate": 6.196561600218023e-06,
      "loss": 0.0021,
      "step": 149920
    },
    {
      "epoch": 0.961604448534953,
      "grad_norm": 0.11569607257843018,
      "learning_rate": 6.19601815284915e-06,
      "loss": 0.0025,
      "step": 149930
    },
    {
      "epoch": 0.9616685854287391,
      "grad_norm": 0.0929485484957695,
      "learning_rate": 6.1954746904933184e-06,
      "loss": 0.0013,
      "step": 149940
    },
    {
      "epoch": 0.9617327223225252,
      "grad_norm": 0.14032530784606934,
      "learning_rate": 6.1949312131573315e-06,
      "loss": 0.0033,
      "step": 149950
    },
    {
      "epoch": 0.9617968592163113,
      "grad_norm": 0.02642715536057949,
      "learning_rate": 6.194387720848003e-06,
      "loss": 0.0023,
      "step": 149960
    },
    {
      "epoch": 0.9618609961100973,
      "grad_norm": 0.15446075797080994,
      "learning_rate": 6.193844213572143e-06,
      "loss": 0.0025,
      "step": 149970
    },
    {
      "epoch": 0.9619251330038835,
      "grad_norm": 0.16909636557102203,
      "learning_rate": 6.193300691336563e-06,
      "loss": 0.0025,
      "step": 149980
    },
    {
      "epoch": 0.9619892698976696,
      "grad_norm": 0.12399910390377045,
      "learning_rate": 6.192757154148071e-06,
      "loss": 0.0042,
      "step": 149990
    },
    {
      "epoch": 0.9620534067914557,
      "grad_norm": 0.1202448159456253,
      "learning_rate": 6.192213602013481e-06,
      "loss": 0.0015,
      "step": 150000
    },
    {
      "epoch": 0.9621175436852418,
      "grad_norm": 0.20809784531593323,
      "learning_rate": 6.191670034939602e-06,
      "loss": 0.0021,
      "step": 150010
    },
    {
      "epoch": 0.9621816805790279,
      "grad_norm": 0.03879670053720474,
      "learning_rate": 6.191126452933246e-06,
      "loss": 0.0022,
      "step": 150020
    },
    {
      "epoch": 0.962245817472814,
      "grad_norm": 0.1990572065114975,
      "learning_rate": 6.190582856001222e-06,
      "loss": 0.0031,
      "step": 150030
    },
    {
      "epoch": 0.9623099543666,
      "grad_norm": 0.04990717023611069,
      "learning_rate": 6.190039244150348e-06,
      "loss": 0.0027,
      "step": 150040
    },
    {
      "epoch": 0.9623740912603862,
      "grad_norm": 0.10544680804014206,
      "learning_rate": 6.189495617387428e-06,
      "loss": 0.0017,
      "step": 150050
    },
    {
      "epoch": 0.9624382281541722,
      "grad_norm": 0.18796586990356445,
      "learning_rate": 6.1889519757192795e-06,
      "loss": 0.0025,
      "step": 150060
    },
    {
      "epoch": 0.9625023650479584,
      "grad_norm": 0.07660584151744843,
      "learning_rate": 6.1884083191527125e-06,
      "loss": 0.0015,
      "step": 150070
    },
    {
      "epoch": 0.9625665019417444,
      "grad_norm": 0.10856002569198608,
      "learning_rate": 6.187864647694541e-06,
      "loss": 0.0016,
      "step": 150080
    },
    {
      "epoch": 0.9626306388355306,
      "grad_norm": 0.10406231135129929,
      "learning_rate": 6.187320961351575e-06,
      "loss": 0.0043,
      "step": 150090
    },
    {
      "epoch": 0.9626947757293166,
      "grad_norm": 0.06723621487617493,
      "learning_rate": 6.18677726013063e-06,
      "loss": 0.0022,
      "step": 150100
    },
    {
      "epoch": 0.9627589126231028,
      "grad_norm": 0.13492071628570557,
      "learning_rate": 6.186233544038517e-06,
      "loss": 0.0018,
      "step": 150110
    },
    {
      "epoch": 0.9628230495168888,
      "grad_norm": 0.09268075972795486,
      "learning_rate": 6.18568981308205e-06,
      "loss": 0.0026,
      "step": 150120
    },
    {
      "epoch": 0.962887186410675,
      "grad_norm": 0.08964299410581589,
      "learning_rate": 6.185146067268042e-06,
      "loss": 0.0021,
      "step": 150130
    },
    {
      "epoch": 0.9629513233044611,
      "grad_norm": 0.05227648839354515,
      "learning_rate": 6.1846023066033066e-06,
      "loss": 0.0056,
      "step": 150140
    },
    {
      "epoch": 0.9630154601982471,
      "grad_norm": 0.0577087476849556,
      "learning_rate": 6.184058531094659e-06,
      "loss": 0.0016,
      "step": 150150
    },
    {
      "epoch": 0.9630795970920333,
      "grad_norm": 0.04068737104535103,
      "learning_rate": 6.183514740748911e-06,
      "loss": 0.0021,
      "step": 150160
    },
    {
      "epoch": 0.9631437339858193,
      "grad_norm": 0.1221189871430397,
      "learning_rate": 6.182970935572877e-06,
      "loss": 0.0017,
      "step": 150170
    },
    {
      "epoch": 0.9632078708796055,
      "grad_norm": 0.11458059400320053,
      "learning_rate": 6.1824271155733715e-06,
      "loss": 0.0013,
      "step": 150180
    },
    {
      "epoch": 0.9632720077733915,
      "grad_norm": 0.10058407485485077,
      "learning_rate": 6.1818832807572106e-06,
      "loss": 0.0018,
      "step": 150190
    },
    {
      "epoch": 0.9633361446671777,
      "grad_norm": 0.1607140749692917,
      "learning_rate": 6.181339431131205e-06,
      "loss": 0.0029,
      "step": 150200
    },
    {
      "epoch": 0.9634002815609637,
      "grad_norm": 0.4767524302005768,
      "learning_rate": 6.1807955667021755e-06,
      "loss": 0.0028,
      "step": 150210
    },
    {
      "epoch": 0.9634644184547498,
      "grad_norm": 0.10355031490325928,
      "learning_rate": 6.180251687476932e-06,
      "loss": 0.0014,
      "step": 150220
    },
    {
      "epoch": 0.9635285553485359,
      "grad_norm": 0.19108106195926666,
      "learning_rate": 6.179707793462292e-06,
      "loss": 0.0019,
      "step": 150230
    },
    {
      "epoch": 0.963592692242322,
      "grad_norm": 0.13633432984352112,
      "learning_rate": 6.179163884665068e-06,
      "loss": 0.0017,
      "step": 150240
    },
    {
      "epoch": 0.9636568291361081,
      "grad_norm": 0.05463302135467529,
      "learning_rate": 6.1786199610920804e-06,
      "loss": 0.0022,
      "step": 150250
    },
    {
      "epoch": 0.9637209660298942,
      "grad_norm": 0.03905477002263069,
      "learning_rate": 6.17807602275014e-06,
      "loss": 0.0012,
      "step": 150260
    },
    {
      "epoch": 0.9637851029236802,
      "grad_norm": 0.05089954286813736,
      "learning_rate": 6.177532069646066e-06,
      "loss": 0.0025,
      "step": 150270
    },
    {
      "epoch": 0.9638492398174664,
      "grad_norm": 0.05219513922929764,
      "learning_rate": 6.176988101786675e-06,
      "loss": 0.0011,
      "step": 150280
    },
    {
      "epoch": 0.9639133767112525,
      "grad_norm": 0.1544886827468872,
      "learning_rate": 6.176444119178779e-06,
      "loss": 0.002,
      "step": 150290
    },
    {
      "epoch": 0.9639775136050386,
      "grad_norm": 0.21321026980876923,
      "learning_rate": 6.1759001218292e-06,
      "loss": 0.0019,
      "step": 150300
    },
    {
      "epoch": 0.9640416504988247,
      "grad_norm": 0.05800163745880127,
      "learning_rate": 6.175356109744751e-06,
      "loss": 0.0016,
      "step": 150310
    },
    {
      "epoch": 0.9641057873926108,
      "grad_norm": 0.08184763789176941,
      "learning_rate": 6.174812082932251e-06,
      "loss": 0.0012,
      "step": 150320
    },
    {
      "epoch": 0.9641699242863969,
      "grad_norm": 0.16461877524852753,
      "learning_rate": 6.174268041398514e-06,
      "loss": 0.002,
      "step": 150330
    },
    {
      "epoch": 0.964234061180183,
      "grad_norm": 0.10538577288389206,
      "learning_rate": 6.17372398515036e-06,
      "loss": 0.0011,
      "step": 150340
    },
    {
      "epoch": 0.9642981980739691,
      "grad_norm": 0.07570752501487732,
      "learning_rate": 6.173179914194606e-06,
      "loss": 0.0018,
      "step": 150350
    },
    {
      "epoch": 0.9643623349677551,
      "grad_norm": 0.04098653048276901,
      "learning_rate": 6.172635828538069e-06,
      "loss": 0.0018,
      "step": 150360
    },
    {
      "epoch": 0.9644264718615413,
      "grad_norm": 0.10436978936195374,
      "learning_rate": 6.172091728187568e-06,
      "loss": 0.0037,
      "step": 150370
    },
    {
      "epoch": 0.9644906087553273,
      "grad_norm": 0.059966154396533966,
      "learning_rate": 6.17154761314992e-06,
      "loss": 0.0034,
      "step": 150380
    },
    {
      "epoch": 0.9645547456491135,
      "grad_norm": 0.14894920587539673,
      "learning_rate": 6.171003483431941e-06,
      "loss": 0.003,
      "step": 150390
    },
    {
      "epoch": 0.9646188825428995,
      "grad_norm": 0.105308398604393,
      "learning_rate": 6.170459339040453e-06,
      "loss": 0.0038,
      "step": 150400
    },
    {
      "epoch": 0.9646830194366857,
      "grad_norm": 0.07003585249185562,
      "learning_rate": 6.169915179982272e-06,
      "loss": 0.0029,
      "step": 150410
    },
    {
      "epoch": 0.9647471563304718,
      "grad_norm": 0.10759350657463074,
      "learning_rate": 6.16937100626422e-06,
      "loss": 0.0029,
      "step": 150420
    },
    {
      "epoch": 0.9648112932242578,
      "grad_norm": 0.12816260755062103,
      "learning_rate": 6.168826817893111e-06,
      "loss": 0.0017,
      "step": 150430
    },
    {
      "epoch": 0.964875430118044,
      "grad_norm": 0.08761973679065704,
      "learning_rate": 6.168282614875768e-06,
      "loss": 0.0011,
      "step": 150440
    },
    {
      "epoch": 0.96493956701183,
      "grad_norm": 0.1492198407649994,
      "learning_rate": 6.167738397219008e-06,
      "loss": 0.0033,
      "step": 150450
    },
    {
      "epoch": 0.9650037039056162,
      "grad_norm": 0.6813758611679077,
      "learning_rate": 6.167194164929652e-06,
      "loss": 0.0084,
      "step": 150460
    },
    {
      "epoch": 0.9650678407994022,
      "grad_norm": 0.1323823630809784,
      "learning_rate": 6.166649918014518e-06,
      "loss": 0.0034,
      "step": 150470
    },
    {
      "epoch": 0.9651319776931884,
      "grad_norm": 0.048957593739032745,
      "learning_rate": 6.166105656480426e-06,
      "loss": 0.0017,
      "step": 150480
    },
    {
      "epoch": 0.9651961145869744,
      "grad_norm": 0.10954766720533371,
      "learning_rate": 6.165561380334199e-06,
      "loss": 0.0019,
      "step": 150490
    },
    {
      "epoch": 0.9652602514807606,
      "grad_norm": 0.06525308638811111,
      "learning_rate": 6.165017089582654e-06,
      "loss": 0.0021,
      "step": 150500
    },
    {
      "epoch": 0.9653243883745466,
      "grad_norm": 0.042362719774246216,
      "learning_rate": 6.1644727842326134e-06,
      "loss": 0.003,
      "step": 150510
    },
    {
      "epoch": 0.9653885252683327,
      "grad_norm": 0.26810696721076965,
      "learning_rate": 6.163928464290895e-06,
      "loss": 0.0035,
      "step": 150520
    },
    {
      "epoch": 0.9654526621621188,
      "grad_norm": 0.12526772916316986,
      "learning_rate": 6.1633841297643215e-06,
      "loss": 0.0024,
      "step": 150530
    },
    {
      "epoch": 0.9655167990559049,
      "grad_norm": 0.04780496284365654,
      "learning_rate": 6.162839780659713e-06,
      "loss": 0.0018,
      "step": 150540
    },
    {
      "epoch": 0.965580935949691,
      "grad_norm": 0.02072593756020069,
      "learning_rate": 6.162295416983892e-06,
      "loss": 0.003,
      "step": 150550
    },
    {
      "epoch": 0.9656450728434771,
      "grad_norm": 0.10294642299413681,
      "learning_rate": 6.161751038743678e-06,
      "loss": 0.002,
      "step": 150560
    },
    {
      "epoch": 0.9657092097372633,
      "grad_norm": 0.09494373947381973,
      "learning_rate": 6.161206645945893e-06,
      "loss": 0.0022,
      "step": 150570
    },
    {
      "epoch": 0.9657733466310493,
      "grad_norm": 0.18811984360218048,
      "learning_rate": 6.160662238597359e-06,
      "loss": 0.0021,
      "step": 150580
    },
    {
      "epoch": 0.9658374835248354,
      "grad_norm": 0.08301776647567749,
      "learning_rate": 6.160117816704898e-06,
      "loss": 0.0019,
      "step": 150590
    },
    {
      "epoch": 0.9659016204186215,
      "grad_norm": 0.21943216025829315,
      "learning_rate": 6.159573380275331e-06,
      "loss": 0.0018,
      "step": 150600
    },
    {
      "epoch": 0.9659657573124076,
      "grad_norm": 0.0833280086517334,
      "learning_rate": 6.1590289293154825e-06,
      "loss": 0.002,
      "step": 150610
    },
    {
      "epoch": 0.9660298942061937,
      "grad_norm": 0.6742835640907288,
      "learning_rate": 6.1584844638321705e-06,
      "loss": 0.0028,
      "step": 150620
    },
    {
      "epoch": 0.9660940310999798,
      "grad_norm": 0.30826130509376526,
      "learning_rate": 6.1579399838322216e-06,
      "loss": 0.0015,
      "step": 150630
    },
    {
      "epoch": 0.9661581679937659,
      "grad_norm": 0.24817430973052979,
      "learning_rate": 6.1573954893224576e-06,
      "loss": 0.0016,
      "step": 150640
    },
    {
      "epoch": 0.966222304887552,
      "grad_norm": 0.11512727290391922,
      "learning_rate": 6.1568509803097e-06,
      "loss": 0.0014,
      "step": 150650
    },
    {
      "epoch": 0.966286441781338,
      "grad_norm": 0.0032293221447616816,
      "learning_rate": 6.1563064568007735e-06,
      "loss": 0.0035,
      "step": 150660
    },
    {
      "epoch": 0.9663505786751242,
      "grad_norm": 0.06835031509399414,
      "learning_rate": 6.155761918802501e-06,
      "loss": 0.0024,
      "step": 150670
    },
    {
      "epoch": 0.9664147155689102,
      "grad_norm": 0.13923440873622894,
      "learning_rate": 6.155217366321705e-06,
      "loss": 0.0015,
      "step": 150680
    },
    {
      "epoch": 0.9664788524626964,
      "grad_norm": 0.16111516952514648,
      "learning_rate": 6.15467279936521e-06,
      "loss": 0.0015,
      "step": 150690
    },
    {
      "epoch": 0.9665429893564825,
      "grad_norm": 0.10677210241556168,
      "learning_rate": 6.15412821793984e-06,
      "loss": 0.0014,
      "step": 150700
    },
    {
      "epoch": 0.9666071262502686,
      "grad_norm": 0.10801635682582855,
      "learning_rate": 6.153583622052417e-06,
      "loss": 0.0018,
      "step": 150710
    },
    {
      "epoch": 0.9666712631440547,
      "grad_norm": 0.19023743271827698,
      "learning_rate": 6.153039011709767e-06,
      "loss": 0.0026,
      "step": 150720
    },
    {
      "epoch": 0.9667354000378408,
      "grad_norm": 0.07062771171331406,
      "learning_rate": 6.1524943869187145e-06,
      "loss": 0.0021,
      "step": 150730
    },
    {
      "epoch": 0.9667995369316269,
      "grad_norm": 0.051830243319272995,
      "learning_rate": 6.151949747686085e-06,
      "loss": 0.0016,
      "step": 150740
    },
    {
      "epoch": 0.9668636738254129,
      "grad_norm": 0.0025231295730918646,
      "learning_rate": 6.151405094018701e-06,
      "loss": 0.001,
      "step": 150750
    },
    {
      "epoch": 0.9669278107191991,
      "grad_norm": 0.2973938584327698,
      "learning_rate": 6.1508604259233885e-06,
      "loss": 0.0022,
      "step": 150760
    },
    {
      "epoch": 0.9669919476129851,
      "grad_norm": 0.091313935816288,
      "learning_rate": 6.150315743406972e-06,
      "loss": 0.0012,
      "step": 150770
    },
    {
      "epoch": 0.9670560845067713,
      "grad_norm": 0.0033550853841006756,
      "learning_rate": 6.149771046476278e-06,
      "loss": 0.0019,
      "step": 150780
    },
    {
      "epoch": 0.9671202214005573,
      "grad_norm": 0.006705216597765684,
      "learning_rate": 6.14922633513813e-06,
      "loss": 0.0026,
      "step": 150790
    },
    {
      "epoch": 0.9671843582943435,
      "grad_norm": 0.10822317749261856,
      "learning_rate": 6.1486816093993555e-06,
      "loss": 0.0021,
      "step": 150800
    },
    {
      "epoch": 0.9672484951881295,
      "grad_norm": 0.07211865484714508,
      "learning_rate": 6.148136869266778e-06,
      "loss": 0.0018,
      "step": 150810
    },
    {
      "epoch": 0.9673126320819156,
      "grad_norm": 0.10985633730888367,
      "learning_rate": 6.147592114747225e-06,
      "loss": 0.0021,
      "step": 150820
    },
    {
      "epoch": 0.9673767689757017,
      "grad_norm": 0.29983705282211304,
      "learning_rate": 6.147047345847524e-06,
      "loss": 0.0029,
      "step": 150830
    },
    {
      "epoch": 0.9674409058694878,
      "grad_norm": 0.06410083919763565,
      "learning_rate": 6.1465025625745e-06,
      "loss": 0.0012,
      "step": 150840
    },
    {
      "epoch": 0.967505042763274,
      "grad_norm": 0.3119731545448303,
      "learning_rate": 6.14595776493498e-06,
      "loss": 0.0022,
      "step": 150850
    },
    {
      "epoch": 0.96756917965706,
      "grad_norm": 0.17304588854312897,
      "learning_rate": 6.1454129529357885e-06,
      "loss": 0.003,
      "step": 150860
    },
    {
      "epoch": 0.9676333165508462,
      "grad_norm": 0.11323154717683792,
      "learning_rate": 6.144868126583755e-06,
      "loss": 0.0034,
      "step": 150870
    },
    {
      "epoch": 0.9676974534446322,
      "grad_norm": 0.11185193061828613,
      "learning_rate": 6.1443232858857045e-06,
      "loss": 0.002,
      "step": 150880
    },
    {
      "epoch": 0.9677615903384184,
      "grad_norm": 0.09502777457237244,
      "learning_rate": 6.143778430848467e-06,
      "loss": 0.0027,
      "step": 150890
    },
    {
      "epoch": 0.9678257272322044,
      "grad_norm": 0.11737873405218124,
      "learning_rate": 6.1432335614788675e-06,
      "loss": 0.0017,
      "step": 150900
    },
    {
      "epoch": 0.9678898641259905,
      "grad_norm": 0.14510773122310638,
      "learning_rate": 6.142688677783736e-06,
      "loss": 0.0019,
      "step": 150910
    },
    {
      "epoch": 0.9679540010197766,
      "grad_norm": 0.16018038988113403,
      "learning_rate": 6.142143779769896e-06,
      "loss": 0.0022,
      "step": 150920
    },
    {
      "epoch": 0.9680181379135627,
      "grad_norm": 0.028643252328038216,
      "learning_rate": 6.141598867444181e-06,
      "loss": 0.0032,
      "step": 150930
    },
    {
      "epoch": 0.9680822748073488,
      "grad_norm": 0.10917980223894119,
      "learning_rate": 6.141053940813414e-06,
      "loss": 0.0019,
      "step": 150940
    },
    {
      "epoch": 0.9681464117011349,
      "grad_norm": 0.12204719334840775,
      "learning_rate": 6.140508999884427e-06,
      "loss": 0.0017,
      "step": 150950
    },
    {
      "epoch": 0.968210548594921,
      "grad_norm": 0.7267501354217529,
      "learning_rate": 6.139964044664046e-06,
      "loss": 0.0021,
      "step": 150960
    },
    {
      "epoch": 0.9682746854887071,
      "grad_norm": 0.2687130272388458,
      "learning_rate": 6.139419075159101e-06,
      "loss": 0.0014,
      "step": 150970
    },
    {
      "epoch": 0.9683388223824932,
      "grad_norm": 0.22710467875003815,
      "learning_rate": 6.138874091376421e-06,
      "loss": 0.0025,
      "step": 150980
    },
    {
      "epoch": 0.9684029592762793,
      "grad_norm": 0.05579902231693268,
      "learning_rate": 6.1383290933228345e-06,
      "loss": 0.0024,
      "step": 150990
    },
    {
      "epoch": 0.9684670961700654,
      "grad_norm": 0.10017244517803192,
      "learning_rate": 6.137784081005171e-06,
      "loss": 0.0017,
      "step": 151000
    },
    {
      "epoch": 0.9685312330638515,
      "grad_norm": 0.05790210887789726,
      "learning_rate": 6.13723905443026e-06,
      "loss": 0.0018,
      "step": 151010
    },
    {
      "epoch": 0.9685953699576376,
      "grad_norm": 0.14402702450752258,
      "learning_rate": 6.136694013604932e-06,
      "loss": 0.0037,
      "step": 151020
    },
    {
      "epoch": 0.9686595068514237,
      "grad_norm": 0.2564027011394501,
      "learning_rate": 6.136148958536014e-06,
      "loss": 0.0024,
      "step": 151030
    },
    {
      "epoch": 0.9687236437452098,
      "grad_norm": 0.3121611475944519,
      "learning_rate": 6.135603889230337e-06,
      "loss": 0.0032,
      "step": 151040
    },
    {
      "epoch": 0.9687877806389958,
      "grad_norm": 0.01328402291983366,
      "learning_rate": 6.1350588056947325e-06,
      "loss": 0.0018,
      "step": 151050
    },
    {
      "epoch": 0.968851917532782,
      "grad_norm": 0.034130360931158066,
      "learning_rate": 6.134513707936031e-06,
      "loss": 0.0015,
      "step": 151060
    },
    {
      "epoch": 0.968916054426568,
      "grad_norm": 0.05559052526950836,
      "learning_rate": 6.13396859596106e-06,
      "loss": 0.002,
      "step": 151070
    },
    {
      "epoch": 0.9689801913203542,
      "grad_norm": 0.09205931425094604,
      "learning_rate": 6.133423469776654e-06,
      "loss": 0.0018,
      "step": 151080
    },
    {
      "epoch": 0.9690443282141402,
      "grad_norm": 0.09927406907081604,
      "learning_rate": 6.13287832938964e-06,
      "loss": 0.002,
      "step": 151090
    },
    {
      "epoch": 0.9691084651079264,
      "grad_norm": 0.08456621319055557,
      "learning_rate": 6.132333174806851e-06,
      "loss": 0.0021,
      "step": 151100
    },
    {
      "epoch": 0.9691726020017124,
      "grad_norm": 0.0676923468708992,
      "learning_rate": 6.131788006035119e-06,
      "loss": 0.002,
      "step": 151110
    },
    {
      "epoch": 0.9692367388954986,
      "grad_norm": 0.01282727625221014,
      "learning_rate": 6.131242823081275e-06,
      "loss": 0.0019,
      "step": 151120
    },
    {
      "epoch": 0.9693008757892847,
      "grad_norm": 0.10641659796237946,
      "learning_rate": 6.130697625952149e-06,
      "loss": 0.0022,
      "step": 151130
    },
    {
      "epoch": 0.9693650126830707,
      "grad_norm": 0.10187271982431412,
      "learning_rate": 6.130152414654574e-06,
      "loss": 0.0039,
      "step": 151140
    },
    {
      "epoch": 0.9694291495768569,
      "grad_norm": 0.05137203633785248,
      "learning_rate": 6.129607189195381e-06,
      "loss": 0.0015,
      "step": 151150
    },
    {
      "epoch": 0.9694932864706429,
      "grad_norm": 0.05988229066133499,
      "learning_rate": 6.129061949581403e-06,
      "loss": 0.0014,
      "step": 151160
    },
    {
      "epoch": 0.9695574233644291,
      "grad_norm": 0.015055349096655846,
      "learning_rate": 6.128516695819472e-06,
      "loss": 0.0032,
      "step": 151170
    },
    {
      "epoch": 0.9696215602582151,
      "grad_norm": 0.1137191578745842,
      "learning_rate": 6.12797142791642e-06,
      "loss": 0.002,
      "step": 151180
    },
    {
      "epoch": 0.9696856971520013,
      "grad_norm": 0.06393451988697052,
      "learning_rate": 6.12742614587908e-06,
      "loss": 0.0022,
      "step": 151190
    },
    {
      "epoch": 0.9697498340457873,
      "grad_norm": 0.03888490051031113,
      "learning_rate": 6.126880849714284e-06,
      "loss": 0.0014,
      "step": 151200
    },
    {
      "epoch": 0.9698139709395734,
      "grad_norm": 0.03833886235952377,
      "learning_rate": 6.126335539428867e-06,
      "loss": 0.0027,
      "step": 151210
    },
    {
      "epoch": 0.9698781078333595,
      "grad_norm": 0.06993034482002258,
      "learning_rate": 6.1257902150296585e-06,
      "loss": 0.0039,
      "step": 151220
    },
    {
      "epoch": 0.9699422447271456,
      "grad_norm": 0.04533940553665161,
      "learning_rate": 6.125244876523496e-06,
      "loss": 0.0015,
      "step": 151230
    },
    {
      "epoch": 0.9700063816209317,
      "grad_norm": 0.05946129187941551,
      "learning_rate": 6.1246995239172105e-06,
      "loss": 0.0013,
      "step": 151240
    },
    {
      "epoch": 0.9700705185147178,
      "grad_norm": 0.0785088837146759,
      "learning_rate": 6.124154157217637e-06,
      "loss": 0.001,
      "step": 151250
    },
    {
      "epoch": 0.970134655408504,
      "grad_norm": 0.09543221443891525,
      "learning_rate": 6.123608776431606e-06,
      "loss": 0.0032,
      "step": 151260
    },
    {
      "epoch": 0.97019879230229,
      "grad_norm": 0.03559909388422966,
      "learning_rate": 6.123063381565957e-06,
      "loss": 0.0023,
      "step": 151270
    },
    {
      "epoch": 0.9702629291960762,
      "grad_norm": 0.10257601737976074,
      "learning_rate": 6.122517972627518e-06,
      "loss": 0.002,
      "step": 151280
    },
    {
      "epoch": 0.9703270660898622,
      "grad_norm": 0.2212112545967102,
      "learning_rate": 6.121972549623129e-06,
      "loss": 0.0034,
      "step": 151290
    },
    {
      "epoch": 0.9703912029836483,
      "grad_norm": 0.04329894483089447,
      "learning_rate": 6.121427112559622e-06,
      "loss": 0.0024,
      "step": 151300
    },
    {
      "epoch": 0.9704553398774344,
      "grad_norm": 0.03108963370323181,
      "learning_rate": 6.120881661443831e-06,
      "loss": 0.0026,
      "step": 151310
    },
    {
      "epoch": 0.9705194767712205,
      "grad_norm": 0.006353151053190231,
      "learning_rate": 6.1203361962825915e-06,
      "loss": 0.0014,
      "step": 151320
    },
    {
      "epoch": 0.9705836136650066,
      "grad_norm": 0.1244301050901413,
      "learning_rate": 6.1197907170827385e-06,
      "loss": 0.002,
      "step": 151330
    },
    {
      "epoch": 0.9706477505587927,
      "grad_norm": 0.14038756489753723,
      "learning_rate": 6.119245223851109e-06,
      "loss": 0.0008,
      "step": 151340
    },
    {
      "epoch": 0.9707118874525787,
      "grad_norm": 0.1935957670211792,
      "learning_rate": 6.1186997165945364e-06,
      "loss": 0.0014,
      "step": 151350
    },
    {
      "epoch": 0.9707760243463649,
      "grad_norm": 0.03064032457768917,
      "learning_rate": 6.118154195319857e-06,
      "loss": 0.0023,
      "step": 151360
    },
    {
      "epoch": 0.9708401612401509,
      "grad_norm": 0.20530641078948975,
      "learning_rate": 6.117608660033904e-06,
      "loss": 0.0022,
      "step": 151370
    },
    {
      "epoch": 0.9709042981339371,
      "grad_norm": 0.009169038385152817,
      "learning_rate": 6.117063110743518e-06,
      "loss": 0.0023,
      "step": 151380
    },
    {
      "epoch": 0.9709684350277231,
      "grad_norm": 0.02863924391567707,
      "learning_rate": 6.116517547455533e-06,
      "loss": 0.0047,
      "step": 151390
    },
    {
      "epoch": 0.9710325719215093,
      "grad_norm": 0.0908161923289299,
      "learning_rate": 6.115971970176785e-06,
      "loss": 0.002,
      "step": 151400
    },
    {
      "epoch": 0.9710967088152954,
      "grad_norm": 0.39482712745666504,
      "learning_rate": 6.11542637891411e-06,
      "loss": 0.0033,
      "step": 151410
    },
    {
      "epoch": 0.9711608457090815,
      "grad_norm": 0.0724111795425415,
      "learning_rate": 6.114880773674347e-06,
      "loss": 0.0013,
      "step": 151420
    },
    {
      "epoch": 0.9712249826028676,
      "grad_norm": 0.014322632923722267,
      "learning_rate": 6.1143351544643285e-06,
      "loss": 0.0028,
      "step": 151430
    },
    {
      "epoch": 0.9712891194966536,
      "grad_norm": 0.050286244601011276,
      "learning_rate": 6.113789521290896e-06,
      "loss": 0.004,
      "step": 151440
    },
    {
      "epoch": 0.9713532563904398,
      "grad_norm": 0.08359697461128235,
      "learning_rate": 6.113243874160884e-06,
      "loss": 0.0031,
      "step": 151450
    },
    {
      "epoch": 0.9714173932842258,
      "grad_norm": 0.19185268878936768,
      "learning_rate": 6.112698213081131e-06,
      "loss": 0.0023,
      "step": 151460
    },
    {
      "epoch": 0.971481530178012,
      "grad_norm": 0.10571939498186111,
      "learning_rate": 6.112152538058473e-06,
      "loss": 0.001,
      "step": 151470
    },
    {
      "epoch": 0.971545667071798,
      "grad_norm": 0.09915226697921753,
      "learning_rate": 6.111606849099749e-06,
      "loss": 0.0028,
      "step": 151480
    },
    {
      "epoch": 0.9716098039655842,
      "grad_norm": 0.0881165862083435,
      "learning_rate": 6.111061146211797e-06,
      "loss": 0.002,
      "step": 151490
    },
    {
      "epoch": 0.9716739408593702,
      "grad_norm": 0.21176064014434814,
      "learning_rate": 6.1105154294014565e-06,
      "loss": 0.0029,
      "step": 151500
    },
    {
      "epoch": 0.9717380777531563,
      "grad_norm": 0.09044768661260605,
      "learning_rate": 6.109969698675562e-06,
      "loss": 0.0034,
      "step": 151510
    },
    {
      "epoch": 0.9718022146469424,
      "grad_norm": 0.13650836050510406,
      "learning_rate": 6.109423954040956e-06,
      "loss": 0.0032,
      "step": 151520
    },
    {
      "epoch": 0.9718663515407285,
      "grad_norm": 0.08207231014966965,
      "learning_rate": 6.108878195504473e-06,
      "loss": 0.0029,
      "step": 151530
    },
    {
      "epoch": 0.9719304884345147,
      "grad_norm": 0.12176788598299026,
      "learning_rate": 6.108332423072953e-06,
      "loss": 0.0029,
      "step": 151540
    },
    {
      "epoch": 0.9719946253283007,
      "grad_norm": 0.09918273985385895,
      "learning_rate": 6.107786636753238e-06,
      "loss": 0.0026,
      "step": 151550
    },
    {
      "epoch": 0.9720587622220869,
      "grad_norm": 0.01861521415412426,
      "learning_rate": 6.107240836552163e-06,
      "loss": 0.0018,
      "step": 151560
    },
    {
      "epoch": 0.9721228991158729,
      "grad_norm": 0.08306858688592911,
      "learning_rate": 6.106695022476571e-06,
      "loss": 0.0017,
      "step": 151570
    },
    {
      "epoch": 0.9721870360096591,
      "grad_norm": 0.16298122704029083,
      "learning_rate": 6.106149194533297e-06,
      "loss": 0.0023,
      "step": 151580
    },
    {
      "epoch": 0.9722511729034451,
      "grad_norm": 0.08419349044561386,
      "learning_rate": 6.105603352729184e-06,
      "loss": 0.0022,
      "step": 151590
    },
    {
      "epoch": 0.9723153097972312,
      "grad_norm": 0.0572366900742054,
      "learning_rate": 6.105057497071071e-06,
      "loss": 0.0024,
      "step": 151600
    },
    {
      "epoch": 0.9723794466910173,
      "grad_norm": 0.049051739275455475,
      "learning_rate": 6.104511627565799e-06,
      "loss": 0.0021,
      "step": 151610
    },
    {
      "epoch": 0.9724435835848034,
      "grad_norm": 0.15118838846683502,
      "learning_rate": 6.103965744220205e-06,
      "loss": 0.002,
      "step": 151620
    },
    {
      "epoch": 0.9725077204785895,
      "grad_norm": 0.04342101886868477,
      "learning_rate": 6.103419847041132e-06,
      "loss": 0.0017,
      "step": 151630
    },
    {
      "epoch": 0.9725718573723756,
      "grad_norm": 0.04763609543442726,
      "learning_rate": 6.1028739360354195e-06,
      "loss": 0.0019,
      "step": 151640
    },
    {
      "epoch": 0.9726359942661617,
      "grad_norm": 0.026795970275998116,
      "learning_rate": 6.102328011209909e-06,
      "loss": 0.0013,
      "step": 151650
    },
    {
      "epoch": 0.9727001311599478,
      "grad_norm": 0.04624636471271515,
      "learning_rate": 6.10178207257144e-06,
      "loss": 0.0029,
      "step": 151660
    },
    {
      "epoch": 0.9727642680537338,
      "grad_norm": 0.08322446048259735,
      "learning_rate": 6.101236120126853e-06,
      "loss": 0.003,
      "step": 151670
    },
    {
      "epoch": 0.97282840494752,
      "grad_norm": 0.07855825871229172,
      "learning_rate": 6.100690153882993e-06,
      "loss": 0.0017,
      "step": 151680
    },
    {
      "epoch": 0.9728925418413061,
      "grad_norm": 0.09713820368051529,
      "learning_rate": 6.100144173846697e-06,
      "loss": 0.0015,
      "step": 151690
    },
    {
      "epoch": 0.9729566787350922,
      "grad_norm": 0.14672186970710754,
      "learning_rate": 6.099598180024809e-06,
      "loss": 0.002,
      "step": 151700
    },
    {
      "epoch": 0.9730208156288783,
      "grad_norm": 0.016113052144646645,
      "learning_rate": 6.099052172424169e-06,
      "loss": 0.0012,
      "step": 151710
    },
    {
      "epoch": 0.9730849525226644,
      "grad_norm": 0.19978825747966766,
      "learning_rate": 6.098506151051621e-06,
      "loss": 0.0045,
      "step": 151720
    },
    {
      "epoch": 0.9731490894164505,
      "grad_norm": 0.12873505055904388,
      "learning_rate": 6.0979601159140055e-06,
      "loss": 0.0013,
      "step": 151730
    },
    {
      "epoch": 0.9732132263102365,
      "grad_norm": 0.30069053173065186,
      "learning_rate": 6.097414067018164e-06,
      "loss": 0.0034,
      "step": 151740
    },
    {
      "epoch": 0.9732773632040227,
      "grad_norm": 0.05176600441336632,
      "learning_rate": 6.0968680043709416e-06,
      "loss": 0.0036,
      "step": 151750
    },
    {
      "epoch": 0.9733415000978087,
      "grad_norm": 0.06669841706752777,
      "learning_rate": 6.0963219279791786e-06,
      "loss": 0.0028,
      "step": 151760
    },
    {
      "epoch": 0.9734056369915949,
      "grad_norm": 0.21331311762332916,
      "learning_rate": 6.095775837849718e-06,
      "loss": 0.0019,
      "step": 151770
    },
    {
      "epoch": 0.9734697738853809,
      "grad_norm": 0.07572337985038757,
      "learning_rate": 6.095229733989403e-06,
      "loss": 0.0024,
      "step": 151780
    },
    {
      "epoch": 0.9735339107791671,
      "grad_norm": 0.13380320370197296,
      "learning_rate": 6.094683616405076e-06,
      "loss": 0.0019,
      "step": 151790
    },
    {
      "epoch": 0.9735980476729531,
      "grad_norm": 0.018392594531178474,
      "learning_rate": 6.094137485103583e-06,
      "loss": 0.0013,
      "step": 151800
    },
    {
      "epoch": 0.9736621845667393,
      "grad_norm": 0.23304355144500732,
      "learning_rate": 6.093591340091763e-06,
      "loss": 0.0016,
      "step": 151810
    },
    {
      "epoch": 0.9737263214605253,
      "grad_norm": 0.09823311865329742,
      "learning_rate": 6.093045181376462e-06,
      "loss": 0.0022,
      "step": 151820
    },
    {
      "epoch": 0.9737904583543114,
      "grad_norm": 0.025532811880111694,
      "learning_rate": 6.092499008964526e-06,
      "loss": 0.001,
      "step": 151830
    },
    {
      "epoch": 0.9738545952480976,
      "grad_norm": 0.0746760368347168,
      "learning_rate": 6.091952822862794e-06,
      "loss": 0.0023,
      "step": 151840
    },
    {
      "epoch": 0.9739187321418836,
      "grad_norm": 0.020240366458892822,
      "learning_rate": 6.091406623078116e-06,
      "loss": 0.0073,
      "step": 151850
    },
    {
      "epoch": 0.9739828690356698,
      "grad_norm": 0.040793247520923615,
      "learning_rate": 6.09086040961733e-06,
      "loss": 0.0018,
      "step": 151860
    },
    {
      "epoch": 0.9740470059294558,
      "grad_norm": 0.13199807703495026,
      "learning_rate": 6.090314182487284e-06,
      "loss": 0.0052,
      "step": 151870
    },
    {
      "epoch": 0.974111142823242,
      "grad_norm": 0.16162413358688354,
      "learning_rate": 6.089767941694822e-06,
      "loss": 0.002,
      "step": 151880
    },
    {
      "epoch": 0.974175279717028,
      "grad_norm": 0.06183703988790512,
      "learning_rate": 6.0892216872467895e-06,
      "loss": 0.0016,
      "step": 151890
    },
    {
      "epoch": 0.9742394166108141,
      "grad_norm": 0.07440250366926193,
      "learning_rate": 6.0886754191500296e-06,
      "loss": 0.0018,
      "step": 151900
    },
    {
      "epoch": 0.9743035535046002,
      "grad_norm": 0.041807644069194794,
      "learning_rate": 6.088129137411391e-06,
      "loss": 0.0038,
      "step": 151910
    },
    {
      "epoch": 0.9743676903983863,
      "grad_norm": 0.07359262555837631,
      "learning_rate": 6.087582842037715e-06,
      "loss": 0.0033,
      "step": 151920
    },
    {
      "epoch": 0.9744318272921724,
      "grad_norm": 0.09813549369573593,
      "learning_rate": 6.08703653303585e-06,
      "loss": 0.0017,
      "step": 151930
    },
    {
      "epoch": 0.9744959641859585,
      "grad_norm": 0.169961079955101,
      "learning_rate": 6.086490210412638e-06,
      "loss": 0.0019,
      "step": 151940
    },
    {
      "epoch": 0.9745601010797446,
      "grad_norm": 0.24065662920475006,
      "learning_rate": 6.0859438741749286e-06,
      "loss": 0.0052,
      "step": 151950
    },
    {
      "epoch": 0.9746242379735307,
      "grad_norm": 0.08111688494682312,
      "learning_rate": 6.085397524329566e-06,
      "loss": 0.001,
      "step": 151960
    },
    {
      "epoch": 0.9746883748673169,
      "grad_norm": 0.08499859273433685,
      "learning_rate": 6.0848511608833984e-06,
      "loss": 0.0035,
      "step": 151970
    },
    {
      "epoch": 0.9747525117611029,
      "grad_norm": 0.07363910973072052,
      "learning_rate": 6.0843047838432675e-06,
      "loss": 0.0029,
      "step": 151980
    },
    {
      "epoch": 0.974816648654889,
      "grad_norm": 0.05126811936497688,
      "learning_rate": 6.083758393216025e-06,
      "loss": 0.0018,
      "step": 151990
    },
    {
      "epoch": 0.9748807855486751,
      "grad_norm": 0.11984333395957947,
      "learning_rate": 6.083211989008514e-06,
      "loss": 0.0017,
      "step": 152000
    },
    {
      "epoch": 0.9749449224424612,
      "grad_norm": 0.1173730194568634,
      "learning_rate": 6.082665571227584e-06,
      "loss": 0.0025,
      "step": 152010
    },
    {
      "epoch": 0.9750090593362473,
      "grad_norm": 0.1031876653432846,
      "learning_rate": 6.08211913988008e-06,
      "loss": 0.003,
      "step": 152020
    },
    {
      "epoch": 0.9750731962300334,
      "grad_norm": 0.11623068153858185,
      "learning_rate": 6.0815726949728484e-06,
      "loss": 0.0016,
      "step": 152030
    },
    {
      "epoch": 0.9751373331238194,
      "grad_norm": 0.07475735247135162,
      "learning_rate": 6.081026236512739e-06,
      "loss": 0.0026,
      "step": 152040
    },
    {
      "epoch": 0.9752014700176056,
      "grad_norm": 0.022157786414027214,
      "learning_rate": 6.080479764506598e-06,
      "loss": 0.0022,
      "step": 152050
    },
    {
      "epoch": 0.9752656069113916,
      "grad_norm": 0.09656548500061035,
      "learning_rate": 6.079933278961274e-06,
      "loss": 0.0038,
      "step": 152060
    },
    {
      "epoch": 0.9753297438051778,
      "grad_norm": 0.12655095756053925,
      "learning_rate": 6.079386779883614e-06,
      "loss": 0.0019,
      "step": 152070
    },
    {
      "epoch": 0.9753938806989638,
      "grad_norm": 0.019430868327617645,
      "learning_rate": 6.078840267280467e-06,
      "loss": 0.0012,
      "step": 152080
    },
    {
      "epoch": 0.97545801759275,
      "grad_norm": 0.058887068182229996,
      "learning_rate": 6.078293741158678e-06,
      "loss": 0.0017,
      "step": 152090
    },
    {
      "epoch": 0.975522154486536,
      "grad_norm": 0.07945910841226578,
      "learning_rate": 6.0777472015251e-06,
      "loss": 0.002,
      "step": 152100
    },
    {
      "epoch": 0.9755862913803222,
      "grad_norm": 0.06385931372642517,
      "learning_rate": 6.077200648386579e-06,
      "loss": 0.0034,
      "step": 152110
    },
    {
      "epoch": 0.9756504282741083,
      "grad_norm": 0.09562142193317413,
      "learning_rate": 6.076654081749964e-06,
      "loss": 0.0032,
      "step": 152120
    },
    {
      "epoch": 0.9757145651678943,
      "grad_norm": 0.08291531354188919,
      "learning_rate": 6.076107501622103e-06,
      "loss": 0.0029,
      "step": 152130
    },
    {
      "epoch": 0.9757787020616805,
      "grad_norm": 0.02452937886118889,
      "learning_rate": 6.075560908009847e-06,
      "loss": 0.0014,
      "step": 152140
    },
    {
      "epoch": 0.9758428389554665,
      "grad_norm": 0.17652659118175507,
      "learning_rate": 6.075014300920042e-06,
      "loss": 0.0071,
      "step": 152150
    },
    {
      "epoch": 0.9759069758492527,
      "grad_norm": 0.07808490842580795,
      "learning_rate": 6.074467680359542e-06,
      "loss": 0.0016,
      "step": 152160
    },
    {
      "epoch": 0.9759711127430387,
      "grad_norm": 0.07041225582361221,
      "learning_rate": 6.073921046335193e-06,
      "loss": 0.0016,
      "step": 152170
    },
    {
      "epoch": 0.9760352496368249,
      "grad_norm": 0.15718582272529602,
      "learning_rate": 6.0733743988538455e-06,
      "loss": 0.0021,
      "step": 152180
    },
    {
      "epoch": 0.9760993865306109,
      "grad_norm": 0.00888227578252554,
      "learning_rate": 6.072827737922351e-06,
      "loss": 0.0014,
      "step": 152190
    },
    {
      "epoch": 0.976163523424397,
      "grad_norm": 0.23891635239124298,
      "learning_rate": 6.072281063547556e-06,
      "loss": 0.0017,
      "step": 152200
    },
    {
      "epoch": 0.9762276603181831,
      "grad_norm": 0.10157041996717453,
      "learning_rate": 6.071734375736314e-06,
      "loss": 0.0033,
      "step": 152210
    },
    {
      "epoch": 0.9762917972119692,
      "grad_norm": 0.16198943555355072,
      "learning_rate": 6.071187674495475e-06,
      "loss": 0.0033,
      "step": 152220
    },
    {
      "epoch": 0.9763559341057553,
      "grad_norm": 0.14039942622184753,
      "learning_rate": 6.070640959831888e-06,
      "loss": 0.0018,
      "step": 152230
    },
    {
      "epoch": 0.9764200709995414,
      "grad_norm": 0.0957934558391571,
      "learning_rate": 6.0700942317524054e-06,
      "loss": 0.0022,
      "step": 152240
    },
    {
      "epoch": 0.9764842078933276,
      "grad_norm": 0.1538853943347931,
      "learning_rate": 6.069547490263876e-06,
      "loss": 0.0024,
      "step": 152250
    },
    {
      "epoch": 0.9765483447871136,
      "grad_norm": 0.4327002465724945,
      "learning_rate": 6.069000735373153e-06,
      "loss": 0.0025,
      "step": 152260
    },
    {
      "epoch": 0.9766124816808998,
      "grad_norm": 0.07163172215223312,
      "learning_rate": 6.068453967087087e-06,
      "loss": 0.0017,
      "step": 152270
    },
    {
      "epoch": 0.9766766185746858,
      "grad_norm": 0.15471971035003662,
      "learning_rate": 6.067907185412528e-06,
      "loss": 0.0022,
      "step": 152280
    },
    {
      "epoch": 0.9767407554684719,
      "grad_norm": 0.04025840014219284,
      "learning_rate": 6.067360390356331e-06,
      "loss": 0.0018,
      "step": 152290
    },
    {
      "epoch": 0.976804892362258,
      "grad_norm": 0.03471701219677925,
      "learning_rate": 6.066813581925344e-06,
      "loss": 0.0015,
      "step": 152300
    },
    {
      "epoch": 0.9768690292560441,
      "grad_norm": 0.08985839784145355,
      "learning_rate": 6.06626676012642e-06,
      "loss": 0.0036,
      "step": 152310
    },
    {
      "epoch": 0.9769331661498302,
      "grad_norm": 0.0038460553623735905,
      "learning_rate": 6.065719924966412e-06,
      "loss": 0.0013,
      "step": 152320
    },
    {
      "epoch": 0.9769973030436163,
      "grad_norm": 0.14992263913154602,
      "learning_rate": 6.065173076452172e-06,
      "loss": 0.0017,
      "step": 152330
    },
    {
      "epoch": 0.9770614399374024,
      "grad_norm": 0.016167109832167625,
      "learning_rate": 6.064626214590552e-06,
      "loss": 0.0014,
      "step": 152340
    },
    {
      "epoch": 0.9771255768311885,
      "grad_norm": 0.14940617978572845,
      "learning_rate": 6.064079339388404e-06,
      "loss": 0.0033,
      "step": 152350
    },
    {
      "epoch": 0.9771897137249745,
      "grad_norm": 0.07109025120735168,
      "learning_rate": 6.063532450852582e-06,
      "loss": 0.0049,
      "step": 152360
    },
    {
      "epoch": 0.9772538506187607,
      "grad_norm": 0.11286960542201996,
      "learning_rate": 6.062985548989939e-06,
      "loss": 0.0061,
      "step": 152370
    },
    {
      "epoch": 0.9773179875125467,
      "grad_norm": 0.20279167592525482,
      "learning_rate": 6.062438633807326e-06,
      "loss": 0.0017,
      "step": 152380
    },
    {
      "epoch": 0.9773821244063329,
      "grad_norm": 0.019253971055150032,
      "learning_rate": 6.061891705311597e-06,
      "loss": 0.0019,
      "step": 152390
    },
    {
      "epoch": 0.977446261300119,
      "grad_norm": 0.12412141263484955,
      "learning_rate": 6.061344763509608e-06,
      "loss": 0.0021,
      "step": 152400
    },
    {
      "epoch": 0.9775103981939051,
      "grad_norm": 0.058596231043338776,
      "learning_rate": 6.060797808408209e-06,
      "loss": 0.0022,
      "step": 152410
    },
    {
      "epoch": 0.9775745350876912,
      "grad_norm": 0.027793683111667633,
      "learning_rate": 6.060250840014257e-06,
      "loss": 0.0021,
      "step": 152420
    },
    {
      "epoch": 0.9776386719814772,
      "grad_norm": 0.07490064948797226,
      "learning_rate": 6.0597038583346026e-06,
      "loss": 0.0022,
      "step": 152430
    },
    {
      "epoch": 0.9777028088752634,
      "grad_norm": 0.029315203428268433,
      "learning_rate": 6.059156863376102e-06,
      "loss": 0.0023,
      "step": 152440
    },
    {
      "epoch": 0.9777669457690494,
      "grad_norm": 0.14921054244041443,
      "learning_rate": 6.058609855145608e-06,
      "loss": 0.0027,
      "step": 152450
    },
    {
      "epoch": 0.9778310826628356,
      "grad_norm": 0.06638119369745255,
      "learning_rate": 6.058062833649978e-06,
      "loss": 0.0035,
      "step": 152460
    },
    {
      "epoch": 0.9778952195566216,
      "grad_norm": 0.09822449833154678,
      "learning_rate": 6.057515798896063e-06,
      "loss": 0.0024,
      "step": 152470
    },
    {
      "epoch": 0.9779593564504078,
      "grad_norm": 0.11892236769199371,
      "learning_rate": 6.0569687508907205e-06,
      "loss": 0.0037,
      "step": 152480
    },
    {
      "epoch": 0.9780234933441938,
      "grad_norm": 0.03293519467115402,
      "learning_rate": 6.056421689640804e-06,
      "loss": 0.0018,
      "step": 152490
    },
    {
      "epoch": 0.97808763023798,
      "grad_norm": 0.33398517966270447,
      "learning_rate": 6.055874615153168e-06,
      "loss": 0.0045,
      "step": 152500
    },
    {
      "epoch": 0.978151767131766,
      "grad_norm": 0.22353917360305786,
      "learning_rate": 6.0553275274346675e-06,
      "loss": 0.0017,
      "step": 152510
    },
    {
      "epoch": 0.9782159040255521,
      "grad_norm": 0.0990443080663681,
      "learning_rate": 6.054780426492161e-06,
      "loss": 0.0057,
      "step": 152520
    },
    {
      "epoch": 0.9782800409193383,
      "grad_norm": 0.10360170900821686,
      "learning_rate": 6.0542333123325e-06,
      "loss": 0.0021,
      "step": 152530
    },
    {
      "epoch": 0.9783441778131243,
      "grad_norm": 0.089264877140522,
      "learning_rate": 6.053686184962543e-06,
      "loss": 0.0016,
      "step": 152540
    },
    {
      "epoch": 0.9784083147069105,
      "grad_norm": 0.20201466977596283,
      "learning_rate": 6.053139044389146e-06,
      "loss": 0.0046,
      "step": 152550
    },
    {
      "epoch": 0.9784724516006965,
      "grad_norm": 0.06233484297990799,
      "learning_rate": 6.052591890619162e-06,
      "loss": 0.0019,
      "step": 152560
    },
    {
      "epoch": 0.9785365884944827,
      "grad_norm": 0.12971296906471252,
      "learning_rate": 6.052044723659451e-06,
      "loss": 0.002,
      "step": 152570
    },
    {
      "epoch": 0.9786007253882687,
      "grad_norm": 0.033529192209243774,
      "learning_rate": 6.051497543516867e-06,
      "loss": 0.0022,
      "step": 152580
    },
    {
      "epoch": 0.9786648622820548,
      "grad_norm": 0.34596654772758484,
      "learning_rate": 6.050950350198268e-06,
      "loss": 0.0042,
      "step": 152590
    },
    {
      "epoch": 0.9787289991758409,
      "grad_norm": 0.16789695620536804,
      "learning_rate": 6.050403143710508e-06,
      "loss": 0.0017,
      "step": 152600
    },
    {
      "epoch": 0.978793136069627,
      "grad_norm": 0.00438518263399601,
      "learning_rate": 6.049855924060449e-06,
      "loss": 0.0015,
      "step": 152610
    },
    {
      "epoch": 0.9788572729634131,
      "grad_norm": 0.16074934601783752,
      "learning_rate": 6.049308691254943e-06,
      "loss": 0.0028,
      "step": 152620
    },
    {
      "epoch": 0.9789214098571992,
      "grad_norm": 0.13383474946022034,
      "learning_rate": 6.048761445300848e-06,
      "loss": 0.0025,
      "step": 152630
    },
    {
      "epoch": 0.9789855467509853,
      "grad_norm": 0.10629253834486008,
      "learning_rate": 6.048214186205024e-06,
      "loss": 0.0029,
      "step": 152640
    },
    {
      "epoch": 0.9790496836447714,
      "grad_norm": 0.026463015004992485,
      "learning_rate": 6.047666913974328e-06,
      "loss": 0.0007,
      "step": 152650
    },
    {
      "epoch": 0.9791138205385574,
      "grad_norm": 0.09996064752340317,
      "learning_rate": 6.047119628615616e-06,
      "loss": 0.0016,
      "step": 152660
    },
    {
      "epoch": 0.9791779574323436,
      "grad_norm": 0.09978724271059036,
      "learning_rate": 6.0465723301357456e-06,
      "loss": 0.0022,
      "step": 152670
    },
    {
      "epoch": 0.9792420943261297,
      "grad_norm": 0.08587059378623962,
      "learning_rate": 6.046025018541577e-06,
      "loss": 0.0017,
      "step": 152680
    },
    {
      "epoch": 0.9793062312199158,
      "grad_norm": 0.10156667977571487,
      "learning_rate": 6.045477693839968e-06,
      "loss": 0.0016,
      "step": 152690
    },
    {
      "epoch": 0.9793703681137019,
      "grad_norm": 0.05822337046265602,
      "learning_rate": 6.044930356037775e-06,
      "loss": 0.0061,
      "step": 152700
    },
    {
      "epoch": 0.979434505007488,
      "grad_norm": 0.18206116557121277,
      "learning_rate": 6.0443830051418596e-06,
      "loss": 0.0012,
      "step": 152710
    },
    {
      "epoch": 0.9794986419012741,
      "grad_norm": 0.11949578672647476,
      "learning_rate": 6.043835641159077e-06,
      "loss": 0.0024,
      "step": 152720
    },
    {
      "epoch": 0.9795627787950602,
      "grad_norm": 0.0030041232239454985,
      "learning_rate": 6.043288264096287e-06,
      "loss": 0.0012,
      "step": 152730
    },
    {
      "epoch": 0.9796269156888463,
      "grad_norm": 0.0638008862733841,
      "learning_rate": 6.0427408739603525e-06,
      "loss": 0.0023,
      "step": 152740
    },
    {
      "epoch": 0.9796910525826323,
      "grad_norm": 0.49946504831314087,
      "learning_rate": 6.0421934707581264e-06,
      "loss": 0.0021,
      "step": 152750
    },
    {
      "epoch": 0.9797551894764185,
      "grad_norm": 0.192970871925354,
      "learning_rate": 6.041646054496474e-06,
      "loss": 0.0018,
      "step": 152760
    },
    {
      "epoch": 0.9798193263702045,
      "grad_norm": 0.10488071292638779,
      "learning_rate": 6.04109862518225e-06,
      "loss": 0.0012,
      "step": 152770
    },
    {
      "epoch": 0.9798834632639907,
      "grad_norm": 0.14728890359401703,
      "learning_rate": 6.040551182822317e-06,
      "loss": 0.0015,
      "step": 152780
    },
    {
      "epoch": 0.9799476001577767,
      "grad_norm": 0.09117814153432846,
      "learning_rate": 6.0400037274235345e-06,
      "loss": 0.0014,
      "step": 152790
    },
    {
      "epoch": 0.9800117370515629,
      "grad_norm": 0.07769643515348434,
      "learning_rate": 6.039456258992762e-06,
      "loss": 0.0015,
      "step": 152800
    },
    {
      "epoch": 0.980075873945349,
      "grad_norm": 0.14587846398353577,
      "learning_rate": 6.038908777536858e-06,
      "loss": 0.0017,
      "step": 152810
    },
    {
      "epoch": 0.980140010839135,
      "grad_norm": 0.08112621307373047,
      "learning_rate": 6.038361283062687e-06,
      "loss": 0.0022,
      "step": 152820
    },
    {
      "epoch": 0.9802041477329212,
      "grad_norm": 0.09943347424268723,
      "learning_rate": 6.037813775577105e-06,
      "loss": 0.0027,
      "step": 152830
    },
    {
      "epoch": 0.9802682846267072,
      "grad_norm": 0.08695618063211441,
      "learning_rate": 6.037266255086977e-06,
      "loss": 0.0013,
      "step": 152840
    },
    {
      "epoch": 0.9803324215204934,
      "grad_norm": 0.08281394094228745,
      "learning_rate": 6.036718721599159e-06,
      "loss": 0.0032,
      "step": 152850
    },
    {
      "epoch": 0.9803965584142794,
      "grad_norm": 0.10669806599617004,
      "learning_rate": 6.036171175120515e-06,
      "loss": 0.0024,
      "step": 152860
    },
    {
      "epoch": 0.9804606953080656,
      "grad_norm": 0.12385310232639313,
      "learning_rate": 6.035623615657906e-06,
      "loss": 0.0012,
      "step": 152870
    },
    {
      "epoch": 0.9805248322018516,
      "grad_norm": 0.031515467911958694,
      "learning_rate": 6.035076043218193e-06,
      "loss": 0.0058,
      "step": 152880
    },
    {
      "epoch": 0.9805889690956378,
      "grad_norm": 0.06842345744371414,
      "learning_rate": 6.034528457808238e-06,
      "loss": 0.0015,
      "step": 152890
    },
    {
      "epoch": 0.9806531059894238,
      "grad_norm": 0.08759213984012604,
      "learning_rate": 6.033980859434902e-06,
      "loss": 0.0019,
      "step": 152900
    },
    {
      "epoch": 0.9807172428832099,
      "grad_norm": 0.19722624123096466,
      "learning_rate": 6.0334332481050465e-06,
      "loss": 0.0035,
      "step": 152910
    },
    {
      "epoch": 0.980781379776996,
      "grad_norm": 0.03208306431770325,
      "learning_rate": 6.032885623825533e-06,
      "loss": 0.0017,
      "step": 152920
    },
    {
      "epoch": 0.9808455166707821,
      "grad_norm": 0.11563814431428909,
      "learning_rate": 6.032337986603225e-06,
      "loss": 0.0031,
      "step": 152930
    },
    {
      "epoch": 0.9809096535645682,
      "grad_norm": 0.030628079548478127,
      "learning_rate": 6.031790336444985e-06,
      "loss": 0.0016,
      "step": 152940
    },
    {
      "epoch": 0.9809737904583543,
      "grad_norm": 0.22559551894664764,
      "learning_rate": 6.031242673357673e-06,
      "loss": 0.0036,
      "step": 152950
    },
    {
      "epoch": 0.9810379273521405,
      "grad_norm": 0.009765544906258583,
      "learning_rate": 6.030694997348154e-06,
      "loss": 0.0009,
      "step": 152960
    },
    {
      "epoch": 0.9811020642459265,
      "grad_norm": 0.026936056092381477,
      "learning_rate": 6.030147308423292e-06,
      "loss": 0.0026,
      "step": 152970
    },
    {
      "epoch": 0.9811662011397126,
      "grad_norm": 0.047007352113723755,
      "learning_rate": 6.029599606589945e-06,
      "loss": 0.0022,
      "step": 152980
    },
    {
      "epoch": 0.9812303380334987,
      "grad_norm": 0.19669876992702484,
      "learning_rate": 6.029051891854981e-06,
      "loss": 0.0035,
      "step": 152990
    },
    {
      "epoch": 0.9812944749272848,
      "grad_norm": 0.004361078608781099,
      "learning_rate": 6.028504164225261e-06,
      "loss": 0.0013,
      "step": 153000
    },
    {
      "epoch": 0.9813586118210709,
      "grad_norm": 0.07227908819913864,
      "learning_rate": 6.027956423707649e-06,
      "loss": 0.0023,
      "step": 153010
    },
    {
      "epoch": 0.981422748714857,
      "grad_norm": 0.30987516045570374,
      "learning_rate": 6.0274086703090065e-06,
      "loss": 0.0041,
      "step": 153020
    },
    {
      "epoch": 0.9814868856086431,
      "grad_norm": 0.1345183104276657,
      "learning_rate": 6.0268609040362e-06,
      "loss": 0.0013,
      "step": 153030
    },
    {
      "epoch": 0.9815510225024292,
      "grad_norm": 0.09673765301704407,
      "learning_rate": 6.026313124896093e-06,
      "loss": 0.0026,
      "step": 153040
    },
    {
      "epoch": 0.9816151593962152,
      "grad_norm": 0.0684819445014,
      "learning_rate": 6.02576533289555e-06,
      "loss": 0.0011,
      "step": 153050
    },
    {
      "epoch": 0.9816792962900014,
      "grad_norm": 0.03442485257983208,
      "learning_rate": 6.025217528041433e-06,
      "loss": 0.0021,
      "step": 153060
    },
    {
      "epoch": 0.9817434331837874,
      "grad_norm": 0.14554892480373383,
      "learning_rate": 6.024669710340609e-06,
      "loss": 0.0021,
      "step": 153070
    },
    {
      "epoch": 0.9818075700775736,
      "grad_norm": 0.04231959581375122,
      "learning_rate": 6.024121879799939e-06,
      "loss": 0.0021,
      "step": 153080
    },
    {
      "epoch": 0.9818717069713596,
      "grad_norm": 0.21806782484054565,
      "learning_rate": 6.023574036426291e-06,
      "loss": 0.0014,
      "step": 153090
    },
    {
      "epoch": 0.9819358438651458,
      "grad_norm": 0.1455833613872528,
      "learning_rate": 6.023026180226529e-06,
      "loss": 0.0024,
      "step": 153100
    },
    {
      "epoch": 0.9819999807589319,
      "grad_norm": 0.07845161110162735,
      "learning_rate": 6.022478311207517e-06,
      "loss": 0.0029,
      "step": 153110
    },
    {
      "epoch": 0.982064117652718,
      "grad_norm": 0.009655271656811237,
      "learning_rate": 6.021930429376122e-06,
      "loss": 0.0036,
      "step": 153120
    },
    {
      "epoch": 0.9821282545465041,
      "grad_norm": 0.08508338779211044,
      "learning_rate": 6.021382534739207e-06,
      "loss": 0.0024,
      "step": 153130
    },
    {
      "epoch": 0.9821923914402901,
      "grad_norm": 0.04608561098575592,
      "learning_rate": 6.020834627303641e-06,
      "loss": 0.0023,
      "step": 153140
    },
    {
      "epoch": 0.9822565283340763,
      "grad_norm": 0.13315847516059875,
      "learning_rate": 6.020286707076286e-06,
      "loss": 0.0019,
      "step": 153150
    },
    {
      "epoch": 0.9823206652278623,
      "grad_norm": 0.024399518966674805,
      "learning_rate": 6.01973877406401e-06,
      "loss": 0.002,
      "step": 153160
    },
    {
      "epoch": 0.9823848021216485,
      "grad_norm": 0.06183422729372978,
      "learning_rate": 6.019190828273678e-06,
      "loss": 0.0022,
      "step": 153170
    },
    {
      "epoch": 0.9824489390154345,
      "grad_norm": 0.18734464049339294,
      "learning_rate": 6.018642869712158e-06,
      "loss": 0.0024,
      "step": 153180
    },
    {
      "epoch": 0.9825130759092207,
      "grad_norm": 0.10502655804157257,
      "learning_rate": 6.018094898386313e-06,
      "loss": 0.0016,
      "step": 153190
    },
    {
      "epoch": 0.9825772128030067,
      "grad_norm": 0.09715384989976883,
      "learning_rate": 6.017546914303012e-06,
      "loss": 0.002,
      "step": 153200
    },
    {
      "epoch": 0.9826413496967928,
      "grad_norm": 0.00909672025591135,
      "learning_rate": 6.01699891746912e-06,
      "loss": 0.0021,
      "step": 153210
    },
    {
      "epoch": 0.9827054865905789,
      "grad_norm": 0.06395453959703445,
      "learning_rate": 6.016450907891507e-06,
      "loss": 0.0016,
      "step": 153220
    },
    {
      "epoch": 0.982769623484365,
      "grad_norm": 0.0025687352754175663,
      "learning_rate": 6.015902885577035e-06,
      "loss": 0.0023,
      "step": 153230
    },
    {
      "epoch": 0.9828337603781512,
      "grad_norm": 0.08915159851312637,
      "learning_rate": 6.015354850532574e-06,
      "loss": 0.0018,
      "step": 153240
    },
    {
      "epoch": 0.9828978972719372,
      "grad_norm": 0.06700330972671509,
      "learning_rate": 6.014806802764993e-06,
      "loss": 0.0011,
      "step": 153250
    },
    {
      "epoch": 0.9829620341657234,
      "grad_norm": 0.06353837996721268,
      "learning_rate": 6.014258742281156e-06,
      "loss": 0.0019,
      "step": 153260
    },
    {
      "epoch": 0.9830261710595094,
      "grad_norm": 0.05698194354772568,
      "learning_rate": 6.013710669087933e-06,
      "loss": 0.0014,
      "step": 153270
    },
    {
      "epoch": 0.9830903079532956,
      "grad_norm": 0.1636301726102829,
      "learning_rate": 6.01316258319219e-06,
      "loss": 0.0021,
      "step": 153280
    },
    {
      "epoch": 0.9831544448470816,
      "grad_norm": 0.04641463980078697,
      "learning_rate": 6.012614484600797e-06,
      "loss": 0.003,
      "step": 153290
    },
    {
      "epoch": 0.9832185817408677,
      "grad_norm": 0.1287623941898346,
      "learning_rate": 6.01206637332062e-06,
      "loss": 0.0021,
      "step": 153300
    },
    {
      "epoch": 0.9832827186346538,
      "grad_norm": 0.1351720243692398,
      "learning_rate": 6.0115182493585274e-06,
      "loss": 0.0029,
      "step": 153310
    },
    {
      "epoch": 0.9833468555284399,
      "grad_norm": 0.20169375836849213,
      "learning_rate": 6.010970112721388e-06,
      "loss": 0.0034,
      "step": 153320
    },
    {
      "epoch": 0.983410992422226,
      "grad_norm": 0.11158497631549835,
      "learning_rate": 6.010421963416073e-06,
      "loss": 0.0022,
      "step": 153330
    },
    {
      "epoch": 0.9834751293160121,
      "grad_norm": 0.05959600582718849,
      "learning_rate": 6.009873801449446e-06,
      "loss": 0.002,
      "step": 153340
    },
    {
      "epoch": 0.9835392662097981,
      "grad_norm": 0.042437419295310974,
      "learning_rate": 6.00932562682838e-06,
      "loss": 0.0026,
      "step": 153350
    },
    {
      "epoch": 0.9836034031035843,
      "grad_norm": 0.23058632016181946,
      "learning_rate": 6.008777439559742e-06,
      "loss": 0.0014,
      "step": 153360
    },
    {
      "epoch": 0.9836675399973703,
      "grad_norm": 0.06583657115697861,
      "learning_rate": 6.008229239650403e-06,
      "loss": 0.0019,
      "step": 153370
    },
    {
      "epoch": 0.9837316768911565,
      "grad_norm": 0.2148008495569229,
      "learning_rate": 6.0076810271072284e-06,
      "loss": 0.0021,
      "step": 153380
    },
    {
      "epoch": 0.9837958137849426,
      "grad_norm": 0.08241555839776993,
      "learning_rate": 6.0071328019370915e-06,
      "loss": 0.003,
      "step": 153390
    },
    {
      "epoch": 0.9838599506787287,
      "grad_norm": 0.1243964210152626,
      "learning_rate": 6.00658456414686e-06,
      "loss": 0.0024,
      "step": 153400
    },
    {
      "epoch": 0.9839240875725148,
      "grad_norm": 0.12397447973489761,
      "learning_rate": 6.006036313743407e-06,
      "loss": 0.0032,
      "step": 153410
    },
    {
      "epoch": 0.9839882244663009,
      "grad_norm": 0.1622055172920227,
      "learning_rate": 6.0054880507335974e-06,
      "loss": 0.0029,
      "step": 153420
    },
    {
      "epoch": 0.984052361360087,
      "grad_norm": 0.10634105652570724,
      "learning_rate": 6.004939775124306e-06,
      "loss": 0.0023,
      "step": 153430
    },
    {
      "epoch": 0.984116498253873,
      "grad_norm": 0.015624837949872017,
      "learning_rate": 6.0043914869224e-06,
      "loss": 0.0015,
      "step": 153440
    },
    {
      "epoch": 0.9841806351476592,
      "grad_norm": 0.08204245567321777,
      "learning_rate": 6.00384318613475e-06,
      "loss": 0.0037,
      "step": 153450
    },
    {
      "epoch": 0.9842447720414452,
      "grad_norm": 0.0344320572912693,
      "learning_rate": 6.003294872768229e-06,
      "loss": 0.0054,
      "step": 153460
    },
    {
      "epoch": 0.9843089089352314,
      "grad_norm": 0.1673690527677536,
      "learning_rate": 6.002746546829705e-06,
      "loss": 0.0033,
      "step": 153470
    },
    {
      "epoch": 0.9843730458290174,
      "grad_norm": 0.09421828389167786,
      "learning_rate": 6.002198208326052e-06,
      "loss": 0.0037,
      "step": 153480
    },
    {
      "epoch": 0.9844371827228036,
      "grad_norm": 0.12709495425224304,
      "learning_rate": 6.001649857264137e-06,
      "loss": 0.002,
      "step": 153490
    },
    {
      "epoch": 0.9845013196165896,
      "grad_norm": 0.22577492892742157,
      "learning_rate": 6.001101493650835e-06,
      "loss": 0.0021,
      "step": 153500
    },
    {
      "epoch": 0.9845654565103757,
      "grad_norm": 0.10200013220310211,
      "learning_rate": 6.0005531174930145e-06,
      "loss": 0.0027,
      "step": 153510
    },
    {
      "epoch": 0.9846295934041619,
      "grad_norm": 0.046134013682603836,
      "learning_rate": 6.00000472879755e-06,
      "loss": 0.0017,
      "step": 153520
    },
    {
      "epoch": 0.9846937302979479,
      "grad_norm": 0.05572868511080742,
      "learning_rate": 5.9994563275713106e-06,
      "loss": 0.0018,
      "step": 153530
    },
    {
      "epoch": 0.9847578671917341,
      "grad_norm": 0.06650326400995255,
      "learning_rate": 5.998907913821169e-06,
      "loss": 0.0018,
      "step": 153540
    },
    {
      "epoch": 0.9848220040855201,
      "grad_norm": 0.0946025401353836,
      "learning_rate": 5.998359487553996e-06,
      "loss": 0.0022,
      "step": 153550
    },
    {
      "epoch": 0.9848861409793063,
      "grad_norm": 0.06810727715492249,
      "learning_rate": 5.997811048776668e-06,
      "loss": 0.0026,
      "step": 153560
    },
    {
      "epoch": 0.9849502778730923,
      "grad_norm": 0.06308284401893616,
      "learning_rate": 5.997262597496052e-06,
      "loss": 0.0014,
      "step": 153570
    },
    {
      "epoch": 0.9850144147668785,
      "grad_norm": 0.13130496442317963,
      "learning_rate": 5.9967141337190234e-06,
      "loss": 0.0021,
      "step": 153580
    },
    {
      "epoch": 0.9850785516606645,
      "grad_norm": 0.15117326378822327,
      "learning_rate": 5.996165657452455e-06,
      "loss": 0.0022,
      "step": 153590
    },
    {
      "epoch": 0.9851426885544506,
      "grad_norm": 0.017763545736670494,
      "learning_rate": 5.99561716870322e-06,
      "loss": 0.0018,
      "step": 153600
    },
    {
      "epoch": 0.9852068254482367,
      "grad_norm": 0.29442140460014343,
      "learning_rate": 5.995068667478189e-06,
      "loss": 0.0035,
      "step": 153610
    },
    {
      "epoch": 0.9852709623420228,
      "grad_norm": 0.08706725388765335,
      "learning_rate": 5.994520153784235e-06,
      "loss": 0.002,
      "step": 153620
    },
    {
      "epoch": 0.9853350992358089,
      "grad_norm": 0.1322675347328186,
      "learning_rate": 5.993971627628235e-06,
      "loss": 0.0022,
      "step": 153630
    },
    {
      "epoch": 0.985399236129595,
      "grad_norm": 0.11531005799770355,
      "learning_rate": 5.9934230890170595e-06,
      "loss": 0.0012,
      "step": 153640
    },
    {
      "epoch": 0.985463373023381,
      "grad_norm": 0.06310612708330154,
      "learning_rate": 5.992874537957583e-06,
      "loss": 0.0024,
      "step": 153650
    },
    {
      "epoch": 0.9855275099171672,
      "grad_norm": 0.10038796067237854,
      "learning_rate": 5.992325974456678e-06,
      "loss": 0.0017,
      "step": 153660
    },
    {
      "epoch": 0.9855916468109533,
      "grad_norm": 0.1436949521303177,
      "learning_rate": 5.99177739852122e-06,
      "loss": 0.0026,
      "step": 153670
    },
    {
      "epoch": 0.9856557837047394,
      "grad_norm": 0.030475756153464317,
      "learning_rate": 5.991228810158081e-06,
      "loss": 0.0033,
      "step": 153680
    },
    {
      "epoch": 0.9857199205985255,
      "grad_norm": 0.24867470562458038,
      "learning_rate": 5.99068020937414e-06,
      "loss": 0.0023,
      "step": 153690
    },
    {
      "epoch": 0.9857840574923116,
      "grad_norm": 0.06362586468458176,
      "learning_rate": 5.990131596176265e-06,
      "loss": 0.002,
      "step": 153700
    },
    {
      "epoch": 0.9858481943860977,
      "grad_norm": 0.2324649840593338,
      "learning_rate": 5.989582970571335e-06,
      "loss": 0.0026,
      "step": 153710
    },
    {
      "epoch": 0.9859123312798838,
      "grad_norm": 0.1523626446723938,
      "learning_rate": 5.989034332566221e-06,
      "loss": 0.0026,
      "step": 153720
    },
    {
      "epoch": 0.9859764681736699,
      "grad_norm": 0.09870940446853638,
      "learning_rate": 5.988485682167801e-06,
      "loss": 0.003,
      "step": 153730
    },
    {
      "epoch": 0.9860406050674559,
      "grad_norm": 0.037186942994594574,
      "learning_rate": 5.98793701938295e-06,
      "loss": 0.001,
      "step": 153740
    },
    {
      "epoch": 0.9861047419612421,
      "grad_norm": 0.06786558777093887,
      "learning_rate": 5.987388344218541e-06,
      "loss": 0.0032,
      "step": 153750
    },
    {
      "epoch": 0.9861688788550281,
      "grad_norm": 0.2750147879123688,
      "learning_rate": 5.986839656681451e-06,
      "loss": 0.0027,
      "step": 153760
    },
    {
      "epoch": 0.9862330157488143,
      "grad_norm": 0.010936897248029709,
      "learning_rate": 5.986290956778554e-06,
      "loss": 0.0016,
      "step": 153770
    },
    {
      "epoch": 0.9862971526426003,
      "grad_norm": 0.07299142330884933,
      "learning_rate": 5.985742244516727e-06,
      "loss": 0.0028,
      "step": 153780
    },
    {
      "epoch": 0.9863612895363865,
      "grad_norm": 0.17001250386238098,
      "learning_rate": 5.9851935199028446e-06,
      "loss": 0.0043,
      "step": 153790
    },
    {
      "epoch": 0.9864254264301726,
      "grad_norm": 0.03887956961989403,
      "learning_rate": 5.984644782943784e-06,
      "loss": 0.0032,
      "step": 153800
    },
    {
      "epoch": 0.9864895633239587,
      "grad_norm": 0.06760495901107788,
      "learning_rate": 5.98409603364642e-06,
      "loss": 0.0018,
      "step": 153810
    },
    {
      "epoch": 0.9865537002177448,
      "grad_norm": 0.21361862123012543,
      "learning_rate": 5.98354727201763e-06,
      "loss": 0.0028,
      "step": 153820
    },
    {
      "epoch": 0.9866178371115308,
      "grad_norm": 0.14941273629665375,
      "learning_rate": 5.982998498064289e-06,
      "loss": 0.005,
      "step": 153830
    },
    {
      "epoch": 0.986681974005317,
      "grad_norm": 0.06045301631093025,
      "learning_rate": 5.982449711793275e-06,
      "loss": 0.0035,
      "step": 153840
    },
    {
      "epoch": 0.986746110899103,
      "grad_norm": 0.15081606805324554,
      "learning_rate": 5.981900913211464e-06,
      "loss": 0.0024,
      "step": 153850
    },
    {
      "epoch": 0.9868102477928892,
      "grad_norm": 0.08579900860786438,
      "learning_rate": 5.981352102325733e-06,
      "loss": 0.0041,
      "step": 153860
    },
    {
      "epoch": 0.9868743846866752,
      "grad_norm": 0.2710689902305603,
      "learning_rate": 5.980803279142958e-06,
      "loss": 0.0016,
      "step": 153870
    },
    {
      "epoch": 0.9869385215804614,
      "grad_norm": 0.04658711329102516,
      "learning_rate": 5.9802544436700185e-06,
      "loss": 0.0032,
      "step": 153880
    },
    {
      "epoch": 0.9870026584742474,
      "grad_norm": 0.15454958379268646,
      "learning_rate": 5.979705595913787e-06,
      "loss": 0.004,
      "step": 153890
    },
    {
      "epoch": 0.9870667953680335,
      "grad_norm": 0.06560371816158295,
      "learning_rate": 5.979156735881148e-06,
      "loss": 0.0009,
      "step": 153900
    },
    {
      "epoch": 0.9871309322618196,
      "grad_norm": 0.0811891257762909,
      "learning_rate": 5.9786078635789725e-06,
      "loss": 0.003,
      "step": 153910
    },
    {
      "epoch": 0.9871950691556057,
      "grad_norm": 0.06769920885562897,
      "learning_rate": 5.978058979014143e-06,
      "loss": 0.0018,
      "step": 153920
    },
    {
      "epoch": 0.9872592060493918,
      "grad_norm": 0.14531618356704712,
      "learning_rate": 5.977510082193534e-06,
      "loss": 0.0022,
      "step": 153930
    },
    {
      "epoch": 0.9873233429431779,
      "grad_norm": 0.12348922342061996,
      "learning_rate": 5.976961173124025e-06,
      "loss": 0.0019,
      "step": 153940
    },
    {
      "epoch": 0.9873874798369641,
      "grad_norm": 0.14243733882904053,
      "learning_rate": 5.976412251812496e-06,
      "loss": 0.0024,
      "step": 153950
    },
    {
      "epoch": 0.9874516167307501,
      "grad_norm": 0.39899900555610657,
      "learning_rate": 5.975863318265822e-06,
      "loss": 0.0028,
      "step": 153960
    },
    {
      "epoch": 0.9875157536245363,
      "grad_norm": 0.06399998068809509,
      "learning_rate": 5.975314372490884e-06,
      "loss": 0.0018,
      "step": 153970
    },
    {
      "epoch": 0.9875798905183223,
      "grad_norm": 0.2605004906654358,
      "learning_rate": 5.974765414494559e-06,
      "loss": 0.0023,
      "step": 153980
    },
    {
      "epoch": 0.9876440274121084,
      "grad_norm": 0.14542421698570251,
      "learning_rate": 5.974216444283728e-06,
      "loss": 0.0023,
      "step": 153990
    },
    {
      "epoch": 0.9877081643058945,
      "grad_norm": 0.04210500419139862,
      "learning_rate": 5.9736674618652676e-06,
      "loss": 0.002,
      "step": 154000
    },
    {
      "epoch": 0.9877723011996806,
      "grad_norm": 0.05088496953248978,
      "learning_rate": 5.9731184672460595e-06,
      "loss": 0.0026,
      "step": 154010
    },
    {
      "epoch": 0.9878364380934667,
      "grad_norm": 0.12533022463321686,
      "learning_rate": 5.9725694604329795e-06,
      "loss": 0.0022,
      "step": 154020
    },
    {
      "epoch": 0.9879005749872528,
      "grad_norm": 0.22114339470863342,
      "learning_rate": 5.972020441432911e-06,
      "loss": 0.0019,
      "step": 154030
    },
    {
      "epoch": 0.9879647118810388,
      "grad_norm": 0.0327499695122242,
      "learning_rate": 5.971471410252729e-06,
      "loss": 0.0023,
      "step": 154040
    },
    {
      "epoch": 0.988028848774825,
      "grad_norm": 0.03905919939279556,
      "learning_rate": 5.9709223668993185e-06,
      "loss": 0.0016,
      "step": 154050
    },
    {
      "epoch": 0.988092985668611,
      "grad_norm": 0.1768200546503067,
      "learning_rate": 5.970373311379556e-06,
      "loss": 0.0021,
      "step": 154060
    },
    {
      "epoch": 0.9881571225623972,
      "grad_norm": 0.0032357678283005953,
      "learning_rate": 5.9698242437003226e-06,
      "loss": 0.0024,
      "step": 154070
    },
    {
      "epoch": 0.9882212594561833,
      "grad_norm": 0.0414033979177475,
      "learning_rate": 5.969275163868498e-06,
      "loss": 0.0031,
      "step": 154080
    },
    {
      "epoch": 0.9882853963499694,
      "grad_norm": 0.14489038288593292,
      "learning_rate": 5.968726071890962e-06,
      "loss": 0.0013,
      "step": 154090
    },
    {
      "epoch": 0.9883495332437555,
      "grad_norm": 0.1014803797006607,
      "learning_rate": 5.968176967774598e-06,
      "loss": 0.0024,
      "step": 154100
    },
    {
      "epoch": 0.9884136701375416,
      "grad_norm": 0.05862985551357269,
      "learning_rate": 5.967627851526283e-06,
      "loss": 0.0027,
      "step": 154110
    },
    {
      "epoch": 0.9884778070313277,
      "grad_norm": 0.16196656227111816,
      "learning_rate": 5.9670787231529005e-06,
      "loss": 0.0022,
      "step": 154120
    },
    {
      "epoch": 0.9885419439251137,
      "grad_norm": 0.043797098100185394,
      "learning_rate": 5.96652958266133e-06,
      "loss": 0.0031,
      "step": 154130
    },
    {
      "epoch": 0.9886060808188999,
      "grad_norm": 0.07587334513664246,
      "learning_rate": 5.965980430058453e-06,
      "loss": 0.0022,
      "step": 154140
    },
    {
      "epoch": 0.9886702177126859,
      "grad_norm": 0.11381826549768448,
      "learning_rate": 5.96543126535115e-06,
      "loss": 0.0009,
      "step": 154150
    },
    {
      "epoch": 0.9887343546064721,
      "grad_norm": 0.040695976465940475,
      "learning_rate": 5.964882088546305e-06,
      "loss": 0.0029,
      "step": 154160
    },
    {
      "epoch": 0.9887984915002581,
      "grad_norm": 0.10237430036067963,
      "learning_rate": 5.9643328996507956e-06,
      "loss": 0.0015,
      "step": 154170
    },
    {
      "epoch": 0.9888626283940443,
      "grad_norm": 0.09190256893634796,
      "learning_rate": 5.963783698671508e-06,
      "loss": 0.0011,
      "step": 154180
    },
    {
      "epoch": 0.9889267652878303,
      "grad_norm": 0.026083897799253464,
      "learning_rate": 5.96323448561532e-06,
      "loss": 0.0031,
      "step": 154190
    },
    {
      "epoch": 0.9889909021816164,
      "grad_norm": 0.04641636833548546,
      "learning_rate": 5.9626852604891175e-06,
      "loss": 0.0015,
      "step": 154200
    },
    {
      "epoch": 0.9890550390754025,
      "grad_norm": 0.040112677961587906,
      "learning_rate": 5.962136023299779e-06,
      "loss": 0.0017,
      "step": 154210
    },
    {
      "epoch": 0.9891191759691886,
      "grad_norm": 0.05133315920829773,
      "learning_rate": 5.96158677405419e-06,
      "loss": 0.0025,
      "step": 154220
    },
    {
      "epoch": 0.9891833128629748,
      "grad_norm": 0.08732984215021133,
      "learning_rate": 5.96103751275923e-06,
      "loss": 0.0018,
      "step": 154230
    },
    {
      "epoch": 0.9892474497567608,
      "grad_norm": 0.041648976504802704,
      "learning_rate": 5.9604882394217844e-06,
      "loss": 0.0018,
      "step": 154240
    },
    {
      "epoch": 0.989311586650547,
      "grad_norm": 0.11416812241077423,
      "learning_rate": 5.959938954048734e-06,
      "loss": 0.0021,
      "step": 154250
    },
    {
      "epoch": 0.989375723544333,
      "grad_norm": 0.0667480081319809,
      "learning_rate": 5.959389656646964e-06,
      "loss": 0.0025,
      "step": 154260
    },
    {
      "epoch": 0.9894398604381192,
      "grad_norm": 0.12027747929096222,
      "learning_rate": 5.958840347223353e-06,
      "loss": 0.001,
      "step": 154270
    },
    {
      "epoch": 0.9895039973319052,
      "grad_norm": 0.13249050080776215,
      "learning_rate": 5.958291025784789e-06,
      "loss": 0.0037,
      "step": 154280
    },
    {
      "epoch": 0.9895681342256913,
      "grad_norm": 0.1557137817144394,
      "learning_rate": 5.957741692338152e-06,
      "loss": 0.0026,
      "step": 154290
    },
    {
      "epoch": 0.9896322711194774,
      "grad_norm": 0.01759687066078186,
      "learning_rate": 5.9571923468903285e-06,
      "loss": 0.002,
      "step": 154300
    },
    {
      "epoch": 0.9896964080132635,
      "grad_norm": 0.04367992654442787,
      "learning_rate": 5.956642989448201e-06,
      "loss": 0.0019,
      "step": 154310
    },
    {
      "epoch": 0.9897605449070496,
      "grad_norm": 0.18906162679195404,
      "learning_rate": 5.956093620018652e-06,
      "loss": 0.0038,
      "step": 154320
    },
    {
      "epoch": 0.9898246818008357,
      "grad_norm": 0.07389261573553085,
      "learning_rate": 5.955544238608568e-06,
      "loss": 0.002,
      "step": 154330
    },
    {
      "epoch": 0.9898888186946218,
      "grad_norm": 0.18635764718055725,
      "learning_rate": 5.954994845224831e-06,
      "loss": 0.0028,
      "step": 154340
    },
    {
      "epoch": 0.9899529555884079,
      "grad_norm": 0.07302852720022202,
      "learning_rate": 5.954445439874327e-06,
      "loss": 0.0023,
      "step": 154350
    },
    {
      "epoch": 0.990017092482194,
      "grad_norm": 0.048571355640888214,
      "learning_rate": 5.953896022563938e-06,
      "loss": 0.0014,
      "step": 154360
    },
    {
      "epoch": 0.9900812293759801,
      "grad_norm": 0.3239956200122833,
      "learning_rate": 5.953346593300552e-06,
      "loss": 0.0027,
      "step": 154370
    },
    {
      "epoch": 0.9901453662697662,
      "grad_norm": 0.05844731256365776,
      "learning_rate": 5.95279715209105e-06,
      "loss": 0.0018,
      "step": 154380
    },
    {
      "epoch": 0.9902095031635523,
      "grad_norm": 0.0586627796292305,
      "learning_rate": 5.9522476989423195e-06,
      "loss": 0.0048,
      "step": 154390
    },
    {
      "epoch": 0.9902736400573384,
      "grad_norm": 0.12048666924238205,
      "learning_rate": 5.951698233861244e-06,
      "loss": 0.0018,
      "step": 154400
    },
    {
      "epoch": 0.9903377769511245,
      "grad_norm": 0.0442906953394413,
      "learning_rate": 5.951148756854711e-06,
      "loss": 0.0018,
      "step": 154410
    },
    {
      "epoch": 0.9904019138449106,
      "grad_norm": 0.18814098834991455,
      "learning_rate": 5.950599267929603e-06,
      "loss": 0.0031,
      "step": 154420
    },
    {
      "epoch": 0.9904660507386966,
      "grad_norm": 0.04006590694189072,
      "learning_rate": 5.950049767092807e-06,
      "loss": 0.0018,
      "step": 154430
    },
    {
      "epoch": 0.9905301876324828,
      "grad_norm": 0.05012265965342522,
      "learning_rate": 5.9495002543512085e-06,
      "loss": 0.0023,
      "step": 154440
    },
    {
      "epoch": 0.9905943245262688,
      "grad_norm": 0.11422944068908691,
      "learning_rate": 5.948950729711694e-06,
      "loss": 0.0015,
      "step": 154450
    },
    {
      "epoch": 0.990658461420055,
      "grad_norm": 0.02385176159441471,
      "learning_rate": 5.948401193181148e-06,
      "loss": 0.0026,
      "step": 154460
    },
    {
      "epoch": 0.990722598313841,
      "grad_norm": 0.3480783998966217,
      "learning_rate": 5.947851644766456e-06,
      "loss": 0.0035,
      "step": 154470
    },
    {
      "epoch": 0.9907867352076272,
      "grad_norm": 0.08721592277288437,
      "learning_rate": 5.947302084474508e-06,
      "loss": 0.0025,
      "step": 154480
    },
    {
      "epoch": 0.9908508721014132,
      "grad_norm": 0.06368129700422287,
      "learning_rate": 5.946752512312186e-06,
      "loss": 0.0029,
      "step": 154490
    },
    {
      "epoch": 0.9909150089951994,
      "grad_norm": 0.14339937269687653,
      "learning_rate": 5.94620292828638e-06,
      "loss": 0.0041,
      "step": 154500
    },
    {
      "epoch": 0.9909791458889855,
      "grad_norm": 0.11102969199419022,
      "learning_rate": 5.945653332403973e-06,
      "loss": 0.0048,
      "step": 154510
    },
    {
      "epoch": 0.9910432827827715,
      "grad_norm": 0.06453274190425873,
      "learning_rate": 5.945103724671856e-06,
      "loss": 0.0018,
      "step": 154520
    },
    {
      "epoch": 0.9911074196765577,
      "grad_norm": 0.16440774500370026,
      "learning_rate": 5.944554105096911e-06,
      "loss": 0.0029,
      "step": 154530
    },
    {
      "epoch": 0.9911715565703437,
      "grad_norm": 0.08864869177341461,
      "learning_rate": 5.944004473686031e-06,
      "loss": 0.0034,
      "step": 154540
    },
    {
      "epoch": 0.9912356934641299,
      "grad_norm": 0.10180410742759705,
      "learning_rate": 5.943454830446097e-06,
      "loss": 0.0025,
      "step": 154550
    },
    {
      "epoch": 0.9912998303579159,
      "grad_norm": 0.2581990957260132,
      "learning_rate": 5.9429051753840025e-06,
      "loss": 0.0033,
      "step": 154560
    },
    {
      "epoch": 0.9913639672517021,
      "grad_norm": 0.10305880755186081,
      "learning_rate": 5.942355508506632e-06,
      "loss": 0.0014,
      "step": 154570
    },
    {
      "epoch": 0.9914281041454881,
      "grad_norm": 0.006125007756054401,
      "learning_rate": 5.941805829820873e-06,
      "loss": 0.0019,
      "step": 154580
    },
    {
      "epoch": 0.9914922410392742,
      "grad_norm": 0.09211353957653046,
      "learning_rate": 5.941256139333613e-06,
      "loss": 0.0019,
      "step": 154590
    },
    {
      "epoch": 0.9915563779330603,
      "grad_norm": 0.17578403651714325,
      "learning_rate": 5.940706437051743e-06,
      "loss": 0.0019,
      "step": 154600
    },
    {
      "epoch": 0.9916205148268464,
      "grad_norm": 0.1794659048318863,
      "learning_rate": 5.940156722982146e-06,
      "loss": 0.0018,
      "step": 154610
    },
    {
      "epoch": 0.9916846517206325,
      "grad_norm": 0.20038414001464844,
      "learning_rate": 5.9396069971317176e-06,
      "loss": 0.0026,
      "step": 154620
    },
    {
      "epoch": 0.9917487886144186,
      "grad_norm": 0.09909538179636002,
      "learning_rate": 5.9390572595073385e-06,
      "loss": 0.0019,
      "step": 154630
    },
    {
      "epoch": 0.9918129255082047,
      "grad_norm": 0.056216951459646225,
      "learning_rate": 5.938507510115901e-06,
      "loss": 0.0011,
      "step": 154640
    },
    {
      "epoch": 0.9918770624019908,
      "grad_norm": 0.0871502012014389,
      "learning_rate": 5.937957748964295e-06,
      "loss": 0.0025,
      "step": 154650
    },
    {
      "epoch": 0.991941199295777,
      "grad_norm": 0.35571610927581787,
      "learning_rate": 5.937407976059407e-06,
      "loss": 0.0025,
      "step": 154660
    },
    {
      "epoch": 0.992005336189563,
      "grad_norm": 0.2239132970571518,
      "learning_rate": 5.936858191408128e-06,
      "loss": 0.0024,
      "step": 154670
    },
    {
      "epoch": 0.9920694730833491,
      "grad_norm": 0.06647442281246185,
      "learning_rate": 5.936308395017346e-06,
      "loss": 0.0028,
      "step": 154680
    },
    {
      "epoch": 0.9921336099771352,
      "grad_norm": 0.09694932401180267,
      "learning_rate": 5.93575858689395e-06,
      "loss": 0.0033,
      "step": 154690
    },
    {
      "epoch": 0.9921977468709213,
      "grad_norm": 0.18823418021202087,
      "learning_rate": 5.935208767044831e-06,
      "loss": 0.0019,
      "step": 154700
    },
    {
      "epoch": 0.9922618837647074,
      "grad_norm": 0.11843463778495789,
      "learning_rate": 5.934658935476878e-06,
      "loss": 0.0014,
      "step": 154710
    },
    {
      "epoch": 0.9923260206584935,
      "grad_norm": 0.2229308784008026,
      "learning_rate": 5.93410909219698e-06,
      "loss": 0.0026,
      "step": 154720
    },
    {
      "epoch": 0.9923901575522796,
      "grad_norm": 0.008210033178329468,
      "learning_rate": 5.933559237212027e-06,
      "loss": 0.0014,
      "step": 154730
    },
    {
      "epoch": 0.9924542944460657,
      "grad_norm": 0.05394374579191208,
      "learning_rate": 5.933009370528911e-06,
      "loss": 0.0018,
      "step": 154740
    },
    {
      "epoch": 0.9925184313398517,
      "grad_norm": 0.13600340485572815,
      "learning_rate": 5.9324594921545195e-06,
      "loss": 0.0017,
      "step": 154750
    },
    {
      "epoch": 0.9925825682336379,
      "grad_norm": 0.04914219304919243,
      "learning_rate": 5.931909602095744e-06,
      "loss": 0.0025,
      "step": 154760
    },
    {
      "epoch": 0.9926467051274239,
      "grad_norm": 0.09887050837278366,
      "learning_rate": 5.931359700359476e-06,
      "loss": 0.0015,
      "step": 154770
    },
    {
      "epoch": 0.9927108420212101,
      "grad_norm": 0.006539844907820225,
      "learning_rate": 5.930809786952605e-06,
      "loss": 0.004,
      "step": 154780
    },
    {
      "epoch": 0.9927749789149962,
      "grad_norm": 0.09673035889863968,
      "learning_rate": 5.930259861882022e-06,
      "loss": 0.0024,
      "step": 154790
    },
    {
      "epoch": 0.9928391158087823,
      "grad_norm": 0.3432390093803406,
      "learning_rate": 5.929709925154619e-06,
      "loss": 0.0025,
      "step": 154800
    },
    {
      "epoch": 0.9929032527025684,
      "grad_norm": 0.30649974942207336,
      "learning_rate": 5.9291599767772844e-06,
      "loss": 0.0035,
      "step": 154810
    },
    {
      "epoch": 0.9929673895963544,
      "grad_norm": 0.33777597546577454,
      "learning_rate": 5.928610016756913e-06,
      "loss": 0.0024,
      "step": 154820
    },
    {
      "epoch": 0.9930315264901406,
      "grad_norm": 0.13778480887413025,
      "learning_rate": 5.928060045100392e-06,
      "loss": 0.0015,
      "step": 154830
    },
    {
      "epoch": 0.9930956633839266,
      "grad_norm": 0.06798578053712845,
      "learning_rate": 5.927510061814618e-06,
      "loss": 0.0027,
      "step": 154840
    },
    {
      "epoch": 0.9931598002777128,
      "grad_norm": 0.2452346831560135,
      "learning_rate": 5.9269600669064785e-06,
      "loss": 0.003,
      "step": 154850
    },
    {
      "epoch": 0.9932239371714988,
      "grad_norm": 0.10779757797718048,
      "learning_rate": 5.9264100603828676e-06,
      "loss": 0.0016,
      "step": 154860
    },
    {
      "epoch": 0.993288074065285,
      "grad_norm": 0.07469695806503296,
      "learning_rate": 5.925860042250676e-06,
      "loss": 0.0029,
      "step": 154870
    },
    {
      "epoch": 0.993352210959071,
      "grad_norm": 0.2282448261976242,
      "learning_rate": 5.925310012516797e-06,
      "loss": 0.0022,
      "step": 154880
    },
    {
      "epoch": 0.9934163478528572,
      "grad_norm": 0.2129080593585968,
      "learning_rate": 5.924759971188122e-06,
      "loss": 0.0028,
      "step": 154890
    },
    {
      "epoch": 0.9934804847466432,
      "grad_norm": 0.06968335807323456,
      "learning_rate": 5.924209918271544e-06,
      "loss": 0.0023,
      "step": 154900
    },
    {
      "epoch": 0.9935446216404293,
      "grad_norm": 0.23044312000274658,
      "learning_rate": 5.923659853773954e-06,
      "loss": 0.002,
      "step": 154910
    },
    {
      "epoch": 0.9936087585342154,
      "grad_norm": 0.3593491315841675,
      "learning_rate": 5.923109777702247e-06,
      "loss": 0.003,
      "step": 154920
    },
    {
      "epoch": 0.9936728954280015,
      "grad_norm": 0.1535368710756302,
      "learning_rate": 5.922559690063313e-06,
      "loss": 0.0022,
      "step": 154930
    },
    {
      "epoch": 0.9937370323217877,
      "grad_norm": 0.04858865216374397,
      "learning_rate": 5.9220095908640495e-06,
      "loss": 0.0017,
      "step": 154940
    },
    {
      "epoch": 0.9938011692155737,
      "grad_norm": 0.020658988505601883,
      "learning_rate": 5.9214594801113445e-06,
      "loss": 0.0028,
      "step": 154950
    },
    {
      "epoch": 0.9938653061093599,
      "grad_norm": 0.11902821063995361,
      "learning_rate": 5.920909357812096e-06,
      "loss": 0.0025,
      "step": 154960
    },
    {
      "epoch": 0.9939294430031459,
      "grad_norm": 0.059825241565704346,
      "learning_rate": 5.9203592239731935e-06,
      "loss": 0.0024,
      "step": 154970
    },
    {
      "epoch": 0.993993579896932,
      "grad_norm": 0.07929425686597824,
      "learning_rate": 5.919809078601533e-06,
      "loss": 0.0017,
      "step": 154980
    },
    {
      "epoch": 0.9940577167907181,
      "grad_norm": 0.10015887022018433,
      "learning_rate": 5.919258921704005e-06,
      "loss": 0.0023,
      "step": 154990
    },
    {
      "epoch": 0.9941218536845042,
      "grad_norm": 0.05563042685389519,
      "learning_rate": 5.918708753287507e-06,
      "loss": 0.002,
      "step": 155000
    },
    {
      "epoch": 0.9941859905782903,
      "grad_norm": 0.1357041895389557,
      "learning_rate": 5.918158573358934e-06,
      "loss": 0.0023,
      "step": 155010
    },
    {
      "epoch": 0.9942501274720764,
      "grad_norm": 0.19267672300338745,
      "learning_rate": 5.917608381925175e-06,
      "loss": 0.0023,
      "step": 155020
    },
    {
      "epoch": 0.9943142643658625,
      "grad_norm": 0.06331752240657806,
      "learning_rate": 5.917058178993129e-06,
      "loss": 0.0022,
      "step": 155030
    },
    {
      "epoch": 0.9943784012596486,
      "grad_norm": 0.0785926952958107,
      "learning_rate": 5.916507964569688e-06,
      "loss": 0.0021,
      "step": 155040
    },
    {
      "epoch": 0.9944425381534346,
      "grad_norm": 0.13848865032196045,
      "learning_rate": 5.9159577386617474e-06,
      "loss": 0.0025,
      "step": 155050
    },
    {
      "epoch": 0.9945066750472208,
      "grad_norm": 0.04544994607567787,
      "learning_rate": 5.9154075012762e-06,
      "loss": 0.0023,
      "step": 155060
    },
    {
      "epoch": 0.9945708119410069,
      "grad_norm": 0.03753521293401718,
      "learning_rate": 5.914857252419946e-06,
      "loss": 0.0016,
      "step": 155070
    },
    {
      "epoch": 0.994634948834793,
      "grad_norm": 0.008138508535921574,
      "learning_rate": 5.914306992099873e-06,
      "loss": 0.0101,
      "step": 155080
    },
    {
      "epoch": 0.9946990857285791,
      "grad_norm": 0.11182375997304916,
      "learning_rate": 5.913756720322883e-06,
      "loss": 0.0017,
      "step": 155090
    },
    {
      "epoch": 0.9947632226223652,
      "grad_norm": 0.13418009877204895,
      "learning_rate": 5.9132064370958654e-06,
      "loss": 0.0017,
      "step": 155100
    },
    {
      "epoch": 0.9948273595161513,
      "grad_norm": 0.19320058822631836,
      "learning_rate": 5.912656142425721e-06,
      "loss": 0.0025,
      "step": 155110
    },
    {
      "epoch": 0.9948914964099373,
      "grad_norm": 0.04462080076336861,
      "learning_rate": 5.912105836319341e-06,
      "loss": 0.0026,
      "step": 155120
    },
    {
      "epoch": 0.9949556333037235,
      "grad_norm": 0.2760769724845886,
      "learning_rate": 5.911555518783624e-06,
      "loss": 0.0015,
      "step": 155130
    },
    {
      "epoch": 0.9950197701975095,
      "grad_norm": 0.11476902663707733,
      "learning_rate": 5.911005189825464e-06,
      "loss": 0.0026,
      "step": 155140
    },
    {
      "epoch": 0.9950839070912957,
      "grad_norm": 0.08436945080757141,
      "learning_rate": 5.910454849451759e-06,
      "loss": 0.0029,
      "step": 155150
    },
    {
      "epoch": 0.9951480439850817,
      "grad_norm": 0.0776107907295227,
      "learning_rate": 5.909904497669403e-06,
      "loss": 0.0018,
      "step": 155160
    },
    {
      "epoch": 0.9952121808788679,
      "grad_norm": 0.10979221016168594,
      "learning_rate": 5.9093541344852935e-06,
      "loss": 0.0023,
      "step": 155170
    },
    {
      "epoch": 0.9952763177726539,
      "grad_norm": 0.10220757871866226,
      "learning_rate": 5.908803759906328e-06,
      "loss": 0.0023,
      "step": 155180
    },
    {
      "epoch": 0.99534045466644,
      "grad_norm": 0.042567286640405655,
      "learning_rate": 5.908253373939401e-06,
      "loss": 0.0021,
      "step": 155190
    },
    {
      "epoch": 0.9954045915602261,
      "grad_norm": 0.04859263077378273,
      "learning_rate": 5.907702976591411e-06,
      "loss": 0.0014,
      "step": 155200
    },
    {
      "epoch": 0.9954687284540122,
      "grad_norm": 0.3271733522415161,
      "learning_rate": 5.907152567869252e-06,
      "loss": 0.0023,
      "step": 155210
    },
    {
      "epoch": 0.9955328653477984,
      "grad_norm": 0.24706242978572845,
      "learning_rate": 5.906602147779824e-06,
      "loss": 0.0021,
      "step": 155220
    },
    {
      "epoch": 0.9955970022415844,
      "grad_norm": 0.09822478890419006,
      "learning_rate": 5.906051716330023e-06,
      "loss": 0.0011,
      "step": 155230
    },
    {
      "epoch": 0.9956611391353706,
      "grad_norm": 0.05063209682703018,
      "learning_rate": 5.905501273526748e-06,
      "loss": 0.0028,
      "step": 155240
    },
    {
      "epoch": 0.9957252760291566,
      "grad_norm": 0.08223580569028854,
      "learning_rate": 5.904950819376892e-06,
      "loss": 0.0025,
      "step": 155250
    },
    {
      "epoch": 0.9957894129229428,
      "grad_norm": 0.011897790245711803,
      "learning_rate": 5.904400353887358e-06,
      "loss": 0.0013,
      "step": 155260
    },
    {
      "epoch": 0.9958535498167288,
      "grad_norm": 0.07343225181102753,
      "learning_rate": 5.90384987706504e-06,
      "loss": 0.0014,
      "step": 155270
    },
    {
      "epoch": 0.995917686710515,
      "grad_norm": 0.033276788890361786,
      "learning_rate": 5.903299388916837e-06,
      "loss": 0.0012,
      "step": 155280
    },
    {
      "epoch": 0.995981823604301,
      "grad_norm": 0.06330505758523941,
      "learning_rate": 5.9027488894496475e-06,
      "loss": 0.0019,
      "step": 155290
    },
    {
      "epoch": 0.9960459604980871,
      "grad_norm": 0.12273988127708435,
      "learning_rate": 5.90219837867037e-06,
      "loss": 0.002,
      "step": 155300
    },
    {
      "epoch": 0.9961100973918732,
      "grad_norm": 0.11730191111564636,
      "learning_rate": 5.901647856585899e-06,
      "loss": 0.0016,
      "step": 155310
    },
    {
      "epoch": 0.9961742342856593,
      "grad_norm": 0.03262682631611824,
      "learning_rate": 5.9010973232031395e-06,
      "loss": 0.0018,
      "step": 155320
    },
    {
      "epoch": 0.9962383711794454,
      "grad_norm": 0.10133693367242813,
      "learning_rate": 5.900546778528985e-06,
      "loss": 0.0018,
      "step": 155330
    },
    {
      "epoch": 0.9963025080732315,
      "grad_norm": 0.44229885935783386,
      "learning_rate": 5.899996222570336e-06,
      "loss": 0.0039,
      "step": 155340
    },
    {
      "epoch": 0.9963666449670177,
      "grad_norm": 0.06712737679481506,
      "learning_rate": 5.899445655334091e-06,
      "loss": 0.0017,
      "step": 155350
    },
    {
      "epoch": 0.9964307818608037,
      "grad_norm": 0.028598163276910782,
      "learning_rate": 5.898895076827148e-06,
      "loss": 0.0038,
      "step": 155360
    },
    {
      "epoch": 0.9964949187545898,
      "grad_norm": 0.02832469530403614,
      "learning_rate": 5.898344487056409e-06,
      "loss": 0.0029,
      "step": 155370
    },
    {
      "epoch": 0.9965590556483759,
      "grad_norm": 0.14169839024543762,
      "learning_rate": 5.897793886028771e-06,
      "loss": 0.0028,
      "step": 155380
    },
    {
      "epoch": 0.996623192542162,
      "grad_norm": 0.10831230133771896,
      "learning_rate": 5.897243273751134e-06,
      "loss": 0.0017,
      "step": 155390
    },
    {
      "epoch": 0.9966873294359481,
      "grad_norm": 0.1491563618183136,
      "learning_rate": 5.896692650230396e-06,
      "loss": 0.0032,
      "step": 155400
    },
    {
      "epoch": 0.9967514663297342,
      "grad_norm": 0.14146079123020172,
      "learning_rate": 5.896142015473462e-06,
      "loss": 0.0022,
      "step": 155410
    },
    {
      "epoch": 0.9968156032235203,
      "grad_norm": 0.08077463507652283,
      "learning_rate": 5.895591369487225e-06,
      "loss": 0.0021,
      "step": 155420
    },
    {
      "epoch": 0.9968797401173064,
      "grad_norm": 0.08268241584300995,
      "learning_rate": 5.895040712278589e-06,
      "loss": 0.0012,
      "step": 155430
    },
    {
      "epoch": 0.9969438770110924,
      "grad_norm": 0.019326310604810715,
      "learning_rate": 5.894490043854453e-06,
      "loss": 0.0024,
      "step": 155440
    },
    {
      "epoch": 0.9970080139048786,
      "grad_norm": 0.03024476021528244,
      "learning_rate": 5.893939364221718e-06,
      "loss": 0.0012,
      "step": 155450
    },
    {
      "epoch": 0.9970721507986646,
      "grad_norm": 0.12479130178689957,
      "learning_rate": 5.893388673387282e-06,
      "loss": 0.0027,
      "step": 155460
    },
    {
      "epoch": 0.9971362876924508,
      "grad_norm": 0.16635245084762573,
      "learning_rate": 5.89283797135805e-06,
      "loss": 0.0014,
      "step": 155470
    },
    {
      "epoch": 0.9972004245862368,
      "grad_norm": 0.11341109126806259,
      "learning_rate": 5.892287258140919e-06,
      "loss": 0.0028,
      "step": 155480
    },
    {
      "epoch": 0.997264561480023,
      "grad_norm": 0.06706415116786957,
      "learning_rate": 5.8917365337427904e-06,
      "loss": 0.0033,
      "step": 155490
    },
    {
      "epoch": 0.9973286983738091,
      "grad_norm": 0.12035437673330307,
      "learning_rate": 5.891185798170567e-06,
      "loss": 0.0023,
      "step": 155500
    },
    {
      "epoch": 0.9973928352675951,
      "grad_norm": 0.13776159286499023,
      "learning_rate": 5.890635051431146e-06,
      "loss": 0.0015,
      "step": 155510
    },
    {
      "epoch": 0.9974569721613813,
      "grad_norm": 0.11789148300886154,
      "learning_rate": 5.890084293531434e-06,
      "loss": 0.0022,
      "step": 155520
    },
    {
      "epoch": 0.9975211090551673,
      "grad_norm": 0.08669178932905197,
      "learning_rate": 5.889533524478328e-06,
      "loss": 0.0025,
      "step": 155530
    },
    {
      "epoch": 0.9975852459489535,
      "grad_norm": 0.15548285841941833,
      "learning_rate": 5.888982744278733e-06,
      "loss": 0.0021,
      "step": 155540
    },
    {
      "epoch": 0.9976493828427395,
      "grad_norm": 0.07993265241384506,
      "learning_rate": 5.888431952939548e-06,
      "loss": 0.0019,
      "step": 155550
    },
    {
      "epoch": 0.9977135197365257,
      "grad_norm": 0.05593022331595421,
      "learning_rate": 5.887881150467676e-06,
      "loss": 0.0033,
      "step": 155560
    },
    {
      "epoch": 0.9977776566303117,
      "grad_norm": 0.13642093539237976,
      "learning_rate": 5.887330336870017e-06,
      "loss": 0.0017,
      "step": 155570
    },
    {
      "epoch": 0.9978417935240979,
      "grad_norm": 0.11116381734609604,
      "learning_rate": 5.886779512153477e-06,
      "loss": 0.002,
      "step": 155580
    },
    {
      "epoch": 0.9979059304178839,
      "grad_norm": 0.26490068435668945,
      "learning_rate": 5.886228676324953e-06,
      "loss": 0.0016,
      "step": 155590
    },
    {
      "epoch": 0.99797006731167,
      "grad_norm": 0.09997473657131195,
      "learning_rate": 5.885677829391353e-06,
      "loss": 0.0022,
      "step": 155600
    },
    {
      "epoch": 0.9980342042054561,
      "grad_norm": 0.06118928641080856,
      "learning_rate": 5.885126971359576e-06,
      "loss": 0.0012,
      "step": 155610
    },
    {
      "epoch": 0.9980983410992422,
      "grad_norm": 0.025356553494930267,
      "learning_rate": 5.884576102236526e-06,
      "loss": 0.0024,
      "step": 155620
    },
    {
      "epoch": 0.9981624779930284,
      "grad_norm": 0.31826135516166687,
      "learning_rate": 5.884025222029104e-06,
      "loss": 0.0027,
      "step": 155630
    },
    {
      "epoch": 0.9982266148868144,
      "grad_norm": 0.005015498027205467,
      "learning_rate": 5.8834743307442145e-06,
      "loss": 0.0021,
      "step": 155640
    },
    {
      "epoch": 0.9982907517806006,
      "grad_norm": 0.1632152497768402,
      "learning_rate": 5.88292342838876e-06,
      "loss": 0.0028,
      "step": 155650
    },
    {
      "epoch": 0.9983548886743866,
      "grad_norm": 0.0382956862449646,
      "learning_rate": 5.882372514969644e-06,
      "loss": 0.0035,
      "step": 155660
    },
    {
      "epoch": 0.9984190255681727,
      "grad_norm": 0.11817727982997894,
      "learning_rate": 5.881821590493772e-06,
      "loss": 0.0019,
      "step": 155670
    },
    {
      "epoch": 0.9984831624619588,
      "grad_norm": 0.12771518528461456,
      "learning_rate": 5.881270654968042e-06,
      "loss": 0.0027,
      "step": 155680
    },
    {
      "epoch": 0.9985472993557449,
      "grad_norm": 0.03132863715291023,
      "learning_rate": 5.880719708399363e-06,
      "loss": 0.0056,
      "step": 155690
    },
    {
      "epoch": 0.998611436249531,
      "grad_norm": 0.12629881501197815,
      "learning_rate": 5.880168750794635e-06,
      "loss": 0.0017,
      "step": 155700
    },
    {
      "epoch": 0.9986755731433171,
      "grad_norm": 0.07862812280654907,
      "learning_rate": 5.879617782160765e-06,
      "loss": 0.0017,
      "step": 155710
    },
    {
      "epoch": 0.9987397100371032,
      "grad_norm": 0.046499527990818024,
      "learning_rate": 5.8790668025046536e-06,
      "loss": 0.0031,
      "step": 155720
    },
    {
      "epoch": 0.9988038469308893,
      "grad_norm": 0.11387953907251358,
      "learning_rate": 5.878515811833209e-06,
      "loss": 0.0023,
      "step": 155730
    },
    {
      "epoch": 0.9988679838246753,
      "grad_norm": 0.14124180376529694,
      "learning_rate": 5.877964810153333e-06,
      "loss": 0.0026,
      "step": 155740
    },
    {
      "epoch": 0.9989321207184615,
      "grad_norm": 0.005234932526946068,
      "learning_rate": 5.87741379747193e-06,
      "loss": 0.0034,
      "step": 155750
    },
    {
      "epoch": 0.9989962576122475,
      "grad_norm": 0.1285937875509262,
      "learning_rate": 5.876862773795905e-06,
      "loss": 0.0017,
      "step": 155760
    },
    {
      "epoch": 0.9990603945060337,
      "grad_norm": 0.10089916735887527,
      "learning_rate": 5.876311739132164e-06,
      "loss": 0.0014,
      "step": 155770
    },
    {
      "epoch": 0.9991245313998198,
      "grad_norm": 0.06541144847869873,
      "learning_rate": 5.875760693487607e-06,
      "loss": 0.0014,
      "step": 155780
    },
    {
      "epoch": 0.9991886682936059,
      "grad_norm": 0.011267658323049545,
      "learning_rate": 5.875209636869147e-06,
      "loss": 0.0019,
      "step": 155790
    },
    {
      "epoch": 0.999252805187392,
      "grad_norm": 0.09115643799304962,
      "learning_rate": 5.874658569283682e-06,
      "loss": 0.0026,
      "step": 155800
    },
    {
      "epoch": 0.999316942081178,
      "grad_norm": 0.09521937370300293,
      "learning_rate": 5.8741074907381215e-06,
      "loss": 0.0013,
      "step": 155810
    },
    {
      "epoch": 0.9993810789749642,
      "grad_norm": 0.051444701850414276,
      "learning_rate": 5.873556401239369e-06,
      "loss": 0.0018,
      "step": 155820
    },
    {
      "epoch": 0.9994452158687502,
      "grad_norm": 0.009652688167989254,
      "learning_rate": 5.87300530079433e-06,
      "loss": 0.0018,
      "step": 155830
    },
    {
      "epoch": 0.9995093527625364,
      "grad_norm": 0.26273030042648315,
      "learning_rate": 5.87245418940991e-06,
      "loss": 0.0015,
      "step": 155840
    },
    {
      "epoch": 0.9995734896563224,
      "grad_norm": 0.05302487686276436,
      "learning_rate": 5.871903067093017e-06,
      "loss": 0.0034,
      "step": 155850
    },
    {
      "epoch": 0.9996376265501086,
      "grad_norm": 0.23049956560134888,
      "learning_rate": 5.871351933850555e-06,
      "loss": 0.0028,
      "step": 155860
    },
    {
      "epoch": 0.9997017634438946,
      "grad_norm": 0.01473341602832079,
      "learning_rate": 5.870800789689431e-06,
      "loss": 0.0023,
      "step": 155870
    },
    {
      "epoch": 0.9997659003376808,
      "grad_norm": 0.07029661536216736,
      "learning_rate": 5.87024963461655e-06,
      "loss": 0.0032,
      "step": 155880
    },
    {
      "epoch": 0.9998300372314668,
      "grad_norm": 0.058633968234062195,
      "learning_rate": 5.869698468638818e-06,
      "loss": 0.0017,
      "step": 155890
    },
    {
      "epoch": 0.9998941741252529,
      "grad_norm": 0.14927196502685547,
      "learning_rate": 5.869147291763145e-06,
      "loss": 0.0023,
      "step": 155900
    },
    {
      "epoch": 0.9999583110190391,
      "grad_norm": 0.01744919642806053,
      "learning_rate": 5.8685961039964334e-06,
      "loss": 0.0018,
      "step": 155910
    },
    {
      "epoch": 1.0000224479128252,
      "grad_norm": 0.09784498065710068,
      "learning_rate": 5.868044905345594e-06,
      "loss": 0.0013,
      "step": 155920
    },
    {
      "epoch": 1.0000865848066112,
      "grad_norm": 0.01963322050869465,
      "learning_rate": 5.867493695817529e-06,
      "loss": 0.0011,
      "step": 155930
    },
    {
      "epoch": 1.0001507217003973,
      "grad_norm": 0.0918341651558876,
      "learning_rate": 5.86694247541915e-06,
      "loss": 0.0009,
      "step": 155940
    },
    {
      "epoch": 1.0002148585941835,
      "grad_norm": 0.1195211410522461,
      "learning_rate": 5.866391244157361e-06,
      "loss": 0.0023,
      "step": 155950
    },
    {
      "epoch": 1.0002789954879696,
      "grad_norm": 0.07075042277574539,
      "learning_rate": 5.865840002039072e-06,
      "loss": 0.0012,
      "step": 155960
    },
    {
      "epoch": 1.0003431323817555,
      "grad_norm": 0.11134473234415054,
      "learning_rate": 5.8652887490711865e-06,
      "loss": 0.0015,
      "step": 155970
    },
    {
      "epoch": 1.0004072692755417,
      "grad_norm": 0.023944471031427383,
      "learning_rate": 5.864737485260616e-06,
      "loss": 0.0008,
      "step": 155980
    },
    {
      "epoch": 1.0004714061693278,
      "grad_norm": 0.026289818808436394,
      "learning_rate": 5.864186210614267e-06,
      "loss": 0.0008,
      "step": 155990
    },
    {
      "epoch": 1.000535543063114,
      "grad_norm": 0.0795121043920517,
      "learning_rate": 5.863634925139046e-06,
      "loss": 0.0011,
      "step": 156000
    },
    {
      "epoch": 1.0005996799569,
      "grad_norm": 0.13002263009548187,
      "learning_rate": 5.863083628841864e-06,
      "loss": 0.0022,
      "step": 156010
    },
    {
      "epoch": 1.000663816850686,
      "grad_norm": 0.02560841664671898,
      "learning_rate": 5.862532321729625e-06,
      "loss": 0.0018,
      "step": 156020
    },
    {
      "epoch": 1.0007279537444722,
      "grad_norm": 0.1948590725660324,
      "learning_rate": 5.8619810038092415e-06,
      "loss": 0.0025,
      "step": 156030
    },
    {
      "epoch": 1.0007920906382584,
      "grad_norm": 0.10446181893348694,
      "learning_rate": 5.8614296750876184e-06,
      "loss": 0.001,
      "step": 156040
    },
    {
      "epoch": 1.0008562275320443,
      "grad_norm": 0.01794133521616459,
      "learning_rate": 5.860878335571668e-06,
      "loss": 0.001,
      "step": 156050
    },
    {
      "epoch": 1.0009203644258304,
      "grad_norm": 0.06554514169692993,
      "learning_rate": 5.8603269852682956e-06,
      "loss": 0.0014,
      "step": 156060
    },
    {
      "epoch": 1.0009845013196166,
      "grad_norm": 0.11025381088256836,
      "learning_rate": 5.859775624184411e-06,
      "loss": 0.0032,
      "step": 156070
    },
    {
      "epoch": 1.0010486382134027,
      "grad_norm": 0.2035963237285614,
      "learning_rate": 5.859224252326922e-06,
      "loss": 0.0029,
      "step": 156080
    },
    {
      "epoch": 1.0011127751071889,
      "grad_norm": 0.17359669506549835,
      "learning_rate": 5.858672869702741e-06,
      "loss": 0.0018,
      "step": 156090
    },
    {
      "epoch": 1.0011769120009748,
      "grad_norm": 0.11393879354000092,
      "learning_rate": 5.858121476318774e-06,
      "loss": 0.0023,
      "step": 156100
    },
    {
      "epoch": 1.001241048894761,
      "grad_norm": 0.04529412463307381,
      "learning_rate": 5.857570072181933e-06,
      "loss": 0.0017,
      "step": 156110
    },
    {
      "epoch": 1.001305185788547,
      "grad_norm": 0.0755084976553917,
      "learning_rate": 5.857018657299125e-06,
      "loss": 0.0018,
      "step": 156120
    },
    {
      "epoch": 1.0013693226823333,
      "grad_norm": 0.033754877746105194,
      "learning_rate": 5.856467231677262e-06,
      "loss": 0.0011,
      "step": 156130
    },
    {
      "epoch": 1.0014334595761192,
      "grad_norm": 0.08789921551942825,
      "learning_rate": 5.855915795323251e-06,
      "loss": 0.002,
      "step": 156140
    },
    {
      "epoch": 1.0014975964699053,
      "grad_norm": 0.07098563760519028,
      "learning_rate": 5.8553643482440045e-06,
      "loss": 0.0017,
      "step": 156150
    },
    {
      "epoch": 1.0015617333636915,
      "grad_norm": 0.28439974784851074,
      "learning_rate": 5.854812890446431e-06,
      "loss": 0.002,
      "step": 156160
    },
    {
      "epoch": 1.0016258702574776,
      "grad_norm": 0.13283652067184448,
      "learning_rate": 5.8542614219374416e-06,
      "loss": 0.003,
      "step": 156170
    },
    {
      "epoch": 1.0016900071512636,
      "grad_norm": 0.12271107733249664,
      "learning_rate": 5.853709942723945e-06,
      "loss": 0.0025,
      "step": 156180
    },
    {
      "epoch": 1.0017541440450497,
      "grad_norm": 0.17057596147060394,
      "learning_rate": 5.853158452812854e-06,
      "loss": 0.002,
      "step": 156190
    },
    {
      "epoch": 1.0018182809388358,
      "grad_norm": 0.08077952265739441,
      "learning_rate": 5.852606952211076e-06,
      "loss": 0.0016,
      "step": 156200
    },
    {
      "epoch": 1.001882417832622,
      "grad_norm": 0.07943473011255264,
      "learning_rate": 5.852055440925524e-06,
      "loss": 0.0015,
      "step": 156210
    },
    {
      "epoch": 1.0019465547264081,
      "grad_norm": 0.08325579017400742,
      "learning_rate": 5.851503918963109e-06,
      "loss": 0.0015,
      "step": 156220
    },
    {
      "epoch": 1.002010691620194,
      "grad_norm": 0.2811479866504669,
      "learning_rate": 5.850952386330741e-06,
      "loss": 0.0023,
      "step": 156230
    },
    {
      "epoch": 1.0020748285139802,
      "grad_norm": 0.0512402318418026,
      "learning_rate": 5.850400843035333e-06,
      "loss": 0.0015,
      "step": 156240
    },
    {
      "epoch": 1.0021389654077664,
      "grad_norm": 0.10010890662670135,
      "learning_rate": 5.849849289083792e-06,
      "loss": 0.0016,
      "step": 156250
    },
    {
      "epoch": 1.0022031023015525,
      "grad_norm": 0.1832588016986847,
      "learning_rate": 5.849297724483034e-06,
      "loss": 0.0027,
      "step": 156260
    },
    {
      "epoch": 1.0022672391953384,
      "grad_norm": 0.12112399190664291,
      "learning_rate": 5.848746149239968e-06,
      "loss": 0.0023,
      "step": 156270
    },
    {
      "epoch": 1.0023313760891246,
      "grad_norm": 0.11804680526256561,
      "learning_rate": 5.848194563361507e-06,
      "loss": 0.0008,
      "step": 156280
    },
    {
      "epoch": 1.0023955129829107,
      "grad_norm": 0.10468452423810959,
      "learning_rate": 5.84764296685456e-06,
      "loss": 0.001,
      "step": 156290
    },
    {
      "epoch": 1.002459649876697,
      "grad_norm": 0.14020980894565582,
      "learning_rate": 5.847091359726043e-06,
      "loss": 0.0015,
      "step": 156300
    },
    {
      "epoch": 1.0025237867704828,
      "grad_norm": 0.33313900232315063,
      "learning_rate": 5.846539741982864e-06,
      "loss": 0.0023,
      "step": 156310
    },
    {
      "epoch": 1.002587923664269,
      "grad_norm": 0.013997141271829605,
      "learning_rate": 5.845988113631937e-06,
      "loss": 0.0013,
      "step": 156320
    },
    {
      "epoch": 1.0026520605580551,
      "grad_norm": 0.05682804808020592,
      "learning_rate": 5.845436474680175e-06,
      "loss": 0.0018,
      "step": 156330
    },
    {
      "epoch": 1.0027161974518413,
      "grad_norm": 0.0879458412528038,
      "learning_rate": 5.844884825134489e-06,
      "loss": 0.0009,
      "step": 156340
    },
    {
      "epoch": 1.0027803343456274,
      "grad_norm": 0.09567782282829285,
      "learning_rate": 5.844333165001794e-06,
      "loss": 0.0022,
      "step": 156350
    },
    {
      "epoch": 1.0028444712394133,
      "grad_norm": 0.023266635835170746,
      "learning_rate": 5.843781494289001e-06,
      "loss": 0.0026,
      "step": 156360
    },
    {
      "epoch": 1.0029086081331995,
      "grad_norm": 0.007620238699018955,
      "learning_rate": 5.843229813003022e-06,
      "loss": 0.0013,
      "step": 156370
    },
    {
      "epoch": 1.0029727450269856,
      "grad_norm": 0.19264978170394897,
      "learning_rate": 5.84267812115077e-06,
      "loss": 0.0019,
      "step": 156380
    },
    {
      "epoch": 1.0030368819207718,
      "grad_norm": 0.12446966022253036,
      "learning_rate": 5.84212641873916e-06,
      "loss": 0.0018,
      "step": 156390
    },
    {
      "epoch": 1.0031010188145577,
      "grad_norm": 0.20501558482646942,
      "learning_rate": 5.841574705775103e-06,
      "loss": 0.0027,
      "step": 156400
    },
    {
      "epoch": 1.0031651557083439,
      "grad_norm": 0.11769339442253113,
      "learning_rate": 5.841022982265515e-06,
      "loss": 0.003,
      "step": 156410
    },
    {
      "epoch": 1.00322929260213,
      "grad_norm": 0.09512394666671753,
      "learning_rate": 5.8404712482173064e-06,
      "loss": 0.0015,
      "step": 156420
    },
    {
      "epoch": 1.0032934294959162,
      "grad_norm": 0.08365820348262787,
      "learning_rate": 5.839919503637394e-06,
      "loss": 0.0026,
      "step": 156430
    },
    {
      "epoch": 1.003357566389702,
      "grad_norm": 0.17679233849048615,
      "learning_rate": 5.839367748532689e-06,
      "loss": 0.0019,
      "step": 156440
    },
    {
      "epoch": 1.0034217032834882,
      "grad_norm": 0.30964887142181396,
      "learning_rate": 5.838815982910107e-06,
      "loss": 0.0028,
      "step": 156450
    },
    {
      "epoch": 1.0034858401772744,
      "grad_norm": 0.10647038370370865,
      "learning_rate": 5.8382642067765586e-06,
      "loss": 0.0015,
      "step": 156460
    },
    {
      "epoch": 1.0035499770710605,
      "grad_norm": 0.14962203800678253,
      "learning_rate": 5.837712420138962e-06,
      "loss": 0.0052,
      "step": 156470
    },
    {
      "epoch": 1.0036141139648467,
      "grad_norm": 0.08887304365634918,
      "learning_rate": 5.83716062300423e-06,
      "loss": 0.0026,
      "step": 156480
    },
    {
      "epoch": 1.0036782508586326,
      "grad_norm": 0.12420698255300522,
      "learning_rate": 5.836608815379279e-06,
      "loss": 0.0034,
      "step": 156490
    },
    {
      "epoch": 1.0037423877524188,
      "grad_norm": 0.006879259832203388,
      "learning_rate": 5.8360569972710206e-06,
      "loss": 0.0032,
      "step": 156500
    },
    {
      "epoch": 1.003806524646205,
      "grad_norm": 0.04893284663558006,
      "learning_rate": 5.83550516868637e-06,
      "loss": 0.0022,
      "step": 156510
    },
    {
      "epoch": 1.003870661539991,
      "grad_norm": 0.2606249749660492,
      "learning_rate": 5.834953329632243e-06,
      "loss": 0.0019,
      "step": 156520
    },
    {
      "epoch": 1.003934798433777,
      "grad_norm": 0.1031232699751854,
      "learning_rate": 5.834401480115554e-06,
      "loss": 0.0021,
      "step": 156530
    },
    {
      "epoch": 1.0039989353275631,
      "grad_norm": 0.0825612023472786,
      "learning_rate": 5.833849620143217e-06,
      "loss": 0.0014,
      "step": 156540
    },
    {
      "epoch": 1.0040630722213493,
      "grad_norm": 0.023826727643609047,
      "learning_rate": 5.833297749722149e-06,
      "loss": 0.0016,
      "step": 156550
    },
    {
      "epoch": 1.0041272091151354,
      "grad_norm": 0.12111756950616837,
      "learning_rate": 5.832745868859265e-06,
      "loss": 0.0013,
      "step": 156560
    },
    {
      "epoch": 1.0041913460089213,
      "grad_norm": 0.16003498435020447,
      "learning_rate": 5.832193977561479e-06,
      "loss": 0.0022,
      "step": 156570
    },
    {
      "epoch": 1.0042554829027075,
      "grad_norm": 0.05248863250017166,
      "learning_rate": 5.8316420758357094e-06,
      "loss": 0.0013,
      "step": 156580
    },
    {
      "epoch": 1.0043196197964936,
      "grad_norm": 0.14343620836734772,
      "learning_rate": 5.83109016368887e-06,
      "loss": 0.0016,
      "step": 156590
    },
    {
      "epoch": 1.0043837566902798,
      "grad_norm": 0.2597375810146332,
      "learning_rate": 5.830538241127876e-06,
      "loss": 0.0013,
      "step": 156600
    },
    {
      "epoch": 1.0044478935840657,
      "grad_norm": 0.3479706943035126,
      "learning_rate": 5.8299863081596445e-06,
      "loss": 0.0045,
      "step": 156610
    },
    {
      "epoch": 1.0045120304778519,
      "grad_norm": 0.07276829332113266,
      "learning_rate": 5.829434364791091e-06,
      "loss": 0.0014,
      "step": 156620
    },
    {
      "epoch": 1.004576167371638,
      "grad_norm": 0.01598191447556019,
      "learning_rate": 5.828882411029133e-06,
      "loss": 0.0019,
      "step": 156630
    },
    {
      "epoch": 1.0046403042654242,
      "grad_norm": 0.05684473738074303,
      "learning_rate": 5.828330446880686e-06,
      "loss": 0.0024,
      "step": 156640
    },
    {
      "epoch": 1.0047044411592103,
      "grad_norm": 0.06137927994132042,
      "learning_rate": 5.827778472352666e-06,
      "loss": 0.002,
      "step": 156650
    },
    {
      "epoch": 1.0047685780529962,
      "grad_norm": 0.15637855231761932,
      "learning_rate": 5.82722648745199e-06,
      "loss": 0.002,
      "step": 156660
    },
    {
      "epoch": 1.0048327149467824,
      "grad_norm": 0.09243360161781311,
      "learning_rate": 5.8266744921855754e-06,
      "loss": 0.0014,
      "step": 156670
    },
    {
      "epoch": 1.0048968518405685,
      "grad_norm": 0.17562691867351532,
      "learning_rate": 5.826122486560338e-06,
      "loss": 0.0018,
      "step": 156680
    },
    {
      "epoch": 1.0049609887343547,
      "grad_norm": 0.051522646099328995,
      "learning_rate": 5.825570470583196e-06,
      "loss": 0.0015,
      "step": 156690
    },
    {
      "epoch": 1.0050251256281406,
      "grad_norm": 0.07359160482883453,
      "learning_rate": 5.8250184442610646e-06,
      "loss": 0.0019,
      "step": 156700
    },
    {
      "epoch": 1.0050892625219268,
      "grad_norm": 0.09471216052770615,
      "learning_rate": 5.824466407600865e-06,
      "loss": 0.0019,
      "step": 156710
    },
    {
      "epoch": 1.005153399415713,
      "grad_norm": 0.14061139523983002,
      "learning_rate": 5.8239143606095095e-06,
      "loss": 0.0015,
      "step": 156720
    },
    {
      "epoch": 1.005217536309499,
      "grad_norm": 0.04952032119035721,
      "learning_rate": 5.82336230329392e-06,
      "loss": 0.0021,
      "step": 156730
    },
    {
      "epoch": 1.005281673203285,
      "grad_norm": 0.02180866152048111,
      "learning_rate": 5.822810235661011e-06,
      "loss": 0.0075,
      "step": 156740
    },
    {
      "epoch": 1.0053458100970711,
      "grad_norm": 0.2020481824874878,
      "learning_rate": 5.822258157717704e-06,
      "loss": 0.0019,
      "step": 156750
    },
    {
      "epoch": 1.0054099469908573,
      "grad_norm": 0.0793982520699501,
      "learning_rate": 5.821706069470913e-06,
      "loss": 0.001,
      "step": 156760
    },
    {
      "epoch": 1.0054740838846434,
      "grad_norm": 0.13440971076488495,
      "learning_rate": 5.821153970927558e-06,
      "loss": 0.0014,
      "step": 156770
    },
    {
      "epoch": 1.0055382207784296,
      "grad_norm": 0.12930013239383698,
      "learning_rate": 5.820601862094556e-06,
      "loss": 0.0018,
      "step": 156780
    },
    {
      "epoch": 1.0056023576722155,
      "grad_norm": 0.02354401908814907,
      "learning_rate": 5.8200497429788275e-06,
      "loss": 0.0016,
      "step": 156790
    },
    {
      "epoch": 1.0056664945660017,
      "grad_norm": 0.0333816297352314,
      "learning_rate": 5.819497613587288e-06,
      "loss": 0.0022,
      "step": 156800
    },
    {
      "epoch": 1.0057306314597878,
      "grad_norm": 0.030010614544153214,
      "learning_rate": 5.81894547392686e-06,
      "loss": 0.0021,
      "step": 156810
    },
    {
      "epoch": 1.005794768353574,
      "grad_norm": 0.06879100948572159,
      "learning_rate": 5.818393324004458e-06,
      "loss": 0.0017,
      "step": 156820
    },
    {
      "epoch": 1.0058589052473599,
      "grad_norm": 0.0591420978307724,
      "learning_rate": 5.817841163827004e-06,
      "loss": 0.002,
      "step": 156830
    },
    {
      "epoch": 1.005923042141146,
      "grad_norm": 0.03935656696557999,
      "learning_rate": 5.8172889934014135e-06,
      "loss": 0.0023,
      "step": 156840
    },
    {
      "epoch": 1.0059871790349322,
      "grad_norm": 0.07783859223127365,
      "learning_rate": 5.81673681273461e-06,
      "loss": 0.0018,
      "step": 156850
    },
    {
      "epoch": 1.0060513159287183,
      "grad_norm": 0.03830932825803757,
      "learning_rate": 5.8161846218335095e-06,
      "loss": 0.0026,
      "step": 156860
    },
    {
      "epoch": 1.0061154528225043,
      "grad_norm": 0.050277289003133774,
      "learning_rate": 5.815632420705033e-06,
      "loss": 0.0017,
      "step": 156870
    },
    {
      "epoch": 1.0061795897162904,
      "grad_norm": 0.10210917890071869,
      "learning_rate": 5.815080209356099e-06,
      "loss": 0.0014,
      "step": 156880
    },
    {
      "epoch": 1.0062437266100765,
      "grad_norm": 0.00457958597689867,
      "learning_rate": 5.8145279877936275e-06,
      "loss": 0.0016,
      "step": 156890
    },
    {
      "epoch": 1.0063078635038627,
      "grad_norm": 0.0677952691912651,
      "learning_rate": 5.813975756024538e-06,
      "loss": 0.0017,
      "step": 156900
    },
    {
      "epoch": 1.0063720003976488,
      "grad_norm": 0.13198815286159515,
      "learning_rate": 5.8134235140557505e-06,
      "loss": 0.0019,
      "step": 156910
    },
    {
      "epoch": 1.0064361372914348,
      "grad_norm": 0.21554039418697357,
      "learning_rate": 5.812871261894186e-06,
      "loss": 0.0013,
      "step": 156920
    },
    {
      "epoch": 1.006500274185221,
      "grad_norm": 0.11818122863769531,
      "learning_rate": 5.8123189995467625e-06,
      "loss": 0.0016,
      "step": 156930
    },
    {
      "epoch": 1.006564411079007,
      "grad_norm": 0.17103898525238037,
      "learning_rate": 5.811766727020403e-06,
      "loss": 0.0016,
      "step": 156940
    },
    {
      "epoch": 1.0066285479727932,
      "grad_norm": 0.12196987867355347,
      "learning_rate": 5.811214444322024e-06,
      "loss": 0.0014,
      "step": 156950
    },
    {
      "epoch": 1.0066926848665791,
      "grad_norm": 0.05216473713517189,
      "learning_rate": 5.81066215145855e-06,
      "loss": 0.0012,
      "step": 156960
    },
    {
      "epoch": 1.0067568217603653,
      "grad_norm": 0.02345466986298561,
      "learning_rate": 5.8101098484368985e-06,
      "loss": 0.0014,
      "step": 156970
    },
    {
      "epoch": 1.0068209586541514,
      "grad_norm": 0.06300746649503708,
      "learning_rate": 5.809557535263993e-06,
      "loss": 0.0011,
      "step": 156980
    },
    {
      "epoch": 1.0068850955479376,
      "grad_norm": 0.11322963982820511,
      "learning_rate": 5.809005211946752e-06,
      "loss": 0.0024,
      "step": 156990
    },
    {
      "epoch": 1.0069492324417235,
      "grad_norm": 0.03215669095516205,
      "learning_rate": 5.808452878492098e-06,
      "loss": 0.0013,
      "step": 157000
    },
    {
      "epoch": 1.0070133693355097,
      "grad_norm": 0.02180369570851326,
      "learning_rate": 5.8079005349069516e-06,
      "loss": 0.0019,
      "step": 157010
    },
    {
      "epoch": 1.0070775062292958,
      "grad_norm": 0.20294401049613953,
      "learning_rate": 5.807348181198235e-06,
      "loss": 0.0031,
      "step": 157020
    },
    {
      "epoch": 1.007141643123082,
      "grad_norm": 0.13072869181632996,
      "learning_rate": 5.806795817372867e-06,
      "loss": 0.0025,
      "step": 157030
    },
    {
      "epoch": 1.0072057800168681,
      "grad_norm": 0.17437884211540222,
      "learning_rate": 5.806243443437771e-06,
      "loss": 0.0019,
      "step": 157040
    },
    {
      "epoch": 1.007269916910654,
      "grad_norm": 0.2740894556045532,
      "learning_rate": 5.805691059399869e-06,
      "loss": 0.0023,
      "step": 157050
    },
    {
      "epoch": 1.0073340538044402,
      "grad_norm": 0.055262111127376556,
      "learning_rate": 5.8051386652660815e-06,
      "loss": 0.0013,
      "step": 157060
    },
    {
      "epoch": 1.0073981906982263,
      "grad_norm": 0.05367450788617134,
      "learning_rate": 5.804586261043332e-06,
      "loss": 0.0026,
      "step": 157070
    },
    {
      "epoch": 1.0074623275920125,
      "grad_norm": 0.02127639576792717,
      "learning_rate": 5.8040338467385416e-06,
      "loss": 0.0014,
      "step": 157080
    },
    {
      "epoch": 1.0075264644857984,
      "grad_norm": 0.08617508411407471,
      "learning_rate": 5.803481422358632e-06,
      "loss": 0.0011,
      "step": 157090
    },
    {
      "epoch": 1.0075906013795846,
      "grad_norm": 0.08567603677511215,
      "learning_rate": 5.802928987910527e-06,
      "loss": 0.0018,
      "step": 157100
    },
    {
      "epoch": 1.0076547382733707,
      "grad_norm": 0.11544500291347504,
      "learning_rate": 5.802376543401147e-06,
      "loss": 0.0034,
      "step": 157110
    },
    {
      "epoch": 1.0077188751671569,
      "grad_norm": 0.15631578862667084,
      "learning_rate": 5.801824088837414e-06,
      "loss": 0.0019,
      "step": 157120
    },
    {
      "epoch": 1.0077830120609428,
      "grad_norm": 0.059519752860069275,
      "learning_rate": 5.801271624226255e-06,
      "loss": 0.0013,
      "step": 157130
    },
    {
      "epoch": 1.007847148954729,
      "grad_norm": 0.10539231449365616,
      "learning_rate": 5.800719149574588e-06,
      "loss": 0.0007,
      "step": 157140
    },
    {
      "epoch": 1.007911285848515,
      "grad_norm": 0.24564160406589508,
      "learning_rate": 5.80016666488934e-06,
      "loss": 0.0013,
      "step": 157150
    },
    {
      "epoch": 1.0079754227423012,
      "grad_norm": 0.07587874680757523,
      "learning_rate": 5.799614170177429e-06,
      "loss": 0.0036,
      "step": 157160
    },
    {
      "epoch": 1.0080395596360872,
      "grad_norm": 0.08727598190307617,
      "learning_rate": 5.799061665445782e-06,
      "loss": 0.0017,
      "step": 157170
    },
    {
      "epoch": 1.0081036965298733,
      "grad_norm": 0.07284372299909592,
      "learning_rate": 5.798509150701321e-06,
      "loss": 0.0021,
      "step": 157180
    },
    {
      "epoch": 1.0081678334236595,
      "grad_norm": 0.033877890557050705,
      "learning_rate": 5.79795662595097e-06,
      "loss": 0.002,
      "step": 157190
    },
    {
      "epoch": 1.0082319703174456,
      "grad_norm": 0.054351892322301865,
      "learning_rate": 5.797404091201653e-06,
      "loss": 0.001,
      "step": 157200
    },
    {
      "epoch": 1.0082961072112318,
      "grad_norm": 0.34961628913879395,
      "learning_rate": 5.796851546460292e-06,
      "loss": 0.0041,
      "step": 157210
    },
    {
      "epoch": 1.0083602441050177,
      "grad_norm": 0.06319102644920349,
      "learning_rate": 5.796298991733812e-06,
      "loss": 0.0018,
      "step": 157220
    },
    {
      "epoch": 1.0084243809988038,
      "grad_norm": 0.1530248373746872,
      "learning_rate": 5.795746427029136e-06,
      "loss": 0.0019,
      "step": 157230
    },
    {
      "epoch": 1.00848851789259,
      "grad_norm": 0.17742373049259186,
      "learning_rate": 5.795193852353187e-06,
      "loss": 0.001,
      "step": 157240
    },
    {
      "epoch": 1.0085526547863761,
      "grad_norm": 0.028137022629380226,
      "learning_rate": 5.794641267712894e-06,
      "loss": 0.0041,
      "step": 157250
    },
    {
      "epoch": 1.008616791680162,
      "grad_norm": 0.0809083953499794,
      "learning_rate": 5.794088673115175e-06,
      "loss": 0.0027,
      "step": 157260
    },
    {
      "epoch": 1.0086809285739482,
      "grad_norm": 0.028177518397569656,
      "learning_rate": 5.793536068566959e-06,
      "loss": 0.0011,
      "step": 157270
    },
    {
      "epoch": 1.0087450654677343,
      "grad_norm": 0.1421087384223938,
      "learning_rate": 5.79298345407517e-06,
      "loss": 0.0018,
      "step": 157280
    },
    {
      "epoch": 1.0088092023615205,
      "grad_norm": 0.13536611199378967,
      "learning_rate": 5.792430829646729e-06,
      "loss": 0.0033,
      "step": 157290
    },
    {
      "epoch": 1.0088733392553064,
      "grad_norm": 0.08360431343317032,
      "learning_rate": 5.791878195288565e-06,
      "loss": 0.0019,
      "step": 157300
    },
    {
      "epoch": 1.0089374761490926,
      "grad_norm": 0.053090013563632965,
      "learning_rate": 5.7913255510076e-06,
      "loss": 0.0038,
      "step": 157310
    },
    {
      "epoch": 1.0090016130428787,
      "grad_norm": 0.19956789910793304,
      "learning_rate": 5.7907728968107635e-06,
      "loss": 0.0022,
      "step": 157320
    },
    {
      "epoch": 1.0090657499366649,
      "grad_norm": 0.07794226706027985,
      "learning_rate": 5.790220232704974e-06,
      "loss": 0.0008,
      "step": 157330
    },
    {
      "epoch": 1.009129886830451,
      "grad_norm": 0.023244792595505714,
      "learning_rate": 5.789667558697163e-06,
      "loss": 0.0035,
      "step": 157340
    },
    {
      "epoch": 1.009194023724237,
      "grad_norm": 0.01863129809498787,
      "learning_rate": 5.789114874794251e-06,
      "loss": 0.0017,
      "step": 157350
    },
    {
      "epoch": 1.009258160618023,
      "grad_norm": 0.13609835505485535,
      "learning_rate": 5.788562181003167e-06,
      "loss": 0.002,
      "step": 157360
    },
    {
      "epoch": 1.0093222975118092,
      "grad_norm": 0.040573108941316605,
      "learning_rate": 5.788009477330834e-06,
      "loss": 0.002,
      "step": 157370
    },
    {
      "epoch": 1.0093864344055954,
      "grad_norm": 0.167245015501976,
      "learning_rate": 5.78745676378418e-06,
      "loss": 0.0026,
      "step": 157380
    },
    {
      "epoch": 1.0094505712993813,
      "grad_norm": 0.012963230721652508,
      "learning_rate": 5.786904040370128e-06,
      "loss": 0.0017,
      "step": 157390
    },
    {
      "epoch": 1.0095147081931675,
      "grad_norm": 0.13187387585639954,
      "learning_rate": 5.786351307095608e-06,
      "loss": 0.0017,
      "step": 157400
    },
    {
      "epoch": 1.0095788450869536,
      "grad_norm": 0.07060447335243225,
      "learning_rate": 5.785798563967544e-06,
      "loss": 0.0011,
      "step": 157410
    },
    {
      "epoch": 1.0096429819807398,
      "grad_norm": 0.06462021172046661,
      "learning_rate": 5.785245810992861e-06,
      "loss": 0.0017,
      "step": 157420
    },
    {
      "epoch": 1.0097071188745257,
      "grad_norm": 0.04903660714626312,
      "learning_rate": 5.784693048178489e-06,
      "loss": 0.0019,
      "step": 157430
    },
    {
      "epoch": 1.0097712557683118,
      "grad_norm": 0.18517525494098663,
      "learning_rate": 5.7841402755313504e-06,
      "loss": 0.0025,
      "step": 157440
    },
    {
      "epoch": 1.009835392662098,
      "grad_norm": 0.19298167526721954,
      "learning_rate": 5.783587493058376e-06,
      "loss": 0.0022,
      "step": 157450
    },
    {
      "epoch": 1.0098995295558841,
      "grad_norm": 0.004793180152773857,
      "learning_rate": 5.783034700766487e-06,
      "loss": 0.0025,
      "step": 157460
    },
    {
      "epoch": 1.0099636664496703,
      "grad_norm": 0.10318158566951752,
      "learning_rate": 5.782481898662616e-06,
      "loss": 0.0031,
      "step": 157470
    },
    {
      "epoch": 1.0100278033434562,
      "grad_norm": 0.04313928261399269,
      "learning_rate": 5.781929086753687e-06,
      "loss": 0.0048,
      "step": 157480
    },
    {
      "epoch": 1.0100919402372424,
      "grad_norm": 0.011004787869751453,
      "learning_rate": 5.7813762650466275e-06,
      "loss": 0.0015,
      "step": 157490
    },
    {
      "epoch": 1.0101560771310285,
      "grad_norm": 0.13316594064235687,
      "learning_rate": 5.7808234335483644e-06,
      "loss": 0.0016,
      "step": 157500
    },
    {
      "epoch": 1.0102202140248147,
      "grad_norm": 0.1499415785074234,
      "learning_rate": 5.780270592265827e-06,
      "loss": 0.0023,
      "step": 157510
    },
    {
      "epoch": 1.0102843509186006,
      "grad_norm": 0.0686010867357254,
      "learning_rate": 5.7797177412059414e-06,
      "loss": 0.0031,
      "step": 157520
    },
    {
      "epoch": 1.0103484878123867,
      "grad_norm": 0.04718811437487602,
      "learning_rate": 5.779164880375635e-06,
      "loss": 0.0018,
      "step": 157530
    },
    {
      "epoch": 1.0104126247061729,
      "grad_norm": 0.05614418536424637,
      "learning_rate": 5.778612009781834e-06,
      "loss": 0.0015,
      "step": 157540
    },
    {
      "epoch": 1.010476761599959,
      "grad_norm": 0.11073336005210876,
      "learning_rate": 5.778059129431469e-06,
      "loss": 0.0015,
      "step": 157550
    },
    {
      "epoch": 1.010540898493745,
      "grad_norm": 0.006827982142567635,
      "learning_rate": 5.777506239331468e-06,
      "loss": 0.0028,
      "step": 157560
    },
    {
      "epoch": 1.010605035387531,
      "grad_norm": 0.15427444875240326,
      "learning_rate": 5.776953339488758e-06,
      "loss": 0.0018,
      "step": 157570
    },
    {
      "epoch": 1.0106691722813173,
      "grad_norm": 0.24773503839969635,
      "learning_rate": 5.776400429910267e-06,
      "loss": 0.0031,
      "step": 157580
    },
    {
      "epoch": 1.0107333091751034,
      "grad_norm": 0.03619404137134552,
      "learning_rate": 5.7758475106029236e-06,
      "loss": 0.0012,
      "step": 157590
    },
    {
      "epoch": 1.0107974460688895,
      "grad_norm": 0.07290811091661453,
      "learning_rate": 5.7752945815736574e-06,
      "loss": 0.001,
      "step": 157600
    },
    {
      "epoch": 1.0108615829626755,
      "grad_norm": 0.052982788532972336,
      "learning_rate": 5.7747416428293935e-06,
      "loss": 0.0014,
      "step": 157610
    },
    {
      "epoch": 1.0109257198564616,
      "grad_norm": 0.021865442395210266,
      "learning_rate": 5.774188694377066e-06,
      "loss": 0.0025,
      "step": 157620
    },
    {
      "epoch": 1.0109898567502478,
      "grad_norm": 0.2732413411140442,
      "learning_rate": 5.773635736223598e-06,
      "loss": 0.0019,
      "step": 157630
    },
    {
      "epoch": 1.011053993644034,
      "grad_norm": 0.36807775497436523,
      "learning_rate": 5.773082768375924e-06,
      "loss": 0.0037,
      "step": 157640
    },
    {
      "epoch": 1.0111181305378198,
      "grad_norm": 0.08362676203250885,
      "learning_rate": 5.772529790840968e-06,
      "loss": 0.0022,
      "step": 157650
    },
    {
      "epoch": 1.011182267431606,
      "grad_norm": 0.1202588900923729,
      "learning_rate": 5.771976803625664e-06,
      "loss": 0.0018,
      "step": 157660
    },
    {
      "epoch": 1.0112464043253921,
      "grad_norm": 0.08706741034984589,
      "learning_rate": 5.771423806736938e-06,
      "loss": 0.0025,
      "step": 157670
    },
    {
      "epoch": 1.0113105412191783,
      "grad_norm": 0.12375295907258987,
      "learning_rate": 5.770870800181721e-06,
      "loss": 0.0063,
      "step": 157680
    },
    {
      "epoch": 1.0113746781129642,
      "grad_norm": 0.034464508295059204,
      "learning_rate": 5.77031778396694e-06,
      "loss": 0.001,
      "step": 157690
    },
    {
      "epoch": 1.0114388150067504,
      "grad_norm": 0.15537671744823456,
      "learning_rate": 5.769764758099528e-06,
      "loss": 0.0017,
      "step": 157700
    },
    {
      "epoch": 1.0115029519005365,
      "grad_norm": 0.10750894993543625,
      "learning_rate": 5.769211722586413e-06,
      "loss": 0.0015,
      "step": 157710
    },
    {
      "epoch": 1.0115670887943227,
      "grad_norm": 0.03844396024942398,
      "learning_rate": 5.7686586774345265e-06,
      "loss": 0.0027,
      "step": 157720
    },
    {
      "epoch": 1.0116312256881086,
      "grad_norm": 0.02411005087196827,
      "learning_rate": 5.768105622650796e-06,
      "loss": 0.0014,
      "step": 157730
    },
    {
      "epoch": 1.0116953625818947,
      "grad_norm": 0.08573506027460098,
      "learning_rate": 5.767552558242153e-06,
      "loss": 0.0011,
      "step": 157740
    },
    {
      "epoch": 1.011759499475681,
      "grad_norm": 0.14994917809963226,
      "learning_rate": 5.766999484215529e-06,
      "loss": 0.0016,
      "step": 157750
    },
    {
      "epoch": 1.011823636369467,
      "grad_norm": 0.11491748690605164,
      "learning_rate": 5.766446400577852e-06,
      "loss": 0.0013,
      "step": 157760
    },
    {
      "epoch": 1.0118877732632532,
      "grad_norm": 0.05437860265374184,
      "learning_rate": 5.765893307336055e-06,
      "loss": 0.005,
      "step": 157770
    },
    {
      "epoch": 1.0119519101570391,
      "grad_norm": 0.16857770085334778,
      "learning_rate": 5.765340204497066e-06,
      "loss": 0.0015,
      "step": 157780
    },
    {
      "epoch": 1.0120160470508253,
      "grad_norm": 0.1880984902381897,
      "learning_rate": 5.76478709206782e-06,
      "loss": 0.0022,
      "step": 157790
    },
    {
      "epoch": 1.0120801839446114,
      "grad_norm": 0.09091341495513916,
      "learning_rate": 5.764233970055243e-06,
      "loss": 0.0021,
      "step": 157800
    },
    {
      "epoch": 1.0121443208383976,
      "grad_norm": 0.041069868952035904,
      "learning_rate": 5.763680838466269e-06,
      "loss": 0.0011,
      "step": 157810
    },
    {
      "epoch": 1.0122084577321835,
      "grad_norm": 0.07052157074213028,
      "learning_rate": 5.763127697307828e-06,
      "loss": 0.0039,
      "step": 157820
    },
    {
      "epoch": 1.0122725946259696,
      "grad_norm": 0.11400876939296722,
      "learning_rate": 5.762574546586852e-06,
      "loss": 0.0014,
      "step": 157830
    },
    {
      "epoch": 1.0123367315197558,
      "grad_norm": 0.0032041589729487896,
      "learning_rate": 5.76202138631027e-06,
      "loss": 0.0015,
      "step": 157840
    },
    {
      "epoch": 1.012400868413542,
      "grad_norm": 0.1367321014404297,
      "learning_rate": 5.7614682164850176e-06,
      "loss": 0.002,
      "step": 157850
    },
    {
      "epoch": 1.0124650053073279,
      "grad_norm": 0.1709347665309906,
      "learning_rate": 5.7609150371180234e-06,
      "loss": 0.0012,
      "step": 157860
    },
    {
      "epoch": 1.012529142201114,
      "grad_norm": 0.12345244735479355,
      "learning_rate": 5.760361848216221e-06,
      "loss": 0.0019,
      "step": 157870
    },
    {
      "epoch": 1.0125932790949002,
      "grad_norm": 0.04564058780670166,
      "learning_rate": 5.759808649786539e-06,
      "loss": 0.0013,
      "step": 157880
    },
    {
      "epoch": 1.0126574159886863,
      "grad_norm": 0.2694692313671112,
      "learning_rate": 5.759255441835914e-06,
      "loss": 0.0031,
      "step": 157890
    },
    {
      "epoch": 1.0127215528824725,
      "grad_norm": 0.15322737395763397,
      "learning_rate": 5.758702224371274e-06,
      "loss": 0.0022,
      "step": 157900
    },
    {
      "epoch": 1.0127856897762584,
      "grad_norm": 0.11926686018705368,
      "learning_rate": 5.758148997399553e-06,
      "loss": 0.0011,
      "step": 157910
    },
    {
      "epoch": 1.0128498266700445,
      "grad_norm": 0.056959182024002075,
      "learning_rate": 5.7575957609276845e-06,
      "loss": 0.0017,
      "step": 157920
    },
    {
      "epoch": 1.0129139635638307,
      "grad_norm": 0.05914493277668953,
      "learning_rate": 5.757042514962599e-06,
      "loss": 0.0027,
      "step": 157930
    },
    {
      "epoch": 1.0129781004576168,
      "grad_norm": 0.0504847951233387,
      "learning_rate": 5.756489259511228e-06,
      "loss": 0.002,
      "step": 157940
    },
    {
      "epoch": 1.0130422373514028,
      "grad_norm": 0.07084763795137405,
      "learning_rate": 5.755935994580508e-06,
      "loss": 0.0014,
      "step": 157950
    },
    {
      "epoch": 1.013106374245189,
      "grad_norm": 0.11007464677095413,
      "learning_rate": 5.75538272017737e-06,
      "loss": 0.0014,
      "step": 157960
    },
    {
      "epoch": 1.013170511138975,
      "grad_norm": 0.028109176084399223,
      "learning_rate": 5.754829436308745e-06,
      "loss": 0.0023,
      "step": 157970
    },
    {
      "epoch": 1.0132346480327612,
      "grad_norm": 0.05885981023311615,
      "learning_rate": 5.75427614298157e-06,
      "loss": 0.0014,
      "step": 157980
    },
    {
      "epoch": 1.0132987849265471,
      "grad_norm": 0.020199328660964966,
      "learning_rate": 5.753722840202772e-06,
      "loss": 0.0013,
      "step": 157990
    },
    {
      "epoch": 1.0133629218203333,
      "grad_norm": 0.29015612602233887,
      "learning_rate": 5.753169527979292e-06,
      "loss": 0.002,
      "step": 158000
    },
    {
      "epoch": 1.0134270587141194,
      "grad_norm": 0.27271541953086853,
      "learning_rate": 5.752616206318057e-06,
      "loss": 0.003,
      "step": 158010
    },
    {
      "epoch": 1.0134911956079056,
      "grad_norm": 0.11496245115995407,
      "learning_rate": 5.752062875226006e-06,
      "loss": 0.0018,
      "step": 158020
    },
    {
      "epoch": 1.0135553325016917,
      "grad_norm": 0.06919115036725998,
      "learning_rate": 5.751509534710066e-06,
      "loss": 0.0032,
      "step": 158030
    },
    {
      "epoch": 1.0136194693954776,
      "grad_norm": 0.04158717393875122,
      "learning_rate": 5.750956184777176e-06,
      "loss": 0.0012,
      "step": 158040
    },
    {
      "epoch": 1.0136836062892638,
      "grad_norm": 0.11757807433605194,
      "learning_rate": 5.750402825434269e-06,
      "loss": 0.0039,
      "step": 158050
    },
    {
      "epoch": 1.01374774318305,
      "grad_norm": 0.021763648837804794,
      "learning_rate": 5.749849456688279e-06,
      "loss": 0.0009,
      "step": 158060
    },
    {
      "epoch": 1.013811880076836,
      "grad_norm": 0.02642359957098961,
      "learning_rate": 5.749296078546137e-06,
      "loss": 0.0023,
      "step": 158070
    },
    {
      "epoch": 1.013876016970622,
      "grad_norm": 0.029139714315533638,
      "learning_rate": 5.748742691014781e-06,
      "loss": 0.0017,
      "step": 158080
    },
    {
      "epoch": 1.0139401538644082,
      "grad_norm": 0.20166301727294922,
      "learning_rate": 5.748189294101144e-06,
      "loss": 0.0022,
      "step": 158090
    },
    {
      "epoch": 1.0140042907581943,
      "grad_norm": 0.05873652175068855,
      "learning_rate": 5.747635887812161e-06,
      "loss": 0.0011,
      "step": 158100
    },
    {
      "epoch": 1.0140684276519805,
      "grad_norm": 0.12319185584783554,
      "learning_rate": 5.747082472154764e-06,
      "loss": 0.0012,
      "step": 158110
    },
    {
      "epoch": 1.0141325645457664,
      "grad_norm": 0.07644834369421005,
      "learning_rate": 5.746529047135891e-06,
      "loss": 0.0018,
      "step": 158120
    },
    {
      "epoch": 1.0141967014395525,
      "grad_norm": 0.04374532029032707,
      "learning_rate": 5.745975612762477e-06,
      "loss": 0.004,
      "step": 158130
    },
    {
      "epoch": 1.0142608383333387,
      "grad_norm": 0.0785631611943245,
      "learning_rate": 5.7454221690414524e-06,
      "loss": 0.0026,
      "step": 158140
    },
    {
      "epoch": 1.0143249752271248,
      "grad_norm": 0.12788932025432587,
      "learning_rate": 5.744868715979758e-06,
      "loss": 0.0026,
      "step": 158150
    },
    {
      "epoch": 1.0143891121209108,
      "grad_norm": 0.1302683800458908,
      "learning_rate": 5.7443152535843235e-06,
      "loss": 0.0027,
      "step": 158160
    },
    {
      "epoch": 1.014453249014697,
      "grad_norm": 0.23391394317150116,
      "learning_rate": 5.743761781862089e-06,
      "loss": 0.0016,
      "step": 158170
    },
    {
      "epoch": 1.014517385908483,
      "grad_norm": 0.1533360779285431,
      "learning_rate": 5.7432083008199876e-06,
      "loss": 0.0039,
      "step": 158180
    },
    {
      "epoch": 1.0145815228022692,
      "grad_norm": 0.03090854361653328,
      "learning_rate": 5.742654810464955e-06,
      "loss": 0.0045,
      "step": 158190
    },
    {
      "epoch": 1.0146456596960554,
      "grad_norm": 0.1253776103258133,
      "learning_rate": 5.742101310803926e-06,
      "loss": 0.0023,
      "step": 158200
    },
    {
      "epoch": 1.0147097965898413,
      "grad_norm": 0.1397521197795868,
      "learning_rate": 5.741547801843839e-06,
      "loss": 0.0015,
      "step": 158210
    },
    {
      "epoch": 1.0147739334836274,
      "grad_norm": 0.03552157059311867,
      "learning_rate": 5.740994283591626e-06,
      "loss": 0.0017,
      "step": 158220
    },
    {
      "epoch": 1.0148380703774136,
      "grad_norm": 0.10333353281021118,
      "learning_rate": 5.740440756054226e-06,
      "loss": 0.0012,
      "step": 158230
    },
    {
      "epoch": 1.0149022072711997,
      "grad_norm": 0.0401446633040905,
      "learning_rate": 5.739887219238574e-06,
      "loss": 0.0018,
      "step": 158240
    },
    {
      "epoch": 1.0149663441649857,
      "grad_norm": 0.2060912698507309,
      "learning_rate": 5.739333673151606e-06,
      "loss": 0.0015,
      "step": 158250
    },
    {
      "epoch": 1.0150304810587718,
      "grad_norm": 0.07493453472852707,
      "learning_rate": 5.738780117800259e-06,
      "loss": 0.0016,
      "step": 158260
    },
    {
      "epoch": 1.015094617952558,
      "grad_norm": 0.43839237093925476,
      "learning_rate": 5.738226553191468e-06,
      "loss": 0.0015,
      "step": 158270
    },
    {
      "epoch": 1.015158754846344,
      "grad_norm": 0.09643466025590897,
      "learning_rate": 5.737672979332173e-06,
      "loss": 0.0025,
      "step": 158280
    },
    {
      "epoch": 1.01522289174013,
      "grad_norm": 0.13256953656673431,
      "learning_rate": 5.737119396229307e-06,
      "loss": 0.0024,
      "step": 158290
    },
    {
      "epoch": 1.0152870286339162,
      "grad_norm": 0.24836604297161102,
      "learning_rate": 5.736565803889809e-06,
      "loss": 0.002,
      "step": 158300
    },
    {
      "epoch": 1.0153511655277023,
      "grad_norm": 0.027909912168979645,
      "learning_rate": 5.736012202320614e-06,
      "loss": 0.002,
      "step": 158310
    },
    {
      "epoch": 1.0154153024214885,
      "grad_norm": 0.1804720014333725,
      "learning_rate": 5.735458591528661e-06,
      "loss": 0.0025,
      "step": 158320
    },
    {
      "epoch": 1.0154794393152746,
      "grad_norm": 0.08508358895778656,
      "learning_rate": 5.734904971520885e-06,
      "loss": 0.0007,
      "step": 158330
    },
    {
      "epoch": 1.0155435762090606,
      "grad_norm": 0.2949533462524414,
      "learning_rate": 5.734351342304227e-06,
      "loss": 0.0041,
      "step": 158340
    },
    {
      "epoch": 1.0156077131028467,
      "grad_norm": 0.22988532483577728,
      "learning_rate": 5.7337977038856186e-06,
      "loss": 0.001,
      "step": 158350
    },
    {
      "epoch": 1.0156718499966328,
      "grad_norm": 0.015086526051163673,
      "learning_rate": 5.733244056272003e-06,
      "loss": 0.0028,
      "step": 158360
    },
    {
      "epoch": 1.015735986890419,
      "grad_norm": 0.1951785832643509,
      "learning_rate": 5.732690399470313e-06,
      "loss": 0.0026,
      "step": 158370
    },
    {
      "epoch": 1.015800123784205,
      "grad_norm": 0.07197275757789612,
      "learning_rate": 5.73213673348749e-06,
      "loss": 0.0015,
      "step": 158380
    },
    {
      "epoch": 1.015864260677991,
      "grad_norm": 0.30711835622787476,
      "learning_rate": 5.7315830583304695e-06,
      "loss": 0.002,
      "step": 158390
    },
    {
      "epoch": 1.0159283975717772,
      "grad_norm": 0.053068507462739944,
      "learning_rate": 5.731029374006192e-06,
      "loss": 0.0014,
      "step": 158400
    },
    {
      "epoch": 1.0159925344655634,
      "grad_norm": 0.05254053696990013,
      "learning_rate": 5.730475680521593e-06,
      "loss": 0.0023,
      "step": 158410
    },
    {
      "epoch": 1.0160566713593493,
      "grad_norm": 0.05597158893942833,
      "learning_rate": 5.7299219778836125e-06,
      "loss": 0.0016,
      "step": 158420
    },
    {
      "epoch": 1.0161208082531354,
      "grad_norm": 0.14286933839321136,
      "learning_rate": 5.729368266099186e-06,
      "loss": 0.0028,
      "step": 158430
    },
    {
      "epoch": 1.0161849451469216,
      "grad_norm": 0.04776264727115631,
      "learning_rate": 5.728814545175256e-06,
      "loss": 0.0015,
      "step": 158440
    },
    {
      "epoch": 1.0162490820407077,
      "grad_norm": 0.07032934576272964,
      "learning_rate": 5.728260815118759e-06,
      "loss": 0.0022,
      "step": 158450
    },
    {
      "epoch": 1.016313218934494,
      "grad_norm": 0.1985510289669037,
      "learning_rate": 5.727707075936632e-06,
      "loss": 0.0021,
      "step": 158460
    },
    {
      "epoch": 1.0163773558282798,
      "grad_norm": 0.05037515610456467,
      "learning_rate": 5.727153327635816e-06,
      "loss": 0.0025,
      "step": 158470
    },
    {
      "epoch": 1.016441492722066,
      "grad_norm": 0.05321979522705078,
      "learning_rate": 5.726599570223249e-06,
      "loss": 0.0014,
      "step": 158480
    },
    {
      "epoch": 1.0165056296158521,
      "grad_norm": 0.23675918579101562,
      "learning_rate": 5.726045803705871e-06,
      "loss": 0.0021,
      "step": 158490
    },
    {
      "epoch": 1.0165697665096383,
      "grad_norm": 0.0526350662112236,
      "learning_rate": 5.725492028090619e-06,
      "loss": 0.0023,
      "step": 158500
    },
    {
      "epoch": 1.0166339034034242,
      "grad_norm": 0.0651823952794075,
      "learning_rate": 5.7249382433844335e-06,
      "loss": 0.0051,
      "step": 158510
    },
    {
      "epoch": 1.0166980402972103,
      "grad_norm": 0.08328308165073395,
      "learning_rate": 5.724384449594253e-06,
      "loss": 0.0012,
      "step": 158520
    },
    {
      "epoch": 1.0167621771909965,
      "grad_norm": 0.1616986244916916,
      "learning_rate": 5.72383064672702e-06,
      "loss": 0.0014,
      "step": 158530
    },
    {
      "epoch": 1.0168263140847826,
      "grad_norm": 0.11192335188388824,
      "learning_rate": 5.72327683478967e-06,
      "loss": 0.001,
      "step": 158540
    },
    {
      "epoch": 1.0168904509785686,
      "grad_norm": 0.08834236860275269,
      "learning_rate": 5.722723013789144e-06,
      "loss": 0.0017,
      "step": 158550
    },
    {
      "epoch": 1.0169545878723547,
      "grad_norm": 0.019593428820371628,
      "learning_rate": 5.722169183732383e-06,
      "loss": 0.0013,
      "step": 158560
    },
    {
      "epoch": 1.0170187247661409,
      "grad_norm": 0.3054475784301758,
      "learning_rate": 5.721615344626327e-06,
      "loss": 0.0017,
      "step": 158570
    },
    {
      "epoch": 1.017082861659927,
      "grad_norm": 0.07718338072299957,
      "learning_rate": 5.721061496477913e-06,
      "loss": 0.0025,
      "step": 158580
    },
    {
      "epoch": 1.017146998553713,
      "grad_norm": 0.15053144097328186,
      "learning_rate": 5.720507639294084e-06,
      "loss": 0.0016,
      "step": 158590
    },
    {
      "epoch": 1.017211135447499,
      "grad_norm": 0.1066005602478981,
      "learning_rate": 5.719953773081779e-06,
      "loss": 0.0024,
      "step": 158600
    },
    {
      "epoch": 1.0172752723412852,
      "grad_norm": 0.15089969336986542,
      "learning_rate": 5.719399897847938e-06,
      "loss": 0.0026,
      "step": 158610
    },
    {
      "epoch": 1.0173394092350714,
      "grad_norm": 0.09610164165496826,
      "learning_rate": 5.718846013599504e-06,
      "loss": 0.0015,
      "step": 158620
    },
    {
      "epoch": 1.0174035461288575,
      "grad_norm": 0.16944530606269836,
      "learning_rate": 5.718292120343414e-06,
      "loss": 0.0017,
      "step": 158630
    },
    {
      "epoch": 1.0174676830226435,
      "grad_norm": 0.08744259178638458,
      "learning_rate": 5.7177382180866115e-06,
      "loss": 0.001,
      "step": 158640
    },
    {
      "epoch": 1.0175318199164296,
      "grad_norm": 0.22927281260490417,
      "learning_rate": 5.717184306836036e-06,
      "loss": 0.003,
      "step": 158650
    },
    {
      "epoch": 1.0175959568102158,
      "grad_norm": 0.04453054070472717,
      "learning_rate": 5.716630386598628e-06,
      "loss": 0.0013,
      "step": 158660
    },
    {
      "epoch": 1.017660093704002,
      "grad_norm": 0.09694628417491913,
      "learning_rate": 5.716076457381329e-06,
      "loss": 0.0025,
      "step": 158670
    },
    {
      "epoch": 1.0177242305977878,
      "grad_norm": 0.05184922739863396,
      "learning_rate": 5.715522519191081e-06,
      "loss": 0.0019,
      "step": 158680
    },
    {
      "epoch": 1.017788367491574,
      "grad_norm": 0.026720192283391953,
      "learning_rate": 5.7149685720348235e-06,
      "loss": 0.0017,
      "step": 158690
    },
    {
      "epoch": 1.0178525043853601,
      "grad_norm": 0.08675410598516464,
      "learning_rate": 5.7144146159195e-06,
      "loss": 0.0077,
      "step": 158700
    },
    {
      "epoch": 1.0179166412791463,
      "grad_norm": 0.15461985766887665,
      "learning_rate": 5.71386065085205e-06,
      "loss": 0.0019,
      "step": 158710
    },
    {
      "epoch": 1.0179807781729322,
      "grad_norm": 0.01148910541087389,
      "learning_rate": 5.7133066768394165e-06,
      "loss": 0.0022,
      "step": 158720
    },
    {
      "epoch": 1.0180449150667183,
      "grad_norm": 0.07814265042543411,
      "learning_rate": 5.712752693888539e-06,
      "loss": 0.0022,
      "step": 158730
    },
    {
      "epoch": 1.0181090519605045,
      "grad_norm": 0.1866464614868164,
      "learning_rate": 5.712198702006363e-06,
      "loss": 0.0029,
      "step": 158740
    },
    {
      "epoch": 1.0181731888542906,
      "grad_norm": 0.12418685853481293,
      "learning_rate": 5.711644701199827e-06,
      "loss": 0.0022,
      "step": 158750
    },
    {
      "epoch": 1.0182373257480768,
      "grad_norm": 0.047584936022758484,
      "learning_rate": 5.711090691475874e-06,
      "loss": 0.0016,
      "step": 158760
    },
    {
      "epoch": 1.0183014626418627,
      "grad_norm": 0.12022664397954941,
      "learning_rate": 5.710536672841446e-06,
      "loss": 0.002,
      "step": 158770
    },
    {
      "epoch": 1.0183655995356489,
      "grad_norm": 0.1303713619709015,
      "learning_rate": 5.709982645303487e-06,
      "loss": 0.0028,
      "step": 158780
    },
    {
      "epoch": 1.018429736429435,
      "grad_norm": 0.10671041160821915,
      "learning_rate": 5.709428608868937e-06,
      "loss": 0.0012,
      "step": 158790
    },
    {
      "epoch": 1.0184938733232212,
      "grad_norm": 0.2769628167152405,
      "learning_rate": 5.70887456354474e-06,
      "loss": 0.0013,
      "step": 158800
    },
    {
      "epoch": 1.018558010217007,
      "grad_norm": 0.04816930368542671,
      "learning_rate": 5.708320509337839e-06,
      "loss": 0.0022,
      "step": 158810
    },
    {
      "epoch": 1.0186221471107932,
      "grad_norm": 0.009474515914916992,
      "learning_rate": 5.707766446255174e-06,
      "loss": 0.0016,
      "step": 158820
    },
    {
      "epoch": 1.0186862840045794,
      "grad_norm": 0.061292871832847595,
      "learning_rate": 5.707212374303691e-06,
      "loss": 0.0023,
      "step": 158830
    },
    {
      "epoch": 1.0187504208983655,
      "grad_norm": 0.06681514531373978,
      "learning_rate": 5.706658293490331e-06,
      "loss": 0.0025,
      "step": 158840
    },
    {
      "epoch": 1.0188145577921515,
      "grad_norm": 0.42780590057373047,
      "learning_rate": 5.706104203822038e-06,
      "loss": 0.0018,
      "step": 158850
    },
    {
      "epoch": 1.0188786946859376,
      "grad_norm": 0.025822345167398453,
      "learning_rate": 5.705550105305754e-06,
      "loss": 0.0019,
      "step": 158860
    },
    {
      "epoch": 1.0189428315797238,
      "grad_norm": 0.21344026923179626,
      "learning_rate": 5.704995997948424e-06,
      "loss": 0.0023,
      "step": 158870
    },
    {
      "epoch": 1.01900696847351,
      "grad_norm": 0.09337637573480606,
      "learning_rate": 5.704441881756989e-06,
      "loss": 0.0018,
      "step": 158880
    },
    {
      "epoch": 1.019071105367296,
      "grad_norm": 0.06411875039339066,
      "learning_rate": 5.703887756738394e-06,
      "loss": 0.0028,
      "step": 158890
    },
    {
      "epoch": 1.019135242261082,
      "grad_norm": 0.1680552363395691,
      "learning_rate": 5.703333622899583e-06,
      "loss": 0.0021,
      "step": 158900
    },
    {
      "epoch": 1.0191993791548681,
      "grad_norm": 0.07275000214576721,
      "learning_rate": 5.702779480247499e-06,
      "loss": 0.0019,
      "step": 158910
    },
    {
      "epoch": 1.0192635160486543,
      "grad_norm": 0.11038286238908768,
      "learning_rate": 5.702225328789085e-06,
      "loss": 0.0018,
      "step": 158920
    },
    {
      "epoch": 1.0193276529424404,
      "grad_norm": 0.14055235683918,
      "learning_rate": 5.701671168531287e-06,
      "loss": 0.0012,
      "step": 158930
    },
    {
      "epoch": 1.0193917898362264,
      "grad_norm": 0.01654187962412834,
      "learning_rate": 5.701116999481047e-06,
      "loss": 0.0011,
      "step": 158940
    },
    {
      "epoch": 1.0194559267300125,
      "grad_norm": 0.06248614937067032,
      "learning_rate": 5.70056282164531e-06,
      "loss": 0.002,
      "step": 158950
    },
    {
      "epoch": 1.0195200636237987,
      "grad_norm": 0.1035081297159195,
      "learning_rate": 5.700008635031021e-06,
      "loss": 0.0021,
      "step": 158960
    },
    {
      "epoch": 1.0195842005175848,
      "grad_norm": 0.1015050858259201,
      "learning_rate": 5.699454439645121e-06,
      "loss": 0.0022,
      "step": 158970
    },
    {
      "epoch": 1.0196483374113707,
      "grad_norm": 0.2446785718202591,
      "learning_rate": 5.698900235494559e-06,
      "loss": 0.0017,
      "step": 158980
    },
    {
      "epoch": 1.0197124743051569,
      "grad_norm": 0.02375601790845394,
      "learning_rate": 5.698346022586276e-06,
      "loss": 0.0055,
      "step": 158990
    },
    {
      "epoch": 1.019776611198943,
      "grad_norm": 0.08211459219455719,
      "learning_rate": 5.6977918009272205e-06,
      "loss": 0.0012,
      "step": 159000
    },
    {
      "epoch": 1.0198407480927292,
      "grad_norm": 0.05030378699302673,
      "learning_rate": 5.6972375705243345e-06,
      "loss": 0.0019,
      "step": 159010
    },
    {
      "epoch": 1.0199048849865153,
      "grad_norm": 0.10835476964712143,
      "learning_rate": 5.696683331384563e-06,
      "loss": 0.0014,
      "step": 159020
    },
    {
      "epoch": 1.0199690218803013,
      "grad_norm": 0.13494984805583954,
      "learning_rate": 5.6961290835148505e-06,
      "loss": 0.0031,
      "step": 159030
    },
    {
      "epoch": 1.0200331587740874,
      "grad_norm": 0.040680792182683945,
      "learning_rate": 5.6955748269221445e-06,
      "loss": 0.001,
      "step": 159040
    },
    {
      "epoch": 1.0200972956678735,
      "grad_norm": 0.07557686418294907,
      "learning_rate": 5.695020561613388e-06,
      "loss": 0.0017,
      "step": 159050
    },
    {
      "epoch": 1.0201614325616597,
      "grad_norm": 0.10823880881071091,
      "learning_rate": 5.694466287595528e-06,
      "loss": 0.0021,
      "step": 159060
    },
    {
      "epoch": 1.0202255694554456,
      "grad_norm": 0.03785393759608269,
      "learning_rate": 5.693912004875508e-06,
      "loss": 0.0032,
      "step": 159070
    },
    {
      "epoch": 1.0202897063492318,
      "grad_norm": 0.2585473656654358,
      "learning_rate": 5.693357713460276e-06,
      "loss": 0.0026,
      "step": 159080
    },
    {
      "epoch": 1.020353843243018,
      "grad_norm": 0.21977251768112183,
      "learning_rate": 5.692803413356774e-06,
      "loss": 0.0022,
      "step": 159090
    },
    {
      "epoch": 1.020417980136804,
      "grad_norm": 0.0885687991976738,
      "learning_rate": 5.692249104571951e-06,
      "loss": 0.0009,
      "step": 159100
    },
    {
      "epoch": 1.02048211703059,
      "grad_norm": 0.10606767237186432,
      "learning_rate": 5.6916947871127535e-06,
      "loss": 0.0032,
      "step": 159110
    },
    {
      "epoch": 1.0205462539243761,
      "grad_norm": 0.19184932112693787,
      "learning_rate": 5.6911404609861245e-06,
      "loss": 0.0018,
      "step": 159120
    },
    {
      "epoch": 1.0206103908181623,
      "grad_norm": 0.1128690093755722,
      "learning_rate": 5.690586126199012e-06,
      "loss": 0.0023,
      "step": 159130
    },
    {
      "epoch": 1.0206745277119484,
      "grad_norm": 0.12977398931980133,
      "learning_rate": 5.690031782758362e-06,
      "loss": 0.0024,
      "step": 159140
    },
    {
      "epoch": 1.0207386646057346,
      "grad_norm": 0.20466890931129456,
      "learning_rate": 5.6894774306711194e-06,
      "loss": 0.0018,
      "step": 159150
    },
    {
      "epoch": 1.0208028014995205,
      "grad_norm": 0.13838985562324524,
      "learning_rate": 5.688923069944232e-06,
      "loss": 0.0017,
      "step": 159160
    },
    {
      "epoch": 1.0208669383933067,
      "grad_norm": 0.07172899693250656,
      "learning_rate": 5.6883687005846475e-06,
      "loss": 0.0013,
      "step": 159170
    },
    {
      "epoch": 1.0209310752870928,
      "grad_norm": 0.014331742189824581,
      "learning_rate": 5.687814322599311e-06,
      "loss": 0.0026,
      "step": 159180
    },
    {
      "epoch": 1.020995212180879,
      "grad_norm": 0.34021681547164917,
      "learning_rate": 5.6872599359951695e-06,
      "loss": 0.0016,
      "step": 159190
    },
    {
      "epoch": 1.021059349074665,
      "grad_norm": 0.3392062485218048,
      "learning_rate": 5.686705540779169e-06,
      "loss": 0.0014,
      "step": 159200
    },
    {
      "epoch": 1.021123485968451,
      "grad_norm": 0.11961116641759872,
      "learning_rate": 5.686151136958258e-06,
      "loss": 0.0019,
      "step": 159210
    },
    {
      "epoch": 1.0211876228622372,
      "grad_norm": 0.04757817089557648,
      "learning_rate": 5.685596724539383e-06,
      "loss": 0.0018,
      "step": 159220
    },
    {
      "epoch": 1.0212517597560233,
      "grad_norm": 0.009403668344020844,
      "learning_rate": 5.685042303529491e-06,
      "loss": 0.0021,
      "step": 159230
    },
    {
      "epoch": 1.0213158966498093,
      "grad_norm": 0.05813450366258621,
      "learning_rate": 5.684487873935528e-06,
      "loss": 0.0029,
      "step": 159240
    },
    {
      "epoch": 1.0213800335435954,
      "grad_norm": 0.09060076624155045,
      "learning_rate": 5.683933435764445e-06,
      "loss": 0.0018,
      "step": 159250
    },
    {
      "epoch": 1.0214441704373816,
      "grad_norm": 0.06484679132699966,
      "learning_rate": 5.6833789890231846e-06,
      "loss": 0.0034,
      "step": 159260
    },
    {
      "epoch": 1.0215083073311677,
      "grad_norm": 0.024051832035183907,
      "learning_rate": 5.682824533718699e-06,
      "loss": 0.0029,
      "step": 159270
    },
    {
      "epoch": 1.0215724442249536,
      "grad_norm": 0.07731301337480545,
      "learning_rate": 5.682270069857934e-06,
      "loss": 0.0021,
      "step": 159280
    },
    {
      "epoch": 1.0216365811187398,
      "grad_norm": 0.08671291172504425,
      "learning_rate": 5.681715597447838e-06,
      "loss": 0.0012,
      "step": 159290
    },
    {
      "epoch": 1.021700718012526,
      "grad_norm": 0.14244763553142548,
      "learning_rate": 5.681161116495356e-06,
      "loss": 0.0014,
      "step": 159300
    },
    {
      "epoch": 1.021764854906312,
      "grad_norm": 0.030874190852046013,
      "learning_rate": 5.6806066270074385e-06,
      "loss": 0.003,
      "step": 159310
    },
    {
      "epoch": 1.0218289918000982,
      "grad_norm": 0.041841261088848114,
      "learning_rate": 5.680052128991036e-06,
      "loss": 0.004,
      "step": 159320
    },
    {
      "epoch": 1.0218931286938842,
      "grad_norm": 0.1903955489397049,
      "learning_rate": 5.679497622453093e-06,
      "loss": 0.0014,
      "step": 159330
    },
    {
      "epoch": 1.0219572655876703,
      "grad_norm": 0.10998605191707611,
      "learning_rate": 5.67894310740056e-06,
      "loss": 0.0013,
      "step": 159340
    },
    {
      "epoch": 1.0220214024814565,
      "grad_norm": 0.032601822167634964,
      "learning_rate": 5.678388583840383e-06,
      "loss": 0.0009,
      "step": 159350
    },
    {
      "epoch": 1.0220855393752426,
      "grad_norm": 0.13070465624332428,
      "learning_rate": 5.677834051779513e-06,
      "loss": 0.003,
      "step": 159360
    },
    {
      "epoch": 1.0221496762690285,
      "grad_norm": 0.07420849800109863,
      "learning_rate": 5.677279511224898e-06,
      "loss": 0.0021,
      "step": 159370
    },
    {
      "epoch": 1.0222138131628147,
      "grad_norm": 0.08946508914232254,
      "learning_rate": 5.676724962183487e-06,
      "loss": 0.0018,
      "step": 159380
    },
    {
      "epoch": 1.0222779500566008,
      "grad_norm": 0.012758632190525532,
      "learning_rate": 5.676170404662227e-06,
      "loss": 0.0016,
      "step": 159390
    },
    {
      "epoch": 1.022342086950387,
      "grad_norm": 0.09677497297525406,
      "learning_rate": 5.6756158386680705e-06,
      "loss": 0.0014,
      "step": 159400
    },
    {
      "epoch": 1.022406223844173,
      "grad_norm": 0.04973935708403587,
      "learning_rate": 5.675061264207964e-06,
      "loss": 0.0018,
      "step": 159410
    },
    {
      "epoch": 1.022470360737959,
      "grad_norm": 0.21082596480846405,
      "learning_rate": 5.674506681288857e-06,
      "loss": 0.0017,
      "step": 159420
    },
    {
      "epoch": 1.0225344976317452,
      "grad_norm": 0.16517893970012665,
      "learning_rate": 5.6739520899177e-06,
      "loss": 0.0021,
      "step": 159430
    },
    {
      "epoch": 1.0225986345255313,
      "grad_norm": 0.04861932992935181,
      "learning_rate": 5.673397490101441e-06,
      "loss": 0.0026,
      "step": 159440
    },
    {
      "epoch": 1.0226627714193175,
      "grad_norm": 0.09634580463171005,
      "learning_rate": 5.6728428818470306e-06,
      "loss": 0.0031,
      "step": 159450
    },
    {
      "epoch": 1.0227269083131034,
      "grad_norm": 0.0037230979651212692,
      "learning_rate": 5.6722882651614165e-06,
      "loss": 0.003,
      "step": 159460
    },
    {
      "epoch": 1.0227910452068896,
      "grad_norm": 0.053129348903894424,
      "learning_rate": 5.6717336400515524e-06,
      "loss": 0.0012,
      "step": 159470
    },
    {
      "epoch": 1.0228551821006757,
      "grad_norm": 0.005225110333412886,
      "learning_rate": 5.671179006524383e-06,
      "loss": 0.0011,
      "step": 159480
    },
    {
      "epoch": 1.0229193189944619,
      "grad_norm": 0.11354188621044159,
      "learning_rate": 5.6706243645868645e-06,
      "loss": 0.0022,
      "step": 159490
    },
    {
      "epoch": 1.0229834558882478,
      "grad_norm": 0.0946168527007103,
      "learning_rate": 5.6700697142459404e-06,
      "loss": 0.0018,
      "step": 159500
    },
    {
      "epoch": 1.023047592782034,
      "grad_norm": 0.09126311540603638,
      "learning_rate": 5.669515055508565e-06,
      "loss": 0.0018,
      "step": 159510
    },
    {
      "epoch": 1.02311172967582,
      "grad_norm": 0.04860438406467438,
      "learning_rate": 5.668960388381687e-06,
      "loss": 0.0021,
      "step": 159520
    },
    {
      "epoch": 1.0231758665696062,
      "grad_norm": 0.0699162483215332,
      "learning_rate": 5.668405712872257e-06,
      "loss": 0.0026,
      "step": 159530
    },
    {
      "epoch": 1.0232400034633922,
      "grad_norm": 0.028246797621250153,
      "learning_rate": 5.667851028987226e-06,
      "loss": 0.0014,
      "step": 159540
    },
    {
      "epoch": 1.0233041403571783,
      "grad_norm": 0.06132975593209267,
      "learning_rate": 5.667296336733545e-06,
      "loss": 0.0037,
      "step": 159550
    },
    {
      "epoch": 1.0233682772509645,
      "grad_norm": 0.04671689122915268,
      "learning_rate": 5.666741636118164e-06,
      "loss": 0.0023,
      "step": 159560
    },
    {
      "epoch": 1.0234324141447506,
      "grad_norm": 0.06894919276237488,
      "learning_rate": 5.666186927148033e-06,
      "loss": 0.0018,
      "step": 159570
    },
    {
      "epoch": 1.0234965510385368,
      "grad_norm": 0.12566019594669342,
      "learning_rate": 5.665632209830104e-06,
      "loss": 0.003,
      "step": 159580
    },
    {
      "epoch": 1.0235606879323227,
      "grad_norm": 0.20520325005054474,
      "learning_rate": 5.665077484171329e-06,
      "loss": 0.0027,
      "step": 159590
    },
    {
      "epoch": 1.0236248248261088,
      "grad_norm": 0.1556227058172226,
      "learning_rate": 5.664522750178656e-06,
      "loss": 0.0019,
      "step": 159600
    },
    {
      "epoch": 1.023688961719895,
      "grad_norm": 0.2768998146057129,
      "learning_rate": 5.663968007859039e-06,
      "loss": 0.0015,
      "step": 159610
    },
    {
      "epoch": 1.0237530986136811,
      "grad_norm": 0.09073995053768158,
      "learning_rate": 5.663413257219426e-06,
      "loss": 0.0016,
      "step": 159620
    },
    {
      "epoch": 1.023817235507467,
      "grad_norm": 0.10886190086603165,
      "learning_rate": 5.662858498266774e-06,
      "loss": 0.0016,
      "step": 159630
    },
    {
      "epoch": 1.0238813724012532,
      "grad_norm": 0.04841167852282524,
      "learning_rate": 5.66230373100803e-06,
      "loss": 0.0019,
      "step": 159640
    },
    {
      "epoch": 1.0239455092950394,
      "grad_norm": 0.034094445407390594,
      "learning_rate": 5.661748955450147e-06,
      "loss": 0.0009,
      "step": 159650
    },
    {
      "epoch": 1.0240096461888255,
      "grad_norm": 0.05819038674235344,
      "learning_rate": 5.6611941716000765e-06,
      "loss": 0.0018,
      "step": 159660
    },
    {
      "epoch": 1.0240737830826114,
      "grad_norm": 0.19978097081184387,
      "learning_rate": 5.66063937946477e-06,
      "loss": 0.002,
      "step": 159670
    },
    {
      "epoch": 1.0241379199763976,
      "grad_norm": 0.12936922907829285,
      "learning_rate": 5.660084579051181e-06,
      "loss": 0.0017,
      "step": 159680
    },
    {
      "epoch": 1.0242020568701837,
      "grad_norm": 0.026892701163887978,
      "learning_rate": 5.659529770366259e-06,
      "loss": 0.0015,
      "step": 159690
    },
    {
      "epoch": 1.0242661937639699,
      "grad_norm": 0.04490223154425621,
      "learning_rate": 5.658974953416959e-06,
      "loss": 0.0019,
      "step": 159700
    },
    {
      "epoch": 1.0243303306577558,
      "grad_norm": 0.06686296314001083,
      "learning_rate": 5.658420128210231e-06,
      "loss": 0.0028,
      "step": 159710
    },
    {
      "epoch": 1.024394467551542,
      "grad_norm": 0.004353153053671122,
      "learning_rate": 5.6578652947530286e-06,
      "loss": 0.0012,
      "step": 159720
    },
    {
      "epoch": 1.024458604445328,
      "grad_norm": 0.06313478946685791,
      "learning_rate": 5.657310453052304e-06,
      "loss": 0.006,
      "step": 159730
    },
    {
      "epoch": 1.0245227413391143,
      "grad_norm": 0.08250018954277039,
      "learning_rate": 5.65675560311501e-06,
      "loss": 0.0015,
      "step": 159740
    },
    {
      "epoch": 1.0245868782329004,
      "grad_norm": 0.12308786809444427,
      "learning_rate": 5.656200744948098e-06,
      "loss": 0.0015,
      "step": 159750
    },
    {
      "epoch": 1.0246510151266863,
      "grad_norm": 0.9503151774406433,
      "learning_rate": 5.655645878558522e-06,
      "loss": 0.002,
      "step": 159760
    },
    {
      "epoch": 1.0247151520204725,
      "grad_norm": 0.03679700568318367,
      "learning_rate": 5.655091003953235e-06,
      "loss": 0.0016,
      "step": 159770
    },
    {
      "epoch": 1.0247792889142586,
      "grad_norm": 0.1574845314025879,
      "learning_rate": 5.65453612113919e-06,
      "loss": 0.0048,
      "step": 159780
    },
    {
      "epoch": 1.0248434258080448,
      "grad_norm": 0.09560144692659378,
      "learning_rate": 5.653981230123338e-06,
      "loss": 0.0017,
      "step": 159790
    },
    {
      "epoch": 1.0249075627018307,
      "grad_norm": 0.02885236032307148,
      "learning_rate": 5.653426330912636e-06,
      "loss": 0.0027,
      "step": 159800
    },
    {
      "epoch": 1.0249716995956168,
      "grad_norm": 0.168492391705513,
      "learning_rate": 5.652871423514033e-06,
      "loss": 0.003,
      "step": 159810
    },
    {
      "epoch": 1.025035836489403,
      "grad_norm": 0.033362600952386856,
      "learning_rate": 5.652316507934485e-06,
      "loss": 0.0011,
      "step": 159820
    },
    {
      "epoch": 1.0250999733831891,
      "grad_norm": 0.08808814734220505,
      "learning_rate": 5.651761584180945e-06,
      "loss": 0.0019,
      "step": 159830
    },
    {
      "epoch": 1.025164110276975,
      "grad_norm": 0.045800428837537766,
      "learning_rate": 5.651206652260366e-06,
      "loss": 0.0016,
      "step": 159840
    },
    {
      "epoch": 1.0252282471707612,
      "grad_norm": 0.1911831498146057,
      "learning_rate": 5.650651712179703e-06,
      "loss": 0.0023,
      "step": 159850
    },
    {
      "epoch": 1.0252923840645474,
      "grad_norm": 0.14668892323970795,
      "learning_rate": 5.650096763945909e-06,
      "loss": 0.0028,
      "step": 159860
    },
    {
      "epoch": 1.0253565209583335,
      "grad_norm": 0.13308972120285034,
      "learning_rate": 5.6495418075659396e-06,
      "loss": 0.0021,
      "step": 159870
    },
    {
      "epoch": 1.0254206578521197,
      "grad_norm": 0.0018908643396571279,
      "learning_rate": 5.648986843046745e-06,
      "loss": 0.0029,
      "step": 159880
    },
    {
      "epoch": 1.0254847947459056,
      "grad_norm": 0.03336883708834648,
      "learning_rate": 5.648431870395281e-06,
      "loss": 0.0027,
      "step": 159890
    },
    {
      "epoch": 1.0255489316396917,
      "grad_norm": 0.006976774428039789,
      "learning_rate": 5.6478768896185035e-06,
      "loss": 0.0012,
      "step": 159900
    },
    {
      "epoch": 1.025613068533478,
      "grad_norm": 0.005612578243017197,
      "learning_rate": 5.647321900723364e-06,
      "loss": 0.0012,
      "step": 159910
    },
    {
      "epoch": 1.025677205427264,
      "grad_norm": 0.15938295423984528,
      "learning_rate": 5.646766903716819e-06,
      "loss": 0.0027,
      "step": 159920
    },
    {
      "epoch": 1.02574134232105,
      "grad_norm": 0.10505570471286774,
      "learning_rate": 5.646211898605823e-06,
      "loss": 0.002,
      "step": 159930
    },
    {
      "epoch": 1.0258054792148361,
      "grad_norm": 0.011524339206516743,
      "learning_rate": 5.645656885397331e-06,
      "loss": 0.0014,
      "step": 159940
    },
    {
      "epoch": 1.0258696161086223,
      "grad_norm": 0.23351573944091797,
      "learning_rate": 5.645101864098296e-06,
      "loss": 0.0024,
      "step": 159950
    },
    {
      "epoch": 1.0259337530024084,
      "grad_norm": 0.21484927833080292,
      "learning_rate": 5.644546834715673e-06,
      "loss": 0.0034,
      "step": 159960
    },
    {
      "epoch": 1.0259978898961943,
      "grad_norm": 0.031179411336779594,
      "learning_rate": 5.643991797256418e-06,
      "loss": 0.0012,
      "step": 159970
    },
    {
      "epoch": 1.0260620267899805,
      "grad_norm": 0.1138547882437706,
      "learning_rate": 5.643436751727485e-06,
      "loss": 0.0024,
      "step": 159980
    },
    {
      "epoch": 1.0261261636837666,
      "grad_norm": 0.10956770181655884,
      "learning_rate": 5.642881698135831e-06,
      "loss": 0.0011,
      "step": 159990
    },
    {
      "epoch": 1.0261903005775528,
      "grad_norm": 0.06331674754619598,
      "learning_rate": 5.642326636488409e-06,
      "loss": 0.0024,
      "step": 160000
    },
    {
      "epoch": 1.026254437471339,
      "grad_norm": 0.13225817680358887,
      "learning_rate": 5.641771566792176e-06,
      "loss": 0.0014,
      "step": 160010
    },
    {
      "epoch": 1.0263185743651249,
      "grad_norm": 0.18962177634239197,
      "learning_rate": 5.641216489054084e-06,
      "loss": 0.0023,
      "step": 160020
    },
    {
      "epoch": 1.026382711258911,
      "grad_norm": 0.20619508624076843,
      "learning_rate": 5.640661403281094e-06,
      "loss": 0.002,
      "step": 160030
    },
    {
      "epoch": 1.0264468481526972,
      "grad_norm": 0.17232321202754974,
      "learning_rate": 5.640106309480158e-06,
      "loss": 0.0021,
      "step": 160040
    },
    {
      "epoch": 1.0265109850464833,
      "grad_norm": 0.07825799286365509,
      "learning_rate": 5.639551207658232e-06,
      "loss": 0.0028,
      "step": 160050
    },
    {
      "epoch": 1.0265751219402692,
      "grad_norm": 0.09236768633127213,
      "learning_rate": 5.638996097822271e-06,
      "loss": 0.0021,
      "step": 160060
    },
    {
      "epoch": 1.0266392588340554,
      "grad_norm": 0.03546392545104027,
      "learning_rate": 5.638440979979234e-06,
      "loss": 0.001,
      "step": 160070
    },
    {
      "epoch": 1.0267033957278415,
      "grad_norm": 0.22646042704582214,
      "learning_rate": 5.637885854136076e-06,
      "loss": 0.0014,
      "step": 160080
    },
    {
      "epoch": 1.0267675326216277,
      "grad_norm": 0.0024000273551791906,
      "learning_rate": 5.6373307202997496e-06,
      "loss": 0.0012,
      "step": 160090
    },
    {
      "epoch": 1.0268316695154136,
      "grad_norm": 0.025319969281554222,
      "learning_rate": 5.636775578477216e-06,
      "loss": 0.0014,
      "step": 160100
    },
    {
      "epoch": 1.0268958064091998,
      "grad_norm": 0.09891081601381302,
      "learning_rate": 5.636220428675429e-06,
      "loss": 0.0013,
      "step": 160110
    },
    {
      "epoch": 1.026959943302986,
      "grad_norm": 0.09997166693210602,
      "learning_rate": 5.635665270901345e-06,
      "loss": 0.0015,
      "step": 160120
    },
    {
      "epoch": 1.027024080196772,
      "grad_norm": 0.047390688210725784,
      "learning_rate": 5.635110105161921e-06,
      "loss": 0.0019,
      "step": 160130
    },
    {
      "epoch": 1.027088217090558,
      "grad_norm": 0.051536865532398224,
      "learning_rate": 5.634554931464113e-06,
      "loss": 0.0029,
      "step": 160140
    },
    {
      "epoch": 1.0271523539843441,
      "grad_norm": 0.13702762126922607,
      "learning_rate": 5.633999749814879e-06,
      "loss": 0.0014,
      "step": 160150
    },
    {
      "epoch": 1.0272164908781303,
      "grad_norm": 0.04055806249380112,
      "learning_rate": 5.633444560221174e-06,
      "loss": 0.0011,
      "step": 160160
    },
    {
      "epoch": 1.0272806277719164,
      "grad_norm": 0.06640499830245972,
      "learning_rate": 5.6328893626899575e-06,
      "loss": 0.0013,
      "step": 160170
    },
    {
      "epoch": 1.0273447646657026,
      "grad_norm": 0.11427941173315048,
      "learning_rate": 5.6323341572281835e-06,
      "loss": 0.0013,
      "step": 160180
    },
    {
      "epoch": 1.0274089015594885,
      "grad_norm": 0.14953333139419556,
      "learning_rate": 5.631778943842812e-06,
      "loss": 0.0011,
      "step": 160190
    },
    {
      "epoch": 1.0274730384532746,
      "grad_norm": 0.02691211923956871,
      "learning_rate": 5.6312237225407965e-06,
      "loss": 0.0006,
      "step": 160200
    },
    {
      "epoch": 1.0275371753470608,
      "grad_norm": 0.08211901783943176,
      "learning_rate": 5.630668493329099e-06,
      "loss": 0.0019,
      "step": 160210
    },
    {
      "epoch": 1.027601312240847,
      "grad_norm": 0.08376616984605789,
      "learning_rate": 5.630113256214673e-06,
      "loss": 0.0029,
      "step": 160220
    },
    {
      "epoch": 1.0276654491346329,
      "grad_norm": 0.07124131172895432,
      "learning_rate": 5.629558011204479e-06,
      "loss": 0.0015,
      "step": 160230
    },
    {
      "epoch": 1.027729586028419,
      "grad_norm": 0.014714283868670464,
      "learning_rate": 5.6290027583054704e-06,
      "loss": 0.0021,
      "step": 160240
    },
    {
      "epoch": 1.0277937229222052,
      "grad_norm": 0.07588217407464981,
      "learning_rate": 5.628447497524611e-06,
      "loss": 0.0016,
      "step": 160250
    },
    {
      "epoch": 1.0278578598159913,
      "grad_norm": 0.18099211156368256,
      "learning_rate": 5.6278922288688535e-06,
      "loss": 0.0017,
      "step": 160260
    },
    {
      "epoch": 1.0279219967097772,
      "grad_norm": 0.10302896797657013,
      "learning_rate": 5.627336952345158e-06,
      "loss": 0.0055,
      "step": 160270
    },
    {
      "epoch": 1.0279861336035634,
      "grad_norm": 0.06174756959080696,
      "learning_rate": 5.626781667960483e-06,
      "loss": 0.0016,
      "step": 160280
    },
    {
      "epoch": 1.0280502704973495,
      "grad_norm": 0.121791310608387,
      "learning_rate": 5.626226375721784e-06,
      "loss": 0.0024,
      "step": 160290
    },
    {
      "epoch": 1.0281144073911357,
      "grad_norm": 0.19790798425674438,
      "learning_rate": 5.625671075636021e-06,
      "loss": 0.0017,
      "step": 160300
    },
    {
      "epoch": 1.0281785442849218,
      "grad_norm": 0.12741999328136444,
      "learning_rate": 5.625115767710154e-06,
      "loss": 0.0023,
      "step": 160310
    },
    {
      "epoch": 1.0282426811787078,
      "grad_norm": 0.11417978256940842,
      "learning_rate": 5.624560451951138e-06,
      "loss": 0.0072,
      "step": 160320
    },
    {
      "epoch": 1.028306818072494,
      "grad_norm": 0.15278507769107819,
      "learning_rate": 5.624005128365935e-06,
      "loss": 0.0015,
      "step": 160330
    },
    {
      "epoch": 1.02837095496628,
      "grad_norm": 0.06196146830916405,
      "learning_rate": 5.6234497969615e-06,
      "loss": 0.003,
      "step": 160340
    },
    {
      "epoch": 1.0284350918600662,
      "grad_norm": 0.07730269432067871,
      "learning_rate": 5.622894457744794e-06,
      "loss": 0.0009,
      "step": 160350
    },
    {
      "epoch": 1.0284992287538521,
      "grad_norm": 0.16394564509391785,
      "learning_rate": 5.622339110722775e-06,
      "loss": 0.002,
      "step": 160360
    },
    {
      "epoch": 1.0285633656476383,
      "grad_norm": 0.04401955008506775,
      "learning_rate": 5.621783755902402e-06,
      "loss": 0.0009,
      "step": 160370
    },
    {
      "epoch": 1.0286275025414244,
      "grad_norm": 0.057977233082056046,
      "learning_rate": 5.621228393290636e-06,
      "loss": 0.0018,
      "step": 160380
    },
    {
      "epoch": 1.0286916394352106,
      "grad_norm": 0.05147771164774895,
      "learning_rate": 5.620673022894433e-06,
      "loss": 0.0022,
      "step": 160390
    },
    {
      "epoch": 1.0287557763289965,
      "grad_norm": 0.02774648182094097,
      "learning_rate": 5.620117644720754e-06,
      "loss": 0.0014,
      "step": 160400
    },
    {
      "epoch": 1.0288199132227827,
      "grad_norm": 0.1690646857023239,
      "learning_rate": 5.619562258776556e-06,
      "loss": 0.002,
      "step": 160410
    },
    {
      "epoch": 1.0288840501165688,
      "grad_norm": 0.09440110623836517,
      "learning_rate": 5.6190068650688025e-06,
      "loss": 0.0019,
      "step": 160420
    },
    {
      "epoch": 1.028948187010355,
      "grad_norm": 0.08699806034564972,
      "learning_rate": 5.618451463604448e-06,
      "loss": 0.0015,
      "step": 160430
    },
    {
      "epoch": 1.029012323904141,
      "grad_norm": 0.07111769914627075,
      "learning_rate": 5.617896054390457e-06,
      "loss": 0.0019,
      "step": 160440
    },
    {
      "epoch": 1.029076460797927,
      "grad_norm": 0.17294718325138092,
      "learning_rate": 5.617340637433785e-06,
      "loss": 0.0021,
      "step": 160450
    },
    {
      "epoch": 1.0291405976917132,
      "grad_norm": 0.04871895909309387,
      "learning_rate": 5.6167852127413955e-06,
      "loss": 0.0015,
      "step": 160460
    },
    {
      "epoch": 1.0292047345854993,
      "grad_norm": 0.05924814194440842,
      "learning_rate": 5.616229780320246e-06,
      "loss": 0.0016,
      "step": 160470
    },
    {
      "epoch": 1.0292688714792855,
      "grad_norm": 0.10790330916643143,
      "learning_rate": 5.615674340177296e-06,
      "loss": 0.0008,
      "step": 160480
    },
    {
      "epoch": 1.0293330083730714,
      "grad_norm": 0.20627789199352264,
      "learning_rate": 5.615118892319507e-06,
      "loss": 0.0032,
      "step": 160490
    },
    {
      "epoch": 1.0293971452668575,
      "grad_norm": 0.08072004467248917,
      "learning_rate": 5.614563436753839e-06,
      "loss": 0.0014,
      "step": 160500
    },
    {
      "epoch": 1.0294612821606437,
      "grad_norm": 0.10128284990787506,
      "learning_rate": 5.6140079734872514e-06,
      "loss": 0.0041,
      "step": 160510
    },
    {
      "epoch": 1.0295254190544298,
      "grad_norm": 0.07587765157222748,
      "learning_rate": 5.613452502526704e-06,
      "loss": 0.001,
      "step": 160520
    },
    {
      "epoch": 1.0295895559482158,
      "grad_norm": 0.06863145530223846,
      "learning_rate": 5.61289702387916e-06,
      "loss": 0.0013,
      "step": 160530
    },
    {
      "epoch": 1.029653692842002,
      "grad_norm": 0.17337927222251892,
      "learning_rate": 5.6123415375515775e-06,
      "loss": 0.0038,
      "step": 160540
    },
    {
      "epoch": 1.029717829735788,
      "grad_norm": 0.01374752726405859,
      "learning_rate": 5.611786043550918e-06,
      "loss": 0.0012,
      "step": 160550
    },
    {
      "epoch": 1.0297819666295742,
      "grad_norm": 0.06106782332062721,
      "learning_rate": 5.611230541884143e-06,
      "loss": 0.0017,
      "step": 160560
    },
    {
      "epoch": 1.0298461035233604,
      "grad_norm": 0.06003543734550476,
      "learning_rate": 5.610675032558211e-06,
      "loss": 0.0013,
      "step": 160570
    },
    {
      "epoch": 1.0299102404171463,
      "grad_norm": 0.07925518602132797,
      "learning_rate": 5.610119515580086e-06,
      "loss": 0.0023,
      "step": 160580
    },
    {
      "epoch": 1.0299743773109324,
      "grad_norm": 0.1774548888206482,
      "learning_rate": 5.609563990956727e-06,
      "loss": 0.0013,
      "step": 160590
    },
    {
      "epoch": 1.0300385142047186,
      "grad_norm": 0.19831368327140808,
      "learning_rate": 5.609008458695095e-06,
      "loss": 0.003,
      "step": 160600
    },
    {
      "epoch": 1.0301026510985047,
      "grad_norm": 0.029156174510717392,
      "learning_rate": 5.608452918802152e-06,
      "loss": 0.0017,
      "step": 160610
    },
    {
      "epoch": 1.0301667879922907,
      "grad_norm": 0.06537648290395737,
      "learning_rate": 5.607897371284858e-06,
      "loss": 0.0014,
      "step": 160620
    },
    {
      "epoch": 1.0302309248860768,
      "grad_norm": 0.1173158586025238,
      "learning_rate": 5.607341816150177e-06,
      "loss": 0.0024,
      "step": 160630
    },
    {
      "epoch": 1.030295061779863,
      "grad_norm": 0.0031485699582844973,
      "learning_rate": 5.6067862534050675e-06,
      "loss": 0.0022,
      "step": 160640
    },
    {
      "epoch": 1.0303591986736491,
      "grad_norm": 0.07841501384973526,
      "learning_rate": 5.606230683056495e-06,
      "loss": 0.0012,
      "step": 160650
    },
    {
      "epoch": 1.030423335567435,
      "grad_norm": 0.09447195380926132,
      "learning_rate": 5.605675105111416e-06,
      "loss": 0.002,
      "step": 160660
    },
    {
      "epoch": 1.0304874724612212,
      "grad_norm": 0.3483569324016571,
      "learning_rate": 5.605119519576796e-06,
      "loss": 0.0019,
      "step": 160670
    },
    {
      "epoch": 1.0305516093550073,
      "grad_norm": 0.0792723298072815,
      "learning_rate": 5.604563926459596e-06,
      "loss": 0.0029,
      "step": 160680
    },
    {
      "epoch": 1.0306157462487935,
      "grad_norm": 0.0928591936826706,
      "learning_rate": 5.604008325766779e-06,
      "loss": 0.0023,
      "step": 160690
    },
    {
      "epoch": 1.0306798831425796,
      "grad_norm": 0.05987996980547905,
      "learning_rate": 5.603452717505304e-06,
      "loss": 0.0016,
      "step": 160700
    },
    {
      "epoch": 1.0307440200363656,
      "grad_norm": 0.1154266893863678,
      "learning_rate": 5.602897101682137e-06,
      "loss": 0.0014,
      "step": 160710
    },
    {
      "epoch": 1.0308081569301517,
      "grad_norm": 0.09690660983324051,
      "learning_rate": 5.602341478304238e-06,
      "loss": 0.0022,
      "step": 160720
    },
    {
      "epoch": 1.0308722938239379,
      "grad_norm": 0.12438417971134186,
      "learning_rate": 5.601785847378569e-06,
      "loss": 0.0021,
      "step": 160730
    },
    {
      "epoch": 1.030936430717724,
      "grad_norm": 0.03502361848950386,
      "learning_rate": 5.601230208912094e-06,
      "loss": 0.0024,
      "step": 160740
    },
    {
      "epoch": 1.03100056761151,
      "grad_norm": 0.05004703998565674,
      "learning_rate": 5.6006745629117745e-06,
      "loss": 0.0015,
      "step": 160750
    },
    {
      "epoch": 1.031064704505296,
      "grad_norm": 0.05685212463140488,
      "learning_rate": 5.600118909384574e-06,
      "loss": 0.0022,
      "step": 160760
    },
    {
      "epoch": 1.0311288413990822,
      "grad_norm": 0.016093585640192032,
      "learning_rate": 5.599563248337454e-06,
      "loss": 0.0009,
      "step": 160770
    },
    {
      "epoch": 1.0311929782928684,
      "grad_norm": 0.2270883470773697,
      "learning_rate": 5.599007579777378e-06,
      "loss": 0.0035,
      "step": 160780
    },
    {
      "epoch": 1.0312571151866543,
      "grad_norm": 0.1231345608830452,
      "learning_rate": 5.598451903711309e-06,
      "loss": 0.0033,
      "step": 160790
    },
    {
      "epoch": 1.0313212520804405,
      "grad_norm": 0.11265743523836136,
      "learning_rate": 5.59789622014621e-06,
      "loss": 0.0029,
      "step": 160800
    },
    {
      "epoch": 1.0313853889742266,
      "grad_norm": 0.1232166662812233,
      "learning_rate": 5.597340529089045e-06,
      "loss": 0.0025,
      "step": 160810
    },
    {
      "epoch": 1.0314495258680128,
      "grad_norm": 0.09138888865709305,
      "learning_rate": 5.596784830546776e-06,
      "loss": 0.0029,
      "step": 160820
    },
    {
      "epoch": 1.0315136627617987,
      "grad_norm": 0.0028286930173635483,
      "learning_rate": 5.596229124526364e-06,
      "loss": 0.0018,
      "step": 160830
    },
    {
      "epoch": 1.0315777996555848,
      "grad_norm": 0.14618414640426636,
      "learning_rate": 5.595673411034779e-06,
      "loss": 0.0014,
      "step": 160840
    },
    {
      "epoch": 1.031641936549371,
      "grad_norm": 0.06368950754404068,
      "learning_rate": 5.595117690078977e-06,
      "loss": 0.0022,
      "step": 160850
    },
    {
      "epoch": 1.0317060734431571,
      "grad_norm": 0.014613325707614422,
      "learning_rate": 5.594561961665927e-06,
      "loss": 0.0018,
      "step": 160860
    },
    {
      "epoch": 1.0317702103369433,
      "grad_norm": 0.047521304339170456,
      "learning_rate": 5.594006225802589e-06,
      "loss": 0.0019,
      "step": 160870
    },
    {
      "epoch": 1.0318343472307292,
      "grad_norm": 0.0785314068198204,
      "learning_rate": 5.59345048249593e-06,
      "loss": 0.0018,
      "step": 160880
    },
    {
      "epoch": 1.0318984841245153,
      "grad_norm": 0.0063042123802006245,
      "learning_rate": 5.592894731752912e-06,
      "loss": 0.0022,
      "step": 160890
    },
    {
      "epoch": 1.0319626210183015,
      "grad_norm": 0.029178213328123093,
      "learning_rate": 5.592338973580499e-06,
      "loss": 0.0015,
      "step": 160900
    },
    {
      "epoch": 1.0320267579120876,
      "grad_norm": 0.13169245421886444,
      "learning_rate": 5.591783207985656e-06,
      "loss": 0.0021,
      "step": 160910
    },
    {
      "epoch": 1.0320908948058736,
      "grad_norm": 0.1766376793384552,
      "learning_rate": 5.591227434975345e-06,
      "loss": 0.0015,
      "step": 160920
    },
    {
      "epoch": 1.0321550316996597,
      "grad_norm": 0.03069526143372059,
      "learning_rate": 5.590671654556533e-06,
      "loss": 0.0014,
      "step": 160930
    },
    {
      "epoch": 1.0322191685934459,
      "grad_norm": 0.16416823863983154,
      "learning_rate": 5.590115866736182e-06,
      "loss": 0.0019,
      "step": 160940
    },
    {
      "epoch": 1.032283305487232,
      "grad_norm": 0.07134131342172623,
      "learning_rate": 5.589560071521259e-06,
      "loss": 0.0015,
      "step": 160950
    },
    {
      "epoch": 1.032347442381018,
      "grad_norm": 0.06377499550580978,
      "learning_rate": 5.589004268918725e-06,
      "loss": 0.0026,
      "step": 160960
    },
    {
      "epoch": 1.032411579274804,
      "grad_norm": 0.012781238183379173,
      "learning_rate": 5.588448458935548e-06,
      "loss": 0.0017,
      "step": 160970
    },
    {
      "epoch": 1.0324757161685902,
      "grad_norm": 0.021176699548959732,
      "learning_rate": 5.5878926415786906e-06,
      "loss": 0.0018,
      "step": 160980
    },
    {
      "epoch": 1.0325398530623764,
      "grad_norm": 0.023555800318717957,
      "learning_rate": 5.587336816855119e-06,
      "loss": 0.0018,
      "step": 160990
    },
    {
      "epoch": 1.0326039899561625,
      "grad_norm": 0.06162658706307411,
      "learning_rate": 5.586780984771797e-06,
      "loss": 0.0014,
      "step": 161000
    },
    {
      "epoch": 1.0326681268499485,
      "grad_norm": 0.13439489901065826,
      "learning_rate": 5.5862251453356885e-06,
      "loss": 0.0025,
      "step": 161010
    },
    {
      "epoch": 1.0327322637437346,
      "grad_norm": 0.12414365261793137,
      "learning_rate": 5.585669298553762e-06,
      "loss": 0.0012,
      "step": 161020
    },
    {
      "epoch": 1.0327964006375208,
      "grad_norm": 0.04816731438040733,
      "learning_rate": 5.585113444432979e-06,
      "loss": 0.0026,
      "step": 161030
    },
    {
      "epoch": 1.032860537531307,
      "grad_norm": 0.10992000252008438,
      "learning_rate": 5.584557582980308e-06,
      "loss": 0.0006,
      "step": 161040
    },
    {
      "epoch": 1.0329246744250928,
      "grad_norm": 0.45915332436561584,
      "learning_rate": 5.584001714202711e-06,
      "loss": 0.0011,
      "step": 161050
    },
    {
      "epoch": 1.032988811318879,
      "grad_norm": 0.036483284085989,
      "learning_rate": 5.583445838107156e-06,
      "loss": 0.0023,
      "step": 161060
    },
    {
      "epoch": 1.0330529482126651,
      "grad_norm": 0.07692298293113708,
      "learning_rate": 5.5828899547006065e-06,
      "loss": 0.0015,
      "step": 161070
    },
    {
      "epoch": 1.0331170851064513,
      "grad_norm": 0.10438433289527893,
      "learning_rate": 5.582334063990031e-06,
      "loss": 0.0017,
      "step": 161080
    },
    {
      "epoch": 1.0331812220002372,
      "grad_norm": 0.14653724431991577,
      "learning_rate": 5.581778165982392e-06,
      "loss": 0.002,
      "step": 161090
    },
    {
      "epoch": 1.0332453588940234,
      "grad_norm": 0.11509974300861359,
      "learning_rate": 5.581222260684659e-06,
      "loss": 0.0017,
      "step": 161100
    },
    {
      "epoch": 1.0333094957878095,
      "grad_norm": 0.20483095943927765,
      "learning_rate": 5.580666348103794e-06,
      "loss": 0.0025,
      "step": 161110
    },
    {
      "epoch": 1.0333736326815957,
      "grad_norm": 0.02480766735970974,
      "learning_rate": 5.5801104282467656e-06,
      "loss": 0.0023,
      "step": 161120
    },
    {
      "epoch": 1.0334377695753818,
      "grad_norm": 0.12931233644485474,
      "learning_rate": 5.579554501120538e-06,
      "loss": 0.0022,
      "step": 161130
    },
    {
      "epoch": 1.0335019064691677,
      "grad_norm": 0.3534706234931946,
      "learning_rate": 5.578998566732079e-06,
      "loss": 0.0028,
      "step": 161140
    },
    {
      "epoch": 1.0335660433629539,
      "grad_norm": 0.18303950130939484,
      "learning_rate": 5.578442625088353e-06,
      "loss": 0.0018,
      "step": 161150
    },
    {
      "epoch": 1.03363018025674,
      "grad_norm": 0.10803184658288956,
      "learning_rate": 5.5778866761963295e-06,
      "loss": 0.0018,
      "step": 161160
    },
    {
      "epoch": 1.0336943171505262,
      "grad_norm": 0.09353943169116974,
      "learning_rate": 5.577330720062971e-06,
      "loss": 0.002,
      "step": 161170
    },
    {
      "epoch": 1.033758454044312,
      "grad_norm": 0.013269330374896526,
      "learning_rate": 5.576774756695247e-06,
      "loss": 0.0015,
      "step": 161180
    },
    {
      "epoch": 1.0338225909380983,
      "grad_norm": 0.019221922382712364,
      "learning_rate": 5.576218786100123e-06,
      "loss": 0.0008,
      "step": 161190
    },
    {
      "epoch": 1.0338867278318844,
      "grad_norm": 0.2175097018480301,
      "learning_rate": 5.575662808284565e-06,
      "loss": 0.0025,
      "step": 161200
    },
    {
      "epoch": 1.0339508647256705,
      "grad_norm": 0.06522596627473831,
      "learning_rate": 5.5751068232555414e-06,
      "loss": 0.0015,
      "step": 161210
    },
    {
      "epoch": 1.0340150016194565,
      "grad_norm": 0.06038666516542435,
      "learning_rate": 5.574550831020018e-06,
      "loss": 0.0013,
      "step": 161220
    },
    {
      "epoch": 1.0340791385132426,
      "grad_norm": 0.06672700494527817,
      "learning_rate": 5.5739948315849625e-06,
      "loss": 0.0011,
      "step": 161230
    },
    {
      "epoch": 1.0341432754070288,
      "grad_norm": 0.07927929610013962,
      "learning_rate": 5.5734388249573404e-06,
      "loss": 0.0011,
      "step": 161240
    },
    {
      "epoch": 1.034207412300815,
      "grad_norm": 0.6204650402069092,
      "learning_rate": 5.572882811144121e-06,
      "loss": 0.0028,
      "step": 161250
    },
    {
      "epoch": 1.0342715491946008,
      "grad_norm": 0.1258297860622406,
      "learning_rate": 5.57232679015227e-06,
      "loss": 0.0021,
      "step": 161260
    },
    {
      "epoch": 1.034335686088387,
      "grad_norm": 0.1383894979953766,
      "learning_rate": 5.571770761988756e-06,
      "loss": 0.0032,
      "step": 161270
    },
    {
      "epoch": 1.0343998229821731,
      "grad_norm": 0.08100338280200958,
      "learning_rate": 5.5712147266605455e-06,
      "loss": 0.0024,
      "step": 161280
    },
    {
      "epoch": 1.0344639598759593,
      "grad_norm": 0.15081030130386353,
      "learning_rate": 5.570658684174606e-06,
      "loss": 0.002,
      "step": 161290
    },
    {
      "epoch": 1.0345280967697454,
      "grad_norm": 0.18181532621383667,
      "learning_rate": 5.570102634537905e-06,
      "loss": 0.0019,
      "step": 161300
    },
    {
      "epoch": 1.0345922336635314,
      "grad_norm": 0.5333714485168457,
      "learning_rate": 5.569546577757412e-06,
      "loss": 0.002,
      "step": 161310
    },
    {
      "epoch": 1.0346563705573175,
      "grad_norm": 0.1616603285074234,
      "learning_rate": 5.568990513840092e-06,
      "loss": 0.0023,
      "step": 161320
    },
    {
      "epoch": 1.0347205074511037,
      "grad_norm": 0.03816382586956024,
      "learning_rate": 5.568434442792915e-06,
      "loss": 0.0034,
      "step": 161330
    },
    {
      "epoch": 1.0347846443448898,
      "grad_norm": 0.04157206788659096,
      "learning_rate": 5.5678783646228466e-06,
      "loss": 0.0014,
      "step": 161340
    },
    {
      "epoch": 1.0348487812386757,
      "grad_norm": 0.10471261292695999,
      "learning_rate": 5.5673222793368575e-06,
      "loss": 0.0016,
      "step": 161350
    },
    {
      "epoch": 1.034912918132462,
      "grad_norm": 0.124302439391613,
      "learning_rate": 5.566766186941914e-06,
      "loss": 0.0017,
      "step": 161360
    },
    {
      "epoch": 1.034977055026248,
      "grad_norm": 0.09578802436590195,
      "learning_rate": 5.566210087444986e-06,
      "loss": 0.0018,
      "step": 161370
    },
    {
      "epoch": 1.0350411919200342,
      "grad_norm": 0.09537933766841888,
      "learning_rate": 5.565653980853042e-06,
      "loss": 0.0019,
      "step": 161380
    },
    {
      "epoch": 1.0351053288138201,
      "grad_norm": 0.11257769912481308,
      "learning_rate": 5.5650978671730474e-06,
      "loss": 0.0012,
      "step": 161390
    },
    {
      "epoch": 1.0351694657076063,
      "grad_norm": 0.16283220052719116,
      "learning_rate": 5.564541746411974e-06,
      "loss": 0.0017,
      "step": 161400
    },
    {
      "epoch": 1.0352336026013924,
      "grad_norm": 0.07024915516376495,
      "learning_rate": 5.5639856185767885e-06,
      "loss": 0.0026,
      "step": 161410
    },
    {
      "epoch": 1.0352977394951786,
      "grad_norm": 0.06933924555778503,
      "learning_rate": 5.56342948367446e-06,
      "loss": 0.0021,
      "step": 161420
    },
    {
      "epoch": 1.0353618763889647,
      "grad_norm": 0.09991660714149475,
      "learning_rate": 5.562873341711958e-06,
      "loss": 0.0029,
      "step": 161430
    },
    {
      "epoch": 1.0354260132827506,
      "grad_norm": 0.07846569269895554,
      "learning_rate": 5.562317192696248e-06,
      "loss": 0.0019,
      "step": 161440
    },
    {
      "epoch": 1.0354901501765368,
      "grad_norm": 0.06706856936216354,
      "learning_rate": 5.561761036634304e-06,
      "loss": 0.0022,
      "step": 161450
    },
    {
      "epoch": 1.035554287070323,
      "grad_norm": 0.22754397988319397,
      "learning_rate": 5.561204873533093e-06,
      "loss": 0.0021,
      "step": 161460
    },
    {
      "epoch": 1.035618423964109,
      "grad_norm": 0.012594277039170265,
      "learning_rate": 5.5606487033995835e-06,
      "loss": 0.0017,
      "step": 161470
    },
    {
      "epoch": 1.035682560857895,
      "grad_norm": 0.11615916341543198,
      "learning_rate": 5.560092526240746e-06,
      "loss": 0.0012,
      "step": 161480
    },
    {
      "epoch": 1.0357466977516812,
      "grad_norm": 0.01409253478050232,
      "learning_rate": 5.559536342063548e-06,
      "loss": 0.001,
      "step": 161490
    },
    {
      "epoch": 1.0358108346454673,
      "grad_norm": 0.10805962234735489,
      "learning_rate": 5.5589801508749595e-06,
      "loss": 0.0016,
      "step": 161500
    },
    {
      "epoch": 1.0358749715392535,
      "grad_norm": 0.22180473804473877,
      "learning_rate": 5.558423952681949e-06,
      "loss": 0.0019,
      "step": 161510
    },
    {
      "epoch": 1.0359391084330394,
      "grad_norm": 0.2480551153421402,
      "learning_rate": 5.55786774749149e-06,
      "loss": 0.0017,
      "step": 161520
    },
    {
      "epoch": 1.0360032453268255,
      "grad_norm": 0.08151610195636749,
      "learning_rate": 5.557311535310548e-06,
      "loss": 0.0013,
      "step": 161530
    },
    {
      "epoch": 1.0360673822206117,
      "grad_norm": 0.03832826763391495,
      "learning_rate": 5.556755316146094e-06,
      "loss": 0.0014,
      "step": 161540
    },
    {
      "epoch": 1.0361315191143978,
      "grad_norm": 0.13015124201774597,
      "learning_rate": 5.556199090005098e-06,
      "loss": 0.0022,
      "step": 161550
    },
    {
      "epoch": 1.036195656008184,
      "grad_norm": 0.15211138129234314,
      "learning_rate": 5.5556428568945306e-06,
      "loss": 0.0016,
      "step": 161560
    },
    {
      "epoch": 1.03625979290197,
      "grad_norm": 0.057695139199495316,
      "learning_rate": 5.55508661682136e-06,
      "loss": 0.0021,
      "step": 161570
    },
    {
      "epoch": 1.036323929795756,
      "grad_norm": 0.17335347831249237,
      "learning_rate": 5.554530369792558e-06,
      "loss": 0.0022,
      "step": 161580
    },
    {
      "epoch": 1.0363880666895422,
      "grad_norm": 0.03899018466472626,
      "learning_rate": 5.553974115815094e-06,
      "loss": 0.0014,
      "step": 161590
    },
    {
      "epoch": 1.0364522035833283,
      "grad_norm": 0.11029978096485138,
      "learning_rate": 5.553417854895938e-06,
      "loss": 0.0034,
      "step": 161600
    },
    {
      "epoch": 1.0365163404771143,
      "grad_norm": 0.07199625670909882,
      "learning_rate": 5.552861587042062e-06,
      "loss": 0.0032,
      "step": 161610
    },
    {
      "epoch": 1.0365804773709004,
      "grad_norm": 0.12434910982847214,
      "learning_rate": 5.552305312260435e-06,
      "loss": 0.0011,
      "step": 161620
    },
    {
      "epoch": 1.0366446142646866,
      "grad_norm": 0.11052077263593674,
      "learning_rate": 5.551749030558027e-06,
      "loss": 0.0037,
      "step": 161630
    },
    {
      "epoch": 1.0367087511584727,
      "grad_norm": 0.374609112739563,
      "learning_rate": 5.5511927419418084e-06,
      "loss": 0.0017,
      "step": 161640
    },
    {
      "epoch": 1.0367728880522586,
      "grad_norm": 0.0357525460422039,
      "learning_rate": 5.550636446418754e-06,
      "loss": 0.0022,
      "step": 161650
    },
    {
      "epoch": 1.0368370249460448,
      "grad_norm": 0.20752693712711334,
      "learning_rate": 5.550080143995827e-06,
      "loss": 0.0034,
      "step": 161660
    },
    {
      "epoch": 1.036901161839831,
      "grad_norm": 0.05699139088392258,
      "learning_rate": 5.549523834680006e-06,
      "loss": 0.0019,
      "step": 161670
    },
    {
      "epoch": 1.036965298733617,
      "grad_norm": 0.03660396486520767,
      "learning_rate": 5.5489675184782575e-06,
      "loss": 0.0025,
      "step": 161680
    },
    {
      "epoch": 1.037029435627403,
      "grad_norm": 0.049404483288526535,
      "learning_rate": 5.548411195397554e-06,
      "loss": 0.0028,
      "step": 161690
    },
    {
      "epoch": 1.0370935725211892,
      "grad_norm": 0.08149591833353043,
      "learning_rate": 5.547854865444866e-06,
      "loss": 0.0019,
      "step": 161700
    },
    {
      "epoch": 1.0371577094149753,
      "grad_norm": 0.10058657079935074,
      "learning_rate": 5.547298528627165e-06,
      "loss": 0.0015,
      "step": 161710
    },
    {
      "epoch": 1.0372218463087615,
      "grad_norm": 0.07720974087715149,
      "learning_rate": 5.546742184951422e-06,
      "loss": 0.0017,
      "step": 161720
    },
    {
      "epoch": 1.0372859832025476,
      "grad_norm": 0.1567755937576294,
      "learning_rate": 5.546185834424609e-06,
      "loss": 0.0034,
      "step": 161730
    },
    {
      "epoch": 1.0373501200963335,
      "grad_norm": 0.015572108328342438,
      "learning_rate": 5.545629477053697e-06,
      "loss": 0.0017,
      "step": 161740
    },
    {
      "epoch": 1.0374142569901197,
      "grad_norm": 0.031676653772592545,
      "learning_rate": 5.545073112845657e-06,
      "loss": 0.0008,
      "step": 161750
    },
    {
      "epoch": 1.0374783938839058,
      "grad_norm": 0.0611712783575058,
      "learning_rate": 5.544516741807463e-06,
      "loss": 0.0019,
      "step": 161760
    },
    {
      "epoch": 1.037542530777692,
      "grad_norm": 0.049083299934864044,
      "learning_rate": 5.543960363946083e-06,
      "loss": 0.0023,
      "step": 161770
    },
    {
      "epoch": 1.037606667671478,
      "grad_norm": 0.1100301593542099,
      "learning_rate": 5.543403979268494e-06,
      "loss": 0.0026,
      "step": 161780
    },
    {
      "epoch": 1.037670804565264,
      "grad_norm": 0.3066883087158203,
      "learning_rate": 5.542847587781661e-06,
      "loss": 0.0045,
      "step": 161790
    },
    {
      "epoch": 1.0377349414590502,
      "grad_norm": 0.05478224158287048,
      "learning_rate": 5.542291189492564e-06,
      "loss": 0.0022,
      "step": 161800
    },
    {
      "epoch": 1.0377990783528364,
      "grad_norm": 0.11197340488433838,
      "learning_rate": 5.541734784408167e-06,
      "loss": 0.0018,
      "step": 161810
    },
    {
      "epoch": 1.0378632152466223,
      "grad_norm": 0.06742127239704132,
      "learning_rate": 5.541178372535447e-06,
      "loss": 0.0018,
      "step": 161820
    },
    {
      "epoch": 1.0379273521404084,
      "grad_norm": 0.11610394716262817,
      "learning_rate": 5.5406219538813745e-06,
      "loss": 0.003,
      "step": 161830
    },
    {
      "epoch": 1.0379914890341946,
      "grad_norm": 0.17880384624004364,
      "learning_rate": 5.540065528452925e-06,
      "loss": 0.0024,
      "step": 161840
    },
    {
      "epoch": 1.0380556259279807,
      "grad_norm": 0.02726311981678009,
      "learning_rate": 5.539509096257066e-06,
      "loss": 0.0016,
      "step": 161850
    },
    {
      "epoch": 1.0381197628217669,
      "grad_norm": 0.006308823358267546,
      "learning_rate": 5.538952657300775e-06,
      "loss": 0.0017,
      "step": 161860
    },
    {
      "epoch": 1.0381838997155528,
      "grad_norm": 0.02790677733719349,
      "learning_rate": 5.53839621159102e-06,
      "loss": 0.0024,
      "step": 161870
    },
    {
      "epoch": 1.038248036609339,
      "grad_norm": 0.2501605451107025,
      "learning_rate": 5.537839759134776e-06,
      "loss": 0.0022,
      "step": 161880
    },
    {
      "epoch": 1.038312173503125,
      "grad_norm": 0.10462504625320435,
      "learning_rate": 5.537283299939016e-06,
      "loss": 0.0021,
      "step": 161890
    },
    {
      "epoch": 1.0383763103969113,
      "grad_norm": 0.033071622252464294,
      "learning_rate": 5.536726834010712e-06,
      "loss": 0.0014,
      "step": 161900
    },
    {
      "epoch": 1.0384404472906972,
      "grad_norm": 0.04038606584072113,
      "learning_rate": 5.536170361356836e-06,
      "loss": 0.0032,
      "step": 161910
    },
    {
      "epoch": 1.0385045841844833,
      "grad_norm": 0.24799920618534088,
      "learning_rate": 5.535613881984363e-06,
      "loss": 0.0031,
      "step": 161920
    },
    {
      "epoch": 1.0385687210782695,
      "grad_norm": 0.0339103527367115,
      "learning_rate": 5.535057395900266e-06,
      "loss": 0.0013,
      "step": 161930
    },
    {
      "epoch": 1.0386328579720556,
      "grad_norm": 0.10233575850725174,
      "learning_rate": 5.5345009031115175e-06,
      "loss": 0.0015,
      "step": 161940
    },
    {
      "epoch": 1.0386969948658415,
      "grad_norm": 0.1514114886522293,
      "learning_rate": 5.53394440362509e-06,
      "loss": 0.0029,
      "step": 161950
    },
    {
      "epoch": 1.0387611317596277,
      "grad_norm": 0.11045278608798981,
      "learning_rate": 5.5333878974479575e-06,
      "loss": 0.0021,
      "step": 161960
    },
    {
      "epoch": 1.0388252686534138,
      "grad_norm": 0.07066120207309723,
      "learning_rate": 5.532831384587094e-06,
      "loss": 0.0031,
      "step": 161970
    },
    {
      "epoch": 1.0388894055472,
      "grad_norm": 0.26015713810920715,
      "learning_rate": 5.532274865049472e-06,
      "loss": 0.0064,
      "step": 161980
    },
    {
      "epoch": 1.0389535424409861,
      "grad_norm": 0.14190037548542023,
      "learning_rate": 5.531718338842066e-06,
      "loss": 0.0024,
      "step": 161990
    },
    {
      "epoch": 1.039017679334772,
      "grad_norm": 0.086383156478405,
      "learning_rate": 5.531161805971848e-06,
      "loss": 0.0011,
      "step": 162000
    },
    {
      "epoch": 1.0390818162285582,
      "grad_norm": 0.05792558938264847,
      "learning_rate": 5.530605266445795e-06,
      "loss": 0.0039,
      "step": 162010
    },
    {
      "epoch": 1.0391459531223444,
      "grad_norm": 0.059819743037223816,
      "learning_rate": 5.530048720270876e-06,
      "loss": 0.0015,
      "step": 162020
    },
    {
      "epoch": 1.0392100900161305,
      "grad_norm": 0.1345950812101364,
      "learning_rate": 5.52949216745407e-06,
      "loss": 0.001,
      "step": 162030
    },
    {
      "epoch": 1.0392742269099164,
      "grad_norm": 0.08652999997138977,
      "learning_rate": 5.528935608002348e-06,
      "loss": 0.002,
      "step": 162040
    },
    {
      "epoch": 1.0393383638037026,
      "grad_norm": 0.034290753304958344,
      "learning_rate": 5.528379041922686e-06,
      "loss": 0.0012,
      "step": 162050
    },
    {
      "epoch": 1.0394025006974887,
      "grad_norm": 0.04874274507164955,
      "learning_rate": 5.527822469222055e-06,
      "loss": 0.0034,
      "step": 162060
    },
    {
      "epoch": 1.039466637591275,
      "grad_norm": 0.034296587109565735,
      "learning_rate": 5.527265889907431e-06,
      "loss": 0.0017,
      "step": 162070
    },
    {
      "epoch": 1.0395307744850608,
      "grad_norm": 0.06329309940338135,
      "learning_rate": 5.52670930398579e-06,
      "loss": 0.0014,
      "step": 162080
    },
    {
      "epoch": 1.039594911378847,
      "grad_norm": 0.1566283255815506,
      "learning_rate": 5.526152711464104e-06,
      "loss": 0.0018,
      "step": 162090
    },
    {
      "epoch": 1.0396590482726331,
      "grad_norm": 0.20953132212162018,
      "learning_rate": 5.525596112349351e-06,
      "loss": 0.0015,
      "step": 162100
    },
    {
      "epoch": 1.0397231851664193,
      "grad_norm": 0.16059410572052002,
      "learning_rate": 5.5250395066485e-06,
      "loss": 0.0007,
      "step": 162110
    },
    {
      "epoch": 1.0397873220602052,
      "grad_norm": 0.032380376011133194,
      "learning_rate": 5.52448289436853e-06,
      "loss": 0.0017,
      "step": 162120
    },
    {
      "epoch": 1.0398514589539913,
      "grad_norm": 0.11249042302370071,
      "learning_rate": 5.523926275516413e-06,
      "loss": 0.0024,
      "step": 162130
    },
    {
      "epoch": 1.0399155958477775,
      "grad_norm": 0.1949484944343567,
      "learning_rate": 5.523369650099128e-06,
      "loss": 0.0033,
      "step": 162140
    },
    {
      "epoch": 1.0399797327415636,
      "grad_norm": 0.2602667510509491,
      "learning_rate": 5.522813018123646e-06,
      "loss": 0.0026,
      "step": 162150
    },
    {
      "epoch": 1.0400438696353498,
      "grad_norm": 0.1616698056459427,
      "learning_rate": 5.522256379596943e-06,
      "loss": 0.0014,
      "step": 162160
    },
    {
      "epoch": 1.0401080065291357,
      "grad_norm": 0.20122376084327698,
      "learning_rate": 5.521699734525995e-06,
      "loss": 0.0018,
      "step": 162170
    },
    {
      "epoch": 1.0401721434229219,
      "grad_norm": 0.0360245555639267,
      "learning_rate": 5.521143082917776e-06,
      "loss": 0.0009,
      "step": 162180
    },
    {
      "epoch": 1.040236280316708,
      "grad_norm": 0.06166728213429451,
      "learning_rate": 5.520586424779262e-06,
      "loss": 0.0017,
      "step": 162190
    },
    {
      "epoch": 1.0403004172104942,
      "grad_norm": 0.058297354727983475,
      "learning_rate": 5.520029760117428e-06,
      "loss": 0.0018,
      "step": 162200
    },
    {
      "epoch": 1.04036455410428,
      "grad_norm": 0.11905381828546524,
      "learning_rate": 5.519473088939247e-06,
      "loss": 0.0019,
      "step": 162210
    },
    {
      "epoch": 1.0404286909980662,
      "grad_norm": 0.06737947463989258,
      "learning_rate": 5.5189164112517e-06,
      "loss": 0.0009,
      "step": 162220
    },
    {
      "epoch": 1.0404928278918524,
      "grad_norm": 0.14862266182899475,
      "learning_rate": 5.518359727061757e-06,
      "loss": 0.0027,
      "step": 162230
    },
    {
      "epoch": 1.0405569647856385,
      "grad_norm": 0.15508893132209778,
      "learning_rate": 5.517803036376399e-06,
      "loss": 0.0017,
      "step": 162240
    },
    {
      "epoch": 1.0406211016794247,
      "grad_norm": 0.09064960479736328,
      "learning_rate": 5.5172463392025975e-06,
      "loss": 0.0028,
      "step": 162250
    },
    {
      "epoch": 1.0406852385732106,
      "grad_norm": 0.1738806515932083,
      "learning_rate": 5.51668963554733e-06,
      "loss": 0.0019,
      "step": 162260
    },
    {
      "epoch": 1.0407493754669968,
      "grad_norm": 0.3050050735473633,
      "learning_rate": 5.5161329254175715e-06,
      "loss": 0.003,
      "step": 162270
    },
    {
      "epoch": 1.040813512360783,
      "grad_norm": 0.04364198073744774,
      "learning_rate": 5.515576208820299e-06,
      "loss": 0.0014,
      "step": 162280
    },
    {
      "epoch": 1.040877649254569,
      "grad_norm": 0.05057888850569725,
      "learning_rate": 5.515019485762488e-06,
      "loss": 0.0019,
      "step": 162290
    },
    {
      "epoch": 1.040941786148355,
      "grad_norm": 0.1384192705154419,
      "learning_rate": 5.514462756251114e-06,
      "loss": 0.0025,
      "step": 162300
    },
    {
      "epoch": 1.0410059230421411,
      "grad_norm": 0.03384971618652344,
      "learning_rate": 5.513906020293156e-06,
      "loss": 0.0017,
      "step": 162310
    },
    {
      "epoch": 1.0410700599359273,
      "grad_norm": 0.07956333458423615,
      "learning_rate": 5.513349277895587e-06,
      "loss": 0.0013,
      "step": 162320
    },
    {
      "epoch": 1.0411341968297134,
      "grad_norm": 0.07815622538328171,
      "learning_rate": 5.512792529065385e-06,
      "loss": 0.0022,
      "step": 162330
    },
    {
      "epoch": 1.0411983337234993,
      "grad_norm": 0.08500978350639343,
      "learning_rate": 5.512235773809526e-06,
      "loss": 0.0013,
      "step": 162340
    },
    {
      "epoch": 1.0412624706172855,
      "grad_norm": 0.08306215703487396,
      "learning_rate": 5.511679012134987e-06,
      "loss": 0.0014,
      "step": 162350
    },
    {
      "epoch": 1.0413266075110716,
      "grad_norm": 0.28995710611343384,
      "learning_rate": 5.511122244048744e-06,
      "loss": 0.0017,
      "step": 162360
    },
    {
      "epoch": 1.0413907444048578,
      "grad_norm": 0.16887091100215912,
      "learning_rate": 5.510565469557774e-06,
      "loss": 0.001,
      "step": 162370
    },
    {
      "epoch": 1.0414548812986437,
      "grad_norm": 0.12421572208404541,
      "learning_rate": 5.510008688669053e-06,
      "loss": 0.0026,
      "step": 162380
    },
    {
      "epoch": 1.0415190181924299,
      "grad_norm": 0.09098473191261292,
      "learning_rate": 5.50945190138956e-06,
      "loss": 0.0023,
      "step": 162390
    },
    {
      "epoch": 1.041583155086216,
      "grad_norm": 0.15700791776180267,
      "learning_rate": 5.508895107726269e-06,
      "loss": 0.0015,
      "step": 162400
    },
    {
      "epoch": 1.0416472919800022,
      "grad_norm": 0.13414964079856873,
      "learning_rate": 5.50833830768616e-06,
      "loss": 0.0012,
      "step": 162410
    },
    {
      "epoch": 1.0417114288737883,
      "grad_norm": 0.1681639552116394,
      "learning_rate": 5.507781501276207e-06,
      "loss": 0.0026,
      "step": 162420
    },
    {
      "epoch": 1.0417755657675742,
      "grad_norm": 0.14209985733032227,
      "learning_rate": 5.50722468850339e-06,
      "loss": 0.0025,
      "step": 162430
    },
    {
      "epoch": 1.0418397026613604,
      "grad_norm": 0.11714351177215576,
      "learning_rate": 5.506667869374685e-06,
      "loss": 0.0027,
      "step": 162440
    },
    {
      "epoch": 1.0419038395551465,
      "grad_norm": 0.07386796921491623,
      "learning_rate": 5.50611104389707e-06,
      "loss": 0.0015,
      "step": 162450
    },
    {
      "epoch": 1.0419679764489327,
      "grad_norm": 0.3755844533443451,
      "learning_rate": 5.505554212077522e-06,
      "loss": 0.0019,
      "step": 162460
    },
    {
      "epoch": 1.0420321133427186,
      "grad_norm": 0.035572972148656845,
      "learning_rate": 5.504997373923018e-06,
      "loss": 0.0013,
      "step": 162470
    },
    {
      "epoch": 1.0420962502365048,
      "grad_norm": 0.04977955296635628,
      "learning_rate": 5.504440529440536e-06,
      "loss": 0.0012,
      "step": 162480
    },
    {
      "epoch": 1.042160387130291,
      "grad_norm": 0.05770646035671234,
      "learning_rate": 5.503883678637053e-06,
      "loss": 0.0013,
      "step": 162490
    },
    {
      "epoch": 1.042224524024077,
      "grad_norm": 0.006688651163130999,
      "learning_rate": 5.503326821519549e-06,
      "loss": 0.0018,
      "step": 162500
    },
    {
      "epoch": 1.042288660917863,
      "grad_norm": 0.009053279645740986,
      "learning_rate": 5.502769958094999e-06,
      "loss": 0.0013,
      "step": 162510
    },
    {
      "epoch": 1.0423527978116491,
      "grad_norm": 0.013610010035336018,
      "learning_rate": 5.502213088370383e-06,
      "loss": 0.0014,
      "step": 162520
    },
    {
      "epoch": 1.0424169347054353,
      "grad_norm": 0.007402785588055849,
      "learning_rate": 5.501656212352676e-06,
      "loss": 0.0013,
      "step": 162530
    },
    {
      "epoch": 1.0424810715992214,
      "grad_norm": 0.2344551831483841,
      "learning_rate": 5.50109933004886e-06,
      "loss": 0.0013,
      "step": 162540
    },
    {
      "epoch": 1.0425452084930076,
      "grad_norm": 0.08997918665409088,
      "learning_rate": 5.500542441465911e-06,
      "loss": 0.0019,
      "step": 162550
    },
    {
      "epoch": 1.0426093453867935,
      "grad_norm": 0.13036172091960907,
      "learning_rate": 5.499985546610808e-06,
      "loss": 0.0018,
      "step": 162560
    },
    {
      "epoch": 1.0426734822805797,
      "grad_norm": 0.1589774489402771,
      "learning_rate": 5.499428645490527e-06,
      "loss": 0.0018,
      "step": 162570
    },
    {
      "epoch": 1.0427376191743658,
      "grad_norm": 0.12301304936408997,
      "learning_rate": 5.498871738112048e-06,
      "loss": 0.0024,
      "step": 162580
    },
    {
      "epoch": 1.042801756068152,
      "grad_norm": 0.09392508119344711,
      "learning_rate": 5.498314824482351e-06,
      "loss": 0.0023,
      "step": 162590
    },
    {
      "epoch": 1.0428658929619379,
      "grad_norm": 0.01861235871911049,
      "learning_rate": 5.497757904608414e-06,
      "loss": 0.0022,
      "step": 162600
    },
    {
      "epoch": 1.042930029855724,
      "grad_norm": 0.09133550524711609,
      "learning_rate": 5.497200978497212e-06,
      "loss": 0.0022,
      "step": 162610
    },
    {
      "epoch": 1.0429941667495102,
      "grad_norm": 0.2399628758430481,
      "learning_rate": 5.4966440461557295e-06,
      "loss": 0.0012,
      "step": 162620
    },
    {
      "epoch": 1.0430583036432963,
      "grad_norm": 0.19869330525398254,
      "learning_rate": 5.49608710759094e-06,
      "loss": 0.0034,
      "step": 162630
    },
    {
      "epoch": 1.0431224405370823,
      "grad_norm": 0.1504802405834198,
      "learning_rate": 5.4955301628098246e-06,
      "loss": 0.0021,
      "step": 162640
    },
    {
      "epoch": 1.0431865774308684,
      "grad_norm": 0.0682467445731163,
      "learning_rate": 5.494973211819363e-06,
      "loss": 0.0034,
      "step": 162650
    },
    {
      "epoch": 1.0432507143246545,
      "grad_norm": 0.15123432874679565,
      "learning_rate": 5.494416254626533e-06,
      "loss": 0.0028,
      "step": 162660
    },
    {
      "epoch": 1.0433148512184407,
      "grad_norm": 0.050001200288534164,
      "learning_rate": 5.493859291238313e-06,
      "loss": 0.0017,
      "step": 162670
    },
    {
      "epoch": 1.0433789881122268,
      "grad_norm": 0.04714712128043175,
      "learning_rate": 5.493302321661684e-06,
      "loss": 0.0013,
      "step": 162680
    },
    {
      "epoch": 1.0434431250060128,
      "grad_norm": 0.04859734699130058,
      "learning_rate": 5.492745345903625e-06,
      "loss": 0.0013,
      "step": 162690
    },
    {
      "epoch": 1.043507261899799,
      "grad_norm": 0.04963231831789017,
      "learning_rate": 5.492188363971114e-06,
      "loss": 0.0023,
      "step": 162700
    },
    {
      "epoch": 1.043571398793585,
      "grad_norm": 0.14139969646930695,
      "learning_rate": 5.491631375871131e-06,
      "loss": 0.0025,
      "step": 162710
    },
    {
      "epoch": 1.0436355356873712,
      "grad_norm": 0.08460360765457153,
      "learning_rate": 5.491074381610655e-06,
      "loss": 0.0019,
      "step": 162720
    },
    {
      "epoch": 1.0436996725811571,
      "grad_norm": 0.019998345524072647,
      "learning_rate": 5.490517381196667e-06,
      "loss": 0.0008,
      "step": 162730
    },
    {
      "epoch": 1.0437638094749433,
      "grad_norm": 0.2822844386100769,
      "learning_rate": 5.489960374636145e-06,
      "loss": 0.0014,
      "step": 162740
    },
    {
      "epoch": 1.0438279463687294,
      "grad_norm": 0.14481887221336365,
      "learning_rate": 5.489403361936071e-06,
      "loss": 0.0011,
      "step": 162750
    },
    {
      "epoch": 1.0438920832625156,
      "grad_norm": 0.2062295526266098,
      "learning_rate": 5.488846343103421e-06,
      "loss": 0.0018,
      "step": 162760
    },
    {
      "epoch": 1.0439562201563015,
      "grad_norm": 0.08027106523513794,
      "learning_rate": 5.488289318145177e-06,
      "loss": 0.002,
      "step": 162770
    },
    {
      "epoch": 1.0440203570500877,
      "grad_norm": 0.0023542765993624926,
      "learning_rate": 5.48773228706832e-06,
      "loss": 0.0013,
      "step": 162780
    },
    {
      "epoch": 1.0440844939438738,
      "grad_norm": 0.1881311982870102,
      "learning_rate": 5.4871752498798284e-06,
      "loss": 0.0028,
      "step": 162790
    },
    {
      "epoch": 1.04414863083766,
      "grad_norm": 0.01573358289897442,
      "learning_rate": 5.4866182065866824e-06,
      "loss": 0.0016,
      "step": 162800
    },
    {
      "epoch": 1.044212767731446,
      "grad_norm": 0.0733632743358612,
      "learning_rate": 5.486061157195862e-06,
      "loss": 0.0016,
      "step": 162810
    },
    {
      "epoch": 1.044276904625232,
      "grad_norm": 0.12339013069868088,
      "learning_rate": 5.485504101714349e-06,
      "loss": 0.0029,
      "step": 162820
    },
    {
      "epoch": 1.0443410415190182,
      "grad_norm": 0.13790413737297058,
      "learning_rate": 5.484947040149122e-06,
      "loss": 0.005,
      "step": 162830
    },
    {
      "epoch": 1.0444051784128043,
      "grad_norm": 0.2742564082145691,
      "learning_rate": 5.4843899725071635e-06,
      "loss": 0.0032,
      "step": 162840
    },
    {
      "epoch": 1.0444693153065905,
      "grad_norm": 0.06959738582372665,
      "learning_rate": 5.483832898795452e-06,
      "loss": 0.001,
      "step": 162850
    },
    {
      "epoch": 1.0445334522003764,
      "grad_norm": 0.11053453385829926,
      "learning_rate": 5.483275819020967e-06,
      "loss": 0.0013,
      "step": 162860
    },
    {
      "epoch": 1.0445975890941626,
      "grad_norm": 0.05350212752819061,
      "learning_rate": 5.482718733190691e-06,
      "loss": 0.0013,
      "step": 162870
    },
    {
      "epoch": 1.0446617259879487,
      "grad_norm": 0.10666050016880035,
      "learning_rate": 5.4821616413116055e-06,
      "loss": 0.0026,
      "step": 162880
    },
    {
      "epoch": 1.0447258628817349,
      "grad_norm": 0.08770793676376343,
      "learning_rate": 5.481604543390688e-06,
      "loss": 0.0011,
      "step": 162890
    },
    {
      "epoch": 1.0447899997755208,
      "grad_norm": 0.03143342584371567,
      "learning_rate": 5.481047439434923e-06,
      "loss": 0.0013,
      "step": 162900
    },
    {
      "epoch": 1.044854136669307,
      "grad_norm": 0.023717986419796944,
      "learning_rate": 5.480490329451289e-06,
      "loss": 0.0009,
      "step": 162910
    },
    {
      "epoch": 1.044918273563093,
      "grad_norm": 0.035895369946956635,
      "learning_rate": 5.479933213446768e-06,
      "loss": 0.0013,
      "step": 162920
    },
    {
      "epoch": 1.0449824104568792,
      "grad_norm": 0.13733455538749695,
      "learning_rate": 5.479376091428341e-06,
      "loss": 0.0025,
      "step": 162930
    },
    {
      "epoch": 1.0450465473506652,
      "grad_norm": 0.10928450524806976,
      "learning_rate": 5.478818963402989e-06,
      "loss": 0.0034,
      "step": 162940
    },
    {
      "epoch": 1.0451106842444513,
      "grad_norm": 0.017055131494998932,
      "learning_rate": 5.478261829377694e-06,
      "loss": 0.0022,
      "step": 162950
    },
    {
      "epoch": 1.0451748211382375,
      "grad_norm": 0.28910472989082336,
      "learning_rate": 5.477704689359435e-06,
      "loss": 0.0023,
      "step": 162960
    },
    {
      "epoch": 1.0452389580320236,
      "grad_norm": 0.03387433663010597,
      "learning_rate": 5.477147543355195e-06,
      "loss": 0.0018,
      "step": 162970
    },
    {
      "epoch": 1.0453030949258098,
      "grad_norm": 0.029093138873577118,
      "learning_rate": 5.476590391371956e-06,
      "loss": 0.002,
      "step": 162980
    },
    {
      "epoch": 1.0453672318195957,
      "grad_norm": 0.02908753789961338,
      "learning_rate": 5.476033233416697e-06,
      "loss": 0.0025,
      "step": 162990
    },
    {
      "epoch": 1.0454313687133818,
      "grad_norm": 0.20391933619976044,
      "learning_rate": 5.4754760694964035e-06,
      "loss": 0.0022,
      "step": 163000
    },
    {
      "epoch": 1.045495505607168,
      "grad_norm": 0.056450072675943375,
      "learning_rate": 5.4749188996180545e-06,
      "loss": 0.002,
      "step": 163010
    },
    {
      "epoch": 1.0455596425009541,
      "grad_norm": 0.28571149706840515,
      "learning_rate": 5.4743617237886315e-06,
      "loss": 0.002,
      "step": 163020
    },
    {
      "epoch": 1.04562377939474,
      "grad_norm": 0.05657358467578888,
      "learning_rate": 5.4738045420151184e-06,
      "loss": 0.0014,
      "step": 163030
    },
    {
      "epoch": 1.0456879162885262,
      "grad_norm": 0.24326853454113007,
      "learning_rate": 5.473247354304495e-06,
      "loss": 0.0013,
      "step": 163040
    },
    {
      "epoch": 1.0457520531823123,
      "grad_norm": 0.11728103458881378,
      "learning_rate": 5.472690160663745e-06,
      "loss": 0.0022,
      "step": 163050
    },
    {
      "epoch": 1.0458161900760985,
      "grad_norm": 0.026328377425670624,
      "learning_rate": 5.4721329610998475e-06,
      "loss": 0.0028,
      "step": 163060
    },
    {
      "epoch": 1.0458803269698844,
      "grad_norm": 0.05746069177985191,
      "learning_rate": 5.471575755619788e-06,
      "loss": 0.0016,
      "step": 163070
    },
    {
      "epoch": 1.0459444638636706,
      "grad_norm": 0.02624838799238205,
      "learning_rate": 5.471018544230546e-06,
      "loss": 0.0016,
      "step": 163080
    },
    {
      "epoch": 1.0460086007574567,
      "grad_norm": 0.06770443171262741,
      "learning_rate": 5.470461326939107e-06,
      "loss": 0.0019,
      "step": 163090
    },
    {
      "epoch": 1.0460727376512429,
      "grad_norm": 0.09298177063465118,
      "learning_rate": 5.46990410375245e-06,
      "loss": 0.0033,
      "step": 163100
    },
    {
      "epoch": 1.046136874545029,
      "grad_norm": 0.18723025918006897,
      "learning_rate": 5.469346874677559e-06,
      "loss": 0.0022,
      "step": 163110
    },
    {
      "epoch": 1.046201011438815,
      "grad_norm": 0.07983297109603882,
      "learning_rate": 5.468789639721416e-06,
      "loss": 0.002,
      "step": 163120
    },
    {
      "epoch": 1.046265148332601,
      "grad_norm": 0.15169022977352142,
      "learning_rate": 5.468232398891004e-06,
      "loss": 0.0012,
      "step": 163130
    },
    {
      "epoch": 1.0463292852263872,
      "grad_norm": 0.02613871730864048,
      "learning_rate": 5.4676751521933055e-06,
      "loss": 0.0007,
      "step": 163140
    },
    {
      "epoch": 1.0463934221201734,
      "grad_norm": 0.03867525979876518,
      "learning_rate": 5.467117899635302e-06,
      "loss": 0.0011,
      "step": 163150
    },
    {
      "epoch": 1.0464575590139593,
      "grad_norm": 0.2037162482738495,
      "learning_rate": 5.466560641223979e-06,
      "loss": 0.0027,
      "step": 163160
    },
    {
      "epoch": 1.0465216959077455,
      "grad_norm": 0.008311222307384014,
      "learning_rate": 5.466003376966317e-06,
      "loss": 0.0013,
      "step": 163170
    },
    {
      "epoch": 1.0465858328015316,
      "grad_norm": 0.024490531533956528,
      "learning_rate": 5.4654461068693e-06,
      "loss": 0.0023,
      "step": 163180
    },
    {
      "epoch": 1.0466499696953178,
      "grad_norm": 0.005609770305454731,
      "learning_rate": 5.4648888309399104e-06,
      "loss": 0.0018,
      "step": 163190
    },
    {
      "epoch": 1.0467141065891037,
      "grad_norm": 0.18306541442871094,
      "learning_rate": 5.464331549185131e-06,
      "loss": 0.0022,
      "step": 163200
    },
    {
      "epoch": 1.0467782434828898,
      "grad_norm": 0.08656865358352661,
      "learning_rate": 5.463774261611946e-06,
      "loss": 0.0015,
      "step": 163210
    },
    {
      "epoch": 1.046842380376676,
      "grad_norm": 0.042053062468767166,
      "learning_rate": 5.463216968227339e-06,
      "loss": 0.002,
      "step": 163220
    },
    {
      "epoch": 1.0469065172704621,
      "grad_norm": 0.10596334934234619,
      "learning_rate": 5.4626596690382905e-06,
      "loss": 0.0013,
      "step": 163230
    },
    {
      "epoch": 1.046970654164248,
      "grad_norm": 0.09371647983789444,
      "learning_rate": 5.462102364051787e-06,
      "loss": 0.0016,
      "step": 163240
    },
    {
      "epoch": 1.0470347910580342,
      "grad_norm": 0.07989853620529175,
      "learning_rate": 5.46154505327481e-06,
      "loss": 0.0021,
      "step": 163250
    },
    {
      "epoch": 1.0470989279518204,
      "grad_norm": 0.04100971296429634,
      "learning_rate": 5.460987736714344e-06,
      "loss": 0.0019,
      "step": 163260
    },
    {
      "epoch": 1.0471630648456065,
      "grad_norm": 0.041018567979335785,
      "learning_rate": 5.460430414377371e-06,
      "loss": 0.0017,
      "step": 163270
    },
    {
      "epoch": 1.0472272017393927,
      "grad_norm": 0.0300181545317173,
      "learning_rate": 5.459873086270876e-06,
      "loss": 0.0022,
      "step": 163280
    },
    {
      "epoch": 1.0472913386331786,
      "grad_norm": 0.26561641693115234,
      "learning_rate": 5.459315752401843e-06,
      "loss": 0.0027,
      "step": 163290
    },
    {
      "epoch": 1.0473554755269647,
      "grad_norm": 0.10002687573432922,
      "learning_rate": 5.4587584127772566e-06,
      "loss": 0.0012,
      "step": 163300
    },
    {
      "epoch": 1.0474196124207509,
      "grad_norm": 0.07808393239974976,
      "learning_rate": 5.4582010674040985e-06,
      "loss": 0.0024,
      "step": 163310
    },
    {
      "epoch": 1.047483749314537,
      "grad_norm": 0.07971341907978058,
      "learning_rate": 5.457643716289354e-06,
      "loss": 0.0014,
      "step": 163320
    },
    {
      "epoch": 1.047547886208323,
      "grad_norm": 0.03393528610467911,
      "learning_rate": 5.457086359440006e-06,
      "loss": 0.0021,
      "step": 163330
    },
    {
      "epoch": 1.047612023102109,
      "grad_norm": 0.24799518287181854,
      "learning_rate": 5.456528996863038e-06,
      "loss": 0.0019,
      "step": 163340
    },
    {
      "epoch": 1.0476761599958953,
      "grad_norm": 0.10391895473003387,
      "learning_rate": 5.4559716285654385e-06,
      "loss": 0.0018,
      "step": 163350
    },
    {
      "epoch": 1.0477402968896814,
      "grad_norm": 0.07402423769235611,
      "learning_rate": 5.4554142545541854e-06,
      "loss": 0.0022,
      "step": 163360
    },
    {
      "epoch": 1.0478044337834673,
      "grad_norm": 0.09968770295381546,
      "learning_rate": 5.454856874836268e-06,
      "loss": 0.0026,
      "step": 163370
    },
    {
      "epoch": 1.0478685706772535,
      "grad_norm": 0.16183412075042725,
      "learning_rate": 5.454299489418669e-06,
      "loss": 0.0012,
      "step": 163380
    },
    {
      "epoch": 1.0479327075710396,
      "grad_norm": 0.09767698496580124,
      "learning_rate": 5.4537420983083725e-06,
      "loss": 0.0014,
      "step": 163390
    },
    {
      "epoch": 1.0479968444648258,
      "grad_norm": 0.11284589767456055,
      "learning_rate": 5.453184701512362e-06,
      "loss": 0.0029,
      "step": 163400
    },
    {
      "epoch": 1.048060981358612,
      "grad_norm": 0.029451711103320122,
      "learning_rate": 5.452627299037625e-06,
      "loss": 0.0012,
      "step": 163410
    },
    {
      "epoch": 1.0481251182523978,
      "grad_norm": 0.14543209969997406,
      "learning_rate": 5.452069890891143e-06,
      "loss": 0.0011,
      "step": 163420
    },
    {
      "epoch": 1.048189255146184,
      "grad_norm": 0.14817456901073456,
      "learning_rate": 5.451512477079904e-06,
      "loss": 0.0019,
      "step": 163430
    },
    {
      "epoch": 1.0482533920399701,
      "grad_norm": 0.012029202654957771,
      "learning_rate": 5.450955057610889e-06,
      "loss": 0.0023,
      "step": 163440
    },
    {
      "epoch": 1.0483175289337563,
      "grad_norm": 0.06852412223815918,
      "learning_rate": 5.450397632491087e-06,
      "loss": 0.0025,
      "step": 163450
    },
    {
      "epoch": 1.0483816658275422,
      "grad_norm": 0.057883575558662415,
      "learning_rate": 5.449840201727479e-06,
      "loss": 0.002,
      "step": 163460
    },
    {
      "epoch": 1.0484458027213284,
      "grad_norm": 0.1246136873960495,
      "learning_rate": 5.449282765327053e-06,
      "loss": 0.0009,
      "step": 163470
    },
    {
      "epoch": 1.0485099396151145,
      "grad_norm": 0.058335140347480774,
      "learning_rate": 5.4487253232967915e-06,
      "loss": 0.0009,
      "step": 163480
    },
    {
      "epoch": 1.0485740765089007,
      "grad_norm": 0.17894196510314941,
      "learning_rate": 5.448167875643681e-06,
      "loss": 0.0012,
      "step": 163490
    },
    {
      "epoch": 1.0486382134026866,
      "grad_norm": 0.06329632550477982,
      "learning_rate": 5.447610422374709e-06,
      "loss": 0.0012,
      "step": 163500
    },
    {
      "epoch": 1.0487023502964727,
      "grad_norm": 0.0986696183681488,
      "learning_rate": 5.447052963496855e-06,
      "loss": 0.0022,
      "step": 163510
    },
    {
      "epoch": 1.048766487190259,
      "grad_norm": 0.03802545368671417,
      "learning_rate": 5.44649549901711e-06,
      "loss": 0.0021,
      "step": 163520
    },
    {
      "epoch": 1.048830624084045,
      "grad_norm": 0.11675931513309479,
      "learning_rate": 5.445938028942456e-06,
      "loss": 0.003,
      "step": 163530
    },
    {
      "epoch": 1.0488947609778312,
      "grad_norm": 0.062087420374155045,
      "learning_rate": 5.445380553279883e-06,
      "loss": 0.0018,
      "step": 163540
    },
    {
      "epoch": 1.0489588978716171,
      "grad_norm": 0.08303030580282211,
      "learning_rate": 5.4448230720363705e-06,
      "loss": 0.0032,
      "step": 163550
    },
    {
      "epoch": 1.0490230347654033,
      "grad_norm": 0.07520963251590729,
      "learning_rate": 5.4442655852189086e-06,
      "loss": 0.0015,
      "step": 163560
    },
    {
      "epoch": 1.0490871716591894,
      "grad_norm": 0.06167283281683922,
      "learning_rate": 5.44370809283448e-06,
      "loss": 0.0013,
      "step": 163570
    },
    {
      "epoch": 1.0491513085529756,
      "grad_norm": 0.1583038866519928,
      "learning_rate": 5.443150594890073e-06,
      "loss": 0.0037,
      "step": 163580
    },
    {
      "epoch": 1.0492154454467615,
      "grad_norm": 0.07042469084262848,
      "learning_rate": 5.442593091392671e-06,
      "loss": 0.0033,
      "step": 163590
    },
    {
      "epoch": 1.0492795823405476,
      "grad_norm": 0.16713710129261017,
      "learning_rate": 5.4420355823492624e-06,
      "loss": 0.0014,
      "step": 163600
    },
    {
      "epoch": 1.0493437192343338,
      "grad_norm": 0.04701533913612366,
      "learning_rate": 5.441478067766831e-06,
      "loss": 0.0011,
      "step": 163610
    },
    {
      "epoch": 1.04940785612812,
      "grad_norm": 0.07531076669692993,
      "learning_rate": 5.440920547652364e-06,
      "loss": 0.002,
      "step": 163620
    },
    {
      "epoch": 1.0494719930219059,
      "grad_norm": 0.022455720230937004,
      "learning_rate": 5.440363022012849e-06,
      "loss": 0.0019,
      "step": 163630
    },
    {
      "epoch": 1.049536129915692,
      "grad_norm": 0.16131709516048431,
      "learning_rate": 5.439805490855269e-06,
      "loss": 0.0023,
      "step": 163640
    },
    {
      "epoch": 1.0496002668094782,
      "grad_norm": 0.07646039873361588,
      "learning_rate": 5.439247954186613e-06,
      "loss": 0.0018,
      "step": 163650
    },
    {
      "epoch": 1.0496644037032643,
      "grad_norm": 0.0025511719286441803,
      "learning_rate": 5.438690412013865e-06,
      "loss": 0.002,
      "step": 163660
    },
    {
      "epoch": 1.0497285405970502,
      "grad_norm": 0.07018695771694183,
      "learning_rate": 5.438132864344013e-06,
      "loss": 0.0017,
      "step": 163670
    },
    {
      "epoch": 1.0497926774908364,
      "grad_norm": 0.079002745449543,
      "learning_rate": 5.4375753111840435e-06,
      "loss": 0.0009,
      "step": 163680
    },
    {
      "epoch": 1.0498568143846225,
      "grad_norm": 0.11101429909467697,
      "learning_rate": 5.437017752540943e-06,
      "loss": 0.0019,
      "step": 163690
    },
    {
      "epoch": 1.0499209512784087,
      "grad_norm": 0.07598632574081421,
      "learning_rate": 5.4364601884216974e-06,
      "loss": 0.0013,
      "step": 163700
    },
    {
      "epoch": 1.0499850881721948,
      "grad_norm": 0.12365309149026871,
      "learning_rate": 5.435902618833293e-06,
      "loss": 0.0034,
      "step": 163710
    },
    {
      "epoch": 1.0500492250659808,
      "grad_norm": 0.05901951342821121,
      "learning_rate": 5.435345043782717e-06,
      "loss": 0.0013,
      "step": 163720
    },
    {
      "epoch": 1.050113361959767,
      "grad_norm": 0.01743759587407112,
      "learning_rate": 5.434787463276959e-06,
      "loss": 0.0014,
      "step": 163730
    },
    {
      "epoch": 1.050177498853553,
      "grad_norm": 0.13494150340557098,
      "learning_rate": 5.434229877322999e-06,
      "loss": 0.0023,
      "step": 163740
    },
    {
      "epoch": 1.0502416357473392,
      "grad_norm": 0.09043806046247482,
      "learning_rate": 5.433672285927831e-06,
      "loss": 0.0016,
      "step": 163750
    },
    {
      "epoch": 1.0503057726411251,
      "grad_norm": 0.03344087675213814,
      "learning_rate": 5.433114689098439e-06,
      "loss": 0.0014,
      "step": 163760
    },
    {
      "epoch": 1.0503699095349113,
      "grad_norm": 0.023698963224887848,
      "learning_rate": 5.432557086841811e-06,
      "loss": 0.0024,
      "step": 163770
    },
    {
      "epoch": 1.0504340464286974,
      "grad_norm": 0.22035405039787292,
      "learning_rate": 5.4319994791649336e-06,
      "loss": 0.0016,
      "step": 163780
    },
    {
      "epoch": 1.0504981833224836,
      "grad_norm": 0.020499106496572495,
      "learning_rate": 5.431441866074793e-06,
      "loss": 0.001,
      "step": 163790
    },
    {
      "epoch": 1.0505623202162697,
      "grad_norm": 0.03326013311743736,
      "learning_rate": 5.430884247578379e-06,
      "loss": 0.0016,
      "step": 163800
    },
    {
      "epoch": 1.0506264571100556,
      "grad_norm": 0.24346397817134857,
      "learning_rate": 5.4303266236826756e-06,
      "loss": 0.0017,
      "step": 163810
    },
    {
      "epoch": 1.0506905940038418,
      "grad_norm": 0.21626178920269012,
      "learning_rate": 5.429768994394673e-06,
      "loss": 0.0026,
      "step": 163820
    },
    {
      "epoch": 1.050754730897628,
      "grad_norm": 0.03512268513441086,
      "learning_rate": 5.429211359721357e-06,
      "loss": 0.001,
      "step": 163830
    },
    {
      "epoch": 1.050818867791414,
      "grad_norm": 0.09154361486434937,
      "learning_rate": 5.4286537196697165e-06,
      "loss": 0.0013,
      "step": 163840
    },
    {
      "epoch": 1.0508830046852,
      "grad_norm": 0.2012699842453003,
      "learning_rate": 5.428096074246738e-06,
      "loss": 0.0014,
      "step": 163850
    },
    {
      "epoch": 1.0509471415789862,
      "grad_norm": 0.07336778938770294,
      "learning_rate": 5.427538423459411e-06,
      "loss": 0.0015,
      "step": 163860
    },
    {
      "epoch": 1.0510112784727723,
      "grad_norm": 0.10989232361316681,
      "learning_rate": 5.426980767314721e-06,
      "loss": 0.0013,
      "step": 163870
    },
    {
      "epoch": 1.0510754153665585,
      "grad_norm": 0.007294870913028717,
      "learning_rate": 5.426423105819658e-06,
      "loss": 0.0019,
      "step": 163880
    },
    {
      "epoch": 1.0511395522603444,
      "grad_norm": 0.13803347945213318,
      "learning_rate": 5.425865438981207e-06,
      "loss": 0.0016,
      "step": 163890
    },
    {
      "epoch": 1.0512036891541305,
      "grad_norm": 0.0911358967423439,
      "learning_rate": 5.425307766806359e-06,
      "loss": 0.0015,
      "step": 163900
    },
    {
      "epoch": 1.0512678260479167,
      "grad_norm": 0.18064068257808685,
      "learning_rate": 5.4247500893021e-06,
      "loss": 0.0019,
      "step": 163910
    },
    {
      "epoch": 1.0513319629417028,
      "grad_norm": 0.1788107454776764,
      "learning_rate": 5.4241924064754195e-06,
      "loss": 0.0025,
      "step": 163920
    },
    {
      "epoch": 1.0513960998354888,
      "grad_norm": 0.06165863946080208,
      "learning_rate": 5.423634718333304e-06,
      "loss": 0.002,
      "step": 163930
    },
    {
      "epoch": 1.051460236729275,
      "grad_norm": 0.05214349180459976,
      "learning_rate": 5.423077024882743e-06,
      "loss": 0.0011,
      "step": 163940
    },
    {
      "epoch": 1.051524373623061,
      "grad_norm": 0.10383394360542297,
      "learning_rate": 5.422519326130725e-06,
      "loss": 0.0027,
      "step": 163950
    },
    {
      "epoch": 1.0515885105168472,
      "grad_norm": 0.1499389261007309,
      "learning_rate": 5.421961622084239e-06,
      "loss": 0.0026,
      "step": 163960
    },
    {
      "epoch": 1.0516526474106334,
      "grad_norm": 0.17220918834209442,
      "learning_rate": 5.42140391275027e-06,
      "loss": 0.0012,
      "step": 163970
    },
    {
      "epoch": 1.0517167843044193,
      "grad_norm": 0.058838699012994766,
      "learning_rate": 5.420846198135808e-06,
      "loss": 0.0022,
      "step": 163980
    },
    {
      "epoch": 1.0517809211982054,
      "grad_norm": 0.011541483923792839,
      "learning_rate": 5.420288478247846e-06,
      "loss": 0.0028,
      "step": 163990
    },
    {
      "epoch": 1.0518450580919916,
      "grad_norm": 0.13648605346679688,
      "learning_rate": 5.419730753093366e-06,
      "loss": 0.0037,
      "step": 164000
    },
    {
      "epoch": 1.0519091949857777,
      "grad_norm": 0.12824368476867676,
      "learning_rate": 5.419173022679361e-06,
      "loss": 0.0023,
      "step": 164010
    },
    {
      "epoch": 1.0519733318795637,
      "grad_norm": 0.01217710506170988,
      "learning_rate": 5.418615287012818e-06,
      "loss": 0.0018,
      "step": 164020
    },
    {
      "epoch": 1.0520374687733498,
      "grad_norm": 0.05712335929274559,
      "learning_rate": 5.418057546100726e-06,
      "loss": 0.0011,
      "step": 164030
    },
    {
      "epoch": 1.052101605667136,
      "grad_norm": 0.04380573332309723,
      "learning_rate": 5.417499799950075e-06,
      "loss": 0.0021,
      "step": 164040
    },
    {
      "epoch": 1.052165742560922,
      "grad_norm": 0.015386135317385197,
      "learning_rate": 5.416942048567853e-06,
      "loss": 0.0016,
      "step": 164050
    },
    {
      "epoch": 1.052229879454708,
      "grad_norm": 0.10284862667322159,
      "learning_rate": 5.416384291961049e-06,
      "loss": 0.0023,
      "step": 164060
    },
    {
      "epoch": 1.0522940163484942,
      "grad_norm": 0.09124301373958588,
      "learning_rate": 5.415826530136653e-06,
      "loss": 0.0016,
      "step": 164070
    },
    {
      "epoch": 1.0523581532422803,
      "grad_norm": 0.005526949185878038,
      "learning_rate": 5.415268763101652e-06,
      "loss": 0.0009,
      "step": 164080
    },
    {
      "epoch": 1.0524222901360665,
      "grad_norm": 0.1582203507423401,
      "learning_rate": 5.414710990863038e-06,
      "loss": 0.002,
      "step": 164090
    },
    {
      "epoch": 1.0524864270298526,
      "grad_norm": 0.046717043966054916,
      "learning_rate": 5.4141532134277976e-06,
      "loss": 0.0012,
      "step": 164100
    },
    {
      "epoch": 1.0525505639236385,
      "grad_norm": 0.19061030447483063,
      "learning_rate": 5.413595430802923e-06,
      "loss": 0.0018,
      "step": 164110
    },
    {
      "epoch": 1.0526147008174247,
      "grad_norm": 0.3718620836734772,
      "learning_rate": 5.413037642995399e-06,
      "loss": 0.0046,
      "step": 164120
    },
    {
      "epoch": 1.0526788377112108,
      "grad_norm": 0.03796644136309624,
      "learning_rate": 5.41247985001222e-06,
      "loss": 0.0016,
      "step": 164130
    },
    {
      "epoch": 1.052742974604997,
      "grad_norm": 0.14698080718517303,
      "learning_rate": 5.411922051860373e-06,
      "loss": 0.0022,
      "step": 164140
    },
    {
      "epoch": 1.052807111498783,
      "grad_norm": 0.05480699986219406,
      "learning_rate": 5.41136424854685e-06,
      "loss": 0.0026,
      "step": 164150
    },
    {
      "epoch": 1.052871248392569,
      "grad_norm": 0.11102596670389175,
      "learning_rate": 5.410806440078637e-06,
      "loss": 0.0017,
      "step": 164160
    },
    {
      "epoch": 1.0529353852863552,
      "grad_norm": 0.057191140949726105,
      "learning_rate": 5.4102486264627285e-06,
      "loss": 0.0028,
      "step": 164170
    },
    {
      "epoch": 1.0529995221801414,
      "grad_norm": 0.08213865011930466,
      "learning_rate": 5.409690807706108e-06,
      "loss": 0.0017,
      "step": 164180
    },
    {
      "epoch": 1.0530636590739273,
      "grad_norm": 0.0631614625453949,
      "learning_rate": 5.409132983815771e-06,
      "loss": 0.0013,
      "step": 164190
    },
    {
      "epoch": 1.0531277959677134,
      "grad_norm": 0.030576270073652267,
      "learning_rate": 5.408575154798705e-06,
      "loss": 0.0013,
      "step": 164200
    },
    {
      "epoch": 1.0531919328614996,
      "grad_norm": 0.28509676456451416,
      "learning_rate": 5.4080173206619e-06,
      "loss": 0.0029,
      "step": 164210
    },
    {
      "epoch": 1.0532560697552857,
      "grad_norm": 0.11434341222047806,
      "learning_rate": 5.407459481412347e-06,
      "loss": 0.0016,
      "step": 164220
    },
    {
      "epoch": 1.0533202066490719,
      "grad_norm": 0.11937174946069717,
      "learning_rate": 5.406901637057035e-06,
      "loss": 0.0038,
      "step": 164230
    },
    {
      "epoch": 1.0533843435428578,
      "grad_norm": 0.024643611162900925,
      "learning_rate": 5.406343787602955e-06,
      "loss": 0.0009,
      "step": 164240
    },
    {
      "epoch": 1.053448480436644,
      "grad_norm": 0.06495968252420425,
      "learning_rate": 5.405785933057097e-06,
      "loss": 0.0022,
      "step": 164250
    },
    {
      "epoch": 1.0535126173304301,
      "grad_norm": 0.05198515206575394,
      "learning_rate": 5.4052280734264515e-06,
      "loss": 0.0013,
      "step": 164260
    },
    {
      "epoch": 1.0535767542242163,
      "grad_norm": 0.041042618453502655,
      "learning_rate": 5.4046702087180066e-06,
      "loss": 0.0028,
      "step": 164270
    },
    {
      "epoch": 1.0536408911180022,
      "grad_norm": 0.05644712597131729,
      "learning_rate": 5.4041123389387574e-06,
      "loss": 0.0012,
      "step": 164280
    },
    {
      "epoch": 1.0537050280117883,
      "grad_norm": 0.048884227871894836,
      "learning_rate": 5.40355446409569e-06,
      "loss": 0.0012,
      "step": 164290
    },
    {
      "epoch": 1.0537691649055745,
      "grad_norm": 0.05040355026721954,
      "learning_rate": 5.4029965841957985e-06,
      "loss": 0.0023,
      "step": 164300
    },
    {
      "epoch": 1.0538333017993606,
      "grad_norm": 0.06720659881830215,
      "learning_rate": 5.40243869924607e-06,
      "loss": 0.0013,
      "step": 164310
    },
    {
      "epoch": 1.0538974386931466,
      "grad_norm": 0.12022145837545395,
      "learning_rate": 5.401880809253498e-06,
      "loss": 0.0013,
      "step": 164320
    },
    {
      "epoch": 1.0539615755869327,
      "grad_norm": 0.0682666078209877,
      "learning_rate": 5.401322914225071e-06,
      "loss": 0.0016,
      "step": 164330
    },
    {
      "epoch": 1.0540257124807189,
      "grad_norm": 0.07866264879703522,
      "learning_rate": 5.400765014167782e-06,
      "loss": 0.0016,
      "step": 164340
    },
    {
      "epoch": 1.054089849374505,
      "grad_norm": 0.1084682047367096,
      "learning_rate": 5.400207109088622e-06,
      "loss": 0.0021,
      "step": 164350
    },
    {
      "epoch": 1.054153986268291,
      "grad_norm": 0.05660034343600273,
      "learning_rate": 5.399649198994578e-06,
      "loss": 0.0025,
      "step": 164360
    },
    {
      "epoch": 1.054218123162077,
      "grad_norm": 0.07491907477378845,
      "learning_rate": 5.399091283892647e-06,
      "loss": 0.0043,
      "step": 164370
    },
    {
      "epoch": 1.0542822600558632,
      "grad_norm": 0.04296877607703209,
      "learning_rate": 5.398533363789815e-06,
      "loss": 0.0019,
      "step": 164380
    },
    {
      "epoch": 1.0543463969496494,
      "grad_norm": 0.008898581378161907,
      "learning_rate": 5.397975438693077e-06,
      "loss": 0.0015,
      "step": 164390
    },
    {
      "epoch": 1.0544105338434355,
      "grad_norm": 0.05675172060728073,
      "learning_rate": 5.3974175086094205e-06,
      "loss": 0.0022,
      "step": 164400
    },
    {
      "epoch": 1.0544746707372215,
      "grad_norm": 0.023167546838521957,
      "learning_rate": 5.39685957354584e-06,
      "loss": 0.0018,
      "step": 164410
    },
    {
      "epoch": 1.0545388076310076,
      "grad_norm": 0.03619055449962616,
      "learning_rate": 5.396301633509324e-06,
      "loss": 0.0015,
      "step": 164420
    },
    {
      "epoch": 1.0546029445247938,
      "grad_norm": 0.0982537716627121,
      "learning_rate": 5.395743688506867e-06,
      "loss": 0.0014,
      "step": 164430
    },
    {
      "epoch": 1.05466708141858,
      "grad_norm": 0.088741734623909,
      "learning_rate": 5.395185738545456e-06,
      "loss": 0.0019,
      "step": 164440
    },
    {
      "epoch": 1.0547312183123658,
      "grad_norm": 0.0686255544424057,
      "learning_rate": 5.394627783632088e-06,
      "loss": 0.0025,
      "step": 164450
    },
    {
      "epoch": 1.054795355206152,
      "grad_norm": 0.07357365638017654,
      "learning_rate": 5.394069823773751e-06,
      "loss": 0.0023,
      "step": 164460
    },
    {
      "epoch": 1.0548594920999381,
      "grad_norm": 0.08306479454040527,
      "learning_rate": 5.393511858977437e-06,
      "loss": 0.0022,
      "step": 164470
    },
    {
      "epoch": 1.0549236289937243,
      "grad_norm": 0.05139013007283211,
      "learning_rate": 5.392953889250139e-06,
      "loss": 0.0032,
      "step": 164480
    },
    {
      "epoch": 1.0549877658875102,
      "grad_norm": 0.06120975688099861,
      "learning_rate": 5.392395914598847e-06,
      "loss": 0.0011,
      "step": 164490
    },
    {
      "epoch": 1.0550519027812963,
      "grad_norm": 0.047991979867219925,
      "learning_rate": 5.391837935030553e-06,
      "loss": 0.0031,
      "step": 164500
    },
    {
      "epoch": 1.0551160396750825,
      "grad_norm": 0.16589823365211487,
      "learning_rate": 5.391279950552252e-06,
      "loss": 0.0016,
      "step": 164510
    },
    {
      "epoch": 1.0551801765688686,
      "grad_norm": 0.06756390631198883,
      "learning_rate": 5.3907219611709315e-06,
      "loss": 0.0013,
      "step": 164520
    },
    {
      "epoch": 1.0552443134626548,
      "grad_norm": 0.040023867040872574,
      "learning_rate": 5.390163966893587e-06,
      "loss": 0.0054,
      "step": 164530
    },
    {
      "epoch": 1.0553084503564407,
      "grad_norm": 0.1901106983423233,
      "learning_rate": 5.389605967727208e-06,
      "loss": 0.0009,
      "step": 164540
    },
    {
      "epoch": 1.0553725872502269,
      "grad_norm": 0.1705317199230194,
      "learning_rate": 5.389047963678788e-06,
      "loss": 0.0031,
      "step": 164550
    },
    {
      "epoch": 1.055436724144013,
      "grad_norm": 0.07654978334903717,
      "learning_rate": 5.3884899547553195e-06,
      "loss": 0.0013,
      "step": 164560
    },
    {
      "epoch": 1.0555008610377992,
      "grad_norm": 0.04382341727614403,
      "learning_rate": 5.387931940963793e-06,
      "loss": 0.0007,
      "step": 164570
    },
    {
      "epoch": 1.055564997931585,
      "grad_norm": 0.007866988889873028,
      "learning_rate": 5.3873739223112045e-06,
      "loss": 0.0087,
      "step": 164580
    },
    {
      "epoch": 1.0556291348253712,
      "grad_norm": 0.17084579169750214,
      "learning_rate": 5.386815898804541e-06,
      "loss": 0.003,
      "step": 164590
    },
    {
      "epoch": 1.0556932717191574,
      "grad_norm": 0.04956797882914543,
      "learning_rate": 5.3862578704508005e-06,
      "loss": 0.0008,
      "step": 164600
    },
    {
      "epoch": 1.0557574086129435,
      "grad_norm": 0.10130944103002548,
      "learning_rate": 5.385699837256971e-06,
      "loss": 0.0014,
      "step": 164610
    },
    {
      "epoch": 1.0558215455067295,
      "grad_norm": 0.039703045040369034,
      "learning_rate": 5.385141799230048e-06,
      "loss": 0.0015,
      "step": 164620
    },
    {
      "epoch": 1.0558856824005156,
      "grad_norm": 0.03739073872566223,
      "learning_rate": 5.384583756377022e-06,
      "loss": 0.0011,
      "step": 164630
    },
    {
      "epoch": 1.0559498192943018,
      "grad_norm": 0.1988016813993454,
      "learning_rate": 5.384025708704887e-06,
      "loss": 0.0008,
      "step": 164640
    },
    {
      "epoch": 1.056013956188088,
      "grad_norm": 0.5328776836395264,
      "learning_rate": 5.383467656220636e-06,
      "loss": 0.002,
      "step": 164650
    },
    {
      "epoch": 1.056078093081874,
      "grad_norm": 0.19334547221660614,
      "learning_rate": 5.382909598931262e-06,
      "loss": 0.0013,
      "step": 164660
    },
    {
      "epoch": 1.05614222997566,
      "grad_norm": 0.1458764374256134,
      "learning_rate": 5.382351536843755e-06,
      "loss": 0.0016,
      "step": 164670
    },
    {
      "epoch": 1.0562063668694461,
      "grad_norm": 0.05135856196284294,
      "learning_rate": 5.38179346996511e-06,
      "loss": 0.0016,
      "step": 164680
    },
    {
      "epoch": 1.0562705037632323,
      "grad_norm": 0.004306500777602196,
      "learning_rate": 5.381235398302321e-06,
      "loss": 0.0016,
      "step": 164690
    },
    {
      "epoch": 1.0563346406570184,
      "grad_norm": 0.09824187308549881,
      "learning_rate": 5.38067732186238e-06,
      "loss": 0.0019,
      "step": 164700
    },
    {
      "epoch": 1.0563987775508044,
      "grad_norm": 0.10251548886299133,
      "learning_rate": 5.3801192406522805e-06,
      "loss": 0.0016,
      "step": 164710
    },
    {
      "epoch": 1.0564629144445905,
      "grad_norm": 0.07564540952444077,
      "learning_rate": 5.379561154679014e-06,
      "loss": 0.0018,
      "step": 164720
    },
    {
      "epoch": 1.0565270513383767,
      "grad_norm": 0.0791371539235115,
      "learning_rate": 5.379003063949576e-06,
      "loss": 0.0012,
      "step": 164730
    },
    {
      "epoch": 1.0565911882321628,
      "grad_norm": 0.14272333681583405,
      "learning_rate": 5.3784449684709585e-06,
      "loss": 0.0023,
      "step": 164740
    },
    {
      "epoch": 1.0566553251259487,
      "grad_norm": 0.22609438002109528,
      "learning_rate": 5.377886868250155e-06,
      "loss": 0.003,
      "step": 164750
    },
    {
      "epoch": 1.0567194620197349,
      "grad_norm": 0.1458098441362381,
      "learning_rate": 5.377328763294158e-06,
      "loss": 0.0011,
      "step": 164760
    },
    {
      "epoch": 1.056783598913521,
      "grad_norm": 0.10944253951311111,
      "learning_rate": 5.376770653609964e-06,
      "loss": 0.0022,
      "step": 164770
    },
    {
      "epoch": 1.0568477358073072,
      "grad_norm": 0.07059153914451599,
      "learning_rate": 5.376212539204563e-06,
      "loss": 0.0032,
      "step": 164780
    },
    {
      "epoch": 1.056911872701093,
      "grad_norm": 0.010585133917629719,
      "learning_rate": 5.37565442008495e-06,
      "loss": 0.0018,
      "step": 164790
    },
    {
      "epoch": 1.0569760095948793,
      "grad_norm": 0.34683093428611755,
      "learning_rate": 5.375096296258119e-06,
      "loss": 0.0022,
      "step": 164800
    },
    {
      "epoch": 1.0570401464886654,
      "grad_norm": 0.3620857894420624,
      "learning_rate": 5.374538167731064e-06,
      "loss": 0.0012,
      "step": 164810
    },
    {
      "epoch": 1.0571042833824515,
      "grad_norm": 0.01606067828834057,
      "learning_rate": 5.373980034510776e-06,
      "loss": 0.0039,
      "step": 164820
    },
    {
      "epoch": 1.0571684202762377,
      "grad_norm": 0.14603322744369507,
      "learning_rate": 5.373421896604252e-06,
      "loss": 0.0015,
      "step": 164830
    },
    {
      "epoch": 1.0572325571700236,
      "grad_norm": 0.2772385776042938,
      "learning_rate": 5.372863754018486e-06,
      "loss": 0.0029,
      "step": 164840
    },
    {
      "epoch": 1.0572966940638098,
      "grad_norm": 0.0029333140701055527,
      "learning_rate": 5.372305606760468e-06,
      "loss": 0.0012,
      "step": 164850
    },
    {
      "epoch": 1.057360830957596,
      "grad_norm": 0.020547224208712578,
      "learning_rate": 5.371747454837197e-06,
      "loss": 0.0013,
      "step": 164860
    },
    {
      "epoch": 1.057424967851382,
      "grad_norm": 0.010254314169287682,
      "learning_rate": 5.371189298255664e-06,
      "loss": 0.0013,
      "step": 164870
    },
    {
      "epoch": 1.057489104745168,
      "grad_norm": 0.04542948305606842,
      "learning_rate": 5.370631137022864e-06,
      "loss": 0.0015,
      "step": 164880
    },
    {
      "epoch": 1.0575532416389541,
      "grad_norm": 0.08468092978000641,
      "learning_rate": 5.370072971145789e-06,
      "loss": 0.0013,
      "step": 164890
    },
    {
      "epoch": 1.0576173785327403,
      "grad_norm": 0.13973210752010345,
      "learning_rate": 5.369514800631438e-06,
      "loss": 0.002,
      "step": 164900
    },
    {
      "epoch": 1.0576815154265264,
      "grad_norm": 0.056624799966812134,
      "learning_rate": 5.368956625486801e-06,
      "loss": 0.002,
      "step": 164910
    },
    {
      "epoch": 1.0577456523203124,
      "grad_norm": 0.003999556880444288,
      "learning_rate": 5.368398445718875e-06,
      "loss": 0.0012,
      "step": 164920
    },
    {
      "epoch": 1.0578097892140985,
      "grad_norm": 0.016761040315032005,
      "learning_rate": 5.367840261334652e-06,
      "loss": 0.0016,
      "step": 164930
    },
    {
      "epoch": 1.0578739261078847,
      "grad_norm": 0.1763841211795807,
      "learning_rate": 5.367282072341128e-06,
      "loss": 0.0022,
      "step": 164940
    },
    {
      "epoch": 1.0579380630016708,
      "grad_norm": 0.13624563813209534,
      "learning_rate": 5.366723878745297e-06,
      "loss": 0.0017,
      "step": 164950
    },
    {
      "epoch": 1.058002199895457,
      "grad_norm": 0.06088218465447426,
      "learning_rate": 5.366165680554154e-06,
      "loss": 0.0032,
      "step": 164960
    },
    {
      "epoch": 1.058066336789243,
      "grad_norm": 0.046900175511837006,
      "learning_rate": 5.365607477774693e-06,
      "loss": 0.0014,
      "step": 164970
    },
    {
      "epoch": 1.058130473683029,
      "grad_norm": 0.1951567828655243,
      "learning_rate": 5.36504927041391e-06,
      "loss": 0.0028,
      "step": 164980
    },
    {
      "epoch": 1.0581946105768152,
      "grad_norm": 0.11414501070976257,
      "learning_rate": 5.364491058478797e-06,
      "loss": 0.0015,
      "step": 164990
    },
    {
      "epoch": 1.0582587474706013,
      "grad_norm": 0.13751763105392456,
      "learning_rate": 5.3639328419763525e-06,
      "loss": 0.0012,
      "step": 165000
    },
    {
      "epoch": 1.0583228843643873,
      "grad_norm": 0.0704054906964302,
      "learning_rate": 5.363374620913567e-06,
      "loss": 0.0038,
      "step": 165010
    },
    {
      "epoch": 1.0583870212581734,
      "grad_norm": 0.015720652416348457,
      "learning_rate": 5.36281639529744e-06,
      "loss": 0.0016,
      "step": 165020
    },
    {
      "epoch": 1.0584511581519596,
      "grad_norm": 0.09596427530050278,
      "learning_rate": 5.362258165134964e-06,
      "loss": 0.0021,
      "step": 165030
    },
    {
      "epoch": 1.0585152950457457,
      "grad_norm": 0.07461073249578476,
      "learning_rate": 5.361699930433133e-06,
      "loss": 0.0021,
      "step": 165040
    },
    {
      "epoch": 1.0585794319395316,
      "grad_norm": 0.003062749747186899,
      "learning_rate": 5.361141691198944e-06,
      "loss": 0.0006,
      "step": 165050
    },
    {
      "epoch": 1.0586435688333178,
      "grad_norm": 0.04284300282597542,
      "learning_rate": 5.360583447439391e-06,
      "loss": 0.0018,
      "step": 165060
    },
    {
      "epoch": 1.058707705727104,
      "grad_norm": 0.4961910545825958,
      "learning_rate": 5.3600251991614715e-06,
      "loss": 0.0022,
      "step": 165070
    },
    {
      "epoch": 1.05877184262089,
      "grad_norm": 0.04706254601478577,
      "learning_rate": 5.359466946372177e-06,
      "loss": 0.0013,
      "step": 165080
    },
    {
      "epoch": 1.0588359795146762,
      "grad_norm": 0.07458434998989105,
      "learning_rate": 5.358908689078506e-06,
      "loss": 0.0017,
      "step": 165090
    },
    {
      "epoch": 1.0589001164084622,
      "grad_norm": 0.11897841095924377,
      "learning_rate": 5.358350427287452e-06,
      "loss": 0.0021,
      "step": 165100
    },
    {
      "epoch": 1.0589642533022483,
      "grad_norm": 0.09279097616672516,
      "learning_rate": 5.357792161006011e-06,
      "loss": 0.0019,
      "step": 165110
    },
    {
      "epoch": 1.0590283901960345,
      "grad_norm": 0.35574957728385925,
      "learning_rate": 5.357233890241178e-06,
      "loss": 0.0023,
      "step": 165120
    },
    {
      "epoch": 1.0590925270898206,
      "grad_norm": 0.021752241998910904,
      "learning_rate": 5.356675614999951e-06,
      "loss": 0.0019,
      "step": 165130
    },
    {
      "epoch": 1.0591566639836065,
      "grad_norm": 0.10023855417966843,
      "learning_rate": 5.356117335289321e-06,
      "loss": 0.0034,
      "step": 165140
    },
    {
      "epoch": 1.0592208008773927,
      "grad_norm": 0.07290735840797424,
      "learning_rate": 5.355559051116289e-06,
      "loss": 0.0017,
      "step": 165150
    },
    {
      "epoch": 1.0592849377711788,
      "grad_norm": 0.04673566296696663,
      "learning_rate": 5.3550007624878465e-06,
      "loss": 0.0019,
      "step": 165160
    },
    {
      "epoch": 1.059349074664965,
      "grad_norm": 0.06728404015302658,
      "learning_rate": 5.354442469410991e-06,
      "loss": 0.0016,
      "step": 165170
    },
    {
      "epoch": 1.059413211558751,
      "grad_norm": 0.05923718586564064,
      "learning_rate": 5.353884171892717e-06,
      "loss": 0.0015,
      "step": 165180
    },
    {
      "epoch": 1.059477348452537,
      "grad_norm": 0.037422552704811096,
      "learning_rate": 5.3533258699400225e-06,
      "loss": 0.0014,
      "step": 165190
    },
    {
      "epoch": 1.0595414853463232,
      "grad_norm": 0.0014998482074588537,
      "learning_rate": 5.352767563559902e-06,
      "loss": 0.0022,
      "step": 165200
    },
    {
      "epoch": 1.0596056222401093,
      "grad_norm": 0.03718739375472069,
      "learning_rate": 5.352209252759352e-06,
      "loss": 0.001,
      "step": 165210
    },
    {
      "epoch": 1.0596697591338953,
      "grad_norm": 0.1297600120306015,
      "learning_rate": 5.351650937545368e-06,
      "loss": 0.0022,
      "step": 165220
    },
    {
      "epoch": 1.0597338960276814,
      "grad_norm": 0.07515675574541092,
      "learning_rate": 5.3510926179249465e-06,
      "loss": 0.0026,
      "step": 165230
    },
    {
      "epoch": 1.0597980329214676,
      "grad_norm": 0.02211691625416279,
      "learning_rate": 5.350534293905084e-06,
      "loss": 0.002,
      "step": 165240
    },
    {
      "epoch": 1.0598621698152537,
      "grad_norm": 0.06225384399294853,
      "learning_rate": 5.349975965492776e-06,
      "loss": 0.0021,
      "step": 165250
    },
    {
      "epoch": 1.0599263067090399,
      "grad_norm": 0.12936356663703918,
      "learning_rate": 5.34941763269502e-06,
      "loss": 0.0014,
      "step": 165260
    },
    {
      "epoch": 1.0599904436028258,
      "grad_norm": 0.0758969858288765,
      "learning_rate": 5.348859295518809e-06,
      "loss": 0.0023,
      "step": 165270
    },
    {
      "epoch": 1.060054580496612,
      "grad_norm": 0.14664317667484283,
      "learning_rate": 5.348300953971144e-06,
      "loss": 0.001,
      "step": 165280
    },
    {
      "epoch": 1.060118717390398,
      "grad_norm": 0.03142325207591057,
      "learning_rate": 5.347742608059017e-06,
      "loss": 0.0016,
      "step": 165290
    },
    {
      "epoch": 1.0601828542841842,
      "grad_norm": 0.01865263469517231,
      "learning_rate": 5.347184257789428e-06,
      "loss": 0.0033,
      "step": 165300
    },
    {
      "epoch": 1.0602469911779702,
      "grad_norm": 0.12045206874608994,
      "learning_rate": 5.346625903169372e-06,
      "loss": 0.0024,
      "step": 165310
    },
    {
      "epoch": 1.0603111280717563,
      "grad_norm": 0.15061624348163605,
      "learning_rate": 5.346067544205846e-06,
      "loss": 0.003,
      "step": 165320
    },
    {
      "epoch": 1.0603752649655425,
      "grad_norm": 0.23420719802379608,
      "learning_rate": 5.3455091809058456e-06,
      "loss": 0.0021,
      "step": 165330
    },
    {
      "epoch": 1.0604394018593286,
      "grad_norm": 0.10741454362869263,
      "learning_rate": 5.34495081327637e-06,
      "loss": 0.0013,
      "step": 165340
    },
    {
      "epoch": 1.0605035387531148,
      "grad_norm": 0.03739853575825691,
      "learning_rate": 5.344392441324412e-06,
      "loss": 0.0006,
      "step": 165350
    },
    {
      "epoch": 1.0605676756469007,
      "grad_norm": 0.25066447257995605,
      "learning_rate": 5.343834065056972e-06,
      "loss": 0.0024,
      "step": 165360
    },
    {
      "epoch": 1.0606318125406868,
      "grad_norm": 0.08378434181213379,
      "learning_rate": 5.343275684481044e-06,
      "loss": 0.0011,
      "step": 165370
    },
    {
      "epoch": 1.060695949434473,
      "grad_norm": 0.02976437658071518,
      "learning_rate": 5.342717299603628e-06,
      "loss": 0.0031,
      "step": 165380
    },
    {
      "epoch": 1.0607600863282591,
      "grad_norm": 0.0805118978023529,
      "learning_rate": 5.342158910431717e-06,
      "loss": 0.0017,
      "step": 165390
    },
    {
      "epoch": 1.060824223222045,
      "grad_norm": 0.05659005045890808,
      "learning_rate": 5.341600516972312e-06,
      "loss": 0.002,
      "step": 165400
    },
    {
      "epoch": 1.0608883601158312,
      "grad_norm": 0.0376061350107193,
      "learning_rate": 5.341042119232409e-06,
      "loss": 0.0014,
      "step": 165410
    },
    {
      "epoch": 1.0609524970096174,
      "grad_norm": 0.04787229374051094,
      "learning_rate": 5.340483717219003e-06,
      "loss": 0.0016,
      "step": 165420
    },
    {
      "epoch": 1.0610166339034035,
      "grad_norm": 0.061529211699962616,
      "learning_rate": 5.339925310939094e-06,
      "loss": 0.0015,
      "step": 165430
    },
    {
      "epoch": 1.0610807707971894,
      "grad_norm": 0.06910432875156403,
      "learning_rate": 5.339366900399677e-06,
      "loss": 0.001,
      "step": 165440
    },
    {
      "epoch": 1.0611449076909756,
      "grad_norm": 0.041936878114938736,
      "learning_rate": 5.33880848560775e-06,
      "loss": 0.0019,
      "step": 165450
    },
    {
      "epoch": 1.0612090445847617,
      "grad_norm": 0.021885309368371964,
      "learning_rate": 5.338250066570311e-06,
      "loss": 0.002,
      "step": 165460
    },
    {
      "epoch": 1.0612731814785479,
      "grad_norm": 0.1264055222272873,
      "learning_rate": 5.337691643294357e-06,
      "loss": 0.002,
      "step": 165470
    },
    {
      "epoch": 1.0613373183723338,
      "grad_norm": 0.12011872977018356,
      "learning_rate": 5.337133215786885e-06,
      "loss": 0.0015,
      "step": 165480
    },
    {
      "epoch": 1.06140145526612,
      "grad_norm": 0.16393841803073883,
      "learning_rate": 5.336574784054894e-06,
      "loss": 0.0009,
      "step": 165490
    },
    {
      "epoch": 1.061465592159906,
      "grad_norm": 0.0875626727938652,
      "learning_rate": 5.336016348105379e-06,
      "loss": 0.0014,
      "step": 165500
    },
    {
      "epoch": 1.0615297290536923,
      "grad_norm": 0.10251889377832413,
      "learning_rate": 5.33545790794534e-06,
      "loss": 0.002,
      "step": 165510
    },
    {
      "epoch": 1.0615938659474784,
      "grad_norm": 0.04495257884263992,
      "learning_rate": 5.334899463581773e-06,
      "loss": 0.0017,
      "step": 165520
    },
    {
      "epoch": 1.0616580028412643,
      "grad_norm": 0.11217251420021057,
      "learning_rate": 5.334341015021676e-06,
      "loss": 0.0016,
      "step": 165530
    },
    {
      "epoch": 1.0617221397350505,
      "grad_norm": 0.002483774209395051,
      "learning_rate": 5.333782562272049e-06,
      "loss": 0.0008,
      "step": 165540
    },
    {
      "epoch": 1.0617862766288366,
      "grad_norm": 0.18159668147563934,
      "learning_rate": 5.333224105339884e-06,
      "loss": 0.0018,
      "step": 165550
    },
    {
      "epoch": 1.0618504135226228,
      "grad_norm": 0.29747647047042847,
      "learning_rate": 5.332665644232188e-06,
      "loss": 0.0025,
      "step": 165560
    },
    {
      "epoch": 1.0619145504164087,
      "grad_norm": 0.039432551711797714,
      "learning_rate": 5.33210717895595e-06,
      "loss": 0.0024,
      "step": 165570
    },
    {
      "epoch": 1.0619786873101948,
      "grad_norm": 0.057367291301488876,
      "learning_rate": 5.331548709518174e-06,
      "loss": 0.0022,
      "step": 165580
    },
    {
      "epoch": 1.062042824203981,
      "grad_norm": 0.11943091452121735,
      "learning_rate": 5.330990235925853e-06,
      "loss": 0.0011,
      "step": 165590
    },
    {
      "epoch": 1.0621069610977671,
      "grad_norm": 0.006093861069530249,
      "learning_rate": 5.330431758185991e-06,
      "loss": 0.0015,
      "step": 165600
    },
    {
      "epoch": 1.062171097991553,
      "grad_norm": 0.11781761795282364,
      "learning_rate": 5.32987327630558e-06,
      "loss": 0.0013,
      "step": 165610
    },
    {
      "epoch": 1.0622352348853392,
      "grad_norm": 0.21625033020973206,
      "learning_rate": 5.329314790291623e-06,
      "loss": 0.0028,
      "step": 165620
    },
    {
      "epoch": 1.0622993717791254,
      "grad_norm": 0.09720658510923386,
      "learning_rate": 5.328756300151116e-06,
      "loss": 0.0014,
      "step": 165630
    },
    {
      "epoch": 1.0623635086729115,
      "grad_norm": 0.078163743019104,
      "learning_rate": 5.328197805891058e-06,
      "loss": 0.0033,
      "step": 165640
    },
    {
      "epoch": 1.0624276455666977,
      "grad_norm": 0.09998316317796707,
      "learning_rate": 5.327639307518444e-06,
      "loss": 0.0024,
      "step": 165650
    },
    {
      "epoch": 1.0624917824604836,
      "grad_norm": 0.14844781160354614,
      "learning_rate": 5.327080805040278e-06,
      "loss": 0.0019,
      "step": 165660
    },
    {
      "epoch": 1.0625559193542697,
      "grad_norm": 0.09575474262237549,
      "learning_rate": 5.326522298463556e-06,
      "loss": 0.0029,
      "step": 165670
    },
    {
      "epoch": 1.0626200562480559,
      "grad_norm": 0.06723390519618988,
      "learning_rate": 5.325963787795275e-06,
      "loss": 0.0008,
      "step": 165680
    },
    {
      "epoch": 1.062684193141842,
      "grad_norm": 0.014396176673471928,
      "learning_rate": 5.325405273042435e-06,
      "loss": 0.0008,
      "step": 165690
    },
    {
      "epoch": 1.062748330035628,
      "grad_norm": 0.05677991360425949,
      "learning_rate": 5.3248467542120354e-06,
      "loss": 0.0015,
      "step": 165700
    },
    {
      "epoch": 1.0628124669294141,
      "grad_norm": 0.05875563621520996,
      "learning_rate": 5.324288231311072e-06,
      "loss": 0.0007,
      "step": 165710
    },
    {
      "epoch": 1.0628766038232003,
      "grad_norm": 0.26707419753074646,
      "learning_rate": 5.323729704346547e-06,
      "loss": 0.0022,
      "step": 165720
    },
    {
      "epoch": 1.0629407407169864,
      "grad_norm": 0.06900002807378769,
      "learning_rate": 5.323171173325457e-06,
      "loss": 0.0014,
      "step": 165730
    },
    {
      "epoch": 1.0630048776107723,
      "grad_norm": 0.0487680621445179,
      "learning_rate": 5.3226126382548e-06,
      "loss": 0.0018,
      "step": 165740
    },
    {
      "epoch": 1.0630690145045585,
      "grad_norm": 0.08753561228513718,
      "learning_rate": 5.322054099141578e-06,
      "loss": 0.0033,
      "step": 165750
    },
    {
      "epoch": 1.0631331513983446,
      "grad_norm": 0.04199390485882759,
      "learning_rate": 5.321495555992787e-06,
      "loss": 0.0016,
      "step": 165760
    },
    {
      "epoch": 1.0631972882921308,
      "grad_norm": 0.17838846147060394,
      "learning_rate": 5.320937008815427e-06,
      "loss": 0.0015,
      "step": 165770
    },
    {
      "epoch": 1.063261425185917,
      "grad_norm": 0.09121342748403549,
      "learning_rate": 5.320378457616498e-06,
      "loss": 0.0007,
      "step": 165780
    },
    {
      "epoch": 1.0633255620797029,
      "grad_norm": 0.10484384745359421,
      "learning_rate": 5.319819902402996e-06,
      "loss": 0.0013,
      "step": 165790
    },
    {
      "epoch": 1.063389698973489,
      "grad_norm": 0.1039414256811142,
      "learning_rate": 5.319261343181923e-06,
      "loss": 0.0019,
      "step": 165800
    },
    {
      "epoch": 1.0634538358672752,
      "grad_norm": 0.05744036287069321,
      "learning_rate": 5.3187027799602775e-06,
      "loss": 0.0027,
      "step": 165810
    },
    {
      "epoch": 1.0635179727610613,
      "grad_norm": 0.018517503514885902,
      "learning_rate": 5.318144212745058e-06,
      "loss": 0.002,
      "step": 165820
    },
    {
      "epoch": 1.0635821096548472,
      "grad_norm": 0.12544725835323334,
      "learning_rate": 5.3175856415432645e-06,
      "loss": 0.0026,
      "step": 165830
    },
    {
      "epoch": 1.0636462465486334,
      "grad_norm": 0.07395651936531067,
      "learning_rate": 5.317027066361895e-06,
      "loss": 0.0021,
      "step": 165840
    },
    {
      "epoch": 1.0637103834424195,
      "grad_norm": 0.07024452090263367,
      "learning_rate": 5.316468487207951e-06,
      "loss": 0.0034,
      "step": 165850
    },
    {
      "epoch": 1.0637745203362057,
      "grad_norm": 0.10968262702226639,
      "learning_rate": 5.31590990408843e-06,
      "loss": 0.002,
      "step": 165860
    },
    {
      "epoch": 1.0638386572299916,
      "grad_norm": 0.11000864207744598,
      "learning_rate": 5.315351317010332e-06,
      "loss": 0.002,
      "step": 165870
    },
    {
      "epoch": 1.0639027941237778,
      "grad_norm": 0.015025916509330273,
      "learning_rate": 5.314792725980657e-06,
      "loss": 0.0012,
      "step": 165880
    },
    {
      "epoch": 1.063966931017564,
      "grad_norm": 0.0507902167737484,
      "learning_rate": 5.314234131006403e-06,
      "loss": 0.0024,
      "step": 165890
    },
    {
      "epoch": 1.06403106791135,
      "grad_norm": 0.09992703795433044,
      "learning_rate": 5.313675532094572e-06,
      "loss": 0.0015,
      "step": 165900
    },
    {
      "epoch": 1.064095204805136,
      "grad_norm": 0.07698260992765427,
      "learning_rate": 5.3131169292521625e-06,
      "loss": 0.0016,
      "step": 165910
    },
    {
      "epoch": 1.0641593416989221,
      "grad_norm": 0.04310933127999306,
      "learning_rate": 5.312558322486174e-06,
      "loss": 0.0011,
      "step": 165920
    },
    {
      "epoch": 1.0642234785927083,
      "grad_norm": 0.22994256019592285,
      "learning_rate": 5.311999711803605e-06,
      "loss": 0.0014,
      "step": 165930
    },
    {
      "epoch": 1.0642876154864944,
      "grad_norm": 0.15693657100200653,
      "learning_rate": 5.311441097211458e-06,
      "loss": 0.0013,
      "step": 165940
    },
    {
      "epoch": 1.0643517523802806,
      "grad_norm": 0.07755468785762787,
      "learning_rate": 5.310882478716731e-06,
      "loss": 0.0014,
      "step": 165950
    },
    {
      "epoch": 1.0644158892740665,
      "grad_norm": 0.05102725327014923,
      "learning_rate": 5.310323856326426e-06,
      "loss": 0.0022,
      "step": 165960
    },
    {
      "epoch": 1.0644800261678526,
      "grad_norm": 0.39134690165519714,
      "learning_rate": 5.309765230047539e-06,
      "loss": 0.0016,
      "step": 165970
    },
    {
      "epoch": 1.0645441630616388,
      "grad_norm": 0.060010071843862534,
      "learning_rate": 5.309206599887074e-06,
      "loss": 0.0017,
      "step": 165980
    },
    {
      "epoch": 1.064608299955425,
      "grad_norm": 0.03553006052970886,
      "learning_rate": 5.308647965852028e-06,
      "loss": 0.0022,
      "step": 165990
    },
    {
      "epoch": 1.0646724368492109,
      "grad_norm": 0.05214393138885498,
      "learning_rate": 5.308089327949403e-06,
      "loss": 0.0019,
      "step": 166000
    },
    {
      "epoch": 1.064736573742997,
      "grad_norm": 0.05157877504825592,
      "learning_rate": 5.3075306861861975e-06,
      "loss": 0.0015,
      "step": 166010
    },
    {
      "epoch": 1.0648007106367832,
      "grad_norm": 0.14204972982406616,
      "learning_rate": 5.306972040569415e-06,
      "loss": 0.003,
      "step": 166020
    },
    {
      "epoch": 1.0648648475305693,
      "grad_norm": 0.05269039794802666,
      "learning_rate": 5.306413391106051e-06,
      "loss": 0.0015,
      "step": 166030
    },
    {
      "epoch": 1.0649289844243552,
      "grad_norm": 0.08166810870170593,
      "learning_rate": 5.305854737803109e-06,
      "loss": 0.0012,
      "step": 166040
    },
    {
      "epoch": 1.0649931213181414,
      "grad_norm": 0.14019718766212463,
      "learning_rate": 5.305296080667588e-06,
      "loss": 0.0009,
      "step": 166050
    },
    {
      "epoch": 1.0650572582119275,
      "grad_norm": 0.12255555391311646,
      "learning_rate": 5.30473741970649e-06,
      "loss": 0.0011,
      "step": 166060
    },
    {
      "epoch": 1.0651213951057137,
      "grad_norm": 0.06505221873521805,
      "learning_rate": 5.304178754926813e-06,
      "loss": 0.0023,
      "step": 166070
    },
    {
      "epoch": 1.0651855319994998,
      "grad_norm": 0.00911052618175745,
      "learning_rate": 5.303620086335561e-06,
      "loss": 0.0025,
      "step": 166080
    },
    {
      "epoch": 1.0652496688932858,
      "grad_norm": 0.12398932874202728,
      "learning_rate": 5.30306141393973e-06,
      "loss": 0.0032,
      "step": 166090
    },
    {
      "epoch": 1.065313805787072,
      "grad_norm": 0.14058513939380646,
      "learning_rate": 5.302502737746325e-06,
      "loss": 0.0012,
      "step": 166100
    },
    {
      "epoch": 1.065377942680858,
      "grad_norm": 0.15345489978790283,
      "learning_rate": 5.301944057762344e-06,
      "loss": 0.0027,
      "step": 166110
    },
    {
      "epoch": 1.0654420795746442,
      "grad_norm": 0.07064566761255264,
      "learning_rate": 5.301385373994786e-06,
      "loss": 0.0019,
      "step": 166120
    },
    {
      "epoch": 1.0655062164684301,
      "grad_norm": 0.1936577558517456,
      "learning_rate": 5.300826686450656e-06,
      "loss": 0.0024,
      "step": 166130
    },
    {
      "epoch": 1.0655703533622163,
      "grad_norm": 0.1288345903158188,
      "learning_rate": 5.300267995136953e-06,
      "loss": 0.0016,
      "step": 166140
    },
    {
      "epoch": 1.0656344902560024,
      "grad_norm": 0.19684065878391266,
      "learning_rate": 5.299709300060677e-06,
      "loss": 0.002,
      "step": 166150
    },
    {
      "epoch": 1.0656986271497886,
      "grad_norm": 0.007303610909730196,
      "learning_rate": 5.299150601228828e-06,
      "loss": 0.0039,
      "step": 166160
    },
    {
      "epoch": 1.0657627640435745,
      "grad_norm": 0.08438844233751297,
      "learning_rate": 5.29859189864841e-06,
      "loss": 0.0015,
      "step": 166170
    },
    {
      "epoch": 1.0658269009373607,
      "grad_norm": 0.09508020430803299,
      "learning_rate": 5.29803319232642e-06,
      "loss": 0.0016,
      "step": 166180
    },
    {
      "epoch": 1.0658910378311468,
      "grad_norm": 0.016180196776986122,
      "learning_rate": 5.297474482269863e-06,
      "loss": 0.0014,
      "step": 166190
    },
    {
      "epoch": 1.065955174724933,
      "grad_norm": 0.2816241681575775,
      "learning_rate": 5.2969157684857375e-06,
      "loss": 0.0031,
      "step": 166200
    },
    {
      "epoch": 1.066019311618719,
      "grad_norm": 0.0024928152561187744,
      "learning_rate": 5.296357050981046e-06,
      "loss": 0.0031,
      "step": 166210
    },
    {
      "epoch": 1.066083448512505,
      "grad_norm": 0.06070615351200104,
      "learning_rate": 5.2957983297627866e-06,
      "loss": 0.0022,
      "step": 166220
    },
    {
      "epoch": 1.0661475854062912,
      "grad_norm": 0.021533237770199776,
      "learning_rate": 5.295239604837965e-06,
      "loss": 0.001,
      "step": 166230
    },
    {
      "epoch": 1.0662117223000773,
      "grad_norm": 0.43801963329315186,
      "learning_rate": 5.2946808762135785e-06,
      "loss": 0.0025,
      "step": 166240
    },
    {
      "epoch": 1.0662758591938635,
      "grad_norm": 0.10720543563365936,
      "learning_rate": 5.29412214389663e-06,
      "loss": 0.002,
      "step": 166250
    },
    {
      "epoch": 1.0663399960876494,
      "grad_norm": 0.03996255621314049,
      "learning_rate": 5.293563407894122e-06,
      "loss": 0.0023,
      "step": 166260
    },
    {
      "epoch": 1.0664041329814355,
      "grad_norm": 0.10009384155273438,
      "learning_rate": 5.293004668213055e-06,
      "loss": 0.0028,
      "step": 166270
    },
    {
      "epoch": 1.0664682698752217,
      "grad_norm": 0.16202972829341888,
      "learning_rate": 5.2924459248604284e-06,
      "loss": 0.0013,
      "step": 166280
    },
    {
      "epoch": 1.0665324067690078,
      "grad_norm": 0.09987245500087738,
      "learning_rate": 5.291887177843247e-06,
      "loss": 0.0026,
      "step": 166290
    },
    {
      "epoch": 1.0665965436627938,
      "grad_norm": 0.14244654774665833,
      "learning_rate": 5.291328427168511e-06,
      "loss": 0.0021,
      "step": 166300
    },
    {
      "epoch": 1.06666068055658,
      "grad_norm": 0.1552550196647644,
      "learning_rate": 5.290769672843219e-06,
      "loss": 0.0019,
      "step": 166310
    },
    {
      "epoch": 1.066724817450366,
      "grad_norm": 0.06859882175922394,
      "learning_rate": 5.290210914874377e-06,
      "loss": 0.0013,
      "step": 166320
    },
    {
      "epoch": 1.0667889543441522,
      "grad_norm": 0.01697307638823986,
      "learning_rate": 5.2896521532689845e-06,
      "loss": 0.0023,
      "step": 166330
    },
    {
      "epoch": 1.0668530912379381,
      "grad_norm": 0.12605133652687073,
      "learning_rate": 5.289093388034045e-06,
      "loss": 0.0013,
      "step": 166340
    },
    {
      "epoch": 1.0669172281317243,
      "grad_norm": 0.04196297377347946,
      "learning_rate": 5.288534619176556e-06,
      "loss": 0.0017,
      "step": 166350
    },
    {
      "epoch": 1.0669813650255104,
      "grad_norm": 0.04504567012190819,
      "learning_rate": 5.287975846703525e-06,
      "loss": 0.0013,
      "step": 166360
    },
    {
      "epoch": 1.0670455019192966,
      "grad_norm": 0.059784576296806335,
      "learning_rate": 5.287417070621947e-06,
      "loss": 0.0019,
      "step": 166370
    },
    {
      "epoch": 1.0671096388130827,
      "grad_norm": 0.026613537222146988,
      "learning_rate": 5.2868582909388296e-06,
      "loss": 0.0012,
      "step": 166380
    },
    {
      "epoch": 1.0671737757068687,
      "grad_norm": 0.25359833240509033,
      "learning_rate": 5.286299507661174e-06,
      "loss": 0.0017,
      "step": 166390
    },
    {
      "epoch": 1.0672379126006548,
      "grad_norm": 0.0881662517786026,
      "learning_rate": 5.28574072079598e-06,
      "loss": 0.0023,
      "step": 166400
    },
    {
      "epoch": 1.067302049494441,
      "grad_norm": 0.05822227522730827,
      "learning_rate": 5.285181930350251e-06,
      "loss": 0.0012,
      "step": 166410
    },
    {
      "epoch": 1.0673661863882271,
      "grad_norm": 0.05762477219104767,
      "learning_rate": 5.284623136330988e-06,
      "loss": 0.0015,
      "step": 166420
    },
    {
      "epoch": 1.067430323282013,
      "grad_norm": 0.20448440313339233,
      "learning_rate": 5.2840643387451924e-06,
      "loss": 0.0024,
      "step": 166430
    },
    {
      "epoch": 1.0674944601757992,
      "grad_norm": 0.1820959895849228,
      "learning_rate": 5.28350553759987e-06,
      "loss": 0.0025,
      "step": 166440
    },
    {
      "epoch": 1.0675585970695853,
      "grad_norm": 0.14897200465202332,
      "learning_rate": 5.282946732902019e-06,
      "loss": 0.0033,
      "step": 166450
    },
    {
      "epoch": 1.0676227339633715,
      "grad_norm": 0.013246786780655384,
      "learning_rate": 5.282387924658644e-06,
      "loss": 0.0024,
      "step": 166460
    },
    {
      "epoch": 1.0676868708571576,
      "grad_norm": 0.017587296664714813,
      "learning_rate": 5.2818291128767466e-06,
      "loss": 0.0014,
      "step": 166470
    },
    {
      "epoch": 1.0677510077509436,
      "grad_norm": 0.03909731283783913,
      "learning_rate": 5.281270297563329e-06,
      "loss": 0.0017,
      "step": 166480
    },
    {
      "epoch": 1.0678151446447297,
      "grad_norm": 0.03878255560994148,
      "learning_rate": 5.2807114787253935e-06,
      "loss": 0.0017,
      "step": 166490
    },
    {
      "epoch": 1.0678792815385159,
      "grad_norm": 0.04676719754934311,
      "learning_rate": 5.280152656369941e-06,
      "loss": 0.0013,
      "step": 166500
    },
    {
      "epoch": 1.067943418432302,
      "grad_norm": 0.038249071687459946,
      "learning_rate": 5.279593830503979e-06,
      "loss": 0.002,
      "step": 166510
    },
    {
      "epoch": 1.068007555326088,
      "grad_norm": 0.053986068814992905,
      "learning_rate": 5.279035001134503e-06,
      "loss": 0.0016,
      "step": 166520
    },
    {
      "epoch": 1.068071692219874,
      "grad_norm": 0.11400827020406723,
      "learning_rate": 5.278476168268522e-06,
      "loss": 0.0012,
      "step": 166530
    },
    {
      "epoch": 1.0681358291136602,
      "grad_norm": 0.03897121176123619,
      "learning_rate": 5.277917331913033e-06,
      "loss": 0.001,
      "step": 166540
    },
    {
      "epoch": 1.0681999660074464,
      "grad_norm": 0.1743035912513733,
      "learning_rate": 5.277358492075042e-06,
      "loss": 0.0016,
      "step": 166550
    },
    {
      "epoch": 1.0682641029012323,
      "grad_norm": 0.04287755861878395,
      "learning_rate": 5.27679964876155e-06,
      "loss": 0.0009,
      "step": 166560
    },
    {
      "epoch": 1.0683282397950185,
      "grad_norm": 0.1285584568977356,
      "learning_rate": 5.276240801979563e-06,
      "loss": 0.005,
      "step": 166570
    },
    {
      "epoch": 1.0683923766888046,
      "grad_norm": 0.021582355722784996,
      "learning_rate": 5.275681951736079e-06,
      "loss": 0.0016,
      "step": 166580
    },
    {
      "epoch": 1.0684565135825908,
      "grad_norm": 0.2306017279624939,
      "learning_rate": 5.275123098038104e-06,
      "loss": 0.0027,
      "step": 166590
    },
    {
      "epoch": 1.0685206504763767,
      "grad_norm": 0.05208316445350647,
      "learning_rate": 5.27456424089264e-06,
      "loss": 0.0009,
      "step": 166600
    },
    {
      "epoch": 1.0685847873701628,
      "grad_norm": 0.3376952111721039,
      "learning_rate": 5.274005380306689e-06,
      "loss": 0.0029,
      "step": 166610
    },
    {
      "epoch": 1.068648924263949,
      "grad_norm": 0.058090854436159134,
      "learning_rate": 5.273446516287258e-06,
      "loss": 0.0019,
      "step": 166620
    },
    {
      "epoch": 1.0687130611577351,
      "grad_norm": 0.054444171488285065,
      "learning_rate": 5.272887648841343e-06,
      "loss": 0.0022,
      "step": 166630
    },
    {
      "epoch": 1.0687771980515213,
      "grad_norm": 0.07919225096702576,
      "learning_rate": 5.272328777975952e-06,
      "loss": 0.0015,
      "step": 166640
    },
    {
      "epoch": 1.0688413349453072,
      "grad_norm": 0.05821329727768898,
      "learning_rate": 5.271769903698086e-06,
      "loss": 0.0018,
      "step": 166650
    },
    {
      "epoch": 1.0689054718390933,
      "grad_norm": 0.04042275249958038,
      "learning_rate": 5.271211026014751e-06,
      "loss": 0.0012,
      "step": 166660
    },
    {
      "epoch": 1.0689696087328795,
      "grad_norm": 0.12543560564517975,
      "learning_rate": 5.270652144932946e-06,
      "loss": 0.0083,
      "step": 166670
    },
    {
      "epoch": 1.0690337456266656,
      "grad_norm": 0.22743694484233856,
      "learning_rate": 5.2700932604596774e-06,
      "loss": 0.0039,
      "step": 166680
    },
    {
      "epoch": 1.0690978825204516,
      "grad_norm": 0.02137444168329239,
      "learning_rate": 5.269534372601946e-06,
      "loss": 0.0036,
      "step": 166690
    },
    {
      "epoch": 1.0691620194142377,
      "grad_norm": 0.018823714926838875,
      "learning_rate": 5.268975481366758e-06,
      "loss": 0.001,
      "step": 166700
    },
    {
      "epoch": 1.0692261563080239,
      "grad_norm": 0.19334377348423004,
      "learning_rate": 5.268416586761113e-06,
      "loss": 0.0023,
      "step": 166710
    },
    {
      "epoch": 1.06929029320181,
      "grad_norm": 0.052227046340703964,
      "learning_rate": 5.267857688792019e-06,
      "loss": 0.0014,
      "step": 166720
    },
    {
      "epoch": 1.069354430095596,
      "grad_norm": 0.1632692813873291,
      "learning_rate": 5.267298787466474e-06,
      "loss": 0.0024,
      "step": 166730
    },
    {
      "epoch": 1.069418566989382,
      "grad_norm": 0.08603489398956299,
      "learning_rate": 5.266739882791485e-06,
      "loss": 0.0012,
      "step": 166740
    },
    {
      "epoch": 1.0694827038831682,
      "grad_norm": 0.2116672843694687,
      "learning_rate": 5.2661809747740554e-06,
      "loss": 0.0018,
      "step": 166750
    },
    {
      "epoch": 1.0695468407769544,
      "grad_norm": 0.13084769248962402,
      "learning_rate": 5.265622063421186e-06,
      "loss": 0.001,
      "step": 166760
    },
    {
      "epoch": 1.0696109776707403,
      "grad_norm": 0.057312317192554474,
      "learning_rate": 5.265063148739885e-06,
      "loss": 0.0019,
      "step": 166770
    },
    {
      "epoch": 1.0696751145645265,
      "grad_norm": 0.11602775007486343,
      "learning_rate": 5.26450423073715e-06,
      "loss": 0.0014,
      "step": 166780
    },
    {
      "epoch": 1.0697392514583126,
      "grad_norm": 0.04075274243950844,
      "learning_rate": 5.2639453094199924e-06,
      "loss": 0.0023,
      "step": 166790
    },
    {
      "epoch": 1.0698033883520988,
      "grad_norm": 0.12920819222927094,
      "learning_rate": 5.263386384795407e-06,
      "loss": 0.0022,
      "step": 166800
    },
    {
      "epoch": 1.069867525245885,
      "grad_norm": 0.2572612464427948,
      "learning_rate": 5.262827456870404e-06,
      "loss": 0.0013,
      "step": 166810
    },
    {
      "epoch": 1.0699316621396708,
      "grad_norm": 0.14153194427490234,
      "learning_rate": 5.262268525651985e-06,
      "loss": 0.0016,
      "step": 166820
    },
    {
      "epoch": 1.069995799033457,
      "grad_norm": 0.0538192056119442,
      "learning_rate": 5.261709591147154e-06,
      "loss": 0.001,
      "step": 166830
    },
    {
      "epoch": 1.0700599359272431,
      "grad_norm": 0.09090349823236465,
      "learning_rate": 5.261150653362914e-06,
      "loss": 0.0036,
      "step": 166840
    },
    {
      "epoch": 1.0701240728210293,
      "grad_norm": 0.08936433494091034,
      "learning_rate": 5.260591712306271e-06,
      "loss": 0.0014,
      "step": 166850
    },
    {
      "epoch": 1.0701882097148152,
      "grad_norm": 0.1939397007226944,
      "learning_rate": 5.2600327679842264e-06,
      "loss": 0.0014,
      "step": 166860
    },
    {
      "epoch": 1.0702523466086014,
      "grad_norm": 0.058595966547727585,
      "learning_rate": 5.259473820403788e-06,
      "loss": 0.0016,
      "step": 166870
    },
    {
      "epoch": 1.0703164835023875,
      "grad_norm": 0.14676587283611298,
      "learning_rate": 5.258914869571954e-06,
      "loss": 0.0022,
      "step": 166880
    },
    {
      "epoch": 1.0703806203961737,
      "grad_norm": 0.18484529852867126,
      "learning_rate": 5.2583559154957334e-06,
      "loss": 0.0018,
      "step": 166890
    },
    {
      "epoch": 1.0704447572899598,
      "grad_norm": 0.10423953086137772,
      "learning_rate": 5.257796958182127e-06,
      "loss": 0.0013,
      "step": 166900
    },
    {
      "epoch": 1.0705088941837457,
      "grad_norm": 0.07524426281452179,
      "learning_rate": 5.257237997638143e-06,
      "loss": 0.0014,
      "step": 166910
    },
    {
      "epoch": 1.0705730310775319,
      "grad_norm": 0.03524652123451233,
      "learning_rate": 5.256679033870781e-06,
      "loss": 0.0024,
      "step": 166920
    },
    {
      "epoch": 1.070637167971318,
      "grad_norm": 0.14464613795280457,
      "learning_rate": 5.256120066887049e-06,
      "loss": 0.002,
      "step": 166930
    },
    {
      "epoch": 1.0707013048651042,
      "grad_norm": 0.15725456178188324,
      "learning_rate": 5.255561096693947e-06,
      "loss": 0.0014,
      "step": 166940
    },
    {
      "epoch": 1.07076544175889,
      "grad_norm": 0.0924675464630127,
      "learning_rate": 5.255002123298482e-06,
      "loss": 0.0018,
      "step": 166950
    },
    {
      "epoch": 1.0708295786526763,
      "grad_norm": 0.13643372058868408,
      "learning_rate": 5.254443146707661e-06,
      "loss": 0.0012,
      "step": 166960
    },
    {
      "epoch": 1.0708937155464624,
      "grad_norm": 0.027466630563139915,
      "learning_rate": 5.253884166928482e-06,
      "loss": 0.0017,
      "step": 166970
    },
    {
      "epoch": 1.0709578524402485,
      "grad_norm": 0.062290776520967484,
      "learning_rate": 5.253325183967956e-06,
      "loss": 0.0013,
      "step": 166980
    },
    {
      "epoch": 1.0710219893340345,
      "grad_norm": 0.1609766185283661,
      "learning_rate": 5.252766197833082e-06,
      "loss": 0.0019,
      "step": 166990
    },
    {
      "epoch": 1.0710861262278206,
      "grad_norm": 0.01689579151570797,
      "learning_rate": 5.252207208530869e-06,
      "loss": 0.0015,
      "step": 167000
    },
    {
      "epoch": 1.0711502631216068,
      "grad_norm": 0.015078687109053135,
      "learning_rate": 5.251648216068318e-06,
      "loss": 0.0024,
      "step": 167010
    },
    {
      "epoch": 1.071214400015393,
      "grad_norm": 0.11737733334302902,
      "learning_rate": 5.251089220452436e-06,
      "loss": 0.002,
      "step": 167020
    },
    {
      "epoch": 1.0712785369091788,
      "grad_norm": 0.1474132090806961,
      "learning_rate": 5.250530221690225e-06,
      "loss": 0.0021,
      "step": 167030
    },
    {
      "epoch": 1.071342673802965,
      "grad_norm": 0.12184314429759979,
      "learning_rate": 5.249971219788693e-06,
      "loss": 0.0019,
      "step": 167040
    },
    {
      "epoch": 1.0714068106967511,
      "grad_norm": 0.04866328462958336,
      "learning_rate": 5.2494122147548424e-06,
      "loss": 0.002,
      "step": 167050
    },
    {
      "epoch": 1.0714709475905373,
      "grad_norm": 0.07479048520326614,
      "learning_rate": 5.248853206595679e-06,
      "loss": 0.0029,
      "step": 167060
    },
    {
      "epoch": 1.0715350844843234,
      "grad_norm": 0.21589982509613037,
      "learning_rate": 5.248294195318205e-06,
      "loss": 0.0018,
      "step": 167070
    },
    {
      "epoch": 1.0715992213781094,
      "grad_norm": 0.02624637447297573,
      "learning_rate": 5.24773518092943e-06,
      "loss": 0.0019,
      "step": 167080
    },
    {
      "epoch": 1.0716633582718955,
      "grad_norm": 0.0176876001060009,
      "learning_rate": 5.247176163436354e-06,
      "loss": 0.0013,
      "step": 167090
    },
    {
      "epoch": 1.0717274951656817,
      "grad_norm": 0.0757228434085846,
      "learning_rate": 5.246617142845985e-06,
      "loss": 0.0016,
      "step": 167100
    },
    {
      "epoch": 1.0717916320594678,
      "grad_norm": 0.12094847112894058,
      "learning_rate": 5.246058119165327e-06,
      "loss": 0.0014,
      "step": 167110
    },
    {
      "epoch": 1.0718557689532537,
      "grad_norm": 0.02124212682247162,
      "learning_rate": 5.245499092401384e-06,
      "loss": 0.0031,
      "step": 167120
    },
    {
      "epoch": 1.0719199058470399,
      "grad_norm": 0.07685308158397675,
      "learning_rate": 5.244940062561164e-06,
      "loss": 0.0024,
      "step": 167130
    },
    {
      "epoch": 1.071984042740826,
      "grad_norm": 0.06877698004245758,
      "learning_rate": 5.244381029651667e-06,
      "loss": 0.0014,
      "step": 167140
    },
    {
      "epoch": 1.0720481796346122,
      "grad_norm": 0.0037930766120553017,
      "learning_rate": 5.243821993679904e-06,
      "loss": 0.0026,
      "step": 167150
    },
    {
      "epoch": 1.0721123165283981,
      "grad_norm": 0.08032867312431335,
      "learning_rate": 5.243262954652875e-06,
      "loss": 0.0014,
      "step": 167160
    },
    {
      "epoch": 1.0721764534221843,
      "grad_norm": 0.11195434629917145,
      "learning_rate": 5.242703912577588e-06,
      "loss": 0.0017,
      "step": 167170
    },
    {
      "epoch": 1.0722405903159704,
      "grad_norm": 0.07904714345932007,
      "learning_rate": 5.242144867461047e-06,
      "loss": 0.0012,
      "step": 167180
    },
    {
      "epoch": 1.0723047272097566,
      "grad_norm": 0.13505138456821442,
      "learning_rate": 5.241585819310258e-06,
      "loss": 0.0027,
      "step": 167190
    },
    {
      "epoch": 1.0723688641035425,
      "grad_norm": 0.12018563598394394,
      "learning_rate": 5.241026768132227e-06,
      "loss": 0.0014,
      "step": 167200
    },
    {
      "epoch": 1.0724330009973286,
      "grad_norm": 0.014983379282057285,
      "learning_rate": 5.240467713933957e-06,
      "loss": 0.0015,
      "step": 167210
    },
    {
      "epoch": 1.0724971378911148,
      "grad_norm": 0.12585660815238953,
      "learning_rate": 5.239908656722455e-06,
      "loss": 0.0019,
      "step": 167220
    },
    {
      "epoch": 1.072561274784901,
      "grad_norm": 0.13963164389133453,
      "learning_rate": 5.239349596504728e-06,
      "loss": 0.0024,
      "step": 167230
    },
    {
      "epoch": 1.072625411678687,
      "grad_norm": 0.1950705200433731,
      "learning_rate": 5.238790533287778e-06,
      "loss": 0.001,
      "step": 167240
    },
    {
      "epoch": 1.072689548572473,
      "grad_norm": 0.1334400475025177,
      "learning_rate": 5.2382314670786114e-06,
      "loss": 0.0011,
      "step": 167250
    },
    {
      "epoch": 1.0727536854662592,
      "grad_norm": 0.15896040201187134,
      "learning_rate": 5.237672397884235e-06,
      "loss": 0.0014,
      "step": 167260
    },
    {
      "epoch": 1.0728178223600453,
      "grad_norm": 0.03526372089982033,
      "learning_rate": 5.2371133257116535e-06,
      "loss": 0.0018,
      "step": 167270
    },
    {
      "epoch": 1.0728819592538315,
      "grad_norm": 0.06044013053178787,
      "learning_rate": 5.236554250567873e-06,
      "loss": 0.0023,
      "step": 167280
    },
    {
      "epoch": 1.0729460961476174,
      "grad_norm": 0.11396222561597824,
      "learning_rate": 5.235995172459898e-06,
      "loss": 0.0019,
      "step": 167290
    },
    {
      "epoch": 1.0730102330414035,
      "grad_norm": 0.03265209496021271,
      "learning_rate": 5.235436091394735e-06,
      "loss": 0.0029,
      "step": 167300
    },
    {
      "epoch": 1.0730743699351897,
      "grad_norm": 0.11949179321527481,
      "learning_rate": 5.234877007379388e-06,
      "loss": 0.0019,
      "step": 167310
    },
    {
      "epoch": 1.0731385068289758,
      "grad_norm": 0.18657705187797546,
      "learning_rate": 5.234317920420867e-06,
      "loss": 0.0035,
      "step": 167320
    },
    {
      "epoch": 1.073202643722762,
      "grad_norm": 0.11149043589830399,
      "learning_rate": 5.233758830526173e-06,
      "loss": 0.0013,
      "step": 167330
    },
    {
      "epoch": 1.073266780616548,
      "grad_norm": 0.0194416344165802,
      "learning_rate": 5.233199737702314e-06,
      "loss": 0.0024,
      "step": 167340
    },
    {
      "epoch": 1.073330917510334,
      "grad_norm": 0.06449584662914276,
      "learning_rate": 5.232640641956297e-06,
      "loss": 0.0013,
      "step": 167350
    },
    {
      "epoch": 1.0733950544041202,
      "grad_norm": 0.2929050028324127,
      "learning_rate": 5.232081543295126e-06,
      "loss": 0.0017,
      "step": 167360
    },
    {
      "epoch": 1.0734591912979063,
      "grad_norm": 0.0034871790558099747,
      "learning_rate": 5.231522441725806e-06,
      "loss": 0.0019,
      "step": 167370
    },
    {
      "epoch": 1.0735233281916923,
      "grad_norm": 0.01426205039024353,
      "learning_rate": 5.2309633372553456e-06,
      "loss": 0.002,
      "step": 167380
    },
    {
      "epoch": 1.0735874650854784,
      "grad_norm": 0.002224660012871027,
      "learning_rate": 5.23040422989075e-06,
      "loss": 0.0012,
      "step": 167390
    },
    {
      "epoch": 1.0736516019792646,
      "grad_norm": 0.05631047859787941,
      "learning_rate": 5.229845119639025e-06,
      "loss": 0.0015,
      "step": 167400
    },
    {
      "epoch": 1.0737157388730507,
      "grad_norm": 0.07507754117250443,
      "learning_rate": 5.229286006507174e-06,
      "loss": 0.0029,
      "step": 167410
    },
    {
      "epoch": 1.0737798757668366,
      "grad_norm": 0.1423020362854004,
      "learning_rate": 5.228726890502207e-06,
      "loss": 0.0019,
      "step": 167420
    },
    {
      "epoch": 1.0738440126606228,
      "grad_norm": 0.12582719326019287,
      "learning_rate": 5.228167771631129e-06,
      "loss": 0.0018,
      "step": 167430
    },
    {
      "epoch": 1.073908149554409,
      "grad_norm": 0.059179939329624176,
      "learning_rate": 5.227608649900944e-06,
      "loss": 0.001,
      "step": 167440
    },
    {
      "epoch": 1.073972286448195,
      "grad_norm": 0.11064352840185165,
      "learning_rate": 5.2270495253186615e-06,
      "loss": 0.0014,
      "step": 167450
    },
    {
      "epoch": 1.074036423341981,
      "grad_norm": 0.05817147344350815,
      "learning_rate": 5.226490397891285e-06,
      "loss": 0.0029,
      "step": 167460
    },
    {
      "epoch": 1.0741005602357672,
      "grad_norm": 0.08750585466623306,
      "learning_rate": 5.225931267625824e-06,
      "loss": 0.0012,
      "step": 167470
    },
    {
      "epoch": 1.0741646971295533,
      "grad_norm": 0.08948834985494614,
      "learning_rate": 5.22537213452928e-06,
      "loss": 0.0023,
      "step": 167480
    },
    {
      "epoch": 1.0742288340233395,
      "grad_norm": 0.08464883267879486,
      "learning_rate": 5.224812998608665e-06,
      "loss": 0.0014,
      "step": 167490
    },
    {
      "epoch": 1.0742929709171256,
      "grad_norm": 0.1435505747795105,
      "learning_rate": 5.22425385987098e-06,
      "loss": 0.0022,
      "step": 167500
    },
    {
      "epoch": 1.0743571078109115,
      "grad_norm": 0.05624101310968399,
      "learning_rate": 5.2236947183232356e-06,
      "loss": 0.003,
      "step": 167510
    },
    {
      "epoch": 1.0744212447046977,
      "grad_norm": 0.048413850367069244,
      "learning_rate": 5.2231355739724355e-06,
      "loss": 0.0026,
      "step": 167520
    },
    {
      "epoch": 1.0744853815984838,
      "grad_norm": 0.1629071682691574,
      "learning_rate": 5.222576426825588e-06,
      "loss": 0.0016,
      "step": 167530
    },
    {
      "epoch": 1.07454951849227,
      "grad_norm": 0.0808134377002716,
      "learning_rate": 5.2220172768896986e-06,
      "loss": 0.0028,
      "step": 167540
    },
    {
      "epoch": 1.074613655386056,
      "grad_norm": 0.29669323563575745,
      "learning_rate": 5.221458124171774e-06,
      "loss": 0.0045,
      "step": 167550
    },
    {
      "epoch": 1.074677792279842,
      "grad_norm": 0.052580058574676514,
      "learning_rate": 5.220898968678821e-06,
      "loss": 0.0029,
      "step": 167560
    },
    {
      "epoch": 1.0747419291736282,
      "grad_norm": 0.05036512017250061,
      "learning_rate": 5.220339810417844e-06,
      "loss": 0.0017,
      "step": 167570
    },
    {
      "epoch": 1.0748060660674144,
      "grad_norm": 0.004044400528073311,
      "learning_rate": 5.219780649395852e-06,
      "loss": 0.0009,
      "step": 167580
    },
    {
      "epoch": 1.0748702029612003,
      "grad_norm": 0.0637139230966568,
      "learning_rate": 5.219221485619854e-06,
      "loss": 0.0013,
      "step": 167590
    },
    {
      "epoch": 1.0749343398549864,
      "grad_norm": 0.05439816042780876,
      "learning_rate": 5.218662319096852e-06,
      "loss": 0.0014,
      "step": 167600
    },
    {
      "epoch": 1.0749984767487726,
      "grad_norm": 0.1431669443845749,
      "learning_rate": 5.218103149833856e-06,
      "loss": 0.002,
      "step": 167610
    },
    {
      "epoch": 1.0750626136425587,
      "grad_norm": 0.06507054716348648,
      "learning_rate": 5.217543977837871e-06,
      "loss": 0.0011,
      "step": 167620
    },
    {
      "epoch": 1.0751267505363449,
      "grad_norm": 0.11313388496637344,
      "learning_rate": 5.2169848031159065e-06,
      "loss": 0.0016,
      "step": 167630
    },
    {
      "epoch": 1.0751908874301308,
      "grad_norm": 0.1550733745098114,
      "learning_rate": 5.216425625674965e-06,
      "loss": 0.0012,
      "step": 167640
    },
    {
      "epoch": 1.075255024323917,
      "grad_norm": 0.05684865638613701,
      "learning_rate": 5.215866445522056e-06,
      "loss": 0.0015,
      "step": 167650
    },
    {
      "epoch": 1.075319161217703,
      "grad_norm": 0.11183615028858185,
      "learning_rate": 5.215307262664187e-06,
      "loss": 0.0031,
      "step": 167660
    },
    {
      "epoch": 1.0753832981114892,
      "grad_norm": 0.05034491792321205,
      "learning_rate": 5.214748077108363e-06,
      "loss": 0.0026,
      "step": 167670
    },
    {
      "epoch": 1.0754474350052752,
      "grad_norm": 0.055223409086465836,
      "learning_rate": 5.214188888861593e-06,
      "loss": 0.0018,
      "step": 167680
    },
    {
      "epoch": 1.0755115718990613,
      "grad_norm": 0.006556411739438772,
      "learning_rate": 5.213629697930883e-06,
      "loss": 0.0018,
      "step": 167690
    },
    {
      "epoch": 1.0755757087928475,
      "grad_norm": 0.1441619098186493,
      "learning_rate": 5.213070504323241e-06,
      "loss": 0.0014,
      "step": 167700
    },
    {
      "epoch": 1.0756398456866336,
      "grad_norm": 0.1584549993276596,
      "learning_rate": 5.212511308045672e-06,
      "loss": 0.0036,
      "step": 167710
    },
    {
      "epoch": 1.0757039825804195,
      "grad_norm": 0.14993031322956085,
      "learning_rate": 5.211952109105185e-06,
      "loss": 0.0018,
      "step": 167720
    },
    {
      "epoch": 1.0757681194742057,
      "grad_norm": 0.2008506953716278,
      "learning_rate": 5.211392907508785e-06,
      "loss": 0.001,
      "step": 167730
    },
    {
      "epoch": 1.0758322563679918,
      "grad_norm": 0.20662356913089752,
      "learning_rate": 5.210833703263484e-06,
      "loss": 0.0022,
      "step": 167740
    },
    {
      "epoch": 1.075896393261778,
      "grad_norm": 0.1540125012397766,
      "learning_rate": 5.210274496376284e-06,
      "loss": 0.001,
      "step": 167750
    },
    {
      "epoch": 1.0759605301555641,
      "grad_norm": 0.13144925236701965,
      "learning_rate": 5.209715286854195e-06,
      "loss": 0.0029,
      "step": 167760
    },
    {
      "epoch": 1.07602466704935,
      "grad_norm": 0.06378944963216782,
      "learning_rate": 5.209156074704222e-06,
      "loss": 0.0034,
      "step": 167770
    },
    {
      "epoch": 1.0760888039431362,
      "grad_norm": 0.0328482985496521,
      "learning_rate": 5.208596859933376e-06,
      "loss": 0.0026,
      "step": 167780
    },
    {
      "epoch": 1.0761529408369224,
      "grad_norm": 0.0747612789273262,
      "learning_rate": 5.208037642548661e-06,
      "loss": 0.0012,
      "step": 167790
    },
    {
      "epoch": 1.0762170777307085,
      "grad_norm": 0.13654185831546783,
      "learning_rate": 5.2074784225570855e-06,
      "loss": 0.0013,
      "step": 167800
    },
    {
      "epoch": 1.0762812146244944,
      "grad_norm": 0.13586686551570892,
      "learning_rate": 5.206919199965657e-06,
      "loss": 0.0022,
      "step": 167810
    },
    {
      "epoch": 1.0763453515182806,
      "grad_norm": 0.03900545835494995,
      "learning_rate": 5.206359974781384e-06,
      "loss": 0.0009,
      "step": 167820
    },
    {
      "epoch": 1.0764094884120667,
      "grad_norm": 0.15561453998088837,
      "learning_rate": 5.205800747011274e-06,
      "loss": 0.0019,
      "step": 167830
    },
    {
      "epoch": 1.0764736253058529,
      "grad_norm": 0.12779083847999573,
      "learning_rate": 5.205241516662331e-06,
      "loss": 0.0012,
      "step": 167840
    },
    {
      "epoch": 1.0765377621996388,
      "grad_norm": 0.17967472970485687,
      "learning_rate": 5.204682283741567e-06,
      "loss": 0.0071,
      "step": 167850
    },
    {
      "epoch": 1.076601899093425,
      "grad_norm": 0.010842211544513702,
      "learning_rate": 5.2041230482559865e-06,
      "loss": 0.0034,
      "step": 167860
    },
    {
      "epoch": 1.0766660359872111,
      "grad_norm": 0.013351025059819221,
      "learning_rate": 5.2035638102126e-06,
      "loss": 0.001,
      "step": 167870
    },
    {
      "epoch": 1.0767301728809973,
      "grad_norm": 0.22351759672164917,
      "learning_rate": 5.20300456961841e-06,
      "loss": 0.0034,
      "step": 167880
    },
    {
      "epoch": 1.0767943097747832,
      "grad_norm": 0.03208925202488899,
      "learning_rate": 5.202445326480432e-06,
      "loss": 0.0027,
      "step": 167890
    },
    {
      "epoch": 1.0768584466685693,
      "grad_norm": 0.10712125897407532,
      "learning_rate": 5.201886080805666e-06,
      "loss": 0.0015,
      "step": 167900
    },
    {
      "epoch": 1.0769225835623555,
      "grad_norm": 0.05341329425573349,
      "learning_rate": 5.201326832601125e-06,
      "loss": 0.0014,
      "step": 167910
    },
    {
      "epoch": 1.0769867204561416,
      "grad_norm": 0.1631436049938202,
      "learning_rate": 5.200767581873815e-06,
      "loss": 0.0025,
      "step": 167920
    },
    {
      "epoch": 1.0770508573499278,
      "grad_norm": 0.19283632934093475,
      "learning_rate": 5.200208328630743e-06,
      "loss": 0.0035,
      "step": 167930
    },
    {
      "epoch": 1.0771149942437137,
      "grad_norm": 0.032223086804151535,
      "learning_rate": 5.199649072878917e-06,
      "loss": 0.0015,
      "step": 167940
    },
    {
      "epoch": 1.0771791311374999,
      "grad_norm": 0.12343668192625046,
      "learning_rate": 5.199089814625345e-06,
      "loss": 0.0017,
      "step": 167950
    },
    {
      "epoch": 1.077243268031286,
      "grad_norm": 0.2309725284576416,
      "learning_rate": 5.198530553877036e-06,
      "loss": 0.0031,
      "step": 167960
    },
    {
      "epoch": 1.0773074049250722,
      "grad_norm": 0.06467442959547043,
      "learning_rate": 5.197971290640999e-06,
      "loss": 0.001,
      "step": 167970
    },
    {
      "epoch": 1.077371541818858,
      "grad_norm": 0.19849643111228943,
      "learning_rate": 5.1974120249242374e-06,
      "loss": 0.0015,
      "step": 167980
    },
    {
      "epoch": 1.0774356787126442,
      "grad_norm": 0.0701538547873497,
      "learning_rate": 5.196852756733765e-06,
      "loss": 0.0021,
      "step": 167990
    },
    {
      "epoch": 1.0774998156064304,
      "grad_norm": 0.10209740698337555,
      "learning_rate": 5.1962934860765845e-06,
      "loss": 0.0021,
      "step": 168000
    },
    {
      "epoch": 1.0775639525002165,
      "grad_norm": 0.08195840567350388,
      "learning_rate": 5.195734212959706e-06,
      "loss": 0.0013,
      "step": 168010
    },
    {
      "epoch": 1.0776280893940027,
      "grad_norm": 0.011830402538180351,
      "learning_rate": 5.195174937390141e-06,
      "loss": 0.002,
      "step": 168020
    },
    {
      "epoch": 1.0776922262877886,
      "grad_norm": 0.05599063262343407,
      "learning_rate": 5.194615659374892e-06,
      "loss": 0.002,
      "step": 168030
    },
    {
      "epoch": 1.0777563631815748,
      "grad_norm": 0.044811591506004333,
      "learning_rate": 5.194056378920971e-06,
      "loss": 0.0016,
      "step": 168040
    },
    {
      "epoch": 1.077820500075361,
      "grad_norm": 0.10353126376867294,
      "learning_rate": 5.193497096035384e-06,
      "loss": 0.0017,
      "step": 168050
    },
    {
      "epoch": 1.077884636969147,
      "grad_norm": 0.006115327589213848,
      "learning_rate": 5.19293781072514e-06,
      "loss": 0.0013,
      "step": 168060
    },
    {
      "epoch": 1.077948773862933,
      "grad_norm": 0.024776805192232132,
      "learning_rate": 5.192378522997248e-06,
      "loss": 0.0015,
      "step": 168070
    },
    {
      "epoch": 1.0780129107567191,
      "grad_norm": 0.034329865127801895,
      "learning_rate": 5.1918192328587155e-06,
      "loss": 0.0009,
      "step": 168080
    },
    {
      "epoch": 1.0780770476505053,
      "grad_norm": 0.03323132172226906,
      "learning_rate": 5.1912599403165495e-06,
      "loss": 0.0014,
      "step": 168090
    },
    {
      "epoch": 1.0781411845442914,
      "grad_norm": 0.06228185445070267,
      "learning_rate": 5.190700645377762e-06,
      "loss": 0.0023,
      "step": 168100
    },
    {
      "epoch": 1.0782053214380773,
      "grad_norm": 0.012968168593943119,
      "learning_rate": 5.190141348049358e-06,
      "loss": 0.0018,
      "step": 168110
    },
    {
      "epoch": 1.0782694583318635,
      "grad_norm": 0.24061404168605804,
      "learning_rate": 5.189582048338348e-06,
      "loss": 0.0025,
      "step": 168120
    },
    {
      "epoch": 1.0783335952256496,
      "grad_norm": 0.060437899082899094,
      "learning_rate": 5.189022746251737e-06,
      "loss": 0.0011,
      "step": 168130
    },
    {
      "epoch": 1.0783977321194358,
      "grad_norm": 0.06799522787332535,
      "learning_rate": 5.188463441796538e-06,
      "loss": 0.0024,
      "step": 168140
    },
    {
      "epoch": 1.0784618690132217,
      "grad_norm": 0.04411352425813675,
      "learning_rate": 5.187904134979758e-06,
      "loss": 0.0014,
      "step": 168150
    },
    {
      "epoch": 1.0785260059070079,
      "grad_norm": 0.08532782644033432,
      "learning_rate": 5.187344825808403e-06,
      "loss": 0.0015,
      "step": 168160
    },
    {
      "epoch": 1.078590142800794,
      "grad_norm": 0.05462755635380745,
      "learning_rate": 5.186785514289484e-06,
      "loss": 0.0016,
      "step": 168170
    },
    {
      "epoch": 1.0786542796945802,
      "grad_norm": 0.07991321384906769,
      "learning_rate": 5.186226200430009e-06,
      "loss": 0.0025,
      "step": 168180
    },
    {
      "epoch": 1.0787184165883663,
      "grad_norm": 0.14190146327018738,
      "learning_rate": 5.185666884236987e-06,
      "loss": 0.0021,
      "step": 168190
    },
    {
      "epoch": 1.0787825534821522,
      "grad_norm": 0.02735942229628563,
      "learning_rate": 5.185107565717426e-06,
      "loss": 0.0028,
      "step": 168200
    },
    {
      "epoch": 1.0788466903759384,
      "grad_norm": 0.05069974437355995,
      "learning_rate": 5.184548244878336e-06,
      "loss": 0.0025,
      "step": 168210
    },
    {
      "epoch": 1.0789108272697245,
      "grad_norm": 0.02090507000684738,
      "learning_rate": 5.183988921726723e-06,
      "loss": 0.0028,
      "step": 168220
    },
    {
      "epoch": 1.0789749641635107,
      "grad_norm": 0.1305733025074005,
      "learning_rate": 5.183429596269598e-06,
      "loss": 0.003,
      "step": 168230
    },
    {
      "epoch": 1.0790391010572966,
      "grad_norm": 0.13104850053787231,
      "learning_rate": 5.182870268513969e-06,
      "loss": 0.0015,
      "step": 168240
    },
    {
      "epoch": 1.0791032379510828,
      "grad_norm": 0.1724039614200592,
      "learning_rate": 5.182310938466845e-06,
      "loss": 0.0025,
      "step": 168250
    },
    {
      "epoch": 1.079167374844869,
      "grad_norm": 0.08119799941778183,
      "learning_rate": 5.181751606135233e-06,
      "loss": 0.0011,
      "step": 168260
    },
    {
      "epoch": 1.079231511738655,
      "grad_norm": 0.043051231652498245,
      "learning_rate": 5.1811922715261464e-06,
      "loss": 0.0012,
      "step": 168270
    },
    {
      "epoch": 1.079295648632441,
      "grad_norm": 0.05889670178294182,
      "learning_rate": 5.180632934646589e-06,
      "loss": 0.0014,
      "step": 168280
    },
    {
      "epoch": 1.0793597855262271,
      "grad_norm": 0.05956847965717316,
      "learning_rate": 5.180073595503572e-06,
      "loss": 0.0013,
      "step": 168290
    },
    {
      "epoch": 1.0794239224200133,
      "grad_norm": 0.10754121840000153,
      "learning_rate": 5.179514254104106e-06,
      "loss": 0.001,
      "step": 168300
    },
    {
      "epoch": 1.0794880593137994,
      "grad_norm": 0.13765943050384521,
      "learning_rate": 5.178954910455195e-06,
      "loss": 0.0017,
      "step": 168310
    },
    {
      "epoch": 1.0795521962075854,
      "grad_norm": 0.08927934616804123,
      "learning_rate": 5.178395564563854e-06,
      "loss": 0.0014,
      "step": 168320
    },
    {
      "epoch": 1.0796163331013715,
      "grad_norm": 0.04931335151195526,
      "learning_rate": 5.1778362164370875e-06,
      "loss": 0.0016,
      "step": 168330
    },
    {
      "epoch": 1.0796804699951577,
      "grad_norm": 0.011807376518845558,
      "learning_rate": 5.1772768660819065e-06,
      "loss": 0.0007,
      "step": 168340
    },
    {
      "epoch": 1.0797446068889438,
      "grad_norm": 0.1913766860961914,
      "learning_rate": 5.17671751350532e-06,
      "loss": 0.0022,
      "step": 168350
    },
    {
      "epoch": 1.07980874378273,
      "grad_norm": 0.12997888028621674,
      "learning_rate": 5.176158158714335e-06,
      "loss": 0.0018,
      "step": 168360
    },
    {
      "epoch": 1.0798728806765159,
      "grad_norm": 0.10905221104621887,
      "learning_rate": 5.175598801715964e-06,
      "loss": 0.002,
      "step": 168370
    },
    {
      "epoch": 1.079937017570302,
      "grad_norm": 0.05260107293725014,
      "learning_rate": 5.1750394425172146e-06,
      "loss": 0.0029,
      "step": 168380
    },
    {
      "epoch": 1.0800011544640882,
      "grad_norm": 0.15370966494083405,
      "learning_rate": 5.174480081125095e-06,
      "loss": 0.0025,
      "step": 168390
    },
    {
      "epoch": 1.0800652913578743,
      "grad_norm": 0.10007897019386292,
      "learning_rate": 5.173920717546617e-06,
      "loss": 0.001,
      "step": 168400
    },
    {
      "epoch": 1.0801294282516603,
      "grad_norm": 0.13257601857185364,
      "learning_rate": 5.173361351788786e-06,
      "loss": 0.0015,
      "step": 168410
    },
    {
      "epoch": 1.0801935651454464,
      "grad_norm": 0.06499084830284119,
      "learning_rate": 5.172801983858615e-06,
      "loss": 0.0011,
      "step": 168420
    },
    {
      "epoch": 1.0802577020392325,
      "grad_norm": 0.08412903547286987,
      "learning_rate": 5.172242613763111e-06,
      "loss": 0.0016,
      "step": 168430
    },
    {
      "epoch": 1.0803218389330187,
      "grad_norm": 0.05546203628182411,
      "learning_rate": 5.171683241509284e-06,
      "loss": 0.0017,
      "step": 168440
    },
    {
      "epoch": 1.0803859758268048,
      "grad_norm": 0.10209858417510986,
      "learning_rate": 5.171123867104143e-06,
      "loss": 0.0022,
      "step": 168450
    },
    {
      "epoch": 1.0804501127205908,
      "grad_norm": 0.01484663411974907,
      "learning_rate": 5.170564490554698e-06,
      "loss": 0.0012,
      "step": 168460
    },
    {
      "epoch": 1.080514249614377,
      "grad_norm": 0.05575484409928322,
      "learning_rate": 5.170005111867957e-06,
      "loss": 0.0017,
      "step": 168470
    },
    {
      "epoch": 1.080578386508163,
      "grad_norm": 0.04578129202127457,
      "learning_rate": 5.169445731050933e-06,
      "loss": 0.0013,
      "step": 168480
    },
    {
      "epoch": 1.0806425234019492,
      "grad_norm": 0.123416967689991,
      "learning_rate": 5.16888634811063e-06,
      "loss": 0.0036,
      "step": 168490
    },
    {
      "epoch": 1.0807066602957351,
      "grad_norm": 0.03386322036385536,
      "learning_rate": 5.168326963054061e-06,
      "loss": 0.0016,
      "step": 168500
    },
    {
      "epoch": 1.0807707971895213,
      "grad_norm": 0.0203689094632864,
      "learning_rate": 5.1677675758882366e-06,
      "loss": 0.0015,
      "step": 168510
    },
    {
      "epoch": 1.0808349340833074,
      "grad_norm": 0.0576355904340744,
      "learning_rate": 5.167208186620163e-06,
      "loss": 0.0015,
      "step": 168520
    },
    {
      "epoch": 1.0808990709770936,
      "grad_norm": 0.027288375422358513,
      "learning_rate": 5.1666487952568525e-06,
      "loss": 0.0022,
      "step": 168530
    },
    {
      "epoch": 1.0809632078708795,
      "grad_norm": 0.04407395422458649,
      "learning_rate": 5.166089401805312e-06,
      "loss": 0.0029,
      "step": 168540
    },
    {
      "epoch": 1.0810273447646657,
      "grad_norm": 0.25099894404411316,
      "learning_rate": 5.165530006272555e-06,
      "loss": 0.0018,
      "step": 168550
    },
    {
      "epoch": 1.0810914816584518,
      "grad_norm": 0.15912792086601257,
      "learning_rate": 5.1649706086655856e-06,
      "loss": 0.002,
      "step": 168560
    },
    {
      "epoch": 1.081155618552238,
      "grad_norm": 0.08991295844316483,
      "learning_rate": 5.164411208991419e-06,
      "loss": 0.001,
      "step": 168570
    },
    {
      "epoch": 1.081219755446024,
      "grad_norm": 0.05349145457148552,
      "learning_rate": 5.163851807257061e-06,
      "loss": 0.0022,
      "step": 168580
    },
    {
      "epoch": 1.08128389233981,
      "grad_norm": 0.07239469140768051,
      "learning_rate": 5.163292403469523e-06,
      "loss": 0.0011,
      "step": 168590
    },
    {
      "epoch": 1.0813480292335962,
      "grad_norm": 0.2052467167377472,
      "learning_rate": 5.162732997635814e-06,
      "loss": 0.0015,
      "step": 168600
    },
    {
      "epoch": 1.0814121661273823,
      "grad_norm": 0.18487276136875153,
      "learning_rate": 5.162173589762946e-06,
      "loss": 0.0043,
      "step": 168610
    },
    {
      "epoch": 1.0814763030211685,
      "grad_norm": 0.11724483221769333,
      "learning_rate": 5.1616141798579255e-06,
      "loss": 0.0013,
      "step": 168620
    },
    {
      "epoch": 1.0815404399149544,
      "grad_norm": 0.0919305756688118,
      "learning_rate": 5.161054767927765e-06,
      "loss": 0.0021,
      "step": 168630
    },
    {
      "epoch": 1.0816045768087406,
      "grad_norm": 0.10012010484933853,
      "learning_rate": 5.160495353979472e-06,
      "loss": 0.0018,
      "step": 168640
    },
    {
      "epoch": 1.0816687137025267,
      "grad_norm": 0.1580265313386917,
      "learning_rate": 5.159935938020058e-06,
      "loss": 0.0021,
      "step": 168650
    },
    {
      "epoch": 1.0817328505963129,
      "grad_norm": 0.05440143495798111,
      "learning_rate": 5.159376520056532e-06,
      "loss": 0.0009,
      "step": 168660
    },
    {
      "epoch": 1.0817969874900988,
      "grad_norm": 0.09930847585201263,
      "learning_rate": 5.158817100095904e-06,
      "loss": 0.001,
      "step": 168670
    },
    {
      "epoch": 1.081861124383885,
      "grad_norm": 0.04801664873957634,
      "learning_rate": 5.158257678145185e-06,
      "loss": 0.0035,
      "step": 168680
    },
    {
      "epoch": 1.081925261277671,
      "grad_norm": 0.07561906427145004,
      "learning_rate": 5.157698254211384e-06,
      "loss": 0.0011,
      "step": 168690
    },
    {
      "epoch": 1.0819893981714572,
      "grad_norm": 0.01799064502120018,
      "learning_rate": 5.157138828301512e-06,
      "loss": 0.0031,
      "step": 168700
    },
    {
      "epoch": 1.0820535350652432,
      "grad_norm": 0.15777525305747986,
      "learning_rate": 5.156579400422577e-06,
      "loss": 0.0011,
      "step": 168710
    },
    {
      "epoch": 1.0821176719590293,
      "grad_norm": 0.051772113889455795,
      "learning_rate": 5.15601997058159e-06,
      "loss": 0.0019,
      "step": 168720
    },
    {
      "epoch": 1.0821818088528155,
      "grad_norm": 0.37480801343917847,
      "learning_rate": 5.155460538785562e-06,
      "loss": 0.0037,
      "step": 168730
    },
    {
      "epoch": 1.0822459457466016,
      "grad_norm": 0.07378017157316208,
      "learning_rate": 5.154901105041502e-06,
      "loss": 0.0027,
      "step": 168740
    },
    {
      "epoch": 1.0823100826403875,
      "grad_norm": 0.1874859482049942,
      "learning_rate": 5.154341669356421e-06,
      "loss": 0.0014,
      "step": 168750
    },
    {
      "epoch": 1.0823742195341737,
      "grad_norm": 0.09158900380134583,
      "learning_rate": 5.153782231737329e-06,
      "loss": 0.0025,
      "step": 168760
    },
    {
      "epoch": 1.0824383564279598,
      "grad_norm": 0.015413496643304825,
      "learning_rate": 5.1532227921912335e-06,
      "loss": 0.0015,
      "step": 168770
    },
    {
      "epoch": 1.082502493321746,
      "grad_norm": 0.06986375898122787,
      "learning_rate": 5.152663350725149e-06,
      "loss": 0.0004,
      "step": 168780
    },
    {
      "epoch": 1.0825666302155321,
      "grad_norm": 0.26239410042762756,
      "learning_rate": 5.152103907346083e-06,
      "loss": 0.0015,
      "step": 168790
    },
    {
      "epoch": 1.082630767109318,
      "grad_norm": 0.004599603824317455,
      "learning_rate": 5.151544462061047e-06,
      "loss": 0.0027,
      "step": 168800
    },
    {
      "epoch": 1.0826949040031042,
      "grad_norm": 0.12062758952379227,
      "learning_rate": 5.150985014877049e-06,
      "loss": 0.002,
      "step": 168810
    },
    {
      "epoch": 1.0827590408968903,
      "grad_norm": 0.026245219632983208,
      "learning_rate": 5.150425565801103e-06,
      "loss": 0.0018,
      "step": 168820
    },
    {
      "epoch": 1.0828231777906765,
      "grad_norm": 0.07881046831607819,
      "learning_rate": 5.149866114840215e-06,
      "loss": 0.0023,
      "step": 168830
    },
    {
      "epoch": 1.0828873146844624,
      "grad_norm": 0.007862749509513378,
      "learning_rate": 5.149306662001399e-06,
      "loss": 0.0022,
      "step": 168840
    },
    {
      "epoch": 1.0829514515782486,
      "grad_norm": 0.13096725940704346,
      "learning_rate": 5.1487472072916624e-06,
      "loss": 0.0016,
      "step": 168850
    },
    {
      "epoch": 1.0830155884720347,
      "grad_norm": 0.08981695771217346,
      "learning_rate": 5.148187750718018e-06,
      "loss": 0.0026,
      "step": 168860
    },
    {
      "epoch": 1.0830797253658209,
      "grad_norm": 0.12101157754659653,
      "learning_rate": 5.1476282922874756e-06,
      "loss": 0.0018,
      "step": 168870
    },
    {
      "epoch": 1.083143862259607,
      "grad_norm": 0.3226020336151123,
      "learning_rate": 5.147068832007044e-06,
      "loss": 0.002,
      "step": 168880
    },
    {
      "epoch": 1.083207999153393,
      "grad_norm": 0.05597984045743942,
      "learning_rate": 5.146509369883735e-06,
      "loss": 0.001,
      "step": 168890
    },
    {
      "epoch": 1.083272136047179,
      "grad_norm": 0.271222859621048,
      "learning_rate": 5.145949905924559e-06,
      "loss": 0.002,
      "step": 168900
    },
    {
      "epoch": 1.0833362729409652,
      "grad_norm": 0.08601807802915573,
      "learning_rate": 5.145390440136527e-06,
      "loss": 0.0013,
      "step": 168910
    },
    {
      "epoch": 1.0834004098347514,
      "grad_norm": 0.11945120990276337,
      "learning_rate": 5.144830972526649e-06,
      "loss": 0.0014,
      "step": 168920
    },
    {
      "epoch": 1.0834645467285373,
      "grad_norm": 0.09423484653234482,
      "learning_rate": 5.144271503101935e-06,
      "loss": 0.0011,
      "step": 168930
    },
    {
      "epoch": 1.0835286836223235,
      "grad_norm": 0.04961375147104263,
      "learning_rate": 5.143712031869396e-06,
      "loss": 0.0011,
      "step": 168940
    },
    {
      "epoch": 1.0835928205161096,
      "grad_norm": 0.06484077870845795,
      "learning_rate": 5.143152558836042e-06,
      "loss": 0.0013,
      "step": 168950
    },
    {
      "epoch": 1.0836569574098958,
      "grad_norm": 0.042897775769233704,
      "learning_rate": 5.142593084008884e-06,
      "loss": 0.0009,
      "step": 168960
    },
    {
      "epoch": 1.0837210943036817,
      "grad_norm": 0.0597703717648983,
      "learning_rate": 5.142033607394934e-06,
      "loss": 0.0018,
      "step": 168970
    },
    {
      "epoch": 1.0837852311974678,
      "grad_norm": 0.0672975704073906,
      "learning_rate": 5.1414741290012006e-06,
      "loss": 0.0018,
      "step": 168980
    },
    {
      "epoch": 1.083849368091254,
      "grad_norm": 0.14853569865226746,
      "learning_rate": 5.140914648834696e-06,
      "loss": 0.0018,
      "step": 168990
    },
    {
      "epoch": 1.0839135049850401,
      "grad_norm": 0.06412370502948761,
      "learning_rate": 5.140355166902429e-06,
      "loss": 0.0011,
      "step": 169000
    },
    {
      "epoch": 1.083977641878826,
      "grad_norm": 0.11806584894657135,
      "learning_rate": 5.139795683211412e-06,
      "loss": 0.0059,
      "step": 169010
    },
    {
      "epoch": 1.0840417787726122,
      "grad_norm": 0.0590679794549942,
      "learning_rate": 5.139236197768654e-06,
      "loss": 0.0015,
      "step": 169020
    },
    {
      "epoch": 1.0841059156663984,
      "grad_norm": 0.1295815408229828,
      "learning_rate": 5.138676710581169e-06,
      "loss": 0.0007,
      "step": 169030
    },
    {
      "epoch": 1.0841700525601845,
      "grad_norm": 0.18576541543006897,
      "learning_rate": 5.138117221655965e-06,
      "loss": 0.0018,
      "step": 169040
    },
    {
      "epoch": 1.0842341894539707,
      "grad_norm": 0.13462528586387634,
      "learning_rate": 5.137557731000052e-06,
      "loss": 0.0018,
      "step": 169050
    },
    {
      "epoch": 1.0842983263477566,
      "grad_norm": 0.04111761972308159,
      "learning_rate": 5.136998238620444e-06,
      "loss": 0.0019,
      "step": 169060
    },
    {
      "epoch": 1.0843624632415427,
      "grad_norm": 0.21343253552913666,
      "learning_rate": 5.13643874452415e-06,
      "loss": 0.0028,
      "step": 169070
    },
    {
      "epoch": 1.0844266001353289,
      "grad_norm": 0.13954845070838928,
      "learning_rate": 5.135879248718181e-06,
      "loss": 0.0007,
      "step": 169080
    },
    {
      "epoch": 1.084490737029115,
      "grad_norm": 0.07925862073898315,
      "learning_rate": 5.1353197512095474e-06,
      "loss": 0.0015,
      "step": 169090
    },
    {
      "epoch": 1.084554873922901,
      "grad_norm": 0.05541840195655823,
      "learning_rate": 5.13476025200526e-06,
      "loss": 0.0014,
      "step": 169100
    },
    {
      "epoch": 1.084619010816687,
      "grad_norm": 0.18047375977039337,
      "learning_rate": 5.134200751112331e-06,
      "loss": 0.0017,
      "step": 169110
    },
    {
      "epoch": 1.0846831477104732,
      "grad_norm": 0.06819538027048111,
      "learning_rate": 5.133641248537772e-06,
      "loss": 0.0018,
      "step": 169120
    },
    {
      "epoch": 1.0847472846042594,
      "grad_norm": 0.09758296608924866,
      "learning_rate": 5.133081744288591e-06,
      "loss": 0.0031,
      "step": 169130
    },
    {
      "epoch": 1.0848114214980453,
      "grad_norm": 0.11406062543392181,
      "learning_rate": 5.132522238371801e-06,
      "loss": 0.0028,
      "step": 169140
    },
    {
      "epoch": 1.0848755583918315,
      "grad_norm": 0.1453990787267685,
      "learning_rate": 5.131962730794412e-06,
      "loss": 0.0015,
      "step": 169150
    },
    {
      "epoch": 1.0849396952856176,
      "grad_norm": 0.05270492285490036,
      "learning_rate": 5.131403221563437e-06,
      "loss": 0.0024,
      "step": 169160
    },
    {
      "epoch": 1.0850038321794038,
      "grad_norm": 0.2334631234407425,
      "learning_rate": 5.130843710685885e-06,
      "loss": 0.0022,
      "step": 169170
    },
    {
      "epoch": 1.08506796907319,
      "grad_norm": 0.1925625205039978,
      "learning_rate": 5.130284198168769e-06,
      "loss": 0.0011,
      "step": 169180
    },
    {
      "epoch": 1.0851321059669758,
      "grad_norm": 0.00834691897034645,
      "learning_rate": 5.129724684019098e-06,
      "loss": 0.0012,
      "step": 169190
    },
    {
      "epoch": 1.085196242860762,
      "grad_norm": 0.10060546547174454,
      "learning_rate": 5.1291651682438835e-06,
      "loss": 0.0013,
      "step": 169200
    },
    {
      "epoch": 1.0852603797545481,
      "grad_norm": 0.07248177379369736,
      "learning_rate": 5.128605650850137e-06,
      "loss": 0.0024,
      "step": 169210
    },
    {
      "epoch": 1.0853245166483343,
      "grad_norm": 0.11243856698274612,
      "learning_rate": 5.128046131844869e-06,
      "loss": 0.0012,
      "step": 169220
    },
    {
      "epoch": 1.0853886535421202,
      "grad_norm": 0.05346961319446564,
      "learning_rate": 5.127486611235094e-06,
      "loss": 0.0013,
      "step": 169230
    },
    {
      "epoch": 1.0854527904359064,
      "grad_norm": 0.11817878484725952,
      "learning_rate": 5.1269270890278186e-06,
      "loss": 0.0017,
      "step": 169240
    },
    {
      "epoch": 1.0855169273296925,
      "grad_norm": 0.11961852759122849,
      "learning_rate": 5.126367565230057e-06,
      "loss": 0.0021,
      "step": 169250
    },
    {
      "epoch": 1.0855810642234787,
      "grad_norm": 0.02972625009715557,
      "learning_rate": 5.125808039848819e-06,
      "loss": 0.0015,
      "step": 169260
    },
    {
      "epoch": 1.0856452011172646,
      "grad_norm": 0.23780551552772522,
      "learning_rate": 5.125248512891118e-06,
      "loss": 0.0025,
      "step": 169270
    },
    {
      "epoch": 1.0857093380110507,
      "grad_norm": 0.08383630216121674,
      "learning_rate": 5.124688984363962e-06,
      "loss": 0.0029,
      "step": 169280
    },
    {
      "epoch": 1.0857734749048369,
      "grad_norm": 0.077206090092659,
      "learning_rate": 5.124129454274363e-06,
      "loss": 0.0014,
      "step": 169290
    },
    {
      "epoch": 1.085837611798623,
      "grad_norm": 0.09887336194515228,
      "learning_rate": 5.123569922629335e-06,
      "loss": 0.0016,
      "step": 169300
    },
    {
      "epoch": 1.0859017486924092,
      "grad_norm": 0.16712264716625214,
      "learning_rate": 5.123010389435888e-06,
      "loss": 0.0023,
      "step": 169310
    },
    {
      "epoch": 1.0859658855861951,
      "grad_norm": 0.2610989809036255,
      "learning_rate": 5.122450854701031e-06,
      "loss": 0.0017,
      "step": 169320
    },
    {
      "epoch": 1.0860300224799813,
      "grad_norm": 0.12500953674316406,
      "learning_rate": 5.1218913184317785e-06,
      "loss": 0.0014,
      "step": 169330
    },
    {
      "epoch": 1.0860941593737674,
      "grad_norm": 0.04894835501909256,
      "learning_rate": 5.121331780635139e-06,
      "loss": 0.0018,
      "step": 169340
    },
    {
      "epoch": 1.0861582962675536,
      "grad_norm": 0.0937710553407669,
      "learning_rate": 5.1207722413181275e-06,
      "loss": 0.0017,
      "step": 169350
    },
    {
      "epoch": 1.0862224331613395,
      "grad_norm": 0.13560210168361664,
      "learning_rate": 5.120212700487753e-06,
      "loss": 0.0026,
      "step": 169360
    },
    {
      "epoch": 1.0862865700551256,
      "grad_norm": 0.05050661042332649,
      "learning_rate": 5.119653158151027e-06,
      "loss": 0.0008,
      "step": 169370
    },
    {
      "epoch": 1.0863507069489118,
      "grad_norm": 0.02115233615040779,
      "learning_rate": 5.119093614314963e-06,
      "loss": 0.0023,
      "step": 169380
    },
    {
      "epoch": 1.086414843842698,
      "grad_norm": 0.06858409941196442,
      "learning_rate": 5.118534068986568e-06,
      "loss": 0.0013,
      "step": 169390
    },
    {
      "epoch": 1.0864789807364839,
      "grad_norm": 0.03543127328157425,
      "learning_rate": 5.11797452217286e-06,
      "loss": 0.0013,
      "step": 169400
    },
    {
      "epoch": 1.08654311763027,
      "grad_norm": 0.09301551431417465,
      "learning_rate": 5.1174149738808434e-06,
      "loss": 0.0014,
      "step": 169410
    },
    {
      "epoch": 1.0866072545240562,
      "grad_norm": 0.07186420261859894,
      "learning_rate": 5.116855424117535e-06,
      "loss": 0.0031,
      "step": 169420
    },
    {
      "epoch": 1.0866713914178423,
      "grad_norm": 0.09942799061536789,
      "learning_rate": 5.116295872889944e-06,
      "loss": 0.0009,
      "step": 169430
    },
    {
      "epoch": 1.0867355283116282,
      "grad_norm": 0.0785299763083458,
      "learning_rate": 5.115736320205083e-06,
      "loss": 0.0011,
      "step": 169440
    },
    {
      "epoch": 1.0867996652054144,
      "grad_norm": 0.13406498730182648,
      "learning_rate": 5.115176766069964e-06,
      "loss": 0.0013,
      "step": 169450
    },
    {
      "epoch": 1.0868638020992005,
      "grad_norm": 0.095372274518013,
      "learning_rate": 5.114617210491597e-06,
      "loss": 0.0014,
      "step": 169460
    },
    {
      "epoch": 1.0869279389929867,
      "grad_norm": 0.2556072771549225,
      "learning_rate": 5.114057653476995e-06,
      "loss": 0.0013,
      "step": 169470
    },
    {
      "epoch": 1.0869920758867728,
      "grad_norm": 0.07904695719480515,
      "learning_rate": 5.113498095033167e-06,
      "loss": 0.0012,
      "step": 169480
    },
    {
      "epoch": 1.0870562127805588,
      "grad_norm": 0.054306890815496445,
      "learning_rate": 5.1129385351671276e-06,
      "loss": 0.002,
      "step": 169490
    },
    {
      "epoch": 1.087120349674345,
      "grad_norm": 0.04829462617635727,
      "learning_rate": 5.11237897388589e-06,
      "loss": 0.001,
      "step": 169500
    },
    {
      "epoch": 1.087184486568131,
      "grad_norm": 0.10848630219697952,
      "learning_rate": 5.111819411196462e-06,
      "loss": 0.0011,
      "step": 169510
    },
    {
      "epoch": 1.0872486234619172,
      "grad_norm": 0.06292233616113663,
      "learning_rate": 5.1112598471058565e-06,
      "loss": 0.0022,
      "step": 169520
    },
    {
      "epoch": 1.0873127603557031,
      "grad_norm": 0.0438341461122036,
      "learning_rate": 5.110700281621085e-06,
      "loss": 0.001,
      "step": 169530
    },
    {
      "epoch": 1.0873768972494893,
      "grad_norm": 0.004697320517152548,
      "learning_rate": 5.110140714749161e-06,
      "loss": 0.0012,
      "step": 169540
    },
    {
      "epoch": 1.0874410341432754,
      "grad_norm": 0.09202726185321808,
      "learning_rate": 5.109581146497094e-06,
      "loss": 0.0015,
      "step": 169550
    },
    {
      "epoch": 1.0875051710370616,
      "grad_norm": 0.3498629629611969,
      "learning_rate": 5.109021576871898e-06,
      "loss": 0.0042,
      "step": 169560
    },
    {
      "epoch": 1.0875693079308477,
      "grad_norm": 0.05647391080856323,
      "learning_rate": 5.108462005880584e-06,
      "loss": 0.0012,
      "step": 169570
    },
    {
      "epoch": 1.0876334448246336,
      "grad_norm": 0.04639855772256851,
      "learning_rate": 5.107902433530163e-06,
      "loss": 0.0014,
      "step": 169580
    },
    {
      "epoch": 1.0876975817184198,
      "grad_norm": 0.1349206417798996,
      "learning_rate": 5.107342859827648e-06,
      "loss": 0.0021,
      "step": 169590
    },
    {
      "epoch": 1.087761718612206,
      "grad_norm": 0.028598807752132416,
      "learning_rate": 5.1067832847800495e-06,
      "loss": 0.0024,
      "step": 169600
    },
    {
      "epoch": 1.087825855505992,
      "grad_norm": 0.07462950795888901,
      "learning_rate": 5.10622370839438e-06,
      "loss": 0.0017,
      "step": 169610
    },
    {
      "epoch": 1.087889992399778,
      "grad_norm": 0.11065831035375595,
      "learning_rate": 5.105664130677651e-06,
      "loss": 0.0025,
      "step": 169620
    },
    {
      "epoch": 1.0879541292935642,
      "grad_norm": 0.15748350322246552,
      "learning_rate": 5.105104551636877e-06,
      "loss": 0.0025,
      "step": 169630
    },
    {
      "epoch": 1.0880182661873503,
      "grad_norm": 0.09333031624555588,
      "learning_rate": 5.104544971279067e-06,
      "loss": 0.0018,
      "step": 169640
    },
    {
      "epoch": 1.0880824030811365,
      "grad_norm": 0.15592175722122192,
      "learning_rate": 5.103985389611233e-06,
      "loss": 0.0031,
      "step": 169650
    },
    {
      "epoch": 1.0881465399749224,
      "grad_norm": 0.12185309827327728,
      "learning_rate": 5.103425806640387e-06,
      "loss": 0.0017,
      "step": 169660
    },
    {
      "epoch": 1.0882106768687085,
      "grad_norm": 0.11560821533203125,
      "learning_rate": 5.102866222373544e-06,
      "loss": 0.0012,
      "step": 169670
    },
    {
      "epoch": 1.0882748137624947,
      "grad_norm": 0.038899004459381104,
      "learning_rate": 5.102306636817712e-06,
      "loss": 0.0017,
      "step": 169680
    },
    {
      "epoch": 1.0883389506562808,
      "grad_norm": 0.053067803382873535,
      "learning_rate": 5.101747049979906e-06,
      "loss": 0.0025,
      "step": 169690
    },
    {
      "epoch": 1.0884030875500668,
      "grad_norm": 0.081266388297081,
      "learning_rate": 5.101187461867135e-06,
      "loss": 0.0018,
      "step": 169700
    },
    {
      "epoch": 1.088467224443853,
      "grad_norm": 0.2550745904445648,
      "learning_rate": 5.100627872486415e-06,
      "loss": 0.0017,
      "step": 169710
    },
    {
      "epoch": 1.088531361337639,
      "grad_norm": 0.1860969513654709,
      "learning_rate": 5.1000682818447546e-06,
      "loss": 0.0013,
      "step": 169720
    },
    {
      "epoch": 1.0885954982314252,
      "grad_norm": 0.4227175712585449,
      "learning_rate": 5.099508689949166e-06,
      "loss": 0.0027,
      "step": 169730
    },
    {
      "epoch": 1.0886596351252114,
      "grad_norm": 0.16643047332763672,
      "learning_rate": 5.098949096806665e-06,
      "loss": 0.0025,
      "step": 169740
    },
    {
      "epoch": 1.0887237720189973,
      "grad_norm": 0.20546631515026093,
      "learning_rate": 5.098389502424259e-06,
      "loss": 0.0015,
      "step": 169750
    },
    {
      "epoch": 1.0887879089127834,
      "grad_norm": 0.024851031601428986,
      "learning_rate": 5.0978299068089635e-06,
      "loss": 0.0016,
      "step": 169760
    },
    {
      "epoch": 1.0888520458065696,
      "grad_norm": 0.4731219708919525,
      "learning_rate": 5.097270309967788e-06,
      "loss": 0.0039,
      "step": 169770
    },
    {
      "epoch": 1.0889161827003557,
      "grad_norm": 0.14127282798290253,
      "learning_rate": 5.096710711907748e-06,
      "loss": 0.002,
      "step": 169780
    },
    {
      "epoch": 1.0889803195941417,
      "grad_norm": 0.023380475118756294,
      "learning_rate": 5.096151112635851e-06,
      "loss": 0.0015,
      "step": 169790
    },
    {
      "epoch": 1.0890444564879278,
      "grad_norm": 0.14559754729270935,
      "learning_rate": 5.095591512159114e-06,
      "loss": 0.0017,
      "step": 169800
    },
    {
      "epoch": 1.089108593381714,
      "grad_norm": 0.04824541509151459,
      "learning_rate": 5.0950319104845455e-06,
      "loss": 0.0013,
      "step": 169810
    },
    {
      "epoch": 1.0891727302755,
      "grad_norm": 0.068598173558712,
      "learning_rate": 5.0944723076191605e-06,
      "loss": 0.0025,
      "step": 169820
    },
    {
      "epoch": 1.089236867169286,
      "grad_norm": 0.022282173857092857,
      "learning_rate": 5.09391270356997e-06,
      "loss": 0.0017,
      "step": 169830
    },
    {
      "epoch": 1.0893010040630722,
      "grad_norm": 0.08425862342119217,
      "learning_rate": 5.093353098343986e-06,
      "loss": 0.0033,
      "step": 169840
    },
    {
      "epoch": 1.0893651409568583,
      "grad_norm": 0.09538999944925308,
      "learning_rate": 5.09279349194822e-06,
      "loss": 0.0016,
      "step": 169850
    },
    {
      "epoch": 1.0894292778506445,
      "grad_norm": 0.16932521760463715,
      "learning_rate": 5.092233884389685e-06,
      "loss": 0.0021,
      "step": 169860
    },
    {
      "epoch": 1.0894934147444304,
      "grad_norm": 0.17279605567455292,
      "learning_rate": 5.0916742756753934e-06,
      "loss": 0.0014,
      "step": 169870
    },
    {
      "epoch": 1.0895575516382165,
      "grad_norm": 0.09527702629566193,
      "learning_rate": 5.09111466581236e-06,
      "loss": 0.001,
      "step": 169880
    },
    {
      "epoch": 1.0896216885320027,
      "grad_norm": 0.02729853428900242,
      "learning_rate": 5.090555054807592e-06,
      "loss": 0.0011,
      "step": 169890
    },
    {
      "epoch": 1.0896858254257888,
      "grad_norm": 0.042803362011909485,
      "learning_rate": 5.089995442668106e-06,
      "loss": 0.0021,
      "step": 169900
    },
    {
      "epoch": 1.089749962319575,
      "grad_norm": 0.03751653805375099,
      "learning_rate": 5.089435829400913e-06,
      "loss": 0.0012,
      "step": 169910
    },
    {
      "epoch": 1.089814099213361,
      "grad_norm": 0.10222543776035309,
      "learning_rate": 5.088876215013024e-06,
      "loss": 0.0014,
      "step": 169920
    },
    {
      "epoch": 1.089878236107147,
      "grad_norm": 0.04198543727397919,
      "learning_rate": 5.0883165995114536e-06,
      "loss": 0.002,
      "step": 169930
    },
    {
      "epoch": 1.0899423730009332,
      "grad_norm": 0.05111617222428322,
      "learning_rate": 5.087756982903212e-06,
      "loss": 0.0036,
      "step": 169940
    },
    {
      "epoch": 1.0900065098947194,
      "grad_norm": 0.16573524475097656,
      "learning_rate": 5.087197365195315e-06,
      "loss": 0.0035,
      "step": 169950
    },
    {
      "epoch": 1.0900706467885053,
      "grad_norm": 0.16933542490005493,
      "learning_rate": 5.08663774639477e-06,
      "loss": 0.0023,
      "step": 169960
    },
    {
      "epoch": 1.0901347836822914,
      "grad_norm": 0.2192375808954239,
      "learning_rate": 5.086078126508593e-06,
      "loss": 0.0027,
      "step": 169970
    },
    {
      "epoch": 1.0901989205760776,
      "grad_norm": 0.008624457754194736,
      "learning_rate": 5.085518505543796e-06,
      "loss": 0.0021,
      "step": 169980
    },
    {
      "epoch": 1.0902630574698637,
      "grad_norm": 0.004744633100926876,
      "learning_rate": 5.084958883507392e-06,
      "loss": 0.0015,
      "step": 169990
    },
    {
      "epoch": 1.0903271943636499,
      "grad_norm": 0.13655802607536316,
      "learning_rate": 5.08439926040639e-06,
      "loss": 0.0014,
      "step": 170000
    },
    {
      "epoch": 1.0903913312574358,
      "grad_norm": 0.03899003937840462,
      "learning_rate": 5.083839636247807e-06,
      "loss": 0.0022,
      "step": 170010
    },
    {
      "epoch": 1.090455468151222,
      "grad_norm": 0.08106729388237,
      "learning_rate": 5.0832800110386515e-06,
      "loss": 0.0018,
      "step": 170020
    },
    {
      "epoch": 1.0905196050450081,
      "grad_norm": 0.08407328277826309,
      "learning_rate": 5.08272038478594e-06,
      "loss": 0.0011,
      "step": 170030
    },
    {
      "epoch": 1.0905837419387943,
      "grad_norm": 0.05661727860569954,
      "learning_rate": 5.0821607574966816e-06,
      "loss": 0.0008,
      "step": 170040
    },
    {
      "epoch": 1.0906478788325802,
      "grad_norm": 0.044677119702100754,
      "learning_rate": 5.081601129177891e-06,
      "loss": 0.0028,
      "step": 170050
    },
    {
      "epoch": 1.0907120157263663,
      "grad_norm": 0.0651017352938652,
      "learning_rate": 5.081041499836579e-06,
      "loss": 0.002,
      "step": 170060
    },
    {
      "epoch": 1.0907761526201525,
      "grad_norm": 0.11894354224205017,
      "learning_rate": 5.0804818694797595e-06,
      "loss": 0.0019,
      "step": 170070
    },
    {
      "epoch": 1.0908402895139386,
      "grad_norm": 0.11313174664974213,
      "learning_rate": 5.079922238114445e-06,
      "loss": 0.0018,
      "step": 170080
    },
    {
      "epoch": 1.0909044264077246,
      "grad_norm": 0.08070102334022522,
      "learning_rate": 5.079362605747648e-06,
      "loss": 0.0012,
      "step": 170090
    },
    {
      "epoch": 1.0909685633015107,
      "grad_norm": 0.13334757089614868,
      "learning_rate": 5.07880297238638e-06,
      "loss": 0.0016,
      "step": 170100
    },
    {
      "epoch": 1.0910327001952969,
      "grad_norm": 0.052368223667144775,
      "learning_rate": 5.078243338037654e-06,
      "loss": 0.0016,
      "step": 170110
    },
    {
      "epoch": 1.091096837089083,
      "grad_norm": 0.19929657876491547,
      "learning_rate": 5.077683702708485e-06,
      "loss": 0.001,
      "step": 170120
    },
    {
      "epoch": 1.091160973982869,
      "grad_norm": 0.1845984309911728,
      "learning_rate": 5.077124066405882e-06,
      "loss": 0.0022,
      "step": 170130
    },
    {
      "epoch": 1.091225110876655,
      "grad_norm": 0.09705594182014465,
      "learning_rate": 5.07656442913686e-06,
      "loss": 0.0027,
      "step": 170140
    },
    {
      "epoch": 1.0912892477704412,
      "grad_norm": 0.04924891144037247,
      "learning_rate": 5.076004790908431e-06,
      "loss": 0.0036,
      "step": 170150
    },
    {
      "epoch": 1.0913533846642274,
      "grad_norm": 0.02453775703907013,
      "learning_rate": 5.075445151727607e-06,
      "loss": 0.0014,
      "step": 170160
    },
    {
      "epoch": 1.0914175215580135,
      "grad_norm": 0.0791323110461235,
      "learning_rate": 5.074885511601402e-06,
      "loss": 0.0012,
      "step": 170170
    },
    {
      "epoch": 1.0914816584517995,
      "grad_norm": 0.16462315618991852,
      "learning_rate": 5.074325870536828e-06,
      "loss": 0.0022,
      "step": 170180
    },
    {
      "epoch": 1.0915457953455856,
      "grad_norm": 0.01788683421909809,
      "learning_rate": 5.073766228540897e-06,
      "loss": 0.0011,
      "step": 170190
    },
    {
      "epoch": 1.0916099322393717,
      "grad_norm": 0.08107317239046097,
      "learning_rate": 5.073206585620622e-06,
      "loss": 0.0012,
      "step": 170200
    },
    {
      "epoch": 1.091674069133158,
      "grad_norm": 0.05761834979057312,
      "learning_rate": 5.072646941783018e-06,
      "loss": 0.0022,
      "step": 170210
    },
    {
      "epoch": 1.0917382060269438,
      "grad_norm": 0.010808841325342655,
      "learning_rate": 5.072087297035095e-06,
      "loss": 0.0016,
      "step": 170220
    },
    {
      "epoch": 1.09180234292073,
      "grad_norm": 0.08491594344377518,
      "learning_rate": 5.071527651383867e-06,
      "loss": 0.0025,
      "step": 170230
    },
    {
      "epoch": 1.0918664798145161,
      "grad_norm": 0.43619436025619507,
      "learning_rate": 5.0709680048363466e-06,
      "loss": 0.0024,
      "step": 170240
    },
    {
      "epoch": 1.0919306167083023,
      "grad_norm": 0.10487474501132965,
      "learning_rate": 5.070408357399544e-06,
      "loss": 0.0018,
      "step": 170250
    },
    {
      "epoch": 1.0919947536020882,
      "grad_norm": 0.06518199294805527,
      "learning_rate": 5.069848709080476e-06,
      "loss": 0.0023,
      "step": 170260
    },
    {
      "epoch": 1.0920588904958743,
      "grad_norm": 0.08357866108417511,
      "learning_rate": 5.069289059886156e-06,
      "loss": 0.0021,
      "step": 170270
    },
    {
      "epoch": 1.0921230273896605,
      "grad_norm": 0.0606221966445446,
      "learning_rate": 5.068729409823591e-06,
      "loss": 0.0013,
      "step": 170280
    },
    {
      "epoch": 1.0921871642834466,
      "grad_norm": 0.136131152510643,
      "learning_rate": 5.0681697588997995e-06,
      "loss": 0.0033,
      "step": 170290
    },
    {
      "epoch": 1.0922513011772326,
      "grad_norm": 0.1497824490070343,
      "learning_rate": 5.0676101071217905e-06,
      "loss": 0.0013,
      "step": 170300
    },
    {
      "epoch": 1.0923154380710187,
      "grad_norm": 0.012980225495994091,
      "learning_rate": 5.067050454496582e-06,
      "loss": 0.001,
      "step": 170310
    },
    {
      "epoch": 1.0923795749648049,
      "grad_norm": 0.13039623200893402,
      "learning_rate": 5.06649080103118e-06,
      "loss": 0.0012,
      "step": 170320
    },
    {
      "epoch": 1.092443711858591,
      "grad_norm": 0.2750491201877594,
      "learning_rate": 5.065931146732602e-06,
      "loss": 0.0028,
      "step": 170330
    },
    {
      "epoch": 1.0925078487523772,
      "grad_norm": 0.20053251087665558,
      "learning_rate": 5.065371491607859e-06,
      "loss": 0.0012,
      "step": 170340
    },
    {
      "epoch": 1.092571985646163,
      "grad_norm": 0.106719009578228,
      "learning_rate": 5.0648118356639654e-06,
      "loss": 0.0015,
      "step": 170350
    },
    {
      "epoch": 1.0926361225399492,
      "grad_norm": 0.08977479487657547,
      "learning_rate": 5.064252178907933e-06,
      "loss": 0.0029,
      "step": 170360
    },
    {
      "epoch": 1.0927002594337354,
      "grad_norm": 0.10772501677274704,
      "learning_rate": 5.063692521346775e-06,
      "loss": 0.0041,
      "step": 170370
    },
    {
      "epoch": 1.0927643963275215,
      "grad_norm": 0.001853933441452682,
      "learning_rate": 5.063132862987502e-06,
      "loss": 0.0029,
      "step": 170380
    },
    {
      "epoch": 1.0928285332213075,
      "grad_norm": 0.031118426471948624,
      "learning_rate": 5.062573203837132e-06,
      "loss": 0.0011,
      "step": 170390
    },
    {
      "epoch": 1.0928926701150936,
      "grad_norm": 0.08649729937314987,
      "learning_rate": 5.062013543902673e-06,
      "loss": 0.0016,
      "step": 170400
    },
    {
      "epoch": 1.0929568070088798,
      "grad_norm": 0.05934537202119827,
      "learning_rate": 5.061453883191141e-06,
      "loss": 0.0018,
      "step": 170410
    },
    {
      "epoch": 1.093020943902666,
      "grad_norm": 0.05794312432408333,
      "learning_rate": 5.060894221709548e-06,
      "loss": 0.0017,
      "step": 170420
    },
    {
      "epoch": 1.093085080796452,
      "grad_norm": 0.06355617940425873,
      "learning_rate": 5.060334559464905e-06,
      "loss": 0.0013,
      "step": 170430
    },
    {
      "epoch": 1.093149217690238,
      "grad_norm": 0.07892221957445145,
      "learning_rate": 5.05977489646423e-06,
      "loss": 0.001,
      "step": 170440
    },
    {
      "epoch": 1.0932133545840241,
      "grad_norm": 0.13034789264202118,
      "learning_rate": 5.0592152327145316e-06,
      "loss": 0.0015,
      "step": 170450
    },
    {
      "epoch": 1.0932774914778103,
      "grad_norm": 0.07530619204044342,
      "learning_rate": 5.058655568222823e-06,
      "loss": 0.0016,
      "step": 170460
    },
    {
      "epoch": 1.0933416283715964,
      "grad_norm": 0.11381381005048752,
      "learning_rate": 5.05809590299612e-06,
      "loss": 0.0016,
      "step": 170470
    },
    {
      "epoch": 1.0934057652653824,
      "grad_norm": 0.24479657411575317,
      "learning_rate": 5.057536237041433e-06,
      "loss": 0.0037,
      "step": 170480
    },
    {
      "epoch": 1.0934699021591685,
      "grad_norm": 0.04012688249349594,
      "learning_rate": 5.056976570365775e-06,
      "loss": 0.0019,
      "step": 170490
    },
    {
      "epoch": 1.0935340390529547,
      "grad_norm": 0.08466518670320511,
      "learning_rate": 5.0564169029761615e-06,
      "loss": 0.0008,
      "step": 170500
    },
    {
      "epoch": 1.0935981759467408,
      "grad_norm": 0.16974368691444397,
      "learning_rate": 5.055857234879602e-06,
      "loss": 0.0014,
      "step": 170510
    },
    {
      "epoch": 1.0936623128405267,
      "grad_norm": 0.027867648750543594,
      "learning_rate": 5.055297566083113e-06,
      "loss": 0.0012,
      "step": 170520
    },
    {
      "epoch": 1.0937264497343129,
      "grad_norm": 0.4133170545101166,
      "learning_rate": 5.054737896593704e-06,
      "loss": 0.0035,
      "step": 170530
    },
    {
      "epoch": 1.093790586628099,
      "grad_norm": 0.07212081551551819,
      "learning_rate": 5.054178226418392e-06,
      "loss": 0.0015,
      "step": 170540
    },
    {
      "epoch": 1.0938547235218852,
      "grad_norm": 0.135950967669487,
      "learning_rate": 5.053618555564187e-06,
      "loss": 0.001,
      "step": 170550
    },
    {
      "epoch": 1.093918860415671,
      "grad_norm": 0.2264435589313507,
      "learning_rate": 5.053058884038103e-06,
      "loss": 0.0021,
      "step": 170560
    },
    {
      "epoch": 1.0939829973094573,
      "grad_norm": 0.0787590816617012,
      "learning_rate": 5.052499211847155e-06,
      "loss": 0.0016,
      "step": 170570
    },
    {
      "epoch": 1.0940471342032434,
      "grad_norm": 0.005399610847234726,
      "learning_rate": 5.051939538998352e-06,
      "loss": 0.0014,
      "step": 170580
    },
    {
      "epoch": 1.0941112710970295,
      "grad_norm": 0.057541925460100174,
      "learning_rate": 5.0513798654987116e-06,
      "loss": 0.0014,
      "step": 170590
    },
    {
      "epoch": 1.0941754079908157,
      "grad_norm": 0.11907067894935608,
      "learning_rate": 5.050820191355243e-06,
      "loss": 0.0012,
      "step": 170600
    },
    {
      "epoch": 1.0942395448846016,
      "grad_norm": 0.16507697105407715,
      "learning_rate": 5.050260516574963e-06,
      "loss": 0.0016,
      "step": 170610
    },
    {
      "epoch": 1.0943036817783878,
      "grad_norm": 0.06532784551382065,
      "learning_rate": 5.049700841164881e-06,
      "loss": 0.0008,
      "step": 170620
    },
    {
      "epoch": 1.094367818672174,
      "grad_norm": 0.06784983724355698,
      "learning_rate": 5.049141165132012e-06,
      "loss": 0.0026,
      "step": 170630
    },
    {
      "epoch": 1.09443195556596,
      "grad_norm": 0.04982329532504082,
      "learning_rate": 5.0485814884833695e-06,
      "loss": 0.0022,
      "step": 170640
    },
    {
      "epoch": 1.094496092459746,
      "grad_norm": 0.18449866771697998,
      "learning_rate": 5.048021811225966e-06,
      "loss": 0.0023,
      "step": 170650
    },
    {
      "epoch": 1.0945602293535321,
      "grad_norm": 0.3333454430103302,
      "learning_rate": 5.047462133366815e-06,
      "loss": 0.0021,
      "step": 170660
    },
    {
      "epoch": 1.0946243662473183,
      "grad_norm": 0.13253095746040344,
      "learning_rate": 5.046902454912931e-06,
      "loss": 0.0008,
      "step": 170670
    },
    {
      "epoch": 1.0946885031411044,
      "grad_norm": 0.05479660630226135,
      "learning_rate": 5.0463427758713234e-06,
      "loss": 0.0013,
      "step": 170680
    },
    {
      "epoch": 1.0947526400348904,
      "grad_norm": 0.10303059965372086,
      "learning_rate": 5.04578309624901e-06,
      "loss": 0.0024,
      "step": 170690
    },
    {
      "epoch": 1.0948167769286765,
      "grad_norm": 0.06154974550008774,
      "learning_rate": 5.045223416053e-06,
      "loss": 0.0008,
      "step": 170700
    },
    {
      "epoch": 1.0948809138224627,
      "grad_norm": 0.06780397891998291,
      "learning_rate": 5.044663735290309e-06,
      "loss": 0.0031,
      "step": 170710
    },
    {
      "epoch": 1.0949450507162488,
      "grad_norm": 0.23820923268795013,
      "learning_rate": 5.04410405396795e-06,
      "loss": 0.0024,
      "step": 170720
    },
    {
      "epoch": 1.095009187610035,
      "grad_norm": 0.1282614916563034,
      "learning_rate": 5.0435443720929344e-06,
      "loss": 0.0013,
      "step": 170730
    },
    {
      "epoch": 1.0950733245038209,
      "grad_norm": 0.060902614146471024,
      "learning_rate": 5.042984689672278e-06,
      "loss": 0.0022,
      "step": 170740
    },
    {
      "epoch": 1.095137461397607,
      "grad_norm": 0.21891924738883972,
      "learning_rate": 5.042425006712992e-06,
      "loss": 0.0014,
      "step": 170750
    },
    {
      "epoch": 1.0952015982913932,
      "grad_norm": 0.014451575465500355,
      "learning_rate": 5.041865323222089e-06,
      "loss": 0.0004,
      "step": 170760
    },
    {
      "epoch": 1.0952657351851793,
      "grad_norm": 0.044007159769535065,
      "learning_rate": 5.041305639206585e-06,
      "loss": 0.0006,
      "step": 170770
    },
    {
      "epoch": 1.0953298720789653,
      "grad_norm": 0.0831480324268341,
      "learning_rate": 5.040745954673492e-06,
      "loss": 0.002,
      "step": 170780
    },
    {
      "epoch": 1.0953940089727514,
      "grad_norm": 0.13365034759044647,
      "learning_rate": 5.040186269629823e-06,
      "loss": 0.0015,
      "step": 170790
    },
    {
      "epoch": 1.0954581458665376,
      "grad_norm": 0.19277553260326385,
      "learning_rate": 5.039626584082592e-06,
      "loss": 0.0016,
      "step": 170800
    },
    {
      "epoch": 1.0955222827603237,
      "grad_norm": 0.13959717750549316,
      "learning_rate": 5.03906689803881e-06,
      "loss": 0.0013,
      "step": 170810
    },
    {
      "epoch": 1.0955864196541096,
      "grad_norm": 0.25731614232063293,
      "learning_rate": 5.038507211505493e-06,
      "loss": 0.0015,
      "step": 170820
    },
    {
      "epoch": 1.0956505565478958,
      "grad_norm": 0.07411252707242966,
      "learning_rate": 5.0379475244896525e-06,
      "loss": 0.0025,
      "step": 170830
    },
    {
      "epoch": 1.095714693441682,
      "grad_norm": 0.03390123322606087,
      "learning_rate": 5.0373878369983035e-06,
      "loss": 0.003,
      "step": 170840
    },
    {
      "epoch": 1.095778830335468,
      "grad_norm": 0.036655180156230927,
      "learning_rate": 5.036828149038456e-06,
      "loss": 0.0019,
      "step": 170850
    },
    {
      "epoch": 1.0958429672292542,
      "grad_norm": 0.022341804578900337,
      "learning_rate": 5.036268460617128e-06,
      "loss": 0.002,
      "step": 170860
    },
    {
      "epoch": 1.0959071041230402,
      "grad_norm": 0.12349379062652588,
      "learning_rate": 5.035708771741328e-06,
      "loss": 0.0026,
      "step": 170870
    },
    {
      "epoch": 1.0959712410168263,
      "grad_norm": 0.02770046703517437,
      "learning_rate": 5.035149082418074e-06,
      "loss": 0.001,
      "step": 170880
    },
    {
      "epoch": 1.0960353779106125,
      "grad_norm": 0.10235182195901871,
      "learning_rate": 5.034589392654374e-06,
      "loss": 0.0016,
      "step": 170890
    },
    {
      "epoch": 1.0960995148043986,
      "grad_norm": 0.0583653599023819,
      "learning_rate": 5.034029702457246e-06,
      "loss": 0.0013,
      "step": 170900
    },
    {
      "epoch": 1.0961636516981845,
      "grad_norm": 0.09971941262483597,
      "learning_rate": 5.033470011833701e-06,
      "loss": 0.0006,
      "step": 170910
    },
    {
      "epoch": 1.0962277885919707,
      "grad_norm": 0.16799309849739075,
      "learning_rate": 5.0329103207907526e-06,
      "loss": 0.0019,
      "step": 170920
    },
    {
      "epoch": 1.0962919254857568,
      "grad_norm": 0.054120440036058426,
      "learning_rate": 5.0323506293354156e-06,
      "loss": 0.0013,
      "step": 170930
    },
    {
      "epoch": 1.096356062379543,
      "grad_norm": 0.01602933183312416,
      "learning_rate": 5.031790937474701e-06,
      "loss": 0.0017,
      "step": 170940
    },
    {
      "epoch": 1.096420199273329,
      "grad_norm": 0.03627896308898926,
      "learning_rate": 5.031231245215623e-06,
      "loss": 0.002,
      "step": 170950
    },
    {
      "epoch": 1.096484336167115,
      "grad_norm": 0.11360101401805878,
      "learning_rate": 5.030671552565196e-06,
      "loss": 0.0011,
      "step": 170960
    },
    {
      "epoch": 1.0965484730609012,
      "grad_norm": 0.1410820335149765,
      "learning_rate": 5.030111859530432e-06,
      "loss": 0.0021,
      "step": 170970
    },
    {
      "epoch": 1.0966126099546873,
      "grad_norm": 0.15676705539226532,
      "learning_rate": 5.029552166118345e-06,
      "loss": 0.0016,
      "step": 170980
    },
    {
      "epoch": 1.0966767468484733,
      "grad_norm": 0.03369128704071045,
      "learning_rate": 5.028992472335949e-06,
      "loss": 0.0019,
      "step": 170990
    },
    {
      "epoch": 1.0967408837422594,
      "grad_norm": 0.05345707759261131,
      "learning_rate": 5.0284327781902556e-06,
      "loss": 0.0019,
      "step": 171000
    },
    {
      "epoch": 1.0968050206360456,
      "grad_norm": 0.3105112314224243,
      "learning_rate": 5.027873083688279e-06,
      "loss": 0.0029,
      "step": 171010
    },
    {
      "epoch": 1.0968691575298317,
      "grad_norm": 0.11570075154304504,
      "learning_rate": 5.027313388837032e-06,
      "loss": 0.0017,
      "step": 171020
    },
    {
      "epoch": 1.0969332944236179,
      "grad_norm": 0.025331120938062668,
      "learning_rate": 5.026753693643531e-06,
      "loss": 0.0015,
      "step": 171030
    },
    {
      "epoch": 1.0969974313174038,
      "grad_norm": 0.07241302728652954,
      "learning_rate": 5.0261939981147855e-06,
      "loss": 0.0012,
      "step": 171040
    },
    {
      "epoch": 1.09706156821119,
      "grad_norm": 0.08699067682027817,
      "learning_rate": 5.025634302257811e-06,
      "loss": 0.0016,
      "step": 171050
    },
    {
      "epoch": 1.097125705104976,
      "grad_norm": 0.04733594134449959,
      "learning_rate": 5.025074606079621e-06,
      "loss": 0.0016,
      "step": 171060
    },
    {
      "epoch": 1.0971898419987622,
      "grad_norm": 0.22508540749549866,
      "learning_rate": 5.024514909587228e-06,
      "loss": 0.0018,
      "step": 171070
    },
    {
      "epoch": 1.0972539788925482,
      "grad_norm": 0.09042717516422272,
      "learning_rate": 5.023955212787644e-06,
      "loss": 0.0011,
      "step": 171080
    },
    {
      "epoch": 1.0973181157863343,
      "grad_norm": 0.2919725477695465,
      "learning_rate": 5.0233955156878864e-06,
      "loss": 0.0016,
      "step": 171090
    },
    {
      "epoch": 1.0973822526801205,
      "grad_norm": 0.07168257981538773,
      "learning_rate": 5.0228358182949645e-06,
      "loss": 0.0014,
      "step": 171100
    },
    {
      "epoch": 1.0974463895739066,
      "grad_norm": 0.10578005760908127,
      "learning_rate": 5.0222761206158945e-06,
      "loss": 0.0037,
      "step": 171110
    },
    {
      "epoch": 1.0975105264676928,
      "grad_norm": 0.029910454526543617,
      "learning_rate": 5.021716422657688e-06,
      "loss": 0.0009,
      "step": 171120
    },
    {
      "epoch": 1.0975746633614787,
      "grad_norm": 0.033341534435749054,
      "learning_rate": 5.02115672442736e-06,
      "loss": 0.0037,
      "step": 171130
    },
    {
      "epoch": 1.0976388002552648,
      "grad_norm": 0.09771548211574554,
      "learning_rate": 5.020597025931923e-06,
      "loss": 0.0032,
      "step": 171140
    },
    {
      "epoch": 1.097702937149051,
      "grad_norm": 0.03211095929145813,
      "learning_rate": 5.02003732717839e-06,
      "loss": 0.0015,
      "step": 171150
    },
    {
      "epoch": 1.0977670740428371,
      "grad_norm": 0.04773737117648125,
      "learning_rate": 5.019477628173775e-06,
      "loss": 0.002,
      "step": 171160
    },
    {
      "epoch": 1.097831210936623,
      "grad_norm": 0.03710533678531647,
      "learning_rate": 5.018917928925092e-06,
      "loss": 0.0013,
      "step": 171170
    },
    {
      "epoch": 1.0978953478304092,
      "grad_norm": 0.6000286936759949,
      "learning_rate": 5.018358229439354e-06,
      "loss": 0.0025,
      "step": 171180
    },
    {
      "epoch": 1.0979594847241954,
      "grad_norm": 0.013128839433193207,
      "learning_rate": 5.017798529723573e-06,
      "loss": 0.0015,
      "step": 171190
    },
    {
      "epoch": 1.0980236216179815,
      "grad_norm": 0.01816055364906788,
      "learning_rate": 5.017238829784766e-06,
      "loss": 0.0015,
      "step": 171200
    },
    {
      "epoch": 1.0980877585117674,
      "grad_norm": 0.0966353639960289,
      "learning_rate": 5.016679129629942e-06,
      "loss": 0.0016,
      "step": 171210
    },
    {
      "epoch": 1.0981518954055536,
      "grad_norm": 0.18456101417541504,
      "learning_rate": 5.016119429266118e-06,
      "loss": 0.0019,
      "step": 171220
    },
    {
      "epoch": 1.0982160322993397,
      "grad_norm": 0.07079457491636276,
      "learning_rate": 5.015559728700304e-06,
      "loss": 0.0029,
      "step": 171230
    },
    {
      "epoch": 1.0982801691931259,
      "grad_norm": 0.0704764872789383,
      "learning_rate": 5.015000027939518e-06,
      "loss": 0.0018,
      "step": 171240
    },
    {
      "epoch": 1.0983443060869118,
      "grad_norm": 0.14907720685005188,
      "learning_rate": 5.014440326990769e-06,
      "loss": 0.0025,
      "step": 171250
    },
    {
      "epoch": 1.098408442980698,
      "grad_norm": 0.049817949533462524,
      "learning_rate": 5.013880625861072e-06,
      "loss": 0.0012,
      "step": 171260
    },
    {
      "epoch": 1.098472579874484,
      "grad_norm": 0.08189506828784943,
      "learning_rate": 5.013320924557443e-06,
      "loss": 0.0026,
      "step": 171270
    },
    {
      "epoch": 1.0985367167682702,
      "grad_norm": 0.13400381803512573,
      "learning_rate": 5.012761223086891e-06,
      "loss": 0.002,
      "step": 171280
    },
    {
      "epoch": 1.0986008536620564,
      "grad_norm": 0.034556783735752106,
      "learning_rate": 5.012201521456435e-06,
      "loss": 0.0018,
      "step": 171290
    },
    {
      "epoch": 1.0986649905558423,
      "grad_norm": 0.026275919750332832,
      "learning_rate": 5.011641819673082e-06,
      "loss": 0.0025,
      "step": 171300
    },
    {
      "epoch": 1.0987291274496285,
      "grad_norm": 0.06027844920754433,
      "learning_rate": 5.01108211774385e-06,
      "loss": 0.001,
      "step": 171310
    },
    {
      "epoch": 1.0987932643434146,
      "grad_norm": 0.07787550240755081,
      "learning_rate": 5.010522415675751e-06,
      "loss": 0.0012,
      "step": 171320
    },
    {
      "epoch": 1.0988574012372008,
      "grad_norm": 0.10380320250988007,
      "learning_rate": 5.009962713475799e-06,
      "loss": 0.0009,
      "step": 171330
    },
    {
      "epoch": 1.0989215381309867,
      "grad_norm": 0.20597901940345764,
      "learning_rate": 5.009403011151006e-06,
      "loss": 0.0038,
      "step": 171340
    },
    {
      "epoch": 1.0989856750247728,
      "grad_norm": 0.05098361149430275,
      "learning_rate": 5.008843308708389e-06,
      "loss": 0.0012,
      "step": 171350
    },
    {
      "epoch": 1.099049811918559,
      "grad_norm": 0.03223688155412674,
      "learning_rate": 5.008283606154957e-06,
      "loss": 0.0007,
      "step": 171360
    },
    {
      "epoch": 1.0991139488123451,
      "grad_norm": 0.023974584415555,
      "learning_rate": 5.007723903497725e-06,
      "loss": 0.0015,
      "step": 171370
    },
    {
      "epoch": 1.099178085706131,
      "grad_norm": 0.03929032385349274,
      "learning_rate": 5.007164200743708e-06,
      "loss": 0.0029,
      "step": 171380
    },
    {
      "epoch": 1.0992422225999172,
      "grad_norm": 0.06479243189096451,
      "learning_rate": 5.0066044978999185e-06,
      "loss": 0.002,
      "step": 171390
    },
    {
      "epoch": 1.0993063594937034,
      "grad_norm": 0.09911324083805084,
      "learning_rate": 5.006044794973368e-06,
      "loss": 0.0013,
      "step": 171400
    },
    {
      "epoch": 1.0993704963874895,
      "grad_norm": 0.09658053517341614,
      "learning_rate": 5.005485091971074e-06,
      "loss": 0.0008,
      "step": 171410
    },
    {
      "epoch": 1.0994346332812754,
      "grad_norm": 0.0684509128332138,
      "learning_rate": 5.004925388900047e-06,
      "loss": 0.0033,
      "step": 171420
    },
    {
      "epoch": 1.0994987701750616,
      "grad_norm": 0.06243107095360756,
      "learning_rate": 5.004365685767303e-06,
      "loss": 0.0022,
      "step": 171430
    },
    {
      "epoch": 1.0995629070688477,
      "grad_norm": 0.17232149839401245,
      "learning_rate": 5.003805982579851e-06,
      "loss": 0.0016,
      "step": 171440
    },
    {
      "epoch": 1.0996270439626339,
      "grad_norm": 0.07372458279132843,
      "learning_rate": 5.00324627934471e-06,
      "loss": 0.0018,
      "step": 171450
    },
    {
      "epoch": 1.09969118085642,
      "grad_norm": 0.04588346928358078,
      "learning_rate": 5.002686576068889e-06,
      "loss": 0.0022,
      "step": 171460
    },
    {
      "epoch": 1.099755317750206,
      "grad_norm": 0.04140771925449371,
      "learning_rate": 5.002126872759404e-06,
      "loss": 0.0014,
      "step": 171470
    },
    {
      "epoch": 1.0998194546439921,
      "grad_norm": 0.04907546937465668,
      "learning_rate": 5.001567169423268e-06,
      "loss": 0.0013,
      "step": 171480
    },
    {
      "epoch": 1.0998835915377783,
      "grad_norm": 0.07015169411897659,
      "learning_rate": 5.001007466067493e-06,
      "loss": 0.0016,
      "step": 171490
    },
    {
      "epoch": 1.0999477284315644,
      "grad_norm": 0.0467178113758564,
      "learning_rate": 5.000447762699095e-06,
      "loss": 0.0021,
      "step": 171500
    },
    {
      "epoch": 1.1000118653253503,
      "grad_norm": 0.16069184243679047,
      "learning_rate": 4.999888059325087e-06,
      "loss": 0.0043,
      "step": 171510
    },
    {
      "epoch": 1.1000760022191365,
      "grad_norm": 0.12354295700788498,
      "learning_rate": 4.999328355952481e-06,
      "loss": 0.0028,
      "step": 171520
    },
    {
      "epoch": 1.1001401391129226,
      "grad_norm": 0.04084980860352516,
      "learning_rate": 4.99876865258829e-06,
      "loss": 0.0023,
      "step": 171530
    },
    {
      "epoch": 1.1002042760067088,
      "grad_norm": 0.2395259439945221,
      "learning_rate": 4.998208949239531e-06,
      "loss": 0.0015,
      "step": 171540
    },
    {
      "epoch": 1.100268412900495,
      "grad_norm": 0.14555861055850983,
      "learning_rate": 4.997649245913213e-06,
      "loss": 0.0015,
      "step": 171550
    },
    {
      "epoch": 1.1003325497942809,
      "grad_norm": 0.008490769192576408,
      "learning_rate": 4.9970895426163515e-06,
      "loss": 0.0032,
      "step": 171560
    },
    {
      "epoch": 1.100396686688067,
      "grad_norm": 0.11735638231039047,
      "learning_rate": 4.996529839355961e-06,
      "loss": 0.0016,
      "step": 171570
    },
    {
      "epoch": 1.1004608235818532,
      "grad_norm": 0.058760784566402435,
      "learning_rate": 4.995970136139054e-06,
      "loss": 0.0009,
      "step": 171580
    },
    {
      "epoch": 1.1005249604756393,
      "grad_norm": 0.2136850357055664,
      "learning_rate": 4.995410432972643e-06,
      "loss": 0.003,
      "step": 171590
    },
    {
      "epoch": 1.1005890973694252,
      "grad_norm": 0.09907729923725128,
      "learning_rate": 4.994850729863744e-06,
      "loss": 0.0021,
      "step": 171600
    },
    {
      "epoch": 1.1006532342632114,
      "grad_norm": 0.041656531393527985,
      "learning_rate": 4.994291026819368e-06,
      "loss": 0.0015,
      "step": 171610
    },
    {
      "epoch": 1.1007173711569975,
      "grad_norm": 0.07977510988712311,
      "learning_rate": 4.993731323846531e-06,
      "loss": 0.0016,
      "step": 171620
    },
    {
      "epoch": 1.1007815080507837,
      "grad_norm": 0.011695623397827148,
      "learning_rate": 4.993171620952243e-06,
      "loss": 0.0012,
      "step": 171630
    },
    {
      "epoch": 1.1008456449445696,
      "grad_norm": 0.10950488597154617,
      "learning_rate": 4.992611918143521e-06,
      "loss": 0.001,
      "step": 171640
    },
    {
      "epoch": 1.1009097818383557,
      "grad_norm": 0.06872362643480301,
      "learning_rate": 4.992052215427376e-06,
      "loss": 0.0034,
      "step": 171650
    },
    {
      "epoch": 1.100973918732142,
      "grad_norm": 0.12201790511608124,
      "learning_rate": 4.9914925128108235e-06,
      "loss": 0.0017,
      "step": 171660
    },
    {
      "epoch": 1.101038055625928,
      "grad_norm": 0.2046944946050644,
      "learning_rate": 4.990932810300876e-06,
      "loss": 0.0022,
      "step": 171670
    },
    {
      "epoch": 1.101102192519714,
      "grad_norm": 0.18569254875183105,
      "learning_rate": 4.9903731079045444e-06,
      "loss": 0.0017,
      "step": 171680
    },
    {
      "epoch": 1.1011663294135001,
      "grad_norm": 0.06764160841703415,
      "learning_rate": 4.989813405628848e-06,
      "loss": 0.0012,
      "step": 171690
    },
    {
      "epoch": 1.1012304663072863,
      "grad_norm": 0.06488970667123795,
      "learning_rate": 4.989253703480794e-06,
      "loss": 0.0012,
      "step": 171700
    },
    {
      "epoch": 1.1012946032010724,
      "grad_norm": 0.10774311423301697,
      "learning_rate": 4.988694001467401e-06,
      "loss": 0.0008,
      "step": 171710
    },
    {
      "epoch": 1.1013587400948586,
      "grad_norm": 0.12963354587554932,
      "learning_rate": 4.988134299595679e-06,
      "loss": 0.0009,
      "step": 171720
    },
    {
      "epoch": 1.1014228769886445,
      "grad_norm": 0.12337248772382736,
      "learning_rate": 4.987574597872643e-06,
      "loss": 0.0018,
      "step": 171730
    },
    {
      "epoch": 1.1014870138824306,
      "grad_norm": 0.2637619972229004,
      "learning_rate": 4.987014896305307e-06,
      "loss": 0.002,
      "step": 171740
    },
    {
      "epoch": 1.1015511507762168,
      "grad_norm": 0.13124048709869385,
      "learning_rate": 4.986455194900683e-06,
      "loss": 0.0016,
      "step": 171750
    },
    {
      "epoch": 1.101615287670003,
      "grad_norm": 0.22929047048091888,
      "learning_rate": 4.9858954936657845e-06,
      "loss": 0.0022,
      "step": 171760
    },
    {
      "epoch": 1.1016794245637889,
      "grad_norm": 0.06770718097686768,
      "learning_rate": 4.9853357926076265e-06,
      "loss": 0.0024,
      "step": 171770
    },
    {
      "epoch": 1.101743561457575,
      "grad_norm": 0.1008705347776413,
      "learning_rate": 4.98477609173322e-06,
      "loss": 0.0032,
      "step": 171780
    },
    {
      "epoch": 1.1018076983513612,
      "grad_norm": 0.10125420987606049,
      "learning_rate": 4.984216391049582e-06,
      "loss": 0.0019,
      "step": 171790
    },
    {
      "epoch": 1.1018718352451473,
      "grad_norm": 0.0495884008705616,
      "learning_rate": 4.983656690563722e-06,
      "loss": 0.0034,
      "step": 171800
    },
    {
      "epoch": 1.1019359721389332,
      "grad_norm": 0.006137746851891279,
      "learning_rate": 4.983096990282657e-06,
      "loss": 0.0012,
      "step": 171810
    },
    {
      "epoch": 1.1020001090327194,
      "grad_norm": 0.039007000625133514,
      "learning_rate": 4.982537290213398e-06,
      "loss": 0.001,
      "step": 171820
    },
    {
      "epoch": 1.1020642459265055,
      "grad_norm": 0.01688682846724987,
      "learning_rate": 4.981977590362959e-06,
      "loss": 0.0009,
      "step": 171830
    },
    {
      "epoch": 1.1021283828202917,
      "grad_norm": 0.135686457157135,
      "learning_rate": 4.981417890738354e-06,
      "loss": 0.0037,
      "step": 171840
    },
    {
      "epoch": 1.1021925197140776,
      "grad_norm": 0.0436604842543602,
      "learning_rate": 4.980858191346595e-06,
      "loss": 0.0015,
      "step": 171850
    },
    {
      "epoch": 1.1022566566078638,
      "grad_norm": 0.1291503757238388,
      "learning_rate": 4.980298492194699e-06,
      "loss": 0.0018,
      "step": 171860
    },
    {
      "epoch": 1.10232079350165,
      "grad_norm": 0.27027952671051025,
      "learning_rate": 4.979738793289675e-06,
      "loss": 0.0024,
      "step": 171870
    },
    {
      "epoch": 1.102384930395436,
      "grad_norm": 0.014753194525837898,
      "learning_rate": 4.9791790946385395e-06,
      "loss": 0.0018,
      "step": 171880
    },
    {
      "epoch": 1.1024490672892222,
      "grad_norm": 0.039702508598566055,
      "learning_rate": 4.978619396248304e-06,
      "loss": 0.0015,
      "step": 171890
    },
    {
      "epoch": 1.1025132041830081,
      "grad_norm": 0.08132674545049667,
      "learning_rate": 4.978059698125983e-06,
      "loss": 0.0017,
      "step": 171900
    },
    {
      "epoch": 1.1025773410767943,
      "grad_norm": 0.2327902913093567,
      "learning_rate": 4.977500000278589e-06,
      "loss": 0.0025,
      "step": 171910
    },
    {
      "epoch": 1.1026414779705804,
      "grad_norm": 0.01180790364742279,
      "learning_rate": 4.976940302713138e-06,
      "loss": 0.0019,
      "step": 171920
    },
    {
      "epoch": 1.1027056148643666,
      "grad_norm": 0.08049006760120392,
      "learning_rate": 4.976380605436639e-06,
      "loss": 0.0012,
      "step": 171930
    },
    {
      "epoch": 1.1027697517581525,
      "grad_norm": 0.02945333532989025,
      "learning_rate": 4.97582090845611e-06,
      "loss": 0.0019,
      "step": 171940
    },
    {
      "epoch": 1.1028338886519387,
      "grad_norm": 0.12358007580041885,
      "learning_rate": 4.975261211778561e-06,
      "loss": 0.0008,
      "step": 171950
    },
    {
      "epoch": 1.1028980255457248,
      "grad_norm": 0.1592055857181549,
      "learning_rate": 4.974701515411008e-06,
      "loss": 0.0009,
      "step": 171960
    },
    {
      "epoch": 1.102962162439511,
      "grad_norm": 0.22565963864326477,
      "learning_rate": 4.974141819360461e-06,
      "loss": 0.001,
      "step": 171970
    },
    {
      "epoch": 1.103026299333297,
      "grad_norm": 0.2617228627204895,
      "learning_rate": 4.973582123633936e-06,
      "loss": 0.0016,
      "step": 171980
    },
    {
      "epoch": 1.103090436227083,
      "grad_norm": 0.01660790480673313,
      "learning_rate": 4.9730224282384475e-06,
      "loss": 0.0014,
      "step": 171990
    },
    {
      "epoch": 1.1031545731208692,
      "grad_norm": 0.009054155088961124,
      "learning_rate": 4.972462733181006e-06,
      "loss": 0.0014,
      "step": 172000
    },
    {
      "epoch": 1.1032187100146553,
      "grad_norm": 0.09953287988901138,
      "learning_rate": 4.971903038468627e-06,
      "loss": 0.0018,
      "step": 172010
    },
    {
      "epoch": 1.1032828469084415,
      "grad_norm": 0.0722479522228241,
      "learning_rate": 4.971343344108322e-06,
      "loss": 0.0014,
      "step": 172020
    },
    {
      "epoch": 1.1033469838022274,
      "grad_norm": 0.11470306664705276,
      "learning_rate": 4.970783650107107e-06,
      "loss": 0.0014,
      "step": 172030
    },
    {
      "epoch": 1.1034111206960135,
      "grad_norm": 0.03702601417899132,
      "learning_rate": 4.970223956471992e-06,
      "loss": 0.0013,
      "step": 172040
    },
    {
      "epoch": 1.1034752575897997,
      "grad_norm": 0.17161774635314941,
      "learning_rate": 4.969664263209993e-06,
      "loss": 0.0026,
      "step": 172050
    },
    {
      "epoch": 1.1035393944835858,
      "grad_norm": 0.12393510341644287,
      "learning_rate": 4.969104570328122e-06,
      "loss": 0.0011,
      "step": 172060
    },
    {
      "epoch": 1.1036035313773718,
      "grad_norm": 0.09345302730798721,
      "learning_rate": 4.968544877833393e-06,
      "loss": 0.0024,
      "step": 172070
    },
    {
      "epoch": 1.103667668271158,
      "grad_norm": 0.2323468178510666,
      "learning_rate": 4.967985185732819e-06,
      "loss": 0.0017,
      "step": 172080
    },
    {
      "epoch": 1.103731805164944,
      "grad_norm": 0.06289396435022354,
      "learning_rate": 4.967425494033414e-06,
      "loss": 0.0011,
      "step": 172090
    },
    {
      "epoch": 1.1037959420587302,
      "grad_norm": 0.03909099102020264,
      "learning_rate": 4.96686580274219e-06,
      "loss": 0.0041,
      "step": 172100
    },
    {
      "epoch": 1.1038600789525161,
      "grad_norm": 0.1267758309841156,
      "learning_rate": 4.9663061118661615e-06,
      "loss": 0.0013,
      "step": 172110
    },
    {
      "epoch": 1.1039242158463023,
      "grad_norm": 0.07649413496255875,
      "learning_rate": 4.965746421412342e-06,
      "loss": 0.0025,
      "step": 172120
    },
    {
      "epoch": 1.1039883527400884,
      "grad_norm": 0.019668733701109886,
      "learning_rate": 4.9651867313877435e-06,
      "loss": 0.0014,
      "step": 172130
    },
    {
      "epoch": 1.1040524896338746,
      "grad_norm": 0.11278221011161804,
      "learning_rate": 4.9646270417993814e-06,
      "loss": 0.0012,
      "step": 172140
    },
    {
      "epoch": 1.1041166265276607,
      "grad_norm": 0.20489659905433655,
      "learning_rate": 4.964067352654267e-06,
      "loss": 0.0014,
      "step": 172150
    },
    {
      "epoch": 1.1041807634214467,
      "grad_norm": 0.06713900715112686,
      "learning_rate": 4.963507663959414e-06,
      "loss": 0.0023,
      "step": 172160
    },
    {
      "epoch": 1.1042449003152328,
      "grad_norm": 0.12010787427425385,
      "learning_rate": 4.962947975721836e-06,
      "loss": 0.0024,
      "step": 172170
    },
    {
      "epoch": 1.104309037209019,
      "grad_norm": 0.10179290920495987,
      "learning_rate": 4.962388287948548e-06,
      "loss": 0.0009,
      "step": 172180
    },
    {
      "epoch": 1.104373174102805,
      "grad_norm": 0.07030438631772995,
      "learning_rate": 4.961828600646559e-06,
      "loss": 0.0013,
      "step": 172190
    },
    {
      "epoch": 1.104437310996591,
      "grad_norm": 0.1282183825969696,
      "learning_rate": 4.961268913822887e-06,
      "loss": 0.0008,
      "step": 172200
    },
    {
      "epoch": 1.1045014478903772,
      "grad_norm": 0.18232722580432892,
      "learning_rate": 4.960709227484542e-06,
      "loss": 0.0013,
      "step": 172210
    },
    {
      "epoch": 1.1045655847841633,
      "grad_norm": 0.00621228851377964,
      "learning_rate": 4.960149541638539e-06,
      "loss": 0.0013,
      "step": 172220
    },
    {
      "epoch": 1.1046297216779495,
      "grad_norm": 0.034846991300582886,
      "learning_rate": 4.9595898562918905e-06,
      "loss": 0.0014,
      "step": 172230
    },
    {
      "epoch": 1.1046938585717354,
      "grad_norm": 0.2832137942314148,
      "learning_rate": 4.959030171451611e-06,
      "loss": 0.0014,
      "step": 172240
    },
    {
      "epoch": 1.1047579954655216,
      "grad_norm": 0.14668747782707214,
      "learning_rate": 4.9584704871247114e-06,
      "loss": 0.0015,
      "step": 172250
    },
    {
      "epoch": 1.1048221323593077,
      "grad_norm": 0.08905400335788727,
      "learning_rate": 4.9579108033182065e-06,
      "loss": 0.0022,
      "step": 172260
    },
    {
      "epoch": 1.1048862692530939,
      "grad_norm": 0.09415697306394577,
      "learning_rate": 4.957351120039109e-06,
      "loss": 0.0021,
      "step": 172270
    },
    {
      "epoch": 1.10495040614688,
      "grad_norm": 0.15120045840740204,
      "learning_rate": 4.956791437294433e-06,
      "loss": 0.0026,
      "step": 172280
    },
    {
      "epoch": 1.105014543040666,
      "grad_norm": 0.06795934587717056,
      "learning_rate": 4.956231755091192e-06,
      "loss": 0.0015,
      "step": 172290
    },
    {
      "epoch": 1.105078679934452,
      "grad_norm": 0.09866806119680405,
      "learning_rate": 4.955672073436397e-06,
      "loss": 0.0013,
      "step": 172300
    },
    {
      "epoch": 1.1051428168282382,
      "grad_norm": 0.06745009869337082,
      "learning_rate": 4.9551123923370635e-06,
      "loss": 0.0014,
      "step": 172310
    },
    {
      "epoch": 1.1052069537220244,
      "grad_norm": 0.09626159071922302,
      "learning_rate": 4.954552711800203e-06,
      "loss": 0.0021,
      "step": 172320
    },
    {
      "epoch": 1.1052710906158103,
      "grad_norm": 0.021619094535708427,
      "learning_rate": 4.953993031832831e-06,
      "loss": 0.0015,
      "step": 172330
    },
    {
      "epoch": 1.1053352275095965,
      "grad_norm": 0.045136865228414536,
      "learning_rate": 4.953433352441957e-06,
      "loss": 0.0018,
      "step": 172340
    },
    {
      "epoch": 1.1053993644033826,
      "grad_norm": 0.1451915055513382,
      "learning_rate": 4.952873673634598e-06,
      "loss": 0.0022,
      "step": 172350
    },
    {
      "epoch": 1.1054635012971687,
      "grad_norm": 0.07904060184955597,
      "learning_rate": 4.952313995417764e-06,
      "loss": 0.0012,
      "step": 172360
    },
    {
      "epoch": 1.1055276381909547,
      "grad_norm": 0.07658032327890396,
      "learning_rate": 4.95175431779847e-06,
      "loss": 0.0019,
      "step": 172370
    },
    {
      "epoch": 1.1055917750847408,
      "grad_norm": 0.02343565784394741,
      "learning_rate": 4.951194640783729e-06,
      "loss": 0.0009,
      "step": 172380
    },
    {
      "epoch": 1.105655911978527,
      "grad_norm": 0.018991762772202492,
      "learning_rate": 4.950634964380556e-06,
      "loss": 0.0017,
      "step": 172390
    },
    {
      "epoch": 1.1057200488723131,
      "grad_norm": 0.054553721100091934,
      "learning_rate": 4.95007528859596e-06,
      "loss": 0.0015,
      "step": 172400
    },
    {
      "epoch": 1.1057841857660993,
      "grad_norm": 0.03292379528284073,
      "learning_rate": 4.949515613436958e-06,
      "loss": 0.0023,
      "step": 172410
    },
    {
      "epoch": 1.1058483226598852,
      "grad_norm": 0.06313908100128174,
      "learning_rate": 4.94895593891056e-06,
      "loss": 0.0019,
      "step": 172420
    },
    {
      "epoch": 1.1059124595536713,
      "grad_norm": 0.21744129061698914,
      "learning_rate": 4.9483962650237815e-06,
      "loss": 0.0032,
      "step": 172430
    },
    {
      "epoch": 1.1059765964474575,
      "grad_norm": 0.09825435280799866,
      "learning_rate": 4.947836591783634e-06,
      "loss": 0.0009,
      "step": 172440
    },
    {
      "epoch": 1.1060407333412436,
      "grad_norm": 0.1280527561903,
      "learning_rate": 4.947276919197132e-06,
      "loss": 0.0014,
      "step": 172450
    },
    {
      "epoch": 1.1061048702350296,
      "grad_norm": 0.08048540353775024,
      "learning_rate": 4.946717247271287e-06,
      "loss": 0.0046,
      "step": 172460
    },
    {
      "epoch": 1.1061690071288157,
      "grad_norm": 0.16083328425884247,
      "learning_rate": 4.946157576013114e-06,
      "loss": 0.0022,
      "step": 172470
    },
    {
      "epoch": 1.1062331440226019,
      "grad_norm": 0.0801376923918724,
      "learning_rate": 4.9455979054296245e-06,
      "loss": 0.0029,
      "step": 172480
    },
    {
      "epoch": 1.106297280916388,
      "grad_norm": 0.04717901721596718,
      "learning_rate": 4.945038235527833e-06,
      "loss": 0.0008,
      "step": 172490
    },
    {
      "epoch": 1.106361417810174,
      "grad_norm": 0.026985157281160355,
      "learning_rate": 4.94447856631475e-06,
      "loss": 0.0016,
      "step": 172500
    },
    {
      "epoch": 1.10642555470396,
      "grad_norm": 0.26381930708885193,
      "learning_rate": 4.943918897797391e-06,
      "loss": 0.0018,
      "step": 172510
    },
    {
      "epoch": 1.1064896915977462,
      "grad_norm": 0.16577814519405365,
      "learning_rate": 4.9433592299827685e-06,
      "loss": 0.0015,
      "step": 172520
    },
    {
      "epoch": 1.1065538284915324,
      "grad_norm": 0.037880945950746536,
      "learning_rate": 4.942799562877895e-06,
      "loss": 0.0009,
      "step": 172530
    },
    {
      "epoch": 1.1066179653853183,
      "grad_norm": 0.23803481459617615,
      "learning_rate": 4.942239896489785e-06,
      "loss": 0.0016,
      "step": 172540
    },
    {
      "epoch": 1.1066821022791045,
      "grad_norm": 0.041799016296863556,
      "learning_rate": 4.941680230825449e-06,
      "loss": 0.0022,
      "step": 172550
    },
    {
      "epoch": 1.1067462391728906,
      "grad_norm": 0.04286804422736168,
      "learning_rate": 4.941120565891902e-06,
      "loss": 0.0016,
      "step": 172560
    },
    {
      "epoch": 1.1068103760666768,
      "grad_norm": 0.49322232604026794,
      "learning_rate": 4.940560901696156e-06,
      "loss": 0.0024,
      "step": 172570
    },
    {
      "epoch": 1.106874512960463,
      "grad_norm": 0.08396865427494049,
      "learning_rate": 4.940001238245225e-06,
      "loss": 0.0013,
      "step": 172580
    },
    {
      "epoch": 1.1069386498542488,
      "grad_norm": 0.1122252568602562,
      "learning_rate": 4.9394415755461205e-06,
      "loss": 0.0022,
      "step": 172590
    },
    {
      "epoch": 1.107002786748035,
      "grad_norm": 0.01022116094827652,
      "learning_rate": 4.9388819136058575e-06,
      "loss": 0.0012,
      "step": 172600
    },
    {
      "epoch": 1.1070669236418211,
      "grad_norm": 0.04051017016172409,
      "learning_rate": 4.938322252431447e-06,
      "loss": 0.0021,
      "step": 172610
    },
    {
      "epoch": 1.1071310605356073,
      "grad_norm": 0.05071176216006279,
      "learning_rate": 4.937762592029903e-06,
      "loss": 0.0012,
      "step": 172620
    },
    {
      "epoch": 1.1071951974293932,
      "grad_norm": 0.1994679868221283,
      "learning_rate": 4.937202932408238e-06,
      "loss": 0.0023,
      "step": 172630
    },
    {
      "epoch": 1.1072593343231794,
      "grad_norm": 0.15775303542613983,
      "learning_rate": 4.936643273573466e-06,
      "loss": 0.0027,
      "step": 172640
    },
    {
      "epoch": 1.1073234712169655,
      "grad_norm": 0.3187548518180847,
      "learning_rate": 4.936083615532598e-06,
      "loss": 0.0016,
      "step": 172650
    },
    {
      "epoch": 1.1073876081107517,
      "grad_norm": 0.0747041329741478,
      "learning_rate": 4.9355239582926485e-06,
      "loss": 0.0016,
      "step": 172660
    },
    {
      "epoch": 1.1074517450045378,
      "grad_norm": 0.059891946613788605,
      "learning_rate": 4.934964301860629e-06,
      "loss": 0.0012,
      "step": 172670
    },
    {
      "epoch": 1.1075158818983237,
      "grad_norm": 0.06337383389472961,
      "learning_rate": 4.934404646243553e-06,
      "loss": 0.0024,
      "step": 172680
    },
    {
      "epoch": 1.1075800187921099,
      "grad_norm": 0.1260882169008255,
      "learning_rate": 4.933844991448436e-06,
      "loss": 0.0023,
      "step": 172690
    },
    {
      "epoch": 1.107644155685896,
      "grad_norm": 0.03950232267379761,
      "learning_rate": 4.9332853374822865e-06,
      "loss": 0.0009,
      "step": 172700
    },
    {
      "epoch": 1.1077082925796822,
      "grad_norm": 0.005105361342430115,
      "learning_rate": 4.932725684352121e-06,
      "loss": 0.0015,
      "step": 172710
    },
    {
      "epoch": 1.107772429473468,
      "grad_norm": 0.1591700166463852,
      "learning_rate": 4.932166032064949e-06,
      "loss": 0.0019,
      "step": 172720
    },
    {
      "epoch": 1.1078365663672542,
      "grad_norm": 0.04399140551686287,
      "learning_rate": 4.931606380627786e-06,
      "loss": 0.0016,
      "step": 172730
    },
    {
      "epoch": 1.1079007032610404,
      "grad_norm": 0.2605571448802948,
      "learning_rate": 4.931046730047644e-06,
      "loss": 0.0015,
      "step": 172740
    },
    {
      "epoch": 1.1079648401548265,
      "grad_norm": 0.0853683203458786,
      "learning_rate": 4.930487080331536e-06,
      "loss": 0.0022,
      "step": 172750
    },
    {
      "epoch": 1.1080289770486125,
      "grad_norm": 0.2431391328573227,
      "learning_rate": 4.929927431486474e-06,
      "loss": 0.0012,
      "step": 172760
    },
    {
      "epoch": 1.1080931139423986,
      "grad_norm": 0.06449903547763824,
      "learning_rate": 4.929367783519471e-06,
      "loss": 0.0019,
      "step": 172770
    },
    {
      "epoch": 1.1081572508361848,
      "grad_norm": 0.0941937267780304,
      "learning_rate": 4.928808136437541e-06,
      "loss": 0.0023,
      "step": 172780
    },
    {
      "epoch": 1.108221387729971,
      "grad_norm": 0.019459959119558334,
      "learning_rate": 4.928248490247696e-06,
      "loss": 0.0012,
      "step": 172790
    },
    {
      "epoch": 1.1082855246237568,
      "grad_norm": 0.15396666526794434,
      "learning_rate": 4.927688844956947e-06,
      "loss": 0.0022,
      "step": 172800
    },
    {
      "epoch": 1.108349661517543,
      "grad_norm": 0.0548369325697422,
      "learning_rate": 4.927129200572311e-06,
      "loss": 0.0007,
      "step": 172810
    },
    {
      "epoch": 1.1084137984113291,
      "grad_norm": 0.014471418224275112,
      "learning_rate": 4.926569557100797e-06,
      "loss": 0.0106,
      "step": 172820
    },
    {
      "epoch": 1.1084779353051153,
      "grad_norm": 0.09189534187316895,
      "learning_rate": 4.926009914549418e-06,
      "loss": 0.0016,
      "step": 172830
    },
    {
      "epoch": 1.1085420721989014,
      "grad_norm": 0.04406943917274475,
      "learning_rate": 4.925450272925189e-06,
      "loss": 0.0012,
      "step": 172840
    },
    {
      "epoch": 1.1086062090926874,
      "grad_norm": 0.12239224463701248,
      "learning_rate": 4.92489063223512e-06,
      "loss": 0.0023,
      "step": 172850
    },
    {
      "epoch": 1.1086703459864735,
      "grad_norm": 0.12262288480997086,
      "learning_rate": 4.924330992486227e-06,
      "loss": 0.0015,
      "step": 172860
    },
    {
      "epoch": 1.1087344828802597,
      "grad_norm": 0.05069934204220772,
      "learning_rate": 4.923771353685519e-06,
      "loss": 0.0013,
      "step": 172870
    },
    {
      "epoch": 1.1087986197740458,
      "grad_norm": 0.5160471200942993,
      "learning_rate": 4.923211715840012e-06,
      "loss": 0.0013,
      "step": 172880
    },
    {
      "epoch": 1.1088627566678317,
      "grad_norm": 0.07519219815731049,
      "learning_rate": 4.922652078956716e-06,
      "loss": 0.001,
      "step": 172890
    },
    {
      "epoch": 1.1089268935616179,
      "grad_norm": 0.24941521883010864,
      "learning_rate": 4.922092443042645e-06,
      "loss": 0.003,
      "step": 172900
    },
    {
      "epoch": 1.108991030455404,
      "grad_norm": 0.34581393003463745,
      "learning_rate": 4.921532808104811e-06,
      "loss": 0.002,
      "step": 172910
    },
    {
      "epoch": 1.1090551673491902,
      "grad_norm": 0.13571126759052277,
      "learning_rate": 4.9209731741502275e-06,
      "loss": 0.001,
      "step": 172920
    },
    {
      "epoch": 1.1091193042429761,
      "grad_norm": 0.020635267719626427,
      "learning_rate": 4.920413541185906e-06,
      "loss": 0.0016,
      "step": 172930
    },
    {
      "epoch": 1.1091834411367623,
      "grad_norm": 0.08942442387342453,
      "learning_rate": 4.919853909218861e-06,
      "loss": 0.0014,
      "step": 172940
    },
    {
      "epoch": 1.1092475780305484,
      "grad_norm": 0.05175596475601196,
      "learning_rate": 4.919294278256102e-06,
      "loss": 0.0014,
      "step": 172950
    },
    {
      "epoch": 1.1093117149243346,
      "grad_norm": 0.10908970981836319,
      "learning_rate": 4.918734648304644e-06,
      "loss": 0.0016,
      "step": 172960
    },
    {
      "epoch": 1.1093758518181205,
      "grad_norm": 0.4310110807418823,
      "learning_rate": 4.9181750193715e-06,
      "loss": 0.0037,
      "step": 172970
    },
    {
      "epoch": 1.1094399887119066,
      "grad_norm": 0.07702956348657608,
      "learning_rate": 4.91761539146368e-06,
      "loss": 0.003,
      "step": 172980
    },
    {
      "epoch": 1.1095041256056928,
      "grad_norm": 0.10723809152841568,
      "learning_rate": 4.917055764588199e-06,
      "loss": 0.0016,
      "step": 172990
    },
    {
      "epoch": 1.109568262499479,
      "grad_norm": 0.01755482703447342,
      "learning_rate": 4.916496138752068e-06,
      "loss": 0.0027,
      "step": 173000
    },
    {
      "epoch": 1.109632399393265,
      "grad_norm": 0.07348005473613739,
      "learning_rate": 4.915936513962301e-06,
      "loss": 0.0018,
      "step": 173010
    },
    {
      "epoch": 1.109696536287051,
      "grad_norm": 0.06056353077292442,
      "learning_rate": 4.915376890225909e-06,
      "loss": 0.002,
      "step": 173020
    },
    {
      "epoch": 1.1097606731808372,
      "grad_norm": 0.011018101125955582,
      "learning_rate": 4.914817267549905e-06,
      "loss": 0.0011,
      "step": 173030
    },
    {
      "epoch": 1.1098248100746233,
      "grad_norm": 0.04618438333272934,
      "learning_rate": 4.914257645941301e-06,
      "loss": 0.0009,
      "step": 173040
    },
    {
      "epoch": 1.1098889469684095,
      "grad_norm": 0.09414304047822952,
      "learning_rate": 4.913698025407111e-06,
      "loss": 0.0018,
      "step": 173050
    },
    {
      "epoch": 1.1099530838621954,
      "grad_norm": 0.042383674532175064,
      "learning_rate": 4.913138405954345e-06,
      "loss": 0.0012,
      "step": 173060
    },
    {
      "epoch": 1.1100172207559815,
      "grad_norm": 0.04793866351246834,
      "learning_rate": 4.912578787590018e-06,
      "loss": 0.0008,
      "step": 173070
    },
    {
      "epoch": 1.1100813576497677,
      "grad_norm": 0.03588831424713135,
      "learning_rate": 4.912019170321141e-06,
      "loss": 0.0006,
      "step": 173080
    },
    {
      "epoch": 1.1101454945435538,
      "grad_norm": 0.1358732283115387,
      "learning_rate": 4.9114595541547274e-06,
      "loss": 0.0023,
      "step": 173090
    },
    {
      "epoch": 1.11020963143734,
      "grad_norm": 0.10910496860742569,
      "learning_rate": 4.910899939097787e-06,
      "loss": 0.0013,
      "step": 173100
    },
    {
      "epoch": 1.110273768331126,
      "grad_norm": 0.06244830787181854,
      "learning_rate": 4.910340325157335e-06,
      "loss": 0.0019,
      "step": 173110
    },
    {
      "epoch": 1.110337905224912,
      "grad_norm": 0.06858513504266739,
      "learning_rate": 4.909780712340381e-06,
      "loss": 0.0009,
      "step": 173120
    },
    {
      "epoch": 1.1104020421186982,
      "grad_norm": 0.23532763123512268,
      "learning_rate": 4.909221100653942e-06,
      "loss": 0.0029,
      "step": 173130
    },
    {
      "epoch": 1.1104661790124843,
      "grad_norm": 0.01755049265921116,
      "learning_rate": 4.908661490105026e-06,
      "loss": 0.0019,
      "step": 173140
    },
    {
      "epoch": 1.1105303159062703,
      "grad_norm": 0.1809210330247879,
      "learning_rate": 4.908101880700648e-06,
      "loss": 0.0019,
      "step": 173150
    },
    {
      "epoch": 1.1105944528000564,
      "grad_norm": 0.1378970891237259,
      "learning_rate": 4.907542272447819e-06,
      "loss": 0.0026,
      "step": 173160
    },
    {
      "epoch": 1.1106585896938426,
      "grad_norm": 0.1288081407546997,
      "learning_rate": 4.906982665353551e-06,
      "loss": 0.001,
      "step": 173170
    },
    {
      "epoch": 1.1107227265876287,
      "grad_norm": 0.22905784845352173,
      "learning_rate": 4.906423059424856e-06,
      "loss": 0.0014,
      "step": 173180
    },
    {
      "epoch": 1.1107868634814146,
      "grad_norm": 0.17911554872989655,
      "learning_rate": 4.9058634546687474e-06,
      "loss": 0.003,
      "step": 173190
    },
    {
      "epoch": 1.1108510003752008,
      "grad_norm": 0.34671086072921753,
      "learning_rate": 4.905303851092237e-06,
      "loss": 0.0021,
      "step": 173200
    },
    {
      "epoch": 1.110915137268987,
      "grad_norm": 0.07665128260850906,
      "learning_rate": 4.904744248702338e-06,
      "loss": 0.001,
      "step": 173210
    },
    {
      "epoch": 1.110979274162773,
      "grad_norm": 0.20960409939289093,
      "learning_rate": 4.90418464750606e-06,
      "loss": 0.001,
      "step": 173220
    },
    {
      "epoch": 1.111043411056559,
      "grad_norm": 0.1022791862487793,
      "learning_rate": 4.903625047510418e-06,
      "loss": 0.0009,
      "step": 173230
    },
    {
      "epoch": 1.1111075479503452,
      "grad_norm": 0.03949712589383125,
      "learning_rate": 4.903065448722423e-06,
      "loss": 0.0017,
      "step": 173240
    },
    {
      "epoch": 1.1111716848441313,
      "grad_norm": 0.10221009701490402,
      "learning_rate": 4.902505851149086e-06,
      "loss": 0.0015,
      "step": 173250
    },
    {
      "epoch": 1.1112358217379175,
      "grad_norm": 0.09457410126924515,
      "learning_rate": 4.901946254797422e-06,
      "loss": 0.0008,
      "step": 173260
    },
    {
      "epoch": 1.1112999586317036,
      "grad_norm": 0.11744176596403122,
      "learning_rate": 4.90138665967444e-06,
      "loss": 0.0011,
      "step": 173270
    },
    {
      "epoch": 1.1113640955254895,
      "grad_norm": 0.16001646220684052,
      "learning_rate": 4.9008270657871545e-06,
      "loss": 0.0015,
      "step": 173280
    },
    {
      "epoch": 1.1114282324192757,
      "grad_norm": 0.10524749755859375,
      "learning_rate": 4.900267473142576e-06,
      "loss": 0.0025,
      "step": 173290
    },
    {
      "epoch": 1.1114923693130618,
      "grad_norm": 0.26658308506011963,
      "learning_rate": 4.899707881747719e-06,
      "loss": 0.0021,
      "step": 173300
    },
    {
      "epoch": 1.111556506206848,
      "grad_norm": 0.057558316737413406,
      "learning_rate": 4.899148291609593e-06,
      "loss": 0.001,
      "step": 173310
    },
    {
      "epoch": 1.111620643100634,
      "grad_norm": 0.04493267089128494,
      "learning_rate": 4.898588702735211e-06,
      "loss": 0.002,
      "step": 173320
    },
    {
      "epoch": 1.11168477999442,
      "grad_norm": 0.0999845489859581,
      "learning_rate": 4.898029115131584e-06,
      "loss": 0.002,
      "step": 173330
    },
    {
      "epoch": 1.1117489168882062,
      "grad_norm": 0.0713239312171936,
      "learning_rate": 4.897469528805727e-06,
      "loss": 0.0021,
      "step": 173340
    },
    {
      "epoch": 1.1118130537819924,
      "grad_norm": 0.07967229932546616,
      "learning_rate": 4.896909943764649e-06,
      "loss": 0.001,
      "step": 173350
    },
    {
      "epoch": 1.1118771906757783,
      "grad_norm": 0.1508815437555313,
      "learning_rate": 4.896350360015363e-06,
      "loss": 0.002,
      "step": 173360
    },
    {
      "epoch": 1.1119413275695644,
      "grad_norm": 0.12152405083179474,
      "learning_rate": 4.895790777564881e-06,
      "loss": 0.0012,
      "step": 173370
    },
    {
      "epoch": 1.1120054644633506,
      "grad_norm": 0.053805720061063766,
      "learning_rate": 4.895231196420215e-06,
      "loss": 0.0013,
      "step": 173380
    },
    {
      "epoch": 1.1120696013571367,
      "grad_norm": 0.05876936763525009,
      "learning_rate": 4.894671616588377e-06,
      "loss": 0.0009,
      "step": 173390
    },
    {
      "epoch": 1.1121337382509227,
      "grad_norm": 0.10661150515079498,
      "learning_rate": 4.894112038076379e-06,
      "loss": 0.001,
      "step": 173400
    },
    {
      "epoch": 1.1121978751447088,
      "grad_norm": 0.06447902321815491,
      "learning_rate": 4.893552460891234e-06,
      "loss": 0.0015,
      "step": 173410
    },
    {
      "epoch": 1.112262012038495,
      "grad_norm": 0.01974887028336525,
      "learning_rate": 4.892992885039951e-06,
      "loss": 0.0012,
      "step": 173420
    },
    {
      "epoch": 1.112326148932281,
      "grad_norm": 0.20772825181484222,
      "learning_rate": 4.892433310529545e-06,
      "loss": 0.0024,
      "step": 173430
    },
    {
      "epoch": 1.1123902858260672,
      "grad_norm": 0.029797615483403206,
      "learning_rate": 4.891873737367025e-06,
      "loss": 0.0014,
      "step": 173440
    },
    {
      "epoch": 1.1124544227198532,
      "grad_norm": 0.3723593056201935,
      "learning_rate": 4.891314165559406e-06,
      "loss": 0.003,
      "step": 173450
    },
    {
      "epoch": 1.1125185596136393,
      "grad_norm": 0.06410173326730728,
      "learning_rate": 4.890754595113697e-06,
      "loss": 0.0022,
      "step": 173460
    },
    {
      "epoch": 1.1125826965074255,
      "grad_norm": 0.04267742112278938,
      "learning_rate": 4.890195026036912e-06,
      "loss": 0.0015,
      "step": 173470
    },
    {
      "epoch": 1.1126468334012116,
      "grad_norm": 0.18614718317985535,
      "learning_rate": 4.8896354583360606e-06,
      "loss": 0.0015,
      "step": 173480
    },
    {
      "epoch": 1.1127109702949975,
      "grad_norm": 0.06482775509357452,
      "learning_rate": 4.889075892018157e-06,
      "loss": 0.0029,
      "step": 173490
    },
    {
      "epoch": 1.1127751071887837,
      "grad_norm": 0.049155496060848236,
      "learning_rate": 4.88851632709021e-06,
      "loss": 0.0012,
      "step": 173500
    },
    {
      "epoch": 1.1128392440825698,
      "grad_norm": 0.08097552508115768,
      "learning_rate": 4.887956763559235e-06,
      "loss": 0.0021,
      "step": 173510
    },
    {
      "epoch": 1.112903380976356,
      "grad_norm": 0.12890399992465973,
      "learning_rate": 4.88739720143224e-06,
      "loss": 0.0035,
      "step": 173520
    },
    {
      "epoch": 1.1129675178701421,
      "grad_norm": 0.04127669334411621,
      "learning_rate": 4.88683764071624e-06,
      "loss": 0.0014,
      "step": 173530
    },
    {
      "epoch": 1.113031654763928,
      "grad_norm": 0.060111287981271744,
      "learning_rate": 4.886278081418245e-06,
      "loss": 0.0033,
      "step": 173540
    },
    {
      "epoch": 1.1130957916577142,
      "grad_norm": 0.10613421350717545,
      "learning_rate": 4.8857185235452665e-06,
      "loss": 0.001,
      "step": 173550
    },
    {
      "epoch": 1.1131599285515004,
      "grad_norm": 0.2460632473230362,
      "learning_rate": 4.885158967104317e-06,
      "loss": 0.0017,
      "step": 173560
    },
    {
      "epoch": 1.1132240654452865,
      "grad_norm": 0.05923857167363167,
      "learning_rate": 4.884599412102408e-06,
      "loss": 0.0017,
      "step": 173570
    },
    {
      "epoch": 1.1132882023390724,
      "grad_norm": 0.05139973387122154,
      "learning_rate": 4.88403985854655e-06,
      "loss": 0.0018,
      "step": 173580
    },
    {
      "epoch": 1.1133523392328586,
      "grad_norm": 0.13422226905822754,
      "learning_rate": 4.883480306443756e-06,
      "loss": 0.0029,
      "step": 173590
    },
    {
      "epoch": 1.1134164761266447,
      "grad_norm": 0.030855519697070122,
      "learning_rate": 4.882920755801037e-06,
      "loss": 0.0014,
      "step": 173600
    },
    {
      "epoch": 1.1134806130204309,
      "grad_norm": 0.06292476505041122,
      "learning_rate": 4.8823612066254035e-06,
      "loss": 0.0088,
      "step": 173610
    },
    {
      "epoch": 1.1135447499142168,
      "grad_norm": 0.2309848517179489,
      "learning_rate": 4.88180165892387e-06,
      "loss": 0.0046,
      "step": 173620
    },
    {
      "epoch": 1.113608886808003,
      "grad_norm": 0.028664829209446907,
      "learning_rate": 4.881242112703445e-06,
      "loss": 0.0015,
      "step": 173630
    },
    {
      "epoch": 1.113673023701789,
      "grad_norm": 0.18330270051956177,
      "learning_rate": 4.8806825679711424e-06,
      "loss": 0.0018,
      "step": 173640
    },
    {
      "epoch": 1.1137371605955753,
      "grad_norm": 0.10287895798683167,
      "learning_rate": 4.880123024733971e-06,
      "loss": 0.0011,
      "step": 173650
    },
    {
      "epoch": 1.1138012974893612,
      "grad_norm": 0.0056119938381016254,
      "learning_rate": 4.879563482998945e-06,
      "loss": 0.0028,
      "step": 173660
    },
    {
      "epoch": 1.1138654343831473,
      "grad_norm": 0.029498474672436714,
      "learning_rate": 4.879003942773074e-06,
      "loss": 0.0021,
      "step": 173670
    },
    {
      "epoch": 1.1139295712769335,
      "grad_norm": 0.3677642345428467,
      "learning_rate": 4.8784444040633685e-06,
      "loss": 0.0021,
      "step": 173680
    },
    {
      "epoch": 1.1139937081707196,
      "grad_norm": 0.06543778628110886,
      "learning_rate": 4.8778848668768446e-06,
      "loss": 0.0015,
      "step": 173690
    },
    {
      "epoch": 1.1140578450645058,
      "grad_norm": 0.13932010531425476,
      "learning_rate": 4.8773253312205085e-06,
      "loss": 0.0024,
      "step": 173700
    },
    {
      "epoch": 1.1141219819582917,
      "grad_norm": 0.1845843493938446,
      "learning_rate": 4.8767657971013746e-06,
      "loss": 0.002,
      "step": 173710
    },
    {
      "epoch": 1.1141861188520779,
      "grad_norm": 0.05792969465255737,
      "learning_rate": 4.876206264526452e-06,
      "loss": 0.0011,
      "step": 173720
    },
    {
      "epoch": 1.114250255745864,
      "grad_norm": 0.3107650876045227,
      "learning_rate": 4.875646733502755e-06,
      "loss": 0.0029,
      "step": 173730
    },
    {
      "epoch": 1.1143143926396502,
      "grad_norm": 0.008603124879300594,
      "learning_rate": 4.875087204037292e-06,
      "loss": 0.0023,
      "step": 173740
    },
    {
      "epoch": 1.114378529533436,
      "grad_norm": 0.07859355211257935,
      "learning_rate": 4.8745276761370776e-06,
      "loss": 0.0008,
      "step": 173750
    },
    {
      "epoch": 1.1144426664272222,
      "grad_norm": 0.062411241233348846,
      "learning_rate": 4.873968149809119e-06,
      "loss": 0.0013,
      "step": 173760
    },
    {
      "epoch": 1.1145068033210084,
      "grad_norm": 0.08033911138772964,
      "learning_rate": 4.873408625060431e-06,
      "loss": 0.0012,
      "step": 173770
    },
    {
      "epoch": 1.1145709402147945,
      "grad_norm": 0.05683527886867523,
      "learning_rate": 4.872849101898022e-06,
      "loss": 0.0024,
      "step": 173780
    },
    {
      "epoch": 1.1146350771085805,
      "grad_norm": 0.10068879276514053,
      "learning_rate": 4.872289580328906e-06,
      "loss": 0.0026,
      "step": 173790
    },
    {
      "epoch": 1.1146992140023666,
      "grad_norm": 0.056821130216121674,
      "learning_rate": 4.871730060360091e-06,
      "loss": 0.0029,
      "step": 173800
    },
    {
      "epoch": 1.1147633508961527,
      "grad_norm": 0.21229983866214752,
      "learning_rate": 4.871170541998591e-06,
      "loss": 0.0025,
      "step": 173810
    },
    {
      "epoch": 1.114827487789939,
      "grad_norm": 0.12080463021993637,
      "learning_rate": 4.870611025251417e-06,
      "loss": 0.0016,
      "step": 173820
    },
    {
      "epoch": 1.114891624683725,
      "grad_norm": 0.04254530742764473,
      "learning_rate": 4.870051510125578e-06,
      "loss": 0.0014,
      "step": 173830
    },
    {
      "epoch": 1.114955761577511,
      "grad_norm": 0.12097093462944031,
      "learning_rate": 4.8694919966280875e-06,
      "loss": 0.0009,
      "step": 173840
    },
    {
      "epoch": 1.1150198984712971,
      "grad_norm": 0.020847784355282784,
      "learning_rate": 4.8689324847659556e-06,
      "loss": 0.002,
      "step": 173850
    },
    {
      "epoch": 1.1150840353650833,
      "grad_norm": 0.27949169278144836,
      "learning_rate": 4.868372974546193e-06,
      "loss": 0.0043,
      "step": 173860
    },
    {
      "epoch": 1.1151481722588694,
      "grad_norm": 0.005513959098607302,
      "learning_rate": 4.867813465975811e-06,
      "loss": 0.0013,
      "step": 173870
    },
    {
      "epoch": 1.1152123091526553,
      "grad_norm": 0.017530998215079308,
      "learning_rate": 4.867253959061821e-06,
      "loss": 0.0016,
      "step": 173880
    },
    {
      "epoch": 1.1152764460464415,
      "grad_norm": 0.20806366205215454,
      "learning_rate": 4.866694453811232e-06,
      "loss": 0.0019,
      "step": 173890
    },
    {
      "epoch": 1.1153405829402276,
      "grad_norm": 0.08184632658958435,
      "learning_rate": 4.866134950231058e-06,
      "loss": 0.0021,
      "step": 173900
    },
    {
      "epoch": 1.1154047198340138,
      "grad_norm": 0.0540790855884552,
      "learning_rate": 4.8655754483283095e-06,
      "loss": 0.0014,
      "step": 173910
    },
    {
      "epoch": 1.1154688567277997,
      "grad_norm": 0.09348951280117035,
      "learning_rate": 4.8650159481099955e-06,
      "loss": 0.0014,
      "step": 173920
    },
    {
      "epoch": 1.1155329936215859,
      "grad_norm": 0.039533596485853195,
      "learning_rate": 4.86445644958313e-06,
      "loss": 0.002,
      "step": 173930
    },
    {
      "epoch": 1.115597130515372,
      "grad_norm": 0.32928287982940674,
      "learning_rate": 4.86389695275472e-06,
      "loss": 0.0026,
      "step": 173940
    },
    {
      "epoch": 1.1156612674091582,
      "grad_norm": 0.10675010085105896,
      "learning_rate": 4.863337457631779e-06,
      "loss": 0.0032,
      "step": 173950
    },
    {
      "epoch": 1.1157254043029443,
      "grad_norm": 0.09021563082933426,
      "learning_rate": 4.862777964221318e-06,
      "loss": 0.0016,
      "step": 173960
    },
    {
      "epoch": 1.1157895411967302,
      "grad_norm": 0.13088293373584747,
      "learning_rate": 4.862218472530347e-06,
      "loss": 0.0007,
      "step": 173970
    },
    {
      "epoch": 1.1158536780905164,
      "grad_norm": 0.06926416605710983,
      "learning_rate": 4.861658982565877e-06,
      "loss": 0.002,
      "step": 173980
    },
    {
      "epoch": 1.1159178149843025,
      "grad_norm": 0.2240966260433197,
      "learning_rate": 4.8610994943349185e-06,
      "loss": 0.0011,
      "step": 173990
    },
    {
      "epoch": 1.1159819518780887,
      "grad_norm": 0.010228103958070278,
      "learning_rate": 4.860540007844484e-06,
      "loss": 0.0019,
      "step": 174000
    },
    {
      "epoch": 1.1160460887718746,
      "grad_norm": 0.12614555656909943,
      "learning_rate": 4.85998052310158e-06,
      "loss": 0.0013,
      "step": 174010
    },
    {
      "epoch": 1.1161102256656608,
      "grad_norm": 0.06303328275680542,
      "learning_rate": 4.859421040113223e-06,
      "loss": 0.0014,
      "step": 174020
    },
    {
      "epoch": 1.116174362559447,
      "grad_norm": 0.04897018149495125,
      "learning_rate": 4.85886155888642e-06,
      "loss": 0.0013,
      "step": 174030
    },
    {
      "epoch": 1.116238499453233,
      "grad_norm": 0.14528539776802063,
      "learning_rate": 4.858302079428182e-06,
      "loss": 0.0021,
      "step": 174040
    },
    {
      "epoch": 1.116302636347019,
      "grad_norm": 0.08426199853420258,
      "learning_rate": 4.85774260174552e-06,
      "loss": 0.0013,
      "step": 174050
    },
    {
      "epoch": 1.1163667732408051,
      "grad_norm": 0.027923064306378365,
      "learning_rate": 4.857183125845446e-06,
      "loss": 0.0016,
      "step": 174060
    },
    {
      "epoch": 1.1164309101345913,
      "grad_norm": 0.14836865663528442,
      "learning_rate": 4.8566236517349685e-06,
      "loss": 0.0012,
      "step": 174070
    },
    {
      "epoch": 1.1164950470283774,
      "grad_norm": 0.17420117557048798,
      "learning_rate": 4.856064179421099e-06,
      "loss": 0.0033,
      "step": 174080
    },
    {
      "epoch": 1.1165591839221634,
      "grad_norm": 0.03189292922616005,
      "learning_rate": 4.855504708910849e-06,
      "loss": 0.0015,
      "step": 174090
    },
    {
      "epoch": 1.1166233208159495,
      "grad_norm": 0.02180171012878418,
      "learning_rate": 4.854945240211228e-06,
      "loss": 0.0012,
      "step": 174100
    },
    {
      "epoch": 1.1166874577097357,
      "grad_norm": 0.005740481428802013,
      "learning_rate": 4.854385773329247e-06,
      "loss": 0.0028,
      "step": 174110
    },
    {
      "epoch": 1.1167515946035218,
      "grad_norm": 0.11378605663776398,
      "learning_rate": 4.8538263082719155e-06,
      "loss": 0.0016,
      "step": 174120
    },
    {
      "epoch": 1.116815731497308,
      "grad_norm": 0.04232391342520714,
      "learning_rate": 4.853266845046246e-06,
      "loss": 0.0015,
      "step": 174130
    },
    {
      "epoch": 1.1168798683910939,
      "grad_norm": 0.08134877681732178,
      "learning_rate": 4.852707383659246e-06,
      "loss": 0.0013,
      "step": 174140
    },
    {
      "epoch": 1.11694400528488,
      "grad_norm": 0.13055117428302765,
      "learning_rate": 4.85214792411793e-06,
      "loss": 0.0012,
      "step": 174150
    },
    {
      "epoch": 1.1170081421786662,
      "grad_norm": 0.027094565331935883,
      "learning_rate": 4.851588466429305e-06,
      "loss": 0.0011,
      "step": 174160
    },
    {
      "epoch": 1.1170722790724523,
      "grad_norm": 0.05294719710946083,
      "learning_rate": 4.851029010600383e-06,
      "loss": 0.0013,
      "step": 174170
    },
    {
      "epoch": 1.1171364159662382,
      "grad_norm": 0.2595820426940918,
      "learning_rate": 4.850469556638173e-06,
      "loss": 0.0036,
      "step": 174180
    },
    {
      "epoch": 1.1172005528600244,
      "grad_norm": 0.08011367917060852,
      "learning_rate": 4.849910104549687e-06,
      "loss": 0.0022,
      "step": 174190
    },
    {
      "epoch": 1.1172646897538105,
      "grad_norm": 0.05550553277134895,
      "learning_rate": 4.849350654341934e-06,
      "loss": 0.0008,
      "step": 174200
    },
    {
      "epoch": 1.1173288266475967,
      "grad_norm": 0.06627099961042404,
      "learning_rate": 4.848791206021927e-06,
      "loss": 0.0027,
      "step": 174210
    },
    {
      "epoch": 1.1173929635413828,
      "grad_norm": 0.05866013467311859,
      "learning_rate": 4.848231759596672e-06,
      "loss": 0.0022,
      "step": 174220
    },
    {
      "epoch": 1.1174571004351688,
      "grad_norm": 0.05826423689723015,
      "learning_rate": 4.847672315073182e-06,
      "loss": 0.0028,
      "step": 174230
    },
    {
      "epoch": 1.117521237328955,
      "grad_norm": 0.098142109811306,
      "learning_rate": 4.847112872458467e-06,
      "loss": 0.0016,
      "step": 174240
    },
    {
      "epoch": 1.117585374222741,
      "grad_norm": 0.13749217987060547,
      "learning_rate": 4.8465534317595375e-06,
      "loss": 0.0022,
      "step": 174250
    },
    {
      "epoch": 1.1176495111165272,
      "grad_norm": 0.08138950914144516,
      "learning_rate": 4.845993992983404e-06,
      "loss": 0.0018,
      "step": 174260
    },
    {
      "epoch": 1.1177136480103131,
      "grad_norm": 0.1339346468448639,
      "learning_rate": 4.8454345561370734e-06,
      "loss": 0.0017,
      "step": 174270
    },
    {
      "epoch": 1.1177777849040993,
      "grad_norm": 0.17551785707473755,
      "learning_rate": 4.84487512122756e-06,
      "loss": 0.0019,
      "step": 174280
    },
    {
      "epoch": 1.1178419217978854,
      "grad_norm": 0.044013574719429016,
      "learning_rate": 4.844315688261871e-06,
      "loss": 0.0008,
      "step": 174290
    },
    {
      "epoch": 1.1179060586916716,
      "grad_norm": 0.031187549233436584,
      "learning_rate": 4.8437562572470195e-06,
      "loss": 0.001,
      "step": 174300
    },
    {
      "epoch": 1.1179701955854575,
      "grad_norm": 0.04440230131149292,
      "learning_rate": 4.843196828190012e-06,
      "loss": 0.0032,
      "step": 174310
    },
    {
      "epoch": 1.1180343324792437,
      "grad_norm": 0.036718934774398804,
      "learning_rate": 4.842637401097861e-06,
      "loss": 0.0012,
      "step": 174320
    },
    {
      "epoch": 1.1180984693730298,
      "grad_norm": 0.07007744163274765,
      "learning_rate": 4.842077975977575e-06,
      "loss": 0.0021,
      "step": 174330
    },
    {
      "epoch": 1.118162606266816,
      "grad_norm": 0.11462406814098358,
      "learning_rate": 4.841518552836167e-06,
      "loss": 0.0019,
      "step": 174340
    },
    {
      "epoch": 1.1182267431606019,
      "grad_norm": 0.06285811215639114,
      "learning_rate": 4.840959131680643e-06,
      "loss": 0.0031,
      "step": 174350
    },
    {
      "epoch": 1.118290880054388,
      "grad_norm": 0.07035444676876068,
      "learning_rate": 4.840399712518015e-06,
      "loss": 0.0014,
      "step": 174360
    },
    {
      "epoch": 1.1183550169481742,
      "grad_norm": 0.015017986297607422,
      "learning_rate": 4.839840295355292e-06,
      "loss": 0.0005,
      "step": 174370
    },
    {
      "epoch": 1.1184191538419603,
      "grad_norm": 0.07589603215456009,
      "learning_rate": 4.839280880199485e-06,
      "loss": 0.0014,
      "step": 174380
    },
    {
      "epoch": 1.1184832907357465,
      "grad_norm": 0.16272950172424316,
      "learning_rate": 4.838721467057605e-06,
      "loss": 0.0013,
      "step": 174390
    },
    {
      "epoch": 1.1185474276295324,
      "grad_norm": 0.09640228748321533,
      "learning_rate": 4.838162055936658e-06,
      "loss": 0.0029,
      "step": 174400
    },
    {
      "epoch": 1.1186115645233186,
      "grad_norm": 0.18411511182785034,
      "learning_rate": 4.837602646843658e-06,
      "loss": 0.0017,
      "step": 174410
    },
    {
      "epoch": 1.1186757014171047,
      "grad_norm": 0.040111999958753586,
      "learning_rate": 4.8370432397856115e-06,
      "loss": 0.0013,
      "step": 174420
    },
    {
      "epoch": 1.1187398383108909,
      "grad_norm": 0.02608310803771019,
      "learning_rate": 4.836483834769531e-06,
      "loss": 0.0017,
      "step": 174430
    },
    {
      "epoch": 1.1188039752046768,
      "grad_norm": 0.17651981115341187,
      "learning_rate": 4.835924431802423e-06,
      "loss": 0.0011,
      "step": 174440
    },
    {
      "epoch": 1.118868112098463,
      "grad_norm": 0.26195093989372253,
      "learning_rate": 4.835365030891301e-06,
      "loss": 0.001,
      "step": 174450
    },
    {
      "epoch": 1.118932248992249,
      "grad_norm": 0.12918025255203247,
      "learning_rate": 4.834805632043172e-06,
      "loss": 0.0034,
      "step": 174460
    },
    {
      "epoch": 1.1189963858860352,
      "grad_norm": 0.15502963960170746,
      "learning_rate": 4.834246235265048e-06,
      "loss": 0.0015,
      "step": 174470
    },
    {
      "epoch": 1.1190605227798212,
      "grad_norm": 0.0754869133234024,
      "learning_rate": 4.833686840563935e-06,
      "loss": 0.0019,
      "step": 174480
    },
    {
      "epoch": 1.1191246596736073,
      "grad_norm": 0.07098864763975143,
      "learning_rate": 4.833127447946846e-06,
      "loss": 0.0015,
      "step": 174490
    },
    {
      "epoch": 1.1191887965673935,
      "grad_norm": 0.024236060678958893,
      "learning_rate": 4.832568057420788e-06,
      "loss": 0.0066,
      "step": 174500
    },
    {
      "epoch": 1.1192529334611796,
      "grad_norm": 0.0722365453839302,
      "learning_rate": 4.832008668992774e-06,
      "loss": 0.0015,
      "step": 174510
    },
    {
      "epoch": 1.1193170703549655,
      "grad_norm": 0.026653124019503593,
      "learning_rate": 4.831449282669809e-06,
      "loss": 0.0039,
      "step": 174520
    },
    {
      "epoch": 1.1193812072487517,
      "grad_norm": 0.09310638904571533,
      "learning_rate": 4.830889898458906e-06,
      "loss": 0.0008,
      "step": 174530
    },
    {
      "epoch": 1.1194453441425378,
      "grad_norm": 0.057517159730196,
      "learning_rate": 4.830330516367074e-06,
      "loss": 0.0017,
      "step": 174540
    },
    {
      "epoch": 1.119509481036324,
      "grad_norm": 0.3690768778324127,
      "learning_rate": 4.829771136401321e-06,
      "loss": 0.0016,
      "step": 174550
    },
    {
      "epoch": 1.1195736179301101,
      "grad_norm": 0.14363223314285278,
      "learning_rate": 4.829211758568658e-06,
      "loss": 0.0025,
      "step": 174560
    },
    {
      "epoch": 1.119637754823896,
      "grad_norm": 0.10250331461429596,
      "learning_rate": 4.828652382876093e-06,
      "loss": 0.002,
      "step": 174570
    },
    {
      "epoch": 1.1197018917176822,
      "grad_norm": 0.04894329234957695,
      "learning_rate": 4.8280930093306376e-06,
      "loss": 0.0016,
      "step": 174580
    },
    {
      "epoch": 1.1197660286114683,
      "grad_norm": 0.13075852394104004,
      "learning_rate": 4.827533637939298e-06,
      "loss": 0.0013,
      "step": 174590
    },
    {
      "epoch": 1.1198301655052545,
      "grad_norm": 0.07651156187057495,
      "learning_rate": 4.826974268709087e-06,
      "loss": 0.0013,
      "step": 174600
    },
    {
      "epoch": 1.1198943023990404,
      "grad_norm": 0.05896775797009468,
      "learning_rate": 4.826414901647009e-06,
      "loss": 0.0017,
      "step": 174610
    },
    {
      "epoch": 1.1199584392928266,
      "grad_norm": 0.18786385655403137,
      "learning_rate": 4.825855536760079e-06,
      "loss": 0.0018,
      "step": 174620
    },
    {
      "epoch": 1.1200225761866127,
      "grad_norm": 0.096187062561512,
      "learning_rate": 4.825296174055303e-06,
      "loss": 0.0019,
      "step": 174630
    },
    {
      "epoch": 1.1200867130803989,
      "grad_norm": 0.36271122097969055,
      "learning_rate": 4.824736813539691e-06,
      "loss": 0.0022,
      "step": 174640
    },
    {
      "epoch": 1.120150849974185,
      "grad_norm": 0.15491759777069092,
      "learning_rate": 4.824177455220249e-06,
      "loss": 0.0016,
      "step": 174650
    },
    {
      "epoch": 1.120214986867971,
      "grad_norm": 0.040893521159887314,
      "learning_rate": 4.823618099103992e-06,
      "loss": 0.001,
      "step": 174660
    },
    {
      "epoch": 1.120279123761757,
      "grad_norm": 0.10148506611585617,
      "learning_rate": 4.823058745197926e-06,
      "loss": 0.0016,
      "step": 174670
    },
    {
      "epoch": 1.1203432606555432,
      "grad_norm": 0.06153487041592598,
      "learning_rate": 4.822499393509061e-06,
      "loss": 0.0014,
      "step": 174680
    },
    {
      "epoch": 1.1204073975493294,
      "grad_norm": 0.05802586302161217,
      "learning_rate": 4.821940044044404e-06,
      "loss": 0.0009,
      "step": 174690
    },
    {
      "epoch": 1.1204715344431153,
      "grad_norm": 0.0913996621966362,
      "learning_rate": 4.8213806968109675e-06,
      "loss": 0.0021,
      "step": 174700
    },
    {
      "epoch": 1.1205356713369015,
      "grad_norm": 0.055139727890491486,
      "learning_rate": 4.820821351815757e-06,
      "loss": 0.0012,
      "step": 174710
    },
    {
      "epoch": 1.1205998082306876,
      "grad_norm": 0.07627728581428528,
      "learning_rate": 4.820262009065784e-06,
      "loss": 0.001,
      "step": 174720
    },
    {
      "epoch": 1.1206639451244738,
      "grad_norm": 0.020158030092716217,
      "learning_rate": 4.819702668568056e-06,
      "loss": 0.0026,
      "step": 174730
    },
    {
      "epoch": 1.1207280820182597,
      "grad_norm": 0.16587579250335693,
      "learning_rate": 4.819143330329582e-06,
      "loss": 0.0027,
      "step": 174740
    },
    {
      "epoch": 1.1207922189120458,
      "grad_norm": 0.007621010299772024,
      "learning_rate": 4.818583994357372e-06,
      "loss": 0.0013,
      "step": 174750
    },
    {
      "epoch": 1.120856355805832,
      "grad_norm": 0.10911239683628082,
      "learning_rate": 4.818024660658434e-06,
      "loss": 0.0022,
      "step": 174760
    },
    {
      "epoch": 1.1209204926996181,
      "grad_norm": 0.07024364918470383,
      "learning_rate": 4.817465329239776e-06,
      "loss": 0.0013,
      "step": 174770
    },
    {
      "epoch": 1.120984629593404,
      "grad_norm": 0.12400852888822556,
      "learning_rate": 4.816906000108409e-06,
      "loss": 0.0018,
      "step": 174780
    },
    {
      "epoch": 1.1210487664871902,
      "grad_norm": 0.059531401842832565,
      "learning_rate": 4.81634667327134e-06,
      "loss": 0.003,
      "step": 174790
    },
    {
      "epoch": 1.1211129033809764,
      "grad_norm": 0.17386078834533691,
      "learning_rate": 4.815787348735579e-06,
      "loss": 0.0027,
      "step": 174800
    },
    {
      "epoch": 1.1211770402747625,
      "grad_norm": 0.22123262286186218,
      "learning_rate": 4.815228026508135e-06,
      "loss": 0.0016,
      "step": 174810
    },
    {
      "epoch": 1.1212411771685487,
      "grad_norm": 0.37798458337783813,
      "learning_rate": 4.814668706596014e-06,
      "loss": 0.0156,
      "step": 174820
    },
    {
      "epoch": 1.1213053140623346,
      "grad_norm": 0.2819300889968872,
      "learning_rate": 4.814109389006228e-06,
      "loss": 0.0019,
      "step": 174830
    },
    {
      "epoch": 1.1213694509561207,
      "grad_norm": 0.1365143060684204,
      "learning_rate": 4.813550073745783e-06,
      "loss": 0.0012,
      "step": 174840
    },
    {
      "epoch": 1.1214335878499069,
      "grad_norm": 0.07883584499359131,
      "learning_rate": 4.81299076082169e-06,
      "loss": 0.0017,
      "step": 174850
    },
    {
      "epoch": 1.121497724743693,
      "grad_norm": 0.273945152759552,
      "learning_rate": 4.812431450240956e-06,
      "loss": 0.0014,
      "step": 174860
    },
    {
      "epoch": 1.121561861637479,
      "grad_norm": 0.22897343337535858,
      "learning_rate": 4.811872142010591e-06,
      "loss": 0.0017,
      "step": 174870
    },
    {
      "epoch": 1.121625998531265,
      "grad_norm": 0.1994645744562149,
      "learning_rate": 4.8113128361376e-06,
      "loss": 0.0026,
      "step": 174880
    },
    {
      "epoch": 1.1216901354250512,
      "grad_norm": 0.052137140184640884,
      "learning_rate": 4.810753532628997e-06,
      "loss": 0.0021,
      "step": 174890
    },
    {
      "epoch": 1.1217542723188374,
      "grad_norm": 0.049453653395175934,
      "learning_rate": 4.810194231491785e-06,
      "loss": 0.0008,
      "step": 174900
    },
    {
      "epoch": 1.1218184092126233,
      "grad_norm": 0.09021912515163422,
      "learning_rate": 4.809634932732976e-06,
      "loss": 0.0017,
      "step": 174910
    },
    {
      "epoch": 1.1218825461064095,
      "grad_norm": 0.11592074483633041,
      "learning_rate": 4.809075636359578e-06,
      "loss": 0.0023,
      "step": 174920
    },
    {
      "epoch": 1.1219466830001956,
      "grad_norm": 0.01169917918741703,
      "learning_rate": 4.808516342378597e-06,
      "loss": 0.0023,
      "step": 174930
    },
    {
      "epoch": 1.1220108198939818,
      "grad_norm": 0.0716211125254631,
      "learning_rate": 4.807957050797045e-06,
      "loss": 0.0035,
      "step": 174940
    },
    {
      "epoch": 1.1220749567877677,
      "grad_norm": 0.14507988095283508,
      "learning_rate": 4.807397761621926e-06,
      "loss": 0.0023,
      "step": 174950
    },
    {
      "epoch": 1.1221390936815538,
      "grad_norm": 0.04453321173787117,
      "learning_rate": 4.806838474860253e-06,
      "loss": 0.0014,
      "step": 174960
    },
    {
      "epoch": 1.12220323057534,
      "grad_norm": 0.12254642695188522,
      "learning_rate": 4.806279190519031e-06,
      "loss": 0.0012,
      "step": 174970
    },
    {
      "epoch": 1.1222673674691261,
      "grad_norm": 0.005856363568454981,
      "learning_rate": 4.80571990860527e-06,
      "loss": 0.0014,
      "step": 174980
    },
    {
      "epoch": 1.1223315043629123,
      "grad_norm": 0.06946040689945221,
      "learning_rate": 4.805160629125976e-06,
      "loss": 0.0018,
      "step": 174990
    },
    {
      "epoch": 1.1223956412566982,
      "grad_norm": 0.1354217231273651,
      "learning_rate": 4.80460135208816e-06,
      "loss": 0.0014,
      "step": 175000
    },
    {
      "epoch": 1.1224597781504844,
      "grad_norm": 0.18229800462722778,
      "learning_rate": 4.804042077498828e-06,
      "loss": 0.0035,
      "step": 175010
    },
    {
      "epoch": 1.1225239150442705,
      "grad_norm": 0.16235755383968353,
      "learning_rate": 4.803482805364989e-06,
      "loss": 0.0014,
      "step": 175020
    },
    {
      "epoch": 1.1225880519380567,
      "grad_norm": 0.3516307473182678,
      "learning_rate": 4.80292353569365e-06,
      "loss": 0.0009,
      "step": 175030
    },
    {
      "epoch": 1.1226521888318426,
      "grad_norm": 0.18743622303009033,
      "learning_rate": 4.8023642684918216e-06,
      "loss": 0.0032,
      "step": 175040
    },
    {
      "epoch": 1.1227163257256287,
      "grad_norm": 0.04604848101735115,
      "learning_rate": 4.801805003766509e-06,
      "loss": 0.002,
      "step": 175050
    },
    {
      "epoch": 1.1227804626194149,
      "grad_norm": 0.06320274621248245,
      "learning_rate": 4.801245741524721e-06,
      "loss": 0.0021,
      "step": 175060
    },
    {
      "epoch": 1.122844599513201,
      "grad_norm": 0.020902549847960472,
      "learning_rate": 4.8006864817734665e-06,
      "loss": 0.001,
      "step": 175070
    },
    {
      "epoch": 1.1229087364069872,
      "grad_norm": 0.1629502773284912,
      "learning_rate": 4.800127224519753e-06,
      "loss": 0.0024,
      "step": 175080
    },
    {
      "epoch": 1.1229728733007731,
      "grad_norm": 0.014271809719502926,
      "learning_rate": 4.799567969770588e-06,
      "loss": 0.0018,
      "step": 175090
    },
    {
      "epoch": 1.1230370101945593,
      "grad_norm": 0.11975744366645813,
      "learning_rate": 4.799008717532979e-06,
      "loss": 0.003,
      "step": 175100
    },
    {
      "epoch": 1.1231011470883454,
      "grad_norm": 0.07389917969703674,
      "learning_rate": 4.798449467813936e-06,
      "loss": 0.0012,
      "step": 175110
    },
    {
      "epoch": 1.1231652839821316,
      "grad_norm": 0.11914652585983276,
      "learning_rate": 4.7978902206204634e-06,
      "loss": 0.0012,
      "step": 175120
    },
    {
      "epoch": 1.1232294208759175,
      "grad_norm": 0.03024482913315296,
      "learning_rate": 4.797330975959572e-06,
      "loss": 0.0008,
      "step": 175130
    },
    {
      "epoch": 1.1232935577697036,
      "grad_norm": 0.04781707376241684,
      "learning_rate": 4.796771733838268e-06,
      "loss": 0.001,
      "step": 175140
    },
    {
      "epoch": 1.1233576946634898,
      "grad_norm": 0.08307913690805435,
      "learning_rate": 4.79621249426356e-06,
      "loss": 0.0023,
      "step": 175150
    },
    {
      "epoch": 1.123421831557276,
      "grad_norm": 0.0447513610124588,
      "learning_rate": 4.795653257242454e-06,
      "loss": 0.0015,
      "step": 175160
    },
    {
      "epoch": 1.1234859684510619,
      "grad_norm": 0.13397501409053802,
      "learning_rate": 4.795094022781959e-06,
      "loss": 0.0013,
      "step": 175170
    },
    {
      "epoch": 1.123550105344848,
      "grad_norm": 0.16908450424671173,
      "learning_rate": 4.794534790889083e-06,
      "loss": 0.0012,
      "step": 175180
    },
    {
      "epoch": 1.1236142422386342,
      "grad_norm": 0.2635178565979004,
      "learning_rate": 4.793975561570832e-06,
      "loss": 0.0027,
      "step": 175190
    },
    {
      "epoch": 1.1236783791324203,
      "grad_norm": 0.03389308601617813,
      "learning_rate": 4.793416334834215e-06,
      "loss": 0.0032,
      "step": 175200
    },
    {
      "epoch": 1.1237425160262062,
      "grad_norm": 0.02887616865336895,
      "learning_rate": 4.79285711068624e-06,
      "loss": 0.0022,
      "step": 175210
    },
    {
      "epoch": 1.1238066529199924,
      "grad_norm": 0.07931360602378845,
      "learning_rate": 4.792297889133911e-06,
      "loss": 0.0018,
      "step": 175220
    },
    {
      "epoch": 1.1238707898137785,
      "grad_norm": 0.035023000091314316,
      "learning_rate": 4.791738670184239e-06,
      "loss": 0.0016,
      "step": 175230
    },
    {
      "epoch": 1.1239349267075647,
      "grad_norm": 0.12552867829799652,
      "learning_rate": 4.791179453844232e-06,
      "loss": 0.0014,
      "step": 175240
    },
    {
      "epoch": 1.1239990636013508,
      "grad_norm": 0.05981716141104698,
      "learning_rate": 4.790620240120893e-06,
      "loss": 0.0018,
      "step": 175250
    },
    {
      "epoch": 1.1240632004951367,
      "grad_norm": 0.025745639577507973,
      "learning_rate": 4.7900610290212335e-06,
      "loss": 0.0021,
      "step": 175260
    },
    {
      "epoch": 1.124127337388923,
      "grad_norm": 0.05925753712654114,
      "learning_rate": 4.789501820552259e-06,
      "loss": 0.0035,
      "step": 175270
    },
    {
      "epoch": 1.124191474282709,
      "grad_norm": 0.05949115753173828,
      "learning_rate": 4.788942614720978e-06,
      "loss": 0.0027,
      "step": 175280
    },
    {
      "epoch": 1.1242556111764952,
      "grad_norm": 0.10865527391433716,
      "learning_rate": 4.7883834115343955e-06,
      "loss": 0.0018,
      "step": 175290
    },
    {
      "epoch": 1.1243197480702811,
      "grad_norm": 0.11768241971731186,
      "learning_rate": 4.787824210999521e-06,
      "loss": 0.0015,
      "step": 175300
    },
    {
      "epoch": 1.1243838849640673,
      "grad_norm": 0.012277359142899513,
      "learning_rate": 4.78726501312336e-06,
      "loss": 0.0024,
      "step": 175310
    },
    {
      "epoch": 1.1244480218578534,
      "grad_norm": 0.049140699207782745,
      "learning_rate": 4.786705817912921e-06,
      "loss": 0.0005,
      "step": 175320
    },
    {
      "epoch": 1.1245121587516396,
      "grad_norm": 0.1447220891714096,
      "learning_rate": 4.78614662537521e-06,
      "loss": 0.0016,
      "step": 175330
    },
    {
      "epoch": 1.1245762956454255,
      "grad_norm": 0.1460440307855606,
      "learning_rate": 4.785587435517236e-06,
      "loss": 0.0027,
      "step": 175340
    },
    {
      "epoch": 1.1246404325392116,
      "grad_norm": 0.17473335564136505,
      "learning_rate": 4.785028248346003e-06,
      "loss": 0.0016,
      "step": 175350
    },
    {
      "epoch": 1.1247045694329978,
      "grad_norm": 0.006431942339986563,
      "learning_rate": 4.784469063868521e-06,
      "loss": 0.0023,
      "step": 175360
    },
    {
      "epoch": 1.124768706326784,
      "grad_norm": 0.10810311138629913,
      "learning_rate": 4.783909882091795e-06,
      "loss": 0.0018,
      "step": 175370
    },
    {
      "epoch": 1.1248328432205699,
      "grad_norm": 0.04965673387050629,
      "learning_rate": 4.783350703022832e-06,
      "loss": 0.0016,
      "step": 175380
    },
    {
      "epoch": 1.124896980114356,
      "grad_norm": 0.0024077396374195814,
      "learning_rate": 4.782791526668641e-06,
      "loss": 0.0022,
      "step": 175390
    },
    {
      "epoch": 1.1249611170081422,
      "grad_norm": 0.05500560626387596,
      "learning_rate": 4.7822323530362256e-06,
      "loss": 0.001,
      "step": 175400
    },
    {
      "epoch": 1.1250252539019283,
      "grad_norm": 0.07789402455091476,
      "learning_rate": 4.781673182132596e-06,
      "loss": 0.0015,
      "step": 175410
    },
    {
      "epoch": 1.1250893907957145,
      "grad_norm": 0.07948887348175049,
      "learning_rate": 4.781114013964756e-06,
      "loss": 0.0021,
      "step": 175420
    },
    {
      "epoch": 1.1251535276895004,
      "grad_norm": 0.2515040636062622,
      "learning_rate": 4.7805548485397134e-06,
      "loss": 0.0021,
      "step": 175430
    },
    {
      "epoch": 1.1252176645832865,
      "grad_norm": 0.1324034333229065,
      "learning_rate": 4.779995685864478e-06,
      "loss": 0.0025,
      "step": 175440
    },
    {
      "epoch": 1.1252818014770727,
      "grad_norm": 0.017076067626476288,
      "learning_rate": 4.779436525946052e-06,
      "loss": 0.0008,
      "step": 175450
    },
    {
      "epoch": 1.1253459383708588,
      "grad_norm": 0.030340420082211494,
      "learning_rate": 4.778877368791445e-06,
      "loss": 0.0016,
      "step": 175460
    },
    {
      "epoch": 1.1254100752646448,
      "grad_norm": 0.004908331669867039,
      "learning_rate": 4.7783182144076616e-06,
      "loss": 0.0008,
      "step": 175470
    },
    {
      "epoch": 1.125474212158431,
      "grad_norm": 0.0785120502114296,
      "learning_rate": 4.777759062801711e-06,
      "loss": 0.0022,
      "step": 175480
    },
    {
      "epoch": 1.125538349052217,
      "grad_norm": 0.02501097321510315,
      "learning_rate": 4.777199913980596e-06,
      "loss": 0.0015,
      "step": 175490
    },
    {
      "epoch": 1.1256024859460032,
      "grad_norm": 0.06175035238265991,
      "learning_rate": 4.776640767951327e-06,
      "loss": 0.0014,
      "step": 175500
    },
    {
      "epoch": 1.1256666228397894,
      "grad_norm": 0.023419445380568504,
      "learning_rate": 4.776081624720908e-06,
      "loss": 0.0019,
      "step": 175510
    },
    {
      "epoch": 1.1257307597335753,
      "grad_norm": 0.8936617374420166,
      "learning_rate": 4.7755224842963464e-06,
      "loss": 0.004,
      "step": 175520
    },
    {
      "epoch": 1.1257948966273614,
      "grad_norm": 0.04048040136694908,
      "learning_rate": 4.77496334668465e-06,
      "loss": 0.0012,
      "step": 175530
    },
    {
      "epoch": 1.1258590335211476,
      "grad_norm": 0.027657775208353996,
      "learning_rate": 4.7744042118928215e-06,
      "loss": 0.0014,
      "step": 175540
    },
    {
      "epoch": 1.1259231704149337,
      "grad_norm": 0.06054976209998131,
      "learning_rate": 4.773845079927871e-06,
      "loss": 0.0029,
      "step": 175550
    },
    {
      "epoch": 1.1259873073087197,
      "grad_norm": 0.06709885597229004,
      "learning_rate": 4.773285950796802e-06,
      "loss": 0.0017,
      "step": 175560
    },
    {
      "epoch": 1.1260514442025058,
      "grad_norm": 0.06731562316417694,
      "learning_rate": 4.772726824506623e-06,
      "loss": 0.0015,
      "step": 175570
    },
    {
      "epoch": 1.126115581096292,
      "grad_norm": 0.023892225697636604,
      "learning_rate": 4.772167701064338e-06,
      "loss": 0.0005,
      "step": 175580
    },
    {
      "epoch": 1.126179717990078,
      "grad_norm": 0.2002870738506317,
      "learning_rate": 4.771608580476957e-06,
      "loss": 0.0005,
      "step": 175590
    },
    {
      "epoch": 1.126243854883864,
      "grad_norm": 0.02563397027552128,
      "learning_rate": 4.7710494627514815e-06,
      "loss": 0.0016,
      "step": 175600
    },
    {
      "epoch": 1.1263079917776502,
      "grad_norm": 0.13682080805301666,
      "learning_rate": 4.770490347894922e-06,
      "loss": 0.0018,
      "step": 175610
    },
    {
      "epoch": 1.1263721286714363,
      "grad_norm": 0.22487889230251312,
      "learning_rate": 4.76993123591428e-06,
      "loss": 0.002,
      "step": 175620
    },
    {
      "epoch": 1.1264362655652225,
      "grad_norm": 0.14215342700481415,
      "learning_rate": 4.7693721268165655e-06,
      "loss": 0.0016,
      "step": 175630
    },
    {
      "epoch": 1.1265004024590084,
      "grad_norm": 0.07587555795907974,
      "learning_rate": 4.768813020608782e-06,
      "loss": 0.0011,
      "step": 175640
    },
    {
      "epoch": 1.1265645393527945,
      "grad_norm": 0.10688536614179611,
      "learning_rate": 4.768253917297937e-06,
      "loss": 0.0014,
      "step": 175650
    },
    {
      "epoch": 1.1266286762465807,
      "grad_norm": 0.17090702056884766,
      "learning_rate": 4.767694816891037e-06,
      "loss": 0.0032,
      "step": 175660
    },
    {
      "epoch": 1.1266928131403668,
      "grad_norm": 0.10280928760766983,
      "learning_rate": 4.767135719395086e-06,
      "loss": 0.0013,
      "step": 175670
    },
    {
      "epoch": 1.126756950034153,
      "grad_norm": 0.03153495863080025,
      "learning_rate": 4.766576624817091e-06,
      "loss": 0.0056,
      "step": 175680
    },
    {
      "epoch": 1.126821086927939,
      "grad_norm": 0.0600753091275692,
      "learning_rate": 4.766017533164058e-06,
      "loss": 0.0013,
      "step": 175690
    },
    {
      "epoch": 1.126885223821725,
      "grad_norm": 0.0316365584731102,
      "learning_rate": 4.765458444442993e-06,
      "loss": 0.0012,
      "step": 175700
    },
    {
      "epoch": 1.1269493607155112,
      "grad_norm": 0.001820790464989841,
      "learning_rate": 4.7648993586609e-06,
      "loss": 0.0019,
      "step": 175710
    },
    {
      "epoch": 1.1270134976092974,
      "grad_norm": 0.011880267411470413,
      "learning_rate": 4.764340275824787e-06,
      "loss": 0.0015,
      "step": 175720
    },
    {
      "epoch": 1.1270776345030833,
      "grad_norm": 0.03209491819143295,
      "learning_rate": 4.763781195941659e-06,
      "loss": 0.0014,
      "step": 175730
    },
    {
      "epoch": 1.1271417713968694,
      "grad_norm": 0.22056789696216583,
      "learning_rate": 4.763222119018521e-06,
      "loss": 0.0027,
      "step": 175740
    },
    {
      "epoch": 1.1272059082906556,
      "grad_norm": 0.12804964184761047,
      "learning_rate": 4.762663045062379e-06,
      "loss": 0.0021,
      "step": 175750
    },
    {
      "epoch": 1.1272700451844417,
      "grad_norm": 0.05203420668840408,
      "learning_rate": 4.762103974080239e-06,
      "loss": 0.0027,
      "step": 175760
    },
    {
      "epoch": 1.1273341820782279,
      "grad_norm": 0.06643746048212051,
      "learning_rate": 4.761544906079106e-06,
      "loss": 0.0018,
      "step": 175770
    },
    {
      "epoch": 1.1273983189720138,
      "grad_norm": 0.16261453926563263,
      "learning_rate": 4.760985841065985e-06,
      "loss": 0.0013,
      "step": 175780
    },
    {
      "epoch": 1.1274624558658,
      "grad_norm": 0.06322963535785675,
      "learning_rate": 4.760426779047884e-06,
      "loss": 0.002,
      "step": 175790
    },
    {
      "epoch": 1.127526592759586,
      "grad_norm": 0.24147634208202362,
      "learning_rate": 4.759867720031804e-06,
      "loss": 0.0012,
      "step": 175800
    },
    {
      "epoch": 1.127590729653372,
      "grad_norm": 0.18048495054244995,
      "learning_rate": 4.759308664024756e-06,
      "loss": 0.0011,
      "step": 175810
    },
    {
      "epoch": 1.1276548665471582,
      "grad_norm": 0.07934437692165375,
      "learning_rate": 4.75874961103374e-06,
      "loss": 0.0018,
      "step": 175820
    },
    {
      "epoch": 1.1277190034409443,
      "grad_norm": 0.08311376720666885,
      "learning_rate": 4.758190561065766e-06,
      "loss": 0.0017,
      "step": 175830
    },
    {
      "epoch": 1.1277831403347305,
      "grad_norm": 0.2344166338443756,
      "learning_rate": 4.757631514127836e-06,
      "loss": 0.0021,
      "step": 175840
    },
    {
      "epoch": 1.1278472772285166,
      "grad_norm": 0.11134286969900131,
      "learning_rate": 4.757072470226957e-06,
      "loss": 0.0013,
      "step": 175850
    },
    {
      "epoch": 1.1279114141223026,
      "grad_norm": 0.04196736961603165,
      "learning_rate": 4.756513429370132e-06,
      "loss": 0.0019,
      "step": 175860
    },
    {
      "epoch": 1.1279755510160887,
      "grad_norm": 0.07570148259401321,
      "learning_rate": 4.755954391564369e-06,
      "loss": 0.0013,
      "step": 175870
    },
    {
      "epoch": 1.1280396879098749,
      "grad_norm": 0.14124351739883423,
      "learning_rate": 4.755395356816672e-06,
      "loss": 0.0011,
      "step": 175880
    },
    {
      "epoch": 1.128103824803661,
      "grad_norm": 0.03067854233086109,
      "learning_rate": 4.754836325134046e-06,
      "loss": 0.0011,
      "step": 175890
    },
    {
      "epoch": 1.128167961697447,
      "grad_norm": 0.018530651926994324,
      "learning_rate": 4.7542772965234945e-06,
      "loss": 0.0009,
      "step": 175900
    },
    {
      "epoch": 1.128232098591233,
      "grad_norm": 0.13074007630348206,
      "learning_rate": 4.753718270992026e-06,
      "loss": 0.0011,
      "step": 175910
    },
    {
      "epoch": 1.1282962354850192,
      "grad_norm": 0.033416058868169785,
      "learning_rate": 4.753159248546643e-06,
      "loss": 0.0011,
      "step": 175920
    },
    {
      "epoch": 1.1283603723788054,
      "grad_norm": 0.06768912076950073,
      "learning_rate": 4.75260022919435e-06,
      "loss": 0.0011,
      "step": 175930
    },
    {
      "epoch": 1.1284245092725915,
      "grad_norm": 0.3239620625972748,
      "learning_rate": 4.752041212942154e-06,
      "loss": 0.0018,
      "step": 175940
    },
    {
      "epoch": 1.1284886461663775,
      "grad_norm": 0.14764080941677094,
      "learning_rate": 4.7514821997970585e-06,
      "loss": 0.0022,
      "step": 175950
    },
    {
      "epoch": 1.1285527830601636,
      "grad_norm": 0.09093287587165833,
      "learning_rate": 4.750923189766069e-06,
      "loss": 0.0018,
      "step": 175960
    },
    {
      "epoch": 1.1286169199539497,
      "grad_norm": 0.10408314317464828,
      "learning_rate": 4.7503641828561895e-06,
      "loss": 0.0033,
      "step": 175970
    },
    {
      "epoch": 1.128681056847736,
      "grad_norm": 0.5459677577018738,
      "learning_rate": 4.749805179074426e-06,
      "loss": 0.0024,
      "step": 175980
    },
    {
      "epoch": 1.1287451937415218,
      "grad_norm": 0.14407488703727722,
      "learning_rate": 4.749246178427781e-06,
      "loss": 0.0016,
      "step": 175990
    },
    {
      "epoch": 1.128809330635308,
      "grad_norm": 0.08890324831008911,
      "learning_rate": 4.748687180923262e-06,
      "loss": 0.0014,
      "step": 176000
    },
    {
      "epoch": 1.1288734675290941,
      "grad_norm": 0.1855904757976532,
      "learning_rate": 4.748128186567871e-06,
      "loss": 0.0029,
      "step": 176010
    },
    {
      "epoch": 1.1289376044228803,
      "grad_norm": 0.14686384797096252,
      "learning_rate": 4.7475691953686154e-06,
      "loss": 0.0011,
      "step": 176020
    },
    {
      "epoch": 1.1290017413166662,
      "grad_norm": 0.01927570253610611,
      "learning_rate": 4.747010207332497e-06,
      "loss": 0.0017,
      "step": 176030
    },
    {
      "epoch": 1.1290658782104523,
      "grad_norm": 0.09091423451900482,
      "learning_rate": 4.7464512224665225e-06,
      "loss": 0.0017,
      "step": 176040
    },
    {
      "epoch": 1.1291300151042385,
      "grad_norm": 0.028048235923051834,
      "learning_rate": 4.745892240777694e-06,
      "loss": 0.002,
      "step": 176050
    },
    {
      "epoch": 1.1291941519980246,
      "grad_norm": 0.024108463898301125,
      "learning_rate": 4.745333262273019e-06,
      "loss": 0.003,
      "step": 176060
    },
    {
      "epoch": 1.1292582888918106,
      "grad_norm": 0.14120781421661377,
      "learning_rate": 4.7447742869594986e-06,
      "loss": 0.0033,
      "step": 176070
    },
    {
      "epoch": 1.1293224257855967,
      "grad_norm": 0.01841743476688862,
      "learning_rate": 4.744215314844138e-06,
      "loss": 0.0014,
      "step": 176080
    },
    {
      "epoch": 1.1293865626793829,
      "grad_norm": 0.017169052734971046,
      "learning_rate": 4.743656345933945e-06,
      "loss": 0.0026,
      "step": 176090
    },
    {
      "epoch": 1.129450699573169,
      "grad_norm": 0.13318702578544617,
      "learning_rate": 4.74309738023592e-06,
      "loss": 0.0017,
      "step": 176100
    },
    {
      "epoch": 1.1295148364669552,
      "grad_norm": 0.1167878583073616,
      "learning_rate": 4.7425384177570686e-06,
      "loss": 0.0012,
      "step": 176110
    },
    {
      "epoch": 1.129578973360741,
      "grad_norm": 0.008401491679251194,
      "learning_rate": 4.741979458504394e-06,
      "loss": 0.0013,
      "step": 176120
    },
    {
      "epoch": 1.1296431102545272,
      "grad_norm": 0.0819748267531395,
      "learning_rate": 4.741420502484903e-06,
      "loss": 0.0024,
      "step": 176130
    },
    {
      "epoch": 1.1297072471483134,
      "grad_norm": 0.004283037036657333,
      "learning_rate": 4.740861549705595e-06,
      "loss": 0.0022,
      "step": 176140
    },
    {
      "epoch": 1.1297713840420995,
      "grad_norm": 0.061530210077762604,
      "learning_rate": 4.740302600173479e-06,
      "loss": 0.002,
      "step": 176150
    },
    {
      "epoch": 1.1298355209358855,
      "grad_norm": 0.0016988730058073997,
      "learning_rate": 4.739743653895557e-06,
      "loss": 0.0007,
      "step": 176160
    },
    {
      "epoch": 1.1298996578296716,
      "grad_norm": 0.10308869183063507,
      "learning_rate": 4.739184710878833e-06,
      "loss": 0.0014,
      "step": 176170
    },
    {
      "epoch": 1.1299637947234578,
      "grad_norm": 0.01512821763753891,
      "learning_rate": 4.73862577113031e-06,
      "loss": 0.0034,
      "step": 176180
    },
    {
      "epoch": 1.130027931617244,
      "grad_norm": 0.10347221046686172,
      "learning_rate": 4.738066834656993e-06,
      "loss": 0.0022,
      "step": 176190
    },
    {
      "epoch": 1.13009206851103,
      "grad_norm": 0.13395991921424866,
      "learning_rate": 4.737507901465886e-06,
      "loss": 0.002,
      "step": 176200
    },
    {
      "epoch": 1.130156205404816,
      "grad_norm": 0.08036302775144577,
      "learning_rate": 4.736948971563993e-06,
      "loss": 0.0018,
      "step": 176210
    },
    {
      "epoch": 1.1302203422986021,
      "grad_norm": 0.05972522497177124,
      "learning_rate": 4.736390044958317e-06,
      "loss": 0.0018,
      "step": 176220
    },
    {
      "epoch": 1.1302844791923883,
      "grad_norm": 0.3460785746574402,
      "learning_rate": 4.735831121655863e-06,
      "loss": 0.003,
      "step": 176230
    },
    {
      "epoch": 1.1303486160861742,
      "grad_norm": 0.08491852134466171,
      "learning_rate": 4.735272201663633e-06,
      "loss": 0.0011,
      "step": 176240
    },
    {
      "epoch": 1.1304127529799604,
      "grad_norm": 0.0766621083021164,
      "learning_rate": 4.7347132849886315e-06,
      "loss": 0.0011,
      "step": 176250
    },
    {
      "epoch": 1.1304768898737465,
      "grad_norm": 0.054548390209674835,
      "learning_rate": 4.7341543716378625e-06,
      "loss": 0.0022,
      "step": 176260
    },
    {
      "epoch": 1.1305410267675327,
      "grad_norm": 0.04001692309975624,
      "learning_rate": 4.73359546161833e-06,
      "loss": 0.0009,
      "step": 176270
    },
    {
      "epoch": 1.1306051636613188,
      "grad_norm": 0.11134858429431915,
      "learning_rate": 4.733036554937035e-06,
      "loss": 0.0021,
      "step": 176280
    },
    {
      "epoch": 1.1306693005551047,
      "grad_norm": 0.14896228909492493,
      "learning_rate": 4.732477651600985e-06,
      "loss": 0.001,
      "step": 176290
    },
    {
      "epoch": 1.1307334374488909,
      "grad_norm": 0.07890468835830688,
      "learning_rate": 4.731918751617179e-06,
      "loss": 0.0025,
      "step": 176300
    },
    {
      "epoch": 1.130797574342677,
      "grad_norm": 0.07719369977712631,
      "learning_rate": 4.731359854992624e-06,
      "loss": 0.0018,
      "step": 176310
    },
    {
      "epoch": 1.1308617112364632,
      "grad_norm": 0.11463964730501175,
      "learning_rate": 4.730800961734321e-06,
      "loss": 0.0012,
      "step": 176320
    },
    {
      "epoch": 1.130925848130249,
      "grad_norm": 0.024333612993359566,
      "learning_rate": 4.730242071849275e-06,
      "loss": 0.0018,
      "step": 176330
    },
    {
      "epoch": 1.1309899850240352,
      "grad_norm": 0.32956039905548096,
      "learning_rate": 4.729683185344488e-06,
      "loss": 0.0019,
      "step": 176340
    },
    {
      "epoch": 1.1310541219178214,
      "grad_norm": 0.07506587356328964,
      "learning_rate": 4.729124302226965e-06,
      "loss": 0.0013,
      "step": 176350
    },
    {
      "epoch": 1.1311182588116075,
      "grad_norm": 0.08916376531124115,
      "learning_rate": 4.728565422503708e-06,
      "loss": 0.0012,
      "step": 176360
    },
    {
      "epoch": 1.1311823957053937,
      "grad_norm": 0.01957746036350727,
      "learning_rate": 4.728006546181718e-06,
      "loss": 0.0009,
      "step": 176370
    },
    {
      "epoch": 1.1312465325991796,
      "grad_norm": 0.22332695126533508,
      "learning_rate": 4.727447673268003e-06,
      "loss": 0.003,
      "step": 176380
    },
    {
      "epoch": 1.1313106694929658,
      "grad_norm": 0.04445593059062958,
      "learning_rate": 4.726888803769562e-06,
      "loss": 0.0015,
      "step": 176390
    },
    {
      "epoch": 1.131374806386752,
      "grad_norm": 0.15768571197986603,
      "learning_rate": 4.726329937693401e-06,
      "loss": 0.0015,
      "step": 176400
    },
    {
      "epoch": 1.131438943280538,
      "grad_norm": 0.013671383261680603,
      "learning_rate": 4.72577107504652e-06,
      "loss": 0.0014,
      "step": 176410
    },
    {
      "epoch": 1.131503080174324,
      "grad_norm": 0.3652510941028595,
      "learning_rate": 4.725212215835925e-06,
      "loss": 0.0021,
      "step": 176420
    },
    {
      "epoch": 1.1315672170681101,
      "grad_norm": 0.15738549828529358,
      "learning_rate": 4.724653360068616e-06,
      "loss": 0.0019,
      "step": 176430
    },
    {
      "epoch": 1.1316313539618963,
      "grad_norm": 0.09711114317178726,
      "learning_rate": 4.724094507751598e-06,
      "loss": 0.0062,
      "step": 176440
    },
    {
      "epoch": 1.1316954908556824,
      "grad_norm": 0.2600592076778412,
      "learning_rate": 4.723535658891872e-06,
      "loss": 0.002,
      "step": 176450
    },
    {
      "epoch": 1.1317596277494686,
      "grad_norm": 0.09903906285762787,
      "learning_rate": 4.722976813496443e-06,
      "loss": 0.0024,
      "step": 176460
    },
    {
      "epoch": 1.1318237646432545,
      "grad_norm": 0.09826955199241638,
      "learning_rate": 4.7224179715723105e-06,
      "loss": 0.0022,
      "step": 176470
    },
    {
      "epoch": 1.1318879015370407,
      "grad_norm": 0.22338512539863586,
      "learning_rate": 4.721859133126481e-06,
      "loss": 0.0068,
      "step": 176480
    },
    {
      "epoch": 1.1319520384308268,
      "grad_norm": 0.08771412074565887,
      "learning_rate": 4.721300298165954e-06,
      "loss": 0.0006,
      "step": 176490
    },
    {
      "epoch": 1.1320161753246127,
      "grad_norm": 0.16829173266887665,
      "learning_rate": 4.720741466697734e-06,
      "loss": 0.0012,
      "step": 176500
    },
    {
      "epoch": 1.1320803122183989,
      "grad_norm": 0.15693411231040955,
      "learning_rate": 4.7201826387288245e-06,
      "loss": 0.0015,
      "step": 176510
    },
    {
      "epoch": 1.132144449112185,
      "grad_norm": 0.20173071324825287,
      "learning_rate": 4.719623814266224e-06,
      "loss": 0.0013,
      "step": 176520
    },
    {
      "epoch": 1.1322085860059712,
      "grad_norm": 0.1597852110862732,
      "learning_rate": 4.71906499331694e-06,
      "loss": 0.002,
      "step": 176530
    },
    {
      "epoch": 1.1322727228997573,
      "grad_norm": 0.08356647938489914,
      "learning_rate": 4.71850617588797e-06,
      "loss": 0.0028,
      "step": 176540
    },
    {
      "epoch": 1.1323368597935433,
      "grad_norm": 0.07380029559135437,
      "learning_rate": 4.717947361986321e-06,
      "loss": 0.0017,
      "step": 176550
    },
    {
      "epoch": 1.1324009966873294,
      "grad_norm": 0.07805683463811874,
      "learning_rate": 4.717388551618992e-06,
      "loss": 0.0009,
      "step": 176560
    },
    {
      "epoch": 1.1324651335811156,
      "grad_norm": 0.11640407145023346,
      "learning_rate": 4.716829744792987e-06,
      "loss": 0.0023,
      "step": 176570
    },
    {
      "epoch": 1.1325292704749017,
      "grad_norm": 0.021123992279171944,
      "learning_rate": 4.7162709415153065e-06,
      "loss": 0.0025,
      "step": 176580
    },
    {
      "epoch": 1.1325934073686876,
      "grad_norm": 0.048886850476264954,
      "learning_rate": 4.715712141792955e-06,
      "loss": 0.0017,
      "step": 176590
    },
    {
      "epoch": 1.1326575442624738,
      "grad_norm": 0.08189143240451813,
      "learning_rate": 4.715153345632933e-06,
      "loss": 0.0026,
      "step": 176600
    },
    {
      "epoch": 1.13272168115626,
      "grad_norm": 0.047019582241773605,
      "learning_rate": 4.714594553042244e-06,
      "loss": 0.0009,
      "step": 176610
    },
    {
      "epoch": 1.132785818050046,
      "grad_norm": 0.0060059260576963425,
      "learning_rate": 4.7140357640278875e-06,
      "loss": 0.0019,
      "step": 176620
    },
    {
      "epoch": 1.1328499549438322,
      "grad_norm": 0.16405409574508667,
      "learning_rate": 4.7134769785968685e-06,
      "loss": 0.0017,
      "step": 176630
    },
    {
      "epoch": 1.1329140918376182,
      "grad_norm": 0.014878307469189167,
      "learning_rate": 4.7129181967561876e-06,
      "loss": 0.0011,
      "step": 176640
    },
    {
      "epoch": 1.1329782287314043,
      "grad_norm": 0.07795178890228271,
      "learning_rate": 4.712359418512846e-06,
      "loss": 0.001,
      "step": 176650
    },
    {
      "epoch": 1.1330423656251905,
      "grad_norm": 0.2748396694660187,
      "learning_rate": 4.711800643873848e-06,
      "loss": 0.0021,
      "step": 176660
    },
    {
      "epoch": 1.1331065025189766,
      "grad_norm": 0.018300451338291168,
      "learning_rate": 4.711241872846193e-06,
      "loss": 0.002,
      "step": 176670
    },
    {
      "epoch": 1.1331706394127625,
      "grad_norm": 0.03134751319885254,
      "learning_rate": 4.710683105436884e-06,
      "loss": 0.0008,
      "step": 176680
    },
    {
      "epoch": 1.1332347763065487,
      "grad_norm": 0.10682273656129837,
      "learning_rate": 4.710124341652922e-06,
      "loss": 0.0015,
      "step": 176690
    },
    {
      "epoch": 1.1332989132003348,
      "grad_norm": 0.17113542556762695,
      "learning_rate": 4.70956558150131e-06,
      "loss": 0.0019,
      "step": 176700
    },
    {
      "epoch": 1.133363050094121,
      "grad_norm": 0.26188206672668457,
      "learning_rate": 4.709006824989048e-06,
      "loss": 0.0031,
      "step": 176710
    },
    {
      "epoch": 1.133427186987907,
      "grad_norm": 0.47190165519714355,
      "learning_rate": 4.708448072123139e-06,
      "loss": 0.0017,
      "step": 176720
    },
    {
      "epoch": 1.133491323881693,
      "grad_norm": 0.15740633010864258,
      "learning_rate": 4.707889322910584e-06,
      "loss": 0.0039,
      "step": 176730
    },
    {
      "epoch": 1.1335554607754792,
      "grad_norm": 0.0610303096473217,
      "learning_rate": 4.707330577358385e-06,
      "loss": 0.0009,
      "step": 176740
    },
    {
      "epoch": 1.1336195976692653,
      "grad_norm": 0.05434420332312584,
      "learning_rate": 4.7067718354735405e-06,
      "loss": 0.0014,
      "step": 176750
    },
    {
      "epoch": 1.1336837345630513,
      "grad_norm": 0.1731656938791275,
      "learning_rate": 4.706213097263057e-06,
      "loss": 0.0019,
      "step": 176760
    },
    {
      "epoch": 1.1337478714568374,
      "grad_norm": 0.044647444039583206,
      "learning_rate": 4.705654362733933e-06,
      "loss": 0.0009,
      "step": 176770
    },
    {
      "epoch": 1.1338120083506236,
      "grad_norm": 0.08089784532785416,
      "learning_rate": 4.705095631893169e-06,
      "loss": 0.0095,
      "step": 176780
    },
    {
      "epoch": 1.1338761452444097,
      "grad_norm": 0.09667832404375076,
      "learning_rate": 4.704536904747769e-06,
      "loss": 0.0039,
      "step": 176790
    },
    {
      "epoch": 1.1339402821381959,
      "grad_norm": 0.06942083686590195,
      "learning_rate": 4.703978181304732e-06,
      "loss": 0.0008,
      "step": 176800
    },
    {
      "epoch": 1.1340044190319818,
      "grad_norm": 0.06267499178647995,
      "learning_rate": 4.70341946157106e-06,
      "loss": 0.0018,
      "step": 176810
    },
    {
      "epoch": 1.134068555925768,
      "grad_norm": 0.021174170076847076,
      "learning_rate": 4.702860745553754e-06,
      "loss": 0.0011,
      "step": 176820
    },
    {
      "epoch": 1.134132692819554,
      "grad_norm": 0.13701143860816956,
      "learning_rate": 4.702302033259816e-06,
      "loss": 0.001,
      "step": 176830
    },
    {
      "epoch": 1.1341968297133402,
      "grad_norm": 0.26283982396125793,
      "learning_rate": 4.701743324696245e-06,
      "loss": 0.0023,
      "step": 176840
    },
    {
      "epoch": 1.1342609666071262,
      "grad_norm": 0.06547980010509491,
      "learning_rate": 4.701184619870045e-06,
      "loss": 0.0007,
      "step": 176850
    },
    {
      "epoch": 1.1343251035009123,
      "grad_norm": 0.028654005378484726,
      "learning_rate": 4.700625918788214e-06,
      "loss": 0.0015,
      "step": 176860
    },
    {
      "epoch": 1.1343892403946985,
      "grad_norm": 0.037154026329517365,
      "learning_rate": 4.700067221457755e-06,
      "loss": 0.0013,
      "step": 176870
    },
    {
      "epoch": 1.1344533772884846,
      "grad_norm": 0.15115104615688324,
      "learning_rate": 4.699508527885667e-06,
      "loss": 0.0015,
      "step": 176880
    },
    {
      "epoch": 1.1345175141822708,
      "grad_norm": 0.082732193171978,
      "learning_rate": 4.6989498380789525e-06,
      "loss": 0.0008,
      "step": 176890
    },
    {
      "epoch": 1.1345816510760567,
      "grad_norm": 0.10726206004619598,
      "learning_rate": 4.698391152044611e-06,
      "loss": 0.0012,
      "step": 176900
    },
    {
      "epoch": 1.1346457879698428,
      "grad_norm": 0.08087017387151718,
      "learning_rate": 4.697832469789646e-06,
      "loss": 0.0018,
      "step": 176910
    },
    {
      "epoch": 1.134709924863629,
      "grad_norm": 0.012344476766884327,
      "learning_rate": 4.697273791321054e-06,
      "loss": 0.0028,
      "step": 176920
    },
    {
      "epoch": 1.134774061757415,
      "grad_norm": 0.07980614900588989,
      "learning_rate": 4.696715116645838e-06,
      "loss": 0.001,
      "step": 176930
    },
    {
      "epoch": 1.134838198651201,
      "grad_norm": 0.05606852471828461,
      "learning_rate": 4.696156445770999e-06,
      "loss": 0.004,
      "step": 176940
    },
    {
      "epoch": 1.1349023355449872,
      "grad_norm": 0.2644835412502289,
      "learning_rate": 4.695597778703535e-06,
      "loss": 0.0015,
      "step": 176950
    },
    {
      "epoch": 1.1349664724387734,
      "grad_norm": 0.17666363716125488,
      "learning_rate": 4.695039115450449e-06,
      "loss": 0.0015,
      "step": 176960
    },
    {
      "epoch": 1.1350306093325595,
      "grad_norm": 0.02572619542479515,
      "learning_rate": 4.694480456018743e-06,
      "loss": 0.0009,
      "step": 176970
    },
    {
      "epoch": 1.1350947462263454,
      "grad_norm": 0.07502380013465881,
      "learning_rate": 4.693921800415413e-06,
      "loss": 0.0018,
      "step": 176980
    },
    {
      "epoch": 1.1351588831201316,
      "grad_norm": 0.03207005187869072,
      "learning_rate": 4.693363148647464e-06,
      "loss": 0.0011,
      "step": 176990
    },
    {
      "epoch": 1.1352230200139177,
      "grad_norm": 0.25719138979911804,
      "learning_rate": 4.692804500721892e-06,
      "loss": 0.0016,
      "step": 177000
    },
    {
      "epoch": 1.1352871569077039,
      "grad_norm": 0.0666014552116394,
      "learning_rate": 4.6922458566457e-06,
      "loss": 0.0025,
      "step": 177010
    },
    {
      "epoch": 1.1353512938014898,
      "grad_norm": 0.14155615866184235,
      "learning_rate": 4.691687216425887e-06,
      "loss": 0.0026,
      "step": 177020
    },
    {
      "epoch": 1.135415430695276,
      "grad_norm": 0.02840949408710003,
      "learning_rate": 4.691128580069455e-06,
      "loss": 0.0018,
      "step": 177030
    },
    {
      "epoch": 1.135479567589062,
      "grad_norm": 0.06482551246881485,
      "learning_rate": 4.6905699475834e-06,
      "loss": 0.0015,
      "step": 177040
    },
    {
      "epoch": 1.1355437044828482,
      "grad_norm": 0.20891007781028748,
      "learning_rate": 4.6900113189747255e-06,
      "loss": 0.0017,
      "step": 177050
    },
    {
      "epoch": 1.1356078413766344,
      "grad_norm": 0.15411852300167084,
      "learning_rate": 4.689452694250432e-06,
      "loss": 0.0026,
      "step": 177060
    },
    {
      "epoch": 1.1356719782704203,
      "grad_norm": 0.08161000162363052,
      "learning_rate": 4.688894073417517e-06,
      "loss": 0.0032,
      "step": 177070
    },
    {
      "epoch": 1.1357361151642065,
      "grad_norm": 0.1426655650138855,
      "learning_rate": 4.688335456482983e-06,
      "loss": 0.0015,
      "step": 177080
    },
    {
      "epoch": 1.1358002520579926,
      "grad_norm": 0.145971417427063,
      "learning_rate": 4.687776843453826e-06,
      "loss": 0.0017,
      "step": 177090
    },
    {
      "epoch": 1.1358643889517788,
      "grad_norm": 0.07309820502996445,
      "learning_rate": 4.687218234337051e-06,
      "loss": 0.0014,
      "step": 177100
    },
    {
      "epoch": 1.1359285258455647,
      "grad_norm": 0.16422462463378906,
      "learning_rate": 4.686659629139652e-06,
      "loss": 0.0052,
      "step": 177110
    },
    {
      "epoch": 1.1359926627393508,
      "grad_norm": 0.05093219876289368,
      "learning_rate": 4.686101027868635e-06,
      "loss": 0.0023,
      "step": 177120
    },
    {
      "epoch": 1.136056799633137,
      "grad_norm": 0.05091457441449165,
      "learning_rate": 4.685542430530995e-06,
      "loss": 0.003,
      "step": 177130
    },
    {
      "epoch": 1.1361209365269231,
      "grad_norm": 0.1133175641298294,
      "learning_rate": 4.6849838371337335e-06,
      "loss": 0.002,
      "step": 177140
    },
    {
      "epoch": 1.136185073420709,
      "grad_norm": 0.06035906448960304,
      "learning_rate": 4.684425247683849e-06,
      "loss": 0.0018,
      "step": 177150
    },
    {
      "epoch": 1.1362492103144952,
      "grad_norm": 0.07222796976566315,
      "learning_rate": 4.683866662188343e-06,
      "loss": 0.0018,
      "step": 177160
    },
    {
      "epoch": 1.1363133472082814,
      "grad_norm": 0.009993140585720539,
      "learning_rate": 4.683308080654211e-06,
      "loss": 0.0013,
      "step": 177170
    },
    {
      "epoch": 1.1363774841020675,
      "grad_norm": 0.08702369034290314,
      "learning_rate": 4.682749503088458e-06,
      "loss": 0.0015,
      "step": 177180
    },
    {
      "epoch": 1.1364416209958534,
      "grad_norm": 0.11006546020507812,
      "learning_rate": 4.6821909294980785e-06,
      "loss": 0.0018,
      "step": 177190
    },
    {
      "epoch": 1.1365057578896396,
      "grad_norm": 0.05190633237361908,
      "learning_rate": 4.6816323598900745e-06,
      "loss": 0.0013,
      "step": 177200
    },
    {
      "epoch": 1.1365698947834257,
      "grad_norm": 0.17885461449623108,
      "learning_rate": 4.681073794271445e-06,
      "loss": 0.0017,
      "step": 177210
    },
    {
      "epoch": 1.1366340316772119,
      "grad_norm": 0.07297784090042114,
      "learning_rate": 4.680515232649188e-06,
      "loss": 0.0014,
      "step": 177220
    },
    {
      "epoch": 1.136698168570998,
      "grad_norm": 0.05929350480437279,
      "learning_rate": 4.679956675030304e-06,
      "loss": 0.0024,
      "step": 177230
    },
    {
      "epoch": 1.136762305464784,
      "grad_norm": 0.06213223561644554,
      "learning_rate": 4.679398121421791e-06,
      "loss": 0.0012,
      "step": 177240
    },
    {
      "epoch": 1.13682644235857,
      "grad_norm": 0.05396553874015808,
      "learning_rate": 4.678839571830649e-06,
      "loss": 0.0009,
      "step": 177250
    },
    {
      "epoch": 1.1368905792523563,
      "grad_norm": 0.20179583132266998,
      "learning_rate": 4.678281026263876e-06,
      "loss": 0.0025,
      "step": 177260
    },
    {
      "epoch": 1.1369547161461424,
      "grad_norm": 0.08537101000547409,
      "learning_rate": 4.6777224847284745e-06,
      "loss": 0.0014,
      "step": 177270
    },
    {
      "epoch": 1.1370188530399283,
      "grad_norm": 0.05290082469582558,
      "learning_rate": 4.677163947231438e-06,
      "loss": 0.0019,
      "step": 177280
    },
    {
      "epoch": 1.1370829899337145,
      "grad_norm": 0.04260869696736336,
      "learning_rate": 4.676605413779769e-06,
      "loss": 0.002,
      "step": 177290
    },
    {
      "epoch": 1.1371471268275006,
      "grad_norm": 0.024307915940880775,
      "learning_rate": 4.676046884380464e-06,
      "loss": 0.0014,
      "step": 177300
    },
    {
      "epoch": 1.1372112637212868,
      "grad_norm": 0.05193071812391281,
      "learning_rate": 4.675488359040524e-06,
      "loss": 0.0091,
      "step": 177310
    },
    {
      "epoch": 1.137275400615073,
      "grad_norm": 0.07003153860569,
      "learning_rate": 4.6749298377669464e-06,
      "loss": 0.0022,
      "step": 177320
    },
    {
      "epoch": 1.1373395375088589,
      "grad_norm": 0.0878596380352974,
      "learning_rate": 4.674371320566731e-06,
      "loss": 0.0011,
      "step": 177330
    },
    {
      "epoch": 1.137403674402645,
      "grad_norm": 0.019946128129959106,
      "learning_rate": 4.673812807446875e-06,
      "loss": 0.0011,
      "step": 177340
    },
    {
      "epoch": 1.1374678112964312,
      "grad_norm": 0.08865027874708176,
      "learning_rate": 4.673254298414378e-06,
      "loss": 0.0014,
      "step": 177350
    },
    {
      "epoch": 1.137531948190217,
      "grad_norm": 0.060385555028915405,
      "learning_rate": 4.672695793476238e-06,
      "loss": 0.0015,
      "step": 177360
    },
    {
      "epoch": 1.1375960850840032,
      "grad_norm": 0.05343933776021004,
      "learning_rate": 4.672137292639453e-06,
      "loss": 0.001,
      "step": 177370
    },
    {
      "epoch": 1.1376602219777894,
      "grad_norm": 0.1751498132944107,
      "learning_rate": 4.6715787959110235e-06,
      "loss": 0.0019,
      "step": 177380
    },
    {
      "epoch": 1.1377243588715755,
      "grad_norm": 0.06436331570148468,
      "learning_rate": 4.671020303297945e-06,
      "loss": 0.0013,
      "step": 177390
    },
    {
      "epoch": 1.1377884957653617,
      "grad_norm": 0.1137598305940628,
      "learning_rate": 4.670461814807218e-06,
      "loss": 0.0021,
      "step": 177400
    },
    {
      "epoch": 1.1378526326591476,
      "grad_norm": 0.06752078235149384,
      "learning_rate": 4.66990333044584e-06,
      "loss": 0.0014,
      "step": 177410
    },
    {
      "epoch": 1.1379167695529337,
      "grad_norm": 0.056894451379776,
      "learning_rate": 4.669344850220809e-06,
      "loss": 0.0036,
      "step": 177420
    },
    {
      "epoch": 1.13798090644672,
      "grad_norm": 0.14695769548416138,
      "learning_rate": 4.668786374139123e-06,
      "loss": 0.0023,
      "step": 177430
    },
    {
      "epoch": 1.138045043340506,
      "grad_norm": 0.14649368822574615,
      "learning_rate": 4.668227902207782e-06,
      "loss": 0.0016,
      "step": 177440
    },
    {
      "epoch": 1.138109180234292,
      "grad_norm": 0.06601706147193909,
      "learning_rate": 4.66766943443378e-06,
      "loss": 0.0014,
      "step": 177450
    },
    {
      "epoch": 1.1381733171280781,
      "grad_norm": 0.05942267179489136,
      "learning_rate": 4.667110970824119e-06,
      "loss": 0.0006,
      "step": 177460
    },
    {
      "epoch": 1.1382374540218643,
      "grad_norm": 0.14742408692836761,
      "learning_rate": 4.666552511385795e-06,
      "loss": 0.0018,
      "step": 177470
    },
    {
      "epoch": 1.1383015909156504,
      "grad_norm": 0.1130622997879982,
      "learning_rate": 4.665994056125806e-06,
      "loss": 0.0011,
      "step": 177480
    },
    {
      "epoch": 1.1383657278094366,
      "grad_norm": 0.03683945909142494,
      "learning_rate": 4.665435605051151e-06,
      "loss": 0.0038,
      "step": 177490
    },
    {
      "epoch": 1.1384298647032225,
      "grad_norm": 0.07685775309801102,
      "learning_rate": 4.664877158168827e-06,
      "loss": 0.0008,
      "step": 177500
    },
    {
      "epoch": 1.1384940015970086,
      "grad_norm": 0.195655956864357,
      "learning_rate": 4.664318715485832e-06,
      "loss": 0.0012,
      "step": 177510
    },
    {
      "epoch": 1.1385581384907948,
      "grad_norm": 0.002429707907140255,
      "learning_rate": 4.6637602770091614e-06,
      "loss": 0.0035,
      "step": 177520
    },
    {
      "epoch": 1.138622275384581,
      "grad_norm": 0.09719683229923248,
      "learning_rate": 4.663201842745818e-06,
      "loss": 0.0013,
      "step": 177530
    },
    {
      "epoch": 1.1386864122783669,
      "grad_norm": 0.03722328320145607,
      "learning_rate": 4.662643412702793e-06,
      "loss": 0.002,
      "step": 177540
    },
    {
      "epoch": 1.138750549172153,
      "grad_norm": 0.015547439455986023,
      "learning_rate": 4.662084986887089e-06,
      "loss": 0.0016,
      "step": 177550
    },
    {
      "epoch": 1.1388146860659392,
      "grad_norm": 0.11467263847589493,
      "learning_rate": 4.661526565305701e-06,
      "loss": 0.0014,
      "step": 177560
    },
    {
      "epoch": 1.1388788229597253,
      "grad_norm": 0.06219152361154556,
      "learning_rate": 4.660968147965628e-06,
      "loss": 0.0024,
      "step": 177570
    },
    {
      "epoch": 1.1389429598535112,
      "grad_norm": 0.05296056717634201,
      "learning_rate": 4.6604097348738655e-06,
      "loss": 0.0015,
      "step": 177580
    },
    {
      "epoch": 1.1390070967472974,
      "grad_norm": 0.0316719226539135,
      "learning_rate": 4.6598513260374125e-06,
      "loss": 0.0018,
      "step": 177590
    },
    {
      "epoch": 1.1390712336410835,
      "grad_norm": 0.10980919748544693,
      "learning_rate": 4.659292921463264e-06,
      "loss": 0.0008,
      "step": 177600
    },
    {
      "epoch": 1.1391353705348697,
      "grad_norm": 0.09972270578145981,
      "learning_rate": 4.658734521158421e-06,
      "loss": 0.0014,
      "step": 177610
    },
    {
      "epoch": 1.1391995074286556,
      "grad_norm": 0.03266414627432823,
      "learning_rate": 4.658176125129876e-06,
      "loss": 0.0018,
      "step": 177620
    },
    {
      "epoch": 1.1392636443224418,
      "grad_norm": 0.013438074849545956,
      "learning_rate": 4.657617733384628e-06,
      "loss": 0.0015,
      "step": 177630
    },
    {
      "epoch": 1.139327781216228,
      "grad_norm": 0.06101763993501663,
      "learning_rate": 4.657059345929677e-06,
      "loss": 0.0018,
      "step": 177640
    },
    {
      "epoch": 1.139391918110014,
      "grad_norm": 0.06609926372766495,
      "learning_rate": 4.656500962772016e-06,
      "loss": 0.0012,
      "step": 177650
    },
    {
      "epoch": 1.1394560550038002,
      "grad_norm": 0.22577722370624542,
      "learning_rate": 4.655942583918644e-06,
      "loss": 0.0015,
      "step": 177660
    },
    {
      "epoch": 1.1395201918975861,
      "grad_norm": 0.2436590939760208,
      "learning_rate": 4.655384209376557e-06,
      "loss": 0.0024,
      "step": 177670
    },
    {
      "epoch": 1.1395843287913723,
      "grad_norm": 0.04356739670038223,
      "learning_rate": 4.654825839152753e-06,
      "loss": 0.0017,
      "step": 177680
    },
    {
      "epoch": 1.1396484656851584,
      "grad_norm": 0.07654337584972382,
      "learning_rate": 4.6542674732542265e-06,
      "loss": 0.0023,
      "step": 177690
    },
    {
      "epoch": 1.1397126025789446,
      "grad_norm": 0.12162031978368759,
      "learning_rate": 4.653709111687978e-06,
      "loss": 0.002,
      "step": 177700
    },
    {
      "epoch": 1.1397767394727305,
      "grad_norm": 0.027826106175780296,
      "learning_rate": 4.653150754460999e-06,
      "loss": 0.0017,
      "step": 177710
    },
    {
      "epoch": 1.1398408763665167,
      "grad_norm": 0.03190990164875984,
      "learning_rate": 4.652592401580288e-06,
      "loss": 0.0014,
      "step": 177720
    },
    {
      "epoch": 1.1399050132603028,
      "grad_norm": 0.20037496089935303,
      "learning_rate": 4.652034053052846e-06,
      "loss": 0.0019,
      "step": 177730
    },
    {
      "epoch": 1.139969150154089,
      "grad_norm": 0.18473303318023682,
      "learning_rate": 4.651475708885663e-06,
      "loss": 0.0014,
      "step": 177740
    },
    {
      "epoch": 1.140033287047875,
      "grad_norm": 0.23001250624656677,
      "learning_rate": 4.6509173690857405e-06,
      "loss": 0.0012,
      "step": 177750
    },
    {
      "epoch": 1.140097423941661,
      "grad_norm": 0.09116110950708389,
      "learning_rate": 4.650359033660073e-06,
      "loss": 0.0012,
      "step": 177760
    },
    {
      "epoch": 1.1401615608354472,
      "grad_norm": 0.11727047711610794,
      "learning_rate": 4.649800702615656e-06,
      "loss": 0.0016,
      "step": 177770
    },
    {
      "epoch": 1.1402256977292333,
      "grad_norm": 0.17801989614963531,
      "learning_rate": 4.649242375959486e-06,
      "loss": 0.0016,
      "step": 177780
    },
    {
      "epoch": 1.1402898346230192,
      "grad_norm": 0.06194036081433296,
      "learning_rate": 4.64868405369856e-06,
      "loss": 0.0012,
      "step": 177790
    },
    {
      "epoch": 1.1403539715168054,
      "grad_norm": 0.10502773523330688,
      "learning_rate": 4.648125735839875e-06,
      "loss": 0.0015,
      "step": 177800
    },
    {
      "epoch": 1.1404181084105915,
      "grad_norm": 0.06816016882658005,
      "learning_rate": 4.6475674223904246e-06,
      "loss": 0.0032,
      "step": 177810
    },
    {
      "epoch": 1.1404822453043777,
      "grad_norm": 0.05017755180597305,
      "learning_rate": 4.647009113357207e-06,
      "loss": 0.0012,
      "step": 177820
    },
    {
      "epoch": 1.1405463821981638,
      "grad_norm": 0.11262434720993042,
      "learning_rate": 4.6464508087472175e-06,
      "loss": 0.0012,
      "step": 177830
    },
    {
      "epoch": 1.1406105190919498,
      "grad_norm": 0.035311099141836166,
      "learning_rate": 4.645892508567452e-06,
      "loss": 0.0015,
      "step": 177840
    },
    {
      "epoch": 1.140674655985736,
      "grad_norm": 0.04332641884684563,
      "learning_rate": 4.645334212824906e-06,
      "loss": 0.0044,
      "step": 177850
    },
    {
      "epoch": 1.140738792879522,
      "grad_norm": 0.0032705175690352917,
      "learning_rate": 4.644775921526577e-06,
      "loss": 0.0015,
      "step": 177860
    },
    {
      "epoch": 1.1408029297733082,
      "grad_norm": 0.11926588416099548,
      "learning_rate": 4.644217634679458e-06,
      "loss": 0.0012,
      "step": 177870
    },
    {
      "epoch": 1.1408670666670941,
      "grad_norm": 0.19587744772434235,
      "learning_rate": 4.6436593522905484e-06,
      "loss": 0.0038,
      "step": 177880
    },
    {
      "epoch": 1.1409312035608803,
      "grad_norm": 0.08402374386787415,
      "learning_rate": 4.643101074366839e-06,
      "loss": 0.0018,
      "step": 177890
    },
    {
      "epoch": 1.1409953404546664,
      "grad_norm": 0.09392978250980377,
      "learning_rate": 4.6425428009153295e-06,
      "loss": 0.0011,
      "step": 177900
    },
    {
      "epoch": 1.1410594773484526,
      "grad_norm": 0.1906268298625946,
      "learning_rate": 4.641984531943015e-06,
      "loss": 0.0021,
      "step": 177910
    },
    {
      "epoch": 1.1411236142422387,
      "grad_norm": 0.020670609548687935,
      "learning_rate": 4.6414262674568885e-06,
      "loss": 0.0012,
      "step": 177920
    },
    {
      "epoch": 1.1411877511360247,
      "grad_norm": 0.1239483579993248,
      "learning_rate": 4.640868007463949e-06,
      "loss": 0.0013,
      "step": 177930
    },
    {
      "epoch": 1.1412518880298108,
      "grad_norm": 0.230666384100914,
      "learning_rate": 4.640309751971189e-06,
      "loss": 0.0024,
      "step": 177940
    },
    {
      "epoch": 1.141316024923597,
      "grad_norm": 0.07052595913410187,
      "learning_rate": 4.639751500985606e-06,
      "loss": 0.0014,
      "step": 177950
    },
    {
      "epoch": 1.141380161817383,
      "grad_norm": 0.061315130442380905,
      "learning_rate": 4.639193254514192e-06,
      "loss": 0.0016,
      "step": 177960
    },
    {
      "epoch": 1.141444298711169,
      "grad_norm": 0.08418230712413788,
      "learning_rate": 4.638635012563946e-06,
      "loss": 0.0011,
      "step": 177970
    },
    {
      "epoch": 1.1415084356049552,
      "grad_norm": 0.05300601199269295,
      "learning_rate": 4.638076775141861e-06,
      "loss": 0.0024,
      "step": 177980
    },
    {
      "epoch": 1.1415725724987413,
      "grad_norm": 0.02021416835486889,
      "learning_rate": 4.637518542254934e-06,
      "loss": 0.0018,
      "step": 177990
    },
    {
      "epoch": 1.1416367093925275,
      "grad_norm": 0.006741389166563749,
      "learning_rate": 4.6369603139101566e-06,
      "loss": 0.001,
      "step": 178000
    },
    {
      "epoch": 1.1417008462863136,
      "grad_norm": 0.040095072239637375,
      "learning_rate": 4.636402090114527e-06,
      "loss": 0.0037,
      "step": 178010
    },
    {
      "epoch": 1.1417649831800996,
      "grad_norm": 0.1890638768672943,
      "learning_rate": 4.635843870875038e-06,
      "loss": 0.0025,
      "step": 178020
    },
    {
      "epoch": 1.1418291200738857,
      "grad_norm": 0.007839915342628956,
      "learning_rate": 4.6352856561986885e-06,
      "loss": 0.0016,
      "step": 178030
    },
    {
      "epoch": 1.1418932569676719,
      "grad_norm": 0.054487310349941254,
      "learning_rate": 4.634727446092468e-06,
      "loss": 0.0024,
      "step": 178040
    },
    {
      "epoch": 1.1419573938614578,
      "grad_norm": 0.1597786843776703,
      "learning_rate": 4.6341692405633725e-06,
      "loss": 0.002,
      "step": 178050
    },
    {
      "epoch": 1.142021530755244,
      "grad_norm": 0.012354972772300243,
      "learning_rate": 4.6336110396184e-06,
      "loss": 0.003,
      "step": 178060
    },
    {
      "epoch": 1.14208566764903,
      "grad_norm": 0.14595657587051392,
      "learning_rate": 4.633052843264542e-06,
      "loss": 0.001,
      "step": 178070
    },
    {
      "epoch": 1.1421498045428162,
      "grad_norm": 0.08125365525484085,
      "learning_rate": 4.632494651508795e-06,
      "loss": 0.0017,
      "step": 178080
    },
    {
      "epoch": 1.1422139414366024,
      "grad_norm": 0.05589194595813751,
      "learning_rate": 4.631936464358152e-06,
      "loss": 0.0014,
      "step": 178090
    },
    {
      "epoch": 1.1422780783303883,
      "grad_norm": 0.013811212033033371,
      "learning_rate": 4.631378281819609e-06,
      "loss": 0.0017,
      "step": 178100
    },
    {
      "epoch": 1.1423422152241745,
      "grad_norm": 0.09815085679292679,
      "learning_rate": 4.630820103900158e-06,
      "loss": 0.0019,
      "step": 178110
    },
    {
      "epoch": 1.1424063521179606,
      "grad_norm": 0.05617450922727585,
      "learning_rate": 4.630261930606797e-06,
      "loss": 0.0015,
      "step": 178120
    },
    {
      "epoch": 1.1424704890117467,
      "grad_norm": 0.04724203050136566,
      "learning_rate": 4.6297037619465164e-06,
      "loss": 0.0012,
      "step": 178130
    },
    {
      "epoch": 1.1425346259055327,
      "grad_norm": 0.06500443071126938,
      "learning_rate": 4.629145597926314e-06,
      "loss": 0.0013,
      "step": 178140
    },
    {
      "epoch": 1.1425987627993188,
      "grad_norm": 0.0620625838637352,
      "learning_rate": 4.628587438553181e-06,
      "loss": 0.0019,
      "step": 178150
    },
    {
      "epoch": 1.142662899693105,
      "grad_norm": 0.14742448925971985,
      "learning_rate": 4.628029283834113e-06,
      "loss": 0.0023,
      "step": 178160
    },
    {
      "epoch": 1.1427270365868911,
      "grad_norm": 0.07294030487537384,
      "learning_rate": 4.6274711337761045e-06,
      "loss": 0.0011,
      "step": 178170
    },
    {
      "epoch": 1.1427911734806773,
      "grad_norm": 0.036900971084833145,
      "learning_rate": 4.6269129883861486e-06,
      "loss": 0.0017,
      "step": 178180
    },
    {
      "epoch": 1.1428553103744632,
      "grad_norm": 0.0273374542593956,
      "learning_rate": 4.626354847671239e-06,
      "loss": 0.0006,
      "step": 178190
    },
    {
      "epoch": 1.1429194472682493,
      "grad_norm": 0.11742100119590759,
      "learning_rate": 4.625796711638371e-06,
      "loss": 0.0011,
      "step": 178200
    },
    {
      "epoch": 1.1429835841620355,
      "grad_norm": 0.004678953904658556,
      "learning_rate": 4.625238580294538e-06,
      "loss": 0.0014,
      "step": 178210
    },
    {
      "epoch": 1.1430477210558216,
      "grad_norm": 0.34453120827674866,
      "learning_rate": 4.624680453646732e-06,
      "loss": 0.0012,
      "step": 178220
    },
    {
      "epoch": 1.1431118579496076,
      "grad_norm": 0.06346111744642258,
      "learning_rate": 4.62412233170195e-06,
      "loss": 0.0015,
      "step": 178230
    },
    {
      "epoch": 1.1431759948433937,
      "grad_norm": 0.06714435666799545,
      "learning_rate": 4.623564214467182e-06,
      "loss": 0.0019,
      "step": 178240
    },
    {
      "epoch": 1.1432401317371799,
      "grad_norm": 0.1403575986623764,
      "learning_rate": 4.6230061019494244e-06,
      "loss": 0.002,
      "step": 178250
    },
    {
      "epoch": 1.143304268630966,
      "grad_norm": 0.2506372034549713,
      "learning_rate": 4.62244799415567e-06,
      "loss": 0.0035,
      "step": 178260
    },
    {
      "epoch": 1.143368405524752,
      "grad_norm": 0.10004667192697525,
      "learning_rate": 4.6218898910929115e-06,
      "loss": 0.0017,
      "step": 178270
    },
    {
      "epoch": 1.143432542418538,
      "grad_norm": 0.040075477212667465,
      "learning_rate": 4.621331792768143e-06,
      "loss": 0.0011,
      "step": 178280
    },
    {
      "epoch": 1.1434966793123242,
      "grad_norm": 0.11705781519412994,
      "learning_rate": 4.6207736991883575e-06,
      "loss": 0.0022,
      "step": 178290
    },
    {
      "epoch": 1.1435608162061104,
      "grad_norm": 0.1692320704460144,
      "learning_rate": 4.620215610360549e-06,
      "loss": 0.0009,
      "step": 178300
    },
    {
      "epoch": 1.1436249530998963,
      "grad_norm": 0.09076292812824249,
      "learning_rate": 4.619657526291711e-06,
      "loss": 0.0019,
      "step": 178310
    },
    {
      "epoch": 1.1436890899936825,
      "grad_norm": 0.18240463733673096,
      "learning_rate": 4.619099446988834e-06,
      "loss": 0.0011,
      "step": 178320
    },
    {
      "epoch": 1.1437532268874686,
      "grad_norm": 0.05098031088709831,
      "learning_rate": 4.6185413724589145e-06,
      "loss": 0.0051,
      "step": 178330
    },
    {
      "epoch": 1.1438173637812548,
      "grad_norm": 0.08685196936130524,
      "learning_rate": 4.617983302708945e-06,
      "loss": 0.0014,
      "step": 178340
    },
    {
      "epoch": 1.143881500675041,
      "grad_norm": 0.048797864466905594,
      "learning_rate": 4.617425237745916e-06,
      "loss": 0.0017,
      "step": 178350
    },
    {
      "epoch": 1.1439456375688268,
      "grad_norm": 0.049440789967775345,
      "learning_rate": 4.616867177576824e-06,
      "loss": 0.0016,
      "step": 178360
    },
    {
      "epoch": 1.144009774462613,
      "grad_norm": 0.07405377179384232,
      "learning_rate": 4.616309122208659e-06,
      "loss": 0.0022,
      "step": 178370
    },
    {
      "epoch": 1.1440739113563991,
      "grad_norm": 0.08482597768306732,
      "learning_rate": 4.615751071648415e-06,
      "loss": 0.0013,
      "step": 178380
    },
    {
      "epoch": 1.1441380482501853,
      "grad_norm": 0.20172375440597534,
      "learning_rate": 4.615193025903084e-06,
      "loss": 0.0016,
      "step": 178390
    },
    {
      "epoch": 1.1442021851439712,
      "grad_norm": 0.1691344678401947,
      "learning_rate": 4.614634984979661e-06,
      "loss": 0.002,
      "step": 178400
    },
    {
      "epoch": 1.1442663220377574,
      "grad_norm": 0.010287750512361526,
      "learning_rate": 4.614076948885136e-06,
      "loss": 0.0011,
      "step": 178410
    },
    {
      "epoch": 1.1443304589315435,
      "grad_norm": 0.09773595631122589,
      "learning_rate": 4.6135189176265035e-06,
      "loss": 0.0015,
      "step": 178420
    },
    {
      "epoch": 1.1443945958253297,
      "grad_norm": 0.23118939995765686,
      "learning_rate": 4.612960891210754e-06,
      "loss": 0.0012,
      "step": 178430
    },
    {
      "epoch": 1.1444587327191158,
      "grad_norm": 0.023415856063365936,
      "learning_rate": 4.612402869644882e-06,
      "loss": 0.0012,
      "step": 178440
    },
    {
      "epoch": 1.1445228696129017,
      "grad_norm": 0.06214074790477753,
      "learning_rate": 4.611844852935878e-06,
      "loss": 0.003,
      "step": 178450
    },
    {
      "epoch": 1.1445870065066879,
      "grad_norm": 0.08905795216560364,
      "learning_rate": 4.611286841090738e-06,
      "loss": 0.0015,
      "step": 178460
    },
    {
      "epoch": 1.144651143400474,
      "grad_norm": 0.07331910729408264,
      "learning_rate": 4.610728834116448e-06,
      "loss": 0.0017,
      "step": 178470
    },
    {
      "epoch": 1.14471528029426,
      "grad_norm": 0.10792273283004761,
      "learning_rate": 4.610170832020005e-06,
      "loss": 0.0016,
      "step": 178480
    },
    {
      "epoch": 1.144779417188046,
      "grad_norm": 0.06411322206258774,
      "learning_rate": 4.6096128348084e-06,
      "loss": 0.0009,
      "step": 178490
    },
    {
      "epoch": 1.1448435540818322,
      "grad_norm": 0.004302809480577707,
      "learning_rate": 4.609054842488627e-06,
      "loss": 0.0027,
      "step": 178500
    },
    {
      "epoch": 1.1449076909756184,
      "grad_norm": 0.05978574976325035,
      "learning_rate": 4.608496855067675e-06,
      "loss": 0.0022,
      "step": 178510
    },
    {
      "epoch": 1.1449718278694045,
      "grad_norm": 0.11780686676502228,
      "learning_rate": 4.607938872552537e-06,
      "loss": 0.0015,
      "step": 178520
    },
    {
      "epoch": 1.1450359647631905,
      "grad_norm": 0.1133837178349495,
      "learning_rate": 4.6073808949502054e-06,
      "loss": 0.0012,
      "step": 178530
    },
    {
      "epoch": 1.1451001016569766,
      "grad_norm": 0.12923049926757812,
      "learning_rate": 4.606822922267673e-06,
      "loss": 0.002,
      "step": 178540
    },
    {
      "epoch": 1.1451642385507628,
      "grad_norm": 0.045091308653354645,
      "learning_rate": 4.606264954511929e-06,
      "loss": 0.0025,
      "step": 178550
    },
    {
      "epoch": 1.145228375444549,
      "grad_norm": 0.10687202215194702,
      "learning_rate": 4.605706991689967e-06,
      "loss": 0.0013,
      "step": 178560
    },
    {
      "epoch": 1.1452925123383348,
      "grad_norm": 0.11072013527154922,
      "learning_rate": 4.605149033808778e-06,
      "loss": 0.0013,
      "step": 178570
    },
    {
      "epoch": 1.145356649232121,
      "grad_norm": 0.23663799464702606,
      "learning_rate": 4.604591080875354e-06,
      "loss": 0.0033,
      "step": 178580
    },
    {
      "epoch": 1.1454207861259071,
      "grad_norm": 0.0915546789765358,
      "learning_rate": 4.604033132896686e-06,
      "loss": 0.0008,
      "step": 178590
    },
    {
      "epoch": 1.1454849230196933,
      "grad_norm": 0.008109216578304768,
      "learning_rate": 4.603475189879765e-06,
      "loss": 0.0015,
      "step": 178600
    },
    {
      "epoch": 1.1455490599134794,
      "grad_norm": 0.26255640387535095,
      "learning_rate": 4.602917251831586e-06,
      "loss": 0.0017,
      "step": 178610
    },
    {
      "epoch": 1.1456131968072654,
      "grad_norm": 0.034671954810619354,
      "learning_rate": 4.602359318759135e-06,
      "loss": 0.001,
      "step": 178620
    },
    {
      "epoch": 1.1456773337010515,
      "grad_norm": 0.09453405439853668,
      "learning_rate": 4.6018013906694075e-06,
      "loss": 0.0025,
      "step": 178630
    },
    {
      "epoch": 1.1457414705948377,
      "grad_norm": 0.11215243488550186,
      "learning_rate": 4.601243467569392e-06,
      "loss": 0.0007,
      "step": 178640
    },
    {
      "epoch": 1.1458056074886238,
      "grad_norm": 0.06050289794802666,
      "learning_rate": 4.600685549466081e-06,
      "loss": 0.0022,
      "step": 178650
    },
    {
      "epoch": 1.1458697443824097,
      "grad_norm": 0.11573351174592972,
      "learning_rate": 4.600127636366466e-06,
      "loss": 0.0018,
      "step": 178660
    },
    {
      "epoch": 1.1459338812761959,
      "grad_norm": 0.08109661936759949,
      "learning_rate": 4.599569728277538e-06,
      "loss": 0.0023,
      "step": 178670
    },
    {
      "epoch": 1.145998018169982,
      "grad_norm": 0.11803097277879715,
      "learning_rate": 4.599011825206287e-06,
      "loss": 0.0016,
      "step": 178680
    },
    {
      "epoch": 1.1460621550637682,
      "grad_norm": 0.15141747891902924,
      "learning_rate": 4.598453927159704e-06,
      "loss": 0.0015,
      "step": 178690
    },
    {
      "epoch": 1.146126291957554,
      "grad_norm": 0.05062533915042877,
      "learning_rate": 4.59789603414478e-06,
      "loss": 0.0007,
      "step": 178700
    },
    {
      "epoch": 1.1461904288513403,
      "grad_norm": 0.2250877469778061,
      "learning_rate": 4.597338146168507e-06,
      "loss": 0.0015,
      "step": 178710
    },
    {
      "epoch": 1.1462545657451264,
      "grad_norm": 0.1328151375055313,
      "learning_rate": 4.596780263237875e-06,
      "loss": 0.0012,
      "step": 178720
    },
    {
      "epoch": 1.1463187026389126,
      "grad_norm": 0.04682164266705513,
      "learning_rate": 4.596222385359873e-06,
      "loss": 0.0015,
      "step": 178730
    },
    {
      "epoch": 1.1463828395326985,
      "grad_norm": 0.15815287828445435,
      "learning_rate": 4.5956645125414945e-06,
      "loss": 0.0014,
      "step": 178740
    },
    {
      "epoch": 1.1464469764264846,
      "grad_norm": 0.0862506851553917,
      "learning_rate": 4.595106644789727e-06,
      "loss": 0.0012,
      "step": 178750
    },
    {
      "epoch": 1.1465111133202708,
      "grad_norm": 0.10812395811080933,
      "learning_rate": 4.594548782111564e-06,
      "loss": 0.0011,
      "step": 178760
    },
    {
      "epoch": 1.146575250214057,
      "grad_norm": 0.015825750306248665,
      "learning_rate": 4.593990924513995e-06,
      "loss": 0.0006,
      "step": 178770
    },
    {
      "epoch": 1.146639387107843,
      "grad_norm": 0.09403803199529648,
      "learning_rate": 4.593433072004009e-06,
      "loss": 0.0019,
      "step": 178780
    },
    {
      "epoch": 1.146703524001629,
      "grad_norm": 0.08201845735311508,
      "learning_rate": 4.592875224588597e-06,
      "loss": 0.0015,
      "step": 178790
    },
    {
      "epoch": 1.1467676608954152,
      "grad_norm": 0.06644584983587265,
      "learning_rate": 4.59231738227475e-06,
      "loss": 0.0012,
      "step": 178800
    },
    {
      "epoch": 1.1468317977892013,
      "grad_norm": 0.1825915426015854,
      "learning_rate": 4.591759545069457e-06,
      "loss": 0.0019,
      "step": 178810
    },
    {
      "epoch": 1.1468959346829875,
      "grad_norm": 0.05877283215522766,
      "learning_rate": 4.591201712979709e-06,
      "loss": 0.0009,
      "step": 178820
    },
    {
      "epoch": 1.1469600715767734,
      "grad_norm": 0.28253012895584106,
      "learning_rate": 4.590643886012496e-06,
      "loss": 0.0028,
      "step": 178830
    },
    {
      "epoch": 1.1470242084705595,
      "grad_norm": 0.00932693388313055,
      "learning_rate": 4.590086064174807e-06,
      "loss": 0.0016,
      "step": 178840
    },
    {
      "epoch": 1.1470883453643457,
      "grad_norm": 0.051429156213998795,
      "learning_rate": 4.589528247473633e-06,
      "loss": 0.0007,
      "step": 178850
    },
    {
      "epoch": 1.1471524822581318,
      "grad_norm": 0.027532080188393593,
      "learning_rate": 4.588970435915964e-06,
      "loss": 0.0009,
      "step": 178860
    },
    {
      "epoch": 1.147216619151918,
      "grad_norm": 0.25374674797058105,
      "learning_rate": 4.588412629508788e-06,
      "loss": 0.0016,
      "step": 178870
    },
    {
      "epoch": 1.147280756045704,
      "grad_norm": 0.12644684314727783,
      "learning_rate": 4.587854828259097e-06,
      "loss": 0.0018,
      "step": 178880
    },
    {
      "epoch": 1.14734489293949,
      "grad_norm": 0.12913778424263,
      "learning_rate": 4.5872970321738784e-06,
      "loss": 0.0021,
      "step": 178890
    },
    {
      "epoch": 1.1474090298332762,
      "grad_norm": 0.14622823894023895,
      "learning_rate": 4.586739241260123e-06,
      "loss": 0.0014,
      "step": 178900
    },
    {
      "epoch": 1.1474731667270621,
      "grad_norm": 0.18271958827972412,
      "learning_rate": 4.586181455524821e-06,
      "loss": 0.0014,
      "step": 178910
    },
    {
      "epoch": 1.1475373036208483,
      "grad_norm": 0.09339526295661926,
      "learning_rate": 4.58562367497496e-06,
      "loss": 0.0022,
      "step": 178920
    },
    {
      "epoch": 1.1476014405146344,
      "grad_norm": 0.009551736526191235,
      "learning_rate": 4.585065899617532e-06,
      "loss": 0.0006,
      "step": 178930
    },
    {
      "epoch": 1.1476655774084206,
      "grad_norm": 0.05664249137043953,
      "learning_rate": 4.584508129459524e-06,
      "loss": 0.0009,
      "step": 178940
    },
    {
      "epoch": 1.1477297143022067,
      "grad_norm": 0.13133834302425385,
      "learning_rate": 4.583950364507927e-06,
      "loss": 0.0016,
      "step": 178950
    },
    {
      "epoch": 1.1477938511959926,
      "grad_norm": 0.05462121590971947,
      "learning_rate": 4.583392604769728e-06,
      "loss": 0.001,
      "step": 178960
    },
    {
      "epoch": 1.1478579880897788,
      "grad_norm": 0.11168855428695679,
      "learning_rate": 4.582834850251919e-06,
      "loss": 0.0011,
      "step": 178970
    },
    {
      "epoch": 1.147922124983565,
      "grad_norm": 0.10457391291856766,
      "learning_rate": 4.582277100961485e-06,
      "loss": 0.0019,
      "step": 178980
    },
    {
      "epoch": 1.147986261877351,
      "grad_norm": 0.10076499730348587,
      "learning_rate": 4.581719356905421e-06,
      "loss": 0.0014,
      "step": 178990
    },
    {
      "epoch": 1.148050398771137,
      "grad_norm": 0.04545801877975464,
      "learning_rate": 4.581161618090709e-06,
      "loss": 0.0013,
      "step": 179000
    },
    {
      "epoch": 1.1481145356649232,
      "grad_norm": 0.07611383497714996,
      "learning_rate": 4.580603884524343e-06,
      "loss": 0.0009,
      "step": 179010
    },
    {
      "epoch": 1.1481786725587093,
      "grad_norm": 0.16176478564739227,
      "learning_rate": 4.5800461562133095e-06,
      "loss": 0.002,
      "step": 179020
    },
    {
      "epoch": 1.1482428094524955,
      "grad_norm": 0.10310656577348709,
      "learning_rate": 4.579488433164599e-06,
      "loss": 0.0014,
      "step": 179030
    },
    {
      "epoch": 1.1483069463462816,
      "grad_norm": 0.1659184992313385,
      "learning_rate": 4.578930715385197e-06,
      "loss": 0.0019,
      "step": 179040
    },
    {
      "epoch": 1.1483710832400675,
      "grad_norm": 0.15081000328063965,
      "learning_rate": 4.5783730028820935e-06,
      "loss": 0.0011,
      "step": 179050
    },
    {
      "epoch": 1.1484352201338537,
      "grad_norm": 0.0816895142197609,
      "learning_rate": 4.57781529566228e-06,
      "loss": 0.0018,
      "step": 179060
    },
    {
      "epoch": 1.1484993570276398,
      "grad_norm": 0.10530539602041245,
      "learning_rate": 4.577257593732741e-06,
      "loss": 0.0013,
      "step": 179070
    },
    {
      "epoch": 1.148563493921426,
      "grad_norm": 0.28155598044395447,
      "learning_rate": 4.576699897100468e-06,
      "loss": 0.0045,
      "step": 179080
    },
    {
      "epoch": 1.148627630815212,
      "grad_norm": 0.07171224057674408,
      "learning_rate": 4.576142205772447e-06,
      "loss": 0.0008,
      "step": 179090
    },
    {
      "epoch": 1.148691767708998,
      "grad_norm": 0.028919516131281853,
      "learning_rate": 4.575584519755668e-06,
      "loss": 0.0023,
      "step": 179100
    },
    {
      "epoch": 1.1487559046027842,
      "grad_norm": 0.11434286832809448,
      "learning_rate": 4.575026839057116e-06,
      "loss": 0.0021,
      "step": 179110
    },
    {
      "epoch": 1.1488200414965704,
      "grad_norm": 0.022055277600884438,
      "learning_rate": 4.5744691636837835e-06,
      "loss": 0.0009,
      "step": 179120
    },
    {
      "epoch": 1.1488841783903563,
      "grad_norm": 0.03330404683947563,
      "learning_rate": 4.573911493642655e-06,
      "loss": 0.0019,
      "step": 179130
    },
    {
      "epoch": 1.1489483152841424,
      "grad_norm": 0.030812516808509827,
      "learning_rate": 4.5733538289407215e-06,
      "loss": 0.0009,
      "step": 179140
    },
    {
      "epoch": 1.1490124521779286,
      "grad_norm": 0.0154456477612257,
      "learning_rate": 4.572796169584968e-06,
      "loss": 0.0009,
      "step": 179150
    },
    {
      "epoch": 1.1490765890717147,
      "grad_norm": 0.18655256927013397,
      "learning_rate": 4.572238515582385e-06,
      "loss": 0.0007,
      "step": 179160
    },
    {
      "epoch": 1.1491407259655007,
      "grad_norm": 0.056279927492141724,
      "learning_rate": 4.571680866939959e-06,
      "loss": 0.0028,
      "step": 179170
    },
    {
      "epoch": 1.1492048628592868,
      "grad_norm": 0.1965060830116272,
      "learning_rate": 4.571123223664677e-06,
      "loss": 0.0012,
      "step": 179180
    },
    {
      "epoch": 1.149268999753073,
      "grad_norm": 0.05386898294091225,
      "learning_rate": 4.570565585763528e-06,
      "loss": 0.0013,
      "step": 179190
    },
    {
      "epoch": 1.149333136646859,
      "grad_norm": 0.23607292771339417,
      "learning_rate": 4.570007953243499e-06,
      "loss": 0.0022,
      "step": 179200
    },
    {
      "epoch": 1.1493972735406452,
      "grad_norm": 0.030810948461294174,
      "learning_rate": 4.5694503261115784e-06,
      "loss": 0.0025,
      "step": 179210
    },
    {
      "epoch": 1.1494614104344312,
      "grad_norm": 0.05335972458124161,
      "learning_rate": 4.568892704374751e-06,
      "loss": 0.0014,
      "step": 179220
    },
    {
      "epoch": 1.1495255473282173,
      "grad_norm": 0.009272546507418156,
      "learning_rate": 4.568335088040009e-06,
      "loss": 0.0018,
      "step": 179230
    },
    {
      "epoch": 1.1495896842220035,
      "grad_norm": 0.022868212312459946,
      "learning_rate": 4.567777477114334e-06,
      "loss": 0.0008,
      "step": 179240
    },
    {
      "epoch": 1.1496538211157896,
      "grad_norm": 0.0760836973786354,
      "learning_rate": 4.567219871604716e-06,
      "loss": 0.0016,
      "step": 179250
    },
    {
      "epoch": 1.1497179580095755,
      "grad_norm": 0.35971537232398987,
      "learning_rate": 4.566662271518144e-06,
      "loss": 0.0008,
      "step": 179260
    },
    {
      "epoch": 1.1497820949033617,
      "grad_norm": 0.07496869564056396,
      "learning_rate": 4.566104676861603e-06,
      "loss": 0.0006,
      "step": 179270
    },
    {
      "epoch": 1.1498462317971478,
      "grad_norm": 0.12314856052398682,
      "learning_rate": 4.5655470876420814e-06,
      "loss": 0.0046,
      "step": 179280
    },
    {
      "epoch": 1.149910368690934,
      "grad_norm": 0.08336050063371658,
      "learning_rate": 4.564989503866564e-06,
      "loss": 0.0017,
      "step": 179290
    },
    {
      "epoch": 1.1499745055847201,
      "grad_norm": 0.2124306708574295,
      "learning_rate": 4.5644319255420405e-06,
      "loss": 0.0024,
      "step": 179300
    },
    {
      "epoch": 1.150038642478506,
      "grad_norm": 0.018601980060338974,
      "learning_rate": 4.563874352675495e-06,
      "loss": 0.0016,
      "step": 179310
    },
    {
      "epoch": 1.1501027793722922,
      "grad_norm": 0.061966847628355026,
      "learning_rate": 4.5633167852739154e-06,
      "loss": 0.001,
      "step": 179320
    },
    {
      "epoch": 1.1501669162660784,
      "grad_norm": 0.11580870300531387,
      "learning_rate": 4.56275922334429e-06,
      "loss": 0.0015,
      "step": 179330
    },
    {
      "epoch": 1.1502310531598643,
      "grad_norm": 0.0410192608833313,
      "learning_rate": 4.562201666893602e-06,
      "loss": 0.0012,
      "step": 179340
    },
    {
      "epoch": 1.1502951900536504,
      "grad_norm": 0.08091183006763458,
      "learning_rate": 4.561644115928842e-06,
      "loss": 0.0013,
      "step": 179350
    },
    {
      "epoch": 1.1503593269474366,
      "grad_norm": 0.17633156478405,
      "learning_rate": 4.5610865704569925e-06,
      "loss": 0.0016,
      "step": 179360
    },
    {
      "epoch": 1.1504234638412227,
      "grad_norm": 0.037320394068956375,
      "learning_rate": 4.560529030485044e-06,
      "loss": 0.0026,
      "step": 179370
    },
    {
      "epoch": 1.1504876007350089,
      "grad_norm": 0.15806110203266144,
      "learning_rate": 4.559971496019979e-06,
      "loss": 0.0027,
      "step": 179380
    },
    {
      "epoch": 1.1505517376287948,
      "grad_norm": 0.033490147441625595,
      "learning_rate": 4.559413967068788e-06,
      "loss": 0.0015,
      "step": 179390
    },
    {
      "epoch": 1.150615874522581,
      "grad_norm": 0.07470483332872391,
      "learning_rate": 4.558856443638452e-06,
      "loss": 0.0011,
      "step": 179400
    },
    {
      "epoch": 1.150680011416367,
      "grad_norm": 0.03577504679560661,
      "learning_rate": 4.558298925735963e-06,
      "loss": 0.0011,
      "step": 179410
    },
    {
      "epoch": 1.1507441483101533,
      "grad_norm": 0.06793781369924545,
      "learning_rate": 4.5577414133683016e-06,
      "loss": 0.0021,
      "step": 179420
    },
    {
      "epoch": 1.1508082852039392,
      "grad_norm": 0.04992002621293068,
      "learning_rate": 4.557183906542458e-06,
      "loss": 0.0023,
      "step": 179430
    },
    {
      "epoch": 1.1508724220977253,
      "grad_norm": 0.24769335985183716,
      "learning_rate": 4.556626405265415e-06,
      "loss": 0.0013,
      "step": 179440
    },
    {
      "epoch": 1.1509365589915115,
      "grad_norm": 0.06733288615942001,
      "learning_rate": 4.55606890954416e-06,
      "loss": 0.0014,
      "step": 179450
    },
    {
      "epoch": 1.1510006958852976,
      "grad_norm": 0.025171691551804543,
      "learning_rate": 4.555511419385681e-06,
      "loss": 0.0025,
      "step": 179460
    },
    {
      "epoch": 1.1510648327790838,
      "grad_norm": 0.13776138424873352,
      "learning_rate": 4.554953934796959e-06,
      "loss": 0.0017,
      "step": 179470
    },
    {
      "epoch": 1.1511289696728697,
      "grad_norm": 0.05418798699975014,
      "learning_rate": 4.554396455784985e-06,
      "loss": 0.0016,
      "step": 179480
    },
    {
      "epoch": 1.1511931065666559,
      "grad_norm": 0.05372137576341629,
      "learning_rate": 4.5538389823567394e-06,
      "loss": 0.0019,
      "step": 179490
    },
    {
      "epoch": 1.151257243460442,
      "grad_norm": 0.01216061506420374,
      "learning_rate": 4.553281514519212e-06,
      "loss": 0.0014,
      "step": 179500
    },
    {
      "epoch": 1.1513213803542282,
      "grad_norm": 0.05967012792825699,
      "learning_rate": 4.552724052279386e-06,
      "loss": 0.0021,
      "step": 179510
    },
    {
      "epoch": 1.151385517248014,
      "grad_norm": 0.06882903724908829,
      "learning_rate": 4.552166595644248e-06,
      "loss": 0.001,
      "step": 179520
    },
    {
      "epoch": 1.1514496541418002,
      "grad_norm": 0.07349782437086105,
      "learning_rate": 4.55160914462078e-06,
      "loss": 0.0022,
      "step": 179530
    },
    {
      "epoch": 1.1515137910355864,
      "grad_norm": 0.023654308170080185,
      "learning_rate": 4.551051699215973e-06,
      "loss": 0.0014,
      "step": 179540
    },
    {
      "epoch": 1.1515779279293725,
      "grad_norm": 0.199869766831398,
      "learning_rate": 4.550494259436807e-06,
      "loss": 0.0014,
      "step": 179550
    },
    {
      "epoch": 1.1516420648231587,
      "grad_norm": 0.0609150230884552,
      "learning_rate": 4.54993682529027e-06,
      "loss": 0.0018,
      "step": 179560
    },
    {
      "epoch": 1.1517062017169446,
      "grad_norm": 0.12952852249145508,
      "learning_rate": 4.549379396783346e-06,
      "loss": 0.001,
      "step": 179570
    },
    {
      "epoch": 1.1517703386107307,
      "grad_norm": 0.06888972222805023,
      "learning_rate": 4.548821973923021e-06,
      "loss": 0.0006,
      "step": 179580
    },
    {
      "epoch": 1.151834475504517,
      "grad_norm": 0.11483610421419144,
      "learning_rate": 4.548264556716277e-06,
      "loss": 0.0016,
      "step": 179590
    },
    {
      "epoch": 1.1518986123983028,
      "grad_norm": 0.09012481570243835,
      "learning_rate": 4.5477071451701025e-06,
      "loss": 0.0014,
      "step": 179600
    },
    {
      "epoch": 1.151962749292089,
      "grad_norm": 0.02963605895638466,
      "learning_rate": 4.547149739291481e-06,
      "loss": 0.0018,
      "step": 179610
    },
    {
      "epoch": 1.1520268861858751,
      "grad_norm": 0.06909776479005814,
      "learning_rate": 4.546592339087396e-06,
      "loss": 0.0013,
      "step": 179620
    },
    {
      "epoch": 1.1520910230796613,
      "grad_norm": 0.19308145344257355,
      "learning_rate": 4.546034944564834e-06,
      "loss": 0.0023,
      "step": 179630
    },
    {
      "epoch": 1.1521551599734474,
      "grad_norm": 0.09878735989332199,
      "learning_rate": 4.545477555730778e-06,
      "loss": 0.0012,
      "step": 179640
    },
    {
      "epoch": 1.1522192968672333,
      "grad_norm": 0.07624828815460205,
      "learning_rate": 4.5449201725922135e-06,
      "loss": 0.0023,
      "step": 179650
    },
    {
      "epoch": 1.1522834337610195,
      "grad_norm": 0.24515004456043243,
      "learning_rate": 4.544362795156123e-06,
      "loss": 0.0021,
      "step": 179660
    },
    {
      "epoch": 1.1523475706548056,
      "grad_norm": 0.13114790618419647,
      "learning_rate": 4.543805423429495e-06,
      "loss": 0.0019,
      "step": 179670
    },
    {
      "epoch": 1.1524117075485918,
      "grad_norm": 0.05208076909184456,
      "learning_rate": 4.5432480574193085e-06,
      "loss": 0.0007,
      "step": 179680
    },
    {
      "epoch": 1.1524758444423777,
      "grad_norm": 0.09520614892244339,
      "learning_rate": 4.5426906971325515e-06,
      "loss": 0.0015,
      "step": 179690
    },
    {
      "epoch": 1.1525399813361639,
      "grad_norm": 0.04604622349143028,
      "learning_rate": 4.542133342576207e-06,
      "loss": 0.0011,
      "step": 179700
    },
    {
      "epoch": 1.15260411822995,
      "grad_norm": 0.053334563970565796,
      "learning_rate": 4.541575993757259e-06,
      "loss": 0.0015,
      "step": 179710
    },
    {
      "epoch": 1.1526682551237362,
      "grad_norm": 0.08976370096206665,
      "learning_rate": 4.541018650682691e-06,
      "loss": 0.003,
      "step": 179720
    },
    {
      "epoch": 1.1527323920175223,
      "grad_norm": 0.15434318780899048,
      "learning_rate": 4.540461313359487e-06,
      "loss": 0.0023,
      "step": 179730
    },
    {
      "epoch": 1.1527965289113082,
      "grad_norm": 0.14458923041820526,
      "learning_rate": 4.5399039817946315e-06,
      "loss": 0.0014,
      "step": 179740
    },
    {
      "epoch": 1.1528606658050944,
      "grad_norm": 0.007658865302801132,
      "learning_rate": 4.539346655995108e-06,
      "loss": 0.0024,
      "step": 179750
    },
    {
      "epoch": 1.1529248026988805,
      "grad_norm": 0.047519613057374954,
      "learning_rate": 4.5387893359679e-06,
      "loss": 0.0012,
      "step": 179760
    },
    {
      "epoch": 1.1529889395926667,
      "grad_norm": 0.79911208152771,
      "learning_rate": 4.538232021719992e-06,
      "loss": 0.002,
      "step": 179770
    },
    {
      "epoch": 1.1530530764864526,
      "grad_norm": 0.17785044014453888,
      "learning_rate": 4.537674713258367e-06,
      "loss": 0.0017,
      "step": 179780
    },
    {
      "epoch": 1.1531172133802388,
      "grad_norm": 0.16367937624454498,
      "learning_rate": 4.537117410590006e-06,
      "loss": 0.0011,
      "step": 179790
    },
    {
      "epoch": 1.153181350274025,
      "grad_norm": 0.1835791915655136,
      "learning_rate": 4.536560113721897e-06,
      "loss": 0.0013,
      "step": 179800
    },
    {
      "epoch": 1.153245487167811,
      "grad_norm": 0.03765160217881203,
      "learning_rate": 4.53600282266102e-06,
      "loss": 0.0016,
      "step": 179810
    },
    {
      "epoch": 1.153309624061597,
      "grad_norm": 0.04303538054227829,
      "learning_rate": 4.535445537414359e-06,
      "loss": 0.0015,
      "step": 179820
    },
    {
      "epoch": 1.1533737609553831,
      "grad_norm": 0.004960588179528713,
      "learning_rate": 4.534888257988898e-06,
      "loss": 0.001,
      "step": 179830
    },
    {
      "epoch": 1.1534378978491693,
      "grad_norm": 0.07767927646636963,
      "learning_rate": 4.5343309843916194e-06,
      "loss": 0.0028,
      "step": 179840
    },
    {
      "epoch": 1.1535020347429554,
      "grad_norm": 0.04093633219599724,
      "learning_rate": 4.533773716629506e-06,
      "loss": 0.0016,
      "step": 179850
    },
    {
      "epoch": 1.1535661716367414,
      "grad_norm": 0.03396964073181152,
      "learning_rate": 4.533216454709542e-06,
      "loss": 0.0011,
      "step": 179860
    },
    {
      "epoch": 1.1536303085305275,
      "grad_norm": 0.043116357177495956,
      "learning_rate": 4.532659198638708e-06,
      "loss": 0.0018,
      "step": 179870
    },
    {
      "epoch": 1.1536944454243137,
      "grad_norm": 0.09401235729455948,
      "learning_rate": 4.53210194842399e-06,
      "loss": 0.0027,
      "step": 179880
    },
    {
      "epoch": 1.1537585823180998,
      "grad_norm": 0.17905250191688538,
      "learning_rate": 4.531544704072367e-06,
      "loss": 0.0008,
      "step": 179890
    },
    {
      "epoch": 1.153822719211886,
      "grad_norm": 0.07407578080892563,
      "learning_rate": 4.530987465590825e-06,
      "loss": 0.001,
      "step": 179900
    },
    {
      "epoch": 1.1538868561056719,
      "grad_norm": 0.040732480585575104,
      "learning_rate": 4.5304302329863456e-06,
      "loss": 0.0019,
      "step": 179910
    },
    {
      "epoch": 1.153950992999458,
      "grad_norm": 0.003030008403584361,
      "learning_rate": 4.52987300626591e-06,
      "loss": 0.0007,
      "step": 179920
    },
    {
      "epoch": 1.1540151298932442,
      "grad_norm": 0.1177324652671814,
      "learning_rate": 4.529315785436502e-06,
      "loss": 0.0018,
      "step": 179930
    },
    {
      "epoch": 1.1540792667870303,
      "grad_norm": 0.03247679024934769,
      "learning_rate": 4.5287585705051035e-06,
      "loss": 0.0009,
      "step": 179940
    },
    {
      "epoch": 1.1541434036808162,
      "grad_norm": 0.11346208304166794,
      "learning_rate": 4.5282013614786976e-06,
      "loss": 0.0016,
      "step": 179950
    },
    {
      "epoch": 1.1542075405746024,
      "grad_norm": 0.017800046131014824,
      "learning_rate": 4.527644158364265e-06,
      "loss": 0.0013,
      "step": 179960
    },
    {
      "epoch": 1.1542716774683885,
      "grad_norm": 0.11046946048736572,
      "learning_rate": 4.527086961168789e-06,
      "loss": 0.0019,
      "step": 179970
    },
    {
      "epoch": 1.1543358143621747,
      "grad_norm": 0.04605415090918541,
      "learning_rate": 4.526529769899251e-06,
      "loss": 0.0012,
      "step": 179980
    },
    {
      "epoch": 1.1543999512559608,
      "grad_norm": 0.040764909237623215,
      "learning_rate": 4.525972584562635e-06,
      "loss": 0.0014,
      "step": 179990
    },
    {
      "epoch": 1.1544640881497468,
      "grad_norm": 0.05737825483083725,
      "learning_rate": 4.525415405165919e-06,
      "loss": 0.002,
      "step": 180000
    },
    {
      "epoch": 1.154528225043533,
      "grad_norm": 0.05237356200814247,
      "learning_rate": 4.524858231716088e-06,
      "loss": 0.0022,
      "step": 180010
    },
    {
      "epoch": 1.154592361937319,
      "grad_norm": 0.11954223364591599,
      "learning_rate": 4.524301064220121e-06,
      "loss": 0.0019,
      "step": 180020
    },
    {
      "epoch": 1.154656498831105,
      "grad_norm": 0.022486301138997078,
      "learning_rate": 4.523743902685004e-06,
      "loss": 0.0032,
      "step": 180030
    },
    {
      "epoch": 1.1547206357248911,
      "grad_norm": 0.08458501845598221,
      "learning_rate": 4.523186747117715e-06,
      "loss": 0.0019,
      "step": 180040
    },
    {
      "epoch": 1.1547847726186773,
      "grad_norm": 0.016316739842295647,
      "learning_rate": 4.522629597525239e-06,
      "loss": 0.0009,
      "step": 180050
    },
    {
      "epoch": 1.1548489095124634,
      "grad_norm": 0.0723700150847435,
      "learning_rate": 4.5220724539145535e-06,
      "loss": 0.0037,
      "step": 180060
    },
    {
      "epoch": 1.1549130464062496,
      "grad_norm": 0.1529541164636612,
      "learning_rate": 4.521515316292643e-06,
      "loss": 0.0016,
      "step": 180070
    },
    {
      "epoch": 1.1549771833000355,
      "grad_norm": 0.07079547643661499,
      "learning_rate": 4.5209581846664865e-06,
      "loss": 0.0025,
      "step": 180080
    },
    {
      "epoch": 1.1550413201938217,
      "grad_norm": 0.05314158648252487,
      "learning_rate": 4.520401059043067e-06,
      "loss": 0.0011,
      "step": 180090
    },
    {
      "epoch": 1.1551054570876078,
      "grad_norm": 0.011576839722692966,
      "learning_rate": 4.519843939429364e-06,
      "loss": 0.0017,
      "step": 180100
    },
    {
      "epoch": 1.155169593981394,
      "grad_norm": 0.01953614130616188,
      "learning_rate": 4.519286825832361e-06,
      "loss": 0.0021,
      "step": 180110
    },
    {
      "epoch": 1.1552337308751799,
      "grad_norm": 0.11375365406274796,
      "learning_rate": 4.518729718259036e-06,
      "loss": 0.0015,
      "step": 180120
    },
    {
      "epoch": 1.155297867768966,
      "grad_norm": 0.060757264494895935,
      "learning_rate": 4.518172616716374e-06,
      "loss": 0.0008,
      "step": 180130
    },
    {
      "epoch": 1.1553620046627522,
      "grad_norm": 0.02616855502128601,
      "learning_rate": 4.517615521211351e-06,
      "loss": 0.0011,
      "step": 180140
    },
    {
      "epoch": 1.1554261415565383,
      "grad_norm": 0.035598281770944595,
      "learning_rate": 4.517058431750952e-06,
      "loss": 0.002,
      "step": 180150
    },
    {
      "epoch": 1.1554902784503245,
      "grad_norm": 0.08569321781396866,
      "learning_rate": 4.516501348342155e-06,
      "loss": 0.001,
      "step": 180160
    },
    {
      "epoch": 1.1555544153441104,
      "grad_norm": 0.030544036999344826,
      "learning_rate": 4.515944270991941e-06,
      "loss": 0.0009,
      "step": 180170
    },
    {
      "epoch": 1.1556185522378966,
      "grad_norm": 0.013408086262643337,
      "learning_rate": 4.515387199707293e-06,
      "loss": 0.0015,
      "step": 180180
    },
    {
      "epoch": 1.1556826891316827,
      "grad_norm": 0.028382427990436554,
      "learning_rate": 4.514830134495188e-06,
      "loss": 0.0012,
      "step": 180190
    },
    {
      "epoch": 1.1557468260254689,
      "grad_norm": 0.05651894956827164,
      "learning_rate": 4.51427307536261e-06,
      "loss": 0.0009,
      "step": 180200
    },
    {
      "epoch": 1.1558109629192548,
      "grad_norm": 0.19445115327835083,
      "learning_rate": 4.513716022316536e-06,
      "loss": 0.0015,
      "step": 180210
    },
    {
      "epoch": 1.155875099813041,
      "grad_norm": 0.3186042606830597,
      "learning_rate": 4.513158975363949e-06,
      "loss": 0.0035,
      "step": 180220
    },
    {
      "epoch": 1.155939236706827,
      "grad_norm": 0.09063344448804855,
      "learning_rate": 4.512601934511827e-06,
      "loss": 0.0018,
      "step": 180230
    },
    {
      "epoch": 1.1560033736006132,
      "grad_norm": 0.16141769289970398,
      "learning_rate": 4.5120448997671515e-06,
      "loss": 0.0016,
      "step": 180240
    },
    {
      "epoch": 1.1560675104943992,
      "grad_norm": 0.19638694822788239,
      "learning_rate": 4.511487871136901e-06,
      "loss": 0.0022,
      "step": 180250
    },
    {
      "epoch": 1.1561316473881853,
      "grad_norm": 0.07716047018766403,
      "learning_rate": 4.510930848628059e-06,
      "loss": 0.0018,
      "step": 180260
    },
    {
      "epoch": 1.1561957842819715,
      "grad_norm": 0.02431539073586464,
      "learning_rate": 4.510373832247601e-06,
      "loss": 0.0014,
      "step": 180270
    },
    {
      "epoch": 1.1562599211757576,
      "grad_norm": 0.07726875692605972,
      "learning_rate": 4.5098168220025104e-06,
      "loss": 0.0014,
      "step": 180280
    },
    {
      "epoch": 1.1563240580695435,
      "grad_norm": 0.09850283712148666,
      "learning_rate": 4.509259817899764e-06,
      "loss": 0.0009,
      "step": 180290
    },
    {
      "epoch": 1.1563881949633297,
      "grad_norm": 0.008121363818645477,
      "learning_rate": 4.508702819946343e-06,
      "loss": 0.0022,
      "step": 180300
    },
    {
      "epoch": 1.1564523318571158,
      "grad_norm": 0.036567483097314835,
      "learning_rate": 4.508145828149227e-06,
      "loss": 0.0034,
      "step": 180310
    },
    {
      "epoch": 1.156516468750902,
      "grad_norm": 0.0764060914516449,
      "learning_rate": 4.507588842515395e-06,
      "loss": 0.0013,
      "step": 180320
    },
    {
      "epoch": 1.1565806056446881,
      "grad_norm": 0.017248669639229774,
      "learning_rate": 4.507031863051828e-06,
      "loss": 0.0014,
      "step": 180330
    },
    {
      "epoch": 1.156644742538474,
      "grad_norm": 0.09835748374462128,
      "learning_rate": 4.506474889765502e-06,
      "loss": 0.0016,
      "step": 180340
    },
    {
      "epoch": 1.1567088794322602,
      "grad_norm": 0.46733415126800537,
      "learning_rate": 4.5059179226633995e-06,
      "loss": 0.0025,
      "step": 180350
    },
    {
      "epoch": 1.1567730163260463,
      "grad_norm": 0.06276906281709671,
      "learning_rate": 4.505360961752498e-06,
      "loss": 0.0012,
      "step": 180360
    },
    {
      "epoch": 1.1568371532198325,
      "grad_norm": 0.15917713940143585,
      "learning_rate": 4.5048040070397795e-06,
      "loss": 0.0012,
      "step": 180370
    },
    {
      "epoch": 1.1569012901136184,
      "grad_norm": 0.07704220712184906,
      "learning_rate": 4.504247058532218e-06,
      "loss": 0.0015,
      "step": 180380
    },
    {
      "epoch": 1.1569654270074046,
      "grad_norm": 0.12476428598165512,
      "learning_rate": 4.5036901162367964e-06,
      "loss": 0.0009,
      "step": 180390
    },
    {
      "epoch": 1.1570295639011907,
      "grad_norm": 0.46193230152130127,
      "learning_rate": 4.503133180160491e-06,
      "loss": 0.0017,
      "step": 180400
    },
    {
      "epoch": 1.1570937007949769,
      "grad_norm": 0.04050949588418007,
      "learning_rate": 4.502576250310284e-06,
      "loss": 0.0015,
      "step": 180410
    },
    {
      "epoch": 1.157157837688763,
      "grad_norm": 0.19467301666736603,
      "learning_rate": 4.50201932669315e-06,
      "loss": 0.0009,
      "step": 180420
    },
    {
      "epoch": 1.157221974582549,
      "grad_norm": 0.46448013186454773,
      "learning_rate": 4.5014624093160715e-06,
      "loss": 0.0016,
      "step": 180430
    },
    {
      "epoch": 1.157286111476335,
      "grad_norm": 0.2336282879114151,
      "learning_rate": 4.500905498186024e-06,
      "loss": 0.0014,
      "step": 180440
    },
    {
      "epoch": 1.1573502483701212,
      "grad_norm": 0.16977283358573914,
      "learning_rate": 4.500348593309988e-06,
      "loss": 0.0022,
      "step": 180450
    },
    {
      "epoch": 1.1574143852639072,
      "grad_norm": 0.034816574305295944,
      "learning_rate": 4.499791694694942e-06,
      "loss": 0.0025,
      "step": 180460
    },
    {
      "epoch": 1.1574785221576933,
      "grad_norm": 0.04419064521789551,
      "learning_rate": 4.499234802347862e-06,
      "loss": 0.0017,
      "step": 180470
    },
    {
      "epoch": 1.1575426590514795,
      "grad_norm": 0.19311970472335815,
      "learning_rate": 4.4986779162757284e-06,
      "loss": 0.0024,
      "step": 180480
    },
    {
      "epoch": 1.1576067959452656,
      "grad_norm": 0.10714399814605713,
      "learning_rate": 4.498121036485519e-06,
      "loss": 0.0018,
      "step": 180490
    },
    {
      "epoch": 1.1576709328390518,
      "grad_norm": 0.3718133568763733,
      "learning_rate": 4.497564162984212e-06,
      "loss": 0.0039,
      "step": 180500
    },
    {
      "epoch": 1.1577350697328377,
      "grad_norm": 0.039395760744810104,
      "learning_rate": 4.497007295778784e-06,
      "loss": 0.0015,
      "step": 180510
    },
    {
      "epoch": 1.1577992066266238,
      "grad_norm": 0.08320226520299911,
      "learning_rate": 4.496450434876215e-06,
      "loss": 0.0008,
      "step": 180520
    },
    {
      "epoch": 1.15786334352041,
      "grad_norm": 0.10056677460670471,
      "learning_rate": 4.495893580283482e-06,
      "loss": 0.001,
      "step": 180530
    },
    {
      "epoch": 1.1579274804141961,
      "grad_norm": 0.3260595202445984,
      "learning_rate": 4.495336732007562e-06,
      "loss": 0.0033,
      "step": 180540
    },
    {
      "epoch": 1.157991617307982,
      "grad_norm": 0.07820891588926315,
      "learning_rate": 4.494779890055434e-06,
      "loss": 0.0019,
      "step": 180550
    },
    {
      "epoch": 1.1580557542017682,
      "grad_norm": 0.10223450511693954,
      "learning_rate": 4.494223054434076e-06,
      "loss": 0.0017,
      "step": 180560
    },
    {
      "epoch": 1.1581198910955544,
      "grad_norm": 0.13785037398338318,
      "learning_rate": 4.493666225150462e-06,
      "loss": 0.0012,
      "step": 180570
    },
    {
      "epoch": 1.1581840279893405,
      "grad_norm": 0.008315321058034897,
      "learning_rate": 4.4931094022115755e-06,
      "loss": 0.0009,
      "step": 180580
    },
    {
      "epoch": 1.1582481648831267,
      "grad_norm": 0.009771126322448254,
      "learning_rate": 4.492552585624387e-06,
      "loss": 0.0026,
      "step": 180590
    },
    {
      "epoch": 1.1583123017769126,
      "grad_norm": 0.043915245682001114,
      "learning_rate": 4.491995775395879e-06,
      "loss": 0.0015,
      "step": 180600
    },
    {
      "epoch": 1.1583764386706987,
      "grad_norm": 0.05186295881867409,
      "learning_rate": 4.491438971533027e-06,
      "loss": 0.0012,
      "step": 180610
    },
    {
      "epoch": 1.1584405755644849,
      "grad_norm": 0.04817241430282593,
      "learning_rate": 4.490882174042808e-06,
      "loss": 0.0018,
      "step": 180620
    },
    {
      "epoch": 1.158504712458271,
      "grad_norm": 0.022592414170503616,
      "learning_rate": 4.490325382932199e-06,
      "loss": 0.0024,
      "step": 180630
    },
    {
      "epoch": 1.158568849352057,
      "grad_norm": 0.06265100836753845,
      "learning_rate": 4.489768598208177e-06,
      "loss": 0.0037,
      "step": 180640
    },
    {
      "epoch": 1.158632986245843,
      "grad_norm": 0.05838491767644882,
      "learning_rate": 4.48921181987772e-06,
      "loss": 0.0017,
      "step": 180650
    },
    {
      "epoch": 1.1586971231396292,
      "grad_norm": 0.0408768430352211,
      "learning_rate": 4.488655047947803e-06,
      "loss": 0.0016,
      "step": 180660
    },
    {
      "epoch": 1.1587612600334154,
      "grad_norm": 0.09586689621210098,
      "learning_rate": 4.488098282425405e-06,
      "loss": 0.0009,
      "step": 180670
    },
    {
      "epoch": 1.1588253969272013,
      "grad_norm": 0.06256552040576935,
      "learning_rate": 4.4875415233175e-06,
      "loss": 0.0031,
      "step": 180680
    },
    {
      "epoch": 1.1588895338209875,
      "grad_norm": 0.2888277769088745,
      "learning_rate": 4.486984770631067e-06,
      "loss": 0.0021,
      "step": 180690
    },
    {
      "epoch": 1.1589536707147736,
      "grad_norm": 0.0159669890999794,
      "learning_rate": 4.486428024373081e-06,
      "loss": 0.001,
      "step": 180700
    },
    {
      "epoch": 1.1590178076085598,
      "grad_norm": 0.04025883972644806,
      "learning_rate": 4.485871284550519e-06,
      "loss": 0.0021,
      "step": 180710
    },
    {
      "epoch": 1.1590819445023457,
      "grad_norm": 0.131544291973114,
      "learning_rate": 4.485314551170357e-06,
      "loss": 0.0016,
      "step": 180720
    },
    {
      "epoch": 1.1591460813961318,
      "grad_norm": 0.12201915681362152,
      "learning_rate": 4.4847578242395715e-06,
      "loss": 0.0017,
      "step": 180730
    },
    {
      "epoch": 1.159210218289918,
      "grad_norm": 0.019768880680203438,
      "learning_rate": 4.484201103765139e-06,
      "loss": 0.0016,
      "step": 180740
    },
    {
      "epoch": 1.1592743551837041,
      "grad_norm": 0.1573418527841568,
      "learning_rate": 4.483644389754034e-06,
      "loss": 0.0019,
      "step": 180750
    },
    {
      "epoch": 1.1593384920774903,
      "grad_norm": 0.15485508739948273,
      "learning_rate": 4.483087682213236e-06,
      "loss": 0.0015,
      "step": 180760
    },
    {
      "epoch": 1.1594026289712762,
      "grad_norm": 0.06819141656160355,
      "learning_rate": 4.482530981149717e-06,
      "loss": 0.0015,
      "step": 180770
    },
    {
      "epoch": 1.1594667658650624,
      "grad_norm": 0.34133756160736084,
      "learning_rate": 4.481974286570454e-06,
      "loss": 0.0012,
      "step": 180780
    },
    {
      "epoch": 1.1595309027588485,
      "grad_norm": 0.10663978010416031,
      "learning_rate": 4.481417598482426e-06,
      "loss": 0.0016,
      "step": 180790
    },
    {
      "epoch": 1.1595950396526347,
      "grad_norm": 0.09885643422603607,
      "learning_rate": 4.480860916892604e-06,
      "loss": 0.0027,
      "step": 180800
    },
    {
      "epoch": 1.1596591765464206,
      "grad_norm": 0.07657206803560257,
      "learning_rate": 4.480304241807967e-06,
      "loss": 0.0017,
      "step": 180810
    },
    {
      "epoch": 1.1597233134402067,
      "grad_norm": 0.021962689235806465,
      "learning_rate": 4.479747573235488e-06,
      "loss": 0.0013,
      "step": 180820
    },
    {
      "epoch": 1.1597874503339929,
      "grad_norm": 0.06031438335776329,
      "learning_rate": 4.479190911182145e-06,
      "loss": 0.0024,
      "step": 180830
    },
    {
      "epoch": 1.159851587227779,
      "grad_norm": 0.014199194498360157,
      "learning_rate": 4.4786342556549115e-06,
      "loss": 0.0015,
      "step": 180840
    },
    {
      "epoch": 1.1599157241215652,
      "grad_norm": 0.03586658835411072,
      "learning_rate": 4.478077606660764e-06,
      "loss": 0.001,
      "step": 180850
    },
    {
      "epoch": 1.159979861015351,
      "grad_norm": 0.16620095074176788,
      "learning_rate": 4.477520964206676e-06,
      "loss": 0.0014,
      "step": 180860
    },
    {
      "epoch": 1.1600439979091373,
      "grad_norm": 0.12353146821260452,
      "learning_rate": 4.476964328299624e-06,
      "loss": 0.0028,
      "step": 180870
    },
    {
      "epoch": 1.1601081348029234,
      "grad_norm": 0.012884975410997868,
      "learning_rate": 4.4764076989465845e-06,
      "loss": 0.0019,
      "step": 180880
    },
    {
      "epoch": 1.1601722716967093,
      "grad_norm": 0.08514288067817688,
      "learning_rate": 4.475851076154529e-06,
      "loss": 0.0016,
      "step": 180890
    },
    {
      "epoch": 1.1602364085904955,
      "grad_norm": 0.06588922441005707,
      "learning_rate": 4.475294459930435e-06,
      "loss": 0.0024,
      "step": 180900
    },
    {
      "epoch": 1.1603005454842816,
      "grad_norm": 0.05888795480132103,
      "learning_rate": 4.4747378502812755e-06,
      "loss": 0.0017,
      "step": 180910
    },
    {
      "epoch": 1.1603646823780678,
      "grad_norm": 0.062410056591033936,
      "learning_rate": 4.474181247214027e-06,
      "loss": 0.0032,
      "step": 180920
    },
    {
      "epoch": 1.160428819271854,
      "grad_norm": 0.08351512253284454,
      "learning_rate": 4.473624650735663e-06,
      "loss": 0.0018,
      "step": 180930
    },
    {
      "epoch": 1.1604929561656399,
      "grad_norm": 0.09259256720542908,
      "learning_rate": 4.473068060853159e-06,
      "loss": 0.0038,
      "step": 180940
    },
    {
      "epoch": 1.160557093059426,
      "grad_norm": 0.05656282231211662,
      "learning_rate": 4.472511477573487e-06,
      "loss": 0.002,
      "step": 180950
    },
    {
      "epoch": 1.1606212299532122,
      "grad_norm": 0.04642675444483757,
      "learning_rate": 4.471954900903625e-06,
      "loss": 0.0014,
      "step": 180960
    },
    {
      "epoch": 1.1606853668469983,
      "grad_norm": 0.09179715067148209,
      "learning_rate": 4.471398330850544e-06,
      "loss": 0.0021,
      "step": 180970
    },
    {
      "epoch": 1.1607495037407842,
      "grad_norm": 0.00894141849130392,
      "learning_rate": 4.470841767421221e-06,
      "loss": 0.0015,
      "step": 180980
    },
    {
      "epoch": 1.1608136406345704,
      "grad_norm": 0.012759844772517681,
      "learning_rate": 4.470285210622628e-06,
      "loss": 0.0011,
      "step": 180990
    },
    {
      "epoch": 1.1608777775283565,
      "grad_norm": 0.2721600830554962,
      "learning_rate": 4.4697286604617405e-06,
      "loss": 0.0022,
      "step": 181000
    },
    {
      "epoch": 1.1609419144221427,
      "grad_norm": 0.009490939788520336,
      "learning_rate": 4.469172116945531e-06,
      "loss": 0.0016,
      "step": 181010
    },
    {
      "epoch": 1.1610060513159288,
      "grad_norm": 0.028010396286845207,
      "learning_rate": 4.468615580080974e-06,
      "loss": 0.0017,
      "step": 181020
    },
    {
      "epoch": 1.1610701882097147,
      "grad_norm": 0.05484509468078613,
      "learning_rate": 4.468059049875045e-06,
      "loss": 0.001,
      "step": 181030
    },
    {
      "epoch": 1.161134325103501,
      "grad_norm": 0.07855560630559921,
      "learning_rate": 4.467502526334715e-06,
      "loss": 0.0005,
      "step": 181040
    },
    {
      "epoch": 1.161198461997287,
      "grad_norm": 0.13370922207832336,
      "learning_rate": 4.466946009466961e-06,
      "loss": 0.0031,
      "step": 181050
    },
    {
      "epoch": 1.1612625988910732,
      "grad_norm": 0.05313895642757416,
      "learning_rate": 4.466389499278751e-06,
      "loss": 0.0017,
      "step": 181060
    },
    {
      "epoch": 1.1613267357848591,
      "grad_norm": 0.12520724534988403,
      "learning_rate": 4.4658329957770655e-06,
      "loss": 0.0018,
      "step": 181070
    },
    {
      "epoch": 1.1613908726786453,
      "grad_norm": 0.1403217315673828,
      "learning_rate": 4.465276498968872e-06,
      "loss": 0.0022,
      "step": 181080
    },
    {
      "epoch": 1.1614550095724314,
      "grad_norm": 0.08550545573234558,
      "learning_rate": 4.464720008861147e-06,
      "loss": 0.0017,
      "step": 181090
    },
    {
      "epoch": 1.1615191464662176,
      "grad_norm": 0.05030914023518562,
      "learning_rate": 4.464163525460862e-06,
      "loss": 0.0018,
      "step": 181100
    },
    {
      "epoch": 1.1615832833600037,
      "grad_norm": 0.042203955352306366,
      "learning_rate": 4.463607048774993e-06,
      "loss": 0.0022,
      "step": 181110
    },
    {
      "epoch": 1.1616474202537896,
      "grad_norm": 0.12186636030673981,
      "learning_rate": 4.463050578810509e-06,
      "loss": 0.0025,
      "step": 181120
    },
    {
      "epoch": 1.1617115571475758,
      "grad_norm": 0.0916713997721672,
      "learning_rate": 4.462494115574387e-06,
      "loss": 0.0013,
      "step": 181130
    },
    {
      "epoch": 1.161775694041362,
      "grad_norm": 0.24145111441612244,
      "learning_rate": 4.461937659073595e-06,
      "loss": 0.0026,
      "step": 181140
    },
    {
      "epoch": 1.1618398309351479,
      "grad_norm": 0.07024310529232025,
      "learning_rate": 4.461381209315111e-06,
      "loss": 0.002,
      "step": 181150
    },
    {
      "epoch": 1.161903967828934,
      "grad_norm": 0.039558593183755875,
      "learning_rate": 4.460824766305904e-06,
      "loss": 0.0015,
      "step": 181160
    },
    {
      "epoch": 1.1619681047227202,
      "grad_norm": 0.35094472765922546,
      "learning_rate": 4.4602683300529484e-06,
      "loss": 0.0027,
      "step": 181170
    },
    {
      "epoch": 1.1620322416165063,
      "grad_norm": 0.030600082129240036,
      "learning_rate": 4.459711900563217e-06,
      "loss": 0.0017,
      "step": 181180
    },
    {
      "epoch": 1.1620963785102925,
      "grad_norm": 0.05822151154279709,
      "learning_rate": 4.459155477843681e-06,
      "loss": 0.0019,
      "step": 181190
    },
    {
      "epoch": 1.1621605154040784,
      "grad_norm": 0.03679274767637253,
      "learning_rate": 4.458599061901314e-06,
      "loss": 0.0014,
      "step": 181200
    },
    {
      "epoch": 1.1622246522978645,
      "grad_norm": 0.07842778414487839,
      "learning_rate": 4.458042652743087e-06,
      "loss": 0.0015,
      "step": 181210
    },
    {
      "epoch": 1.1622887891916507,
      "grad_norm": 0.08766324818134308,
      "learning_rate": 4.457486250375974e-06,
      "loss": 0.0016,
      "step": 181220
    },
    {
      "epoch": 1.1623529260854368,
      "grad_norm": 0.11263548582792282,
      "learning_rate": 4.456929854806944e-06,
      "loss": 0.0019,
      "step": 181230
    },
    {
      "epoch": 1.1624170629792228,
      "grad_norm": 0.03930861875414848,
      "learning_rate": 4.456373466042974e-06,
      "loss": 0.0011,
      "step": 181240
    },
    {
      "epoch": 1.162481199873009,
      "grad_norm": 0.16990233957767487,
      "learning_rate": 4.45581708409103e-06,
      "loss": 0.0023,
      "step": 181250
    },
    {
      "epoch": 1.162545336766795,
      "grad_norm": 0.07150521129369736,
      "learning_rate": 4.455260708958089e-06,
      "loss": 0.0014,
      "step": 181260
    },
    {
      "epoch": 1.1626094736605812,
      "grad_norm": 0.04427288472652435,
      "learning_rate": 4.454704340651119e-06,
      "loss": 0.0017,
      "step": 181270
    },
    {
      "epoch": 1.1626736105543674,
      "grad_norm": 0.033710263669490814,
      "learning_rate": 4.454147979177095e-06,
      "loss": 0.0015,
      "step": 181280
    },
    {
      "epoch": 1.1627377474481533,
      "grad_norm": 0.06993027776479721,
      "learning_rate": 4.4535916245429855e-06,
      "loss": 0.0014,
      "step": 181290
    },
    {
      "epoch": 1.1628018843419394,
      "grad_norm": 0.04299509897828102,
      "learning_rate": 4.4530352767557635e-06,
      "loss": 0.0007,
      "step": 181300
    },
    {
      "epoch": 1.1628660212357256,
      "grad_norm": 0.19627784192562103,
      "learning_rate": 4.452478935822402e-06,
      "loss": 0.0024,
      "step": 181310
    },
    {
      "epoch": 1.1629301581295117,
      "grad_norm": 0.0833597183227539,
      "learning_rate": 4.451922601749869e-06,
      "loss": 0.001,
      "step": 181320
    },
    {
      "epoch": 1.1629942950232977,
      "grad_norm": 0.11564180999994278,
      "learning_rate": 4.451366274545138e-06,
      "loss": 0.0021,
      "step": 181330
    },
    {
      "epoch": 1.1630584319170838,
      "grad_norm": 0.05993635207414627,
      "learning_rate": 4.4508099542151795e-06,
      "loss": 0.002,
      "step": 181340
    },
    {
      "epoch": 1.16312256881087,
      "grad_norm": 0.11969758570194244,
      "learning_rate": 4.450253640766966e-06,
      "loss": 0.0014,
      "step": 181350
    },
    {
      "epoch": 1.163186705704656,
      "grad_norm": 0.07571439445018768,
      "learning_rate": 4.449697334207465e-06,
      "loss": 0.001,
      "step": 181360
    },
    {
      "epoch": 1.163250842598442,
      "grad_norm": 0.39648115634918213,
      "learning_rate": 4.449141034543653e-06,
      "loss": 0.0023,
      "step": 181370
    },
    {
      "epoch": 1.1633149794922282,
      "grad_norm": 0.04993022233247757,
      "learning_rate": 4.448584741782495e-06,
      "loss": 0.0014,
      "step": 181380
    },
    {
      "epoch": 1.1633791163860143,
      "grad_norm": 0.04748213663697243,
      "learning_rate": 4.448028455930965e-06,
      "loss": 0.0025,
      "step": 181390
    },
    {
      "epoch": 1.1634432532798005,
      "grad_norm": 0.26994580030441284,
      "learning_rate": 4.447472176996033e-06,
      "loss": 0.002,
      "step": 181400
    },
    {
      "epoch": 1.1635073901735864,
      "grad_norm": 0.058311715722084045,
      "learning_rate": 4.44691590498467e-06,
      "loss": 0.0015,
      "step": 181410
    },
    {
      "epoch": 1.1635715270673725,
      "grad_norm": 0.019568253308534622,
      "learning_rate": 4.4463596399038444e-06,
      "loss": 0.0018,
      "step": 181420
    },
    {
      "epoch": 1.1636356639611587,
      "grad_norm": 0.07359485328197479,
      "learning_rate": 4.44580338176053e-06,
      "loss": 0.0029,
      "step": 181430
    },
    {
      "epoch": 1.1636998008549448,
      "grad_norm": 0.07900948077440262,
      "learning_rate": 4.445247130561694e-06,
      "loss": 0.0013,
      "step": 181440
    },
    {
      "epoch": 1.163763937748731,
      "grad_norm": 0.21989136934280396,
      "learning_rate": 4.444690886314307e-06,
      "loss": 0.0023,
      "step": 181450
    },
    {
      "epoch": 1.163828074642517,
      "grad_norm": 0.040261659771203995,
      "learning_rate": 4.4441346490253425e-06,
      "loss": 0.0014,
      "step": 181460
    },
    {
      "epoch": 1.163892211536303,
      "grad_norm": 0.05164617672562599,
      "learning_rate": 4.443578418701766e-06,
      "loss": 0.0018,
      "step": 181470
    },
    {
      "epoch": 1.1639563484300892,
      "grad_norm": 0.12991642951965332,
      "learning_rate": 4.443022195350551e-06,
      "loss": 0.0011,
      "step": 181480
    },
    {
      "epoch": 1.1640204853238754,
      "grad_norm": 0.1465480476617813,
      "learning_rate": 4.4424659789786655e-06,
      "loss": 0.002,
      "step": 181490
    },
    {
      "epoch": 1.1640846222176613,
      "grad_norm": 0.051542460918426514,
      "learning_rate": 4.4419097695930806e-06,
      "loss": 0.0018,
      "step": 181500
    },
    {
      "epoch": 1.1641487591114474,
      "grad_norm": 0.01723652519285679,
      "learning_rate": 4.441353567200763e-06,
      "loss": 0.0007,
      "step": 181510
    },
    {
      "epoch": 1.1642128960052336,
      "grad_norm": 0.03253493085503578,
      "learning_rate": 4.4407973718086876e-06,
      "loss": 0.0035,
      "step": 181520
    },
    {
      "epoch": 1.1642770328990197,
      "grad_norm": 0.4729132652282715,
      "learning_rate": 4.440241183423818e-06,
      "loss": 0.0011,
      "step": 181530
    },
    {
      "epoch": 1.1643411697928059,
      "grad_norm": 0.18357275426387787,
      "learning_rate": 4.439685002053128e-06,
      "loss": 0.0023,
      "step": 181540
    },
    {
      "epoch": 1.1644053066865918,
      "grad_norm": 0.06634698808193207,
      "learning_rate": 4.439128827703583e-06,
      "loss": 0.0023,
      "step": 181550
    },
    {
      "epoch": 1.164469443580378,
      "grad_norm": 0.16304150223731995,
      "learning_rate": 4.438572660382156e-06,
      "loss": 0.0016,
      "step": 181560
    },
    {
      "epoch": 1.164533580474164,
      "grad_norm": 0.030540192499756813,
      "learning_rate": 4.438016500095815e-06,
      "loss": 0.0016,
      "step": 181570
    },
    {
      "epoch": 1.16459771736795,
      "grad_norm": 0.054729048162698746,
      "learning_rate": 4.43746034685153e-06,
      "loss": 0.0008,
      "step": 181580
    },
    {
      "epoch": 1.1646618542617362,
      "grad_norm": 0.008067057467997074,
      "learning_rate": 4.436904200656267e-06,
      "loss": 0.0021,
      "step": 181590
    },
    {
      "epoch": 1.1647259911555223,
      "grad_norm": 0.36333948373794556,
      "learning_rate": 4.436348061516999e-06,
      "loss": 0.0009,
      "step": 181600
    },
    {
      "epoch": 1.1647901280493085,
      "grad_norm": 0.008159461431205273,
      "learning_rate": 4.435791929440691e-06,
      "loss": 0.0013,
      "step": 181610
    },
    {
      "epoch": 1.1648542649430946,
      "grad_norm": 0.05168044939637184,
      "learning_rate": 4.435235804434314e-06,
      "loss": 0.001,
      "step": 181620
    },
    {
      "epoch": 1.1649184018368806,
      "grad_norm": 0.05114075914025307,
      "learning_rate": 4.434679686504836e-06,
      "loss": 0.0011,
      "step": 181630
    },
    {
      "epoch": 1.1649825387306667,
      "grad_norm": 0.024549264460802078,
      "learning_rate": 4.4341235756592255e-06,
      "loss": 0.0016,
      "step": 181640
    },
    {
      "epoch": 1.1650466756244529,
      "grad_norm": 0.06715276092290878,
      "learning_rate": 4.43356747190445e-06,
      "loss": 0.0012,
      "step": 181650
    },
    {
      "epoch": 1.165110812518239,
      "grad_norm": 0.06436097621917725,
      "learning_rate": 4.433011375247481e-06,
      "loss": 0.001,
      "step": 181660
    },
    {
      "epoch": 1.165174949412025,
      "grad_norm": 0.13928824663162231,
      "learning_rate": 4.4324552856952826e-06,
      "loss": 0.0016,
      "step": 181670
    },
    {
      "epoch": 1.165239086305811,
      "grad_norm": 0.18230652809143066,
      "learning_rate": 4.431899203254827e-06,
      "loss": 0.0028,
      "step": 181680
    },
    {
      "epoch": 1.1653032231995972,
      "grad_norm": 0.09898312389850616,
      "learning_rate": 4.4313431279330785e-06,
      "loss": 0.0015,
      "step": 181690
    },
    {
      "epoch": 1.1653673600933834,
      "grad_norm": 0.12875409424304962,
      "learning_rate": 4.430787059737009e-06,
      "loss": 0.0021,
      "step": 181700
    },
    {
      "epoch": 1.1654314969871695,
      "grad_norm": 0.15830400586128235,
      "learning_rate": 4.430230998673582e-06,
      "loss": 0.0013,
      "step": 181710
    },
    {
      "epoch": 1.1654956338809555,
      "grad_norm": 0.18438397347927094,
      "learning_rate": 4.42967494474977e-06,
      "loss": 0.0013,
      "step": 181720
    },
    {
      "epoch": 1.1655597707747416,
      "grad_norm": 0.1835208535194397,
      "learning_rate": 4.429118897972538e-06,
      "loss": 0.0019,
      "step": 181730
    },
    {
      "epoch": 1.1656239076685277,
      "grad_norm": 0.14134618639945984,
      "learning_rate": 4.428562858348853e-06,
      "loss": 0.001,
      "step": 181740
    },
    {
      "epoch": 1.165688044562314,
      "grad_norm": 0.10361126065254211,
      "learning_rate": 4.428006825885686e-06,
      "loss": 0.0021,
      "step": 181750
    },
    {
      "epoch": 1.1657521814560998,
      "grad_norm": 0.11805353313684464,
      "learning_rate": 4.427450800590001e-06,
      "loss": 0.0019,
      "step": 181760
    },
    {
      "epoch": 1.165816318349886,
      "grad_norm": 0.13512155413627625,
      "learning_rate": 4.4268947824687665e-06,
      "loss": 0.0012,
      "step": 181770
    },
    {
      "epoch": 1.1658804552436721,
      "grad_norm": 0.029260262846946716,
      "learning_rate": 4.42633877152895e-06,
      "loss": 0.0013,
      "step": 181780
    },
    {
      "epoch": 1.1659445921374583,
      "grad_norm": 0.1251649260520935,
      "learning_rate": 4.4257827677775185e-06,
      "loss": 0.002,
      "step": 181790
    },
    {
      "epoch": 1.1660087290312442,
      "grad_norm": 0.03417348861694336,
      "learning_rate": 4.425226771221439e-06,
      "loss": 0.0008,
      "step": 181800
    },
    {
      "epoch": 1.1660728659250303,
      "grad_norm": 0.02154257707297802,
      "learning_rate": 4.42467078186768e-06,
      "loss": 0.0011,
      "step": 181810
    },
    {
      "epoch": 1.1661370028188165,
      "grad_norm": 0.14985492825508118,
      "learning_rate": 4.424114799723207e-06,
      "loss": 0.0022,
      "step": 181820
    },
    {
      "epoch": 1.1662011397126026,
      "grad_norm": 0.16996338963508606,
      "learning_rate": 4.423558824794987e-06,
      "loss": 0.0022,
      "step": 181830
    },
    {
      "epoch": 1.1662652766063886,
      "grad_norm": 0.06400048732757568,
      "learning_rate": 4.423002857089987e-06,
      "loss": 0.0016,
      "step": 181840
    },
    {
      "epoch": 1.1663294135001747,
      "grad_norm": 0.014239252544939518,
      "learning_rate": 4.422446896615174e-06,
      "loss": 0.0011,
      "step": 181850
    },
    {
      "epoch": 1.1663935503939609,
      "grad_norm": 0.13408030569553375,
      "learning_rate": 4.421890943377512e-06,
      "loss": 0.0033,
      "step": 181860
    },
    {
      "epoch": 1.166457687287747,
      "grad_norm": 0.12906645238399506,
      "learning_rate": 4.421334997383971e-06,
      "loss": 0.0012,
      "step": 181870
    },
    {
      "epoch": 1.1665218241815332,
      "grad_norm": 0.014006029814481735,
      "learning_rate": 4.420779058641517e-06,
      "loss": 0.0012,
      "step": 181880
    },
    {
      "epoch": 1.166585961075319,
      "grad_norm": 0.0008519539842382073,
      "learning_rate": 4.420223127157114e-06,
      "loss": 0.0006,
      "step": 181890
    },
    {
      "epoch": 1.1666500979691052,
      "grad_norm": 0.05692529305815697,
      "learning_rate": 4.419667202937731e-06,
      "loss": 0.0016,
      "step": 181900
    },
    {
      "epoch": 1.1667142348628914,
      "grad_norm": 0.023376788944005966,
      "learning_rate": 4.4191112859903305e-06,
      "loss": 0.0019,
      "step": 181910
    },
    {
      "epoch": 1.1667783717566775,
      "grad_norm": 0.09955395013093948,
      "learning_rate": 4.418555376321883e-06,
      "loss": 0.0018,
      "step": 181920
    },
    {
      "epoch": 1.1668425086504635,
      "grad_norm": 0.49170419573783875,
      "learning_rate": 4.41799947393935e-06,
      "loss": 0.0014,
      "step": 181930
    },
    {
      "epoch": 1.1669066455442496,
      "grad_norm": 0.1264665573835373,
      "learning_rate": 4.4174435788497025e-06,
      "loss": 0.0017,
      "step": 181940
    },
    {
      "epoch": 1.1669707824380358,
      "grad_norm": 0.041081368923187256,
      "learning_rate": 4.416887691059901e-06,
      "loss": 0.0014,
      "step": 181950
    },
    {
      "epoch": 1.167034919331822,
      "grad_norm": 0.05159052088856697,
      "learning_rate": 4.4163318105769145e-06,
      "loss": 0.0009,
      "step": 181960
    },
    {
      "epoch": 1.167099056225608,
      "grad_norm": 0.04607924073934555,
      "learning_rate": 4.415775937407706e-06,
      "loss": 0.0008,
      "step": 181970
    },
    {
      "epoch": 1.167163193119394,
      "grad_norm": 0.23263250291347504,
      "learning_rate": 4.415220071559244e-06,
      "loss": 0.0027,
      "step": 181980
    },
    {
      "epoch": 1.1672273300131801,
      "grad_norm": 0.14163386821746826,
      "learning_rate": 4.4146642130384925e-06,
      "loss": 0.0011,
      "step": 181990
    },
    {
      "epoch": 1.1672914669069663,
      "grad_norm": 0.1206795796751976,
      "learning_rate": 4.4141083618524175e-06,
      "loss": 0.0019,
      "step": 182000
    },
    {
      "epoch": 1.1673556038007522,
      "grad_norm": 0.02696717530488968,
      "learning_rate": 4.413552518007981e-06,
      "loss": 0.0016,
      "step": 182010
    },
    {
      "epoch": 1.1674197406945384,
      "grad_norm": 0.07129814475774765,
      "learning_rate": 4.412996681512152e-06,
      "loss": 0.0012,
      "step": 182020
    },
    {
      "epoch": 1.1674838775883245,
      "grad_norm": 0.09845761209726334,
      "learning_rate": 4.412440852371894e-06,
      "loss": 0.0019,
      "step": 182030
    },
    {
      "epoch": 1.1675480144821107,
      "grad_norm": 0.07310761511325836,
      "learning_rate": 4.411885030594172e-06,
      "loss": 0.0016,
      "step": 182040
    },
    {
      "epoch": 1.1676121513758968,
      "grad_norm": 0.030018869787454605,
      "learning_rate": 4.411329216185951e-06,
      "loss": 0.0017,
      "step": 182050
    },
    {
      "epoch": 1.1676762882696827,
      "grad_norm": 0.17031699419021606,
      "learning_rate": 4.410773409154195e-06,
      "loss": 0.0029,
      "step": 182060
    },
    {
      "epoch": 1.1677404251634689,
      "grad_norm": 0.09744299203157425,
      "learning_rate": 4.4102176095058705e-06,
      "loss": 0.0028,
      "step": 182070
    },
    {
      "epoch": 1.167804562057255,
      "grad_norm": 0.014784046448767185,
      "learning_rate": 4.40966181724794e-06,
      "loss": 0.0011,
      "step": 182080
    },
    {
      "epoch": 1.1678686989510412,
      "grad_norm": 0.08462289720773697,
      "learning_rate": 4.40910603238737e-06,
      "loss": 0.0015,
      "step": 182090
    },
    {
      "epoch": 1.167932835844827,
      "grad_norm": 0.12970688939094543,
      "learning_rate": 4.408550254931122e-06,
      "loss": 0.0018,
      "step": 182100
    },
    {
      "epoch": 1.1679969727386132,
      "grad_norm": 0.14601962268352509,
      "learning_rate": 4.407994484886163e-06,
      "loss": 0.0024,
      "step": 182110
    },
    {
      "epoch": 1.1680611096323994,
      "grad_norm": 0.29048222303390503,
      "learning_rate": 4.407438722259456e-06,
      "loss": 0.0028,
      "step": 182120
    },
    {
      "epoch": 1.1681252465261855,
      "grad_norm": 0.22827433049678802,
      "learning_rate": 4.406882967057966e-06,
      "loss": 0.0023,
      "step": 182130
    },
    {
      "epoch": 1.1681893834199717,
      "grad_norm": 0.14444595575332642,
      "learning_rate": 4.4063272192886555e-06,
      "loss": 0.0017,
      "step": 182140
    },
    {
      "epoch": 1.1682535203137576,
      "grad_norm": 0.08912989497184753,
      "learning_rate": 4.405771478958488e-06,
      "loss": 0.0022,
      "step": 182150
    },
    {
      "epoch": 1.1683176572075438,
      "grad_norm": 0.006506162695586681,
      "learning_rate": 4.405215746074431e-06,
      "loss": 0.001,
      "step": 182160
    },
    {
      "epoch": 1.16838179410133,
      "grad_norm": 0.16769647598266602,
      "learning_rate": 4.404660020643444e-06,
      "loss": 0.0025,
      "step": 182170
    },
    {
      "epoch": 1.168445930995116,
      "grad_norm": 0.06533028930425644,
      "learning_rate": 4.404104302672494e-06,
      "loss": 0.0031,
      "step": 182180
    },
    {
      "epoch": 1.168510067888902,
      "grad_norm": 0.23282302916049957,
      "learning_rate": 4.403548592168541e-06,
      "loss": 0.0027,
      "step": 182190
    },
    {
      "epoch": 1.1685742047826881,
      "grad_norm": 0.03258645161986351,
      "learning_rate": 4.4029928891385525e-06,
      "loss": 0.0008,
      "step": 182200
    },
    {
      "epoch": 1.1686383416764743,
      "grad_norm": 0.09809160977602005,
      "learning_rate": 4.4024371935894885e-06,
      "loss": 0.001,
      "step": 182210
    },
    {
      "epoch": 1.1687024785702604,
      "grad_norm": 0.07028601318597794,
      "learning_rate": 4.401881505528314e-06,
      "loss": 0.0013,
      "step": 182220
    },
    {
      "epoch": 1.1687666154640464,
      "grad_norm": 0.12784184515476227,
      "learning_rate": 4.401325824961991e-06,
      "loss": 0.0019,
      "step": 182230
    },
    {
      "epoch": 1.1688307523578325,
      "grad_norm": 0.0225824024528265,
      "learning_rate": 4.400770151897484e-06,
      "loss": 0.0014,
      "step": 182240
    },
    {
      "epoch": 1.1688948892516187,
      "grad_norm": 0.3825037181377411,
      "learning_rate": 4.400214486341755e-06,
      "loss": 0.0026,
      "step": 182250
    },
    {
      "epoch": 1.1689590261454048,
      "grad_norm": 0.11530786007642746,
      "learning_rate": 4.399658828301768e-06,
      "loss": 0.0019,
      "step": 182260
    },
    {
      "epoch": 1.1690231630391907,
      "grad_norm": 0.12124911695718765,
      "learning_rate": 4.399103177784484e-06,
      "loss": 0.0015,
      "step": 182270
    },
    {
      "epoch": 1.1690872999329769,
      "grad_norm": 0.038760680705308914,
      "learning_rate": 4.398547534796867e-06,
      "loss": 0.0016,
      "step": 182280
    },
    {
      "epoch": 1.169151436826763,
      "grad_norm": 0.0614054910838604,
      "learning_rate": 4.397991899345879e-06,
      "loss": 0.0016,
      "step": 182290
    },
    {
      "epoch": 1.1692155737205492,
      "grad_norm": 0.11623246222734451,
      "learning_rate": 4.397436271438482e-06,
      "loss": 0.0013,
      "step": 182300
    },
    {
      "epoch": 1.1692797106143353,
      "grad_norm": 0.14545269310474396,
      "learning_rate": 4.396880651081638e-06,
      "loss": 0.0019,
      "step": 182310
    },
    {
      "epoch": 1.1693438475081213,
      "grad_norm": 0.07068877667188644,
      "learning_rate": 4.396325038282313e-06,
      "loss": 0.0031,
      "step": 182320
    },
    {
      "epoch": 1.1694079844019074,
      "grad_norm": 0.09615910798311234,
      "learning_rate": 4.395769433047466e-06,
      "loss": 0.0015,
      "step": 182330
    },
    {
      "epoch": 1.1694721212956936,
      "grad_norm": 0.03125937655568123,
      "learning_rate": 4.395213835384061e-06,
      "loss": 0.0021,
      "step": 182340
    },
    {
      "epoch": 1.1695362581894797,
      "grad_norm": 0.21899282932281494,
      "learning_rate": 4.3946582452990565e-06,
      "loss": 0.0023,
      "step": 182350
    },
    {
      "epoch": 1.1696003950832656,
      "grad_norm": 0.09769497811794281,
      "learning_rate": 4.394102662799418e-06,
      "loss": 0.0024,
      "step": 182360
    },
    {
      "epoch": 1.1696645319770518,
      "grad_norm": 0.023459237068891525,
      "learning_rate": 4.3935470878921076e-06,
      "loss": 0.0018,
      "step": 182370
    },
    {
      "epoch": 1.169728668870838,
      "grad_norm": 0.1508914679288864,
      "learning_rate": 4.392991520584084e-06,
      "loss": 0.0007,
      "step": 182380
    },
    {
      "epoch": 1.169792805764624,
      "grad_norm": 0.2519901394844055,
      "learning_rate": 4.392435960882311e-06,
      "loss": 0.0013,
      "step": 182390
    },
    {
      "epoch": 1.1698569426584102,
      "grad_norm": 0.11893133074045181,
      "learning_rate": 4.39188040879375e-06,
      "loss": 0.0008,
      "step": 182400
    },
    {
      "epoch": 1.1699210795521962,
      "grad_norm": 0.1509605050086975,
      "learning_rate": 4.391324864325361e-06,
      "loss": 0.004,
      "step": 182410
    },
    {
      "epoch": 1.1699852164459823,
      "grad_norm": 0.1959735006093979,
      "learning_rate": 4.390769327484106e-06,
      "loss": 0.0026,
      "step": 182420
    },
    {
      "epoch": 1.1700493533397684,
      "grad_norm": 0.04640708118677139,
      "learning_rate": 4.390213798276949e-06,
      "loss": 0.0005,
      "step": 182430
    },
    {
      "epoch": 1.1701134902335544,
      "grad_norm": 0.02035406418144703,
      "learning_rate": 4.3896582767108465e-06,
      "loss": 0.003,
      "step": 182440
    },
    {
      "epoch": 1.1701776271273405,
      "grad_norm": 0.032201845198869705,
      "learning_rate": 4.389102762792764e-06,
      "loss": 0.0017,
      "step": 182450
    },
    {
      "epoch": 1.1702417640211267,
      "grad_norm": 0.018200211226940155,
      "learning_rate": 4.388547256529659e-06,
      "loss": 0.0013,
      "step": 182460
    },
    {
      "epoch": 1.1703059009149128,
      "grad_norm": 0.10391952842473984,
      "learning_rate": 4.387991757928495e-06,
      "loss": 0.0013,
      "step": 182470
    },
    {
      "epoch": 1.170370037808699,
      "grad_norm": 0.08152526617050171,
      "learning_rate": 4.387436266996231e-06,
      "loss": 0.0009,
      "step": 182480
    },
    {
      "epoch": 1.170434174702485,
      "grad_norm": 0.1280585378408432,
      "learning_rate": 4.386880783739829e-06,
      "loss": 0.0023,
      "step": 182490
    },
    {
      "epoch": 1.170498311596271,
      "grad_norm": 0.02015378326177597,
      "learning_rate": 4.386325308166249e-06,
      "loss": 0.0011,
      "step": 182500
    },
    {
      "epoch": 1.1705624484900572,
      "grad_norm": 0.13980241119861603,
      "learning_rate": 4.38576984028245e-06,
      "loss": 0.0013,
      "step": 182510
    },
    {
      "epoch": 1.1706265853838433,
      "grad_norm": 0.08781352639198303,
      "learning_rate": 4.385214380095395e-06,
      "loss": 0.0017,
      "step": 182520
    },
    {
      "epoch": 1.1706907222776293,
      "grad_norm": 0.19003696739673615,
      "learning_rate": 4.3846589276120435e-06,
      "loss": 0.0026,
      "step": 182530
    },
    {
      "epoch": 1.1707548591714154,
      "grad_norm": 0.025535663589835167,
      "learning_rate": 4.384103482839354e-06,
      "loss": 0.0012,
      "step": 182540
    },
    {
      "epoch": 1.1708189960652016,
      "grad_norm": 0.119336798787117,
      "learning_rate": 4.383548045784291e-06,
      "loss": 0.002,
      "step": 182550
    },
    {
      "epoch": 1.1708831329589877,
      "grad_norm": 0.05758288875222206,
      "learning_rate": 4.382992616453809e-06,
      "loss": 0.002,
      "step": 182560
    },
    {
      "epoch": 1.1709472698527739,
      "grad_norm": 0.025762133300304413,
      "learning_rate": 4.38243719485487e-06,
      "loss": 0.0023,
      "step": 182570
    },
    {
      "epoch": 1.1710114067465598,
      "grad_norm": 0.2017737179994583,
      "learning_rate": 4.381881780994436e-06,
      "loss": 0.0042,
      "step": 182580
    },
    {
      "epoch": 1.171075543640346,
      "grad_norm": 0.008837364614009857,
      "learning_rate": 4.381326374879464e-06,
      "loss": 0.0018,
      "step": 182590
    },
    {
      "epoch": 1.171139680534132,
      "grad_norm": 0.0664515420794487,
      "learning_rate": 4.380770976516916e-06,
      "loss": 0.0033,
      "step": 182600
    },
    {
      "epoch": 1.1712038174279182,
      "grad_norm": 0.07497132569551468,
      "learning_rate": 4.380215585913749e-06,
      "loss": 0.0015,
      "step": 182610
    },
    {
      "epoch": 1.1712679543217042,
      "grad_norm": 0.10247337073087692,
      "learning_rate": 4.3796602030769245e-06,
      "loss": 0.0026,
      "step": 182620
    },
    {
      "epoch": 1.1713320912154903,
      "grad_norm": 0.07052148878574371,
      "learning_rate": 4.3791048280133995e-06,
      "loss": 0.0019,
      "step": 182630
    },
    {
      "epoch": 1.1713962281092765,
      "grad_norm": 0.17377875745296478,
      "learning_rate": 4.378549460730136e-06,
      "loss": 0.0011,
      "step": 182640
    },
    {
      "epoch": 1.1714603650030626,
      "grad_norm": 0.10041338950395584,
      "learning_rate": 4.377994101234092e-06,
      "loss": 0.0014,
      "step": 182650
    },
    {
      "epoch": 1.1715245018968488,
      "grad_norm": 0.04387228563427925,
      "learning_rate": 4.377438749532226e-06,
      "loss": 0.001,
      "step": 182660
    },
    {
      "epoch": 1.1715886387906347,
      "grad_norm": 0.14890913665294647,
      "learning_rate": 4.376883405631497e-06,
      "loss": 0.0015,
      "step": 182670
    },
    {
      "epoch": 1.1716527756844208,
      "grad_norm": 0.11959514766931534,
      "learning_rate": 4.376328069538865e-06,
      "loss": 0.0017,
      "step": 182680
    },
    {
      "epoch": 1.171716912578207,
      "grad_norm": 0.1006447970867157,
      "learning_rate": 4.3757727412612874e-06,
      "loss": 0.0011,
      "step": 182690
    },
    {
      "epoch": 1.171781049471993,
      "grad_norm": 0.137902170419693,
      "learning_rate": 4.375217420805725e-06,
      "loss": 0.0019,
      "step": 182700
    },
    {
      "epoch": 1.171845186365779,
      "grad_norm": 0.1885043978691101,
      "learning_rate": 4.374662108179133e-06,
      "loss": 0.0009,
      "step": 182710
    },
    {
      "epoch": 1.1719093232595652,
      "grad_norm": 0.0794215202331543,
      "learning_rate": 4.374106803388472e-06,
      "loss": 0.0021,
      "step": 182720
    },
    {
      "epoch": 1.1719734601533514,
      "grad_norm": 0.3572762906551361,
      "learning_rate": 4.373551506440701e-06,
      "loss": 0.0031,
      "step": 182730
    },
    {
      "epoch": 1.1720375970471375,
      "grad_norm": 0.02289850264787674,
      "learning_rate": 4.372996217342776e-06,
      "loss": 0.0028,
      "step": 182740
    },
    {
      "epoch": 1.1721017339409234,
      "grad_norm": 0.02222614735364914,
      "learning_rate": 4.3724409361016585e-06,
      "loss": 0.0011,
      "step": 182750
    },
    {
      "epoch": 1.1721658708347096,
      "grad_norm": 0.11943326145410538,
      "learning_rate": 4.371885662724302e-06,
      "loss": 0.0019,
      "step": 182760
    },
    {
      "epoch": 1.1722300077284957,
      "grad_norm": 0.04962974786758423,
      "learning_rate": 4.3713303972176705e-06,
      "loss": 0.0011,
      "step": 182770
    },
    {
      "epoch": 1.1722941446222819,
      "grad_norm": 0.09099887311458588,
      "learning_rate": 4.370775139588715e-06,
      "loss": 0.0019,
      "step": 182780
    },
    {
      "epoch": 1.1723582815160678,
      "grad_norm": 0.08444700390100479,
      "learning_rate": 4.370219889844399e-06,
      "loss": 0.0018,
      "step": 182790
    },
    {
      "epoch": 1.172422418409854,
      "grad_norm": 0.1226532906293869,
      "learning_rate": 4.3696646479916766e-06,
      "loss": 0.0016,
      "step": 182800
    },
    {
      "epoch": 1.17248655530364,
      "grad_norm": 0.05741210654377937,
      "learning_rate": 4.369109414037508e-06,
      "loss": 0.001,
      "step": 182810
    },
    {
      "epoch": 1.1725506921974262,
      "grad_norm": 0.08488724380731583,
      "learning_rate": 4.368554187988847e-06,
      "loss": 0.0013,
      "step": 182820
    },
    {
      "epoch": 1.1726148290912124,
      "grad_norm": 0.2592843174934387,
      "learning_rate": 4.367998969852656e-06,
      "loss": 0.0012,
      "step": 182830
    },
    {
      "epoch": 1.1726789659849983,
      "grad_norm": 0.03624601289629936,
      "learning_rate": 4.367443759635888e-06,
      "loss": 0.0008,
      "step": 182840
    },
    {
      "epoch": 1.1727431028787845,
      "grad_norm": 0.07296912372112274,
      "learning_rate": 4.366888557345503e-06,
      "loss": 0.0014,
      "step": 182850
    },
    {
      "epoch": 1.1728072397725706,
      "grad_norm": 0.08172671496868134,
      "learning_rate": 4.366333362988455e-06,
      "loss": 0.0016,
      "step": 182860
    },
    {
      "epoch": 1.1728713766663568,
      "grad_norm": 0.12368035316467285,
      "learning_rate": 4.3657781765717026e-06,
      "loss": 0.002,
      "step": 182870
    },
    {
      "epoch": 1.1729355135601427,
      "grad_norm": 0.0981890931725502,
      "learning_rate": 4.365222998102205e-06,
      "loss": 0.0026,
      "step": 182880
    },
    {
      "epoch": 1.1729996504539288,
      "grad_norm": 0.08738455921411514,
      "learning_rate": 4.364667827586915e-06,
      "loss": 0.0011,
      "step": 182890
    },
    {
      "epoch": 1.173063787347715,
      "grad_norm": 0.08175847679376602,
      "learning_rate": 4.364112665032793e-06,
      "loss": 0.0013,
      "step": 182900
    },
    {
      "epoch": 1.1731279242415011,
      "grad_norm": 0.14890995621681213,
      "learning_rate": 4.363557510446792e-06,
      "loss": 0.0016,
      "step": 182910
    },
    {
      "epoch": 1.173192061135287,
      "grad_norm": 0.04660869762301445,
      "learning_rate": 4.363002363835872e-06,
      "loss": 0.0019,
      "step": 182920
    },
    {
      "epoch": 1.1732561980290732,
      "grad_norm": 0.003985627554357052,
      "learning_rate": 4.3624472252069865e-06,
      "loss": 0.0008,
      "step": 182930
    },
    {
      "epoch": 1.1733203349228594,
      "grad_norm": 0.0942840576171875,
      "learning_rate": 4.361892094567094e-06,
      "loss": 0.0018,
      "step": 182940
    },
    {
      "epoch": 1.1733844718166455,
      "grad_norm": 0.10258271545171738,
      "learning_rate": 4.361336971923149e-06,
      "loss": 0.0014,
      "step": 182950
    },
    {
      "epoch": 1.1734486087104314,
      "grad_norm": 0.12442044168710709,
      "learning_rate": 4.360781857282108e-06,
      "loss": 0.001,
      "step": 182960
    },
    {
      "epoch": 1.1735127456042176,
      "grad_norm": 0.07931763678789139,
      "learning_rate": 4.360226750650926e-06,
      "loss": 0.0019,
      "step": 182970
    },
    {
      "epoch": 1.1735768824980037,
      "grad_norm": 0.018039437010884285,
      "learning_rate": 4.359671652036562e-06,
      "loss": 0.001,
      "step": 182980
    },
    {
      "epoch": 1.1736410193917899,
      "grad_norm": 0.03374273329973221,
      "learning_rate": 4.359116561445969e-06,
      "loss": 0.004,
      "step": 182990
    },
    {
      "epoch": 1.173705156285576,
      "grad_norm": 0.003924379590898752,
      "learning_rate": 4.358561478886103e-06,
      "loss": 0.0023,
      "step": 183000
    },
    {
      "epoch": 1.173705156285576,
      "eval_loss": 0.0024000562261790037,
      "eval_runtime": 3.3153,
      "eval_samples_per_second": 60.326,
      "eval_steps_per_second": 15.082,
      "step": 183000
    },
    {
      "epoch": 1.173769293179362,
      "grad_norm": 0.05327145382761955,
      "learning_rate": 4.358006404363921e-06,
      "loss": 0.0025,
      "step": 183010
    },
    {
      "epoch": 1.173833430073148,
      "grad_norm": 0.014441578648984432,
      "learning_rate": 4.357451337886377e-06,
      "loss": 0.0015,
      "step": 183020
    },
    {
      "epoch": 1.1738975669669343,
      "grad_norm": 0.18125401437282562,
      "learning_rate": 4.356896279460428e-06,
      "loss": 0.0013,
      "step": 183030
    },
    {
      "epoch": 1.1739617038607204,
      "grad_norm": 0.05702318251132965,
      "learning_rate": 4.356341229093028e-06,
      "loss": 0.0022,
      "step": 183040
    },
    {
      "epoch": 1.1740258407545063,
      "grad_norm": 0.22551800310611725,
      "learning_rate": 4.355786186791132e-06,
      "loss": 0.0019,
      "step": 183050
    },
    {
      "epoch": 1.1740899776482925,
      "grad_norm": 0.04299104958772659,
      "learning_rate": 4.355231152561694e-06,
      "loss": 0.002,
      "step": 183060
    },
    {
      "epoch": 1.1741541145420786,
      "grad_norm": 0.14469648897647858,
      "learning_rate": 4.354676126411672e-06,
      "loss": 0.0016,
      "step": 183070
    },
    {
      "epoch": 1.1742182514358648,
      "grad_norm": 0.06454824656248093,
      "learning_rate": 4.354121108348018e-06,
      "loss": 0.0011,
      "step": 183080
    },
    {
      "epoch": 1.174282388329651,
      "grad_norm": 0.09450476616621017,
      "learning_rate": 4.353566098377689e-06,
      "loss": 0.0013,
      "step": 183090
    },
    {
      "epoch": 1.1743465252234369,
      "grad_norm": 0.08068623393774033,
      "learning_rate": 4.35301109650764e-06,
      "loss": 0.0017,
      "step": 183100
    },
    {
      "epoch": 1.174410662117223,
      "grad_norm": 0.06034604832530022,
      "learning_rate": 4.352456102744823e-06,
      "loss": 0.0016,
      "step": 183110
    },
    {
      "epoch": 1.1744747990110092,
      "grad_norm": 0.06156300753355026,
      "learning_rate": 4.3519011170961945e-06,
      "loss": 0.0013,
      "step": 183120
    },
    {
      "epoch": 1.174538935904795,
      "grad_norm": 0.017585139721632004,
      "learning_rate": 4.351346139568708e-06,
      "loss": 0.0016,
      "step": 183130
    },
    {
      "epoch": 1.1746030727985812,
      "grad_norm": 0.08957716077566147,
      "learning_rate": 4.3507911701693175e-06,
      "loss": 0.0021,
      "step": 183140
    },
    {
      "epoch": 1.1746672096923674,
      "grad_norm": 0.018210873007774353,
      "learning_rate": 4.350236208904978e-06,
      "loss": 0.0022,
      "step": 183150
    },
    {
      "epoch": 1.1747313465861535,
      "grad_norm": 0.26158273220062256,
      "learning_rate": 4.349681255782643e-06,
      "loss": 0.0011,
      "step": 183160
    },
    {
      "epoch": 1.1747954834799397,
      "grad_norm": 0.08066187053918839,
      "learning_rate": 4.3491263108092675e-06,
      "loss": 0.0012,
      "step": 183170
    },
    {
      "epoch": 1.1748596203737256,
      "grad_norm": 0.07623965293169022,
      "learning_rate": 4.348571373991803e-06,
      "loss": 0.0006,
      "step": 183180
    },
    {
      "epoch": 1.1749237572675117,
      "grad_norm": 0.11561774462461472,
      "learning_rate": 4.348016445337206e-06,
      "loss": 0.0015,
      "step": 183190
    },
    {
      "epoch": 1.174987894161298,
      "grad_norm": 0.1563999056816101,
      "learning_rate": 4.347461524852429e-06,
      "loss": 0.0013,
      "step": 183200
    },
    {
      "epoch": 1.175052031055084,
      "grad_norm": 0.0295012928545475,
      "learning_rate": 4.346906612544425e-06,
      "loss": 0.0006,
      "step": 183210
    },
    {
      "epoch": 1.17511616794887,
      "grad_norm": 0.08366100490093231,
      "learning_rate": 4.346351708420147e-06,
      "loss": 0.0021,
      "step": 183220
    },
    {
      "epoch": 1.1751803048426561,
      "grad_norm": 0.05583413690328598,
      "learning_rate": 4.345796812486552e-06,
      "loss": 0.0014,
      "step": 183230
    },
    {
      "epoch": 1.1752444417364423,
      "grad_norm": 0.09528223425149918,
      "learning_rate": 4.345241924750588e-06,
      "loss": 0.0021,
      "step": 183240
    },
    {
      "epoch": 1.1753085786302284,
      "grad_norm": 0.08164983987808228,
      "learning_rate": 4.344687045219212e-06,
      "loss": 0.0012,
      "step": 183250
    },
    {
      "epoch": 1.1753727155240146,
      "grad_norm": 0.027242237702012062,
      "learning_rate": 4.344132173899374e-06,
      "loss": 0.002,
      "step": 183260
    },
    {
      "epoch": 1.1754368524178005,
      "grad_norm": 0.20938929915428162,
      "learning_rate": 4.343577310798028e-06,
      "loss": 0.0012,
      "step": 183270
    },
    {
      "epoch": 1.1755009893115866,
      "grad_norm": 0.14060692489147186,
      "learning_rate": 4.3430224559221305e-06,
      "loss": 0.0017,
      "step": 183280
    },
    {
      "epoch": 1.1755651262053728,
      "grad_norm": 0.18034429848194122,
      "learning_rate": 4.342467609278629e-06,
      "loss": 0.0012,
      "step": 183290
    },
    {
      "epoch": 1.175629263099159,
      "grad_norm": 0.029719088226556778,
      "learning_rate": 4.341912770874479e-06,
      "loss": 0.0012,
      "step": 183300
    },
    {
      "epoch": 1.1756933999929449,
      "grad_norm": 0.06746480613946915,
      "learning_rate": 4.341357940716631e-06,
      "loss": 0.0014,
      "step": 183310
    },
    {
      "epoch": 1.175757536886731,
      "grad_norm": 0.07167258113622665,
      "learning_rate": 4.340803118812042e-06,
      "loss": 0.0023,
      "step": 183320
    },
    {
      "epoch": 1.1758216737805172,
      "grad_norm": 0.4483695924282074,
      "learning_rate": 4.340248305167658e-06,
      "loss": 0.003,
      "step": 183330
    },
    {
      "epoch": 1.1758858106743033,
      "grad_norm": 0.029790734872221947,
      "learning_rate": 4.339693499790436e-06,
      "loss": 0.0028,
      "step": 183340
    },
    {
      "epoch": 1.1759499475680892,
      "grad_norm": 0.10472667962312698,
      "learning_rate": 4.3391387026873246e-06,
      "loss": 0.001,
      "step": 183350
    },
    {
      "epoch": 1.1760140844618754,
      "grad_norm": 0.17653949558734894,
      "learning_rate": 4.3385839138652796e-06,
      "loss": 0.002,
      "step": 183360
    },
    {
      "epoch": 1.1760782213556615,
      "grad_norm": 0.043926745653152466,
      "learning_rate": 4.338029133331249e-06,
      "loss": 0.0015,
      "step": 183370
    },
    {
      "epoch": 1.1761423582494477,
      "grad_norm": 0.14336556196212769,
      "learning_rate": 4.3374743610921886e-06,
      "loss": 0.0014,
      "step": 183380
    },
    {
      "epoch": 1.1762064951432336,
      "grad_norm": 0.04863275960087776,
      "learning_rate": 4.336919597155046e-06,
      "loss": 0.0022,
      "step": 183390
    },
    {
      "epoch": 1.1762706320370198,
      "grad_norm": 0.06869960576295853,
      "learning_rate": 4.336364841526775e-06,
      "loss": 0.0009,
      "step": 183400
    },
    {
      "epoch": 1.176334768930806,
      "grad_norm": 0.17010624706745148,
      "learning_rate": 4.335810094214327e-06,
      "loss": 0.0025,
      "step": 183410
    },
    {
      "epoch": 1.176398905824592,
      "grad_norm": 0.03901492431759834,
      "learning_rate": 4.335255355224653e-06,
      "loss": 0.0018,
      "step": 183420
    },
    {
      "epoch": 1.1764630427183782,
      "grad_norm": 0.10105922818183899,
      "learning_rate": 4.334700624564706e-06,
      "loss": 0.0012,
      "step": 183430
    },
    {
      "epoch": 1.1765271796121641,
      "grad_norm": 0.07125986367464066,
      "learning_rate": 4.334145902241434e-06,
      "loss": 0.0019,
      "step": 183440
    },
    {
      "epoch": 1.1765913165059503,
      "grad_norm": 0.022353487089276314,
      "learning_rate": 4.333591188261791e-06,
      "loss": 0.0011,
      "step": 183450
    },
    {
      "epoch": 1.1766554533997364,
      "grad_norm": 0.07350007444620132,
      "learning_rate": 4.333036482632725e-06,
      "loss": 0.0008,
      "step": 183460
    },
    {
      "epoch": 1.1767195902935226,
      "grad_norm": 0.1433224380016327,
      "learning_rate": 4.332481785361191e-06,
      "loss": 0.0016,
      "step": 183470
    },
    {
      "epoch": 1.1767837271873085,
      "grad_norm": 0.04887795448303223,
      "learning_rate": 4.331927096454135e-06,
      "loss": 0.0012,
      "step": 183480
    },
    {
      "epoch": 1.1768478640810947,
      "grad_norm": 0.13807959854602814,
      "learning_rate": 4.331372415918511e-06,
      "loss": 0.0023,
      "step": 183490
    },
    {
      "epoch": 1.1769120009748808,
      "grad_norm": 0.04304143041372299,
      "learning_rate": 4.330817743761267e-06,
      "loss": 0.0008,
      "step": 183500
    },
    {
      "epoch": 1.176976137868667,
      "grad_norm": 0.09120394289493561,
      "learning_rate": 4.330263079989358e-06,
      "loss": 0.0012,
      "step": 183510
    },
    {
      "epoch": 1.177040274762453,
      "grad_norm": 0.07101622223854065,
      "learning_rate": 4.3297084246097285e-06,
      "loss": 0.0013,
      "step": 183520
    },
    {
      "epoch": 1.177104411656239,
      "grad_norm": 0.07521035522222519,
      "learning_rate": 4.329153777629333e-06,
      "loss": 0.0016,
      "step": 183530
    },
    {
      "epoch": 1.1771685485500252,
      "grad_norm": 0.14560550451278687,
      "learning_rate": 4.328599139055119e-06,
      "loss": 0.0017,
      "step": 183540
    },
    {
      "epoch": 1.1772326854438113,
      "grad_norm": 0.08797882497310638,
      "learning_rate": 4.328044508894039e-06,
      "loss": 0.0024,
      "step": 183550
    },
    {
      "epoch": 1.1772968223375972,
      "grad_norm": 0.1541559100151062,
      "learning_rate": 4.32748988715304e-06,
      "loss": 0.0018,
      "step": 183560
    },
    {
      "epoch": 1.1773609592313834,
      "grad_norm": 0.1894298642873764,
      "learning_rate": 4.326935273839074e-06,
      "loss": 0.0018,
      "step": 183570
    },
    {
      "epoch": 1.1774250961251695,
      "grad_norm": 0.05562448874115944,
      "learning_rate": 4.326380668959091e-06,
      "loss": 0.0017,
      "step": 183580
    },
    {
      "epoch": 1.1774892330189557,
      "grad_norm": 0.14997988939285278,
      "learning_rate": 4.3258260725200374e-06,
      "loss": 0.0023,
      "step": 183590
    },
    {
      "epoch": 1.1775533699127418,
      "grad_norm": 0.23341451585292816,
      "learning_rate": 4.325271484528867e-06,
      "loss": 0.0018,
      "step": 183600
    },
    {
      "epoch": 1.1776175068065278,
      "grad_norm": 0.2117612212896347,
      "learning_rate": 4.324716904992527e-06,
      "loss": 0.0022,
      "step": 183610
    },
    {
      "epoch": 1.177681643700314,
      "grad_norm": 0.18917398154735565,
      "learning_rate": 4.3241623339179665e-06,
      "loss": 0.003,
      "step": 183620
    },
    {
      "epoch": 1.1777457805941,
      "grad_norm": 0.07607435435056686,
      "learning_rate": 4.323607771312134e-06,
      "loss": 0.0015,
      "step": 183630
    },
    {
      "epoch": 1.1778099174878862,
      "grad_norm": 0.08164612948894501,
      "learning_rate": 4.323053217181981e-06,
      "loss": 0.0018,
      "step": 183640
    },
    {
      "epoch": 1.1778740543816721,
      "grad_norm": 0.03821833059191704,
      "learning_rate": 4.322498671534453e-06,
      "loss": 0.0016,
      "step": 183650
    },
    {
      "epoch": 1.1779381912754583,
      "grad_norm": 0.018987243995070457,
      "learning_rate": 4.321944134376503e-06,
      "loss": 0.0017,
      "step": 183660
    },
    {
      "epoch": 1.1780023281692444,
      "grad_norm": 0.0716153010725975,
      "learning_rate": 4.321389605715076e-06,
      "loss": 0.001,
      "step": 183670
    },
    {
      "epoch": 1.1780664650630306,
      "grad_norm": 0.12727735936641693,
      "learning_rate": 4.3208350855571235e-06,
      "loss": 0.0013,
      "step": 183680
    },
    {
      "epoch": 1.1781306019568167,
      "grad_norm": 0.3035914897918701,
      "learning_rate": 4.320280573909592e-06,
      "loss": 0.002,
      "step": 183690
    },
    {
      "epoch": 1.1781947388506027,
      "grad_norm": 0.22567832469940186,
      "learning_rate": 4.3197260707794315e-06,
      "loss": 0.0017,
      "step": 183700
    },
    {
      "epoch": 1.1782588757443888,
      "grad_norm": 0.12561731040477753,
      "learning_rate": 4.319171576173588e-06,
      "loss": 0.0006,
      "step": 183710
    },
    {
      "epoch": 1.178323012638175,
      "grad_norm": 0.11747381836175919,
      "learning_rate": 4.318617090099012e-06,
      "loss": 0.0017,
      "step": 183720
    },
    {
      "epoch": 1.178387149531961,
      "grad_norm": 0.10898769646883011,
      "learning_rate": 4.3180626125626515e-06,
      "loss": 0.0027,
      "step": 183730
    },
    {
      "epoch": 1.178451286425747,
      "grad_norm": 0.31439948081970215,
      "learning_rate": 4.317508143571453e-06,
      "loss": 0.0032,
      "step": 183740
    },
    {
      "epoch": 1.1785154233195332,
      "grad_norm": 0.0507233664393425,
      "learning_rate": 4.316953683132365e-06,
      "loss": 0.0019,
      "step": 183750
    },
    {
      "epoch": 1.1785795602133193,
      "grad_norm": 0.08064527064561844,
      "learning_rate": 4.316399231252336e-06,
      "loss": 0.0012,
      "step": 183760
    },
    {
      "epoch": 1.1786436971071055,
      "grad_norm": 0.04937786981463432,
      "learning_rate": 4.3158447879383145e-06,
      "loss": 0.0018,
      "step": 183770
    },
    {
      "epoch": 1.1787078340008914,
      "grad_norm": 0.17828668653964996,
      "learning_rate": 4.315290353197244e-06,
      "loss": 0.0014,
      "step": 183780
    },
    {
      "epoch": 1.1787719708946776,
      "grad_norm": 0.08378896117210388,
      "learning_rate": 4.314735927036078e-06,
      "loss": 0.002,
      "step": 183790
    },
    {
      "epoch": 1.1788361077884637,
      "grad_norm": 0.092821404337883,
      "learning_rate": 4.314181509461758e-06,
      "loss": 0.0017,
      "step": 183800
    },
    {
      "epoch": 1.1789002446822499,
      "grad_norm": 0.013002404943108559,
      "learning_rate": 4.313627100481234e-06,
      "loss": 0.0017,
      "step": 183810
    },
    {
      "epoch": 1.1789643815760358,
      "grad_norm": 0.004985139239579439,
      "learning_rate": 4.313072700101454e-06,
      "loss": 0.0013,
      "step": 183820
    },
    {
      "epoch": 1.179028518469822,
      "grad_norm": 0.11837707459926605,
      "learning_rate": 4.312518308329364e-06,
      "loss": 0.0019,
      "step": 183830
    },
    {
      "epoch": 1.179092655363608,
      "grad_norm": 0.22910629212856293,
      "learning_rate": 4.311963925171908e-06,
      "loss": 0.0011,
      "step": 183840
    },
    {
      "epoch": 1.1791567922573942,
      "grad_norm": 0.01819387823343277,
      "learning_rate": 4.31140955063604e-06,
      "loss": 0.002,
      "step": 183850
    },
    {
      "epoch": 1.1792209291511804,
      "grad_norm": 0.06624264270067215,
      "learning_rate": 4.3108551847287e-06,
      "loss": 0.001,
      "step": 183860
    },
    {
      "epoch": 1.1792850660449663,
      "grad_norm": 0.16025210916996002,
      "learning_rate": 4.310300827456838e-06,
      "loss": 0.0017,
      "step": 183870
    },
    {
      "epoch": 1.1793492029387525,
      "grad_norm": 0.11636680364608765,
      "learning_rate": 4.309746478827399e-06,
      "loss": 0.0022,
      "step": 183880
    },
    {
      "epoch": 1.1794133398325386,
      "grad_norm": 0.005025495775043964,
      "learning_rate": 4.3091921388473305e-06,
      "loss": 0.0017,
      "step": 183890
    },
    {
      "epoch": 1.1794774767263247,
      "grad_norm": 0.04582427814602852,
      "learning_rate": 4.308637807523578e-06,
      "loss": 0.0009,
      "step": 183900
    },
    {
      "epoch": 1.1795416136201107,
      "grad_norm": 0.10134973376989365,
      "learning_rate": 4.308083484863088e-06,
      "loss": 0.0016,
      "step": 183910
    },
    {
      "epoch": 1.1796057505138968,
      "grad_norm": 0.013092214241623878,
      "learning_rate": 4.307529170872806e-06,
      "loss": 0.0028,
      "step": 183920
    },
    {
      "epoch": 1.179669887407683,
      "grad_norm": 0.06885766983032227,
      "learning_rate": 4.30697486555968e-06,
      "loss": 0.0032,
      "step": 183930
    },
    {
      "epoch": 1.1797340243014691,
      "grad_norm": 0.02871783636510372,
      "learning_rate": 4.306420568930652e-06,
      "loss": 0.0019,
      "step": 183940
    },
    {
      "epoch": 1.1797981611952553,
      "grad_norm": 0.04335256665945053,
      "learning_rate": 4.305866280992671e-06,
      "loss": 0.0006,
      "step": 183950
    },
    {
      "epoch": 1.1798622980890412,
      "grad_norm": 0.061511993408203125,
      "learning_rate": 4.305312001752682e-06,
      "loss": 0.0012,
      "step": 183960
    },
    {
      "epoch": 1.1799264349828273,
      "grad_norm": 0.07183343172073364,
      "learning_rate": 4.30475773121763e-06,
      "loss": 0.0012,
      "step": 183970
    },
    {
      "epoch": 1.1799905718766135,
      "grad_norm": 0.109437957406044,
      "learning_rate": 4.30420346939446e-06,
      "loss": 0.0013,
      "step": 183980
    },
    {
      "epoch": 1.1800547087703994,
      "grad_norm": 0.15795163810253143,
      "learning_rate": 4.303649216290117e-06,
      "loss": 0.0023,
      "step": 183990
    },
    {
      "epoch": 1.1801188456641856,
      "grad_norm": 0.12437650561332703,
      "learning_rate": 4.303094971911548e-06,
      "loss": 0.002,
      "step": 184000
    },
    {
      "epoch": 1.1801829825579717,
      "grad_norm": 0.07483706623315811,
      "learning_rate": 4.302540736265697e-06,
      "loss": 0.0017,
      "step": 184010
    },
    {
      "epoch": 1.1802471194517579,
      "grad_norm": 0.00944068469107151,
      "learning_rate": 4.301986509359509e-06,
      "loss": 0.0016,
      "step": 184020
    },
    {
      "epoch": 1.180311256345544,
      "grad_norm": 0.09464728832244873,
      "learning_rate": 4.301432291199929e-06,
      "loss": 0.0019,
      "step": 184030
    },
    {
      "epoch": 1.18037539323933,
      "grad_norm": 0.0372900627553463,
      "learning_rate": 4.300878081793902e-06,
      "loss": 0.0011,
      "step": 184040
    },
    {
      "epoch": 1.180439530133116,
      "grad_norm": 0.23563946783542633,
      "learning_rate": 4.300323881148371e-06,
      "loss": 0.0013,
      "step": 184050
    },
    {
      "epoch": 1.1805036670269022,
      "grad_norm": 0.17711228132247925,
      "learning_rate": 4.299769689270284e-06,
      "loss": 0.002,
      "step": 184060
    },
    {
      "epoch": 1.1805678039206884,
      "grad_norm": 0.12385842949151993,
      "learning_rate": 4.299215506166581e-06,
      "loss": 0.0014,
      "step": 184070
    },
    {
      "epoch": 1.1806319408144743,
      "grad_norm": 0.04377797245979309,
      "learning_rate": 4.29866133184421e-06,
      "loss": 0.0012,
      "step": 184080
    },
    {
      "epoch": 1.1806960777082605,
      "grad_norm": 0.07182195037603378,
      "learning_rate": 4.2981071663101135e-06,
      "loss": 0.0014,
      "step": 184090
    },
    {
      "epoch": 1.1807602146020466,
      "grad_norm": 0.47728675603866577,
      "learning_rate": 4.297553009571236e-06,
      "loss": 0.0013,
      "step": 184100
    },
    {
      "epoch": 1.1808243514958328,
      "grad_norm": 0.05756598711013794,
      "learning_rate": 4.2969988616345205e-06,
      "loss": 0.0014,
      "step": 184110
    },
    {
      "epoch": 1.180888488389619,
      "grad_norm": 0.04733746498823166,
      "learning_rate": 4.296444722506911e-06,
      "loss": 0.0017,
      "step": 184120
    },
    {
      "epoch": 1.1809526252834048,
      "grad_norm": 0.09320191293954849,
      "learning_rate": 4.295890592195354e-06,
      "loss": 0.002,
      "step": 184130
    },
    {
      "epoch": 1.181016762177191,
      "grad_norm": 0.10781149566173553,
      "learning_rate": 4.29533647070679e-06,
      "loss": 0.001,
      "step": 184140
    },
    {
      "epoch": 1.1810808990709771,
      "grad_norm": 0.07742494344711304,
      "learning_rate": 4.294782358048164e-06,
      "loss": 0.0007,
      "step": 184150
    },
    {
      "epoch": 1.1811450359647633,
      "grad_norm": 0.11263524740934372,
      "learning_rate": 4.294228254226418e-06,
      "loss": 0.0012,
      "step": 184160
    },
    {
      "epoch": 1.1812091728585492,
      "grad_norm": 0.14303967356681824,
      "learning_rate": 4.293674159248498e-06,
      "loss": 0.0016,
      "step": 184170
    },
    {
      "epoch": 1.1812733097523354,
      "grad_norm": 0.07030445337295532,
      "learning_rate": 4.293120073121345e-06,
      "loss": 0.0015,
      "step": 184180
    },
    {
      "epoch": 1.1813374466461215,
      "grad_norm": 0.06604164838790894,
      "learning_rate": 4.292565995851903e-06,
      "loss": 0.0016,
      "step": 184190
    },
    {
      "epoch": 1.1814015835399077,
      "grad_norm": 0.04574945569038391,
      "learning_rate": 4.292011927447114e-06,
      "loss": 0.0011,
      "step": 184200
    },
    {
      "epoch": 1.1814657204336938,
      "grad_norm": 0.21651262044906616,
      "learning_rate": 4.2914578679139215e-06,
      "loss": 0.0012,
      "step": 184210
    },
    {
      "epoch": 1.1815298573274797,
      "grad_norm": 0.007107165176421404,
      "learning_rate": 4.290903817259269e-06,
      "loss": 0.0009,
      "step": 184220
    },
    {
      "epoch": 1.1815939942212659,
      "grad_norm": 0.08238311111927032,
      "learning_rate": 4.290349775490098e-06,
      "loss": 0.0019,
      "step": 184230
    },
    {
      "epoch": 1.181658131115052,
      "grad_norm": 0.026919526979327202,
      "learning_rate": 4.289795742613351e-06,
      "loss": 0.0009,
      "step": 184240
    },
    {
      "epoch": 1.181722268008838,
      "grad_norm": 0.021025434136390686,
      "learning_rate": 4.289241718635972e-06,
      "loss": 0.001,
      "step": 184250
    },
    {
      "epoch": 1.181786404902624,
      "grad_norm": 0.008072410710155964,
      "learning_rate": 4.288687703564901e-06,
      "loss": 0.0012,
      "step": 184260
    },
    {
      "epoch": 1.1818505417964102,
      "grad_norm": 0.09548212587833405,
      "learning_rate": 4.288133697407082e-06,
      "loss": 0.0014,
      "step": 184270
    },
    {
      "epoch": 1.1819146786901964,
      "grad_norm": 0.1398821920156479,
      "learning_rate": 4.287579700169457e-06,
      "loss": 0.0024,
      "step": 184280
    },
    {
      "epoch": 1.1819788155839825,
      "grad_norm": 0.12649239599704742,
      "learning_rate": 4.287025711858966e-06,
      "loss": 0.0027,
      "step": 184290
    },
    {
      "epoch": 1.1820429524777685,
      "grad_norm": 0.07590962946414948,
      "learning_rate": 4.2864717324825546e-06,
      "loss": 0.0023,
      "step": 184300
    },
    {
      "epoch": 1.1821070893715546,
      "grad_norm": 0.09035839885473251,
      "learning_rate": 4.285917762047161e-06,
      "loss": 0.002,
      "step": 184310
    },
    {
      "epoch": 1.1821712262653408,
      "grad_norm": 0.2527417242527008,
      "learning_rate": 4.285363800559728e-06,
      "loss": 0.0015,
      "step": 184320
    },
    {
      "epoch": 1.182235363159127,
      "grad_norm": 0.07209846377372742,
      "learning_rate": 4.284809848027198e-06,
      "loss": 0.001,
      "step": 184330
    },
    {
      "epoch": 1.1822995000529128,
      "grad_norm": 0.10733459144830704,
      "learning_rate": 4.2842559044565126e-06,
      "loss": 0.0014,
      "step": 184340
    },
    {
      "epoch": 1.182363636946699,
      "grad_norm": 0.021622339263558388,
      "learning_rate": 4.28370196985461e-06,
      "loss": 0.0018,
      "step": 184350
    },
    {
      "epoch": 1.1824277738404851,
      "grad_norm": 0.18197694420814514,
      "learning_rate": 4.283148044228435e-06,
      "loss": 0.001,
      "step": 184360
    },
    {
      "epoch": 1.1824919107342713,
      "grad_norm": 0.07683996856212616,
      "learning_rate": 4.282594127584928e-06,
      "loss": 0.0018,
      "step": 184370
    },
    {
      "epoch": 1.1825560476280574,
      "grad_norm": 0.025498775765299797,
      "learning_rate": 4.282040219931028e-06,
      "loss": 0.0017,
      "step": 184380
    },
    {
      "epoch": 1.1826201845218434,
      "grad_norm": 0.22006182372570038,
      "learning_rate": 4.281486321273678e-06,
      "loss": 0.0016,
      "step": 184390
    },
    {
      "epoch": 1.1826843214156295,
      "grad_norm": 0.19186091423034668,
      "learning_rate": 4.280932431619818e-06,
      "loss": 0.0037,
      "step": 184400
    },
    {
      "epoch": 1.1827484583094157,
      "grad_norm": 0.06715280562639236,
      "learning_rate": 4.280378550976388e-06,
      "loss": 0.0015,
      "step": 184410
    },
    {
      "epoch": 1.1828125952032016,
      "grad_norm": 0.1739199012517929,
      "learning_rate": 4.279824679350328e-06,
      "loss": 0.0025,
      "step": 184420
    },
    {
      "epoch": 1.1828767320969877,
      "grad_norm": 0.13191701471805573,
      "learning_rate": 4.2792708167485805e-06,
      "loss": 0.0022,
      "step": 184430
    },
    {
      "epoch": 1.1829408689907739,
      "grad_norm": 0.01909823529422283,
      "learning_rate": 4.278716963178085e-06,
      "loss": 0.0018,
      "step": 184440
    },
    {
      "epoch": 1.18300500588456,
      "grad_norm": 0.08328601717948914,
      "learning_rate": 4.2781631186457815e-06,
      "loss": 0.0026,
      "step": 184450
    },
    {
      "epoch": 1.1830691427783462,
      "grad_norm": 0.06265491992235184,
      "learning_rate": 4.27760928315861e-06,
      "loss": 0.0018,
      "step": 184460
    },
    {
      "epoch": 1.183133279672132,
      "grad_norm": 0.011295688338577747,
      "learning_rate": 4.2770554567235104e-06,
      "loss": 0.0025,
      "step": 184470
    },
    {
      "epoch": 1.1831974165659183,
      "grad_norm": 0.09248891472816467,
      "learning_rate": 4.276501639347423e-06,
      "loss": 0.0036,
      "step": 184480
    },
    {
      "epoch": 1.1832615534597044,
      "grad_norm": 0.0904218927025795,
      "learning_rate": 4.275947831037287e-06,
      "loss": 0.0016,
      "step": 184490
    },
    {
      "epoch": 1.1833256903534906,
      "grad_norm": 0.02539573423564434,
      "learning_rate": 4.275394031800041e-06,
      "loss": 0.0017,
      "step": 184500
    },
    {
      "epoch": 1.1833898272472765,
      "grad_norm": 0.1429576724767685,
      "learning_rate": 4.274840241642627e-06,
      "loss": 0.0019,
      "step": 184510
    },
    {
      "epoch": 1.1834539641410626,
      "grad_norm": 0.09809023141860962,
      "learning_rate": 4.274286460571981e-06,
      "loss": 0.002,
      "step": 184520
    },
    {
      "epoch": 1.1835181010348488,
      "grad_norm": 0.004953475669026375,
      "learning_rate": 4.273732688595047e-06,
      "loss": 0.0004,
      "step": 184530
    },
    {
      "epoch": 1.183582237928635,
      "grad_norm": 0.02463797852396965,
      "learning_rate": 4.27317892571876e-06,
      "loss": 0.0029,
      "step": 184540
    },
    {
      "epoch": 1.183646374822421,
      "grad_norm": 0.03655168041586876,
      "learning_rate": 4.272625171950061e-06,
      "loss": 0.0006,
      "step": 184550
    },
    {
      "epoch": 1.183710511716207,
      "grad_norm": 0.1086079478263855,
      "learning_rate": 4.272071427295887e-06,
      "loss": 0.0014,
      "step": 184560
    },
    {
      "epoch": 1.1837746486099932,
      "grad_norm": 0.005866074003279209,
      "learning_rate": 4.27151769176318e-06,
      "loss": 0.0014,
      "step": 184570
    },
    {
      "epoch": 1.1838387855037793,
      "grad_norm": 0.050801679491996765,
      "learning_rate": 4.270963965358877e-06,
      "loss": 0.0011,
      "step": 184580
    },
    {
      "epoch": 1.1839029223975654,
      "grad_norm": 0.014840539544820786,
      "learning_rate": 4.270410248089915e-06,
      "loss": 0.0008,
      "step": 184590
    },
    {
      "epoch": 1.1839670592913514,
      "grad_norm": 0.08119440078735352,
      "learning_rate": 4.269856539963236e-06,
      "loss": 0.0017,
      "step": 184600
    },
    {
      "epoch": 1.1840311961851375,
      "grad_norm": 0.6708487272262573,
      "learning_rate": 4.269302840985774e-06,
      "loss": 0.0014,
      "step": 184610
    },
    {
      "epoch": 1.1840953330789237,
      "grad_norm": 0.028421934694051743,
      "learning_rate": 4.26874915116447e-06,
      "loss": 0.0004,
      "step": 184620
    },
    {
      "epoch": 1.1841594699727098,
      "grad_norm": 0.3628992438316345,
      "learning_rate": 4.268195470506263e-06,
      "loss": 0.0013,
      "step": 184630
    },
    {
      "epoch": 1.184223606866496,
      "grad_norm": 0.014118288643658161,
      "learning_rate": 4.267641799018089e-06,
      "loss": 0.0012,
      "step": 184640
    },
    {
      "epoch": 1.184287743760282,
      "grad_norm": 0.014202838763594627,
      "learning_rate": 4.267088136706888e-06,
      "loss": 0.0011,
      "step": 184650
    },
    {
      "epoch": 1.184351880654068,
      "grad_norm": 0.05995582044124603,
      "learning_rate": 4.266534483579595e-06,
      "loss": 0.0038,
      "step": 184660
    },
    {
      "epoch": 1.1844160175478542,
      "grad_norm": 0.07392224669456482,
      "learning_rate": 4.26598083964315e-06,
      "loss": 0.0021,
      "step": 184670
    },
    {
      "epoch": 1.1844801544416401,
      "grad_norm": 0.04062545299530029,
      "learning_rate": 4.26542720490449e-06,
      "loss": 0.0011,
      "step": 184680
    },
    {
      "epoch": 1.1845442913354263,
      "grad_norm": 0.020193586125969887,
      "learning_rate": 4.26487357937055e-06,
      "loss": 0.0009,
      "step": 184690
    },
    {
      "epoch": 1.1846084282292124,
      "grad_norm": 0.13494987785816193,
      "learning_rate": 4.264319963048272e-06,
      "loss": 0.0012,
      "step": 184700
    },
    {
      "epoch": 1.1846725651229986,
      "grad_norm": 0.09614096581935883,
      "learning_rate": 4.263766355944589e-06,
      "loss": 0.001,
      "step": 184710
    },
    {
      "epoch": 1.1847367020167847,
      "grad_norm": 0.10363675653934479,
      "learning_rate": 4.263212758066441e-06,
      "loss": 0.0015,
      "step": 184720
    },
    {
      "epoch": 1.1848008389105706,
      "grad_norm": 0.1137445792555809,
      "learning_rate": 4.262659169420762e-06,
      "loss": 0.0014,
      "step": 184730
    },
    {
      "epoch": 1.1848649758043568,
      "grad_norm": 0.06114194169640541,
      "learning_rate": 4.262105590014492e-06,
      "loss": 0.0008,
      "step": 184740
    },
    {
      "epoch": 1.184929112698143,
      "grad_norm": 0.047655001282691956,
      "learning_rate": 4.261552019854566e-06,
      "loss": 0.0013,
      "step": 184750
    },
    {
      "epoch": 1.184993249591929,
      "grad_norm": 0.01726294681429863,
      "learning_rate": 4.260998458947922e-06,
      "loss": 0.003,
      "step": 184760
    },
    {
      "epoch": 1.185057386485715,
      "grad_norm": 0.031104477122426033,
      "learning_rate": 4.260444907301494e-06,
      "loss": 0.0021,
      "step": 184770
    },
    {
      "epoch": 1.1851215233795012,
      "grad_norm": 0.037550829350948334,
      "learning_rate": 4.259891364922221e-06,
      "loss": 0.0007,
      "step": 184780
    },
    {
      "epoch": 1.1851856602732873,
      "grad_norm": 0.13632871210575104,
      "learning_rate": 4.259337831817037e-06,
      "loss": 0.0016,
      "step": 184790
    },
    {
      "epoch": 1.1852497971670735,
      "grad_norm": 0.017351537942886353,
      "learning_rate": 4.25878430799288e-06,
      "loss": 0.0023,
      "step": 184800
    },
    {
      "epoch": 1.1853139340608596,
      "grad_norm": 0.11563106626272202,
      "learning_rate": 4.258230793456685e-06,
      "loss": 0.0012,
      "step": 184810
    },
    {
      "epoch": 1.1853780709546455,
      "grad_norm": 0.059192679822444916,
      "learning_rate": 4.257677288215388e-06,
      "loss": 0.0024,
      "step": 184820
    },
    {
      "epoch": 1.1854422078484317,
      "grad_norm": 0.09861285239458084,
      "learning_rate": 4.257123792275925e-06,
      "loss": 0.0016,
      "step": 184830
    },
    {
      "epoch": 1.1855063447422178,
      "grad_norm": 0.32034093141555786,
      "learning_rate": 4.256570305645233e-06,
      "loss": 0.0026,
      "step": 184840
    },
    {
      "epoch": 1.185570481636004,
      "grad_norm": 0.19994030892848969,
      "learning_rate": 4.256016828330245e-06,
      "loss": 0.0023,
      "step": 184850
    },
    {
      "epoch": 1.18563461852979,
      "grad_norm": 0.12482646107673645,
      "learning_rate": 4.255463360337898e-06,
      "loss": 0.0027,
      "step": 184860
    },
    {
      "epoch": 1.185698755423576,
      "grad_norm": 0.17621171474456787,
      "learning_rate": 4.254909901675128e-06,
      "loss": 0.0011,
      "step": 184870
    },
    {
      "epoch": 1.1857628923173622,
      "grad_norm": 0.0534793920814991,
      "learning_rate": 4.254356452348869e-06,
      "loss": 0.001,
      "step": 184880
    },
    {
      "epoch": 1.1858270292111484,
      "grad_norm": 0.1162564679980278,
      "learning_rate": 4.253803012366058e-06,
      "loss": 0.0024,
      "step": 184890
    },
    {
      "epoch": 1.1858911661049343,
      "grad_norm": 0.006866424344480038,
      "learning_rate": 4.253249581733626e-06,
      "loss": 0.0014,
      "step": 184900
    },
    {
      "epoch": 1.1859553029987204,
      "grad_norm": 0.0665106326341629,
      "learning_rate": 4.252696160458513e-06,
      "loss": 0.0012,
      "step": 184910
    },
    {
      "epoch": 1.1860194398925066,
      "grad_norm": 0.1312754899263382,
      "learning_rate": 4.252142748547649e-06,
      "loss": 0.0019,
      "step": 184920
    },
    {
      "epoch": 1.1860835767862927,
      "grad_norm": 0.12573768198490143,
      "learning_rate": 4.2515893460079725e-06,
      "loss": 0.0022,
      "step": 184930
    },
    {
      "epoch": 1.1861477136800787,
      "grad_norm": 0.04333149641752243,
      "learning_rate": 4.251035952846415e-06,
      "loss": 0.0012,
      "step": 184940
    },
    {
      "epoch": 1.1862118505738648,
      "grad_norm": 0.02859134040772915,
      "learning_rate": 4.250482569069913e-06,
      "loss": 0.0022,
      "step": 184950
    },
    {
      "epoch": 1.186275987467651,
      "grad_norm": 0.08747965097427368,
      "learning_rate": 4.249929194685401e-06,
      "loss": 0.002,
      "step": 184960
    },
    {
      "epoch": 1.186340124361437,
      "grad_norm": 0.005212867632508278,
      "learning_rate": 4.24937582969981e-06,
      "loss": 0.001,
      "step": 184970
    },
    {
      "epoch": 1.1864042612552232,
      "grad_norm": 0.08888068795204163,
      "learning_rate": 4.248822474120078e-06,
      "loss": 0.0023,
      "step": 184980
    },
    {
      "epoch": 1.1864683981490092,
      "grad_norm": 0.04227092117071152,
      "learning_rate": 4.2482691279531365e-06,
      "loss": 0.0016,
      "step": 184990
    },
    {
      "epoch": 1.1865325350427953,
      "grad_norm": 0.11541248857975006,
      "learning_rate": 4.247715791205921e-06,
      "loss": 0.0016,
      "step": 185000
    },
    {
      "epoch": 1.1865966719365815,
      "grad_norm": 0.27483734488487244,
      "learning_rate": 4.247162463885363e-06,
      "loss": 0.0013,
      "step": 185010
    },
    {
      "epoch": 1.1866608088303676,
      "grad_norm": 0.04213074594736099,
      "learning_rate": 4.246609145998399e-06,
      "loss": 0.0022,
      "step": 185020
    },
    {
      "epoch": 1.1867249457241535,
      "grad_norm": 0.048484332859516144,
      "learning_rate": 4.2460558375519585e-06,
      "loss": 0.0038,
      "step": 185030
    },
    {
      "epoch": 1.1867890826179397,
      "grad_norm": 0.05892828106880188,
      "learning_rate": 4.2455025385529795e-06,
      "loss": 0.0015,
      "step": 185040
    },
    {
      "epoch": 1.1868532195117258,
      "grad_norm": 0.030466919764876366,
      "learning_rate": 4.2449492490083924e-06,
      "loss": 0.0016,
      "step": 185050
    },
    {
      "epoch": 1.186917356405512,
      "grad_norm": 0.07284680008888245,
      "learning_rate": 4.24439596892513e-06,
      "loss": 0.0012,
      "step": 185060
    },
    {
      "epoch": 1.1869814932992981,
      "grad_norm": 0.03833737596869469,
      "learning_rate": 4.243842698310126e-06,
      "loss": 0.001,
      "step": 185070
    },
    {
      "epoch": 1.187045630193084,
      "grad_norm": 0.1180029883980751,
      "learning_rate": 4.243289437170316e-06,
      "loss": 0.0016,
      "step": 185080
    },
    {
      "epoch": 1.1871097670868702,
      "grad_norm": 0.10787667334079742,
      "learning_rate": 4.242736185512627e-06,
      "loss": 0.002,
      "step": 185090
    },
    {
      "epoch": 1.1871739039806564,
      "grad_norm": 0.11979163438081741,
      "learning_rate": 4.2421829433439965e-06,
      "loss": 0.0014,
      "step": 185100
    },
    {
      "epoch": 1.1872380408744423,
      "grad_norm": 0.28249624371528625,
      "learning_rate": 4.241629710671355e-06,
      "loss": 0.0021,
      "step": 185110
    },
    {
      "epoch": 1.1873021777682284,
      "grad_norm": 0.12857112288475037,
      "learning_rate": 4.241076487501634e-06,
      "loss": 0.0019,
      "step": 185120
    },
    {
      "epoch": 1.1873663146620146,
      "grad_norm": 0.031790681183338165,
      "learning_rate": 4.240523273841768e-06,
      "loss": 0.0037,
      "step": 185130
    },
    {
      "epoch": 1.1874304515558007,
      "grad_norm": 0.16708554327487946,
      "learning_rate": 4.239970069698688e-06,
      "loss": 0.0016,
      "step": 185140
    },
    {
      "epoch": 1.1874945884495869,
      "grad_norm": 0.07488280534744263,
      "learning_rate": 4.239416875079327e-06,
      "loss": 0.0012,
      "step": 185150
    },
    {
      "epoch": 1.1875587253433728,
      "grad_norm": 0.005816930904984474,
      "learning_rate": 4.238863689990615e-06,
      "loss": 0.0014,
      "step": 185160
    },
    {
      "epoch": 1.187622862237159,
      "grad_norm": 0.053667064756155014,
      "learning_rate": 4.238310514439487e-06,
      "loss": 0.001,
      "step": 185170
    },
    {
      "epoch": 1.187686999130945,
      "grad_norm": 0.13718998432159424,
      "learning_rate": 4.23775734843287e-06,
      "loss": 0.0021,
      "step": 185180
    },
    {
      "epoch": 1.1877511360247313,
      "grad_norm": 0.06031809374690056,
      "learning_rate": 4.2372041919777e-06,
      "loss": 0.0015,
      "step": 185190
    },
    {
      "epoch": 1.1878152729185172,
      "grad_norm": 0.11881702393293381,
      "learning_rate": 4.236651045080905e-06,
      "loss": 0.0021,
      "step": 185200
    },
    {
      "epoch": 1.1878794098123033,
      "grad_norm": 0.06994739919900894,
      "learning_rate": 4.236097907749419e-06,
      "loss": 0.0016,
      "step": 185210
    },
    {
      "epoch": 1.1879435467060895,
      "grad_norm": 0.13131435215473175,
      "learning_rate": 4.2355447799901725e-06,
      "loss": 0.0009,
      "step": 185220
    },
    {
      "epoch": 1.1880076835998756,
      "grad_norm": 0.019481072202324867,
      "learning_rate": 4.234991661810096e-06,
      "loss": 0.0022,
      "step": 185230
    },
    {
      "epoch": 1.1880718204936618,
      "grad_norm": 0.08921419829130173,
      "learning_rate": 4.2344385532161205e-06,
      "loss": 0.0022,
      "step": 185240
    },
    {
      "epoch": 1.1881359573874477,
      "grad_norm": 0.09565063565969467,
      "learning_rate": 4.233885454215178e-06,
      "loss": 0.0019,
      "step": 185250
    },
    {
      "epoch": 1.1882000942812339,
      "grad_norm": 0.13402371108531952,
      "learning_rate": 4.233332364814197e-06,
      "loss": 0.001,
      "step": 185260
    },
    {
      "epoch": 1.18826423117502,
      "grad_norm": 0.18935437500476837,
      "learning_rate": 4.232779285020109e-06,
      "loss": 0.0011,
      "step": 185270
    },
    {
      "epoch": 1.1883283680688062,
      "grad_norm": 0.00180476950481534,
      "learning_rate": 4.232226214839845e-06,
      "loss": 0.0005,
      "step": 185280
    },
    {
      "epoch": 1.188392504962592,
      "grad_norm": 0.11445319652557373,
      "learning_rate": 4.231673154280336e-06,
      "loss": 0.0024,
      "step": 185290
    },
    {
      "epoch": 1.1884566418563782,
      "grad_norm": 0.07702426612377167,
      "learning_rate": 4.231120103348512e-06,
      "loss": 0.0008,
      "step": 185300
    },
    {
      "epoch": 1.1885207787501644,
      "grad_norm": 0.1344224214553833,
      "learning_rate": 4.230567062051301e-06,
      "loss": 0.0011,
      "step": 185310
    },
    {
      "epoch": 1.1885849156439505,
      "grad_norm": 0.1541663408279419,
      "learning_rate": 4.230014030395637e-06,
      "loss": 0.0015,
      "step": 185320
    },
    {
      "epoch": 1.1886490525377365,
      "grad_norm": 0.06729359924793243,
      "learning_rate": 4.229461008388446e-06,
      "loss": 0.0006,
      "step": 185330
    },
    {
      "epoch": 1.1887131894315226,
      "grad_norm": 0.09541870653629303,
      "learning_rate": 4.22890799603666e-06,
      "loss": 0.0026,
      "step": 185340
    },
    {
      "epoch": 1.1887773263253087,
      "grad_norm": 0.11226841062307358,
      "learning_rate": 4.228354993347207e-06,
      "loss": 0.0019,
      "step": 185350
    },
    {
      "epoch": 1.188841463219095,
      "grad_norm": 0.05591916665434837,
      "learning_rate": 4.227802000327018e-06,
      "loss": 0.0011,
      "step": 185360
    },
    {
      "epoch": 1.1889056001128808,
      "grad_norm": 0.027111517265439034,
      "learning_rate": 4.227249016983021e-06,
      "loss": 0.001,
      "step": 185370
    },
    {
      "epoch": 1.188969737006667,
      "grad_norm": 0.11823323369026184,
      "learning_rate": 4.226696043322145e-06,
      "loss": 0.0014,
      "step": 185380
    },
    {
      "epoch": 1.1890338739004531,
      "grad_norm": 0.06831841170787811,
      "learning_rate": 4.2261430793513215e-06,
      "loss": 0.0017,
      "step": 185390
    },
    {
      "epoch": 1.1890980107942393,
      "grad_norm": 0.17060433328151703,
      "learning_rate": 4.225590125077479e-06,
      "loss": 0.0013,
      "step": 185400
    },
    {
      "epoch": 1.1891621476880254,
      "grad_norm": 0.063255675137043,
      "learning_rate": 4.2250371805075455e-06,
      "loss": 0.0025,
      "step": 185410
    },
    {
      "epoch": 1.1892262845818113,
      "grad_norm": 0.04975636675953865,
      "learning_rate": 4.224484245648451e-06,
      "loss": 0.0016,
      "step": 185420
    },
    {
      "epoch": 1.1892904214755975,
      "grad_norm": 0.12806348502635956,
      "learning_rate": 4.223931320507121e-06,
      "loss": 0.0046,
      "step": 185430
    },
    {
      "epoch": 1.1893545583693836,
      "grad_norm": 0.047647055238485336,
      "learning_rate": 4.223378405090487e-06,
      "loss": 0.0017,
      "step": 185440
    },
    {
      "epoch": 1.1894186952631698,
      "grad_norm": 0.09078560769557953,
      "learning_rate": 4.222825499405477e-06,
      "loss": 0.0015,
      "step": 185450
    },
    {
      "epoch": 1.1894828321569557,
      "grad_norm": 0.05512853339314461,
      "learning_rate": 4.222272603459019e-06,
      "loss": 0.0011,
      "step": 185460
    },
    {
      "epoch": 1.1895469690507419,
      "grad_norm": 0.23027078807353973,
      "learning_rate": 4.22171971725804e-06,
      "loss": 0.0026,
      "step": 185470
    },
    {
      "epoch": 1.189611105944528,
      "grad_norm": 0.10206654667854309,
      "learning_rate": 4.221166840809472e-06,
      "loss": 0.0022,
      "step": 185480
    },
    {
      "epoch": 1.1896752428383142,
      "grad_norm": 0.0034956876188516617,
      "learning_rate": 4.220613974120237e-06,
      "loss": 0.0014,
      "step": 185490
    },
    {
      "epoch": 1.1897393797321003,
      "grad_norm": 0.011971892789006233,
      "learning_rate": 4.220061117197268e-06,
      "loss": 0.0029,
      "step": 185500
    },
    {
      "epoch": 1.1898035166258862,
      "grad_norm": 0.2612984776496887,
      "learning_rate": 4.219508270047489e-06,
      "loss": 0.0018,
      "step": 185510
    },
    {
      "epoch": 1.1898676535196724,
      "grad_norm": 0.045755352824926376,
      "learning_rate": 4.21895543267783e-06,
      "loss": 0.0015,
      "step": 185520
    },
    {
      "epoch": 1.1899317904134585,
      "grad_norm": 0.13760651648044586,
      "learning_rate": 4.218402605095218e-06,
      "loss": 0.001,
      "step": 185530
    },
    {
      "epoch": 1.1899959273072445,
      "grad_norm": 0.03199196606874466,
      "learning_rate": 4.2178497873065784e-06,
      "loss": 0.0014,
      "step": 185540
    },
    {
      "epoch": 1.1900600642010306,
      "grad_norm": 0.11631520092487335,
      "learning_rate": 4.217296979318841e-06,
      "loss": 0.0014,
      "step": 185550
    },
    {
      "epoch": 1.1901242010948168,
      "grad_norm": 0.15202581882476807,
      "learning_rate": 4.216744181138932e-06,
      "loss": 0.0021,
      "step": 185560
    },
    {
      "epoch": 1.190188337988603,
      "grad_norm": 0.15063878893852234,
      "learning_rate": 4.216191392773779e-06,
      "loss": 0.0026,
      "step": 185570
    },
    {
      "epoch": 1.190252474882389,
      "grad_norm": 0.20501667261123657,
      "learning_rate": 4.215638614230306e-06,
      "loss": 0.0011,
      "step": 185580
    },
    {
      "epoch": 1.190316611776175,
      "grad_norm": 0.009964863769710064,
      "learning_rate": 4.215085845515444e-06,
      "loss": 0.0021,
      "step": 185590
    },
    {
      "epoch": 1.1903807486699611,
      "grad_norm": 0.097783662378788,
      "learning_rate": 4.214533086636115e-06,
      "loss": 0.0022,
      "step": 185600
    },
    {
      "epoch": 1.1904448855637473,
      "grad_norm": 0.20840519666671753,
      "learning_rate": 4.21398033759925e-06,
      "loss": 0.0021,
      "step": 185610
    },
    {
      "epoch": 1.1905090224575334,
      "grad_norm": 0.2879226505756378,
      "learning_rate": 4.213427598411771e-06,
      "loss": 0.0024,
      "step": 185620
    },
    {
      "epoch": 1.1905731593513194,
      "grad_norm": 0.00205128057859838,
      "learning_rate": 4.212874869080609e-06,
      "loss": 0.0013,
      "step": 185630
    },
    {
      "epoch": 1.1906372962451055,
      "grad_norm": 0.26543793082237244,
      "learning_rate": 4.2123221496126844e-06,
      "loss": 0.001,
      "step": 185640
    },
    {
      "epoch": 1.1907014331388917,
      "grad_norm": 0.14912432432174683,
      "learning_rate": 4.211769440014929e-06,
      "loss": 0.0012,
      "step": 185650
    },
    {
      "epoch": 1.1907655700326778,
      "grad_norm": 0.15746237337589264,
      "learning_rate": 4.211216740294264e-06,
      "loss": 0.0012,
      "step": 185660
    },
    {
      "epoch": 1.190829706926464,
      "grad_norm": 0.0665852278470993,
      "learning_rate": 4.210664050457618e-06,
      "loss": 0.0011,
      "step": 185670
    },
    {
      "epoch": 1.1908938438202499,
      "grad_norm": 0.12235524505376816,
      "learning_rate": 4.210111370511915e-06,
      "loss": 0.002,
      "step": 185680
    },
    {
      "epoch": 1.190957980714036,
      "grad_norm": 0.06834182143211365,
      "learning_rate": 4.20955870046408e-06,
      "loss": 0.0011,
      "step": 185690
    },
    {
      "epoch": 1.1910221176078222,
      "grad_norm": 0.029557999223470688,
      "learning_rate": 4.2090060403210405e-06,
      "loss": 0.0015,
      "step": 185700
    },
    {
      "epoch": 1.1910862545016083,
      "grad_norm": 0.09428150951862335,
      "learning_rate": 4.20845339008972e-06,
      "loss": 0.0019,
      "step": 185710
    },
    {
      "epoch": 1.1911503913953942,
      "grad_norm": 0.05321458354592323,
      "learning_rate": 4.207900749777046e-06,
      "loss": 0.0009,
      "step": 185720
    },
    {
      "epoch": 1.1912145282891804,
      "grad_norm": 0.10113570839166641,
      "learning_rate": 4.207348119389941e-06,
      "loss": 0.0015,
      "step": 185730
    },
    {
      "epoch": 1.1912786651829665,
      "grad_norm": 0.06574378907680511,
      "learning_rate": 4.206795498935331e-06,
      "loss": 0.0013,
      "step": 185740
    },
    {
      "epoch": 1.1913428020767527,
      "grad_norm": 0.06594528257846832,
      "learning_rate": 4.206242888420139e-06,
      "loss": 0.0009,
      "step": 185750
    },
    {
      "epoch": 1.1914069389705388,
      "grad_norm": 0.03930355980992317,
      "learning_rate": 4.205690287851291e-06,
      "loss": 0.0023,
      "step": 185760
    },
    {
      "epoch": 1.1914710758643248,
      "grad_norm": 0.14651933312416077,
      "learning_rate": 4.205137697235712e-06,
      "loss": 0.0017,
      "step": 185770
    },
    {
      "epoch": 1.191535212758111,
      "grad_norm": 0.02262321673333645,
      "learning_rate": 4.204585116580325e-06,
      "loss": 0.0008,
      "step": 185780
    },
    {
      "epoch": 1.191599349651897,
      "grad_norm": 0.00292450119741261,
      "learning_rate": 4.204032545892056e-06,
      "loss": 0.0014,
      "step": 185790
    },
    {
      "epoch": 1.191663486545683,
      "grad_norm": 0.1417255699634552,
      "learning_rate": 4.203479985177827e-06,
      "loss": 0.0025,
      "step": 185800
    },
    {
      "epoch": 1.1917276234394691,
      "grad_norm": 0.030075931921601295,
      "learning_rate": 4.202927434444564e-06,
      "loss": 0.0018,
      "step": 185810
    },
    {
      "epoch": 1.1917917603332553,
      "grad_norm": 0.10921056568622589,
      "learning_rate": 4.2023748936991885e-06,
      "loss": 0.0015,
      "step": 185820
    },
    {
      "epoch": 1.1918558972270414,
      "grad_norm": 0.1951446533203125,
      "learning_rate": 4.2018223629486275e-06,
      "loss": 0.0018,
      "step": 185830
    },
    {
      "epoch": 1.1919200341208276,
      "grad_norm": 0.1522318720817566,
      "learning_rate": 4.201269842199801e-06,
      "loss": 0.001,
      "step": 185840
    },
    {
      "epoch": 1.1919841710146135,
      "grad_norm": 0.07772860676050186,
      "learning_rate": 4.200717331459636e-06,
      "loss": 0.0012,
      "step": 185850
    },
    {
      "epoch": 1.1920483079083997,
      "grad_norm": 0.13243059813976288,
      "learning_rate": 4.200164830735053e-06,
      "loss": 0.0027,
      "step": 185860
    },
    {
      "epoch": 1.1921124448021858,
      "grad_norm": 0.09397582709789276,
      "learning_rate": 4.1996123400329765e-06,
      "loss": 0.0021,
      "step": 185870
    },
    {
      "epoch": 1.192176581695972,
      "grad_norm": 0.06552623957395554,
      "learning_rate": 4.199059859360329e-06,
      "loss": 0.0018,
      "step": 185880
    },
    {
      "epoch": 1.1922407185897579,
      "grad_norm": 0.015774687752127647,
      "learning_rate": 4.198507388724034e-06,
      "loss": 0.0006,
      "step": 185890
    },
    {
      "epoch": 1.192304855483544,
      "grad_norm": 0.12974444031715393,
      "learning_rate": 4.197954928131015e-06,
      "loss": 0.0031,
      "step": 185900
    },
    {
      "epoch": 1.1923689923773302,
      "grad_norm": 0.08671166747808456,
      "learning_rate": 4.1974024775881935e-06,
      "loss": 0.002,
      "step": 185910
    },
    {
      "epoch": 1.1924331292711163,
      "grad_norm": 0.07681435346603394,
      "learning_rate": 4.196850037102492e-06,
      "loss": 0.0014,
      "step": 185920
    },
    {
      "epoch": 1.1924972661649025,
      "grad_norm": 0.1029948741197586,
      "learning_rate": 4.196297606680835e-06,
      "loss": 0.0027,
      "step": 185930
    },
    {
      "epoch": 1.1925614030586884,
      "grad_norm": 0.12408502399921417,
      "learning_rate": 4.195745186330142e-06,
      "loss": 0.0021,
      "step": 185940
    },
    {
      "epoch": 1.1926255399524746,
      "grad_norm": 0.19858390092849731,
      "learning_rate": 4.1951927760573375e-06,
      "loss": 0.0022,
      "step": 185950
    },
    {
      "epoch": 1.1926896768462607,
      "grad_norm": 0.030382227152585983,
      "learning_rate": 4.194640375869342e-06,
      "loss": 0.0012,
      "step": 185960
    },
    {
      "epoch": 1.1927538137400466,
      "grad_norm": 0.15151441097259521,
      "learning_rate": 4.194087985773078e-06,
      "loss": 0.0025,
      "step": 185970
    },
    {
      "epoch": 1.1928179506338328,
      "grad_norm": 0.08990645408630371,
      "learning_rate": 4.193535605775468e-06,
      "loss": 0.002,
      "step": 185980
    },
    {
      "epoch": 1.192882087527619,
      "grad_norm": 0.1135379821062088,
      "learning_rate": 4.192983235883433e-06,
      "loss": 0.001,
      "step": 185990
    },
    {
      "epoch": 1.192946224421405,
      "grad_norm": 0.07216603308916092,
      "learning_rate": 4.192430876103896e-06,
      "loss": 0.0009,
      "step": 186000
    },
    {
      "epoch": 1.1930103613151912,
      "grad_norm": 0.0714292898774147,
      "learning_rate": 4.191878526443776e-06,
      "loss": 0.0016,
      "step": 186010
    },
    {
      "epoch": 1.1930744982089772,
      "grad_norm": 0.00590480025857687,
      "learning_rate": 4.191326186909996e-06,
      "loss": 0.0013,
      "step": 186020
    },
    {
      "epoch": 1.1931386351027633,
      "grad_norm": 0.0996394157409668,
      "learning_rate": 4.190773857509476e-06,
      "loss": 0.002,
      "step": 186030
    },
    {
      "epoch": 1.1932027719965494,
      "grad_norm": 0.2634541988372803,
      "learning_rate": 4.19022153824914e-06,
      "loss": 0.0042,
      "step": 186040
    },
    {
      "epoch": 1.1932669088903356,
      "grad_norm": 0.23692826926708221,
      "learning_rate": 4.189669229135904e-06,
      "loss": 0.0022,
      "step": 186050
    },
    {
      "epoch": 1.1933310457841215,
      "grad_norm": 0.2381436675786972,
      "learning_rate": 4.189116930176694e-06,
      "loss": 0.0025,
      "step": 186060
    },
    {
      "epoch": 1.1933951826779077,
      "grad_norm": 0.06733310967683792,
      "learning_rate": 4.188564641378427e-06,
      "loss": 0.001,
      "step": 186070
    },
    {
      "epoch": 1.1934593195716938,
      "grad_norm": 0.030237965285778046,
      "learning_rate": 4.188012362748026e-06,
      "loss": 0.0018,
      "step": 186080
    },
    {
      "epoch": 1.19352345646548,
      "grad_norm": 0.29854798316955566,
      "learning_rate": 4.18746009429241e-06,
      "loss": 0.0014,
      "step": 186090
    },
    {
      "epoch": 1.1935875933592661,
      "grad_norm": 0.10096389055252075,
      "learning_rate": 4.1869078360185e-06,
      "loss": 0.0027,
      "step": 186100
    },
    {
      "epoch": 1.193651730253052,
      "grad_norm": 0.06695820391178131,
      "learning_rate": 4.186355587933215e-06,
      "loss": 0.0007,
      "step": 186110
    },
    {
      "epoch": 1.1937158671468382,
      "grad_norm": 0.0895819142460823,
      "learning_rate": 4.185803350043476e-06,
      "loss": 0.0025,
      "step": 186120
    },
    {
      "epoch": 1.1937800040406243,
      "grad_norm": 0.13711774349212646,
      "learning_rate": 4.185251122356204e-06,
      "loss": 0.0014,
      "step": 186130
    },
    {
      "epoch": 1.1938441409344105,
      "grad_norm": 0.31235578656196594,
      "learning_rate": 4.184698904878316e-06,
      "loss": 0.0023,
      "step": 186140
    },
    {
      "epoch": 1.1939082778281964,
      "grad_norm": 0.05649501457810402,
      "learning_rate": 4.184146697616735e-06,
      "loss": 0.0024,
      "step": 186150
    },
    {
      "epoch": 1.1939724147219826,
      "grad_norm": 0.015902062878012657,
      "learning_rate": 4.183594500578379e-06,
      "loss": 0.0011,
      "step": 186160
    },
    {
      "epoch": 1.1940365516157687,
      "grad_norm": 0.06633001565933228,
      "learning_rate": 4.183042313770166e-06,
      "loss": 0.0018,
      "step": 186170
    },
    {
      "epoch": 1.1941006885095549,
      "grad_norm": 0.032952580600976944,
      "learning_rate": 4.1824901371990175e-06,
      "loss": 0.0013,
      "step": 186180
    },
    {
      "epoch": 1.194164825403341,
      "grad_norm": 0.10262813419103622,
      "learning_rate": 4.181937970871852e-06,
      "loss": 0.0016,
      "step": 186190
    },
    {
      "epoch": 1.194228962297127,
      "grad_norm": 0.10096637904644012,
      "learning_rate": 4.181385814795589e-06,
      "loss": 0.0025,
      "step": 186200
    },
    {
      "epoch": 1.194293099190913,
      "grad_norm": 0.03392057120800018,
      "learning_rate": 4.180833668977145e-06,
      "loss": 0.0024,
      "step": 186210
    },
    {
      "epoch": 1.1943572360846992,
      "grad_norm": 0.0377911701798439,
      "learning_rate": 4.180281533423444e-06,
      "loss": 0.0018,
      "step": 186220
    },
    {
      "epoch": 1.1944213729784852,
      "grad_norm": 0.16003689169883728,
      "learning_rate": 4.179729408141398e-06,
      "loss": 0.0019,
      "step": 186230
    },
    {
      "epoch": 1.1944855098722713,
      "grad_norm": 0.08728373795747757,
      "learning_rate": 4.179177293137929e-06,
      "loss": 0.0011,
      "step": 186240
    },
    {
      "epoch": 1.1945496467660575,
      "grad_norm": 0.06210342422127724,
      "learning_rate": 4.178625188419957e-06,
      "loss": 0.0014,
      "step": 186250
    },
    {
      "epoch": 1.1946137836598436,
      "grad_norm": 0.17774981260299683,
      "learning_rate": 4.178073093994397e-06,
      "loss": 0.0014,
      "step": 186260
    },
    {
      "epoch": 1.1946779205536298,
      "grad_norm": 0.013091071508824825,
      "learning_rate": 4.177521009868169e-06,
      "loss": 0.0018,
      "step": 186270
    },
    {
      "epoch": 1.1947420574474157,
      "grad_norm": 0.04876753315329552,
      "learning_rate": 4.17696893604819e-06,
      "loss": 0.0012,
      "step": 186280
    },
    {
      "epoch": 1.1948061943412018,
      "grad_norm": 0.023355981335043907,
      "learning_rate": 4.176416872541379e-06,
      "loss": 0.0009,
      "step": 186290
    },
    {
      "epoch": 1.194870331234988,
      "grad_norm": 0.052153367549180984,
      "learning_rate": 4.175864819354653e-06,
      "loss": 0.0016,
      "step": 186300
    },
    {
      "epoch": 1.1949344681287741,
      "grad_norm": 0.10881591588258743,
      "learning_rate": 4.175312776494931e-06,
      "loss": 0.0037,
      "step": 186310
    },
    {
      "epoch": 1.19499860502256,
      "grad_norm": 0.1673198789358139,
      "learning_rate": 4.1747607439691275e-06,
      "loss": 0.0014,
      "step": 186320
    },
    {
      "epoch": 1.1950627419163462,
      "grad_norm": 0.043181002140045166,
      "learning_rate": 4.174208721784162e-06,
      "loss": 0.0009,
      "step": 186330
    },
    {
      "epoch": 1.1951268788101324,
      "grad_norm": 0.048188578337430954,
      "learning_rate": 4.173656709946952e-06,
      "loss": 0.002,
      "step": 186340
    },
    {
      "epoch": 1.1951910157039185,
      "grad_norm": 0.07603321224451065,
      "learning_rate": 4.173104708464414e-06,
      "loss": 0.0016,
      "step": 186350
    },
    {
      "epoch": 1.1952551525977047,
      "grad_norm": 0.04499555751681328,
      "learning_rate": 4.172552717343464e-06,
      "loss": 0.001,
      "step": 186360
    },
    {
      "epoch": 1.1953192894914906,
      "grad_norm": 0.08442261815071106,
      "learning_rate": 4.172000736591022e-06,
      "loss": 0.0017,
      "step": 186370
    },
    {
      "epoch": 1.1953834263852767,
      "grad_norm": 0.2813013195991516,
      "learning_rate": 4.171448766214e-06,
      "loss": 0.0014,
      "step": 186380
    },
    {
      "epoch": 1.1954475632790629,
      "grad_norm": 0.038745105266571045,
      "learning_rate": 4.170896806219319e-06,
      "loss": 0.0023,
      "step": 186390
    },
    {
      "epoch": 1.195511700172849,
      "grad_norm": 0.006805906537920237,
      "learning_rate": 4.170344856613893e-06,
      "loss": 0.0025,
      "step": 186400
    },
    {
      "epoch": 1.195575837066635,
      "grad_norm": 0.04771881923079491,
      "learning_rate": 4.169792917404638e-06,
      "loss": 0.0012,
      "step": 186410
    },
    {
      "epoch": 1.195639973960421,
      "grad_norm": 0.20832189917564392,
      "learning_rate": 4.169240988598472e-06,
      "loss": 0.0012,
      "step": 186420
    },
    {
      "epoch": 1.1957041108542072,
      "grad_norm": 0.04566992074251175,
      "learning_rate": 4.1686890702023095e-06,
      "loss": 0.0015,
      "step": 186430
    },
    {
      "epoch": 1.1957682477479934,
      "grad_norm": 0.0640728622674942,
      "learning_rate": 4.1681371622230685e-06,
      "loss": 0.003,
      "step": 186440
    },
    {
      "epoch": 1.1958323846417793,
      "grad_norm": 0.09161300212144852,
      "learning_rate": 4.1675852646676615e-06,
      "loss": 0.0019,
      "step": 186450
    },
    {
      "epoch": 1.1958965215355655,
      "grad_norm": 0.03013503924012184,
      "learning_rate": 4.167033377543008e-06,
      "loss": 0.0007,
      "step": 186460
    },
    {
      "epoch": 1.1959606584293516,
      "grad_norm": 0.1995771825313568,
      "learning_rate": 4.166481500856019e-06,
      "loss": 0.0012,
      "step": 186470
    },
    {
      "epoch": 1.1960247953231378,
      "grad_norm": 0.0016916844760999084,
      "learning_rate": 4.165929634613615e-06,
      "loss": 0.0011,
      "step": 186480
    },
    {
      "epoch": 1.1960889322169237,
      "grad_norm": 0.11496210843324661,
      "learning_rate": 4.1653777788227075e-06,
      "loss": 0.003,
      "step": 186490
    },
    {
      "epoch": 1.1961530691107098,
      "grad_norm": 0.06641450524330139,
      "learning_rate": 4.164825933490214e-06,
      "loss": 0.0013,
      "step": 186500
    },
    {
      "epoch": 1.196217206004496,
      "grad_norm": 0.12112980335950851,
      "learning_rate": 4.164274098623047e-06,
      "loss": 0.0019,
      "step": 186510
    },
    {
      "epoch": 1.1962813428982821,
      "grad_norm": 0.08294852077960968,
      "learning_rate": 4.163722274228124e-06,
      "loss": 0.001,
      "step": 186520
    },
    {
      "epoch": 1.1963454797920683,
      "grad_norm": 0.09129433333873749,
      "learning_rate": 4.163170460312358e-06,
      "loss": 0.0024,
      "step": 186530
    },
    {
      "epoch": 1.1964096166858542,
      "grad_norm": 0.159889355301857,
      "learning_rate": 4.162618656882663e-06,
      "loss": 0.0011,
      "step": 186540
    },
    {
      "epoch": 1.1964737535796404,
      "grad_norm": 0.045074403285980225,
      "learning_rate": 4.162066863945957e-06,
      "loss": 0.0009,
      "step": 186550
    },
    {
      "epoch": 1.1965378904734265,
      "grad_norm": 0.092170350253582,
      "learning_rate": 4.161515081509151e-06,
      "loss": 0.0007,
      "step": 186560
    },
    {
      "epoch": 1.1966020273672127,
      "grad_norm": 0.009705310687422752,
      "learning_rate": 4.160963309579161e-06,
      "loss": 0.002,
      "step": 186570
    },
    {
      "epoch": 1.1966661642609986,
      "grad_norm": 0.06332354247570038,
      "learning_rate": 4.160411548162899e-06,
      "loss": 0.0012,
      "step": 186580
    },
    {
      "epoch": 1.1967303011547847,
      "grad_norm": 0.1114995926618576,
      "learning_rate": 4.159859797267282e-06,
      "loss": 0.0011,
      "step": 186590
    },
    {
      "epoch": 1.1967944380485709,
      "grad_norm": 0.04476296901702881,
      "learning_rate": 4.15930805689922e-06,
      "loss": 0.0014,
      "step": 186600
    },
    {
      "epoch": 1.196858574942357,
      "grad_norm": 0.043185412883758545,
      "learning_rate": 4.15875632706563e-06,
      "loss": 0.0017,
      "step": 186610
    },
    {
      "epoch": 1.1969227118361432,
      "grad_norm": 0.09429524093866348,
      "learning_rate": 4.158204607773425e-06,
      "loss": 0.0017,
      "step": 186620
    },
    {
      "epoch": 1.196986848729929,
      "grad_norm": 0.053793590515851974,
      "learning_rate": 4.157652899029518e-06,
      "loss": 0.0017,
      "step": 186630
    },
    {
      "epoch": 1.1970509856237153,
      "grad_norm": 0.1297648549079895,
      "learning_rate": 4.157101200840821e-06,
      "loss": 0.0018,
      "step": 186640
    },
    {
      "epoch": 1.1971151225175014,
      "grad_norm": 0.021457897499203682,
      "learning_rate": 4.15654951321425e-06,
      "loss": 0.0008,
      "step": 186650
    },
    {
      "epoch": 1.1971792594112873,
      "grad_norm": 0.06702035665512085,
      "learning_rate": 4.1559978361567145e-06,
      "loss": 0.001,
      "step": 186660
    },
    {
      "epoch": 1.1972433963050735,
      "grad_norm": 0.08455851674079895,
      "learning_rate": 4.15544616967513e-06,
      "loss": 0.0019,
      "step": 186670
    },
    {
      "epoch": 1.1973075331988596,
      "grad_norm": 0.13208948075771332,
      "learning_rate": 4.15489451377641e-06,
      "loss": 0.0013,
      "step": 186680
    },
    {
      "epoch": 1.1973716700926458,
      "grad_norm": 0.08098907023668289,
      "learning_rate": 4.154342868467463e-06,
      "loss": 0.001,
      "step": 186690
    },
    {
      "epoch": 1.197435806986432,
      "grad_norm": 0.07625220715999603,
      "learning_rate": 4.153791233755207e-06,
      "loss": 0.001,
      "step": 186700
    },
    {
      "epoch": 1.1974999438802179,
      "grad_norm": 0.13422337174415588,
      "learning_rate": 4.15323960964655e-06,
      "loss": 0.0012,
      "step": 186710
    },
    {
      "epoch": 1.197564080774004,
      "grad_norm": 0.07337094098329544,
      "learning_rate": 4.152687996148407e-06,
      "loss": 0.0021,
      "step": 186720
    },
    {
      "epoch": 1.1976282176677902,
      "grad_norm": 0.11599288135766983,
      "learning_rate": 4.152136393267688e-06,
      "loss": 0.0023,
      "step": 186730
    },
    {
      "epoch": 1.1976923545615763,
      "grad_norm": 0.045450177043676376,
      "learning_rate": 4.151584801011306e-06,
      "loss": 0.0024,
      "step": 186740
    },
    {
      "epoch": 1.1977564914553622,
      "grad_norm": 0.13330058753490448,
      "learning_rate": 4.1510332193861725e-06,
      "loss": 0.0014,
      "step": 186750
    },
    {
      "epoch": 1.1978206283491484,
      "grad_norm": 0.09642498195171356,
      "learning_rate": 4.1504816483992005e-06,
      "loss": 0.0017,
      "step": 186760
    },
    {
      "epoch": 1.1978847652429345,
      "grad_norm": 0.05989083647727966,
      "learning_rate": 4.1499300880573e-06,
      "loss": 0.0015,
      "step": 186770
    },
    {
      "epoch": 1.1979489021367207,
      "grad_norm": 0.12496864050626755,
      "learning_rate": 4.149378538367383e-06,
      "loss": 0.001,
      "step": 186780
    },
    {
      "epoch": 1.1980130390305068,
      "grad_norm": 0.04755425453186035,
      "learning_rate": 4.14882699933636e-06,
      "loss": 0.0013,
      "step": 186790
    },
    {
      "epoch": 1.1980771759242927,
      "grad_norm": 0.14057539403438568,
      "learning_rate": 4.148275470971145e-06,
      "loss": 0.0029,
      "step": 186800
    },
    {
      "epoch": 1.198141312818079,
      "grad_norm": 0.14890030026435852,
      "learning_rate": 4.147723953278646e-06,
      "loss": 0.0012,
      "step": 186810
    },
    {
      "epoch": 1.198205449711865,
      "grad_norm": 0.04012436792254448,
      "learning_rate": 4.1471724462657745e-06,
      "loss": 0.0008,
      "step": 186820
    },
    {
      "epoch": 1.1982695866056512,
      "grad_norm": 0.16928061842918396,
      "learning_rate": 4.1466209499394424e-06,
      "loss": 0.0021,
      "step": 186830
    },
    {
      "epoch": 1.1983337234994371,
      "grad_norm": 0.2661837935447693,
      "learning_rate": 4.14606946430656e-06,
      "loss": 0.001,
      "step": 186840
    },
    {
      "epoch": 1.1983978603932233,
      "grad_norm": 0.09210334718227386,
      "learning_rate": 4.145517989374037e-06,
      "loss": 0.0034,
      "step": 186850
    },
    {
      "epoch": 1.1984619972870094,
      "grad_norm": 0.02662188746035099,
      "learning_rate": 4.1449665251487844e-06,
      "loss": 0.0013,
      "step": 186860
    },
    {
      "epoch": 1.1985261341807956,
      "grad_norm": 0.13997481763362885,
      "learning_rate": 4.144415071637713e-06,
      "loss": 0.0028,
      "step": 186870
    },
    {
      "epoch": 1.1985902710745815,
      "grad_norm": 0.0501549169421196,
      "learning_rate": 4.143863628847731e-06,
      "loss": 0.0014,
      "step": 186880
    },
    {
      "epoch": 1.1986544079683676,
      "grad_norm": 0.06980713456869125,
      "learning_rate": 4.1433121967857505e-06,
      "loss": 0.0025,
      "step": 186890
    },
    {
      "epoch": 1.1987185448621538,
      "grad_norm": 0.0924331322312355,
      "learning_rate": 4.1427607754586794e-06,
      "loss": 0.0017,
      "step": 186900
    },
    {
      "epoch": 1.19878268175594,
      "grad_norm": 0.004991814494132996,
      "learning_rate": 4.142209364873428e-06,
      "loss": 0.0019,
      "step": 186910
    },
    {
      "epoch": 1.1988468186497259,
      "grad_norm": 0.18857893347740173,
      "learning_rate": 4.141657965036907e-06,
      "loss": 0.0014,
      "step": 186920
    },
    {
      "epoch": 1.198910955543512,
      "grad_norm": 0.05855089798569679,
      "learning_rate": 4.141106575956026e-06,
      "loss": 0.0019,
      "step": 186930
    },
    {
      "epoch": 1.1989750924372982,
      "grad_norm": 0.18651893734931946,
      "learning_rate": 4.140555197637692e-06,
      "loss": 0.0022,
      "step": 186940
    },
    {
      "epoch": 1.1990392293310843,
      "grad_norm": 0.09940574318170547,
      "learning_rate": 4.140003830088817e-06,
      "loss": 0.001,
      "step": 186950
    },
    {
      "epoch": 1.1991033662248705,
      "grad_norm": 0.0719454437494278,
      "learning_rate": 4.139452473316308e-06,
      "loss": 0.0009,
      "step": 186960
    },
    {
      "epoch": 1.1991675031186564,
      "grad_norm": 0.06848664581775665,
      "learning_rate": 4.138901127327075e-06,
      "loss": 0.0025,
      "step": 186970
    },
    {
      "epoch": 1.1992316400124425,
      "grad_norm": 0.03205366060137749,
      "learning_rate": 4.138349792128026e-06,
      "loss": 0.0022,
      "step": 186980
    },
    {
      "epoch": 1.1992957769062287,
      "grad_norm": 0.06204473599791527,
      "learning_rate": 4.137798467726071e-06,
      "loss": 0.0021,
      "step": 186990
    },
    {
      "epoch": 1.1993599138000148,
      "grad_norm": 0.17390339076519012,
      "learning_rate": 4.137247154128116e-06,
      "loss": 0.001,
      "step": 187000
    },
    {
      "epoch": 1.1994240506938008,
      "grad_norm": 0.14126235246658325,
      "learning_rate": 4.136695851341071e-06,
      "loss": 0.0025,
      "step": 187010
    },
    {
      "epoch": 1.199488187587587,
      "grad_norm": 0.07059001922607422,
      "learning_rate": 4.136144559371844e-06,
      "loss": 0.0015,
      "step": 187020
    },
    {
      "epoch": 1.199552324481373,
      "grad_norm": 0.022132446989417076,
      "learning_rate": 4.135593278227344e-06,
      "loss": 0.0009,
      "step": 187030
    },
    {
      "epoch": 1.1996164613751592,
      "grad_norm": 0.0732007697224617,
      "learning_rate": 4.135042007914477e-06,
      "loss": 0.0011,
      "step": 187040
    },
    {
      "epoch": 1.1996805982689454,
      "grad_norm": 0.007789216935634613,
      "learning_rate": 4.134490748440152e-06,
      "loss": 0.0019,
      "step": 187050
    },
    {
      "epoch": 1.1997447351627313,
      "grad_norm": 0.047100041061639786,
      "learning_rate": 4.133939499811276e-06,
      "loss": 0.0006,
      "step": 187060
    },
    {
      "epoch": 1.1998088720565174,
      "grad_norm": 0.06680265069007874,
      "learning_rate": 4.133388262034758e-06,
      "loss": 0.0017,
      "step": 187070
    },
    {
      "epoch": 1.1998730089503036,
      "grad_norm": 0.04481464624404907,
      "learning_rate": 4.132837035117504e-06,
      "loss": 0.0015,
      "step": 187080
    },
    {
      "epoch": 1.1999371458440895,
      "grad_norm": 0.050326891243457794,
      "learning_rate": 4.1322858190664215e-06,
      "loss": 0.0012,
      "step": 187090
    },
    {
      "epoch": 1.2000012827378757,
      "grad_norm": 0.006268339231610298,
      "learning_rate": 4.131734613888418e-06,
      "loss": 0.0012,
      "step": 187100
    },
    {
      "epoch": 1.2000654196316618,
      "grad_norm": 0.11378975212574005,
      "learning_rate": 4.1311834195904e-06,
      "loss": 0.0013,
      "step": 187110
    },
    {
      "epoch": 1.200129556525448,
      "grad_norm": 0.08344443142414093,
      "learning_rate": 4.130632236179276e-06,
      "loss": 0.0007,
      "step": 187120
    },
    {
      "epoch": 1.200193693419234,
      "grad_norm": 0.10512486845254898,
      "learning_rate": 4.1300810636619496e-06,
      "loss": 0.0008,
      "step": 187130
    },
    {
      "epoch": 1.20025783031302,
      "grad_norm": 0.08657458424568176,
      "learning_rate": 4.129529902045331e-06,
      "loss": 0.0011,
      "step": 187140
    },
    {
      "epoch": 1.2003219672068062,
      "grad_norm": 0.015405516140162945,
      "learning_rate": 4.128978751336322e-06,
      "loss": 0.001,
      "step": 187150
    },
    {
      "epoch": 1.2003861041005923,
      "grad_norm": 0.05375182628631592,
      "learning_rate": 4.128427611541836e-06,
      "loss": 0.0014,
      "step": 187160
    },
    {
      "epoch": 1.2004502409943785,
      "grad_norm": 0.23793500661849976,
      "learning_rate": 4.127876482668771e-06,
      "loss": 0.0033,
      "step": 187170
    },
    {
      "epoch": 1.2005143778881644,
      "grad_norm": 0.08199360966682434,
      "learning_rate": 4.127325364724039e-06,
      "loss": 0.0015,
      "step": 187180
    },
    {
      "epoch": 1.2005785147819505,
      "grad_norm": 0.011276515200734138,
      "learning_rate": 4.126774257714543e-06,
      "loss": 0.0025,
      "step": 187190
    },
    {
      "epoch": 1.2006426516757367,
      "grad_norm": 0.008790786378085613,
      "learning_rate": 4.126223161647191e-06,
      "loss": 0.0014,
      "step": 187200
    },
    {
      "epoch": 1.2007067885695228,
      "grad_norm": 0.15151667594909668,
      "learning_rate": 4.125672076528886e-06,
      "loss": 0.0013,
      "step": 187210
    },
    {
      "epoch": 1.200770925463309,
      "grad_norm": 0.03541888669133186,
      "learning_rate": 4.125121002366535e-06,
      "loss": 0.0012,
      "step": 187220
    },
    {
      "epoch": 1.200835062357095,
      "grad_norm": 0.1084711104631424,
      "learning_rate": 4.124569939167043e-06,
      "loss": 0.0012,
      "step": 187230
    },
    {
      "epoch": 1.200899199250881,
      "grad_norm": 0.04629380628466606,
      "learning_rate": 4.124018886937315e-06,
      "loss": 0.0015,
      "step": 187240
    },
    {
      "epoch": 1.2009633361446672,
      "grad_norm": 0.042566780000925064,
      "learning_rate": 4.1234678456842575e-06,
      "loss": 0.0012,
      "step": 187250
    },
    {
      "epoch": 1.2010274730384534,
      "grad_norm": 0.05979502201080322,
      "learning_rate": 4.122916815414773e-06,
      "loss": 0.001,
      "step": 187260
    },
    {
      "epoch": 1.2010916099322393,
      "grad_norm": 0.02722802944481373,
      "learning_rate": 4.122365796135769e-06,
      "loss": 0.0019,
      "step": 187270
    },
    {
      "epoch": 1.2011557468260254,
      "grad_norm": 0.1804477423429489,
      "learning_rate": 4.1218147878541485e-06,
      "loss": 0.0012,
      "step": 187280
    },
    {
      "epoch": 1.2012198837198116,
      "grad_norm": 0.017981648445129395,
      "learning_rate": 4.121263790576816e-06,
      "loss": 0.0011,
      "step": 187290
    },
    {
      "epoch": 1.2012840206135977,
      "grad_norm": 0.029202884063124657,
      "learning_rate": 4.120712804310676e-06,
      "loss": 0.0016,
      "step": 187300
    },
    {
      "epoch": 1.2013481575073839,
      "grad_norm": 0.03807961195707321,
      "learning_rate": 4.120161829062633e-06,
      "loss": 0.0017,
      "step": 187310
    },
    {
      "epoch": 1.2014122944011698,
      "grad_norm": 0.061058592051267624,
      "learning_rate": 4.119610864839592e-06,
      "loss": 0.0019,
      "step": 187320
    },
    {
      "epoch": 1.201476431294956,
      "grad_norm": 0.05313233286142349,
      "learning_rate": 4.1190599116484554e-06,
      "loss": 0.0014,
      "step": 187330
    },
    {
      "epoch": 1.201540568188742,
      "grad_norm": 0.015431663021445274,
      "learning_rate": 4.118508969496127e-06,
      "loss": 0.0011,
      "step": 187340
    },
    {
      "epoch": 1.201604705082528,
      "grad_norm": 0.023360364139080048,
      "learning_rate": 4.117958038389512e-06,
      "loss": 0.0009,
      "step": 187350
    },
    {
      "epoch": 1.2016688419763142,
      "grad_norm": 0.20973680913448334,
      "learning_rate": 4.117407118335513e-06,
      "loss": 0.002,
      "step": 187360
    },
    {
      "epoch": 1.2017329788701003,
      "grad_norm": 0.0811159685254097,
      "learning_rate": 4.116856209341034e-06,
      "loss": 0.006,
      "step": 187370
    },
    {
      "epoch": 1.2017971157638865,
      "grad_norm": 0.2112804502248764,
      "learning_rate": 4.116305311412978e-06,
      "loss": 0.0018,
      "step": 187380
    },
    {
      "epoch": 1.2018612526576726,
      "grad_norm": 0.09988492727279663,
      "learning_rate": 4.115754424558247e-06,
      "loss": 0.001,
      "step": 187390
    },
    {
      "epoch": 1.2019253895514586,
      "grad_norm": 0.18315206468105316,
      "learning_rate": 4.115203548783746e-06,
      "loss": 0.0026,
      "step": 187400
    },
    {
      "epoch": 1.2019895264452447,
      "grad_norm": 0.07950010895729065,
      "learning_rate": 4.114652684096375e-06,
      "loss": 0.0022,
      "step": 187410
    },
    {
      "epoch": 1.2020536633390309,
      "grad_norm": 0.12662889063358307,
      "learning_rate": 4.114101830503041e-06,
      "loss": 0.0021,
      "step": 187420
    },
    {
      "epoch": 1.202117800232817,
      "grad_norm": 0.15062464773654938,
      "learning_rate": 4.1135509880106425e-06,
      "loss": 0.002,
      "step": 187430
    },
    {
      "epoch": 1.202181937126603,
      "grad_norm": 0.07892768830060959,
      "learning_rate": 4.113000156626086e-06,
      "loss": 0.0025,
      "step": 187440
    },
    {
      "epoch": 1.202246074020389,
      "grad_norm": 0.4865648150444031,
      "learning_rate": 4.112449336356269e-06,
      "loss": 0.0038,
      "step": 187450
    },
    {
      "epoch": 1.2023102109141752,
      "grad_norm": 0.03853175416588783,
      "learning_rate": 4.111898527208097e-06,
      "loss": 0.001,
      "step": 187460
    },
    {
      "epoch": 1.2023743478079614,
      "grad_norm": 0.009135020896792412,
      "learning_rate": 4.111347729188471e-06,
      "loss": 0.0013,
      "step": 187470
    },
    {
      "epoch": 1.2024384847017475,
      "grad_norm": 0.05706874281167984,
      "learning_rate": 4.110796942304294e-06,
      "loss": 0.0019,
      "step": 187480
    },
    {
      "epoch": 1.2025026215955334,
      "grad_norm": 0.11223699152469635,
      "learning_rate": 4.1102461665624655e-06,
      "loss": 0.0015,
      "step": 187490
    },
    {
      "epoch": 1.2025667584893196,
      "grad_norm": 0.17788241803646088,
      "learning_rate": 4.1096954019698895e-06,
      "loss": 0.002,
      "step": 187500
    },
    {
      "epoch": 1.2026308953831057,
      "grad_norm": 0.042884886264801025,
      "learning_rate": 4.109144648533466e-06,
      "loss": 0.0015,
      "step": 187510
    },
    {
      "epoch": 1.2026950322768917,
      "grad_norm": 0.1537342518568039,
      "learning_rate": 4.108593906260096e-06,
      "loss": 0.0019,
      "step": 187520
    },
    {
      "epoch": 1.2027591691706778,
      "grad_norm": 0.12015387415885925,
      "learning_rate": 4.108043175156682e-06,
      "loss": 0.0015,
      "step": 187530
    },
    {
      "epoch": 1.202823306064464,
      "grad_norm": 0.11079380661249161,
      "learning_rate": 4.107492455230124e-06,
      "loss": 0.0013,
      "step": 187540
    },
    {
      "epoch": 1.2028874429582501,
      "grad_norm": 0.04511912539601326,
      "learning_rate": 4.106941746487325e-06,
      "loss": 0.0017,
      "step": 187550
    },
    {
      "epoch": 1.2029515798520363,
      "grad_norm": 0.02514353021979332,
      "learning_rate": 4.106391048935183e-06,
      "loss": 0.0012,
      "step": 187560
    },
    {
      "epoch": 1.2030157167458222,
      "grad_norm": 0.07930656522512436,
      "learning_rate": 4.105840362580601e-06,
      "loss": 0.0017,
      "step": 187570
    },
    {
      "epoch": 1.2030798536396083,
      "grad_norm": 0.3285084366798401,
      "learning_rate": 4.1052896874304756e-06,
      "loss": 0.0006,
      "step": 187580
    },
    {
      "epoch": 1.2031439905333945,
      "grad_norm": 0.058521948754787445,
      "learning_rate": 4.1047390234917124e-06,
      "loss": 0.0017,
      "step": 187590
    },
    {
      "epoch": 1.2032081274271806,
      "grad_norm": 0.15210366249084473,
      "learning_rate": 4.104188370771208e-06,
      "loss": 0.0028,
      "step": 187600
    },
    {
      "epoch": 1.2032722643209666,
      "grad_norm": 0.1036456972360611,
      "learning_rate": 4.103637729275864e-06,
      "loss": 0.0015,
      "step": 187610
    },
    {
      "epoch": 1.2033364012147527,
      "grad_norm": 0.6061391234397888,
      "learning_rate": 4.103087099012579e-06,
      "loss": 0.0023,
      "step": 187620
    },
    {
      "epoch": 1.2034005381085389,
      "grad_norm": 0.31037306785583496,
      "learning_rate": 4.102536479988255e-06,
      "loss": 0.0023,
      "step": 187630
    },
    {
      "epoch": 1.203464675002325,
      "grad_norm": 0.08877530694007874,
      "learning_rate": 4.101985872209789e-06,
      "loss": 0.001,
      "step": 187640
    },
    {
      "epoch": 1.2035288118961112,
      "grad_norm": 0.37374791502952576,
      "learning_rate": 4.101435275684083e-06,
      "loss": 0.0023,
      "step": 187650
    },
    {
      "epoch": 1.203592948789897,
      "grad_norm": 0.04684571921825409,
      "learning_rate": 4.100884690418035e-06,
      "loss": 0.0016,
      "step": 187660
    },
    {
      "epoch": 1.2036570856836832,
      "grad_norm": 0.11537224054336548,
      "learning_rate": 4.100334116418542e-06,
      "loss": 0.0016,
      "step": 187670
    },
    {
      "epoch": 1.2037212225774694,
      "grad_norm": 0.0287192240357399,
      "learning_rate": 4.099783553692507e-06,
      "loss": 0.0013,
      "step": 187680
    },
    {
      "epoch": 1.2037853594712555,
      "grad_norm": 0.0894940122961998,
      "learning_rate": 4.0992330022468285e-06,
      "loss": 0.0012,
      "step": 187690
    },
    {
      "epoch": 1.2038494963650415,
      "grad_norm": 0.056172262877225876,
      "learning_rate": 4.098682462088403e-06,
      "loss": 0.0015,
      "step": 187700
    },
    {
      "epoch": 1.2039136332588276,
      "grad_norm": 0.0038929262664169073,
      "learning_rate": 4.098131933224132e-06,
      "loss": 0.007,
      "step": 187710
    },
    {
      "epoch": 1.2039777701526138,
      "grad_norm": 0.07283658534288406,
      "learning_rate": 4.0975814156609106e-06,
      "loss": 0.0026,
      "step": 187720
    },
    {
      "epoch": 1.2040419070464,
      "grad_norm": 0.07869023084640503,
      "learning_rate": 4.097030909405641e-06,
      "loss": 0.0022,
      "step": 187730
    },
    {
      "epoch": 1.204106043940186,
      "grad_norm": 0.007316358853131533,
      "learning_rate": 4.0964804144652176e-06,
      "loss": 0.0014,
      "step": 187740
    },
    {
      "epoch": 1.204170180833972,
      "grad_norm": 0.07600446045398712,
      "learning_rate": 4.095929930846542e-06,
      "loss": 0.0016,
      "step": 187750
    },
    {
      "epoch": 1.2042343177277581,
      "grad_norm": 0.055925771594047546,
      "learning_rate": 4.095379458556509e-06,
      "loss": 0.0008,
      "step": 187760
    },
    {
      "epoch": 1.2042984546215443,
      "grad_norm": 0.09064169973134995,
      "learning_rate": 4.094828997602019e-06,
      "loss": 0.0016,
      "step": 187770
    },
    {
      "epoch": 1.2043625915153302,
      "grad_norm": 0.04664687439799309,
      "learning_rate": 4.094278547989967e-06,
      "loss": 0.0015,
      "step": 187780
    },
    {
      "epoch": 1.2044267284091164,
      "grad_norm": 0.22472380101680756,
      "learning_rate": 4.0937281097272545e-06,
      "loss": 0.0014,
      "step": 187790
    },
    {
      "epoch": 1.2044908653029025,
      "grad_norm": 0.028136566281318665,
      "learning_rate": 4.093177682820775e-06,
      "loss": 0.0016,
      "step": 187800
    },
    {
      "epoch": 1.2045550021966887,
      "grad_norm": 0.46360525488853455,
      "learning_rate": 4.092627267277427e-06,
      "loss": 0.0014,
      "step": 187810
    },
    {
      "epoch": 1.2046191390904748,
      "grad_norm": 0.11265137046575546,
      "learning_rate": 4.092076863104109e-06,
      "loss": 0.0013,
      "step": 187820
    },
    {
      "epoch": 1.2046832759842607,
      "grad_norm": 0.293995201587677,
      "learning_rate": 4.0915264703077154e-06,
      "loss": 0.0017,
      "step": 187830
    },
    {
      "epoch": 1.2047474128780469,
      "grad_norm": 0.16387131810188293,
      "learning_rate": 4.090976088895145e-06,
      "loss": 0.0025,
      "step": 187840
    },
    {
      "epoch": 1.204811549771833,
      "grad_norm": 0.1470591127872467,
      "learning_rate": 4.090425718873294e-06,
      "loss": 0.0018,
      "step": 187850
    },
    {
      "epoch": 1.2048756866656192,
      "grad_norm": 0.01563205197453499,
      "learning_rate": 4.089875360249059e-06,
      "loss": 0.0019,
      "step": 187860
    },
    {
      "epoch": 1.204939823559405,
      "grad_norm": 0.08468756824731827,
      "learning_rate": 4.089325013029335e-06,
      "loss": 0.0012,
      "step": 187870
    },
    {
      "epoch": 1.2050039604531912,
      "grad_norm": 0.09946174174547195,
      "learning_rate": 4.0887746772210215e-06,
      "loss": 0.0006,
      "step": 187880
    },
    {
      "epoch": 1.2050680973469774,
      "grad_norm": 0.057054974138736725,
      "learning_rate": 4.088224352831011e-06,
      "loss": 0.0031,
      "step": 187890
    },
    {
      "epoch": 1.2051322342407635,
      "grad_norm": 0.2079000622034073,
      "learning_rate": 4.0876740398662015e-06,
      "loss": 0.0021,
      "step": 187900
    },
    {
      "epoch": 1.2051963711345497,
      "grad_norm": 0.05940638482570648,
      "learning_rate": 4.0871237383334885e-06,
      "loss": 0.0021,
      "step": 187910
    },
    {
      "epoch": 1.2052605080283356,
      "grad_norm": 0.028715074062347412,
      "learning_rate": 4.086573448239768e-06,
      "loss": 0.0014,
      "step": 187920
    },
    {
      "epoch": 1.2053246449221218,
      "grad_norm": 0.23639696836471558,
      "learning_rate": 4.086023169591934e-06,
      "loss": 0.0016,
      "step": 187930
    },
    {
      "epoch": 1.205388781815908,
      "grad_norm": 0.07437796145677567,
      "learning_rate": 4.085472902396882e-06,
      "loss": 0.0013,
      "step": 187940
    },
    {
      "epoch": 1.205452918709694,
      "grad_norm": 0.042225006967782974,
      "learning_rate": 4.08492264666151e-06,
      "loss": 0.0008,
      "step": 187950
    },
    {
      "epoch": 1.20551705560348,
      "grad_norm": 0.1663634181022644,
      "learning_rate": 4.08437240239271e-06,
      "loss": 0.0012,
      "step": 187960
    },
    {
      "epoch": 1.2055811924972661,
      "grad_norm": 0.08279787749052048,
      "learning_rate": 4.083822169597379e-06,
      "loss": 0.0011,
      "step": 187970
    },
    {
      "epoch": 1.2056453293910523,
      "grad_norm": 0.07078959792852402,
      "learning_rate": 4.08327194828241e-06,
      "loss": 0.0038,
      "step": 187980
    },
    {
      "epoch": 1.2057094662848384,
      "grad_norm": 0.013743606396019459,
      "learning_rate": 4.082721738454701e-06,
      "loss": 0.0018,
      "step": 187990
    },
    {
      "epoch": 1.2057736031786244,
      "grad_norm": 0.01445829588919878,
      "learning_rate": 4.082171540121141e-06,
      "loss": 0.0017,
      "step": 188000
    },
    {
      "epoch": 1.2058377400724105,
      "grad_norm": 0.08816064149141312,
      "learning_rate": 4.08162135328863e-06,
      "loss": 0.0018,
      "step": 188010
    },
    {
      "epoch": 1.2059018769661967,
      "grad_norm": 0.08746741712093353,
      "learning_rate": 4.081071177964058e-06,
      "loss": 0.0019,
      "step": 188020
    },
    {
      "epoch": 1.2059660138599828,
      "grad_norm": 0.029904184862971306,
      "learning_rate": 4.080521014154323e-06,
      "loss": 0.0022,
      "step": 188030
    },
    {
      "epoch": 1.2060301507537687,
      "grad_norm": 0.07069440931081772,
      "learning_rate": 4.079970861866315e-06,
      "loss": 0.0014,
      "step": 188040
    },
    {
      "epoch": 1.2060942876475549,
      "grad_norm": 0.08658213168382645,
      "learning_rate": 4.079420721106931e-06,
      "loss": 0.0014,
      "step": 188050
    },
    {
      "epoch": 1.206158424541341,
      "grad_norm": 0.05616581067442894,
      "learning_rate": 4.078870591883062e-06,
      "loss": 0.0013,
      "step": 188060
    },
    {
      "epoch": 1.2062225614351272,
      "grad_norm": 0.02350578084588051,
      "learning_rate": 4.078320474201604e-06,
      "loss": 0.0011,
      "step": 188070
    },
    {
      "epoch": 1.2062866983289133,
      "grad_norm": 0.04117431864142418,
      "learning_rate": 4.077770368069448e-06,
      "loss": 0.0013,
      "step": 188080
    },
    {
      "epoch": 1.2063508352226993,
      "grad_norm": 0.04477888345718384,
      "learning_rate": 4.077220273493488e-06,
      "loss": 0.0019,
      "step": 188090
    },
    {
      "epoch": 1.2064149721164854,
      "grad_norm": 0.14388786256313324,
      "learning_rate": 4.076670190480619e-06,
      "loss": 0.0029,
      "step": 188100
    },
    {
      "epoch": 1.2064791090102716,
      "grad_norm": 0.015612471848726273,
      "learning_rate": 4.076120119037731e-06,
      "loss": 0.0016,
      "step": 188110
    },
    {
      "epoch": 1.2065432459040577,
      "grad_norm": 0.03613625839352608,
      "learning_rate": 4.075570059171719e-06,
      "loss": 0.0026,
      "step": 188120
    },
    {
      "epoch": 1.2066073827978436,
      "grad_norm": 0.05011439695954323,
      "learning_rate": 4.075020010889475e-06,
      "loss": 0.0011,
      "step": 188130
    },
    {
      "epoch": 1.2066715196916298,
      "grad_norm": 0.05125297233462334,
      "learning_rate": 4.074469974197892e-06,
      "loss": 0.002,
      "step": 188140
    },
    {
      "epoch": 1.206735656585416,
      "grad_norm": 0.10764499753713608,
      "learning_rate": 4.07391994910386e-06,
      "loss": 0.0017,
      "step": 188150
    },
    {
      "epoch": 1.206799793479202,
      "grad_norm": 0.005238198209553957,
      "learning_rate": 4.073369935614274e-06,
      "loss": 0.0016,
      "step": 188160
    },
    {
      "epoch": 1.2068639303729882,
      "grad_norm": 0.04456604644656181,
      "learning_rate": 4.072819933736025e-06,
      "loss": 0.0014,
      "step": 188170
    },
    {
      "epoch": 1.2069280672667742,
      "grad_norm": 0.14825411140918732,
      "learning_rate": 4.072269943476005e-06,
      "loss": 0.0013,
      "step": 188180
    },
    {
      "epoch": 1.2069922041605603,
      "grad_norm": 0.04572425037622452,
      "learning_rate": 4.071719964841104e-06,
      "loss": 0.0013,
      "step": 188190
    },
    {
      "epoch": 1.2070563410543464,
      "grad_norm": 0.29229500889778137,
      "learning_rate": 4.0711699978382174e-06,
      "loss": 0.0015,
      "step": 188200
    },
    {
      "epoch": 1.2071204779481324,
      "grad_norm": 0.25380653142929077,
      "learning_rate": 4.0706200424742335e-06,
      "loss": 0.0017,
      "step": 188210
    },
    {
      "epoch": 1.2071846148419185,
      "grad_norm": 0.0806683674454689,
      "learning_rate": 4.0700700987560455e-06,
      "loss": 0.0015,
      "step": 188220
    },
    {
      "epoch": 1.2072487517357047,
      "grad_norm": 0.04647250473499298,
      "learning_rate": 4.069520166690543e-06,
      "loss": 0.002,
      "step": 188230
    },
    {
      "epoch": 1.2073128886294908,
      "grad_norm": 0.03942198306322098,
      "learning_rate": 4.068970246284617e-06,
      "loss": 0.0019,
      "step": 188240
    },
    {
      "epoch": 1.207377025523277,
      "grad_norm": 0.31986311078071594,
      "learning_rate": 4.06842033754516e-06,
      "loss": 0.0012,
      "step": 188250
    },
    {
      "epoch": 1.207441162417063,
      "grad_norm": 0.17748479545116425,
      "learning_rate": 4.067870440479062e-06,
      "loss": 0.0021,
      "step": 188260
    },
    {
      "epoch": 1.207505299310849,
      "grad_norm": 0.08459348231554031,
      "learning_rate": 4.067320555093214e-06,
      "loss": 0.0011,
      "step": 188270
    },
    {
      "epoch": 1.2075694362046352,
      "grad_norm": 0.06454211473464966,
      "learning_rate": 4.066770681394506e-06,
      "loss": 0.0009,
      "step": 188280
    },
    {
      "epoch": 1.2076335730984213,
      "grad_norm": 0.11118533462285995,
      "learning_rate": 4.066220819389829e-06,
      "loss": 0.0036,
      "step": 188290
    },
    {
      "epoch": 1.2076977099922073,
      "grad_norm": 0.025810057297348976,
      "learning_rate": 4.06567096908607e-06,
      "loss": 0.0012,
      "step": 188300
    },
    {
      "epoch": 1.2077618468859934,
      "grad_norm": 0.09371408075094223,
      "learning_rate": 4.065121130490124e-06,
      "loss": 0.0012,
      "step": 188310
    },
    {
      "epoch": 1.2078259837797796,
      "grad_norm": 0.018266141414642334,
      "learning_rate": 4.064571303608877e-06,
      "loss": 0.0022,
      "step": 188320
    },
    {
      "epoch": 1.2078901206735657,
      "grad_norm": 0.14298661053180695,
      "learning_rate": 4.064021488449221e-06,
      "loss": 0.0028,
      "step": 188330
    },
    {
      "epoch": 1.2079542575673519,
      "grad_norm": 0.019672038033604622,
      "learning_rate": 4.063471685018043e-06,
      "loss": 0.0007,
      "step": 188340
    },
    {
      "epoch": 1.2080183944611378,
      "grad_norm": 0.0716666504740715,
      "learning_rate": 4.062921893322236e-06,
      "loss": 0.002,
      "step": 188350
    },
    {
      "epoch": 1.208082531354924,
      "grad_norm": 0.10400012135505676,
      "learning_rate": 4.0623721133686866e-06,
      "loss": 0.0016,
      "step": 188360
    },
    {
      "epoch": 1.20814666824871,
      "grad_norm": 0.05531886965036392,
      "learning_rate": 4.061822345164286e-06,
      "loss": 0.0029,
      "step": 188370
    },
    {
      "epoch": 1.2082108051424962,
      "grad_norm": 0.02966497465968132,
      "learning_rate": 4.0612725887159195e-06,
      "loss": 0.0018,
      "step": 188380
    },
    {
      "epoch": 1.2082749420362822,
      "grad_norm": 0.07674681395292282,
      "learning_rate": 4.060722844030479e-06,
      "loss": 0.0014,
      "step": 188390
    },
    {
      "epoch": 1.2083390789300683,
      "grad_norm": 0.030127577483654022,
      "learning_rate": 4.060173111114854e-06,
      "loss": 0.001,
      "step": 188400
    },
    {
      "epoch": 1.2084032158238545,
      "grad_norm": 0.08961062878370285,
      "learning_rate": 4.05962338997593e-06,
      "loss": 0.001,
      "step": 188410
    },
    {
      "epoch": 1.2084673527176406,
      "grad_norm": 0.035098202526569366,
      "learning_rate": 4.059073680620599e-06,
      "loss": 0.0009,
      "step": 188420
    },
    {
      "epoch": 1.2085314896114265,
      "grad_norm": 0.04928535968065262,
      "learning_rate": 4.058523983055745e-06,
      "loss": 0.0028,
      "step": 188430
    },
    {
      "epoch": 1.2085956265052127,
      "grad_norm": 0.021472038701176643,
      "learning_rate": 4.057974297288258e-06,
      "loss": 0.002,
      "step": 188440
    },
    {
      "epoch": 1.2086597633989988,
      "grad_norm": 0.135204017162323,
      "learning_rate": 4.057424623325028e-06,
      "loss": 0.0012,
      "step": 188450
    },
    {
      "epoch": 1.208723900292785,
      "grad_norm": 0.006811929401010275,
      "learning_rate": 4.056874961172941e-06,
      "loss": 0.0027,
      "step": 188460
    },
    {
      "epoch": 1.208788037186571,
      "grad_norm": 0.09208974242210388,
      "learning_rate": 4.056325310838885e-06,
      "loss": 0.0017,
      "step": 188470
    },
    {
      "epoch": 1.208852174080357,
      "grad_norm": 0.161551833152771,
      "learning_rate": 4.055775672329746e-06,
      "loss": 0.0011,
      "step": 188480
    },
    {
      "epoch": 1.2089163109741432,
      "grad_norm": 0.01946430653333664,
      "learning_rate": 4.055226045652414e-06,
      "loss": 0.0015,
      "step": 188490
    },
    {
      "epoch": 1.2089804478679294,
      "grad_norm": 0.08680370450019836,
      "learning_rate": 4.054676430813774e-06,
      "loss": 0.0018,
      "step": 188500
    },
    {
      "epoch": 1.2090445847617155,
      "grad_norm": 0.08462203294038773,
      "learning_rate": 4.054126827820714e-06,
      "loss": 0.0014,
      "step": 188510
    },
    {
      "epoch": 1.2091087216555014,
      "grad_norm": 0.06439220905303955,
      "learning_rate": 4.053577236680122e-06,
      "loss": 0.0011,
      "step": 188520
    },
    {
      "epoch": 1.2091728585492876,
      "grad_norm": 0.07089382410049438,
      "learning_rate": 4.053027657398882e-06,
      "loss": 0.0014,
      "step": 188530
    },
    {
      "epoch": 1.2092369954430737,
      "grad_norm": 0.025199666619300842,
      "learning_rate": 4.052478089983883e-06,
      "loss": 0.0009,
      "step": 188540
    },
    {
      "epoch": 1.2093011323368599,
      "grad_norm": 0.039101798087358475,
      "learning_rate": 4.051928534442011e-06,
      "loss": 0.0014,
      "step": 188550
    },
    {
      "epoch": 1.2093652692306458,
      "grad_norm": 0.10798163712024689,
      "learning_rate": 4.051378990780153e-06,
      "loss": 0.0013,
      "step": 188560
    },
    {
      "epoch": 1.209429406124432,
      "grad_norm": 0.07373690605163574,
      "learning_rate": 4.050829459005192e-06,
      "loss": 0.0013,
      "step": 188570
    },
    {
      "epoch": 1.209493543018218,
      "grad_norm": 0.18303203582763672,
      "learning_rate": 4.050279939124018e-06,
      "loss": 0.0015,
      "step": 188580
    },
    {
      "epoch": 1.2095576799120042,
      "grad_norm": 0.04567175731062889,
      "learning_rate": 4.049730431143514e-06,
      "loss": 0.0015,
      "step": 188590
    },
    {
      "epoch": 1.2096218168057904,
      "grad_norm": 0.07699242234230042,
      "learning_rate": 4.049180935070567e-06,
      "loss": 0.0009,
      "step": 188600
    },
    {
      "epoch": 1.2096859536995763,
      "grad_norm": 0.07643191516399384,
      "learning_rate": 4.048631450912063e-06,
      "loss": 0.002,
      "step": 188610
    },
    {
      "epoch": 1.2097500905933625,
      "grad_norm": 0.05156608670949936,
      "learning_rate": 4.048081978674886e-06,
      "loss": 0.0012,
      "step": 188620
    },
    {
      "epoch": 1.2098142274871486,
      "grad_norm": 0.3705749213695526,
      "learning_rate": 4.047532518365923e-06,
      "loss": 0.0035,
      "step": 188630
    },
    {
      "epoch": 1.2098783643809345,
      "grad_norm": 0.043635543435811996,
      "learning_rate": 4.046983069992058e-06,
      "loss": 0.0019,
      "step": 188640
    },
    {
      "epoch": 1.2099425012747207,
      "grad_norm": 0.0662478655576706,
      "learning_rate": 4.046433633560176e-06,
      "loss": 0.0024,
      "step": 188650
    },
    {
      "epoch": 1.2100066381685068,
      "grad_norm": 0.1132093146443367,
      "learning_rate": 4.045884209077162e-06,
      "loss": 0.0023,
      "step": 188660
    },
    {
      "epoch": 1.210070775062293,
      "grad_norm": 0.022825339809060097,
      "learning_rate": 4.045334796549901e-06,
      "loss": 0.0017,
      "step": 188670
    },
    {
      "epoch": 1.2101349119560791,
      "grad_norm": 0.006263209972530603,
      "learning_rate": 4.044785395985277e-06,
      "loss": 0.0016,
      "step": 188680
    },
    {
      "epoch": 1.210199048849865,
      "grad_norm": 0.052848342806100845,
      "learning_rate": 4.044236007390176e-06,
      "loss": 0.0022,
      "step": 188690
    },
    {
      "epoch": 1.2102631857436512,
      "grad_norm": 0.023211365565657616,
      "learning_rate": 4.043686630771479e-06,
      "loss": 0.0021,
      "step": 188700
    },
    {
      "epoch": 1.2103273226374374,
      "grad_norm": 0.03306528925895691,
      "learning_rate": 4.043137266136074e-06,
      "loss": 0.0024,
      "step": 188710
    },
    {
      "epoch": 1.2103914595312235,
      "grad_norm": 0.14097969233989716,
      "learning_rate": 4.042587913490841e-06,
      "loss": 0.002,
      "step": 188720
    },
    {
      "epoch": 1.2104555964250094,
      "grad_norm": 0.08027029782533646,
      "learning_rate": 4.042038572842667e-06,
      "loss": 0.0032,
      "step": 188730
    },
    {
      "epoch": 1.2105197333187956,
      "grad_norm": 0.014452873729169369,
      "learning_rate": 4.041489244198432e-06,
      "loss": 0.0011,
      "step": 188740
    },
    {
      "epoch": 1.2105838702125817,
      "grad_norm": 0.07585106045007706,
      "learning_rate": 4.040939927565025e-06,
      "loss": 0.0021,
      "step": 188750
    },
    {
      "epoch": 1.2106480071063679,
      "grad_norm": 0.14823010563850403,
      "learning_rate": 4.0403906229493236e-06,
      "loss": 0.0037,
      "step": 188760
    },
    {
      "epoch": 1.210712144000154,
      "grad_norm": 0.058975305408239365,
      "learning_rate": 4.039841330358216e-06,
      "loss": 0.0018,
      "step": 188770
    },
    {
      "epoch": 1.21077628089394,
      "grad_norm": 0.06043016538023949,
      "learning_rate": 4.039292049798581e-06,
      "loss": 0.0014,
      "step": 188780
    },
    {
      "epoch": 1.210840417787726,
      "grad_norm": 0.005526017397642136,
      "learning_rate": 4.0387427812773025e-06,
      "loss": 0.0013,
      "step": 188790
    },
    {
      "epoch": 1.2109045546815123,
      "grad_norm": 0.09409480541944504,
      "learning_rate": 4.0381935248012664e-06,
      "loss": 0.0016,
      "step": 188800
    },
    {
      "epoch": 1.2109686915752984,
      "grad_norm": 0.11467521637678146,
      "learning_rate": 4.0376442803773504e-06,
      "loss": 0.0024,
      "step": 188810
    },
    {
      "epoch": 1.2110328284690843,
      "grad_norm": 0.018768569454550743,
      "learning_rate": 4.037095048012441e-06,
      "loss": 0.0018,
      "step": 188820
    },
    {
      "epoch": 1.2110969653628705,
      "grad_norm": 0.1887967586517334,
      "learning_rate": 4.0365458277134174e-06,
      "loss": 0.0018,
      "step": 188830
    },
    {
      "epoch": 1.2111611022566566,
      "grad_norm": 0.13555163145065308,
      "learning_rate": 4.035996619487165e-06,
      "loss": 0.0023,
      "step": 188840
    },
    {
      "epoch": 1.2112252391504428,
      "grad_norm": 0.0160826463252306,
      "learning_rate": 4.0354474233405624e-06,
      "loss": 0.0039,
      "step": 188850
    },
    {
      "epoch": 1.2112893760442287,
      "grad_norm": 0.053795114159584045,
      "learning_rate": 4.0348982392804934e-06,
      "loss": 0.002,
      "step": 188860
    },
    {
      "epoch": 1.2113535129380149,
      "grad_norm": 0.16556724905967712,
      "learning_rate": 4.034349067313839e-06,
      "loss": 0.0016,
      "step": 188870
    },
    {
      "epoch": 1.211417649831801,
      "grad_norm": 0.08626790344715118,
      "learning_rate": 4.0337999074474824e-06,
      "loss": 0.0012,
      "step": 188880
    },
    {
      "epoch": 1.2114817867255872,
      "grad_norm": 0.2976478040218353,
      "learning_rate": 4.033250759688301e-06,
      "loss": 0.0014,
      "step": 188890
    },
    {
      "epoch": 1.211545923619373,
      "grad_norm": 1.178741216659546,
      "learning_rate": 4.03270162404318e-06,
      "loss": 0.0019,
      "step": 188900
    },
    {
      "epoch": 1.2116100605131592,
      "grad_norm": 0.04616977274417877,
      "learning_rate": 4.0321525005189984e-06,
      "loss": 0.0028,
      "step": 188910
    },
    {
      "epoch": 1.2116741974069454,
      "grad_norm": 0.025343716144561768,
      "learning_rate": 4.031603389122639e-06,
      "loss": 0.0007,
      "step": 188920
    },
    {
      "epoch": 1.2117383343007315,
      "grad_norm": 0.03278636932373047,
      "learning_rate": 4.03105428986098e-06,
      "loss": 0.0019,
      "step": 188930
    },
    {
      "epoch": 1.2118024711945177,
      "grad_norm": 0.1040743812918663,
      "learning_rate": 4.030505202740903e-06,
      "loss": 0.002,
      "step": 188940
    },
    {
      "epoch": 1.2118666080883036,
      "grad_norm": 0.13504858314990997,
      "learning_rate": 4.02995612776929e-06,
      "loss": 0.002,
      "step": 188950
    },
    {
      "epoch": 1.2119307449820897,
      "grad_norm": 0.1743793785572052,
      "learning_rate": 4.029407064953019e-06,
      "loss": 0.0016,
      "step": 188960
    },
    {
      "epoch": 1.211994881875876,
      "grad_norm": 0.0381179116666317,
      "learning_rate": 4.028858014298971e-06,
      "loss": 0.0015,
      "step": 188970
    },
    {
      "epoch": 1.212059018769662,
      "grad_norm": 0.12197365611791611,
      "learning_rate": 4.028308975814027e-06,
      "loss": 0.0014,
      "step": 188980
    },
    {
      "epoch": 1.212123155663448,
      "grad_norm": 0.14338000118732452,
      "learning_rate": 4.0277599495050655e-06,
      "loss": 0.0008,
      "step": 188990
    },
    {
      "epoch": 1.2121872925572341,
      "grad_norm": 0.05253363400697708,
      "learning_rate": 4.0272109353789665e-06,
      "loss": 0.0017,
      "step": 189000
    },
    {
      "epoch": 1.2122514294510203,
      "grad_norm": 0.13448677957057953,
      "learning_rate": 4.02666193344261e-06,
      "loss": 0.0011,
      "step": 189010
    },
    {
      "epoch": 1.2123155663448064,
      "grad_norm": 0.07924114912748337,
      "learning_rate": 4.026112943702874e-06,
      "loss": 0.0019,
      "step": 189020
    },
    {
      "epoch": 1.2123797032385926,
      "grad_norm": 0.16758912801742554,
      "learning_rate": 4.025563966166641e-06,
      "loss": 0.0019,
      "step": 189030
    },
    {
      "epoch": 1.2124438401323785,
      "grad_norm": 0.13628177344799042,
      "learning_rate": 4.025015000840786e-06,
      "loss": 0.0018,
      "step": 189040
    },
    {
      "epoch": 1.2125079770261646,
      "grad_norm": 0.006133451592177153,
      "learning_rate": 4.024466047732192e-06,
      "loss": 0.0009,
      "step": 189050
    },
    {
      "epoch": 1.2125721139199508,
      "grad_norm": 0.06621977686882019,
      "learning_rate": 4.023917106847733e-06,
      "loss": 0.0023,
      "step": 189060
    },
    {
      "epoch": 1.2126362508137367,
      "grad_norm": 0.190654456615448,
      "learning_rate": 4.023368178194292e-06,
      "loss": 0.0013,
      "step": 189070
    },
    {
      "epoch": 1.2127003877075229,
      "grad_norm": 0.04014454036951065,
      "learning_rate": 4.022819261778745e-06,
      "loss": 0.0013,
      "step": 189080
    },
    {
      "epoch": 1.212764524601309,
      "grad_norm": 0.07272602617740631,
      "learning_rate": 4.022270357607971e-06,
      "loss": 0.0015,
      "step": 189090
    },
    {
      "epoch": 1.2128286614950952,
      "grad_norm": 0.012182219885289669,
      "learning_rate": 4.021721465688849e-06,
      "loss": 0.0009,
      "step": 189100
    },
    {
      "epoch": 1.2128927983888813,
      "grad_norm": 0.08933998644351959,
      "learning_rate": 4.021172586028255e-06,
      "loss": 0.0007,
      "step": 189110
    },
    {
      "epoch": 1.2129569352826672,
      "grad_norm": 0.10960163176059723,
      "learning_rate": 4.020623718633069e-06,
      "loss": 0.0016,
      "step": 189120
    },
    {
      "epoch": 1.2130210721764534,
      "grad_norm": 0.04663372039794922,
      "learning_rate": 4.020074863510167e-06,
      "loss": 0.0009,
      "step": 189130
    },
    {
      "epoch": 1.2130852090702395,
      "grad_norm": 0.1042468473315239,
      "learning_rate": 4.019526020666429e-06,
      "loss": 0.0023,
      "step": 189140
    },
    {
      "epoch": 1.2131493459640257,
      "grad_norm": 0.02915082685649395,
      "learning_rate": 4.018977190108729e-06,
      "loss": 0.0014,
      "step": 189150
    },
    {
      "epoch": 1.2132134828578116,
      "grad_norm": 0.06580620259046555,
      "learning_rate": 4.018428371843946e-06,
      "loss": 0.0015,
      "step": 189160
    },
    {
      "epoch": 1.2132776197515978,
      "grad_norm": 0.05101105570793152,
      "learning_rate": 4.017879565878957e-06,
      "loss": 0.0008,
      "step": 189170
    },
    {
      "epoch": 1.213341756645384,
      "grad_norm": 0.15847432613372803,
      "learning_rate": 4.01733077222064e-06,
      "loss": 0.0015,
      "step": 189180
    },
    {
      "epoch": 1.21340589353917,
      "grad_norm": 0.1405104398727417,
      "learning_rate": 4.01678199087587e-06,
      "loss": 0.0022,
      "step": 189190
    },
    {
      "epoch": 1.2134700304329562,
      "grad_norm": 0.045638903975486755,
      "learning_rate": 4.016233221851525e-06,
      "loss": 0.0009,
      "step": 189200
    },
    {
      "epoch": 1.2135341673267421,
      "grad_norm": 0.061501361429691315,
      "learning_rate": 4.015684465154477e-06,
      "loss": 0.0027,
      "step": 189210
    },
    {
      "epoch": 1.2135983042205283,
      "grad_norm": 0.1591787338256836,
      "learning_rate": 4.015135720791611e-06,
      "loss": 0.0012,
      "step": 189220
    },
    {
      "epoch": 1.2136624411143144,
      "grad_norm": 0.18680493533611298,
      "learning_rate": 4.014586988769796e-06,
      "loss": 0.0012,
      "step": 189230
    },
    {
      "epoch": 1.2137265780081006,
      "grad_norm": 0.10476205497980118,
      "learning_rate": 4.014038269095911e-06,
      "loss": 0.0025,
      "step": 189240
    },
    {
      "epoch": 1.2137907149018865,
      "grad_norm": 0.11433443427085876,
      "learning_rate": 4.013489561776831e-06,
      "loss": 0.0022,
      "step": 189250
    },
    {
      "epoch": 1.2138548517956727,
      "grad_norm": 0.1732834428548813,
      "learning_rate": 4.012940866819432e-06,
      "loss": 0.0018,
      "step": 189260
    },
    {
      "epoch": 1.2139189886894588,
      "grad_norm": 0.08467309176921844,
      "learning_rate": 4.012392184230589e-06,
      "loss": 0.0021,
      "step": 189270
    },
    {
      "epoch": 1.213983125583245,
      "grad_norm": 0.06844529509544373,
      "learning_rate": 4.011843514017178e-06,
      "loss": 0.0013,
      "step": 189280
    },
    {
      "epoch": 1.214047262477031,
      "grad_norm": 0.1597221940755844,
      "learning_rate": 4.0112948561860745e-06,
      "loss": 0.0037,
      "step": 189290
    },
    {
      "epoch": 1.214111399370817,
      "grad_norm": 0.10579892247915268,
      "learning_rate": 4.010746210744153e-06,
      "loss": 0.0011,
      "step": 189300
    },
    {
      "epoch": 1.2141755362646032,
      "grad_norm": 0.07977374643087387,
      "learning_rate": 4.010197577698287e-06,
      "loss": 0.0011,
      "step": 189310
    },
    {
      "epoch": 1.2142396731583893,
      "grad_norm": 0.1373594105243683,
      "learning_rate": 4.0096489570553554e-06,
      "loss": 0.0016,
      "step": 189320
    },
    {
      "epoch": 1.2143038100521752,
      "grad_norm": 0.03002183511853218,
      "learning_rate": 4.009100348822228e-06,
      "loss": 0.0035,
      "step": 189330
    },
    {
      "epoch": 1.2143679469459614,
      "grad_norm": 0.1346518099308014,
      "learning_rate": 4.0085517530057825e-06,
      "loss": 0.0012,
      "step": 189340
    },
    {
      "epoch": 1.2144320838397475,
      "grad_norm": 0.07705426216125488,
      "learning_rate": 4.008003169612891e-06,
      "loss": 0.0008,
      "step": 189350
    },
    {
      "epoch": 1.2144962207335337,
      "grad_norm": 0.12079522758722305,
      "learning_rate": 4.007454598650429e-06,
      "loss": 0.0012,
      "step": 189360
    },
    {
      "epoch": 1.2145603576273198,
      "grad_norm": 0.43452203273773193,
      "learning_rate": 4.0069060401252715e-06,
      "loss": 0.0028,
      "step": 189370
    },
    {
      "epoch": 1.2146244945211058,
      "grad_norm": 0.06812382489442825,
      "learning_rate": 4.006357494044289e-06,
      "loss": 0.0022,
      "step": 189380
    },
    {
      "epoch": 1.214688631414892,
      "grad_norm": 0.18466074764728546,
      "learning_rate": 4.005808960414359e-06,
      "loss": 0.0022,
      "step": 189390
    },
    {
      "epoch": 1.214752768308678,
      "grad_norm": 0.08517183363437653,
      "learning_rate": 4.005260439242354e-06,
      "loss": 0.0009,
      "step": 189400
    },
    {
      "epoch": 1.2148169052024642,
      "grad_norm": 0.16007645428180695,
      "learning_rate": 4.004711930535145e-06,
      "loss": 0.001,
      "step": 189410
    },
    {
      "epoch": 1.2148810420962501,
      "grad_norm": 0.04669779911637306,
      "learning_rate": 4.004163434299606e-06,
      "loss": 0.0013,
      "step": 189420
    },
    {
      "epoch": 1.2149451789900363,
      "grad_norm": 0.0766749456524849,
      "learning_rate": 4.003614950542614e-06,
      "loss": 0.0026,
      "step": 189430
    },
    {
      "epoch": 1.2150093158838224,
      "grad_norm": 0.03686497360467911,
      "learning_rate": 4.003066479271035e-06,
      "loss": 0.0038,
      "step": 189440
    },
    {
      "epoch": 1.2150734527776086,
      "grad_norm": 0.05317874625325203,
      "learning_rate": 4.002518020491748e-06,
      "loss": 0.0022,
      "step": 189450
    },
    {
      "epoch": 1.2151375896713947,
      "grad_norm": 0.0613647922873497,
      "learning_rate": 4.001969574211623e-06,
      "loss": 0.0016,
      "step": 189460
    },
    {
      "epoch": 1.2152017265651807,
      "grad_norm": 0.20941519737243652,
      "learning_rate": 4.001421140437533e-06,
      "loss": 0.001,
      "step": 189470
    },
    {
      "epoch": 1.2152658634589668,
      "grad_norm": 0.043185021728277206,
      "learning_rate": 4.000872719176347e-06,
      "loss": 0.0012,
      "step": 189480
    },
    {
      "epoch": 1.215330000352753,
      "grad_norm": 0.03593922033905983,
      "learning_rate": 4.000324310434943e-06,
      "loss": 0.0018,
      "step": 189490
    },
    {
      "epoch": 1.215394137246539,
      "grad_norm": 0.14687849581241608,
      "learning_rate": 3.999775914220188e-06,
      "loss": 0.0013,
      "step": 189500
    },
    {
      "epoch": 1.215458274140325,
      "grad_norm": 0.043099600821733475,
      "learning_rate": 3.9992275305389555e-06,
      "loss": 0.0022,
      "step": 189510
    },
    {
      "epoch": 1.2155224110341112,
      "grad_norm": 0.10225958377122879,
      "learning_rate": 3.998679159398118e-06,
      "loss": 0.0019,
      "step": 189520
    },
    {
      "epoch": 1.2155865479278973,
      "grad_norm": 0.051480717957019806,
      "learning_rate": 3.9981308008045464e-06,
      "loss": 0.001,
      "step": 189530
    },
    {
      "epoch": 1.2156506848216835,
      "grad_norm": 0.11180371046066284,
      "learning_rate": 3.997582454765112e-06,
      "loss": 0.0018,
      "step": 189540
    },
    {
      "epoch": 1.2157148217154694,
      "grad_norm": 0.07143254578113556,
      "learning_rate": 3.997034121286685e-06,
      "loss": 0.0014,
      "step": 189550
    },
    {
      "epoch": 1.2157789586092556,
      "grad_norm": 0.08327856659889221,
      "learning_rate": 3.9964858003761385e-06,
      "loss": 0.0014,
      "step": 189560
    },
    {
      "epoch": 1.2158430955030417,
      "grad_norm": 0.06372194737195969,
      "learning_rate": 3.995937492040341e-06,
      "loss": 0.0011,
      "step": 189570
    },
    {
      "epoch": 1.2159072323968279,
      "grad_norm": 0.29604244232177734,
      "learning_rate": 3.995389196286165e-06,
      "loss": 0.003,
      "step": 189580
    },
    {
      "epoch": 1.2159713692906138,
      "grad_norm": 0.11042928695678711,
      "learning_rate": 3.99484091312048e-06,
      "loss": 0.0022,
      "step": 189590
    },
    {
      "epoch": 1.2160355061844,
      "grad_norm": 0.070286825299263,
      "learning_rate": 3.9942926425501574e-06,
      "loss": 0.0019,
      "step": 189600
    },
    {
      "epoch": 1.216099643078186,
      "grad_norm": 0.016711633652448654,
      "learning_rate": 3.993744384582065e-06,
      "loss": 0.0029,
      "step": 189610
    },
    {
      "epoch": 1.2161637799719722,
      "grad_norm": 0.030171751976013184,
      "learning_rate": 3.993196139223077e-06,
      "loss": 0.0022,
      "step": 189620
    },
    {
      "epoch": 1.2162279168657584,
      "grad_norm": 0.1727365404367447,
      "learning_rate": 3.99264790648006e-06,
      "loss": 0.0015,
      "step": 189630
    },
    {
      "epoch": 1.2162920537595443,
      "grad_norm": 0.00967460311949253,
      "learning_rate": 3.992099686359883e-06,
      "loss": 0.0017,
      "step": 189640
    },
    {
      "epoch": 1.2163561906533304,
      "grad_norm": 0.1014925017952919,
      "learning_rate": 3.99155147886942e-06,
      "loss": 0.0019,
      "step": 189650
    },
    {
      "epoch": 1.2164203275471166,
      "grad_norm": 0.20213359594345093,
      "learning_rate": 3.9910032840155355e-06,
      "loss": 0.0013,
      "step": 189660
    },
    {
      "epoch": 1.2164844644409027,
      "grad_norm": 0.045818161219358444,
      "learning_rate": 3.990455101805102e-06,
      "loss": 0.0018,
      "step": 189670
    },
    {
      "epoch": 1.2165486013346887,
      "grad_norm": 0.06107959896326065,
      "learning_rate": 3.989906932244987e-06,
      "loss": 0.0013,
      "step": 189680
    },
    {
      "epoch": 1.2166127382284748,
      "grad_norm": 0.10544388741254807,
      "learning_rate": 3.989358775342062e-06,
      "loss": 0.0012,
      "step": 189690
    },
    {
      "epoch": 1.216676875122261,
      "grad_norm": 0.05415049567818642,
      "learning_rate": 3.9888106311031914e-06,
      "loss": 0.0021,
      "step": 189700
    },
    {
      "epoch": 1.2167410120160471,
      "grad_norm": 0.2054257094860077,
      "learning_rate": 3.988262499535248e-06,
      "loss": 0.0014,
      "step": 189710
    },
    {
      "epoch": 1.2168051489098333,
      "grad_norm": 0.1231830045580864,
      "learning_rate": 3.987714380645097e-06,
      "loss": 0.0018,
      "step": 189720
    },
    {
      "epoch": 1.2168692858036192,
      "grad_norm": 0.025744184851646423,
      "learning_rate": 3.987166274439609e-06,
      "loss": 0.0007,
      "step": 189730
    },
    {
      "epoch": 1.2169334226974053,
      "grad_norm": 0.0822049155831337,
      "learning_rate": 3.9866181809256515e-06,
      "loss": 0.0014,
      "step": 189740
    },
    {
      "epoch": 1.2169975595911915,
      "grad_norm": 0.014606112614274025,
      "learning_rate": 3.986070100110093e-06,
      "loss": 0.0018,
      "step": 189750
    },
    {
      "epoch": 1.2170616964849774,
      "grad_norm": 0.08604684472084045,
      "learning_rate": 3.9855220319998e-06,
      "loss": 0.0026,
      "step": 189760
    },
    {
      "epoch": 1.2171258333787636,
      "grad_norm": 0.0861067995429039,
      "learning_rate": 3.984973976601641e-06,
      "loss": 0.0013,
      "step": 189770
    },
    {
      "epoch": 1.2171899702725497,
      "grad_norm": 0.29283806681632996,
      "learning_rate": 3.9844259339224835e-06,
      "loss": 0.0016,
      "step": 189780
    },
    {
      "epoch": 1.2172541071663359,
      "grad_norm": 0.10819019377231598,
      "learning_rate": 3.983877903969195e-06,
      "loss": 0.0013,
      "step": 189790
    },
    {
      "epoch": 1.217318244060122,
      "grad_norm": 0.062073834240436554,
      "learning_rate": 3.983329886748643e-06,
      "loss": 0.0009,
      "step": 189800
    },
    {
      "epoch": 1.217382380953908,
      "grad_norm": 0.09325316548347473,
      "learning_rate": 3.982781882267693e-06,
      "loss": 0.0019,
      "step": 189810
    },
    {
      "epoch": 1.217446517847694,
      "grad_norm": 0.14637985825538635,
      "learning_rate": 3.9822338905332145e-06,
      "loss": 0.0023,
      "step": 189820
    },
    {
      "epoch": 1.2175106547414802,
      "grad_norm": 0.05455147847533226,
      "learning_rate": 3.981685911552072e-06,
      "loss": 0.0009,
      "step": 189830
    },
    {
      "epoch": 1.2175747916352664,
      "grad_norm": 0.040302470326423645,
      "learning_rate": 3.981137945331133e-06,
      "loss": 0.0016,
      "step": 189840
    },
    {
      "epoch": 1.2176389285290523,
      "grad_norm": 0.1425817310810089,
      "learning_rate": 3.980589991877263e-06,
      "loss": 0.0019,
      "step": 189850
    },
    {
      "epoch": 1.2177030654228385,
      "grad_norm": 0.1364559829235077,
      "learning_rate": 3.98004205119733e-06,
      "loss": 0.0013,
      "step": 189860
    },
    {
      "epoch": 1.2177672023166246,
      "grad_norm": 0.1536790430545807,
      "learning_rate": 3.9794941232981985e-06,
      "loss": 0.0014,
      "step": 189870
    },
    {
      "epoch": 1.2178313392104108,
      "grad_norm": 0.09738431870937347,
      "learning_rate": 3.978946208186736e-06,
      "loss": 0.0017,
      "step": 189880
    },
    {
      "epoch": 1.217895476104197,
      "grad_norm": 0.06185058876872063,
      "learning_rate": 3.978398305869805e-06,
      "loss": 0.0011,
      "step": 189890
    },
    {
      "epoch": 1.2179596129979828,
      "grad_norm": 0.20613211393356323,
      "learning_rate": 3.977850416354275e-06,
      "loss": 0.001,
      "step": 189900
    },
    {
      "epoch": 1.218023749891769,
      "grad_norm": 0.16253237426280975,
      "learning_rate": 3.9773025396470095e-06,
      "loss": 0.0014,
      "step": 189910
    },
    {
      "epoch": 1.2180878867855551,
      "grad_norm": 0.11593171209096909,
      "learning_rate": 3.976754675754875e-06,
      "loss": 0.0017,
      "step": 189920
    },
    {
      "epoch": 1.2181520236793413,
      "grad_norm": 0.16497549414634705,
      "learning_rate": 3.976206824684734e-06,
      "loss": 0.001,
      "step": 189930
    },
    {
      "epoch": 1.2182161605731272,
      "grad_norm": 0.14754164218902588,
      "learning_rate": 3.975658986443454e-06,
      "loss": 0.0029,
      "step": 189940
    },
    {
      "epoch": 1.2182802974669134,
      "grad_norm": 0.09771906584501266,
      "learning_rate": 3.975111161037899e-06,
      "loss": 0.0021,
      "step": 189950
    },
    {
      "epoch": 1.2183444343606995,
      "grad_norm": 0.08924322575330734,
      "learning_rate": 3.974563348474933e-06,
      "loss": 0.0017,
      "step": 189960
    },
    {
      "epoch": 1.2184085712544857,
      "grad_norm": 0.021901067346334457,
      "learning_rate": 3.9740155487614196e-06,
      "loss": 0.0011,
      "step": 189970
    },
    {
      "epoch": 1.2184727081482716,
      "grad_norm": 0.07129445672035217,
      "learning_rate": 3.9734677619042276e-06,
      "loss": 0.0022,
      "step": 189980
    },
    {
      "epoch": 1.2185368450420577,
      "grad_norm": 0.13516168296337128,
      "learning_rate": 3.972919987910217e-06,
      "loss": 0.0028,
      "step": 189990
    },
    {
      "epoch": 1.2186009819358439,
      "grad_norm": 0.15106359124183655,
      "learning_rate": 3.972372226786253e-06,
      "loss": 0.0011,
      "step": 190000
    },
    {
      "epoch": 1.21866511882963,
      "grad_norm": 0.1575210988521576,
      "learning_rate": 3.9718244785392e-06,
      "loss": 0.0014,
      "step": 190010
    },
    {
      "epoch": 1.218729255723416,
      "grad_norm": 0.03287835791707039,
      "learning_rate": 3.9712767431759216e-06,
      "loss": 0.0013,
      "step": 190020
    },
    {
      "epoch": 1.218793392617202,
      "grad_norm": 0.16302621364593506,
      "learning_rate": 3.9707290207032785e-06,
      "loss": 0.0017,
      "step": 190030
    },
    {
      "epoch": 1.2188575295109882,
      "grad_norm": 0.03399442136287689,
      "learning_rate": 3.970181311128139e-06,
      "loss": 0.0008,
      "step": 190040
    },
    {
      "epoch": 1.2189216664047744,
      "grad_norm": 0.010786185972392559,
      "learning_rate": 3.9696336144573625e-06,
      "loss": 0.0011,
      "step": 190050
    },
    {
      "epoch": 1.2189858032985605,
      "grad_norm": 0.044280312955379486,
      "learning_rate": 3.969085930697812e-06,
      "loss": 0.001,
      "step": 190060
    },
    {
      "epoch": 1.2190499401923465,
      "grad_norm": 0.009728172793984413,
      "learning_rate": 3.968538259856354e-06,
      "loss": 0.001,
      "step": 190070
    },
    {
      "epoch": 1.2191140770861326,
      "grad_norm": 0.029003439471125603,
      "learning_rate": 3.9679906019398475e-06,
      "loss": 0.0021,
      "step": 190080
    },
    {
      "epoch": 1.2191782139799188,
      "grad_norm": 0.07220987975597382,
      "learning_rate": 3.967442956955157e-06,
      "loss": 0.0012,
      "step": 190090
    },
    {
      "epoch": 1.219242350873705,
      "grad_norm": 0.2752041816711426,
      "learning_rate": 3.966895324909143e-06,
      "loss": 0.0076,
      "step": 190100
    },
    {
      "epoch": 1.2193064877674908,
      "grad_norm": 0.024868693202733994,
      "learning_rate": 3.96634770580867e-06,
      "loss": 0.0018,
      "step": 190110
    },
    {
      "epoch": 1.219370624661277,
      "grad_norm": 0.07499602437019348,
      "learning_rate": 3.965800099660598e-06,
      "loss": 0.0016,
      "step": 190120
    },
    {
      "epoch": 1.2194347615550631,
      "grad_norm": 0.2519768476486206,
      "learning_rate": 3.965252506471791e-06,
      "loss": 0.0018,
      "step": 190130
    },
    {
      "epoch": 1.2194988984488493,
      "grad_norm": 0.02175537869334221,
      "learning_rate": 3.964704926249109e-06,
      "loss": 0.0006,
      "step": 190140
    },
    {
      "epoch": 1.2195630353426354,
      "grad_norm": 0.04755594581365585,
      "learning_rate": 3.964157358999415e-06,
      "loss": 0.0009,
      "step": 190150
    },
    {
      "epoch": 1.2196271722364214,
      "grad_norm": 0.09518511593341827,
      "learning_rate": 3.963609804729568e-06,
      "loss": 0.0012,
      "step": 190160
    },
    {
      "epoch": 1.2196913091302075,
      "grad_norm": 0.12631119787693024,
      "learning_rate": 3.963062263446432e-06,
      "loss": 0.0015,
      "step": 190170
    },
    {
      "epoch": 1.2197554460239937,
      "grad_norm": 0.08314120769500732,
      "learning_rate": 3.962514735156867e-06,
      "loss": 0.0026,
      "step": 190180
    },
    {
      "epoch": 1.2198195829177796,
      "grad_norm": 0.1589404046535492,
      "learning_rate": 3.961967219867734e-06,
      "loss": 0.003,
      "step": 190190
    },
    {
      "epoch": 1.2198837198115657,
      "grad_norm": 0.04208589717745781,
      "learning_rate": 3.961419717585892e-06,
      "loss": 0.002,
      "step": 190200
    },
    {
      "epoch": 1.2199478567053519,
      "grad_norm": 0.09710143506526947,
      "learning_rate": 3.960872228318204e-06,
      "loss": 0.0014,
      "step": 190210
    },
    {
      "epoch": 1.220011993599138,
      "grad_norm": 0.06639356911182404,
      "learning_rate": 3.96032475207153e-06,
      "loss": 0.0015,
      "step": 190220
    },
    {
      "epoch": 1.2200761304929242,
      "grad_norm": 0.01445319876074791,
      "learning_rate": 3.95977728885273e-06,
      "loss": 0.0011,
      "step": 190230
    },
    {
      "epoch": 1.22014026738671,
      "grad_norm": 0.07329652458429337,
      "learning_rate": 3.959229838668665e-06,
      "loss": 0.0022,
      "step": 190240
    },
    {
      "epoch": 1.2202044042804963,
      "grad_norm": 0.03763607144355774,
      "learning_rate": 3.958682401526191e-06,
      "loss": 0.0031,
      "step": 190250
    },
    {
      "epoch": 1.2202685411742824,
      "grad_norm": 0.10107597708702087,
      "learning_rate": 3.9581349774321736e-06,
      "loss": 0.0012,
      "step": 190260
    },
    {
      "epoch": 1.2203326780680686,
      "grad_norm": 0.07468398660421371,
      "learning_rate": 3.957587566393468e-06,
      "loss": 0.0015,
      "step": 190270
    },
    {
      "epoch": 1.2203968149618545,
      "grad_norm": 0.1615559607744217,
      "learning_rate": 3.957040168416936e-06,
      "loss": 0.0012,
      "step": 190280
    },
    {
      "epoch": 1.2204609518556406,
      "grad_norm": 0.03799016401171684,
      "learning_rate": 3.956492783509436e-06,
      "loss": 0.002,
      "step": 190290
    },
    {
      "epoch": 1.2205250887494268,
      "grad_norm": 0.06666628271341324,
      "learning_rate": 3.955945411677827e-06,
      "loss": 0.0011,
      "step": 190300
    },
    {
      "epoch": 1.220589225643213,
      "grad_norm": 0.06822290271520615,
      "learning_rate": 3.955398052928968e-06,
      "loss": 0.0012,
      "step": 190310
    },
    {
      "epoch": 1.220653362536999,
      "grad_norm": 0.030858352780342102,
      "learning_rate": 3.954850707269718e-06,
      "loss": 0.0022,
      "step": 190320
    },
    {
      "epoch": 1.220717499430785,
      "grad_norm": 0.09506650269031525,
      "learning_rate": 3.9543033747069355e-06,
      "loss": 0.0009,
      "step": 190330
    },
    {
      "epoch": 1.2207816363245712,
      "grad_norm": 0.014402428641915321,
      "learning_rate": 3.95375605524748e-06,
      "loss": 0.0014,
      "step": 190340
    },
    {
      "epoch": 1.2208457732183573,
      "grad_norm": 0.11958342790603638,
      "learning_rate": 3.953208748898208e-06,
      "loss": 0.0012,
      "step": 190350
    },
    {
      "epoch": 1.2209099101121434,
      "grad_norm": 0.07521621882915497,
      "learning_rate": 3.952661455665978e-06,
      "loss": 0.0016,
      "step": 190360
    },
    {
      "epoch": 1.2209740470059294,
      "grad_norm": 0.11715155839920044,
      "learning_rate": 3.952114175557649e-06,
      "loss": 0.0051,
      "step": 190370
    },
    {
      "epoch": 1.2210381838997155,
      "grad_norm": 0.0628410205245018,
      "learning_rate": 3.951566908580078e-06,
      "loss": 0.0013,
      "step": 190380
    },
    {
      "epoch": 1.2211023207935017,
      "grad_norm": 0.01683405227959156,
      "learning_rate": 3.951019654740124e-06,
      "loss": 0.0009,
      "step": 190390
    },
    {
      "epoch": 1.2211664576872878,
      "grad_norm": 0.17968235909938812,
      "learning_rate": 3.950472414044642e-06,
      "loss": 0.0014,
      "step": 190400
    },
    {
      "epoch": 1.2212305945810737,
      "grad_norm": 0.05620647221803665,
      "learning_rate": 3.949925186500492e-06,
      "loss": 0.0024,
      "step": 190410
    },
    {
      "epoch": 1.22129473147486,
      "grad_norm": 0.12351801246404648,
      "learning_rate": 3.949377972114529e-06,
      "loss": 0.0014,
      "step": 190420
    },
    {
      "epoch": 1.221358868368646,
      "grad_norm": 0.1264180839061737,
      "learning_rate": 3.948830770893612e-06,
      "loss": 0.0019,
      "step": 190430
    },
    {
      "epoch": 1.2214230052624322,
      "grad_norm": 0.11731204390525818,
      "learning_rate": 3.948283582844595e-06,
      "loss": 0.0027,
      "step": 190440
    },
    {
      "epoch": 1.2214871421562181,
      "grad_norm": 0.06938435882329941,
      "learning_rate": 3.9477364079743386e-06,
      "loss": 0.0011,
      "step": 190450
    },
    {
      "epoch": 1.2215512790500043,
      "grad_norm": 0.06576994806528091,
      "learning_rate": 3.947189246289695e-06,
      "loss": 0.0016,
      "step": 190460
    },
    {
      "epoch": 1.2216154159437904,
      "grad_norm": 0.09067405760288239,
      "learning_rate": 3.946642097797524e-06,
      "loss": 0.0016,
      "step": 190470
    },
    {
      "epoch": 1.2216795528375766,
      "grad_norm": 0.0050344220362603664,
      "learning_rate": 3.946094962504679e-06,
      "loss": 0.0013,
      "step": 190480
    },
    {
      "epoch": 1.2217436897313627,
      "grad_norm": 0.05471819266676903,
      "learning_rate": 3.945547840418017e-06,
      "loss": 0.0011,
      "step": 190490
    },
    {
      "epoch": 1.2218078266251486,
      "grad_norm": 0.12141404300928116,
      "learning_rate": 3.945000731544396e-06,
      "loss": 0.0047,
      "step": 190500
    },
    {
      "epoch": 1.2218719635189348,
      "grad_norm": 0.011203257367014885,
      "learning_rate": 3.944453635890668e-06,
      "loss": 0.0017,
      "step": 190510
    },
    {
      "epoch": 1.221936100412721,
      "grad_norm": 0.15600796043872833,
      "learning_rate": 3.943906553463693e-06,
      "loss": 0.0044,
      "step": 190520
    },
    {
      "epoch": 1.222000237306507,
      "grad_norm": 0.060006964951753616,
      "learning_rate": 3.943359484270321e-06,
      "loss": 0.0025,
      "step": 190530
    },
    {
      "epoch": 1.222064374200293,
      "grad_norm": 0.06969252228736877,
      "learning_rate": 3.9428124283174116e-06,
      "loss": 0.0012,
      "step": 190540
    },
    {
      "epoch": 1.2221285110940792,
      "grad_norm": 0.1027282178401947,
      "learning_rate": 3.942265385611817e-06,
      "loss": 0.0013,
      "step": 190550
    },
    {
      "epoch": 1.2221926479878653,
      "grad_norm": 0.010004268027842045,
      "learning_rate": 3.941718356160393e-06,
      "loss": 0.0009,
      "step": 190560
    },
    {
      "epoch": 1.2222567848816515,
      "grad_norm": 0.04696183651685715,
      "learning_rate": 3.941171339969995e-06,
      "loss": 0.0028,
      "step": 190570
    },
    {
      "epoch": 1.2223209217754376,
      "grad_norm": 0.02193659543991089,
      "learning_rate": 3.9406243370474776e-06,
      "loss": 0.0019,
      "step": 190580
    },
    {
      "epoch": 1.2223850586692235,
      "grad_norm": 0.11040711402893066,
      "learning_rate": 3.940077347399693e-06,
      "loss": 0.0019,
      "step": 190590
    },
    {
      "epoch": 1.2224491955630097,
      "grad_norm": 0.08952760696411133,
      "learning_rate": 3.939530371033496e-06,
      "loss": 0.0018,
      "step": 190600
    },
    {
      "epoch": 1.2225133324567958,
      "grad_norm": 0.0674658939242363,
      "learning_rate": 3.938983407955742e-06,
      "loss": 0.001,
      "step": 190610
    },
    {
      "epoch": 1.2225774693505818,
      "grad_norm": 0.16172173619270325,
      "learning_rate": 3.938436458173286e-06,
      "loss": 0.0022,
      "step": 190620
    },
    {
      "epoch": 1.222641606244368,
      "grad_norm": 0.03608992323279381,
      "learning_rate": 3.937889521692978e-06,
      "loss": 0.0013,
      "step": 190630
    },
    {
      "epoch": 1.222705743138154,
      "grad_norm": 0.05420034006237984,
      "learning_rate": 3.9373425985216725e-06,
      "loss": 0.0022,
      "step": 190640
    },
    {
      "epoch": 1.2227698800319402,
      "grad_norm": 0.038899801671504974,
      "learning_rate": 3.936795688666226e-06,
      "loss": 0.002,
      "step": 190650
    },
    {
      "epoch": 1.2228340169257264,
      "grad_norm": 0.14675188064575195,
      "learning_rate": 3.936248792133488e-06,
      "loss": 0.002,
      "step": 190660
    },
    {
      "epoch": 1.2228981538195123,
      "grad_norm": 0.03291318193078041,
      "learning_rate": 3.935701908930314e-06,
      "loss": 0.0014,
      "step": 190670
    },
    {
      "epoch": 1.2229622907132984,
      "grad_norm": 0.22007884085178375,
      "learning_rate": 3.935155039063555e-06,
      "loss": 0.0022,
      "step": 190680
    },
    {
      "epoch": 1.2230264276070846,
      "grad_norm": 0.023851877078413963,
      "learning_rate": 3.934608182540065e-06,
      "loss": 0.0006,
      "step": 190690
    },
    {
      "epoch": 1.2230905645008707,
      "grad_norm": 0.01468751858919859,
      "learning_rate": 3.934061339366694e-06,
      "loss": 0.0015,
      "step": 190700
    },
    {
      "epoch": 1.2231547013946567,
      "grad_norm": 0.08310827612876892,
      "learning_rate": 3.933514509550298e-06,
      "loss": 0.0014,
      "step": 190710
    },
    {
      "epoch": 1.2232188382884428,
      "grad_norm": 0.0643293634057045,
      "learning_rate": 3.932967693097727e-06,
      "loss": 0.0061,
      "step": 190720
    },
    {
      "epoch": 1.223282975182229,
      "grad_norm": 0.025908587500452995,
      "learning_rate": 3.932420890015834e-06,
      "loss": 0.0015,
      "step": 190730
    },
    {
      "epoch": 1.223347112076015,
      "grad_norm": 0.0017419327050447464,
      "learning_rate": 3.931874100311468e-06,
      "loss": 0.0029,
      "step": 190740
    },
    {
      "epoch": 1.2234112489698012,
      "grad_norm": 0.07896079123020172,
      "learning_rate": 3.931327323991484e-06,
      "loss": 0.0016,
      "step": 190750
    },
    {
      "epoch": 1.2234753858635872,
      "grad_norm": 0.03480801731348038,
      "learning_rate": 3.930780561062733e-06,
      "loss": 0.0015,
      "step": 190760
    },
    {
      "epoch": 1.2235395227573733,
      "grad_norm": 0.10877152532339096,
      "learning_rate": 3.930233811532066e-06,
      "loss": 0.001,
      "step": 190770
    },
    {
      "epoch": 1.2236036596511595,
      "grad_norm": 0.14966726303100586,
      "learning_rate": 3.929687075406332e-06,
      "loss": 0.0016,
      "step": 190780
    },
    {
      "epoch": 1.2236677965449456,
      "grad_norm": 0.09734129905700684,
      "learning_rate": 3.929140352692387e-06,
      "loss": 0.0013,
      "step": 190790
    },
    {
      "epoch": 1.2237319334387315,
      "grad_norm": 0.11551859229803085,
      "learning_rate": 3.928593643397076e-06,
      "loss": 0.0016,
      "step": 190800
    },
    {
      "epoch": 1.2237960703325177,
      "grad_norm": 0.07634082436561584,
      "learning_rate": 3.928046947527254e-06,
      "loss": 0.0014,
      "step": 190810
    },
    {
      "epoch": 1.2238602072263038,
      "grad_norm": 0.04561929404735565,
      "learning_rate": 3.927500265089769e-06,
      "loss": 0.0021,
      "step": 190820
    },
    {
      "epoch": 1.22392434412009,
      "grad_norm": 0.026405436918139458,
      "learning_rate": 3.926953596091473e-06,
      "loss": 0.0006,
      "step": 190830
    },
    {
      "epoch": 1.2239884810138761,
      "grad_norm": 0.05360418185591698,
      "learning_rate": 3.926406940539215e-06,
      "loss": 0.0008,
      "step": 190840
    },
    {
      "epoch": 1.224052617907662,
      "grad_norm": 0.10631723701953888,
      "learning_rate": 3.925860298439845e-06,
      "loss": 0.0015,
      "step": 190850
    },
    {
      "epoch": 1.2241167548014482,
      "grad_norm": 0.044981710612773895,
      "learning_rate": 3.925313669800213e-06,
      "loss": 0.0016,
      "step": 190860
    },
    {
      "epoch": 1.2241808916952344,
      "grad_norm": 0.006090010050684214,
      "learning_rate": 3.92476705462717e-06,
      "loss": 0.0021,
      "step": 190870
    },
    {
      "epoch": 1.2242450285890203,
      "grad_norm": 0.07183487713336945,
      "learning_rate": 3.924220452927563e-06,
      "loss": 0.0028,
      "step": 190880
    },
    {
      "epoch": 1.2243091654828064,
      "grad_norm": 0.10996677726507187,
      "learning_rate": 3.9236738647082435e-06,
      "loss": 0.0014,
      "step": 190890
    },
    {
      "epoch": 1.2243733023765926,
      "grad_norm": 0.12826906144618988,
      "learning_rate": 3.923127289976059e-06,
      "loss": 0.0012,
      "step": 190900
    },
    {
      "epoch": 1.2244374392703787,
      "grad_norm": 0.06789213418960571,
      "learning_rate": 3.922580728737858e-06,
      "loss": 0.0011,
      "step": 190910
    },
    {
      "epoch": 1.2245015761641649,
      "grad_norm": 0.18491800129413605,
      "learning_rate": 3.922034181000492e-06,
      "loss": 0.0033,
      "step": 190920
    },
    {
      "epoch": 1.2245657130579508,
      "grad_norm": 0.04658259078860283,
      "learning_rate": 3.921487646770808e-06,
      "loss": 0.0015,
      "step": 190930
    },
    {
      "epoch": 1.224629849951737,
      "grad_norm": 0.16573143005371094,
      "learning_rate": 3.9209411260556555e-06,
      "loss": 0.0019,
      "step": 190940
    },
    {
      "epoch": 1.224693986845523,
      "grad_norm": 0.04675838351249695,
      "learning_rate": 3.92039461886188e-06,
      "loss": 0.0013,
      "step": 190950
    },
    {
      "epoch": 1.2247581237393093,
      "grad_norm": 0.025246139615774155,
      "learning_rate": 3.919848125196333e-06,
      "loss": 0.0014,
      "step": 190960
    },
    {
      "epoch": 1.2248222606330952,
      "grad_norm": 0.046005018055438995,
      "learning_rate": 3.919301645065861e-06,
      "loss": 0.0042,
      "step": 190970
    },
    {
      "epoch": 1.2248863975268813,
      "grad_norm": 0.14488635957241058,
      "learning_rate": 3.918755178477311e-06,
      "loss": 0.0034,
      "step": 190980
    },
    {
      "epoch": 1.2249505344206675,
      "grad_norm": 0.0798855572938919,
      "learning_rate": 3.918208725437531e-06,
      "loss": 0.002,
      "step": 190990
    },
    {
      "epoch": 1.2250146713144536,
      "grad_norm": 0.03376210480928421,
      "learning_rate": 3.9176622859533695e-06,
      "loss": 0.002,
      "step": 191000
    },
    {
      "epoch": 1.2250788082082398,
      "grad_norm": 0.07722453027963638,
      "learning_rate": 3.917115860031673e-06,
      "loss": 0.0014,
      "step": 191010
    },
    {
      "epoch": 1.2251429451020257,
      "grad_norm": 0.17970149219036102,
      "learning_rate": 3.91656944767929e-06,
      "loss": 0.0018,
      "step": 191020
    },
    {
      "epoch": 1.2252070819958119,
      "grad_norm": 0.2177116721868515,
      "learning_rate": 3.916023048903063e-06,
      "loss": 0.0013,
      "step": 191030
    },
    {
      "epoch": 1.225271218889598,
      "grad_norm": 0.1654694676399231,
      "learning_rate": 3.915476663709845e-06,
      "loss": 0.0024,
      "step": 191040
    },
    {
      "epoch": 1.2253353557833842,
      "grad_norm": 0.10967286676168442,
      "learning_rate": 3.914930292106477e-06,
      "loss": 0.002,
      "step": 191050
    },
    {
      "epoch": 1.22539949267717,
      "grad_norm": 0.09925783425569534,
      "learning_rate": 3.914383934099809e-06,
      "loss": 0.0017,
      "step": 191060
    },
    {
      "epoch": 1.2254636295709562,
      "grad_norm": 0.2531794309616089,
      "learning_rate": 3.913837589696687e-06,
      "loss": 0.0018,
      "step": 191070
    },
    {
      "epoch": 1.2255277664647424,
      "grad_norm": 0.025740792974829674,
      "learning_rate": 3.913291258903955e-06,
      "loss": 0.001,
      "step": 191080
    },
    {
      "epoch": 1.2255919033585285,
      "grad_norm": 0.02483651600778103,
      "learning_rate": 3.912744941728461e-06,
      "loss": 0.0018,
      "step": 191090
    },
    {
      "epoch": 1.2256560402523144,
      "grad_norm": 0.04175538569688797,
      "learning_rate": 3.912198638177049e-06,
      "loss": 0.0017,
      "step": 191100
    },
    {
      "epoch": 1.2257201771461006,
      "grad_norm": 0.2951119840145111,
      "learning_rate": 3.911652348256567e-06,
      "loss": 0.0015,
      "step": 191110
    },
    {
      "epoch": 1.2257843140398867,
      "grad_norm": 0.3507690131664276,
      "learning_rate": 3.911106071973858e-06,
      "loss": 0.0015,
      "step": 191120
    },
    {
      "epoch": 1.225848450933673,
      "grad_norm": 0.06317662447690964,
      "learning_rate": 3.910559809335769e-06,
      "loss": 0.002,
      "step": 191130
    },
    {
      "epoch": 1.2259125878274588,
      "grad_norm": 0.0028338609263300896,
      "learning_rate": 3.910013560349143e-06,
      "loss": 0.0012,
      "step": 191140
    },
    {
      "epoch": 1.225976724721245,
      "grad_norm": 0.03527184948325157,
      "learning_rate": 3.909467325020826e-06,
      "loss": 0.0017,
      "step": 191150
    },
    {
      "epoch": 1.2260408616150311,
      "grad_norm": 0.014810839667916298,
      "learning_rate": 3.908921103357663e-06,
      "loss": 0.0022,
      "step": 191160
    },
    {
      "epoch": 1.2261049985088173,
      "grad_norm": 0.1442299634218216,
      "learning_rate": 3.908374895366499e-06,
      "loss": 0.0015,
      "step": 191170
    },
    {
      "epoch": 1.2261691354026034,
      "grad_norm": 0.15262559056282043,
      "learning_rate": 3.907828701054177e-06,
      "loss": 0.0017,
      "step": 191180
    },
    {
      "epoch": 1.2262332722963893,
      "grad_norm": 0.03376549482345581,
      "learning_rate": 3.907282520427543e-06,
      "loss": 0.0015,
      "step": 191190
    },
    {
      "epoch": 1.2262974091901755,
      "grad_norm": 0.16245616972446442,
      "learning_rate": 3.9067363534934385e-06,
      "loss": 0.0017,
      "step": 191200
    },
    {
      "epoch": 1.2263615460839616,
      "grad_norm": 0.09553241729736328,
      "learning_rate": 3.906190200258709e-06,
      "loss": 0.0025,
      "step": 191210
    },
    {
      "epoch": 1.2264256829777478,
      "grad_norm": 0.05627526715397835,
      "learning_rate": 3.9056440607302e-06,
      "loss": 0.0011,
      "step": 191220
    },
    {
      "epoch": 1.2264898198715337,
      "grad_norm": 0.028180358931422234,
      "learning_rate": 3.905097934914751e-06,
      "loss": 0.002,
      "step": 191230
    },
    {
      "epoch": 1.2265539567653199,
      "grad_norm": 0.07987452298402786,
      "learning_rate": 3.9045518228192085e-06,
      "loss": 0.0014,
      "step": 191240
    },
    {
      "epoch": 1.226618093659106,
      "grad_norm": 0.04825502634048462,
      "learning_rate": 3.904005724450413e-06,
      "loss": 0.0015,
      "step": 191250
    },
    {
      "epoch": 1.2266822305528922,
      "grad_norm": 0.0545743964612484,
      "learning_rate": 3.903459639815212e-06,
      "loss": 0.0023,
      "step": 191260
    },
    {
      "epoch": 1.2267463674466783,
      "grad_norm": 0.1914282590150833,
      "learning_rate": 3.902913568920443e-06,
      "loss": 0.0016,
      "step": 191270
    },
    {
      "epoch": 1.2268105043404642,
      "grad_norm": 0.05901381000876427,
      "learning_rate": 3.902367511772952e-06,
      "loss": 0.0011,
      "step": 191280
    },
    {
      "epoch": 1.2268746412342504,
      "grad_norm": 0.07726976275444031,
      "learning_rate": 3.90182146837958e-06,
      "loss": 0.0011,
      "step": 191290
    },
    {
      "epoch": 1.2269387781280365,
      "grad_norm": 0.08275720477104187,
      "learning_rate": 3.901275438747171e-06,
      "loss": 0.0018,
      "step": 191300
    },
    {
      "epoch": 1.2270029150218225,
      "grad_norm": 0.09165611863136292,
      "learning_rate": 3.900729422882564e-06,
      "loss": 0.0014,
      "step": 191310
    },
    {
      "epoch": 1.2270670519156086,
      "grad_norm": 0.0632534846663475,
      "learning_rate": 3.900183420792605e-06,
      "loss": 0.0014,
      "step": 191320
    },
    {
      "epoch": 1.2271311888093948,
      "grad_norm": 0.024312160909175873,
      "learning_rate": 3.899637432484132e-06,
      "loss": 0.0022,
      "step": 191330
    },
    {
      "epoch": 1.227195325703181,
      "grad_norm": 0.05788853392004967,
      "learning_rate": 3.899091457963989e-06,
      "loss": 0.0012,
      "step": 191340
    },
    {
      "epoch": 1.227259462596967,
      "grad_norm": 0.0016625310527160764,
      "learning_rate": 3.898545497239018e-06,
      "loss": 0.0011,
      "step": 191350
    },
    {
      "epoch": 1.227323599490753,
      "grad_norm": 0.13773536682128906,
      "learning_rate": 3.897999550316057e-06,
      "loss": 0.002,
      "step": 191360
    },
    {
      "epoch": 1.2273877363845391,
      "grad_norm": 0.0493728332221508,
      "learning_rate": 3.8974536172019514e-06,
      "loss": 0.0017,
      "step": 191370
    },
    {
      "epoch": 1.2274518732783253,
      "grad_norm": 0.1362447440624237,
      "learning_rate": 3.896907697903538e-06,
      "loss": 0.0019,
      "step": 191380
    },
    {
      "epoch": 1.2275160101721114,
      "grad_norm": 0.0662623718380928,
      "learning_rate": 3.896361792427662e-06,
      "loss": 0.0015,
      "step": 191390
    },
    {
      "epoch": 1.2275801470658974,
      "grad_norm": 0.264980286359787,
      "learning_rate": 3.895815900781159e-06,
      "loss": 0.0027,
      "step": 191400
    },
    {
      "epoch": 1.2276442839596835,
      "grad_norm": 0.20691430568695068,
      "learning_rate": 3.895270022970873e-06,
      "loss": 0.002,
      "step": 191410
    },
    {
      "epoch": 1.2277084208534697,
      "grad_norm": 0.045289549976587296,
      "learning_rate": 3.8947241590036425e-06,
      "loss": 0.0023,
      "step": 191420
    },
    {
      "epoch": 1.2277725577472558,
      "grad_norm": 0.07116792351007462,
      "learning_rate": 3.8941783088863094e-06,
      "loss": 0.0022,
      "step": 191430
    },
    {
      "epoch": 1.227836694641042,
      "grad_norm": 0.09005143493413925,
      "learning_rate": 3.893632472625711e-06,
      "loss": 0.0018,
      "step": 191440
    },
    {
      "epoch": 1.2279008315348279,
      "grad_norm": 0.001668777083978057,
      "learning_rate": 3.89308665022869e-06,
      "loss": 0.0017,
      "step": 191450
    },
    {
      "epoch": 1.227964968428614,
      "grad_norm": 0.11686328798532486,
      "learning_rate": 3.892540841702083e-06,
      "loss": 0.0012,
      "step": 191460
    },
    {
      "epoch": 1.2280291053224002,
      "grad_norm": 0.04654378071427345,
      "learning_rate": 3.8919950470527326e-06,
      "loss": 0.0018,
      "step": 191470
    },
    {
      "epoch": 1.2280932422161863,
      "grad_norm": 0.0269235260784626,
      "learning_rate": 3.891449266287474e-06,
      "loss": 0.0007,
      "step": 191480
    },
    {
      "epoch": 1.2281573791099722,
      "grad_norm": 0.11890744417905807,
      "learning_rate": 3.890903499413148e-06,
      "loss": 0.0028,
      "step": 191490
    },
    {
      "epoch": 1.2282215160037584,
      "grad_norm": 0.17330829799175262,
      "learning_rate": 3.890357746436594e-06,
      "loss": 0.0032,
      "step": 191500
    },
    {
      "epoch": 1.2282856528975445,
      "grad_norm": 0.04410916566848755,
      "learning_rate": 3.8898120073646515e-06,
      "loss": 0.0006,
      "step": 191510
    },
    {
      "epoch": 1.2283497897913307,
      "grad_norm": 0.046860549598932266,
      "learning_rate": 3.889266282204158e-06,
      "loss": 0.002,
      "step": 191520
    },
    {
      "epoch": 1.2284139266851166,
      "grad_norm": 0.060270313173532486,
      "learning_rate": 3.888720570961952e-06,
      "loss": 0.0012,
      "step": 191530
    },
    {
      "epoch": 1.2284780635789028,
      "grad_norm": 0.2538609802722931,
      "learning_rate": 3.888174873644871e-06,
      "loss": 0.0019,
      "step": 191540
    },
    {
      "epoch": 1.228542200472689,
      "grad_norm": 0.04857063293457031,
      "learning_rate": 3.887629190259754e-06,
      "loss": 0.0013,
      "step": 191550
    },
    {
      "epoch": 1.228606337366475,
      "grad_norm": 0.16587914526462555,
      "learning_rate": 3.887083520813438e-06,
      "loss": 0.0052,
      "step": 191560
    },
    {
      "epoch": 1.228670474260261,
      "grad_norm": 0.050497233867645264,
      "learning_rate": 3.886537865312761e-06,
      "loss": 0.0007,
      "step": 191570
    },
    {
      "epoch": 1.2287346111540471,
      "grad_norm": 0.010793877765536308,
      "learning_rate": 3.88599222376456e-06,
      "loss": 0.0021,
      "step": 191580
    },
    {
      "epoch": 1.2287987480478333,
      "grad_norm": 0.1318853795528412,
      "learning_rate": 3.885446596175673e-06,
      "loss": 0.0014,
      "step": 191590
    },
    {
      "epoch": 1.2288628849416194,
      "grad_norm": 0.17666688561439514,
      "learning_rate": 3.884900982552936e-06,
      "loss": 0.0017,
      "step": 191600
    },
    {
      "epoch": 1.2289270218354056,
      "grad_norm": 0.04485802352428436,
      "learning_rate": 3.884355382903187e-06,
      "loss": 0.0012,
      "step": 191610
    },
    {
      "epoch": 1.2289911587291915,
      "grad_norm": 0.05705088749527931,
      "learning_rate": 3.8838097972332625e-06,
      "loss": 0.0005,
      "step": 191620
    },
    {
      "epoch": 1.2290552956229777,
      "grad_norm": 0.12357481569051743,
      "learning_rate": 3.883264225549999e-06,
      "loss": 0.0012,
      "step": 191630
    },
    {
      "epoch": 1.2291194325167638,
      "grad_norm": 0.08283048868179321,
      "learning_rate": 3.882718667860233e-06,
      "loss": 0.0009,
      "step": 191640
    },
    {
      "epoch": 1.22918356941055,
      "grad_norm": 0.041520070284605026,
      "learning_rate": 3.8821731241708e-06,
      "loss": 0.0045,
      "step": 191650
    },
    {
      "epoch": 1.2292477063043359,
      "grad_norm": 0.030925313010811806,
      "learning_rate": 3.881627594488538e-06,
      "loss": 0.0031,
      "step": 191660
    },
    {
      "epoch": 1.229311843198122,
      "grad_norm": 0.06241489574313164,
      "learning_rate": 3.88108207882028e-06,
      "loss": 0.0036,
      "step": 191670
    },
    {
      "epoch": 1.2293759800919082,
      "grad_norm": 0.0917147770524025,
      "learning_rate": 3.8805365771728646e-06,
      "loss": 0.0022,
      "step": 191680
    },
    {
      "epoch": 1.2294401169856943,
      "grad_norm": 0.31132322549819946,
      "learning_rate": 3.879991089553125e-06,
      "loss": 0.0018,
      "step": 191690
    },
    {
      "epoch": 1.2295042538794805,
      "grad_norm": 0.061488717794418335,
      "learning_rate": 3.879445615967897e-06,
      "loss": 0.0008,
      "step": 191700
    },
    {
      "epoch": 1.2295683907732664,
      "grad_norm": 0.24310879409313202,
      "learning_rate": 3.878900156424017e-06,
      "loss": 0.0019,
      "step": 191710
    },
    {
      "epoch": 1.2296325276670526,
      "grad_norm": 0.00773091334849596,
      "learning_rate": 3.87835471092832e-06,
      "loss": 0.0012,
      "step": 191720
    },
    {
      "epoch": 1.2296966645608387,
      "grad_norm": 0.055883124470710754,
      "learning_rate": 3.877809279487638e-06,
      "loss": 0.0017,
      "step": 191730
    },
    {
      "epoch": 1.2297608014546246,
      "grad_norm": 0.06533777713775635,
      "learning_rate": 3.87726386210881e-06,
      "loss": 0.0012,
      "step": 191740
    },
    {
      "epoch": 1.2298249383484108,
      "grad_norm": 0.038957130163908005,
      "learning_rate": 3.876718458798667e-06,
      "loss": 0.0011,
      "step": 191750
    },
    {
      "epoch": 1.229889075242197,
      "grad_norm": 0.04319208115339279,
      "learning_rate": 3.876173069564045e-06,
      "loss": 0.0008,
      "step": 191760
    },
    {
      "epoch": 1.229953212135983,
      "grad_norm": 0.03332489728927612,
      "learning_rate": 3.875627694411778e-06,
      "loss": 0.0011,
      "step": 191770
    },
    {
      "epoch": 1.2300173490297692,
      "grad_norm": 0.05945609509944916,
      "learning_rate": 3.875082333348699e-06,
      "loss": 0.0015,
      "step": 191780
    },
    {
      "epoch": 1.2300814859235552,
      "grad_norm": 0.11271724104881287,
      "learning_rate": 3.874536986381643e-06,
      "loss": 0.0015,
      "step": 191790
    },
    {
      "epoch": 1.2301456228173413,
      "grad_norm": 0.06493068486452103,
      "learning_rate": 3.873991653517441e-06,
      "loss": 0.0011,
      "step": 191800
    },
    {
      "epoch": 1.2302097597111274,
      "grad_norm": 0.008765444159507751,
      "learning_rate": 3.87344633476293e-06,
      "loss": 0.0024,
      "step": 191810
    },
    {
      "epoch": 1.2302738966049136,
      "grad_norm": 0.0825427919626236,
      "learning_rate": 3.872901030124941e-06,
      "loss": 0.001,
      "step": 191820
    },
    {
      "epoch": 1.2303380334986995,
      "grad_norm": 0.16172820329666138,
      "learning_rate": 3.8723557396103084e-06,
      "loss": 0.0012,
      "step": 191830
    },
    {
      "epoch": 1.2304021703924857,
      "grad_norm": 0.05429816246032715,
      "learning_rate": 3.871810463225863e-06,
      "loss": 0.002,
      "step": 191840
    },
    {
      "epoch": 1.2304663072862718,
      "grad_norm": 0.008258895017206669,
      "learning_rate": 3.8712652009784404e-06,
      "loss": 0.0019,
      "step": 191850
    },
    {
      "epoch": 1.230530444180058,
      "grad_norm": 0.20216979086399078,
      "learning_rate": 3.8707199528748694e-06,
      "loss": 0.0014,
      "step": 191860
    },
    {
      "epoch": 1.2305945810738441,
      "grad_norm": 0.01203607302159071,
      "learning_rate": 3.870174718921987e-06,
      "loss": 0.0009,
      "step": 191870
    },
    {
      "epoch": 1.23065871796763,
      "grad_norm": 0.012646661140024662,
      "learning_rate": 3.869629499126621e-06,
      "loss": 0.0013,
      "step": 191880
    },
    {
      "epoch": 1.2307228548614162,
      "grad_norm": 0.0920863151550293,
      "learning_rate": 3.869084293495606e-06,
      "loss": 0.0021,
      "step": 191890
    },
    {
      "epoch": 1.2307869917552023,
      "grad_norm": 0.01504463329911232,
      "learning_rate": 3.868539102035772e-06,
      "loss": 0.0015,
      "step": 191900
    },
    {
      "epoch": 1.2308511286489885,
      "grad_norm": 0.05006036534905434,
      "learning_rate": 3.867993924753952e-06,
      "loss": 0.0011,
      "step": 191910
    },
    {
      "epoch": 1.2309152655427744,
      "grad_norm": 0.20346948504447937,
      "learning_rate": 3.867448761656979e-06,
      "loss": 0.0019,
      "step": 191920
    },
    {
      "epoch": 1.2309794024365606,
      "grad_norm": 0.0386669859290123,
      "learning_rate": 3.86690361275168e-06,
      "loss": 0.0015,
      "step": 191930
    },
    {
      "epoch": 1.2310435393303467,
      "grad_norm": 0.05314434692263603,
      "learning_rate": 3.8663584780448905e-06,
      "loss": 0.0019,
      "step": 191940
    },
    {
      "epoch": 1.2311076762241329,
      "grad_norm": 0.02070430852472782,
      "learning_rate": 3.865813357543438e-06,
      "loss": 0.0025,
      "step": 191950
    },
    {
      "epoch": 1.2311718131179188,
      "grad_norm": 0.1890772134065628,
      "learning_rate": 3.865268251254156e-06,
      "loss": 0.0019,
      "step": 191960
    },
    {
      "epoch": 1.231235950011705,
      "grad_norm": 0.06091101095080376,
      "learning_rate": 3.864723159183873e-06,
      "loss": 0.0019,
      "step": 191970
    },
    {
      "epoch": 1.231300086905491,
      "grad_norm": 0.08680860698223114,
      "learning_rate": 3.86417808133942e-06,
      "loss": 0.0012,
      "step": 191980
    },
    {
      "epoch": 1.2313642237992772,
      "grad_norm": 0.0516524463891983,
      "learning_rate": 3.863633017727628e-06,
      "loss": 0.0015,
      "step": 191990
    },
    {
      "epoch": 1.2314283606930632,
      "grad_norm": 0.0733397975564003,
      "learning_rate": 3.863087968355327e-06,
      "loss": 0.0059,
      "step": 192000
    },
    {
      "epoch": 1.2314924975868493,
      "grad_norm": 0.03720054402947426,
      "learning_rate": 3.862542933229345e-06,
      "loss": 0.001,
      "step": 192010
    },
    {
      "epoch": 1.2315566344806355,
      "grad_norm": 0.03688714653253555,
      "learning_rate": 3.861997912356513e-06,
      "loss": 0.0018,
      "step": 192020
    },
    {
      "epoch": 1.2316207713744216,
      "grad_norm": 0.05015870928764343,
      "learning_rate": 3.861452905743661e-06,
      "loss": 0.0015,
      "step": 192030
    },
    {
      "epoch": 1.2316849082682078,
      "grad_norm": 0.15229342877864838,
      "learning_rate": 3.8609079133976176e-06,
      "loss": 0.0012,
      "step": 192040
    },
    {
      "epoch": 1.2317490451619937,
      "grad_norm": 0.16295719146728516,
      "learning_rate": 3.860362935325213e-06,
      "loss": 0.0018,
      "step": 192050
    },
    {
      "epoch": 1.2318131820557798,
      "grad_norm": 0.035660114139318466,
      "learning_rate": 3.859817971533273e-06,
      "loss": 0.0017,
      "step": 192060
    },
    {
      "epoch": 1.231877318949566,
      "grad_norm": 0.030620815232396126,
      "learning_rate": 3.859273022028631e-06,
      "loss": 0.0026,
      "step": 192070
    },
    {
      "epoch": 1.2319414558433521,
      "grad_norm": 0.1859862059354782,
      "learning_rate": 3.858728086818112e-06,
      "loss": 0.0014,
      "step": 192080
    },
    {
      "epoch": 1.232005592737138,
      "grad_norm": 0.009169869124889374,
      "learning_rate": 3.8581831659085474e-06,
      "loss": 0.0015,
      "step": 192090
    },
    {
      "epoch": 1.2320697296309242,
      "grad_norm": 0.01506097987294197,
      "learning_rate": 3.8576382593067616e-06,
      "loss": 0.0011,
      "step": 192100
    },
    {
      "epoch": 1.2321338665247104,
      "grad_norm": 0.03212521970272064,
      "learning_rate": 3.857093367019587e-06,
      "loss": 0.0014,
      "step": 192110
    },
    {
      "epoch": 1.2321980034184965,
      "grad_norm": 0.23379331827163696,
      "learning_rate": 3.856548489053847e-06,
      "loss": 0.0022,
      "step": 192120
    },
    {
      "epoch": 1.2322621403122827,
      "grad_norm": 0.0456843227148056,
      "learning_rate": 3.856003625416374e-06,
      "loss": 0.0013,
      "step": 192130
    },
    {
      "epoch": 1.2323262772060686,
      "grad_norm": 0.07977654039859772,
      "learning_rate": 3.855458776113991e-06,
      "loss": 0.0015,
      "step": 192140
    },
    {
      "epoch": 1.2323904140998547,
      "grad_norm": 0.09059882909059525,
      "learning_rate": 3.85491394115353e-06,
      "loss": 0.0009,
      "step": 192150
    },
    {
      "epoch": 1.2324545509936409,
      "grad_norm": 0.23698201775550842,
      "learning_rate": 3.854369120541814e-06,
      "loss": 0.0018,
      "step": 192160
    },
    {
      "epoch": 1.2325186878874268,
      "grad_norm": 0.2211540937423706,
      "learning_rate": 3.853824314285672e-06,
      "loss": 0.0021,
      "step": 192170
    },
    {
      "epoch": 1.232582824781213,
      "grad_norm": 0.0691179633140564,
      "learning_rate": 3.85327952239193e-06,
      "loss": 0.0018,
      "step": 192180
    },
    {
      "epoch": 1.232646961674999,
      "grad_norm": 0.03404263034462929,
      "learning_rate": 3.852734744867415e-06,
      "loss": 0.0014,
      "step": 192190
    },
    {
      "epoch": 1.2327110985687852,
      "grad_norm": 0.022241869941353798,
      "learning_rate": 3.852189981718955e-06,
      "loss": 0.0021,
      "step": 192200
    },
    {
      "epoch": 1.2327752354625714,
      "grad_norm": 0.10520397126674652,
      "learning_rate": 3.851645232953373e-06,
      "loss": 0.0012,
      "step": 192210
    },
    {
      "epoch": 1.2328393723563573,
      "grad_norm": 0.13404682278633118,
      "learning_rate": 3.851100498577499e-06,
      "loss": 0.0015,
      "step": 192220
    },
    {
      "epoch": 1.2329035092501435,
      "grad_norm": 0.07412949204444885,
      "learning_rate": 3.8505557785981555e-06,
      "loss": 0.0013,
      "step": 192230
    },
    {
      "epoch": 1.2329676461439296,
      "grad_norm": 0.06194952502846718,
      "learning_rate": 3.8500110730221705e-06,
      "loss": 0.0012,
      "step": 192240
    },
    {
      "epoch": 1.2330317830377158,
      "grad_norm": 0.09382586926221848,
      "learning_rate": 3.849466381856367e-06,
      "loss": 0.0011,
      "step": 192250
    },
    {
      "epoch": 1.2330959199315017,
      "grad_norm": 0.08719581365585327,
      "learning_rate": 3.848921705107574e-06,
      "loss": 0.0029,
      "step": 192260
    },
    {
      "epoch": 1.2331600568252878,
      "grad_norm": 0.012377760373055935,
      "learning_rate": 3.8483770427826125e-06,
      "loss": 0.0016,
      "step": 192270
    },
    {
      "epoch": 1.233224193719074,
      "grad_norm": 0.00940409954637289,
      "learning_rate": 3.84783239488831e-06,
      "loss": 0.0022,
      "step": 192280
    },
    {
      "epoch": 1.2332883306128601,
      "grad_norm": 0.006755847949534655,
      "learning_rate": 3.847287761431493e-06,
      "loss": 0.0015,
      "step": 192290
    },
    {
      "epoch": 1.2333524675066463,
      "grad_norm": 0.07054046541452408,
      "learning_rate": 3.846743142418983e-06,
      "loss": 0.0016,
      "step": 192300
    },
    {
      "epoch": 1.2334166044004322,
      "grad_norm": 0.3278670310974121,
      "learning_rate": 3.846198537857606e-06,
      "loss": 0.001,
      "step": 192310
    },
    {
      "epoch": 1.2334807412942184,
      "grad_norm": 0.038489580154418945,
      "learning_rate": 3.845653947754186e-06,
      "loss": 0.0008,
      "step": 192320
    },
    {
      "epoch": 1.2335448781880045,
      "grad_norm": 0.38875246047973633,
      "learning_rate": 3.845109372115545e-06,
      "loss": 0.0021,
      "step": 192330
    },
    {
      "epoch": 1.2336090150817907,
      "grad_norm": 0.07522208243608475,
      "learning_rate": 3.844564810948511e-06,
      "loss": 0.0011,
      "step": 192340
    },
    {
      "epoch": 1.2336731519755766,
      "grad_norm": 0.06156201288104057,
      "learning_rate": 3.844020264259906e-06,
      "loss": 0.0014,
      "step": 192350
    },
    {
      "epoch": 1.2337372888693627,
      "grad_norm": 0.024865301325917244,
      "learning_rate": 3.843475732056553e-06,
      "loss": 0.0012,
      "step": 192360
    },
    {
      "epoch": 1.2338014257631489,
      "grad_norm": 0.09704294055700302,
      "learning_rate": 3.842931214345275e-06,
      "loss": 0.0014,
      "step": 192370
    },
    {
      "epoch": 1.233865562656935,
      "grad_norm": 0.05504216253757477,
      "learning_rate": 3.8423867111328965e-06,
      "loss": 0.0014,
      "step": 192380
    },
    {
      "epoch": 1.2339296995507212,
      "grad_norm": 0.17114925384521484,
      "learning_rate": 3.841842222426239e-06,
      "loss": 0.0024,
      "step": 192390
    },
    {
      "epoch": 1.233993836444507,
      "grad_norm": 0.15212135016918182,
      "learning_rate": 3.8412977482321275e-06,
      "loss": 0.0015,
      "step": 192400
    },
    {
      "epoch": 1.2340579733382933,
      "grad_norm": 0.11747518926858902,
      "learning_rate": 3.840753288557382e-06,
      "loss": 0.0013,
      "step": 192410
    },
    {
      "epoch": 1.2341221102320794,
      "grad_norm": 0.016201209276914597,
      "learning_rate": 3.840208843408827e-06,
      "loss": 0.0015,
      "step": 192420
    },
    {
      "epoch": 1.2341862471258653,
      "grad_norm": 0.0847243070602417,
      "learning_rate": 3.8396644127932835e-06,
      "loss": 0.0015,
      "step": 192430
    },
    {
      "epoch": 1.2342503840196515,
      "grad_norm": 0.13864751160144806,
      "learning_rate": 3.839119996717576e-06,
      "loss": 0.0028,
      "step": 192440
    },
    {
      "epoch": 1.2343145209134376,
      "grad_norm": 0.09642212837934494,
      "learning_rate": 3.838575595188522e-06,
      "loss": 0.0016,
      "step": 192450
    },
    {
      "epoch": 1.2343786578072238,
      "grad_norm": 0.01905280351638794,
      "learning_rate": 3.838031208212947e-06,
      "loss": 0.0016,
      "step": 192460
    },
    {
      "epoch": 1.23444279470101,
      "grad_norm": 0.06971075385808945,
      "learning_rate": 3.837486835797672e-06,
      "loss": 0.0009,
      "step": 192470
    },
    {
      "epoch": 1.2345069315947959,
      "grad_norm": 0.03184329718351364,
      "learning_rate": 3.836942477949517e-06,
      "loss": 0.0019,
      "step": 192480
    },
    {
      "epoch": 1.234571068488582,
      "grad_norm": 0.04337267577648163,
      "learning_rate": 3.836398134675304e-06,
      "loss": 0.0018,
      "step": 192490
    },
    {
      "epoch": 1.2346352053823682,
      "grad_norm": 1.0930646657943726,
      "learning_rate": 3.835853805981854e-06,
      "loss": 0.0012,
      "step": 192500
    },
    {
      "epoch": 1.2346993422761543,
      "grad_norm": 0.06390325725078583,
      "learning_rate": 3.835309491875989e-06,
      "loss": 0.0013,
      "step": 192510
    },
    {
      "epoch": 1.2347634791699402,
      "grad_norm": 0.07073578983545303,
      "learning_rate": 3.834765192364527e-06,
      "loss": 0.0027,
      "step": 192520
    },
    {
      "epoch": 1.2348276160637264,
      "grad_norm": 0.05255277454853058,
      "learning_rate": 3.834220907454291e-06,
      "loss": 0.0014,
      "step": 192530
    },
    {
      "epoch": 1.2348917529575125,
      "grad_norm": 0.01460962649434805,
      "learning_rate": 3.833676637152099e-06,
      "loss": 0.0013,
      "step": 192540
    },
    {
      "epoch": 1.2349558898512987,
      "grad_norm": 0.02730894461274147,
      "learning_rate": 3.833132381464773e-06,
      "loss": 0.001,
      "step": 192550
    },
    {
      "epoch": 1.2350200267450848,
      "grad_norm": 0.011701928451657295,
      "learning_rate": 3.832588140399131e-06,
      "loss": 0.001,
      "step": 192560
    },
    {
      "epoch": 1.2350841636388707,
      "grad_norm": 0.08750313520431519,
      "learning_rate": 3.832043913961997e-06,
      "loss": 0.0016,
      "step": 192570
    },
    {
      "epoch": 1.235148300532657,
      "grad_norm": 0.06760246306657791,
      "learning_rate": 3.8314997021601846e-06,
      "loss": 0.0025,
      "step": 192580
    },
    {
      "epoch": 1.235212437426443,
      "grad_norm": 0.02466864138841629,
      "learning_rate": 3.830955505000518e-06,
      "loss": 0.0027,
      "step": 192590
    },
    {
      "epoch": 1.2352765743202292,
      "grad_norm": 0.02973290905356407,
      "learning_rate": 3.830411322489812e-06,
      "loss": 0.0014,
      "step": 192600
    },
    {
      "epoch": 1.2353407112140151,
      "grad_norm": 0.04889494553208351,
      "learning_rate": 3.829867154634889e-06,
      "loss": 0.0016,
      "step": 192610
    },
    {
      "epoch": 1.2354048481078013,
      "grad_norm": 0.10187707841396332,
      "learning_rate": 3.829323001442568e-06,
      "loss": 0.0013,
      "step": 192620
    },
    {
      "epoch": 1.2354689850015874,
      "grad_norm": 0.1182231605052948,
      "learning_rate": 3.8287788629196655e-06,
      "loss": 0.0017,
      "step": 192630
    },
    {
      "epoch": 1.2355331218953736,
      "grad_norm": 0.07204227894544601,
      "learning_rate": 3.8282347390730015e-06,
      "loss": 0.0006,
      "step": 192640
    },
    {
      "epoch": 1.2355972587891595,
      "grad_norm": 0.09443531930446625,
      "learning_rate": 3.827690629909393e-06,
      "loss": 0.001,
      "step": 192650
    },
    {
      "epoch": 1.2356613956829456,
      "grad_norm": 0.03801809251308441,
      "learning_rate": 3.82714653543566e-06,
      "loss": 0.001,
      "step": 192660
    },
    {
      "epoch": 1.2357255325767318,
      "grad_norm": 0.10350924730300903,
      "learning_rate": 3.8266024556586164e-06,
      "loss": 0.001,
      "step": 192670
    },
    {
      "epoch": 1.235789669470518,
      "grad_norm": 0.13763123750686646,
      "learning_rate": 3.826058390585086e-06,
      "loss": 0.003,
      "step": 192680
    },
    {
      "epoch": 1.2358538063643039,
      "grad_norm": 0.09756220132112503,
      "learning_rate": 3.82551434022188e-06,
      "loss": 0.0008,
      "step": 192690
    },
    {
      "epoch": 1.23591794325809,
      "grad_norm": 0.09421417117118835,
      "learning_rate": 3.824970304575821e-06,
      "loss": 0.0017,
      "step": 192700
    },
    {
      "epoch": 1.2359820801518762,
      "grad_norm": 0.08317472785711288,
      "learning_rate": 3.824426283653723e-06,
      "loss": 0.0016,
      "step": 192710
    },
    {
      "epoch": 1.2360462170456623,
      "grad_norm": 0.17168138921260834,
      "learning_rate": 3.823882277462404e-06,
      "loss": 0.0015,
      "step": 192720
    },
    {
      "epoch": 1.2361103539394485,
      "grad_norm": 0.025490155443549156,
      "learning_rate": 3.82333828600868e-06,
      "loss": 0.0011,
      "step": 192730
    },
    {
      "epoch": 1.2361744908332344,
      "grad_norm": 0.13631509244441986,
      "learning_rate": 3.822794309299369e-06,
      "loss": 0.0009,
      "step": 192740
    },
    {
      "epoch": 1.2362386277270205,
      "grad_norm": 0.28083887696266174,
      "learning_rate": 3.822250347341286e-06,
      "loss": 0.0026,
      "step": 192750
    },
    {
      "epoch": 1.2363027646208067,
      "grad_norm": 0.17060667276382446,
      "learning_rate": 3.8217064001412475e-06,
      "loss": 0.0015,
      "step": 192760
    },
    {
      "epoch": 1.2363669015145928,
      "grad_norm": 0.10084377229213715,
      "learning_rate": 3.821162467706071e-06,
      "loss": 0.0011,
      "step": 192770
    },
    {
      "epoch": 1.2364310384083788,
      "grad_norm": 0.13612842559814453,
      "learning_rate": 3.820618550042571e-06,
      "loss": 0.003,
      "step": 192780
    },
    {
      "epoch": 1.236495175302165,
      "grad_norm": 0.030563458800315857,
      "learning_rate": 3.8200746471575635e-06,
      "loss": 0.0029,
      "step": 192790
    },
    {
      "epoch": 1.236559312195951,
      "grad_norm": 0.053338564932346344,
      "learning_rate": 3.819530759057863e-06,
      "loss": 0.0017,
      "step": 192800
    },
    {
      "epoch": 1.2366234490897372,
      "grad_norm": 0.02539745718240738,
      "learning_rate": 3.818986885750287e-06,
      "loss": 0.0023,
      "step": 192810
    },
    {
      "epoch": 1.2366875859835234,
      "grad_norm": 0.03159044310450554,
      "learning_rate": 3.818443027241648e-06,
      "loss": 0.0016,
      "step": 192820
    },
    {
      "epoch": 1.2367517228773093,
      "grad_norm": 0.11882440745830536,
      "learning_rate": 3.817899183538765e-06,
      "loss": 0.0015,
      "step": 192830
    },
    {
      "epoch": 1.2368158597710954,
      "grad_norm": 0.1520904302597046,
      "learning_rate": 3.8173553546484475e-06,
      "loss": 0.0014,
      "step": 192840
    },
    {
      "epoch": 1.2368799966648816,
      "grad_norm": 0.025404080748558044,
      "learning_rate": 3.816811540577513e-06,
      "loss": 0.0013,
      "step": 192850
    },
    {
      "epoch": 1.2369441335586675,
      "grad_norm": 0.23404638469219208,
      "learning_rate": 3.8162677413327755e-06,
      "loss": 0.0015,
      "step": 192860
    },
    {
      "epoch": 1.2370082704524537,
      "grad_norm": 0.17075438797473907,
      "learning_rate": 3.815723956921051e-06,
      "loss": 0.0019,
      "step": 192870
    },
    {
      "epoch": 1.2370724073462398,
      "grad_norm": 0.1039583832025528,
      "learning_rate": 3.815180187349149e-06,
      "loss": 0.0018,
      "step": 192880
    },
    {
      "epoch": 1.237136544240026,
      "grad_norm": 0.0913623720407486,
      "learning_rate": 3.8146364326238884e-06,
      "loss": 0.0011,
      "step": 192890
    },
    {
      "epoch": 1.237200681133812,
      "grad_norm": 0.11237393319606781,
      "learning_rate": 3.8140926927520795e-06,
      "loss": 0.0037,
      "step": 192900
    },
    {
      "epoch": 1.237264818027598,
      "grad_norm": 0.12427417188882828,
      "learning_rate": 3.8135489677405358e-06,
      "loss": 0.0015,
      "step": 192910
    },
    {
      "epoch": 1.2373289549213842,
      "grad_norm": 0.010974938049912453,
      "learning_rate": 3.813005257596073e-06,
      "loss": 0.0017,
      "step": 192920
    },
    {
      "epoch": 1.2373930918151703,
      "grad_norm": 0.08517418056726456,
      "learning_rate": 3.8124615623255012e-06,
      "loss": 0.0021,
      "step": 192930
    },
    {
      "epoch": 1.2374572287089565,
      "grad_norm": 0.006680286023765802,
      "learning_rate": 3.8119178819356362e-06,
      "loss": 0.0019,
      "step": 192940
    },
    {
      "epoch": 1.2375213656027424,
      "grad_norm": 0.2067411094903946,
      "learning_rate": 3.811374216433288e-06,
      "loss": 0.0011,
      "step": 192950
    },
    {
      "epoch": 1.2375855024965285,
      "grad_norm": 0.3387925326824188,
      "learning_rate": 3.8108305658252716e-06,
      "loss": 0.0016,
      "step": 192960
    },
    {
      "epoch": 1.2376496393903147,
      "grad_norm": 0.09132348001003265,
      "learning_rate": 3.810286930118397e-06,
      "loss": 0.0021,
      "step": 192970
    },
    {
      "epoch": 1.2377137762841008,
      "grad_norm": 0.20602019131183624,
      "learning_rate": 3.809743309319478e-06,
      "loss": 0.0019,
      "step": 192980
    },
    {
      "epoch": 1.237777913177887,
      "grad_norm": 0.16022828221321106,
      "learning_rate": 3.8091997034353256e-06,
      "loss": 0.0028,
      "step": 192990
    },
    {
      "epoch": 1.237842050071673,
      "grad_norm": 0.0357336662709713,
      "learning_rate": 3.808656112472753e-06,
      "loss": 0.0039,
      "step": 193000
    },
    {
      "epoch": 1.237906186965459,
      "grad_norm": 0.0015852287178859115,
      "learning_rate": 3.8081125364385697e-06,
      "loss": 0.0023,
      "step": 193010
    },
    {
      "epoch": 1.2379703238592452,
      "grad_norm": 0.06841090321540833,
      "learning_rate": 3.807568975339588e-06,
      "loss": 0.0012,
      "step": 193020
    },
    {
      "epoch": 1.2380344607530314,
      "grad_norm": 0.0926680862903595,
      "learning_rate": 3.80702542918262e-06,
      "loss": 0.0025,
      "step": 193030
    },
    {
      "epoch": 1.2380985976468173,
      "grad_norm": 0.1930045485496521,
      "learning_rate": 3.806481897974474e-06,
      "loss": 0.0018,
      "step": 193040
    },
    {
      "epoch": 1.2381627345406034,
      "grad_norm": 0.10639449208974838,
      "learning_rate": 3.8059383817219644e-06,
      "loss": 0.0014,
      "step": 193050
    },
    {
      "epoch": 1.2382268714343896,
      "grad_norm": 0.11121968179941177,
      "learning_rate": 3.805394880431901e-06,
      "loss": 0.0025,
      "step": 193060
    },
    {
      "epoch": 1.2382910083281757,
      "grad_norm": 0.02013281360268593,
      "learning_rate": 3.804851394111092e-06,
      "loss": 0.001,
      "step": 193070
    },
    {
      "epoch": 1.2383551452219617,
      "grad_norm": 0.3018694818019867,
      "learning_rate": 3.8043079227663504e-06,
      "loss": 0.0021,
      "step": 193080
    },
    {
      "epoch": 1.2384192821157478,
      "grad_norm": 0.06745024770498276,
      "learning_rate": 3.8037644664044845e-06,
      "loss": 0.0017,
      "step": 193090
    },
    {
      "epoch": 1.238483419009534,
      "grad_norm": 0.04946301877498627,
      "learning_rate": 3.803221025032305e-06,
      "loss": 0.001,
      "step": 193100
    },
    {
      "epoch": 1.23854755590332,
      "grad_norm": 0.011163666844367981,
      "learning_rate": 3.802677598656621e-06,
      "loss": 0.0028,
      "step": 193110
    },
    {
      "epoch": 1.238611692797106,
      "grad_norm": 0.0768004059791565,
      "learning_rate": 3.802134187284244e-06,
      "loss": 0.0014,
      "step": 193120
    },
    {
      "epoch": 1.2386758296908922,
      "grad_norm": 0.10020840913057327,
      "learning_rate": 3.8015907909219797e-06,
      "loss": 0.0007,
      "step": 193130
    },
    {
      "epoch": 1.2387399665846783,
      "grad_norm": 0.005092281382530928,
      "learning_rate": 3.80104740957664e-06,
      "loss": 0.001,
      "step": 193140
    },
    {
      "epoch": 1.2388041034784645,
      "grad_norm": 0.03567197546362877,
      "learning_rate": 3.8005040432550334e-06,
      "loss": 0.0019,
      "step": 193150
    },
    {
      "epoch": 1.2388682403722506,
      "grad_norm": 0.1282946914434433,
      "learning_rate": 3.7999606919639687e-06,
      "loss": 0.0027,
      "step": 193160
    },
    {
      "epoch": 1.2389323772660366,
      "grad_norm": 0.020889919251203537,
      "learning_rate": 3.799417355710253e-06,
      "loss": 0.0013,
      "step": 193170
    },
    {
      "epoch": 1.2389965141598227,
      "grad_norm": 0.08679716289043427,
      "learning_rate": 3.798874034500696e-06,
      "loss": 0.002,
      "step": 193180
    },
    {
      "epoch": 1.2390606510536089,
      "grad_norm": 0.11425179988145828,
      "learning_rate": 3.798330728342107e-06,
      "loss": 0.0013,
      "step": 193190
    },
    {
      "epoch": 1.239124787947395,
      "grad_norm": 0.032662127166986465,
      "learning_rate": 3.797787437241292e-06,
      "loss": 0.0012,
      "step": 193200
    },
    {
      "epoch": 1.239188924841181,
      "grad_norm": 0.1015615239739418,
      "learning_rate": 3.79724416120506e-06,
      "loss": 0.0022,
      "step": 193210
    },
    {
      "epoch": 1.239253061734967,
      "grad_norm": 0.27617600560188293,
      "learning_rate": 3.7967009002402176e-06,
      "loss": 0.0016,
      "step": 193220
    },
    {
      "epoch": 1.2393171986287532,
      "grad_norm": 0.11621396988630295,
      "learning_rate": 3.7961576543535745e-06,
      "loss": 0.0015,
      "step": 193230
    },
    {
      "epoch": 1.2393813355225394,
      "grad_norm": 0.028171326965093613,
      "learning_rate": 3.7956144235519343e-06,
      "loss": 0.0016,
      "step": 193240
    },
    {
      "epoch": 1.2394454724163255,
      "grad_norm": 0.039086826145648956,
      "learning_rate": 3.795071207842108e-06,
      "loss": 0.0024,
      "step": 193250
    },
    {
      "epoch": 1.2395096093101114,
      "grad_norm": 0.1428215205669403,
      "learning_rate": 3.794528007230899e-06,
      "loss": 0.0018,
      "step": 193260
    },
    {
      "epoch": 1.2395737462038976,
      "grad_norm": 0.006755185779184103,
      "learning_rate": 3.793984821725117e-06,
      "loss": 0.003,
      "step": 193270
    },
    {
      "epoch": 1.2396378830976837,
      "grad_norm": 0.023260431364178658,
      "learning_rate": 3.793441651331566e-06,
      "loss": 0.0026,
      "step": 193280
    },
    {
      "epoch": 1.2397020199914697,
      "grad_norm": 0.03583270683884621,
      "learning_rate": 3.7928984960570542e-06,
      "loss": 0.0008,
      "step": 193290
    },
    {
      "epoch": 1.2397661568852558,
      "grad_norm": 0.04842666909098625,
      "learning_rate": 3.792355355908386e-06,
      "loss": 0.0012,
      "step": 193300
    },
    {
      "epoch": 1.239830293779042,
      "grad_norm": 0.07545134425163269,
      "learning_rate": 3.7918122308923682e-06,
      "loss": 0.0017,
      "step": 193310
    },
    {
      "epoch": 1.2398944306728281,
      "grad_norm": 0.12448123842477798,
      "learning_rate": 3.7912691210158083e-06,
      "loss": 0.0015,
      "step": 193320
    },
    {
      "epoch": 1.2399585675666143,
      "grad_norm": 0.07896067202091217,
      "learning_rate": 3.7907260262855083e-06,
      "loss": 0.0014,
      "step": 193330
    },
    {
      "epoch": 1.2400227044604002,
      "grad_norm": 0.11490961909294128,
      "learning_rate": 3.790182946708277e-06,
      "loss": 0.0018,
      "step": 193340
    },
    {
      "epoch": 1.2400868413541863,
      "grad_norm": 0.12536735832691193,
      "learning_rate": 3.7896398822909166e-06,
      "loss": 0.0007,
      "step": 193350
    },
    {
      "epoch": 1.2401509782479725,
      "grad_norm": 0.04173259809613228,
      "learning_rate": 3.789096833040235e-06,
      "loss": 0.0013,
      "step": 193360
    },
    {
      "epoch": 1.2402151151417586,
      "grad_norm": 0.11744063347578049,
      "learning_rate": 3.788553798963034e-06,
      "loss": 0.001,
      "step": 193370
    },
    {
      "epoch": 1.2402792520355446,
      "grad_norm": 0.05858948081731796,
      "learning_rate": 3.788010780066121e-06,
      "loss": 0.0014,
      "step": 193380
    },
    {
      "epoch": 1.2403433889293307,
      "grad_norm": 0.30108389258384705,
      "learning_rate": 3.7874677763562982e-06,
      "loss": 0.0008,
      "step": 193390
    },
    {
      "epoch": 1.2404075258231169,
      "grad_norm": 0.07916703075170517,
      "learning_rate": 3.7869247878403714e-06,
      "loss": 0.0015,
      "step": 193400
    },
    {
      "epoch": 1.240471662716903,
      "grad_norm": 0.039126984775066376,
      "learning_rate": 3.7863818145251437e-06,
      "loss": 0.0009,
      "step": 193410
    },
    {
      "epoch": 1.2405357996106892,
      "grad_norm": 0.08537046611309052,
      "learning_rate": 3.7858388564174197e-06,
      "loss": 0.0014,
      "step": 193420
    },
    {
      "epoch": 1.240599936504475,
      "grad_norm": 0.052776217460632324,
      "learning_rate": 3.7852959135240016e-06,
      "loss": 0.0009,
      "step": 193430
    },
    {
      "epoch": 1.2406640733982612,
      "grad_norm": 0.03654957935214043,
      "learning_rate": 3.7847529858516948e-06,
      "loss": 0.0017,
      "step": 193440
    },
    {
      "epoch": 1.2407282102920474,
      "grad_norm": 0.12807244062423706,
      "learning_rate": 3.7842100734073006e-06,
      "loss": 0.0012,
      "step": 193450
    },
    {
      "epoch": 1.2407923471858335,
      "grad_norm": 0.04586097598075867,
      "learning_rate": 3.7836671761976228e-06,
      "loss": 0.0009,
      "step": 193460
    },
    {
      "epoch": 1.2408564840796195,
      "grad_norm": 0.028308771550655365,
      "learning_rate": 3.783124294229466e-06,
      "loss": 0.0028,
      "step": 193470
    },
    {
      "epoch": 1.2409206209734056,
      "grad_norm": 0.08119162172079086,
      "learning_rate": 3.7825814275096306e-06,
      "loss": 0.0013,
      "step": 193480
    },
    {
      "epoch": 1.2409847578671918,
      "grad_norm": 0.06081259623169899,
      "learning_rate": 3.782038576044921e-06,
      "loss": 0.0016,
      "step": 193490
    },
    {
      "epoch": 1.241048894760978,
      "grad_norm": 0.06799715757369995,
      "learning_rate": 3.781495739842137e-06,
      "loss": 0.002,
      "step": 193500
    },
    {
      "epoch": 1.2411130316547638,
      "grad_norm": 0.049341876059770584,
      "learning_rate": 3.7809529189080837e-06,
      "loss": 0.0017,
      "step": 193510
    },
    {
      "epoch": 1.24117716854855,
      "grad_norm": 0.052619002759456635,
      "learning_rate": 3.78041011324956e-06,
      "loss": 0.0014,
      "step": 193520
    },
    {
      "epoch": 1.2412413054423361,
      "grad_norm": 0.09561476856470108,
      "learning_rate": 3.779867322873371e-06,
      "loss": 0.0016,
      "step": 193530
    },
    {
      "epoch": 1.2413054423361223,
      "grad_norm": 0.056680865585803986,
      "learning_rate": 3.779324547786315e-06,
      "loss": 0.0018,
      "step": 193540
    },
    {
      "epoch": 1.2413695792299082,
      "grad_norm": 0.09741806983947754,
      "learning_rate": 3.778781787995196e-06,
      "loss": 0.0023,
      "step": 193550
    },
    {
      "epoch": 1.2414337161236944,
      "grad_norm": 0.05924725532531738,
      "learning_rate": 3.778239043506813e-06,
      "loss": 0.0011,
      "step": 193560
    },
    {
      "epoch": 1.2414978530174805,
      "grad_norm": 0.05741953104734421,
      "learning_rate": 3.777696314327969e-06,
      "loss": 0.0025,
      "step": 193570
    },
    {
      "epoch": 1.2415619899112667,
      "grad_norm": 0.09546720236539841,
      "learning_rate": 3.777153600465462e-06,
      "loss": 0.0016,
      "step": 193580
    },
    {
      "epoch": 1.2416261268050528,
      "grad_norm": 0.009045228362083435,
      "learning_rate": 3.776610901926096e-06,
      "loss": 0.005,
      "step": 193590
    },
    {
      "epoch": 1.2416902636988387,
      "grad_norm": 0.19120807945728302,
      "learning_rate": 3.7760682187166685e-06,
      "loss": 0.0016,
      "step": 193600
    },
    {
      "epoch": 1.2417544005926249,
      "grad_norm": 0.14705318212509155,
      "learning_rate": 3.7755255508439813e-06,
      "loss": 0.0019,
      "step": 193610
    },
    {
      "epoch": 1.241818537486411,
      "grad_norm": 0.06438015401363373,
      "learning_rate": 3.7749828983148352e-06,
      "loss": 0.0028,
      "step": 193620
    },
    {
      "epoch": 1.2418826743801972,
      "grad_norm": 0.0024094143882393837,
      "learning_rate": 3.7744402611360276e-06,
      "loss": 0.0009,
      "step": 193630
    },
    {
      "epoch": 1.241946811273983,
      "grad_norm": 0.04401683062314987,
      "learning_rate": 3.77389763931436e-06,
      "loss": 0.0017,
      "step": 193640
    },
    {
      "epoch": 1.2420109481677692,
      "grad_norm": 0.031052129343152046,
      "learning_rate": 3.7733550328566308e-06,
      "loss": 0.0015,
      "step": 193650
    },
    {
      "epoch": 1.2420750850615554,
      "grad_norm": 0.07920389622449875,
      "learning_rate": 3.772812441769641e-06,
      "loss": 0.0013,
      "step": 193660
    },
    {
      "epoch": 1.2421392219553415,
      "grad_norm": 0.02331819385290146,
      "learning_rate": 3.7722698660601864e-06,
      "loss": 0.0017,
      "step": 193670
    },
    {
      "epoch": 1.2422033588491277,
      "grad_norm": 0.147809699177742,
      "learning_rate": 3.7717273057350697e-06,
      "loss": 0.001,
      "step": 193680
    },
    {
      "epoch": 1.2422674957429136,
      "grad_norm": 0.22028706967830658,
      "learning_rate": 3.7711847608010867e-06,
      "loss": 0.0018,
      "step": 193690
    },
    {
      "epoch": 1.2423316326366998,
      "grad_norm": 0.05800031125545502,
      "learning_rate": 3.770642231265037e-06,
      "loss": 0.001,
      "step": 193700
    },
    {
      "epoch": 1.242395769530486,
      "grad_norm": 0.18299369513988495,
      "learning_rate": 3.7700997171337184e-06,
      "loss": 0.0011,
      "step": 193710
    },
    {
      "epoch": 1.2424599064242718,
      "grad_norm": 0.20967887341976166,
      "learning_rate": 3.76955721841393e-06,
      "loss": 0.0027,
      "step": 193720
    },
    {
      "epoch": 1.242524043318058,
      "grad_norm": 0.011691104620695114,
      "learning_rate": 3.769014735112468e-06,
      "loss": 0.0011,
      "step": 193730
    },
    {
      "epoch": 1.2425881802118441,
      "grad_norm": 0.34137067198753357,
      "learning_rate": 3.7684722672361328e-06,
      "loss": 0.001,
      "step": 193740
    },
    {
      "epoch": 1.2426523171056303,
      "grad_norm": 0.0727938637137413,
      "learning_rate": 3.7679298147917186e-06,
      "loss": 0.0016,
      "step": 193750
    },
    {
      "epoch": 1.2427164539994164,
      "grad_norm": 0.04760196432471275,
      "learning_rate": 3.767387377786025e-06,
      "loss": 0.0013,
      "step": 193760
    },
    {
      "epoch": 1.2427805908932024,
      "grad_norm": 0.28860580921173096,
      "learning_rate": 3.766844956225849e-06,
      "loss": 0.0012,
      "step": 193770
    },
    {
      "epoch": 1.2428447277869885,
      "grad_norm": 0.04993847385048866,
      "learning_rate": 3.766302550117986e-06,
      "loss": 0.0013,
      "step": 193780
    },
    {
      "epoch": 1.2429088646807747,
      "grad_norm": 0.06602156162261963,
      "learning_rate": 3.765760159469235e-06,
      "loss": 0.0029,
      "step": 193790
    },
    {
      "epoch": 1.2429730015745608,
      "grad_norm": 0.027851253747940063,
      "learning_rate": 3.765217784286389e-06,
      "loss": 0.0009,
      "step": 193800
    },
    {
      "epoch": 1.2430371384683467,
      "grad_norm": 0.10883858799934387,
      "learning_rate": 3.764675424576248e-06,
      "loss": 0.0011,
      "step": 193810
    },
    {
      "epoch": 1.2431012753621329,
      "grad_norm": 0.053645309060811996,
      "learning_rate": 3.764133080345608e-06,
      "loss": 0.0011,
      "step": 193820
    },
    {
      "epoch": 1.243165412255919,
      "grad_norm": 0.11289020627737045,
      "learning_rate": 3.763590751601262e-06,
      "loss": 0.0022,
      "step": 193830
    },
    {
      "epoch": 1.2432295491497052,
      "grad_norm": 0.06499308347702026,
      "learning_rate": 3.763048438350009e-06,
      "loss": 0.0059,
      "step": 193840
    },
    {
      "epoch": 1.2432936860434913,
      "grad_norm": 0.028254959732294083,
      "learning_rate": 3.762506140598642e-06,
      "loss": 0.0014,
      "step": 193850
    },
    {
      "epoch": 1.2433578229372773,
      "grad_norm": 0.07718160003423691,
      "learning_rate": 3.7619638583539587e-06,
      "loss": 0.0014,
      "step": 193860
    },
    {
      "epoch": 1.2434219598310634,
      "grad_norm": 0.036642514169216156,
      "learning_rate": 3.7614215916227515e-06,
      "loss": 0.0012,
      "step": 193870
    },
    {
      "epoch": 1.2434860967248496,
      "grad_norm": 0.09880058467388153,
      "learning_rate": 3.760879340411817e-06,
      "loss": 0.0009,
      "step": 193880
    },
    {
      "epoch": 1.2435502336186357,
      "grad_norm": 0.10891883075237274,
      "learning_rate": 3.7603371047279517e-06,
      "loss": 0.0015,
      "step": 193890
    },
    {
      "epoch": 1.2436143705124216,
      "grad_norm": 0.2018737643957138,
      "learning_rate": 3.759794884577947e-06,
      "loss": 0.0014,
      "step": 193900
    },
    {
      "epoch": 1.2436785074062078,
      "grad_norm": 0.0744648426771164,
      "learning_rate": 3.7592526799686e-06,
      "loss": 0.0009,
      "step": 193910
    },
    {
      "epoch": 1.243742644299994,
      "grad_norm": 0.07834817469120026,
      "learning_rate": 3.7587104909067024e-06,
      "loss": 0.003,
      "step": 193920
    },
    {
      "epoch": 1.24380678119378,
      "grad_norm": 0.06488928943872452,
      "learning_rate": 3.758168317399051e-06,
      "loss": 0.0017,
      "step": 193930
    },
    {
      "epoch": 1.2438709180875662,
      "grad_norm": 0.08567792922258377,
      "learning_rate": 3.757626159452436e-06,
      "loss": 0.0008,
      "step": 193940
    },
    {
      "epoch": 1.2439350549813522,
      "grad_norm": 0.07470423728227615,
      "learning_rate": 3.757084017073655e-06,
      "loss": 0.0012,
      "step": 193950
    },
    {
      "epoch": 1.2439991918751383,
      "grad_norm": 0.03506644070148468,
      "learning_rate": 3.7565418902694984e-06,
      "loss": 0.0015,
      "step": 193960
    },
    {
      "epoch": 1.2440633287689244,
      "grad_norm": 0.029685180634260178,
      "learning_rate": 3.7559997790467616e-06,
      "loss": 0.0024,
      "step": 193970
    },
    {
      "epoch": 1.2441274656627104,
      "grad_norm": 0.1393328756093979,
      "learning_rate": 3.755457683412236e-06,
      "loss": 0.0013,
      "step": 193980
    },
    {
      "epoch": 1.2441916025564965,
      "grad_norm": 0.08571834862232208,
      "learning_rate": 3.754915603372715e-06,
      "loss": 0.0022,
      "step": 193990
    },
    {
      "epoch": 1.2442557394502827,
      "grad_norm": 0.06823822110891342,
      "learning_rate": 3.7543735389349913e-06,
      "loss": 0.0021,
      "step": 194000
    },
    {
      "epoch": 1.2443198763440688,
      "grad_norm": 0.022043902426958084,
      "learning_rate": 3.753831490105858e-06,
      "loss": 0.0014,
      "step": 194010
    },
    {
      "epoch": 1.244384013237855,
      "grad_norm": 0.09544781595468521,
      "learning_rate": 3.7532894568921057e-06,
      "loss": 0.0019,
      "step": 194020
    },
    {
      "epoch": 1.244448150131641,
      "grad_norm": 0.21995370090007782,
      "learning_rate": 3.752747439300528e-06,
      "loss": 0.0015,
      "step": 194030
    },
    {
      "epoch": 1.244512287025427,
      "grad_norm": 0.050760120153427124,
      "learning_rate": 3.752205437337917e-06,
      "loss": 0.0014,
      "step": 194040
    },
    {
      "epoch": 1.2445764239192132,
      "grad_norm": 0.12651541829109192,
      "learning_rate": 3.7516634510110628e-06,
      "loss": 0.0031,
      "step": 194050
    },
    {
      "epoch": 1.2446405608129993,
      "grad_norm": 0.08600185811519623,
      "learning_rate": 3.751121480326758e-06,
      "loss": 0.0017,
      "step": 194060
    },
    {
      "epoch": 1.2447046977067853,
      "grad_norm": 0.3223367929458618,
      "learning_rate": 3.750579525291793e-06,
      "loss": 0.0033,
      "step": 194070
    },
    {
      "epoch": 1.2447688346005714,
      "grad_norm": 0.2447756975889206,
      "learning_rate": 3.7500375859129613e-06,
      "loss": 0.0015,
      "step": 194080
    },
    {
      "epoch": 1.2448329714943576,
      "grad_norm": 0.0507052056491375,
      "learning_rate": 3.7494956621970503e-06,
      "loss": 0.008,
      "step": 194090
    },
    {
      "epoch": 1.2448971083881437,
      "grad_norm": 0.07987413555383682,
      "learning_rate": 3.748953754150854e-06,
      "loss": 0.0016,
      "step": 194100
    },
    {
      "epoch": 1.2449612452819299,
      "grad_norm": 0.055381108075380325,
      "learning_rate": 3.7484118617811593e-06,
      "loss": 0.0012,
      "step": 194110
    },
    {
      "epoch": 1.2450253821757158,
      "grad_norm": 0.06680621951818466,
      "learning_rate": 3.7478699850947597e-06,
      "loss": 0.0029,
      "step": 194120
    },
    {
      "epoch": 1.245089519069502,
      "grad_norm": 0.15559466183185577,
      "learning_rate": 3.7473281240984426e-06,
      "loss": 0.0026,
      "step": 194130
    },
    {
      "epoch": 1.245153655963288,
      "grad_norm": 0.11484317481517792,
      "learning_rate": 3.7467862787990018e-06,
      "loss": 0.0015,
      "step": 194140
    },
    {
      "epoch": 1.2452177928570742,
      "grad_norm": 0.11385003477334976,
      "learning_rate": 3.7462444492032223e-06,
      "loss": 0.0025,
      "step": 194150
    },
    {
      "epoch": 1.2452819297508602,
      "grad_norm": 0.027558807283639908,
      "learning_rate": 3.7457026353178965e-06,
      "loss": 0.0023,
      "step": 194160
    },
    {
      "epoch": 1.2453460666446463,
      "grad_norm": 0.036400206387043,
      "learning_rate": 3.7451608371498137e-06,
      "loss": 0.0012,
      "step": 194170
    },
    {
      "epoch": 1.2454102035384325,
      "grad_norm": 0.0022631220053881407,
      "learning_rate": 3.7446190547057622e-06,
      "loss": 0.0023,
      "step": 194180
    },
    {
      "epoch": 1.2454743404322186,
      "grad_norm": 0.04237014800310135,
      "learning_rate": 3.7440772879925313e-06,
      "loss": 0.001,
      "step": 194190
    },
    {
      "epoch": 1.2455384773260045,
      "grad_norm": 0.08778490126132965,
      "learning_rate": 3.743535537016909e-06,
      "loss": 0.0017,
      "step": 194200
    },
    {
      "epoch": 1.2456026142197907,
      "grad_norm": 0.057334933429956436,
      "learning_rate": 3.742993801785686e-06,
      "loss": 0.0017,
      "step": 194210
    },
    {
      "epoch": 1.2456667511135768,
      "grad_norm": 0.08290781825780869,
      "learning_rate": 3.742452082305647e-06,
      "loss": 0.0028,
      "step": 194220
    },
    {
      "epoch": 1.245730888007363,
      "grad_norm": 0.042327698320150375,
      "learning_rate": 3.7419103785835835e-06,
      "loss": 0.001,
      "step": 194230
    },
    {
      "epoch": 1.245795024901149,
      "grad_norm": 0.0877726674079895,
      "learning_rate": 3.7413686906262813e-06,
      "loss": 0.0018,
      "step": 194240
    },
    {
      "epoch": 1.245859161794935,
      "grad_norm": 0.08212432265281677,
      "learning_rate": 3.7408270184405293e-06,
      "loss": 0.0013,
      "step": 194250
    },
    {
      "epoch": 1.2459232986887212,
      "grad_norm": 0.047862276434898376,
      "learning_rate": 3.740285362033115e-06,
      "loss": 0.0015,
      "step": 194260
    },
    {
      "epoch": 1.2459874355825074,
      "grad_norm": 0.03472644463181496,
      "learning_rate": 3.739743721410825e-06,
      "loss": 0.0019,
      "step": 194270
    },
    {
      "epoch": 1.2460515724762935,
      "grad_norm": 0.045804962515830994,
      "learning_rate": 3.7392020965804464e-06,
      "loss": 0.001,
      "step": 194280
    },
    {
      "epoch": 1.2461157093700794,
      "grad_norm": 0.03998519852757454,
      "learning_rate": 3.7386604875487675e-06,
      "loss": 0.0008,
      "step": 194290
    },
    {
      "epoch": 1.2461798462638656,
      "grad_norm": 0.031306635588407516,
      "learning_rate": 3.738118894322574e-06,
      "loss": 0.0014,
      "step": 194300
    },
    {
      "epoch": 1.2462439831576517,
      "grad_norm": 0.12886030972003937,
      "learning_rate": 3.7375773169086515e-06,
      "loss": 0.0012,
      "step": 194310
    },
    {
      "epoch": 1.2463081200514379,
      "grad_norm": 0.0673123300075531,
      "learning_rate": 3.7370357553137895e-06,
      "loss": 0.0012,
      "step": 194320
    },
    {
      "epoch": 1.2463722569452238,
      "grad_norm": 0.0477038212120533,
      "learning_rate": 3.73649420954477e-06,
      "loss": 0.0022,
      "step": 194330
    },
    {
      "epoch": 1.24643639383901,
      "grad_norm": 0.0366206169128418,
      "learning_rate": 3.735952679608382e-06,
      "loss": 0.0012,
      "step": 194340
    },
    {
      "epoch": 1.246500530732796,
      "grad_norm": 0.06619582325220108,
      "learning_rate": 3.73541116551141e-06,
      "loss": 0.0026,
      "step": 194350
    },
    {
      "epoch": 1.2465646676265822,
      "grad_norm": 0.0772685706615448,
      "learning_rate": 3.734869667260641e-06,
      "loss": 0.0008,
      "step": 194360
    },
    {
      "epoch": 1.2466288045203684,
      "grad_norm": 0.08183934539556503,
      "learning_rate": 3.734328184862858e-06,
      "loss": 0.0009,
      "step": 194370
    },
    {
      "epoch": 1.2466929414141543,
      "grad_norm": 0.3086586892604828,
      "learning_rate": 3.733786718324849e-06,
      "loss": 0.0023,
      "step": 194380
    },
    {
      "epoch": 1.2467570783079405,
      "grad_norm": 0.07298220694065094,
      "learning_rate": 3.7332452676533954e-06,
      "loss": 0.0013,
      "step": 194390
    },
    {
      "epoch": 1.2468212152017266,
      "grad_norm": 0.022465460002422333,
      "learning_rate": 3.732703832855286e-06,
      "loss": 0.0017,
      "step": 194400
    },
    {
      "epoch": 1.2468853520955125,
      "grad_norm": 0.04418754205107689,
      "learning_rate": 3.732162413937301e-06,
      "loss": 0.0007,
      "step": 194410
    },
    {
      "epoch": 1.2469494889892987,
      "grad_norm": 0.13295075297355652,
      "learning_rate": 3.731621010906229e-06,
      "loss": 0.0017,
      "step": 194420
    },
    {
      "epoch": 1.2470136258830848,
      "grad_norm": 0.10908317565917969,
      "learning_rate": 3.7310796237688506e-06,
      "loss": 0.0008,
      "step": 194430
    },
    {
      "epoch": 1.247077762776871,
      "grad_norm": 0.2138085514307022,
      "learning_rate": 3.7305382525319527e-06,
      "loss": 0.0018,
      "step": 194440
    },
    {
      "epoch": 1.2471418996706571,
      "grad_norm": 0.09427808225154877,
      "learning_rate": 3.729996897202317e-06,
      "loss": 0.0007,
      "step": 194450
    },
    {
      "epoch": 1.247206036564443,
      "grad_norm": 0.11858411878347397,
      "learning_rate": 3.7294555577867277e-06,
      "loss": 0.0011,
      "step": 194460
    },
    {
      "epoch": 1.2472701734582292,
      "grad_norm": 0.08571857213973999,
      "learning_rate": 3.7289142342919695e-06,
      "loss": 0.0008,
      "step": 194470
    },
    {
      "epoch": 1.2473343103520154,
      "grad_norm": 0.0037415132392197847,
      "learning_rate": 3.7283729267248226e-06,
      "loss": 0.0012,
      "step": 194480
    },
    {
      "epoch": 1.2473984472458015,
      "grad_norm": 0.027030210942029953,
      "learning_rate": 3.727831635092073e-06,
      "loss": 0.0009,
      "step": 194490
    },
    {
      "epoch": 1.2474625841395874,
      "grad_norm": 0.06800957024097443,
      "learning_rate": 3.7272903594005015e-06,
      "loss": 0.0009,
      "step": 194500
    },
    {
      "epoch": 1.2475267210333736,
      "grad_norm": 0.07498878240585327,
      "learning_rate": 3.7267490996568924e-06,
      "loss": 0.0014,
      "step": 194510
    },
    {
      "epoch": 1.2475908579271597,
      "grad_norm": 0.0452398918569088,
      "learning_rate": 3.726207855868026e-06,
      "loss": 0.0023,
      "step": 194520
    },
    {
      "epoch": 1.2476549948209459,
      "grad_norm": 0.1459784060716629,
      "learning_rate": 3.7256666280406866e-06,
      "loss": 0.0011,
      "step": 194530
    },
    {
      "epoch": 1.247719131714732,
      "grad_norm": 0.15869516134262085,
      "learning_rate": 3.7251254161816543e-06,
      "loss": 0.0018,
      "step": 194540
    },
    {
      "epoch": 1.247783268608518,
      "grad_norm": 0.01833445392549038,
      "learning_rate": 3.7245842202977124e-06,
      "loss": 0.0008,
      "step": 194550
    },
    {
      "epoch": 1.247847405502304,
      "grad_norm": 0.06426620483398438,
      "learning_rate": 3.7240430403956407e-06,
      "loss": 0.0013,
      "step": 194560
    },
    {
      "epoch": 1.2479115423960903,
      "grad_norm": 0.094499871134758,
      "learning_rate": 3.7235018764822206e-06,
      "loss": 0.0022,
      "step": 194570
    },
    {
      "epoch": 1.2479756792898764,
      "grad_norm": 0.06422349065542221,
      "learning_rate": 3.7229607285642354e-06,
      "loss": 0.0026,
      "step": 194580
    },
    {
      "epoch": 1.2480398161836623,
      "grad_norm": 0.04425198957324028,
      "learning_rate": 3.7224195966484656e-06,
      "loss": 0.0022,
      "step": 194590
    },
    {
      "epoch": 1.2481039530774485,
      "grad_norm": 0.025903604924678802,
      "learning_rate": 3.7218784807416906e-06,
      "loss": 0.0014,
      "step": 194600
    },
    {
      "epoch": 1.2481680899712346,
      "grad_norm": 0.05532994121313095,
      "learning_rate": 3.7213373808506925e-06,
      "loss": 0.001,
      "step": 194610
    },
    {
      "epoch": 1.2482322268650208,
      "grad_norm": 0.20678691565990448,
      "learning_rate": 3.720796296982251e-06,
      "loss": 0.0019,
      "step": 194620
    },
    {
      "epoch": 1.2482963637588067,
      "grad_norm": 0.041809648275375366,
      "learning_rate": 3.720255229143146e-06,
      "loss": 0.0011,
      "step": 194630
    },
    {
      "epoch": 1.2483605006525929,
      "grad_norm": 0.04943133518099785,
      "learning_rate": 3.719714177340157e-06,
      "loss": 0.0019,
      "step": 194640
    },
    {
      "epoch": 1.248424637546379,
      "grad_norm": 0.11647520959377289,
      "learning_rate": 3.719173141580065e-06,
      "loss": 0.0031,
      "step": 194650
    },
    {
      "epoch": 1.2484887744401651,
      "grad_norm": 0.10881810635328293,
      "learning_rate": 3.718632121869648e-06,
      "loss": 0.0015,
      "step": 194660
    },
    {
      "epoch": 1.248552911333951,
      "grad_norm": 0.07503806054592133,
      "learning_rate": 3.7180911182156874e-06,
      "loss": 0.0011,
      "step": 194670
    },
    {
      "epoch": 1.2486170482277372,
      "grad_norm": 0.113071009516716,
      "learning_rate": 3.7175501306249607e-06,
      "loss": 0.0009,
      "step": 194680
    },
    {
      "epoch": 1.2486811851215234,
      "grad_norm": 0.017294185236096382,
      "learning_rate": 3.7170091591042478e-06,
      "loss": 0.0017,
      "step": 194690
    },
    {
      "epoch": 1.2487453220153095,
      "grad_norm": 0.01370396837592125,
      "learning_rate": 3.716468203660326e-06,
      "loss": 0.0017,
      "step": 194700
    },
    {
      "epoch": 1.2488094589090957,
      "grad_norm": 0.11356718838214874,
      "learning_rate": 3.715927264299977e-06,
      "loss": 0.0006,
      "step": 194710
    },
    {
      "epoch": 1.2488735958028816,
      "grad_norm": 0.1287156343460083,
      "learning_rate": 3.7153863410299747e-06,
      "loss": 0.0013,
      "step": 194720
    },
    {
      "epoch": 1.2489377326966677,
      "grad_norm": 0.04905663803219795,
      "learning_rate": 3.7148454338571e-06,
      "loss": 0.0012,
      "step": 194730
    },
    {
      "epoch": 1.249001869590454,
      "grad_norm": 0.2283930629491806,
      "learning_rate": 3.714304542788132e-06,
      "loss": 0.0018,
      "step": 194740
    },
    {
      "epoch": 1.24906600648424,
      "grad_norm": 0.05940965563058853,
      "learning_rate": 3.7137636678298457e-06,
      "loss": 0.0017,
      "step": 194750
    },
    {
      "epoch": 1.249130143378026,
      "grad_norm": 0.07616443932056427,
      "learning_rate": 3.713222808989021e-06,
      "loss": 0.0011,
      "step": 194760
    },
    {
      "epoch": 1.2491942802718121,
      "grad_norm": 0.19304047524929047,
      "learning_rate": 3.712681966272433e-06,
      "loss": 0.0023,
      "step": 194770
    },
    {
      "epoch": 1.2492584171655983,
      "grad_norm": 0.008352248929440975,
      "learning_rate": 3.7121411396868614e-06,
      "loss": 0.0026,
      "step": 194780
    },
    {
      "epoch": 1.2493225540593844,
      "grad_norm": 0.0745990201830864,
      "learning_rate": 3.71160032923908e-06,
      "loss": 0.0015,
      "step": 194790
    },
    {
      "epoch": 1.2493866909531706,
      "grad_norm": 0.029070056974887848,
      "learning_rate": 3.7110595349358684e-06,
      "loss": 0.0013,
      "step": 194800
    },
    {
      "epoch": 1.2494508278469565,
      "grad_norm": 0.17112939059734344,
      "learning_rate": 3.7105187567840016e-06,
      "loss": 0.0012,
      "step": 194810
    },
    {
      "epoch": 1.2495149647407426,
      "grad_norm": 0.0685807392001152,
      "learning_rate": 3.709977994790257e-06,
      "loss": 0.0014,
      "step": 194820
    },
    {
      "epoch": 1.2495791016345288,
      "grad_norm": 0.11804874986410141,
      "learning_rate": 3.7094372489614087e-06,
      "loss": 0.0022,
      "step": 194830
    },
    {
      "epoch": 1.2496432385283147,
      "grad_norm": 0.02202121540904045,
      "learning_rate": 3.708896519304236e-06,
      "loss": 0.0013,
      "step": 194840
    },
    {
      "epoch": 1.2497073754221009,
      "grad_norm": 0.2399996519088745,
      "learning_rate": 3.7083558058255107e-06,
      "loss": 0.0017,
      "step": 194850
    },
    {
      "epoch": 1.249771512315887,
      "grad_norm": 0.07525387406349182,
      "learning_rate": 3.7078151085320117e-06,
      "loss": 0.0018,
      "step": 194860
    },
    {
      "epoch": 1.2498356492096732,
      "grad_norm": 0.09805445373058319,
      "learning_rate": 3.707274427430512e-06,
      "loss": 0.0014,
      "step": 194870
    },
    {
      "epoch": 1.2498997861034593,
      "grad_norm": 0.14419735968112946,
      "learning_rate": 3.706733762527788e-06,
      "loss": 0.0025,
      "step": 194880
    },
    {
      "epoch": 1.2499639229972452,
      "grad_norm": 0.11062031984329224,
      "learning_rate": 3.706193113830614e-06,
      "loss": 0.001,
      "step": 194890
    },
    {
      "epoch": 1.2500280598910314,
      "grad_norm": 0.041483424603939056,
      "learning_rate": 3.705652481345765e-06,
      "loss": 0.0012,
      "step": 194900
    },
    {
      "epoch": 1.2500921967848175,
      "grad_norm": 0.08496949076652527,
      "learning_rate": 3.705111865080016e-06,
      "loss": 0.0011,
      "step": 194910
    },
    {
      "epoch": 1.2501563336786037,
      "grad_norm": 0.050424862653017044,
      "learning_rate": 3.70457126504014e-06,
      "loss": 0.0029,
      "step": 194920
    },
    {
      "epoch": 1.2502204705723896,
      "grad_norm": 0.15065942704677582,
      "learning_rate": 3.704030681232913e-06,
      "loss": 0.0016,
      "step": 194930
    },
    {
      "epoch": 1.2502846074661758,
      "grad_norm": 0.07603470981121063,
      "learning_rate": 3.703490113665106e-06,
      "loss": 0.0026,
      "step": 194940
    },
    {
      "epoch": 1.250348744359962,
      "grad_norm": 0.1060481071472168,
      "learning_rate": 3.702949562343496e-06,
      "loss": 0.0011,
      "step": 194950
    },
    {
      "epoch": 1.250412881253748,
      "grad_norm": 0.12667687237262726,
      "learning_rate": 3.7024090272748547e-06,
      "loss": 0.0019,
      "step": 194960
    },
    {
      "epoch": 1.2504770181475342,
      "grad_norm": 0.04505964368581772,
      "learning_rate": 3.7018685084659556e-06,
      "loss": 0.0009,
      "step": 194970
    },
    {
      "epoch": 1.2505411550413201,
      "grad_norm": 0.0329747311770916,
      "learning_rate": 3.701328005923571e-06,
      "loss": 0.002,
      "step": 194980
    },
    {
      "epoch": 1.2506052919351063,
      "grad_norm": 0.049436502158641815,
      "learning_rate": 3.700787519654476e-06,
      "loss": 0.0012,
      "step": 194990
    },
    {
      "epoch": 1.2506694288288924,
      "grad_norm": 0.06754521280527115,
      "learning_rate": 3.700247049665441e-06,
      "loss": 0.0018,
      "step": 195000
    },
    {
      "epoch": 1.2507335657226784,
      "grad_norm": 0.028261858969926834,
      "learning_rate": 3.6997065959632406e-06,
      "loss": 0.0007,
      "step": 195010
    },
    {
      "epoch": 1.2507977026164645,
      "grad_norm": 0.0627891793847084,
      "learning_rate": 3.699166158554644e-06,
      "loss": 0.0009,
      "step": 195020
    },
    {
      "epoch": 1.2508618395102507,
      "grad_norm": 0.06759367138147354,
      "learning_rate": 3.698625737446426e-06,
      "loss": 0.0011,
      "step": 195030
    },
    {
      "epoch": 1.2509259764040368,
      "grad_norm": 0.05140414461493492,
      "learning_rate": 3.698085332645358e-06,
      "loss": 0.0014,
      "step": 195040
    },
    {
      "epoch": 1.250990113297823,
      "grad_norm": 0.11699049174785614,
      "learning_rate": 3.6975449441582102e-06,
      "loss": 0.0011,
      "step": 195050
    },
    {
      "epoch": 1.251054250191609,
      "grad_norm": 0.1346450001001358,
      "learning_rate": 3.6970045719917562e-06,
      "loss": 0.0014,
      "step": 195060
    },
    {
      "epoch": 1.251118387085395,
      "grad_norm": 0.04619720205664635,
      "learning_rate": 3.6964642161527652e-06,
      "loss": 0.0009,
      "step": 195070
    },
    {
      "epoch": 1.2511825239791812,
      "grad_norm": 0.030455434694886208,
      "learning_rate": 3.6959238766480105e-06,
      "loss": 0.0018,
      "step": 195080
    },
    {
      "epoch": 1.2512466608729673,
      "grad_norm": 0.0943940281867981,
      "learning_rate": 3.6953835534842603e-06,
      "loss": 0.0015,
      "step": 195090
    },
    {
      "epoch": 1.2513107977667532,
      "grad_norm": 0.13932521641254425,
      "learning_rate": 3.6948432466682875e-06,
      "loss": 0.0009,
      "step": 195100
    },
    {
      "epoch": 1.2513749346605394,
      "grad_norm": 0.15518449246883392,
      "learning_rate": 3.694302956206861e-06,
      "loss": 0.0012,
      "step": 195110
    },
    {
      "epoch": 1.2514390715543255,
      "grad_norm": 0.04630433768033981,
      "learning_rate": 3.6937626821067525e-06,
      "loss": 0.0019,
      "step": 195120
    },
    {
      "epoch": 1.2515032084481117,
      "grad_norm": 0.044243235141038895,
      "learning_rate": 3.6932224243747296e-06,
      "loss": 0.0011,
      "step": 195130
    },
    {
      "epoch": 1.2515673453418978,
      "grad_norm": 0.11855106055736542,
      "learning_rate": 3.692682183017565e-06,
      "loss": 0.0012,
      "step": 195140
    },
    {
      "epoch": 1.2516314822356838,
      "grad_norm": 0.07241770625114441,
      "learning_rate": 3.6921419580420265e-06,
      "loss": 0.0008,
      "step": 195150
    },
    {
      "epoch": 1.25169561912947,
      "grad_norm": 0.06299462169408798,
      "learning_rate": 3.691601749454883e-06,
      "loss": 0.0029,
      "step": 195160
    },
    {
      "epoch": 1.251759756023256,
      "grad_norm": 0.04221626743674278,
      "learning_rate": 3.6910615572629066e-06,
      "loss": 0.0012,
      "step": 195170
    },
    {
      "epoch": 1.2518238929170422,
      "grad_norm": 0.09005774557590485,
      "learning_rate": 3.690521381472863e-06,
      "loss": 0.0014,
      "step": 195180
    },
    {
      "epoch": 1.2518880298108281,
      "grad_norm": 0.07019791007041931,
      "learning_rate": 3.6899812220915233e-06,
      "loss": 0.0006,
      "step": 195190
    },
    {
      "epoch": 1.2519521667046143,
      "grad_norm": 0.07578397542238235,
      "learning_rate": 3.689441079125654e-06,
      "loss": 0.0019,
      "step": 195200
    },
    {
      "epoch": 1.2520163035984004,
      "grad_norm": 0.08820551633834839,
      "learning_rate": 3.6889009525820264e-06,
      "loss": 0.0022,
      "step": 195210
    },
    {
      "epoch": 1.2520804404921866,
      "grad_norm": 0.03909339755773544,
      "learning_rate": 3.688360842467405e-06,
      "loss": 0.0013,
      "step": 195220
    },
    {
      "epoch": 1.2521445773859727,
      "grad_norm": 0.04139334708452225,
      "learning_rate": 3.687820748788561e-06,
      "loss": 0.0007,
      "step": 195230
    },
    {
      "epoch": 1.2522087142797587,
      "grad_norm": 0.07054650038480759,
      "learning_rate": 3.6872806715522597e-06,
      "loss": 0.0043,
      "step": 195240
    },
    {
      "epoch": 1.2522728511735448,
      "grad_norm": 0.25521770119667053,
      "learning_rate": 3.686740610765271e-06,
      "loss": 0.0026,
      "step": 195250
    },
    {
      "epoch": 1.252336988067331,
      "grad_norm": 0.28910234570503235,
      "learning_rate": 3.68620056643436e-06,
      "loss": 0.0015,
      "step": 195260
    },
    {
      "epoch": 1.2524011249611169,
      "grad_norm": 0.005549044813960791,
      "learning_rate": 3.6856605385662957e-06,
      "loss": 0.0017,
      "step": 195270
    },
    {
      "epoch": 1.252465261854903,
      "grad_norm": 0.17769359052181244,
      "learning_rate": 3.685120527167843e-06,
      "loss": 0.0013,
      "step": 195280
    },
    {
      "epoch": 1.2525293987486892,
      "grad_norm": 0.036505963653326035,
      "learning_rate": 3.684580532245771e-06,
      "loss": 0.0016,
      "step": 195290
    },
    {
      "epoch": 1.2525935356424753,
      "grad_norm": 0.05288667976856232,
      "learning_rate": 3.684040553806844e-06,
      "loss": 0.001,
      "step": 195300
    },
    {
      "epoch": 1.2526576725362615,
      "grad_norm": 0.005735976621508598,
      "learning_rate": 3.6835005918578293e-06,
      "loss": 0.0023,
      "step": 195310
    },
    {
      "epoch": 1.2527218094300476,
      "grad_norm": 0.15252001583576202,
      "learning_rate": 3.6829606464054945e-06,
      "loss": 0.002,
      "step": 195320
    },
    {
      "epoch": 1.2527859463238336,
      "grad_norm": 0.074702188372612,
      "learning_rate": 3.6824207174566018e-06,
      "loss": 0.0012,
      "step": 195330
    },
    {
      "epoch": 1.2528500832176197,
      "grad_norm": 0.06723365187644958,
      "learning_rate": 3.68188080501792e-06,
      "loss": 0.0013,
      "step": 195340
    },
    {
      "epoch": 1.2529142201114059,
      "grad_norm": 0.09764128923416138,
      "learning_rate": 3.6813409090962148e-06,
      "loss": 0.0018,
      "step": 195350
    },
    {
      "epoch": 1.2529783570051918,
      "grad_norm": 0.02114538475871086,
      "learning_rate": 3.68080102969825e-06,
      "loss": 0.0025,
      "step": 195360
    },
    {
      "epoch": 1.253042493898978,
      "grad_norm": 0.10238329321146011,
      "learning_rate": 3.680261166830791e-06,
      "loss": 0.0011,
      "step": 195370
    },
    {
      "epoch": 1.253106630792764,
      "grad_norm": 0.29062432050704956,
      "learning_rate": 3.679721320500603e-06,
      "loss": 0.0011,
      "step": 195380
    },
    {
      "epoch": 1.2531707676865502,
      "grad_norm": 0.129994198679924,
      "learning_rate": 3.6791814907144505e-06,
      "loss": 0.002,
      "step": 195390
    },
    {
      "epoch": 1.2532349045803364,
      "grad_norm": 0.04898214712738991,
      "learning_rate": 3.678641677479098e-06,
      "loss": 0.0009,
      "step": 195400
    },
    {
      "epoch": 1.2532990414741223,
      "grad_norm": 0.08681108057498932,
      "learning_rate": 3.6781018808013097e-06,
      "loss": 0.0017,
      "step": 195410
    },
    {
      "epoch": 1.2533631783679084,
      "grad_norm": 0.07974491268396378,
      "learning_rate": 3.6775621006878487e-06,
      "loss": 0.0018,
      "step": 195420
    },
    {
      "epoch": 1.2534273152616946,
      "grad_norm": 0.05810292437672615,
      "learning_rate": 3.6770223371454797e-06,
      "loss": 0.0018,
      "step": 195430
    },
    {
      "epoch": 1.2534914521554805,
      "grad_norm": 0.021946711465716362,
      "learning_rate": 3.6764825901809673e-06,
      "loss": 0.0015,
      "step": 195440
    },
    {
      "epoch": 1.2535555890492667,
      "grad_norm": 0.1336854249238968,
      "learning_rate": 3.6759428598010738e-06,
      "loss": 0.0011,
      "step": 195450
    },
    {
      "epoch": 1.2536197259430528,
      "grad_norm": 0.07914441078901291,
      "learning_rate": 3.675403146012563e-06,
      "loss": 0.0007,
      "step": 195460
    },
    {
      "epoch": 1.253683862836839,
      "grad_norm": 0.05006599798798561,
      "learning_rate": 3.6748634488221955e-06,
      "loss": 0.0013,
      "step": 195470
    },
    {
      "epoch": 1.2537479997306251,
      "grad_norm": 0.14474913477897644,
      "learning_rate": 3.6743237682367385e-06,
      "loss": 0.0012,
      "step": 195480
    },
    {
      "epoch": 1.2538121366244113,
      "grad_norm": 0.03993711993098259,
      "learning_rate": 3.67378410426295e-06,
      "loss": 0.0009,
      "step": 195490
    },
    {
      "epoch": 1.2538762735181972,
      "grad_norm": 0.002169192535802722,
      "learning_rate": 3.6732444569075964e-06,
      "loss": 0.0012,
      "step": 195500
    },
    {
      "epoch": 1.2539404104119833,
      "grad_norm": 0.011117948219180107,
      "learning_rate": 3.672704826177437e-06,
      "loss": 0.002,
      "step": 195510
    },
    {
      "epoch": 1.2540045473057695,
      "grad_norm": 0.026117559522390366,
      "learning_rate": 3.672165212079235e-06,
      "loss": 0.0014,
      "step": 195520
    },
    {
      "epoch": 1.2540686841995554,
      "grad_norm": 0.06564907729625702,
      "learning_rate": 3.671625614619751e-06,
      "loss": 0.0009,
      "step": 195530
    },
    {
      "epoch": 1.2541328210933416,
      "grad_norm": 0.3652568459510803,
      "learning_rate": 3.6710860338057485e-06,
      "loss": 0.0027,
      "step": 195540
    },
    {
      "epoch": 1.2541969579871277,
      "grad_norm": 0.08847367018461227,
      "learning_rate": 3.6705464696439875e-06,
      "loss": 0.0015,
      "step": 195550
    },
    {
      "epoch": 1.2542610948809139,
      "grad_norm": 0.0992204025387764,
      "learning_rate": 3.6700069221412295e-06,
      "loss": 0.0013,
      "step": 195560
    },
    {
      "epoch": 1.2543252317747,
      "grad_norm": 0.03547690063714981,
      "learning_rate": 3.6694673913042345e-06,
      "loss": 0.0017,
      "step": 195570
    },
    {
      "epoch": 1.254389368668486,
      "grad_norm": 0.05089672654867172,
      "learning_rate": 3.668927877139764e-06,
      "loss": 0.0013,
      "step": 195580
    },
    {
      "epoch": 1.254453505562272,
      "grad_norm": 0.015458229929208755,
      "learning_rate": 3.668388379654579e-06,
      "loss": 0.0021,
      "step": 195590
    },
    {
      "epoch": 1.2545176424560582,
      "grad_norm": 0.1813635677099228,
      "learning_rate": 3.6678488988554393e-06,
      "loss": 0.0027,
      "step": 195600
    },
    {
      "epoch": 1.2545817793498444,
      "grad_norm": 0.0283439289778471,
      "learning_rate": 3.6673094347491055e-06,
      "loss": 0.0012,
      "step": 195610
    },
    {
      "epoch": 1.2546459162436303,
      "grad_norm": 0.06188954412937164,
      "learning_rate": 3.6667699873423363e-06,
      "loss": 0.0014,
      "step": 195620
    },
    {
      "epoch": 1.2547100531374165,
      "grad_norm": 0.0191205944865942,
      "learning_rate": 3.6662305566418926e-06,
      "loss": 0.0011,
      "step": 195630
    },
    {
      "epoch": 1.2547741900312026,
      "grad_norm": 0.018281230702996254,
      "learning_rate": 3.665691142654532e-06,
      "loss": 0.002,
      "step": 195640
    },
    {
      "epoch": 1.2548383269249888,
      "grad_norm": 0.06265388429164886,
      "learning_rate": 3.665151745387017e-06,
      "loss": 0.0017,
      "step": 195650
    },
    {
      "epoch": 1.254902463818775,
      "grad_norm": 0.19270409643650055,
      "learning_rate": 3.664612364846103e-06,
      "loss": 0.0025,
      "step": 195660
    },
    {
      "epoch": 1.2549666007125608,
      "grad_norm": 0.033852968364953995,
      "learning_rate": 3.664073001038551e-06,
      "loss": 0.0015,
      "step": 195670
    },
    {
      "epoch": 1.255030737606347,
      "grad_norm": 0.00962325744330883,
      "learning_rate": 3.6635336539711188e-06,
      "loss": 0.0013,
      "step": 195680
    },
    {
      "epoch": 1.2550948745001331,
      "grad_norm": 0.15461745858192444,
      "learning_rate": 3.6629943236505655e-06,
      "loss": 0.0015,
      "step": 195690
    },
    {
      "epoch": 1.255159011393919,
      "grad_norm": 0.03359571844339371,
      "learning_rate": 3.662455010083648e-06,
      "loss": 0.0016,
      "step": 195700
    },
    {
      "epoch": 1.2552231482877052,
      "grad_norm": 0.11851087957620621,
      "learning_rate": 3.6619157132771265e-06,
      "loss": 0.001,
      "step": 195710
    },
    {
      "epoch": 1.2552872851814914,
      "grad_norm": 0.014558130875229836,
      "learning_rate": 3.661376433237756e-06,
      "loss": 0.0042,
      "step": 195720
    },
    {
      "epoch": 1.2553514220752775,
      "grad_norm": 0.039042163640260696,
      "learning_rate": 3.6608371699722954e-06,
      "loss": 0.001,
      "step": 195730
    },
    {
      "epoch": 1.2554155589690636,
      "grad_norm": 0.05535927414894104,
      "learning_rate": 3.6602979234875036e-06,
      "loss": 0.0019,
      "step": 195740
    },
    {
      "epoch": 1.2554796958628498,
      "grad_norm": 0.03454405069351196,
      "learning_rate": 3.6597586937901353e-06,
      "loss": 0.0049,
      "step": 195750
    },
    {
      "epoch": 1.2555438327566357,
      "grad_norm": 0.008221879601478577,
      "learning_rate": 3.6592194808869498e-06,
      "loss": 0.0011,
      "step": 195760
    },
    {
      "epoch": 1.2556079696504219,
      "grad_norm": 0.004276182036846876,
      "learning_rate": 3.6586802847847e-06,
      "loss": 0.0011,
      "step": 195770
    },
    {
      "epoch": 1.255672106544208,
      "grad_norm": 0.10540279000997543,
      "learning_rate": 3.658141105490147e-06,
      "loss": 0.0024,
      "step": 195780
    },
    {
      "epoch": 1.255736243437994,
      "grad_norm": 0.16785524785518646,
      "learning_rate": 3.657601943010044e-06,
      "loss": 0.0016,
      "step": 195790
    },
    {
      "epoch": 1.25580038033178,
      "grad_norm": 0.1500561386346817,
      "learning_rate": 3.6570627973511487e-06,
      "loss": 0.001,
      "step": 195800
    },
    {
      "epoch": 1.2558645172255662,
      "grad_norm": 0.15236826241016388,
      "learning_rate": 3.6565236685202154e-06,
      "loss": 0.0018,
      "step": 195810
    },
    {
      "epoch": 1.2559286541193524,
      "grad_norm": 0.17366255819797516,
      "learning_rate": 3.6559845565240014e-06,
      "loss": 0.002,
      "step": 195820
    },
    {
      "epoch": 1.2559927910131385,
      "grad_norm": 0.08815667033195496,
      "learning_rate": 3.6554454613692613e-06,
      "loss": 0.0019,
      "step": 195830
    },
    {
      "epoch": 1.2560569279069245,
      "grad_norm": 0.00293713784776628,
      "learning_rate": 3.6549063830627506e-06,
      "loss": 0.0029,
      "step": 195840
    },
    {
      "epoch": 1.2561210648007106,
      "grad_norm": 0.04876834899187088,
      "learning_rate": 3.654367321611224e-06,
      "loss": 0.0018,
      "step": 195850
    },
    {
      "epoch": 1.2561852016944968,
      "grad_norm": 0.04722928628325462,
      "learning_rate": 3.653828277021437e-06,
      "loss": 0.0018,
      "step": 195860
    },
    {
      "epoch": 1.2562493385882827,
      "grad_norm": 0.002980181248858571,
      "learning_rate": 3.6532892493001433e-06,
      "loss": 0.002,
      "step": 195870
    },
    {
      "epoch": 1.2563134754820688,
      "grad_norm": 0.04678097739815712,
      "learning_rate": 3.652750238454097e-06,
      "loss": 0.0012,
      "step": 195880
    },
    {
      "epoch": 1.256377612375855,
      "grad_norm": 0.11403562128543854,
      "learning_rate": 3.652211244490055e-06,
      "loss": 0.0019,
      "step": 195890
    },
    {
      "epoch": 1.2564417492696411,
      "grad_norm": 0.08019541203975677,
      "learning_rate": 3.6516722674147675e-06,
      "loss": 0.0017,
      "step": 195900
    },
    {
      "epoch": 1.2565058861634273,
      "grad_norm": 0.3112826347351074,
      "learning_rate": 3.651133307234992e-06,
      "loss": 0.0016,
      "step": 195910
    },
    {
      "epoch": 1.2565700230572134,
      "grad_norm": 0.07466558367013931,
      "learning_rate": 3.6505943639574782e-06,
      "loss": 0.0017,
      "step": 195920
    },
    {
      "epoch": 1.2566341599509994,
      "grad_norm": 0.06018274277448654,
      "learning_rate": 3.650055437588983e-06,
      "loss": 0.0025,
      "step": 195930
    },
    {
      "epoch": 1.2566982968447855,
      "grad_norm": 0.14753106236457825,
      "learning_rate": 3.6495165281362573e-06,
      "loss": 0.0019,
      "step": 195940
    },
    {
      "epoch": 1.2567624337385717,
      "grad_norm": 0.09227093309164047,
      "learning_rate": 3.6489776356060546e-06,
      "loss": 0.0014,
      "step": 195950
    },
    {
      "epoch": 1.2568265706323576,
      "grad_norm": 0.019015872851014137,
      "learning_rate": 3.6484387600051276e-06,
      "loss": 0.0013,
      "step": 195960
    },
    {
      "epoch": 1.2568907075261437,
      "grad_norm": 0.006597777362912893,
      "learning_rate": 3.64789990134023e-06,
      "loss": 0.001,
      "step": 195970
    },
    {
      "epoch": 1.2569548444199299,
      "grad_norm": 0.0689472034573555,
      "learning_rate": 3.647361059618111e-06,
      "loss": 0.0021,
      "step": 195980
    },
    {
      "epoch": 1.257018981313716,
      "grad_norm": 0.12803134322166443,
      "learning_rate": 3.646822234845527e-06,
      "loss": 0.0016,
      "step": 195990
    },
    {
      "epoch": 1.2570831182075022,
      "grad_norm": 0.05919404327869415,
      "learning_rate": 3.646283427029226e-06,
      "loss": 0.0024,
      "step": 196000
    },
    {
      "epoch": 1.257147255101288,
      "grad_norm": 0.2608695328235626,
      "learning_rate": 3.6457446361759603e-06,
      "loss": 0.0022,
      "step": 196010
    },
    {
      "epoch": 1.2572113919950743,
      "grad_norm": 0.02095581591129303,
      "learning_rate": 3.6452058622924836e-06,
      "loss": 0.0016,
      "step": 196020
    },
    {
      "epoch": 1.2572755288888604,
      "grad_norm": 0.1533578336238861,
      "learning_rate": 3.6446671053855453e-06,
      "loss": 0.0015,
      "step": 196030
    },
    {
      "epoch": 1.2573396657826466,
      "grad_norm": 0.13423244655132294,
      "learning_rate": 3.6441283654618975e-06,
      "loss": 0.0016,
      "step": 196040
    },
    {
      "epoch": 1.2574038026764325,
      "grad_norm": 0.09977749735116959,
      "learning_rate": 3.643589642528289e-06,
      "loss": 0.0025,
      "step": 196050
    },
    {
      "epoch": 1.2574679395702186,
      "grad_norm": 0.10623639076948166,
      "learning_rate": 3.6430509365914723e-06,
      "loss": 0.0012,
      "step": 196060
    },
    {
      "epoch": 1.2575320764640048,
      "grad_norm": 0.08471555262804031,
      "learning_rate": 3.642512247658197e-06,
      "loss": 0.0018,
      "step": 196070
    },
    {
      "epoch": 1.257596213357791,
      "grad_norm": 0.043866001069545746,
      "learning_rate": 3.6419735757352144e-06,
      "loss": 0.0016,
      "step": 196080
    },
    {
      "epoch": 1.257660350251577,
      "grad_norm": 0.027464529499411583,
      "learning_rate": 3.641434920829272e-06,
      "loss": 0.0015,
      "step": 196090
    },
    {
      "epoch": 1.257724487145363,
      "grad_norm": 0.0067048026248812675,
      "learning_rate": 3.6408962829471205e-06,
      "loss": 0.0012,
      "step": 196100
    },
    {
      "epoch": 1.2577886240391492,
      "grad_norm": 0.007362800184637308,
      "learning_rate": 3.6403576620955117e-06,
      "loss": 0.0015,
      "step": 196110
    },
    {
      "epoch": 1.2578527609329353,
      "grad_norm": 0.10874085128307343,
      "learning_rate": 3.6398190582811922e-06,
      "loss": 0.0023,
      "step": 196120
    },
    {
      "epoch": 1.2579168978267212,
      "grad_norm": 0.05695396289229393,
      "learning_rate": 3.639280471510912e-06,
      "loss": 0.0011,
      "step": 196130
    },
    {
      "epoch": 1.2579810347205074,
      "grad_norm": 0.02212439477443695,
      "learning_rate": 3.6387419017914204e-06,
      "loss": 0.0017,
      "step": 196140
    },
    {
      "epoch": 1.2580451716142935,
      "grad_norm": 0.04459254443645477,
      "learning_rate": 3.638203349129465e-06,
      "loss": 0.0011,
      "step": 196150
    },
    {
      "epoch": 1.2581093085080797,
      "grad_norm": 0.10103648900985718,
      "learning_rate": 3.637664813531796e-06,
      "loss": 0.001,
      "step": 196160
    },
    {
      "epoch": 1.2581734454018658,
      "grad_norm": 0.04929735139012337,
      "learning_rate": 3.6371262950051593e-06,
      "loss": 0.0016,
      "step": 196170
    },
    {
      "epoch": 1.258237582295652,
      "grad_norm": 0.060934390872716904,
      "learning_rate": 3.6365877935563055e-06,
      "loss": 0.0018,
      "step": 196180
    },
    {
      "epoch": 1.258301719189438,
      "grad_norm": 0.007423396687954664,
      "learning_rate": 3.63604930919198e-06,
      "loss": 0.0008,
      "step": 196190
    },
    {
      "epoch": 1.258365856083224,
      "grad_norm": 0.039432283490896225,
      "learning_rate": 3.635510841918932e-06,
      "loss": 0.0013,
      "step": 196200
    },
    {
      "epoch": 1.2584299929770102,
      "grad_norm": 0.003703708527609706,
      "learning_rate": 3.634972391743908e-06,
      "loss": 0.001,
      "step": 196210
    },
    {
      "epoch": 1.2584941298707961,
      "grad_norm": 0.17724007368087769,
      "learning_rate": 3.6344339586736568e-06,
      "loss": 0.002,
      "step": 196220
    },
    {
      "epoch": 1.2585582667645823,
      "grad_norm": 0.11132031679153442,
      "learning_rate": 3.633895542714922e-06,
      "loss": 0.0013,
      "step": 196230
    },
    {
      "epoch": 1.2586224036583684,
      "grad_norm": 0.11109644174575806,
      "learning_rate": 3.6333571438744542e-06,
      "loss": 0.0012,
      "step": 196240
    },
    {
      "epoch": 1.2586865405521546,
      "grad_norm": 0.08999479562044144,
      "learning_rate": 3.632818762158996e-06,
      "loss": 0.0026,
      "step": 196250
    },
    {
      "epoch": 1.2587506774459407,
      "grad_norm": 0.08881781250238419,
      "learning_rate": 3.632280397575298e-06,
      "loss": 0.0019,
      "step": 196260
    },
    {
      "epoch": 1.2588148143397266,
      "grad_norm": 0.0790308266878128,
      "learning_rate": 3.6317420501301027e-06,
      "loss": 0.0022,
      "step": 196270
    },
    {
      "epoch": 1.2588789512335128,
      "grad_norm": 0.034874871373176575,
      "learning_rate": 3.6312037198301576e-06,
      "loss": 0.0019,
      "step": 196280
    },
    {
      "epoch": 1.258943088127299,
      "grad_norm": 0.08324344456195831,
      "learning_rate": 3.6306654066822087e-06,
      "loss": 0.001,
      "step": 196290
    },
    {
      "epoch": 1.259007225021085,
      "grad_norm": 0.009703489020466805,
      "learning_rate": 3.630127110693e-06,
      "loss": 0.0012,
      "step": 196300
    },
    {
      "epoch": 1.259071361914871,
      "grad_norm": 0.3019203245639801,
      "learning_rate": 3.629588831869279e-06,
      "loss": 0.0013,
      "step": 196310
    },
    {
      "epoch": 1.2591354988086572,
      "grad_norm": 0.110463447868824,
      "learning_rate": 3.629050570217788e-06,
      "loss": 0.0063,
      "step": 196320
    },
    {
      "epoch": 1.2591996357024433,
      "grad_norm": 0.0900101438164711,
      "learning_rate": 3.628512325745275e-06,
      "loss": 0.0014,
      "step": 196330
    },
    {
      "epoch": 1.2592637725962295,
      "grad_norm": 0.11893414705991745,
      "learning_rate": 3.6279740984584804e-06,
      "loss": 0.0016,
      "step": 196340
    },
    {
      "epoch": 1.2593279094900156,
      "grad_norm": 0.03216111287474632,
      "learning_rate": 3.6274358883641525e-06,
      "loss": 0.002,
      "step": 196350
    },
    {
      "epoch": 1.2593920463838015,
      "grad_norm": 0.12460221350193024,
      "learning_rate": 3.626897695469033e-06,
      "loss": 0.0019,
      "step": 196360
    },
    {
      "epoch": 1.2594561832775877,
      "grad_norm": 0.08771844208240509,
      "learning_rate": 3.626359519779868e-06,
      "loss": 0.0012,
      "step": 196370
    },
    {
      "epoch": 1.2595203201713738,
      "grad_norm": 0.07012537121772766,
      "learning_rate": 3.6258213613033983e-06,
      "loss": 0.0008,
      "step": 196380
    },
    {
      "epoch": 1.2595844570651598,
      "grad_norm": 0.051890429109334946,
      "learning_rate": 3.6252832200463708e-06,
      "loss": 0.0014,
      "step": 196390
    },
    {
      "epoch": 1.259648593958946,
      "grad_norm": 0.06170021370053291,
      "learning_rate": 3.624745096015525e-06,
      "loss": 0.0009,
      "step": 196400
    },
    {
      "epoch": 1.259712730852732,
      "grad_norm": 0.17516081035137177,
      "learning_rate": 3.624206989217608e-06,
      "loss": 0.0023,
      "step": 196410
    },
    {
      "epoch": 1.2597768677465182,
      "grad_norm": 0.09847494959831238,
      "learning_rate": 3.6236688996593593e-06,
      "loss": 0.0012,
      "step": 196420
    },
    {
      "epoch": 1.2598410046403044,
      "grad_norm": 0.07129335403442383,
      "learning_rate": 3.623130827347523e-06,
      "loss": 0.0009,
      "step": 196430
    },
    {
      "epoch": 1.2599051415340905,
      "grad_norm": 0.010882940143346786,
      "learning_rate": 3.622592772288842e-06,
      "loss": 0.0012,
      "step": 196440
    },
    {
      "epoch": 1.2599692784278764,
      "grad_norm": 0.22375863790512085,
      "learning_rate": 3.622054734490058e-06,
      "loss": 0.0025,
      "step": 196450
    },
    {
      "epoch": 1.2600334153216626,
      "grad_norm": 0.1855129450559616,
      "learning_rate": 3.6215167139579133e-06,
      "loss": 0.0033,
      "step": 196460
    },
    {
      "epoch": 1.2600975522154487,
      "grad_norm": 0.08187199383974075,
      "learning_rate": 3.620978710699148e-06,
      "loss": 0.002,
      "step": 196470
    },
    {
      "epoch": 1.2601616891092347,
      "grad_norm": 0.0629785805940628,
      "learning_rate": 3.6204407247205066e-06,
      "loss": 0.0013,
      "step": 196480
    },
    {
      "epoch": 1.2602258260030208,
      "grad_norm": 0.10981062799692154,
      "learning_rate": 3.619902756028728e-06,
      "loss": 0.0026,
      "step": 196490
    },
    {
      "epoch": 1.260289962896807,
      "grad_norm": 0.015897827222943306,
      "learning_rate": 3.6193648046305553e-06,
      "loss": 0.0015,
      "step": 196500
    },
    {
      "epoch": 1.260354099790593,
      "grad_norm": 0.10848476737737656,
      "learning_rate": 3.618826870532727e-06,
      "loss": 0.0021,
      "step": 196510
    },
    {
      "epoch": 1.2604182366843792,
      "grad_norm": 0.023016218096017838,
      "learning_rate": 3.618288953741986e-06,
      "loss": 0.0024,
      "step": 196520
    },
    {
      "epoch": 1.2604823735781652,
      "grad_norm": 0.0073949359357357025,
      "learning_rate": 3.6177510542650705e-06,
      "loss": 0.0027,
      "step": 196530
    },
    {
      "epoch": 1.2605465104719513,
      "grad_norm": 0.15369051694869995,
      "learning_rate": 3.617213172108724e-06,
      "loss": 0.0008,
      "step": 196540
    },
    {
      "epoch": 1.2606106473657375,
      "grad_norm": 0.3139120638370514,
      "learning_rate": 3.6166753072796834e-06,
      "loss": 0.0035,
      "step": 196550
    },
    {
      "epoch": 1.2606747842595234,
      "grad_norm": 0.09201942384243011,
      "learning_rate": 3.616137459784691e-06,
      "loss": 0.0035,
      "step": 196560
    },
    {
      "epoch": 1.2607389211533095,
      "grad_norm": 0.12248411774635315,
      "learning_rate": 3.615599629630484e-06,
      "loss": 0.0009,
      "step": 196570
    },
    {
      "epoch": 1.2608030580470957,
      "grad_norm": 0.132732555270195,
      "learning_rate": 3.615061816823803e-06,
      "loss": 0.0011,
      "step": 196580
    },
    {
      "epoch": 1.2608671949408818,
      "grad_norm": 0.038817718625068665,
      "learning_rate": 3.6145240213713884e-06,
      "loss": 0.001,
      "step": 196590
    },
    {
      "epoch": 1.260931331834668,
      "grad_norm": 0.04961353912949562,
      "learning_rate": 3.613986243279977e-06,
      "loss": 0.0015,
      "step": 196600
    },
    {
      "epoch": 1.2609954687284541,
      "grad_norm": 0.01886233128607273,
      "learning_rate": 3.61344848255631e-06,
      "loss": 0.0017,
      "step": 196610
    },
    {
      "epoch": 1.26105960562224,
      "grad_norm": 0.03975687548518181,
      "learning_rate": 3.6129107392071227e-06,
      "loss": 0.0012,
      "step": 196620
    },
    {
      "epoch": 1.2611237425160262,
      "grad_norm": 0.048187654465436935,
      "learning_rate": 3.6123730132391565e-06,
      "loss": 0.0038,
      "step": 196630
    },
    {
      "epoch": 1.2611878794098124,
      "grad_norm": 0.07477468997240067,
      "learning_rate": 3.611835304659147e-06,
      "loss": 0.002,
      "step": 196640
    },
    {
      "epoch": 1.2612520163035983,
      "grad_norm": 0.007399613503366709,
      "learning_rate": 3.6112976134738342e-06,
      "loss": 0.0014,
      "step": 196650
    },
    {
      "epoch": 1.2613161531973844,
      "grad_norm": 0.12148452550172806,
      "learning_rate": 3.610759939689954e-06,
      "loss": 0.0011,
      "step": 196660
    },
    {
      "epoch": 1.2613802900911706,
      "grad_norm": 0.09646649658679962,
      "learning_rate": 3.6102222833142452e-06,
      "loss": 0.0021,
      "step": 196670
    },
    {
      "epoch": 1.2614444269849567,
      "grad_norm": 0.013172101229429245,
      "learning_rate": 3.609684644353444e-06,
      "loss": 0.001,
      "step": 196680
    },
    {
      "epoch": 1.2615085638787429,
      "grad_norm": 0.052144456654787064,
      "learning_rate": 3.6091470228142888e-06,
      "loss": 0.0032,
      "step": 196690
    },
    {
      "epoch": 1.2615727007725288,
      "grad_norm": 0.06968370079994202,
      "learning_rate": 3.6086094187035148e-06,
      "loss": 0.0009,
      "step": 196700
    },
    {
      "epoch": 1.261636837666315,
      "grad_norm": 0.05902137607336044,
      "learning_rate": 3.6080718320278595e-06,
      "loss": 0.0012,
      "step": 196710
    },
    {
      "epoch": 1.261700974560101,
      "grad_norm": 0.13474467396736145,
      "learning_rate": 3.607534262794058e-06,
      "loss": 0.0016,
      "step": 196720
    },
    {
      "epoch": 1.2617651114538873,
      "grad_norm": 0.02709767036139965,
      "learning_rate": 3.6069967110088477e-06,
      "loss": 0.0015,
      "step": 196730
    },
    {
      "epoch": 1.2618292483476732,
      "grad_norm": 0.13237573206424713,
      "learning_rate": 3.606459176678965e-06,
      "loss": 0.0013,
      "step": 196740
    },
    {
      "epoch": 1.2618933852414593,
      "grad_norm": 0.10878362506628036,
      "learning_rate": 3.6059216598111437e-06,
      "loss": 0.001,
      "step": 196750
    },
    {
      "epoch": 1.2619575221352455,
      "grad_norm": 0.05179942399263382,
      "learning_rate": 3.6053841604121213e-06,
      "loss": 0.0016,
      "step": 196760
    },
    {
      "epoch": 1.2620216590290316,
      "grad_norm": 0.014150524511933327,
      "learning_rate": 3.6048466784886316e-06,
      "loss": 0.0037,
      "step": 196770
    },
    {
      "epoch": 1.2620857959228178,
      "grad_norm": 0.14497435092926025,
      "learning_rate": 3.6043092140474107e-06,
      "loss": 0.0012,
      "step": 196780
    },
    {
      "epoch": 1.2621499328166037,
      "grad_norm": 0.12207596004009247,
      "learning_rate": 3.6037717670951916e-06,
      "loss": 0.0019,
      "step": 196790
    },
    {
      "epoch": 1.2622140697103899,
      "grad_norm": 0.13010969758033752,
      "learning_rate": 3.603234337638711e-06,
      "loss": 0.0021,
      "step": 196800
    },
    {
      "epoch": 1.262278206604176,
      "grad_norm": 0.14584875106811523,
      "learning_rate": 3.602696925684702e-06,
      "loss": 0.0011,
      "step": 196810
    },
    {
      "epoch": 1.262342343497962,
      "grad_norm": 0.04134012758731842,
      "learning_rate": 3.6021595312399003e-06,
      "loss": 0.0008,
      "step": 196820
    },
    {
      "epoch": 1.262406480391748,
      "grad_norm": 0.03213493153452873,
      "learning_rate": 3.6016221543110373e-06,
      "loss": 0.0012,
      "step": 196830
    },
    {
      "epoch": 1.2624706172855342,
      "grad_norm": 0.061812907457351685,
      "learning_rate": 3.6010847949048487e-06,
      "loss": 0.0023,
      "step": 196840
    },
    {
      "epoch": 1.2625347541793204,
      "grad_norm": 0.004594553727656603,
      "learning_rate": 3.600547453028067e-06,
      "loss": 0.002,
      "step": 196850
    },
    {
      "epoch": 1.2625988910731065,
      "grad_norm": 0.011848662979900837,
      "learning_rate": 3.6000101286874254e-06,
      "loss": 0.002,
      "step": 196860
    },
    {
      "epoch": 1.2626630279668927,
      "grad_norm": 0.007808353286236525,
      "learning_rate": 3.599472821889657e-06,
      "loss": 0.001,
      "step": 196870
    },
    {
      "epoch": 1.2627271648606786,
      "grad_norm": 0.042417317628860474,
      "learning_rate": 3.598935532641497e-06,
      "loss": 0.0015,
      "step": 196880
    },
    {
      "epoch": 1.2627913017544647,
      "grad_norm": 0.16768033802509308,
      "learning_rate": 3.598398260949675e-06,
      "loss": 0.0021,
      "step": 196890
    },
    {
      "epoch": 1.262855438648251,
      "grad_norm": 0.0659007728099823,
      "learning_rate": 3.597861006820926e-06,
      "loss": 0.001,
      "step": 196900
    },
    {
      "epoch": 1.2629195755420368,
      "grad_norm": 0.006474540568888187,
      "learning_rate": 3.597323770261979e-06,
      "loss": 0.0014,
      "step": 196910
    },
    {
      "epoch": 1.262983712435823,
      "grad_norm": 0.20306262373924255,
      "learning_rate": 3.5967865512795684e-06,
      "loss": 0.0014,
      "step": 196920
    },
    {
      "epoch": 1.2630478493296091,
      "grad_norm": 0.10493241995573044,
      "learning_rate": 3.5962493498804244e-06,
      "loss": 0.0019,
      "step": 196930
    },
    {
      "epoch": 1.2631119862233953,
      "grad_norm": 0.12724147737026215,
      "learning_rate": 3.595712166071281e-06,
      "loss": 0.0013,
      "step": 196940
    },
    {
      "epoch": 1.2631761231171814,
      "grad_norm": 0.014561139978468418,
      "learning_rate": 3.595174999858866e-06,
      "loss": 0.0013,
      "step": 196950
    },
    {
      "epoch": 1.2632402600109673,
      "grad_norm": 0.05698208510875702,
      "learning_rate": 3.5946378512499137e-06,
      "loss": 0.002,
      "step": 196960
    },
    {
      "epoch": 1.2633043969047535,
      "grad_norm": 0.12596088647842407,
      "learning_rate": 3.5941007202511526e-06,
      "loss": 0.0011,
      "step": 196970
    },
    {
      "epoch": 1.2633685337985396,
      "grad_norm": 0.08804207295179367,
      "learning_rate": 3.5935636068693147e-06,
      "loss": 0.0008,
      "step": 196980
    },
    {
      "epoch": 1.2634326706923256,
      "grad_norm": 0.15855345129966736,
      "learning_rate": 3.593026511111129e-06,
      "loss": 0.0016,
      "step": 196990
    },
    {
      "epoch": 1.2634968075861117,
      "grad_norm": 0.16150519251823425,
      "learning_rate": 3.592489432983327e-06,
      "loss": 0.0027,
      "step": 197000
    },
    {
      "epoch": 1.2635609444798979,
      "grad_norm": 0.08217993378639221,
      "learning_rate": 3.5919523724926396e-06,
      "loss": 0.0012,
      "step": 197010
    },
    {
      "epoch": 1.263625081373684,
      "grad_norm": 0.04182827100157738,
      "learning_rate": 3.5914153296457936e-06,
      "loss": 0.0015,
      "step": 197020
    },
    {
      "epoch": 1.2636892182674702,
      "grad_norm": 0.09404095262289047,
      "learning_rate": 3.5908783044495215e-06,
      "loss": 0.0013,
      "step": 197030
    },
    {
      "epoch": 1.2637533551612563,
      "grad_norm": 0.11161879450082779,
      "learning_rate": 3.5903412969105507e-06,
      "loss": 0.002,
      "step": 197040
    },
    {
      "epoch": 1.2638174920550422,
      "grad_norm": 0.08677653223276138,
      "learning_rate": 3.5898043070356114e-06,
      "loss": 0.0014,
      "step": 197050
    },
    {
      "epoch": 1.2638816289488284,
      "grad_norm": 0.13284312188625336,
      "learning_rate": 3.589267334831431e-06,
      "loss": 0.007,
      "step": 197060
    },
    {
      "epoch": 1.2639457658426145,
      "grad_norm": 0.1533959060907364,
      "learning_rate": 3.58873038030474e-06,
      "loss": 0.0016,
      "step": 197070
    },
    {
      "epoch": 1.2640099027364005,
      "grad_norm": 0.28346744179725647,
      "learning_rate": 3.5881934434622655e-06,
      "loss": 0.0015,
      "step": 197080
    },
    {
      "epoch": 1.2640740396301866,
      "grad_norm": 0.025374893099069595,
      "learning_rate": 3.5876565243107377e-06,
      "loss": 0.0013,
      "step": 197090
    },
    {
      "epoch": 1.2641381765239728,
      "grad_norm": 0.0803384780883789,
      "learning_rate": 3.5871196228568807e-06,
      "loss": 0.0014,
      "step": 197100
    },
    {
      "epoch": 1.264202313417759,
      "grad_norm": 0.04518803954124451,
      "learning_rate": 3.5865827391074265e-06,
      "loss": 0.0011,
      "step": 197110
    },
    {
      "epoch": 1.264266450311545,
      "grad_norm": 0.08995330333709717,
      "learning_rate": 3.5860458730690996e-06,
      "loss": 0.0013,
      "step": 197120
    },
    {
      "epoch": 1.264330587205331,
      "grad_norm": 0.06358237564563751,
      "learning_rate": 3.585509024748628e-06,
      "loss": 0.0008,
      "step": 197130
    },
    {
      "epoch": 1.2643947240991171,
      "grad_norm": 0.05482769012451172,
      "learning_rate": 3.5849721941527405e-06,
      "loss": 0.002,
      "step": 197140
    },
    {
      "epoch": 1.2644588609929033,
      "grad_norm": 0.019504649564623833,
      "learning_rate": 3.5844353812881617e-06,
      "loss": 0.0016,
      "step": 197150
    },
    {
      "epoch": 1.2645229978866894,
      "grad_norm": 0.05306436866521835,
      "learning_rate": 3.5838985861616205e-06,
      "loss": 0.0008,
      "step": 197160
    },
    {
      "epoch": 1.2645871347804754,
      "grad_norm": 0.09248397499322891,
      "learning_rate": 3.5833618087798406e-06,
      "loss": 0.0017,
      "step": 197170
    },
    {
      "epoch": 1.2646512716742615,
      "grad_norm": 0.04191691800951958,
      "learning_rate": 3.582825049149551e-06,
      "loss": 0.0011,
      "step": 197180
    },
    {
      "epoch": 1.2647154085680476,
      "grad_norm": 0.009918549098074436,
      "learning_rate": 3.582288307277475e-06,
      "loss": 0.0011,
      "step": 197190
    },
    {
      "epoch": 1.2647795454618338,
      "grad_norm": 0.007974385283887386,
      "learning_rate": 3.5817515831703413e-06,
      "loss": 0.0015,
      "step": 197200
    },
    {
      "epoch": 1.26484368235562,
      "grad_norm": 0.1345491111278534,
      "learning_rate": 3.5812148768348724e-06,
      "loss": 0.0017,
      "step": 197210
    },
    {
      "epoch": 1.2649078192494059,
      "grad_norm": 0.07586783915758133,
      "learning_rate": 3.5806781882777963e-06,
      "loss": 0.0028,
      "step": 197220
    },
    {
      "epoch": 1.264971956143192,
      "grad_norm": 0.060350608080625534,
      "learning_rate": 3.580141517505836e-06,
      "loss": 0.0016,
      "step": 197230
    },
    {
      "epoch": 1.2650360930369782,
      "grad_norm": 0.07729600369930267,
      "learning_rate": 3.5796048645257185e-06,
      "loss": 0.0013,
      "step": 197240
    },
    {
      "epoch": 1.265100229930764,
      "grad_norm": 0.11275080591440201,
      "learning_rate": 3.579068229344166e-06,
      "loss": 0.0021,
      "step": 197250
    },
    {
      "epoch": 1.2651643668245502,
      "grad_norm": 0.03644602745771408,
      "learning_rate": 3.5785316119679047e-06,
      "loss": 0.0024,
      "step": 197260
    },
    {
      "epoch": 1.2652285037183364,
      "grad_norm": 0.13671661913394928,
      "learning_rate": 3.5779950124036574e-06,
      "loss": 0.003,
      "step": 197270
    },
    {
      "epoch": 1.2652926406121225,
      "grad_norm": 0.008413456380367279,
      "learning_rate": 3.5774584306581482e-06,
      "loss": 0.0009,
      "step": 197280
    },
    {
      "epoch": 1.2653567775059087,
      "grad_norm": 0.040338508784770966,
      "learning_rate": 3.576921866738104e-06,
      "loss": 0.0029,
      "step": 197290
    },
    {
      "epoch": 1.2654209143996948,
      "grad_norm": 0.012562062591314316,
      "learning_rate": 3.576385320650244e-06,
      "loss": 0.0023,
      "step": 197300
    },
    {
      "epoch": 1.2654850512934808,
      "grad_norm": 0.011955642141401768,
      "learning_rate": 3.5758487924012942e-06,
      "loss": 0.0013,
      "step": 197310
    },
    {
      "epoch": 1.265549188187267,
      "grad_norm": 0.10264961421489716,
      "learning_rate": 3.575312281997976e-06,
      "loss": 0.0016,
      "step": 197320
    },
    {
      "epoch": 1.265613325081053,
      "grad_norm": 0.06680679321289062,
      "learning_rate": 3.5747757894470147e-06,
      "loss": 0.0009,
      "step": 197330
    },
    {
      "epoch": 1.265677461974839,
      "grad_norm": 0.03692662715911865,
      "learning_rate": 3.57423931475513e-06,
      "loss": 0.0017,
      "step": 197340
    },
    {
      "epoch": 1.2657415988686251,
      "grad_norm": 0.1036771759390831,
      "learning_rate": 3.5737028579290468e-06,
      "loss": 0.0022,
      "step": 197350
    },
    {
      "epoch": 1.2658057357624113,
      "grad_norm": 0.11271841078996658,
      "learning_rate": 3.5731664189754845e-06,
      "loss": 0.001,
      "step": 197360
    },
    {
      "epoch": 1.2658698726561974,
      "grad_norm": 0.09648439288139343,
      "learning_rate": 3.572629997901168e-06,
      "loss": 0.0018,
      "step": 197370
    },
    {
      "epoch": 1.2659340095499836,
      "grad_norm": 0.04070811718702316,
      "learning_rate": 3.572093594712817e-06,
      "loss": 0.0011,
      "step": 197380
    },
    {
      "epoch": 1.2659981464437695,
      "grad_norm": 0.02357848361134529,
      "learning_rate": 3.5715572094171548e-06,
      "loss": 0.0013,
      "step": 197390
    },
    {
      "epoch": 1.2660622833375557,
      "grad_norm": 0.0517885759472847,
      "learning_rate": 3.5710208420209003e-06,
      "loss": 0.0013,
      "step": 197400
    },
    {
      "epoch": 1.2661264202313418,
      "grad_norm": 0.12000174075365067,
      "learning_rate": 3.5704844925307775e-06,
      "loss": 0.0021,
      "step": 197410
    },
    {
      "epoch": 1.2661905571251277,
      "grad_norm": 0.06956811994314194,
      "learning_rate": 3.5699481609535047e-06,
      "loss": 0.0022,
      "step": 197420
    },
    {
      "epoch": 1.2662546940189139,
      "grad_norm": 0.05584663152694702,
      "learning_rate": 3.569411847295803e-06,
      "loss": 0.001,
      "step": 197430
    },
    {
      "epoch": 1.2663188309127,
      "grad_norm": 0.10756741464138031,
      "learning_rate": 3.5688755515643943e-06,
      "loss": 0.0021,
      "step": 197440
    },
    {
      "epoch": 1.2663829678064862,
      "grad_norm": 0.06895508617162704,
      "learning_rate": 3.5683392737659967e-06,
      "loss": 0.0008,
      "step": 197450
    },
    {
      "epoch": 1.2664471047002723,
      "grad_norm": 0.11756862699985504,
      "learning_rate": 3.567803013907332e-06,
      "loss": 0.0027,
      "step": 197460
    },
    {
      "epoch": 1.2665112415940585,
      "grad_norm": 0.20282131433486938,
      "learning_rate": 3.567266771995119e-06,
      "loss": 0.0018,
      "step": 197470
    },
    {
      "epoch": 1.2665753784878444,
      "grad_norm": 0.12561297416687012,
      "learning_rate": 3.5667305480360776e-06,
      "loss": 0.0012,
      "step": 197480
    },
    {
      "epoch": 1.2666395153816306,
      "grad_norm": 0.1691175252199173,
      "learning_rate": 3.5661943420369263e-06,
      "loss": 0.0014,
      "step": 197490
    },
    {
      "epoch": 1.2667036522754167,
      "grad_norm": 0.08796347677707672,
      "learning_rate": 3.5656581540043843e-06,
      "loss": 0.0016,
      "step": 197500
    },
    {
      "epoch": 1.2667677891692026,
      "grad_norm": 0.072384774684906,
      "learning_rate": 3.5651219839451712e-06,
      "loss": 0.0013,
      "step": 197510
    },
    {
      "epoch": 1.2668319260629888,
      "grad_norm": 0.09652019292116165,
      "learning_rate": 3.5645858318660053e-06,
      "loss": 0.0005,
      "step": 197520
    },
    {
      "epoch": 1.266896062956775,
      "grad_norm": 0.0962836816906929,
      "learning_rate": 3.5640496977736044e-06,
      "loss": 0.0016,
      "step": 197530
    },
    {
      "epoch": 1.266960199850561,
      "grad_norm": 0.12505964934825897,
      "learning_rate": 3.563513581674688e-06,
      "loss": 0.003,
      "step": 197540
    },
    {
      "epoch": 1.2670243367443472,
      "grad_norm": 0.2275208979845047,
      "learning_rate": 3.562977483575971e-06,
      "loss": 0.0016,
      "step": 197550
    },
    {
      "epoch": 1.2670884736381332,
      "grad_norm": 0.11794976890087128,
      "learning_rate": 3.562441403484176e-06,
      "loss": 0.002,
      "step": 197560
    },
    {
      "epoch": 1.2671526105319193,
      "grad_norm": 0.2552700936794281,
      "learning_rate": 3.561905341406015e-06,
      "loss": 0.0016,
      "step": 197570
    },
    {
      "epoch": 1.2672167474257054,
      "grad_norm": 0.09053491055965424,
      "learning_rate": 3.5613692973482083e-06,
      "loss": 0.0017,
      "step": 197580
    },
    {
      "epoch": 1.2672808843194916,
      "grad_norm": 0.10517799854278564,
      "learning_rate": 3.5608332713174743e-06,
      "loss": 0.0008,
      "step": 197590
    },
    {
      "epoch": 1.2673450212132775,
      "grad_norm": 0.03843609243631363,
      "learning_rate": 3.5602972633205257e-06,
      "loss": 0.0007,
      "step": 197600
    },
    {
      "epoch": 1.2674091581070637,
      "grad_norm": 0.07241566479206085,
      "learning_rate": 3.559761273364083e-06,
      "loss": 0.0013,
      "step": 197610
    },
    {
      "epoch": 1.2674732950008498,
      "grad_norm": 0.04209311306476593,
      "learning_rate": 3.55922530145486e-06,
      "loss": 0.0011,
      "step": 197620
    },
    {
      "epoch": 1.267537431894636,
      "grad_norm": 0.0937863141298294,
      "learning_rate": 3.558689347599572e-06,
      "loss": 0.0011,
      "step": 197630
    },
    {
      "epoch": 1.2676015687884221,
      "grad_norm": 0.20069929957389832,
      "learning_rate": 3.5581534118049394e-06,
      "loss": 0.0026,
      "step": 197640
    },
    {
      "epoch": 1.267665705682208,
      "grad_norm": 0.10028954595327377,
      "learning_rate": 3.5576174940776735e-06,
      "loss": 0.0015,
      "step": 197650
    },
    {
      "epoch": 1.2677298425759942,
      "grad_norm": 0.3673674166202545,
      "learning_rate": 3.557081594424493e-06,
      "loss": 0.0026,
      "step": 197660
    },
    {
      "epoch": 1.2677939794697803,
      "grad_norm": 0.10088752955198288,
      "learning_rate": 3.5565457128521096e-06,
      "loss": 0.0011,
      "step": 197670
    },
    {
      "epoch": 1.2678581163635663,
      "grad_norm": 0.09068503230810165,
      "learning_rate": 3.5560098493672413e-06,
      "loss": 0.0017,
      "step": 197680
    },
    {
      "epoch": 1.2679222532573524,
      "grad_norm": 0.08756272494792938,
      "learning_rate": 3.5554740039766e-06,
      "loss": 0.0017,
      "step": 197690
    },
    {
      "epoch": 1.2679863901511386,
      "grad_norm": 0.19590413570404053,
      "learning_rate": 3.5549381766869027e-06,
      "loss": 0.0021,
      "step": 197700
    },
    {
      "epoch": 1.2680505270449247,
      "grad_norm": 0.11053825169801712,
      "learning_rate": 3.5544023675048634e-06,
      "loss": 0.0014,
      "step": 197710
    },
    {
      "epoch": 1.2681146639387109,
      "grad_norm": 0.02658925950527191,
      "learning_rate": 3.553866576437195e-06,
      "loss": 0.0016,
      "step": 197720
    },
    {
      "epoch": 1.268178800832497,
      "grad_norm": 0.039638325572013855,
      "learning_rate": 3.553330803490612e-06,
      "loss": 0.0011,
      "step": 197730
    },
    {
      "epoch": 1.268242937726283,
      "grad_norm": 0.04033947363495827,
      "learning_rate": 3.5527950486718275e-06,
      "loss": 0.001,
      "step": 197740
    },
    {
      "epoch": 1.268307074620069,
      "grad_norm": 0.361795037984848,
      "learning_rate": 3.552259311987557e-06,
      "loss": 0.0047,
      "step": 197750
    },
    {
      "epoch": 1.2683712115138552,
      "grad_norm": 0.10919544100761414,
      "learning_rate": 3.55172359344451e-06,
      "loss": 0.0014,
      "step": 197760
    },
    {
      "epoch": 1.2684353484076412,
      "grad_norm": 0.08699346333742142,
      "learning_rate": 3.5511878930494028e-06,
      "loss": 0.0011,
      "step": 197770
    },
    {
      "epoch": 1.2684994853014273,
      "grad_norm": 0.020397085696458817,
      "learning_rate": 3.550652210808946e-06,
      "loss": 0.0011,
      "step": 197780
    },
    {
      "epoch": 1.2685636221952135,
      "grad_norm": 0.4498283267021179,
      "learning_rate": 3.5501165467298538e-06,
      "loss": 0.0036,
      "step": 197790
    },
    {
      "epoch": 1.2686277590889996,
      "grad_norm": 0.007840072736144066,
      "learning_rate": 3.5495809008188357e-06,
      "loss": 0.0009,
      "step": 197800
    },
    {
      "epoch": 1.2686918959827858,
      "grad_norm": 0.16372261941432953,
      "learning_rate": 3.5490452730826073e-06,
      "loss": 0.0015,
      "step": 197810
    },
    {
      "epoch": 1.2687560328765717,
      "grad_norm": 0.13359896838665009,
      "learning_rate": 3.548509663527877e-06,
      "loss": 0.0021,
      "step": 197820
    },
    {
      "epoch": 1.2688201697703578,
      "grad_norm": 0.063264861702919,
      "learning_rate": 3.54797407216136e-06,
      "loss": 0.0012,
      "step": 197830
    },
    {
      "epoch": 1.268884306664144,
      "grad_norm": 0.08021801710128784,
      "learning_rate": 3.5474384989897637e-06,
      "loss": 0.0013,
      "step": 197840
    },
    {
      "epoch": 1.2689484435579301,
      "grad_norm": 0.07261926680803299,
      "learning_rate": 3.546902944019801e-06,
      "loss": 0.0014,
      "step": 197850
    },
    {
      "epoch": 1.269012580451716,
      "grad_norm": 0.02300884947180748,
      "learning_rate": 3.546367407258185e-06,
      "loss": 0.0015,
      "step": 197860
    },
    {
      "epoch": 1.2690767173455022,
      "grad_norm": 0.050239454954862595,
      "learning_rate": 3.545831888711623e-06,
      "loss": 0.0013,
      "step": 197870
    },
    {
      "epoch": 1.2691408542392884,
      "grad_norm": 0.10576912015676498,
      "learning_rate": 3.545296388386827e-06,
      "loss": 0.0024,
      "step": 197880
    },
    {
      "epoch": 1.2692049911330745,
      "grad_norm": 0.0587228424847126,
      "learning_rate": 3.5447609062905062e-06,
      "loss": 0.0008,
      "step": 197890
    },
    {
      "epoch": 1.2692691280268606,
      "grad_norm": 0.13328789174556732,
      "learning_rate": 3.544225442429372e-06,
      "loss": 0.002,
      "step": 197900
    },
    {
      "epoch": 1.2693332649206466,
      "grad_norm": 0.07111755758523941,
      "learning_rate": 3.5436899968101334e-06,
      "loss": 0.0007,
      "step": 197910
    },
    {
      "epoch": 1.2693974018144327,
      "grad_norm": 0.08294267952442169,
      "learning_rate": 3.5431545694394996e-06,
      "loss": 0.0009,
      "step": 197920
    },
    {
      "epoch": 1.2694615387082189,
      "grad_norm": 0.11793050915002823,
      "learning_rate": 3.5426191603241805e-06,
      "loss": 0.0018,
      "step": 197930
    },
    {
      "epoch": 1.2695256756020048,
      "grad_norm": 0.06387563049793243,
      "learning_rate": 3.5420837694708853e-06,
      "loss": 0.0016,
      "step": 197940
    },
    {
      "epoch": 1.269589812495791,
      "grad_norm": 0.18619148433208466,
      "learning_rate": 3.541548396886322e-06,
      "loss": 0.0014,
      "step": 197950
    },
    {
      "epoch": 1.269653949389577,
      "grad_norm": 0.07638446241617203,
      "learning_rate": 3.5410130425771993e-06,
      "loss": 0.0047,
      "step": 197960
    },
    {
      "epoch": 1.2697180862833632,
      "grad_norm": 0.14112454652786255,
      "learning_rate": 3.540477706550226e-06,
      "loss": 0.0019,
      "step": 197970
    },
    {
      "epoch": 1.2697822231771494,
      "grad_norm": 0.10112203657627106,
      "learning_rate": 3.53994238881211e-06,
      "loss": 0.0012,
      "step": 197980
    },
    {
      "epoch": 1.2698463600709355,
      "grad_norm": 0.05991072580218315,
      "learning_rate": 3.5394070893695598e-06,
      "loss": 0.0016,
      "step": 197990
    },
    {
      "epoch": 1.2699104969647215,
      "grad_norm": 0.012906217016279697,
      "learning_rate": 3.538871808229283e-06,
      "loss": 0.0017,
      "step": 198000
    },
    {
      "epoch": 1.2699746338585076,
      "grad_norm": 0.12138118594884872,
      "learning_rate": 3.538336545397986e-06,
      "loss": 0.0036,
      "step": 198010
    },
    {
      "epoch": 1.2700387707522938,
      "grad_norm": 0.20799575746059418,
      "learning_rate": 3.5378013008823765e-06,
      "loss": 0.0016,
      "step": 198020
    },
    {
      "epoch": 1.2701029076460797,
      "grad_norm": 0.05465441942214966,
      "learning_rate": 3.537266074689163e-06,
      "loss": 0.0015,
      "step": 198030
    },
    {
      "epoch": 1.2701670445398658,
      "grad_norm": 0.07412154972553253,
      "learning_rate": 3.53673086682505e-06,
      "loss": 0.0012,
      "step": 198040
    },
    {
      "epoch": 1.270231181433652,
      "grad_norm": 0.0858510211110115,
      "learning_rate": 3.5361956772967453e-06,
      "loss": 0.0017,
      "step": 198050
    },
    {
      "epoch": 1.2702953183274381,
      "grad_norm": 0.05278509110212326,
      "learning_rate": 3.5356605061109545e-06,
      "loss": 0.0015,
      "step": 198060
    },
    {
      "epoch": 1.2703594552212243,
      "grad_norm": 0.05897562950849533,
      "learning_rate": 3.5351253532743847e-06,
      "loss": 0.001,
      "step": 198070
    },
    {
      "epoch": 1.2704235921150102,
      "grad_norm": 0.12483041733503342,
      "learning_rate": 3.534590218793741e-06,
      "loss": 0.0031,
      "step": 198080
    },
    {
      "epoch": 1.2704877290087964,
      "grad_norm": 0.15329022705554962,
      "learning_rate": 3.534055102675729e-06,
      "loss": 0.0019,
      "step": 198090
    },
    {
      "epoch": 1.2705518659025825,
      "grad_norm": 0.09980618208646774,
      "learning_rate": 3.5335200049270537e-06,
      "loss": 0.0019,
      "step": 198100
    },
    {
      "epoch": 1.2706160027963684,
      "grad_norm": 0.004729558248072863,
      "learning_rate": 3.5329849255544226e-06,
      "loss": 0.0023,
      "step": 198110
    },
    {
      "epoch": 1.2706801396901546,
      "grad_norm": 0.020556021481752396,
      "learning_rate": 3.5324498645645368e-06,
      "loss": 0.0009,
      "step": 198120
    },
    {
      "epoch": 1.2707442765839407,
      "grad_norm": 0.037179723381996155,
      "learning_rate": 3.531914821964103e-06,
      "loss": 0.0016,
      "step": 198130
    },
    {
      "epoch": 1.2708084134777269,
      "grad_norm": 0.289968341588974,
      "learning_rate": 3.5313797977598276e-06,
      "loss": 0.0021,
      "step": 198140
    },
    {
      "epoch": 1.270872550371513,
      "grad_norm": 0.1045757308602333,
      "learning_rate": 3.530844791958411e-06,
      "loss": 0.0009,
      "step": 198150
    },
    {
      "epoch": 1.2709366872652992,
      "grad_norm": 0.16217878460884094,
      "learning_rate": 3.5303098045665607e-06,
      "loss": 0.0012,
      "step": 198160
    },
    {
      "epoch": 1.271000824159085,
      "grad_norm": 0.07823773473501205,
      "learning_rate": 3.529774835590978e-06,
      "loss": 0.0015,
      "step": 198170
    },
    {
      "epoch": 1.2710649610528713,
      "grad_norm": 0.08206193894147873,
      "learning_rate": 3.529239885038368e-06,
      "loss": 0.0006,
      "step": 198180
    },
    {
      "epoch": 1.2711290979466574,
      "grad_norm": 0.15786902606487274,
      "learning_rate": 3.5287049529154325e-06,
      "loss": 0.0016,
      "step": 198190
    },
    {
      "epoch": 1.2711932348404433,
      "grad_norm": 0.018060648813843727,
      "learning_rate": 3.528170039228877e-06,
      "loss": 0.001,
      "step": 198200
    },
    {
      "epoch": 1.2712573717342295,
      "grad_norm": 0.11646221578121185,
      "learning_rate": 3.5276351439854013e-06,
      "loss": 0.0019,
      "step": 198210
    },
    {
      "epoch": 1.2713215086280156,
      "grad_norm": 0.13924136757850647,
      "learning_rate": 3.527100267191711e-06,
      "loss": 0.0014,
      "step": 198220
    },
    {
      "epoch": 1.2713856455218018,
      "grad_norm": 0.19885863363742828,
      "learning_rate": 3.526565408854505e-06,
      "loss": 0.0021,
      "step": 198230
    },
    {
      "epoch": 1.271449782415588,
      "grad_norm": 0.08520758152008057,
      "learning_rate": 3.526030568980489e-06,
      "loss": 0.0011,
      "step": 198240
    },
    {
      "epoch": 1.2715139193093739,
      "grad_norm": 0.07740870863199234,
      "learning_rate": 3.525495747576363e-06,
      "loss": 0.0011,
      "step": 198250
    },
    {
      "epoch": 1.27157805620316,
      "grad_norm": 0.05698053911328316,
      "learning_rate": 3.5249609446488293e-06,
      "loss": 0.0022,
      "step": 198260
    },
    {
      "epoch": 1.2716421930969461,
      "grad_norm": 0.10947985202074051,
      "learning_rate": 3.5244261602045882e-06,
      "loss": 0.0012,
      "step": 198270
    },
    {
      "epoch": 1.2717063299907323,
      "grad_norm": 0.03766002878546715,
      "learning_rate": 3.523891394250342e-06,
      "loss": 0.0012,
      "step": 198280
    },
    {
      "epoch": 1.2717704668845182,
      "grad_norm": 0.05742770805954933,
      "learning_rate": 3.523356646792793e-06,
      "loss": 0.001,
      "step": 198290
    },
    {
      "epoch": 1.2718346037783044,
      "grad_norm": 0.22869835793972015,
      "learning_rate": 3.522821917838639e-06,
      "loss": 0.0016,
      "step": 198300
    },
    {
      "epoch": 1.2718987406720905,
      "grad_norm": 0.034320537000894547,
      "learning_rate": 3.5222872073945835e-06,
      "loss": 0.0011,
      "step": 198310
    },
    {
      "epoch": 1.2719628775658767,
      "grad_norm": 0.016153911128640175,
      "learning_rate": 3.5217525154673243e-06,
      "loss": 0.0014,
      "step": 198320
    },
    {
      "epoch": 1.2720270144596628,
      "grad_norm": 0.0031908380333334208,
      "learning_rate": 3.5212178420635635e-06,
      "loss": 0.0015,
      "step": 198330
    },
    {
      "epoch": 1.2720911513534487,
      "grad_norm": 0.031341325491666794,
      "learning_rate": 3.5206831871899994e-06,
      "loss": 0.0018,
      "step": 198340
    },
    {
      "epoch": 1.272155288247235,
      "grad_norm": 0.11928802728652954,
      "learning_rate": 3.520148550853333e-06,
      "loss": 0.0017,
      "step": 198350
    },
    {
      "epoch": 1.272219425141021,
      "grad_norm": 0.0955466702580452,
      "learning_rate": 3.5196139330602615e-06,
      "loss": 0.0025,
      "step": 198360
    },
    {
      "epoch": 1.272283562034807,
      "grad_norm": 0.0947674885392189,
      "learning_rate": 3.5190793338174865e-06,
      "loss": 0.002,
      "step": 198370
    },
    {
      "epoch": 1.2723476989285931,
      "grad_norm": 0.0821545422077179,
      "learning_rate": 3.518544753131705e-06,
      "loss": 0.0017,
      "step": 198380
    },
    {
      "epoch": 1.2724118358223793,
      "grad_norm": 0.10674940794706345,
      "learning_rate": 3.5180101910096175e-06,
      "loss": 0.0011,
      "step": 198390
    },
    {
      "epoch": 1.2724759727161654,
      "grad_norm": 0.16639532148838043,
      "learning_rate": 3.5174756474579188e-06,
      "loss": 0.0011,
      "step": 198400
    },
    {
      "epoch": 1.2725401096099516,
      "grad_norm": 0.0952320322394371,
      "learning_rate": 3.5169411224833124e-06,
      "loss": 0.0019,
      "step": 198410
    },
    {
      "epoch": 1.2726042465037377,
      "grad_norm": 0.07761390507221222,
      "learning_rate": 3.5164066160924924e-06,
      "loss": 0.0017,
      "step": 198420
    },
    {
      "epoch": 1.2726683833975236,
      "grad_norm": 0.22065448760986328,
      "learning_rate": 3.5158721282921592e-06,
      "loss": 0.0024,
      "step": 198430
    },
    {
      "epoch": 1.2727325202913098,
      "grad_norm": 0.08059868961572647,
      "learning_rate": 3.5153376590890074e-06,
      "loss": 0.0017,
      "step": 198440
    },
    {
      "epoch": 1.272796657185096,
      "grad_norm": 0.1412181556224823,
      "learning_rate": 3.5148032084897375e-06,
      "loss": 0.001,
      "step": 198450
    },
    {
      "epoch": 1.2728607940788819,
      "grad_norm": 0.13328877091407776,
      "learning_rate": 3.5142687765010443e-06,
      "loss": 0.0014,
      "step": 198460
    },
    {
      "epoch": 1.272924930972668,
      "grad_norm": 0.0085500068962574,
      "learning_rate": 3.513734363129625e-06,
      "loss": 0.0009,
      "step": 198470
    },
    {
      "epoch": 1.2729890678664542,
      "grad_norm": 0.1177920401096344,
      "learning_rate": 3.513199968382176e-06,
      "loss": 0.002,
      "step": 198480
    },
    {
      "epoch": 1.2730532047602403,
      "grad_norm": 0.12280911952257156,
      "learning_rate": 3.5126655922653953e-06,
      "loss": 0.0014,
      "step": 198490
    },
    {
      "epoch": 1.2731173416540265,
      "grad_norm": 0.007653483655303717,
      "learning_rate": 3.512131234785976e-06,
      "loss": 0.0012,
      "step": 198500
    },
    {
      "epoch": 1.2731814785478124,
      "grad_norm": 0.0819479450583458,
      "learning_rate": 3.5115968959506164e-06,
      "loss": 0.0022,
      "step": 198510
    },
    {
      "epoch": 1.2732456154415985,
      "grad_norm": 0.05366813763976097,
      "learning_rate": 3.511062575766011e-06,
      "loss": 0.0016,
      "step": 198520
    },
    {
      "epoch": 1.2733097523353847,
      "grad_norm": 0.017890458926558495,
      "learning_rate": 3.510528274238857e-06,
      "loss": 0.0009,
      "step": 198530
    },
    {
      "epoch": 1.2733738892291706,
      "grad_norm": 0.014618410728871822,
      "learning_rate": 3.509993991375847e-06,
      "loss": 0.0018,
      "step": 198540
    },
    {
      "epoch": 1.2734380261229568,
      "grad_norm": 0.136849507689476,
      "learning_rate": 3.509459727183677e-06,
      "loss": 0.0017,
      "step": 198550
    },
    {
      "epoch": 1.273502163016743,
      "grad_norm": 0.05653097480535507,
      "learning_rate": 3.5089254816690425e-06,
      "loss": 0.0025,
      "step": 198560
    },
    {
      "epoch": 1.273566299910529,
      "grad_norm": 0.14733551442623138,
      "learning_rate": 3.5083912548386367e-06,
      "loss": 0.0013,
      "step": 198570
    },
    {
      "epoch": 1.2736304368043152,
      "grad_norm": 0.3166915774345398,
      "learning_rate": 3.5078570466991556e-06,
      "loss": 0.0022,
      "step": 198580
    },
    {
      "epoch": 1.2736945736981014,
      "grad_norm": 0.07953108847141266,
      "learning_rate": 3.5073228572572914e-06,
      "loss": 0.002,
      "step": 198590
    },
    {
      "epoch": 1.2737587105918873,
      "grad_norm": 0.09786271303892136,
      "learning_rate": 3.5067886865197397e-06,
      "loss": 0.0042,
      "step": 198600
    },
    {
      "epoch": 1.2738228474856734,
      "grad_norm": 0.06881196796894073,
      "learning_rate": 3.5062545344931915e-06,
      "loss": 0.0011,
      "step": 198610
    },
    {
      "epoch": 1.2738869843794596,
      "grad_norm": 0.007864178158342838,
      "learning_rate": 3.5057204011843427e-06,
      "loss": 0.0022,
      "step": 198620
    },
    {
      "epoch": 1.2739511212732455,
      "grad_norm": 0.15022243559360504,
      "learning_rate": 3.5051862865998843e-06,
      "loss": 0.0028,
      "step": 198630
    },
    {
      "epoch": 1.2740152581670317,
      "grad_norm": 0.13478389382362366,
      "learning_rate": 3.50465219074651e-06,
      "loss": 0.0015,
      "step": 198640
    },
    {
      "epoch": 1.2740793950608178,
      "grad_norm": 0.30176836252212524,
      "learning_rate": 3.5041181136309133e-06,
      "loss": 0.0009,
      "step": 198650
    },
    {
      "epoch": 1.274143531954604,
      "grad_norm": 0.054974887520074844,
      "learning_rate": 3.5035840552597855e-06,
      "loss": 0.0013,
      "step": 198660
    },
    {
      "epoch": 1.27420766884839,
      "grad_norm": 0.11352365463972092,
      "learning_rate": 3.5030500156398182e-06,
      "loss": 0.0014,
      "step": 198670
    },
    {
      "epoch": 1.274271805742176,
      "grad_norm": 0.08226925879716873,
      "learning_rate": 3.5025159947777055e-06,
      "loss": 0.0011,
      "step": 198680
    },
    {
      "epoch": 1.2743359426359622,
      "grad_norm": 0.07407528907060623,
      "learning_rate": 3.5019819926801357e-06,
      "loss": 0.0013,
      "step": 198690
    },
    {
      "epoch": 1.2744000795297483,
      "grad_norm": 0.15557901561260223,
      "learning_rate": 3.5014480093538027e-06,
      "loss": 0.0017,
      "step": 198700
    },
    {
      "epoch": 1.2744642164235345,
      "grad_norm": 0.07050323486328125,
      "learning_rate": 3.500914044805398e-06,
      "loss": 0.0007,
      "step": 198710
    },
    {
      "epoch": 1.2745283533173204,
      "grad_norm": 0.09494238346815109,
      "learning_rate": 3.5003800990416117e-06,
      "loss": 0.0018,
      "step": 198720
    },
    {
      "epoch": 1.2745924902111065,
      "grad_norm": 0.05900660529732704,
      "learning_rate": 3.499846172069134e-06,
      "loss": 0.001,
      "step": 198730
    },
    {
      "epoch": 1.2746566271048927,
      "grad_norm": 0.07017749547958374,
      "learning_rate": 3.4993122638946555e-06,
      "loss": 0.0032,
      "step": 198740
    },
    {
      "epoch": 1.2747207639986788,
      "grad_norm": 0.04791183024644852,
      "learning_rate": 3.498778374524868e-06,
      "loss": 0.0015,
      "step": 198750
    },
    {
      "epoch": 1.274784900892465,
      "grad_norm": 0.10131379961967468,
      "learning_rate": 3.498244503966459e-06,
      "loss": 0.0026,
      "step": 198760
    },
    {
      "epoch": 1.274849037786251,
      "grad_norm": 0.11673296242952347,
      "learning_rate": 3.497710652226122e-06,
      "loss": 0.0009,
      "step": 198770
    },
    {
      "epoch": 1.274913174680037,
      "grad_norm": 0.032778967171907425,
      "learning_rate": 3.4971768193105414e-06,
      "loss": 0.001,
      "step": 198780
    },
    {
      "epoch": 1.2749773115738232,
      "grad_norm": 0.044548217207193375,
      "learning_rate": 3.4966430052264112e-06,
      "loss": 0.0014,
      "step": 198790
    },
    {
      "epoch": 1.2750414484676091,
      "grad_norm": 0.12870903313159943,
      "learning_rate": 3.496109209980417e-06,
      "loss": 0.0018,
      "step": 198800
    },
    {
      "epoch": 1.2751055853613953,
      "grad_norm": 0.2227126806974411,
      "learning_rate": 3.4955754335792513e-06,
      "loss": 0.0012,
      "step": 198810
    },
    {
      "epoch": 1.2751697222551814,
      "grad_norm": 0.09919065237045288,
      "learning_rate": 3.4950416760295987e-06,
      "loss": 0.0013,
      "step": 198820
    },
    {
      "epoch": 1.2752338591489676,
      "grad_norm": 0.18281763792037964,
      "learning_rate": 3.4945079373381496e-06,
      "loss": 0.0024,
      "step": 198830
    },
    {
      "epoch": 1.2752979960427537,
      "grad_norm": 0.06877867132425308,
      "learning_rate": 3.493974217511593e-06,
      "loss": 0.0015,
      "step": 198840
    },
    {
      "epoch": 1.2753621329365399,
      "grad_norm": 0.03930193558335304,
      "learning_rate": 3.4934405165566153e-06,
      "loss": 0.0029,
      "step": 198850
    },
    {
      "epoch": 1.2754262698303258,
      "grad_norm": 0.058170489966869354,
      "learning_rate": 3.492906834479905e-06,
      "loss": 0.0008,
      "step": 198860
    },
    {
      "epoch": 1.275490406724112,
      "grad_norm": 0.02838187851011753,
      "learning_rate": 3.4923731712881483e-06,
      "loss": 0.0007,
      "step": 198870
    },
    {
      "epoch": 1.275554543617898,
      "grad_norm": 0.20764176547527313,
      "learning_rate": 3.4918395269880345e-06,
      "loss": 0.002,
      "step": 198880
    },
    {
      "epoch": 1.275618680511684,
      "grad_norm": 0.18721093237400055,
      "learning_rate": 3.491305901586248e-06,
      "loss": 0.0011,
      "step": 198890
    },
    {
      "epoch": 1.2756828174054702,
      "grad_norm": 0.027380438521504402,
      "learning_rate": 3.4907722950894785e-06,
      "loss": 0.001,
      "step": 198900
    },
    {
      "epoch": 1.2757469542992563,
      "grad_norm": 0.04297863319516182,
      "learning_rate": 3.4902387075044085e-06,
      "loss": 0.0016,
      "step": 198910
    },
    {
      "epoch": 1.2758110911930425,
      "grad_norm": 0.013423757627606392,
      "learning_rate": 3.489705138837729e-06,
      "loss": 0.0024,
      "step": 198920
    },
    {
      "epoch": 1.2758752280868286,
      "grad_norm": 0.21460917592048645,
      "learning_rate": 3.4891715890961215e-06,
      "loss": 0.0032,
      "step": 198930
    },
    {
      "epoch": 1.2759393649806146,
      "grad_norm": 0.09087955951690674,
      "learning_rate": 3.488638058286275e-06,
      "loss": 0.0011,
      "step": 198940
    },
    {
      "epoch": 1.2760035018744007,
      "grad_norm": 0.18837442994117737,
      "learning_rate": 3.488104546414873e-06,
      "loss": 0.0017,
      "step": 198950
    },
    {
      "epoch": 1.2760676387681869,
      "grad_norm": 0.02254685014486313,
      "learning_rate": 3.487571053488602e-06,
      "loss": 0.0009,
      "step": 198960
    },
    {
      "epoch": 1.2761317756619728,
      "grad_norm": 0.10164394229650497,
      "learning_rate": 3.4870375795141463e-06,
      "loss": 0.001,
      "step": 198970
    },
    {
      "epoch": 1.276195912555759,
      "grad_norm": 0.12734057009220123,
      "learning_rate": 3.4865041244981906e-06,
      "loss": 0.0015,
      "step": 198980
    },
    {
      "epoch": 1.276260049449545,
      "grad_norm": 0.04082336649298668,
      "learning_rate": 3.4859706884474207e-06,
      "loss": 0.0007,
      "step": 198990
    },
    {
      "epoch": 1.2763241863433312,
      "grad_norm": 0.2712097764015198,
      "learning_rate": 3.48543727136852e-06,
      "loss": 0.0023,
      "step": 199000
    },
    {
      "epoch": 1.2763883232371174,
      "grad_norm": 0.1718614399433136,
      "learning_rate": 3.484903873268173e-06,
      "loss": 0.0029,
      "step": 199010
    },
    {
      "epoch": 1.2764524601309035,
      "grad_norm": 0.04521147906780243,
      "learning_rate": 3.484370494153062e-06,
      "loss": 0.0008,
      "step": 199020
    },
    {
      "epoch": 1.2765165970246894,
      "grad_norm": 0.04617341235280037,
      "learning_rate": 3.483837134029874e-06,
      "loss": 0.0015,
      "step": 199030
    },
    {
      "epoch": 1.2765807339184756,
      "grad_norm": 0.057335786521434784,
      "learning_rate": 3.483303792905288e-06,
      "loss": 0.0011,
      "step": 199040
    },
    {
      "epoch": 1.2766448708122617,
      "grad_norm": 0.010286826640367508,
      "learning_rate": 3.4827704707859917e-06,
      "loss": 0.0026,
      "step": 199050
    },
    {
      "epoch": 1.2767090077060477,
      "grad_norm": 0.014647098258137703,
      "learning_rate": 3.482237167678664e-06,
      "loss": 0.0021,
      "step": 199060
    },
    {
      "epoch": 1.2767731445998338,
      "grad_norm": 0.0392424501478672,
      "learning_rate": 3.481703883589991e-06,
      "loss": 0.0019,
      "step": 199070
    },
    {
      "epoch": 1.27683728149362,
      "grad_norm": 0.08423841744661331,
      "learning_rate": 3.4811706185266514e-06,
      "loss": 0.0018,
      "step": 199080
    },
    {
      "epoch": 1.2769014183874061,
      "grad_norm": 0.08013307303190231,
      "learning_rate": 3.480637372495331e-06,
      "loss": 0.0015,
      "step": 199090
    },
    {
      "epoch": 1.2769655552811923,
      "grad_norm": 0.43441662192344666,
      "learning_rate": 3.4801041455027094e-06,
      "loss": 0.0016,
      "step": 199100
    },
    {
      "epoch": 1.2770296921749782,
      "grad_norm": 0.15558330714702606,
      "learning_rate": 3.47957093755547e-06,
      "loss": 0.0012,
      "step": 199110
    },
    {
      "epoch": 1.2770938290687643,
      "grad_norm": 0.07779597491025925,
      "learning_rate": 3.479037748660292e-06,
      "loss": 0.0012,
      "step": 199120
    },
    {
      "epoch": 1.2771579659625505,
      "grad_norm": 0.1093960627913475,
      "learning_rate": 3.4785045788238582e-06,
      "loss": 0.0017,
      "step": 199130
    },
    {
      "epoch": 1.2772221028563366,
      "grad_norm": 0.0921272486448288,
      "learning_rate": 3.4779714280528507e-06,
      "loss": 0.0014,
      "step": 199140
    },
    {
      "epoch": 1.2772862397501226,
      "grad_norm": 0.004315617028623819,
      "learning_rate": 3.4774382963539475e-06,
      "loss": 0.0018,
      "step": 199150
    },
    {
      "epoch": 1.2773503766439087,
      "grad_norm": 0.1703188568353653,
      "learning_rate": 3.4769051837338303e-06,
      "loss": 0.0021,
      "step": 199160
    },
    {
      "epoch": 1.2774145135376949,
      "grad_norm": 0.1463497132062912,
      "learning_rate": 3.4763720901991815e-06,
      "loss": 0.0009,
      "step": 199170
    },
    {
      "epoch": 1.277478650431481,
      "grad_norm": 0.018715834245085716,
      "learning_rate": 3.475839015756678e-06,
      "loss": 0.0018,
      "step": 199180
    },
    {
      "epoch": 1.2775427873252672,
      "grad_norm": 0.08665967732667923,
      "learning_rate": 3.475305960413002e-06,
      "loss": 0.001,
      "step": 199190
    },
    {
      "epoch": 1.277606924219053,
      "grad_norm": 0.024206936359405518,
      "learning_rate": 3.4747729241748308e-06,
      "loss": 0.0009,
      "step": 199200
    },
    {
      "epoch": 1.2776710611128392,
      "grad_norm": 0.1050475686788559,
      "learning_rate": 3.4742399070488464e-06,
      "loss": 0.0011,
      "step": 199210
    },
    {
      "epoch": 1.2777351980066254,
      "grad_norm": 0.156494602560997,
      "learning_rate": 3.473706909041725e-06,
      "loss": 0.0017,
      "step": 199220
    },
    {
      "epoch": 1.2777993349004113,
      "grad_norm": 0.07716964930295944,
      "learning_rate": 3.4731739301601482e-06,
      "loss": 0.0014,
      "step": 199230
    },
    {
      "epoch": 1.2778634717941975,
      "grad_norm": 0.2524273991584778,
      "learning_rate": 3.472640970410792e-06,
      "loss": 0.0037,
      "step": 199240
    },
    {
      "epoch": 1.2779276086879836,
      "grad_norm": 0.05351315438747406,
      "learning_rate": 3.4721080298003364e-06,
      "loss": 0.0009,
      "step": 199250
    },
    {
      "epoch": 1.2779917455817698,
      "grad_norm": 0.08141786605119705,
      "learning_rate": 3.47157510833546e-06,
      "loss": 0.0015,
      "step": 199260
    },
    {
      "epoch": 1.278055882475556,
      "grad_norm": 0.09562107920646667,
      "learning_rate": 3.4710422060228384e-06,
      "loss": 0.0018,
      "step": 199270
    },
    {
      "epoch": 1.278120019369342,
      "grad_norm": 0.03243833780288696,
      "learning_rate": 3.470509322869152e-06,
      "loss": 0.0013,
      "step": 199280
    },
    {
      "epoch": 1.278184156263128,
      "grad_norm": 0.10546350479125977,
      "learning_rate": 3.469976458881076e-06,
      "loss": 0.0011,
      "step": 199290
    },
    {
      "epoch": 1.2782482931569141,
      "grad_norm": 0.09585004299879074,
      "learning_rate": 3.469443614065289e-06,
      "loss": 0.0011,
      "step": 199300
    },
    {
      "epoch": 1.2783124300507003,
      "grad_norm": 0.13913419842720032,
      "learning_rate": 3.4689107884284666e-06,
      "loss": 0.0023,
      "step": 199310
    },
    {
      "epoch": 1.2783765669444862,
      "grad_norm": 0.09405070543289185,
      "learning_rate": 3.4683779819772877e-06,
      "loss": 0.0015,
      "step": 199320
    },
    {
      "epoch": 1.2784407038382724,
      "grad_norm": 0.07846540957689285,
      "learning_rate": 3.467845194718426e-06,
      "loss": 0.0019,
      "step": 199330
    },
    {
      "epoch": 1.2785048407320585,
      "grad_norm": 0.09102808684110641,
      "learning_rate": 3.46731242665856e-06,
      "loss": 0.0011,
      "step": 199340
    },
    {
      "epoch": 1.2785689776258446,
      "grad_norm": 0.15697677433490753,
      "learning_rate": 3.466779677804364e-06,
      "loss": 0.0016,
      "step": 199350
    },
    {
      "epoch": 1.2786331145196308,
      "grad_norm": 0.029571913182735443,
      "learning_rate": 3.466246948162515e-06,
      "loss": 0.0012,
      "step": 199360
    },
    {
      "epoch": 1.2786972514134167,
      "grad_norm": 0.04870545119047165,
      "learning_rate": 3.4657142377396876e-06,
      "loss": 0.0015,
      "step": 199370
    },
    {
      "epoch": 1.2787613883072029,
      "grad_norm": 0.059037283062934875,
      "learning_rate": 3.4651815465425577e-06,
      "loss": 0.001,
      "step": 199380
    },
    {
      "epoch": 1.278825525200989,
      "grad_norm": 0.047760073095560074,
      "learning_rate": 3.464648874577799e-06,
      "loss": 0.0012,
      "step": 199390
    },
    {
      "epoch": 1.2788896620947752,
      "grad_norm": 0.1385759860277176,
      "learning_rate": 3.4641162218520875e-06,
      "loss": 0.0015,
      "step": 199400
    },
    {
      "epoch": 1.278953798988561,
      "grad_norm": 0.06457321345806122,
      "learning_rate": 3.463583588372098e-06,
      "loss": 0.0013,
      "step": 199410
    },
    {
      "epoch": 1.2790179358823472,
      "grad_norm": 0.08694420754909515,
      "learning_rate": 3.4630509741445027e-06,
      "loss": 0.0011,
      "step": 199420
    },
    {
      "epoch": 1.2790820727761334,
      "grad_norm": 0.18832582235336304,
      "learning_rate": 3.4625183791759793e-06,
      "loss": 0.001,
      "step": 199430
    },
    {
      "epoch": 1.2791462096699195,
      "grad_norm": 0.07723814249038696,
      "learning_rate": 3.461985803473198e-06,
      "loss": 0.0014,
      "step": 199440
    },
    {
      "epoch": 1.2792103465637057,
      "grad_norm": 0.10175478458404541,
      "learning_rate": 3.4614532470428346e-06,
      "loss": 0.0014,
      "step": 199450
    },
    {
      "epoch": 1.2792744834574916,
      "grad_norm": 0.1596599817276001,
      "learning_rate": 3.460920709891561e-06,
      "loss": 0.0012,
      "step": 199460
    },
    {
      "epoch": 1.2793386203512778,
      "grad_norm": 0.015792615711688995,
      "learning_rate": 3.4603881920260516e-06,
      "loss": 0.0011,
      "step": 199470
    },
    {
      "epoch": 1.279402757245064,
      "grad_norm": 0.11992931365966797,
      "learning_rate": 3.459855693452977e-06,
      "loss": 0.0012,
      "step": 199480
    },
    {
      "epoch": 1.2794668941388498,
      "grad_norm": 0.05175817757844925,
      "learning_rate": 3.4593232141790134e-06,
      "loss": 0.001,
      "step": 199490
    },
    {
      "epoch": 1.279531031032636,
      "grad_norm": 0.12967315316200256,
      "learning_rate": 3.458790754210829e-06,
      "loss": 0.003,
      "step": 199500
    },
    {
      "epoch": 1.2795951679264221,
      "grad_norm": 0.04164445027709007,
      "learning_rate": 3.4582583135550995e-06,
      "loss": 0.0012,
      "step": 199510
    },
    {
      "epoch": 1.2796593048202083,
      "grad_norm": 0.052503302693367004,
      "learning_rate": 3.457725892218494e-06,
      "loss": 0.0009,
      "step": 199520
    },
    {
      "epoch": 1.2797234417139944,
      "grad_norm": 0.18722525238990784,
      "learning_rate": 3.4571934902076866e-06,
      "loss": 0.0015,
      "step": 199530
    },
    {
      "epoch": 1.2797875786077806,
      "grad_norm": 0.163239523768425,
      "learning_rate": 3.4566611075293455e-06,
      "loss": 0.0012,
      "step": 199540
    },
    {
      "epoch": 1.2798517155015665,
      "grad_norm": 0.07324928045272827,
      "learning_rate": 3.456128744190145e-06,
      "loss": 0.0013,
      "step": 199550
    },
    {
      "epoch": 1.2799158523953527,
      "grad_norm": 0.09207472205162048,
      "learning_rate": 3.4555964001967544e-06,
      "loss": 0.0018,
      "step": 199560
    },
    {
      "epoch": 1.2799799892891388,
      "grad_norm": 0.07849516719579697,
      "learning_rate": 3.455064075555845e-06,
      "loss": 0.0015,
      "step": 199570
    },
    {
      "epoch": 1.2800441261829247,
      "grad_norm": 0.09054175764322281,
      "learning_rate": 3.454531770274087e-06,
      "loss": 0.0031,
      "step": 199580
    },
    {
      "epoch": 1.2801082630767109,
      "grad_norm": 0.01391641702502966,
      "learning_rate": 3.45399948435815e-06,
      "loss": 0.0015,
      "step": 199590
    },
    {
      "epoch": 1.280172399970497,
      "grad_norm": 0.04894445464015007,
      "learning_rate": 3.453467217814705e-06,
      "loss": 0.0011,
      "step": 199600
    },
    {
      "epoch": 1.2802365368642832,
      "grad_norm": 0.0364600270986557,
      "learning_rate": 3.45293497065042e-06,
      "loss": 0.0019,
      "step": 199610
    },
    {
      "epoch": 1.2803006737580693,
      "grad_norm": 0.011941754259169102,
      "learning_rate": 3.4524027428719664e-06,
      "loss": 0.0015,
      "step": 199620
    },
    {
      "epoch": 1.2803648106518553,
      "grad_norm": 0.0831613540649414,
      "learning_rate": 3.4518705344860122e-06,
      "loss": 0.0009,
      "step": 199630
    },
    {
      "epoch": 1.2804289475456414,
      "grad_norm": 0.19725944101810455,
      "learning_rate": 3.4513383454992268e-06,
      "loss": 0.0016,
      "step": 199640
    },
    {
      "epoch": 1.2804930844394276,
      "grad_norm": 0.03266933560371399,
      "learning_rate": 3.4508061759182783e-06,
      "loss": 0.0024,
      "step": 199650
    },
    {
      "epoch": 1.2805572213332135,
      "grad_norm": 0.07610664516687393,
      "learning_rate": 3.4502740257498363e-06,
      "loss": 0.0009,
      "step": 199660
    },
    {
      "epoch": 1.2806213582269996,
      "grad_norm": 0.12725989520549774,
      "learning_rate": 3.4497418950005673e-06,
      "loss": 0.0006,
      "step": 199670
    },
    {
      "epoch": 1.2806854951207858,
      "grad_norm": 0.08239573985338211,
      "learning_rate": 3.4492097836771405e-06,
      "loss": 0.0018,
      "step": 199680
    },
    {
      "epoch": 1.280749632014572,
      "grad_norm": 0.313229501247406,
      "learning_rate": 3.4486776917862242e-06,
      "loss": 0.0016,
      "step": 199690
    },
    {
      "epoch": 1.280813768908358,
      "grad_norm": 0.07731663435697556,
      "learning_rate": 3.448145619334484e-06,
      "loss": 0.0017,
      "step": 199700
    },
    {
      "epoch": 1.2808779058021442,
      "grad_norm": 0.0655188336968422,
      "learning_rate": 3.4476135663285905e-06,
      "loss": 0.0036,
      "step": 199710
    },
    {
      "epoch": 1.2809420426959301,
      "grad_norm": 0.012748525477945805,
      "learning_rate": 3.447081532775206e-06,
      "loss": 0.0009,
      "step": 199720
    },
    {
      "epoch": 1.2810061795897163,
      "grad_norm": 0.06347203254699707,
      "learning_rate": 3.4465495186810016e-06,
      "loss": 0.0012,
      "step": 199730
    },
    {
      "epoch": 1.2810703164835024,
      "grad_norm": 0.23465777933597565,
      "learning_rate": 3.4460175240526407e-06,
      "loss": 0.0019,
      "step": 199740
    },
    {
      "epoch": 1.2811344533772884,
      "grad_norm": 0.06513968110084534,
      "learning_rate": 3.445485548896792e-06,
      "loss": 0.0011,
      "step": 199750
    },
    {
      "epoch": 1.2811985902710745,
      "grad_norm": 0.10150118172168732,
      "learning_rate": 3.444953593220119e-06,
      "loss": 0.0023,
      "step": 199760
    },
    {
      "epoch": 1.2812627271648607,
      "grad_norm": 0.12035603821277618,
      "learning_rate": 3.44442165702929e-06,
      "loss": 0.0019,
      "step": 199770
    },
    {
      "epoch": 1.2813268640586468,
      "grad_norm": 0.18347781896591187,
      "learning_rate": 3.4438897403309686e-06,
      "loss": 0.002,
      "step": 199780
    },
    {
      "epoch": 1.281391000952433,
      "grad_norm": 0.033217377960681915,
      "learning_rate": 3.4433578431318216e-06,
      "loss": 0.0017,
      "step": 199790
    },
    {
      "epoch": 1.281455137846219,
      "grad_norm": 0.1518799513578415,
      "learning_rate": 3.4428259654385123e-06,
      "loss": 0.0063,
      "step": 199800
    },
    {
      "epoch": 1.281519274740005,
      "grad_norm": 0.08355305343866348,
      "learning_rate": 3.4422941072577076e-06,
      "loss": 0.0015,
      "step": 199810
    },
    {
      "epoch": 1.2815834116337912,
      "grad_norm": 0.11808858811855316,
      "learning_rate": 3.4417622685960704e-06,
      "loss": 0.002,
      "step": 199820
    },
    {
      "epoch": 1.2816475485275773,
      "grad_norm": 0.06710100173950195,
      "learning_rate": 3.4412304494602654e-06,
      "loss": 0.003,
      "step": 199830
    },
    {
      "epoch": 1.2817116854213633,
      "grad_norm": 0.15470421314239502,
      "learning_rate": 3.4406986498569576e-06,
      "loss": 0.0018,
      "step": 199840
    },
    {
      "epoch": 1.2817758223151494,
      "grad_norm": 0.17547038197517395,
      "learning_rate": 3.4401668697928094e-06,
      "loss": 0.0009,
      "step": 199850
    },
    {
      "epoch": 1.2818399592089356,
      "grad_norm": 0.044894181191921234,
      "learning_rate": 3.4396351092744864e-06,
      "loss": 0.0009,
      "step": 199860
    },
    {
      "epoch": 1.2819040961027217,
      "grad_norm": 0.23050405085086823,
      "learning_rate": 3.4391033683086492e-06,
      "loss": 0.0046,
      "step": 199870
    },
    {
      "epoch": 1.2819682329965079,
      "grad_norm": 0.016979113221168518,
      "learning_rate": 3.4385716469019637e-06,
      "loss": 0.0009,
      "step": 199880
    },
    {
      "epoch": 1.2820323698902938,
      "grad_norm": 0.04797173663973808,
      "learning_rate": 3.43803994506109e-06,
      "loss": 0.0027,
      "step": 199890
    },
    {
      "epoch": 1.28209650678408,
      "grad_norm": 0.08218462020158768,
      "learning_rate": 3.4375082627926935e-06,
      "loss": 0.0017,
      "step": 199900
    },
    {
      "epoch": 1.282160643677866,
      "grad_norm": 0.2115473449230194,
      "learning_rate": 3.436976600103434e-06,
      "loss": 0.0009,
      "step": 199910
    },
    {
      "epoch": 1.282224780571652,
      "grad_norm": 0.08777006715536118,
      "learning_rate": 3.436444956999976e-06,
      "loss": 0.0016,
      "step": 199920
    },
    {
      "epoch": 1.2822889174654382,
      "grad_norm": 0.012184024788439274,
      "learning_rate": 3.4359133334889773e-06,
      "loss": 0.0006,
      "step": 199930
    },
    {
      "epoch": 1.2823530543592243,
      "grad_norm": 0.13068903982639313,
      "learning_rate": 3.435381729577104e-06,
      "loss": 0.0024,
      "step": 199940
    },
    {
      "epoch": 1.2824171912530105,
      "grad_norm": 0.04617651551961899,
      "learning_rate": 3.434850145271016e-06,
      "loss": 0.0012,
      "step": 199950
    },
    {
      "epoch": 1.2824813281467966,
      "grad_norm": 0.012335099279880524,
      "learning_rate": 3.434318580577375e-06,
      "loss": 0.0009,
      "step": 199960
    },
    {
      "epoch": 1.2825454650405828,
      "grad_norm": 0.09198637306690216,
      "learning_rate": 3.4337870355028404e-06,
      "loss": 0.0011,
      "step": 199970
    },
    {
      "epoch": 1.2826096019343687,
      "grad_norm": 0.4390782415866852,
      "learning_rate": 3.433255510054074e-06,
      "loss": 0.0034,
      "step": 199980
    },
    {
      "epoch": 1.2826737388281548,
      "grad_norm": 0.08791748434305191,
      "learning_rate": 3.432724004237736e-06,
      "loss": 0.001,
      "step": 199990
    },
    {
      "epoch": 1.282737875721941,
      "grad_norm": 0.1234566867351532,
      "learning_rate": 3.4321925180604863e-06,
      "loss": 0.0023,
      "step": 200000
    },
    {
      "epoch": 1.282802012615727,
      "grad_norm": 0.1424219161272049,
      "learning_rate": 3.431661051528984e-06,
      "loss": 0.0008,
      "step": 200010
    },
    {
      "epoch": 1.282866149509513,
      "grad_norm": 0.10655222088098526,
      "learning_rate": 3.431129604649891e-06,
      "loss": 0.0017,
      "step": 200020
    },
    {
      "epoch": 1.2829302864032992,
      "grad_norm": 0.15433649718761444,
      "learning_rate": 3.4305981774298636e-06,
      "loss": 0.0014,
      "step": 200030
    },
    {
      "epoch": 1.2829944232970854,
      "grad_norm": 0.15786220133304596,
      "learning_rate": 3.430066769875564e-06,
      "loss": 0.0013,
      "step": 200040
    },
    {
      "epoch": 1.2830585601908715,
      "grad_norm": 0.09799902141094208,
      "learning_rate": 3.4295353819936495e-06,
      "loss": 0.0021,
      "step": 200050
    },
    {
      "epoch": 1.2831226970846574,
      "grad_norm": 0.16291195154190063,
      "learning_rate": 3.429004013790779e-06,
      "loss": 0.0034,
      "step": 200060
    },
    {
      "epoch": 1.2831868339784436,
      "grad_norm": 0.19031967222690582,
      "learning_rate": 3.4284726652736113e-06,
      "loss": 0.0027,
      "step": 200070
    },
    {
      "epoch": 1.2832509708722297,
      "grad_norm": 0.1304776966571808,
      "learning_rate": 3.4279413364488036e-06,
      "loss": 0.0008,
      "step": 200080
    },
    {
      "epoch": 1.2833151077660157,
      "grad_norm": 0.04710371419787407,
      "learning_rate": 3.427410027323015e-06,
      "loss": 0.0011,
      "step": 200090
    },
    {
      "epoch": 1.2833792446598018,
      "grad_norm": 0.25472894310951233,
      "learning_rate": 3.426878737902901e-06,
      "loss": 0.0033,
      "step": 200100
    },
    {
      "epoch": 1.283443381553588,
      "grad_norm": 0.05922931805253029,
      "learning_rate": 3.4263474681951233e-06,
      "loss": 0.0013,
      "step": 200110
    },
    {
      "epoch": 1.283507518447374,
      "grad_norm": 0.05359066650271416,
      "learning_rate": 3.4258162182063347e-06,
      "loss": 0.0013,
      "step": 200120
    },
    {
      "epoch": 1.2835716553411602,
      "grad_norm": 0.10407944023609161,
      "learning_rate": 3.425284987943195e-06,
      "loss": 0.0017,
      "step": 200130
    },
    {
      "epoch": 1.2836357922349464,
      "grad_norm": 0.042828936129808426,
      "learning_rate": 3.424753777412359e-06,
      "loss": 0.0007,
      "step": 200140
    },
    {
      "epoch": 1.2836999291287323,
      "grad_norm": 0.21707478165626526,
      "learning_rate": 3.424222586620485e-06,
      "loss": 0.0009,
      "step": 200150
    },
    {
      "epoch": 1.2837640660225185,
      "grad_norm": 0.14766912162303925,
      "learning_rate": 3.423691415574227e-06,
      "loss": 0.0013,
      "step": 200160
    },
    {
      "epoch": 1.2838282029163046,
      "grad_norm": 0.11225715279579163,
      "learning_rate": 3.423160264280243e-06,
      "loss": 0.0018,
      "step": 200170
    },
    {
      "epoch": 1.2838923398100905,
      "grad_norm": 0.09696038067340851,
      "learning_rate": 3.4226291327451877e-06,
      "loss": 0.0013,
      "step": 200180
    },
    {
      "epoch": 1.2839564767038767,
      "grad_norm": 0.04146742448210716,
      "learning_rate": 3.422098020975717e-06,
      "loss": 0.001,
      "step": 200190
    },
    {
      "epoch": 1.2840206135976628,
      "grad_norm": 0.014756908640265465,
      "learning_rate": 3.4215669289784847e-06,
      "loss": 0.0016,
      "step": 200200
    },
    {
      "epoch": 1.284084750491449,
      "grad_norm": 0.06989117711782455,
      "learning_rate": 3.4210358567601477e-06,
      "loss": 0.0009,
      "step": 200210
    },
    {
      "epoch": 1.2841488873852351,
      "grad_norm": 0.14492103457450867,
      "learning_rate": 3.4205048043273594e-06,
      "loss": 0.002,
      "step": 200220
    },
    {
      "epoch": 1.284213024279021,
      "grad_norm": 0.06944374740123749,
      "learning_rate": 3.4199737716867753e-06,
      "loss": 0.0008,
      "step": 200230
    },
    {
      "epoch": 1.2842771611728072,
      "grad_norm": 0.07351551949977875,
      "learning_rate": 3.4194427588450485e-06,
      "loss": 0.0011,
      "step": 200240
    },
    {
      "epoch": 1.2843412980665934,
      "grad_norm": 0.6855619549751282,
      "learning_rate": 3.4189117658088332e-06,
      "loss": 0.0021,
      "step": 200250
    },
    {
      "epoch": 1.2844054349603795,
      "grad_norm": 0.12001349776983261,
      "learning_rate": 3.418380792584785e-06,
      "loss": 0.0008,
      "step": 200260
    },
    {
      "epoch": 1.2844695718541654,
      "grad_norm": 0.04269237071275711,
      "learning_rate": 3.4178498391795546e-06,
      "loss": 0.0025,
      "step": 200270
    },
    {
      "epoch": 1.2845337087479516,
      "grad_norm": 0.04182310402393341,
      "learning_rate": 3.4173189055997973e-06,
      "loss": 0.0008,
      "step": 200280
    },
    {
      "epoch": 1.2845978456417377,
      "grad_norm": 0.17033688724040985,
      "learning_rate": 3.416787991852164e-06,
      "loss": 0.0016,
      "step": 200290
    },
    {
      "epoch": 1.2846619825355239,
      "grad_norm": 0.0695016160607338,
      "learning_rate": 3.41625709794331e-06,
      "loss": 0.001,
      "step": 200300
    },
    {
      "epoch": 1.28472611942931,
      "grad_norm": 0.026342421770095825,
      "learning_rate": 3.4157262238798857e-06,
      "loss": 0.0023,
      "step": 200310
    },
    {
      "epoch": 1.284790256323096,
      "grad_norm": 0.05523712933063507,
      "learning_rate": 3.415195369668545e-06,
      "loss": 0.0005,
      "step": 200320
    },
    {
      "epoch": 1.284854393216882,
      "grad_norm": 0.02777286432683468,
      "learning_rate": 3.4146645353159372e-06,
      "loss": 0.001,
      "step": 200330
    },
    {
      "epoch": 1.2849185301106683,
      "grad_norm": 0.0390302874147892,
      "learning_rate": 3.414133720828717e-06,
      "loss": 0.0014,
      "step": 200340
    },
    {
      "epoch": 1.2849826670044542,
      "grad_norm": 0.03369353339076042,
      "learning_rate": 3.4136029262135345e-06,
      "loss": 0.0009,
      "step": 200350
    },
    {
      "epoch": 1.2850468038982403,
      "grad_norm": 0.047813788056373596,
      "learning_rate": 3.4130721514770416e-06,
      "loss": 0.002,
      "step": 200360
    },
    {
      "epoch": 1.2851109407920265,
      "grad_norm": 0.04628170654177666,
      "learning_rate": 3.412541396625888e-06,
      "loss": 0.0034,
      "step": 200370
    },
    {
      "epoch": 1.2851750776858126,
      "grad_norm": 0.011677786707878113,
      "learning_rate": 3.412010661666726e-06,
      "loss": 0.0015,
      "step": 200380
    },
    {
      "epoch": 1.2852392145795988,
      "grad_norm": 0.11463964730501175,
      "learning_rate": 3.4114799466062043e-06,
      "loss": 0.002,
      "step": 200390
    },
    {
      "epoch": 1.285303351473385,
      "grad_norm": 0.05677622929215431,
      "learning_rate": 3.4109492514509746e-06,
      "loss": 0.0013,
      "step": 200400
    },
    {
      "epoch": 1.2853674883671709,
      "grad_norm": 0.07445462048053741,
      "learning_rate": 3.410418576207687e-06,
      "loss": 0.0026,
      "step": 200410
    },
    {
      "epoch": 1.285431625260957,
      "grad_norm": 0.03686346486210823,
      "learning_rate": 3.4098879208829895e-06,
      "loss": 0.0015,
      "step": 200420
    },
    {
      "epoch": 1.2854957621547431,
      "grad_norm": 0.16804060339927673,
      "learning_rate": 3.4093572854835346e-06,
      "loss": 0.0017,
      "step": 200430
    },
    {
      "epoch": 1.285559899048529,
      "grad_norm": 0.007531581912189722,
      "learning_rate": 3.4088266700159677e-06,
      "loss": 0.0014,
      "step": 200440
    },
    {
      "epoch": 1.2856240359423152,
      "grad_norm": 0.15468910336494446,
      "learning_rate": 3.4082960744869414e-06,
      "loss": 0.0027,
      "step": 200450
    },
    {
      "epoch": 1.2856881728361014,
      "grad_norm": 0.3094685673713684,
      "learning_rate": 3.4077654989031017e-06,
      "loss": 0.0024,
      "step": 200460
    },
    {
      "epoch": 1.2857523097298875,
      "grad_norm": 0.04157442972064018,
      "learning_rate": 3.4072349432710992e-06,
      "loss": 0.0013,
      "step": 200470
    },
    {
      "epoch": 1.2858164466236737,
      "grad_norm": 0.02950865402817726,
      "learning_rate": 3.406704407597581e-06,
      "loss": 0.0009,
      "step": 200480
    },
    {
      "epoch": 1.2858805835174596,
      "grad_norm": 0.016590271145105362,
      "learning_rate": 3.4061738918891952e-06,
      "loss": 0.0022,
      "step": 200490
    },
    {
      "epoch": 1.2859447204112457,
      "grad_norm": 0.026118848472833633,
      "learning_rate": 3.405643396152589e-06,
      "loss": 0.0016,
      "step": 200500
    },
    {
      "epoch": 1.286008857305032,
      "grad_norm": 0.0074649169109761715,
      "learning_rate": 3.4051129203944117e-06,
      "loss": 0.0019,
      "step": 200510
    },
    {
      "epoch": 1.2860729941988178,
      "grad_norm": 0.03584938496351242,
      "learning_rate": 3.4045824646213083e-06,
      "loss": 0.0018,
      "step": 200520
    },
    {
      "epoch": 1.286137131092604,
      "grad_norm": 0.24542315304279327,
      "learning_rate": 3.404052028839927e-06,
      "loss": 0.0039,
      "step": 200530
    },
    {
      "epoch": 1.2862012679863901,
      "grad_norm": 0.12411786615848541,
      "learning_rate": 3.4035216130569147e-06,
      "loss": 0.0005,
      "step": 200540
    },
    {
      "epoch": 1.2862654048801763,
      "grad_norm": 0.03921869397163391,
      "learning_rate": 3.402991217278917e-06,
      "loss": 0.001,
      "step": 200550
    },
    {
      "epoch": 1.2863295417739624,
      "grad_norm": 0.1604154407978058,
      "learning_rate": 3.402460841512582e-06,
      "loss": 0.0021,
      "step": 200560
    },
    {
      "epoch": 1.2863936786677486,
      "grad_norm": 0.048219580203294754,
      "learning_rate": 3.401930485764553e-06,
      "loss": 0.0008,
      "step": 200570
    },
    {
      "epoch": 1.2864578155615345,
      "grad_norm": 0.09091049432754517,
      "learning_rate": 3.4014001500414784e-06,
      "loss": 0.001,
      "step": 200580
    },
    {
      "epoch": 1.2865219524553206,
      "grad_norm": 0.05343707278370857,
      "learning_rate": 3.400869834350001e-06,
      "loss": 0.0017,
      "step": 200590
    },
    {
      "epoch": 1.2865860893491068,
      "grad_norm": 0.013386723585426807,
      "learning_rate": 3.4003395386967692e-06,
      "loss": 0.0023,
      "step": 200600
    },
    {
      "epoch": 1.2866502262428927,
      "grad_norm": 0.09424804896116257,
      "learning_rate": 3.399809263088425e-06,
      "loss": 0.0026,
      "step": 200610
    },
    {
      "epoch": 1.2867143631366789,
      "grad_norm": 0.0879783183336258,
      "learning_rate": 3.3992790075316152e-06,
      "loss": 0.0032,
      "step": 200620
    },
    {
      "epoch": 1.286778500030465,
      "grad_norm": 0.12786969542503357,
      "learning_rate": 3.398748772032982e-06,
      "loss": 0.0017,
      "step": 200630
    },
    {
      "epoch": 1.2868426369242512,
      "grad_norm": 0.11305782198905945,
      "learning_rate": 3.398218556599173e-06,
      "loss": 0.0022,
      "step": 200640
    },
    {
      "epoch": 1.2869067738180373,
      "grad_norm": 0.3130101263523102,
      "learning_rate": 3.3976883612368284e-06,
      "loss": 0.0023,
      "step": 200650
    },
    {
      "epoch": 1.2869709107118232,
      "grad_norm": 0.09633992612361908,
      "learning_rate": 3.397158185952595e-06,
      "loss": 0.002,
      "step": 200660
    },
    {
      "epoch": 1.2870350476056094,
      "grad_norm": 0.05635792389512062,
      "learning_rate": 3.396628030753114e-06,
      "loss": 0.0011,
      "step": 200670
    },
    {
      "epoch": 1.2870991844993955,
      "grad_norm": 0.03625056520104408,
      "learning_rate": 3.39609789564503e-06,
      "loss": 0.0008,
      "step": 200680
    },
    {
      "epoch": 1.2871633213931817,
      "grad_norm": 0.06167403236031532,
      "learning_rate": 3.395567780634987e-06,
      "loss": 0.0015,
      "step": 200690
    },
    {
      "epoch": 1.2872274582869676,
      "grad_norm": 0.036683231592178345,
      "learning_rate": 3.395037685729624e-06,
      "loss": 0.0011,
      "step": 200700
    },
    {
      "epoch": 1.2872915951807538,
      "grad_norm": 0.09551402181386948,
      "learning_rate": 3.3945076109355866e-06,
      "loss": 0.0013,
      "step": 200710
    },
    {
      "epoch": 1.28735573207454,
      "grad_norm": 0.2631330192089081,
      "learning_rate": 3.3939775562595173e-06,
      "loss": 0.0028,
      "step": 200720
    },
    {
      "epoch": 1.287419868968326,
      "grad_norm": 0.014659715816378593,
      "learning_rate": 3.3934475217080563e-06,
      "loss": 0.0024,
      "step": 200730
    },
    {
      "epoch": 1.2874840058621122,
      "grad_norm": 0.08092175424098969,
      "learning_rate": 3.3929175072878475e-06,
      "loss": 0.0029,
      "step": 200740
    },
    {
      "epoch": 1.2875481427558981,
      "grad_norm": 0.17531432211399078,
      "learning_rate": 3.3923875130055284e-06,
      "loss": 0.0032,
      "step": 200750
    },
    {
      "epoch": 1.2876122796496843,
      "grad_norm": 0.06754172593355179,
      "learning_rate": 3.391857538867745e-06,
      "loss": 0.0015,
      "step": 200760
    },
    {
      "epoch": 1.2876764165434704,
      "grad_norm": 0.04346820339560509,
      "learning_rate": 3.3913275848811346e-06,
      "loss": 0.0015,
      "step": 200770
    },
    {
      "epoch": 1.2877405534372564,
      "grad_norm": 0.048482779413461685,
      "learning_rate": 3.39079765105234e-06,
      "loss": 0.0005,
      "step": 200780
    },
    {
      "epoch": 1.2878046903310425,
      "grad_norm": 0.3787684142589569,
      "learning_rate": 3.3902677373880005e-06,
      "loss": 0.0012,
      "step": 200790
    },
    {
      "epoch": 1.2878688272248286,
      "grad_norm": 0.06011801213026047,
      "learning_rate": 3.3897378438947557e-06,
      "loss": 0.0005,
      "step": 200800
    },
    {
      "epoch": 1.2879329641186148,
      "grad_norm": 0.04073334485292435,
      "learning_rate": 3.389207970579249e-06,
      "loss": 0.0016,
      "step": 200810
    },
    {
      "epoch": 1.287997101012401,
      "grad_norm": 0.052258770912885666,
      "learning_rate": 3.3886781174481156e-06,
      "loss": 0.001,
      "step": 200820
    },
    {
      "epoch": 1.288061237906187,
      "grad_norm": 0.06269658356904984,
      "learning_rate": 3.3881482845079983e-06,
      "loss": 0.0022,
      "step": 200830
    },
    {
      "epoch": 1.288125374799973,
      "grad_norm": 0.12791575491428375,
      "learning_rate": 3.3876184717655337e-06,
      "loss": 0.002,
      "step": 200840
    },
    {
      "epoch": 1.2881895116937592,
      "grad_norm": 0.11520896852016449,
      "learning_rate": 3.387088679227364e-06,
      "loss": 0.0034,
      "step": 200850
    },
    {
      "epoch": 1.2882536485875453,
      "grad_norm": 0.059444136917591095,
      "learning_rate": 3.386558906900124e-06,
      "loss": 0.001,
      "step": 200860
    },
    {
      "epoch": 1.2883177854813312,
      "grad_norm": 0.05978221446275711,
      "learning_rate": 3.3860291547904545e-06,
      "loss": 0.001,
      "step": 200870
    },
    {
      "epoch": 1.2883819223751174,
      "grad_norm": 0.06346441805362701,
      "learning_rate": 3.385499422904993e-06,
      "loss": 0.0014,
      "step": 200880
    },
    {
      "epoch": 1.2884460592689035,
      "grad_norm": 0.08050800859928131,
      "learning_rate": 3.384969711250379e-06,
      "loss": 0.0012,
      "step": 200890
    },
    {
      "epoch": 1.2885101961626897,
      "grad_norm": 0.05469054728746414,
      "learning_rate": 3.384440019833247e-06,
      "loss": 0.0017,
      "step": 200900
    },
    {
      "epoch": 1.2885743330564758,
      "grad_norm": 0.262292742729187,
      "learning_rate": 3.3839103486602365e-06,
      "loss": 0.0047,
      "step": 200910
    },
    {
      "epoch": 1.2886384699502618,
      "grad_norm": 0.16505391895771027,
      "learning_rate": 3.3833806977379845e-06,
      "loss": 0.0008,
      "step": 200920
    },
    {
      "epoch": 1.288702606844048,
      "grad_norm": 0.08264858275651932,
      "learning_rate": 3.3828510670731285e-06,
      "loss": 0.0016,
      "step": 200930
    },
    {
      "epoch": 1.288766743737834,
      "grad_norm": 0.05656171962618828,
      "learning_rate": 3.382321456672303e-06,
      "loss": 0.0011,
      "step": 200940
    },
    {
      "epoch": 1.2888308806316202,
      "grad_norm": 0.07540126889944077,
      "learning_rate": 3.3817918665421463e-06,
      "loss": 0.0008,
      "step": 200950
    },
    {
      "epoch": 1.2888950175254061,
      "grad_norm": 0.08787442743778229,
      "learning_rate": 3.381262296689295e-06,
      "loss": 0.001,
      "step": 200960
    },
    {
      "epoch": 1.2889591544191923,
      "grad_norm": 0.06413739174604416,
      "learning_rate": 3.380732747120383e-06,
      "loss": 0.0014,
      "step": 200970
    },
    {
      "epoch": 1.2890232913129784,
      "grad_norm": 0.04403886944055557,
      "learning_rate": 3.3802032178420473e-06,
      "loss": 0.0008,
      "step": 200980
    },
    {
      "epoch": 1.2890874282067646,
      "grad_norm": 0.09438447654247284,
      "learning_rate": 3.379673708860922e-06,
      "loss": 0.0014,
      "step": 200990
    },
    {
      "epoch": 1.2891515651005507,
      "grad_norm": 0.017013145610690117,
      "learning_rate": 3.379144220183645e-06,
      "loss": 0.0009,
      "step": 201000
    },
    {
      "epoch": 1.2892157019943367,
      "grad_norm": 0.013759625144302845,
      "learning_rate": 3.3786147518168476e-06,
      "loss": 0.0014,
      "step": 201010
    },
    {
      "epoch": 1.2892798388881228,
      "grad_norm": 0.11689459532499313,
      "learning_rate": 3.378085303767168e-06,
      "loss": 0.0014,
      "step": 201020
    },
    {
      "epoch": 1.289343975781909,
      "grad_norm": 0.08445636928081512,
      "learning_rate": 3.377555876041237e-06,
      "loss": 0.0017,
      "step": 201030
    },
    {
      "epoch": 1.2894081126756949,
      "grad_norm": 0.02180854044854641,
      "learning_rate": 3.377026468645691e-06,
      "loss": 0.001,
      "step": 201040
    },
    {
      "epoch": 1.289472249569481,
      "grad_norm": 0.04893924295902252,
      "learning_rate": 3.3764970815871633e-06,
      "loss": 0.001,
      "step": 201050
    },
    {
      "epoch": 1.2895363864632672,
      "grad_norm": 0.06638488173484802,
      "learning_rate": 3.375967714872288e-06,
      "loss": 0.0011,
      "step": 201060
    },
    {
      "epoch": 1.2896005233570533,
      "grad_norm": 0.050552599132061005,
      "learning_rate": 3.375438368507697e-06,
      "loss": 0.0018,
      "step": 201070
    },
    {
      "epoch": 1.2896646602508395,
      "grad_norm": 0.43968114256858826,
      "learning_rate": 3.374909042500025e-06,
      "loss": 0.0029,
      "step": 201080
    },
    {
      "epoch": 1.2897287971446256,
      "grad_norm": 0.07450058311223984,
      "learning_rate": 3.3743797368559035e-06,
      "loss": 0.0015,
      "step": 201090
    },
    {
      "epoch": 1.2897929340384116,
      "grad_norm": 0.03740369901061058,
      "learning_rate": 3.3738504515819654e-06,
      "loss": 0.0013,
      "step": 201100
    },
    {
      "epoch": 1.2898570709321977,
      "grad_norm": 0.06631144881248474,
      "learning_rate": 3.3733211866848447e-06,
      "loss": 0.001,
      "step": 201110
    },
    {
      "epoch": 1.2899212078259839,
      "grad_norm": 0.10576901584863663,
      "learning_rate": 3.372791942171171e-06,
      "loss": 0.001,
      "step": 201120
    },
    {
      "epoch": 1.2899853447197698,
      "grad_norm": 0.053370777517557144,
      "learning_rate": 3.3722627180475774e-06,
      "loss": 0.0021,
      "step": 201130
    },
    {
      "epoch": 1.290049481613556,
      "grad_norm": 0.15974250435829163,
      "learning_rate": 3.3717335143206952e-06,
      "loss": 0.0035,
      "step": 201140
    },
    {
      "epoch": 1.290113618507342,
      "grad_norm": 0.03818970546126366,
      "learning_rate": 3.3712043309971567e-06,
      "loss": 0.0014,
      "step": 201150
    },
    {
      "epoch": 1.2901777554011282,
      "grad_norm": 0.12067518383264542,
      "learning_rate": 3.370675168083591e-06,
      "loss": 0.0012,
      "step": 201160
    },
    {
      "epoch": 1.2902418922949144,
      "grad_norm": 0.053740669041872025,
      "learning_rate": 3.37014602558663e-06,
      "loss": 0.0018,
      "step": 201170
    },
    {
      "epoch": 1.2903060291887003,
      "grad_norm": 0.10573643445968628,
      "learning_rate": 3.3696169035129046e-06,
      "loss": 0.0021,
      "step": 201180
    },
    {
      "epoch": 1.2903701660824864,
      "grad_norm": 0.09610594063997269,
      "learning_rate": 3.369087801869045e-06,
      "loss": 0.0009,
      "step": 201190
    },
    {
      "epoch": 1.2904343029762726,
      "grad_norm": 0.0808788537979126,
      "learning_rate": 3.3685587206616793e-06,
      "loss": 0.0017,
      "step": 201200
    },
    {
      "epoch": 1.2904984398700585,
      "grad_norm": 0.03952576592564583,
      "learning_rate": 3.3680296598974406e-06,
      "loss": 0.0009,
      "step": 201210
    },
    {
      "epoch": 1.2905625767638447,
      "grad_norm": 0.24757136404514313,
      "learning_rate": 3.3675006195829554e-06,
      "loss": 0.0027,
      "step": 201220
    },
    {
      "epoch": 1.2906267136576308,
      "grad_norm": 0.15349692106246948,
      "learning_rate": 3.366971599724855e-06,
      "loss": 0.0023,
      "step": 201230
    },
    {
      "epoch": 1.290690850551417,
      "grad_norm": 0.04370679333806038,
      "learning_rate": 3.366442600329766e-06,
      "loss": 0.0014,
      "step": 201240
    },
    {
      "epoch": 1.2907549874452031,
      "grad_norm": 0.03327915817499161,
      "learning_rate": 3.3659136214043197e-06,
      "loss": 0.0013,
      "step": 201250
    },
    {
      "epoch": 1.2908191243389893,
      "grad_norm": 0.11615893989801407,
      "learning_rate": 3.365384662955144e-06,
      "loss": 0.0015,
      "step": 201260
    },
    {
      "epoch": 1.2908832612327752,
      "grad_norm": 0.04019397497177124,
      "learning_rate": 3.3648557249888657e-06,
      "loss": 0.0007,
      "step": 201270
    },
    {
      "epoch": 1.2909473981265613,
      "grad_norm": 0.08190418779850006,
      "learning_rate": 3.364326807512115e-06,
      "loss": 0.0025,
      "step": 201280
    },
    {
      "epoch": 1.2910115350203475,
      "grad_norm": 0.09775541722774506,
      "learning_rate": 3.3637979105315175e-06,
      "loss": 0.0009,
      "step": 201290
    },
    {
      "epoch": 1.2910756719141334,
      "grad_norm": 0.045272570103406906,
      "learning_rate": 3.363269034053702e-06,
      "loss": 0.0012,
      "step": 201300
    },
    {
      "epoch": 1.2911398088079196,
      "grad_norm": 0.12820862233638763,
      "learning_rate": 3.3627401780852952e-06,
      "loss": 0.0024,
      "step": 201310
    },
    {
      "epoch": 1.2912039457017057,
      "grad_norm": 0.16251589357852936,
      "learning_rate": 3.362211342632925e-06,
      "loss": 0.0018,
      "step": 201320
    },
    {
      "epoch": 1.2912680825954919,
      "grad_norm": 0.041628845036029816,
      "learning_rate": 3.3616825277032154e-06,
      "loss": 0.0008,
      "step": 201330
    },
    {
      "epoch": 1.291332219489278,
      "grad_norm": 0.05819320306181908,
      "learning_rate": 3.361153733302796e-06,
      "loss": 0.0023,
      "step": 201340
    },
    {
      "epoch": 1.291396356383064,
      "grad_norm": 0.051456280052661896,
      "learning_rate": 3.3606249594382905e-06,
      "loss": 0.0022,
      "step": 201350
    },
    {
      "epoch": 1.29146049327685,
      "grad_norm": 0.11780291050672531,
      "learning_rate": 3.360096206116327e-06,
      "loss": 0.0022,
      "step": 201360
    },
    {
      "epoch": 1.2915246301706362,
      "grad_norm": 0.0045365807600319386,
      "learning_rate": 3.359567473343529e-06,
      "loss": 0.0007,
      "step": 201370
    },
    {
      "epoch": 1.2915887670644224,
      "grad_norm": 0.04459698870778084,
      "learning_rate": 3.359038761126523e-06,
      "loss": 0.0018,
      "step": 201380
    },
    {
      "epoch": 1.2916529039582083,
      "grad_norm": 0.4114333689212799,
      "learning_rate": 3.358510069471935e-06,
      "loss": 0.0018,
      "step": 201390
    },
    {
      "epoch": 1.2917170408519945,
      "grad_norm": 0.17456431686878204,
      "learning_rate": 3.3579813983863884e-06,
      "loss": 0.0015,
      "step": 201400
    },
    {
      "epoch": 1.2917811777457806,
      "grad_norm": 0.2068919539451599,
      "learning_rate": 3.357452747876509e-06,
      "loss": 0.002,
      "step": 201410
    },
    {
      "epoch": 1.2918453146395668,
      "grad_norm": 0.0049244496040046215,
      "learning_rate": 3.3569241179489197e-06,
      "loss": 0.0018,
      "step": 201420
    },
    {
      "epoch": 1.291909451533353,
      "grad_norm": 0.03238717094063759,
      "learning_rate": 3.356395508610246e-06,
      "loss": 0.001,
      "step": 201430
    },
    {
      "epoch": 1.2919735884271388,
      "grad_norm": 0.15247681736946106,
      "learning_rate": 3.3558669198671113e-06,
      "loss": 0.0014,
      "step": 201440
    },
    {
      "epoch": 1.292037725320925,
      "grad_norm": 0.06962453573942184,
      "learning_rate": 3.3553383517261395e-06,
      "loss": 0.0021,
      "step": 201450
    },
    {
      "epoch": 1.2921018622147111,
      "grad_norm": 0.047201476991176605,
      "learning_rate": 3.3548098041939515e-06,
      "loss": 0.0012,
      "step": 201460
    },
    {
      "epoch": 1.292165999108497,
      "grad_norm": 0.11179137974977493,
      "learning_rate": 3.3542812772771737e-06,
      "loss": 0.0016,
      "step": 201470
    },
    {
      "epoch": 1.2922301360022832,
      "grad_norm": 0.27629971504211426,
      "learning_rate": 3.3537527709824286e-06,
      "loss": 0.0015,
      "step": 201480
    },
    {
      "epoch": 1.2922942728960694,
      "grad_norm": 0.05948624014854431,
      "learning_rate": 3.3532242853163366e-06,
      "loss": 0.0023,
      "step": 201490
    },
    {
      "epoch": 1.2923584097898555,
      "grad_norm": 0.0237751342356205,
      "learning_rate": 3.3526958202855227e-06,
      "loss": 0.0014,
      "step": 201500
    },
    {
      "epoch": 1.2924225466836416,
      "grad_norm": 0.02121659927070141,
      "learning_rate": 3.352167375896606e-06,
      "loss": 0.0013,
      "step": 201510
    },
    {
      "epoch": 1.2924866835774278,
      "grad_norm": 0.08977126330137253,
      "learning_rate": 3.35163895215621e-06,
      "loss": 0.0013,
      "step": 201520
    },
    {
      "epoch": 1.2925508204712137,
      "grad_norm": 0.1702510118484497,
      "learning_rate": 3.3511105490709572e-06,
      "loss": 0.0008,
      "step": 201530
    },
    {
      "epoch": 1.2926149573649999,
      "grad_norm": 0.21891912817955017,
      "learning_rate": 3.3505821666474657e-06,
      "loss": 0.0022,
      "step": 201540
    },
    {
      "epoch": 1.292679094258786,
      "grad_norm": 0.06410212069749832,
      "learning_rate": 3.3500538048923603e-06,
      "loss": 0.0012,
      "step": 201550
    },
    {
      "epoch": 1.292743231152572,
      "grad_norm": 0.070395328104496,
      "learning_rate": 3.3495254638122593e-06,
      "loss": 0.0018,
      "step": 201560
    },
    {
      "epoch": 1.292807368046358,
      "grad_norm": 0.07691259682178497,
      "learning_rate": 3.348997143413785e-06,
      "loss": 0.0026,
      "step": 201570
    },
    {
      "epoch": 1.2928715049401442,
      "grad_norm": 0.10150125622749329,
      "learning_rate": 3.348468843703554e-06,
      "loss": 0.0021,
      "step": 201580
    },
    {
      "epoch": 1.2929356418339304,
      "grad_norm": 0.03191756084561348,
      "learning_rate": 3.347940564688191e-06,
      "loss": 0.0013,
      "step": 201590
    },
    {
      "epoch": 1.2929997787277165,
      "grad_norm": 0.13160976767539978,
      "learning_rate": 3.3474123063743114e-06,
      "loss": 0.0018,
      "step": 201600
    },
    {
      "epoch": 1.2930639156215025,
      "grad_norm": 0.07916834205389023,
      "learning_rate": 3.3468840687685383e-06,
      "loss": 0.0014,
      "step": 201610
    },
    {
      "epoch": 1.2931280525152886,
      "grad_norm": 0.010582217015326023,
      "learning_rate": 3.3463558518774888e-06,
      "loss": 0.0017,
      "step": 201620
    },
    {
      "epoch": 1.2931921894090748,
      "grad_norm": 0.0967269316315651,
      "learning_rate": 3.3458276557077823e-06,
      "loss": 0.0011,
      "step": 201630
    },
    {
      "epoch": 1.2932563263028607,
      "grad_norm": 0.07778430730104446,
      "learning_rate": 3.3452994802660367e-06,
      "loss": 0.0009,
      "step": 201640
    },
    {
      "epoch": 1.2933204631966468,
      "grad_norm": 0.15006621181964874,
      "learning_rate": 3.3447713255588717e-06,
      "loss": 0.0021,
      "step": 201650
    },
    {
      "epoch": 1.293384600090433,
      "grad_norm": 0.02836078777909279,
      "learning_rate": 3.344243191592905e-06,
      "loss": 0.0017,
      "step": 201660
    },
    {
      "epoch": 1.2934487369842191,
      "grad_norm": 0.23487643897533417,
      "learning_rate": 3.343715078374755e-06,
      "loss": 0.0018,
      "step": 201670
    },
    {
      "epoch": 1.2935128738780053,
      "grad_norm": 0.01940734125673771,
      "learning_rate": 3.3431869859110387e-06,
      "loss": 0.0017,
      "step": 201680
    },
    {
      "epoch": 1.2935770107717914,
      "grad_norm": 0.061511676758527756,
      "learning_rate": 3.342658914208373e-06,
      "loss": 0.0017,
      "step": 201690
    },
    {
      "epoch": 1.2936411476655774,
      "grad_norm": 0.049369022250175476,
      "learning_rate": 3.342130863273376e-06,
      "loss": 0.0018,
      "step": 201700
    },
    {
      "epoch": 1.2937052845593635,
      "grad_norm": 0.05317164212465286,
      "learning_rate": 3.3416028331126638e-06,
      "loss": 0.0011,
      "step": 201710
    },
    {
      "epoch": 1.2937694214531497,
      "grad_norm": 0.025014236569404602,
      "learning_rate": 3.3410748237328537e-06,
      "loss": 0.0011,
      "step": 201720
    },
    {
      "epoch": 1.2938335583469356,
      "grad_norm": 0.025604791939258575,
      "learning_rate": 3.340546835140561e-06,
      "loss": 0.0024,
      "step": 201730
    },
    {
      "epoch": 1.2938976952407217,
      "grad_norm": 0.03990999609231949,
      "learning_rate": 3.340018867342404e-06,
      "loss": 0.0015,
      "step": 201740
    },
    {
      "epoch": 1.2939618321345079,
      "grad_norm": 0.10523956269025803,
      "learning_rate": 3.3394909203449953e-06,
      "loss": 0.0012,
      "step": 201750
    },
    {
      "epoch": 1.294025969028294,
      "grad_norm": 0.06308460980653763,
      "learning_rate": 3.338962994154954e-06,
      "loss": 0.0013,
      "step": 201760
    },
    {
      "epoch": 1.2940901059220802,
      "grad_norm": 0.06793212890625,
      "learning_rate": 3.338435088778891e-06,
      "loss": 0.0011,
      "step": 201770
    },
    {
      "epoch": 1.294154242815866,
      "grad_norm": 0.03244560584425926,
      "learning_rate": 3.3379072042234263e-06,
      "loss": 0.0015,
      "step": 201780
    },
    {
      "epoch": 1.2942183797096523,
      "grad_norm": 0.12184517830610275,
      "learning_rate": 3.3373793404951704e-06,
      "loss": 0.0015,
      "step": 201790
    },
    {
      "epoch": 1.2942825166034384,
      "grad_norm": 0.03447624668478966,
      "learning_rate": 3.3368514976007394e-06,
      "loss": 0.0012,
      "step": 201800
    },
    {
      "epoch": 1.2943466534972246,
      "grad_norm": 0.07783801108598709,
      "learning_rate": 3.3363236755467498e-06,
      "loss": 0.0019,
      "step": 201810
    },
    {
      "epoch": 1.2944107903910105,
      "grad_norm": 0.06861250102519989,
      "learning_rate": 3.335795874339812e-06,
      "loss": 0.0014,
      "step": 201820
    },
    {
      "epoch": 1.2944749272847966,
      "grad_norm": 0.03938888758420944,
      "learning_rate": 3.3352680939865417e-06,
      "loss": 0.0024,
      "step": 201830
    },
    {
      "epoch": 1.2945390641785828,
      "grad_norm": 0.09895730018615723,
      "learning_rate": 3.3347403344935513e-06,
      "loss": 0.0012,
      "step": 201840
    },
    {
      "epoch": 1.294603201072369,
      "grad_norm": 0.19778893887996674,
      "learning_rate": 3.334212595867456e-06,
      "loss": 0.002,
      "step": 201850
    },
    {
      "epoch": 1.294667337966155,
      "grad_norm": 0.059761881828308105,
      "learning_rate": 3.333684878114866e-06,
      "loss": 0.0009,
      "step": 201860
    },
    {
      "epoch": 1.294731474859941,
      "grad_norm": 0.2661759555339813,
      "learning_rate": 3.333157181242397e-06,
      "loss": 0.0011,
      "step": 201870
    },
    {
      "epoch": 1.2947956117537271,
      "grad_norm": 0.059345848858356476,
      "learning_rate": 3.332629505256658e-06,
      "loss": 0.0008,
      "step": 201880
    },
    {
      "epoch": 1.2948597486475133,
      "grad_norm": 0.04656009376049042,
      "learning_rate": 3.3321018501642642e-06,
      "loss": 0.0013,
      "step": 201890
    },
    {
      "epoch": 1.2949238855412992,
      "grad_norm": 0.1327023208141327,
      "learning_rate": 3.3315742159718256e-06,
      "loss": 0.0018,
      "step": 201900
    },
    {
      "epoch": 1.2949880224350854,
      "grad_norm": 0.05535271763801575,
      "learning_rate": 3.3310466026859555e-06,
      "loss": 0.001,
      "step": 201910
    },
    {
      "epoch": 1.2950521593288715,
      "grad_norm": 0.20933690667152405,
      "learning_rate": 3.3305190103132627e-06,
      "loss": 0.0011,
      "step": 201920
    },
    {
      "epoch": 1.2951162962226577,
      "grad_norm": 0.6115767359733582,
      "learning_rate": 3.3299914388603614e-06,
      "loss": 0.0018,
      "step": 201930
    },
    {
      "epoch": 1.2951804331164438,
      "grad_norm": 0.09265130013227463,
      "learning_rate": 3.3294638883338592e-06,
      "loss": 0.0005,
      "step": 201940
    },
    {
      "epoch": 1.29524457001023,
      "grad_norm": 0.044155530631542206,
      "learning_rate": 3.328936358740369e-06,
      "loss": 0.0018,
      "step": 201950
    },
    {
      "epoch": 1.295308706904016,
      "grad_norm": 0.0016786607448011637,
      "learning_rate": 3.328408850086502e-06,
      "loss": 0.0014,
      "step": 201960
    },
    {
      "epoch": 1.295372843797802,
      "grad_norm": 0.07835692912340164,
      "learning_rate": 3.3278813623788646e-06,
      "loss": 0.0025,
      "step": 201970
    },
    {
      "epoch": 1.2954369806915882,
      "grad_norm": 0.06830964237451553,
      "learning_rate": 3.3273538956240704e-06,
      "loss": 0.0007,
      "step": 201980
    },
    {
      "epoch": 1.2955011175853741,
      "grad_norm": 0.0658147931098938,
      "learning_rate": 3.326826449828726e-06,
      "loss": 0.002,
      "step": 201990
    },
    {
      "epoch": 1.2955652544791603,
      "grad_norm": 0.13605722784996033,
      "learning_rate": 3.326299024999443e-06,
      "loss": 0.0017,
      "step": 202000
    },
    {
      "epoch": 1.2956293913729464,
      "grad_norm": 0.045622818171978,
      "learning_rate": 3.325771621142829e-06,
      "loss": 0.0013,
      "step": 202010
    },
    {
      "epoch": 1.2956935282667326,
      "grad_norm": 0.07887290418148041,
      "learning_rate": 3.3252442382654937e-06,
      "loss": 0.0015,
      "step": 202020
    },
    {
      "epoch": 1.2957576651605187,
      "grad_norm": 0.006410001777112484,
      "learning_rate": 3.3247168763740446e-06,
      "loss": 0.0016,
      "step": 202030
    },
    {
      "epoch": 1.2958218020543046,
      "grad_norm": 0.05459125339984894,
      "learning_rate": 3.3241895354750903e-06,
      "loss": 0.0009,
      "step": 202040
    },
    {
      "epoch": 1.2958859389480908,
      "grad_norm": 0.02828606776893139,
      "learning_rate": 3.3236622155752386e-06,
      "loss": 0.0005,
      "step": 202050
    },
    {
      "epoch": 1.295950075841877,
      "grad_norm": 0.017393292859196663,
      "learning_rate": 3.323134916681098e-06,
      "loss": 0.0021,
      "step": 202060
    },
    {
      "epoch": 1.2960142127356629,
      "grad_norm": 0.047871265560388565,
      "learning_rate": 3.3226076387992744e-06,
      "loss": 0.0012,
      "step": 202070
    },
    {
      "epoch": 1.296078349629449,
      "grad_norm": 0.35553261637687683,
      "learning_rate": 3.3220803819363765e-06,
      "loss": 0.0012,
      "step": 202080
    },
    {
      "epoch": 1.2961424865232352,
      "grad_norm": 0.13506869971752167,
      "learning_rate": 3.3215531460990103e-06,
      "loss": 0.0011,
      "step": 202090
    },
    {
      "epoch": 1.2962066234170213,
      "grad_norm": 0.09168893843889236,
      "learning_rate": 3.3210259312937826e-06,
      "loss": 0.0029,
      "step": 202100
    },
    {
      "epoch": 1.2962707603108075,
      "grad_norm": 0.09741288423538208,
      "learning_rate": 3.320498737527301e-06,
      "loss": 0.0014,
      "step": 202110
    },
    {
      "epoch": 1.2963348972045936,
      "grad_norm": 0.0015789240133017302,
      "learning_rate": 3.3199715648061693e-06,
      "loss": 0.0009,
      "step": 202120
    },
    {
      "epoch": 1.2963990340983795,
      "grad_norm": 0.08950482308864594,
      "learning_rate": 3.3194444131369957e-06,
      "loss": 0.0005,
      "step": 202130
    },
    {
      "epoch": 1.2964631709921657,
      "grad_norm": 0.0017960354452952743,
      "learning_rate": 3.318917282526384e-06,
      "loss": 0.0028,
      "step": 202140
    },
    {
      "epoch": 1.2965273078859518,
      "grad_norm": 0.06722722947597504,
      "learning_rate": 3.31839017298094e-06,
      "loss": 0.0015,
      "step": 202150
    },
    {
      "epoch": 1.2965914447797378,
      "grad_norm": 0.3773379623889923,
      "learning_rate": 3.317863084507269e-06,
      "loss": 0.0023,
      "step": 202160
    },
    {
      "epoch": 1.296655581673524,
      "grad_norm": 0.09458949416875839,
      "learning_rate": 3.3173360171119766e-06,
      "loss": 0.0017,
      "step": 202170
    },
    {
      "epoch": 1.29671971856731,
      "grad_norm": 0.02694663032889366,
      "learning_rate": 3.316808970801666e-06,
      "loss": 0.0018,
      "step": 202180
    },
    {
      "epoch": 1.2967838554610962,
      "grad_norm": 0.07020226866006851,
      "learning_rate": 3.316281945582942e-06,
      "loss": 0.0019,
      "step": 202190
    },
    {
      "epoch": 1.2968479923548824,
      "grad_norm": 0.20809036493301392,
      "learning_rate": 3.315754941462408e-06,
      "loss": 0.0018,
      "step": 202200
    },
    {
      "epoch": 1.2969121292486683,
      "grad_norm": 0.07734943926334381,
      "learning_rate": 3.3152279584466697e-06,
      "loss": 0.002,
      "step": 202210
    },
    {
      "epoch": 1.2969762661424544,
      "grad_norm": 0.08855696767568588,
      "learning_rate": 3.3147009965423273e-06,
      "loss": 0.0023,
      "step": 202220
    },
    {
      "epoch": 1.2970404030362406,
      "grad_norm": 0.0852886363863945,
      "learning_rate": 3.3141740557559854e-06,
      "loss": 0.0009,
      "step": 202230
    },
    {
      "epoch": 1.2971045399300267,
      "grad_norm": 0.06497237831354141,
      "learning_rate": 3.313647136094248e-06,
      "loss": 0.0024,
      "step": 202240
    },
    {
      "epoch": 1.2971686768238126,
      "grad_norm": 0.1269858032464981,
      "learning_rate": 3.3131202375637183e-06,
      "loss": 0.0011,
      "step": 202250
    },
    {
      "epoch": 1.2972328137175988,
      "grad_norm": 0.03508013114333153,
      "learning_rate": 3.312593360170997e-06,
      "loss": 0.0019,
      "step": 202260
    },
    {
      "epoch": 1.297296950611385,
      "grad_norm": 0.07374950498342514,
      "learning_rate": 3.312066503922688e-06,
      "loss": 0.0017,
      "step": 202270
    },
    {
      "epoch": 1.297361087505171,
      "grad_norm": 0.01798911578953266,
      "learning_rate": 3.3115396688253907e-06,
      "loss": 0.0023,
      "step": 202280
    },
    {
      "epoch": 1.2974252243989572,
      "grad_norm": 0.17165930569171906,
      "learning_rate": 3.311012854885709e-06,
      "loss": 0.0015,
      "step": 202290
    },
    {
      "epoch": 1.2974893612927432,
      "grad_norm": 0.32450318336486816,
      "learning_rate": 3.3104860621102426e-06,
      "loss": 0.0015,
      "step": 202300
    },
    {
      "epoch": 1.2975534981865293,
      "grad_norm": 0.03590445592999458,
      "learning_rate": 3.3099592905055944e-06,
      "loss": 0.0017,
      "step": 202310
    },
    {
      "epoch": 1.2976176350803155,
      "grad_norm": 0.11629247665405273,
      "learning_rate": 3.3094325400783627e-06,
      "loss": 0.0019,
      "step": 202320
    },
    {
      "epoch": 1.2976817719741014,
      "grad_norm": 0.0783039852976799,
      "learning_rate": 3.3089058108351507e-06,
      "loss": 0.0017,
      "step": 202330
    },
    {
      "epoch": 1.2977459088678875,
      "grad_norm": 0.1197119727730751,
      "learning_rate": 3.308379102782556e-06,
      "loss": 0.0012,
      "step": 202340
    },
    {
      "epoch": 1.2978100457616737,
      "grad_norm": 0.0840274766087532,
      "learning_rate": 3.3078524159271817e-06,
      "loss": 0.0014,
      "step": 202350
    },
    {
      "epoch": 1.2978741826554598,
      "grad_norm": 0.06660163402557373,
      "learning_rate": 3.307325750275625e-06,
      "loss": 0.0027,
      "step": 202360
    },
    {
      "epoch": 1.297938319549246,
      "grad_norm": 0.10205428302288055,
      "learning_rate": 3.3067991058344866e-06,
      "loss": 0.001,
      "step": 202370
    },
    {
      "epoch": 1.2980024564430321,
      "grad_norm": 0.2115112692117691,
      "learning_rate": 3.3062724826103664e-06,
      "loss": 0.0028,
      "step": 202380
    },
    {
      "epoch": 1.298066593336818,
      "grad_norm": 0.44339513778686523,
      "learning_rate": 3.3057458806098616e-06,
      "loss": 0.0019,
      "step": 202390
    },
    {
      "epoch": 1.2981307302306042,
      "grad_norm": 0.007985997945070267,
      "learning_rate": 3.3052192998395722e-06,
      "loss": 0.0015,
      "step": 202400
    },
    {
      "epoch": 1.2981948671243904,
      "grad_norm": 0.020172620192170143,
      "learning_rate": 3.304692740306096e-06,
      "loss": 0.0016,
      "step": 202410
    },
    {
      "epoch": 1.2982590040181763,
      "grad_norm": 0.015433751977980137,
      "learning_rate": 3.3041662020160314e-06,
      "loss": 0.0011,
      "step": 202420
    },
    {
      "epoch": 1.2983231409119624,
      "grad_norm": 0.05586683005094528,
      "learning_rate": 3.3036396849759755e-06,
      "loss": 0.0019,
      "step": 202430
    },
    {
      "epoch": 1.2983872778057486,
      "grad_norm": 0.04309661686420441,
      "learning_rate": 3.303113189192528e-06,
      "loss": 0.0018,
      "step": 202440
    },
    {
      "epoch": 1.2984514146995347,
      "grad_norm": 0.09828870743513107,
      "learning_rate": 3.3025867146722844e-06,
      "loss": 0.0009,
      "step": 202450
    },
    {
      "epoch": 1.2985155515933209,
      "grad_norm": 0.04972812533378601,
      "learning_rate": 3.3020602614218434e-06,
      "loss": 0.001,
      "step": 202460
    },
    {
      "epoch": 1.2985796884871068,
      "grad_norm": 0.17699630558490753,
      "learning_rate": 3.301533829447799e-06,
      "loss": 0.0019,
      "step": 202470
    },
    {
      "epoch": 1.298643825380893,
      "grad_norm": 0.03271789848804474,
      "learning_rate": 3.301007418756751e-06,
      "loss": 0.002,
      "step": 202480
    },
    {
      "epoch": 1.298707962274679,
      "grad_norm": 0.02394431084394455,
      "learning_rate": 3.3004810293552936e-06,
      "loss": 0.0015,
      "step": 202490
    },
    {
      "epoch": 1.2987720991684653,
      "grad_norm": 0.13419033586978912,
      "learning_rate": 3.299954661250023e-06,
      "loss": 0.0022,
      "step": 202500
    },
    {
      "epoch": 1.2988362360622512,
      "grad_norm": 0.11430850625038147,
      "learning_rate": 3.2994283144475368e-06,
      "loss": 0.0014,
      "step": 202510
    },
    {
      "epoch": 1.2989003729560373,
      "grad_norm": 0.007823416963219643,
      "learning_rate": 3.298901988954428e-06,
      "loss": 0.001,
      "step": 202520
    },
    {
      "epoch": 1.2989645098498235,
      "grad_norm": 0.10548365116119385,
      "learning_rate": 3.298375684777294e-06,
      "loss": 0.002,
      "step": 202530
    },
    {
      "epoch": 1.2990286467436096,
      "grad_norm": 0.031271420419216156,
      "learning_rate": 3.297849401922728e-06,
      "loss": 0.0009,
      "step": 202540
    },
    {
      "epoch": 1.2990927836373958,
      "grad_norm": 0.04257241263985634,
      "learning_rate": 3.2973231403973265e-06,
      "loss": 0.0017,
      "step": 202550
    },
    {
      "epoch": 1.2991569205311817,
      "grad_norm": 0.12754876911640167,
      "learning_rate": 3.2967969002076816e-06,
      "loss": 0.0011,
      "step": 202560
    },
    {
      "epoch": 1.2992210574249679,
      "grad_norm": 0.111076220870018,
      "learning_rate": 3.2962706813603904e-06,
      "loss": 0.0022,
      "step": 202570
    },
    {
      "epoch": 1.299285194318754,
      "grad_norm": 0.05372251942753792,
      "learning_rate": 3.295744483862044e-06,
      "loss": 0.0011,
      "step": 202580
    },
    {
      "epoch": 1.29934933121254,
      "grad_norm": 0.1858329176902771,
      "learning_rate": 3.295218307719238e-06,
      "loss": 0.0033,
      "step": 202590
    },
    {
      "epoch": 1.299413468106326,
      "grad_norm": 0.1459503173828125,
      "learning_rate": 3.2946921529385645e-06,
      "loss": 0.0017,
      "step": 202600
    },
    {
      "epoch": 1.2994776050001122,
      "grad_norm": 0.17990605533123016,
      "learning_rate": 3.294166019526617e-06,
      "loss": 0.003,
      "step": 202610
    },
    {
      "epoch": 1.2995417418938984,
      "grad_norm": 0.5927944183349609,
      "learning_rate": 3.2936399074899884e-06,
      "loss": 0.0012,
      "step": 202620
    },
    {
      "epoch": 1.2996058787876845,
      "grad_norm": 0.11808649450540543,
      "learning_rate": 3.2931138168352715e-06,
      "loss": 0.0015,
      "step": 202630
    },
    {
      "epoch": 1.2996700156814707,
      "grad_norm": 0.05937317758798599,
      "learning_rate": 3.292587747569058e-06,
      "loss": 0.0011,
      "step": 202640
    },
    {
      "epoch": 1.2997341525752566,
      "grad_norm": 0.0408489927649498,
      "learning_rate": 3.2920616996979406e-06,
      "loss": 0.0019,
      "step": 202650
    },
    {
      "epoch": 1.2997982894690427,
      "grad_norm": 0.1323373168706894,
      "learning_rate": 3.2915356732285116e-06,
      "loss": 0.0011,
      "step": 202660
    },
    {
      "epoch": 1.299862426362829,
      "grad_norm": 0.28618696331977844,
      "learning_rate": 3.29100966816736e-06,
      "loss": 0.0021,
      "step": 202670
    },
    {
      "epoch": 1.2999265632566148,
      "grad_norm": 0.06622526049613953,
      "learning_rate": 3.2904836845210796e-06,
      "loss": 0.0008,
      "step": 202680
    },
    {
      "epoch": 1.299990700150401,
      "grad_norm": 0.017771361395716667,
      "learning_rate": 3.2899577222962604e-06,
      "loss": 0.0011,
      "step": 202690
    },
    {
      "epoch": 1.3000548370441871,
      "grad_norm": 0.04204076528549194,
      "learning_rate": 3.2894317814994934e-06,
      "loss": 0.0011,
      "step": 202700
    },
    {
      "epoch": 1.3001189739379733,
      "grad_norm": 0.05336964130401611,
      "learning_rate": 3.2889058621373674e-06,
      "loss": 0.0015,
      "step": 202710
    },
    {
      "epoch": 1.3001831108317594,
      "grad_norm": 0.0902399867773056,
      "learning_rate": 3.288379964216476e-06,
      "loss": 0.0011,
      "step": 202720
    },
    {
      "epoch": 1.3002472477255453,
      "grad_norm": 0.1343262940645218,
      "learning_rate": 3.287854087743405e-06,
      "loss": 0.0015,
      "step": 202730
    },
    {
      "epoch": 1.3003113846193315,
      "grad_norm": 0.10670477896928787,
      "learning_rate": 3.287328232724747e-06,
      "loss": 0.0011,
      "step": 202740
    },
    {
      "epoch": 1.3003755215131176,
      "grad_norm": 0.01778768189251423,
      "learning_rate": 3.2868023991670895e-06,
      "loss": 0.0013,
      "step": 202750
    },
    {
      "epoch": 1.3004396584069036,
      "grad_norm": 0.06783147901296616,
      "learning_rate": 3.286276587077023e-06,
      "loss": 0.0013,
      "step": 202760
    },
    {
      "epoch": 1.3005037953006897,
      "grad_norm": 0.045526184141635895,
      "learning_rate": 3.2857507964611347e-06,
      "loss": 0.0011,
      "step": 202770
    },
    {
      "epoch": 1.3005679321944759,
      "grad_norm": 0.04004519432783127,
      "learning_rate": 3.2852250273260155e-06,
      "loss": 0.003,
      "step": 202780
    },
    {
      "epoch": 1.300632069088262,
      "grad_norm": 0.08523906022310257,
      "learning_rate": 3.2846992796782507e-06,
      "loss": 0.0016,
      "step": 202790
    },
    {
      "epoch": 1.3006962059820482,
      "grad_norm": 0.14614760875701904,
      "learning_rate": 3.28417355352443e-06,
      "loss": 0.0009,
      "step": 202800
    },
    {
      "epoch": 1.3007603428758343,
      "grad_norm": 0.08484523743391037,
      "learning_rate": 3.283647848871142e-06,
      "loss": 0.0009,
      "step": 202810
    },
    {
      "epoch": 1.3008244797696202,
      "grad_norm": 0.08262226730585098,
      "learning_rate": 3.2831221657249723e-06,
      "loss": 0.0014,
      "step": 202820
    },
    {
      "epoch": 1.3008886166634064,
      "grad_norm": 0.09609952569007874,
      "learning_rate": 3.28259650409251e-06,
      "loss": 0.001,
      "step": 202830
    },
    {
      "epoch": 1.3009527535571925,
      "grad_norm": 0.036871932446956635,
      "learning_rate": 3.2820708639803396e-06,
      "loss": 0.0021,
      "step": 202840
    },
    {
      "epoch": 1.3010168904509785,
      "grad_norm": 0.218247190117836,
      "learning_rate": 3.28154524539505e-06,
      "loss": 0.0014,
      "step": 202850
    },
    {
      "epoch": 1.3010810273447646,
      "grad_norm": 0.2301071435213089,
      "learning_rate": 3.2810196483432255e-06,
      "loss": 0.0017,
      "step": 202860
    },
    {
      "epoch": 1.3011451642385508,
      "grad_norm": 0.07522774487733841,
      "learning_rate": 3.280494072831455e-06,
      "loss": 0.0007,
      "step": 202870
    },
    {
      "epoch": 1.301209301132337,
      "grad_norm": 0.11489323526620865,
      "learning_rate": 3.279968518866321e-06,
      "loss": 0.0028,
      "step": 202880
    },
    {
      "epoch": 1.301273438026123,
      "grad_norm": 0.01686144433915615,
      "learning_rate": 3.2794429864544126e-06,
      "loss": 0.0012,
      "step": 202890
    },
    {
      "epoch": 1.301337574919909,
      "grad_norm": 0.011467288248240948,
      "learning_rate": 3.278917475602311e-06,
      "loss": 0.0017,
      "step": 202900
    },
    {
      "epoch": 1.3014017118136951,
      "grad_norm": 0.13039171695709229,
      "learning_rate": 3.278391986316606e-06,
      "loss": 0.0011,
      "step": 202910
    },
    {
      "epoch": 1.3014658487074813,
      "grad_norm": 0.019838912412524223,
      "learning_rate": 3.2778665186038776e-06,
      "loss": 0.0009,
      "step": 202920
    },
    {
      "epoch": 1.3015299856012674,
      "grad_norm": 0.05785344913601875,
      "learning_rate": 3.277341072470714e-06,
      "loss": 0.0015,
      "step": 202930
    },
    {
      "epoch": 1.3015941224950534,
      "grad_norm": 0.04003293067216873,
      "learning_rate": 3.276815647923697e-06,
      "loss": 0.0016,
      "step": 202940
    },
    {
      "epoch": 1.3016582593888395,
      "grad_norm": 0.11377974599599838,
      "learning_rate": 3.276290244969411e-06,
      "loss": 0.0015,
      "step": 202950
    },
    {
      "epoch": 1.3017223962826256,
      "grad_norm": 0.025902625173330307,
      "learning_rate": 3.2757648636144413e-06,
      "loss": 0.0017,
      "step": 202960
    },
    {
      "epoch": 1.3017865331764118,
      "grad_norm": 0.08864615112543106,
      "learning_rate": 3.2752395038653693e-06,
      "loss": 0.0005,
      "step": 202970
    },
    {
      "epoch": 1.301850670070198,
      "grad_norm": 0.12506291270256042,
      "learning_rate": 3.2747141657287806e-06,
      "loss": 0.0026,
      "step": 202980
    },
    {
      "epoch": 1.3019148069639839,
      "grad_norm": 0.05017014592885971,
      "learning_rate": 3.2741888492112534e-06,
      "loss": 0.0016,
      "step": 202990
    },
    {
      "epoch": 1.30197894385777,
      "grad_norm": 0.0649946928024292,
      "learning_rate": 3.2736635543193753e-06,
      "loss": 0.002,
      "step": 203000
    },
    {
      "epoch": 1.3020430807515562,
      "grad_norm": 0.3042117655277252,
      "learning_rate": 3.2731382810597267e-06,
      "loss": 0.0015,
      "step": 203010
    },
    {
      "epoch": 1.302107217645342,
      "grad_norm": 0.013076065108180046,
      "learning_rate": 3.2726130294388892e-06,
      "loss": 0.0011,
      "step": 203020
    },
    {
      "epoch": 1.3021713545391282,
      "grad_norm": 0.12743835151195526,
      "learning_rate": 3.272087799463446e-06,
      "loss": 0.0014,
      "step": 203030
    },
    {
      "epoch": 1.3022354914329144,
      "grad_norm": 0.04182353615760803,
      "learning_rate": 3.271562591139976e-06,
      "loss": 0.0012,
      "step": 203040
    },
    {
      "epoch": 1.3022996283267005,
      "grad_norm": 0.13796482980251312,
      "learning_rate": 3.2710374044750638e-06,
      "loss": 0.0015,
      "step": 203050
    },
    {
      "epoch": 1.3023637652204867,
      "grad_norm": 0.021277979016304016,
      "learning_rate": 3.270512239475287e-06,
      "loss": 0.0012,
      "step": 203060
    },
    {
      "epoch": 1.3024279021142728,
      "grad_norm": 0.003118517342954874,
      "learning_rate": 3.269987096147228e-06,
      "loss": 0.0019,
      "step": 203070
    },
    {
      "epoch": 1.3024920390080588,
      "grad_norm": 0.042708396911621094,
      "learning_rate": 3.269461974497468e-06,
      "loss": 0.001,
      "step": 203080
    },
    {
      "epoch": 1.302556175901845,
      "grad_norm": 0.12176183611154556,
      "learning_rate": 3.2689368745325854e-06,
      "loss": 0.0033,
      "step": 203090
    },
    {
      "epoch": 1.302620312795631,
      "grad_norm": 0.09801366925239563,
      "learning_rate": 3.2684117962591624e-06,
      "loss": 0.0009,
      "step": 203100
    },
    {
      "epoch": 1.302684449689417,
      "grad_norm": 0.020501457154750824,
      "learning_rate": 3.2678867396837753e-06,
      "loss": 0.0012,
      "step": 203110
    },
    {
      "epoch": 1.3027485865832031,
      "grad_norm": 0.1360543817281723,
      "learning_rate": 3.267361704813007e-06,
      "loss": 0.0013,
      "step": 203120
    },
    {
      "epoch": 1.3028127234769893,
      "grad_norm": 0.16810686886310577,
      "learning_rate": 3.2668366916534334e-06,
      "loss": 0.0016,
      "step": 203130
    },
    {
      "epoch": 1.3028768603707754,
      "grad_norm": 0.05818602070212364,
      "learning_rate": 3.2663117002116363e-06,
      "loss": 0.0012,
      "step": 203140
    },
    {
      "epoch": 1.3029409972645616,
      "grad_norm": 0.026946822181344032,
      "learning_rate": 3.2657867304941913e-06,
      "loss": 0.0012,
      "step": 203150
    },
    {
      "epoch": 1.3030051341583475,
      "grad_norm": 0.01007855124771595,
      "learning_rate": 3.265261782507679e-06,
      "loss": 0.001,
      "step": 203160
    },
    {
      "epoch": 1.3030692710521337,
      "grad_norm": 0.06413888186216354,
      "learning_rate": 3.264736856258675e-06,
      "loss": 0.0008,
      "step": 203170
    },
    {
      "epoch": 1.3031334079459198,
      "grad_norm": 0.09537085145711899,
      "learning_rate": 3.2642119517537597e-06,
      "loss": 0.0012,
      "step": 203180
    },
    {
      "epoch": 1.3031975448397057,
      "grad_norm": 0.1698506772518158,
      "learning_rate": 3.263687068999508e-06,
      "loss": 0.0013,
      "step": 203190
    },
    {
      "epoch": 1.3032616817334919,
      "grad_norm": 0.09424944967031479,
      "learning_rate": 3.2631622080024992e-06,
      "loss": 0.0018,
      "step": 203200
    },
    {
      "epoch": 1.303325818627278,
      "grad_norm": 0.040745921432971954,
      "learning_rate": 3.262637368769309e-06,
      "loss": 0.0012,
      "step": 203210
    },
    {
      "epoch": 1.3033899555210642,
      "grad_norm": 0.05310925841331482,
      "learning_rate": 3.2621125513065132e-06,
      "loss": 0.001,
      "step": 203220
    },
    {
      "epoch": 1.3034540924148503,
      "grad_norm": 0.06866107881069183,
      "learning_rate": 3.261587755620691e-06,
      "loss": 0.0005,
      "step": 203230
    },
    {
      "epoch": 1.3035182293086365,
      "grad_norm": 0.028232771903276443,
      "learning_rate": 3.261062981718415e-06,
      "loss": 0.0016,
      "step": 203240
    },
    {
      "epoch": 1.3035823662024224,
      "grad_norm": 0.11395175009965897,
      "learning_rate": 3.2605382296062643e-06,
      "loss": 0.0014,
      "step": 203250
    },
    {
      "epoch": 1.3036465030962086,
      "grad_norm": 0.05476236343383789,
      "learning_rate": 3.2600134992908115e-06,
      "loss": 0.0016,
      "step": 203260
    },
    {
      "epoch": 1.3037106399899947,
      "grad_norm": 0.01740194670855999,
      "learning_rate": 3.259488790778634e-06,
      "loss": 0.0035,
      "step": 203270
    },
    {
      "epoch": 1.3037747768837806,
      "grad_norm": 0.05768860876560211,
      "learning_rate": 3.258964104076305e-06,
      "loss": 0.0014,
      "step": 203280
    },
    {
      "epoch": 1.3038389137775668,
      "grad_norm": 0.013930111192166805,
      "learning_rate": 3.2584394391904008e-06,
      "loss": 0.0011,
      "step": 203290
    },
    {
      "epoch": 1.303903050671353,
      "grad_norm": 0.040215425193309784,
      "learning_rate": 3.257914796127494e-06,
      "loss": 0.0015,
      "step": 203300
    },
    {
      "epoch": 1.303967187565139,
      "grad_norm": 0.0704159140586853,
      "learning_rate": 3.2573901748941615e-06,
      "loss": 0.0012,
      "step": 203310
    },
    {
      "epoch": 1.3040313244589252,
      "grad_norm": 0.03699749708175659,
      "learning_rate": 3.2568655754969736e-06,
      "loss": 0.0011,
      "step": 203320
    },
    {
      "epoch": 1.3040954613527111,
      "grad_norm": 0.007879172451794147,
      "learning_rate": 3.2563409979425076e-06,
      "loss": 0.0043,
      "step": 203330
    },
    {
      "epoch": 1.3041595982464973,
      "grad_norm": 0.057002849876880646,
      "learning_rate": 3.255816442237334e-06,
      "loss": 0.0023,
      "step": 203340
    },
    {
      "epoch": 1.3042237351402834,
      "grad_norm": 0.05223952978849411,
      "learning_rate": 3.2552919083880273e-06,
      "loss": 0.0034,
      "step": 203350
    },
    {
      "epoch": 1.3042878720340696,
      "grad_norm": 0.14209382236003876,
      "learning_rate": 3.25476739640116e-06,
      "loss": 0.0014,
      "step": 203360
    },
    {
      "epoch": 1.3043520089278555,
      "grad_norm": 0.1263195425271988,
      "learning_rate": 3.254242906283304e-06,
      "loss": 0.0018,
      "step": 203370
    },
    {
      "epoch": 1.3044161458216417,
      "grad_norm": 0.008843827061355114,
      "learning_rate": 3.2537184380410324e-06,
      "loss": 0.0018,
      "step": 203380
    },
    {
      "epoch": 1.3044802827154278,
      "grad_norm": 0.051901768893003464,
      "learning_rate": 3.253193991680917e-06,
      "loss": 0.0041,
      "step": 203390
    },
    {
      "epoch": 1.304544419609214,
      "grad_norm": 0.04232050105929375,
      "learning_rate": 3.2526695672095297e-06,
      "loss": 0.0014,
      "step": 203400
    },
    {
      "epoch": 1.3046085565030001,
      "grad_norm": 0.04878690093755722,
      "learning_rate": 3.25214516463344e-06,
      "loss": 0.0014,
      "step": 203410
    },
    {
      "epoch": 1.304672693396786,
      "grad_norm": 0.04045327007770538,
      "learning_rate": 3.2516207839592225e-06,
      "loss": 0.0015,
      "step": 203420
    },
    {
      "epoch": 1.3047368302905722,
      "grad_norm": 0.024009767919778824,
      "learning_rate": 3.2510964251934447e-06,
      "loss": 0.002,
      "step": 203430
    },
    {
      "epoch": 1.3048009671843583,
      "grad_norm": 0.05732010677456856,
      "learning_rate": 3.2505720883426793e-06,
      "loss": 0.0012,
      "step": 203440
    },
    {
      "epoch": 1.3048651040781443,
      "grad_norm": 0.025046516209840775,
      "learning_rate": 3.2500477734134954e-06,
      "loss": 0.0013,
      "step": 203450
    },
    {
      "epoch": 1.3049292409719304,
      "grad_norm": 0.05671786889433861,
      "learning_rate": 3.2495234804124644e-06,
      "loss": 0.0011,
      "step": 203460
    },
    {
      "epoch": 1.3049933778657166,
      "grad_norm": 0.019109154120087624,
      "learning_rate": 3.248999209346154e-06,
      "loss": 0.0007,
      "step": 203470
    },
    {
      "epoch": 1.3050575147595027,
      "grad_norm": 0.09901275485754013,
      "learning_rate": 3.2484749602211363e-06,
      "loss": 0.0016,
      "step": 203480
    },
    {
      "epoch": 1.3051216516532889,
      "grad_norm": 0.03167424350976944,
      "learning_rate": 3.2479507330439788e-06,
      "loss": 0.0016,
      "step": 203490
    },
    {
      "epoch": 1.305185788547075,
      "grad_norm": 0.08982674032449722,
      "learning_rate": 3.2474265278212495e-06,
      "loss": 0.0014,
      "step": 203500
    },
    {
      "epoch": 1.305249925440861,
      "grad_norm": 0.08459123969078064,
      "learning_rate": 3.2469023445595197e-06,
      "loss": 0.0012,
      "step": 203510
    },
    {
      "epoch": 1.305314062334647,
      "grad_norm": 0.025580601766705513,
      "learning_rate": 3.246378183265356e-06,
      "loss": 0.0019,
      "step": 203520
    },
    {
      "epoch": 1.3053781992284332,
      "grad_norm": 0.023656470701098442,
      "learning_rate": 3.2458540439453273e-06,
      "loss": 0.0016,
      "step": 203530
    },
    {
      "epoch": 1.3054423361222192,
      "grad_norm": 0.13090649247169495,
      "learning_rate": 3.245329926606e-06,
      "loss": 0.0013,
      "step": 203540
    },
    {
      "epoch": 1.3055064730160053,
      "grad_norm": 0.02131395787000656,
      "learning_rate": 3.244805831253944e-06,
      "loss": 0.0009,
      "step": 203550
    },
    {
      "epoch": 1.3055706099097915,
      "grad_norm": 0.022348342463374138,
      "learning_rate": 3.2442817578957253e-06,
      "loss": 0.0013,
      "step": 203560
    },
    {
      "epoch": 1.3056347468035776,
      "grad_norm": 0.03129712492227554,
      "learning_rate": 3.243757706537911e-06,
      "loss": 0.0006,
      "step": 203570
    },
    {
      "epoch": 1.3056988836973638,
      "grad_norm": 0.09756151586771011,
      "learning_rate": 3.243233677187067e-06,
      "loss": 0.0034,
      "step": 203580
    },
    {
      "epoch": 1.3057630205911497,
      "grad_norm": 0.086264967918396,
      "learning_rate": 3.242709669849762e-06,
      "loss": 0.0018,
      "step": 203590
    },
    {
      "epoch": 1.3058271574849358,
      "grad_norm": 0.0013568139402195811,
      "learning_rate": 3.242185684532559e-06,
      "loss": 0.0017,
      "step": 203600
    },
    {
      "epoch": 1.305891294378722,
      "grad_norm": 0.0020245579071342945,
      "learning_rate": 3.2416617212420263e-06,
      "loss": 0.002,
      "step": 203610
    },
    {
      "epoch": 1.305955431272508,
      "grad_norm": 0.0683000385761261,
      "learning_rate": 3.241137779984729e-06,
      "loss": 0.0009,
      "step": 203620
    },
    {
      "epoch": 1.306019568166294,
      "grad_norm": 0.02258705347776413,
      "learning_rate": 3.2406138607672334e-06,
      "loss": 0.0016,
      "step": 203630
    },
    {
      "epoch": 1.3060837050600802,
      "grad_norm": 0.06761499494314194,
      "learning_rate": 3.2400899635961014e-06,
      "loss": 0.0012,
      "step": 203640
    },
    {
      "epoch": 1.3061478419538664,
      "grad_norm": 0.20406056940555573,
      "learning_rate": 3.2395660884779007e-06,
      "loss": 0.0027,
      "step": 203650
    },
    {
      "epoch": 1.3062119788476525,
      "grad_norm": 0.12400095909833908,
      "learning_rate": 3.2390422354191964e-06,
      "loss": 0.001,
      "step": 203660
    },
    {
      "epoch": 1.3062761157414386,
      "grad_norm": 0.054169487208127975,
      "learning_rate": 3.2385184044265495e-06,
      "loss": 0.001,
      "step": 203670
    },
    {
      "epoch": 1.3063402526352246,
      "grad_norm": 0.02641112729907036,
      "learning_rate": 3.237994595506527e-06,
      "loss": 0.0014,
      "step": 203680
    },
    {
      "epoch": 1.3064043895290107,
      "grad_norm": 0.11586663126945496,
      "learning_rate": 3.2374708086656916e-06,
      "loss": 0.0023,
      "step": 203690
    },
    {
      "epoch": 1.3064685264227969,
      "grad_norm": 0.11719394475221634,
      "learning_rate": 3.2369470439106065e-06,
      "loss": 0.0022,
      "step": 203700
    },
    {
      "epoch": 1.3065326633165828,
      "grad_norm": 0.1640966832637787,
      "learning_rate": 3.236423301247835e-06,
      "loss": 0.0021,
      "step": 203710
    },
    {
      "epoch": 1.306596800210369,
      "grad_norm": 0.08211014419794083,
      "learning_rate": 3.23589958068394e-06,
      "loss": 0.0021,
      "step": 203720
    },
    {
      "epoch": 1.306660937104155,
      "grad_norm": 0.03999345377087593,
      "learning_rate": 3.235375882225483e-06,
      "loss": 0.0009,
      "step": 203730
    },
    {
      "epoch": 1.3067250739979412,
      "grad_norm": 0.02322913520038128,
      "learning_rate": 3.2348522058790287e-06,
      "loss": 0.0016,
      "step": 203740
    },
    {
      "epoch": 1.3067892108917274,
      "grad_norm": 0.14985918998718262,
      "learning_rate": 3.234328551651137e-06,
      "loss": 0.0008,
      "step": 203750
    },
    {
      "epoch": 1.3068533477855133,
      "grad_norm": 0.07556784898042679,
      "learning_rate": 3.2338049195483696e-06,
      "loss": 0.0011,
      "step": 203760
    },
    {
      "epoch": 1.3069174846792995,
      "grad_norm": 0.06410104036331177,
      "learning_rate": 3.2332813095772897e-06,
      "loss": 0.0015,
      "step": 203770
    },
    {
      "epoch": 1.3069816215730856,
      "grad_norm": 0.10119643062353134,
      "learning_rate": 3.232757721744458e-06,
      "loss": 0.003,
      "step": 203780
    },
    {
      "epoch": 1.3070457584668718,
      "grad_norm": 0.2246820330619812,
      "learning_rate": 3.232234156056435e-06,
      "loss": 0.0012,
      "step": 203790
    },
    {
      "epoch": 1.3071098953606577,
      "grad_norm": 0.09073050320148468,
      "learning_rate": 3.2317106125197816e-06,
      "loss": 0.0012,
      "step": 203800
    },
    {
      "epoch": 1.3071740322544438,
      "grad_norm": 0.04273083433508873,
      "learning_rate": 3.231187091141057e-06,
      "loss": 0.0011,
      "step": 203810
    },
    {
      "epoch": 1.30723816914823,
      "grad_norm": 0.1075194776058197,
      "learning_rate": 3.2306635919268235e-06,
      "loss": 0.0027,
      "step": 203820
    },
    {
      "epoch": 1.3073023060420161,
      "grad_norm": 0.03114445134997368,
      "learning_rate": 3.23014011488364e-06,
      "loss": 0.001,
      "step": 203830
    },
    {
      "epoch": 1.3073664429358023,
      "grad_norm": 0.04628058150410652,
      "learning_rate": 3.229616660018065e-06,
      "loss": 0.0013,
      "step": 203840
    },
    {
      "epoch": 1.3074305798295882,
      "grad_norm": 0.05554237216711044,
      "learning_rate": 3.229093227336658e-06,
      "loss": 0.0008,
      "step": 203850
    },
    {
      "epoch": 1.3074947167233744,
      "grad_norm": 0.04712125286459923,
      "learning_rate": 3.22856981684598e-06,
      "loss": 0.0028,
      "step": 203860
    },
    {
      "epoch": 1.3075588536171605,
      "grad_norm": 0.05716227367520332,
      "learning_rate": 3.228046428552587e-06,
      "loss": 0.0025,
      "step": 203870
    },
    {
      "epoch": 1.3076229905109464,
      "grad_norm": 0.025026440620422363,
      "learning_rate": 3.2275230624630395e-06,
      "loss": 0.0012,
      "step": 203880
    },
    {
      "epoch": 1.3076871274047326,
      "grad_norm": 0.17739082872867584,
      "learning_rate": 3.2269997185838935e-06,
      "loss": 0.0012,
      "step": 203890
    },
    {
      "epoch": 1.3077512642985187,
      "grad_norm": 0.04024747386574745,
      "learning_rate": 3.2264763969217094e-06,
      "loss": 0.0012,
      "step": 203900
    },
    {
      "epoch": 1.3078154011923049,
      "grad_norm": 0.011896566487848759,
      "learning_rate": 3.225953097483043e-06,
      "loss": 0.0008,
      "step": 203910
    },
    {
      "epoch": 1.307879538086091,
      "grad_norm": 0.06987634301185608,
      "learning_rate": 3.2254298202744516e-06,
      "loss": 0.0015,
      "step": 203920
    },
    {
      "epoch": 1.3079436749798772,
      "grad_norm": 0.0587100051343441,
      "learning_rate": 3.2249065653024948e-06,
      "loss": 0.0013,
      "step": 203930
    },
    {
      "epoch": 1.308007811873663,
      "grad_norm": 0.03610633313655853,
      "learning_rate": 3.224383332573725e-06,
      "loss": 0.0013,
      "step": 203940
    },
    {
      "epoch": 1.3080719487674493,
      "grad_norm": 0.09071983397006989,
      "learning_rate": 3.223860122094703e-06,
      "loss": 0.0016,
      "step": 203950
    },
    {
      "epoch": 1.3081360856612354,
      "grad_norm": 0.21553845703601837,
      "learning_rate": 3.2233369338719816e-06,
      "loss": 0.0016,
      "step": 203960
    },
    {
      "epoch": 1.3082002225550213,
      "grad_norm": 0.06973786652088165,
      "learning_rate": 3.2228137679121198e-06,
      "loss": 0.0016,
      "step": 203970
    },
    {
      "epoch": 1.3082643594488075,
      "grad_norm": 0.10736697912216187,
      "learning_rate": 3.22229062422167e-06,
      "loss": 0.0011,
      "step": 203980
    },
    {
      "epoch": 1.3083284963425936,
      "grad_norm": 0.04209981858730316,
      "learning_rate": 3.2217675028071905e-06,
      "loss": 0.0006,
      "step": 203990
    },
    {
      "epoch": 1.3083926332363798,
      "grad_norm": 0.14557282626628876,
      "learning_rate": 3.2212444036752335e-06,
      "loss": 0.0014,
      "step": 204000
    },
    {
      "epoch": 1.308456770130166,
      "grad_norm": 0.11464909464120865,
      "learning_rate": 3.220721326832357e-06,
      "loss": 0.0013,
      "step": 204010
    },
    {
      "epoch": 1.3085209070239519,
      "grad_norm": 0.0317268893122673,
      "learning_rate": 3.2201982722851127e-06,
      "loss": 0.0011,
      "step": 204020
    },
    {
      "epoch": 1.308585043917738,
      "grad_norm": 0.13303732872009277,
      "learning_rate": 3.2196752400400573e-06,
      "loss": 0.0024,
      "step": 204030
    },
    {
      "epoch": 1.3086491808115241,
      "grad_norm": 0.03285135701298714,
      "learning_rate": 3.219152230103742e-06,
      "loss": 0.0023,
      "step": 204040
    },
    {
      "epoch": 1.3087133177053103,
      "grad_norm": 0.12097574770450592,
      "learning_rate": 3.2186292424827236e-06,
      "loss": 0.0025,
      "step": 204050
    },
    {
      "epoch": 1.3087774545990962,
      "grad_norm": 0.07854858040809631,
      "learning_rate": 3.2181062771835524e-06,
      "loss": 0.0019,
      "step": 204060
    },
    {
      "epoch": 1.3088415914928824,
      "grad_norm": 0.0640355572104454,
      "learning_rate": 3.2175833342127834e-06,
      "loss": 0.0015,
      "step": 204070
    },
    {
      "epoch": 1.3089057283866685,
      "grad_norm": 0.10648632049560547,
      "learning_rate": 3.21706041357697e-06,
      "loss": 0.0021,
      "step": 204080
    },
    {
      "epoch": 1.3089698652804547,
      "grad_norm": 0.06015864014625549,
      "learning_rate": 3.216537515282663e-06,
      "loss": 0.0017,
      "step": 204090
    },
    {
      "epoch": 1.3090340021742408,
      "grad_norm": 0.13343878090381622,
      "learning_rate": 3.216014639336416e-06,
      "loss": 0.0017,
      "step": 204100
    },
    {
      "epoch": 1.3090981390680267,
      "grad_norm": 0.13053520023822784,
      "learning_rate": 3.2154917857447797e-06,
      "loss": 0.0028,
      "step": 204110
    },
    {
      "epoch": 1.309162275961813,
      "grad_norm": 0.03047873079776764,
      "learning_rate": 3.214968954514308e-06,
      "loss": 0.002,
      "step": 204120
    },
    {
      "epoch": 1.309226412855599,
      "grad_norm": 0.09146006405353546,
      "learning_rate": 3.21444614565155e-06,
      "loss": 0.0018,
      "step": 204130
    },
    {
      "epoch": 1.309290549749385,
      "grad_norm": 0.024952685460448265,
      "learning_rate": 3.2139233591630593e-06,
      "loss": 0.0008,
      "step": 204140
    },
    {
      "epoch": 1.3093546866431711,
      "grad_norm": 0.12278847396373749,
      "learning_rate": 3.213400595055384e-06,
      "loss": 0.0014,
      "step": 204150
    },
    {
      "epoch": 1.3094188235369573,
      "grad_norm": 0.03779635950922966,
      "learning_rate": 3.2128778533350767e-06,
      "loss": 0.0009,
      "step": 204160
    },
    {
      "epoch": 1.3094829604307434,
      "grad_norm": 0.012328768149018288,
      "learning_rate": 3.2123551340086868e-06,
      "loss": 0.001,
      "step": 204170
    },
    {
      "epoch": 1.3095470973245296,
      "grad_norm": 0.070521779358387,
      "learning_rate": 3.2118324370827654e-06,
      "loss": 0.0013,
      "step": 204180
    },
    {
      "epoch": 1.3096112342183155,
      "grad_norm": 0.061533670872449875,
      "learning_rate": 3.2113097625638606e-06,
      "loss": 0.001,
      "step": 204190
    },
    {
      "epoch": 1.3096753711121016,
      "grad_norm": 0.33698606491088867,
      "learning_rate": 3.2107871104585238e-06,
      "loss": 0.0034,
      "step": 204200
    },
    {
      "epoch": 1.3097395080058878,
      "grad_norm": 0.06746309995651245,
      "learning_rate": 3.210264480773302e-06,
      "loss": 0.0007,
      "step": 204210
    },
    {
      "epoch": 1.309803644899674,
      "grad_norm": 0.006428118795156479,
      "learning_rate": 3.209741873514746e-06,
      "loss": 0.0007,
      "step": 204220
    },
    {
      "epoch": 1.3098677817934599,
      "grad_norm": 0.1033342257142067,
      "learning_rate": 3.209219288689404e-06,
      "loss": 0.002,
      "step": 204230
    },
    {
      "epoch": 1.309931918687246,
      "grad_norm": 0.14611609280109406,
      "learning_rate": 3.2086967263038236e-06,
      "loss": 0.0017,
      "step": 204240
    },
    {
      "epoch": 1.3099960555810322,
      "grad_norm": 0.12221353501081467,
      "learning_rate": 3.2081741863645543e-06,
      "loss": 0.0011,
      "step": 204250
    },
    {
      "epoch": 1.3100601924748183,
      "grad_norm": 0.02614627592265606,
      "learning_rate": 3.207651668878142e-06,
      "loss": 0.001,
      "step": 204260
    },
    {
      "epoch": 1.3101243293686045,
      "grad_norm": 0.4079796075820923,
      "learning_rate": 3.207129173851137e-06,
      "loss": 0.0018,
      "step": 204270
    },
    {
      "epoch": 1.3101884662623904,
      "grad_norm": 0.19869622588157654,
      "learning_rate": 3.206606701290083e-06,
      "loss": 0.0018,
      "step": 204280
    },
    {
      "epoch": 1.3102526031561765,
      "grad_norm": 0.05197782814502716,
      "learning_rate": 3.20608425120153e-06,
      "loss": 0.0018,
      "step": 204290
    },
    {
      "epoch": 1.3103167400499627,
      "grad_norm": 0.017865058034658432,
      "learning_rate": 3.2055618235920226e-06,
      "loss": 0.001,
      "step": 204300
    },
    {
      "epoch": 1.3103808769437486,
      "grad_norm": 0.04671267047524452,
      "learning_rate": 3.205039418468109e-06,
      "loss": 0.0013,
      "step": 204310
    },
    {
      "epoch": 1.3104450138375348,
      "grad_norm": 0.016545766964554787,
      "learning_rate": 3.2045170358363333e-06,
      "loss": 0.0011,
      "step": 204320
    },
    {
      "epoch": 1.310509150731321,
      "grad_norm": 0.12884260714054108,
      "learning_rate": 3.2039946757032423e-06,
      "loss": 0.0023,
      "step": 204330
    },
    {
      "epoch": 1.310573287625107,
      "grad_norm": 0.05837222933769226,
      "learning_rate": 3.203472338075382e-06,
      "loss": 0.002,
      "step": 204340
    },
    {
      "epoch": 1.3106374245188932,
      "grad_norm": 0.06461817026138306,
      "learning_rate": 3.202950022959297e-06,
      "loss": 0.0012,
      "step": 204350
    },
    {
      "epoch": 1.3107015614126794,
      "grad_norm": 0.14301276206970215,
      "learning_rate": 3.2024277303615323e-06,
      "loss": 0.001,
      "step": 204360
    },
    {
      "epoch": 1.3107656983064653,
      "grad_norm": 0.024913959205150604,
      "learning_rate": 3.2019054602886334e-06,
      "loss": 0.001,
      "step": 204370
    },
    {
      "epoch": 1.3108298352002514,
      "grad_norm": 0.06206691637635231,
      "learning_rate": 3.2013832127471445e-06,
      "loss": 0.0015,
      "step": 204380
    },
    {
      "epoch": 1.3108939720940376,
      "grad_norm": 0.008530830033123493,
      "learning_rate": 3.2008609877436087e-06,
      "loss": 0.0008,
      "step": 204390
    },
    {
      "epoch": 1.3109581089878235,
      "grad_norm": 0.10439231246709824,
      "learning_rate": 3.2003387852845714e-06,
      "loss": 0.0011,
      "step": 204400
    },
    {
      "epoch": 1.3110222458816096,
      "grad_norm": 0.059677399694919586,
      "learning_rate": 3.1998166053765746e-06,
      "loss": 0.0017,
      "step": 204410
    },
    {
      "epoch": 1.3110863827753958,
      "grad_norm": 0.121797576546669,
      "learning_rate": 3.199294448026162e-06,
      "loss": 0.0012,
      "step": 204420
    },
    {
      "epoch": 1.311150519669182,
      "grad_norm": 0.04569542035460472,
      "learning_rate": 3.198772313239877e-06,
      "loss": 0.0021,
      "step": 204430
    },
    {
      "epoch": 1.311214656562968,
      "grad_norm": 0.03476222604513168,
      "learning_rate": 3.1982502010242633e-06,
      "loss": 0.0008,
      "step": 204440
    },
    {
      "epoch": 1.311278793456754,
      "grad_norm": 0.08952942490577698,
      "learning_rate": 3.1977281113858615e-06,
      "loss": 0.001,
      "step": 204450
    },
    {
      "epoch": 1.3113429303505402,
      "grad_norm": 0.10412070155143738,
      "learning_rate": 3.197206044331215e-06,
      "loss": 0.0035,
      "step": 204460
    },
    {
      "epoch": 1.3114070672443263,
      "grad_norm": 0.07542761415243149,
      "learning_rate": 3.196683999866864e-06,
      "loss": 0.0013,
      "step": 204470
    },
    {
      "epoch": 1.3114712041381125,
      "grad_norm": 0.056854650378227234,
      "learning_rate": 3.1961619779993524e-06,
      "loss": 0.0014,
      "step": 204480
    },
    {
      "epoch": 1.3115353410318984,
      "grad_norm": 0.04510832577943802,
      "learning_rate": 3.19563997873522e-06,
      "loss": 0.0006,
      "step": 204490
    },
    {
      "epoch": 1.3115994779256845,
      "grad_norm": 0.1429297924041748,
      "learning_rate": 3.195118002081008e-06,
      "loss": 0.002,
      "step": 204500
    },
    {
      "epoch": 1.3116636148194707,
      "grad_norm": 0.05365953966975212,
      "learning_rate": 3.194596048043258e-06,
      "loss": 0.0011,
      "step": 204510
    },
    {
      "epoch": 1.3117277517132568,
      "grad_norm": 0.017902769148349762,
      "learning_rate": 3.1940741166285088e-06,
      "loss": 0.0017,
      "step": 204520
    },
    {
      "epoch": 1.311791888607043,
      "grad_norm": 0.09438332170248032,
      "learning_rate": 3.193552207843301e-06,
      "loss": 0.001,
      "step": 204530
    },
    {
      "epoch": 1.311856025500829,
      "grad_norm": 0.054045744240283966,
      "learning_rate": 3.1930303216941773e-06,
      "loss": 0.001,
      "step": 204540
    },
    {
      "epoch": 1.311920162394615,
      "grad_norm": 0.09531563520431519,
      "learning_rate": 3.192508458187673e-06,
      "loss": 0.0038,
      "step": 204550
    },
    {
      "epoch": 1.3119842992884012,
      "grad_norm": 0.04847593978047371,
      "learning_rate": 3.1919866173303316e-06,
      "loss": 0.0005,
      "step": 204560
    },
    {
      "epoch": 1.3120484361821871,
      "grad_norm": 0.09458363056182861,
      "learning_rate": 3.1914647991286886e-06,
      "loss": 0.0013,
      "step": 204570
    },
    {
      "epoch": 1.3121125730759733,
      "grad_norm": 0.24712799489498138,
      "learning_rate": 3.190943003589285e-06,
      "loss": 0.0017,
      "step": 204580
    },
    {
      "epoch": 1.3121767099697594,
      "grad_norm": 0.05329671502113342,
      "learning_rate": 3.1904212307186576e-06,
      "loss": 0.0027,
      "step": 204590
    },
    {
      "epoch": 1.3122408468635456,
      "grad_norm": 0.0810551792383194,
      "learning_rate": 3.189899480523347e-06,
      "loss": 0.0019,
      "step": 204600
    },
    {
      "epoch": 1.3123049837573317,
      "grad_norm": 0.11516894400119781,
      "learning_rate": 3.189377753009888e-06,
      "loss": 0.0019,
      "step": 204610
    },
    {
      "epoch": 1.3123691206511179,
      "grad_norm": 0.011796182952821255,
      "learning_rate": 3.1888560481848195e-06,
      "loss": 0.0014,
      "step": 204620
    },
    {
      "epoch": 1.3124332575449038,
      "grad_norm": 0.07899974286556244,
      "learning_rate": 3.188334366054681e-06,
      "loss": 0.0011,
      "step": 204630
    },
    {
      "epoch": 1.31249739443869,
      "grad_norm": 0.08089365065097809,
      "learning_rate": 3.187812706626007e-06,
      "loss": 0.0022,
      "step": 204640
    },
    {
      "epoch": 1.312561531332476,
      "grad_norm": 0.12527890503406525,
      "learning_rate": 3.1872910699053343e-06,
      "loss": 0.0013,
      "step": 204650
    },
    {
      "epoch": 1.312625668226262,
      "grad_norm": 0.0680837482213974,
      "learning_rate": 3.1867694558992e-06,
      "loss": 0.0018,
      "step": 204660
    },
    {
      "epoch": 1.3126898051200482,
      "grad_norm": 0.048125989735126495,
      "learning_rate": 3.1862478646141413e-06,
      "loss": 0.0021,
      "step": 204670
    },
    {
      "epoch": 1.3127539420138343,
      "grad_norm": 0.042054932564496994,
      "learning_rate": 3.185726296056692e-06,
      "loss": 0.0007,
      "step": 204680
    },
    {
      "epoch": 1.3128180789076205,
      "grad_norm": 0.11337859183549881,
      "learning_rate": 3.18520475023339e-06,
      "loss": 0.001,
      "step": 204690
    },
    {
      "epoch": 1.3128822158014066,
      "grad_norm": 0.10501635074615479,
      "learning_rate": 3.184683227150768e-06,
      "loss": 0.0009,
      "step": 204700
    },
    {
      "epoch": 1.3129463526951926,
      "grad_norm": 0.044662632048130035,
      "learning_rate": 3.1841617268153647e-06,
      "loss": 0.0026,
      "step": 204710
    },
    {
      "epoch": 1.3130104895889787,
      "grad_norm": 0.12997126579284668,
      "learning_rate": 3.1836402492337105e-06,
      "loss": 0.0016,
      "step": 204720
    },
    {
      "epoch": 1.3130746264827649,
      "grad_norm": 0.08274967968463898,
      "learning_rate": 3.1831187944123435e-06,
      "loss": 0.0012,
      "step": 204730
    },
    {
      "epoch": 1.3131387633765508,
      "grad_norm": 0.02656915783882141,
      "learning_rate": 3.1825973623577954e-06,
      "loss": 0.0019,
      "step": 204740
    },
    {
      "epoch": 1.313202900270337,
      "grad_norm": 0.024764280766248703,
      "learning_rate": 3.1820759530766026e-06,
      "loss": 0.0021,
      "step": 204750
    },
    {
      "epoch": 1.313267037164123,
      "grad_norm": 0.057234641164541245,
      "learning_rate": 3.1815545665752966e-06,
      "loss": 0.0011,
      "step": 204760
    },
    {
      "epoch": 1.3133311740579092,
      "grad_norm": 0.10282032936811447,
      "learning_rate": 3.1810332028604106e-06,
      "loss": 0.0011,
      "step": 204770
    },
    {
      "epoch": 1.3133953109516954,
      "grad_norm": 0.06660402566194534,
      "learning_rate": 3.18051186193848e-06,
      "loss": 0.0008,
      "step": 204780
    },
    {
      "epoch": 1.3134594478454815,
      "grad_norm": 0.09857561439275742,
      "learning_rate": 3.179990543816035e-06,
      "loss": 0.0013,
      "step": 204790
    },
    {
      "epoch": 1.3135235847392674,
      "grad_norm": 0.019108690321445465,
      "learning_rate": 3.179469248499611e-06,
      "loss": 0.0009,
      "step": 204800
    },
    {
      "epoch": 1.3135877216330536,
      "grad_norm": 0.008525248616933823,
      "learning_rate": 3.1789479759957366e-06,
      "loss": 0.0018,
      "step": 204810
    },
    {
      "epoch": 1.3136518585268397,
      "grad_norm": 0.06714153289794922,
      "learning_rate": 3.1784267263109467e-06,
      "loss": 0.0016,
      "step": 204820
    },
    {
      "epoch": 1.3137159954206257,
      "grad_norm": 0.0973796471953392,
      "learning_rate": 3.177905499451771e-06,
      "loss": 0.001,
      "step": 204830
    },
    {
      "epoch": 1.3137801323144118,
      "grad_norm": 0.039386920630931854,
      "learning_rate": 3.1773842954247423e-06,
      "loss": 0.0006,
      "step": 204840
    },
    {
      "epoch": 1.313844269208198,
      "grad_norm": 0.09478655457496643,
      "learning_rate": 3.1768631142363902e-06,
      "loss": 0.0012,
      "step": 204850
    },
    {
      "epoch": 1.3139084061019841,
      "grad_norm": 0.10358711332082748,
      "learning_rate": 3.1763419558932473e-06,
      "loss": 0.0014,
      "step": 204860
    },
    {
      "epoch": 1.3139725429957703,
      "grad_norm": 0.12413185834884644,
      "learning_rate": 3.175820820401842e-06,
      "loss": 0.0011,
      "step": 204870
    },
    {
      "epoch": 1.3140366798895562,
      "grad_norm": 0.03639234974980354,
      "learning_rate": 3.1752997077687063e-06,
      "loss": 0.0013,
      "step": 204880
    },
    {
      "epoch": 1.3141008167833423,
      "grad_norm": 0.25842973589897156,
      "learning_rate": 3.1747786180003687e-06,
      "loss": 0.0013,
      "step": 204890
    },
    {
      "epoch": 1.3141649536771285,
      "grad_norm": 0.11785202473402023,
      "learning_rate": 3.17425755110336e-06,
      "loss": 0.0007,
      "step": 204900
    },
    {
      "epoch": 1.3142290905709146,
      "grad_norm": 0.209381565451622,
      "learning_rate": 3.1737365070842084e-06,
      "loss": 0.0027,
      "step": 204910
    },
    {
      "epoch": 1.3142932274647006,
      "grad_norm": 0.10094190388917923,
      "learning_rate": 3.1732154859494436e-06,
      "loss": 0.0029,
      "step": 204920
    },
    {
      "epoch": 1.3143573643584867,
      "grad_norm": 0.12673942744731903,
      "learning_rate": 3.1726944877055955e-06,
      "loss": 0.0009,
      "step": 204930
    },
    {
      "epoch": 1.3144215012522729,
      "grad_norm": 0.146784707903862,
      "learning_rate": 3.17217351235919e-06,
      "loss": 0.001,
      "step": 204940
    },
    {
      "epoch": 1.314485638146059,
      "grad_norm": 0.025839293375611305,
      "learning_rate": 3.171652559916758e-06,
      "loss": 0.0015,
      "step": 204950
    },
    {
      "epoch": 1.3145497750398452,
      "grad_norm": 0.16551709175109863,
      "learning_rate": 3.171131630384825e-06,
      "loss": 0.0016,
      "step": 204960
    },
    {
      "epoch": 1.314613911933631,
      "grad_norm": 0.10145284980535507,
      "learning_rate": 3.1706107237699214e-06,
      "loss": 0.0015,
      "step": 204970
    },
    {
      "epoch": 1.3146780488274172,
      "grad_norm": 0.11628744751214981,
      "learning_rate": 3.1700898400785707e-06,
      "loss": 0.0023,
      "step": 204980
    },
    {
      "epoch": 1.3147421857212034,
      "grad_norm": 0.13586631417274475,
      "learning_rate": 3.169568979317304e-06,
      "loss": 0.0011,
      "step": 204990
    },
    {
      "epoch": 1.3148063226149893,
      "grad_norm": 0.09128490835428238,
      "learning_rate": 3.169048141492644e-06,
      "loss": 0.0018,
      "step": 205000
    },
    {
      "epoch": 1.3148704595087755,
      "grad_norm": 0.15765392780303955,
      "learning_rate": 3.168527326611122e-06,
      "loss": 0.0024,
      "step": 205010
    },
    {
      "epoch": 1.3149345964025616,
      "grad_norm": 0.1263551115989685,
      "learning_rate": 3.16800653467926e-06,
      "loss": 0.0012,
      "step": 205020
    },
    {
      "epoch": 1.3149987332963478,
      "grad_norm": 0.02617996372282505,
      "learning_rate": 3.1674857657035863e-06,
      "loss": 0.0007,
      "step": 205030
    },
    {
      "epoch": 1.315062870190134,
      "grad_norm": 0.04973071441054344,
      "learning_rate": 3.1669650196906243e-06,
      "loss": 0.0013,
      "step": 205040
    },
    {
      "epoch": 1.31512700708392,
      "grad_norm": 0.0865158811211586,
      "learning_rate": 3.1664442966469016e-06,
      "loss": 0.001,
      "step": 205050
    },
    {
      "epoch": 1.315191143977706,
      "grad_norm": 0.058341436088085175,
      "learning_rate": 3.1659235965789416e-06,
      "loss": 0.0015,
      "step": 205060
    },
    {
      "epoch": 1.3152552808714921,
      "grad_norm": 0.04228498786687851,
      "learning_rate": 3.16540291949327e-06,
      "loss": 0.0024,
      "step": 205070
    },
    {
      "epoch": 1.3153194177652783,
      "grad_norm": 0.10057054460048676,
      "learning_rate": 3.1648822653964117e-06,
      "loss": 0.0013,
      "step": 205080
    },
    {
      "epoch": 1.3153835546590642,
      "grad_norm": 0.1200881227850914,
      "learning_rate": 3.164361634294889e-06,
      "loss": 0.0011,
      "step": 205090
    },
    {
      "epoch": 1.3154476915528504,
      "grad_norm": 0.08531105518341064,
      "learning_rate": 3.163841026195228e-06,
      "loss": 0.0009,
      "step": 205100
    },
    {
      "epoch": 1.3155118284466365,
      "grad_norm": 0.21463236212730408,
      "learning_rate": 3.1633204411039498e-06,
      "loss": 0.0017,
      "step": 205110
    },
    {
      "epoch": 1.3155759653404226,
      "grad_norm": 0.019257543608546257,
      "learning_rate": 3.162799879027581e-06,
      "loss": 0.0025,
      "step": 205120
    },
    {
      "epoch": 1.3156401022342088,
      "grad_norm": 0.17167098820209503,
      "learning_rate": 3.1622793399726415e-06,
      "loss": 0.002,
      "step": 205130
    },
    {
      "epoch": 1.3157042391279947,
      "grad_norm": 0.03428174555301666,
      "learning_rate": 3.1617588239456557e-06,
      "loss": 0.001,
      "step": 205140
    },
    {
      "epoch": 1.3157683760217809,
      "grad_norm": 0.06448190659284592,
      "learning_rate": 3.1612383309531452e-06,
      "loss": 0.0013,
      "step": 205150
    },
    {
      "epoch": 1.315832512915567,
      "grad_norm": 0.03777821734547615,
      "learning_rate": 3.160717861001633e-06,
      "loss": 0.0013,
      "step": 205160
    },
    {
      "epoch": 1.315896649809353,
      "grad_norm": 0.418274462223053,
      "learning_rate": 3.1601974140976404e-06,
      "loss": 0.0015,
      "step": 205170
    },
    {
      "epoch": 1.315960786703139,
      "grad_norm": 0.07767599821090698,
      "learning_rate": 3.15967699024769e-06,
      "loss": 0.0005,
      "step": 205180
    },
    {
      "epoch": 1.3160249235969252,
      "grad_norm": 0.06448792666196823,
      "learning_rate": 3.1591565894583005e-06,
      "loss": 0.0006,
      "step": 205190
    },
    {
      "epoch": 1.3160890604907114,
      "grad_norm": 0.34093043208122253,
      "learning_rate": 3.158636211735995e-06,
      "loss": 0.0017,
      "step": 205200
    },
    {
      "epoch": 1.3161531973844975,
      "grad_norm": 0.11581496149301529,
      "learning_rate": 3.1581158570872936e-06,
      "loss": 0.0025,
      "step": 205210
    },
    {
      "epoch": 1.3162173342782837,
      "grad_norm": 0.13731037080287933,
      "learning_rate": 3.1575955255187173e-06,
      "loss": 0.0018,
      "step": 205220
    },
    {
      "epoch": 1.3162814711720696,
      "grad_norm": 0.06086349859833717,
      "learning_rate": 3.157075217036787e-06,
      "loss": 0.0023,
      "step": 205230
    },
    {
      "epoch": 1.3163456080658558,
      "grad_norm": 0.0504254549741745,
      "learning_rate": 3.1565549316480192e-06,
      "loss": 0.0014,
      "step": 205240
    },
    {
      "epoch": 1.316409744959642,
      "grad_norm": 0.0985199511051178,
      "learning_rate": 3.1560346693589372e-06,
      "loss": 0.0017,
      "step": 205250
    },
    {
      "epoch": 1.3164738818534278,
      "grad_norm": 0.03697388619184494,
      "learning_rate": 3.1555144301760576e-06,
      "loss": 0.0014,
      "step": 205260
    },
    {
      "epoch": 1.316538018747214,
      "grad_norm": 0.16653156280517578,
      "learning_rate": 3.1549942141059014e-06,
      "loss": 0.0029,
      "step": 205270
    },
    {
      "epoch": 1.3166021556410001,
      "grad_norm": 0.09604401141405106,
      "learning_rate": 3.154474021154985e-06,
      "loss": 0.0006,
      "step": 205280
    },
    {
      "epoch": 1.3166662925347863,
      "grad_norm": 0.10724960267543793,
      "learning_rate": 3.153953851329827e-06,
      "loss": 0.0008,
      "step": 205290
    },
    {
      "epoch": 1.3167304294285724,
      "grad_norm": 0.022921591997146606,
      "learning_rate": 3.15343370463695e-06,
      "loss": 0.0006,
      "step": 205300
    },
    {
      "epoch": 1.3167945663223584,
      "grad_norm": 0.023986762389540672,
      "learning_rate": 3.152913581082866e-06,
      "loss": 0.0012,
      "step": 205310
    },
    {
      "epoch": 1.3168587032161445,
      "grad_norm": 0.3583962321281433,
      "learning_rate": 3.1523934806740965e-06,
      "loss": 0.001,
      "step": 205320
    },
    {
      "epoch": 1.3169228401099307,
      "grad_norm": 0.0671306699514389,
      "learning_rate": 3.151873403417156e-06,
      "loss": 0.0008,
      "step": 205330
    },
    {
      "epoch": 1.3169869770037168,
      "grad_norm": 0.1393251121044159,
      "learning_rate": 3.151353349318562e-06,
      "loss": 0.0016,
      "step": 205340
    },
    {
      "epoch": 1.3170511138975027,
      "grad_norm": 0.09741727262735367,
      "learning_rate": 3.1508333183848337e-06,
      "loss": 0.0018,
      "step": 205350
    },
    {
      "epoch": 1.3171152507912889,
      "grad_norm": 0.022273661568760872,
      "learning_rate": 3.1503133106224844e-06,
      "loss": 0.0032,
      "step": 205360
    },
    {
      "epoch": 1.317179387685075,
      "grad_norm": 0.05251382291316986,
      "learning_rate": 3.149793326038032e-06,
      "loss": 0.002,
      "step": 205370
    },
    {
      "epoch": 1.3172435245788612,
      "grad_norm": 0.13902156054973602,
      "learning_rate": 3.1492733646379903e-06,
      "loss": 0.001,
      "step": 205380
    },
    {
      "epoch": 1.3173076614726473,
      "grad_norm": 0.07224063575267792,
      "learning_rate": 3.148753426428877e-06,
      "loss": 0.0024,
      "step": 205390
    },
    {
      "epoch": 1.3173717983664333,
      "grad_norm": 0.06636285781860352,
      "learning_rate": 3.148233511417206e-06,
      "loss": 0.0008,
      "step": 205400
    },
    {
      "epoch": 1.3174359352602194,
      "grad_norm": 0.05493977665901184,
      "learning_rate": 3.1477136196094926e-06,
      "loss": 0.001,
      "step": 205410
    },
    {
      "epoch": 1.3175000721540056,
      "grad_norm": 0.026711495593190193,
      "learning_rate": 3.1471937510122515e-06,
      "loss": 0.0013,
      "step": 205420
    },
    {
      "epoch": 1.3175642090477915,
      "grad_norm": 0.18658442795276642,
      "learning_rate": 3.146673905631997e-06,
      "loss": 0.0014,
      "step": 205430
    },
    {
      "epoch": 1.3176283459415776,
      "grad_norm": 0.09886343777179718,
      "learning_rate": 3.1461540834752423e-06,
      "loss": 0.0009,
      "step": 205440
    },
    {
      "epoch": 1.3176924828353638,
      "grad_norm": 0.06812240928411484,
      "learning_rate": 3.1456342845485032e-06,
      "loss": 0.0015,
      "step": 205450
    },
    {
      "epoch": 1.31775661972915,
      "grad_norm": 0.05942175164818764,
      "learning_rate": 3.1451145088582903e-06,
      "loss": 0.0016,
      "step": 205460
    },
    {
      "epoch": 1.317820756622936,
      "grad_norm": 0.13044652342796326,
      "learning_rate": 3.144594756411118e-06,
      "loss": 0.0012,
      "step": 205470
    },
    {
      "epoch": 1.3178848935167222,
      "grad_norm": 0.011788193136453629,
      "learning_rate": 3.1440750272135013e-06,
      "loss": 0.0015,
      "step": 205480
    },
    {
      "epoch": 1.3179490304105081,
      "grad_norm": 0.07968220859766006,
      "learning_rate": 3.1435553212719495e-06,
      "loss": 0.0022,
      "step": 205490
    },
    {
      "epoch": 1.3180131673042943,
      "grad_norm": 0.2021966427564621,
      "learning_rate": 3.1430356385929774e-06,
      "loss": 0.0011,
      "step": 205500
    },
    {
      "epoch": 1.3180773041980804,
      "grad_norm": 0.07877519726753235,
      "learning_rate": 3.1425159791830947e-06,
      "loss": 0.0018,
      "step": 205510
    },
    {
      "epoch": 1.3181414410918664,
      "grad_norm": 0.10816589742898941,
      "learning_rate": 3.1419963430488155e-06,
      "loss": 0.0017,
      "step": 205520
    },
    {
      "epoch": 1.3182055779856525,
      "grad_norm": 0.24587659537792206,
      "learning_rate": 3.141476730196649e-06,
      "loss": 0.0011,
      "step": 205530
    },
    {
      "epoch": 1.3182697148794387,
      "grad_norm": 0.058466412127017975,
      "learning_rate": 3.1409571406331076e-06,
      "loss": 0.001,
      "step": 205540
    },
    {
      "epoch": 1.3183338517732248,
      "grad_norm": 0.04519858956336975,
      "learning_rate": 3.140437574364702e-06,
      "loss": 0.002,
      "step": 205550
    },
    {
      "epoch": 1.318397988667011,
      "grad_norm": 0.12952274084091187,
      "learning_rate": 3.1399180313979425e-06,
      "loss": 0.001,
      "step": 205560
    },
    {
      "epoch": 1.318462125560797,
      "grad_norm": 0.006296441424638033,
      "learning_rate": 3.1393985117393392e-06,
      "loss": 0.0011,
      "step": 205570
    },
    {
      "epoch": 1.318526262454583,
      "grad_norm": 0.06386483460664749,
      "learning_rate": 3.1388790153954034e-06,
      "loss": 0.0025,
      "step": 205580
    },
    {
      "epoch": 1.3185903993483692,
      "grad_norm": 0.024813035503029823,
      "learning_rate": 3.138359542372642e-06,
      "loss": 0.0012,
      "step": 205590
    },
    {
      "epoch": 1.3186545362421551,
      "grad_norm": 0.1011386513710022,
      "learning_rate": 3.1378400926775675e-06,
      "loss": 0.0016,
      "step": 205600
    },
    {
      "epoch": 1.3187186731359413,
      "grad_norm": 0.08756448328495026,
      "learning_rate": 3.137320666316687e-06,
      "loss": 0.0013,
      "step": 205610
    },
    {
      "epoch": 1.3187828100297274,
      "grad_norm": 0.06075456738471985,
      "learning_rate": 3.1368012632965088e-06,
      "loss": 0.001,
      "step": 205620
    },
    {
      "epoch": 1.3188469469235136,
      "grad_norm": 0.07185124605894089,
      "learning_rate": 3.136281883623544e-06,
      "loss": 0.0009,
      "step": 205630
    },
    {
      "epoch": 1.3189110838172997,
      "grad_norm": 0.05814353749155998,
      "learning_rate": 3.135762527304298e-06,
      "loss": 0.0013,
      "step": 205640
    },
    {
      "epoch": 1.3189752207110859,
      "grad_norm": 0.17648516595363617,
      "learning_rate": 3.135243194345281e-06,
      "loss": 0.0014,
      "step": 205650
    },
    {
      "epoch": 1.3190393576048718,
      "grad_norm": 0.016797835007309914,
      "learning_rate": 3.134723884752998e-06,
      "loss": 0.0032,
      "step": 205660
    },
    {
      "epoch": 1.319103494498658,
      "grad_norm": 0.08698790520429611,
      "learning_rate": 3.1342045985339598e-06,
      "loss": 0.001,
      "step": 205670
    },
    {
      "epoch": 1.319167631392444,
      "grad_norm": 0.08872439712285995,
      "learning_rate": 3.1336853356946695e-06,
      "loss": 0.0008,
      "step": 205680
    },
    {
      "epoch": 1.31923176828623,
      "grad_norm": 0.2343255579471588,
      "learning_rate": 3.133166096241637e-06,
      "loss": 0.0015,
      "step": 205690
    },
    {
      "epoch": 1.3192959051800162,
      "grad_norm": 0.03147163614630699,
      "learning_rate": 3.132646880181367e-06,
      "loss": 0.0033,
      "step": 205700
    },
    {
      "epoch": 1.3193600420738023,
      "grad_norm": 0.059024930000305176,
      "learning_rate": 3.1321276875203664e-06,
      "loss": 0.0022,
      "step": 205710
    },
    {
      "epoch": 1.3194241789675885,
      "grad_norm": 0.09479758143424988,
      "learning_rate": 3.13160851826514e-06,
      "loss": 0.0007,
      "step": 205720
    },
    {
      "epoch": 1.3194883158613746,
      "grad_norm": 0.08601735532283783,
      "learning_rate": 3.131089372422196e-06,
      "loss": 0.0009,
      "step": 205730
    },
    {
      "epoch": 1.3195524527551605,
      "grad_norm": 0.07166753709316254,
      "learning_rate": 3.130570249998036e-06,
      "loss": 0.0017,
      "step": 205740
    },
    {
      "epoch": 1.3196165896489467,
      "grad_norm": 0.026545114815235138,
      "learning_rate": 3.130051150999168e-06,
      "loss": 0.001,
      "step": 205750
    },
    {
      "epoch": 1.3196807265427328,
      "grad_norm": 0.031212162226438522,
      "learning_rate": 3.1295320754320946e-06,
      "loss": 0.0019,
      "step": 205760
    },
    {
      "epoch": 1.319744863436519,
      "grad_norm": 0.21869143843650818,
      "learning_rate": 3.129013023303321e-06,
      "loss": 0.0019,
      "step": 205770
    },
    {
      "epoch": 1.319809000330305,
      "grad_norm": 0.03965316712856293,
      "learning_rate": 3.128493994619353e-06,
      "loss": 0.002,
      "step": 205780
    },
    {
      "epoch": 1.319873137224091,
      "grad_norm": 0.03560245782136917,
      "learning_rate": 3.127974989386691e-06,
      "loss": 0.0024,
      "step": 205790
    },
    {
      "epoch": 1.3199372741178772,
      "grad_norm": 0.06623739004135132,
      "learning_rate": 3.1274560076118423e-06,
      "loss": 0.0009,
      "step": 205800
    },
    {
      "epoch": 1.3200014110116634,
      "grad_norm": 0.06063716113567352,
      "learning_rate": 3.126937049301306e-06,
      "loss": 0.002,
      "step": 205810
    },
    {
      "epoch": 1.3200655479054495,
      "grad_norm": 0.08815579116344452,
      "learning_rate": 3.1264181144615877e-06,
      "loss": 0.0011,
      "step": 205820
    },
    {
      "epoch": 1.3201296847992354,
      "grad_norm": 0.04997415840625763,
      "learning_rate": 3.12589920309919e-06,
      "loss": 0.0019,
      "step": 205830
    },
    {
      "epoch": 1.3201938216930216,
      "grad_norm": 0.09462188929319382,
      "learning_rate": 3.1253803152206153e-06,
      "loss": 0.0014,
      "step": 205840
    },
    {
      "epoch": 1.3202579585868077,
      "grad_norm": 0.2312559187412262,
      "learning_rate": 3.1248614508323632e-06,
      "loss": 0.0014,
      "step": 205850
    },
    {
      "epoch": 1.3203220954805936,
      "grad_norm": 0.05016280338168144,
      "learning_rate": 3.124342609940939e-06,
      "loss": 0.0013,
      "step": 205860
    },
    {
      "epoch": 1.3203862323743798,
      "grad_norm": 0.11266014724969864,
      "learning_rate": 3.1238237925528414e-06,
      "loss": 0.0018,
      "step": 205870
    },
    {
      "epoch": 1.320450369268166,
      "grad_norm": 0.06293545663356781,
      "learning_rate": 3.1233049986745727e-06,
      "loss": 0.0017,
      "step": 205880
    },
    {
      "epoch": 1.320514506161952,
      "grad_norm": 0.022486304864287376,
      "learning_rate": 3.122786228312633e-06,
      "loss": 0.0014,
      "step": 205890
    },
    {
      "epoch": 1.3205786430557382,
      "grad_norm": 0.2625771760940552,
      "learning_rate": 3.1222674814735256e-06,
      "loss": 0.0017,
      "step": 205900
    },
    {
      "epoch": 1.3206427799495244,
      "grad_norm": 0.0930396020412445,
      "learning_rate": 3.121748758163746e-06,
      "loss": 0.0015,
      "step": 205910
    },
    {
      "epoch": 1.3207069168433103,
      "grad_norm": 0.02263970486819744,
      "learning_rate": 3.121230058389798e-06,
      "loss": 0.0008,
      "step": 205920
    },
    {
      "epoch": 1.3207710537370965,
      "grad_norm": 0.2657495141029358,
      "learning_rate": 3.120711382158181e-06,
      "loss": 0.0016,
      "step": 205930
    },
    {
      "epoch": 1.3208351906308826,
      "grad_norm": 0.13245417177677155,
      "learning_rate": 3.120192729475392e-06,
      "loss": 0.0011,
      "step": 205940
    },
    {
      "epoch": 1.3208993275246685,
      "grad_norm": 0.24873270094394684,
      "learning_rate": 3.1196741003479324e-06,
      "loss": 0.0016,
      "step": 205950
    },
    {
      "epoch": 1.3209634644184547,
      "grad_norm": 0.02384812943637371,
      "learning_rate": 3.1191554947823e-06,
      "loss": 0.0013,
      "step": 205960
    },
    {
      "epoch": 1.3210276013122408,
      "grad_norm": 0.03216005489230156,
      "learning_rate": 3.1186369127849934e-06,
      "loss": 0.0037,
      "step": 205970
    },
    {
      "epoch": 1.321091738206027,
      "grad_norm": 0.039360105991363525,
      "learning_rate": 3.118118354362511e-06,
      "loss": 0.0012,
      "step": 205980
    },
    {
      "epoch": 1.3211558750998131,
      "grad_norm": 0.19784843921661377,
      "learning_rate": 3.1175998195213508e-06,
      "loss": 0.0012,
      "step": 205990
    },
    {
      "epoch": 1.321220011993599,
      "grad_norm": 0.06966821849346161,
      "learning_rate": 3.1170813082680094e-06,
      "loss": 0.0024,
      "step": 206000
    },
    {
      "epoch": 1.3212841488873852,
      "grad_norm": 0.02089749276638031,
      "learning_rate": 3.116562820608986e-06,
      "loss": 0.0007,
      "step": 206010
    },
    {
      "epoch": 1.3213482857811714,
      "grad_norm": 0.039188411086797714,
      "learning_rate": 3.116044356550776e-06,
      "loss": 0.0016,
      "step": 206020
    },
    {
      "epoch": 1.3214124226749575,
      "grad_norm": 0.13055889308452606,
      "learning_rate": 3.1155259160998767e-06,
      "loss": 0.0017,
      "step": 206030
    },
    {
      "epoch": 1.3214765595687434,
      "grad_norm": 0.1189856305718422,
      "learning_rate": 3.1150074992627844e-06,
      "loss": 0.0012,
      "step": 206040
    },
    {
      "epoch": 1.3215406964625296,
      "grad_norm": 0.08534970134496689,
      "learning_rate": 3.114489106045995e-06,
      "loss": 0.0015,
      "step": 206050
    },
    {
      "epoch": 1.3216048333563157,
      "grad_norm": 0.05722140148282051,
      "learning_rate": 3.1139707364560035e-06,
      "loss": 0.0012,
      "step": 206060
    },
    {
      "epoch": 1.3216689702501019,
      "grad_norm": 0.04165022820234299,
      "learning_rate": 3.1134523904993088e-06,
      "loss": 0.0009,
      "step": 206070
    },
    {
      "epoch": 1.321733107143888,
      "grad_norm": 0.09680133312940598,
      "learning_rate": 3.112934068182402e-06,
      "loss": 0.001,
      "step": 206080
    },
    {
      "epoch": 1.321797244037674,
      "grad_norm": 0.1394525021314621,
      "learning_rate": 3.112415769511782e-06,
      "loss": 0.0012,
      "step": 206090
    },
    {
      "epoch": 1.32186138093146,
      "grad_norm": 0.06444688141345978,
      "learning_rate": 3.11189749449394e-06,
      "loss": 0.001,
      "step": 206100
    },
    {
      "epoch": 1.3219255178252463,
      "grad_norm": 0.13324974477291107,
      "learning_rate": 3.111379243135373e-06,
      "loss": 0.0016,
      "step": 206110
    },
    {
      "epoch": 1.3219896547190322,
      "grad_norm": 0.21047130227088928,
      "learning_rate": 3.1108610154425733e-06,
      "loss": 0.0016,
      "step": 206120
    },
    {
      "epoch": 1.3220537916128183,
      "grad_norm": 0.03437228500843048,
      "learning_rate": 3.110342811422036e-06,
      "loss": 0.0018,
      "step": 206130
    },
    {
      "epoch": 1.3221179285066045,
      "grad_norm": 0.07776031643152237,
      "learning_rate": 3.109824631080252e-06,
      "loss": 0.0017,
      "step": 206140
    },
    {
      "epoch": 1.3221820654003906,
      "grad_norm": 0.0849027931690216,
      "learning_rate": 3.1093064744237177e-06,
      "loss": 0.0012,
      "step": 206150
    },
    {
      "epoch": 1.3222462022941768,
      "grad_norm": 0.005785741843283176,
      "learning_rate": 3.108788341458924e-06,
      "loss": 0.0028,
      "step": 206160
    },
    {
      "epoch": 1.322310339187963,
      "grad_norm": 0.1434858739376068,
      "learning_rate": 3.108270232192364e-06,
      "loss": 0.0007,
      "step": 206170
    },
    {
      "epoch": 1.3223744760817489,
      "grad_norm": 0.12548981606960297,
      "learning_rate": 3.1077521466305304e-06,
      "loss": 0.0012,
      "step": 206180
    },
    {
      "epoch": 1.322438612975535,
      "grad_norm": 0.2050582319498062,
      "learning_rate": 3.1072340847799137e-06,
      "loss": 0.0025,
      "step": 206190
    },
    {
      "epoch": 1.3225027498693211,
      "grad_norm": 0.18678708374500275,
      "learning_rate": 3.106716046647008e-06,
      "loss": 0.0016,
      "step": 206200
    },
    {
      "epoch": 1.322566886763107,
      "grad_norm": 0.12373144179582596,
      "learning_rate": 3.106198032238302e-06,
      "loss": 0.0015,
      "step": 206210
    },
    {
      "epoch": 1.3226310236568932,
      "grad_norm": 0.013299357146024704,
      "learning_rate": 3.10568004156029e-06,
      "loss": 0.0015,
      "step": 206220
    },
    {
      "epoch": 1.3226951605506794,
      "grad_norm": 0.23023241758346558,
      "learning_rate": 3.1051620746194595e-06,
      "loss": 0.0019,
      "step": 206230
    },
    {
      "epoch": 1.3227592974444655,
      "grad_norm": 0.054234281182289124,
      "learning_rate": 3.104644131422303e-06,
      "loss": 0.0018,
      "step": 206240
    },
    {
      "epoch": 1.3228234343382517,
      "grad_norm": 0.1322822868824005,
      "learning_rate": 3.1041262119753097e-06,
      "loss": 0.0017,
      "step": 206250
    },
    {
      "epoch": 1.3228875712320376,
      "grad_norm": 0.16184493899345398,
      "learning_rate": 3.1036083162849706e-06,
      "loss": 0.0016,
      "step": 206260
    },
    {
      "epoch": 1.3229517081258237,
      "grad_norm": 0.057139068841934204,
      "learning_rate": 3.1030904443577736e-06,
      "loss": 0.0008,
      "step": 206270
    },
    {
      "epoch": 1.32301584501961,
      "grad_norm": 0.042741693556308746,
      "learning_rate": 3.10257259620021e-06,
      "loss": 0.0012,
      "step": 206280
    },
    {
      "epoch": 1.3230799819133958,
      "grad_norm": 0.008121524937450886,
      "learning_rate": 3.1020547718187673e-06,
      "loss": 0.0019,
      "step": 206290
    },
    {
      "epoch": 1.323144118807182,
      "grad_norm": 0.09957344830036163,
      "learning_rate": 3.101536971219935e-06,
      "loss": 0.002,
      "step": 206300
    },
    {
      "epoch": 1.3232082557009681,
      "grad_norm": 0.028351394459605217,
      "learning_rate": 3.1010191944102007e-06,
      "loss": 0.001,
      "step": 206310
    },
    {
      "epoch": 1.3232723925947543,
      "grad_norm": 0.046477172523736954,
      "learning_rate": 3.1005014413960534e-06,
      "loss": 0.0031,
      "step": 206320
    },
    {
      "epoch": 1.3233365294885404,
      "grad_norm": 0.21906918287277222,
      "learning_rate": 3.0999837121839817e-06,
      "loss": 0.0015,
      "step": 206330
    },
    {
      "epoch": 1.3234006663823266,
      "grad_norm": 0.03294725343585014,
      "learning_rate": 3.09946600678047e-06,
      "loss": 0.0011,
      "step": 206340
    },
    {
      "epoch": 1.3234648032761125,
      "grad_norm": 0.048295244574546814,
      "learning_rate": 3.098948325192009e-06,
      "loss": 0.0013,
      "step": 206350
    },
    {
      "epoch": 1.3235289401698986,
      "grad_norm": 0.1487571895122528,
      "learning_rate": 3.098430667425083e-06,
      "loss": 0.0016,
      "step": 206360
    },
    {
      "epoch": 1.3235930770636848,
      "grad_norm": 0.07452013343572617,
      "learning_rate": 3.097913033486182e-06,
      "loss": 0.0011,
      "step": 206370
    },
    {
      "epoch": 1.3236572139574707,
      "grad_norm": 0.011145166121423244,
      "learning_rate": 3.097395423381787e-06,
      "loss": 0.001,
      "step": 206380
    },
    {
      "epoch": 1.3237213508512569,
      "grad_norm": 0.05800461024045944,
      "learning_rate": 3.0968778371183893e-06,
      "loss": 0.0011,
      "step": 206390
    },
    {
      "epoch": 1.323785487745043,
      "grad_norm": 0.033336617052555084,
      "learning_rate": 3.0963602747024717e-06,
      "loss": 0.0014,
      "step": 206400
    },
    {
      "epoch": 1.3238496246388292,
      "grad_norm": 0.06338828057050705,
      "learning_rate": 3.0958427361405207e-06,
      "loss": 0.0008,
      "step": 206410
    },
    {
      "epoch": 1.3239137615326153,
      "grad_norm": 0.09692800045013428,
      "learning_rate": 3.0953252214390207e-06,
      "loss": 0.0019,
      "step": 206420
    },
    {
      "epoch": 1.3239778984264012,
      "grad_norm": 0.2921394407749176,
      "learning_rate": 3.094807730604458e-06,
      "loss": 0.0012,
      "step": 206430
    },
    {
      "epoch": 1.3240420353201874,
      "grad_norm": 0.04953905940055847,
      "learning_rate": 3.094290263643315e-06,
      "loss": 0.0019,
      "step": 206440
    },
    {
      "epoch": 1.3241061722139735,
      "grad_norm": 0.047770023345947266,
      "learning_rate": 3.093772820562077e-06,
      "loss": 0.0019,
      "step": 206450
    },
    {
      "epoch": 1.3241703091077597,
      "grad_norm": 0.06971164792776108,
      "learning_rate": 3.0932554013672277e-06,
      "loss": 0.0013,
      "step": 206460
    },
    {
      "epoch": 1.3242344460015456,
      "grad_norm": 0.16142980754375458,
      "learning_rate": 3.0927380060652512e-06,
      "loss": 0.0019,
      "step": 206470
    },
    {
      "epoch": 1.3242985828953318,
      "grad_norm": 0.14970916509628296,
      "learning_rate": 3.092220634662631e-06,
      "loss": 0.0013,
      "step": 206480
    },
    {
      "epoch": 1.324362719789118,
      "grad_norm": 0.03817512094974518,
      "learning_rate": 3.091703287165849e-06,
      "loss": 0.0013,
      "step": 206490
    },
    {
      "epoch": 1.324426856682904,
      "grad_norm": 0.0302517581731081,
      "learning_rate": 3.0911859635813896e-06,
      "loss": 0.0014,
      "step": 206500
    },
    {
      "epoch": 1.3244909935766902,
      "grad_norm": 0.03984169661998749,
      "learning_rate": 3.0906686639157335e-06,
      "loss": 0.0008,
      "step": 206510
    },
    {
      "epoch": 1.3245551304704761,
      "grad_norm": 0.09860210120677948,
      "learning_rate": 3.090151388175364e-06,
      "loss": 0.0032,
      "step": 206520
    },
    {
      "epoch": 1.3246192673642623,
      "grad_norm": 0.1225770115852356,
      "learning_rate": 3.089634136366762e-06,
      "loss": 0.0011,
      "step": 206530
    },
    {
      "epoch": 1.3246834042580484,
      "grad_norm": 0.028699776157736778,
      "learning_rate": 3.0891169084964106e-06,
      "loss": 0.0008,
      "step": 206540
    },
    {
      "epoch": 1.3247475411518344,
      "grad_norm": 0.01483263447880745,
      "learning_rate": 3.088599704570789e-06,
      "loss": 0.0011,
      "step": 206550
    },
    {
      "epoch": 1.3248116780456205,
      "grad_norm": 0.08767399191856384,
      "learning_rate": 3.0880825245963796e-06,
      "loss": 0.0017,
      "step": 206560
    },
    {
      "epoch": 1.3248758149394066,
      "grad_norm": 0.11258254200220108,
      "learning_rate": 3.0875653685796624e-06,
      "loss": 0.0012,
      "step": 206570
    },
    {
      "epoch": 1.3249399518331928,
      "grad_norm": 0.012691079638898373,
      "learning_rate": 3.087048236527119e-06,
      "loss": 0.0015,
      "step": 206580
    },
    {
      "epoch": 1.325004088726979,
      "grad_norm": 0.06870340555906296,
      "learning_rate": 3.086531128445227e-06,
      "loss": 0.001,
      "step": 206590
    },
    {
      "epoch": 1.325068225620765,
      "grad_norm": 0.3031693994998932,
      "learning_rate": 3.0860140443404686e-06,
      "loss": 0.0024,
      "step": 206600
    },
    {
      "epoch": 1.325132362514551,
      "grad_norm": 0.1562807261943817,
      "learning_rate": 3.0854969842193206e-06,
      "loss": 0.0008,
      "step": 206610
    },
    {
      "epoch": 1.3251964994083372,
      "grad_norm": 0.1482306867837906,
      "learning_rate": 3.084979948088265e-06,
      "loss": 0.0012,
      "step": 206620
    },
    {
      "epoch": 1.3252606363021233,
      "grad_norm": 0.19808292388916016,
      "learning_rate": 3.0844629359537788e-06,
      "loss": 0.0022,
      "step": 206630
    },
    {
      "epoch": 1.3253247731959092,
      "grad_norm": 0.18784743547439575,
      "learning_rate": 3.0839459478223404e-06,
      "loss": 0.0023,
      "step": 206640
    },
    {
      "epoch": 1.3253889100896954,
      "grad_norm": 0.06134854257106781,
      "learning_rate": 3.08342898370043e-06,
      "loss": 0.0016,
      "step": 206650
    },
    {
      "epoch": 1.3254530469834815,
      "grad_norm": 0.05111366882920265,
      "learning_rate": 3.082912043594523e-06,
      "loss": 0.0012,
      "step": 206660
    },
    {
      "epoch": 1.3255171838772677,
      "grad_norm": 0.0958254337310791,
      "learning_rate": 3.0823951275111e-06,
      "loss": 0.0012,
      "step": 206670
    },
    {
      "epoch": 1.3255813207710538,
      "grad_norm": 0.075313039124012,
      "learning_rate": 3.0818782354566346e-06,
      "loss": 0.0017,
      "step": 206680
    },
    {
      "epoch": 1.3256454576648398,
      "grad_norm": 0.07610893994569778,
      "learning_rate": 3.0813613674376064e-06,
      "loss": 0.0012,
      "step": 206690
    },
    {
      "epoch": 1.325709594558626,
      "grad_norm": 0.01589725725352764,
      "learning_rate": 3.080844523460491e-06,
      "loss": 0.0016,
      "step": 206700
    },
    {
      "epoch": 1.325773731452412,
      "grad_norm": 0.06068975478410721,
      "learning_rate": 3.080327703531767e-06,
      "loss": 0.0052,
      "step": 206710
    },
    {
      "epoch": 1.325837868346198,
      "grad_norm": 0.06539834290742874,
      "learning_rate": 3.0798109076579074e-06,
      "loss": 0.0018,
      "step": 206720
    },
    {
      "epoch": 1.3259020052399841,
      "grad_norm": 0.06082136556506157,
      "learning_rate": 3.07929413584539e-06,
      "loss": 0.002,
      "step": 206730
    },
    {
      "epoch": 1.3259661421337703,
      "grad_norm": 0.1441694051027298,
      "learning_rate": 3.0787773881006887e-06,
      "loss": 0.002,
      "step": 206740
    },
    {
      "epoch": 1.3260302790275564,
      "grad_norm": 0.03906743600964546,
      "learning_rate": 3.0782606644302816e-06,
      "loss": 0.0015,
      "step": 206750
    },
    {
      "epoch": 1.3260944159213426,
      "grad_norm": 0.014613121747970581,
      "learning_rate": 3.0777439648406404e-06,
      "loss": 0.001,
      "step": 206760
    },
    {
      "epoch": 1.3261585528151287,
      "grad_norm": 0.06862866133451462,
      "learning_rate": 3.0772272893382416e-06,
      "loss": 0.0014,
      "step": 206770
    },
    {
      "epoch": 1.3262226897089147,
      "grad_norm": 0.05738750472664833,
      "learning_rate": 3.07671063792956e-06,
      "loss": 0.0035,
      "step": 206780
    },
    {
      "epoch": 1.3262868266027008,
      "grad_norm": 0.10485716909170151,
      "learning_rate": 3.0761940106210665e-06,
      "loss": 0.0019,
      "step": 206790
    },
    {
      "epoch": 1.326350963496487,
      "grad_norm": 0.08685830235481262,
      "learning_rate": 3.0756774074192385e-06,
      "loss": 0.0021,
      "step": 206800
    },
    {
      "epoch": 1.3264151003902729,
      "grad_norm": 0.10155344754457474,
      "learning_rate": 3.075160828330548e-06,
      "loss": 0.0026,
      "step": 206810
    },
    {
      "epoch": 1.326479237284059,
      "grad_norm": 0.09273690730333328,
      "learning_rate": 3.0746442733614655e-06,
      "loss": 0.0014,
      "step": 206820
    },
    {
      "epoch": 1.3265433741778452,
      "grad_norm": 0.10257264971733093,
      "learning_rate": 3.074127742518469e-06,
      "loss": 0.0016,
      "step": 206830
    },
    {
      "epoch": 1.3266075110716313,
      "grad_norm": 0.0632232204079628,
      "learning_rate": 3.0736112358080274e-06,
      "loss": 0.0012,
      "step": 206840
    },
    {
      "epoch": 1.3266716479654175,
      "grad_norm": 0.14479869604110718,
      "learning_rate": 3.073094753236614e-06,
      "loss": 0.0023,
      "step": 206850
    },
    {
      "epoch": 1.3267357848592034,
      "grad_norm": 0.10152845829725266,
      "learning_rate": 3.072578294810701e-06,
      "loss": 0.0005,
      "step": 206860
    },
    {
      "epoch": 1.3267999217529896,
      "grad_norm": 0.0303544569760561,
      "learning_rate": 3.072061860536759e-06,
      "loss": 0.002,
      "step": 206870
    },
    {
      "epoch": 1.3268640586467757,
      "grad_norm": 0.05967644974589348,
      "learning_rate": 3.071545450421259e-06,
      "loss": 0.0019,
      "step": 206880
    },
    {
      "epoch": 1.3269281955405619,
      "grad_norm": 0.03381352126598358,
      "learning_rate": 3.0710290644706732e-06,
      "loss": 0.0024,
      "step": 206890
    },
    {
      "epoch": 1.3269923324343478,
      "grad_norm": 0.01893250085413456,
      "learning_rate": 3.0705127026914726e-06,
      "loss": 0.0009,
      "step": 206900
    },
    {
      "epoch": 1.327056469328134,
      "grad_norm": 0.128096342086792,
      "learning_rate": 3.069996365090126e-06,
      "loss": 0.0012,
      "step": 206910
    },
    {
      "epoch": 1.32712060622192,
      "grad_norm": 0.08662473410367966,
      "learning_rate": 3.069480051673105e-06,
      "loss": 0.0011,
      "step": 206920
    },
    {
      "epoch": 1.3271847431157062,
      "grad_norm": 0.019852977246046066,
      "learning_rate": 3.0689637624468783e-06,
      "loss": 0.0012,
      "step": 206930
    },
    {
      "epoch": 1.3272488800094924,
      "grad_norm": 0.1014653667807579,
      "learning_rate": 3.0684474974179163e-06,
      "loss": 0.0014,
      "step": 206940
    },
    {
      "epoch": 1.3273130169032783,
      "grad_norm": 0.03182956203818321,
      "learning_rate": 3.0679312565926865e-06,
      "loss": 0.0014,
      "step": 206950
    },
    {
      "epoch": 1.3273771537970644,
      "grad_norm": 0.2458362728357315,
      "learning_rate": 3.0674150399776603e-06,
      "loss": 0.002,
      "step": 206960
    },
    {
      "epoch": 1.3274412906908506,
      "grad_norm": 0.09880047291517258,
      "learning_rate": 3.066898847579303e-06,
      "loss": 0.001,
      "step": 206970
    },
    {
      "epoch": 1.3275054275846365,
      "grad_norm": 0.09074307978153229,
      "learning_rate": 3.0663826794040863e-06,
      "loss": 0.0008,
      "step": 206980
    },
    {
      "epoch": 1.3275695644784227,
      "grad_norm": 0.020720677450299263,
      "learning_rate": 3.065866535458476e-06,
      "loss": 0.0017,
      "step": 206990
    },
    {
      "epoch": 1.3276337013722088,
      "grad_norm": 0.07115289568901062,
      "learning_rate": 3.0653504157489407e-06,
      "loss": 0.002,
      "step": 207000
    },
    {
      "epoch": 1.327697838265995,
      "grad_norm": 0.15864945948123932,
      "learning_rate": 3.064834320281946e-06,
      "loss": 0.0012,
      "step": 207010
    },
    {
      "epoch": 1.3277619751597811,
      "grad_norm": 0.09046715497970581,
      "learning_rate": 3.064318249063962e-06,
      "loss": 0.0022,
      "step": 207020
    },
    {
      "epoch": 1.3278261120535673,
      "grad_norm": 0.07268185168504715,
      "learning_rate": 3.063802202101453e-06,
      "loss": 0.0007,
      "step": 207030
    },
    {
      "epoch": 1.3278902489473532,
      "grad_norm": 0.10096162557601929,
      "learning_rate": 3.0632861794008858e-06,
      "loss": 0.001,
      "step": 207040
    },
    {
      "epoch": 1.3279543858411393,
      "grad_norm": 0.11104384809732437,
      "learning_rate": 3.062770180968728e-06,
      "loss": 0.0012,
      "step": 207050
    },
    {
      "epoch": 1.3280185227349255,
      "grad_norm": 0.13525117933750153,
      "learning_rate": 3.062254206811443e-06,
      "loss": 0.0016,
      "step": 207060
    },
    {
      "epoch": 1.3280826596287114,
      "grad_norm": 0.026424003764986992,
      "learning_rate": 3.0617382569354993e-06,
      "loss": 0.003,
      "step": 207070
    },
    {
      "epoch": 1.3281467965224976,
      "grad_norm": 0.1317053735256195,
      "learning_rate": 3.0612223313473587e-06,
      "loss": 0.0011,
      "step": 207080
    },
    {
      "epoch": 1.3282109334162837,
      "grad_norm": 0.17248418927192688,
      "learning_rate": 3.0607064300534893e-06,
      "loss": 0.0013,
      "step": 207090
    },
    {
      "epoch": 1.3282750703100699,
      "grad_norm": 0.041471950709819794,
      "learning_rate": 3.060190553060353e-06,
      "loss": 0.0014,
      "step": 207100
    },
    {
      "epoch": 1.328339207203856,
      "grad_norm": 0.008954065851867199,
      "learning_rate": 3.059674700374417e-06,
      "loss": 0.0009,
      "step": 207110
    },
    {
      "epoch": 1.328403344097642,
      "grad_norm": 0.20160478353500366,
      "learning_rate": 3.0591588720021427e-06,
      "loss": 0.0016,
      "step": 207120
    },
    {
      "epoch": 1.328467480991428,
      "grad_norm": 0.06734311580657959,
      "learning_rate": 3.0586430679499956e-06,
      "loss": 0.0016,
      "step": 207130
    },
    {
      "epoch": 1.3285316178852142,
      "grad_norm": 0.15078191459178925,
      "learning_rate": 3.0581272882244374e-06,
      "loss": 0.0034,
      "step": 207140
    },
    {
      "epoch": 1.3285957547790002,
      "grad_norm": 0.06340338289737701,
      "learning_rate": 3.0576115328319334e-06,
      "loss": 0.0015,
      "step": 207150
    },
    {
      "epoch": 1.3286598916727863,
      "grad_norm": 0.028959598392248154,
      "learning_rate": 3.057095801778943e-06,
      "loss": 0.0007,
      "step": 207160
    },
    {
      "epoch": 1.3287240285665725,
      "grad_norm": 0.1481342911720276,
      "learning_rate": 3.0565800950719317e-06,
      "loss": 0.0013,
      "step": 207170
    },
    {
      "epoch": 1.3287881654603586,
      "grad_norm": 0.031880080699920654,
      "learning_rate": 3.0560644127173614e-06,
      "loss": 0.0013,
      "step": 207180
    },
    {
      "epoch": 1.3288523023541448,
      "grad_norm": 0.08684264123439789,
      "learning_rate": 3.0555487547216927e-06,
      "loss": 0.0016,
      "step": 207190
    },
    {
      "epoch": 1.328916439247931,
      "grad_norm": 0.2443089485168457,
      "learning_rate": 3.0550331210913887e-06,
      "loss": 0.0036,
      "step": 207200
    },
    {
      "epoch": 1.3289805761417168,
      "grad_norm": 0.09374405443668365,
      "learning_rate": 3.0545175118329085e-06,
      "loss": 0.0012,
      "step": 207210
    },
    {
      "epoch": 1.329044713035503,
      "grad_norm": 0.02969098649919033,
      "learning_rate": 3.0540019269527155e-06,
      "loss": 0.0008,
      "step": 207220
    },
    {
      "epoch": 1.3291088499292891,
      "grad_norm": 0.12158779054880142,
      "learning_rate": 3.0534863664572686e-06,
      "loss": 0.0016,
      "step": 207230
    },
    {
      "epoch": 1.329172986823075,
      "grad_norm": 0.013619618490338326,
      "learning_rate": 3.052970830353029e-06,
      "loss": 0.0007,
      "step": 207240
    },
    {
      "epoch": 1.3292371237168612,
      "grad_norm": 0.12970755994319916,
      "learning_rate": 3.0524553186464555e-06,
      "loss": 0.0019,
      "step": 207250
    },
    {
      "epoch": 1.3293012606106474,
      "grad_norm": 0.15070787072181702,
      "learning_rate": 3.0519398313440094e-06,
      "loss": 0.0009,
      "step": 207260
    },
    {
      "epoch": 1.3293653975044335,
      "grad_norm": 0.31933197379112244,
      "learning_rate": 3.0514243684521496e-06,
      "loss": 0.0017,
      "step": 207270
    },
    {
      "epoch": 1.3294295343982196,
      "grad_norm": 0.09706978499889374,
      "learning_rate": 3.0509089299773354e-06,
      "loss": 0.0015,
      "step": 207280
    },
    {
      "epoch": 1.3294936712920056,
      "grad_norm": 0.021569713950157166,
      "learning_rate": 3.0503935159260234e-06,
      "loss": 0.0012,
      "step": 207290
    },
    {
      "epoch": 1.3295578081857917,
      "grad_norm": 0.031222287565469742,
      "learning_rate": 3.0498781263046763e-06,
      "loss": 0.001,
      "step": 207300
    },
    {
      "epoch": 1.3296219450795779,
      "grad_norm": 0.0353911817073822,
      "learning_rate": 3.0493627611197483e-06,
      "loss": 0.004,
      "step": 207310
    },
    {
      "epoch": 1.329686081973364,
      "grad_norm": 0.2569740414619446,
      "learning_rate": 3.0488474203776996e-06,
      "loss": 0.003,
      "step": 207320
    },
    {
      "epoch": 1.32975021886715,
      "grad_norm": 0.12398944050073624,
      "learning_rate": 3.048332104084988e-06,
      "loss": 0.0023,
      "step": 207330
    },
    {
      "epoch": 1.329814355760936,
      "grad_norm": 0.07236245274543762,
      "learning_rate": 3.047816812248069e-06,
      "loss": 0.0013,
      "step": 207340
    },
    {
      "epoch": 1.3298784926547222,
      "grad_norm": 0.1026138961315155,
      "learning_rate": 3.047301544873401e-06,
      "loss": 0.0018,
      "step": 207350
    },
    {
      "epoch": 1.3299426295485084,
      "grad_norm": 0.0675225779414177,
      "learning_rate": 3.04678630196744e-06,
      "loss": 0.0011,
      "step": 207360
    },
    {
      "epoch": 1.3300067664422945,
      "grad_norm": 0.1957700103521347,
      "learning_rate": 3.0462710835366426e-06,
      "loss": 0.0014,
      "step": 207370
    },
    {
      "epoch": 1.3300709033360805,
      "grad_norm": 0.2093002051115036,
      "learning_rate": 3.045755889587465e-06,
      "loss": 0.0015,
      "step": 207380
    },
    {
      "epoch": 1.3301350402298666,
      "grad_norm": 0.07760210335254669,
      "learning_rate": 3.0452407201263633e-06,
      "loss": 0.0014,
      "step": 207390
    },
    {
      "epoch": 1.3301991771236528,
      "grad_norm": 0.3380934000015259,
      "learning_rate": 3.044725575159791e-06,
      "loss": 0.001,
      "step": 207400
    },
    {
      "epoch": 1.3302633140174387,
      "grad_norm": 0.13043218851089478,
      "learning_rate": 3.044210454694206e-06,
      "loss": 0.0019,
      "step": 207410
    },
    {
      "epoch": 1.3303274509112248,
      "grad_norm": 0.010576743632555008,
      "learning_rate": 3.0436953587360607e-06,
      "loss": 0.0019,
      "step": 207420
    },
    {
      "epoch": 1.330391587805011,
      "grad_norm": 0.13296468555927277,
      "learning_rate": 3.0431802872918114e-06,
      "loss": 0.0079,
      "step": 207430
    },
    {
      "epoch": 1.3304557246987971,
      "grad_norm": 0.06005353108048439,
      "learning_rate": 3.0426652403679115e-06,
      "loss": 0.0024,
      "step": 207440
    },
    {
      "epoch": 1.3305198615925833,
      "grad_norm": 0.1177959218621254,
      "learning_rate": 3.042150217970815e-06,
      "loss": 0.0009,
      "step": 207450
    },
    {
      "epoch": 1.3305839984863694,
      "grad_norm": 0.005490519572049379,
      "learning_rate": 3.0416352201069753e-06,
      "loss": 0.0015,
      "step": 207460
    },
    {
      "epoch": 1.3306481353801554,
      "grad_norm": 0.42011868953704834,
      "learning_rate": 3.041120246782846e-06,
      "loss": 0.0025,
      "step": 207470
    },
    {
      "epoch": 1.3307122722739415,
      "grad_norm": 0.2927258312702179,
      "learning_rate": 3.0406052980048807e-06,
      "loss": 0.0008,
      "step": 207480
    },
    {
      "epoch": 1.3307764091677277,
      "grad_norm": 0.2572489380836487,
      "learning_rate": 3.04009037377953e-06,
      "loss": 0.0016,
      "step": 207490
    },
    {
      "epoch": 1.3308405460615136,
      "grad_norm": 0.022028865292668343,
      "learning_rate": 3.039575474113249e-06,
      "loss": 0.0009,
      "step": 207500
    },
    {
      "epoch": 1.3309046829552997,
      "grad_norm": 0.08524273335933685,
      "learning_rate": 3.0390605990124875e-06,
      "loss": 0.001,
      "step": 207510
    },
    {
      "epoch": 1.3309688198490859,
      "grad_norm": 0.03125406801700592,
      "learning_rate": 3.0385457484836987e-06,
      "loss": 0.0008,
      "step": 207520
    },
    {
      "epoch": 1.331032956742872,
      "grad_norm": 0.2823640704154968,
      "learning_rate": 3.038030922533333e-06,
      "loss": 0.0026,
      "step": 207530
    },
    {
      "epoch": 1.3310970936366582,
      "grad_norm": 0.0846913754940033,
      "learning_rate": 3.0375161211678426e-06,
      "loss": 0.001,
      "step": 207540
    },
    {
      "epoch": 1.331161230530444,
      "grad_norm": 0.13597342371940613,
      "learning_rate": 3.0370013443936773e-06,
      "loss": 0.0014,
      "step": 207550
    },
    {
      "epoch": 1.3312253674242303,
      "grad_norm": 0.06595153361558914,
      "learning_rate": 3.036486592217289e-06,
      "loss": 0.0016,
      "step": 207560
    },
    {
      "epoch": 1.3312895043180164,
      "grad_norm": 0.0930728167295456,
      "learning_rate": 3.035971864645126e-06,
      "loss": 0.0012,
      "step": 207570
    },
    {
      "epoch": 1.3313536412118026,
      "grad_norm": 0.018666686490178108,
      "learning_rate": 3.0354571616836398e-06,
      "loss": 0.0013,
      "step": 207580
    },
    {
      "epoch": 1.3314177781055885,
      "grad_norm": 0.07905947417020798,
      "learning_rate": 3.0349424833392765e-06,
      "loss": 0.0016,
      "step": 207590
    },
    {
      "epoch": 1.3314819149993746,
      "grad_norm": 0.14699843525886536,
      "learning_rate": 3.0344278296184913e-06,
      "loss": 0.0014,
      "step": 207600
    },
    {
      "epoch": 1.3315460518931608,
      "grad_norm": 0.12528088688850403,
      "learning_rate": 3.0339132005277294e-06,
      "loss": 0.0011,
      "step": 207610
    },
    {
      "epoch": 1.331610188786947,
      "grad_norm": 0.09150472283363342,
      "learning_rate": 3.033398596073441e-06,
      "loss": 0.0011,
      "step": 207620
    },
    {
      "epoch": 1.331674325680733,
      "grad_norm": 0.016758302226662636,
      "learning_rate": 3.032884016262073e-06,
      "loss": 0.0009,
      "step": 207630
    },
    {
      "epoch": 1.331738462574519,
      "grad_norm": 0.10561700910329819,
      "learning_rate": 3.0323694611000753e-06,
      "loss": 0.001,
      "step": 207640
    },
    {
      "epoch": 1.3318025994683051,
      "grad_norm": 0.06780321896076202,
      "learning_rate": 3.031854930593893e-06,
      "loss": 0.003,
      "step": 207650
    },
    {
      "epoch": 1.3318667363620913,
      "grad_norm": 0.0012612667633220553,
      "learning_rate": 3.0313404247499763e-06,
      "loss": 0.0018,
      "step": 207660
    },
    {
      "epoch": 1.3319308732558772,
      "grad_norm": 0.009958547540009022,
      "learning_rate": 3.03082594357477e-06,
      "loss": 0.0015,
      "step": 207670
    },
    {
      "epoch": 1.3319950101496634,
      "grad_norm": 0.019118620082736015,
      "learning_rate": 3.030311487074723e-06,
      "loss": 0.0041,
      "step": 207680
    },
    {
      "epoch": 1.3320591470434495,
      "grad_norm": 0.1367831826210022,
      "learning_rate": 3.0297970552562793e-06,
      "loss": 0.0021,
      "step": 207690
    },
    {
      "epoch": 1.3321232839372357,
      "grad_norm": 0.14062421023845673,
      "learning_rate": 3.0292826481258887e-06,
      "loss": 0.0018,
      "step": 207700
    },
    {
      "epoch": 1.3321874208310218,
      "grad_norm": 0.05483415350317955,
      "learning_rate": 3.028768265689993e-06,
      "loss": 0.0007,
      "step": 207710
    },
    {
      "epoch": 1.332251557724808,
      "grad_norm": 0.09449375420808792,
      "learning_rate": 3.028253907955041e-06,
      "loss": 0.0006,
      "step": 207720
    },
    {
      "epoch": 1.332315694618594,
      "grad_norm": 0.279594361782074,
      "learning_rate": 3.0277395749274757e-06,
      "loss": 0.0012,
      "step": 207730
    },
    {
      "epoch": 1.33237983151238,
      "grad_norm": 0.12302122265100479,
      "learning_rate": 3.027225266613743e-06,
      "loss": 0.0016,
      "step": 207740
    },
    {
      "epoch": 1.3324439684061662,
      "grad_norm": 0.11502698063850403,
      "learning_rate": 3.026710983020289e-06,
      "loss": 0.0016,
      "step": 207750
    },
    {
      "epoch": 1.3325081052999521,
      "grad_norm": 0.030506573617458344,
      "learning_rate": 3.026196724153555e-06,
      "loss": 0.0012,
      "step": 207760
    },
    {
      "epoch": 1.3325722421937383,
      "grad_norm": 0.07216215878725052,
      "learning_rate": 3.0256824900199886e-06,
      "loss": 0.0015,
      "step": 207770
    },
    {
      "epoch": 1.3326363790875244,
      "grad_norm": 0.06424123793840408,
      "learning_rate": 3.0251682806260298e-06,
      "loss": 0.0011,
      "step": 207780
    },
    {
      "epoch": 1.3327005159813106,
      "grad_norm": 0.05916834995150566,
      "learning_rate": 3.0246540959781257e-06,
      "loss": 0.0011,
      "step": 207790
    },
    {
      "epoch": 1.3327646528750967,
      "grad_norm": 0.3115961253643036,
      "learning_rate": 3.0241399360827157e-06,
      "loss": 0.0021,
      "step": 207800
    },
    {
      "epoch": 1.3328287897688826,
      "grad_norm": 0.05232136696577072,
      "learning_rate": 3.023625800946246e-06,
      "loss": 0.0013,
      "step": 207810
    },
    {
      "epoch": 1.3328929266626688,
      "grad_norm": 0.15427418053150177,
      "learning_rate": 3.0231116905751554e-06,
      "loss": 0.0013,
      "step": 207820
    },
    {
      "epoch": 1.332957063556455,
      "grad_norm": 0.07504183799028397,
      "learning_rate": 3.022597604975891e-06,
      "loss": 0.0011,
      "step": 207830
    },
    {
      "epoch": 1.3330212004502409,
      "grad_norm": 0.02206926792860031,
      "learning_rate": 3.0220835441548896e-06,
      "loss": 0.0006,
      "step": 207840
    },
    {
      "epoch": 1.333085337344027,
      "grad_norm": 0.04043566435575485,
      "learning_rate": 3.021569508118596e-06,
      "loss": 0.0011,
      "step": 207850
    },
    {
      "epoch": 1.3331494742378132,
      "grad_norm": 0.06484486162662506,
      "learning_rate": 3.0210554968734495e-06,
      "loss": 0.0014,
      "step": 207860
    },
    {
      "epoch": 1.3332136111315993,
      "grad_norm": 0.05188364535570145,
      "learning_rate": 3.0205415104258934e-06,
      "loss": 0.0033,
      "step": 207870
    },
    {
      "epoch": 1.3332777480253855,
      "grad_norm": 0.1079925149679184,
      "learning_rate": 3.020027548782366e-06,
      "loss": 0.0011,
      "step": 207880
    },
    {
      "epoch": 1.3333418849191716,
      "grad_norm": 0.1064772754907608,
      "learning_rate": 3.019513611949308e-06,
      "loss": 0.0007,
      "step": 207890
    },
    {
      "epoch": 1.3334060218129575,
      "grad_norm": 0.09681066870689392,
      "learning_rate": 3.0189996999331606e-06,
      "loss": 0.0009,
      "step": 207900
    },
    {
      "epoch": 1.3334701587067437,
      "grad_norm": 0.06284935772418976,
      "learning_rate": 3.018485812740363e-06,
      "loss": 0.0012,
      "step": 207910
    },
    {
      "epoch": 1.3335342956005298,
      "grad_norm": 0.05767025426030159,
      "learning_rate": 3.0179719503773543e-06,
      "loss": 0.0015,
      "step": 207920
    },
    {
      "epoch": 1.3335984324943158,
      "grad_norm": 0.09643470495939255,
      "learning_rate": 3.017458112850572e-06,
      "loss": 0.0024,
      "step": 207930
    },
    {
      "epoch": 1.333662569388102,
      "grad_norm": 0.13457752764225006,
      "learning_rate": 3.0169443001664585e-06,
      "loss": 0.0024,
      "step": 207940
    },
    {
      "epoch": 1.333726706281888,
      "grad_norm": 0.06010362133383751,
      "learning_rate": 3.0164305123314484e-06,
      "loss": 0.001,
      "step": 207950
    },
    {
      "epoch": 1.3337908431756742,
      "grad_norm": 0.04925031214952469,
      "learning_rate": 3.0159167493519836e-06,
      "loss": 0.0008,
      "step": 207960
    },
    {
      "epoch": 1.3338549800694603,
      "grad_norm": 0.08997722715139389,
      "learning_rate": 3.0154030112344985e-06,
      "loss": 0.0011,
      "step": 207970
    },
    {
      "epoch": 1.3339191169632463,
      "grad_norm": 0.13654863834381104,
      "learning_rate": 3.014889297985433e-06,
      "loss": 0.0018,
      "step": 207980
    },
    {
      "epoch": 1.3339832538570324,
      "grad_norm": 0.18833725154399872,
      "learning_rate": 3.014375609611222e-06,
      "loss": 0.0014,
      "step": 207990
    },
    {
      "epoch": 1.3340473907508186,
      "grad_norm": 0.10654346644878387,
      "learning_rate": 3.013861946118305e-06,
      "loss": 0.0012,
      "step": 208000
    },
    {
      "epoch": 1.3341115276446047,
      "grad_norm": 0.046681273728609085,
      "learning_rate": 3.0133483075131165e-06,
      "loss": 0.0031,
      "step": 208010
    },
    {
      "epoch": 1.3341756645383906,
      "grad_norm": 0.23288017511367798,
      "learning_rate": 3.0128346938020938e-06,
      "loss": 0.0012,
      "step": 208020
    },
    {
      "epoch": 1.3342398014321768,
      "grad_norm": 0.039609394967556,
      "learning_rate": 3.012321104991673e-06,
      "loss": 0.0016,
      "step": 208030
    },
    {
      "epoch": 1.334303938325963,
      "grad_norm": 0.14077578485012054,
      "learning_rate": 3.0118075410882886e-06,
      "loss": 0.002,
      "step": 208040
    },
    {
      "epoch": 1.334368075219749,
      "grad_norm": 0.0437595397233963,
      "learning_rate": 3.0112940020983773e-06,
      "loss": 0.001,
      "step": 208050
    },
    {
      "epoch": 1.3344322121135352,
      "grad_norm": 0.21762453019618988,
      "learning_rate": 3.0107804880283726e-06,
      "loss": 0.0011,
      "step": 208060
    },
    {
      "epoch": 1.3344963490073212,
      "grad_norm": 0.21231357753276825,
      "learning_rate": 3.0102669988847112e-06,
      "loss": 0.0015,
      "step": 208070
    },
    {
      "epoch": 1.3345604859011073,
      "grad_norm": 0.18998970091342926,
      "learning_rate": 3.009753534673825e-06,
      "loss": 0.0013,
      "step": 208080
    },
    {
      "epoch": 1.3346246227948935,
      "grad_norm": 0.004365130327641964,
      "learning_rate": 3.0092400954021507e-06,
      "loss": 0.0004,
      "step": 208090
    },
    {
      "epoch": 1.3346887596886794,
      "grad_norm": 0.011583004146814346,
      "learning_rate": 3.00872668107612e-06,
      "loss": 0.0018,
      "step": 208100
    },
    {
      "epoch": 1.3347528965824655,
      "grad_norm": 0.13997438549995422,
      "learning_rate": 3.0082132917021677e-06,
      "loss": 0.0022,
      "step": 208110
    },
    {
      "epoch": 1.3348170334762517,
      "grad_norm": 0.08748896420001984,
      "learning_rate": 3.0076999272867256e-06,
      "loss": 0.0011,
      "step": 208120
    },
    {
      "epoch": 1.3348811703700378,
      "grad_norm": 0.1436755210161209,
      "learning_rate": 3.007186587836228e-06,
      "loss": 0.0018,
      "step": 208130
    },
    {
      "epoch": 1.334945307263824,
      "grad_norm": 0.05570292845368385,
      "learning_rate": 3.0066732733571064e-06,
      "loss": 0.0016,
      "step": 208140
    },
    {
      "epoch": 1.3350094441576101,
      "grad_norm": 0.13503798842430115,
      "learning_rate": 3.006159983855794e-06,
      "loss": 0.0019,
      "step": 208150
    },
    {
      "epoch": 1.335073581051396,
      "grad_norm": 0.033364132046699524,
      "learning_rate": 3.0056467193387207e-06,
      "loss": 0.001,
      "step": 208160
    },
    {
      "epoch": 1.3351377179451822,
      "grad_norm": 0.03804779797792435,
      "learning_rate": 3.0051334798123192e-06,
      "loss": 0.002,
      "step": 208170
    },
    {
      "epoch": 1.3352018548389684,
      "grad_norm": 0.07062738388776779,
      "learning_rate": 3.0046202652830226e-06,
      "loss": 0.0014,
      "step": 208180
    },
    {
      "epoch": 1.3352659917327543,
      "grad_norm": 0.046227481216192245,
      "learning_rate": 3.004107075757259e-06,
      "loss": 0.0011,
      "step": 208190
    },
    {
      "epoch": 1.3353301286265404,
      "grad_norm": 0.0878714993596077,
      "learning_rate": 3.0035939112414603e-06,
      "loss": 0.001,
      "step": 208200
    },
    {
      "epoch": 1.3353942655203266,
      "grad_norm": 0.09363577514886856,
      "learning_rate": 3.0030807717420562e-06,
      "loss": 0.0015,
      "step": 208210
    },
    {
      "epoch": 1.3354584024141127,
      "grad_norm": 0.18899093568325043,
      "learning_rate": 3.002567657265479e-06,
      "loss": 0.0019,
      "step": 208220
    },
    {
      "epoch": 1.3355225393078989,
      "grad_norm": 0.00698311859741807,
      "learning_rate": 3.0020545678181547e-06,
      "loss": 0.0012,
      "step": 208230
    },
    {
      "epoch": 1.3355866762016848,
      "grad_norm": 0.1096363216638565,
      "learning_rate": 3.0015415034065155e-06,
      "loss": 0.0022,
      "step": 208240
    },
    {
      "epoch": 1.335650813095471,
      "grad_norm": 0.10652846097946167,
      "learning_rate": 3.001028464036989e-06,
      "loss": 0.0024,
      "step": 208250
    },
    {
      "epoch": 1.335714949989257,
      "grad_norm": 0.05483076721429825,
      "learning_rate": 3.0005154497160054e-06,
      "loss": 0.0016,
      "step": 208260
    },
    {
      "epoch": 1.335779086883043,
      "grad_norm": 0.08351773768663406,
      "learning_rate": 3.000002460449991e-06,
      "loss": 0.0015,
      "step": 208270
    },
    {
      "epoch": 1.3358432237768292,
      "grad_norm": 0.059546101838350296,
      "learning_rate": 2.999489496245377e-06,
      "loss": 0.0011,
      "step": 208280
    },
    {
      "epoch": 1.3359073606706153,
      "grad_norm": 0.1459716260433197,
      "learning_rate": 2.998976557108587e-06,
      "loss": 0.001,
      "step": 208290
    },
    {
      "epoch": 1.3359714975644015,
      "grad_norm": 0.48013800382614136,
      "learning_rate": 2.998463643046053e-06,
      "loss": 0.0024,
      "step": 208300
    },
    {
      "epoch": 1.3360356344581876,
      "grad_norm": 0.06204058229923248,
      "learning_rate": 2.9979507540641985e-06,
      "loss": 0.0022,
      "step": 208310
    },
    {
      "epoch": 1.3360997713519738,
      "grad_norm": 0.0903536006808281,
      "learning_rate": 2.997437890169452e-06,
      "loss": 0.0009,
      "step": 208320
    },
    {
      "epoch": 1.3361639082457597,
      "grad_norm": 0.04781496152281761,
      "learning_rate": 2.9969250513682408e-06,
      "loss": 0.0015,
      "step": 208330
    },
    {
      "epoch": 1.3362280451395459,
      "grad_norm": 0.15592113137245178,
      "learning_rate": 2.996412237666989e-06,
      "loss": 0.0008,
      "step": 208340
    },
    {
      "epoch": 1.336292182033332,
      "grad_norm": 0.3037644624710083,
      "learning_rate": 2.9958994490721257e-06,
      "loss": 0.0019,
      "step": 208350
    },
    {
      "epoch": 1.336356318927118,
      "grad_norm": 0.019961733371019363,
      "learning_rate": 2.9953866855900716e-06,
      "loss": 0.0014,
      "step": 208360
    },
    {
      "epoch": 1.336420455820904,
      "grad_norm": 0.1020020991563797,
      "learning_rate": 2.994873947227257e-06,
      "loss": 0.0011,
      "step": 208370
    },
    {
      "epoch": 1.3364845927146902,
      "grad_norm": 0.10029342025518417,
      "learning_rate": 2.9943612339901052e-06,
      "loss": 0.0013,
      "step": 208380
    },
    {
      "epoch": 1.3365487296084764,
      "grad_norm": 0.04823361337184906,
      "learning_rate": 2.9938485458850396e-06,
      "loss": 0.0016,
      "step": 208390
    },
    {
      "epoch": 1.3366128665022625,
      "grad_norm": 0.053015630692243576,
      "learning_rate": 2.9933358829184867e-06,
      "loss": 0.0019,
      "step": 208400
    },
    {
      "epoch": 1.3366770033960484,
      "grad_norm": 0.0986398383975029,
      "learning_rate": 2.992823245096868e-06,
      "loss": 0.0016,
      "step": 208410
    },
    {
      "epoch": 1.3367411402898346,
      "grad_norm": 0.10989446192979813,
      "learning_rate": 2.99231063242661e-06,
      "loss": 0.0018,
      "step": 208420
    },
    {
      "epoch": 1.3368052771836207,
      "grad_norm": 0.03991841524839401,
      "learning_rate": 2.9917980449141336e-06,
      "loss": 0.001,
      "step": 208430
    },
    {
      "epoch": 1.336869414077407,
      "grad_norm": 0.08873841166496277,
      "learning_rate": 2.991285482565862e-06,
      "loss": 0.0012,
      "step": 208440
    },
    {
      "epoch": 1.3369335509711928,
      "grad_norm": 0.24083659052848816,
      "learning_rate": 2.9907729453882213e-06,
      "loss": 0.0016,
      "step": 208450
    },
    {
      "epoch": 1.336997687864979,
      "grad_norm": 0.13094668090343475,
      "learning_rate": 2.9902604333876293e-06,
      "loss": 0.0016,
      "step": 208460
    },
    {
      "epoch": 1.3370618247587651,
      "grad_norm": 0.02347368746995926,
      "learning_rate": 2.9897479465705127e-06,
      "loss": 0.0009,
      "step": 208470
    },
    {
      "epoch": 1.3371259616525513,
      "grad_norm": 0.20845891535282135,
      "learning_rate": 2.989235484943289e-06,
      "loss": 0.0024,
      "step": 208480
    },
    {
      "epoch": 1.3371900985463374,
      "grad_norm": 0.0839250236749649,
      "learning_rate": 2.9887230485123834e-06,
      "loss": 0.0011,
      "step": 208490
    },
    {
      "epoch": 1.3372542354401233,
      "grad_norm": 0.03127298131585121,
      "learning_rate": 2.9882106372842147e-06,
      "loss": 0.0015,
      "step": 208500
    },
    {
      "epoch": 1.3373183723339095,
      "grad_norm": 0.051121849566698074,
      "learning_rate": 2.9876982512652042e-06,
      "loss": 0.0017,
      "step": 208510
    },
    {
      "epoch": 1.3373825092276956,
      "grad_norm": 0.09151150286197662,
      "learning_rate": 2.987185890461773e-06,
      "loss": 0.0012,
      "step": 208520
    },
    {
      "epoch": 1.3374466461214816,
      "grad_norm": 0.039352428168058395,
      "learning_rate": 2.986673554880342e-06,
      "loss": 0.0022,
      "step": 208530
    },
    {
      "epoch": 1.3375107830152677,
      "grad_norm": 0.056529849767684937,
      "learning_rate": 2.9861612445273287e-06,
      "loss": 0.0009,
      "step": 208540
    },
    {
      "epoch": 1.3375749199090539,
      "grad_norm": 0.02233145758509636,
      "learning_rate": 2.9856489594091555e-06,
      "loss": 0.0016,
      "step": 208550
    },
    {
      "epoch": 1.33763905680284,
      "grad_norm": 0.1725313365459442,
      "learning_rate": 2.9851366995322393e-06,
      "loss": 0.0015,
      "step": 208560
    },
    {
      "epoch": 1.3377031936966262,
      "grad_norm": 0.004792868159711361,
      "learning_rate": 2.984624464903002e-06,
      "loss": 0.0013,
      "step": 208570
    },
    {
      "epoch": 1.3377673305904123,
      "grad_norm": 0.045336589217185974,
      "learning_rate": 2.984112255527859e-06,
      "loss": 0.0008,
      "step": 208580
    },
    {
      "epoch": 1.3378314674841982,
      "grad_norm": 0.08572478592395782,
      "learning_rate": 2.9836000714132297e-06,
      "loss": 0.0009,
      "step": 208590
    },
    {
      "epoch": 1.3378956043779844,
      "grad_norm": 0.041878607124090195,
      "learning_rate": 2.9830879125655343e-06,
      "loss": 0.0008,
      "step": 208600
    },
    {
      "epoch": 1.3379597412717705,
      "grad_norm": 0.03267974033951759,
      "learning_rate": 2.9825757789911874e-06,
      "loss": 0.0011,
      "step": 208610
    },
    {
      "epoch": 1.3380238781655565,
      "grad_norm": 0.006213339976966381,
      "learning_rate": 2.9820636706966087e-06,
      "loss": 0.0017,
      "step": 208620
    },
    {
      "epoch": 1.3380880150593426,
      "grad_norm": 0.09869519621133804,
      "learning_rate": 2.981551587688214e-06,
      "loss": 0.001,
      "step": 208630
    },
    {
      "epoch": 1.3381521519531288,
      "grad_norm": 0.06236009672284126,
      "learning_rate": 2.9810395299724214e-06,
      "loss": 0.0012,
      "step": 208640
    },
    {
      "epoch": 1.338216288846915,
      "grad_norm": 0.10517404228448868,
      "learning_rate": 2.9805274975556453e-06,
      "loss": 0.0013,
      "step": 208650
    },
    {
      "epoch": 1.338280425740701,
      "grad_norm": 0.014546317979693413,
      "learning_rate": 2.9800154904443034e-06,
      "loss": 0.0009,
      "step": 208660
    },
    {
      "epoch": 1.338344562634487,
      "grad_norm": 0.03995361179113388,
      "learning_rate": 2.979503508644811e-06,
      "loss": 0.0008,
      "step": 208670
    },
    {
      "epoch": 1.3384086995282731,
      "grad_norm": 0.07176606357097626,
      "learning_rate": 2.9789915521635837e-06,
      "loss": 0.0007,
      "step": 208680
    },
    {
      "epoch": 1.3384728364220593,
      "grad_norm": 0.03574638441205025,
      "learning_rate": 2.9784796210070368e-06,
      "loss": 0.0016,
      "step": 208690
    },
    {
      "epoch": 1.3385369733158452,
      "grad_norm": 0.057587962597608566,
      "learning_rate": 2.977967715181585e-06,
      "loss": 0.0008,
      "step": 208700
    },
    {
      "epoch": 1.3386011102096314,
      "grad_norm": 0.11445263028144836,
      "learning_rate": 2.977455834693642e-06,
      "loss": 0.0013,
      "step": 208710
    },
    {
      "epoch": 1.3386652471034175,
      "grad_norm": 0.023943578824400902,
      "learning_rate": 2.9769439795496247e-06,
      "loss": 0.002,
      "step": 208720
    },
    {
      "epoch": 1.3387293839972036,
      "grad_norm": 0.2436387985944748,
      "learning_rate": 2.9764321497559435e-06,
      "loss": 0.0028,
      "step": 208730
    },
    {
      "epoch": 1.3387935208909898,
      "grad_norm": 0.033301565796136856,
      "learning_rate": 2.975920345319013e-06,
      "loss": 0.001,
      "step": 208740
    },
    {
      "epoch": 1.338857657784776,
      "grad_norm": 0.09697895497083664,
      "learning_rate": 2.9754085662452494e-06,
      "loss": 0.002,
      "step": 208750
    },
    {
      "epoch": 1.3389217946785619,
      "grad_norm": 0.04288691282272339,
      "learning_rate": 2.9748968125410617e-06,
      "loss": 0.0029,
      "step": 208760
    },
    {
      "epoch": 1.338985931572348,
      "grad_norm": 0.06362093240022659,
      "learning_rate": 2.9743850842128657e-06,
      "loss": 0.002,
      "step": 208770
    },
    {
      "epoch": 1.3390500684661342,
      "grad_norm": 0.04038805514574051,
      "learning_rate": 2.973873381267071e-06,
      "loss": 0.0011,
      "step": 208780
    },
    {
      "epoch": 1.33911420535992,
      "grad_norm": 0.020341023802757263,
      "learning_rate": 2.973361703710092e-06,
      "loss": 0.0009,
      "step": 208790
    },
    {
      "epoch": 1.3391783422537062,
      "grad_norm": 0.03332603722810745,
      "learning_rate": 2.9728500515483383e-06,
      "loss": 0.0011,
      "step": 208800
    },
    {
      "epoch": 1.3392424791474924,
      "grad_norm": 0.06648506224155426,
      "learning_rate": 2.972338424788223e-06,
      "loss": 0.0024,
      "step": 208810
    },
    {
      "epoch": 1.3393066160412785,
      "grad_norm": 0.008569886907935143,
      "learning_rate": 2.971826823436156e-06,
      "loss": 0.0014,
      "step": 208820
    },
    {
      "epoch": 1.3393707529350647,
      "grad_norm": 0.15676802396774292,
      "learning_rate": 2.9713152474985485e-06,
      "loss": 0.0012,
      "step": 208830
    },
    {
      "epoch": 1.3394348898288506,
      "grad_norm": 0.0957457646727562,
      "learning_rate": 2.9708036969818106e-06,
      "loss": 0.0013,
      "step": 208840
    },
    {
      "epoch": 1.3394990267226368,
      "grad_norm": 0.016529573127627373,
      "learning_rate": 2.970292171892353e-06,
      "loss": 0.0012,
      "step": 208850
    },
    {
      "epoch": 1.339563163616423,
      "grad_norm": 0.11083996295928955,
      "learning_rate": 2.9697806722365845e-06,
      "loss": 0.001,
      "step": 208860
    },
    {
      "epoch": 1.339627300510209,
      "grad_norm": 0.14161944389343262,
      "learning_rate": 2.9692691980209153e-06,
      "loss": 0.0116,
      "step": 208870
    },
    {
      "epoch": 1.339691437403995,
      "grad_norm": 0.07772056758403778,
      "learning_rate": 2.968757749251755e-06,
      "loss": 0.0008,
      "step": 208880
    },
    {
      "epoch": 1.3397555742977811,
      "grad_norm": 0.053042441606521606,
      "learning_rate": 2.968246325935511e-06,
      "loss": 0.0013,
      "step": 208890
    },
    {
      "epoch": 1.3398197111915673,
      "grad_norm": 0.027649134397506714,
      "learning_rate": 2.9677349280785937e-06,
      "loss": 0.0013,
      "step": 208900
    },
    {
      "epoch": 1.3398838480853534,
      "grad_norm": 0.029094886034727097,
      "learning_rate": 2.9672235556874085e-06,
      "loss": 0.0012,
      "step": 208910
    },
    {
      "epoch": 1.3399479849791396,
      "grad_norm": 0.0654555931687355,
      "learning_rate": 2.966712208768367e-06,
      "loss": 0.0014,
      "step": 208920
    },
    {
      "epoch": 1.3400121218729255,
      "grad_norm": 0.07652648538351059,
      "learning_rate": 2.966200887327873e-06,
      "loss": 0.0012,
      "step": 208930
    },
    {
      "epoch": 1.3400762587667117,
      "grad_norm": 0.0004798894515261054,
      "learning_rate": 2.9656895913723365e-06,
      "loss": 0.0012,
      "step": 208940
    },
    {
      "epoch": 1.3401403956604978,
      "grad_norm": 0.020167993381619453,
      "learning_rate": 2.965178320908163e-06,
      "loss": 0.0015,
      "step": 208950
    },
    {
      "epoch": 1.3402045325542837,
      "grad_norm": 0.049159470945596695,
      "learning_rate": 2.9646670759417595e-06,
      "loss": 0.0014,
      "step": 208960
    },
    {
      "epoch": 1.3402686694480699,
      "grad_norm": 0.23737210035324097,
      "learning_rate": 2.9641558564795315e-06,
      "loss": 0.0018,
      "step": 208970
    },
    {
      "epoch": 1.340332806341856,
      "grad_norm": 0.11929892003536224,
      "learning_rate": 2.963644662527887e-06,
      "loss": 0.0025,
      "step": 208980
    },
    {
      "epoch": 1.3403969432356422,
      "grad_norm": 0.09051161259412766,
      "learning_rate": 2.963133494093229e-06,
      "loss": 0.0017,
      "step": 208990
    },
    {
      "epoch": 1.3404610801294283,
      "grad_norm": 0.060351449996232986,
      "learning_rate": 2.962622351181964e-06,
      "loss": 0.0019,
      "step": 209000
    },
    {
      "epoch": 1.3405252170232145,
      "grad_norm": 0.05083949491381645,
      "learning_rate": 2.9621112338004978e-06,
      "loss": 0.0019,
      "step": 209010
    },
    {
      "epoch": 1.3405893539170004,
      "grad_norm": 0.11279977858066559,
      "learning_rate": 2.961600141955233e-06,
      "loss": 0.0012,
      "step": 209020
    },
    {
      "epoch": 1.3406534908107866,
      "grad_norm": 0.06853433698415756,
      "learning_rate": 2.961089075652577e-06,
      "loss": 0.0007,
      "step": 209030
    },
    {
      "epoch": 1.3407176277045727,
      "grad_norm": 0.08879048377275467,
      "learning_rate": 2.9605780348989305e-06,
      "loss": 0.0018,
      "step": 209040
    },
    {
      "epoch": 1.3407817645983586,
      "grad_norm": 0.06870577484369278,
      "learning_rate": 2.9600670197006997e-06,
      "loss": 0.001,
      "step": 209050
    },
    {
      "epoch": 1.3408459014921448,
      "grad_norm": 0.13088449835777283,
      "learning_rate": 2.959556030064287e-06,
      "loss": 0.0013,
      "step": 209060
    },
    {
      "epoch": 1.340910038385931,
      "grad_norm": 0.08416897058486938,
      "learning_rate": 2.9590450659960958e-06,
      "loss": 0.0057,
      "step": 209070
    },
    {
      "epoch": 1.340974175279717,
      "grad_norm": 0.05533396080136299,
      "learning_rate": 2.9585341275025277e-06,
      "loss": 0.0021,
      "step": 209080
    },
    {
      "epoch": 1.3410383121735032,
      "grad_norm": 0.019153635948896408,
      "learning_rate": 2.958023214589987e-06,
      "loss": 0.0026,
      "step": 209090
    },
    {
      "epoch": 1.3411024490672891,
      "grad_norm": 0.19014737010002136,
      "learning_rate": 2.957512327264873e-06,
      "loss": 0.0015,
      "step": 209100
    },
    {
      "epoch": 1.3411665859610753,
      "grad_norm": 0.22254988551139832,
      "learning_rate": 2.9570014655335917e-06,
      "loss": 0.0013,
      "step": 209110
    },
    {
      "epoch": 1.3412307228548614,
      "grad_norm": 0.03758307546377182,
      "learning_rate": 2.95649062940254e-06,
      "loss": 0.0011,
      "step": 209120
    },
    {
      "epoch": 1.3412948597486476,
      "grad_norm": 0.0588693767786026,
      "learning_rate": 2.9559798188781208e-06,
      "loss": 0.0017,
      "step": 209130
    },
    {
      "epoch": 1.3413589966424335,
      "grad_norm": 0.05812332406640053,
      "learning_rate": 2.955469033966737e-06,
      "loss": 0.0011,
      "step": 209140
    },
    {
      "epoch": 1.3414231335362197,
      "grad_norm": 0.09899233281612396,
      "learning_rate": 2.9549582746747872e-06,
      "loss": 0.0015,
      "step": 209150
    },
    {
      "epoch": 1.3414872704300058,
      "grad_norm": 0.021945517510175705,
      "learning_rate": 2.9544475410086715e-06,
      "loss": 0.0006,
      "step": 209160
    },
    {
      "epoch": 1.341551407323792,
      "grad_norm": 0.007202756125479937,
      "learning_rate": 2.95393683297479e-06,
      "loss": 0.0024,
      "step": 209170
    },
    {
      "epoch": 1.3416155442175781,
      "grad_norm": 0.007555840536952019,
      "learning_rate": 2.9534261505795426e-06,
      "loss": 0.0006,
      "step": 209180
    },
    {
      "epoch": 1.341679681111364,
      "grad_norm": 0.24433951079845428,
      "learning_rate": 2.952915493829328e-06,
      "loss": 0.003,
      "step": 209190
    },
    {
      "epoch": 1.3417438180051502,
      "grad_norm": 0.04254264757037163,
      "learning_rate": 2.9524048627305455e-06,
      "loss": 0.0013,
      "step": 209200
    },
    {
      "epoch": 1.3418079548989363,
      "grad_norm": 0.2464970499277115,
      "learning_rate": 2.9518942572895937e-06,
      "loss": 0.0035,
      "step": 209210
    },
    {
      "epoch": 1.3418720917927223,
      "grad_norm": 0.1973329335451126,
      "learning_rate": 2.95138367751287e-06,
      "loss": 0.0022,
      "step": 209220
    },
    {
      "epoch": 1.3419362286865084,
      "grad_norm": 0.06293509900569916,
      "learning_rate": 2.950873123406773e-06,
      "loss": 0.0011,
      "step": 209230
    },
    {
      "epoch": 1.3420003655802946,
      "grad_norm": 0.03426901251077652,
      "learning_rate": 2.9503625949777003e-06,
      "loss": 0.0023,
      "step": 209240
    },
    {
      "epoch": 1.3420645024740807,
      "grad_norm": 0.22345155477523804,
      "learning_rate": 2.94985209223205e-06,
      "loss": 0.0006,
      "step": 209250
    },
    {
      "epoch": 1.3421286393678669,
      "grad_norm": 0.011404551565647125,
      "learning_rate": 2.9493416151762173e-06,
      "loss": 0.0012,
      "step": 209260
    },
    {
      "epoch": 1.342192776261653,
      "grad_norm": 0.06039094179868698,
      "learning_rate": 2.9488311638166e-06,
      "loss": 0.0012,
      "step": 209270
    },
    {
      "epoch": 1.342256913155439,
      "grad_norm": 0.024154851213097572,
      "learning_rate": 2.948320738159594e-06,
      "loss": 0.0017,
      "step": 209280
    },
    {
      "epoch": 1.342321050049225,
      "grad_norm": 0.13007420301437378,
      "learning_rate": 2.947810338211595e-06,
      "loss": 0.0019,
      "step": 209290
    },
    {
      "epoch": 1.3423851869430112,
      "grad_norm": 0.1291894167661667,
      "learning_rate": 2.947299963979e-06,
      "loss": 0.0016,
      "step": 209300
    },
    {
      "epoch": 1.3424493238367972,
      "grad_norm": 0.024135563522577286,
      "learning_rate": 2.946789615468203e-06,
      "loss": 0.0012,
      "step": 209310
    },
    {
      "epoch": 1.3425134607305833,
      "grad_norm": 0.07220418006181717,
      "learning_rate": 2.9462792926856002e-06,
      "loss": 0.0017,
      "step": 209320
    },
    {
      "epoch": 1.3425775976243695,
      "grad_norm": 0.08306324481964111,
      "learning_rate": 2.9457689956375847e-06,
      "loss": 0.0009,
      "step": 209330
    },
    {
      "epoch": 1.3426417345181556,
      "grad_norm": 0.15459685027599335,
      "learning_rate": 2.945258724330553e-06,
      "loss": 0.002,
      "step": 209340
    },
    {
      "epoch": 1.3427058714119418,
      "grad_norm": 0.06739833205938339,
      "learning_rate": 2.944748478770897e-06,
      "loss": 0.0017,
      "step": 209350
    },
    {
      "epoch": 1.3427700083057277,
      "grad_norm": 0.03368918597698212,
      "learning_rate": 2.944238258965012e-06,
      "loss": 0.0012,
      "step": 209360
    },
    {
      "epoch": 1.3428341451995138,
      "grad_norm": 0.1809409111738205,
      "learning_rate": 2.94372806491929e-06,
      "loss": 0.002,
      "step": 209370
    },
    {
      "epoch": 1.3428982820933,
      "grad_norm": 0.10510099679231644,
      "learning_rate": 2.943217896640126e-06,
      "loss": 0.0014,
      "step": 209380
    },
    {
      "epoch": 1.342962418987086,
      "grad_norm": 0.1288333386182785,
      "learning_rate": 2.942707754133911e-06,
      "loss": 0.0043,
      "step": 209390
    },
    {
      "epoch": 1.343026555880872,
      "grad_norm": 0.1137746199965477,
      "learning_rate": 2.9421976374070393e-06,
      "loss": 0.0015,
      "step": 209400
    },
    {
      "epoch": 1.3430906927746582,
      "grad_norm": 0.009048394858837128,
      "learning_rate": 2.9416875464659e-06,
      "loss": 0.0012,
      "step": 209410
    },
    {
      "epoch": 1.3431548296684443,
      "grad_norm": 0.1014087125658989,
      "learning_rate": 2.941177481316888e-06,
      "loss": 0.0009,
      "step": 209420
    },
    {
      "epoch": 1.3432189665622305,
      "grad_norm": 0.07595310360193253,
      "learning_rate": 2.9406674419663935e-06,
      "loss": 0.0014,
      "step": 209430
    },
    {
      "epoch": 1.3432831034560166,
      "grad_norm": 0.164398655295372,
      "learning_rate": 2.9401574284208068e-06,
      "loss": 0.0041,
      "step": 209440
    },
    {
      "epoch": 1.3433472403498026,
      "grad_norm": 0.10600937157869339,
      "learning_rate": 2.939647440686521e-06,
      "loss": 0.0013,
      "step": 209450
    },
    {
      "epoch": 1.3434113772435887,
      "grad_norm": 0.23413997888565063,
      "learning_rate": 2.9391374787699243e-06,
      "loss": 0.0017,
      "step": 209460
    },
    {
      "epoch": 1.3434755141373749,
      "grad_norm": 0.04459110274910927,
      "learning_rate": 2.938627542677409e-06,
      "loss": 0.0013,
      "step": 209470
    },
    {
      "epoch": 1.3435396510311608,
      "grad_norm": 0.08799204230308533,
      "learning_rate": 2.938117632415363e-06,
      "loss": 0.0012,
      "step": 209480
    },
    {
      "epoch": 1.343603787924947,
      "grad_norm": 0.09129440039396286,
      "learning_rate": 2.9376077479901767e-06,
      "loss": 0.0017,
      "step": 209490
    },
    {
      "epoch": 1.343667924818733,
      "grad_norm": 0.04995739459991455,
      "learning_rate": 2.937097889408239e-06,
      "loss": 0.0012,
      "step": 209500
    },
    {
      "epoch": 1.3437320617125192,
      "grad_norm": 0.07876341789960861,
      "learning_rate": 2.93658805667594e-06,
      "loss": 0.0034,
      "step": 209510
    },
    {
      "epoch": 1.3437961986063054,
      "grad_norm": 0.06519567221403122,
      "learning_rate": 2.9360782497996664e-06,
      "loss": 0.0013,
      "step": 209520
    },
    {
      "epoch": 1.3438603355000913,
      "grad_norm": 0.07680631428956985,
      "learning_rate": 2.935568468785809e-06,
      "loss": 0.0018,
      "step": 209530
    },
    {
      "epoch": 1.3439244723938775,
      "grad_norm": 0.1277865618467331,
      "learning_rate": 2.9350587136407517e-06,
      "loss": 0.0009,
      "step": 209540
    },
    {
      "epoch": 1.3439886092876636,
      "grad_norm": 0.04611477628350258,
      "learning_rate": 2.9345489843708863e-06,
      "loss": 0.0009,
      "step": 209550
    },
    {
      "epoch": 1.3440527461814498,
      "grad_norm": 0.1415407657623291,
      "learning_rate": 2.9340392809825976e-06,
      "loss": 0.0013,
      "step": 209560
    },
    {
      "epoch": 1.3441168830752357,
      "grad_norm": 0.14277635514736176,
      "learning_rate": 2.9335296034822737e-06,
      "loss": 0.0012,
      "step": 209570
    },
    {
      "epoch": 1.3441810199690218,
      "grad_norm": 0.1470096856355667,
      "learning_rate": 2.9330199518763002e-06,
      "loss": 0.001,
      "step": 209580
    },
    {
      "epoch": 1.344245156862808,
      "grad_norm": 0.06251547485589981,
      "learning_rate": 2.932510326171063e-06,
      "loss": 0.0011,
      "step": 209590
    },
    {
      "epoch": 1.3443092937565941,
      "grad_norm": 0.14789541065692902,
      "learning_rate": 2.932000726372951e-06,
      "loss": 0.0018,
      "step": 209600
    },
    {
      "epoch": 1.3443734306503803,
      "grad_norm": 0.03225419297814369,
      "learning_rate": 2.9314911524883466e-06,
      "loss": 0.0011,
      "step": 209610
    },
    {
      "epoch": 1.3444375675441662,
      "grad_norm": 0.11897369474172592,
      "learning_rate": 2.930981604523637e-06,
      "loss": 0.0019,
      "step": 209620
    },
    {
      "epoch": 1.3445017044379524,
      "grad_norm": 0.062377601861953735,
      "learning_rate": 2.930472082485206e-06,
      "loss": 0.0013,
      "step": 209630
    },
    {
      "epoch": 1.3445658413317385,
      "grad_norm": 0.10768963396549225,
      "learning_rate": 2.92996258637944e-06,
      "loss": 0.002,
      "step": 209640
    },
    {
      "epoch": 1.3446299782255244,
      "grad_norm": 0.13047385215759277,
      "learning_rate": 2.9294531162127216e-06,
      "loss": 0.0014,
      "step": 209650
    },
    {
      "epoch": 1.3446941151193106,
      "grad_norm": 0.03815917298197746,
      "learning_rate": 2.9289436719914353e-06,
      "loss": 0.0013,
      "step": 209660
    },
    {
      "epoch": 1.3447582520130967,
      "grad_norm": 0.392093688249588,
      "learning_rate": 2.928434253721965e-06,
      "loss": 0.0023,
      "step": 209670
    },
    {
      "epoch": 1.3448223889068829,
      "grad_norm": 0.14308254420757294,
      "learning_rate": 2.9279248614106948e-06,
      "loss": 0.0044,
      "step": 209680
    },
    {
      "epoch": 1.344886525800669,
      "grad_norm": 0.09599711000919342,
      "learning_rate": 2.9274154950640053e-06,
      "loss": 0.0013,
      "step": 209690
    },
    {
      "epoch": 1.3449506626944552,
      "grad_norm": 0.09055355936288834,
      "learning_rate": 2.926906154688283e-06,
      "loss": 0.0016,
      "step": 209700
    },
    {
      "epoch": 1.345014799588241,
      "grad_norm": 0.09291297197341919,
      "learning_rate": 2.9263968402899064e-06,
      "loss": 0.0013,
      "step": 209710
    },
    {
      "epoch": 1.3450789364820273,
      "grad_norm": 0.08032003790140152,
      "learning_rate": 2.9258875518752595e-06,
      "loss": 0.0018,
      "step": 209720
    },
    {
      "epoch": 1.3451430733758134,
      "grad_norm": 0.007770441472530365,
      "learning_rate": 2.9253782894507248e-06,
      "loss": 0.0015,
      "step": 209730
    },
    {
      "epoch": 1.3452072102695993,
      "grad_norm": 0.02273477427661419,
      "learning_rate": 2.924869053022682e-06,
      "loss": 0.0012,
      "step": 209740
    },
    {
      "epoch": 1.3452713471633855,
      "grad_norm": 0.07384491711854935,
      "learning_rate": 2.924359842597514e-06,
      "loss": 0.0021,
      "step": 209750
    },
    {
      "epoch": 1.3453354840571716,
      "grad_norm": 0.03380212560296059,
      "learning_rate": 2.9238506581815997e-06,
      "loss": 0.0066,
      "step": 209760
    },
    {
      "epoch": 1.3453996209509578,
      "grad_norm": 0.009146971628069878,
      "learning_rate": 2.9233414997813213e-06,
      "loss": 0.0029,
      "step": 209770
    },
    {
      "epoch": 1.345463757844744,
      "grad_norm": 0.07387570291757584,
      "learning_rate": 2.922832367403058e-06,
      "loss": 0.0018,
      "step": 209780
    },
    {
      "epoch": 1.3455278947385299,
      "grad_norm": 0.039091628044843674,
      "learning_rate": 2.9223232610531894e-06,
      "loss": 0.001,
      "step": 209790
    },
    {
      "epoch": 1.345592031632316,
      "grad_norm": 0.09510093182325363,
      "learning_rate": 2.9218141807380947e-06,
      "loss": 0.0022,
      "step": 209800
    },
    {
      "epoch": 1.3456561685261021,
      "grad_norm": 0.07570884376764297,
      "learning_rate": 2.9213051264641546e-06,
      "loss": 0.0013,
      "step": 209810
    },
    {
      "epoch": 1.345720305419888,
      "grad_norm": 0.026463299989700317,
      "learning_rate": 2.9207960982377457e-06,
      "loss": 0.0022,
      "step": 209820
    },
    {
      "epoch": 1.3457844423136742,
      "grad_norm": 0.05057981610298157,
      "learning_rate": 2.9202870960652486e-06,
      "loss": 0.0022,
      "step": 209830
    },
    {
      "epoch": 1.3458485792074604,
      "grad_norm": 0.1473167985677719,
      "learning_rate": 2.91977811995304e-06,
      "loss": 0.0009,
      "step": 209840
    },
    {
      "epoch": 1.3459127161012465,
      "grad_norm": 0.05318130925297737,
      "learning_rate": 2.9192691699074993e-06,
      "loss": 0.0006,
      "step": 209850
    },
    {
      "epoch": 1.3459768529950327,
      "grad_norm": 0.09713903814554214,
      "learning_rate": 2.9187602459350016e-06,
      "loss": 0.0018,
      "step": 209860
    },
    {
      "epoch": 1.3460409898888188,
      "grad_norm": 0.039324406534433365,
      "learning_rate": 2.9182513480419262e-06,
      "loss": 0.001,
      "step": 209870
    },
    {
      "epoch": 1.3461051267826047,
      "grad_norm": 0.049848772585392,
      "learning_rate": 2.917742476234648e-06,
      "loss": 0.0017,
      "step": 209880
    },
    {
      "epoch": 1.346169263676391,
      "grad_norm": 0.1317090094089508,
      "learning_rate": 2.917233630519544e-06,
      "loss": 0.0012,
      "step": 209890
    },
    {
      "epoch": 1.346233400570177,
      "grad_norm": 0.17113488912582397,
      "learning_rate": 2.9167248109029934e-06,
      "loss": 0.0017,
      "step": 209900
    },
    {
      "epoch": 1.346297537463963,
      "grad_norm": 0.030517883598804474,
      "learning_rate": 2.91621601739137e-06,
      "loss": 0.0018,
      "step": 209910
    },
    {
      "epoch": 1.3463616743577491,
      "grad_norm": 0.10433640331029892,
      "learning_rate": 2.9157072499910487e-06,
      "loss": 0.0017,
      "step": 209920
    },
    {
      "epoch": 1.3464258112515353,
      "grad_norm": 0.045640259981155396,
      "learning_rate": 2.915198508708404e-06,
      "loss": 0.001,
      "step": 209930
    },
    {
      "epoch": 1.3464899481453214,
      "grad_norm": 0.030804645270109177,
      "learning_rate": 2.9146897935498133e-06,
      "loss": 0.0013,
      "step": 209940
    },
    {
      "epoch": 1.3465540850391076,
      "grad_norm": 0.1606251299381256,
      "learning_rate": 2.91418110452165e-06,
      "loss": 0.0015,
      "step": 209950
    },
    {
      "epoch": 1.3466182219328935,
      "grad_norm": 0.006368404719978571,
      "learning_rate": 2.9136724416302887e-06,
      "loss": 0.0013,
      "step": 209960
    },
    {
      "epoch": 1.3466823588266796,
      "grad_norm": 0.00760697154328227,
      "learning_rate": 2.9131638048821e-06,
      "loss": 0.0006,
      "step": 209970
    },
    {
      "epoch": 1.3467464957204658,
      "grad_norm": 0.052641209214925766,
      "learning_rate": 2.9126551942834625e-06,
      "loss": 0.001,
      "step": 209980
    },
    {
      "epoch": 1.346810632614252,
      "grad_norm": 0.011346855200827122,
      "learning_rate": 2.912146609840747e-06,
      "loss": 0.0015,
      "step": 209990
    },
    {
      "epoch": 1.3468747695080379,
      "grad_norm": 0.1732601821422577,
      "learning_rate": 2.911638051560325e-06,
      "loss": 0.0035,
      "step": 210000
    },
    {
      "epoch": 1.346938906401824,
      "grad_norm": 0.029262270778417587,
      "learning_rate": 2.911129519448573e-06,
      "loss": 0.0012,
      "step": 210010
    },
    {
      "epoch": 1.3470030432956102,
      "grad_norm": 0.10278654098510742,
      "learning_rate": 2.9106210135118608e-06,
      "loss": 0.0011,
      "step": 210020
    },
    {
      "epoch": 1.3470671801893963,
      "grad_norm": 0.08118873089551926,
      "learning_rate": 2.91011253375656e-06,
      "loss": 0.0006,
      "step": 210030
    },
    {
      "epoch": 1.3471313170831825,
      "grad_norm": 0.1753062754869461,
      "learning_rate": 2.9096040801890413e-06,
      "loss": 0.002,
      "step": 210040
    },
    {
      "epoch": 1.3471954539769684,
      "grad_norm": 0.09469661861658096,
      "learning_rate": 2.9090956528156793e-06,
      "loss": 0.0015,
      "step": 210050
    },
    {
      "epoch": 1.3472595908707545,
      "grad_norm": 0.04578879848122597,
      "learning_rate": 2.908587251642843e-06,
      "loss": 0.0012,
      "step": 210060
    },
    {
      "epoch": 1.3473237277645407,
      "grad_norm": 0.12490465492010117,
      "learning_rate": 2.9080788766769036e-06,
      "loss": 0.0017,
      "step": 210070
    },
    {
      "epoch": 1.3473878646583266,
      "grad_norm": 0.12824605405330658,
      "learning_rate": 2.9075705279242284e-06,
      "loss": 0.0012,
      "step": 210080
    },
    {
      "epoch": 1.3474520015521128,
      "grad_norm": 0.06864582747220993,
      "learning_rate": 2.9070622053911925e-06,
      "loss": 0.0011,
      "step": 210090
    },
    {
      "epoch": 1.347516138445899,
      "grad_norm": 0.182582825422287,
      "learning_rate": 2.9065539090841623e-06,
      "loss": 0.0019,
      "step": 210100
    },
    {
      "epoch": 1.347580275339685,
      "grad_norm": 0.006342856679111719,
      "learning_rate": 2.9060456390095082e-06,
      "loss": 0.0019,
      "step": 210110
    },
    {
      "epoch": 1.3476444122334712,
      "grad_norm": 0.020330281928181648,
      "learning_rate": 2.9055373951735966e-06,
      "loss": 0.0017,
      "step": 210120
    },
    {
      "epoch": 1.3477085491272573,
      "grad_norm": 0.14057806134223938,
      "learning_rate": 2.905029177582801e-06,
      "loss": 0.0011,
      "step": 210130
    },
    {
      "epoch": 1.3477726860210433,
      "grad_norm": 0.0436885803937912,
      "learning_rate": 2.9045209862434864e-06,
      "loss": 0.0006,
      "step": 210140
    },
    {
      "epoch": 1.3478368229148294,
      "grad_norm": 0.459991991519928,
      "learning_rate": 2.9040128211620195e-06,
      "loss": 0.0012,
      "step": 210150
    },
    {
      "epoch": 1.3479009598086156,
      "grad_norm": 0.09436246752738953,
      "learning_rate": 2.903504682344772e-06,
      "loss": 0.0013,
      "step": 210160
    },
    {
      "epoch": 1.3479650967024015,
      "grad_norm": 0.13223658502101898,
      "learning_rate": 2.902996569798109e-06,
      "loss": 0.0033,
      "step": 210170
    },
    {
      "epoch": 1.3480292335961876,
      "grad_norm": 0.13869208097457886,
      "learning_rate": 2.9024884835283978e-06,
      "loss": 0.0013,
      "step": 210180
    },
    {
      "epoch": 1.3480933704899738,
      "grad_norm": 0.1832425892353058,
      "learning_rate": 2.901980423542003e-06,
      "loss": 0.0014,
      "step": 210190
    },
    {
      "epoch": 1.34815750738376,
      "grad_norm": 0.03327884525060654,
      "learning_rate": 2.9014723898452946e-06,
      "loss": 0.0013,
      "step": 210200
    },
    {
      "epoch": 1.348221644277546,
      "grad_norm": 0.003701785346493125,
      "learning_rate": 2.9009643824446377e-06,
      "loss": 0.0029,
      "step": 210210
    },
    {
      "epoch": 1.348285781171332,
      "grad_norm": 0.08859538286924362,
      "learning_rate": 2.9004564013463964e-06,
      "loss": 0.0012,
      "step": 210220
    },
    {
      "epoch": 1.3483499180651182,
      "grad_norm": 0.00830372329801321,
      "learning_rate": 2.899948446556936e-06,
      "loss": 0.0007,
      "step": 210230
    },
    {
      "epoch": 1.3484140549589043,
      "grad_norm": 0.044589921832084656,
      "learning_rate": 2.8994405180826234e-06,
      "loss": 0.0013,
      "step": 210240
    },
    {
      "epoch": 1.3484781918526902,
      "grad_norm": 0.07862818986177444,
      "learning_rate": 2.8989326159298225e-06,
      "loss": 0.002,
      "step": 210250
    },
    {
      "epoch": 1.3485423287464764,
      "grad_norm": 0.06766141206026077,
      "learning_rate": 2.8984247401048983e-06,
      "loss": 0.0015,
      "step": 210260
    },
    {
      "epoch": 1.3486064656402625,
      "grad_norm": 0.06975769996643066,
      "learning_rate": 2.8979168906142114e-06,
      "loss": 0.0018,
      "step": 210270
    },
    {
      "epoch": 1.3486706025340487,
      "grad_norm": 0.08193672448396683,
      "learning_rate": 2.8974090674641304e-06,
      "loss": 0.0022,
      "step": 210280
    },
    {
      "epoch": 1.3487347394278348,
      "grad_norm": 0.10991158336400986,
      "learning_rate": 2.896901270661017e-06,
      "loss": 0.001,
      "step": 210290
    },
    {
      "epoch": 1.348798876321621,
      "grad_norm": 0.08881914615631104,
      "learning_rate": 2.896393500211231e-06,
      "loss": 0.0021,
      "step": 210300
    },
    {
      "epoch": 1.348863013215407,
      "grad_norm": 0.06696899235248566,
      "learning_rate": 2.8958857561211394e-06,
      "loss": 0.0018,
      "step": 210310
    },
    {
      "epoch": 1.348927150109193,
      "grad_norm": 0.04735543578863144,
      "learning_rate": 2.8953780383971043e-06,
      "loss": 0.0013,
      "step": 210320
    },
    {
      "epoch": 1.3489912870029792,
      "grad_norm": 0.15906420350074768,
      "learning_rate": 2.8948703470454854e-06,
      "loss": 0.0017,
      "step": 210330
    },
    {
      "epoch": 1.3490554238967651,
      "grad_norm": 0.040080178529024124,
      "learning_rate": 2.8943626820726446e-06,
      "loss": 0.0009,
      "step": 210340
    },
    {
      "epoch": 1.3491195607905513,
      "grad_norm": 0.055117517709732056,
      "learning_rate": 2.8938550434849454e-06,
      "loss": 0.0015,
      "step": 210350
    },
    {
      "epoch": 1.3491836976843374,
      "grad_norm": 0.2058384120464325,
      "learning_rate": 2.8933474312887477e-06,
      "loss": 0.0044,
      "step": 210360
    },
    {
      "epoch": 1.3492478345781236,
      "grad_norm": 0.13802611827850342,
      "learning_rate": 2.8928398454904127e-06,
      "loss": 0.0016,
      "step": 210370
    },
    {
      "epoch": 1.3493119714719097,
      "grad_norm": 0.15008658170700073,
      "learning_rate": 2.8923322860962986e-06,
      "loss": 0.0014,
      "step": 210380
    },
    {
      "epoch": 1.3493761083656957,
      "grad_norm": 0.03854681923985481,
      "learning_rate": 2.8918247531127696e-06,
      "loss": 0.001,
      "step": 210390
    },
    {
      "epoch": 1.3494402452594818,
      "grad_norm": 0.0791763886809349,
      "learning_rate": 2.8913172465461824e-06,
      "loss": 0.001,
      "step": 210400
    },
    {
      "epoch": 1.349504382153268,
      "grad_norm": 0.1294926255941391,
      "learning_rate": 2.890809766402897e-06,
      "loss": 0.0016,
      "step": 210410
    },
    {
      "epoch": 1.349568519047054,
      "grad_norm": 0.06261022388935089,
      "learning_rate": 2.8903023126892714e-06,
      "loss": 0.0011,
      "step": 210420
    },
    {
      "epoch": 1.34963265594084,
      "grad_norm": 0.06749068945646286,
      "learning_rate": 2.8897948854116676e-06,
      "loss": 0.0016,
      "step": 210430
    },
    {
      "epoch": 1.3496967928346262,
      "grad_norm": 0.24904891848564148,
      "learning_rate": 2.8892874845764414e-06,
      "loss": 0.0017,
      "step": 210440
    },
    {
      "epoch": 1.3497609297284123,
      "grad_norm": 0.07309553027153015,
      "learning_rate": 2.8887801101899495e-06,
      "loss": 0.0018,
      "step": 210450
    },
    {
      "epoch": 1.3498250666221985,
      "grad_norm": 0.08268938958644867,
      "learning_rate": 2.888272762258554e-06,
      "loss": 0.0007,
      "step": 210460
    },
    {
      "epoch": 1.3498892035159846,
      "grad_norm": 0.11225002259016037,
      "learning_rate": 2.8877654407886102e-06,
      "loss": 0.0026,
      "step": 210470
    },
    {
      "epoch": 1.3499533404097706,
      "grad_norm": 0.12459053844213486,
      "learning_rate": 2.8872581457864747e-06,
      "loss": 0.0027,
      "step": 210480
    },
    {
      "epoch": 1.3500174773035567,
      "grad_norm": 0.09756121784448624,
      "learning_rate": 2.886750877258503e-06,
      "loss": 0.0016,
      "step": 210490
    },
    {
      "epoch": 1.3500816141973428,
      "grad_norm": 0.05711054429411888,
      "learning_rate": 2.8862436352110543e-06,
      "loss": 0.0014,
      "step": 210500
    },
    {
      "epoch": 1.3501457510911288,
      "grad_norm": 0.2802426218986511,
      "learning_rate": 2.8857364196504843e-06,
      "loss": 0.0022,
      "step": 210510
    },
    {
      "epoch": 1.350209887984915,
      "grad_norm": 0.01650386117398739,
      "learning_rate": 2.8852292305831473e-06,
      "loss": 0.0007,
      "step": 210520
    },
    {
      "epoch": 1.350274024878701,
      "grad_norm": 0.03775866702198982,
      "learning_rate": 2.8847220680153983e-06,
      "loss": 0.0008,
      "step": 210530
    },
    {
      "epoch": 1.3503381617724872,
      "grad_norm": 0.07077381014823914,
      "learning_rate": 2.884214931953595e-06,
      "loss": 0.0015,
      "step": 210540
    },
    {
      "epoch": 1.3504022986662734,
      "grad_norm": 0.0790569856762886,
      "learning_rate": 2.8837078224040904e-06,
      "loss": 0.0021,
      "step": 210550
    },
    {
      "epoch": 1.3504664355600595,
      "grad_norm": 0.08557630330324173,
      "learning_rate": 2.8832007393732404e-06,
      "loss": 0.0006,
      "step": 210560
    },
    {
      "epoch": 1.3505305724538454,
      "grad_norm": 0.034811779856681824,
      "learning_rate": 2.8826936828673953e-06,
      "loss": 0.0011,
      "step": 210570
    },
    {
      "epoch": 1.3505947093476316,
      "grad_norm": 0.01789713464677334,
      "learning_rate": 2.8821866528929133e-06,
      "loss": 0.0011,
      "step": 210580
    },
    {
      "epoch": 1.3506588462414177,
      "grad_norm": 0.11298376321792603,
      "learning_rate": 2.8816796494561468e-06,
      "loss": 0.0009,
      "step": 210590
    },
    {
      "epoch": 1.3507229831352037,
      "grad_norm": 0.0268674548715353,
      "learning_rate": 2.881172672563446e-06,
      "loss": 0.0038,
      "step": 210600
    },
    {
      "epoch": 1.3507871200289898,
      "grad_norm": 0.07442095130681992,
      "learning_rate": 2.8806657222211677e-06,
      "loss": 0.001,
      "step": 210610
    },
    {
      "epoch": 1.350851256922776,
      "grad_norm": 0.05544663220643997,
      "learning_rate": 2.8801587984356623e-06,
      "loss": 0.0011,
      "step": 210620
    },
    {
      "epoch": 1.3509153938165621,
      "grad_norm": 0.03473219275474548,
      "learning_rate": 2.879651901213283e-06,
      "loss": 0.0005,
      "step": 210630
    },
    {
      "epoch": 1.3509795307103483,
      "grad_norm": 0.039361413568258286,
      "learning_rate": 2.8791450305603773e-06,
      "loss": 0.0016,
      "step": 210640
    },
    {
      "epoch": 1.3510436676041342,
      "grad_norm": 0.0213471706956625,
      "learning_rate": 2.8786381864833014e-06,
      "loss": 0.0008,
      "step": 210650
    },
    {
      "epoch": 1.3511078044979203,
      "grad_norm": 0.03584079071879387,
      "learning_rate": 2.8781313689884073e-06,
      "loss": 0.002,
      "step": 210660
    },
    {
      "epoch": 1.3511719413917065,
      "grad_norm": 0.08981915563344955,
      "learning_rate": 2.877624578082043e-06,
      "loss": 0.0016,
      "step": 210670
    },
    {
      "epoch": 1.3512360782854926,
      "grad_norm": 0.08490435034036636,
      "learning_rate": 2.8771178137705595e-06,
      "loss": 0.0018,
      "step": 210680
    },
    {
      "epoch": 1.3513002151792786,
      "grad_norm": 0.07120810449123383,
      "learning_rate": 2.876611076060305e-06,
      "loss": 0.002,
      "step": 210690
    },
    {
      "epoch": 1.3513643520730647,
      "grad_norm": 0.04556477442383766,
      "learning_rate": 2.876104364957634e-06,
      "loss": 0.0021,
      "step": 210700
    },
    {
      "epoch": 1.3514284889668509,
      "grad_norm": 0.0411594994366169,
      "learning_rate": 2.875597680468892e-06,
      "loss": 0.0014,
      "step": 210710
    },
    {
      "epoch": 1.351492625860637,
      "grad_norm": 0.13774462044239044,
      "learning_rate": 2.8750910226004287e-06,
      "loss": 0.002,
      "step": 210720
    },
    {
      "epoch": 1.3515567627544232,
      "grad_norm": 0.2863829731941223,
      "learning_rate": 2.8745843913585946e-06,
      "loss": 0.0016,
      "step": 210730
    },
    {
      "epoch": 1.351620899648209,
      "grad_norm": 0.02558436617255211,
      "learning_rate": 2.874077786749738e-06,
      "loss": 0.0011,
      "step": 210740
    },
    {
      "epoch": 1.3516850365419952,
      "grad_norm": 0.10683951526880264,
      "learning_rate": 2.8735712087802055e-06,
      "loss": 0.0038,
      "step": 210750
    },
    {
      "epoch": 1.3517491734357814,
      "grad_norm": 0.07721851021051407,
      "learning_rate": 2.8730646574563437e-06,
      "loss": 0.0022,
      "step": 210760
    },
    {
      "epoch": 1.3518133103295673,
      "grad_norm": 0.024722186848521233,
      "learning_rate": 2.8725581327845033e-06,
      "loss": 0.002,
      "step": 210770
    },
    {
      "epoch": 1.3518774472233535,
      "grad_norm": 0.2504269480705261,
      "learning_rate": 2.8720516347710305e-06,
      "loss": 0.0026,
      "step": 210780
    },
    {
      "epoch": 1.3519415841171396,
      "grad_norm": 0.059513602405786514,
      "learning_rate": 2.8715451634222713e-06,
      "loss": 0.0016,
      "step": 210790
    },
    {
      "epoch": 1.3520057210109258,
      "grad_norm": 0.04534576088190079,
      "learning_rate": 2.8710387187445697e-06,
      "loss": 0.002,
      "step": 210800
    },
    {
      "epoch": 1.352069857904712,
      "grad_norm": 0.005066273733973503,
      "learning_rate": 2.8705323007442774e-06,
      "loss": 0.0007,
      "step": 210810
    },
    {
      "epoch": 1.352133994798498,
      "grad_norm": 0.04880732670426369,
      "learning_rate": 2.870025909427737e-06,
      "loss": 0.0007,
      "step": 210820
    },
    {
      "epoch": 1.352198131692284,
      "grad_norm": 0.10243187099695206,
      "learning_rate": 2.869519544801295e-06,
      "loss": 0.0015,
      "step": 210830
    },
    {
      "epoch": 1.3522622685860701,
      "grad_norm": 0.08573068678379059,
      "learning_rate": 2.8690132068712926e-06,
      "loss": 0.0012,
      "step": 210840
    },
    {
      "epoch": 1.3523264054798563,
      "grad_norm": 0.15311363339424133,
      "learning_rate": 2.8685068956440803e-06,
      "loss": 0.0012,
      "step": 210850
    },
    {
      "epoch": 1.3523905423736422,
      "grad_norm": 0.2949298918247223,
      "learning_rate": 2.868000611126e-06,
      "loss": 0.0019,
      "step": 210860
    },
    {
      "epoch": 1.3524546792674284,
      "grad_norm": 0.02028181403875351,
      "learning_rate": 2.8674943533233936e-06,
      "loss": 0.0014,
      "step": 210870
    },
    {
      "epoch": 1.3525188161612145,
      "grad_norm": 0.05514419823884964,
      "learning_rate": 2.8669881222426086e-06,
      "loss": 0.0019,
      "step": 210880
    },
    {
      "epoch": 1.3525829530550006,
      "grad_norm": 0.03877931460738182,
      "learning_rate": 2.8664819178899876e-06,
      "loss": 0.0011,
      "step": 210890
    },
    {
      "epoch": 1.3526470899487868,
      "grad_norm": 0.04640922695398331,
      "learning_rate": 2.8659757402718724e-06,
      "loss": 0.0021,
      "step": 210900
    },
    {
      "epoch": 1.3527112268425727,
      "grad_norm": 0.008041328750550747,
      "learning_rate": 2.865469589394605e-06,
      "loss": 0.0009,
      "step": 210910
    },
    {
      "epoch": 1.3527753637363589,
      "grad_norm": 0.0224269088357687,
      "learning_rate": 2.864963465264531e-06,
      "loss": 0.0008,
      "step": 210920
    },
    {
      "epoch": 1.352839500630145,
      "grad_norm": 0.07984951883554459,
      "learning_rate": 2.8644573678879907e-06,
      "loss": 0.0008,
      "step": 210930
    },
    {
      "epoch": 1.352903637523931,
      "grad_norm": 0.00870156567543745,
      "learning_rate": 2.8639512972713253e-06,
      "loss": 0.0021,
      "step": 210940
    },
    {
      "epoch": 1.352967774417717,
      "grad_norm": 0.029904576018452644,
      "learning_rate": 2.8634452534208747e-06,
      "loss": 0.0007,
      "step": 210950
    },
    {
      "epoch": 1.3530319113115032,
      "grad_norm": 0.01964622177183628,
      "learning_rate": 2.862939236342984e-06,
      "loss": 0.002,
      "step": 210960
    },
    {
      "epoch": 1.3530960482052894,
      "grad_norm": 0.05954081937670708,
      "learning_rate": 2.8624332460439928e-06,
      "loss": 0.0007,
      "step": 210970
    },
    {
      "epoch": 1.3531601850990755,
      "grad_norm": 0.07545377314090729,
      "learning_rate": 2.8619272825302402e-06,
      "loss": 0.0011,
      "step": 210980
    },
    {
      "epoch": 1.3532243219928617,
      "grad_norm": 0.07013353705406189,
      "learning_rate": 2.8614213458080653e-06,
      "loss": 0.0017,
      "step": 210990
    },
    {
      "epoch": 1.3532884588866476,
      "grad_norm": 0.011843344196677208,
      "learning_rate": 2.860915435883811e-06,
      "loss": 0.0005,
      "step": 211000
    },
    {
      "epoch": 1.3533525957804338,
      "grad_norm": 0.07124199718236923,
      "learning_rate": 2.860409552763815e-06,
      "loss": 0.0011,
      "step": 211010
    },
    {
      "epoch": 1.35341673267422,
      "grad_norm": 0.19323082268238068,
      "learning_rate": 2.8599036964544147e-06,
      "loss": 0.0014,
      "step": 211020
    },
    {
      "epoch": 1.3534808695680058,
      "grad_norm": 0.045922085642814636,
      "learning_rate": 2.8593978669619522e-06,
      "loss": 0.0007,
      "step": 211030
    },
    {
      "epoch": 1.353545006461792,
      "grad_norm": 0.42880502343177795,
      "learning_rate": 2.858892064292764e-06,
      "loss": 0.0008,
      "step": 211040
    },
    {
      "epoch": 1.3536091433555781,
      "grad_norm": 0.06439348310232162,
      "learning_rate": 2.858386288453189e-06,
      "loss": 0.0007,
      "step": 211050
    },
    {
      "epoch": 1.3536732802493643,
      "grad_norm": 0.13360503315925598,
      "learning_rate": 2.8578805394495623e-06,
      "loss": 0.0013,
      "step": 211060
    },
    {
      "epoch": 1.3537374171431504,
      "grad_norm": 0.10818067193031311,
      "learning_rate": 2.8573748172882254e-06,
      "loss": 0.0012,
      "step": 211070
    },
    {
      "epoch": 1.3538015540369364,
      "grad_norm": 0.1894940733909607,
      "learning_rate": 2.8568691219755133e-06,
      "loss": 0.0031,
      "step": 211080
    },
    {
      "epoch": 1.3538656909307225,
      "grad_norm": 0.02911425195634365,
      "learning_rate": 2.856363453517763e-06,
      "loss": 0.001,
      "step": 211090
    },
    {
      "epoch": 1.3539298278245087,
      "grad_norm": 0.15279807150363922,
      "learning_rate": 2.855857811921309e-06,
      "loss": 0.0015,
      "step": 211100
    },
    {
      "epoch": 1.3539939647182948,
      "grad_norm": 0.004417724907398224,
      "learning_rate": 2.8553521971924903e-06,
      "loss": 0.0009,
      "step": 211110
    },
    {
      "epoch": 1.3540581016120807,
      "grad_norm": 0.14657741785049438,
      "learning_rate": 2.8548466093376415e-06,
      "loss": 0.0008,
      "step": 211120
    },
    {
      "epoch": 1.3541222385058669,
      "grad_norm": 0.474303275346756,
      "learning_rate": 2.8543410483630974e-06,
      "loss": 0.0025,
      "step": 211130
    },
    {
      "epoch": 1.354186375399653,
      "grad_norm": 0.08662206679582596,
      "learning_rate": 2.853835514275192e-06,
      "loss": 0.0017,
      "step": 211140
    },
    {
      "epoch": 1.3542505122934392,
      "grad_norm": 0.0414050854742527,
      "learning_rate": 2.853330007080264e-06,
      "loss": 0.0012,
      "step": 211150
    },
    {
      "epoch": 1.3543146491872253,
      "grad_norm": 0.1096024289727211,
      "learning_rate": 2.8528245267846446e-06,
      "loss": 0.0016,
      "step": 211160
    },
    {
      "epoch": 1.3543787860810113,
      "grad_norm": 0.19043126702308655,
      "learning_rate": 2.852319073394666e-06,
      "loss": 0.001,
      "step": 211170
    },
    {
      "epoch": 1.3544429229747974,
      "grad_norm": 0.045862916857004166,
      "learning_rate": 2.851813646916667e-06,
      "loss": 0.0011,
      "step": 211180
    },
    {
      "epoch": 1.3545070598685836,
      "grad_norm": 0.1857152283191681,
      "learning_rate": 2.8513082473569775e-06,
      "loss": 0.0022,
      "step": 211190
    },
    {
      "epoch": 1.3545711967623695,
      "grad_norm": 0.09480617940425873,
      "learning_rate": 2.850802874721932e-06,
      "loss": 0.0023,
      "step": 211200
    },
    {
      "epoch": 1.3546353336561556,
      "grad_norm": 0.01438390463590622,
      "learning_rate": 2.8502975290178604e-06,
      "loss": 0.0008,
      "step": 211210
    },
    {
      "epoch": 1.3546994705499418,
      "grad_norm": 0.053283028304576874,
      "learning_rate": 2.849792210251099e-06,
      "loss": 0.0012,
      "step": 211220
    },
    {
      "epoch": 1.354763607443728,
      "grad_norm": 0.014264298602938652,
      "learning_rate": 2.849286918427978e-06,
      "loss": 0.0013,
      "step": 211230
    },
    {
      "epoch": 1.354827744337514,
      "grad_norm": 0.017701173201203346,
      "learning_rate": 2.8487816535548286e-06,
      "loss": 0.0051,
      "step": 211240
    },
    {
      "epoch": 1.3548918812313002,
      "grad_norm": 0.03873252496123314,
      "learning_rate": 2.848276415637981e-06,
      "loss": 0.0009,
      "step": 211250
    },
    {
      "epoch": 1.3549560181250861,
      "grad_norm": 0.015451488085091114,
      "learning_rate": 2.8477712046837697e-06,
      "loss": 0.0015,
      "step": 211260
    },
    {
      "epoch": 1.3550201550188723,
      "grad_norm": 0.08412735164165497,
      "learning_rate": 2.8472660206985237e-06,
      "loss": 0.0011,
      "step": 211270
    },
    {
      "epoch": 1.3550842919126584,
      "grad_norm": 0.06948544085025787,
      "learning_rate": 2.84676086368857e-06,
      "loss": 0.0023,
      "step": 211280
    },
    {
      "epoch": 1.3551484288064444,
      "grad_norm": 0.07955335080623627,
      "learning_rate": 2.846255733660245e-06,
      "loss": 0.0008,
      "step": 211290
    },
    {
      "epoch": 1.3552125657002305,
      "grad_norm": 0.0016547439154237509,
      "learning_rate": 2.8457506306198733e-06,
      "loss": 0.0007,
      "step": 211300
    },
    {
      "epoch": 1.3552767025940167,
      "grad_norm": 0.0419953428208828,
      "learning_rate": 2.8452455545737866e-06,
      "loss": 0.0013,
      "step": 211310
    },
    {
      "epoch": 1.3553408394878028,
      "grad_norm": 0.14658237993717194,
      "learning_rate": 2.8447405055283117e-06,
      "loss": 0.001,
      "step": 211320
    },
    {
      "epoch": 1.355404976381589,
      "grad_norm": 0.0616254098713398,
      "learning_rate": 2.8442354834897793e-06,
      "loss": 0.0008,
      "step": 211330
    },
    {
      "epoch": 1.355469113275375,
      "grad_norm": 0.013792560435831547,
      "learning_rate": 2.843730488464518e-06,
      "loss": 0.0017,
      "step": 211340
    },
    {
      "epoch": 1.355533250169161,
      "grad_norm": 0.062466397881507874,
      "learning_rate": 2.8432255204588542e-06,
      "loss": 0.0011,
      "step": 211350
    },
    {
      "epoch": 1.3555973870629472,
      "grad_norm": 0.0883217379450798,
      "learning_rate": 2.8427205794791146e-06,
      "loss": 0.0012,
      "step": 211360
    },
    {
      "epoch": 1.3556615239567331,
      "grad_norm": 0.11724873632192612,
      "learning_rate": 2.8422156655316297e-06,
      "loss": 0.0014,
      "step": 211370
    },
    {
      "epoch": 1.3557256608505193,
      "grad_norm": 0.07271930575370789,
      "learning_rate": 2.8417107786227244e-06,
      "loss": 0.0012,
      "step": 211380
    },
    {
      "epoch": 1.3557897977443054,
      "grad_norm": 0.06599129736423492,
      "learning_rate": 2.841205918758726e-06,
      "loss": 0.0016,
      "step": 211390
    },
    {
      "epoch": 1.3558539346380916,
      "grad_norm": 0.0861605778336525,
      "learning_rate": 2.8407010859459577e-06,
      "loss": 0.0024,
      "step": 211400
    },
    {
      "epoch": 1.3559180715318777,
      "grad_norm": 0.12029363214969635,
      "learning_rate": 2.84019628019075e-06,
      "loss": 0.001,
      "step": 211410
    },
    {
      "epoch": 1.3559822084256639,
      "grad_norm": 0.10664620250463486,
      "learning_rate": 2.839691501499425e-06,
      "loss": 0.0026,
      "step": 211420
    },
    {
      "epoch": 1.3560463453194498,
      "grad_norm": 0.06973236799240112,
      "learning_rate": 2.8391867498783117e-06,
      "loss": 0.0014,
      "step": 211430
    },
    {
      "epoch": 1.356110482213236,
      "grad_norm": 0.07850681245326996,
      "learning_rate": 2.83868202533373e-06,
      "loss": 0.0014,
      "step": 211440
    },
    {
      "epoch": 1.356174619107022,
      "grad_norm": 0.07012001425027847,
      "learning_rate": 2.838177327872009e-06,
      "loss": 0.0015,
      "step": 211450
    },
    {
      "epoch": 1.356238756000808,
      "grad_norm": 0.05666491761803627,
      "learning_rate": 2.8376726574994722e-06,
      "loss": 0.0025,
      "step": 211460
    },
    {
      "epoch": 1.3563028928945942,
      "grad_norm": 0.13862815499305725,
      "learning_rate": 2.8371680142224422e-06,
      "loss": 0.0019,
      "step": 211470
    },
    {
      "epoch": 1.3563670297883803,
      "grad_norm": 0.1430405229330063,
      "learning_rate": 2.83666339804724e-06,
      "loss": 0.001,
      "step": 211480
    },
    {
      "epoch": 1.3564311666821665,
      "grad_norm": 0.04503755271434784,
      "learning_rate": 2.836158808980194e-06,
      "loss": 0.0011,
      "step": 211490
    },
    {
      "epoch": 1.3564953035759526,
      "grad_norm": 0.22144374251365662,
      "learning_rate": 2.8356542470276242e-06,
      "loss": 0.0012,
      "step": 211500
    },
    {
      "epoch": 1.3565594404697385,
      "grad_norm": 0.1325940489768982,
      "learning_rate": 2.8351497121958536e-06,
      "loss": 0.001,
      "step": 211510
    },
    {
      "epoch": 1.3566235773635247,
      "grad_norm": 0.159137561917305,
      "learning_rate": 2.8346452044912033e-06,
      "loss": 0.0027,
      "step": 211520
    },
    {
      "epoch": 1.3566877142573108,
      "grad_norm": 0.043283578008413315,
      "learning_rate": 2.834140723919997e-06,
      "loss": 0.001,
      "step": 211530
    },
    {
      "epoch": 1.356751851151097,
      "grad_norm": 0.14707869291305542,
      "learning_rate": 2.8336362704885554e-06,
      "loss": 0.001,
      "step": 211540
    },
    {
      "epoch": 1.356815988044883,
      "grad_norm": 0.10080327838659286,
      "learning_rate": 2.833131844203198e-06,
      "loss": 0.0017,
      "step": 211550
    },
    {
      "epoch": 1.356880124938669,
      "grad_norm": 0.09956194460391998,
      "learning_rate": 2.8326274450702497e-06,
      "loss": 0.0018,
      "step": 211560
    },
    {
      "epoch": 1.3569442618324552,
      "grad_norm": 0.24537304043769836,
      "learning_rate": 2.8321230730960274e-06,
      "loss": 0.0013,
      "step": 211570
    },
    {
      "epoch": 1.3570083987262413,
      "grad_norm": 0.0012119578896090388,
      "learning_rate": 2.831618728286853e-06,
      "loss": 0.0006,
      "step": 211580
    },
    {
      "epoch": 1.3570725356200275,
      "grad_norm": 0.06794049590826035,
      "learning_rate": 2.8311144106490435e-06,
      "loss": 0.0015,
      "step": 211590
    },
    {
      "epoch": 1.3571366725138134,
      "grad_norm": 0.1144900992512703,
      "learning_rate": 2.8306101201889226e-06,
      "loss": 0.0014,
      "step": 211600
    },
    {
      "epoch": 1.3572008094075996,
      "grad_norm": 0.0597737655043602,
      "learning_rate": 2.830105856912807e-06,
      "loss": 0.0013,
      "step": 211610
    },
    {
      "epoch": 1.3572649463013857,
      "grad_norm": 0.08430390805006027,
      "learning_rate": 2.8296016208270162e-06,
      "loss": 0.0017,
      "step": 211620
    },
    {
      "epoch": 1.3573290831951716,
      "grad_norm": 0.06578479707241058,
      "learning_rate": 2.829097411937867e-06,
      "loss": 0.0006,
      "step": 211630
    },
    {
      "epoch": 1.3573932200889578,
      "grad_norm": 0.16109733283519745,
      "learning_rate": 2.8285932302516794e-06,
      "loss": 0.002,
      "step": 211640
    },
    {
      "epoch": 1.357457356982744,
      "grad_norm": 0.021322287619113922,
      "learning_rate": 2.828089075774771e-06,
      "loss": 0.0006,
      "step": 211650
    },
    {
      "epoch": 1.35752149387653,
      "grad_norm": 0.08686596155166626,
      "learning_rate": 2.827584948513459e-06,
      "loss": 0.0011,
      "step": 211660
    },
    {
      "epoch": 1.3575856307703162,
      "grad_norm": 0.031903836876153946,
      "learning_rate": 2.8270808484740576e-06,
      "loss": 0.0011,
      "step": 211670
    },
    {
      "epoch": 1.3576497676641024,
      "grad_norm": 0.12447334825992584,
      "learning_rate": 2.8265767756628883e-06,
      "loss": 0.0016,
      "step": 211680
    },
    {
      "epoch": 1.3577139045578883,
      "grad_norm": 0.28986862301826477,
      "learning_rate": 2.8260727300862655e-06,
      "loss": 0.0021,
      "step": 211690
    },
    {
      "epoch": 1.3577780414516745,
      "grad_norm": 0.024045251309871674,
      "learning_rate": 2.825568711750503e-06,
      "loss": 0.0024,
      "step": 211700
    },
    {
      "epoch": 1.3578421783454606,
      "grad_norm": 0.10672322660684586,
      "learning_rate": 2.825064720661921e-06,
      "loss": 0.0017,
      "step": 211710
    },
    {
      "epoch": 1.3579063152392465,
      "grad_norm": 0.06658641993999481,
      "learning_rate": 2.824560756826832e-06,
      "loss": 0.0016,
      "step": 211720
    },
    {
      "epoch": 1.3579704521330327,
      "grad_norm": 0.011623168364167213,
      "learning_rate": 2.8240568202515517e-06,
      "loss": 0.0015,
      "step": 211730
    },
    {
      "epoch": 1.3580345890268188,
      "grad_norm": 0.029520215466618538,
      "learning_rate": 2.823552910942392e-06,
      "loss": 0.0012,
      "step": 211740
    },
    {
      "epoch": 1.358098725920605,
      "grad_norm": 0.04139503836631775,
      "learning_rate": 2.823049028905672e-06,
      "loss": 0.0004,
      "step": 211750
    },
    {
      "epoch": 1.3581628628143911,
      "grad_norm": 0.11593854427337646,
      "learning_rate": 2.8225451741477037e-06,
      "loss": 0.0016,
      "step": 211760
    },
    {
      "epoch": 1.358226999708177,
      "grad_norm": 0.02130136266350746,
      "learning_rate": 2.8220413466748008e-06,
      "loss": 0.0017,
      "step": 211770
    },
    {
      "epoch": 1.3582911366019632,
      "grad_norm": 0.07793150842189789,
      "learning_rate": 2.8215375464932748e-06,
      "loss": 0.0014,
      "step": 211780
    },
    {
      "epoch": 1.3583552734957494,
      "grad_norm": 0.07984554022550583,
      "learning_rate": 2.8210337736094417e-06,
      "loss": 0.0008,
      "step": 211790
    },
    {
      "epoch": 1.3584194103895353,
      "grad_norm": 0.04542630538344383,
      "learning_rate": 2.8205300280296123e-06,
      "loss": 0.0023,
      "step": 211800
    },
    {
      "epoch": 1.3584835472833214,
      "grad_norm": 0.05402369797229767,
      "learning_rate": 2.8200263097601e-06,
      "loss": 0.0013,
      "step": 211810
    },
    {
      "epoch": 1.3585476841771076,
      "grad_norm": 0.09029842913150787,
      "learning_rate": 2.8195226188072135e-06,
      "loss": 0.0014,
      "step": 211820
    },
    {
      "epoch": 1.3586118210708937,
      "grad_norm": 0.06096288934350014,
      "learning_rate": 2.8190189551772695e-06,
      "loss": 0.0016,
      "step": 211830
    },
    {
      "epoch": 1.3586759579646799,
      "grad_norm": 0.09616797417402267,
      "learning_rate": 2.818515318876576e-06,
      "loss": 0.0008,
      "step": 211840
    },
    {
      "epoch": 1.358740094858466,
      "grad_norm": 0.09054041653871536,
      "learning_rate": 2.818011709911443e-06,
      "loss": 0.0017,
      "step": 211850
    },
    {
      "epoch": 1.358804231752252,
      "grad_norm": 0.10492989420890808,
      "learning_rate": 2.8175081282881843e-06,
      "loss": 0.0008,
      "step": 211860
    },
    {
      "epoch": 1.358868368646038,
      "grad_norm": 0.49018484354019165,
      "learning_rate": 2.8170045740131093e-06,
      "loss": 0.0048,
      "step": 211870
    },
    {
      "epoch": 1.3589325055398243,
      "grad_norm": 0.30528512597084045,
      "learning_rate": 2.816501047092527e-06,
      "loss": 0.0017,
      "step": 211880
    },
    {
      "epoch": 1.3589966424336102,
      "grad_norm": 0.035619039088487625,
      "learning_rate": 2.815997547532745e-06,
      "loss": 0.0013,
      "step": 211890
    },
    {
      "epoch": 1.3590607793273963,
      "grad_norm": 0.08016897737979889,
      "learning_rate": 2.8154940753400763e-06,
      "loss": 0.0013,
      "step": 211900
    },
    {
      "epoch": 1.3591249162211825,
      "grad_norm": 0.058081433176994324,
      "learning_rate": 2.8149906305208285e-06,
      "loss": 0.0007,
      "step": 211910
    },
    {
      "epoch": 1.3591890531149686,
      "grad_norm": 0.03151792287826538,
      "learning_rate": 2.8144872130813097e-06,
      "loss": 0.0017,
      "step": 211920
    },
    {
      "epoch": 1.3592531900087548,
      "grad_norm": 0.20671361684799194,
      "learning_rate": 2.813983823027826e-06,
      "loss": 0.0025,
      "step": 211930
    },
    {
      "epoch": 1.3593173269025407,
      "grad_norm": 0.0331951342523098,
      "learning_rate": 2.8134804603666887e-06,
      "loss": 0.0017,
      "step": 211940
    },
    {
      "epoch": 1.3593814637963268,
      "grad_norm": 0.09501717239618301,
      "learning_rate": 2.812977125104205e-06,
      "loss": 0.0032,
      "step": 211950
    },
    {
      "epoch": 1.359445600690113,
      "grad_norm": 0.14686033129692078,
      "learning_rate": 2.8124738172466805e-06,
      "loss": 0.0013,
      "step": 211960
    },
    {
      "epoch": 1.3595097375838991,
      "grad_norm": 0.08381103724241257,
      "learning_rate": 2.81197053680042e-06,
      "loss": 0.0018,
      "step": 211970
    },
    {
      "epoch": 1.359573874477685,
      "grad_norm": 0.17146036028862,
      "learning_rate": 2.811467283771735e-06,
      "loss": 0.0013,
      "step": 211980
    },
    {
      "epoch": 1.3596380113714712,
      "grad_norm": 0.19006793200969696,
      "learning_rate": 2.810964058166929e-06,
      "loss": 0.0018,
      "step": 211990
    },
    {
      "epoch": 1.3597021482652574,
      "grad_norm": 0.023726634681224823,
      "learning_rate": 2.8104608599923055e-06,
      "loss": 0.0029,
      "step": 212000
    },
    {
      "epoch": 1.3597662851590435,
      "grad_norm": 0.07961488515138626,
      "learning_rate": 2.8099576892541746e-06,
      "loss": 0.0015,
      "step": 212010
    },
    {
      "epoch": 1.3598304220528297,
      "grad_norm": 0.17800407111644745,
      "learning_rate": 2.809454545958839e-06,
      "loss": 0.0013,
      "step": 212020
    },
    {
      "epoch": 1.3598945589466156,
      "grad_norm": 0.12595905363559723,
      "learning_rate": 2.808951430112603e-06,
      "loss": 0.0012,
      "step": 212030
    },
    {
      "epoch": 1.3599586958404017,
      "grad_norm": 0.09590614587068558,
      "learning_rate": 2.80844834172177e-06,
      "loss": 0.0019,
      "step": 212040
    },
    {
      "epoch": 1.360022832734188,
      "grad_norm": 0.03253532946109772,
      "learning_rate": 2.8079452807926477e-06,
      "loss": 0.0007,
      "step": 212050
    },
    {
      "epoch": 1.3600869696279738,
      "grad_norm": 0.0494966059923172,
      "learning_rate": 2.8074422473315376e-06,
      "loss": 0.0015,
      "step": 212060
    },
    {
      "epoch": 1.36015110652176,
      "grad_norm": 0.0349554680287838,
      "learning_rate": 2.8069392413447425e-06,
      "loss": 0.0033,
      "step": 212070
    },
    {
      "epoch": 1.3602152434155461,
      "grad_norm": 0.0674915537238121,
      "learning_rate": 2.806436262838565e-06,
      "loss": 0.0021,
      "step": 212080
    },
    {
      "epoch": 1.3602793803093323,
      "grad_norm": 0.0689818412065506,
      "learning_rate": 2.8059333118193103e-06,
      "loss": 0.0014,
      "step": 212090
    },
    {
      "epoch": 1.3603435172031184,
      "grad_norm": 0.052752457559108734,
      "learning_rate": 2.80543038829328e-06,
      "loss": 0.001,
      "step": 212100
    },
    {
      "epoch": 1.3604076540969046,
      "grad_norm": 0.12120924890041351,
      "learning_rate": 2.8049274922667745e-06,
      "loss": 0.0013,
      "step": 212110
    },
    {
      "epoch": 1.3604717909906905,
      "grad_norm": 0.23242832720279694,
      "learning_rate": 2.8044246237460944e-06,
      "loss": 0.0013,
      "step": 212120
    },
    {
      "epoch": 1.3605359278844766,
      "grad_norm": 0.15054942667484283,
      "learning_rate": 2.8039217827375455e-06,
      "loss": 0.0027,
      "step": 212130
    },
    {
      "epoch": 1.3606000647782628,
      "grad_norm": 0.06705403327941895,
      "learning_rate": 2.8034189692474257e-06,
      "loss": 0.0011,
      "step": 212140
    },
    {
      "epoch": 1.3606642016720487,
      "grad_norm": 0.06358052045106888,
      "learning_rate": 2.802916183282034e-06,
      "loss": 0.001,
      "step": 212150
    },
    {
      "epoch": 1.3607283385658349,
      "grad_norm": 0.027282526716589928,
      "learning_rate": 2.8024134248476746e-06,
      "loss": 0.0043,
      "step": 212160
    },
    {
      "epoch": 1.360792475459621,
      "grad_norm": 0.016235968098044395,
      "learning_rate": 2.8019106939506458e-06,
      "loss": 0.0017,
      "step": 212170
    },
    {
      "epoch": 1.3608566123534072,
      "grad_norm": 0.10143746435642242,
      "learning_rate": 2.8014079905972467e-06,
      "loss": 0.0014,
      "step": 212180
    },
    {
      "epoch": 1.3609207492471933,
      "grad_norm": 0.08578982949256897,
      "learning_rate": 2.800905314793776e-06,
      "loss": 0.0014,
      "step": 212190
    },
    {
      "epoch": 1.3609848861409792,
      "grad_norm": 0.14914195239543915,
      "learning_rate": 2.8004026665465324e-06,
      "loss": 0.002,
      "step": 212200
    },
    {
      "epoch": 1.3610490230347654,
      "grad_norm": 0.14445620775222778,
      "learning_rate": 2.7999000458618176e-06,
      "loss": 0.0009,
      "step": 212210
    },
    {
      "epoch": 1.3611131599285515,
      "grad_norm": 0.04894477128982544,
      "learning_rate": 2.799397452745927e-06,
      "loss": 0.0009,
      "step": 212220
    },
    {
      "epoch": 1.3611772968223377,
      "grad_norm": 0.09352356940507889,
      "learning_rate": 2.7988948872051596e-06,
      "loss": 0.0014,
      "step": 212230
    },
    {
      "epoch": 1.3612414337161236,
      "grad_norm": 0.021016614511609077,
      "learning_rate": 2.798392349245811e-06,
      "loss": 0.0011,
      "step": 212240
    },
    {
      "epoch": 1.3613055706099098,
      "grad_norm": 0.038588326424360275,
      "learning_rate": 2.7978898388741805e-06,
      "loss": 0.001,
      "step": 212250
    },
    {
      "epoch": 1.361369707503696,
      "grad_norm": 0.25138717889785767,
      "learning_rate": 2.7973873560965647e-06,
      "loss": 0.0015,
      "step": 212260
    },
    {
      "epoch": 1.361433844397482,
      "grad_norm": 0.06818045675754547,
      "learning_rate": 2.796884900919258e-06,
      "loss": 0.0015,
      "step": 212270
    },
    {
      "epoch": 1.3614979812912682,
      "grad_norm": 0.07317348569631577,
      "learning_rate": 2.796382473348559e-06,
      "loss": 0.0012,
      "step": 212280
    },
    {
      "epoch": 1.3615621181850541,
      "grad_norm": 0.12382987886667252,
      "learning_rate": 2.795880073390763e-06,
      "loss": 0.0012,
      "step": 212290
    },
    {
      "epoch": 1.3616262550788403,
      "grad_norm": 0.12995034456253052,
      "learning_rate": 2.7953777010521656e-06,
      "loss": 0.0044,
      "step": 212300
    },
    {
      "epoch": 1.3616903919726264,
      "grad_norm": 0.016672872006893158,
      "learning_rate": 2.7948753563390585e-06,
      "loss": 0.0021,
      "step": 212310
    },
    {
      "epoch": 1.3617545288664124,
      "grad_norm": 0.07339946180582047,
      "learning_rate": 2.7943730392577416e-06,
      "loss": 0.0013,
      "step": 212320
    },
    {
      "epoch": 1.3618186657601985,
      "grad_norm": 0.1623236984014511,
      "learning_rate": 2.793870749814506e-06,
      "loss": 0.0015,
      "step": 212330
    },
    {
      "epoch": 1.3618828026539846,
      "grad_norm": 0.11059384793043137,
      "learning_rate": 2.7933684880156475e-06,
      "loss": 0.0015,
      "step": 212340
    },
    {
      "epoch": 1.3619469395477708,
      "grad_norm": 0.06458184868097305,
      "learning_rate": 2.792866253867457e-06,
      "loss": 0.001,
      "step": 212350
    },
    {
      "epoch": 1.362011076441557,
      "grad_norm": 0.032684240490198135,
      "learning_rate": 2.7923640473762307e-06,
      "loss": 0.0006,
      "step": 212360
    },
    {
      "epoch": 1.362075213335343,
      "grad_norm": 0.14689889550209045,
      "learning_rate": 2.791861868548261e-06,
      "loss": 0.003,
      "step": 212370
    },
    {
      "epoch": 1.362139350229129,
      "grad_norm": 0.039106424897909164,
      "learning_rate": 2.791359717389841e-06,
      "loss": 0.0016,
      "step": 212380
    },
    {
      "epoch": 1.3622034871229152,
      "grad_norm": 0.16570846736431122,
      "learning_rate": 2.79085759390726e-06,
      "loss": 0.002,
      "step": 212390
    },
    {
      "epoch": 1.3622676240167013,
      "grad_norm": 0.02795327454805374,
      "learning_rate": 2.7903554981068135e-06,
      "loss": 0.0014,
      "step": 212400
    },
    {
      "epoch": 1.3623317609104872,
      "grad_norm": 0.08838541805744171,
      "learning_rate": 2.789853429994792e-06,
      "loss": 0.0012,
      "step": 212410
    },
    {
      "epoch": 1.3623958978042734,
      "grad_norm": 0.09232831746339798,
      "learning_rate": 2.7893513895774856e-06,
      "loss": 0.002,
      "step": 212420
    },
    {
      "epoch": 1.3624600346980595,
      "grad_norm": 0.07343509048223495,
      "learning_rate": 2.7888493768611867e-06,
      "loss": 0.0023,
      "step": 212430
    },
    {
      "epoch": 1.3625241715918457,
      "grad_norm": 0.044543057680130005,
      "learning_rate": 2.7883473918521864e-06,
      "loss": 0.0028,
      "step": 212440
    },
    {
      "epoch": 1.3625883084856318,
      "grad_norm": 0.693203330039978,
      "learning_rate": 2.787845434556774e-06,
      "loss": 0.0024,
      "step": 212450
    },
    {
      "epoch": 1.3626524453794178,
      "grad_norm": 0.20361362397670746,
      "learning_rate": 2.787343504981237e-06,
      "loss": 0.0016,
      "step": 212460
    },
    {
      "epoch": 1.362716582273204,
      "grad_norm": 0.2686789035797119,
      "learning_rate": 2.786841603131869e-06,
      "loss": 0.0023,
      "step": 212470
    },
    {
      "epoch": 1.36278071916699,
      "grad_norm": 0.09884592890739441,
      "learning_rate": 2.786339729014958e-06,
      "loss": 0.0015,
      "step": 212480
    },
    {
      "epoch": 1.362844856060776,
      "grad_norm": 0.021662624552845955,
      "learning_rate": 2.7858378826367914e-06,
      "loss": 0.0015,
      "step": 212490
    },
    {
      "epoch": 1.3629089929545621,
      "grad_norm": 0.03224967420101166,
      "learning_rate": 2.7853360640036577e-06,
      "loss": 0.0014,
      "step": 212500
    },
    {
      "epoch": 1.3629731298483483,
      "grad_norm": 0.1176484078168869,
      "learning_rate": 2.784834273121847e-06,
      "loss": 0.0013,
      "step": 212510
    },
    {
      "epoch": 1.3630372667421344,
      "grad_norm": 0.10865365713834763,
      "learning_rate": 2.7843325099976467e-06,
      "loss": 0.0011,
      "step": 212520
    },
    {
      "epoch": 1.3631014036359206,
      "grad_norm": 0.026730181649327278,
      "learning_rate": 2.7838307746373427e-06,
      "loss": 0.0011,
      "step": 212530
    },
    {
      "epoch": 1.3631655405297067,
      "grad_norm": 0.1156671792268753,
      "learning_rate": 2.783329067047222e-06,
      "loss": 0.0011,
      "step": 212540
    },
    {
      "epoch": 1.3632296774234927,
      "grad_norm": 0.09126152098178864,
      "learning_rate": 2.7828273872335738e-06,
      "loss": 0.0012,
      "step": 212550
    },
    {
      "epoch": 1.3632938143172788,
      "grad_norm": 0.021603649482131004,
      "learning_rate": 2.7823257352026826e-06,
      "loss": 0.0019,
      "step": 212560
    },
    {
      "epoch": 1.363357951211065,
      "grad_norm": 0.009608804248273373,
      "learning_rate": 2.7818241109608333e-06,
      "loss": 0.0019,
      "step": 212570
    },
    {
      "epoch": 1.3634220881048509,
      "grad_norm": 0.07074563205242157,
      "learning_rate": 2.781322514514315e-06,
      "loss": 0.0034,
      "step": 212580
    },
    {
      "epoch": 1.363486224998637,
      "grad_norm": 0.16082030534744263,
      "learning_rate": 2.7808209458694117e-06,
      "loss": 0.0013,
      "step": 212590
    },
    {
      "epoch": 1.3635503618924232,
      "grad_norm": 0.06372162699699402,
      "learning_rate": 2.7803194050324078e-06,
      "loss": 0.0015,
      "step": 212600
    },
    {
      "epoch": 1.3636144987862093,
      "grad_norm": 0.00480593740940094,
      "learning_rate": 2.7798178920095865e-06,
      "loss": 0.0018,
      "step": 212610
    },
    {
      "epoch": 1.3636786356799955,
      "grad_norm": 0.07020413130521774,
      "learning_rate": 2.7793164068072353e-06,
      "loss": 0.0011,
      "step": 212620
    },
    {
      "epoch": 1.3637427725737814,
      "grad_norm": 0.0518869049847126,
      "learning_rate": 2.7788149494316364e-06,
      "loss": 0.0013,
      "step": 212630
    },
    {
      "epoch": 1.3638069094675676,
      "grad_norm": 0.04014595225453377,
      "learning_rate": 2.7783135198890743e-06,
      "loss": 0.0021,
      "step": 212640
    },
    {
      "epoch": 1.3638710463613537,
      "grad_norm": 0.012117862701416016,
      "learning_rate": 2.7778121181858285e-06,
      "loss": 0.001,
      "step": 212650
    },
    {
      "epoch": 1.3639351832551398,
      "grad_norm": 0.0709318220615387,
      "learning_rate": 2.7773107443281877e-06,
      "loss": 0.0023,
      "step": 212660
    },
    {
      "epoch": 1.3639993201489258,
      "grad_norm": 0.013128736987709999,
      "learning_rate": 2.776809398322432e-06,
      "loss": 0.0012,
      "step": 212670
    },
    {
      "epoch": 1.364063457042712,
      "grad_norm": 0.0026822881773114204,
      "learning_rate": 2.7763080801748434e-06,
      "loss": 0.0009,
      "step": 212680
    },
    {
      "epoch": 1.364127593936498,
      "grad_norm": 0.008389109745621681,
      "learning_rate": 2.7758067898917018e-06,
      "loss": 0.0013,
      "step": 212690
    },
    {
      "epoch": 1.3641917308302842,
      "grad_norm": 0.018718862906098366,
      "learning_rate": 2.775305527479293e-06,
      "loss": 0.0009,
      "step": 212700
    },
    {
      "epoch": 1.3642558677240704,
      "grad_norm": 0.09819506108760834,
      "learning_rate": 2.774804292943895e-06,
      "loss": 0.002,
      "step": 212710
    },
    {
      "epoch": 1.3643200046178563,
      "grad_norm": 0.008475651033222675,
      "learning_rate": 2.774303086291788e-06,
      "loss": 0.0011,
      "step": 212720
    },
    {
      "epoch": 1.3643841415116424,
      "grad_norm": 0.05434533953666687,
      "learning_rate": 2.773801907529256e-06,
      "loss": 0.0009,
      "step": 212730
    },
    {
      "epoch": 1.3644482784054286,
      "grad_norm": 0.017436152324080467,
      "learning_rate": 2.773300756662577e-06,
      "loss": 0.0009,
      "step": 212740
    },
    {
      "epoch": 1.3645124152992145,
      "grad_norm": 0.05302887409925461,
      "learning_rate": 2.772799633698031e-06,
      "loss": 0.0015,
      "step": 212750
    },
    {
      "epoch": 1.3645765521930007,
      "grad_norm": 0.09424480050802231,
      "learning_rate": 2.7722985386418957e-06,
      "loss": 0.0011,
      "step": 212760
    },
    {
      "epoch": 1.3646406890867868,
      "grad_norm": 0.007592364680022001,
      "learning_rate": 2.7717974715004534e-06,
      "loss": 0.0008,
      "step": 212770
    },
    {
      "epoch": 1.364704825980573,
      "grad_norm": 0.03045915625989437,
      "learning_rate": 2.771296432279982e-06,
      "loss": 0.001,
      "step": 212780
    },
    {
      "epoch": 1.3647689628743591,
      "grad_norm": 0.040771279484033585,
      "learning_rate": 2.7707954209867584e-06,
      "loss": 0.0003,
      "step": 212790
    },
    {
      "epoch": 1.3648330997681453,
      "grad_norm": 0.15090760588645935,
      "learning_rate": 2.770294437627059e-06,
      "loss": 0.0013,
      "step": 212800
    },
    {
      "epoch": 1.3648972366619312,
      "grad_norm": 0.1347474604845047,
      "learning_rate": 2.769793482207167e-06,
      "loss": 0.0013,
      "step": 212810
    },
    {
      "epoch": 1.3649613735557173,
      "grad_norm": 0.07247631996870041,
      "learning_rate": 2.769292554733356e-06,
      "loss": 0.0014,
      "step": 212820
    },
    {
      "epoch": 1.3650255104495035,
      "grad_norm": 0.15575724840164185,
      "learning_rate": 2.768791655211903e-06,
      "loss": 0.0022,
      "step": 212830
    },
    {
      "epoch": 1.3650896473432894,
      "grad_norm": 0.0690620094537735,
      "learning_rate": 2.7682907836490834e-06,
      "loss": 0.0006,
      "step": 212840
    },
    {
      "epoch": 1.3651537842370756,
      "grad_norm": 0.020767055451869965,
      "learning_rate": 2.767789940051178e-06,
      "loss": 0.0006,
      "step": 212850
    },
    {
      "epoch": 1.3652179211308617,
      "grad_norm": 0.07257429510354996,
      "learning_rate": 2.7672891244244593e-06,
      "loss": 0.0007,
      "step": 212860
    },
    {
      "epoch": 1.3652820580246479,
      "grad_norm": 0.007378075271844864,
      "learning_rate": 2.7667883367752024e-06,
      "loss": 0.0006,
      "step": 212870
    },
    {
      "epoch": 1.365346194918434,
      "grad_norm": 0.09765730053186417,
      "learning_rate": 2.7662875771096847e-06,
      "loss": 0.0026,
      "step": 212880
    },
    {
      "epoch": 1.36541033181222,
      "grad_norm": 0.04521140828728676,
      "learning_rate": 2.765786845434181e-06,
      "loss": 0.0018,
      "step": 212890
    },
    {
      "epoch": 1.365474468706006,
      "grad_norm": 0.14008601009845734,
      "learning_rate": 2.765286141754964e-06,
      "loss": 0.0014,
      "step": 212900
    },
    {
      "epoch": 1.3655386055997922,
      "grad_norm": 0.20395956933498383,
      "learning_rate": 2.7647854660783067e-06,
      "loss": 0.0012,
      "step": 212910
    },
    {
      "epoch": 1.3656027424935782,
      "grad_norm": 0.11808919161558151,
      "learning_rate": 2.7642848184104876e-06,
      "loss": 0.0017,
      "step": 212920
    },
    {
      "epoch": 1.3656668793873643,
      "grad_norm": 0.16811446845531464,
      "learning_rate": 2.7637841987577775e-06,
      "loss": 0.0024,
      "step": 212930
    },
    {
      "epoch": 1.3657310162811505,
      "grad_norm": 0.09902886301279068,
      "learning_rate": 2.763283607126449e-06,
      "loss": 0.0011,
      "step": 212940
    },
    {
      "epoch": 1.3657951531749366,
      "grad_norm": 0.18342621624469757,
      "learning_rate": 2.7627830435227744e-06,
      "loss": 0.0019,
      "step": 212950
    },
    {
      "epoch": 1.3658592900687228,
      "grad_norm": 0.03563809022307396,
      "learning_rate": 2.762282507953027e-06,
      "loss": 0.0015,
      "step": 212960
    },
    {
      "epoch": 1.365923426962509,
      "grad_norm": 0.004312540870159864,
      "learning_rate": 2.761782000423481e-06,
      "loss": 0.0015,
      "step": 212970
    },
    {
      "epoch": 1.3659875638562948,
      "grad_norm": 0.08165877312421799,
      "learning_rate": 2.7612815209404052e-06,
      "loss": 0.0017,
      "step": 212980
    },
    {
      "epoch": 1.366051700750081,
      "grad_norm": 0.05772337317466736,
      "learning_rate": 2.760781069510071e-06,
      "loss": 0.0023,
      "step": 212990
    },
    {
      "epoch": 1.3661158376438671,
      "grad_norm": 0.0420399084687233,
      "learning_rate": 2.760280646138752e-06,
      "loss": 0.0012,
      "step": 213000
    },
    {
      "epoch": 1.366179974537653,
      "grad_norm": 0.04587164148688316,
      "learning_rate": 2.759780250832717e-06,
      "loss": 0.001,
      "step": 213010
    },
    {
      "epoch": 1.3662441114314392,
      "grad_norm": 0.1739441156387329,
      "learning_rate": 2.759279883598237e-06,
      "loss": 0.0011,
      "step": 213020
    },
    {
      "epoch": 1.3663082483252253,
      "grad_norm": 0.29219284653663635,
      "learning_rate": 2.7587795444415787e-06,
      "loss": 0.0018,
      "step": 213030
    },
    {
      "epoch": 1.3663723852190115,
      "grad_norm": 0.0494915209710598,
      "learning_rate": 2.7582792333690174e-06,
      "loss": 0.0017,
      "step": 213040
    },
    {
      "epoch": 1.3664365221127976,
      "grad_norm": 0.14492544531822205,
      "learning_rate": 2.757778950386819e-06,
      "loss": 0.0012,
      "step": 213050
    },
    {
      "epoch": 1.3665006590065836,
      "grad_norm": 0.14945034682750702,
      "learning_rate": 2.7572786955012527e-06,
      "loss": 0.0008,
      "step": 213060
    },
    {
      "epoch": 1.3665647959003697,
      "grad_norm": 0.1814311295747757,
      "learning_rate": 2.756778468718585e-06,
      "loss": 0.0023,
      "step": 213070
    },
    {
      "epoch": 1.3666289327941559,
      "grad_norm": 0.11481145769357681,
      "learning_rate": 2.7562782700450883e-06,
      "loss": 0.0016,
      "step": 213080
    },
    {
      "epoch": 1.366693069687942,
      "grad_norm": 0.05379915237426758,
      "learning_rate": 2.7557780994870298e-06,
      "loss": 0.0011,
      "step": 213090
    },
    {
      "epoch": 1.366757206581728,
      "grad_norm": 0.09029082208871841,
      "learning_rate": 2.7552779570506743e-06,
      "loss": 0.0015,
      "step": 213100
    },
    {
      "epoch": 1.366821343475514,
      "grad_norm": 0.20908387005329132,
      "learning_rate": 2.754777842742289e-06,
      "loss": 0.001,
      "step": 213110
    },
    {
      "epoch": 1.3668854803693002,
      "grad_norm": 0.10487104207277298,
      "learning_rate": 2.7542777565681434e-06,
      "loss": 0.0008,
      "step": 213120
    },
    {
      "epoch": 1.3669496172630864,
      "grad_norm": 0.13100507855415344,
      "learning_rate": 2.7537776985345023e-06,
      "loss": 0.0015,
      "step": 213130
    },
    {
      "epoch": 1.3670137541568725,
      "grad_norm": 0.01375681720674038,
      "learning_rate": 2.7532776686476302e-06,
      "loss": 0.0024,
      "step": 213140
    },
    {
      "epoch": 1.3670778910506585,
      "grad_norm": 0.1232491284608841,
      "learning_rate": 2.752777666913797e-06,
      "loss": 0.0155,
      "step": 213150
    },
    {
      "epoch": 1.3671420279444446,
      "grad_norm": 0.057872459292411804,
      "learning_rate": 2.7522776933392657e-06,
      "loss": 0.0014,
      "step": 213160
    },
    {
      "epoch": 1.3672061648382308,
      "grad_norm": 0.23585091531276703,
      "learning_rate": 2.751777747930301e-06,
      "loss": 0.0025,
      "step": 213170
    },
    {
      "epoch": 1.3672703017320167,
      "grad_norm": 0.10261121392250061,
      "learning_rate": 2.751277830693166e-06,
      "loss": 0.0016,
      "step": 213180
    },
    {
      "epoch": 1.3673344386258028,
      "grad_norm": 0.1702728569507599,
      "learning_rate": 2.7507779416341286e-06,
      "loss": 0.0019,
      "step": 213190
    },
    {
      "epoch": 1.367398575519589,
      "grad_norm": 0.1250714808702469,
      "learning_rate": 2.7502780807594513e-06,
      "loss": 0.0023,
      "step": 213200
    },
    {
      "epoch": 1.3674627124133751,
      "grad_norm": 0.049842942506074905,
      "learning_rate": 2.7497782480753976e-06,
      "loss": 0.0018,
      "step": 213210
    },
    {
      "epoch": 1.3675268493071613,
      "grad_norm": 0.1522701531648636,
      "learning_rate": 2.749278443588229e-06,
      "loss": 0.0012,
      "step": 213220
    },
    {
      "epoch": 1.3675909862009474,
      "grad_norm": 0.157740518450737,
      "learning_rate": 2.7487786673042115e-06,
      "loss": 0.0011,
      "step": 213230
    },
    {
      "epoch": 1.3676551230947334,
      "grad_norm": 0.1830168515443802,
      "learning_rate": 2.748278919229607e-06,
      "loss": 0.0038,
      "step": 213240
    },
    {
      "epoch": 1.3677192599885195,
      "grad_norm": 0.13043898344039917,
      "learning_rate": 2.7477791993706737e-06,
      "loss": 0.0017,
      "step": 213250
    },
    {
      "epoch": 1.3677833968823057,
      "grad_norm": 0.11714408546686172,
      "learning_rate": 2.74727950773368e-06,
      "loss": 0.0012,
      "step": 213260
    },
    {
      "epoch": 1.3678475337760916,
      "grad_norm": 0.10734383761882782,
      "learning_rate": 2.7467798443248827e-06,
      "loss": 0.0009,
      "step": 213270
    },
    {
      "epoch": 1.3679116706698777,
      "grad_norm": 0.14148421585559845,
      "learning_rate": 2.7462802091505452e-06,
      "loss": 0.0011,
      "step": 213280
    },
    {
      "epoch": 1.3679758075636639,
      "grad_norm": 0.11280106753110886,
      "learning_rate": 2.7457806022169253e-06,
      "loss": 0.0017,
      "step": 213290
    },
    {
      "epoch": 1.36803994445745,
      "grad_norm": 0.02165481075644493,
      "learning_rate": 2.745281023530287e-06,
      "loss": 0.001,
      "step": 213300
    },
    {
      "epoch": 1.3681040813512362,
      "grad_norm": 0.08553720265626907,
      "learning_rate": 2.7447814730968904e-06,
      "loss": 0.0014,
      "step": 213310
    },
    {
      "epoch": 1.368168218245022,
      "grad_norm": 0.10544779151678085,
      "learning_rate": 2.7442819509229924e-06,
      "loss": 0.0009,
      "step": 213320
    },
    {
      "epoch": 1.3682323551388083,
      "grad_norm": 0.051266372203826904,
      "learning_rate": 2.743782457014852e-06,
      "loss": 0.0004,
      "step": 213330
    },
    {
      "epoch": 1.3682964920325944,
      "grad_norm": 0.027301175519824028,
      "learning_rate": 2.7432829913787316e-06,
      "loss": 0.001,
      "step": 213340
    },
    {
      "epoch": 1.3683606289263803,
      "grad_norm": 0.11781658232212067,
      "learning_rate": 2.742783554020888e-06,
      "loss": 0.001,
      "step": 213350
    },
    {
      "epoch": 1.3684247658201665,
      "grad_norm": 0.1479956954717636,
      "learning_rate": 2.74228414494758e-06,
      "loss": 0.0031,
      "step": 213360
    },
    {
      "epoch": 1.3684889027139526,
      "grad_norm": 0.13904689252376556,
      "learning_rate": 2.7417847641650642e-06,
      "loss": 0.0008,
      "step": 213370
    },
    {
      "epoch": 1.3685530396077388,
      "grad_norm": 0.04022185504436493,
      "learning_rate": 2.7412854116795996e-06,
      "loss": 0.0008,
      "step": 213380
    },
    {
      "epoch": 1.368617176501525,
      "grad_norm": 0.04929453134536743,
      "learning_rate": 2.740786087497444e-06,
      "loss": 0.0014,
      "step": 213390
    },
    {
      "epoch": 1.368681313395311,
      "grad_norm": 0.16074256598949432,
      "learning_rate": 2.740286791624851e-06,
      "loss": 0.0017,
      "step": 213400
    },
    {
      "epoch": 1.368745450289097,
      "grad_norm": 0.02083122730255127,
      "learning_rate": 2.7397875240680816e-06,
      "loss": 0.0013,
      "step": 213410
    },
    {
      "epoch": 1.3688095871828831,
      "grad_norm": 0.0831194743514061,
      "learning_rate": 2.7392882848333902e-06,
      "loss": 0.0019,
      "step": 213420
    },
    {
      "epoch": 1.3688737240766693,
      "grad_norm": 0.08221425116062164,
      "learning_rate": 2.738789073927032e-06,
      "loss": 0.0022,
      "step": 213430
    },
    {
      "epoch": 1.3689378609704552,
      "grad_norm": 0.13809236884117126,
      "learning_rate": 2.738289891355261e-06,
      "loss": 0.0022,
      "step": 213440
    },
    {
      "epoch": 1.3690019978642414,
      "grad_norm": 0.04812110587954521,
      "learning_rate": 2.737790737124336e-06,
      "loss": 0.001,
      "step": 213450
    },
    {
      "epoch": 1.3690661347580275,
      "grad_norm": 0.0709865540266037,
      "learning_rate": 2.7372916112405097e-06,
      "loss": 0.0011,
      "step": 213460
    },
    {
      "epoch": 1.3691302716518137,
      "grad_norm": 0.11143574863672256,
      "learning_rate": 2.7367925137100364e-06,
      "loss": 0.0008,
      "step": 213470
    },
    {
      "epoch": 1.3691944085455998,
      "grad_norm": 0.21924534440040588,
      "learning_rate": 2.7362934445391686e-06,
      "loss": 0.0015,
      "step": 213480
    },
    {
      "epoch": 1.3692585454393857,
      "grad_norm": 0.11467649787664413,
      "learning_rate": 2.7357944037341644e-06,
      "loss": 0.0029,
      "step": 213490
    },
    {
      "epoch": 1.369322682333172,
      "grad_norm": 0.04351959004998207,
      "learning_rate": 2.7352953913012737e-06,
      "loss": 0.001,
      "step": 213500
    },
    {
      "epoch": 1.369386819226958,
      "grad_norm": 0.11486602574586868,
      "learning_rate": 2.7347964072467503e-06,
      "loss": 0.0014,
      "step": 213510
    },
    {
      "epoch": 1.3694509561207442,
      "grad_norm": 0.07183624058961868,
      "learning_rate": 2.734297451576845e-06,
      "loss": 0.0016,
      "step": 213520
    },
    {
      "epoch": 1.3695150930145301,
      "grad_norm": 0.07201708853244781,
      "learning_rate": 2.7337985242978144e-06,
      "loss": 0.0004,
      "step": 213530
    },
    {
      "epoch": 1.3695792299083163,
      "grad_norm": 0.1011878177523613,
      "learning_rate": 2.733299625415907e-06,
      "loss": 0.0012,
      "step": 213540
    },
    {
      "epoch": 1.3696433668021024,
      "grad_norm": 0.1852533221244812,
      "learning_rate": 2.732800754937374e-06,
      "loss": 0.0015,
      "step": 213550
    },
    {
      "epoch": 1.3697075036958886,
      "grad_norm": 0.14826343953609467,
      "learning_rate": 2.7323019128684703e-06,
      "loss": 0.0017,
      "step": 213560
    },
    {
      "epoch": 1.3697716405896747,
      "grad_norm": 0.03523099422454834,
      "learning_rate": 2.731803099215443e-06,
      "loss": 0.0005,
      "step": 213570
    },
    {
      "epoch": 1.3698357774834606,
      "grad_norm": 0.16956846415996552,
      "learning_rate": 2.7313043139845452e-06,
      "loss": 0.0013,
      "step": 213580
    },
    {
      "epoch": 1.3698999143772468,
      "grad_norm": 0.12136631458997726,
      "learning_rate": 2.730805557182023e-06,
      "loss": 0.0012,
      "step": 213590
    },
    {
      "epoch": 1.369964051271033,
      "grad_norm": 0.15683306753635406,
      "learning_rate": 2.730306828814131e-06,
      "loss": 0.0015,
      "step": 213600
    },
    {
      "epoch": 1.3700281881648189,
      "grad_norm": 0.00855253729969263,
      "learning_rate": 2.729808128887117e-06,
      "loss": 0.0028,
      "step": 213610
    },
    {
      "epoch": 1.370092325058605,
      "grad_norm": 0.09091190993785858,
      "learning_rate": 2.7293094574072295e-06,
      "loss": 0.0015,
      "step": 213620
    },
    {
      "epoch": 1.3701564619523912,
      "grad_norm": 0.06666356325149536,
      "learning_rate": 2.728810814380715e-06,
      "loss": 0.001,
      "step": 213630
    },
    {
      "epoch": 1.3702205988461773,
      "grad_norm": 0.14627714455127716,
      "learning_rate": 2.7283121998138264e-06,
      "loss": 0.0009,
      "step": 213640
    },
    {
      "epoch": 1.3702847357399635,
      "grad_norm": 0.13065119087696075,
      "learning_rate": 2.7278136137128097e-06,
      "loss": 0.0012,
      "step": 213650
    },
    {
      "epoch": 1.3703488726337496,
      "grad_norm": 0.013947075232863426,
      "learning_rate": 2.727315056083912e-06,
      "loss": 0.0016,
      "step": 213660
    },
    {
      "epoch": 1.3704130095275355,
      "grad_norm": 0.0493902862071991,
      "learning_rate": 2.726816526933379e-06,
      "loss": 0.0008,
      "step": 213670
    },
    {
      "epoch": 1.3704771464213217,
      "grad_norm": 0.01956510730087757,
      "learning_rate": 2.7263180262674617e-06,
      "loss": 0.001,
      "step": 213680
    },
    {
      "epoch": 1.3705412833151078,
      "grad_norm": 0.08916755765676498,
      "learning_rate": 2.725819554092404e-06,
      "loss": 0.0017,
      "step": 213690
    },
    {
      "epoch": 1.3706054202088938,
      "grad_norm": 0.15487140417099,
      "learning_rate": 2.7253211104144504e-06,
      "loss": 0.0016,
      "step": 213700
    },
    {
      "epoch": 1.37066955710268,
      "grad_norm": 0.017485613003373146,
      "learning_rate": 2.724822695239852e-06,
      "loss": 0.0017,
      "step": 213710
    },
    {
      "epoch": 1.370733693996466,
      "grad_norm": 0.11065894365310669,
      "learning_rate": 2.724324308574849e-06,
      "loss": 0.0012,
      "step": 213720
    },
    {
      "epoch": 1.3707978308902522,
      "grad_norm": 0.1534070074558258,
      "learning_rate": 2.7238259504256904e-06,
      "loss": 0.0015,
      "step": 213730
    },
    {
      "epoch": 1.3708619677840383,
      "grad_norm": 0.0838715210556984,
      "learning_rate": 2.7233276207986196e-06,
      "loss": 0.0021,
      "step": 213740
    },
    {
      "epoch": 1.3709261046778243,
      "grad_norm": 0.17326146364212036,
      "learning_rate": 2.722829319699879e-06,
      "loss": 0.0018,
      "step": 213750
    },
    {
      "epoch": 1.3709902415716104,
      "grad_norm": 0.07356669008731842,
      "learning_rate": 2.7223310471357165e-06,
      "loss": 0.001,
      "step": 213760
    },
    {
      "epoch": 1.3710543784653966,
      "grad_norm": 0.17165057361125946,
      "learning_rate": 2.721832803112373e-06,
      "loss": 0.0017,
      "step": 213770
    },
    {
      "epoch": 1.3711185153591827,
      "grad_norm": 0.11722222715616226,
      "learning_rate": 2.721334587636094e-06,
      "loss": 0.0022,
      "step": 213780
    },
    {
      "epoch": 1.3711826522529686,
      "grad_norm": 0.14171120524406433,
      "learning_rate": 2.720836400713118e-06,
      "loss": 0.0026,
      "step": 213790
    },
    {
      "epoch": 1.3712467891467548,
      "grad_norm": 0.19339559972286224,
      "learning_rate": 2.7203382423496938e-06,
      "loss": 0.001,
      "step": 213800
    },
    {
      "epoch": 1.371310926040541,
      "grad_norm": 0.06070531904697418,
      "learning_rate": 2.71984011255206e-06,
      "loss": 0.001,
      "step": 213810
    },
    {
      "epoch": 1.371375062934327,
      "grad_norm": 0.024490779265761375,
      "learning_rate": 2.7193420113264566e-06,
      "loss": 0.0009,
      "step": 213820
    },
    {
      "epoch": 1.3714391998281132,
      "grad_norm": 0.1451950967311859,
      "learning_rate": 2.718843938679131e-06,
      "loss": 0.0027,
      "step": 213830
    },
    {
      "epoch": 1.3715033367218992,
      "grad_norm": 0.039366308599710464,
      "learning_rate": 2.718345894616321e-06,
      "loss": 0.0017,
      "step": 213840
    },
    {
      "epoch": 1.3715674736156853,
      "grad_norm": 0.06168130040168762,
      "learning_rate": 2.717847879144267e-06,
      "loss": 0.0013,
      "step": 213850
    },
    {
      "epoch": 1.3716316105094715,
      "grad_norm": 0.01753639057278633,
      "learning_rate": 2.717349892269208e-06,
      "loss": 0.0022,
      "step": 213860
    },
    {
      "epoch": 1.3716957474032574,
      "grad_norm": 0.023625411093235016,
      "learning_rate": 2.7168519339973887e-06,
      "loss": 0.0009,
      "step": 213870
    },
    {
      "epoch": 1.3717598842970435,
      "grad_norm": 0.2518775165081024,
      "learning_rate": 2.716354004335047e-06,
      "loss": 0.0018,
      "step": 213880
    },
    {
      "epoch": 1.3718240211908297,
      "grad_norm": 0.05535493046045303,
      "learning_rate": 2.71585610328842e-06,
      "loss": 0.0008,
      "step": 213890
    },
    {
      "epoch": 1.3718881580846158,
      "grad_norm": 0.18779072165489197,
      "learning_rate": 2.7153582308637485e-06,
      "loss": 0.0018,
      "step": 213900
    },
    {
      "epoch": 1.371952294978402,
      "grad_norm": 0.06707113981246948,
      "learning_rate": 2.714860387067272e-06,
      "loss": 0.0004,
      "step": 213910
    },
    {
      "epoch": 1.3720164318721881,
      "grad_norm": 0.12983901798725128,
      "learning_rate": 2.714362571905228e-06,
      "loss": 0.0012,
      "step": 213920
    },
    {
      "epoch": 1.372080568765974,
      "grad_norm": 0.06924022734165192,
      "learning_rate": 2.7138647853838546e-06,
      "loss": 0.0013,
      "step": 213930
    },
    {
      "epoch": 1.3721447056597602,
      "grad_norm": 0.06443751603364944,
      "learning_rate": 2.713367027509387e-06,
      "loss": 0.0014,
      "step": 213940
    },
    {
      "epoch": 1.3722088425535464,
      "grad_norm": 0.05126974359154701,
      "learning_rate": 2.712869298288068e-06,
      "loss": 0.0024,
      "step": 213950
    },
    {
      "epoch": 1.3722729794473323,
      "grad_norm": 0.10546655207872391,
      "learning_rate": 2.71237159772613e-06,
      "loss": 0.0018,
      "step": 213960
    },
    {
      "epoch": 1.3723371163411184,
      "grad_norm": 0.18614305555820465,
      "learning_rate": 2.7118739258298094e-06,
      "loss": 0.001,
      "step": 213970
    },
    {
      "epoch": 1.3724012532349046,
      "grad_norm": 0.05279052257537842,
      "learning_rate": 2.711376282605346e-06,
      "loss": 0.0012,
      "step": 213980
    },
    {
      "epoch": 1.3724653901286907,
      "grad_norm": 0.005365374032407999,
      "learning_rate": 2.710878668058973e-06,
      "loss": 0.0015,
      "step": 213990
    },
    {
      "epoch": 1.3725295270224769,
      "grad_norm": 0.06920674443244934,
      "learning_rate": 2.7103810821969264e-06,
      "loss": 0.0012,
      "step": 214000
    },
    {
      "epoch": 1.3725936639162628,
      "grad_norm": 0.06458396464586258,
      "learning_rate": 2.709883525025439e-06,
      "loss": 0.0022,
      "step": 214010
    },
    {
      "epoch": 1.372657800810049,
      "grad_norm": 0.10044469684362411,
      "learning_rate": 2.7093859965507496e-06,
      "loss": 0.0017,
      "step": 214020
    },
    {
      "epoch": 1.372721937703835,
      "grad_norm": 0.1628868579864502,
      "learning_rate": 2.708888496779092e-06,
      "loss": 0.0017,
      "step": 214030
    },
    {
      "epoch": 1.372786074597621,
      "grad_norm": 0.07906507700681686,
      "learning_rate": 2.708391025716698e-06,
      "loss": 0.0011,
      "step": 214040
    },
    {
      "epoch": 1.3728502114914072,
      "grad_norm": 0.0659327283501625,
      "learning_rate": 2.7078935833698e-06,
      "loss": 0.0016,
      "step": 214050
    },
    {
      "epoch": 1.3729143483851933,
      "grad_norm": 0.08163490891456604,
      "learning_rate": 2.707396169744636e-06,
      "loss": 0.0013,
      "step": 214060
    },
    {
      "epoch": 1.3729784852789795,
      "grad_norm": 0.07006902247667313,
      "learning_rate": 2.706898784847436e-06,
      "loss": 0.0016,
      "step": 214070
    },
    {
      "epoch": 1.3730426221727656,
      "grad_norm": 0.12226539850234985,
      "learning_rate": 2.7064014286844337e-06,
      "loss": 0.0015,
      "step": 214080
    },
    {
      "epoch": 1.3731067590665518,
      "grad_norm": 0.07548625767230988,
      "learning_rate": 2.7059041012618583e-06,
      "loss": 0.0013,
      "step": 214090
    },
    {
      "epoch": 1.3731708959603377,
      "grad_norm": 0.016052765771746635,
      "learning_rate": 2.7054068025859463e-06,
      "loss": 0.0025,
      "step": 214100
    },
    {
      "epoch": 1.3732350328541238,
      "grad_norm": 0.009808254428207874,
      "learning_rate": 2.704909532662927e-06,
      "loss": 0.0013,
      "step": 214110
    },
    {
      "epoch": 1.37329916974791,
      "grad_norm": 0.16814613342285156,
      "learning_rate": 2.7044122914990296e-06,
      "loss": 0.0012,
      "step": 214120
    },
    {
      "epoch": 1.373363306641696,
      "grad_norm": 0.2443917989730835,
      "learning_rate": 2.703915079100489e-06,
      "loss": 0.0014,
      "step": 214130
    },
    {
      "epoch": 1.373427443535482,
      "grad_norm": 0.10541815310716629,
      "learning_rate": 2.7034178954735325e-06,
      "loss": 0.0012,
      "step": 214140
    },
    {
      "epoch": 1.3734915804292682,
      "grad_norm": 0.03079073876142502,
      "learning_rate": 2.702920740624392e-06,
      "loss": 0.0008,
      "step": 214150
    },
    {
      "epoch": 1.3735557173230544,
      "grad_norm": 0.05303191766142845,
      "learning_rate": 2.7024236145592942e-06,
      "loss": 0.0016,
      "step": 214160
    },
    {
      "epoch": 1.3736198542168405,
      "grad_norm": 0.03506298363208771,
      "learning_rate": 2.701926517284472e-06,
      "loss": 0.0012,
      "step": 214170
    },
    {
      "epoch": 1.3736839911106264,
      "grad_norm": 0.043985120952129364,
      "learning_rate": 2.701429448806153e-06,
      "loss": 0.0027,
      "step": 214180
    },
    {
      "epoch": 1.3737481280044126,
      "grad_norm": 0.11024189740419388,
      "learning_rate": 2.7009324091305667e-06,
      "loss": 0.0018,
      "step": 214190
    },
    {
      "epoch": 1.3738122648981987,
      "grad_norm": 0.006982483901083469,
      "learning_rate": 2.7004353982639375e-06,
      "loss": 0.0009,
      "step": 214200
    },
    {
      "epoch": 1.373876401791985,
      "grad_norm": 0.02276814542710781,
      "learning_rate": 2.6999384162124987e-06,
      "loss": 0.0012,
      "step": 214210
    },
    {
      "epoch": 1.3739405386857708,
      "grad_norm": 0.1164424791932106,
      "learning_rate": 2.699441462982475e-06,
      "loss": 0.0009,
      "step": 214220
    },
    {
      "epoch": 1.374004675579557,
      "grad_norm": 0.09967939555644989,
      "learning_rate": 2.698944538580095e-06,
      "loss": 0.0007,
      "step": 214230
    },
    {
      "epoch": 1.3740688124733431,
      "grad_norm": 0.10169283300638199,
      "learning_rate": 2.6984476430115815e-06,
      "loss": 0.001,
      "step": 214240
    },
    {
      "epoch": 1.3741329493671293,
      "grad_norm": 0.202654629945755,
      "learning_rate": 2.6979507762831657e-06,
      "loss": 0.0013,
      "step": 214250
    },
    {
      "epoch": 1.3741970862609154,
      "grad_norm": 0.17559348046779633,
      "learning_rate": 2.6974539384010722e-06,
      "loss": 0.0014,
      "step": 214260
    },
    {
      "epoch": 1.3742612231547013,
      "grad_norm": 0.11496882885694504,
      "learning_rate": 2.6969571293715247e-06,
      "loss": 0.0018,
      "step": 214270
    },
    {
      "epoch": 1.3743253600484875,
      "grad_norm": 0.14585083723068237,
      "learning_rate": 2.6964603492007523e-06,
      "loss": 0.0011,
      "step": 214280
    },
    {
      "epoch": 1.3743894969422736,
      "grad_norm": 0.15473544597625732,
      "learning_rate": 2.6959635978949785e-06,
      "loss": 0.0013,
      "step": 214290
    },
    {
      "epoch": 1.3744536338360596,
      "grad_norm": 0.006460663862526417,
      "learning_rate": 2.6954668754604264e-06,
      "loss": 0.0008,
      "step": 214300
    },
    {
      "epoch": 1.3745177707298457,
      "grad_norm": 0.1418498307466507,
      "learning_rate": 2.69497018190332e-06,
      "loss": 0.002,
      "step": 214310
    },
    {
      "epoch": 1.3745819076236319,
      "grad_norm": 0.06111471354961395,
      "learning_rate": 2.694473517229886e-06,
      "loss": 0.0011,
      "step": 214320
    },
    {
      "epoch": 1.374646044517418,
      "grad_norm": 0.07857625931501389,
      "learning_rate": 2.6939768814463472e-06,
      "loss": 0.0013,
      "step": 214330
    },
    {
      "epoch": 1.3747101814112042,
      "grad_norm": 0.041253603994846344,
      "learning_rate": 2.6934802745589265e-06,
      "loss": 0.0014,
      "step": 214340
    },
    {
      "epoch": 1.3747743183049903,
      "grad_norm": 0.0759199857711792,
      "learning_rate": 2.692983696573844e-06,
      "loss": 0.0023,
      "step": 214350
    },
    {
      "epoch": 1.3748384551987762,
      "grad_norm": 0.01071109902113676,
      "learning_rate": 2.692487147497327e-06,
      "loss": 0.0008,
      "step": 214360
    },
    {
      "epoch": 1.3749025920925624,
      "grad_norm": 0.13840435445308685,
      "learning_rate": 2.6919906273355947e-06,
      "loss": 0.0016,
      "step": 214370
    },
    {
      "epoch": 1.3749667289863485,
      "grad_norm": 0.11041421443223953,
      "learning_rate": 2.691494136094869e-06,
      "loss": 0.0013,
      "step": 214380
    },
    {
      "epoch": 1.3750308658801345,
      "grad_norm": 0.08328073471784592,
      "learning_rate": 2.69099767378137e-06,
      "loss": 0.001,
      "step": 214390
    },
    {
      "epoch": 1.3750950027739206,
      "grad_norm": 0.11724260449409485,
      "learning_rate": 2.6905012404013227e-06,
      "loss": 0.0015,
      "step": 214400
    },
    {
      "epoch": 1.3751591396677068,
      "grad_norm": 0.005760582629591227,
      "learning_rate": 2.690004835960945e-06,
      "loss": 0.0007,
      "step": 214410
    },
    {
      "epoch": 1.375223276561493,
      "grad_norm": 0.032109711319208145,
      "learning_rate": 2.6895084604664556e-06,
      "loss": 0.0013,
      "step": 214420
    },
    {
      "epoch": 1.375287413455279,
      "grad_norm": 0.06785446405410767,
      "learning_rate": 2.689012113924079e-06,
      "loss": 0.0019,
      "step": 214430
    },
    {
      "epoch": 1.375351550349065,
      "grad_norm": 0.001295348978601396,
      "learning_rate": 2.6885157963400315e-06,
      "loss": 0.0005,
      "step": 214440
    },
    {
      "epoch": 1.3754156872428511,
      "grad_norm": 0.007197576574981213,
      "learning_rate": 2.6880195077205333e-06,
      "loss": 0.0028,
      "step": 214450
    },
    {
      "epoch": 1.3754798241366373,
      "grad_norm": 0.0747002586722374,
      "learning_rate": 2.6875232480718014e-06,
      "loss": 0.0007,
      "step": 214460
    },
    {
      "epoch": 1.3755439610304232,
      "grad_norm": 0.08409067243337631,
      "learning_rate": 2.687027017400058e-06,
      "loss": 0.0013,
      "step": 214470
    },
    {
      "epoch": 1.3756080979242093,
      "grad_norm": 0.06002112478017807,
      "learning_rate": 2.6865308157115167e-06,
      "loss": 0.0012,
      "step": 214480
    },
    {
      "epoch": 1.3756722348179955,
      "grad_norm": 0.17536841332912445,
      "learning_rate": 2.6860346430123995e-06,
      "loss": 0.0023,
      "step": 214490
    },
    {
      "epoch": 1.3757363717117816,
      "grad_norm": 0.04730650782585144,
      "learning_rate": 2.685538499308922e-06,
      "loss": 0.0016,
      "step": 214500
    },
    {
      "epoch": 1.3758005086055678,
      "grad_norm": 0.03322217985987663,
      "learning_rate": 2.6850423846073e-06,
      "loss": 0.0015,
      "step": 214510
    },
    {
      "epoch": 1.375864645499354,
      "grad_norm": 0.03648802265524864,
      "learning_rate": 2.6845462989137528e-06,
      "loss": 0.0009,
      "step": 214520
    },
    {
      "epoch": 1.3759287823931399,
      "grad_norm": 0.04079330340027809,
      "learning_rate": 2.6840502422344955e-06,
      "loss": 0.0014,
      "step": 214530
    },
    {
      "epoch": 1.375992919286926,
      "grad_norm": 0.0969911515712738,
      "learning_rate": 2.683554214575742e-06,
      "loss": 0.002,
      "step": 214540
    },
    {
      "epoch": 1.3760570561807122,
      "grad_norm": 0.29174312949180603,
      "learning_rate": 2.683058215943711e-06,
      "loss": 0.0029,
      "step": 214550
    },
    {
      "epoch": 1.376121193074498,
      "grad_norm": 0.026156960055232048,
      "learning_rate": 2.682562246344617e-06,
      "loss": 0.0014,
      "step": 214560
    },
    {
      "epoch": 1.3761853299682842,
      "grad_norm": 0.0699620246887207,
      "learning_rate": 2.682066305784674e-06,
      "loss": 0.0022,
      "step": 214570
    },
    {
      "epoch": 1.3762494668620704,
      "grad_norm": 0.06607739627361298,
      "learning_rate": 2.6815703942700954e-06,
      "loss": 0.0009,
      "step": 214580
    },
    {
      "epoch": 1.3763136037558565,
      "grad_norm": 0.021971357986330986,
      "learning_rate": 2.681074511807098e-06,
      "loss": 0.0018,
      "step": 214590
    },
    {
      "epoch": 1.3763777406496427,
      "grad_norm": 0.062438640743494034,
      "learning_rate": 2.680578658401895e-06,
      "loss": 0.001,
      "step": 214600
    },
    {
      "epoch": 1.3764418775434286,
      "grad_norm": 0.11185724288225174,
      "learning_rate": 2.680082834060699e-06,
      "loss": 0.0011,
      "step": 214610
    },
    {
      "epoch": 1.3765060144372148,
      "grad_norm": 0.00310879317112267,
      "learning_rate": 2.6795870387897205e-06,
      "loss": 0.0008,
      "step": 214620
    },
    {
      "epoch": 1.376570151331001,
      "grad_norm": 0.06282266974449158,
      "learning_rate": 2.6790912725951766e-06,
      "loss": 0.0008,
      "step": 214630
    },
    {
      "epoch": 1.376634288224787,
      "grad_norm": 0.30241039395332336,
      "learning_rate": 2.678595535483278e-06,
      "loss": 0.0026,
      "step": 214640
    },
    {
      "epoch": 1.376698425118573,
      "grad_norm": 0.035346679389476776,
      "learning_rate": 2.6780998274602376e-06,
      "loss": 0.0022,
      "step": 214650
    },
    {
      "epoch": 1.3767625620123591,
      "grad_norm": 0.0476689487695694,
      "learning_rate": 2.677604148532262e-06,
      "loss": 0.0019,
      "step": 214660
    },
    {
      "epoch": 1.3768266989061453,
      "grad_norm": 0.1020970419049263,
      "learning_rate": 2.6771084987055694e-06,
      "loss": 0.0013,
      "step": 214670
    },
    {
      "epoch": 1.3768908357999314,
      "grad_norm": 0.18499650061130524,
      "learning_rate": 2.6766128779863677e-06,
      "loss": 0.0018,
      "step": 214680
    },
    {
      "epoch": 1.3769549726937176,
      "grad_norm": 0.028281651437282562,
      "learning_rate": 2.6761172863808645e-06,
      "loss": 0.0023,
      "step": 214690
    },
    {
      "epoch": 1.3770191095875035,
      "grad_norm": 0.047749750316143036,
      "learning_rate": 2.675621723895274e-06,
      "loss": 0.0008,
      "step": 214700
    },
    {
      "epoch": 1.3770832464812897,
      "grad_norm": 0.12156499922275543,
      "learning_rate": 2.6751261905358055e-06,
      "loss": 0.0015,
      "step": 214710
    },
    {
      "epoch": 1.3771473833750758,
      "grad_norm": 0.06353110820055008,
      "learning_rate": 2.6746306863086668e-06,
      "loss": 0.0007,
      "step": 214720
    },
    {
      "epoch": 1.3772115202688617,
      "grad_norm": 0.09507710486650467,
      "learning_rate": 2.674135211220066e-06,
      "loss": 0.0014,
      "step": 214730
    },
    {
      "epoch": 1.3772756571626479,
      "grad_norm": 0.18407589197158813,
      "learning_rate": 2.673639765276215e-06,
      "loss": 0.0021,
      "step": 214740
    },
    {
      "epoch": 1.377339794056434,
      "grad_norm": 0.0212041474878788,
      "learning_rate": 2.673144348483321e-06,
      "loss": 0.0016,
      "step": 214750
    },
    {
      "epoch": 1.3774039309502202,
      "grad_norm": 0.1254289299249649,
      "learning_rate": 2.6726489608475905e-06,
      "loss": 0.0022,
      "step": 214760
    },
    {
      "epoch": 1.3774680678440063,
      "grad_norm": 0.2893882989883423,
      "learning_rate": 2.6721536023752297e-06,
      "loss": 0.0015,
      "step": 214770
    },
    {
      "epoch": 1.3775322047377925,
      "grad_norm": 0.0767577588558197,
      "learning_rate": 2.6716582730724505e-06,
      "loss": 0.0007,
      "step": 214780
    },
    {
      "epoch": 1.3775963416315784,
      "grad_norm": 0.02579805813729763,
      "learning_rate": 2.6711629729454568e-06,
      "loss": 0.0007,
      "step": 214790
    },
    {
      "epoch": 1.3776604785253646,
      "grad_norm": 0.05162614583969116,
      "learning_rate": 2.6706677020004558e-06,
      "loss": 0.0025,
      "step": 214800
    },
    {
      "epoch": 1.3777246154191507,
      "grad_norm": 0.04368935525417328,
      "learning_rate": 2.6701724602436506e-06,
      "loss": 0.0019,
      "step": 214810
    },
    {
      "epoch": 1.3777887523129366,
      "grad_norm": 0.14431016147136688,
      "learning_rate": 2.6696772476812527e-06,
      "loss": 0.0012,
      "step": 214820
    },
    {
      "epoch": 1.3778528892067228,
      "grad_norm": 0.05523088201880455,
      "learning_rate": 2.669182064319463e-06,
      "loss": 0.001,
      "step": 214830
    },
    {
      "epoch": 1.377917026100509,
      "grad_norm": 0.06342064589262009,
      "learning_rate": 2.6686869101644867e-06,
      "loss": 0.0013,
      "step": 214840
    },
    {
      "epoch": 1.377981162994295,
      "grad_norm": 0.08518742024898529,
      "learning_rate": 2.6681917852225315e-06,
      "loss": 0.0016,
      "step": 214850
    },
    {
      "epoch": 1.3780452998880812,
      "grad_norm": 0.01703224517405033,
      "learning_rate": 2.6676966894998003e-06,
      "loss": 0.0012,
      "step": 214860
    },
    {
      "epoch": 1.3781094367818671,
      "grad_norm": 0.10826463997364044,
      "learning_rate": 2.6672016230024955e-06,
      "loss": 0.001,
      "step": 214870
    },
    {
      "epoch": 1.3781735736756533,
      "grad_norm": 0.1422380805015564,
      "learning_rate": 2.6667065857368203e-06,
      "loss": 0.002,
      "step": 214880
    },
    {
      "epoch": 1.3782377105694394,
      "grad_norm": 0.10455601662397385,
      "learning_rate": 2.6662115777089805e-06,
      "loss": 0.0014,
      "step": 214890
    },
    {
      "epoch": 1.3783018474632254,
      "grad_norm": 0.026486804708838463,
      "learning_rate": 2.6657165989251777e-06,
      "loss": 0.0009,
      "step": 214900
    },
    {
      "epoch": 1.3783659843570115,
      "grad_norm": 0.12323911488056183,
      "learning_rate": 2.665221649391614e-06,
      "loss": 0.0025,
      "step": 214910
    },
    {
      "epoch": 1.3784301212507977,
      "grad_norm": 0.08645933866500854,
      "learning_rate": 2.66472672911449e-06,
      "loss": 0.0019,
      "step": 214920
    },
    {
      "epoch": 1.3784942581445838,
      "grad_norm": 0.004943096078932285,
      "learning_rate": 2.664231838100011e-06,
      "loss": 0.0014,
      "step": 214930
    },
    {
      "epoch": 1.37855839503837,
      "grad_norm": 0.006645245011895895,
      "learning_rate": 2.663736976354377e-06,
      "loss": 0.0008,
      "step": 214940
    },
    {
      "epoch": 1.3786225319321561,
      "grad_norm": 0.08196337521076202,
      "learning_rate": 2.663242143883786e-06,
      "loss": 0.0011,
      "step": 214950
    },
    {
      "epoch": 1.378686668825942,
      "grad_norm": 0.04176272079348564,
      "learning_rate": 2.6627473406944426e-06,
      "loss": 0.0007,
      "step": 214960
    },
    {
      "epoch": 1.3787508057197282,
      "grad_norm": 0.02642265520989895,
      "learning_rate": 2.662252566792546e-06,
      "loss": 0.0014,
      "step": 214970
    },
    {
      "epoch": 1.3788149426135143,
      "grad_norm": 0.07213565707206726,
      "learning_rate": 2.661757822184295e-06,
      "loss": 0.0014,
      "step": 214980
    },
    {
      "epoch": 1.3788790795073003,
      "grad_norm": 0.02878113090991974,
      "learning_rate": 2.6612631068758876e-06,
      "loss": 0.0008,
      "step": 214990
    },
    {
      "epoch": 1.3789432164010864,
      "grad_norm": 0.10633989423513412,
      "learning_rate": 2.6607684208735275e-06,
      "loss": 0.0007,
      "step": 215000
    },
    {
      "epoch": 1.3790073532948726,
      "grad_norm": 0.05150622874498367,
      "learning_rate": 2.6602737641834105e-06,
      "loss": 0.0021,
      "step": 215010
    },
    {
      "epoch": 1.3790714901886587,
      "grad_norm": 0.05158638954162598,
      "learning_rate": 2.6597791368117354e-06,
      "loss": 0.0013,
      "step": 215020
    },
    {
      "epoch": 1.3791356270824449,
      "grad_norm": 0.07146673649549484,
      "learning_rate": 2.659284538764698e-06,
      "loss": 0.0005,
      "step": 215030
    },
    {
      "epoch": 1.3791997639762308,
      "grad_norm": 0.2744249105453491,
      "learning_rate": 2.6587899700485008e-06,
      "loss": 0.0029,
      "step": 215040
    },
    {
      "epoch": 1.379263900870017,
      "grad_norm": 0.14671044051647186,
      "learning_rate": 2.6582954306693383e-06,
      "loss": 0.0013,
      "step": 215050
    },
    {
      "epoch": 1.379328037763803,
      "grad_norm": 0.19606345891952515,
      "learning_rate": 2.6578009206334076e-06,
      "loss": 0.0015,
      "step": 215060
    },
    {
      "epoch": 1.3793921746575892,
      "grad_norm": 0.07785133272409439,
      "learning_rate": 2.657306439946903e-06,
      "loss": 0.003,
      "step": 215070
    },
    {
      "epoch": 1.3794563115513752,
      "grad_norm": 0.017010271549224854,
      "learning_rate": 2.6568119886160255e-06,
      "loss": 0.0012,
      "step": 215080
    },
    {
      "epoch": 1.3795204484451613,
      "grad_norm": 0.047582823783159256,
      "learning_rate": 2.6563175666469683e-06,
      "loss": 0.0015,
      "step": 215090
    },
    {
      "epoch": 1.3795845853389475,
      "grad_norm": 0.09325523674488068,
      "learning_rate": 2.655823174045925e-06,
      "loss": 0.0015,
      "step": 215100
    },
    {
      "epoch": 1.3796487222327336,
      "grad_norm": 0.10350356996059418,
      "learning_rate": 2.6553288108190943e-06,
      "loss": 0.001,
      "step": 215110
    },
    {
      "epoch": 1.3797128591265198,
      "grad_norm": 0.03015591949224472,
      "learning_rate": 2.6548344769726696e-06,
      "loss": 0.0017,
      "step": 215120
    },
    {
      "epoch": 1.3797769960203057,
      "grad_norm": 0.0348021537065506,
      "learning_rate": 2.654340172512845e-06,
      "loss": 0.0008,
      "step": 215130
    },
    {
      "epoch": 1.3798411329140918,
      "grad_norm": 0.04149578511714935,
      "learning_rate": 2.6538458974458128e-06,
      "loss": 0.0015,
      "step": 215140
    },
    {
      "epoch": 1.379905269807878,
      "grad_norm": 0.04008261114358902,
      "learning_rate": 2.65335165177777e-06,
      "loss": 0.0008,
      "step": 215150
    },
    {
      "epoch": 1.379969406701664,
      "grad_norm": 0.026236629113554955,
      "learning_rate": 2.652857435514908e-06,
      "loss": 0.001,
      "step": 215160
    },
    {
      "epoch": 1.38003354359545,
      "grad_norm": 0.15502652525901794,
      "learning_rate": 2.65236324866342e-06,
      "loss": 0.0012,
      "step": 215170
    },
    {
      "epoch": 1.3800976804892362,
      "grad_norm": 0.0263929795473814,
      "learning_rate": 2.6518690912294957e-06,
      "loss": 0.0016,
      "step": 215180
    },
    {
      "epoch": 1.3801618173830223,
      "grad_norm": 0.16136017441749573,
      "learning_rate": 2.651374963219333e-06,
      "loss": 0.0017,
      "step": 215190
    },
    {
      "epoch": 1.3802259542768085,
      "grad_norm": 0.10542784631252289,
      "learning_rate": 2.6508808646391193e-06,
      "loss": 0.0024,
      "step": 215200
    },
    {
      "epoch": 1.3802900911705946,
      "grad_norm": 0.14071311056613922,
      "learning_rate": 2.650386795495048e-06,
      "loss": 0.0021,
      "step": 215210
    },
    {
      "epoch": 1.3803542280643806,
      "grad_norm": 0.16117265820503235,
      "learning_rate": 2.6498927557933072e-06,
      "loss": 0.0012,
      "step": 215220
    },
    {
      "epoch": 1.3804183649581667,
      "grad_norm": 0.029635494574904442,
      "learning_rate": 2.6493987455400923e-06,
      "loss": 0.0014,
      "step": 215230
    },
    {
      "epoch": 1.3804825018519529,
      "grad_norm": 0.11301979422569275,
      "learning_rate": 2.64890476474159e-06,
      "loss": 0.0023,
      "step": 215240
    },
    {
      "epoch": 1.3805466387457388,
      "grad_norm": 0.03197695314884186,
      "learning_rate": 2.6484108134039903e-06,
      "loss": 0.0016,
      "step": 215250
    },
    {
      "epoch": 1.380610775639525,
      "grad_norm": 0.05557979643344879,
      "learning_rate": 2.6479168915334834e-06,
      "loss": 0.0029,
      "step": 215260
    },
    {
      "epoch": 1.380674912533311,
      "grad_norm": 0.13175056874752045,
      "learning_rate": 2.647422999136261e-06,
      "loss": 0.0018,
      "step": 215270
    },
    {
      "epoch": 1.3807390494270972,
      "grad_norm": 0.06291336566209793,
      "learning_rate": 2.646929136218509e-06,
      "loss": 0.0015,
      "step": 215280
    },
    {
      "epoch": 1.3808031863208834,
      "grad_norm": 0.12071221321821213,
      "learning_rate": 2.6464353027864177e-06,
      "loss": 0.0018,
      "step": 215290
    },
    {
      "epoch": 1.3808673232146693,
      "grad_norm": 0.09303078055381775,
      "learning_rate": 2.645941498846172e-06,
      "loss": 0.0012,
      "step": 215300
    },
    {
      "epoch": 1.3809314601084555,
      "grad_norm": 0.03669220209121704,
      "learning_rate": 2.645447724403963e-06,
      "loss": 0.0018,
      "step": 215310
    },
    {
      "epoch": 1.3809955970022416,
      "grad_norm": 0.16656337678432465,
      "learning_rate": 2.6449539794659775e-06,
      "loss": 0.0038,
      "step": 215320
    },
    {
      "epoch": 1.3810597338960278,
      "grad_norm": 0.08406962454319,
      "learning_rate": 2.6444602640384008e-06,
      "loss": 0.0009,
      "step": 215330
    },
    {
      "epoch": 1.3811238707898137,
      "grad_norm": 0.043212637305259705,
      "learning_rate": 2.6439665781274193e-06,
      "loss": 0.0014,
      "step": 215340
    },
    {
      "epoch": 1.3811880076835998,
      "grad_norm": 0.025267785415053368,
      "learning_rate": 2.6434729217392223e-06,
      "loss": 0.0022,
      "step": 215350
    },
    {
      "epoch": 1.381252144577386,
      "grad_norm": 0.15569503605365753,
      "learning_rate": 2.6429792948799935e-06,
      "loss": 0.0012,
      "step": 215360
    },
    {
      "epoch": 1.3813162814711721,
      "grad_norm": 0.07439172267913818,
      "learning_rate": 2.6424856975559177e-06,
      "loss": 0.0011,
      "step": 215370
    },
    {
      "epoch": 1.3813804183649583,
      "grad_norm": 0.15520213544368744,
      "learning_rate": 2.6419921297731823e-06,
      "loss": 0.0012,
      "step": 215380
    },
    {
      "epoch": 1.3814445552587442,
      "grad_norm": 0.012888588942587376,
      "learning_rate": 2.64149859153797e-06,
      "loss": 0.0012,
      "step": 215390
    },
    {
      "epoch": 1.3815086921525304,
      "grad_norm": 0.05528208240866661,
      "learning_rate": 2.641005082856467e-06,
      "loss": 0.0022,
      "step": 215400
    },
    {
      "epoch": 1.3815728290463165,
      "grad_norm": 0.13787513971328735,
      "learning_rate": 2.6405116037348534e-06,
      "loss": 0.0011,
      "step": 215410
    },
    {
      "epoch": 1.3816369659401024,
      "grad_norm": 0.14957444369792938,
      "learning_rate": 2.640018154179318e-06,
      "loss": 0.0014,
      "step": 215420
    },
    {
      "epoch": 1.3817011028338886,
      "grad_norm": 0.06772604584693909,
      "learning_rate": 2.6395247341960424e-06,
      "loss": 0.0007,
      "step": 215430
    },
    {
      "epoch": 1.3817652397276747,
      "grad_norm": 0.05124377831816673,
      "learning_rate": 2.6390313437912083e-06,
      "loss": 0.0015,
      "step": 215440
    },
    {
      "epoch": 1.3818293766214609,
      "grad_norm": 0.13782000541687012,
      "learning_rate": 2.638537982970997e-06,
      "loss": 0.0012,
      "step": 215450
    },
    {
      "epoch": 1.381893513515247,
      "grad_norm": 0.06755907833576202,
      "learning_rate": 2.638044651741594e-06,
      "loss": 0.002,
      "step": 215460
    },
    {
      "epoch": 1.3819576504090332,
      "grad_norm": 0.0827302485704422,
      "learning_rate": 2.6375513501091803e-06,
      "loss": 0.0008,
      "step": 215470
    },
    {
      "epoch": 1.382021787302819,
      "grad_norm": 0.26965776085853577,
      "learning_rate": 2.637058078079936e-06,
      "loss": 0.0017,
      "step": 215480
    },
    {
      "epoch": 1.3820859241966053,
      "grad_norm": 0.04469255357980728,
      "learning_rate": 2.6365648356600414e-06,
      "loss": 0.0011,
      "step": 215490
    },
    {
      "epoch": 1.3821500610903914,
      "grad_norm": 0.2724514305591583,
      "learning_rate": 2.6360716228556805e-06,
      "loss": 0.0014,
      "step": 215500
    },
    {
      "epoch": 1.3822141979841773,
      "grad_norm": 0.05116065591573715,
      "learning_rate": 2.6355784396730313e-06,
      "loss": 0.0015,
      "step": 215510
    },
    {
      "epoch": 1.3822783348779635,
      "grad_norm": 0.0161726213991642,
      "learning_rate": 2.635085286118272e-06,
      "loss": 0.0008,
      "step": 215520
    },
    {
      "epoch": 1.3823424717717496,
      "grad_norm": 0.0626625269651413,
      "learning_rate": 2.634592162197586e-06,
      "loss": 0.0015,
      "step": 215530
    },
    {
      "epoch": 1.3824066086655358,
      "grad_norm": 0.2643812596797943,
      "learning_rate": 2.6340990679171508e-06,
      "loss": 0.0021,
      "step": 215540
    },
    {
      "epoch": 1.382470745559322,
      "grad_norm": 0.3324837386608124,
      "learning_rate": 2.6336060032831457e-06,
      "loss": 0.0028,
      "step": 215550
    },
    {
      "epoch": 1.3825348824531078,
      "grad_norm": 0.11031866818666458,
      "learning_rate": 2.6331129683017465e-06,
      "loss": 0.001,
      "step": 215560
    },
    {
      "epoch": 1.382599019346894,
      "grad_norm": 0.06517376005649567,
      "learning_rate": 2.6326199629791347e-06,
      "loss": 0.0011,
      "step": 215570
    },
    {
      "epoch": 1.3826631562406801,
      "grad_norm": 0.060064904391765594,
      "learning_rate": 2.632126987321488e-06,
      "loss": 0.0021,
      "step": 215580
    },
    {
      "epoch": 1.382727293134466,
      "grad_norm": 0.05003609135746956,
      "learning_rate": 2.6316340413349817e-06,
      "loss": 0.0011,
      "step": 215590
    },
    {
      "epoch": 1.3827914300282522,
      "grad_norm": 0.01165304146707058,
      "learning_rate": 2.6311411250257914e-06,
      "loss": 0.0009,
      "step": 215600
    },
    {
      "epoch": 1.3828555669220384,
      "grad_norm": 0.0635843276977539,
      "learning_rate": 2.6306482384000974e-06,
      "loss": 0.0027,
      "step": 215610
    },
    {
      "epoch": 1.3829197038158245,
      "grad_norm": 0.023913035169243813,
      "learning_rate": 2.630155381464075e-06,
      "loss": 0.0012,
      "step": 215620
    },
    {
      "epoch": 1.3829838407096107,
      "grad_norm": 0.029272524639964104,
      "learning_rate": 2.6296625542238995e-06,
      "loss": 0.0021,
      "step": 215630
    },
    {
      "epoch": 1.3830479776033968,
      "grad_norm": 0.002349895890802145,
      "learning_rate": 2.6291697566857436e-06,
      "loss": 0.0009,
      "step": 215640
    },
    {
      "epoch": 1.3831121144971827,
      "grad_norm": 0.038314275443553925,
      "learning_rate": 2.6286769888557878e-06,
      "loss": 0.002,
      "step": 215650
    },
    {
      "epoch": 1.383176251390969,
      "grad_norm": 0.04881530627608299,
      "learning_rate": 2.6281842507402034e-06,
      "loss": 0.0018,
      "step": 215660
    },
    {
      "epoch": 1.383240388284755,
      "grad_norm": 0.04537365213036537,
      "learning_rate": 2.6276915423451645e-06,
      "loss": 0.0006,
      "step": 215670
    },
    {
      "epoch": 1.383304525178541,
      "grad_norm": 0.07577371597290039,
      "learning_rate": 2.6271988636768465e-06,
      "loss": 0.0017,
      "step": 215680
    },
    {
      "epoch": 1.3833686620723271,
      "grad_norm": 0.18578864634037018,
      "learning_rate": 2.6267062147414237e-06,
      "loss": 0.0019,
      "step": 215690
    },
    {
      "epoch": 1.3834327989661133,
      "grad_norm": 0.06750933080911636,
      "learning_rate": 2.626213595545067e-06,
      "loss": 0.0014,
      "step": 215700
    },
    {
      "epoch": 1.3834969358598994,
      "grad_norm": 0.052284397184848785,
      "learning_rate": 2.62572100609395e-06,
      "loss": 0.0012,
      "step": 215710
    },
    {
      "epoch": 1.3835610727536856,
      "grad_norm": 0.12799391150474548,
      "learning_rate": 2.6252284463942468e-06,
      "loss": 0.0025,
      "step": 215720
    },
    {
      "epoch": 1.3836252096474715,
      "grad_norm": 0.11345848441123962,
      "learning_rate": 2.624735916452129e-06,
      "loss": 0.0012,
      "step": 215730
    },
    {
      "epoch": 1.3836893465412576,
      "grad_norm": 0.06445007771253586,
      "learning_rate": 2.624243416273767e-06,
      "loss": 0.0017,
      "step": 215740
    },
    {
      "epoch": 1.3837534834350438,
      "grad_norm": 0.07878265529870987,
      "learning_rate": 2.623750945865331e-06,
      "loss": 0.0019,
      "step": 215750
    },
    {
      "epoch": 1.38381762032883,
      "grad_norm": 0.13059911131858826,
      "learning_rate": 2.6232585052329964e-06,
      "loss": 0.0014,
      "step": 215760
    },
    {
      "epoch": 1.3838817572226159,
      "grad_norm": 0.09587746113538742,
      "learning_rate": 2.6227660943829315e-06,
      "loss": 0.001,
      "step": 215770
    },
    {
      "epoch": 1.383945894116402,
      "grad_norm": 0.1065930724143982,
      "learning_rate": 2.622273713321306e-06,
      "loss": 0.0028,
      "step": 215780
    },
    {
      "epoch": 1.3840100310101882,
      "grad_norm": 0.062053028494119644,
      "learning_rate": 2.621781362054288e-06,
      "loss": 0.0014,
      "step": 215790
    },
    {
      "epoch": 1.3840741679039743,
      "grad_norm": 0.05019015073776245,
      "learning_rate": 2.6212890405880516e-06,
      "loss": 0.0016,
      "step": 215800
    },
    {
      "epoch": 1.3841383047977605,
      "grad_norm": 0.10837151855230331,
      "learning_rate": 2.620796748928763e-06,
      "loss": 0.0019,
      "step": 215810
    },
    {
      "epoch": 1.3842024416915464,
      "grad_norm": 0.04327791556715965,
      "learning_rate": 2.62030448708259e-06,
      "loss": 0.0013,
      "step": 215820
    },
    {
      "epoch": 1.3842665785853325,
      "grad_norm": 0.14705218374729156,
      "learning_rate": 2.6198122550557047e-06,
      "loss": 0.0013,
      "step": 215830
    },
    {
      "epoch": 1.3843307154791187,
      "grad_norm": 0.062407251447439194,
      "learning_rate": 2.619320052854273e-06,
      "loss": 0.0017,
      "step": 215840
    },
    {
      "epoch": 1.3843948523729046,
      "grad_norm": 0.0543547049164772,
      "learning_rate": 2.618827880484462e-06,
      "loss": 0.0021,
      "step": 215850
    },
    {
      "epoch": 1.3844589892666908,
      "grad_norm": 0.05428921431303024,
      "learning_rate": 2.6183357379524376e-06,
      "loss": 0.0013,
      "step": 215860
    },
    {
      "epoch": 1.384523126160477,
      "grad_norm": 0.04059014469385147,
      "learning_rate": 2.6178436252643703e-06,
      "loss": 0.0051,
      "step": 215870
    },
    {
      "epoch": 1.384587263054263,
      "grad_norm": 0.05927395448088646,
      "learning_rate": 2.617351542426425e-06,
      "loss": 0.0026,
      "step": 215880
    },
    {
      "epoch": 1.3846513999480492,
      "grad_norm": 0.24868649244308472,
      "learning_rate": 2.6168594894447687e-06,
      "loss": 0.0013,
      "step": 215890
    },
    {
      "epoch": 1.3847155368418353,
      "grad_norm": 0.09808999300003052,
      "learning_rate": 2.6163674663255635e-06,
      "loss": 0.0013,
      "step": 215900
    },
    {
      "epoch": 1.3847796737356213,
      "grad_norm": 0.04929284378886223,
      "learning_rate": 2.615875473074979e-06,
      "loss": 0.0009,
      "step": 215910
    },
    {
      "epoch": 1.3848438106294074,
      "grad_norm": 0.030347779393196106,
      "learning_rate": 2.6153835096991793e-06,
      "loss": 0.0012,
      "step": 215920
    },
    {
      "epoch": 1.3849079475231936,
      "grad_norm": 0.1598779261112213,
      "learning_rate": 2.614891576204328e-06,
      "loss": 0.0043,
      "step": 215930
    },
    {
      "epoch": 1.3849720844169795,
      "grad_norm": 0.3156174123287201,
      "learning_rate": 2.614399672596588e-06,
      "loss": 0.0023,
      "step": 215940
    },
    {
      "epoch": 1.3850362213107656,
      "grad_norm": 0.14330817759037018,
      "learning_rate": 2.613907798882128e-06,
      "loss": 0.0018,
      "step": 215950
    },
    {
      "epoch": 1.3851003582045518,
      "grad_norm": 0.30222201347351074,
      "learning_rate": 2.6134159550671072e-06,
      "loss": 0.0008,
      "step": 215960
    },
    {
      "epoch": 1.385164495098338,
      "grad_norm": 0.07947871834039688,
      "learning_rate": 2.612924141157689e-06,
      "loss": 0.0016,
      "step": 215970
    },
    {
      "epoch": 1.385228631992124,
      "grad_norm": 0.1046389490365982,
      "learning_rate": 2.612432357160039e-06,
      "loss": 0.0012,
      "step": 215980
    },
    {
      "epoch": 1.38529276888591,
      "grad_norm": 0.005430108867585659,
      "learning_rate": 2.6119406030803186e-06,
      "loss": 0.0015,
      "step": 215990
    },
    {
      "epoch": 1.3853569057796962,
      "grad_norm": 0.08242988586425781,
      "learning_rate": 2.6114488789246882e-06,
      "loss": 0.0005,
      "step": 216000
    },
    {
      "epoch": 1.3854210426734823,
      "grad_norm": 0.03959507867693901,
      "learning_rate": 2.610957184699309e-06,
      "loss": 0.001,
      "step": 216010
    },
    {
      "epoch": 1.3854851795672682,
      "grad_norm": 0.03189368173480034,
      "learning_rate": 2.6104655204103435e-06,
      "loss": 0.0014,
      "step": 216020
    },
    {
      "epoch": 1.3855493164610544,
      "grad_norm": 0.0017806560499593616,
      "learning_rate": 2.6099738860639547e-06,
      "loss": 0.0021,
      "step": 216030
    },
    {
      "epoch": 1.3856134533548405,
      "grad_norm": 0.02551179751753807,
      "learning_rate": 2.6094822816663013e-06,
      "loss": 0.0007,
      "step": 216040
    },
    {
      "epoch": 1.3856775902486267,
      "grad_norm": 0.03730745241045952,
      "learning_rate": 2.6089907072235436e-06,
      "loss": 0.0009,
      "step": 216050
    },
    {
      "epoch": 1.3857417271424128,
      "grad_norm": 0.09445857256650925,
      "learning_rate": 2.608499162741839e-06,
      "loss": 0.0027,
      "step": 216060
    },
    {
      "epoch": 1.385805864036199,
      "grad_norm": 0.12052536755800247,
      "learning_rate": 2.608007648227351e-06,
      "loss": 0.0018,
      "step": 216070
    },
    {
      "epoch": 1.385870000929985,
      "grad_norm": 0.04084416851401329,
      "learning_rate": 2.6075161636862368e-06,
      "loss": 0.0009,
      "step": 216080
    },
    {
      "epoch": 1.385934137823771,
      "grad_norm": 0.01814129576086998,
      "learning_rate": 2.607024709124653e-06,
      "loss": 0.0014,
      "step": 216090
    },
    {
      "epoch": 1.3859982747175572,
      "grad_norm": 0.35136717557907104,
      "learning_rate": 2.606533284548762e-06,
      "loss": 0.0013,
      "step": 216100
    },
    {
      "epoch": 1.3860624116113431,
      "grad_norm": 0.03940017893910408,
      "learning_rate": 2.6060418899647188e-06,
      "loss": 0.002,
      "step": 216110
    },
    {
      "epoch": 1.3861265485051293,
      "grad_norm": 0.01433576736599207,
      "learning_rate": 2.6055505253786827e-06,
      "loss": 0.0008,
      "step": 216120
    },
    {
      "epoch": 1.3861906853989154,
      "grad_norm": 0.016492605209350586,
      "learning_rate": 2.605059190796807e-06,
      "loss": 0.0024,
      "step": 216130
    },
    {
      "epoch": 1.3862548222927016,
      "grad_norm": 0.08506309986114502,
      "learning_rate": 2.604567886225254e-06,
      "loss": 0.0016,
      "step": 216140
    },
    {
      "epoch": 1.3863189591864877,
      "grad_norm": 0.081624835729599,
      "learning_rate": 2.6040766116701764e-06,
      "loss": 0.0014,
      "step": 216150
    },
    {
      "epoch": 1.3863830960802737,
      "grad_norm": 0.04499850049614906,
      "learning_rate": 2.603585367137732e-06,
      "loss": 0.0009,
      "step": 216160
    },
    {
      "epoch": 1.3864472329740598,
      "grad_norm": 0.025472009554505348,
      "learning_rate": 2.6030941526340737e-06,
      "loss": 0.0013,
      "step": 216170
    },
    {
      "epoch": 1.386511369867846,
      "grad_norm": 0.5035187005996704,
      "learning_rate": 2.60260296816536e-06,
      "loss": 0.0027,
      "step": 216180
    },
    {
      "epoch": 1.386575506761632,
      "grad_norm": 0.07505131512880325,
      "learning_rate": 2.602111813737745e-06,
      "loss": 0.0013,
      "step": 216190
    },
    {
      "epoch": 1.386639643655418,
      "grad_norm": 0.13377723097801208,
      "learning_rate": 2.6016206893573825e-06,
      "loss": 0.0015,
      "step": 216200
    },
    {
      "epoch": 1.3867037805492042,
      "grad_norm": 0.28014901280403137,
      "learning_rate": 2.601129595030425e-06,
      "loss": 0.0014,
      "step": 216210
    },
    {
      "epoch": 1.3867679174429903,
      "grad_norm": 0.008496072143316269,
      "learning_rate": 2.60063853076303e-06,
      "loss": 0.0023,
      "step": 216220
    },
    {
      "epoch": 1.3868320543367765,
      "grad_norm": 0.08742309361696243,
      "learning_rate": 2.6001474965613493e-06,
      "loss": 0.0008,
      "step": 216230
    },
    {
      "epoch": 1.3868961912305626,
      "grad_norm": 0.10961107164621353,
      "learning_rate": 2.5996564924315332e-06,
      "loss": 0.0024,
      "step": 216240
    },
    {
      "epoch": 1.3869603281243486,
      "grad_norm": 0.1274971067905426,
      "learning_rate": 2.5991655183797394e-06,
      "loss": 0.0028,
      "step": 216250
    },
    {
      "epoch": 1.3870244650181347,
      "grad_norm": 0.07134983688592911,
      "learning_rate": 2.5986745744121168e-06,
      "loss": 0.0016,
      "step": 216260
    },
    {
      "epoch": 1.3870886019119208,
      "grad_norm": 0.07663287222385406,
      "learning_rate": 2.5981836605348174e-06,
      "loss": 0.0007,
      "step": 216270
    },
    {
      "epoch": 1.3871527388057068,
      "grad_norm": 0.15232987701892853,
      "learning_rate": 2.597692776753993e-06,
      "loss": 0.0028,
      "step": 216280
    },
    {
      "epoch": 1.387216875699493,
      "grad_norm": 0.04419165849685669,
      "learning_rate": 2.597201923075796e-06,
      "loss": 0.0006,
      "step": 216290
    },
    {
      "epoch": 1.387281012593279,
      "grad_norm": 0.051558490842580795,
      "learning_rate": 2.596711099506376e-06,
      "loss": 0.0016,
      "step": 216300
    },
    {
      "epoch": 1.3873451494870652,
      "grad_norm": 0.041023433208465576,
      "learning_rate": 2.596220306051884e-06,
      "loss": 0.0012,
      "step": 216310
    },
    {
      "epoch": 1.3874092863808514,
      "grad_norm": 0.04962332919239998,
      "learning_rate": 2.595729542718467e-06,
      "loss": 0.0015,
      "step": 216320
    },
    {
      "epoch": 1.3874734232746375,
      "grad_norm": 0.023958439007401466,
      "learning_rate": 2.59523880951228e-06,
      "loss": 0.0007,
      "step": 216330
    },
    {
      "epoch": 1.3875375601684234,
      "grad_norm": 0.0671641007065773,
      "learning_rate": 2.5947481064394688e-06,
      "loss": 0.0006,
      "step": 216340
    },
    {
      "epoch": 1.3876016970622096,
      "grad_norm": 0.13715045154094696,
      "learning_rate": 2.5942574335061826e-06,
      "loss": 0.001,
      "step": 216350
    },
    {
      "epoch": 1.3876658339559957,
      "grad_norm": 0.1365310549736023,
      "learning_rate": 2.593766790718568e-06,
      "loss": 0.0012,
      "step": 216360
    },
    {
      "epoch": 1.3877299708497817,
      "grad_norm": 0.06390999257564545,
      "learning_rate": 2.593276178082778e-06,
      "loss": 0.0014,
      "step": 216370
    },
    {
      "epoch": 1.3877941077435678,
      "grad_norm": 0.047890402376651764,
      "learning_rate": 2.5927855956049576e-06,
      "loss": 0.0015,
      "step": 216380
    },
    {
      "epoch": 1.387858244637354,
      "grad_norm": 0.12084142118692398,
      "learning_rate": 2.5922950432912515e-06,
      "loss": 0.0011,
      "step": 216390
    },
    {
      "epoch": 1.3879223815311401,
      "grad_norm": 0.027084503322839737,
      "learning_rate": 2.5918045211478116e-06,
      "loss": 0.0017,
      "step": 216400
    },
    {
      "epoch": 1.3879865184249263,
      "grad_norm": 0.1095394641160965,
      "learning_rate": 2.5913140291807814e-06,
      "loss": 0.0017,
      "step": 216410
    },
    {
      "epoch": 1.3880506553187122,
      "grad_norm": 0.06118927150964737,
      "learning_rate": 2.590823567396309e-06,
      "loss": 0.0012,
      "step": 216420
    },
    {
      "epoch": 1.3881147922124983,
      "grad_norm": 0.13544799387454987,
      "learning_rate": 2.5903331358005364e-06,
      "loss": 0.0024,
      "step": 216430
    },
    {
      "epoch": 1.3881789291062845,
      "grad_norm": 0.04669766500592232,
      "learning_rate": 2.5898427343996136e-06,
      "loss": 0.0017,
      "step": 216440
    },
    {
      "epoch": 1.3882430660000704,
      "grad_norm": 0.17931942641735077,
      "learning_rate": 2.589352363199684e-06,
      "loss": 0.0016,
      "step": 216450
    },
    {
      "epoch": 1.3883072028938566,
      "grad_norm": 0.042885906994342804,
      "learning_rate": 2.588862022206893e-06,
      "loss": 0.001,
      "step": 216460
    },
    {
      "epoch": 1.3883713397876427,
      "grad_norm": 0.025600217282772064,
      "learning_rate": 2.588371711427381e-06,
      "loss": 0.0035,
      "step": 216470
    },
    {
      "epoch": 1.3884354766814289,
      "grad_norm": 0.23572660982608795,
      "learning_rate": 2.587881430867298e-06,
      "loss": 0.0009,
      "step": 216480
    },
    {
      "epoch": 1.388499613575215,
      "grad_norm": 0.025940345600247383,
      "learning_rate": 2.5873911805327835e-06,
      "loss": 0.0011,
      "step": 216490
    },
    {
      "epoch": 1.3885637504690012,
      "grad_norm": 0.06433112919330597,
      "learning_rate": 2.586900960429982e-06,
      "loss": 0.0011,
      "step": 216500
    },
    {
      "epoch": 1.388627887362787,
      "grad_norm": 0.06253568828105927,
      "learning_rate": 2.5864107705650356e-06,
      "loss": 0.0012,
      "step": 216510
    },
    {
      "epoch": 1.3886920242565732,
      "grad_norm": 0.04966779425740242,
      "learning_rate": 2.585920610944087e-06,
      "loss": 0.0029,
      "step": 216520
    },
    {
      "epoch": 1.3887561611503594,
      "grad_norm": 0.0927979052066803,
      "learning_rate": 2.58543048157328e-06,
      "loss": 0.0015,
      "step": 216530
    },
    {
      "epoch": 1.3888202980441453,
      "grad_norm": 0.029799990355968475,
      "learning_rate": 2.584940382458752e-06,
      "loss": 0.0026,
      "step": 216540
    },
    {
      "epoch": 1.3888844349379315,
      "grad_norm": 0.023917008191347122,
      "learning_rate": 2.584450313606649e-06,
      "loss": 0.0006,
      "step": 216550
    },
    {
      "epoch": 1.3889485718317176,
      "grad_norm": 0.009702787734568119,
      "learning_rate": 2.5839602750231106e-06,
      "loss": 0.0007,
      "step": 216560
    },
    {
      "epoch": 1.3890127087255038,
      "grad_norm": 0.092452771961689,
      "learning_rate": 2.5834702667142765e-06,
      "loss": 0.0012,
      "step": 216570
    },
    {
      "epoch": 1.38907684561929,
      "grad_norm": 0.057914964854717255,
      "learning_rate": 2.5829802886862845e-06,
      "loss": 0.0011,
      "step": 216580
    },
    {
      "epoch": 1.3891409825130758,
      "grad_norm": 0.11667285114526749,
      "learning_rate": 2.582490340945279e-06,
      "loss": 0.0015,
      "step": 216590
    },
    {
      "epoch": 1.389205119406862,
      "grad_norm": 0.12485719472169876,
      "learning_rate": 2.582000423497397e-06,
      "loss": 0.0015,
      "step": 216600
    },
    {
      "epoch": 1.3892692563006481,
      "grad_norm": 0.06877756863832474,
      "learning_rate": 2.5815105363487792e-06,
      "loss": 0.0014,
      "step": 216610
    },
    {
      "epoch": 1.3893333931944343,
      "grad_norm": 0.10036937892436981,
      "learning_rate": 2.5810206795055593e-06,
      "loss": 0.0021,
      "step": 216620
    },
    {
      "epoch": 1.3893975300882202,
      "grad_norm": 0.053731124848127365,
      "learning_rate": 2.5805308529738814e-06,
      "loss": 0.0013,
      "step": 216630
    },
    {
      "epoch": 1.3894616669820063,
      "grad_norm": 0.03762771934270859,
      "learning_rate": 2.5800410567598823e-06,
      "loss": 0.0012,
      "step": 216640
    },
    {
      "epoch": 1.3895258038757925,
      "grad_norm": 0.0443277582526207,
      "learning_rate": 2.5795512908696953e-06,
      "loss": 0.0017,
      "step": 216650
    },
    {
      "epoch": 1.3895899407695786,
      "grad_norm": 0.054403021931648254,
      "learning_rate": 2.579061555309462e-06,
      "loss": 0.0018,
      "step": 216660
    },
    {
      "epoch": 1.3896540776633648,
      "grad_norm": 0.09955320507287979,
      "learning_rate": 2.5785718500853186e-06,
      "loss": 0.0021,
      "step": 216670
    },
    {
      "epoch": 1.3897182145571507,
      "grad_norm": 0.22564105689525604,
      "learning_rate": 2.5780821752034003e-06,
      "loss": 0.0012,
      "step": 216680
    },
    {
      "epoch": 1.3897823514509369,
      "grad_norm": 0.11402849853038788,
      "learning_rate": 2.5775925306698408e-06,
      "loss": 0.0018,
      "step": 216690
    },
    {
      "epoch": 1.389846488344723,
      "grad_norm": 0.1865815371274948,
      "learning_rate": 2.57710291649078e-06,
      "loss": 0.0011,
      "step": 216700
    },
    {
      "epoch": 1.389910625238509,
      "grad_norm": 0.036226801574230194,
      "learning_rate": 2.576613332672352e-06,
      "loss": 0.0016,
      "step": 216710
    },
    {
      "epoch": 1.389974762132295,
      "grad_norm": 0.09096764773130417,
      "learning_rate": 2.5761237792206904e-06,
      "loss": 0.0013,
      "step": 216720
    },
    {
      "epoch": 1.3900388990260812,
      "grad_norm": 0.03716425597667694,
      "learning_rate": 2.575634256141929e-06,
      "loss": 0.0007,
      "step": 216730
    },
    {
      "epoch": 1.3901030359198674,
      "grad_norm": 0.04131145030260086,
      "learning_rate": 2.5751447634422037e-06,
      "loss": 0.0013,
      "step": 216740
    },
    {
      "epoch": 1.3901671728136535,
      "grad_norm": 0.06567847728729248,
      "learning_rate": 2.5746553011276488e-06,
      "loss": 0.0015,
      "step": 216750
    },
    {
      "epoch": 1.3902313097074397,
      "grad_norm": 0.03881232813000679,
      "learning_rate": 2.574165869204396e-06,
      "loss": 0.002,
      "step": 216760
    },
    {
      "epoch": 1.3902954466012256,
      "grad_norm": 0.15144580602645874,
      "learning_rate": 2.5736764676785758e-06,
      "loss": 0.0014,
      "step": 216770
    },
    {
      "epoch": 1.3903595834950118,
      "grad_norm": 0.0609845295548439,
      "learning_rate": 2.5731870965563244e-06,
      "loss": 0.0017,
      "step": 216780
    },
    {
      "epoch": 1.390423720388798,
      "grad_norm": 0.07088928669691086,
      "learning_rate": 2.572697755843775e-06,
      "loss": 0.0009,
      "step": 216790
    },
    {
      "epoch": 1.3904878572825838,
      "grad_norm": 0.16477276384830475,
      "learning_rate": 2.5722084455470564e-06,
      "loss": 0.0023,
      "step": 216800
    },
    {
      "epoch": 1.39055199417637,
      "grad_norm": 0.021267952397465706,
      "learning_rate": 2.5717191656723005e-06,
      "loss": 0.0005,
      "step": 216810
    },
    {
      "epoch": 1.3906161310701561,
      "grad_norm": 0.03198913484811783,
      "learning_rate": 2.5712299162256404e-06,
      "loss": 0.001,
      "step": 216820
    },
    {
      "epoch": 1.3906802679639423,
      "grad_norm": 0.025601960718631744,
      "learning_rate": 2.5707406972132053e-06,
      "loss": 0.0022,
      "step": 216830
    },
    {
      "epoch": 1.3907444048577284,
      "grad_norm": 0.0331542082130909,
      "learning_rate": 2.5702515086411252e-06,
      "loss": 0.0008,
      "step": 216840
    },
    {
      "epoch": 1.3908085417515144,
      "grad_norm": 0.1769125908613205,
      "learning_rate": 2.5697623505155283e-06,
      "loss": 0.0014,
      "step": 216850
    },
    {
      "epoch": 1.3908726786453005,
      "grad_norm": 0.04075758531689644,
      "learning_rate": 2.569273222842548e-06,
      "loss": 0.0014,
      "step": 216860
    },
    {
      "epoch": 1.3909368155390867,
      "grad_norm": 0.09046512097120285,
      "learning_rate": 2.568784125628312e-06,
      "loss": 0.0016,
      "step": 216870
    },
    {
      "epoch": 1.3910009524328728,
      "grad_norm": 0.05227779224514961,
      "learning_rate": 2.568295058878948e-06,
      "loss": 0.0012,
      "step": 216880
    },
    {
      "epoch": 1.3910650893266587,
      "grad_norm": 0.084907665848732,
      "learning_rate": 2.5678060226005826e-06,
      "loss": 0.0014,
      "step": 216890
    },
    {
      "epoch": 1.3911292262204449,
      "grad_norm": 0.08666041493415833,
      "learning_rate": 2.567317016799348e-06,
      "loss": 0.0012,
      "step": 216900
    },
    {
      "epoch": 1.391193363114231,
      "grad_norm": 0.16974017024040222,
      "learning_rate": 2.56682804148137e-06,
      "loss": 0.0015,
      "step": 216910
    },
    {
      "epoch": 1.3912575000080172,
      "grad_norm": 0.16871021687984467,
      "learning_rate": 2.5663390966527734e-06,
      "loss": 0.0013,
      "step": 216920
    },
    {
      "epoch": 1.3913216369018033,
      "grad_norm": 0.034537214785814285,
      "learning_rate": 2.5658501823196903e-06,
      "loss": 0.0016,
      "step": 216930
    },
    {
      "epoch": 1.3913857737955893,
      "grad_norm": 0.038643766194581985,
      "learning_rate": 2.5653612984882425e-06,
      "loss": 0.0008,
      "step": 216940
    },
    {
      "epoch": 1.3914499106893754,
      "grad_norm": 0.09909739345312119,
      "learning_rate": 2.564872445164559e-06,
      "loss": 0.0011,
      "step": 216950
    },
    {
      "epoch": 1.3915140475831616,
      "grad_norm": 0.07292510569095612,
      "learning_rate": 2.564383622354761e-06,
      "loss": 0.0018,
      "step": 216960
    },
    {
      "epoch": 1.3915781844769475,
      "grad_norm": 0.028376858681440353,
      "learning_rate": 2.5638948300649795e-06,
      "loss": 0.0015,
      "step": 216970
    },
    {
      "epoch": 1.3916423213707336,
      "grad_norm": 0.016274534165859222,
      "learning_rate": 2.563406068301337e-06,
      "loss": 0.0017,
      "step": 216980
    },
    {
      "epoch": 1.3917064582645198,
      "grad_norm": 0.022052470594644547,
      "learning_rate": 2.5629173370699575e-06,
      "loss": 0.0006,
      "step": 216990
    },
    {
      "epoch": 1.391770595158306,
      "grad_norm": 0.019932232797145844,
      "learning_rate": 2.5624286363769635e-06,
      "loss": 0.0015,
      "step": 217000
    },
    {
      "epoch": 1.391834732052092,
      "grad_norm": 0.07324390858411789,
      "learning_rate": 2.561939966228483e-06,
      "loss": 0.0013,
      "step": 217010
    },
    {
      "epoch": 1.3918988689458782,
      "grad_norm": 0.1380704641342163,
      "learning_rate": 2.5614513266306374e-06,
      "loss": 0.0019,
      "step": 217020
    },
    {
      "epoch": 1.3919630058396641,
      "grad_norm": 0.09188283234834671,
      "learning_rate": 2.560962717589549e-06,
      "loss": 0.0006,
      "step": 217030
    },
    {
      "epoch": 1.3920271427334503,
      "grad_norm": 0.07266037911176682,
      "learning_rate": 2.5604741391113384e-06,
      "loss": 0.0014,
      "step": 217040
    },
    {
      "epoch": 1.3920912796272364,
      "grad_norm": 0.077426478266716,
      "learning_rate": 2.5599855912021334e-06,
      "loss": 0.0016,
      "step": 217050
    },
    {
      "epoch": 1.3921554165210224,
      "grad_norm": 0.053078003227710724,
      "learning_rate": 2.5594970738680526e-06,
      "loss": 0.0026,
      "step": 217060
    },
    {
      "epoch": 1.3922195534148085,
      "grad_norm": 0.002474989043548703,
      "learning_rate": 2.559008587115216e-06,
      "loss": 0.0007,
      "step": 217070
    },
    {
      "epoch": 1.3922836903085947,
      "grad_norm": 0.06136370450258255,
      "learning_rate": 2.5585201309497477e-06,
      "loss": 0.0036,
      "step": 217080
    },
    {
      "epoch": 1.3923478272023808,
      "grad_norm": 0.11670518666505814,
      "learning_rate": 2.558031705377767e-06,
      "loss": 0.001,
      "step": 217090
    },
    {
      "epoch": 1.392411964096167,
      "grad_norm": 0.0774407833814621,
      "learning_rate": 2.5575433104053945e-06,
      "loss": 0.0012,
      "step": 217100
    },
    {
      "epoch": 1.392476100989953,
      "grad_norm": 0.18761223554611206,
      "learning_rate": 2.557054946038748e-06,
      "loss": 0.0011,
      "step": 217110
    },
    {
      "epoch": 1.392540237883739,
      "grad_norm": 0.1756909042596817,
      "learning_rate": 2.5565666122839507e-06,
      "loss": 0.0013,
      "step": 217120
    },
    {
      "epoch": 1.3926043747775252,
      "grad_norm": 0.044896550476551056,
      "learning_rate": 2.5560783091471203e-06,
      "loss": 0.0015,
      "step": 217130
    },
    {
      "epoch": 1.3926685116713111,
      "grad_norm": 0.255946546792984,
      "learning_rate": 2.555590036634375e-06,
      "loss": 0.0028,
      "step": 217140
    },
    {
      "epoch": 1.3927326485650973,
      "grad_norm": 0.06978725641965866,
      "learning_rate": 2.555101794751832e-06,
      "loss": 0.0014,
      "step": 217150
    },
    {
      "epoch": 1.3927967854588834,
      "grad_norm": 0.03100358322262764,
      "learning_rate": 2.554613583505613e-06,
      "loss": 0.0016,
      "step": 217160
    },
    {
      "epoch": 1.3928609223526696,
      "grad_norm": 0.04570413753390312,
      "learning_rate": 2.5541254029018324e-06,
      "loss": 0.0016,
      "step": 217170
    },
    {
      "epoch": 1.3929250592464557,
      "grad_norm": 0.0653790831565857,
      "learning_rate": 2.5536372529466093e-06,
      "loss": 0.0016,
      "step": 217180
    },
    {
      "epoch": 1.3929891961402419,
      "grad_norm": 0.19819806516170502,
      "learning_rate": 2.553149133646058e-06,
      "loss": 0.0021,
      "step": 217190
    },
    {
      "epoch": 1.3930533330340278,
      "grad_norm": 0.1131213903427124,
      "learning_rate": 2.5526610450062983e-06,
      "loss": 0.0012,
      "step": 217200
    },
    {
      "epoch": 1.393117469927814,
      "grad_norm": 0.09659481793642044,
      "learning_rate": 2.5521729870334454e-06,
      "loss": 0.001,
      "step": 217210
    },
    {
      "epoch": 1.3931816068216,
      "grad_norm": 0.05584048852324486,
      "learning_rate": 2.5516849597336125e-06,
      "loss": 0.0009,
      "step": 217220
    },
    {
      "epoch": 1.393245743715386,
      "grad_norm": 0.12968671321868896,
      "learning_rate": 2.551196963112918e-06,
      "loss": 0.0025,
      "step": 217230
    },
    {
      "epoch": 1.3933098806091722,
      "grad_norm": 0.15531879663467407,
      "learning_rate": 2.5507089971774758e-06,
      "loss": 0.0015,
      "step": 217240
    },
    {
      "epoch": 1.3933740175029583,
      "grad_norm": 0.0011866064742207527,
      "learning_rate": 2.5502210619334013e-06,
      "loss": 0.0009,
      "step": 217250
    },
    {
      "epoch": 1.3934381543967445,
      "grad_norm": 0.06492505222558975,
      "learning_rate": 2.5497331573868046e-06,
      "loss": 0.0014,
      "step": 217260
    },
    {
      "epoch": 1.3935022912905306,
      "grad_norm": 0.07042653113603592,
      "learning_rate": 2.5492452835438055e-06,
      "loss": 0.0017,
      "step": 217270
    },
    {
      "epoch": 1.3935664281843165,
      "grad_norm": 0.008883239701390266,
      "learning_rate": 2.5487574404105135e-06,
      "loss": 0.0012,
      "step": 217280
    },
    {
      "epoch": 1.3936305650781027,
      "grad_norm": 0.5036296248435974,
      "learning_rate": 2.5482696279930435e-06,
      "loss": 0.001,
      "step": 217290
    },
    {
      "epoch": 1.3936947019718888,
      "grad_norm": 0.07283367216587067,
      "learning_rate": 2.547781846297505e-06,
      "loss": 0.0012,
      "step": 217300
    },
    {
      "epoch": 1.393758838865675,
      "grad_norm": 0.017748216167092323,
      "learning_rate": 2.5472940953300146e-06,
      "loss": 0.002,
      "step": 217310
    },
    {
      "epoch": 1.393822975759461,
      "grad_norm": 0.10280224680900574,
      "learning_rate": 2.546806375096682e-06,
      "loss": 0.0009,
      "step": 217320
    },
    {
      "epoch": 1.393887112653247,
      "grad_norm": 0.017901591956615448,
      "learning_rate": 2.5463186856036194e-06,
      "loss": 0.0017,
      "step": 217330
    },
    {
      "epoch": 1.3939512495470332,
      "grad_norm": 0.050007883459329605,
      "learning_rate": 2.5458310268569343e-06,
      "loss": 0.0013,
      "step": 217340
    },
    {
      "epoch": 1.3940153864408193,
      "grad_norm": 0.08111511170864105,
      "learning_rate": 2.545343398862743e-06,
      "loss": 0.001,
      "step": 217350
    },
    {
      "epoch": 1.3940795233346055,
      "grad_norm": 0.13947226107120514,
      "learning_rate": 2.544855801627153e-06,
      "loss": 0.0018,
      "step": 217360
    },
    {
      "epoch": 1.3941436602283914,
      "grad_norm": 0.04099639132618904,
      "learning_rate": 2.5443682351562727e-06,
      "loss": 0.0009,
      "step": 217370
    },
    {
      "epoch": 1.3942077971221776,
      "grad_norm": 0.0020156067330390215,
      "learning_rate": 2.543880699456215e-06,
      "loss": 0.0006,
      "step": 217380
    },
    {
      "epoch": 1.3942719340159637,
      "grad_norm": 0.001230254303663969,
      "learning_rate": 2.543393194533088e-06,
      "loss": 0.0008,
      "step": 217390
    },
    {
      "epoch": 1.3943360709097496,
      "grad_norm": 0.07310499995946884,
      "learning_rate": 2.542905720393e-06,
      "loss": 0.0019,
      "step": 217400
    },
    {
      "epoch": 1.3944002078035358,
      "grad_norm": 0.11221767961978912,
      "learning_rate": 2.542418277042057e-06,
      "loss": 0.0008,
      "step": 217410
    },
    {
      "epoch": 1.394464344697322,
      "grad_norm": 0.19872452318668365,
      "learning_rate": 2.5419308644863717e-06,
      "loss": 0.0018,
      "step": 217420
    },
    {
      "epoch": 1.394528481591108,
      "grad_norm": 0.0393829271197319,
      "learning_rate": 2.541443482732049e-06,
      "loss": 0.0011,
      "step": 217430
    },
    {
      "epoch": 1.3945926184848942,
      "grad_norm": 0.012060633860528469,
      "learning_rate": 2.540956131785197e-06,
      "loss": 0.0013,
      "step": 217440
    },
    {
      "epoch": 1.3946567553786804,
      "grad_norm": 0.0534462034702301,
      "learning_rate": 2.54046881165192e-06,
      "loss": 0.0018,
      "step": 217450
    },
    {
      "epoch": 1.3947208922724663,
      "grad_norm": 0.013612331822514534,
      "learning_rate": 2.5399815223383286e-06,
      "loss": 0.0011,
      "step": 217460
    },
    {
      "epoch": 1.3947850291662525,
      "grad_norm": 0.05973077192902565,
      "learning_rate": 2.539494263850527e-06,
      "loss": 0.001,
      "step": 217470
    },
    {
      "epoch": 1.3948491660600386,
      "grad_norm": 0.0985378846526146,
      "learning_rate": 2.5390070361946205e-06,
      "loss": 0.0012,
      "step": 217480
    },
    {
      "epoch": 1.3949133029538245,
      "grad_norm": 0.22525636851787567,
      "learning_rate": 2.538519839376713e-06,
      "loss": 0.0013,
      "step": 217490
    },
    {
      "epoch": 1.3949774398476107,
      "grad_norm": 0.1863296777009964,
      "learning_rate": 2.5380326734029127e-06,
      "loss": 0.001,
      "step": 217500
    },
    {
      "epoch": 1.3950415767413968,
      "grad_norm": 0.0970177873969078,
      "learning_rate": 2.5375455382793223e-06,
      "loss": 0.0016,
      "step": 217510
    },
    {
      "epoch": 1.395105713635183,
      "grad_norm": 0.06821413338184357,
      "learning_rate": 2.5370584340120453e-06,
      "loss": 0.0016,
      "step": 217520
    },
    {
      "epoch": 1.3951698505289691,
      "grad_norm": 0.06517302244901657,
      "learning_rate": 2.5365713606071874e-06,
      "loss": 0.0009,
      "step": 217530
    },
    {
      "epoch": 1.395233987422755,
      "grad_norm": 0.012658648192882538,
      "learning_rate": 2.5360843180708516e-06,
      "loss": 0.0014,
      "step": 217540
    },
    {
      "epoch": 1.3952981243165412,
      "grad_norm": 0.16197888553142548,
      "learning_rate": 2.5355973064091378e-06,
      "loss": 0.0014,
      "step": 217550
    },
    {
      "epoch": 1.3953622612103274,
      "grad_norm": 0.11316743493080139,
      "learning_rate": 2.535110325628153e-06,
      "loss": 0.0012,
      "step": 217560
    },
    {
      "epoch": 1.3954263981041133,
      "grad_norm": 0.42287132143974304,
      "learning_rate": 2.5346233757339956e-06,
      "loss": 0.0028,
      "step": 217570
    },
    {
      "epoch": 1.3954905349978994,
      "grad_norm": 0.06374204158782959,
      "learning_rate": 2.534136456732771e-06,
      "loss": 0.002,
      "step": 217580
    },
    {
      "epoch": 1.3955546718916856,
      "grad_norm": 0.18178769946098328,
      "learning_rate": 2.5336495686305796e-06,
      "loss": 0.0011,
      "step": 217590
    },
    {
      "epoch": 1.3956188087854717,
      "grad_norm": 0.147816002368927,
      "learning_rate": 2.5331627114335213e-06,
      "loss": 0.0023,
      "step": 217600
    },
    {
      "epoch": 1.3956829456792579,
      "grad_norm": 0.2753295302391052,
      "learning_rate": 2.532675885147695e-06,
      "loss": 0.0028,
      "step": 217610
    },
    {
      "epoch": 1.395747082573044,
      "grad_norm": 0.07842117547988892,
      "learning_rate": 2.5321890897792057e-06,
      "loss": 0.0015,
      "step": 217620
    },
    {
      "epoch": 1.39581121946683,
      "grad_norm": 0.04347194358706474,
      "learning_rate": 2.53170232533415e-06,
      "loss": 0.0018,
      "step": 217630
    },
    {
      "epoch": 1.395875356360616,
      "grad_norm": 0.11565347760915756,
      "learning_rate": 2.531215591818627e-06,
      "loss": 0.001,
      "step": 217640
    },
    {
      "epoch": 1.3959394932544023,
      "grad_norm": 0.2027328461408615,
      "learning_rate": 2.5307288892387395e-06,
      "loss": 0.0023,
      "step": 217650
    },
    {
      "epoch": 1.3960036301481882,
      "grad_norm": 0.03963017091155052,
      "learning_rate": 2.5302422176005824e-06,
      "loss": 0.0011,
      "step": 217660
    },
    {
      "epoch": 1.3960677670419743,
      "grad_norm": 0.22279760241508484,
      "learning_rate": 2.5297555769102564e-06,
      "loss": 0.0024,
      "step": 217670
    },
    {
      "epoch": 1.3961319039357605,
      "grad_norm": 0.12143002450466156,
      "learning_rate": 2.5292689671738567e-06,
      "loss": 0.0011,
      "step": 217680
    },
    {
      "epoch": 1.3961960408295466,
      "grad_norm": 0.046854518353939056,
      "learning_rate": 2.5287823883974837e-06,
      "loss": 0.0032,
      "step": 217690
    },
    {
      "epoch": 1.3962601777233328,
      "grad_norm": 0.06825122237205505,
      "learning_rate": 2.528295840587234e-06,
      "loss": 0.0015,
      "step": 217700
    },
    {
      "epoch": 1.3963243146171187,
      "grad_norm": 0.10842904448509216,
      "learning_rate": 2.5278093237492042e-06,
      "loss": 0.0023,
      "step": 217710
    },
    {
      "epoch": 1.3963884515109048,
      "grad_norm": 0.016050230711698532,
      "learning_rate": 2.527322837889488e-06,
      "loss": 0.0012,
      "step": 217720
    },
    {
      "epoch": 1.396452588404691,
      "grad_norm": 0.1674279272556305,
      "learning_rate": 2.526836383014186e-06,
      "loss": 0.0011,
      "step": 217730
    },
    {
      "epoch": 1.3965167252984771,
      "grad_norm": 0.2426510751247406,
      "learning_rate": 2.526349959129392e-06,
      "loss": 0.002,
      "step": 217740
    },
    {
      "epoch": 1.396580862192263,
      "grad_norm": 0.0029317704029381275,
      "learning_rate": 2.5258635662412007e-06,
      "loss": 0.0015,
      "step": 217750
    },
    {
      "epoch": 1.3966449990860492,
      "grad_norm": 0.15908031165599823,
      "learning_rate": 2.525377204355705e-06,
      "loss": 0.0011,
      "step": 217760
    },
    {
      "epoch": 1.3967091359798354,
      "grad_norm": 0.008479459211230278,
      "learning_rate": 2.5248908734790035e-06,
      "loss": 0.0011,
      "step": 217770
    },
    {
      "epoch": 1.3967732728736215,
      "grad_norm": 0.024743076413869858,
      "learning_rate": 2.5244045736171883e-06,
      "loss": 0.0014,
      "step": 217780
    },
    {
      "epoch": 1.3968374097674077,
      "grad_norm": 0.16069556772708893,
      "learning_rate": 2.5239183047763514e-06,
      "loss": 0.0012,
      "step": 217790
    },
    {
      "epoch": 1.3969015466611936,
      "grad_norm": 0.06781712919473648,
      "learning_rate": 2.523432066962589e-06,
      "loss": 0.0005,
      "step": 217800
    },
    {
      "epoch": 1.3969656835549797,
      "grad_norm": 0.13242220878601074,
      "learning_rate": 2.5229458601819934e-06,
      "loss": 0.0016,
      "step": 217810
    },
    {
      "epoch": 1.397029820448766,
      "grad_norm": 0.03151949495077133,
      "learning_rate": 2.522459684440657e-06,
      "loss": 0.0014,
      "step": 217820
    },
    {
      "epoch": 1.3970939573425518,
      "grad_norm": 0.09716480225324631,
      "learning_rate": 2.5219735397446686e-06,
      "loss": 0.0006,
      "step": 217830
    },
    {
      "epoch": 1.397158094236338,
      "grad_norm": 0.12030621618032455,
      "learning_rate": 2.521487426100125e-06,
      "loss": 0.0009,
      "step": 217840
    },
    {
      "epoch": 1.3972222311301241,
      "grad_norm": 0.06489705294370651,
      "learning_rate": 2.521001343513115e-06,
      "loss": 0.0007,
      "step": 217850
    },
    {
      "epoch": 1.3972863680239103,
      "grad_norm": 0.13355490565299988,
      "learning_rate": 2.5205152919897304e-06,
      "loss": 0.0017,
      "step": 217860
    },
    {
      "epoch": 1.3973505049176964,
      "grad_norm": 0.020991241559386253,
      "learning_rate": 2.5200292715360588e-06,
      "loss": 0.0007,
      "step": 217870
    },
    {
      "epoch": 1.3974146418114826,
      "grad_norm": 0.10901031643152237,
      "learning_rate": 2.519543282158195e-06,
      "loss": 0.0012,
      "step": 217880
    },
    {
      "epoch": 1.3974787787052685,
      "grad_norm": 0.04401480406522751,
      "learning_rate": 2.5190573238622266e-06,
      "loss": 0.0021,
      "step": 217890
    },
    {
      "epoch": 1.3975429155990546,
      "grad_norm": 0.026542415842413902,
      "learning_rate": 2.5185713966542435e-06,
      "loss": 0.001,
      "step": 217900
    },
    {
      "epoch": 1.3976070524928408,
      "grad_norm": 0.09319578111171722,
      "learning_rate": 2.5180855005403316e-06,
      "loss": 0.0016,
      "step": 217910
    },
    {
      "epoch": 1.3976711893866267,
      "grad_norm": 0.09715227037668228,
      "learning_rate": 2.517599635526584e-06,
      "loss": 0.0012,
      "step": 217920
    },
    {
      "epoch": 1.3977353262804129,
      "grad_norm": 0.157813161611557,
      "learning_rate": 2.517113801619088e-06,
      "loss": 0.0022,
      "step": 217930
    },
    {
      "epoch": 1.397799463174199,
      "grad_norm": 0.09481701999902725,
      "learning_rate": 2.5166279988239283e-06,
      "loss": 0.0014,
      "step": 217940
    },
    {
      "epoch": 1.3978636000679852,
      "grad_norm": 0.046457622200250626,
      "learning_rate": 2.5161422271471964e-06,
      "loss": 0.0011,
      "step": 217950
    },
    {
      "epoch": 1.3979277369617713,
      "grad_norm": 0.0647912546992302,
      "learning_rate": 2.5156564865949774e-06,
      "loss": 0.0027,
      "step": 217960
    },
    {
      "epoch": 1.3979918738555572,
      "grad_norm": 0.05202125012874603,
      "learning_rate": 2.515170777173358e-06,
      "loss": 0.0016,
      "step": 217970
    },
    {
      "epoch": 1.3980560107493434,
      "grad_norm": 0.13212570548057556,
      "learning_rate": 2.514685098888423e-06,
      "loss": 0.0014,
      "step": 217980
    },
    {
      "epoch": 1.3981201476431295,
      "grad_norm": 0.019459405913949013,
      "learning_rate": 2.514199451746262e-06,
      "loss": 0.0008,
      "step": 217990
    },
    {
      "epoch": 1.3981842845369155,
      "grad_norm": 0.054449573159217834,
      "learning_rate": 2.513713835752959e-06,
      "loss": 0.0008,
      "step": 218000
    },
    {
      "epoch": 1.3982484214307016,
      "grad_norm": 0.08177845180034637,
      "learning_rate": 2.513228250914598e-06,
      "loss": 0.001,
      "step": 218010
    },
    {
      "epoch": 1.3983125583244878,
      "grad_norm": 0.05067718029022217,
      "learning_rate": 2.512742697237262e-06,
      "loss": 0.0014,
      "step": 218020
    },
    {
      "epoch": 1.398376695218274,
      "grad_norm": 0.1587323546409607,
      "learning_rate": 2.5122571747270397e-06,
      "loss": 0.0009,
      "step": 218030
    },
    {
      "epoch": 1.39844083211206,
      "grad_norm": 0.08408209681510925,
      "learning_rate": 2.511771683390013e-06,
      "loss": 0.0014,
      "step": 218040
    },
    {
      "epoch": 1.3985049690058462,
      "grad_norm": 0.011681445874273777,
      "learning_rate": 2.5112862232322654e-06,
      "loss": 0.0009,
      "step": 218050
    },
    {
      "epoch": 1.3985691058996321,
      "grad_norm": 0.034590236842632294,
      "learning_rate": 2.510800794259879e-06,
      "loss": 0.0012,
      "step": 218060
    },
    {
      "epoch": 1.3986332427934183,
      "grad_norm": 0.03324037417769432,
      "learning_rate": 2.5103153964789386e-06,
      "loss": 0.0009,
      "step": 218070
    },
    {
      "epoch": 1.3986973796872044,
      "grad_norm": 0.019120942801237106,
      "learning_rate": 2.5098300298955257e-06,
      "loss": 0.0019,
      "step": 218080
    },
    {
      "epoch": 1.3987615165809903,
      "grad_norm": 0.1737535446882248,
      "learning_rate": 2.5093446945157206e-06,
      "loss": 0.0012,
      "step": 218090
    },
    {
      "epoch": 1.3988256534747765,
      "grad_norm": 0.04208459332585335,
      "learning_rate": 2.508859390345608e-06,
      "loss": 0.0018,
      "step": 218100
    },
    {
      "epoch": 1.3988897903685626,
      "grad_norm": 0.04920925945043564,
      "learning_rate": 2.508374117391268e-06,
      "loss": 0.0015,
      "step": 218110
    },
    {
      "epoch": 1.3989539272623488,
      "grad_norm": 0.027007663622498512,
      "learning_rate": 2.507888875658781e-06,
      "loss": 0.0015,
      "step": 218120
    },
    {
      "epoch": 1.399018064156135,
      "grad_norm": 0.09326251596212387,
      "learning_rate": 2.5074036651542256e-06,
      "loss": 0.0014,
      "step": 218130
    },
    {
      "epoch": 1.3990822010499209,
      "grad_norm": 0.033007364720106125,
      "learning_rate": 2.506918485883686e-06,
      "loss": 0.0011,
      "step": 218140
    },
    {
      "epoch": 1.399146337943707,
      "grad_norm": 0.3016238510608673,
      "learning_rate": 2.506433337853239e-06,
      "loss": 0.0016,
      "step": 218150
    },
    {
      "epoch": 1.3992104748374932,
      "grad_norm": 0.025880996137857437,
      "learning_rate": 2.5059482210689645e-06,
      "loss": 0.0011,
      "step": 218160
    },
    {
      "epoch": 1.3992746117312793,
      "grad_norm": 0.0889752134680748,
      "learning_rate": 2.505463135536939e-06,
      "loss": 0.0008,
      "step": 218170
    },
    {
      "epoch": 1.3993387486250652,
      "grad_norm": 0.08893129974603653,
      "learning_rate": 2.5049780812632457e-06,
      "loss": 0.0018,
      "step": 218180
    },
    {
      "epoch": 1.3994028855188514,
      "grad_norm": 0.05968914180994034,
      "learning_rate": 2.50449305825396e-06,
      "loss": 0.0017,
      "step": 218190
    },
    {
      "epoch": 1.3994670224126375,
      "grad_norm": 0.06970475614070892,
      "learning_rate": 2.50400806651516e-06,
      "loss": 0.0008,
      "step": 218200
    },
    {
      "epoch": 1.3995311593064237,
      "grad_norm": 0.16024905443191528,
      "learning_rate": 2.503523106052921e-06,
      "loss": 0.0015,
      "step": 218210
    },
    {
      "epoch": 1.3995952962002098,
      "grad_norm": 0.06745795160531998,
      "learning_rate": 2.5030381768733223e-06,
      "loss": 0.0008,
      "step": 218220
    },
    {
      "epoch": 1.3996594330939958,
      "grad_norm": 0.07381200790405273,
      "learning_rate": 2.5025532789824414e-06,
      "loss": 0.0008,
      "step": 218230
    },
    {
      "epoch": 1.399723569987782,
      "grad_norm": 0.08183572441339493,
      "learning_rate": 2.5020684123863503e-06,
      "loss": 0.0012,
      "step": 218240
    },
    {
      "epoch": 1.399787706881568,
      "grad_norm": 0.0365438275039196,
      "learning_rate": 2.501583577091129e-06,
      "loss": 0.0011,
      "step": 218250
    },
    {
      "epoch": 1.399851843775354,
      "grad_norm": 0.0748719871044159,
      "learning_rate": 2.501098773102851e-06,
      "loss": 0.001,
      "step": 218260
    },
    {
      "epoch": 1.3999159806691401,
      "grad_norm": 0.04895782470703125,
      "learning_rate": 2.5006140004275913e-06,
      "loss": 0.0015,
      "step": 218270
    },
    {
      "epoch": 1.3999801175629263,
      "grad_norm": 0.022944876924157143,
      "learning_rate": 2.500129259071423e-06,
      "loss": 0.0009,
      "step": 218280
    },
    {
      "epoch": 1.4000442544567124,
      "grad_norm": 0.057753268629312515,
      "learning_rate": 2.4996445490404237e-06,
      "loss": 0.0006,
      "step": 218290
    },
    {
      "epoch": 1.4001083913504986,
      "grad_norm": 0.05829066038131714,
      "learning_rate": 2.499159870340665e-06,
      "loss": 0.0007,
      "step": 218300
    },
    {
      "epoch": 1.4001725282442847,
      "grad_norm": 0.11744900047779083,
      "learning_rate": 2.498675222978218e-06,
      "loss": 0.0007,
      "step": 218310
    },
    {
      "epoch": 1.4002366651380707,
      "grad_norm": 0.047571297734975815,
      "learning_rate": 2.4981906069591603e-06,
      "loss": 0.0011,
      "step": 218320
    },
    {
      "epoch": 1.4003008020318568,
      "grad_norm": 0.06034592166543007,
      "learning_rate": 2.497706022289561e-06,
      "loss": 0.0008,
      "step": 218330
    },
    {
      "epoch": 1.400364938925643,
      "grad_norm": 0.12814146280288696,
      "learning_rate": 2.4972214689754948e-06,
      "loss": 0.001,
      "step": 218340
    },
    {
      "epoch": 1.4004290758194289,
      "grad_norm": 0.005401868373155594,
      "learning_rate": 2.496736947023033e-06,
      "loss": 0.0011,
      "step": 218350
    },
    {
      "epoch": 1.400493212713215,
      "grad_norm": 0.17886695265769958,
      "learning_rate": 2.4962524564382435e-06,
      "loss": 0.0053,
      "step": 218360
    },
    {
      "epoch": 1.4005573496070012,
      "grad_norm": 0.005190698895603418,
      "learning_rate": 2.4957679972272023e-06,
      "loss": 0.0014,
      "step": 218370
    },
    {
      "epoch": 1.4006214865007873,
      "grad_norm": 0.0008826800039969385,
      "learning_rate": 2.4952835693959782e-06,
      "loss": 0.0007,
      "step": 218380
    },
    {
      "epoch": 1.4006856233945735,
      "grad_norm": 0.15037375688552856,
      "learning_rate": 2.494799172950641e-06,
      "loss": 0.001,
      "step": 218390
    },
    {
      "epoch": 1.4007497602883594,
      "grad_norm": 0.32553181052207947,
      "learning_rate": 2.4943148078972586e-06,
      "loss": 0.0011,
      "step": 218400
    },
    {
      "epoch": 1.4008138971821456,
      "grad_norm": 0.011083441786468029,
      "learning_rate": 2.4938304742419045e-06,
      "loss": 0.0007,
      "step": 218410
    },
    {
      "epoch": 1.4008780340759317,
      "grad_norm": 0.16851891577243805,
      "learning_rate": 2.4933461719906455e-06,
      "loss": 0.001,
      "step": 218420
    },
    {
      "epoch": 1.4009421709697178,
      "grad_norm": 0.19955788552761078,
      "learning_rate": 2.49286190114955e-06,
      "loss": 0.001,
      "step": 218430
    },
    {
      "epoch": 1.4010063078635038,
      "grad_norm": 0.07683160156011581,
      "learning_rate": 2.492377661724686e-06,
      "loss": 0.0007,
      "step": 218440
    },
    {
      "epoch": 1.40107044475729,
      "grad_norm": 0.025752320885658264,
      "learning_rate": 2.4918934537221228e-06,
      "loss": 0.0011,
      "step": 218450
    },
    {
      "epoch": 1.401134581651076,
      "grad_norm": 0.25075531005859375,
      "learning_rate": 2.4914092771479276e-06,
      "loss": 0.0028,
      "step": 218460
    },
    {
      "epoch": 1.4011987185448622,
      "grad_norm": 0.07408151775598526,
      "learning_rate": 2.4909251320081674e-06,
      "loss": 0.001,
      "step": 218470
    },
    {
      "epoch": 1.4012628554386484,
      "grad_norm": 0.10768114030361176,
      "learning_rate": 2.490441018308906e-06,
      "loss": 0.002,
      "step": 218480
    },
    {
      "epoch": 1.4013269923324343,
      "grad_norm": 0.21442459523677826,
      "learning_rate": 2.489956936056214e-06,
      "loss": 0.0013,
      "step": 218490
    },
    {
      "epoch": 1.4013911292262204,
      "grad_norm": 0.20392319560050964,
      "learning_rate": 2.489472885256156e-06,
      "loss": 0.0013,
      "step": 218500
    },
    {
      "epoch": 1.4014552661200066,
      "grad_norm": 0.022533521056175232,
      "learning_rate": 2.4889888659147947e-06,
      "loss": 0.0006,
      "step": 218510
    },
    {
      "epoch": 1.4015194030137925,
      "grad_norm": 0.0441557802259922,
      "learning_rate": 2.4885048780382e-06,
      "loss": 0.0043,
      "step": 218520
    },
    {
      "epoch": 1.4015835399075787,
      "grad_norm": 0.04519323632121086,
      "learning_rate": 2.4880209216324335e-06,
      "loss": 0.0013,
      "step": 218530
    },
    {
      "epoch": 1.4016476768013648,
      "grad_norm": 0.07286236435174942,
      "learning_rate": 2.4875369967035607e-06,
      "loss": 0.0011,
      "step": 218540
    },
    {
      "epoch": 1.401711813695151,
      "grad_norm": 0.02912171557545662,
      "learning_rate": 2.487053103257642e-06,
      "loss": 0.0011,
      "step": 218550
    },
    {
      "epoch": 1.4017759505889371,
      "grad_norm": 0.16293995082378387,
      "learning_rate": 2.486569241300747e-06,
      "loss": 0.0021,
      "step": 218560
    },
    {
      "epoch": 1.4018400874827233,
      "grad_norm": 0.1001424491405487,
      "learning_rate": 2.4860854108389353e-06,
      "loss": 0.0015,
      "step": 218570
    },
    {
      "epoch": 1.4019042243765092,
      "grad_norm": 0.09254918247461319,
      "learning_rate": 2.48560161187827e-06,
      "loss": 0.0011,
      "step": 218580
    },
    {
      "epoch": 1.4019683612702953,
      "grad_norm": 0.024134088307619095,
      "learning_rate": 2.4851178444248112e-06,
      "loss": 0.0011,
      "step": 218590
    },
    {
      "epoch": 1.4020324981640815,
      "grad_norm": 0.03021530620753765,
      "learning_rate": 2.484634108484626e-06,
      "loss": 0.0016,
      "step": 218600
    },
    {
      "epoch": 1.4020966350578674,
      "grad_norm": 0.05822308361530304,
      "learning_rate": 2.4841504040637726e-06,
      "loss": 0.0008,
      "step": 218610
    },
    {
      "epoch": 1.4021607719516536,
      "grad_norm": 0.5333791971206665,
      "learning_rate": 2.483666731168313e-06,
      "loss": 0.0039,
      "step": 218620
    },
    {
      "epoch": 1.4022249088454397,
      "grad_norm": 0.05283120274543762,
      "learning_rate": 2.4831830898043054e-06,
      "loss": 0.0013,
      "step": 218630
    },
    {
      "epoch": 1.4022890457392259,
      "grad_norm": 0.12147217243909836,
      "learning_rate": 2.4826994799778143e-06,
      "loss": 0.0007,
      "step": 218640
    },
    {
      "epoch": 1.402353182633012,
      "grad_norm": 0.07120306044816971,
      "learning_rate": 2.4822159016948975e-06,
      "loss": 0.0013,
      "step": 218650
    },
    {
      "epoch": 1.402417319526798,
      "grad_norm": 0.0498415008187294,
      "learning_rate": 2.4817323549616134e-06,
      "loss": 0.0014,
      "step": 218660
    },
    {
      "epoch": 1.402481456420584,
      "grad_norm": 0.04400544986128807,
      "learning_rate": 2.4812488397840246e-06,
      "loss": 0.0009,
      "step": 218670
    },
    {
      "epoch": 1.4025455933143702,
      "grad_norm": 0.023885253816843033,
      "learning_rate": 2.480765356168188e-06,
      "loss": 0.0006,
      "step": 218680
    },
    {
      "epoch": 1.4026097302081562,
      "grad_norm": 0.09544313699007034,
      "learning_rate": 2.480281904120162e-06,
      "loss": 0.0027,
      "step": 218690
    },
    {
      "epoch": 1.4026738671019423,
      "grad_norm": 0.034635111689567566,
      "learning_rate": 2.479798483646002e-06,
      "loss": 0.0008,
      "step": 218700
    },
    {
      "epoch": 1.4027380039957285,
      "grad_norm": 0.10411494970321655,
      "learning_rate": 2.4793150947517706e-06,
      "loss": 0.0025,
      "step": 218710
    },
    {
      "epoch": 1.4028021408895146,
      "grad_norm": 0.06943442672491074,
      "learning_rate": 2.4788317374435227e-06,
      "loss": 0.002,
      "step": 218720
    },
    {
      "epoch": 1.4028662777833008,
      "grad_norm": 0.11532283574342728,
      "learning_rate": 2.478348411727315e-06,
      "loss": 0.0011,
      "step": 218730
    },
    {
      "epoch": 1.402930414677087,
      "grad_norm": 0.08214882761240005,
      "learning_rate": 2.4778651176092018e-06,
      "loss": 0.0024,
      "step": 218740
    },
    {
      "epoch": 1.4029945515708728,
      "grad_norm": 0.1970735341310501,
      "learning_rate": 2.4773818550952435e-06,
      "loss": 0.0009,
      "step": 218750
    },
    {
      "epoch": 1.403058688464659,
      "grad_norm": 0.6348500847816467,
      "learning_rate": 2.4768986241914926e-06,
      "loss": 0.0009,
      "step": 218760
    },
    {
      "epoch": 1.4031228253584451,
      "grad_norm": 0.059380240738391876,
      "learning_rate": 2.476415424904004e-06,
      "loss": 0.0009,
      "step": 218770
    },
    {
      "epoch": 1.403186962252231,
      "grad_norm": 0.22803273797035217,
      "learning_rate": 2.4759322572388357e-06,
      "loss": 0.0011,
      "step": 218780
    },
    {
      "epoch": 1.4032510991460172,
      "grad_norm": 0.1198737621307373,
      "learning_rate": 2.475449121202039e-06,
      "loss": 0.0012,
      "step": 218790
    },
    {
      "epoch": 1.4033152360398033,
      "grad_norm": 0.1235053688287735,
      "learning_rate": 2.4749660167996702e-06,
      "loss": 0.0007,
      "step": 218800
    },
    {
      "epoch": 1.4033793729335895,
      "grad_norm": 0.034360114485025406,
      "learning_rate": 2.47448294403778e-06,
      "loss": 0.0017,
      "step": 218810
    },
    {
      "epoch": 1.4034435098273756,
      "grad_norm": 0.16802871227264404,
      "learning_rate": 2.473999902922425e-06,
      "loss": 0.0015,
      "step": 218820
    },
    {
      "epoch": 1.4035076467211616,
      "grad_norm": 0.001983431400731206,
      "learning_rate": 2.4735168934596567e-06,
      "loss": 0.0007,
      "step": 218830
    },
    {
      "epoch": 1.4035717836149477,
      "grad_norm": 0.12489792704582214,
      "learning_rate": 2.4730339156555266e-06,
      "loss": 0.0008,
      "step": 218840
    },
    {
      "epoch": 1.4036359205087339,
      "grad_norm": 0.37852925062179565,
      "learning_rate": 2.4725509695160863e-06,
      "loss": 0.0026,
      "step": 218850
    },
    {
      "epoch": 1.40370005740252,
      "grad_norm": 0.1418200582265854,
      "learning_rate": 2.472068055047391e-06,
      "loss": 0.0022,
      "step": 218860
    },
    {
      "epoch": 1.403764194296306,
      "grad_norm": 0.02384890243411064,
      "learning_rate": 2.4715851722554886e-06,
      "loss": 0.0022,
      "step": 218870
    },
    {
      "epoch": 1.403828331190092,
      "grad_norm": 0.10498286038637161,
      "learning_rate": 2.471102321146432e-06,
      "loss": 0.0009,
      "step": 218880
    },
    {
      "epoch": 1.4038924680838782,
      "grad_norm": 0.05484259873628616,
      "learning_rate": 2.470619501726268e-06,
      "loss": 0.001,
      "step": 218890
    },
    {
      "epoch": 1.4039566049776644,
      "grad_norm": 0.11952408403158188,
      "learning_rate": 2.470136714001052e-06,
      "loss": 0.0046,
      "step": 218900
    },
    {
      "epoch": 1.4040207418714505,
      "grad_norm": 0.0364367812871933,
      "learning_rate": 2.46965395797683e-06,
      "loss": 0.0006,
      "step": 218910
    },
    {
      "epoch": 1.4040848787652365,
      "grad_norm": 0.06499797850847244,
      "learning_rate": 2.469171233659651e-06,
      "loss": 0.0013,
      "step": 218920
    },
    {
      "epoch": 1.4041490156590226,
      "grad_norm": 0.05752279981970787,
      "learning_rate": 2.468688541055568e-06,
      "loss": 0.0008,
      "step": 218930
    },
    {
      "epoch": 1.4042131525528088,
      "grad_norm": 0.09352587163448334,
      "learning_rate": 2.4682058801706256e-06,
      "loss": 0.001,
      "step": 218940
    },
    {
      "epoch": 1.4042772894465947,
      "grad_norm": 0.00825769267976284,
      "learning_rate": 2.4677232510108732e-06,
      "loss": 0.0038,
      "step": 218950
    },
    {
      "epoch": 1.4043414263403808,
      "grad_norm": 0.062322016805410385,
      "learning_rate": 2.467240653582356e-06,
      "loss": 0.001,
      "step": 218960
    },
    {
      "epoch": 1.404405563234167,
      "grad_norm": 0.09102189540863037,
      "learning_rate": 2.4667580878911265e-06,
      "loss": 0.0011,
      "step": 218970
    },
    {
      "epoch": 1.4044697001279531,
      "grad_norm": 0.23846127092838287,
      "learning_rate": 2.4662755539432277e-06,
      "loss": 0.002,
      "step": 218980
    },
    {
      "epoch": 1.4045338370217393,
      "grad_norm": 0.06262645870447159,
      "learning_rate": 2.4657930517447076e-06,
      "loss": 0.0011,
      "step": 218990
    },
    {
      "epoch": 1.4045979739155254,
      "grad_norm": 0.0890268087387085,
      "learning_rate": 2.4653105813016095e-06,
      "loss": 0.0022,
      "step": 219000
    },
    {
      "epoch": 1.4046621108093114,
      "grad_norm": 0.017491163685917854,
      "learning_rate": 2.464828142619983e-06,
      "loss": 0.0008,
      "step": 219010
    },
    {
      "epoch": 1.4047262477030975,
      "grad_norm": 0.05489290505647659,
      "learning_rate": 2.4643457357058718e-06,
      "loss": 0.0013,
      "step": 219020
    },
    {
      "epoch": 1.4047903845968837,
      "grad_norm": 0.15791866183280945,
      "learning_rate": 2.463863360565321e-06,
      "loss": 0.0012,
      "step": 219030
    },
    {
      "epoch": 1.4048545214906696,
      "grad_norm": 0.10572674870491028,
      "learning_rate": 2.463381017204373e-06,
      "loss": 0.0014,
      "step": 219040
    },
    {
      "epoch": 1.4049186583844557,
      "grad_norm": 0.6303642988204956,
      "learning_rate": 2.462898705629076e-06,
      "loss": 0.0014,
      "step": 219050
    },
    {
      "epoch": 1.4049827952782419,
      "grad_norm": 0.052637193351984024,
      "learning_rate": 2.4624164258454707e-06,
      "loss": 0.0009,
      "step": 219060
    },
    {
      "epoch": 1.405046932172028,
      "grad_norm": 0.10681580752134323,
      "learning_rate": 2.4619341778595994e-06,
      "loss": 0.0024,
      "step": 219070
    },
    {
      "epoch": 1.4051110690658142,
      "grad_norm": 0.051800213754177094,
      "learning_rate": 2.4614519616775075e-06,
      "loss": 0.0042,
      "step": 219080
    },
    {
      "epoch": 1.4051752059596,
      "grad_norm": 0.11543156206607819,
      "learning_rate": 2.460969777305238e-06,
      "loss": 0.0012,
      "step": 219090
    },
    {
      "epoch": 1.4052393428533863,
      "grad_norm": 0.05147361382842064,
      "learning_rate": 2.4604876247488317e-06,
      "loss": 0.0017,
      "step": 219100
    },
    {
      "epoch": 1.4053034797471724,
      "grad_norm": 0.09925807267427444,
      "learning_rate": 2.4600055040143308e-06,
      "loss": 0.0006,
      "step": 219110
    },
    {
      "epoch": 1.4053676166409583,
      "grad_norm": 0.11557259410619736,
      "learning_rate": 2.459523415107774e-06,
      "loss": 0.0012,
      "step": 219120
    },
    {
      "epoch": 1.4054317535347445,
      "grad_norm": 0.07576493918895721,
      "learning_rate": 2.459041358035207e-06,
      "loss": 0.0041,
      "step": 219130
    },
    {
      "epoch": 1.4054958904285306,
      "grad_norm": 0.07128460705280304,
      "learning_rate": 2.458559332802667e-06,
      "loss": 0.0013,
      "step": 219140
    },
    {
      "epoch": 1.4055600273223168,
      "grad_norm": 0.07296294718980789,
      "learning_rate": 2.4580773394161954e-06,
      "loss": 0.0011,
      "step": 219150
    },
    {
      "epoch": 1.405624164216103,
      "grad_norm": 0.11052407324314117,
      "learning_rate": 2.4575953778818295e-06,
      "loss": 0.0009,
      "step": 219160
    },
    {
      "epoch": 1.405688301109889,
      "grad_norm": 0.035698726773262024,
      "learning_rate": 2.457113448205612e-06,
      "loss": 0.0009,
      "step": 219170
    },
    {
      "epoch": 1.405752438003675,
      "grad_norm": 0.11392034590244293,
      "learning_rate": 2.4566315503935813e-06,
      "loss": 0.0017,
      "step": 219180
    },
    {
      "epoch": 1.4058165748974611,
      "grad_norm": 0.12178272753953934,
      "learning_rate": 2.4561496844517717e-06,
      "loss": 0.0011,
      "step": 219190
    },
    {
      "epoch": 1.4058807117912473,
      "grad_norm": 0.1255182921886444,
      "learning_rate": 2.4556678503862276e-06,
      "loss": 0.0028,
      "step": 219200
    },
    {
      "epoch": 1.4059448486850332,
      "grad_norm": 0.06408878415822983,
      "learning_rate": 2.4551860482029834e-06,
      "loss": 0.0008,
      "step": 219210
    },
    {
      "epoch": 1.4060089855788194,
      "grad_norm": 0.061047110706567764,
      "learning_rate": 2.454704277908077e-06,
      "loss": 0.0011,
      "step": 219220
    },
    {
      "epoch": 1.4060731224726055,
      "grad_norm": 0.04080299288034439,
      "learning_rate": 2.4542225395075433e-06,
      "loss": 0.0014,
      "step": 219230
    },
    {
      "epoch": 1.4061372593663917,
      "grad_norm": 0.011935588903725147,
      "learning_rate": 2.453740833007422e-06,
      "loss": 0.0029,
      "step": 219240
    },
    {
      "epoch": 1.4062013962601778,
      "grad_norm": 0.1219937726855278,
      "learning_rate": 2.453259158413748e-06,
      "loss": 0.0013,
      "step": 219250
    },
    {
      "epoch": 1.4062655331539637,
      "grad_norm": 0.03379470854997635,
      "learning_rate": 2.4527775157325568e-06,
      "loss": 0.0007,
      "step": 219260
    },
    {
      "epoch": 1.40632967004775,
      "grad_norm": 0.0320466049015522,
      "learning_rate": 2.4522959049698817e-06,
      "loss": 0.0009,
      "step": 219270
    },
    {
      "epoch": 1.406393806941536,
      "grad_norm": 0.07323256880044937,
      "learning_rate": 2.4518143261317624e-06,
      "loss": 0.0013,
      "step": 219280
    },
    {
      "epoch": 1.4064579438353222,
      "grad_norm": 0.04394732415676117,
      "learning_rate": 2.45133277922423e-06,
      "loss": 0.0011,
      "step": 219290
    },
    {
      "epoch": 1.4065220807291081,
      "grad_norm": 0.04408658668398857,
      "learning_rate": 2.45085126425332e-06,
      "loss": 0.0008,
      "step": 219300
    },
    {
      "epoch": 1.4065862176228943,
      "grad_norm": 0.10613957792520523,
      "learning_rate": 2.4503697812250633e-06,
      "loss": 0.0009,
      "step": 219310
    },
    {
      "epoch": 1.4066503545166804,
      "grad_norm": 0.26895952224731445,
      "learning_rate": 2.4498883301454972e-06,
      "loss": 0.0022,
      "step": 219320
    },
    {
      "epoch": 1.4067144914104666,
      "grad_norm": 0.05045659467577934,
      "learning_rate": 2.449406911020653e-06,
      "loss": 0.0007,
      "step": 219330
    },
    {
      "epoch": 1.4067786283042527,
      "grad_norm": 0.02221301943063736,
      "learning_rate": 2.448925523856561e-06,
      "loss": 0.0009,
      "step": 219340
    },
    {
      "epoch": 1.4068427651980386,
      "grad_norm": 0.0894397646188736,
      "learning_rate": 2.448444168659258e-06,
      "loss": 0.0016,
      "step": 219350
    },
    {
      "epoch": 1.4069069020918248,
      "grad_norm": 0.12538520991802216,
      "learning_rate": 2.4479628454347726e-06,
      "loss": 0.0017,
      "step": 219360
    },
    {
      "epoch": 1.406971038985611,
      "grad_norm": 0.05737454816699028,
      "learning_rate": 2.4474815541891366e-06,
      "loss": 0.001,
      "step": 219370
    },
    {
      "epoch": 1.4070351758793969,
      "grad_norm": 0.11879002302885056,
      "learning_rate": 2.4470002949283795e-06,
      "loss": 0.0009,
      "step": 219380
    },
    {
      "epoch": 1.407099312773183,
      "grad_norm": 0.14492501318454742,
      "learning_rate": 2.4465190676585343e-06,
      "loss": 0.0013,
      "step": 219390
    },
    {
      "epoch": 1.4071634496669692,
      "grad_norm": 0.07216792553663254,
      "learning_rate": 2.446037872385631e-06,
      "loss": 0.0014,
      "step": 219400
    },
    {
      "epoch": 1.4072275865607553,
      "grad_norm": 0.10150250792503357,
      "learning_rate": 2.445556709115698e-06,
      "loss": 0.0012,
      "step": 219410
    },
    {
      "epoch": 1.4072917234545415,
      "grad_norm": 0.11663208901882172,
      "learning_rate": 2.4450755778547637e-06,
      "loss": 0.0009,
      "step": 219420
    },
    {
      "epoch": 1.4073558603483276,
      "grad_norm": 0.11327947676181793,
      "learning_rate": 2.44459447860886e-06,
      "loss": 0.001,
      "step": 219430
    },
    {
      "epoch": 1.4074199972421135,
      "grad_norm": 0.08630412817001343,
      "learning_rate": 2.444113411384013e-06,
      "loss": 0.0013,
      "step": 219440
    },
    {
      "epoch": 1.4074841341358997,
      "grad_norm": 0.06219204142689705,
      "learning_rate": 2.443632376186253e-06,
      "loss": 0.0018,
      "step": 219450
    },
    {
      "epoch": 1.4075482710296858,
      "grad_norm": 0.28655245900154114,
      "learning_rate": 2.4431513730216034e-06,
      "loss": 0.0017,
      "step": 219460
    },
    {
      "epoch": 1.4076124079234718,
      "grad_norm": 0.18242192268371582,
      "learning_rate": 2.442670401896097e-06,
      "loss": 0.0025,
      "step": 219470
    },
    {
      "epoch": 1.407676544817258,
      "grad_norm": 0.16763143241405487,
      "learning_rate": 2.4421894628157577e-06,
      "loss": 0.0043,
      "step": 219480
    },
    {
      "epoch": 1.407740681711044,
      "grad_norm": 0.017439650371670723,
      "learning_rate": 2.441708555786611e-06,
      "loss": 0.0021,
      "step": 219490
    },
    {
      "epoch": 1.4078048186048302,
      "grad_norm": 0.029463430866599083,
      "learning_rate": 2.441227680814686e-06,
      "loss": 0.0019,
      "step": 219500
    },
    {
      "epoch": 1.4078689554986163,
      "grad_norm": 0.13877111673355103,
      "learning_rate": 2.440746837906007e-06,
      "loss": 0.0018,
      "step": 219510
    },
    {
      "epoch": 1.4079330923924023,
      "grad_norm": 0.20989437401294708,
      "learning_rate": 2.440266027066599e-06,
      "loss": 0.0022,
      "step": 219520
    },
    {
      "epoch": 1.4079972292861884,
      "grad_norm": 0.07056647539138794,
      "learning_rate": 2.439785248302486e-06,
      "loss": 0.0013,
      "step": 219530
    },
    {
      "epoch": 1.4080613661799746,
      "grad_norm": 0.1661432981491089,
      "learning_rate": 2.4393045016196946e-06,
      "loss": 0.0019,
      "step": 219540
    },
    {
      "epoch": 1.4081255030737605,
      "grad_norm": 0.035545654594898224,
      "learning_rate": 2.4388237870242483e-06,
      "loss": 0.0019,
      "step": 219550
    },
    {
      "epoch": 1.4081896399675466,
      "grad_norm": 0.04041499271988869,
      "learning_rate": 2.43834310452217e-06,
      "loss": 0.0008,
      "step": 219560
    },
    {
      "epoch": 1.4082537768613328,
      "grad_norm": 0.011176417581737041,
      "learning_rate": 2.4378624541194816e-06,
      "loss": 0.0011,
      "step": 219570
    },
    {
      "epoch": 1.408317913755119,
      "grad_norm": 0.1042177826166153,
      "learning_rate": 2.4373818358222102e-06,
      "loss": 0.0011,
      "step": 219580
    },
    {
      "epoch": 1.408382050648905,
      "grad_norm": 0.014329710975289345,
      "learning_rate": 2.436901249636375e-06,
      "loss": 0.0008,
      "step": 219590
    },
    {
      "epoch": 1.4084461875426912,
      "grad_norm": 0.08484010398387909,
      "learning_rate": 2.436420695568e-06,
      "loss": 0.0022,
      "step": 219600
    },
    {
      "epoch": 1.4085103244364772,
      "grad_norm": 0.14927707612514496,
      "learning_rate": 2.4359401736231035e-06,
      "loss": 0.0012,
      "step": 219610
    },
    {
      "epoch": 1.4085744613302633,
      "grad_norm": 0.0252826064825058,
      "learning_rate": 2.435459683807711e-06,
      "loss": 0.0013,
      "step": 219620
    },
    {
      "epoch": 1.4086385982240495,
      "grad_norm": 0.11428942531347275,
      "learning_rate": 2.4349792261278416e-06,
      "loss": 0.0017,
      "step": 219630
    },
    {
      "epoch": 1.4087027351178354,
      "grad_norm": 0.008819522336125374,
      "learning_rate": 2.434498800589514e-06,
      "loss": 0.0009,
      "step": 219640
    },
    {
      "epoch": 1.4087668720116215,
      "grad_norm": 0.20233199000358582,
      "learning_rate": 2.434018407198751e-06,
      "loss": 0.0008,
      "step": 219650
    },
    {
      "epoch": 1.4088310089054077,
      "grad_norm": 0.12796108424663544,
      "learning_rate": 2.433538045961572e-06,
      "loss": 0.001,
      "step": 219660
    },
    {
      "epoch": 1.4088951457991938,
      "grad_norm": 0.030231822282075882,
      "learning_rate": 2.4330577168839953e-06,
      "loss": 0.001,
      "step": 219670
    },
    {
      "epoch": 1.40895928269298,
      "grad_norm": 0.015168020501732826,
      "learning_rate": 2.4325774199720382e-06,
      "loss": 0.001,
      "step": 219680
    },
    {
      "epoch": 1.409023419586766,
      "grad_norm": 0.14559921622276306,
      "learning_rate": 2.4320971552317223e-06,
      "loss": 0.0011,
      "step": 219690
    },
    {
      "epoch": 1.409087556480552,
      "grad_norm": 0.07003036141395569,
      "learning_rate": 2.431616922669065e-06,
      "loss": 0.0012,
      "step": 219700
    },
    {
      "epoch": 1.4091516933743382,
      "grad_norm": 0.12697798013687134,
      "learning_rate": 2.431136722290083e-06,
      "loss": 0.0011,
      "step": 219710
    },
    {
      "epoch": 1.4092158302681244,
      "grad_norm": 0.039292674511671066,
      "learning_rate": 2.4306565541007914e-06,
      "loss": 0.0015,
      "step": 219720
    },
    {
      "epoch": 1.4092799671619103,
      "grad_norm": 0.022258060052990913,
      "learning_rate": 2.4301764181072118e-06,
      "loss": 0.0012,
      "step": 219730
    },
    {
      "epoch": 1.4093441040556964,
      "grad_norm": 0.055856283754110336,
      "learning_rate": 2.4296963143153586e-06,
      "loss": 0.0009,
      "step": 219740
    },
    {
      "epoch": 1.4094082409494826,
      "grad_norm": 0.2606965899467468,
      "learning_rate": 2.429216242731247e-06,
      "loss": 0.0024,
      "step": 219750
    },
    {
      "epoch": 1.4094723778432687,
      "grad_norm": 0.08281605690717697,
      "learning_rate": 2.428736203360892e-06,
      "loss": 0.0012,
      "step": 219760
    },
    {
      "epoch": 1.4095365147370549,
      "grad_norm": 0.007565287407487631,
      "learning_rate": 2.428256196210311e-06,
      "loss": 0.0018,
      "step": 219770
    },
    {
      "epoch": 1.4096006516308408,
      "grad_norm": 0.0865720584988594,
      "learning_rate": 2.4277762212855186e-06,
      "loss": 0.0009,
      "step": 219780
    },
    {
      "epoch": 1.409664788524627,
      "grad_norm": 0.005745246075093746,
      "learning_rate": 2.4272962785925265e-06,
      "loss": 0.0036,
      "step": 219790
    },
    {
      "epoch": 1.409728925418413,
      "grad_norm": 0.05915892496705055,
      "learning_rate": 2.4268163681373526e-06,
      "loss": 0.0013,
      "step": 219800
    },
    {
      "epoch": 1.409793062312199,
      "grad_norm": 0.007312369532883167,
      "learning_rate": 2.4263364899260083e-06,
      "loss": 0.0008,
      "step": 219810
    },
    {
      "epoch": 1.4098571992059852,
      "grad_norm": 0.09697067737579346,
      "learning_rate": 2.425856643964507e-06,
      "loss": 0.0016,
      "step": 219820
    },
    {
      "epoch": 1.4099213360997713,
      "grad_norm": 0.10934215039014816,
      "learning_rate": 2.4253768302588606e-06,
      "loss": 0.0006,
      "step": 219830
    },
    {
      "epoch": 1.4099854729935575,
      "grad_norm": 0.05217369273304939,
      "learning_rate": 2.4248970488150834e-06,
      "loss": 0.0008,
      "step": 219840
    },
    {
      "epoch": 1.4100496098873436,
      "grad_norm": 0.04329046979546547,
      "learning_rate": 2.4244172996391857e-06,
      "loss": 0.001,
      "step": 219850
    },
    {
      "epoch": 1.4101137467811298,
      "grad_norm": 0.06280362606048584,
      "learning_rate": 2.423937582737181e-06,
      "loss": 0.0011,
      "step": 219860
    },
    {
      "epoch": 1.4101778836749157,
      "grad_norm": 0.0004947282723151147,
      "learning_rate": 2.42345789811508e-06,
      "loss": 0.0008,
      "step": 219870
    },
    {
      "epoch": 1.4102420205687018,
      "grad_norm": 0.24201305210590363,
      "learning_rate": 2.422978245778891e-06,
      "loss": 0.0019,
      "step": 219880
    },
    {
      "epoch": 1.410306157462488,
      "grad_norm": 0.025508929044008255,
      "learning_rate": 2.4224986257346284e-06,
      "loss": 0.0024,
      "step": 219890
    },
    {
      "epoch": 1.410370294356274,
      "grad_norm": 0.10216487944126129,
      "learning_rate": 2.4220190379882996e-06,
      "loss": 0.0011,
      "step": 219900
    },
    {
      "epoch": 1.41043443125006,
      "grad_norm": 0.053904447704553604,
      "learning_rate": 2.421539482545913e-06,
      "loss": 0.0011,
      "step": 219910
    },
    {
      "epoch": 1.4104985681438462,
      "grad_norm": 0.0518166646361351,
      "learning_rate": 2.4210599594134822e-06,
      "loss": 0.002,
      "step": 219920
    },
    {
      "epoch": 1.4105627050376324,
      "grad_norm": 0.10732424259185791,
      "learning_rate": 2.4205804685970126e-06,
      "loss": 0.0012,
      "step": 219930
    },
    {
      "epoch": 1.4106268419314185,
      "grad_norm": 0.07041086256504059,
      "learning_rate": 2.420101010102513e-06,
      "loss": 0.0012,
      "step": 219940
    },
    {
      "epoch": 1.4106909788252044,
      "grad_norm": 0.05897292494773865,
      "learning_rate": 2.4196215839359898e-06,
      "loss": 0.0031,
      "step": 219950
    },
    {
      "epoch": 1.4107551157189906,
      "grad_norm": 0.03396107628941536,
      "learning_rate": 2.4191421901034547e-06,
      "loss": 0.0023,
      "step": 219960
    },
    {
      "epoch": 1.4108192526127767,
      "grad_norm": 0.14326755702495575,
      "learning_rate": 2.4186628286109116e-06,
      "loss": 0.0033,
      "step": 219970
    },
    {
      "epoch": 1.410883389506563,
      "grad_norm": 0.10896255820989609,
      "learning_rate": 2.418183499464369e-06,
      "loss": 0.0014,
      "step": 219980
    },
    {
      "epoch": 1.4109475264003488,
      "grad_norm": 0.09689708054065704,
      "learning_rate": 2.4177042026698294e-06,
      "loss": 0.0013,
      "step": 219990
    },
    {
      "epoch": 1.411011663294135,
      "grad_norm": 0.04778188467025757,
      "learning_rate": 2.4172249382333046e-06,
      "loss": 0.0029,
      "step": 220000
    },
    {
      "epoch": 1.4110758001879211,
      "grad_norm": 0.04208539426326752,
      "learning_rate": 2.416745706160797e-06,
      "loss": 0.0012,
      "step": 220010
    },
    {
      "epoch": 1.4111399370817073,
      "grad_norm": 0.05550342798233032,
      "learning_rate": 2.4162665064583112e-06,
      "loss": 0.0012,
      "step": 220020
    },
    {
      "epoch": 1.4112040739754934,
      "grad_norm": 0.08074969053268433,
      "learning_rate": 2.4157873391318514e-06,
      "loss": 0.002,
      "step": 220030
    },
    {
      "epoch": 1.4112682108692793,
      "grad_norm": 0.10883128643035889,
      "learning_rate": 2.4153082041874248e-06,
      "loss": 0.001,
      "step": 220040
    },
    {
      "epoch": 1.4113323477630655,
      "grad_norm": 0.11793199181556702,
      "learning_rate": 2.4148291016310338e-06,
      "loss": 0.0012,
      "step": 220050
    },
    {
      "epoch": 1.4113964846568516,
      "grad_norm": 0.030807217583060265,
      "learning_rate": 2.4143500314686797e-06,
      "loss": 0.001,
      "step": 220060
    },
    {
      "epoch": 1.4114606215506376,
      "grad_norm": 0.09902536869049072,
      "learning_rate": 2.41387099370637e-06,
      "loss": 0.0009,
      "step": 220070
    },
    {
      "epoch": 1.4115247584444237,
      "grad_norm": 0.018539322540163994,
      "learning_rate": 2.4133919883501043e-06,
      "loss": 0.0019,
      "step": 220080
    },
    {
      "epoch": 1.4115888953382099,
      "grad_norm": 0.004996740259230137,
      "learning_rate": 2.412913015405886e-06,
      "loss": 0.0007,
      "step": 220090
    },
    {
      "epoch": 1.411653032231996,
      "grad_norm": 0.1072591245174408,
      "learning_rate": 2.412434074879715e-06,
      "loss": 0.0014,
      "step": 220100
    },
    {
      "epoch": 1.4117171691257822,
      "grad_norm": 0.23977114260196686,
      "learning_rate": 2.411955166777596e-06,
      "loss": 0.0017,
      "step": 220110
    },
    {
      "epoch": 1.4117813060195683,
      "grad_norm": 0.08273182064294815,
      "learning_rate": 2.4114762911055282e-06,
      "loss": 0.0016,
      "step": 220120
    },
    {
      "epoch": 1.4118454429133542,
      "grad_norm": 0.03678559139370918,
      "learning_rate": 2.4109974478695124e-06,
      "loss": 0.0015,
      "step": 220130
    },
    {
      "epoch": 1.4119095798071404,
      "grad_norm": 0.16956031322479248,
      "learning_rate": 2.4105186370755475e-06,
      "loss": 0.001,
      "step": 220140
    },
    {
      "epoch": 1.4119737167009265,
      "grad_norm": 0.03240702301263809,
      "learning_rate": 2.4100398587296365e-06,
      "loss": 0.0018,
      "step": 220150
    },
    {
      "epoch": 1.4120378535947125,
      "grad_norm": 0.20417430996894836,
      "learning_rate": 2.4095611128377767e-06,
      "loss": 0.0009,
      "step": 220160
    },
    {
      "epoch": 1.4121019904884986,
      "grad_norm": 0.031480804085731506,
      "learning_rate": 2.4090823994059674e-06,
      "loss": 0.0007,
      "step": 220170
    },
    {
      "epoch": 1.4121661273822848,
      "grad_norm": 0.045793939381837845,
      "learning_rate": 2.4086037184402062e-06,
      "loss": 0.0009,
      "step": 220180
    },
    {
      "epoch": 1.412230264276071,
      "grad_norm": 0.051683228462934494,
      "learning_rate": 2.4081250699464937e-06,
      "loss": 0.0007,
      "step": 220190
    },
    {
      "epoch": 1.412294401169857,
      "grad_norm": 0.39436984062194824,
      "learning_rate": 2.407646453930827e-06,
      "loss": 0.0011,
      "step": 220200
    },
    {
      "epoch": 1.412358538063643,
      "grad_norm": 0.036083146929740906,
      "learning_rate": 2.407167870399201e-06,
      "loss": 0.0011,
      "step": 220210
    },
    {
      "epoch": 1.4124226749574291,
      "grad_norm": 0.10143862664699554,
      "learning_rate": 2.4066893193576157e-06,
      "loss": 0.0008,
      "step": 220220
    },
    {
      "epoch": 1.4124868118512153,
      "grad_norm": 0.1924578696489334,
      "learning_rate": 2.4062108008120676e-06,
      "loss": 0.0018,
      "step": 220230
    },
    {
      "epoch": 1.4125509487450012,
      "grad_norm": 0.09693806618452072,
      "learning_rate": 2.4057323147685513e-06,
      "loss": 0.0015,
      "step": 220240
    },
    {
      "epoch": 1.4126150856387873,
      "grad_norm": 0.17619289457798004,
      "learning_rate": 2.405253861233062e-06,
      "loss": 0.0009,
      "step": 220250
    },
    {
      "epoch": 1.4126792225325735,
      "grad_norm": 0.03629330173134804,
      "learning_rate": 2.4047754402115974e-06,
      "loss": 0.0012,
      "step": 220260
    },
    {
      "epoch": 1.4127433594263596,
      "grad_norm": 0.06974545866250992,
      "learning_rate": 2.4042970517101516e-06,
      "loss": 0.001,
      "step": 220270
    },
    {
      "epoch": 1.4128074963201458,
      "grad_norm": 0.17821314930915833,
      "learning_rate": 2.403818695734719e-06,
      "loss": 0.0007,
      "step": 220280
    },
    {
      "epoch": 1.412871633213932,
      "grad_norm": 0.02364698238670826,
      "learning_rate": 2.403340372291292e-06,
      "loss": 0.0008,
      "step": 220290
    },
    {
      "epoch": 1.4129357701077179,
      "grad_norm": 0.00746135413646698,
      "learning_rate": 2.402862081385867e-06,
      "loss": 0.0025,
      "step": 220300
    },
    {
      "epoch": 1.412999907001504,
      "grad_norm": 0.13213053345680237,
      "learning_rate": 2.402383823024437e-06,
      "loss": 0.001,
      "step": 220310
    },
    {
      "epoch": 1.4130640438952902,
      "grad_norm": 0.04691189527511597,
      "learning_rate": 2.4019055972129933e-06,
      "loss": 0.0007,
      "step": 220320
    },
    {
      "epoch": 1.413128180789076,
      "grad_norm": 0.03543318808078766,
      "learning_rate": 2.401427403957528e-06,
      "loss": 0.001,
      "step": 220330
    },
    {
      "epoch": 1.4131923176828622,
      "grad_norm": 0.19499273598194122,
      "learning_rate": 2.400949243264037e-06,
      "loss": 0.0028,
      "step": 220340
    },
    {
      "epoch": 1.4132564545766484,
      "grad_norm": 0.023899951949715614,
      "learning_rate": 2.4004711151385086e-06,
      "loss": 0.0021,
      "step": 220350
    },
    {
      "epoch": 1.4133205914704345,
      "grad_norm": 0.06745419651269913,
      "learning_rate": 2.399993019586933e-06,
      "loss": 0.0029,
      "step": 220360
    },
    {
      "epoch": 1.4133847283642207,
      "grad_norm": 0.37230822443962097,
      "learning_rate": 2.399514956615306e-06,
      "loss": 0.0016,
      "step": 220370
    },
    {
      "epoch": 1.4134488652580066,
      "grad_norm": 0.03617741912603378,
      "learning_rate": 2.3990369262296143e-06,
      "loss": 0.0016,
      "step": 220380
    },
    {
      "epoch": 1.4135130021517928,
      "grad_norm": 0.17530429363250732,
      "learning_rate": 2.398558928435849e-06,
      "loss": 0.0013,
      "step": 220390
    },
    {
      "epoch": 1.413577139045579,
      "grad_norm": 0.11301398277282715,
      "learning_rate": 2.3980809632399975e-06,
      "loss": 0.0015,
      "step": 220400
    },
    {
      "epoch": 1.413641275939365,
      "grad_norm": 0.056706465780735016,
      "learning_rate": 2.397603030648053e-06,
      "loss": 0.0016,
      "step": 220410
    },
    {
      "epoch": 1.413705412833151,
      "grad_norm": 0.09726312756538391,
      "learning_rate": 2.3971251306660027e-06,
      "loss": 0.0017,
      "step": 220420
    },
    {
      "epoch": 1.4137695497269371,
      "grad_norm": 0.07375887036323547,
      "learning_rate": 2.3966472632998345e-06,
      "loss": 0.0015,
      "step": 220430
    },
    {
      "epoch": 1.4138336866207233,
      "grad_norm": 0.1622479259967804,
      "learning_rate": 2.3961694285555347e-06,
      "loss": 0.0012,
      "step": 220440
    },
    {
      "epoch": 1.4138978235145094,
      "grad_norm": 0.14091528952121735,
      "learning_rate": 2.3956916264390952e-06,
      "loss": 0.0012,
      "step": 220450
    },
    {
      "epoch": 1.4139619604082956,
      "grad_norm": 0.1288294792175293,
      "learning_rate": 2.3952138569565003e-06,
      "loss": 0.0015,
      "step": 220460
    },
    {
      "epoch": 1.4140260973020815,
      "grad_norm": 0.20033596456050873,
      "learning_rate": 2.394736120113736e-06,
      "loss": 0.0016,
      "step": 220470
    },
    {
      "epoch": 1.4140902341958677,
      "grad_norm": 0.15678802132606506,
      "learning_rate": 2.394258415916792e-06,
      "loss": 0.0025,
      "step": 220480
    },
    {
      "epoch": 1.4141543710896538,
      "grad_norm": 0.006190237123519182,
      "learning_rate": 2.393780744371652e-06,
      "loss": 0.0006,
      "step": 220490
    },
    {
      "epoch": 1.4142185079834397,
      "grad_norm": 0.029406633228063583,
      "learning_rate": 2.393303105484302e-06,
      "loss": 0.0031,
      "step": 220500
    },
    {
      "epoch": 1.4142826448772259,
      "grad_norm": 0.04776811972260475,
      "learning_rate": 2.3928254992607253e-06,
      "loss": 0.0004,
      "step": 220510
    },
    {
      "epoch": 1.414346781771012,
      "grad_norm": 0.09439283609390259,
      "learning_rate": 2.3923479257069105e-06,
      "loss": 0.0018,
      "step": 220520
    },
    {
      "epoch": 1.4144109186647982,
      "grad_norm": 0.0398576594889164,
      "learning_rate": 2.3918703848288396e-06,
      "loss": 0.0015,
      "step": 220530
    },
    {
      "epoch": 1.4144750555585843,
      "grad_norm": 0.06165533885359764,
      "learning_rate": 2.391392876632497e-06,
      "loss": 0.0009,
      "step": 220540
    },
    {
      "epoch": 1.4145391924523705,
      "grad_norm": 0.01763291470706463,
      "learning_rate": 2.3909154011238634e-06,
      "loss": 0.0006,
      "step": 220550
    },
    {
      "epoch": 1.4146033293461564,
      "grad_norm": 0.21147936582565308,
      "learning_rate": 2.390437958308927e-06,
      "loss": 0.0024,
      "step": 220560
    },
    {
      "epoch": 1.4146674662399426,
      "grad_norm": 0.24931780993938446,
      "learning_rate": 2.3899605481936678e-06,
      "loss": 0.0014,
      "step": 220570
    },
    {
      "epoch": 1.4147316031337287,
      "grad_norm": 0.11878237873315811,
      "learning_rate": 2.3894831707840682e-06,
      "loss": 0.0011,
      "step": 220580
    },
    {
      "epoch": 1.4147957400275146,
      "grad_norm": 0.08321509510278702,
      "learning_rate": 2.3890058260861086e-06,
      "loss": 0.0018,
      "step": 220590
    },
    {
      "epoch": 1.4148598769213008,
      "grad_norm": 0.425908625125885,
      "learning_rate": 2.388528514105774e-06,
      "loss": 0.0012,
      "step": 220600
    },
    {
      "epoch": 1.414924013815087,
      "grad_norm": 0.09089425206184387,
      "learning_rate": 2.3880512348490412e-06,
      "loss": 0.0023,
      "step": 220610
    },
    {
      "epoch": 1.414988150708873,
      "grad_norm": 0.07154347002506256,
      "learning_rate": 2.3875739883218953e-06,
      "loss": 0.0014,
      "step": 220620
    },
    {
      "epoch": 1.4150522876026592,
      "grad_norm": 0.12015087157487869,
      "learning_rate": 2.3870967745303124e-06,
      "loss": 0.001,
      "step": 220630
    },
    {
      "epoch": 1.4151164244964451,
      "grad_norm": 0.17344476282596588,
      "learning_rate": 2.3866195934802765e-06,
      "loss": 0.0014,
      "step": 220640
    },
    {
      "epoch": 1.4151805613902313,
      "grad_norm": 0.024174202233552933,
      "learning_rate": 2.3861424451777643e-06,
      "loss": 0.0013,
      "step": 220650
    },
    {
      "epoch": 1.4152446982840174,
      "grad_norm": 0.06502870470285416,
      "learning_rate": 2.3856653296287557e-06,
      "loss": 0.0015,
      "step": 220660
    },
    {
      "epoch": 1.4153088351778034,
      "grad_norm": 0.0520755797624588,
      "learning_rate": 2.3851882468392275e-06,
      "loss": 0.0007,
      "step": 220670
    },
    {
      "epoch": 1.4153729720715895,
      "grad_norm": 0.10467202961444855,
      "learning_rate": 2.3847111968151605e-06,
      "loss": 0.001,
      "step": 220680
    },
    {
      "epoch": 1.4154371089653757,
      "grad_norm": 0.07006729394197464,
      "learning_rate": 2.384234179562532e-06,
      "loss": 0.0013,
      "step": 220690
    },
    {
      "epoch": 1.4155012458591618,
      "grad_norm": 0.05697028711438179,
      "learning_rate": 2.383757195087318e-06,
      "loss": 0.0016,
      "step": 220700
    },
    {
      "epoch": 1.415565382752948,
      "grad_norm": 0.07601667940616608,
      "learning_rate": 2.383280243395495e-06,
      "loss": 0.0011,
      "step": 220710
    },
    {
      "epoch": 1.4156295196467341,
      "grad_norm": 0.1048382818698883,
      "learning_rate": 2.3828033244930423e-06,
      "loss": 0.0011,
      "step": 220720
    },
    {
      "epoch": 1.41569365654052,
      "grad_norm": 0.04876832291483879,
      "learning_rate": 2.382326438385934e-06,
      "loss": 0.003,
      "step": 220730
    },
    {
      "epoch": 1.4157577934343062,
      "grad_norm": 0.12612700462341309,
      "learning_rate": 2.381849585080144e-06,
      "loss": 0.0009,
      "step": 220740
    },
    {
      "epoch": 1.4158219303280923,
      "grad_norm": 0.08230948448181152,
      "learning_rate": 2.381372764581653e-06,
      "loss": 0.0018,
      "step": 220750
    },
    {
      "epoch": 1.4158860672218783,
      "grad_norm": 0.21726293861865997,
      "learning_rate": 2.3808959768964317e-06,
      "loss": 0.002,
      "step": 220760
    },
    {
      "epoch": 1.4159502041156644,
      "grad_norm": 0.13905607163906097,
      "learning_rate": 2.3804192220304565e-06,
      "loss": 0.0011,
      "step": 220770
    },
    {
      "epoch": 1.4160143410094506,
      "grad_norm": 0.07558777183294296,
      "learning_rate": 2.3799424999896983e-06,
      "loss": 0.0014,
      "step": 220780
    },
    {
      "epoch": 1.4160784779032367,
      "grad_norm": 0.08255726844072342,
      "learning_rate": 2.379465810780135e-06,
      "loss": 0.001,
      "step": 220790
    },
    {
      "epoch": 1.4161426147970229,
      "grad_norm": 0.01143329031765461,
      "learning_rate": 2.378989154407738e-06,
      "loss": 0.0012,
      "step": 220800
    },
    {
      "epoch": 1.4162067516908088,
      "grad_norm": 0.18541577458381653,
      "learning_rate": 2.37851253087848e-06,
      "loss": 0.0018,
      "step": 220810
    },
    {
      "epoch": 1.416270888584595,
      "grad_norm": 0.10772611200809479,
      "learning_rate": 2.378035940198332e-06,
      "loss": 0.003,
      "step": 220820
    },
    {
      "epoch": 1.416335025478381,
      "grad_norm": 0.12028023600578308,
      "learning_rate": 2.377559382373269e-06,
      "loss": 0.0026,
      "step": 220830
    },
    {
      "epoch": 1.4163991623721672,
      "grad_norm": 0.052633851766586304,
      "learning_rate": 2.3770828574092614e-06,
      "loss": 0.0013,
      "step": 220840
    },
    {
      "epoch": 1.4164632992659532,
      "grad_norm": 0.10359262675046921,
      "learning_rate": 2.3766063653122807e-06,
      "loss": 0.0025,
      "step": 220850
    },
    {
      "epoch": 1.4165274361597393,
      "grad_norm": 0.04219682142138481,
      "learning_rate": 2.376129906088294e-06,
      "loss": 0.0011,
      "step": 220860
    },
    {
      "epoch": 1.4165915730535255,
      "grad_norm": 0.08940007537603378,
      "learning_rate": 2.375653479743278e-06,
      "loss": 0.0014,
      "step": 220870
    },
    {
      "epoch": 1.4166557099473116,
      "grad_norm": 0.0020488423760980368,
      "learning_rate": 2.3751770862831985e-06,
      "loss": 0.002,
      "step": 220880
    },
    {
      "epoch": 1.4167198468410978,
      "grad_norm": 0.06613387912511826,
      "learning_rate": 2.374700725714025e-06,
      "loss": 0.0009,
      "step": 220890
    },
    {
      "epoch": 1.4167839837348837,
      "grad_norm": 0.04910537600517273,
      "learning_rate": 2.374224398041729e-06,
      "loss": 0.0013,
      "step": 220900
    },
    {
      "epoch": 1.4168481206286698,
      "grad_norm": 0.09544882923364639,
      "learning_rate": 2.3737481032722782e-06,
      "loss": 0.0016,
      "step": 220910
    },
    {
      "epoch": 1.416912257522456,
      "grad_norm": 0.01734582521021366,
      "learning_rate": 2.3732718414116403e-06,
      "loss": 0.0007,
      "step": 220920
    },
    {
      "epoch": 1.416976394416242,
      "grad_norm": 0.06376821547746658,
      "learning_rate": 2.372795612465782e-06,
      "loss": 0.001,
      "step": 220930
    },
    {
      "epoch": 1.417040531310028,
      "grad_norm": 0.08851666003465652,
      "learning_rate": 2.372319416440674e-06,
      "loss": 0.0021,
      "step": 220940
    },
    {
      "epoch": 1.4171046682038142,
      "grad_norm": 0.06368345767259598,
      "learning_rate": 2.3718432533422813e-06,
      "loss": 0.0018,
      "step": 220950
    },
    {
      "epoch": 1.4171688050976003,
      "grad_norm": 0.06570161134004593,
      "learning_rate": 2.3713671231765718e-06,
      "loss": 0.0012,
      "step": 220960
    },
    {
      "epoch": 1.4172329419913865,
      "grad_norm": 0.08560293912887573,
      "learning_rate": 2.3708910259495082e-06,
      "loss": 0.0005,
      "step": 220970
    },
    {
      "epoch": 1.4172970788851726,
      "grad_norm": 0.15812426805496216,
      "learning_rate": 2.3704149616670615e-06,
      "loss": 0.0012,
      "step": 220980
    },
    {
      "epoch": 1.4173612157789586,
      "grad_norm": 0.309256374835968,
      "learning_rate": 2.3699389303351946e-06,
      "loss": 0.0012,
      "step": 220990
    },
    {
      "epoch": 1.4174253526727447,
      "grad_norm": 0.03634125366806984,
      "learning_rate": 2.3694629319598724e-06,
      "loss": 0.0011,
      "step": 221000
    },
    {
      "epoch": 1.4174894895665309,
      "grad_norm": 0.11074472218751907,
      "learning_rate": 2.3689869665470575e-06,
      "loss": 0.0015,
      "step": 221010
    },
    {
      "epoch": 1.4175536264603168,
      "grad_norm": 0.01838160865008831,
      "learning_rate": 2.3685110341027183e-06,
      "loss": 0.001,
      "step": 221020
    },
    {
      "epoch": 1.417617763354103,
      "grad_norm": 0.16619424521923065,
      "learning_rate": 2.368035134632817e-06,
      "loss": 0.0018,
      "step": 221030
    },
    {
      "epoch": 1.417681900247889,
      "grad_norm": 0.10735753178596497,
      "learning_rate": 2.3675592681433142e-06,
      "loss": 0.001,
      "step": 221040
    },
    {
      "epoch": 1.4177460371416752,
      "grad_norm": 0.0535336509346962,
      "learning_rate": 2.3670834346401773e-06,
      "loss": 0.0011,
      "step": 221050
    },
    {
      "epoch": 1.4178101740354614,
      "grad_norm": 0.004761670250445604,
      "learning_rate": 2.3666076341293664e-06,
      "loss": 0.0013,
      "step": 221060
    },
    {
      "epoch": 1.4178743109292473,
      "grad_norm": 0.043905891478061676,
      "learning_rate": 2.3661318666168444e-06,
      "loss": 0.0009,
      "step": 221070
    },
    {
      "epoch": 1.4179384478230335,
      "grad_norm": 0.06458326429128647,
      "learning_rate": 2.36565613210857e-06,
      "loss": 0.0011,
      "step": 221080
    },
    {
      "epoch": 1.4180025847168196,
      "grad_norm": 0.24945113062858582,
      "learning_rate": 2.3651804306105094e-06,
      "loss": 0.0016,
      "step": 221090
    },
    {
      "epoch": 1.4180667216106055,
      "grad_norm": 0.025887012481689453,
      "learning_rate": 2.364704762128621e-06,
      "loss": 0.0014,
      "step": 221100
    },
    {
      "epoch": 1.4181308585043917,
      "grad_norm": 0.04768415540456772,
      "learning_rate": 2.364229126668865e-06,
      "loss": 0.001,
      "step": 221110
    },
    {
      "epoch": 1.4181949953981778,
      "grad_norm": 0.14740866422653198,
      "learning_rate": 2.3637535242372002e-06,
      "loss": 0.001,
      "step": 221120
    },
    {
      "epoch": 1.418259132291964,
      "grad_norm": 0.05745894834399223,
      "learning_rate": 2.363277954839589e-06,
      "loss": 0.0015,
      "step": 221130
    },
    {
      "epoch": 1.4183232691857501,
      "grad_norm": 0.07447264343500137,
      "learning_rate": 2.3628024184819903e-06,
      "loss": 0.0011,
      "step": 221140
    },
    {
      "epoch": 1.4183874060795363,
      "grad_norm": 0.03310972824692726,
      "learning_rate": 2.362326915170362e-06,
      "loss": 0.0026,
      "step": 221150
    },
    {
      "epoch": 1.4184515429733222,
      "grad_norm": 0.04251949116587639,
      "learning_rate": 2.3618514449106606e-06,
      "loss": 0.0013,
      "step": 221160
    },
    {
      "epoch": 1.4185156798671084,
      "grad_norm": 0.03919053077697754,
      "learning_rate": 2.361376007708848e-06,
      "loss": 0.0016,
      "step": 221170
    },
    {
      "epoch": 1.4185798167608945,
      "grad_norm": 0.04729697108268738,
      "learning_rate": 2.3609006035708793e-06,
      "loss": 0.0009,
      "step": 221180
    },
    {
      "epoch": 1.4186439536546804,
      "grad_norm": 0.06630100309848785,
      "learning_rate": 2.3604252325027106e-06,
      "loss": 0.0011,
      "step": 221190
    },
    {
      "epoch": 1.4187080905484666,
      "grad_norm": 0.0515078529715538,
      "learning_rate": 2.359949894510302e-06,
      "loss": 0.0005,
      "step": 221200
    },
    {
      "epoch": 1.4187722274422527,
      "grad_norm": 0.07539720833301544,
      "learning_rate": 2.3594745895996083e-06,
      "loss": 0.0012,
      "step": 221210
    },
    {
      "epoch": 1.4188363643360389,
      "grad_norm": 0.022874781861901283,
      "learning_rate": 2.3589993177765854e-06,
      "loss": 0.0021,
      "step": 221220
    },
    {
      "epoch": 1.418900501229825,
      "grad_norm": 0.03904331475496292,
      "learning_rate": 2.3585240790471862e-06,
      "loss": 0.0006,
      "step": 221230
    },
    {
      "epoch": 1.418964638123611,
      "grad_norm": 0.07771909236907959,
      "learning_rate": 2.3580488734173703e-06,
      "loss": 0.0008,
      "step": 221240
    },
    {
      "epoch": 1.419028775017397,
      "grad_norm": 0.08516875654459,
      "learning_rate": 2.35757370089309e-06,
      "loss": 0.0005,
      "step": 221250
    },
    {
      "epoch": 1.4190929119111833,
      "grad_norm": 0.053227923810482025,
      "learning_rate": 2.3570985614803e-06,
      "loss": 0.0016,
      "step": 221260
    },
    {
      "epoch": 1.4191570488049694,
      "grad_norm": 0.12357582151889801,
      "learning_rate": 2.3566234551849514e-06,
      "loss": 0.0011,
      "step": 221270
    },
    {
      "epoch": 1.4192211856987553,
      "grad_norm": 0.1287904679775238,
      "learning_rate": 2.356148382013002e-06,
      "loss": 0.0012,
      "step": 221280
    },
    {
      "epoch": 1.4192853225925415,
      "grad_norm": 0.16033320128917694,
      "learning_rate": 2.3556733419704036e-06,
      "loss": 0.0107,
      "step": 221290
    },
    {
      "epoch": 1.4193494594863276,
      "grad_norm": 0.08888640999794006,
      "learning_rate": 2.355198335063108e-06,
      "loss": 0.0022,
      "step": 221300
    },
    {
      "epoch": 1.4194135963801138,
      "grad_norm": 0.13831044733524323,
      "learning_rate": 2.354723361297065e-06,
      "loss": 0.0012,
      "step": 221310
    },
    {
      "epoch": 1.4194777332739,
      "grad_norm": 0.10133393108844757,
      "learning_rate": 2.3542484206782313e-06,
      "loss": 0.0024,
      "step": 221320
    },
    {
      "epoch": 1.4195418701676858,
      "grad_norm": 0.1301027089357376,
      "learning_rate": 2.3537735132125555e-06,
      "loss": 0.001,
      "step": 221330
    },
    {
      "epoch": 1.419606007061472,
      "grad_norm": 0.22867679595947266,
      "learning_rate": 2.3532986389059868e-06,
      "loss": 0.0009,
      "step": 221340
    },
    {
      "epoch": 1.4196701439552581,
      "grad_norm": 0.036261752247810364,
      "learning_rate": 2.35282379776448e-06,
      "loss": 0.0005,
      "step": 221350
    },
    {
      "epoch": 1.419734280849044,
      "grad_norm": 0.02306281588971615,
      "learning_rate": 2.352348989793983e-06,
      "loss": 0.0019,
      "step": 221360
    },
    {
      "epoch": 1.4197984177428302,
      "grad_norm": 0.1047319546341896,
      "learning_rate": 2.3518742150004447e-06,
      "loss": 0.0012,
      "step": 221370
    },
    {
      "epoch": 1.4198625546366164,
      "grad_norm": 0.07301588356494904,
      "learning_rate": 2.351399473389813e-06,
      "loss": 0.0007,
      "step": 221380
    },
    {
      "epoch": 1.4199266915304025,
      "grad_norm": 0.2079521119594574,
      "learning_rate": 2.35092476496804e-06,
      "loss": 0.0008,
      "step": 221390
    },
    {
      "epoch": 1.4199908284241887,
      "grad_norm": 0.09328661859035492,
      "learning_rate": 2.350450089741074e-06,
      "loss": 0.002,
      "step": 221400
    },
    {
      "epoch": 1.4200549653179748,
      "grad_norm": 0.13493525981903076,
      "learning_rate": 2.3499754477148623e-06,
      "loss": 0.0011,
      "step": 221410
    },
    {
      "epoch": 1.4201191022117607,
      "grad_norm": 0.03143034875392914,
      "learning_rate": 2.3495008388953523e-06,
      "loss": 0.0006,
      "step": 221420
    },
    {
      "epoch": 1.420183239105547,
      "grad_norm": 0.10459790378808975,
      "learning_rate": 2.349026263288489e-06,
      "loss": 0.0019,
      "step": 221430
    },
    {
      "epoch": 1.420247375999333,
      "grad_norm": 0.06142476946115494,
      "learning_rate": 2.348551720900223e-06,
      "loss": 0.0011,
      "step": 221440
    },
    {
      "epoch": 1.420311512893119,
      "grad_norm": 0.3689160645008087,
      "learning_rate": 2.348077211736499e-06,
      "loss": 0.0022,
      "step": 221450
    },
    {
      "epoch": 1.4203756497869051,
      "grad_norm": 0.06635049730539322,
      "learning_rate": 2.347602735803261e-06,
      "loss": 0.0014,
      "step": 221460
    },
    {
      "epoch": 1.4204397866806913,
      "grad_norm": 0.07540370523929596,
      "learning_rate": 2.347128293106458e-06,
      "loss": 0.001,
      "step": 221470
    },
    {
      "epoch": 1.4205039235744774,
      "grad_norm": 0.12796327471733093,
      "learning_rate": 2.3466538836520337e-06,
      "loss": 0.0011,
      "step": 221480
    },
    {
      "epoch": 1.4205680604682636,
      "grad_norm": 0.1090424507856369,
      "learning_rate": 2.3461795074459327e-06,
      "loss": 0.0016,
      "step": 221490
    },
    {
      "epoch": 1.4206321973620495,
      "grad_norm": 0.0716809332370758,
      "learning_rate": 2.345705164494097e-06,
      "loss": 0.0009,
      "step": 221500
    },
    {
      "epoch": 1.4206963342558356,
      "grad_norm": 0.03105447068810463,
      "learning_rate": 2.345230854802475e-06,
      "loss": 0.001,
      "step": 221510
    },
    {
      "epoch": 1.4207604711496218,
      "grad_norm": 0.03507548198103905,
      "learning_rate": 2.344756578377007e-06,
      "loss": 0.0014,
      "step": 221520
    },
    {
      "epoch": 1.420824608043408,
      "grad_norm": 0.12229614704847336,
      "learning_rate": 2.344282335223637e-06,
      "loss": 0.0012,
      "step": 221530
    },
    {
      "epoch": 1.4208887449371939,
      "grad_norm": 0.19771310687065125,
      "learning_rate": 2.343808125348305e-06,
      "loss": 0.0021,
      "step": 221540
    },
    {
      "epoch": 1.42095288183098,
      "grad_norm": 0.21561577916145325,
      "learning_rate": 2.3433339487569586e-06,
      "loss": 0.0014,
      "step": 221550
    },
    {
      "epoch": 1.4210170187247662,
      "grad_norm": 0.006400718353688717,
      "learning_rate": 2.342859805455536e-06,
      "loss": 0.0011,
      "step": 221560
    },
    {
      "epoch": 1.4210811556185523,
      "grad_norm": 0.042118512094020844,
      "learning_rate": 2.3423856954499783e-06,
      "loss": 0.0028,
      "step": 221570
    },
    {
      "epoch": 1.4211452925123385,
      "grad_norm": 0.18844324350357056,
      "learning_rate": 2.341911618746226e-06,
      "loss": 0.0024,
      "step": 221580
    },
    {
      "epoch": 1.4212094294061244,
      "grad_norm": 0.15068097412586212,
      "learning_rate": 2.341437575350223e-06,
      "loss": 0.0012,
      "step": 221590
    },
    {
      "epoch": 1.4212735662999105,
      "grad_norm": 0.12804320454597473,
      "learning_rate": 2.3409635652679068e-06,
      "loss": 0.0013,
      "step": 221600
    },
    {
      "epoch": 1.4213377031936967,
      "grad_norm": 0.22675733268260956,
      "learning_rate": 2.3404895885052156e-06,
      "loss": 0.0025,
      "step": 221610
    },
    {
      "epoch": 1.4214018400874826,
      "grad_norm": 0.01598006673157215,
      "learning_rate": 2.340015645068093e-06,
      "loss": 0.0016,
      "step": 221620
    },
    {
      "epoch": 1.4214659769812688,
      "grad_norm": 0.0054878066293895245,
      "learning_rate": 2.3395417349624744e-06,
      "loss": 0.0011,
      "step": 221630
    },
    {
      "epoch": 1.421530113875055,
      "grad_norm": 0.0077287182211875916,
      "learning_rate": 2.3390678581943e-06,
      "loss": 0.0008,
      "step": 221640
    },
    {
      "epoch": 1.421594250768841,
      "grad_norm": 0.11687889695167542,
      "learning_rate": 2.3385940147695053e-06,
      "loss": 0.0013,
      "step": 221650
    },
    {
      "epoch": 1.4216583876626272,
      "grad_norm": 0.09536377340555191,
      "learning_rate": 2.338120204694031e-06,
      "loss": 0.0012,
      "step": 221660
    },
    {
      "epoch": 1.4217225245564131,
      "grad_norm": 0.05910412594676018,
      "learning_rate": 2.3376464279738133e-06,
      "loss": 0.001,
      "step": 221670
    },
    {
      "epoch": 1.4217866614501993,
      "grad_norm": 0.03483949229121208,
      "learning_rate": 2.3371726846147886e-06,
      "loss": 0.0018,
      "step": 221680
    },
    {
      "epoch": 1.4218507983439854,
      "grad_norm": 0.026624565944075584,
      "learning_rate": 2.3366989746228914e-06,
      "loss": 0.0007,
      "step": 221690
    },
    {
      "epoch": 1.4219149352377716,
      "grad_norm": 0.015865344554185867,
      "learning_rate": 2.3362252980040614e-06,
      "loss": 0.0018,
      "step": 221700
    },
    {
      "epoch": 1.4219790721315575,
      "grad_norm": 0.03469441086053848,
      "learning_rate": 2.3357516547642316e-06,
      "loss": 0.001,
      "step": 221710
    },
    {
      "epoch": 1.4220432090253436,
      "grad_norm": 0.14101850986480713,
      "learning_rate": 2.335278044909338e-06,
      "loss": 0.0012,
      "step": 221720
    },
    {
      "epoch": 1.4221073459191298,
      "grad_norm": 0.03780830278992653,
      "learning_rate": 2.3348044684453138e-06,
      "loss": 0.0013,
      "step": 221730
    },
    {
      "epoch": 1.422171482812916,
      "grad_norm": 0.044076595455408096,
      "learning_rate": 2.3343309253780953e-06,
      "loss": 0.0008,
      "step": 221740
    },
    {
      "epoch": 1.422235619706702,
      "grad_norm": 0.11094122380018234,
      "learning_rate": 2.3338574157136155e-06,
      "loss": 0.0016,
      "step": 221750
    },
    {
      "epoch": 1.422299756600488,
      "grad_norm": 0.1418447047472,
      "learning_rate": 2.3333839394578067e-06,
      "loss": 0.0019,
      "step": 221760
    },
    {
      "epoch": 1.4223638934942742,
      "grad_norm": 0.07506245374679565,
      "learning_rate": 2.3329104966166045e-06,
      "loss": 0.0007,
      "step": 221770
    },
    {
      "epoch": 1.4224280303880603,
      "grad_norm": 0.018780136480927467,
      "learning_rate": 2.3324370871959393e-06,
      "loss": 0.0009,
      "step": 221780
    },
    {
      "epoch": 1.4224921672818462,
      "grad_norm": 0.1984405517578125,
      "learning_rate": 2.3319637112017447e-06,
      "loss": 0.0018,
      "step": 221790
    },
    {
      "epoch": 1.4225563041756324,
      "grad_norm": 0.1646849513053894,
      "learning_rate": 2.3314903686399493e-06,
      "loss": 0.0012,
      "step": 221800
    },
    {
      "epoch": 1.4226204410694185,
      "grad_norm": 0.02109588496387005,
      "learning_rate": 2.3310170595164884e-06,
      "loss": 0.0011,
      "step": 221810
    },
    {
      "epoch": 1.4226845779632047,
      "grad_norm": 0.24100464582443237,
      "learning_rate": 2.3305437838372913e-06,
      "loss": 0.0017,
      "step": 221820
    },
    {
      "epoch": 1.4227487148569908,
      "grad_norm": 0.05594455450773239,
      "learning_rate": 2.3300705416082888e-06,
      "loss": 0.0009,
      "step": 221830
    },
    {
      "epoch": 1.422812851750777,
      "grad_norm": 0.14231975376605988,
      "learning_rate": 2.329597332835409e-06,
      "loss": 0.0014,
      "step": 221840
    },
    {
      "epoch": 1.422876988644563,
      "grad_norm": 0.008400573395192623,
      "learning_rate": 2.329124157524584e-06,
      "loss": 0.002,
      "step": 221850
    },
    {
      "epoch": 1.422941125538349,
      "grad_norm": 0.19097235798835754,
      "learning_rate": 2.3286510156817426e-06,
      "loss": 0.0027,
      "step": 221860
    },
    {
      "epoch": 1.4230052624321352,
      "grad_norm": 0.028585517778992653,
      "learning_rate": 2.3281779073128135e-06,
      "loss": 0.0009,
      "step": 221870
    },
    {
      "epoch": 1.4230693993259211,
      "grad_norm": 0.009553569369018078,
      "learning_rate": 2.3277048324237222e-06,
      "loss": 0.0009,
      "step": 221880
    },
    {
      "epoch": 1.4231335362197073,
      "grad_norm": 0.12681032717227936,
      "learning_rate": 2.327231791020402e-06,
      "loss": 0.0019,
      "step": 221890
    },
    {
      "epoch": 1.4231976731134934,
      "grad_norm": 0.00455677043646574,
      "learning_rate": 2.326758783108777e-06,
      "loss": 0.0013,
      "step": 221900
    },
    {
      "epoch": 1.4232618100072796,
      "grad_norm": 0.008080567233264446,
      "learning_rate": 2.326285808694773e-06,
      "loss": 0.0013,
      "step": 221910
    },
    {
      "epoch": 1.4233259469010657,
      "grad_norm": 0.0857100561261177,
      "learning_rate": 2.325812867784321e-06,
      "loss": 0.001,
      "step": 221920
    },
    {
      "epoch": 1.4233900837948517,
      "grad_norm": 0.19669803977012634,
      "learning_rate": 2.3253399603833448e-06,
      "loss": 0.0025,
      "step": 221930
    },
    {
      "epoch": 1.4234542206886378,
      "grad_norm": 0.10235214233398438,
      "learning_rate": 2.3248670864977706e-06,
      "loss": 0.0011,
      "step": 221940
    },
    {
      "epoch": 1.423518357582424,
      "grad_norm": 0.2035292088985443,
      "learning_rate": 2.324394246133522e-06,
      "loss": 0.0012,
      "step": 221950
    },
    {
      "epoch": 1.42358249447621,
      "grad_norm": 0.05004888400435448,
      "learning_rate": 2.323921439296527e-06,
      "loss": 0.0014,
      "step": 221960
    },
    {
      "epoch": 1.423646631369996,
      "grad_norm": 0.017358137294650078,
      "learning_rate": 2.323448665992709e-06,
      "loss": 0.0049,
      "step": 221970
    },
    {
      "epoch": 1.4237107682637822,
      "grad_norm": 0.022545376792550087,
      "learning_rate": 2.322975926227992e-06,
      "loss": 0.0011,
      "step": 221980
    },
    {
      "epoch": 1.4237749051575683,
      "grad_norm": 0.06279610842466354,
      "learning_rate": 2.322503220008299e-06,
      "loss": 0.0015,
      "step": 221990
    },
    {
      "epoch": 1.4238390420513545,
      "grad_norm": 0.0311025008559227,
      "learning_rate": 2.322030547339555e-06,
      "loss": 0.0006,
      "step": 222000
    },
    {
      "epoch": 1.4239031789451406,
      "grad_norm": 0.1442340612411499,
      "learning_rate": 2.321557908227682e-06,
      "loss": 0.0017,
      "step": 222010
    },
    {
      "epoch": 1.4239673158389266,
      "grad_norm": 0.05317877233028412,
      "learning_rate": 2.321085302678604e-06,
      "loss": 0.0014,
      "step": 222020
    },
    {
      "epoch": 1.4240314527327127,
      "grad_norm": 0.05801217257976532,
      "learning_rate": 2.320612730698239e-06,
      "loss": 0.0012,
      "step": 222030
    },
    {
      "epoch": 1.4240955896264988,
      "grad_norm": 0.15718373656272888,
      "learning_rate": 2.3201401922925132e-06,
      "loss": 0.0018,
      "step": 222040
    },
    {
      "epoch": 1.4241597265202848,
      "grad_norm": 0.040582649409770966,
      "learning_rate": 2.3196676874673464e-06,
      "loss": 0.001,
      "step": 222050
    },
    {
      "epoch": 1.424223863414071,
      "grad_norm": 0.0020285227801650763,
      "learning_rate": 2.3191952162286576e-06,
      "loss": 0.0009,
      "step": 222060
    },
    {
      "epoch": 1.424288000307857,
      "grad_norm": 0.04461637884378433,
      "learning_rate": 2.3187227785823696e-06,
      "loss": 0.0012,
      "step": 222070
    },
    {
      "epoch": 1.4243521372016432,
      "grad_norm": 0.07278832048177719,
      "learning_rate": 2.318250374534402e-06,
      "loss": 0.0021,
      "step": 222080
    },
    {
      "epoch": 1.4244162740954294,
      "grad_norm": 0.05430842563509941,
      "learning_rate": 2.3177780040906735e-06,
      "loss": 0.002,
      "step": 222090
    },
    {
      "epoch": 1.4244804109892155,
      "grad_norm": 0.1495872437953949,
      "learning_rate": 2.317305667257102e-06,
      "loss": 0.001,
      "step": 222100
    },
    {
      "epoch": 1.4245445478830014,
      "grad_norm": 0.10312050580978394,
      "learning_rate": 2.3168333640396097e-06,
      "loss": 0.0009,
      "step": 222110
    },
    {
      "epoch": 1.4246086847767876,
      "grad_norm": 0.1687595248222351,
      "learning_rate": 2.316361094444113e-06,
      "loss": 0.0008,
      "step": 222120
    },
    {
      "epoch": 1.4246728216705737,
      "grad_norm": 0.0017926287837326527,
      "learning_rate": 2.3158888584765295e-06,
      "loss": 0.0009,
      "step": 222130
    },
    {
      "epoch": 1.4247369585643597,
      "grad_norm": 0.020870963111519814,
      "learning_rate": 2.3154166561427748e-06,
      "loss": 0.0013,
      "step": 222140
    },
    {
      "epoch": 1.4248010954581458,
      "grad_norm": 0.03439674898982048,
      "learning_rate": 2.314944487448769e-06,
      "loss": 0.0007,
      "step": 222150
    },
    {
      "epoch": 1.424865232351932,
      "grad_norm": 0.07166893780231476,
      "learning_rate": 2.3144723524004286e-06,
      "loss": 0.0015,
      "step": 222160
    },
    {
      "epoch": 1.4249293692457181,
      "grad_norm": 0.02373315952718258,
      "learning_rate": 2.3140002510036693e-06,
      "loss": 0.001,
      "step": 222170
    },
    {
      "epoch": 1.4249935061395043,
      "grad_norm": 0.07971645891666412,
      "learning_rate": 2.3135281832644045e-06,
      "loss": 0.0009,
      "step": 222180
    },
    {
      "epoch": 1.4250576430332902,
      "grad_norm": 0.12441080808639526,
      "learning_rate": 2.3130561491885538e-06,
      "loss": 0.001,
      "step": 222190
    },
    {
      "epoch": 1.4251217799270763,
      "grad_norm": 0.09575378149747849,
      "learning_rate": 2.3125841487820293e-06,
      "loss": 0.0008,
      "step": 222200
    },
    {
      "epoch": 1.4251859168208625,
      "grad_norm": 0.0775391086935997,
      "learning_rate": 2.312112182050747e-06,
      "loss": 0.0011,
      "step": 222210
    },
    {
      "epoch": 1.4252500537146484,
      "grad_norm": 0.022984299808740616,
      "learning_rate": 2.3116402490006173e-06,
      "loss": 0.0014,
      "step": 222220
    },
    {
      "epoch": 1.4253141906084346,
      "grad_norm": 0.028220055624842644,
      "learning_rate": 2.311168349637559e-06,
      "loss": 0.0028,
      "step": 222230
    },
    {
      "epoch": 1.4253783275022207,
      "grad_norm": 0.05225247144699097,
      "learning_rate": 2.310696483967483e-06,
      "loss": 0.0028,
      "step": 222240
    },
    {
      "epoch": 1.4254424643960069,
      "grad_norm": 0.07785075157880783,
      "learning_rate": 2.310224651996302e-06,
      "loss": 0.0007,
      "step": 222250
    },
    {
      "epoch": 1.425506601289793,
      "grad_norm": 0.15400755405426025,
      "learning_rate": 2.3097528537299263e-06,
      "loss": 0.0017,
      "step": 222260
    },
    {
      "epoch": 1.4255707381835792,
      "grad_norm": 0.029253359884023666,
      "learning_rate": 2.309281089174272e-06,
      "loss": 0.002,
      "step": 222270
    },
    {
      "epoch": 1.425634875077365,
      "grad_norm": 0.0710631012916565,
      "learning_rate": 2.308809358335249e-06,
      "loss": 0.0012,
      "step": 222280
    },
    {
      "epoch": 1.4256990119711512,
      "grad_norm": 0.07595669478178024,
      "learning_rate": 2.3083376612187676e-06,
      "loss": 0.0014,
      "step": 222290
    },
    {
      "epoch": 1.4257631488649374,
      "grad_norm": 0.08065669238567352,
      "learning_rate": 2.307865997830737e-06,
      "loss": 0.0009,
      "step": 222300
    },
    {
      "epoch": 1.4258272857587233,
      "grad_norm": 0.04362192004919052,
      "learning_rate": 2.3073943681770723e-06,
      "loss": 0.0018,
      "step": 222310
    },
    {
      "epoch": 1.4258914226525095,
      "grad_norm": 0.07967459410429001,
      "learning_rate": 2.30692277226368e-06,
      "loss": 0.0011,
      "step": 222320
    },
    {
      "epoch": 1.4259555595462956,
      "grad_norm": 0.06728452444076538,
      "learning_rate": 2.3064512100964683e-06,
      "loss": 0.0023,
      "step": 222330
    },
    {
      "epoch": 1.4260196964400818,
      "grad_norm": 0.08936009556055069,
      "learning_rate": 2.3059796816813494e-06,
      "loss": 0.0012,
      "step": 222340
    },
    {
      "epoch": 1.426083833333868,
      "grad_norm": 0.06235896050930023,
      "learning_rate": 2.3055081870242314e-06,
      "loss": 0.0013,
      "step": 222350
    },
    {
      "epoch": 1.4261479702276538,
      "grad_norm": 0.16244564950466156,
      "learning_rate": 2.3050367261310207e-06,
      "loss": 0.0015,
      "step": 222360
    },
    {
      "epoch": 1.42621210712144,
      "grad_norm": 0.06734142452478409,
      "learning_rate": 2.3045652990076245e-06,
      "loss": 0.0017,
      "step": 222370
    },
    {
      "epoch": 1.4262762440152261,
      "grad_norm": 0.18772275745868683,
      "learning_rate": 2.3040939056599533e-06,
      "loss": 0.0011,
      "step": 222380
    },
    {
      "epoch": 1.4263403809090123,
      "grad_norm": 0.27170705795288086,
      "learning_rate": 2.3036225460939115e-06,
      "loss": 0.0012,
      "step": 222390
    },
    {
      "epoch": 1.4264045178027982,
      "grad_norm": 0.042087312787771225,
      "learning_rate": 2.303151220315407e-06,
      "loss": 0.0007,
      "step": 222400
    },
    {
      "epoch": 1.4264686546965843,
      "grad_norm": 0.08965945243835449,
      "learning_rate": 2.302679928330343e-06,
      "loss": 0.0011,
      "step": 222410
    },
    {
      "epoch": 1.4265327915903705,
      "grad_norm": 0.05479520559310913,
      "learning_rate": 2.3022086701446292e-06,
      "loss": 0.0009,
      "step": 222420
    },
    {
      "epoch": 1.4265969284841566,
      "grad_norm": 0.06495223194360733,
      "learning_rate": 2.301737445764169e-06,
      "loss": 0.0006,
      "step": 222430
    },
    {
      "epoch": 1.4266610653779428,
      "grad_norm": 0.17523230612277985,
      "learning_rate": 2.301266255194865e-06,
      "loss": 0.0015,
      "step": 222440
    },
    {
      "epoch": 1.4267252022717287,
      "grad_norm": 0.06626857817173004,
      "learning_rate": 2.3007950984426254e-06,
      "loss": 0.0017,
      "step": 222450
    },
    {
      "epoch": 1.4267893391655149,
      "grad_norm": 0.2895970046520233,
      "learning_rate": 2.300323975513352e-06,
      "loss": 0.003,
      "step": 222460
    },
    {
      "epoch": 1.426853476059301,
      "grad_norm": 0.06247478350996971,
      "learning_rate": 2.2998528864129495e-06,
      "loss": 0.0006,
      "step": 222470
    },
    {
      "epoch": 1.426917612953087,
      "grad_norm": 0.044517505913972855,
      "learning_rate": 2.2993818311473175e-06,
      "loss": 0.0012,
      "step": 222480
    },
    {
      "epoch": 1.426981749846873,
      "grad_norm": 0.05764401704072952,
      "learning_rate": 2.2989108097223635e-06,
      "loss": 0.0011,
      "step": 222490
    },
    {
      "epoch": 1.4270458867406592,
      "grad_norm": 0.041933346539735794,
      "learning_rate": 2.2984398221439873e-06,
      "loss": 0.0008,
      "step": 222500
    },
    {
      "epoch": 1.4271100236344454,
      "grad_norm": 0.06117698922753334,
      "learning_rate": 2.297968868418091e-06,
      "loss": 0.0012,
      "step": 222510
    },
    {
      "epoch": 1.4271741605282315,
      "grad_norm": 0.04828890040516853,
      "learning_rate": 2.2974979485505737e-06,
      "loss": 0.0021,
      "step": 222520
    },
    {
      "epoch": 1.4272382974220177,
      "grad_norm": 0.08611414581537247,
      "learning_rate": 2.2970270625473407e-06,
      "loss": 0.0026,
      "step": 222530
    },
    {
      "epoch": 1.4273024343158036,
      "grad_norm": 0.017762210220098495,
      "learning_rate": 2.29655621041429e-06,
      "loss": 0.0007,
      "step": 222540
    },
    {
      "epoch": 1.4273665712095898,
      "grad_norm": 0.1064884141087532,
      "learning_rate": 2.2960853921573227e-06,
      "loss": 0.001,
      "step": 222550
    },
    {
      "epoch": 1.427430708103376,
      "grad_norm": 0.08850418776273727,
      "learning_rate": 2.295614607782336e-06,
      "loss": 0.0031,
      "step": 222560
    },
    {
      "epoch": 1.4274948449971618,
      "grad_norm": 0.09420426189899445,
      "learning_rate": 2.295143857295233e-06,
      "loss": 0.002,
      "step": 222570
    },
    {
      "epoch": 1.427558981890948,
      "grad_norm": 0.03913057968020439,
      "learning_rate": 2.29467314070191e-06,
      "loss": 0.0008,
      "step": 222580
    },
    {
      "epoch": 1.4276231187847341,
      "grad_norm": 0.03345941752195358,
      "learning_rate": 2.2942024580082643e-06,
      "loss": 0.0016,
      "step": 222590
    },
    {
      "epoch": 1.4276872556785203,
      "grad_norm": 0.1600685566663742,
      "learning_rate": 2.2937318092201975e-06,
      "loss": 0.0014,
      "step": 222600
    },
    {
      "epoch": 1.4277513925723064,
      "grad_norm": 0.18811281025409698,
      "learning_rate": 2.2932611943436055e-06,
      "loss": 0.0024,
      "step": 222610
    },
    {
      "epoch": 1.4278155294660924,
      "grad_norm": 0.10699602216482162,
      "learning_rate": 2.2927906133843853e-06,
      "loss": 0.0008,
      "step": 222620
    },
    {
      "epoch": 1.4278796663598785,
      "grad_norm": 0.0679832175374031,
      "learning_rate": 2.2923200663484315e-06,
      "loss": 0.0013,
      "step": 222630
    },
    {
      "epoch": 1.4279438032536647,
      "grad_norm": 0.03156968206167221,
      "learning_rate": 2.2918495532416444e-06,
      "loss": 0.0014,
      "step": 222640
    },
    {
      "epoch": 1.4280079401474506,
      "grad_norm": 0.09393204748630524,
      "learning_rate": 2.291379074069918e-06,
      "loss": 0.0007,
      "step": 222650
    },
    {
      "epoch": 1.4280720770412367,
      "grad_norm": 0.04491729289293289,
      "learning_rate": 2.290908628839147e-06,
      "loss": 0.0011,
      "step": 222660
    },
    {
      "epoch": 1.4281362139350229,
      "grad_norm": 0.3788091540336609,
      "learning_rate": 2.290438217555226e-06,
      "loss": 0.0053,
      "step": 222670
    },
    {
      "epoch": 1.428200350828809,
      "grad_norm": 0.08632595092058182,
      "learning_rate": 2.2899678402240516e-06,
      "loss": 0.001,
      "step": 222680
    },
    {
      "epoch": 1.4282644877225952,
      "grad_norm": 0.20936362445354462,
      "learning_rate": 2.2894974968515176e-06,
      "loss": 0.002,
      "step": 222690
    },
    {
      "epoch": 1.4283286246163813,
      "grad_norm": 0.0063627613708376884,
      "learning_rate": 2.289027187443517e-06,
      "loss": 0.0004,
      "step": 222700
    },
    {
      "epoch": 1.4283927615101673,
      "grad_norm": 0.019038209691643715,
      "learning_rate": 2.2885569120059415e-06,
      "loss": 0.0019,
      "step": 222710
    },
    {
      "epoch": 1.4284568984039534,
      "grad_norm": 0.06620853394269943,
      "learning_rate": 2.288086670544687e-06,
      "loss": 0.0013,
      "step": 222720
    },
    {
      "epoch": 1.4285210352977395,
      "grad_norm": 0.04086804389953613,
      "learning_rate": 2.2876164630656455e-06,
      "loss": 0.0014,
      "step": 222730
    },
    {
      "epoch": 1.4285851721915255,
      "grad_norm": 0.03152358904480934,
      "learning_rate": 2.287146289574706e-06,
      "loss": 0.0012,
      "step": 222740
    },
    {
      "epoch": 1.4286493090853116,
      "grad_norm": 0.2787061631679535,
      "learning_rate": 2.286676150077764e-06,
      "loss": 0.0018,
      "step": 222750
    },
    {
      "epoch": 1.4287134459790978,
      "grad_norm": 0.021292492747306824,
      "learning_rate": 2.2862060445807094e-06,
      "loss": 0.0009,
      "step": 222760
    },
    {
      "epoch": 1.428777582872884,
      "grad_norm": 0.06899149715900421,
      "learning_rate": 2.285735973089432e-06,
      "loss": 0.0008,
      "step": 222770
    },
    {
      "epoch": 1.42884171976667,
      "grad_norm": 0.08117298036813736,
      "learning_rate": 2.2852659356098215e-06,
      "loss": 0.0007,
      "step": 222780
    },
    {
      "epoch": 1.428905856660456,
      "grad_norm": 0.024597637355327606,
      "learning_rate": 2.284795932147771e-06,
      "loss": 0.0009,
      "step": 222790
    },
    {
      "epoch": 1.4289699935542421,
      "grad_norm": 0.10690958052873611,
      "learning_rate": 2.2843259627091676e-06,
      "loss": 0.0009,
      "step": 222800
    },
    {
      "epoch": 1.4290341304480283,
      "grad_norm": 0.0217173770070076,
      "learning_rate": 2.2838560272999006e-06,
      "loss": 0.001,
      "step": 222810
    },
    {
      "epoch": 1.4290982673418144,
      "grad_norm": 0.0378669835627079,
      "learning_rate": 2.283386125925857e-06,
      "loss": 0.001,
      "step": 222820
    },
    {
      "epoch": 1.4291624042356004,
      "grad_norm": 0.05231548845767975,
      "learning_rate": 2.2829162585929287e-06,
      "loss": 0.0012,
      "step": 222830
    },
    {
      "epoch": 1.4292265411293865,
      "grad_norm": 0.08824943006038666,
      "learning_rate": 2.2824464253070017e-06,
      "loss": 0.0007,
      "step": 222840
    },
    {
      "epoch": 1.4292906780231727,
      "grad_norm": 0.01857241988182068,
      "learning_rate": 2.281976626073963e-06,
      "loss": 0.0006,
      "step": 222850
    },
    {
      "epoch": 1.4293548149169588,
      "grad_norm": 0.12173985689878464,
      "learning_rate": 2.281506860899698e-06,
      "loss": 0.0019,
      "step": 222860
    },
    {
      "epoch": 1.429418951810745,
      "grad_norm": 0.08885741978883743,
      "learning_rate": 2.281037129790097e-06,
      "loss": 0.005,
      "step": 222870
    },
    {
      "epoch": 1.429483088704531,
      "grad_norm": 0.054606009274721146,
      "learning_rate": 2.2805674327510436e-06,
      "loss": 0.0015,
      "step": 222880
    },
    {
      "epoch": 1.429547225598317,
      "grad_norm": 0.03347182646393776,
      "learning_rate": 2.280097769788422e-06,
      "loss": 0.0012,
      "step": 222890
    },
    {
      "epoch": 1.4296113624921032,
      "grad_norm": 0.1394302248954773,
      "learning_rate": 2.279628140908121e-06,
      "loss": 0.0007,
      "step": 222900
    },
    {
      "epoch": 1.4296754993858891,
      "grad_norm": 0.044962868094444275,
      "learning_rate": 2.2791585461160222e-06,
      "loss": 0.0014,
      "step": 222910
    },
    {
      "epoch": 1.4297396362796753,
      "grad_norm": 0.12735018134117126,
      "learning_rate": 2.278688985418013e-06,
      "loss": 0.0032,
      "step": 222920
    },
    {
      "epoch": 1.4298037731734614,
      "grad_norm": 0.21698321402072906,
      "learning_rate": 2.278219458819975e-06,
      "loss": 0.0018,
      "step": 222930
    },
    {
      "epoch": 1.4298679100672476,
      "grad_norm": 0.01788831502199173,
      "learning_rate": 2.277749966327791e-06,
      "loss": 0.0008,
      "step": 222940
    },
    {
      "epoch": 1.4299320469610337,
      "grad_norm": 0.1518547683954239,
      "learning_rate": 2.2772805079473482e-06,
      "loss": 0.0016,
      "step": 222950
    },
    {
      "epoch": 1.4299961838548199,
      "grad_norm": 0.10488652437925339,
      "learning_rate": 2.276811083684526e-06,
      "loss": 0.0022,
      "step": 222960
    },
    {
      "epoch": 1.4300603207486058,
      "grad_norm": 0.06715139001607895,
      "learning_rate": 2.2763416935452064e-06,
      "loss": 0.0028,
      "step": 222970
    },
    {
      "epoch": 1.430124457642392,
      "grad_norm": 0.10721565783023834,
      "learning_rate": 2.275872337535271e-06,
      "loss": 0.0007,
      "step": 222980
    },
    {
      "epoch": 1.430188594536178,
      "grad_norm": 0.14324164390563965,
      "learning_rate": 2.275403015660604e-06,
      "loss": 0.0019,
      "step": 222990
    },
    {
      "epoch": 1.430252731429964,
      "grad_norm": 0.12048004567623138,
      "learning_rate": 2.274933727927084e-06,
      "loss": 0.0008,
      "step": 223000
    },
    {
      "epoch": 1.4303168683237502,
      "grad_norm": 0.09764410555362701,
      "learning_rate": 2.2744644743405903e-06,
      "loss": 0.0015,
      "step": 223010
    },
    {
      "epoch": 1.4303810052175363,
      "grad_norm": 0.1218528151512146,
      "learning_rate": 2.273995254907006e-06,
      "loss": 0.0008,
      "step": 223020
    },
    {
      "epoch": 1.4304451421113225,
      "grad_norm": 0.08948386460542679,
      "learning_rate": 2.2735260696322097e-06,
      "loss": 0.0015,
      "step": 223030
    },
    {
      "epoch": 1.4305092790051086,
      "grad_norm": 0.056394536048173904,
      "learning_rate": 2.2730569185220807e-06,
      "loss": 0.001,
      "step": 223040
    },
    {
      "epoch": 1.4305734158988945,
      "grad_norm": 0.11490045487880707,
      "learning_rate": 2.2725878015824954e-06,
      "loss": 0.0019,
      "step": 223050
    },
    {
      "epoch": 1.4306375527926807,
      "grad_norm": 0.03355501964688301,
      "learning_rate": 2.272118718819336e-06,
      "loss": 0.0007,
      "step": 223060
    },
    {
      "epoch": 1.4307016896864668,
      "grad_norm": 0.052423711866140366,
      "learning_rate": 2.2716496702384787e-06,
      "loss": 0.0015,
      "step": 223070
    },
    {
      "epoch": 1.4307658265802528,
      "grad_norm": 0.10398556292057037,
      "learning_rate": 2.271180655845801e-06,
      "loss": 0.0012,
      "step": 223080
    },
    {
      "epoch": 1.430829963474039,
      "grad_norm": 0.07617367804050446,
      "learning_rate": 2.2707116756471783e-06,
      "loss": 0.0017,
      "step": 223090
    },
    {
      "epoch": 1.430894100367825,
      "grad_norm": 0.0866626724600792,
      "learning_rate": 2.2702427296484903e-06,
      "loss": 0.0023,
      "step": 223100
    },
    {
      "epoch": 1.4309582372616112,
      "grad_norm": 0.1561611443758011,
      "learning_rate": 2.2697738178556118e-06,
      "loss": 0.0013,
      "step": 223110
    },
    {
      "epoch": 1.4310223741553973,
      "grad_norm": 0.04214513301849365,
      "learning_rate": 2.269304940274419e-06,
      "loss": 0.0012,
      "step": 223120
    },
    {
      "epoch": 1.4310865110491835,
      "grad_norm": 0.03349952772259712,
      "learning_rate": 2.2688360969107847e-06,
      "loss": 0.0012,
      "step": 223130
    },
    {
      "epoch": 1.4311506479429694,
      "grad_norm": 0.14768363535404205,
      "learning_rate": 2.2683672877705876e-06,
      "loss": 0.0009,
      "step": 223140
    },
    {
      "epoch": 1.4312147848367556,
      "grad_norm": 0.06566477566957474,
      "learning_rate": 2.2678985128597016e-06,
      "loss": 0.0013,
      "step": 223150
    },
    {
      "epoch": 1.4312789217305417,
      "grad_norm": 0.09701356291770935,
      "learning_rate": 2.2674297721839973e-06,
      "loss": 0.0012,
      "step": 223160
    },
    {
      "epoch": 1.4313430586243276,
      "grad_norm": 0.10374977439641953,
      "learning_rate": 2.266961065749353e-06,
      "loss": 0.0009,
      "step": 223170
    },
    {
      "epoch": 1.4314071955181138,
      "grad_norm": 0.004872579593211412,
      "learning_rate": 2.2664923935616397e-06,
      "loss": 0.0017,
      "step": 223180
    },
    {
      "epoch": 1.4314713324119,
      "grad_norm": 0.06999684870243073,
      "learning_rate": 2.26602375562673e-06,
      "loss": 0.0009,
      "step": 223190
    },
    {
      "epoch": 1.431535469305686,
      "grad_norm": 0.05309106409549713,
      "learning_rate": 2.2655551519504955e-06,
      "loss": 0.0012,
      "step": 223200
    },
    {
      "epoch": 1.4315996061994722,
      "grad_norm": 0.1891021579504013,
      "learning_rate": 2.2650865825388103e-06,
      "loss": 0.0008,
      "step": 223210
    },
    {
      "epoch": 1.4316637430932582,
      "grad_norm": 0.06925483047962189,
      "learning_rate": 2.2646180473975454e-06,
      "loss": 0.0008,
      "step": 223220
    },
    {
      "epoch": 1.4317278799870443,
      "grad_norm": 0.16873957216739655,
      "learning_rate": 2.2641495465325707e-06,
      "loss": 0.0006,
      "step": 223230
    },
    {
      "epoch": 1.4317920168808305,
      "grad_norm": 0.12946078181266785,
      "learning_rate": 2.263681079949756e-06,
      "loss": 0.0011,
      "step": 223240
    },
    {
      "epoch": 1.4318561537746166,
      "grad_norm": 0.04395328834652901,
      "learning_rate": 2.263212647654975e-06,
      "loss": 0.0017,
      "step": 223250
    },
    {
      "epoch": 1.4319202906684025,
      "grad_norm": 0.11219099164009094,
      "learning_rate": 2.262744249654095e-06,
      "loss": 0.0015,
      "step": 223260
    },
    {
      "epoch": 1.4319844275621887,
      "grad_norm": 0.04642408341169357,
      "learning_rate": 2.2622758859529866e-06,
      "loss": 0.0011,
      "step": 223270
    },
    {
      "epoch": 1.4320485644559748,
      "grad_norm": 0.02733568102121353,
      "learning_rate": 2.261807556557516e-06,
      "loss": 0.0015,
      "step": 223280
    },
    {
      "epoch": 1.432112701349761,
      "grad_norm": 0.08607029914855957,
      "learning_rate": 2.261339261473555e-06,
      "loss": 0.001,
      "step": 223290
    },
    {
      "epoch": 1.4321768382435471,
      "grad_norm": 0.14403820037841797,
      "learning_rate": 2.2608710007069706e-06,
      "loss": 0.0015,
      "step": 223300
    },
    {
      "epoch": 1.432240975137333,
      "grad_norm": 0.09584439545869827,
      "learning_rate": 2.260402774263628e-06,
      "loss": 0.0013,
      "step": 223310
    },
    {
      "epoch": 1.4323051120311192,
      "grad_norm": 0.012513621710240841,
      "learning_rate": 2.259934582149399e-06,
      "loss": 0.0009,
      "step": 223320
    },
    {
      "epoch": 1.4323692489249054,
      "grad_norm": 0.09077856689691544,
      "learning_rate": 2.2594664243701474e-06,
      "loss": 0.0023,
      "step": 223330
    },
    {
      "epoch": 1.4324333858186913,
      "grad_norm": 0.09028846770524979,
      "learning_rate": 2.2589983009317405e-06,
      "loss": 0.0011,
      "step": 223340
    },
    {
      "epoch": 1.4324975227124774,
      "grad_norm": 0.05965743586421013,
      "learning_rate": 2.258530211840042e-06,
      "loss": 0.001,
      "step": 223350
    },
    {
      "epoch": 1.4325616596062636,
      "grad_norm": 0.007242657709866762,
      "learning_rate": 2.258062157100921e-06,
      "loss": 0.0015,
      "step": 223360
    },
    {
      "epoch": 1.4326257965000497,
      "grad_norm": 0.2847703695297241,
      "learning_rate": 2.2575941367202407e-06,
      "loss": 0.0015,
      "step": 223370
    },
    {
      "epoch": 1.4326899333938359,
      "grad_norm": 0.02166147716343403,
      "learning_rate": 2.257126150703866e-06,
      "loss": 0.0008,
      "step": 223380
    },
    {
      "epoch": 1.432754070287622,
      "grad_norm": 0.08338756859302521,
      "learning_rate": 2.2566581990576592e-06,
      "loss": 0.0009,
      "step": 223390
    },
    {
      "epoch": 1.432818207181408,
      "grad_norm": 0.2065654993057251,
      "learning_rate": 2.2561902817874877e-06,
      "loss": 0.0016,
      "step": 223400
    },
    {
      "epoch": 1.432882344075194,
      "grad_norm": 0.09669201076030731,
      "learning_rate": 2.2557223988992127e-06,
      "loss": 0.0021,
      "step": 223410
    },
    {
      "epoch": 1.4329464809689803,
      "grad_norm": 0.04104436933994293,
      "learning_rate": 2.255254550398697e-06,
      "loss": 0.0009,
      "step": 223420
    },
    {
      "epoch": 1.4330106178627662,
      "grad_norm": 0.03617101535201073,
      "learning_rate": 2.2547867362918026e-06,
      "loss": 0.0008,
      "step": 223430
    },
    {
      "epoch": 1.4330747547565523,
      "grad_norm": 0.07173281908035278,
      "learning_rate": 2.2543189565843938e-06,
      "loss": 0.0024,
      "step": 223440
    },
    {
      "epoch": 1.4331388916503385,
      "grad_norm": 0.06350976228713989,
      "learning_rate": 2.2538512112823303e-06,
      "loss": 0.0008,
      "step": 223450
    },
    {
      "epoch": 1.4332030285441246,
      "grad_norm": 0.004857209045439959,
      "learning_rate": 2.253383500391472e-06,
      "loss": 0.0007,
      "step": 223460
    },
    {
      "epoch": 1.4332671654379108,
      "grad_norm": 0.0834898054599762,
      "learning_rate": 2.2529158239176835e-06,
      "loss": 0.0011,
      "step": 223470
    },
    {
      "epoch": 1.4333313023316967,
      "grad_norm": 0.09479456394910812,
      "learning_rate": 2.252448181866823e-06,
      "loss": 0.0017,
      "step": 223480
    },
    {
      "epoch": 1.4333954392254828,
      "grad_norm": 0.2217305600643158,
      "learning_rate": 2.2519805742447504e-06,
      "loss": 0.0022,
      "step": 223490
    },
    {
      "epoch": 1.433459576119269,
      "grad_norm": 0.10038892924785614,
      "learning_rate": 2.2515130010573234e-06,
      "loss": 0.0014,
      "step": 223500
    },
    {
      "epoch": 1.4335237130130551,
      "grad_norm": 0.07915161550045013,
      "learning_rate": 2.251045462310405e-06,
      "loss": 0.0015,
      "step": 223510
    },
    {
      "epoch": 1.433587849906841,
      "grad_norm": 0.0070143286138772964,
      "learning_rate": 2.250577958009851e-06,
      "loss": 0.0015,
      "step": 223520
    },
    {
      "epoch": 1.4336519868006272,
      "grad_norm": 0.11248385161161423,
      "learning_rate": 2.2501104881615205e-06,
      "loss": 0.0016,
      "step": 223530
    },
    {
      "epoch": 1.4337161236944134,
      "grad_norm": 0.003413565456867218,
      "learning_rate": 2.2496430527712687e-06,
      "loss": 0.0014,
      "step": 223540
    },
    {
      "epoch": 1.4337802605881995,
      "grad_norm": 0.12429042160511017,
      "learning_rate": 2.2491756518449576e-06,
      "loss": 0.0009,
      "step": 223550
    },
    {
      "epoch": 1.4338443974819857,
      "grad_norm": 0.16163988411426544,
      "learning_rate": 2.2487082853884413e-06,
      "loss": 0.0016,
      "step": 223560
    },
    {
      "epoch": 1.4339085343757716,
      "grad_norm": 0.15755464136600494,
      "learning_rate": 2.248240953407576e-06,
      "loss": 0.0011,
      "step": 223570
    },
    {
      "epoch": 1.4339726712695577,
      "grad_norm": 0.04327183961868286,
      "learning_rate": 2.247773655908217e-06,
      "loss": 0.001,
      "step": 223580
    },
    {
      "epoch": 1.434036808163344,
      "grad_norm": 0.024731261655688286,
      "learning_rate": 2.2473063928962222e-06,
      "loss": 0.0021,
      "step": 223590
    },
    {
      "epoch": 1.4341009450571298,
      "grad_norm": 0.25946247577667236,
      "learning_rate": 2.2468391643774464e-06,
      "loss": 0.0017,
      "step": 223600
    },
    {
      "epoch": 1.434165081950916,
      "grad_norm": 0.039348818361759186,
      "learning_rate": 2.2463719703577414e-06,
      "loss": 0.001,
      "step": 223610
    },
    {
      "epoch": 1.4342292188447021,
      "grad_norm": 0.08746260404586792,
      "learning_rate": 2.2459048108429653e-06,
      "loss": 0.0011,
      "step": 223620
    },
    {
      "epoch": 1.4342933557384883,
      "grad_norm": 0.04812712222337723,
      "learning_rate": 2.2454376858389704e-06,
      "loss": 0.0015,
      "step": 223630
    },
    {
      "epoch": 1.4343574926322744,
      "grad_norm": 0.025965077802538872,
      "learning_rate": 2.24497059535161e-06,
      "loss": 0.0006,
      "step": 223640
    },
    {
      "epoch": 1.4344216295260606,
      "grad_norm": 0.13733486831188202,
      "learning_rate": 2.244503539386735e-06,
      "loss": 0.0013,
      "step": 223650
    },
    {
      "epoch": 1.4344857664198465,
      "grad_norm": 0.030825136229395866,
      "learning_rate": 2.244036517950202e-06,
      "loss": 0.0013,
      "step": 223660
    },
    {
      "epoch": 1.4345499033136326,
      "grad_norm": 0.019314579665660858,
      "learning_rate": 2.2435695310478587e-06,
      "loss": 0.0012,
      "step": 223670
    },
    {
      "epoch": 1.4346140402074188,
      "grad_norm": 0.07342612743377686,
      "learning_rate": 2.2431025786855614e-06,
      "loss": 0.0023,
      "step": 223680
    },
    {
      "epoch": 1.4346781771012047,
      "grad_norm": 0.0634700134396553,
      "learning_rate": 2.2426356608691597e-06,
      "loss": 0.0024,
      "step": 223690
    },
    {
      "epoch": 1.4347423139949909,
      "grad_norm": 0.011114954017102718,
      "learning_rate": 2.2421687776045016e-06,
      "loss": 0.0012,
      "step": 223700
    },
    {
      "epoch": 1.434806450888777,
      "grad_norm": 0.0006286423886194825,
      "learning_rate": 2.241701928897441e-06,
      "loss": 0.0014,
      "step": 223710
    },
    {
      "epoch": 1.4348705877825632,
      "grad_norm": 0.06351950019598007,
      "learning_rate": 2.241235114753827e-06,
      "loss": 0.0006,
      "step": 223720
    },
    {
      "epoch": 1.4349347246763493,
      "grad_norm": 0.05163135752081871,
      "learning_rate": 2.240768335179507e-06,
      "loss": 0.0008,
      "step": 223730
    },
    {
      "epoch": 1.4349988615701352,
      "grad_norm": 0.014766098000109196,
      "learning_rate": 2.2403015901803333e-06,
      "loss": 0.0006,
      "step": 223740
    },
    {
      "epoch": 1.4350629984639214,
      "grad_norm": 0.034667257219552994,
      "learning_rate": 2.2398348797621528e-06,
      "loss": 0.0011,
      "step": 223750
    },
    {
      "epoch": 1.4351271353577075,
      "grad_norm": 0.025669017806649208,
      "learning_rate": 2.2393682039308147e-06,
      "loss": 0.0021,
      "step": 223760
    },
    {
      "epoch": 1.4351912722514935,
      "grad_norm": 0.04082329198718071,
      "learning_rate": 2.2389015626921634e-06,
      "loss": 0.0004,
      "step": 223770
    },
    {
      "epoch": 1.4352554091452796,
      "grad_norm": 0.05806554853916168,
      "learning_rate": 2.238434956052051e-06,
      "loss": 0.0041,
      "step": 223780
    },
    {
      "epoch": 1.4353195460390658,
      "grad_norm": 0.10913745313882828,
      "learning_rate": 2.237968384016322e-06,
      "loss": 0.001,
      "step": 223790
    },
    {
      "epoch": 1.435383682932852,
      "grad_norm": 0.04209039360284805,
      "learning_rate": 2.237501846590823e-06,
      "loss": 0.0009,
      "step": 223800
    },
    {
      "epoch": 1.435447819826638,
      "grad_norm": 0.1168067678809166,
      "learning_rate": 2.2370353437813992e-06,
      "loss": 0.001,
      "step": 223810
    },
    {
      "epoch": 1.4355119567204242,
      "grad_norm": 0.05427335202693939,
      "learning_rate": 2.2365688755938986e-06,
      "loss": 0.0013,
      "step": 223820
    },
    {
      "epoch": 1.4355760936142101,
      "grad_norm": 0.08518806844949722,
      "learning_rate": 2.236102442034165e-06,
      "loss": 0.0023,
      "step": 223830
    },
    {
      "epoch": 1.4356402305079963,
      "grad_norm": 0.3001389801502228,
      "learning_rate": 2.2356360431080427e-06,
      "loss": 0.0023,
      "step": 223840
    },
    {
      "epoch": 1.4357043674017824,
      "grad_norm": 0.10559581220149994,
      "learning_rate": 2.235169678821375e-06,
      "loss": 0.0016,
      "step": 223850
    },
    {
      "epoch": 1.4357685042955683,
      "grad_norm": 0.07048898935317993,
      "learning_rate": 2.234703349180009e-06,
      "loss": 0.001,
      "step": 223860
    },
    {
      "epoch": 1.4358326411893545,
      "grad_norm": 0.11938511580228806,
      "learning_rate": 2.2342370541897867e-06,
      "loss": 0.0007,
      "step": 223870
    },
    {
      "epoch": 1.4358967780831406,
      "grad_norm": 0.09405989199876785,
      "learning_rate": 2.2337707938565485e-06,
      "loss": 0.0008,
      "step": 223880
    },
    {
      "epoch": 1.4359609149769268,
      "grad_norm": 0.11577337235212326,
      "learning_rate": 2.233304568186141e-06,
      "loss": 0.0043,
      "step": 223890
    },
    {
      "epoch": 1.436025051870713,
      "grad_norm": 0.6096563339233398,
      "learning_rate": 2.232838377184405e-06,
      "loss": 0.0039,
      "step": 223900
    },
    {
      "epoch": 1.4360891887644989,
      "grad_norm": 0.0584028996527195,
      "learning_rate": 2.2323722208571813e-06,
      "loss": 0.0016,
      "step": 223910
    },
    {
      "epoch": 1.436153325658285,
      "grad_norm": 0.058712344616651535,
      "learning_rate": 2.23190609921031e-06,
      "loss": 0.0012,
      "step": 223920
    },
    {
      "epoch": 1.4362174625520712,
      "grad_norm": 0.022509023547172546,
      "learning_rate": 2.231440012249636e-06,
      "loss": 0.0013,
      "step": 223930
    },
    {
      "epoch": 1.4362815994458573,
      "grad_norm": 0.1276644617319107,
      "learning_rate": 2.2309739599809967e-06,
      "loss": 0.0015,
      "step": 223940
    },
    {
      "epoch": 1.4363457363396432,
      "grad_norm": 0.08014482259750366,
      "learning_rate": 2.2305079424102328e-06,
      "loss": 0.0012,
      "step": 223950
    },
    {
      "epoch": 1.4364098732334294,
      "grad_norm": 0.2725202441215515,
      "learning_rate": 2.230041959543182e-06,
      "loss": 0.0027,
      "step": 223960
    },
    {
      "epoch": 1.4364740101272155,
      "grad_norm": 0.24841365218162537,
      "learning_rate": 2.229576011385687e-06,
      "loss": 0.0027,
      "step": 223970
    },
    {
      "epoch": 1.4365381470210017,
      "grad_norm": 0.0672532245516777,
      "learning_rate": 2.229110097943584e-06,
      "loss": 0.001,
      "step": 223980
    },
    {
      "epoch": 1.4366022839147878,
      "grad_norm": 0.12797874212265015,
      "learning_rate": 2.2286442192227126e-06,
      "loss": 0.0017,
      "step": 223990
    },
    {
      "epoch": 1.4366664208085738,
      "grad_norm": 0.10601053386926651,
      "learning_rate": 2.228178375228907e-06,
      "loss": 0.0011,
      "step": 224000
    },
    {
      "epoch": 1.43673055770236,
      "grad_norm": 0.028491705656051636,
      "learning_rate": 2.22771256596801e-06,
      "loss": 0.0022,
      "step": 224010
    },
    {
      "epoch": 1.436794694596146,
      "grad_norm": 0.01704668626189232,
      "learning_rate": 2.2272467914458557e-06,
      "loss": 0.0008,
      "step": 224020
    },
    {
      "epoch": 1.436858831489932,
      "grad_norm": 0.053478218615055084,
      "learning_rate": 2.2267810516682793e-06,
      "loss": 0.0008,
      "step": 224030
    },
    {
      "epoch": 1.4369229683837181,
      "grad_norm": 0.13523180782794952,
      "learning_rate": 2.226315346641119e-06,
      "loss": 0.0015,
      "step": 224040
    },
    {
      "epoch": 1.4369871052775043,
      "grad_norm": 0.035574037581682205,
      "learning_rate": 2.2258496763702107e-06,
      "loss": 0.0021,
      "step": 224050
    },
    {
      "epoch": 1.4370512421712904,
      "grad_norm": 0.01855083554983139,
      "learning_rate": 2.2253840408613885e-06,
      "loss": 0.001,
      "step": 224060
    },
    {
      "epoch": 1.4371153790650766,
      "grad_norm": 0.004640830680727959,
      "learning_rate": 2.2249184401204854e-06,
      "loss": 0.0029,
      "step": 224070
    },
    {
      "epoch": 1.4371795159588627,
      "grad_norm": 0.02465139515697956,
      "learning_rate": 2.2244528741533388e-06,
      "loss": 0.0025,
      "step": 224080
    },
    {
      "epoch": 1.4372436528526487,
      "grad_norm": 0.2250843048095703,
      "learning_rate": 2.223987342965782e-06,
      "loss": 0.0023,
      "step": 224090
    },
    {
      "epoch": 1.4373077897464348,
      "grad_norm": 0.21969127655029297,
      "learning_rate": 2.2235218465636477e-06,
      "loss": 0.0025,
      "step": 224100
    },
    {
      "epoch": 1.437371926640221,
      "grad_norm": 0.13385099172592163,
      "learning_rate": 2.223056384952767e-06,
      "loss": 0.0012,
      "step": 224110
    },
    {
      "epoch": 1.4374360635340069,
      "grad_norm": 0.08626440167427063,
      "learning_rate": 2.222590958138976e-06,
      "loss": 0.0012,
      "step": 224120
    },
    {
      "epoch": 1.437500200427793,
      "grad_norm": 0.029763473197817802,
      "learning_rate": 2.222125566128106e-06,
      "loss": 0.0016,
      "step": 224130
    },
    {
      "epoch": 1.4375643373215792,
      "grad_norm": 0.037327855825424194,
      "learning_rate": 2.2216602089259852e-06,
      "loss": 0.0012,
      "step": 224140
    },
    {
      "epoch": 1.4376284742153653,
      "grad_norm": 0.09519702196121216,
      "learning_rate": 2.2211948865384503e-06,
      "loss": 0.0007,
      "step": 224150
    },
    {
      "epoch": 1.4376926111091515,
      "grad_norm": 0.08022797852754593,
      "learning_rate": 2.220729598971329e-06,
      "loss": 0.0015,
      "step": 224160
    },
    {
      "epoch": 1.4377567480029374,
      "grad_norm": 0.20702189207077026,
      "learning_rate": 2.2202643462304525e-06,
      "loss": 0.0022,
      "step": 224170
    },
    {
      "epoch": 1.4378208848967236,
      "grad_norm": 0.11773733794689178,
      "learning_rate": 2.2197991283216486e-06,
      "loss": 0.0017,
      "step": 224180
    },
    {
      "epoch": 1.4378850217905097,
      "grad_norm": 0.019129108637571335,
      "learning_rate": 2.21933394525075e-06,
      "loss": 0.0008,
      "step": 224190
    },
    {
      "epoch": 1.4379491586842956,
      "grad_norm": 0.08117559552192688,
      "learning_rate": 2.2188687970235846e-06,
      "loss": 0.0024,
      "step": 224200
    },
    {
      "epoch": 1.4380132955780818,
      "grad_norm": 0.062179937958717346,
      "learning_rate": 2.2184036836459805e-06,
      "loss": 0.001,
      "step": 224210
    },
    {
      "epoch": 1.438077432471868,
      "grad_norm": 0.2298547625541687,
      "learning_rate": 2.2179386051237652e-06,
      "loss": 0.0015,
      "step": 224220
    },
    {
      "epoch": 1.438141569365654,
      "grad_norm": 0.11879534274339676,
      "learning_rate": 2.2174735614627685e-06,
      "loss": 0.0011,
      "step": 224230
    },
    {
      "epoch": 1.4382057062594402,
      "grad_norm": 0.2470162957906723,
      "learning_rate": 2.2170085526688166e-06,
      "loss": 0.0014,
      "step": 224240
    },
    {
      "epoch": 1.4382698431532264,
      "grad_norm": 0.012568817473948002,
      "learning_rate": 2.2165435787477376e-06,
      "loss": 0.0012,
      "step": 224250
    },
    {
      "epoch": 1.4383339800470123,
      "grad_norm": 0.06997620314359665,
      "learning_rate": 2.216078639705354e-06,
      "loss": 0.0011,
      "step": 224260
    },
    {
      "epoch": 1.4383981169407984,
      "grad_norm": 0.03136972337961197,
      "learning_rate": 2.2156137355474967e-06,
      "loss": 0.004,
      "step": 224270
    },
    {
      "epoch": 1.4384622538345846,
      "grad_norm": 0.047197647392749786,
      "learning_rate": 2.21514886627999e-06,
      "loss": 0.001,
      "step": 224280
    },
    {
      "epoch": 1.4385263907283705,
      "grad_norm": 0.04600229859352112,
      "learning_rate": 2.2146840319086554e-06,
      "loss": 0.0016,
      "step": 224290
    },
    {
      "epoch": 1.4385905276221567,
      "grad_norm": 0.03101489320397377,
      "learning_rate": 2.214219232439323e-06,
      "loss": 0.001,
      "step": 224300
    },
    {
      "epoch": 1.4386546645159428,
      "grad_norm": 0.2436067909002304,
      "learning_rate": 2.2137544678778145e-06,
      "loss": 0.0021,
      "step": 224310
    },
    {
      "epoch": 1.438718801409729,
      "grad_norm": 0.07516494393348694,
      "learning_rate": 2.2132897382299532e-06,
      "loss": 0.0021,
      "step": 224320
    },
    {
      "epoch": 1.4387829383035151,
      "grad_norm": 0.12822185456752777,
      "learning_rate": 2.2128250435015618e-06,
      "loss": 0.0012,
      "step": 224330
    },
    {
      "epoch": 1.438847075197301,
      "grad_norm": 0.0714908316731453,
      "learning_rate": 2.212360383698467e-06,
      "loss": 0.0009,
      "step": 224340
    },
    {
      "epoch": 1.4389112120910872,
      "grad_norm": 0.10385997593402863,
      "learning_rate": 2.211895758826488e-06,
      "loss": 0.0008,
      "step": 224350
    },
    {
      "epoch": 1.4389753489848733,
      "grad_norm": 0.06676853448152542,
      "learning_rate": 2.211431168891448e-06,
      "loss": 0.0009,
      "step": 224360
    },
    {
      "epoch": 1.4390394858786595,
      "grad_norm": 0.10909586399793625,
      "learning_rate": 2.210966613899167e-06,
      "loss": 0.0013,
      "step": 224370
    },
    {
      "epoch": 1.4391036227724454,
      "grad_norm": 0.09288477152585983,
      "learning_rate": 2.2105020938554687e-06,
      "loss": 0.0009,
      "step": 224380
    },
    {
      "epoch": 1.4391677596662316,
      "grad_norm": 0.07626969367265701,
      "learning_rate": 2.2100376087661736e-06,
      "loss": 0.0016,
      "step": 224390
    },
    {
      "epoch": 1.4392318965600177,
      "grad_norm": 0.07371053844690323,
      "learning_rate": 2.2095731586371007e-06,
      "loss": 0.0006,
      "step": 224400
    },
    {
      "epoch": 1.4392960334538039,
      "grad_norm": 0.03710919991135597,
      "learning_rate": 2.209108743474069e-06,
      "loss": 0.0028,
      "step": 224410
    },
    {
      "epoch": 1.43936017034759,
      "grad_norm": 0.10803516954183578,
      "learning_rate": 2.208644363282901e-06,
      "loss": 0.0011,
      "step": 224420
    },
    {
      "epoch": 1.439424307241376,
      "grad_norm": 0.12155324965715408,
      "learning_rate": 2.2081800180694135e-06,
      "loss": 0.0009,
      "step": 224430
    },
    {
      "epoch": 1.439488444135162,
      "grad_norm": 0.0163084976375103,
      "learning_rate": 2.2077157078394244e-06,
      "loss": 0.0017,
      "step": 224440
    },
    {
      "epoch": 1.4395525810289482,
      "grad_norm": 0.07152056694030762,
      "learning_rate": 2.2072514325987532e-06,
      "loss": 0.0009,
      "step": 224450
    },
    {
      "epoch": 1.4396167179227342,
      "grad_norm": 0.03252122923731804,
      "learning_rate": 2.2067871923532186e-06,
      "loss": 0.0008,
      "step": 224460
    },
    {
      "epoch": 1.4396808548165203,
      "grad_norm": 0.07736662775278091,
      "learning_rate": 2.206322987108638e-06,
      "loss": 0.0017,
      "step": 224470
    },
    {
      "epoch": 1.4397449917103065,
      "grad_norm": 0.0351140983402729,
      "learning_rate": 2.2058588168708257e-06,
      "loss": 0.0023,
      "step": 224480
    },
    {
      "epoch": 1.4398091286040926,
      "grad_norm": 0.11911506950855255,
      "learning_rate": 2.2053946816455985e-06,
      "loss": 0.0012,
      "step": 224490
    },
    {
      "epoch": 1.4398732654978788,
      "grad_norm": 0.11416828632354736,
      "learning_rate": 2.2049305814387746e-06,
      "loss": 0.0016,
      "step": 224500
    },
    {
      "epoch": 1.439937402391665,
      "grad_norm": 0.08931335061788559,
      "learning_rate": 2.204466516256168e-06,
      "loss": 0.001,
      "step": 224510
    },
    {
      "epoch": 1.4400015392854508,
      "grad_norm": 0.1183854192495346,
      "learning_rate": 2.204002486103594e-06,
      "loss": 0.0015,
      "step": 224520
    },
    {
      "epoch": 1.440065676179237,
      "grad_norm": 0.08022509515285492,
      "learning_rate": 2.203538490986865e-06,
      "loss": 0.0014,
      "step": 224530
    },
    {
      "epoch": 1.4401298130730231,
      "grad_norm": 0.05324234440922737,
      "learning_rate": 2.2030745309117995e-06,
      "loss": 0.0015,
      "step": 224540
    },
    {
      "epoch": 1.440193949966809,
      "grad_norm": 0.09862230718135834,
      "learning_rate": 2.2026106058842094e-06,
      "loss": 0.0025,
      "step": 224550
    },
    {
      "epoch": 1.4402580868605952,
      "grad_norm": 0.024814411997795105,
      "learning_rate": 2.2021467159099055e-06,
      "loss": 0.0012,
      "step": 224560
    },
    {
      "epoch": 1.4403222237543813,
      "grad_norm": 0.07038882374763489,
      "learning_rate": 2.2016828609947044e-06,
      "loss": 0.0017,
      "step": 224570
    },
    {
      "epoch": 1.4403863606481675,
      "grad_norm": 0.06504993885755539,
      "learning_rate": 2.201219041144417e-06,
      "loss": 0.0011,
      "step": 224580
    },
    {
      "epoch": 1.4404504975419536,
      "grad_norm": 0.029528573155403137,
      "learning_rate": 2.200755256364856e-06,
      "loss": 0.0018,
      "step": 224590
    },
    {
      "epoch": 1.4405146344357396,
      "grad_norm": 0.03728478401899338,
      "learning_rate": 2.20029150666183e-06,
      "loss": 0.001,
      "step": 224600
    },
    {
      "epoch": 1.4405787713295257,
      "grad_norm": 0.0857059583067894,
      "learning_rate": 2.1998277920411536e-06,
      "loss": 0.0011,
      "step": 224610
    },
    {
      "epoch": 1.4406429082233119,
      "grad_norm": 0.058394331485033035,
      "learning_rate": 2.199364112508637e-06,
      "loss": 0.0011,
      "step": 224620
    },
    {
      "epoch": 1.4407070451170978,
      "grad_norm": 0.0408020094037056,
      "learning_rate": 2.1989004680700893e-06,
      "loss": 0.0014,
      "step": 224630
    },
    {
      "epoch": 1.440771182010884,
      "grad_norm": 0.06660662591457367,
      "learning_rate": 2.1984368587313186e-06,
      "loss": 0.0015,
      "step": 224640
    },
    {
      "epoch": 1.44083531890467,
      "grad_norm": 0.02765144221484661,
      "learning_rate": 2.197973284498139e-06,
      "loss": 0.0009,
      "step": 224650
    },
    {
      "epoch": 1.4408994557984562,
      "grad_norm": 0.13821285963058472,
      "learning_rate": 2.197509745376356e-06,
      "loss": 0.0025,
      "step": 224660
    },
    {
      "epoch": 1.4409635926922424,
      "grad_norm": 0.07172411680221558,
      "learning_rate": 2.197046241371779e-06,
      "loss": 0.0015,
      "step": 224670
    },
    {
      "epoch": 1.4410277295860285,
      "grad_norm": 0.06663408875465393,
      "learning_rate": 2.196582772490214e-06,
      "loss": 0.0018,
      "step": 224680
    },
    {
      "epoch": 1.4410918664798145,
      "grad_norm": 0.021494129672646523,
      "learning_rate": 2.196119338737472e-06,
      "loss": 0.0014,
      "step": 224690
    },
    {
      "epoch": 1.4411560033736006,
      "grad_norm": 0.043371740728616714,
      "learning_rate": 2.195655940119359e-06,
      "loss": 0.0019,
      "step": 224700
    },
    {
      "epoch": 1.4412201402673868,
      "grad_norm": 0.03123489022254944,
      "learning_rate": 2.1951925766416795e-06,
      "loss": 0.0007,
      "step": 224710
    },
    {
      "epoch": 1.4412842771611727,
      "grad_norm": 0.11745089292526245,
      "learning_rate": 2.194729248310243e-06,
      "loss": 0.0015,
      "step": 224720
    },
    {
      "epoch": 1.4413484140549588,
      "grad_norm": 0.03425348922610283,
      "learning_rate": 2.1942659551308536e-06,
      "loss": 0.0017,
      "step": 224730
    },
    {
      "epoch": 1.441412550948745,
      "grad_norm": 0.15983393788337708,
      "learning_rate": 2.193802697109318e-06,
      "loss": 0.0014,
      "step": 224740
    },
    {
      "epoch": 1.4414766878425311,
      "grad_norm": 0.09651978313922882,
      "learning_rate": 2.1933394742514376e-06,
      "loss": 0.002,
      "step": 224750
    },
    {
      "epoch": 1.4415408247363173,
      "grad_norm": 0.034692175686359406,
      "learning_rate": 2.1928762865630215e-06,
      "loss": 0.0017,
      "step": 224760
    },
    {
      "epoch": 1.4416049616301032,
      "grad_norm": 0.0047668395563960075,
      "learning_rate": 2.1924131340498716e-06,
      "loss": 0.0014,
      "step": 224770
    },
    {
      "epoch": 1.4416690985238894,
      "grad_norm": 0.08082887530326843,
      "learning_rate": 2.191950016717792e-06,
      "loss": 0.0011,
      "step": 224780
    },
    {
      "epoch": 1.4417332354176755,
      "grad_norm": 0.018363196402788162,
      "learning_rate": 2.1914869345725836e-06,
      "loss": 0.0028,
      "step": 224790
    },
    {
      "epoch": 1.4417973723114617,
      "grad_norm": 0.0388919971883297,
      "learning_rate": 2.1910238876200533e-06,
      "loss": 0.0006,
      "step": 224800
    },
    {
      "epoch": 1.4418615092052476,
      "grad_norm": 0.05781109258532524,
      "learning_rate": 2.190560875866001e-06,
      "loss": 0.001,
      "step": 224810
    },
    {
      "epoch": 1.4419256460990337,
      "grad_norm": 0.04180987924337387,
      "learning_rate": 2.1900978993162284e-06,
      "loss": 0.0015,
      "step": 224820
    },
    {
      "epoch": 1.4419897829928199,
      "grad_norm": 0.05864508077502251,
      "learning_rate": 2.1896349579765356e-06,
      "loss": 0.001,
      "step": 224830
    },
    {
      "epoch": 1.442053919886606,
      "grad_norm": 0.045534711331129074,
      "learning_rate": 2.1891720518527276e-06,
      "loss": 0.0015,
      "step": 224840
    },
    {
      "epoch": 1.4421180567803922,
      "grad_norm": 0.24148043990135193,
      "learning_rate": 2.188709180950603e-06,
      "loss": 0.0025,
      "step": 224850
    },
    {
      "epoch": 1.442182193674178,
      "grad_norm": 0.05315045267343521,
      "learning_rate": 2.188246345275959e-06,
      "loss": 0.0023,
      "step": 224860
    },
    {
      "epoch": 1.4422463305679643,
      "grad_norm": 0.039694495499134064,
      "learning_rate": 2.1877835448346e-06,
      "loss": 0.0027,
      "step": 224870
    },
    {
      "epoch": 1.4423104674617504,
      "grad_norm": 0.12355036288499832,
      "learning_rate": 2.1873207796323227e-06,
      "loss": 0.0011,
      "step": 224880
    },
    {
      "epoch": 1.4423746043555363,
      "grad_norm": 0.003701229579746723,
      "learning_rate": 2.186858049674926e-06,
      "loss": 0.0012,
      "step": 224890
    },
    {
      "epoch": 1.4424387412493225,
      "grad_norm": 0.04159437492489815,
      "learning_rate": 2.186395354968207e-06,
      "loss": 0.001,
      "step": 224900
    },
    {
      "epoch": 1.4425028781431086,
      "grad_norm": 0.041146691888570786,
      "learning_rate": 2.185932695517967e-06,
      "loss": 0.0014,
      "step": 224910
    },
    {
      "epoch": 1.4425670150368948,
      "grad_norm": 0.08462332934141159,
      "learning_rate": 2.1854700713300015e-06,
      "loss": 0.0032,
      "step": 224920
    },
    {
      "epoch": 1.442631151930681,
      "grad_norm": 0.10139299184083939,
      "learning_rate": 2.1850074824101068e-06,
      "loss": 0.0014,
      "step": 224930
    },
    {
      "epoch": 1.442695288824467,
      "grad_norm": 0.25560396909713745,
      "learning_rate": 2.184544928764079e-06,
      "loss": 0.002,
      "step": 224940
    },
    {
      "epoch": 1.442759425718253,
      "grad_norm": 0.10392890125513077,
      "learning_rate": 2.1840824103977175e-06,
      "loss": 0.0014,
      "step": 224950
    },
    {
      "epoch": 1.4428235626120391,
      "grad_norm": 0.04784127697348595,
      "learning_rate": 2.183619927316816e-06,
      "loss": 0.0017,
      "step": 224960
    },
    {
      "epoch": 1.4428876995058253,
      "grad_norm": 0.11029145866632462,
      "learning_rate": 2.183157479527169e-06,
      "loss": 0.0015,
      "step": 224970
    },
    {
      "epoch": 1.4429518363996112,
      "grad_norm": 0.05119152367115021,
      "learning_rate": 2.1826950670345702e-06,
      "loss": 0.0017,
      "step": 224980
    },
    {
      "epoch": 1.4430159732933974,
      "grad_norm": 0.1125304326415062,
      "learning_rate": 2.1822326898448175e-06,
      "loss": 0.0013,
      "step": 224990
    },
    {
      "epoch": 1.4430801101871835,
      "grad_norm": 0.0023442390374839306,
      "learning_rate": 2.181770347963704e-06,
      "loss": 0.0007,
      "step": 225000
    },
    {
      "epoch": 1.4431442470809697,
      "grad_norm": 0.04524316266179085,
      "learning_rate": 2.1813080413970196e-06,
      "loss": 0.0007,
      "step": 225010
    },
    {
      "epoch": 1.4432083839747558,
      "grad_norm": 0.07342428714036942,
      "learning_rate": 2.1808457701505615e-06,
      "loss": 0.0007,
      "step": 225020
    },
    {
      "epoch": 1.4432725208685417,
      "grad_norm": 0.02977115474641323,
      "learning_rate": 2.180383534230121e-06,
      "loss": 0.0012,
      "step": 225030
    },
    {
      "epoch": 1.443336657762328,
      "grad_norm": 0.29319748282432556,
      "learning_rate": 2.1799213336414897e-06,
      "loss": 0.0014,
      "step": 225040
    },
    {
      "epoch": 1.443400794656114,
      "grad_norm": 0.10643231123685837,
      "learning_rate": 2.1794591683904582e-06,
      "loss": 0.0021,
      "step": 225050
    },
    {
      "epoch": 1.4434649315499002,
      "grad_norm": 0.011669746600091457,
      "learning_rate": 2.178997038482821e-06,
      "loss": 0.0015,
      "step": 225060
    },
    {
      "epoch": 1.4435290684436861,
      "grad_norm": 0.019415656104683876,
      "learning_rate": 2.1785349439243664e-06,
      "loss": 0.0008,
      "step": 225070
    },
    {
      "epoch": 1.4435932053374723,
      "grad_norm": 0.09339431673288345,
      "learning_rate": 2.178072884720886e-06,
      "loss": 0.0009,
      "step": 225080
    },
    {
      "epoch": 1.4436573422312584,
      "grad_norm": 0.03157876431941986,
      "learning_rate": 2.177610860878167e-06,
      "loss": 0.0009,
      "step": 225090
    },
    {
      "epoch": 1.4437214791250446,
      "grad_norm": 0.07756247371435165,
      "learning_rate": 2.1771488724020037e-06,
      "loss": 0.0015,
      "step": 225100
    },
    {
      "epoch": 1.4437856160188307,
      "grad_norm": 0.07325758785009384,
      "learning_rate": 2.1766869192981814e-06,
      "loss": 0.0023,
      "step": 225110
    },
    {
      "epoch": 1.4438497529126166,
      "grad_norm": 0.03857431933283806,
      "learning_rate": 2.17622500157249e-06,
      "loss": 0.0007,
      "step": 225120
    },
    {
      "epoch": 1.4439138898064028,
      "grad_norm": 0.001007712329737842,
      "learning_rate": 2.1757631192307162e-06,
      "loss": 0.0013,
      "step": 225130
    },
    {
      "epoch": 1.443978026700189,
      "grad_norm": 0.042061109095811844,
      "learning_rate": 2.1753012722786502e-06,
      "loss": 0.0009,
      "step": 225140
    },
    {
      "epoch": 1.4440421635939749,
      "grad_norm": 0.062346864491701126,
      "learning_rate": 2.1748394607220786e-06,
      "loss": 0.0012,
      "step": 225150
    },
    {
      "epoch": 1.444106300487761,
      "grad_norm": 0.053323470056056976,
      "learning_rate": 2.174377684566785e-06,
      "loss": 0.0015,
      "step": 225160
    },
    {
      "epoch": 1.4441704373815472,
      "grad_norm": 0.04464549198746681,
      "learning_rate": 2.1739159438185608e-06,
      "loss": 0.0024,
      "step": 225170
    },
    {
      "epoch": 1.4442345742753333,
      "grad_norm": 0.034703902900218964,
      "learning_rate": 2.1734542384831898e-06,
      "loss": 0.0043,
      "step": 225180
    },
    {
      "epoch": 1.4442987111691195,
      "grad_norm": 0.022805416956543922,
      "learning_rate": 2.172992568566457e-06,
      "loss": 0.0015,
      "step": 225190
    },
    {
      "epoch": 1.4443628480629056,
      "grad_norm": 0.0011144662275910378,
      "learning_rate": 2.1725309340741464e-06,
      "loss": 0.0011,
      "step": 225200
    },
    {
      "epoch": 1.4444269849566915,
      "grad_norm": 0.09161978960037231,
      "learning_rate": 2.1720693350120433e-06,
      "loss": 0.002,
      "step": 225210
    },
    {
      "epoch": 1.4444911218504777,
      "grad_norm": 0.03671182692050934,
      "learning_rate": 2.1716077713859345e-06,
      "loss": 0.0019,
      "step": 225220
    },
    {
      "epoch": 1.4445552587442638,
      "grad_norm": 0.14577430486679077,
      "learning_rate": 2.171146243201602e-06,
      "loss": 0.0022,
      "step": 225230
    },
    {
      "epoch": 1.4446193956380498,
      "grad_norm": 0.055364321917295456,
      "learning_rate": 2.1706847504648286e-06,
      "loss": 0.0016,
      "step": 225240
    },
    {
      "epoch": 1.444683532531836,
      "grad_norm": 0.016351599246263504,
      "learning_rate": 2.1702232931813956e-06,
      "loss": 0.0011,
      "step": 225250
    },
    {
      "epoch": 1.444747669425622,
      "grad_norm": 0.04354793205857277,
      "learning_rate": 2.169761871357089e-06,
      "loss": 0.0011,
      "step": 225260
    },
    {
      "epoch": 1.4448118063194082,
      "grad_norm": 0.04667946696281433,
      "learning_rate": 2.169300484997689e-06,
      "loss": 0.0007,
      "step": 225270
    },
    {
      "epoch": 1.4448759432131943,
      "grad_norm": 0.039875585585832596,
      "learning_rate": 2.1688391341089748e-06,
      "loss": 0.001,
      "step": 225280
    },
    {
      "epoch": 1.4449400801069803,
      "grad_norm": 0.1300768107175827,
      "learning_rate": 2.168377818696732e-06,
      "loss": 0.0021,
      "step": 225290
    },
    {
      "epoch": 1.4450042170007664,
      "grad_norm": 0.07227897644042969,
      "learning_rate": 2.1679165387667386e-06,
      "loss": 0.0008,
      "step": 225300
    },
    {
      "epoch": 1.4450683538945526,
      "grad_norm": 0.07283520698547363,
      "learning_rate": 2.167455294324775e-06,
      "loss": 0.0011,
      "step": 225310
    },
    {
      "epoch": 1.4451324907883385,
      "grad_norm": 0.011886750347912312,
      "learning_rate": 2.166994085376619e-06,
      "loss": 0.0006,
      "step": 225320
    },
    {
      "epoch": 1.4451966276821246,
      "grad_norm": 0.030358009040355682,
      "learning_rate": 2.1665329119280537e-06,
      "loss": 0.0015,
      "step": 225330
    },
    {
      "epoch": 1.4452607645759108,
      "grad_norm": 0.0716029554605484,
      "learning_rate": 2.1660717739848565e-06,
      "loss": 0.0017,
      "step": 225340
    },
    {
      "epoch": 1.445324901469697,
      "grad_norm": 0.0746510773897171,
      "learning_rate": 2.1656106715528045e-06,
      "loss": 0.0025,
      "step": 225350
    },
    {
      "epoch": 1.445389038363483,
      "grad_norm": 0.0027565364725887775,
      "learning_rate": 2.165149604637676e-06,
      "loss": 0.0017,
      "step": 225360
    },
    {
      "epoch": 1.4454531752572692,
      "grad_norm": 0.06034074351191521,
      "learning_rate": 2.1646885732452492e-06,
      "loss": 0.0007,
      "step": 225370
    },
    {
      "epoch": 1.4455173121510552,
      "grad_norm": 0.06511419266462326,
      "learning_rate": 2.164227577381302e-06,
      "loss": 0.0017,
      "step": 225380
    },
    {
      "epoch": 1.4455814490448413,
      "grad_norm": 0.03042871318757534,
      "learning_rate": 2.1637666170516096e-06,
      "loss": 0.0006,
      "step": 225390
    },
    {
      "epoch": 1.4456455859386275,
      "grad_norm": 0.13117969036102295,
      "learning_rate": 2.163305692261947e-06,
      "loss": 0.0017,
      "step": 225400
    },
    {
      "epoch": 1.4457097228324134,
      "grad_norm": 0.14985649287700653,
      "learning_rate": 2.1628448030180933e-06,
      "loss": 0.0012,
      "step": 225410
    },
    {
      "epoch": 1.4457738597261995,
      "grad_norm": 0.004519871901720762,
      "learning_rate": 2.1623839493258213e-06,
      "loss": 0.0013,
      "step": 225420
    },
    {
      "epoch": 1.4458379966199857,
      "grad_norm": 0.0251761544495821,
      "learning_rate": 2.161923131190905e-06,
      "loss": 0.0006,
      "step": 225430
    },
    {
      "epoch": 1.4459021335137718,
      "grad_norm": 0.05949077755212784,
      "learning_rate": 2.161462348619122e-06,
      "loss": 0.002,
      "step": 225440
    },
    {
      "epoch": 1.445966270407558,
      "grad_norm": 0.04768809303641319,
      "learning_rate": 2.1610016016162443e-06,
      "loss": 0.0011,
      "step": 225450
    },
    {
      "epoch": 1.446030407301344,
      "grad_norm": 0.005062917247414589,
      "learning_rate": 2.1605408901880458e-06,
      "loss": 0.0007,
      "step": 225460
    },
    {
      "epoch": 1.44609454419513,
      "grad_norm": 0.10849764943122864,
      "learning_rate": 2.160080214340297e-06,
      "loss": 0.0016,
      "step": 225470
    },
    {
      "epoch": 1.4461586810889162,
      "grad_norm": 0.01204383559525013,
      "learning_rate": 2.159619574078775e-06,
      "loss": 0.0015,
      "step": 225480
    },
    {
      "epoch": 1.4462228179827024,
      "grad_norm": 0.03433408588171005,
      "learning_rate": 2.1591589694092496e-06,
      "loss": 0.0012,
      "step": 225490
    },
    {
      "epoch": 1.4462869548764883,
      "grad_norm": 0.12959915399551392,
      "learning_rate": 2.158698400337493e-06,
      "loss": 0.0008,
      "step": 225500
    },
    {
      "epoch": 1.4463510917702744,
      "grad_norm": 0.015933355316519737,
      "learning_rate": 2.1582378668692738e-06,
      "loss": 0.0009,
      "step": 225510
    },
    {
      "epoch": 1.4464152286640606,
      "grad_norm": 0.005979140289127827,
      "learning_rate": 2.157777369010367e-06,
      "loss": 0.0022,
      "step": 225520
    },
    {
      "epoch": 1.4464793655578467,
      "grad_norm": 0.08340150117874146,
      "learning_rate": 2.1573169067665413e-06,
      "loss": 0.0019,
      "step": 225530
    },
    {
      "epoch": 1.4465435024516329,
      "grad_norm": 0.01124319713562727,
      "learning_rate": 2.156856480143566e-06,
      "loss": 0.0015,
      "step": 225540
    },
    {
      "epoch": 1.4466076393454188,
      "grad_norm": 0.05296528711915016,
      "learning_rate": 2.1563960891472096e-06,
      "loss": 0.0006,
      "step": 225550
    },
    {
      "epoch": 1.446671776239205,
      "grad_norm": 0.06643268465995789,
      "learning_rate": 2.1559357337832436e-06,
      "loss": 0.0009,
      "step": 225560
    },
    {
      "epoch": 1.446735913132991,
      "grad_norm": 0.06128658354282379,
      "learning_rate": 2.1554754140574364e-06,
      "loss": 0.0008,
      "step": 225570
    },
    {
      "epoch": 1.446800050026777,
      "grad_norm": 0.049856431782245636,
      "learning_rate": 2.1550151299755527e-06,
      "loss": 0.0012,
      "step": 225580
    },
    {
      "epoch": 1.4468641869205632,
      "grad_norm": 0.07697924226522446,
      "learning_rate": 2.1545548815433647e-06,
      "loss": 0.0013,
      "step": 225590
    },
    {
      "epoch": 1.4469283238143493,
      "grad_norm": 0.05229620262980461,
      "learning_rate": 2.154094668766638e-06,
      "loss": 0.0012,
      "step": 225600
    },
    {
      "epoch": 1.4469924607081355,
      "grad_norm": 0.07874248921871185,
      "learning_rate": 2.1536344916511387e-06,
      "loss": 0.0022,
      "step": 225610
    },
    {
      "epoch": 1.4470565976019216,
      "grad_norm": 0.03423517569899559,
      "learning_rate": 2.1531743502026318e-06,
      "loss": 0.0009,
      "step": 225620
    },
    {
      "epoch": 1.4471207344957078,
      "grad_norm": 0.21702076494693756,
      "learning_rate": 2.1527142444268866e-06,
      "loss": 0.0015,
      "step": 225630
    },
    {
      "epoch": 1.4471848713894937,
      "grad_norm": 0.0025034372229129076,
      "learning_rate": 2.152254174329667e-06,
      "loss": 0.002,
      "step": 225640
    },
    {
      "epoch": 1.4472490082832798,
      "grad_norm": 0.0951765701174736,
      "learning_rate": 2.1517941399167373e-06,
      "loss": 0.0017,
      "step": 225650
    },
    {
      "epoch": 1.447313145177066,
      "grad_norm": 0.01177617721259594,
      "learning_rate": 2.1513341411938614e-06,
      "loss": 0.0009,
      "step": 225660
    },
    {
      "epoch": 1.447377282070852,
      "grad_norm": 0.04554301127791405,
      "learning_rate": 2.1508741781668064e-06,
      "loss": 0.0012,
      "step": 225670
    },
    {
      "epoch": 1.447441418964638,
      "grad_norm": 0.07917334139347076,
      "learning_rate": 2.150414250841334e-06,
      "loss": 0.0012,
      "step": 225680
    },
    {
      "epoch": 1.4475055558584242,
      "grad_norm": 0.21603022515773773,
      "learning_rate": 2.1499543592232075e-06,
      "loss": 0.001,
      "step": 225690
    },
    {
      "epoch": 1.4475696927522104,
      "grad_norm": 0.2080763280391693,
      "learning_rate": 2.149494503318188e-06,
      "loss": 0.0013,
      "step": 225700
    },
    {
      "epoch": 1.4476338296459965,
      "grad_norm": 0.12338167428970337,
      "learning_rate": 2.149034683132042e-06,
      "loss": 0.002,
      "step": 225710
    },
    {
      "epoch": 1.4476979665397824,
      "grad_norm": 0.06685017049312592,
      "learning_rate": 2.1485748986705285e-06,
      "loss": 0.0005,
      "step": 225720
    },
    {
      "epoch": 1.4477621034335686,
      "grad_norm": 0.17813804745674133,
      "learning_rate": 2.148115149939407e-06,
      "loss": 0.001,
      "step": 225730
    },
    {
      "epoch": 1.4478262403273547,
      "grad_norm": 0.11285127699375153,
      "learning_rate": 2.1476554369444438e-06,
      "loss": 0.001,
      "step": 225740
    },
    {
      "epoch": 1.4478903772211407,
      "grad_norm": 0.05985259637236595,
      "learning_rate": 2.1471957596913957e-06,
      "loss": 0.0012,
      "step": 225750
    },
    {
      "epoch": 1.4479545141149268,
      "grad_norm": 0.06689460575580597,
      "learning_rate": 2.146736118186024e-06,
      "loss": 0.0015,
      "step": 225760
    },
    {
      "epoch": 1.448018651008713,
      "grad_norm": 0.05374080315232277,
      "learning_rate": 2.146276512434086e-06,
      "loss": 0.0011,
      "step": 225770
    },
    {
      "epoch": 1.4480827879024991,
      "grad_norm": 0.003331320360302925,
      "learning_rate": 2.1458169424413445e-06,
      "loss": 0.0006,
      "step": 225780
    },
    {
      "epoch": 1.4481469247962853,
      "grad_norm": 0.03331978991627693,
      "learning_rate": 2.145357408213557e-06,
      "loss": 0.0009,
      "step": 225790
    },
    {
      "epoch": 1.4482110616900714,
      "grad_norm": 0.18725016713142395,
      "learning_rate": 2.144897909756481e-06,
      "loss": 0.0011,
      "step": 225800
    },
    {
      "epoch": 1.4482751985838573,
      "grad_norm": 0.1007186695933342,
      "learning_rate": 2.144438447075873e-06,
      "loss": 0.0016,
      "step": 225810
    },
    {
      "epoch": 1.4483393354776435,
      "grad_norm": 0.02833637408912182,
      "learning_rate": 2.143979020177494e-06,
      "loss": 0.0011,
      "step": 225820
    },
    {
      "epoch": 1.4484034723714296,
      "grad_norm": 0.06066592410206795,
      "learning_rate": 2.143519629067099e-06,
      "loss": 0.0019,
      "step": 225830
    },
    {
      "epoch": 1.4484676092652156,
      "grad_norm": 0.1573522984981537,
      "learning_rate": 2.143060273750444e-06,
      "loss": 0.0018,
      "step": 225840
    },
    {
      "epoch": 1.4485317461590017,
      "grad_norm": 0.045502424240112305,
      "learning_rate": 2.142600954233284e-06,
      "loss": 0.0022,
      "step": 225850
    },
    {
      "epoch": 1.4485958830527879,
      "grad_norm": 0.0707055926322937,
      "learning_rate": 2.142141670521379e-06,
      "loss": 0.0008,
      "step": 225860
    },
    {
      "epoch": 1.448660019946574,
      "grad_norm": 0.22369283437728882,
      "learning_rate": 2.14168242262048e-06,
      "loss": 0.0022,
      "step": 225870
    },
    {
      "epoch": 1.4487241568403602,
      "grad_norm": 0.071811243891716,
      "learning_rate": 2.141223210536342e-06,
      "loss": 0.0015,
      "step": 225880
    },
    {
      "epoch": 1.448788293734146,
      "grad_norm": 0.03724861890077591,
      "learning_rate": 2.140764034274722e-06,
      "loss": 0.0012,
      "step": 225890
    },
    {
      "epoch": 1.4488524306279322,
      "grad_norm": 0.11753320693969727,
      "learning_rate": 2.140304893841372e-06,
      "loss": 0.0019,
      "step": 225900
    },
    {
      "epoch": 1.4489165675217184,
      "grad_norm": 0.04428223520517349,
      "learning_rate": 2.1398457892420454e-06,
      "loss": 0.0011,
      "step": 225910
    },
    {
      "epoch": 1.4489807044155045,
      "grad_norm": 0.05743437632918358,
      "learning_rate": 2.1393867204824936e-06,
      "loss": 0.001,
      "step": 225920
    },
    {
      "epoch": 1.4490448413092905,
      "grad_norm": 0.08938732743263245,
      "learning_rate": 2.1389276875684727e-06,
      "loss": 0.0011,
      "step": 225930
    },
    {
      "epoch": 1.4491089782030766,
      "grad_norm": 0.00183452432975173,
      "learning_rate": 2.138468690505732e-06,
      "loss": 0.0018,
      "step": 225940
    },
    {
      "epoch": 1.4491731150968628,
      "grad_norm": 0.12324772030115128,
      "learning_rate": 2.138009729300024e-06,
      "loss": 0.0023,
      "step": 225950
    },
    {
      "epoch": 1.449237251990649,
      "grad_norm": 0.002847007242962718,
      "learning_rate": 2.137550803957098e-06,
      "loss": 0.0008,
      "step": 225960
    },
    {
      "epoch": 1.449301388884435,
      "grad_norm": 0.13642844557762146,
      "learning_rate": 2.137091914482706e-06,
      "loss": 0.0009,
      "step": 225970
    },
    {
      "epoch": 1.449365525778221,
      "grad_norm": 0.01883120648562908,
      "learning_rate": 2.1366330608826004e-06,
      "loss": 0.0011,
      "step": 225980
    },
    {
      "epoch": 1.4494296626720071,
      "grad_norm": 0.04838910698890686,
      "learning_rate": 2.136174243162528e-06,
      "loss": 0.0016,
      "step": 225990
    },
    {
      "epoch": 1.4494937995657933,
      "grad_norm": 0.14629925787448883,
      "learning_rate": 2.135715461328238e-06,
      "loss": 0.0023,
      "step": 226000
    },
    {
      "epoch": 1.4495579364595792,
      "grad_norm": 0.17296472191810608,
      "learning_rate": 2.1352567153854825e-06,
      "loss": 0.0011,
      "step": 226010
    },
    {
      "epoch": 1.4496220733533653,
      "grad_norm": 0.10599583387374878,
      "learning_rate": 2.1347980053400076e-06,
      "loss": 0.001,
      "step": 226020
    },
    {
      "epoch": 1.4496862102471515,
      "grad_norm": 0.11168216913938522,
      "learning_rate": 2.1343393311975612e-06,
      "loss": 0.0016,
      "step": 226030
    },
    {
      "epoch": 1.4497503471409376,
      "grad_norm": 0.155610591173172,
      "learning_rate": 2.133880692963889e-06,
      "loss": 0.0019,
      "step": 226040
    },
    {
      "epoch": 1.4498144840347238,
      "grad_norm": 0.03576747328042984,
      "learning_rate": 2.133422090644742e-06,
      "loss": 0.0008,
      "step": 226050
    },
    {
      "epoch": 1.44987862092851,
      "grad_norm": 0.06736559420824051,
      "learning_rate": 2.1329635242458653e-06,
      "loss": 0.0013,
      "step": 226060
    },
    {
      "epoch": 1.4499427578222959,
      "grad_norm": 0.04706592112779617,
      "learning_rate": 2.1325049937730045e-06,
      "loss": 0.0007,
      "step": 226070
    },
    {
      "epoch": 1.450006894716082,
      "grad_norm": 0.16581174731254578,
      "learning_rate": 2.1320464992319035e-06,
      "loss": 0.002,
      "step": 226080
    },
    {
      "epoch": 1.4500710316098682,
      "grad_norm": 0.1650496870279312,
      "learning_rate": 2.131588040628312e-06,
      "loss": 0.0014,
      "step": 226090
    },
    {
      "epoch": 1.450135168503654,
      "grad_norm": 0.19719888269901276,
      "learning_rate": 2.1311296179679715e-06,
      "loss": 0.0007,
      "step": 226100
    },
    {
      "epoch": 1.4501993053974402,
      "grad_norm": 0.02815418690443039,
      "learning_rate": 2.1306712312566257e-06,
      "loss": 0.0011,
      "step": 226110
    },
    {
      "epoch": 1.4502634422912264,
      "grad_norm": 0.05859525874257088,
      "learning_rate": 2.1302128805000223e-06,
      "loss": 0.0011,
      "step": 226120
    },
    {
      "epoch": 1.4503275791850125,
      "grad_norm": 0.118864044547081,
      "learning_rate": 2.129754565703902e-06,
      "loss": 0.0025,
      "step": 226130
    },
    {
      "epoch": 1.4503917160787987,
      "grad_norm": 0.09736819565296173,
      "learning_rate": 2.1292962868740084e-06,
      "loss": 0.0036,
      "step": 226140
    },
    {
      "epoch": 1.4504558529725846,
      "grad_norm": 0.1656569093465805,
      "learning_rate": 2.128838044016083e-06,
      "loss": 0.0022,
      "step": 226150
    },
    {
      "epoch": 1.4505199898663708,
      "grad_norm": 0.0474659688770771,
      "learning_rate": 2.1283798371358694e-06,
      "loss": 0.0007,
      "step": 226160
    },
    {
      "epoch": 1.450584126760157,
      "grad_norm": 0.0348595455288887,
      "learning_rate": 2.1279216662391094e-06,
      "loss": 0.0005,
      "step": 226170
    },
    {
      "epoch": 1.4506482636539428,
      "grad_norm": 0.09374228119850159,
      "learning_rate": 2.1274635313315434e-06,
      "loss": 0.0015,
      "step": 226180
    },
    {
      "epoch": 1.450712400547729,
      "grad_norm": 0.17273084819316864,
      "learning_rate": 2.127005432418911e-06,
      "loss": 0.0015,
      "step": 226190
    },
    {
      "epoch": 1.4507765374415151,
      "grad_norm": 0.24151749908924103,
      "learning_rate": 2.126547369506955e-06,
      "loss": 0.002,
      "step": 226200
    },
    {
      "epoch": 1.4508406743353013,
      "grad_norm": 0.010909780859947205,
      "learning_rate": 2.1260893426014145e-06,
      "loss": 0.0003,
      "step": 226210
    },
    {
      "epoch": 1.4509048112290874,
      "grad_norm": 0.09100042283535004,
      "learning_rate": 2.1256313517080284e-06,
      "loss": 0.0007,
      "step": 226220
    },
    {
      "epoch": 1.4509689481228736,
      "grad_norm": 0.12479095906019211,
      "learning_rate": 2.125173396832534e-06,
      "loss": 0.0024,
      "step": 226230
    },
    {
      "epoch": 1.4510330850166595,
      "grad_norm": 0.1252729892730713,
      "learning_rate": 2.1247154779806726e-06,
      "loss": 0.0007,
      "step": 226240
    },
    {
      "epoch": 1.4510972219104457,
      "grad_norm": 0.05272957310080528,
      "learning_rate": 2.1242575951581823e-06,
      "loss": 0.0021,
      "step": 226250
    },
    {
      "epoch": 1.4511613588042318,
      "grad_norm": 0.01092084962874651,
      "learning_rate": 2.1237997483707974e-06,
      "loss": 0.0016,
      "step": 226260
    },
    {
      "epoch": 1.4512254956980177,
      "grad_norm": 0.09231545776128769,
      "learning_rate": 2.1233419376242587e-06,
      "loss": 0.0018,
      "step": 226270
    },
    {
      "epoch": 1.4512896325918039,
      "grad_norm": 0.08255818486213684,
      "learning_rate": 2.122884162924302e-06,
      "loss": 0.0039,
      "step": 226280
    },
    {
      "epoch": 1.45135376948559,
      "grad_norm": 0.050203077495098114,
      "learning_rate": 2.1224264242766625e-06,
      "loss": 0.001,
      "step": 226290
    },
    {
      "epoch": 1.4514179063793762,
      "grad_norm": 0.13449114561080933,
      "learning_rate": 2.1219687216870747e-06,
      "loss": 0.0025,
      "step": 226300
    },
    {
      "epoch": 1.4514820432731623,
      "grad_norm": 0.047507259994745255,
      "learning_rate": 2.121511055161278e-06,
      "loss": 0.0008,
      "step": 226310
    },
    {
      "epoch": 1.4515461801669483,
      "grad_norm": 0.22218233346939087,
      "learning_rate": 2.1210534247050045e-06,
      "loss": 0.0017,
      "step": 226320
    },
    {
      "epoch": 1.4516103170607344,
      "grad_norm": 0.09313784539699554,
      "learning_rate": 2.1205958303239892e-06,
      "loss": 0.0026,
      "step": 226330
    },
    {
      "epoch": 1.4516744539545205,
      "grad_norm": 0.25622984766960144,
      "learning_rate": 2.1201382720239644e-06,
      "loss": 0.0022,
      "step": 226340
    },
    {
      "epoch": 1.4517385908483067,
      "grad_norm": 0.08348685503005981,
      "learning_rate": 2.119680749810667e-06,
      "loss": 0.0006,
      "step": 226350
    },
    {
      "epoch": 1.4518027277420926,
      "grad_norm": 0.007414282765239477,
      "learning_rate": 2.1192232636898285e-06,
      "loss": 0.0028,
      "step": 226360
    },
    {
      "epoch": 1.4518668646358788,
      "grad_norm": 0.06608813256025314,
      "learning_rate": 2.118765813667181e-06,
      "loss": 0.0006,
      "step": 226370
    },
    {
      "epoch": 1.451931001529665,
      "grad_norm": 0.07415993511676788,
      "learning_rate": 2.1183083997484554e-06,
      "loss": 0.001,
      "step": 226380
    },
    {
      "epoch": 1.451995138423451,
      "grad_norm": 0.09311382472515106,
      "learning_rate": 2.1178510219393873e-06,
      "loss": 0.0017,
      "step": 226390
    },
    {
      "epoch": 1.4520592753172372,
      "grad_norm": 0.04318315535783768,
      "learning_rate": 2.117393680245705e-06,
      "loss": 0.0015,
      "step": 226400
    },
    {
      "epoch": 1.4521234122110231,
      "grad_norm": 0.028543667867779732,
      "learning_rate": 2.116936374673139e-06,
      "loss": 0.0005,
      "step": 226410
    },
    {
      "epoch": 1.4521875491048093,
      "grad_norm": 0.0671742632985115,
      "learning_rate": 2.116479105227422e-06,
      "loss": 0.0019,
      "step": 226420
    },
    {
      "epoch": 1.4522516859985954,
      "grad_norm": 0.07890462130308151,
      "learning_rate": 2.1160218719142827e-06,
      "loss": 0.0014,
      "step": 226430
    },
    {
      "epoch": 1.4523158228923814,
      "grad_norm": 0.08221903443336487,
      "learning_rate": 2.1155646747394505e-06,
      "loss": 0.0016,
      "step": 226440
    },
    {
      "epoch": 1.4523799597861675,
      "grad_norm": 0.2833114564418793,
      "learning_rate": 2.1151075137086522e-06,
      "loss": 0.0014,
      "step": 226450
    },
    {
      "epoch": 1.4524440966799537,
      "grad_norm": 0.11079540848731995,
      "learning_rate": 2.1146503888276214e-06,
      "loss": 0.0016,
      "step": 226460
    },
    {
      "epoch": 1.4525082335737398,
      "grad_norm": 0.18482473492622375,
      "learning_rate": 2.1141933001020826e-06,
      "loss": 0.0015,
      "step": 226470
    },
    {
      "epoch": 1.452572370467526,
      "grad_norm": 0.03776835277676582,
      "learning_rate": 2.1137362475377643e-06,
      "loss": 0.0018,
      "step": 226480
    },
    {
      "epoch": 1.4526365073613121,
      "grad_norm": 0.018275147303938866,
      "learning_rate": 2.1132792311403922e-06,
      "loss": 0.0011,
      "step": 226490
    },
    {
      "epoch": 1.452700644255098,
      "grad_norm": 0.01807580702006817,
      "learning_rate": 2.112822250915696e-06,
      "loss": 0.0009,
      "step": 226500
    },
    {
      "epoch": 1.4527647811488842,
      "grad_norm": 0.05914665013551712,
      "learning_rate": 2.1123653068694e-06,
      "loss": 0.0009,
      "step": 226510
    },
    {
      "epoch": 1.4528289180426703,
      "grad_norm": 0.006592872552573681,
      "learning_rate": 2.111908399007231e-06,
      "loss": 0.001,
      "step": 226520
    },
    {
      "epoch": 1.4528930549364563,
      "grad_norm": 0.04994434863328934,
      "learning_rate": 2.111451527334912e-06,
      "loss": 0.001,
      "step": 226530
    },
    {
      "epoch": 1.4529571918302424,
      "grad_norm": 0.04584444686770439,
      "learning_rate": 2.1109946918581715e-06,
      "loss": 0.0011,
      "step": 226540
    },
    {
      "epoch": 1.4530213287240286,
      "grad_norm": 0.0671549141407013,
      "learning_rate": 2.1105378925827317e-06,
      "loss": 0.0015,
      "step": 226550
    },
    {
      "epoch": 1.4530854656178147,
      "grad_norm": 0.05410590395331383,
      "learning_rate": 2.110081129514316e-06,
      "loss": 0.0008,
      "step": 226560
    },
    {
      "epoch": 1.4531496025116009,
      "grad_norm": 0.09354747831821442,
      "learning_rate": 2.109624402658651e-06,
      "loss": 0.0014,
      "step": 226570
    },
    {
      "epoch": 1.4532137394053868,
      "grad_norm": 0.0746600553393364,
      "learning_rate": 2.1091677120214577e-06,
      "loss": 0.0013,
      "step": 226580
    },
    {
      "epoch": 1.453277876299173,
      "grad_norm": 0.08220326155424118,
      "learning_rate": 2.108711057608459e-06,
      "loss": 0.0013,
      "step": 226590
    },
    {
      "epoch": 1.453342013192959,
      "grad_norm": 0.09785089641809464,
      "learning_rate": 2.108254439425375e-06,
      "loss": 0.0017,
      "step": 226600
    },
    {
      "epoch": 1.4534061500867452,
      "grad_norm": 0.016085678711533546,
      "learning_rate": 2.107797857477932e-06,
      "loss": 0.0009,
      "step": 226610
    },
    {
      "epoch": 1.4534702869805312,
      "grad_norm": 0.11783857643604279,
      "learning_rate": 2.1073413117718488e-06,
      "loss": 0.0009,
      "step": 226620
    },
    {
      "epoch": 1.4535344238743173,
      "grad_norm": 0.2694298326969147,
      "learning_rate": 2.1068848023128464e-06,
      "loss": 0.0016,
      "step": 226630
    },
    {
      "epoch": 1.4535985607681035,
      "grad_norm": 0.10376250743865967,
      "learning_rate": 2.1064283291066427e-06,
      "loss": 0.0018,
      "step": 226640
    },
    {
      "epoch": 1.4536626976618896,
      "grad_norm": 0.030363505706191063,
      "learning_rate": 2.105971892158962e-06,
      "loss": 0.0014,
      "step": 226650
    },
    {
      "epoch": 1.4537268345556758,
      "grad_norm": 0.020838048309087753,
      "learning_rate": 2.105515491475522e-06,
      "loss": 0.0019,
      "step": 226660
    },
    {
      "epoch": 1.4537909714494617,
      "grad_norm": 0.08939540386199951,
      "learning_rate": 2.105059127062042e-06,
      "loss": 0.0018,
      "step": 226670
    },
    {
      "epoch": 1.4538551083432478,
      "grad_norm": 0.1325758397579193,
      "learning_rate": 2.104602798924238e-06,
      "loss": 0.0007,
      "step": 226680
    },
    {
      "epoch": 1.453919245237034,
      "grad_norm": 0.060680270195007324,
      "learning_rate": 2.104146507067832e-06,
      "loss": 0.0013,
      "step": 226690
    },
    {
      "epoch": 1.45398338213082,
      "grad_norm": 0.08166959881782532,
      "learning_rate": 2.1036902514985396e-06,
      "loss": 0.0011,
      "step": 226700
    },
    {
      "epoch": 1.454047519024606,
      "grad_norm": 0.15600843727588654,
      "learning_rate": 2.1032340322220766e-06,
      "loss": 0.0009,
      "step": 226710
    },
    {
      "epoch": 1.4541116559183922,
      "grad_norm": 0.006155435461550951,
      "learning_rate": 2.1027778492441636e-06,
      "loss": 0.0009,
      "step": 226720
    },
    {
      "epoch": 1.4541757928121783,
      "grad_norm": 0.03900561481714249,
      "learning_rate": 2.1023217025705143e-06,
      "loss": 0.0013,
      "step": 226730
    },
    {
      "epoch": 1.4542399297059645,
      "grad_norm": 0.04580393061041832,
      "learning_rate": 2.1018655922068436e-06,
      "loss": 0.0013,
      "step": 226740
    },
    {
      "epoch": 1.4543040665997506,
      "grad_norm": 0.00805017538368702,
      "learning_rate": 2.1014095181588703e-06,
      "loss": 0.0011,
      "step": 226750
    },
    {
      "epoch": 1.4543682034935366,
      "grad_norm": 0.10442374646663666,
      "learning_rate": 2.1009534804323045e-06,
      "loss": 0.0007,
      "step": 226760
    },
    {
      "epoch": 1.4544323403873227,
      "grad_norm": 0.15928815305233002,
      "learning_rate": 2.1004974790328665e-06,
      "loss": 0.0007,
      "step": 226770
    },
    {
      "epoch": 1.4544964772811089,
      "grad_norm": 0.010206308215856552,
      "learning_rate": 2.1000415139662667e-06,
      "loss": 0.0016,
      "step": 226780
    },
    {
      "epoch": 1.4545606141748948,
      "grad_norm": 0.25000104308128357,
      "learning_rate": 2.099585585238219e-06,
      "loss": 0.0015,
      "step": 226790
    },
    {
      "epoch": 1.454624751068681,
      "grad_norm": 0.05201121047139168,
      "learning_rate": 2.0991296928544357e-06,
      "loss": 0.001,
      "step": 226800
    },
    {
      "epoch": 1.454688887962467,
      "grad_norm": 0.010511612519621849,
      "learning_rate": 2.098673836820632e-06,
      "loss": 0.0006,
      "step": 226810
    },
    {
      "epoch": 1.4547530248562532,
      "grad_norm": 0.08724335581064224,
      "learning_rate": 2.098218017142519e-06,
      "loss": 0.0024,
      "step": 226820
    },
    {
      "epoch": 1.4548171617500394,
      "grad_norm": 0.14763133227825165,
      "learning_rate": 2.0977622338258056e-06,
      "loss": 0.0009,
      "step": 226830
    },
    {
      "epoch": 1.4548812986438253,
      "grad_norm": 0.17446410655975342,
      "learning_rate": 2.0973064868762084e-06,
      "loss": 0.0012,
      "step": 226840
    },
    {
      "epoch": 1.4549454355376115,
      "grad_norm": 0.028958257287740707,
      "learning_rate": 2.096850776299435e-06,
      "loss": 0.0014,
      "step": 226850
    },
    {
      "epoch": 1.4550095724313976,
      "grad_norm": 0.1068197637796402,
      "learning_rate": 2.0963951021011965e-06,
      "loss": 0.0013,
      "step": 226860
    },
    {
      "epoch": 1.4550737093251835,
      "grad_norm": 0.016174737364053726,
      "learning_rate": 2.0959394642871998e-06,
      "loss": 0.0008,
      "step": 226870
    },
    {
      "epoch": 1.4551378462189697,
      "grad_norm": 0.07588784396648407,
      "learning_rate": 2.09548386286316e-06,
      "loss": 0.0029,
      "step": 226880
    },
    {
      "epoch": 1.4552019831127558,
      "grad_norm": 0.02686605043709278,
      "learning_rate": 2.0950282978347826e-06,
      "loss": 0.0011,
      "step": 226890
    },
    {
      "epoch": 1.455266120006542,
      "grad_norm": 0.05538346245884895,
      "learning_rate": 2.094572769207777e-06,
      "loss": 0.0007,
      "step": 226900
    },
    {
      "epoch": 1.4553302569003281,
      "grad_norm": 0.08940229564905167,
      "learning_rate": 2.094117276987849e-06,
      "loss": 0.0012,
      "step": 226910
    },
    {
      "epoch": 1.4553943937941143,
      "grad_norm": 0.022856472060084343,
      "learning_rate": 2.09366182118071e-06,
      "loss": 0.0007,
      "step": 226920
    },
    {
      "epoch": 1.4554585306879002,
      "grad_norm": 0.0348527617752552,
      "learning_rate": 2.093206401792066e-06,
      "loss": 0.0011,
      "step": 226930
    },
    {
      "epoch": 1.4555226675816864,
      "grad_norm": 0.13491670787334442,
      "learning_rate": 2.0927510188276233e-06,
      "loss": 0.0017,
      "step": 226940
    },
    {
      "epoch": 1.4555868044754725,
      "grad_norm": 0.05308113247156143,
      "learning_rate": 2.092295672293086e-06,
      "loss": 0.0017,
      "step": 226950
    },
    {
      "epoch": 1.4556509413692584,
      "grad_norm": 0.06905005872249603,
      "learning_rate": 2.091840362194164e-06,
      "loss": 0.0014,
      "step": 226960
    },
    {
      "epoch": 1.4557150782630446,
      "grad_norm": 0.2603563666343689,
      "learning_rate": 2.0913850885365603e-06,
      "loss": 0.0007,
      "step": 226970
    },
    {
      "epoch": 1.4557792151568307,
      "grad_norm": 0.05765556916594505,
      "learning_rate": 2.090929851325979e-06,
      "loss": 0.0011,
      "step": 226980
    },
    {
      "epoch": 1.4558433520506169,
      "grad_norm": 0.06922811269760132,
      "learning_rate": 2.0904746505681273e-06,
      "loss": 0.0013,
      "step": 226990
    },
    {
      "epoch": 1.455907488944403,
      "grad_norm": 0.044415734708309174,
      "learning_rate": 2.0900194862687077e-06,
      "loss": 0.0007,
      "step": 227000
    },
    {
      "epoch": 1.455971625838189,
      "grad_norm": 0.1271929293870926,
      "learning_rate": 2.0895643584334233e-06,
      "loss": 0.0015,
      "step": 227010
    },
    {
      "epoch": 1.456035762731975,
      "grad_norm": 0.058753691613674164,
      "learning_rate": 2.0891092670679764e-06,
      "loss": 0.0005,
      "step": 227020
    },
    {
      "epoch": 1.4560998996257613,
      "grad_norm": 0.0076009416952729225,
      "learning_rate": 2.088654212178072e-06,
      "loss": 0.001,
      "step": 227030
    },
    {
      "epoch": 1.4561640365195474,
      "grad_norm": 0.06783096492290497,
      "learning_rate": 2.0881991937694114e-06,
      "loss": 0.001,
      "step": 227040
    },
    {
      "epoch": 1.4562281734133333,
      "grad_norm": 0.03720617666840553,
      "learning_rate": 2.0877442118476952e-06,
      "loss": 0.0014,
      "step": 227050
    },
    {
      "epoch": 1.4562923103071195,
      "grad_norm": 0.1779642254114151,
      "learning_rate": 2.0872892664186246e-06,
      "loss": 0.0024,
      "step": 227060
    },
    {
      "epoch": 1.4563564472009056,
      "grad_norm": 0.07750676572322845,
      "learning_rate": 2.0868343574879025e-06,
      "loss": 0.001,
      "step": 227070
    },
    {
      "epoch": 1.4564205840946918,
      "grad_norm": 0.11309818923473358,
      "learning_rate": 2.086379485061228e-06,
      "loss": 0.0008,
      "step": 227080
    },
    {
      "epoch": 1.456484720988478,
      "grad_norm": 0.14714545011520386,
      "learning_rate": 2.0859246491443007e-06,
      "loss": 0.0022,
      "step": 227090
    },
    {
      "epoch": 1.4565488578822638,
      "grad_norm": 0.10395433753728867,
      "learning_rate": 2.0854698497428182e-06,
      "loss": 0.0017,
      "step": 227100
    },
    {
      "epoch": 1.45661299477605,
      "grad_norm": 0.042010944336652756,
      "learning_rate": 2.085015086862484e-06,
      "loss": 0.0012,
      "step": 227110
    },
    {
      "epoch": 1.4566771316698361,
      "grad_norm": 0.0029472862370312214,
      "learning_rate": 2.0845603605089936e-06,
      "loss": 0.0011,
      "step": 227120
    },
    {
      "epoch": 1.456741268563622,
      "grad_norm": 0.00931242760270834,
      "learning_rate": 2.0841056706880437e-06,
      "loss": 0.0007,
      "step": 227130
    },
    {
      "epoch": 1.4568054054574082,
      "grad_norm": 0.11214316636323929,
      "learning_rate": 2.0836510174053355e-06,
      "loss": 0.0049,
      "step": 227140
    },
    {
      "epoch": 1.4568695423511944,
      "grad_norm": 0.10304298996925354,
      "learning_rate": 2.0831964006665644e-06,
      "loss": 0.0007,
      "step": 227150
    },
    {
      "epoch": 1.4569336792449805,
      "grad_norm": 0.044274721294641495,
      "learning_rate": 2.082741820477427e-06,
      "loss": 0.0007,
      "step": 227160
    },
    {
      "epoch": 1.4569978161387667,
      "grad_norm": 0.10848188400268555,
      "learning_rate": 2.0822872768436177e-06,
      "loss": 0.0012,
      "step": 227170
    },
    {
      "epoch": 1.4570619530325528,
      "grad_norm": 0.0019466973608359694,
      "learning_rate": 2.081832769770836e-06,
      "loss": 0.0013,
      "step": 227180
    },
    {
      "epoch": 1.4571260899263387,
      "grad_norm": 0.09821311384439468,
      "learning_rate": 2.081378299264775e-06,
      "loss": 0.001,
      "step": 227190
    },
    {
      "epoch": 1.457190226820125,
      "grad_norm": 0.10010962188243866,
      "learning_rate": 2.0809238653311305e-06,
      "loss": 0.001,
      "step": 227200
    },
    {
      "epoch": 1.457254363713911,
      "grad_norm": 0.07062168419361115,
      "learning_rate": 2.080469467975593e-06,
      "loss": 0.0012,
      "step": 227210
    },
    {
      "epoch": 1.457318500607697,
      "grad_norm": 0.050293177366256714,
      "learning_rate": 2.080015107203863e-06,
      "loss": 0.0011,
      "step": 227220
    },
    {
      "epoch": 1.4573826375014831,
      "grad_norm": 0.032794367522001266,
      "learning_rate": 2.0795607830216295e-06,
      "loss": 0.0014,
      "step": 227230
    },
    {
      "epoch": 1.4574467743952693,
      "grad_norm": 0.25751736760139465,
      "learning_rate": 2.0791064954345873e-06,
      "loss": 0.0013,
      "step": 227240
    },
    {
      "epoch": 1.4575109112890554,
      "grad_norm": 0.14639756083488464,
      "learning_rate": 2.0786522444484257e-06,
      "loss": 0.0015,
      "step": 227250
    },
    {
      "epoch": 1.4575750481828416,
      "grad_norm": 0.048349980264902115,
      "learning_rate": 2.078198030068842e-06,
      "loss": 0.0014,
      "step": 227260
    },
    {
      "epoch": 1.4576391850766275,
      "grad_norm": 0.05995164439082146,
      "learning_rate": 2.077743852301525e-06,
      "loss": 0.0008,
      "step": 227270
    },
    {
      "epoch": 1.4577033219704136,
      "grad_norm": 0.04522133246064186,
      "learning_rate": 2.077289711152164e-06,
      "loss": 0.0008,
      "step": 227280
    },
    {
      "epoch": 1.4577674588641998,
      "grad_norm": 0.17663951218128204,
      "learning_rate": 2.0768356066264532e-06,
      "loss": 0.0009,
      "step": 227290
    },
    {
      "epoch": 1.4578315957579857,
      "grad_norm": 0.006134978495538235,
      "learning_rate": 2.0763815387300817e-06,
      "loss": 0.001,
      "step": 227300
    },
    {
      "epoch": 1.4578957326517719,
      "grad_norm": 0.03722411394119263,
      "learning_rate": 2.075927507468739e-06,
      "loss": 0.0013,
      "step": 227310
    },
    {
      "epoch": 1.457959869545558,
      "grad_norm": 0.15329819917678833,
      "learning_rate": 2.075473512848113e-06,
      "loss": 0.0014,
      "step": 227320
    },
    {
      "epoch": 1.4580240064393442,
      "grad_norm": 0.062296733260154724,
      "learning_rate": 2.075019554873895e-06,
      "loss": 0.0008,
      "step": 227330
    },
    {
      "epoch": 1.4580881433331303,
      "grad_norm": 0.12302592396736145,
      "learning_rate": 2.074565633551773e-06,
      "loss": 0.0027,
      "step": 227340
    },
    {
      "epoch": 1.4581522802269165,
      "grad_norm": 0.09618202596902847,
      "learning_rate": 2.0741117488874347e-06,
      "loss": 0.0013,
      "step": 227350
    },
    {
      "epoch": 1.4582164171207024,
      "grad_norm": 0.024977801367640495,
      "learning_rate": 2.0736579008865647e-06,
      "loss": 0.0016,
      "step": 227360
    },
    {
      "epoch": 1.4582805540144885,
      "grad_norm": 0.08137394487857819,
      "learning_rate": 2.073204089554855e-06,
      "loss": 0.0011,
      "step": 227370
    },
    {
      "epoch": 1.4583446909082747,
      "grad_norm": 0.008541885763406754,
      "learning_rate": 2.0727503148979894e-06,
      "loss": 0.0021,
      "step": 227380
    },
    {
      "epoch": 1.4584088278020606,
      "grad_norm": 0.05951809138059616,
      "learning_rate": 2.0722965769216548e-06,
      "loss": 0.0008,
      "step": 227390
    },
    {
      "epoch": 1.4584729646958468,
      "grad_norm": 0.03887954726815224,
      "learning_rate": 2.0718428756315346e-06,
      "loss": 0.0013,
      "step": 227400
    },
    {
      "epoch": 1.458537101589633,
      "grad_norm": 0.01444583386182785,
      "learning_rate": 2.0713892110333173e-06,
      "loss": 0.0015,
      "step": 227410
    },
    {
      "epoch": 1.458601238483419,
      "grad_norm": 0.07182031124830246,
      "learning_rate": 2.070935583132687e-06,
      "loss": 0.0021,
      "step": 227420
    },
    {
      "epoch": 1.4586653753772052,
      "grad_norm": 0.09865667670965195,
      "learning_rate": 2.0704819919353246e-06,
      "loss": 0.001,
      "step": 227430
    },
    {
      "epoch": 1.4587295122709911,
      "grad_norm": 0.09204527735710144,
      "learning_rate": 2.0700284374469185e-06,
      "loss": 0.0017,
      "step": 227440
    },
    {
      "epoch": 1.4587936491647773,
      "grad_norm": 0.017316631972789764,
      "learning_rate": 2.069574919673151e-06,
      "loss": 0.0016,
      "step": 227450
    },
    {
      "epoch": 1.4588577860585634,
      "grad_norm": 0.1520482748746872,
      "learning_rate": 2.0691214386197027e-06,
      "loss": 0.0015,
      "step": 227460
    },
    {
      "epoch": 1.4589219229523496,
      "grad_norm": 0.053171031177043915,
      "learning_rate": 2.068667994292257e-06,
      "loss": 0.0011,
      "step": 227470
    },
    {
      "epoch": 1.4589860598461355,
      "grad_norm": 0.03078148141503334,
      "learning_rate": 2.068214586696497e-06,
      "loss": 0.0005,
      "step": 227480
    },
    {
      "epoch": 1.4590501967399216,
      "grad_norm": 0.14004504680633545,
      "learning_rate": 2.0677612158381043e-06,
      "loss": 0.0011,
      "step": 227490
    },
    {
      "epoch": 1.4591143336337078,
      "grad_norm": 0.05386776477098465,
      "learning_rate": 2.067307881722757e-06,
      "loss": 0.0022,
      "step": 227500
    },
    {
      "epoch": 1.459178470527494,
      "grad_norm": 0.040485769510269165,
      "learning_rate": 2.06685458435614e-06,
      "loss": 0.0013,
      "step": 227510
    },
    {
      "epoch": 1.45924260742128,
      "grad_norm": 0.20594653487205505,
      "learning_rate": 2.066401323743929e-06,
      "loss": 0.0011,
      "step": 227520
    },
    {
      "epoch": 1.459306744315066,
      "grad_norm": 0.22772714495658875,
      "learning_rate": 2.0659480998918085e-06,
      "loss": 0.0007,
      "step": 227530
    },
    {
      "epoch": 1.4593708812088522,
      "grad_norm": 0.14145562052726746,
      "learning_rate": 2.0654949128054548e-06,
      "loss": 0.0007,
      "step": 227540
    },
    {
      "epoch": 1.4594350181026383,
      "grad_norm": 0.057922638952732086,
      "learning_rate": 2.0650417624905454e-06,
      "loss": 0.0006,
      "step": 227550
    },
    {
      "epoch": 1.4594991549964242,
      "grad_norm": 0.09675946831703186,
      "learning_rate": 2.064588648952762e-06,
      "loss": 0.0017,
      "step": 227560
    },
    {
      "epoch": 1.4595632918902104,
      "grad_norm": 0.06728710979223251,
      "learning_rate": 2.064135572197781e-06,
      "loss": 0.0011,
      "step": 227570
    },
    {
      "epoch": 1.4596274287839965,
      "grad_norm": 0.008515803143382072,
      "learning_rate": 2.0636825322312794e-06,
      "loss": 0.001,
      "step": 227580
    },
    {
      "epoch": 1.4596915656777827,
      "grad_norm": 0.06523746252059937,
      "learning_rate": 2.0632295290589326e-06,
      "loss": 0.001,
      "step": 227590
    },
    {
      "epoch": 1.4597557025715688,
      "grad_norm": 0.05311696231365204,
      "learning_rate": 2.0627765626864197e-06,
      "loss": 0.001,
      "step": 227600
    },
    {
      "epoch": 1.459819839465355,
      "grad_norm": 0.14546065032482147,
      "learning_rate": 2.062323633119417e-06,
      "loss": 0.0013,
      "step": 227610
    },
    {
      "epoch": 1.459883976359141,
      "grad_norm": 0.06579133868217468,
      "learning_rate": 2.061870740363598e-06,
      "loss": 0.0008,
      "step": 227620
    },
    {
      "epoch": 1.459948113252927,
      "grad_norm": 0.05342791602015495,
      "learning_rate": 2.061417884424636e-06,
      "loss": 0.0011,
      "step": 227630
    },
    {
      "epoch": 1.4600122501467132,
      "grad_norm": 0.058039650321006775,
      "learning_rate": 2.0609650653082108e-06,
      "loss": 0.0009,
      "step": 227640
    },
    {
      "epoch": 1.4600763870404991,
      "grad_norm": 0.08071907609701157,
      "learning_rate": 2.0605122830199936e-06,
      "loss": 0.0006,
      "step": 227650
    },
    {
      "epoch": 1.4601405239342853,
      "grad_norm": 0.12008611112833023,
      "learning_rate": 2.0600595375656584e-06,
      "loss": 0.0014,
      "step": 227660
    },
    {
      "epoch": 1.4602046608280714,
      "grad_norm": 0.04277434200048447,
      "learning_rate": 2.0596068289508766e-06,
      "loss": 0.0018,
      "step": 227670
    },
    {
      "epoch": 1.4602687977218576,
      "grad_norm": 0.18823428452014923,
      "learning_rate": 2.059154157181324e-06,
      "loss": 0.0011,
      "step": 227680
    },
    {
      "epoch": 1.4603329346156437,
      "grad_norm": 0.06276317685842514,
      "learning_rate": 2.058701522262672e-06,
      "loss": 0.0031,
      "step": 227690
    },
    {
      "epoch": 1.4603970715094297,
      "grad_norm": 0.03390146791934967,
      "learning_rate": 2.0582489242005897e-06,
      "loss": 0.0008,
      "step": 227700
    },
    {
      "epoch": 1.4604612084032158,
      "grad_norm": 0.18487323820590973,
      "learning_rate": 2.057796363000753e-06,
      "loss": 0.0018,
      "step": 227710
    },
    {
      "epoch": 1.460525345297002,
      "grad_norm": 0.14110855758190155,
      "learning_rate": 2.057343838668831e-06,
      "loss": 0.0007,
      "step": 227720
    },
    {
      "epoch": 1.4605894821907879,
      "grad_norm": 0.10188963264226913,
      "learning_rate": 2.0568913512104934e-06,
      "loss": 0.001,
      "step": 227730
    },
    {
      "epoch": 1.460653619084574,
      "grad_norm": 0.1425182968378067,
      "learning_rate": 2.056438900631409e-06,
      "loss": 0.0019,
      "step": 227740
    },
    {
      "epoch": 1.4607177559783602,
      "grad_norm": 0.0360860750079155,
      "learning_rate": 2.05598648693725e-06,
      "loss": 0.0008,
      "step": 227750
    },
    {
      "epoch": 1.4607818928721463,
      "grad_norm": 0.028131086379289627,
      "learning_rate": 2.055534110133686e-06,
      "loss": 0.0009,
      "step": 227760
    },
    {
      "epoch": 1.4608460297659325,
      "grad_norm": 0.24010440707206726,
      "learning_rate": 2.0550817702263824e-06,
      "loss": 0.0026,
      "step": 227770
    },
    {
      "epoch": 1.4609101666597186,
      "grad_norm": 0.14275701344013214,
      "learning_rate": 2.0546294672210075e-06,
      "loss": 0.0014,
      "step": 227780
    },
    {
      "epoch": 1.4609743035535045,
      "grad_norm": 0.06743670254945755,
      "learning_rate": 2.0541772011232327e-06,
      "loss": 0.0009,
      "step": 227790
    },
    {
      "epoch": 1.4610384404472907,
      "grad_norm": 0.052330587059259415,
      "learning_rate": 2.053724971938722e-06,
      "loss": 0.0015,
      "step": 227800
    },
    {
      "epoch": 1.4611025773410768,
      "grad_norm": 0.10786642879247665,
      "learning_rate": 2.053272779673144e-06,
      "loss": 0.0019,
      "step": 227810
    },
    {
      "epoch": 1.4611667142348628,
      "grad_norm": 0.0722866803407669,
      "learning_rate": 2.0528206243321618e-06,
      "loss": 0.0006,
      "step": 227820
    },
    {
      "epoch": 1.461230851128649,
      "grad_norm": 0.2144901007413864,
      "learning_rate": 2.0523685059214452e-06,
      "loss": 0.0012,
      "step": 227830
    },
    {
      "epoch": 1.461294988022435,
      "grad_norm": 0.11507140100002289,
      "learning_rate": 2.0519164244466583e-06,
      "loss": 0.0014,
      "step": 227840
    },
    {
      "epoch": 1.4613591249162212,
      "grad_norm": 0.052062615752220154,
      "learning_rate": 2.051464379913463e-06,
      "loss": 0.0007,
      "step": 227850
    },
    {
      "epoch": 1.4614232618100074,
      "grad_norm": 0.0925881564617157,
      "learning_rate": 2.051012372327529e-06,
      "loss": 0.0015,
      "step": 227860
    },
    {
      "epoch": 1.4614873987037933,
      "grad_norm": 0.07081978023052216,
      "learning_rate": 2.0505604016945176e-06,
      "loss": 0.0031,
      "step": 227870
    },
    {
      "epoch": 1.4615515355975794,
      "grad_norm": 0.14238618314266205,
      "learning_rate": 2.0501084680200915e-06,
      "loss": 0.0016,
      "step": 227880
    },
    {
      "epoch": 1.4616156724913656,
      "grad_norm": 0.07094322890043259,
      "learning_rate": 2.049656571309913e-06,
      "loss": 0.0012,
      "step": 227890
    },
    {
      "epoch": 1.4616798093851517,
      "grad_norm": 0.011230080388486385,
      "learning_rate": 2.0492047115696483e-06,
      "loss": 0.0014,
      "step": 227900
    },
    {
      "epoch": 1.4617439462789377,
      "grad_norm": 0.04832952097058296,
      "learning_rate": 2.0487528888049574e-06,
      "loss": 0.0011,
      "step": 227910
    },
    {
      "epoch": 1.4618080831727238,
      "grad_norm": 0.03298899158835411,
      "learning_rate": 2.048301103021502e-06,
      "loss": 0.0027,
      "step": 227920
    },
    {
      "epoch": 1.46187222006651,
      "grad_norm": 0.1945764720439911,
      "learning_rate": 2.0478493542249417e-06,
      "loss": 0.0019,
      "step": 227930
    },
    {
      "epoch": 1.4619363569602961,
      "grad_norm": 0.02438526414334774,
      "learning_rate": 2.04739764242094e-06,
      "loss": 0.0006,
      "step": 227940
    },
    {
      "epoch": 1.4620004938540823,
      "grad_norm": 0.2396112084388733,
      "learning_rate": 2.0469459676151566e-06,
      "loss": 0.0012,
      "step": 227950
    },
    {
      "epoch": 1.4620646307478682,
      "grad_norm": 0.06391119956970215,
      "learning_rate": 2.046494329813249e-06,
      "loss": 0.0009,
      "step": 227960
    },
    {
      "epoch": 1.4621287676416543,
      "grad_norm": 0.2302028387784958,
      "learning_rate": 2.04604272902088e-06,
      "loss": 0.002,
      "step": 227970
    },
    {
      "epoch": 1.4621929045354405,
      "grad_norm": 0.02100854367017746,
      "learning_rate": 2.0455911652437066e-06,
      "loss": 0.0114,
      "step": 227980
    },
    {
      "epoch": 1.4622570414292264,
      "grad_norm": 0.007054073270410299,
      "learning_rate": 2.045139638487387e-06,
      "loss": 0.0012,
      "step": 227990
    },
    {
      "epoch": 1.4623211783230126,
      "grad_norm": 0.15014074742794037,
      "learning_rate": 2.0446881487575785e-06,
      "loss": 0.0014,
      "step": 228000
    },
    {
      "epoch": 1.4623853152167987,
      "grad_norm": 0.1818699687719345,
      "learning_rate": 2.0442366960599415e-06,
      "loss": 0.0018,
      "step": 228010
    },
    {
      "epoch": 1.4624494521105849,
      "grad_norm": 0.16917431354522705,
      "learning_rate": 2.0437852804001305e-06,
      "loss": 0.0021,
      "step": 228020
    },
    {
      "epoch": 1.462513589004371,
      "grad_norm": 0.18695397675037384,
      "learning_rate": 2.0433339017838032e-06,
      "loss": 0.001,
      "step": 228030
    },
    {
      "epoch": 1.4625777258981572,
      "grad_norm": 0.10904522985219955,
      "learning_rate": 2.0428825602166132e-06,
      "loss": 0.0009,
      "step": 228040
    },
    {
      "epoch": 1.462641862791943,
      "grad_norm": 0.007502240594476461,
      "learning_rate": 2.04243125570422e-06,
      "loss": 0.0007,
      "step": 228050
    },
    {
      "epoch": 1.4627059996857292,
      "grad_norm": 0.12464974075555801,
      "learning_rate": 2.041979988252277e-06,
      "loss": 0.0014,
      "step": 228060
    },
    {
      "epoch": 1.4627701365795154,
      "grad_norm": 0.12314174324274063,
      "learning_rate": 2.0415287578664383e-06,
      "loss": 0.0014,
      "step": 228070
    },
    {
      "epoch": 1.4628342734733013,
      "grad_norm": 0.006173981353640556,
      "learning_rate": 2.041077564552357e-06,
      "loss": 0.0008,
      "step": 228080
    },
    {
      "epoch": 1.4628984103670875,
      "grad_norm": 0.048853591084480286,
      "learning_rate": 2.0406264083156905e-06,
      "loss": 0.0007,
      "step": 228090
    },
    {
      "epoch": 1.4629625472608736,
      "grad_norm": 0.04714033752679825,
      "learning_rate": 2.0401752891620908e-06,
      "loss": 0.0011,
      "step": 228100
    },
    {
      "epoch": 1.4630266841546598,
      "grad_norm": 0.01387617364525795,
      "learning_rate": 2.039724207097207e-06,
      "loss": 0.0008,
      "step": 228110
    },
    {
      "epoch": 1.463090821048446,
      "grad_norm": 0.03569384664297104,
      "learning_rate": 2.0392731621266974e-06,
      "loss": 0.001,
      "step": 228120
    },
    {
      "epoch": 1.4631549579422318,
      "grad_norm": 0.010431944392621517,
      "learning_rate": 2.038822154256211e-06,
      "loss": 0.0022,
      "step": 228130
    },
    {
      "epoch": 1.463219094836018,
      "grad_norm": 0.08307234197854996,
      "learning_rate": 2.038371183491399e-06,
      "loss": 0.0019,
      "step": 228140
    },
    {
      "epoch": 1.4632832317298041,
      "grad_norm": 0.16739168763160706,
      "learning_rate": 2.037920249837911e-06,
      "loss": 0.0012,
      "step": 228150
    },
    {
      "epoch": 1.4633473686235903,
      "grad_norm": 0.13285072147846222,
      "learning_rate": 2.0374693533014007e-06,
      "loss": 0.0008,
      "step": 228160
    },
    {
      "epoch": 1.4634115055173762,
      "grad_norm": 0.03239947184920311,
      "learning_rate": 2.037018493887517e-06,
      "loss": 0.0014,
      "step": 228170
    },
    {
      "epoch": 1.4634756424111623,
      "grad_norm": 0.03774484246969223,
      "learning_rate": 2.0365676716019095e-06,
      "loss": 0.002,
      "step": 228180
    },
    {
      "epoch": 1.4635397793049485,
      "grad_norm": 0.007905969396233559,
      "learning_rate": 2.0361168864502246e-06,
      "loss": 0.0011,
      "step": 228190
    },
    {
      "epoch": 1.4636039161987346,
      "grad_norm": 0.12281139940023422,
      "learning_rate": 2.0356661384381155e-06,
      "loss": 0.0012,
      "step": 228200
    },
    {
      "epoch": 1.4636680530925208,
      "grad_norm": 0.0835944414138794,
      "learning_rate": 2.035215427571228e-06,
      "loss": 0.0013,
      "step": 228210
    },
    {
      "epoch": 1.4637321899863067,
      "grad_norm": 0.0011360858334228396,
      "learning_rate": 2.0347647538552105e-06,
      "loss": 0.0007,
      "step": 228220
    },
    {
      "epoch": 1.4637963268800929,
      "grad_norm": 0.1155383288860321,
      "learning_rate": 2.0343141172957077e-06,
      "loss": 0.0053,
      "step": 228230
    },
    {
      "epoch": 1.463860463773879,
      "grad_norm": 0.16193082928657532,
      "learning_rate": 2.0338635178983706e-06,
      "loss": 0.0029,
      "step": 228240
    },
    {
      "epoch": 1.463924600667665,
      "grad_norm": 0.019801519811153412,
      "learning_rate": 2.033412955668844e-06,
      "loss": 0.0014,
      "step": 228250
    },
    {
      "epoch": 1.463988737561451,
      "grad_norm": 0.015870848670601845,
      "learning_rate": 2.0329624306127705e-06,
      "loss": 0.0005,
      "step": 228260
    },
    {
      "epoch": 1.4640528744552372,
      "grad_norm": 0.07326389104127884,
      "learning_rate": 2.032511942735798e-06,
      "loss": 0.0014,
      "step": 228270
    },
    {
      "epoch": 1.4641170113490234,
      "grad_norm": 0.014281432144343853,
      "learning_rate": 2.0320614920435744e-06,
      "loss": 0.0014,
      "step": 228280
    },
    {
      "epoch": 1.4641811482428095,
      "grad_norm": 0.0774741917848587,
      "learning_rate": 2.0316110785417414e-06,
      "loss": 0.0007,
      "step": 228290
    },
    {
      "epoch": 1.4642452851365957,
      "grad_norm": 0.014599175192415714,
      "learning_rate": 2.0311607022359424e-06,
      "loss": 0.002,
      "step": 228300
    },
    {
      "epoch": 1.4643094220303816,
      "grad_norm": 0.09474321454763412,
      "learning_rate": 2.0307103631318202e-06,
      "loss": 0.0013,
      "step": 228310
    },
    {
      "epoch": 1.4643735589241678,
      "grad_norm": 0.1536872386932373,
      "learning_rate": 2.0302600612350217e-06,
      "loss": 0.0011,
      "step": 228320
    },
    {
      "epoch": 1.464437695817954,
      "grad_norm": 0.09565769135951996,
      "learning_rate": 2.029809796551186e-06,
      "loss": 0.0012,
      "step": 228330
    },
    {
      "epoch": 1.4645018327117398,
      "grad_norm": 0.0661081001162529,
      "learning_rate": 2.0293595690859567e-06,
      "loss": 0.0008,
      "step": 228340
    },
    {
      "epoch": 1.464565969605526,
      "grad_norm": 0.02482161857187748,
      "learning_rate": 2.028909378844973e-06,
      "loss": 0.0009,
      "step": 228350
    },
    {
      "epoch": 1.4646301064993121,
      "grad_norm": 0.11742063611745834,
      "learning_rate": 2.028459225833881e-06,
      "loss": 0.0007,
      "step": 228360
    },
    {
      "epoch": 1.4646942433930983,
      "grad_norm": 0.05392173305153847,
      "learning_rate": 2.0280091100583175e-06,
      "loss": 0.0014,
      "step": 228370
    },
    {
      "epoch": 1.4647583802868844,
      "grad_norm": 0.014185560867190361,
      "learning_rate": 2.027559031523923e-06,
      "loss": 0.0015,
      "step": 228380
    },
    {
      "epoch": 1.4648225171806704,
      "grad_norm": 0.013879602774977684,
      "learning_rate": 2.0271089902363394e-06,
      "loss": 0.0008,
      "step": 228390
    },
    {
      "epoch": 1.4648866540744565,
      "grad_norm": 0.011047734878957272,
      "learning_rate": 2.0266589862012047e-06,
      "loss": 0.003,
      "step": 228400
    },
    {
      "epoch": 1.4649507909682427,
      "grad_norm": 0.024848774075508118,
      "learning_rate": 2.0262090194241585e-06,
      "loss": 0.0006,
      "step": 228410
    },
    {
      "epoch": 1.4650149278620286,
      "grad_norm": 0.023540036752820015,
      "learning_rate": 2.0257590899108364e-06,
      "loss": 0.0013,
      "step": 228420
    },
    {
      "epoch": 1.4650790647558147,
      "grad_norm": 0.04605693742632866,
      "learning_rate": 2.025309197666881e-06,
      "loss": 0.0022,
      "step": 228430
    },
    {
      "epoch": 1.4651432016496009,
      "grad_norm": 0.14158214628696442,
      "learning_rate": 2.0248593426979268e-06,
      "loss": 0.0009,
      "step": 228440
    },
    {
      "epoch": 1.465207338543387,
      "grad_norm": 0.11355796456336975,
      "learning_rate": 2.0244095250096115e-06,
      "loss": 0.0013,
      "step": 228450
    },
    {
      "epoch": 1.4652714754371732,
      "grad_norm": 0.08286843448877335,
      "learning_rate": 2.0239597446075692e-06,
      "loss": 0.0004,
      "step": 228460
    },
    {
      "epoch": 1.4653356123309593,
      "grad_norm": 0.08982168883085251,
      "learning_rate": 2.02351000149744e-06,
      "loss": 0.0006,
      "step": 228470
    },
    {
      "epoch": 1.4653997492247453,
      "grad_norm": 0.06646949797868729,
      "learning_rate": 2.023060295684859e-06,
      "loss": 0.0016,
      "step": 228480
    },
    {
      "epoch": 1.4654638861185314,
      "grad_norm": 0.016652148216962814,
      "learning_rate": 2.0226106271754587e-06,
      "loss": 0.0012,
      "step": 228490
    },
    {
      "epoch": 1.4655280230123175,
      "grad_norm": 0.05408656597137451,
      "learning_rate": 2.022160995974874e-06,
      "loss": 0.0013,
      "step": 228500
    },
    {
      "epoch": 1.4655921599061035,
      "grad_norm": 0.023705298081040382,
      "learning_rate": 2.0217114020887416e-06,
      "loss": 0.0013,
      "step": 228510
    },
    {
      "epoch": 1.4656562967998896,
      "grad_norm": 0.06567654013633728,
      "learning_rate": 2.0212618455226947e-06,
      "loss": 0.0012,
      "step": 228520
    },
    {
      "epoch": 1.4657204336936758,
      "grad_norm": 0.12240105867385864,
      "learning_rate": 2.0208123262823633e-06,
      "loss": 0.0019,
      "step": 228530
    },
    {
      "epoch": 1.465784570587462,
      "grad_norm": 0.0822448655962944,
      "learning_rate": 2.020362844373385e-06,
      "loss": 0.0011,
      "step": 228540
    },
    {
      "epoch": 1.465848707481248,
      "grad_norm": 0.0353735014796257,
      "learning_rate": 2.0199133998013892e-06,
      "loss": 0.0013,
      "step": 228550
    },
    {
      "epoch": 1.465912844375034,
      "grad_norm": 0.050326962023973465,
      "learning_rate": 2.0194639925720083e-06,
      "loss": 0.0007,
      "step": 228560
    },
    {
      "epoch": 1.4659769812688201,
      "grad_norm": 0.06709080189466476,
      "learning_rate": 2.0190146226908726e-06,
      "loss": 0.001,
      "step": 228570
    },
    {
      "epoch": 1.4660411181626063,
      "grad_norm": 0.23265334963798523,
      "learning_rate": 2.0185652901636147e-06,
      "loss": 0.0018,
      "step": 228580
    },
    {
      "epoch": 1.4661052550563924,
      "grad_norm": 0.0635458454489708,
      "learning_rate": 2.0181159949958655e-06,
      "loss": 0.0012,
      "step": 228590
    },
    {
      "epoch": 1.4661693919501784,
      "grad_norm": 0.06184995546936989,
      "learning_rate": 2.017666737193254e-06,
      "loss": 0.0009,
      "step": 228600
    },
    {
      "epoch": 1.4662335288439645,
      "grad_norm": 0.04848029091954231,
      "learning_rate": 2.0172175167614076e-06,
      "loss": 0.0005,
      "step": 228610
    },
    {
      "epoch": 1.4662976657377507,
      "grad_norm": 0.03075077198445797,
      "learning_rate": 2.0167683337059597e-06,
      "loss": 0.0012,
      "step": 228620
    },
    {
      "epoch": 1.4663618026315368,
      "grad_norm": 0.09495376795530319,
      "learning_rate": 2.016319188032536e-06,
      "loss": 0.0014,
      "step": 228630
    },
    {
      "epoch": 1.466425939525323,
      "grad_norm": 0.015723099932074547,
      "learning_rate": 2.0158700797467663e-06,
      "loss": 0.0013,
      "step": 228640
    },
    {
      "epoch": 1.466490076419109,
      "grad_norm": 0.017615314573049545,
      "learning_rate": 2.015421008854274e-06,
      "loss": 0.0009,
      "step": 228650
    },
    {
      "epoch": 1.466554213312895,
      "grad_norm": 0.008050485514104366,
      "learning_rate": 2.0149719753606927e-06,
      "loss": 0.0021,
      "step": 228660
    },
    {
      "epoch": 1.4666183502066812,
      "grad_norm": 0.0542878732085228,
      "learning_rate": 2.014522979271645e-06,
      "loss": 0.0009,
      "step": 228670
    },
    {
      "epoch": 1.4666824871004671,
      "grad_norm": 0.057286377996206284,
      "learning_rate": 2.0140740205927567e-06,
      "loss": 0.0009,
      "step": 228680
    },
    {
      "epoch": 1.4667466239942533,
      "grad_norm": 0.09914997220039368,
      "learning_rate": 2.013625099329657e-06,
      "loss": 0.0021,
      "step": 228690
    },
    {
      "epoch": 1.4668107608880394,
      "grad_norm": 0.014204105362296104,
      "learning_rate": 2.0131762154879688e-06,
      "loss": 0.0006,
      "step": 228700
    },
    {
      "epoch": 1.4668748977818256,
      "grad_norm": 0.10019271820783615,
      "learning_rate": 2.012727369073317e-06,
      "loss": 0.0016,
      "step": 228710
    },
    {
      "epoch": 1.4669390346756117,
      "grad_norm": 0.12342901527881622,
      "learning_rate": 2.012278560091324e-06,
      "loss": 0.0018,
      "step": 228720
    },
    {
      "epoch": 1.4670031715693979,
      "grad_norm": 0.031092116609215736,
      "learning_rate": 2.011829788547619e-06,
      "loss": 0.0011,
      "step": 228730
    },
    {
      "epoch": 1.4670673084631838,
      "grad_norm": 0.038071438670158386,
      "learning_rate": 2.0113810544478213e-06,
      "loss": 0.0006,
      "step": 228740
    },
    {
      "epoch": 1.46713144535697,
      "grad_norm": 0.02734973281621933,
      "learning_rate": 2.0109323577975553e-06,
      "loss": 0.0012,
      "step": 228750
    },
    {
      "epoch": 1.467195582250756,
      "grad_norm": 0.1813708245754242,
      "learning_rate": 2.0104836986024416e-06,
      "loss": 0.0014,
      "step": 228760
    },
    {
      "epoch": 1.467259719144542,
      "grad_norm": 0.03087799809873104,
      "learning_rate": 2.0100350768681054e-06,
      "loss": 0.0015,
      "step": 228770
    },
    {
      "epoch": 1.4673238560383282,
      "grad_norm": 0.04848742485046387,
      "learning_rate": 2.0095864926001664e-06,
      "loss": 0.0018,
      "step": 228780
    },
    {
      "epoch": 1.4673879929321143,
      "grad_norm": 0.10303984582424164,
      "learning_rate": 2.0091379458042458e-06,
      "loss": 0.001,
      "step": 228790
    },
    {
      "epoch": 1.4674521298259005,
      "grad_norm": 0.12709379196166992,
      "learning_rate": 2.008689436485963e-06,
      "loss": 0.0011,
      "step": 228800
    },
    {
      "epoch": 1.4675162667196866,
      "grad_norm": 0.0049931686371564865,
      "learning_rate": 2.0082409646509406e-06,
      "loss": 0.0004,
      "step": 228810
    },
    {
      "epoch": 1.4675804036134725,
      "grad_norm": 0.09220277518033981,
      "learning_rate": 2.0077925303047975e-06,
      "loss": 0.0014,
      "step": 228820
    },
    {
      "epoch": 1.4676445405072587,
      "grad_norm": 0.049771569669246674,
      "learning_rate": 2.007344133453151e-06,
      "loss": 0.0007,
      "step": 228830
    },
    {
      "epoch": 1.4677086774010448,
      "grad_norm": 0.17239652574062347,
      "learning_rate": 2.006895774101622e-06,
      "loss": 0.0007,
      "step": 228840
    },
    {
      "epoch": 1.4677728142948308,
      "grad_norm": 0.044996198266744614,
      "learning_rate": 2.0064474522558287e-06,
      "loss": 0.0017,
      "step": 228850
    },
    {
      "epoch": 1.467836951188617,
      "grad_norm": 0.016608484089374542,
      "learning_rate": 2.0059991679213884e-06,
      "loss": 0.001,
      "step": 228860
    },
    {
      "epoch": 1.467901088082403,
      "grad_norm": 0.05175834149122238,
      "learning_rate": 2.0055509211039166e-06,
      "loss": 0.0007,
      "step": 228870
    },
    {
      "epoch": 1.4679652249761892,
      "grad_norm": 0.1386476755142212,
      "learning_rate": 2.005102711809033e-06,
      "loss": 0.0009,
      "step": 228880
    },
    {
      "epoch": 1.4680293618699753,
      "grad_norm": 0.1274915337562561,
      "learning_rate": 2.004654540042354e-06,
      "loss": 0.0015,
      "step": 228890
    },
    {
      "epoch": 1.4680934987637615,
      "grad_norm": 0.18073634803295135,
      "learning_rate": 2.0042064058094934e-06,
      "loss": 0.0017,
      "step": 228900
    },
    {
      "epoch": 1.4681576356575474,
      "grad_norm": 0.053620342165231705,
      "learning_rate": 2.003758309116066e-06,
      "loss": 0.0017,
      "step": 228910
    },
    {
      "epoch": 1.4682217725513336,
      "grad_norm": 0.03877738118171692,
      "learning_rate": 2.0033102499676904e-06,
      "loss": 0.0008,
      "step": 228920
    },
    {
      "epoch": 1.4682859094451197,
      "grad_norm": 0.05406402796506882,
      "learning_rate": 2.002862228369979e-06,
      "loss": 0.001,
      "step": 228930
    },
    {
      "epoch": 1.4683500463389056,
      "grad_norm": 0.05574464052915573,
      "learning_rate": 2.0024142443285455e-06,
      "loss": 0.0017,
      "step": 228940
    },
    {
      "epoch": 1.4684141832326918,
      "grad_norm": 0.1522226333618164,
      "learning_rate": 2.001966297849003e-06,
      "loss": 0.001,
      "step": 228950
    },
    {
      "epoch": 1.468478320126478,
      "grad_norm": 0.08424288779497147,
      "learning_rate": 2.001518388936966e-06,
      "loss": 0.001,
      "step": 228960
    },
    {
      "epoch": 1.468542457020264,
      "grad_norm": 0.0023943937849253416,
      "learning_rate": 2.001070517598048e-06,
      "loss": 0.0005,
      "step": 228970
    },
    {
      "epoch": 1.4686065939140502,
      "grad_norm": 0.12704569101333618,
      "learning_rate": 2.000622683837857e-06,
      "loss": 0.0009,
      "step": 228980
    },
    {
      "epoch": 1.4686707308078362,
      "grad_norm": 0.10262220352888107,
      "learning_rate": 2.000174887662009e-06,
      "loss": 0.0008,
      "step": 228990
    },
    {
      "epoch": 1.4687348677016223,
      "grad_norm": 0.041612930595874786,
      "learning_rate": 1.9997271290761137e-06,
      "loss": 0.0017,
      "step": 229000
    },
    {
      "epoch": 1.4687990045954085,
      "grad_norm": 0.10134538263082504,
      "learning_rate": 1.999279408085782e-06,
      "loss": 0.0014,
      "step": 229010
    },
    {
      "epoch": 1.4688631414891946,
      "grad_norm": 0.0696941614151001,
      "learning_rate": 1.998831724696622e-06,
      "loss": 0.002,
      "step": 229020
    },
    {
      "epoch": 1.4689272783829805,
      "grad_norm": 0.22982709109783173,
      "learning_rate": 1.9983840789142474e-06,
      "loss": 0.003,
      "step": 229030
    },
    {
      "epoch": 1.4689914152767667,
      "grad_norm": 0.019180903211236,
      "learning_rate": 1.997936470744263e-06,
      "loss": 0.001,
      "step": 229040
    },
    {
      "epoch": 1.4690555521705528,
      "grad_norm": 0.06342509388923645,
      "learning_rate": 1.997488900192283e-06,
      "loss": 0.0012,
      "step": 229050
    },
    {
      "epoch": 1.469119689064339,
      "grad_norm": 0.07248317450284958,
      "learning_rate": 1.997041367263913e-06,
      "loss": 0.0009,
      "step": 229060
    },
    {
      "epoch": 1.4691838259581251,
      "grad_norm": 0.06499683111906052,
      "learning_rate": 1.9965938719647582e-06,
      "loss": 0.0006,
      "step": 229070
    },
    {
      "epoch": 1.469247962851911,
      "grad_norm": 0.15315909683704376,
      "learning_rate": 1.9961464143004306e-06,
      "loss": 0.0008,
      "step": 229080
    },
    {
      "epoch": 1.4693120997456972,
      "grad_norm": 0.09500911831855774,
      "learning_rate": 1.9956989942765353e-06,
      "loss": 0.0011,
      "step": 229090
    },
    {
      "epoch": 1.4693762366394834,
      "grad_norm": 0.03162289038300514,
      "learning_rate": 1.9952516118986775e-06,
      "loss": 0.0018,
      "step": 229100
    },
    {
      "epoch": 1.4694403735332693,
      "grad_norm": 0.013298217207193375,
      "learning_rate": 1.9948042671724655e-06,
      "loss": 0.0014,
      "step": 229110
    },
    {
      "epoch": 1.4695045104270554,
      "grad_norm": 0.0648544505238533,
      "learning_rate": 1.9943569601035045e-06,
      "loss": 0.0017,
      "step": 229120
    },
    {
      "epoch": 1.4695686473208416,
      "grad_norm": 0.35140907764434814,
      "learning_rate": 1.9939096906973985e-06,
      "loss": 0.0025,
      "step": 229130
    },
    {
      "epoch": 1.4696327842146277,
      "grad_norm": 0.03840072825551033,
      "learning_rate": 1.9934624589597506e-06,
      "loss": 0.0005,
      "step": 229140
    },
    {
      "epoch": 1.4696969211084139,
      "grad_norm": 0.016085276380181313,
      "learning_rate": 1.9930152648961686e-06,
      "loss": 0.0011,
      "step": 229150
    },
    {
      "epoch": 1.4697610580022,
      "grad_norm": 0.0642709732055664,
      "learning_rate": 1.992568108512255e-06,
      "loss": 0.0016,
      "step": 229160
    },
    {
      "epoch": 1.469825194895986,
      "grad_norm": 0.12338618189096451,
      "learning_rate": 1.9921209898136117e-06,
      "loss": 0.0011,
      "step": 229170
    },
    {
      "epoch": 1.469889331789772,
      "grad_norm": 0.07437125593423843,
      "learning_rate": 1.9916739088058406e-06,
      "loss": 0.0021,
      "step": 229180
    },
    {
      "epoch": 1.4699534686835583,
      "grad_norm": 0.02611471898853779,
      "learning_rate": 1.991226865494547e-06,
      "loss": 0.0013,
      "step": 229190
    },
    {
      "epoch": 1.4700176055773442,
      "grad_norm": 0.06297191232442856,
      "learning_rate": 1.990779859885332e-06,
      "loss": 0.0018,
      "step": 229200
    },
    {
      "epoch": 1.4700817424711303,
      "grad_norm": 0.08756767213344574,
      "learning_rate": 1.990332891983795e-06,
      "loss": 0.0012,
      "step": 229210
    },
    {
      "epoch": 1.4701458793649165,
      "grad_norm": 0.03923594579100609,
      "learning_rate": 1.9898859617955363e-06,
      "loss": 0.0007,
      "step": 229220
    },
    {
      "epoch": 1.4702100162587026,
      "grad_norm": 0.04241190478205681,
      "learning_rate": 1.9894390693261595e-06,
      "loss": 0.0012,
      "step": 229230
    },
    {
      "epoch": 1.4702741531524888,
      "grad_norm": 0.18856562674045563,
      "learning_rate": 1.988992214581263e-06,
      "loss": 0.0012,
      "step": 229240
    },
    {
      "epoch": 1.4703382900462747,
      "grad_norm": 0.07870613783597946,
      "learning_rate": 1.9885453975664443e-06,
      "loss": 0.0008,
      "step": 229250
    },
    {
      "epoch": 1.4704024269400608,
      "grad_norm": 0.04163774102926254,
      "learning_rate": 1.988098618287306e-06,
      "loss": 0.0013,
      "step": 229260
    },
    {
      "epoch": 1.470466563833847,
      "grad_norm": 0.07428430020809174,
      "learning_rate": 1.987651876749444e-06,
      "loss": 0.0011,
      "step": 229270
    },
    {
      "epoch": 1.470530700727633,
      "grad_norm": 0.10029874742031097,
      "learning_rate": 1.9872051729584567e-06,
      "loss": 0.0024,
      "step": 229280
    },
    {
      "epoch": 1.470594837621419,
      "grad_norm": 0.052629999816417694,
      "learning_rate": 1.9867585069199407e-06,
      "loss": 0.0011,
      "step": 229290
    },
    {
      "epoch": 1.4706589745152052,
      "grad_norm": 0.060825642198324203,
      "learning_rate": 1.986311878639495e-06,
      "loss": 0.001,
      "step": 229300
    },
    {
      "epoch": 1.4707231114089914,
      "grad_norm": 0.0829131156206131,
      "learning_rate": 1.9858652881227162e-06,
      "loss": 0.0011,
      "step": 229310
    },
    {
      "epoch": 1.4707872483027775,
      "grad_norm": 0.03715585544705391,
      "learning_rate": 1.9854187353751996e-06,
      "loss": 0.0009,
      "step": 229320
    },
    {
      "epoch": 1.4708513851965637,
      "grad_norm": 0.20530331134796143,
      "learning_rate": 1.9849722204025385e-06,
      "loss": 0.0027,
      "step": 229330
    },
    {
      "epoch": 1.4709155220903496,
      "grad_norm": 0.13014550507068634,
      "learning_rate": 1.984525743210332e-06,
      "loss": 0.002,
      "step": 229340
    },
    {
      "epoch": 1.4709796589841357,
      "grad_norm": 0.16381634771823883,
      "learning_rate": 1.9840793038041733e-06,
      "loss": 0.0008,
      "step": 229350
    },
    {
      "epoch": 1.471043795877922,
      "grad_norm": 0.05737442895770073,
      "learning_rate": 1.9836329021896567e-06,
      "loss": 0.0005,
      "step": 229360
    },
    {
      "epoch": 1.4711079327717078,
      "grad_norm": 0.06670114398002625,
      "learning_rate": 1.983186538372373e-06,
      "loss": 0.0022,
      "step": 229370
    },
    {
      "epoch": 1.471172069665494,
      "grad_norm": 0.18281473219394684,
      "learning_rate": 1.9827402123579204e-06,
      "loss": 0.0019,
      "step": 229380
    },
    {
      "epoch": 1.4712362065592801,
      "grad_norm": 0.06288686394691467,
      "learning_rate": 1.982293924151889e-06,
      "loss": 0.0012,
      "step": 229390
    },
    {
      "epoch": 1.4713003434530663,
      "grad_norm": 0.12208051234483719,
      "learning_rate": 1.98184767375987e-06,
      "loss": 0.0008,
      "step": 229400
    },
    {
      "epoch": 1.4713644803468524,
      "grad_norm": 0.06759564578533173,
      "learning_rate": 1.981401461187459e-06,
      "loss": 0.0006,
      "step": 229410
    },
    {
      "epoch": 1.4714286172406383,
      "grad_norm": 0.10475599765777588,
      "learning_rate": 1.9809552864402437e-06,
      "loss": 0.0014,
      "step": 229420
    },
    {
      "epoch": 1.4714927541344245,
      "grad_norm": 0.1359255462884903,
      "learning_rate": 1.980509149523817e-06,
      "loss": 0.0011,
      "step": 229430
    },
    {
      "epoch": 1.4715568910282106,
      "grad_norm": 0.07360190153121948,
      "learning_rate": 1.9800630504437672e-06,
      "loss": 0.0014,
      "step": 229440
    },
    {
      "epoch": 1.4716210279219968,
      "grad_norm": 0.17912261188030243,
      "learning_rate": 1.9796169892056873e-06,
      "loss": 0.0026,
      "step": 229450
    },
    {
      "epoch": 1.4716851648157827,
      "grad_norm": 0.11364316940307617,
      "learning_rate": 1.9791709658151647e-06,
      "loss": 0.0019,
      "step": 229460
    },
    {
      "epoch": 1.4717493017095689,
      "grad_norm": 0.008240544237196445,
      "learning_rate": 1.9787249802777897e-06,
      "loss": 0.0008,
      "step": 229470
    },
    {
      "epoch": 1.471813438603355,
      "grad_norm": 0.10144824534654617,
      "learning_rate": 1.9782790325991474e-06,
      "loss": 0.0013,
      "step": 229480
    },
    {
      "epoch": 1.4718775754971412,
      "grad_norm": 0.11618056893348694,
      "learning_rate": 1.9778331227848307e-06,
      "loss": 0.0011,
      "step": 229490
    },
    {
      "epoch": 1.4719417123909273,
      "grad_norm": 0.06600002944469452,
      "learning_rate": 1.9773872508404244e-06,
      "loss": 0.0049,
      "step": 229500
    },
    {
      "epoch": 1.4720058492847132,
      "grad_norm": 0.027490144595503807,
      "learning_rate": 1.976941416771516e-06,
      "loss": 0.0007,
      "step": 229510
    },
    {
      "epoch": 1.4720699861784994,
      "grad_norm": 0.07673504948616028,
      "learning_rate": 1.976495620583691e-06,
      "loss": 0.0033,
      "step": 229520
    },
    {
      "epoch": 1.4721341230722855,
      "grad_norm": 0.024866528809070587,
      "learning_rate": 1.9760498622825385e-06,
      "loss": 0.0008,
      "step": 229530
    },
    {
      "epoch": 1.4721982599660715,
      "grad_norm": 0.14301668107509613,
      "learning_rate": 1.975604141873642e-06,
      "loss": 0.0024,
      "step": 229540
    },
    {
      "epoch": 1.4722623968598576,
      "grad_norm": 0.3289594054222107,
      "learning_rate": 1.9751584593625855e-06,
      "loss": 0.003,
      "step": 229550
    },
    {
      "epoch": 1.4723265337536438,
      "grad_norm": 0.20589904487133026,
      "learning_rate": 1.974712814754957e-06,
      "loss": 0.0012,
      "step": 229560
    },
    {
      "epoch": 1.47239067064743,
      "grad_norm": 0.05686769261956215,
      "learning_rate": 1.9742672080563395e-06,
      "loss": 0.0009,
      "step": 229570
    },
    {
      "epoch": 1.472454807541216,
      "grad_norm": 0.07623685151338577,
      "learning_rate": 1.973821639272316e-06,
      "loss": 0.0013,
      "step": 229580
    },
    {
      "epoch": 1.4725189444350022,
      "grad_norm": 0.09793905168771744,
      "learning_rate": 1.9733761084084687e-06,
      "loss": 0.0012,
      "step": 229590
    },
    {
      "epoch": 1.4725830813287881,
      "grad_norm": 0.03943793103098869,
      "learning_rate": 1.972930615470383e-06,
      "loss": 0.0011,
      "step": 229600
    },
    {
      "epoch": 1.4726472182225743,
      "grad_norm": 0.0713537186384201,
      "learning_rate": 1.97248516046364e-06,
      "loss": 0.0006,
      "step": 229610
    },
    {
      "epoch": 1.4727113551163604,
      "grad_norm": 0.07061032205820084,
      "learning_rate": 1.9720397433938227e-06,
      "loss": 0.0012,
      "step": 229620
    },
    {
      "epoch": 1.4727754920101463,
      "grad_norm": 0.06999431550502777,
      "learning_rate": 1.9715943642665094e-06,
      "loss": 0.0012,
      "step": 229630
    },
    {
      "epoch": 1.4728396289039325,
      "grad_norm": 0.030215317383408546,
      "learning_rate": 1.9711490230872843e-06,
      "loss": 0.0007,
      "step": 229640
    },
    {
      "epoch": 1.4729037657977186,
      "grad_norm": 0.07707786560058594,
      "learning_rate": 1.970703719861727e-06,
      "loss": 0.0015,
      "step": 229650
    },
    {
      "epoch": 1.4729679026915048,
      "grad_norm": 0.04843216761946678,
      "learning_rate": 1.970258454595415e-06,
      "loss": 0.0014,
      "step": 229660
    },
    {
      "epoch": 1.473032039585291,
      "grad_norm": 0.12600107491016388,
      "learning_rate": 1.969813227293932e-06,
      "loss": 0.0012,
      "step": 229670
    },
    {
      "epoch": 1.4730961764790769,
      "grad_norm": 0.041636981070041656,
      "learning_rate": 1.9693680379628543e-06,
      "loss": 0.0023,
      "step": 229680
    },
    {
      "epoch": 1.473160313372863,
      "grad_norm": 0.11662957817316055,
      "learning_rate": 1.968922886607762e-06,
      "loss": 0.0025,
      "step": 229690
    },
    {
      "epoch": 1.4732244502666492,
      "grad_norm": 0.09565620124340057,
      "learning_rate": 1.9684777732342296e-06,
      "loss": 0.0012,
      "step": 229700
    },
    {
      "epoch": 1.4732885871604353,
      "grad_norm": 0.005680976435542107,
      "learning_rate": 1.9680326978478393e-06,
      "loss": 0.0009,
      "step": 229710
    },
    {
      "epoch": 1.4733527240542212,
      "grad_norm": 0.01353106927126646,
      "learning_rate": 1.9675876604541665e-06,
      "loss": 0.001,
      "step": 229720
    },
    {
      "epoch": 1.4734168609480074,
      "grad_norm": 0.0855177491903305,
      "learning_rate": 1.9671426610587873e-06,
      "loss": 0.0017,
      "step": 229730
    },
    {
      "epoch": 1.4734809978417935,
      "grad_norm": 0.0511997751891613,
      "learning_rate": 1.9666976996672764e-06,
      "loss": 0.0014,
      "step": 229740
    },
    {
      "epoch": 1.4735451347355797,
      "grad_norm": 0.053706057369709015,
      "learning_rate": 1.966252776285213e-06,
      "loss": 0.0014,
      "step": 229750
    },
    {
      "epoch": 1.4736092716293658,
      "grad_norm": 0.016953039914369583,
      "learning_rate": 1.9658078909181705e-06,
      "loss": 0.0022,
      "step": 229760
    },
    {
      "epoch": 1.4736734085231518,
      "grad_norm": 0.030180564150214195,
      "learning_rate": 1.9653630435717237e-06,
      "loss": 0.0011,
      "step": 229770
    },
    {
      "epoch": 1.473737545416938,
      "grad_norm": 0.03028823249042034,
      "learning_rate": 1.964918234251445e-06,
      "loss": 0.0007,
      "step": 229780
    },
    {
      "epoch": 1.473801682310724,
      "grad_norm": 0.12614458799362183,
      "learning_rate": 1.9644734629629114e-06,
      "loss": 0.0017,
      "step": 229790
    },
    {
      "epoch": 1.47386581920451,
      "grad_norm": 0.11758097261190414,
      "learning_rate": 1.9640287297116933e-06,
      "loss": 0.0018,
      "step": 229800
    },
    {
      "epoch": 1.4739299560982961,
      "grad_norm": 0.038990456610918045,
      "learning_rate": 1.9635840345033667e-06,
      "loss": 0.0016,
      "step": 229810
    },
    {
      "epoch": 1.4739940929920823,
      "grad_norm": 0.08074386417865753,
      "learning_rate": 1.9631393773435006e-06,
      "loss": 0.0008,
      "step": 229820
    },
    {
      "epoch": 1.4740582298858684,
      "grad_norm": 0.061525046825408936,
      "learning_rate": 1.9626947582376704e-06,
      "loss": 0.0017,
      "step": 229830
    },
    {
      "epoch": 1.4741223667796546,
      "grad_norm": 0.1928442120552063,
      "learning_rate": 1.9622501771914454e-06,
      "loss": 0.0011,
      "step": 229840
    },
    {
      "epoch": 1.4741865036734407,
      "grad_norm": 0.012257641181349754,
      "learning_rate": 1.9618056342103965e-06,
      "loss": 0.0006,
      "step": 229850
    },
    {
      "epoch": 1.4742506405672267,
      "grad_norm": 0.05344541743397713,
      "learning_rate": 1.9613611293000927e-06,
      "loss": 0.0011,
      "step": 229860
    },
    {
      "epoch": 1.4743147774610128,
      "grad_norm": 0.08402790874242783,
      "learning_rate": 1.9609166624661076e-06,
      "loss": 0.0023,
      "step": 229870
    },
    {
      "epoch": 1.474378914354799,
      "grad_norm": 0.0019207809818908572,
      "learning_rate": 1.9604722337140086e-06,
      "loss": 0.0006,
      "step": 229880
    },
    {
      "epoch": 1.4744430512485849,
      "grad_norm": 0.01550615206360817,
      "learning_rate": 1.960027843049365e-06,
      "loss": 0.0016,
      "step": 229890
    },
    {
      "epoch": 1.474507188142371,
      "grad_norm": 0.10568740963935852,
      "learning_rate": 1.9595834904777434e-06,
      "loss": 0.0023,
      "step": 229900
    },
    {
      "epoch": 1.4745713250361572,
      "grad_norm": 0.1658589094877243,
      "learning_rate": 1.9591391760047153e-06,
      "loss": 0.0007,
      "step": 229910
    },
    {
      "epoch": 1.4746354619299433,
      "grad_norm": 0.3455018401145935,
      "learning_rate": 1.9586948996358463e-06,
      "loss": 0.003,
      "step": 229920
    },
    {
      "epoch": 1.4746995988237295,
      "grad_norm": 0.025642002001404762,
      "learning_rate": 1.9582506613767023e-06,
      "loss": 0.0045,
      "step": 229930
    },
    {
      "epoch": 1.4747637357175154,
      "grad_norm": 0.18122798204421997,
      "learning_rate": 1.957806461232853e-06,
      "loss": 0.0018,
      "step": 229940
    },
    {
      "epoch": 1.4748278726113015,
      "grad_norm": 0.047227855771780014,
      "learning_rate": 1.957362299209863e-06,
      "loss": 0.0008,
      "step": 229950
    },
    {
      "epoch": 1.4748920095050877,
      "grad_norm": 0.059645794332027435,
      "learning_rate": 1.9569181753132982e-06,
      "loss": 0.0014,
      "step": 229960
    },
    {
      "epoch": 1.4749561463988736,
      "grad_norm": 0.044832829385995865,
      "learning_rate": 1.9564740895487217e-06,
      "loss": 0.0013,
      "step": 229970
    },
    {
      "epoch": 1.4750202832926598,
      "grad_norm": 0.009424911811947823,
      "learning_rate": 1.9560300419217015e-06,
      "loss": 0.0013,
      "step": 229980
    },
    {
      "epoch": 1.475084420186446,
      "grad_norm": 0.08541758358478546,
      "learning_rate": 1.9555860324378005e-06,
      "loss": 0.0012,
      "step": 229990
    },
    {
      "epoch": 1.475148557080232,
      "grad_norm": 0.08901616185903549,
      "learning_rate": 1.955142061102582e-06,
      "loss": 0.0011,
      "step": 230000
    },
    {
      "epoch": 1.4752126939740182,
      "grad_norm": 0.09359852969646454,
      "learning_rate": 1.954698127921609e-06,
      "loss": 0.0015,
      "step": 230010
    },
    {
      "epoch": 1.4752768308678044,
      "grad_norm": 0.1309373825788498,
      "learning_rate": 1.9542542329004456e-06,
      "loss": 0.0013,
      "step": 230020
    },
    {
      "epoch": 1.4753409677615903,
      "grad_norm": 0.12006743997335434,
      "learning_rate": 1.9538103760446537e-06,
      "loss": 0.0013,
      "step": 230030
    },
    {
      "epoch": 1.4754051046553764,
      "grad_norm": 0.06533078104257584,
      "learning_rate": 1.953366557359795e-06,
      "loss": 0.0012,
      "step": 230040
    },
    {
      "epoch": 1.4754692415491626,
      "grad_norm": 0.0857715979218483,
      "learning_rate": 1.9529227768514286e-06,
      "loss": 0.0018,
      "step": 230050
    },
    {
      "epoch": 1.4755333784429485,
      "grad_norm": 0.02155408076941967,
      "learning_rate": 1.9524790345251193e-06,
      "loss": 0.0012,
      "step": 230060
    },
    {
      "epoch": 1.4755975153367347,
      "grad_norm": 0.05905630812048912,
      "learning_rate": 1.952035330386426e-06,
      "loss": 0.0007,
      "step": 230070
    },
    {
      "epoch": 1.4756616522305208,
      "grad_norm": 0.014506034553050995,
      "learning_rate": 1.951591664440907e-06,
      "loss": 0.0005,
      "step": 230080
    },
    {
      "epoch": 1.475725789124307,
      "grad_norm": 0.24017755687236786,
      "learning_rate": 1.9511480366941245e-06,
      "loss": 0.0019,
      "step": 230090
    },
    {
      "epoch": 1.4757899260180931,
      "grad_norm": 0.031952865421772,
      "learning_rate": 1.9507044471516358e-06,
      "loss": 0.0012,
      "step": 230100
    },
    {
      "epoch": 1.475854062911879,
      "grad_norm": 0.15895450115203857,
      "learning_rate": 1.9502608958189995e-06,
      "loss": 0.0014,
      "step": 230110
    },
    {
      "epoch": 1.4759181998056652,
      "grad_norm": 0.260976105928421,
      "learning_rate": 1.9498173827017724e-06,
      "loss": 0.0016,
      "step": 230120
    },
    {
      "epoch": 1.4759823366994513,
      "grad_norm": 0.030834907665848732,
      "learning_rate": 1.949373907805515e-06,
      "loss": 0.0009,
      "step": 230130
    },
    {
      "epoch": 1.4760464735932375,
      "grad_norm": 0.03008120320737362,
      "learning_rate": 1.948930471135783e-06,
      "loss": 0.0012,
      "step": 230140
    },
    {
      "epoch": 1.4761106104870234,
      "grad_norm": 0.19386741518974304,
      "learning_rate": 1.948487072698133e-06,
      "loss": 0.0009,
      "step": 230150
    },
    {
      "epoch": 1.4761747473808096,
      "grad_norm": 0.0772666484117508,
      "learning_rate": 1.948043712498119e-06,
      "loss": 0.0008,
      "step": 230160
    },
    {
      "epoch": 1.4762388842745957,
      "grad_norm": 0.007100511807948351,
      "learning_rate": 1.9476003905413e-06,
      "loss": 0.0022,
      "step": 230170
    },
    {
      "epoch": 1.4763030211683819,
      "grad_norm": 0.09792197495698929,
      "learning_rate": 1.9471571068332295e-06,
      "loss": 0.0008,
      "step": 230180
    },
    {
      "epoch": 1.476367158062168,
      "grad_norm": 0.04310609772801399,
      "learning_rate": 1.946713861379463e-06,
      "loss": 0.001,
      "step": 230190
    },
    {
      "epoch": 1.476431294955954,
      "grad_norm": 0.08101110905408859,
      "learning_rate": 1.9462706541855514e-06,
      "loss": 0.0017,
      "step": 230200
    },
    {
      "epoch": 1.47649543184974,
      "grad_norm": 0.05013372004032135,
      "learning_rate": 1.945827485257053e-06,
      "loss": 0.0014,
      "step": 230210
    },
    {
      "epoch": 1.4765595687435262,
      "grad_norm": 0.38013651967048645,
      "learning_rate": 1.945384354599519e-06,
      "loss": 0.002,
      "step": 230220
    },
    {
      "epoch": 1.4766237056373122,
      "grad_norm": 0.06051740422844887,
      "learning_rate": 1.9449412622185004e-06,
      "loss": 0.0006,
      "step": 230230
    },
    {
      "epoch": 1.4766878425310983,
      "grad_norm": 0.001617398695088923,
      "learning_rate": 1.944498208119553e-06,
      "loss": 0.0014,
      "step": 230240
    },
    {
      "epoch": 1.4767519794248845,
      "grad_norm": 0.18351486325263977,
      "learning_rate": 1.9440551923082266e-06,
      "loss": 0.0012,
      "step": 230250
    },
    {
      "epoch": 1.4768161163186706,
      "grad_norm": 0.0058859786950051785,
      "learning_rate": 1.9436122147900733e-06,
      "loss": 0.0009,
      "step": 230260
    },
    {
      "epoch": 1.4768802532124568,
      "grad_norm": 0.0904245376586914,
      "learning_rate": 1.943169275570641e-06,
      "loss": 0.0016,
      "step": 230270
    },
    {
      "epoch": 1.476944390106243,
      "grad_norm": 0.08254070580005646,
      "learning_rate": 1.9427263746554843e-06,
      "loss": 0.0014,
      "step": 230280
    },
    {
      "epoch": 1.4770085270000288,
      "grad_norm": 0.054536763578653336,
      "learning_rate": 1.942283512050151e-06,
      "loss": 0.0006,
      "step": 230290
    },
    {
      "epoch": 1.477072663893815,
      "grad_norm": 0.058397021144628525,
      "learning_rate": 1.94184068776019e-06,
      "loss": 0.0011,
      "step": 230300
    },
    {
      "epoch": 1.4771368007876011,
      "grad_norm": 0.009756239131093025,
      "learning_rate": 1.9413979017911504e-06,
      "loss": 0.0015,
      "step": 230310
    },
    {
      "epoch": 1.477200937681387,
      "grad_norm": 0.02611563727259636,
      "learning_rate": 1.940955154148582e-06,
      "loss": 0.0013,
      "step": 230320
    },
    {
      "epoch": 1.4772650745751732,
      "grad_norm": 0.15121859312057495,
      "learning_rate": 1.940512444838032e-06,
      "loss": 0.0011,
      "step": 230330
    },
    {
      "epoch": 1.4773292114689593,
      "grad_norm": 0.11068111658096313,
      "learning_rate": 1.9400697738650474e-06,
      "loss": 0.0011,
      "step": 230340
    },
    {
      "epoch": 1.4773933483627455,
      "grad_norm": 0.014362462796270847,
      "learning_rate": 1.9396271412351737e-06,
      "loss": 0.0009,
      "step": 230350
    },
    {
      "epoch": 1.4774574852565316,
      "grad_norm": 0.07700163125991821,
      "learning_rate": 1.939184546953961e-06,
      "loss": 0.001,
      "step": 230360
    },
    {
      "epoch": 1.4775216221503176,
      "grad_norm": 0.05529524013400078,
      "learning_rate": 1.9387419910269536e-06,
      "loss": 0.0013,
      "step": 230370
    },
    {
      "epoch": 1.4775857590441037,
      "grad_norm": 0.12332439422607422,
      "learning_rate": 1.9382994734596948e-06,
      "loss": 0.0022,
      "step": 230380
    },
    {
      "epoch": 1.4776498959378899,
      "grad_norm": 0.1013512909412384,
      "learning_rate": 1.937856994257734e-06,
      "loss": 0.001,
      "step": 230390
    },
    {
      "epoch": 1.4777140328316758,
      "grad_norm": 0.02150823548436165,
      "learning_rate": 1.9374145534266133e-06,
      "loss": 0.0011,
      "step": 230400
    },
    {
      "epoch": 1.477778169725462,
      "grad_norm": 0.042761147022247314,
      "learning_rate": 1.936972150971877e-06,
      "loss": 0.0019,
      "step": 230410
    },
    {
      "epoch": 1.477842306619248,
      "grad_norm": 0.16678951680660248,
      "learning_rate": 1.936529786899067e-06,
      "loss": 0.001,
      "step": 230420
    },
    {
      "epoch": 1.4779064435130342,
      "grad_norm": 0.07481800019741058,
      "learning_rate": 1.93608746121373e-06,
      "loss": 0.0012,
      "step": 230430
    },
    {
      "epoch": 1.4779705804068204,
      "grad_norm": 0.020131539553403854,
      "learning_rate": 1.9356451739214067e-06,
      "loss": 0.0005,
      "step": 230440
    },
    {
      "epoch": 1.4780347173006065,
      "grad_norm": 0.0174866896122694,
      "learning_rate": 1.93520292502764e-06,
      "loss": 0.0011,
      "step": 230450
    },
    {
      "epoch": 1.4780988541943925,
      "grad_norm": 0.03777153789997101,
      "learning_rate": 1.934760714537969e-06,
      "loss": 0.0013,
      "step": 230460
    },
    {
      "epoch": 1.4781629910881786,
      "grad_norm": 0.021376833319664,
      "learning_rate": 1.934318542457938e-06,
      "loss": 0.0008,
      "step": 230470
    },
    {
      "epoch": 1.4782271279819648,
      "grad_norm": 0.047228723764419556,
      "learning_rate": 1.9338764087930873e-06,
      "loss": 0.0013,
      "step": 230480
    },
    {
      "epoch": 1.4782912648757507,
      "grad_norm": 0.026556644588708878,
      "learning_rate": 1.9334343135489564e-06,
      "loss": 0.0012,
      "step": 230490
    },
    {
      "epoch": 1.4783554017695368,
      "grad_norm": 0.157155841588974,
      "learning_rate": 1.9329922567310833e-06,
      "loss": 0.0023,
      "step": 230500
    },
    {
      "epoch": 1.478419538663323,
      "grad_norm": 0.08791355788707733,
      "learning_rate": 1.9325502383450114e-06,
      "loss": 0.0013,
      "step": 230510
    },
    {
      "epoch": 1.4784836755571091,
      "grad_norm": 0.21543775498867035,
      "learning_rate": 1.932108258396277e-06,
      "loss": 0.002,
      "step": 230520
    },
    {
      "epoch": 1.4785478124508953,
      "grad_norm": 0.1585678607225418,
      "learning_rate": 1.931666316890417e-06,
      "loss": 0.0024,
      "step": 230530
    },
    {
      "epoch": 1.4786119493446812,
      "grad_norm": 0.036143168807029724,
      "learning_rate": 1.931224413832973e-06,
      "loss": 0.0007,
      "step": 230540
    },
    {
      "epoch": 1.4786760862384674,
      "grad_norm": 0.07579084485769272,
      "learning_rate": 1.9307825492294798e-06,
      "loss": 0.0013,
      "step": 230550
    },
    {
      "epoch": 1.4787402231322535,
      "grad_norm": 0.2692042589187622,
      "learning_rate": 1.930340723085475e-06,
      "loss": 0.0019,
      "step": 230560
    },
    {
      "epoch": 1.4788043600260397,
      "grad_norm": 0.10390500724315643,
      "learning_rate": 1.929898935406493e-06,
      "loss": 0.0009,
      "step": 230570
    },
    {
      "epoch": 1.4788684969198256,
      "grad_norm": 0.03958144411444664,
      "learning_rate": 1.929457186198071e-06,
      "loss": 0.001,
      "step": 230580
    },
    {
      "epoch": 1.4789326338136117,
      "grad_norm": 0.06917309761047363,
      "learning_rate": 1.9290154754657474e-06,
      "loss": 0.0011,
      "step": 230590
    },
    {
      "epoch": 1.4789967707073979,
      "grad_norm": 0.05723357945680618,
      "learning_rate": 1.928573803215055e-06,
      "loss": 0.0012,
      "step": 230600
    },
    {
      "epoch": 1.479060907601184,
      "grad_norm": 0.10763806849718094,
      "learning_rate": 1.928132169451527e-06,
      "loss": 0.001,
      "step": 230610
    },
    {
      "epoch": 1.4791250444949702,
      "grad_norm": 0.21228329837322235,
      "learning_rate": 1.927690574180697e-06,
      "loss": 0.0017,
      "step": 230620
    },
    {
      "epoch": 1.479189181388756,
      "grad_norm": 0.10918084532022476,
      "learning_rate": 1.9272490174081015e-06,
      "loss": 0.0011,
      "step": 230630
    },
    {
      "epoch": 1.4792533182825423,
      "grad_norm": 0.061497002840042114,
      "learning_rate": 1.9268074991392722e-06,
      "loss": 0.002,
      "step": 230640
    },
    {
      "epoch": 1.4793174551763284,
      "grad_norm": 0.11596682667732239,
      "learning_rate": 1.9263660193797397e-06,
      "loss": 0.0013,
      "step": 230650
    },
    {
      "epoch": 1.4793815920701143,
      "grad_norm": 0.0664464458823204,
      "learning_rate": 1.9259245781350387e-06,
      "loss": 0.0018,
      "step": 230660
    },
    {
      "epoch": 1.4794457289639005,
      "grad_norm": 0.06875289231538773,
      "learning_rate": 1.9254831754107e-06,
      "loss": 0.0006,
      "step": 230670
    },
    {
      "epoch": 1.4795098658576866,
      "grad_norm": 0.2601640224456787,
      "learning_rate": 1.925041811212255e-06,
      "loss": 0.0019,
      "step": 230680
    },
    {
      "epoch": 1.4795740027514728,
      "grad_norm": 0.10457682609558105,
      "learning_rate": 1.9246004855452314e-06,
      "loss": 0.001,
      "step": 230690
    },
    {
      "epoch": 1.479638139645259,
      "grad_norm": 0.16158710420131683,
      "learning_rate": 1.9241591984151636e-06,
      "loss": 0.0008,
      "step": 230700
    },
    {
      "epoch": 1.479702276539045,
      "grad_norm": 0.11226168274879456,
      "learning_rate": 1.9237179498275792e-06,
      "loss": 0.0007,
      "step": 230710
    },
    {
      "epoch": 1.479766413432831,
      "grad_norm": 0.09860129654407501,
      "learning_rate": 1.923276739788008e-06,
      "loss": 0.0016,
      "step": 230720
    },
    {
      "epoch": 1.4798305503266171,
      "grad_norm": 0.07697834819555283,
      "learning_rate": 1.922835568301976e-06,
      "loss": 0.0023,
      "step": 230730
    },
    {
      "epoch": 1.4798946872204033,
      "grad_norm": 0.10646352916955948,
      "learning_rate": 1.922394435375015e-06,
      "loss": 0.0014,
      "step": 230740
    },
    {
      "epoch": 1.4799588241141892,
      "grad_norm": 0.06855232268571854,
      "learning_rate": 1.9219533410126517e-06,
      "loss": 0.0013,
      "step": 230750
    },
    {
      "epoch": 1.4800229610079754,
      "grad_norm": 0.1098976582288742,
      "learning_rate": 1.9215122852204125e-06,
      "loss": 0.0007,
      "step": 230760
    },
    {
      "epoch": 1.4800870979017615,
      "grad_norm": 0.15310770273208618,
      "learning_rate": 1.9210712680038236e-06,
      "loss": 0.0009,
      "step": 230770
    },
    {
      "epoch": 1.4801512347955477,
      "grad_norm": 0.02309637889266014,
      "learning_rate": 1.920630289368413e-06,
      "loss": 0.0024,
      "step": 230780
    },
    {
      "epoch": 1.4802153716893338,
      "grad_norm": 0.12637177109718323,
      "learning_rate": 1.920189349319706e-06,
      "loss": 0.0017,
      "step": 230790
    },
    {
      "epoch": 1.4802795085831197,
      "grad_norm": 0.07150278985500336,
      "learning_rate": 1.919748447863226e-06,
      "loss": 0.0011,
      "step": 230800
    },
    {
      "epoch": 1.480343645476906,
      "grad_norm": 0.07935880124568939,
      "learning_rate": 1.9193075850045006e-06,
      "loss": 0.0011,
      "step": 230810
    },
    {
      "epoch": 1.480407782370692,
      "grad_norm": 0.011261014267802238,
      "learning_rate": 1.9188667607490533e-06,
      "loss": 0.0012,
      "step": 230820
    },
    {
      "epoch": 1.480471919264478,
      "grad_norm": 0.10852573066949844,
      "learning_rate": 1.9184259751024076e-06,
      "loss": 0.001,
      "step": 230830
    },
    {
      "epoch": 1.4805360561582641,
      "grad_norm": 0.044947221875190735,
      "learning_rate": 1.917985228070085e-06,
      "loss": 0.0018,
      "step": 230840
    },
    {
      "epoch": 1.4806001930520503,
      "grad_norm": 0.0234210267663002,
      "learning_rate": 1.917544519657612e-06,
      "loss": 0.0007,
      "step": 230850
    },
    {
      "epoch": 1.4806643299458364,
      "grad_norm": 0.10135694593191147,
      "learning_rate": 1.917103849870509e-06,
      "loss": 0.0023,
      "step": 230860
    },
    {
      "epoch": 1.4807284668396226,
      "grad_norm": 0.011645328253507614,
      "learning_rate": 1.916663218714298e-06,
      "loss": 0.0006,
      "step": 230870
    },
    {
      "epoch": 1.4807926037334087,
      "grad_norm": 0.08838336914777756,
      "learning_rate": 1.9162226261944988e-06,
      "loss": 0.0008,
      "step": 230880
    },
    {
      "epoch": 1.4808567406271946,
      "grad_norm": 0.06066448986530304,
      "learning_rate": 1.915782072316636e-06,
      "loss": 0.0012,
      "step": 230890
    },
    {
      "epoch": 1.4809208775209808,
      "grad_norm": 0.2691197395324707,
      "learning_rate": 1.9153415570862275e-06,
      "loss": 0.0015,
      "step": 230900
    },
    {
      "epoch": 1.480985014414767,
      "grad_norm": 0.00545494956895709,
      "learning_rate": 1.9149010805087947e-06,
      "loss": 0.0015,
      "step": 230910
    },
    {
      "epoch": 1.4810491513085529,
      "grad_norm": 0.06407414376735687,
      "learning_rate": 1.9144606425898545e-06,
      "loss": 0.0021,
      "step": 230920
    },
    {
      "epoch": 1.481113288202339,
      "grad_norm": 0.024175606667995453,
      "learning_rate": 1.9140202433349288e-06,
      "loss": 0.0005,
      "step": 230930
    },
    {
      "epoch": 1.4811774250961252,
      "grad_norm": 0.10273516178131104,
      "learning_rate": 1.913579882749535e-06,
      "loss": 0.0013,
      "step": 230940
    },
    {
      "epoch": 1.4812415619899113,
      "grad_norm": 0.065152108669281,
      "learning_rate": 1.9131395608391893e-06,
      "loss": 0.0007,
      "step": 230950
    },
    {
      "epoch": 1.4813056988836975,
      "grad_norm": 0.05232097581028938,
      "learning_rate": 1.9126992776094133e-06,
      "loss": 0.0007,
      "step": 230960
    },
    {
      "epoch": 1.4813698357774834,
      "grad_norm": 0.03106767311692238,
      "learning_rate": 1.9122590330657215e-06,
      "loss": 0.001,
      "step": 230970
    },
    {
      "epoch": 1.4814339726712695,
      "grad_norm": 0.10196743160486221,
      "learning_rate": 1.9118188272136305e-06,
      "loss": 0.001,
      "step": 230980
    },
    {
      "epoch": 1.4814981095650557,
      "grad_norm": 0.08105448633432388,
      "learning_rate": 1.9113786600586555e-06,
      "loss": 0.0008,
      "step": 230990
    },
    {
      "epoch": 1.4815622464588418,
      "grad_norm": 0.0760020911693573,
      "learning_rate": 1.9109385316063146e-06,
      "loss": 0.0009,
      "step": 231000
    },
    {
      "epoch": 1.4816263833526278,
      "grad_norm": 0.02205081842839718,
      "learning_rate": 1.910498441862122e-06,
      "loss": 0.001,
      "step": 231010
    },
    {
      "epoch": 1.481690520246414,
      "grad_norm": 0.09554306417703629,
      "learning_rate": 1.9100583908315917e-06,
      "loss": 0.0009,
      "step": 231020
    },
    {
      "epoch": 1.4817546571402,
      "grad_norm": 0.008705779910087585,
      "learning_rate": 1.9096183785202365e-06,
      "loss": 0.0017,
      "step": 231030
    },
    {
      "epoch": 1.4818187940339862,
      "grad_norm": 0.014837021008133888,
      "learning_rate": 1.9091784049335733e-06,
      "loss": 0.0016,
      "step": 231040
    },
    {
      "epoch": 1.4818829309277723,
      "grad_norm": 0.08490317314863205,
      "learning_rate": 1.908738470077114e-06,
      "loss": 0.0009,
      "step": 231050
    },
    {
      "epoch": 1.4819470678215583,
      "grad_norm": 0.2210465967655182,
      "learning_rate": 1.9082985739563703e-06,
      "loss": 0.0023,
      "step": 231060
    },
    {
      "epoch": 1.4820112047153444,
      "grad_norm": 0.09002111852169037,
      "learning_rate": 1.9078587165768537e-06,
      "loss": 0.0008,
      "step": 231070
    },
    {
      "epoch": 1.4820753416091306,
      "grad_norm": 0.07184798270463943,
      "learning_rate": 1.9074188979440796e-06,
      "loss": 0.0008,
      "step": 231080
    },
    {
      "epoch": 1.4821394785029165,
      "grad_norm": 0.04037946090102196,
      "learning_rate": 1.906979118063556e-06,
      "loss": 0.0016,
      "step": 231090
    },
    {
      "epoch": 1.4822036153967026,
      "grad_norm": 0.007163423113524914,
      "learning_rate": 1.9065393769407931e-06,
      "loss": 0.0009,
      "step": 231100
    },
    {
      "epoch": 1.4822677522904888,
      "grad_norm": 0.3299335837364197,
      "learning_rate": 1.9060996745813049e-06,
      "loss": 0.002,
      "step": 231110
    },
    {
      "epoch": 1.482331889184275,
      "grad_norm": 0.09048480540513992,
      "learning_rate": 1.905660010990598e-06,
      "loss": 0.0008,
      "step": 231120
    },
    {
      "epoch": 1.482396026078061,
      "grad_norm": 0.0650215595960617,
      "learning_rate": 1.9052203861741837e-06,
      "loss": 0.0021,
      "step": 231130
    },
    {
      "epoch": 1.4824601629718472,
      "grad_norm": 0.0951550230383873,
      "learning_rate": 1.9047808001375679e-06,
      "loss": 0.0009,
      "step": 231140
    },
    {
      "epoch": 1.4825242998656332,
      "grad_norm": 0.027630243450403214,
      "learning_rate": 1.904341252886262e-06,
      "loss": 0.0006,
      "step": 231150
    },
    {
      "epoch": 1.4825884367594193,
      "grad_norm": 0.1635369211435318,
      "learning_rate": 1.9039017444257728e-06,
      "loss": 0.0028,
      "step": 231160
    },
    {
      "epoch": 1.4826525736532055,
      "grad_norm": 0.06888850033283234,
      "learning_rate": 1.9034622747616078e-06,
      "loss": 0.0014,
      "step": 231170
    },
    {
      "epoch": 1.4827167105469914,
      "grad_norm": 0.10007062554359436,
      "learning_rate": 1.9030228438992716e-06,
      "loss": 0.001,
      "step": 231180
    },
    {
      "epoch": 1.4827808474407775,
      "grad_norm": 0.058289770036935806,
      "learning_rate": 1.9025834518442737e-06,
      "loss": 0.0008,
      "step": 231190
    },
    {
      "epoch": 1.4828449843345637,
      "grad_norm": 0.052228596061468124,
      "learning_rate": 1.9021440986021196e-06,
      "loss": 0.001,
      "step": 231200
    },
    {
      "epoch": 1.4829091212283498,
      "grad_norm": 0.050283003598451614,
      "learning_rate": 1.9017047841783137e-06,
      "loss": 0.001,
      "step": 231210
    },
    {
      "epoch": 1.482973258122136,
      "grad_norm": 0.16117650270462036,
      "learning_rate": 1.90126550857836e-06,
      "loss": 0.0013,
      "step": 231220
    },
    {
      "epoch": 1.483037395015922,
      "grad_norm": 0.024334462359547615,
      "learning_rate": 1.9008262718077653e-06,
      "loss": 0.0016,
      "step": 231230
    },
    {
      "epoch": 1.483101531909708,
      "grad_norm": 0.04623173549771309,
      "learning_rate": 1.9003870738720321e-06,
      "loss": 0.0008,
      "step": 231240
    },
    {
      "epoch": 1.4831656688034942,
      "grad_norm": 0.15785768628120422,
      "learning_rate": 1.8999479147766625e-06,
      "loss": 0.0018,
      "step": 231250
    },
    {
      "epoch": 1.4832298056972804,
      "grad_norm": 0.20995059609413147,
      "learning_rate": 1.8995087945271634e-06,
      "loss": 0.0021,
      "step": 231260
    },
    {
      "epoch": 1.4832939425910663,
      "grad_norm": 0.02085651084780693,
      "learning_rate": 1.8990697131290341e-06,
      "loss": 0.0014,
      "step": 231270
    },
    {
      "epoch": 1.4833580794848524,
      "grad_norm": 0.13163235783576965,
      "learning_rate": 1.8986306705877782e-06,
      "loss": 0.0011,
      "step": 231280
    },
    {
      "epoch": 1.4834222163786386,
      "grad_norm": 0.02859206683933735,
      "learning_rate": 1.8981916669088947e-06,
      "loss": 0.0015,
      "step": 231290
    },
    {
      "epoch": 1.4834863532724247,
      "grad_norm": 0.45611247420310974,
      "learning_rate": 1.897752702097888e-06,
      "loss": 0.0146,
      "step": 231300
    },
    {
      "epoch": 1.4835504901662109,
      "grad_norm": 0.29005199670791626,
      "learning_rate": 1.8973137761602572e-06,
      "loss": 0.0014,
      "step": 231310
    },
    {
      "epoch": 1.4836146270599968,
      "grad_norm": 0.03413574770092964,
      "learning_rate": 1.8968748891015021e-06,
      "loss": 0.0035,
      "step": 231320
    },
    {
      "epoch": 1.483678763953783,
      "grad_norm": 0.037481214851140976,
      "learning_rate": 1.896436040927121e-06,
      "loss": 0.0012,
      "step": 231330
    },
    {
      "epoch": 1.483742900847569,
      "grad_norm": 0.0579238161444664,
      "learning_rate": 1.895997231642614e-06,
      "loss": 0.0007,
      "step": 231340
    },
    {
      "epoch": 1.483807037741355,
      "grad_norm": 0.027532974258065224,
      "learning_rate": 1.8955584612534821e-06,
      "loss": 0.0018,
      "step": 231350
    },
    {
      "epoch": 1.4838711746351412,
      "grad_norm": 0.11463338881731033,
      "learning_rate": 1.8951197297652214e-06,
      "loss": 0.0013,
      "step": 231360
    },
    {
      "epoch": 1.4839353115289273,
      "grad_norm": 0.09564340114593506,
      "learning_rate": 1.8946810371833273e-06,
      "loss": 0.0017,
      "step": 231370
    },
    {
      "epoch": 1.4839994484227135,
      "grad_norm": 0.04451832175254822,
      "learning_rate": 1.8942423835133012e-06,
      "loss": 0.0012,
      "step": 231380
    },
    {
      "epoch": 1.4840635853164996,
      "grad_norm": 0.09709803760051727,
      "learning_rate": 1.8938037687606376e-06,
      "loss": 0.0006,
      "step": 231390
    },
    {
      "epoch": 1.4841277222102858,
      "grad_norm": 0.05841127783060074,
      "learning_rate": 1.8933651929308328e-06,
      "loss": 0.0015,
      "step": 231400
    },
    {
      "epoch": 1.4841918591040717,
      "grad_norm": 0.05287181958556175,
      "learning_rate": 1.8929266560293807e-06,
      "loss": 0.0008,
      "step": 231410
    },
    {
      "epoch": 1.4842559959978578,
      "grad_norm": 0.0692257508635521,
      "learning_rate": 1.8924881580617804e-06,
      "loss": 0.0005,
      "step": 231420
    },
    {
      "epoch": 1.484320132891644,
      "grad_norm": 0.07513230293989182,
      "learning_rate": 1.8920496990335235e-06,
      "loss": 0.0005,
      "step": 231430
    },
    {
      "epoch": 1.48438426978543,
      "grad_norm": 0.2818194031715393,
      "learning_rate": 1.8916112789501057e-06,
      "loss": 0.0016,
      "step": 231440
    },
    {
      "epoch": 1.484448406679216,
      "grad_norm": 0.012203599326312542,
      "learning_rate": 1.891172897817018e-06,
      "loss": 0.0012,
      "step": 231450
    },
    {
      "epoch": 1.4845125435730022,
      "grad_norm": 0.02082725055515766,
      "learning_rate": 1.8907345556397577e-06,
      "loss": 0.0011,
      "step": 231460
    },
    {
      "epoch": 1.4845766804667884,
      "grad_norm": 0.12193366885185242,
      "learning_rate": 1.8902962524238156e-06,
      "loss": 0.0009,
      "step": 231470
    },
    {
      "epoch": 1.4846408173605745,
      "grad_norm": 0.1180684044957161,
      "learning_rate": 1.8898579881746842e-06,
      "loss": 0.0015,
      "step": 231480
    },
    {
      "epoch": 1.4847049542543604,
      "grad_norm": 0.0809510126709938,
      "learning_rate": 1.8894197628978528e-06,
      "loss": 0.0012,
      "step": 231490
    },
    {
      "epoch": 1.4847690911481466,
      "grad_norm": 0.17628474533557892,
      "learning_rate": 1.8889815765988168e-06,
      "loss": 0.0015,
      "step": 231500
    },
    {
      "epoch": 1.4848332280419327,
      "grad_norm": 0.05376352742314339,
      "learning_rate": 1.8885434292830652e-06,
      "loss": 0.0037,
      "step": 231510
    },
    {
      "epoch": 1.4848973649357187,
      "grad_norm": 0.02923703007400036,
      "learning_rate": 1.8881053209560857e-06,
      "loss": 0.0009,
      "step": 231520
    },
    {
      "epoch": 1.4849615018295048,
      "grad_norm": 0.026265686377882957,
      "learning_rate": 1.8876672516233729e-06,
      "loss": 0.0013,
      "step": 231530
    },
    {
      "epoch": 1.485025638723291,
      "grad_norm": 0.06677863746881485,
      "learning_rate": 1.8872292212904136e-06,
      "loss": 0.0018,
      "step": 231540
    },
    {
      "epoch": 1.4850897756170771,
      "grad_norm": 0.06208925321698189,
      "learning_rate": 1.8867912299626973e-06,
      "loss": 0.0014,
      "step": 231550
    },
    {
      "epoch": 1.4851539125108633,
      "grad_norm": 0.1605261266231537,
      "learning_rate": 1.8863532776457094e-06,
      "loss": 0.0018,
      "step": 231560
    },
    {
      "epoch": 1.4852180494046494,
      "grad_norm": 0.039795663207769394,
      "learning_rate": 1.8859153643449424e-06,
      "loss": 0.0008,
      "step": 231570
    },
    {
      "epoch": 1.4852821862984353,
      "grad_norm": 0.0179697647690773,
      "learning_rate": 1.8854774900658812e-06,
      "loss": 0.0011,
      "step": 231580
    },
    {
      "epoch": 1.4853463231922215,
      "grad_norm": 0.1934625506401062,
      "learning_rate": 1.8850396548140133e-06,
      "loss": 0.0011,
      "step": 231590
    },
    {
      "epoch": 1.4854104600860076,
      "grad_norm": 0.07390321791172028,
      "learning_rate": 1.8846018585948228e-06,
      "loss": 0.0013,
      "step": 231600
    },
    {
      "epoch": 1.4854745969797936,
      "grad_norm": 0.06746132671833038,
      "learning_rate": 1.8841641014137995e-06,
      "loss": 0.0008,
      "step": 231610
    },
    {
      "epoch": 1.4855387338735797,
      "grad_norm": 0.06024260073900223,
      "learning_rate": 1.8837263832764268e-06,
      "loss": 0.0018,
      "step": 231620
    },
    {
      "epoch": 1.4856028707673659,
      "grad_norm": 0.06886883825063705,
      "learning_rate": 1.8832887041881876e-06,
      "loss": 0.0014,
      "step": 231630
    },
    {
      "epoch": 1.485667007661152,
      "grad_norm": 0.05687910318374634,
      "learning_rate": 1.8828510641545705e-06,
      "loss": 0.0008,
      "step": 231640
    },
    {
      "epoch": 1.4857311445549382,
      "grad_norm": 0.08611790090799332,
      "learning_rate": 1.8824134631810575e-06,
      "loss": 0.0021,
      "step": 231650
    },
    {
      "epoch": 1.485795281448724,
      "grad_norm": 0.028475651517510414,
      "learning_rate": 1.8819759012731314e-06,
      "loss": 0.0015,
      "step": 231660
    },
    {
      "epoch": 1.4858594183425102,
      "grad_norm": 0.08990021049976349,
      "learning_rate": 1.8815383784362745e-06,
      "loss": 0.0017,
      "step": 231670
    },
    {
      "epoch": 1.4859235552362964,
      "grad_norm": 0.04032914713025093,
      "learning_rate": 1.8811008946759718e-06,
      "loss": 0.0006,
      "step": 231680
    },
    {
      "epoch": 1.4859876921300825,
      "grad_norm": 0.061242811381816864,
      "learning_rate": 1.880663449997704e-06,
      "loss": 0.0014,
      "step": 231690
    },
    {
      "epoch": 1.4860518290238685,
      "grad_norm": 0.05517999827861786,
      "learning_rate": 1.8802260444069526e-06,
      "loss": 0.0022,
      "step": 231700
    },
    {
      "epoch": 1.4861159659176546,
      "grad_norm": 0.046025365591049194,
      "learning_rate": 1.8797886779091968e-06,
      "loss": 0.0012,
      "step": 231710
    },
    {
      "epoch": 1.4861801028114408,
      "grad_norm": 0.01580861397087574,
      "learning_rate": 1.879351350509921e-06,
      "loss": 0.0009,
      "step": 231720
    },
    {
      "epoch": 1.486244239705227,
      "grad_norm": 0.115535669028759,
      "learning_rate": 1.8789140622146025e-06,
      "loss": 0.0009,
      "step": 231730
    },
    {
      "epoch": 1.486308376599013,
      "grad_norm": 0.08806314319372177,
      "learning_rate": 1.8784768130287211e-06,
      "loss": 0.001,
      "step": 231740
    },
    {
      "epoch": 1.486372513492799,
      "grad_norm": 0.17606164515018463,
      "learning_rate": 1.8780396029577551e-06,
      "loss": 0.0015,
      "step": 231750
    },
    {
      "epoch": 1.4864366503865851,
      "grad_norm": 0.01907699555158615,
      "learning_rate": 1.8776024320071856e-06,
      "loss": 0.0012,
      "step": 231760
    },
    {
      "epoch": 1.4865007872803713,
      "grad_norm": 0.09007932245731354,
      "learning_rate": 1.8771653001824891e-06,
      "loss": 0.0007,
      "step": 231770
    },
    {
      "epoch": 1.4865649241741572,
      "grad_norm": 0.11589324474334717,
      "learning_rate": 1.876728207489142e-06,
      "loss": 0.0009,
      "step": 231780
    },
    {
      "epoch": 1.4866290610679433,
      "grad_norm": 0.05713924765586853,
      "learning_rate": 1.8762911539326234e-06,
      "loss": 0.0021,
      "step": 231790
    },
    {
      "epoch": 1.4866931979617295,
      "grad_norm": 0.04543076828122139,
      "learning_rate": 1.8758541395184099e-06,
      "loss": 0.001,
      "step": 231800
    },
    {
      "epoch": 1.4867573348555156,
      "grad_norm": 0.0943898931145668,
      "learning_rate": 1.8754171642519763e-06,
      "loss": 0.0013,
      "step": 231810
    },
    {
      "epoch": 1.4868214717493018,
      "grad_norm": 0.03184668347239494,
      "learning_rate": 1.8749802281387975e-06,
      "loss": 0.001,
      "step": 231820
    },
    {
      "epoch": 1.486885608643088,
      "grad_norm": 0.06906899809837341,
      "learning_rate": 1.8745433311843514e-06,
      "loss": 0.0023,
      "step": 231830
    },
    {
      "epoch": 1.4869497455368739,
      "grad_norm": 0.050902411341667175,
      "learning_rate": 1.8741064733941112e-06,
      "loss": 0.001,
      "step": 231840
    },
    {
      "epoch": 1.48701388243066,
      "grad_norm": 0.0598636195063591,
      "learning_rate": 1.8736696547735505e-06,
      "loss": 0.0016,
      "step": 231850
    },
    {
      "epoch": 1.4870780193244462,
      "grad_norm": 0.12425347417593002,
      "learning_rate": 1.873232875328142e-06,
      "loss": 0.0024,
      "step": 231860
    },
    {
      "epoch": 1.487142156218232,
      "grad_norm": 0.04981955885887146,
      "learning_rate": 1.8727961350633622e-06,
      "loss": 0.001,
      "step": 231870
    },
    {
      "epoch": 1.4872062931120182,
      "grad_norm": 0.0955575481057167,
      "learning_rate": 1.8723594339846813e-06,
      "loss": 0.0014,
      "step": 231880
    },
    {
      "epoch": 1.4872704300058044,
      "grad_norm": 0.04553009197115898,
      "learning_rate": 1.8719227720975725e-06,
      "loss": 0.0019,
      "step": 231890
    },
    {
      "epoch": 1.4873345668995905,
      "grad_norm": 0.09013471007347107,
      "learning_rate": 1.8714861494075048e-06,
      "loss": 0.0012,
      "step": 231900
    },
    {
      "epoch": 1.4873987037933767,
      "grad_norm": 0.07908090204000473,
      "learning_rate": 1.8710495659199535e-06,
      "loss": 0.0026,
      "step": 231910
    },
    {
      "epoch": 1.4874628406871626,
      "grad_norm": 0.08770821243524551,
      "learning_rate": 1.8706130216403872e-06,
      "loss": 0.0026,
      "step": 231920
    },
    {
      "epoch": 1.4875269775809488,
      "grad_norm": 0.02814689837396145,
      "learning_rate": 1.8701765165742746e-06,
      "loss": 0.0008,
      "step": 231930
    },
    {
      "epoch": 1.487591114474735,
      "grad_norm": 0.05187418684363365,
      "learning_rate": 1.8697400507270891e-06,
      "loss": 0.0011,
      "step": 231940
    },
    {
      "epoch": 1.4876552513685208,
      "grad_norm": 0.05968382582068443,
      "learning_rate": 1.8693036241042978e-06,
      "loss": 0.0016,
      "step": 231950
    },
    {
      "epoch": 1.487719388262307,
      "grad_norm": 0.08687455952167511,
      "learning_rate": 1.8688672367113691e-06,
      "loss": 0.0005,
      "step": 231960
    },
    {
      "epoch": 1.4877835251560931,
      "grad_norm": 0.04117651283740997,
      "learning_rate": 1.8684308885537704e-06,
      "loss": 0.0014,
      "step": 231970
    },
    {
      "epoch": 1.4878476620498793,
      "grad_norm": 0.10416851192712784,
      "learning_rate": 1.8679945796369725e-06,
      "loss": 0.001,
      "step": 231980
    },
    {
      "epoch": 1.4879117989436654,
      "grad_norm": 0.059600580483675,
      "learning_rate": 1.8675583099664413e-06,
      "loss": 0.001,
      "step": 231990
    },
    {
      "epoch": 1.4879759358374516,
      "grad_norm": 0.04816209897398949,
      "learning_rate": 1.8671220795476424e-06,
      "loss": 0.0005,
      "step": 232000
    },
    {
      "epoch": 1.4880400727312375,
      "grad_norm": 0.018877891823649406,
      "learning_rate": 1.8666858883860417e-06,
      "loss": 0.0016,
      "step": 232010
    },
    {
      "epoch": 1.4881042096250237,
      "grad_norm": 0.11844898760318756,
      "learning_rate": 1.866249736487108e-06,
      "loss": 0.0009,
      "step": 232020
    },
    {
      "epoch": 1.4881683465188098,
      "grad_norm": 0.07525431364774704,
      "learning_rate": 1.8658136238563041e-06,
      "loss": 0.0008,
      "step": 232030
    },
    {
      "epoch": 1.4882324834125957,
      "grad_norm": 0.08898802101612091,
      "learning_rate": 1.8653775504990962e-06,
      "loss": 0.0013,
      "step": 232040
    },
    {
      "epoch": 1.4882966203063819,
      "grad_norm": 0.11834735423326492,
      "learning_rate": 1.8649415164209455e-06,
      "loss": 0.0009,
      "step": 232050
    },
    {
      "epoch": 1.488360757200168,
      "grad_norm": 0.09863437712192535,
      "learning_rate": 1.8645055216273206e-06,
      "loss": 0.001,
      "step": 232060
    },
    {
      "epoch": 1.4884248940939542,
      "grad_norm": 0.07873088121414185,
      "learning_rate": 1.8640695661236818e-06,
      "loss": 0.0014,
      "step": 232070
    },
    {
      "epoch": 1.4884890309877403,
      "grad_norm": 0.030200175940990448,
      "learning_rate": 1.863633649915491e-06,
      "loss": 0.0028,
      "step": 232080
    },
    {
      "epoch": 1.4885531678815263,
      "grad_norm": 0.038778483867645264,
      "learning_rate": 1.863197773008214e-06,
      "loss": 0.001,
      "step": 232090
    },
    {
      "epoch": 1.4886173047753124,
      "grad_norm": 0.06872984766960144,
      "learning_rate": 1.862761935407309e-06,
      "loss": 0.0007,
      "step": 232100
    },
    {
      "epoch": 1.4886814416690985,
      "grad_norm": 0.15356433391571045,
      "learning_rate": 1.8623261371182405e-06,
      "loss": 0.0011,
      "step": 232110
    },
    {
      "epoch": 1.4887455785628847,
      "grad_norm": 0.06912410259246826,
      "learning_rate": 1.861890378146468e-06,
      "loss": 0.0012,
      "step": 232120
    },
    {
      "epoch": 1.4888097154566706,
      "grad_norm": 0.010716703720390797,
      "learning_rate": 1.8614546584974503e-06,
      "loss": 0.0021,
      "step": 232130
    },
    {
      "epoch": 1.4888738523504568,
      "grad_norm": 0.28601473569869995,
      "learning_rate": 1.8610189781766507e-06,
      "loss": 0.0026,
      "step": 232140
    },
    {
      "epoch": 1.488937989244243,
      "grad_norm": 0.0352194719016552,
      "learning_rate": 1.8605833371895266e-06,
      "loss": 0.0009,
      "step": 232150
    },
    {
      "epoch": 1.489002126138029,
      "grad_norm": 0.2125111073255539,
      "learning_rate": 1.860147735541537e-06,
      "loss": 0.001,
      "step": 232160
    },
    {
      "epoch": 1.4890662630318152,
      "grad_norm": 0.08166087418794632,
      "learning_rate": 1.8597121732381384e-06,
      "loss": 0.0007,
      "step": 232170
    },
    {
      "epoch": 1.4891303999256011,
      "grad_norm": 0.08743234723806381,
      "learning_rate": 1.8592766502847925e-06,
      "loss": 0.0012,
      "step": 232180
    },
    {
      "epoch": 1.4891945368193873,
      "grad_norm": 0.06199329346418381,
      "learning_rate": 1.8588411666869544e-06,
      "loss": 0.0008,
      "step": 232190
    },
    {
      "epoch": 1.4892586737131734,
      "grad_norm": 0.023580171167850494,
      "learning_rate": 1.8584057224500801e-06,
      "loss": 0.0007,
      "step": 232200
    },
    {
      "epoch": 1.4893228106069594,
      "grad_norm": 0.01461642887443304,
      "learning_rate": 1.8579703175796293e-06,
      "loss": 0.0012,
      "step": 232210
    },
    {
      "epoch": 1.4893869475007455,
      "grad_norm": 0.007897169329226017,
      "learning_rate": 1.8575349520810553e-06,
      "loss": 0.0013,
      "step": 232220
    },
    {
      "epoch": 1.4894510843945317,
      "grad_norm": 0.04625585302710533,
      "learning_rate": 1.8570996259598152e-06,
      "loss": 0.0013,
      "step": 232230
    },
    {
      "epoch": 1.4895152212883178,
      "grad_norm": 0.05158587545156479,
      "learning_rate": 1.8566643392213607e-06,
      "loss": 0.0009,
      "step": 232240
    },
    {
      "epoch": 1.489579358182104,
      "grad_norm": 0.045665670186281204,
      "learning_rate": 1.8562290918711502e-06,
      "loss": 0.0006,
      "step": 232250
    },
    {
      "epoch": 1.4896434950758901,
      "grad_norm": 0.003189841518178582,
      "learning_rate": 1.8557938839146362e-06,
      "loss": 0.0026,
      "step": 232260
    },
    {
      "epoch": 1.489707631969676,
      "grad_norm": 0.16319698095321655,
      "learning_rate": 1.8553587153572716e-06,
      "loss": 0.0016,
      "step": 232270
    },
    {
      "epoch": 1.4897717688634622,
      "grad_norm": 0.17188839614391327,
      "learning_rate": 1.8549235862045085e-06,
      "loss": 0.0015,
      "step": 232280
    },
    {
      "epoch": 1.4898359057572483,
      "grad_norm": 0.09171821922063828,
      "learning_rate": 1.8544884964618021e-06,
      "loss": 0.0013,
      "step": 232290
    },
    {
      "epoch": 1.4899000426510343,
      "grad_norm": 0.060288283973932266,
      "learning_rate": 1.8540534461346027e-06,
      "loss": 0.0014,
      "step": 232300
    },
    {
      "epoch": 1.4899641795448204,
      "grad_norm": 0.05129498243331909,
      "learning_rate": 1.853618435228362e-06,
      "loss": 0.0013,
      "step": 232310
    },
    {
      "epoch": 1.4900283164386066,
      "grad_norm": 0.11541414260864258,
      "learning_rate": 1.8531834637485297e-06,
      "loss": 0.0021,
      "step": 232320
    },
    {
      "epoch": 1.4900924533323927,
      "grad_norm": 0.08661103248596191,
      "learning_rate": 1.852748531700559e-06,
      "loss": 0.0017,
      "step": 232330
    },
    {
      "epoch": 1.4901565902261789,
      "grad_norm": 0.05976423621177673,
      "learning_rate": 1.8523136390898982e-06,
      "loss": 0.0017,
      "step": 232340
    },
    {
      "epoch": 1.4902207271199648,
      "grad_norm": 0.11782398074865341,
      "learning_rate": 1.851878785921996e-06,
      "loss": 0.0012,
      "step": 232350
    },
    {
      "epoch": 1.490284864013751,
      "grad_norm": 0.03229362145066261,
      "learning_rate": 1.8514439722023043e-06,
      "loss": 0.0017,
      "step": 232360
    },
    {
      "epoch": 1.490349000907537,
      "grad_norm": 0.20834103226661682,
      "learning_rate": 1.8510091979362693e-06,
      "loss": 0.002,
      "step": 232370
    },
    {
      "epoch": 1.490413137801323,
      "grad_norm": 0.08213939517736435,
      "learning_rate": 1.85057446312934e-06,
      "loss": 0.0009,
      "step": 232380
    },
    {
      "epoch": 1.4904772746951092,
      "grad_norm": 0.06685581803321838,
      "learning_rate": 1.850139767786962e-06,
      "loss": 0.0009,
      "step": 232390
    },
    {
      "epoch": 1.4905414115888953,
      "grad_norm": 0.1658812165260315,
      "learning_rate": 1.8497051119145854e-06,
      "loss": 0.0011,
      "step": 232400
    },
    {
      "epoch": 1.4906055484826815,
      "grad_norm": 0.10321488976478577,
      "learning_rate": 1.849270495517655e-06,
      "loss": 0.0012,
      "step": 232410
    },
    {
      "epoch": 1.4906696853764676,
      "grad_norm": 0.11038286238908768,
      "learning_rate": 1.8488359186016168e-06,
      "loss": 0.0012,
      "step": 232420
    },
    {
      "epoch": 1.4907338222702538,
      "grad_norm": 0.1400638371706009,
      "learning_rate": 1.848401381171916e-06,
      "loss": 0.002,
      "step": 232430
    },
    {
      "epoch": 1.4907979591640397,
      "grad_norm": 0.006335919257253408,
      "learning_rate": 1.8479668832339992e-06,
      "loss": 0.0012,
      "step": 232440
    },
    {
      "epoch": 1.4908620960578258,
      "grad_norm": 0.1237395852804184,
      "learning_rate": 1.8475324247933103e-06,
      "loss": 0.0014,
      "step": 232450
    },
    {
      "epoch": 1.490926232951612,
      "grad_norm": 0.046299826353788376,
      "learning_rate": 1.847098005855293e-06,
      "loss": 0.0024,
      "step": 232460
    },
    {
      "epoch": 1.490990369845398,
      "grad_norm": 0.0490373857319355,
      "learning_rate": 1.8466636264253895e-06,
      "loss": 0.001,
      "step": 232470
    },
    {
      "epoch": 1.491054506739184,
      "grad_norm": 0.09268473088741302,
      "learning_rate": 1.8462292865090458e-06,
      "loss": 0.0014,
      "step": 232480
    },
    {
      "epoch": 1.4911186436329702,
      "grad_norm": 0.07769180834293365,
      "learning_rate": 1.8457949861117035e-06,
      "loss": 0.0016,
      "step": 232490
    },
    {
      "epoch": 1.4911827805267563,
      "grad_norm": 0.03023562766611576,
      "learning_rate": 1.8453607252388022e-06,
      "loss": 0.0027,
      "step": 232500
    },
    {
      "epoch": 1.4912469174205425,
      "grad_norm": 0.1785176545381546,
      "learning_rate": 1.8449265038957882e-06,
      "loss": 0.0019,
      "step": 232510
    },
    {
      "epoch": 1.4913110543143284,
      "grad_norm": 0.02365567721426487,
      "learning_rate": 1.844492322088099e-06,
      "loss": 0.0009,
      "step": 232520
    },
    {
      "epoch": 1.4913751912081146,
      "grad_norm": 0.02931872569024563,
      "learning_rate": 1.8440581798211765e-06,
      "loss": 0.0008,
      "step": 232530
    },
    {
      "epoch": 1.4914393281019007,
      "grad_norm": 0.13341720402240753,
      "learning_rate": 1.8436240771004587e-06,
      "loss": 0.0012,
      "step": 232540
    },
    {
      "epoch": 1.4915034649956869,
      "grad_norm": 0.043740373104810715,
      "learning_rate": 1.843190013931389e-06,
      "loss": 0.0007,
      "step": 232550
    },
    {
      "epoch": 1.4915676018894728,
      "grad_norm": 0.13154718279838562,
      "learning_rate": 1.8427559903194049e-06,
      "loss": 0.0014,
      "step": 232560
    },
    {
      "epoch": 1.491631738783259,
      "grad_norm": 0.0338272899389267,
      "learning_rate": 1.8423220062699442e-06,
      "loss": 0.0009,
      "step": 232570
    },
    {
      "epoch": 1.491695875677045,
      "grad_norm": 0.06865895539522171,
      "learning_rate": 1.8418880617884438e-06,
      "loss": 0.0013,
      "step": 232580
    },
    {
      "epoch": 1.4917600125708312,
      "grad_norm": 0.033184465020895004,
      "learning_rate": 1.841454156880345e-06,
      "loss": 0.0009,
      "step": 232590
    },
    {
      "epoch": 1.4918241494646174,
      "grad_norm": 0.021861722692847252,
      "learning_rate": 1.8410202915510827e-06,
      "loss": 0.0007,
      "step": 232600
    },
    {
      "epoch": 1.4918882863584033,
      "grad_norm": 0.028117230162024498,
      "learning_rate": 1.840586465806094e-06,
      "loss": 0.0009,
      "step": 232610
    },
    {
      "epoch": 1.4919524232521895,
      "grad_norm": 0.035108163952827454,
      "learning_rate": 1.8401526796508135e-06,
      "loss": 0.0005,
      "step": 232620
    },
    {
      "epoch": 1.4920165601459756,
      "grad_norm": 0.1248382031917572,
      "learning_rate": 1.8397189330906794e-06,
      "loss": 0.0017,
      "step": 232630
    },
    {
      "epoch": 1.4920806970397615,
      "grad_norm": 0.08763181418180466,
      "learning_rate": 1.839285226131126e-06,
      "loss": 0.0008,
      "step": 232640
    },
    {
      "epoch": 1.4921448339335477,
      "grad_norm": 0.016087012365460396,
      "learning_rate": 1.8388515587775857e-06,
      "loss": 0.002,
      "step": 232650
    },
    {
      "epoch": 1.4922089708273338,
      "grad_norm": 0.06415554136037827,
      "learning_rate": 1.8384179310354966e-06,
      "loss": 0.0007,
      "step": 232660
    },
    {
      "epoch": 1.49227310772112,
      "grad_norm": 0.0232497900724411,
      "learning_rate": 1.83798434291029e-06,
      "loss": 0.0013,
      "step": 232670
    },
    {
      "epoch": 1.4923372446149061,
      "grad_norm": 0.10639127343893051,
      "learning_rate": 1.8375507944074e-06,
      "loss": 0.0012,
      "step": 232680
    },
    {
      "epoch": 1.4924013815086923,
      "grad_norm": 0.03844868019223213,
      "learning_rate": 1.8371172855322567e-06,
      "loss": 0.0022,
      "step": 232690
    },
    {
      "epoch": 1.4924655184024782,
      "grad_norm": 0.05413614958524704,
      "learning_rate": 1.8366838162902956e-06,
      "loss": 0.001,
      "step": 232700
    },
    {
      "epoch": 1.4925296552962644,
      "grad_norm": 0.0037933976855129004,
      "learning_rate": 1.8362503866869474e-06,
      "loss": 0.0012,
      "step": 232710
    },
    {
      "epoch": 1.4925937921900505,
      "grad_norm": 0.0512174591422081,
      "learning_rate": 1.8358169967276435e-06,
      "loss": 0.0011,
      "step": 232720
    },
    {
      "epoch": 1.4926579290838364,
      "grad_norm": 0.04273045063018799,
      "learning_rate": 1.8353836464178115e-06,
      "loss": 0.0008,
      "step": 232730
    },
    {
      "epoch": 1.4927220659776226,
      "grad_norm": 0.03175722062587738,
      "learning_rate": 1.834950335762886e-06,
      "loss": 0.0013,
      "step": 232740
    },
    {
      "epoch": 1.4927862028714087,
      "grad_norm": 0.07337773591279984,
      "learning_rate": 1.8345170647682953e-06,
      "loss": 0.0009,
      "step": 232750
    },
    {
      "epoch": 1.4928503397651949,
      "grad_norm": 0.03609747067093849,
      "learning_rate": 1.8340838334394679e-06,
      "loss": 0.0012,
      "step": 232760
    },
    {
      "epoch": 1.492914476658981,
      "grad_norm": 0.08420764654874802,
      "learning_rate": 1.8336506417818306e-06,
      "loss": 0.0016,
      "step": 232770
    },
    {
      "epoch": 1.492978613552767,
      "grad_norm": 0.11893970519304276,
      "learning_rate": 1.833217489800816e-06,
      "loss": 0.0025,
      "step": 232780
    },
    {
      "epoch": 1.493042750446553,
      "grad_norm": 0.023968994617462158,
      "learning_rate": 1.8327843775018495e-06,
      "loss": 0.001,
      "step": 232790
    },
    {
      "epoch": 1.4931068873403393,
      "grad_norm": 0.13853110373020172,
      "learning_rate": 1.8323513048903563e-06,
      "loss": 0.0013,
      "step": 232800
    },
    {
      "epoch": 1.4931710242341254,
      "grad_norm": 0.04194265604019165,
      "learning_rate": 1.8319182719717665e-06,
      "loss": 0.001,
      "step": 232810
    },
    {
      "epoch": 1.4932351611279113,
      "grad_norm": 0.028974315151572227,
      "learning_rate": 1.8314852787515053e-06,
      "loss": 0.0007,
      "step": 232820
    },
    {
      "epoch": 1.4932992980216975,
      "grad_norm": 0.01592678390443325,
      "learning_rate": 1.8310523252349982e-06,
      "loss": 0.0019,
      "step": 232830
    },
    {
      "epoch": 1.4933634349154836,
      "grad_norm": 0.024703364819288254,
      "learning_rate": 1.8306194114276687e-06,
      "loss": 0.002,
      "step": 232840
    },
    {
      "epoch": 1.4934275718092698,
      "grad_norm": 0.13141685724258423,
      "learning_rate": 1.8301865373349448e-06,
      "loss": 0.0014,
      "step": 232850
    },
    {
      "epoch": 1.493491708703056,
      "grad_norm": 0.10000346601009369,
      "learning_rate": 1.8297537029622487e-06,
      "loss": 0.0012,
      "step": 232860
    },
    {
      "epoch": 1.4935558455968418,
      "grad_norm": 0.021517250686883926,
      "learning_rate": 1.8293209083150027e-06,
      "loss": 0.0008,
      "step": 232870
    },
    {
      "epoch": 1.493619982490628,
      "grad_norm": 0.032985907047986984,
      "learning_rate": 1.8288881533986335e-06,
      "loss": 0.0012,
      "step": 232880
    },
    {
      "epoch": 1.4936841193844141,
      "grad_norm": 0.17923562228679657,
      "learning_rate": 1.8284554382185604e-06,
      "loss": 0.0019,
      "step": 232890
    },
    {
      "epoch": 1.4937482562782,
      "grad_norm": 0.048135485500097275,
      "learning_rate": 1.8280227627802093e-06,
      "loss": 0.0011,
      "step": 232900
    },
    {
      "epoch": 1.4938123931719862,
      "grad_norm": 0.2491879016160965,
      "learning_rate": 1.827590127089e-06,
      "loss": 0.0008,
      "step": 232910
    },
    {
      "epoch": 1.4938765300657724,
      "grad_norm": 0.06684473156929016,
      "learning_rate": 1.8271575311503514e-06,
      "loss": 0.0005,
      "step": 232920
    },
    {
      "epoch": 1.4939406669595585,
      "grad_norm": 0.13278910517692566,
      "learning_rate": 1.8267249749696886e-06,
      "loss": 0.0012,
      "step": 232930
    },
    {
      "epoch": 1.4940048038533447,
      "grad_norm": 0.016812479123473167,
      "learning_rate": 1.82629245855243e-06,
      "loss": 0.0019,
      "step": 232940
    },
    {
      "epoch": 1.4940689407471308,
      "grad_norm": 0.0660853311419487,
      "learning_rate": 1.8258599819039952e-06,
      "loss": 0.0007,
      "step": 232950
    },
    {
      "epoch": 1.4941330776409167,
      "grad_norm": 0.05519381910562515,
      "learning_rate": 1.825427545029801e-06,
      "loss": 0.0006,
      "step": 232960
    },
    {
      "epoch": 1.4941972145347029,
      "grad_norm": 0.04533018544316292,
      "learning_rate": 1.8249951479352707e-06,
      "loss": 0.0006,
      "step": 232970
    },
    {
      "epoch": 1.494261351428489,
      "grad_norm": 0.05164116993546486,
      "learning_rate": 1.8245627906258201e-06,
      "loss": 0.0023,
      "step": 232980
    },
    {
      "epoch": 1.494325488322275,
      "grad_norm": 0.012422630563378334,
      "learning_rate": 1.8241304731068671e-06,
      "loss": 0.0008,
      "step": 232990
    },
    {
      "epoch": 1.4943896252160611,
      "grad_norm": 0.17956890165805817,
      "learning_rate": 1.8236981953838278e-06,
      "loss": 0.0011,
      "step": 233000
    },
    {
      "epoch": 1.4944537621098473,
      "grad_norm": 0.12385495007038116,
      "learning_rate": 1.8232659574621215e-06,
      "loss": 0.0009,
      "step": 233010
    },
    {
      "epoch": 1.4945178990036334,
      "grad_norm": 0.14306782186031342,
      "learning_rate": 1.8228337593471635e-06,
      "loss": 0.0016,
      "step": 233020
    },
    {
      "epoch": 1.4945820358974196,
      "grad_norm": 0.05540962889790535,
      "learning_rate": 1.8224016010443685e-06,
      "loss": 0.0006,
      "step": 233030
    },
    {
      "epoch": 1.4946461727912055,
      "grad_norm": 0.18952268362045288,
      "learning_rate": 1.8219694825591511e-06,
      "loss": 0.0007,
      "step": 233040
    },
    {
      "epoch": 1.4947103096849916,
      "grad_norm": 0.2383899837732315,
      "learning_rate": 1.8215374038969292e-06,
      "loss": 0.0015,
      "step": 233050
    },
    {
      "epoch": 1.4947744465787778,
      "grad_norm": 0.1607988029718399,
      "learning_rate": 1.8211053650631155e-06,
      "loss": 0.0008,
      "step": 233060
    },
    {
      "epoch": 1.4948385834725637,
      "grad_norm": 0.04241831228137016,
      "learning_rate": 1.820673366063121e-06,
      "loss": 0.0021,
      "step": 233070
    },
    {
      "epoch": 1.4949027203663499,
      "grad_norm": 0.10005103796720505,
      "learning_rate": 1.820241406902364e-06,
      "loss": 0.0019,
      "step": 233080
    },
    {
      "epoch": 1.494966857260136,
      "grad_norm": 0.04470842704176903,
      "learning_rate": 1.8198094875862543e-06,
      "loss": 0.0014,
      "step": 233090
    },
    {
      "epoch": 1.4950309941539222,
      "grad_norm": 0.021665040403604507,
      "learning_rate": 1.8193776081202048e-06,
      "loss": 0.0009,
      "step": 233100
    },
    {
      "epoch": 1.4950951310477083,
      "grad_norm": 0.057689227163791656,
      "learning_rate": 1.8189457685096246e-06,
      "loss": 0.0027,
      "step": 233110
    },
    {
      "epoch": 1.4951592679414945,
      "grad_norm": 0.02833256684243679,
      "learning_rate": 1.8185139687599297e-06,
      "loss": 0.0011,
      "step": 233120
    },
    {
      "epoch": 1.4952234048352804,
      "grad_norm": 0.0563499815762043,
      "learning_rate": 1.8180822088765288e-06,
      "loss": 0.001,
      "step": 233130
    },
    {
      "epoch": 1.4952875417290665,
      "grad_norm": 0.016010211780667305,
      "learning_rate": 1.8176504888648312e-06,
      "loss": 0.0008,
      "step": 233140
    },
    {
      "epoch": 1.4953516786228527,
      "grad_norm": 0.042862121015787125,
      "learning_rate": 1.8172188087302461e-06,
      "loss": 0.0014,
      "step": 233150
    },
    {
      "epoch": 1.4954158155166386,
      "grad_norm": 0.057495709508657455,
      "learning_rate": 1.8167871684781856e-06,
      "loss": 0.001,
      "step": 233160
    },
    {
      "epoch": 1.4954799524104248,
      "grad_norm": 0.08322770148515701,
      "learning_rate": 1.816355568114057e-06,
      "loss": 0.0011,
      "step": 233170
    },
    {
      "epoch": 1.495544089304211,
      "grad_norm": 0.04599379003047943,
      "learning_rate": 1.8159240076432682e-06,
      "loss": 0.0012,
      "step": 233180
    },
    {
      "epoch": 1.495608226197997,
      "grad_norm": 0.08682698011398315,
      "learning_rate": 1.815492487071226e-06,
      "loss": 0.0013,
      "step": 233190
    },
    {
      "epoch": 1.4956723630917832,
      "grad_norm": 0.08469253778457642,
      "learning_rate": 1.8150610064033404e-06,
      "loss": 0.0016,
      "step": 233200
    },
    {
      "epoch": 1.4957364999855691,
      "grad_norm": 0.0031433776021003723,
      "learning_rate": 1.8146295656450164e-06,
      "loss": 0.0011,
      "step": 233210
    },
    {
      "epoch": 1.4958006368793553,
      "grad_norm": 0.029707908630371094,
      "learning_rate": 1.8141981648016587e-06,
      "loss": 0.0007,
      "step": 233220
    },
    {
      "epoch": 1.4958647737731414,
      "grad_norm": 0.08981744199991226,
      "learning_rate": 1.8137668038786765e-06,
      "loss": 0.0024,
      "step": 233230
    },
    {
      "epoch": 1.4959289106669276,
      "grad_norm": 0.29700136184692383,
      "learning_rate": 1.8133354828814736e-06,
      "loss": 0.0013,
      "step": 233240
    },
    {
      "epoch": 1.4959930475607135,
      "grad_norm": 0.019079823046922684,
      "learning_rate": 1.8129042018154546e-06,
      "loss": 0.0015,
      "step": 233250
    },
    {
      "epoch": 1.4960571844544996,
      "grad_norm": 0.09006282687187195,
      "learning_rate": 1.8124729606860225e-06,
      "loss": 0.0008,
      "step": 233260
    },
    {
      "epoch": 1.4961213213482858,
      "grad_norm": 0.13939876854419708,
      "learning_rate": 1.8120417594985834e-06,
      "loss": 0.001,
      "step": 233270
    },
    {
      "epoch": 1.496185458242072,
      "grad_norm": 0.1402837634086609,
      "learning_rate": 1.8116105982585397e-06,
      "loss": 0.0012,
      "step": 233280
    },
    {
      "epoch": 1.496249595135858,
      "grad_norm": 0.1734987199306488,
      "learning_rate": 1.8111794769712937e-06,
      "loss": 0.0038,
      "step": 233290
    },
    {
      "epoch": 1.496313732029644,
      "grad_norm": 0.05005556717514992,
      "learning_rate": 1.8107483956422462e-06,
      "loss": 0.0024,
      "step": 233300
    },
    {
      "epoch": 1.4963778689234302,
      "grad_norm": 0.1206803098320961,
      "learning_rate": 1.8103173542768026e-06,
      "loss": 0.0017,
      "step": 233310
    },
    {
      "epoch": 1.4964420058172163,
      "grad_norm": 0.14100918173789978,
      "learning_rate": 1.8098863528803616e-06,
      "loss": 0.0013,
      "step": 233320
    },
    {
      "epoch": 1.4965061427110022,
      "grad_norm": 0.16625139117240906,
      "learning_rate": 1.8094553914583235e-06,
      "loss": 0.001,
      "step": 233330
    },
    {
      "epoch": 1.4965702796047884,
      "grad_norm": 0.019594913348555565,
      "learning_rate": 1.8090244700160909e-06,
      "loss": 0.0007,
      "step": 233340
    },
    {
      "epoch": 1.4966344164985745,
      "grad_norm": 0.04821164533495903,
      "learning_rate": 1.8085935885590627e-06,
      "loss": 0.0012,
      "step": 233350
    },
    {
      "epoch": 1.4966985533923607,
      "grad_norm": 0.04124406352639198,
      "learning_rate": 1.8081627470926372e-06,
      "loss": 0.0033,
      "step": 233360
    },
    {
      "epoch": 1.4967626902861468,
      "grad_norm": 0.08549986779689789,
      "learning_rate": 1.807731945622212e-06,
      "loss": 0.0015,
      "step": 233370
    },
    {
      "epoch": 1.496826827179933,
      "grad_norm": 0.07066275924444199,
      "learning_rate": 1.807301184153189e-06,
      "loss": 0.0008,
      "step": 233380
    },
    {
      "epoch": 1.496890964073719,
      "grad_norm": 0.00854248832911253,
      "learning_rate": 1.806870462690964e-06,
      "loss": 0.0028,
      "step": 233390
    },
    {
      "epoch": 1.496955100967505,
      "grad_norm": 0.10337699204683304,
      "learning_rate": 1.806439781240934e-06,
      "loss": 0.0007,
      "step": 233400
    },
    {
      "epoch": 1.4970192378612912,
      "grad_norm": 0.06711836904287338,
      "learning_rate": 1.806009139808494e-06,
      "loss": 0.0011,
      "step": 233410
    },
    {
      "epoch": 1.4970833747550771,
      "grad_norm": 0.07099798321723938,
      "learning_rate": 1.8055785383990442e-06,
      "loss": 0.0016,
      "step": 233420
    },
    {
      "epoch": 1.4971475116488633,
      "grad_norm": 0.00854500662535429,
      "learning_rate": 1.805147977017978e-06,
      "loss": 0.0011,
      "step": 233430
    },
    {
      "epoch": 1.4972116485426494,
      "grad_norm": 0.026742344722151756,
      "learning_rate": 1.8047174556706914e-06,
      "loss": 0.0013,
      "step": 233440
    },
    {
      "epoch": 1.4972757854364356,
      "grad_norm": 0.00517117977142334,
      "learning_rate": 1.804286974362577e-06,
      "loss": 0.0013,
      "step": 233450
    },
    {
      "epoch": 1.4973399223302217,
      "grad_norm": 0.08385056257247925,
      "learning_rate": 1.803856533099032e-06,
      "loss": 0.0014,
      "step": 233460
    },
    {
      "epoch": 1.4974040592240077,
      "grad_norm": 0.06410916149616241,
      "learning_rate": 1.803426131885449e-06,
      "loss": 0.0014,
      "step": 233470
    },
    {
      "epoch": 1.4974681961177938,
      "grad_norm": 0.16342325508594513,
      "learning_rate": 1.80299577072722e-06,
      "loss": 0.0014,
      "step": 233480
    },
    {
      "epoch": 1.49753233301158,
      "grad_norm": 0.12051145732402802,
      "learning_rate": 1.80256544962974e-06,
      "loss": 0.0014,
      "step": 233490
    },
    {
      "epoch": 1.4975964699053659,
      "grad_norm": 0.10693907737731934,
      "learning_rate": 1.8021351685984001e-06,
      "loss": 0.001,
      "step": 233500
    },
    {
      "epoch": 1.497660606799152,
      "grad_norm": 0.12433069199323654,
      "learning_rate": 1.8017049276385924e-06,
      "loss": 0.0015,
      "step": 233510
    },
    {
      "epoch": 1.4977247436929382,
      "grad_norm": 0.023863285779953003,
      "learning_rate": 1.8012747267557057e-06,
      "loss": 0.0018,
      "step": 233520
    },
    {
      "epoch": 1.4977888805867243,
      "grad_norm": 0.040423374623060226,
      "learning_rate": 1.8008445659551344e-06,
      "loss": 0.002,
      "step": 233530
    },
    {
      "epoch": 1.4978530174805105,
      "grad_norm": 0.109131820499897,
      "learning_rate": 1.8004144452422672e-06,
      "loss": 0.0013,
      "step": 233540
    },
    {
      "epoch": 1.4979171543742966,
      "grad_norm": 0.11498875170946121,
      "learning_rate": 1.799984364622494e-06,
      "loss": 0.0013,
      "step": 233550
    },
    {
      "epoch": 1.4979812912680825,
      "grad_norm": 0.07260105013847351,
      "learning_rate": 1.7995543241012014e-06,
      "loss": 0.0011,
      "step": 233560
    },
    {
      "epoch": 1.4980454281618687,
      "grad_norm": 0.11735733598470688,
      "learning_rate": 1.799124323683783e-06,
      "loss": 0.0018,
      "step": 233570
    },
    {
      "epoch": 1.4981095650556548,
      "grad_norm": 0.06831369549036026,
      "learning_rate": 1.7986943633756238e-06,
      "loss": 0.0008,
      "step": 233580
    },
    {
      "epoch": 1.4981737019494408,
      "grad_norm": 0.045636653900146484,
      "learning_rate": 1.7982644431821122e-06,
      "loss": 0.0016,
      "step": 233590
    },
    {
      "epoch": 1.498237838843227,
      "grad_norm": 0.12789089977741241,
      "learning_rate": 1.7978345631086342e-06,
      "loss": 0.0007,
      "step": 233600
    },
    {
      "epoch": 1.498301975737013,
      "grad_norm": 0.0596468448638916,
      "learning_rate": 1.797404723160579e-06,
      "loss": 0.0017,
      "step": 233610
    },
    {
      "epoch": 1.4983661126307992,
      "grad_norm": 0.05858698487281799,
      "learning_rate": 1.7969749233433314e-06,
      "loss": 0.0012,
      "step": 233620
    },
    {
      "epoch": 1.4984302495245854,
      "grad_norm": 0.4808373749256134,
      "learning_rate": 1.7965451636622755e-06,
      "loss": 0.0013,
      "step": 233630
    },
    {
      "epoch": 1.4984943864183713,
      "grad_norm": 0.058651797473430634,
      "learning_rate": 1.7961154441227986e-06,
      "loss": 0.001,
      "step": 233640
    },
    {
      "epoch": 1.4985585233121574,
      "grad_norm": 0.14847874641418457,
      "learning_rate": 1.795685764730286e-06,
      "loss": 0.0014,
      "step": 233650
    },
    {
      "epoch": 1.4986226602059436,
      "grad_norm": 0.030869372189044952,
      "learning_rate": 1.7952561254901214e-06,
      "loss": 0.0009,
      "step": 233660
    },
    {
      "epoch": 1.4986867970997297,
      "grad_norm": 0.0621880367398262,
      "learning_rate": 1.7948265264076876e-06,
      "loss": 0.0006,
      "step": 233670
    },
    {
      "epoch": 1.4987509339935157,
      "grad_norm": 0.14163510501384735,
      "learning_rate": 1.7943969674883666e-06,
      "loss": 0.0007,
      "step": 233680
    },
    {
      "epoch": 1.4988150708873018,
      "grad_norm": 0.032575275748968124,
      "learning_rate": 1.7939674487375442e-06,
      "loss": 0.0008,
      "step": 233690
    },
    {
      "epoch": 1.498879207781088,
      "grad_norm": 0.012831765227019787,
      "learning_rate": 1.793537970160601e-06,
      "loss": 0.0017,
      "step": 233700
    },
    {
      "epoch": 1.4989433446748741,
      "grad_norm": 0.20057491958141327,
      "learning_rate": 1.7931085317629187e-06,
      "loss": 0.002,
      "step": 233710
    },
    {
      "epoch": 1.4990074815686603,
      "grad_norm": 0.1405409276485443,
      "learning_rate": 1.7926791335498767e-06,
      "loss": 0.0021,
      "step": 233720
    },
    {
      "epoch": 1.4990716184624462,
      "grad_norm": 0.028645657002925873,
      "learning_rate": 1.7922497755268592e-06,
      "loss": 0.001,
      "step": 233730
    },
    {
      "epoch": 1.4991357553562323,
      "grad_norm": 0.02900051139295101,
      "learning_rate": 1.7918204576992443e-06,
      "loss": 0.001,
      "step": 233740
    },
    {
      "epoch": 1.4991998922500185,
      "grad_norm": 0.013951257802546024,
      "learning_rate": 1.7913911800724104e-06,
      "loss": 0.0007,
      "step": 233750
    },
    {
      "epoch": 1.4992640291438044,
      "grad_norm": 0.044111721217632294,
      "learning_rate": 1.7909619426517399e-06,
      "loss": 0.0007,
      "step": 233760
    },
    {
      "epoch": 1.4993281660375906,
      "grad_norm": 0.02437959797680378,
      "learning_rate": 1.7905327454426096e-06,
      "loss": 0.0008,
      "step": 233770
    },
    {
      "epoch": 1.4993923029313767,
      "grad_norm": 0.15801259875297546,
      "learning_rate": 1.7901035884503976e-06,
      "loss": 0.0015,
      "step": 233780
    },
    {
      "epoch": 1.4994564398251629,
      "grad_norm": 0.12411099672317505,
      "learning_rate": 1.7896744716804805e-06,
      "loss": 0.0008,
      "step": 233790
    },
    {
      "epoch": 1.499520576718949,
      "grad_norm": 0.0033227538224309683,
      "learning_rate": 1.7892453951382383e-06,
      "loss": 0.0006,
      "step": 233800
    },
    {
      "epoch": 1.4995847136127352,
      "grad_norm": 0.046695925295352936,
      "learning_rate": 1.7888163588290453e-06,
      "loss": 0.0015,
      "step": 233810
    },
    {
      "epoch": 1.499648850506521,
      "grad_norm": 0.12680582702159882,
      "learning_rate": 1.7883873627582794e-06,
      "loss": 0.0016,
      "step": 233820
    },
    {
      "epoch": 1.4997129874003072,
      "grad_norm": 0.2100486159324646,
      "learning_rate": 1.7879584069313127e-06,
      "loss": 0.0031,
      "step": 233830
    },
    {
      "epoch": 1.4997771242940934,
      "grad_norm": 0.06719649583101273,
      "learning_rate": 1.7875294913535246e-06,
      "loss": 0.0013,
      "step": 233840
    },
    {
      "epoch": 1.4998412611878793,
      "grad_norm": 0.016559133306145668,
      "learning_rate": 1.787100616030288e-06,
      "loss": 0.0009,
      "step": 233850
    },
    {
      "epoch": 1.4999053980816655,
      "grad_norm": 0.0677509531378746,
      "learning_rate": 1.7866717809669765e-06,
      "loss": 0.0021,
      "step": 233860
    },
    {
      "epoch": 1.4999695349754516,
      "grad_norm": 0.17736130952835083,
      "learning_rate": 1.7862429861689628e-06,
      "loss": 0.0012,
      "step": 233870
    },
    {
      "epoch": 1.5000336718692378,
      "grad_norm": 0.06214691326022148,
      "learning_rate": 1.7858142316416232e-06,
      "loss": 0.001,
      "step": 233880
    },
    {
      "epoch": 1.500097808763024,
      "grad_norm": 0.08927460014820099,
      "learning_rate": 1.7853855173903284e-06,
      "loss": 0.0014,
      "step": 233890
    },
    {
      "epoch": 1.50016194565681,
      "grad_norm": 0.075447678565979,
      "learning_rate": 1.7849568434204484e-06,
      "loss": 0.0011,
      "step": 233900
    },
    {
      "epoch": 1.500226082550596,
      "grad_norm": 0.007112096529453993,
      "learning_rate": 1.7845282097373589e-06,
      "loss": 0.001,
      "step": 233910
    },
    {
      "epoch": 1.5002902194443821,
      "grad_norm": 0.028644679114222527,
      "learning_rate": 1.7840996163464285e-06,
      "loss": 0.0013,
      "step": 233920
    },
    {
      "epoch": 1.500354356338168,
      "grad_norm": 0.03432578966021538,
      "learning_rate": 1.783671063253029e-06,
      "loss": 0.0016,
      "step": 233930
    },
    {
      "epoch": 1.5004184932319542,
      "grad_norm": 0.03648257628083229,
      "learning_rate": 1.7832425504625278e-06,
      "loss": 0.0012,
      "step": 233940
    },
    {
      "epoch": 1.5004826301257403,
      "grad_norm": 0.05791352316737175,
      "learning_rate": 1.7828140779802984e-06,
      "loss": 0.001,
      "step": 233950
    },
    {
      "epoch": 1.5005467670195265,
      "grad_norm": 0.13585078716278076,
      "learning_rate": 1.7823856458117072e-06,
      "loss": 0.001,
      "step": 233960
    },
    {
      "epoch": 1.5006109039133126,
      "grad_norm": 0.007192340213805437,
      "learning_rate": 1.7819572539621239e-06,
      "loss": 0.0008,
      "step": 233970
    },
    {
      "epoch": 1.5006750408070988,
      "grad_norm": 0.18648450076580048,
      "learning_rate": 1.7815289024369142e-06,
      "loss": 0.0013,
      "step": 233980
    },
    {
      "epoch": 1.5007391777008847,
      "grad_norm": 0.0246658306568861,
      "learning_rate": 1.7811005912414498e-06,
      "loss": 0.0007,
      "step": 233990
    },
    {
      "epoch": 1.5008033145946709,
      "grad_norm": 0.09747783094644547,
      "learning_rate": 1.7806723203810949e-06,
      "loss": 0.0015,
      "step": 234000
    },
    {
      "epoch": 1.500867451488457,
      "grad_norm": 0.12793777883052826,
      "learning_rate": 1.7802440898612166e-06,
      "loss": 0.0022,
      "step": 234010
    },
    {
      "epoch": 1.500931588382243,
      "grad_norm": 0.0568283386528492,
      "learning_rate": 1.7798158996871795e-06,
      "loss": 0.0022,
      "step": 234020
    },
    {
      "epoch": 1.500995725276029,
      "grad_norm": 0.056288767606019974,
      "learning_rate": 1.7793877498643524e-06,
      "loss": 0.0028,
      "step": 234030
    },
    {
      "epoch": 1.5010598621698152,
      "grad_norm": 0.27575936913490295,
      "learning_rate": 1.7789596403980985e-06,
      "loss": 0.0013,
      "step": 234040
    },
    {
      "epoch": 1.5011239990636014,
      "grad_norm": 0.01003996655344963,
      "learning_rate": 1.7785315712937805e-06,
      "loss": 0.0025,
      "step": 234050
    },
    {
      "epoch": 1.5011881359573875,
      "grad_norm": 0.07622134685516357,
      "learning_rate": 1.778103542556766e-06,
      "loss": 0.0015,
      "step": 234060
    },
    {
      "epoch": 1.5012522728511737,
      "grad_norm": 0.004882059525698423,
      "learning_rate": 1.7776755541924169e-06,
      "loss": 0.0006,
      "step": 234070
    },
    {
      "epoch": 1.5013164097449596,
      "grad_norm": 0.1184699684381485,
      "learning_rate": 1.7772476062060963e-06,
      "loss": 0.0022,
      "step": 234080
    },
    {
      "epoch": 1.5013805466387458,
      "grad_norm": 0.18747399747371674,
      "learning_rate": 1.7768196986031643e-06,
      "loss": 0.0018,
      "step": 234090
    },
    {
      "epoch": 1.5014446835325317,
      "grad_norm": 0.0672893077135086,
      "learning_rate": 1.7763918313889867e-06,
      "loss": 0.0014,
      "step": 234100
    },
    {
      "epoch": 1.5015088204263178,
      "grad_norm": 0.04679718241095543,
      "learning_rate": 1.7759640045689235e-06,
      "loss": 0.0014,
      "step": 234110
    },
    {
      "epoch": 1.501572957320104,
      "grad_norm": 0.051002953201532364,
      "learning_rate": 1.775536218148336e-06,
      "loss": 0.001,
      "step": 234120
    },
    {
      "epoch": 1.5016370942138901,
      "grad_norm": 0.0018901376752182841,
      "learning_rate": 1.7751084721325818e-06,
      "loss": 0.001,
      "step": 234130
    },
    {
      "epoch": 1.5017012311076763,
      "grad_norm": 0.04492555931210518,
      "learning_rate": 1.774680766527025e-06,
      "loss": 0.0013,
      "step": 234140
    },
    {
      "epoch": 1.5017653680014624,
      "grad_norm": 0.00817918125540018,
      "learning_rate": 1.7742531013370235e-06,
      "loss": 0.002,
      "step": 234150
    },
    {
      "epoch": 1.5018295048952486,
      "grad_norm": 0.0407526008784771,
      "learning_rate": 1.7738254765679364e-06,
      "loss": 0.0026,
      "step": 234160
    },
    {
      "epoch": 1.5018936417890345,
      "grad_norm": 0.020076848566532135,
      "learning_rate": 1.7733978922251193e-06,
      "loss": 0.0015,
      "step": 234170
    },
    {
      "epoch": 1.5019577786828207,
      "grad_norm": 0.17962588369846344,
      "learning_rate": 1.7729703483139348e-06,
      "loss": 0.0013,
      "step": 234180
    },
    {
      "epoch": 1.5020219155766066,
      "grad_norm": 0.06566172093153,
      "learning_rate": 1.772542844839738e-06,
      "loss": 0.0015,
      "step": 234190
    },
    {
      "epoch": 1.5020860524703927,
      "grad_norm": 0.02670532651245594,
      "learning_rate": 1.7721153818078845e-06,
      "loss": 0.0006,
      "step": 234200
    },
    {
      "epoch": 1.5021501893641789,
      "grad_norm": 0.13443441689014435,
      "learning_rate": 1.771687959223734e-06,
      "loss": 0.0012,
      "step": 234210
    },
    {
      "epoch": 1.502214326257965,
      "grad_norm": 0.12669919431209564,
      "learning_rate": 1.77126057709264e-06,
      "loss": 0.0008,
      "step": 234220
    },
    {
      "epoch": 1.5022784631517512,
      "grad_norm": 0.020429085940122604,
      "learning_rate": 1.7708332354199592e-06,
      "loss": 0.0006,
      "step": 234230
    },
    {
      "epoch": 1.5023426000455373,
      "grad_norm": 0.2921810448169708,
      "learning_rate": 1.7704059342110435e-06,
      "loss": 0.0013,
      "step": 234240
    },
    {
      "epoch": 1.5024067369393233,
      "grad_norm": 0.33811965584754944,
      "learning_rate": 1.7699786734712515e-06,
      "loss": 0.0034,
      "step": 234250
    },
    {
      "epoch": 1.5024708738331094,
      "grad_norm": 0.1620689481496811,
      "learning_rate": 1.7695514532059355e-06,
      "loss": 0.0011,
      "step": 234260
    },
    {
      "epoch": 1.5025350107268953,
      "grad_norm": 0.05258363485336304,
      "learning_rate": 1.7691242734204483e-06,
      "loss": 0.0019,
      "step": 234270
    },
    {
      "epoch": 1.5025991476206815,
      "grad_norm": 0.0416279137134552,
      "learning_rate": 1.7686971341201413e-06,
      "loss": 0.0017,
      "step": 234280
    },
    {
      "epoch": 1.5026632845144676,
      "grad_norm": 0.10482791811227798,
      "learning_rate": 1.7682700353103704e-06,
      "loss": 0.0013,
      "step": 234290
    },
    {
      "epoch": 1.5027274214082538,
      "grad_norm": 0.08230151236057281,
      "learning_rate": 1.7678429769964856e-06,
      "loss": 0.0011,
      "step": 234300
    },
    {
      "epoch": 1.50279155830204,
      "grad_norm": 0.2138132005929947,
      "learning_rate": 1.7674159591838385e-06,
      "loss": 0.0012,
      "step": 234310
    },
    {
      "epoch": 1.502855695195826,
      "grad_norm": 0.3787125051021576,
      "learning_rate": 1.7669889818777774e-06,
      "loss": 0.0009,
      "step": 234320
    },
    {
      "epoch": 1.5029198320896122,
      "grad_norm": 0.08036356419324875,
      "learning_rate": 1.7665620450836574e-06,
      "loss": 0.0008,
      "step": 234330
    },
    {
      "epoch": 1.5029839689833981,
      "grad_norm": 0.02135232463479042,
      "learning_rate": 1.7661351488068257e-06,
      "loss": 0.0005,
      "step": 234340
    },
    {
      "epoch": 1.5030481058771843,
      "grad_norm": 0.055322859436273575,
      "learning_rate": 1.7657082930526298e-06,
      "loss": 0.0014,
      "step": 234350
    },
    {
      "epoch": 1.5031122427709702,
      "grad_norm": 0.05045409873127937,
      "learning_rate": 1.7652814778264222e-06,
      "loss": 0.0014,
      "step": 234360
    },
    {
      "epoch": 1.5031763796647564,
      "grad_norm": 0.03834830969572067,
      "learning_rate": 1.7648547031335495e-06,
      "loss": 0.002,
      "step": 234370
    },
    {
      "epoch": 1.5032405165585425,
      "grad_norm": 0.006903606466948986,
      "learning_rate": 1.7644279689793598e-06,
      "loss": 0.0013,
      "step": 234380
    },
    {
      "epoch": 1.5033046534523287,
      "grad_norm": 0.07117222249507904,
      "learning_rate": 1.7640012753691982e-06,
      "loss": 0.0008,
      "step": 234390
    },
    {
      "epoch": 1.5033687903461148,
      "grad_norm": 0.012395719066262245,
      "learning_rate": 1.7635746223084132e-06,
      "loss": 0.0016,
      "step": 234400
    },
    {
      "epoch": 1.503432927239901,
      "grad_norm": 0.07572619616985321,
      "learning_rate": 1.7631480098023535e-06,
      "loss": 0.0015,
      "step": 234410
    },
    {
      "epoch": 1.503497064133687,
      "grad_norm": 0.0958796814084053,
      "learning_rate": 1.7627214378563618e-06,
      "loss": 0.0008,
      "step": 234420
    },
    {
      "epoch": 1.503561201027473,
      "grad_norm": 0.02184971049427986,
      "learning_rate": 1.7622949064757843e-06,
      "loss": 0.0007,
      "step": 234430
    },
    {
      "epoch": 1.5036253379212592,
      "grad_norm": 0.179481640458107,
      "learning_rate": 1.7618684156659643e-06,
      "loss": 0.0013,
      "step": 234440
    },
    {
      "epoch": 1.5036894748150451,
      "grad_norm": 0.0073411292396485806,
      "learning_rate": 1.761441965432249e-06,
      "loss": 0.0004,
      "step": 234450
    },
    {
      "epoch": 1.5037536117088313,
      "grad_norm": 0.16384795308113098,
      "learning_rate": 1.7610155557799803e-06,
      "loss": 0.0022,
      "step": 234460
    },
    {
      "epoch": 1.5038177486026174,
      "grad_norm": 0.057359084486961365,
      "learning_rate": 1.7605891867144997e-06,
      "loss": 0.0031,
      "step": 234470
    },
    {
      "epoch": 1.5038818854964036,
      "grad_norm": 0.04206176847219467,
      "learning_rate": 1.7601628582411535e-06,
      "loss": 0.001,
      "step": 234480
    },
    {
      "epoch": 1.5039460223901897,
      "grad_norm": 0.00203709676861763,
      "learning_rate": 1.7597365703652819e-06,
      "loss": 0.0012,
      "step": 234490
    },
    {
      "epoch": 1.5040101592839759,
      "grad_norm": 0.059611234813928604,
      "learning_rate": 1.7593103230922276e-06,
      "loss": 0.0015,
      "step": 234500
    },
    {
      "epoch": 1.5040742961777618,
      "grad_norm": 0.04611333832144737,
      "learning_rate": 1.7588841164273285e-06,
      "loss": 0.0007,
      "step": 234510
    },
    {
      "epoch": 1.504138433071548,
      "grad_norm": 0.02621953934431076,
      "learning_rate": 1.7584579503759303e-06,
      "loss": 0.001,
      "step": 234520
    },
    {
      "epoch": 1.5042025699653339,
      "grad_norm": 0.05761010944843292,
      "learning_rate": 1.7580318249433698e-06,
      "loss": 0.0009,
      "step": 234530
    },
    {
      "epoch": 1.50426670685912,
      "grad_norm": 0.03204973414540291,
      "learning_rate": 1.7576057401349882e-06,
      "loss": 0.0012,
      "step": 234540
    },
    {
      "epoch": 1.5043308437529062,
      "grad_norm": 0.02957121655344963,
      "learning_rate": 1.7571796959561221e-06,
      "loss": 0.0009,
      "step": 234550
    },
    {
      "epoch": 1.5043949806466923,
      "grad_norm": 0.1492205262184143,
      "learning_rate": 1.7567536924121137e-06,
      "loss": 0.0019,
      "step": 234560
    },
    {
      "epoch": 1.5044591175404785,
      "grad_norm": 0.049612388014793396,
      "learning_rate": 1.7563277295082993e-06,
      "loss": 0.0008,
      "step": 234570
    },
    {
      "epoch": 1.5045232544342646,
      "grad_norm": 0.267138808965683,
      "learning_rate": 1.7559018072500173e-06,
      "loss": 0.002,
      "step": 234580
    },
    {
      "epoch": 1.5045873913280507,
      "grad_norm": 0.060233224183321,
      "learning_rate": 1.7554759256426017e-06,
      "loss": 0.0032,
      "step": 234590
    },
    {
      "epoch": 1.5046515282218367,
      "grad_norm": 0.017759747803211212,
      "learning_rate": 1.7550500846913937e-06,
      "loss": 0.0023,
      "step": 234600
    },
    {
      "epoch": 1.5047156651156228,
      "grad_norm": 0.040159065276384354,
      "learning_rate": 1.7546242844017269e-06,
      "loss": 0.0004,
      "step": 234610
    },
    {
      "epoch": 1.5047798020094088,
      "grad_norm": 0.14336469769477844,
      "learning_rate": 1.7541985247789362e-06,
      "loss": 0.0012,
      "step": 234620
    },
    {
      "epoch": 1.504843938903195,
      "grad_norm": 0.023215042427182198,
      "learning_rate": 1.7537728058283588e-06,
      "loss": 0.0023,
      "step": 234630
    },
    {
      "epoch": 1.504908075796981,
      "grad_norm": 0.008424768224358559,
      "learning_rate": 1.7533471275553281e-06,
      "loss": 0.0012,
      "step": 234640
    },
    {
      "epoch": 1.5049722126907672,
      "grad_norm": 0.008528662845492363,
      "learning_rate": 1.7529214899651787e-06,
      "loss": 0.0005,
      "step": 234650
    },
    {
      "epoch": 1.5050363495845533,
      "grad_norm": 0.027422424405813217,
      "learning_rate": 1.7524958930632418e-06,
      "loss": 0.0012,
      "step": 234660
    },
    {
      "epoch": 1.5051004864783395,
      "grad_norm": 0.037562474608421326,
      "learning_rate": 1.7520703368548541e-06,
      "loss": 0.0007,
      "step": 234670
    },
    {
      "epoch": 1.5051646233721254,
      "grad_norm": 0.08821986615657806,
      "learning_rate": 1.7516448213453463e-06,
      "loss": 0.0013,
      "step": 234680
    },
    {
      "epoch": 1.5052287602659116,
      "grad_norm": 0.044554661959409714,
      "learning_rate": 1.7512193465400506e-06,
      "loss": 0.0016,
      "step": 234690
    },
    {
      "epoch": 1.5052928971596975,
      "grad_norm": 0.031345486640930176,
      "learning_rate": 1.7507939124442963e-06,
      "loss": 0.003,
      "step": 234700
    },
    {
      "epoch": 1.5053570340534836,
      "grad_norm": 0.07827392965555191,
      "learning_rate": 1.7503685190634185e-06,
      "loss": 0.0018,
      "step": 234710
    },
    {
      "epoch": 1.5054211709472698,
      "grad_norm": 0.013680531643331051,
      "learning_rate": 1.7499431664027456e-06,
      "loss": 0.0012,
      "step": 234720
    },
    {
      "epoch": 1.505485307841056,
      "grad_norm": 0.05149161443114281,
      "learning_rate": 1.7495178544676072e-06,
      "loss": 0.0011,
      "step": 234730
    },
    {
      "epoch": 1.505549444734842,
      "grad_norm": 0.1659402996301651,
      "learning_rate": 1.7490925832633315e-06,
      "loss": 0.0013,
      "step": 234740
    },
    {
      "epoch": 1.5056135816286282,
      "grad_norm": 0.06993673741817474,
      "learning_rate": 1.7486673527952513e-06,
      "loss": 0.0019,
      "step": 234750
    },
    {
      "epoch": 1.5056777185224144,
      "grad_norm": 0.10557802021503448,
      "learning_rate": 1.7482421630686925e-06,
      "loss": 0.0011,
      "step": 234760
    },
    {
      "epoch": 1.5057418554162003,
      "grad_norm": 0.005402317736297846,
      "learning_rate": 1.7478170140889822e-06,
      "loss": 0.0006,
      "step": 234770
    },
    {
      "epoch": 1.5058059923099865,
      "grad_norm": 0.07126693427562714,
      "learning_rate": 1.7473919058614502e-06,
      "loss": 0.0008,
      "step": 234780
    },
    {
      "epoch": 1.5058701292037724,
      "grad_norm": 0.05289481580257416,
      "learning_rate": 1.7469668383914223e-06,
      "loss": 0.0008,
      "step": 234790
    },
    {
      "epoch": 1.5059342660975585,
      "grad_norm": 0.09493552893400192,
      "learning_rate": 1.7465418116842248e-06,
      "loss": 0.0013,
      "step": 234800
    },
    {
      "epoch": 1.5059984029913447,
      "grad_norm": 0.10266318917274475,
      "learning_rate": 1.7461168257451822e-06,
      "loss": 0.0008,
      "step": 234810
    },
    {
      "epoch": 1.5060625398851308,
      "grad_norm": 0.04841642081737518,
      "learning_rate": 1.7456918805796224e-06,
      "loss": 0.0014,
      "step": 234820
    },
    {
      "epoch": 1.506126676778917,
      "grad_norm": 0.15532329678535461,
      "learning_rate": 1.7452669761928697e-06,
      "loss": 0.002,
      "step": 234830
    },
    {
      "epoch": 1.5061908136727031,
      "grad_norm": 0.10063052922487259,
      "learning_rate": 1.7448421125902476e-06,
      "loss": 0.0008,
      "step": 234840
    },
    {
      "epoch": 1.506254950566489,
      "grad_norm": 0.05639910325407982,
      "learning_rate": 1.744417289777079e-06,
      "loss": 0.0022,
      "step": 234850
    },
    {
      "epoch": 1.5063190874602752,
      "grad_norm": 0.04638287425041199,
      "learning_rate": 1.74399250775869e-06,
      "loss": 0.0009,
      "step": 234860
    },
    {
      "epoch": 1.5063832243540614,
      "grad_norm": 0.056501440703868866,
      "learning_rate": 1.7435677665404016e-06,
      "loss": 0.001,
      "step": 234870
    },
    {
      "epoch": 1.5064473612478473,
      "grad_norm": 0.056153517216444016,
      "learning_rate": 1.7431430661275367e-06,
      "loss": 0.0015,
      "step": 234880
    },
    {
      "epoch": 1.5065114981416334,
      "grad_norm": 0.0634610652923584,
      "learning_rate": 1.7427184065254149e-06,
      "loss": 0.0008,
      "step": 234890
    },
    {
      "epoch": 1.5065756350354196,
      "grad_norm": 0.11776232719421387,
      "learning_rate": 1.7422937877393613e-06,
      "loss": 0.0023,
      "step": 234900
    },
    {
      "epoch": 1.5066397719292057,
      "grad_norm": 0.0619511753320694,
      "learning_rate": 1.7418692097746949e-06,
      "loss": 0.0012,
      "step": 234910
    },
    {
      "epoch": 1.5067039088229919,
      "grad_norm": 0.10542446374893188,
      "learning_rate": 1.741444672636734e-06,
      "loss": 0.0015,
      "step": 234920
    },
    {
      "epoch": 1.506768045716778,
      "grad_norm": 0.023263636976480484,
      "learning_rate": 1.7410201763308022e-06,
      "loss": 0.0012,
      "step": 234930
    },
    {
      "epoch": 1.506832182610564,
      "grad_norm": 0.028178084641695023,
      "learning_rate": 1.7405957208622158e-06,
      "loss": 0.001,
      "step": 234940
    },
    {
      "epoch": 1.50689631950435,
      "grad_norm": 0.08219774812459946,
      "learning_rate": 1.7401713062362957e-06,
      "loss": 0.0011,
      "step": 234950
    },
    {
      "epoch": 1.506960456398136,
      "grad_norm": 0.16498440504074097,
      "learning_rate": 1.7397469324583567e-06,
      "loss": 0.0009,
      "step": 234960
    },
    {
      "epoch": 1.5070245932919222,
      "grad_norm": 0.013587296940386295,
      "learning_rate": 1.7393225995337198e-06,
      "loss": 0.0016,
      "step": 234970
    },
    {
      "epoch": 1.5070887301857083,
      "grad_norm": 0.051530029624700546,
      "learning_rate": 1.7388983074677018e-06,
      "loss": 0.0006,
      "step": 234980
    },
    {
      "epoch": 1.5071528670794945,
      "grad_norm": 0.04435497522354126,
      "learning_rate": 1.7384740562656188e-06,
      "loss": 0.0007,
      "step": 234990
    },
    {
      "epoch": 1.5072170039732806,
      "grad_norm": 0.08541511744260788,
      "learning_rate": 1.7380498459327844e-06,
      "loss": 0.0017,
      "step": 235000
    },
    {
      "epoch": 1.5072811408670668,
      "grad_norm": 0.0650327205657959,
      "learning_rate": 1.7376256764745191e-06,
      "loss": 0.0013,
      "step": 235010
    },
    {
      "epoch": 1.507345277760853,
      "grad_norm": 0.06552031636238098,
      "learning_rate": 1.7372015478961357e-06,
      "loss": 0.0011,
      "step": 235020
    },
    {
      "epoch": 1.5074094146546388,
      "grad_norm": 0.051747698336839676,
      "learning_rate": 1.736777460202948e-06,
      "loss": 0.0015,
      "step": 235030
    },
    {
      "epoch": 1.507473551548425,
      "grad_norm": 0.06316728889942169,
      "learning_rate": 1.7363534134002702e-06,
      "loss": 0.0021,
      "step": 235040
    },
    {
      "epoch": 1.507537688442211,
      "grad_norm": 0.05629882588982582,
      "learning_rate": 1.7359294074934175e-06,
      "loss": 0.0013,
      "step": 235050
    },
    {
      "epoch": 1.507601825335997,
      "grad_norm": 0.044074613600969315,
      "learning_rate": 1.7355054424877027e-06,
      "loss": 0.0008,
      "step": 235060
    },
    {
      "epoch": 1.5076659622297832,
      "grad_norm": 0.015521244145929813,
      "learning_rate": 1.7350815183884355e-06,
      "loss": 0.0013,
      "step": 235070
    },
    {
      "epoch": 1.5077300991235694,
      "grad_norm": 0.030993226915597916,
      "learning_rate": 1.7346576352009326e-06,
      "loss": 0.0008,
      "step": 235080
    },
    {
      "epoch": 1.5077942360173555,
      "grad_norm": 0.02457059919834137,
      "learning_rate": 1.734233792930503e-06,
      "loss": 0.0007,
      "step": 235090
    },
    {
      "epoch": 1.5078583729111417,
      "grad_norm": 0.14740078151226044,
      "learning_rate": 1.733809991582458e-06,
      "loss": 0.0007,
      "step": 235100
    },
    {
      "epoch": 1.5079225098049276,
      "grad_norm": 0.29353275895118713,
      "learning_rate": 1.7333862311621064e-06,
      "loss": 0.0016,
      "step": 235110
    },
    {
      "epoch": 1.5079866466987137,
      "grad_norm": 0.030130386352539062,
      "learning_rate": 1.7329625116747617e-06,
      "loss": 0.0007,
      "step": 235120
    },
    {
      "epoch": 1.5080507835924997,
      "grad_norm": 0.06130951642990112,
      "learning_rate": 1.7325388331257315e-06,
      "loss": 0.0015,
      "step": 235130
    },
    {
      "epoch": 1.5081149204862858,
      "grad_norm": 0.08484799414873123,
      "learning_rate": 1.7321151955203252e-06,
      "loss": 0.001,
      "step": 235140
    },
    {
      "epoch": 1.508179057380072,
      "grad_norm": 0.09276269376277924,
      "learning_rate": 1.7316915988638488e-06,
      "loss": 0.0014,
      "step": 235150
    },
    {
      "epoch": 1.5082431942738581,
      "grad_norm": 0.06885452568531036,
      "learning_rate": 1.731268043161613e-06,
      "loss": 0.0009,
      "step": 235160
    },
    {
      "epoch": 1.5083073311676443,
      "grad_norm": 0.08210889250040054,
      "learning_rate": 1.7308445284189262e-06,
      "loss": 0.0011,
      "step": 235170
    },
    {
      "epoch": 1.5083714680614304,
      "grad_norm": 0.04074737802147865,
      "learning_rate": 1.7304210546410944e-06,
      "loss": 0.0013,
      "step": 235180
    },
    {
      "epoch": 1.5084356049552166,
      "grad_norm": 0.006532349623739719,
      "learning_rate": 1.7299976218334215e-06,
      "loss": 0.0022,
      "step": 235190
    },
    {
      "epoch": 1.5084997418490025,
      "grad_norm": 0.06377600133419037,
      "learning_rate": 1.729574230001217e-06,
      "loss": 0.0008,
      "step": 235200
    },
    {
      "epoch": 1.5085638787427886,
      "grad_norm": 0.35375145077705383,
      "learning_rate": 1.7291508791497845e-06,
      "loss": 0.0016,
      "step": 235210
    },
    {
      "epoch": 1.5086280156365746,
      "grad_norm": 0.04930626600980759,
      "learning_rate": 1.728727569284429e-06,
      "loss": 0.0008,
      "step": 235220
    },
    {
      "epoch": 1.5086921525303607,
      "grad_norm": 0.1520637422800064,
      "learning_rate": 1.728304300410454e-06,
      "loss": 0.0027,
      "step": 235230
    },
    {
      "epoch": 1.5087562894241469,
      "grad_norm": 0.1196468248963356,
      "learning_rate": 1.7278810725331657e-06,
      "loss": 0.0013,
      "step": 235240
    },
    {
      "epoch": 1.508820426317933,
      "grad_norm": 0.1320398598909378,
      "learning_rate": 1.7274578856578661e-06,
      "loss": 0.0016,
      "step": 235250
    },
    {
      "epoch": 1.5088845632117192,
      "grad_norm": 0.04028046503663063,
      "learning_rate": 1.7270347397898585e-06,
      "loss": 0.0007,
      "step": 235260
    },
    {
      "epoch": 1.5089487001055053,
      "grad_norm": 0.09058906137943268,
      "learning_rate": 1.7266116349344425e-06,
      "loss": 0.0015,
      "step": 235270
    },
    {
      "epoch": 1.5090128369992915,
      "grad_norm": 0.002862320514395833,
      "learning_rate": 1.726188571096924e-06,
      "loss": 0.0016,
      "step": 235280
    },
    {
      "epoch": 1.5090769738930774,
      "grad_norm": 0.03169231116771698,
      "learning_rate": 1.7257655482826024e-06,
      "loss": 0.0015,
      "step": 235290
    },
    {
      "epoch": 1.5091411107868635,
      "grad_norm": 0.0070249466225504875,
      "learning_rate": 1.7253425664967787e-06,
      "loss": 0.0015,
      "step": 235300
    },
    {
      "epoch": 1.5092052476806495,
      "grad_norm": 0.050528980791568756,
      "learning_rate": 1.7249196257447515e-06,
      "loss": 0.0017,
      "step": 235310
    },
    {
      "epoch": 1.5092693845744356,
      "grad_norm": 0.16391852498054504,
      "learning_rate": 1.724496726031823e-06,
      "loss": 0.0014,
      "step": 235320
    },
    {
      "epoch": 1.5093335214682218,
      "grad_norm": 0.10123415291309357,
      "learning_rate": 1.7240738673632923e-06,
      "loss": 0.0023,
      "step": 235330
    },
    {
      "epoch": 1.509397658362008,
      "grad_norm": 0.0017691326793283224,
      "learning_rate": 1.7236510497444552e-06,
      "loss": 0.0009,
      "step": 235340
    },
    {
      "epoch": 1.509461795255794,
      "grad_norm": 0.015997136011719704,
      "learning_rate": 1.7232282731806137e-06,
      "loss": 0.0022,
      "step": 235350
    },
    {
      "epoch": 1.5095259321495802,
      "grad_norm": 0.04225035384297371,
      "learning_rate": 1.722805537677064e-06,
      "loss": 0.0008,
      "step": 235360
    },
    {
      "epoch": 1.5095900690433661,
      "grad_norm": 0.012225965037941933,
      "learning_rate": 1.7223828432391031e-06,
      "loss": 0.0008,
      "step": 235370
    },
    {
      "epoch": 1.5096542059371523,
      "grad_norm": 0.06684717535972595,
      "learning_rate": 1.7219601898720256e-06,
      "loss": 0.0014,
      "step": 235380
    },
    {
      "epoch": 1.5097183428309382,
      "grad_norm": 0.21332567930221558,
      "learning_rate": 1.7215375775811315e-06,
      "loss": 0.0014,
      "step": 235390
    },
    {
      "epoch": 1.5097824797247243,
      "grad_norm": 0.058626167476177216,
      "learning_rate": 1.7211150063717153e-06,
      "loss": 0.0011,
      "step": 235400
    },
    {
      "epoch": 1.5098466166185105,
      "grad_norm": 0.15646931529045105,
      "learning_rate": 1.7206924762490706e-06,
      "loss": 0.0011,
      "step": 235410
    },
    {
      "epoch": 1.5099107535122966,
      "grad_norm": 0.013812965713441372,
      "learning_rate": 1.7202699872184919e-06,
      "loss": 0.0005,
      "step": 235420
    },
    {
      "epoch": 1.5099748904060828,
      "grad_norm": 0.11039616912603378,
      "learning_rate": 1.7198475392852754e-06,
      "loss": 0.001,
      "step": 235430
    },
    {
      "epoch": 1.510039027299869,
      "grad_norm": 0.010939417406916618,
      "learning_rate": 1.7194251324547134e-06,
      "loss": 0.0006,
      "step": 235440
    },
    {
      "epoch": 1.510103164193655,
      "grad_norm": 0.10745838284492493,
      "learning_rate": 1.719002766732098e-06,
      "loss": 0.0015,
      "step": 235450
    },
    {
      "epoch": 1.510167301087441,
      "grad_norm": 0.10931447893381119,
      "learning_rate": 1.7185804421227242e-06,
      "loss": 0.0013,
      "step": 235460
    },
    {
      "epoch": 1.5102314379812272,
      "grad_norm": 0.017885947600007057,
      "learning_rate": 1.7181581586318823e-06,
      "loss": 0.0012,
      "step": 235470
    },
    {
      "epoch": 1.510295574875013,
      "grad_norm": 0.0009909237269312143,
      "learning_rate": 1.717735916264865e-06,
      "loss": 0.0009,
      "step": 235480
    },
    {
      "epoch": 1.5103597117687992,
      "grad_norm": 0.06282038241624832,
      "learning_rate": 1.7173137150269597e-06,
      "loss": 0.0009,
      "step": 235490
    },
    {
      "epoch": 1.5104238486625854,
      "grad_norm": 0.08964784443378448,
      "learning_rate": 1.716891554923462e-06,
      "loss": 0.001,
      "step": 235500
    },
    {
      "epoch": 1.5104879855563715,
      "grad_norm": 0.034846872091293335,
      "learning_rate": 1.7164694359596595e-06,
      "loss": 0.0011,
      "step": 235510
    },
    {
      "epoch": 1.5105521224501577,
      "grad_norm": 0.06375276297330856,
      "learning_rate": 1.7160473581408415e-06,
      "loss": 0.0028,
      "step": 235520
    },
    {
      "epoch": 1.5106162593439438,
      "grad_norm": 0.11003471910953522,
      "learning_rate": 1.7156253214722956e-06,
      "loss": 0.0005,
      "step": 235530
    },
    {
      "epoch": 1.5106803962377298,
      "grad_norm": 0.23813316226005554,
      "learning_rate": 1.7152033259593137e-06,
      "loss": 0.0025,
      "step": 235540
    },
    {
      "epoch": 1.510744533131516,
      "grad_norm": 0.009359259158372879,
      "learning_rate": 1.7147813716071815e-06,
      "loss": 0.0012,
      "step": 235550
    },
    {
      "epoch": 1.510808670025302,
      "grad_norm": 0.08566562831401825,
      "learning_rate": 1.7143594584211864e-06,
      "loss": 0.0016,
      "step": 235560
    },
    {
      "epoch": 1.510872806919088,
      "grad_norm": 0.0824081152677536,
      "learning_rate": 1.7139375864066137e-06,
      "loss": 0.0012,
      "step": 235570
    },
    {
      "epoch": 1.5109369438128741,
      "grad_norm": 0.026241037994623184,
      "learning_rate": 1.7135157555687537e-06,
      "loss": 0.0013,
      "step": 235580
    },
    {
      "epoch": 1.5110010807066603,
      "grad_norm": 0.3644322454929352,
      "learning_rate": 1.7130939659128897e-06,
      "loss": 0.0026,
      "step": 235590
    },
    {
      "epoch": 1.5110652176004464,
      "grad_norm": 0.015625417232513428,
      "learning_rate": 1.712672217444306e-06,
      "loss": 0.0014,
      "step": 235600
    },
    {
      "epoch": 1.5111293544942326,
      "grad_norm": 0.057176608592271805,
      "learning_rate": 1.7122505101682901e-06,
      "loss": 0.0012,
      "step": 235610
    },
    {
      "epoch": 1.5111934913880187,
      "grad_norm": 0.006172089837491512,
      "learning_rate": 1.7118288440901255e-06,
      "loss": 0.0016,
      "step": 235620
    },
    {
      "epoch": 1.5112576282818047,
      "grad_norm": 0.023188453167676926,
      "learning_rate": 1.7114072192150954e-06,
      "loss": 0.0025,
      "step": 235630
    },
    {
      "epoch": 1.5113217651755908,
      "grad_norm": 0.06347131729125977,
      "learning_rate": 1.7109856355484817e-06,
      "loss": 0.001,
      "step": 235640
    },
    {
      "epoch": 1.5113859020693767,
      "grad_norm": 0.04603705182671547,
      "learning_rate": 1.7105640930955697e-06,
      "loss": 0.0005,
      "step": 235650
    },
    {
      "epoch": 1.5114500389631629,
      "grad_norm": 0.15269768238067627,
      "learning_rate": 1.710142591861641e-06,
      "loss": 0.0018,
      "step": 235660
    },
    {
      "epoch": 1.511514175856949,
      "grad_norm": 0.06650374829769135,
      "learning_rate": 1.7097211318519769e-06,
      "loss": 0.0011,
      "step": 235670
    },
    {
      "epoch": 1.5115783127507352,
      "grad_norm": 0.007529903668910265,
      "learning_rate": 1.7092997130718564e-06,
      "loss": 0.0009,
      "step": 235680
    },
    {
      "epoch": 1.5116424496445213,
      "grad_norm": 0.0774153545498848,
      "learning_rate": 1.7088783355265643e-06,
      "loss": 0.003,
      "step": 235690
    },
    {
      "epoch": 1.5117065865383075,
      "grad_norm": 0.06964616477489471,
      "learning_rate": 1.708456999221379e-06,
      "loss": 0.0008,
      "step": 235700
    },
    {
      "epoch": 1.5117707234320936,
      "grad_norm": 0.15337079763412476,
      "learning_rate": 1.7080357041615797e-06,
      "loss": 0.0007,
      "step": 235710
    },
    {
      "epoch": 1.5118348603258795,
      "grad_norm": 0.05863737314939499,
      "learning_rate": 1.707614450352444e-06,
      "loss": 0.0009,
      "step": 235720
    },
    {
      "epoch": 1.5118989972196657,
      "grad_norm": 0.21024541556835175,
      "learning_rate": 1.707193237799254e-06,
      "loss": 0.001,
      "step": 235730
    },
    {
      "epoch": 1.5119631341134516,
      "grad_norm": 0.12702852487564087,
      "learning_rate": 1.7067720665072861e-06,
      "loss": 0.002,
      "step": 235740
    },
    {
      "epoch": 1.5120272710072378,
      "grad_norm": 0.07385962456464767,
      "learning_rate": 1.7063509364818158e-06,
      "loss": 0.001,
      "step": 235750
    },
    {
      "epoch": 1.512091407901024,
      "grad_norm": 0.06825349479913712,
      "learning_rate": 1.7059298477281243e-06,
      "loss": 0.0011,
      "step": 235760
    },
    {
      "epoch": 1.51215554479481,
      "grad_norm": 0.1421101689338684,
      "learning_rate": 1.7055088002514864e-06,
      "loss": 0.001,
      "step": 235770
    },
    {
      "epoch": 1.5122196816885962,
      "grad_norm": 0.08823364228010178,
      "learning_rate": 1.705087794057177e-06,
      "loss": 0.0012,
      "step": 235780
    },
    {
      "epoch": 1.5122838185823824,
      "grad_norm": 0.028474045917391777,
      "learning_rate": 1.704666829150471e-06,
      "loss": 0.0008,
      "step": 235790
    },
    {
      "epoch": 1.5123479554761683,
      "grad_norm": 0.03342844173312187,
      "learning_rate": 1.7042459055366467e-06,
      "loss": 0.0009,
      "step": 235800
    },
    {
      "epoch": 1.5124120923699544,
      "grad_norm": 0.061231207102537155,
      "learning_rate": 1.7038250232209764e-06,
      "loss": 0.001,
      "step": 235810
    },
    {
      "epoch": 1.5124762292637404,
      "grad_norm": 0.04442818462848663,
      "learning_rate": 1.7034041822087343e-06,
      "loss": 0.0012,
      "step": 235820
    },
    {
      "epoch": 1.5125403661575265,
      "grad_norm": 0.14335766434669495,
      "learning_rate": 1.7029833825051922e-06,
      "loss": 0.002,
      "step": 235830
    },
    {
      "epoch": 1.5126045030513127,
      "grad_norm": 0.09852230548858643,
      "learning_rate": 1.7025626241156263e-06,
      "loss": 0.0012,
      "step": 235840
    },
    {
      "epoch": 1.5126686399450988,
      "grad_norm": 0.11228171736001968,
      "learning_rate": 1.7021419070453076e-06,
      "loss": 0.0012,
      "step": 235850
    },
    {
      "epoch": 1.512732776838885,
      "grad_norm": 0.11434180289506912,
      "learning_rate": 1.7017212312995074e-06,
      "loss": 0.0038,
      "step": 235860
    },
    {
      "epoch": 1.5127969137326711,
      "grad_norm": 0.03543776646256447,
      "learning_rate": 1.7013005968834962e-06,
      "loss": 0.001,
      "step": 235870
    },
    {
      "epoch": 1.5128610506264573,
      "grad_norm": 0.04616093263030052,
      "learning_rate": 1.7008800038025469e-06,
      "loss": 0.0034,
      "step": 235880
    },
    {
      "epoch": 1.5129251875202432,
      "grad_norm": 0.06281872093677521,
      "learning_rate": 1.70045945206193e-06,
      "loss": 0.0012,
      "step": 235890
    },
    {
      "epoch": 1.5129893244140293,
      "grad_norm": 0.14692969620227814,
      "learning_rate": 1.7000389416669115e-06,
      "loss": 0.0016,
      "step": 235900
    },
    {
      "epoch": 1.5130534613078153,
      "grad_norm": 0.00830824300646782,
      "learning_rate": 1.699618472622766e-06,
      "loss": 0.0017,
      "step": 235910
    },
    {
      "epoch": 1.5131175982016014,
      "grad_norm": 0.03200173005461693,
      "learning_rate": 1.69919804493476e-06,
      "loss": 0.0007,
      "step": 235920
    },
    {
      "epoch": 1.5131817350953876,
      "grad_norm": 0.11528050899505615,
      "learning_rate": 1.6987776586081595e-06,
      "loss": 0.0009,
      "step": 235930
    },
    {
      "epoch": 1.5132458719891737,
      "grad_norm": 0.06540056318044662,
      "learning_rate": 1.6983573136482355e-06,
      "loss": 0.0015,
      "step": 235940
    },
    {
      "epoch": 1.5133100088829599,
      "grad_norm": 0.1667458862066269,
      "learning_rate": 1.6979370100602533e-06,
      "loss": 0.0022,
      "step": 235950
    },
    {
      "epoch": 1.513374145776746,
      "grad_norm": 0.030666688457131386,
      "learning_rate": 1.6975167478494814e-06,
      "loss": 0.0015,
      "step": 235960
    },
    {
      "epoch": 1.513438282670532,
      "grad_norm": 0.1034374013543129,
      "learning_rate": 1.6970965270211847e-06,
      "loss": 0.0026,
      "step": 235970
    },
    {
      "epoch": 1.513502419564318,
      "grad_norm": 0.15055187046527863,
      "learning_rate": 1.6966763475806297e-06,
      "loss": 0.0016,
      "step": 235980
    },
    {
      "epoch": 1.5135665564581042,
      "grad_norm": 0.00941514316946268,
      "learning_rate": 1.6962562095330787e-06,
      "loss": 0.0014,
      "step": 235990
    },
    {
      "epoch": 1.5136306933518902,
      "grad_norm": 0.11294365674257278,
      "learning_rate": 1.6958361128838002e-06,
      "loss": 0.0018,
      "step": 236000
    },
    {
      "epoch": 1.5136948302456763,
      "grad_norm": 0.02906079776585102,
      "learning_rate": 1.6954160576380568e-06,
      "loss": 0.0011,
      "step": 236010
    },
    {
      "epoch": 1.5137589671394625,
      "grad_norm": 0.058754779398441315,
      "learning_rate": 1.6949960438011104e-06,
      "loss": 0.0007,
      "step": 236020
    },
    {
      "epoch": 1.5138231040332486,
      "grad_norm": 0.10251867771148682,
      "learning_rate": 1.6945760713782273e-06,
      "loss": 0.0013,
      "step": 236030
    },
    {
      "epoch": 1.5138872409270347,
      "grad_norm": 0.008062995970249176,
      "learning_rate": 1.694156140374668e-06,
      "loss": 0.0008,
      "step": 236040
    },
    {
      "epoch": 1.513951377820821,
      "grad_norm": 0.12957392632961273,
      "learning_rate": 1.6937362507956955e-06,
      "loss": 0.0007,
      "step": 236050
    },
    {
      "epoch": 1.5140155147146068,
      "grad_norm": 0.05608009546995163,
      "learning_rate": 1.6933164026465688e-06,
      "loss": 0.0008,
      "step": 236060
    },
    {
      "epoch": 1.514079651608393,
      "grad_norm": 0.07690117508172989,
      "learning_rate": 1.6928965959325527e-06,
      "loss": 0.0012,
      "step": 236070
    },
    {
      "epoch": 1.514143788502179,
      "grad_norm": 0.05481581389904022,
      "learning_rate": 1.6924768306589056e-06,
      "loss": 0.0008,
      "step": 236080
    },
    {
      "epoch": 1.514207925395965,
      "grad_norm": 0.0975133627653122,
      "learning_rate": 1.6920571068308871e-06,
      "loss": 0.0013,
      "step": 236090
    },
    {
      "epoch": 1.5142720622897512,
      "grad_norm": 0.029730889946222305,
      "learning_rate": 1.6916374244537564e-06,
      "loss": 0.0009,
      "step": 236100
    },
    {
      "epoch": 1.5143361991835373,
      "grad_norm": 0.0672646313905716,
      "learning_rate": 1.6912177835327749e-06,
      "loss": 0.0009,
      "step": 236110
    },
    {
      "epoch": 1.5144003360773235,
      "grad_norm": 0.05796928331255913,
      "learning_rate": 1.6907981840731984e-06,
      "loss": 0.0015,
      "step": 236120
    },
    {
      "epoch": 1.5144644729711096,
      "grad_norm": 0.13216884434223175,
      "learning_rate": 1.6903786260802868e-06,
      "loss": 0.0022,
      "step": 236130
    },
    {
      "epoch": 1.5145286098648958,
      "grad_norm": 0.08694811165332794,
      "learning_rate": 1.6899591095592939e-06,
      "loss": 0.0009,
      "step": 236140
    },
    {
      "epoch": 1.5145927467586817,
      "grad_norm": 0.19647616147994995,
      "learning_rate": 1.6895396345154812e-06,
      "loss": 0.0021,
      "step": 236150
    },
    {
      "epoch": 1.5146568836524679,
      "grad_norm": 0.025054756551980972,
      "learning_rate": 1.6891202009541024e-06,
      "loss": 0.0014,
      "step": 236160
    },
    {
      "epoch": 1.5147210205462538,
      "grad_norm": 0.09283973276615143,
      "learning_rate": 1.6887008088804124e-06,
      "loss": 0.0015,
      "step": 236170
    },
    {
      "epoch": 1.51478515744004,
      "grad_norm": 0.07384605705738068,
      "learning_rate": 1.6882814582996693e-06,
      "loss": 0.0014,
      "step": 236180
    },
    {
      "epoch": 1.514849294333826,
      "grad_norm": 0.09666016697883606,
      "learning_rate": 1.6878621492171265e-06,
      "loss": 0.0006,
      "step": 236190
    },
    {
      "epoch": 1.5149134312276122,
      "grad_norm": 0.224246546626091,
      "learning_rate": 1.6874428816380379e-06,
      "loss": 0.0011,
      "step": 236200
    },
    {
      "epoch": 1.5149775681213984,
      "grad_norm": 0.05371362343430519,
      "learning_rate": 1.6870236555676561e-06,
      "loss": 0.0013,
      "step": 236210
    },
    {
      "epoch": 1.5150417050151845,
      "grad_norm": 0.034396693110466,
      "learning_rate": 1.6866044710112366e-06,
      "loss": 0.0013,
      "step": 236220
    },
    {
      "epoch": 1.5151058419089705,
      "grad_norm": 0.16512033343315125,
      "learning_rate": 1.686185327974032e-06,
      "loss": 0.0017,
      "step": 236230
    },
    {
      "epoch": 1.5151699788027566,
      "grad_norm": 0.057726021856069565,
      "learning_rate": 1.6857662264612934e-06,
      "loss": 0.0005,
      "step": 236240
    },
    {
      "epoch": 1.5152341156965425,
      "grad_norm": 0.08658675104379654,
      "learning_rate": 1.6853471664782706e-06,
      "loss": 0.0011,
      "step": 236250
    },
    {
      "epoch": 1.5152982525903287,
      "grad_norm": 0.15499667823314667,
      "learning_rate": 1.6849281480302187e-06,
      "loss": 0.0013,
      "step": 236260
    },
    {
      "epoch": 1.5153623894841148,
      "grad_norm": 0.09169743955135345,
      "learning_rate": 1.684509171122386e-06,
      "loss": 0.0019,
      "step": 236270
    },
    {
      "epoch": 1.515426526377901,
      "grad_norm": 0.005327288061380386,
      "learning_rate": 1.6840902357600235e-06,
      "loss": 0.0012,
      "step": 236280
    },
    {
      "epoch": 1.5154906632716871,
      "grad_norm": 0.04662052541971207,
      "learning_rate": 1.683671341948378e-06,
      "loss": 0.0015,
      "step": 236290
    },
    {
      "epoch": 1.5155548001654733,
      "grad_norm": 0.05670848861336708,
      "learning_rate": 1.6832524896927027e-06,
      "loss": 0.0014,
      "step": 236300
    },
    {
      "epoch": 1.5156189370592594,
      "grad_norm": 0.026139413937926292,
      "learning_rate": 1.6828336789982442e-06,
      "loss": 0.0128,
      "step": 236310
    },
    {
      "epoch": 1.5156830739530454,
      "grad_norm": 0.10478154569864273,
      "learning_rate": 1.6824149098702486e-06,
      "loss": 0.0009,
      "step": 236320
    },
    {
      "epoch": 1.5157472108468315,
      "grad_norm": 0.05814342573285103,
      "learning_rate": 1.6819961823139668e-06,
      "loss": 0.0009,
      "step": 236330
    },
    {
      "epoch": 1.5158113477406174,
      "grad_norm": 0.09379421174526215,
      "learning_rate": 1.6815774963346443e-06,
      "loss": 0.0011,
      "step": 236340
    },
    {
      "epoch": 1.5158754846344036,
      "grad_norm": 0.01812293566763401,
      "learning_rate": 1.6811588519375276e-06,
      "loss": 0.0012,
      "step": 236350
    },
    {
      "epoch": 1.5159396215281897,
      "grad_norm": 0.09073521941900253,
      "learning_rate": 1.6807402491278607e-06,
      "loss": 0.0018,
      "step": 236360
    },
    {
      "epoch": 1.5160037584219759,
      "grad_norm": 0.09301798045635223,
      "learning_rate": 1.6803216879108919e-06,
      "loss": 0.0008,
      "step": 236370
    },
    {
      "epoch": 1.516067895315762,
      "grad_norm": 0.26336193084716797,
      "learning_rate": 1.6799031682918654e-06,
      "loss": 0.0011,
      "step": 236380
    },
    {
      "epoch": 1.5161320322095482,
      "grad_norm": 0.02151043340563774,
      "learning_rate": 1.6794846902760249e-06,
      "loss": 0.0009,
      "step": 236390
    },
    {
      "epoch": 1.516196169103334,
      "grad_norm": 0.12041262537240982,
      "learning_rate": 1.6790662538686126e-06,
      "loss": 0.0009,
      "step": 236400
    },
    {
      "epoch": 1.5162603059971203,
      "grad_norm": 0.059923578053712845,
      "learning_rate": 1.6786478590748756e-06,
      "loss": 0.0008,
      "step": 236410
    },
    {
      "epoch": 1.5163244428909064,
      "grad_norm": 0.053933363407850266,
      "learning_rate": 1.678229505900054e-06,
      "loss": 0.0014,
      "step": 236420
    },
    {
      "epoch": 1.5163885797846923,
      "grad_norm": 0.045569125562906265,
      "learning_rate": 1.6778111943493914e-06,
      "loss": 0.001,
      "step": 236430
    },
    {
      "epoch": 1.5164527166784785,
      "grad_norm": 0.12082313746213913,
      "learning_rate": 1.6773929244281266e-06,
      "loss": 0.002,
      "step": 236440
    },
    {
      "epoch": 1.5165168535722646,
      "grad_norm": 0.1428079605102539,
      "learning_rate": 1.676974696141505e-06,
      "loss": 0.0025,
      "step": 236450
    },
    {
      "epoch": 1.5165809904660508,
      "grad_norm": 0.19562873244285583,
      "learning_rate": 1.6765565094947656e-06,
      "loss": 0.0019,
      "step": 236460
    },
    {
      "epoch": 1.516645127359837,
      "grad_norm": 0.11447757482528687,
      "learning_rate": 1.6761383644931468e-06,
      "loss": 0.0016,
      "step": 236470
    },
    {
      "epoch": 1.516709264253623,
      "grad_norm": 0.10640858113765717,
      "learning_rate": 1.675720261141891e-06,
      "loss": 0.0009,
      "step": 236480
    },
    {
      "epoch": 1.516773401147409,
      "grad_norm": 0.026647871360182762,
      "learning_rate": 1.6753021994462366e-06,
      "loss": 0.0012,
      "step": 236490
    },
    {
      "epoch": 1.5168375380411951,
      "grad_norm": 0.08135776221752167,
      "learning_rate": 1.6748841794114218e-06,
      "loss": 0.0015,
      "step": 236500
    },
    {
      "epoch": 1.516901674934981,
      "grad_norm": 0.02475298009812832,
      "learning_rate": 1.6744662010426828e-06,
      "loss": 0.0015,
      "step": 236510
    },
    {
      "epoch": 1.5169658118287672,
      "grad_norm": 0.026814064010977745,
      "learning_rate": 1.6740482643452605e-06,
      "loss": 0.0028,
      "step": 236520
    },
    {
      "epoch": 1.5170299487225534,
      "grad_norm": 0.08240596950054169,
      "learning_rate": 1.6736303693243904e-06,
      "loss": 0.0006,
      "step": 236530
    },
    {
      "epoch": 1.5170940856163395,
      "grad_norm": 0.0356484018266201,
      "learning_rate": 1.6732125159853097e-06,
      "loss": 0.0013,
      "step": 236540
    },
    {
      "epoch": 1.5171582225101257,
      "grad_norm": 0.06164219230413437,
      "learning_rate": 1.6727947043332515e-06,
      "loss": 0.0013,
      "step": 236550
    },
    {
      "epoch": 1.5172223594039118,
      "grad_norm": 0.03566623479127884,
      "learning_rate": 1.6723769343734552e-06,
      "loss": 0.0013,
      "step": 236560
    },
    {
      "epoch": 1.517286496297698,
      "grad_norm": 0.061721839010715485,
      "learning_rate": 1.6719592061111545e-06,
      "loss": 0.001,
      "step": 236570
    },
    {
      "epoch": 1.5173506331914839,
      "grad_norm": 0.1656349152326584,
      "learning_rate": 1.6715415195515838e-06,
      "loss": 0.0014,
      "step": 236580
    },
    {
      "epoch": 1.51741477008527,
      "grad_norm": 0.053992122411727905,
      "learning_rate": 1.6711238746999742e-06,
      "loss": 0.0007,
      "step": 236590
    },
    {
      "epoch": 1.517478906979056,
      "grad_norm": 0.0747436061501503,
      "learning_rate": 1.6707062715615635e-06,
      "loss": 0.0008,
      "step": 236600
    },
    {
      "epoch": 1.5175430438728421,
      "grad_norm": 0.06978696584701538,
      "learning_rate": 1.6702887101415833e-06,
      "loss": 0.0022,
      "step": 236610
    },
    {
      "epoch": 1.5176071807666283,
      "grad_norm": 0.15435577929019928,
      "learning_rate": 1.6698711904452625e-06,
      "loss": 0.0014,
      "step": 236620
    },
    {
      "epoch": 1.5176713176604144,
      "grad_norm": 0.077399343252182,
      "learning_rate": 1.6694537124778383e-06,
      "loss": 0.0004,
      "step": 236630
    },
    {
      "epoch": 1.5177354545542006,
      "grad_norm": 0.17518320679664612,
      "learning_rate": 1.669036276244539e-06,
      "loss": 0.0021,
      "step": 236640
    },
    {
      "epoch": 1.5177995914479867,
      "grad_norm": 0.047856125980615616,
      "learning_rate": 1.6686188817505955e-06,
      "loss": 0.0102,
      "step": 236650
    },
    {
      "epoch": 1.5178637283417726,
      "grad_norm": 0.06831234693527222,
      "learning_rate": 1.6682015290012372e-06,
      "loss": 0.0018,
      "step": 236660
    },
    {
      "epoch": 1.5179278652355588,
      "grad_norm": 0.048504315316677094,
      "learning_rate": 1.6677842180016963e-06,
      "loss": 0.001,
      "step": 236670
    },
    {
      "epoch": 1.5179920021293447,
      "grad_norm": 0.017820101231336594,
      "learning_rate": 1.6673669487572003e-06,
      "loss": 0.0017,
      "step": 236680
    },
    {
      "epoch": 1.5180561390231309,
      "grad_norm": 0.07297179102897644,
      "learning_rate": 1.6669497212729785e-06,
      "loss": 0.0007,
      "step": 236690
    },
    {
      "epoch": 1.518120275916917,
      "grad_norm": 0.027433795854449272,
      "learning_rate": 1.6665325355542573e-06,
      "loss": 0.0009,
      "step": 236700
    },
    {
      "epoch": 1.5181844128107032,
      "grad_norm": 0.0877075120806694,
      "learning_rate": 1.6661153916062656e-06,
      "loss": 0.0019,
      "step": 236710
    },
    {
      "epoch": 1.5182485497044893,
      "grad_norm": 0.25056546926498413,
      "learning_rate": 1.6656982894342327e-06,
      "loss": 0.001,
      "step": 236720
    },
    {
      "epoch": 1.5183126865982755,
      "grad_norm": 0.14040808379650116,
      "learning_rate": 1.6652812290433829e-06,
      "loss": 0.0012,
      "step": 236730
    },
    {
      "epoch": 1.5183768234920616,
      "grad_norm": 0.15343688428401947,
      "learning_rate": 1.6648642104389413e-06,
      "loss": 0.0009,
      "step": 236740
    },
    {
      "epoch": 1.5184409603858475,
      "grad_norm": 0.13074518740177155,
      "learning_rate": 1.6644472336261363e-06,
      "loss": 0.0013,
      "step": 236750
    },
    {
      "epoch": 1.5185050972796337,
      "grad_norm": 0.03336082026362419,
      "learning_rate": 1.6640302986101908e-06,
      "loss": 0.0033,
      "step": 236760
    },
    {
      "epoch": 1.5185692341734196,
      "grad_norm": 0.10020921379327774,
      "learning_rate": 1.6636134053963304e-06,
      "loss": 0.0011,
      "step": 236770
    },
    {
      "epoch": 1.5186333710672058,
      "grad_norm": 0.1573495715856552,
      "learning_rate": 1.6631965539897765e-06,
      "loss": 0.002,
      "step": 236780
    },
    {
      "epoch": 1.518697507960992,
      "grad_norm": 0.03277795761823654,
      "learning_rate": 1.6627797443957567e-06,
      "loss": 0.0006,
      "step": 236790
    },
    {
      "epoch": 1.518761644854778,
      "grad_norm": 0.03879289701581001,
      "learning_rate": 1.6623629766194914e-06,
      "loss": 0.0022,
      "step": 236800
    },
    {
      "epoch": 1.5188257817485642,
      "grad_norm": 0.13124696910381317,
      "learning_rate": 1.6619462506662037e-06,
      "loss": 0.0012,
      "step": 236810
    },
    {
      "epoch": 1.5188899186423503,
      "grad_norm": 0.08229606598615646,
      "learning_rate": 1.661529566541113e-06,
      "loss": 0.0023,
      "step": 236820
    },
    {
      "epoch": 1.5189540555361365,
      "grad_norm": 0.07102247327566147,
      "learning_rate": 1.661112924249445e-06,
      "loss": 0.0008,
      "step": 236830
    },
    {
      "epoch": 1.5190181924299224,
      "grad_norm": 0.04948734492063522,
      "learning_rate": 1.660696323796418e-06,
      "loss": 0.0017,
      "step": 236840
    },
    {
      "epoch": 1.5190823293237086,
      "grad_norm": 0.013510667718946934,
      "learning_rate": 1.6602797651872527e-06,
      "loss": 0.0008,
      "step": 236850
    },
    {
      "epoch": 1.5191464662174945,
      "grad_norm": 0.0725022703409195,
      "learning_rate": 1.659863248427167e-06,
      "loss": 0.0007,
      "step": 236860
    },
    {
      "epoch": 1.5192106031112806,
      "grad_norm": 0.08922433108091354,
      "learning_rate": 1.6594467735213838e-06,
      "loss": 0.0007,
      "step": 236870
    },
    {
      "epoch": 1.5192747400050668,
      "grad_norm": 0.013453935272991657,
      "learning_rate": 1.6590303404751195e-06,
      "loss": 0.001,
      "step": 236880
    },
    {
      "epoch": 1.519338876898853,
      "grad_norm": 0.0375833734869957,
      "learning_rate": 1.6586139492935922e-06,
      "loss": 0.0015,
      "step": 236890
    },
    {
      "epoch": 1.519403013792639,
      "grad_norm": 0.14216962456703186,
      "learning_rate": 1.658197599982021e-06,
      "loss": 0.0021,
      "step": 236900
    },
    {
      "epoch": 1.5194671506864252,
      "grad_norm": 0.08312832564115524,
      "learning_rate": 1.6577812925456221e-06,
      "loss": 0.0015,
      "step": 236910
    },
    {
      "epoch": 1.5195312875802112,
      "grad_norm": 0.0007432155543938279,
      "learning_rate": 1.6573650269896125e-06,
      "loss": 0.001,
      "step": 236920
    },
    {
      "epoch": 1.5195954244739973,
      "grad_norm": 0.13070756196975708,
      "learning_rate": 1.6569488033192066e-06,
      "loss": 0.001,
      "step": 236930
    },
    {
      "epoch": 1.5196595613677832,
      "grad_norm": 0.016720326617360115,
      "learning_rate": 1.6565326215396227e-06,
      "loss": 0.0016,
      "step": 236940
    },
    {
      "epoch": 1.5197236982615694,
      "grad_norm": 0.03852033615112305,
      "learning_rate": 1.656116481656075e-06,
      "loss": 0.001,
      "step": 236950
    },
    {
      "epoch": 1.5197878351553555,
      "grad_norm": 0.24319837987422943,
      "learning_rate": 1.6557003836737773e-06,
      "loss": 0.0019,
      "step": 236960
    },
    {
      "epoch": 1.5198519720491417,
      "grad_norm": 0.02781999483704567,
      "learning_rate": 1.6552843275979424e-06,
      "loss": 0.0008,
      "step": 236970
    },
    {
      "epoch": 1.5199161089429278,
      "grad_norm": 0.1802973598241806,
      "learning_rate": 1.6548683134337873e-06,
      "loss": 0.0019,
      "step": 236980
    },
    {
      "epoch": 1.519980245836714,
      "grad_norm": 0.02273796685039997,
      "learning_rate": 1.6544523411865226e-06,
      "loss": 0.0015,
      "step": 236990
    },
    {
      "epoch": 1.5200443827305001,
      "grad_norm": 0.03507988527417183,
      "learning_rate": 1.654036410861361e-06,
      "loss": 0.0021,
      "step": 237000
    },
    {
      "epoch": 1.520108519624286,
      "grad_norm": 0.07036600261926651,
      "learning_rate": 1.6536205224635132e-06,
      "loss": 0.0021,
      "step": 237010
    },
    {
      "epoch": 1.5201726565180722,
      "grad_norm": 0.06622333824634552,
      "learning_rate": 1.6532046759981934e-06,
      "loss": 0.0004,
      "step": 237020
    },
    {
      "epoch": 1.5202367934118581,
      "grad_norm": 0.06492827832698822,
      "learning_rate": 1.652788871470611e-06,
      "loss": 0.0011,
      "step": 237030
    },
    {
      "epoch": 1.5203009303056443,
      "grad_norm": 0.09560419619083405,
      "learning_rate": 1.652373108885974e-06,
      "loss": 0.0025,
      "step": 237040
    },
    {
      "epoch": 1.5203650671994304,
      "grad_norm": 0.3649875223636627,
      "learning_rate": 1.6519573882494966e-06,
      "loss": 0.0021,
      "step": 237050
    },
    {
      "epoch": 1.5204292040932166,
      "grad_norm": 0.07574035227298737,
      "learning_rate": 1.6515417095663861e-06,
      "loss": 0.0016,
      "step": 237060
    },
    {
      "epoch": 1.5204933409870027,
      "grad_norm": 0.006303989328444004,
      "learning_rate": 1.651126072841851e-06,
      "loss": 0.0011,
      "step": 237070
    },
    {
      "epoch": 1.5205574778807889,
      "grad_norm": 0.05555268004536629,
      "learning_rate": 1.650710478081098e-06,
      "loss": 0.0011,
      "step": 237080
    },
    {
      "epoch": 1.5206216147745748,
      "grad_norm": 0.33239391446113586,
      "learning_rate": 1.6502949252893374e-06,
      "loss": 0.0026,
      "step": 237090
    },
    {
      "epoch": 1.520685751668361,
      "grad_norm": 0.02349112741649151,
      "learning_rate": 1.6498794144717761e-06,
      "loss": 0.003,
      "step": 237100
    },
    {
      "epoch": 1.520749888562147,
      "grad_norm": 0.13826599717140198,
      "learning_rate": 1.6494639456336198e-06,
      "loss": 0.001,
      "step": 237110
    },
    {
      "epoch": 1.520814025455933,
      "grad_norm": 0.04017220437526703,
      "learning_rate": 1.6490485187800731e-06,
      "loss": 0.001,
      "step": 237120
    },
    {
      "epoch": 1.5208781623497192,
      "grad_norm": 0.05414574220776558,
      "learning_rate": 1.6486331339163452e-06,
      "loss": 0.0017,
      "step": 237130
    },
    {
      "epoch": 1.5209422992435053,
      "grad_norm": 0.03123663365840912,
      "learning_rate": 1.6482177910476393e-06,
      "loss": 0.001,
      "step": 237140
    },
    {
      "epoch": 1.5210064361372915,
      "grad_norm": 0.054575365036726,
      "learning_rate": 1.6478024901791583e-06,
      "loss": 0.0011,
      "step": 237150
    },
    {
      "epoch": 1.5210705730310776,
      "grad_norm": 0.25057604908943176,
      "learning_rate": 1.6473872313161094e-06,
      "loss": 0.0015,
      "step": 237160
    },
    {
      "epoch": 1.5211347099248638,
      "grad_norm": 0.05438575521111488,
      "learning_rate": 1.6469720144636946e-06,
      "loss": 0.0012,
      "step": 237170
    },
    {
      "epoch": 1.5211988468186497,
      "grad_norm": 0.13939109444618225,
      "learning_rate": 1.646556839627117e-06,
      "loss": 0.0018,
      "step": 237180
    },
    {
      "epoch": 1.5212629837124358,
      "grad_norm": 0.19489020109176636,
      "learning_rate": 1.646141706811577e-06,
      "loss": 0.0014,
      "step": 237190
    },
    {
      "epoch": 1.5213271206062218,
      "grad_norm": 0.04001326486468315,
      "learning_rate": 1.64572661602228e-06,
      "loss": 0.0011,
      "step": 237200
    },
    {
      "epoch": 1.521391257500008,
      "grad_norm": 0.07058380544185638,
      "learning_rate": 1.645311567264426e-06,
      "loss": 0.001,
      "step": 237210
    },
    {
      "epoch": 1.521455394393794,
      "grad_norm": 0.14475677907466888,
      "learning_rate": 1.6448965605432155e-06,
      "loss": 0.0017,
      "step": 237220
    },
    {
      "epoch": 1.5215195312875802,
      "grad_norm": 0.20844435691833496,
      "learning_rate": 1.6444815958638471e-06,
      "loss": 0.001,
      "step": 237230
    },
    {
      "epoch": 1.5215836681813664,
      "grad_norm": 0.09307936578989029,
      "learning_rate": 1.6440666732315248e-06,
      "loss": 0.0009,
      "step": 237240
    },
    {
      "epoch": 1.5216478050751525,
      "grad_norm": 0.1734868884086609,
      "learning_rate": 1.6436517926514444e-06,
      "loss": 0.0012,
      "step": 237250
    },
    {
      "epoch": 1.5217119419689387,
      "grad_norm": 0.02707342617213726,
      "learning_rate": 1.6432369541288068e-06,
      "loss": 0.0007,
      "step": 237260
    },
    {
      "epoch": 1.5217760788627246,
      "grad_norm": 0.0013223428977653384,
      "learning_rate": 1.642822157668807e-06,
      "loss": 0.0005,
      "step": 237270
    },
    {
      "epoch": 1.5218402157565107,
      "grad_norm": 0.10929550230503082,
      "learning_rate": 1.6424074032766469e-06,
      "loss": 0.0063,
      "step": 237280
    },
    {
      "epoch": 1.5219043526502967,
      "grad_norm": 0.03251928463578224,
      "learning_rate": 1.6419926909575217e-06,
      "loss": 0.0013,
      "step": 237290
    },
    {
      "epoch": 1.5219684895440828,
      "grad_norm": 0.037062376737594604,
      "learning_rate": 1.641578020716626e-06,
      "loss": 0.0012,
      "step": 237300
    },
    {
      "epoch": 1.522032626437869,
      "grad_norm": 0.14792048931121826,
      "learning_rate": 1.6411633925591597e-06,
      "loss": 0.0027,
      "step": 237310
    },
    {
      "epoch": 1.5220967633316551,
      "grad_norm": 0.04708145186305046,
      "learning_rate": 1.6407488064903165e-06,
      "loss": 0.0022,
      "step": 237320
    },
    {
      "epoch": 1.5221609002254413,
      "grad_norm": 0.025157131254673004,
      "learning_rate": 1.6403342625152918e-06,
      "loss": 0.0014,
      "step": 237330
    },
    {
      "epoch": 1.5222250371192274,
      "grad_norm": 0.04621124267578125,
      "learning_rate": 1.6399197606392787e-06,
      "loss": 0.0006,
      "step": 237340
    },
    {
      "epoch": 1.5222891740130133,
      "grad_norm": 0.18020835518836975,
      "learning_rate": 1.6395053008674732e-06,
      "loss": 0.0018,
      "step": 237350
    },
    {
      "epoch": 1.5223533109067995,
      "grad_norm": 0.02694026753306389,
      "learning_rate": 1.6390908832050683e-06,
      "loss": 0.0023,
      "step": 237360
    },
    {
      "epoch": 1.5224174478005854,
      "grad_norm": 0.021097462624311447,
      "learning_rate": 1.638676507657257e-06,
      "loss": 0.0011,
      "step": 237370
    },
    {
      "epoch": 1.5224815846943716,
      "grad_norm": 0.1032140851020813,
      "learning_rate": 1.6382621742292292e-06,
      "loss": 0.001,
      "step": 237380
    },
    {
      "epoch": 1.5225457215881577,
      "grad_norm": 0.026876067742705345,
      "learning_rate": 1.6378478829261806e-06,
      "loss": 0.001,
      "step": 237390
    },
    {
      "epoch": 1.5226098584819439,
      "grad_norm": 0.042561158537864685,
      "learning_rate": 1.6374336337533015e-06,
      "loss": 0.0004,
      "step": 237400
    },
    {
      "epoch": 1.52267399537573,
      "grad_norm": 0.06443782150745392,
      "learning_rate": 1.637019426715782e-06,
      "loss": 0.0013,
      "step": 237410
    },
    {
      "epoch": 1.5227381322695162,
      "grad_norm": 0.0814862996339798,
      "learning_rate": 1.6366052618188105e-06,
      "loss": 0.0006,
      "step": 237420
    },
    {
      "epoch": 1.5228022691633023,
      "grad_norm": 0.09525315463542938,
      "learning_rate": 1.6361911390675806e-06,
      "loss": 0.0011,
      "step": 237430
    },
    {
      "epoch": 1.5228664060570882,
      "grad_norm": 0.27986252307891846,
      "learning_rate": 1.63577705846728e-06,
      "loss": 0.0005,
      "step": 237440
    },
    {
      "epoch": 1.5229305429508744,
      "grad_norm": 0.043385230004787445,
      "learning_rate": 1.6353630200230951e-06,
      "loss": 0.0012,
      "step": 237450
    },
    {
      "epoch": 1.5229946798446603,
      "grad_norm": 0.05866341292858124,
      "learning_rate": 1.634949023740217e-06,
      "loss": 0.0016,
      "step": 237460
    },
    {
      "epoch": 1.5230588167384465,
      "grad_norm": 0.06352284550666809,
      "learning_rate": 1.6345350696238339e-06,
      "loss": 0.0016,
      "step": 237470
    },
    {
      "epoch": 1.5231229536322326,
      "grad_norm": 0.01987922377884388,
      "learning_rate": 1.6341211576791317e-06,
      "loss": 0.0011,
      "step": 237480
    },
    {
      "epoch": 1.5231870905260187,
      "grad_norm": 0.28827333450317383,
      "learning_rate": 1.633707287911297e-06,
      "loss": 0.0015,
      "step": 237490
    },
    {
      "epoch": 1.523251227419805,
      "grad_norm": 0.018926220014691353,
      "learning_rate": 1.633293460325514e-06,
      "loss": 0.0014,
      "step": 237500
    },
    {
      "epoch": 1.523315364313591,
      "grad_norm": 0.14290401339530945,
      "learning_rate": 1.6328796749269715e-06,
      "loss": 0.0015,
      "step": 237510
    },
    {
      "epoch": 1.523379501207377,
      "grad_norm": 0.040627621114254,
      "learning_rate": 1.6324659317208536e-06,
      "loss": 0.0011,
      "step": 237520
    },
    {
      "epoch": 1.5234436381011631,
      "grad_norm": 0.11878825724124908,
      "learning_rate": 1.6320522307123443e-06,
      "loss": 0.0007,
      "step": 237530
    },
    {
      "epoch": 1.5235077749949493,
      "grad_norm": 0.07322997599840164,
      "learning_rate": 1.6316385719066258e-06,
      "loss": 0.0012,
      "step": 237540
    },
    {
      "epoch": 1.5235719118887352,
      "grad_norm": 0.07629713416099548,
      "learning_rate": 1.631224955308885e-06,
      "loss": 0.0021,
      "step": 237550
    },
    {
      "epoch": 1.5236360487825213,
      "grad_norm": 0.11985758692026138,
      "learning_rate": 1.6308113809243026e-06,
      "loss": 0.0016,
      "step": 237560
    },
    {
      "epoch": 1.5237001856763075,
      "grad_norm": 0.06853505223989487,
      "learning_rate": 1.6303978487580607e-06,
      "loss": 0.0011,
      "step": 237570
    },
    {
      "epoch": 1.5237643225700936,
      "grad_norm": 0.06833287328481674,
      "learning_rate": 1.629984358815343e-06,
      "loss": 0.0005,
      "step": 237580
    },
    {
      "epoch": 1.5238284594638798,
      "grad_norm": 0.07232213020324707,
      "learning_rate": 1.6295709111013302e-06,
      "loss": 0.0006,
      "step": 237590
    },
    {
      "epoch": 1.523892596357666,
      "grad_norm": 0.04098551347851753,
      "learning_rate": 1.6291575056212029e-06,
      "loss": 0.001,
      "step": 237600
    },
    {
      "epoch": 1.5239567332514519,
      "grad_norm": 0.16344477236270905,
      "learning_rate": 1.6287441423801386e-06,
      "loss": 0.0019,
      "step": 237610
    },
    {
      "epoch": 1.524020870145238,
      "grad_norm": 0.003833626862615347,
      "learning_rate": 1.628330821383322e-06,
      "loss": 0.001,
      "step": 237620
    },
    {
      "epoch": 1.524085007039024,
      "grad_norm": 0.13997580111026764,
      "learning_rate": 1.6279175426359295e-06,
      "loss": 0.0017,
      "step": 237630
    },
    {
      "epoch": 1.52414914393281,
      "grad_norm": 0.15628334879875183,
      "learning_rate": 1.6275043061431406e-06,
      "loss": 0.0006,
      "step": 237640
    },
    {
      "epoch": 1.5242132808265962,
      "grad_norm": 0.20656491816043854,
      "learning_rate": 1.6270911119101313e-06,
      "loss": 0.0017,
      "step": 237650
    },
    {
      "epoch": 1.5242774177203824,
      "grad_norm": 0.03471900895237923,
      "learning_rate": 1.6266779599420823e-06,
      "loss": 0.0013,
      "step": 237660
    },
    {
      "epoch": 1.5243415546141685,
      "grad_norm": 0.15333348512649536,
      "learning_rate": 1.6262648502441697e-06,
      "loss": 0.0009,
      "step": 237670
    },
    {
      "epoch": 1.5244056915079547,
      "grad_norm": 0.06840582937002182,
      "learning_rate": 1.6258517828215692e-06,
      "loss": 0.001,
      "step": 237680
    },
    {
      "epoch": 1.5244698284017408,
      "grad_norm": 0.01534581370651722,
      "learning_rate": 1.6254387576794562e-06,
      "loss": 0.0007,
      "step": 237690
    },
    {
      "epoch": 1.5245339652955268,
      "grad_norm": 0.1612095683813095,
      "learning_rate": 1.6250257748230086e-06,
      "loss": 0.0014,
      "step": 237700
    },
    {
      "epoch": 1.524598102189313,
      "grad_norm": 0.06327960640192032,
      "learning_rate": 1.6246128342574002e-06,
      "loss": 0.0022,
      "step": 237710
    },
    {
      "epoch": 1.5246622390830988,
      "grad_norm": 0.03815099224448204,
      "learning_rate": 1.624199935987804e-06,
      "loss": 0.0008,
      "step": 237720
    },
    {
      "epoch": 1.524726375976885,
      "grad_norm": 0.10183555632829666,
      "learning_rate": 1.6237870800193966e-06,
      "loss": 0.0019,
      "step": 237730
    },
    {
      "epoch": 1.5247905128706711,
      "grad_norm": 0.30045872926712036,
      "learning_rate": 1.6233742663573503e-06,
      "loss": 0.0024,
      "step": 237740
    },
    {
      "epoch": 1.5248546497644573,
      "grad_norm": 0.029955746605992317,
      "learning_rate": 1.6229614950068374e-06,
      "loss": 0.001,
      "step": 237750
    },
    {
      "epoch": 1.5249187866582434,
      "grad_norm": 0.01774410903453827,
      "learning_rate": 1.6225487659730288e-06,
      "loss": 0.0026,
      "step": 237760
    },
    {
      "epoch": 1.5249829235520296,
      "grad_norm": 0.07637742906808853,
      "learning_rate": 1.6221360792611002e-06,
      "loss": 0.0009,
      "step": 237770
    },
    {
      "epoch": 1.5250470604458155,
      "grad_norm": 0.05893915146589279,
      "learning_rate": 1.6217234348762202e-06,
      "loss": 0.0016,
      "step": 237780
    },
    {
      "epoch": 1.5251111973396017,
      "grad_norm": 0.025320112705230713,
      "learning_rate": 1.62131083282356e-06,
      "loss": 0.0012,
      "step": 237790
    },
    {
      "epoch": 1.5251753342333876,
      "grad_norm": 0.24433287978172302,
      "learning_rate": 1.6208982731082879e-06,
      "loss": 0.0014,
      "step": 237800
    },
    {
      "epoch": 1.5252394711271737,
      "grad_norm": 0.12436637282371521,
      "learning_rate": 1.6204857557355775e-06,
      "loss": 0.0011,
      "step": 237810
    },
    {
      "epoch": 1.5253036080209599,
      "grad_norm": 0.11953530460596085,
      "learning_rate": 1.6200732807105956e-06,
      "loss": 0.0023,
      "step": 237820
    },
    {
      "epoch": 1.525367744914746,
      "grad_norm": 0.06200635805726051,
      "learning_rate": 1.6196608480385112e-06,
      "loss": 0.0009,
      "step": 237830
    },
    {
      "epoch": 1.5254318818085322,
      "grad_norm": 0.14688628911972046,
      "learning_rate": 1.6192484577244904e-06,
      "loss": 0.0012,
      "step": 237840
    },
    {
      "epoch": 1.5254960187023183,
      "grad_norm": 0.0018314635381102562,
      "learning_rate": 1.6188361097737043e-06,
      "loss": 0.0008,
      "step": 237850
    },
    {
      "epoch": 1.5255601555961045,
      "grad_norm": 0.05635478347539902,
      "learning_rate": 1.6184238041913187e-06,
      "loss": 0.001,
      "step": 237860
    },
    {
      "epoch": 1.5256242924898904,
      "grad_norm": 0.0111836614087224,
      "learning_rate": 1.6180115409824976e-06,
      "loss": 0.0013,
      "step": 237870
    },
    {
      "epoch": 1.5256884293836765,
      "grad_norm": 0.004777786787599325,
      "learning_rate": 1.6175993201524103e-06,
      "loss": 0.0012,
      "step": 237880
    },
    {
      "epoch": 1.5257525662774625,
      "grad_norm": 0.08003873378038406,
      "learning_rate": 1.6171871417062218e-06,
      "loss": 0.0017,
      "step": 237890
    },
    {
      "epoch": 1.5258167031712486,
      "grad_norm": 0.024621382355690002,
      "learning_rate": 1.6167750056490955e-06,
      "loss": 0.0021,
      "step": 237900
    },
    {
      "epoch": 1.5258808400650348,
      "grad_norm": 0.08933726698160172,
      "learning_rate": 1.6163629119861945e-06,
      "loss": 0.0013,
      "step": 237910
    },
    {
      "epoch": 1.525944976958821,
      "grad_norm": 0.038684818893671036,
      "learning_rate": 1.6159508607226865e-06,
      "loss": 0.0009,
      "step": 237920
    },
    {
      "epoch": 1.526009113852607,
      "grad_norm": 0.1163448691368103,
      "learning_rate": 1.6155388518637327e-06,
      "loss": 0.0011,
      "step": 237930
    },
    {
      "epoch": 1.5260732507463932,
      "grad_norm": 0.07280529290437698,
      "learning_rate": 1.6151268854144958e-06,
      "loss": 0.0012,
      "step": 237940
    },
    {
      "epoch": 1.5261373876401791,
      "grad_norm": 0.09436652809381485,
      "learning_rate": 1.614714961380136e-06,
      "loss": 0.0008,
      "step": 237950
    },
    {
      "epoch": 1.5262015245339653,
      "grad_norm": 0.039469748735427856,
      "learning_rate": 1.6143030797658194e-06,
      "loss": 0.0021,
      "step": 237960
    },
    {
      "epoch": 1.5262656614277514,
      "grad_norm": 0.009606624953448772,
      "learning_rate": 1.6138912405767048e-06,
      "loss": 0.0004,
      "step": 237970
    },
    {
      "epoch": 1.5263297983215374,
      "grad_norm": 0.18109820783138275,
      "learning_rate": 1.6134794438179534e-06,
      "loss": 0.002,
      "step": 237980
    },
    {
      "epoch": 1.5263939352153235,
      "grad_norm": 0.1913572996854782,
      "learning_rate": 1.6130676894947228e-06,
      "loss": 0.0026,
      "step": 237990
    },
    {
      "epoch": 1.5264580721091097,
      "grad_norm": 0.027836723253130913,
      "learning_rate": 1.6126559776121764e-06,
      "loss": 0.0008,
      "step": 238000
    },
    {
      "epoch": 1.5265222090028958,
      "grad_norm": 0.11762955784797668,
      "learning_rate": 1.6122443081754713e-06,
      "loss": 0.0009,
      "step": 238010
    },
    {
      "epoch": 1.526586345896682,
      "grad_norm": 0.06743264943361282,
      "learning_rate": 1.611832681189765e-06,
      "loss": 0.0015,
      "step": 238020
    },
    {
      "epoch": 1.526650482790468,
      "grad_norm": 0.005941579584032297,
      "learning_rate": 1.611421096660218e-06,
      "loss": 0.0012,
      "step": 238030
    },
    {
      "epoch": 1.526714619684254,
      "grad_norm": 0.13325421512126923,
      "learning_rate": 1.6110095545919863e-06,
      "loss": 0.0016,
      "step": 238040
    },
    {
      "epoch": 1.5267787565780402,
      "grad_norm": 0.032834045588970184,
      "learning_rate": 1.6105980549902273e-06,
      "loss": 0.0006,
      "step": 238050
    },
    {
      "epoch": 1.5268428934718261,
      "grad_norm": 0.09828424453735352,
      "learning_rate": 1.610186597860095e-06,
      "loss": 0.0011,
      "step": 238060
    },
    {
      "epoch": 1.5269070303656123,
      "grad_norm": 0.11012592911720276,
      "learning_rate": 1.6097751832067492e-06,
      "loss": 0.0016,
      "step": 238070
    },
    {
      "epoch": 1.5269711672593984,
      "grad_norm": 0.10598249733448029,
      "learning_rate": 1.6093638110353432e-06,
      "loss": 0.0011,
      "step": 238080
    },
    {
      "epoch": 1.5270353041531846,
      "grad_norm": 0.007695229258388281,
      "learning_rate": 1.6089524813510321e-06,
      "loss": 0.0009,
      "step": 238090
    },
    {
      "epoch": 1.5270994410469707,
      "grad_norm": 0.2664151191711426,
      "learning_rate": 1.6085411941589685e-06,
      "loss": 0.0028,
      "step": 238100
    },
    {
      "epoch": 1.5271635779407569,
      "grad_norm": 0.14196178317070007,
      "learning_rate": 1.6081299494643087e-06,
      "loss": 0.0017,
      "step": 238110
    },
    {
      "epoch": 1.527227714834543,
      "grad_norm": 0.1498994082212448,
      "learning_rate": 1.607718747272205e-06,
      "loss": 0.0007,
      "step": 238120
    },
    {
      "epoch": 1.527291851728329,
      "grad_norm": 0.10175507515668869,
      "learning_rate": 1.6073075875878097e-06,
      "loss": 0.0053,
      "step": 238130
    },
    {
      "epoch": 1.527355988622115,
      "grad_norm": 0.007635892368853092,
      "learning_rate": 1.6068964704162736e-06,
      "loss": 0.0018,
      "step": 238140
    },
    {
      "epoch": 1.527420125515901,
      "grad_norm": 0.022415148094296455,
      "learning_rate": 1.6064853957627513e-06,
      "loss": 0.0009,
      "step": 238150
    },
    {
      "epoch": 1.5274842624096872,
      "grad_norm": 0.18747855722904205,
      "learning_rate": 1.6060743636323923e-06,
      "loss": 0.0016,
      "step": 238160
    },
    {
      "epoch": 1.5275483993034733,
      "grad_norm": 0.126417875289917,
      "learning_rate": 1.6056633740303456e-06,
      "loss": 0.0016,
      "step": 238170
    },
    {
      "epoch": 1.5276125361972595,
      "grad_norm": 0.1865348368883133,
      "learning_rate": 1.6052524269617641e-06,
      "loss": 0.0019,
      "step": 238180
    },
    {
      "epoch": 1.5276766730910456,
      "grad_norm": 0.20272643864154816,
      "learning_rate": 1.604841522431796e-06,
      "loss": 0.0011,
      "step": 238190
    },
    {
      "epoch": 1.5277408099848317,
      "grad_norm": 0.08572908490896225,
      "learning_rate": 1.60443066044559e-06,
      "loss": 0.0015,
      "step": 238200
    },
    {
      "epoch": 1.5278049468786177,
      "grad_norm": 0.09908934682607651,
      "learning_rate": 1.6040198410082936e-06,
      "loss": 0.0028,
      "step": 238210
    },
    {
      "epoch": 1.5278690837724038,
      "grad_norm": 0.009173417463898659,
      "learning_rate": 1.6036090641250568e-06,
      "loss": 0.002,
      "step": 238220
    },
    {
      "epoch": 1.5279332206661898,
      "grad_norm": 0.04009650647640228,
      "learning_rate": 1.603198329801024e-06,
      "loss": 0.0011,
      "step": 238230
    },
    {
      "epoch": 1.527997357559976,
      "grad_norm": 0.08476738631725311,
      "learning_rate": 1.6027876380413453e-06,
      "loss": 0.0011,
      "step": 238240
    },
    {
      "epoch": 1.528061494453762,
      "grad_norm": 0.19845592975616455,
      "learning_rate": 1.602376988851166e-06,
      "loss": 0.0016,
      "step": 238250
    },
    {
      "epoch": 1.5281256313475482,
      "grad_norm": 0.09100116789340973,
      "learning_rate": 1.601966382235629e-06,
      "loss": 0.0014,
      "step": 238260
    },
    {
      "epoch": 1.5281897682413343,
      "grad_norm": 0.019955476745963097,
      "learning_rate": 1.6015558181998842e-06,
      "loss": 0.0009,
      "step": 238270
    },
    {
      "epoch": 1.5282539051351205,
      "grad_norm": 0.07049417495727539,
      "learning_rate": 1.6011452967490732e-06,
      "loss": 0.0016,
      "step": 238280
    },
    {
      "epoch": 1.5283180420289066,
      "grad_norm": 0.01399952918291092,
      "learning_rate": 1.6007348178883391e-06,
      "loss": 0.0018,
      "step": 238290
    },
    {
      "epoch": 1.5283821789226926,
      "grad_norm": 0.01669745333492756,
      "learning_rate": 1.6003243816228292e-06,
      "loss": 0.0007,
      "step": 238300
    },
    {
      "epoch": 1.5284463158164787,
      "grad_norm": 0.07379358261823654,
      "learning_rate": 1.5999139879576842e-06,
      "loss": 0.0015,
      "step": 238310
    },
    {
      "epoch": 1.5285104527102646,
      "grad_norm": 0.014796216040849686,
      "learning_rate": 1.5995036368980472e-06,
      "loss": 0.0006,
      "step": 238320
    },
    {
      "epoch": 1.5285745896040508,
      "grad_norm": 0.09059396386146545,
      "learning_rate": 1.599093328449058e-06,
      "loss": 0.001,
      "step": 238330
    },
    {
      "epoch": 1.528638726497837,
      "grad_norm": 0.39618799090385437,
      "learning_rate": 1.5986830626158618e-06,
      "loss": 0.003,
      "step": 238340
    },
    {
      "epoch": 1.528702863391623,
      "grad_norm": 0.021616334095597267,
      "learning_rate": 1.598272839403598e-06,
      "loss": 0.0006,
      "step": 238350
    },
    {
      "epoch": 1.5287670002854092,
      "grad_norm": 0.0020888724830001593,
      "learning_rate": 1.5978626588174061e-06,
      "loss": 0.0007,
      "step": 238360
    },
    {
      "epoch": 1.5288311371791954,
      "grad_norm": 0.012589544989168644,
      "learning_rate": 1.5974525208624253e-06,
      "loss": 0.0022,
      "step": 238370
    },
    {
      "epoch": 1.5288952740729815,
      "grad_norm": 0.004906138405203819,
      "learning_rate": 1.5970424255437977e-06,
      "loss": 0.0014,
      "step": 238380
    },
    {
      "epoch": 1.5289594109667675,
      "grad_norm": 0.0733562558889389,
      "learning_rate": 1.5966323728666606e-06,
      "loss": 0.0008,
      "step": 238390
    },
    {
      "epoch": 1.5290235478605536,
      "grad_norm": 0.03818623349070549,
      "learning_rate": 1.5962223628361523e-06,
      "loss": 0.0016,
      "step": 238400
    },
    {
      "epoch": 1.5290876847543395,
      "grad_norm": 0.0623963326215744,
      "learning_rate": 1.5958123954574084e-06,
      "loss": 0.001,
      "step": 238410
    },
    {
      "epoch": 1.5291518216481257,
      "grad_norm": 0.08566606044769287,
      "learning_rate": 1.5954024707355703e-06,
      "loss": 0.0009,
      "step": 238420
    },
    {
      "epoch": 1.5292159585419118,
      "grad_norm": 0.036736611276865005,
      "learning_rate": 1.5949925886757722e-06,
      "loss": 0.0012,
      "step": 238430
    },
    {
      "epoch": 1.529280095435698,
      "grad_norm": 0.05121005326509476,
      "learning_rate": 1.5945827492831484e-06,
      "loss": 0.0014,
      "step": 238440
    },
    {
      "epoch": 1.5293442323294841,
      "grad_norm": 0.07882276922464371,
      "learning_rate": 1.594172952562839e-06,
      "loss": 0.0013,
      "step": 238450
    },
    {
      "epoch": 1.5294083692232703,
      "grad_norm": 0.10243957489728928,
      "learning_rate": 1.5937631985199764e-06,
      "loss": 0.0065,
      "step": 238460
    },
    {
      "epoch": 1.5294725061170562,
      "grad_norm": 0.11681561917066574,
      "learning_rate": 1.5933534871596952e-06,
      "loss": 0.0012,
      "step": 238470
    },
    {
      "epoch": 1.5295366430108424,
      "grad_norm": 0.026318980380892754,
      "learning_rate": 1.5929438184871277e-06,
      "loss": 0.0012,
      "step": 238480
    },
    {
      "epoch": 1.5296007799046283,
      "grad_norm": 0.05359599366784096,
      "learning_rate": 1.5925341925074112e-06,
      "loss": 0.0052,
      "step": 238490
    },
    {
      "epoch": 1.5296649167984144,
      "grad_norm": 0.054443489760160446,
      "learning_rate": 1.5921246092256758e-06,
      "loss": 0.0008,
      "step": 238500
    },
    {
      "epoch": 1.5297290536922006,
      "grad_norm": 0.003965005744248629,
      "learning_rate": 1.591715068647055e-06,
      "loss": 0.0005,
      "step": 238510
    },
    {
      "epoch": 1.5297931905859867,
      "grad_norm": 0.034113626927137375,
      "learning_rate": 1.5913055707766788e-06,
      "loss": 0.0019,
      "step": 238520
    },
    {
      "epoch": 1.5298573274797729,
      "grad_norm": 0.012696630321443081,
      "learning_rate": 1.5908961156196818e-06,
      "loss": 0.0011,
      "step": 238530
    },
    {
      "epoch": 1.529921464373559,
      "grad_norm": 0.13981802761554718,
      "learning_rate": 1.5904867031811926e-06,
      "loss": 0.001,
      "step": 238540
    },
    {
      "epoch": 1.5299856012673452,
      "grad_norm": 0.10802409797906876,
      "learning_rate": 1.5900773334663417e-06,
      "loss": 0.0008,
      "step": 238550
    },
    {
      "epoch": 1.530049738161131,
      "grad_norm": 0.07713709026575089,
      "learning_rate": 1.5896680064802573e-06,
      "loss": 0.0015,
      "step": 238560
    },
    {
      "epoch": 1.5301138750549172,
      "grad_norm": 0.08833787590265274,
      "learning_rate": 1.589258722228072e-06,
      "loss": 0.0012,
      "step": 238570
    },
    {
      "epoch": 1.5301780119487032,
      "grad_norm": 0.02495083026587963,
      "learning_rate": 1.5888494807149118e-06,
      "loss": 0.0007,
      "step": 238580
    },
    {
      "epoch": 1.5302421488424893,
      "grad_norm": 0.12479456514120102,
      "learning_rate": 1.5884402819459044e-06,
      "loss": 0.001,
      "step": 238590
    },
    {
      "epoch": 1.5303062857362755,
      "grad_norm": 0.04055798798799515,
      "learning_rate": 1.5880311259261806e-06,
      "loss": 0.0012,
      "step": 238600
    },
    {
      "epoch": 1.5303704226300616,
      "grad_norm": 0.06542731076478958,
      "learning_rate": 1.5876220126608643e-06,
      "loss": 0.0011,
      "step": 238610
    },
    {
      "epoch": 1.5304345595238478,
      "grad_norm": 0.07246585935354233,
      "learning_rate": 1.5872129421550836e-06,
      "loss": 0.0017,
      "step": 238620
    },
    {
      "epoch": 1.530498696417634,
      "grad_norm": 0.06228356808423996,
      "learning_rate": 1.586803914413962e-06,
      "loss": 0.0013,
      "step": 238630
    },
    {
      "epoch": 1.5305628333114198,
      "grad_norm": 0.004853600636124611,
      "learning_rate": 1.5863949294426284e-06,
      "loss": 0.0009,
      "step": 238640
    },
    {
      "epoch": 1.530626970205206,
      "grad_norm": 0.0747649073600769,
      "learning_rate": 1.5859859872462058e-06,
      "loss": 0.0012,
      "step": 238650
    },
    {
      "epoch": 1.5306911070989921,
      "grad_norm": 0.0786866620182991,
      "learning_rate": 1.5855770878298188e-06,
      "loss": 0.001,
      "step": 238660
    },
    {
      "epoch": 1.530755243992778,
      "grad_norm": 0.0508040115237236,
      "learning_rate": 1.58516823119859e-06,
      "loss": 0.0016,
      "step": 238670
    },
    {
      "epoch": 1.5308193808865642,
      "grad_norm": 0.05188383534550667,
      "learning_rate": 1.5847594173576447e-06,
      "loss": 0.0016,
      "step": 238680
    },
    {
      "epoch": 1.5308835177803504,
      "grad_norm": 0.08505114167928696,
      "learning_rate": 1.584350646312105e-06,
      "loss": 0.0012,
      "step": 238690
    },
    {
      "epoch": 1.5309476546741365,
      "grad_norm": 0.10715041309595108,
      "learning_rate": 1.5839419180670935e-06,
      "loss": 0.0009,
      "step": 238700
    },
    {
      "epoch": 1.5310117915679227,
      "grad_norm": 0.0594995841383934,
      "learning_rate": 1.5835332326277287e-06,
      "loss": 0.0008,
      "step": 238710
    },
    {
      "epoch": 1.5310759284617088,
      "grad_norm": 0.0047898245975375175,
      "learning_rate": 1.583124589999136e-06,
      "loss": 0.0005,
      "step": 238720
    },
    {
      "epoch": 1.5311400653554947,
      "grad_norm": 0.19723492860794067,
      "learning_rate": 1.5827159901864342e-06,
      "loss": 0.0012,
      "step": 238730
    },
    {
      "epoch": 1.5312042022492809,
      "grad_norm": 0.053472548723220825,
      "learning_rate": 1.5823074331947418e-06,
      "loss": 0.0008,
      "step": 238740
    },
    {
      "epoch": 1.5312683391430668,
      "grad_norm": 0.15455129742622375,
      "learning_rate": 1.5818989190291816e-06,
      "loss": 0.0007,
      "step": 238750
    },
    {
      "epoch": 1.531332476036853,
      "grad_norm": 0.06002508103847504,
      "learning_rate": 1.5814904476948707e-06,
      "loss": 0.0008,
      "step": 238760
    },
    {
      "epoch": 1.5313966129306391,
      "grad_norm": 0.1651560366153717,
      "learning_rate": 1.5810820191969278e-06,
      "loss": 0.0009,
      "step": 238770
    },
    {
      "epoch": 1.5314607498244253,
      "grad_norm": 0.22709596157073975,
      "learning_rate": 1.5806736335404688e-06,
      "loss": 0.0018,
      "step": 238780
    },
    {
      "epoch": 1.5315248867182114,
      "grad_norm": 0.12977087497711182,
      "learning_rate": 1.5802652907306148e-06,
      "loss": 0.0014,
      "step": 238790
    },
    {
      "epoch": 1.5315890236119976,
      "grad_norm": 0.06415867060422897,
      "learning_rate": 1.5798569907724804e-06,
      "loss": 0.001,
      "step": 238800
    },
    {
      "epoch": 1.5316531605057837,
      "grad_norm": 0.12139426171779633,
      "learning_rate": 1.5794487336711827e-06,
      "loss": 0.0014,
      "step": 238810
    },
    {
      "epoch": 1.5317172973995696,
      "grad_norm": 0.025406209751963615,
      "learning_rate": 1.5790405194318354e-06,
      "loss": 0.0012,
      "step": 238820
    },
    {
      "epoch": 1.5317814342933558,
      "grad_norm": 0.01880517229437828,
      "learning_rate": 1.5786323480595562e-06,
      "loss": 0.0011,
      "step": 238830
    },
    {
      "epoch": 1.5318455711871417,
      "grad_norm": 0.04645241051912308,
      "learning_rate": 1.5782242195594594e-06,
      "loss": 0.0009,
      "step": 238840
    },
    {
      "epoch": 1.5319097080809279,
      "grad_norm": 0.13483476638793945,
      "learning_rate": 1.5778161339366572e-06,
      "loss": 0.0013,
      "step": 238850
    },
    {
      "epoch": 1.531973844974714,
      "grad_norm": 0.06605926156044006,
      "learning_rate": 1.5774080911962657e-06,
      "loss": 0.0016,
      "step": 238860
    },
    {
      "epoch": 1.5320379818685002,
      "grad_norm": 0.15618863701820374,
      "learning_rate": 1.5770000913433974e-06,
      "loss": 0.0016,
      "step": 238870
    },
    {
      "epoch": 1.5321021187622863,
      "grad_norm": 0.0538754016160965,
      "learning_rate": 1.5765921343831642e-06,
      "loss": 0.001,
      "step": 238880
    },
    {
      "epoch": 1.5321662556560725,
      "grad_norm": 0.031060660257935524,
      "learning_rate": 1.5761842203206767e-06,
      "loss": 0.0009,
      "step": 238890
    },
    {
      "epoch": 1.5322303925498584,
      "grad_norm": 0.02626187354326248,
      "learning_rate": 1.5757763491610494e-06,
      "loss": 0.0003,
      "step": 238900
    },
    {
      "epoch": 1.5322945294436445,
      "grad_norm": 0.146221324801445,
      "learning_rate": 1.5753685209093917e-06,
      "loss": 0.0011,
      "step": 238910
    },
    {
      "epoch": 1.5323586663374305,
      "grad_norm": 0.16122718155384064,
      "learning_rate": 1.5749607355708142e-06,
      "loss": 0.0012,
      "step": 238920
    },
    {
      "epoch": 1.5324228032312166,
      "grad_norm": 0.06554492563009262,
      "learning_rate": 1.5745529931504243e-06,
      "loss": 0.0006,
      "step": 238930
    },
    {
      "epoch": 1.5324869401250028,
      "grad_norm": 0.15830129384994507,
      "learning_rate": 1.5741452936533358e-06,
      "loss": 0.0015,
      "step": 238940
    },
    {
      "epoch": 1.532551077018789,
      "grad_norm": 0.23866844177246094,
      "learning_rate": 1.5737376370846547e-06,
      "loss": 0.0024,
      "step": 238950
    },
    {
      "epoch": 1.532615213912575,
      "grad_norm": 0.16389961540699005,
      "learning_rate": 1.5733300234494903e-06,
      "loss": 0.0014,
      "step": 238960
    },
    {
      "epoch": 1.5326793508063612,
      "grad_norm": 0.023869449272751808,
      "learning_rate": 1.5729224527529474e-06,
      "loss": 0.0009,
      "step": 238970
    },
    {
      "epoch": 1.5327434877001473,
      "grad_norm": 0.044073037803173065,
      "learning_rate": 1.5725149250001377e-06,
      "loss": 0.0019,
      "step": 238980
    },
    {
      "epoch": 1.5328076245939333,
      "grad_norm": 0.0290644820779562,
      "learning_rate": 1.5721074401961633e-06,
      "loss": 0.0005,
      "step": 238990
    },
    {
      "epoch": 1.5328717614877194,
      "grad_norm": 0.055113472044467926,
      "learning_rate": 1.5716999983461344e-06,
      "loss": 0.0011,
      "step": 239000
    },
    {
      "epoch": 1.5329358983815053,
      "grad_norm": 0.007592161186039448,
      "learning_rate": 1.5712925994551536e-06,
      "loss": 0.0024,
      "step": 239010
    },
    {
      "epoch": 1.5330000352752915,
      "grad_norm": 0.14502429962158203,
      "learning_rate": 1.5708852435283283e-06,
      "loss": 0.002,
      "step": 239020
    },
    {
      "epoch": 1.5330641721690776,
      "grad_norm": 0.1949986070394516,
      "learning_rate": 1.5704779305707613e-06,
      "loss": 0.0028,
      "step": 239030
    },
    {
      "epoch": 1.5331283090628638,
      "grad_norm": 0.06624270975589752,
      "learning_rate": 1.570070660587557e-06,
      "loss": 0.001,
      "step": 239040
    },
    {
      "epoch": 1.53319244595665,
      "grad_norm": 0.044614873826503754,
      "learning_rate": 1.5696634335838172e-06,
      "loss": 0.0024,
      "step": 239050
    },
    {
      "epoch": 1.533256582850436,
      "grad_norm": 0.0646383985877037,
      "learning_rate": 1.569256249564648e-06,
      "loss": 0.0008,
      "step": 239060
    },
    {
      "epoch": 1.533320719744222,
      "grad_norm": 0.025071272626519203,
      "learning_rate": 1.5688491085351499e-06,
      "loss": 0.0009,
      "step": 239070
    },
    {
      "epoch": 1.5333848566380082,
      "grad_norm": 0.027934769168496132,
      "learning_rate": 1.5684420105004245e-06,
      "loss": 0.0005,
      "step": 239080
    },
    {
      "epoch": 1.5334489935317943,
      "grad_norm": 0.1389744132757187,
      "learning_rate": 1.5680349554655716e-06,
      "loss": 0.0016,
      "step": 239090
    },
    {
      "epoch": 1.5335131304255802,
      "grad_norm": 0.025445854291319847,
      "learning_rate": 1.567627943435695e-06,
      "loss": 0.0005,
      "step": 239100
    },
    {
      "epoch": 1.5335772673193664,
      "grad_norm": 0.0023066888097673655,
      "learning_rate": 1.5672209744158935e-06,
      "loss": 0.0007,
      "step": 239110
    },
    {
      "epoch": 1.5336414042131525,
      "grad_norm": 0.11088065803050995,
      "learning_rate": 1.5668140484112649e-06,
      "loss": 0.0015,
      "step": 239120
    },
    {
      "epoch": 1.5337055411069387,
      "grad_norm": 0.09681985527276993,
      "learning_rate": 1.5664071654269114e-06,
      "loss": 0.001,
      "step": 239130
    },
    {
      "epoch": 1.5337696780007248,
      "grad_norm": 0.04255175217986107,
      "learning_rate": 1.5660003254679302e-06,
      "loss": 0.0012,
      "step": 239140
    },
    {
      "epoch": 1.533833814894511,
      "grad_norm": 0.05023466795682907,
      "learning_rate": 1.565593528539419e-06,
      "loss": 0.001,
      "step": 239150
    },
    {
      "epoch": 1.533897951788297,
      "grad_norm": 0.026395542547106743,
      "learning_rate": 1.5651867746464743e-06,
      "loss": 0.0015,
      "step": 239160
    },
    {
      "epoch": 1.533962088682083,
      "grad_norm": 0.03340686485171318,
      "learning_rate": 1.564780063794195e-06,
      "loss": 0.0006,
      "step": 239170
    },
    {
      "epoch": 1.534026225575869,
      "grad_norm": 0.011547114700078964,
      "learning_rate": 1.5643733959876772e-06,
      "loss": 0.0013,
      "step": 239180
    },
    {
      "epoch": 1.5340903624696551,
      "grad_norm": 0.02173069305717945,
      "learning_rate": 1.5639667712320161e-06,
      "loss": 0.0015,
      "step": 239190
    },
    {
      "epoch": 1.5341544993634413,
      "grad_norm": 0.10959511250257492,
      "learning_rate": 1.5635601895323054e-06,
      "loss": 0.0019,
      "step": 239200
    },
    {
      "epoch": 1.5342186362572274,
      "grad_norm": 0.0560414157807827,
      "learning_rate": 1.563153650893643e-06,
      "loss": 0.0021,
      "step": 239210
    },
    {
      "epoch": 1.5342827731510136,
      "grad_norm": 0.060674846172332764,
      "learning_rate": 1.5627471553211216e-06,
      "loss": 0.001,
      "step": 239220
    },
    {
      "epoch": 1.5343469100447997,
      "grad_norm": 0.06418942660093307,
      "learning_rate": 1.562340702819835e-06,
      "loss": 0.0004,
      "step": 239230
    },
    {
      "epoch": 1.5344110469385859,
      "grad_norm": 0.16282793879508972,
      "learning_rate": 1.561934293394875e-06,
      "loss": 0.0013,
      "step": 239240
    },
    {
      "epoch": 1.5344751838323718,
      "grad_norm": 0.08704288303852081,
      "learning_rate": 1.5615279270513367e-06,
      "loss": 0.0011,
      "step": 239250
    },
    {
      "epoch": 1.534539320726158,
      "grad_norm": 0.12967506051063538,
      "learning_rate": 1.5611216037943105e-06,
      "loss": 0.0011,
      "step": 239260
    },
    {
      "epoch": 1.5346034576199439,
      "grad_norm": 0.038755375891923904,
      "learning_rate": 1.5607153236288874e-06,
      "loss": 0.0005,
      "step": 239270
    },
    {
      "epoch": 1.53466759451373,
      "grad_norm": 0.048440322279930115,
      "learning_rate": 1.5603090865601605e-06,
      "loss": 0.0006,
      "step": 239280
    },
    {
      "epoch": 1.5347317314075162,
      "grad_norm": 0.03399894759058952,
      "learning_rate": 1.559902892593219e-06,
      "loss": 0.0029,
      "step": 239290
    },
    {
      "epoch": 1.5347958683013023,
      "grad_norm": 0.16211147606372833,
      "learning_rate": 1.5594967417331536e-06,
      "loss": 0.0008,
      "step": 239300
    },
    {
      "epoch": 1.5348600051950885,
      "grad_norm": 0.08329859375953674,
      "learning_rate": 1.5590906339850504e-06,
      "loss": 0.0009,
      "step": 239310
    },
    {
      "epoch": 1.5349241420888746,
      "grad_norm": 0.24930784106254578,
      "learning_rate": 1.558684569354003e-06,
      "loss": 0.0014,
      "step": 239320
    },
    {
      "epoch": 1.5349882789826605,
      "grad_norm": 0.06121993437409401,
      "learning_rate": 1.5582785478450968e-06,
      "loss": 0.0011,
      "step": 239330
    },
    {
      "epoch": 1.5350524158764467,
      "grad_norm": 0.1486685872077942,
      "learning_rate": 1.5578725694634207e-06,
      "loss": 0.001,
      "step": 239340
    },
    {
      "epoch": 1.5351165527702326,
      "grad_norm": 0.04166239872574806,
      "learning_rate": 1.5574666342140598e-06,
      "loss": 0.001,
      "step": 239350
    },
    {
      "epoch": 1.5351806896640188,
      "grad_norm": 0.027320783585309982,
      "learning_rate": 1.5570607421021032e-06,
      "loss": 0.0008,
      "step": 239360
    },
    {
      "epoch": 1.535244826557805,
      "grad_norm": 0.12129577994346619,
      "learning_rate": 1.556654893132637e-06,
      "loss": 0.0011,
      "step": 239370
    },
    {
      "epoch": 1.535308963451591,
      "grad_norm": 0.03990350291132927,
      "learning_rate": 1.5562490873107456e-06,
      "loss": 0.0012,
      "step": 239380
    },
    {
      "epoch": 1.5353731003453772,
      "grad_norm": 0.09766650199890137,
      "learning_rate": 1.5558433246415123e-06,
      "loss": 0.0016,
      "step": 239390
    },
    {
      "epoch": 1.5354372372391634,
      "grad_norm": 0.22297364473342896,
      "learning_rate": 1.5554376051300258e-06,
      "loss": 0.002,
      "step": 239400
    },
    {
      "epoch": 1.5355013741329495,
      "grad_norm": 0.06788810342550278,
      "learning_rate": 1.5550319287813675e-06,
      "loss": 0.0013,
      "step": 239410
    },
    {
      "epoch": 1.5355655110267354,
      "grad_norm": 0.1882997751235962,
      "learning_rate": 1.5546262956006197e-06,
      "loss": 0.0016,
      "step": 239420
    },
    {
      "epoch": 1.5356296479205216,
      "grad_norm": 0.01625201851129532,
      "learning_rate": 1.5542207055928688e-06,
      "loss": 0.0006,
      "step": 239430
    },
    {
      "epoch": 1.5356937848143075,
      "grad_norm": 0.04416879639029503,
      "learning_rate": 1.553815158763195e-06,
      "loss": 0.0009,
      "step": 239440
    },
    {
      "epoch": 1.5357579217080937,
      "grad_norm": 0.06767366826534271,
      "learning_rate": 1.55340965511668e-06,
      "loss": 0.0008,
      "step": 239450
    },
    {
      "epoch": 1.5358220586018798,
      "grad_norm": 0.003938053268939257,
      "learning_rate": 1.5530041946584035e-06,
      "loss": 0.0007,
      "step": 239460
    },
    {
      "epoch": 1.535886195495666,
      "grad_norm": 0.12138058990240097,
      "learning_rate": 1.5525987773934499e-06,
      "loss": 0.0011,
      "step": 239470
    },
    {
      "epoch": 1.535950332389452,
      "grad_norm": 0.06975291669368744,
      "learning_rate": 1.552193403326897e-06,
      "loss": 0.0011,
      "step": 239480
    },
    {
      "epoch": 1.5360144692832383,
      "grad_norm": 0.07297302782535553,
      "learning_rate": 1.5517880724638258e-06,
      "loss": 0.0016,
      "step": 239490
    },
    {
      "epoch": 1.5360786061770242,
      "grad_norm": 0.19614681601524353,
      "learning_rate": 1.5513827848093115e-06,
      "loss": 0.0013,
      "step": 239500
    },
    {
      "epoch": 1.5361427430708103,
      "grad_norm": 0.04527741298079491,
      "learning_rate": 1.5509775403684385e-06,
      "loss": 0.0006,
      "step": 239510
    },
    {
      "epoch": 1.5362068799645965,
      "grad_norm": 0.07013577222824097,
      "learning_rate": 1.5505723391462813e-06,
      "loss": 0.0014,
      "step": 239520
    },
    {
      "epoch": 1.5362710168583824,
      "grad_norm": 0.047192350029945374,
      "learning_rate": 1.550167181147918e-06,
      "loss": 0.0015,
      "step": 239530
    },
    {
      "epoch": 1.5363351537521686,
      "grad_norm": 0.1701509952545166,
      "learning_rate": 1.5497620663784236e-06,
      "loss": 0.0009,
      "step": 239540
    },
    {
      "epoch": 1.5363992906459547,
      "grad_norm": 0.06133796274662018,
      "learning_rate": 1.5493569948428783e-06,
      "loss": 0.0011,
      "step": 239550
    },
    {
      "epoch": 1.5364634275397409,
      "grad_norm": 0.08307241648435593,
      "learning_rate": 1.5489519665463559e-06,
      "loss": 0.001,
      "step": 239560
    },
    {
      "epoch": 1.536527564433527,
      "grad_norm": 0.06485676020383835,
      "learning_rate": 1.5485469814939303e-06,
      "loss": 0.001,
      "step": 239570
    },
    {
      "epoch": 1.5365917013273132,
      "grad_norm": 0.0776052251458168,
      "learning_rate": 1.5481420396906793e-06,
      "loss": 0.0025,
      "step": 239580
    },
    {
      "epoch": 1.536655838221099,
      "grad_norm": 0.04508012533187866,
      "learning_rate": 1.5477371411416753e-06,
      "loss": 0.0012,
      "step": 239590
    },
    {
      "epoch": 1.5367199751148852,
      "grad_norm": 0.03473687916994095,
      "learning_rate": 1.5473322858519924e-06,
      "loss": 0.001,
      "step": 239600
    },
    {
      "epoch": 1.5367841120086712,
      "grad_norm": 0.10467745363712311,
      "learning_rate": 1.546927473826702e-06,
      "loss": 0.0022,
      "step": 239610
    },
    {
      "epoch": 1.5368482489024573,
      "grad_norm": 0.11767324060201645,
      "learning_rate": 1.5465227050708797e-06,
      "loss": 0.0009,
      "step": 239620
    },
    {
      "epoch": 1.5369123857962435,
      "grad_norm": 0.033213626593351364,
      "learning_rate": 1.5461179795895963e-06,
      "loss": 0.0011,
      "step": 239630
    },
    {
      "epoch": 1.5369765226900296,
      "grad_norm": 0.03318759426474571,
      "learning_rate": 1.545713297387923e-06,
      "loss": 0.0008,
      "step": 239640
    },
    {
      "epoch": 1.5370406595838157,
      "grad_norm": 0.21723335981369019,
      "learning_rate": 1.5453086584709286e-06,
      "loss": 0.0021,
      "step": 239650
    },
    {
      "epoch": 1.537104796477602,
      "grad_norm": 0.2502657175064087,
      "learning_rate": 1.5449040628436884e-06,
      "loss": 0.0019,
      "step": 239660
    },
    {
      "epoch": 1.537168933371388,
      "grad_norm": 0.03230128437280655,
      "learning_rate": 1.5444995105112686e-06,
      "loss": 0.0006,
      "step": 239670
    },
    {
      "epoch": 1.537233070265174,
      "grad_norm": 0.12768696248531342,
      "learning_rate": 1.5440950014787404e-06,
      "loss": 0.0026,
      "step": 239680
    },
    {
      "epoch": 1.5372972071589601,
      "grad_norm": 0.07762544602155685,
      "learning_rate": 1.5436905357511694e-06,
      "loss": 0.0033,
      "step": 239690
    },
    {
      "epoch": 1.537361344052746,
      "grad_norm": 0.06683100759983063,
      "learning_rate": 1.5432861133336285e-06,
      "loss": 0.0008,
      "step": 239700
    },
    {
      "epoch": 1.5374254809465322,
      "grad_norm": 0.2564246356487274,
      "learning_rate": 1.5428817342311825e-06,
      "loss": 0.0033,
      "step": 239710
    },
    {
      "epoch": 1.5374896178403183,
      "grad_norm": 0.15359140932559967,
      "learning_rate": 1.5424773984488978e-06,
      "loss": 0.0017,
      "step": 239720
    },
    {
      "epoch": 1.5375537547341045,
      "grad_norm": 0.015263685956597328,
      "learning_rate": 1.5420731059918436e-06,
      "loss": 0.0015,
      "step": 239730
    },
    {
      "epoch": 1.5376178916278906,
      "grad_norm": 0.07375656068325043,
      "learning_rate": 1.5416688568650856e-06,
      "loss": 0.001,
      "step": 239740
    },
    {
      "epoch": 1.5376820285216768,
      "grad_norm": 0.03587993606925011,
      "learning_rate": 1.5412646510736878e-06,
      "loss": 0.0009,
      "step": 239750
    },
    {
      "epoch": 1.5377461654154627,
      "grad_norm": 0.11929059773683548,
      "learning_rate": 1.540860488622714e-06,
      "loss": 0.0011,
      "step": 239760
    },
    {
      "epoch": 1.5378103023092489,
      "grad_norm": 0.08358773589134216,
      "learning_rate": 1.5404563695172309e-06,
      "loss": 0.001,
      "step": 239770
    },
    {
      "epoch": 1.5378744392030348,
      "grad_norm": 0.009313437156379223,
      "learning_rate": 1.5400522937623035e-06,
      "loss": 0.0005,
      "step": 239780
    },
    {
      "epoch": 1.537938576096821,
      "grad_norm": 0.1198863759636879,
      "learning_rate": 1.5396482613629937e-06,
      "loss": 0.0019,
      "step": 239790
    },
    {
      "epoch": 1.538002712990607,
      "grad_norm": 0.032433778047561646,
      "learning_rate": 1.539244272324364e-06,
      "loss": 0.0016,
      "step": 239800
    },
    {
      "epoch": 1.5380668498843932,
      "grad_norm": 0.042525772005319595,
      "learning_rate": 1.5388403266514756e-06,
      "loss": 0.0004,
      "step": 239810
    },
    {
      "epoch": 1.5381309867781794,
      "grad_norm": 0.21475815773010254,
      "learning_rate": 1.5384364243493932e-06,
      "loss": 0.001,
      "step": 239820
    },
    {
      "epoch": 1.5381951236719655,
      "grad_norm": 0.09270118921995163,
      "learning_rate": 1.5380325654231764e-06,
      "loss": 0.0006,
      "step": 239830
    },
    {
      "epoch": 1.5382592605657517,
      "grad_norm": 0.05335585027933121,
      "learning_rate": 1.5376287498778841e-06,
      "loss": 0.0016,
      "step": 239840
    },
    {
      "epoch": 1.5383233974595376,
      "grad_norm": 0.046682726591825485,
      "learning_rate": 1.5372249777185793e-06,
      "loss": 0.0013,
      "step": 239850
    },
    {
      "epoch": 1.5383875343533238,
      "grad_norm": 0.02034573256969452,
      "learning_rate": 1.5368212489503208e-06,
      "loss": 0.0015,
      "step": 239860
    },
    {
      "epoch": 1.5384516712471097,
      "grad_norm": 0.08550441265106201,
      "learning_rate": 1.536417563578167e-06,
      "loss": 0.0009,
      "step": 239870
    },
    {
      "epoch": 1.5385158081408958,
      "grad_norm": 0.024634793400764465,
      "learning_rate": 1.5360139216071746e-06,
      "loss": 0.0015,
      "step": 239880
    },
    {
      "epoch": 1.538579945034682,
      "grad_norm": 0.0035681596491485834,
      "learning_rate": 1.5356103230424057e-06,
      "loss": 0.0015,
      "step": 239890
    },
    {
      "epoch": 1.5386440819284681,
      "grad_norm": 0.18973346054553986,
      "learning_rate": 1.535206767888915e-06,
      "loss": 0.0012,
      "step": 239900
    },
    {
      "epoch": 1.5387082188222543,
      "grad_norm": 0.014074346981942654,
      "learning_rate": 1.5348032561517596e-06,
      "loss": 0.0014,
      "step": 239910
    },
    {
      "epoch": 1.5387723557160404,
      "grad_norm": 0.13889437913894653,
      "learning_rate": 1.5343997878359945e-06,
      "loss": 0.0014,
      "step": 239920
    },
    {
      "epoch": 1.5388364926098266,
      "grad_norm": 0.19967631995677948,
      "learning_rate": 1.5339963629466787e-06,
      "loss": 0.0023,
      "step": 239930
    },
    {
      "epoch": 1.5389006295036125,
      "grad_norm": 0.03955758363008499,
      "learning_rate": 1.5335929814888656e-06,
      "loss": 0.0013,
      "step": 239940
    },
    {
      "epoch": 1.5389647663973987,
      "grad_norm": 0.05924701690673828,
      "learning_rate": 1.5331896434676096e-06,
      "loss": 0.0006,
      "step": 239950
    },
    {
      "epoch": 1.5390289032911846,
      "grad_norm": 0.06783130764961243,
      "learning_rate": 1.5327863488879635e-06,
      "loss": 0.0007,
      "step": 239960
    },
    {
      "epoch": 1.5390930401849707,
      "grad_norm": 0.009174141101539135,
      "learning_rate": 1.532383097754984e-06,
      "loss": 0.0017,
      "step": 239970
    },
    {
      "epoch": 1.5391571770787569,
      "grad_norm": 0.5989391803741455,
      "learning_rate": 1.5319798900737226e-06,
      "loss": 0.0021,
      "step": 239980
    },
    {
      "epoch": 1.539221313972543,
      "grad_norm": 0.09201222658157349,
      "learning_rate": 1.5315767258492304e-06,
      "loss": 0.0008,
      "step": 239990
    },
    {
      "epoch": 1.5392854508663292,
      "grad_norm": 0.048332955688238144,
      "learning_rate": 1.5311736050865616e-06,
      "loss": 0.0011,
      "step": 240000
    },
    {
      "epoch": 1.5393495877601153,
      "grad_norm": 0.2488795667886734,
      "learning_rate": 1.530770527790767e-06,
      "loss": 0.0021,
      "step": 240010
    },
    {
      "epoch": 1.5394137246539012,
      "grad_norm": 0.04531652107834816,
      "learning_rate": 1.5303674939668971e-06,
      "loss": 0.001,
      "step": 240020
    },
    {
      "epoch": 1.5394778615476874,
      "grad_norm": 0.012171772308647633,
      "learning_rate": 1.529964503620001e-06,
      "loss": 0.0013,
      "step": 240030
    },
    {
      "epoch": 1.5395419984414733,
      "grad_norm": 0.019362622871994972,
      "learning_rate": 1.5295615567551307e-06,
      "loss": 0.0012,
      "step": 240040
    },
    {
      "epoch": 1.5396061353352595,
      "grad_norm": 0.01918141543865204,
      "learning_rate": 1.5291586533773351e-06,
      "loss": 0.0007,
      "step": 240050
    },
    {
      "epoch": 1.5396702722290456,
      "grad_norm": 0.050688810646533966,
      "learning_rate": 1.5287557934916615e-06,
      "loss": 0.001,
      "step": 240060
    },
    {
      "epoch": 1.5397344091228318,
      "grad_norm": 0.07629798352718353,
      "learning_rate": 1.5283529771031568e-06,
      "loss": 0.0027,
      "step": 240070
    },
    {
      "epoch": 1.539798546016618,
      "grad_norm": 0.08818703144788742,
      "learning_rate": 1.5279502042168726e-06,
      "loss": 0.0013,
      "step": 240080
    },
    {
      "epoch": 1.539862682910404,
      "grad_norm": 0.07173456251621246,
      "learning_rate": 1.527547474837854e-06,
      "loss": 0.0015,
      "step": 240090
    },
    {
      "epoch": 1.5399268198041902,
      "grad_norm": 0.04634719341993332,
      "learning_rate": 1.5271447889711466e-06,
      "loss": 0.0005,
      "step": 240100
    },
    {
      "epoch": 1.5399909566979761,
      "grad_norm": 0.0994994193315506,
      "learning_rate": 1.5267421466217958e-06,
      "loss": 0.001,
      "step": 240110
    },
    {
      "epoch": 1.5400550935917623,
      "grad_norm": 0.20136234164237976,
      "learning_rate": 1.52633954779485e-06,
      "loss": 0.0009,
      "step": 240120
    },
    {
      "epoch": 1.5401192304855482,
      "grad_norm": 0.09540402889251709,
      "learning_rate": 1.525936992495352e-06,
      "loss": 0.0008,
      "step": 240130
    },
    {
      "epoch": 1.5401833673793344,
      "grad_norm": 0.18061742186546326,
      "learning_rate": 1.5255344807283445e-06,
      "loss": 0.0011,
      "step": 240140
    },
    {
      "epoch": 1.5402475042731205,
      "grad_norm": 0.04126046970486641,
      "learning_rate": 1.525132012498875e-06,
      "loss": 0.0013,
      "step": 240150
    },
    {
      "epoch": 1.5403116411669067,
      "grad_norm": 0.090067058801651,
      "learning_rate": 1.524729587811985e-06,
      "loss": 0.0003,
      "step": 240160
    },
    {
      "epoch": 1.5403757780606928,
      "grad_norm": 0.006891035940498114,
      "learning_rate": 1.5243272066727167e-06,
      "loss": 0.002,
      "step": 240170
    },
    {
      "epoch": 1.540439914954479,
      "grad_norm": 0.06617258489131927,
      "learning_rate": 1.5239248690861109e-06,
      "loss": 0.0012,
      "step": 240180
    },
    {
      "epoch": 1.5405040518482649,
      "grad_norm": 0.14071519672870636,
      "learning_rate": 1.5235225750572124e-06,
      "loss": 0.0012,
      "step": 240190
    },
    {
      "epoch": 1.540568188742051,
      "grad_norm": 0.04268965870141983,
      "learning_rate": 1.5231203245910608e-06,
      "loss": 0.0008,
      "step": 240200
    },
    {
      "epoch": 1.5406323256358372,
      "grad_norm": 0.07042060047388077,
      "learning_rate": 1.522718117692697e-06,
      "loss": 0.0013,
      "step": 240210
    },
    {
      "epoch": 1.5406964625296231,
      "grad_norm": 0.03442941606044769,
      "learning_rate": 1.522315954367158e-06,
      "loss": 0.0007,
      "step": 240220
    },
    {
      "epoch": 1.5407605994234093,
      "grad_norm": 0.05070033669471741,
      "learning_rate": 1.5219138346194873e-06,
      "loss": 0.0018,
      "step": 240230
    },
    {
      "epoch": 1.5408247363171954,
      "grad_norm": 0.09447235614061356,
      "learning_rate": 1.5215117584547219e-06,
      "loss": 0.0018,
      "step": 240240
    },
    {
      "epoch": 1.5408888732109816,
      "grad_norm": 0.07985574752092361,
      "learning_rate": 1.5211097258779e-06,
      "loss": 0.0031,
      "step": 240250
    },
    {
      "epoch": 1.5409530101047677,
      "grad_norm": 0.08799561113119125,
      "learning_rate": 1.5207077368940587e-06,
      "loss": 0.0009,
      "step": 240260
    },
    {
      "epoch": 1.5410171469985539,
      "grad_norm": 0.07382316887378693,
      "learning_rate": 1.520305791508237e-06,
      "loss": 0.0017,
      "step": 240270
    },
    {
      "epoch": 1.5410812838923398,
      "grad_norm": 0.10267768800258636,
      "learning_rate": 1.5199038897254709e-06,
      "loss": 0.0012,
      "step": 240280
    },
    {
      "epoch": 1.541145420786126,
      "grad_norm": 0.06217135861515999,
      "learning_rate": 1.5195020315507947e-06,
      "loss": 0.0009,
      "step": 240290
    },
    {
      "epoch": 1.5412095576799119,
      "grad_norm": 0.07185224443674088,
      "learning_rate": 1.5191002169892472e-06,
      "loss": 0.0006,
      "step": 240300
    },
    {
      "epoch": 1.541273694573698,
      "grad_norm": 0.0013148905709385872,
      "learning_rate": 1.5186984460458614e-06,
      "loss": 0.0008,
      "step": 240310
    },
    {
      "epoch": 1.5413378314674842,
      "grad_norm": 0.06411554664373398,
      "learning_rate": 1.5182967187256725e-06,
      "loss": 0.0011,
      "step": 240320
    },
    {
      "epoch": 1.5414019683612703,
      "grad_norm": 0.16318665444850922,
      "learning_rate": 1.5178950350337123e-06,
      "loss": 0.0016,
      "step": 240330
    },
    {
      "epoch": 1.5414661052550565,
      "grad_norm": 0.21505574882030487,
      "learning_rate": 1.5174933949750176e-06,
      "loss": 0.0015,
      "step": 240340
    },
    {
      "epoch": 1.5415302421488426,
      "grad_norm": 0.1235521212220192,
      "learning_rate": 1.5170917985546191e-06,
      "loss": 0.0013,
      "step": 240350
    },
    {
      "epoch": 1.5415943790426287,
      "grad_norm": 0.16214965283870697,
      "learning_rate": 1.51669024577755e-06,
      "loss": 0.0013,
      "step": 240360
    },
    {
      "epoch": 1.5416585159364147,
      "grad_norm": 0.04509899392724037,
      "learning_rate": 1.51628873664884e-06,
      "loss": 0.0012,
      "step": 240370
    },
    {
      "epoch": 1.5417226528302008,
      "grad_norm": 0.04285109043121338,
      "learning_rate": 1.5158872711735234e-06,
      "loss": 0.0009,
      "step": 240380
    },
    {
      "epoch": 1.5417867897239868,
      "grad_norm": 0.043615441769361496,
      "learning_rate": 1.5154858493566294e-06,
      "loss": 0.0031,
      "step": 240390
    },
    {
      "epoch": 1.541850926617773,
      "grad_norm": 0.01779012195765972,
      "learning_rate": 1.5150844712031882e-06,
      "loss": 0.0005,
      "step": 240400
    },
    {
      "epoch": 1.541915063511559,
      "grad_norm": 0.08573545515537262,
      "learning_rate": 1.5146831367182275e-06,
      "loss": 0.0033,
      "step": 240410
    },
    {
      "epoch": 1.5419792004053452,
      "grad_norm": 0.026266250759363174,
      "learning_rate": 1.5142818459067792e-06,
      "loss": 0.0012,
      "step": 240420
    },
    {
      "epoch": 1.5420433372991313,
      "grad_norm": 0.13313041627407074,
      "learning_rate": 1.5138805987738715e-06,
      "loss": 0.001,
      "step": 240430
    },
    {
      "epoch": 1.5421074741929175,
      "grad_norm": 0.11342868208885193,
      "learning_rate": 1.5134793953245291e-06,
      "loss": 0.0012,
      "step": 240440
    },
    {
      "epoch": 1.5421716110867034,
      "grad_norm": 0.09087756276130676,
      "learning_rate": 1.513078235563783e-06,
      "loss": 0.0014,
      "step": 240450
    },
    {
      "epoch": 1.5422357479804896,
      "grad_norm": 0.046084512025117874,
      "learning_rate": 1.5126771194966595e-06,
      "loss": 0.0017,
      "step": 240460
    },
    {
      "epoch": 1.5422998848742755,
      "grad_norm": 0.03875308111310005,
      "learning_rate": 1.5122760471281833e-06,
      "loss": 0.0007,
      "step": 240470
    },
    {
      "epoch": 1.5423640217680616,
      "grad_norm": 0.1758212000131607,
      "learning_rate": 1.5118750184633796e-06,
      "loss": 0.0009,
      "step": 240480
    },
    {
      "epoch": 1.5424281586618478,
      "grad_norm": 0.060830261558294296,
      "learning_rate": 1.5114740335072764e-06,
      "loss": 0.0009,
      "step": 240490
    },
    {
      "epoch": 1.542492295555634,
      "grad_norm": 0.10947144776582718,
      "learning_rate": 1.5110730922648969e-06,
      "loss": 0.0028,
      "step": 240500
    },
    {
      "epoch": 1.54255643244942,
      "grad_norm": 0.09652144461870193,
      "learning_rate": 1.510672194741265e-06,
      "loss": 0.0009,
      "step": 240510
    },
    {
      "epoch": 1.5426205693432062,
      "grad_norm": 0.09939628094434738,
      "learning_rate": 1.5102713409414028e-06,
      "loss": 0.0007,
      "step": 240520
    },
    {
      "epoch": 1.5426847062369924,
      "grad_norm": 0.05884421616792679,
      "learning_rate": 1.5098705308703344e-06,
      "loss": 0.0017,
      "step": 240530
    },
    {
      "epoch": 1.5427488431307783,
      "grad_norm": 0.013617325574159622,
      "learning_rate": 1.5094697645330841e-06,
      "loss": 0.0006,
      "step": 240540
    },
    {
      "epoch": 1.5428129800245645,
      "grad_norm": 0.09015436470508575,
      "learning_rate": 1.5090690419346726e-06,
      "loss": 0.001,
      "step": 240550
    },
    {
      "epoch": 1.5428771169183504,
      "grad_norm": 0.12895537912845612,
      "learning_rate": 1.5086683630801197e-06,
      "loss": 0.0027,
      "step": 240560
    },
    {
      "epoch": 1.5429412538121365,
      "grad_norm": 0.03182035684585571,
      "learning_rate": 1.5082677279744485e-06,
      "loss": 0.0017,
      "step": 240570
    },
    {
      "epoch": 1.5430053907059227,
      "grad_norm": 0.007782531436532736,
      "learning_rate": 1.5078671366226783e-06,
      "loss": 0.0008,
      "step": 240580
    },
    {
      "epoch": 1.5430695275997088,
      "grad_norm": 0.051679827272892,
      "learning_rate": 1.5074665890298285e-06,
      "loss": 0.0009,
      "step": 240590
    },
    {
      "epoch": 1.543133664493495,
      "grad_norm": 0.11652804166078568,
      "learning_rate": 1.5070660852009173e-06,
      "loss": 0.0015,
      "step": 240600
    },
    {
      "epoch": 1.5431978013872811,
      "grad_norm": 0.09914771467447281,
      "learning_rate": 1.5066656251409656e-06,
      "loss": 0.0017,
      "step": 240610
    },
    {
      "epoch": 1.543261938281067,
      "grad_norm": 0.04378265142440796,
      "learning_rate": 1.5062652088549907e-06,
      "loss": 0.001,
      "step": 240620
    },
    {
      "epoch": 1.5433260751748532,
      "grad_norm": 0.06831402331590652,
      "learning_rate": 1.5058648363480088e-06,
      "loss": 0.0008,
      "step": 240630
    },
    {
      "epoch": 1.5433902120686394,
      "grad_norm": 0.052788347005844116,
      "learning_rate": 1.5054645076250368e-06,
      "loss": 0.0031,
      "step": 240640
    },
    {
      "epoch": 1.5434543489624253,
      "grad_norm": 0.006500875577330589,
      "learning_rate": 1.5050642226910938e-06,
      "loss": 0.0003,
      "step": 240650
    },
    {
      "epoch": 1.5435184858562114,
      "grad_norm": 0.08855942636728287,
      "learning_rate": 1.5046639815511932e-06,
      "loss": 0.0016,
      "step": 240660
    },
    {
      "epoch": 1.5435826227499976,
      "grad_norm": 0.03987079858779907,
      "learning_rate": 1.5042637842103514e-06,
      "loss": 0.0012,
      "step": 240670
    },
    {
      "epoch": 1.5436467596437837,
      "grad_norm": 0.09060493856668472,
      "learning_rate": 1.5038636306735815e-06,
      "loss": 0.0008,
      "step": 240680
    },
    {
      "epoch": 1.5437108965375699,
      "grad_norm": 0.029795430600643158,
      "learning_rate": 1.5034635209458998e-06,
      "loss": 0.0009,
      "step": 240690
    },
    {
      "epoch": 1.543775033431356,
      "grad_norm": 0.10571033507585526,
      "learning_rate": 1.5030634550323198e-06,
      "loss": 0.0007,
      "step": 240700
    },
    {
      "epoch": 1.543839170325142,
      "grad_norm": 0.05886177346110344,
      "learning_rate": 1.502663432937852e-06,
      "loss": 0.0021,
      "step": 240710
    },
    {
      "epoch": 1.543903307218928,
      "grad_norm": 0.07125724852085114,
      "learning_rate": 1.5022634546675124e-06,
      "loss": 0.0013,
      "step": 240720
    },
    {
      "epoch": 1.543967444112714,
      "grad_norm": 0.15605969727039337,
      "learning_rate": 1.5018635202263115e-06,
      "loss": 0.002,
      "step": 240730
    },
    {
      "epoch": 1.5440315810065002,
      "grad_norm": 0.06762544065713882,
      "learning_rate": 1.5014636296192607e-06,
      "loss": 0.0014,
      "step": 240740
    },
    {
      "epoch": 1.5440957179002863,
      "grad_norm": 0.0794999822974205,
      "learning_rate": 1.5010637828513702e-06,
      "loss": 0.0018,
      "step": 240750
    },
    {
      "epoch": 1.5441598547940725,
      "grad_norm": 0.07379082590341568,
      "learning_rate": 1.5006639799276518e-06,
      "loss": 0.0009,
      "step": 240760
    },
    {
      "epoch": 1.5442239916878586,
      "grad_norm": 0.06329061836004257,
      "learning_rate": 1.5002642208531154e-06,
      "loss": 0.0017,
      "step": 240770
    },
    {
      "epoch": 1.5442881285816448,
      "grad_norm": 0.0044218008406460285,
      "learning_rate": 1.4998645056327687e-06,
      "loss": 0.0012,
      "step": 240780
    },
    {
      "epoch": 1.544352265475431,
      "grad_norm": 0.13181927800178528,
      "learning_rate": 1.4994648342716205e-06,
      "loss": 0.0017,
      "step": 240790
    },
    {
      "epoch": 1.5444164023692168,
      "grad_norm": 0.06018625944852829,
      "learning_rate": 1.4990652067746808e-06,
      "loss": 0.0006,
      "step": 240800
    },
    {
      "epoch": 1.544480539263003,
      "grad_norm": 0.03753812611103058,
      "learning_rate": 1.498665623146956e-06,
      "loss": 0.0007,
      "step": 240810
    },
    {
      "epoch": 1.544544676156789,
      "grad_norm": 0.10939286649227142,
      "learning_rate": 1.4982660833934521e-06,
      "loss": 0.0009,
      "step": 240820
    },
    {
      "epoch": 1.544608813050575,
      "grad_norm": 0.05444050952792168,
      "learning_rate": 1.4978665875191784e-06,
      "loss": 0.0012,
      "step": 240830
    },
    {
      "epoch": 1.5446729499443612,
      "grad_norm": 0.03752541542053223,
      "learning_rate": 1.4974671355291393e-06,
      "loss": 0.0017,
      "step": 240840
    },
    {
      "epoch": 1.5447370868381474,
      "grad_norm": 0.0661284327507019,
      "learning_rate": 1.49706772742834e-06,
      "loss": 0.001,
      "step": 240850
    },
    {
      "epoch": 1.5448012237319335,
      "grad_norm": 0.22921940684318542,
      "learning_rate": 1.4966683632217843e-06,
      "loss": 0.0011,
      "step": 240860
    },
    {
      "epoch": 1.5448653606257197,
      "grad_norm": 0.16091661155223846,
      "learning_rate": 1.496269042914479e-06,
      "loss": 0.001,
      "step": 240870
    },
    {
      "epoch": 1.5449294975195056,
      "grad_norm": 0.19941392540931702,
      "learning_rate": 1.4958697665114268e-06,
      "loss": 0.0015,
      "step": 240880
    },
    {
      "epoch": 1.5449936344132917,
      "grad_norm": 0.05221044272184372,
      "learning_rate": 1.4954705340176312e-06,
      "loss": 0.0013,
      "step": 240890
    },
    {
      "epoch": 1.5450577713070777,
      "grad_norm": 0.12489048391580582,
      "learning_rate": 1.4950713454380922e-06,
      "loss": 0.0013,
      "step": 240900
    },
    {
      "epoch": 1.5451219082008638,
      "grad_norm": 0.0692276880145073,
      "learning_rate": 1.4946722007778164e-06,
      "loss": 0.0007,
      "step": 240910
    },
    {
      "epoch": 1.54518604509465,
      "grad_norm": 0.07412971556186676,
      "learning_rate": 1.4942731000418026e-06,
      "loss": 0.0017,
      "step": 240920
    },
    {
      "epoch": 1.5452501819884361,
      "grad_norm": 0.12431248277425766,
      "learning_rate": 1.4938740432350525e-06,
      "loss": 0.0008,
      "step": 240930
    },
    {
      "epoch": 1.5453143188822223,
      "grad_norm": 0.019095465540885925,
      "learning_rate": 1.493475030362565e-06,
      "loss": 0.0011,
      "step": 240940
    },
    {
      "epoch": 1.5453784557760084,
      "grad_norm": 0.09157022833824158,
      "learning_rate": 1.4930760614293432e-06,
      "loss": 0.0009,
      "step": 240950
    },
    {
      "epoch": 1.5454425926697946,
      "grad_norm": 0.30129361152648926,
      "learning_rate": 1.492677136440384e-06,
      "loss": 0.0017,
      "step": 240960
    },
    {
      "epoch": 1.5455067295635805,
      "grad_norm": 0.07098261266946793,
      "learning_rate": 1.4922782554006859e-06,
      "loss": 0.0011,
      "step": 240970
    },
    {
      "epoch": 1.5455708664573666,
      "grad_norm": 0.0788029134273529,
      "learning_rate": 1.4918794183152497e-06,
      "loss": 0.0027,
      "step": 240980
    },
    {
      "epoch": 1.5456350033511526,
      "grad_norm": 0.004866393748670816,
      "learning_rate": 1.4914806251890717e-06,
      "loss": 0.0018,
      "step": 240990
    },
    {
      "epoch": 1.5456991402449387,
      "grad_norm": 0.033007655292749405,
      "learning_rate": 1.491081876027149e-06,
      "loss": 0.0014,
      "step": 241000
    },
    {
      "epoch": 1.5457632771387249,
      "grad_norm": 0.08490005135536194,
      "learning_rate": 1.4906831708344767e-06,
      "loss": 0.0015,
      "step": 241010
    },
    {
      "epoch": 1.545827414032511,
      "grad_norm": 0.034369032829999924,
      "learning_rate": 1.4902845096160534e-06,
      "loss": 0.0021,
      "step": 241020
    },
    {
      "epoch": 1.5458915509262972,
      "grad_norm": 0.17891032993793488,
      "learning_rate": 1.489885892376874e-06,
      "loss": 0.0013,
      "step": 241030
    },
    {
      "epoch": 1.5459556878200833,
      "grad_norm": 0.031783584505319595,
      "learning_rate": 1.4894873191219329e-06,
      "loss": 0.0009,
      "step": 241040
    },
    {
      "epoch": 1.5460198247138692,
      "grad_norm": 0.06378553062677383,
      "learning_rate": 1.4890887898562228e-06,
      "loss": 0.0005,
      "step": 241050
    },
    {
      "epoch": 1.5460839616076554,
      "grad_norm": 0.10979326069355011,
      "learning_rate": 1.4886903045847412e-06,
      "loss": 0.0021,
      "step": 241060
    },
    {
      "epoch": 1.5461480985014415,
      "grad_norm": 0.015144680626690388,
      "learning_rate": 1.4882918633124794e-06,
      "loss": 0.0011,
      "step": 241070
    },
    {
      "epoch": 1.5462122353952275,
      "grad_norm": 0.07192160934209824,
      "learning_rate": 1.4878934660444305e-06,
      "loss": 0.0025,
      "step": 241080
    },
    {
      "epoch": 1.5462763722890136,
      "grad_norm": 0.14805211126804352,
      "learning_rate": 1.4874951127855847e-06,
      "loss": 0.002,
      "step": 241090
    },
    {
      "epoch": 1.5463405091827997,
      "grad_norm": 0.048894334584474564,
      "learning_rate": 1.4870968035409371e-06,
      "loss": 0.0007,
      "step": 241100
    },
    {
      "epoch": 1.546404646076586,
      "grad_norm": 0.04320673272013664,
      "learning_rate": 1.4866985383154775e-06,
      "loss": 0.0019,
      "step": 241110
    },
    {
      "epoch": 1.546468782970372,
      "grad_norm": 0.25919121503829956,
      "learning_rate": 1.486300317114195e-06,
      "loss": 0.0071,
      "step": 241120
    },
    {
      "epoch": 1.5465329198641582,
      "grad_norm": 0.10468928515911102,
      "learning_rate": 1.4859021399420813e-06,
      "loss": 0.0015,
      "step": 241130
    },
    {
      "epoch": 1.5465970567579441,
      "grad_norm": 0.329969197511673,
      "learning_rate": 1.485504006804126e-06,
      "loss": 0.0014,
      "step": 241140
    },
    {
      "epoch": 1.5466611936517303,
      "grad_norm": 0.24276787042617798,
      "learning_rate": 1.4851059177053167e-06,
      "loss": 0.0009,
      "step": 241150
    },
    {
      "epoch": 1.5467253305455162,
      "grad_norm": 0.011052812449634075,
      "learning_rate": 1.4847078726506409e-06,
      "loss": 0.0013,
      "step": 241160
    },
    {
      "epoch": 1.5467894674393023,
      "grad_norm": 0.06549164652824402,
      "learning_rate": 1.4843098716450893e-06,
      "loss": 0.0008,
      "step": 241170
    },
    {
      "epoch": 1.5468536043330885,
      "grad_norm": 0.24330514669418335,
      "learning_rate": 1.483911914693648e-06,
      "loss": 0.002,
      "step": 241180
    },
    {
      "epoch": 1.5469177412268746,
      "grad_norm": 0.3681797981262207,
      "learning_rate": 1.4835140018013033e-06,
      "loss": 0.0023,
      "step": 241190
    },
    {
      "epoch": 1.5469818781206608,
      "grad_norm": 0.012566491961479187,
      "learning_rate": 1.4831161329730392e-06,
      "loss": 0.0016,
      "step": 241200
    },
    {
      "epoch": 1.547046015014447,
      "grad_norm": 0.128724604845047,
      "learning_rate": 1.4827183082138457e-06,
      "loss": 0.0009,
      "step": 241210
    },
    {
      "epoch": 1.547110151908233,
      "grad_norm": 0.12924446165561676,
      "learning_rate": 1.482320527528705e-06,
      "loss": 0.001,
      "step": 241220
    },
    {
      "epoch": 1.547174288802019,
      "grad_norm": 0.04774237424135208,
      "learning_rate": 1.4819227909226025e-06,
      "loss": 0.001,
      "step": 241230
    },
    {
      "epoch": 1.5472384256958052,
      "grad_norm": 0.01152403000742197,
      "learning_rate": 1.4815250984005203e-06,
      "loss": 0.001,
      "step": 241240
    },
    {
      "epoch": 1.547302562589591,
      "grad_norm": 0.014252632856369019,
      "learning_rate": 1.4811274499674444e-06,
      "loss": 0.0016,
      "step": 241250
    },
    {
      "epoch": 1.5473666994833772,
      "grad_norm": 0.16456815600395203,
      "learning_rate": 1.480729845628357e-06,
      "loss": 0.0006,
      "step": 241260
    },
    {
      "epoch": 1.5474308363771634,
      "grad_norm": 0.013565735891461372,
      "learning_rate": 1.480332285388238e-06,
      "loss": 0.001,
      "step": 241270
    },
    {
      "epoch": 1.5474949732709495,
      "grad_norm": 0.09282597154378891,
      "learning_rate": 1.4799347692520722e-06,
      "loss": 0.0016,
      "step": 241280
    },
    {
      "epoch": 1.5475591101647357,
      "grad_norm": 0.028873804956674576,
      "learning_rate": 1.4795372972248378e-06,
      "loss": 0.001,
      "step": 241290
    },
    {
      "epoch": 1.5476232470585218,
      "grad_norm": 0.08476100862026215,
      "learning_rate": 1.4791398693115195e-06,
      "loss": 0.0012,
      "step": 241300
    },
    {
      "epoch": 1.5476873839523078,
      "grad_norm": 0.22884000837802887,
      "learning_rate": 1.478742485517094e-06,
      "loss": 0.001,
      "step": 241310
    },
    {
      "epoch": 1.547751520846094,
      "grad_norm": 0.019175561144948006,
      "learning_rate": 1.4783451458465409e-06,
      "loss": 0.0029,
      "step": 241320
    },
    {
      "epoch": 1.5478156577398798,
      "grad_norm": 0.029204215854406357,
      "learning_rate": 1.477947850304841e-06,
      "loss": 0.001,
      "step": 241330
    },
    {
      "epoch": 1.547879794633666,
      "grad_norm": 0.00403748732060194,
      "learning_rate": 1.4775505988969723e-06,
      "loss": 0.0006,
      "step": 241340
    },
    {
      "epoch": 1.5479439315274521,
      "grad_norm": 0.040969155728816986,
      "learning_rate": 1.477153391627912e-06,
      "loss": 0.001,
      "step": 241350
    },
    {
      "epoch": 1.5480080684212383,
      "grad_norm": 0.0407954677939415,
      "learning_rate": 1.4767562285026355e-06,
      "loss": 0.0018,
      "step": 241360
    },
    {
      "epoch": 1.5480722053150244,
      "grad_norm": 0.10405800491571426,
      "learning_rate": 1.4763591095261233e-06,
      "loss": 0.0011,
      "step": 241370
    },
    {
      "epoch": 1.5481363422088106,
      "grad_norm": 0.25405964255332947,
      "learning_rate": 1.4759620347033493e-06,
      "loss": 0.002,
      "step": 241380
    },
    {
      "epoch": 1.5482004791025967,
      "grad_norm": 0.13294708728790283,
      "learning_rate": 1.4755650040392888e-06,
      "loss": 0.0012,
      "step": 241390
    },
    {
      "epoch": 1.5482646159963827,
      "grad_norm": 0.05426466092467308,
      "learning_rate": 1.4751680175389188e-06,
      "loss": 0.001,
      "step": 241400
    },
    {
      "epoch": 1.5483287528901688,
      "grad_norm": 0.09514619410037994,
      "learning_rate": 1.4747710752072125e-06,
      "loss": 0.0011,
      "step": 241410
    },
    {
      "epoch": 1.5483928897839547,
      "grad_norm": 0.16939108073711395,
      "learning_rate": 1.4743741770491443e-06,
      "loss": 0.0015,
      "step": 241420
    },
    {
      "epoch": 1.5484570266777409,
      "grad_norm": 0.14232565462589264,
      "learning_rate": 1.4739773230696858e-06,
      "loss": 0.0015,
      "step": 241430
    },
    {
      "epoch": 1.548521163571527,
      "grad_norm": 0.08041399717330933,
      "learning_rate": 1.4735805132738135e-06,
      "loss": 0.0009,
      "step": 241440
    },
    {
      "epoch": 1.5485853004653132,
      "grad_norm": 0.06873752921819687,
      "learning_rate": 1.4731837476664967e-06,
      "loss": 0.0006,
      "step": 241450
    },
    {
      "epoch": 1.5486494373590993,
      "grad_norm": 0.0607171431183815,
      "learning_rate": 1.472787026252709e-06,
      "loss": 0.0017,
      "step": 241460
    },
    {
      "epoch": 1.5487135742528855,
      "grad_norm": 0.07531336694955826,
      "learning_rate": 1.4723903490374186e-06,
      "loss": 0.0009,
      "step": 241470
    },
    {
      "epoch": 1.5487777111466716,
      "grad_norm": 0.1684890240430832,
      "learning_rate": 1.4719937160256004e-06,
      "loss": 0.0012,
      "step": 241480
    },
    {
      "epoch": 1.5488418480404575,
      "grad_norm": 0.05508307367563248,
      "learning_rate": 1.4715971272222217e-06,
      "loss": 0.0009,
      "step": 241490
    },
    {
      "epoch": 1.5489059849342437,
      "grad_norm": 0.15135546028614044,
      "learning_rate": 1.471200582632253e-06,
      "loss": 0.0014,
      "step": 241500
    },
    {
      "epoch": 1.5489701218280296,
      "grad_norm": 0.005747431889176369,
      "learning_rate": 1.4708040822606618e-06,
      "loss": 0.0005,
      "step": 241510
    },
    {
      "epoch": 1.5490342587218158,
      "grad_norm": 0.12473436444997787,
      "learning_rate": 1.4704076261124183e-06,
      "loss": 0.0011,
      "step": 241520
    },
    {
      "epoch": 1.549098395615602,
      "grad_norm": 0.032045476138591766,
      "learning_rate": 1.4700112141924905e-06,
      "loss": 0.0011,
      "step": 241530
    },
    {
      "epoch": 1.549162532509388,
      "grad_norm": 0.13783413171768188,
      "learning_rate": 1.4696148465058436e-06,
      "loss": 0.0015,
      "step": 241540
    },
    {
      "epoch": 1.5492266694031742,
      "grad_norm": 0.06455527991056442,
      "learning_rate": 1.4692185230574475e-06,
      "loss": 0.0014,
      "step": 241550
    },
    {
      "epoch": 1.5492908062969604,
      "grad_norm": 0.04851328209042549,
      "learning_rate": 1.4688222438522658e-06,
      "loss": 0.0014,
      "step": 241560
    },
    {
      "epoch": 1.5493549431907463,
      "grad_norm": 0.02985536865890026,
      "learning_rate": 1.4684260088952663e-06,
      "loss": 0.0013,
      "step": 241570
    },
    {
      "epoch": 1.5494190800845324,
      "grad_norm": 0.013616991229355335,
      "learning_rate": 1.4680298181914105e-06,
      "loss": 0.0035,
      "step": 241580
    },
    {
      "epoch": 1.5494832169783184,
      "grad_norm": 0.001772357034496963,
      "learning_rate": 1.4676336717456668e-06,
      "loss": 0.0009,
      "step": 241590
    },
    {
      "epoch": 1.5495473538721045,
      "grad_norm": 0.07002270221710205,
      "learning_rate": 1.4672375695629982e-06,
      "loss": 0.0012,
      "step": 241600
    },
    {
      "epoch": 1.5496114907658907,
      "grad_norm": 0.07833683490753174,
      "learning_rate": 1.4668415116483675e-06,
      "loss": 0.0012,
      "step": 241610
    },
    {
      "epoch": 1.5496756276596768,
      "grad_norm": 0.3495336174964905,
      "learning_rate": 1.4664454980067362e-06,
      "loss": 0.002,
      "step": 241620
    },
    {
      "epoch": 1.549739764553463,
      "grad_norm": 0.06402402371168137,
      "learning_rate": 1.4660495286430699e-06,
      "loss": 0.0016,
      "step": 241630
    },
    {
      "epoch": 1.549803901447249,
      "grad_norm": 0.05305173620581627,
      "learning_rate": 1.4656536035623286e-06,
      "loss": 0.0013,
      "step": 241640
    },
    {
      "epoch": 1.5498680383410353,
      "grad_norm": 0.1703844666481018,
      "learning_rate": 1.4652577227694737e-06,
      "loss": 0.001,
      "step": 241650
    },
    {
      "epoch": 1.5499321752348212,
      "grad_norm": 0.01733911968767643,
      "learning_rate": 1.4648618862694636e-06,
      "loss": 0.0009,
      "step": 241660
    },
    {
      "epoch": 1.5499963121286073,
      "grad_norm": 0.08173426240682602,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.0012,
      "step": 241670
    },
    {
      "epoch": 1.5500604490223933,
      "grad_norm": 0.006655998528003693,
      "learning_rate": 1.4640703461678285e-06,
      "loss": 0.0007,
      "step": 241680
    },
    {
      "epoch": 1.5501245859161794,
      "grad_norm": 0.06824464350938797,
      "learning_rate": 1.4636746425761183e-06,
      "loss": 0.001,
      "step": 241690
    },
    {
      "epoch": 1.5501887228099656,
      "grad_norm": 0.04127271845936775,
      "learning_rate": 1.463278983297094e-06,
      "loss": 0.0004,
      "step": 241700
    },
    {
      "epoch": 1.5502528597037517,
      "grad_norm": 0.08334328234195709,
      "learning_rate": 1.4628833683357114e-06,
      "loss": 0.0016,
      "step": 241710
    },
    {
      "epoch": 1.5503169965975379,
      "grad_norm": 0.07784296572208405,
      "learning_rate": 1.4624877976969282e-06,
      "loss": 0.001,
      "step": 241720
    },
    {
      "epoch": 1.550381133491324,
      "grad_norm": 0.19750766456127167,
      "learning_rate": 1.4620922713856994e-06,
      "loss": 0.0027,
      "step": 241730
    },
    {
      "epoch": 1.55044527038511,
      "grad_norm": 0.2604537606239319,
      "learning_rate": 1.4616967894069845e-06,
      "loss": 0.0015,
      "step": 241740
    },
    {
      "epoch": 1.550509407278896,
      "grad_norm": 0.08659055083990097,
      "learning_rate": 1.4613013517657377e-06,
      "loss": 0.0009,
      "step": 241750
    },
    {
      "epoch": 1.5505735441726822,
      "grad_norm": 0.019921543076634407,
      "learning_rate": 1.4609059584669139e-06,
      "loss": 0.0008,
      "step": 241760
    },
    {
      "epoch": 1.5506376810664682,
      "grad_norm": 0.001479236176237464,
      "learning_rate": 1.4605106095154664e-06,
      "loss": 0.0007,
      "step": 241770
    },
    {
      "epoch": 1.5507018179602543,
      "grad_norm": 0.07658054679632187,
      "learning_rate": 1.4601153049163525e-06,
      "loss": 0.0012,
      "step": 241780
    },
    {
      "epoch": 1.5507659548540405,
      "grad_norm": 0.10939273238182068,
      "learning_rate": 1.4597200446745235e-06,
      "loss": 0.0015,
      "step": 241790
    },
    {
      "epoch": 1.5508300917478266,
      "grad_norm": 0.008693967945873737,
      "learning_rate": 1.4593248287949324e-06,
      "loss": 0.0003,
      "step": 241800
    },
    {
      "epoch": 1.5508942286416127,
      "grad_norm": 0.2519848942756653,
      "learning_rate": 1.4589296572825302e-06,
      "loss": 0.0017,
      "step": 241810
    },
    {
      "epoch": 1.550958365535399,
      "grad_norm": 0.022765902802348137,
      "learning_rate": 1.4585345301422715e-06,
      "loss": 0.001,
      "step": 241820
    },
    {
      "epoch": 1.5510225024291848,
      "grad_norm": 0.02760549820959568,
      "learning_rate": 1.4581394473791067e-06,
      "loss": 0.0014,
      "step": 241830
    },
    {
      "epoch": 1.551086639322971,
      "grad_norm": 0.15331031382083893,
      "learning_rate": 1.4577444089979837e-06,
      "loss": 0.0014,
      "step": 241840
    },
    {
      "epoch": 1.551150776216757,
      "grad_norm": 0.03634566441178322,
      "learning_rate": 1.457349415003857e-06,
      "loss": 0.0019,
      "step": 241850
    },
    {
      "epoch": 1.551214913110543,
      "grad_norm": 0.07276799529790878,
      "learning_rate": 1.4569544654016737e-06,
      "loss": 0.001,
      "step": 241860
    },
    {
      "epoch": 1.5512790500043292,
      "grad_norm": 0.09183241426944733,
      "learning_rate": 1.4565595601963833e-06,
      "loss": 0.0015,
      "step": 241870
    },
    {
      "epoch": 1.5513431868981153,
      "grad_norm": 0.03877013549208641,
      "learning_rate": 1.4561646993929323e-06,
      "loss": 0.0007,
      "step": 241880
    },
    {
      "epoch": 1.5514073237919015,
      "grad_norm": 0.03545617312192917,
      "learning_rate": 1.4557698829962724e-06,
      "loss": 0.0014,
      "step": 241890
    },
    {
      "epoch": 1.5514714606856876,
      "grad_norm": 0.29456937313079834,
      "learning_rate": 1.4553751110113484e-06,
      "loss": 0.0008,
      "step": 241900
    },
    {
      "epoch": 1.5515355975794738,
      "grad_norm": 0.20917344093322754,
      "learning_rate": 1.4549803834431076e-06,
      "loss": 0.0014,
      "step": 241910
    },
    {
      "epoch": 1.5515997344732597,
      "grad_norm": 0.03617027401924133,
      "learning_rate": 1.4545857002964948e-06,
      "loss": 0.0008,
      "step": 241920
    },
    {
      "epoch": 1.5516638713670459,
      "grad_norm": 0.1423276662826538,
      "learning_rate": 1.4541910615764587e-06,
      "loss": 0.0012,
      "step": 241930
    },
    {
      "epoch": 1.5517280082608318,
      "grad_norm": 0.08496689796447754,
      "learning_rate": 1.4537964672879422e-06,
      "loss": 0.0016,
      "step": 241940
    },
    {
      "epoch": 1.551792145154618,
      "grad_norm": 0.049270179122686386,
      "learning_rate": 1.453401917435891e-06,
      "loss": 0.0013,
      "step": 241950
    },
    {
      "epoch": 1.551856282048404,
      "grad_norm": 0.039366334676742554,
      "learning_rate": 1.4530074120252468e-06,
      "loss": 0.0009,
      "step": 241960
    },
    {
      "epoch": 1.5519204189421902,
      "grad_norm": 0.1436959207057953,
      "learning_rate": 1.452612951060956e-06,
      "loss": 0.002,
      "step": 241970
    },
    {
      "epoch": 1.5519845558359764,
      "grad_norm": 0.05382394790649414,
      "learning_rate": 1.4522185345479606e-06,
      "loss": 0.0019,
      "step": 241980
    },
    {
      "epoch": 1.5520486927297625,
      "grad_norm": 0.12463415414094925,
      "learning_rate": 1.451824162491201e-06,
      "loss": 0.0008,
      "step": 241990
    },
    {
      "epoch": 1.5521128296235485,
      "grad_norm": 0.15907691419124603,
      "learning_rate": 1.451429834895622e-06,
      "loss": 0.0012,
      "step": 242000
    },
    {
      "epoch": 1.5521769665173346,
      "grad_norm": 0.0672299712896347,
      "learning_rate": 1.4510355517661628e-06,
      "loss": 0.0013,
      "step": 242010
    },
    {
      "epoch": 1.5522411034111205,
      "grad_norm": 0.1971874237060547,
      "learning_rate": 1.4506413131077652e-06,
      "loss": 0.0022,
      "step": 242020
    },
    {
      "epoch": 1.5523052403049067,
      "grad_norm": 0.07928959280252457,
      "learning_rate": 1.4502471189253665e-06,
      "loss": 0.0023,
      "step": 242030
    },
    {
      "epoch": 1.5523693771986928,
      "grad_norm": 0.019396668300032616,
      "learning_rate": 1.44985296922391e-06,
      "loss": 0.0013,
      "step": 242040
    },
    {
      "epoch": 1.552433514092479,
      "grad_norm": 0.010010046884417534,
      "learning_rate": 1.4494588640083334e-06,
      "loss": 0.0014,
      "step": 242050
    },
    {
      "epoch": 1.5524976509862651,
      "grad_norm": 0.12092921137809753,
      "learning_rate": 1.449064803283573e-06,
      "loss": 0.0012,
      "step": 242060
    },
    {
      "epoch": 1.5525617878800513,
      "grad_norm": 0.2388553023338318,
      "learning_rate": 1.4486707870545701e-06,
      "loss": 0.0019,
      "step": 242070
    },
    {
      "epoch": 1.5526259247738374,
      "grad_norm": 0.004728635307401419,
      "learning_rate": 1.4482768153262583e-06,
      "loss": 0.0015,
      "step": 242080
    },
    {
      "epoch": 1.5526900616676234,
      "grad_norm": 0.10769958049058914,
      "learning_rate": 1.4478828881035783e-06,
      "loss": 0.0009,
      "step": 242090
    },
    {
      "epoch": 1.5527541985614095,
      "grad_norm": 0.1482594609260559,
      "learning_rate": 1.4474890053914647e-06,
      "loss": 0.0011,
      "step": 242100
    },
    {
      "epoch": 1.5528183354551954,
      "grad_norm": 0.05761986970901489,
      "learning_rate": 1.4470951671948508e-06,
      "loss": 0.001,
      "step": 242110
    },
    {
      "epoch": 1.5528824723489816,
      "grad_norm": 0.008168700151145458,
      "learning_rate": 1.4467013735186752e-06,
      "loss": 0.0012,
      "step": 242120
    },
    {
      "epoch": 1.5529466092427677,
      "grad_norm": 0.09967954456806183,
      "learning_rate": 1.4463076243678714e-06,
      "loss": 0.0017,
      "step": 242130
    },
    {
      "epoch": 1.5530107461365539,
      "grad_norm": 0.04914829134941101,
      "learning_rate": 1.4459139197473725e-06,
      "loss": 0.0009,
      "step": 242140
    },
    {
      "epoch": 1.55307488303034,
      "grad_norm": 0.009536635130643845,
      "learning_rate": 1.445520259662111e-06,
      "loss": 0.0011,
      "step": 242150
    },
    {
      "epoch": 1.5531390199241262,
      "grad_norm": 0.08600223809480667,
      "learning_rate": 1.445126644117022e-06,
      "loss": 0.001,
      "step": 242160
    },
    {
      "epoch": 1.553203156817912,
      "grad_norm": 0.034827813506126404,
      "learning_rate": 1.4447330731170372e-06,
      "loss": 0.0005,
      "step": 242170
    },
    {
      "epoch": 1.5532672937116982,
      "grad_norm": 0.07760866731405258,
      "learning_rate": 1.444339546667088e-06,
      "loss": 0.002,
      "step": 242180
    },
    {
      "epoch": 1.5533314306054844,
      "grad_norm": 0.004131306428462267,
      "learning_rate": 1.4439460647721032e-06,
      "loss": 0.0008,
      "step": 242190
    },
    {
      "epoch": 1.5533955674992703,
      "grad_norm": 0.15165965259075165,
      "learning_rate": 1.4435526274370176e-06,
      "loss": 0.0013,
      "step": 242200
    },
    {
      "epoch": 1.5534597043930565,
      "grad_norm": 0.07331079989671707,
      "learning_rate": 1.4431592346667595e-06,
      "loss": 0.0006,
      "step": 242210
    },
    {
      "epoch": 1.5535238412868426,
      "grad_norm": 0.10047654062509537,
      "learning_rate": 1.4427658864662585e-06,
      "loss": 0.0015,
      "step": 242220
    },
    {
      "epoch": 1.5535879781806288,
      "grad_norm": 0.1110883429646492,
      "learning_rate": 1.4423725828404411e-06,
      "loss": 0.0015,
      "step": 242230
    },
    {
      "epoch": 1.553652115074415,
      "grad_norm": 0.0955362617969513,
      "learning_rate": 1.4419793237942397e-06,
      "loss": 0.0012,
      "step": 242240
    },
    {
      "epoch": 1.553716251968201,
      "grad_norm": 0.043106190860271454,
      "learning_rate": 1.4415861093325805e-06,
      "loss": 0.0008,
      "step": 242250
    },
    {
      "epoch": 1.553780388861987,
      "grad_norm": 0.008243992924690247,
      "learning_rate": 1.4411929394603884e-06,
      "loss": 0.0025,
      "step": 242260
    },
    {
      "epoch": 1.5538445257557731,
      "grad_norm": 0.05364123731851578,
      "learning_rate": 1.440799814182594e-06,
      "loss": 0.0016,
      "step": 242270
    },
    {
      "epoch": 1.553908662649559,
      "grad_norm": 0.043370697647333145,
      "learning_rate": 1.4404067335041216e-06,
      "loss": 0.0013,
      "step": 242280
    },
    {
      "epoch": 1.5539727995433452,
      "grad_norm": 0.04179252311587334,
      "learning_rate": 1.4400136974298972e-06,
      "loss": 0.0011,
      "step": 242290
    },
    {
      "epoch": 1.5540369364371314,
      "grad_norm": 0.00294058327563107,
      "learning_rate": 1.4396207059648438e-06,
      "loss": 0.0014,
      "step": 242300
    },
    {
      "epoch": 1.5541010733309175,
      "grad_norm": 0.07783500850200653,
      "learning_rate": 1.4392277591138886e-06,
      "loss": 0.0018,
      "step": 242310
    },
    {
      "epoch": 1.5541652102247037,
      "grad_norm": 0.10862353444099426,
      "learning_rate": 1.438834856881955e-06,
      "loss": 0.0007,
      "step": 242320
    },
    {
      "epoch": 1.5542293471184898,
      "grad_norm": 0.12284133583307266,
      "learning_rate": 1.438441999273965e-06,
      "loss": 0.0015,
      "step": 242330
    },
    {
      "epoch": 1.554293484012276,
      "grad_norm": 0.05826522037386894,
      "learning_rate": 1.4380491862948415e-06,
      "loss": 0.0015,
      "step": 242340
    },
    {
      "epoch": 1.5543576209060619,
      "grad_norm": 0.0899071991443634,
      "learning_rate": 1.4376564179495085e-06,
      "loss": 0.0033,
      "step": 242350
    },
    {
      "epoch": 1.554421757799848,
      "grad_norm": 0.03002636320888996,
      "learning_rate": 1.4372636942428863e-06,
      "loss": 0.0022,
      "step": 242360
    },
    {
      "epoch": 1.554485894693634,
      "grad_norm": 0.16218431293964386,
      "learning_rate": 1.436871015179897e-06,
      "loss": 0.0014,
      "step": 242370
    },
    {
      "epoch": 1.5545500315874201,
      "grad_norm": 0.02801361121237278,
      "learning_rate": 1.436478380765458e-06,
      "loss": 0.0011,
      "step": 242380
    },
    {
      "epoch": 1.5546141684812063,
      "grad_norm": 0.07750725001096725,
      "learning_rate": 1.436085791004494e-06,
      "loss": 0.0015,
      "step": 242390
    },
    {
      "epoch": 1.5546783053749924,
      "grad_norm": 0.15338924527168274,
      "learning_rate": 1.4356932459019218e-06,
      "loss": 0.0015,
      "step": 242400
    },
    {
      "epoch": 1.5547424422687786,
      "grad_norm": 0.21196606755256653,
      "learning_rate": 1.4353007454626588e-06,
      "loss": 0.001,
      "step": 242410
    },
    {
      "epoch": 1.5548065791625647,
      "grad_norm": 0.0607130192220211,
      "learning_rate": 1.4349082896916273e-06,
      "loss": 0.001,
      "step": 242420
    },
    {
      "epoch": 1.5548707160563506,
      "grad_norm": 0.12506140768527985,
      "learning_rate": 1.434515878593742e-06,
      "loss": 0.0018,
      "step": 242430
    },
    {
      "epoch": 1.5549348529501368,
      "grad_norm": 0.11620277166366577,
      "learning_rate": 1.4341235121739216e-06,
      "loss": 0.002,
      "step": 242440
    },
    {
      "epoch": 1.5549989898439227,
      "grad_norm": 0.12097823619842529,
      "learning_rate": 1.4337311904370804e-06,
      "loss": 0.0019,
      "step": 242450
    },
    {
      "epoch": 1.5550631267377089,
      "grad_norm": 0.03330722823739052,
      "learning_rate": 1.4333389133881375e-06,
      "loss": 0.0025,
      "step": 242460
    },
    {
      "epoch": 1.555127263631495,
      "grad_norm": 0.017193831503391266,
      "learning_rate": 1.432946681032007e-06,
      "loss": 0.0005,
      "step": 242470
    },
    {
      "epoch": 1.5551914005252812,
      "grad_norm": 0.08073843270540237,
      "learning_rate": 1.4325544933736047e-06,
      "loss": 0.0008,
      "step": 242480
    },
    {
      "epoch": 1.5552555374190673,
      "grad_norm": 0.0536016970872879,
      "learning_rate": 1.4321623504178416e-06,
      "loss": 0.0012,
      "step": 242490
    },
    {
      "epoch": 1.5553196743128535,
      "grad_norm": 0.004435609094798565,
      "learning_rate": 1.4317702521696364e-06,
      "loss": 0.0014,
      "step": 242500
    },
    {
      "epoch": 1.5553838112066396,
      "grad_norm": 0.1516411006450653,
      "learning_rate": 1.4313781986338998e-06,
      "loss": 0.0011,
      "step": 242510
    },
    {
      "epoch": 1.5554479481004255,
      "grad_norm": 0.07027588039636612,
      "learning_rate": 1.4309861898155453e-06,
      "loss": 0.0015,
      "step": 242520
    },
    {
      "epoch": 1.5555120849942117,
      "grad_norm": 0.06725809723138809,
      "learning_rate": 1.430594225719482e-06,
      "loss": 0.001,
      "step": 242530
    },
    {
      "epoch": 1.5555762218879976,
      "grad_norm": 0.16939279437065125,
      "learning_rate": 1.4302023063506265e-06,
      "loss": 0.0015,
      "step": 242540
    },
    {
      "epoch": 1.5556403587817837,
      "grad_norm": 0.0354507751762867,
      "learning_rate": 1.4298104317138873e-06,
      "loss": 0.0013,
      "step": 242550
    },
    {
      "epoch": 1.55570449567557,
      "grad_norm": 0.019089942798018456,
      "learning_rate": 1.4294186018141732e-06,
      "loss": 0.0006,
      "step": 242560
    },
    {
      "epoch": 1.555768632569356,
      "grad_norm": 0.03581133484840393,
      "learning_rate": 1.4290268166563975e-06,
      "loss": 0.0012,
      "step": 242570
    },
    {
      "epoch": 1.5558327694631422,
      "grad_norm": 0.05963753163814545,
      "learning_rate": 1.4286350762454682e-06,
      "loss": 0.0015,
      "step": 242580
    },
    {
      "epoch": 1.5558969063569283,
      "grad_norm": 0.07994631677865982,
      "learning_rate": 1.4282433805862933e-06,
      "loss": 0.001,
      "step": 242590
    },
    {
      "epoch": 1.5559610432507143,
      "grad_norm": 0.09030909091234207,
      "learning_rate": 1.4278517296837807e-06,
      "loss": 0.0021,
      "step": 242600
    },
    {
      "epoch": 1.5560251801445004,
      "grad_norm": 0.042248863726854324,
      "learning_rate": 1.4274601235428399e-06,
      "loss": 0.001,
      "step": 242610
    },
    {
      "epoch": 1.5560893170382866,
      "grad_norm": 0.11223195493221283,
      "learning_rate": 1.4270685621683772e-06,
      "loss": 0.0015,
      "step": 242620
    },
    {
      "epoch": 1.5561534539320725,
      "grad_norm": 0.003470065537840128,
      "learning_rate": 1.4266770455652984e-06,
      "loss": 0.0015,
      "step": 242630
    },
    {
      "epoch": 1.5562175908258586,
      "grad_norm": 0.09900204092264175,
      "learning_rate": 1.426285573738509e-06,
      "loss": 0.0006,
      "step": 242640
    },
    {
      "epoch": 1.5562817277196448,
      "grad_norm": 0.06904611736536026,
      "learning_rate": 1.4258941466929171e-06,
      "loss": 0.0003,
      "step": 242650
    },
    {
      "epoch": 1.556345864613431,
      "grad_norm": 0.06262891739606857,
      "learning_rate": 1.4255027644334257e-06,
      "loss": 0.0012,
      "step": 242660
    },
    {
      "epoch": 1.556410001507217,
      "grad_norm": 0.11479669064283371,
      "learning_rate": 1.4251114269649379e-06,
      "loss": 0.001,
      "step": 242670
    },
    {
      "epoch": 1.5564741384010032,
      "grad_norm": 0.19921335577964783,
      "learning_rate": 1.4247201342923605e-06,
      "loss": 0.0015,
      "step": 242680
    },
    {
      "epoch": 1.5565382752947892,
      "grad_norm": 0.023834677413105965,
      "learning_rate": 1.4243288864205945e-06,
      "loss": 0.001,
      "step": 242690
    },
    {
      "epoch": 1.5566024121885753,
      "grad_norm": 0.05833800137042999,
      "learning_rate": 1.4239376833545437e-06,
      "loss": 0.001,
      "step": 242700
    },
    {
      "epoch": 1.5566665490823612,
      "grad_norm": 0.16412882506847382,
      "learning_rate": 1.4235465250991076e-06,
      "loss": 0.0015,
      "step": 242710
    },
    {
      "epoch": 1.5567306859761474,
      "grad_norm": 0.03514918312430382,
      "learning_rate": 1.4231554116591912e-06,
      "loss": 0.0013,
      "step": 242720
    },
    {
      "epoch": 1.5567948228699335,
      "grad_norm": 0.026855146512389183,
      "learning_rate": 1.4227643430396938e-06,
      "loss": 0.001,
      "step": 242730
    },
    {
      "epoch": 1.5568589597637197,
      "grad_norm": 0.047292310744524,
      "learning_rate": 1.422373319245516e-06,
      "loss": 0.0013,
      "step": 242740
    },
    {
      "epoch": 1.5569230966575058,
      "grad_norm": 0.07855642586946487,
      "learning_rate": 1.4219823402815559e-06,
      "loss": 0.0014,
      "step": 242750
    },
    {
      "epoch": 1.556987233551292,
      "grad_norm": 0.02176888845860958,
      "learning_rate": 1.4215914061527152e-06,
      "loss": 0.0011,
      "step": 242760
    },
    {
      "epoch": 1.5570513704450781,
      "grad_norm": 0.12822465598583221,
      "learning_rate": 1.421200516863892e-06,
      "loss": 0.0004,
      "step": 242770
    },
    {
      "epoch": 1.557115507338864,
      "grad_norm": 0.07081867754459381,
      "learning_rate": 1.4208096724199843e-06,
      "loss": 0.0014,
      "step": 242780
    },
    {
      "epoch": 1.5571796442326502,
      "grad_norm": 0.11112422496080399,
      "learning_rate": 1.4204188728258877e-06,
      "loss": 0.0017,
      "step": 242790
    },
    {
      "epoch": 1.5572437811264361,
      "grad_norm": 0.025584004819393158,
      "learning_rate": 1.4200281180865023e-06,
      "loss": 0.0008,
      "step": 242800
    },
    {
      "epoch": 1.5573079180202223,
      "grad_norm": 0.1054731011390686,
      "learning_rate": 1.4196374082067231e-06,
      "loss": 0.0016,
      "step": 242810
    },
    {
      "epoch": 1.5573720549140084,
      "grad_norm": 0.08611925691366196,
      "learning_rate": 1.4192467431914446e-06,
      "loss": 0.0007,
      "step": 242820
    },
    {
      "epoch": 1.5574361918077946,
      "grad_norm": 0.043091047555208206,
      "learning_rate": 1.4188561230455632e-06,
      "loss": 0.0015,
      "step": 242830
    },
    {
      "epoch": 1.5575003287015807,
      "grad_norm": 0.03824600949883461,
      "learning_rate": 1.4184655477739763e-06,
      "loss": 0.0008,
      "step": 242840
    },
    {
      "epoch": 1.5575644655953669,
      "grad_norm": 0.03904568403959274,
      "learning_rate": 1.4180750173815756e-06,
      "loss": 0.0018,
      "step": 242850
    },
    {
      "epoch": 1.5576286024891528,
      "grad_norm": 0.10202863812446594,
      "learning_rate": 1.417684531873254e-06,
      "loss": 0.0014,
      "step": 242860
    },
    {
      "epoch": 1.557692739382939,
      "grad_norm": 0.035224538296461105,
      "learning_rate": 1.4172940912539045e-06,
      "loss": 0.0025,
      "step": 242870
    },
    {
      "epoch": 1.5577568762767249,
      "grad_norm": 0.12914836406707764,
      "learning_rate": 1.4169036955284227e-06,
      "loss": 0.0009,
      "step": 242880
    },
    {
      "epoch": 1.557821013170511,
      "grad_norm": 0.030876312404870987,
      "learning_rate": 1.4165133447016976e-06,
      "loss": 0.0008,
      "step": 242890
    },
    {
      "epoch": 1.5578851500642972,
      "grad_norm": 0.1389792114496231,
      "learning_rate": 1.4161230387786217e-06,
      "loss": 0.0019,
      "step": 242900
    },
    {
      "epoch": 1.5579492869580833,
      "grad_norm": 0.009573440998792648,
      "learning_rate": 1.415732777764084e-06,
      "loss": 0.001,
      "step": 242910
    },
    {
      "epoch": 1.5580134238518695,
      "grad_norm": 0.05179290100932121,
      "learning_rate": 1.4153425616629773e-06,
      "loss": 0.001,
      "step": 242920
    },
    {
      "epoch": 1.5580775607456556,
      "grad_norm": 0.04464186728000641,
      "learning_rate": 1.414952390480191e-06,
      "loss": 0.0012,
      "step": 242930
    },
    {
      "epoch": 1.5581416976394418,
      "grad_norm": 0.07343059033155441,
      "learning_rate": 1.4145622642206113e-06,
      "loss": 0.0021,
      "step": 242940
    },
    {
      "epoch": 1.5582058345332277,
      "grad_norm": 0.0018839394906535745,
      "learning_rate": 1.4141721828891303e-06,
      "loss": 0.0006,
      "step": 242950
    },
    {
      "epoch": 1.5582699714270138,
      "grad_norm": 0.03524048253893852,
      "learning_rate": 1.4137821464906349e-06,
      "loss": 0.0017,
      "step": 242960
    },
    {
      "epoch": 1.5583341083207998,
      "grad_norm": 0.07066163420677185,
      "learning_rate": 1.4133921550300122e-06,
      "loss": 0.0008,
      "step": 242970
    },
    {
      "epoch": 1.558398245214586,
      "grad_norm": 0.04661337658762932,
      "learning_rate": 1.4130022085121475e-06,
      "loss": 0.0008,
      "step": 242980
    },
    {
      "epoch": 1.558462382108372,
      "grad_norm": 0.024441925808787346,
      "learning_rate": 1.4126123069419307e-06,
      "loss": 0.0012,
      "step": 242990
    },
    {
      "epoch": 1.5585265190021582,
      "grad_norm": 0.11181625723838806,
      "learning_rate": 1.412222450324245e-06,
      "loss": 0.0017,
      "step": 243000
    },
    {
      "epoch": 1.5585906558959444,
      "grad_norm": 0.0369691401720047,
      "learning_rate": 1.4118326386639764e-06,
      "loss": 0.0031,
      "step": 243010
    },
    {
      "epoch": 1.5586547927897305,
      "grad_norm": 0.002632532501593232,
      "learning_rate": 1.4114428719660078e-06,
      "loss": 0.0006,
      "step": 243020
    },
    {
      "epoch": 1.5587189296835167,
      "grad_norm": 0.13097435235977173,
      "learning_rate": 1.411053150235226e-06,
      "loss": 0.0015,
      "step": 243030
    },
    {
      "epoch": 1.5587830665773026,
      "grad_norm": 0.11004441231489182,
      "learning_rate": 1.4106634734765135e-06,
      "loss": 0.0017,
      "step": 243040
    },
    {
      "epoch": 1.5588472034710887,
      "grad_norm": 0.1229131743311882,
      "learning_rate": 1.4102738416947525e-06,
      "loss": 0.0019,
      "step": 243050
    },
    {
      "epoch": 1.5589113403648747,
      "grad_norm": 0.03745304048061371,
      "learning_rate": 1.409884254894825e-06,
      "loss": 0.0019,
      "step": 243060
    },
    {
      "epoch": 1.5589754772586608,
      "grad_norm": 0.018195126205682755,
      "learning_rate": 1.4094947130816144e-06,
      "loss": 0.0014,
      "step": 243070
    },
    {
      "epoch": 1.559039614152447,
      "grad_norm": 0.12419760972261429,
      "learning_rate": 1.4091052162600017e-06,
      "loss": 0.0011,
      "step": 243080
    },
    {
      "epoch": 1.559103751046233,
      "grad_norm": 0.32271960377693176,
      "learning_rate": 1.4087157644348648e-06,
      "loss": 0.0016,
      "step": 243090
    },
    {
      "epoch": 1.5591678879400193,
      "grad_norm": 0.03508146479725838,
      "learning_rate": 1.4083263576110885e-06,
      "loss": 0.0009,
      "step": 243100
    },
    {
      "epoch": 1.5592320248338054,
      "grad_norm": 0.08492378890514374,
      "learning_rate": 1.4079369957935491e-06,
      "loss": 0.001,
      "step": 243110
    },
    {
      "epoch": 1.5592961617275913,
      "grad_norm": 0.3226666748523712,
      "learning_rate": 1.4075476789871267e-06,
      "loss": 0.0009,
      "step": 243120
    },
    {
      "epoch": 1.5593602986213775,
      "grad_norm": 0.03416430577635765,
      "learning_rate": 1.4071584071966976e-06,
      "loss": 0.001,
      "step": 243130
    },
    {
      "epoch": 1.5594244355151634,
      "grad_norm": 0.062238939106464386,
      "learning_rate": 1.4067691804271433e-06,
      "loss": 0.0007,
      "step": 243140
    },
    {
      "epoch": 1.5594885724089496,
      "grad_norm": 0.10507658123970032,
      "learning_rate": 1.4063799986833388e-06,
      "loss": 0.0014,
      "step": 243150
    },
    {
      "epoch": 1.5595527093027357,
      "grad_norm": 0.10475585609674454,
      "learning_rate": 1.4059908619701612e-06,
      "loss": 0.0009,
      "step": 243160
    },
    {
      "epoch": 1.5596168461965219,
      "grad_norm": 0.006586553994566202,
      "learning_rate": 1.4056017702924858e-06,
      "loss": 0.0018,
      "step": 243170
    },
    {
      "epoch": 1.559680983090308,
      "grad_norm": 0.1432701051235199,
      "learning_rate": 1.40521272365519e-06,
      "loss": 0.0011,
      "step": 243180
    },
    {
      "epoch": 1.5597451199840942,
      "grad_norm": 0.016658388078212738,
      "learning_rate": 1.4048237220631484e-06,
      "loss": 0.0004,
      "step": 243190
    },
    {
      "epoch": 1.5598092568778803,
      "grad_norm": 0.09033188968896866,
      "learning_rate": 1.4044347655212343e-06,
      "loss": 0.001,
      "step": 243200
    },
    {
      "epoch": 1.5598733937716662,
      "grad_norm": 0.09436356276273727,
      "learning_rate": 1.4040458540343215e-06,
      "loss": 0.0014,
      "step": 243210
    },
    {
      "epoch": 1.5599375306654524,
      "grad_norm": 0.12356075644493103,
      "learning_rate": 1.4036569876072853e-06,
      "loss": 0.0009,
      "step": 243220
    },
    {
      "epoch": 1.5600016675592383,
      "grad_norm": 0.07750432938337326,
      "learning_rate": 1.4032681662449976e-06,
      "loss": 0.001,
      "step": 243230
    },
    {
      "epoch": 1.5600658044530245,
      "grad_norm": 0.0030151098035275936,
      "learning_rate": 1.4028793899523285e-06,
      "loss": 0.0008,
      "step": 243240
    },
    {
      "epoch": 1.5601299413468106,
      "grad_norm": 0.032839518040418625,
      "learning_rate": 1.402490658734153e-06,
      "loss": 0.0009,
      "step": 243250
    },
    {
      "epoch": 1.5601940782405967,
      "grad_norm": 0.06680786609649658,
      "learning_rate": 1.4021019725953405e-06,
      "loss": 0.0007,
      "step": 243260
    },
    {
      "epoch": 1.560258215134383,
      "grad_norm": 0.05233680456876755,
      "learning_rate": 1.4017133315407622e-06,
      "loss": 0.0008,
      "step": 243270
    },
    {
      "epoch": 1.560322352028169,
      "grad_norm": 0.04931798204779625,
      "learning_rate": 1.4013247355752858e-06,
      "loss": 0.0009,
      "step": 243280
    },
    {
      "epoch": 1.560386488921955,
      "grad_norm": 0.0011380620999261737,
      "learning_rate": 1.4009361847037833e-06,
      "loss": 0.0014,
      "step": 243290
    },
    {
      "epoch": 1.5604506258157411,
      "grad_norm": 0.0662262886762619,
      "learning_rate": 1.4005476789311233e-06,
      "loss": 0.0022,
      "step": 243300
    },
    {
      "epoch": 1.5605147627095273,
      "grad_norm": 0.15421129763126373,
      "learning_rate": 1.4001592182621732e-06,
      "loss": 0.0014,
      "step": 243310
    },
    {
      "epoch": 1.5605788996033132,
      "grad_norm": 0.12788690626621246,
      "learning_rate": 1.3997708027017993e-06,
      "loss": 0.0009,
      "step": 243320
    },
    {
      "epoch": 1.5606430364970993,
      "grad_norm": 0.1507088989019394,
      "learning_rate": 1.3993824322548721e-06,
      "loss": 0.0013,
      "step": 243330
    },
    {
      "epoch": 1.5607071733908855,
      "grad_norm": 0.09280886501073837,
      "learning_rate": 1.3989941069262558e-06,
      "loss": 0.001,
      "step": 243340
    },
    {
      "epoch": 1.5607713102846716,
      "grad_norm": 0.06773676723241806,
      "learning_rate": 1.3986058267208174e-06,
      "loss": 0.0009,
      "step": 243350
    },
    {
      "epoch": 1.5608354471784578,
      "grad_norm": 0.1687345951795578,
      "learning_rate": 1.3982175916434204e-06,
      "loss": 0.0012,
      "step": 243360
    },
    {
      "epoch": 1.560899584072244,
      "grad_norm": 0.10132893174886703,
      "learning_rate": 1.3978294016989324e-06,
      "loss": 0.0007,
      "step": 243370
    },
    {
      "epoch": 1.5609637209660299,
      "grad_norm": 0.030050620436668396,
      "learning_rate": 1.3974412568922163e-06,
      "loss": 0.0012,
      "step": 243380
    },
    {
      "epoch": 1.561027857859816,
      "grad_norm": 0.07169070839881897,
      "learning_rate": 1.397053157228135e-06,
      "loss": 0.0005,
      "step": 243390
    },
    {
      "epoch": 1.561091994753602,
      "grad_norm": 0.1907750368118286,
      "learning_rate": 1.3966651027115536e-06,
      "loss": 0.003,
      "step": 243400
    },
    {
      "epoch": 1.561156131647388,
      "grad_norm": 0.03356742486357689,
      "learning_rate": 1.3962770933473336e-06,
      "loss": 0.0012,
      "step": 243410
    },
    {
      "epoch": 1.5612202685411742,
      "grad_norm": 0.05747605115175247,
      "learning_rate": 1.3958891291403376e-06,
      "loss": 0.0016,
      "step": 243420
    },
    {
      "epoch": 1.5612844054349604,
      "grad_norm": 0.05902548506855965,
      "learning_rate": 1.3955012100954246e-06,
      "loss": 0.0016,
      "step": 243430
    },
    {
      "epoch": 1.5613485423287465,
      "grad_norm": 0.09562114626169205,
      "learning_rate": 1.3951133362174595e-06,
      "loss": 0.0006,
      "step": 243440
    },
    {
      "epoch": 1.5614126792225327,
      "grad_norm": 0.04821772128343582,
      "learning_rate": 1.394725507511301e-06,
      "loss": 0.0011,
      "step": 243450
    },
    {
      "epoch": 1.5614768161163188,
      "grad_norm": 0.04168270155787468,
      "learning_rate": 1.3943377239818078e-06,
      "loss": 0.0006,
      "step": 243460
    },
    {
      "epoch": 1.5615409530101048,
      "grad_norm": 0.127223938703537,
      "learning_rate": 1.3939499856338384e-06,
      "loss": 0.0013,
      "step": 243470
    },
    {
      "epoch": 1.561605089903891,
      "grad_norm": 0.0999385342001915,
      "learning_rate": 1.3935622924722546e-06,
      "loss": 0.0009,
      "step": 243480
    },
    {
      "epoch": 1.5616692267976768,
      "grad_norm": 0.15948669612407684,
      "learning_rate": 1.3931746445019134e-06,
      "loss": 0.0019,
      "step": 243490
    },
    {
      "epoch": 1.561733363691463,
      "grad_norm": 0.09697796404361725,
      "learning_rate": 1.3927870417276707e-06,
      "loss": 0.0012,
      "step": 243500
    },
    {
      "epoch": 1.5617975005852491,
      "grad_norm": 0.07811637222766876,
      "learning_rate": 1.3923994841543836e-06,
      "loss": 0.0007,
      "step": 243510
    },
    {
      "epoch": 1.5618616374790353,
      "grad_norm": 0.0629824548959732,
      "learning_rate": 1.3920119717869102e-06,
      "loss": 0.0015,
      "step": 243520
    },
    {
      "epoch": 1.5619257743728214,
      "grad_norm": 0.1268361210823059,
      "learning_rate": 1.3916245046301058e-06,
      "loss": 0.0009,
      "step": 243530
    },
    {
      "epoch": 1.5619899112666076,
      "grad_norm": 0.023304827511310577,
      "learning_rate": 1.3912370826888232e-06,
      "loss": 0.001,
      "step": 243540
    },
    {
      "epoch": 1.5620540481603935,
      "grad_norm": 0.04373375326395035,
      "learning_rate": 1.390849705967921e-06,
      "loss": 0.0011,
      "step": 243550
    },
    {
      "epoch": 1.5621181850541797,
      "grad_norm": 0.04374853894114494,
      "learning_rate": 1.3904623744722517e-06,
      "loss": 0.0028,
      "step": 243560
    },
    {
      "epoch": 1.5621823219479656,
      "grad_norm": 0.0007684393785893917,
      "learning_rate": 1.3900750882066683e-06,
      "loss": 0.0012,
      "step": 243570
    },
    {
      "epoch": 1.5622464588417517,
      "grad_norm": 0.04392901062965393,
      "learning_rate": 1.3896878471760222e-06,
      "loss": 0.0011,
      "step": 243580
    },
    {
      "epoch": 1.5623105957355379,
      "grad_norm": 0.07232941687107086,
      "learning_rate": 1.3893006513851676e-06,
      "loss": 0.0012,
      "step": 243590
    },
    {
      "epoch": 1.562374732629324,
      "grad_norm": 0.026323221623897552,
      "learning_rate": 1.3889135008389582e-06,
      "loss": 0.0015,
      "step": 243600
    },
    {
      "epoch": 1.5624388695231102,
      "grad_norm": 0.06476714462041855,
      "learning_rate": 1.3885263955422434e-06,
      "loss": 0.0008,
      "step": 243610
    },
    {
      "epoch": 1.5625030064168963,
      "grad_norm": 0.1663721352815628,
      "learning_rate": 1.388139335499874e-06,
      "loss": 0.0016,
      "step": 243620
    },
    {
      "epoch": 1.5625671433106825,
      "grad_norm": 0.06588054448366165,
      "learning_rate": 1.3877523207166982e-06,
      "loss": 0.0015,
      "step": 243630
    },
    {
      "epoch": 1.5626312802044684,
      "grad_norm": 0.0852726399898529,
      "learning_rate": 1.3873653511975694e-06,
      "loss": 0.0013,
      "step": 243640
    },
    {
      "epoch": 1.5626954170982545,
      "grad_norm": 0.0007486481335945427,
      "learning_rate": 1.3869784269473347e-06,
      "loss": 0.0012,
      "step": 243650
    },
    {
      "epoch": 1.5627595539920405,
      "grad_norm": 0.00634699035435915,
      "learning_rate": 1.386591547970841e-06,
      "loss": 0.0026,
      "step": 243660
    },
    {
      "epoch": 1.5628236908858266,
      "grad_norm": 0.033408116549253464,
      "learning_rate": 1.3862047142729385e-06,
      "loss": 0.0012,
      "step": 243670
    },
    {
      "epoch": 1.5628878277796128,
      "grad_norm": 0.09897807240486145,
      "learning_rate": 1.3858179258584743e-06,
      "loss": 0.0007,
      "step": 243680
    },
    {
      "epoch": 1.562951964673399,
      "grad_norm": 0.04074937850236893,
      "learning_rate": 1.3854311827322942e-06,
      "loss": 0.002,
      "step": 243690
    },
    {
      "epoch": 1.563016101567185,
      "grad_norm": 0.04539360851049423,
      "learning_rate": 1.3850444848992433e-06,
      "loss": 0.0007,
      "step": 243700
    },
    {
      "epoch": 1.5630802384609712,
      "grad_norm": 0.013356728479266167,
      "learning_rate": 1.3846578323641702e-06,
      "loss": 0.0006,
      "step": 243710
    },
    {
      "epoch": 1.5631443753547571,
      "grad_norm": 0.15393126010894775,
      "learning_rate": 1.3842712251319185e-06,
      "loss": 0.0012,
      "step": 243720
    },
    {
      "epoch": 1.5632085122485433,
      "grad_norm": 0.056364256888628006,
      "learning_rate": 1.3838846632073316e-06,
      "loss": 0.0014,
      "step": 243730
    },
    {
      "epoch": 1.5632726491423294,
      "grad_norm": 0.04982461407780647,
      "learning_rate": 1.3834981465952535e-06,
      "loss": 0.0018,
      "step": 243740
    },
    {
      "epoch": 1.5633367860361154,
      "grad_norm": 0.07150271534919739,
      "learning_rate": 1.3831116753005297e-06,
      "loss": 0.0015,
      "step": 243750
    },
    {
      "epoch": 1.5634009229299015,
      "grad_norm": 0.047069478780031204,
      "learning_rate": 1.3827252493280014e-06,
      "loss": 0.0011,
      "step": 243760
    },
    {
      "epoch": 1.5634650598236877,
      "grad_norm": 0.1507633626461029,
      "learning_rate": 1.3823388686825112e-06,
      "loss": 0.0009,
      "step": 243770
    },
    {
      "epoch": 1.5635291967174738,
      "grad_norm": 0.12099693715572357,
      "learning_rate": 1.3819525333688989e-06,
      "loss": 0.0011,
      "step": 243780
    },
    {
      "epoch": 1.56359333361126,
      "grad_norm": 0.020730815827846527,
      "learning_rate": 1.3815662433920084e-06,
      "loss": 0.0012,
      "step": 243790
    },
    {
      "epoch": 1.563657470505046,
      "grad_norm": 0.007649121806025505,
      "learning_rate": 1.3811799987566794e-06,
      "loss": 0.0022,
      "step": 243800
    },
    {
      "epoch": 1.563721607398832,
      "grad_norm": 0.014868194237351418,
      "learning_rate": 1.3807937994677494e-06,
      "loss": 0.0009,
      "step": 243810
    },
    {
      "epoch": 1.5637857442926182,
      "grad_norm": 0.15606722235679626,
      "learning_rate": 1.3804076455300614e-06,
      "loss": 0.0013,
      "step": 243820
    },
    {
      "epoch": 1.5638498811864041,
      "grad_norm": 0.016681639477610588,
      "learning_rate": 1.3800215369484521e-06,
      "loss": 0.0011,
      "step": 243830
    },
    {
      "epoch": 1.5639140180801903,
      "grad_norm": 0.0037738841492682695,
      "learning_rate": 1.3796354737277607e-06,
      "loss": 0.0006,
      "step": 243840
    },
    {
      "epoch": 1.5639781549739764,
      "grad_norm": 0.03227373585104942,
      "learning_rate": 1.3792494558728226e-06,
      "loss": 0.0021,
      "step": 243850
    },
    {
      "epoch": 1.5640422918677626,
      "grad_norm": 0.07239340245723724,
      "learning_rate": 1.378863483388478e-06,
      "loss": 0.0012,
      "step": 243860
    },
    {
      "epoch": 1.5641064287615487,
      "grad_norm": 0.06314176321029663,
      "learning_rate": 1.3784775562795617e-06,
      "loss": 0.0011,
      "step": 243870
    },
    {
      "epoch": 1.5641705656553349,
      "grad_norm": 0.06834463030099869,
      "learning_rate": 1.3780916745509099e-06,
      "loss": 0.0011,
      "step": 243880
    },
    {
      "epoch": 1.564234702549121,
      "grad_norm": 0.039398193359375,
      "learning_rate": 1.3777058382073566e-06,
      "loss": 0.0004,
      "step": 243890
    },
    {
      "epoch": 1.564298839442907,
      "grad_norm": 0.1559874713420868,
      "learning_rate": 1.3773200472537396e-06,
      "loss": 0.0013,
      "step": 243900
    },
    {
      "epoch": 1.564362976336693,
      "grad_norm": 0.028561554849147797,
      "learning_rate": 1.3769343016948911e-06,
      "loss": 0.0008,
      "step": 243910
    },
    {
      "epoch": 1.564427113230479,
      "grad_norm": 0.042328860610723495,
      "learning_rate": 1.3765486015356455e-06,
      "loss": 0.0008,
      "step": 243920
    },
    {
      "epoch": 1.5644912501242652,
      "grad_norm": 0.08399327844381332,
      "learning_rate": 1.376162946780834e-06,
      "loss": 0.0008,
      "step": 243930
    },
    {
      "epoch": 1.5645553870180513,
      "grad_norm": 0.08977387845516205,
      "learning_rate": 1.3757773374352918e-06,
      "loss": 0.0006,
      "step": 243940
    },
    {
      "epoch": 1.5646195239118375,
      "grad_norm": 0.005761916283518076,
      "learning_rate": 1.3753917735038503e-06,
      "loss": 0.0015,
      "step": 243950
    },
    {
      "epoch": 1.5646836608056236,
      "grad_norm": 0.00901766400784254,
      "learning_rate": 1.3750062549913383e-06,
      "loss": 0.0018,
      "step": 243960
    },
    {
      "epoch": 1.5647477976994097,
      "grad_norm": 0.05700463801622391,
      "learning_rate": 1.374620781902591e-06,
      "loss": 0.0016,
      "step": 243970
    },
    {
      "epoch": 1.5648119345931957,
      "grad_norm": 0.11297713965177536,
      "learning_rate": 1.374235354242436e-06,
      "loss": 0.0015,
      "step": 243980
    },
    {
      "epoch": 1.5648760714869818,
      "grad_norm": 0.03825245797634125,
      "learning_rate": 1.3738499720157028e-06,
      "loss": 0.0005,
      "step": 243990
    },
    {
      "epoch": 1.5649402083807677,
      "grad_norm": 0.11012471467256546,
      "learning_rate": 1.3734646352272202e-06,
      "loss": 0.0018,
      "step": 244000
    },
    {
      "epoch": 1.5649402083807677,
      "eval_loss": 0.0021118263248354197,
      "eval_runtime": 3.3204,
      "eval_samples_per_second": 60.233,
      "eval_steps_per_second": 15.058,
      "step": 244000
    },
    {
      "epoch": 1.565004345274554,
      "grad_norm": 0.051630180329084396,
      "learning_rate": 1.3730793438818184e-06,
      "loss": 0.0011,
      "step": 244010
    },
    {
      "epoch": 1.56506848216834,
      "grad_norm": 0.1360810250043869,
      "learning_rate": 1.372694097984325e-06,
      "loss": 0.0008,
      "step": 244020
    },
    {
      "epoch": 1.5651326190621262,
      "grad_norm": 0.027779752388596535,
      "learning_rate": 1.3723088975395671e-06,
      "loss": 0.0006,
      "step": 244030
    },
    {
      "epoch": 1.5651967559559123,
      "grad_norm": 0.01636786386370659,
      "learning_rate": 1.3719237425523695e-06,
      "loss": 0.0009,
      "step": 244040
    },
    {
      "epoch": 1.5652608928496985,
      "grad_norm": 0.017599863931536674,
      "learning_rate": 1.371538633027562e-06,
      "loss": 0.0008,
      "step": 244050
    },
    {
      "epoch": 1.5653250297434846,
      "grad_norm": 0.08058811724185944,
      "learning_rate": 1.3711535689699684e-06,
      "loss": 0.0011,
      "step": 244060
    },
    {
      "epoch": 1.5653891666372706,
      "grad_norm": 0.17835091054439545,
      "learning_rate": 1.3707685503844142e-06,
      "loss": 0.0023,
      "step": 244070
    },
    {
      "epoch": 1.5654533035310567,
      "grad_norm": 0.05642031878232956,
      "learning_rate": 1.3703835772757229e-06,
      "loss": 0.0014,
      "step": 244080
    },
    {
      "epoch": 1.5655174404248426,
      "grad_norm": 0.03311365097761154,
      "learning_rate": 1.3699986496487206e-06,
      "loss": 0.0011,
      "step": 244090
    },
    {
      "epoch": 1.5655815773186288,
      "grad_norm": 0.004921520594507456,
      "learning_rate": 1.369613767508229e-06,
      "loss": 0.0013,
      "step": 244100
    },
    {
      "epoch": 1.565645714212415,
      "grad_norm": 0.024157697334885597,
      "learning_rate": 1.3692289308590706e-06,
      "loss": 0.0014,
      "step": 244110
    },
    {
      "epoch": 1.565709851106201,
      "grad_norm": 0.003105662763118744,
      "learning_rate": 1.3688441397060698e-06,
      "loss": 0.0012,
      "step": 244120
    },
    {
      "epoch": 1.5657739879999872,
      "grad_norm": 0.3345547020435333,
      "learning_rate": 1.3684593940540468e-06,
      "loss": 0.0008,
      "step": 244130
    },
    {
      "epoch": 1.5658381248937734,
      "grad_norm": 0.010019885376095772,
      "learning_rate": 1.3680746939078237e-06,
      "loss": 0.0017,
      "step": 244140
    },
    {
      "epoch": 1.5659022617875593,
      "grad_norm": 0.05200977995991707,
      "learning_rate": 1.3676900392722186e-06,
      "loss": 0.0014,
      "step": 244150
    },
    {
      "epoch": 1.5659663986813455,
      "grad_norm": 0.014318738132715225,
      "learning_rate": 1.3673054301520545e-06,
      "loss": 0.0008,
      "step": 244160
    },
    {
      "epoch": 1.5660305355751316,
      "grad_norm": 0.10073602199554443,
      "learning_rate": 1.36692086655215e-06,
      "loss": 0.0006,
      "step": 244170
    },
    {
      "epoch": 1.5660946724689175,
      "grad_norm": 0.04899398609995842,
      "learning_rate": 1.3665363484773237e-06,
      "loss": 0.0009,
      "step": 244180
    },
    {
      "epoch": 1.5661588093627037,
      "grad_norm": 0.03722739592194557,
      "learning_rate": 1.3661518759323916e-06,
      "loss": 0.0009,
      "step": 244190
    },
    {
      "epoch": 1.5662229462564898,
      "grad_norm": 0.14708061516284943,
      "learning_rate": 1.3657674489221756e-06,
      "loss": 0.001,
      "step": 244200
    },
    {
      "epoch": 1.566287083150276,
      "grad_norm": 0.16129310429096222,
      "learning_rate": 1.3653830674514906e-06,
      "loss": 0.0011,
      "step": 244210
    },
    {
      "epoch": 1.5663512200440621,
      "grad_norm": 0.04316999763250351,
      "learning_rate": 1.3649987315251534e-06,
      "loss": 0.0007,
      "step": 244220
    },
    {
      "epoch": 1.5664153569378483,
      "grad_norm": 0.6100648641586304,
      "learning_rate": 1.3646144411479784e-06,
      "loss": 0.0018,
      "step": 244230
    },
    {
      "epoch": 1.5664794938316342,
      "grad_norm": 0.07713521271944046,
      "learning_rate": 1.3642301963247845e-06,
      "loss": 0.0007,
      "step": 244240
    },
    {
      "epoch": 1.5665436307254204,
      "grad_norm": 0.014485377818346024,
      "learning_rate": 1.363845997060384e-06,
      "loss": 0.0008,
      "step": 244250
    },
    {
      "epoch": 1.5666077676192063,
      "grad_norm": 0.05376194417476654,
      "learning_rate": 1.3634618433595908e-06,
      "loss": 0.0047,
      "step": 244260
    },
    {
      "epoch": 1.5666719045129924,
      "grad_norm": 0.018168184906244278,
      "learning_rate": 1.3630777352272212e-06,
      "loss": 0.0008,
      "step": 244270
    },
    {
      "epoch": 1.5667360414067786,
      "grad_norm": 0.05175149813294411,
      "learning_rate": 1.3626936726680867e-06,
      "loss": 0.0009,
      "step": 244280
    },
    {
      "epoch": 1.5668001783005647,
      "grad_norm": 0.0020207969937473536,
      "learning_rate": 1.3623096556870003e-06,
      "loss": 0.0009,
      "step": 244290
    },
    {
      "epoch": 1.5668643151943509,
      "grad_norm": 0.11341476440429688,
      "learning_rate": 1.3619256842887724e-06,
      "loss": 0.0005,
      "step": 244300
    },
    {
      "epoch": 1.566928452088137,
      "grad_norm": 0.0669422447681427,
      "learning_rate": 1.361541758478217e-06,
      "loss": 0.0006,
      "step": 244310
    },
    {
      "epoch": 1.5669925889819232,
      "grad_norm": 0.12491405010223389,
      "learning_rate": 1.3611578782601436e-06,
      "loss": 0.001,
      "step": 244320
    },
    {
      "epoch": 1.567056725875709,
      "grad_norm": 0.21512703597545624,
      "learning_rate": 1.3607740436393624e-06,
      "loss": 0.0007,
      "step": 244330
    },
    {
      "epoch": 1.5671208627694952,
      "grad_norm": 0.0756843164563179,
      "learning_rate": 1.3603902546206826e-06,
      "loss": 0.0004,
      "step": 244340
    },
    {
      "epoch": 1.5671849996632812,
      "grad_norm": 0.01778343692421913,
      "learning_rate": 1.360006511208915e-06,
      "loss": 0.0012,
      "step": 244350
    },
    {
      "epoch": 1.5672491365570673,
      "grad_norm": 0.03670975938439369,
      "learning_rate": 1.3596228134088657e-06,
      "loss": 0.0011,
      "step": 244360
    },
    {
      "epoch": 1.5673132734508535,
      "grad_norm": 0.13029594719409943,
      "learning_rate": 1.3592391612253465e-06,
      "loss": 0.0018,
      "step": 244370
    },
    {
      "epoch": 1.5673774103446396,
      "grad_norm": 0.04360777512192726,
      "learning_rate": 1.3588555546631605e-06,
      "loss": 0.0015,
      "step": 244380
    },
    {
      "epoch": 1.5674415472384258,
      "grad_norm": 0.03885827213525772,
      "learning_rate": 1.3584719937271184e-06,
      "loss": 0.0006,
      "step": 244390
    },
    {
      "epoch": 1.567505684132212,
      "grad_norm": 0.07572092860937119,
      "learning_rate": 1.3580884784220243e-06,
      "loss": 0.0025,
      "step": 244400
    },
    {
      "epoch": 1.5675698210259978,
      "grad_norm": 0.013776548206806183,
      "learning_rate": 1.357705008752685e-06,
      "loss": 0.0005,
      "step": 244410
    },
    {
      "epoch": 1.567633957919784,
      "grad_norm": 0.09575240314006805,
      "learning_rate": 1.3573215847239034e-06,
      "loss": 0.0011,
      "step": 244420
    },
    {
      "epoch": 1.56769809481357,
      "grad_norm": 0.05258919671177864,
      "learning_rate": 1.3569382063404873e-06,
      "loss": 0.0008,
      "step": 244430
    },
    {
      "epoch": 1.567762231707356,
      "grad_norm": 0.04056648537516594,
      "learning_rate": 1.3565548736072394e-06,
      "loss": 0.0007,
      "step": 244440
    },
    {
      "epoch": 1.5678263686011422,
      "grad_norm": 0.017687644809484482,
      "learning_rate": 1.356171586528962e-06,
      "loss": 0.001,
      "step": 244450
    },
    {
      "epoch": 1.5678905054949284,
      "grad_norm": 0.08572002500295639,
      "learning_rate": 1.3557883451104581e-06,
      "loss": 0.001,
      "step": 244460
    },
    {
      "epoch": 1.5679546423887145,
      "grad_norm": 0.028661692515015602,
      "learning_rate": 1.3554051493565317e-06,
      "loss": 0.0024,
      "step": 244470
    },
    {
      "epoch": 1.5680187792825007,
      "grad_norm": 0.10158579796552658,
      "learning_rate": 1.3550219992719838e-06,
      "loss": 0.0015,
      "step": 244480
    },
    {
      "epoch": 1.5680829161762868,
      "grad_norm": 0.012906504794955254,
      "learning_rate": 1.3546388948616152e-06,
      "loss": 0.001,
      "step": 244490
    },
    {
      "epoch": 1.5681470530700727,
      "grad_norm": 0.07433255761861801,
      "learning_rate": 1.354255836130225e-06,
      "loss": 0.002,
      "step": 244500
    },
    {
      "epoch": 1.5682111899638589,
      "grad_norm": 0.06329713761806488,
      "learning_rate": 1.3538728230826165e-06,
      "loss": 0.0006,
      "step": 244510
    },
    {
      "epoch": 1.5682753268576448,
      "grad_norm": 0.16259931027889252,
      "learning_rate": 1.353489855723587e-06,
      "loss": 0.0009,
      "step": 244520
    },
    {
      "epoch": 1.568339463751431,
      "grad_norm": 0.017799677327275276,
      "learning_rate": 1.3531069340579345e-06,
      "loss": 0.0021,
      "step": 244530
    },
    {
      "epoch": 1.568403600645217,
      "grad_norm": 0.16020628809928894,
      "learning_rate": 1.35272405809046e-06,
      "loss": 0.0012,
      "step": 244540
    },
    {
      "epoch": 1.5684677375390033,
      "grad_norm": 0.0366208590567112,
      "learning_rate": 1.3523412278259595e-06,
      "loss": 0.0013,
      "step": 244550
    },
    {
      "epoch": 1.5685318744327894,
      "grad_norm": 0.11714744567871094,
      "learning_rate": 1.3519584432692305e-06,
      "loss": 0.0016,
      "step": 244560
    },
    {
      "epoch": 1.5685960113265756,
      "grad_norm": 0.026635903865098953,
      "learning_rate": 1.3515757044250683e-06,
      "loss": 0.003,
      "step": 244570
    },
    {
      "epoch": 1.5686601482203617,
      "grad_norm": 0.012604729272425175,
      "learning_rate": 1.3511930112982713e-06,
      "loss": 0.0006,
      "step": 244580
    },
    {
      "epoch": 1.5687242851141476,
      "grad_norm": 0.12273949384689331,
      "learning_rate": 1.3508103638936332e-06,
      "loss": 0.0012,
      "step": 244590
    },
    {
      "epoch": 1.5687884220079338,
      "grad_norm": 0.1338241696357727,
      "learning_rate": 1.3504277622159495e-06,
      "loss": 0.002,
      "step": 244600
    },
    {
      "epoch": 1.5688525589017197,
      "grad_norm": 0.02673218958079815,
      "learning_rate": 1.3500452062700132e-06,
      "loss": 0.001,
      "step": 244610
    },
    {
      "epoch": 1.5689166957955059,
      "grad_norm": 0.09174564480781555,
      "learning_rate": 1.3496626960606202e-06,
      "loss": 0.0014,
      "step": 244620
    },
    {
      "epoch": 1.568980832689292,
      "grad_norm": 0.069851815700531,
      "learning_rate": 1.3492802315925623e-06,
      "loss": 0.0026,
      "step": 244630
    },
    {
      "epoch": 1.5690449695830782,
      "grad_norm": 0.12030567228794098,
      "learning_rate": 1.3488978128706309e-06,
      "loss": 0.0012,
      "step": 244640
    },
    {
      "epoch": 1.5691091064768643,
      "grad_norm": 0.1622411608695984,
      "learning_rate": 1.3485154398996209e-06,
      "loss": 0.0026,
      "step": 244650
    },
    {
      "epoch": 1.5691732433706505,
      "grad_norm": 0.010083461180329323,
      "learning_rate": 1.3481331126843216e-06,
      "loss": 0.0008,
      "step": 244660
    },
    {
      "epoch": 1.5692373802644364,
      "grad_norm": 0.14678412675857544,
      "learning_rate": 1.347750831229525e-06,
      "loss": 0.0014,
      "step": 244670
    },
    {
      "epoch": 1.5693015171582225,
      "grad_norm": 0.07875506579875946,
      "learning_rate": 1.347368595540019e-06,
      "loss": 0.0021,
      "step": 244680
    },
    {
      "epoch": 1.5693656540520085,
      "grad_norm": 0.03652176260948181,
      "learning_rate": 1.3469864056205962e-06,
      "loss": 0.0005,
      "step": 244690
    },
    {
      "epoch": 1.5694297909457946,
      "grad_norm": 0.14352712035179138,
      "learning_rate": 1.3466042614760449e-06,
      "loss": 0.0022,
      "step": 244700
    },
    {
      "epoch": 1.5694939278395807,
      "grad_norm": 0.40054547786712646,
      "learning_rate": 1.3462221631111533e-06,
      "loss": 0.0047,
      "step": 244710
    },
    {
      "epoch": 1.569558064733367,
      "grad_norm": 0.08101187646389008,
      "learning_rate": 1.3458401105307073e-06,
      "loss": 0.0009,
      "step": 244720
    },
    {
      "epoch": 1.569622201627153,
      "grad_norm": 0.1281840205192566,
      "learning_rate": 1.3454581037394981e-06,
      "loss": 0.0007,
      "step": 244730
    },
    {
      "epoch": 1.5696863385209392,
      "grad_norm": 0.05733279883861542,
      "learning_rate": 1.345076142742311e-06,
      "loss": 0.0011,
      "step": 244740
    },
    {
      "epoch": 1.5697504754147253,
      "grad_norm": 0.36465564370155334,
      "learning_rate": 1.3446942275439317e-06,
      "loss": 0.0047,
      "step": 244750
    },
    {
      "epoch": 1.5698146123085113,
      "grad_norm": 0.0006671261508017778,
      "learning_rate": 1.3443123581491446e-06,
      "loss": 0.001,
      "step": 244760
    },
    {
      "epoch": 1.5698787492022974,
      "grad_norm": 0.05007866397500038,
      "learning_rate": 1.343930534562738e-06,
      "loss": 0.0011,
      "step": 244770
    },
    {
      "epoch": 1.5699428860960833,
      "grad_norm": 0.05445602163672447,
      "learning_rate": 1.3435487567894944e-06,
      "loss": 0.0012,
      "step": 244780
    },
    {
      "epoch": 1.5700070229898695,
      "grad_norm": 0.17970049381256104,
      "learning_rate": 1.3431670248341965e-06,
      "loss": 0.0011,
      "step": 244790
    },
    {
      "epoch": 1.5700711598836556,
      "grad_norm": 0.0806487500667572,
      "learning_rate": 1.3427853387016315e-06,
      "loss": 0.001,
      "step": 244800
    },
    {
      "epoch": 1.5701352967774418,
      "grad_norm": 0.11729707568883896,
      "learning_rate": 1.3424036983965793e-06,
      "loss": 0.0014,
      "step": 244810
    },
    {
      "epoch": 1.570199433671228,
      "grad_norm": 0.08794651180505753,
      "learning_rate": 1.3420221039238235e-06,
      "loss": 0.0015,
      "step": 244820
    },
    {
      "epoch": 1.570263570565014,
      "grad_norm": 0.09539645165205002,
      "learning_rate": 1.3416405552881429e-06,
      "loss": 0.002,
      "step": 244830
    },
    {
      "epoch": 1.5703277074588,
      "grad_norm": 0.08823912590742111,
      "learning_rate": 1.3412590524943226e-06,
      "loss": 0.0014,
      "step": 244840
    },
    {
      "epoch": 1.5703918443525862,
      "grad_norm": 0.11009037494659424,
      "learning_rate": 1.340877595547141e-06,
      "loss": 0.0013,
      "step": 244850
    },
    {
      "epoch": 1.5704559812463723,
      "grad_norm": 0.09003601223230362,
      "learning_rate": 1.3404961844513786e-06,
      "loss": 0.0014,
      "step": 244860
    },
    {
      "epoch": 1.5705201181401582,
      "grad_norm": 0.007920468226075172,
      "learning_rate": 1.3401148192118124e-06,
      "loss": 0.0013,
      "step": 244870
    },
    {
      "epoch": 1.5705842550339444,
      "grad_norm": 0.004617893602699041,
      "learning_rate": 1.3397334998332252e-06,
      "loss": 0.0006,
      "step": 244880
    },
    {
      "epoch": 1.5706483919277305,
      "grad_norm": 0.10888317227363586,
      "learning_rate": 1.339352226320393e-06,
      "loss": 0.0012,
      "step": 244890
    },
    {
      "epoch": 1.5707125288215167,
      "grad_norm": 0.0815168246626854,
      "learning_rate": 1.3389709986780935e-06,
      "loss": 0.0012,
      "step": 244900
    },
    {
      "epoch": 1.5707766657153028,
      "grad_norm": 0.09804923832416534,
      "learning_rate": 1.3385898169111028e-06,
      "loss": 0.0017,
      "step": 244910
    },
    {
      "epoch": 1.570840802609089,
      "grad_norm": 0.016148488968610764,
      "learning_rate": 1.3382086810241995e-06,
      "loss": 0.0006,
      "step": 244920
    },
    {
      "epoch": 1.570904939502875,
      "grad_norm": 0.02452525682747364,
      "learning_rate": 1.337827591022159e-06,
      "loss": 0.0024,
      "step": 244930
    },
    {
      "epoch": 1.570969076396661,
      "grad_norm": 0.10429196804761887,
      "learning_rate": 1.3374465469097536e-06,
      "loss": 0.0015,
      "step": 244940
    },
    {
      "epoch": 1.571033213290447,
      "grad_norm": 0.01082930900156498,
      "learning_rate": 1.3370655486917627e-06,
      "loss": 0.0016,
      "step": 244950
    },
    {
      "epoch": 1.5710973501842331,
      "grad_norm": 0.030714496970176697,
      "learning_rate": 1.3366845963729586e-06,
      "loss": 0.0006,
      "step": 244960
    },
    {
      "epoch": 1.5711614870780193,
      "grad_norm": 0.13202424347400665,
      "learning_rate": 1.3363036899581145e-06,
      "loss": 0.0016,
      "step": 244970
    },
    {
      "epoch": 1.5712256239718054,
      "grad_norm": 0.05811617895960808,
      "learning_rate": 1.3359228294520017e-06,
      "loss": 0.0008,
      "step": 244980
    },
    {
      "epoch": 1.5712897608655916,
      "grad_norm": 0.014790991321206093,
      "learning_rate": 1.3355420148593961e-06,
      "loss": 0.0016,
      "step": 244990
    },
    {
      "epoch": 1.5713538977593777,
      "grad_norm": 0.12423159182071686,
      "learning_rate": 1.3351612461850682e-06,
      "loss": 0.0009,
      "step": 245000
    },
    {
      "epoch": 1.5714180346531639,
      "grad_norm": 0.03937933221459389,
      "learning_rate": 1.334780523433789e-06,
      "loss": 0.0006,
      "step": 245010
    },
    {
      "epoch": 1.5714821715469498,
      "grad_norm": 0.13783757388591766,
      "learning_rate": 1.334399846610328e-06,
      "loss": 0.0017,
      "step": 245020
    },
    {
      "epoch": 1.571546308440736,
      "grad_norm": 0.03599138185381889,
      "learning_rate": 1.3340192157194581e-06,
      "loss": 0.0012,
      "step": 245030
    },
    {
      "epoch": 1.5716104453345219,
      "grad_norm": 0.30758318305015564,
      "learning_rate": 1.3336386307659477e-06,
      "loss": 0.0015,
      "step": 245040
    },
    {
      "epoch": 1.571674582228308,
      "grad_norm": 0.01626293547451496,
      "learning_rate": 1.333258091754565e-06,
      "loss": 0.0016,
      "step": 245050
    },
    {
      "epoch": 1.5717387191220942,
      "grad_norm": 0.13304047286510468,
      "learning_rate": 1.3328775986900782e-06,
      "loss": 0.0007,
      "step": 245060
    },
    {
      "epoch": 1.5718028560158803,
      "grad_norm": 0.034299153834581375,
      "learning_rate": 1.3324971515772566e-06,
      "loss": 0.0009,
      "step": 245070
    },
    {
      "epoch": 1.5718669929096665,
      "grad_norm": 0.05086997151374817,
      "learning_rate": 1.3321167504208677e-06,
      "loss": 0.0004,
      "step": 245080
    },
    {
      "epoch": 1.5719311298034526,
      "grad_norm": 0.10344081372022629,
      "learning_rate": 1.331736395225675e-06,
      "loss": 0.0006,
      "step": 245090
    },
    {
      "epoch": 1.5719952666972385,
      "grad_norm": 0.21350620687007904,
      "learning_rate": 1.3313560859964493e-06,
      "loss": 0.0017,
      "step": 245100
    },
    {
      "epoch": 1.5720594035910247,
      "grad_norm": 0.18175390362739563,
      "learning_rate": 1.3309758227379533e-06,
      "loss": 0.0012,
      "step": 245110
    },
    {
      "epoch": 1.5721235404848106,
      "grad_norm": 0.06544242054224014,
      "learning_rate": 1.3305956054549508e-06,
      "loss": 0.0008,
      "step": 245120
    },
    {
      "epoch": 1.5721876773785968,
      "grad_norm": 0.0935523584485054,
      "learning_rate": 1.3302154341522093e-06,
      "loss": 0.0005,
      "step": 245130
    },
    {
      "epoch": 1.572251814272383,
      "grad_norm": 0.11086033284664154,
      "learning_rate": 1.32983530883449e-06,
      "loss": 0.0016,
      "step": 245140
    },
    {
      "epoch": 1.572315951166169,
      "grad_norm": 0.038969676941633224,
      "learning_rate": 1.3294552295065588e-06,
      "loss": 0.0019,
      "step": 245150
    },
    {
      "epoch": 1.5723800880599552,
      "grad_norm": 0.02033829689025879,
      "learning_rate": 1.3290751961731769e-06,
      "loss": 0.0021,
      "step": 245160
    },
    {
      "epoch": 1.5724442249537414,
      "grad_norm": 0.20166823267936707,
      "learning_rate": 1.3286952088391058e-06,
      "loss": 0.0023,
      "step": 245170
    },
    {
      "epoch": 1.5725083618475275,
      "grad_norm": 0.05726677551865578,
      "learning_rate": 1.3283152675091065e-06,
      "loss": 0.0008,
      "step": 245180
    },
    {
      "epoch": 1.5725724987413134,
      "grad_norm": 0.05271828919649124,
      "learning_rate": 1.3279353721879429e-06,
      "loss": 0.0008,
      "step": 245190
    },
    {
      "epoch": 1.5726366356350996,
      "grad_norm": 0.3088632822036743,
      "learning_rate": 1.3275555228803726e-06,
      "loss": 0.002,
      "step": 245200
    },
    {
      "epoch": 1.5727007725288855,
      "grad_norm": 0.08474955707788467,
      "learning_rate": 1.3271757195911556e-06,
      "loss": 0.0006,
      "step": 245210
    },
    {
      "epoch": 1.5727649094226717,
      "grad_norm": 0.07594852894544601,
      "learning_rate": 1.3267959623250526e-06,
      "loss": 0.0053,
      "step": 245220
    },
    {
      "epoch": 1.5728290463164578,
      "grad_norm": 0.03569547086954117,
      "learning_rate": 1.3264162510868216e-06,
      "loss": 0.0004,
      "step": 245230
    },
    {
      "epoch": 1.572893183210244,
      "grad_norm": 0.008282431401312351,
      "learning_rate": 1.3260365858812208e-06,
      "loss": 0.0008,
      "step": 245240
    },
    {
      "epoch": 1.57295732010403,
      "grad_norm": 0.0521833673119545,
      "learning_rate": 1.3256569667130053e-06,
      "loss": 0.0012,
      "step": 245250
    },
    {
      "epoch": 1.5730214569978163,
      "grad_norm": 0.14588910341262817,
      "learning_rate": 1.3252773935869357e-06,
      "loss": 0.0019,
      "step": 245260
    },
    {
      "epoch": 1.5730855938916022,
      "grad_norm": 0.11893438547849655,
      "learning_rate": 1.3248978665077667e-06,
      "loss": 0.0008,
      "step": 245270
    },
    {
      "epoch": 1.5731497307853883,
      "grad_norm": 0.04997969791293144,
      "learning_rate": 1.324518385480254e-06,
      "loss": 0.0009,
      "step": 245280
    },
    {
      "epoch": 1.5732138676791745,
      "grad_norm": 0.10066279768943787,
      "learning_rate": 1.3241389505091513e-06,
      "loss": 0.0019,
      "step": 245290
    },
    {
      "epoch": 1.5732780045729604,
      "grad_norm": 0.13450700044631958,
      "learning_rate": 1.323759561599216e-06,
      "loss": 0.0014,
      "step": 245300
    },
    {
      "epoch": 1.5733421414667466,
      "grad_norm": 0.11641976237297058,
      "learning_rate": 1.3233802187552007e-06,
      "loss": 0.0023,
      "step": 245310
    },
    {
      "epoch": 1.5734062783605327,
      "grad_norm": 0.042439937591552734,
      "learning_rate": 1.3230009219818596e-06,
      "loss": 0.001,
      "step": 245320
    },
    {
      "epoch": 1.5734704152543189,
      "grad_norm": 0.027549132704734802,
      "learning_rate": 1.322621671283943e-06,
      "loss": 0.0012,
      "step": 245330
    },
    {
      "epoch": 1.573534552148105,
      "grad_norm": 0.03462856262922287,
      "learning_rate": 1.322242466666206e-06,
      "loss": 0.001,
      "step": 245340
    },
    {
      "epoch": 1.5735986890418912,
      "grad_norm": 0.0055800797417759895,
      "learning_rate": 1.3218633081333997e-06,
      "loss": 0.0023,
      "step": 245350
    },
    {
      "epoch": 1.573662825935677,
      "grad_norm": 0.05559716001152992,
      "learning_rate": 1.3214841956902735e-06,
      "loss": 0.0011,
      "step": 245360
    },
    {
      "epoch": 1.5737269628294632,
      "grad_norm": 0.4655143618583679,
      "learning_rate": 1.3211051293415805e-06,
      "loss": 0.0014,
      "step": 245370
    },
    {
      "epoch": 1.5737910997232492,
      "grad_norm": 0.08998212218284607,
      "learning_rate": 1.3207261090920698e-06,
      "loss": 0.0012,
      "step": 245380
    },
    {
      "epoch": 1.5738552366170353,
      "grad_norm": 0.02035832405090332,
      "learning_rate": 1.3203471349464903e-06,
      "loss": 0.0014,
      "step": 245390
    },
    {
      "epoch": 1.5739193735108215,
      "grad_norm": 0.14554955065250397,
      "learning_rate": 1.3199682069095899e-06,
      "loss": 0.0013,
      "step": 245400
    },
    {
      "epoch": 1.5739835104046076,
      "grad_norm": 0.20792771875858307,
      "learning_rate": 1.3195893249861192e-06,
      "loss": 0.0009,
      "step": 245410
    },
    {
      "epoch": 1.5740476472983937,
      "grad_norm": 0.15382707118988037,
      "learning_rate": 1.3192104891808244e-06,
      "loss": 0.0012,
      "step": 245420
    },
    {
      "epoch": 1.57411178419218,
      "grad_norm": 0.01003416907042265,
      "learning_rate": 1.318831699498453e-06,
      "loss": 0.0003,
      "step": 245430
    },
    {
      "epoch": 1.574175921085966,
      "grad_norm": 0.06154649332165718,
      "learning_rate": 1.3184529559437498e-06,
      "loss": 0.0019,
      "step": 245440
    },
    {
      "epoch": 1.574240057979752,
      "grad_norm": 0.23963524401187897,
      "learning_rate": 1.3180742585214635e-06,
      "loss": 0.0015,
      "step": 245450
    },
    {
      "epoch": 1.5743041948735381,
      "grad_norm": 0.18416304886341095,
      "learning_rate": 1.3176956072363384e-06,
      "loss": 0.0014,
      "step": 245460
    },
    {
      "epoch": 1.574368331767324,
      "grad_norm": 0.05095000937581062,
      "learning_rate": 1.3173170020931187e-06,
      "loss": 0.0006,
      "step": 245470
    },
    {
      "epoch": 1.5744324686611102,
      "grad_norm": 0.03930210322141647,
      "learning_rate": 1.3169384430965483e-06,
      "loss": 0.0013,
      "step": 245480
    },
    {
      "epoch": 1.5744966055548963,
      "grad_norm": 0.11303441971540451,
      "learning_rate": 1.316559930251372e-06,
      "loss": 0.0011,
      "step": 245490
    },
    {
      "epoch": 1.5745607424486825,
      "grad_norm": 0.10995394736528397,
      "learning_rate": 1.3161814635623327e-06,
      "loss": 0.0009,
      "step": 245500
    },
    {
      "epoch": 1.5746248793424686,
      "grad_norm": 0.0212956964969635,
      "learning_rate": 1.3158030430341707e-06,
      "loss": 0.0006,
      "step": 245510
    },
    {
      "epoch": 1.5746890162362548,
      "grad_norm": 0.004086362197995186,
      "learning_rate": 1.3154246686716315e-06,
      "loss": 0.001,
      "step": 245520
    },
    {
      "epoch": 1.5747531531300407,
      "grad_norm": 0.10942432284355164,
      "learning_rate": 1.3150463404794544e-06,
      "loss": 0.0017,
      "step": 245530
    },
    {
      "epoch": 1.5748172900238269,
      "grad_norm": 0.1038489043712616,
      "learning_rate": 1.31466805846238e-06,
      "loss": 0.001,
      "step": 245540
    },
    {
      "epoch": 1.5748814269176128,
      "grad_norm": 0.06717261672019958,
      "learning_rate": 1.3142898226251478e-06,
      "loss": 0.0012,
      "step": 245550
    },
    {
      "epoch": 1.574945563811399,
      "grad_norm": 0.10273327678442001,
      "learning_rate": 1.3139116329724994e-06,
      "loss": 0.0011,
      "step": 245560
    },
    {
      "epoch": 1.575009700705185,
      "grad_norm": 0.374106228351593,
      "learning_rate": 1.3135334895091723e-06,
      "loss": 0.0018,
      "step": 245570
    },
    {
      "epoch": 1.5750738375989712,
      "grad_norm": 0.13840097188949585,
      "learning_rate": 1.3131553922399053e-06,
      "loss": 0.001,
      "step": 245580
    },
    {
      "epoch": 1.5751379744927574,
      "grad_norm": 0.030233973637223244,
      "learning_rate": 1.3127773411694356e-06,
      "loss": 0.001,
      "step": 245590
    },
    {
      "epoch": 1.5752021113865435,
      "grad_norm": 0.06467781215906143,
      "learning_rate": 1.3123993363025018e-06,
      "loss": 0.001,
      "step": 245600
    },
    {
      "epoch": 1.5752662482803297,
      "grad_norm": 0.08963991701602936,
      "learning_rate": 1.3120213776438395e-06,
      "loss": 0.0016,
      "step": 245610
    },
    {
      "epoch": 1.5753303851741156,
      "grad_norm": 0.11008745431900024,
      "learning_rate": 1.3116434651981857e-06,
      "loss": 0.0009,
      "step": 245620
    },
    {
      "epoch": 1.5753945220679018,
      "grad_norm": 0.09108911454677582,
      "learning_rate": 1.3112655989702733e-06,
      "loss": 0.0012,
      "step": 245630
    },
    {
      "epoch": 1.5754586589616877,
      "grad_norm": 0.10402543097734451,
      "learning_rate": 1.3108877789648412e-06,
      "loss": 0.0006,
      "step": 245640
    },
    {
      "epoch": 1.5755227958554738,
      "grad_norm": 0.18037079274654388,
      "learning_rate": 1.3105100051866214e-06,
      "loss": 0.0021,
      "step": 245650
    },
    {
      "epoch": 1.57558693274926,
      "grad_norm": 0.26776322722435,
      "learning_rate": 1.3101322776403468e-06,
      "loss": 0.002,
      "step": 245660
    },
    {
      "epoch": 1.5756510696430461,
      "grad_norm": 0.11705923825502396,
      "learning_rate": 1.309754596330754e-06,
      "loss": 0.0017,
      "step": 245670
    },
    {
      "epoch": 1.5757152065368323,
      "grad_norm": 0.08753743767738342,
      "learning_rate": 1.3093769612625729e-06,
      "loss": 0.001,
      "step": 245680
    },
    {
      "epoch": 1.5757793434306184,
      "grad_norm": 0.05915207788348198,
      "learning_rate": 1.3089993724405358e-06,
      "loss": 0.0009,
      "step": 245690
    },
    {
      "epoch": 1.5758434803244044,
      "grad_norm": 0.23128096759319305,
      "learning_rate": 1.3086218298693732e-06,
      "loss": 0.001,
      "step": 245700
    },
    {
      "epoch": 1.5759076172181905,
      "grad_norm": 0.08078275620937347,
      "learning_rate": 1.3082443335538185e-06,
      "loss": 0.0004,
      "step": 245710
    },
    {
      "epoch": 1.5759717541119767,
      "grad_norm": 0.11938001960515976,
      "learning_rate": 1.3078668834986013e-06,
      "loss": 0.001,
      "step": 245720
    },
    {
      "epoch": 1.5760358910057626,
      "grad_norm": 0.1649659276008606,
      "learning_rate": 1.3074894797084508e-06,
      "loss": 0.0015,
      "step": 245730
    },
    {
      "epoch": 1.5761000278995487,
      "grad_norm": 0.12822167575359344,
      "learning_rate": 1.3071121221880945e-06,
      "loss": 0.0008,
      "step": 245740
    },
    {
      "epoch": 1.5761641647933349,
      "grad_norm": 0.06904137134552002,
      "learning_rate": 1.306734810942264e-06,
      "loss": 0.0005,
      "step": 245750
    },
    {
      "epoch": 1.576228301687121,
      "grad_norm": 0.010498573072254658,
      "learning_rate": 1.3063575459756855e-06,
      "loss": 0.0018,
      "step": 245760
    },
    {
      "epoch": 1.5762924385809072,
      "grad_norm": 0.017474694177508354,
      "learning_rate": 1.3059803272930871e-06,
      "loss": 0.0006,
      "step": 245770
    },
    {
      "epoch": 1.5763565754746933,
      "grad_norm": 0.066599041223526,
      "learning_rate": 1.3056031548991937e-06,
      "loss": 0.0015,
      "step": 245780
    },
    {
      "epoch": 1.5764207123684792,
      "grad_norm": 0.01642470993101597,
      "learning_rate": 1.3052260287987351e-06,
      "loss": 0.0011,
      "step": 245790
    },
    {
      "epoch": 1.5764848492622654,
      "grad_norm": 0.09349098056554794,
      "learning_rate": 1.3048489489964344e-06,
      "loss": 0.0011,
      "step": 245800
    },
    {
      "epoch": 1.5765489861560513,
      "grad_norm": 0.1274600625038147,
      "learning_rate": 1.304471915497016e-06,
      "loss": 0.0011,
      "step": 245810
    },
    {
      "epoch": 1.5766131230498375,
      "grad_norm": 0.12524327635765076,
      "learning_rate": 1.3040949283052069e-06,
      "loss": 0.0017,
      "step": 245820
    },
    {
      "epoch": 1.5766772599436236,
      "grad_norm": 0.01571890339255333,
      "learning_rate": 1.30371798742573e-06,
      "loss": 0.0013,
      "step": 245830
    },
    {
      "epoch": 1.5767413968374098,
      "grad_norm": 0.0890233963727951,
      "learning_rate": 1.3033410928633083e-06,
      "loss": 0.0015,
      "step": 245840
    },
    {
      "epoch": 1.576805533731196,
      "grad_norm": 0.10049532353878021,
      "learning_rate": 1.3029642446226632e-06,
      "loss": 0.0009,
      "step": 245850
    },
    {
      "epoch": 1.576869670624982,
      "grad_norm": 0.06716424971818924,
      "learning_rate": 1.3025874427085193e-06,
      "loss": 0.0005,
      "step": 245860
    },
    {
      "epoch": 1.5769338075187682,
      "grad_norm": 0.04313703626394272,
      "learning_rate": 1.302210687125598e-06,
      "loss": 0.001,
      "step": 245870
    },
    {
      "epoch": 1.5769979444125541,
      "grad_norm": 0.04454496130347252,
      "learning_rate": 1.3018339778786193e-06,
      "loss": 0.0008,
      "step": 245880
    },
    {
      "epoch": 1.5770620813063403,
      "grad_norm": 0.28844645619392395,
      "learning_rate": 1.3014573149723025e-06,
      "loss": 0.0016,
      "step": 245890
    },
    {
      "epoch": 1.5771262182001262,
      "grad_norm": 0.0529690720140934,
      "learning_rate": 1.301080698411369e-06,
      "loss": 0.0023,
      "step": 245900
    },
    {
      "epoch": 1.5771903550939124,
      "grad_norm": 0.07780619710683823,
      "learning_rate": 1.300704128200539e-06,
      "loss": 0.0007,
      "step": 245910
    },
    {
      "epoch": 1.5772544919876985,
      "grad_norm": 0.040550705045461655,
      "learning_rate": 1.3003276043445307e-06,
      "loss": 0.0013,
      "step": 245920
    },
    {
      "epoch": 1.5773186288814847,
      "grad_norm": 0.04916974902153015,
      "learning_rate": 1.2999511268480598e-06,
      "loss": 0.0007,
      "step": 245930
    },
    {
      "epoch": 1.5773827657752708,
      "grad_norm": 0.006002301815897226,
      "learning_rate": 1.2995746957158466e-06,
      "loss": 0.001,
      "step": 245940
    },
    {
      "epoch": 1.577446902669057,
      "grad_norm": 0.10196304321289062,
      "learning_rate": 1.299198310952608e-06,
      "loss": 0.0009,
      "step": 245950
    },
    {
      "epoch": 1.5775110395628429,
      "grad_norm": 0.035125818103551865,
      "learning_rate": 1.298821972563059e-06,
      "loss": 0.0017,
      "step": 245960
    },
    {
      "epoch": 1.577575176456629,
      "grad_norm": 0.011328734457492828,
      "learning_rate": 1.2984456805519146e-06,
      "loss": 0.0007,
      "step": 245970
    },
    {
      "epoch": 1.577639313350415,
      "grad_norm": 0.007865257561206818,
      "learning_rate": 1.2980694349238925e-06,
      "loss": 0.0013,
      "step": 245980
    },
    {
      "epoch": 1.577703450244201,
      "grad_norm": 0.023871131241321564,
      "learning_rate": 1.2976932356837064e-06,
      "loss": 0.0011,
      "step": 245990
    },
    {
      "epoch": 1.5777675871379873,
      "grad_norm": 0.060003168880939484,
      "learning_rate": 1.2973170828360698e-06,
      "loss": 0.0008,
      "step": 246000
    },
    {
      "epoch": 1.5778317240317734,
      "grad_norm": 0.04944562539458275,
      "learning_rate": 1.2969409763856954e-06,
      "loss": 0.0009,
      "step": 246010
    },
    {
      "epoch": 1.5778958609255596,
      "grad_norm": 0.13507193326950073,
      "learning_rate": 1.2965649163372985e-06,
      "loss": 0.0013,
      "step": 246020
    },
    {
      "epoch": 1.5779599978193457,
      "grad_norm": 0.009508341550827026,
      "learning_rate": 1.2961889026955898e-06,
      "loss": 0.0008,
      "step": 246030
    },
    {
      "epoch": 1.5780241347131319,
      "grad_norm": 0.15763916075229645,
      "learning_rate": 1.2958129354652809e-06,
      "loss": 0.0021,
      "step": 246040
    },
    {
      "epoch": 1.5780882716069178,
      "grad_norm": 0.1158578023314476,
      "learning_rate": 1.295437014651082e-06,
      "loss": 0.003,
      "step": 246050
    },
    {
      "epoch": 1.578152408500704,
      "grad_norm": 0.08165355771780014,
      "learning_rate": 1.2950611402577068e-06,
      "loss": 0.002,
      "step": 246060
    },
    {
      "epoch": 1.5782165453944899,
      "grad_norm": 0.11118276417255402,
      "learning_rate": 1.2946853122898629e-06,
      "loss": 0.0008,
      "step": 246070
    },
    {
      "epoch": 1.578280682288276,
      "grad_norm": 0.025926100090146065,
      "learning_rate": 1.2943095307522592e-06,
      "loss": 0.0012,
      "step": 246080
    },
    {
      "epoch": 1.5783448191820622,
      "grad_norm": 0.0540606826543808,
      "learning_rate": 1.2939337956496062e-06,
      "loss": 0.0008,
      "step": 246090
    },
    {
      "epoch": 1.5784089560758483,
      "grad_norm": 0.07775036990642548,
      "learning_rate": 1.2935581069866122e-06,
      "loss": 0.0028,
      "step": 246100
    },
    {
      "epoch": 1.5784730929696345,
      "grad_norm": 0.09212201088666916,
      "learning_rate": 1.2931824647679837e-06,
      "loss": 0.0008,
      "step": 246110
    },
    {
      "epoch": 1.5785372298634206,
      "grad_norm": 0.013664871454238892,
      "learning_rate": 1.2928068689984268e-06,
      "loss": 0.0016,
      "step": 246120
    },
    {
      "epoch": 1.5786013667572067,
      "grad_norm": 0.04631384462118149,
      "learning_rate": 1.2924313196826504e-06,
      "loss": 0.0007,
      "step": 246130
    },
    {
      "epoch": 1.5786655036509927,
      "grad_norm": 0.08882184326648712,
      "learning_rate": 1.29205581682536e-06,
      "loss": 0.0011,
      "step": 246140
    },
    {
      "epoch": 1.5787296405447788,
      "grad_norm": 0.016570575535297394,
      "learning_rate": 1.29168036043126e-06,
      "loss": 0.0008,
      "step": 246150
    },
    {
      "epoch": 1.5787937774385647,
      "grad_norm": 0.11887503415346146,
      "learning_rate": 1.2913049505050535e-06,
      "loss": 0.0006,
      "step": 246160
    },
    {
      "epoch": 1.578857914332351,
      "grad_norm": 0.08305158466100693,
      "learning_rate": 1.2909295870514482e-06,
      "loss": 0.0012,
      "step": 246170
    },
    {
      "epoch": 1.578922051226137,
      "grad_norm": 0.0012892017839476466,
      "learning_rate": 1.2905542700751461e-06,
      "loss": 0.0005,
      "step": 246180
    },
    {
      "epoch": 1.5789861881199232,
      "grad_norm": 0.005535861011594534,
      "learning_rate": 1.2901789995808506e-06,
      "loss": 0.0005,
      "step": 246190
    },
    {
      "epoch": 1.5790503250137093,
      "grad_norm": 0.020395416766405106,
      "learning_rate": 1.2898037755732613e-06,
      "loss": 0.0087,
      "step": 246200
    },
    {
      "epoch": 1.5791144619074955,
      "grad_norm": 0.0902070477604866,
      "learning_rate": 1.2894285980570842e-06,
      "loss": 0.0016,
      "step": 246210
    },
    {
      "epoch": 1.5791785988012814,
      "grad_norm": 0.06517552584409714,
      "learning_rate": 1.2890534670370187e-06,
      "loss": 0.0015,
      "step": 246220
    },
    {
      "epoch": 1.5792427356950676,
      "grad_norm": 0.05345199629664421,
      "learning_rate": 1.288678382517764e-06,
      "loss": 0.0011,
      "step": 246230
    },
    {
      "epoch": 1.5793068725888535,
      "grad_norm": 0.07760341465473175,
      "learning_rate": 1.2883033445040228e-06,
      "loss": 0.0009,
      "step": 246240
    },
    {
      "epoch": 1.5793710094826396,
      "grad_norm": 0.1683817356824875,
      "learning_rate": 1.2879283530004932e-06,
      "loss": 0.0009,
      "step": 246250
    },
    {
      "epoch": 1.5794351463764258,
      "grad_norm": 0.0635901466012001,
      "learning_rate": 1.2875534080118751e-06,
      "loss": 0.0012,
      "step": 246260
    },
    {
      "epoch": 1.579499283270212,
      "grad_norm": 0.03265683725476265,
      "learning_rate": 1.287178509542864e-06,
      "loss": 0.0019,
      "step": 246270
    },
    {
      "epoch": 1.579563420163998,
      "grad_norm": 0.08981554955244064,
      "learning_rate": 1.2868036575981613e-06,
      "loss": 0.0011,
      "step": 246280
    },
    {
      "epoch": 1.5796275570577842,
      "grad_norm": 0.052812643349170685,
      "learning_rate": 1.2864288521824625e-06,
      "loss": 0.0022,
      "step": 246290
    },
    {
      "epoch": 1.5796916939515704,
      "grad_norm": 0.030123114585876465,
      "learning_rate": 1.2860540933004645e-06,
      "loss": 0.0009,
      "step": 246300
    },
    {
      "epoch": 1.5797558308453563,
      "grad_norm": 0.027613233774900436,
      "learning_rate": 1.2856793809568613e-06,
      "loss": 0.0012,
      "step": 246310
    },
    {
      "epoch": 1.5798199677391425,
      "grad_norm": 0.07647305727005005,
      "learning_rate": 1.2853047151563513e-06,
      "loss": 0.0013,
      "step": 246320
    },
    {
      "epoch": 1.5798841046329284,
      "grad_norm": 0.09239564836025238,
      "learning_rate": 1.2849300959036287e-06,
      "loss": 0.0009,
      "step": 246330
    },
    {
      "epoch": 1.5799482415267145,
      "grad_norm": 0.05426836013793945,
      "learning_rate": 1.2845555232033852e-06,
      "loss": 0.0009,
      "step": 246340
    },
    {
      "epoch": 1.5800123784205007,
      "grad_norm": 0.051104720681905746,
      "learning_rate": 1.2841809970603176e-06,
      "loss": 0.0009,
      "step": 246350
    },
    {
      "epoch": 1.5800765153142868,
      "grad_norm": 0.15226265788078308,
      "learning_rate": 1.2838065174791182e-06,
      "loss": 0.0012,
      "step": 246360
    },
    {
      "epoch": 1.580140652208073,
      "grad_norm": 0.07135067135095596,
      "learning_rate": 1.2834320844644788e-06,
      "loss": 0.0026,
      "step": 246370
    },
    {
      "epoch": 1.5802047891018591,
      "grad_norm": 0.01079430803656578,
      "learning_rate": 1.2830576980210906e-06,
      "loss": 0.0013,
      "step": 246380
    },
    {
      "epoch": 1.580268925995645,
      "grad_norm": 0.051921162754297256,
      "learning_rate": 1.282683358153647e-06,
      "loss": 0.0012,
      "step": 246390
    },
    {
      "epoch": 1.5803330628894312,
      "grad_norm": 0.0851326659321785,
      "learning_rate": 1.2823090648668375e-06,
      "loss": 0.0019,
      "step": 246400
    },
    {
      "epoch": 1.5803971997832171,
      "grad_norm": 0.042742229998111725,
      "learning_rate": 1.281934818165353e-06,
      "loss": 0.002,
      "step": 246410
    },
    {
      "epoch": 1.5804613366770033,
      "grad_norm": 0.11216244101524353,
      "learning_rate": 1.2815606180538804e-06,
      "loss": 0.0023,
      "step": 246420
    },
    {
      "epoch": 1.5805254735707894,
      "grad_norm": 0.010424695909023285,
      "learning_rate": 1.2811864645371124e-06,
      "loss": 0.0007,
      "step": 246430
    },
    {
      "epoch": 1.5805896104645756,
      "grad_norm": 0.04899062588810921,
      "learning_rate": 1.2808123576197356e-06,
      "loss": 0.0013,
      "step": 246440
    },
    {
      "epoch": 1.5806537473583617,
      "grad_norm": 0.1219649612903595,
      "learning_rate": 1.2804382973064383e-06,
      "loss": 0.0012,
      "step": 246450
    },
    {
      "epoch": 1.5807178842521479,
      "grad_norm": 0.018149536103010178,
      "learning_rate": 1.2800642836019062e-06,
      "loss": 0.0009,
      "step": 246460
    },
    {
      "epoch": 1.580782021145934,
      "grad_norm": 0.11209143698215485,
      "learning_rate": 1.2796903165108282e-06,
      "loss": 0.0025,
      "step": 246470
    },
    {
      "epoch": 1.58084615803972,
      "grad_norm": 0.024495825171470642,
      "learning_rate": 1.2793163960378896e-06,
      "loss": 0.001,
      "step": 246480
    },
    {
      "epoch": 1.580910294933506,
      "grad_norm": 0.08707740902900696,
      "learning_rate": 1.2789425221877743e-06,
      "loss": 0.0014,
      "step": 246490
    },
    {
      "epoch": 1.580974431827292,
      "grad_norm": 0.09110743552446365,
      "learning_rate": 1.2785686949651704e-06,
      "loss": 0.0011,
      "step": 246500
    },
    {
      "epoch": 1.5810385687210782,
      "grad_norm": 0.02921900525689125,
      "learning_rate": 1.2781949143747603e-06,
      "loss": 0.0006,
      "step": 246510
    },
    {
      "epoch": 1.5811027056148643,
      "grad_norm": 0.08461631089448929,
      "learning_rate": 1.277821180421227e-06,
      "loss": 0.0012,
      "step": 246520
    },
    {
      "epoch": 1.5811668425086505,
      "grad_norm": 0.09141971170902252,
      "learning_rate": 1.277447493109254e-06,
      "loss": 0.001,
      "step": 246530
    },
    {
      "epoch": 1.5812309794024366,
      "grad_norm": 0.0596066415309906,
      "learning_rate": 1.2770738524435255e-06,
      "loss": 0.0017,
      "step": 246540
    },
    {
      "epoch": 1.5812951162962228,
      "grad_norm": 0.07852587848901749,
      "learning_rate": 1.2767002584287224e-06,
      "loss": 0.002,
      "step": 246550
    },
    {
      "epoch": 1.581359253190009,
      "grad_norm": 0.0018973442493006587,
      "learning_rate": 1.276326711069526e-06,
      "loss": 0.0006,
      "step": 246560
    },
    {
      "epoch": 1.5814233900837948,
      "grad_norm": 0.12400216609239578,
      "learning_rate": 1.2759532103706157e-06,
      "loss": 0.002,
      "step": 246570
    },
    {
      "epoch": 1.581487526977581,
      "grad_norm": 0.07634778320789337,
      "learning_rate": 1.275579756336675e-06,
      "loss": 0.0007,
      "step": 246580
    },
    {
      "epoch": 1.581551663871367,
      "grad_norm": 0.025330260396003723,
      "learning_rate": 1.2752063489723816e-06,
      "loss": 0.0006,
      "step": 246590
    },
    {
      "epoch": 1.581615800765153,
      "grad_norm": 0.07662412524223328,
      "learning_rate": 1.2748329882824146e-06,
      "loss": 0.0016,
      "step": 246600
    },
    {
      "epoch": 1.5816799376589392,
      "grad_norm": 0.06542450189590454,
      "learning_rate": 1.274459674271451e-06,
      "loss": 0.0023,
      "step": 246610
    },
    {
      "epoch": 1.5817440745527254,
      "grad_norm": 0.037552472203969955,
      "learning_rate": 1.2740864069441716e-06,
      "loss": 0.0006,
      "step": 246620
    },
    {
      "epoch": 1.5818082114465115,
      "grad_norm": 0.0035921186208724976,
      "learning_rate": 1.2737131863052526e-06,
      "loss": 0.0018,
      "step": 246630
    },
    {
      "epoch": 1.5818723483402977,
      "grad_norm": 0.1471884846687317,
      "learning_rate": 1.2733400123593692e-06,
      "loss": 0.001,
      "step": 246640
    },
    {
      "epoch": 1.5819364852340836,
      "grad_norm": 0.04834791645407677,
      "learning_rate": 1.2729668851111987e-06,
      "loss": 0.001,
      "step": 246650
    },
    {
      "epoch": 1.5820006221278697,
      "grad_norm": 0.08057847619056702,
      "learning_rate": 1.2725938045654185e-06,
      "loss": 0.0017,
      "step": 246660
    },
    {
      "epoch": 1.5820647590216557,
      "grad_norm": 0.07877923548221588,
      "learning_rate": 1.272220770726702e-06,
      "loss": 0.0007,
      "step": 246670
    },
    {
      "epoch": 1.5821288959154418,
      "grad_norm": 0.07737217843532562,
      "learning_rate": 1.2718477835997229e-06,
      "loss": 0.001,
      "step": 246680
    },
    {
      "epoch": 1.582193032809228,
      "grad_norm": 0.08197028189897537,
      "learning_rate": 1.2714748431891545e-06,
      "loss": 0.0011,
      "step": 246690
    },
    {
      "epoch": 1.582257169703014,
      "grad_norm": 0.20422714948654175,
      "learning_rate": 1.2711019494996723e-06,
      "loss": 0.002,
      "step": 246700
    },
    {
      "epoch": 1.5823213065968003,
      "grad_norm": 0.15007200837135315,
      "learning_rate": 1.2707291025359475e-06,
      "loss": 0.001,
      "step": 246710
    },
    {
      "epoch": 1.5823854434905864,
      "grad_norm": 0.08318213373422623,
      "learning_rate": 1.2703563023026527e-06,
      "loss": 0.0017,
      "step": 246720
    },
    {
      "epoch": 1.5824495803843726,
      "grad_norm": 0.03313397616147995,
      "learning_rate": 1.269983548804457e-06,
      "loss": 0.0016,
      "step": 246730
    },
    {
      "epoch": 1.5825137172781585,
      "grad_norm": 0.11351385712623596,
      "learning_rate": 1.2696108420460352e-06,
      "loss": 0.0008,
      "step": 246740
    },
    {
      "epoch": 1.5825778541719446,
      "grad_norm": 0.07731682062149048,
      "learning_rate": 1.2692381820320554e-06,
      "loss": 0.0015,
      "step": 246750
    },
    {
      "epoch": 1.5826419910657306,
      "grad_norm": 0.13076956570148468,
      "learning_rate": 1.2688655687671864e-06,
      "loss": 0.0007,
      "step": 246760
    },
    {
      "epoch": 1.5827061279595167,
      "grad_norm": 0.13700582087039948,
      "learning_rate": 1.2684930022560992e-06,
      "loss": 0.0017,
      "step": 246770
    },
    {
      "epoch": 1.5827702648533029,
      "grad_norm": 0.08193062245845795,
      "learning_rate": 1.2681204825034621e-06,
      "loss": 0.0007,
      "step": 246780
    },
    {
      "epoch": 1.582834401747089,
      "grad_norm": 0.08168069273233414,
      "learning_rate": 1.2677480095139427e-06,
      "loss": 0.0011,
      "step": 246790
    },
    {
      "epoch": 1.5828985386408752,
      "grad_norm": 0.1415787637233734,
      "learning_rate": 1.2673755832922064e-06,
      "loss": 0.0013,
      "step": 246800
    },
    {
      "epoch": 1.5829626755346613,
      "grad_norm": 0.08244706690311432,
      "learning_rate": 1.2670032038429225e-06,
      "loss": 0.0015,
      "step": 246810
    },
    {
      "epoch": 1.5830268124284472,
      "grad_norm": 0.017394885420799255,
      "learning_rate": 1.266630871170757e-06,
      "loss": 0.0009,
      "step": 246820
    },
    {
      "epoch": 1.5830909493222334,
      "grad_norm": 0.047168806195259094,
      "learning_rate": 1.266258585280375e-06,
      "loss": 0.0004,
      "step": 246830
    },
    {
      "epoch": 1.5831550862160195,
      "grad_norm": 0.019143542274832726,
      "learning_rate": 1.2658863461764398e-06,
      "loss": 0.0008,
      "step": 246840
    },
    {
      "epoch": 1.5832192231098055,
      "grad_norm": 0.12793073058128357,
      "learning_rate": 1.265514153863619e-06,
      "loss": 0.0008,
      "step": 246850
    },
    {
      "epoch": 1.5832833600035916,
      "grad_norm": 0.028690339997410774,
      "learning_rate": 1.2651420083465748e-06,
      "loss": 0.0007,
      "step": 246860
    },
    {
      "epoch": 1.5833474968973777,
      "grad_norm": 0.060283955186605453,
      "learning_rate": 1.2647699096299703e-06,
      "loss": 0.0018,
      "step": 246870
    },
    {
      "epoch": 1.583411633791164,
      "grad_norm": 0.1377541571855545,
      "learning_rate": 1.2643978577184668e-06,
      "loss": 0.0026,
      "step": 246880
    },
    {
      "epoch": 1.58347577068495,
      "grad_norm": 0.005368927028030157,
      "learning_rate": 1.2640258526167298e-06,
      "loss": 0.0013,
      "step": 246890
    },
    {
      "epoch": 1.5835399075787362,
      "grad_norm": 0.0832783579826355,
      "learning_rate": 1.2636538943294186e-06,
      "loss": 0.0013,
      "step": 246900
    },
    {
      "epoch": 1.5836040444725221,
      "grad_norm": 0.05644654110074043,
      "learning_rate": 1.2632819828611931e-06,
      "loss": 0.0021,
      "step": 246910
    },
    {
      "epoch": 1.5836681813663083,
      "grad_norm": 0.06918276101350784,
      "learning_rate": 1.2629101182167164e-06,
      "loss": 0.0031,
      "step": 246920
    },
    {
      "epoch": 1.5837323182600942,
      "grad_norm": 0.10213880240917206,
      "learning_rate": 1.262538300400647e-06,
      "loss": 0.0038,
      "step": 246930
    },
    {
      "epoch": 1.5837964551538803,
      "grad_norm": 0.11579898744821548,
      "learning_rate": 1.2621665294176433e-06,
      "loss": 0.0011,
      "step": 246940
    },
    {
      "epoch": 1.5838605920476665,
      "grad_norm": 0.04726351425051689,
      "learning_rate": 1.2617948052723633e-06,
      "loss": 0.0008,
      "step": 246950
    },
    {
      "epoch": 1.5839247289414526,
      "grad_norm": 0.1257619559764862,
      "learning_rate": 1.2614231279694678e-06,
      "loss": 0.0012,
      "step": 246960
    },
    {
      "epoch": 1.5839888658352388,
      "grad_norm": 0.040273942053318024,
      "learning_rate": 1.2610514975136118e-06,
      "loss": 0.0006,
      "step": 246970
    },
    {
      "epoch": 1.584053002729025,
      "grad_norm": 0.19437241554260254,
      "learning_rate": 1.2606799139094529e-06,
      "loss": 0.0017,
      "step": 246980
    },
    {
      "epoch": 1.584117139622811,
      "grad_norm": 0.16303570568561554,
      "learning_rate": 1.2603083771616459e-06,
      "loss": 0.0012,
      "step": 246990
    },
    {
      "epoch": 1.584181276516597,
      "grad_norm": 0.050766024738550186,
      "learning_rate": 1.2599368872748491e-06,
      "loss": 0.0008,
      "step": 247000
    },
    {
      "epoch": 1.5842454134103832,
      "grad_norm": 0.11944007128477097,
      "learning_rate": 1.2595654442537158e-06,
      "loss": 0.002,
      "step": 247010
    },
    {
      "epoch": 1.584309550304169,
      "grad_norm": 0.0842583030462265,
      "learning_rate": 1.259194048102901e-06,
      "loss": 0.0011,
      "step": 247020
    },
    {
      "epoch": 1.5843736871979552,
      "grad_norm": 0.09873521327972412,
      "learning_rate": 1.2588226988270564e-06,
      "loss": 0.0014,
      "step": 247030
    },
    {
      "epoch": 1.5844378240917414,
      "grad_norm": 0.07508692145347595,
      "learning_rate": 1.2584513964308393e-06,
      "loss": 0.0011,
      "step": 247040
    },
    {
      "epoch": 1.5845019609855275,
      "grad_norm": 0.10218508541584015,
      "learning_rate": 1.2580801409188997e-06,
      "loss": 0.0009,
      "step": 247050
    },
    {
      "epoch": 1.5845660978793137,
      "grad_norm": 0.08793481439352036,
      "learning_rate": 1.257708932295889e-06,
      "loss": 0.0009,
      "step": 247060
    },
    {
      "epoch": 1.5846302347730998,
      "grad_norm": 0.04642735421657562,
      "learning_rate": 1.2573377705664613e-06,
      "loss": 0.0015,
      "step": 247070
    },
    {
      "epoch": 1.5846943716668858,
      "grad_norm": 0.1290217638015747,
      "learning_rate": 1.2569666557352662e-06,
      "loss": 0.0011,
      "step": 247080
    },
    {
      "epoch": 1.584758508560672,
      "grad_norm": 0.06868187338113785,
      "learning_rate": 1.256595587806954e-06,
      "loss": 0.0013,
      "step": 247090
    },
    {
      "epoch": 1.5848226454544578,
      "grad_norm": 0.05559730902314186,
      "learning_rate": 1.2562245667861727e-06,
      "loss": 0.0006,
      "step": 247100
    },
    {
      "epoch": 1.584886782348244,
      "grad_norm": 0.09011746197938919,
      "learning_rate": 1.2558535926775757e-06,
      "loss": 0.0034,
      "step": 247110
    },
    {
      "epoch": 1.5849509192420301,
      "grad_norm": 0.07295114547014236,
      "learning_rate": 1.2554826654858081e-06,
      "loss": 0.0007,
      "step": 247120
    },
    {
      "epoch": 1.5850150561358163,
      "grad_norm": 0.14845941960811615,
      "learning_rate": 1.2551117852155192e-06,
      "loss": 0.0011,
      "step": 247130
    },
    {
      "epoch": 1.5850791930296024,
      "grad_norm": 0.0017486442811787128,
      "learning_rate": 1.2547409518713543e-06,
      "loss": 0.001,
      "step": 247140
    },
    {
      "epoch": 1.5851433299233886,
      "grad_norm": 0.0336947999894619,
      "learning_rate": 1.2543701654579637e-06,
      "loss": 0.001,
      "step": 247150
    },
    {
      "epoch": 1.5852074668171747,
      "grad_norm": 0.22223295271396637,
      "learning_rate": 1.253999425979992e-06,
      "loss": 0.0007,
      "step": 247160
    },
    {
      "epoch": 1.5852716037109607,
      "grad_norm": 0.0942402258515358,
      "learning_rate": 1.2536287334420848e-06,
      "loss": 0.0018,
      "step": 247170
    },
    {
      "epoch": 1.5853357406047468,
      "grad_norm": 0.029933787882328033,
      "learning_rate": 1.2532580878488854e-06,
      "loss": 0.0024,
      "step": 247180
    },
    {
      "epoch": 1.5853998774985327,
      "grad_norm": 0.12282829731702805,
      "learning_rate": 1.2528874892050414e-06,
      "loss": 0.0012,
      "step": 247190
    },
    {
      "epoch": 1.5854640143923189,
      "grad_norm": 0.05031512677669525,
      "learning_rate": 1.2525169375151953e-06,
      "loss": 0.0007,
      "step": 247200
    },
    {
      "epoch": 1.585528151286105,
      "grad_norm": 0.06032634153962135,
      "learning_rate": 1.2521464327839882e-06,
      "loss": 0.0009,
      "step": 247210
    },
    {
      "epoch": 1.5855922881798912,
      "grad_norm": 0.057162925601005554,
      "learning_rate": 1.2517759750160667e-06,
      "loss": 0.0011,
      "step": 247220
    },
    {
      "epoch": 1.5856564250736773,
      "grad_norm": 0.07621420919895172,
      "learning_rate": 1.2514055642160716e-06,
      "loss": 0.0036,
      "step": 247230
    },
    {
      "epoch": 1.5857205619674635,
      "grad_norm": 0.0914858728647232,
      "learning_rate": 1.2510352003886432e-06,
      "loss": 0.0009,
      "step": 247240
    },
    {
      "epoch": 1.5857846988612494,
      "grad_norm": 0.0333448126912117,
      "learning_rate": 1.2506648835384221e-06,
      "loss": 0.0013,
      "step": 247250
    },
    {
      "epoch": 1.5858488357550355,
      "grad_norm": 0.05894660949707031,
      "learning_rate": 1.2502946136700507e-06,
      "loss": 0.0018,
      "step": 247260
    },
    {
      "epoch": 1.5859129726488217,
      "grad_norm": 0.08252187818288803,
      "learning_rate": 1.2499243907881676e-06,
      "loss": 0.0009,
      "step": 247270
    },
    {
      "epoch": 1.5859771095426076,
      "grad_norm": 0.020556898787617683,
      "learning_rate": 1.2495542148974121e-06,
      "loss": 0.0022,
      "step": 247280
    },
    {
      "epoch": 1.5860412464363938,
      "grad_norm": 0.09331326931715012,
      "learning_rate": 1.2491840860024212e-06,
      "loss": 0.0015,
      "step": 247290
    },
    {
      "epoch": 1.58610538333018,
      "grad_norm": 0.1433272659778595,
      "learning_rate": 1.2488140041078362e-06,
      "loss": 0.0012,
      "step": 247300
    },
    {
      "epoch": 1.586169520223966,
      "grad_norm": 0.21614956855773926,
      "learning_rate": 1.2484439692182926e-06,
      "loss": 0.0017,
      "step": 247310
    },
    {
      "epoch": 1.5862336571177522,
      "grad_norm": 0.02656024694442749,
      "learning_rate": 1.2480739813384268e-06,
      "loss": 0.0014,
      "step": 247320
    },
    {
      "epoch": 1.5862977940115384,
      "grad_norm": 0.12968099117279053,
      "learning_rate": 1.2477040404728741e-06,
      "loss": 0.0017,
      "step": 247330
    },
    {
      "epoch": 1.5863619309053243,
      "grad_norm": 0.015749948099255562,
      "learning_rate": 1.2473341466262734e-06,
      "loss": 0.0015,
      "step": 247340
    },
    {
      "epoch": 1.5864260677991104,
      "grad_norm": 0.07174625247716904,
      "learning_rate": 1.2469642998032577e-06,
      "loss": 0.001,
      "step": 247350
    },
    {
      "epoch": 1.5864902046928964,
      "grad_norm": 0.05340525507926941,
      "learning_rate": 1.24659450000846e-06,
      "loss": 0.0005,
      "step": 247360
    },
    {
      "epoch": 1.5865543415866825,
      "grad_norm": 0.15386264026165009,
      "learning_rate": 1.2462247472465172e-06,
      "loss": 0.0018,
      "step": 247370
    },
    {
      "epoch": 1.5866184784804687,
      "grad_norm": 0.02587992511689663,
      "learning_rate": 1.245855041522061e-06,
      "loss": 0.0012,
      "step": 247380
    },
    {
      "epoch": 1.5866826153742548,
      "grad_norm": 0.21163691580295563,
      "learning_rate": 1.245485382839724e-06,
      "loss": 0.0023,
      "step": 247390
    },
    {
      "epoch": 1.586746752268041,
      "grad_norm": 0.26643458008766174,
      "learning_rate": 1.2451157712041374e-06,
      "loss": 0.0034,
      "step": 247400
    },
    {
      "epoch": 1.586810889161827,
      "grad_norm": 0.04723113775253296,
      "learning_rate": 1.244746206619935e-06,
      "loss": 0.0009,
      "step": 247410
    },
    {
      "epoch": 1.5868750260556133,
      "grad_norm": 0.01681624911725521,
      "learning_rate": 1.2443766890917452e-06,
      "loss": 0.0006,
      "step": 247420
    },
    {
      "epoch": 1.5869391629493992,
      "grad_norm": 0.014629567973315716,
      "learning_rate": 1.2440072186242008e-06,
      "loss": 0.0021,
      "step": 247430
    },
    {
      "epoch": 1.5870032998431853,
      "grad_norm": 0.06550707668066025,
      "learning_rate": 1.243637795221931e-06,
      "loss": 0.0012,
      "step": 247440
    },
    {
      "epoch": 1.5870674367369713,
      "grad_norm": 0.10899477452039719,
      "learning_rate": 1.2432684188895616e-06,
      "loss": 0.0006,
      "step": 247450
    },
    {
      "epoch": 1.5871315736307574,
      "grad_norm": 0.10097301006317139,
      "learning_rate": 1.242899089631726e-06,
      "loss": 0.0006,
      "step": 247460
    },
    {
      "epoch": 1.5871957105245436,
      "grad_norm": 0.12945745885372162,
      "learning_rate": 1.2425298074530502e-06,
      "loss": 0.0012,
      "step": 247470
    },
    {
      "epoch": 1.5872598474183297,
      "grad_norm": 0.06421015411615372,
      "learning_rate": 1.2421605723581593e-06,
      "loss": 0.0013,
      "step": 247480
    },
    {
      "epoch": 1.5873239843121159,
      "grad_norm": 0.06142069771885872,
      "learning_rate": 1.2417913843516838e-06,
      "loss": 0.0015,
      "step": 247490
    },
    {
      "epoch": 1.587388121205902,
      "grad_norm": 0.09488467872142792,
      "learning_rate": 1.2414222434382483e-06,
      "loss": 0.0015,
      "step": 247500
    },
    {
      "epoch": 1.587452258099688,
      "grad_norm": 0.037198759615421295,
      "learning_rate": 1.2410531496224782e-06,
      "loss": 0.0013,
      "step": 247510
    },
    {
      "epoch": 1.587516394993474,
      "grad_norm": 0.04325252026319504,
      "learning_rate": 1.2406841029089972e-06,
      "loss": 0.0013,
      "step": 247520
    },
    {
      "epoch": 1.58758053188726,
      "grad_norm": 0.0072103929705917835,
      "learning_rate": 1.240315103302433e-06,
      "loss": 0.0006,
      "step": 247530
    },
    {
      "epoch": 1.5876446687810462,
      "grad_norm": 0.0305376797914505,
      "learning_rate": 1.2399461508074067e-06,
      "loss": 0.002,
      "step": 247540
    },
    {
      "epoch": 1.5877088056748323,
      "grad_norm": 0.10464880615472794,
      "learning_rate": 1.239577245428543e-06,
      "loss": 0.0017,
      "step": 247550
    },
    {
      "epoch": 1.5877729425686185,
      "grad_norm": 0.18822775781154633,
      "learning_rate": 1.2392083871704624e-06,
      "loss": 0.0025,
      "step": 247560
    },
    {
      "epoch": 1.5878370794624046,
      "grad_norm": 0.07013606280088425,
      "learning_rate": 1.2388395760377896e-06,
      "loss": 0.0012,
      "step": 247570
    },
    {
      "epoch": 1.5879012163561907,
      "grad_norm": 0.03351839259266853,
      "learning_rate": 1.2384708120351458e-06,
      "loss": 0.0008,
      "step": 247580
    },
    {
      "epoch": 1.587965353249977,
      "grad_norm": 0.04853971675038338,
      "learning_rate": 1.2381020951671502e-06,
      "loss": 0.0018,
      "step": 247590
    },
    {
      "epoch": 1.5880294901437628,
      "grad_norm": 0.08642683178186417,
      "learning_rate": 1.2377334254384232e-06,
      "loss": 0.0012,
      "step": 247600
    },
    {
      "epoch": 1.588093627037549,
      "grad_norm": 0.08164126425981522,
      "learning_rate": 1.2373648028535862e-06,
      "loss": 0.0013,
      "step": 247610
    },
    {
      "epoch": 1.588157763931335,
      "grad_norm": 0.12997667491436005,
      "learning_rate": 1.2369962274172575e-06,
      "loss": 0.0012,
      "step": 247620
    },
    {
      "epoch": 1.588221900825121,
      "grad_norm": 0.42265182733535767,
      "learning_rate": 1.2366276991340537e-06,
      "loss": 0.0017,
      "step": 247630
    },
    {
      "epoch": 1.5882860377189072,
      "grad_norm": 0.025046564638614655,
      "learning_rate": 1.2362592180085963e-06,
      "loss": 0.0005,
      "step": 247640
    },
    {
      "epoch": 1.5883501746126933,
      "grad_norm": 0.010669128969311714,
      "learning_rate": 1.2358907840455004e-06,
      "loss": 0.0013,
      "step": 247650
    },
    {
      "epoch": 1.5884143115064795,
      "grad_norm": 0.019329898059368134,
      "learning_rate": 1.2355223972493835e-06,
      "loss": 0.0017,
      "step": 247660
    },
    {
      "epoch": 1.5884784484002656,
      "grad_norm": 0.032687701284885406,
      "learning_rate": 1.23515405762486e-06,
      "loss": 0.0011,
      "step": 247670
    },
    {
      "epoch": 1.5885425852940518,
      "grad_norm": 0.028135398402810097,
      "learning_rate": 1.2347857651765483e-06,
      "loss": 0.0005,
      "step": 247680
    },
    {
      "epoch": 1.5886067221878377,
      "grad_norm": 0.018525226041674614,
      "learning_rate": 1.2344175199090613e-06,
      "loss": 0.0011,
      "step": 247690
    },
    {
      "epoch": 1.5886708590816239,
      "grad_norm": 0.045167919248342514,
      "learning_rate": 1.234049321827015e-06,
      "loss": 0.0017,
      "step": 247700
    },
    {
      "epoch": 1.5887349959754098,
      "grad_norm": 0.10461270064115524,
      "learning_rate": 1.23368117093502e-06,
      "loss": 0.0018,
      "step": 247710
    },
    {
      "epoch": 1.588799132869196,
      "grad_norm": 0.06808051466941833,
      "learning_rate": 1.233313067237693e-06,
      "loss": 0.0021,
      "step": 247720
    },
    {
      "epoch": 1.588863269762982,
      "grad_norm": 0.0033471379429101944,
      "learning_rate": 1.2329450107396456e-06,
      "loss": 0.0013,
      "step": 247730
    },
    {
      "epoch": 1.5889274066567682,
      "grad_norm": 0.1340583711862564,
      "learning_rate": 1.2325770014454897e-06,
      "loss": 0.0012,
      "step": 247740
    },
    {
      "epoch": 1.5889915435505544,
      "grad_norm": 0.029970020055770874,
      "learning_rate": 1.2322090393598352e-06,
      "loss": 0.0015,
      "step": 247750
    },
    {
      "epoch": 1.5890556804443405,
      "grad_norm": 0.034044284373521805,
      "learning_rate": 1.2318411244872952e-06,
      "loss": 0.001,
      "step": 247760
    },
    {
      "epoch": 1.5891198173381265,
      "grad_norm": 0.06569530069828033,
      "learning_rate": 1.231473256832479e-06,
      "loss": 0.0005,
      "step": 247770
    },
    {
      "epoch": 1.5891839542319126,
      "grad_norm": 0.1073935404419899,
      "learning_rate": 1.2311054363999948e-06,
      "loss": 0.0016,
      "step": 247780
    },
    {
      "epoch": 1.5892480911256985,
      "grad_norm": 0.011293603107333183,
      "learning_rate": 1.2307376631944545e-06,
      "loss": 0.0011,
      "step": 247790
    },
    {
      "epoch": 1.5893122280194847,
      "grad_norm": 0.004708931315690279,
      "learning_rate": 1.2303699372204653e-06,
      "loss": 0.0018,
      "step": 247800
    },
    {
      "epoch": 1.5893763649132708,
      "grad_norm": 0.0963144302368164,
      "learning_rate": 1.230002258482635e-06,
      "loss": 0.001,
      "step": 247810
    },
    {
      "epoch": 1.589440501807057,
      "grad_norm": 0.009647555649280548,
      "learning_rate": 1.229634626985569e-06,
      "loss": 0.0012,
      "step": 247820
    },
    {
      "epoch": 1.5895046387008431,
      "grad_norm": 0.09639555215835571,
      "learning_rate": 1.2292670427338777e-06,
      "loss": 0.0017,
      "step": 247830
    },
    {
      "epoch": 1.5895687755946293,
      "grad_norm": 0.04794756695628166,
      "learning_rate": 1.2288995057321645e-06,
      "loss": 0.0012,
      "step": 247840
    },
    {
      "epoch": 1.5896329124884154,
      "grad_norm": 0.10684481263160706,
      "learning_rate": 1.2285320159850362e-06,
      "loss": 0.0025,
      "step": 247850
    },
    {
      "epoch": 1.5896970493822014,
      "grad_norm": 0.03137262538075447,
      "learning_rate": 1.2281645734970953e-06,
      "loss": 0.0009,
      "step": 247860
    },
    {
      "epoch": 1.5897611862759875,
      "grad_norm": 0.036332130432128906,
      "learning_rate": 1.2277971782729503e-06,
      "loss": 0.0007,
      "step": 247870
    },
    {
      "epoch": 1.5898253231697734,
      "grad_norm": 0.0729440450668335,
      "learning_rate": 1.2274298303172017e-06,
      "loss": 0.0021,
      "step": 247880
    },
    {
      "epoch": 1.5898894600635596,
      "grad_norm": 0.03348057344555855,
      "learning_rate": 1.2270625296344541e-06,
      "loss": 0.0017,
      "step": 247890
    },
    {
      "epoch": 1.5899535969573457,
      "grad_norm": 0.00622418150305748,
      "learning_rate": 1.2266952762293078e-06,
      "loss": 0.002,
      "step": 247900
    },
    {
      "epoch": 1.5900177338511319,
      "grad_norm": 0.07951758056879044,
      "learning_rate": 1.2263280701063678e-06,
      "loss": 0.0008,
      "step": 247910
    },
    {
      "epoch": 1.590081870744918,
      "grad_norm": 0.06803692132234573,
      "learning_rate": 1.2259609112702342e-06,
      "loss": 0.0011,
      "step": 247920
    },
    {
      "epoch": 1.5901460076387042,
      "grad_norm": 0.11719736456871033,
      "learning_rate": 1.2255937997255064e-06,
      "loss": 0.0011,
      "step": 247930
    },
    {
      "epoch": 1.59021014453249,
      "grad_norm": 0.07720184326171875,
      "learning_rate": 1.2252267354767866e-06,
      "loss": 0.0018,
      "step": 247940
    },
    {
      "epoch": 1.5902742814262762,
      "grad_norm": 0.10507367551326752,
      "learning_rate": 1.2248597185286742e-06,
      "loss": 0.0012,
      "step": 247950
    },
    {
      "epoch": 1.5903384183200622,
      "grad_norm": 0.08309304714202881,
      "learning_rate": 1.2244927488857678e-06,
      "loss": 0.0006,
      "step": 247960
    },
    {
      "epoch": 1.5904025552138483,
      "grad_norm": 0.020326273515820503,
      "learning_rate": 1.2241258265526635e-06,
      "loss": 0.0008,
      "step": 247970
    },
    {
      "epoch": 1.5904666921076345,
      "grad_norm": 0.02808596007525921,
      "learning_rate": 1.2237589515339627e-06,
      "loss": 0.0006,
      "step": 247980
    },
    {
      "epoch": 1.5905308290014206,
      "grad_norm": 0.08803045749664307,
      "learning_rate": 1.223392123834261e-06,
      "loss": 0.001,
      "step": 247990
    },
    {
      "epoch": 1.5905949658952068,
      "grad_norm": 0.023076031357049942,
      "learning_rate": 1.2230253434581558e-06,
      "loss": 0.0012,
      "step": 248000
    },
    {
      "epoch": 1.590659102788993,
      "grad_norm": 0.08916833996772766,
      "learning_rate": 1.2226586104102407e-06,
      "loss": 0.0014,
      "step": 248010
    },
    {
      "epoch": 1.590723239682779,
      "grad_norm": 0.004460031166672707,
      "learning_rate": 1.2222919246951136e-06,
      "loss": 0.0011,
      "step": 248020
    },
    {
      "epoch": 1.590787376576565,
      "grad_norm": 0.11745785176753998,
      "learning_rate": 1.2219252863173692e-06,
      "loss": 0.0012,
      "step": 248030
    },
    {
      "epoch": 1.5908515134703511,
      "grad_norm": 0.04275527969002724,
      "learning_rate": 1.2215586952815994e-06,
      "loss": 0.0014,
      "step": 248040
    },
    {
      "epoch": 1.590915650364137,
      "grad_norm": 0.10326236486434937,
      "learning_rate": 1.2211921515924014e-06,
      "loss": 0.0007,
      "step": 248050
    },
    {
      "epoch": 1.5909797872579232,
      "grad_norm": 0.015973402187228203,
      "learning_rate": 1.2208256552543657e-06,
      "loss": 0.001,
      "step": 248060
    },
    {
      "epoch": 1.5910439241517094,
      "grad_norm": 0.03947042301297188,
      "learning_rate": 1.220459206272086e-06,
      "loss": 0.0051,
      "step": 248070
    },
    {
      "epoch": 1.5911080610454955,
      "grad_norm": 0.2029953896999359,
      "learning_rate": 1.220092804650152e-06,
      "loss": 0.0009,
      "step": 248080
    },
    {
      "epoch": 1.5911721979392817,
      "grad_norm": 0.04274044185876846,
      "learning_rate": 1.2197264503931584e-06,
      "loss": 0.0006,
      "step": 248090
    },
    {
      "epoch": 1.5912363348330678,
      "grad_norm": 0.13463370501995087,
      "learning_rate": 1.219360143505694e-06,
      "loss": 0.0018,
      "step": 248100
    },
    {
      "epoch": 1.591300471726854,
      "grad_norm": 0.04177447780966759,
      "learning_rate": 1.2189938839923487e-06,
      "loss": 0.0016,
      "step": 248110
    },
    {
      "epoch": 1.5913646086206399,
      "grad_norm": 0.04472409188747406,
      "learning_rate": 1.218627671857711e-06,
      "loss": 0.0017,
      "step": 248120
    },
    {
      "epoch": 1.591428745514426,
      "grad_norm": 0.04733484610915184,
      "learning_rate": 1.2182615071063724e-06,
      "loss": 0.0016,
      "step": 248130
    },
    {
      "epoch": 1.591492882408212,
      "grad_norm": 0.012340148910880089,
      "learning_rate": 1.2178953897429202e-06,
      "loss": 0.0008,
      "step": 248140
    },
    {
      "epoch": 1.591557019301998,
      "grad_norm": 0.1512427031993866,
      "learning_rate": 1.2175293197719413e-06,
      "loss": 0.0008,
      "step": 248150
    },
    {
      "epoch": 1.5916211561957843,
      "grad_norm": 0.1732378602027893,
      "learning_rate": 1.2171632971980225e-06,
      "loss": 0.0014,
      "step": 248160
    },
    {
      "epoch": 1.5916852930895704,
      "grad_norm": 0.020908314734697342,
      "learning_rate": 1.2167973220257517e-06,
      "loss": 0.0011,
      "step": 248170
    },
    {
      "epoch": 1.5917494299833566,
      "grad_norm": 0.0377255454659462,
      "learning_rate": 1.216431394259715e-06,
      "loss": 0.0017,
      "step": 248180
    },
    {
      "epoch": 1.5918135668771427,
      "grad_norm": 0.06305290758609772,
      "learning_rate": 1.2160655139044953e-06,
      "loss": 0.0008,
      "step": 248190
    },
    {
      "epoch": 1.5918777037709286,
      "grad_norm": 0.16171769797801971,
      "learning_rate": 1.2156996809646792e-06,
      "loss": 0.0028,
      "step": 248200
    },
    {
      "epoch": 1.5919418406647148,
      "grad_norm": 0.010947500355541706,
      "learning_rate": 1.2153338954448518e-06,
      "loss": 0.0006,
      "step": 248210
    },
    {
      "epoch": 1.5920059775585007,
      "grad_norm": 0.007494628429412842,
      "learning_rate": 1.214968157349596e-06,
      "loss": 0.0013,
      "step": 248220
    },
    {
      "epoch": 1.5920701144522869,
      "grad_norm": 0.026027843356132507,
      "learning_rate": 1.2146024666834944e-06,
      "loss": 0.0013,
      "step": 248230
    },
    {
      "epoch": 1.592134251346073,
      "grad_norm": 0.06451816856861115,
      "learning_rate": 1.214236823451127e-06,
      "loss": 0.0015,
      "step": 248240
    },
    {
      "epoch": 1.5921983882398592,
      "grad_norm": 0.10120180249214172,
      "learning_rate": 1.2138712276570802e-06,
      "loss": 0.0007,
      "step": 248250
    },
    {
      "epoch": 1.5922625251336453,
      "grad_norm": 0.1435142308473587,
      "learning_rate": 1.2135056793059325e-06,
      "loss": 0.0008,
      "step": 248260
    },
    {
      "epoch": 1.5923266620274314,
      "grad_norm": 0.04157077521085739,
      "learning_rate": 1.2131401784022646e-06,
      "loss": 0.001,
      "step": 248270
    },
    {
      "epoch": 1.5923907989212176,
      "grad_norm": 0.06308285892009735,
      "learning_rate": 1.2127747249506554e-06,
      "loss": 0.0009,
      "step": 248280
    },
    {
      "epoch": 1.5924549358150035,
      "grad_norm": 0.05132739245891571,
      "learning_rate": 1.212409318955687e-06,
      "loss": 0.001,
      "step": 248290
    },
    {
      "epoch": 1.5925190727087897,
      "grad_norm": 0.03832164406776428,
      "learning_rate": 1.212043960421937e-06,
      "loss": 0.0007,
      "step": 248300
    },
    {
      "epoch": 1.5925832096025756,
      "grad_norm": 0.019853752106428146,
      "learning_rate": 1.2116786493539818e-06,
      "loss": 0.0008,
      "step": 248310
    },
    {
      "epoch": 1.5926473464963617,
      "grad_norm": 0.02733612060546875,
      "learning_rate": 1.2113133857564018e-06,
      "loss": 0.0014,
      "step": 248320
    },
    {
      "epoch": 1.592711483390148,
      "grad_norm": 0.007777262479066849,
      "learning_rate": 1.2109481696337732e-06,
      "loss": 0.0013,
      "step": 248330
    },
    {
      "epoch": 1.592775620283934,
      "grad_norm": 0.0869772881269455,
      "learning_rate": 1.2105830009906716e-06,
      "loss": 0.0012,
      "step": 248340
    },
    {
      "epoch": 1.5928397571777202,
      "grad_norm": 0.042146243155002594,
      "learning_rate": 1.2102178798316722e-06,
      "loss": 0.0013,
      "step": 248350
    },
    {
      "epoch": 1.5929038940715063,
      "grad_norm": 0.06737937778234482,
      "learning_rate": 1.2098528061613523e-06,
      "loss": 0.0014,
      "step": 248360
    },
    {
      "epoch": 1.5929680309652923,
      "grad_norm": 0.18272994458675385,
      "learning_rate": 1.2094877799842858e-06,
      "loss": 0.0014,
      "step": 248370
    },
    {
      "epoch": 1.5930321678590784,
      "grad_norm": 0.03615164756774902,
      "learning_rate": 1.2091228013050466e-06,
      "loss": 0.0006,
      "step": 248380
    },
    {
      "epoch": 1.5930963047528646,
      "grad_norm": 0.027232449501752853,
      "learning_rate": 1.2087578701282065e-06,
      "loss": 0.0017,
      "step": 248390
    },
    {
      "epoch": 1.5931604416466505,
      "grad_norm": 0.05842263624072075,
      "learning_rate": 1.208392986458341e-06,
      "loss": 0.0015,
      "step": 248400
    },
    {
      "epoch": 1.5932245785404366,
      "grad_norm": 0.05320729687809944,
      "learning_rate": 1.2080281503000214e-06,
      "loss": 0.0017,
      "step": 248410
    },
    {
      "epoch": 1.5932887154342228,
      "grad_norm": 0.03330248221755028,
      "learning_rate": 1.2076633616578192e-06,
      "loss": 0.0008,
      "step": 248420
    },
    {
      "epoch": 1.593352852328009,
      "grad_norm": 0.09903709590435028,
      "learning_rate": 1.2072986205363041e-06,
      "loss": 0.0013,
      "step": 248430
    },
    {
      "epoch": 1.593416989221795,
      "grad_norm": 0.1738354116678238,
      "learning_rate": 1.2069339269400493e-06,
      "loss": 0.0012,
      "step": 248440
    },
    {
      "epoch": 1.5934811261155812,
      "grad_norm": 0.04957576468586922,
      "learning_rate": 1.206569280873623e-06,
      "loss": 0.0008,
      "step": 248450
    },
    {
      "epoch": 1.5935452630093672,
      "grad_norm": 0.03456306830048561,
      "learning_rate": 1.2062046823415936e-06,
      "loss": 0.0009,
      "step": 248460
    },
    {
      "epoch": 1.5936093999031533,
      "grad_norm": 0.037498582154512405,
      "learning_rate": 1.2058401313485318e-06,
      "loss": 0.0006,
      "step": 248470
    },
    {
      "epoch": 1.5936735367969392,
      "grad_norm": 0.027157435193657875,
      "learning_rate": 1.2054756278990053e-06,
      "loss": 0.0012,
      "step": 248480
    },
    {
      "epoch": 1.5937376736907254,
      "grad_norm": 0.11722490191459656,
      "learning_rate": 1.205111171997581e-06,
      "loss": 0.0013,
      "step": 248490
    },
    {
      "epoch": 1.5938018105845115,
      "grad_norm": 0.04586930572986603,
      "learning_rate": 1.204746763648824e-06,
      "loss": 0.0012,
      "step": 248500
    },
    {
      "epoch": 1.5938659474782977,
      "grad_norm": 0.0408138781785965,
      "learning_rate": 1.2043824028573049e-06,
      "loss": 0.0011,
      "step": 248510
    },
    {
      "epoch": 1.5939300843720838,
      "grad_norm": 0.2502545714378357,
      "learning_rate": 1.204018089627586e-06,
      "loss": 0.0014,
      "step": 248520
    },
    {
      "epoch": 1.59399422126587,
      "grad_norm": 0.0775919258594513,
      "learning_rate": 1.2036538239642336e-06,
      "loss": 0.001,
      "step": 248530
    },
    {
      "epoch": 1.5940583581596561,
      "grad_norm": 0.07798996567726135,
      "learning_rate": 1.2032896058718109e-06,
      "loss": 0.0012,
      "step": 248540
    },
    {
      "epoch": 1.594122495053442,
      "grad_norm": 0.08263470977544785,
      "learning_rate": 1.2029254353548841e-06,
      "loss": 0.0008,
      "step": 248550
    },
    {
      "epoch": 1.5941866319472282,
      "grad_norm": 0.11589933186769485,
      "learning_rate": 1.2025613124180158e-06,
      "loss": 0.0009,
      "step": 248560
    },
    {
      "epoch": 1.5942507688410141,
      "grad_norm": 0.07335735857486725,
      "learning_rate": 1.2021972370657676e-06,
      "loss": 0.001,
      "step": 248570
    },
    {
      "epoch": 1.5943149057348003,
      "grad_norm": 0.07589791715145111,
      "learning_rate": 1.2018332093027014e-06,
      "loss": 0.0004,
      "step": 248580
    },
    {
      "epoch": 1.5943790426285864,
      "grad_norm": 0.06774845719337463,
      "learning_rate": 1.20146922913338e-06,
      "loss": 0.0007,
      "step": 248590
    },
    {
      "epoch": 1.5944431795223726,
      "grad_norm": 0.09183141589164734,
      "learning_rate": 1.2011052965623648e-06,
      "loss": 0.0011,
      "step": 248600
    },
    {
      "epoch": 1.5945073164161587,
      "grad_norm": 0.1584300845861435,
      "learning_rate": 1.200741411594214e-06,
      "loss": 0.001,
      "step": 248610
    },
    {
      "epoch": 1.5945714533099449,
      "grad_norm": 0.05256221443414688,
      "learning_rate": 1.2003775742334894e-06,
      "loss": 0.0008,
      "step": 248620
    },
    {
      "epoch": 1.5946355902037308,
      "grad_norm": 0.04152030125260353,
      "learning_rate": 1.2000137844847497e-06,
      "loss": 0.0011,
      "step": 248630
    },
    {
      "epoch": 1.594699727097517,
      "grad_norm": 0.06338044255971909,
      "learning_rate": 1.199650042352553e-06,
      "loss": 0.0009,
      "step": 248640
    },
    {
      "epoch": 1.5947638639913029,
      "grad_norm": 0.05661037191748619,
      "learning_rate": 1.1992863478414563e-06,
      "loss": 0.0006,
      "step": 248650
    },
    {
      "epoch": 1.594828000885089,
      "grad_norm": 0.11422684043645859,
      "learning_rate": 1.198922700956019e-06,
      "loss": 0.0017,
      "step": 248660
    },
    {
      "epoch": 1.5948921377788752,
      "grad_norm": 0.06677471846342087,
      "learning_rate": 1.198559101700797e-06,
      "loss": 0.001,
      "step": 248670
    },
    {
      "epoch": 1.5949562746726613,
      "grad_norm": 0.22182701528072357,
      "learning_rate": 1.1981955500803461e-06,
      "loss": 0.0013,
      "step": 248680
    },
    {
      "epoch": 1.5950204115664475,
      "grad_norm": 0.03416885808110237,
      "learning_rate": 1.1978320460992204e-06,
      "loss": 0.001,
      "step": 248690
    },
    {
      "epoch": 1.5950845484602336,
      "grad_norm": 0.16863662004470825,
      "learning_rate": 1.1974685897619786e-06,
      "loss": 0.0017,
      "step": 248700
    },
    {
      "epoch": 1.5951486853540198,
      "grad_norm": 0.2889634966850281,
      "learning_rate": 1.1971051810731726e-06,
      "loss": 0.0012,
      "step": 248710
    },
    {
      "epoch": 1.5952128222478057,
      "grad_norm": 0.09325895458459854,
      "learning_rate": 1.1967418200373565e-06,
      "loss": 0.0032,
      "step": 248720
    },
    {
      "epoch": 1.5952769591415918,
      "grad_norm": 0.15792018175125122,
      "learning_rate": 1.1963785066590827e-06,
      "loss": 0.0009,
      "step": 248730
    },
    {
      "epoch": 1.5953410960353778,
      "grad_norm": 0.06073154881596565,
      "learning_rate": 1.1960152409429055e-06,
      "loss": 0.0013,
      "step": 248740
    },
    {
      "epoch": 1.595405232929164,
      "grad_norm": 0.0029850320424884558,
      "learning_rate": 1.195652022893376e-06,
      "loss": 0.0006,
      "step": 248750
    },
    {
      "epoch": 1.59546936982295,
      "grad_norm": 0.06377291679382324,
      "learning_rate": 1.1952888525150441e-06,
      "loss": 0.0011,
      "step": 248760
    },
    {
      "epoch": 1.5955335067167362,
      "grad_norm": 0.05814136937260628,
      "learning_rate": 1.1949257298124639e-06,
      "loss": 0.0013,
      "step": 248770
    },
    {
      "epoch": 1.5955976436105224,
      "grad_norm": 0.14128145575523376,
      "learning_rate": 1.1945626547901835e-06,
      "loss": 0.0008,
      "step": 248780
    },
    {
      "epoch": 1.5956617805043085,
      "grad_norm": 0.12270446121692657,
      "learning_rate": 1.1941996274527528e-06,
      "loss": 0.0021,
      "step": 248790
    },
    {
      "epoch": 1.5957259173980944,
      "grad_norm": 0.02536684088408947,
      "learning_rate": 1.1938366478047192e-06,
      "loss": 0.0012,
      "step": 248800
    },
    {
      "epoch": 1.5957900542918806,
      "grad_norm": 0.12321645766496658,
      "learning_rate": 1.1934737158506342e-06,
      "loss": 0.0014,
      "step": 248810
    },
    {
      "epoch": 1.5958541911856667,
      "grad_norm": 0.046535737812519073,
      "learning_rate": 1.193110831595044e-06,
      "loss": 0.001,
      "step": 248820
    },
    {
      "epoch": 1.5959183280794527,
      "grad_norm": 0.07171449065208435,
      "learning_rate": 1.1927479950424958e-06,
      "loss": 0.0009,
      "step": 248830
    },
    {
      "epoch": 1.5959824649732388,
      "grad_norm": 0.07412496209144592,
      "learning_rate": 1.192385206197535e-06,
      "loss": 0.0009,
      "step": 248840
    },
    {
      "epoch": 1.596046601867025,
      "grad_norm": 0.026464832946658134,
      "learning_rate": 1.19202246506471e-06,
      "loss": 0.0008,
      "step": 248850
    },
    {
      "epoch": 1.596110738760811,
      "grad_norm": 0.1299687623977661,
      "learning_rate": 1.1916597716485651e-06,
      "loss": 0.0007,
      "step": 248860
    },
    {
      "epoch": 1.5961748756545973,
      "grad_norm": 0.009185717441141605,
      "learning_rate": 1.191297125953645e-06,
      "loss": 0.0013,
      "step": 248870
    },
    {
      "epoch": 1.5962390125483834,
      "grad_norm": 0.060487210750579834,
      "learning_rate": 1.1909345279844925e-06,
      "loss": 0.0022,
      "step": 248880
    },
    {
      "epoch": 1.5963031494421693,
      "grad_norm": 0.04519510641694069,
      "learning_rate": 1.1905719777456537e-06,
      "loss": 0.0004,
      "step": 248890
    },
    {
      "epoch": 1.5963672863359555,
      "grad_norm": 0.03752319887280464,
      "learning_rate": 1.190209475241671e-06,
      "loss": 0.0019,
      "step": 248900
    },
    {
      "epoch": 1.5964314232297414,
      "grad_norm": 0.01071302779018879,
      "learning_rate": 1.189847020477085e-06,
      "loss": 0.0011,
      "step": 248910
    },
    {
      "epoch": 1.5964955601235276,
      "grad_norm": 0.10751044005155563,
      "learning_rate": 1.1894846134564402e-06,
      "loss": 0.0011,
      "step": 248920
    },
    {
      "epoch": 1.5965596970173137,
      "grad_norm": 0.0483706034719944,
      "learning_rate": 1.1891222541842767e-06,
      "loss": 0.0009,
      "step": 248930
    },
    {
      "epoch": 1.5966238339110999,
      "grad_norm": 0.1471995711326599,
      "learning_rate": 1.1887599426651353e-06,
      "loss": 0.0025,
      "step": 248940
    },
    {
      "epoch": 1.596687970804886,
      "grad_norm": 0.10588864982128143,
      "learning_rate": 1.1883976789035533e-06,
      "loss": 0.0008,
      "step": 248950
    },
    {
      "epoch": 1.5967521076986722,
      "grad_norm": 0.06941290944814682,
      "learning_rate": 1.188035462904073e-06,
      "loss": 0.0005,
      "step": 248960
    },
    {
      "epoch": 1.5968162445924583,
      "grad_norm": 0.006535564083606005,
      "learning_rate": 1.1876732946712344e-06,
      "loss": 0.0016,
      "step": 248970
    },
    {
      "epoch": 1.5968803814862442,
      "grad_norm": 0.07977228611707687,
      "learning_rate": 1.1873111742095739e-06,
      "loss": 0.0009,
      "step": 248980
    },
    {
      "epoch": 1.5969445183800304,
      "grad_norm": 0.0827580988407135,
      "learning_rate": 1.186949101523629e-06,
      "loss": 0.0017,
      "step": 248990
    },
    {
      "epoch": 1.5970086552738163,
      "grad_norm": 0.08859512954950333,
      "learning_rate": 1.186587076617936e-06,
      "loss": 0.0007,
      "step": 249000
    },
    {
      "epoch": 1.5970727921676025,
      "grad_norm": 0.0018137848237529397,
      "learning_rate": 1.186225099497033e-06,
      "loss": 0.0011,
      "step": 249010
    },
    {
      "epoch": 1.5971369290613886,
      "grad_norm": 0.2125101238489151,
      "learning_rate": 1.185863170165456e-06,
      "loss": 0.002,
      "step": 249020
    },
    {
      "epoch": 1.5972010659551747,
      "grad_norm": 0.008683650754392147,
      "learning_rate": 1.1855012886277377e-06,
      "loss": 0.0007,
      "step": 249030
    },
    {
      "epoch": 1.597265202848961,
      "grad_norm": 0.153066024184227,
      "learning_rate": 1.185139454888416e-06,
      "loss": 0.0016,
      "step": 249040
    },
    {
      "epoch": 1.597329339742747,
      "grad_norm": 0.062498725950717926,
      "learning_rate": 1.1847776689520228e-06,
      "loss": 0.0012,
      "step": 249050
    },
    {
      "epoch": 1.597393476636533,
      "grad_norm": 0.10802311450242996,
      "learning_rate": 1.1844159308230924e-06,
      "loss": 0.001,
      "step": 249060
    },
    {
      "epoch": 1.5974576135303191,
      "grad_norm": 0.457075834274292,
      "learning_rate": 1.1840542405061562e-06,
      "loss": 0.0018,
      "step": 249070
    },
    {
      "epoch": 1.597521750424105,
      "grad_norm": 0.06506223976612091,
      "learning_rate": 1.183692598005749e-06,
      "loss": 0.0004,
      "step": 249080
    },
    {
      "epoch": 1.5975858873178912,
      "grad_norm": 0.15700525045394897,
      "learning_rate": 1.1833310033264006e-06,
      "loss": 0.0011,
      "step": 249090
    },
    {
      "epoch": 1.5976500242116773,
      "grad_norm": 0.024480275809764862,
      "learning_rate": 1.182969456472643e-06,
      "loss": 0.001,
      "step": 249100
    },
    {
      "epoch": 1.5977141611054635,
      "grad_norm": 0.48664844036102295,
      "learning_rate": 1.182607957449004e-06,
      "loss": 0.0009,
      "step": 249110
    },
    {
      "epoch": 1.5977782979992496,
      "grad_norm": 0.0950402095913887,
      "learning_rate": 1.1822465062600175e-06,
      "loss": 0.0016,
      "step": 249120
    },
    {
      "epoch": 1.5978424348930358,
      "grad_norm": 0.004985040053725243,
      "learning_rate": 1.1818851029102108e-06,
      "loss": 0.0011,
      "step": 249130
    },
    {
      "epoch": 1.597906571786822,
      "grad_norm": 0.0396341010928154,
      "learning_rate": 1.181523747404112e-06,
      "loss": 0.0019,
      "step": 249140
    },
    {
      "epoch": 1.5979707086806079,
      "grad_norm": 0.08913438767194748,
      "learning_rate": 1.181162439746249e-06,
      "loss": 0.0009,
      "step": 249150
    },
    {
      "epoch": 1.598034845574394,
      "grad_norm": 0.055692918598651886,
      "learning_rate": 1.1808011799411507e-06,
      "loss": 0.0009,
      "step": 249160
    },
    {
      "epoch": 1.59809898246818,
      "grad_norm": 0.07682760059833527,
      "learning_rate": 1.180439967993343e-06,
      "loss": 0.0007,
      "step": 249170
    },
    {
      "epoch": 1.598163119361966,
      "grad_norm": 0.17688927054405212,
      "learning_rate": 1.1800788039073519e-06,
      "loss": 0.0017,
      "step": 249180
    },
    {
      "epoch": 1.5982272562557522,
      "grad_norm": 0.12076979875564575,
      "learning_rate": 1.1797176876877037e-06,
      "loss": 0.0009,
      "step": 249190
    },
    {
      "epoch": 1.5982913931495384,
      "grad_norm": 0.030076563358306885,
      "learning_rate": 1.1793566193389239e-06,
      "loss": 0.0011,
      "step": 249200
    },
    {
      "epoch": 1.5983555300433245,
      "grad_norm": 0.17592324316501617,
      "learning_rate": 1.1789955988655361e-06,
      "loss": 0.0018,
      "step": 249210
    },
    {
      "epoch": 1.5984196669371107,
      "grad_norm": 0.16255876421928406,
      "learning_rate": 1.1786346262720628e-06,
      "loss": 0.0018,
      "step": 249220
    },
    {
      "epoch": 1.5984838038308968,
      "grad_norm": 0.04533237963914871,
      "learning_rate": 1.1782737015630302e-06,
      "loss": 0.0012,
      "step": 249230
    },
    {
      "epoch": 1.5985479407246828,
      "grad_norm": 0.0027576833963394165,
      "learning_rate": 1.1779128247429594e-06,
      "loss": 0.0012,
      "step": 249240
    },
    {
      "epoch": 1.598612077618469,
      "grad_norm": 0.0029595845844596624,
      "learning_rate": 1.1775519958163723e-06,
      "loss": 0.0017,
      "step": 249250
    },
    {
      "epoch": 1.5986762145122548,
      "grad_norm": 0.07872213423252106,
      "learning_rate": 1.1771912147877896e-06,
      "loss": 0.0012,
      "step": 249260
    },
    {
      "epoch": 1.598740351406041,
      "grad_norm": 0.025151098147034645,
      "learning_rate": 1.1768304816617344e-06,
      "loss": 0.0011,
      "step": 249270
    },
    {
      "epoch": 1.5988044882998271,
      "grad_norm": 0.041764579713344574,
      "learning_rate": 1.1764697964427253e-06,
      "loss": 0.0012,
      "step": 249280
    },
    {
      "epoch": 1.5988686251936133,
      "grad_norm": 0.039692021906375885,
      "learning_rate": 1.1761091591352825e-06,
      "loss": 0.0014,
      "step": 249290
    },
    {
      "epoch": 1.5989327620873994,
      "grad_norm": 0.09026049077510834,
      "learning_rate": 1.1757485697439235e-06,
      "loss": 0.0009,
      "step": 249300
    },
    {
      "epoch": 1.5989968989811856,
      "grad_norm": 0.026388583704829216,
      "learning_rate": 1.1753880282731694e-06,
      "loss": 0.0022,
      "step": 249310
    },
    {
      "epoch": 1.5990610358749715,
      "grad_norm": 0.044500820338726044,
      "learning_rate": 1.175027534727537e-06,
      "loss": 0.0006,
      "step": 249320
    },
    {
      "epoch": 1.5991251727687577,
      "grad_norm": 0.08768253773450851,
      "learning_rate": 1.1746670891115414e-06,
      "loss": 0.0019,
      "step": 249330
    },
    {
      "epoch": 1.5991893096625436,
      "grad_norm": 0.08217465132474899,
      "learning_rate": 1.1743066914297025e-06,
      "loss": 0.0008,
      "step": 249340
    },
    {
      "epoch": 1.5992534465563297,
      "grad_norm": 0.17016156017780304,
      "learning_rate": 1.1739463416865348e-06,
      "loss": 0.0016,
      "step": 249350
    },
    {
      "epoch": 1.5993175834501159,
      "grad_norm": 0.0020344434306025505,
      "learning_rate": 1.1735860398865546e-06,
      "loss": 0.0012,
      "step": 249360
    },
    {
      "epoch": 1.599381720343902,
      "grad_norm": 0.02283032238483429,
      "learning_rate": 1.1732257860342743e-06,
      "loss": 0.0008,
      "step": 249370
    },
    {
      "epoch": 1.5994458572376882,
      "grad_norm": 0.024351265281438828,
      "learning_rate": 1.172865580134211e-06,
      "loss": 0.0013,
      "step": 249380
    },
    {
      "epoch": 1.5995099941314743,
      "grad_norm": 0.22470252215862274,
      "learning_rate": 1.1725054221908772e-06,
      "loss": 0.0017,
      "step": 249390
    },
    {
      "epoch": 1.5995741310252605,
      "grad_norm": 0.027585037052631378,
      "learning_rate": 1.1721453122087862e-06,
      "loss": 0.001,
      "step": 249400
    },
    {
      "epoch": 1.5996382679190464,
      "grad_norm": 0.09092612564563751,
      "learning_rate": 1.1717852501924487e-06,
      "loss": 0.0007,
      "step": 249410
    },
    {
      "epoch": 1.5997024048128325,
      "grad_norm": 0.04048805311322212,
      "learning_rate": 1.1714252361463791e-06,
      "loss": 0.0005,
      "step": 249420
    },
    {
      "epoch": 1.5997665417066185,
      "grad_norm": 0.10383453220129013,
      "learning_rate": 1.1710652700750879e-06,
      "loss": 0.0013,
      "step": 249430
    },
    {
      "epoch": 1.5998306786004046,
      "grad_norm": 0.03964724391698837,
      "learning_rate": 1.1707053519830852e-06,
      "loss": 0.001,
      "step": 249440
    },
    {
      "epoch": 1.5998948154941908,
      "grad_norm": 0.05186668410897255,
      "learning_rate": 1.17034548187488e-06,
      "loss": 0.0007,
      "step": 249450
    },
    {
      "epoch": 1.599958952387977,
      "grad_norm": 0.17879439890384674,
      "learning_rate": 1.1699856597549842e-06,
      "loss": 0.0006,
      "step": 249460
    },
    {
      "epoch": 1.600023089281763,
      "grad_norm": 0.03267485648393631,
      "learning_rate": 1.1696258856279053e-06,
      "loss": 0.0017,
      "step": 249470
    },
    {
      "epoch": 1.6000872261755492,
      "grad_norm": 0.0976705327630043,
      "learning_rate": 1.1692661594981502e-06,
      "loss": 0.0011,
      "step": 249480
    },
    {
      "epoch": 1.6001513630693351,
      "grad_norm": 0.03639264777302742,
      "learning_rate": 1.1689064813702295e-06,
      "loss": 0.0014,
      "step": 249490
    },
    {
      "epoch": 1.6002154999631213,
      "grad_norm": 0.007496272213757038,
      "learning_rate": 1.1685468512486481e-06,
      "loss": 0.0008,
      "step": 249500
    },
    {
      "epoch": 1.6002796368569072,
      "grad_norm": 0.041140880435705185,
      "learning_rate": 1.1681872691379132e-06,
      "loss": 0.0022,
      "step": 249510
    },
    {
      "epoch": 1.6003437737506934,
      "grad_norm": 0.16042396426200867,
      "learning_rate": 1.1678277350425293e-06,
      "loss": 0.0077,
      "step": 249520
    },
    {
      "epoch": 1.6004079106444795,
      "grad_norm": 0.0523374006152153,
      "learning_rate": 1.1674682489670036e-06,
      "loss": 0.0015,
      "step": 249530
    },
    {
      "epoch": 1.6004720475382657,
      "grad_norm": 0.07310651242733002,
      "learning_rate": 1.1671088109158402e-06,
      "loss": 0.0009,
      "step": 249540
    },
    {
      "epoch": 1.6005361844320518,
      "grad_norm": 0.04398871958255768,
      "learning_rate": 1.1667494208935426e-06,
      "loss": 0.0018,
      "step": 249550
    },
    {
      "epoch": 1.600600321325838,
      "grad_norm": 0.06598342210054398,
      "learning_rate": 1.166390078904613e-06,
      "loss": 0.0009,
      "step": 249560
    },
    {
      "epoch": 1.600664458219624,
      "grad_norm": 0.0393049456179142,
      "learning_rate": 1.1660307849535569e-06,
      "loss": 0.0006,
      "step": 249570
    },
    {
      "epoch": 1.60072859511341,
      "grad_norm": 0.10699399560689926,
      "learning_rate": 1.1656715390448746e-06,
      "loss": 0.0013,
      "step": 249580
    },
    {
      "epoch": 1.6007927320071962,
      "grad_norm": 0.04361341521143913,
      "learning_rate": 1.165312341183069e-06,
      "loss": 0.0014,
      "step": 249590
    },
    {
      "epoch": 1.600856868900982,
      "grad_norm": 0.0037963781505823135,
      "learning_rate": 1.1649531913726392e-06,
      "loss": 0.0013,
      "step": 249600
    },
    {
      "epoch": 1.6009210057947683,
      "grad_norm": 0.003534812480211258,
      "learning_rate": 1.1645940896180874e-06,
      "loss": 0.0007,
      "step": 249610
    },
    {
      "epoch": 1.6009851426885544,
      "grad_norm": 0.0005362582160159945,
      "learning_rate": 1.1642350359239136e-06,
      "loss": 0.0006,
      "step": 249620
    },
    {
      "epoch": 1.6010492795823406,
      "grad_norm": 0.11060915887355804,
      "learning_rate": 1.163876030294614e-06,
      "loss": 0.001,
      "step": 249630
    },
    {
      "epoch": 1.6011134164761267,
      "grad_norm": 0.03232883661985397,
      "learning_rate": 1.1635170727346912e-06,
      "loss": 0.0006,
      "step": 249640
    },
    {
      "epoch": 1.6011775533699129,
      "grad_norm": 0.040262069553136826,
      "learning_rate": 1.1631581632486411e-06,
      "loss": 0.0014,
      "step": 249650
    },
    {
      "epoch": 1.601241690263699,
      "grad_norm": 0.045741088688373566,
      "learning_rate": 1.1627993018409616e-06,
      "loss": 0.0009,
      "step": 249660
    },
    {
      "epoch": 1.601305827157485,
      "grad_norm": 0.03924969211220741,
      "learning_rate": 1.1624404885161478e-06,
      "loss": 0.0004,
      "step": 249670
    },
    {
      "epoch": 1.601369964051271,
      "grad_norm": 0.038819339126348495,
      "learning_rate": 1.1620817232786986e-06,
      "loss": 0.0003,
      "step": 249680
    },
    {
      "epoch": 1.601434100945057,
      "grad_norm": 0.046807948499917984,
      "learning_rate": 1.1617230061331085e-06,
      "loss": 0.0006,
      "step": 249690
    },
    {
      "epoch": 1.6014982378388432,
      "grad_norm": 0.014780609868466854,
      "learning_rate": 1.1613643370838723e-06,
      "loss": 0.0039,
      "step": 249700
    },
    {
      "epoch": 1.6015623747326293,
      "grad_norm": 0.0041016568429768085,
      "learning_rate": 1.1610057161354827e-06,
      "loss": 0.001,
      "step": 249710
    },
    {
      "epoch": 1.6016265116264154,
      "grad_norm": 0.09573370218276978,
      "learning_rate": 1.1606471432924353e-06,
      "loss": 0.0022,
      "step": 249720
    },
    {
      "epoch": 1.6016906485202016,
      "grad_norm": 0.016817208379507065,
      "learning_rate": 1.1602886185592248e-06,
      "loss": 0.0004,
      "step": 249730
    },
    {
      "epoch": 1.6017547854139877,
      "grad_norm": 0.048654478043317795,
      "learning_rate": 1.159930141940342e-06,
      "loss": 0.0015,
      "step": 249740
    },
    {
      "epoch": 1.6018189223077737,
      "grad_norm": 0.07242769747972488,
      "learning_rate": 1.1595717134402778e-06,
      "loss": 0.0029,
      "step": 249750
    },
    {
      "epoch": 1.6018830592015598,
      "grad_norm": 0.01238231360912323,
      "learning_rate": 1.1592133330635258e-06,
      "loss": 0.0008,
      "step": 249760
    },
    {
      "epoch": 1.6019471960953457,
      "grad_norm": 0.04801921918988228,
      "learning_rate": 1.1588550008145756e-06,
      "loss": 0.0008,
      "step": 249770
    },
    {
      "epoch": 1.602011332989132,
      "grad_norm": 0.10079651325941086,
      "learning_rate": 1.1584967166979178e-06,
      "loss": 0.0008,
      "step": 249780
    },
    {
      "epoch": 1.602075469882918,
      "grad_norm": 0.1167251393198967,
      "learning_rate": 1.1581384807180402e-06,
      "loss": 0.0022,
      "step": 249790
    },
    {
      "epoch": 1.6021396067767042,
      "grad_norm": 0.1415635645389557,
      "learning_rate": 1.1577802928794341e-06,
      "loss": 0.0008,
      "step": 249800
    },
    {
      "epoch": 1.6022037436704903,
      "grad_norm": 0.04449000209569931,
      "learning_rate": 1.1574221531865875e-06,
      "loss": 0.0008,
      "step": 249810
    },
    {
      "epoch": 1.6022678805642765,
      "grad_norm": 0.2547750174999237,
      "learning_rate": 1.1570640616439872e-06,
      "loss": 0.0031,
      "step": 249820
    },
    {
      "epoch": 1.6023320174580626,
      "grad_norm": 0.05963004752993584,
      "learning_rate": 1.1567060182561196e-06,
      "loss": 0.0005,
      "step": 249830
    },
    {
      "epoch": 1.6023961543518486,
      "grad_norm": 0.171538308262825,
      "learning_rate": 1.1563480230274738e-06,
      "loss": 0.0012,
      "step": 249840
    },
    {
      "epoch": 1.6024602912456347,
      "grad_norm": 0.17668305337429047,
      "learning_rate": 1.1559900759625336e-06,
      "loss": 0.0013,
      "step": 249850
    },
    {
      "epoch": 1.6025244281394206,
      "grad_norm": 0.01708364672958851,
      "learning_rate": 1.1556321770657858e-06,
      "loss": 0.002,
      "step": 249860
    },
    {
      "epoch": 1.6025885650332068,
      "grad_norm": 0.36142441630363464,
      "learning_rate": 1.1552743263417126e-06,
      "loss": 0.0008,
      "step": 249870
    },
    {
      "epoch": 1.602652701926993,
      "grad_norm": 0.048225026577711105,
      "learning_rate": 1.1549165237948011e-06,
      "loss": 0.0008,
      "step": 249880
    },
    {
      "epoch": 1.602716838820779,
      "grad_norm": 0.0728774294257164,
      "learning_rate": 1.154558769429534e-06,
      "loss": 0.0023,
      "step": 249890
    },
    {
      "epoch": 1.6027809757145652,
      "grad_norm": 0.08327235281467438,
      "learning_rate": 1.1542010632503925e-06,
      "loss": 0.0008,
      "step": 249900
    },
    {
      "epoch": 1.6028451126083514,
      "grad_norm": 0.18347333371639252,
      "learning_rate": 1.1538434052618614e-06,
      "loss": 0.0017,
      "step": 249910
    },
    {
      "epoch": 1.6029092495021373,
      "grad_norm": 0.008992303162813187,
      "learning_rate": 1.153485795468421e-06,
      "loss": 0.0007,
      "step": 249920
    },
    {
      "epoch": 1.6029733863959235,
      "grad_norm": 0.06550116837024689,
      "learning_rate": 1.153128233874553e-06,
      "loss": 0.0019,
      "step": 249930
    },
    {
      "epoch": 1.6030375232897096,
      "grad_norm": 0.07155051827430725,
      "learning_rate": 1.152770720484736e-06,
      "loss": 0.0008,
      "step": 249940
    },
    {
      "epoch": 1.6031016601834955,
      "grad_norm": 0.08370451629161835,
      "learning_rate": 1.1524132553034528e-06,
      "loss": 0.0008,
      "step": 249950
    },
    {
      "epoch": 1.6031657970772817,
      "grad_norm": 0.07363441586494446,
      "learning_rate": 1.1520558383351816e-06,
      "loss": 0.0009,
      "step": 249960
    },
    {
      "epoch": 1.6032299339710678,
      "grad_norm": 0.060655880719423294,
      "learning_rate": 1.1516984695844002e-06,
      "loss": 0.0003,
      "step": 249970
    },
    {
      "epoch": 1.603294070864854,
      "grad_norm": 0.02170303836464882,
      "learning_rate": 1.1513411490555865e-06,
      "loss": 0.0013,
      "step": 249980
    },
    {
      "epoch": 1.6033582077586401,
      "grad_norm": 0.2270852029323578,
      "learning_rate": 1.15098387675322e-06,
      "loss": 0.0008,
      "step": 249990
    },
    {
      "epoch": 1.6034223446524263,
      "grad_norm": 0.06730823218822479,
      "learning_rate": 1.1506266526817766e-06,
      "loss": 0.0011,
      "step": 250000
    },
    {
      "epoch": 1.6034864815462122,
      "grad_norm": 0.004074290860444307,
      "learning_rate": 1.1502694768457317e-06,
      "loss": 0.0011,
      "step": 250010
    },
    {
      "epoch": 1.6035506184399984,
      "grad_norm": 0.09978527575731277,
      "learning_rate": 1.1499123492495607e-06,
      "loss": 0.0003,
      "step": 250020
    },
    {
      "epoch": 1.6036147553337843,
      "grad_norm": 0.035276979207992554,
      "learning_rate": 1.1495552698977414e-06,
      "loss": 0.0007,
      "step": 250030
    },
    {
      "epoch": 1.6036788922275704,
      "grad_norm": 0.13160935044288635,
      "learning_rate": 1.149198238794746e-06,
      "loss": 0.0009,
      "step": 250040
    },
    {
      "epoch": 1.6037430291213566,
      "grad_norm": 0.1706625372171402,
      "learning_rate": 1.1488412559450473e-06,
      "loss": 0.0008,
      "step": 250050
    },
    {
      "epoch": 1.6038071660151427,
      "grad_norm": 0.10292987525463104,
      "learning_rate": 1.1484843213531216e-06,
      "loss": 0.0013,
      "step": 250060
    },
    {
      "epoch": 1.6038713029089289,
      "grad_norm": 0.13077609241008759,
      "learning_rate": 1.1481274350234395e-06,
      "loss": 0.0011,
      "step": 250070
    },
    {
      "epoch": 1.603935439802715,
      "grad_norm": 0.06850180774927139,
      "learning_rate": 1.147770596960474e-06,
      "loss": 0.0037,
      "step": 250080
    },
    {
      "epoch": 1.6039995766965012,
      "grad_norm": 0.04928059130907059,
      "learning_rate": 1.1474138071686947e-06,
      "loss": 0.0012,
      "step": 250090
    },
    {
      "epoch": 1.604063713590287,
      "grad_norm": 0.04577067866921425,
      "learning_rate": 1.1470570656525754e-06,
      "loss": 0.001,
      "step": 250100
    },
    {
      "epoch": 1.6041278504840732,
      "grad_norm": 0.14815518260002136,
      "learning_rate": 1.1467003724165842e-06,
      "loss": 0.0007,
      "step": 250110
    },
    {
      "epoch": 1.6041919873778592,
      "grad_norm": 0.06266386806964874,
      "learning_rate": 1.1463437274651918e-06,
      "loss": 0.001,
      "step": 250120
    },
    {
      "epoch": 1.6042561242716453,
      "grad_norm": 0.002447890816256404,
      "learning_rate": 1.1459871308028647e-06,
      "loss": 0.0011,
      "step": 250130
    },
    {
      "epoch": 1.6043202611654315,
      "grad_norm": 0.12594091892242432,
      "learning_rate": 1.1456305824340746e-06,
      "loss": 0.0009,
      "step": 250140
    },
    {
      "epoch": 1.6043843980592176,
      "grad_norm": 0.06531865894794464,
      "learning_rate": 1.1452740823632885e-06,
      "loss": 0.001,
      "step": 250150
    },
    {
      "epoch": 1.6044485349530038,
      "grad_norm": 0.07496868818998337,
      "learning_rate": 1.1449176305949717e-06,
      "loss": 0.001,
      "step": 250160
    },
    {
      "epoch": 1.60451267184679,
      "grad_norm": 0.08994642645120621,
      "learning_rate": 1.1445612271335933e-06,
      "loss": 0.0014,
      "step": 250170
    },
    {
      "epoch": 1.6045768087405758,
      "grad_norm": 0.17000466585159302,
      "learning_rate": 1.144204871983618e-06,
      "loss": 0.0008,
      "step": 250180
    },
    {
      "epoch": 1.604640945634362,
      "grad_norm": 0.08831164985895157,
      "learning_rate": 1.1438485651495117e-06,
      "loss": 0.0007,
      "step": 250190
    },
    {
      "epoch": 1.604705082528148,
      "grad_norm": 0.0376860573887825,
      "learning_rate": 1.1434923066357374e-06,
      "loss": 0.0006,
      "step": 250200
    },
    {
      "epoch": 1.604769219421934,
      "grad_norm": 0.047595515847206116,
      "learning_rate": 1.143136096446762e-06,
      "loss": 0.002,
      "step": 250210
    },
    {
      "epoch": 1.6048333563157202,
      "grad_norm": 0.1046251505613327,
      "learning_rate": 1.1427799345870478e-06,
      "loss": 0.001,
      "step": 250220
    },
    {
      "epoch": 1.6048974932095064,
      "grad_norm": 0.03919575735926628,
      "learning_rate": 1.1424238210610577e-06,
      "loss": 0.0009,
      "step": 250230
    },
    {
      "epoch": 1.6049616301032925,
      "grad_norm": 0.09882248193025589,
      "learning_rate": 1.1420677558732523e-06,
      "loss": 0.0017,
      "step": 250240
    },
    {
      "epoch": 1.6050257669970787,
      "grad_norm": 0.060251470655202866,
      "learning_rate": 1.1417117390280975e-06,
      "loss": 0.0008,
      "step": 250250
    },
    {
      "epoch": 1.6050899038908648,
      "grad_norm": 0.09244880825281143,
      "learning_rate": 1.141355770530051e-06,
      "loss": 0.0013,
      "step": 250260
    },
    {
      "epoch": 1.6051540407846507,
      "grad_norm": 0.005103874485939741,
      "learning_rate": 1.1409998503835751e-06,
      "loss": 0.0013,
      "step": 250270
    },
    {
      "epoch": 1.6052181776784369,
      "grad_norm": 0.06546703726053238,
      "learning_rate": 1.1406439785931272e-06,
      "loss": 0.0006,
      "step": 250280
    },
    {
      "epoch": 1.6052823145722228,
      "grad_norm": 0.008968470618128777,
      "learning_rate": 1.1402881551631706e-06,
      "loss": 0.0007,
      "step": 250290
    },
    {
      "epoch": 1.605346451466009,
      "grad_norm": 0.16478286683559418,
      "learning_rate": 1.1399323800981616e-06,
      "loss": 0.0028,
      "step": 250300
    },
    {
      "epoch": 1.605410588359795,
      "grad_norm": 0.10185612738132477,
      "learning_rate": 1.139576653402557e-06,
      "loss": 0.0006,
      "step": 250310
    },
    {
      "epoch": 1.6054747252535813,
      "grad_norm": 0.07352697849273682,
      "learning_rate": 1.1392209750808175e-06,
      "loss": 0.0006,
      "step": 250320
    },
    {
      "epoch": 1.6055388621473674,
      "grad_norm": 0.006930564530193806,
      "learning_rate": 1.1388653451373983e-06,
      "loss": 0.0007,
      "step": 250330
    },
    {
      "epoch": 1.6056029990411536,
      "grad_norm": 0.058381129056215286,
      "learning_rate": 1.138509763576756e-06,
      "loss": 0.0008,
      "step": 250340
    },
    {
      "epoch": 1.6056671359349395,
      "grad_norm": 0.01590009219944477,
      "learning_rate": 1.1381542304033443e-06,
      "loss": 0.0008,
      "step": 250350
    },
    {
      "epoch": 1.6057312728287256,
      "grad_norm": 0.07030168175697327,
      "learning_rate": 1.1377987456216222e-06,
      "loss": 0.0014,
      "step": 250360
    },
    {
      "epoch": 1.6057954097225118,
      "grad_norm": 0.0982590839266777,
      "learning_rate": 1.1374433092360416e-06,
      "loss": 0.0017,
      "step": 250370
    },
    {
      "epoch": 1.6058595466162977,
      "grad_norm": 0.17898105084896088,
      "learning_rate": 1.1370879212510572e-06,
      "loss": 0.0011,
      "step": 250380
    },
    {
      "epoch": 1.6059236835100839,
      "grad_norm": 0.38064634799957275,
      "learning_rate": 1.1367325816711205e-06,
      "loss": 0.0016,
      "step": 250390
    },
    {
      "epoch": 1.60598782040387,
      "grad_norm": 0.05069749429821968,
      "learning_rate": 1.1363772905006865e-06,
      "loss": 0.0008,
      "step": 250400
    },
    {
      "epoch": 1.6060519572976562,
      "grad_norm": 0.4793483316898346,
      "learning_rate": 1.1360220477442068e-06,
      "loss": 0.0012,
      "step": 250410
    },
    {
      "epoch": 1.6061160941914423,
      "grad_norm": 0.0342763252556324,
      "learning_rate": 1.135666853406132e-06,
      "loss": 0.0008,
      "step": 250420
    },
    {
      "epoch": 1.6061802310852284,
      "grad_norm": 0.10136108845472336,
      "learning_rate": 1.1353117074909125e-06,
      "loss": 0.0007,
      "step": 250430
    },
    {
      "epoch": 1.6062443679790144,
      "grad_norm": 0.0018044215394183993,
      "learning_rate": 1.1349566100030007e-06,
      "loss": 0.0012,
      "step": 250440
    },
    {
      "epoch": 1.6063085048728005,
      "grad_norm": 0.1991158425807953,
      "learning_rate": 1.1346015609468446e-06,
      "loss": 0.0011,
      "step": 250450
    },
    {
      "epoch": 1.6063726417665865,
      "grad_norm": 0.002864201320335269,
      "learning_rate": 1.134246560326892e-06,
      "loss": 0.0009,
      "step": 250460
    },
    {
      "epoch": 1.6064367786603726,
      "grad_norm": 0.022196555510163307,
      "learning_rate": 1.1338916081475943e-06,
      "loss": 0.0008,
      "step": 250470
    },
    {
      "epoch": 1.6065009155541587,
      "grad_norm": 0.1637929528951645,
      "learning_rate": 1.1335367044133965e-06,
      "loss": 0.001,
      "step": 250480
    },
    {
      "epoch": 1.606565052447945,
      "grad_norm": 0.20712658762931824,
      "learning_rate": 1.1331818491287483e-06,
      "loss": 0.0023,
      "step": 250490
    },
    {
      "epoch": 1.606629189341731,
      "grad_norm": 0.016757261008024216,
      "learning_rate": 1.1328270422980953e-06,
      "loss": 0.0011,
      "step": 250500
    },
    {
      "epoch": 1.6066933262355172,
      "grad_norm": 0.0070112538523972034,
      "learning_rate": 1.1324722839258823e-06,
      "loss": 0.001,
      "step": 250510
    },
    {
      "epoch": 1.6067574631293033,
      "grad_norm": 0.02961020916700363,
      "learning_rate": 1.1321175740165563e-06,
      "loss": 0.0007,
      "step": 250520
    },
    {
      "epoch": 1.6068216000230893,
      "grad_norm": 0.09048043936491013,
      "learning_rate": 1.1317629125745616e-06,
      "loss": 0.0008,
      "step": 250530
    },
    {
      "epoch": 1.6068857369168754,
      "grad_norm": 0.02654772624373436,
      "learning_rate": 1.1314082996043424e-06,
      "loss": 0.0007,
      "step": 250540
    },
    {
      "epoch": 1.6069498738106613,
      "grad_norm": 0.050176508724689484,
      "learning_rate": 1.1310537351103407e-06,
      "loss": 0.0005,
      "step": 250550
    },
    {
      "epoch": 1.6070140107044475,
      "grad_norm": 0.08712086081504822,
      "learning_rate": 1.1306992190970023e-06,
      "loss": 0.0013,
      "step": 250560
    },
    {
      "epoch": 1.6070781475982336,
      "grad_norm": 0.019818203523755074,
      "learning_rate": 1.1303447515687676e-06,
      "loss": 0.0009,
      "step": 250570
    },
    {
      "epoch": 1.6071422844920198,
      "grad_norm": 0.03647455945611,
      "learning_rate": 1.1299903325300775e-06,
      "loss": 0.0008,
      "step": 250580
    },
    {
      "epoch": 1.607206421385806,
      "grad_norm": 0.030462345108389854,
      "learning_rate": 1.129635961985376e-06,
      "loss": 0.0024,
      "step": 250590
    },
    {
      "epoch": 1.607270558279592,
      "grad_norm": 0.004395130090415478,
      "learning_rate": 1.1292816399391022e-06,
      "loss": 0.0016,
      "step": 250600
    },
    {
      "epoch": 1.607334695173378,
      "grad_norm": 0.02232508361339569,
      "learning_rate": 1.1289273663956951e-06,
      "loss": 0.0008,
      "step": 250610
    },
    {
      "epoch": 1.6073988320671642,
      "grad_norm": 0.018504155799746513,
      "learning_rate": 1.128573141359594e-06,
      "loss": 0.0009,
      "step": 250620
    },
    {
      "epoch": 1.60746296896095,
      "grad_norm": 0.08404509723186493,
      "learning_rate": 1.1282189648352393e-06,
      "loss": 0.0006,
      "step": 250630
    },
    {
      "epoch": 1.6075271058547362,
      "grad_norm": 0.04173661395907402,
      "learning_rate": 1.1278648368270683e-06,
      "loss": 0.0011,
      "step": 250640
    },
    {
      "epoch": 1.6075912427485224,
      "grad_norm": 0.064873106777668,
      "learning_rate": 1.1275107573395183e-06,
      "loss": 0.001,
      "step": 250650
    },
    {
      "epoch": 1.6076553796423085,
      "grad_norm": 0.0069096386432647705,
      "learning_rate": 1.1271567263770244e-06,
      "loss": 0.0006,
      "step": 250660
    },
    {
      "epoch": 1.6077195165360947,
      "grad_norm": 0.08207154273986816,
      "learning_rate": 1.126802743944026e-06,
      "loss": 0.0078,
      "step": 250670
    },
    {
      "epoch": 1.6077836534298808,
      "grad_norm": 0.09320911765098572,
      "learning_rate": 1.1264488100449577e-06,
      "loss": 0.0014,
      "step": 250680
    },
    {
      "epoch": 1.607847790323667,
      "grad_norm": 0.043957896530628204,
      "learning_rate": 1.1260949246842539e-06,
      "loss": 0.0011,
      "step": 250690
    },
    {
      "epoch": 1.607911927217453,
      "grad_norm": 0.055686213076114655,
      "learning_rate": 1.1257410878663482e-06,
      "loss": 0.0008,
      "step": 250700
    },
    {
      "epoch": 1.607976064111239,
      "grad_norm": 0.027599196881055832,
      "learning_rate": 1.1253872995956772e-06,
      "loss": 0.0005,
      "step": 250710
    },
    {
      "epoch": 1.608040201005025,
      "grad_norm": 0.03367011249065399,
      "learning_rate": 1.125033559876672e-06,
      "loss": 0.0008,
      "step": 250720
    },
    {
      "epoch": 1.6081043378988111,
      "grad_norm": 0.09349153935909271,
      "learning_rate": 1.1246798687137644e-06,
      "loss": 0.0017,
      "step": 250730
    },
    {
      "epoch": 1.6081684747925973,
      "grad_norm": 0.024527819827198982,
      "learning_rate": 1.124326226111389e-06,
      "loss": 0.0012,
      "step": 250740
    },
    {
      "epoch": 1.6082326116863834,
      "grad_norm": 0.032239992171525955,
      "learning_rate": 1.1239726320739758e-06,
      "loss": 0.0011,
      "step": 250750
    },
    {
      "epoch": 1.6082967485801696,
      "grad_norm": 0.11960357427597046,
      "learning_rate": 1.123619086605956e-06,
      "loss": 0.002,
      "step": 250760
    },
    {
      "epoch": 1.6083608854739557,
      "grad_norm": 0.016015082597732544,
      "learning_rate": 1.1232655897117579e-06,
      "loss": 0.0014,
      "step": 250770
    },
    {
      "epoch": 1.6084250223677417,
      "grad_norm": 0.06744474917650223,
      "learning_rate": 1.1229121413958144e-06,
      "loss": 0.0005,
      "step": 250780
    },
    {
      "epoch": 1.6084891592615278,
      "grad_norm": 0.035480957478284836,
      "learning_rate": 1.1225587416625521e-06,
      "loss": 0.0014,
      "step": 250790
    },
    {
      "epoch": 1.608553296155314,
      "grad_norm": 0.07199794054031372,
      "learning_rate": 1.1222053905163999e-06,
      "loss": 0.0013,
      "step": 250800
    },
    {
      "epoch": 1.6086174330490999,
      "grad_norm": 0.09330529719591141,
      "learning_rate": 1.1218520879617845e-06,
      "loss": 0.003,
      "step": 250810
    },
    {
      "epoch": 1.608681569942886,
      "grad_norm": 0.06037352606654167,
      "learning_rate": 1.1214988340031357e-06,
      "loss": 0.0007,
      "step": 250820
    },
    {
      "epoch": 1.6087457068366722,
      "grad_norm": 0.10916303098201752,
      "learning_rate": 1.121145628644878e-06,
      "loss": 0.0009,
      "step": 250830
    },
    {
      "epoch": 1.6088098437304583,
      "grad_norm": 0.05477078631520271,
      "learning_rate": 1.120792471891438e-06,
      "loss": 0.0017,
      "step": 250840
    },
    {
      "epoch": 1.6088739806242445,
      "grad_norm": 0.14304757118225098,
      "learning_rate": 1.1204393637472394e-06,
      "loss": 0.0037,
      "step": 250850
    },
    {
      "epoch": 1.6089381175180306,
      "grad_norm": 0.08332877606153488,
      "learning_rate": 1.1200863042167093e-06,
      "loss": 0.0012,
      "step": 250860
    },
    {
      "epoch": 1.6090022544118165,
      "grad_norm": 0.04952731728553772,
      "learning_rate": 1.119733293304271e-06,
      "loss": 0.0014,
      "step": 250870
    },
    {
      "epoch": 1.6090663913056027,
      "grad_norm": 0.039637889713048935,
      "learning_rate": 1.1193803310143463e-06,
      "loss": 0.0016,
      "step": 250880
    },
    {
      "epoch": 1.6091305281993886,
      "grad_norm": 0.0378977470099926,
      "learning_rate": 1.119027417351361e-06,
      "loss": 0.0008,
      "step": 250890
    },
    {
      "epoch": 1.6091946650931748,
      "grad_norm": 0.05773506313562393,
      "learning_rate": 1.1186745523197357e-06,
      "loss": 0.0014,
      "step": 250900
    },
    {
      "epoch": 1.609258801986961,
      "grad_norm": 0.11306332051753998,
      "learning_rate": 1.1183217359238924e-06,
      "loss": 0.0009,
      "step": 250910
    },
    {
      "epoch": 1.609322938880747,
      "grad_norm": 0.11664940416812897,
      "learning_rate": 1.1179689681682504e-06,
      "loss": 0.001,
      "step": 250920
    },
    {
      "epoch": 1.6093870757745332,
      "grad_norm": 0.06874285638332367,
      "learning_rate": 1.1176162490572328e-06,
      "loss": 0.0024,
      "step": 250930
    },
    {
      "epoch": 1.6094512126683194,
      "grad_norm": 0.007244887761771679,
      "learning_rate": 1.1172635785952584e-06,
      "loss": 0.0014,
      "step": 250940
    },
    {
      "epoch": 1.6095153495621055,
      "grad_norm": 0.08755049109458923,
      "learning_rate": 1.116910956786747e-06,
      "loss": 0.0016,
      "step": 250950
    },
    {
      "epoch": 1.6095794864558914,
      "grad_norm": 0.11688628792762756,
      "learning_rate": 1.1165583836361143e-06,
      "loss": 0.0011,
      "step": 250960
    },
    {
      "epoch": 1.6096436233496776,
      "grad_norm": 0.03535597771406174,
      "learning_rate": 1.116205859147782e-06,
      "loss": 0.0023,
      "step": 250970
    },
    {
      "epoch": 1.6097077602434635,
      "grad_norm": 0.06629591435194016,
      "learning_rate": 1.1158533833261665e-06,
      "loss": 0.0011,
      "step": 250980
    },
    {
      "epoch": 1.6097718971372497,
      "grad_norm": 0.03955451771616936,
      "learning_rate": 1.1155009561756836e-06,
      "loss": 0.0015,
      "step": 250990
    },
    {
      "epoch": 1.6098360340310358,
      "grad_norm": 0.08270388096570969,
      "learning_rate": 1.1151485777007487e-06,
      "loss": 0.0008,
      "step": 251000
    },
    {
      "epoch": 1.609900170924822,
      "grad_norm": 0.10194525867700577,
      "learning_rate": 1.11479624790578e-06,
      "loss": 0.0016,
      "step": 251010
    },
    {
      "epoch": 1.609964307818608,
      "grad_norm": 0.08749617636203766,
      "learning_rate": 1.1144439667951906e-06,
      "loss": 0.0012,
      "step": 251020
    },
    {
      "epoch": 1.6100284447123943,
      "grad_norm": 0.052722033113241196,
      "learning_rate": 1.1140917343733943e-06,
      "loss": 0.0011,
      "step": 251030
    },
    {
      "epoch": 1.6100925816061802,
      "grad_norm": 0.05054409056901932,
      "learning_rate": 1.1137395506448074e-06,
      "loss": 0.0004,
      "step": 251040
    },
    {
      "epoch": 1.6101567184999663,
      "grad_norm": 0.021686574444174767,
      "learning_rate": 1.1133874156138407e-06,
      "loss": 0.0006,
      "step": 251050
    },
    {
      "epoch": 1.6102208553937523,
      "grad_norm": 0.017417414113879204,
      "learning_rate": 1.1130353292849083e-06,
      "loss": 0.0007,
      "step": 251060
    },
    {
      "epoch": 1.6102849922875384,
      "grad_norm": 0.003439707215875387,
      "learning_rate": 1.1126832916624192e-06,
      "loss": 0.001,
      "step": 251070
    },
    {
      "epoch": 1.6103491291813246,
      "grad_norm": 0.023953434079885483,
      "learning_rate": 1.1123313027507882e-06,
      "loss": 0.0009,
      "step": 251080
    },
    {
      "epoch": 1.6104132660751107,
      "grad_norm": 0.01200141292065382,
      "learning_rate": 1.1119793625544246e-06,
      "loss": 0.0006,
      "step": 251090
    },
    {
      "epoch": 1.6104774029688969,
      "grad_norm": 0.10512842237949371,
      "learning_rate": 1.111627471077738e-06,
      "loss": 0.0012,
      "step": 251100
    },
    {
      "epoch": 1.610541539862683,
      "grad_norm": 0.10881675034761429,
      "learning_rate": 1.111275628325137e-06,
      "loss": 0.0014,
      "step": 251110
    },
    {
      "epoch": 1.6106056767564692,
      "grad_norm": 0.11241288483142853,
      "learning_rate": 1.1109238343010326e-06,
      "loss": 0.0016,
      "step": 251120
    },
    {
      "epoch": 1.610669813650255,
      "grad_norm": 0.10109782963991165,
      "learning_rate": 1.1105720890098327e-06,
      "loss": 0.0016,
      "step": 251130
    },
    {
      "epoch": 1.6107339505440412,
      "grad_norm": 0.22030378878116608,
      "learning_rate": 1.1102203924559441e-06,
      "loss": 0.0008,
      "step": 251140
    },
    {
      "epoch": 1.6107980874378272,
      "grad_norm": 0.0337819904088974,
      "learning_rate": 1.1098687446437722e-06,
      "loss": 0.001,
      "step": 251150
    },
    {
      "epoch": 1.6108622243316133,
      "grad_norm": 0.01923406682908535,
      "learning_rate": 1.1095171455777264e-06,
      "loss": 0.0014,
      "step": 251160
    },
    {
      "epoch": 1.6109263612253995,
      "grad_norm": 0.08116120845079422,
      "learning_rate": 1.109165595262212e-06,
      "loss": 0.0007,
      "step": 251170
    },
    {
      "epoch": 1.6109904981191856,
      "grad_norm": 0.014466686174273491,
      "learning_rate": 1.1088140937016318e-06,
      "loss": 0.0008,
      "step": 251180
    },
    {
      "epoch": 1.6110546350129717,
      "grad_norm": 0.052190881222486496,
      "learning_rate": 1.108462640900393e-06,
      "loss": 0.0013,
      "step": 251190
    },
    {
      "epoch": 1.611118771906758,
      "grad_norm": 0.11071562767028809,
      "learning_rate": 1.1081112368628989e-06,
      "loss": 0.0034,
      "step": 251200
    },
    {
      "epoch": 1.611182908800544,
      "grad_norm": 0.015977121889591217,
      "learning_rate": 1.107759881593552e-06,
      "loss": 0.0006,
      "step": 251210
    },
    {
      "epoch": 1.61124704569433,
      "grad_norm": 0.10085508227348328,
      "learning_rate": 1.1074085750967545e-06,
      "loss": 0.0014,
      "step": 251220
    },
    {
      "epoch": 1.6113111825881161,
      "grad_norm": 0.11096709966659546,
      "learning_rate": 1.107057317376911e-06,
      "loss": 0.001,
      "step": 251230
    },
    {
      "epoch": 1.611375319481902,
      "grad_norm": 0.04569392651319504,
      "learning_rate": 1.1067061084384218e-06,
      "loss": 0.0012,
      "step": 251240
    },
    {
      "epoch": 1.6114394563756882,
      "grad_norm": 0.10049768537282944,
      "learning_rate": 1.1063549482856855e-06,
      "loss": 0.0011,
      "step": 251250
    },
    {
      "epoch": 1.6115035932694743,
      "grad_norm": 0.09229201078414917,
      "learning_rate": 1.1060038369231063e-06,
      "loss": 0.002,
      "step": 251260
    },
    {
      "epoch": 1.6115677301632605,
      "grad_norm": 0.030462343245744705,
      "learning_rate": 1.1056527743550805e-06,
      "loss": 0.0016,
      "step": 251270
    },
    {
      "epoch": 1.6116318670570466,
      "grad_norm": 0.05826370045542717,
      "learning_rate": 1.10530176058601e-06,
      "loss": 0.001,
      "step": 251280
    },
    {
      "epoch": 1.6116960039508328,
      "grad_norm": 0.07355988025665283,
      "learning_rate": 1.104950795620292e-06,
      "loss": 0.0007,
      "step": 251290
    },
    {
      "epoch": 1.6117601408446187,
      "grad_norm": 0.11956194043159485,
      "learning_rate": 1.1045998794623231e-06,
      "loss": 0.0017,
      "step": 251300
    },
    {
      "epoch": 1.6118242777384049,
      "grad_norm": 0.02855396270751953,
      "learning_rate": 1.1042490121165033e-06,
      "loss": 0.0009,
      "step": 251310
    },
    {
      "epoch": 1.6118884146321908,
      "grad_norm": 0.010830280371010303,
      "learning_rate": 1.1038981935872272e-06,
      "loss": 0.0013,
      "step": 251320
    },
    {
      "epoch": 1.611952551525977,
      "grad_norm": 0.06761819124221802,
      "learning_rate": 1.1035474238788912e-06,
      "loss": 0.0012,
      "step": 251330
    },
    {
      "epoch": 1.612016688419763,
      "grad_norm": 0.02414173260331154,
      "learning_rate": 1.1031967029958897e-06,
      "loss": 0.0006,
      "step": 251340
    },
    {
      "epoch": 1.6120808253135492,
      "grad_norm": 0.0437890999019146,
      "learning_rate": 1.1028460309426193e-06,
      "loss": 0.001,
      "step": 251350
    },
    {
      "epoch": 1.6121449622073354,
      "grad_norm": 0.0029892411548644304,
      "learning_rate": 1.102495407723474e-06,
      "loss": 0.0027,
      "step": 251360
    },
    {
      "epoch": 1.6122090991011215,
      "grad_norm": 0.0340309739112854,
      "learning_rate": 1.1021448333428464e-06,
      "loss": 0.0016,
      "step": 251370
    },
    {
      "epoch": 1.6122732359949077,
      "grad_norm": 0.09791620820760727,
      "learning_rate": 1.1017943078051285e-06,
      "loss": 0.0006,
      "step": 251380
    },
    {
      "epoch": 1.6123373728886936,
      "grad_norm": 0.09915640205144882,
      "learning_rate": 1.1014438311147152e-06,
      "loss": 0.0016,
      "step": 251390
    },
    {
      "epoch": 1.6124015097824798,
      "grad_norm": 0.1071750596165657,
      "learning_rate": 1.1010934032759968e-06,
      "loss": 0.001,
      "step": 251400
    },
    {
      "epoch": 1.6124656466762657,
      "grad_norm": 0.1516011655330658,
      "learning_rate": 1.1007430242933653e-06,
      "loss": 0.0036,
      "step": 251410
    },
    {
      "epoch": 1.6125297835700518,
      "grad_norm": 0.09297031909227371,
      "learning_rate": 1.1003926941712085e-06,
      "loss": 0.0008,
      "step": 251420
    },
    {
      "epoch": 1.612593920463838,
      "grad_norm": 0.22856420278549194,
      "learning_rate": 1.1000424129139197e-06,
      "loss": 0.0013,
      "step": 251430
    },
    {
      "epoch": 1.6126580573576241,
      "grad_norm": 0.1917046159505844,
      "learning_rate": 1.0996921805258864e-06,
      "loss": 0.0023,
      "step": 251440
    },
    {
      "epoch": 1.6127221942514103,
      "grad_norm": 0.05130043625831604,
      "learning_rate": 1.0993419970114966e-06,
      "loss": 0.0015,
      "step": 251450
    },
    {
      "epoch": 1.6127863311451964,
      "grad_norm": 0.01813359372317791,
      "learning_rate": 1.098991862375141e-06,
      "loss": 0.0019,
      "step": 251460
    },
    {
      "epoch": 1.6128504680389824,
      "grad_norm": 0.01646178402006626,
      "learning_rate": 1.098641776621205e-06,
      "loss": 0.0012,
      "step": 251470
    },
    {
      "epoch": 1.6129146049327685,
      "grad_norm": 0.2576580345630646,
      "learning_rate": 1.098291739754076e-06,
      "loss": 0.0007,
      "step": 251480
    },
    {
      "epoch": 1.6129787418265547,
      "grad_norm": 0.0731213241815567,
      "learning_rate": 1.0979417517781383e-06,
      "loss": 0.001,
      "step": 251490
    },
    {
      "epoch": 1.6130428787203406,
      "grad_norm": 0.00965266302227974,
      "learning_rate": 1.097591812697781e-06,
      "loss": 0.0015,
      "step": 251500
    },
    {
      "epoch": 1.6131070156141267,
      "grad_norm": 0.075978122651577,
      "learning_rate": 1.0972419225173869e-06,
      "loss": 0.001,
      "step": 251510
    },
    {
      "epoch": 1.6131711525079129,
      "grad_norm": 0.05839274823665619,
      "learning_rate": 1.0968920812413409e-06,
      "loss": 0.0008,
      "step": 251520
    },
    {
      "epoch": 1.613235289401699,
      "grad_norm": 0.12388291209936142,
      "learning_rate": 1.0965422888740252e-06,
      "loss": 0.0016,
      "step": 251530
    },
    {
      "epoch": 1.6132994262954852,
      "grad_norm": 0.06515951454639435,
      "learning_rate": 1.0961925454198257e-06,
      "loss": 0.001,
      "step": 251540
    },
    {
      "epoch": 1.6133635631892713,
      "grad_norm": 0.04512935131788254,
      "learning_rate": 1.0958428508831237e-06,
      "loss": 0.0005,
      "step": 251550
    },
    {
      "epoch": 1.6134277000830572,
      "grad_norm": 0.008951231837272644,
      "learning_rate": 1.0954932052683014e-06,
      "loss": 0.0017,
      "step": 251560
    },
    {
      "epoch": 1.6134918369768434,
      "grad_norm": 0.20069415867328644,
      "learning_rate": 1.0951436085797378e-06,
      "loss": 0.0006,
      "step": 251570
    },
    {
      "epoch": 1.6135559738706293,
      "grad_norm": 0.006812704261392355,
      "learning_rate": 1.0947940608218171e-06,
      "loss": 0.0009,
      "step": 251580
    },
    {
      "epoch": 1.6136201107644155,
      "grad_norm": 0.14547978341579437,
      "learning_rate": 1.094444561998918e-06,
      "loss": 0.0016,
      "step": 251590
    },
    {
      "epoch": 1.6136842476582016,
      "grad_norm": 0.07378130406141281,
      "learning_rate": 1.0940951121154187e-06,
      "loss": 0.0016,
      "step": 251600
    },
    {
      "epoch": 1.6137483845519878,
      "grad_norm": 0.06077616661787033,
      "learning_rate": 1.0937457111757e-06,
      "loss": 0.0024,
      "step": 251610
    },
    {
      "epoch": 1.613812521445774,
      "grad_norm": 0.0864984542131424,
      "learning_rate": 1.0933963591841395e-06,
      "loss": 0.0012,
      "step": 251620
    },
    {
      "epoch": 1.61387665833956,
      "grad_norm": 0.0956496149301529,
      "learning_rate": 1.0930470561451145e-06,
      "loss": 0.0008,
      "step": 251630
    },
    {
      "epoch": 1.6139407952333462,
      "grad_norm": 0.09259046614170074,
      "learning_rate": 1.0926978020630008e-06,
      "loss": 0.0009,
      "step": 251640
    },
    {
      "epoch": 1.6140049321271321,
      "grad_norm": 0.03744299337267876,
      "learning_rate": 1.0923485969421776e-06,
      "loss": 0.0009,
      "step": 251650
    },
    {
      "epoch": 1.6140690690209183,
      "grad_norm": 0.03746733069419861,
      "learning_rate": 1.0919994407870194e-06,
      "loss": 0.0012,
      "step": 251660
    },
    {
      "epoch": 1.6141332059147042,
      "grad_norm": 0.023150503635406494,
      "learning_rate": 1.0916503336019008e-06,
      "loss": 0.0046,
      "step": 251670
    },
    {
      "epoch": 1.6141973428084904,
      "grad_norm": 0.05050637200474739,
      "learning_rate": 1.0913012753911955e-06,
      "loss": 0.0004,
      "step": 251680
    },
    {
      "epoch": 1.6142614797022765,
      "grad_norm": 0.14014700055122375,
      "learning_rate": 1.0909522661592804e-06,
      "loss": 0.0014,
      "step": 251690
    },
    {
      "epoch": 1.6143256165960627,
      "grad_norm": 0.012953144498169422,
      "learning_rate": 1.090603305910527e-06,
      "loss": 0.003,
      "step": 251700
    },
    {
      "epoch": 1.6143897534898488,
      "grad_norm": 0.09172007441520691,
      "learning_rate": 1.0902543946493083e-06,
      "loss": 0.0008,
      "step": 251710
    },
    {
      "epoch": 1.614453890383635,
      "grad_norm": 0.005768218543380499,
      "learning_rate": 1.0899055323799945e-06,
      "loss": 0.0008,
      "step": 251720
    },
    {
      "epoch": 1.6145180272774209,
      "grad_norm": 0.14142289757728577,
      "learning_rate": 1.0895567191069605e-06,
      "loss": 0.0008,
      "step": 251730
    },
    {
      "epoch": 1.614582164171207,
      "grad_norm": 0.07299283146858215,
      "learning_rate": 1.0892079548345758e-06,
      "loss": 0.0011,
      "step": 251740
    },
    {
      "epoch": 1.614646301064993,
      "grad_norm": 0.0013321398291736841,
      "learning_rate": 1.0888592395672087e-06,
      "loss": 0.0009,
      "step": 251750
    },
    {
      "epoch": 1.614710437958779,
      "grad_norm": 0.07189654558897018,
      "learning_rate": 1.0885105733092322e-06,
      "loss": 0.0017,
      "step": 251760
    },
    {
      "epoch": 1.6147745748525653,
      "grad_norm": 0.10784239321947098,
      "learning_rate": 1.0881619560650137e-06,
      "loss": 0.0017,
      "step": 251770
    },
    {
      "epoch": 1.6148387117463514,
      "grad_norm": 0.07716381549835205,
      "learning_rate": 1.0878133878389213e-06,
      "loss": 0.0017,
      "step": 251780
    },
    {
      "epoch": 1.6149028486401376,
      "grad_norm": 0.041925061494112015,
      "learning_rate": 1.0874648686353224e-06,
      "loss": 0.0007,
      "step": 251790
    },
    {
      "epoch": 1.6149669855339237,
      "grad_norm": 0.05732448399066925,
      "learning_rate": 1.0871163984585859e-06,
      "loss": 0.0015,
      "step": 251800
    },
    {
      "epoch": 1.6150311224277099,
      "grad_norm": 0.05569880083203316,
      "learning_rate": 1.0867679773130775e-06,
      "loss": 0.0009,
      "step": 251810
    },
    {
      "epoch": 1.6150952593214958,
      "grad_norm": 0.06757721304893494,
      "learning_rate": 1.0864196052031627e-06,
      "loss": 0.001,
      "step": 251820
    },
    {
      "epoch": 1.615159396215282,
      "grad_norm": 0.3683226406574249,
      "learning_rate": 1.0860712821332064e-06,
      "loss": 0.002,
      "step": 251830
    },
    {
      "epoch": 1.6152235331090679,
      "grad_norm": 0.05867931991815567,
      "learning_rate": 1.0857230081075754e-06,
      "loss": 0.0007,
      "step": 251840
    },
    {
      "epoch": 1.615287670002854,
      "grad_norm": 0.2559502422809601,
      "learning_rate": 1.085374783130632e-06,
      "loss": 0.0009,
      "step": 251850
    },
    {
      "epoch": 1.6153518068966402,
      "grad_norm": 0.08006120473146439,
      "learning_rate": 1.08502660720674e-06,
      "loss": 0.0012,
      "step": 251860
    },
    {
      "epoch": 1.6154159437904263,
      "grad_norm": 0.01886802911758423,
      "learning_rate": 1.0846784803402633e-06,
      "loss": 0.0008,
      "step": 251870
    },
    {
      "epoch": 1.6154800806842124,
      "grad_norm": 0.05986326187849045,
      "learning_rate": 1.0843304025355638e-06,
      "loss": 0.001,
      "step": 251880
    },
    {
      "epoch": 1.6155442175779986,
      "grad_norm": 0.012768320739269257,
      "learning_rate": 1.083982373797003e-06,
      "loss": 0.0011,
      "step": 251890
    },
    {
      "epoch": 1.6156083544717845,
      "grad_norm": 0.13626137375831604,
      "learning_rate": 1.0836343941289395e-06,
      "loss": 0.0006,
      "step": 251900
    },
    {
      "epoch": 1.6156724913655707,
      "grad_norm": 0.06842995434999466,
      "learning_rate": 1.0832864635357382e-06,
      "loss": 0.0011,
      "step": 251910
    },
    {
      "epoch": 1.6157366282593568,
      "grad_norm": 0.11791258305311203,
      "learning_rate": 1.082938582021757e-06,
      "loss": 0.0012,
      "step": 251920
    },
    {
      "epoch": 1.6158007651531427,
      "grad_norm": 0.11330414563417435,
      "learning_rate": 1.082590749591354e-06,
      "loss": 0.0019,
      "step": 251930
    },
    {
      "epoch": 1.615864902046929,
      "grad_norm": 0.013453952036798,
      "learning_rate": 1.0822429662488875e-06,
      "loss": 0.0008,
      "step": 251940
    },
    {
      "epoch": 1.615929038940715,
      "grad_norm": 0.06207551807165146,
      "learning_rate": 1.0818952319987187e-06,
      "loss": 0.0008,
      "step": 251950
    },
    {
      "epoch": 1.6159931758345012,
      "grad_norm": 0.05609503015875816,
      "learning_rate": 1.081547546845202e-06,
      "loss": 0.0009,
      "step": 251960
    },
    {
      "epoch": 1.6160573127282873,
      "grad_norm": 0.2726113498210907,
      "learning_rate": 1.0811999107926958e-06,
      "loss": 0.0015,
      "step": 251970
    },
    {
      "epoch": 1.6161214496220735,
      "grad_norm": 0.04356217011809349,
      "learning_rate": 1.0808523238455532e-06,
      "loss": 0.0008,
      "step": 251980
    },
    {
      "epoch": 1.6161855865158594,
      "grad_norm": 0.04935280233621597,
      "learning_rate": 1.0805047860081335e-06,
      "loss": 0.0014,
      "step": 251990
    },
    {
      "epoch": 1.6162497234096456,
      "grad_norm": 0.15753525495529175,
      "learning_rate": 1.0801572972847907e-06,
      "loss": 0.0025,
      "step": 252000
    },
    {
      "epoch": 1.6163138603034315,
      "grad_norm": 0.0707324743270874,
      "learning_rate": 1.0798098576798766e-06,
      "loss": 0.0007,
      "step": 252010
    },
    {
      "epoch": 1.6163779971972176,
      "grad_norm": 0.05470231920480728,
      "learning_rate": 1.0794624671977465e-06,
      "loss": 0.0015,
      "step": 252020
    },
    {
      "epoch": 1.6164421340910038,
      "grad_norm": 0.13599084317684174,
      "learning_rate": 1.0791151258427557e-06,
      "loss": 0.0015,
      "step": 252030
    },
    {
      "epoch": 1.61650627098479,
      "grad_norm": 0.021795066073536873,
      "learning_rate": 1.0787678336192542e-06,
      "loss": 0.0015,
      "step": 252040
    },
    {
      "epoch": 1.616570407878576,
      "grad_norm": 0.07633227854967117,
      "learning_rate": 1.0784205905315941e-06,
      "loss": 0.0012,
      "step": 252050
    },
    {
      "epoch": 1.6166345447723622,
      "grad_norm": 0.008000954985618591,
      "learning_rate": 1.078073396584125e-06,
      "loss": 0.0011,
      "step": 252060
    },
    {
      "epoch": 1.6166986816661484,
      "grad_norm": 0.01430139597505331,
      "learning_rate": 1.0777262517812014e-06,
      "loss": 0.0009,
      "step": 252070
    },
    {
      "epoch": 1.6167628185599343,
      "grad_norm": 0.05942973494529724,
      "learning_rate": 1.0773791561271706e-06,
      "loss": 0.0012,
      "step": 252080
    },
    {
      "epoch": 1.6168269554537205,
      "grad_norm": 0.004552983678877354,
      "learning_rate": 1.0770321096263825e-06,
      "loss": 0.0011,
      "step": 252090
    },
    {
      "epoch": 1.6168910923475064,
      "grad_norm": 0.12817177176475525,
      "learning_rate": 1.0766851122831845e-06,
      "loss": 0.001,
      "step": 252100
    },
    {
      "epoch": 1.6169552292412925,
      "grad_norm": 0.1432831883430481,
      "learning_rate": 1.0763381641019272e-06,
      "loss": 0.0011,
      "step": 252110
    },
    {
      "epoch": 1.6170193661350787,
      "grad_norm": 0.050865817815065384,
      "learning_rate": 1.075991265086957e-06,
      "loss": 0.0009,
      "step": 252120
    },
    {
      "epoch": 1.6170835030288648,
      "grad_norm": 0.10968196392059326,
      "learning_rate": 1.0756444152426192e-06,
      "loss": 0.0012,
      "step": 252130
    },
    {
      "epoch": 1.617147639922651,
      "grad_norm": 0.08784846216440201,
      "learning_rate": 1.075297614573263e-06,
      "loss": 0.0009,
      "step": 252140
    },
    {
      "epoch": 1.6172117768164371,
      "grad_norm": 0.0864066407084465,
      "learning_rate": 1.0749508630832329e-06,
      "loss": 0.0008,
      "step": 252150
    },
    {
      "epoch": 1.617275913710223,
      "grad_norm": 0.15843220055103302,
      "learning_rate": 1.074604160776873e-06,
      "loss": 0.0015,
      "step": 252160
    },
    {
      "epoch": 1.6173400506040092,
      "grad_norm": 0.12567143142223358,
      "learning_rate": 1.0742575076585276e-06,
      "loss": 0.0016,
      "step": 252170
    },
    {
      "epoch": 1.6174041874977951,
      "grad_norm": 0.048958681523799896,
      "learning_rate": 1.0739109037325423e-06,
      "loss": 0.0006,
      "step": 252180
    },
    {
      "epoch": 1.6174683243915813,
      "grad_norm": 0.08918514847755432,
      "learning_rate": 1.07356434900326e-06,
      "loss": 0.001,
      "step": 252190
    },
    {
      "epoch": 1.6175324612853674,
      "grad_norm": 0.004656277596950531,
      "learning_rate": 1.0732178434750218e-06,
      "loss": 0.0012,
      "step": 252200
    },
    {
      "epoch": 1.6175965981791536,
      "grad_norm": 0.13030418753623962,
      "learning_rate": 1.0728713871521695e-06,
      "loss": 0.0016,
      "step": 252210
    },
    {
      "epoch": 1.6176607350729397,
      "grad_norm": 0.03686818480491638,
      "learning_rate": 1.0725249800390468e-06,
      "loss": 0.0011,
      "step": 252220
    },
    {
      "epoch": 1.6177248719667259,
      "grad_norm": 0.0056649139150977135,
      "learning_rate": 1.0721786221399928e-06,
      "loss": 0.0012,
      "step": 252230
    },
    {
      "epoch": 1.617789008860512,
      "grad_norm": 0.06054788827896118,
      "learning_rate": 1.0718323134593477e-06,
      "loss": 0.0031,
      "step": 252240
    },
    {
      "epoch": 1.617853145754298,
      "grad_norm": 0.0929122045636177,
      "learning_rate": 1.0714860540014504e-06,
      "loss": 0.001,
      "step": 252250
    },
    {
      "epoch": 1.617917282648084,
      "grad_norm": 0.09619086980819702,
      "learning_rate": 1.0711398437706416e-06,
      "loss": 0.0007,
      "step": 252260
    },
    {
      "epoch": 1.61798141954187,
      "grad_norm": 0.06638479977846146,
      "learning_rate": 1.0707936827712584e-06,
      "loss": 0.0013,
      "step": 252270
    },
    {
      "epoch": 1.6180455564356562,
      "grad_norm": 0.172328919172287,
      "learning_rate": 1.0704475710076367e-06,
      "loss": 0.0012,
      "step": 252280
    },
    {
      "epoch": 1.6181096933294423,
      "grad_norm": 0.010810944251716137,
      "learning_rate": 1.0701015084841171e-06,
      "loss": 0.001,
      "step": 252290
    },
    {
      "epoch": 1.6181738302232285,
      "grad_norm": 0.04878225550055504,
      "learning_rate": 1.0697554952050342e-06,
      "loss": 0.0017,
      "step": 252300
    },
    {
      "epoch": 1.6182379671170146,
      "grad_norm": 0.06370214372873306,
      "learning_rate": 1.0694095311747243e-06,
      "loss": 0.0008,
      "step": 252310
    },
    {
      "epoch": 1.6183021040108008,
      "grad_norm": 0.04039781913161278,
      "learning_rate": 1.0690636163975204e-06,
      "loss": 0.0008,
      "step": 252320
    },
    {
      "epoch": 1.6183662409045867,
      "grad_norm": 0.07960638403892517,
      "learning_rate": 1.0687177508777602e-06,
      "loss": 0.0015,
      "step": 252330
    },
    {
      "epoch": 1.6184303777983728,
      "grad_norm": 0.43764522671699524,
      "learning_rate": 1.0683719346197758e-06,
      "loss": 0.0012,
      "step": 252340
    },
    {
      "epoch": 1.618494514692159,
      "grad_norm": 0.03488235920667648,
      "learning_rate": 1.0680261676279014e-06,
      "loss": 0.0007,
      "step": 252350
    },
    {
      "epoch": 1.618558651585945,
      "grad_norm": 0.035237208008766174,
      "learning_rate": 1.067680449906468e-06,
      "loss": 0.0007,
      "step": 252360
    },
    {
      "epoch": 1.618622788479731,
      "grad_norm": 0.07528963685035706,
      "learning_rate": 1.0673347814598101e-06,
      "loss": 0.0012,
      "step": 252370
    },
    {
      "epoch": 1.6186869253735172,
      "grad_norm": 0.20800499618053436,
      "learning_rate": 1.066989162292258e-06,
      "loss": 0.0029,
      "step": 252380
    },
    {
      "epoch": 1.6187510622673034,
      "grad_norm": 0.13910309970378876,
      "learning_rate": 1.0666435924081424e-06,
      "loss": 0.0012,
      "step": 252390
    },
    {
      "epoch": 1.6188151991610895,
      "grad_norm": 0.005096248351037502,
      "learning_rate": 1.0662980718117927e-06,
      "loss": 0.0012,
      "step": 252400
    },
    {
      "epoch": 1.6188793360548757,
      "grad_norm": 0.13247188925743103,
      "learning_rate": 1.065952600507541e-06,
      "loss": 0.0014,
      "step": 252410
    },
    {
      "epoch": 1.6189434729486616,
      "grad_norm": 0.05428318679332733,
      "learning_rate": 1.0656071784997147e-06,
      "loss": 0.0008,
      "step": 252420
    },
    {
      "epoch": 1.6190076098424477,
      "grad_norm": 0.09457910805940628,
      "learning_rate": 1.0652618057926405e-06,
      "loss": 0.0008,
      "step": 252430
    },
    {
      "epoch": 1.6190717467362337,
      "grad_norm": 0.07830148190259933,
      "learning_rate": 1.06491648239065e-06,
      "loss": 0.0015,
      "step": 252440
    },
    {
      "epoch": 1.6191358836300198,
      "grad_norm": 0.05774727463722229,
      "learning_rate": 1.064571208298068e-06,
      "loss": 0.0007,
      "step": 252450
    },
    {
      "epoch": 1.619200020523806,
      "grad_norm": 0.08623258024454117,
      "learning_rate": 1.0642259835192215e-06,
      "loss": 0.0012,
      "step": 252460
    },
    {
      "epoch": 1.619264157417592,
      "grad_norm": 0.051753610372543335,
      "learning_rate": 1.0638808080584346e-06,
      "loss": 0.0024,
      "step": 252470
    },
    {
      "epoch": 1.6193282943113783,
      "grad_norm": 0.028300661593675613,
      "learning_rate": 1.063535681920036e-06,
      "loss": 0.001,
      "step": 252480
    },
    {
      "epoch": 1.6193924312051644,
      "grad_norm": 0.09492386877536774,
      "learning_rate": 1.0631906051083484e-06,
      "loss": 0.0018,
      "step": 252490
    },
    {
      "epoch": 1.6194565680989506,
      "grad_norm": 0.0015070786466822028,
      "learning_rate": 1.0628455776276964e-06,
      "loss": 0.0003,
      "step": 252500
    },
    {
      "epoch": 1.6195207049927365,
      "grad_norm": 0.03143575042486191,
      "learning_rate": 1.0625005994824017e-06,
      "loss": 0.0017,
      "step": 252510
    },
    {
      "epoch": 1.6195848418865226,
      "grad_norm": 0.07570742070674896,
      "learning_rate": 1.06215567067679e-06,
      "loss": 0.0008,
      "step": 252520
    },
    {
      "epoch": 1.6196489787803086,
      "grad_norm": 0.036942899227142334,
      "learning_rate": 1.0618107912151815e-06,
      "loss": 0.0014,
      "step": 252530
    },
    {
      "epoch": 1.6197131156740947,
      "grad_norm": 0.06867800652980804,
      "learning_rate": 1.0614659611018991e-06,
      "loss": 0.0008,
      "step": 252540
    },
    {
      "epoch": 1.6197772525678809,
      "grad_norm": 0.14280128479003906,
      "learning_rate": 1.0611211803412614e-06,
      "loss": 0.0009,
      "step": 252550
    },
    {
      "epoch": 1.619841389461667,
      "grad_norm": 0.03174474462866783,
      "learning_rate": 1.0607764489375915e-06,
      "loss": 0.0014,
      "step": 252560
    },
    {
      "epoch": 1.6199055263554532,
      "grad_norm": 0.24680443108081818,
      "learning_rate": 1.0604317668952084e-06,
      "loss": 0.0011,
      "step": 252570
    },
    {
      "epoch": 1.6199696632492393,
      "grad_norm": 0.05565698444843292,
      "learning_rate": 1.0600871342184294e-06,
      "loss": 0.0009,
      "step": 252580
    },
    {
      "epoch": 1.6200338001430252,
      "grad_norm": 0.2579374313354492,
      "learning_rate": 1.0597425509115756e-06,
      "loss": 0.0009,
      "step": 252590
    },
    {
      "epoch": 1.6200979370368114,
      "grad_norm": 0.007848929613828659,
      "learning_rate": 1.0593980169789636e-06,
      "loss": 0.0016,
      "step": 252600
    },
    {
      "epoch": 1.6201620739305973,
      "grad_norm": 0.004670563619583845,
      "learning_rate": 1.0590535324249113e-06,
      "loss": 0.0008,
      "step": 252610
    },
    {
      "epoch": 1.6202262108243835,
      "grad_norm": 0.03399378061294556,
      "learning_rate": 1.0587090972537327e-06,
      "loss": 0.0017,
      "step": 252620
    },
    {
      "epoch": 1.6202903477181696,
      "grad_norm": 0.10650131106376648,
      "learning_rate": 1.0583647114697483e-06,
      "loss": 0.0008,
      "step": 252630
    },
    {
      "epoch": 1.6203544846119557,
      "grad_norm": 0.1858557015657425,
      "learning_rate": 1.05802037507727e-06,
      "loss": 0.0016,
      "step": 252640
    },
    {
      "epoch": 1.620418621505742,
      "grad_norm": 0.08053423464298248,
      "learning_rate": 1.0576760880806142e-06,
      "loss": 0.0015,
      "step": 252650
    },
    {
      "epoch": 1.620482758399528,
      "grad_norm": 0.09590917080640793,
      "learning_rate": 1.0573318504840935e-06,
      "loss": 0.0017,
      "step": 252660
    },
    {
      "epoch": 1.6205468952933142,
      "grad_norm": 0.03918692097067833,
      "learning_rate": 1.0569876622920232e-06,
      "loss": 0.0003,
      "step": 252670
    },
    {
      "epoch": 1.6206110321871001,
      "grad_norm": 0.14594435691833496,
      "learning_rate": 1.0566435235087157e-06,
      "loss": 0.001,
      "step": 252680
    },
    {
      "epoch": 1.6206751690808863,
      "grad_norm": 0.10581952333450317,
      "learning_rate": 1.0562994341384835e-06,
      "loss": 0.0016,
      "step": 252690
    },
    {
      "epoch": 1.6207393059746722,
      "grad_norm": 0.06139199808239937,
      "learning_rate": 1.0559553941856365e-06,
      "loss": 0.0012,
      "step": 252700
    },
    {
      "epoch": 1.6208034428684583,
      "grad_norm": 0.09698482602834702,
      "learning_rate": 1.0556114036544878e-06,
      "loss": 0.0014,
      "step": 252710
    },
    {
      "epoch": 1.6208675797622445,
      "grad_norm": 0.09465713798999786,
      "learning_rate": 1.055267462549348e-06,
      "loss": 0.0009,
      "step": 252720
    },
    {
      "epoch": 1.6209317166560306,
      "grad_norm": 0.11443684250116348,
      "learning_rate": 1.0549235708745249e-06,
      "loss": 0.0043,
      "step": 252730
    },
    {
      "epoch": 1.6209958535498168,
      "grad_norm": 0.1651214212179184,
      "learning_rate": 1.0545797286343296e-06,
      "loss": 0.001,
      "step": 252740
    },
    {
      "epoch": 1.621059990443603,
      "grad_norm": 0.06214063987135887,
      "learning_rate": 1.0542359358330707e-06,
      "loss": 0.0019,
      "step": 252750
    },
    {
      "epoch": 1.621124127337389,
      "grad_norm": 0.034032728523015976,
      "learning_rate": 1.053892192475055e-06,
      "loss": 0.0013,
      "step": 252760
    },
    {
      "epoch": 1.621188264231175,
      "grad_norm": 0.011761156842112541,
      "learning_rate": 1.0535484985645895e-06,
      "loss": 0.0007,
      "step": 252770
    },
    {
      "epoch": 1.6212524011249612,
      "grad_norm": 0.1583254039287567,
      "learning_rate": 1.0532048541059814e-06,
      "loss": 0.0007,
      "step": 252780
    },
    {
      "epoch": 1.621316538018747,
      "grad_norm": 0.15233290195465088,
      "learning_rate": 1.0528612591035386e-06,
      "loss": 0.0008,
      "step": 252790
    },
    {
      "epoch": 1.6213806749125332,
      "grad_norm": 0.004722410812973976,
      "learning_rate": 1.0525177135615656e-06,
      "loss": 0.0011,
      "step": 252800
    },
    {
      "epoch": 1.6214448118063194,
      "grad_norm": 0.10750328749418259,
      "learning_rate": 1.0521742174843663e-06,
      "loss": 0.0017,
      "step": 252810
    },
    {
      "epoch": 1.6215089487001055,
      "grad_norm": 0.020725587382912636,
      "learning_rate": 1.0518307708762448e-06,
      "loss": 0.0007,
      "step": 252820
    },
    {
      "epoch": 1.6215730855938917,
      "grad_norm": 0.18014080822467804,
      "learning_rate": 1.0514873737415065e-06,
      "loss": 0.001,
      "step": 252830
    },
    {
      "epoch": 1.6216372224876778,
      "grad_norm": 0.010807321406900883,
      "learning_rate": 1.051144026084453e-06,
      "loss": 0.0011,
      "step": 252840
    },
    {
      "epoch": 1.6217013593814638,
      "grad_norm": 0.059095799922943115,
      "learning_rate": 1.0508007279093862e-06,
      "loss": 0.0012,
      "step": 252850
    },
    {
      "epoch": 1.62176549627525,
      "grad_norm": 0.10498159378767014,
      "learning_rate": 1.0504574792206101e-06,
      "loss": 0.0014,
      "step": 252860
    },
    {
      "epoch": 1.6218296331690358,
      "grad_norm": 0.13310320675373077,
      "learning_rate": 1.050114280022424e-06,
      "loss": 0.0014,
      "step": 252870
    },
    {
      "epoch": 1.621893770062822,
      "grad_norm": 0.04748896136879921,
      "learning_rate": 1.0497711303191294e-06,
      "loss": 0.0006,
      "step": 252880
    },
    {
      "epoch": 1.6219579069566081,
      "grad_norm": 0.01747206225991249,
      "learning_rate": 1.049428030115024e-06,
      "loss": 0.0006,
      "step": 252890
    },
    {
      "epoch": 1.6220220438503943,
      "grad_norm": 0.017674749717116356,
      "learning_rate": 1.0490849794144103e-06,
      "loss": 0.0014,
      "step": 252900
    },
    {
      "epoch": 1.6220861807441804,
      "grad_norm": 0.008640342392027378,
      "learning_rate": 1.0487419782215858e-06,
      "loss": 0.0005,
      "step": 252910
    },
    {
      "epoch": 1.6221503176379666,
      "grad_norm": 0.04351125285029411,
      "learning_rate": 1.0483990265408477e-06,
      "loss": 0.0008,
      "step": 252920
    },
    {
      "epoch": 1.6222144545317527,
      "grad_norm": 0.004246466793119907,
      "learning_rate": 1.048056124376493e-06,
      "loss": 0.0007,
      "step": 252930
    },
    {
      "epoch": 1.6222785914255387,
      "grad_norm": 0.013899121433496475,
      "learning_rate": 1.0477132717328208e-06,
      "loss": 0.0004,
      "step": 252940
    },
    {
      "epoch": 1.6223427283193248,
      "grad_norm": 0.1409715712070465,
      "learning_rate": 1.0473704686141261e-06,
      "loss": 0.0024,
      "step": 252950
    },
    {
      "epoch": 1.6224068652131107,
      "grad_norm": 0.024969857186079025,
      "learning_rate": 1.0470277150247039e-06,
      "loss": 0.0016,
      "step": 252960
    },
    {
      "epoch": 1.6224710021068969,
      "grad_norm": 0.16803643107414246,
      "learning_rate": 1.0466850109688487e-06,
      "loss": 0.0032,
      "step": 252970
    },
    {
      "epoch": 1.622535139000683,
      "grad_norm": 0.11032666265964508,
      "learning_rate": 1.0463423564508567e-06,
      "loss": 0.0014,
      "step": 252980
    },
    {
      "epoch": 1.6225992758944692,
      "grad_norm": 0.07094945013523102,
      "learning_rate": 1.0459997514750204e-06,
      "loss": 0.0013,
      "step": 252990
    },
    {
      "epoch": 1.6226634127882553,
      "grad_norm": 0.014905665069818497,
      "learning_rate": 1.0456571960456324e-06,
      "loss": 0.0011,
      "step": 253000
    },
    {
      "epoch": 1.6227275496820415,
      "grad_norm": 0.08949162811040878,
      "learning_rate": 1.0453146901669863e-06,
      "loss": 0.0015,
      "step": 253010
    },
    {
      "epoch": 1.6227916865758274,
      "grad_norm": 0.06372623890638351,
      "learning_rate": 1.0449722338433743e-06,
      "loss": 0.0014,
      "step": 253020
    },
    {
      "epoch": 1.6228558234696135,
      "grad_norm": 0.003106101183220744,
      "learning_rate": 1.0446298270790866e-06,
      "loss": 0.0017,
      "step": 253030
    },
    {
      "epoch": 1.6229199603633997,
      "grad_norm": 0.0415562242269516,
      "learning_rate": 1.0442874698784128e-06,
      "loss": 0.0011,
      "step": 253040
    },
    {
      "epoch": 1.6229840972571856,
      "grad_norm": 0.12806938588619232,
      "learning_rate": 1.0439451622456453e-06,
      "loss": 0.0009,
      "step": 253050
    },
    {
      "epoch": 1.6230482341509718,
      "grad_norm": 0.0016295817913487554,
      "learning_rate": 1.043602904185072e-06,
      "loss": 0.0011,
      "step": 253060
    },
    {
      "epoch": 1.623112371044758,
      "grad_norm": 0.04967926815152168,
      "learning_rate": 1.0432606957009823e-06,
      "loss": 0.0009,
      "step": 253070
    },
    {
      "epoch": 1.623176507938544,
      "grad_norm": 0.08386543393135071,
      "learning_rate": 1.0429185367976625e-06,
      "loss": 0.0006,
      "step": 253080
    },
    {
      "epoch": 1.6232406448323302,
      "grad_norm": 0.05398569256067276,
      "learning_rate": 1.0425764274794032e-06,
      "loss": 0.001,
      "step": 253090
    },
    {
      "epoch": 1.6233047817261164,
      "grad_norm": 0.02327839843928814,
      "learning_rate": 1.0422343677504888e-06,
      "loss": 0.0016,
      "step": 253100
    },
    {
      "epoch": 1.6233689186199023,
      "grad_norm": 0.053075287491083145,
      "learning_rate": 1.0418923576152069e-06,
      "loss": 0.0011,
      "step": 253110
    },
    {
      "epoch": 1.6234330555136884,
      "grad_norm": 0.10003472864627838,
      "learning_rate": 1.0415503970778411e-06,
      "loss": 0.0008,
      "step": 253120
    },
    {
      "epoch": 1.6234971924074744,
      "grad_norm": 0.23413614928722382,
      "learning_rate": 1.041208486142679e-06,
      "loss": 0.0011,
      "step": 253130
    },
    {
      "epoch": 1.6235613293012605,
      "grad_norm": 0.17361804842948914,
      "learning_rate": 1.0408666248140043e-06,
      "loss": 0.0013,
      "step": 253140
    },
    {
      "epoch": 1.6236254661950467,
      "grad_norm": 0.0991687998175621,
      "learning_rate": 1.0405248130960988e-06,
      "loss": 0.0018,
      "step": 253150
    },
    {
      "epoch": 1.6236896030888328,
      "grad_norm": 0.21054257452487946,
      "learning_rate": 1.0401830509932488e-06,
      "loss": 0.0024,
      "step": 253160
    },
    {
      "epoch": 1.623753739982619,
      "grad_norm": 0.13935822248458862,
      "learning_rate": 1.0398413385097345e-06,
      "loss": 0.001,
      "step": 253170
    },
    {
      "epoch": 1.623817876876405,
      "grad_norm": 0.09200213849544525,
      "learning_rate": 1.0394996756498394e-06,
      "loss": 0.0006,
      "step": 253180
    },
    {
      "epoch": 1.6238820137701913,
      "grad_norm": 0.0637202262878418,
      "learning_rate": 1.0391580624178416e-06,
      "loss": 0.0005,
      "step": 253190
    },
    {
      "epoch": 1.6239461506639772,
      "grad_norm": 0.020669076591730118,
      "learning_rate": 1.0388164988180261e-06,
      "loss": 0.0013,
      "step": 253200
    },
    {
      "epoch": 1.6240102875577633,
      "grad_norm": 0.08464609831571579,
      "learning_rate": 1.0384749848546704e-06,
      "loss": 0.0007,
      "step": 253210
    },
    {
      "epoch": 1.6240744244515493,
      "grad_norm": 0.031027456745505333,
      "learning_rate": 1.0381335205320547e-06,
      "loss": 0.0009,
      "step": 253220
    },
    {
      "epoch": 1.6241385613453354,
      "grad_norm": 0.05245270952582359,
      "learning_rate": 1.0377921058544567e-06,
      "loss": 0.0009,
      "step": 253230
    },
    {
      "epoch": 1.6242026982391216,
      "grad_norm": 0.04250001534819603,
      "learning_rate": 1.0374507408261558e-06,
      "loss": 0.0014,
      "step": 253240
    },
    {
      "epoch": 1.6242668351329077,
      "grad_norm": 0.028570299968123436,
      "learning_rate": 1.0371094254514292e-06,
      "loss": 0.0006,
      "step": 253250
    },
    {
      "epoch": 1.6243309720266939,
      "grad_norm": 0.18932245671749115,
      "learning_rate": 1.0367681597345541e-06,
      "loss": 0.0014,
      "step": 253260
    },
    {
      "epoch": 1.62439510892048,
      "grad_norm": 0.003532364498823881,
      "learning_rate": 1.0364269436798053e-06,
      "loss": 0.0006,
      "step": 253270
    },
    {
      "epoch": 1.624459245814266,
      "grad_norm": 0.0064608012326061726,
      "learning_rate": 1.0360857772914601e-06,
      "loss": 0.0008,
      "step": 253280
    },
    {
      "epoch": 1.624523382708052,
      "grad_norm": 0.08528894931077957,
      "learning_rate": 1.0357446605737936e-06,
      "loss": 0.0009,
      "step": 253290
    },
    {
      "epoch": 1.624587519601838,
      "grad_norm": 0.14303399622440338,
      "learning_rate": 1.0354035935310785e-06,
      "loss": 0.0012,
      "step": 253300
    },
    {
      "epoch": 1.6246516564956242,
      "grad_norm": 0.08092918246984482,
      "learning_rate": 1.0350625761675908e-06,
      "loss": 0.0006,
      "step": 253310
    },
    {
      "epoch": 1.6247157933894103,
      "grad_norm": 0.09207907319068909,
      "learning_rate": 1.0347216084876033e-06,
      "loss": 0.0005,
      "step": 253320
    },
    {
      "epoch": 1.6247799302831964,
      "grad_norm": 0.06599237024784088,
      "learning_rate": 1.0343806904953873e-06,
      "loss": 0.0012,
      "step": 253330
    },
    {
      "epoch": 1.6248440671769826,
      "grad_norm": 0.06715761870145798,
      "learning_rate": 1.0340398221952146e-06,
      "loss": 0.0007,
      "step": 253340
    },
    {
      "epoch": 1.6249082040707687,
      "grad_norm": 0.174531027674675,
      "learning_rate": 1.0336990035913586e-06,
      "loss": 0.0008,
      "step": 253350
    },
    {
      "epoch": 1.624972340964555,
      "grad_norm": 0.06355522572994232,
      "learning_rate": 1.0333582346880887e-06,
      "loss": 0.0006,
      "step": 253360
    },
    {
      "epoch": 1.6250364778583408,
      "grad_norm": 0.22836199402809143,
      "learning_rate": 1.0330175154896748e-06,
      "loss": 0.0015,
      "step": 253370
    },
    {
      "epoch": 1.625100614752127,
      "grad_norm": 0.16075459122657776,
      "learning_rate": 1.0326768460003856e-06,
      "loss": 0.0006,
      "step": 253380
    },
    {
      "epoch": 1.625164751645913,
      "grad_norm": 0.011506589129567146,
      "learning_rate": 1.0323362262244923e-06,
      "loss": 0.0005,
      "step": 253390
    },
    {
      "epoch": 1.625228888539699,
      "grad_norm": 0.09543559700250626,
      "learning_rate": 1.0319956561662615e-06,
      "loss": 0.0006,
      "step": 253400
    },
    {
      "epoch": 1.6252930254334852,
      "grad_norm": 0.09004479646682739,
      "learning_rate": 1.0316551358299614e-06,
      "loss": 0.001,
      "step": 253410
    },
    {
      "epoch": 1.6253571623272713,
      "grad_norm": 0.22792024910449982,
      "learning_rate": 1.0313146652198568e-06,
      "loss": 0.0018,
      "step": 253420
    },
    {
      "epoch": 1.6254212992210575,
      "grad_norm": 0.0872546061873436,
      "learning_rate": 1.030974244340217e-06,
      "loss": 0.0013,
      "step": 253430
    },
    {
      "epoch": 1.6254854361148436,
      "grad_norm": 0.04959293082356453,
      "learning_rate": 1.030633873195307e-06,
      "loss": 0.0021,
      "step": 253440
    },
    {
      "epoch": 1.6255495730086296,
      "grad_norm": 0.15947020053863525,
      "learning_rate": 1.0302935517893897e-06,
      "loss": 0.0014,
      "step": 253450
    },
    {
      "epoch": 1.6256137099024157,
      "grad_norm": 0.08254580199718475,
      "learning_rate": 1.029953280126733e-06,
      "loss": 0.0005,
      "step": 253460
    },
    {
      "epoch": 1.6256778467962019,
      "grad_norm": 0.11837852001190186,
      "learning_rate": 1.0296130582115992e-06,
      "loss": 0.0012,
      "step": 253470
    },
    {
      "epoch": 1.6257419836899878,
      "grad_norm": 0.025882352143526077,
      "learning_rate": 1.029272886048251e-06,
      "loss": 0.001,
      "step": 253480
    },
    {
      "epoch": 1.625806120583774,
      "grad_norm": 0.036259740591049194,
      "learning_rate": 1.0289327636409502e-06,
      "loss": 0.0007,
      "step": 253490
    },
    {
      "epoch": 1.62587025747756,
      "grad_norm": 0.20507407188415527,
      "learning_rate": 1.028592690993961e-06,
      "loss": 0.002,
      "step": 253500
    },
    {
      "epoch": 1.6259343943713462,
      "grad_norm": 0.19284480810165405,
      "learning_rate": 1.028252668111544e-06,
      "loss": 0.0016,
      "step": 253510
    },
    {
      "epoch": 1.6259985312651324,
      "grad_norm": 0.05104293301701546,
      "learning_rate": 1.0279126949979596e-06,
      "loss": 0.0009,
      "step": 253520
    },
    {
      "epoch": 1.6260626681589185,
      "grad_norm": 0.06307905167341232,
      "learning_rate": 1.0275727716574663e-06,
      "loss": 0.0006,
      "step": 253530
    },
    {
      "epoch": 1.6261268050527045,
      "grad_norm": 0.026986945420503616,
      "learning_rate": 1.0272328980943269e-06,
      "loss": 0.0012,
      "step": 253540
    },
    {
      "epoch": 1.6261909419464906,
      "grad_norm": 0.19290678203105927,
      "learning_rate": 1.0268930743127976e-06,
      "loss": 0.0009,
      "step": 253550
    },
    {
      "epoch": 1.6262550788402765,
      "grad_norm": 0.003200072795152664,
      "learning_rate": 1.0265533003171379e-06,
      "loss": 0.005,
      "step": 253560
    },
    {
      "epoch": 1.6263192157340627,
      "grad_norm": 0.049872659146785736,
      "learning_rate": 1.026213576111605e-06,
      "loss": 0.001,
      "step": 253570
    },
    {
      "epoch": 1.6263833526278488,
      "grad_norm": 0.12477219849824905,
      "learning_rate": 1.0258739017004565e-06,
      "loss": 0.001,
      "step": 253580
    },
    {
      "epoch": 1.626447489521635,
      "grad_norm": 0.06833262741565704,
      "learning_rate": 1.0255342770879484e-06,
      "loss": 0.0007,
      "step": 253590
    },
    {
      "epoch": 1.6265116264154211,
      "grad_norm": 0.20073255896568298,
      "learning_rate": 1.0251947022783365e-06,
      "loss": 0.0016,
      "step": 253600
    },
    {
      "epoch": 1.6265757633092073,
      "grad_norm": 0.05578920617699623,
      "learning_rate": 1.024855177275874e-06,
      "loss": 0.0012,
      "step": 253610
    },
    {
      "epoch": 1.6266399002029934,
      "grad_norm": 0.2767143249511719,
      "learning_rate": 1.024515702084819e-06,
      "loss": 0.0015,
      "step": 253620
    },
    {
      "epoch": 1.6267040370967794,
      "grad_norm": 0.20888754725456238,
      "learning_rate": 1.0241762767094231e-06,
      "loss": 0.0018,
      "step": 253630
    },
    {
      "epoch": 1.6267681739905655,
      "grad_norm": 0.11888570338487625,
      "learning_rate": 1.0238369011539406e-06,
      "loss": 0.0009,
      "step": 253640
    },
    {
      "epoch": 1.6268323108843514,
      "grad_norm": 0.054734617471694946,
      "learning_rate": 1.0234975754226212e-06,
      "loss": 0.0011,
      "step": 253650
    },
    {
      "epoch": 1.6268964477781376,
      "grad_norm": 0.08102140575647354,
      "learning_rate": 1.0231582995197208e-06,
      "loss": 0.0006,
      "step": 253660
    },
    {
      "epoch": 1.6269605846719237,
      "grad_norm": 0.02955351211130619,
      "learning_rate": 1.0228190734494897e-06,
      "loss": 0.0012,
      "step": 253670
    },
    {
      "epoch": 1.6270247215657099,
      "grad_norm": 0.014702515676617622,
      "learning_rate": 1.0224798972161776e-06,
      "loss": 0.0006,
      "step": 253680
    },
    {
      "epoch": 1.627088858459496,
      "grad_norm": 0.03445766866207123,
      "learning_rate": 1.0221407708240333e-06,
      "loss": 0.0004,
      "step": 253690
    },
    {
      "epoch": 1.6271529953532822,
      "grad_norm": 0.10714350640773773,
      "learning_rate": 1.0218016942773102e-06,
      "loss": 0.0022,
      "step": 253700
    },
    {
      "epoch": 1.627217132247068,
      "grad_norm": 0.21742475032806396,
      "learning_rate": 1.0214626675802547e-06,
      "loss": 0.0008,
      "step": 253710
    },
    {
      "epoch": 1.6272812691408542,
      "grad_norm": 0.05936504900455475,
      "learning_rate": 1.0211236907371141e-06,
      "loss": 0.0011,
      "step": 253720
    },
    {
      "epoch": 1.6273454060346402,
      "grad_norm": 0.4004567265510559,
      "learning_rate": 1.0207847637521385e-06,
      "loss": 0.0041,
      "step": 253730
    },
    {
      "epoch": 1.6274095429284263,
      "grad_norm": 0.1915207952260971,
      "learning_rate": 1.0204458866295736e-06,
      "loss": 0.0027,
      "step": 253740
    },
    {
      "epoch": 1.6274736798222125,
      "grad_norm": 0.10918500274419785,
      "learning_rate": 1.0201070593736661e-06,
      "loss": 0.0011,
      "step": 253750
    },
    {
      "epoch": 1.6275378167159986,
      "grad_norm": 0.09684329479932785,
      "learning_rate": 1.01976828198866e-06,
      "loss": 0.0007,
      "step": 253760
    },
    {
      "epoch": 1.6276019536097848,
      "grad_norm": 0.10650018602609634,
      "learning_rate": 1.0194295544788036e-06,
      "loss": 0.0006,
      "step": 253770
    },
    {
      "epoch": 1.627666090503571,
      "grad_norm": 0.028103556483983994,
      "learning_rate": 1.0190908768483398e-06,
      "loss": 0.001,
      "step": 253780
    },
    {
      "epoch": 1.627730227397357,
      "grad_norm": 0.07362200319766998,
      "learning_rate": 1.0187522491015122e-06,
      "loss": 0.0007,
      "step": 253790
    },
    {
      "epoch": 1.627794364291143,
      "grad_norm": 0.11591235548257828,
      "learning_rate": 1.018413671242563e-06,
      "loss": 0.0011,
      "step": 253800
    },
    {
      "epoch": 1.6278585011849291,
      "grad_norm": 0.14773188531398773,
      "learning_rate": 1.0180751432757374e-06,
      "loss": 0.0013,
      "step": 253810
    },
    {
      "epoch": 1.627922638078715,
      "grad_norm": 0.12079880386590958,
      "learning_rate": 1.0177366652052762e-06,
      "loss": 0.0017,
      "step": 253820
    },
    {
      "epoch": 1.6279867749725012,
      "grad_norm": 0.048897046595811844,
      "learning_rate": 1.0173982370354192e-06,
      "loss": 0.0013,
      "step": 253830
    },
    {
      "epoch": 1.6280509118662874,
      "grad_norm": 0.0130966417491436,
      "learning_rate": 1.01705985877041e-06,
      "loss": 0.0018,
      "step": 253840
    },
    {
      "epoch": 1.6281150487600735,
      "grad_norm": 0.06076972186565399,
      "learning_rate": 1.0167215304144872e-06,
      "loss": 0.0011,
      "step": 253850
    },
    {
      "epoch": 1.6281791856538597,
      "grad_norm": 0.0678781121969223,
      "learning_rate": 1.0163832519718907e-06,
      "loss": 0.0009,
      "step": 253860
    },
    {
      "epoch": 1.6282433225476458,
      "grad_norm": 0.09457668662071228,
      "learning_rate": 1.0160450234468578e-06,
      "loss": 0.0007,
      "step": 253870
    },
    {
      "epoch": 1.6283074594414317,
      "grad_norm": 0.13398870825767517,
      "learning_rate": 1.015706844843629e-06,
      "loss": 0.0017,
      "step": 253880
    },
    {
      "epoch": 1.6283715963352179,
      "grad_norm": 0.13375312089920044,
      "learning_rate": 1.015368716166441e-06,
      "loss": 0.0019,
      "step": 253890
    },
    {
      "epoch": 1.628435733229004,
      "grad_norm": 0.10975903272628784,
      "learning_rate": 1.0150306374195313e-06,
      "loss": 0.0023,
      "step": 253900
    },
    {
      "epoch": 1.62849987012279,
      "grad_norm": 0.13212823867797852,
      "learning_rate": 1.0146926086071334e-06,
      "loss": 0.0013,
      "step": 253910
    },
    {
      "epoch": 1.628564007016576,
      "grad_norm": 0.05868707224726677,
      "learning_rate": 1.0143546297334878e-06,
      "loss": 0.0006,
      "step": 253920
    },
    {
      "epoch": 1.6286281439103623,
      "grad_norm": 0.10828299075365067,
      "learning_rate": 1.0140167008028267e-06,
      "loss": 0.0005,
      "step": 253930
    },
    {
      "epoch": 1.6286922808041484,
      "grad_norm": 0.09012256562709808,
      "learning_rate": 1.013678821819385e-06,
      "loss": 0.0009,
      "step": 253940
    },
    {
      "epoch": 1.6287564176979346,
      "grad_norm": 0.1700572669506073,
      "learning_rate": 1.0133409927873954e-06,
      "loss": 0.0017,
      "step": 253950
    },
    {
      "epoch": 1.6288205545917207,
      "grad_norm": 0.0477653443813324,
      "learning_rate": 1.0130032137110935e-06,
      "loss": 0.001,
      "step": 253960
    },
    {
      "epoch": 1.6288846914855066,
      "grad_norm": 0.12062278389930725,
      "learning_rate": 1.012665484594711e-06,
      "loss": 0.0035,
      "step": 253970
    },
    {
      "epoch": 1.6289488283792928,
      "grad_norm": 0.10053012520074844,
      "learning_rate": 1.0123278054424784e-06,
      "loss": 0.001,
      "step": 253980
    },
    {
      "epoch": 1.6290129652730787,
      "grad_norm": 0.1188049167394638,
      "learning_rate": 1.0119901762586298e-06,
      "loss": 0.0016,
      "step": 253990
    },
    {
      "epoch": 1.6290771021668649,
      "grad_norm": 0.03905399143695831,
      "learning_rate": 1.0116525970473945e-06,
      "loss": 0.0008,
      "step": 254000
    },
    {
      "epoch": 1.629141239060651,
      "grad_norm": 0.026444246992468834,
      "learning_rate": 1.0113150678130024e-06,
      "loss": 0.0006,
      "step": 254010
    },
    {
      "epoch": 1.6292053759544372,
      "grad_norm": 0.08806001394987106,
      "learning_rate": 1.0109775885596818e-06,
      "loss": 0.0008,
      "step": 254020
    },
    {
      "epoch": 1.6292695128482233,
      "grad_norm": 0.21622851490974426,
      "learning_rate": 1.0106401592916644e-06,
      "loss": 0.0012,
      "step": 254030
    },
    {
      "epoch": 1.6293336497420094,
      "grad_norm": 0.09526892006397247,
      "learning_rate": 1.0103027800131765e-06,
      "loss": 0.0009,
      "step": 254040
    },
    {
      "epoch": 1.6293977866357956,
      "grad_norm": 0.17161071300506592,
      "learning_rate": 1.0099654507284467e-06,
      "loss": 0.0014,
      "step": 254050
    },
    {
      "epoch": 1.6294619235295815,
      "grad_norm": 0.06431101262569427,
      "learning_rate": 1.0096281714417e-06,
      "loss": 0.0008,
      "step": 254060
    },
    {
      "epoch": 1.6295260604233677,
      "grad_norm": 0.04051023721694946,
      "learning_rate": 1.009290942157165e-06,
      "loss": 0.0005,
      "step": 254070
    },
    {
      "epoch": 1.6295901973171536,
      "grad_norm": 0.12363533675670624,
      "learning_rate": 1.0089537628790675e-06,
      "loss": 0.0037,
      "step": 254080
    },
    {
      "epoch": 1.6296543342109397,
      "grad_norm": 0.1416805237531662,
      "learning_rate": 1.008616633611631e-06,
      "loss": 0.0064,
      "step": 254090
    },
    {
      "epoch": 1.629718471104726,
      "grad_norm": 0.028971588239073753,
      "learning_rate": 1.0082795543590796e-06,
      "loss": 0.0007,
      "step": 254100
    },
    {
      "epoch": 1.629782607998512,
      "grad_norm": 0.14429962635040283,
      "learning_rate": 1.0079425251256397e-06,
      "loss": 0.0009,
      "step": 254110
    },
    {
      "epoch": 1.6298467448922982,
      "grad_norm": 0.1125936433672905,
      "learning_rate": 1.0076055459155327e-06,
      "loss": 0.0025,
      "step": 254120
    },
    {
      "epoch": 1.6299108817860843,
      "grad_norm": 0.15618206560611725,
      "learning_rate": 1.0072686167329803e-06,
      "loss": 0.0008,
      "step": 254130
    },
    {
      "epoch": 1.6299750186798703,
      "grad_norm": 0.010613231919705868,
      "learning_rate": 1.0069317375822073e-06,
      "loss": 0.0013,
      "step": 254140
    },
    {
      "epoch": 1.6300391555736564,
      "grad_norm": 0.020671576261520386,
      "learning_rate": 1.0065949084674332e-06,
      "loss": 0.0008,
      "step": 254150
    },
    {
      "epoch": 1.6301032924674423,
      "grad_norm": 0.13317108154296875,
      "learning_rate": 1.0062581293928791e-06,
      "loss": 0.002,
      "step": 254160
    },
    {
      "epoch": 1.6301674293612285,
      "grad_norm": 0.1296268254518509,
      "learning_rate": 1.0059214003627638e-06,
      "loss": 0.0014,
      "step": 254170
    },
    {
      "epoch": 1.6302315662550146,
      "grad_norm": 0.3353487551212311,
      "learning_rate": 1.0055847213813085e-06,
      "loss": 0.0017,
      "step": 254180
    },
    {
      "epoch": 1.6302957031488008,
      "grad_norm": 0.025075187906622887,
      "learning_rate": 1.005248092452732e-06,
      "loss": 0.0005,
      "step": 254190
    },
    {
      "epoch": 1.630359840042587,
      "grad_norm": 0.0145181929692626,
      "learning_rate": 1.0049115135812514e-06,
      "loss": 0.0016,
      "step": 254200
    },
    {
      "epoch": 1.630423976936373,
      "grad_norm": 0.07771790772676468,
      "learning_rate": 1.004574984771084e-06,
      "loss": 0.0014,
      "step": 254210
    },
    {
      "epoch": 1.6304881138301592,
      "grad_norm": 0.01438713725656271,
      "learning_rate": 1.0042385060264486e-06,
      "loss": 0.0012,
      "step": 254220
    },
    {
      "epoch": 1.6305522507239452,
      "grad_norm": 0.008997242897748947,
      "learning_rate": 1.0039020773515607e-06,
      "loss": 0.0019,
      "step": 254230
    },
    {
      "epoch": 1.6306163876177313,
      "grad_norm": 0.004172218032181263,
      "learning_rate": 1.0035656987506354e-06,
      "loss": 0.0015,
      "step": 254240
    },
    {
      "epoch": 1.6306805245115172,
      "grad_norm": 0.10217877477407455,
      "learning_rate": 1.0032293702278866e-06,
      "loss": 0.0013,
      "step": 254250
    },
    {
      "epoch": 1.6307446614053034,
      "grad_norm": 0.06913498789072037,
      "learning_rate": 1.0028930917875323e-06,
      "loss": 0.001,
      "step": 254260
    },
    {
      "epoch": 1.6308087982990895,
      "grad_norm": 0.11332852393388748,
      "learning_rate": 1.0025568634337834e-06,
      "loss": 0.0013,
      "step": 254270
    },
    {
      "epoch": 1.6308729351928757,
      "grad_norm": 0.09933172911405563,
      "learning_rate": 1.0022206851708532e-06,
      "loss": 0.0015,
      "step": 254280
    },
    {
      "epoch": 1.6309370720866618,
      "grad_norm": 0.05286799743771553,
      "learning_rate": 1.0018845570029562e-06,
      "loss": 0.0006,
      "step": 254290
    },
    {
      "epoch": 1.631001208980448,
      "grad_norm": 0.03247305378317833,
      "learning_rate": 1.0015484789343027e-06,
      "loss": 0.0014,
      "step": 254300
    },
    {
      "epoch": 1.6310653458742341,
      "grad_norm": 0.17723588645458221,
      "learning_rate": 1.0012124509691034e-06,
      "loss": 0.0016,
      "step": 254310
    },
    {
      "epoch": 1.63112948276802,
      "grad_norm": 0.07294376939535141,
      "learning_rate": 1.0008764731115717e-06,
      "loss": 0.0025,
      "step": 254320
    },
    {
      "epoch": 1.6311936196618062,
      "grad_norm": 0.21547554433345795,
      "learning_rate": 1.000540545365914e-06,
      "loss": 0.0008,
      "step": 254330
    },
    {
      "epoch": 1.6312577565555921,
      "grad_norm": 0.06939996778964996,
      "learning_rate": 1.000204667736343e-06,
      "loss": 0.0014,
      "step": 254340
    },
    {
      "epoch": 1.6313218934493783,
      "grad_norm": 0.14365023374557495,
      "learning_rate": 9.998688402270667e-07,
      "loss": 0.0012,
      "step": 254350
    },
    {
      "epoch": 1.6313860303431644,
      "grad_norm": 0.004856092389672995,
      "learning_rate": 9.995330628422923e-07,
      "loss": 0.0008,
      "step": 254360
    },
    {
      "epoch": 1.6314501672369506,
      "grad_norm": 0.046167537569999695,
      "learning_rate": 9.99197335586226e-07,
      "loss": 0.0011,
      "step": 254370
    },
    {
      "epoch": 1.6315143041307367,
      "grad_norm": 0.15383437275886536,
      "learning_rate": 9.988616584630783e-07,
      "loss": 0.0025,
      "step": 254380
    },
    {
      "epoch": 1.6315784410245229,
      "grad_norm": 0.016113227233290672,
      "learning_rate": 9.985260314770533e-07,
      "loss": 0.0008,
      "step": 254390
    },
    {
      "epoch": 1.6316425779183088,
      "grad_norm": 0.19796684384346008,
      "learning_rate": 9.981904546323555e-07,
      "loss": 0.0031,
      "step": 254400
    },
    {
      "epoch": 1.631706714812095,
      "grad_norm": 0.10615178197622299,
      "learning_rate": 9.978549279331934e-07,
      "loss": 0.0006,
      "step": 254410
    },
    {
      "epoch": 1.6317708517058809,
      "grad_norm": 0.020054157823324203,
      "learning_rate": 9.975194513837687e-07,
      "loss": 0.0011,
      "step": 254420
    },
    {
      "epoch": 1.631834988599667,
      "grad_norm": 0.06269645690917969,
      "learning_rate": 9.971840249882859e-07,
      "loss": 0.0021,
      "step": 254430
    },
    {
      "epoch": 1.6318991254934532,
      "grad_norm": 0.1866074651479721,
      "learning_rate": 9.968486487509466e-07,
      "loss": 0.0013,
      "step": 254440
    },
    {
      "epoch": 1.6319632623872393,
      "grad_norm": 0.12030469626188278,
      "learning_rate": 9.965133226759566e-07,
      "loss": 0.0014,
      "step": 254450
    },
    {
      "epoch": 1.6320273992810255,
      "grad_norm": 0.10025019943714142,
      "learning_rate": 9.961780467675153e-07,
      "loss": 0.001,
      "step": 254460
    },
    {
      "epoch": 1.6320915361748116,
      "grad_norm": 0.07710816711187363,
      "learning_rate": 9.95842821029825e-07,
      "loss": 0.0006,
      "step": 254470
    },
    {
      "epoch": 1.6321556730685978,
      "grad_norm": 0.08118700981140137,
      "learning_rate": 9.955076454670842e-07,
      "loss": 0.0011,
      "step": 254480
    },
    {
      "epoch": 1.6322198099623837,
      "grad_norm": 0.001708224997855723,
      "learning_rate": 9.951725200834961e-07,
      "loss": 0.0008,
      "step": 254490
    },
    {
      "epoch": 1.6322839468561698,
      "grad_norm": 0.1567011922597885,
      "learning_rate": 9.94837444883258e-07,
      "loss": 0.0009,
      "step": 254500
    },
    {
      "epoch": 1.6323480837499558,
      "grad_norm": 0.0870925784111023,
      "learning_rate": 9.9450241987057e-07,
      "loss": 0.0015,
      "step": 254510
    },
    {
      "epoch": 1.632412220643742,
      "grad_norm": 0.05215726047754288,
      "learning_rate": 9.941674450496276e-07,
      "loss": 0.0006,
      "step": 254520
    },
    {
      "epoch": 1.632476357537528,
      "grad_norm": 0.07498734444379807,
      "learning_rate": 9.938325204246313e-07,
      "loss": 0.001,
      "step": 254530
    },
    {
      "epoch": 1.6325404944313142,
      "grad_norm": 0.10156609863042831,
      "learning_rate": 9.934976459997768e-07,
      "loss": 0.0022,
      "step": 254540
    },
    {
      "epoch": 1.6326046313251004,
      "grad_norm": 0.01245181169360876,
      "learning_rate": 9.931628217792593e-07,
      "loss": 0.0014,
      "step": 254550
    },
    {
      "epoch": 1.6326687682188865,
      "grad_norm": 0.12639538943767548,
      "learning_rate": 9.928280477672764e-07,
      "loss": 0.0009,
      "step": 254560
    },
    {
      "epoch": 1.6327329051126724,
      "grad_norm": 0.01934858411550522,
      "learning_rate": 9.924933239680213e-07,
      "loss": 0.0026,
      "step": 254570
    },
    {
      "epoch": 1.6327970420064586,
      "grad_norm": 0.006678812671452761,
      "learning_rate": 9.921586503856894e-07,
      "loss": 0.001,
      "step": 254580
    },
    {
      "epoch": 1.6328611789002447,
      "grad_norm": 0.14333570003509521,
      "learning_rate": 9.918240270244727e-07,
      "loss": 0.0022,
      "step": 254590
    },
    {
      "epoch": 1.6329253157940307,
      "grad_norm": 0.06659631431102753,
      "learning_rate": 9.914894538885671e-07,
      "loss": 0.0011,
      "step": 254600
    },
    {
      "epoch": 1.6329894526878168,
      "grad_norm": 0.020249219611287117,
      "learning_rate": 9.911549309821632e-07,
      "loss": 0.0007,
      "step": 254610
    },
    {
      "epoch": 1.633053589581603,
      "grad_norm": 0.03794463723897934,
      "learning_rate": 9.908204583094533e-07,
      "loss": 0.0008,
      "step": 254620
    },
    {
      "epoch": 1.633117726475389,
      "grad_norm": 0.07175882905721664,
      "learning_rate": 9.904860358746265e-07,
      "loss": 0.0028,
      "step": 254630
    },
    {
      "epoch": 1.6331818633691753,
      "grad_norm": 0.00746492063626647,
      "learning_rate": 9.90151663681877e-07,
      "loss": 0.0006,
      "step": 254640
    },
    {
      "epoch": 1.6332460002629614,
      "grad_norm": 0.019557328894734383,
      "learning_rate": 9.898173417353929e-07,
      "loss": 0.001,
      "step": 254650
    },
    {
      "epoch": 1.6333101371567473,
      "grad_norm": 0.13451708853244781,
      "learning_rate": 9.894830700393637e-07,
      "loss": 0.0017,
      "step": 254660
    },
    {
      "epoch": 1.6333742740505335,
      "grad_norm": 0.0878053605556488,
      "learning_rate": 9.891488485979762e-07,
      "loss": 0.0014,
      "step": 254670
    },
    {
      "epoch": 1.6334384109443194,
      "grad_norm": 0.27910155057907104,
      "learning_rate": 9.888146774154217e-07,
      "loss": 0.0012,
      "step": 254680
    },
    {
      "epoch": 1.6335025478381056,
      "grad_norm": 0.039628490805625916,
      "learning_rate": 9.88480556495886e-07,
      "loss": 0.0015,
      "step": 254690
    },
    {
      "epoch": 1.6335666847318917,
      "grad_norm": 0.10816241055727005,
      "learning_rate": 9.881464858435547e-07,
      "loss": 0.0006,
      "step": 254700
    },
    {
      "epoch": 1.6336308216256779,
      "grad_norm": 0.035373102873563766,
      "learning_rate": 9.878124654626165e-07,
      "loss": 0.0025,
      "step": 254710
    },
    {
      "epoch": 1.633694958519464,
      "grad_norm": 0.08186127990484238,
      "learning_rate": 9.874784953572553e-07,
      "loss": 0.0008,
      "step": 254720
    },
    {
      "epoch": 1.6337590954132502,
      "grad_norm": 0.004696046933531761,
      "learning_rate": 9.871445755316562e-07,
      "loss": 0.0011,
      "step": 254730
    },
    {
      "epoch": 1.6338232323070363,
      "grad_norm": 0.054276540875434875,
      "learning_rate": 9.868107059900024e-07,
      "loss": 0.0013,
      "step": 254740
    },
    {
      "epoch": 1.6338873692008222,
      "grad_norm": 0.03867164999246597,
      "learning_rate": 9.8647688673648e-07,
      "loss": 0.001,
      "step": 254750
    },
    {
      "epoch": 1.6339515060946084,
      "grad_norm": 0.07352355122566223,
      "learning_rate": 9.861431177752706e-07,
      "loss": 0.0012,
      "step": 254760
    },
    {
      "epoch": 1.6340156429883943,
      "grad_norm": 0.05755196884274483,
      "learning_rate": 9.858093991105567e-07,
      "loss": 0.0009,
      "step": 254770
    },
    {
      "epoch": 1.6340797798821804,
      "grad_norm": 0.08385112136602402,
      "learning_rate": 9.854757307465185e-07,
      "loss": 0.0011,
      "step": 254780
    },
    {
      "epoch": 1.6341439167759666,
      "grad_norm": 0.10787700861692429,
      "learning_rate": 9.851421126873395e-07,
      "loss": 0.0018,
      "step": 254790
    },
    {
      "epoch": 1.6342080536697527,
      "grad_norm": 0.051003411412239075,
      "learning_rate": 9.848085449371998e-07,
      "loss": 0.0012,
      "step": 254800
    },
    {
      "epoch": 1.634272190563539,
      "grad_norm": 0.28282269835472107,
      "learning_rate": 9.844750275002784e-07,
      "loss": 0.0013,
      "step": 254810
    },
    {
      "epoch": 1.634336327457325,
      "grad_norm": 0.059521906077861786,
      "learning_rate": 9.841415603807536e-07,
      "loss": 0.001,
      "step": 254820
    },
    {
      "epoch": 1.634400464351111,
      "grad_norm": 0.11180324852466583,
      "learning_rate": 9.838081435828061e-07,
      "loss": 0.0006,
      "step": 254830
    },
    {
      "epoch": 1.6344646012448971,
      "grad_norm": 0.02087988890707493,
      "learning_rate": 9.834747771106128e-07,
      "loss": 0.001,
      "step": 254840
    },
    {
      "epoch": 1.634528738138683,
      "grad_norm": 0.042987458407878876,
      "learning_rate": 9.831414609683503e-07,
      "loss": 0.0013,
      "step": 254850
    },
    {
      "epoch": 1.6345928750324692,
      "grad_norm": 0.08517789840698242,
      "learning_rate": 9.82808195160197e-07,
      "loss": 0.0013,
      "step": 254860
    },
    {
      "epoch": 1.6346570119262553,
      "grad_norm": 0.07343614101409912,
      "learning_rate": 9.824749796903281e-07,
      "loss": 0.001,
      "step": 254870
    },
    {
      "epoch": 1.6347211488200415,
      "grad_norm": 0.04992000758647919,
      "learning_rate": 9.821418145629196e-07,
      "loss": 0.0008,
      "step": 254880
    },
    {
      "epoch": 1.6347852857138276,
      "grad_norm": 0.10211850702762604,
      "learning_rate": 9.818086997821436e-07,
      "loss": 0.0007,
      "step": 254890
    },
    {
      "epoch": 1.6348494226076138,
      "grad_norm": 0.0167512446641922,
      "learning_rate": 9.814756353521782e-07,
      "loss": 0.0013,
      "step": 254900
    },
    {
      "epoch": 1.6349135595014,
      "grad_norm": 0.14388491213321686,
      "learning_rate": 9.811426212771946e-07,
      "loss": 0.0006,
      "step": 254910
    },
    {
      "epoch": 1.6349776963951859,
      "grad_norm": 0.021333513781428337,
      "learning_rate": 9.808096575613668e-07,
      "loss": 0.0006,
      "step": 254920
    },
    {
      "epoch": 1.635041833288972,
      "grad_norm": 0.0028631179593503475,
      "learning_rate": 9.804767442088647e-07,
      "loss": 0.001,
      "step": 254930
    },
    {
      "epoch": 1.635105970182758,
      "grad_norm": 0.052888594567775726,
      "learning_rate": 9.801438812238627e-07,
      "loss": 0.0007,
      "step": 254940
    },
    {
      "epoch": 1.635170107076544,
      "grad_norm": 0.133683443069458,
      "learning_rate": 9.798110686105317e-07,
      "loss": 0.0009,
      "step": 254950
    },
    {
      "epoch": 1.6352342439703302,
      "grad_norm": 0.07119552791118622,
      "learning_rate": 9.794783063730407e-07,
      "loss": 0.0009,
      "step": 254960
    },
    {
      "epoch": 1.6352983808641164,
      "grad_norm": 0.11453062295913696,
      "learning_rate": 9.791455945155588e-07,
      "loss": 0.0008,
      "step": 254970
    },
    {
      "epoch": 1.6353625177579025,
      "grad_norm": 0.04229366406798363,
      "learning_rate": 9.788129330422568e-07,
      "loss": 0.0008,
      "step": 254980
    },
    {
      "epoch": 1.6354266546516887,
      "grad_norm": 0.1479012817144394,
      "learning_rate": 9.784803219573036e-07,
      "loss": 0.0022,
      "step": 254990
    },
    {
      "epoch": 1.6354907915454746,
      "grad_norm": 0.03219074010848999,
      "learning_rate": 9.781477612648644e-07,
      "loss": 0.0007,
      "step": 255000
    },
    {
      "epoch": 1.6355549284392608,
      "grad_norm": 0.05763757973909378,
      "learning_rate": 9.778152509691097e-07,
      "loss": 0.0006,
      "step": 255010
    },
    {
      "epoch": 1.635619065333047,
      "grad_norm": 0.04857158660888672,
      "learning_rate": 9.77482791074204e-07,
      "loss": 0.0008,
      "step": 255020
    },
    {
      "epoch": 1.6356832022268328,
      "grad_norm": 0.08321210741996765,
      "learning_rate": 9.771503815843143e-07,
      "loss": 0.0015,
      "step": 255030
    },
    {
      "epoch": 1.635747339120619,
      "grad_norm": 0.01625431887805462,
      "learning_rate": 9.76818022503604e-07,
      "loss": 0.0012,
      "step": 255040
    },
    {
      "epoch": 1.6358114760144051,
      "grad_norm": 0.1378721445798874,
      "learning_rate": 9.764857138362405e-07,
      "loss": 0.0012,
      "step": 255050
    },
    {
      "epoch": 1.6358756129081913,
      "grad_norm": 0.036976706236600876,
      "learning_rate": 9.76153455586386e-07,
      "loss": 0.0012,
      "step": 255060
    },
    {
      "epoch": 1.6359397498019774,
      "grad_norm": 0.005930387880653143,
      "learning_rate": 9.758212477582052e-07,
      "loss": 0.0005,
      "step": 255070
    },
    {
      "epoch": 1.6360038866957636,
      "grad_norm": 0.09951237589120865,
      "learning_rate": 9.75489090355859e-07,
      "loss": 0.001,
      "step": 255080
    },
    {
      "epoch": 1.6360680235895495,
      "grad_norm": 0.005069872364401817,
      "learning_rate": 9.7515698338351e-07,
      "loss": 0.0007,
      "step": 255090
    },
    {
      "epoch": 1.6361321604833357,
      "grad_norm": 0.01948101818561554,
      "learning_rate": 9.748249268453225e-07,
      "loss": 0.001,
      "step": 255100
    },
    {
      "epoch": 1.6361962973771216,
      "grad_norm": 0.0026461135130375624,
      "learning_rate": 9.744929207454552e-07,
      "loss": 0.0017,
      "step": 255110
    },
    {
      "epoch": 1.6362604342709077,
      "grad_norm": 0.05868653580546379,
      "learning_rate": 9.741609650880674e-07,
      "loss": 0.0016,
      "step": 255120
    },
    {
      "epoch": 1.6363245711646939,
      "grad_norm": 0.2810801565647125,
      "learning_rate": 9.738290598773209e-07,
      "loss": 0.0011,
      "step": 255130
    },
    {
      "epoch": 1.63638870805848,
      "grad_norm": 0.01811443269252777,
      "learning_rate": 9.734972051173742e-07,
      "loss": 0.0017,
      "step": 255140
    },
    {
      "epoch": 1.6364528449522662,
      "grad_norm": 0.15347135066986084,
      "learning_rate": 9.731654008123848e-07,
      "loss": 0.001,
      "step": 255150
    },
    {
      "epoch": 1.6365169818460523,
      "grad_norm": 0.041657716035842896,
      "learning_rate": 9.728336469665096e-07,
      "loss": 0.0019,
      "step": 255160
    },
    {
      "epoch": 1.6365811187398385,
      "grad_norm": 0.05868009477853775,
      "learning_rate": 9.725019435839085e-07,
      "loss": 0.0015,
      "step": 255170
    },
    {
      "epoch": 1.6366452556336244,
      "grad_norm": 0.1378551423549652,
      "learning_rate": 9.721702906687363e-07,
      "loss": 0.0009,
      "step": 255180
    },
    {
      "epoch": 1.6367093925274105,
      "grad_norm": 0.015106326900422573,
      "learning_rate": 9.71838688225149e-07,
      "loss": 0.001,
      "step": 255190
    },
    {
      "epoch": 1.6367735294211965,
      "grad_norm": 0.12061250954866409,
      "learning_rate": 9.715071362573002e-07,
      "loss": 0.0011,
      "step": 255200
    },
    {
      "epoch": 1.6368376663149826,
      "grad_norm": 0.05114997923374176,
      "learning_rate": 9.71175634769348e-07,
      "loss": 0.0009,
      "step": 255210
    },
    {
      "epoch": 1.6369018032087688,
      "grad_norm": 0.009737277403473854,
      "learning_rate": 9.708441837654441e-07,
      "loss": 0.0013,
      "step": 255220
    },
    {
      "epoch": 1.636965940102555,
      "grad_norm": 0.08145773410797119,
      "learning_rate": 9.705127832497423e-07,
      "loss": 0.0009,
      "step": 255230
    },
    {
      "epoch": 1.637030076996341,
      "grad_norm": 0.3063494861125946,
      "learning_rate": 9.701814332263943e-07,
      "loss": 0.0017,
      "step": 255240
    },
    {
      "epoch": 1.6370942138901272,
      "grad_norm": 0.09998942911624908,
      "learning_rate": 9.698501336995536e-07,
      "loss": 0.002,
      "step": 255250
    },
    {
      "epoch": 1.6371583507839131,
      "grad_norm": 0.11387070268392563,
      "learning_rate": 9.695188846733712e-07,
      "loss": 0.0015,
      "step": 255260
    },
    {
      "epoch": 1.6372224876776993,
      "grad_norm": 0.043187957257032394,
      "learning_rate": 9.691876861519967e-07,
      "loss": 0.0024,
      "step": 255270
    },
    {
      "epoch": 1.6372866245714852,
      "grad_norm": 0.12112016975879669,
      "learning_rate": 9.68856538139582e-07,
      "loss": 0.0012,
      "step": 255280
    },
    {
      "epoch": 1.6373507614652714,
      "grad_norm": 0.03679222613573074,
      "learning_rate": 9.685254406402767e-07,
      "loss": 0.0009,
      "step": 255290
    },
    {
      "epoch": 1.6374148983590575,
      "grad_norm": 0.033384256064891815,
      "learning_rate": 9.681943936582282e-07,
      "loss": 0.0008,
      "step": 255300
    },
    {
      "epoch": 1.6374790352528437,
      "grad_norm": 0.07566209882497787,
      "learning_rate": 9.678633971975842e-07,
      "loss": 0.0013,
      "step": 255310
    },
    {
      "epoch": 1.6375431721466298,
      "grad_norm": 0.1315520852804184,
      "learning_rate": 9.67532451262495e-07,
      "loss": 0.0024,
      "step": 255320
    },
    {
      "epoch": 1.637607309040416,
      "grad_norm": 0.042651720345020294,
      "learning_rate": 9.672015558571063e-07,
      "loss": 0.0013,
      "step": 255330
    },
    {
      "epoch": 1.637671445934202,
      "grad_norm": 0.11850108206272125,
      "learning_rate": 9.668707109855646e-07,
      "loss": 0.0015,
      "step": 255340
    },
    {
      "epoch": 1.637735582827988,
      "grad_norm": 0.06423966586589813,
      "learning_rate": 9.665399166520135e-07,
      "loss": 0.0006,
      "step": 255350
    },
    {
      "epoch": 1.6377997197217742,
      "grad_norm": 0.04013515263795853,
      "learning_rate": 9.66209172860601e-07,
      "loss": 0.0007,
      "step": 255360
    },
    {
      "epoch": 1.63786385661556,
      "grad_norm": 0.1369294822216034,
      "learning_rate": 9.658784796154714e-07,
      "loss": 0.0017,
      "step": 255370
    },
    {
      "epoch": 1.6379279935093463,
      "grad_norm": 0.04467201605439186,
      "learning_rate": 9.655478369207665e-07,
      "loss": 0.0018,
      "step": 255380
    },
    {
      "epoch": 1.6379921304031324,
      "grad_norm": 0.01031948160380125,
      "learning_rate": 9.652172447806302e-07,
      "loss": 0.001,
      "step": 255390
    },
    {
      "epoch": 1.6380562672969186,
      "grad_norm": 0.03769548982381821,
      "learning_rate": 9.64886703199206e-07,
      "loss": 0.0009,
      "step": 255400
    },
    {
      "epoch": 1.6381204041907047,
      "grad_norm": 0.02971324324607849,
      "learning_rate": 9.645562121806357e-07,
      "loss": 0.0011,
      "step": 255410
    },
    {
      "epoch": 1.6381845410844909,
      "grad_norm": 0.08204713463783264,
      "learning_rate": 9.642257717290586e-07,
      "loss": 0.0017,
      "step": 255420
    },
    {
      "epoch": 1.6382486779782768,
      "grad_norm": 0.0884409248828888,
      "learning_rate": 9.638953818486185e-07,
      "loss": 0.0009,
      "step": 255430
    },
    {
      "epoch": 1.638312814872063,
      "grad_norm": 0.029870890080928802,
      "learning_rate": 9.635650425434534e-07,
      "loss": 0.001,
      "step": 255440
    },
    {
      "epoch": 1.638376951765849,
      "grad_norm": 0.05527724698185921,
      "learning_rate": 9.632347538177033e-07,
      "loss": 0.0015,
      "step": 255450
    },
    {
      "epoch": 1.638441088659635,
      "grad_norm": 0.013997524045407772,
      "learning_rate": 9.629045156755057e-07,
      "loss": 0.0013,
      "step": 255460
    },
    {
      "epoch": 1.6385052255534212,
      "grad_norm": 0.04378235340118408,
      "learning_rate": 9.62574328121001e-07,
      "loss": 0.0007,
      "step": 255470
    },
    {
      "epoch": 1.6385693624472073,
      "grad_norm": 0.014343179762363434,
      "learning_rate": 9.622441911583253e-07,
      "loss": 0.0012,
      "step": 255480
    },
    {
      "epoch": 1.6386334993409934,
      "grad_norm": 0.05192306637763977,
      "learning_rate": 9.619141047916158e-07,
      "loss": 0.0007,
      "step": 255490
    },
    {
      "epoch": 1.6386976362347796,
      "grad_norm": 0.016144398599863052,
      "learning_rate": 9.615840690250073e-07,
      "loss": 0.0012,
      "step": 255500
    },
    {
      "epoch": 1.6387617731285657,
      "grad_norm": 0.04556126892566681,
      "learning_rate": 9.612540838626378e-07,
      "loss": 0.0019,
      "step": 255510
    },
    {
      "epoch": 1.6388259100223517,
      "grad_norm": 0.015942873433232307,
      "learning_rate": 9.609241493086413e-07,
      "loss": 0.0014,
      "step": 255520
    },
    {
      "epoch": 1.6388900469161378,
      "grad_norm": 0.049452684819698334,
      "learning_rate": 9.605942653671503e-07,
      "loss": 0.0008,
      "step": 255530
    },
    {
      "epoch": 1.6389541838099237,
      "grad_norm": 0.25146812200546265,
      "learning_rate": 9.602644320423011e-07,
      "loss": 0.0013,
      "step": 255540
    },
    {
      "epoch": 1.63901832070371,
      "grad_norm": 0.11109595745801926,
      "learning_rate": 9.59934649338226e-07,
      "loss": 0.0012,
      "step": 255550
    },
    {
      "epoch": 1.639082457597496,
      "grad_norm": 0.09703154861927032,
      "learning_rate": 9.596049172590577e-07,
      "loss": 0.0009,
      "step": 255560
    },
    {
      "epoch": 1.6391465944912822,
      "grad_norm": 0.07881749421358109,
      "learning_rate": 9.592752358089253e-07,
      "loss": 0.0018,
      "step": 255570
    },
    {
      "epoch": 1.6392107313850683,
      "grad_norm": 0.02280495874583721,
      "learning_rate": 9.589456049919638e-07,
      "loss": 0.0006,
      "step": 255580
    },
    {
      "epoch": 1.6392748682788545,
      "grad_norm": 0.070301353931427,
      "learning_rate": 9.586160248123015e-07,
      "loss": 0.0015,
      "step": 255590
    },
    {
      "epoch": 1.6393390051726406,
      "grad_norm": 0.034235622733831406,
      "learning_rate": 9.582864952740694e-07,
      "loss": 0.0012,
      "step": 255600
    },
    {
      "epoch": 1.6394031420664266,
      "grad_norm": 0.010166875086724758,
      "learning_rate": 9.579570163813945e-07,
      "loss": 0.0004,
      "step": 255610
    },
    {
      "epoch": 1.6394672789602127,
      "grad_norm": 0.06755946576595306,
      "learning_rate": 9.576275881384084e-07,
      "loss": 0.0011,
      "step": 255620
    },
    {
      "epoch": 1.6395314158539986,
      "grad_norm": 0.022154757753014565,
      "learning_rate": 9.572982105492373e-07,
      "loss": 0.0009,
      "step": 255630
    },
    {
      "epoch": 1.6395955527477848,
      "grad_norm": 0.12332911789417267,
      "learning_rate": 9.56968883618009e-07,
      "loss": 0.001,
      "step": 255640
    },
    {
      "epoch": 1.639659689641571,
      "grad_norm": 0.02179088443517685,
      "learning_rate": 9.566396073488487e-07,
      "loss": 0.0017,
      "step": 255650
    },
    {
      "epoch": 1.639723826535357,
      "grad_norm": 0.12347091734409332,
      "learning_rate": 9.563103817458857e-07,
      "loss": 0.0028,
      "step": 255660
    },
    {
      "epoch": 1.6397879634291432,
      "grad_norm": 0.010825304314494133,
      "learning_rate": 9.559812068132434e-07,
      "loss": 0.0012,
      "step": 255670
    },
    {
      "epoch": 1.6398521003229294,
      "grad_norm": 0.1055217906832695,
      "learning_rate": 9.556520825550453e-07,
      "loss": 0.0012,
      "step": 255680
    },
    {
      "epoch": 1.6399162372167153,
      "grad_norm": 0.02588566020131111,
      "learning_rate": 9.553230089754184e-07,
      "loss": 0.0013,
      "step": 255690
    },
    {
      "epoch": 1.6399803741105015,
      "grad_norm": 0.05414152890443802,
      "learning_rate": 9.549939860784852e-07,
      "loss": 0.0013,
      "step": 255700
    },
    {
      "epoch": 1.6400445110042874,
      "grad_norm": 0.11356854438781738,
      "learning_rate": 9.546650138683678e-07,
      "loss": 0.0014,
      "step": 255710
    },
    {
      "epoch": 1.6401086478980735,
      "grad_norm": 0.11140979826450348,
      "learning_rate": 9.543360923491878e-07,
      "loss": 0.0019,
      "step": 255720
    },
    {
      "epoch": 1.6401727847918597,
      "grad_norm": 0.03561192378401756,
      "learning_rate": 9.540072215250689e-07,
      "loss": 0.0008,
      "step": 255730
    },
    {
      "epoch": 1.6402369216856458,
      "grad_norm": 0.26978716254234314,
      "learning_rate": 9.536784014001316e-07,
      "loss": 0.0016,
      "step": 255740
    },
    {
      "epoch": 1.640301058579432,
      "grad_norm": 0.0954686626791954,
      "learning_rate": 9.533496319784957e-07,
      "loss": 0.0011,
      "step": 255750
    },
    {
      "epoch": 1.6403651954732181,
      "grad_norm": 0.021966198459267616,
      "learning_rate": 9.530209132642799e-07,
      "loss": 0.002,
      "step": 255760
    },
    {
      "epoch": 1.6404293323670043,
      "grad_norm": 0.1717972755432129,
      "learning_rate": 9.526922452616055e-07,
      "loss": 0.0009,
      "step": 255770
    },
    {
      "epoch": 1.6404934692607902,
      "grad_norm": 0.12879596650600433,
      "learning_rate": 9.523636279745901e-07,
      "loss": 0.0011,
      "step": 255780
    },
    {
      "epoch": 1.6405576061545764,
      "grad_norm": 0.05737556889653206,
      "learning_rate": 9.520350614073509e-07,
      "loss": 0.0011,
      "step": 255790
    },
    {
      "epoch": 1.6406217430483623,
      "grad_norm": 0.10769285261631012,
      "learning_rate": 9.51706545564004e-07,
      "loss": 0.0009,
      "step": 255800
    },
    {
      "epoch": 1.6406858799421484,
      "grad_norm": 0.06674765050411224,
      "learning_rate": 9.513780804486688e-07,
      "loss": 0.0007,
      "step": 255810
    },
    {
      "epoch": 1.6407500168359346,
      "grad_norm": 0.18882626295089722,
      "learning_rate": 9.5104966606546e-07,
      "loss": 0.0025,
      "step": 255820
    },
    {
      "epoch": 1.6408141537297207,
      "grad_norm": 0.3161608576774597,
      "learning_rate": 9.507213024184914e-07,
      "loss": 0.0015,
      "step": 255830
    },
    {
      "epoch": 1.6408782906235069,
      "grad_norm": 0.05102437362074852,
      "learning_rate": 9.503929895118796e-07,
      "loss": 0.0031,
      "step": 255840
    },
    {
      "epoch": 1.640942427517293,
      "grad_norm": 0.14979737997055054,
      "learning_rate": 9.500647273497371e-07,
      "loss": 0.0011,
      "step": 255850
    },
    {
      "epoch": 1.6410065644110792,
      "grad_norm": 0.13412176072597504,
      "learning_rate": 9.497365159361788e-07,
      "loss": 0.0011,
      "step": 255860
    },
    {
      "epoch": 1.641070701304865,
      "grad_norm": 0.025658756494522095,
      "learning_rate": 9.494083552753169e-07,
      "loss": 0.0028,
      "step": 255870
    },
    {
      "epoch": 1.6411348381986512,
      "grad_norm": 0.05824677646160126,
      "learning_rate": 9.490802453712622e-07,
      "loss": 0.0004,
      "step": 255880
    },
    {
      "epoch": 1.6411989750924372,
      "grad_norm": 0.07874336838722229,
      "learning_rate": 9.487521862281279e-07,
      "loss": 0.0011,
      "step": 255890
    },
    {
      "epoch": 1.6412631119862233,
      "grad_norm": 0.3560166656970978,
      "learning_rate": 9.484241778500247e-07,
      "loss": 0.0011,
      "step": 255900
    },
    {
      "epoch": 1.6413272488800095,
      "grad_norm": 0.03607160225510597,
      "learning_rate": 9.480962202410615e-07,
      "loss": 0.0008,
      "step": 255910
    },
    {
      "epoch": 1.6413913857737956,
      "grad_norm": 0.13168150186538696,
      "learning_rate": 9.477683134053478e-07,
      "loss": 0.0017,
      "step": 255920
    },
    {
      "epoch": 1.6414555226675818,
      "grad_norm": 0.16879135370254517,
      "learning_rate": 9.474404573469942e-07,
      "loss": 0.0008,
      "step": 255930
    },
    {
      "epoch": 1.641519659561368,
      "grad_norm": 0.021642599254846573,
      "learning_rate": 9.471126520701079e-07,
      "loss": 0.0011,
      "step": 255940
    },
    {
      "epoch": 1.6415837964551538,
      "grad_norm": 0.044197067618370056,
      "learning_rate": 9.467848975787958e-07,
      "loss": 0.001,
      "step": 255950
    },
    {
      "epoch": 1.64164793334894,
      "grad_norm": 0.09164709597826004,
      "learning_rate": 9.464571938771666e-07,
      "loss": 0.0031,
      "step": 255960
    },
    {
      "epoch": 1.641712070242726,
      "grad_norm": 0.03152585029602051,
      "learning_rate": 9.461295409693261e-07,
      "loss": 0.0011,
      "step": 255970
    },
    {
      "epoch": 1.641776207136512,
      "grad_norm": 0.11910726130008698,
      "learning_rate": 9.458019388593792e-07,
      "loss": 0.0011,
      "step": 255980
    },
    {
      "epoch": 1.6418403440302982,
      "grad_norm": 0.12369369715452194,
      "learning_rate": 9.454743875514305e-07,
      "loss": 0.001,
      "step": 255990
    },
    {
      "epoch": 1.6419044809240844,
      "grad_norm": 0.09841001033782959,
      "learning_rate": 9.451468870495867e-07,
      "loss": 0.0011,
      "step": 256000
    },
    {
      "epoch": 1.6419686178178705,
      "grad_norm": 0.06547404080629349,
      "learning_rate": 9.448194373579505e-07,
      "loss": 0.0011,
      "step": 256010
    },
    {
      "epoch": 1.6420327547116567,
      "grad_norm": 0.07814065366983414,
      "learning_rate": 9.44492038480625e-07,
      "loss": 0.0006,
      "step": 256020
    },
    {
      "epoch": 1.6420968916054428,
      "grad_norm": 0.06707212328910828,
      "learning_rate": 9.441646904217112e-07,
      "loss": 0.0016,
      "step": 256030
    },
    {
      "epoch": 1.6421610284992287,
      "grad_norm": 0.16191032528877258,
      "learning_rate": 9.43837393185314e-07,
      "loss": 0.0017,
      "step": 256040
    },
    {
      "epoch": 1.6422251653930149,
      "grad_norm": 0.2640500068664551,
      "learning_rate": 9.435101467755331e-07,
      "loss": 0.0011,
      "step": 256050
    },
    {
      "epoch": 1.6422893022868008,
      "grad_norm": 0.032831743359565735,
      "learning_rate": 9.431829511964691e-07,
      "loss": 0.0009,
      "step": 256060
    },
    {
      "epoch": 1.642353439180587,
      "grad_norm": 0.07627349346876144,
      "learning_rate": 9.428558064522208e-07,
      "loss": 0.0019,
      "step": 256070
    },
    {
      "epoch": 1.642417576074373,
      "grad_norm": 0.01395474374294281,
      "learning_rate": 9.425287125468902e-07,
      "loss": 0.0014,
      "step": 256080
    },
    {
      "epoch": 1.6424817129681593,
      "grad_norm": 0.07457228749990463,
      "learning_rate": 9.422016694845743e-07,
      "loss": 0.0009,
      "step": 256090
    },
    {
      "epoch": 1.6425458498619454,
      "grad_norm": 0.08780708909034729,
      "learning_rate": 9.418746772693704e-07,
      "loss": 0.0011,
      "step": 256100
    },
    {
      "epoch": 1.6426099867557316,
      "grad_norm": 0.4448014497756958,
      "learning_rate": 9.415477359053787e-07,
      "loss": 0.0033,
      "step": 256110
    },
    {
      "epoch": 1.6426741236495175,
      "grad_norm": 0.1905025988817215,
      "learning_rate": 9.412208453966942e-07,
      "loss": 0.0014,
      "step": 256120
    },
    {
      "epoch": 1.6427382605433036,
      "grad_norm": 0.03435979783535004,
      "learning_rate": 9.408940057474136e-07,
      "loss": 0.0009,
      "step": 256130
    },
    {
      "epoch": 1.6428023974370898,
      "grad_norm": 0.07999928295612335,
      "learning_rate": 9.405672169616304e-07,
      "loss": 0.0017,
      "step": 256140
    },
    {
      "epoch": 1.6428665343308757,
      "grad_norm": 0.0338425487279892,
      "learning_rate": 9.402404790434427e-07,
      "loss": 0.0013,
      "step": 256150
    },
    {
      "epoch": 1.6429306712246619,
      "grad_norm": 0.07838539034128189,
      "learning_rate": 9.399137919969431e-07,
      "loss": 0.0007,
      "step": 256160
    },
    {
      "epoch": 1.642994808118448,
      "grad_norm": 0.02995946630835533,
      "learning_rate": 9.395871558262254e-07,
      "loss": 0.0009,
      "step": 256170
    },
    {
      "epoch": 1.6430589450122342,
      "grad_norm": 0.01181158795952797,
      "learning_rate": 9.392605705353813e-07,
      "loss": 0.001,
      "step": 256180
    },
    {
      "epoch": 1.6431230819060203,
      "grad_norm": 0.24913929402828217,
      "learning_rate": 9.389340361285059e-07,
      "loss": 0.0022,
      "step": 256190
    },
    {
      "epoch": 1.6431872187998064,
      "grad_norm": 0.1120237410068512,
      "learning_rate": 9.386075526096888e-07,
      "loss": 0.0036,
      "step": 256200
    },
    {
      "epoch": 1.6432513556935924,
      "grad_norm": 0.17576159536838531,
      "learning_rate": 9.382811199830227e-07,
      "loss": 0.0013,
      "step": 256210
    },
    {
      "epoch": 1.6433154925873785,
      "grad_norm": 0.09315807372331619,
      "learning_rate": 9.37954738252595e-07,
      "loss": 0.0014,
      "step": 256220
    },
    {
      "epoch": 1.6433796294811645,
      "grad_norm": 0.15503202378749847,
      "learning_rate": 9.376284074224989e-07,
      "loss": 0.001,
      "step": 256230
    },
    {
      "epoch": 1.6434437663749506,
      "grad_norm": 0.13876628875732422,
      "learning_rate": 9.373021274968225e-07,
      "loss": 0.0015,
      "step": 256240
    },
    {
      "epoch": 1.6435079032687367,
      "grad_norm": 0.1393558830022812,
      "learning_rate": 9.369758984796523e-07,
      "loss": 0.001,
      "step": 256250
    },
    {
      "epoch": 1.643572040162523,
      "grad_norm": 0.04984534904360771,
      "learning_rate": 9.366497203750796e-07,
      "loss": 0.0007,
      "step": 256260
    },
    {
      "epoch": 1.643636177056309,
      "grad_norm": 0.08660076558589935,
      "learning_rate": 9.3632359318719e-07,
      "loss": 0.001,
      "step": 256270
    },
    {
      "epoch": 1.6437003139500952,
      "grad_norm": 0.055148929357528687,
      "learning_rate": 9.359975169200696e-07,
      "loss": 0.0007,
      "step": 256280
    },
    {
      "epoch": 1.6437644508438813,
      "grad_norm": 0.13627982139587402,
      "learning_rate": 9.356714915778037e-07,
      "loss": 0.0031,
      "step": 256290
    },
    {
      "epoch": 1.6438285877376673,
      "grad_norm": 0.08665461093187332,
      "learning_rate": 9.353455171644798e-07,
      "loss": 0.0005,
      "step": 256300
    },
    {
      "epoch": 1.6438927246314534,
      "grad_norm": 0.08246444910764694,
      "learning_rate": 9.350195936841822e-07,
      "loss": 0.0012,
      "step": 256310
    },
    {
      "epoch": 1.6439568615252393,
      "grad_norm": 0.04495377838611603,
      "learning_rate": 9.346937211409939e-07,
      "loss": 0.0005,
      "step": 256320
    },
    {
      "epoch": 1.6440209984190255,
      "grad_norm": 0.06548351049423218,
      "learning_rate": 9.34367899538997e-07,
      "loss": 0.001,
      "step": 256330
    },
    {
      "epoch": 1.6440851353128116,
      "grad_norm": 0.09427938610315323,
      "learning_rate": 9.340421288822782e-07,
      "loss": 0.0015,
      "step": 256340
    },
    {
      "epoch": 1.6441492722065978,
      "grad_norm": 0.07400583475828171,
      "learning_rate": 9.337164091749168e-07,
      "loss": 0.0006,
      "step": 256350
    },
    {
      "epoch": 1.644213409100384,
      "grad_norm": 0.19087454676628113,
      "learning_rate": 9.333907404209947e-07,
      "loss": 0.0019,
      "step": 256360
    },
    {
      "epoch": 1.64427754599417,
      "grad_norm": 0.03530288115143776,
      "learning_rate": 9.330651226245924e-07,
      "loss": 0.003,
      "step": 256370
    },
    {
      "epoch": 1.644341682887956,
      "grad_norm": 0.04272129759192467,
      "learning_rate": 9.327395557897911e-07,
      "loss": 0.0018,
      "step": 256380
    },
    {
      "epoch": 1.6444058197817422,
      "grad_norm": 0.1905973255634308,
      "learning_rate": 9.32414039920671e-07,
      "loss": 0.0014,
      "step": 256390
    },
    {
      "epoch": 1.644469956675528,
      "grad_norm": 0.004057542886584997,
      "learning_rate": 9.320885750213083e-07,
      "loss": 0.0008,
      "step": 256400
    },
    {
      "epoch": 1.6445340935693142,
      "grad_norm": 0.02324562892317772,
      "learning_rate": 9.317631610957839e-07,
      "loss": 0.0014,
      "step": 256410
    },
    {
      "epoch": 1.6445982304631004,
      "grad_norm": 0.008663343265652657,
      "learning_rate": 9.314377981481754e-07,
      "loss": 0.001,
      "step": 256420
    },
    {
      "epoch": 1.6446623673568865,
      "grad_norm": 0.06866946071386337,
      "learning_rate": 9.311124861825582e-07,
      "loss": 0.0013,
      "step": 256430
    },
    {
      "epoch": 1.6447265042506727,
      "grad_norm": 0.13391518592834473,
      "learning_rate": 9.307872252030087e-07,
      "loss": 0.0007,
      "step": 256440
    },
    {
      "epoch": 1.6447906411444588,
      "grad_norm": 0.013321959413588047,
      "learning_rate": 9.304620152136052e-07,
      "loss": 0.0013,
      "step": 256450
    },
    {
      "epoch": 1.644854778038245,
      "grad_norm": 0.055423554033041,
      "learning_rate": 9.301368562184204e-07,
      "loss": 0.0008,
      "step": 256460
    },
    {
      "epoch": 1.644918914932031,
      "grad_norm": 0.11180834472179413,
      "learning_rate": 9.298117482215296e-07,
      "loss": 0.0011,
      "step": 256470
    },
    {
      "epoch": 1.644983051825817,
      "grad_norm": 0.34718215465545654,
      "learning_rate": 9.294866912270056e-07,
      "loss": 0.0018,
      "step": 256480
    },
    {
      "epoch": 1.645047188719603,
      "grad_norm": 0.02721422351896763,
      "learning_rate": 9.291616852389235e-07,
      "loss": 0.0005,
      "step": 256490
    },
    {
      "epoch": 1.6451113256133891,
      "grad_norm": 0.006650662515312433,
      "learning_rate": 9.288367302613549e-07,
      "loss": 0.0012,
      "step": 256500
    },
    {
      "epoch": 1.6451754625071753,
      "grad_norm": 0.062067050486803055,
      "learning_rate": 9.285118262983717e-07,
      "loss": 0.0009,
      "step": 256510
    },
    {
      "epoch": 1.6452395994009614,
      "grad_norm": 0.10535760223865509,
      "learning_rate": 9.281869733540443e-07,
      "loss": 0.0012,
      "step": 256520
    },
    {
      "epoch": 1.6453037362947476,
      "grad_norm": 0.06388656049966812,
      "learning_rate": 9.278621714324449e-07,
      "loss": 0.001,
      "step": 256530
    },
    {
      "epoch": 1.6453678731885337,
      "grad_norm": 0.10174249112606049,
      "learning_rate": 9.275374205376431e-07,
      "loss": 0.0014,
      "step": 256540
    },
    {
      "epoch": 1.6454320100823197,
      "grad_norm": 0.11160119622945786,
      "learning_rate": 9.272127206737064e-07,
      "loss": 0.0008,
      "step": 256550
    },
    {
      "epoch": 1.6454961469761058,
      "grad_norm": 0.08544647693634033,
      "learning_rate": 9.268880718447066e-07,
      "loss": 0.001,
      "step": 256560
    },
    {
      "epoch": 1.645560283869892,
      "grad_norm": 0.07638426870107651,
      "learning_rate": 9.265634740547103e-07,
      "loss": 0.0006,
      "step": 256570
    },
    {
      "epoch": 1.6456244207636779,
      "grad_norm": 0.004329252522438765,
      "learning_rate": 9.262389273077849e-07,
      "loss": 0.0009,
      "step": 256580
    },
    {
      "epoch": 1.645688557657464,
      "grad_norm": 0.002356065670028329,
      "learning_rate": 9.25914431607996e-07,
      "loss": 0.0007,
      "step": 256590
    },
    {
      "epoch": 1.6457526945512502,
      "grad_norm": 0.01471665408462286,
      "learning_rate": 9.255899869594121e-07,
      "loss": 0.0015,
      "step": 256600
    },
    {
      "epoch": 1.6458168314450363,
      "grad_norm": 0.053834833204746246,
      "learning_rate": 9.252655933660964e-07,
      "loss": 0.0011,
      "step": 256610
    },
    {
      "epoch": 1.6458809683388225,
      "grad_norm": 0.17222963273525238,
      "learning_rate": 9.249412508321159e-07,
      "loss": 0.0024,
      "step": 256620
    },
    {
      "epoch": 1.6459451052326086,
      "grad_norm": 0.10242783278226852,
      "learning_rate": 9.246169593615345e-07,
      "loss": 0.0007,
      "step": 256630
    },
    {
      "epoch": 1.6460092421263945,
      "grad_norm": 0.11901787668466568,
      "learning_rate": 9.242927189584139e-07,
      "loss": 0.0007,
      "step": 256640
    },
    {
      "epoch": 1.6460733790201807,
      "grad_norm": 0.030185380950570107,
      "learning_rate": 9.239685296268191e-07,
      "loss": 0.001,
      "step": 256650
    },
    {
      "epoch": 1.6461375159139666,
      "grad_norm": 0.07891599088907242,
      "learning_rate": 9.236443913708121e-07,
      "loss": 0.0008,
      "step": 256660
    },
    {
      "epoch": 1.6462016528077528,
      "grad_norm": 0.0033202078193426132,
      "learning_rate": 9.233203041944533e-07,
      "loss": 0.0011,
      "step": 256670
    },
    {
      "epoch": 1.646265789701539,
      "grad_norm": 0.23745547235012054,
      "learning_rate": 9.229962681018051e-07,
      "loss": 0.0014,
      "step": 256680
    },
    {
      "epoch": 1.646329926595325,
      "grad_norm": 0.016615508124232292,
      "learning_rate": 9.226722830969281e-07,
      "loss": 0.0008,
      "step": 256690
    },
    {
      "epoch": 1.6463940634891112,
      "grad_norm": 0.09162529557943344,
      "learning_rate": 9.223483491838814e-07,
      "loss": 0.0014,
      "step": 256700
    },
    {
      "epoch": 1.6464582003828974,
      "grad_norm": 0.09584107249975204,
      "learning_rate": 9.220244663667228e-07,
      "loss": 0.0009,
      "step": 256710
    },
    {
      "epoch": 1.6465223372766835,
      "grad_norm": 0.1296747475862503,
      "learning_rate": 9.217006346495128e-07,
      "loss": 0.0007,
      "step": 256720
    },
    {
      "epoch": 1.6465864741704694,
      "grad_norm": 0.06498198956251144,
      "learning_rate": 9.213768540363094e-07,
      "loss": 0.0011,
      "step": 256730
    },
    {
      "epoch": 1.6466506110642556,
      "grad_norm": 0.09656433016061783,
      "learning_rate": 9.210531245311683e-07,
      "loss": 0.0006,
      "step": 256740
    },
    {
      "epoch": 1.6467147479580415,
      "grad_norm": 0.12453159689903259,
      "learning_rate": 9.207294461381461e-07,
      "loss": 0.0013,
      "step": 256750
    },
    {
      "epoch": 1.6467788848518277,
      "grad_norm": 0.24564692378044128,
      "learning_rate": 9.204058188613002e-07,
      "loss": 0.0035,
      "step": 256760
    },
    {
      "epoch": 1.6468430217456138,
      "grad_norm": 0.08555837720632553,
      "learning_rate": 9.200822427046852e-07,
      "loss": 0.0012,
      "step": 256770
    },
    {
      "epoch": 1.6469071586394,
      "grad_norm": 0.030364805832505226,
      "learning_rate": 9.197587176723555e-07,
      "loss": 0.0009,
      "step": 256780
    },
    {
      "epoch": 1.646971295533186,
      "grad_norm": 0.056859008967876434,
      "learning_rate": 9.194352437683635e-07,
      "loss": 0.0011,
      "step": 256790
    },
    {
      "epoch": 1.6470354324269723,
      "grad_norm": 0.0025923969224095345,
      "learning_rate": 9.191118209967658e-07,
      "loss": 0.0006,
      "step": 256800
    },
    {
      "epoch": 1.6470995693207582,
      "grad_norm": 0.01009892113506794,
      "learning_rate": 9.18788449361614e-07,
      "loss": 0.0006,
      "step": 256810
    },
    {
      "epoch": 1.6471637062145443,
      "grad_norm": 0.12013409286737442,
      "learning_rate": 9.184651288669582e-07,
      "loss": 0.0014,
      "step": 256820
    },
    {
      "epoch": 1.6472278431083303,
      "grad_norm": 0.11209302395582199,
      "learning_rate": 9.181418595168523e-07,
      "loss": 0.0015,
      "step": 256830
    },
    {
      "epoch": 1.6472919800021164,
      "grad_norm": 0.03150581568479538,
      "learning_rate": 9.178186413153468e-07,
      "loss": 0.0012,
      "step": 256840
    },
    {
      "epoch": 1.6473561168959026,
      "grad_norm": 0.040904004126787186,
      "learning_rate": 9.174954742664904e-07,
      "loss": 0.0011,
      "step": 256850
    },
    {
      "epoch": 1.6474202537896887,
      "grad_norm": 0.10026594251394272,
      "learning_rate": 9.171723583743325e-07,
      "loss": 0.0009,
      "step": 256860
    },
    {
      "epoch": 1.6474843906834749,
      "grad_norm": 0.006834973581135273,
      "learning_rate": 9.168492936429246e-07,
      "loss": 0.0006,
      "step": 256870
    },
    {
      "epoch": 1.647548527577261,
      "grad_norm": 0.15026934444904327,
      "learning_rate": 9.165262800763126e-07,
      "loss": 0.0017,
      "step": 256880
    },
    {
      "epoch": 1.6476126644710472,
      "grad_norm": 0.2972351908683777,
      "learning_rate": 9.16203317678545e-07,
      "loss": 0.0016,
      "step": 256890
    },
    {
      "epoch": 1.647676801364833,
      "grad_norm": 0.05031905695796013,
      "learning_rate": 9.158804064536675e-07,
      "loss": 0.0011,
      "step": 256900
    },
    {
      "epoch": 1.6477409382586192,
      "grad_norm": 0.1165706142783165,
      "learning_rate": 9.155575464057282e-07,
      "loss": 0.0011,
      "step": 256910
    },
    {
      "epoch": 1.6478050751524052,
      "grad_norm": 0.06868688762187958,
      "learning_rate": 9.152347375387721e-07,
      "loss": 0.0011,
      "step": 256920
    },
    {
      "epoch": 1.6478692120461913,
      "grad_norm": 0.14829511940479279,
      "learning_rate": 9.14911979856844e-07,
      "loss": 0.0032,
      "step": 256930
    },
    {
      "epoch": 1.6479333489399774,
      "grad_norm": 0.06655465066432953,
      "learning_rate": 9.145892733639872e-07,
      "loss": 0.0004,
      "step": 256940
    },
    {
      "epoch": 1.6479974858337636,
      "grad_norm": 0.0601753368973732,
      "learning_rate": 9.142666180642479e-07,
      "loss": 0.0008,
      "step": 256950
    },
    {
      "epoch": 1.6480616227275497,
      "grad_norm": 0.01680414192378521,
      "learning_rate": 9.139440139616679e-07,
      "loss": 0.001,
      "step": 256960
    },
    {
      "epoch": 1.648125759621336,
      "grad_norm": 0.06670185923576355,
      "learning_rate": 9.136214610602884e-07,
      "loss": 0.0014,
      "step": 256970
    },
    {
      "epoch": 1.6481898965151218,
      "grad_norm": 0.030908556655049324,
      "learning_rate": 9.132989593641534e-07,
      "loss": 0.0005,
      "step": 256980
    },
    {
      "epoch": 1.648254033408908,
      "grad_norm": 0.18697044253349304,
      "learning_rate": 9.129765088773035e-07,
      "loss": 0.001,
      "step": 256990
    },
    {
      "epoch": 1.6483181703026941,
      "grad_norm": 0.4722413122653961,
      "learning_rate": 9.126541096037789e-07,
      "loss": 0.001,
      "step": 257000
    },
    {
      "epoch": 1.64838230719648,
      "grad_norm": 0.026497313752770424,
      "learning_rate": 9.123317615476185e-07,
      "loss": 0.0009,
      "step": 257010
    },
    {
      "epoch": 1.6484464440902662,
      "grad_norm": 0.03699392452836037,
      "learning_rate": 9.12009464712863e-07,
      "loss": 0.0007,
      "step": 257020
    },
    {
      "epoch": 1.6485105809840523,
      "grad_norm": 0.0715600997209549,
      "learning_rate": 9.116872191035514e-07,
      "loss": 0.0005,
      "step": 257030
    },
    {
      "epoch": 1.6485747178778385,
      "grad_norm": 0.05829404294490814,
      "learning_rate": 9.113650247237205e-07,
      "loss": 0.0008,
      "step": 257040
    },
    {
      "epoch": 1.6486388547716246,
      "grad_norm": 0.06432265043258667,
      "learning_rate": 9.110428815774064e-07,
      "loss": 0.0015,
      "step": 257050
    },
    {
      "epoch": 1.6487029916654108,
      "grad_norm": 0.1736939251422882,
      "learning_rate": 9.107207896686487e-07,
      "loss": 0.0017,
      "step": 257060
    },
    {
      "epoch": 1.6487671285591967,
      "grad_norm": 0.02131580002605915,
      "learning_rate": 9.103987490014826e-07,
      "loss": 0.0012,
      "step": 257070
    },
    {
      "epoch": 1.6488312654529829,
      "grad_norm": 0.049157217144966125,
      "learning_rate": 9.100767595799426e-07,
      "loss": 0.001,
      "step": 257080
    },
    {
      "epoch": 1.6488954023467688,
      "grad_norm": 0.0992347002029419,
      "learning_rate": 9.09754821408062e-07,
      "loss": 0.001,
      "step": 257090
    },
    {
      "epoch": 1.648959539240555,
      "grad_norm": 0.08398938924074173,
      "learning_rate": 9.094329344898789e-07,
      "loss": 0.0011,
      "step": 257100
    },
    {
      "epoch": 1.649023676134341,
      "grad_norm": 0.010240110568702221,
      "learning_rate": 9.091110988294244e-07,
      "loss": 0.0011,
      "step": 257110
    },
    {
      "epoch": 1.6490878130281272,
      "grad_norm": 0.018708644434809685,
      "learning_rate": 9.087893144307297e-07,
      "loss": 0.0013,
      "step": 257120
    },
    {
      "epoch": 1.6491519499219134,
      "grad_norm": 0.10159168392419815,
      "learning_rate": 9.084675812978305e-07,
      "loss": 0.0015,
      "step": 257130
    },
    {
      "epoch": 1.6492160868156995,
      "grad_norm": 0.02465192601084709,
      "learning_rate": 9.081458994347569e-07,
      "loss": 0.0007,
      "step": 257140
    },
    {
      "epoch": 1.6492802237094857,
      "grad_norm": 0.09112152457237244,
      "learning_rate": 9.078242688455397e-07,
      "loss": 0.0014,
      "step": 257150
    },
    {
      "epoch": 1.6493443606032716,
      "grad_norm": 0.002627485664561391,
      "learning_rate": 9.075026895342071e-07,
      "loss": 0.0006,
      "step": 257160
    },
    {
      "epoch": 1.6494084974970578,
      "grad_norm": 0.1102386862039566,
      "learning_rate": 9.071811615047926e-07,
      "loss": 0.0021,
      "step": 257170
    },
    {
      "epoch": 1.6494726343908437,
      "grad_norm": 0.16178393363952637,
      "learning_rate": 9.068596847613226e-07,
      "loss": 0.0025,
      "step": 257180
    },
    {
      "epoch": 1.6495367712846298,
      "grad_norm": 0.03818779066205025,
      "learning_rate": 9.065382593078265e-07,
      "loss": 0.0007,
      "step": 257190
    },
    {
      "epoch": 1.649600908178416,
      "grad_norm": 0.0884946957230568,
      "learning_rate": 9.062168851483299e-07,
      "loss": 0.0006,
      "step": 257200
    },
    {
      "epoch": 1.6496650450722021,
      "grad_norm": 0.1244484931230545,
      "learning_rate": 9.058955622868631e-07,
      "loss": 0.0011,
      "step": 257210
    },
    {
      "epoch": 1.6497291819659883,
      "grad_norm": 0.01862574927508831,
      "learning_rate": 9.055742907274511e-07,
      "loss": 0.0012,
      "step": 257220
    },
    {
      "epoch": 1.6497933188597744,
      "grad_norm": 0.012687386944890022,
      "learning_rate": 9.052530704741191e-07,
      "loss": 0.0011,
      "step": 257230
    },
    {
      "epoch": 1.6498574557535604,
      "grad_norm": 0.014412990771234035,
      "learning_rate": 9.049319015308916e-07,
      "loss": 0.0008,
      "step": 257240
    },
    {
      "epoch": 1.6499215926473465,
      "grad_norm": 0.00819874182343483,
      "learning_rate": 9.046107839017954e-07,
      "loss": 0.0018,
      "step": 257250
    },
    {
      "epoch": 1.6499857295411324,
      "grad_norm": 0.08654952794313431,
      "learning_rate": 9.042897175908527e-07,
      "loss": 0.0031,
      "step": 257260
    },
    {
      "epoch": 1.6500498664349186,
      "grad_norm": 0.44848760962486267,
      "learning_rate": 9.039687026020861e-07,
      "loss": 0.0014,
      "step": 257270
    },
    {
      "epoch": 1.6501140033287047,
      "grad_norm": 0.05857900530099869,
      "learning_rate": 9.036477389395204e-07,
      "loss": 0.001,
      "step": 257280
    },
    {
      "epoch": 1.6501781402224909,
      "grad_norm": 0.09503237903118134,
      "learning_rate": 9.033268266071755e-07,
      "loss": 0.0008,
      "step": 257290
    },
    {
      "epoch": 1.650242277116277,
      "grad_norm": 0.13392576575279236,
      "learning_rate": 9.03005965609074e-07,
      "loss": 0.0018,
      "step": 257300
    },
    {
      "epoch": 1.6503064140100632,
      "grad_norm": 0.09986912459135056,
      "learning_rate": 9.026851559492339e-07,
      "loss": 0.0007,
      "step": 257310
    },
    {
      "epoch": 1.6503705509038493,
      "grad_norm": 0.013021270744502544,
      "learning_rate": 9.023643976316787e-07,
      "loss": 0.0018,
      "step": 257320
    },
    {
      "epoch": 1.6504346877976352,
      "grad_norm": 0.025929611176252365,
      "learning_rate": 9.020436906604263e-07,
      "loss": 0.0008,
      "step": 257330
    },
    {
      "epoch": 1.6504988246914214,
      "grad_norm": 0.017933866009116173,
      "learning_rate": 9.017230350394951e-07,
      "loss": 0.0009,
      "step": 257340
    },
    {
      "epoch": 1.6505629615852073,
      "grad_norm": 0.08063351362943649,
      "learning_rate": 9.014024307729019e-07,
      "loss": 0.0018,
      "step": 257350
    },
    {
      "epoch": 1.6506270984789935,
      "grad_norm": 0.018969247117638588,
      "learning_rate": 9.010818778646668e-07,
      "loss": 0.0008,
      "step": 257360
    },
    {
      "epoch": 1.6506912353727796,
      "grad_norm": 0.011855973862111568,
      "learning_rate": 9.00761376318805e-07,
      "loss": 0.0011,
      "step": 257370
    },
    {
      "epoch": 1.6507553722665658,
      "grad_norm": 0.036045532673597336,
      "learning_rate": 9.004409261393321e-07,
      "loss": 0.0011,
      "step": 257380
    },
    {
      "epoch": 1.650819509160352,
      "grad_norm": 0.029906436800956726,
      "learning_rate": 9.001205273302638e-07,
      "loss": 0.0012,
      "step": 257390
    },
    {
      "epoch": 1.650883646054138,
      "grad_norm": 0.13861581683158875,
      "learning_rate": 8.998001798956168e-07,
      "loss": 0.0005,
      "step": 257400
    },
    {
      "epoch": 1.6509477829479242,
      "grad_norm": 0.10438334196805954,
      "learning_rate": 8.994798838394043e-07,
      "loss": 0.0006,
      "step": 257410
    },
    {
      "epoch": 1.6510119198417101,
      "grad_norm": 0.1229967400431633,
      "learning_rate": 8.991596391656388e-07,
      "loss": 0.0012,
      "step": 257420
    },
    {
      "epoch": 1.6510760567354963,
      "grad_norm": 0.05250271409749985,
      "learning_rate": 8.988394458783334e-07,
      "loss": 0.002,
      "step": 257430
    },
    {
      "epoch": 1.6511401936292822,
      "grad_norm": 0.0711706355214119,
      "learning_rate": 8.985193039815015e-07,
      "loss": 0.0007,
      "step": 257440
    },
    {
      "epoch": 1.6512043305230684,
      "grad_norm": 0.12365922331809998,
      "learning_rate": 8.98199213479154e-07,
      "loss": 0.002,
      "step": 257450
    },
    {
      "epoch": 1.6512684674168545,
      "grad_norm": 0.03708312660455704,
      "learning_rate": 8.978791743753018e-07,
      "loss": 0.0012,
      "step": 257460
    },
    {
      "epoch": 1.6513326043106407,
      "grad_norm": 0.04477045685052872,
      "learning_rate": 8.975591866739541e-07,
      "loss": 0.0017,
      "step": 257470
    },
    {
      "epoch": 1.6513967412044268,
      "grad_norm": 0.01089094765484333,
      "learning_rate": 8.972392503791233e-07,
      "loss": 0.0012,
      "step": 257480
    },
    {
      "epoch": 1.651460878098213,
      "grad_norm": 0.014865393750369549,
      "learning_rate": 8.969193654948166e-07,
      "loss": 0.0016,
      "step": 257490
    },
    {
      "epoch": 1.6515250149919989,
      "grad_norm": 0.22002476453781128,
      "learning_rate": 8.965995320250415e-07,
      "loss": 0.001,
      "step": 257500
    },
    {
      "epoch": 1.651589151885785,
      "grad_norm": 0.1237795427441597,
      "learning_rate": 8.962797499738085e-07,
      "loss": 0.0018,
      "step": 257510
    },
    {
      "epoch": 1.651653288779571,
      "grad_norm": 0.01695345714688301,
      "learning_rate": 8.959600193451229e-07,
      "loss": 0.0003,
      "step": 257520
    },
    {
      "epoch": 1.651717425673357,
      "grad_norm": 0.07024957239627838,
      "learning_rate": 8.956403401429909e-07,
      "loss": 0.0006,
      "step": 257530
    },
    {
      "epoch": 1.6517815625671433,
      "grad_norm": 0.003555310657247901,
      "learning_rate": 8.95320712371418e-07,
      "loss": 0.0007,
      "step": 257540
    },
    {
      "epoch": 1.6518456994609294,
      "grad_norm": 0.0036182524636387825,
      "learning_rate": 8.950011360344113e-07,
      "loss": 0.0011,
      "step": 257550
    },
    {
      "epoch": 1.6519098363547156,
      "grad_norm": 0.12117797136306763,
      "learning_rate": 8.946816111359741e-07,
      "loss": 0.0012,
      "step": 257560
    },
    {
      "epoch": 1.6519739732485017,
      "grad_norm": 0.09137025475502014,
      "learning_rate": 8.943621376801103e-07,
      "loss": 0.0008,
      "step": 257570
    },
    {
      "epoch": 1.6520381101422879,
      "grad_norm": 0.09539289027452469,
      "learning_rate": 8.940427156708225e-07,
      "loss": 0.0013,
      "step": 257580
    },
    {
      "epoch": 1.6521022470360738,
      "grad_norm": 0.004371563903987408,
      "learning_rate": 8.937233451121147e-07,
      "loss": 0.0013,
      "step": 257590
    },
    {
      "epoch": 1.65216638392986,
      "grad_norm": 0.08195861428976059,
      "learning_rate": 8.934040260079879e-07,
      "loss": 0.0006,
      "step": 257600
    },
    {
      "epoch": 1.6522305208236459,
      "grad_norm": 0.25516414642333984,
      "learning_rate": 8.930847583624441e-07,
      "loss": 0.0006,
      "step": 257610
    },
    {
      "epoch": 1.652294657717432,
      "grad_norm": 0.0634370967745781,
      "learning_rate": 8.927655421794817e-07,
      "loss": 0.0006,
      "step": 257620
    },
    {
      "epoch": 1.6523587946112182,
      "grad_norm": 0.12932150065898895,
      "learning_rate": 8.924463774631043e-07,
      "loss": 0.0014,
      "step": 257630
    },
    {
      "epoch": 1.6524229315050043,
      "grad_norm": 0.01701982319355011,
      "learning_rate": 8.92127264217309e-07,
      "loss": 0.0036,
      "step": 257640
    },
    {
      "epoch": 1.6524870683987904,
      "grad_norm": 0.10885673016309738,
      "learning_rate": 8.918082024460934e-07,
      "loss": 0.0014,
      "step": 257650
    },
    {
      "epoch": 1.6525512052925766,
      "grad_norm": 0.06675291806459427,
      "learning_rate": 8.914891921534591e-07,
      "loss": 0.0008,
      "step": 257660
    },
    {
      "epoch": 1.6526153421863625,
      "grad_norm": 0.13544189929962158,
      "learning_rate": 8.911702333434008e-07,
      "loss": 0.0008,
      "step": 257670
    },
    {
      "epoch": 1.6526794790801487,
      "grad_norm": 0.3474222421646118,
      "learning_rate": 8.908513260199165e-07,
      "loss": 0.0023,
      "step": 257680
    },
    {
      "epoch": 1.6527436159739348,
      "grad_norm": 0.07608381658792496,
      "learning_rate": 8.905324701870006e-07,
      "loss": 0.001,
      "step": 257690
    },
    {
      "epoch": 1.6528077528677207,
      "grad_norm": 0.10198512673377991,
      "learning_rate": 8.902136658486504e-07,
      "loss": 0.0016,
      "step": 257700
    },
    {
      "epoch": 1.652871889761507,
      "grad_norm": 0.09177470952272415,
      "learning_rate": 8.898949130088607e-07,
      "loss": 0.001,
      "step": 257710
    },
    {
      "epoch": 1.652936026655293,
      "grad_norm": 0.1004767194390297,
      "learning_rate": 8.895762116716256e-07,
      "loss": 0.0009,
      "step": 257720
    },
    {
      "epoch": 1.6530001635490792,
      "grad_norm": 0.06865187734365463,
      "learning_rate": 8.892575618409365e-07,
      "loss": 0.0011,
      "step": 257730
    },
    {
      "epoch": 1.6530643004428653,
      "grad_norm": 0.1096527948975563,
      "learning_rate": 8.889389635207896e-07,
      "loss": 0.0008,
      "step": 257740
    },
    {
      "epoch": 1.6531284373366515,
      "grad_norm": 0.20047782361507416,
      "learning_rate": 8.886204167151752e-07,
      "loss": 0.0018,
      "step": 257750
    },
    {
      "epoch": 1.6531925742304374,
      "grad_norm": 0.16751068830490112,
      "learning_rate": 8.883019214280853e-07,
      "loss": 0.0012,
      "step": 257760
    },
    {
      "epoch": 1.6532567111242236,
      "grad_norm": 0.06954125314950943,
      "learning_rate": 8.879834776635099e-07,
      "loss": 0.001,
      "step": 257770
    },
    {
      "epoch": 1.6533208480180095,
      "grad_norm": 0.009285716339945793,
      "learning_rate": 8.876650854254421e-07,
      "loss": 0.0008,
      "step": 257780
    },
    {
      "epoch": 1.6533849849117956,
      "grad_norm": 0.17273807525634766,
      "learning_rate": 8.873467447178691e-07,
      "loss": 0.0023,
      "step": 257790
    },
    {
      "epoch": 1.6534491218055818,
      "grad_norm": 0.021548360586166382,
      "learning_rate": 8.870284555447794e-07,
      "loss": 0.0008,
      "step": 257800
    },
    {
      "epoch": 1.653513258699368,
      "grad_norm": 0.09591946005821228,
      "learning_rate": 8.867102179101645e-07,
      "loss": 0.0008,
      "step": 257810
    },
    {
      "epoch": 1.653577395593154,
      "grad_norm": 0.007451815530657768,
      "learning_rate": 8.863920318180097e-07,
      "loss": 0.0005,
      "step": 257820
    },
    {
      "epoch": 1.6536415324869402,
      "grad_norm": 0.05306859314441681,
      "learning_rate": 8.860738972723026e-07,
      "loss": 0.0012,
      "step": 257830
    },
    {
      "epoch": 1.6537056693807264,
      "grad_norm": 0.013590164482593536,
      "learning_rate": 8.85755814277029e-07,
      "loss": 0.0016,
      "step": 257840
    },
    {
      "epoch": 1.6537698062745123,
      "grad_norm": 0.31237760186195374,
      "learning_rate": 8.854377828361765e-07,
      "loss": 0.0014,
      "step": 257850
    },
    {
      "epoch": 1.6538339431682985,
      "grad_norm": 0.08388779312372208,
      "learning_rate": 8.851198029537295e-07,
      "loss": 0.0007,
      "step": 257860
    },
    {
      "epoch": 1.6538980800620844,
      "grad_norm": 0.020440956577658653,
      "learning_rate": 8.848018746336717e-07,
      "loss": 0.0006,
      "step": 257870
    },
    {
      "epoch": 1.6539622169558705,
      "grad_norm": 0.03452529013156891,
      "learning_rate": 8.844839978799868e-07,
      "loss": 0.0008,
      "step": 257880
    },
    {
      "epoch": 1.6540263538496567,
      "grad_norm": 0.10935361683368683,
      "learning_rate": 8.841661726966593e-07,
      "loss": 0.0009,
      "step": 257890
    },
    {
      "epoch": 1.6540904907434428,
      "grad_norm": 0.21043239533901215,
      "learning_rate": 8.838483990876718e-07,
      "loss": 0.0016,
      "step": 257900
    },
    {
      "epoch": 1.654154627637229,
      "grad_norm": 0.10948438197374344,
      "learning_rate": 8.835306770570051e-07,
      "loss": 0.0012,
      "step": 257910
    },
    {
      "epoch": 1.6542187645310151,
      "grad_norm": 0.10672084242105484,
      "learning_rate": 8.832130066086403e-07,
      "loss": 0.0005,
      "step": 257920
    },
    {
      "epoch": 1.654282901424801,
      "grad_norm": 0.023894185200333595,
      "learning_rate": 8.828953877465595e-07,
      "loss": 0.0008,
      "step": 257930
    },
    {
      "epoch": 1.6543470383185872,
      "grad_norm": 0.16079075634479523,
      "learning_rate": 8.825778204747426e-07,
      "loss": 0.0008,
      "step": 257940
    },
    {
      "epoch": 1.6544111752123731,
      "grad_norm": 0.019003797322511673,
      "learning_rate": 8.822603047971667e-07,
      "loss": 0.0008,
      "step": 257950
    },
    {
      "epoch": 1.6544753121061593,
      "grad_norm": 0.001286782673560083,
      "learning_rate": 8.819428407178133e-07,
      "loss": 0.0007,
      "step": 257960
    },
    {
      "epoch": 1.6545394489999454,
      "grad_norm": 0.09896622598171234,
      "learning_rate": 8.816254282406589e-07,
      "loss": 0.0008,
      "step": 257970
    },
    {
      "epoch": 1.6546035858937316,
      "grad_norm": 0.12817074358463287,
      "learning_rate": 8.813080673696817e-07,
      "loss": 0.0011,
      "step": 257980
    },
    {
      "epoch": 1.6546677227875177,
      "grad_norm": 0.07022719830274582,
      "learning_rate": 8.809907581088567e-07,
      "loss": 0.0035,
      "step": 257990
    },
    {
      "epoch": 1.6547318596813039,
      "grad_norm": 0.072791688144207,
      "learning_rate": 8.806735004621625e-07,
      "loss": 0.0011,
      "step": 258000
    },
    {
      "epoch": 1.65479599657509,
      "grad_norm": 0.12035372108221054,
      "learning_rate": 8.803562944335731e-07,
      "loss": 0.0032,
      "step": 258010
    },
    {
      "epoch": 1.654860133468876,
      "grad_norm": 0.06667357683181763,
      "learning_rate": 8.800391400270641e-07,
      "loss": 0.0019,
      "step": 258020
    },
    {
      "epoch": 1.654924270362662,
      "grad_norm": 0.11507948487997055,
      "learning_rate": 8.797220372466075e-07,
      "loss": 0.0021,
      "step": 258030
    },
    {
      "epoch": 1.654988407256448,
      "grad_norm": 0.06681770831346512,
      "learning_rate": 8.7940498609618e-07,
      "loss": 0.0008,
      "step": 258040
    },
    {
      "epoch": 1.6550525441502342,
      "grad_norm": 0.125620499253273,
      "learning_rate": 8.790879865797525e-07,
      "loss": 0.0008,
      "step": 258050
    },
    {
      "epoch": 1.6551166810440203,
      "grad_norm": 0.09714993834495544,
      "learning_rate": 8.787710387012982e-07,
      "loss": 0.0007,
      "step": 258060
    },
    {
      "epoch": 1.6551808179378065,
      "grad_norm": 0.005753469653427601,
      "learning_rate": 8.784541424647869e-07,
      "loss": 0.0006,
      "step": 258070
    },
    {
      "epoch": 1.6552449548315926,
      "grad_norm": 0.1216263622045517,
      "learning_rate": 8.78137297874192e-07,
      "loss": 0.001,
      "step": 258080
    },
    {
      "epoch": 1.6553090917253788,
      "grad_norm": 0.026845403015613556,
      "learning_rate": 8.778205049334825e-07,
      "loss": 0.0013,
      "step": 258090
    },
    {
      "epoch": 1.6553732286191647,
      "grad_norm": 0.04864067956805229,
      "learning_rate": 8.775037636466271e-07,
      "loss": 0.0011,
      "step": 258100
    },
    {
      "epoch": 1.6554373655129508,
      "grad_norm": 0.03315316513180733,
      "learning_rate": 8.77187074017597e-07,
      "loss": 0.0019,
      "step": 258110
    },
    {
      "epoch": 1.655501502406737,
      "grad_norm": 0.08306470513343811,
      "learning_rate": 8.768704360503594e-07,
      "loss": 0.0008,
      "step": 258120
    },
    {
      "epoch": 1.655565639300523,
      "grad_norm": 0.07637351006269455,
      "learning_rate": 8.765538497488818e-07,
      "loss": 0.0013,
      "step": 258130
    },
    {
      "epoch": 1.655629776194309,
      "grad_norm": 0.1510131061077118,
      "learning_rate": 8.762373151171305e-07,
      "loss": 0.0026,
      "step": 258140
    },
    {
      "epoch": 1.6556939130880952,
      "grad_norm": 0.1989763081073761,
      "learning_rate": 8.759208321590723e-07,
      "loss": 0.001,
      "step": 258150
    },
    {
      "epoch": 1.6557580499818814,
      "grad_norm": 0.07850760966539383,
      "learning_rate": 8.756044008786751e-07,
      "loss": 0.0009,
      "step": 258160
    },
    {
      "epoch": 1.6558221868756675,
      "grad_norm": 0.03400076925754547,
      "learning_rate": 8.752880212799025e-07,
      "loss": 0.001,
      "step": 258170
    },
    {
      "epoch": 1.6558863237694537,
      "grad_norm": 0.04656525328755379,
      "learning_rate": 8.74971693366719e-07,
      "loss": 0.0007,
      "step": 258180
    },
    {
      "epoch": 1.6559504606632396,
      "grad_norm": 0.12720754742622375,
      "learning_rate": 8.746554171430871e-07,
      "loss": 0.0011,
      "step": 258190
    },
    {
      "epoch": 1.6560145975570257,
      "grad_norm": 0.05890100821852684,
      "learning_rate": 8.743391926129723e-07,
      "loss": 0.002,
      "step": 258200
    },
    {
      "epoch": 1.6560787344508117,
      "grad_norm": 0.19628125429153442,
      "learning_rate": 8.740230197803356e-07,
      "loss": 0.0012,
      "step": 258210
    },
    {
      "epoch": 1.6561428713445978,
      "grad_norm": 0.04698095843195915,
      "learning_rate": 8.737068986491388e-07,
      "loss": 0.0012,
      "step": 258220
    },
    {
      "epoch": 1.656207008238384,
      "grad_norm": 0.031655896455049515,
      "learning_rate": 8.733908292233439e-07,
      "loss": 0.0007,
      "step": 258230
    },
    {
      "epoch": 1.65627114513217,
      "grad_norm": 0.11383464187383652,
      "learning_rate": 8.730748115069116e-07,
      "loss": 0.0011,
      "step": 258240
    },
    {
      "epoch": 1.6563352820259563,
      "grad_norm": 0.015196473337709904,
      "learning_rate": 8.727588455038017e-07,
      "loss": 0.002,
      "step": 258250
    },
    {
      "epoch": 1.6563994189197424,
      "grad_norm": 0.18308258056640625,
      "learning_rate": 8.724429312179716e-07,
      "loss": 0.0013,
      "step": 258260
    },
    {
      "epoch": 1.6564635558135286,
      "grad_norm": 0.007341694552451372,
      "learning_rate": 8.721270686533823e-07,
      "loss": 0.0011,
      "step": 258270
    },
    {
      "epoch": 1.6565276927073145,
      "grad_norm": 0.21798130869865417,
      "learning_rate": 8.718112578139915e-07,
      "loss": 0.001,
      "step": 258280
    },
    {
      "epoch": 1.6565918296011006,
      "grad_norm": 0.0013603041879832745,
      "learning_rate": 8.714954987037561e-07,
      "loss": 0.0009,
      "step": 258290
    },
    {
      "epoch": 1.6566559664948866,
      "grad_norm": 0.11801481246948242,
      "learning_rate": 8.711797913266312e-07,
      "loss": 0.0009,
      "step": 258300
    },
    {
      "epoch": 1.6567201033886727,
      "grad_norm": 0.0527237206697464,
      "learning_rate": 8.708641356865755e-07,
      "loss": 0.0008,
      "step": 258310
    },
    {
      "epoch": 1.6567842402824589,
      "grad_norm": 0.04094007611274719,
      "learning_rate": 8.705485317875434e-07,
      "loss": 0.0013,
      "step": 258320
    },
    {
      "epoch": 1.656848377176245,
      "grad_norm": 0.04453378543257713,
      "learning_rate": 8.702329796334896e-07,
      "loss": 0.0014,
      "step": 258330
    },
    {
      "epoch": 1.6569125140700312,
      "grad_norm": 0.13029615581035614,
      "learning_rate": 8.699174792283666e-07,
      "loss": 0.0027,
      "step": 258340
    },
    {
      "epoch": 1.6569766509638173,
      "grad_norm": 0.11282777786254883,
      "learning_rate": 8.696020305761305e-07,
      "loss": 0.0017,
      "step": 258350
    },
    {
      "epoch": 1.6570407878576032,
      "grad_norm": 0.14039598405361176,
      "learning_rate": 8.692866336807332e-07,
      "loss": 0.0018,
      "step": 258360
    },
    {
      "epoch": 1.6571049247513894,
      "grad_norm": 0.05331820994615555,
      "learning_rate": 8.689712885461249e-07,
      "loss": 0.0008,
      "step": 258370
    },
    {
      "epoch": 1.6571690616451753,
      "grad_norm": 0.025747543200850487,
      "learning_rate": 8.686559951762602e-07,
      "loss": 0.0011,
      "step": 258380
    },
    {
      "epoch": 1.6572331985389614,
      "grad_norm": 0.08664846420288086,
      "learning_rate": 8.683407535750887e-07,
      "loss": 0.0009,
      "step": 258390
    },
    {
      "epoch": 1.6572973354327476,
      "grad_norm": 0.07032646983861923,
      "learning_rate": 8.680255637465601e-07,
      "loss": 0.001,
      "step": 258400
    },
    {
      "epoch": 1.6573614723265337,
      "grad_norm": 0.08050806075334549,
      "learning_rate": 8.677104256946233e-07,
      "loss": 0.0008,
      "step": 258410
    },
    {
      "epoch": 1.65742560922032,
      "grad_norm": 0.07885205745697021,
      "learning_rate": 8.673953394232293e-07,
      "loss": 0.0007,
      "step": 258420
    },
    {
      "epoch": 1.657489746114106,
      "grad_norm": 0.056627798825502396,
      "learning_rate": 8.670803049363252e-07,
      "loss": 0.0018,
      "step": 258430
    },
    {
      "epoch": 1.6575538830078922,
      "grad_norm": 0.09750901162624359,
      "learning_rate": 8.667653222378586e-07,
      "loss": 0.002,
      "step": 258440
    },
    {
      "epoch": 1.6576180199016781,
      "grad_norm": 0.015911463648080826,
      "learning_rate": 8.664503913317751e-07,
      "loss": 0.0008,
      "step": 258450
    },
    {
      "epoch": 1.6576821567954643,
      "grad_norm": 0.03874349594116211,
      "learning_rate": 8.661355122220239e-07,
      "loss": 0.0014,
      "step": 258460
    },
    {
      "epoch": 1.6577462936892502,
      "grad_norm": 0.09996067732572556,
      "learning_rate": 8.658206849125495e-07,
      "loss": 0.001,
      "step": 258470
    },
    {
      "epoch": 1.6578104305830363,
      "grad_norm": 0.07658535242080688,
      "learning_rate": 8.65505909407296e-07,
      "loss": 0.0026,
      "step": 258480
    },
    {
      "epoch": 1.6578745674768225,
      "grad_norm": 0.0181189626455307,
      "learning_rate": 8.651911857102075e-07,
      "loss": 0.0021,
      "step": 258490
    },
    {
      "epoch": 1.6579387043706086,
      "grad_norm": 0.024657348170876503,
      "learning_rate": 8.64876513825229e-07,
      "loss": 0.0012,
      "step": 258500
    },
    {
      "epoch": 1.6580028412643948,
      "grad_norm": 0.1810661256313324,
      "learning_rate": 8.645618937563039e-07,
      "loss": 0.0011,
      "step": 258510
    },
    {
      "epoch": 1.658066978158181,
      "grad_norm": 0.068379245698452,
      "learning_rate": 8.642473255073725e-07,
      "loss": 0.0013,
      "step": 258520
    },
    {
      "epoch": 1.6581311150519669,
      "grad_norm": 0.0052925460040569305,
      "learning_rate": 8.639328090823789e-07,
      "loss": 0.0005,
      "step": 258530
    },
    {
      "epoch": 1.658195251945753,
      "grad_norm": 0.01999668963253498,
      "learning_rate": 8.636183444852631e-07,
      "loss": 0.0023,
      "step": 258540
    },
    {
      "epoch": 1.6582593888395392,
      "grad_norm": 0.059351321309804916,
      "learning_rate": 8.63303931719966e-07,
      "loss": 0.0009,
      "step": 258550
    },
    {
      "epoch": 1.658323525733325,
      "grad_norm": 0.12230983376502991,
      "learning_rate": 8.629895707904256e-07,
      "loss": 0.003,
      "step": 258560
    },
    {
      "epoch": 1.6583876626271112,
      "grad_norm": 0.11306685209274292,
      "learning_rate": 8.62675261700584e-07,
      "loss": 0.0008,
      "step": 258570
    },
    {
      "epoch": 1.6584517995208974,
      "grad_norm": 0.038956791162490845,
      "learning_rate": 8.623610044543779e-07,
      "loss": 0.0025,
      "step": 258580
    },
    {
      "epoch": 1.6585159364146835,
      "grad_norm": 0.10281708836555481,
      "learning_rate": 8.620467990557458e-07,
      "loss": 0.0021,
      "step": 258590
    },
    {
      "epoch": 1.6585800733084697,
      "grad_norm": 0.08976875990629196,
      "learning_rate": 8.617326455086234e-07,
      "loss": 0.0012,
      "step": 258600
    },
    {
      "epoch": 1.6586442102022558,
      "grad_norm": 0.21373844146728516,
      "learning_rate": 8.614185438169498e-07,
      "loss": 0.0017,
      "step": 258610
    },
    {
      "epoch": 1.6587083470960418,
      "grad_norm": 0.05366050451993942,
      "learning_rate": 8.611044939846597e-07,
      "loss": 0.0009,
      "step": 258620
    },
    {
      "epoch": 1.658772483989828,
      "grad_norm": 0.26793211698532104,
      "learning_rate": 8.607904960156882e-07,
      "loss": 0.0014,
      "step": 258630
    },
    {
      "epoch": 1.6588366208836138,
      "grad_norm": 0.05795462056994438,
      "learning_rate": 8.604765499139689e-07,
      "loss": 0.0011,
      "step": 258640
    },
    {
      "epoch": 1.6589007577774,
      "grad_norm": 0.018418608233332634,
      "learning_rate": 8.60162655683438e-07,
      "loss": 0.0005,
      "step": 258650
    },
    {
      "epoch": 1.6589648946711861,
      "grad_norm": 0.07220429927110672,
      "learning_rate": 8.598488133280281e-07,
      "loss": 0.0008,
      "step": 258660
    },
    {
      "epoch": 1.6590290315649723,
      "grad_norm": 0.17509478330612183,
      "learning_rate": 8.595350228516697e-07,
      "loss": 0.0009,
      "step": 258670
    },
    {
      "epoch": 1.6590931684587584,
      "grad_norm": 0.018411090597510338,
      "learning_rate": 8.592212842582981e-07,
      "loss": 0.0014,
      "step": 258680
    },
    {
      "epoch": 1.6591573053525446,
      "grad_norm": 0.10979661345481873,
      "learning_rate": 8.589075975518435e-07,
      "loss": 0.0007,
      "step": 258690
    },
    {
      "epoch": 1.6592214422463307,
      "grad_norm": 0.015542004257440567,
      "learning_rate": 8.585939627362356e-07,
      "loss": 0.0009,
      "step": 258700
    },
    {
      "epoch": 1.6592855791401167,
      "grad_norm": 0.09888337552547455,
      "learning_rate": 8.582803798154044e-07,
      "loss": 0.0012,
      "step": 258710
    },
    {
      "epoch": 1.6593497160339028,
      "grad_norm": 0.08750156313180923,
      "learning_rate": 8.579668487932813e-07,
      "loss": 0.0006,
      "step": 258720
    },
    {
      "epoch": 1.6594138529276887,
      "grad_norm": 0.04003286361694336,
      "learning_rate": 8.576533696737937e-07,
      "loss": 0.0007,
      "step": 258730
    },
    {
      "epoch": 1.6594779898214749,
      "grad_norm": 0.09341073781251907,
      "learning_rate": 8.573399424608703e-07,
      "loss": 0.0013,
      "step": 258740
    },
    {
      "epoch": 1.659542126715261,
      "grad_norm": 0.12593533098697662,
      "learning_rate": 8.570265671584361e-07,
      "loss": 0.0013,
      "step": 258750
    },
    {
      "epoch": 1.6596062636090472,
      "grad_norm": 0.08734682947397232,
      "learning_rate": 8.567132437704217e-07,
      "loss": 0.0007,
      "step": 258760
    },
    {
      "epoch": 1.6596704005028333,
      "grad_norm": 0.03187654912471771,
      "learning_rate": 8.563999723007515e-07,
      "loss": 0.0009,
      "step": 258770
    },
    {
      "epoch": 1.6597345373966195,
      "grad_norm": 0.07724946737289429,
      "learning_rate": 8.560867527533512e-07,
      "loss": 0.0006,
      "step": 258780
    },
    {
      "epoch": 1.6597986742904054,
      "grad_norm": 0.09878451377153397,
      "learning_rate": 8.557735851321442e-07,
      "loss": 0.0013,
      "step": 258790
    },
    {
      "epoch": 1.6598628111841915,
      "grad_norm": 0.005886028986424208,
      "learning_rate": 8.554604694410568e-07,
      "loss": 0.0007,
      "step": 258800
    },
    {
      "epoch": 1.6599269480779775,
      "grad_norm": 0.05878474935889244,
      "learning_rate": 8.55147405684012e-07,
      "loss": 0.0008,
      "step": 258810
    },
    {
      "epoch": 1.6599910849717636,
      "grad_norm": 0.061283230781555176,
      "learning_rate": 8.548343938649312e-07,
      "loss": 0.0013,
      "step": 258820
    },
    {
      "epoch": 1.6600552218655498,
      "grad_norm": 0.06955339014530182,
      "learning_rate": 8.545214339877395e-07,
      "loss": 0.0013,
      "step": 258830
    },
    {
      "epoch": 1.660119358759336,
      "grad_norm": 0.054750386625528336,
      "learning_rate": 8.542085260563565e-07,
      "loss": 0.0016,
      "step": 258840
    },
    {
      "epoch": 1.660183495653122,
      "grad_norm": 0.0939524918794632,
      "learning_rate": 8.538956700747042e-07,
      "loss": 0.0005,
      "step": 258850
    },
    {
      "epoch": 1.6602476325469082,
      "grad_norm": 0.10427170246839523,
      "learning_rate": 8.535828660467005e-07,
      "loss": 0.0018,
      "step": 258860
    },
    {
      "epoch": 1.6603117694406944,
      "grad_norm": 0.18324019014835358,
      "learning_rate": 8.532701139762683e-07,
      "loss": 0.0015,
      "step": 258870
    },
    {
      "epoch": 1.6603759063344803,
      "grad_norm": 0.03328492119908333,
      "learning_rate": 8.529574138673258e-07,
      "loss": 0.001,
      "step": 258880
    },
    {
      "epoch": 1.6604400432282664,
      "grad_norm": 0.058263231068849564,
      "learning_rate": 8.526447657237901e-07,
      "loss": 0.0011,
      "step": 258890
    },
    {
      "epoch": 1.6605041801220524,
      "grad_norm": 0.09907954931259155,
      "learning_rate": 8.523321695495785e-07,
      "loss": 0.0005,
      "step": 258900
    },
    {
      "epoch": 1.6605683170158385,
      "grad_norm": 0.13083921372890472,
      "learning_rate": 8.52019625348609e-07,
      "loss": 0.001,
      "step": 258910
    },
    {
      "epoch": 1.6606324539096247,
      "grad_norm": 0.008092692121863365,
      "learning_rate": 8.517071331247995e-07,
      "loss": 0.001,
      "step": 258920
    },
    {
      "epoch": 1.6606965908034108,
      "grad_norm": 0.08676736056804657,
      "learning_rate": 8.513946928820644e-07,
      "loss": 0.001,
      "step": 258930
    },
    {
      "epoch": 1.660760727697197,
      "grad_norm": 0.05568745732307434,
      "learning_rate": 8.510823046243171e-07,
      "loss": 0.0018,
      "step": 258940
    },
    {
      "epoch": 1.660824864590983,
      "grad_norm": 0.08479592204093933,
      "learning_rate": 8.507699683554755e-07,
      "loss": 0.0013,
      "step": 258950
    },
    {
      "epoch": 1.6608890014847693,
      "grad_norm": 0.27204254269599915,
      "learning_rate": 8.504576840794516e-07,
      "loss": 0.0009,
      "step": 258960
    },
    {
      "epoch": 1.6609531383785552,
      "grad_norm": 0.09179872274398804,
      "learning_rate": 8.501454518001579e-07,
      "loss": 0.0009,
      "step": 258970
    },
    {
      "epoch": 1.6610172752723413,
      "grad_norm": 0.08390912413597107,
      "learning_rate": 8.498332715215068e-07,
      "loss": 0.001,
      "step": 258980
    },
    {
      "epoch": 1.6610814121661273,
      "grad_norm": 0.007385471370071173,
      "learning_rate": 8.495211432474121e-07,
      "loss": 0.0014,
      "step": 258990
    },
    {
      "epoch": 1.6611455490599134,
      "grad_norm": 0.03575975447893143,
      "learning_rate": 8.492090669817837e-07,
      "loss": 0.0009,
      "step": 259000
    },
    {
      "epoch": 1.6612096859536996,
      "grad_norm": 0.12160173803567886,
      "learning_rate": 8.48897042728532e-07,
      "loss": 0.0014,
      "step": 259010
    },
    {
      "epoch": 1.6612738228474857,
      "grad_norm": 0.13452289998531342,
      "learning_rate": 8.485850704915665e-07,
      "loss": 0.0019,
      "step": 259020
    },
    {
      "epoch": 1.6613379597412719,
      "grad_norm": 0.12849852442741394,
      "learning_rate": 8.482731502747976e-07,
      "loss": 0.0012,
      "step": 259030
    },
    {
      "epoch": 1.661402096635058,
      "grad_norm": 0.028622470796108246,
      "learning_rate": 8.479612820821337e-07,
      "loss": 0.0019,
      "step": 259040
    },
    {
      "epoch": 1.661466233528844,
      "grad_norm": 0.11926902085542679,
      "learning_rate": 8.476494659174822e-07,
      "loss": 0.001,
      "step": 259050
    },
    {
      "epoch": 1.66153037042263,
      "grad_norm": 0.11900042742490768,
      "learning_rate": 8.473377017847495e-07,
      "loss": 0.0015,
      "step": 259060
    },
    {
      "epoch": 1.661594507316416,
      "grad_norm": 0.09407677501440048,
      "learning_rate": 8.470259896878442e-07,
      "loss": 0.0021,
      "step": 259070
    },
    {
      "epoch": 1.6616586442102022,
      "grad_norm": 0.09666857123374939,
      "learning_rate": 8.467143296306718e-07,
      "loss": 0.001,
      "step": 259080
    },
    {
      "epoch": 1.6617227811039883,
      "grad_norm": 0.0920310914516449,
      "learning_rate": 8.464027216171356e-07,
      "loss": 0.0015,
      "step": 259090
    },
    {
      "epoch": 1.6617869179977744,
      "grad_norm": 0.09433083236217499,
      "learning_rate": 8.460911656511428e-07,
      "loss": 0.0009,
      "step": 259100
    },
    {
      "epoch": 1.6618510548915606,
      "grad_norm": 0.13251636922359467,
      "learning_rate": 8.45779661736597e-07,
      "loss": 0.0015,
      "step": 259110
    },
    {
      "epoch": 1.6619151917853467,
      "grad_norm": 0.010854779742658138,
      "learning_rate": 8.454682098774003e-07,
      "loss": 0.0002,
      "step": 259120
    },
    {
      "epoch": 1.661979328679133,
      "grad_norm": 0.11809447407722473,
      "learning_rate": 8.451568100774554e-07,
      "loss": 0.0013,
      "step": 259130
    },
    {
      "epoch": 1.6620434655729188,
      "grad_norm": 0.19234363734722137,
      "learning_rate": 8.448454623406666e-07,
      "loss": 0.0006,
      "step": 259140
    },
    {
      "epoch": 1.662107602466705,
      "grad_norm": 0.027598833665251732,
      "learning_rate": 8.445341666709334e-07,
      "loss": 0.0028,
      "step": 259150
    },
    {
      "epoch": 1.662171739360491,
      "grad_norm": 0.03952634334564209,
      "learning_rate": 8.442229230721572e-07,
      "loss": 0.0016,
      "step": 259160
    },
    {
      "epoch": 1.662235876254277,
      "grad_norm": 0.10926931351423264,
      "learning_rate": 8.439117315482359e-07,
      "loss": 0.0004,
      "step": 259170
    },
    {
      "epoch": 1.6623000131480632,
      "grad_norm": 0.061621829867362976,
      "learning_rate": 8.436005921030732e-07,
      "loss": 0.0018,
      "step": 259180
    },
    {
      "epoch": 1.6623641500418493,
      "grad_norm": 0.025558849796652794,
      "learning_rate": 8.432895047405648e-07,
      "loss": 0.0018,
      "step": 259190
    },
    {
      "epoch": 1.6624282869356355,
      "grad_norm": 0.010318063199520111,
      "learning_rate": 8.429784694646098e-07,
      "loss": 0.0013,
      "step": 259200
    },
    {
      "epoch": 1.6624924238294216,
      "grad_norm": 0.10450685024261475,
      "learning_rate": 8.426674862791046e-07,
      "loss": 0.0012,
      "step": 259210
    },
    {
      "epoch": 1.6625565607232076,
      "grad_norm": 0.0915796086192131,
      "learning_rate": 8.423565551879482e-07,
      "loss": 0.0021,
      "step": 259220
    },
    {
      "epoch": 1.6626206976169937,
      "grad_norm": 0.12747132778167725,
      "learning_rate": 8.420456761950357e-07,
      "loss": 0.001,
      "step": 259230
    },
    {
      "epoch": 1.6626848345107799,
      "grad_norm": 0.19605652987957,
      "learning_rate": 8.417348493042609e-07,
      "loss": 0.0021,
      "step": 259240
    },
    {
      "epoch": 1.6627489714045658,
      "grad_norm": 0.033247210085392,
      "learning_rate": 8.414240745195218e-07,
      "loss": 0.0019,
      "step": 259250
    },
    {
      "epoch": 1.662813108298352,
      "grad_norm": 0.12935180962085724,
      "learning_rate": 8.411133518447113e-07,
      "loss": 0.0008,
      "step": 259260
    },
    {
      "epoch": 1.662877245192138,
      "grad_norm": 0.06682842969894409,
      "learning_rate": 8.408026812837222e-07,
      "loss": 0.0011,
      "step": 259270
    },
    {
      "epoch": 1.6629413820859242,
      "grad_norm": 0.07753019034862518,
      "learning_rate": 8.404920628404473e-07,
      "loss": 0.0015,
      "step": 259280
    },
    {
      "epoch": 1.6630055189797104,
      "grad_norm": 0.030806515365839005,
      "learning_rate": 8.40181496518781e-07,
      "loss": 0.0012,
      "step": 259290
    },
    {
      "epoch": 1.6630696558734965,
      "grad_norm": 0.028927655890583992,
      "learning_rate": 8.398709823226131e-07,
      "loss": 0.0013,
      "step": 259300
    },
    {
      "epoch": 1.6631337927672825,
      "grad_norm": 0.3859172463417053,
      "learning_rate": 8.395605202558349e-07,
      "loss": 0.0028,
      "step": 259310
    },
    {
      "epoch": 1.6631979296610686,
      "grad_norm": 0.004698690492659807,
      "learning_rate": 8.392501103223361e-07,
      "loss": 0.0005,
      "step": 259320
    },
    {
      "epoch": 1.6632620665548545,
      "grad_norm": 0.18692345917224884,
      "learning_rate": 8.389397525260079e-07,
      "loss": 0.0013,
      "step": 259330
    },
    {
      "epoch": 1.6633262034486407,
      "grad_norm": 0.10133519023656845,
      "learning_rate": 8.386294468707384e-07,
      "loss": 0.0025,
      "step": 259340
    },
    {
      "epoch": 1.6633903403424268,
      "grad_norm": 0.058625735342502594,
      "learning_rate": 8.383191933604151e-07,
      "loss": 0.0006,
      "step": 259350
    },
    {
      "epoch": 1.663454477236213,
      "grad_norm": 0.14901982247829437,
      "learning_rate": 8.380089919989276e-07,
      "loss": 0.0015,
      "step": 259360
    },
    {
      "epoch": 1.6635186141299991,
      "grad_norm": 0.03865260258316994,
      "learning_rate": 8.376988427901622e-07,
      "loss": 0.0039,
      "step": 259370
    },
    {
      "epoch": 1.6635827510237853,
      "grad_norm": 0.03860464319586754,
      "learning_rate": 8.373887457380047e-07,
      "loss": 0.0011,
      "step": 259380
    },
    {
      "epoch": 1.6636468879175714,
      "grad_norm": 0.06766288727521896,
      "learning_rate": 8.370787008463404e-07,
      "loss": 0.0011,
      "step": 259390
    },
    {
      "epoch": 1.6637110248113574,
      "grad_norm": 0.07876670360565186,
      "learning_rate": 8.367687081190562e-07,
      "loss": 0.0013,
      "step": 259400
    },
    {
      "epoch": 1.6637751617051435,
      "grad_norm": 0.13151301443576813,
      "learning_rate": 8.364587675600355e-07,
      "loss": 0.0018,
      "step": 259410
    },
    {
      "epoch": 1.6638392985989294,
      "grad_norm": 0.040049389004707336,
      "learning_rate": 8.36148879173162e-07,
      "loss": 0.0006,
      "step": 259420
    },
    {
      "epoch": 1.6639034354927156,
      "grad_norm": 0.057279810309410095,
      "learning_rate": 8.358390429623181e-07,
      "loss": 0.0008,
      "step": 259430
    },
    {
      "epoch": 1.6639675723865017,
      "grad_norm": 0.11955063790082932,
      "learning_rate": 8.355292589313879e-07,
      "loss": 0.0009,
      "step": 259440
    },
    {
      "epoch": 1.6640317092802879,
      "grad_norm": 0.1107025220990181,
      "learning_rate": 8.352195270842522e-07,
      "loss": 0.0013,
      "step": 259450
    },
    {
      "epoch": 1.664095846174074,
      "grad_norm": 0.048308372497558594,
      "learning_rate": 8.349098474247924e-07,
      "loss": 0.0007,
      "step": 259460
    },
    {
      "epoch": 1.6641599830678602,
      "grad_norm": 0.03436093404889107,
      "learning_rate": 8.346002199568881e-07,
      "loss": 0.0009,
      "step": 259470
    },
    {
      "epoch": 1.664224119961646,
      "grad_norm": 0.0168591495603323,
      "learning_rate": 8.342906446844212e-07,
      "loss": 0.0051,
      "step": 259480
    },
    {
      "epoch": 1.6642882568554322,
      "grad_norm": 0.07112542539834976,
      "learning_rate": 8.339811216112698e-07,
      "loss": 0.0011,
      "step": 259490
    },
    {
      "epoch": 1.6643523937492182,
      "grad_norm": 0.13269291818141937,
      "learning_rate": 8.336716507413107e-07,
      "loss": 0.0016,
      "step": 259500
    },
    {
      "epoch": 1.6644165306430043,
      "grad_norm": 0.1593928039073944,
      "learning_rate": 8.333622320784246e-07,
      "loss": 0.0038,
      "step": 259510
    },
    {
      "epoch": 1.6644806675367905,
      "grad_norm": 0.09930547326803207,
      "learning_rate": 8.33052865626488e-07,
      "loss": 0.0011,
      "step": 259520
    },
    {
      "epoch": 1.6645448044305766,
      "grad_norm": 0.034947656095027924,
      "learning_rate": 8.327435513893767e-07,
      "loss": 0.0009,
      "step": 259530
    },
    {
      "epoch": 1.6646089413243628,
      "grad_norm": 0.05803503096103668,
      "learning_rate": 8.324342893709664e-07,
      "loss": 0.0015,
      "step": 259540
    },
    {
      "epoch": 1.664673078218149,
      "grad_norm": 0.05112419277429581,
      "learning_rate": 8.321250795751334e-07,
      "loss": 0.0005,
      "step": 259550
    },
    {
      "epoch": 1.664737215111935,
      "grad_norm": 0.04191881790757179,
      "learning_rate": 8.318159220057526e-07,
      "loss": 0.0005,
      "step": 259560
    },
    {
      "epoch": 1.664801352005721,
      "grad_norm": 0.08693952113389969,
      "learning_rate": 8.315068166666967e-07,
      "loss": 0.0023,
      "step": 259570
    },
    {
      "epoch": 1.6648654888995071,
      "grad_norm": 0.025236818939447403,
      "learning_rate": 8.311977635618385e-07,
      "loss": 0.0009,
      "step": 259580
    },
    {
      "epoch": 1.664929625793293,
      "grad_norm": 0.05903942137956619,
      "learning_rate": 8.30888762695053e-07,
      "loss": 0.001,
      "step": 259590
    },
    {
      "epoch": 1.6649937626870792,
      "grad_norm": 0.06373009830713272,
      "learning_rate": 8.30579814070211e-07,
      "loss": 0.0009,
      "step": 259600
    },
    {
      "epoch": 1.6650578995808654,
      "grad_norm": 0.040365107357501984,
      "learning_rate": 8.302709176911838e-07,
      "loss": 0.0013,
      "step": 259610
    },
    {
      "epoch": 1.6651220364746515,
      "grad_norm": 0.13631953299045563,
      "learning_rate": 8.299620735618408e-07,
      "loss": 0.0016,
      "step": 259620
    },
    {
      "epoch": 1.6651861733684377,
      "grad_norm": 0.14050503075122833,
      "learning_rate": 8.296532816860542e-07,
      "loss": 0.001,
      "step": 259630
    },
    {
      "epoch": 1.6652503102622238,
      "grad_norm": 0.07822521030902863,
      "learning_rate": 8.293445420676927e-07,
      "loss": 0.0011,
      "step": 259640
    },
    {
      "epoch": 1.6653144471560097,
      "grad_norm": 0.05127008259296417,
      "learning_rate": 8.290358547106237e-07,
      "loss": 0.001,
      "step": 259650
    },
    {
      "epoch": 1.6653785840497959,
      "grad_norm": 0.07713084667921066,
      "learning_rate": 8.287272196187174e-07,
      "loss": 0.0012,
      "step": 259660
    },
    {
      "epoch": 1.665442720943582,
      "grad_norm": 0.07243801653385162,
      "learning_rate": 8.284186367958402e-07,
      "loss": 0.0028,
      "step": 259670
    },
    {
      "epoch": 1.665506857837368,
      "grad_norm": 0.046422865241765976,
      "learning_rate": 8.28110106245858e-07,
      "loss": 0.0008,
      "step": 259680
    },
    {
      "epoch": 1.665570994731154,
      "grad_norm": 0.3993385136127472,
      "learning_rate": 8.278016279726386e-07,
      "loss": 0.0012,
      "step": 259690
    },
    {
      "epoch": 1.6656351316249403,
      "grad_norm": 0.011559012345969677,
      "learning_rate": 8.274932019800458e-07,
      "loss": 0.0011,
      "step": 259700
    },
    {
      "epoch": 1.6656992685187264,
      "grad_norm": 0.20504416525363922,
      "learning_rate": 8.271848282719463e-07,
      "loss": 0.0024,
      "step": 259710
    },
    {
      "epoch": 1.6657634054125126,
      "grad_norm": 0.007328205741941929,
      "learning_rate": 8.268765068522028e-07,
      "loss": 0.0012,
      "step": 259720
    },
    {
      "epoch": 1.6658275423062987,
      "grad_norm": 0.06418116390705109,
      "learning_rate": 8.265682377246797e-07,
      "loss": 0.002,
      "step": 259730
    },
    {
      "epoch": 1.6658916792000846,
      "grad_norm": 0.0644039437174797,
      "learning_rate": 8.262600208932375e-07,
      "loss": 0.0013,
      "step": 259740
    },
    {
      "epoch": 1.6659558160938708,
      "grad_norm": 0.0812007486820221,
      "learning_rate": 8.259518563617419e-07,
      "loss": 0.001,
      "step": 259750
    },
    {
      "epoch": 1.6660199529876567,
      "grad_norm": 0.22765526175498962,
      "learning_rate": 8.256437441340525e-07,
      "loss": 0.0044,
      "step": 259760
    },
    {
      "epoch": 1.6660840898814429,
      "grad_norm": 0.0717020258307457,
      "learning_rate": 8.253356842140293e-07,
      "loss": 0.0008,
      "step": 259770
    },
    {
      "epoch": 1.666148226775229,
      "grad_norm": 0.11149182170629501,
      "learning_rate": 8.250276766055343e-07,
      "loss": 0.0014,
      "step": 259780
    },
    {
      "epoch": 1.6662123636690152,
      "grad_norm": 0.05467083305120468,
      "learning_rate": 8.247197213124269e-07,
      "loss": 0.0007,
      "step": 259790
    },
    {
      "epoch": 1.6662765005628013,
      "grad_norm": 0.030334288254380226,
      "learning_rate": 8.244118183385652e-07,
      "loss": 0.0021,
      "step": 259800
    },
    {
      "epoch": 1.6663406374565874,
      "grad_norm": 0.044191643595695496,
      "learning_rate": 8.24103967687806e-07,
      "loss": 0.0012,
      "step": 259810
    },
    {
      "epoch": 1.6664047743503736,
      "grad_norm": 0.047531258314847946,
      "learning_rate": 8.237961693640101e-07,
      "loss": 0.0015,
      "step": 259820
    },
    {
      "epoch": 1.6664689112441595,
      "grad_norm": 0.10695334523916245,
      "learning_rate": 8.234884233710333e-07,
      "loss": 0.0012,
      "step": 259830
    },
    {
      "epoch": 1.6665330481379457,
      "grad_norm": 0.06404314935207367,
      "learning_rate": 8.231807297127309e-07,
      "loss": 0.0014,
      "step": 259840
    },
    {
      "epoch": 1.6665971850317316,
      "grad_norm": 0.05008331313729286,
      "learning_rate": 8.228730883929581e-07,
      "loss": 0.0011,
      "step": 259850
    },
    {
      "epoch": 1.6666613219255177,
      "grad_norm": 0.0490146204829216,
      "learning_rate": 8.225654994155718e-07,
      "loss": 0.0019,
      "step": 259860
    },
    {
      "epoch": 1.666725458819304,
      "grad_norm": 0.002646214794367552,
      "learning_rate": 8.222579627844257e-07,
      "loss": 0.0011,
      "step": 259870
    },
    {
      "epoch": 1.66678959571309,
      "grad_norm": 0.008767073974013329,
      "learning_rate": 8.219504785033733e-07,
      "loss": 0.0018,
      "step": 259880
    },
    {
      "epoch": 1.6668537326068762,
      "grad_norm": 0.009336371906101704,
      "learning_rate": 8.216430465762659e-07,
      "loss": 0.001,
      "step": 259890
    },
    {
      "epoch": 1.6669178695006623,
      "grad_norm": 0.047865401953458786,
      "learning_rate": 8.213356670069583e-07,
      "loss": 0.0021,
      "step": 259900
    },
    {
      "epoch": 1.6669820063944483,
      "grad_norm": 0.023457759991288185,
      "learning_rate": 8.210283397993018e-07,
      "loss": 0.0012,
      "step": 259910
    },
    {
      "epoch": 1.6670461432882344,
      "grad_norm": 0.20852166414260864,
      "learning_rate": 8.207210649571451e-07,
      "loss": 0.0031,
      "step": 259920
    },
    {
      "epoch": 1.6671102801820203,
      "grad_norm": 0.017165524885058403,
      "learning_rate": 8.204138424843422e-07,
      "loss": 0.0005,
      "step": 259930
    },
    {
      "epoch": 1.6671744170758065,
      "grad_norm": 0.020751260221004486,
      "learning_rate": 8.201066723847406e-07,
      "loss": 0.0011,
      "step": 259940
    },
    {
      "epoch": 1.6672385539695926,
      "grad_norm": 0.042478736490011215,
      "learning_rate": 8.197995546621895e-07,
      "loss": 0.0004,
      "step": 259950
    },
    {
      "epoch": 1.6673026908633788,
      "grad_norm": 0.05359106883406639,
      "learning_rate": 8.194924893205363e-07,
      "loss": 0.001,
      "step": 259960
    },
    {
      "epoch": 1.667366827757165,
      "grad_norm": 0.1899213045835495,
      "learning_rate": 8.191854763636314e-07,
      "loss": 0.0014,
      "step": 259970
    },
    {
      "epoch": 1.667430964650951,
      "grad_norm": 0.02799876406788826,
      "learning_rate": 8.188785157953205e-07,
      "loss": 0.0007,
      "step": 259980
    },
    {
      "epoch": 1.6674951015447372,
      "grad_norm": 0.0149038415402174,
      "learning_rate": 8.1857160761945e-07,
      "loss": 0.0013,
      "step": 259990
    },
    {
      "epoch": 1.6675592384385232,
      "grad_norm": 0.03285687044262886,
      "learning_rate": 8.182647518398645e-07,
      "loss": 0.0015,
      "step": 260000
    },
    {
      "epoch": 1.6676233753323093,
      "grad_norm": 0.09179037064313889,
      "learning_rate": 8.179579484604116e-07,
      "loss": 0.002,
      "step": 260010
    },
    {
      "epoch": 1.6676875122260952,
      "grad_norm": 0.08945292234420776,
      "learning_rate": 8.176511974849344e-07,
      "loss": 0.0005,
      "step": 260020
    },
    {
      "epoch": 1.6677516491198814,
      "grad_norm": 0.14974276721477509,
      "learning_rate": 8.173444989172763e-07,
      "loss": 0.002,
      "step": 260030
    },
    {
      "epoch": 1.6678157860136675,
      "grad_norm": 0.044630106538534164,
      "learning_rate": 8.170378527612804e-07,
      "loss": 0.0011,
      "step": 260040
    },
    {
      "epoch": 1.6678799229074537,
      "grad_norm": 0.026970086619257927,
      "learning_rate": 8.167312590207905e-07,
      "loss": 0.0006,
      "step": 260050
    },
    {
      "epoch": 1.6679440598012398,
      "grad_norm": 0.019978389143943787,
      "learning_rate": 8.164247176996476e-07,
      "loss": 0.0013,
      "step": 260060
    },
    {
      "epoch": 1.668008196695026,
      "grad_norm": 0.2422911822795868,
      "learning_rate": 8.161182288016922e-07,
      "loss": 0.0009,
      "step": 260070
    },
    {
      "epoch": 1.668072333588812,
      "grad_norm": 0.061068106442689896,
      "learning_rate": 8.158117923307662e-07,
      "loss": 0.0008,
      "step": 260080
    },
    {
      "epoch": 1.668136470482598,
      "grad_norm": 0.020502885803580284,
      "learning_rate": 8.155054082907093e-07,
      "loss": 0.0013,
      "step": 260090
    },
    {
      "epoch": 1.6682006073763842,
      "grad_norm": 0.05441015586256981,
      "learning_rate": 8.151990766853602e-07,
      "loss": 0.0008,
      "step": 260100
    },
    {
      "epoch": 1.6682647442701701,
      "grad_norm": 0.08081910759210587,
      "learning_rate": 8.148927975185561e-07,
      "loss": 0.0013,
      "step": 260110
    },
    {
      "epoch": 1.6683288811639563,
      "grad_norm": 0.0780453011393547,
      "learning_rate": 8.145865707941375e-07,
      "loss": 0.0016,
      "step": 260120
    },
    {
      "epoch": 1.6683930180577424,
      "grad_norm": 0.46676942706108093,
      "learning_rate": 8.142803965159407e-07,
      "loss": 0.0006,
      "step": 260130
    },
    {
      "epoch": 1.6684571549515286,
      "grad_norm": 0.07748177647590637,
      "learning_rate": 8.139742746878021e-07,
      "loss": 0.0019,
      "step": 260140
    },
    {
      "epoch": 1.6685212918453147,
      "grad_norm": 0.05093437433242798,
      "learning_rate": 8.136682053135558e-07,
      "loss": 0.0012,
      "step": 260150
    },
    {
      "epoch": 1.6685854287391009,
      "grad_norm": 0.3314497768878937,
      "learning_rate": 8.133621883970405e-07,
      "loss": 0.0019,
      "step": 260160
    },
    {
      "epoch": 1.6686495656328868,
      "grad_norm": 0.06093122810125351,
      "learning_rate": 8.130562239420886e-07,
      "loss": 0.0011,
      "step": 260170
    },
    {
      "epoch": 1.668713702526673,
      "grad_norm": 0.08686614036560059,
      "learning_rate": 8.127503119525349e-07,
      "loss": 0.0013,
      "step": 260180
    },
    {
      "epoch": 1.6687778394204589,
      "grad_norm": 0.0773782730102539,
      "learning_rate": 8.124444524322111e-07,
      "loss": 0.0023,
      "step": 260190
    },
    {
      "epoch": 1.668841976314245,
      "grad_norm": 0.15026801824569702,
      "learning_rate": 8.121386453849528e-07,
      "loss": 0.0018,
      "step": 260200
    },
    {
      "epoch": 1.6689061132080312,
      "grad_norm": 0.0033452792558819056,
      "learning_rate": 8.118328908145895e-07,
      "loss": 0.0014,
      "step": 260210
    },
    {
      "epoch": 1.6689702501018173,
      "grad_norm": 0.12269829213619232,
      "learning_rate": 8.115271887249527e-07,
      "loss": 0.0008,
      "step": 260220
    },
    {
      "epoch": 1.6690343869956035,
      "grad_norm": 0.050269342958927155,
      "learning_rate": 8.112215391198747e-07,
      "loss": 0.0004,
      "step": 260230
    },
    {
      "epoch": 1.6690985238893896,
      "grad_norm": 0.047745510935783386,
      "learning_rate": 8.109159420031843e-07,
      "loss": 0.0015,
      "step": 260240
    },
    {
      "epoch": 1.6691626607831758,
      "grad_norm": 0.12586195766925812,
      "learning_rate": 8.106103973787116e-07,
      "loss": 0.0015,
      "step": 260250
    },
    {
      "epoch": 1.6692267976769617,
      "grad_norm": 0.22415123879909515,
      "learning_rate": 8.103049052502832e-07,
      "loss": 0.0008,
      "step": 260260
    },
    {
      "epoch": 1.6692909345707478,
      "grad_norm": 0.13841083645820618,
      "learning_rate": 8.099994656217297e-07,
      "loss": 0.0014,
      "step": 260270
    },
    {
      "epoch": 1.6693550714645338,
      "grad_norm": 0.04546148329973221,
      "learning_rate": 8.096940784968776e-07,
      "loss": 0.0013,
      "step": 260280
    },
    {
      "epoch": 1.66941920835832,
      "grad_norm": 0.2795056700706482,
      "learning_rate": 8.093887438795534e-07,
      "loss": 0.0019,
      "step": 260290
    },
    {
      "epoch": 1.669483345252106,
      "grad_norm": 0.18928854167461395,
      "learning_rate": 8.090834617735816e-07,
      "loss": 0.0016,
      "step": 260300
    },
    {
      "epoch": 1.6695474821458922,
      "grad_norm": 0.1277276575565338,
      "learning_rate": 8.087782321827909e-07,
      "loss": 0.0011,
      "step": 260310
    },
    {
      "epoch": 1.6696116190396784,
      "grad_norm": 0.048505738377571106,
      "learning_rate": 8.084730551110043e-07,
      "loss": 0.0012,
      "step": 260320
    },
    {
      "epoch": 1.6696757559334645,
      "grad_norm": 0.040552858263254166,
      "learning_rate": 8.08167930562046e-07,
      "loss": 0.0023,
      "step": 260330
    },
    {
      "epoch": 1.6697398928272504,
      "grad_norm": 0.06651022285223007,
      "learning_rate": 8.078628585397386e-07,
      "loss": 0.001,
      "step": 260340
    },
    {
      "epoch": 1.6698040297210366,
      "grad_norm": 0.13170164823532104,
      "learning_rate": 8.075578390479061e-07,
      "loss": 0.0015,
      "step": 260350
    },
    {
      "epoch": 1.6698681666148225,
      "grad_norm": 0.09963278472423553,
      "learning_rate": 8.072528720903705e-07,
      "loss": 0.0006,
      "step": 260360
    },
    {
      "epoch": 1.6699323035086087,
      "grad_norm": 0.04172990471124649,
      "learning_rate": 8.069479576709521e-07,
      "loss": 0.0013,
      "step": 260370
    },
    {
      "epoch": 1.6699964404023948,
      "grad_norm": 0.0824194923043251,
      "learning_rate": 8.066430957934729e-07,
      "loss": 0.0004,
      "step": 260380
    },
    {
      "epoch": 1.670060577296181,
      "grad_norm": 0.06737431138753891,
      "learning_rate": 8.063382864617536e-07,
      "loss": 0.0012,
      "step": 260390
    },
    {
      "epoch": 1.670124714189967,
      "grad_norm": 0.1556253284215927,
      "learning_rate": 8.060335296796118e-07,
      "loss": 0.0009,
      "step": 260400
    },
    {
      "epoch": 1.6701888510837533,
      "grad_norm": 0.01636998914182186,
      "learning_rate": 8.057288254508667e-07,
      "loss": 0.0008,
      "step": 260410
    },
    {
      "epoch": 1.6702529879775394,
      "grad_norm": 0.010741271078586578,
      "learning_rate": 8.054241737793378e-07,
      "loss": 0.002,
      "step": 260420
    },
    {
      "epoch": 1.6703171248713253,
      "grad_norm": 0.023520736023783684,
      "learning_rate": 8.051195746688423e-07,
      "loss": 0.0008,
      "step": 260430
    },
    {
      "epoch": 1.6703812617651115,
      "grad_norm": 0.03630809485912323,
      "learning_rate": 8.048150281231953e-07,
      "loss": 0.0009,
      "step": 260440
    },
    {
      "epoch": 1.6704453986588974,
      "grad_norm": 0.09315015375614166,
      "learning_rate": 8.04510534146215e-07,
      "loss": 0.0009,
      "step": 260450
    },
    {
      "epoch": 1.6705095355526836,
      "grad_norm": 0.08812452107667923,
      "learning_rate": 8.04206092741715e-07,
      "loss": 0.0013,
      "step": 260460
    },
    {
      "epoch": 1.6705736724464697,
      "grad_norm": 0.10664328932762146,
      "learning_rate": 8.039017039135133e-07,
      "loss": 0.0009,
      "step": 260470
    },
    {
      "epoch": 1.6706378093402559,
      "grad_norm": 0.13169336318969727,
      "learning_rate": 8.035973676654212e-07,
      "loss": 0.0009,
      "step": 260480
    },
    {
      "epoch": 1.670701946234042,
      "grad_norm": 0.19953066110610962,
      "learning_rate": 8.032930840012527e-07,
      "loss": 0.0013,
      "step": 260490
    },
    {
      "epoch": 1.6707660831278281,
      "grad_norm": 0.04290478676557541,
      "learning_rate": 8.02988852924822e-07,
      "loss": 0.0013,
      "step": 260500
    },
    {
      "epoch": 1.6708302200216143,
      "grad_norm": 0.03743245452642441,
      "learning_rate": 8.026846744399408e-07,
      "loss": 0.0012,
      "step": 260510
    },
    {
      "epoch": 1.6708943569154002,
      "grad_norm": 0.1103561744093895,
      "learning_rate": 8.023805485504204e-07,
      "loss": 0.0015,
      "step": 260520
    },
    {
      "epoch": 1.6709584938091864,
      "grad_norm": 0.0029104137793183327,
      "learning_rate": 8.020764752600702e-07,
      "loss": 0.001,
      "step": 260530
    },
    {
      "epoch": 1.6710226307029723,
      "grad_norm": 0.04735827445983887,
      "learning_rate": 8.017724545727035e-07,
      "loss": 0.0012,
      "step": 260540
    },
    {
      "epoch": 1.6710867675967584,
      "grad_norm": 0.12044599652290344,
      "learning_rate": 8.014684864921279e-07,
      "loss": 0.0009,
      "step": 260550
    },
    {
      "epoch": 1.6711509044905446,
      "grad_norm": 0.009900239296257496,
      "learning_rate": 8.011645710221533e-07,
      "loss": 0.0009,
      "step": 260560
    },
    {
      "epoch": 1.6712150413843307,
      "grad_norm": 0.04917309433221817,
      "learning_rate": 8.008607081665864e-07,
      "loss": 0.0005,
      "step": 260570
    },
    {
      "epoch": 1.671279178278117,
      "grad_norm": 0.12651683390140533,
      "learning_rate": 8.005568979292367e-07,
      "loss": 0.0019,
      "step": 260580
    },
    {
      "epoch": 1.671343315171903,
      "grad_norm": 0.0106301698833704,
      "learning_rate": 8.002531403139108e-07,
      "loss": 0.0033,
      "step": 260590
    },
    {
      "epoch": 1.671407452065689,
      "grad_norm": 0.10813277214765549,
      "learning_rate": 7.999494353244142e-07,
      "loss": 0.0009,
      "step": 260600
    },
    {
      "epoch": 1.6714715889594751,
      "grad_norm": 0.05767657235264778,
      "learning_rate": 7.996457829645516e-07,
      "loss": 0.0023,
      "step": 260610
    },
    {
      "epoch": 1.671535725853261,
      "grad_norm": 0.02989642694592476,
      "learning_rate": 7.99342183238131e-07,
      "loss": 0.0009,
      "step": 260620
    },
    {
      "epoch": 1.6715998627470472,
      "grad_norm": 0.053140442818403244,
      "learning_rate": 7.990386361489544e-07,
      "loss": 0.0031,
      "step": 260630
    },
    {
      "epoch": 1.6716639996408333,
      "grad_norm": 0.09720054268836975,
      "learning_rate": 7.987351417008249e-07,
      "loss": 0.0006,
      "step": 260640
    },
    {
      "epoch": 1.6717281365346195,
      "grad_norm": 0.05448118597269058,
      "learning_rate": 7.984316998975484e-07,
      "loss": 0.0018,
      "step": 260650
    },
    {
      "epoch": 1.6717922734284056,
      "grad_norm": 0.12441521883010864,
      "learning_rate": 7.98128310742925e-07,
      "loss": 0.0015,
      "step": 260660
    },
    {
      "epoch": 1.6718564103221918,
      "grad_norm": 0.045425478368997574,
      "learning_rate": 7.978249742407573e-07,
      "loss": 0.0015,
      "step": 260670
    },
    {
      "epoch": 1.671920547215978,
      "grad_norm": 0.019230369478464127,
      "learning_rate": 7.975216903948446e-07,
      "loss": 0.0005,
      "step": 260680
    },
    {
      "epoch": 1.6719846841097639,
      "grad_norm": 0.04719553887844086,
      "learning_rate": 7.972184592089899e-07,
      "loss": 0.0008,
      "step": 260690
    },
    {
      "epoch": 1.67204882100355,
      "grad_norm": 0.04122736304998398,
      "learning_rate": 7.969152806869912e-07,
      "loss": 0.001,
      "step": 260700
    },
    {
      "epoch": 1.672112957897336,
      "grad_norm": 0.07827562838792801,
      "learning_rate": 7.966121548326483e-07,
      "loss": 0.0009,
      "step": 260710
    },
    {
      "epoch": 1.672177094791122,
      "grad_norm": 0.1230246052145958,
      "learning_rate": 7.963090816497576e-07,
      "loss": 0.0015,
      "step": 260720
    },
    {
      "epoch": 1.6722412316849082,
      "grad_norm": 0.5096327066421509,
      "learning_rate": 7.9600606114212e-07,
      "loss": 0.0013,
      "step": 260730
    },
    {
      "epoch": 1.6723053685786944,
      "grad_norm": 0.19383084774017334,
      "learning_rate": 7.957030933135312e-07,
      "loss": 0.0015,
      "step": 260740
    },
    {
      "epoch": 1.6723695054724805,
      "grad_norm": 0.056503500789403915,
      "learning_rate": 7.954001781677872e-07,
      "loss": 0.0018,
      "step": 260750
    },
    {
      "epoch": 1.6724336423662667,
      "grad_norm": 0.16064177453517914,
      "learning_rate": 7.950973157086827e-07,
      "loss": 0.0007,
      "step": 260760
    },
    {
      "epoch": 1.6724977792600526,
      "grad_norm": 0.038757264614105225,
      "learning_rate": 7.947945059400153e-07,
      "loss": 0.0011,
      "step": 260770
    },
    {
      "epoch": 1.6725619161538388,
      "grad_norm": 0.080266073346138,
      "learning_rate": 7.944917488655779e-07,
      "loss": 0.0014,
      "step": 260780
    },
    {
      "epoch": 1.672626053047625,
      "grad_norm": 0.0057196966372430325,
      "learning_rate": 7.941890444891637e-07,
      "loss": 0.0006,
      "step": 260790
    },
    {
      "epoch": 1.6726901899414108,
      "grad_norm": 0.14382421970367432,
      "learning_rate": 7.938863928145674e-07,
      "loss": 0.0012,
      "step": 260800
    },
    {
      "epoch": 1.672754326835197,
      "grad_norm": 0.15724822878837585,
      "learning_rate": 7.935837938455809e-07,
      "loss": 0.0007,
      "step": 260810
    },
    {
      "epoch": 1.6728184637289831,
      "grad_norm": 0.06264682859182358,
      "learning_rate": 7.932812475859958e-07,
      "loss": 0.0006,
      "step": 260820
    },
    {
      "epoch": 1.6728826006227693,
      "grad_norm": 0.1835131198167801,
      "learning_rate": 7.929787540396016e-07,
      "loss": 0.0012,
      "step": 260830
    },
    {
      "epoch": 1.6729467375165554,
      "grad_norm": 0.020518098026514053,
      "learning_rate": 7.926763132101923e-07,
      "loss": 0.0008,
      "step": 260840
    },
    {
      "epoch": 1.6730108744103416,
      "grad_norm": 0.03425362706184387,
      "learning_rate": 7.92373925101555e-07,
      "loss": 0.0012,
      "step": 260850
    },
    {
      "epoch": 1.6730750113041275,
      "grad_norm": 0.010144386440515518,
      "learning_rate": 7.920715897174796e-07,
      "loss": 0.0014,
      "step": 260860
    },
    {
      "epoch": 1.6731391481979137,
      "grad_norm": 0.020300816744565964,
      "learning_rate": 7.917693070617538e-07,
      "loss": 0.001,
      "step": 260870
    },
    {
      "epoch": 1.6732032850916996,
      "grad_norm": 0.14274899661540985,
      "learning_rate": 7.914670771381671e-07,
      "loss": 0.0011,
      "step": 260880
    },
    {
      "epoch": 1.6732674219854857,
      "grad_norm": 0.09293682128190994,
      "learning_rate": 7.911648999505056e-07,
      "loss": 0.0007,
      "step": 260890
    },
    {
      "epoch": 1.6733315588792719,
      "grad_norm": 0.004996034782379866,
      "learning_rate": 7.908627755025561e-07,
      "loss": 0.0012,
      "step": 260900
    },
    {
      "epoch": 1.673395695773058,
      "grad_norm": 0.07686932384967804,
      "learning_rate": 7.905607037981028e-07,
      "loss": 0.001,
      "step": 260910
    },
    {
      "epoch": 1.6734598326668442,
      "grad_norm": 0.10814297199249268,
      "learning_rate": 7.902586848409338e-07,
      "loss": 0.0012,
      "step": 260920
    },
    {
      "epoch": 1.6735239695606303,
      "grad_norm": 0.2382301688194275,
      "learning_rate": 7.899567186348322e-07,
      "loss": 0.0008,
      "step": 260930
    },
    {
      "epoch": 1.6735881064544165,
      "grad_norm": 0.032378844916820526,
      "learning_rate": 7.896548051835801e-07,
      "loss": 0.0013,
      "step": 260940
    },
    {
      "epoch": 1.6736522433482024,
      "grad_norm": 0.0709405466914177,
      "learning_rate": 7.893529444909642e-07,
      "loss": 0.0008,
      "step": 260950
    },
    {
      "epoch": 1.6737163802419885,
      "grad_norm": 0.09986542165279388,
      "learning_rate": 7.890511365607645e-07,
      "loss": 0.0006,
      "step": 260960
    },
    {
      "epoch": 1.6737805171357745,
      "grad_norm": 0.2603265047073364,
      "learning_rate": 7.887493813967639e-07,
      "loss": 0.0011,
      "step": 260970
    },
    {
      "epoch": 1.6738446540295606,
      "grad_norm": 0.03708268702030182,
      "learning_rate": 7.884476790027424e-07,
      "loss": 0.0008,
      "step": 260980
    },
    {
      "epoch": 1.6739087909233468,
      "grad_norm": 0.11559338122606277,
      "learning_rate": 7.881460293824827e-07,
      "loss": 0.0013,
      "step": 260990
    },
    {
      "epoch": 1.673972927817133,
      "grad_norm": 0.039508212357759476,
      "learning_rate": 7.878444325397628e-07,
      "loss": 0.0006,
      "step": 261000
    },
    {
      "epoch": 1.674037064710919,
      "grad_norm": 0.1299441009759903,
      "learning_rate": 7.875428884783631e-07,
      "loss": 0.0012,
      "step": 261010
    },
    {
      "epoch": 1.6741012016047052,
      "grad_norm": 0.07982337474822998,
      "learning_rate": 7.872413972020598e-07,
      "loss": 0.0011,
      "step": 261020
    },
    {
      "epoch": 1.6741653384984911,
      "grad_norm": 0.0564100444316864,
      "learning_rate": 7.869399587146342e-07,
      "loss": 0.0011,
      "step": 261030
    },
    {
      "epoch": 1.6742294753922773,
      "grad_norm": 0.17207171022891998,
      "learning_rate": 7.866385730198623e-07,
      "loss": 0.0006,
      "step": 261040
    },
    {
      "epoch": 1.6742936122860632,
      "grad_norm": 0.033203672617673874,
      "learning_rate": 7.863372401215186e-07,
      "loss": 0.0008,
      "step": 261050
    },
    {
      "epoch": 1.6743577491798494,
      "grad_norm": 0.037467509508132935,
      "learning_rate": 7.860359600233819e-07,
      "loss": 0.0012,
      "step": 261060
    },
    {
      "epoch": 1.6744218860736355,
      "grad_norm": 0.035042110830545425,
      "learning_rate": 7.857347327292269e-07,
      "loss": 0.001,
      "step": 261070
    },
    {
      "epoch": 1.6744860229674217,
      "grad_norm": 0.006249886937439442,
      "learning_rate": 7.854335582428269e-07,
      "loss": 0.0014,
      "step": 261080
    },
    {
      "epoch": 1.6745501598612078,
      "grad_norm": 0.07606332749128342,
      "learning_rate": 7.851324365679557e-07,
      "loss": 0.0017,
      "step": 261090
    },
    {
      "epoch": 1.674614296754994,
      "grad_norm": 0.032138582319021225,
      "learning_rate": 7.848313677083885e-07,
      "loss": 0.0009,
      "step": 261100
    },
    {
      "epoch": 1.67467843364878,
      "grad_norm": 0.06362374871969223,
      "learning_rate": 7.845303516678964e-07,
      "loss": 0.0011,
      "step": 261110
    },
    {
      "epoch": 1.674742570542566,
      "grad_norm": 0.027711695060133934,
      "learning_rate": 7.842293884502523e-07,
      "loss": 0.0012,
      "step": 261120
    },
    {
      "epoch": 1.6748067074363522,
      "grad_norm": 0.034594226628541946,
      "learning_rate": 7.83928478059226e-07,
      "loss": 0.0018,
      "step": 261130
    },
    {
      "epoch": 1.674870844330138,
      "grad_norm": 0.024958940222859383,
      "learning_rate": 7.836276204985894e-07,
      "loss": 0.0009,
      "step": 261140
    },
    {
      "epoch": 1.6749349812239243,
      "grad_norm": 0.11396825313568115,
      "learning_rate": 7.833268157721124e-07,
      "loss": 0.0014,
      "step": 261150
    },
    {
      "epoch": 1.6749991181177104,
      "grad_norm": 0.1585187166929245,
      "learning_rate": 7.830260638835646e-07,
      "loss": 0.001,
      "step": 261160
    },
    {
      "epoch": 1.6750632550114966,
      "grad_norm": 0.027392275631427765,
      "learning_rate": 7.827253648367122e-07,
      "loss": 0.0006,
      "step": 261170
    },
    {
      "epoch": 1.6751273919052827,
      "grad_norm": 0.08222039043903351,
      "learning_rate": 7.824247186353262e-07,
      "loss": 0.001,
      "step": 261180
    },
    {
      "epoch": 1.6751915287990689,
      "grad_norm": 0.06914021819829941,
      "learning_rate": 7.821241252831729e-07,
      "loss": 0.0016,
      "step": 261190
    },
    {
      "epoch": 1.6752556656928548,
      "grad_norm": 0.027489222586154938,
      "learning_rate": 7.818235847840178e-07,
      "loss": 0.001,
      "step": 261200
    },
    {
      "epoch": 1.675319802586641,
      "grad_norm": 0.14536058902740479,
      "learning_rate": 7.815230971416277e-07,
      "loss": 0.0011,
      "step": 261210
    },
    {
      "epoch": 1.675383939480427,
      "grad_norm": 0.05501720681786537,
      "learning_rate": 7.812226623597691e-07,
      "loss": 0.0012,
      "step": 261220
    },
    {
      "epoch": 1.675448076374213,
      "grad_norm": 0.4984779357910156,
      "learning_rate": 7.809222804422062e-07,
      "loss": 0.0067,
      "step": 261230
    },
    {
      "epoch": 1.6755122132679992,
      "grad_norm": 0.14796331524848938,
      "learning_rate": 7.806219513927016e-07,
      "loss": 0.0016,
      "step": 261240
    },
    {
      "epoch": 1.6755763501617853,
      "grad_norm": 0.052086081355810165,
      "learning_rate": 7.803216752150189e-07,
      "loss": 0.0006,
      "step": 261250
    },
    {
      "epoch": 1.6756404870555714,
      "grad_norm": 0.09541890025138855,
      "learning_rate": 7.800214519129223e-07,
      "loss": 0.0012,
      "step": 261260
    },
    {
      "epoch": 1.6757046239493576,
      "grad_norm": 0.11986565589904785,
      "learning_rate": 7.797212814901733e-07,
      "loss": 0.0013,
      "step": 261270
    },
    {
      "epoch": 1.6757687608431437,
      "grad_norm": 0.09449843317270279,
      "learning_rate": 7.794211639505322e-07,
      "loss": 0.0008,
      "step": 261280
    },
    {
      "epoch": 1.6758328977369297,
      "grad_norm": 0.07100570201873779,
      "learning_rate": 7.791210992977599e-07,
      "loss": 0.0014,
      "step": 261290
    },
    {
      "epoch": 1.6758970346307158,
      "grad_norm": 0.04841234162449837,
      "learning_rate": 7.788210875356173e-07,
      "loss": 0.0014,
      "step": 261300
    },
    {
      "epoch": 1.6759611715245017,
      "grad_norm": 0.0634155198931694,
      "learning_rate": 7.78521128667864e-07,
      "loss": 0.0009,
      "step": 261310
    },
    {
      "epoch": 1.676025308418288,
      "grad_norm": 0.04266565665602684,
      "learning_rate": 7.782212226982561e-07,
      "loss": 0.0013,
      "step": 261320
    },
    {
      "epoch": 1.676089445312074,
      "grad_norm": 0.2701398730278015,
      "learning_rate": 7.779213696305554e-07,
      "loss": 0.0013,
      "step": 261330
    },
    {
      "epoch": 1.6761535822058602,
      "grad_norm": 0.09580012410879135,
      "learning_rate": 7.776215694685169e-07,
      "loss": 0.0012,
      "step": 261340
    },
    {
      "epoch": 1.6762177190996463,
      "grad_norm": 0.09963560849428177,
      "learning_rate": 7.773218222158979e-07,
      "loss": 0.0008,
      "step": 261350
    },
    {
      "epoch": 1.6762818559934325,
      "grad_norm": 0.06957130879163742,
      "learning_rate": 7.770221278764534e-07,
      "loss": 0.0015,
      "step": 261360
    },
    {
      "epoch": 1.6763459928872186,
      "grad_norm": 0.08795663714408875,
      "learning_rate": 7.767224864539408e-07,
      "loss": 0.0006,
      "step": 261370
    },
    {
      "epoch": 1.6764101297810046,
      "grad_norm": 0.059120964258909225,
      "learning_rate": 7.764228979521138e-07,
      "loss": 0.0019,
      "step": 261380
    },
    {
      "epoch": 1.6764742666747907,
      "grad_norm": 0.052831437438726425,
      "learning_rate": 7.761233623747261e-07,
      "loss": 0.0009,
      "step": 261390
    },
    {
      "epoch": 1.6765384035685766,
      "grad_norm": 0.024905595928430557,
      "learning_rate": 7.758238797255308e-07,
      "loss": 0.0014,
      "step": 261400
    },
    {
      "epoch": 1.6766025404623628,
      "grad_norm": 0.09078676253557205,
      "learning_rate": 7.755244500082821e-07,
      "loss": 0.0008,
      "step": 261410
    },
    {
      "epoch": 1.676666677356149,
      "grad_norm": 0.055515043437480927,
      "learning_rate": 7.752250732267314e-07,
      "loss": 0.0009,
      "step": 261420
    },
    {
      "epoch": 1.676730814249935,
      "grad_norm": 0.02581734210252762,
      "learning_rate": 7.749257493846301e-07,
      "loss": 0.0005,
      "step": 261430
    },
    {
      "epoch": 1.6767949511437212,
      "grad_norm": 0.2283240556716919,
      "learning_rate": 7.746264784857271e-07,
      "loss": 0.0015,
      "step": 261440
    },
    {
      "epoch": 1.6768590880375074,
      "grad_norm": 0.030655836686491966,
      "learning_rate": 7.743272605337754e-07,
      "loss": 0.0007,
      "step": 261450
    },
    {
      "epoch": 1.6769232249312933,
      "grad_norm": 0.1032186895608902,
      "learning_rate": 7.740280955325236e-07,
      "loss": 0.0014,
      "step": 261460
    },
    {
      "epoch": 1.6769873618250795,
      "grad_norm": 0.05850433185696602,
      "learning_rate": 7.737289834857187e-07,
      "loss": 0.001,
      "step": 261470
    },
    {
      "epoch": 1.6770514987188654,
      "grad_norm": 0.5587889552116394,
      "learning_rate": 7.734299243971111e-07,
      "loss": 0.0005,
      "step": 261480
    },
    {
      "epoch": 1.6771156356126515,
      "grad_norm": 0.21267381310462952,
      "learning_rate": 7.731309182704471e-07,
      "loss": 0.0009,
      "step": 261490
    },
    {
      "epoch": 1.6771797725064377,
      "grad_norm": 0.104644276201725,
      "learning_rate": 7.72831965109474e-07,
      "loss": 0.0017,
      "step": 261500
    },
    {
      "epoch": 1.6772439094002238,
      "grad_norm": 0.05525534972548485,
      "learning_rate": 7.725330649179364e-07,
      "loss": 0.0005,
      "step": 261510
    },
    {
      "epoch": 1.67730804629401,
      "grad_norm": 0.04056679457426071,
      "learning_rate": 7.722342176995817e-07,
      "loss": 0.0013,
      "step": 261520
    },
    {
      "epoch": 1.6773721831877961,
      "grad_norm": 0.016694676131010056,
      "learning_rate": 7.719354234581539e-07,
      "loss": 0.0013,
      "step": 261530
    },
    {
      "epoch": 1.6774363200815823,
      "grad_norm": 0.20411895215511322,
      "learning_rate": 7.716366821973964e-07,
      "loss": 0.0009,
      "step": 261540
    },
    {
      "epoch": 1.6775004569753682,
      "grad_norm": 0.02913813479244709,
      "learning_rate": 7.71337993921053e-07,
      "loss": 0.001,
      "step": 261550
    },
    {
      "epoch": 1.6775645938691544,
      "grad_norm": 0.06505188345909119,
      "learning_rate": 7.710393586328674e-07,
      "loss": 0.0009,
      "step": 261560
    },
    {
      "epoch": 1.6776287307629403,
      "grad_norm": 0.07389794290065765,
      "learning_rate": 7.707407763365809e-07,
      "loss": 0.0016,
      "step": 261570
    },
    {
      "epoch": 1.6776928676567264,
      "grad_norm": 0.13428865373134613,
      "learning_rate": 7.704422470359352e-07,
      "loss": 0.0008,
      "step": 261580
    },
    {
      "epoch": 1.6777570045505126,
      "grad_norm": 0.13568255305290222,
      "learning_rate": 7.701437707346698e-07,
      "loss": 0.0011,
      "step": 261590
    },
    {
      "epoch": 1.6778211414442987,
      "grad_norm": 0.043442100286483765,
      "learning_rate": 7.698453474365275e-07,
      "loss": 0.0006,
      "step": 261600
    },
    {
      "epoch": 1.6778852783380849,
      "grad_norm": 0.07020683586597443,
      "learning_rate": 7.695469771452458e-07,
      "loss": 0.0008,
      "step": 261610
    },
    {
      "epoch": 1.677949415231871,
      "grad_norm": 0.1260029524564743,
      "learning_rate": 7.692486598645632e-07,
      "loss": 0.0008,
      "step": 261620
    },
    {
      "epoch": 1.678013552125657,
      "grad_norm": 0.11657378077507019,
      "learning_rate": 7.689503955982192e-07,
      "loss": 0.0009,
      "step": 261630
    },
    {
      "epoch": 1.678077689019443,
      "grad_norm": 0.18355035781860352,
      "learning_rate": 7.686521843499512e-07,
      "loss": 0.0019,
      "step": 261640
    },
    {
      "epoch": 1.6781418259132292,
      "grad_norm": 0.0605897530913353,
      "learning_rate": 7.683540261234951e-07,
      "loss": 0.0012,
      "step": 261650
    },
    {
      "epoch": 1.6782059628070152,
      "grad_norm": 0.18452134728431702,
      "learning_rate": 7.680559209225863e-07,
      "loss": 0.0027,
      "step": 261660
    },
    {
      "epoch": 1.6782700997008013,
      "grad_norm": 0.027823036536574364,
      "learning_rate": 7.677578687509624e-07,
      "loss": 0.0008,
      "step": 261670
    },
    {
      "epoch": 1.6783342365945875,
      "grad_norm": 0.055968355387449265,
      "learning_rate": 7.674598696123569e-07,
      "loss": 0.0016,
      "step": 261680
    },
    {
      "epoch": 1.6783983734883736,
      "grad_norm": 0.08254148066043854,
      "learning_rate": 7.671619235105043e-07,
      "loss": 0.001,
      "step": 261690
    },
    {
      "epoch": 1.6784625103821598,
      "grad_norm": 0.0800231546163559,
      "learning_rate": 7.668640304491371e-07,
      "loss": 0.0017,
      "step": 261700
    },
    {
      "epoch": 1.678526647275946,
      "grad_norm": 0.1368565559387207,
      "learning_rate": 7.665661904319898e-07,
      "loss": 0.0012,
      "step": 261710
    },
    {
      "epoch": 1.6785907841697318,
      "grad_norm": 0.0406535267829895,
      "learning_rate": 7.662684034627943e-07,
      "loss": 0.0021,
      "step": 261720
    },
    {
      "epoch": 1.678654921063518,
      "grad_norm": 0.07403713464736938,
      "learning_rate": 7.659706695452807e-07,
      "loss": 0.0011,
      "step": 261730
    },
    {
      "epoch": 1.678719057957304,
      "grad_norm": 0.09942308813333511,
      "learning_rate": 7.656729886831798e-07,
      "loss": 0.0008,
      "step": 261740
    },
    {
      "epoch": 1.67878319485109,
      "grad_norm": 0.06193510442972183,
      "learning_rate": 7.653753608802239e-07,
      "loss": 0.0014,
      "step": 261750
    },
    {
      "epoch": 1.6788473317448762,
      "grad_norm": 0.11304626613855362,
      "learning_rate": 7.650777861401415e-07,
      "loss": 0.0011,
      "step": 261760
    },
    {
      "epoch": 1.6789114686386624,
      "grad_norm": 0.2045489251613617,
      "learning_rate": 7.647802644666591e-07,
      "loss": 0.0013,
      "step": 261770
    },
    {
      "epoch": 1.6789756055324485,
      "grad_norm": 0.1679089516401291,
      "learning_rate": 7.644827958635087e-07,
      "loss": 0.0006,
      "step": 261780
    },
    {
      "epoch": 1.6790397424262347,
      "grad_norm": 0.012728678993880749,
      "learning_rate": 7.641853803344157e-07,
      "loss": 0.0024,
      "step": 261790
    },
    {
      "epoch": 1.6791038793200208,
      "grad_norm": 0.15526632964611053,
      "learning_rate": 7.638880178831071e-07,
      "loss": 0.0014,
      "step": 261800
    },
    {
      "epoch": 1.6791680162138067,
      "grad_norm": 0.047305576503276825,
      "learning_rate": 7.635907085133082e-07,
      "loss": 0.0016,
      "step": 261810
    },
    {
      "epoch": 1.6792321531075929,
      "grad_norm": 0.09887392073869705,
      "learning_rate": 7.63293452228746e-07,
      "loss": 0.0009,
      "step": 261820
    },
    {
      "epoch": 1.6792962900013788,
      "grad_norm": 0.09882711619138718,
      "learning_rate": 7.629962490331456e-07,
      "loss": 0.0018,
      "step": 261830
    },
    {
      "epoch": 1.679360426895165,
      "grad_norm": 0.015573171898722649,
      "learning_rate": 7.626990989302296e-07,
      "loss": 0.0006,
      "step": 261840
    },
    {
      "epoch": 1.679424563788951,
      "grad_norm": 0.07946422696113586,
      "learning_rate": 7.624020019237216e-07,
      "loss": 0.0013,
      "step": 261850
    },
    {
      "epoch": 1.6794887006827373,
      "grad_norm": 0.013404951430857182,
      "learning_rate": 7.621049580173461e-07,
      "loss": 0.0006,
      "step": 261860
    },
    {
      "epoch": 1.6795528375765234,
      "grad_norm": 0.03741779923439026,
      "learning_rate": 7.618079672148243e-07,
      "loss": 0.0015,
      "step": 261870
    },
    {
      "epoch": 1.6796169744703096,
      "grad_norm": 0.0537244938313961,
      "learning_rate": 7.615110295198774e-07,
      "loss": 0.001,
      "step": 261880
    },
    {
      "epoch": 1.6796811113640955,
      "grad_norm": 0.07449512183666229,
      "learning_rate": 7.612141449362254e-07,
      "loss": 0.0011,
      "step": 261890
    },
    {
      "epoch": 1.6797452482578816,
      "grad_norm": 0.05764267221093178,
      "learning_rate": 7.609173134675912e-07,
      "loss": 0.0011,
      "step": 261900
    },
    {
      "epoch": 1.6798093851516676,
      "grad_norm": 0.3346785604953766,
      "learning_rate": 7.606205351176921e-07,
      "loss": 0.0029,
      "step": 261910
    },
    {
      "epoch": 1.6798735220454537,
      "grad_norm": 0.04754569008946419,
      "learning_rate": 7.603238098902466e-07,
      "loss": 0.0008,
      "step": 261920
    },
    {
      "epoch": 1.6799376589392399,
      "grad_norm": 0.03843769431114197,
      "learning_rate": 7.600271377889751e-07,
      "loss": 0.0006,
      "step": 261930
    },
    {
      "epoch": 1.680001795833026,
      "grad_norm": 0.040142547339200974,
      "learning_rate": 7.597305188175941e-07,
      "loss": 0.0018,
      "step": 261940
    },
    {
      "epoch": 1.6800659327268122,
      "grad_norm": 0.016071323305368423,
      "learning_rate": 7.594339529798195e-07,
      "loss": 0.0009,
      "step": 261950
    },
    {
      "epoch": 1.6801300696205983,
      "grad_norm": 0.014691210351884365,
      "learning_rate": 7.591374402793672e-07,
      "loss": 0.0007,
      "step": 261960
    },
    {
      "epoch": 1.6801942065143844,
      "grad_norm": 0.11548521369695663,
      "learning_rate": 7.588409807199537e-07,
      "loss": 0.0011,
      "step": 261970
    },
    {
      "epoch": 1.6802583434081704,
      "grad_norm": 0.033998552709817886,
      "learning_rate": 7.585445743052955e-07,
      "loss": 0.0005,
      "step": 261980
    },
    {
      "epoch": 1.6803224803019565,
      "grad_norm": 0.050724223256111145,
      "learning_rate": 7.582482210391046e-07,
      "loss": 0.001,
      "step": 261990
    },
    {
      "epoch": 1.6803866171957424,
      "grad_norm": 0.1336205154657364,
      "learning_rate": 7.579519209250946e-07,
      "loss": 0.0007,
      "step": 262000
    },
    {
      "epoch": 1.6804507540895286,
      "grad_norm": 0.004237947519868612,
      "learning_rate": 7.576556739669782e-07,
      "loss": 0.0011,
      "step": 262010
    },
    {
      "epoch": 1.6805148909833147,
      "grad_norm": 0.10108738392591476,
      "learning_rate": 7.573594801684692e-07,
      "loss": 0.001,
      "step": 262020
    },
    {
      "epoch": 1.680579027877101,
      "grad_norm": 0.06919895857572556,
      "learning_rate": 7.570633395332783e-07,
      "loss": 0.0054,
      "step": 262030
    },
    {
      "epoch": 1.680643164770887,
      "grad_norm": 0.07569548487663269,
      "learning_rate": 7.567672520651148e-07,
      "loss": 0.0011,
      "step": 262040
    },
    {
      "epoch": 1.6807073016646732,
      "grad_norm": 0.020496781915426254,
      "learning_rate": 7.564712177676914e-07,
      "loss": 0.0009,
      "step": 262050
    },
    {
      "epoch": 1.6807714385584593,
      "grad_norm": 0.12332148849964142,
      "learning_rate": 7.561752366447162e-07,
      "loss": 0.0018,
      "step": 262060
    },
    {
      "epoch": 1.6808355754522453,
      "grad_norm": 0.0759996846318245,
      "learning_rate": 7.558793086998989e-07,
      "loss": 0.0012,
      "step": 262070
    },
    {
      "epoch": 1.6808997123460314,
      "grad_norm": 0.004803712945431471,
      "learning_rate": 7.555834339369456e-07,
      "loss": 0.0009,
      "step": 262080
    },
    {
      "epoch": 1.6809638492398173,
      "grad_norm": 0.04900961369276047,
      "learning_rate": 7.55287612359566e-07,
      "loss": 0.0007,
      "step": 262090
    },
    {
      "epoch": 1.6810279861336035,
      "grad_norm": 0.06744549423456192,
      "learning_rate": 7.549918439714671e-07,
      "loss": 0.0028,
      "step": 262100
    },
    {
      "epoch": 1.6810921230273896,
      "grad_norm": 0.09023269265890121,
      "learning_rate": 7.546961287763538e-07,
      "loss": 0.0019,
      "step": 262110
    },
    {
      "epoch": 1.6811562599211758,
      "grad_norm": 0.09666185826063156,
      "learning_rate": 7.544004667779309e-07,
      "loss": 0.0008,
      "step": 262120
    },
    {
      "epoch": 1.681220396814962,
      "grad_norm": 0.045062899589538574,
      "learning_rate": 7.541048579799054e-07,
      "loss": 0.0014,
      "step": 262130
    },
    {
      "epoch": 1.681284533708748,
      "grad_norm": 0.13950562477111816,
      "learning_rate": 7.53809302385981e-07,
      "loss": 0.0011,
      "step": 262140
    },
    {
      "epoch": 1.681348670602534,
      "grad_norm": 0.015909569337964058,
      "learning_rate": 7.535137999998604e-07,
      "loss": 0.0007,
      "step": 262150
    },
    {
      "epoch": 1.6814128074963202,
      "grad_norm": 0.0681985542178154,
      "learning_rate": 7.532183508252455e-07,
      "loss": 0.0007,
      "step": 262160
    },
    {
      "epoch": 1.681476944390106,
      "grad_norm": 0.08667725324630737,
      "learning_rate": 7.529229548658412e-07,
      "loss": 0.0006,
      "step": 262170
    },
    {
      "epoch": 1.6815410812838922,
      "grad_norm": 0.05679089203476906,
      "learning_rate": 7.526276121253479e-07,
      "loss": 0.0014,
      "step": 262180
    },
    {
      "epoch": 1.6816052181776784,
      "grad_norm": 0.14505784213542938,
      "learning_rate": 7.523323226074647e-07,
      "loss": 0.0011,
      "step": 262190
    },
    {
      "epoch": 1.6816693550714645,
      "grad_norm": 0.08209065347909927,
      "learning_rate": 7.520370863158943e-07,
      "loss": 0.0013,
      "step": 262200
    },
    {
      "epoch": 1.6817334919652507,
      "grad_norm": 0.15775032341480255,
      "learning_rate": 7.517419032543355e-07,
      "loss": 0.0018,
      "step": 262210
    },
    {
      "epoch": 1.6817976288590368,
      "grad_norm": 0.1044556275010109,
      "learning_rate": 7.514467734264869e-07,
      "loss": 0.0012,
      "step": 262220
    },
    {
      "epoch": 1.681861765752823,
      "grad_norm": 0.06325215846300125,
      "learning_rate": 7.511516968360449e-07,
      "loss": 0.0012,
      "step": 262230
    },
    {
      "epoch": 1.681925902646609,
      "grad_norm": 0.2255081981420517,
      "learning_rate": 7.508566734867107e-07,
      "loss": 0.0015,
      "step": 262240
    },
    {
      "epoch": 1.681990039540395,
      "grad_norm": 0.04119641333818436,
      "learning_rate": 7.505617033821783e-07,
      "loss": 0.0004,
      "step": 262250
    },
    {
      "epoch": 1.682054176434181,
      "grad_norm": 0.10483510047197342,
      "learning_rate": 7.502667865261454e-07,
      "loss": 0.0011,
      "step": 262260
    },
    {
      "epoch": 1.6821183133279671,
      "grad_norm": 0.1358007788658142,
      "learning_rate": 7.499719229223057e-07,
      "loss": 0.0015,
      "step": 262270
    },
    {
      "epoch": 1.6821824502217533,
      "grad_norm": 0.02440463751554489,
      "learning_rate": 7.496771125743563e-07,
      "loss": 0.0008,
      "step": 262280
    },
    {
      "epoch": 1.6822465871155394,
      "grad_norm": 0.08051423728466034,
      "learning_rate": 7.493823554859903e-07,
      "loss": 0.0016,
      "step": 262290
    },
    {
      "epoch": 1.6823107240093256,
      "grad_norm": 0.06842823326587677,
      "learning_rate": 7.490876516609013e-07,
      "loss": 0.0014,
      "step": 262300
    },
    {
      "epoch": 1.6823748609031117,
      "grad_norm": 0.04999329894781113,
      "learning_rate": 7.48793001102781e-07,
      "loss": 0.0006,
      "step": 262310
    },
    {
      "epoch": 1.6824389977968977,
      "grad_norm": 0.05793393775820732,
      "learning_rate": 7.484984038153237e-07,
      "loss": 0.0012,
      "step": 262320
    },
    {
      "epoch": 1.6825031346906838,
      "grad_norm": 0.14233599603176117,
      "learning_rate": 7.482038598022202e-07,
      "loss": 0.0005,
      "step": 262330
    },
    {
      "epoch": 1.68256727158447,
      "grad_norm": 0.07395566999912262,
      "learning_rate": 7.479093690671596e-07,
      "loss": 0.0007,
      "step": 262340
    },
    {
      "epoch": 1.6826314084782559,
      "grad_norm": 0.12719450891017914,
      "learning_rate": 7.476149316138348e-07,
      "loss": 0.0007,
      "step": 262350
    },
    {
      "epoch": 1.682695545372042,
      "grad_norm": 0.13135480880737305,
      "learning_rate": 7.473205474459339e-07,
      "loss": 0.0013,
      "step": 262360
    },
    {
      "epoch": 1.6827596822658282,
      "grad_norm": 0.09198420494794846,
      "learning_rate": 7.470262165671461e-07,
      "loss": 0.0015,
      "step": 262370
    },
    {
      "epoch": 1.6828238191596143,
      "grad_norm": 0.13832350075244904,
      "learning_rate": 7.467319389811578e-07,
      "loss": 0.0007,
      "step": 262380
    },
    {
      "epoch": 1.6828879560534005,
      "grad_norm": 0.08932088315486908,
      "learning_rate": 7.464377146916595e-07,
      "loss": 0.0012,
      "step": 262390
    },
    {
      "epoch": 1.6829520929471866,
      "grad_norm": 0.0009030341752804816,
      "learning_rate": 7.461435437023368e-07,
      "loss": 0.0017,
      "step": 262400
    },
    {
      "epoch": 1.6830162298409725,
      "grad_norm": 0.09307923167943954,
      "learning_rate": 7.458494260168753e-07,
      "loss": 0.0007,
      "step": 262410
    },
    {
      "epoch": 1.6830803667347587,
      "grad_norm": 0.03176266327500343,
      "learning_rate": 7.455553616389599e-07,
      "loss": 0.0015,
      "step": 262420
    },
    {
      "epoch": 1.6831445036285446,
      "grad_norm": 0.021300937980413437,
      "learning_rate": 7.452613505722778e-07,
      "loss": 0.0011,
      "step": 262430
    },
    {
      "epoch": 1.6832086405223308,
      "grad_norm": 0.01835530996322632,
      "learning_rate": 7.449673928205114e-07,
      "loss": 0.0012,
      "step": 262440
    },
    {
      "epoch": 1.683272777416117,
      "grad_norm": 0.03712568059563637,
      "learning_rate": 7.446734883873452e-07,
      "loss": 0.0009,
      "step": 262450
    },
    {
      "epoch": 1.683336914309903,
      "grad_norm": 0.08349623531103134,
      "learning_rate": 7.4437963727646e-07,
      "loss": 0.0012,
      "step": 262460
    },
    {
      "epoch": 1.6834010512036892,
      "grad_norm": 0.15113046765327454,
      "learning_rate": 7.440858394915401e-07,
      "loss": 0.0011,
      "step": 262470
    },
    {
      "epoch": 1.6834651880974754,
      "grad_norm": 0.2494426816701889,
      "learning_rate": 7.437920950362665e-07,
      "loss": 0.001,
      "step": 262480
    },
    {
      "epoch": 1.6835293249912615,
      "grad_norm": 0.07854034751653671,
      "learning_rate": 7.434984039143189e-07,
      "loss": 0.0011,
      "step": 262490
    },
    {
      "epoch": 1.6835934618850474,
      "grad_norm": 0.03320141136646271,
      "learning_rate": 7.432047661293795e-07,
      "loss": 0.0007,
      "step": 262500
    },
    {
      "epoch": 1.6836575987788336,
      "grad_norm": 0.09847461432218552,
      "learning_rate": 7.429111816851264e-07,
      "loss": 0.0005,
      "step": 262510
    },
    {
      "epoch": 1.6837217356726195,
      "grad_norm": 0.02717522718012333,
      "learning_rate": 7.42617650585239e-07,
      "loss": 0.0011,
      "step": 262520
    },
    {
      "epoch": 1.6837858725664057,
      "grad_norm": 0.2049603909254074,
      "learning_rate": 7.423241728333941e-07,
      "loss": 0.0021,
      "step": 262530
    },
    {
      "epoch": 1.6838500094601918,
      "grad_norm": 0.011489232070744038,
      "learning_rate": 7.420307484332711e-07,
      "loss": 0.0007,
      "step": 262540
    },
    {
      "epoch": 1.683914146353978,
      "grad_norm": 0.10288581997156143,
      "learning_rate": 7.417373773885461e-07,
      "loss": 0.0022,
      "step": 262550
    },
    {
      "epoch": 1.683978283247764,
      "grad_norm": 0.09101805835962296,
      "learning_rate": 7.414440597028949e-07,
      "loss": 0.0017,
      "step": 262560
    },
    {
      "epoch": 1.6840424201415503,
      "grad_norm": 0.01015832182019949,
      "learning_rate": 7.411507953799918e-07,
      "loss": 0.0009,
      "step": 262570
    },
    {
      "epoch": 1.6841065570353362,
      "grad_norm": 0.10437482595443726,
      "learning_rate": 7.408575844235144e-07,
      "loss": 0.0018,
      "step": 262580
    },
    {
      "epoch": 1.6841706939291223,
      "grad_norm": 0.03975355252623558,
      "learning_rate": 7.405644268371359e-07,
      "loss": 0.0028,
      "step": 262590
    },
    {
      "epoch": 1.6842348308229083,
      "grad_norm": 0.12639813125133514,
      "learning_rate": 7.402713226245284e-07,
      "loss": 0.002,
      "step": 262600
    },
    {
      "epoch": 1.6842989677166944,
      "grad_norm": 0.19275104999542236,
      "learning_rate": 7.39978271789365e-07,
      "loss": 0.0017,
      "step": 262610
    },
    {
      "epoch": 1.6843631046104806,
      "grad_norm": 0.018935957923531532,
      "learning_rate": 7.396852743353194e-07,
      "loss": 0.0011,
      "step": 262620
    },
    {
      "epoch": 1.6844272415042667,
      "grad_norm": 0.015886522829532623,
      "learning_rate": 7.393923302660627e-07,
      "loss": 0.0006,
      "step": 262630
    },
    {
      "epoch": 1.6844913783980529,
      "grad_norm": 0.04877450317144394,
      "learning_rate": 7.390994395852636e-07,
      "loss": 0.0025,
      "step": 262640
    },
    {
      "epoch": 1.684555515291839,
      "grad_norm": 0.01553063653409481,
      "learning_rate": 7.388066022965946e-07,
      "loss": 0.0009,
      "step": 262650
    },
    {
      "epoch": 1.6846196521856251,
      "grad_norm": 0.02764376997947693,
      "learning_rate": 7.385138184037244e-07,
      "loss": 0.0011,
      "step": 262660
    },
    {
      "epoch": 1.684683789079411,
      "grad_norm": 0.03206116333603859,
      "learning_rate": 7.382210879103219e-07,
      "loss": 0.0007,
      "step": 262670
    },
    {
      "epoch": 1.6847479259731972,
      "grad_norm": 0.10185568779706955,
      "learning_rate": 7.379284108200535e-07,
      "loss": 0.0025,
      "step": 262680
    },
    {
      "epoch": 1.6848120628669832,
      "grad_norm": 0.05674508213996887,
      "learning_rate": 7.376357871365897e-07,
      "loss": 0.0007,
      "step": 262690
    },
    {
      "epoch": 1.6848761997607693,
      "grad_norm": 0.07970869541168213,
      "learning_rate": 7.373432168635958e-07,
      "loss": 0.0011,
      "step": 262700
    },
    {
      "epoch": 1.6849403366545554,
      "grad_norm": 0.02939658612012863,
      "learning_rate": 7.370507000047378e-07,
      "loss": 0.0007,
      "step": 262710
    },
    {
      "epoch": 1.6850044735483416,
      "grad_norm": 0.03765963017940521,
      "learning_rate": 7.3675823656368e-07,
      "loss": 0.0015,
      "step": 262720
    },
    {
      "epoch": 1.6850686104421277,
      "grad_norm": 0.007897039875388145,
      "learning_rate": 7.364658265440894e-07,
      "loss": 0.0009,
      "step": 262730
    },
    {
      "epoch": 1.685132747335914,
      "grad_norm": 0.09889094531536102,
      "learning_rate": 7.361734699496282e-07,
      "loss": 0.0011,
      "step": 262740
    },
    {
      "epoch": 1.6851968842296998,
      "grad_norm": 0.09377726912498474,
      "learning_rate": 7.358811667839616e-07,
      "loss": 0.0009,
      "step": 262750
    },
    {
      "epoch": 1.685261021123486,
      "grad_norm": 0.17835475504398346,
      "learning_rate": 7.355889170507502e-07,
      "loss": 0.001,
      "step": 262760
    },
    {
      "epoch": 1.6853251580172721,
      "grad_norm": 0.07152102142572403,
      "learning_rate": 7.352967207536587e-07,
      "loss": 0.0009,
      "step": 262770
    },
    {
      "epoch": 1.685389294911058,
      "grad_norm": 0.013958334922790527,
      "learning_rate": 7.350045778963477e-07,
      "loss": 0.0027,
      "step": 262780
    },
    {
      "epoch": 1.6854534318048442,
      "grad_norm": 0.09204661846160889,
      "learning_rate": 7.347124884824769e-07,
      "loss": 0.0012,
      "step": 262790
    },
    {
      "epoch": 1.6855175686986303,
      "grad_norm": 0.1123444065451622,
      "learning_rate": 7.344204525157061e-07,
      "loss": 0.001,
      "step": 262800
    },
    {
      "epoch": 1.6855817055924165,
      "grad_norm": 0.033399078994989395,
      "learning_rate": 7.34128469999697e-07,
      "loss": 0.0013,
      "step": 262810
    },
    {
      "epoch": 1.6856458424862026,
      "grad_norm": 0.09275560826063156,
      "learning_rate": 7.338365409381065e-07,
      "loss": 0.0009,
      "step": 262820
    },
    {
      "epoch": 1.6857099793799888,
      "grad_norm": 0.12270651757717133,
      "learning_rate": 7.335446653345934e-07,
      "loss": 0.0012,
      "step": 262830
    },
    {
      "epoch": 1.6857741162737747,
      "grad_norm": 0.10209984332323074,
      "learning_rate": 7.332528431928133e-07,
      "loss": 0.0011,
      "step": 262840
    },
    {
      "epoch": 1.6858382531675609,
      "grad_norm": 0.04941209778189659,
      "learning_rate": 7.329610745164262e-07,
      "loss": 0.001,
      "step": 262850
    },
    {
      "epoch": 1.6859023900613468,
      "grad_norm": 0.36141422390937805,
      "learning_rate": 7.326693593090861e-07,
      "loss": 0.0008,
      "step": 262860
    },
    {
      "epoch": 1.685966526955133,
      "grad_norm": 0.031035954132676125,
      "learning_rate": 7.323776975744484e-07,
      "loss": 0.0012,
      "step": 262870
    },
    {
      "epoch": 1.686030663848919,
      "grad_norm": 0.029855938628315926,
      "learning_rate": 7.320860893161674e-07,
      "loss": 0.0016,
      "step": 262880
    },
    {
      "epoch": 1.6860948007427052,
      "grad_norm": 0.03218485787510872,
      "learning_rate": 7.317945345378991e-07,
      "loss": 0.0016,
      "step": 262890
    },
    {
      "epoch": 1.6861589376364914,
      "grad_norm": 0.07458245754241943,
      "learning_rate": 7.315030332432959e-07,
      "loss": 0.0016,
      "step": 262900
    },
    {
      "epoch": 1.6862230745302775,
      "grad_norm": 0.08994331955909729,
      "learning_rate": 7.312115854360091e-07,
      "loss": 0.0038,
      "step": 262910
    },
    {
      "epoch": 1.6862872114240637,
      "grad_norm": 0.1847568303346634,
      "learning_rate": 7.309201911196928e-07,
      "loss": 0.0008,
      "step": 262920
    },
    {
      "epoch": 1.6863513483178496,
      "grad_norm": 0.0190438125282526,
      "learning_rate": 7.30628850297998e-07,
      "loss": 0.0009,
      "step": 262930
    },
    {
      "epoch": 1.6864154852116358,
      "grad_norm": 0.06274163722991943,
      "learning_rate": 7.303375629745746e-07,
      "loss": 0.0009,
      "step": 262940
    },
    {
      "epoch": 1.6864796221054217,
      "grad_norm": 0.07734082639217377,
      "learning_rate": 7.300463291530718e-07,
      "loss": 0.0008,
      "step": 262950
    },
    {
      "epoch": 1.6865437589992078,
      "grad_norm": 0.019327746704220772,
      "learning_rate": 7.297551488371418e-07,
      "loss": 0.0032,
      "step": 262960
    },
    {
      "epoch": 1.686607895892994,
      "grad_norm": 0.007246449589729309,
      "learning_rate": 7.294640220304317e-07,
      "loss": 0.0006,
      "step": 262970
    },
    {
      "epoch": 1.6866720327867801,
      "grad_norm": 0.006289687007665634,
      "learning_rate": 7.291729487365889e-07,
      "loss": 0.0009,
      "step": 262980
    },
    {
      "epoch": 1.6867361696805663,
      "grad_norm": 0.1553562432527542,
      "learning_rate": 7.288819289592608e-07,
      "loss": 0.0026,
      "step": 262990
    },
    {
      "epoch": 1.6868003065743524,
      "grad_norm": 0.2426321655511856,
      "learning_rate": 7.285909627020954e-07,
      "loss": 0.0018,
      "step": 263000
    },
    {
      "epoch": 1.6868644434681384,
      "grad_norm": 0.19575554132461548,
      "learning_rate": 7.283000499687382e-07,
      "loss": 0.0006,
      "step": 263010
    },
    {
      "epoch": 1.6869285803619245,
      "grad_norm": 0.1894301176071167,
      "learning_rate": 7.280091907628334e-07,
      "loss": 0.001,
      "step": 263020
    },
    {
      "epoch": 1.6869927172557104,
      "grad_norm": 0.17330430448055267,
      "learning_rate": 7.277183850880271e-07,
      "loss": 0.0014,
      "step": 263030
    },
    {
      "epoch": 1.6870568541494966,
      "grad_norm": 0.10145927220582962,
      "learning_rate": 7.274276329479635e-07,
      "loss": 0.0011,
      "step": 263040
    },
    {
      "epoch": 1.6871209910432827,
      "grad_norm": 0.07053583860397339,
      "learning_rate": 7.271369343462848e-07,
      "loss": 0.0008,
      "step": 263050
    },
    {
      "epoch": 1.6871851279370689,
      "grad_norm": 0.12517698109149933,
      "learning_rate": 7.268462892866329e-07,
      "loss": 0.0009,
      "step": 263060
    },
    {
      "epoch": 1.687249264830855,
      "grad_norm": 0.04059239849448204,
      "learning_rate": 7.265556977726523e-07,
      "loss": 0.0007,
      "step": 263070
    },
    {
      "epoch": 1.6873134017246412,
      "grad_norm": 0.0846027135848999,
      "learning_rate": 7.262651598079829e-07,
      "loss": 0.002,
      "step": 263080
    },
    {
      "epoch": 1.6873775386184273,
      "grad_norm": 0.07531607151031494,
      "learning_rate": 7.259746753962655e-07,
      "loss": 0.0013,
      "step": 263090
    },
    {
      "epoch": 1.6874416755122132,
      "grad_norm": 0.051252126693725586,
      "learning_rate": 7.256842445411389e-07,
      "loss": 0.0006,
      "step": 263100
    },
    {
      "epoch": 1.6875058124059994,
      "grad_norm": 0.05703764408826828,
      "learning_rate": 7.253938672462446e-07,
      "loss": 0.0013,
      "step": 263110
    },
    {
      "epoch": 1.6875699492997853,
      "grad_norm": 0.03158862888813019,
      "learning_rate": 7.251035435152198e-07,
      "loss": 0.001,
      "step": 263120
    },
    {
      "epoch": 1.6876340861935715,
      "grad_norm": 0.02299869991838932,
      "learning_rate": 7.248132733517032e-07,
      "loss": 0.0006,
      "step": 263130
    },
    {
      "epoch": 1.6876982230873576,
      "grad_norm": 0.17391365766525269,
      "learning_rate": 7.245230567593308e-07,
      "loss": 0.0011,
      "step": 263140
    },
    {
      "epoch": 1.6877623599811438,
      "grad_norm": 0.013422899879515171,
      "learning_rate": 7.242328937417409e-07,
      "loss": 0.0008,
      "step": 263150
    },
    {
      "epoch": 1.68782649687493,
      "grad_norm": 0.10892119258642197,
      "learning_rate": 7.239427843025687e-07,
      "loss": 0.0014,
      "step": 263160
    },
    {
      "epoch": 1.687890633768716,
      "grad_norm": 0.1053953766822815,
      "learning_rate": 7.236527284454487e-07,
      "loss": 0.0006,
      "step": 263170
    },
    {
      "epoch": 1.687954770662502,
      "grad_norm": 0.07258967310190201,
      "learning_rate": 7.233627261740167e-07,
      "loss": 0.0025,
      "step": 263180
    },
    {
      "epoch": 1.6880189075562881,
      "grad_norm": 0.00420130742713809,
      "learning_rate": 7.230727774919066e-07,
      "loss": 0.0015,
      "step": 263190
    },
    {
      "epoch": 1.6880830444500743,
      "grad_norm": 0.01840096153318882,
      "learning_rate": 7.227828824027516e-07,
      "loss": 0.0017,
      "step": 263200
    },
    {
      "epoch": 1.6881471813438602,
      "grad_norm": 0.12298330664634705,
      "learning_rate": 7.224930409101827e-07,
      "loss": 0.0007,
      "step": 263210
    },
    {
      "epoch": 1.6882113182376464,
      "grad_norm": 0.014765714295208454,
      "learning_rate": 7.222032530178342e-07,
      "loss": 0.0014,
      "step": 263220
    },
    {
      "epoch": 1.6882754551314325,
      "grad_norm": 0.04172176122665405,
      "learning_rate": 7.219135187293357e-07,
      "loss": 0.0015,
      "step": 263230
    },
    {
      "epoch": 1.6883395920252187,
      "grad_norm": 0.015325518324971199,
      "learning_rate": 7.216238380483188e-07,
      "loss": 0.0027,
      "step": 263240
    },
    {
      "epoch": 1.6884037289190048,
      "grad_norm": 0.07766814529895782,
      "learning_rate": 7.213342109784117e-07,
      "loss": 0.0007,
      "step": 263250
    },
    {
      "epoch": 1.688467865812791,
      "grad_norm": 0.11536253988742828,
      "learning_rate": 7.21044637523246e-07,
      "loss": 0.0013,
      "step": 263260
    },
    {
      "epoch": 1.6885320027065769,
      "grad_norm": 0.12471003085374832,
      "learning_rate": 7.207551176864491e-07,
      "loss": 0.0014,
      "step": 263270
    },
    {
      "epoch": 1.688596139600363,
      "grad_norm": 0.025006921961903572,
      "learning_rate": 7.204656514716491e-07,
      "loss": 0.0007,
      "step": 263280
    },
    {
      "epoch": 1.688660276494149,
      "grad_norm": 0.22946634888648987,
      "learning_rate": 7.201762388824717e-07,
      "loss": 0.001,
      "step": 263290
    },
    {
      "epoch": 1.688724413387935,
      "grad_norm": 0.06974420696496964,
      "learning_rate": 7.198868799225455e-07,
      "loss": 0.0005,
      "step": 263300
    },
    {
      "epoch": 1.6887885502817213,
      "grad_norm": 0.007004431914538145,
      "learning_rate": 7.195975745954964e-07,
      "loss": 0.0014,
      "step": 263310
    },
    {
      "epoch": 1.6888526871755074,
      "grad_norm": 0.05225522071123123,
      "learning_rate": 7.193083229049475e-07,
      "loss": 0.004,
      "step": 263320
    },
    {
      "epoch": 1.6889168240692936,
      "grad_norm": 0.007790552452206612,
      "learning_rate": 7.190191248545253e-07,
      "loss": 0.001,
      "step": 263330
    },
    {
      "epoch": 1.6889809609630797,
      "grad_norm": 0.001854132628068328,
      "learning_rate": 7.187299804478536e-07,
      "loss": 0.0008,
      "step": 263340
    },
    {
      "epoch": 1.6890450978568659,
      "grad_norm": 0.06629274040460587,
      "learning_rate": 7.184408896885547e-07,
      "loss": 0.0009,
      "step": 263350
    },
    {
      "epoch": 1.6891092347506518,
      "grad_norm": 0.0625310093164444,
      "learning_rate": 7.181518525802505e-07,
      "loss": 0.001,
      "step": 263360
    },
    {
      "epoch": 1.689173371644438,
      "grad_norm": 0.04694930836558342,
      "learning_rate": 7.178628691265649e-07,
      "loss": 0.0008,
      "step": 263370
    },
    {
      "epoch": 1.6892375085382239,
      "grad_norm": 0.00229465588927269,
      "learning_rate": 7.175739393311176e-07,
      "loss": 0.0008,
      "step": 263380
    },
    {
      "epoch": 1.68930164543201,
      "grad_norm": 0.11056386679410934,
      "learning_rate": 7.172850631975298e-07,
      "loss": 0.0007,
      "step": 263390
    },
    {
      "epoch": 1.6893657823257962,
      "grad_norm": 0.23887838423252106,
      "learning_rate": 7.169962407294201e-07,
      "loss": 0.0008,
      "step": 263400
    },
    {
      "epoch": 1.6894299192195823,
      "grad_norm": 0.034622084349393845,
      "learning_rate": 7.16707471930409e-07,
      "loss": 0.0014,
      "step": 263410
    },
    {
      "epoch": 1.6894940561133684,
      "grad_norm": 0.011238383129239082,
      "learning_rate": 7.164187568041154e-07,
      "loss": 0.0009,
      "step": 263420
    },
    {
      "epoch": 1.6895581930071546,
      "grad_norm": 0.006505804136395454,
      "learning_rate": 7.161300953541556e-07,
      "loss": 0.001,
      "step": 263430
    },
    {
      "epoch": 1.6896223299009405,
      "grad_norm": 0.19434146583080292,
      "learning_rate": 7.158414875841457e-07,
      "loss": 0.0006,
      "step": 263440
    },
    {
      "epoch": 1.6896864667947267,
      "grad_norm": 0.03185209631919861,
      "learning_rate": 7.155529334977057e-07,
      "loss": 0.0009,
      "step": 263450
    },
    {
      "epoch": 1.6897506036885126,
      "grad_norm": 0.13237544894218445,
      "learning_rate": 7.152644330984493e-07,
      "loss": 0.0009,
      "step": 263460
    },
    {
      "epoch": 1.6898147405822987,
      "grad_norm": 0.03689480945467949,
      "learning_rate": 7.149759863899908e-07,
      "loss": 0.0006,
      "step": 263470
    },
    {
      "epoch": 1.689878877476085,
      "grad_norm": 0.1426839977502823,
      "learning_rate": 7.146875933759467e-07,
      "loss": 0.0011,
      "step": 263480
    },
    {
      "epoch": 1.689943014369871,
      "grad_norm": 0.26433318853378296,
      "learning_rate": 7.143992540599293e-07,
      "loss": 0.0015,
      "step": 263490
    },
    {
      "epoch": 1.6900071512636572,
      "grad_norm": 0.04924926534295082,
      "learning_rate": 7.141109684455517e-07,
      "loss": 0.0014,
      "step": 263500
    },
    {
      "epoch": 1.6900712881574433,
      "grad_norm": 0.09768155217170715,
      "learning_rate": 7.138227365364275e-07,
      "loss": 0.0016,
      "step": 263510
    },
    {
      "epoch": 1.6901354250512295,
      "grad_norm": 0.07902403920888901,
      "learning_rate": 7.135345583361663e-07,
      "loss": 0.001,
      "step": 263520
    },
    {
      "epoch": 1.6901995619450154,
      "grad_norm": 0.09526874125003815,
      "learning_rate": 7.132464338483818e-07,
      "loss": 0.0009,
      "step": 263530
    },
    {
      "epoch": 1.6902636988388016,
      "grad_norm": 0.08909189701080322,
      "learning_rate": 7.129583630766834e-07,
      "loss": 0.0012,
      "step": 263540
    },
    {
      "epoch": 1.6903278357325875,
      "grad_norm": 0.04073075205087662,
      "learning_rate": 7.126703460246809e-07,
      "loss": 0.0018,
      "step": 263550
    },
    {
      "epoch": 1.6903919726263736,
      "grad_norm": 0.09088864177465439,
      "learning_rate": 7.123823826959813e-07,
      "loss": 0.0009,
      "step": 263560
    },
    {
      "epoch": 1.6904561095201598,
      "grad_norm": 0.002805843949317932,
      "learning_rate": 7.120944730941964e-07,
      "loss": 0.0006,
      "step": 263570
    },
    {
      "epoch": 1.690520246413946,
      "grad_norm": 0.0623597651720047,
      "learning_rate": 7.118066172229321e-07,
      "loss": 0.001,
      "step": 263580
    },
    {
      "epoch": 1.690584383307732,
      "grad_norm": 0.1109582707285881,
      "learning_rate": 7.115188150857943e-07,
      "loss": 0.0014,
      "step": 263590
    },
    {
      "epoch": 1.6906485202015182,
      "grad_norm": 0.013657116331160069,
      "learning_rate": 7.11231066686392e-07,
      "loss": 0.0007,
      "step": 263600
    },
    {
      "epoch": 1.6907126570953044,
      "grad_norm": 0.06236774101853371,
      "learning_rate": 7.109433720283299e-07,
      "loss": 0.0006,
      "step": 263610
    },
    {
      "epoch": 1.6907767939890903,
      "grad_norm": 0.025372015312314034,
      "learning_rate": 7.106557311152118e-07,
      "loss": 0.0014,
      "step": 263620
    },
    {
      "epoch": 1.6908409308828765,
      "grad_norm": 0.11978617310523987,
      "learning_rate": 7.103681439506427e-07,
      "loss": 0.0013,
      "step": 263630
    },
    {
      "epoch": 1.6909050677766624,
      "grad_norm": 0.20812635123729706,
      "learning_rate": 7.100806105382275e-07,
      "loss": 0.0016,
      "step": 263640
    },
    {
      "epoch": 1.6909692046704485,
      "grad_norm": 0.056038640439510345,
      "learning_rate": 7.097931308815676e-07,
      "loss": 0.0008,
      "step": 263650
    },
    {
      "epoch": 1.6910333415642347,
      "grad_norm": 0.017140207812190056,
      "learning_rate": 7.095057049842662e-07,
      "loss": 0.0008,
      "step": 263660
    },
    {
      "epoch": 1.6910974784580208,
      "grad_norm": 0.10366932302713394,
      "learning_rate": 7.092183328499241e-07,
      "loss": 0.0009,
      "step": 263670
    },
    {
      "epoch": 1.691161615351807,
      "grad_norm": 0.10067028552293777,
      "learning_rate": 7.089310144821432e-07,
      "loss": 0.0009,
      "step": 263680
    },
    {
      "epoch": 1.6912257522455931,
      "grad_norm": 0.12825380265712738,
      "learning_rate": 7.086437498845239e-07,
      "loss": 0.0012,
      "step": 263690
    },
    {
      "epoch": 1.691289889139379,
      "grad_norm": 0.04232438653707504,
      "learning_rate": 7.083565390606656e-07,
      "loss": 0.0016,
      "step": 263700
    },
    {
      "epoch": 1.6913540260331652,
      "grad_norm": 0.01695011369884014,
      "learning_rate": 7.080693820141654e-07,
      "loss": 0.0007,
      "step": 263710
    },
    {
      "epoch": 1.6914181629269511,
      "grad_norm": 0.0885840430855751,
      "learning_rate": 7.077822787486249e-07,
      "loss": 0.0019,
      "step": 263720
    },
    {
      "epoch": 1.6914822998207373,
      "grad_norm": 0.0035042043309658766,
      "learning_rate": 7.074952292676396e-07,
      "loss": 0.0006,
      "step": 263730
    },
    {
      "epoch": 1.6915464367145234,
      "grad_norm": 0.07663535326719284,
      "learning_rate": 7.07208233574806e-07,
      "loss": 0.0021,
      "step": 263740
    },
    {
      "epoch": 1.6916105736083096,
      "grad_norm": 0.07760876417160034,
      "learning_rate": 7.069212916737223e-07,
      "loss": 0.0012,
      "step": 263750
    },
    {
      "epoch": 1.6916747105020957,
      "grad_norm": 0.05023767799139023,
      "learning_rate": 7.066344035679828e-07,
      "loss": 0.0008,
      "step": 263760
    },
    {
      "epoch": 1.6917388473958819,
      "grad_norm": 0.048591502010822296,
      "learning_rate": 7.063475692611832e-07,
      "loss": 0.0009,
      "step": 263770
    },
    {
      "epoch": 1.691802984289668,
      "grad_norm": 0.08918386697769165,
      "learning_rate": 7.06060788756916e-07,
      "loss": 0.001,
      "step": 263780
    },
    {
      "epoch": 1.691867121183454,
      "grad_norm": 0.031152892857789993,
      "learning_rate": 7.057740620587766e-07,
      "loss": 0.0009,
      "step": 263790
    },
    {
      "epoch": 1.69193125807724,
      "grad_norm": 0.07061514258384705,
      "learning_rate": 7.054873891703573e-07,
      "loss": 0.0009,
      "step": 263800
    },
    {
      "epoch": 1.691995394971026,
      "grad_norm": 0.14635823667049408,
      "learning_rate": 7.052007700952506e-07,
      "loss": 0.001,
      "step": 263810
    },
    {
      "epoch": 1.6920595318648122,
      "grad_norm": 0.10236475616693497,
      "learning_rate": 7.049142048370461e-07,
      "loss": 0.0016,
      "step": 263820
    },
    {
      "epoch": 1.6921236687585983,
      "grad_norm": 0.10531385987997055,
      "learning_rate": 7.046276933993379e-07,
      "loss": 0.0016,
      "step": 263830
    },
    {
      "epoch": 1.6921878056523845,
      "grad_norm": 0.04722611606121063,
      "learning_rate": 7.043412357857144e-07,
      "loss": 0.0015,
      "step": 263840
    },
    {
      "epoch": 1.6922519425461706,
      "grad_norm": 0.03881454840302467,
      "learning_rate": 7.040548319997653e-07,
      "loss": 0.0009,
      "step": 263850
    },
    {
      "epoch": 1.6923160794399568,
      "grad_norm": 0.08548258990049362,
      "learning_rate": 7.03768482045078e-07,
      "loss": 0.0007,
      "step": 263860
    },
    {
      "epoch": 1.6923802163337427,
      "grad_norm": 0.4337688088417053,
      "learning_rate": 7.034821859252433e-07,
      "loss": 0.0016,
      "step": 263870
    },
    {
      "epoch": 1.6924443532275288,
      "grad_norm": 0.09530623257160187,
      "learning_rate": 7.03195943643848e-07,
      "loss": 0.0014,
      "step": 263880
    },
    {
      "epoch": 1.6925084901213148,
      "grad_norm": 0.09382256120443344,
      "learning_rate": 7.029097552044767e-07,
      "loss": 0.0007,
      "step": 263890
    },
    {
      "epoch": 1.692572627015101,
      "grad_norm": 0.07203307747840881,
      "learning_rate": 7.026236206107185e-07,
      "loss": 0.0023,
      "step": 263900
    },
    {
      "epoch": 1.692636763908887,
      "grad_norm": 0.0677449181675911,
      "learning_rate": 7.02337539866158e-07,
      "loss": 0.0013,
      "step": 263910
    },
    {
      "epoch": 1.6927009008026732,
      "grad_norm": 0.09249774366617203,
      "learning_rate": 7.020515129743794e-07,
      "loss": 0.001,
      "step": 263920
    },
    {
      "epoch": 1.6927650376964594,
      "grad_norm": 0.19315855205059052,
      "learning_rate": 7.017655399389656e-07,
      "loss": 0.0009,
      "step": 263930
    },
    {
      "epoch": 1.6928291745902455,
      "grad_norm": 0.04837198182940483,
      "learning_rate": 7.014796207635027e-07,
      "loss": 0.0008,
      "step": 263940
    },
    {
      "epoch": 1.6928933114840317,
      "grad_norm": 0.3748023211956024,
      "learning_rate": 7.011937554515724e-07,
      "loss": 0.0014,
      "step": 263950
    },
    {
      "epoch": 1.6929574483778176,
      "grad_norm": 0.15534189343452454,
      "learning_rate": 7.009079440067567e-07,
      "loss": 0.0011,
      "step": 263960
    },
    {
      "epoch": 1.6930215852716037,
      "grad_norm": 0.04076743870973587,
      "learning_rate": 7.006221864326357e-07,
      "loss": 0.0009,
      "step": 263970
    },
    {
      "epoch": 1.6930857221653897,
      "grad_norm": 0.01948699913918972,
      "learning_rate": 7.00336482732793e-07,
      "loss": 0.0014,
      "step": 263980
    },
    {
      "epoch": 1.6931498590591758,
      "grad_norm": 0.04300503805279732,
      "learning_rate": 7.000508329108063e-07,
      "loss": 0.0008,
      "step": 263990
    },
    {
      "epoch": 1.693213995952962,
      "grad_norm": 0.25529250502586365,
      "learning_rate": 6.997652369702562e-07,
      "loss": 0.0011,
      "step": 264000
    },
    {
      "epoch": 1.693278132846748,
      "grad_norm": 0.03575824573636055,
      "learning_rate": 6.994796949147204e-07,
      "loss": 0.0003,
      "step": 264010
    },
    {
      "epoch": 1.6933422697405343,
      "grad_norm": 0.0745188519358635,
      "learning_rate": 6.99194206747778e-07,
      "loss": 0.0007,
      "step": 264020
    },
    {
      "epoch": 1.6934064066343204,
      "grad_norm": 0.027460288256406784,
      "learning_rate": 6.989087724730059e-07,
      "loss": 0.001,
      "step": 264030
    },
    {
      "epoch": 1.6934705435281066,
      "grad_norm": 0.04555510729551315,
      "learning_rate": 6.9862339209398e-07,
      "loss": 0.0007,
      "step": 264040
    },
    {
      "epoch": 1.6935346804218925,
      "grad_norm": 0.02433430403470993,
      "learning_rate": 6.983380656142785e-07,
      "loss": 0.0012,
      "step": 264050
    },
    {
      "epoch": 1.6935988173156786,
      "grad_norm": 0.06247928366065025,
      "learning_rate": 6.980527930374748e-07,
      "loss": 0.0009,
      "step": 264060
    },
    {
      "epoch": 1.6936629542094646,
      "grad_norm": 0.022036660462617874,
      "learning_rate": 6.977675743671448e-07,
      "loss": 0.0007,
      "step": 264070
    },
    {
      "epoch": 1.6937270911032507,
      "grad_norm": 0.10363399982452393,
      "learning_rate": 6.9748240960686e-07,
      "loss": 0.0019,
      "step": 264080
    },
    {
      "epoch": 1.6937912279970369,
      "grad_norm": 0.041434530168771744,
      "learning_rate": 6.971972987601972e-07,
      "loss": 0.0005,
      "step": 264090
    },
    {
      "epoch": 1.693855364890823,
      "grad_norm": 0.051926229149103165,
      "learning_rate": 6.969122418307272e-07,
      "loss": 0.001,
      "step": 264100
    },
    {
      "epoch": 1.6939195017846091,
      "grad_norm": 0.12077579647302628,
      "learning_rate": 6.966272388220224e-07,
      "loss": 0.0011,
      "step": 264110
    },
    {
      "epoch": 1.6939836386783953,
      "grad_norm": 0.23285746574401855,
      "learning_rate": 6.963422897376526e-07,
      "loss": 0.0012,
      "step": 264120
    },
    {
      "epoch": 1.6940477755721812,
      "grad_norm": 0.08923560380935669,
      "learning_rate": 6.96057394581191e-07,
      "loss": 0.0006,
      "step": 264130
    },
    {
      "epoch": 1.6941119124659674,
      "grad_norm": 0.03662756457924843,
      "learning_rate": 6.95772553356206e-07,
      "loss": 0.0012,
      "step": 264140
    },
    {
      "epoch": 1.6941760493597533,
      "grad_norm": 0.04590865969657898,
      "learning_rate": 6.954877660662673e-07,
      "loss": 0.0004,
      "step": 264150
    },
    {
      "epoch": 1.6942401862535394,
      "grad_norm": 0.07533108443021774,
      "learning_rate": 6.952030327149417e-07,
      "loss": 0.0007,
      "step": 264160
    },
    {
      "epoch": 1.6943043231473256,
      "grad_norm": 0.0856962725520134,
      "learning_rate": 6.949183533058002e-07,
      "loss": 0.0013,
      "step": 264170
    },
    {
      "epoch": 1.6943684600411117,
      "grad_norm": 0.09973156452178955,
      "learning_rate": 6.946337278424087e-07,
      "loss": 0.0009,
      "step": 264180
    },
    {
      "epoch": 1.694432596934898,
      "grad_norm": 0.10728974640369415,
      "learning_rate": 6.943491563283322e-07,
      "loss": 0.001,
      "step": 264190
    },
    {
      "epoch": 1.694496733828684,
      "grad_norm": 0.10567191243171692,
      "learning_rate": 6.94064638767139e-07,
      "loss": 0.0008,
      "step": 264200
    },
    {
      "epoch": 1.6945608707224702,
      "grad_norm": 0.021192071959376335,
      "learning_rate": 6.937801751623935e-07,
      "loss": 0.0013,
      "step": 264210
    },
    {
      "epoch": 1.6946250076162561,
      "grad_norm": 0.006901530083268881,
      "learning_rate": 6.9349576551766e-07,
      "loss": 0.0012,
      "step": 264220
    },
    {
      "epoch": 1.6946891445100423,
      "grad_norm": 0.05930528789758682,
      "learning_rate": 6.932114098365006e-07,
      "loss": 0.0016,
      "step": 264230
    },
    {
      "epoch": 1.6947532814038282,
      "grad_norm": 0.11704102903604507,
      "learning_rate": 6.92927108122482e-07,
      "loss": 0.0014,
      "step": 264240
    },
    {
      "epoch": 1.6948174182976143,
      "grad_norm": 0.08013855665922165,
      "learning_rate": 6.926428603791652e-07,
      "loss": 0.0012,
      "step": 264250
    },
    {
      "epoch": 1.6948815551914005,
      "grad_norm": 0.09500796347856522,
      "learning_rate": 6.923586666101112e-07,
      "loss": 0.0012,
      "step": 264260
    },
    {
      "epoch": 1.6949456920851866,
      "grad_norm": 0.10292333364486694,
      "learning_rate": 6.920745268188811e-07,
      "loss": 0.001,
      "step": 264270
    },
    {
      "epoch": 1.6950098289789728,
      "grad_norm": 0.11834751069545746,
      "learning_rate": 6.917904410090359e-07,
      "loss": 0.0012,
      "step": 264280
    },
    {
      "epoch": 1.695073965872759,
      "grad_norm": 0.03635682910680771,
      "learning_rate": 6.915064091841361e-07,
      "loss": 0.0024,
      "step": 264290
    },
    {
      "epoch": 1.6951381027665449,
      "grad_norm": 0.22080478072166443,
      "learning_rate": 6.912224313477406e-07,
      "loss": 0.0018,
      "step": 264300
    },
    {
      "epoch": 1.695202239660331,
      "grad_norm": 0.0906863883137703,
      "learning_rate": 6.909385075034065e-07,
      "loss": 0.001,
      "step": 264310
    },
    {
      "epoch": 1.6952663765541172,
      "grad_norm": 0.0966591164469719,
      "learning_rate": 6.906546376546936e-07,
      "loss": 0.0009,
      "step": 264320
    },
    {
      "epoch": 1.695330513447903,
      "grad_norm": 0.04243626818060875,
      "learning_rate": 6.90370821805158e-07,
      "loss": 0.0018,
      "step": 264330
    },
    {
      "epoch": 1.6953946503416892,
      "grad_norm": 0.01818661577999592,
      "learning_rate": 6.900870599583559e-07,
      "loss": 0.0008,
      "step": 264340
    },
    {
      "epoch": 1.6954587872354754,
      "grad_norm": 0.0733444094657898,
      "learning_rate": 6.898033521178421e-07,
      "loss": 0.0014,
      "step": 264350
    },
    {
      "epoch": 1.6955229241292615,
      "grad_norm": 0.0397791862487793,
      "learning_rate": 6.895196982871738e-07,
      "loss": 0.0009,
      "step": 264360
    },
    {
      "epoch": 1.6955870610230477,
      "grad_norm": 0.0348215214908123,
      "learning_rate": 6.892360984699042e-07,
      "loss": 0.0034,
      "step": 264370
    },
    {
      "epoch": 1.6956511979168338,
      "grad_norm": 0.056535977870225906,
      "learning_rate": 6.889525526695873e-07,
      "loss": 0.001,
      "step": 264380
    },
    {
      "epoch": 1.6957153348106198,
      "grad_norm": 0.002477928763255477,
      "learning_rate": 6.88669060889775e-07,
      "loss": 0.001,
      "step": 264390
    },
    {
      "epoch": 1.695779471704406,
      "grad_norm": 0.05568133294582367,
      "learning_rate": 6.883856231340219e-07,
      "loss": 0.0008,
      "step": 264400
    },
    {
      "epoch": 1.6958436085981918,
      "grad_norm": 0.3743601143360138,
      "learning_rate": 6.881022394058779e-07,
      "loss": 0.0011,
      "step": 264410
    },
    {
      "epoch": 1.695907745491978,
      "grad_norm": 0.11811339855194092,
      "learning_rate": 6.87818909708895e-07,
      "loss": 0.0014,
      "step": 264420
    },
    {
      "epoch": 1.6959718823857641,
      "grad_norm": 0.10599221289157867,
      "learning_rate": 6.875356340466216e-07,
      "loss": 0.0012,
      "step": 264430
    },
    {
      "epoch": 1.6960360192795503,
      "grad_norm": 0.10912429541349411,
      "learning_rate": 6.872524124226099e-07,
      "loss": 0.001,
      "step": 264440
    },
    {
      "epoch": 1.6961001561733364,
      "grad_norm": 0.04976571723818779,
      "learning_rate": 6.869692448404081e-07,
      "loss": 0.0013,
      "step": 264450
    },
    {
      "epoch": 1.6961642930671226,
      "grad_norm": 0.12095761299133301,
      "learning_rate": 6.866861313035622e-07,
      "loss": 0.0007,
      "step": 264460
    },
    {
      "epoch": 1.6962284299609087,
      "grad_norm": 0.07463040202856064,
      "learning_rate": 6.864030718156234e-07,
      "loss": 0.0008,
      "step": 264470
    },
    {
      "epoch": 1.6962925668546947,
      "grad_norm": 0.04961828142404556,
      "learning_rate": 6.861200663801371e-07,
      "loss": 0.0009,
      "step": 264480
    },
    {
      "epoch": 1.6963567037484808,
      "grad_norm": 0.03794373199343681,
      "learning_rate": 6.858371150006493e-07,
      "loss": 0.0015,
      "step": 264490
    },
    {
      "epoch": 1.6964208406422667,
      "grad_norm": 0.013013191521167755,
      "learning_rate": 6.855542176807051e-07,
      "loss": 0.0004,
      "step": 264500
    },
    {
      "epoch": 1.6964849775360529,
      "grad_norm": 0.018160870298743248,
      "learning_rate": 6.852713744238504e-07,
      "loss": 0.0018,
      "step": 264510
    },
    {
      "epoch": 1.696549114429839,
      "grad_norm": 0.14764241874217987,
      "learning_rate": 6.849885852336297e-07,
      "loss": 0.0023,
      "step": 264520
    },
    {
      "epoch": 1.6966132513236252,
      "grad_norm": 0.05118638649582863,
      "learning_rate": 6.847058501135856e-07,
      "loss": 0.0009,
      "step": 264530
    },
    {
      "epoch": 1.6966773882174113,
      "grad_norm": 0.17762254178524017,
      "learning_rate": 6.844231690672599e-07,
      "loss": 0.001,
      "step": 264540
    },
    {
      "epoch": 1.6967415251111975,
      "grad_norm": 0.07955456525087357,
      "learning_rate": 6.841405420981978e-07,
      "loss": 0.0023,
      "step": 264550
    },
    {
      "epoch": 1.6968056620049834,
      "grad_norm": 0.08135933429002762,
      "learning_rate": 6.838579692099389e-07,
      "loss": 0.0041,
      "step": 264560
    },
    {
      "epoch": 1.6968697988987695,
      "grad_norm": 0.04895102232694626,
      "learning_rate": 6.835754504060244e-07,
      "loss": 0.0013,
      "step": 264570
    },
    {
      "epoch": 1.6969339357925555,
      "grad_norm": 0.12795418500900269,
      "learning_rate": 6.832929856899934e-07,
      "loss": 0.0013,
      "step": 264580
    },
    {
      "epoch": 1.6969980726863416,
      "grad_norm": 0.08570278435945511,
      "learning_rate": 6.830105750653875e-07,
      "loss": 0.001,
      "step": 264590
    },
    {
      "epoch": 1.6970622095801278,
      "grad_norm": 0.03363915905356407,
      "learning_rate": 6.827282185357442e-07,
      "loss": 0.001,
      "step": 264600
    },
    {
      "epoch": 1.697126346473914,
      "grad_norm": 0.03974626213312149,
      "learning_rate": 6.82445916104601e-07,
      "loss": 0.0009,
      "step": 264610
    },
    {
      "epoch": 1.6971904833677,
      "grad_norm": 0.024083245545625687,
      "learning_rate": 6.821636677754967e-07,
      "loss": 0.0009,
      "step": 264620
    },
    {
      "epoch": 1.6972546202614862,
      "grad_norm": 0.036198630928993225,
      "learning_rate": 6.818814735519685e-07,
      "loss": 0.0011,
      "step": 264630
    },
    {
      "epoch": 1.6973187571552724,
      "grad_norm": 0.06512334942817688,
      "learning_rate": 6.815993334375509e-07,
      "loss": 0.0012,
      "step": 264640
    },
    {
      "epoch": 1.6973828940490583,
      "grad_norm": 0.08642488718032837,
      "learning_rate": 6.813172474357788e-07,
      "loss": 0.0009,
      "step": 264650
    },
    {
      "epoch": 1.6974470309428444,
      "grad_norm": 0.025644278153777122,
      "learning_rate": 6.810352155501898e-07,
      "loss": 0.0013,
      "step": 264660
    },
    {
      "epoch": 1.6975111678366304,
      "grad_norm": 0.1016627624630928,
      "learning_rate": 6.807532377843157e-07,
      "loss": 0.0009,
      "step": 264670
    },
    {
      "epoch": 1.6975753047304165,
      "grad_norm": 0.2199431210756302,
      "learning_rate": 6.804713141416907e-07,
      "loss": 0.0009,
      "step": 264680
    },
    {
      "epoch": 1.6976394416242027,
      "grad_norm": 0.008409027941524982,
      "learning_rate": 6.801894446258462e-07,
      "loss": 0.0009,
      "step": 264690
    },
    {
      "epoch": 1.6977035785179888,
      "grad_norm": 0.10022395849227905,
      "learning_rate": 6.799076292403161e-07,
      "loss": 0.0008,
      "step": 264700
    },
    {
      "epoch": 1.697767715411775,
      "grad_norm": 0.11597537249326706,
      "learning_rate": 6.796258679886309e-07,
      "loss": 0.0013,
      "step": 264710
    },
    {
      "epoch": 1.697831852305561,
      "grad_norm": 0.06105426326394081,
      "learning_rate": 6.793441608743206e-07,
      "loss": 0.0018,
      "step": 264720
    },
    {
      "epoch": 1.697895989199347,
      "grad_norm": 0.01340585295110941,
      "learning_rate": 6.790625079009167e-07,
      "loss": 0.0014,
      "step": 264730
    },
    {
      "epoch": 1.6979601260931332,
      "grad_norm": 0.2167714387178421,
      "learning_rate": 6.787809090719477e-07,
      "loss": 0.0018,
      "step": 264740
    },
    {
      "epoch": 1.6980242629869193,
      "grad_norm": 0.249318465590477,
      "learning_rate": 6.784993643909426e-07,
      "loss": 0.0016,
      "step": 264750
    },
    {
      "epoch": 1.6980883998807053,
      "grad_norm": 0.08137817680835724,
      "learning_rate": 6.782178738614276e-07,
      "loss": 0.0005,
      "step": 264760
    },
    {
      "epoch": 1.6981525367744914,
      "grad_norm": 0.0265290979295969,
      "learning_rate": 6.779364374869324e-07,
      "loss": 0.0009,
      "step": 264770
    },
    {
      "epoch": 1.6982166736682776,
      "grad_norm": 0.12711146473884583,
      "learning_rate": 6.776550552709827e-07,
      "loss": 0.0005,
      "step": 264780
    },
    {
      "epoch": 1.6982808105620637,
      "grad_norm": 0.08862216770648956,
      "learning_rate": 6.773737272171043e-07,
      "loss": 0.0007,
      "step": 264790
    },
    {
      "epoch": 1.6983449474558499,
      "grad_norm": 0.08041568100452423,
      "learning_rate": 6.770924533288215e-07,
      "loss": 0.0008,
      "step": 264800
    },
    {
      "epoch": 1.698409084349636,
      "grad_norm": 0.0358443446457386,
      "learning_rate": 6.768112336096605e-07,
      "loss": 0.0009,
      "step": 264810
    },
    {
      "epoch": 1.698473221243422,
      "grad_norm": 0.02899232506752014,
      "learning_rate": 6.765300680631448e-07,
      "loss": 0.0027,
      "step": 264820
    },
    {
      "epoch": 1.698537358137208,
      "grad_norm": 0.012916501611471176,
      "learning_rate": 6.762489566927971e-07,
      "loss": 0.0008,
      "step": 264830
    },
    {
      "epoch": 1.698601495030994,
      "grad_norm": 0.06844992935657501,
      "learning_rate": 6.75967899502139e-07,
      "loss": 0.0011,
      "step": 264840
    },
    {
      "epoch": 1.6986656319247802,
      "grad_norm": 0.1103900671005249,
      "learning_rate": 6.756868964946944e-07,
      "loss": 0.0012,
      "step": 264850
    },
    {
      "epoch": 1.6987297688185663,
      "grad_norm": 0.17282582819461823,
      "learning_rate": 6.754059476739838e-07,
      "loss": 0.0027,
      "step": 264860
    },
    {
      "epoch": 1.6987939057123524,
      "grad_norm": 0.1048382967710495,
      "learning_rate": 6.751250530435261e-07,
      "loss": 0.0019,
      "step": 264870
    },
    {
      "epoch": 1.6988580426061386,
      "grad_norm": 0.03646853566169739,
      "learning_rate": 6.748442126068439e-07,
      "loss": 0.0006,
      "step": 264880
    },
    {
      "epoch": 1.6989221794999247,
      "grad_norm": 0.030923152342438698,
      "learning_rate": 6.745634263674544e-07,
      "loss": 0.0016,
      "step": 264890
    },
    {
      "epoch": 1.698986316393711,
      "grad_norm": 0.061948273330926895,
      "learning_rate": 6.742826943288771e-07,
      "loss": 0.0009,
      "step": 264900
    },
    {
      "epoch": 1.6990504532874968,
      "grad_norm": 0.0893976241350174,
      "learning_rate": 6.740020164946276e-07,
      "loss": 0.0005,
      "step": 264910
    },
    {
      "epoch": 1.699114590181283,
      "grad_norm": 0.15201431512832642,
      "learning_rate": 6.737213928682257e-07,
      "loss": 0.002,
      "step": 264920
    },
    {
      "epoch": 1.699178727075069,
      "grad_norm": 0.03452815115451813,
      "learning_rate": 6.734408234531869e-07,
      "loss": 0.0008,
      "step": 264930
    },
    {
      "epoch": 1.699242863968855,
      "grad_norm": 0.04301924630999565,
      "learning_rate": 6.731603082530269e-07,
      "loss": 0.0013,
      "step": 264940
    },
    {
      "epoch": 1.6993070008626412,
      "grad_norm": 0.0761362686753273,
      "learning_rate": 6.728798472712589e-07,
      "loss": 0.001,
      "step": 264950
    },
    {
      "epoch": 1.6993711377564273,
      "grad_norm": 0.07112876325845718,
      "learning_rate": 6.725994405114006e-07,
      "loss": 0.0007,
      "step": 264960
    },
    {
      "epoch": 1.6994352746502135,
      "grad_norm": 0.05872325971722603,
      "learning_rate": 6.723190879769637e-07,
      "loss": 0.0006,
      "step": 264970
    },
    {
      "epoch": 1.6994994115439996,
      "grad_norm": 0.09243052452802658,
      "learning_rate": 6.720387896714619e-07,
      "loss": 0.0008,
      "step": 264980
    },
    {
      "epoch": 1.6995635484377856,
      "grad_norm": 0.008637145161628723,
      "learning_rate": 6.717585455984061e-07,
      "loss": 0.0008,
      "step": 264990
    },
    {
      "epoch": 1.6996276853315717,
      "grad_norm": 0.05439627543091774,
      "learning_rate": 6.714783557613097e-07,
      "loss": 0.0007,
      "step": 265000
    },
    {
      "epoch": 1.6996918222253576,
      "grad_norm": 0.025693070143461227,
      "learning_rate": 6.711982201636835e-07,
      "loss": 0.0007,
      "step": 265010
    },
    {
      "epoch": 1.6997559591191438,
      "grad_norm": 0.048525068908929825,
      "learning_rate": 6.709181388090357e-07,
      "loss": 0.0006,
      "step": 265020
    },
    {
      "epoch": 1.69982009601293,
      "grad_norm": 0.05880725756287575,
      "learning_rate": 6.706381117008793e-07,
      "loss": 0.0008,
      "step": 265030
    },
    {
      "epoch": 1.699884232906716,
      "grad_norm": 0.11127229779958725,
      "learning_rate": 6.703581388427199e-07,
      "loss": 0.0012,
      "step": 265040
    },
    {
      "epoch": 1.6999483698005022,
      "grad_norm": 0.047811251133680344,
      "learning_rate": 6.70078220238069e-07,
      "loss": 0.0039,
      "step": 265050
    },
    {
      "epoch": 1.7000125066942884,
      "grad_norm": 0.2942372262477875,
      "learning_rate": 6.697983558904325e-07,
      "loss": 0.0011,
      "step": 265060
    },
    {
      "epoch": 1.7000766435880745,
      "grad_norm": 0.041605714708566666,
      "learning_rate": 6.69518545803316e-07,
      "loss": 0.0016,
      "step": 265070
    },
    {
      "epoch": 1.7001407804818605,
      "grad_norm": 0.09894415736198425,
      "learning_rate": 6.692387899802283e-07,
      "loss": 0.0014,
      "step": 265080
    },
    {
      "epoch": 1.7002049173756466,
      "grad_norm": 0.1805630326271057,
      "learning_rate": 6.689590884246733e-07,
      "loss": 0.0005,
      "step": 265090
    },
    {
      "epoch": 1.7002690542694325,
      "grad_norm": 0.04582097753882408,
      "learning_rate": 6.686794411401564e-07,
      "loss": 0.0014,
      "step": 265100
    },
    {
      "epoch": 1.7003331911632187,
      "grad_norm": 0.1644166111946106,
      "learning_rate": 6.683998481301812e-07,
      "loss": 0.0013,
      "step": 265110
    },
    {
      "epoch": 1.7003973280570048,
      "grad_norm": 0.17809748649597168,
      "learning_rate": 6.681203093982519e-07,
      "loss": 0.0013,
      "step": 265120
    },
    {
      "epoch": 1.700461464950791,
      "grad_norm": 0.034539878368377686,
      "learning_rate": 6.678408249478719e-07,
      "loss": 0.0018,
      "step": 265130
    },
    {
      "epoch": 1.7005256018445771,
      "grad_norm": 0.007027273066341877,
      "learning_rate": 6.67561394782541e-07,
      "loss": 0.0009,
      "step": 265140
    },
    {
      "epoch": 1.7005897387383633,
      "grad_norm": 0.0024962613824754953,
      "learning_rate": 6.672820189057628e-07,
      "loss": 0.0013,
      "step": 265150
    },
    {
      "epoch": 1.7006538756321494,
      "grad_norm": 0.3956187069416046,
      "learning_rate": 6.670026973210381e-07,
      "loss": 0.0021,
      "step": 265160
    },
    {
      "epoch": 1.7007180125259354,
      "grad_norm": 0.19051364064216614,
      "learning_rate": 6.667234300318659e-07,
      "loss": 0.0008,
      "step": 265170
    },
    {
      "epoch": 1.7007821494197215,
      "grad_norm": 0.47155243158340454,
      "learning_rate": 6.664442170417457e-07,
      "loss": 0.0011,
      "step": 265180
    },
    {
      "epoch": 1.7008462863135074,
      "grad_norm": 0.06036653369665146,
      "learning_rate": 6.661650583541767e-07,
      "loss": 0.0012,
      "step": 265190
    },
    {
      "epoch": 1.7009104232072936,
      "grad_norm": 0.03911900520324707,
      "learning_rate": 6.658859539726575e-07,
      "loss": 0.001,
      "step": 265200
    },
    {
      "epoch": 1.7009745601010797,
      "grad_norm": 0.025036673992872238,
      "learning_rate": 6.656069039006846e-07,
      "loss": 0.0018,
      "step": 265210
    },
    {
      "epoch": 1.7010386969948659,
      "grad_norm": 0.10333983600139618,
      "learning_rate": 6.653279081417541e-07,
      "loss": 0.0014,
      "step": 265220
    },
    {
      "epoch": 1.701102833888652,
      "grad_norm": 0.15717697143554688,
      "learning_rate": 6.650489666993637e-07,
      "loss": 0.0017,
      "step": 265230
    },
    {
      "epoch": 1.7011669707824382,
      "grad_norm": 0.03343525528907776,
      "learning_rate": 6.647700795770084e-07,
      "loss": 0.0014,
      "step": 265240
    },
    {
      "epoch": 1.701231107676224,
      "grad_norm": 0.028773343190550804,
      "learning_rate": 6.644912467781816e-07,
      "loss": 0.0005,
      "step": 265250
    },
    {
      "epoch": 1.7012952445700102,
      "grad_norm": 0.054235756397247314,
      "learning_rate": 6.642124683063772e-07,
      "loss": 0.0006,
      "step": 265260
    },
    {
      "epoch": 1.7013593814637962,
      "grad_norm": 0.08739051222801208,
      "learning_rate": 6.639337441650906e-07,
      "loss": 0.0011,
      "step": 265270
    },
    {
      "epoch": 1.7014235183575823,
      "grad_norm": 0.17551854252815247,
      "learning_rate": 6.63655074357813e-07,
      "loss": 0.0011,
      "step": 265280
    },
    {
      "epoch": 1.7014876552513685,
      "grad_norm": 0.14753693342208862,
      "learning_rate": 6.633764588880354e-07,
      "loss": 0.0009,
      "step": 265290
    },
    {
      "epoch": 1.7015517921451546,
      "grad_norm": 0.02053745463490486,
      "learning_rate": 6.630978977592512e-07,
      "loss": 0.0006,
      "step": 265300
    },
    {
      "epoch": 1.7016159290389408,
      "grad_norm": 0.04896795004606247,
      "learning_rate": 6.628193909749497e-07,
      "loss": 0.0012,
      "step": 265310
    },
    {
      "epoch": 1.701680065932727,
      "grad_norm": 0.027453456073999405,
      "learning_rate": 6.625409385386211e-07,
      "loss": 0.0008,
      "step": 265320
    },
    {
      "epoch": 1.701744202826513,
      "grad_norm": 0.02024935744702816,
      "learning_rate": 6.622625404537536e-07,
      "loss": 0.0006,
      "step": 265330
    },
    {
      "epoch": 1.701808339720299,
      "grad_norm": 0.14231550693511963,
      "learning_rate": 6.619841967238378e-07,
      "loss": 0.0009,
      "step": 265340
    },
    {
      "epoch": 1.7018724766140851,
      "grad_norm": 0.027371495962142944,
      "learning_rate": 6.617059073523597e-07,
      "loss": 0.0007,
      "step": 265350
    },
    {
      "epoch": 1.701936613507871,
      "grad_norm": 0.10268767923116684,
      "learning_rate": 6.614276723428076e-07,
      "loss": 0.0013,
      "step": 265360
    },
    {
      "epoch": 1.7020007504016572,
      "grad_norm": 0.2743382751941681,
      "learning_rate": 6.611494916986661e-07,
      "loss": 0.0019,
      "step": 265370
    },
    {
      "epoch": 1.7020648872954434,
      "grad_norm": 0.052332375198602676,
      "learning_rate": 6.608713654234239e-07,
      "loss": 0.0011,
      "step": 265380
    },
    {
      "epoch": 1.7021290241892295,
      "grad_norm": 0.001781732658855617,
      "learning_rate": 6.605932935205644e-07,
      "loss": 0.0009,
      "step": 265390
    },
    {
      "epoch": 1.7021931610830157,
      "grad_norm": 0.004635539371520281,
      "learning_rate": 6.603152759935722e-07,
      "loss": 0.0011,
      "step": 265400
    },
    {
      "epoch": 1.7022572979768018,
      "grad_norm": 0.06639865785837173,
      "learning_rate": 6.600373128459298e-07,
      "loss": 0.0023,
      "step": 265410
    },
    {
      "epoch": 1.7023214348705877,
      "grad_norm": 0.12824387848377228,
      "learning_rate": 6.597594040811228e-07,
      "loss": 0.0009,
      "step": 265420
    },
    {
      "epoch": 1.7023855717643739,
      "grad_norm": 0.05182370916008949,
      "learning_rate": 6.59481549702633e-07,
      "loss": 0.0011,
      "step": 265430
    },
    {
      "epoch": 1.7024497086581598,
      "grad_norm": 0.14896632730960846,
      "learning_rate": 6.592037497139397e-07,
      "loss": 0.002,
      "step": 265440
    },
    {
      "epoch": 1.702513845551946,
      "grad_norm": 0.012180998921394348,
      "learning_rate": 6.58926004118527e-07,
      "loss": 0.0018,
      "step": 265450
    },
    {
      "epoch": 1.702577982445732,
      "grad_norm": 0.1370176523923874,
      "learning_rate": 6.586483129198739e-07,
      "loss": 0.0009,
      "step": 265460
    },
    {
      "epoch": 1.7026421193395183,
      "grad_norm": 0.03550407662987709,
      "learning_rate": 6.583706761214603e-07,
      "loss": 0.0008,
      "step": 265470
    },
    {
      "epoch": 1.7027062562333044,
      "grad_norm": 0.21676664054393768,
      "learning_rate": 6.580930937267644e-07,
      "loss": 0.0014,
      "step": 265480
    },
    {
      "epoch": 1.7027703931270906,
      "grad_norm": 0.10779733210802078,
      "learning_rate": 6.578155657392654e-07,
      "loss": 0.0015,
      "step": 265490
    },
    {
      "epoch": 1.7028345300208767,
      "grad_norm": 0.03631667420268059,
      "learning_rate": 6.575380921624414e-07,
      "loss": 0.001,
      "step": 265500
    },
    {
      "epoch": 1.7028986669146626,
      "grad_norm": 0.126630499958992,
      "learning_rate": 6.572606729997682e-07,
      "loss": 0.001,
      "step": 265510
    },
    {
      "epoch": 1.7029628038084488,
      "grad_norm": 0.006550406105816364,
      "learning_rate": 6.569833082547217e-07,
      "loss": 0.0013,
      "step": 265520
    },
    {
      "epoch": 1.7030269407022347,
      "grad_norm": 0.13078109920024872,
      "learning_rate": 6.567059979307788e-07,
      "loss": 0.0019,
      "step": 265530
    },
    {
      "epoch": 1.7030910775960209,
      "grad_norm": 0.0882938951253891,
      "learning_rate": 6.564287420314142e-07,
      "loss": 0.0005,
      "step": 265540
    },
    {
      "epoch": 1.703155214489807,
      "grad_norm": 0.0037018307484686375,
      "learning_rate": 6.561515405601021e-07,
      "loss": 0.0038,
      "step": 265550
    },
    {
      "epoch": 1.7032193513835931,
      "grad_norm": 0.015162280760705471,
      "learning_rate": 6.558743935203138e-07,
      "loss": 0.002,
      "step": 265560
    },
    {
      "epoch": 1.7032834882773793,
      "grad_norm": 0.07943790405988693,
      "learning_rate": 6.555973009155259e-07,
      "loss": 0.0006,
      "step": 265570
    },
    {
      "epoch": 1.7033476251711654,
      "grad_norm": 0.042144257575273514,
      "learning_rate": 6.553202627492084e-07,
      "loss": 0.0004,
      "step": 265580
    },
    {
      "epoch": 1.7034117620649516,
      "grad_norm": 0.1977560967206955,
      "learning_rate": 6.550432790248318e-07,
      "loss": 0.0012,
      "step": 265590
    },
    {
      "epoch": 1.7034758989587375,
      "grad_norm": 0.06872736662626266,
      "learning_rate": 6.547663497458695e-07,
      "loss": 0.0007,
      "step": 265600
    },
    {
      "epoch": 1.7035400358525237,
      "grad_norm": 0.07133138179779053,
      "learning_rate": 6.544894749157904e-07,
      "loss": 0.0015,
      "step": 265610
    },
    {
      "epoch": 1.7036041727463096,
      "grad_norm": 0.09437665343284607,
      "learning_rate": 6.542126545380639e-07,
      "loss": 0.001,
      "step": 265620
    },
    {
      "epoch": 1.7036683096400957,
      "grad_norm": 0.09219174832105637,
      "learning_rate": 6.539358886161573e-07,
      "loss": 0.0007,
      "step": 265630
    },
    {
      "epoch": 1.703732446533882,
      "grad_norm": 0.007705396506935358,
      "learning_rate": 6.536591771535417e-07,
      "loss": 0.006,
      "step": 265640
    },
    {
      "epoch": 1.703796583427668,
      "grad_norm": 0.030093500390648842,
      "learning_rate": 6.533825201536826e-07,
      "loss": 0.0007,
      "step": 265650
    },
    {
      "epoch": 1.7038607203214542,
      "grad_norm": 0.053052615374326706,
      "learning_rate": 6.531059176200472e-07,
      "loss": 0.0027,
      "step": 265660
    },
    {
      "epoch": 1.7039248572152403,
      "grad_norm": 0.10145606845617294,
      "learning_rate": 6.528293695560999e-07,
      "loss": 0.0012,
      "step": 265670
    },
    {
      "epoch": 1.7039889941090263,
      "grad_norm": 0.055144038051366806,
      "learning_rate": 6.525528759653088e-07,
      "loss": 0.0007,
      "step": 265680
    },
    {
      "epoch": 1.7040531310028124,
      "grad_norm": 0.10749773681163788,
      "learning_rate": 6.522764368511375e-07,
      "loss": 0.001,
      "step": 265690
    },
    {
      "epoch": 1.7041172678965983,
      "grad_norm": 0.1056133508682251,
      "learning_rate": 6.520000522170488e-07,
      "loss": 0.0021,
      "step": 265700
    },
    {
      "epoch": 1.7041814047903845,
      "grad_norm": 0.034532222896814346,
      "learning_rate": 6.517237220665068e-07,
      "loss": 0.0008,
      "step": 265710
    },
    {
      "epoch": 1.7042455416841706,
      "grad_norm": 0.15641459822654724,
      "learning_rate": 6.514474464029752e-07,
      "loss": 0.0014,
      "step": 265720
    },
    {
      "epoch": 1.7043096785779568,
      "grad_norm": 0.010333506390452385,
      "learning_rate": 6.511712252299146e-07,
      "loss": 0.0007,
      "step": 265730
    },
    {
      "epoch": 1.704373815471743,
      "grad_norm": 0.013161303475499153,
      "learning_rate": 6.508950585507856e-07,
      "loss": 0.0005,
      "step": 265740
    },
    {
      "epoch": 1.704437952365529,
      "grad_norm": 0.06176147982478142,
      "learning_rate": 6.50618946369051e-07,
      "loss": 0.0014,
      "step": 265750
    },
    {
      "epoch": 1.7045020892593152,
      "grad_norm": 0.12416189163923264,
      "learning_rate": 6.50342888688169e-07,
      "loss": 0.0013,
      "step": 265760
    },
    {
      "epoch": 1.7045662261531012,
      "grad_norm": 0.03362249210476875,
      "learning_rate": 6.500668855115999e-07,
      "loss": 0.001,
      "step": 265770
    },
    {
      "epoch": 1.7046303630468873,
      "grad_norm": 0.004374053794890642,
      "learning_rate": 6.497909368428002e-07,
      "loss": 0.0013,
      "step": 265780
    },
    {
      "epoch": 1.7046944999406732,
      "grad_norm": 0.06230396404862404,
      "learning_rate": 6.495150426852298e-07,
      "loss": 0.0011,
      "step": 265790
    },
    {
      "epoch": 1.7047586368344594,
      "grad_norm": 0.04657968878746033,
      "learning_rate": 6.492392030423444e-07,
      "loss": 0.0015,
      "step": 265800
    },
    {
      "epoch": 1.7048227737282455,
      "grad_norm": 0.00979990791529417,
      "learning_rate": 6.489634179176025e-07,
      "loss": 0.0009,
      "step": 265810
    },
    {
      "epoch": 1.7048869106220317,
      "grad_norm": 0.08935471624135971,
      "learning_rate": 6.486876873144587e-07,
      "loss": 0.001,
      "step": 265820
    },
    {
      "epoch": 1.7049510475158178,
      "grad_norm": 0.021802430972456932,
      "learning_rate": 6.484120112363667e-07,
      "loss": 0.0007,
      "step": 265830
    },
    {
      "epoch": 1.705015184409604,
      "grad_norm": 0.04074636101722717,
      "learning_rate": 6.481363896867832e-07,
      "loss": 0.0018,
      "step": 265840
    },
    {
      "epoch": 1.70507932130339,
      "grad_norm": 0.04045892506837845,
      "learning_rate": 6.478608226691613e-07,
      "loss": 0.0017,
      "step": 265850
    },
    {
      "epoch": 1.705143458197176,
      "grad_norm": 0.009271270595490932,
      "learning_rate": 6.475853101869523e-07,
      "loss": 0.0004,
      "step": 265860
    },
    {
      "epoch": 1.7052075950909622,
      "grad_norm": 0.08048803359270096,
      "learning_rate": 6.473098522436111e-07,
      "loss": 0.0013,
      "step": 265870
    },
    {
      "epoch": 1.7052717319847481,
      "grad_norm": 0.007005834486335516,
      "learning_rate": 6.470344488425884e-07,
      "loss": 0.0018,
      "step": 265880
    },
    {
      "epoch": 1.7053358688785343,
      "grad_norm": 0.01865677908062935,
      "learning_rate": 6.467590999873352e-07,
      "loss": 0.0019,
      "step": 265890
    },
    {
      "epoch": 1.7054000057723204,
      "grad_norm": 0.17345519363880157,
      "learning_rate": 6.464838056813006e-07,
      "loss": 0.0017,
      "step": 265900
    },
    {
      "epoch": 1.7054641426661066,
      "grad_norm": 0.202212855219841,
      "learning_rate": 6.46208565927936e-07,
      "loss": 0.0011,
      "step": 265910
    },
    {
      "epoch": 1.7055282795598927,
      "grad_norm": 0.008908011019229889,
      "learning_rate": 6.4593338073069e-07,
      "loss": 0.0007,
      "step": 265920
    },
    {
      "epoch": 1.7055924164536789,
      "grad_norm": 0.2655397355556488,
      "learning_rate": 6.456582500930103e-07,
      "loss": 0.0007,
      "step": 265930
    },
    {
      "epoch": 1.7056565533474648,
      "grad_norm": 0.2044960856437683,
      "learning_rate": 6.453831740183441e-07,
      "loss": 0.0018,
      "step": 265940
    },
    {
      "epoch": 1.705720690241251,
      "grad_norm": 0.08434181660413742,
      "learning_rate": 6.451081525101399e-07,
      "loss": 0.0013,
      "step": 265950
    },
    {
      "epoch": 1.7057848271350369,
      "grad_norm": 0.004985256120562553,
      "learning_rate": 6.448331855718426e-07,
      "loss": 0.0021,
      "step": 265960
    },
    {
      "epoch": 1.705848964028823,
      "grad_norm": 0.10638520866632462,
      "learning_rate": 6.445582732068984e-07,
      "loss": 0.001,
      "step": 265970
    },
    {
      "epoch": 1.7059131009226092,
      "grad_norm": 0.07057823240756989,
      "learning_rate": 6.442834154187505e-07,
      "loss": 0.0008,
      "step": 265980
    },
    {
      "epoch": 1.7059772378163953,
      "grad_norm": 0.0023433228489011526,
      "learning_rate": 6.440086122108458e-07,
      "loss": 0.0055,
      "step": 265990
    },
    {
      "epoch": 1.7060413747101815,
      "grad_norm": 0.08313106000423431,
      "learning_rate": 6.437338635866258e-07,
      "loss": 0.0012,
      "step": 266000
    },
    {
      "epoch": 1.7061055116039676,
      "grad_norm": 0.06899221241474152,
      "learning_rate": 6.434591695495335e-07,
      "loss": 0.0014,
      "step": 266010
    },
    {
      "epoch": 1.7061696484977538,
      "grad_norm": 0.19987016916275024,
      "learning_rate": 6.431845301030115e-07,
      "loss": 0.0015,
      "step": 266020
    },
    {
      "epoch": 1.7062337853915397,
      "grad_norm": 0.11432585120201111,
      "learning_rate": 6.429099452505023e-07,
      "loss": 0.0018,
      "step": 266030
    },
    {
      "epoch": 1.7062979222853258,
      "grad_norm": 0.13025084137916565,
      "learning_rate": 6.426354149954445e-07,
      "loss": 0.0008,
      "step": 266040
    },
    {
      "epoch": 1.7063620591791118,
      "grad_norm": 0.09899530559778214,
      "learning_rate": 6.423609393412783e-07,
      "loss": 0.0008,
      "step": 266050
    },
    {
      "epoch": 1.706426196072898,
      "grad_norm": 0.0024011684581637383,
      "learning_rate": 6.420865182914454e-07,
      "loss": 0.0006,
      "step": 266060
    },
    {
      "epoch": 1.706490332966684,
      "grad_norm": 0.06698352843523026,
      "learning_rate": 6.418121518493825e-07,
      "loss": 0.0009,
      "step": 266070
    },
    {
      "epoch": 1.7065544698604702,
      "grad_norm": 0.12909150123596191,
      "learning_rate": 6.41537840018528e-07,
      "loss": 0.0007,
      "step": 266080
    },
    {
      "epoch": 1.7066186067542564,
      "grad_norm": 0.12435998022556305,
      "learning_rate": 6.412635828023183e-07,
      "loss": 0.0009,
      "step": 266090
    },
    {
      "epoch": 1.7066827436480425,
      "grad_norm": 0.44490617513656616,
      "learning_rate": 6.409893802041927e-07,
      "loss": 0.0017,
      "step": 266100
    },
    {
      "epoch": 1.7067468805418284,
      "grad_norm": 0.0475863553583622,
      "learning_rate": 6.40715232227585e-07,
      "loss": 0.0011,
      "step": 266110
    },
    {
      "epoch": 1.7068110174356146,
      "grad_norm": 0.043053992092609406,
      "learning_rate": 6.404411388759307e-07,
      "loss": 0.0011,
      "step": 266120
    },
    {
      "epoch": 1.7068751543294005,
      "grad_norm": 0.05306117981672287,
      "learning_rate": 6.401671001526644e-07,
      "loss": 0.0013,
      "step": 266130
    },
    {
      "epoch": 1.7069392912231867,
      "grad_norm": 0.032787881791591644,
      "learning_rate": 6.398931160612204e-07,
      "loss": 0.0007,
      "step": 266140
    },
    {
      "epoch": 1.7070034281169728,
      "grad_norm": 0.027778422459959984,
      "learning_rate": 6.396191866050328e-07,
      "loss": 0.001,
      "step": 266150
    },
    {
      "epoch": 1.707067565010759,
      "grad_norm": 0.2105724811553955,
      "learning_rate": 6.393453117875314e-07,
      "loss": 0.0022,
      "step": 266160
    },
    {
      "epoch": 1.707131701904545,
      "grad_norm": 0.03211251273751259,
      "learning_rate": 6.390714916121505e-07,
      "loss": 0.0006,
      "step": 266170
    },
    {
      "epoch": 1.7071958387983313,
      "grad_norm": 0.0578240305185318,
      "learning_rate": 6.387977260823214e-07,
      "loss": 0.0004,
      "step": 266180
    },
    {
      "epoch": 1.7072599756921174,
      "grad_norm": 0.07418222725391388,
      "learning_rate": 6.385240152014732e-07,
      "loss": 0.0009,
      "step": 266190
    },
    {
      "epoch": 1.7073241125859033,
      "grad_norm": 0.09191425144672394,
      "learning_rate": 6.382503589730349e-07,
      "loss": 0.0011,
      "step": 266200
    },
    {
      "epoch": 1.7073882494796895,
      "grad_norm": 0.051483865827322006,
      "learning_rate": 6.37976757400438e-07,
      "loss": 0.0007,
      "step": 266210
    },
    {
      "epoch": 1.7074523863734754,
      "grad_norm": 0.05915215611457825,
      "learning_rate": 6.377032104871101e-07,
      "loss": 0.0013,
      "step": 266220
    },
    {
      "epoch": 1.7075165232672616,
      "grad_norm": 0.01516183651983738,
      "learning_rate": 6.374297182364786e-07,
      "loss": 0.0012,
      "step": 266230
    },
    {
      "epoch": 1.7075806601610477,
      "grad_norm": 0.04072074219584465,
      "learning_rate": 6.371562806519699e-07,
      "loss": 0.0007,
      "step": 266240
    },
    {
      "epoch": 1.7076447970548339,
      "grad_norm": 0.11622676998376846,
      "learning_rate": 6.368828977370117e-07,
      "loss": 0.0012,
      "step": 266250
    },
    {
      "epoch": 1.70770893394862,
      "grad_norm": 0.12206113338470459,
      "learning_rate": 6.366095694950292e-07,
      "loss": 0.0011,
      "step": 266260
    },
    {
      "epoch": 1.7077730708424061,
      "grad_norm": 0.17869877815246582,
      "learning_rate": 6.363362959294472e-07,
      "loss": 0.0019,
      "step": 266270
    },
    {
      "epoch": 1.707837207736192,
      "grad_norm": 0.11581534892320633,
      "learning_rate": 6.360630770436887e-07,
      "loss": 0.0009,
      "step": 266280
    },
    {
      "epoch": 1.7079013446299782,
      "grad_norm": 0.04566968232393265,
      "learning_rate": 6.357899128411804e-07,
      "loss": 0.0005,
      "step": 266290
    },
    {
      "epoch": 1.7079654815237644,
      "grad_norm": 0.011661666445434093,
      "learning_rate": 6.355168033253428e-07,
      "loss": 0.001,
      "step": 266300
    },
    {
      "epoch": 1.7080296184175503,
      "grad_norm": 0.19823497533798218,
      "learning_rate": 6.352437484995988e-07,
      "loss": 0.0012,
      "step": 266310
    },
    {
      "epoch": 1.7080937553113364,
      "grad_norm": 0.07741516083478928,
      "learning_rate": 6.349707483673706e-07,
      "loss": 0.0042,
      "step": 266320
    },
    {
      "epoch": 1.7081578922051226,
      "grad_norm": 0.05639111250638962,
      "learning_rate": 6.346978029320782e-07,
      "loss": 0.0009,
      "step": 266330
    },
    {
      "epoch": 1.7082220290989087,
      "grad_norm": 0.2746449112892151,
      "learning_rate": 6.34424912197143e-07,
      "loss": 0.0006,
      "step": 266340
    },
    {
      "epoch": 1.708286165992695,
      "grad_norm": 0.08014754205942154,
      "learning_rate": 6.341520761659819e-07,
      "loss": 0.0014,
      "step": 266350
    },
    {
      "epoch": 1.708350302886481,
      "grad_norm": 0.00769168371334672,
      "learning_rate": 6.33879294842017e-07,
      "loss": 0.001,
      "step": 266360
    },
    {
      "epoch": 1.708414439780267,
      "grad_norm": 0.08942686766386032,
      "learning_rate": 6.336065682286647e-07,
      "loss": 0.0011,
      "step": 266370
    },
    {
      "epoch": 1.7084785766740531,
      "grad_norm": 0.08934418857097626,
      "learning_rate": 6.333338963293428e-07,
      "loss": 0.0011,
      "step": 266380
    },
    {
      "epoch": 1.708542713567839,
      "grad_norm": 0.004581431858241558,
      "learning_rate": 6.330612791474666e-07,
      "loss": 0.0012,
      "step": 266390
    },
    {
      "epoch": 1.7086068504616252,
      "grad_norm": 0.0772194117307663,
      "learning_rate": 6.327887166864549e-07,
      "loss": 0.0013,
      "step": 266400
    },
    {
      "epoch": 1.7086709873554113,
      "grad_norm": 0.0690506175160408,
      "learning_rate": 6.325162089497223e-07,
      "loss": 0.0028,
      "step": 266410
    },
    {
      "epoch": 1.7087351242491975,
      "grad_norm": 0.04467339441180229,
      "learning_rate": 6.322437559406825e-07,
      "loss": 0.001,
      "step": 266420
    },
    {
      "epoch": 1.7087992611429836,
      "grad_norm": 0.06026383116841316,
      "learning_rate": 6.31971357662749e-07,
      "loss": 0.0054,
      "step": 266430
    },
    {
      "epoch": 1.7088633980367698,
      "grad_norm": 0.037136465311050415,
      "learning_rate": 6.316990141193368e-07,
      "loss": 0.0008,
      "step": 266440
    },
    {
      "epoch": 1.708927534930556,
      "grad_norm": 0.08824778348207474,
      "learning_rate": 6.314267253138584e-07,
      "loss": 0.0009,
      "step": 266450
    },
    {
      "epoch": 1.7089916718243419,
      "grad_norm": 0.029125427827239037,
      "learning_rate": 6.311544912497241e-07,
      "loss": 0.0007,
      "step": 266460
    },
    {
      "epoch": 1.709055808718128,
      "grad_norm": 0.030814319849014282,
      "learning_rate": 6.308823119303481e-07,
      "loss": 0.0012,
      "step": 266470
    },
    {
      "epoch": 1.709119945611914,
      "grad_norm": 0.0323149673640728,
      "learning_rate": 6.306101873591386e-07,
      "loss": 0.0013,
      "step": 266480
    },
    {
      "epoch": 1.7091840825057,
      "grad_norm": 0.041041210293769836,
      "learning_rate": 6.303381175395062e-07,
      "loss": 0.0019,
      "step": 266490
    },
    {
      "epoch": 1.7092482193994862,
      "grad_norm": 0.0773228108882904,
      "learning_rate": 6.300661024748595e-07,
      "loss": 0.0009,
      "step": 266500
    },
    {
      "epoch": 1.7093123562932724,
      "grad_norm": 0.17675377428531647,
      "learning_rate": 6.297941421686083e-07,
      "loss": 0.0012,
      "step": 266510
    },
    {
      "epoch": 1.7093764931870585,
      "grad_norm": 0.02773331105709076,
      "learning_rate": 6.295222366241599e-07,
      "loss": 0.0012,
      "step": 266520
    },
    {
      "epoch": 1.7094406300808447,
      "grad_norm": 0.0747687891125679,
      "learning_rate": 6.292503858449217e-07,
      "loss": 0.0009,
      "step": 266530
    },
    {
      "epoch": 1.7095047669746306,
      "grad_norm": 0.272279292345047,
      "learning_rate": 6.289785898342987e-07,
      "loss": 0.0025,
      "step": 266540
    },
    {
      "epoch": 1.7095689038684168,
      "grad_norm": 0.011230595409870148,
      "learning_rate": 6.287068485956988e-07,
      "loss": 0.001,
      "step": 266550
    },
    {
      "epoch": 1.7096330407622027,
      "grad_norm": 0.08038709312677383,
      "learning_rate": 6.284351621325263e-07,
      "loss": 0.0011,
      "step": 266560
    },
    {
      "epoch": 1.7096971776559888,
      "grad_norm": 0.03474096208810806,
      "learning_rate": 6.281635304481847e-07,
      "loss": 0.0008,
      "step": 266570
    },
    {
      "epoch": 1.709761314549775,
      "grad_norm": 0.10244005173444748,
      "learning_rate": 6.278919535460781e-07,
      "loss": 0.0007,
      "step": 266580
    },
    {
      "epoch": 1.7098254514435611,
      "grad_norm": 0.06485792249441147,
      "learning_rate": 6.276204314296113e-07,
      "loss": 0.0006,
      "step": 266590
    },
    {
      "epoch": 1.7098895883373473,
      "grad_norm": 0.022783532738685608,
      "learning_rate": 6.273489641021857e-07,
      "loss": 0.0011,
      "step": 266600
    },
    {
      "epoch": 1.7099537252311334,
      "grad_norm": 0.005095191765576601,
      "learning_rate": 6.270775515672023e-07,
      "loss": 0.0009,
      "step": 266610
    },
    {
      "epoch": 1.7100178621249196,
      "grad_norm": 0.082610122859478,
      "learning_rate": 6.268061938280612e-07,
      "loss": 0.0012,
      "step": 266620
    },
    {
      "epoch": 1.7100819990187055,
      "grad_norm": 0.049736011773347855,
      "learning_rate": 6.265348908881658e-07,
      "loss": 0.0011,
      "step": 266630
    },
    {
      "epoch": 1.7101461359124916,
      "grad_norm": 0.06456831097602844,
      "learning_rate": 6.262636427509133e-07,
      "loss": 0.0005,
      "step": 266640
    },
    {
      "epoch": 1.7102102728062776,
      "grad_norm": 0.09709274023771286,
      "learning_rate": 6.259924494197034e-07,
      "loss": 0.0022,
      "step": 266650
    },
    {
      "epoch": 1.7102744097000637,
      "grad_norm": 0.0323229655623436,
      "learning_rate": 6.257213108979326e-07,
      "loss": 0.0009,
      "step": 266660
    },
    {
      "epoch": 1.7103385465938499,
      "grad_norm": 0.11305923014879227,
      "learning_rate": 6.254502271890017e-07,
      "loss": 0.0006,
      "step": 266670
    },
    {
      "epoch": 1.710402683487636,
      "grad_norm": 0.060932356864213943,
      "learning_rate": 6.251791982963057e-07,
      "loss": 0.0012,
      "step": 266680
    },
    {
      "epoch": 1.7104668203814222,
      "grad_norm": 0.10568568855524063,
      "learning_rate": 6.249082242232407e-07,
      "loss": 0.0011,
      "step": 266690
    },
    {
      "epoch": 1.7105309572752083,
      "grad_norm": 0.007015223614871502,
      "learning_rate": 6.246373049732018e-07,
      "loss": 0.0005,
      "step": 266700
    },
    {
      "epoch": 1.7105950941689945,
      "grad_norm": 0.05034947395324707,
      "learning_rate": 6.243664405495858e-07,
      "loss": 0.0032,
      "step": 266710
    },
    {
      "epoch": 1.7106592310627804,
      "grad_norm": 0.17234694957733154,
      "learning_rate": 6.24095630955785e-07,
      "loss": 0.0012,
      "step": 266720
    },
    {
      "epoch": 1.7107233679565665,
      "grad_norm": 0.06869015097618103,
      "learning_rate": 6.238248761951926e-07,
      "loss": 0.0014,
      "step": 266730
    },
    {
      "epoch": 1.7107875048503525,
      "grad_norm": 0.04930241405963898,
      "learning_rate": 6.235541762712027e-07,
      "loss": 0.001,
      "step": 266740
    },
    {
      "epoch": 1.7108516417441386,
      "grad_norm": 0.0738460049033165,
      "learning_rate": 6.232835311872071e-07,
      "loss": 0.0004,
      "step": 266750
    },
    {
      "epoch": 1.7109157786379248,
      "grad_norm": 0.08131708949804306,
      "learning_rate": 6.230129409465968e-07,
      "loss": 0.0018,
      "step": 266760
    },
    {
      "epoch": 1.710979915531711,
      "grad_norm": 0.12405913323163986,
      "learning_rate": 6.22742405552762e-07,
      "loss": 0.0013,
      "step": 266770
    },
    {
      "epoch": 1.711044052425497,
      "grad_norm": 0.039367906749248505,
      "learning_rate": 6.224719250090938e-07,
      "loss": 0.0011,
      "step": 266780
    },
    {
      "epoch": 1.7111081893192832,
      "grad_norm": 0.1633806675672531,
      "learning_rate": 6.222014993189806e-07,
      "loss": 0.001,
      "step": 266790
    },
    {
      "epoch": 1.7111723262130691,
      "grad_norm": 0.008292242884635925,
      "learning_rate": 6.21931128485812e-07,
      "loss": 0.0006,
      "step": 266800
    },
    {
      "epoch": 1.7112364631068553,
      "grad_norm": 0.01935173198580742,
      "learning_rate": 6.216608125129741e-07,
      "loss": 0.0007,
      "step": 266810
    },
    {
      "epoch": 1.7113006000006412,
      "grad_norm": 0.013553868047893047,
      "learning_rate": 6.213905514038559e-07,
      "loss": 0.0011,
      "step": 266820
    },
    {
      "epoch": 1.7113647368944274,
      "grad_norm": 0.2309228479862213,
      "learning_rate": 6.211203451618441e-07,
      "loss": 0.0022,
      "step": 266830
    },
    {
      "epoch": 1.7114288737882135,
      "grad_norm": 0.4577714204788208,
      "learning_rate": 6.208501937903227e-07,
      "loss": 0.0027,
      "step": 266840
    },
    {
      "epoch": 1.7114930106819997,
      "grad_norm": 0.0713023692369461,
      "learning_rate": 6.205800972926789e-07,
      "loss": 0.0013,
      "step": 266850
    },
    {
      "epoch": 1.7115571475757858,
      "grad_norm": 0.02272789739072323,
      "learning_rate": 6.203100556722969e-07,
      "loss": 0.0006,
      "step": 266860
    },
    {
      "epoch": 1.711621284469572,
      "grad_norm": 0.06002601981163025,
      "learning_rate": 6.200400689325597e-07,
      "loss": 0.0011,
      "step": 266870
    },
    {
      "epoch": 1.711685421363358,
      "grad_norm": 0.13648687303066254,
      "learning_rate": 6.197701370768494e-07,
      "loss": 0.0012,
      "step": 266880
    },
    {
      "epoch": 1.711749558257144,
      "grad_norm": 0.05618472024798393,
      "learning_rate": 6.195002601085515e-07,
      "loss": 0.0006,
      "step": 266890
    },
    {
      "epoch": 1.7118136951509302,
      "grad_norm": 0.02872016280889511,
      "learning_rate": 6.192304380310455e-07,
      "loss": 0.001,
      "step": 266900
    },
    {
      "epoch": 1.711877832044716,
      "grad_norm": 0.1917131394147873,
      "learning_rate": 6.189606708477131e-07,
      "loss": 0.001,
      "step": 266910
    },
    {
      "epoch": 1.7119419689385023,
      "grad_norm": 0.009861587546765804,
      "learning_rate": 6.186909585619339e-07,
      "loss": 0.0018,
      "step": 266920
    },
    {
      "epoch": 1.7120061058322884,
      "grad_norm": 0.05207042768597603,
      "learning_rate": 6.18421301177089e-07,
      "loss": 0.0008,
      "step": 266930
    },
    {
      "epoch": 1.7120702427260746,
      "grad_norm": 0.0531453862786293,
      "learning_rate": 6.181516986965569e-07,
      "loss": 0.0008,
      "step": 266940
    },
    {
      "epoch": 1.7121343796198607,
      "grad_norm": 0.06557393819093704,
      "learning_rate": 6.178821511237154e-07,
      "loss": 0.0015,
      "step": 266950
    },
    {
      "epoch": 1.7121985165136469,
      "grad_norm": 0.06830243021249771,
      "learning_rate": 6.176126584619413e-07,
      "loss": 0.0015,
      "step": 266960
    },
    {
      "epoch": 1.7122626534074328,
      "grad_norm": 0.1856570839881897,
      "learning_rate": 6.173432207146135e-07,
      "loss": 0.0022,
      "step": 266970
    },
    {
      "epoch": 1.712326790301219,
      "grad_norm": 0.05492860823869705,
      "learning_rate": 6.170738378851077e-07,
      "loss": 0.0011,
      "step": 266980
    },
    {
      "epoch": 1.7123909271950049,
      "grad_norm": 0.008831695653498173,
      "learning_rate": 6.168045099767983e-07,
      "loss": 0.0009,
      "step": 266990
    },
    {
      "epoch": 1.712455064088791,
      "grad_norm": 0.0019844414200633764,
      "learning_rate": 6.165352369930616e-07,
      "loss": 0.0007,
      "step": 267000
    },
    {
      "epoch": 1.7125192009825771,
      "grad_norm": 0.20661035180091858,
      "learning_rate": 6.162660189372716e-07,
      "loss": 0.0016,
      "step": 267010
    },
    {
      "epoch": 1.7125833378763633,
      "grad_norm": 0.13168048858642578,
      "learning_rate": 6.15996855812801e-07,
      "loss": 0.0011,
      "step": 267020
    },
    {
      "epoch": 1.7126474747701494,
      "grad_norm": 0.04329398646950722,
      "learning_rate": 6.157277476230222e-07,
      "loss": 0.0008,
      "step": 267030
    },
    {
      "epoch": 1.7127116116639356,
      "grad_norm": 0.08694454282522202,
      "learning_rate": 6.154586943713092e-07,
      "loss": 0.001,
      "step": 267040
    },
    {
      "epoch": 1.7127757485577217,
      "grad_norm": 0.06795962154865265,
      "learning_rate": 6.151896960610326e-07,
      "loss": 0.0009,
      "step": 267050
    },
    {
      "epoch": 1.7128398854515077,
      "grad_norm": 0.13790574669837952,
      "learning_rate": 6.149207526955625e-07,
      "loss": 0.0011,
      "step": 267060
    },
    {
      "epoch": 1.7129040223452938,
      "grad_norm": 0.052158210426568985,
      "learning_rate": 6.146518642782684e-07,
      "loss": 0.0023,
      "step": 267070
    },
    {
      "epoch": 1.7129681592390797,
      "grad_norm": 0.015230941586196423,
      "learning_rate": 6.143830308125215e-07,
      "loss": 0.0215,
      "step": 267080
    },
    {
      "epoch": 1.713032296132866,
      "grad_norm": 0.19992482662200928,
      "learning_rate": 6.141142523016896e-07,
      "loss": 0.0025,
      "step": 267090
    },
    {
      "epoch": 1.713096433026652,
      "grad_norm": 0.005246074870228767,
      "learning_rate": 6.138455287491408e-07,
      "loss": 0.0004,
      "step": 267100
    },
    {
      "epoch": 1.7131605699204382,
      "grad_norm": 0.10086475312709808,
      "learning_rate": 6.13576860158241e-07,
      "loss": 0.0011,
      "step": 267110
    },
    {
      "epoch": 1.7132247068142243,
      "grad_norm": 0.0043176934123039246,
      "learning_rate": 6.133082465323592e-07,
      "loss": 0.0012,
      "step": 267120
    },
    {
      "epoch": 1.7132888437080105,
      "grad_norm": 0.2594849169254303,
      "learning_rate": 6.130396878748607e-07,
      "loss": 0.001,
      "step": 267130
    },
    {
      "epoch": 1.7133529806017966,
      "grad_norm": 0.03310069814324379,
      "learning_rate": 6.127711841891087e-07,
      "loss": 0.0008,
      "step": 267140
    },
    {
      "epoch": 1.7134171174955826,
      "grad_norm": 0.06937534362077713,
      "learning_rate": 6.125027354784702e-07,
      "loss": 0.0014,
      "step": 267150
    },
    {
      "epoch": 1.7134812543893687,
      "grad_norm": 0.04508813098073006,
      "learning_rate": 6.122343417463083e-07,
      "loss": 0.0009,
      "step": 267160
    },
    {
      "epoch": 1.7135453912831546,
      "grad_norm": 0.06019587814807892,
      "learning_rate": 6.119660029959862e-07,
      "loss": 0.0013,
      "step": 267170
    },
    {
      "epoch": 1.7136095281769408,
      "grad_norm": 0.01498024258762598,
      "learning_rate": 6.116977192308649e-07,
      "loss": 0.0005,
      "step": 267180
    },
    {
      "epoch": 1.713673665070727,
      "grad_norm": 0.010349423624575138,
      "learning_rate": 6.114294904543084e-07,
      "loss": 0.0013,
      "step": 267190
    },
    {
      "epoch": 1.713737801964513,
      "grad_norm": 0.09700113534927368,
      "learning_rate": 6.111613166696767e-07,
      "loss": 0.001,
      "step": 267200
    },
    {
      "epoch": 1.7138019388582992,
      "grad_norm": 0.22153259813785553,
      "learning_rate": 6.108931978803306e-07,
      "loss": 0.0022,
      "step": 267210
    },
    {
      "epoch": 1.7138660757520854,
      "grad_norm": 0.011544063687324524,
      "learning_rate": 6.106251340896285e-07,
      "loss": 0.0016,
      "step": 267220
    },
    {
      "epoch": 1.7139302126458713,
      "grad_norm": 0.04484653100371361,
      "learning_rate": 6.103571253009316e-07,
      "loss": 0.0007,
      "step": 267230
    },
    {
      "epoch": 1.7139943495396575,
      "grad_norm": 0.15743902325630188,
      "learning_rate": 6.100891715175972e-07,
      "loss": 0.0011,
      "step": 267240
    },
    {
      "epoch": 1.7140584864334434,
      "grad_norm": 0.034544993191957474,
      "learning_rate": 6.098212727429825e-07,
      "loss": 0.0007,
      "step": 267250
    },
    {
      "epoch": 1.7141226233272295,
      "grad_norm": 0.058348409831523895,
      "learning_rate": 6.095534289804444e-07,
      "loss": 0.0016,
      "step": 267260
    },
    {
      "epoch": 1.7141867602210157,
      "grad_norm": 0.10489243268966675,
      "learning_rate": 6.092856402333403e-07,
      "loss": 0.001,
      "step": 267270
    },
    {
      "epoch": 1.7142508971148018,
      "grad_norm": 0.02318517304956913,
      "learning_rate": 6.090179065050256e-07,
      "loss": 0.0008,
      "step": 267280
    },
    {
      "epoch": 1.714315034008588,
      "grad_norm": 0.06452568620443344,
      "learning_rate": 6.087502277988533e-07,
      "loss": 0.001,
      "step": 267290
    },
    {
      "epoch": 1.7143791709023741,
      "grad_norm": 0.05898898467421532,
      "learning_rate": 6.084826041181796e-07,
      "loss": 0.0013,
      "step": 267300
    },
    {
      "epoch": 1.7144433077961603,
      "grad_norm": 0.046029724180698395,
      "learning_rate": 6.08215035466358e-07,
      "loss": 0.0013,
      "step": 267310
    },
    {
      "epoch": 1.7145074446899462,
      "grad_norm": 0.05649979040026665,
      "learning_rate": 6.079475218467407e-07,
      "loss": 0.0009,
      "step": 267320
    },
    {
      "epoch": 1.7145715815837324,
      "grad_norm": 0.09914553165435791,
      "learning_rate": 6.076800632626789e-07,
      "loss": 0.0007,
      "step": 267330
    },
    {
      "epoch": 1.7146357184775183,
      "grad_norm": 0.1583593189716339,
      "learning_rate": 6.074126597175251e-07,
      "loss": 0.001,
      "step": 267340
    },
    {
      "epoch": 1.7146998553713044,
      "grad_norm": 0.00976317748427391,
      "learning_rate": 6.071453112146309e-07,
      "loss": 0.0004,
      "step": 267350
    },
    {
      "epoch": 1.7147639922650906,
      "grad_norm": 0.0697932094335556,
      "learning_rate": 6.068780177573458e-07,
      "loss": 0.0007,
      "step": 267360
    },
    {
      "epoch": 1.7148281291588767,
      "grad_norm": 0.0736718475818634,
      "learning_rate": 6.066107793490183e-07,
      "loss": 0.0013,
      "step": 267370
    },
    {
      "epoch": 1.7148922660526629,
      "grad_norm": 0.037221480160951614,
      "learning_rate": 6.063435959929969e-07,
      "loss": 0.0011,
      "step": 267380
    },
    {
      "epoch": 1.714956402946449,
      "grad_norm": 0.0539449043571949,
      "learning_rate": 6.060764676926317e-07,
      "loss": 0.0009,
      "step": 267390
    },
    {
      "epoch": 1.715020539840235,
      "grad_norm": 0.10681122541427612,
      "learning_rate": 6.058093944512683e-07,
      "loss": 0.0021,
      "step": 267400
    },
    {
      "epoch": 1.715084676734021,
      "grad_norm": 0.06330878287553787,
      "learning_rate": 6.055423762722523e-07,
      "loss": 0.0023,
      "step": 267410
    },
    {
      "epoch": 1.7151488136278072,
      "grad_norm": 0.0955391600728035,
      "learning_rate": 6.052754131589328e-07,
      "loss": 0.0008,
      "step": 267420
    },
    {
      "epoch": 1.7152129505215932,
      "grad_norm": 0.20529933273792267,
      "learning_rate": 6.050085051146525e-07,
      "loss": 0.0021,
      "step": 267430
    },
    {
      "epoch": 1.7152770874153793,
      "grad_norm": 0.08644850552082062,
      "learning_rate": 6.047416521427574e-07,
      "loss": 0.0009,
      "step": 267440
    },
    {
      "epoch": 1.7153412243091655,
      "grad_norm": 0.020164771005511284,
      "learning_rate": 6.044748542465889e-07,
      "loss": 0.004,
      "step": 267450
    },
    {
      "epoch": 1.7154053612029516,
      "grad_norm": 0.061093926429748535,
      "learning_rate": 6.042081114294934e-07,
      "loss": 0.0007,
      "step": 267460
    },
    {
      "epoch": 1.7154694980967378,
      "grad_norm": 0.052653685212135315,
      "learning_rate": 6.039414236948116e-07,
      "loss": 0.0014,
      "step": 267470
    },
    {
      "epoch": 1.715533634990524,
      "grad_norm": 0.13485048711299896,
      "learning_rate": 6.036747910458856e-07,
      "loss": 0.0008,
      "step": 267480
    },
    {
      "epoch": 1.7155977718843098,
      "grad_norm": 0.11048126965761185,
      "learning_rate": 6.034082134860558e-07,
      "loss": 0.0018,
      "step": 267490
    },
    {
      "epoch": 1.715661908778096,
      "grad_norm": 0.19369938969612122,
      "learning_rate": 6.031416910186644e-07,
      "loss": 0.001,
      "step": 267500
    },
    {
      "epoch": 1.715726045671882,
      "grad_norm": 0.026004096493124962,
      "learning_rate": 6.028752236470492e-07,
      "loss": 0.0007,
      "step": 267510
    },
    {
      "epoch": 1.715790182565668,
      "grad_norm": 0.1416434645652771,
      "learning_rate": 6.026088113745509e-07,
      "loss": 0.0009,
      "step": 267520
    },
    {
      "epoch": 1.7158543194594542,
      "grad_norm": 0.02691793628036976,
      "learning_rate": 6.023424542045053e-07,
      "loss": 0.0009,
      "step": 267530
    },
    {
      "epoch": 1.7159184563532404,
      "grad_norm": 0.08257123082876205,
      "learning_rate": 6.020761521402529e-07,
      "loss": 0.0013,
      "step": 267540
    },
    {
      "epoch": 1.7159825932470265,
      "grad_norm": 0.0846862643957138,
      "learning_rate": 6.018099051851295e-07,
      "loss": 0.0005,
      "step": 267550
    },
    {
      "epoch": 1.7160467301408127,
      "grad_norm": 0.07328194379806519,
      "learning_rate": 6.015437133424706e-07,
      "loss": 0.001,
      "step": 267560
    },
    {
      "epoch": 1.7161108670345988,
      "grad_norm": 0.06155369058251381,
      "learning_rate": 6.012775766156131e-07,
      "loss": 0.0028,
      "step": 267570
    },
    {
      "epoch": 1.7161750039283847,
      "grad_norm": 0.07231937348842621,
      "learning_rate": 6.010114950078916e-07,
      "loss": 0.0016,
      "step": 267580
    },
    {
      "epoch": 1.7162391408221709,
      "grad_norm": 0.05894768238067627,
      "learning_rate": 6.0074546852264e-07,
      "loss": 0.0017,
      "step": 267590
    },
    {
      "epoch": 1.7163032777159568,
      "grad_norm": 0.003771646646782756,
      "learning_rate": 6.004794971631905e-07,
      "loss": 0.002,
      "step": 267600
    },
    {
      "epoch": 1.716367414609743,
      "grad_norm": 0.03629041835665703,
      "learning_rate": 6.002135809328785e-07,
      "loss": 0.0017,
      "step": 267610
    },
    {
      "epoch": 1.716431551503529,
      "grad_norm": 0.04036949202418327,
      "learning_rate": 5.999477198350346e-07,
      "loss": 0.0006,
      "step": 267620
    },
    {
      "epoch": 1.7164956883973153,
      "grad_norm": 0.06391513347625732,
      "learning_rate": 5.996819138729904e-07,
      "loss": 0.0015,
      "step": 267630
    },
    {
      "epoch": 1.7165598252911014,
      "grad_norm": 0.03568914532661438,
      "learning_rate": 5.994161630500755e-07,
      "loss": 0.0014,
      "step": 267640
    },
    {
      "epoch": 1.7166239621848876,
      "grad_norm": 0.08988387137651443,
      "learning_rate": 5.991504673696224e-07,
      "loss": 0.0013,
      "step": 267650
    },
    {
      "epoch": 1.7166880990786735,
      "grad_norm": 0.04272187128663063,
      "learning_rate": 5.988848268349589e-07,
      "loss": 0.0015,
      "step": 267660
    },
    {
      "epoch": 1.7167522359724596,
      "grad_norm": 0.11087754368782043,
      "learning_rate": 5.986192414494146e-07,
      "loss": 0.0009,
      "step": 267670
    },
    {
      "epoch": 1.7168163728662456,
      "grad_norm": 0.11257560551166534,
      "learning_rate": 5.983537112163151e-07,
      "loss": 0.0028,
      "step": 267680
    },
    {
      "epoch": 1.7168805097600317,
      "grad_norm": 0.12729914486408234,
      "learning_rate": 5.980882361389906e-07,
      "loss": 0.0004,
      "step": 267690
    },
    {
      "epoch": 1.7169446466538179,
      "grad_norm": 0.09516214579343796,
      "learning_rate": 5.978228162207666e-07,
      "loss": 0.0033,
      "step": 267700
    },
    {
      "epoch": 1.717008783547604,
      "grad_norm": 0.0547090545296669,
      "learning_rate": 5.975574514649679e-07,
      "loss": 0.0009,
      "step": 267710
    },
    {
      "epoch": 1.7170729204413901,
      "grad_norm": 0.20721198618412018,
      "learning_rate": 5.972921418749222e-07,
      "loss": 0.0018,
      "step": 267720
    },
    {
      "epoch": 1.7171370573351763,
      "grad_norm": 0.16879446804523468,
      "learning_rate": 5.97026887453952e-07,
      "loss": 0.0015,
      "step": 267730
    },
    {
      "epoch": 1.7172011942289624,
      "grad_norm": 0.02988448366522789,
      "learning_rate": 5.967616882053818e-07,
      "loss": 0.0008,
      "step": 267740
    },
    {
      "epoch": 1.7172653311227484,
      "grad_norm": 0.033318206667900085,
      "learning_rate": 5.964965441325332e-07,
      "loss": 0.0038,
      "step": 267750
    },
    {
      "epoch": 1.7173294680165345,
      "grad_norm": 0.07636163383722305,
      "learning_rate": 5.962314552387317e-07,
      "loss": 0.0009,
      "step": 267760
    },
    {
      "epoch": 1.7173936049103204,
      "grad_norm": 0.04419083893299103,
      "learning_rate": 5.95966421527297e-07,
      "loss": 0.0006,
      "step": 267770
    },
    {
      "epoch": 1.7174577418041066,
      "grad_norm": 0.009929543361067772,
      "learning_rate": 5.957014430015506e-07,
      "loss": 0.0008,
      "step": 267780
    },
    {
      "epoch": 1.7175218786978927,
      "grad_norm": 0.07677434384822845,
      "learning_rate": 5.954365196648116e-07,
      "loss": 0.0091,
      "step": 267790
    },
    {
      "epoch": 1.717586015591679,
      "grad_norm": 0.042411331087350845,
      "learning_rate": 5.95171651520402e-07,
      "loss": 0.0013,
      "step": 267800
    },
    {
      "epoch": 1.717650152485465,
      "grad_norm": 0.036884430795907974,
      "learning_rate": 5.9490683857164e-07,
      "loss": 0.0022,
      "step": 267810
    },
    {
      "epoch": 1.7177142893792512,
      "grad_norm": 0.19136400520801544,
      "learning_rate": 5.946420808218434e-07,
      "loss": 0.0015,
      "step": 267820
    },
    {
      "epoch": 1.7177784262730371,
      "grad_norm": 0.04579498991370201,
      "learning_rate": 5.94377378274329e-07,
      "loss": 0.0008,
      "step": 267830
    },
    {
      "epoch": 1.7178425631668233,
      "grad_norm": 0.020937394350767136,
      "learning_rate": 5.941127309324157e-07,
      "loss": 0.0007,
      "step": 267840
    },
    {
      "epoch": 1.7179067000606094,
      "grad_norm": 0.06400826573371887,
      "learning_rate": 5.938481387994189e-07,
      "loss": 0.001,
      "step": 267850
    },
    {
      "epoch": 1.7179708369543953,
      "grad_norm": 0.23463304340839386,
      "learning_rate": 5.935836018786523e-07,
      "loss": 0.002,
      "step": 267860
    },
    {
      "epoch": 1.7180349738481815,
      "grad_norm": 0.048746898770332336,
      "learning_rate": 5.933191201734339e-07,
      "loss": 0.0003,
      "step": 267870
    },
    {
      "epoch": 1.7180991107419676,
      "grad_norm": 0.1165740042924881,
      "learning_rate": 5.930546936870762e-07,
      "loss": 0.0009,
      "step": 267880
    },
    {
      "epoch": 1.7181632476357538,
      "grad_norm": 0.06149594485759735,
      "learning_rate": 5.927903224228932e-07,
      "loss": 0.0016,
      "step": 267890
    },
    {
      "epoch": 1.71822738452954,
      "grad_norm": 0.09030690044164658,
      "learning_rate": 5.925260063841959e-07,
      "loss": 0.0011,
      "step": 267900
    },
    {
      "epoch": 1.718291521423326,
      "grad_norm": 0.07678499817848206,
      "learning_rate": 5.922617455742985e-07,
      "loss": 0.0011,
      "step": 267910
    },
    {
      "epoch": 1.718355658317112,
      "grad_norm": 0.18253855407238007,
      "learning_rate": 5.919975399965117e-07,
      "loss": 0.0009,
      "step": 267920
    },
    {
      "epoch": 1.7184197952108982,
      "grad_norm": 0.023488642647862434,
      "learning_rate": 5.917333896541466e-07,
      "loss": 0.001,
      "step": 267930
    },
    {
      "epoch": 1.718483932104684,
      "grad_norm": 0.05363985523581505,
      "learning_rate": 5.914692945505112e-07,
      "loss": 0.0006,
      "step": 267940
    },
    {
      "epoch": 1.7185480689984702,
      "grad_norm": 0.03072049282491207,
      "learning_rate": 5.912052546889175e-07,
      "loss": 0.0012,
      "step": 267950
    },
    {
      "epoch": 1.7186122058922564,
      "grad_norm": 0.12611393630504608,
      "learning_rate": 5.909412700726725e-07,
      "loss": 0.0013,
      "step": 267960
    },
    {
      "epoch": 1.7186763427860425,
      "grad_norm": 0.04329914599657059,
      "learning_rate": 5.906773407050847e-07,
      "loss": 0.0017,
      "step": 267970
    },
    {
      "epoch": 1.7187404796798287,
      "grad_norm": 0.05544757843017578,
      "learning_rate": 5.904134665894601e-07,
      "loss": 0.0013,
      "step": 267980
    },
    {
      "epoch": 1.7188046165736148,
      "grad_norm": 0.0072180950082838535,
      "learning_rate": 5.901496477291069e-07,
      "loss": 0.0008,
      "step": 267990
    },
    {
      "epoch": 1.718868753467401,
      "grad_norm": 0.055339131504297256,
      "learning_rate": 5.898858841273309e-07,
      "loss": 0.0004,
      "step": 268000
    },
    {
      "epoch": 1.718932890361187,
      "grad_norm": 0.05021030455827713,
      "learning_rate": 5.896221757874354e-07,
      "loss": 0.0012,
      "step": 268010
    },
    {
      "epoch": 1.718997027254973,
      "grad_norm": 0.20326822996139526,
      "learning_rate": 5.893585227127269e-07,
      "loss": 0.0009,
      "step": 268020
    },
    {
      "epoch": 1.719061164148759,
      "grad_norm": 0.0853961855173111,
      "learning_rate": 5.890949249065086e-07,
      "loss": 0.0007,
      "step": 268030
    },
    {
      "epoch": 1.7191253010425451,
      "grad_norm": 0.06888889521360397,
      "learning_rate": 5.888313823720831e-07,
      "loss": 0.0015,
      "step": 268040
    },
    {
      "epoch": 1.7191894379363313,
      "grad_norm": 0.11359435319900513,
      "learning_rate": 5.885678951127521e-07,
      "loss": 0.0007,
      "step": 268050
    },
    {
      "epoch": 1.7192535748301174,
      "grad_norm": 0.06066417321562767,
      "learning_rate": 5.883044631318191e-07,
      "loss": 0.0015,
      "step": 268060
    },
    {
      "epoch": 1.7193177117239036,
      "grad_norm": 0.034054309129714966,
      "learning_rate": 5.880410864325847e-07,
      "loss": 0.0017,
      "step": 268070
    },
    {
      "epoch": 1.7193818486176897,
      "grad_norm": 0.04914059489965439,
      "learning_rate": 5.87777765018348e-07,
      "loss": 0.0011,
      "step": 268080
    },
    {
      "epoch": 1.7194459855114756,
      "grad_norm": 0.0036719120107591152,
      "learning_rate": 5.875144988924087e-07,
      "loss": 0.001,
      "step": 268090
    },
    {
      "epoch": 1.7195101224052618,
      "grad_norm": 0.023349685594439507,
      "learning_rate": 5.872512880580661e-07,
      "loss": 0.0008,
      "step": 268100
    },
    {
      "epoch": 1.7195742592990477,
      "grad_norm": 0.03897149860858917,
      "learning_rate": 5.869881325186194e-07,
      "loss": 0.0011,
      "step": 268110
    },
    {
      "epoch": 1.7196383961928339,
      "grad_norm": 0.07962550222873688,
      "learning_rate": 5.867250322773655e-07,
      "loss": 0.0008,
      "step": 268120
    },
    {
      "epoch": 1.71970253308662,
      "grad_norm": 0.055032357573509216,
      "learning_rate": 5.864619873376004e-07,
      "loss": 0.0006,
      "step": 268130
    },
    {
      "epoch": 1.7197666699804062,
      "grad_norm": 0.06639142334461212,
      "learning_rate": 5.861989977026211e-07,
      "loss": 0.0008,
      "step": 268140
    },
    {
      "epoch": 1.7198308068741923,
      "grad_norm": 0.006149010267108679,
      "learning_rate": 5.859360633757238e-07,
      "loss": 0.0017,
      "step": 268150
    },
    {
      "epoch": 1.7198949437679785,
      "grad_norm": 0.0026477861683815718,
      "learning_rate": 5.856731843602015e-07,
      "loss": 0.0007,
      "step": 268160
    },
    {
      "epoch": 1.7199590806617646,
      "grad_norm": 0.15994895994663239,
      "learning_rate": 5.854103606593487e-07,
      "loss": 0.0016,
      "step": 268170
    },
    {
      "epoch": 1.7200232175555505,
      "grad_norm": 0.05148687586188316,
      "learning_rate": 5.851475922764599e-07,
      "loss": 0.0011,
      "step": 268180
    },
    {
      "epoch": 1.7200873544493367,
      "grad_norm": 0.03837016224861145,
      "learning_rate": 5.84884879214827e-07,
      "loss": 0.0008,
      "step": 268190
    },
    {
      "epoch": 1.7201514913431226,
      "grad_norm": 0.14075350761413574,
      "learning_rate": 5.84622221477742e-07,
      "loss": 0.0011,
      "step": 268200
    },
    {
      "epoch": 1.7202156282369088,
      "grad_norm": 0.15034602582454681,
      "learning_rate": 5.843596190684952e-07,
      "loss": 0.0012,
      "step": 268210
    },
    {
      "epoch": 1.720279765130695,
      "grad_norm": 0.03877944499254227,
      "learning_rate": 5.840970719903788e-07,
      "loss": 0.0013,
      "step": 268220
    },
    {
      "epoch": 1.720343902024481,
      "grad_norm": 0.02334008365869522,
      "learning_rate": 5.838345802466827e-07,
      "loss": 0.001,
      "step": 268230
    },
    {
      "epoch": 1.7204080389182672,
      "grad_norm": 0.11830996721982956,
      "learning_rate": 5.835721438406955e-07,
      "loss": 0.0009,
      "step": 268240
    },
    {
      "epoch": 1.7204721758120534,
      "grad_norm": 0.03914085403084755,
      "learning_rate": 5.833097627757039e-07,
      "loss": 0.0021,
      "step": 268250
    },
    {
      "epoch": 1.7205363127058393,
      "grad_norm": 0.11923909187316895,
      "learning_rate": 5.830474370549993e-07,
      "loss": 0.001,
      "step": 268260
    },
    {
      "epoch": 1.7206004495996254,
      "grad_norm": 0.03597554564476013,
      "learning_rate": 5.827851666818668e-07,
      "loss": 0.0013,
      "step": 268270
    },
    {
      "epoch": 1.7206645864934116,
      "grad_norm": 0.019106004387140274,
      "learning_rate": 5.82522951659592e-07,
      "loss": 0.0005,
      "step": 268280
    },
    {
      "epoch": 1.7207287233871975,
      "grad_norm": 0.054433081299066544,
      "learning_rate": 5.822607919914631e-07,
      "loss": 0.0012,
      "step": 268290
    },
    {
      "epoch": 1.7207928602809837,
      "grad_norm": 0.11670947819948196,
      "learning_rate": 5.819986876807638e-07,
      "loss": 0.0014,
      "step": 268300
    },
    {
      "epoch": 1.7208569971747698,
      "grad_norm": 0.001161582418717444,
      "learning_rate": 5.817366387307782e-07,
      "loss": 0.0014,
      "step": 268310
    },
    {
      "epoch": 1.720921134068556,
      "grad_norm": 0.05353518947958946,
      "learning_rate": 5.814746451447889e-07,
      "loss": 0.0007,
      "step": 268320
    },
    {
      "epoch": 1.720985270962342,
      "grad_norm": 0.0915118157863617,
      "learning_rate": 5.812127069260814e-07,
      "loss": 0.001,
      "step": 268330
    },
    {
      "epoch": 1.7210494078561283,
      "grad_norm": 0.03967313468456268,
      "learning_rate": 5.809508240779371e-07,
      "loss": 0.0019,
      "step": 268340
    },
    {
      "epoch": 1.7211135447499142,
      "grad_norm": 0.010144086554646492,
      "learning_rate": 5.806889966036372e-07,
      "loss": 0.0008,
      "step": 268350
    },
    {
      "epoch": 1.7211776816437003,
      "grad_norm": 0.0049812509678304195,
      "learning_rate": 5.804272245064613e-07,
      "loss": 0.0009,
      "step": 268360
    },
    {
      "epoch": 1.7212418185374863,
      "grad_norm": 0.004335071891546249,
      "learning_rate": 5.801655077896917e-07,
      "loss": 0.0007,
      "step": 268370
    },
    {
      "epoch": 1.7213059554312724,
      "grad_norm": 0.07853275537490845,
      "learning_rate": 5.799038464566075e-07,
      "loss": 0.0012,
      "step": 268380
    },
    {
      "epoch": 1.7213700923250586,
      "grad_norm": 0.0015121111646294594,
      "learning_rate": 5.796422405104868e-07,
      "loss": 0.0021,
      "step": 268390
    },
    {
      "epoch": 1.7214342292188447,
      "grad_norm": 0.022543083876371384,
      "learning_rate": 5.793806899546072e-07,
      "loss": 0.0008,
      "step": 268400
    },
    {
      "epoch": 1.7214983661126309,
      "grad_norm": 0.07074515521526337,
      "learning_rate": 5.791191947922481e-07,
      "loss": 0.0017,
      "step": 268410
    },
    {
      "epoch": 1.721562503006417,
      "grad_norm": 0.07344797253608704,
      "learning_rate": 5.788577550266844e-07,
      "loss": 0.0006,
      "step": 268420
    },
    {
      "epoch": 1.7216266399002031,
      "grad_norm": 0.10367267578840256,
      "learning_rate": 5.785963706611925e-07,
      "loss": 0.0008,
      "step": 268430
    },
    {
      "epoch": 1.721690776793989,
      "grad_norm": 0.03328879177570343,
      "learning_rate": 5.783350416990485e-07,
      "loss": 0.0008,
      "step": 268440
    },
    {
      "epoch": 1.7217549136877752,
      "grad_norm": 0.07013612240552902,
      "learning_rate": 5.780737681435267e-07,
      "loss": 0.0007,
      "step": 268450
    },
    {
      "epoch": 1.7218190505815612,
      "grad_norm": 0.027138646692037582,
      "learning_rate": 5.778125499979009e-07,
      "loss": 0.0008,
      "step": 268460
    },
    {
      "epoch": 1.7218831874753473,
      "grad_norm": 0.1326294243335724,
      "learning_rate": 5.775513872654436e-07,
      "loss": 0.0015,
      "step": 268470
    },
    {
      "epoch": 1.7219473243691334,
      "grad_norm": 0.015880443155765533,
      "learning_rate": 5.77290279949429e-07,
      "loss": 0.0016,
      "step": 268480
    },
    {
      "epoch": 1.7220114612629196,
      "grad_norm": 0.14516927301883698,
      "learning_rate": 5.770292280531276e-07,
      "loss": 0.0017,
      "step": 268490
    },
    {
      "epoch": 1.7220755981567057,
      "grad_norm": 0.0028718379326164722,
      "learning_rate": 5.767682315798117e-07,
      "loss": 0.0006,
      "step": 268500
    },
    {
      "epoch": 1.722139735050492,
      "grad_norm": 0.041969895362854004,
      "learning_rate": 5.765072905327496e-07,
      "loss": 0.0025,
      "step": 268510
    },
    {
      "epoch": 1.7222038719442778,
      "grad_norm": 0.06452200561761856,
      "learning_rate": 5.762464049152139e-07,
      "loss": 0.0014,
      "step": 268520
    },
    {
      "epoch": 1.722268008838064,
      "grad_norm": 0.1101684421300888,
      "learning_rate": 5.759855747304722e-07,
      "loss": 0.0008,
      "step": 268530
    },
    {
      "epoch": 1.72233214573185,
      "grad_norm": 0.15614256262779236,
      "learning_rate": 5.757247999817917e-07,
      "loss": 0.001,
      "step": 268540
    },
    {
      "epoch": 1.722396282625636,
      "grad_norm": 0.0508844330906868,
      "learning_rate": 5.754640806724426e-07,
      "loss": 0.001,
      "step": 268550
    },
    {
      "epoch": 1.7224604195194222,
      "grad_norm": 0.017746128141880035,
      "learning_rate": 5.752034168056908e-07,
      "loss": 0.0008,
      "step": 268560
    },
    {
      "epoch": 1.7225245564132083,
      "grad_norm": 0.042515430599451065,
      "learning_rate": 5.74942808384803e-07,
      "loss": 0.0006,
      "step": 268570
    },
    {
      "epoch": 1.7225886933069945,
      "grad_norm": 0.046711161732673645,
      "learning_rate": 5.746822554130427e-07,
      "loss": 0.0013,
      "step": 268580
    },
    {
      "epoch": 1.7226528302007806,
      "grad_norm": 0.04456254094839096,
      "learning_rate": 5.744217578936778e-07,
      "loss": 0.001,
      "step": 268590
    },
    {
      "epoch": 1.7227169670945668,
      "grad_norm": 0.020523108541965485,
      "learning_rate": 5.741613158299714e-07,
      "loss": 0.0008,
      "step": 268600
    },
    {
      "epoch": 1.7227811039883527,
      "grad_norm": 0.023720821365714073,
      "learning_rate": 5.739009292251863e-07,
      "loss": 0.0014,
      "step": 268610
    },
    {
      "epoch": 1.7228452408821389,
      "grad_norm": 0.15102943778038025,
      "learning_rate": 5.736405980825849e-07,
      "loss": 0.0011,
      "step": 268620
    },
    {
      "epoch": 1.7229093777759248,
      "grad_norm": 0.004818478133529425,
      "learning_rate": 5.733803224054319e-07,
      "loss": 0.0008,
      "step": 268630
    },
    {
      "epoch": 1.722973514669711,
      "grad_norm": 0.0735711082816124,
      "learning_rate": 5.731201021969868e-07,
      "loss": 0.0017,
      "step": 268640
    },
    {
      "epoch": 1.723037651563497,
      "grad_norm": 0.21588671207427979,
      "learning_rate": 5.728599374605104e-07,
      "loss": 0.001,
      "step": 268650
    },
    {
      "epoch": 1.7231017884572832,
      "grad_norm": 0.1779114305973053,
      "learning_rate": 5.725998281992618e-07,
      "loss": 0.0026,
      "step": 268660
    },
    {
      "epoch": 1.7231659253510694,
      "grad_norm": 0.07624183595180511,
      "learning_rate": 5.723397744165032e-07,
      "loss": 0.0013,
      "step": 268670
    },
    {
      "epoch": 1.7232300622448555,
      "grad_norm": 0.07083241641521454,
      "learning_rate": 5.720797761154911e-07,
      "loss": 0.0009,
      "step": 268680
    },
    {
      "epoch": 1.7232941991386417,
      "grad_norm": 0.0490984283387661,
      "learning_rate": 5.718198332994835e-07,
      "loss": 0.0019,
      "step": 268690
    },
    {
      "epoch": 1.7233583360324276,
      "grad_norm": 0.20232270658016205,
      "learning_rate": 5.715599459717386e-07,
      "loss": 0.0008,
      "step": 268700
    },
    {
      "epoch": 1.7234224729262138,
      "grad_norm": 0.00593222351744771,
      "learning_rate": 5.713001141355129e-07,
      "loss": 0.0012,
      "step": 268710
    },
    {
      "epoch": 1.7234866098199997,
      "grad_norm": 0.04960239306092262,
      "learning_rate": 5.710403377940621e-07,
      "loss": 0.0012,
      "step": 268720
    },
    {
      "epoch": 1.7235507467137858,
      "grad_norm": 0.07858838140964508,
      "learning_rate": 5.707806169506397e-07,
      "loss": 0.0008,
      "step": 268730
    },
    {
      "epoch": 1.723614883607572,
      "grad_norm": 0.1905651092529297,
      "learning_rate": 5.70520951608503e-07,
      "loss": 0.0011,
      "step": 268740
    },
    {
      "epoch": 1.7236790205013581,
      "grad_norm": 0.043486468493938446,
      "learning_rate": 5.702613417709046e-07,
      "loss": 0.0007,
      "step": 268750
    },
    {
      "epoch": 1.7237431573951443,
      "grad_norm": 0.11901059001684189,
      "learning_rate": 5.700017874410973e-07,
      "loss": 0.0013,
      "step": 268760
    },
    {
      "epoch": 1.7238072942889304,
      "grad_norm": 0.0030484062153846025,
      "learning_rate": 5.697422886223325e-07,
      "loss": 0.0008,
      "step": 268770
    },
    {
      "epoch": 1.7238714311827164,
      "grad_norm": 0.1276695430278778,
      "learning_rate": 5.694828453178641e-07,
      "loss": 0.0011,
      "step": 268780
    },
    {
      "epoch": 1.7239355680765025,
      "grad_norm": 0.0690717101097107,
      "learning_rate": 5.692234575309424e-07,
      "loss": 0.0012,
      "step": 268790
    },
    {
      "epoch": 1.7239997049702884,
      "grad_norm": 0.014858945272862911,
      "learning_rate": 5.689641252648171e-07,
      "loss": 0.0006,
      "step": 268800
    },
    {
      "epoch": 1.7240638418640746,
      "grad_norm": 0.07536854594945908,
      "learning_rate": 5.687048485227376e-07,
      "loss": 0.001,
      "step": 268810
    },
    {
      "epoch": 1.7241279787578607,
      "grad_norm": 0.06748121231794357,
      "learning_rate": 5.684456273079536e-07,
      "loss": 0.0017,
      "step": 268820
    },
    {
      "epoch": 1.7241921156516469,
      "grad_norm": 0.060129065066576004,
      "learning_rate": 5.681864616237137e-07,
      "loss": 0.0009,
      "step": 268830
    },
    {
      "epoch": 1.724256252545433,
      "grad_norm": 0.09345393627882004,
      "learning_rate": 5.679273514732636e-07,
      "loss": 0.0021,
      "step": 268840
    },
    {
      "epoch": 1.7243203894392192,
      "grad_norm": 0.016838548704981804,
      "learning_rate": 5.676682968598523e-07,
      "loss": 0.0021,
      "step": 268850
    },
    {
      "epoch": 1.7243845263330053,
      "grad_norm": 0.006551853846758604,
      "learning_rate": 5.674092977867252e-07,
      "loss": 0.0008,
      "step": 268860
    },
    {
      "epoch": 1.7244486632267912,
      "grad_norm": 0.06902817636728287,
      "learning_rate": 5.67150354257126e-07,
      "loss": 0.0011,
      "step": 268870
    },
    {
      "epoch": 1.7245128001205774,
      "grad_norm": 0.07074497640132904,
      "learning_rate": 5.668914662743025e-07,
      "loss": 0.0009,
      "step": 268880
    },
    {
      "epoch": 1.7245769370143633,
      "grad_norm": 0.12114270031452179,
      "learning_rate": 5.666326338414962e-07,
      "loss": 0.0007,
      "step": 268890
    },
    {
      "epoch": 1.7246410739081495,
      "grad_norm": 0.06865867972373962,
      "learning_rate": 5.66373856961952e-07,
      "loss": 0.0008,
      "step": 268900
    },
    {
      "epoch": 1.7247052108019356,
      "grad_norm": 0.10277780890464783,
      "learning_rate": 5.661151356389127e-07,
      "loss": 0.0004,
      "step": 268910
    },
    {
      "epoch": 1.7247693476957218,
      "grad_norm": 0.02711491286754608,
      "learning_rate": 5.658564698756192e-07,
      "loss": 0.0009,
      "step": 268920
    },
    {
      "epoch": 1.724833484589508,
      "grad_norm": 0.04675956070423126,
      "learning_rate": 5.65597859675312e-07,
      "loss": 0.0015,
      "step": 268930
    },
    {
      "epoch": 1.724897621483294,
      "grad_norm": 0.17096774280071259,
      "learning_rate": 5.653393050412343e-07,
      "loss": 0.0016,
      "step": 268940
    },
    {
      "epoch": 1.72496175837708,
      "grad_norm": 0.12755827605724335,
      "learning_rate": 5.650808059766238e-07,
      "loss": 0.0023,
      "step": 268950
    },
    {
      "epoch": 1.7250258952708661,
      "grad_norm": 0.030916327610611916,
      "learning_rate": 5.648223624847199e-07,
      "loss": 0.0009,
      "step": 268960
    },
    {
      "epoch": 1.7250900321646523,
      "grad_norm": 0.1278943568468094,
      "learning_rate": 5.645639745687625e-07,
      "loss": 0.0011,
      "step": 268970
    },
    {
      "epoch": 1.7251541690584382,
      "grad_norm": 0.009564920328557491,
      "learning_rate": 5.643056422319887e-07,
      "loss": 0.0007,
      "step": 268980
    },
    {
      "epoch": 1.7252183059522244,
      "grad_norm": 0.07532178610563278,
      "learning_rate": 5.640473654776352e-07,
      "loss": 0.0017,
      "step": 268990
    },
    {
      "epoch": 1.7252824428460105,
      "grad_norm": 0.06400129944086075,
      "learning_rate": 5.637891443089372e-07,
      "loss": 0.0006,
      "step": 269000
    },
    {
      "epoch": 1.7253465797397967,
      "grad_norm": 0.0003619497874751687,
      "learning_rate": 5.635309787291332e-07,
      "loss": 0.001,
      "step": 269010
    },
    {
      "epoch": 1.7254107166335828,
      "grad_norm": 0.09920549392700195,
      "learning_rate": 5.632728687414562e-07,
      "loss": 0.0017,
      "step": 269020
    },
    {
      "epoch": 1.725474853527369,
      "grad_norm": 0.031354062259197235,
      "learning_rate": 5.630148143491415e-07,
      "loss": 0.0009,
      "step": 269030
    },
    {
      "epoch": 1.7255389904211549,
      "grad_norm": 0.05927729234099388,
      "learning_rate": 5.627568155554214e-07,
      "loss": 0.0028,
      "step": 269040
    },
    {
      "epoch": 1.725603127314941,
      "grad_norm": 0.0024785532150417566,
      "learning_rate": 5.6249887236353e-07,
      "loss": 0.0029,
      "step": 269050
    },
    {
      "epoch": 1.725667264208727,
      "grad_norm": 0.1858125627040863,
      "learning_rate": 5.622409847766997e-07,
      "loss": 0.0017,
      "step": 269060
    },
    {
      "epoch": 1.725731401102513,
      "grad_norm": 0.1651456207036972,
      "learning_rate": 5.619831527981612e-07,
      "loss": 0.001,
      "step": 269070
    },
    {
      "epoch": 1.7257955379962993,
      "grad_norm": 0.04127020016312599,
      "learning_rate": 5.617253764311447e-07,
      "loss": 0.0009,
      "step": 269080
    },
    {
      "epoch": 1.7258596748900854,
      "grad_norm": 0.01615344174206257,
      "learning_rate": 5.614676556788823e-07,
      "loss": 0.0008,
      "step": 269090
    },
    {
      "epoch": 1.7259238117838716,
      "grad_norm": 0.013099998235702515,
      "learning_rate": 5.612099905446022e-07,
      "loss": 0.001,
      "step": 269100
    },
    {
      "epoch": 1.7259879486776577,
      "grad_norm": 0.08057832717895508,
      "learning_rate": 5.609523810315326e-07,
      "loss": 0.0007,
      "step": 269110
    },
    {
      "epoch": 1.7260520855714439,
      "grad_norm": 0.06329803913831711,
      "learning_rate": 5.60694827142903e-07,
      "loss": 0.0011,
      "step": 269120
    },
    {
      "epoch": 1.7261162224652298,
      "grad_norm": 0.09694880247116089,
      "learning_rate": 5.604373288819398e-07,
      "loss": 0.001,
      "step": 269130
    },
    {
      "epoch": 1.726180359359016,
      "grad_norm": 0.21163076162338257,
      "learning_rate": 5.601798862518698e-07,
      "loss": 0.0021,
      "step": 269140
    },
    {
      "epoch": 1.7262444962528019,
      "grad_norm": 0.1984589546918869,
      "learning_rate": 5.599224992559177e-07,
      "loss": 0.0011,
      "step": 269150
    },
    {
      "epoch": 1.726308633146588,
      "grad_norm": 0.19185489416122437,
      "learning_rate": 5.596651678973114e-07,
      "loss": 0.0015,
      "step": 269160
    },
    {
      "epoch": 1.7263727700403741,
      "grad_norm": 0.20477338135242462,
      "learning_rate": 5.594078921792739e-07,
      "loss": 0.0021,
      "step": 269170
    },
    {
      "epoch": 1.7264369069341603,
      "grad_norm": 0.0421905592083931,
      "learning_rate": 5.591506721050294e-07,
      "loss": 0.0008,
      "step": 269180
    },
    {
      "epoch": 1.7265010438279464,
      "grad_norm": 0.09347651898860931,
      "learning_rate": 5.588935076777991e-07,
      "loss": 0.0025,
      "step": 269190
    },
    {
      "epoch": 1.7265651807217326,
      "grad_norm": 0.05665479600429535,
      "learning_rate": 5.586363989008087e-07,
      "loss": 0.001,
      "step": 269200
    },
    {
      "epoch": 1.7266293176155185,
      "grad_norm": 0.08920852839946747,
      "learning_rate": 5.583793457772785e-07,
      "loss": 0.0013,
      "step": 269210
    },
    {
      "epoch": 1.7266934545093047,
      "grad_norm": 0.02189306542277336,
      "learning_rate": 5.581223483104292e-07,
      "loss": 0.0013,
      "step": 269220
    },
    {
      "epoch": 1.7267575914030906,
      "grad_norm": 0.09904073178768158,
      "learning_rate": 5.578654065034806e-07,
      "loss": 0.001,
      "step": 269230
    },
    {
      "epoch": 1.7268217282968767,
      "grad_norm": 0.059837743639945984,
      "learning_rate": 5.576085203596537e-07,
      "loss": 0.001,
      "step": 269240
    },
    {
      "epoch": 1.726885865190663,
      "grad_norm": 0.09818316251039505,
      "learning_rate": 5.573516898821674e-07,
      "loss": 0.0009,
      "step": 269250
    },
    {
      "epoch": 1.726950002084449,
      "grad_norm": 0.12516026198863983,
      "learning_rate": 5.570949150742383e-07,
      "loss": 0.0014,
      "step": 269260
    },
    {
      "epoch": 1.7270141389782352,
      "grad_norm": 0.1935839205980301,
      "learning_rate": 5.568381959390868e-07,
      "loss": 0.0008,
      "step": 269270
    },
    {
      "epoch": 1.7270782758720213,
      "grad_norm": 0.05724336579442024,
      "learning_rate": 5.565815324799278e-07,
      "loss": 0.0021,
      "step": 269280
    },
    {
      "epoch": 1.7271424127658075,
      "grad_norm": 0.040515750646591187,
      "learning_rate": 5.563249246999774e-07,
      "loss": 0.001,
      "step": 269290
    },
    {
      "epoch": 1.7272065496595934,
      "grad_norm": 0.05509898066520691,
      "learning_rate": 5.560683726024512e-07,
      "loss": 0.0005,
      "step": 269300
    },
    {
      "epoch": 1.7272706865533796,
      "grad_norm": 0.01175855565816164,
      "learning_rate": 5.558118761905651e-07,
      "loss": 0.0006,
      "step": 269310
    },
    {
      "epoch": 1.7273348234471655,
      "grad_norm": 0.15045689046382904,
      "learning_rate": 5.555554354675325e-07,
      "loss": 0.0012,
      "step": 269320
    },
    {
      "epoch": 1.7273989603409516,
      "grad_norm": 0.029485128819942474,
      "learning_rate": 5.552990504365669e-07,
      "loss": 0.0018,
      "step": 269330
    },
    {
      "epoch": 1.7274630972347378,
      "grad_norm": 0.042644254863262177,
      "learning_rate": 5.550427211008791e-07,
      "loss": 0.0007,
      "step": 269340
    },
    {
      "epoch": 1.727527234128524,
      "grad_norm": 0.069488525390625,
      "learning_rate": 5.547864474636844e-07,
      "loss": 0.001,
      "step": 269350
    },
    {
      "epoch": 1.72759137102231,
      "grad_norm": 0.025307761505246162,
      "learning_rate": 5.545302295281924e-07,
      "loss": 0.0022,
      "step": 269360
    },
    {
      "epoch": 1.7276555079160962,
      "grad_norm": 0.1392243653535843,
      "learning_rate": 5.542740672976133e-07,
      "loss": 0.0011,
      "step": 269370
    },
    {
      "epoch": 1.7277196448098822,
      "grad_norm": 0.08107449114322662,
      "learning_rate": 5.540179607751566e-07,
      "loss": 0.0005,
      "step": 269380
    },
    {
      "epoch": 1.7277837817036683,
      "grad_norm": 0.18233659863471985,
      "learning_rate": 5.537619099640335e-07,
      "loss": 0.001,
      "step": 269390
    },
    {
      "epoch": 1.7278479185974545,
      "grad_norm": 0.014074004255235195,
      "learning_rate": 5.535059148674516e-07,
      "loss": 0.0006,
      "step": 269400
    },
    {
      "epoch": 1.7279120554912404,
      "grad_norm": 0.044057007879018784,
      "learning_rate": 5.532499754886173e-07,
      "loss": 0.0005,
      "step": 269410
    },
    {
      "epoch": 1.7279761923850265,
      "grad_norm": 0.12658320367336273,
      "learning_rate": 5.529940918307392e-07,
      "loss": 0.0016,
      "step": 269420
    },
    {
      "epoch": 1.7280403292788127,
      "grad_norm": 0.1030837744474411,
      "learning_rate": 5.527382638970241e-07,
      "loss": 0.0028,
      "step": 269430
    },
    {
      "epoch": 1.7281044661725988,
      "grad_norm": 0.07192330062389374,
      "learning_rate": 5.524824916906768e-07,
      "loss": 0.0012,
      "step": 269440
    },
    {
      "epoch": 1.728168603066385,
      "grad_norm": 0.11342022567987442,
      "learning_rate": 5.522267752149013e-07,
      "loss": 0.0013,
      "step": 269450
    },
    {
      "epoch": 1.7282327399601711,
      "grad_norm": 0.01136010605841875,
      "learning_rate": 5.51971114472904e-07,
      "loss": 0.0006,
      "step": 269460
    },
    {
      "epoch": 1.728296876853957,
      "grad_norm": 0.11182946711778641,
      "learning_rate": 5.517155094678877e-07,
      "loss": 0.0016,
      "step": 269470
    },
    {
      "epoch": 1.7283610137477432,
      "grad_norm": 0.16093896329402924,
      "learning_rate": 5.514599602030557e-07,
      "loss": 0.0021,
      "step": 269480
    },
    {
      "epoch": 1.7284251506415291,
      "grad_norm": 0.1454562246799469,
      "learning_rate": 5.51204466681608e-07,
      "loss": 0.0007,
      "step": 269490
    },
    {
      "epoch": 1.7284892875353153,
      "grad_norm": 0.17309823632240295,
      "learning_rate": 5.509490289067493e-07,
      "loss": 0.0008,
      "step": 269500
    },
    {
      "epoch": 1.7285534244291014,
      "grad_norm": 0.025239739567041397,
      "learning_rate": 5.506936468816787e-07,
      "loss": 0.0013,
      "step": 269510
    },
    {
      "epoch": 1.7286175613228876,
      "grad_norm": 0.05094781517982483,
      "learning_rate": 5.504383206095971e-07,
      "loss": 0.001,
      "step": 269520
    },
    {
      "epoch": 1.7286816982166737,
      "grad_norm": 0.025469429790973663,
      "learning_rate": 5.501830500937016e-07,
      "loss": 0.0004,
      "step": 269530
    },
    {
      "epoch": 1.7287458351104599,
      "grad_norm": 0.030855637043714523,
      "learning_rate": 5.49927835337194e-07,
      "loss": 0.0004,
      "step": 269540
    },
    {
      "epoch": 1.728809972004246,
      "grad_norm": 0.05780933424830437,
      "learning_rate": 5.496726763432714e-07,
      "loss": 0.0012,
      "step": 269550
    },
    {
      "epoch": 1.728874108898032,
      "grad_norm": 0.08418560773134232,
      "learning_rate": 5.494175731151297e-07,
      "loss": 0.0011,
      "step": 269560
    },
    {
      "epoch": 1.728938245791818,
      "grad_norm": 0.09374580532312393,
      "learning_rate": 5.491625256559674e-07,
      "loss": 0.0006,
      "step": 269570
    },
    {
      "epoch": 1.729002382685604,
      "grad_norm": 0.12030810117721558,
      "learning_rate": 5.489075339689792e-07,
      "loss": 0.001,
      "step": 269580
    },
    {
      "epoch": 1.7290665195793902,
      "grad_norm": 0.07812446355819702,
      "learning_rate": 5.486525980573615e-07,
      "loss": 0.0006,
      "step": 269590
    },
    {
      "epoch": 1.7291306564731763,
      "grad_norm": 0.01384024415165186,
      "learning_rate": 5.483977179243066e-07,
      "loss": 0.0018,
      "step": 269600
    },
    {
      "epoch": 1.7291947933669625,
      "grad_norm": 0.054847002029418945,
      "learning_rate": 5.48142893573011e-07,
      "loss": 0.0018,
      "step": 269610
    },
    {
      "epoch": 1.7292589302607486,
      "grad_norm": 0.007031548768281937,
      "learning_rate": 5.478881250066665e-07,
      "loss": 0.0011,
      "step": 269620
    },
    {
      "epoch": 1.7293230671545348,
      "grad_norm": 0.20511378347873688,
      "learning_rate": 5.476334122284649e-07,
      "loss": 0.0017,
      "step": 269630
    },
    {
      "epoch": 1.7293872040483207,
      "grad_norm": 0.09294286370277405,
      "learning_rate": 5.473787552415994e-07,
      "loss": 0.0021,
      "step": 269640
    },
    {
      "epoch": 1.7294513409421068,
      "grad_norm": 0.1279413104057312,
      "learning_rate": 5.471241540492595e-07,
      "loss": 0.0021,
      "step": 269650
    },
    {
      "epoch": 1.7295154778358928,
      "grad_norm": 0.0042890217155218124,
      "learning_rate": 5.468696086546377e-07,
      "loss": 0.0012,
      "step": 269660
    },
    {
      "epoch": 1.729579614729679,
      "grad_norm": 0.06577875465154648,
      "learning_rate": 5.466151190609215e-07,
      "loss": 0.0008,
      "step": 269670
    },
    {
      "epoch": 1.729643751623465,
      "grad_norm": 0.08991734683513641,
      "learning_rate": 5.463606852713005e-07,
      "loss": 0.0015,
      "step": 269680
    },
    {
      "epoch": 1.7297078885172512,
      "grad_norm": 0.08856498450040817,
      "learning_rate": 5.461063072889633e-07,
      "loss": 0.001,
      "step": 269690
    },
    {
      "epoch": 1.7297720254110374,
      "grad_norm": 0.049498848617076874,
      "learning_rate": 5.458519851170979e-07,
      "loss": 0.0014,
      "step": 269700
    },
    {
      "epoch": 1.7298361623048235,
      "grad_norm": 0.02303713746368885,
      "learning_rate": 5.455977187588902e-07,
      "loss": 0.001,
      "step": 269710
    },
    {
      "epoch": 1.7299002991986097,
      "grad_norm": 0.04629380255937576,
      "learning_rate": 5.453435082175256e-07,
      "loss": 0.0006,
      "step": 269720
    },
    {
      "epoch": 1.7299644360923956,
      "grad_norm": 0.05294347181916237,
      "learning_rate": 5.450893534961915e-07,
      "loss": 0.0013,
      "step": 269730
    },
    {
      "epoch": 1.7300285729861817,
      "grad_norm": 0.2689642906188965,
      "learning_rate": 5.448352545980717e-07,
      "loss": 0.0014,
      "step": 269740
    },
    {
      "epoch": 1.7300927098799677,
      "grad_norm": 0.10071782022714615,
      "learning_rate": 5.4458121152635e-07,
      "loss": 0.0007,
      "step": 269750
    },
    {
      "epoch": 1.7301568467737538,
      "grad_norm": 0.23420055210590363,
      "learning_rate": 5.44327224284209e-07,
      "loss": 0.0018,
      "step": 269760
    },
    {
      "epoch": 1.73022098366754,
      "grad_norm": 0.11843283474445343,
      "learning_rate": 5.440732928748338e-07,
      "loss": 0.0006,
      "step": 269770
    },
    {
      "epoch": 1.730285120561326,
      "grad_norm": 0.05789335444569588,
      "learning_rate": 5.438194173014044e-07,
      "loss": 0.0011,
      "step": 269780
    },
    {
      "epoch": 1.7303492574551123,
      "grad_norm": 0.021529171615839005,
      "learning_rate": 5.435655975671029e-07,
      "loss": 0.0011,
      "step": 269790
    },
    {
      "epoch": 1.7304133943488984,
      "grad_norm": 0.02813744731247425,
      "learning_rate": 5.433118336751081e-07,
      "loss": 0.0009,
      "step": 269800
    },
    {
      "epoch": 1.7304775312426843,
      "grad_norm": 0.1608094424009323,
      "learning_rate": 5.430581256286022e-07,
      "loss": 0.0011,
      "step": 269810
    },
    {
      "epoch": 1.7305416681364705,
      "grad_norm": 0.1470644772052765,
      "learning_rate": 5.42804473430763e-07,
      "loss": 0.0022,
      "step": 269820
    },
    {
      "epoch": 1.7306058050302566,
      "grad_norm": 0.029143664985895157,
      "learning_rate": 5.425508770847687e-07,
      "loss": 0.0009,
      "step": 269830
    },
    {
      "epoch": 1.7306699419240426,
      "grad_norm": 0.0821867510676384,
      "learning_rate": 5.422973365937983e-07,
      "loss": 0.001,
      "step": 269840
    },
    {
      "epoch": 1.7307340788178287,
      "grad_norm": 0.004571834579110146,
      "learning_rate": 5.420438519610282e-07,
      "loss": 0.0021,
      "step": 269850
    },
    {
      "epoch": 1.7307982157116149,
      "grad_norm": 0.04594600945711136,
      "learning_rate": 5.417904231896348e-07,
      "loss": 0.0009,
      "step": 269860
    },
    {
      "epoch": 1.730862352605401,
      "grad_norm": 0.1579095721244812,
      "learning_rate": 5.415370502827927e-07,
      "loss": 0.0025,
      "step": 269870
    },
    {
      "epoch": 1.7309264894991871,
      "grad_norm": 0.10846943408250809,
      "learning_rate": 5.412837332436788e-07,
      "loss": 0.0014,
      "step": 269880
    },
    {
      "epoch": 1.7309906263929733,
      "grad_norm": 0.19232290983200073,
      "learning_rate": 5.410304720754661e-07,
      "loss": 0.0014,
      "step": 269890
    },
    {
      "epoch": 1.7310547632867592,
      "grad_norm": 0.08476459980010986,
      "learning_rate": 5.407772667813288e-07,
      "loss": 0.0013,
      "step": 269900
    },
    {
      "epoch": 1.7311189001805454,
      "grad_norm": 0.04643384739756584,
      "learning_rate": 5.405241173644377e-07,
      "loss": 0.0015,
      "step": 269910
    },
    {
      "epoch": 1.7311830370743313,
      "grad_norm": 0.10197514295578003,
      "learning_rate": 5.402710238279679e-07,
      "loss": 0.0008,
      "step": 269920
    },
    {
      "epoch": 1.7312471739681174,
      "grad_norm": 0.12310665845870972,
      "learning_rate": 5.400179861750898e-07,
      "loss": 0.0014,
      "step": 269930
    },
    {
      "epoch": 1.7313113108619036,
      "grad_norm": 0.11327383667230606,
      "learning_rate": 5.39765004408973e-07,
      "loss": 0.0008,
      "step": 269940
    },
    {
      "epoch": 1.7313754477556897,
      "grad_norm": 0.11353304237127304,
      "learning_rate": 5.395120785327878e-07,
      "loss": 0.0021,
      "step": 269950
    },
    {
      "epoch": 1.731439584649476,
      "grad_norm": 0.020031224936246872,
      "learning_rate": 5.392592085497055e-07,
      "loss": 0.0008,
      "step": 269960
    },
    {
      "epoch": 1.731503721543262,
      "grad_norm": 0.04257189854979515,
      "learning_rate": 5.390063944628932e-07,
      "loss": 0.0009,
      "step": 269970
    },
    {
      "epoch": 1.7315678584370482,
      "grad_norm": 0.09973517805337906,
      "learning_rate": 5.387536362755175e-07,
      "loss": 0.0008,
      "step": 269980
    },
    {
      "epoch": 1.7316319953308341,
      "grad_norm": 0.05843477323651314,
      "learning_rate": 5.385009339907487e-07,
      "loss": 0.0009,
      "step": 269990
    },
    {
      "epoch": 1.7316961322246203,
      "grad_norm": 0.09853708744049072,
      "learning_rate": 5.382482876117511e-07,
      "loss": 0.0011,
      "step": 270000
    },
    {
      "epoch": 1.7317602691184062,
      "grad_norm": 0.05056915804743767,
      "learning_rate": 5.379956971416922e-07,
      "loss": 0.001,
      "step": 270010
    },
    {
      "epoch": 1.7318244060121923,
      "grad_norm": 0.15489324927330017,
      "learning_rate": 5.377431625837343e-07,
      "loss": 0.0012,
      "step": 270020
    },
    {
      "epoch": 1.7318885429059785,
      "grad_norm": 0.08503349870443344,
      "learning_rate": 5.374906839410455e-07,
      "loss": 0.0007,
      "step": 270030
    },
    {
      "epoch": 1.7319526797997646,
      "grad_norm": 0.13210877776145935,
      "learning_rate": 5.37238261216787e-07,
      "loss": 0.0013,
      "step": 270040
    },
    {
      "epoch": 1.7320168166935508,
      "grad_norm": 0.06503061950206757,
      "learning_rate": 5.369858944141226e-07,
      "loss": 0.001,
      "step": 270050
    },
    {
      "epoch": 1.732080953587337,
      "grad_norm": 0.038855258375406265,
      "learning_rate": 5.367335835362141e-07,
      "loss": 0.0012,
      "step": 270060
    },
    {
      "epoch": 1.7321450904811229,
      "grad_norm": 0.01080508716404438,
      "learning_rate": 5.364813285862242e-07,
      "loss": 0.0011,
      "step": 270070
    },
    {
      "epoch": 1.732209227374909,
      "grad_norm": 0.013003021478652954,
      "learning_rate": 5.362291295673139e-07,
      "loss": 0.0033,
      "step": 270080
    },
    {
      "epoch": 1.732273364268695,
      "grad_norm": 0.010630217380821705,
      "learning_rate": 5.35976986482642e-07,
      "loss": 0.0011,
      "step": 270090
    },
    {
      "epoch": 1.732337501162481,
      "grad_norm": 0.11707303673028946,
      "learning_rate": 5.357248993353686e-07,
      "loss": 0.0004,
      "step": 270100
    },
    {
      "epoch": 1.7324016380562672,
      "grad_norm": 0.0049722082912921906,
      "learning_rate": 5.35472868128653e-07,
      "loss": 0.0008,
      "step": 270110
    },
    {
      "epoch": 1.7324657749500534,
      "grad_norm": 0.0259892288595438,
      "learning_rate": 5.352208928656539e-07,
      "loss": 0.0006,
      "step": 270120
    },
    {
      "epoch": 1.7325299118438395,
      "grad_norm": 0.05452693626284599,
      "learning_rate": 5.349689735495267e-07,
      "loss": 0.0009,
      "step": 270130
    },
    {
      "epoch": 1.7325940487376257,
      "grad_norm": 0.012979741208255291,
      "learning_rate": 5.347171101834297e-07,
      "loss": 0.0021,
      "step": 270140
    },
    {
      "epoch": 1.7326581856314118,
      "grad_norm": 0.057723190635442734,
      "learning_rate": 5.344653027705193e-07,
      "loss": 0.001,
      "step": 270150
    },
    {
      "epoch": 1.7327223225251978,
      "grad_norm": 0.12237024307250977,
      "learning_rate": 5.342135513139501e-07,
      "loss": 0.0011,
      "step": 270160
    },
    {
      "epoch": 1.732786459418984,
      "grad_norm": 0.05576246604323387,
      "learning_rate": 5.339618558168757e-07,
      "loss": 0.0008,
      "step": 270170
    },
    {
      "epoch": 1.7328505963127698,
      "grad_norm": 0.004119261167943478,
      "learning_rate": 5.33710216282452e-07,
      "loss": 0.0025,
      "step": 270180
    },
    {
      "epoch": 1.732914733206556,
      "grad_norm": 0.0020740805193781853,
      "learning_rate": 5.334586327138313e-07,
      "loss": 0.001,
      "step": 270190
    },
    {
      "epoch": 1.7329788701003421,
      "grad_norm": 0.02012786827981472,
      "learning_rate": 5.332071051141663e-07,
      "loss": 0.0013,
      "step": 270200
    },
    {
      "epoch": 1.7330430069941283,
      "grad_norm": 0.01470683142542839,
      "learning_rate": 5.329556334866071e-07,
      "loss": 0.0025,
      "step": 270210
    },
    {
      "epoch": 1.7331071438879144,
      "grad_norm": 0.09086151421070099,
      "learning_rate": 5.327042178343078e-07,
      "loss": 0.0068,
      "step": 270220
    },
    {
      "epoch": 1.7331712807817006,
      "grad_norm": 0.19252285361289978,
      "learning_rate": 5.324528581604177e-07,
      "loss": 0.0013,
      "step": 270230
    },
    {
      "epoch": 1.7332354176754867,
      "grad_norm": 0.15432101488113403,
      "learning_rate": 5.322015544680848e-07,
      "loss": 0.0019,
      "step": 270240
    },
    {
      "epoch": 1.7332995545692726,
      "grad_norm": 0.10898509621620178,
      "learning_rate": 5.319503067604604e-07,
      "loss": 0.0016,
      "step": 270250
    },
    {
      "epoch": 1.7333636914630588,
      "grad_norm": 0.14141759276390076,
      "learning_rate": 5.31699115040692e-07,
      "loss": 0.0012,
      "step": 270260
    },
    {
      "epoch": 1.7334278283568447,
      "grad_norm": 0.009414039552211761,
      "learning_rate": 5.314479793119271e-07,
      "loss": 0.0022,
      "step": 270270
    },
    {
      "epoch": 1.7334919652506309,
      "grad_norm": 0.13237473368644714,
      "learning_rate": 5.311968995773115e-07,
      "loss": 0.0008,
      "step": 270280
    },
    {
      "epoch": 1.733556102144417,
      "grad_norm": 0.04768795520067215,
      "learning_rate": 5.309458758399938e-07,
      "loss": 0.0038,
      "step": 270290
    },
    {
      "epoch": 1.7336202390382032,
      "grad_norm": 0.12000446021556854,
      "learning_rate": 5.306949081031182e-07,
      "loss": 0.0021,
      "step": 270300
    },
    {
      "epoch": 1.7336843759319893,
      "grad_norm": 0.08753681182861328,
      "learning_rate": 5.304439963698299e-07,
      "loss": 0.0051,
      "step": 270310
    },
    {
      "epoch": 1.7337485128257755,
      "grad_norm": 0.15666162967681885,
      "learning_rate": 5.301931406432715e-07,
      "loss": 0.0009,
      "step": 270320
    },
    {
      "epoch": 1.7338126497195614,
      "grad_norm": 0.04588661715388298,
      "learning_rate": 5.29942340926588e-07,
      "loss": 0.0014,
      "step": 270330
    },
    {
      "epoch": 1.7338767866133475,
      "grad_norm": 0.012371247634291649,
      "learning_rate": 5.296915972229222e-07,
      "loss": 0.0015,
      "step": 270340
    },
    {
      "epoch": 1.7339409235071335,
      "grad_norm": 0.010006232187151909,
      "learning_rate": 5.294409095354158e-07,
      "loss": 0.0007,
      "step": 270350
    },
    {
      "epoch": 1.7340050604009196,
      "grad_norm": 0.06989938765764236,
      "learning_rate": 5.291902778672081e-07,
      "loss": 0.0011,
      "step": 270360
    },
    {
      "epoch": 1.7340691972947058,
      "grad_norm": 0.14981599152088165,
      "learning_rate": 5.289397022214432e-07,
      "loss": 0.0009,
      "step": 270370
    },
    {
      "epoch": 1.734133334188492,
      "grad_norm": 0.014810253866016865,
      "learning_rate": 5.286891826012591e-07,
      "loss": 0.0015,
      "step": 270380
    },
    {
      "epoch": 1.734197471082278,
      "grad_norm": 0.004750589840114117,
      "learning_rate": 5.28438719009794e-07,
      "loss": 0.0012,
      "step": 270390
    },
    {
      "epoch": 1.7342616079760642,
      "grad_norm": 0.004686721134930849,
      "learning_rate": 5.281883114501874e-07,
      "loss": 0.0008,
      "step": 270400
    },
    {
      "epoch": 1.7343257448698504,
      "grad_norm": 0.09390250593423843,
      "learning_rate": 5.279379599255779e-07,
      "loss": 0.001,
      "step": 270410
    },
    {
      "epoch": 1.7343898817636363,
      "grad_norm": 0.10690487176179886,
      "learning_rate": 5.276876644391027e-07,
      "loss": 0.0011,
      "step": 270420
    },
    {
      "epoch": 1.7344540186574224,
      "grad_norm": 0.08885901421308517,
      "learning_rate": 5.274374249938969e-07,
      "loss": 0.0008,
      "step": 270430
    },
    {
      "epoch": 1.7345181555512084,
      "grad_norm": 0.04019516333937645,
      "learning_rate": 5.271872415930956e-07,
      "loss": 0.0012,
      "step": 270440
    },
    {
      "epoch": 1.7345822924449945,
      "grad_norm": 0.027625810354948044,
      "learning_rate": 5.26937114239836e-07,
      "loss": 0.0017,
      "step": 270450
    },
    {
      "epoch": 1.7346464293387807,
      "grad_norm": 0.03272608667612076,
      "learning_rate": 5.266870429372511e-07,
      "loss": 0.0019,
      "step": 270460
    },
    {
      "epoch": 1.7347105662325668,
      "grad_norm": 0.023706305772066116,
      "learning_rate": 5.264370276884751e-07,
      "loss": 0.0023,
      "step": 270470
    },
    {
      "epoch": 1.734774703126353,
      "grad_norm": 0.07645612955093384,
      "learning_rate": 5.261870684966386e-07,
      "loss": 0.0008,
      "step": 270480
    },
    {
      "epoch": 1.734838840020139,
      "grad_norm": 0.11616411805152893,
      "learning_rate": 5.259371653648771e-07,
      "loss": 0.0035,
      "step": 270490
    },
    {
      "epoch": 1.734902976913925,
      "grad_norm": 0.010350296273827553,
      "learning_rate": 5.256873182963201e-07,
      "loss": 0.0005,
      "step": 270500
    },
    {
      "epoch": 1.7349671138077112,
      "grad_norm": 0.19959133863449097,
      "learning_rate": 5.254375272940982e-07,
      "loss": 0.0016,
      "step": 270510
    },
    {
      "epoch": 1.7350312507014973,
      "grad_norm": 0.10781977325677872,
      "learning_rate": 5.251877923613424e-07,
      "loss": 0.0013,
      "step": 270520
    },
    {
      "epoch": 1.7350953875952833,
      "grad_norm": 0.11725848913192749,
      "learning_rate": 5.249381135011822e-07,
      "loss": 0.0016,
      "step": 270530
    },
    {
      "epoch": 1.7351595244890694,
      "grad_norm": 0.08592963963747025,
      "learning_rate": 5.246884907167454e-07,
      "loss": 0.0005,
      "step": 270540
    },
    {
      "epoch": 1.7352236613828556,
      "grad_norm": 0.0015895924298092723,
      "learning_rate": 5.244389240111591e-07,
      "loss": 0.0006,
      "step": 270550
    },
    {
      "epoch": 1.7352877982766417,
      "grad_norm": 0.00917302630841732,
      "learning_rate": 5.241894133875531e-07,
      "loss": 0.0015,
      "step": 270560
    },
    {
      "epoch": 1.7353519351704279,
      "grad_norm": 0.08128763735294342,
      "learning_rate": 5.239399588490524e-07,
      "loss": 0.0016,
      "step": 270570
    },
    {
      "epoch": 1.735416072064214,
      "grad_norm": 0.10477113723754883,
      "learning_rate": 5.236905603987829e-07,
      "loss": 0.001,
      "step": 270580
    },
    {
      "epoch": 1.735480208958,
      "grad_norm": 0.05633659288287163,
      "learning_rate": 5.23441218039869e-07,
      "loss": 0.0007,
      "step": 270590
    },
    {
      "epoch": 1.735544345851786,
      "grad_norm": 0.16217748820781708,
      "learning_rate": 5.231919317754369e-07,
      "loss": 0.0021,
      "step": 270600
    },
    {
      "epoch": 1.735608482745572,
      "grad_norm": 0.11766723543405533,
      "learning_rate": 5.229427016086097e-07,
      "loss": 0.0016,
      "step": 270610
    },
    {
      "epoch": 1.7356726196393581,
      "grad_norm": 0.15290795266628265,
      "learning_rate": 5.226935275425099e-07,
      "loss": 0.0009,
      "step": 270620
    },
    {
      "epoch": 1.7357367565331443,
      "grad_norm": 0.2578399181365967,
      "learning_rate": 5.224444095802589e-07,
      "loss": 0.0015,
      "step": 270630
    },
    {
      "epoch": 1.7358008934269304,
      "grad_norm": 0.04691920056939125,
      "learning_rate": 5.221953477249808e-07,
      "loss": 0.001,
      "step": 270640
    },
    {
      "epoch": 1.7358650303207166,
      "grad_norm": 0.048594821244478226,
      "learning_rate": 5.219463419797949e-07,
      "loss": 0.001,
      "step": 270650
    },
    {
      "epoch": 1.7359291672145027,
      "grad_norm": 0.11301165819168091,
      "learning_rate": 5.216973923478214e-07,
      "loss": 0.001,
      "step": 270660
    },
    {
      "epoch": 1.735993304108289,
      "grad_norm": 0.0807356983423233,
      "learning_rate": 5.214484988321805e-07,
      "loss": 0.0009,
      "step": 270670
    },
    {
      "epoch": 1.7360574410020748,
      "grad_norm": 0.06519937515258789,
      "learning_rate": 5.21199661435991e-07,
      "loss": 0.0008,
      "step": 270680
    },
    {
      "epoch": 1.736121577895861,
      "grad_norm": 0.1602220982313156,
      "learning_rate": 5.209508801623708e-07,
      "loss": 0.0008,
      "step": 270690
    },
    {
      "epoch": 1.736185714789647,
      "grad_norm": 0.051953308284282684,
      "learning_rate": 5.207021550144359e-07,
      "loss": 0.0012,
      "step": 270700
    },
    {
      "epoch": 1.736249851683433,
      "grad_norm": 0.010182995349168777,
      "learning_rate": 5.204534859953053e-07,
      "loss": 0.0009,
      "step": 270710
    },
    {
      "epoch": 1.7363139885772192,
      "grad_norm": 0.1255737692117691,
      "learning_rate": 5.202048731080944e-07,
      "loss": 0.0012,
      "step": 270720
    },
    {
      "epoch": 1.7363781254710053,
      "grad_norm": 0.016661204397678375,
      "learning_rate": 5.199563163559179e-07,
      "loss": 0.0009,
      "step": 270730
    },
    {
      "epoch": 1.7364422623647915,
      "grad_norm": 0.06500456482172012,
      "learning_rate": 5.197078157418894e-07,
      "loss": 0.0004,
      "step": 270740
    },
    {
      "epoch": 1.7365063992585776,
      "grad_norm": 0.06425906717777252,
      "learning_rate": 5.194593712691254e-07,
      "loss": 0.0015,
      "step": 270750
    },
    {
      "epoch": 1.7365705361523636,
      "grad_norm": 0.17025528848171234,
      "learning_rate": 5.192109829407371e-07,
      "loss": 0.0016,
      "step": 270760
    },
    {
      "epoch": 1.7366346730461497,
      "grad_norm": 0.07801761478185654,
      "learning_rate": 5.189626507598377e-07,
      "loss": 0.0011,
      "step": 270770
    },
    {
      "epoch": 1.7366988099399356,
      "grad_norm": 0.10675777494907379,
      "learning_rate": 5.187143747295381e-07,
      "loss": 0.0013,
      "step": 270780
    },
    {
      "epoch": 1.7367629468337218,
      "grad_norm": 0.07541347295045853,
      "learning_rate": 5.184661548529513e-07,
      "loss": 0.0008,
      "step": 270790
    },
    {
      "epoch": 1.736827083727508,
      "grad_norm": 0.07384781539440155,
      "learning_rate": 5.182179911331858e-07,
      "loss": 0.0008,
      "step": 270800
    },
    {
      "epoch": 1.736891220621294,
      "grad_norm": 0.024426599964499474,
      "learning_rate": 5.179698835733515e-07,
      "loss": 0.0011,
      "step": 270810
    },
    {
      "epoch": 1.7369553575150802,
      "grad_norm": 0.05507689341902733,
      "learning_rate": 5.177218321765587e-07,
      "loss": 0.001,
      "step": 270820
    },
    {
      "epoch": 1.7370194944088664,
      "grad_norm": 0.0703328549861908,
      "learning_rate": 5.174738369459148e-07,
      "loss": 0.0006,
      "step": 270830
    },
    {
      "epoch": 1.7370836313026525,
      "grad_norm": 0.047399815171957016,
      "learning_rate": 5.172258978845274e-07,
      "loss": 0.001,
      "step": 270840
    },
    {
      "epoch": 1.7371477681964385,
      "grad_norm": 0.02468428760766983,
      "learning_rate": 5.169780149955017e-07,
      "loss": 0.0029,
      "step": 270850
    },
    {
      "epoch": 1.7372119050902246,
      "grad_norm": 0.00034481112379580736,
      "learning_rate": 5.16730188281947e-07,
      "loss": 0.0014,
      "step": 270860
    },
    {
      "epoch": 1.7372760419840105,
      "grad_norm": 0.13004070520401,
      "learning_rate": 5.164824177469673e-07,
      "loss": 0.0013,
      "step": 270870
    },
    {
      "epoch": 1.7373401788777967,
      "grad_norm": 0.0290438923984766,
      "learning_rate": 5.162347033936671e-07,
      "loss": 0.0007,
      "step": 270880
    },
    {
      "epoch": 1.7374043157715828,
      "grad_norm": 0.012381301261484623,
      "learning_rate": 5.159870452251492e-07,
      "loss": 0.0007,
      "step": 270890
    },
    {
      "epoch": 1.737468452665369,
      "grad_norm": 0.04207621514797211,
      "learning_rate": 5.157394432445195e-07,
      "loss": 0.0022,
      "step": 270900
    },
    {
      "epoch": 1.7375325895591551,
      "grad_norm": 0.0601922944188118,
      "learning_rate": 5.154918974548795e-07,
      "loss": 0.0014,
      "step": 270910
    },
    {
      "epoch": 1.7375967264529413,
      "grad_norm": 0.03622348606586456,
      "learning_rate": 5.15244407859331e-07,
      "loss": 0.0008,
      "step": 270920
    },
    {
      "epoch": 1.7376608633467272,
      "grad_norm": 0.00952757615596056,
      "learning_rate": 5.149969744609745e-07,
      "loss": 0.0004,
      "step": 270930
    },
    {
      "epoch": 1.7377250002405134,
      "grad_norm": 0.0805339366197586,
      "learning_rate": 5.147495972629124e-07,
      "loss": 0.0009,
      "step": 270940
    },
    {
      "epoch": 1.7377891371342995,
      "grad_norm": 0.049618132412433624,
      "learning_rate": 5.145022762682427e-07,
      "loss": 0.001,
      "step": 270950
    },
    {
      "epoch": 1.7378532740280854,
      "grad_norm": 0.07571809738874435,
      "learning_rate": 5.142550114800649e-07,
      "loss": 0.0012,
      "step": 270960
    },
    {
      "epoch": 1.7379174109218716,
      "grad_norm": 0.019243987277150154,
      "learning_rate": 5.140078029014783e-07,
      "loss": 0.0013,
      "step": 270970
    },
    {
      "epoch": 1.7379815478156577,
      "grad_norm": 0.11308278143405914,
      "learning_rate": 5.137606505355802e-07,
      "loss": 0.001,
      "step": 270980
    },
    {
      "epoch": 1.7380456847094439,
      "grad_norm": 0.014698876067996025,
      "learning_rate": 5.135135543854675e-07,
      "loss": 0.0011,
      "step": 270990
    },
    {
      "epoch": 1.73810982160323,
      "grad_norm": 0.041699737310409546,
      "learning_rate": 5.132665144542354e-07,
      "loss": 0.0009,
      "step": 271000
    },
    {
      "epoch": 1.7381739584970162,
      "grad_norm": 0.066974937915802,
      "learning_rate": 5.130195307449815e-07,
      "loss": 0.001,
      "step": 271010
    },
    {
      "epoch": 1.738238095390802,
      "grad_norm": 0.11968187242746353,
      "learning_rate": 5.127726032607994e-07,
      "loss": 0.0016,
      "step": 271020
    },
    {
      "epoch": 1.7383022322845882,
      "grad_norm": 0.21375861763954163,
      "learning_rate": 5.125257320047839e-07,
      "loss": 0.001,
      "step": 271030
    },
    {
      "epoch": 1.7383663691783742,
      "grad_norm": 0.11760486662387848,
      "learning_rate": 5.122789169800269e-07,
      "loss": 0.001,
      "step": 271040
    },
    {
      "epoch": 1.7384305060721603,
      "grad_norm": 0.08709526062011719,
      "learning_rate": 5.120321581896237e-07,
      "loss": 0.0006,
      "step": 271050
    },
    {
      "epoch": 1.7384946429659465,
      "grad_norm": 0.08717769384384155,
      "learning_rate": 5.117854556366647e-07,
      "loss": 0.0016,
      "step": 271060
    },
    {
      "epoch": 1.7385587798597326,
      "grad_norm": 0.047181129455566406,
      "learning_rate": 5.115388093242418e-07,
      "loss": 0.001,
      "step": 271070
    },
    {
      "epoch": 1.7386229167535188,
      "grad_norm": 0.1129605770111084,
      "learning_rate": 5.112922192554443e-07,
      "loss": 0.0012,
      "step": 271080
    },
    {
      "epoch": 1.738687053647305,
      "grad_norm": 0.15419110655784607,
      "learning_rate": 5.110456854333645e-07,
      "loss": 0.001,
      "step": 271090
    },
    {
      "epoch": 1.738751190541091,
      "grad_norm": 0.01996695250272751,
      "learning_rate": 5.107992078610901e-07,
      "loss": 0.0007,
      "step": 271100
    },
    {
      "epoch": 1.738815327434877,
      "grad_norm": 0.0014193024253472686,
      "learning_rate": 5.105527865417098e-07,
      "loss": 0.0008,
      "step": 271110
    },
    {
      "epoch": 1.7388794643286631,
      "grad_norm": 0.0876384973526001,
      "learning_rate": 5.103064214783121e-07,
      "loss": 0.0014,
      "step": 271120
    },
    {
      "epoch": 1.738943601222449,
      "grad_norm": 0.04958288371562958,
      "learning_rate": 5.100601126739835e-07,
      "loss": 0.0007,
      "step": 271130
    },
    {
      "epoch": 1.7390077381162352,
      "grad_norm": 0.036646727472543716,
      "learning_rate": 5.098138601318109e-07,
      "loss": 0.0008,
      "step": 271140
    },
    {
      "epoch": 1.7390718750100214,
      "grad_norm": 0.05905251204967499,
      "learning_rate": 5.095676638548786e-07,
      "loss": 0.0009,
      "step": 271150
    },
    {
      "epoch": 1.7391360119038075,
      "grad_norm": 0.05176056921482086,
      "learning_rate": 5.093215238462728e-07,
      "loss": 0.0006,
      "step": 271160
    },
    {
      "epoch": 1.7392001487975937,
      "grad_norm": 0.06435713917016983,
      "learning_rate": 5.090754401090787e-07,
      "loss": 0.0008,
      "step": 271170
    },
    {
      "epoch": 1.7392642856913798,
      "grad_norm": 0.007998064160346985,
      "learning_rate": 5.088294126463789e-07,
      "loss": 0.001,
      "step": 271180
    },
    {
      "epoch": 1.7393284225851657,
      "grad_norm": 0.016270240768790245,
      "learning_rate": 5.085834414612567e-07,
      "loss": 0.0009,
      "step": 271190
    },
    {
      "epoch": 1.7393925594789519,
      "grad_norm": 0.08200455456972122,
      "learning_rate": 5.083375265567925e-07,
      "loss": 0.0009,
      "step": 271200
    },
    {
      "epoch": 1.7394566963727378,
      "grad_norm": 0.15137460827827454,
      "learning_rate": 5.080916679360703e-07,
      "loss": 0.0008,
      "step": 271210
    },
    {
      "epoch": 1.739520833266524,
      "grad_norm": 0.05072480067610741,
      "learning_rate": 5.078458656021701e-07,
      "loss": 0.001,
      "step": 271220
    },
    {
      "epoch": 1.73958497016031,
      "grad_norm": 0.15417474508285522,
      "learning_rate": 5.076001195581709e-07,
      "loss": 0.0012,
      "step": 271230
    },
    {
      "epoch": 1.7396491070540963,
      "grad_norm": 0.032537128776311874,
      "learning_rate": 5.073544298071531e-07,
      "loss": 0.001,
      "step": 271240
    },
    {
      "epoch": 1.7397132439478824,
      "grad_norm": 0.004596466664224863,
      "learning_rate": 5.071087963521959e-07,
      "loss": 0.0011,
      "step": 271250
    },
    {
      "epoch": 1.7397773808416686,
      "grad_norm": 0.09249936789274216,
      "learning_rate": 5.068632191963758e-07,
      "loss": 0.0009,
      "step": 271260
    },
    {
      "epoch": 1.7398415177354547,
      "grad_norm": 0.013414569199085236,
      "learning_rate": 5.066176983427701e-07,
      "loss": 0.0017,
      "step": 271270
    },
    {
      "epoch": 1.7399056546292406,
      "grad_norm": 0.11013256013393402,
      "learning_rate": 5.063722337944571e-07,
      "loss": 0.0018,
      "step": 271280
    },
    {
      "epoch": 1.7399697915230268,
      "grad_norm": 0.16548310220241547,
      "learning_rate": 5.061268255545115e-07,
      "loss": 0.0025,
      "step": 271290
    },
    {
      "epoch": 1.7400339284168127,
      "grad_norm": 0.15287838876247406,
      "learning_rate": 5.058814736260087e-07,
      "loss": 0.0012,
      "step": 271300
    },
    {
      "epoch": 1.7400980653105989,
      "grad_norm": 0.11675088852643967,
      "learning_rate": 5.056361780120216e-07,
      "loss": 0.0016,
      "step": 271310
    },
    {
      "epoch": 1.740162202204385,
      "grad_norm": 0.04161589592695236,
      "learning_rate": 5.053909387156264e-07,
      "loss": 0.0027,
      "step": 271320
    },
    {
      "epoch": 1.7402263390981711,
      "grad_norm": 0.07786667346954346,
      "learning_rate": 5.051457557398948e-07,
      "loss": 0.0013,
      "step": 271330
    },
    {
      "epoch": 1.7402904759919573,
      "grad_norm": 0.08022894710302353,
      "learning_rate": 5.049006290878993e-07,
      "loss": 0.0024,
      "step": 271340
    },
    {
      "epoch": 1.7403546128857434,
      "grad_norm": 0.10996535420417786,
      "learning_rate": 5.046555587627111e-07,
      "loss": 0.0016,
      "step": 271350
    },
    {
      "epoch": 1.7404187497795294,
      "grad_norm": 0.2262801229953766,
      "learning_rate": 5.044105447674019e-07,
      "loss": 0.0017,
      "step": 271360
    },
    {
      "epoch": 1.7404828866733155,
      "grad_norm": 0.044142354279756546,
      "learning_rate": 5.041655871050416e-07,
      "loss": 0.0008,
      "step": 271370
    },
    {
      "epoch": 1.7405470235671017,
      "grad_norm": 0.16979172825813293,
      "learning_rate": 5.039206857786988e-07,
      "loss": 0.0008,
      "step": 271380
    },
    {
      "epoch": 1.7406111604608876,
      "grad_norm": 0.12596730887889862,
      "learning_rate": 5.036758407914444e-07,
      "loss": 0.0014,
      "step": 271390
    },
    {
      "epoch": 1.7406752973546737,
      "grad_norm": 0.015053360722959042,
      "learning_rate": 5.034310521463449e-07,
      "loss": 0.0011,
      "step": 271400
    },
    {
      "epoch": 1.74073943424846,
      "grad_norm": 0.10763084143400192,
      "learning_rate": 5.031863198464676e-07,
      "loss": 0.0015,
      "step": 271410
    },
    {
      "epoch": 1.740803571142246,
      "grad_norm": 0.08218323439359665,
      "learning_rate": 5.029416438948786e-07,
      "loss": 0.0011,
      "step": 271420
    },
    {
      "epoch": 1.7408677080360322,
      "grad_norm": 0.0035591446794569492,
      "learning_rate": 5.026970242946466e-07,
      "loss": 0.0014,
      "step": 271430
    },
    {
      "epoch": 1.7409318449298183,
      "grad_norm": 0.025590816512703896,
      "learning_rate": 5.024524610488341e-07,
      "loss": 0.0008,
      "step": 271440
    },
    {
      "epoch": 1.7409959818236043,
      "grad_norm": 0.10720288008451462,
      "learning_rate": 5.022079541605074e-07,
      "loss": 0.001,
      "step": 271450
    },
    {
      "epoch": 1.7410601187173904,
      "grad_norm": 0.10830137878656387,
      "learning_rate": 5.019635036327281e-07,
      "loss": 0.0014,
      "step": 271460
    },
    {
      "epoch": 1.7411242556111763,
      "grad_norm": 0.03661242872476578,
      "learning_rate": 5.01719109468562e-07,
      "loss": 0.0005,
      "step": 271470
    },
    {
      "epoch": 1.7411883925049625,
      "grad_norm": 0.04848403111100197,
      "learning_rate": 5.0147477167107e-07,
      "loss": 0.0009,
      "step": 271480
    },
    {
      "epoch": 1.7412525293987486,
      "grad_norm": 0.010289512574672699,
      "learning_rate": 5.01230490243314e-07,
      "loss": 0.0005,
      "step": 271490
    },
    {
      "epoch": 1.7413166662925348,
      "grad_norm": 0.05726618319749832,
      "learning_rate": 5.009862651883546e-07,
      "loss": 0.0012,
      "step": 271500
    },
    {
      "epoch": 1.741380803186321,
      "grad_norm": 0.007875513285398483,
      "learning_rate": 5.00742096509254e-07,
      "loss": 0.0013,
      "step": 271510
    },
    {
      "epoch": 1.741444940080107,
      "grad_norm": 0.12497909367084503,
      "learning_rate": 5.004979842090702e-07,
      "loss": 0.0015,
      "step": 271520
    },
    {
      "epoch": 1.7415090769738932,
      "grad_norm": 0.07876826077699661,
      "learning_rate": 5.002539282908614e-07,
      "loss": 0.0032,
      "step": 271530
    },
    {
      "epoch": 1.7415732138676792,
      "grad_norm": 0.052569467574357986,
      "learning_rate": 5.000099287576876e-07,
      "loss": 0.0009,
      "step": 271540
    },
    {
      "epoch": 1.7416373507614653,
      "grad_norm": 0.1996789127588272,
      "learning_rate": 4.997659856126053e-07,
      "loss": 0.0012,
      "step": 271550
    },
    {
      "epoch": 1.7417014876552512,
      "grad_norm": 0.05426711589097977,
      "learning_rate": 4.995220988586719e-07,
      "loss": 0.001,
      "step": 271560
    },
    {
      "epoch": 1.7417656245490374,
      "grad_norm": 0.16309188306331635,
      "learning_rate": 4.992782684989422e-07,
      "loss": 0.0009,
      "step": 271570
    },
    {
      "epoch": 1.7418297614428235,
      "grad_norm": 0.10447095334529877,
      "learning_rate": 4.990344945364727e-07,
      "loss": 0.0011,
      "step": 271580
    },
    {
      "epoch": 1.7418938983366097,
      "grad_norm": 0.14355474710464478,
      "learning_rate": 4.987907769743183e-07,
      "loss": 0.0004,
      "step": 271590
    },
    {
      "epoch": 1.7419580352303958,
      "grad_norm": 0.06109283119440079,
      "learning_rate": 4.985471158155325e-07,
      "loss": 0.0009,
      "step": 271600
    },
    {
      "epoch": 1.742022172124182,
      "grad_norm": 0.012738275341689587,
      "learning_rate": 4.983035110631674e-07,
      "loss": 0.0013,
      "step": 271610
    },
    {
      "epoch": 1.742086309017968,
      "grad_norm": 0.13235825300216675,
      "learning_rate": 4.980599627202776e-07,
      "loss": 0.0007,
      "step": 271620
    },
    {
      "epoch": 1.742150445911754,
      "grad_norm": 0.05144060403108597,
      "learning_rate": 4.978164707899142e-07,
      "loss": 0.0009,
      "step": 271630
    },
    {
      "epoch": 1.74221458280554,
      "grad_norm": 0.04663949832320213,
      "learning_rate": 4.975730352751273e-07,
      "loss": 0.0005,
      "step": 271640
    },
    {
      "epoch": 1.7422787196993261,
      "grad_norm": 0.03589535877108574,
      "learning_rate": 4.973296561789676e-07,
      "loss": 0.001,
      "step": 271650
    },
    {
      "epoch": 1.7423428565931123,
      "grad_norm": 0.030549991875886917,
      "learning_rate": 4.970863335044867e-07,
      "loss": 0.0006,
      "step": 271660
    },
    {
      "epoch": 1.7424069934868984,
      "grad_norm": 0.03686181828379631,
      "learning_rate": 4.968430672547314e-07,
      "loss": 0.0018,
      "step": 271670
    },
    {
      "epoch": 1.7424711303806846,
      "grad_norm": 0.01910620741546154,
      "learning_rate": 4.965998574327508e-07,
      "loss": 0.0013,
      "step": 271680
    },
    {
      "epoch": 1.7425352672744707,
      "grad_norm": 0.05751664936542511,
      "learning_rate": 4.963567040415929e-07,
      "loss": 0.0009,
      "step": 271690
    },
    {
      "epoch": 1.7425994041682569,
      "grad_norm": 0.014478910714387894,
      "learning_rate": 4.961136070843043e-07,
      "loss": 0.0009,
      "step": 271700
    },
    {
      "epoch": 1.7426635410620428,
      "grad_norm": 0.01925645023584366,
      "learning_rate": 4.958705665639308e-07,
      "loss": 0.0009,
      "step": 271710
    },
    {
      "epoch": 1.742727677955829,
      "grad_norm": 0.05361935496330261,
      "learning_rate": 4.956275824835177e-07,
      "loss": 0.0012,
      "step": 271720
    },
    {
      "epoch": 1.7427918148496149,
      "grad_norm": 0.051223274320364,
      "learning_rate": 4.953846548461105e-07,
      "loss": 0.0013,
      "step": 271730
    },
    {
      "epoch": 1.742855951743401,
      "grad_norm": 0.08455495536327362,
      "learning_rate": 4.951417836547539e-07,
      "loss": 0.002,
      "step": 271740
    },
    {
      "epoch": 1.7429200886371872,
      "grad_norm": 0.05032658204436302,
      "learning_rate": 4.948989689124894e-07,
      "loss": 0.0012,
      "step": 271750
    },
    {
      "epoch": 1.7429842255309733,
      "grad_norm": 0.047818951308727264,
      "learning_rate": 4.946562106223602e-07,
      "loss": 0.0014,
      "step": 271760
    },
    {
      "epoch": 1.7430483624247595,
      "grad_norm": 0.005683389492332935,
      "learning_rate": 4.944135087874097e-07,
      "loss": 0.0017,
      "step": 271770
    },
    {
      "epoch": 1.7431124993185456,
      "grad_norm": 0.024180131033062935,
      "learning_rate": 4.941708634106773e-07,
      "loss": 0.0013,
      "step": 271780
    },
    {
      "epoch": 1.7431766362123318,
      "grad_norm": 0.12976869940757751,
      "learning_rate": 4.939282744952051e-07,
      "loss": 0.0007,
      "step": 271790
    },
    {
      "epoch": 1.7432407731061177,
      "grad_norm": 0.14844253659248352,
      "learning_rate": 4.936857420440306e-07,
      "loss": 0.001,
      "step": 271800
    },
    {
      "epoch": 1.7433049099999038,
      "grad_norm": 0.059311579912900925,
      "learning_rate": 4.934432660601951e-07,
      "loss": 0.0007,
      "step": 271810
    },
    {
      "epoch": 1.7433690468936898,
      "grad_norm": 0.07273940742015839,
      "learning_rate": 4.932008465467369e-07,
      "loss": 0.0019,
      "step": 271820
    },
    {
      "epoch": 1.743433183787476,
      "grad_norm": 0.04856349900364876,
      "learning_rate": 4.929584835066914e-07,
      "loss": 0.001,
      "step": 271830
    },
    {
      "epoch": 1.743497320681262,
      "grad_norm": 0.11532484740018845,
      "learning_rate": 4.927161769430989e-07,
      "loss": 0.0014,
      "step": 271840
    },
    {
      "epoch": 1.7435614575750482,
      "grad_norm": 0.08793105185031891,
      "learning_rate": 4.924739268589934e-07,
      "loss": 0.0012,
      "step": 271850
    },
    {
      "epoch": 1.7436255944688344,
      "grad_norm": 0.057791031897068024,
      "learning_rate": 4.922317332574117e-07,
      "loss": 0.0022,
      "step": 271860
    },
    {
      "epoch": 1.7436897313626205,
      "grad_norm": 0.2020353376865387,
      "learning_rate": 4.919895961413867e-07,
      "loss": 0.0021,
      "step": 271870
    },
    {
      "epoch": 1.7437538682564064,
      "grad_norm": 0.1663372665643692,
      "learning_rate": 4.917475155139545e-07,
      "loss": 0.0013,
      "step": 271880
    },
    {
      "epoch": 1.7438180051501926,
      "grad_norm": 0.1889677196741104,
      "learning_rate": 4.915054913781486e-07,
      "loss": 0.001,
      "step": 271890
    },
    {
      "epoch": 1.7438821420439785,
      "grad_norm": 0.1728479266166687,
      "learning_rate": 4.912635237370006e-07,
      "loss": 0.0012,
      "step": 271900
    },
    {
      "epoch": 1.7439462789377647,
      "grad_norm": 0.06438399851322174,
      "learning_rate": 4.910216125935424e-07,
      "loss": 0.0007,
      "step": 271910
    },
    {
      "epoch": 1.7440104158315508,
      "grad_norm": 0.06103010103106499,
      "learning_rate": 4.907797579508066e-07,
      "loss": 0.0005,
      "step": 271920
    },
    {
      "epoch": 1.744074552725337,
      "grad_norm": 0.04396438226103783,
      "learning_rate": 4.905379598118221e-07,
      "loss": 0.0013,
      "step": 271930
    },
    {
      "epoch": 1.744138689619123,
      "grad_norm": 0.4433193504810333,
      "learning_rate": 4.902962181796211e-07,
      "loss": 0.0027,
      "step": 271940
    },
    {
      "epoch": 1.7442028265129093,
      "grad_norm": 0.006359719205647707,
      "learning_rate": 4.900545330572304e-07,
      "loss": 0.0004,
      "step": 271950
    },
    {
      "epoch": 1.7442669634066954,
      "grad_norm": 0.060575541108846664,
      "learning_rate": 4.898129044476802e-07,
      "loss": 0.0022,
      "step": 271960
    },
    {
      "epoch": 1.7443311003004813,
      "grad_norm": 0.032161895185709,
      "learning_rate": 4.89571332353998e-07,
      "loss": 0.001,
      "step": 271970
    },
    {
      "epoch": 1.7443952371942675,
      "grad_norm": 0.043320585042238235,
      "learning_rate": 4.89329816779211e-07,
      "loss": 0.0019,
      "step": 271980
    },
    {
      "epoch": 1.7444593740880534,
      "grad_norm": 0.0536777563393116,
      "learning_rate": 4.890883577263439e-07,
      "loss": 0.0005,
      "step": 271990
    },
    {
      "epoch": 1.7445235109818396,
      "grad_norm": 0.011271800845861435,
      "learning_rate": 4.888469551984243e-07,
      "loss": 0.0005,
      "step": 272000
    },
    {
      "epoch": 1.7445876478756257,
      "grad_norm": 0.08536845445632935,
      "learning_rate": 4.886056091984764e-07,
      "loss": 0.002,
      "step": 272010
    },
    {
      "epoch": 1.7446517847694119,
      "grad_norm": 0.015522822737693787,
      "learning_rate": 4.883643197295246e-07,
      "loss": 0.0009,
      "step": 272020
    },
    {
      "epoch": 1.744715921663198,
      "grad_norm": 0.09444174915552139,
      "learning_rate": 4.881230867945913e-07,
      "loss": 0.001,
      "step": 272030
    },
    {
      "epoch": 1.7447800585569841,
      "grad_norm": 0.05753888562321663,
      "learning_rate": 4.878819103967014e-07,
      "loss": 0.0024,
      "step": 272040
    },
    {
      "epoch": 1.74484419545077,
      "grad_norm": 0.130371555685997,
      "learning_rate": 4.876407905388758e-07,
      "loss": 0.0007,
      "step": 272050
    },
    {
      "epoch": 1.7449083323445562,
      "grad_norm": 0.13810470700263977,
      "learning_rate": 4.87399727224136e-07,
      "loss": 0.0006,
      "step": 272060
    },
    {
      "epoch": 1.7449724692383424,
      "grad_norm": 0.10591180622577667,
      "learning_rate": 4.871587204555018e-07,
      "loss": 0.0013,
      "step": 272070
    },
    {
      "epoch": 1.7450366061321283,
      "grad_norm": 0.1073022112250328,
      "learning_rate": 4.869177702359951e-07,
      "loss": 0.0008,
      "step": 272080
    },
    {
      "epoch": 1.7451007430259144,
      "grad_norm": 0.036559127271175385,
      "learning_rate": 4.866768765686342e-07,
      "loss": 0.0016,
      "step": 272090
    },
    {
      "epoch": 1.7451648799197006,
      "grad_norm": 0.008580600842833519,
      "learning_rate": 4.864360394564366e-07,
      "loss": 0.0006,
      "step": 272100
    },
    {
      "epoch": 1.7452290168134867,
      "grad_norm": 0.06898060441017151,
      "learning_rate": 4.861952589024222e-07,
      "loss": 0.0015,
      "step": 272110
    },
    {
      "epoch": 1.745293153707273,
      "grad_norm": 0.060005057603120804,
      "learning_rate": 4.859545349096073e-07,
      "loss": 0.0022,
      "step": 272120
    },
    {
      "epoch": 1.745357290601059,
      "grad_norm": 0.04038242995738983,
      "learning_rate": 4.857138674810081e-07,
      "loss": 0.0025,
      "step": 272130
    },
    {
      "epoch": 1.745421427494845,
      "grad_norm": 0.10667458921670914,
      "learning_rate": 4.854732566196397e-07,
      "loss": 0.0015,
      "step": 272140
    },
    {
      "epoch": 1.7454855643886311,
      "grad_norm": 0.10232573002576828,
      "learning_rate": 4.852327023285186e-07,
      "loss": 0.0021,
      "step": 272150
    },
    {
      "epoch": 1.745549701282417,
      "grad_norm": 0.06841970235109329,
      "learning_rate": 4.849922046106581e-07,
      "loss": 0.0015,
      "step": 272160
    },
    {
      "epoch": 1.7456138381762032,
      "grad_norm": 0.04854750633239746,
      "learning_rate": 4.847517634690729e-07,
      "loss": 0.0006,
      "step": 272170
    },
    {
      "epoch": 1.7456779750699893,
      "grad_norm": 0.05373238027095795,
      "learning_rate": 4.845113789067735e-07,
      "loss": 0.0005,
      "step": 272180
    },
    {
      "epoch": 1.7457421119637755,
      "grad_norm": 0.031732819974422455,
      "learning_rate": 4.84271050926775e-07,
      "loss": 0.0013,
      "step": 272190
    },
    {
      "epoch": 1.7458062488575616,
      "grad_norm": 0.0857602208852768,
      "learning_rate": 4.840307795320876e-07,
      "loss": 0.0016,
      "step": 272200
    },
    {
      "epoch": 1.7458703857513478,
      "grad_norm": 0.064203180372715,
      "learning_rate": 4.837905647257207e-07,
      "loss": 0.0048,
      "step": 272210
    },
    {
      "epoch": 1.745934522645134,
      "grad_norm": 0.1207960918545723,
      "learning_rate": 4.835504065106872e-07,
      "loss": 0.0008,
      "step": 272220
    },
    {
      "epoch": 1.7459986595389199,
      "grad_norm": 0.024911632761359215,
      "learning_rate": 4.833103048899946e-07,
      "loss": 0.0013,
      "step": 272230
    },
    {
      "epoch": 1.746062796432706,
      "grad_norm": 0.052240040153265,
      "learning_rate": 4.83070259866652e-07,
      "loss": 0.0006,
      "step": 272240
    },
    {
      "epoch": 1.746126933326492,
      "grad_norm": 0.3158170282840729,
      "learning_rate": 4.828302714436661e-07,
      "loss": 0.0017,
      "step": 272250
    },
    {
      "epoch": 1.746191070220278,
      "grad_norm": 0.10014260560274124,
      "learning_rate": 4.825903396240461e-07,
      "loss": 0.0012,
      "step": 272260
    },
    {
      "epoch": 1.7462552071140642,
      "grad_norm": 0.06072854995727539,
      "learning_rate": 4.823504644107985e-07,
      "loss": 0.0006,
      "step": 272270
    },
    {
      "epoch": 1.7463193440078504,
      "grad_norm": 0.018507638946175575,
      "learning_rate": 4.821106458069275e-07,
      "loss": 0.0012,
      "step": 272280
    },
    {
      "epoch": 1.7463834809016365,
      "grad_norm": 0.023834386840462685,
      "learning_rate": 4.818708838154384e-07,
      "loss": 0.0008,
      "step": 272290
    },
    {
      "epoch": 1.7464476177954227,
      "grad_norm": 0.07017380744218826,
      "learning_rate": 4.816311784393368e-07,
      "loss": 0.0008,
      "step": 272300
    },
    {
      "epoch": 1.7465117546892086,
      "grad_norm": 0.07759445160627365,
      "learning_rate": 4.813915296816263e-07,
      "loss": 0.0034,
      "step": 272310
    },
    {
      "epoch": 1.7465758915829948,
      "grad_norm": 0.025435132905840874,
      "learning_rate": 4.811519375453089e-07,
      "loss": 0.0004,
      "step": 272320
    },
    {
      "epoch": 1.7466400284767807,
      "grad_norm": 0.12056136876344681,
      "learning_rate": 4.809124020333867e-07,
      "loss": 0.0012,
      "step": 272330
    },
    {
      "epoch": 1.7467041653705668,
      "grad_norm": 0.1359078586101532,
      "learning_rate": 4.806729231488622e-07,
      "loss": 0.0011,
      "step": 272340
    },
    {
      "epoch": 1.746768302264353,
      "grad_norm": 0.008710467256605625,
      "learning_rate": 4.804335008947364e-07,
      "loss": 0.0009,
      "step": 272350
    },
    {
      "epoch": 1.7468324391581391,
      "grad_norm": 0.15154539048671722,
      "learning_rate": 4.80194135274008e-07,
      "loss": 0.0006,
      "step": 272360
    },
    {
      "epoch": 1.7468965760519253,
      "grad_norm": 0.16552463173866272,
      "learning_rate": 4.799548262896781e-07,
      "loss": 0.0014,
      "step": 272370
    },
    {
      "epoch": 1.7469607129457114,
      "grad_norm": 0.05478297173976898,
      "learning_rate": 4.797155739447445e-07,
      "loss": 0.0007,
      "step": 272380
    },
    {
      "epoch": 1.7470248498394976,
      "grad_norm": 0.010676765814423561,
      "learning_rate": 4.794763782422057e-07,
      "loss": 0.0006,
      "step": 272390
    },
    {
      "epoch": 1.7470889867332835,
      "grad_norm": 0.06147603318095207,
      "learning_rate": 4.792372391850575e-07,
      "loss": 0.001,
      "step": 272400
    },
    {
      "epoch": 1.7471531236270696,
      "grad_norm": 0.06963121891021729,
      "learning_rate": 4.789981567762986e-07,
      "loss": 0.001,
      "step": 272410
    },
    {
      "epoch": 1.7472172605208556,
      "grad_norm": 0.2152966409921646,
      "learning_rate": 4.787591310189238e-07,
      "loss": 0.0009,
      "step": 272420
    },
    {
      "epoch": 1.7472813974146417,
      "grad_norm": 0.1609608232975006,
      "learning_rate": 4.785201619159285e-07,
      "loss": 0.0017,
      "step": 272430
    },
    {
      "epoch": 1.7473455343084279,
      "grad_norm": 0.006035391241312027,
      "learning_rate": 4.782812494703066e-07,
      "loss": 0.0006,
      "step": 272440
    },
    {
      "epoch": 1.747409671202214,
      "grad_norm": 0.06160098686814308,
      "learning_rate": 4.780423936850526e-07,
      "loss": 0.0009,
      "step": 272450
    },
    {
      "epoch": 1.7474738080960002,
      "grad_norm": 0.06009820103645325,
      "learning_rate": 4.778035945631593e-07,
      "loss": 0.0013,
      "step": 272460
    },
    {
      "epoch": 1.7475379449897863,
      "grad_norm": 0.03193667158484459,
      "learning_rate": 4.775648521076187e-07,
      "loss": 0.0011,
      "step": 272470
    },
    {
      "epoch": 1.7476020818835722,
      "grad_norm": 0.11648708581924438,
      "learning_rate": 4.773261663214218e-07,
      "loss": 0.0023,
      "step": 272480
    },
    {
      "epoch": 1.7476662187773584,
      "grad_norm": 0.056543439626693726,
      "learning_rate": 4.770875372075618e-07,
      "loss": 0.0007,
      "step": 272490
    },
    {
      "epoch": 1.7477303556711445,
      "grad_norm": 0.0710235983133316,
      "learning_rate": 4.768489647690266e-07,
      "loss": 0.006,
      "step": 272500
    },
    {
      "epoch": 1.7477944925649305,
      "grad_norm": 0.06654216349124908,
      "learning_rate": 4.7661044900880636e-07,
      "loss": 0.0006,
      "step": 272510
    },
    {
      "epoch": 1.7478586294587166,
      "grad_norm": 0.3334129750728607,
      "learning_rate": 4.7637198992989075e-07,
      "loss": 0.0011,
      "step": 272520
    },
    {
      "epoch": 1.7479227663525028,
      "grad_norm": 0.06493246555328369,
      "learning_rate": 4.7613358753526686e-07,
      "loss": 0.0009,
      "step": 272530
    },
    {
      "epoch": 1.747986903246289,
      "grad_norm": 0.04161922633647919,
      "learning_rate": 4.7589524182792226e-07,
      "loss": 0.0007,
      "step": 272540
    },
    {
      "epoch": 1.748051040140075,
      "grad_norm": 0.02677379921078682,
      "learning_rate": 4.756569528108429e-07,
      "loss": 0.0004,
      "step": 272550
    },
    {
      "epoch": 1.7481151770338612,
      "grad_norm": 0.002950383350253105,
      "learning_rate": 4.7541872048701585e-07,
      "loss": 0.0009,
      "step": 272560
    },
    {
      "epoch": 1.7481793139276471,
      "grad_norm": 0.06629134714603424,
      "learning_rate": 4.7518054485942653e-07,
      "loss": 0.0011,
      "step": 272570
    },
    {
      "epoch": 1.7482434508214333,
      "grad_norm": 0.022829001769423485,
      "learning_rate": 4.7494242593105867e-07,
      "loss": 0.0006,
      "step": 272580
    },
    {
      "epoch": 1.7483075877152192,
      "grad_norm": 0.020911620929837227,
      "learning_rate": 4.7470436370489535e-07,
      "loss": 0.0003,
      "step": 272590
    },
    {
      "epoch": 1.7483717246090054,
      "grad_norm": 0.040540631860494614,
      "learning_rate": 4.7446635818392093e-07,
      "loss": 0.0008,
      "step": 272600
    },
    {
      "epoch": 1.7484358615027915,
      "grad_norm": 0.06236898899078369,
      "learning_rate": 4.7422840937111804e-07,
      "loss": 0.0011,
      "step": 272610
    },
    {
      "epoch": 1.7484999983965777,
      "grad_norm": 0.0678824707865715,
      "learning_rate": 4.7399051726946754e-07,
      "loss": 0.0009,
      "step": 272620
    },
    {
      "epoch": 1.7485641352903638,
      "grad_norm": 0.00921363290399313,
      "learning_rate": 4.7375268188194934e-07,
      "loss": 0.0008,
      "step": 272630
    },
    {
      "epoch": 1.74862827218415,
      "grad_norm": 0.3204735815525055,
      "learning_rate": 4.73514903211546e-07,
      "loss": 0.0032,
      "step": 272640
    },
    {
      "epoch": 1.748692409077936,
      "grad_norm": 0.03617285192012787,
      "learning_rate": 4.7327718126123577e-07,
      "loss": 0.0015,
      "step": 272650
    },
    {
      "epoch": 1.748756545971722,
      "grad_norm": 0.022229550406336784,
      "learning_rate": 4.7303951603399736e-07,
      "loss": 0.0006,
      "step": 272660
    },
    {
      "epoch": 1.7488206828655082,
      "grad_norm": 0.006759267766028643,
      "learning_rate": 4.7280190753280943e-07,
      "loss": 0.0006,
      "step": 272670
    },
    {
      "epoch": 1.748884819759294,
      "grad_norm": 0.1989506185054779,
      "learning_rate": 4.7256435576064963e-07,
      "loss": 0.0006,
      "step": 272680
    },
    {
      "epoch": 1.7489489566530803,
      "grad_norm": 0.06484976410865784,
      "learning_rate": 4.7232686072049394e-07,
      "loss": 0.0009,
      "step": 272690
    },
    {
      "epoch": 1.7490130935468664,
      "grad_norm": 0.0020673118997365236,
      "learning_rate": 4.720894224153172e-07,
      "loss": 0.001,
      "step": 272700
    },
    {
      "epoch": 1.7490772304406526,
      "grad_norm": 0.10880999267101288,
      "learning_rate": 4.718520408480964e-07,
      "loss": 0.0019,
      "step": 272710
    },
    {
      "epoch": 1.7491413673344387,
      "grad_norm": 0.40636178851127625,
      "learning_rate": 4.716147160218071e-07,
      "loss": 0.001,
      "step": 272720
    },
    {
      "epoch": 1.7492055042282249,
      "grad_norm": 0.316981703042984,
      "learning_rate": 4.713774479394212e-07,
      "loss": 0.0019,
      "step": 272730
    },
    {
      "epoch": 1.7492696411220108,
      "grad_norm": 0.11160846799612045,
      "learning_rate": 4.711402366039125e-07,
      "loss": 0.0017,
      "step": 272740
    },
    {
      "epoch": 1.749333778015797,
      "grad_norm": 0.04493987560272217,
      "learning_rate": 4.7090308201825305e-07,
      "loss": 0.0008,
      "step": 272750
    },
    {
      "epoch": 1.7493979149095829,
      "grad_norm": 0.06282531470060349,
      "learning_rate": 4.70665984185415e-07,
      "loss": 0.0007,
      "step": 272760
    },
    {
      "epoch": 1.749462051803369,
      "grad_norm": 0.08844546228647232,
      "learning_rate": 4.7042894310836975e-07,
      "loss": 0.0009,
      "step": 272770
    },
    {
      "epoch": 1.7495261886971551,
      "grad_norm": 0.06802531331777573,
      "learning_rate": 4.7019195879008603e-07,
      "loss": 0.001,
      "step": 272780
    },
    {
      "epoch": 1.7495903255909413,
      "grad_norm": 0.15322864055633545,
      "learning_rate": 4.699550312335355e-07,
      "loss": 0.001,
      "step": 272790
    },
    {
      "epoch": 1.7496544624847274,
      "grad_norm": 0.13658012449741364,
      "learning_rate": 4.697181604416856e-07,
      "loss": 0.0004,
      "step": 272800
    },
    {
      "epoch": 1.7497185993785136,
      "grad_norm": 0.07306113094091415,
      "learning_rate": 4.694813464175052e-07,
      "loss": 0.0008,
      "step": 272810
    },
    {
      "epoch": 1.7497827362722997,
      "grad_norm": 0.03323390707373619,
      "learning_rate": 4.6924458916396076e-07,
      "loss": 0.0014,
      "step": 272820
    },
    {
      "epoch": 1.7498468731660857,
      "grad_norm": 0.1885232925415039,
      "learning_rate": 4.690078886840199e-07,
      "loss": 0.0016,
      "step": 272830
    },
    {
      "epoch": 1.7499110100598718,
      "grad_norm": 0.09597447514533997,
      "learning_rate": 4.687712449806492e-07,
      "loss": 0.0012,
      "step": 272840
    },
    {
      "epoch": 1.7499751469536577,
      "grad_norm": 0.2891257703304291,
      "learning_rate": 4.6853465805681287e-07,
      "loss": 0.0015,
      "step": 272850
    },
    {
      "epoch": 1.750039283847444,
      "grad_norm": 0.1429951786994934,
      "learning_rate": 4.682981279154747e-07,
      "loss": 0.0007,
      "step": 272860
    },
    {
      "epoch": 1.75010342074123,
      "grad_norm": 0.01997167058289051,
      "learning_rate": 4.680616545596012e-07,
      "loss": 0.0009,
      "step": 272870
    },
    {
      "epoch": 1.7501675576350162,
      "grad_norm": 0.04769844189286232,
      "learning_rate": 4.6782523799215384e-07,
      "loss": 0.0009,
      "step": 272880
    },
    {
      "epoch": 1.7502316945288023,
      "grad_norm": 0.07950573414564133,
      "learning_rate": 4.6758887821609534e-07,
      "loss": 0.0007,
      "step": 272890
    },
    {
      "epoch": 1.7502958314225885,
      "grad_norm": 0.06623969227075577,
      "learning_rate": 4.673525752343866e-07,
      "loss": 0.0005,
      "step": 272900
    },
    {
      "epoch": 1.7503599683163744,
      "grad_norm": 0.03988812863826752,
      "learning_rate": 4.671163290499903e-07,
      "loss": 0.0005,
      "step": 272910
    },
    {
      "epoch": 1.7504241052101606,
      "grad_norm": 0.11533651500940323,
      "learning_rate": 4.668801396658662e-07,
      "loss": 0.001,
      "step": 272920
    },
    {
      "epoch": 1.7504882421039467,
      "grad_norm": 0.08430314064025879,
      "learning_rate": 4.666440070849726e-07,
      "loss": 0.0008,
      "step": 272930
    },
    {
      "epoch": 1.7505523789977326,
      "grad_norm": 0.11702308803796768,
      "learning_rate": 4.664079313102704e-07,
      "loss": 0.0016,
      "step": 272940
    },
    {
      "epoch": 1.7506165158915188,
      "grad_norm": 0.051689263433218,
      "learning_rate": 4.6617191234471726e-07,
      "loss": 0.0005,
      "step": 272950
    },
    {
      "epoch": 1.750680652785305,
      "grad_norm": 0.09569631516933441,
      "learning_rate": 4.6593595019127023e-07,
      "loss": 0.0018,
      "step": 272960
    },
    {
      "epoch": 1.750744789679091,
      "grad_norm": 0.09990435838699341,
      "learning_rate": 4.657000448528848e-07,
      "loss": 0.0012,
      "step": 272970
    },
    {
      "epoch": 1.7508089265728772,
      "grad_norm": 0.09022072702646255,
      "learning_rate": 4.6546419633252016e-07,
      "loss": 0.0015,
      "step": 272980
    },
    {
      "epoch": 1.7508730634666634,
      "grad_norm": 0.20424622297286987,
      "learning_rate": 4.6522840463312904e-07,
      "loss": 0.0007,
      "step": 272990
    },
    {
      "epoch": 1.7509372003604493,
      "grad_norm": 0.18562482297420502,
      "learning_rate": 4.649926697576679e-07,
      "loss": 0.0006,
      "step": 273000
    },
    {
      "epoch": 1.7510013372542355,
      "grad_norm": 0.038311198353767395,
      "learning_rate": 4.6475699170908826e-07,
      "loss": 0.0007,
      "step": 273010
    },
    {
      "epoch": 1.7510654741480214,
      "grad_norm": 0.040191225707530975,
      "learning_rate": 4.645213704903456e-07,
      "loss": 0.0007,
      "step": 273020
    },
    {
      "epoch": 1.7511296110418075,
      "grad_norm": 0.011406010948121548,
      "learning_rate": 4.64285806104392e-07,
      "loss": 0.0021,
      "step": 273030
    },
    {
      "epoch": 1.7511937479355937,
      "grad_norm": 0.08132412284612656,
      "learning_rate": 4.640502985541784e-07,
      "loss": 0.0004,
      "step": 273040
    },
    {
      "epoch": 1.7512578848293798,
      "grad_norm": 0.07212943583726883,
      "learning_rate": 4.638148478426557e-07,
      "loss": 0.0013,
      "step": 273050
    },
    {
      "epoch": 1.751322021723166,
      "grad_norm": 0.055701203644275665,
      "learning_rate": 4.6357945397277615e-07,
      "loss": 0.0009,
      "step": 273060
    },
    {
      "epoch": 1.7513861586169521,
      "grad_norm": 0.10213173925876617,
      "learning_rate": 4.6334411694748785e-07,
      "loss": 0.0013,
      "step": 273070
    },
    {
      "epoch": 1.7514502955107383,
      "grad_norm": 0.047767817974090576,
      "learning_rate": 4.6310883676973895e-07,
      "loss": 0.0006,
      "step": 273080
    },
    {
      "epoch": 1.7515144324045242,
      "grad_norm": 0.07404826581478119,
      "learning_rate": 4.6287361344247995e-07,
      "loss": 0.001,
      "step": 273090
    },
    {
      "epoch": 1.7515785692983104,
      "grad_norm": 0.014767042361199856,
      "learning_rate": 4.6263844696865734e-07,
      "loss": 0.0028,
      "step": 273100
    },
    {
      "epoch": 1.7516427061920963,
      "grad_norm": 0.06555477529764175,
      "learning_rate": 4.6240333735121765e-07,
      "loss": 0.0009,
      "step": 273110
    },
    {
      "epoch": 1.7517068430858824,
      "grad_norm": 0.028044844046235085,
      "learning_rate": 4.621682845931064e-07,
      "loss": 0.0039,
      "step": 273120
    },
    {
      "epoch": 1.7517709799796686,
      "grad_norm": 0.06414242833852768,
      "learning_rate": 4.619332886972699e-07,
      "loss": 0.0009,
      "step": 273130
    },
    {
      "epoch": 1.7518351168734547,
      "grad_norm": 0.29313191771507263,
      "learning_rate": 4.6169834966665326e-07,
      "loss": 0.0013,
      "step": 273140
    },
    {
      "epoch": 1.7518992537672409,
      "grad_norm": 0.2261309027671814,
      "learning_rate": 4.6146346750419957e-07,
      "loss": 0.0013,
      "step": 273150
    },
    {
      "epoch": 1.751963390661027,
      "grad_norm": 0.0017110253684222698,
      "learning_rate": 4.6122864221285144e-07,
      "loss": 0.0011,
      "step": 273160
    },
    {
      "epoch": 1.752027527554813,
      "grad_norm": 0.11952899396419525,
      "learning_rate": 4.6099387379555327e-07,
      "loss": 0.0017,
      "step": 273170
    },
    {
      "epoch": 1.752091664448599,
      "grad_norm": 0.026024747639894485,
      "learning_rate": 4.607591622552454e-07,
      "loss": 0.0005,
      "step": 273180
    },
    {
      "epoch": 1.752155801342385,
      "grad_norm": 0.022143952548503876,
      "learning_rate": 4.6052450759487e-07,
      "loss": 0.0012,
      "step": 273190
    },
    {
      "epoch": 1.7522199382361712,
      "grad_norm": 0.049013178795576096,
      "learning_rate": 4.6028990981736577e-07,
      "loss": 0.0023,
      "step": 273200
    },
    {
      "epoch": 1.7522840751299573,
      "grad_norm": 0.05707814171910286,
      "learning_rate": 4.600553689256737e-07,
      "loss": 0.0027,
      "step": 273210
    },
    {
      "epoch": 1.7523482120237435,
      "grad_norm": 0.23792365193367004,
      "learning_rate": 4.5982088492273316e-07,
      "loss": 0.001,
      "step": 273220
    },
    {
      "epoch": 1.7524123489175296,
      "grad_norm": 0.09483388811349869,
      "learning_rate": 4.595864578114806e-07,
      "loss": 0.0013,
      "step": 273230
    },
    {
      "epoch": 1.7524764858113158,
      "grad_norm": 0.0022677434608340263,
      "learning_rate": 4.5935208759485593e-07,
      "loss": 0.0006,
      "step": 273240
    },
    {
      "epoch": 1.752540622705102,
      "grad_norm": 0.04843544960021973,
      "learning_rate": 4.5911777427579464e-07,
      "loss": 0.0014,
      "step": 273250
    },
    {
      "epoch": 1.7526047595988878,
      "grad_norm": 0.0260971337556839,
      "learning_rate": 4.5888351785723264e-07,
      "loss": 0.0011,
      "step": 273260
    },
    {
      "epoch": 1.752668896492674,
      "grad_norm": 0.04305571690201759,
      "learning_rate": 4.586493183421048e-07,
      "loss": 0.0025,
      "step": 273270
    },
    {
      "epoch": 1.75273303338646,
      "grad_norm": 0.011727412231266499,
      "learning_rate": 4.584151757333477e-07,
      "loss": 0.0007,
      "step": 273280
    },
    {
      "epoch": 1.752797170280246,
      "grad_norm": 0.07655268907546997,
      "learning_rate": 4.58181090033894e-07,
      "loss": 0.0008,
      "step": 273290
    },
    {
      "epoch": 1.7528613071740322,
      "grad_norm": 0.10501454025506973,
      "learning_rate": 4.57947061246678e-07,
      "loss": 0.0009,
      "step": 273300
    },
    {
      "epoch": 1.7529254440678184,
      "grad_norm": 0.016707416623830795,
      "learning_rate": 4.5771308937462957e-07,
      "loss": 0.0017,
      "step": 273310
    },
    {
      "epoch": 1.7529895809616045,
      "grad_norm": 0.03645702451467514,
      "learning_rate": 4.574791744206841e-07,
      "loss": 0.001,
      "step": 273320
    },
    {
      "epoch": 1.7530537178553907,
      "grad_norm": 0.09454400092363358,
      "learning_rate": 4.5724531638777105e-07,
      "loss": 0.0007,
      "step": 273330
    },
    {
      "epoch": 1.7531178547491768,
      "grad_norm": 0.05410655215382576,
      "learning_rate": 4.5701151527882014e-07,
      "loss": 0.0008,
      "step": 273340
    },
    {
      "epoch": 1.7531819916429627,
      "grad_norm": 0.040957093238830566,
      "learning_rate": 4.5677777109676134e-07,
      "loss": 0.0014,
      "step": 273350
    },
    {
      "epoch": 1.7532461285367489,
      "grad_norm": 0.04725410416722298,
      "learning_rate": 4.5654408384452507e-07,
      "loss": 0.0009,
      "step": 273360
    },
    {
      "epoch": 1.7533102654305348,
      "grad_norm": 0.043523017317056656,
      "learning_rate": 4.5631045352503844e-07,
      "loss": 0.004,
      "step": 273370
    },
    {
      "epoch": 1.753374402324321,
      "grad_norm": 0.04890415817499161,
      "learning_rate": 4.560768801412285e-07,
      "loss": 0.0015,
      "step": 273380
    },
    {
      "epoch": 1.753438539218107,
      "grad_norm": 0.12464313954114914,
      "learning_rate": 4.5584336369602355e-07,
      "loss": 0.0012,
      "step": 273390
    },
    {
      "epoch": 1.7535026761118933,
      "grad_norm": 0.07013463228940964,
      "learning_rate": 4.556099041923484e-07,
      "loss": 0.0015,
      "step": 273400
    },
    {
      "epoch": 1.7535668130056794,
      "grad_norm": 0.09630367159843445,
      "learning_rate": 4.553765016331296e-07,
      "loss": 0.0013,
      "step": 273410
    },
    {
      "epoch": 1.7536309498994656,
      "grad_norm": 0.018023649230599403,
      "learning_rate": 4.551431560212899e-07,
      "loss": 0.0008,
      "step": 273420
    },
    {
      "epoch": 1.7536950867932515,
      "grad_norm": 0.07699010521173477,
      "learning_rate": 4.5490986735975573e-07,
      "loss": 0.0006,
      "step": 273430
    },
    {
      "epoch": 1.7537592236870376,
      "grad_norm": 0.02094586379826069,
      "learning_rate": 4.5467663565144924e-07,
      "loss": 0.0018,
      "step": 273440
    },
    {
      "epoch": 1.7538233605808236,
      "grad_norm": 0.10399764031171799,
      "learning_rate": 4.544434608992931e-07,
      "loss": 0.0015,
      "step": 273450
    },
    {
      "epoch": 1.7538874974746097,
      "grad_norm": 0.013622358441352844,
      "learning_rate": 4.5421034310620784e-07,
      "loss": 0.0016,
      "step": 273460
    },
    {
      "epoch": 1.7539516343683959,
      "grad_norm": 0.04761362448334694,
      "learning_rate": 4.5397728227511596e-07,
      "loss": 0.0005,
      "step": 273470
    },
    {
      "epoch": 1.754015771262182,
      "grad_norm": 0.003122537164017558,
      "learning_rate": 4.537442784089391e-07,
      "loss": 0.001,
      "step": 273480
    },
    {
      "epoch": 1.7540799081559681,
      "grad_norm": 0.31933242082595825,
      "learning_rate": 4.535113315105949e-07,
      "loss": 0.0015,
      "step": 273490
    },
    {
      "epoch": 1.7541440450497543,
      "grad_norm": 0.09475366771221161,
      "learning_rate": 4.532784415830027e-07,
      "loss": 0.001,
      "step": 273500
    },
    {
      "epoch": 1.7542081819435404,
      "grad_norm": 0.1212468072772026,
      "learning_rate": 4.530456086290813e-07,
      "loss": 0.0012,
      "step": 273510
    },
    {
      "epoch": 1.7542723188373264,
      "grad_norm": 0.005724911577999592,
      "learning_rate": 4.528128326517489e-07,
      "loss": 0.0005,
      "step": 273520
    },
    {
      "epoch": 1.7543364557311125,
      "grad_norm": 0.027791861444711685,
      "learning_rate": 4.5258011365392096e-07,
      "loss": 0.0006,
      "step": 273530
    },
    {
      "epoch": 1.7544005926248984,
      "grad_norm": 0.09632658958435059,
      "learning_rate": 4.5234745163851346e-07,
      "loss": 0.0011,
      "step": 273540
    },
    {
      "epoch": 1.7544647295186846,
      "grad_norm": 0.08549169450998306,
      "learning_rate": 4.521148466084435e-07,
      "loss": 0.0007,
      "step": 273550
    },
    {
      "epoch": 1.7545288664124707,
      "grad_norm": 0.1568332314491272,
      "learning_rate": 4.5188229856662493e-07,
      "loss": 0.0008,
      "step": 273560
    },
    {
      "epoch": 1.754593003306257,
      "grad_norm": 0.09158727526664734,
      "learning_rate": 4.5164980751597143e-07,
      "loss": 0.0009,
      "step": 273570
    },
    {
      "epoch": 1.754657140200043,
      "grad_norm": 0.06931840628385544,
      "learning_rate": 4.5141737345939573e-07,
      "loss": 0.0009,
      "step": 273580
    },
    {
      "epoch": 1.7547212770938292,
      "grad_norm": 0.00592627702280879,
      "learning_rate": 4.5118499639981216e-07,
      "loss": 0.0007,
      "step": 273590
    },
    {
      "epoch": 1.7547854139876151,
      "grad_norm": 0.15602008998394012,
      "learning_rate": 4.509526763401312e-07,
      "loss": 0.0013,
      "step": 273600
    },
    {
      "epoch": 1.7548495508814013,
      "grad_norm": 0.01364865992218256,
      "learning_rate": 4.507204132832649e-07,
      "loss": 0.001,
      "step": 273610
    },
    {
      "epoch": 1.7549136877751874,
      "grad_norm": 0.020498182624578476,
      "learning_rate": 4.504882072321215e-07,
      "loss": 0.0012,
      "step": 273620
    },
    {
      "epoch": 1.7549778246689733,
      "grad_norm": 0.046825554221868515,
      "learning_rate": 4.5025605818961373e-07,
      "loss": 0.0014,
      "step": 273630
    },
    {
      "epoch": 1.7550419615627595,
      "grad_norm": 0.11790932714939117,
      "learning_rate": 4.500239661586492e-07,
      "loss": 0.001,
      "step": 273640
    },
    {
      "epoch": 1.7551060984565456,
      "grad_norm": 0.018157294020056725,
      "learning_rate": 4.4979193114213506e-07,
      "loss": 0.0015,
      "step": 273650
    },
    {
      "epoch": 1.7551702353503318,
      "grad_norm": 0.06198253110051155,
      "learning_rate": 4.4955995314298117e-07,
      "loss": 0.0038,
      "step": 273660
    },
    {
      "epoch": 1.755234372244118,
      "grad_norm": 0.22804534435272217,
      "learning_rate": 4.493280321640936e-07,
      "loss": 0.0015,
      "step": 273670
    },
    {
      "epoch": 1.755298509137904,
      "grad_norm": 0.0053673069924116135,
      "learning_rate": 4.490961682083772e-07,
      "loss": 0.0006,
      "step": 273680
    },
    {
      "epoch": 1.75536264603169,
      "grad_norm": 0.04484537988901138,
      "learning_rate": 4.48864361278738e-07,
      "loss": 0.0008,
      "step": 273690
    },
    {
      "epoch": 1.7554267829254762,
      "grad_norm": 0.16247770190238953,
      "learning_rate": 4.48632611378082e-07,
      "loss": 0.0011,
      "step": 273700
    },
    {
      "epoch": 1.755490919819262,
      "grad_norm": 0.07464831322431564,
      "learning_rate": 4.4840091850931187e-07,
      "loss": 0.0031,
      "step": 273710
    },
    {
      "epoch": 1.7555550567130482,
      "grad_norm": 0.04992212727665901,
      "learning_rate": 4.4816928267533144e-07,
      "loss": 0.0033,
      "step": 273720
    },
    {
      "epoch": 1.7556191936068344,
      "grad_norm": 0.04683858901262283,
      "learning_rate": 4.479377038790417e-07,
      "loss": 0.0006,
      "step": 273730
    },
    {
      "epoch": 1.7556833305006205,
      "grad_norm": 0.03431175649166107,
      "learning_rate": 4.4770618212334695e-07,
      "loss": 0.0013,
      "step": 273740
    },
    {
      "epoch": 1.7557474673944067,
      "grad_norm": 0.00959078874439001,
      "learning_rate": 4.4747471741114713e-07,
      "loss": 0.0006,
      "step": 273750
    },
    {
      "epoch": 1.7558116042881928,
      "grad_norm": 0.3120195269584656,
      "learning_rate": 4.4724330974534327e-07,
      "loss": 0.0016,
      "step": 273760
    },
    {
      "epoch": 1.755875741181979,
      "grad_norm": 0.20491768419742584,
      "learning_rate": 4.47011959128833e-07,
      "loss": 0.0007,
      "step": 273770
    },
    {
      "epoch": 1.755939878075765,
      "grad_norm": 0.06042253226041794,
      "learning_rate": 4.4678066556451793e-07,
      "loss": 0.0006,
      "step": 273780
    },
    {
      "epoch": 1.756004014969551,
      "grad_norm": 0.08034548163414001,
      "learning_rate": 4.4654942905529577e-07,
      "loss": 0.0009,
      "step": 273790
    },
    {
      "epoch": 1.756068151863337,
      "grad_norm": 0.039186157286167145,
      "learning_rate": 4.463182496040619e-07,
      "loss": 0.0014,
      "step": 273800
    },
    {
      "epoch": 1.7561322887571231,
      "grad_norm": 0.13175533711910248,
      "learning_rate": 4.4608712721371626e-07,
      "loss": 0.0008,
      "step": 273810
    },
    {
      "epoch": 1.7561964256509093,
      "grad_norm": 0.09573426097631454,
      "learning_rate": 4.4585606188715325e-07,
      "loss": 0.0022,
      "step": 273820
    },
    {
      "epoch": 1.7562605625446954,
      "grad_norm": 0.09324001520872116,
      "learning_rate": 4.456250536272688e-07,
      "loss": 0.001,
      "step": 273830
    },
    {
      "epoch": 1.7563246994384816,
      "grad_norm": 0.17717893421649933,
      "learning_rate": 4.4539410243695624e-07,
      "loss": 0.0009,
      "step": 273840
    },
    {
      "epoch": 1.7563888363322677,
      "grad_norm": 0.06860344111919403,
      "learning_rate": 4.451632083191121e-07,
      "loss": 0.0028,
      "step": 273850
    },
    {
      "epoch": 1.7564529732260536,
      "grad_norm": 0.07243098318576813,
      "learning_rate": 4.449323712766279e-07,
      "loss": 0.0023,
      "step": 273860
    },
    {
      "epoch": 1.7565171101198398,
      "grad_norm": 0.1243899017572403,
      "learning_rate": 4.4470159131239645e-07,
      "loss": 0.0009,
      "step": 273870
    },
    {
      "epoch": 1.7565812470136257,
      "grad_norm": 0.10625339299440384,
      "learning_rate": 4.444708684293086e-07,
      "loss": 0.0012,
      "step": 273880
    },
    {
      "epoch": 1.7566453839074119,
      "grad_norm": 0.02705708146095276,
      "learning_rate": 4.442402026302578e-07,
      "loss": 0.0011,
      "step": 273890
    },
    {
      "epoch": 1.756709520801198,
      "grad_norm": 0.008359714411199093,
      "learning_rate": 4.4400959391813323e-07,
      "loss": 0.0005,
      "step": 273900
    },
    {
      "epoch": 1.7567736576949842,
      "grad_norm": 0.006910772528499365,
      "learning_rate": 4.4377904229582434e-07,
      "loss": 0.0014,
      "step": 273910
    },
    {
      "epoch": 1.7568377945887703,
      "grad_norm": 0.12756577134132385,
      "learning_rate": 4.435485477662199e-07,
      "loss": 0.0021,
      "step": 273920
    },
    {
      "epoch": 1.7569019314825565,
      "grad_norm": 0.2769004702568054,
      "learning_rate": 4.433181103322093e-07,
      "loss": 0.001,
      "step": 273930
    },
    {
      "epoch": 1.7569660683763426,
      "grad_norm": 0.0014730626717209816,
      "learning_rate": 4.4308772999667905e-07,
      "loss": 0.0009,
      "step": 273940
    },
    {
      "epoch": 1.7570302052701285,
      "grad_norm": 0.0016581410309299827,
      "learning_rate": 4.428574067625158e-07,
      "loss": 0.0011,
      "step": 273950
    },
    {
      "epoch": 1.7570943421639147,
      "grad_norm": 0.26757508516311646,
      "learning_rate": 4.426271406326066e-07,
      "loss": 0.0026,
      "step": 273960
    },
    {
      "epoch": 1.7571584790577006,
      "grad_norm": 0.01904025487601757,
      "learning_rate": 4.4239693160983696e-07,
      "loss": 0.0014,
      "step": 273970
    },
    {
      "epoch": 1.7572226159514868,
      "grad_norm": 0.04785741865634918,
      "learning_rate": 4.4216677969709076e-07,
      "loss": 0.0013,
      "step": 273980
    },
    {
      "epoch": 1.757286752845273,
      "grad_norm": 0.12253019213676453,
      "learning_rate": 4.419366848972517e-07,
      "loss": 0.0012,
      "step": 273990
    },
    {
      "epoch": 1.757350889739059,
      "grad_norm": 0.1177787259221077,
      "learning_rate": 4.4170664721320424e-07,
      "loss": 0.0013,
      "step": 274000
    },
    {
      "epoch": 1.7574150266328452,
      "grad_norm": 0.10684992372989655,
      "learning_rate": 4.4147666664782984e-07,
      "loss": 0.0008,
      "step": 274010
    },
    {
      "epoch": 1.7574791635266314,
      "grad_norm": 0.012389772571623325,
      "learning_rate": 4.412467432040113e-07,
      "loss": 0.0015,
      "step": 274020
    },
    {
      "epoch": 1.7575433004204173,
      "grad_norm": 0.036616671830415726,
      "learning_rate": 4.4101687688462793e-07,
      "loss": 0.0008,
      "step": 274030
    },
    {
      "epoch": 1.7576074373142034,
      "grad_norm": 0.03530687093734741,
      "learning_rate": 4.4078706769256243e-07,
      "loss": 0.002,
      "step": 274040
    },
    {
      "epoch": 1.7576715742079896,
      "grad_norm": 0.09993518888950348,
      "learning_rate": 4.405573156306936e-07,
      "loss": 0.0015,
      "step": 274050
    },
    {
      "epoch": 1.7577357111017755,
      "grad_norm": 0.07789759337902069,
      "learning_rate": 4.4032762070189917e-07,
      "loss": 0.0012,
      "step": 274060
    },
    {
      "epoch": 1.7577998479955617,
      "grad_norm": 0.11408476531505585,
      "learning_rate": 4.40097982909059e-07,
      "loss": 0.0005,
      "step": 274070
    },
    {
      "epoch": 1.7578639848893478,
      "grad_norm": 0.07201806455850601,
      "learning_rate": 4.3986840225505036e-07,
      "loss": 0.0009,
      "step": 274080
    },
    {
      "epoch": 1.757928121783134,
      "grad_norm": 0.1237541139125824,
      "learning_rate": 4.3963887874274915e-07,
      "loss": 0.001,
      "step": 274090
    },
    {
      "epoch": 1.75799225867692,
      "grad_norm": 0.05876180902123451,
      "learning_rate": 4.3940941237503143e-07,
      "loss": 0.0008,
      "step": 274100
    },
    {
      "epoch": 1.7580563955707063,
      "grad_norm": 0.030227554962038994,
      "learning_rate": 4.3918000315477436e-07,
      "loss": 0.001,
      "step": 274110
    },
    {
      "epoch": 1.7581205324644922,
      "grad_norm": 0.018378805369138718,
      "learning_rate": 4.3895065108485124e-07,
      "loss": 0.0011,
      "step": 274120
    },
    {
      "epoch": 1.7581846693582783,
      "grad_norm": 0.006053559482097626,
      "learning_rate": 4.387213561681364e-07,
      "loss": 0.0008,
      "step": 274130
    },
    {
      "epoch": 1.7582488062520643,
      "grad_norm": 0.07609471678733826,
      "learning_rate": 4.3849211840750137e-07,
      "loss": 0.0013,
      "step": 274140
    },
    {
      "epoch": 1.7583129431458504,
      "grad_norm": 0.13260923326015472,
      "learning_rate": 4.3826293780582175e-07,
      "loss": 0.0007,
      "step": 274150
    },
    {
      "epoch": 1.7583770800396366,
      "grad_norm": 0.05366222932934761,
      "learning_rate": 4.380338143659674e-07,
      "loss": 0.001,
      "step": 274160
    },
    {
      "epoch": 1.7584412169334227,
      "grad_norm": 0.0023712320253252983,
      "learning_rate": 4.3780474809080985e-07,
      "loss": 0.0012,
      "step": 274170
    },
    {
      "epoch": 1.7585053538272089,
      "grad_norm": 0.10853615403175354,
      "learning_rate": 4.3757573898321803e-07,
      "loss": 0.0013,
      "step": 274180
    },
    {
      "epoch": 1.758569490720995,
      "grad_norm": 0.046302780508995056,
      "learning_rate": 4.373467870460646e-07,
      "loss": 0.0014,
      "step": 274190
    },
    {
      "epoch": 1.7586336276147811,
      "grad_norm": 0.045307353138923645,
      "learning_rate": 4.3711789228221615e-07,
      "loss": 0.0016,
      "step": 274200
    },
    {
      "epoch": 1.758697764508567,
      "grad_norm": 0.08859158307313919,
      "learning_rate": 4.368890546945409e-07,
      "loss": 0.0011,
      "step": 274210
    },
    {
      "epoch": 1.7587619014023532,
      "grad_norm": 0.015490715391933918,
      "learning_rate": 4.3666027428590775e-07,
      "loss": 0.0006,
      "step": 274220
    },
    {
      "epoch": 1.7588260382961391,
      "grad_norm": 0.002372288377955556,
      "learning_rate": 4.364315510591821e-07,
      "loss": 0.0008,
      "step": 274230
    },
    {
      "epoch": 1.7588901751899253,
      "grad_norm": 0.11505340039730072,
      "learning_rate": 4.362028850172312e-07,
      "loss": 0.0013,
      "step": 274240
    },
    {
      "epoch": 1.7589543120837114,
      "grad_norm": 0.006366999354213476,
      "learning_rate": 4.3597427616292044e-07,
      "loss": 0.0012,
      "step": 274250
    },
    {
      "epoch": 1.7590184489774976,
      "grad_norm": 0.0795503780245781,
      "learning_rate": 4.3574572449911254e-07,
      "loss": 0.0022,
      "step": 274260
    },
    {
      "epoch": 1.7590825858712837,
      "grad_norm": 0.0016061549540609121,
      "learning_rate": 4.3551723002867364e-07,
      "loss": 0.0011,
      "step": 274270
    },
    {
      "epoch": 1.75914672276507,
      "grad_norm": 0.016629310324788094,
      "learning_rate": 4.352887927544658e-07,
      "loss": 0.0013,
      "step": 274280
    },
    {
      "epoch": 1.7592108596588558,
      "grad_norm": 0.3661215007305145,
      "learning_rate": 4.350604126793523e-07,
      "loss": 0.0012,
      "step": 274290
    },
    {
      "epoch": 1.759274996552642,
      "grad_norm": 0.11058757454156876,
      "learning_rate": 4.348320898061931e-07,
      "loss": 0.0014,
      "step": 274300
    },
    {
      "epoch": 1.759339133446428,
      "grad_norm": 0.0846678763628006,
      "learning_rate": 4.346038241378514e-07,
      "loss": 0.0009,
      "step": 274310
    },
    {
      "epoch": 1.759403270340214,
      "grad_norm": 0.09430932998657227,
      "learning_rate": 4.3437561567718667e-07,
      "loss": 0.0016,
      "step": 274320
    },
    {
      "epoch": 1.7594674072340002,
      "grad_norm": 0.20638717710971832,
      "learning_rate": 4.3414746442705767e-07,
      "loss": 0.0013,
      "step": 274330
    },
    {
      "epoch": 1.7595315441277863,
      "grad_norm": 0.1938396543264389,
      "learning_rate": 4.3391937039032496e-07,
      "loss": 0.0011,
      "step": 274340
    },
    {
      "epoch": 1.7595956810215725,
      "grad_norm": 0.05062471330165863,
      "learning_rate": 4.3369133356984563e-07,
      "loss": 0.0023,
      "step": 274350
    },
    {
      "epoch": 1.7596598179153586,
      "grad_norm": 0.13887950778007507,
      "learning_rate": 4.3346335396847736e-07,
      "loss": 0.0006,
      "step": 274360
    },
    {
      "epoch": 1.7597239548091448,
      "grad_norm": 0.09052564948797226,
      "learning_rate": 4.3323543158907635e-07,
      "loss": 0.0003,
      "step": 274370
    },
    {
      "epoch": 1.7597880917029307,
      "grad_norm": 0.008419949561357498,
      "learning_rate": 4.3300756643450016e-07,
      "loss": 0.0004,
      "step": 274380
    },
    {
      "epoch": 1.7598522285967169,
      "grad_norm": 0.0025691508781164885,
      "learning_rate": 4.3277975850760266e-07,
      "loss": 0.0004,
      "step": 274390
    },
    {
      "epoch": 1.7599163654905028,
      "grad_norm": 0.04689428210258484,
      "learning_rate": 4.325520078112394e-07,
      "loss": 0.0009,
      "step": 274400
    },
    {
      "epoch": 1.759980502384289,
      "grad_norm": 0.04605294391512871,
      "learning_rate": 4.323243143482625e-07,
      "loss": 0.0015,
      "step": 274410
    },
    {
      "epoch": 1.760044639278075,
      "grad_norm": 0.03803924471139908,
      "learning_rate": 4.3209667812152745e-07,
      "loss": 0.0007,
      "step": 274420
    },
    {
      "epoch": 1.7601087761718612,
      "grad_norm": 0.041317418217659,
      "learning_rate": 4.3186909913388587e-07,
      "loss": 0.0014,
      "step": 274430
    },
    {
      "epoch": 1.7601729130656474,
      "grad_norm": 0.050265710800886154,
      "learning_rate": 4.3164157738818935e-07,
      "loss": 0.001,
      "step": 274440
    },
    {
      "epoch": 1.7602370499594335,
      "grad_norm": 0.026506226509809494,
      "learning_rate": 4.3141411288728783e-07,
      "loss": 0.0005,
      "step": 274450
    },
    {
      "epoch": 1.7603011868532195,
      "grad_norm": 0.07355380058288574,
      "learning_rate": 4.3118670563403296e-07,
      "loss": 0.0011,
      "step": 274460
    },
    {
      "epoch": 1.7603653237470056,
      "grad_norm": 0.04049091786146164,
      "learning_rate": 4.309593556312747e-07,
      "loss": 0.0013,
      "step": 274470
    },
    {
      "epoch": 1.7604294606407918,
      "grad_norm": 0.02454739809036255,
      "learning_rate": 4.3073206288185956e-07,
      "loss": 0.0028,
      "step": 274480
    },
    {
      "epoch": 1.7604935975345777,
      "grad_norm": 0.0465393029153347,
      "learning_rate": 4.3050482738863866e-07,
      "loss": 0.0007,
      "step": 274490
    },
    {
      "epoch": 1.7605577344283638,
      "grad_norm": 0.15284514427185059,
      "learning_rate": 4.30277649154458e-07,
      "loss": 0.0007,
      "step": 274500
    },
    {
      "epoch": 1.76062187132215,
      "grad_norm": 0.1650029420852661,
      "learning_rate": 4.300505281821643e-07,
      "loss": 0.001,
      "step": 274510
    },
    {
      "epoch": 1.7606860082159361,
      "grad_norm": 0.022082634270191193,
      "learning_rate": 4.2982346447460245e-07,
      "loss": 0.0002,
      "step": 274520
    },
    {
      "epoch": 1.7607501451097223,
      "grad_norm": 0.0654936209321022,
      "learning_rate": 4.2959645803462014e-07,
      "loss": 0.0007,
      "step": 274530
    },
    {
      "epoch": 1.7608142820035084,
      "grad_norm": 0.13372991979122162,
      "learning_rate": 4.2936950886506014e-07,
      "loss": 0.0016,
      "step": 274540
    },
    {
      "epoch": 1.7608784188972944,
      "grad_norm": 0.02332022599875927,
      "learning_rate": 4.2914261696876736e-07,
      "loss": 0.0008,
      "step": 274550
    },
    {
      "epoch": 1.7609425557910805,
      "grad_norm": 0.0456223227083683,
      "learning_rate": 4.289157823485829e-07,
      "loss": 0.0008,
      "step": 274560
    },
    {
      "epoch": 1.7610066926848664,
      "grad_norm": 0.03842933103442192,
      "learning_rate": 4.2868900500735166e-07,
      "loss": 0.0011,
      "step": 274570
    },
    {
      "epoch": 1.7610708295786526,
      "grad_norm": 0.15203480422496796,
      "learning_rate": 4.284622849479142e-07,
      "loss": 0.0023,
      "step": 274580
    },
    {
      "epoch": 1.7611349664724387,
      "grad_norm": 0.06362316757440567,
      "learning_rate": 4.2823562217311154e-07,
      "loss": 0.0008,
      "step": 274590
    },
    {
      "epoch": 1.7611991033662249,
      "grad_norm": 0.03955472260713577,
      "learning_rate": 4.280090166857831e-07,
      "loss": 0.0009,
      "step": 274600
    },
    {
      "epoch": 1.761263240260011,
      "grad_norm": 0.12518471479415894,
      "learning_rate": 4.2778246848877045e-07,
      "loss": 0.0011,
      "step": 274610
    },
    {
      "epoch": 1.7613273771537972,
      "grad_norm": 0.023512301966547966,
      "learning_rate": 4.2755597758491084e-07,
      "loss": 0.0006,
      "step": 274620
    },
    {
      "epoch": 1.7613915140475833,
      "grad_norm": 0.007494448684155941,
      "learning_rate": 4.2732954397704196e-07,
      "loss": 0.0013,
      "step": 274630
    },
    {
      "epoch": 1.7614556509413692,
      "grad_norm": 0.3269810378551483,
      "learning_rate": 4.2710316766800264e-07,
      "loss": 0.0012,
      "step": 274640
    },
    {
      "epoch": 1.7615197878351554,
      "grad_norm": 0.017750240862369537,
      "learning_rate": 4.2687684866062897e-07,
      "loss": 0.0007,
      "step": 274650
    },
    {
      "epoch": 1.7615839247289413,
      "grad_norm": 0.010823042131960392,
      "learning_rate": 4.266505869577564e-07,
      "loss": 0.0012,
      "step": 274660
    },
    {
      "epoch": 1.7616480616227275,
      "grad_norm": 0.019104070961475372,
      "learning_rate": 4.2642438256221996e-07,
      "loss": 0.0013,
      "step": 274670
    },
    {
      "epoch": 1.7617121985165136,
      "grad_norm": 0.001704822527244687,
      "learning_rate": 4.2619823547685567e-07,
      "loss": 0.0019,
      "step": 274680
    },
    {
      "epoch": 1.7617763354102998,
      "grad_norm": 0.10967900604009628,
      "learning_rate": 4.259721457044963e-07,
      "loss": 0.0012,
      "step": 274690
    },
    {
      "epoch": 1.761840472304086,
      "grad_norm": 0.04793315753340721,
      "learning_rate": 4.2574611324797455e-07,
      "loss": 0.0015,
      "step": 274700
    },
    {
      "epoch": 1.761904609197872,
      "grad_norm": 0.1262887865304947,
      "learning_rate": 4.255201381101226e-07,
      "loss": 0.0013,
      "step": 274710
    },
    {
      "epoch": 1.761968746091658,
      "grad_norm": 0.0792112872004509,
      "learning_rate": 4.2529422029377376e-07,
      "loss": 0.0012,
      "step": 274720
    },
    {
      "epoch": 1.7620328829854441,
      "grad_norm": 0.16655921936035156,
      "learning_rate": 4.250683598017574e-07,
      "loss": 0.0011,
      "step": 274730
    },
    {
      "epoch": 1.76209701987923,
      "grad_norm": 0.10308097302913666,
      "learning_rate": 4.2484255663690466e-07,
      "loss": 0.0008,
      "step": 274740
    },
    {
      "epoch": 1.7621611567730162,
      "grad_norm": 0.0619034506380558,
      "learning_rate": 4.246168108020432e-07,
      "loss": 0.001,
      "step": 274750
    },
    {
      "epoch": 1.7622252936668024,
      "grad_norm": 0.06137575954198837,
      "learning_rate": 4.2439112230000414e-07,
      "loss": 0.0006,
      "step": 274760
    },
    {
      "epoch": 1.7622894305605885,
      "grad_norm": 0.021134236827492714,
      "learning_rate": 4.2416549113361407e-07,
      "loss": 0.0009,
      "step": 274770
    },
    {
      "epoch": 1.7623535674543747,
      "grad_norm": 0.08857366442680359,
      "learning_rate": 4.2393991730570016e-07,
      "loss": 0.0008,
      "step": 274780
    },
    {
      "epoch": 1.7624177043481608,
      "grad_norm": 0.12476522475481033,
      "learning_rate": 4.2371440081909023e-07,
      "loss": 0.0006,
      "step": 274790
    },
    {
      "epoch": 1.762481841241947,
      "grad_norm": 0.017562726512551308,
      "learning_rate": 4.234889416766097e-07,
      "loss": 0.0009,
      "step": 274800
    },
    {
      "epoch": 1.7625459781357329,
      "grad_norm": 0.010540449991822243,
      "learning_rate": 4.232635398810836e-07,
      "loss": 0.0012,
      "step": 274810
    },
    {
      "epoch": 1.762610115029519,
      "grad_norm": 0.03732389584183693,
      "learning_rate": 4.230381954353352e-07,
      "loss": 0.0007,
      "step": 274820
    },
    {
      "epoch": 1.762674251923305,
      "grad_norm": 0.1333763152360916,
      "learning_rate": 4.2281290834219e-07,
      "loss": 0.0013,
      "step": 274830
    },
    {
      "epoch": 1.762738388817091,
      "grad_norm": 0.039669591933488846,
      "learning_rate": 4.225876786044708e-07,
      "loss": 0.0008,
      "step": 274840
    },
    {
      "epoch": 1.7628025257108773,
      "grad_norm": 0.11995638161897659,
      "learning_rate": 4.223625062249992e-07,
      "loss": 0.0013,
      "step": 274850
    },
    {
      "epoch": 1.7628666626046634,
      "grad_norm": 0.04716256260871887,
      "learning_rate": 4.2213739120659626e-07,
      "loss": 0.0006,
      "step": 274860
    },
    {
      "epoch": 1.7629307994984496,
      "grad_norm": 0.15836253762245178,
      "learning_rate": 4.2191233355208415e-07,
      "loss": 0.0008,
      "step": 274870
    },
    {
      "epoch": 1.7629949363922357,
      "grad_norm": 0.07259972393512726,
      "learning_rate": 4.2168733326428235e-07,
      "loss": 0.0008,
      "step": 274880
    },
    {
      "epoch": 1.7630590732860218,
      "grad_norm": 0.05724453553557396,
      "learning_rate": 4.214623903460108e-07,
      "loss": 0.0008,
      "step": 274890
    },
    {
      "epoch": 1.7631232101798078,
      "grad_norm": 0.06909746676683426,
      "learning_rate": 4.2123750480008607e-07,
      "loss": 0.0006,
      "step": 274900
    },
    {
      "epoch": 1.763187347073594,
      "grad_norm": 0.18209044635295868,
      "learning_rate": 4.2101267662932934e-07,
      "loss": 0.0015,
      "step": 274910
    },
    {
      "epoch": 1.7632514839673799,
      "grad_norm": 0.07382436841726303,
      "learning_rate": 4.20787905836556e-07,
      "loss": 0.0006,
      "step": 274920
    },
    {
      "epoch": 1.763315620861166,
      "grad_norm": 0.06956379115581512,
      "learning_rate": 4.2056319242458235e-07,
      "loss": 0.0008,
      "step": 274930
    },
    {
      "epoch": 1.7633797577549521,
      "grad_norm": 0.0836181566119194,
      "learning_rate": 4.2033853639622534e-07,
      "loss": 0.001,
      "step": 274940
    },
    {
      "epoch": 1.7634438946487383,
      "grad_norm": 0.1506095826625824,
      "learning_rate": 4.201139377543001e-07,
      "loss": 0.0011,
      "step": 274950
    },
    {
      "epoch": 1.7635080315425244,
      "grad_norm": 0.01491470541805029,
      "learning_rate": 4.1988939650161986e-07,
      "loss": 0.001,
      "step": 274960
    },
    {
      "epoch": 1.7635721684363106,
      "grad_norm": 0.12581266462802887,
      "learning_rate": 4.1966491264099797e-07,
      "loss": 0.0019,
      "step": 274970
    },
    {
      "epoch": 1.7636363053300965,
      "grad_norm": 0.012968640774488449,
      "learning_rate": 4.1944048617524934e-07,
      "loss": 0.0009,
      "step": 274980
    },
    {
      "epoch": 1.7637004422238827,
      "grad_norm": 0.16107068955898285,
      "learning_rate": 4.1921611710718404e-07,
      "loss": 0.0017,
      "step": 274990
    },
    {
      "epoch": 1.7637645791176686,
      "grad_norm": 0.0013026761589571834,
      "learning_rate": 4.1899180543961524e-07,
      "loss": 0.0026,
      "step": 275000
    },
    {
      "epoch": 1.7638287160114547,
      "grad_norm": 0.005042532924562693,
      "learning_rate": 4.187675511753536e-07,
      "loss": 0.0018,
      "step": 275010
    },
    {
      "epoch": 1.763892852905241,
      "grad_norm": 0.32132092118263245,
      "learning_rate": 4.1854335431720794e-07,
      "loss": 0.0013,
      "step": 275020
    },
    {
      "epoch": 1.763956989799027,
      "grad_norm": 0.05832437425851822,
      "learning_rate": 4.1831921486798876e-07,
      "loss": 0.0009,
      "step": 275030
    },
    {
      "epoch": 1.7640211266928132,
      "grad_norm": 0.20390057563781738,
      "learning_rate": 4.180951328305044e-07,
      "loss": 0.0009,
      "step": 275040
    },
    {
      "epoch": 1.7640852635865993,
      "grad_norm": 0.044995591044425964,
      "learning_rate": 4.1787110820756205e-07,
      "loss": 0.0005,
      "step": 275050
    },
    {
      "epoch": 1.7641494004803855,
      "grad_norm": 0.08996668457984924,
      "learning_rate": 4.1764714100197003e-07,
      "loss": 0.0008,
      "step": 275060
    },
    {
      "epoch": 1.7642135373741714,
      "grad_norm": 0.09389866888523102,
      "learning_rate": 4.174232312165344e-07,
      "loss": 0.0007,
      "step": 275070
    },
    {
      "epoch": 1.7642776742679576,
      "grad_norm": 0.0043205274268984795,
      "learning_rate": 4.171993788540613e-07,
      "loss": 0.0009,
      "step": 275080
    },
    {
      "epoch": 1.7643418111617435,
      "grad_norm": 0.11533765494823456,
      "learning_rate": 4.169755839173539e-07,
      "loss": 0.0011,
      "step": 275090
    },
    {
      "epoch": 1.7644059480555296,
      "grad_norm": 0.008159862831234932,
      "learning_rate": 4.167518464092185e-07,
      "loss": 0.0014,
      "step": 275100
    },
    {
      "epoch": 1.7644700849493158,
      "grad_norm": 0.07528112083673477,
      "learning_rate": 4.165281663324583e-07,
      "loss": 0.0011,
      "step": 275110
    },
    {
      "epoch": 1.764534221843102,
      "grad_norm": 0.04019409045577049,
      "learning_rate": 4.1630454368987605e-07,
      "loss": 0.001,
      "step": 275120
    },
    {
      "epoch": 1.764598358736888,
      "grad_norm": 0.08170922100543976,
      "learning_rate": 4.160809784842729e-07,
      "loss": 0.0013,
      "step": 275130
    },
    {
      "epoch": 1.7646624956306742,
      "grad_norm": 0.16807445883750916,
      "learning_rate": 4.158574707184521e-07,
      "loss": 0.0012,
      "step": 275140
    },
    {
      "epoch": 1.7647266325244602,
      "grad_norm": 0.019600650295615196,
      "learning_rate": 4.156340203952136e-07,
      "loss": 0.0008,
      "step": 275150
    },
    {
      "epoch": 1.7647907694182463,
      "grad_norm": 0.02643808163702488,
      "learning_rate": 4.1541062751735696e-07,
      "loss": 0.001,
      "step": 275160
    },
    {
      "epoch": 1.7648549063120325,
      "grad_norm": 0.15279154479503632,
      "learning_rate": 4.1518729208768096e-07,
      "loss": 0.0011,
      "step": 275170
    },
    {
      "epoch": 1.7649190432058184,
      "grad_norm": 0.1854482889175415,
      "learning_rate": 4.1496401410898557e-07,
      "loss": 0.0013,
      "step": 275180
    },
    {
      "epoch": 1.7649831800996045,
      "grad_norm": 0.06766447424888611,
      "learning_rate": 4.1474079358406806e-07,
      "loss": 0.0009,
      "step": 275190
    },
    {
      "epoch": 1.7650473169933907,
      "grad_norm": 0.13233855366706848,
      "learning_rate": 4.1451763051572444e-07,
      "loss": 0.0015,
      "step": 275200
    },
    {
      "epoch": 1.7651114538871768,
      "grad_norm": 0.041932497173547745,
      "learning_rate": 4.1429452490675315e-07,
      "loss": 0.0012,
      "step": 275210
    },
    {
      "epoch": 1.765175590780963,
      "grad_norm": 0.08011277765035629,
      "learning_rate": 4.1407147675994855e-07,
      "loss": 0.0004,
      "step": 275220
    },
    {
      "epoch": 1.7652397276747491,
      "grad_norm": 0.01628202199935913,
      "learning_rate": 4.138484860781061e-07,
      "loss": 0.002,
      "step": 275230
    },
    {
      "epoch": 1.765303864568535,
      "grad_norm": 0.059850651770830154,
      "learning_rate": 4.136255528640193e-07,
      "loss": 0.001,
      "step": 275240
    },
    {
      "epoch": 1.7653680014623212,
      "grad_norm": 0.029925214126706123,
      "learning_rate": 4.1340267712048245e-07,
      "loss": 0.0017,
      "step": 275250
    },
    {
      "epoch": 1.7654321383561071,
      "grad_norm": 0.06737666577100754,
      "learning_rate": 4.131798588502883e-07,
      "loss": 0.001,
      "step": 275260
    },
    {
      "epoch": 1.7654962752498933,
      "grad_norm": 0.027920279651880264,
      "learning_rate": 4.129570980562286e-07,
      "loss": 0.0009,
      "step": 275270
    },
    {
      "epoch": 1.7655604121436794,
      "grad_norm": 0.05020119249820709,
      "learning_rate": 4.1273439474109387e-07,
      "loss": 0.0008,
      "step": 275280
    },
    {
      "epoch": 1.7656245490374656,
      "grad_norm": 0.034163858741521835,
      "learning_rate": 4.125117489076763e-07,
      "loss": 0.0012,
      "step": 275290
    },
    {
      "epoch": 1.7656886859312517,
      "grad_norm": 0.04959217458963394,
      "learning_rate": 4.122891605587653e-07,
      "loss": 0.0009,
      "step": 275300
    },
    {
      "epoch": 1.7657528228250379,
      "grad_norm": 0.19557246565818787,
      "learning_rate": 4.1206662969715037e-07,
      "loss": 0.0009,
      "step": 275310
    },
    {
      "epoch": 1.765816959718824,
      "grad_norm": 0.10723380744457245,
      "learning_rate": 4.1184415632561815e-07,
      "loss": 0.0011,
      "step": 275320
    },
    {
      "epoch": 1.76588109661261,
      "grad_norm": 0.02903168648481369,
      "learning_rate": 4.116217404469586e-07,
      "loss": 0.0003,
      "step": 275330
    },
    {
      "epoch": 1.765945233506396,
      "grad_norm": 0.01404721848666668,
      "learning_rate": 4.1139938206395837e-07,
      "loss": 0.0018,
      "step": 275340
    },
    {
      "epoch": 1.766009370400182,
      "grad_norm": 0.06477724015712738,
      "learning_rate": 4.111770811794019e-07,
      "loss": 0.0014,
      "step": 275350
    },
    {
      "epoch": 1.7660735072939682,
      "grad_norm": 0.020367076620459557,
      "learning_rate": 4.10954837796077e-07,
      "loss": 0.0009,
      "step": 275360
    },
    {
      "epoch": 1.7661376441877543,
      "grad_norm": 0.03996472433209419,
      "learning_rate": 4.107326519167681e-07,
      "loss": 0.0013,
      "step": 275370
    },
    {
      "epoch": 1.7662017810815405,
      "grad_norm": 0.06612200289964676,
      "learning_rate": 4.1051052354425915e-07,
      "loss": 0.0013,
      "step": 275380
    },
    {
      "epoch": 1.7662659179753266,
      "grad_norm": 0.11471281200647354,
      "learning_rate": 4.1028845268133223e-07,
      "loss": 0.0013,
      "step": 275390
    },
    {
      "epoch": 1.7663300548691128,
      "grad_norm": 0.03563837707042694,
      "learning_rate": 4.1006643933077184e-07,
      "loss": 0.0012,
      "step": 275400
    },
    {
      "epoch": 1.7663941917628987,
      "grad_norm": 0.09462326765060425,
      "learning_rate": 4.0984448349535967e-07,
      "loss": 0.0011,
      "step": 275410
    },
    {
      "epoch": 1.7664583286566848,
      "grad_norm": 0.11228634417057037,
      "learning_rate": 4.096225851778768e-07,
      "loss": 0.0007,
      "step": 275420
    },
    {
      "epoch": 1.7665224655504708,
      "grad_norm": 0.048812948167324066,
      "learning_rate": 4.094007443811021e-07,
      "loss": 0.0006,
      "step": 275430
    },
    {
      "epoch": 1.766586602444257,
      "grad_norm": 0.0776638612151146,
      "learning_rate": 4.091789611078184e-07,
      "loss": 0.0008,
      "step": 275440
    },
    {
      "epoch": 1.766650739338043,
      "grad_norm": 0.12345968186855316,
      "learning_rate": 4.0895723536080343e-07,
      "loss": 0.0015,
      "step": 275450
    },
    {
      "epoch": 1.7667148762318292,
      "grad_norm": 0.01290800143033266,
      "learning_rate": 4.08735567142835e-07,
      "loss": 0.0012,
      "step": 275460
    },
    {
      "epoch": 1.7667790131256154,
      "grad_norm": 0.11628074944019318,
      "learning_rate": 4.085139564566909e-07,
      "loss": 0.0009,
      "step": 275470
    },
    {
      "epoch": 1.7668431500194015,
      "grad_norm": 0.014658177271485329,
      "learning_rate": 4.0829240330514885e-07,
      "loss": 0.001,
      "step": 275480
    },
    {
      "epoch": 1.7669072869131877,
      "grad_norm": 0.07219427824020386,
      "learning_rate": 4.0807090769098447e-07,
      "loss": 0.0012,
      "step": 275490
    },
    {
      "epoch": 1.7669714238069736,
      "grad_norm": 0.005928349681198597,
      "learning_rate": 4.078494696169727e-07,
      "loss": 0.001,
      "step": 275500
    },
    {
      "epoch": 1.7670355607007597,
      "grad_norm": 0.0487651452422142,
      "learning_rate": 4.0762808908589026e-07,
      "loss": 0.0007,
      "step": 275510
    },
    {
      "epoch": 1.7670996975945457,
      "grad_norm": 0.07806593924760818,
      "learning_rate": 4.0740676610050987e-07,
      "loss": 0.0014,
      "step": 275520
    },
    {
      "epoch": 1.7671638344883318,
      "grad_norm": 0.03947426751255989,
      "learning_rate": 4.0718550066360496e-07,
      "loss": 0.0011,
      "step": 275530
    },
    {
      "epoch": 1.767227971382118,
      "grad_norm": 0.11940759420394897,
      "learning_rate": 4.0696429277794713e-07,
      "loss": 0.0009,
      "step": 275540
    },
    {
      "epoch": 1.767292108275904,
      "grad_norm": 0.08800876885652542,
      "learning_rate": 4.067431424463103e-07,
      "loss": 0.0009,
      "step": 275550
    },
    {
      "epoch": 1.7673562451696903,
      "grad_norm": 0.09916777163743973,
      "learning_rate": 4.065220496714645e-07,
      "loss": 0.0013,
      "step": 275560
    },
    {
      "epoch": 1.7674203820634764,
      "grad_norm": 0.042468562722206116,
      "learning_rate": 4.063010144561802e-07,
      "loss": 0.0009,
      "step": 275570
    },
    {
      "epoch": 1.7674845189572623,
      "grad_norm": 0.12761539220809937,
      "learning_rate": 4.0608003680322696e-07,
      "loss": 0.0013,
      "step": 275580
    },
    {
      "epoch": 1.7675486558510485,
      "grad_norm": 0.14592377841472626,
      "learning_rate": 4.0585911671537415e-07,
      "loss": 0.002,
      "step": 275590
    },
    {
      "epoch": 1.7676127927448346,
      "grad_norm": 0.13312381505966187,
      "learning_rate": 4.056382541953907e-07,
      "loss": 0.0007,
      "step": 275600
    },
    {
      "epoch": 1.7676769296386206,
      "grad_norm": 0.01981111615896225,
      "learning_rate": 4.054174492460433e-07,
      "loss": 0.0003,
      "step": 275610
    },
    {
      "epoch": 1.7677410665324067,
      "grad_norm": 0.10597145557403564,
      "learning_rate": 4.051967018700975e-07,
      "loss": 0.001,
      "step": 275620
    },
    {
      "epoch": 1.7678052034261929,
      "grad_norm": 0.011349072679877281,
      "learning_rate": 4.049760120703228e-07,
      "loss": 0.0008,
      "step": 275630
    },
    {
      "epoch": 1.767869340319979,
      "grad_norm": 0.03977828845381737,
      "learning_rate": 4.047553798494819e-07,
      "loss": 0.0011,
      "step": 275640
    },
    {
      "epoch": 1.7679334772137651,
      "grad_norm": 0.06345690786838531,
      "learning_rate": 4.045348052103393e-07,
      "loss": 0.0008,
      "step": 275650
    },
    {
      "epoch": 1.7679976141075513,
      "grad_norm": 0.0018058223649859428,
      "learning_rate": 4.043142881556611e-07,
      "loss": 0.0028,
      "step": 275660
    },
    {
      "epoch": 1.7680617510013372,
      "grad_norm": 0.028217557817697525,
      "learning_rate": 4.040938286882096e-07,
      "loss": 0.001,
      "step": 275670
    },
    {
      "epoch": 1.7681258878951234,
      "grad_norm": 0.054113343358039856,
      "learning_rate": 4.038734268107469e-07,
      "loss": 0.0012,
      "step": 275680
    },
    {
      "epoch": 1.7681900247889093,
      "grad_norm": 0.03654279187321663,
      "learning_rate": 4.0365308252603377e-07,
      "loss": 0.001,
      "step": 275690
    },
    {
      "epoch": 1.7682541616826954,
      "grad_norm": 0.05833207815885544,
      "learning_rate": 4.034327958368339e-07,
      "loss": 0.001,
      "step": 275700
    },
    {
      "epoch": 1.7683182985764816,
      "grad_norm": 0.05758131295442581,
      "learning_rate": 4.032125667459058e-07,
      "loss": 0.0008,
      "step": 275710
    },
    {
      "epoch": 1.7683824354702677,
      "grad_norm": 0.1753859519958496,
      "learning_rate": 4.029923952560094e-07,
      "loss": 0.0018,
      "step": 275720
    },
    {
      "epoch": 1.768446572364054,
      "grad_norm": 0.08543030172586441,
      "learning_rate": 4.027722813699031e-07,
      "loss": 0.0012,
      "step": 275730
    },
    {
      "epoch": 1.76851070925784,
      "grad_norm": 0.01350217405706644,
      "learning_rate": 4.0255222509034685e-07,
      "loss": 0.0007,
      "step": 275740
    },
    {
      "epoch": 1.7685748461516262,
      "grad_norm": 0.12346943467855453,
      "learning_rate": 4.0233222642009627e-07,
      "loss": 0.0007,
      "step": 275750
    },
    {
      "epoch": 1.7686389830454121,
      "grad_norm": 0.032266777008771896,
      "learning_rate": 4.021122853619086e-07,
      "loss": 0.0011,
      "step": 275760
    },
    {
      "epoch": 1.7687031199391983,
      "grad_norm": 0.1392628252506256,
      "learning_rate": 4.0189240191854e-07,
      "loss": 0.0013,
      "step": 275770
    },
    {
      "epoch": 1.7687672568329842,
      "grad_norm": 0.070569708943367,
      "learning_rate": 4.0167257609274645e-07,
      "loss": 0.0007,
      "step": 275780
    },
    {
      "epoch": 1.7688313937267703,
      "grad_norm": 0.026905816048383713,
      "learning_rate": 4.0145280788728147e-07,
      "loss": 0.0006,
      "step": 275790
    },
    {
      "epoch": 1.7688955306205565,
      "grad_norm": 0.042029283940792084,
      "learning_rate": 4.012330973049e-07,
      "loss": 0.0016,
      "step": 275800
    },
    {
      "epoch": 1.7689596675143426,
      "grad_norm": 0.04697522148489952,
      "learning_rate": 4.010134443483532e-07,
      "loss": 0.0007,
      "step": 275810
    },
    {
      "epoch": 1.7690238044081288,
      "grad_norm": 0.23765172064304352,
      "learning_rate": 4.007938490203961e-07,
      "loss": 0.0007,
      "step": 275820
    },
    {
      "epoch": 1.769087941301915,
      "grad_norm": 0.062108952552080154,
      "learning_rate": 4.005743113237787e-07,
      "loss": 0.0007,
      "step": 275830
    },
    {
      "epoch": 1.7691520781957009,
      "grad_norm": 0.06532338261604309,
      "learning_rate": 4.003548312612526e-07,
      "loss": 0.0007,
      "step": 275840
    },
    {
      "epoch": 1.769216215089487,
      "grad_norm": 0.0470815934240818,
      "learning_rate": 4.001354088355669e-07,
      "loss": 0.0011,
      "step": 275850
    },
    {
      "epoch": 1.769280351983273,
      "grad_norm": 0.054930299520492554,
      "learning_rate": 3.999160440494726e-07,
      "loss": 0.0009,
      "step": 275860
    },
    {
      "epoch": 1.769344488877059,
      "grad_norm": 0.04019542038440704,
      "learning_rate": 3.9969673690571807e-07,
      "loss": 0.0005,
      "step": 275870
    },
    {
      "epoch": 1.7694086257708452,
      "grad_norm": 0.10912688821554184,
      "learning_rate": 3.994774874070512e-07,
      "loss": 0.0016,
      "step": 275880
    },
    {
      "epoch": 1.7694727626646314,
      "grad_norm": 0.018208475783467293,
      "learning_rate": 3.9925829555621853e-07,
      "loss": 0.001,
      "step": 275890
    },
    {
      "epoch": 1.7695368995584175,
      "grad_norm": 0.0015009446069598198,
      "learning_rate": 3.9903916135596854e-07,
      "loss": 0.0006,
      "step": 275900
    },
    {
      "epoch": 1.7696010364522037,
      "grad_norm": 0.011169308796525002,
      "learning_rate": 3.9882008480904564e-07,
      "loss": 0.0006,
      "step": 275910
    },
    {
      "epoch": 1.7696651733459898,
      "grad_norm": 0.1605800986289978,
      "learning_rate": 3.986010659181949e-07,
      "loss": 0.0011,
      "step": 275920
    },
    {
      "epoch": 1.7697293102397758,
      "grad_norm": 0.06483528763055801,
      "learning_rate": 3.9838210468616244e-07,
      "loss": 0.0005,
      "step": 275930
    },
    {
      "epoch": 1.769793447133562,
      "grad_norm": 0.03819932043552399,
      "learning_rate": 3.9816320111569107e-07,
      "loss": 0.0007,
      "step": 275940
    },
    {
      "epoch": 1.7698575840273478,
      "grad_norm": 0.07846676558256149,
      "learning_rate": 3.97944355209523e-07,
      "loss": 0.0009,
      "step": 275950
    },
    {
      "epoch": 1.769921720921134,
      "grad_norm": 0.13251625001430511,
      "learning_rate": 3.977255669704011e-07,
      "loss": 0.0007,
      "step": 275960
    },
    {
      "epoch": 1.7699858578149201,
      "grad_norm": 0.0833633616566658,
      "learning_rate": 3.97506836401067e-07,
      "loss": 0.0011,
      "step": 275970
    },
    {
      "epoch": 1.7700499947087063,
      "grad_norm": 0.0796852707862854,
      "learning_rate": 3.9728816350426247e-07,
      "loss": 0.0007,
      "step": 275980
    },
    {
      "epoch": 1.7701141316024924,
      "grad_norm": 0.09615079313516617,
      "learning_rate": 3.9706954828272636e-07,
      "loss": 0.0008,
      "step": 275990
    },
    {
      "epoch": 1.7701782684962786,
      "grad_norm": 0.003064911812543869,
      "learning_rate": 3.9685099073919766e-07,
      "loss": 0.0006,
      "step": 276000
    },
    {
      "epoch": 1.7702424053900645,
      "grad_norm": 0.07344446331262589,
      "learning_rate": 3.9663249087641684e-07,
      "loss": 0.0013,
      "step": 276010
    },
    {
      "epoch": 1.7703065422838506,
      "grad_norm": 0.04304369166493416,
      "learning_rate": 3.964140486971207e-07,
      "loss": 0.0008,
      "step": 276020
    },
    {
      "epoch": 1.7703706791776368,
      "grad_norm": 0.2215459793806076,
      "learning_rate": 3.9619566420404543e-07,
      "loss": 0.0011,
      "step": 276030
    },
    {
      "epoch": 1.7704348160714227,
      "grad_norm": 0.004388847388327122,
      "learning_rate": 3.9597733739993037e-07,
      "loss": 0.0005,
      "step": 276040
    },
    {
      "epoch": 1.7704989529652089,
      "grad_norm": 0.047044143080711365,
      "learning_rate": 3.9575906828750956e-07,
      "loss": 0.002,
      "step": 276050
    },
    {
      "epoch": 1.770563089858995,
      "grad_norm": 0.07690353691577911,
      "learning_rate": 3.9554085686951795e-07,
      "loss": 0.0013,
      "step": 276060
    },
    {
      "epoch": 1.7706272267527812,
      "grad_norm": 0.021951548755168915,
      "learning_rate": 3.9532270314868947e-07,
      "loss": 0.001,
      "step": 276070
    },
    {
      "epoch": 1.7706913636465673,
      "grad_norm": 0.14238323271274567,
      "learning_rate": 3.951046071277592e-07,
      "loss": 0.0015,
      "step": 276080
    },
    {
      "epoch": 1.7707555005403535,
      "grad_norm": 0.04025707021355629,
      "learning_rate": 3.948865688094594e-07,
      "loss": 0.0008,
      "step": 276090
    },
    {
      "epoch": 1.7708196374341394,
      "grad_norm": 0.027796175330877304,
      "learning_rate": 3.9466858819652174e-07,
      "loss": 0.0011,
      "step": 276100
    },
    {
      "epoch": 1.7708837743279255,
      "grad_norm": 0.003692495170980692,
      "learning_rate": 3.944506652916774e-07,
      "loss": 0.0024,
      "step": 276110
    },
    {
      "epoch": 1.7709479112217115,
      "grad_norm": 0.18867793679237366,
      "learning_rate": 3.942328000976586e-07,
      "loss": 0.0021,
      "step": 276120
    },
    {
      "epoch": 1.7710120481154976,
      "grad_norm": 0.0782652497291565,
      "learning_rate": 3.9401499261719487e-07,
      "loss": 0.001,
      "step": 276130
    },
    {
      "epoch": 1.7710761850092838,
      "grad_norm": 0.06763576716184616,
      "learning_rate": 3.937972428530151e-07,
      "loss": 0.0009,
      "step": 276140
    },
    {
      "epoch": 1.77114032190307,
      "grad_norm": 0.00766996992751956,
      "learning_rate": 3.935795508078466e-07,
      "loss": 0.0026,
      "step": 276150
    },
    {
      "epoch": 1.771204458796856,
      "grad_norm": 0.023206427693367004,
      "learning_rate": 3.9336191648441935e-07,
      "loss": 0.0008,
      "step": 276160
    },
    {
      "epoch": 1.7712685956906422,
      "grad_norm": 0.008450353518128395,
      "learning_rate": 3.931443398854601e-07,
      "loss": 0.0013,
      "step": 276170
    },
    {
      "epoch": 1.7713327325844284,
      "grad_norm": 0.027441825717687607,
      "learning_rate": 3.9292682101369326e-07,
      "loss": 0.001,
      "step": 276180
    },
    {
      "epoch": 1.7713968694782143,
      "grad_norm": 0.03814779594540596,
      "learning_rate": 3.9270935987184734e-07,
      "loss": 0.0016,
      "step": 276190
    },
    {
      "epoch": 1.7714610063720004,
      "grad_norm": 0.061811137944459915,
      "learning_rate": 3.924919564626456e-07,
      "loss": 0.0019,
      "step": 276200
    },
    {
      "epoch": 1.7715251432657864,
      "grad_norm": 0.08312131464481354,
      "learning_rate": 3.9227461078881314e-07,
      "loss": 0.001,
      "step": 276210
    },
    {
      "epoch": 1.7715892801595725,
      "grad_norm": 0.057990048080682755,
      "learning_rate": 3.920573228530716e-07,
      "loss": 0.0003,
      "step": 276220
    },
    {
      "epoch": 1.7716534170533587,
      "grad_norm": 0.01682116463780403,
      "learning_rate": 3.918400926581456e-07,
      "loss": 0.0004,
      "step": 276230
    },
    {
      "epoch": 1.7717175539471448,
      "grad_norm": 0.009075605310499668,
      "learning_rate": 3.916229202067573e-07,
      "loss": 0.0009,
      "step": 276240
    },
    {
      "epoch": 1.771781690840931,
      "grad_norm": 0.058820053935050964,
      "learning_rate": 3.914058055016268e-07,
      "loss": 0.0007,
      "step": 276250
    },
    {
      "epoch": 1.771845827734717,
      "grad_norm": 0.06107432767748833,
      "learning_rate": 3.911887485454746e-07,
      "loss": 0.0009,
      "step": 276260
    },
    {
      "epoch": 1.771909964628503,
      "grad_norm": 0.04917144402861595,
      "learning_rate": 3.90971749341022e-07,
      "loss": 0.0007,
      "step": 276270
    },
    {
      "epoch": 1.7719741015222892,
      "grad_norm": 0.11266804486513138,
      "learning_rate": 3.9075480789098786e-07,
      "loss": 0.0012,
      "step": 276280
    },
    {
      "epoch": 1.772038238416075,
      "grad_norm": 0.03330082818865776,
      "learning_rate": 3.9053792419809e-07,
      "loss": 0.0013,
      "step": 276290
    },
    {
      "epoch": 1.7721023753098613,
      "grad_norm": 0.07728027552366257,
      "learning_rate": 3.9032109826504516e-07,
      "loss": 0.0013,
      "step": 276300
    },
    {
      "epoch": 1.7721665122036474,
      "grad_norm": 0.09894976019859314,
      "learning_rate": 3.9010433009457225e-07,
      "loss": 0.0009,
      "step": 276310
    },
    {
      "epoch": 1.7722306490974336,
      "grad_norm": 0.04832207411527634,
      "learning_rate": 3.8988761968938693e-07,
      "loss": 0.0012,
      "step": 276320
    },
    {
      "epoch": 1.7722947859912197,
      "grad_norm": 0.03660057112574577,
      "learning_rate": 3.8967096705220363e-07,
      "loss": 0.002,
      "step": 276330
    },
    {
      "epoch": 1.7723589228850058,
      "grad_norm": 0.059815533459186554,
      "learning_rate": 3.894543721857391e-07,
      "loss": 0.001,
      "step": 276340
    },
    {
      "epoch": 1.772423059778792,
      "grad_norm": 0.08611498773097992,
      "learning_rate": 3.8923783509270616e-07,
      "loss": 0.001,
      "step": 276350
    },
    {
      "epoch": 1.772487196672578,
      "grad_norm": 0.030227821320295334,
      "learning_rate": 3.8902135577581825e-07,
      "loss": 0.0008,
      "step": 276360
    },
    {
      "epoch": 1.772551333566364,
      "grad_norm": 0.09007013589143753,
      "learning_rate": 3.8880493423778755e-07,
      "loss": 0.0008,
      "step": 276370
    },
    {
      "epoch": 1.77261547046015,
      "grad_norm": 0.27165305614471436,
      "learning_rate": 3.885885704813275e-07,
      "loss": 0.002,
      "step": 276380
    },
    {
      "epoch": 1.7726796073539361,
      "grad_norm": 0.01975778490304947,
      "learning_rate": 3.883722645091487e-07,
      "loss": 0.0009,
      "step": 276390
    },
    {
      "epoch": 1.7727437442477223,
      "grad_norm": 0.01760447584092617,
      "learning_rate": 3.881560163239606e-07,
      "loss": 0.0016,
      "step": 276400
    },
    {
      "epoch": 1.7728078811415084,
      "grad_norm": 0.0017506623407825828,
      "learning_rate": 3.8793982592847333e-07,
      "loss": 0.0015,
      "step": 276410
    },
    {
      "epoch": 1.7728720180352946,
      "grad_norm": 0.07523943483829498,
      "learning_rate": 3.8772369332539695e-07,
      "loss": 0.0006,
      "step": 276420
    },
    {
      "epoch": 1.7729361549290807,
      "grad_norm": 0.028977354988455772,
      "learning_rate": 3.8750761851743866e-07,
      "loss": 0.0006,
      "step": 276430
    },
    {
      "epoch": 1.773000291822867,
      "grad_norm": 0.0565430223941803,
      "learning_rate": 3.8729160150730694e-07,
      "loss": 0.0012,
      "step": 276440
    },
    {
      "epoch": 1.7730644287166528,
      "grad_norm": 0.03902720287442207,
      "learning_rate": 3.870756422977068e-07,
      "loss": 0.001,
      "step": 276450
    },
    {
      "epoch": 1.773128565610439,
      "grad_norm": 0.02282317355275154,
      "learning_rate": 3.868597408913466e-07,
      "loss": 0.0008,
      "step": 276460
    },
    {
      "epoch": 1.773192702504225,
      "grad_norm": 0.05966460704803467,
      "learning_rate": 3.8664389729093144e-07,
      "loss": 0.0007,
      "step": 276470
    },
    {
      "epoch": 1.773256839398011,
      "grad_norm": 0.038154445588588715,
      "learning_rate": 3.864281114991636e-07,
      "loss": 0.0011,
      "step": 276480
    },
    {
      "epoch": 1.7733209762917972,
      "grad_norm": 0.07106415182352066,
      "learning_rate": 3.862123835187503e-07,
      "loss": 0.0005,
      "step": 276490
    },
    {
      "epoch": 1.7733851131855833,
      "grad_norm": 0.05983132869005203,
      "learning_rate": 3.8599671335239284e-07,
      "loss": 0.0026,
      "step": 276500
    },
    {
      "epoch": 1.7734492500793695,
      "grad_norm": 0.06823556870222092,
      "learning_rate": 3.8578110100279455e-07,
      "loss": 0.0012,
      "step": 276510
    },
    {
      "epoch": 1.7735133869731556,
      "grad_norm": 0.03860345482826233,
      "learning_rate": 3.85565546472656e-07,
      "loss": 0.0006,
      "step": 276520
    },
    {
      "epoch": 1.7735775238669416,
      "grad_norm": 0.045919954776763916,
      "learning_rate": 3.853500497646784e-07,
      "loss": 0.0007,
      "step": 276530
    },
    {
      "epoch": 1.7736416607607277,
      "grad_norm": 0.0455821193754673,
      "learning_rate": 3.85134610881564e-07,
      "loss": 0.0005,
      "step": 276540
    },
    {
      "epoch": 1.7737057976545136,
      "grad_norm": 0.05296953395009041,
      "learning_rate": 3.849192298260113e-07,
      "loss": 0.0014,
      "step": 276550
    },
    {
      "epoch": 1.7737699345482998,
      "grad_norm": 0.11851990222930908,
      "learning_rate": 3.8470390660071855e-07,
      "loss": 0.0005,
      "step": 276560
    },
    {
      "epoch": 1.773834071442086,
      "grad_norm": 0.06372379511594772,
      "learning_rate": 3.844886412083837e-07,
      "loss": 0.0019,
      "step": 276570
    },
    {
      "epoch": 1.773898208335872,
      "grad_norm": 0.06602787226438522,
      "learning_rate": 3.8427343365170567e-07,
      "loss": 0.0008,
      "step": 276580
    },
    {
      "epoch": 1.7739623452296582,
      "grad_norm": 0.09497271478176117,
      "learning_rate": 3.8405828393338063e-07,
      "loss": 0.0012,
      "step": 276590
    },
    {
      "epoch": 1.7740264821234444,
      "grad_norm": 0.038306642323732376,
      "learning_rate": 3.838431920561031e-07,
      "loss": 0.0014,
      "step": 276600
    },
    {
      "epoch": 1.7740906190172305,
      "grad_norm": 0.15213309228420258,
      "learning_rate": 3.8362815802257035e-07,
      "loss": 0.0014,
      "step": 276610
    },
    {
      "epoch": 1.7741547559110165,
      "grad_norm": 0.04893713817000389,
      "learning_rate": 3.8341318183547583e-07,
      "loss": 0.0005,
      "step": 276620
    },
    {
      "epoch": 1.7742188928048026,
      "grad_norm": 0.07411212474107742,
      "learning_rate": 3.83198263497514e-07,
      "loss": 0.001,
      "step": 276630
    },
    {
      "epoch": 1.7742830296985885,
      "grad_norm": 0.08615033328533173,
      "learning_rate": 3.829834030113766e-07,
      "loss": 0.0008,
      "step": 276640
    },
    {
      "epoch": 1.7743471665923747,
      "grad_norm": 0.03488355875015259,
      "learning_rate": 3.8276860037975704e-07,
      "loss": 0.001,
      "step": 276650
    },
    {
      "epoch": 1.7744113034861608,
      "grad_norm": 0.06693438440561295,
      "learning_rate": 3.825538556053476e-07,
      "loss": 0.0012,
      "step": 276660
    },
    {
      "epoch": 1.774475440379947,
      "grad_norm": 0.10357903689146042,
      "learning_rate": 3.8233916869083845e-07,
      "loss": 0.0009,
      "step": 276670
    },
    {
      "epoch": 1.7745395772737331,
      "grad_norm": 0.21852293610572815,
      "learning_rate": 3.8212453963891837e-07,
      "loss": 0.0009,
      "step": 276680
    },
    {
      "epoch": 1.7746037141675193,
      "grad_norm": 0.11806097626686096,
      "learning_rate": 3.819099684522792e-07,
      "loss": 0.0013,
      "step": 276690
    },
    {
      "epoch": 1.7746678510613052,
      "grad_norm": 0.250988245010376,
      "learning_rate": 3.816954551336088e-07,
      "loss": 0.0019,
      "step": 276700
    },
    {
      "epoch": 1.7747319879550914,
      "grad_norm": 0.04955058917403221,
      "learning_rate": 3.8148099968559447e-07,
      "loss": 0.0004,
      "step": 276710
    },
    {
      "epoch": 1.7747961248488775,
      "grad_norm": 0.05638645961880684,
      "learning_rate": 3.812666021109235e-07,
      "loss": 0.0009,
      "step": 276720
    },
    {
      "epoch": 1.7748602617426634,
      "grad_norm": 0.07330942153930664,
      "learning_rate": 3.810522624122842e-07,
      "loss": 0.0005,
      "step": 276730
    },
    {
      "epoch": 1.7749243986364496,
      "grad_norm": 0.05153987556695938,
      "learning_rate": 3.808379805923607e-07,
      "loss": 0.0009,
      "step": 276740
    },
    {
      "epoch": 1.7749885355302357,
      "grad_norm": 0.09960044920444489,
      "learning_rate": 3.806237566538379e-07,
      "loss": 0.0019,
      "step": 276750
    },
    {
      "epoch": 1.7750526724240219,
      "grad_norm": 0.030330004170536995,
      "learning_rate": 3.8040959059940154e-07,
      "loss": 0.0019,
      "step": 276760
    },
    {
      "epoch": 1.775116809317808,
      "grad_norm": 0.08835256099700928,
      "learning_rate": 3.8019548243173444e-07,
      "loss": 0.0006,
      "step": 276770
    },
    {
      "epoch": 1.7751809462115942,
      "grad_norm": 0.02172674797475338,
      "learning_rate": 3.7998143215352e-07,
      "loss": 0.0009,
      "step": 276780
    },
    {
      "epoch": 1.77524508310538,
      "grad_norm": 0.3368646800518036,
      "learning_rate": 3.797674397674389e-07,
      "loss": 0.002,
      "step": 276790
    },
    {
      "epoch": 1.7753092199991662,
      "grad_norm": 0.039189331233501434,
      "learning_rate": 3.795535052761751e-07,
      "loss": 0.0012,
      "step": 276800
    },
    {
      "epoch": 1.7753733568929522,
      "grad_norm": 0.1254708170890808,
      "learning_rate": 3.7933962868240805e-07,
      "loss": 0.0012,
      "step": 276810
    },
    {
      "epoch": 1.7754374937867383,
      "grad_norm": 0.03662301227450371,
      "learning_rate": 3.7912580998881734e-07,
      "loss": 0.0013,
      "step": 276820
    },
    {
      "epoch": 1.7755016306805245,
      "grad_norm": 0.08107379078865051,
      "learning_rate": 3.7891204919808243e-07,
      "loss": 0.0009,
      "step": 276830
    },
    {
      "epoch": 1.7755657675743106,
      "grad_norm": 0.012480441480875015,
      "learning_rate": 3.786983463128824e-07,
      "loss": 0.0014,
      "step": 276840
    },
    {
      "epoch": 1.7756299044680968,
      "grad_norm": 0.09728964418172836,
      "learning_rate": 3.7848470133589557e-07,
      "loss": 0.0014,
      "step": 276850
    },
    {
      "epoch": 1.775694041361883,
      "grad_norm": 0.1525079309940338,
      "learning_rate": 3.7827111426979755e-07,
      "loss": 0.001,
      "step": 276860
    },
    {
      "epoch": 1.775758178255669,
      "grad_norm": 0.09387625008821487,
      "learning_rate": 3.780575851172652e-07,
      "loss": 0.0014,
      "step": 276870
    },
    {
      "epoch": 1.775822315149455,
      "grad_norm": 0.08427508920431137,
      "learning_rate": 3.7784411388097577e-07,
      "loss": 0.0007,
      "step": 276880
    },
    {
      "epoch": 1.7758864520432411,
      "grad_norm": 0.07875356823205948,
      "learning_rate": 3.776307005636026e-07,
      "loss": 0.0007,
      "step": 276890
    },
    {
      "epoch": 1.775950588937027,
      "grad_norm": 0.04849980026483536,
      "learning_rate": 3.7741734516781925e-07,
      "loss": 0.0007,
      "step": 276900
    },
    {
      "epoch": 1.7760147258308132,
      "grad_norm": 0.050537627190351486,
      "learning_rate": 3.7720404769630124e-07,
      "loss": 0.0015,
      "step": 276910
    },
    {
      "epoch": 1.7760788627245994,
      "grad_norm": 0.25381091237068176,
      "learning_rate": 3.76990808151721e-07,
      "loss": 0.0006,
      "step": 276920
    },
    {
      "epoch": 1.7761429996183855,
      "grad_norm": 0.022908836603164673,
      "learning_rate": 3.76777626536749e-07,
      "loss": 0.0007,
      "step": 276930
    },
    {
      "epoch": 1.7762071365121717,
      "grad_norm": 0.030104728415608406,
      "learning_rate": 3.765645028540571e-07,
      "loss": 0.0007,
      "step": 276940
    },
    {
      "epoch": 1.7762712734059578,
      "grad_norm": 0.03946729749441147,
      "learning_rate": 3.7635143710631706e-07,
      "loss": 0.0018,
      "step": 276950
    },
    {
      "epoch": 1.7763354102997437,
      "grad_norm": 0.08555560559034348,
      "learning_rate": 3.7613842929619837e-07,
      "loss": 0.0008,
      "step": 276960
    },
    {
      "epoch": 1.7763995471935299,
      "grad_norm": 0.046086009591817856,
      "learning_rate": 3.7592547942636946e-07,
      "loss": 0.001,
      "step": 276970
    },
    {
      "epoch": 1.7764636840873158,
      "grad_norm": 0.06498829275369644,
      "learning_rate": 3.757125874994982e-07,
      "loss": 0.0011,
      "step": 276980
    },
    {
      "epoch": 1.776527820981102,
      "grad_norm": 0.04270976409316063,
      "learning_rate": 3.754997535182542e-07,
      "loss": 0.0018,
      "step": 276990
    },
    {
      "epoch": 1.776591957874888,
      "grad_norm": 0.09135537594556808,
      "learning_rate": 3.75286977485303e-07,
      "loss": 0.0012,
      "step": 277000
    },
    {
      "epoch": 1.7766560947686743,
      "grad_norm": 0.030918428674340248,
      "learning_rate": 3.750742594033113e-07,
      "loss": 0.0005,
      "step": 277010
    },
    {
      "epoch": 1.7767202316624604,
      "grad_norm": 0.07951313257217407,
      "learning_rate": 3.748615992749438e-07,
      "loss": 0.0008,
      "step": 277020
    },
    {
      "epoch": 1.7767843685562466,
      "grad_norm": 0.018732817843556404,
      "learning_rate": 3.7464899710286717e-07,
      "loss": 0.0005,
      "step": 277030
    },
    {
      "epoch": 1.7768485054500327,
      "grad_norm": 0.2585754990577698,
      "learning_rate": 3.7443645288974374e-07,
      "loss": 0.0009,
      "step": 277040
    },
    {
      "epoch": 1.7769126423438186,
      "grad_norm": 0.0019640575628727674,
      "learning_rate": 3.742239666382369e-07,
      "loss": 0.0008,
      "step": 277050
    },
    {
      "epoch": 1.7769767792376048,
      "grad_norm": 0.11195459216833115,
      "learning_rate": 3.740115383510107e-07,
      "loss": 0.0008,
      "step": 277060
    },
    {
      "epoch": 1.7770409161313907,
      "grad_norm": 0.03852207213640213,
      "learning_rate": 3.7379916803072567e-07,
      "loss": 0.0008,
      "step": 277070
    },
    {
      "epoch": 1.7771050530251769,
      "grad_norm": 0.09934749454259872,
      "learning_rate": 3.735868556800432e-07,
      "loss": 0.0007,
      "step": 277080
    },
    {
      "epoch": 1.777169189918963,
      "grad_norm": 0.03736523538827896,
      "learning_rate": 3.7337460130162375e-07,
      "loss": 0.0008,
      "step": 277090
    },
    {
      "epoch": 1.7772333268127491,
      "grad_norm": 0.0296183992177248,
      "learning_rate": 3.731624048981275e-07,
      "loss": 0.0008,
      "step": 277100
    },
    {
      "epoch": 1.7772974637065353,
      "grad_norm": 0.03917883336544037,
      "learning_rate": 3.729502664722129e-07,
      "loss": 0.001,
      "step": 277110
    },
    {
      "epoch": 1.7773616006003214,
      "grad_norm": 0.13903005421161652,
      "learning_rate": 3.727381860265389e-07,
      "loss": 0.0012,
      "step": 277120
    },
    {
      "epoch": 1.7774257374941074,
      "grad_norm": 0.12451034784317017,
      "learning_rate": 3.7252616356376117e-07,
      "loss": 0.0005,
      "step": 277130
    },
    {
      "epoch": 1.7774898743878935,
      "grad_norm": 0.0074807340279221535,
      "learning_rate": 3.723141990865392e-07,
      "loss": 0.0007,
      "step": 277140
    },
    {
      "epoch": 1.7775540112816797,
      "grad_norm": 0.1273653656244278,
      "learning_rate": 3.7210229259752706e-07,
      "loss": 0.0011,
      "step": 277150
    },
    {
      "epoch": 1.7776181481754656,
      "grad_norm": 0.04760697856545448,
      "learning_rate": 3.718904440993809e-07,
      "loss": 0.0022,
      "step": 277160
    },
    {
      "epoch": 1.7776822850692517,
      "grad_norm": 0.14566820859909058,
      "learning_rate": 3.716786535947542e-07,
      "loss": 0.001,
      "step": 277170
    },
    {
      "epoch": 1.777746421963038,
      "grad_norm": 0.045055124908685684,
      "learning_rate": 3.714669210863031e-07,
      "loss": 0.0007,
      "step": 277180
    },
    {
      "epoch": 1.777810558856824,
      "grad_norm": 0.013329172506928444,
      "learning_rate": 3.712552465766789e-07,
      "loss": 0.0013,
      "step": 277190
    },
    {
      "epoch": 1.7778746957506102,
      "grad_norm": 0.05233705788850784,
      "learning_rate": 3.710436300685338e-07,
      "loss": 0.0011,
      "step": 277200
    },
    {
      "epoch": 1.7779388326443963,
      "grad_norm": 0.08523570001125336,
      "learning_rate": 3.708320715645214e-07,
      "loss": 0.0016,
      "step": 277210
    },
    {
      "epoch": 1.7780029695381823,
      "grad_norm": 0.054902657866477966,
      "learning_rate": 3.7062057106729166e-07,
      "loss": 0.0014,
      "step": 277220
    },
    {
      "epoch": 1.7780671064319684,
      "grad_norm": 0.004034021403640509,
      "learning_rate": 3.7040912857949474e-07,
      "loss": 0.0015,
      "step": 277230
    },
    {
      "epoch": 1.7781312433257543,
      "grad_norm": 0.020426299422979355,
      "learning_rate": 3.7019774410377905e-07,
      "loss": 0.0011,
      "step": 277240
    },
    {
      "epoch": 1.7781953802195405,
      "grad_norm": 0.272157222032547,
      "learning_rate": 3.699864176427953e-07,
      "loss": 0.0012,
      "step": 277250
    },
    {
      "epoch": 1.7782595171133266,
      "grad_norm": 0.02593575045466423,
      "learning_rate": 3.6977514919919075e-07,
      "loss": 0.0007,
      "step": 277260
    },
    {
      "epoch": 1.7783236540071128,
      "grad_norm": 0.15078026056289673,
      "learning_rate": 3.695639387756128e-07,
      "loss": 0.0007,
      "step": 277270
    },
    {
      "epoch": 1.778387790900899,
      "grad_norm": 0.08224067836999893,
      "learning_rate": 3.6935278637470705e-07,
      "loss": 0.0007,
      "step": 277280
    },
    {
      "epoch": 1.778451927794685,
      "grad_norm": 0.02865617163479328,
      "learning_rate": 3.6914169199912086e-07,
      "loss": 0.0022,
      "step": 277290
    },
    {
      "epoch": 1.7785160646884712,
      "grad_norm": 0.006676721386611462,
      "learning_rate": 3.689306556514993e-07,
      "loss": 0.001,
      "step": 277300
    },
    {
      "epoch": 1.7785802015822572,
      "grad_norm": 0.10433896631002426,
      "learning_rate": 3.6871967733448646e-07,
      "loss": 0.0009,
      "step": 277310
    },
    {
      "epoch": 1.7786443384760433,
      "grad_norm": 0.019970644265413284,
      "learning_rate": 3.685087570507251e-07,
      "loss": 0.0018,
      "step": 277320
    },
    {
      "epoch": 1.7787084753698292,
      "grad_norm": 0.058814890682697296,
      "learning_rate": 3.6829789480285983e-07,
      "loss": 0.0013,
      "step": 277330
    },
    {
      "epoch": 1.7787726122636154,
      "grad_norm": 0.024515783414244652,
      "learning_rate": 3.6808709059353244e-07,
      "loss": 0.0005,
      "step": 277340
    },
    {
      "epoch": 1.7788367491574015,
      "grad_norm": 0.08331423252820969,
      "learning_rate": 3.6787634442538414e-07,
      "loss": 0.0013,
      "step": 277350
    },
    {
      "epoch": 1.7789008860511877,
      "grad_norm": 0.07859785109758377,
      "learning_rate": 3.67665656301055e-07,
      "loss": 0.0007,
      "step": 277360
    },
    {
      "epoch": 1.7789650229449738,
      "grad_norm": 0.010937534272670746,
      "learning_rate": 3.674550262231869e-07,
      "loss": 0.0008,
      "step": 277370
    },
    {
      "epoch": 1.77902915983876,
      "grad_norm": 0.1252000331878662,
      "learning_rate": 3.672444541944176e-07,
      "loss": 0.0009,
      "step": 277380
    },
    {
      "epoch": 1.779093296732546,
      "grad_norm": 0.14219404757022858,
      "learning_rate": 3.670339402173867e-07,
      "loss": 0.0015,
      "step": 277390
    },
    {
      "epoch": 1.779157433626332,
      "grad_norm": 0.1581021100282669,
      "learning_rate": 3.6682348429473044e-07,
      "loss": 0.0008,
      "step": 277400
    },
    {
      "epoch": 1.779221570520118,
      "grad_norm": 0.16713345050811768,
      "learning_rate": 3.6661308642908846e-07,
      "loss": 0.0011,
      "step": 277410
    },
    {
      "epoch": 1.7792857074139041,
      "grad_norm": 0.058367762714624405,
      "learning_rate": 3.664027466230957e-07,
      "loss": 0.001,
      "step": 277420
    },
    {
      "epoch": 1.7793498443076903,
      "grad_norm": 0.012379063293337822,
      "learning_rate": 3.66192464879388e-07,
      "loss": 0.001,
      "step": 277430
    },
    {
      "epoch": 1.7794139812014764,
      "grad_norm": 0.08468479663133621,
      "learning_rate": 3.6598224120059977e-07,
      "loss": 0.0012,
      "step": 277440
    },
    {
      "epoch": 1.7794781180952626,
      "grad_norm": 0.1617065668106079,
      "learning_rate": 3.657720755893668e-07,
      "loss": 0.0017,
      "step": 277450
    },
    {
      "epoch": 1.7795422549890487,
      "grad_norm": 0.03813833370804787,
      "learning_rate": 3.655619680483213e-07,
      "loss": 0.0021,
      "step": 277460
    },
    {
      "epoch": 1.7796063918828349,
      "grad_norm": 0.18245063722133636,
      "learning_rate": 3.6535191858009633e-07,
      "loss": 0.0011,
      "step": 277470
    },
    {
      "epoch": 1.7796705287766208,
      "grad_norm": 0.0410914272069931,
      "learning_rate": 3.6514192718732467e-07,
      "loss": 0.0038,
      "step": 277480
    },
    {
      "epoch": 1.779734665670407,
      "grad_norm": 0.07085295766592026,
      "learning_rate": 3.64931993872637e-07,
      "loss": 0.0009,
      "step": 277490
    },
    {
      "epoch": 1.7797988025641929,
      "grad_norm": 0.18684187531471252,
      "learning_rate": 3.647221186386646e-07,
      "loss": 0.0013,
      "step": 277500
    },
    {
      "epoch": 1.779862939457979,
      "grad_norm": 0.13839027285575867,
      "learning_rate": 3.6451230148803583e-07,
      "loss": 0.0012,
      "step": 277510
    },
    {
      "epoch": 1.7799270763517652,
      "grad_norm": 0.05822043865919113,
      "learning_rate": 3.643025424233815e-07,
      "loss": 0.0008,
      "step": 277520
    },
    {
      "epoch": 1.7799912132455513,
      "grad_norm": 0.013622069731354713,
      "learning_rate": 3.640928414473294e-07,
      "loss": 0.0012,
      "step": 277530
    },
    {
      "epoch": 1.7800553501393375,
      "grad_norm": 0.24358125030994415,
      "learning_rate": 3.6388319856250744e-07,
      "loss": 0.0007,
      "step": 277540
    },
    {
      "epoch": 1.7801194870331236,
      "grad_norm": 0.11311017721891403,
      "learning_rate": 3.636736137715413e-07,
      "loss": 0.0009,
      "step": 277550
    },
    {
      "epoch": 1.7801836239269095,
      "grad_norm": 0.037950318306684494,
      "learning_rate": 3.6346408707705893e-07,
      "loss": 0.0011,
      "step": 277560
    },
    {
      "epoch": 1.7802477608206957,
      "grad_norm": 0.062418099492788315,
      "learning_rate": 3.63254618481686e-07,
      "loss": 0.0012,
      "step": 277570
    },
    {
      "epoch": 1.7803118977144818,
      "grad_norm": 0.01619354449212551,
      "learning_rate": 3.6304520798804587e-07,
      "loss": 0.0008,
      "step": 277580
    },
    {
      "epoch": 1.7803760346082678,
      "grad_norm": 0.030265532433986664,
      "learning_rate": 3.6283585559876265e-07,
      "loss": 0.0014,
      "step": 277590
    },
    {
      "epoch": 1.780440171502054,
      "grad_norm": 0.15462017059326172,
      "learning_rate": 3.626265613164609e-07,
      "loss": 0.0008,
      "step": 277600
    },
    {
      "epoch": 1.78050430839584,
      "grad_norm": 0.0385175496339798,
      "learning_rate": 3.6241732514376295e-07,
      "loss": 0.0007,
      "step": 277610
    },
    {
      "epoch": 1.7805684452896262,
      "grad_norm": 0.08664373308420181,
      "learning_rate": 3.6220814708328945e-07,
      "loss": 0.0011,
      "step": 277620
    },
    {
      "epoch": 1.7806325821834124,
      "grad_norm": 0.007938371039927006,
      "learning_rate": 3.6199902713766334e-07,
      "loss": 0.0006,
      "step": 277630
    },
    {
      "epoch": 1.7806967190771985,
      "grad_norm": 0.1576695740222931,
      "learning_rate": 3.617899653095042e-07,
      "loss": 0.001,
      "step": 277640
    },
    {
      "epoch": 1.7807608559709844,
      "grad_norm": 0.10778723657131195,
      "learning_rate": 3.6158096160143096e-07,
      "loss": 0.0018,
      "step": 277650
    },
    {
      "epoch": 1.7808249928647706,
      "grad_norm": 0.03832042217254639,
      "learning_rate": 3.613720160160633e-07,
      "loss": 0.0016,
      "step": 277660
    },
    {
      "epoch": 1.7808891297585565,
      "grad_norm": 0.14303508400917053,
      "learning_rate": 3.611631285560202e-07,
      "loss": 0.0022,
      "step": 277670
    },
    {
      "epoch": 1.7809532666523427,
      "grad_norm": 0.05030818283557892,
      "learning_rate": 3.609542992239179e-07,
      "loss": 0.0011,
      "step": 277680
    },
    {
      "epoch": 1.7810174035461288,
      "grad_norm": 0.15469315648078918,
      "learning_rate": 3.607455280223743e-07,
      "loss": 0.0024,
      "step": 277690
    },
    {
      "epoch": 1.781081540439915,
      "grad_norm": 0.005536719225347042,
      "learning_rate": 3.6053681495400337e-07,
      "loss": 0.0017,
      "step": 277700
    },
    {
      "epoch": 1.781145677333701,
      "grad_norm": 0.017671020701527596,
      "learning_rate": 3.603281600214231e-07,
      "loss": 0.001,
      "step": 277710
    },
    {
      "epoch": 1.7812098142274873,
      "grad_norm": 0.22595839202404022,
      "learning_rate": 3.601195632272475e-07,
      "loss": 0.0017,
      "step": 277720
    },
    {
      "epoch": 1.7812739511212734,
      "grad_norm": 0.1403406709432602,
      "learning_rate": 3.5991102457408834e-07,
      "loss": 0.0018,
      "step": 277730
    },
    {
      "epoch": 1.7813380880150593,
      "grad_norm": 0.09702686965465546,
      "learning_rate": 3.5970254406456126e-07,
      "loss": 0.0011,
      "step": 277740
    },
    {
      "epoch": 1.7814022249088455,
      "grad_norm": 0.10749652981758118,
      "learning_rate": 3.594941217012776e-07,
      "loss": 0.0007,
      "step": 277750
    },
    {
      "epoch": 1.7814663618026314,
      "grad_norm": 0.054250333458185196,
      "learning_rate": 3.592857574868491e-07,
      "loss": 0.0007,
      "step": 277760
    },
    {
      "epoch": 1.7815304986964176,
      "grad_norm": 0.04100211337208748,
      "learning_rate": 3.5907745142388597e-07,
      "loss": 0.0009,
      "step": 277770
    },
    {
      "epoch": 1.7815946355902037,
      "grad_norm": 0.08517192304134369,
      "learning_rate": 3.588692035149999e-07,
      "loss": 0.0013,
      "step": 277780
    },
    {
      "epoch": 1.7816587724839898,
      "grad_norm": 0.23140820860862732,
      "learning_rate": 3.5866101376279995e-07,
      "loss": 0.002,
      "step": 277790
    },
    {
      "epoch": 1.781722909377776,
      "grad_norm": 0.023030906915664673,
      "learning_rate": 3.5845288216989407e-07,
      "loss": 0.0013,
      "step": 277800
    },
    {
      "epoch": 1.7817870462715621,
      "grad_norm": 0.0510365292429924,
      "learning_rate": 3.5824480873889064e-07,
      "loss": 0.002,
      "step": 277810
    },
    {
      "epoch": 1.781851183165348,
      "grad_norm": 0.018052220344543457,
      "learning_rate": 3.5803679347239775e-07,
      "loss": 0.001,
      "step": 277820
    },
    {
      "epoch": 1.7819153200591342,
      "grad_norm": 0.06128367409110069,
      "learning_rate": 3.578288363730215e-07,
      "loss": 0.0011,
      "step": 277830
    },
    {
      "epoch": 1.7819794569529201,
      "grad_norm": 0.0588102862238884,
      "learning_rate": 3.5762093744336754e-07,
      "loss": 0.0008,
      "step": 277840
    },
    {
      "epoch": 1.7820435938467063,
      "grad_norm": 0.06334016472101212,
      "learning_rate": 3.5741309668604007e-07,
      "loss": 0.0014,
      "step": 277850
    },
    {
      "epoch": 1.7821077307404924,
      "grad_norm": 0.06539490818977356,
      "learning_rate": 3.5720531410364523e-07,
      "loss": 0.0019,
      "step": 277860
    },
    {
      "epoch": 1.7821718676342786,
      "grad_norm": 0.0030477459076792,
      "learning_rate": 3.569975896987865e-07,
      "loss": 0.0012,
      "step": 277870
    },
    {
      "epoch": 1.7822360045280647,
      "grad_norm": 0.2873289883136749,
      "learning_rate": 3.5678992347406517e-07,
      "loss": 0.0009,
      "step": 277880
    },
    {
      "epoch": 1.782300141421851,
      "grad_norm": 0.06125490739941597,
      "learning_rate": 3.5658231543208523e-07,
      "loss": 0.001,
      "step": 277890
    },
    {
      "epoch": 1.782364278315637,
      "grad_norm": 0.001798335462808609,
      "learning_rate": 3.563747655754479e-07,
      "loss": 0.0008,
      "step": 277900
    },
    {
      "epoch": 1.782428415209423,
      "grad_norm": 0.04958849400281906,
      "learning_rate": 3.5616727390675343e-07,
      "loss": 0.0014,
      "step": 277910
    },
    {
      "epoch": 1.7824925521032091,
      "grad_norm": 0.00632866658270359,
      "learning_rate": 3.5595984042860076e-07,
      "loss": 0.0018,
      "step": 277920
    },
    {
      "epoch": 1.782556688996995,
      "grad_norm": 0.048083748668432236,
      "learning_rate": 3.557524651435912e-07,
      "loss": 0.0034,
      "step": 277930
    },
    {
      "epoch": 1.7826208258907812,
      "grad_norm": 0.14223527908325195,
      "learning_rate": 3.555451480543226e-07,
      "loss": 0.0009,
      "step": 277940
    },
    {
      "epoch": 1.7826849627845673,
      "grad_norm": 0.007624818477779627,
      "learning_rate": 3.553378891633924e-07,
      "loss": 0.001,
      "step": 277950
    },
    {
      "epoch": 1.7827490996783535,
      "grad_norm": 0.08095641434192657,
      "learning_rate": 3.551306884733974e-07,
      "loss": 0.0006,
      "step": 277960
    },
    {
      "epoch": 1.7828132365721396,
      "grad_norm": 0.22837106883525848,
      "learning_rate": 3.5492354598693547e-07,
      "loss": 0.0018,
      "step": 277970
    },
    {
      "epoch": 1.7828773734659258,
      "grad_norm": 0.2520412504673004,
      "learning_rate": 3.5471646170660067e-07,
      "loss": 0.0016,
      "step": 277980
    },
    {
      "epoch": 1.782941510359712,
      "grad_norm": 0.05908200144767761,
      "learning_rate": 3.545094356349893e-07,
      "loss": 0.0013,
      "step": 277990
    },
    {
      "epoch": 1.7830056472534979,
      "grad_norm": 0.06598391383886337,
      "learning_rate": 3.543024677746937e-07,
      "loss": 0.0008,
      "step": 278000
    },
    {
      "epoch": 1.783069784147284,
      "grad_norm": 0.013274770230054855,
      "learning_rate": 3.54095558128309e-07,
      "loss": 0.0014,
      "step": 278010
    },
    {
      "epoch": 1.78313392104107,
      "grad_norm": 0.15073542296886444,
      "learning_rate": 3.5388870669842755e-07,
      "loss": 0.0015,
      "step": 278020
    },
    {
      "epoch": 1.783198057934856,
      "grad_norm": 0.01688840240240097,
      "learning_rate": 3.5368191348764013e-07,
      "loss": 0.0012,
      "step": 278030
    },
    {
      "epoch": 1.7832621948286422,
      "grad_norm": 0.0731649175286293,
      "learning_rate": 3.534751784985396e-07,
      "loss": 0.0028,
      "step": 278040
    },
    {
      "epoch": 1.7833263317224284,
      "grad_norm": 0.02688288502395153,
      "learning_rate": 3.532685017337167e-07,
      "loss": 0.0013,
      "step": 278050
    },
    {
      "epoch": 1.7833904686162145,
      "grad_norm": 0.05125247314572334,
      "learning_rate": 3.5306188319575886e-07,
      "loss": 0.0007,
      "step": 278060
    },
    {
      "epoch": 1.7834546055100007,
      "grad_norm": 0.17364509403705597,
      "learning_rate": 3.5285532288725777e-07,
      "loss": 0.0014,
      "step": 278070
    },
    {
      "epoch": 1.7835187424037866,
      "grad_norm": 0.3780440390110016,
      "learning_rate": 3.526488208108003e-07,
      "loss": 0.001,
      "step": 278080
    },
    {
      "epoch": 1.7835828792975728,
      "grad_norm": 0.012333092279732227,
      "learning_rate": 3.52442376968975e-07,
      "loss": 0.0007,
      "step": 278090
    },
    {
      "epoch": 1.7836470161913587,
      "grad_norm": 0.0849262997508049,
      "learning_rate": 3.5223599136436806e-07,
      "loss": 0.0004,
      "step": 278100
    },
    {
      "epoch": 1.7837111530851448,
      "grad_norm": 0.03807647153735161,
      "learning_rate": 3.520296639995663e-07,
      "loss": 0.0003,
      "step": 278110
    },
    {
      "epoch": 1.783775289978931,
      "grad_norm": 0.033494483679533005,
      "learning_rate": 3.5182339487715377e-07,
      "loss": 0.0008,
      "step": 278120
    },
    {
      "epoch": 1.7838394268727171,
      "grad_norm": 0.1657082736492157,
      "learning_rate": 3.516171839997168e-07,
      "loss": 0.0009,
      "step": 278130
    },
    {
      "epoch": 1.7839035637665033,
      "grad_norm": 0.16438868641853333,
      "learning_rate": 3.514110313698388e-07,
      "loss": 0.0031,
      "step": 278140
    },
    {
      "epoch": 1.7839677006602894,
      "grad_norm": 0.3702642321586609,
      "learning_rate": 3.512049369901016e-07,
      "loss": 0.0013,
      "step": 278150
    },
    {
      "epoch": 1.7840318375540756,
      "grad_norm": 0.10200533270835876,
      "learning_rate": 3.509989008630904e-07,
      "loss": 0.0013,
      "step": 278160
    },
    {
      "epoch": 1.7840959744478615,
      "grad_norm": 0.08121436089277267,
      "learning_rate": 3.5079292299138535e-07,
      "loss": 0.0009,
      "step": 278170
    },
    {
      "epoch": 1.7841601113416476,
      "grad_norm": 0.044688694179058075,
      "learning_rate": 3.505870033775677e-07,
      "loss": 0.0009,
      "step": 278180
    },
    {
      "epoch": 1.7842242482354336,
      "grad_norm": 0.038585562258958817,
      "learning_rate": 3.5038114202421705e-07,
      "loss": 0.0015,
      "step": 278190
    },
    {
      "epoch": 1.7842883851292197,
      "grad_norm": 0.024512559175491333,
      "learning_rate": 3.5017533893391466e-07,
      "loss": 0.0005,
      "step": 278200
    },
    {
      "epoch": 1.7843525220230059,
      "grad_norm": 0.039037760347127914,
      "learning_rate": 3.4996959410923793e-07,
      "loss": 0.0011,
      "step": 278210
    },
    {
      "epoch": 1.784416658916792,
      "grad_norm": 0.05863253399729729,
      "learning_rate": 3.497639075527659e-07,
      "loss": 0.0018,
      "step": 278220
    },
    {
      "epoch": 1.7844807958105782,
      "grad_norm": 0.006160229444503784,
      "learning_rate": 3.4955827926707485e-07,
      "loss": 0.0006,
      "step": 278230
    },
    {
      "epoch": 1.7845449327043643,
      "grad_norm": 0.0604865700006485,
      "learning_rate": 3.4935270925474274e-07,
      "loss": 0.001,
      "step": 278240
    },
    {
      "epoch": 1.7846090695981502,
      "grad_norm": 0.07260777056217194,
      "learning_rate": 3.491471975183452e-07,
      "loss": 0.0014,
      "step": 278250
    },
    {
      "epoch": 1.7846732064919364,
      "grad_norm": 0.012632405385375023,
      "learning_rate": 3.48941744060457e-07,
      "loss": 0.0005,
      "step": 278260
    },
    {
      "epoch": 1.7847373433857225,
      "grad_norm": 0.1146724745631218,
      "learning_rate": 3.4873634888365206e-07,
      "loss": 0.0009,
      "step": 278270
    },
    {
      "epoch": 1.7848014802795085,
      "grad_norm": 0.1695510447025299,
      "learning_rate": 3.4853101199050553e-07,
      "loss": 0.0014,
      "step": 278280
    },
    {
      "epoch": 1.7848656171732946,
      "grad_norm": 0.006848647724837065,
      "learning_rate": 3.4832573338358986e-07,
      "loss": 0.0006,
      "step": 278290
    },
    {
      "epoch": 1.7849297540670808,
      "grad_norm": 0.0386105477809906,
      "learning_rate": 3.4812051306547637e-07,
      "loss": 0.0008,
      "step": 278300
    },
    {
      "epoch": 1.784993890960867,
      "grad_norm": 0.0190505962818861,
      "learning_rate": 3.479153510387384e-07,
      "loss": 0.0009,
      "step": 278310
    },
    {
      "epoch": 1.785058027854653,
      "grad_norm": 0.048402898013591766,
      "learning_rate": 3.4771024730594627e-07,
      "loss": 0.001,
      "step": 278320
    },
    {
      "epoch": 1.7851221647484392,
      "grad_norm": 0.025702862069010735,
      "learning_rate": 3.47505201869669e-07,
      "loss": 0.0007,
      "step": 278330
    },
    {
      "epoch": 1.7851863016422251,
      "grad_norm": 0.03252828121185303,
      "learning_rate": 3.473002147324761e-07,
      "loss": 0.0012,
      "step": 278340
    },
    {
      "epoch": 1.7852504385360113,
      "grad_norm": 0.02330513298511505,
      "learning_rate": 3.470952858969379e-07,
      "loss": 0.0007,
      "step": 278350
    },
    {
      "epoch": 1.7853145754297972,
      "grad_norm": 0.023918023332953453,
      "learning_rate": 3.4689041536562053e-07,
      "loss": 0.0007,
      "step": 278360
    },
    {
      "epoch": 1.7853787123235834,
      "grad_norm": 0.06483355164527893,
      "learning_rate": 3.466856031410926e-07,
      "loss": 0.0024,
      "step": 278370
    },
    {
      "epoch": 1.7854428492173695,
      "grad_norm": 0.04346863552927971,
      "learning_rate": 3.4648084922591806e-07,
      "loss": 0.0007,
      "step": 278380
    },
    {
      "epoch": 1.7855069861111557,
      "grad_norm": 0.06587924808263779,
      "learning_rate": 3.4627615362266554e-07,
      "loss": 0.0012,
      "step": 278390
    },
    {
      "epoch": 1.7855711230049418,
      "grad_norm": 0.05727564916014671,
      "learning_rate": 3.4607151633389903e-07,
      "loss": 0.0012,
      "step": 278400
    },
    {
      "epoch": 1.785635259898728,
      "grad_norm": 0.2357838749885559,
      "learning_rate": 3.4586693736218204e-07,
      "loss": 0.0016,
      "step": 278410
    },
    {
      "epoch": 1.785699396792514,
      "grad_norm": 0.002354859374463558,
      "learning_rate": 3.4566241671007806e-07,
      "loss": 0.0005,
      "step": 278420
    },
    {
      "epoch": 1.7857635336863,
      "grad_norm": 0.3780546188354492,
      "learning_rate": 3.4545795438015115e-07,
      "loss": 0.0043,
      "step": 278430
    },
    {
      "epoch": 1.7858276705800862,
      "grad_norm": 0.04920445755124092,
      "learning_rate": 3.452535503749621e-07,
      "loss": 0.0016,
      "step": 278440
    },
    {
      "epoch": 1.785891807473872,
      "grad_norm": 0.13359010219573975,
      "learning_rate": 3.450492046970727e-07,
      "loss": 0.0005,
      "step": 278450
    },
    {
      "epoch": 1.7859559443676583,
      "grad_norm": 0.024329397827386856,
      "learning_rate": 3.4484491734904423e-07,
      "loss": 0.0012,
      "step": 278460
    },
    {
      "epoch": 1.7860200812614444,
      "grad_norm": 0.03433268144726753,
      "learning_rate": 3.446406883334358e-07,
      "loss": 0.0042,
      "step": 278470
    },
    {
      "epoch": 1.7860842181552306,
      "grad_norm": 0.06557345390319824,
      "learning_rate": 3.4443651765280697e-07,
      "loss": 0.0007,
      "step": 278480
    },
    {
      "epoch": 1.7861483550490167,
      "grad_norm": 0.0781419649720192,
      "learning_rate": 3.442324053097146e-07,
      "loss": 0.0011,
      "step": 278490
    },
    {
      "epoch": 1.7862124919428028,
      "grad_norm": 0.009028265252709389,
      "learning_rate": 3.440283513067183e-07,
      "loss": 0.0011,
      "step": 278500
    },
    {
      "epoch": 1.7862766288365888,
      "grad_norm": 0.04159869998693466,
      "learning_rate": 3.4382435564637496e-07,
      "loss": 0.0014,
      "step": 278510
    },
    {
      "epoch": 1.786340765730375,
      "grad_norm": 0.06454797834157944,
      "learning_rate": 3.4362041833123973e-07,
      "loss": 0.0005,
      "step": 278520
    },
    {
      "epoch": 1.7864049026241609,
      "grad_norm": 0.04843491315841675,
      "learning_rate": 3.4341653936386776e-07,
      "loss": 0.0012,
      "step": 278530
    },
    {
      "epoch": 1.786469039517947,
      "grad_norm": 0.017590438947081566,
      "learning_rate": 3.4321271874681595e-07,
      "loss": 0.0009,
      "step": 278540
    },
    {
      "epoch": 1.7865331764117331,
      "grad_norm": 0.08637532591819763,
      "learning_rate": 3.430089564826361e-07,
      "loss": 0.0009,
      "step": 278550
    },
    {
      "epoch": 1.7865973133055193,
      "grad_norm": 0.09738405793905258,
      "learning_rate": 3.4280525257388284e-07,
      "loss": 0.0012,
      "step": 278560
    },
    {
      "epoch": 1.7866614501993054,
      "grad_norm": 0.005397073924541473,
      "learning_rate": 3.4260160702310697e-07,
      "loss": 0.0009,
      "step": 278570
    },
    {
      "epoch": 1.7867255870930916,
      "grad_norm": 0.0016158544458448887,
      "learning_rate": 3.4239801983286305e-07,
      "loss": 0.0006,
      "step": 278580
    },
    {
      "epoch": 1.7867897239868777,
      "grad_norm": 0.004870242904871702,
      "learning_rate": 3.421944910057001e-07,
      "loss": 0.0005,
      "step": 278590
    },
    {
      "epoch": 1.7868538608806637,
      "grad_norm": 0.07605539262294769,
      "learning_rate": 3.4199102054416843e-07,
      "loss": 0.0014,
      "step": 278600
    },
    {
      "epoch": 1.7869179977744498,
      "grad_norm": 0.023373080417513847,
      "learning_rate": 3.4178760845081925e-07,
      "loss": 0.0038,
      "step": 278610
    },
    {
      "epoch": 1.7869821346682357,
      "grad_norm": 0.12973308563232422,
      "learning_rate": 3.415842547282e-07,
      "loss": 0.0012,
      "step": 278620
    },
    {
      "epoch": 1.787046271562022,
      "grad_norm": 0.07479102164506912,
      "learning_rate": 3.413809593788597e-07,
      "loss": 0.0015,
      "step": 278630
    },
    {
      "epoch": 1.787110408455808,
      "grad_norm": 0.009775402024388313,
      "learning_rate": 3.4117772240534474e-07,
      "loss": 0.0008,
      "step": 278640
    },
    {
      "epoch": 1.7871745453495942,
      "grad_norm": 0.17945508658885956,
      "learning_rate": 3.4097454381020356e-07,
      "loss": 0.0009,
      "step": 278650
    },
    {
      "epoch": 1.7872386822433803,
      "grad_norm": 0.20023804903030396,
      "learning_rate": 3.4077142359598024e-07,
      "loss": 0.0011,
      "step": 278660
    },
    {
      "epoch": 1.7873028191371665,
      "grad_norm": 0.09040382504463196,
      "learning_rate": 3.4056836176522166e-07,
      "loss": 0.0008,
      "step": 278670
    },
    {
      "epoch": 1.7873669560309524,
      "grad_norm": 0.041846659034490585,
      "learning_rate": 3.403653583204708e-07,
      "loss": 0.0009,
      "step": 278680
    },
    {
      "epoch": 1.7874310929247386,
      "grad_norm": 0.04396964609622955,
      "learning_rate": 3.4016241326427223e-07,
      "loss": 0.0012,
      "step": 278690
    },
    {
      "epoch": 1.7874952298185247,
      "grad_norm": 0.053148720413446426,
      "learning_rate": 3.399595265991695e-07,
      "loss": 0.0012,
      "step": 278700
    },
    {
      "epoch": 1.7875593667123106,
      "grad_norm": 0.053744640201330185,
      "learning_rate": 3.397566983277045e-07,
      "loss": 0.0013,
      "step": 278710
    },
    {
      "epoch": 1.7876235036060968,
      "grad_norm": 0.1659751683473587,
      "learning_rate": 3.39553928452418e-07,
      "loss": 0.001,
      "step": 278720
    },
    {
      "epoch": 1.787687640499883,
      "grad_norm": 0.12354079633951187,
      "learning_rate": 3.393512169758517e-07,
      "loss": 0.0008,
      "step": 278730
    },
    {
      "epoch": 1.787751777393669,
      "grad_norm": 0.06363257020711899,
      "learning_rate": 3.3914856390054594e-07,
      "loss": 0.001,
      "step": 278740
    },
    {
      "epoch": 1.7878159142874552,
      "grad_norm": 0.05239519476890564,
      "learning_rate": 3.3894596922903923e-07,
      "loss": 0.0008,
      "step": 278750
    },
    {
      "epoch": 1.7878800511812414,
      "grad_norm": 0.035413049161434174,
      "learning_rate": 3.387434329638711e-07,
      "loss": 0.0033,
      "step": 278760
    },
    {
      "epoch": 1.7879441880750273,
      "grad_norm": 0.08839958161115646,
      "learning_rate": 3.385409551075791e-07,
      "loss": 0.0009,
      "step": 278770
    },
    {
      "epoch": 1.7880083249688135,
      "grad_norm": 0.021664628759026527,
      "learning_rate": 3.383385356627006e-07,
      "loss": 0.0213,
      "step": 278780
    },
    {
      "epoch": 1.7880724618625994,
      "grad_norm": 0.07869093120098114,
      "learning_rate": 3.381361746317713e-07,
      "loss": 0.002,
      "step": 278790
    },
    {
      "epoch": 1.7881365987563855,
      "grad_norm": 0.07586157321929932,
      "learning_rate": 3.379338720173275e-07,
      "loss": 0.0009,
      "step": 278800
    },
    {
      "epoch": 1.7882007356501717,
      "grad_norm": 0.10144965350627899,
      "learning_rate": 3.3773162782190505e-07,
      "loss": 0.0009,
      "step": 278810
    },
    {
      "epoch": 1.7882648725439578,
      "grad_norm": 0.1888962835073471,
      "learning_rate": 3.3752944204803626e-07,
      "loss": 0.0011,
      "step": 278820
    },
    {
      "epoch": 1.788329009437744,
      "grad_norm": 0.002420370001345873,
      "learning_rate": 3.373273146982564e-07,
      "loss": 0.0013,
      "step": 278830
    },
    {
      "epoch": 1.7883931463315301,
      "grad_norm": 0.0930965393781662,
      "learning_rate": 3.371252457750973e-07,
      "loss": 0.0015,
      "step": 278840
    },
    {
      "epoch": 1.7884572832253163,
      "grad_norm": 0.08777619153261185,
      "learning_rate": 3.369232352810919e-07,
      "loss": 0.0014,
      "step": 278850
    },
    {
      "epoch": 1.7885214201191022,
      "grad_norm": 0.10435951501131058,
      "learning_rate": 3.3672128321877104e-07,
      "loss": 0.0011,
      "step": 278860
    },
    {
      "epoch": 1.7885855570128883,
      "grad_norm": 0.013379015028476715,
      "learning_rate": 3.365193895906649e-07,
      "loss": 0.0014,
      "step": 278870
    },
    {
      "epoch": 1.7886496939066743,
      "grad_norm": 0.09842795878648758,
      "learning_rate": 3.3631755439930415e-07,
      "loss": 0.0019,
      "step": 278880
    },
    {
      "epoch": 1.7887138308004604,
      "grad_norm": 0.04331498220562935,
      "learning_rate": 3.3611577764721735e-07,
      "loss": 0.001,
      "step": 278890
    },
    {
      "epoch": 1.7887779676942466,
      "grad_norm": 0.14871858060359955,
      "learning_rate": 3.359140593369331e-07,
      "loss": 0.0009,
      "step": 278900
    },
    {
      "epoch": 1.7888421045880327,
      "grad_norm": 0.0670400857925415,
      "learning_rate": 3.357123994709782e-07,
      "loss": 0.0012,
      "step": 278910
    },
    {
      "epoch": 1.7889062414818189,
      "grad_norm": 0.12930205464363098,
      "learning_rate": 3.3551079805188183e-07,
      "loss": 0.0007,
      "step": 278920
    },
    {
      "epoch": 1.788970378375605,
      "grad_norm": 0.025455178692936897,
      "learning_rate": 3.353092550821685e-07,
      "loss": 0.001,
      "step": 278930
    },
    {
      "epoch": 1.789034515269391,
      "grad_norm": 0.05392056703567505,
      "learning_rate": 3.351077705643641e-07,
      "loss": 0.001,
      "step": 278940
    },
    {
      "epoch": 1.789098652163177,
      "grad_norm": 0.05469081178307533,
      "learning_rate": 3.3490634450099203e-07,
      "loss": 0.0007,
      "step": 278950
    },
    {
      "epoch": 1.789162789056963,
      "grad_norm": 0.03116176836192608,
      "learning_rate": 3.347049768945787e-07,
      "loss": 0.001,
      "step": 278960
    },
    {
      "epoch": 1.7892269259507492,
      "grad_norm": 0.08433020114898682,
      "learning_rate": 3.3450366774764655e-07,
      "loss": 0.002,
      "step": 278970
    },
    {
      "epoch": 1.7892910628445353,
      "grad_norm": 0.07080574333667755,
      "learning_rate": 3.343024170627174e-07,
      "loss": 0.0012,
      "step": 278980
    },
    {
      "epoch": 1.7893551997383215,
      "grad_norm": 0.2883979082107544,
      "learning_rate": 3.3410122484231254e-07,
      "loss": 0.0021,
      "step": 278990
    },
    {
      "epoch": 1.7894193366321076,
      "grad_norm": 0.4545295536518097,
      "learning_rate": 3.3390009108895503e-07,
      "loss": 0.0072,
      "step": 279000
    },
    {
      "epoch": 1.7894834735258938,
      "grad_norm": 0.20828689634799957,
      "learning_rate": 3.336990158051645e-07,
      "loss": 0.0018,
      "step": 279010
    },
    {
      "epoch": 1.78954761041968,
      "grad_norm": 0.10651051998138428,
      "learning_rate": 3.334979989934589e-07,
      "loss": 0.0008,
      "step": 279020
    },
    {
      "epoch": 1.7896117473134658,
      "grad_norm": 0.06715662032365799,
      "learning_rate": 3.332970406563596e-07,
      "loss": 0.0008,
      "step": 279030
    },
    {
      "epoch": 1.789675884207252,
      "grad_norm": 0.05971822887659073,
      "learning_rate": 3.330961407963834e-07,
      "loss": 0.0011,
      "step": 279040
    },
    {
      "epoch": 1.789740021101038,
      "grad_norm": 0.04333239048719406,
      "learning_rate": 3.3289529941604836e-07,
      "loss": 0.0006,
      "step": 279050
    },
    {
      "epoch": 1.789804157994824,
      "grad_norm": 0.3848743736743927,
      "learning_rate": 3.3269451651786966e-07,
      "loss": 0.0014,
      "step": 279060
    },
    {
      "epoch": 1.7898682948886102,
      "grad_norm": 0.0060179815627634525,
      "learning_rate": 3.324937921043653e-07,
      "loss": 0.0007,
      "step": 279070
    },
    {
      "epoch": 1.7899324317823964,
      "grad_norm": 0.08131309598684311,
      "learning_rate": 3.322931261780493e-07,
      "loss": 0.0008,
      "step": 279080
    },
    {
      "epoch": 1.7899965686761825,
      "grad_norm": 0.16063711047172546,
      "learning_rate": 3.3209251874143645e-07,
      "loss": 0.0005,
      "step": 279090
    },
    {
      "epoch": 1.7900607055699687,
      "grad_norm": 0.11836498230695724,
      "learning_rate": 3.318919697970396e-07,
      "loss": 0.0018,
      "step": 279100
    },
    {
      "epoch": 1.7901248424637546,
      "grad_norm": 0.012969082221388817,
      "learning_rate": 3.3169147934737355e-07,
      "loss": 0.0018,
      "step": 279110
    },
    {
      "epoch": 1.7901889793575407,
      "grad_norm": 0.13354268670082092,
      "learning_rate": 3.3149104739494954e-07,
      "loss": 0.0009,
      "step": 279120
    },
    {
      "epoch": 1.7902531162513269,
      "grad_norm": 0.023537177592515945,
      "learning_rate": 3.312906739422794e-07,
      "loss": 0.0004,
      "step": 279130
    },
    {
      "epoch": 1.7903172531451128,
      "grad_norm": 0.018479079008102417,
      "learning_rate": 3.310903589918729e-07,
      "loss": 0.0012,
      "step": 279140
    },
    {
      "epoch": 1.790381390038899,
      "grad_norm": 0.07216761261224747,
      "learning_rate": 3.308901025462419e-07,
      "loss": 0.0009,
      "step": 279150
    },
    {
      "epoch": 1.790445526932685,
      "grad_norm": 0.012507503852248192,
      "learning_rate": 3.306899046078949e-07,
      "loss": 0.0005,
      "step": 279160
    },
    {
      "epoch": 1.7905096638264713,
      "grad_norm": 0.05727556720376015,
      "learning_rate": 3.3048976517933984e-07,
      "loss": 0.0012,
      "step": 279170
    },
    {
      "epoch": 1.7905738007202574,
      "grad_norm": 0.06608857214450836,
      "learning_rate": 3.302896842630854e-07,
      "loss": 0.0014,
      "step": 279180
    },
    {
      "epoch": 1.7906379376140436,
      "grad_norm": 0.09351908415555954,
      "learning_rate": 3.300896618616395e-07,
      "loss": 0.0006,
      "step": 279190
    },
    {
      "epoch": 1.7907020745078295,
      "grad_norm": 0.09092344343662262,
      "learning_rate": 3.298896979775068e-07,
      "loss": 0.0011,
      "step": 279200
    },
    {
      "epoch": 1.7907662114016156,
      "grad_norm": 0.13009698688983917,
      "learning_rate": 3.296897926131937e-07,
      "loss": 0.0008,
      "step": 279210
    },
    {
      "epoch": 1.7908303482954016,
      "grad_norm": 0.05622050166130066,
      "learning_rate": 3.2948994577120585e-07,
      "loss": 0.001,
      "step": 279220
    },
    {
      "epoch": 1.7908944851891877,
      "grad_norm": 0.03919670730829239,
      "learning_rate": 3.292901574540469e-07,
      "loss": 0.0005,
      "step": 279230
    },
    {
      "epoch": 1.7909586220829739,
      "grad_norm": 0.08288165181875229,
      "learning_rate": 3.2909042766422037e-07,
      "loss": 0.0007,
      "step": 279240
    },
    {
      "epoch": 1.79102275897676,
      "grad_norm": 0.09894700348377228,
      "learning_rate": 3.2889075640422876e-07,
      "loss": 0.0013,
      "step": 279250
    },
    {
      "epoch": 1.7910868958705461,
      "grad_norm": 0.14007550477981567,
      "learning_rate": 3.2869114367657437e-07,
      "loss": 0.0012,
      "step": 279260
    },
    {
      "epoch": 1.7911510327643323,
      "grad_norm": 0.07528212666511536,
      "learning_rate": 3.284915894837587e-07,
      "loss": 0.0007,
      "step": 279270
    },
    {
      "epoch": 1.7912151696581184,
      "grad_norm": 0.036512792110443115,
      "learning_rate": 3.282920938282824e-07,
      "loss": 0.0007,
      "step": 279280
    },
    {
      "epoch": 1.7912793065519044,
      "grad_norm": 0.08713310956954956,
      "learning_rate": 3.280926567126441e-07,
      "loss": 0.0007,
      "step": 279290
    },
    {
      "epoch": 1.7913434434456905,
      "grad_norm": 0.03951220214366913,
      "learning_rate": 3.278932781393446e-07,
      "loss": 0.0008,
      "step": 279300
    },
    {
      "epoch": 1.7914075803394764,
      "grad_norm": 0.2905653715133667,
      "learning_rate": 3.276939581108812e-07,
      "loss": 0.0018,
      "step": 279310
    },
    {
      "epoch": 1.7914717172332626,
      "grad_norm": 0.030956579372286797,
      "learning_rate": 3.2749469662975096e-07,
      "loss": 0.0009,
      "step": 279320
    },
    {
      "epoch": 1.7915358541270487,
      "grad_norm": 0.11684050410985947,
      "learning_rate": 3.272954936984524e-07,
      "loss": 0.0011,
      "step": 279330
    },
    {
      "epoch": 1.791599991020835,
      "grad_norm": 0.307887464761734,
      "learning_rate": 3.270963493194812e-07,
      "loss": 0.0015,
      "step": 279340
    },
    {
      "epoch": 1.791664127914621,
      "grad_norm": 0.049778252840042114,
      "learning_rate": 3.2689726349533213e-07,
      "loss": 0.0008,
      "step": 279350
    },
    {
      "epoch": 1.7917282648084072,
      "grad_norm": 0.04836338013410568,
      "learning_rate": 3.266982362284993e-07,
      "loss": 0.0012,
      "step": 279360
    },
    {
      "epoch": 1.7917924017021931,
      "grad_norm": 0.03864509239792824,
      "learning_rate": 3.2649926752147843e-07,
      "loss": 0.0011,
      "step": 279370
    },
    {
      "epoch": 1.7918565385959793,
      "grad_norm": 0.004542256239801645,
      "learning_rate": 3.2630035737676203e-07,
      "loss": 0.0007,
      "step": 279380
    },
    {
      "epoch": 1.7919206754897652,
      "grad_norm": 0.26342859864234924,
      "learning_rate": 3.2610150579684195e-07,
      "loss": 0.002,
      "step": 279390
    },
    {
      "epoch": 1.7919848123835513,
      "grad_norm": 0.07165306806564331,
      "learning_rate": 3.259027127842096e-07,
      "loss": 0.002,
      "step": 279400
    },
    {
      "epoch": 1.7920489492773375,
      "grad_norm": 0.09093828499317169,
      "learning_rate": 3.257039783413579e-07,
      "loss": 0.0008,
      "step": 279410
    },
    {
      "epoch": 1.7921130861711236,
      "grad_norm": 0.019808035343885422,
      "learning_rate": 3.2550530247077605e-07,
      "loss": 0.0007,
      "step": 279420
    },
    {
      "epoch": 1.7921772230649098,
      "grad_norm": 0.02690022811293602,
      "learning_rate": 3.253066851749526e-07,
      "loss": 0.0016,
      "step": 279430
    },
    {
      "epoch": 1.792241359958696,
      "grad_norm": 0.04569307714700699,
      "learning_rate": 3.2510812645637824e-07,
      "loss": 0.0008,
      "step": 279440
    },
    {
      "epoch": 1.792305496852482,
      "grad_norm": 0.21289020776748657,
      "learning_rate": 3.2490962631754e-07,
      "loss": 0.0016,
      "step": 279450
    },
    {
      "epoch": 1.792369633746268,
      "grad_norm": 0.09713505208492279,
      "learning_rate": 3.247111847609252e-07,
      "loss": 0.0009,
      "step": 279460
    },
    {
      "epoch": 1.7924337706400542,
      "grad_norm": 0.04144668206572533,
      "learning_rate": 3.2451280178902033e-07,
      "loss": 0.0003,
      "step": 279470
    },
    {
      "epoch": 1.79249790753384,
      "grad_norm": 0.21650367975234985,
      "learning_rate": 3.2431447740431224e-07,
      "loss": 0.0025,
      "step": 279480
    },
    {
      "epoch": 1.7925620444276262,
      "grad_norm": 0.04582495987415314,
      "learning_rate": 3.2411621160928497e-07,
      "loss": 0.001,
      "step": 279490
    },
    {
      "epoch": 1.7926261813214124,
      "grad_norm": 0.11708521842956543,
      "learning_rate": 3.2391800440642384e-07,
      "loss": 0.001,
      "step": 279500
    },
    {
      "epoch": 1.7926903182151985,
      "grad_norm": 0.016406027600169182,
      "learning_rate": 3.237198557982113e-07,
      "loss": 0.0011,
      "step": 279510
    },
    {
      "epoch": 1.7927544551089847,
      "grad_norm": 0.0726175531744957,
      "learning_rate": 3.2352176578713144e-07,
      "loss": 0.0007,
      "step": 279520
    },
    {
      "epoch": 1.7928185920027708,
      "grad_norm": 0.1482061743736267,
      "learning_rate": 3.233237343756662e-07,
      "loss": 0.0027,
      "step": 279530
    },
    {
      "epoch": 1.792882728896557,
      "grad_norm": 0.0564088337123394,
      "learning_rate": 3.2312576156629747e-07,
      "loss": 0.0008,
      "step": 279540
    },
    {
      "epoch": 1.792946865790343,
      "grad_norm": 0.09703831374645233,
      "learning_rate": 3.229278473615044e-07,
      "loss": 0.001,
      "step": 279550
    },
    {
      "epoch": 1.793011002684129,
      "grad_norm": 0.03012095019221306,
      "learning_rate": 3.227299917637683e-07,
      "loss": 0.0014,
      "step": 279560
    },
    {
      "epoch": 1.793075139577915,
      "grad_norm": 0.023541104048490524,
      "learning_rate": 3.2253219477556886e-07,
      "loss": 0.0008,
      "step": 279570
    },
    {
      "epoch": 1.7931392764717011,
      "grad_norm": 0.059293895959854126,
      "learning_rate": 3.22334456399383e-07,
      "loss": 0.001,
      "step": 279580
    },
    {
      "epoch": 1.7932034133654873,
      "grad_norm": 0.03531581535935402,
      "learning_rate": 3.221367766376893e-07,
      "loss": 0.0015,
      "step": 279590
    },
    {
      "epoch": 1.7932675502592734,
      "grad_norm": 0.014433217234909534,
      "learning_rate": 3.2193915549296575e-07,
      "loss": 0.0011,
      "step": 279600
    },
    {
      "epoch": 1.7933316871530596,
      "grad_norm": 0.030258553102612495,
      "learning_rate": 3.2174159296768816e-07,
      "loss": 0.0009,
      "step": 279610
    },
    {
      "epoch": 1.7933958240468457,
      "grad_norm": 0.04009813070297241,
      "learning_rate": 3.2154408906433174e-07,
      "loss": 0.001,
      "step": 279620
    },
    {
      "epoch": 1.7934599609406316,
      "grad_norm": 0.08422304689884186,
      "learning_rate": 3.2134664378537075e-07,
      "loss": 0.0014,
      "step": 279630
    },
    {
      "epoch": 1.7935240978344178,
      "grad_norm": 0.02794818766415119,
      "learning_rate": 3.2114925713328084e-07,
      "loss": 0.0011,
      "step": 279640
    },
    {
      "epoch": 1.7935882347282037,
      "grad_norm": 0.1486392766237259,
      "learning_rate": 3.209519291105351e-07,
      "loss": 0.0008,
      "step": 279650
    },
    {
      "epoch": 1.7936523716219899,
      "grad_norm": 0.033261239528656006,
      "learning_rate": 3.207546597196054e-07,
      "loss": 0.0012,
      "step": 279660
    },
    {
      "epoch": 1.793716508515776,
      "grad_norm": 0.06807536631822586,
      "learning_rate": 3.2055744896296313e-07,
      "loss": 0.0016,
      "step": 279670
    },
    {
      "epoch": 1.7937806454095622,
      "grad_norm": 0.025683945044875145,
      "learning_rate": 3.2036029684308133e-07,
      "loss": 0.0007,
      "step": 279680
    },
    {
      "epoch": 1.7938447823033483,
      "grad_norm": 0.012448211200535297,
      "learning_rate": 3.201632033624297e-07,
      "loss": 0.0017,
      "step": 279690
    },
    {
      "epoch": 1.7939089191971345,
      "grad_norm": 0.00342944567091763,
      "learning_rate": 3.1996616852347673e-07,
      "loss": 0.001,
      "step": 279700
    },
    {
      "epoch": 1.7939730560909206,
      "grad_norm": 0.007278515491634607,
      "learning_rate": 3.197691923286933e-07,
      "loss": 0.0014,
      "step": 279710
    },
    {
      "epoch": 1.7940371929847065,
      "grad_norm": 0.05002468824386597,
      "learning_rate": 3.195722747805463e-07,
      "loss": 0.0005,
      "step": 279720
    },
    {
      "epoch": 1.7941013298784927,
      "grad_norm": 0.03734104707837105,
      "learning_rate": 3.193754158815043e-07,
      "loss": 0.0011,
      "step": 279730
    },
    {
      "epoch": 1.7941654667722786,
      "grad_norm": 0.06979603320360184,
      "learning_rate": 3.191786156340326e-07,
      "loss": 0.0003,
      "step": 279740
    },
    {
      "epoch": 1.7942296036660648,
      "grad_norm": 0.0816522017121315,
      "learning_rate": 3.189818740405992e-07,
      "loss": 0.0009,
      "step": 279750
    },
    {
      "epoch": 1.794293740559851,
      "grad_norm": 0.10368962585926056,
      "learning_rate": 3.1878519110366824e-07,
      "loss": 0.001,
      "step": 279760
    },
    {
      "epoch": 1.794357877453637,
      "grad_norm": 0.0599248968064785,
      "learning_rate": 3.185885668257044e-07,
      "loss": 0.001,
      "step": 279770
    },
    {
      "epoch": 1.7944220143474232,
      "grad_norm": 0.11840607225894928,
      "learning_rate": 3.183920012091707e-07,
      "loss": 0.0006,
      "step": 279780
    },
    {
      "epoch": 1.7944861512412094,
      "grad_norm": 0.24062080681324005,
      "learning_rate": 3.1819549425653186e-07,
      "loss": 0.0024,
      "step": 279790
    },
    {
      "epoch": 1.7945502881349953,
      "grad_norm": 0.08701277524232864,
      "learning_rate": 3.1799904597024976e-07,
      "loss": 0.0005,
      "step": 279800
    },
    {
      "epoch": 1.7946144250287814,
      "grad_norm": 0.11865556240081787,
      "learning_rate": 3.1780265635278583e-07,
      "loss": 0.0005,
      "step": 279810
    },
    {
      "epoch": 1.7946785619225676,
      "grad_norm": 0.10071233659982681,
      "learning_rate": 3.1760632540659975e-07,
      "loss": 0.0007,
      "step": 279820
    },
    {
      "epoch": 1.7947426988163535,
      "grad_norm": 0.029392078518867493,
      "learning_rate": 3.1741005313415394e-07,
      "loss": 0.0007,
      "step": 279830
    },
    {
      "epoch": 1.7948068357101397,
      "grad_norm": 0.08132757991552353,
      "learning_rate": 3.172138395379065e-07,
      "loss": 0.0018,
      "step": 279840
    },
    {
      "epoch": 1.7948709726039258,
      "grad_norm": 0.1531228870153427,
      "learning_rate": 3.1701768462031603e-07,
      "loss": 0.0013,
      "step": 279850
    },
    {
      "epoch": 1.794935109497712,
      "grad_norm": 0.07538601756095886,
      "learning_rate": 3.168215883838416e-07,
      "loss": 0.001,
      "step": 279860
    },
    {
      "epoch": 1.794999246391498,
      "grad_norm": 0.0838082954287529,
      "learning_rate": 3.1662555083093906e-07,
      "loss": 0.0007,
      "step": 279870
    },
    {
      "epoch": 1.7950633832852843,
      "grad_norm": 0.01654684729874134,
      "learning_rate": 3.1642957196406586e-07,
      "loss": 0.0011,
      "step": 279880
    },
    {
      "epoch": 1.7951275201790702,
      "grad_norm": 0.04836370795965195,
      "learning_rate": 3.1623365178567677e-07,
      "loss": 0.0008,
      "step": 279890
    },
    {
      "epoch": 1.7951916570728563,
      "grad_norm": 0.04304756596684456,
      "learning_rate": 3.160377902982287e-07,
      "loss": 0.0007,
      "step": 279900
    },
    {
      "epoch": 1.7952557939666423,
      "grad_norm": 0.07920753955841064,
      "learning_rate": 3.1584198750417403e-07,
      "loss": 0.0012,
      "step": 279910
    },
    {
      "epoch": 1.7953199308604284,
      "grad_norm": 0.07729573547840118,
      "learning_rate": 3.1564624340596704e-07,
      "loss": 0.0011,
      "step": 279920
    },
    {
      "epoch": 1.7953840677542146,
      "grad_norm": 0.13962076604366302,
      "learning_rate": 3.154505580060596e-07,
      "loss": 0.0013,
      "step": 279930
    },
    {
      "epoch": 1.7954482046480007,
      "grad_norm": 0.09074907004833221,
      "learning_rate": 3.152549313069059e-07,
      "loss": 0.0009,
      "step": 279940
    },
    {
      "epoch": 1.7955123415417868,
      "grad_norm": 0.0053997826762497425,
      "learning_rate": 3.1505936331095557e-07,
      "loss": 0.0005,
      "step": 279950
    },
    {
      "epoch": 1.795576478435573,
      "grad_norm": 0.24059101939201355,
      "learning_rate": 3.1486385402066e-07,
      "loss": 0.002,
      "step": 279960
    },
    {
      "epoch": 1.7956406153293591,
      "grad_norm": 0.04343503713607788,
      "learning_rate": 3.1466840343846785e-07,
      "loss": 0.0014,
      "step": 279970
    },
    {
      "epoch": 1.795704752223145,
      "grad_norm": 0.16977642476558685,
      "learning_rate": 3.144730115668304e-07,
      "loss": 0.0015,
      "step": 279980
    },
    {
      "epoch": 1.7957688891169312,
      "grad_norm": 0.026483383029699326,
      "learning_rate": 3.142776784081941e-07,
      "loss": 0.0005,
      "step": 279990
    },
    {
      "epoch": 1.7958330260107171,
      "grad_norm": 0.22658784687519073,
      "learning_rate": 3.14082403965007e-07,
      "loss": 0.0012,
      "step": 280000
    },
    {
      "epoch": 1.7958971629045033,
      "grad_norm": 0.027925442904233932,
      "learning_rate": 3.138871882397171e-07,
      "loss": 0.0009,
      "step": 280010
    },
    {
      "epoch": 1.7959612997982894,
      "grad_norm": 0.06454706937074661,
      "learning_rate": 3.1369203123476964e-07,
      "loss": 0.0016,
      "step": 280020
    },
    {
      "epoch": 1.7960254366920756,
      "grad_norm": 0.056851208209991455,
      "learning_rate": 3.134969329526105e-07,
      "loss": 0.0005,
      "step": 280030
    },
    {
      "epoch": 1.7960895735858617,
      "grad_norm": 0.03306282311677933,
      "learning_rate": 3.133018933956833e-07,
      "loss": 0.0004,
      "step": 280040
    },
    {
      "epoch": 1.796153710479648,
      "grad_norm": 0.018016841262578964,
      "learning_rate": 3.131069125664332e-07,
      "loss": 0.0005,
      "step": 280050
    },
    {
      "epoch": 1.7962178473734338,
      "grad_norm": 0.9242339134216309,
      "learning_rate": 3.1291199046730393e-07,
      "loss": 0.0069,
      "step": 280060
    },
    {
      "epoch": 1.79628198426722,
      "grad_norm": 0.01819530874490738,
      "learning_rate": 3.1271712710073675e-07,
      "loss": 0.0009,
      "step": 280070
    },
    {
      "epoch": 1.796346121161006,
      "grad_norm": 0.013147273100912571,
      "learning_rate": 3.1252232246917314e-07,
      "loss": 0.0006,
      "step": 280080
    },
    {
      "epoch": 1.796410258054792,
      "grad_norm": 0.011244059540331364,
      "learning_rate": 3.1232757657505605e-07,
      "loss": 0.0008,
      "step": 280090
    },
    {
      "epoch": 1.7964743949485782,
      "grad_norm": 0.036678630858659744,
      "learning_rate": 3.1213288942082474e-07,
      "loss": 0.0008,
      "step": 280100
    },
    {
      "epoch": 1.7965385318423643,
      "grad_norm": 0.03329041227698326,
      "learning_rate": 3.119382610089183e-07,
      "loss": 0.0003,
      "step": 280110
    },
    {
      "epoch": 1.7966026687361505,
      "grad_norm": 0.059137530624866486,
      "learning_rate": 3.117436913417754e-07,
      "loss": 0.0012,
      "step": 280120
    },
    {
      "epoch": 1.7966668056299366,
      "grad_norm": 0.05240783840417862,
      "learning_rate": 3.115491804218351e-07,
      "loss": 0.0009,
      "step": 280130
    },
    {
      "epoch": 1.7967309425237228,
      "grad_norm": 0.14184416830539703,
      "learning_rate": 3.1135472825153447e-07,
      "loss": 0.0016,
      "step": 280140
    },
    {
      "epoch": 1.7967950794175087,
      "grad_norm": 0.05168509483337402,
      "learning_rate": 3.111603348333098e-07,
      "loss": 0.0007,
      "step": 280150
    },
    {
      "epoch": 1.7968592163112949,
      "grad_norm": 0.044936347752809525,
      "learning_rate": 3.109660001695974e-07,
      "loss": 0.0017,
      "step": 280160
    },
    {
      "epoch": 1.7969233532050808,
      "grad_norm": 0.04893936589360237,
      "learning_rate": 3.107717242628322e-07,
      "loss": 0.0006,
      "step": 280170
    },
    {
      "epoch": 1.796987490098867,
      "grad_norm": 0.07789768278598785,
      "learning_rate": 3.105775071154493e-07,
      "loss": 0.0026,
      "step": 280180
    },
    {
      "epoch": 1.797051626992653,
      "grad_norm": 0.03610196337103844,
      "learning_rate": 3.1038334872988016e-07,
      "loss": 0.0011,
      "step": 280190
    },
    {
      "epoch": 1.7971157638864392,
      "grad_norm": 0.003820534097030759,
      "learning_rate": 3.1018924910856064e-07,
      "loss": 0.0005,
      "step": 280200
    },
    {
      "epoch": 1.7971799007802254,
      "grad_norm": 0.032815784215927124,
      "learning_rate": 3.099952082539209e-07,
      "loss": 0.0007,
      "step": 280210
    },
    {
      "epoch": 1.7972440376740115,
      "grad_norm": 0.053888414055109024,
      "learning_rate": 3.0980122616839357e-07,
      "loss": 0.0015,
      "step": 280220
    },
    {
      "epoch": 1.7973081745677975,
      "grad_norm": 0.14324425160884857,
      "learning_rate": 3.096073028544083e-07,
      "loss": 0.0012,
      "step": 280230
    },
    {
      "epoch": 1.7973723114615836,
      "grad_norm": 0.08394581079483032,
      "learning_rate": 3.0941343831439597e-07,
      "loss": 0.0009,
      "step": 280240
    },
    {
      "epoch": 1.7974364483553698,
      "grad_norm": 0.04315227270126343,
      "learning_rate": 3.0921963255078625e-07,
      "loss": 0.0009,
      "step": 280250
    },
    {
      "epoch": 1.7975005852491557,
      "grad_norm": 0.011538634076714516,
      "learning_rate": 3.090258855660061e-07,
      "loss": 0.0007,
      "step": 280260
    },
    {
      "epoch": 1.7975647221429418,
      "grad_norm": 0.06566084176301956,
      "learning_rate": 3.0883219736248417e-07,
      "loss": 0.0005,
      "step": 280270
    },
    {
      "epoch": 1.797628859036728,
      "grad_norm": 0.07556148618459702,
      "learning_rate": 3.086385679426479e-07,
      "loss": 0.0012,
      "step": 280280
    },
    {
      "epoch": 1.7976929959305141,
      "grad_norm": 0.11681356281042099,
      "learning_rate": 3.0844499730892376e-07,
      "loss": 0.0008,
      "step": 280290
    },
    {
      "epoch": 1.7977571328243003,
      "grad_norm": 0.051375046372413635,
      "learning_rate": 3.082514854637353e-07,
      "loss": 0.0009,
      "step": 280300
    },
    {
      "epoch": 1.7978212697180864,
      "grad_norm": 0.08355208486318588,
      "learning_rate": 3.0805803240951e-07,
      "loss": 0.0009,
      "step": 280310
    },
    {
      "epoch": 1.7978854066118723,
      "grad_norm": 0.16111016273498535,
      "learning_rate": 3.0786463814867106e-07,
      "loss": 0.0007,
      "step": 280320
    },
    {
      "epoch": 1.7979495435056585,
      "grad_norm": 0.13881529867649078,
      "learning_rate": 3.076713026836414e-07,
      "loss": 0.0012,
      "step": 280330
    },
    {
      "epoch": 1.7980136803994444,
      "grad_norm": 0.17221049964427948,
      "learning_rate": 3.074780260168431e-07,
      "loss": 0.001,
      "step": 280340
    },
    {
      "epoch": 1.7980778172932306,
      "grad_norm": 0.10530173778533936,
      "learning_rate": 3.072848081506996e-07,
      "loss": 0.001,
      "step": 280350
    },
    {
      "epoch": 1.7981419541870167,
      "grad_norm": 0.0914389118552208,
      "learning_rate": 3.0709164908763025e-07,
      "loss": 0.001,
      "step": 280360
    },
    {
      "epoch": 1.7982060910808029,
      "grad_norm": 0.10517113655805588,
      "learning_rate": 3.0689854883005746e-07,
      "loss": 0.0007,
      "step": 280370
    },
    {
      "epoch": 1.798270227974589,
      "grad_norm": 0.002558621345087886,
      "learning_rate": 3.067055073804004e-07,
      "loss": 0.0016,
      "step": 280380
    },
    {
      "epoch": 1.7983343648683752,
      "grad_norm": 0.11635466665029526,
      "learning_rate": 3.0651252474107607e-07,
      "loss": 0.0012,
      "step": 280390
    },
    {
      "epoch": 1.7983985017621613,
      "grad_norm": 0.09409769624471664,
      "learning_rate": 3.063196009145053e-07,
      "loss": 0.0007,
      "step": 280400
    },
    {
      "epoch": 1.7984626386559472,
      "grad_norm": 0.05665315315127373,
      "learning_rate": 3.061267359031045e-07,
      "loss": 0.0015,
      "step": 280410
    },
    {
      "epoch": 1.7985267755497334,
      "grad_norm": 0.025883320719003677,
      "learning_rate": 3.0593392970929004e-07,
      "loss": 0.0011,
      "step": 280420
    },
    {
      "epoch": 1.7985909124435193,
      "grad_norm": 0.1190020889043808,
      "learning_rate": 3.0574118233547834e-07,
      "loss": 0.0011,
      "step": 280430
    },
    {
      "epoch": 1.7986550493373055,
      "grad_norm": 0.06334960460662842,
      "learning_rate": 3.055484937840847e-07,
      "loss": 0.0007,
      "step": 280440
    },
    {
      "epoch": 1.7987191862310916,
      "grad_norm": 0.014685377478599548,
      "learning_rate": 3.053558640575238e-07,
      "loss": 0.0009,
      "step": 280450
    },
    {
      "epoch": 1.7987833231248778,
      "grad_norm": 0.04168510064482689,
      "learning_rate": 3.051632931582077e-07,
      "loss": 0.0007,
      "step": 280460
    },
    {
      "epoch": 1.798847460018664,
      "grad_norm": 0.0745929628610611,
      "learning_rate": 3.0497078108855214e-07,
      "loss": 0.001,
      "step": 280470
    },
    {
      "epoch": 1.79891159691245,
      "grad_norm": 0.06373525410890579,
      "learning_rate": 3.047783278509681e-07,
      "loss": 0.0013,
      "step": 280480
    },
    {
      "epoch": 1.798975733806236,
      "grad_norm": 0.09083700180053711,
      "learning_rate": 3.0458593344786737e-07,
      "loss": 0.001,
      "step": 280490
    },
    {
      "epoch": 1.7990398707000221,
      "grad_norm": 0.08681952208280563,
      "learning_rate": 3.0439359788165977e-07,
      "loss": 0.001,
      "step": 280500
    },
    {
      "epoch": 1.799104007593808,
      "grad_norm": 0.07414544373750687,
      "learning_rate": 3.042013211547573e-07,
      "loss": 0.0009,
      "step": 280510
    },
    {
      "epoch": 1.7991681444875942,
      "grad_norm": 0.005429443903267384,
      "learning_rate": 3.0400910326956803e-07,
      "loss": 0.0012,
      "step": 280520
    },
    {
      "epoch": 1.7992322813813804,
      "grad_norm": 0.00619896687567234,
      "learning_rate": 3.0381694422850107e-07,
      "loss": 0.0008,
      "step": 280530
    },
    {
      "epoch": 1.7992964182751665,
      "grad_norm": 0.04532570391893387,
      "learning_rate": 3.0362484403396286e-07,
      "loss": 0.0005,
      "step": 280540
    },
    {
      "epoch": 1.7993605551689527,
      "grad_norm": 0.2156553417444229,
      "learning_rate": 3.0343280268836315e-07,
      "loss": 0.0014,
      "step": 280550
    },
    {
      "epoch": 1.7994246920627388,
      "grad_norm": 0.02299816720187664,
      "learning_rate": 3.032408201941067e-07,
      "loss": 0.0007,
      "step": 280560
    },
    {
      "epoch": 1.799488828956525,
      "grad_norm": 0.22444169223308563,
      "learning_rate": 3.0304889655359884e-07,
      "loss": 0.0017,
      "step": 280570
    },
    {
      "epoch": 1.7995529658503109,
      "grad_norm": 0.18806947767734528,
      "learning_rate": 3.0285703176924587e-07,
      "loss": 0.0008,
      "step": 280580
    },
    {
      "epoch": 1.799617102744097,
      "grad_norm": 0.11151966452598572,
      "learning_rate": 3.0266522584345147e-07,
      "loss": 0.0012,
      "step": 280590
    },
    {
      "epoch": 1.799681239637883,
      "grad_norm": 0.0499407984316349,
      "learning_rate": 3.024734787786188e-07,
      "loss": 0.0017,
      "step": 280600
    },
    {
      "epoch": 1.799745376531669,
      "grad_norm": 0.019490646198391914,
      "learning_rate": 3.022817905771497e-07,
      "loss": 0.0006,
      "step": 280610
    },
    {
      "epoch": 1.7998095134254553,
      "grad_norm": 0.04655775800347328,
      "learning_rate": 3.020901612414484e-07,
      "loss": 0.0006,
      "step": 280620
    },
    {
      "epoch": 1.7998736503192414,
      "grad_norm": 0.06663881987333298,
      "learning_rate": 3.018985907739147e-07,
      "loss": 0.0004,
      "step": 280630
    },
    {
      "epoch": 1.7999377872130276,
      "grad_norm": 0.010833787731826305,
      "learning_rate": 3.0170707917694885e-07,
      "loss": 0.0011,
      "step": 280640
    },
    {
      "epoch": 1.8000019241068137,
      "grad_norm": 0.09659522026777267,
      "learning_rate": 3.0151562645295116e-07,
      "loss": 0.0011,
      "step": 280650
    },
    {
      "epoch": 1.8000660610005996,
      "grad_norm": 0.08453579992055893,
      "learning_rate": 3.0132423260432084e-07,
      "loss": 0.0007,
      "step": 280660
    },
    {
      "epoch": 1.8001301978943858,
      "grad_norm": 0.03279484063386917,
      "learning_rate": 3.011328976334565e-07,
      "loss": 0.002,
      "step": 280670
    },
    {
      "epoch": 1.800194334788172,
      "grad_norm": 0.30616578459739685,
      "learning_rate": 3.009416215427546e-07,
      "loss": 0.001,
      "step": 280680
    },
    {
      "epoch": 1.8002584716819579,
      "grad_norm": 0.046144042164087296,
      "learning_rate": 3.0075040433461213e-07,
      "loss": 0.0014,
      "step": 280690
    },
    {
      "epoch": 1.800322608575744,
      "grad_norm": 0.14769472181797028,
      "learning_rate": 3.005592460114265e-07,
      "loss": 0.0006,
      "step": 280700
    },
    {
      "epoch": 1.8003867454695301,
      "grad_norm": 0.09131012111902237,
      "learning_rate": 3.003681465755914e-07,
      "loss": 0.0013,
      "step": 280710
    },
    {
      "epoch": 1.8004508823633163,
      "grad_norm": 0.08563213795423508,
      "learning_rate": 3.001771060295022e-07,
      "loss": 0.0007,
      "step": 280720
    },
    {
      "epoch": 1.8005150192571024,
      "grad_norm": 0.0615537166595459,
      "learning_rate": 2.9998612437555365e-07,
      "loss": 0.0005,
      "step": 280730
    },
    {
      "epoch": 1.8005791561508886,
      "grad_norm": 0.008855455555021763,
      "learning_rate": 2.997952016161376e-07,
      "loss": 0.002,
      "step": 280740
    },
    {
      "epoch": 1.8006432930446745,
      "grad_norm": 0.07653220742940903,
      "learning_rate": 2.9960433775364674e-07,
      "loss": 0.0006,
      "step": 280750
    },
    {
      "epoch": 1.8007074299384607,
      "grad_norm": 0.0029082682449370623,
      "learning_rate": 2.994135327904724e-07,
      "loss": 0.0005,
      "step": 280760
    },
    {
      "epoch": 1.8007715668322466,
      "grad_norm": 0.10379134863615036,
      "learning_rate": 2.992227867290065e-07,
      "loss": 0.0008,
      "step": 280770
    },
    {
      "epoch": 1.8008357037260327,
      "grad_norm": 0.06596032530069351,
      "learning_rate": 2.990320995716389e-07,
      "loss": 0.002,
      "step": 280780
    },
    {
      "epoch": 1.800899840619819,
      "grad_norm": 0.08790586143732071,
      "learning_rate": 2.988414713207588e-07,
      "loss": 0.0007,
      "step": 280790
    },
    {
      "epoch": 1.800963977513605,
      "grad_norm": 0.04251585528254509,
      "learning_rate": 2.986509019787537e-07,
      "loss": 0.0008,
      "step": 280800
    },
    {
      "epoch": 1.8010281144073912,
      "grad_norm": 0.06360335648059845,
      "learning_rate": 2.984603915480139e-07,
      "loss": 0.0014,
      "step": 280810
    },
    {
      "epoch": 1.8010922513011773,
      "grad_norm": 0.03347192704677582,
      "learning_rate": 2.982699400309258e-07,
      "loss": 0.0007,
      "step": 280820
    },
    {
      "epoch": 1.8011563881949635,
      "grad_norm": 0.10828061401844025,
      "learning_rate": 2.9807954742987534e-07,
      "loss": 0.0016,
      "step": 280830
    },
    {
      "epoch": 1.8012205250887494,
      "grad_norm": 0.018905604258179665,
      "learning_rate": 2.978892137472478e-07,
      "loss": 0.0025,
      "step": 280840
    },
    {
      "epoch": 1.8012846619825356,
      "grad_norm": 0.06821412593126297,
      "learning_rate": 2.976989389854301e-07,
      "loss": 0.0028,
      "step": 280850
    },
    {
      "epoch": 1.8013487988763215,
      "grad_norm": 0.09935334324836731,
      "learning_rate": 2.975087231468049e-07,
      "loss": 0.001,
      "step": 280860
    },
    {
      "epoch": 1.8014129357701076,
      "grad_norm": 0.0035753112751990557,
      "learning_rate": 2.973185662337558e-07,
      "loss": 0.0013,
      "step": 280870
    },
    {
      "epoch": 1.8014770726638938,
      "grad_norm": 0.1754932999610901,
      "learning_rate": 2.971284682486664e-07,
      "loss": 0.0013,
      "step": 280880
    },
    {
      "epoch": 1.80154120955768,
      "grad_norm": 0.0921214148402214,
      "learning_rate": 2.969384291939187e-07,
      "loss": 0.0011,
      "step": 280890
    },
    {
      "epoch": 1.801605346451466,
      "grad_norm": 0.1432764232158661,
      "learning_rate": 2.967484490718936e-07,
      "loss": 0.0012,
      "step": 280900
    },
    {
      "epoch": 1.8016694833452522,
      "grad_norm": 0.140858456492424,
      "learning_rate": 2.9655852788497087e-07,
      "loss": 0.0026,
      "step": 280910
    },
    {
      "epoch": 1.8017336202390382,
      "grad_norm": 0.14326424896717072,
      "learning_rate": 2.963686656355319e-07,
      "loss": 0.0021,
      "step": 280920
    },
    {
      "epoch": 1.8017977571328243,
      "grad_norm": 0.12035234272480011,
      "learning_rate": 2.9617886232595596e-07,
      "loss": 0.0019,
      "step": 280930
    },
    {
      "epoch": 1.8018618940266102,
      "grad_norm": 0.05423832684755325,
      "learning_rate": 2.9598911795862e-07,
      "loss": 0.0006,
      "step": 280940
    },
    {
      "epoch": 1.8019260309203964,
      "grad_norm": 0.07330697029829025,
      "learning_rate": 2.957994325359015e-07,
      "loss": 0.0021,
      "step": 280950
    },
    {
      "epoch": 1.8019901678141825,
      "grad_norm": 0.07319719344377518,
      "learning_rate": 2.9560980606017917e-07,
      "loss": 0.0011,
      "step": 280960
    },
    {
      "epoch": 1.8020543047079687,
      "grad_norm": 0.013270118273794651,
      "learning_rate": 2.954202385338284e-07,
      "loss": 0.0007,
      "step": 280970
    },
    {
      "epoch": 1.8021184416017548,
      "grad_norm": 0.004556635860353708,
      "learning_rate": 2.9523072995922387e-07,
      "loss": 0.0007,
      "step": 280980
    },
    {
      "epoch": 1.802182578495541,
      "grad_norm": 0.11629913747310638,
      "learning_rate": 2.9504128033874035e-07,
      "loss": 0.001,
      "step": 280990
    },
    {
      "epoch": 1.8022467153893271,
      "grad_norm": 0.11676929891109467,
      "learning_rate": 2.9485188967475263e-07,
      "loss": 0.0014,
      "step": 281000
    },
    {
      "epoch": 1.802310852283113,
      "grad_norm": 0.019509943202137947,
      "learning_rate": 2.9466255796963327e-07,
      "loss": 0.0008,
      "step": 281010
    },
    {
      "epoch": 1.8023749891768992,
      "grad_norm": 0.21810665726661682,
      "learning_rate": 2.944732852257548e-07,
      "loss": 0.0008,
      "step": 281020
    },
    {
      "epoch": 1.8024391260706851,
      "grad_norm": 0.009947647340595722,
      "learning_rate": 2.942840714454892e-07,
      "loss": 0.0005,
      "step": 281030
    },
    {
      "epoch": 1.8025032629644713,
      "grad_norm": 0.17453940212726593,
      "learning_rate": 2.9409491663120736e-07,
      "loss": 0.0008,
      "step": 281040
    },
    {
      "epoch": 1.8025673998582574,
      "grad_norm": 0.06499435007572174,
      "learning_rate": 2.9390582078528016e-07,
      "loss": 0.001,
      "step": 281050
    },
    {
      "epoch": 1.8026315367520436,
      "grad_norm": 0.09117833524942398,
      "learning_rate": 2.9371678391007516e-07,
      "loss": 0.002,
      "step": 281060
    },
    {
      "epoch": 1.8026956736458297,
      "grad_norm": 0.03328854963183403,
      "learning_rate": 2.9352780600796274e-07,
      "loss": 0.0009,
      "step": 281070
    },
    {
      "epoch": 1.8027598105396159,
      "grad_norm": 0.18647193908691406,
      "learning_rate": 2.933388870813114e-07,
      "loss": 0.0006,
      "step": 281080
    },
    {
      "epoch": 1.802823947433402,
      "grad_norm": 0.12472377717494965,
      "learning_rate": 2.9315002713248717e-07,
      "loss": 0.0007,
      "step": 281090
    },
    {
      "epoch": 1.802888084327188,
      "grad_norm": 0.07018400728702545,
      "learning_rate": 2.9296122616385646e-07,
      "loss": 0.0011,
      "step": 281100
    },
    {
      "epoch": 1.802952221220974,
      "grad_norm": 0.0946158692240715,
      "learning_rate": 2.9277248417778624e-07,
      "loss": 0.0009,
      "step": 281110
    },
    {
      "epoch": 1.80301635811476,
      "grad_norm": 0.030763905495405197,
      "learning_rate": 2.925838011766402e-07,
      "loss": 0.0015,
      "step": 281120
    },
    {
      "epoch": 1.8030804950085462,
      "grad_norm": 0.10628003627061844,
      "learning_rate": 2.923951771627842e-07,
      "loss": 0.0006,
      "step": 281130
    },
    {
      "epoch": 1.8031446319023323,
      "grad_norm": 0.003375616390258074,
      "learning_rate": 2.9220661213858024e-07,
      "loss": 0.0019,
      "step": 281140
    },
    {
      "epoch": 1.8032087687961185,
      "grad_norm": 0.08748704940080643,
      "learning_rate": 2.9201810610639316e-07,
      "loss": 0.0024,
      "step": 281150
    },
    {
      "epoch": 1.8032729056899046,
      "grad_norm": 0.050831131637096405,
      "learning_rate": 2.9182965906858375e-07,
      "loss": 0.0008,
      "step": 281160
    },
    {
      "epoch": 1.8033370425836908,
      "grad_norm": 0.15017278492450714,
      "learning_rate": 2.9164127102751347e-07,
      "loss": 0.0006,
      "step": 281170
    },
    {
      "epoch": 1.8034011794774767,
      "grad_norm": 0.04059740900993347,
      "learning_rate": 2.9145294198554275e-07,
      "loss": 0.0005,
      "step": 281180
    },
    {
      "epoch": 1.8034653163712628,
      "grad_norm": 0.07971083372831345,
      "learning_rate": 2.9126467194503236e-07,
      "loss": 0.0016,
      "step": 281190
    },
    {
      "epoch": 1.8035294532650488,
      "grad_norm": 0.09287826716899872,
      "learning_rate": 2.910764609083405e-07,
      "loss": 0.0009,
      "step": 281200
    },
    {
      "epoch": 1.803593590158835,
      "grad_norm": 0.08002916723489761,
      "learning_rate": 2.908883088778264e-07,
      "loss": 0.0012,
      "step": 281210
    },
    {
      "epoch": 1.803657727052621,
      "grad_norm": 0.13122466206550598,
      "learning_rate": 2.90700215855847e-07,
      "loss": 0.0015,
      "step": 281220
    },
    {
      "epoch": 1.8037218639464072,
      "grad_norm": 0.001795127522200346,
      "learning_rate": 2.905121818447598e-07,
      "loss": 0.0004,
      "step": 281230
    },
    {
      "epoch": 1.8037860008401934,
      "grad_norm": 0.08344212174415588,
      "learning_rate": 2.9032420684692085e-07,
      "loss": 0.0007,
      "step": 281240
    },
    {
      "epoch": 1.8038501377339795,
      "grad_norm": 0.04882136732339859,
      "learning_rate": 2.901362908646854e-07,
      "loss": 0.0009,
      "step": 281250
    },
    {
      "epoch": 1.8039142746277657,
      "grad_norm": 0.047372933477163315,
      "learning_rate": 2.8994843390040773e-07,
      "loss": 0.0012,
      "step": 281260
    },
    {
      "epoch": 1.8039784115215516,
      "grad_norm": 0.16857458651065826,
      "learning_rate": 2.8976063595644313e-07,
      "loss": 0.0012,
      "step": 281270
    },
    {
      "epoch": 1.8040425484153377,
      "grad_norm": 0.057195521891117096,
      "learning_rate": 2.8957289703514413e-07,
      "loss": 0.0009,
      "step": 281280
    },
    {
      "epoch": 1.8041066853091237,
      "grad_norm": 0.07683075964450836,
      "learning_rate": 2.893852171388628e-07,
      "loss": 0.0009,
      "step": 281290
    },
    {
      "epoch": 1.8041708222029098,
      "grad_norm": 0.0845569521188736,
      "learning_rate": 2.891975962699517e-07,
      "loss": 0.0017,
      "step": 281300
    },
    {
      "epoch": 1.804234959096696,
      "grad_norm": 0.12468987703323364,
      "learning_rate": 2.8901003443076113e-07,
      "loss": 0.0013,
      "step": 281310
    },
    {
      "epoch": 1.804299095990482,
      "grad_norm": 0.0669076219201088,
      "learning_rate": 2.888225316236426e-07,
      "loss": 0.0004,
      "step": 281320
    },
    {
      "epoch": 1.8043632328842683,
      "grad_norm": 0.010885614901781082,
      "learning_rate": 2.8863508785094305e-07,
      "loss": 0.0018,
      "step": 281330
    },
    {
      "epoch": 1.8044273697780544,
      "grad_norm": 0.09682101011276245,
      "learning_rate": 2.884477031150146e-07,
      "loss": 0.0013,
      "step": 281340
    },
    {
      "epoch": 1.8044915066718403,
      "grad_norm": 0.016672270372509956,
      "learning_rate": 2.8826037741820357e-07,
      "loss": 0.0014,
      "step": 281350
    },
    {
      "epoch": 1.8045556435656265,
      "grad_norm": 0.08441387116909027,
      "learning_rate": 2.8807311076285705e-07,
      "loss": 0.0012,
      "step": 281360
    },
    {
      "epoch": 1.8046197804594124,
      "grad_norm": 0.002428964478895068,
      "learning_rate": 2.8788590315132146e-07,
      "loss": 0.0009,
      "step": 281370
    },
    {
      "epoch": 1.8046839173531986,
      "grad_norm": 0.021949807181954384,
      "learning_rate": 2.876987545859439e-07,
      "loss": 0.0013,
      "step": 281380
    },
    {
      "epoch": 1.8047480542469847,
      "grad_norm": 0.016486287117004395,
      "learning_rate": 2.8751166506906904e-07,
      "loss": 0.0008,
      "step": 281390
    },
    {
      "epoch": 1.8048121911407708,
      "grad_norm": 0.08122608065605164,
      "learning_rate": 2.873246346030406e-07,
      "loss": 0.0008,
      "step": 281400
    },
    {
      "epoch": 1.804876328034557,
      "grad_norm": 0.01998315192759037,
      "learning_rate": 2.871376631902023e-07,
      "loss": 0.0008,
      "step": 281410
    },
    {
      "epoch": 1.8049404649283431,
      "grad_norm": 0.05530071258544922,
      "learning_rate": 2.8695075083289727e-07,
      "loss": 0.0008,
      "step": 281420
    },
    {
      "epoch": 1.8050046018221293,
      "grad_norm": 0.00521240197122097,
      "learning_rate": 2.8676389753346856e-07,
      "loss": 0.0023,
      "step": 281430
    },
    {
      "epoch": 1.8050687387159152,
      "grad_norm": 0.004941079765558243,
      "learning_rate": 2.865771032942555e-07,
      "loss": 0.001,
      "step": 281440
    },
    {
      "epoch": 1.8051328756097014,
      "grad_norm": 0.02477288246154785,
      "learning_rate": 2.863903681176006e-07,
      "loss": 0.0011,
      "step": 281450
    },
    {
      "epoch": 1.8051970125034873,
      "grad_norm": 0.454287052154541,
      "learning_rate": 2.8620369200584363e-07,
      "loss": 0.0029,
      "step": 281460
    },
    {
      "epoch": 1.8052611493972734,
      "grad_norm": 0.19727076590061188,
      "learning_rate": 2.8601707496132283e-07,
      "loss": 0.0009,
      "step": 281470
    },
    {
      "epoch": 1.8053252862910596,
      "grad_norm": 0.03952730819582939,
      "learning_rate": 2.858305169863768e-07,
      "loss": 0.0009,
      "step": 281480
    },
    {
      "epoch": 1.8053894231848457,
      "grad_norm": 0.09670371562242508,
      "learning_rate": 2.856440180833442e-07,
      "loss": 0.0009,
      "step": 281490
    },
    {
      "epoch": 1.805453560078632,
      "grad_norm": 0.045216407626867294,
      "learning_rate": 2.85457578254561e-07,
      "loss": 0.0022,
      "step": 281500
    },
    {
      "epoch": 1.805517696972418,
      "grad_norm": 0.1280045360326767,
      "learning_rate": 2.852711975023642e-07,
      "loss": 0.0012,
      "step": 281510
    },
    {
      "epoch": 1.8055818338662042,
      "grad_norm": 0.07866466045379639,
      "learning_rate": 2.8508487582908804e-07,
      "loss": 0.0006,
      "step": 281520
    },
    {
      "epoch": 1.8056459707599901,
      "grad_norm": 0.25505441427230835,
      "learning_rate": 2.8489861323706844e-07,
      "loss": 0.0013,
      "step": 281530
    },
    {
      "epoch": 1.8057101076537763,
      "grad_norm": 0.057884521782398224,
      "learning_rate": 2.847124097286397e-07,
      "loss": 0.0015,
      "step": 281540
    },
    {
      "epoch": 1.8057742445475622,
      "grad_norm": 0.12495142966508865,
      "learning_rate": 2.8452626530613314e-07,
      "loss": 0.0008,
      "step": 281550
    },
    {
      "epoch": 1.8058383814413483,
      "grad_norm": 0.022214218974113464,
      "learning_rate": 2.843401799718837e-07,
      "loss": 0.0006,
      "step": 281560
    },
    {
      "epoch": 1.8059025183351345,
      "grad_norm": 0.028516335412859917,
      "learning_rate": 2.8415415372822164e-07,
      "loss": 0.0013,
      "step": 281570
    },
    {
      "epoch": 1.8059666552289206,
      "grad_norm": 0.08100759983062744,
      "learning_rate": 2.839681865774785e-07,
      "loss": 0.0035,
      "step": 281580
    },
    {
      "epoch": 1.8060307921227068,
      "grad_norm": 0.20059432089328766,
      "learning_rate": 2.83782278521984e-07,
      "loss": 0.0027,
      "step": 281590
    },
    {
      "epoch": 1.806094929016493,
      "grad_norm": 0.020009048283100128,
      "learning_rate": 2.835964295640686e-07,
      "loss": 0.0013,
      "step": 281600
    },
    {
      "epoch": 1.8061590659102789,
      "grad_norm": 0.06283935159444809,
      "learning_rate": 2.83410639706061e-07,
      "loss": 0.0008,
      "step": 281610
    },
    {
      "epoch": 1.806223202804065,
      "grad_norm": 0.041120126843452454,
      "learning_rate": 2.832249089502886e-07,
      "loss": 0.0017,
      "step": 281620
    },
    {
      "epoch": 1.806287339697851,
      "grad_norm": 0.10960491001605988,
      "learning_rate": 2.8303923729907866e-07,
      "loss": 0.0007,
      "step": 281630
    },
    {
      "epoch": 1.806351476591637,
      "grad_norm": 0.15289542078971863,
      "learning_rate": 2.828536247547592e-07,
      "loss": 0.001,
      "step": 281640
    },
    {
      "epoch": 1.8064156134854232,
      "grad_norm": 0.013551203534007072,
      "learning_rate": 2.826680713196545e-07,
      "loss": 0.001,
      "step": 281650
    },
    {
      "epoch": 1.8064797503792094,
      "grad_norm": 0.09905155748128891,
      "learning_rate": 2.8248257699609095e-07,
      "loss": 0.0007,
      "step": 281660
    },
    {
      "epoch": 1.8065438872729955,
      "grad_norm": 0.038905225694179535,
      "learning_rate": 2.8229714178639125e-07,
      "loss": 0.0004,
      "step": 281670
    },
    {
      "epoch": 1.8066080241667817,
      "grad_norm": 0.10164858400821686,
      "learning_rate": 2.8211176569288065e-07,
      "loss": 0.0017,
      "step": 281680
    },
    {
      "epoch": 1.8066721610605678,
      "grad_norm": 0.00880381092429161,
      "learning_rate": 2.8192644871788135e-07,
      "loss": 0.0011,
      "step": 281690
    },
    {
      "epoch": 1.8067362979543538,
      "grad_norm": 0.06276466697454453,
      "learning_rate": 2.817411908637152e-07,
      "loss": 0.0018,
      "step": 281700
    },
    {
      "epoch": 1.80680043484814,
      "grad_norm": 0.089014932513237,
      "learning_rate": 2.815559921327049e-07,
      "loss": 0.0013,
      "step": 281710
    },
    {
      "epoch": 1.8068645717419258,
      "grad_norm": 0.7244876623153687,
      "learning_rate": 2.8137085252716965e-07,
      "loss": 0.001,
      "step": 281720
    },
    {
      "epoch": 1.806928708635712,
      "grad_norm": 0.08272527158260345,
      "learning_rate": 2.811857720494304e-07,
      "loss": 0.0021,
      "step": 281730
    },
    {
      "epoch": 1.8069928455294981,
      "grad_norm": 0.09128829091787338,
      "learning_rate": 2.8100075070180535e-07,
      "loss": 0.0011,
      "step": 281740
    },
    {
      "epoch": 1.8070569824232843,
      "grad_norm": 0.1517772674560547,
      "learning_rate": 2.808157884866136e-07,
      "loss": 0.0012,
      "step": 281750
    },
    {
      "epoch": 1.8071211193170704,
      "grad_norm": 0.14261551201343536,
      "learning_rate": 2.8063088540617345e-07,
      "loss": 0.0011,
      "step": 281760
    },
    {
      "epoch": 1.8071852562108566,
      "grad_norm": 0.05205344036221504,
      "learning_rate": 2.8044604146280074e-07,
      "loss": 0.0005,
      "step": 281770
    },
    {
      "epoch": 1.8072493931046425,
      "grad_norm": 0.058461472392082214,
      "learning_rate": 2.802612566588114e-07,
      "loss": 0.0017,
      "step": 281780
    },
    {
      "epoch": 1.8073135299984286,
      "grad_norm": 0.05705380439758301,
      "learning_rate": 2.800765309965225e-07,
      "loss": 0.0011,
      "step": 281790
    },
    {
      "epoch": 1.8073776668922148,
      "grad_norm": 0.0523599311709404,
      "learning_rate": 2.7989186447824834e-07,
      "loss": 0.001,
      "step": 281800
    },
    {
      "epoch": 1.8074418037860007,
      "grad_norm": 0.07191793620586395,
      "learning_rate": 2.7970725710630196e-07,
      "loss": 0.0008,
      "step": 281810
    },
    {
      "epoch": 1.8075059406797869,
      "grad_norm": 0.08435988426208496,
      "learning_rate": 2.7952270888299657e-07,
      "loss": 0.0016,
      "step": 281820
    },
    {
      "epoch": 1.807570077573573,
      "grad_norm": 0.006404428742825985,
      "learning_rate": 2.793382198106465e-07,
      "loss": 0.0007,
      "step": 281830
    },
    {
      "epoch": 1.8076342144673592,
      "grad_norm": 0.0974842756986618,
      "learning_rate": 2.7915378989156193e-07,
      "loss": 0.0008,
      "step": 281840
    },
    {
      "epoch": 1.8076983513611453,
      "grad_norm": 0.05271318182349205,
      "learning_rate": 2.789694191280534e-07,
      "loss": 0.0012,
      "step": 281850
    },
    {
      "epoch": 1.8077624882549315,
      "grad_norm": 0.04372788220643997,
      "learning_rate": 2.7878510752243294e-07,
      "loss": 0.0006,
      "step": 281860
    },
    {
      "epoch": 1.8078266251487174,
      "grad_norm": 0.06479683518409729,
      "learning_rate": 2.7860085507700975e-07,
      "loss": 0.0011,
      "step": 281870
    },
    {
      "epoch": 1.8078907620425035,
      "grad_norm": 0.10288426280021667,
      "learning_rate": 2.784166617940914e-07,
      "loss": 0.0019,
      "step": 281880
    },
    {
      "epoch": 1.8079548989362895,
      "grad_norm": 0.09815745800733566,
      "learning_rate": 2.7823252767598675e-07,
      "loss": 0.0008,
      "step": 281890
    },
    {
      "epoch": 1.8080190358300756,
      "grad_norm": 0.02030204050242901,
      "learning_rate": 2.780484527250027e-07,
      "loss": 0.0006,
      "step": 281900
    },
    {
      "epoch": 1.8080831727238618,
      "grad_norm": 0.018891580402851105,
      "learning_rate": 2.7786443694344745e-07,
      "loss": 0.0023,
      "step": 281910
    },
    {
      "epoch": 1.808147309617648,
      "grad_norm": 0.005142646841704845,
      "learning_rate": 2.776804803336253e-07,
      "loss": 0.0007,
      "step": 281920
    },
    {
      "epoch": 1.808211446511434,
      "grad_norm": 0.10481272637844086,
      "learning_rate": 2.7749658289784156e-07,
      "loss": 0.0011,
      "step": 281930
    },
    {
      "epoch": 1.8082755834052202,
      "grad_norm": 0.011906717903912067,
      "learning_rate": 2.773127446384005e-07,
      "loss": 0.0014,
      "step": 281940
    },
    {
      "epoch": 1.8083397202990064,
      "grad_norm": 0.11263014376163483,
      "learning_rate": 2.771289655576065e-07,
      "loss": 0.0009,
      "step": 281950
    },
    {
      "epoch": 1.8084038571927923,
      "grad_norm": 0.03821062296628952,
      "learning_rate": 2.76945245657762e-07,
      "loss": 0.0007,
      "step": 281960
    },
    {
      "epoch": 1.8084679940865784,
      "grad_norm": 0.008966549299657345,
      "learning_rate": 2.7676158494116814e-07,
      "loss": 0.0023,
      "step": 281970
    },
    {
      "epoch": 1.8085321309803644,
      "grad_norm": 0.00295799458399415,
      "learning_rate": 2.765779834101279e-07,
      "loss": 0.0011,
      "step": 281980
    },
    {
      "epoch": 1.8085962678741505,
      "grad_norm": 0.06502897292375565,
      "learning_rate": 2.7639444106694167e-07,
      "loss": 0.0012,
      "step": 281990
    },
    {
      "epoch": 1.8086604047679367,
      "grad_norm": 0.01179441250860691,
      "learning_rate": 2.7621095791390883e-07,
      "loss": 0.0007,
      "step": 282000
    },
    {
      "epoch": 1.8087245416617228,
      "grad_norm": 0.07027900218963623,
      "learning_rate": 2.7602753395332803e-07,
      "loss": 0.0008,
      "step": 282010
    },
    {
      "epoch": 1.808788678555509,
      "grad_norm": 0.10831480473279953,
      "learning_rate": 2.758441691874991e-07,
      "loss": 0.0012,
      "step": 282020
    },
    {
      "epoch": 1.808852815449295,
      "grad_norm": 0.07933515310287476,
      "learning_rate": 2.756608636187191e-07,
      "loss": 0.0011,
      "step": 282030
    },
    {
      "epoch": 1.808916952343081,
      "grad_norm": 0.05271373316645622,
      "learning_rate": 2.7547761724928515e-07,
      "loss": 0.0005,
      "step": 282040
    },
    {
      "epoch": 1.8089810892368672,
      "grad_norm": 0.06447888910770416,
      "learning_rate": 2.7529443008149193e-07,
      "loss": 0.0009,
      "step": 282050
    },
    {
      "epoch": 1.809045226130653,
      "grad_norm": 0.04930812120437622,
      "learning_rate": 2.7511130211763773e-07,
      "loss": 0.0008,
      "step": 282060
    },
    {
      "epoch": 1.8091093630244393,
      "grad_norm": 0.05755256116390228,
      "learning_rate": 2.749282333600151e-07,
      "loss": 0.0015,
      "step": 282070
    },
    {
      "epoch": 1.8091734999182254,
      "grad_norm": 0.11437110602855682,
      "learning_rate": 2.7474522381091883e-07,
      "loss": 0.0014,
      "step": 282080
    },
    {
      "epoch": 1.8092376368120116,
      "grad_norm": 0.029845241457223892,
      "learning_rate": 2.745622734726411e-07,
      "loss": 0.0026,
      "step": 282090
    },
    {
      "epoch": 1.8093017737057977,
      "grad_norm": 0.14690016210079193,
      "learning_rate": 2.743793823474761e-07,
      "loss": 0.0005,
      "step": 282100
    },
    {
      "epoch": 1.8093659105995838,
      "grad_norm": 0.06361275166273117,
      "learning_rate": 2.7419655043771474e-07,
      "loss": 0.0018,
      "step": 282110
    },
    {
      "epoch": 1.80943004749337,
      "grad_norm": 0.04907489940524101,
      "learning_rate": 2.7401377774564754e-07,
      "loss": 0.0009,
      "step": 282120
    },
    {
      "epoch": 1.809494184387156,
      "grad_norm": 0.08087644726037979,
      "learning_rate": 2.7383106427356645e-07,
      "loss": 0.0007,
      "step": 282130
    },
    {
      "epoch": 1.809558321280942,
      "grad_norm": 0.07917283475399017,
      "learning_rate": 2.736484100237591e-07,
      "loss": 0.0011,
      "step": 282140
    },
    {
      "epoch": 1.809622458174728,
      "grad_norm": 0.05790490657091141,
      "learning_rate": 2.734658149985153e-07,
      "loss": 0.001,
      "step": 282150
    },
    {
      "epoch": 1.8096865950685141,
      "grad_norm": 0.22517386078834534,
      "learning_rate": 2.7328327920012223e-07,
      "loss": 0.0011,
      "step": 282160
    },
    {
      "epoch": 1.8097507319623003,
      "grad_norm": 0.06193617731332779,
      "learning_rate": 2.7310080263086847e-07,
      "loss": 0.0013,
      "step": 282170
    },
    {
      "epoch": 1.8098148688560864,
      "grad_norm": 0.02145802229642868,
      "learning_rate": 2.7291838529304005e-07,
      "loss": 0.0012,
      "step": 282180
    },
    {
      "epoch": 1.8098790057498726,
      "grad_norm": 0.057392776012420654,
      "learning_rate": 2.727360271889223e-07,
      "loss": 0.001,
      "step": 282190
    },
    {
      "epoch": 1.8099431426436587,
      "grad_norm": 0.0896657407283783,
      "learning_rate": 2.725537283208002e-07,
      "loss": 0.0012,
      "step": 282200
    },
    {
      "epoch": 1.8100072795374447,
      "grad_norm": 0.0054199169389903545,
      "learning_rate": 2.7237148869095955e-07,
      "loss": 0.0006,
      "step": 282210
    },
    {
      "epoch": 1.8100714164312308,
      "grad_norm": 0.10844659805297852,
      "learning_rate": 2.7218930830168247e-07,
      "loss": 0.0005,
      "step": 282220
    },
    {
      "epoch": 1.810135553325017,
      "grad_norm": 0.10597152262926102,
      "learning_rate": 2.720071871552521e-07,
      "loss": 0.001,
      "step": 282230
    },
    {
      "epoch": 1.810199690218803,
      "grad_norm": 0.04104155674576759,
      "learning_rate": 2.7182512525395055e-07,
      "loss": 0.0006,
      "step": 282240
    },
    {
      "epoch": 1.810263827112589,
      "grad_norm": 0.07009421288967133,
      "learning_rate": 2.7164312260005986e-07,
      "loss": 0.0011,
      "step": 282250
    },
    {
      "epoch": 1.8103279640063752,
      "grad_norm": 0.1096557155251503,
      "learning_rate": 2.714611791958599e-07,
      "loss": 0.001,
      "step": 282260
    },
    {
      "epoch": 1.8103921009001613,
      "grad_norm": 0.19809821248054504,
      "learning_rate": 2.7127929504363047e-07,
      "loss": 0.001,
      "step": 282270
    },
    {
      "epoch": 1.8104562377939475,
      "grad_norm": 0.052505962550640106,
      "learning_rate": 2.7109747014565146e-07,
      "loss": 0.0012,
      "step": 282280
    },
    {
      "epoch": 1.8105203746877336,
      "grad_norm": 0.029269928112626076,
      "learning_rate": 2.70915704504201e-07,
      "loss": 0.0009,
      "step": 282290
    },
    {
      "epoch": 1.8105845115815196,
      "grad_norm": 0.12355450540781021,
      "learning_rate": 2.707339981215568e-07,
      "loss": 0.0037,
      "step": 282300
    },
    {
      "epoch": 1.8106486484753057,
      "grad_norm": 0.06523124873638153,
      "learning_rate": 2.705523509999941e-07,
      "loss": 0.0007,
      "step": 282310
    },
    {
      "epoch": 1.8107127853690916,
      "grad_norm": 0.25369080901145935,
      "learning_rate": 2.703707631417918e-07,
      "loss": 0.0019,
      "step": 282320
    },
    {
      "epoch": 1.8107769222628778,
      "grad_norm": 0.02460874617099762,
      "learning_rate": 2.701892345492241e-07,
      "loss": 0.0007,
      "step": 282330
    },
    {
      "epoch": 1.810841059156664,
      "grad_norm": 0.06876687705516815,
      "learning_rate": 2.7000776522456586e-07,
      "loss": 0.0012,
      "step": 282340
    },
    {
      "epoch": 1.81090519605045,
      "grad_norm": 0.047911740839481354,
      "learning_rate": 2.698263551700897e-07,
      "loss": 0.0008,
      "step": 282350
    },
    {
      "epoch": 1.8109693329442362,
      "grad_norm": 0.05371854081749916,
      "learning_rate": 2.696450043880705e-07,
      "loss": 0.0008,
      "step": 282360
    },
    {
      "epoch": 1.8110334698380224,
      "grad_norm": 0.010016710497438908,
      "learning_rate": 2.694637128807803e-07,
      "loss": 0.001,
      "step": 282370
    },
    {
      "epoch": 1.8110976067318085,
      "grad_norm": 0.0178392231464386,
      "learning_rate": 2.692824806504907e-07,
      "loss": 0.0009,
      "step": 282380
    },
    {
      "epoch": 1.8111617436255945,
      "grad_norm": 0.044535715132951736,
      "learning_rate": 2.6910130769947194e-07,
      "loss": 0.0005,
      "step": 282390
    },
    {
      "epoch": 1.8112258805193806,
      "grad_norm": 0.12505608797073364,
      "learning_rate": 2.6892019402999515e-07,
      "loss": 0.001,
      "step": 282400
    },
    {
      "epoch": 1.8112900174131665,
      "grad_norm": 0.011626729741692543,
      "learning_rate": 2.687391396443301e-07,
      "loss": 0.0011,
      "step": 282410
    },
    {
      "epoch": 1.8113541543069527,
      "grad_norm": 0.07806049287319183,
      "learning_rate": 2.6855814454474437e-07,
      "loss": 0.0012,
      "step": 282420
    },
    {
      "epoch": 1.8114182912007388,
      "grad_norm": 0.16373804211616516,
      "learning_rate": 2.683772087335068e-07,
      "loss": 0.001,
      "step": 282430
    },
    {
      "epoch": 1.811482428094525,
      "grad_norm": 0.11224564164876938,
      "learning_rate": 2.68196332212885e-07,
      "loss": 0.0011,
      "step": 282440
    },
    {
      "epoch": 1.8115465649883111,
      "grad_norm": 0.017139988020062447,
      "learning_rate": 2.6801551498514434e-07,
      "loss": 0.0012,
      "step": 282450
    },
    {
      "epoch": 1.8116107018820973,
      "grad_norm": 0.05777726694941521,
      "learning_rate": 2.678347570525508e-07,
      "loss": 0.0011,
      "step": 282460
    },
    {
      "epoch": 1.8116748387758832,
      "grad_norm": 0.07671855390071869,
      "learning_rate": 2.676540584173709e-07,
      "loss": 0.001,
      "step": 282470
    },
    {
      "epoch": 1.8117389756696693,
      "grad_norm": 0.0009651641012169421,
      "learning_rate": 2.674734190818673e-07,
      "loss": 0.0007,
      "step": 282480
    },
    {
      "epoch": 1.8118031125634553,
      "grad_norm": 0.13907742500305176,
      "learning_rate": 2.6729283904830426e-07,
      "loss": 0.0007,
      "step": 282490
    },
    {
      "epoch": 1.8118672494572414,
      "grad_norm": 0.07996530830860138,
      "learning_rate": 2.671123183189434e-07,
      "loss": 0.001,
      "step": 282500
    },
    {
      "epoch": 1.8119313863510276,
      "grad_norm": 0.062143053859472275,
      "learning_rate": 2.669318568960483e-07,
      "loss": 0.0006,
      "step": 282510
    },
    {
      "epoch": 1.8119955232448137,
      "grad_norm": 0.23566403985023499,
      "learning_rate": 2.6675145478188003e-07,
      "loss": 0.0014,
      "step": 282520
    },
    {
      "epoch": 1.8120596601385999,
      "grad_norm": 0.04087146371603012,
      "learning_rate": 2.66571111978699e-07,
      "loss": 0.0008,
      "step": 282530
    },
    {
      "epoch": 1.812123797032386,
      "grad_norm": 0.0075711836107075214,
      "learning_rate": 2.66390828488764e-07,
      "loss": 0.0014,
      "step": 282540
    },
    {
      "epoch": 1.8121879339261722,
      "grad_norm": 0.03502900153398514,
      "learning_rate": 2.662106043143353e-07,
      "loss": 0.0018,
      "step": 282550
    },
    {
      "epoch": 1.812252070819958,
      "grad_norm": 0.035383906215429306,
      "learning_rate": 2.660304394576713e-07,
      "loss": 0.001,
      "step": 282560
    },
    {
      "epoch": 1.8123162077137442,
      "grad_norm": 0.42719870805740356,
      "learning_rate": 2.658503339210289e-07,
      "loss": 0.0017,
      "step": 282570
    },
    {
      "epoch": 1.8123803446075302,
      "grad_norm": 0.13851973414421082,
      "learning_rate": 2.6567028770666525e-07,
      "loss": 0.0012,
      "step": 282580
    },
    {
      "epoch": 1.8124444815013163,
      "grad_norm": 0.02888554334640503,
      "learning_rate": 2.654903008168369e-07,
      "loss": 0.0004,
      "step": 282590
    },
    {
      "epoch": 1.8125086183951025,
      "grad_norm": 0.05228433385491371,
      "learning_rate": 2.653103732537987e-07,
      "loss": 0.0012,
      "step": 282600
    },
    {
      "epoch": 1.8125727552888886,
      "grad_norm": 0.06343115121126175,
      "learning_rate": 2.651305050198044e-07,
      "loss": 0.0011,
      "step": 282610
    },
    {
      "epoch": 1.8126368921826748,
      "grad_norm": 0.016875579953193665,
      "learning_rate": 2.6495069611710944e-07,
      "loss": 0.0016,
      "step": 282620
    },
    {
      "epoch": 1.812701029076461,
      "grad_norm": 0.07538674771785736,
      "learning_rate": 2.64770946547967e-07,
      "loss": 0.0013,
      "step": 282630
    },
    {
      "epoch": 1.812765165970247,
      "grad_norm": 0.06355781108140945,
      "learning_rate": 2.64591256314628e-07,
      "loss": 0.0006,
      "step": 282640
    },
    {
      "epoch": 1.812829302864033,
      "grad_norm": 0.06597629189491272,
      "learning_rate": 2.644116254193446e-07,
      "loss": 0.0009,
      "step": 282650
    },
    {
      "epoch": 1.8128934397578191,
      "grad_norm": 0.09554356336593628,
      "learning_rate": 2.6423205386436834e-07,
      "loss": 0.0012,
      "step": 282660
    },
    {
      "epoch": 1.812957576651605,
      "grad_norm": 0.20009282231330872,
      "learning_rate": 2.640525416519496e-07,
      "loss": 0.0015,
      "step": 282670
    },
    {
      "epoch": 1.8130217135453912,
      "grad_norm": 0.06588278710842133,
      "learning_rate": 2.6387308878433713e-07,
      "loss": 0.0009,
      "step": 282680
    },
    {
      "epoch": 1.8130858504391774,
      "grad_norm": 0.06026269868016243,
      "learning_rate": 2.636936952637786e-07,
      "loss": 0.0012,
      "step": 282690
    },
    {
      "epoch": 1.8131499873329635,
      "grad_norm": 0.24051862955093384,
      "learning_rate": 2.635143610925245e-07,
      "loss": 0.0009,
      "step": 282700
    },
    {
      "epoch": 1.8132141242267497,
      "grad_norm": 0.1734119951725006,
      "learning_rate": 2.6333508627282014e-07,
      "loss": 0.002,
      "step": 282710
    },
    {
      "epoch": 1.8132782611205358,
      "grad_norm": 0.023682259023189545,
      "learning_rate": 2.6315587080691205e-07,
      "loss": 0.0009,
      "step": 282720
    },
    {
      "epoch": 1.8133423980143217,
      "grad_norm": 0.18128125369548798,
      "learning_rate": 2.629767146970463e-07,
      "loss": 0.0019,
      "step": 282730
    },
    {
      "epoch": 1.8134065349081079,
      "grad_norm": 0.04932695999741554,
      "learning_rate": 2.627976179454683e-07,
      "loss": 0.0017,
      "step": 282740
    },
    {
      "epoch": 1.8134706718018938,
      "grad_norm": 0.030380118638277054,
      "learning_rate": 2.626185805544218e-07,
      "loss": 0.0007,
      "step": 282750
    },
    {
      "epoch": 1.81353480869568,
      "grad_norm": 0.13956759870052338,
      "learning_rate": 2.6243960252614997e-07,
      "loss": 0.0006,
      "step": 282760
    },
    {
      "epoch": 1.813598945589466,
      "grad_norm": 0.13544772565364838,
      "learning_rate": 2.622606838628949e-07,
      "loss": 0.0012,
      "step": 282770
    },
    {
      "epoch": 1.8136630824832523,
      "grad_norm": 0.06107968091964722,
      "learning_rate": 2.6208182456690044e-07,
      "loss": 0.0005,
      "step": 282780
    },
    {
      "epoch": 1.8137272193770384,
      "grad_norm": 0.13277402520179749,
      "learning_rate": 2.6190302464040693e-07,
      "loss": 0.0015,
      "step": 282790
    },
    {
      "epoch": 1.8137913562708246,
      "grad_norm": 0.024062488228082657,
      "learning_rate": 2.6172428408565485e-07,
      "loss": 0.001,
      "step": 282800
    },
    {
      "epoch": 1.8138554931646107,
      "grad_norm": 0.06757381558418274,
      "learning_rate": 2.615456029048835e-07,
      "loss": 0.0009,
      "step": 282810
    },
    {
      "epoch": 1.8139196300583966,
      "grad_norm": 0.1526866853237152,
      "learning_rate": 2.613669811003322e-07,
      "loss": 0.0009,
      "step": 282820
    },
    {
      "epoch": 1.8139837669521828,
      "grad_norm": 0.06795057654380798,
      "learning_rate": 2.611884186742397e-07,
      "loss": 0.0013,
      "step": 282830
    },
    {
      "epoch": 1.8140479038459687,
      "grad_norm": 0.0031683510169386864,
      "learning_rate": 2.610099156288426e-07,
      "loss": 0.0013,
      "step": 282840
    },
    {
      "epoch": 1.8141120407397548,
      "grad_norm": 0.04895066097378731,
      "learning_rate": 2.608314719663785e-07,
      "loss": 0.0008,
      "step": 282850
    },
    {
      "epoch": 1.814176177633541,
      "grad_norm": 0.006946314591914415,
      "learning_rate": 2.6065308768908335e-07,
      "loss": 0.0014,
      "step": 282860
    },
    {
      "epoch": 1.8142403145273271,
      "grad_norm": 0.08963599801063538,
      "learning_rate": 2.604747627991922e-07,
      "loss": 0.0008,
      "step": 282870
    },
    {
      "epoch": 1.8143044514211133,
      "grad_norm": 0.15497802197933197,
      "learning_rate": 2.602964972989386e-07,
      "loss": 0.0021,
      "step": 282880
    },
    {
      "epoch": 1.8143685883148994,
      "grad_norm": 0.0698586031794548,
      "learning_rate": 2.601182911905581e-07,
      "loss": 0.0007,
      "step": 282890
    },
    {
      "epoch": 1.8144327252086854,
      "grad_norm": 0.047951746731996536,
      "learning_rate": 2.599401444762828e-07,
      "loss": 0.0009,
      "step": 282900
    },
    {
      "epoch": 1.8144968621024715,
      "grad_norm": 0.02806960605084896,
      "learning_rate": 2.5976205715834533e-07,
      "loss": 0.0006,
      "step": 282910
    },
    {
      "epoch": 1.8145609989962574,
      "grad_norm": 0.06804698705673218,
      "learning_rate": 2.5958402923897616e-07,
      "loss": 0.001,
      "step": 282920
    },
    {
      "epoch": 1.8146251358900436,
      "grad_norm": 0.14505638182163239,
      "learning_rate": 2.5940606072040795e-07,
      "loss": 0.0014,
      "step": 282930
    },
    {
      "epoch": 1.8146892727838297,
      "grad_norm": 0.031782738864421844,
      "learning_rate": 2.5922815160487003e-07,
      "loss": 0.0006,
      "step": 282940
    },
    {
      "epoch": 1.814753409677616,
      "grad_norm": 0.11933920532464981,
      "learning_rate": 2.590503018945911e-07,
      "loss": 0.0019,
      "step": 282950
    },
    {
      "epoch": 1.814817546571402,
      "grad_norm": 0.05999316647648811,
      "learning_rate": 2.5887251159180005e-07,
      "loss": 0.0006,
      "step": 282960
    },
    {
      "epoch": 1.8148816834651882,
      "grad_norm": 0.06679647415876389,
      "learning_rate": 2.58694780698725e-07,
      "loss": 0.0007,
      "step": 282970
    },
    {
      "epoch": 1.8149458203589743,
      "grad_norm": 0.10151786357164383,
      "learning_rate": 2.585171092175931e-07,
      "loss": 0.0017,
      "step": 282980
    },
    {
      "epoch": 1.8150099572527603,
      "grad_norm": 0.04049347713589668,
      "learning_rate": 2.5833949715063035e-07,
      "loss": 0.0011,
      "step": 282990
    },
    {
      "epoch": 1.8150740941465464,
      "grad_norm": 0.06263840943574905,
      "learning_rate": 2.581619445000627e-07,
      "loss": 0.0004,
      "step": 283000
    },
    {
      "epoch": 1.8151382310403323,
      "grad_norm": 0.04469954967498779,
      "learning_rate": 2.5798445126811455e-07,
      "loss": 0.0003,
      "step": 283010
    },
    {
      "epoch": 1.8152023679341185,
      "grad_norm": 0.11815565824508667,
      "learning_rate": 2.5780701745701073e-07,
      "loss": 0.0014,
      "step": 283020
    },
    {
      "epoch": 1.8152665048279046,
      "grad_norm": 0.030553000047802925,
      "learning_rate": 2.576296430689734e-07,
      "loss": 0.0013,
      "step": 283030
    },
    {
      "epoch": 1.8153306417216908,
      "grad_norm": 0.04269164428114891,
      "learning_rate": 2.5745232810622634e-07,
      "loss": 0.0015,
      "step": 283040
    },
    {
      "epoch": 1.815394778615477,
      "grad_norm": 0.04332788288593292,
      "learning_rate": 2.572750725709916e-07,
      "loss": 0.0009,
      "step": 283050
    },
    {
      "epoch": 1.815458915509263,
      "grad_norm": 0.4782222807407379,
      "learning_rate": 2.5709787646548967e-07,
      "loss": 0.0019,
      "step": 283060
    },
    {
      "epoch": 1.8155230524030492,
      "grad_norm": 0.09612571448087692,
      "learning_rate": 2.569207397919399e-07,
      "loss": 0.0007,
      "step": 283070
    },
    {
      "epoch": 1.8155871892968352,
      "grad_norm": 0.11113991588354111,
      "learning_rate": 2.5674366255256433e-07,
      "loss": 0.0012,
      "step": 283080
    },
    {
      "epoch": 1.8156513261906213,
      "grad_norm": 0.10252439230680466,
      "learning_rate": 2.5656664474958014e-07,
      "loss": 0.001,
      "step": 283090
    },
    {
      "epoch": 1.8157154630844072,
      "grad_norm": 0.041576530784368515,
      "learning_rate": 2.5638968638520613e-07,
      "loss": 0.001,
      "step": 283100
    },
    {
      "epoch": 1.8157795999781934,
      "grad_norm": 0.05730690062046051,
      "learning_rate": 2.562127874616588e-07,
      "loss": 0.0033,
      "step": 283110
    },
    {
      "epoch": 1.8158437368719795,
      "grad_norm": 0.06864220649003983,
      "learning_rate": 2.560359479811564e-07,
      "loss": 0.001,
      "step": 283120
    },
    {
      "epoch": 1.8159078737657657,
      "grad_norm": 0.06728434562683105,
      "learning_rate": 2.5585916794591437e-07,
      "loss": 0.0014,
      "step": 283130
    },
    {
      "epoch": 1.8159720106595518,
      "grad_norm": 0.11732706427574158,
      "learning_rate": 2.556824473581465e-07,
      "loss": 0.0011,
      "step": 283140
    },
    {
      "epoch": 1.816036147553338,
      "grad_norm": 0.029023462906479836,
      "learning_rate": 2.5550578622006885e-07,
      "loss": 0.0014,
      "step": 283150
    },
    {
      "epoch": 1.816100284447124,
      "grad_norm": 0.027747154235839844,
      "learning_rate": 2.5532918453389456e-07,
      "loss": 0.0008,
      "step": 283160
    },
    {
      "epoch": 1.81616442134091,
      "grad_norm": 0.0065784878097474575,
      "learning_rate": 2.551526423018369e-07,
      "loss": 0.0008,
      "step": 283170
    },
    {
      "epoch": 1.816228558234696,
      "grad_norm": 0.06812255829572678,
      "learning_rate": 2.549761595261069e-07,
      "loss": 0.0009,
      "step": 283180
    },
    {
      "epoch": 1.8162926951284821,
      "grad_norm": 0.15295302867889404,
      "learning_rate": 2.5479973620891716e-07,
      "loss": 0.001,
      "step": 283190
    },
    {
      "epoch": 1.8163568320222683,
      "grad_norm": 0.06800895184278488,
      "learning_rate": 2.5462337235247824e-07,
      "loss": 0.0013,
      "step": 283200
    },
    {
      "epoch": 1.8164209689160544,
      "grad_norm": 0.1769361048936844,
      "learning_rate": 2.5444706795899996e-07,
      "loss": 0.0016,
      "step": 283210
    },
    {
      "epoch": 1.8164851058098406,
      "grad_norm": 0.02690737694501877,
      "learning_rate": 2.5427082303069116e-07,
      "loss": 0.0016,
      "step": 283220
    },
    {
      "epoch": 1.8165492427036267,
      "grad_norm": 0.023523934185504913,
      "learning_rate": 2.5409463756976117e-07,
      "loss": 0.0019,
      "step": 283230
    },
    {
      "epoch": 1.8166133795974129,
      "grad_norm": 0.09415130317211151,
      "learning_rate": 2.539185115784171e-07,
      "loss": 0.0006,
      "step": 283240
    },
    {
      "epoch": 1.8166775164911988,
      "grad_norm": 0.01110894326120615,
      "learning_rate": 2.537424450588655e-07,
      "loss": 0.0005,
      "step": 283250
    },
    {
      "epoch": 1.816741653384985,
      "grad_norm": 0.023204559460282326,
      "learning_rate": 2.5356643801331346e-07,
      "loss": 0.0011,
      "step": 283260
    },
    {
      "epoch": 1.8168057902787709,
      "grad_norm": 0.017421049997210503,
      "learning_rate": 2.533904904439666e-07,
      "loss": 0.0007,
      "step": 283270
    },
    {
      "epoch": 1.816869927172557,
      "grad_norm": 0.03126649558544159,
      "learning_rate": 2.532146023530291e-07,
      "loss": 0.0012,
      "step": 283280
    },
    {
      "epoch": 1.8169340640663432,
      "grad_norm": 0.06004180759191513,
      "learning_rate": 2.5303877374270426e-07,
      "loss": 0.0009,
      "step": 283290
    },
    {
      "epoch": 1.8169982009601293,
      "grad_norm": 0.015872960910201073,
      "learning_rate": 2.5286300461519754e-07,
      "loss": 0.0013,
      "step": 283300
    },
    {
      "epoch": 1.8170623378539155,
      "grad_norm": 0.014313939958810806,
      "learning_rate": 2.526872949727094e-07,
      "loss": 0.0006,
      "step": 283310
    },
    {
      "epoch": 1.8171264747477016,
      "grad_norm": 0.06666509062051773,
      "learning_rate": 2.525116448174425e-07,
      "loss": 0.0008,
      "step": 283320
    },
    {
      "epoch": 1.8171906116414875,
      "grad_norm": 0.20529133081436157,
      "learning_rate": 2.523360541515968e-07,
      "loss": 0.001,
      "step": 283330
    },
    {
      "epoch": 1.8172547485352737,
      "grad_norm": 0.11298438906669617,
      "learning_rate": 2.5216052297737437e-07,
      "loss": 0.001,
      "step": 283340
    },
    {
      "epoch": 1.8173188854290598,
      "grad_norm": 0.08906236290931702,
      "learning_rate": 2.5198505129697406e-07,
      "loss": 0.0007,
      "step": 283350
    },
    {
      "epoch": 1.8173830223228458,
      "grad_norm": 0.024035204201936722,
      "learning_rate": 2.51809639112594e-07,
      "loss": 0.0007,
      "step": 283360
    },
    {
      "epoch": 1.817447159216632,
      "grad_norm": 0.12804993987083435,
      "learning_rate": 2.516342864264321e-07,
      "loss": 0.001,
      "step": 283370
    },
    {
      "epoch": 1.817511296110418,
      "grad_norm": 0.03600526973605156,
      "learning_rate": 2.514589932406869e-07,
      "loss": 0.0011,
      "step": 283380
    },
    {
      "epoch": 1.8175754330042042,
      "grad_norm": 0.017293879762291908,
      "learning_rate": 2.51283759557554e-07,
      "loss": 0.0004,
      "step": 283390
    },
    {
      "epoch": 1.8176395698979904,
      "grad_norm": 0.043514467775821686,
      "learning_rate": 2.5110858537922944e-07,
      "loss": 0.0008,
      "step": 283400
    },
    {
      "epoch": 1.8177037067917765,
      "grad_norm": 0.12087168544530869,
      "learning_rate": 2.509334707079081e-07,
      "loss": 0.001,
      "step": 283410
    },
    {
      "epoch": 1.8177678436855624,
      "grad_norm": 0.15204310417175293,
      "learning_rate": 2.507584155457843e-07,
      "loss": 0.0009,
      "step": 283420
    },
    {
      "epoch": 1.8178319805793486,
      "grad_norm": 0.1251821219921112,
      "learning_rate": 2.5058341989505244e-07,
      "loss": 0.0007,
      "step": 283430
    },
    {
      "epoch": 1.8178961174731345,
      "grad_norm": 0.05553895980119705,
      "learning_rate": 2.5040848375790463e-07,
      "loss": 0.0016,
      "step": 283440
    },
    {
      "epoch": 1.8179602543669207,
      "grad_norm": 0.094098299741745,
      "learning_rate": 2.5023360713653244e-07,
      "loss": 0.0009,
      "step": 283450
    },
    {
      "epoch": 1.8180243912607068,
      "grad_norm": 0.050325796008110046,
      "learning_rate": 2.5005879003312805e-07,
      "loss": 0.0009,
      "step": 283460
    },
    {
      "epoch": 1.818088528154493,
      "grad_norm": 0.03272897005081177,
      "learning_rate": 2.4988403244988246e-07,
      "loss": 0.0009,
      "step": 283470
    },
    {
      "epoch": 1.818152665048279,
      "grad_norm": 0.03459775820374489,
      "learning_rate": 2.4970933438898394e-07,
      "loss": 0.0007,
      "step": 283480
    },
    {
      "epoch": 1.8182168019420653,
      "grad_norm": 0.010766434483230114,
      "learning_rate": 2.495346958526224e-07,
      "loss": 0.0013,
      "step": 283490
    },
    {
      "epoch": 1.8182809388358514,
      "grad_norm": 0.12688294053077698,
      "learning_rate": 2.4936011684298657e-07,
      "loss": 0.0011,
      "step": 283500
    },
    {
      "epoch": 1.8183450757296373,
      "grad_norm": 0.19279158115386963,
      "learning_rate": 2.4918559736226364e-07,
      "loss": 0.0025,
      "step": 283510
    },
    {
      "epoch": 1.8184092126234235,
      "grad_norm": 0.07030703127384186,
      "learning_rate": 2.490111374126403e-07,
      "loss": 0.0006,
      "step": 283520
    },
    {
      "epoch": 1.8184733495172094,
      "grad_norm": 0.10207617282867432,
      "learning_rate": 2.488367369963035e-07,
      "loss": 0.0006,
      "step": 283530
    },
    {
      "epoch": 1.8185374864109956,
      "grad_norm": 0.045885562896728516,
      "learning_rate": 2.4866239611543774e-07,
      "loss": 0.0011,
      "step": 283540
    },
    {
      "epoch": 1.8186016233047817,
      "grad_norm": 0.1675550937652588,
      "learning_rate": 2.484881147722279e-07,
      "loss": 0.0013,
      "step": 283550
    },
    {
      "epoch": 1.8186657601985678,
      "grad_norm": 0.08820303529500961,
      "learning_rate": 2.483138929688572e-07,
      "loss": 0.0007,
      "step": 283560
    },
    {
      "epoch": 1.818729897092354,
      "grad_norm": 0.07986678183078766,
      "learning_rate": 2.481397307075101e-07,
      "loss": 0.0009,
      "step": 283570
    },
    {
      "epoch": 1.8187940339861401,
      "grad_norm": 0.026571124792099,
      "learning_rate": 2.4796562799036806e-07,
      "loss": 0.0007,
      "step": 283580
    },
    {
      "epoch": 1.818858170879926,
      "grad_norm": 0.01867252215743065,
      "learning_rate": 2.4779158481961327e-07,
      "loss": 0.0014,
      "step": 283590
    },
    {
      "epoch": 1.8189223077737122,
      "grad_norm": 0.02360674925148487,
      "learning_rate": 2.476176011974252e-07,
      "loss": 0.0034,
      "step": 283600
    },
    {
      "epoch": 1.8189864446674981,
      "grad_norm": 0.10880941897630692,
      "learning_rate": 2.474436771259864e-07,
      "loss": 0.001,
      "step": 283610
    },
    {
      "epoch": 1.8190505815612843,
      "grad_norm": 0.05003716051578522,
      "learning_rate": 2.4726981260747405e-07,
      "loss": 0.0018,
      "step": 283620
    },
    {
      "epoch": 1.8191147184550704,
      "grad_norm": 0.06162728741765022,
      "learning_rate": 2.4709600764406813e-07,
      "loss": 0.0015,
      "step": 283630
    },
    {
      "epoch": 1.8191788553488566,
      "grad_norm": 0.13946183025836945,
      "learning_rate": 2.469222622379458e-07,
      "loss": 0.001,
      "step": 283640
    },
    {
      "epoch": 1.8192429922426427,
      "grad_norm": 0.004668702837079763,
      "learning_rate": 2.467485763912847e-07,
      "loss": 0.0013,
      "step": 283650
    },
    {
      "epoch": 1.819307129136429,
      "grad_norm": 0.015313344076275826,
      "learning_rate": 2.465749501062609e-07,
      "loss": 0.0009,
      "step": 283660
    },
    {
      "epoch": 1.819371266030215,
      "grad_norm": 0.12870186567306519,
      "learning_rate": 2.464013833850498e-07,
      "loss": 0.0019,
      "step": 283670
    },
    {
      "epoch": 1.819435402924001,
      "grad_norm": 0.06950012594461441,
      "learning_rate": 2.462278762298276e-07,
      "loss": 0.0019,
      "step": 283680
    },
    {
      "epoch": 1.8194995398177871,
      "grad_norm": 0.08987858146429062,
      "learning_rate": 2.4605442864276685e-07,
      "loss": 0.0005,
      "step": 283690
    },
    {
      "epoch": 1.819563676711573,
      "grad_norm": 0.003726414404809475,
      "learning_rate": 2.4588104062604257e-07,
      "loss": 0.0008,
      "step": 283700
    },
    {
      "epoch": 1.8196278136053592,
      "grad_norm": 0.09197190403938293,
      "learning_rate": 2.457077121818252e-07,
      "loss": 0.0009,
      "step": 283710
    },
    {
      "epoch": 1.8196919504991453,
      "grad_norm": 0.04489089176058769,
      "learning_rate": 2.4553444331228916e-07,
      "loss": 0.001,
      "step": 283720
    },
    {
      "epoch": 1.8197560873929315,
      "grad_norm": 0.09359274804592133,
      "learning_rate": 2.453612340196043e-07,
      "loss": 0.0014,
      "step": 283730
    },
    {
      "epoch": 1.8198202242867176,
      "grad_norm": 0.6765559315681458,
      "learning_rate": 2.4518808430594123e-07,
      "loss": 0.0067,
      "step": 283740
    },
    {
      "epoch": 1.8198843611805038,
      "grad_norm": 0.0031113028526306152,
      "learning_rate": 2.450149941734692e-07,
      "loss": 0.0005,
      "step": 283750
    },
    {
      "epoch": 1.8199484980742897,
      "grad_norm": 0.010244605131447315,
      "learning_rate": 2.448419636243582e-07,
      "loss": 0.0008,
      "step": 283760
    },
    {
      "epoch": 1.8200126349680759,
      "grad_norm": 0.013047495856881142,
      "learning_rate": 2.4466899266077594e-07,
      "loss": 0.0009,
      "step": 283770
    },
    {
      "epoch": 1.820076771861862,
      "grad_norm": 0.12160877883434296,
      "learning_rate": 2.4449608128488955e-07,
      "loss": 0.0015,
      "step": 283780
    },
    {
      "epoch": 1.820140908755648,
      "grad_norm": 0.07258080691099167,
      "learning_rate": 2.443232294988657e-07,
      "loss": 0.0026,
      "step": 283790
    },
    {
      "epoch": 1.820205045649434,
      "grad_norm": 0.13584206998348236,
      "learning_rate": 2.4415043730487084e-07,
      "loss": 0.0012,
      "step": 283800
    },
    {
      "epoch": 1.8202691825432202,
      "grad_norm": 0.03979848325252533,
      "learning_rate": 2.439777047050701e-07,
      "loss": 0.0014,
      "step": 283810
    },
    {
      "epoch": 1.8203333194370064,
      "grad_norm": 0.0566716194152832,
      "learning_rate": 2.438050317016272e-07,
      "loss": 0.0026,
      "step": 283820
    },
    {
      "epoch": 1.8203974563307925,
      "grad_norm": 0.09523949772119522,
      "learning_rate": 2.43632418296707e-07,
      "loss": 0.0007,
      "step": 283830
    },
    {
      "epoch": 1.8204615932245787,
      "grad_norm": 0.21772432327270508,
      "learning_rate": 2.434598644924724e-07,
      "loss": 0.0012,
      "step": 283840
    },
    {
      "epoch": 1.8205257301183646,
      "grad_norm": 0.029549170285463333,
      "learning_rate": 2.432873702910843e-07,
      "loss": 0.0018,
      "step": 283850
    },
    {
      "epoch": 1.8205898670121508,
      "grad_norm": 0.018051231279969215,
      "learning_rate": 2.431149356947049e-07,
      "loss": 0.0025,
      "step": 283860
    },
    {
      "epoch": 1.8206540039059367,
      "grad_norm": 0.004038951825350523,
      "learning_rate": 2.429425607054958e-07,
      "loss": 0.0009,
      "step": 283870
    },
    {
      "epoch": 1.8207181407997228,
      "grad_norm": 0.11834798008203506,
      "learning_rate": 2.4277024532561586e-07,
      "loss": 0.0008,
      "step": 283880
    },
    {
      "epoch": 1.820782277693509,
      "grad_norm": 0.07488865405321121,
      "learning_rate": 2.4259798955722504e-07,
      "loss": 0.0006,
      "step": 283890
    },
    {
      "epoch": 1.8208464145872951,
      "grad_norm": 0.0020906440913677216,
      "learning_rate": 2.4242579340248096e-07,
      "loss": 0.0006,
      "step": 283900
    },
    {
      "epoch": 1.8209105514810813,
      "grad_norm": 0.03752012923359871,
      "learning_rate": 2.4225365686354195e-07,
      "loss": 0.0004,
      "step": 283910
    },
    {
      "epoch": 1.8209746883748674,
      "grad_norm": 0.31805944442749023,
      "learning_rate": 2.420815799425652e-07,
      "loss": 0.0007,
      "step": 283920
    },
    {
      "epoch": 1.8210388252686536,
      "grad_norm": 0.05737582966685295,
      "learning_rate": 2.4190956264170673e-07,
      "loss": 0.0003,
      "step": 283930
    },
    {
      "epoch": 1.8211029621624395,
      "grad_norm": 0.03898722305893898,
      "learning_rate": 2.417376049631215e-07,
      "loss": 0.0005,
      "step": 283940
    },
    {
      "epoch": 1.8211670990562256,
      "grad_norm": 0.11303561925888062,
      "learning_rate": 2.41565706908965e-07,
      "loss": 0.0021,
      "step": 283950
    },
    {
      "epoch": 1.8212312359500116,
      "grad_norm": 0.07213585823774338,
      "learning_rate": 2.4139386848139156e-07,
      "loss": 0.0008,
      "step": 283960
    },
    {
      "epoch": 1.8212953728437977,
      "grad_norm": 0.0378580316901207,
      "learning_rate": 2.412220896825529e-07,
      "loss": 0.001,
      "step": 283970
    },
    {
      "epoch": 1.8213595097375839,
      "grad_norm": 0.027136176824569702,
      "learning_rate": 2.4105037051460277e-07,
      "loss": 0.0007,
      "step": 283980
    },
    {
      "epoch": 1.82142364663137,
      "grad_norm": 0.1134774312376976,
      "learning_rate": 2.408787109796934e-07,
      "loss": 0.0009,
      "step": 283990
    },
    {
      "epoch": 1.8214877835251562,
      "grad_norm": 0.06808760017156601,
      "learning_rate": 2.4070711107997467e-07,
      "loss": 0.0013,
      "step": 284000
    },
    {
      "epoch": 1.8215519204189423,
      "grad_norm": 0.03774027153849602,
      "learning_rate": 2.4053557081759657e-07,
      "loss": 0.0008,
      "step": 284010
    },
    {
      "epoch": 1.8216160573127282,
      "grad_norm": 0.04879768192768097,
      "learning_rate": 2.403640901947102e-07,
      "loss": 0.0015,
      "step": 284020
    },
    {
      "epoch": 1.8216801942065144,
      "grad_norm": 0.0474613681435585,
      "learning_rate": 2.401926692134632e-07,
      "loss": 0.0008,
      "step": 284030
    },
    {
      "epoch": 1.8217443311003003,
      "grad_norm": 0.14003092050552368,
      "learning_rate": 2.400213078760039e-07,
      "loss": 0.0015,
      "step": 284040
    },
    {
      "epoch": 1.8218084679940865,
      "grad_norm": 0.10043643414974213,
      "learning_rate": 2.3985000618447884e-07,
      "loss": 0.0014,
      "step": 284050
    },
    {
      "epoch": 1.8218726048878726,
      "grad_norm": 0.08790823072195053,
      "learning_rate": 2.3967876414103586e-07,
      "loss": 0.0006,
      "step": 284060
    },
    {
      "epoch": 1.8219367417816588,
      "grad_norm": 0.05833596736192703,
      "learning_rate": 2.3950758174781984e-07,
      "loss": 0.0011,
      "step": 284070
    },
    {
      "epoch": 1.822000878675445,
      "grad_norm": 0.05277148261666298,
      "learning_rate": 2.393364590069769e-07,
      "loss": 0.0017,
      "step": 284080
    },
    {
      "epoch": 1.822065015569231,
      "grad_norm": 0.03449109569191933,
      "learning_rate": 2.391653959206491e-07,
      "loss": 0.003,
      "step": 284090
    },
    {
      "epoch": 1.8221291524630172,
      "grad_norm": 0.12976112961769104,
      "learning_rate": 2.3899439249098264e-07,
      "loss": 0.0011,
      "step": 284100
    },
    {
      "epoch": 1.8221932893568031,
      "grad_norm": 0.04298318549990654,
      "learning_rate": 2.3882344872011854e-07,
      "loss": 0.0012,
      "step": 284110
    },
    {
      "epoch": 1.8222574262505893,
      "grad_norm": 0.0845508873462677,
      "learning_rate": 2.386525646101989e-07,
      "loss": 0.003,
      "step": 284120
    },
    {
      "epoch": 1.8223215631443752,
      "grad_norm": 0.01747436821460724,
      "learning_rate": 2.3848174016336657e-07,
      "loss": 0.002,
      "step": 284130
    },
    {
      "epoch": 1.8223857000381614,
      "grad_norm": 0.017732495442032814,
      "learning_rate": 2.3831097538176086e-07,
      "loss": 0.0008,
      "step": 284140
    },
    {
      "epoch": 1.8224498369319475,
      "grad_norm": 0.04959338530898094,
      "learning_rate": 2.3814027026752118e-07,
      "loss": 0.0008,
      "step": 284150
    },
    {
      "epoch": 1.8225139738257337,
      "grad_norm": 0.027588363736867905,
      "learning_rate": 2.3796962482278697e-07,
      "loss": 0.001,
      "step": 284160
    },
    {
      "epoch": 1.8225781107195198,
      "grad_norm": 0.09255266934633255,
      "learning_rate": 2.377990390496976e-07,
      "loss": 0.0013,
      "step": 284170
    },
    {
      "epoch": 1.822642247613306,
      "grad_norm": 0.028061656281352043,
      "learning_rate": 2.376285129503897e-07,
      "loss": 0.0018,
      "step": 284180
    },
    {
      "epoch": 1.822706384507092,
      "grad_norm": 0.007896821945905685,
      "learning_rate": 2.3745804652699933e-07,
      "loss": 0.0011,
      "step": 284190
    },
    {
      "epoch": 1.822770521400878,
      "grad_norm": 0.05308055505156517,
      "learning_rate": 2.372876397816637e-07,
      "loss": 0.0008,
      "step": 284200
    },
    {
      "epoch": 1.8228346582946642,
      "grad_norm": 0.05304637551307678,
      "learning_rate": 2.3711729271651774e-07,
      "loss": 0.0018,
      "step": 284210
    },
    {
      "epoch": 1.82289879518845,
      "grad_norm": 0.0017153015360236168,
      "learning_rate": 2.3694700533369641e-07,
      "loss": 0.0007,
      "step": 284220
    },
    {
      "epoch": 1.8229629320822363,
      "grad_norm": 0.015460219234228134,
      "learning_rate": 2.36776777635333e-07,
      "loss": 0.0013,
      "step": 284230
    },
    {
      "epoch": 1.8230270689760224,
      "grad_norm": 0.049073509871959686,
      "learning_rate": 2.3660660962356086e-07,
      "loss": 0.0018,
      "step": 284240
    },
    {
      "epoch": 1.8230912058698086,
      "grad_norm": 0.11652687191963196,
      "learning_rate": 2.364365013005121e-07,
      "loss": 0.0016,
      "step": 284250
    },
    {
      "epoch": 1.8231553427635947,
      "grad_norm": 0.008816922083497047,
      "learning_rate": 2.3626645266831893e-07,
      "loss": 0.0019,
      "step": 284260
    },
    {
      "epoch": 1.8232194796573808,
      "grad_norm": 0.03230687603354454,
      "learning_rate": 2.3609646372911133e-07,
      "loss": 0.0008,
      "step": 284270
    },
    {
      "epoch": 1.8232836165511668,
      "grad_norm": 0.07474133372306824,
      "learning_rate": 2.3592653448501924e-07,
      "loss": 0.001,
      "step": 284280
    },
    {
      "epoch": 1.823347753444953,
      "grad_norm": 0.0740194246172905,
      "learning_rate": 2.3575666493817373e-07,
      "loss": 0.0008,
      "step": 284290
    },
    {
      "epoch": 1.8234118903387388,
      "grad_norm": 0.07866983115673065,
      "learning_rate": 2.3558685509070146e-07,
      "loss": 0.0018,
      "step": 284300
    },
    {
      "epoch": 1.823476027232525,
      "grad_norm": 0.004387131426483393,
      "learning_rate": 2.3541710494473124e-07,
      "loss": 0.0011,
      "step": 284310
    },
    {
      "epoch": 1.8235401641263111,
      "grad_norm": 0.03439735248684883,
      "learning_rate": 2.352474145023892e-07,
      "loss": 0.0007,
      "step": 284320
    },
    {
      "epoch": 1.8236043010200973,
      "grad_norm": 0.11432670801877975,
      "learning_rate": 2.3507778376580303e-07,
      "loss": 0.0006,
      "step": 284330
    },
    {
      "epoch": 1.8236684379138834,
      "grad_norm": 0.05256522446870804,
      "learning_rate": 2.3490821273709774e-07,
      "loss": 0.0015,
      "step": 284340
    },
    {
      "epoch": 1.8237325748076696,
      "grad_norm": 0.11117374151945114,
      "learning_rate": 2.3473870141839772e-07,
      "loss": 0.001,
      "step": 284350
    },
    {
      "epoch": 1.8237967117014557,
      "grad_norm": 0.06200719624757767,
      "learning_rate": 2.345692498118274e-07,
      "loss": 0.001,
      "step": 284360
    },
    {
      "epoch": 1.8238608485952417,
      "grad_norm": 0.044623877853155136,
      "learning_rate": 2.3439985791951058e-07,
      "loss": 0.0008,
      "step": 284370
    },
    {
      "epoch": 1.8239249854890278,
      "grad_norm": 0.018259888514876366,
      "learning_rate": 2.3423052574356898e-07,
      "loss": 0.0006,
      "step": 284380
    },
    {
      "epoch": 1.8239891223828137,
      "grad_norm": 0.0028362676966935396,
      "learning_rate": 2.3406125328612472e-07,
      "loss": 0.0004,
      "step": 284390
    },
    {
      "epoch": 1.8240532592766,
      "grad_norm": 0.014020822010934353,
      "learning_rate": 2.3389204054930005e-07,
      "loss": 0.0011,
      "step": 284400
    },
    {
      "epoch": 1.824117396170386,
      "grad_norm": 0.021346304565668106,
      "learning_rate": 2.3372288753521433e-07,
      "loss": 0.0011,
      "step": 284410
    },
    {
      "epoch": 1.8241815330641722,
      "grad_norm": 0.12803484499454498,
      "learning_rate": 2.3355379424598702e-07,
      "loss": 0.0016,
      "step": 284420
    },
    {
      "epoch": 1.8242456699579583,
      "grad_norm": 0.07047094404697418,
      "learning_rate": 2.333847606837364e-07,
      "loss": 0.005,
      "step": 284430
    },
    {
      "epoch": 1.8243098068517445,
      "grad_norm": 0.21268074214458466,
      "learning_rate": 2.332157868505819e-07,
      "loss": 0.0006,
      "step": 284440
    },
    {
      "epoch": 1.8243739437455304,
      "grad_norm": 0.06028466299176216,
      "learning_rate": 2.330468727486407e-07,
      "loss": 0.002,
      "step": 284450
    },
    {
      "epoch": 1.8244380806393166,
      "grad_norm": 0.020619206130504608,
      "learning_rate": 2.328780183800289e-07,
      "loss": 0.0008,
      "step": 284460
    },
    {
      "epoch": 1.8245022175331025,
      "grad_norm": 0.004687672946602106,
      "learning_rate": 2.32709223746862e-07,
      "loss": 0.0024,
      "step": 284470
    },
    {
      "epoch": 1.8245663544268886,
      "grad_norm": 0.007976637221872807,
      "learning_rate": 2.325404888512567e-07,
      "loss": 0.0005,
      "step": 284480
    },
    {
      "epoch": 1.8246304913206748,
      "grad_norm": 0.07855124026536942,
      "learning_rate": 2.323718136953257e-07,
      "loss": 0.0008,
      "step": 284490
    },
    {
      "epoch": 1.824694628214461,
      "grad_norm": 0.1139468252658844,
      "learning_rate": 2.3220319828118342e-07,
      "loss": 0.0008,
      "step": 284500
    },
    {
      "epoch": 1.824758765108247,
      "grad_norm": 0.0543724000453949,
      "learning_rate": 2.320346426109421e-07,
      "loss": 0.0006,
      "step": 284510
    },
    {
      "epoch": 1.8248229020020332,
      "grad_norm": 0.024786775931715965,
      "learning_rate": 2.3186614668671504e-07,
      "loss": 0.0007,
      "step": 284520
    },
    {
      "epoch": 1.8248870388958194,
      "grad_norm": 0.07218330353498459,
      "learning_rate": 2.3169771051061274e-07,
      "loss": 0.0008,
      "step": 284530
    },
    {
      "epoch": 1.8249511757896053,
      "grad_norm": 0.023913349956274033,
      "learning_rate": 2.315293340847452e-07,
      "loss": 0.0013,
      "step": 284540
    },
    {
      "epoch": 1.8250153126833915,
      "grad_norm": 0.07372229546308517,
      "learning_rate": 2.3136101741122352e-07,
      "loss": 0.0013,
      "step": 284550
    },
    {
      "epoch": 1.8250794495771774,
      "grad_norm": 0.04746563732624054,
      "learning_rate": 2.311927604921571e-07,
      "loss": 0.0011,
      "step": 284560
    },
    {
      "epoch": 1.8251435864709635,
      "grad_norm": 0.08703367412090302,
      "learning_rate": 2.3102456332965317e-07,
      "loss": 0.001,
      "step": 284570
    },
    {
      "epoch": 1.8252077233647497,
      "grad_norm": 0.10042295604944229,
      "learning_rate": 2.3085642592581892e-07,
      "loss": 0.0009,
      "step": 284580
    },
    {
      "epoch": 1.8252718602585358,
      "grad_norm": 0.059683751314878464,
      "learning_rate": 2.306883482827632e-07,
      "loss": 0.0013,
      "step": 284590
    },
    {
      "epoch": 1.825335997152322,
      "grad_norm": 0.21237105131149292,
      "learning_rate": 2.3052033040259048e-07,
      "loss": 0.0009,
      "step": 284600
    },
    {
      "epoch": 1.8254001340461081,
      "grad_norm": 0.047385189682245255,
      "learning_rate": 2.3035237228740735e-07,
      "loss": 0.0006,
      "step": 284610
    },
    {
      "epoch": 1.8254642709398943,
      "grad_norm": 0.02855606935918331,
      "learning_rate": 2.3018447393931664e-07,
      "loss": 0.0013,
      "step": 284620
    },
    {
      "epoch": 1.8255284078336802,
      "grad_norm": 0.011425605043768883,
      "learning_rate": 2.300166353604244e-07,
      "loss": 0.0012,
      "step": 284630
    },
    {
      "epoch": 1.8255925447274663,
      "grad_norm": 0.04062425717711449,
      "learning_rate": 2.2984885655283285e-07,
      "loss": 0.0014,
      "step": 284640
    },
    {
      "epoch": 1.8256566816212523,
      "grad_norm": 0.17988109588623047,
      "learning_rate": 2.2968113751864417e-07,
      "loss": 0.0007,
      "step": 284650
    },
    {
      "epoch": 1.8257208185150384,
      "grad_norm": 0.013708336278796196,
      "learning_rate": 2.295134782599595e-07,
      "loss": 0.0016,
      "step": 284660
    },
    {
      "epoch": 1.8257849554088246,
      "grad_norm": 0.03549422323703766,
      "learning_rate": 2.2934587877888105e-07,
      "loss": 0.0014,
      "step": 284670
    },
    {
      "epoch": 1.8258490923026107,
      "grad_norm": 0.02807396836578846,
      "learning_rate": 2.291783390775082e-07,
      "loss": 0.0015,
      "step": 284680
    },
    {
      "epoch": 1.8259132291963969,
      "grad_norm": 0.041874222457408905,
      "learning_rate": 2.2901085915793986e-07,
      "loss": 0.0019,
      "step": 284690
    },
    {
      "epoch": 1.825977366090183,
      "grad_norm": 0.1899644285440445,
      "learning_rate": 2.2884343902227547e-07,
      "loss": 0.0018,
      "step": 284700
    },
    {
      "epoch": 1.826041502983969,
      "grad_norm": 0.07992995530366898,
      "learning_rate": 2.2867607867261332e-07,
      "loss": 0.0013,
      "step": 284710
    },
    {
      "epoch": 1.826105639877755,
      "grad_norm": 0.04594377428293228,
      "learning_rate": 2.2850877811104955e-07,
      "loss": 0.0019,
      "step": 284720
    },
    {
      "epoch": 1.826169776771541,
      "grad_norm": 0.09122800827026367,
      "learning_rate": 2.2834153733968023e-07,
      "loss": 0.0011,
      "step": 284730
    },
    {
      "epoch": 1.8262339136653272,
      "grad_norm": 0.07416558265686035,
      "learning_rate": 2.2817435636060258e-07,
      "loss": 0.0014,
      "step": 284740
    },
    {
      "epoch": 1.8262980505591133,
      "grad_norm": 0.0800134614109993,
      "learning_rate": 2.280072351759105e-07,
      "loss": 0.0003,
      "step": 284750
    },
    {
      "epoch": 1.8263621874528995,
      "grad_norm": 0.04095055162906647,
      "learning_rate": 2.2784017378769784e-07,
      "loss": 0.0017,
      "step": 284760
    },
    {
      "epoch": 1.8264263243466856,
      "grad_norm": 0.0665946677327156,
      "learning_rate": 2.276731721980585e-07,
      "loss": 0.0007,
      "step": 284770
    },
    {
      "epoch": 1.8264904612404718,
      "grad_norm": 0.06196899712085724,
      "learning_rate": 2.2750623040908527e-07,
      "loss": 0.0006,
      "step": 284780
    },
    {
      "epoch": 1.826554598134258,
      "grad_norm": 0.05296425148844719,
      "learning_rate": 2.2733934842286975e-07,
      "loss": 0.0011,
      "step": 284790
    },
    {
      "epoch": 1.8266187350280438,
      "grad_norm": 0.026383887976408005,
      "learning_rate": 2.271725262415031e-07,
      "loss": 0.0018,
      "step": 284800
    },
    {
      "epoch": 1.82668287192183,
      "grad_norm": 0.06328203529119492,
      "learning_rate": 2.2700576386707584e-07,
      "loss": 0.0007,
      "step": 284810
    },
    {
      "epoch": 1.826747008815616,
      "grad_norm": 0.007411746773868799,
      "learning_rate": 2.2683906130167742e-07,
      "loss": 0.0006,
      "step": 284820
    },
    {
      "epoch": 1.826811145709402,
      "grad_norm": 0.13926292955875397,
      "learning_rate": 2.266724185473973e-07,
      "loss": 0.0007,
      "step": 284830
    },
    {
      "epoch": 1.8268752826031882,
      "grad_norm": 0.014773097820580006,
      "learning_rate": 2.2650583560632266e-07,
      "loss": 0.0005,
      "step": 284840
    },
    {
      "epoch": 1.8269394194969744,
      "grad_norm": 0.04269682243466377,
      "learning_rate": 2.2633931248054187e-07,
      "loss": 0.0007,
      "step": 284850
    },
    {
      "epoch": 1.8270035563907605,
      "grad_norm": 0.017428934574127197,
      "learning_rate": 2.26172849172141e-07,
      "loss": 0.0009,
      "step": 284860
    },
    {
      "epoch": 1.8270676932845467,
      "grad_norm": 0.060745932161808014,
      "learning_rate": 2.2600644568320618e-07,
      "loss": 0.001,
      "step": 284870
    },
    {
      "epoch": 1.8271318301783326,
      "grad_norm": 0.0948595181107521,
      "learning_rate": 2.258401020158224e-07,
      "loss": 0.0012,
      "step": 284880
    },
    {
      "epoch": 1.8271959670721187,
      "grad_norm": 0.03808773308992386,
      "learning_rate": 2.2567381817207412e-07,
      "loss": 0.0017,
      "step": 284890
    },
    {
      "epoch": 1.8272601039659049,
      "grad_norm": 0.01708976924419403,
      "learning_rate": 2.2550759415404577e-07,
      "loss": 0.0009,
      "step": 284900
    },
    {
      "epoch": 1.8273242408596908,
      "grad_norm": 0.04463403299450874,
      "learning_rate": 2.2534142996381902e-07,
      "loss": 0.0007,
      "step": 284910
    },
    {
      "epoch": 1.827388377753477,
      "grad_norm": 0.049894046038389206,
      "learning_rate": 2.251753256034761e-07,
      "loss": 0.001,
      "step": 284920
    },
    {
      "epoch": 1.827452514647263,
      "grad_norm": 0.09170003980398178,
      "learning_rate": 2.2500928107509923e-07,
      "loss": 0.0005,
      "step": 284930
    },
    {
      "epoch": 1.8275166515410493,
      "grad_norm": 0.0323021374642849,
      "learning_rate": 2.2484329638076896e-07,
      "loss": 0.0008,
      "step": 284940
    },
    {
      "epoch": 1.8275807884348354,
      "grad_norm": 0.02366352640092373,
      "learning_rate": 2.2467737152256474e-07,
      "loss": 0.0024,
      "step": 284950
    },
    {
      "epoch": 1.8276449253286216,
      "grad_norm": 0.03775579854846001,
      "learning_rate": 2.2451150650256548e-07,
      "loss": 0.0008,
      "step": 284960
    },
    {
      "epoch": 1.8277090622224075,
      "grad_norm": 0.0916261374950409,
      "learning_rate": 2.243457013228506e-07,
      "loss": 0.0015,
      "step": 284970
    },
    {
      "epoch": 1.8277731991161936,
      "grad_norm": 0.06144082918763161,
      "learning_rate": 2.2417995598549735e-07,
      "loss": 0.0023,
      "step": 284980
    },
    {
      "epoch": 1.8278373360099796,
      "grad_norm": 0.44196850061416626,
      "learning_rate": 2.2401427049258239e-07,
      "loss": 0.0015,
      "step": 284990
    },
    {
      "epoch": 1.8279014729037657,
      "grad_norm": 0.5167443752288818,
      "learning_rate": 2.238486448461813e-07,
      "loss": 0.0034,
      "step": 285000
    },
    {
      "epoch": 1.8279656097975518,
      "grad_norm": 0.031271178275346756,
      "learning_rate": 2.2368307904837072e-07,
      "loss": 0.0015,
      "step": 285010
    },
    {
      "epoch": 1.828029746691338,
      "grad_norm": 0.05039814114570618,
      "learning_rate": 2.2351757310122458e-07,
      "loss": 0.0015,
      "step": 285020
    },
    {
      "epoch": 1.8280938835851241,
      "grad_norm": 0.0538700595498085,
      "learning_rate": 2.233521270068173e-07,
      "loss": 0.0013,
      "step": 285030
    },
    {
      "epoch": 1.8281580204789103,
      "grad_norm": 0.011439146474003792,
      "learning_rate": 2.2318674076722114e-07,
      "loss": 0.0005,
      "step": 285040
    },
    {
      "epoch": 1.8282221573726964,
      "grad_norm": 0.030994849279522896,
      "learning_rate": 2.230214143845094e-07,
      "loss": 0.0006,
      "step": 285050
    },
    {
      "epoch": 1.8282862942664824,
      "grad_norm": 0.04261430725455284,
      "learning_rate": 2.2285614786075382e-07,
      "loss": 0.0008,
      "step": 285060
    },
    {
      "epoch": 1.8283504311602685,
      "grad_norm": 0.010509190149605274,
      "learning_rate": 2.2269094119802438e-07,
      "loss": 0.0011,
      "step": 285070
    },
    {
      "epoch": 1.8284145680540544,
      "grad_norm": 0.07600738108158112,
      "learning_rate": 2.2252579439839105e-07,
      "loss": 0.0008,
      "step": 285080
    },
    {
      "epoch": 1.8284787049478406,
      "grad_norm": 0.04539591446518898,
      "learning_rate": 2.22360707463925e-07,
      "loss": 0.0006,
      "step": 285090
    },
    {
      "epoch": 1.8285428418416267,
      "grad_norm": 0.0028495138976722956,
      "learning_rate": 2.2219568039669347e-07,
      "loss": 0.0015,
      "step": 285100
    },
    {
      "epoch": 1.828606978735413,
      "grad_norm": 0.07312221825122833,
      "learning_rate": 2.2203071319876422e-07,
      "loss": 0.0015,
      "step": 285110
    },
    {
      "epoch": 1.828671115629199,
      "grad_norm": 0.060156095772981644,
      "learning_rate": 2.2186580587220562e-07,
      "loss": 0.0012,
      "step": 285120
    },
    {
      "epoch": 1.8287352525229852,
      "grad_norm": 0.18297149240970612,
      "learning_rate": 2.2170095841908322e-07,
      "loss": 0.0012,
      "step": 285130
    },
    {
      "epoch": 1.8287993894167711,
      "grad_norm": 0.06519244611263275,
      "learning_rate": 2.2153617084146316e-07,
      "loss": 0.0016,
      "step": 285140
    },
    {
      "epoch": 1.8288635263105573,
      "grad_norm": 0.16978222131729126,
      "learning_rate": 2.2137144314140879e-07,
      "loss": 0.0008,
      "step": 285150
    },
    {
      "epoch": 1.8289276632043432,
      "grad_norm": 0.061636921018362045,
      "learning_rate": 2.2120677532098677e-07,
      "loss": 0.0005,
      "step": 285160
    },
    {
      "epoch": 1.8289918000981293,
      "grad_norm": 0.07009384781122208,
      "learning_rate": 2.2104216738225882e-07,
      "loss": 0.0017,
      "step": 285170
    },
    {
      "epoch": 1.8290559369919155,
      "grad_norm": 0.09590679407119751,
      "learning_rate": 2.2087761932728768e-07,
      "loss": 0.0015,
      "step": 285180
    },
    {
      "epoch": 1.8291200738857016,
      "grad_norm": 0.06868293881416321,
      "learning_rate": 2.207131311581351e-07,
      "loss": 0.0012,
      "step": 285190
    },
    {
      "epoch": 1.8291842107794878,
      "grad_norm": 0.06806357949972153,
      "learning_rate": 2.205487028768638e-07,
      "loss": 0.0016,
      "step": 285200
    },
    {
      "epoch": 1.829248347673274,
      "grad_norm": 0.040417108684778214,
      "learning_rate": 2.2038433448553275e-07,
      "loss": 0.0007,
      "step": 285210
    },
    {
      "epoch": 1.82931248456706,
      "grad_norm": 0.06480779498815536,
      "learning_rate": 2.2022002598620084e-07,
      "loss": 0.0024,
      "step": 285220
    },
    {
      "epoch": 1.829376621460846,
      "grad_norm": 0.051375679671764374,
      "learning_rate": 2.200557773809292e-07,
      "loss": 0.001,
      "step": 285230
    },
    {
      "epoch": 1.8294407583546322,
      "grad_norm": 0.049823835492134094,
      "learning_rate": 2.1989158867177506e-07,
      "loss": 0.0004,
      "step": 285240
    },
    {
      "epoch": 1.829504895248418,
      "grad_norm": 0.1641385555267334,
      "learning_rate": 2.1972745986079513e-07,
      "loss": 0.0011,
      "step": 285250
    },
    {
      "epoch": 1.8295690321422042,
      "grad_norm": 0.0011703071650117636,
      "learning_rate": 2.1956339095004608e-07,
      "loss": 0.0014,
      "step": 285260
    },
    {
      "epoch": 1.8296331690359904,
      "grad_norm": 0.1630324125289917,
      "learning_rate": 2.1939938194158517e-07,
      "loss": 0.0015,
      "step": 285270
    },
    {
      "epoch": 1.8296973059297765,
      "grad_norm": 0.007984250783920288,
      "learning_rate": 2.192354328374663e-07,
      "loss": 0.0013,
      "step": 285280
    },
    {
      "epoch": 1.8297614428235627,
      "grad_norm": 0.03937610611319542,
      "learning_rate": 2.1907154363974392e-07,
      "loss": 0.0013,
      "step": 285290
    },
    {
      "epoch": 1.8298255797173488,
      "grad_norm": 0.04710065573453903,
      "learning_rate": 2.18907714350472e-07,
      "loss": 0.0012,
      "step": 285300
    },
    {
      "epoch": 1.8298897166111348,
      "grad_norm": 0.10098345577716827,
      "learning_rate": 2.187439449717038e-07,
      "loss": 0.0008,
      "step": 285310
    },
    {
      "epoch": 1.829953853504921,
      "grad_norm": 0.015499911271035671,
      "learning_rate": 2.1858023550549113e-07,
      "loss": 0.0009,
      "step": 285320
    },
    {
      "epoch": 1.830017990398707,
      "grad_norm": 0.04580983147025108,
      "learning_rate": 2.1841658595388503e-07,
      "loss": 0.0009,
      "step": 285330
    },
    {
      "epoch": 1.830082127292493,
      "grad_norm": 0.06571226567029953,
      "learning_rate": 2.1825299631893615e-07,
      "loss": 0.0008,
      "step": 285340
    },
    {
      "epoch": 1.8301462641862791,
      "grad_norm": 0.10527317970991135,
      "learning_rate": 2.1808946660269503e-07,
      "loss": 0.0015,
      "step": 285350
    },
    {
      "epoch": 1.8302104010800653,
      "grad_norm": 0.023459885269403458,
      "learning_rate": 2.179259968072106e-07,
      "loss": 0.0008,
      "step": 285360
    },
    {
      "epoch": 1.8302745379738514,
      "grad_norm": 0.08581940084695816,
      "learning_rate": 2.1776258693453068e-07,
      "loss": 0.001,
      "step": 285370
    },
    {
      "epoch": 1.8303386748676376,
      "grad_norm": 0.10950316488742828,
      "learning_rate": 2.1759923698670415e-07,
      "loss": 0.0014,
      "step": 285380
    },
    {
      "epoch": 1.8304028117614237,
      "grad_norm": 0.24941828846931458,
      "learning_rate": 2.1743594696577662e-07,
      "loss": 0.0021,
      "step": 285390
    },
    {
      "epoch": 1.8304669486552096,
      "grad_norm": 0.099155955016613,
      "learning_rate": 2.172727168737948e-07,
      "loss": 0.0005,
      "step": 285400
    },
    {
      "epoch": 1.8305310855489958,
      "grad_norm": 0.039132773876190186,
      "learning_rate": 2.171095467128037e-07,
      "loss": 0.0007,
      "step": 285410
    },
    {
      "epoch": 1.8305952224427817,
      "grad_norm": 0.12010085582733154,
      "learning_rate": 2.1694643648484892e-07,
      "loss": 0.0013,
      "step": 285420
    },
    {
      "epoch": 1.8306593593365679,
      "grad_norm": 0.04474587365984917,
      "learning_rate": 2.1678338619197325e-07,
      "loss": 0.0006,
      "step": 285430
    },
    {
      "epoch": 1.830723496230354,
      "grad_norm": 0.08558463305234909,
      "learning_rate": 2.1662039583622064e-07,
      "loss": 0.0008,
      "step": 285440
    },
    {
      "epoch": 1.8307876331241402,
      "grad_norm": 0.005080286413431168,
      "learning_rate": 2.164574654196322e-07,
      "loss": 0.0015,
      "step": 285450
    },
    {
      "epoch": 1.8308517700179263,
      "grad_norm": 0.055858634412288666,
      "learning_rate": 2.1629459494425131e-07,
      "loss": 0.001,
      "step": 285460
    },
    {
      "epoch": 1.8309159069117125,
      "grad_norm": 0.018276643007993698,
      "learning_rate": 2.1613178441211745e-07,
      "loss": 0.0006,
      "step": 285470
    },
    {
      "epoch": 1.8309800438054986,
      "grad_norm": 0.04330141469836235,
      "learning_rate": 2.1596903382527178e-07,
      "loss": 0.0009,
      "step": 285480
    },
    {
      "epoch": 1.8310441806992845,
      "grad_norm": 0.061170391738414764,
      "learning_rate": 2.1580634318575265e-07,
      "loss": 0.0021,
      "step": 285490
    },
    {
      "epoch": 1.8311083175930707,
      "grad_norm": 0.016236858442425728,
      "learning_rate": 2.1564371249560012e-07,
      "loss": 0.0005,
      "step": 285500
    },
    {
      "epoch": 1.8311724544868566,
      "grad_norm": 0.19410637021064758,
      "learning_rate": 2.1548114175685086e-07,
      "loss": 0.0022,
      "step": 285510
    },
    {
      "epoch": 1.8312365913806428,
      "grad_norm": 0.13653182983398438,
      "learning_rate": 2.153186309715416e-07,
      "loss": 0.0013,
      "step": 285520
    },
    {
      "epoch": 1.831300728274429,
      "grad_norm": 0.2021653950214386,
      "learning_rate": 2.1515618014171014e-07,
      "loss": 0.0038,
      "step": 285530
    },
    {
      "epoch": 1.831364865168215,
      "grad_norm": 0.15602578222751617,
      "learning_rate": 2.1499378926939152e-07,
      "loss": 0.0007,
      "step": 285540
    },
    {
      "epoch": 1.8314290020620012,
      "grad_norm": 0.01986047253012657,
      "learning_rate": 2.1483145835662021e-07,
      "loss": 0.0008,
      "step": 285550
    },
    {
      "epoch": 1.8314931389557874,
      "grad_norm": 0.09452295303344727,
      "learning_rate": 2.1466918740543075e-07,
      "loss": 0.001,
      "step": 285560
    },
    {
      "epoch": 1.8315572758495733,
      "grad_norm": 0.09902264922857285,
      "learning_rate": 2.1450697641785646e-07,
      "loss": 0.0009,
      "step": 285570
    },
    {
      "epoch": 1.8316214127433594,
      "grad_norm": 0.002422323450446129,
      "learning_rate": 2.143448253959296e-07,
      "loss": 0.0011,
      "step": 285580
    },
    {
      "epoch": 1.8316855496371454,
      "grad_norm": 0.1396687924861908,
      "learning_rate": 2.1418273434168303e-07,
      "loss": 0.001,
      "step": 285590
    },
    {
      "epoch": 1.8317496865309315,
      "grad_norm": 0.5556753873825073,
      "learning_rate": 2.140207032571462e-07,
      "loss": 0.0022,
      "step": 285600
    },
    {
      "epoch": 1.8318138234247177,
      "grad_norm": 0.052254196256399155,
      "learning_rate": 2.1385873214435082e-07,
      "loss": 0.0004,
      "step": 285610
    },
    {
      "epoch": 1.8318779603185038,
      "grad_norm": 0.129730686545372,
      "learning_rate": 2.136968210053264e-07,
      "loss": 0.0012,
      "step": 285620
    },
    {
      "epoch": 1.83194209721229,
      "grad_norm": 0.1199645921587944,
      "learning_rate": 2.1353496984210187e-07,
      "loss": 0.0017,
      "step": 285630
    },
    {
      "epoch": 1.832006234106076,
      "grad_norm": 0.01759057678282261,
      "learning_rate": 2.1337317865670393e-07,
      "loss": 0.0006,
      "step": 285640
    },
    {
      "epoch": 1.8320703709998623,
      "grad_norm": 0.21323983371257782,
      "learning_rate": 2.1321144745116206e-07,
      "loss": 0.0011,
      "step": 285650
    },
    {
      "epoch": 1.8321345078936482,
      "grad_norm": 0.023735884577035904,
      "learning_rate": 2.1304977622750135e-07,
      "loss": 0.0004,
      "step": 285660
    },
    {
      "epoch": 1.8321986447874343,
      "grad_norm": 0.09157714247703552,
      "learning_rate": 2.1288816498774733e-07,
      "loss": 0.0034,
      "step": 285670
    },
    {
      "epoch": 1.8322627816812203,
      "grad_norm": 0.05838904529809952,
      "learning_rate": 2.1272661373392734e-07,
      "loss": 0.0014,
      "step": 285680
    },
    {
      "epoch": 1.8323269185750064,
      "grad_norm": 0.14488168060779572,
      "learning_rate": 2.125651224680636e-07,
      "loss": 0.0014,
      "step": 285690
    },
    {
      "epoch": 1.8323910554687926,
      "grad_norm": 0.13726484775543213,
      "learning_rate": 2.1240369119218062e-07,
      "loss": 0.0007,
      "step": 285700
    },
    {
      "epoch": 1.8324551923625787,
      "grad_norm": 0.03192019462585449,
      "learning_rate": 2.1224231990830013e-07,
      "loss": 0.001,
      "step": 285710
    },
    {
      "epoch": 1.8325193292563648,
      "grad_norm": 0.0978066623210907,
      "learning_rate": 2.120810086184455e-07,
      "loss": 0.001,
      "step": 285720
    },
    {
      "epoch": 1.832583466150151,
      "grad_norm": 0.0054242261685431,
      "learning_rate": 2.1191975732463843e-07,
      "loss": 0.0005,
      "step": 285730
    },
    {
      "epoch": 1.832647603043937,
      "grad_norm": 0.054589856415987015,
      "learning_rate": 2.1175856602889843e-07,
      "loss": 0.0012,
      "step": 285740
    },
    {
      "epoch": 1.832711739937723,
      "grad_norm": 0.1770869642496109,
      "learning_rate": 2.115974347332461e-07,
      "loss": 0.0007,
      "step": 285750
    },
    {
      "epoch": 1.8327758768315092,
      "grad_norm": 0.12868542969226837,
      "learning_rate": 2.114363634396993e-07,
      "loss": 0.0013,
      "step": 285760
    },
    {
      "epoch": 1.8328400137252951,
      "grad_norm": 0.08010537177324295,
      "learning_rate": 2.112753521502775e-07,
      "loss": 0.0015,
      "step": 285770
    },
    {
      "epoch": 1.8329041506190813,
      "grad_norm": 0.17784012854099274,
      "learning_rate": 2.1111440086699854e-07,
      "loss": 0.0013,
      "step": 285780
    },
    {
      "epoch": 1.8329682875128674,
      "grad_norm": 0.21651434898376465,
      "learning_rate": 2.10953509591878e-07,
      "loss": 0.0016,
      "step": 285790
    },
    {
      "epoch": 1.8330324244066536,
      "grad_norm": 0.0007737832493148744,
      "learning_rate": 2.1079267832693317e-07,
      "loss": 0.0006,
      "step": 285800
    },
    {
      "epoch": 1.8330965613004397,
      "grad_norm": 0.0973321944475174,
      "learning_rate": 2.1063190707417858e-07,
      "loss": 0.0013,
      "step": 285810
    },
    {
      "epoch": 1.833160698194226,
      "grad_norm": 0.015331475995481014,
      "learning_rate": 2.1047119583562926e-07,
      "loss": 0.0023,
      "step": 285820
    },
    {
      "epoch": 1.8332248350880118,
      "grad_norm": 0.09351062774658203,
      "learning_rate": 2.103105446132986e-07,
      "loss": 0.0016,
      "step": 285830
    },
    {
      "epoch": 1.833288971981798,
      "grad_norm": 0.06600060313940048,
      "learning_rate": 2.1014995340919998e-07,
      "loss": 0.0008,
      "step": 285840
    },
    {
      "epoch": 1.833353108875584,
      "grad_norm": 0.06511164456605911,
      "learning_rate": 2.0998942222534624e-07,
      "loss": 0.0013,
      "step": 285850
    },
    {
      "epoch": 1.83341724576937,
      "grad_norm": 0.01580311357975006,
      "learning_rate": 2.0982895106374802e-07,
      "loss": 0.001,
      "step": 285860
    },
    {
      "epoch": 1.8334813826631562,
      "grad_norm": 0.09493190795183182,
      "learning_rate": 2.0966853992641644e-07,
      "loss": 0.0013,
      "step": 285870
    },
    {
      "epoch": 1.8335455195569423,
      "grad_norm": 0.09049346297979355,
      "learning_rate": 2.095081888153616e-07,
      "loss": 0.0012,
      "step": 285880
    },
    {
      "epoch": 1.8336096564507285,
      "grad_norm": 0.1104205846786499,
      "learning_rate": 2.0934789773259355e-07,
      "loss": 0.001,
      "step": 285890
    },
    {
      "epoch": 1.8336737933445146,
      "grad_norm": 0.010128783993422985,
      "learning_rate": 2.0918766668011959e-07,
      "loss": 0.0015,
      "step": 285900
    },
    {
      "epoch": 1.8337379302383008,
      "grad_norm": 0.03706973418593407,
      "learning_rate": 2.0902749565994751e-07,
      "loss": 0.0024,
      "step": 285910
    },
    {
      "epoch": 1.8338020671320867,
      "grad_norm": 0.04837115854024887,
      "learning_rate": 2.088673846740863e-07,
      "loss": 0.0007,
      "step": 285920
    },
    {
      "epoch": 1.8338662040258729,
      "grad_norm": 0.08827932924032211,
      "learning_rate": 2.0870733372454045e-07,
      "loss": 0.0012,
      "step": 285930
    },
    {
      "epoch": 1.8339303409196588,
      "grad_norm": 0.12149443477392197,
      "learning_rate": 2.0854734281331502e-07,
      "loss": 0.0014,
      "step": 285940
    },
    {
      "epoch": 1.833994477813445,
      "grad_norm": 0.11144936829805374,
      "learning_rate": 2.0838741194241675e-07,
      "loss": 0.0015,
      "step": 285950
    },
    {
      "epoch": 1.834058614707231,
      "grad_norm": 0.08110851794481277,
      "learning_rate": 2.0822754111384846e-07,
      "loss": 0.0037,
      "step": 285960
    },
    {
      "epoch": 1.8341227516010172,
      "grad_norm": 0.024030309170484543,
      "learning_rate": 2.0806773032961414e-07,
      "loss": 0.0011,
      "step": 285970
    },
    {
      "epoch": 1.8341868884948034,
      "grad_norm": 0.0025723432190716267,
      "learning_rate": 2.0790797959171494e-07,
      "loss": 0.0006,
      "step": 285980
    },
    {
      "epoch": 1.8342510253885895,
      "grad_norm": 0.08574619889259338,
      "learning_rate": 2.077482889021548e-07,
      "loss": 0.0018,
      "step": 285990
    },
    {
      "epoch": 1.8343151622823755,
      "grad_norm": 0.039204102009534836,
      "learning_rate": 2.0758865826293274e-07,
      "loss": 0.0016,
      "step": 286000
    },
    {
      "epoch": 1.8343792991761616,
      "grad_norm": 0.03519874066114426,
      "learning_rate": 2.0742908767605042e-07,
      "loss": 0.0009,
      "step": 286010
    },
    {
      "epoch": 1.8344434360699475,
      "grad_norm": 0.11675609648227692,
      "learning_rate": 2.0726957714350626e-07,
      "loss": 0.0004,
      "step": 286020
    },
    {
      "epoch": 1.8345075729637337,
      "grad_norm": 0.01122630387544632,
      "learning_rate": 2.071101266672998e-07,
      "loss": 0.0007,
      "step": 286030
    },
    {
      "epoch": 1.8345717098575198,
      "grad_norm": 0.10321499407291412,
      "learning_rate": 2.069507362494294e-07,
      "loss": 0.0006,
      "step": 286040
    },
    {
      "epoch": 1.834635846751306,
      "grad_norm": 0.055491261184215546,
      "learning_rate": 2.0679140589189128e-07,
      "loss": 0.0015,
      "step": 286050
    },
    {
      "epoch": 1.8346999836450921,
      "grad_norm": 0.08569338172674179,
      "learning_rate": 2.0663213559668215e-07,
      "loss": 0.0003,
      "step": 286060
    },
    {
      "epoch": 1.8347641205388783,
      "grad_norm": 0.17324191331863403,
      "learning_rate": 2.0647292536579877e-07,
      "loss": 0.0004,
      "step": 286070
    },
    {
      "epoch": 1.8348282574326644,
      "grad_norm": 0.10901259630918503,
      "learning_rate": 2.0631377520123563e-07,
      "loss": 0.0005,
      "step": 286080
    },
    {
      "epoch": 1.8348923943264503,
      "grad_norm": 0.027271650731563568,
      "learning_rate": 2.0615468510498616e-07,
      "loss": 0.0006,
      "step": 286090
    },
    {
      "epoch": 1.8349565312202365,
      "grad_norm": 0.16619277000427246,
      "learning_rate": 2.0599565507904539e-07,
      "loss": 0.0016,
      "step": 286100
    },
    {
      "epoch": 1.8350206681140224,
      "grad_norm": 0.03477492928504944,
      "learning_rate": 2.0583668512540512e-07,
      "loss": 0.0006,
      "step": 286110
    },
    {
      "epoch": 1.8350848050078086,
      "grad_norm": 0.04874105006456375,
      "learning_rate": 2.0567777524605704e-07,
      "loss": 0.0009,
      "step": 286120
    },
    {
      "epoch": 1.8351489419015947,
      "grad_norm": 0.026079542934894562,
      "learning_rate": 2.055189254429929e-07,
      "loss": 0.0009,
      "step": 286130
    },
    {
      "epoch": 1.8352130787953809,
      "grad_norm": 0.08594903349876404,
      "learning_rate": 2.0536013571820336e-07,
      "loss": 0.0011,
      "step": 286140
    },
    {
      "epoch": 1.835277215689167,
      "grad_norm": 0.1158260926604271,
      "learning_rate": 2.0520140607367845e-07,
      "loss": 0.0013,
      "step": 286150
    },
    {
      "epoch": 1.8353413525829532,
      "grad_norm": 0.13311445713043213,
      "learning_rate": 2.050427365114066e-07,
      "loss": 0.001,
      "step": 286160
    },
    {
      "epoch": 1.8354054894767393,
      "grad_norm": 0.04005707800388336,
      "learning_rate": 2.0488412703337512e-07,
      "loss": 0.0008,
      "step": 286170
    },
    {
      "epoch": 1.8354696263705252,
      "grad_norm": 0.1504872590303421,
      "learning_rate": 2.0472557764157352e-07,
      "loss": 0.0011,
      "step": 286180
    },
    {
      "epoch": 1.8355337632643114,
      "grad_norm": 0.05102236196398735,
      "learning_rate": 2.04567088337988e-07,
      "loss": 0.0009,
      "step": 286190
    },
    {
      "epoch": 1.8355979001580973,
      "grad_norm": 0.05543968081474304,
      "learning_rate": 2.0440865912460362e-07,
      "loss": 0.0017,
      "step": 286200
    },
    {
      "epoch": 1.8356620370518835,
      "grad_norm": 0.16379182040691376,
      "learning_rate": 2.0425029000340546e-07,
      "loss": 0.0012,
      "step": 286210
    },
    {
      "epoch": 1.8357261739456696,
      "grad_norm": 0.04222199320793152,
      "learning_rate": 2.0409198097637917e-07,
      "loss": 0.0007,
      "step": 286220
    },
    {
      "epoch": 1.8357903108394558,
      "grad_norm": 0.2061203569173813,
      "learning_rate": 2.0393373204550815e-07,
      "loss": 0.0007,
      "step": 286230
    },
    {
      "epoch": 1.835854447733242,
      "grad_norm": 0.07101015746593475,
      "learning_rate": 2.0377554321277414e-07,
      "loss": 0.0007,
      "step": 286240
    },
    {
      "epoch": 1.835918584627028,
      "grad_norm": 0.02212749794125557,
      "learning_rate": 2.036174144801617e-07,
      "loss": 0.0028,
      "step": 286250
    },
    {
      "epoch": 1.835982721520814,
      "grad_norm": 0.05497482046484947,
      "learning_rate": 2.0345934584965034e-07,
      "loss": 0.001,
      "step": 286260
    },
    {
      "epoch": 1.8360468584146001,
      "grad_norm": 0.014361917041242123,
      "learning_rate": 2.0330133732322177e-07,
      "loss": 0.0005,
      "step": 286270
    },
    {
      "epoch": 1.836110995308386,
      "grad_norm": 0.2536948025226593,
      "learning_rate": 2.03143388902855e-07,
      "loss": 0.001,
      "step": 286280
    },
    {
      "epoch": 1.8361751322021722,
      "grad_norm": 0.13963524997234344,
      "learning_rate": 2.0298550059053012e-07,
      "loss": 0.0022,
      "step": 286290
    },
    {
      "epoch": 1.8362392690959584,
      "grad_norm": 0.16756793856620789,
      "learning_rate": 2.0282767238822553e-07,
      "loss": 0.0012,
      "step": 286300
    },
    {
      "epoch": 1.8363034059897445,
      "grad_norm": 0.30188626050949097,
      "learning_rate": 2.0266990429791911e-07,
      "loss": 0.0015,
      "step": 286310
    },
    {
      "epoch": 1.8363675428835307,
      "grad_norm": 0.255096435546875,
      "learning_rate": 2.025121963215859e-07,
      "loss": 0.0011,
      "step": 286320
    },
    {
      "epoch": 1.8364316797773168,
      "grad_norm": 0.0976017639040947,
      "learning_rate": 2.0235454846120496e-07,
      "loss": 0.0015,
      "step": 286330
    },
    {
      "epoch": 1.836495816671103,
      "grad_norm": 0.007329453714191914,
      "learning_rate": 2.0219696071875016e-07,
      "loss": 0.0015,
      "step": 286340
    },
    {
      "epoch": 1.8365599535648889,
      "grad_norm": 0.08448395878076553,
      "learning_rate": 2.020394330961961e-07,
      "loss": 0.001,
      "step": 286350
    },
    {
      "epoch": 1.836624090458675,
      "grad_norm": 0.054196055978536606,
      "learning_rate": 2.0188196559551621e-07,
      "loss": 0.0025,
      "step": 286360
    },
    {
      "epoch": 1.836688227352461,
      "grad_norm": 0.00047856580931693316,
      "learning_rate": 2.0172455821868554e-07,
      "loss": 0.0009,
      "step": 286370
    },
    {
      "epoch": 1.836752364246247,
      "grad_norm": 0.02655697613954544,
      "learning_rate": 2.0156721096767528e-07,
      "loss": 0.0006,
      "step": 286380
    },
    {
      "epoch": 1.8368165011400333,
      "grad_norm": 0.008494804613292217,
      "learning_rate": 2.0140992384445668e-07,
      "loss": 0.0009,
      "step": 286390
    },
    {
      "epoch": 1.8368806380338194,
      "grad_norm": 0.06913774460554123,
      "learning_rate": 2.0125269685100203e-07,
      "loss": 0.001,
      "step": 286400
    },
    {
      "epoch": 1.8369447749276056,
      "grad_norm": 0.053152233362197876,
      "learning_rate": 2.0109552998928027e-07,
      "loss": 0.0006,
      "step": 286410
    },
    {
      "epoch": 1.8370089118213917,
      "grad_norm": 0.1499040573835373,
      "learning_rate": 2.0093842326126156e-07,
      "loss": 0.0016,
      "step": 286420
    },
    {
      "epoch": 1.8370730487151776,
      "grad_norm": 0.09256197512149811,
      "learning_rate": 2.0078137666891374e-07,
      "loss": 0.0011,
      "step": 286430
    },
    {
      "epoch": 1.8371371856089638,
      "grad_norm": 0.02435123547911644,
      "learning_rate": 2.006243902142052e-07,
      "loss": 0.0005,
      "step": 286440
    },
    {
      "epoch": 1.83720132250275,
      "grad_norm": 0.0038517331704497337,
      "learning_rate": 2.0046746389910388e-07,
      "loss": 0.0006,
      "step": 286450
    },
    {
      "epoch": 1.8372654593965358,
      "grad_norm": 0.10937990993261337,
      "learning_rate": 2.0031059772557482e-07,
      "loss": 0.0008,
      "step": 286460
    },
    {
      "epoch": 1.837329596290322,
      "grad_norm": 0.038557618856430054,
      "learning_rate": 2.0015379169558368e-07,
      "loss": 0.0026,
      "step": 286470
    },
    {
      "epoch": 1.8373937331841081,
      "grad_norm": 0.06308803707361221,
      "learning_rate": 1.9999704581109614e-07,
      "loss": 0.0012,
      "step": 286480
    },
    {
      "epoch": 1.8374578700778943,
      "grad_norm": 0.052062951028347015,
      "learning_rate": 1.9984036007407724e-07,
      "loss": 0.0007,
      "step": 286490
    },
    {
      "epoch": 1.8375220069716804,
      "grad_norm": 0.05477140098810196,
      "learning_rate": 1.9968373448648826e-07,
      "loss": 0.001,
      "step": 286500
    },
    {
      "epoch": 1.8375861438654666,
      "grad_norm": 0.08055808395147324,
      "learning_rate": 1.9952716905029313e-07,
      "loss": 0.0016,
      "step": 286510
    },
    {
      "epoch": 1.8376502807592525,
      "grad_norm": 0.2582389712333679,
      "learning_rate": 1.9937066376745363e-07,
      "loss": 0.001,
      "step": 286520
    },
    {
      "epoch": 1.8377144176530387,
      "grad_norm": 0.09943609684705734,
      "learning_rate": 1.992142186399304e-07,
      "loss": 0.0013,
      "step": 286530
    },
    {
      "epoch": 1.8377785545468246,
      "grad_norm": 0.002249883022159338,
      "learning_rate": 1.990578336696841e-07,
      "loss": 0.0005,
      "step": 286540
    },
    {
      "epoch": 1.8378426914406107,
      "grad_norm": 0.08084165304899216,
      "learning_rate": 1.989015088586743e-07,
      "loss": 0.0014,
      "step": 286550
    },
    {
      "epoch": 1.837906828334397,
      "grad_norm": 0.03168710693717003,
      "learning_rate": 1.9874524420886e-07,
      "loss": 0.001,
      "step": 286560
    },
    {
      "epoch": 1.837970965228183,
      "grad_norm": 0.050417460501194,
      "learning_rate": 1.9858903972219955e-07,
      "loss": 0.0007,
      "step": 286570
    },
    {
      "epoch": 1.8380351021219692,
      "grad_norm": 0.07067926228046417,
      "learning_rate": 1.9843289540064926e-07,
      "loss": 0.0006,
      "step": 286580
    },
    {
      "epoch": 1.8380992390157553,
      "grad_norm": 0.034939754754304886,
      "learning_rate": 1.9827681124616638e-07,
      "loss": 0.0008,
      "step": 286590
    },
    {
      "epoch": 1.8381633759095415,
      "grad_norm": 0.07707566022872925,
      "learning_rate": 1.9812078726070716e-07,
      "loss": 0.0008,
      "step": 286600
    },
    {
      "epoch": 1.8382275128033274,
      "grad_norm": 0.08685056120157242,
      "learning_rate": 1.979648234462267e-07,
      "loss": 0.0007,
      "step": 286610
    },
    {
      "epoch": 1.8382916496971136,
      "grad_norm": 0.044056981801986694,
      "learning_rate": 1.97808919804679e-07,
      "loss": 0.0008,
      "step": 286620
    },
    {
      "epoch": 1.8383557865908995,
      "grad_norm": 0.08716120570898056,
      "learning_rate": 1.9765307633801635e-07,
      "loss": 0.0012,
      "step": 286630
    },
    {
      "epoch": 1.8384199234846856,
      "grad_norm": 0.02711053565144539,
      "learning_rate": 1.9749729304819386e-07,
      "loss": 0.0021,
      "step": 286640
    },
    {
      "epoch": 1.8384840603784718,
      "grad_norm": 0.07572808116674423,
      "learning_rate": 1.973415699371628e-07,
      "loss": 0.0009,
      "step": 286650
    },
    {
      "epoch": 1.838548197272258,
      "grad_norm": 0.01647782139480114,
      "learning_rate": 1.9718590700687323e-07,
      "loss": 0.0018,
      "step": 286660
    },
    {
      "epoch": 1.838612334166044,
      "grad_norm": 0.06854109466075897,
      "learning_rate": 1.970303042592775e-07,
      "loss": 0.0005,
      "step": 286670
    },
    {
      "epoch": 1.8386764710598302,
      "grad_norm": 0.06197073310613632,
      "learning_rate": 1.9687476169632458e-07,
      "loss": 0.0004,
      "step": 286680
    },
    {
      "epoch": 1.8387406079536162,
      "grad_norm": 0.02185492031276226,
      "learning_rate": 1.9671927931996403e-07,
      "loss": 0.001,
      "step": 286690
    },
    {
      "epoch": 1.8388047448474023,
      "grad_norm": 0.05141449347138405,
      "learning_rate": 1.965638571321432e-07,
      "loss": 0.0019,
      "step": 286700
    },
    {
      "epoch": 1.8388688817411882,
      "grad_norm": 0.052674420177936554,
      "learning_rate": 1.9640849513481052e-07,
      "loss": 0.0011,
      "step": 286710
    },
    {
      "epoch": 1.8389330186349744,
      "grad_norm": 0.0528746098279953,
      "learning_rate": 1.9625319332991223e-07,
      "loss": 0.0018,
      "step": 286720
    },
    {
      "epoch": 1.8389971555287605,
      "grad_norm": 0.18851709365844727,
      "learning_rate": 1.9609795171939506e-07,
      "loss": 0.0019,
      "step": 286730
    },
    {
      "epoch": 1.8390612924225467,
      "grad_norm": 0.03728310018777847,
      "learning_rate": 1.9594277030520303e-07,
      "loss": 0.0011,
      "step": 286740
    },
    {
      "epoch": 1.8391254293163328,
      "grad_norm": 0.12406738102436066,
      "learning_rate": 1.9578764908928183e-07,
      "loss": 0.0014,
      "step": 286750
    },
    {
      "epoch": 1.839189566210119,
      "grad_norm": 0.09355689585208893,
      "learning_rate": 1.9563258807357543e-07,
      "loss": 0.001,
      "step": 286760
    },
    {
      "epoch": 1.8392537031039051,
      "grad_norm": 0.06668778508901596,
      "learning_rate": 1.9547758726002618e-07,
      "loss": 0.0035,
      "step": 286770
    },
    {
      "epoch": 1.839317839997691,
      "grad_norm": 0.05535772442817688,
      "learning_rate": 1.9532264665057588e-07,
      "loss": 0.003,
      "step": 286780
    },
    {
      "epoch": 1.8393819768914772,
      "grad_norm": 0.0019860744941979647,
      "learning_rate": 1.9516776624716738e-07,
      "loss": 0.0008,
      "step": 286790
    },
    {
      "epoch": 1.8394461137852631,
      "grad_norm": 0.07493186742067337,
      "learning_rate": 1.9501294605174026e-07,
      "loss": 0.0005,
      "step": 286800
    },
    {
      "epoch": 1.8395102506790493,
      "grad_norm": 0.0036910579074174166,
      "learning_rate": 1.9485818606623464e-07,
      "loss": 0.0006,
      "step": 286810
    },
    {
      "epoch": 1.8395743875728354,
      "grad_norm": 0.12099093198776245,
      "learning_rate": 1.9470348629259118e-07,
      "loss": 0.0007,
      "step": 286820
    },
    {
      "epoch": 1.8396385244666216,
      "grad_norm": 0.06820711493492126,
      "learning_rate": 1.9454884673274666e-07,
      "loss": 0.0007,
      "step": 286830
    },
    {
      "epoch": 1.8397026613604077,
      "grad_norm": 0.05245767533779144,
      "learning_rate": 1.9439426738864008e-07,
      "loss": 0.0033,
      "step": 286840
    },
    {
      "epoch": 1.8397667982541939,
      "grad_norm": 0.03281093016266823,
      "learning_rate": 1.942397482622066e-07,
      "loss": 0.0009,
      "step": 286850
    },
    {
      "epoch": 1.8398309351479798,
      "grad_norm": 0.006717719603329897,
      "learning_rate": 1.940852893553846e-07,
      "loss": 0.0007,
      "step": 286860
    },
    {
      "epoch": 1.839895072041766,
      "grad_norm": 0.07139433175325394,
      "learning_rate": 1.9393089067010872e-07,
      "loss": 0.0008,
      "step": 286870
    },
    {
      "epoch": 1.839959208935552,
      "grad_norm": 0.10973839461803436,
      "learning_rate": 1.9377655220831348e-07,
      "loss": 0.0008,
      "step": 286880
    },
    {
      "epoch": 1.840023345829338,
      "grad_norm": 0.027526378631591797,
      "learning_rate": 1.936222739719329e-07,
      "loss": 0.0011,
      "step": 286890
    },
    {
      "epoch": 1.8400874827231242,
      "grad_norm": 0.17372477054595947,
      "learning_rate": 1.9346805596290042e-07,
      "loss": 0.001,
      "step": 286900
    },
    {
      "epoch": 1.8401516196169103,
      "grad_norm": 0.04875880479812622,
      "learning_rate": 1.9331389818314838e-07,
      "loss": 0.0009,
      "step": 286910
    },
    {
      "epoch": 1.8402157565106965,
      "grad_norm": 0.11518640071153641,
      "learning_rate": 1.9315980063460805e-07,
      "loss": 0.0019,
      "step": 286920
    },
    {
      "epoch": 1.8402798934044826,
      "grad_norm": 0.2383868396282196,
      "learning_rate": 1.9300576331921116e-07,
      "loss": 0.0012,
      "step": 286930
    },
    {
      "epoch": 1.8403440302982688,
      "grad_norm": 0.03028559498488903,
      "learning_rate": 1.9285178623888788e-07,
      "loss": 0.0022,
      "step": 286940
    },
    {
      "epoch": 1.8404081671920547,
      "grad_norm": 0.1163666620850563,
      "learning_rate": 1.9269786939556722e-07,
      "loss": 0.0004,
      "step": 286950
    },
    {
      "epoch": 1.8404723040858408,
      "grad_norm": 0.019930845126509666,
      "learning_rate": 1.9254401279117763e-07,
      "loss": 0.001,
      "step": 286960
    },
    {
      "epoch": 1.8405364409796268,
      "grad_norm": 0.0351133793592453,
      "learning_rate": 1.9239021642764754e-07,
      "loss": 0.0007,
      "step": 286970
    },
    {
      "epoch": 1.840600577873413,
      "grad_norm": 0.08575202524662018,
      "learning_rate": 1.9223648030690433e-07,
      "loss": 0.001,
      "step": 286980
    },
    {
      "epoch": 1.840664714767199,
      "grad_norm": 0.01145631168037653,
      "learning_rate": 1.9208280443087369e-07,
      "loss": 0.0008,
      "step": 286990
    },
    {
      "epoch": 1.8407288516609852,
      "grad_norm": 0.03613721579313278,
      "learning_rate": 1.9192918880148125e-07,
      "loss": 0.0007,
      "step": 287000
    },
    {
      "epoch": 1.8407929885547714,
      "grad_norm": 0.03389430791139603,
      "learning_rate": 1.9177563342065332e-07,
      "loss": 0.0021,
      "step": 287010
    },
    {
      "epoch": 1.8408571254485575,
      "grad_norm": 0.0285571850836277,
      "learning_rate": 1.9162213829031273e-07,
      "loss": 0.001,
      "step": 287020
    },
    {
      "epoch": 1.8409212623423437,
      "grad_norm": 0.0828038901090622,
      "learning_rate": 1.9146870341238356e-07,
      "loss": 0.0008,
      "step": 287030
    },
    {
      "epoch": 1.8409853992361296,
      "grad_norm": 0.1025981530547142,
      "learning_rate": 1.91315328788787e-07,
      "loss": 0.0009,
      "step": 287040
    },
    {
      "epoch": 1.8410495361299157,
      "grad_norm": 0.043404772877693176,
      "learning_rate": 1.9116201442144712e-07,
      "loss": 0.001,
      "step": 287050
    },
    {
      "epoch": 1.8411136730237017,
      "grad_norm": 0.09349111467599869,
      "learning_rate": 1.91008760312284e-07,
      "loss": 0.0007,
      "step": 287060
    },
    {
      "epoch": 1.8411778099174878,
      "grad_norm": 0.15410931408405304,
      "learning_rate": 1.9085556646321724e-07,
      "loss": 0.0018,
      "step": 287070
    },
    {
      "epoch": 1.841241946811274,
      "grad_norm": 0.005862515885382891,
      "learning_rate": 1.9070243287616753e-07,
      "loss": 0.0013,
      "step": 287080
    },
    {
      "epoch": 1.84130608370506,
      "grad_norm": 0.06634867936372757,
      "learning_rate": 1.905493595530544e-07,
      "loss": 0.0008,
      "step": 287090
    },
    {
      "epoch": 1.8413702205988463,
      "grad_norm": 0.035512588918209076,
      "learning_rate": 1.9039634649579418e-07,
      "loss": 0.0006,
      "step": 287100
    },
    {
      "epoch": 1.8414343574926324,
      "grad_norm": 0.057515282183885574,
      "learning_rate": 1.902433937063053e-07,
      "loss": 0.0008,
      "step": 287110
    },
    {
      "epoch": 1.8414984943864183,
      "grad_norm": 0.11442890763282776,
      "learning_rate": 1.9009050118650397e-07,
      "loss": 0.0008,
      "step": 287120
    },
    {
      "epoch": 1.8415626312802045,
      "grad_norm": 0.02416716329753399,
      "learning_rate": 1.89937668938307e-07,
      "loss": 0.001,
      "step": 287130
    },
    {
      "epoch": 1.8416267681739904,
      "grad_norm": 0.20283135771751404,
      "learning_rate": 1.8978489696362846e-07,
      "loss": 0.0056,
      "step": 287140
    },
    {
      "epoch": 1.8416909050677766,
      "grad_norm": 0.011368321254849434,
      "learning_rate": 1.8963218526438233e-07,
      "loss": 0.001,
      "step": 287150
    },
    {
      "epoch": 1.8417550419615627,
      "grad_norm": 0.07025032490491867,
      "learning_rate": 1.894795338424832e-07,
      "loss": 0.0003,
      "step": 287160
    },
    {
      "epoch": 1.8418191788553488,
      "grad_norm": 0.04045814648270607,
      "learning_rate": 1.8932694269984397e-07,
      "loss": 0.0015,
      "step": 287170
    },
    {
      "epoch": 1.841883315749135,
      "grad_norm": 0.017666665837168694,
      "learning_rate": 1.8917441183837593e-07,
      "loss": 0.0019,
      "step": 287180
    },
    {
      "epoch": 1.8419474526429211,
      "grad_norm": 0.0366133488714695,
      "learning_rate": 1.8902194125999029e-07,
      "loss": 0.0008,
      "step": 287190
    },
    {
      "epoch": 1.8420115895367073,
      "grad_norm": 0.055339265614748,
      "learning_rate": 1.8886953096659833e-07,
      "loss": 0.0011,
      "step": 287200
    },
    {
      "epoch": 1.8420757264304932,
      "grad_norm": 0.13355812430381775,
      "learning_rate": 1.8871718096011015e-07,
      "loss": 0.0019,
      "step": 287210
    },
    {
      "epoch": 1.8421398633242794,
      "grad_norm": 0.0653081163764,
      "learning_rate": 1.8856489124243372e-07,
      "loss": 0.0006,
      "step": 287220
    },
    {
      "epoch": 1.8422040002180653,
      "grad_norm": 0.015454231761395931,
      "learning_rate": 1.8841266181547803e-07,
      "loss": 0.001,
      "step": 287230
    },
    {
      "epoch": 1.8422681371118514,
      "grad_norm": 0.029873233288526535,
      "learning_rate": 1.88260492681151e-07,
      "loss": 0.0005,
      "step": 287240
    },
    {
      "epoch": 1.8423322740056376,
      "grad_norm": 0.058013807982206345,
      "learning_rate": 1.8810838384135777e-07,
      "loss": 0.0017,
      "step": 287250
    },
    {
      "epoch": 1.8423964108994237,
      "grad_norm": 0.22602404654026031,
      "learning_rate": 1.8795633529800627e-07,
      "loss": 0.0015,
      "step": 287260
    },
    {
      "epoch": 1.84246054779321,
      "grad_norm": 0.03169405832886696,
      "learning_rate": 1.8780434705300054e-07,
      "loss": 0.0013,
      "step": 287270
    },
    {
      "epoch": 1.842524684686996,
      "grad_norm": 0.06978941708803177,
      "learning_rate": 1.8765241910824573e-07,
      "loss": 0.0005,
      "step": 287280
    },
    {
      "epoch": 1.842588821580782,
      "grad_norm": 0.12041551619768143,
      "learning_rate": 1.8750055146564582e-07,
      "loss": 0.0012,
      "step": 287290
    },
    {
      "epoch": 1.8426529584745681,
      "grad_norm": 0.15691913664340973,
      "learning_rate": 1.873487441271038e-07,
      "loss": 0.0017,
      "step": 287300
    },
    {
      "epoch": 1.8427170953683543,
      "grad_norm": 0.06942026317119598,
      "learning_rate": 1.8719699709452032e-07,
      "loss": 0.0009,
      "step": 287310
    },
    {
      "epoch": 1.8427812322621402,
      "grad_norm": 0.05978382006287575,
      "learning_rate": 1.870453103697989e-07,
      "loss": 0.0012,
      "step": 287320
    },
    {
      "epoch": 1.8428453691559263,
      "grad_norm": 0.08361706137657166,
      "learning_rate": 1.8689368395483964e-07,
      "loss": 0.0006,
      "step": 287330
    },
    {
      "epoch": 1.8429095060497125,
      "grad_norm": 0.1487368941307068,
      "learning_rate": 1.8674211785154162e-07,
      "loss": 0.0015,
      "step": 287340
    },
    {
      "epoch": 1.8429736429434986,
      "grad_norm": 0.02252854034304619,
      "learning_rate": 1.865906120618055e-07,
      "loss": 0.001,
      "step": 287350
    },
    {
      "epoch": 1.8430377798372848,
      "grad_norm": 0.00402287021279335,
      "learning_rate": 1.8643916658752925e-07,
      "loss": 0.0007,
      "step": 287360
    },
    {
      "epoch": 1.843101916731071,
      "grad_norm": 0.05641627311706543,
      "learning_rate": 1.8628778143061078e-07,
      "loss": 0.0015,
      "step": 287370
    },
    {
      "epoch": 1.8431660536248569,
      "grad_norm": 0.004220783710479736,
      "learning_rate": 1.8613645659294522e-07,
      "loss": 0.0003,
      "step": 287380
    },
    {
      "epoch": 1.843230190518643,
      "grad_norm": 0.08366972208023071,
      "learning_rate": 1.859851920764316e-07,
      "loss": 0.0012,
      "step": 287390
    },
    {
      "epoch": 1.843294327412429,
      "grad_norm": 0.007999827153980732,
      "learning_rate": 1.8583398788296403e-07,
      "loss": 0.0008,
      "step": 287400
    },
    {
      "epoch": 1.843358464306215,
      "grad_norm": 0.08857898414134979,
      "learning_rate": 1.8568284401443648e-07,
      "loss": 0.0013,
      "step": 287410
    },
    {
      "epoch": 1.8434226012000012,
      "grad_norm": 0.041703127324581146,
      "learning_rate": 1.8553176047274357e-07,
      "loss": 0.0009,
      "step": 287420
    },
    {
      "epoch": 1.8434867380937874,
      "grad_norm": 0.07411111146211624,
      "learning_rate": 1.8538073725977935e-07,
      "loss": 0.0006,
      "step": 287430
    },
    {
      "epoch": 1.8435508749875735,
      "grad_norm": 0.13202106952667236,
      "learning_rate": 1.8522977437743505e-07,
      "loss": 0.0013,
      "step": 287440
    },
    {
      "epoch": 1.8436150118813597,
      "grad_norm": 0.03775406628847122,
      "learning_rate": 1.850788718276031e-07,
      "loss": 0.0009,
      "step": 287450
    },
    {
      "epoch": 1.8436791487751458,
      "grad_norm": 0.07211952656507492,
      "learning_rate": 1.8492802961217305e-07,
      "loss": 0.0007,
      "step": 287460
    },
    {
      "epoch": 1.8437432856689318,
      "grad_norm": 0.002661585807800293,
      "learning_rate": 1.8477724773303728e-07,
      "loss": 0.0013,
      "step": 287470
    },
    {
      "epoch": 1.843807422562718,
      "grad_norm": 0.22028471529483795,
      "learning_rate": 1.846265261920832e-07,
      "loss": 0.0016,
      "step": 287480
    },
    {
      "epoch": 1.8438715594565038,
      "grad_norm": 0.052171215415000916,
      "learning_rate": 1.8447586499120042e-07,
      "loss": 0.0012,
      "step": 287490
    },
    {
      "epoch": 1.84393569635029,
      "grad_norm": 0.1524595320224762,
      "learning_rate": 1.843252641322768e-07,
      "loss": 0.001,
      "step": 287500
    },
    {
      "epoch": 1.8439998332440761,
      "grad_norm": 0.018931686878204346,
      "learning_rate": 1.8417472361719923e-07,
      "loss": 0.001,
      "step": 287510
    },
    {
      "epoch": 1.8440639701378623,
      "grad_norm": 0.005724246148020029,
      "learning_rate": 1.8402424344785452e-07,
      "loss": 0.0014,
      "step": 287520
    },
    {
      "epoch": 1.8441281070316484,
      "grad_norm": 0.05640290305018425,
      "learning_rate": 1.8387382362612727e-07,
      "loss": 0.0011,
      "step": 287530
    },
    {
      "epoch": 1.8441922439254346,
      "grad_norm": 0.19624315202236176,
      "learning_rate": 1.8372346415390374e-07,
      "loss": 0.0009,
      "step": 287540
    },
    {
      "epoch": 1.8442563808192205,
      "grad_norm": 0.0690305233001709,
      "learning_rate": 1.8357316503306688e-07,
      "loss": 0.0006,
      "step": 287550
    },
    {
      "epoch": 1.8443205177130066,
      "grad_norm": 0.04752447083592415,
      "learning_rate": 1.8342292626550074e-07,
      "loss": 0.0006,
      "step": 287560
    },
    {
      "epoch": 1.8443846546067926,
      "grad_norm": 0.061734091490507126,
      "learning_rate": 1.8327274785308713e-07,
      "loss": 0.0013,
      "step": 287570
    },
    {
      "epoch": 1.8444487915005787,
      "grad_norm": 0.03917500004172325,
      "learning_rate": 1.8312262979770955e-07,
      "loss": 0.0017,
      "step": 287580
    },
    {
      "epoch": 1.8445129283943649,
      "grad_norm": 0.10373364388942719,
      "learning_rate": 1.8297257210124708e-07,
      "loss": 0.0018,
      "step": 287590
    },
    {
      "epoch": 1.844577065288151,
      "grad_norm": 0.05342405289411545,
      "learning_rate": 1.8282257476558153e-07,
      "loss": 0.0013,
      "step": 287600
    },
    {
      "epoch": 1.8446412021819372,
      "grad_norm": 0.03532358258962631,
      "learning_rate": 1.8267263779259138e-07,
      "loss": 0.0013,
      "step": 287610
    },
    {
      "epoch": 1.8447053390757233,
      "grad_norm": 0.04092554748058319,
      "learning_rate": 1.8252276118415624e-07,
      "loss": 0.0009,
      "step": 287620
    },
    {
      "epoch": 1.8447694759695095,
      "grad_norm": 0.037707142531871796,
      "learning_rate": 1.8237294494215408e-07,
      "loss": 0.0013,
      "step": 287630
    },
    {
      "epoch": 1.8448336128632954,
      "grad_norm": 0.049918580800294876,
      "learning_rate": 1.822231890684617e-07,
      "loss": 0.0013,
      "step": 287640
    },
    {
      "epoch": 1.8448977497570815,
      "grad_norm": 0.02753879502415657,
      "learning_rate": 1.8207349356495595e-07,
      "loss": 0.0005,
      "step": 287650
    },
    {
      "epoch": 1.8449618866508675,
      "grad_norm": 0.22618646919727325,
      "learning_rate": 1.8192385843351313e-07,
      "loss": 0.0006,
      "step": 287660
    },
    {
      "epoch": 1.8450260235446536,
      "grad_norm": 0.08298692107200623,
      "learning_rate": 1.8177428367600725e-07,
      "loss": 0.0008,
      "step": 287670
    },
    {
      "epoch": 1.8450901604384398,
      "grad_norm": 0.17711129784584045,
      "learning_rate": 1.8162476929431295e-07,
      "loss": 0.0018,
      "step": 287680
    },
    {
      "epoch": 1.845154297332226,
      "grad_norm": 0.014852141961455345,
      "learning_rate": 1.8147531529030483e-07,
      "loss": 0.0015,
      "step": 287690
    },
    {
      "epoch": 1.845218434226012,
      "grad_norm": 0.07084935158491135,
      "learning_rate": 1.8132592166585416e-07,
      "loss": 0.001,
      "step": 287700
    },
    {
      "epoch": 1.8452825711197982,
      "grad_norm": 0.03970756009221077,
      "learning_rate": 1.8117658842283392e-07,
      "loss": 0.0012,
      "step": 287710
    },
    {
      "epoch": 1.8453467080135844,
      "grad_norm": 0.14913849532604218,
      "learning_rate": 1.8102731556311425e-07,
      "loss": 0.0016,
      "step": 287720
    },
    {
      "epoch": 1.8454108449073703,
      "grad_norm": 0.05938960239291191,
      "learning_rate": 1.8087810308856703e-07,
      "loss": 0.0014,
      "step": 287730
    },
    {
      "epoch": 1.8454749818011564,
      "grad_norm": 0.03920167312026024,
      "learning_rate": 1.8072895100106124e-07,
      "loss": 0.0011,
      "step": 287740
    },
    {
      "epoch": 1.8455391186949424,
      "grad_norm": 0.0677153617143631,
      "learning_rate": 1.8057985930246603e-07,
      "loss": 0.0018,
      "step": 287750
    },
    {
      "epoch": 1.8456032555887285,
      "grad_norm": 0.362377405166626,
      "learning_rate": 1.804308279946493e-07,
      "loss": 0.0008,
      "step": 287760
    },
    {
      "epoch": 1.8456673924825147,
      "grad_norm": 0.044777531176805496,
      "learning_rate": 1.80281857079479e-07,
      "loss": 0.0005,
      "step": 287770
    },
    {
      "epoch": 1.8457315293763008,
      "grad_norm": 0.03310674801468849,
      "learning_rate": 1.8013294655882142e-07,
      "loss": 0.0006,
      "step": 287780
    },
    {
      "epoch": 1.845795666270087,
      "grad_norm": 0.010308466851711273,
      "learning_rate": 1.7998409643454228e-07,
      "loss": 0.0006,
      "step": 287790
    },
    {
      "epoch": 1.845859803163873,
      "grad_norm": 0.07554610818624496,
      "learning_rate": 1.7983530670850846e-07,
      "loss": 0.0027,
      "step": 287800
    },
    {
      "epoch": 1.845923940057659,
      "grad_norm": 0.0661851018667221,
      "learning_rate": 1.7968657738258232e-07,
      "loss": 0.0011,
      "step": 287810
    },
    {
      "epoch": 1.8459880769514452,
      "grad_norm": 0.006231280043721199,
      "learning_rate": 1.7953790845862907e-07,
      "loss": 0.0008,
      "step": 287820
    },
    {
      "epoch": 1.846052213845231,
      "grad_norm": 0.23022891581058502,
      "learning_rate": 1.7938929993850996e-07,
      "loss": 0.0013,
      "step": 287830
    },
    {
      "epoch": 1.8461163507390173,
      "grad_norm": 0.020045816898345947,
      "learning_rate": 1.792407518240885e-07,
      "loss": 0.0006,
      "step": 287840
    },
    {
      "epoch": 1.8461804876328034,
      "grad_norm": 0.1408000886440277,
      "learning_rate": 1.7909226411722603e-07,
      "loss": 0.0022,
      "step": 287850
    },
    {
      "epoch": 1.8462446245265896,
      "grad_norm": 0.007024731952697039,
      "learning_rate": 1.7894383681978321e-07,
      "loss": 0.0011,
      "step": 287860
    },
    {
      "epoch": 1.8463087614203757,
      "grad_norm": 0.045289646834135056,
      "learning_rate": 1.7879546993361918e-07,
      "loss": 0.0007,
      "step": 287870
    },
    {
      "epoch": 1.8463728983141618,
      "grad_norm": 0.03839179128408432,
      "learning_rate": 1.7864716346059352e-07,
      "loss": 0.0008,
      "step": 287880
    },
    {
      "epoch": 1.846437035207948,
      "grad_norm": 0.02492174133658409,
      "learning_rate": 1.784989174025653e-07,
      "loss": 0.001,
      "step": 287890
    },
    {
      "epoch": 1.846501172101734,
      "grad_norm": 0.0650896206498146,
      "learning_rate": 1.7835073176139083e-07,
      "loss": 0.0015,
      "step": 287900
    },
    {
      "epoch": 1.84656530899552,
      "grad_norm": 0.10479025542736053,
      "learning_rate": 1.7820260653892752e-07,
      "loss": 0.0017,
      "step": 287910
    },
    {
      "epoch": 1.846629445889306,
      "grad_norm": 0.2658521831035614,
      "learning_rate": 1.7805454173703218e-07,
      "loss": 0.001,
      "step": 287920
    },
    {
      "epoch": 1.8466935827830921,
      "grad_norm": 0.12911981344223022,
      "learning_rate": 1.7790653735756002e-07,
      "loss": 0.0011,
      "step": 287930
    },
    {
      "epoch": 1.8467577196768783,
      "grad_norm": 0.11474470794200897,
      "learning_rate": 1.77758593402364e-07,
      "loss": 0.0005,
      "step": 287940
    },
    {
      "epoch": 1.8468218565706644,
      "grad_norm": 0.051907800137996674,
      "learning_rate": 1.7761070987329988e-07,
      "loss": 0.0005,
      "step": 287950
    },
    {
      "epoch": 1.8468859934644506,
      "grad_norm": 0.06339803338050842,
      "learning_rate": 1.7746288677222058e-07,
      "loss": 0.0017,
      "step": 287960
    },
    {
      "epoch": 1.8469501303582367,
      "grad_norm": 0.05739164352416992,
      "learning_rate": 1.773151241009774e-07,
      "loss": 0.0019,
      "step": 287970
    },
    {
      "epoch": 1.8470142672520227,
      "grad_norm": 0.048748426139354706,
      "learning_rate": 1.7716742186142222e-07,
      "loss": 0.001,
      "step": 287980
    },
    {
      "epoch": 1.8470784041458088,
      "grad_norm": 0.03448588401079178,
      "learning_rate": 1.7701978005540633e-07,
      "loss": 0.0013,
      "step": 287990
    },
    {
      "epoch": 1.847142541039595,
      "grad_norm": 0.03690283000469208,
      "learning_rate": 1.768721986847799e-07,
      "loss": 0.0012,
      "step": 288000
    },
    {
      "epoch": 1.847206677933381,
      "grad_norm": 0.03804676979780197,
      "learning_rate": 1.7672467775139145e-07,
      "loss": 0.001,
      "step": 288010
    },
    {
      "epoch": 1.847270814827167,
      "grad_norm": 0.06943704932928085,
      "learning_rate": 1.7657721725708953e-07,
      "loss": 0.0021,
      "step": 288020
    },
    {
      "epoch": 1.8473349517209532,
      "grad_norm": 0.08496490865945816,
      "learning_rate": 1.7642981720372264e-07,
      "loss": 0.0012,
      "step": 288030
    },
    {
      "epoch": 1.8473990886147393,
      "grad_norm": 0.03908595070242882,
      "learning_rate": 1.7628247759313765e-07,
      "loss": 0.0007,
      "step": 288040
    },
    {
      "epoch": 1.8474632255085255,
      "grad_norm": 0.011440551839768887,
      "learning_rate": 1.7613519842718086e-07,
      "loss": 0.0015,
      "step": 288050
    },
    {
      "epoch": 1.8475273624023116,
      "grad_norm": 0.10329517722129822,
      "learning_rate": 1.7598797970769688e-07,
      "loss": 0.0015,
      "step": 288060
    },
    {
      "epoch": 1.8475914992960976,
      "grad_norm": 0.030152389779686928,
      "learning_rate": 1.7584082143653204e-07,
      "loss": 0.0008,
      "step": 288070
    },
    {
      "epoch": 1.8476556361898837,
      "grad_norm": 0.05683228000998497,
      "learning_rate": 1.7569372361552927e-07,
      "loss": 0.0009,
      "step": 288080
    },
    {
      "epoch": 1.8477197730836696,
      "grad_norm": 0.15313345193862915,
      "learning_rate": 1.755466862465316e-07,
      "loss": 0.0008,
      "step": 288090
    },
    {
      "epoch": 1.8477839099774558,
      "grad_norm": 0.10386233776807785,
      "learning_rate": 1.7539970933138195e-07,
      "loss": 0.0015,
      "step": 288100
    },
    {
      "epoch": 1.847848046871242,
      "grad_norm": 0.011520123109221458,
      "learning_rate": 1.7525279287192277e-07,
      "loss": 0.0005,
      "step": 288110
    },
    {
      "epoch": 1.847912183765028,
      "grad_norm": 0.08517508208751678,
      "learning_rate": 1.7510593686999367e-07,
      "loss": 0.0024,
      "step": 288120
    },
    {
      "epoch": 1.8479763206588142,
      "grad_norm": 0.011454598978161812,
      "learning_rate": 1.7495914132743597e-07,
      "loss": 0.0011,
      "step": 288130
    },
    {
      "epoch": 1.8480404575526004,
      "grad_norm": 0.1312156468629837,
      "learning_rate": 1.7481240624608764e-07,
      "loss": 0.001,
      "step": 288140
    },
    {
      "epoch": 1.8481045944463865,
      "grad_norm": 0.036829717457294464,
      "learning_rate": 1.7466573162778944e-07,
      "loss": 0.0005,
      "step": 288150
    },
    {
      "epoch": 1.8481687313401725,
      "grad_norm": 0.07705144584178925,
      "learning_rate": 1.7451911747437766e-07,
      "loss": 0.0009,
      "step": 288160
    },
    {
      "epoch": 1.8482328682339586,
      "grad_norm": 0.06838127970695496,
      "learning_rate": 1.743725637876903e-07,
      "loss": 0.0009,
      "step": 288170
    },
    {
      "epoch": 1.8482970051277445,
      "grad_norm": 0.029356516897678375,
      "learning_rate": 1.742260705695631e-07,
      "loss": 0.0013,
      "step": 288180
    },
    {
      "epoch": 1.8483611420215307,
      "grad_norm": 0.08610998094081879,
      "learning_rate": 1.740796378218329e-07,
      "loss": 0.0019,
      "step": 288190
    },
    {
      "epoch": 1.8484252789153168,
      "grad_norm": 0.07729338854551315,
      "learning_rate": 1.7393326554633382e-07,
      "loss": 0.0009,
      "step": 288200
    },
    {
      "epoch": 1.848489415809103,
      "grad_norm": 0.13719235360622406,
      "learning_rate": 1.7378695374489883e-07,
      "loss": 0.0006,
      "step": 288210
    },
    {
      "epoch": 1.8485535527028891,
      "grad_norm": 0.06897895038127899,
      "learning_rate": 1.736407024193637e-07,
      "loss": 0.001,
      "step": 288220
    },
    {
      "epoch": 1.8486176895966753,
      "grad_norm": 0.06548663973808289,
      "learning_rate": 1.7349451157156026e-07,
      "loss": 0.0004,
      "step": 288230
    },
    {
      "epoch": 1.8486818264904612,
      "grad_norm": 0.10674738883972168,
      "learning_rate": 1.733483812033193e-07,
      "loss": 0.001,
      "step": 288240
    },
    {
      "epoch": 1.8487459633842473,
      "grad_norm": 0.03931476175785065,
      "learning_rate": 1.732023113164727e-07,
      "loss": 0.0008,
      "step": 288250
    },
    {
      "epoch": 1.8488101002780333,
      "grad_norm": 0.06040503457188606,
      "learning_rate": 1.730563019128506e-07,
      "loss": 0.0007,
      "step": 288260
    },
    {
      "epoch": 1.8488742371718194,
      "grad_norm": 0.0705578550696373,
      "learning_rate": 1.7291035299428328e-07,
      "loss": 0.0008,
      "step": 288270
    },
    {
      "epoch": 1.8489383740656056,
      "grad_norm": 0.08032902330160141,
      "learning_rate": 1.7276446456259922e-07,
      "loss": 0.001,
      "step": 288280
    },
    {
      "epoch": 1.8490025109593917,
      "grad_norm": 0.03601422905921936,
      "learning_rate": 1.7261863661962585e-07,
      "loss": 0.0006,
      "step": 288290
    },
    {
      "epoch": 1.8490666478531779,
      "grad_norm": 0.030967695638537407,
      "learning_rate": 1.7247286916719174e-07,
      "loss": 0.001,
      "step": 288300
    },
    {
      "epoch": 1.849130784746964,
      "grad_norm": 0.051623016595840454,
      "learning_rate": 1.723271622071221e-07,
      "loss": 0.0009,
      "step": 288310
    },
    {
      "epoch": 1.8491949216407502,
      "grad_norm": 0.09495478868484497,
      "learning_rate": 1.7218151574124376e-07,
      "loss": 0.0006,
      "step": 288320
    },
    {
      "epoch": 1.849259058534536,
      "grad_norm": 0.03560183197259903,
      "learning_rate": 1.7203592977138083e-07,
      "loss": 0.0009,
      "step": 288330
    },
    {
      "epoch": 1.8493231954283222,
      "grad_norm": 0.025246890261769295,
      "learning_rate": 1.718904042993591e-07,
      "loss": 0.0007,
      "step": 288340
    },
    {
      "epoch": 1.8493873323221082,
      "grad_norm": 0.0935717299580574,
      "learning_rate": 1.7174493932700098e-07,
      "loss": 0.0013,
      "step": 288350
    },
    {
      "epoch": 1.8494514692158943,
      "grad_norm": 0.06764940172433853,
      "learning_rate": 1.715995348561289e-07,
      "loss": 0.0012,
      "step": 288360
    },
    {
      "epoch": 1.8495156061096805,
      "grad_norm": 0.16031967103481293,
      "learning_rate": 1.7145419088856585e-07,
      "loss": 0.0006,
      "step": 288370
    },
    {
      "epoch": 1.8495797430034666,
      "grad_norm": 0.042350515723228455,
      "learning_rate": 1.713089074261326e-07,
      "loss": 0.0008,
      "step": 288380
    },
    {
      "epoch": 1.8496438798972528,
      "grad_norm": 0.18568123877048492,
      "learning_rate": 1.7116368447064991e-07,
      "loss": 0.0009,
      "step": 288390
    },
    {
      "epoch": 1.849708016791039,
      "grad_norm": 0.15407635271549225,
      "learning_rate": 1.7101852202393688e-07,
      "loss": 0.0007,
      "step": 288400
    },
    {
      "epoch": 1.8497721536848248,
      "grad_norm": 0.06498847901821136,
      "learning_rate": 1.7087342008781372e-07,
      "loss": 0.0014,
      "step": 288410
    },
    {
      "epoch": 1.849836290578611,
      "grad_norm": 0.11050275713205338,
      "learning_rate": 1.7072837866409785e-07,
      "loss": 0.001,
      "step": 288420
    },
    {
      "epoch": 1.8499004274723971,
      "grad_norm": 0.07677308470010757,
      "learning_rate": 1.7058339775460675e-07,
      "loss": 0.0017,
      "step": 288430
    },
    {
      "epoch": 1.849964564366183,
      "grad_norm": 0.07862634211778641,
      "learning_rate": 1.704384773611567e-07,
      "loss": 0.001,
      "step": 288440
    },
    {
      "epoch": 1.8500287012599692,
      "grad_norm": 0.024905717000365257,
      "learning_rate": 1.7029361748556462e-07,
      "loss": 0.0027,
      "step": 288450
    },
    {
      "epoch": 1.8500928381537554,
      "grad_norm": 0.39429962635040283,
      "learning_rate": 1.7014881812964511e-07,
      "loss": 0.0046,
      "step": 288460
    },
    {
      "epoch": 1.8501569750475415,
      "grad_norm": 0.12610678374767303,
      "learning_rate": 1.7000407929521289e-07,
      "loss": 0.0012,
      "step": 288470
    },
    {
      "epoch": 1.8502211119413277,
      "grad_norm": 0.1045580506324768,
      "learning_rate": 1.6985940098408093e-07,
      "loss": 0.0014,
      "step": 288480
    },
    {
      "epoch": 1.8502852488351138,
      "grad_norm": 0.0926932767033577,
      "learning_rate": 1.6971478319806334e-07,
      "loss": 0.0017,
      "step": 288490
    },
    {
      "epoch": 1.8503493857288997,
      "grad_norm": 0.02598356083035469,
      "learning_rate": 1.69570225938972e-07,
      "loss": 0.0013,
      "step": 288500
    },
    {
      "epoch": 1.8504135226226859,
      "grad_norm": 0.1517276167869568,
      "learning_rate": 1.6942572920861712e-07,
      "loss": 0.0012,
      "step": 288510
    },
    {
      "epoch": 1.8504776595164718,
      "grad_norm": 0.18101486563682556,
      "learning_rate": 1.6928129300881057e-07,
      "loss": 0.002,
      "step": 288520
    },
    {
      "epoch": 1.850541796410258,
      "grad_norm": 0.11176485568284988,
      "learning_rate": 1.6913691734136206e-07,
      "loss": 0.0014,
      "step": 288530
    },
    {
      "epoch": 1.850605933304044,
      "grad_norm": 0.21926063299179077,
      "learning_rate": 1.6899260220808068e-07,
      "loss": 0.0011,
      "step": 288540
    },
    {
      "epoch": 1.8506700701978303,
      "grad_norm": 0.01746945083141327,
      "learning_rate": 1.6884834761077385e-07,
      "loss": 0.0007,
      "step": 288550
    },
    {
      "epoch": 1.8507342070916164,
      "grad_norm": 0.010603145696222782,
      "learning_rate": 1.6870415355125068e-07,
      "loss": 0.0009,
      "step": 288560
    },
    {
      "epoch": 1.8507983439854025,
      "grad_norm": 0.01538256648927927,
      "learning_rate": 1.68560020031317e-07,
      "loss": 0.0017,
      "step": 288570
    },
    {
      "epoch": 1.8508624808791887,
      "grad_norm": 0.1104760617017746,
      "learning_rate": 1.6841594705277964e-07,
      "loss": 0.0006,
      "step": 288580
    },
    {
      "epoch": 1.8509266177729746,
      "grad_norm": 0.002728406572714448,
      "learning_rate": 1.6827193461744275e-07,
      "loss": 0.0013,
      "step": 288590
    },
    {
      "epoch": 1.8509907546667608,
      "grad_norm": 0.14359810948371887,
      "learning_rate": 1.681279827271126e-07,
      "loss": 0.0011,
      "step": 288600
    },
    {
      "epoch": 1.8510548915605467,
      "grad_norm": 0.13374969363212585,
      "learning_rate": 1.6798409138359172e-07,
      "loss": 0.0008,
      "step": 288610
    },
    {
      "epoch": 1.8511190284543328,
      "grad_norm": 0.09471452981233597,
      "learning_rate": 1.6784026058868418e-07,
      "loss": 0.0005,
      "step": 288620
    },
    {
      "epoch": 1.851183165348119,
      "grad_norm": 0.0014831902226433158,
      "learning_rate": 1.6769649034419078e-07,
      "loss": 0.0012,
      "step": 288630
    },
    {
      "epoch": 1.8512473022419051,
      "grad_norm": 0.15719948709011078,
      "learning_rate": 1.6755278065191395e-07,
      "loss": 0.0015,
      "step": 288640
    },
    {
      "epoch": 1.8513114391356913,
      "grad_norm": 0.07202229648828506,
      "learning_rate": 1.67409131513655e-07,
      "loss": 0.0008,
      "step": 288650
    },
    {
      "epoch": 1.8513755760294774,
      "grad_norm": 0.1010279506444931,
      "learning_rate": 1.672655429312131e-07,
      "loss": 0.0007,
      "step": 288660
    },
    {
      "epoch": 1.8514397129232634,
      "grad_norm": 0.0597020722925663,
      "learning_rate": 1.671220149063879e-07,
      "loss": 0.0006,
      "step": 288670
    },
    {
      "epoch": 1.8515038498170495,
      "grad_norm": 0.01989404484629631,
      "learning_rate": 1.6697854744097853e-07,
      "loss": 0.001,
      "step": 288680
    },
    {
      "epoch": 1.8515679867108354,
      "grad_norm": 0.14670448005199432,
      "learning_rate": 1.668351405367813e-07,
      "loss": 0.0018,
      "step": 288690
    },
    {
      "epoch": 1.8516321236046216,
      "grad_norm": 0.038981057703495026,
      "learning_rate": 1.6669179419559366e-07,
      "loss": 0.0006,
      "step": 288700
    },
    {
      "epoch": 1.8516962604984077,
      "grad_norm": 0.14221884310245514,
      "learning_rate": 1.665485084192131e-07,
      "loss": 0.0011,
      "step": 288710
    },
    {
      "epoch": 1.851760397392194,
      "grad_norm": 0.06356054544448853,
      "learning_rate": 1.6640528320943372e-07,
      "loss": 0.0017,
      "step": 288720
    },
    {
      "epoch": 1.85182453428598,
      "grad_norm": 0.05507383123040199,
      "learning_rate": 1.6626211856805075e-07,
      "loss": 0.0008,
      "step": 288730
    },
    {
      "epoch": 1.8518886711797662,
      "grad_norm": 0.08880084753036499,
      "learning_rate": 1.6611901449685775e-07,
      "loss": 0.0007,
      "step": 288740
    },
    {
      "epoch": 1.8519528080735523,
      "grad_norm": 0.002199749927967787,
      "learning_rate": 1.659759709976483e-07,
      "loss": 0.0018,
      "step": 288750
    },
    {
      "epoch": 1.8520169449673383,
      "grad_norm": 0.13226568698883057,
      "learning_rate": 1.6583298807221537e-07,
      "loss": 0.0009,
      "step": 288760
    },
    {
      "epoch": 1.8520810818611244,
      "grad_norm": 0.029226161539554596,
      "learning_rate": 1.6569006572234869e-07,
      "loss": 0.0005,
      "step": 288770
    },
    {
      "epoch": 1.8521452187549103,
      "grad_norm": 0.03023863211274147,
      "learning_rate": 1.6554720394984124e-07,
      "loss": 0.0009,
      "step": 288780
    },
    {
      "epoch": 1.8522093556486965,
      "grad_norm": 0.07122538983821869,
      "learning_rate": 1.6540440275648274e-07,
      "loss": 0.0008,
      "step": 288790
    },
    {
      "epoch": 1.8522734925424826,
      "grad_norm": 0.055547356605529785,
      "learning_rate": 1.6526166214406226e-07,
      "loss": 0.0014,
      "step": 288800
    },
    {
      "epoch": 1.8523376294362688,
      "grad_norm": 0.03576844185590744,
      "learning_rate": 1.651189821143684e-07,
      "loss": 0.0012,
      "step": 288810
    },
    {
      "epoch": 1.852401766330055,
      "grad_norm": 0.07118990272283554,
      "learning_rate": 1.6497636266918858e-07,
      "loss": 0.0017,
      "step": 288820
    },
    {
      "epoch": 1.852465903223841,
      "grad_norm": 0.15735577046871185,
      "learning_rate": 1.6483380381031145e-07,
      "loss": 0.0011,
      "step": 288830
    },
    {
      "epoch": 1.852530040117627,
      "grad_norm": 0.0425572507083416,
      "learning_rate": 1.6469130553952219e-07,
      "loss": 0.0011,
      "step": 288840
    },
    {
      "epoch": 1.8525941770114132,
      "grad_norm": 0.08285803347826004,
      "learning_rate": 1.645488678586066e-07,
      "loss": 0.0006,
      "step": 288850
    },
    {
      "epoch": 1.8526583139051993,
      "grad_norm": 0.0034415412228554487,
      "learning_rate": 1.6440649076934878e-07,
      "loss": 0.001,
      "step": 288860
    },
    {
      "epoch": 1.8527224507989852,
      "grad_norm": 0.12120576947927475,
      "learning_rate": 1.6426417427353403e-07,
      "loss": 0.0012,
      "step": 288870
    },
    {
      "epoch": 1.8527865876927714,
      "grad_norm": 0.00897813867777586,
      "learning_rate": 1.6412191837294534e-07,
      "loss": 0.0015,
      "step": 288880
    },
    {
      "epoch": 1.8528507245865575,
      "grad_norm": 0.06584618985652924,
      "learning_rate": 1.6397972306936516e-07,
      "loss": 0.0015,
      "step": 288890
    },
    {
      "epoch": 1.8529148614803437,
      "grad_norm": 0.02014937251806259,
      "learning_rate": 1.6383758836457487e-07,
      "loss": 0.0006,
      "step": 288900
    },
    {
      "epoch": 1.8529789983741298,
      "grad_norm": 0.02061321958899498,
      "learning_rate": 1.636955142603569e-07,
      "loss": 0.0021,
      "step": 288910
    },
    {
      "epoch": 1.853043135267916,
      "grad_norm": 0.4016263484954834,
      "learning_rate": 1.6355350075848986e-07,
      "loss": 0.004,
      "step": 288920
    },
    {
      "epoch": 1.853107272161702,
      "grad_norm": 0.05717543885111809,
      "learning_rate": 1.6341154786075396e-07,
      "loss": 0.0005,
      "step": 288930
    },
    {
      "epoch": 1.853171409055488,
      "grad_norm": 0.03883775696158409,
      "learning_rate": 1.6326965556892836e-07,
      "loss": 0.0014,
      "step": 288940
    },
    {
      "epoch": 1.853235545949274,
      "grad_norm": 0.06243137642741203,
      "learning_rate": 1.631278238847911e-07,
      "loss": 0.001,
      "step": 288950
    },
    {
      "epoch": 1.8532996828430601,
      "grad_norm": 0.12922696769237518,
      "learning_rate": 1.629860528101185e-07,
      "loss": 0.0009,
      "step": 288960
    },
    {
      "epoch": 1.8533638197368463,
      "grad_norm": 0.03439302369952202,
      "learning_rate": 1.6284434234668746e-07,
      "loss": 0.0007,
      "step": 288970
    },
    {
      "epoch": 1.8534279566306324,
      "grad_norm": 0.04470491781830788,
      "learning_rate": 1.6270269249627435e-07,
      "loss": 0.0011,
      "step": 288980
    },
    {
      "epoch": 1.8534920935244186,
      "grad_norm": 0.030634136870503426,
      "learning_rate": 1.6256110326065332e-07,
      "loss": 0.0013,
      "step": 288990
    },
    {
      "epoch": 1.8535562304182047,
      "grad_norm": 0.015092063695192337,
      "learning_rate": 1.6241957464159907e-07,
      "loss": 0.0005,
      "step": 289000
    },
    {
      "epoch": 1.8536203673119909,
      "grad_norm": 0.09808158874511719,
      "learning_rate": 1.6227810664088462e-07,
      "loss": 0.0018,
      "step": 289010
    },
    {
      "epoch": 1.8536845042057768,
      "grad_norm": 0.07334356009960175,
      "learning_rate": 1.6213669926028352e-07,
      "loss": 0.0012,
      "step": 289020
    },
    {
      "epoch": 1.853748641099563,
      "grad_norm": 0.03621971979737282,
      "learning_rate": 1.6199535250156717e-07,
      "loss": 0.0011,
      "step": 289030
    },
    {
      "epoch": 1.8538127779933489,
      "grad_norm": 0.03564497083425522,
      "learning_rate": 1.618540663665058e-07,
      "loss": 0.0009,
      "step": 289040
    },
    {
      "epoch": 1.853876914887135,
      "grad_norm": 0.0414310023188591,
      "learning_rate": 1.6171284085687133e-07,
      "loss": 0.0014,
      "step": 289050
    },
    {
      "epoch": 1.8539410517809212,
      "grad_norm": 0.041223831474781036,
      "learning_rate": 1.6157167597443345e-07,
      "loss": 0.0004,
      "step": 289060
    },
    {
      "epoch": 1.8540051886747073,
      "grad_norm": 0.0021375659853219986,
      "learning_rate": 1.6143057172095965e-07,
      "loss": 0.0016,
      "step": 289070
    },
    {
      "epoch": 1.8540693255684935,
      "grad_norm": 0.017594551667571068,
      "learning_rate": 1.6128952809821852e-07,
      "loss": 0.0017,
      "step": 289080
    },
    {
      "epoch": 1.8541334624622796,
      "grad_norm": 0.16080540418624878,
      "learning_rate": 1.6114854510797862e-07,
      "loss": 0.0015,
      "step": 289090
    },
    {
      "epoch": 1.8541975993560655,
      "grad_norm": 0.024991154670715332,
      "learning_rate": 1.6100762275200576e-07,
      "loss": 0.0017,
      "step": 289100
    },
    {
      "epoch": 1.8542617362498517,
      "grad_norm": 0.10999748855829239,
      "learning_rate": 1.6086676103206577e-07,
      "loss": 0.0007,
      "step": 289110
    },
    {
      "epoch": 1.8543258731436376,
      "grad_norm": 0.11018364131450653,
      "learning_rate": 1.607259599499228e-07,
      "loss": 0.0006,
      "step": 289120
    },
    {
      "epoch": 1.8543900100374238,
      "grad_norm": 0.0660175308585167,
      "learning_rate": 1.6058521950734262e-07,
      "loss": 0.001,
      "step": 289130
    },
    {
      "epoch": 1.85445414693121,
      "grad_norm": 0.06319452077150345,
      "learning_rate": 1.6044453970608886e-07,
      "loss": 0.0007,
      "step": 289140
    },
    {
      "epoch": 1.854518283824996,
      "grad_norm": 0.19534561038017273,
      "learning_rate": 1.6030392054792344e-07,
      "loss": 0.002,
      "step": 289150
    },
    {
      "epoch": 1.8545824207187822,
      "grad_norm": 0.03305432200431824,
      "learning_rate": 1.6016336203460825e-07,
      "loss": 0.0008,
      "step": 289160
    },
    {
      "epoch": 1.8546465576125684,
      "grad_norm": 0.06634438782930374,
      "learning_rate": 1.600228641679058e-07,
      "loss": 0.0009,
      "step": 289170
    },
    {
      "epoch": 1.8547106945063545,
      "grad_norm": 0.01888250932097435,
      "learning_rate": 1.5988242694957577e-07,
      "loss": 0.0015,
      "step": 289180
    },
    {
      "epoch": 1.8547748314001404,
      "grad_norm": 0.009038791060447693,
      "learning_rate": 1.597420503813779e-07,
      "loss": 0.0007,
      "step": 289190
    },
    {
      "epoch": 1.8548389682939266,
      "grad_norm": 0.13109901547431946,
      "learning_rate": 1.5960173446507187e-07,
      "loss": 0.0039,
      "step": 289200
    },
    {
      "epoch": 1.8549031051877125,
      "grad_norm": 0.04096861928701401,
      "learning_rate": 1.5946147920241516e-07,
      "loss": 0.0007,
      "step": 289210
    },
    {
      "epoch": 1.8549672420814987,
      "grad_norm": 0.07823286950588226,
      "learning_rate": 1.5932128459516582e-07,
      "loss": 0.0009,
      "step": 289220
    },
    {
      "epoch": 1.8550313789752848,
      "grad_norm": 0.044161245226860046,
      "learning_rate": 1.5918115064507966e-07,
      "loss": 0.0006,
      "step": 289230
    },
    {
      "epoch": 1.855095515869071,
      "grad_norm": 0.1879640519618988,
      "learning_rate": 1.5904107735391415e-07,
      "loss": 0.0007,
      "step": 289240
    },
    {
      "epoch": 1.855159652762857,
      "grad_norm": 0.056915298104286194,
      "learning_rate": 1.589010647234235e-07,
      "loss": 0.0017,
      "step": 289250
    },
    {
      "epoch": 1.8552237896566433,
      "grad_norm": 0.18126241862773895,
      "learning_rate": 1.5876111275536231e-07,
      "loss": 0.0013,
      "step": 289260
    },
    {
      "epoch": 1.8552879265504294,
      "grad_norm": 0.10514125972986221,
      "learning_rate": 1.586212214514843e-07,
      "loss": 0.0004,
      "step": 289270
    },
    {
      "epoch": 1.8553520634442153,
      "grad_norm": 0.05540881305932999,
      "learning_rate": 1.5848139081354242e-07,
      "loss": 0.0026,
      "step": 289280
    },
    {
      "epoch": 1.8554162003380015,
      "grad_norm": 0.08097107708454132,
      "learning_rate": 1.583416208432892e-07,
      "loss": 0.0014,
      "step": 289290
    },
    {
      "epoch": 1.8554803372317874,
      "grad_norm": 0.07919275015592575,
      "learning_rate": 1.58201911542476e-07,
      "loss": 0.0018,
      "step": 289300
    },
    {
      "epoch": 1.8555444741255736,
      "grad_norm": 0.14636743068695068,
      "learning_rate": 1.58062262912852e-07,
      "loss": 0.0009,
      "step": 289310
    },
    {
      "epoch": 1.8556086110193597,
      "grad_norm": 0.1234036535024643,
      "learning_rate": 1.5792267495616963e-07,
      "loss": 0.0023,
      "step": 289320
    },
    {
      "epoch": 1.8556727479131458,
      "grad_norm": 0.08552830666303635,
      "learning_rate": 1.5778314767417645e-07,
      "loss": 0.0009,
      "step": 289330
    },
    {
      "epoch": 1.855736884806932,
      "grad_norm": 0.0010238487739115953,
      "learning_rate": 1.5764368106862048e-07,
      "loss": 0.0015,
      "step": 289340
    },
    {
      "epoch": 1.8558010217007181,
      "grad_norm": 0.19790436327457428,
      "learning_rate": 1.5750427514125034e-07,
      "loss": 0.0008,
      "step": 289350
    },
    {
      "epoch": 1.855865158594504,
      "grad_norm": 0.03808882459998131,
      "learning_rate": 1.5736492989381292e-07,
      "loss": 0.0007,
      "step": 289360
    },
    {
      "epoch": 1.8559292954882902,
      "grad_norm": 0.00809218268841505,
      "learning_rate": 1.5722564532805352e-07,
      "loss": 0.0008,
      "step": 289370
    },
    {
      "epoch": 1.8559934323820761,
      "grad_norm": 0.02840617671608925,
      "learning_rate": 1.570864214457174e-07,
      "loss": 0.001,
      "step": 289380
    },
    {
      "epoch": 1.8560575692758623,
      "grad_norm": 0.007927249185740948,
      "learning_rate": 1.569472582485504e-07,
      "loss": 0.0014,
      "step": 289390
    },
    {
      "epoch": 1.8561217061696484,
      "grad_norm": 0.03396786376833916,
      "learning_rate": 1.56808155738295e-07,
      "loss": 0.0008,
      "step": 289400
    },
    {
      "epoch": 1.8561858430634346,
      "grad_norm": 0.007886813022196293,
      "learning_rate": 1.5666911391669537e-07,
      "loss": 0.0016,
      "step": 289410
    },
    {
      "epoch": 1.8562499799572207,
      "grad_norm": 0.06851857155561447,
      "learning_rate": 1.5653013278549234e-07,
      "loss": 0.0011,
      "step": 289420
    },
    {
      "epoch": 1.856314116851007,
      "grad_norm": 0.26010021567344666,
      "learning_rate": 1.5639121234642895e-07,
      "loss": 0.0009,
      "step": 289430
    },
    {
      "epoch": 1.856378253744793,
      "grad_norm": 0.03365808352828026,
      "learning_rate": 1.562523526012455e-07,
      "loss": 0.0008,
      "step": 289440
    },
    {
      "epoch": 1.856442390638579,
      "grad_norm": 0.0676041916012764,
      "learning_rate": 1.5611355355168167e-07,
      "loss": 0.0014,
      "step": 289450
    },
    {
      "epoch": 1.8565065275323651,
      "grad_norm": 0.10150793194770813,
      "learning_rate": 1.559748151994761e-07,
      "loss": 0.0012,
      "step": 289460
    },
    {
      "epoch": 1.856570664426151,
      "grad_norm": 0.01773030310869217,
      "learning_rate": 1.5583613754636906e-07,
      "loss": 0.0013,
      "step": 289470
    },
    {
      "epoch": 1.8566348013199372,
      "grad_norm": 0.3465256094932556,
      "learning_rate": 1.5569752059409748e-07,
      "loss": 0.0016,
      "step": 289480
    },
    {
      "epoch": 1.8566989382137233,
      "grad_norm": 0.10367559641599655,
      "learning_rate": 1.5555896434439722e-07,
      "loss": 0.0012,
      "step": 289490
    },
    {
      "epoch": 1.8567630751075095,
      "grad_norm": 0.2588544487953186,
      "learning_rate": 1.5542046879900573e-07,
      "loss": 0.0021,
      "step": 289500
    },
    {
      "epoch": 1.8568272120012956,
      "grad_norm": 0.08816307038068771,
      "learning_rate": 1.5528203395965835e-07,
      "loss": 0.0004,
      "step": 289510
    },
    {
      "epoch": 1.8568913488950818,
      "grad_norm": 0.01976894959807396,
      "learning_rate": 1.5514365982808975e-07,
      "loss": 0.001,
      "step": 289520
    },
    {
      "epoch": 1.8569554857888677,
      "grad_norm": 0.01618194580078125,
      "learning_rate": 1.5500534640603304e-07,
      "loss": 0.0007,
      "step": 289530
    },
    {
      "epoch": 1.8570196226826539,
      "grad_norm": 0.09402556717395782,
      "learning_rate": 1.5486709369522236e-07,
      "loss": 0.0012,
      "step": 289540
    },
    {
      "epoch": 1.85708375957644,
      "grad_norm": 0.06722892820835114,
      "learning_rate": 1.5472890169738909e-07,
      "loss": 0.0009,
      "step": 289550
    },
    {
      "epoch": 1.857147896470226,
      "grad_norm": 0.24180275201797485,
      "learning_rate": 1.545907704142663e-07,
      "loss": 0.0009,
      "step": 289560
    },
    {
      "epoch": 1.857212033364012,
      "grad_norm": 0.14629188179969788,
      "learning_rate": 1.5445269984758427e-07,
      "loss": 0.0015,
      "step": 289570
    },
    {
      "epoch": 1.8572761702577982,
      "grad_norm": 0.03688056394457817,
      "learning_rate": 1.5431468999907218e-07,
      "loss": 0.0013,
      "step": 289580
    },
    {
      "epoch": 1.8573403071515844,
      "grad_norm": 0.06282052397727966,
      "learning_rate": 1.5417674087046087e-07,
      "loss": 0.0029,
      "step": 289590
    },
    {
      "epoch": 1.8574044440453705,
      "grad_norm": 0.07084057480096817,
      "learning_rate": 1.540388524634784e-07,
      "loss": 0.0009,
      "step": 289600
    },
    {
      "epoch": 1.8574685809391567,
      "grad_norm": 0.02749285101890564,
      "learning_rate": 1.539010247798517e-07,
      "loss": 0.0003,
      "step": 289610
    },
    {
      "epoch": 1.8575327178329426,
      "grad_norm": 0.13757991790771484,
      "learning_rate": 1.5376325782130885e-07,
      "loss": 0.0014,
      "step": 289620
    },
    {
      "epoch": 1.8575968547267288,
      "grad_norm": 0.00561791704967618,
      "learning_rate": 1.5362555158957626e-07,
      "loss": 0.0017,
      "step": 289630
    },
    {
      "epoch": 1.8576609916205147,
      "grad_norm": 0.1288556158542633,
      "learning_rate": 1.534879060863792e-07,
      "loss": 0.0012,
      "step": 289640
    },
    {
      "epoch": 1.8577251285143008,
      "grad_norm": 0.07304224371910095,
      "learning_rate": 1.5335032131344185e-07,
      "loss": 0.0024,
      "step": 289650
    },
    {
      "epoch": 1.857789265408087,
      "grad_norm": 0.018483763560652733,
      "learning_rate": 1.532127972724895e-07,
      "loss": 0.0009,
      "step": 289660
    },
    {
      "epoch": 1.8578534023018731,
      "grad_norm": 0.033974286168813705,
      "learning_rate": 1.5307533396524466e-07,
      "loss": 0.0012,
      "step": 289670
    },
    {
      "epoch": 1.8579175391956593,
      "grad_norm": 0.017961300909519196,
      "learning_rate": 1.5293793139342984e-07,
      "loss": 0.001,
      "step": 289680
    },
    {
      "epoch": 1.8579816760894454,
      "grad_norm": 0.08534421771764755,
      "learning_rate": 1.5280058955876642e-07,
      "loss": 0.0025,
      "step": 289690
    },
    {
      "epoch": 1.8580458129832316,
      "grad_norm": 0.05251190811395645,
      "learning_rate": 1.5266330846297638e-07,
      "loss": 0.0022,
      "step": 289700
    },
    {
      "epoch": 1.8581099498770175,
      "grad_norm": 0.03260777145624161,
      "learning_rate": 1.5252608810777946e-07,
      "loss": 0.0004,
      "step": 289710
    },
    {
      "epoch": 1.8581740867708036,
      "grad_norm": 0.08294403553009033,
      "learning_rate": 1.5238892849489483e-07,
      "loss": 0.001,
      "step": 289720
    },
    {
      "epoch": 1.8582382236645896,
      "grad_norm": 0.0010562815004959702,
      "learning_rate": 1.5225182962604112e-07,
      "loss": 0.0015,
      "step": 289730
    },
    {
      "epoch": 1.8583023605583757,
      "grad_norm": 0.05841523036360741,
      "learning_rate": 1.521147915029375e-07,
      "loss": 0.0009,
      "step": 289740
    },
    {
      "epoch": 1.8583664974521619,
      "grad_norm": 0.024304818361997604,
      "learning_rate": 1.5197781412729984e-07,
      "loss": 0.0022,
      "step": 289750
    },
    {
      "epoch": 1.858430634345948,
      "grad_norm": 0.17089597880840302,
      "learning_rate": 1.518408975008445e-07,
      "loss": 0.001,
      "step": 289760
    },
    {
      "epoch": 1.8584947712397342,
      "grad_norm": 0.008853795938193798,
      "learning_rate": 1.5170404162528796e-07,
      "loss": 0.0008,
      "step": 289770
    },
    {
      "epoch": 1.8585589081335203,
      "grad_norm": 0.032498959451913834,
      "learning_rate": 1.5156724650234545e-07,
      "loss": 0.0006,
      "step": 289780
    },
    {
      "epoch": 1.8586230450273062,
      "grad_norm": 0.029728278517723083,
      "learning_rate": 1.5143051213373007e-07,
      "loss": 0.0009,
      "step": 289790
    },
    {
      "epoch": 1.8586871819210924,
      "grad_norm": 0.08156143873929977,
      "learning_rate": 1.5129383852115486e-07,
      "loss": 0.0011,
      "step": 289800
    },
    {
      "epoch": 1.8587513188148783,
      "grad_norm": 0.016715293750166893,
      "learning_rate": 1.5115722566633406e-07,
      "loss": 0.0009,
      "step": 289810
    },
    {
      "epoch": 1.8588154557086645,
      "grad_norm": 0.05936663597822189,
      "learning_rate": 1.510206735709785e-07,
      "loss": 0.0019,
      "step": 289820
    },
    {
      "epoch": 1.8588795926024506,
      "grad_norm": 0.02317112125456333,
      "learning_rate": 1.5088418223679956e-07,
      "loss": 0.0007,
      "step": 289830
    },
    {
      "epoch": 1.8589437294962368,
      "grad_norm": 0.0055950540117919445,
      "learning_rate": 1.5074775166550647e-07,
      "loss": 0.0012,
      "step": 289840
    },
    {
      "epoch": 1.859007866390023,
      "grad_norm": 0.21792720258235931,
      "learning_rate": 1.5061138185881063e-07,
      "loss": 0.0015,
      "step": 289850
    },
    {
      "epoch": 1.859072003283809,
      "grad_norm": 0.13058669865131378,
      "learning_rate": 1.5047507281842012e-07,
      "loss": 0.0011,
      "step": 289860
    },
    {
      "epoch": 1.8591361401775952,
      "grad_norm": 0.012549787759780884,
      "learning_rate": 1.5033882454604244e-07,
      "loss": 0.0013,
      "step": 289870
    },
    {
      "epoch": 1.8592002770713811,
      "grad_norm": 0.2569255232810974,
      "learning_rate": 1.5020263704338513e-07,
      "loss": 0.0017,
      "step": 289880
    },
    {
      "epoch": 1.8592644139651673,
      "grad_norm": 0.09576437622308731,
      "learning_rate": 1.5006651031215513e-07,
      "loss": 0.0011,
      "step": 289890
    },
    {
      "epoch": 1.8593285508589532,
      "grad_norm": 0.058203332126140594,
      "learning_rate": 1.4993044435405836e-07,
      "loss": 0.001,
      "step": 289900
    },
    {
      "epoch": 1.8593926877527394,
      "grad_norm": 0.1404581069946289,
      "learning_rate": 1.497944391707984e-07,
      "loss": 0.0009,
      "step": 289910
    },
    {
      "epoch": 1.8594568246465255,
      "grad_norm": 0.002524998504668474,
      "learning_rate": 1.4965849476408113e-07,
      "loss": 0.0012,
      "step": 289920
    },
    {
      "epoch": 1.8595209615403117,
      "grad_norm": 0.1650613397359848,
      "learning_rate": 1.4952261113560963e-07,
      "loss": 0.0011,
      "step": 289930
    },
    {
      "epoch": 1.8595850984340978,
      "grad_norm": 0.049357227981090546,
      "learning_rate": 1.4938678828708642e-07,
      "loss": 0.0005,
      "step": 289940
    },
    {
      "epoch": 1.859649235327884,
      "grad_norm": 0.12375776469707489,
      "learning_rate": 1.4925102622021293e-07,
      "loss": 0.0021,
      "step": 289950
    },
    {
      "epoch": 1.8597133722216699,
      "grad_norm": 0.11767815053462982,
      "learning_rate": 1.4911532493669113e-07,
      "loss": 0.0012,
      "step": 289960
    },
    {
      "epoch": 1.859777509115456,
      "grad_norm": 0.14198528230190277,
      "learning_rate": 1.489796844382213e-07,
      "loss": 0.0021,
      "step": 289970
    },
    {
      "epoch": 1.8598416460092422,
      "grad_norm": 0.02816985361278057,
      "learning_rate": 1.4884410472650269e-07,
      "loss": 0.0004,
      "step": 289980
    },
    {
      "epoch": 1.859905782903028,
      "grad_norm": 0.1383359581232071,
      "learning_rate": 1.4870858580323445e-07,
      "loss": 0.0008,
      "step": 289990
    },
    {
      "epoch": 1.8599699197968143,
      "grad_norm": 0.10833131521940231,
      "learning_rate": 1.4857312767011521e-07,
      "loss": 0.0009,
      "step": 290000
    },
    {
      "epoch": 1.8600340566906004,
      "grad_norm": 0.13313689827919006,
      "learning_rate": 1.48437730328842e-07,
      "loss": 0.0011,
      "step": 290010
    },
    {
      "epoch": 1.8600981935843866,
      "grad_norm": 0.06083110347390175,
      "learning_rate": 1.4830239378111123e-07,
      "loss": 0.0007,
      "step": 290020
    },
    {
      "epoch": 1.8601623304781727,
      "grad_norm": 0.0641220286488533,
      "learning_rate": 1.4816711802861816e-07,
      "loss": 0.0013,
      "step": 290030
    },
    {
      "epoch": 1.8602264673719588,
      "grad_norm": 0.00814772117882967,
      "learning_rate": 1.4803190307305982e-07,
      "loss": 0.0004,
      "step": 290040
    },
    {
      "epoch": 1.8602906042657448,
      "grad_norm": 0.039491720497608185,
      "learning_rate": 1.478967489161287e-07,
      "loss": 0.0011,
      "step": 290050
    },
    {
      "epoch": 1.860354741159531,
      "grad_norm": 0.0898330882191658,
      "learning_rate": 1.4776165555951904e-07,
      "loss": 0.0011,
      "step": 290060
    },
    {
      "epoch": 1.8604188780533168,
      "grad_norm": 0.04195317625999451,
      "learning_rate": 1.4762662300492392e-07,
      "loss": 0.0006,
      "step": 290070
    },
    {
      "epoch": 1.860483014947103,
      "grad_norm": 0.08919844031333923,
      "learning_rate": 1.474916512540353e-07,
      "loss": 0.0006,
      "step": 290080
    },
    {
      "epoch": 1.8605471518408891,
      "grad_norm": 0.08594311773777008,
      "learning_rate": 1.4735674030854408e-07,
      "loss": 0.0013,
      "step": 290090
    },
    {
      "epoch": 1.8606112887346753,
      "grad_norm": 0.020756598562002182,
      "learning_rate": 1.472218901701411e-07,
      "loss": 0.0015,
      "step": 290100
    },
    {
      "epoch": 1.8606754256284614,
      "grad_norm": 0.04540959745645523,
      "learning_rate": 1.4708710084051613e-07,
      "loss": 0.0006,
      "step": 290110
    },
    {
      "epoch": 1.8607395625222476,
      "grad_norm": 0.014379561878740788,
      "learning_rate": 1.4695237232135785e-07,
      "loss": 0.0005,
      "step": 290120
    },
    {
      "epoch": 1.8608036994160337,
      "grad_norm": 0.07023920118808746,
      "learning_rate": 1.468177046143554e-07,
      "loss": 0.0008,
      "step": 290130
    },
    {
      "epoch": 1.8608678363098197,
      "grad_norm": 0.1338580846786499,
      "learning_rate": 1.4668309772119471e-07,
      "loss": 0.0012,
      "step": 290140
    },
    {
      "epoch": 1.8609319732036058,
      "grad_norm": 0.08372864127159119,
      "learning_rate": 1.4654855164356386e-07,
      "loss": 0.0032,
      "step": 290150
    },
    {
      "epoch": 1.8609961100973917,
      "grad_norm": 0.02373339794576168,
      "learning_rate": 1.464140663831487e-07,
      "loss": 0.001,
      "step": 290160
    },
    {
      "epoch": 1.861060246991178,
      "grad_norm": 0.2806231677532196,
      "learning_rate": 1.46279641941634e-07,
      "loss": 0.001,
      "step": 290170
    },
    {
      "epoch": 1.861124383884964,
      "grad_norm": 0.04377453774213791,
      "learning_rate": 1.4614527832070347e-07,
      "loss": 0.0006,
      "step": 290180
    },
    {
      "epoch": 1.8611885207787502,
      "grad_norm": 0.27308493852615356,
      "learning_rate": 1.4601097552204235e-07,
      "loss": 0.0011,
      "step": 290190
    },
    {
      "epoch": 1.8612526576725363,
      "grad_norm": 0.09190023690462112,
      "learning_rate": 1.4587673354733267e-07,
      "loss": 0.0014,
      "step": 290200
    },
    {
      "epoch": 1.8613167945663225,
      "grad_norm": 0.1294102668762207,
      "learning_rate": 1.4574255239825642e-07,
      "loss": 0.0017,
      "step": 290210
    },
    {
      "epoch": 1.8613809314601084,
      "grad_norm": 0.021928202360868454,
      "learning_rate": 1.4560843207649557e-07,
      "loss": 0.0013,
      "step": 290220
    },
    {
      "epoch": 1.8614450683538946,
      "grad_norm": 0.09823820739984512,
      "learning_rate": 1.4547437258373098e-07,
      "loss": 0.0009,
      "step": 290230
    },
    {
      "epoch": 1.8615092052476805,
      "grad_norm": 0.08892396092414856,
      "learning_rate": 1.4534037392164136e-07,
      "loss": 0.0005,
      "step": 290240
    },
    {
      "epoch": 1.8615733421414666,
      "grad_norm": 0.055825598537921906,
      "learning_rate": 1.4520643609190588e-07,
      "loss": 0.0018,
      "step": 290250
    },
    {
      "epoch": 1.8616374790352528,
      "grad_norm": 0.016030259430408478,
      "learning_rate": 1.4507255909620431e-07,
      "loss": 0.0008,
      "step": 290260
    },
    {
      "epoch": 1.861701615929039,
      "grad_norm": 0.1098199263215065,
      "learning_rate": 1.4493874293621312e-07,
      "loss": 0.0022,
      "step": 290270
    },
    {
      "epoch": 1.861765752822825,
      "grad_norm": 0.011548800393939018,
      "learning_rate": 1.4480498761360984e-07,
      "loss": 0.0009,
      "step": 290280
    },
    {
      "epoch": 1.8618298897166112,
      "grad_norm": 0.16268345713615417,
      "learning_rate": 1.4467129313006868e-07,
      "loss": 0.0017,
      "step": 290290
    },
    {
      "epoch": 1.8618940266103974,
      "grad_norm": 0.09902417659759521,
      "learning_rate": 1.4453765948726717e-07,
      "loss": 0.0011,
      "step": 290300
    },
    {
      "epoch": 1.8619581635041833,
      "grad_norm": 0.03196365013718605,
      "learning_rate": 1.4440408668687844e-07,
      "loss": 0.0012,
      "step": 290310
    },
    {
      "epoch": 1.8620223003979695,
      "grad_norm": 0.11622941493988037,
      "learning_rate": 1.4427057473057727e-07,
      "loss": 0.0008,
      "step": 290320
    },
    {
      "epoch": 1.8620864372917554,
      "grad_norm": 0.06088544428348541,
      "learning_rate": 1.4413712362003562e-07,
      "loss": 0.001,
      "step": 290330
    },
    {
      "epoch": 1.8621505741855415,
      "grad_norm": 0.11155184358358383,
      "learning_rate": 1.440037333569272e-07,
      "loss": 0.001,
      "step": 290340
    },
    {
      "epoch": 1.8622147110793277,
      "grad_norm": 0.03502899408340454,
      "learning_rate": 1.4387040394292175e-07,
      "loss": 0.001,
      "step": 290350
    },
    {
      "epoch": 1.8622788479731138,
      "grad_norm": 0.006140367593616247,
      "learning_rate": 1.4373713537969125e-07,
      "loss": 0.001,
      "step": 290360
    },
    {
      "epoch": 1.8623429848669,
      "grad_norm": 0.2765370309352875,
      "learning_rate": 1.436039276689044e-07,
      "loss": 0.0011,
      "step": 290370
    },
    {
      "epoch": 1.8624071217606861,
      "grad_norm": 0.06028538942337036,
      "learning_rate": 1.434707808122321e-07,
      "loss": 0.0009,
      "step": 290380
    },
    {
      "epoch": 1.862471258654472,
      "grad_norm": 0.17979082465171814,
      "learning_rate": 1.433376948113413e-07,
      "loss": 0.0019,
      "step": 290390
    },
    {
      "epoch": 1.8625353955482582,
      "grad_norm": 0.13455578684806824,
      "learning_rate": 1.4320466966790015e-07,
      "loss": 0.0009,
      "step": 290400
    },
    {
      "epoch": 1.8625995324420443,
      "grad_norm": 0.01863623596727848,
      "learning_rate": 1.4307170538357563e-07,
      "loss": 0.0014,
      "step": 290410
    },
    {
      "epoch": 1.8626636693358303,
      "grad_norm": 0.10759945958852768,
      "learning_rate": 1.4293880196003363e-07,
      "loss": 0.0009,
      "step": 290420
    },
    {
      "epoch": 1.8627278062296164,
      "grad_norm": 0.0366380400955677,
      "learning_rate": 1.4280595939894005e-07,
      "loss": 0.0006,
      "step": 290430
    },
    {
      "epoch": 1.8627919431234026,
      "grad_norm": 0.08533468842506409,
      "learning_rate": 1.4267317770195966e-07,
      "loss": 0.0018,
      "step": 290440
    },
    {
      "epoch": 1.8628560800171887,
      "grad_norm": 0.04634246230125427,
      "learning_rate": 1.4254045687075502e-07,
      "loss": 0.0039,
      "step": 290450
    },
    {
      "epoch": 1.8629202169109749,
      "grad_norm": 0.02340097166597843,
      "learning_rate": 1.4240779690699035e-07,
      "loss": 0.0011,
      "step": 290460
    },
    {
      "epoch": 1.862984353804761,
      "grad_norm": 0.017465023323893547,
      "learning_rate": 1.4227519781232767e-07,
      "loss": 0.001,
      "step": 290470
    },
    {
      "epoch": 1.863048490698547,
      "grad_norm": 0.06610596925020218,
      "learning_rate": 1.4214265958842787e-07,
      "loss": 0.0008,
      "step": 290480
    },
    {
      "epoch": 1.863112627592333,
      "grad_norm": 0.0007582298712804914,
      "learning_rate": 1.4201018223695295e-07,
      "loss": 0.0022,
      "step": 290490
    },
    {
      "epoch": 1.863176764486119,
      "grad_norm": 0.09911002218723297,
      "learning_rate": 1.418777657595627e-07,
      "loss": 0.0018,
      "step": 290500
    },
    {
      "epoch": 1.8632409013799052,
      "grad_norm": 0.0666009783744812,
      "learning_rate": 1.4174541015791633e-07,
      "loss": 0.0007,
      "step": 290510
    },
    {
      "epoch": 1.8633050382736913,
      "grad_norm": 0.05044683441519737,
      "learning_rate": 1.4161311543367084e-07,
      "loss": 0.001,
      "step": 290520
    },
    {
      "epoch": 1.8633691751674775,
      "grad_norm": 0.09418749809265137,
      "learning_rate": 1.414808815884866e-07,
      "loss": 0.0011,
      "step": 290530
    },
    {
      "epoch": 1.8634333120612636,
      "grad_norm": 0.032017726451158524,
      "learning_rate": 1.413487086240184e-07,
      "loss": 0.0007,
      "step": 290540
    },
    {
      "epoch": 1.8634974489550498,
      "grad_norm": 0.04044421762228012,
      "learning_rate": 1.4121659654192377e-07,
      "loss": 0.0011,
      "step": 290550
    },
    {
      "epoch": 1.863561585848836,
      "grad_norm": 0.04872051253914833,
      "learning_rate": 1.4108454534385696e-07,
      "loss": 0.0011,
      "step": 290560
    },
    {
      "epoch": 1.8636257227426218,
      "grad_norm": 0.12192614376544952,
      "learning_rate": 1.4095255503147443e-07,
      "loss": 0.0007,
      "step": 290570
    },
    {
      "epoch": 1.863689859636408,
      "grad_norm": 0.09276101738214493,
      "learning_rate": 1.4082062560642872e-07,
      "loss": 0.0012,
      "step": 290580
    },
    {
      "epoch": 1.863753996530194,
      "grad_norm": 0.0316673219203949,
      "learning_rate": 1.4068875707037355e-07,
      "loss": 0.0015,
      "step": 290590
    },
    {
      "epoch": 1.86381813342398,
      "grad_norm": 0.04883475974202156,
      "learning_rate": 1.4055694942496035e-07,
      "loss": 0.0009,
      "step": 290600
    },
    {
      "epoch": 1.8638822703177662,
      "grad_norm": 0.014902455732226372,
      "learning_rate": 1.4042520267184224e-07,
      "loss": 0.0008,
      "step": 290610
    },
    {
      "epoch": 1.8639464072115524,
      "grad_norm": 0.028056534007191658,
      "learning_rate": 1.4029351681266955e-07,
      "loss": 0.0024,
      "step": 290620
    },
    {
      "epoch": 1.8640105441053385,
      "grad_norm": 0.046454910188913345,
      "learning_rate": 1.4016189184909156e-07,
      "loss": 0.0009,
      "step": 290630
    },
    {
      "epoch": 1.8640746809991247,
      "grad_norm": 0.06369485706090927,
      "learning_rate": 1.400303277827586e-07,
      "loss": 0.0011,
      "step": 290640
    },
    {
      "epoch": 1.8641388178929106,
      "grad_norm": 0.01218261756002903,
      "learning_rate": 1.3989882461531933e-07,
      "loss": 0.0016,
      "step": 290650
    },
    {
      "epoch": 1.8642029547866967,
      "grad_norm": 0.021232986822724342,
      "learning_rate": 1.3976738234842136e-07,
      "loss": 0.0008,
      "step": 290660
    },
    {
      "epoch": 1.8642670916804827,
      "grad_norm": 0.08904554694890976,
      "learning_rate": 1.396360009837111e-07,
      "loss": 0.0007,
      "step": 290670
    },
    {
      "epoch": 1.8643312285742688,
      "grad_norm": 0.038141753524541855,
      "learning_rate": 1.3950468052283562e-07,
      "loss": 0.0006,
      "step": 290680
    },
    {
      "epoch": 1.864395365468055,
      "grad_norm": 0.04452548548579216,
      "learning_rate": 1.3937342096744077e-07,
      "loss": 0.0008,
      "step": 290690
    },
    {
      "epoch": 1.864459502361841,
      "grad_norm": 0.03853532299399376,
      "learning_rate": 1.3924222231917028e-07,
      "loss": 0.0011,
      "step": 290700
    },
    {
      "epoch": 1.8645236392556273,
      "grad_norm": 0.07612650841474533,
      "learning_rate": 1.3911108457966837e-07,
      "loss": 0.0012,
      "step": 290710
    },
    {
      "epoch": 1.8645877761494134,
      "grad_norm": 0.03762242570519447,
      "learning_rate": 1.3898000775057928e-07,
      "loss": 0.0016,
      "step": 290720
    },
    {
      "epoch": 1.8646519130431995,
      "grad_norm": 0.017636556178331375,
      "learning_rate": 1.388489918335445e-07,
      "loss": 0.0006,
      "step": 290730
    },
    {
      "epoch": 1.8647160499369855,
      "grad_norm": 0.050190914422273636,
      "learning_rate": 1.3871803683020546e-07,
      "loss": 0.0007,
      "step": 290740
    },
    {
      "epoch": 1.8647801868307716,
      "grad_norm": 0.0019875033758580685,
      "learning_rate": 1.3858714274220474e-07,
      "loss": 0.0004,
      "step": 290750
    },
    {
      "epoch": 1.8648443237245576,
      "grad_norm": 0.15975508093833923,
      "learning_rate": 1.3845630957118107e-07,
      "loss": 0.0014,
      "step": 290760
    },
    {
      "epoch": 1.8649084606183437,
      "grad_norm": 0.04271375387907028,
      "learning_rate": 1.3832553731877474e-07,
      "loss": 0.0009,
      "step": 290770
    },
    {
      "epoch": 1.8649725975121298,
      "grad_norm": 0.043441157788038254,
      "learning_rate": 1.3819482598662281e-07,
      "loss": 0.001,
      "step": 290780
    },
    {
      "epoch": 1.865036734405916,
      "grad_norm": 0.0461297333240509,
      "learning_rate": 1.3806417557636564e-07,
      "loss": 0.0011,
      "step": 290790
    },
    {
      "epoch": 1.8651008712997021,
      "grad_norm": 0.00915750116109848,
      "learning_rate": 1.379335860896386e-07,
      "loss": 0.0016,
      "step": 290800
    },
    {
      "epoch": 1.8651650081934883,
      "grad_norm": 0.06905308365821838,
      "learning_rate": 1.378030575280792e-07,
      "loss": 0.0015,
      "step": 290810
    },
    {
      "epoch": 1.8652291450872744,
      "grad_norm": 0.04653427377343178,
      "learning_rate": 1.3767258989332122e-07,
      "loss": 0.0007,
      "step": 290820
    },
    {
      "epoch": 1.8652932819810604,
      "grad_norm": 0.11974623054265976,
      "learning_rate": 1.3754218318700164e-07,
      "loss": 0.0008,
      "step": 290830
    },
    {
      "epoch": 1.8653574188748465,
      "grad_norm": 0.060955844819545746,
      "learning_rate": 1.374118374107536e-07,
      "loss": 0.0016,
      "step": 290840
    },
    {
      "epoch": 1.8654215557686324,
      "grad_norm": 0.1308441460132599,
      "learning_rate": 1.3728155256621078e-07,
      "loss": 0.0014,
      "step": 290850
    },
    {
      "epoch": 1.8654856926624186,
      "grad_norm": 0.2624158561229706,
      "learning_rate": 1.3715132865500468e-07,
      "loss": 0.0039,
      "step": 290860
    },
    {
      "epoch": 1.8655498295562047,
      "grad_norm": 0.06166985630989075,
      "learning_rate": 1.3702116567876788e-07,
      "loss": 0.0012,
      "step": 290870
    },
    {
      "epoch": 1.865613966449991,
      "grad_norm": 0.06867585331201553,
      "learning_rate": 1.3689106363913186e-07,
      "loss": 0.0012,
      "step": 290880
    },
    {
      "epoch": 1.865678103343777,
      "grad_norm": 0.232400581240654,
      "learning_rate": 1.3676102253772583e-07,
      "loss": 0.0016,
      "step": 290890
    },
    {
      "epoch": 1.8657422402375632,
      "grad_norm": 0.0076700723730027676,
      "learning_rate": 1.3663104237618075e-07,
      "loss": 0.0011,
      "step": 290900
    },
    {
      "epoch": 1.8658063771313491,
      "grad_norm": 0.01567370444536209,
      "learning_rate": 1.365011231561242e-07,
      "loss": 0.0013,
      "step": 290910
    },
    {
      "epoch": 1.8658705140251353,
      "grad_norm": 0.09104197472333908,
      "learning_rate": 1.3637126487918428e-07,
      "loss": 0.0009,
      "step": 290920
    },
    {
      "epoch": 1.8659346509189212,
      "grad_norm": 0.03871174529194832,
      "learning_rate": 1.3624146754698808e-07,
      "loss": 0.0009,
      "step": 290930
    },
    {
      "epoch": 1.8659987878127073,
      "grad_norm": 0.2143871784210205,
      "learning_rate": 1.3611173116116316e-07,
      "loss": 0.0012,
      "step": 290940
    },
    {
      "epoch": 1.8660629247064935,
      "grad_norm": 0.004860973916947842,
      "learning_rate": 1.3598205572333378e-07,
      "loss": 0.002,
      "step": 290950
    },
    {
      "epoch": 1.8661270616002796,
      "grad_norm": 0.039312057197093964,
      "learning_rate": 1.3585244123512587e-07,
      "loss": 0.0013,
      "step": 290960
    },
    {
      "epoch": 1.8661911984940658,
      "grad_norm": 0.040601156651973724,
      "learning_rate": 1.357228876981631e-07,
      "loss": 0.0006,
      "step": 290970
    },
    {
      "epoch": 1.866255335387852,
      "grad_norm": 0.06531794369220734,
      "learning_rate": 1.3559339511406865e-07,
      "loss": 0.0008,
      "step": 290980
    },
    {
      "epoch": 1.866319472281638,
      "grad_norm": 0.07218613475561142,
      "learning_rate": 1.354639634844662e-07,
      "loss": 0.0007,
      "step": 290990
    },
    {
      "epoch": 1.866383609175424,
      "grad_norm": 0.03746568039059639,
      "learning_rate": 1.3533459281097673e-07,
      "loss": 0.0003,
      "step": 291000
    },
    {
      "epoch": 1.8664477460692102,
      "grad_norm": 0.05084928497672081,
      "learning_rate": 1.3520528309522108e-07,
      "loss": 0.0006,
      "step": 291010
    },
    {
      "epoch": 1.866511882962996,
      "grad_norm": 0.14735764265060425,
      "learning_rate": 1.3507603433882022e-07,
      "loss": 0.0015,
      "step": 291020
    },
    {
      "epoch": 1.8665760198567822,
      "grad_norm": 0.03813185170292854,
      "learning_rate": 1.3494684654339397e-07,
      "loss": 0.0011,
      "step": 291030
    },
    {
      "epoch": 1.8666401567505684,
      "grad_norm": 0.2427544891834259,
      "learning_rate": 1.348177197105599e-07,
      "loss": 0.0018,
      "step": 291040
    },
    {
      "epoch": 1.8667042936443545,
      "grad_norm": 0.027156217023730278,
      "learning_rate": 1.346886538419373e-07,
      "loss": 0.0003,
      "step": 291050
    },
    {
      "epoch": 1.8667684305381407,
      "grad_norm": 0.0016352023230865598,
      "learning_rate": 1.3455964893914376e-07,
      "loss": 0.0005,
      "step": 291060
    },
    {
      "epoch": 1.8668325674319268,
      "grad_norm": 0.3087085783481598,
      "learning_rate": 1.3443070500379464e-07,
      "loss": 0.0027,
      "step": 291070
    },
    {
      "epoch": 1.8668967043257128,
      "grad_norm": 0.19220387935638428,
      "learning_rate": 1.343018220375053e-07,
      "loss": 0.0016,
      "step": 291080
    },
    {
      "epoch": 1.866960841219499,
      "grad_norm": 0.06612735986709595,
      "learning_rate": 1.3417300004189226e-07,
      "loss": 0.0005,
      "step": 291090
    },
    {
      "epoch": 1.867024978113285,
      "grad_norm": 0.019334839656949043,
      "learning_rate": 1.340442390185692e-07,
      "loss": 0.0011,
      "step": 291100
    },
    {
      "epoch": 1.867089115007071,
      "grad_norm": 0.049330078065395355,
      "learning_rate": 1.3391553896914933e-07,
      "loss": 0.0007,
      "step": 291110
    },
    {
      "epoch": 1.8671532519008571,
      "grad_norm": 0.05242971330881119,
      "learning_rate": 1.3378689989524573e-07,
      "loss": 0.0004,
      "step": 291120
    },
    {
      "epoch": 1.8672173887946433,
      "grad_norm": 0.18560273945331573,
      "learning_rate": 1.336583217984694e-07,
      "loss": 0.0011,
      "step": 291130
    },
    {
      "epoch": 1.8672815256884294,
      "grad_norm": 0.03429925814270973,
      "learning_rate": 1.3352980468043285e-07,
      "loss": 0.0018,
      "step": 291140
    },
    {
      "epoch": 1.8673456625822156,
      "grad_norm": 0.03484424576163292,
      "learning_rate": 1.33401348542746e-07,
      "loss": 0.0009,
      "step": 291150
    },
    {
      "epoch": 1.8674097994760017,
      "grad_norm": 0.015022116713225842,
      "learning_rate": 1.3327295338701806e-07,
      "loss": 0.0011,
      "step": 291160
    },
    {
      "epoch": 1.8674739363697876,
      "grad_norm": 0.1105656549334526,
      "learning_rate": 1.3314461921485834e-07,
      "loss": 0.0007,
      "step": 291170
    },
    {
      "epoch": 1.8675380732635738,
      "grad_norm": 0.03682788088917732,
      "learning_rate": 1.3301634602787494e-07,
      "loss": 0.0019,
      "step": 291180
    },
    {
      "epoch": 1.8676022101573597,
      "grad_norm": 0.0016587432473897934,
      "learning_rate": 1.3288813382767496e-07,
      "loss": 0.001,
      "step": 291190
    },
    {
      "epoch": 1.8676663470511459,
      "grad_norm": 0.0696180984377861,
      "learning_rate": 1.3275998261586486e-07,
      "loss": 0.0011,
      "step": 291200
    },
    {
      "epoch": 1.867730483944932,
      "grad_norm": 0.019244924187660217,
      "learning_rate": 1.3263189239405116e-07,
      "loss": 0.0005,
      "step": 291210
    },
    {
      "epoch": 1.8677946208387182,
      "grad_norm": 0.02816018834710121,
      "learning_rate": 1.3250386316383813e-07,
      "loss": 0.0009,
      "step": 291220
    },
    {
      "epoch": 1.8678587577325043,
      "grad_norm": 0.03528784587979317,
      "learning_rate": 1.323758949268311e-07,
      "loss": 0.0011,
      "step": 291230
    },
    {
      "epoch": 1.8679228946262905,
      "grad_norm": 0.07470495998859406,
      "learning_rate": 1.322479876846322e-07,
      "loss": 0.0014,
      "step": 291240
    },
    {
      "epoch": 1.8679870315200766,
      "grad_norm": 0.020319901406764984,
      "learning_rate": 1.321201414388451e-07,
      "loss": 0.0017,
      "step": 291250
    },
    {
      "epoch": 1.8680511684138625,
      "grad_norm": 0.03774775192141533,
      "learning_rate": 1.3199235619107187e-07,
      "loss": 0.0008,
      "step": 291260
    },
    {
      "epoch": 1.8681153053076487,
      "grad_norm": 0.05965312942862511,
      "learning_rate": 1.3186463194291343e-07,
      "loss": 0.0029,
      "step": 291270
    },
    {
      "epoch": 1.8681794422014346,
      "grad_norm": 0.012127332389354706,
      "learning_rate": 1.3173696869597018e-07,
      "loss": 0.0014,
      "step": 291280
    },
    {
      "epoch": 1.8682435790952208,
      "grad_norm": 0.05256694555282593,
      "learning_rate": 1.3160936645184253e-07,
      "loss": 0.0012,
      "step": 291290
    },
    {
      "epoch": 1.868307715989007,
      "grad_norm": 0.04609822481870651,
      "learning_rate": 1.314818252121286e-07,
      "loss": 0.0013,
      "step": 291300
    },
    {
      "epoch": 1.868371852882793,
      "grad_norm": 0.07213814556598663,
      "learning_rate": 1.3135434497842658e-07,
      "loss": 0.0008,
      "step": 291310
    },
    {
      "epoch": 1.8684359897765792,
      "grad_norm": 0.017824608832597733,
      "learning_rate": 1.3122692575233464e-07,
      "loss": 0.0013,
      "step": 291320
    },
    {
      "epoch": 1.8685001266703654,
      "grad_norm": 0.12118804454803467,
      "learning_rate": 1.3109956753544872e-07,
      "loss": 0.0006,
      "step": 291330
    },
    {
      "epoch": 1.8685642635641513,
      "grad_norm": 0.11525218933820724,
      "learning_rate": 1.3097227032936534e-07,
      "loss": 0.0019,
      "step": 291340
    },
    {
      "epoch": 1.8686284004579374,
      "grad_norm": 0.02752411551773548,
      "learning_rate": 1.3084503413567874e-07,
      "loss": 0.0009,
      "step": 291350
    },
    {
      "epoch": 1.8686925373517234,
      "grad_norm": 0.007287966553121805,
      "learning_rate": 1.3071785895598376e-07,
      "loss": 0.0011,
      "step": 291360
    },
    {
      "epoch": 1.8687566742455095,
      "grad_norm": 0.028855368494987488,
      "learning_rate": 1.3059074479187472e-07,
      "loss": 0.0017,
      "step": 291370
    },
    {
      "epoch": 1.8688208111392957,
      "grad_norm": 0.09461076557636261,
      "learning_rate": 1.304636916449431e-07,
      "loss": 0.002,
      "step": 291380
    },
    {
      "epoch": 1.8688849480330818,
      "grad_norm": 0.08592013269662857,
      "learning_rate": 1.3033669951678153e-07,
      "loss": 0.0013,
      "step": 291390
    },
    {
      "epoch": 1.868949084926868,
      "grad_norm": 0.15071649849414825,
      "learning_rate": 1.3020976840898203e-07,
      "loss": 0.0009,
      "step": 291400
    },
    {
      "epoch": 1.869013221820654,
      "grad_norm": 0.039613980799913406,
      "learning_rate": 1.3008289832313448e-07,
      "loss": 0.0008,
      "step": 291410
    },
    {
      "epoch": 1.8690773587144403,
      "grad_norm": 0.041813723742961884,
      "learning_rate": 1.299560892608287e-07,
      "loss": 0.0016,
      "step": 291420
    },
    {
      "epoch": 1.8691414956082262,
      "grad_norm": 0.01769079640507698,
      "learning_rate": 1.2982934122365287e-07,
      "loss": 0.0004,
      "step": 291430
    },
    {
      "epoch": 1.8692056325020123,
      "grad_norm": 0.14418870210647583,
      "learning_rate": 1.2970265421319684e-07,
      "loss": 0.0013,
      "step": 291440
    },
    {
      "epoch": 1.8692697693957983,
      "grad_norm": 0.08726982027292252,
      "learning_rate": 1.295760282310471e-07,
      "loss": 0.0013,
      "step": 291450
    },
    {
      "epoch": 1.8693339062895844,
      "grad_norm": 0.03900137543678284,
      "learning_rate": 1.2944946327879014e-07,
      "loss": 0.0009,
      "step": 291460
    },
    {
      "epoch": 1.8693980431833706,
      "grad_norm": 0.044279251247644424,
      "learning_rate": 1.293229593580131e-07,
      "loss": 0.0004,
      "step": 291470
    },
    {
      "epoch": 1.8694621800771567,
      "grad_norm": 0.06156221777200699,
      "learning_rate": 1.2919651647030017e-07,
      "loss": 0.0007,
      "step": 291480
    },
    {
      "epoch": 1.8695263169709428,
      "grad_norm": 0.01173117384314537,
      "learning_rate": 1.2907013461723573e-07,
      "loss": 0.001,
      "step": 291490
    },
    {
      "epoch": 1.869590453864729,
      "grad_norm": 0.0110355569049716,
      "learning_rate": 1.2894381380040344e-07,
      "loss": 0.001,
      "step": 291500
    },
    {
      "epoch": 1.869654590758515,
      "grad_norm": 0.11326061934232712,
      "learning_rate": 1.2881755402138706e-07,
      "loss": 0.001,
      "step": 291510
    },
    {
      "epoch": 1.869718727652301,
      "grad_norm": 0.04712875187397003,
      "learning_rate": 1.2869135528176812e-07,
      "loss": 0.0009,
      "step": 291520
    },
    {
      "epoch": 1.8697828645460872,
      "grad_norm": 0.34151312708854675,
      "learning_rate": 1.2856521758312756e-07,
      "loss": 0.0021,
      "step": 291530
    },
    {
      "epoch": 1.8698470014398731,
      "grad_norm": 0.07753254473209381,
      "learning_rate": 1.2843914092704633e-07,
      "loss": 0.001,
      "step": 291540
    },
    {
      "epoch": 1.8699111383336593,
      "grad_norm": 0.1782429963350296,
      "learning_rate": 1.2831312531510488e-07,
      "loss": 0.0013,
      "step": 291550
    },
    {
      "epoch": 1.8699752752274454,
      "grad_norm": 0.08619780838489532,
      "learning_rate": 1.2818717074888133e-07,
      "loss": 0.0011,
      "step": 291560
    },
    {
      "epoch": 1.8700394121212316,
      "grad_norm": 9.459776878356934,
      "learning_rate": 1.2806127722995443e-07,
      "loss": 0.0158,
      "step": 291570
    },
    {
      "epoch": 1.8701035490150177,
      "grad_norm": 0.07566432654857635,
      "learning_rate": 1.2793544475990127e-07,
      "loss": 0.0016,
      "step": 291580
    },
    {
      "epoch": 1.870167685908804,
      "grad_norm": 0.05713646113872528,
      "learning_rate": 1.2780967334030002e-07,
      "loss": 0.0006,
      "step": 291590
    },
    {
      "epoch": 1.8702318228025898,
      "grad_norm": 0.049163222312927246,
      "learning_rate": 1.27683962972725e-07,
      "loss": 0.0035,
      "step": 291600
    },
    {
      "epoch": 1.870295959696376,
      "grad_norm": 0.13116882741451263,
      "learning_rate": 1.2755831365875159e-07,
      "loss": 0.0006,
      "step": 291610
    },
    {
      "epoch": 1.870360096590162,
      "grad_norm": 0.10171689838171005,
      "learning_rate": 1.2743272539995576e-07,
      "loss": 0.0006,
      "step": 291620
    },
    {
      "epoch": 1.870424233483948,
      "grad_norm": 0.07907585054636002,
      "learning_rate": 1.2730719819790959e-07,
      "loss": 0.0023,
      "step": 291630
    },
    {
      "epoch": 1.8704883703777342,
      "grad_norm": 0.07575205713510513,
      "learning_rate": 1.2718173205418739e-07,
      "loss": 0.0009,
      "step": 291640
    },
    {
      "epoch": 1.8705525072715203,
      "grad_norm": 0.02001216448843479,
      "learning_rate": 1.2705632697035953e-07,
      "loss": 0.0009,
      "step": 291650
    },
    {
      "epoch": 1.8706166441653065,
      "grad_norm": 0.02850428968667984,
      "learning_rate": 1.2693098294799922e-07,
      "loss": 0.0011,
      "step": 291660
    },
    {
      "epoch": 1.8706807810590926,
      "grad_norm": 0.04260598495602608,
      "learning_rate": 1.2680569998867686e-07,
      "loss": 0.0019,
      "step": 291670
    },
    {
      "epoch": 1.8707449179528788,
      "grad_norm": 0.033002011477947235,
      "learning_rate": 1.266804780939612e-07,
      "loss": 0.0006,
      "step": 291680
    },
    {
      "epoch": 1.8708090548466647,
      "grad_norm": 0.022915298119187355,
      "learning_rate": 1.2655531726542214e-07,
      "loss": 0.0003,
      "step": 291690
    },
    {
      "epoch": 1.8708731917404509,
      "grad_norm": 0.2629983425140381,
      "learning_rate": 1.2643021750462837e-07,
      "loss": 0.0016,
      "step": 291700
    },
    {
      "epoch": 1.8709373286342368,
      "grad_norm": 0.11415312439203262,
      "learning_rate": 1.2630517881314697e-07,
      "loss": 0.0017,
      "step": 291710
    },
    {
      "epoch": 1.871001465528023,
      "grad_norm": 0.057738933712244034,
      "learning_rate": 1.2618020119254448e-07,
      "loss": 0.0008,
      "step": 291720
    },
    {
      "epoch": 1.871065602421809,
      "grad_norm": 0.00794194731861353,
      "learning_rate": 1.2605528464438743e-07,
      "loss": 0.001,
      "step": 291730
    },
    {
      "epoch": 1.8711297393155952,
      "grad_norm": 0.09125705808401108,
      "learning_rate": 1.2593042917024122e-07,
      "loss": 0.001,
      "step": 291740
    },
    {
      "epoch": 1.8711938762093814,
      "grad_norm": 0.06937754154205322,
      "learning_rate": 1.2580563477167018e-07,
      "loss": 0.0019,
      "step": 291750
    },
    {
      "epoch": 1.8712580131031675,
      "grad_norm": 0.01868765987455845,
      "learning_rate": 1.256809014502375e-07,
      "loss": 0.0014,
      "step": 291760
    },
    {
      "epoch": 1.8713221499969535,
      "grad_norm": 0.0665220245718956,
      "learning_rate": 1.2555622920750743e-07,
      "loss": 0.0009,
      "step": 291770
    },
    {
      "epoch": 1.8713862868907396,
      "grad_norm": 0.13614019751548767,
      "learning_rate": 1.2543161804504156e-07,
      "loss": 0.0016,
      "step": 291780
    },
    {
      "epoch": 1.8714504237845255,
      "grad_norm": 0.05044696480035782,
      "learning_rate": 1.2530706796440085e-07,
      "loss": 0.0007,
      "step": 291790
    },
    {
      "epoch": 1.8715145606783117,
      "grad_norm": 0.10428164899349213,
      "learning_rate": 1.2518257896714626e-07,
      "loss": 0.001,
      "step": 291800
    },
    {
      "epoch": 1.8715786975720978,
      "grad_norm": 0.06369077414274216,
      "learning_rate": 1.2505815105483876e-07,
      "loss": 0.0009,
      "step": 291810
    },
    {
      "epoch": 1.871642834465884,
      "grad_norm": 0.3190913200378418,
      "learning_rate": 1.2493378422903658e-07,
      "loss": 0.0013,
      "step": 291820
    },
    {
      "epoch": 1.8717069713596701,
      "grad_norm": 0.01415115687996149,
      "learning_rate": 1.2480947849129788e-07,
      "loss": 0.0015,
      "step": 291830
    },
    {
      "epoch": 1.8717711082534563,
      "grad_norm": 0.09167692065238953,
      "learning_rate": 1.2468523384318033e-07,
      "loss": 0.0008,
      "step": 291840
    },
    {
      "epoch": 1.8718352451472424,
      "grad_norm": 0.001591764623299241,
      "learning_rate": 1.24561050286241e-07,
      "loss": 0.0006,
      "step": 291850
    },
    {
      "epoch": 1.8718993820410283,
      "grad_norm": 0.034697454422712326,
      "learning_rate": 1.2443692782203697e-07,
      "loss": 0.002,
      "step": 291860
    },
    {
      "epoch": 1.8719635189348145,
      "grad_norm": 0.015665624290704727,
      "learning_rate": 1.243128664521226e-07,
      "loss": 0.0007,
      "step": 291870
    },
    {
      "epoch": 1.8720276558286004,
      "grad_norm": 0.08221752196550369,
      "learning_rate": 1.2418886617805214e-07,
      "loss": 0.0015,
      "step": 291880
    },
    {
      "epoch": 1.8720917927223866,
      "grad_norm": 0.2477082461118698,
      "learning_rate": 1.240649270013805e-07,
      "loss": 0.0016,
      "step": 291890
    },
    {
      "epoch": 1.8721559296161727,
      "grad_norm": 0.09488316625356674,
      "learning_rate": 1.2394104892365976e-07,
      "loss": 0.0027,
      "step": 291900
    },
    {
      "epoch": 1.8722200665099589,
      "grad_norm": 0.02428659237921238,
      "learning_rate": 1.2381723194644258e-07,
      "loss": 0.0011,
      "step": 291910
    },
    {
      "epoch": 1.872284203403745,
      "grad_norm": 0.06166722625494003,
      "learning_rate": 1.2369347607127991e-07,
      "loss": 0.0009,
      "step": 291920
    },
    {
      "epoch": 1.8723483402975312,
      "grad_norm": 0.005926318001002073,
      "learning_rate": 1.2356978129972387e-07,
      "loss": 0.001,
      "step": 291930
    },
    {
      "epoch": 1.872412477191317,
      "grad_norm": 0.09185931086540222,
      "learning_rate": 1.2344614763332374e-07,
      "loss": 0.0015,
      "step": 291940
    },
    {
      "epoch": 1.8724766140851032,
      "grad_norm": 0.059110403060913086,
      "learning_rate": 1.2332257507362778e-07,
      "loss": 0.0015,
      "step": 291950
    },
    {
      "epoch": 1.8725407509788894,
      "grad_norm": 0.06216234341263771,
      "learning_rate": 1.231990636221858e-07,
      "loss": 0.0025,
      "step": 291960
    },
    {
      "epoch": 1.8726048878726753,
      "grad_norm": 0.0533917061984539,
      "learning_rate": 1.2307561328054441e-07,
      "loss": 0.001,
      "step": 291970
    },
    {
      "epoch": 1.8726690247664615,
      "grad_norm": 0.11403600871562958,
      "learning_rate": 1.2295222405025176e-07,
      "loss": 0.001,
      "step": 291980
    },
    {
      "epoch": 1.8727331616602476,
      "grad_norm": 0.05329390987753868,
      "learning_rate": 1.228288959328533e-07,
      "loss": 0.001,
      "step": 291990
    },
    {
      "epoch": 1.8727972985540338,
      "grad_norm": 0.008496430702507496,
      "learning_rate": 1.227056289298939e-07,
      "loss": 0.0015,
      "step": 292000
    },
    {
      "epoch": 1.87286143544782,
      "grad_norm": 0.10404878854751587,
      "learning_rate": 1.22582423042919e-07,
      "loss": 0.0013,
      "step": 292010
    },
    {
      "epoch": 1.872925572341606,
      "grad_norm": 0.018436910584568977,
      "learning_rate": 1.2245927827347237e-07,
      "loss": 0.0011,
      "step": 292020
    },
    {
      "epoch": 1.872989709235392,
      "grad_norm": 0.014300175942480564,
      "learning_rate": 1.223361946230961e-07,
      "loss": 0.001,
      "step": 292030
    },
    {
      "epoch": 1.8730538461291781,
      "grad_norm": 0.09425853192806244,
      "learning_rate": 1.2221317209333395e-07,
      "loss": 0.0006,
      "step": 292040
    },
    {
      "epoch": 1.873117983022964,
      "grad_norm": 0.016019126400351524,
      "learning_rate": 1.220902106857269e-07,
      "loss": 0.0005,
      "step": 292050
    },
    {
      "epoch": 1.8731821199167502,
      "grad_norm": 0.07877396792173386,
      "learning_rate": 1.2196731040181598e-07,
      "loss": 0.0012,
      "step": 292060
    },
    {
      "epoch": 1.8732462568105364,
      "grad_norm": 0.2555936872959137,
      "learning_rate": 1.218444712431399e-07,
      "loss": 0.0014,
      "step": 292070
    },
    {
      "epoch": 1.8733103937043225,
      "grad_norm": 0.07026442885398865,
      "learning_rate": 1.2172169321123973e-07,
      "loss": 0.0007,
      "step": 292080
    },
    {
      "epoch": 1.8733745305981087,
      "grad_norm": 0.01781727746129036,
      "learning_rate": 1.215989763076536e-07,
      "loss": 0.0003,
      "step": 292090
    },
    {
      "epoch": 1.8734386674918948,
      "grad_norm": 0.03243421018123627,
      "learning_rate": 1.214763205339181e-07,
      "loss": 0.0013,
      "step": 292100
    },
    {
      "epoch": 1.873502804385681,
      "grad_norm": 0.19241560995578766,
      "learning_rate": 1.2135372589157092e-07,
      "loss": 0.0015,
      "step": 292110
    },
    {
      "epoch": 1.8735669412794669,
      "grad_norm": 0.03661670535802841,
      "learning_rate": 1.2123119238214854e-07,
      "loss": 0.0029,
      "step": 292120
    },
    {
      "epoch": 1.873631078173253,
      "grad_norm": 0.024726325646042824,
      "learning_rate": 1.2110872000718643e-07,
      "loss": 0.0006,
      "step": 292130
    },
    {
      "epoch": 1.873695215067039,
      "grad_norm": 0.04537238925695419,
      "learning_rate": 1.2098630876821894e-07,
      "loss": 0.0023,
      "step": 292140
    },
    {
      "epoch": 1.873759351960825,
      "grad_norm": 0.14653237164020538,
      "learning_rate": 1.2086395866677925e-07,
      "loss": 0.0006,
      "step": 292150
    },
    {
      "epoch": 1.8738234888546113,
      "grad_norm": 0.03436683490872383,
      "learning_rate": 1.207416697044017e-07,
      "loss": 0.002,
      "step": 292160
    },
    {
      "epoch": 1.8738876257483974,
      "grad_norm": 0.11190580576658249,
      "learning_rate": 1.2061944188261787e-07,
      "loss": 0.0009,
      "step": 292170
    },
    {
      "epoch": 1.8739517626421835,
      "grad_norm": 0.013377520255744457,
      "learning_rate": 1.2049727520295983e-07,
      "loss": 0.0005,
      "step": 292180
    },
    {
      "epoch": 1.8740158995359697,
      "grad_norm": 0.0461856946349144,
      "learning_rate": 1.203751696669586e-07,
      "loss": 0.0012,
      "step": 292190
    },
    {
      "epoch": 1.8740800364297556,
      "grad_norm": 0.09250977635383606,
      "learning_rate": 1.2025312527614353e-07,
      "loss": 0.0007,
      "step": 292200
    },
    {
      "epoch": 1.8741441733235418,
      "grad_norm": 0.04103543981909752,
      "learning_rate": 1.2013114203204447e-07,
      "loss": 0.0005,
      "step": 292210
    },
    {
      "epoch": 1.8742083102173277,
      "grad_norm": 0.003423841670155525,
      "learning_rate": 1.2000921993618962e-07,
      "loss": 0.0014,
      "step": 292220
    },
    {
      "epoch": 1.8742724471111138,
      "grad_norm": 0.0074290018528699875,
      "learning_rate": 1.198873589901073e-07,
      "loss": 0.0031,
      "step": 292230
    },
    {
      "epoch": 1.8743365840049,
      "grad_norm": 0.016086000949144363,
      "learning_rate": 1.1976555919532395e-07,
      "loss": 0.001,
      "step": 292240
    },
    {
      "epoch": 1.8744007208986861,
      "grad_norm": 0.037323419004678726,
      "learning_rate": 1.1964382055336566e-07,
      "loss": 0.0012,
      "step": 292250
    },
    {
      "epoch": 1.8744648577924723,
      "grad_norm": 0.047648780047893524,
      "learning_rate": 1.195221430657584e-07,
      "loss": 0.0012,
      "step": 292260
    },
    {
      "epoch": 1.8745289946862584,
      "grad_norm": 0.06828763335943222,
      "learning_rate": 1.1940052673402703e-07,
      "loss": 0.0016,
      "step": 292270
    },
    {
      "epoch": 1.8745931315800446,
      "grad_norm": 0.019493915140628815,
      "learning_rate": 1.192789715596948e-07,
      "loss": 0.0007,
      "step": 292280
    },
    {
      "epoch": 1.8746572684738305,
      "grad_norm": 0.10813542455434799,
      "learning_rate": 1.1915747754428553e-07,
      "loss": 0.0006,
      "step": 292290
    },
    {
      "epoch": 1.8747214053676167,
      "grad_norm": 0.15127013623714447,
      "learning_rate": 1.1903604468932128e-07,
      "loss": 0.0081,
      "step": 292300
    },
    {
      "epoch": 1.8747855422614026,
      "grad_norm": 0.23425142467021942,
      "learning_rate": 1.1891467299632365e-07,
      "loss": 0.0044,
      "step": 292310
    },
    {
      "epoch": 1.8748496791551887,
      "grad_norm": 0.0018174074357375503,
      "learning_rate": 1.1879336246681361e-07,
      "loss": 0.0017,
      "step": 292320
    },
    {
      "epoch": 1.874913816048975,
      "grad_norm": 0.06480545550584793,
      "learning_rate": 1.1867211310231108e-07,
      "loss": 0.0006,
      "step": 292330
    },
    {
      "epoch": 1.874977952942761,
      "grad_norm": 0.13981083035469055,
      "learning_rate": 1.1855092490433595e-07,
      "loss": 0.0009,
      "step": 292340
    },
    {
      "epoch": 1.8750420898365472,
      "grad_norm": 0.14205895364284515,
      "learning_rate": 1.1842979787440645e-07,
      "loss": 0.001,
      "step": 292350
    },
    {
      "epoch": 1.8751062267303333,
      "grad_norm": 0.023094238713383675,
      "learning_rate": 1.183087320140408e-07,
      "loss": 0.0009,
      "step": 292360
    },
    {
      "epoch": 1.8751703636241195,
      "grad_norm": 0.11857487261295319,
      "learning_rate": 1.1818772732475447e-07,
      "loss": 0.0016,
      "step": 292370
    },
    {
      "epoch": 1.8752345005179054,
      "grad_norm": 0.0800149217247963,
      "learning_rate": 1.1806678380806569e-07,
      "loss": 0.0011,
      "step": 292380
    },
    {
      "epoch": 1.8752986374116916,
      "grad_norm": 0.14757139980793,
      "learning_rate": 1.1794590146548934e-07,
      "loss": 0.0024,
      "step": 292390
    },
    {
      "epoch": 1.8753627743054775,
      "grad_norm": 0.13186165690422058,
      "learning_rate": 1.1782508029853979e-07,
      "loss": 0.0009,
      "step": 292400
    },
    {
      "epoch": 1.8754269111992636,
      "grad_norm": 0.010632836259901524,
      "learning_rate": 1.177043203087308e-07,
      "loss": 0.0006,
      "step": 292410
    },
    {
      "epoch": 1.8754910480930498,
      "grad_norm": 0.07655154913663864,
      "learning_rate": 1.1758362149757674e-07,
      "loss": 0.0013,
      "step": 292420
    },
    {
      "epoch": 1.875555184986836,
      "grad_norm": 0.060383353382349014,
      "learning_rate": 1.1746298386658917e-07,
      "loss": 0.0007,
      "step": 292430
    },
    {
      "epoch": 1.875619321880622,
      "grad_norm": 0.3088397681713104,
      "learning_rate": 1.173424074172791e-07,
      "loss": 0.0011,
      "step": 292440
    },
    {
      "epoch": 1.8756834587744082,
      "grad_norm": 0.09891100227832794,
      "learning_rate": 1.172218921511592e-07,
      "loss": 0.0023,
      "step": 292450
    },
    {
      "epoch": 1.8757475956681942,
      "grad_norm": 0.07089928537607193,
      "learning_rate": 1.1710143806973829e-07,
      "loss": 0.0023,
      "step": 292460
    },
    {
      "epoch": 1.8758117325619803,
      "grad_norm": 0.07698842138051987,
      "learning_rate": 1.1698104517452624e-07,
      "loss": 0.001,
      "step": 292470
    },
    {
      "epoch": 1.8758758694557662,
      "grad_norm": 0.1335075944662094,
      "learning_rate": 1.1686071346703076e-07,
      "loss": 0.0014,
      "step": 292480
    },
    {
      "epoch": 1.8759400063495524,
      "grad_norm": 0.07194697111845016,
      "learning_rate": 1.167404429487612e-07,
      "loss": 0.0018,
      "step": 292490
    },
    {
      "epoch": 1.8760041432433385,
      "grad_norm": 0.030050817877054214,
      "learning_rate": 1.166202336212241e-07,
      "loss": 0.0008,
      "step": 292500
    },
    {
      "epoch": 1.8760682801371247,
      "grad_norm": 0.052116841077804565,
      "learning_rate": 1.1650008548592551e-07,
      "loss": 0.0013,
      "step": 292510
    },
    {
      "epoch": 1.8761324170309108,
      "grad_norm": 0.015215110033750534,
      "learning_rate": 1.1637999854437032e-07,
      "loss": 0.0006,
      "step": 292520
    },
    {
      "epoch": 1.876196553924697,
      "grad_norm": 0.014021230861544609,
      "learning_rate": 1.1625997279806456e-07,
      "loss": 0.0014,
      "step": 292530
    },
    {
      "epoch": 1.8762606908184831,
      "grad_norm": 0.048613812774419785,
      "learning_rate": 1.1614000824851201e-07,
      "loss": 0.0014,
      "step": 292540
    },
    {
      "epoch": 1.876324827712269,
      "grad_norm": 0.06975982338190079,
      "learning_rate": 1.1602010489721538e-07,
      "loss": 0.0019,
      "step": 292550
    },
    {
      "epoch": 1.8763889646060552,
      "grad_norm": 0.021208981052041054,
      "learning_rate": 1.1590026274567678e-07,
      "loss": 0.001,
      "step": 292560
    },
    {
      "epoch": 1.8764531014998411,
      "grad_norm": 0.06615134328603745,
      "learning_rate": 1.1578048179539891e-07,
      "loss": 0.0011,
      "step": 292570
    },
    {
      "epoch": 1.8765172383936273,
      "grad_norm": 0.11063316464424133,
      "learning_rate": 1.1566076204788223e-07,
      "loss": 0.0007,
      "step": 292580
    },
    {
      "epoch": 1.8765813752874134,
      "grad_norm": 0.06497619301080704,
      "learning_rate": 1.1554110350462722e-07,
      "loss": 0.0016,
      "step": 292590
    },
    {
      "epoch": 1.8766455121811996,
      "grad_norm": 0.023190032690763474,
      "learning_rate": 1.1542150616713266e-07,
      "loss": 0.0011,
      "step": 292600
    },
    {
      "epoch": 1.8767096490749857,
      "grad_norm": 0.09450613707304001,
      "learning_rate": 1.1530197003689736e-07,
      "loss": 0.0009,
      "step": 292610
    },
    {
      "epoch": 1.8767737859687719,
      "grad_norm": 0.0028610366862267256,
      "learning_rate": 1.1518249511542012e-07,
      "loss": 0.0015,
      "step": 292620
    },
    {
      "epoch": 1.8768379228625578,
      "grad_norm": 0.08985843509435654,
      "learning_rate": 1.1506308140419752e-07,
      "loss": 0.0012,
      "step": 292630
    },
    {
      "epoch": 1.876902059756344,
      "grad_norm": 0.429986834526062,
      "learning_rate": 1.1494372890472506e-07,
      "loss": 0.0013,
      "step": 292640
    },
    {
      "epoch": 1.87696619665013,
      "grad_norm": 0.061802733689546585,
      "learning_rate": 1.1482443761849926e-07,
      "loss": 0.0006,
      "step": 292650
    },
    {
      "epoch": 1.877030333543916,
      "grad_norm": 0.10510675609111786,
      "learning_rate": 1.1470520754701453e-07,
      "loss": 0.0016,
      "step": 292660
    },
    {
      "epoch": 1.8770944704377022,
      "grad_norm": 0.006245863623917103,
      "learning_rate": 1.1458603869176521e-07,
      "loss": 0.0005,
      "step": 292670
    },
    {
      "epoch": 1.8771586073314883,
      "grad_norm": 0.030436299741268158,
      "learning_rate": 1.14466931054244e-07,
      "loss": 0.0007,
      "step": 292680
    },
    {
      "epoch": 1.8772227442252745,
      "grad_norm": 0.040905144065618515,
      "learning_rate": 1.1434788463594415e-07,
      "loss": 0.001,
      "step": 292690
    },
    {
      "epoch": 1.8772868811190606,
      "grad_norm": 0.013966300524771214,
      "learning_rate": 1.1422889943835725e-07,
      "loss": 0.0007,
      "step": 292700
    },
    {
      "epoch": 1.8773510180128468,
      "grad_norm": 0.11456148326396942,
      "learning_rate": 1.141099754629732e-07,
      "loss": 0.0007,
      "step": 292710
    },
    {
      "epoch": 1.8774151549066327,
      "grad_norm": 0.020847106352448463,
      "learning_rate": 1.139911127112836e-07,
      "loss": 0.0006,
      "step": 292720
    },
    {
      "epoch": 1.8774792918004188,
      "grad_norm": 0.1931428462266922,
      "learning_rate": 1.1387231118477782e-07,
      "loss": 0.0013,
      "step": 292730
    },
    {
      "epoch": 1.8775434286942048,
      "grad_norm": 0.08069287240505219,
      "learning_rate": 1.1375357088494354e-07,
      "loss": 0.0012,
      "step": 292740
    },
    {
      "epoch": 1.877607565587991,
      "grad_norm": 0.021612413227558136,
      "learning_rate": 1.1363489181326848e-07,
      "loss": 0.0006,
      "step": 292750
    },
    {
      "epoch": 1.877671702481777,
      "grad_norm": 0.016337022185325623,
      "learning_rate": 1.1351627397124143e-07,
      "loss": 0.001,
      "step": 292760
    },
    {
      "epoch": 1.8777358393755632,
      "grad_norm": 0.2647327780723572,
      "learning_rate": 1.1339771736034788e-07,
      "loss": 0.0013,
      "step": 292770
    },
    {
      "epoch": 1.8777999762693494,
      "grad_norm": 0.046598881483078,
      "learning_rate": 1.1327922198207275e-07,
      "loss": 0.0015,
      "step": 292780
    },
    {
      "epoch": 1.8778641131631355,
      "grad_norm": 0.05295758694410324,
      "learning_rate": 1.1316078783790152e-07,
      "loss": 0.0008,
      "step": 292790
    },
    {
      "epoch": 1.8779282500569217,
      "grad_norm": 0.05909512937068939,
      "learning_rate": 1.1304241492931855e-07,
      "loss": 0.0006,
      "step": 292800
    },
    {
      "epoch": 1.8779923869507076,
      "grad_norm": 0.06993626803159714,
      "learning_rate": 1.1292410325780656e-07,
      "loss": 0.0009,
      "step": 292810
    },
    {
      "epoch": 1.8780565238444937,
      "grad_norm": 0.03092687390744686,
      "learning_rate": 1.1280585282484824e-07,
      "loss": 0.0021,
      "step": 292820
    },
    {
      "epoch": 1.8781206607382797,
      "grad_norm": 0.20055820047855377,
      "learning_rate": 1.1268766363192518e-07,
      "loss": 0.0016,
      "step": 292830
    },
    {
      "epoch": 1.8781847976320658,
      "grad_norm": 0.019928058609366417,
      "learning_rate": 1.1256953568051898e-07,
      "loss": 0.0007,
      "step": 292840
    },
    {
      "epoch": 1.878248934525852,
      "grad_norm": 0.05944271385669708,
      "learning_rate": 1.1245146897210957e-07,
      "loss": 0.0011,
      "step": 292850
    },
    {
      "epoch": 1.878313071419638,
      "grad_norm": 0.002654408337548375,
      "learning_rate": 1.1233346350817575e-07,
      "loss": 0.0015,
      "step": 292860
    },
    {
      "epoch": 1.8783772083134243,
      "grad_norm": 0.04330388456583023,
      "learning_rate": 1.1221551929019748e-07,
      "loss": 0.0004,
      "step": 292870
    },
    {
      "epoch": 1.8784413452072104,
      "grad_norm": 0.02954934909939766,
      "learning_rate": 1.1209763631965187e-07,
      "loss": 0.0015,
      "step": 292880
    },
    {
      "epoch": 1.8785054821009963,
      "grad_norm": 0.007893679663538933,
      "learning_rate": 1.119798145980161e-07,
      "loss": 0.001,
      "step": 292890
    },
    {
      "epoch": 1.8785696189947825,
      "grad_norm": 0.014690901152789593,
      "learning_rate": 1.118620541267662e-07,
      "loss": 0.0013,
      "step": 292900
    },
    {
      "epoch": 1.8786337558885684,
      "grad_norm": 0.061036258935928345,
      "learning_rate": 1.1174435490737879e-07,
      "loss": 0.0013,
      "step": 292910
    },
    {
      "epoch": 1.8786978927823546,
      "grad_norm": 0.049692727625370026,
      "learning_rate": 1.116267169413282e-07,
      "loss": 0.0016,
      "step": 292920
    },
    {
      "epoch": 1.8787620296761407,
      "grad_norm": 0.046990640461444855,
      "learning_rate": 1.115091402300883e-07,
      "loss": 0.0004,
      "step": 292930
    },
    {
      "epoch": 1.8788261665699268,
      "grad_norm": 0.13081590831279755,
      "learning_rate": 1.1139162477513233e-07,
      "loss": 0.0012,
      "step": 292940
    },
    {
      "epoch": 1.878890303463713,
      "grad_norm": 0.039909522980451584,
      "learning_rate": 1.112741705779341e-07,
      "loss": 0.001,
      "step": 292950
    },
    {
      "epoch": 1.8789544403574991,
      "grad_norm": 0.07774998992681503,
      "learning_rate": 1.1115677763996357e-07,
      "loss": 0.0013,
      "step": 292960
    },
    {
      "epoch": 1.8790185772512853,
      "grad_norm": 0.04683222994208336,
      "learning_rate": 1.1103944596269345e-07,
      "loss": 0.0007,
      "step": 292970
    },
    {
      "epoch": 1.8790827141450712,
      "grad_norm": 0.11634768545627594,
      "learning_rate": 1.10922175547592e-07,
      "loss": 0.0007,
      "step": 292980
    },
    {
      "epoch": 1.8791468510388574,
      "grad_norm": 0.04241019859910011,
      "learning_rate": 1.1080496639613081e-07,
      "loss": 0.0005,
      "step": 292990
    },
    {
      "epoch": 1.8792109879326433,
      "grad_norm": 0.02268884889781475,
      "learning_rate": 1.1068781850977706e-07,
      "loss": 0.0008,
      "step": 293000
    },
    {
      "epoch": 1.8792751248264294,
      "grad_norm": 0.14664621651172638,
      "learning_rate": 1.1057073188999956e-07,
      "loss": 0.0011,
      "step": 293010
    },
    {
      "epoch": 1.8793392617202156,
      "grad_norm": 0.031727563589811325,
      "learning_rate": 1.1045370653826492e-07,
      "loss": 0.0004,
      "step": 293020
    },
    {
      "epoch": 1.8794033986140017,
      "grad_norm": 0.20842571556568146,
      "learning_rate": 1.1033674245604031e-07,
      "loss": 0.0017,
      "step": 293030
    },
    {
      "epoch": 1.879467535507788,
      "grad_norm": 0.03360797464847565,
      "learning_rate": 1.1021983964479122e-07,
      "loss": 0.0013,
      "step": 293040
    },
    {
      "epoch": 1.879531672401574,
      "grad_norm": 0.09648806601762772,
      "learning_rate": 1.101029981059809e-07,
      "loss": 0.001,
      "step": 293050
    },
    {
      "epoch": 1.87959580929536,
      "grad_norm": 0.05264410376548767,
      "learning_rate": 1.09986217841076e-07,
      "loss": 0.0003,
      "step": 293060
    },
    {
      "epoch": 1.8796599461891461,
      "grad_norm": 0.11550859361886978,
      "learning_rate": 1.0986949885153809e-07,
      "loss": 0.0004,
      "step": 293070
    },
    {
      "epoch": 1.8797240830829323,
      "grad_norm": 0.03810092434287071,
      "learning_rate": 1.09752841138831e-07,
      "loss": 0.0013,
      "step": 293080
    },
    {
      "epoch": 1.8797882199767182,
      "grad_norm": 0.11439960449934006,
      "learning_rate": 1.0963624470441469e-07,
      "loss": 0.0025,
      "step": 293090
    },
    {
      "epoch": 1.8798523568705043,
      "grad_norm": 0.05327378585934639,
      "learning_rate": 1.0951970954975188e-07,
      "loss": 0.001,
      "step": 293100
    },
    {
      "epoch": 1.8799164937642905,
      "grad_norm": 0.08586519211530685,
      "learning_rate": 1.0940323567630251e-07,
      "loss": 0.0008,
      "step": 293110
    },
    {
      "epoch": 1.8799806306580766,
      "grad_norm": 0.05707564949989319,
      "learning_rate": 1.092868230855254e-07,
      "loss": 0.0012,
      "step": 293120
    },
    {
      "epoch": 1.8800447675518628,
      "grad_norm": 0.2404910773038864,
      "learning_rate": 1.0917047177887997e-07,
      "loss": 0.0026,
      "step": 293130
    },
    {
      "epoch": 1.880108904445649,
      "grad_norm": 0.026872379705309868,
      "learning_rate": 1.090541817578239e-07,
      "loss": 0.0009,
      "step": 293140
    },
    {
      "epoch": 1.8801730413394349,
      "grad_norm": 0.037973977625370026,
      "learning_rate": 1.0893795302381438e-07,
      "loss": 0.0011,
      "step": 293150
    },
    {
      "epoch": 1.880237178233221,
      "grad_norm": 0.09424316138029099,
      "learning_rate": 1.0882178557830802e-07,
      "loss": 0.0017,
      "step": 293160
    },
    {
      "epoch": 1.880301315127007,
      "grad_norm": 0.08930182456970215,
      "learning_rate": 1.0870567942276033e-07,
      "loss": 0.0016,
      "step": 293170
    },
    {
      "epoch": 1.880365452020793,
      "grad_norm": 0.2537005543708801,
      "learning_rate": 1.0858963455862625e-07,
      "loss": 0.0017,
      "step": 293180
    },
    {
      "epoch": 1.8804295889145792,
      "grad_norm": 0.07175569981336594,
      "learning_rate": 1.0847365098736018e-07,
      "loss": 0.0011,
      "step": 293190
    },
    {
      "epoch": 1.8804937258083654,
      "grad_norm": 0.06721962243318558,
      "learning_rate": 1.0835772871041428e-07,
      "loss": 0.0015,
      "step": 293200
    },
    {
      "epoch": 1.8805578627021515,
      "grad_norm": 0.06867531687021255,
      "learning_rate": 1.0824186772924295e-07,
      "loss": 0.0009,
      "step": 293210
    },
    {
      "epoch": 1.8806219995959377,
      "grad_norm": 0.01514175534248352,
      "learning_rate": 1.0812606804529668e-07,
      "loss": 0.0007,
      "step": 293220
    },
    {
      "epoch": 1.8806861364897238,
      "grad_norm": 0.0059837414883077145,
      "learning_rate": 1.0801032966002712e-07,
      "loss": 0.0017,
      "step": 293230
    },
    {
      "epoch": 1.8807502733835098,
      "grad_norm": 0.25319117307662964,
      "learning_rate": 1.0789465257488418e-07,
      "loss": 0.0017,
      "step": 293240
    },
    {
      "epoch": 1.880814410277296,
      "grad_norm": 0.048505526036024094,
      "learning_rate": 1.0777903679131785e-07,
      "loss": 0.0004,
      "step": 293250
    },
    {
      "epoch": 1.8808785471710818,
      "grad_norm": 0.002642439678311348,
      "learning_rate": 1.0766348231077639e-07,
      "loss": 0.0004,
      "step": 293260
    },
    {
      "epoch": 1.880942684064868,
      "grad_norm": 0.003176505444571376,
      "learning_rate": 1.075479891347081e-07,
      "loss": 0.0004,
      "step": 293270
    },
    {
      "epoch": 1.8810068209586541,
      "grad_norm": 0.0015448590274900198,
      "learning_rate": 1.0743255726455959e-07,
      "loss": 0.001,
      "step": 293280
    },
    {
      "epoch": 1.8810709578524403,
      "grad_norm": 0.024752216413617134,
      "learning_rate": 1.073171867017786e-07,
      "loss": 0.0006,
      "step": 293290
    },
    {
      "epoch": 1.8811350947462264,
      "grad_norm": 0.38314008712768555,
      "learning_rate": 1.0720187744781008e-07,
      "loss": 0.002,
      "step": 293300
    },
    {
      "epoch": 1.8811992316400126,
      "grad_norm": 0.02777690440416336,
      "learning_rate": 1.0708662950409787e-07,
      "loss": 0.0005,
      "step": 293310
    },
    {
      "epoch": 1.8812633685337985,
      "grad_norm": 0.13420148193836212,
      "learning_rate": 1.0697144287208805e-07,
      "loss": 0.0011,
      "step": 293320
    },
    {
      "epoch": 1.8813275054275846,
      "grad_norm": 0.09384496510028839,
      "learning_rate": 1.0685631755322279e-07,
      "loss": 0.0008,
      "step": 293330
    },
    {
      "epoch": 1.8813916423213706,
      "grad_norm": 0.20899826288223267,
      "learning_rate": 1.0674125354894483e-07,
      "loss": 0.0035,
      "step": 293340
    },
    {
      "epoch": 1.8814557792151567,
      "grad_norm": 0.04801175370812416,
      "learning_rate": 1.0662625086069579e-07,
      "loss": 0.0006,
      "step": 293350
    },
    {
      "epoch": 1.8815199161089429,
      "grad_norm": 0.003630199935287237,
      "learning_rate": 1.0651130948991728e-07,
      "loss": 0.0009,
      "step": 293360
    },
    {
      "epoch": 1.881584053002729,
      "grad_norm": 0.09173569083213806,
      "learning_rate": 1.0639642943804984e-07,
      "loss": 0.0024,
      "step": 293370
    },
    {
      "epoch": 1.8816481898965152,
      "grad_norm": 0.07374726980924606,
      "learning_rate": 1.0628161070653175e-07,
      "loss": 0.0005,
      "step": 293380
    },
    {
      "epoch": 1.8817123267903013,
      "grad_norm": 0.05393248423933983,
      "learning_rate": 1.0616685329680354e-07,
      "loss": 0.0016,
      "step": 293390
    },
    {
      "epoch": 1.8817764636840875,
      "grad_norm": 0.05182025954127312,
      "learning_rate": 1.0605215721030126e-07,
      "loss": 0.0009,
      "step": 293400
    },
    {
      "epoch": 1.8818406005778734,
      "grad_norm": 0.08506152778863907,
      "learning_rate": 1.0593752244846378e-07,
      "loss": 0.0006,
      "step": 293410
    },
    {
      "epoch": 1.8819047374716595,
      "grad_norm": 0.09758177399635315,
      "learning_rate": 1.0582294901272661e-07,
      "loss": 0.0012,
      "step": 293420
    },
    {
      "epoch": 1.8819688743654455,
      "grad_norm": 0.03581222519278526,
      "learning_rate": 1.0570843690452526e-07,
      "loss": 0.0005,
      "step": 293430
    },
    {
      "epoch": 1.8820330112592316,
      "grad_norm": 0.2432924509048462,
      "learning_rate": 1.0559398612529581e-07,
      "loss": 0.0011,
      "step": 293440
    },
    {
      "epoch": 1.8820971481530178,
      "grad_norm": 0.15219354629516602,
      "learning_rate": 1.0547959667647157e-07,
      "loss": 0.0006,
      "step": 293450
    },
    {
      "epoch": 1.882161285046804,
      "grad_norm": 0.017942845821380615,
      "learning_rate": 1.0536526855948637e-07,
      "loss": 0.0008,
      "step": 293460
    },
    {
      "epoch": 1.88222542194059,
      "grad_norm": 0.045608971267938614,
      "learning_rate": 1.0525100177577185e-07,
      "loss": 0.0006,
      "step": 293470
    },
    {
      "epoch": 1.8822895588343762,
      "grad_norm": 0.28920862078666687,
      "learning_rate": 1.0513679632676077e-07,
      "loss": 0.0021,
      "step": 293480
    },
    {
      "epoch": 1.8823536957281621,
      "grad_norm": 0.05183442682027817,
      "learning_rate": 1.0502265221388419e-07,
      "loss": 0.0018,
      "step": 293490
    },
    {
      "epoch": 1.8824178326219483,
      "grad_norm": 0.015167763456702232,
      "learning_rate": 1.0490856943857153e-07,
      "loss": 0.0017,
      "step": 293500
    },
    {
      "epoch": 1.8824819695157344,
      "grad_norm": 0.18725934624671936,
      "learning_rate": 1.047945480022533e-07,
      "loss": 0.0009,
      "step": 293510
    },
    {
      "epoch": 1.8825461064095204,
      "grad_norm": 0.13862395286560059,
      "learning_rate": 1.0468058790635782e-07,
      "loss": 0.0017,
      "step": 293520
    },
    {
      "epoch": 1.8826102433033065,
      "grad_norm": 0.012817653827369213,
      "learning_rate": 1.0456668915231338e-07,
      "loss": 0.0004,
      "step": 293530
    },
    {
      "epoch": 1.8826743801970927,
      "grad_norm": 0.10296519845724106,
      "learning_rate": 1.0445285174154717e-07,
      "loss": 0.0011,
      "step": 293540
    },
    {
      "epoch": 1.8827385170908788,
      "grad_norm": 0.1026385948061943,
      "learning_rate": 1.043390756754853e-07,
      "loss": 0.0014,
      "step": 293550
    },
    {
      "epoch": 1.882802653984665,
      "grad_norm": 0.01244757603853941,
      "learning_rate": 1.0422536095555381e-07,
      "loss": 0.0022,
      "step": 293560
    },
    {
      "epoch": 1.882866790878451,
      "grad_norm": 0.0076684970408678055,
      "learning_rate": 1.0411170758317768e-07,
      "loss": 0.0016,
      "step": 293570
    },
    {
      "epoch": 1.882930927772237,
      "grad_norm": 0.125825434923172,
      "learning_rate": 1.0399811555978024e-07,
      "loss": 0.001,
      "step": 293580
    },
    {
      "epoch": 1.8829950646660232,
      "grad_norm": 0.02257315255701542,
      "learning_rate": 1.0388458488678588e-07,
      "loss": 0.0008,
      "step": 293590
    },
    {
      "epoch": 1.883059201559809,
      "grad_norm": 0.03127681463956833,
      "learning_rate": 1.0377111556561682e-07,
      "loss": 0.0056,
      "step": 293600
    },
    {
      "epoch": 1.8831233384535953,
      "grad_norm": 0.07549407333135605,
      "learning_rate": 1.0365770759769522e-07,
      "loss": 0.0017,
      "step": 293610
    },
    {
      "epoch": 1.8831874753473814,
      "grad_norm": 0.08330096304416656,
      "learning_rate": 1.0354436098444165e-07,
      "loss": 0.0006,
      "step": 293620
    },
    {
      "epoch": 1.8832516122411675,
      "grad_norm": 0.04965728893876076,
      "learning_rate": 1.0343107572727663e-07,
      "loss": 0.0004,
      "step": 293630
    },
    {
      "epoch": 1.8833157491349537,
      "grad_norm": 0.1543658822774887,
      "learning_rate": 1.033178518276201e-07,
      "loss": 0.0015,
      "step": 293640
    },
    {
      "epoch": 1.8833798860287398,
      "grad_norm": 0.013736913911998272,
      "learning_rate": 1.0320468928689043e-07,
      "loss": 0.0008,
      "step": 293650
    },
    {
      "epoch": 1.883444022922526,
      "grad_norm": 0.06732528656721115,
      "learning_rate": 1.0309158810650532e-07,
      "loss": 0.0004,
      "step": 293660
    },
    {
      "epoch": 1.883508159816312,
      "grad_norm": 0.02943151257932186,
      "learning_rate": 1.0297854828788311e-07,
      "loss": 0.0005,
      "step": 293670
    },
    {
      "epoch": 1.883572296710098,
      "grad_norm": 0.008330841548740864,
      "learning_rate": 1.0286556983243878e-07,
      "loss": 0.0008,
      "step": 293680
    },
    {
      "epoch": 1.883636433603884,
      "grad_norm": 0.02544136717915535,
      "learning_rate": 1.0275265274158952e-07,
      "loss": 0.001,
      "step": 293690
    },
    {
      "epoch": 1.8837005704976701,
      "grad_norm": 0.13496889173984528,
      "learning_rate": 1.0263979701674865e-07,
      "loss": 0.0014,
      "step": 293700
    },
    {
      "epoch": 1.8837647073914563,
      "grad_norm": 0.05403325334191322,
      "learning_rate": 1.0252700265933169e-07,
      "loss": 0.0009,
      "step": 293710
    },
    {
      "epoch": 1.8838288442852424,
      "grad_norm": 0.05685772746801376,
      "learning_rate": 1.024142696707514e-07,
      "loss": 0.0008,
      "step": 293720
    },
    {
      "epoch": 1.8838929811790286,
      "grad_norm": 0.010242694057524204,
      "learning_rate": 1.0230159805242058e-07,
      "loss": 0.0011,
      "step": 293730
    },
    {
      "epoch": 1.8839571180728147,
      "grad_norm": 0.050826091319322586,
      "learning_rate": 1.0218898780575138e-07,
      "loss": 0.0006,
      "step": 293740
    },
    {
      "epoch": 1.8840212549666007,
      "grad_norm": 0.10138603299856186,
      "learning_rate": 1.0207643893215435e-07,
      "loss": 0.0008,
      "step": 293750
    },
    {
      "epoch": 1.8840853918603868,
      "grad_norm": 0.06282013654708862,
      "learning_rate": 1.0196395143304006e-07,
      "loss": 0.003,
      "step": 293760
    },
    {
      "epoch": 1.8841495287541727,
      "grad_norm": 0.05487838387489319,
      "learning_rate": 1.018515253098179e-07,
      "loss": 0.0009,
      "step": 293770
    },
    {
      "epoch": 1.884213665647959,
      "grad_norm": 0.10411953926086426,
      "learning_rate": 1.0173916056389677e-07,
      "loss": 0.0011,
      "step": 293780
    },
    {
      "epoch": 1.884277802541745,
      "grad_norm": 0.167263925075531,
      "learning_rate": 1.0162685719668497e-07,
      "loss": 0.0021,
      "step": 293790
    },
    {
      "epoch": 1.8843419394355312,
      "grad_norm": 0.03700994327664375,
      "learning_rate": 1.0151461520958971e-07,
      "loss": 0.0011,
      "step": 293800
    },
    {
      "epoch": 1.8844060763293173,
      "grad_norm": 0.09405221045017242,
      "learning_rate": 1.01402434604016e-07,
      "loss": 0.0018,
      "step": 293810
    },
    {
      "epoch": 1.8844702132231035,
      "grad_norm": 0.01762114278972149,
      "learning_rate": 1.0129031538137213e-07,
      "loss": 0.0007,
      "step": 293820
    },
    {
      "epoch": 1.8845343501168896,
      "grad_norm": 0.16988490521907806,
      "learning_rate": 1.0117825754306088e-07,
      "loss": 0.002,
      "step": 293830
    },
    {
      "epoch": 1.8845984870106756,
      "grad_norm": 0.08364865928888321,
      "learning_rate": 1.0106626109048778e-07,
      "loss": 0.001,
      "step": 293840
    },
    {
      "epoch": 1.8846626239044617,
      "grad_norm": 0.0017609879141673446,
      "learning_rate": 1.0095432602505506e-07,
      "loss": 0.0005,
      "step": 293850
    },
    {
      "epoch": 1.8847267607982476,
      "grad_norm": 0.0023259019944816828,
      "learning_rate": 1.0084245234816603e-07,
      "loss": 0.0009,
      "step": 293860
    },
    {
      "epoch": 1.8847908976920338,
      "grad_norm": 0.013131055980920792,
      "learning_rate": 1.007306400612229e-07,
      "loss": 0.0006,
      "step": 293870
    },
    {
      "epoch": 1.88485503458582,
      "grad_norm": 0.056270454078912735,
      "learning_rate": 1.0061888916562568e-07,
      "loss": 0.0012,
      "step": 293880
    },
    {
      "epoch": 1.884919171479606,
      "grad_norm": 0.01898212917149067,
      "learning_rate": 1.0050719966277544e-07,
      "loss": 0.0007,
      "step": 293890
    },
    {
      "epoch": 1.8849833083733922,
      "grad_norm": 0.15191014111042023,
      "learning_rate": 1.003955715540722e-07,
      "loss": 0.0015,
      "step": 293900
    },
    {
      "epoch": 1.8850474452671784,
      "grad_norm": 0.0716463252902031,
      "learning_rate": 1.0028400484091372e-07,
      "loss": 0.0011,
      "step": 293910
    },
    {
      "epoch": 1.8851115821609645,
      "grad_norm": 0.06062595546245575,
      "learning_rate": 1.0017249952469776e-07,
      "loss": 0.0016,
      "step": 293920
    },
    {
      "epoch": 1.8851757190547505,
      "grad_norm": 0.09747257828712463,
      "learning_rate": 1.0006105560682322e-07,
      "loss": 0.0014,
      "step": 293930
    },
    {
      "epoch": 1.8852398559485366,
      "grad_norm": 0.022606564685702324,
      "learning_rate": 9.994967308868564e-08,
      "loss": 0.0007,
      "step": 293940
    },
    {
      "epoch": 1.8853039928423225,
      "grad_norm": 0.09071505814790726,
      "learning_rate": 9.983835197168001e-08,
      "loss": 0.0008,
      "step": 293950
    },
    {
      "epoch": 1.8853681297361087,
      "grad_norm": 0.016011979430913925,
      "learning_rate": 9.972709225720189e-08,
      "loss": 0.0008,
      "step": 293960
    },
    {
      "epoch": 1.8854322666298948,
      "grad_norm": 0.05072588101029396,
      "learning_rate": 9.961589394664628e-08,
      "loss": 0.0016,
      "step": 293970
    },
    {
      "epoch": 1.885496403523681,
      "grad_norm": 0.0029339883476495743,
      "learning_rate": 9.950475704140539e-08,
      "loss": 0.0016,
      "step": 293980
    },
    {
      "epoch": 1.8855605404174671,
      "grad_norm": 0.037333372980356216,
      "learning_rate": 9.939368154287198e-08,
      "loss": 0.0015,
      "step": 293990
    },
    {
      "epoch": 1.8856246773112533,
      "grad_norm": 0.013276482932269573,
      "learning_rate": 9.928266745243831e-08,
      "loss": 0.0017,
      "step": 294000
    },
    {
      "epoch": 1.8856888142050392,
      "grad_norm": 0.06680195778608322,
      "learning_rate": 9.91717147714949e-08,
      "loss": 0.0011,
      "step": 294010
    },
    {
      "epoch": 1.8857529510988253,
      "grad_norm": 0.06643390655517578,
      "learning_rate": 9.906082350143343e-08,
      "loss": 0.001,
      "step": 294020
    },
    {
      "epoch": 1.8858170879926113,
      "grad_norm": 0.08585356175899506,
      "learning_rate": 9.894999364364166e-08,
      "loss": 0.0011,
      "step": 294030
    },
    {
      "epoch": 1.8858812248863974,
      "grad_norm": 0.045736897736787796,
      "learning_rate": 9.883922519950961e-08,
      "loss": 0.0007,
      "step": 294040
    },
    {
      "epoch": 1.8859453617801836,
      "grad_norm": 0.06042035296559334,
      "learning_rate": 9.872851817042451e-08,
      "loss": 0.0006,
      "step": 294050
    },
    {
      "epoch": 1.8860094986739697,
      "grad_norm": 0.0751265287399292,
      "learning_rate": 9.861787255777411e-08,
      "loss": 0.0008,
      "step": 294060
    },
    {
      "epoch": 1.8860736355677559,
      "grad_norm": 0.04391670227050781,
      "learning_rate": 9.850728836294455e-08,
      "loss": 0.0008,
      "step": 294070
    },
    {
      "epoch": 1.886137772461542,
      "grad_norm": 0.11632103472948074,
      "learning_rate": 9.839676558732248e-08,
      "loss": 0.0012,
      "step": 294080
    },
    {
      "epoch": 1.8862019093553282,
      "grad_norm": 0.1448919028043747,
      "learning_rate": 9.828630423229124e-08,
      "loss": 0.002,
      "step": 294090
    },
    {
      "epoch": 1.886266046249114,
      "grad_norm": 0.021598802879452705,
      "learning_rate": 9.81759042992364e-08,
      "loss": 0.0008,
      "step": 294100
    },
    {
      "epoch": 1.8863301831429002,
      "grad_norm": 0.05400576442480087,
      "learning_rate": 9.806556578954019e-08,
      "loss": 0.0004,
      "step": 294110
    },
    {
      "epoch": 1.8863943200366862,
      "grad_norm": 0.1336008459329605,
      "learning_rate": 9.795528870458593e-08,
      "loss": 0.0009,
      "step": 294120
    },
    {
      "epoch": 1.8864584569304723,
      "grad_norm": 0.24765513837337494,
      "learning_rate": 9.784507304575586e-08,
      "loss": 0.0012,
      "step": 294130
    },
    {
      "epoch": 1.8865225938242585,
      "grad_norm": 0.09768116474151611,
      "learning_rate": 9.773491881442998e-08,
      "loss": 0.0009,
      "step": 294140
    },
    {
      "epoch": 1.8865867307180446,
      "grad_norm": 0.10387171059846878,
      "learning_rate": 9.762482601198886e-08,
      "loss": 0.0017,
      "step": 294150
    },
    {
      "epoch": 1.8866508676118308,
      "grad_norm": 0.06271151453256607,
      "learning_rate": 9.751479463981306e-08,
      "loss": 0.0006,
      "step": 294160
    },
    {
      "epoch": 1.886715004505617,
      "grad_norm": 0.15673549473285675,
      "learning_rate": 9.740482469928036e-08,
      "loss": 0.0009,
      "step": 294170
    },
    {
      "epoch": 1.8867791413994028,
      "grad_norm": 0.14968831837177277,
      "learning_rate": 9.729491619176912e-08,
      "loss": 0.0018,
      "step": 294180
    },
    {
      "epoch": 1.886843278293189,
      "grad_norm": 0.13565237820148468,
      "learning_rate": 9.718506911865655e-08,
      "loss": 0.0007,
      "step": 294190
    },
    {
      "epoch": 1.8869074151869751,
      "grad_norm": 0.07846157252788544,
      "learning_rate": 9.707528348131878e-08,
      "loss": 0.0015,
      "step": 294200
    },
    {
      "epoch": 1.886971552080761,
      "grad_norm": 0.09204721450805664,
      "learning_rate": 9.69655592811325e-08,
      "loss": 0.0004,
      "step": 294210
    },
    {
      "epoch": 1.8870356889745472,
      "grad_norm": 0.003562049474567175,
      "learning_rate": 9.685589651947102e-08,
      "loss": 0.0012,
      "step": 294220
    },
    {
      "epoch": 1.8870998258683334,
      "grad_norm": 0.059349425137043,
      "learning_rate": 9.67462951977094e-08,
      "loss": 0.001,
      "step": 294230
    },
    {
      "epoch": 1.8871639627621195,
      "grad_norm": 0.08028772473335266,
      "learning_rate": 9.66367553172215e-08,
      "loss": 0.0009,
      "step": 294240
    },
    {
      "epoch": 1.8872280996559057,
      "grad_norm": 0.0028649321757256985,
      "learning_rate": 9.652727687937957e-08,
      "loss": 0.0007,
      "step": 294250
    },
    {
      "epoch": 1.8872922365496918,
      "grad_norm": 0.16632534563541412,
      "learning_rate": 9.641785988555529e-08,
      "loss": 0.0018,
      "step": 294260
    },
    {
      "epoch": 1.8873563734434777,
      "grad_norm": 0.05998115986585617,
      "learning_rate": 9.630850433711925e-08,
      "loss": 0.0007,
      "step": 294270
    },
    {
      "epoch": 1.8874205103372639,
      "grad_norm": 0.2951284945011139,
      "learning_rate": 9.619921023544254e-08,
      "loss": 0.0014,
      "step": 294280
    },
    {
      "epoch": 1.8874846472310498,
      "grad_norm": 0.05850505083799362,
      "learning_rate": 9.608997758189465e-08,
      "loss": 0.0008,
      "step": 294290
    },
    {
      "epoch": 1.887548784124836,
      "grad_norm": 0.005478013306856155,
      "learning_rate": 9.598080637784335e-08,
      "loss": 0.0005,
      "step": 294300
    },
    {
      "epoch": 1.887612921018622,
      "grad_norm": 0.13967640697956085,
      "learning_rate": 9.587169662465811e-08,
      "loss": 0.001,
      "step": 294310
    },
    {
      "epoch": 1.8876770579124083,
      "grad_norm": 0.06771519780158997,
      "learning_rate": 9.576264832370508e-08,
      "loss": 0.0007,
      "step": 294320
    },
    {
      "epoch": 1.8877411948061944,
      "grad_norm": 0.17400945723056793,
      "learning_rate": 9.565366147635147e-08,
      "loss": 0.0008,
      "step": 294330
    },
    {
      "epoch": 1.8878053316999805,
      "grad_norm": 0.05246680974960327,
      "learning_rate": 9.554473608396175e-08,
      "loss": 0.0007,
      "step": 294340
    },
    {
      "epoch": 1.8878694685937667,
      "grad_norm": 0.016703316941857338,
      "learning_rate": 9.543587214790261e-08,
      "loss": 0.0017,
      "step": 294350
    },
    {
      "epoch": 1.8879336054875526,
      "grad_norm": 0.014375735074281693,
      "learning_rate": 9.532706966953686e-08,
      "loss": 0.002,
      "step": 294360
    },
    {
      "epoch": 1.8879977423813388,
      "grad_norm": 0.1126503273844719,
      "learning_rate": 9.52183286502284e-08,
      "loss": 0.001,
      "step": 294370
    },
    {
      "epoch": 1.8880618792751247,
      "grad_norm": 0.0054759010672569275,
      "learning_rate": 9.510964909133946e-08,
      "loss": 0.0011,
      "step": 294380
    },
    {
      "epoch": 1.8881260161689108,
      "grad_norm": 0.05257393419742584,
      "learning_rate": 9.500103099423174e-08,
      "loss": 0.0011,
      "step": 294390
    },
    {
      "epoch": 1.888190153062697,
      "grad_norm": 0.05237250030040741,
      "learning_rate": 9.489247436026749e-08,
      "loss": 0.0037,
      "step": 294400
    },
    {
      "epoch": 1.8882542899564831,
      "grad_norm": 0.13146939873695374,
      "learning_rate": 9.478397919080506e-08,
      "loss": 0.0014,
      "step": 294410
    },
    {
      "epoch": 1.8883184268502693,
      "grad_norm": 0.01931230165064335,
      "learning_rate": 9.467554548720615e-08,
      "loss": 0.0009,
      "step": 294420
    },
    {
      "epoch": 1.8883825637440554,
      "grad_norm": 0.1467883288860321,
      "learning_rate": 9.456717325082798e-08,
      "loss": 0.002,
      "step": 294430
    },
    {
      "epoch": 1.8884467006378414,
      "grad_norm": 0.02824421413242817,
      "learning_rate": 9.445886248302949e-08,
      "loss": 0.0005,
      "step": 294440
    },
    {
      "epoch": 1.8885108375316275,
      "grad_norm": 0.038735803216695786,
      "learning_rate": 9.435061318516625e-08,
      "loss": 0.0011,
      "step": 294450
    },
    {
      "epoch": 1.8885749744254134,
      "grad_norm": 0.03149021416902542,
      "learning_rate": 9.424242535859662e-08,
      "loss": 0.0004,
      "step": 294460
    },
    {
      "epoch": 1.8886391113191996,
      "grad_norm": 0.034077052026987076,
      "learning_rate": 9.413429900467564e-08,
      "loss": 0.0007,
      "step": 294470
    },
    {
      "epoch": 1.8887032482129857,
      "grad_norm": 0.00893963873386383,
      "learning_rate": 9.402623412475775e-08,
      "loss": 0.0005,
      "step": 294480
    },
    {
      "epoch": 1.888767385106772,
      "grad_norm": 0.010076693259179592,
      "learning_rate": 9.391823072019746e-08,
      "loss": 0.0003,
      "step": 294490
    },
    {
      "epoch": 1.888831522000558,
      "grad_norm": 0.0885658785700798,
      "learning_rate": 9.38102887923481e-08,
      "loss": 0.0016,
      "step": 294500
    },
    {
      "epoch": 1.8888956588943442,
      "grad_norm": 0.053332455456256866,
      "learning_rate": 9.37024083425625e-08,
      "loss": 0.001,
      "step": 294510
    },
    {
      "epoch": 1.8889597957881303,
      "grad_norm": 0.07140041142702103,
      "learning_rate": 9.359458937219179e-08,
      "loss": 0.0025,
      "step": 294520
    },
    {
      "epoch": 1.8890239326819163,
      "grad_norm": 0.06093136593699455,
      "learning_rate": 9.348683188258712e-08,
      "loss": 0.001,
      "step": 294530
    },
    {
      "epoch": 1.8890880695757024,
      "grad_norm": 0.44862136244773865,
      "learning_rate": 9.337913587509961e-08,
      "loss": 0.0027,
      "step": 294540
    },
    {
      "epoch": 1.8891522064694883,
      "grad_norm": 0.14340335130691528,
      "learning_rate": 9.327150135107821e-08,
      "loss": 0.0015,
      "step": 294550
    },
    {
      "epoch": 1.8892163433632745,
      "grad_norm": 0.07659434527158737,
      "learning_rate": 9.316392831187126e-08,
      "loss": 0.0016,
      "step": 294560
    },
    {
      "epoch": 1.8892804802570606,
      "grad_norm": 0.0966777428984642,
      "learning_rate": 9.305641675882771e-08,
      "loss": 0.0012,
      "step": 294570
    },
    {
      "epoch": 1.8893446171508468,
      "grad_norm": 0.28905001282691956,
      "learning_rate": 9.294896669329423e-08,
      "loss": 0.0021,
      "step": 294580
    },
    {
      "epoch": 1.889408754044633,
      "grad_norm": 0.12563137710094452,
      "learning_rate": 9.284157811661698e-08,
      "loss": 0.0015,
      "step": 294590
    },
    {
      "epoch": 1.889472890938419,
      "grad_norm": 0.2730359137058258,
      "learning_rate": 9.273425103014155e-08,
      "loss": 0.0022,
      "step": 294600
    },
    {
      "epoch": 1.889537027832205,
      "grad_norm": 0.07535937428474426,
      "learning_rate": 9.262698543521353e-08,
      "loss": 0.0008,
      "step": 294610
    },
    {
      "epoch": 1.8896011647259912,
      "grad_norm": 0.35285791754722595,
      "learning_rate": 9.251978133317629e-08,
      "loss": 0.0023,
      "step": 294620
    },
    {
      "epoch": 1.8896653016197773,
      "grad_norm": 0.08534836769104004,
      "learning_rate": 9.241263872537376e-08,
      "loss": 0.0007,
      "step": 294630
    },
    {
      "epoch": 1.8897294385135632,
      "grad_norm": 0.08622516691684723,
      "learning_rate": 9.230555761314819e-08,
      "loss": 0.0018,
      "step": 294640
    },
    {
      "epoch": 1.8897935754073494,
      "grad_norm": 0.06464492529630661,
      "learning_rate": 9.219853799784129e-08,
      "loss": 0.0008,
      "step": 294650
    },
    {
      "epoch": 1.8898577123011355,
      "grad_norm": 0.0783517137169838,
      "learning_rate": 9.209157988079475e-08,
      "loss": 0.0018,
      "step": 294660
    },
    {
      "epoch": 1.8899218491949217,
      "grad_norm": 0.057245105504989624,
      "learning_rate": 9.198468326334809e-08,
      "loss": 0.0009,
      "step": 294670
    },
    {
      "epoch": 1.8899859860887078,
      "grad_norm": 0.05029996857047081,
      "learning_rate": 9.187784814684076e-08,
      "loss": 0.0009,
      "step": 294680
    },
    {
      "epoch": 1.890050122982494,
      "grad_norm": 0.10758306831121445,
      "learning_rate": 9.177107453261225e-08,
      "loss": 0.0008,
      "step": 294690
    },
    {
      "epoch": 1.89011425987628,
      "grad_norm": 0.2533423602581024,
      "learning_rate": 9.166436242199983e-08,
      "loss": 0.0023,
      "step": 294700
    },
    {
      "epoch": 1.890178396770066,
      "grad_norm": 0.020279565826058388,
      "learning_rate": 9.155771181634076e-08,
      "loss": 0.0011,
      "step": 294710
    },
    {
      "epoch": 1.890242533663852,
      "grad_norm": 0.1295294463634491,
      "learning_rate": 9.145112271697176e-08,
      "loss": 0.0007,
      "step": 294720
    },
    {
      "epoch": 1.8903066705576381,
      "grad_norm": 0.03561263158917427,
      "learning_rate": 9.134459512522842e-08,
      "loss": 0.0015,
      "step": 294730
    },
    {
      "epoch": 1.8903708074514243,
      "grad_norm": 0.016167763620615005,
      "learning_rate": 9.123812904244522e-08,
      "loss": 0.001,
      "step": 294740
    },
    {
      "epoch": 1.8904349443452104,
      "grad_norm": 0.055394627153873444,
      "learning_rate": 9.11317244699561e-08,
      "loss": 0.0008,
      "step": 294750
    },
    {
      "epoch": 1.8904990812389966,
      "grad_norm": 0.029049696400761604,
      "learning_rate": 9.102538140909555e-08,
      "loss": 0.0015,
      "step": 294760
    },
    {
      "epoch": 1.8905632181327827,
      "grad_norm": 0.0370810441672802,
      "learning_rate": 9.091909986119474e-08,
      "loss": 0.0009,
      "step": 294770
    },
    {
      "epoch": 1.8906273550265689,
      "grad_norm": 0.17660130560398102,
      "learning_rate": 9.081287982758647e-08,
      "loss": 0.0007,
      "step": 294780
    },
    {
      "epoch": 1.8906914919203548,
      "grad_norm": 0.15603278577327728,
      "learning_rate": 9.070672130960134e-08,
      "loss": 0.0015,
      "step": 294790
    },
    {
      "epoch": 1.890755628814141,
      "grad_norm": 0.0040861391462385654,
      "learning_rate": 9.060062430856942e-08,
      "loss": 0.0004,
      "step": 294800
    },
    {
      "epoch": 1.8908197657079269,
      "grad_norm": 0.09123212099075317,
      "learning_rate": 9.049458882582075e-08,
      "loss": 0.0006,
      "step": 294810
    },
    {
      "epoch": 1.890883902601713,
      "grad_norm": 0.04571864753961563,
      "learning_rate": 9.038861486268313e-08,
      "loss": 0.0011,
      "step": 294820
    },
    {
      "epoch": 1.8909480394954992,
      "grad_norm": 0.192408949136734,
      "learning_rate": 9.028270242048498e-08,
      "loss": 0.0009,
      "step": 294830
    },
    {
      "epoch": 1.8910121763892853,
      "grad_norm": 0.04279579594731331,
      "learning_rate": 9.017685150055411e-08,
      "loss": 0.0004,
      "step": 294840
    },
    {
      "epoch": 1.8910763132830715,
      "grad_norm": 0.19916805624961853,
      "learning_rate": 9.007106210421613e-08,
      "loss": 0.0032,
      "step": 294850
    },
    {
      "epoch": 1.8911404501768576,
      "grad_norm": 0.03907819092273712,
      "learning_rate": 8.996533423279608e-08,
      "loss": 0.001,
      "step": 294860
    },
    {
      "epoch": 1.8912045870706435,
      "grad_norm": 0.03859139606356621,
      "learning_rate": 8.985966788762013e-08,
      "loss": 0.0008,
      "step": 294870
    },
    {
      "epoch": 1.8912687239644297,
      "grad_norm": 0.05505898594856262,
      "learning_rate": 8.975406307001222e-08,
      "loss": 0.0008,
      "step": 294880
    },
    {
      "epoch": 1.8913328608582156,
      "grad_norm": 0.0035464514512568712,
      "learning_rate": 8.964851978129463e-08,
      "loss": 0.0006,
      "step": 294890
    },
    {
      "epoch": 1.8913969977520018,
      "grad_norm": 0.016381246969103813,
      "learning_rate": 8.954303802279019e-08,
      "loss": 0.0013,
      "step": 294900
    },
    {
      "epoch": 1.891461134645788,
      "grad_norm": 0.04539335519075394,
      "learning_rate": 8.943761779582116e-08,
      "loss": 0.0006,
      "step": 294910
    },
    {
      "epoch": 1.891525271539574,
      "grad_norm": 0.04665771871805191,
      "learning_rate": 8.933225910170818e-08,
      "loss": 0.0004,
      "step": 294920
    },
    {
      "epoch": 1.8915894084333602,
      "grad_norm": 0.26177358627319336,
      "learning_rate": 8.922696194177238e-08,
      "loss": 0.001,
      "step": 294930
    },
    {
      "epoch": 1.8916535453271464,
      "grad_norm": 0.005626978352665901,
      "learning_rate": 8.912172631733162e-08,
      "loss": 0.0004,
      "step": 294940
    },
    {
      "epoch": 1.8917176822209325,
      "grad_norm": 0.3913898766040802,
      "learning_rate": 8.90165522297054e-08,
      "loss": 0.0024,
      "step": 294950
    },
    {
      "epoch": 1.8917818191147184,
      "grad_norm": 0.045536499470472336,
      "learning_rate": 8.89114396802121e-08,
      "loss": 0.0016,
      "step": 294960
    },
    {
      "epoch": 1.8918459560085046,
      "grad_norm": 0.051582399755716324,
      "learning_rate": 8.880638867016844e-08,
      "loss": 0.0006,
      "step": 294970
    },
    {
      "epoch": 1.8919100929022905,
      "grad_norm": 0.1261383295059204,
      "learning_rate": 8.870139920089005e-08,
      "loss": 0.0009,
      "step": 294980
    },
    {
      "epoch": 1.8919742297960767,
      "grad_norm": 0.07379081100225449,
      "learning_rate": 8.859647127369364e-08,
      "loss": 0.0011,
      "step": 294990
    },
    {
      "epoch": 1.8920383666898628,
      "grad_norm": 0.06457892805337906,
      "learning_rate": 8.849160488989317e-08,
      "loss": 0.0009,
      "step": 295000
    },
    {
      "epoch": 1.892102503583649,
      "grad_norm": 0.08591438084840775,
      "learning_rate": 8.838680005080368e-08,
      "loss": 0.003,
      "step": 295010
    },
    {
      "epoch": 1.892166640477435,
      "grad_norm": 0.08735421299934387,
      "learning_rate": 8.828205675773749e-08,
      "loss": 0.001,
      "step": 295020
    },
    {
      "epoch": 1.8922307773712213,
      "grad_norm": 0.037042561918497086,
      "learning_rate": 8.81773750120074e-08,
      "loss": 0.0011,
      "step": 295030
    },
    {
      "epoch": 1.8922949142650072,
      "grad_norm": 0.03372732177376747,
      "learning_rate": 8.807275481492572e-08,
      "loss": 0.0013,
      "step": 295040
    },
    {
      "epoch": 1.8923590511587933,
      "grad_norm": 0.03540458902716637,
      "learning_rate": 8.79681961678025e-08,
      "loss": 0.0009,
      "step": 295050
    },
    {
      "epoch": 1.8924231880525795,
      "grad_norm": 0.006590025965124369,
      "learning_rate": 8.786369907194836e-08,
      "loss": 0.0005,
      "step": 295060
    },
    {
      "epoch": 1.8924873249463654,
      "grad_norm": 0.11606483906507492,
      "learning_rate": 8.775926352867281e-08,
      "loss": 0.0016,
      "step": 295070
    },
    {
      "epoch": 1.8925514618401515,
      "grad_norm": 0.04718036204576492,
      "learning_rate": 8.765488953928425e-08,
      "loss": 0.0008,
      "step": 295080
    },
    {
      "epoch": 1.8926155987339377,
      "grad_norm": 0.0750744491815567,
      "learning_rate": 8.755057710509108e-08,
      "loss": 0.0011,
      "step": 295090
    },
    {
      "epoch": 1.8926797356277238,
      "grad_norm": 0.0894516259431839,
      "learning_rate": 8.744632622739946e-08,
      "loss": 0.0014,
      "step": 295100
    },
    {
      "epoch": 1.89274387252151,
      "grad_norm": 0.039059873670339584,
      "learning_rate": 8.734213690751725e-08,
      "loss": 0.0011,
      "step": 295110
    },
    {
      "epoch": 1.8928080094152961,
      "grad_norm": 0.1326218545436859,
      "learning_rate": 8.72380091467484e-08,
      "loss": 0.0017,
      "step": 295120
    },
    {
      "epoch": 1.892872146309082,
      "grad_norm": 0.07302480936050415,
      "learning_rate": 8.713394294639799e-08,
      "loss": 0.0009,
      "step": 295130
    },
    {
      "epoch": 1.8929362832028682,
      "grad_norm": 0.08233803510665894,
      "learning_rate": 8.702993830777162e-08,
      "loss": 0.001,
      "step": 295140
    },
    {
      "epoch": 1.8930004200966541,
      "grad_norm": 0.004406723193824291,
      "learning_rate": 8.692599523217049e-08,
      "loss": 0.0007,
      "step": 295150
    },
    {
      "epoch": 1.8930645569904403,
      "grad_norm": 0.10290886461734772,
      "learning_rate": 8.682211372089855e-08,
      "loss": 0.0006,
      "step": 295160
    },
    {
      "epoch": 1.8931286938842264,
      "grad_norm": 0.06705625355243683,
      "learning_rate": 8.671829377525642e-08,
      "loss": 0.0019,
      "step": 295170
    },
    {
      "epoch": 1.8931928307780126,
      "grad_norm": 0.06751684844493866,
      "learning_rate": 8.66145353965453e-08,
      "loss": 0.0009,
      "step": 295180
    },
    {
      "epoch": 1.8932569676717987,
      "grad_norm": 0.12947224080562592,
      "learning_rate": 8.651083858606635e-08,
      "loss": 0.0005,
      "step": 295190
    },
    {
      "epoch": 1.893321104565585,
      "grad_norm": 0.03041689656674862,
      "learning_rate": 8.640720334511799e-08,
      "loss": 0.0009,
      "step": 295200
    },
    {
      "epoch": 1.893385241459371,
      "grad_norm": 0.13694101572036743,
      "learning_rate": 8.630362967499862e-08,
      "loss": 0.0013,
      "step": 295210
    },
    {
      "epoch": 1.893449378353157,
      "grad_norm": 0.17339839041233063,
      "learning_rate": 8.620011757700719e-08,
      "loss": 0.0016,
      "step": 295220
    },
    {
      "epoch": 1.8935135152469431,
      "grad_norm": 0.06717420369386673,
      "learning_rate": 8.60966670524399e-08,
      "loss": 0.0015,
      "step": 295230
    },
    {
      "epoch": 1.893577652140729,
      "grad_norm": 0.05104643851518631,
      "learning_rate": 8.599327810259295e-08,
      "loss": 0.0012,
      "step": 295240
    },
    {
      "epoch": 1.8936417890345152,
      "grad_norm": 0.021760204806923866,
      "learning_rate": 8.588995072876249e-08,
      "loss": 0.0006,
      "step": 295250
    },
    {
      "epoch": 1.8937059259283013,
      "grad_norm": 0.07978024333715439,
      "learning_rate": 8.578668493224306e-08,
      "loss": 0.0011,
      "step": 295260
    },
    {
      "epoch": 1.8937700628220875,
      "grad_norm": 0.3168538808822632,
      "learning_rate": 8.568348071432863e-08,
      "loss": 0.0024,
      "step": 295270
    },
    {
      "epoch": 1.8938341997158736,
      "grad_norm": 0.07114940881729126,
      "learning_rate": 8.55803380763115e-08,
      "loss": 0.0022,
      "step": 295280
    },
    {
      "epoch": 1.8938983366096598,
      "grad_norm": 0.021724779158830643,
      "learning_rate": 8.547725701948618e-08,
      "loss": 0.0006,
      "step": 295290
    },
    {
      "epoch": 1.8939624735034457,
      "grad_norm": 0.03300023078918457,
      "learning_rate": 8.537423754514274e-08,
      "loss": 0.001,
      "step": 295300
    },
    {
      "epoch": 1.8940266103972319,
      "grad_norm": 0.07515677064657211,
      "learning_rate": 8.527127965457293e-08,
      "loss": 0.0012,
      "step": 295310
    },
    {
      "epoch": 1.8940907472910178,
      "grad_norm": 0.05644484981894493,
      "learning_rate": 8.516838334906574e-08,
      "loss": 0.0012,
      "step": 295320
    },
    {
      "epoch": 1.894154884184804,
      "grad_norm": 0.008623950183391571,
      "learning_rate": 8.506554862991179e-08,
      "loss": 0.0009,
      "step": 295330
    },
    {
      "epoch": 1.89421902107859,
      "grad_norm": 0.04799478501081467,
      "learning_rate": 8.496277549839893e-08,
      "loss": 0.0014,
      "step": 295340
    },
    {
      "epoch": 1.8942831579723762,
      "grad_norm": 0.050887249410152435,
      "learning_rate": 8.486006395581559e-08,
      "loss": 0.001,
      "step": 295350
    },
    {
      "epoch": 1.8943472948661624,
      "grad_norm": 0.2912452518939972,
      "learning_rate": 8.475741400344794e-08,
      "loss": 0.0008,
      "step": 295360
    },
    {
      "epoch": 1.8944114317599485,
      "grad_norm": 0.05949672311544418,
      "learning_rate": 8.465482564258332e-08,
      "loss": 0.0012,
      "step": 295370
    },
    {
      "epoch": 1.8944755686537347,
      "grad_norm": 0.05052575096487999,
      "learning_rate": 8.455229887450622e-08,
      "loss": 0.0007,
      "step": 295380
    },
    {
      "epoch": 1.8945397055475206,
      "grad_norm": 0.014839448034763336,
      "learning_rate": 8.44498337005023e-08,
      "loss": 0.0011,
      "step": 295390
    },
    {
      "epoch": 1.8946038424413068,
      "grad_norm": 0.10445275902748108,
      "learning_rate": 8.434743012185442e-08,
      "loss": 0.0012,
      "step": 295400
    },
    {
      "epoch": 1.8946679793350927,
      "grad_norm": 0.16053244471549988,
      "learning_rate": 8.424508813984711e-08,
      "loss": 0.0005,
      "step": 295410
    },
    {
      "epoch": 1.8947321162288788,
      "grad_norm": 0.010030178353190422,
      "learning_rate": 8.414280775576156e-08,
      "loss": 0.0005,
      "step": 295420
    },
    {
      "epoch": 1.894796253122665,
      "grad_norm": 0.18217097222805023,
      "learning_rate": 8.404058897088008e-08,
      "loss": 0.0014,
      "step": 295430
    },
    {
      "epoch": 1.8948603900164511,
      "grad_norm": 0.16161665320396423,
      "learning_rate": 8.393843178648331e-08,
      "loss": 0.0012,
      "step": 295440
    },
    {
      "epoch": 1.8949245269102373,
      "grad_norm": 0.0912969708442688,
      "learning_rate": 8.383633620385134e-08,
      "loss": 0.0021,
      "step": 295450
    },
    {
      "epoch": 1.8949886638040234,
      "grad_norm": 0.0579666867852211,
      "learning_rate": 8.373430222426427e-08,
      "loss": 0.0013,
      "step": 295460
    },
    {
      "epoch": 1.8950528006978096,
      "grad_norm": 0.19168339669704437,
      "learning_rate": 8.363232984899938e-08,
      "loss": 0.0007,
      "step": 295470
    },
    {
      "epoch": 1.8951169375915955,
      "grad_norm": 0.11395134776830673,
      "learning_rate": 8.353041907933512e-08,
      "loss": 0.0011,
      "step": 295480
    },
    {
      "epoch": 1.8951810744853816,
      "grad_norm": 0.05215844139456749,
      "learning_rate": 8.342856991654879e-08,
      "loss": 0.0008,
      "step": 295490
    },
    {
      "epoch": 1.8952452113791676,
      "grad_norm": 0.012780345045030117,
      "learning_rate": 8.332678236191605e-08,
      "loss": 0.0007,
      "step": 295500
    },
    {
      "epoch": 1.8953093482729537,
      "grad_norm": 0.00840887613594532,
      "learning_rate": 8.322505641671252e-08,
      "loss": 0.0011,
      "step": 295510
    },
    {
      "epoch": 1.8953734851667399,
      "grad_norm": 0.011757910251617432,
      "learning_rate": 8.312339208221331e-08,
      "loss": 0.0008,
      "step": 295520
    },
    {
      "epoch": 1.895437622060526,
      "grad_norm": 0.08257078379392624,
      "learning_rate": 8.302178935969186e-08,
      "loss": 0.0006,
      "step": 295530
    },
    {
      "epoch": 1.8955017589543122,
      "grad_norm": 0.02417915128171444,
      "learning_rate": 8.292024825042155e-08,
      "loss": 0.0015,
      "step": 295540
    },
    {
      "epoch": 1.8955658958480983,
      "grad_norm": 0.10538596659898758,
      "learning_rate": 8.281876875567418e-08,
      "loss": 0.0017,
      "step": 295550
    },
    {
      "epoch": 1.8956300327418842,
      "grad_norm": 0.07173268496990204,
      "learning_rate": 8.271735087672261e-08,
      "loss": 0.0011,
      "step": 295560
    },
    {
      "epoch": 1.8956941696356704,
      "grad_norm": 0.10954654961824417,
      "learning_rate": 8.261599461483694e-08,
      "loss": 0.001,
      "step": 295570
    },
    {
      "epoch": 1.8957583065294563,
      "grad_norm": 0.040432389825582504,
      "learning_rate": 8.25146999712867e-08,
      "loss": 0.0014,
      "step": 295580
    },
    {
      "epoch": 1.8958224434232425,
      "grad_norm": 0.04825109243392944,
      "learning_rate": 8.241346694734198e-08,
      "loss": 0.0011,
      "step": 295590
    },
    {
      "epoch": 1.8958865803170286,
      "grad_norm": 0.138762965798378,
      "learning_rate": 8.231229554427123e-08,
      "loss": 0.0007,
      "step": 295600
    },
    {
      "epoch": 1.8959507172108148,
      "grad_norm": 0.14435996115207672,
      "learning_rate": 8.221118576334231e-08,
      "loss": 0.0011,
      "step": 295610
    },
    {
      "epoch": 1.896014854104601,
      "grad_norm": 0.031372375786304474,
      "learning_rate": 8.211013760582142e-08,
      "loss": 0.0027,
      "step": 295620
    },
    {
      "epoch": 1.896078990998387,
      "grad_norm": 0.0296054445207119,
      "learning_rate": 8.200915107297535e-08,
      "loss": 0.0012,
      "step": 295630
    },
    {
      "epoch": 1.8961431278921732,
      "grad_norm": 0.12075857073068619,
      "learning_rate": 8.190822616606975e-08,
      "loss": 0.0011,
      "step": 295640
    },
    {
      "epoch": 1.8962072647859591,
      "grad_norm": 0.040819283574819565,
      "learning_rate": 8.180736288636915e-08,
      "loss": 0.0021,
      "step": 295650
    },
    {
      "epoch": 1.8962714016797453,
      "grad_norm": 0.07715485244989395,
      "learning_rate": 8.170656123513643e-08,
      "loss": 0.0012,
      "step": 295660
    },
    {
      "epoch": 1.8963355385735312,
      "grad_norm": 0.015074445866048336,
      "learning_rate": 8.160582121363614e-08,
      "loss": 0.0006,
      "step": 295670
    },
    {
      "epoch": 1.8963996754673174,
      "grad_norm": 0.09128915518522263,
      "learning_rate": 8.150514282312949e-08,
      "loss": 0.001,
      "step": 295680
    },
    {
      "epoch": 1.8964638123611035,
      "grad_norm": 0.10052846372127533,
      "learning_rate": 8.140452606487937e-08,
      "loss": 0.0011,
      "step": 295690
    },
    {
      "epoch": 1.8965279492548897,
      "grad_norm": 0.017802821472287178,
      "learning_rate": 8.130397094014475e-08,
      "loss": 0.0011,
      "step": 295700
    },
    {
      "epoch": 1.8965920861486758,
      "grad_norm": 0.0714261382818222,
      "learning_rate": 8.120347745018798e-08,
      "loss": 0.0012,
      "step": 295710
    },
    {
      "epoch": 1.896656223042462,
      "grad_norm": 0.016490206122398376,
      "learning_rate": 8.110304559626635e-08,
      "loss": 0.0004,
      "step": 295720
    },
    {
      "epoch": 1.8967203599362479,
      "grad_norm": 0.1430540829896927,
      "learning_rate": 8.100267537963947e-08,
      "loss": 0.0015,
      "step": 295730
    },
    {
      "epoch": 1.896784496830034,
      "grad_norm": 0.1116696298122406,
      "learning_rate": 8.090236680156404e-08,
      "loss": 0.0007,
      "step": 295740
    },
    {
      "epoch": 1.8968486337238202,
      "grad_norm": 0.018631864339113235,
      "learning_rate": 8.0802119863298e-08,
      "loss": 0.0009,
      "step": 295750
    },
    {
      "epoch": 1.896912770617606,
      "grad_norm": 0.05354061350226402,
      "learning_rate": 8.070193456609699e-08,
      "loss": 0.0008,
      "step": 295760
    },
    {
      "epoch": 1.8969769075113923,
      "grad_norm": 0.0025101928040385246,
      "learning_rate": 8.060181091121667e-08,
      "loss": 0.0008,
      "step": 295770
    },
    {
      "epoch": 1.8970410444051784,
      "grad_norm": 0.09185576438903809,
      "learning_rate": 8.050174889991103e-08,
      "loss": 0.0017,
      "step": 295780
    },
    {
      "epoch": 1.8971051812989645,
      "grad_norm": 0.0862610936164856,
      "learning_rate": 8.040174853343464e-08,
      "loss": 0.0005,
      "step": 295790
    },
    {
      "epoch": 1.8971693181927507,
      "grad_norm": 0.024376560002565384,
      "learning_rate": 8.030180981304036e-08,
      "loss": 0.0013,
      "step": 295800
    },
    {
      "epoch": 1.8972334550865368,
      "grad_norm": 0.05399715155363083,
      "learning_rate": 8.020193273998055e-08,
      "loss": 0.0012,
      "step": 295810
    },
    {
      "epoch": 1.8972975919803228,
      "grad_norm": 0.029330408200621605,
      "learning_rate": 8.010211731550643e-08,
      "loss": 0.0015,
      "step": 295820
    },
    {
      "epoch": 1.897361728874109,
      "grad_norm": 0.0124303437769413,
      "learning_rate": 8.00023635408692e-08,
      "loss": 0.0006,
      "step": 295830
    },
    {
      "epoch": 1.8974258657678948,
      "grad_norm": 0.33522889018058777,
      "learning_rate": 7.990267141731845e-08,
      "loss": 0.0012,
      "step": 295840
    },
    {
      "epoch": 1.897490002661681,
      "grad_norm": 0.04323439672589302,
      "learning_rate": 7.980304094610314e-08,
      "loss": 0.0007,
      "step": 295850
    },
    {
      "epoch": 1.8975541395554671,
      "grad_norm": 0.007989304140210152,
      "learning_rate": 7.970347212847285e-08,
      "loss": 0.0009,
      "step": 295860
    },
    {
      "epoch": 1.8976182764492533,
      "grad_norm": 0.06763707101345062,
      "learning_rate": 7.960396496567436e-08,
      "loss": 0.0012,
      "step": 295870
    },
    {
      "epoch": 1.8976824133430394,
      "grad_norm": 0.045833926647901535,
      "learning_rate": 7.9504519458955e-08,
      "loss": 0.0012,
      "step": 295880
    },
    {
      "epoch": 1.8977465502368256,
      "grad_norm": 0.06413446366786957,
      "learning_rate": 7.940513560955986e-08,
      "loss": 0.001,
      "step": 295890
    },
    {
      "epoch": 1.8978106871306117,
      "grad_norm": 0.08513320982456207,
      "learning_rate": 7.930581341873577e-08,
      "loss": 0.0006,
      "step": 295900
    },
    {
      "epoch": 1.8978748240243977,
      "grad_norm": 0.11804290115833282,
      "learning_rate": 7.920655288772672e-08,
      "loss": 0.0012,
      "step": 295910
    },
    {
      "epoch": 1.8979389609181838,
      "grad_norm": 0.018304867669939995,
      "learning_rate": 7.910735401777613e-08,
      "loss": 0.0013,
      "step": 295920
    },
    {
      "epoch": 1.8980030978119697,
      "grad_norm": 0.07030729949474335,
      "learning_rate": 7.900821681012693e-08,
      "loss": 0.0008,
      "step": 295930
    },
    {
      "epoch": 1.898067234705756,
      "grad_norm": 0.024078192189335823,
      "learning_rate": 7.8909141266022e-08,
      "loss": 0.0004,
      "step": 295940
    },
    {
      "epoch": 1.898131371599542,
      "grad_norm": 0.047128576785326004,
      "learning_rate": 7.881012738670257e-08,
      "loss": 0.001,
      "step": 295950
    },
    {
      "epoch": 1.8981955084933282,
      "grad_norm": 0.06363033503293991,
      "learning_rate": 7.871117517340987e-08,
      "loss": 0.0009,
      "step": 295960
    },
    {
      "epoch": 1.8982596453871143,
      "grad_norm": 0.3511105179786682,
      "learning_rate": 7.861228462738235e-08,
      "loss": 0.001,
      "step": 295970
    },
    {
      "epoch": 1.8983237822809005,
      "grad_norm": 0.05861715227365494,
      "learning_rate": 7.851345574986124e-08,
      "loss": 0.002,
      "step": 295980
    },
    {
      "epoch": 1.8983879191746864,
      "grad_norm": 0.017348650842905045,
      "learning_rate": 7.841468854208334e-08,
      "loss": 0.0005,
      "step": 295990
    },
    {
      "epoch": 1.8984520560684726,
      "grad_norm": 0.03482039272785187,
      "learning_rate": 7.831598300528653e-08,
      "loss": 0.0011,
      "step": 296000
    },
    {
      "epoch": 1.8985161929622585,
      "grad_norm": 0.005001869518309832,
      "learning_rate": 7.821733914070816e-08,
      "loss": 0.0009,
      "step": 296010
    },
    {
      "epoch": 1.8985803298560446,
      "grad_norm": 0.14488859474658966,
      "learning_rate": 7.811875694958448e-08,
      "loss": 0.0025,
      "step": 296020
    },
    {
      "epoch": 1.8986444667498308,
      "grad_norm": 0.09802429378032684,
      "learning_rate": 7.802023643315005e-08,
      "loss": 0.0012,
      "step": 296030
    },
    {
      "epoch": 1.898708603643617,
      "grad_norm": 0.018383031710982323,
      "learning_rate": 7.792177759263941e-08,
      "loss": 0.001,
      "step": 296040
    },
    {
      "epoch": 1.898772740537403,
      "grad_norm": 0.1427089124917984,
      "learning_rate": 7.782338042928716e-08,
      "loss": 0.0019,
      "step": 296050
    },
    {
      "epoch": 1.8988368774311892,
      "grad_norm": 0.02857111766934395,
      "learning_rate": 7.772504494432564e-08,
      "loss": 0.0018,
      "step": 296060
    },
    {
      "epoch": 1.8989010143249754,
      "grad_norm": 0.12065998464822769,
      "learning_rate": 7.76267711389872e-08,
      "loss": 0.0008,
      "step": 296070
    },
    {
      "epoch": 1.8989651512187613,
      "grad_norm": 0.10027211904525757,
      "learning_rate": 7.752855901450306e-08,
      "loss": 0.0006,
      "step": 296080
    },
    {
      "epoch": 1.8990292881125475,
      "grad_norm": 0.14700154960155487,
      "learning_rate": 7.743040857210393e-08,
      "loss": 0.0025,
      "step": 296090
    },
    {
      "epoch": 1.8990934250063334,
      "grad_norm": 0.03631043806672096,
      "learning_rate": 7.733231981302047e-08,
      "loss": 0.001,
      "step": 296100
    },
    {
      "epoch": 1.8991575619001195,
      "grad_norm": 0.05204037204384804,
      "learning_rate": 7.72342927384806e-08,
      "loss": 0.0016,
      "step": 296110
    },
    {
      "epoch": 1.8992216987939057,
      "grad_norm": 0.02272013947367668,
      "learning_rate": 7.713632734971388e-08,
      "loss": 0.0006,
      "step": 296120
    },
    {
      "epoch": 1.8992858356876918,
      "grad_norm": 0.06338615715503693,
      "learning_rate": 7.703842364794711e-08,
      "loss": 0.0014,
      "step": 296130
    },
    {
      "epoch": 1.899349972581478,
      "grad_norm": 0.07160641252994537,
      "learning_rate": 7.694058163440766e-08,
      "loss": 0.001,
      "step": 296140
    },
    {
      "epoch": 1.8994141094752641,
      "grad_norm": 0.03875984996557236,
      "learning_rate": 7.684280131032062e-08,
      "loss": 0.0007,
      "step": 296150
    },
    {
      "epoch": 1.89947824636905,
      "grad_norm": 0.07607729732990265,
      "learning_rate": 7.674508267691172e-08,
      "loss": 0.0008,
      "step": 296160
    },
    {
      "epoch": 1.8995423832628362,
      "grad_norm": 0.034141361713409424,
      "learning_rate": 7.664742573540607e-08,
      "loss": 0.0006,
      "step": 296170
    },
    {
      "epoch": 1.8996065201566223,
      "grad_norm": 0.04295650124549866,
      "learning_rate": 7.654983048702658e-08,
      "loss": 0.0008,
      "step": 296180
    },
    {
      "epoch": 1.8996706570504083,
      "grad_norm": 0.013392886146903038,
      "learning_rate": 7.645229693299617e-08,
      "loss": 0.0006,
      "step": 296190
    },
    {
      "epoch": 1.8997347939441944,
      "grad_norm": 0.03318767994642258,
      "learning_rate": 7.635482507453773e-08,
      "loss": 0.0007,
      "step": 296200
    },
    {
      "epoch": 1.8997989308379806,
      "grad_norm": 0.039401594549417496,
      "learning_rate": 7.625741491287197e-08,
      "loss": 0.0011,
      "step": 296210
    },
    {
      "epoch": 1.8998630677317667,
      "grad_norm": 0.03686876967549324,
      "learning_rate": 7.616006644922014e-08,
      "loss": 0.0064,
      "step": 296220
    },
    {
      "epoch": 1.8999272046255529,
      "grad_norm": 0.22167079150676727,
      "learning_rate": 7.606277968480125e-08,
      "loss": 0.0012,
      "step": 296230
    },
    {
      "epoch": 1.899991341519339,
      "grad_norm": 0.036673545837402344,
      "learning_rate": 7.596555462083488e-08,
      "loss": 0.0008,
      "step": 296240
    },
    {
      "epoch": 1.900055478413125,
      "grad_norm": 0.048822399228811264,
      "learning_rate": 7.586839125853951e-08,
      "loss": 0.0014,
      "step": 296250
    },
    {
      "epoch": 1.900119615306911,
      "grad_norm": 0.23170608282089233,
      "learning_rate": 7.577128959913193e-08,
      "loss": 0.0014,
      "step": 296260
    },
    {
      "epoch": 1.900183752200697,
      "grad_norm": 0.01074785366654396,
      "learning_rate": 7.567424964382953e-08,
      "loss": 0.0004,
      "step": 296270
    },
    {
      "epoch": 1.9002478890944832,
      "grad_norm": 0.08027832210063934,
      "learning_rate": 7.557727139384852e-08,
      "loss": 0.0021,
      "step": 296280
    },
    {
      "epoch": 1.9003120259882693,
      "grad_norm": 0.09156271070241928,
      "learning_rate": 7.54803548504035e-08,
      "loss": 0.0011,
      "step": 296290
    },
    {
      "epoch": 1.9003761628820555,
      "grad_norm": 0.09900987148284912,
      "learning_rate": 7.538350001470907e-08,
      "loss": 0.0017,
      "step": 296300
    },
    {
      "epoch": 1.9004402997758416,
      "grad_norm": 0.0034116050228476524,
      "learning_rate": 7.528670688797868e-08,
      "loss": 0.0009,
      "step": 296310
    },
    {
      "epoch": 1.9005044366696278,
      "grad_norm": 0.04635313153266907,
      "learning_rate": 7.518997547142637e-08,
      "loss": 0.0006,
      "step": 296320
    },
    {
      "epoch": 1.900568573563414,
      "grad_norm": 0.16556833684444427,
      "learning_rate": 7.509330576626284e-08,
      "loss": 0.0021,
      "step": 296330
    },
    {
      "epoch": 1.9006327104571998,
      "grad_norm": 0.08494561910629272,
      "learning_rate": 7.499669777369933e-08,
      "loss": 0.0018,
      "step": 296340
    },
    {
      "epoch": 1.900696847350986,
      "grad_norm": 0.21394364535808563,
      "learning_rate": 7.490015149494823e-08,
      "loss": 0.0033,
      "step": 296350
    },
    {
      "epoch": 1.900760984244772,
      "grad_norm": 0.09006308764219284,
      "learning_rate": 7.480366693121744e-08,
      "loss": 0.0011,
      "step": 296360
    },
    {
      "epoch": 1.900825121138558,
      "grad_norm": 0.044618889689445496,
      "learning_rate": 7.47072440837171e-08,
      "loss": 0.001,
      "step": 296370
    },
    {
      "epoch": 1.9008892580323442,
      "grad_norm": 0.06901489198207855,
      "learning_rate": 7.46108829536546e-08,
      "loss": 0.0011,
      "step": 296380
    },
    {
      "epoch": 1.9009533949261304,
      "grad_norm": 0.15160177648067474,
      "learning_rate": 7.451458354223784e-08,
      "loss": 0.0012,
      "step": 296390
    },
    {
      "epoch": 1.9010175318199165,
      "grad_norm": 0.1087668314576149,
      "learning_rate": 7.441834585067365e-08,
      "loss": 0.0018,
      "step": 296400
    },
    {
      "epoch": 1.9010816687137027,
      "grad_norm": 0.0020599712152034044,
      "learning_rate": 7.432216988016771e-08,
      "loss": 0.0008,
      "step": 296410
    },
    {
      "epoch": 1.9011458056074886,
      "grad_norm": 0.07600867748260498,
      "learning_rate": 7.422605563192575e-08,
      "loss": 0.0005,
      "step": 296420
    },
    {
      "epoch": 1.9012099425012747,
      "grad_norm": 0.03156861662864685,
      "learning_rate": 7.41300031071518e-08,
      "loss": 0.0021,
      "step": 296430
    },
    {
      "epoch": 1.9012740793950607,
      "grad_norm": 0.05616842210292816,
      "learning_rate": 7.403401230704876e-08,
      "loss": 0.0009,
      "step": 296440
    },
    {
      "epoch": 1.9013382162888468,
      "grad_norm": 0.036105792969465256,
      "learning_rate": 7.39380832328207e-08,
      "loss": 0.0005,
      "step": 296450
    },
    {
      "epoch": 1.901402353182633,
      "grad_norm": 0.1208905577659607,
      "learning_rate": 7.384221588566831e-08,
      "loss": 0.0008,
      "step": 296460
    },
    {
      "epoch": 1.901466490076419,
      "grad_norm": 0.10680118948221207,
      "learning_rate": 7.374641026679396e-08,
      "loss": 0.0006,
      "step": 296470
    },
    {
      "epoch": 1.9015306269702053,
      "grad_norm": 0.1479857712984085,
      "learning_rate": 7.365066637739837e-08,
      "loss": 0.0013,
      "step": 296480
    },
    {
      "epoch": 1.9015947638639914,
      "grad_norm": 0.13697151839733124,
      "learning_rate": 7.355498421868001e-08,
      "loss": 0.0011,
      "step": 296490
    },
    {
      "epoch": 1.9016589007577775,
      "grad_norm": 0.1833760142326355,
      "learning_rate": 7.345936379183904e-08,
      "loss": 0.0009,
      "step": 296500
    },
    {
      "epoch": 1.9017230376515635,
      "grad_norm": 0.012452002614736557,
      "learning_rate": 7.336380509807284e-08,
      "loss": 0.0004,
      "step": 296510
    },
    {
      "epoch": 1.9017871745453496,
      "grad_norm": 0.06796858459711075,
      "learning_rate": 7.326830813857933e-08,
      "loss": 0.0015,
      "step": 296520
    },
    {
      "epoch": 1.9018513114391356,
      "grad_norm": 0.04385437071323395,
      "learning_rate": 7.317287291455478e-08,
      "loss": 0.0012,
      "step": 296530
    },
    {
      "epoch": 1.9019154483329217,
      "grad_norm": 0.05254632607102394,
      "learning_rate": 7.3077499427196e-08,
      "loss": 0.002,
      "step": 296540
    },
    {
      "epoch": 1.9019795852267078,
      "grad_norm": 0.06706543266773224,
      "learning_rate": 7.298218767769705e-08,
      "loss": 0.0008,
      "step": 296550
    },
    {
      "epoch": 1.902043722120494,
      "grad_norm": 0.024213319644331932,
      "learning_rate": 7.288693766725253e-08,
      "loss": 0.0011,
      "step": 296560
    },
    {
      "epoch": 1.9021078590142801,
      "grad_norm": 0.06082941219210625,
      "learning_rate": 7.279174939705536e-08,
      "loss": 0.0021,
      "step": 296570
    },
    {
      "epoch": 1.9021719959080663,
      "grad_norm": 0.10337240248918533,
      "learning_rate": 7.26966228682996e-08,
      "loss": 0.0013,
      "step": 296580
    },
    {
      "epoch": 1.9022361328018522,
      "grad_norm": 0.2524295449256897,
      "learning_rate": 7.260155808217706e-08,
      "loss": 0.0011,
      "step": 296590
    },
    {
      "epoch": 1.9023002696956384,
      "grad_norm": 0.05802197381854057,
      "learning_rate": 7.250655503987792e-08,
      "loss": 0.0007,
      "step": 296600
    },
    {
      "epoch": 1.9023644065894245,
      "grad_norm": 0.06177850067615509,
      "learning_rate": 7.241161374259343e-08,
      "loss": 0.0011,
      "step": 296610
    },
    {
      "epoch": 1.9024285434832104,
      "grad_norm": 0.1176663413643837,
      "learning_rate": 7.23167341915132e-08,
      "loss": 0.0006,
      "step": 296620
    },
    {
      "epoch": 1.9024926803769966,
      "grad_norm": 0.12713593244552612,
      "learning_rate": 7.222191638782572e-08,
      "loss": 0.0011,
      "step": 296630
    },
    {
      "epoch": 1.9025568172707827,
      "grad_norm": 0.006985232699662447,
      "learning_rate": 7.212716033272005e-08,
      "loss": 0.001,
      "step": 296640
    },
    {
      "epoch": 1.902620954164569,
      "grad_norm": 0.12933291494846344,
      "learning_rate": 7.203246602738245e-08,
      "loss": 0.0012,
      "step": 296650
    },
    {
      "epoch": 1.902685091058355,
      "grad_norm": 0.016291489824652672,
      "learning_rate": 7.193783347300032e-08,
      "loss": 0.0006,
      "step": 296660
    },
    {
      "epoch": 1.9027492279521412,
      "grad_norm": 0.01926431432366371,
      "learning_rate": 7.184326267075936e-08,
      "loss": 0.001,
      "step": 296670
    },
    {
      "epoch": 1.9028133648459271,
      "grad_norm": 0.024936731904745102,
      "learning_rate": 7.174875362184363e-08,
      "loss": 0.0013,
      "step": 296680
    },
    {
      "epoch": 1.9028775017397133,
      "grad_norm": 0.0974646806716919,
      "learning_rate": 7.165430632743886e-08,
      "loss": 0.0021,
      "step": 296690
    },
    {
      "epoch": 1.9029416386334992,
      "grad_norm": 0.08997154980897903,
      "learning_rate": 7.155992078872742e-08,
      "loss": 0.0003,
      "step": 296700
    },
    {
      "epoch": 1.9030057755272853,
      "grad_norm": 0.08475396037101746,
      "learning_rate": 7.146559700689337e-08,
      "loss": 0.002,
      "step": 296710
    },
    {
      "epoch": 1.9030699124210715,
      "grad_norm": 0.00956734549254179,
      "learning_rate": 7.137133498311633e-08,
      "loss": 0.002,
      "step": 296720
    },
    {
      "epoch": 1.9031340493148576,
      "grad_norm": 0.04105890542268753,
      "learning_rate": 7.127713471857977e-08,
      "loss": 0.0011,
      "step": 296730
    },
    {
      "epoch": 1.9031981862086438,
      "grad_norm": 0.04018397629261017,
      "learning_rate": 7.118299621446334e-08,
      "loss": 0.001,
      "step": 296740
    },
    {
      "epoch": 1.90326232310243,
      "grad_norm": 0.19481207430362701,
      "learning_rate": 7.108891947194662e-08,
      "loss": 0.0012,
      "step": 296750
    },
    {
      "epoch": 1.903326459996216,
      "grad_norm": 0.05149256810545921,
      "learning_rate": 7.099490449220758e-08,
      "loss": 0.001,
      "step": 296760
    },
    {
      "epoch": 1.903390596890002,
      "grad_norm": 0.04618384689092636,
      "learning_rate": 7.090095127642582e-08,
      "loss": 0.0012,
      "step": 296770
    },
    {
      "epoch": 1.9034547337837882,
      "grad_norm": 0.13499963283538818,
      "learning_rate": 7.080705982577817e-08,
      "loss": 0.0017,
      "step": 296780
    },
    {
      "epoch": 1.903518870677574,
      "grad_norm": 0.07416088879108429,
      "learning_rate": 7.071323014144039e-08,
      "loss": 0.0014,
      "step": 296790
    },
    {
      "epoch": 1.9035830075713602,
      "grad_norm": 0.041787635535001755,
      "learning_rate": 7.061946222458871e-08,
      "loss": 0.0014,
      "step": 296800
    },
    {
      "epoch": 1.9036471444651464,
      "grad_norm": 0.13607001304626465,
      "learning_rate": 7.052575607639833e-08,
      "loss": 0.0012,
      "step": 296810
    },
    {
      "epoch": 1.9037112813589325,
      "grad_norm": 0.0958867073059082,
      "learning_rate": 7.043211169804332e-08,
      "loss": 0.001,
      "step": 296820
    },
    {
      "epoch": 1.9037754182527187,
      "grad_norm": 0.14104056358337402,
      "learning_rate": 7.03385290906966e-08,
      "loss": 0.0008,
      "step": 296830
    },
    {
      "epoch": 1.9038395551465048,
      "grad_norm": 0.1130591556429863,
      "learning_rate": 7.024500825553172e-08,
      "loss": 0.0034,
      "step": 296840
    },
    {
      "epoch": 1.9039036920402908,
      "grad_norm": 0.10772310197353363,
      "learning_rate": 7.015154919371991e-08,
      "loss": 0.0011,
      "step": 296850
    },
    {
      "epoch": 1.903967828934077,
      "grad_norm": 0.13072291016578674,
      "learning_rate": 7.00581519064325e-08,
      "loss": 0.001,
      "step": 296860
    },
    {
      "epoch": 1.9040319658278628,
      "grad_norm": 0.050700295716524124,
      "learning_rate": 6.996481639483966e-08,
      "loss": 0.0011,
      "step": 296870
    },
    {
      "epoch": 1.904096102721649,
      "grad_norm": 0.026685891672968864,
      "learning_rate": 6.987154266011154e-08,
      "loss": 0.0012,
      "step": 296880
    },
    {
      "epoch": 1.9041602396154351,
      "grad_norm": 0.007372669875621796,
      "learning_rate": 6.977833070341667e-08,
      "loss": 0.002,
      "step": 296890
    },
    {
      "epoch": 1.9042243765092213,
      "grad_norm": 0.16585969924926758,
      "learning_rate": 6.968518052592244e-08,
      "loss": 0.0037,
      "step": 296900
    },
    {
      "epoch": 1.9042885134030074,
      "grad_norm": 0.011847295798361301,
      "learning_rate": 6.959209212879625e-08,
      "loss": 0.0007,
      "step": 296910
    },
    {
      "epoch": 1.9043526502967936,
      "grad_norm": 0.05392090231180191,
      "learning_rate": 6.949906551320551e-08,
      "loss": 0.0007,
      "step": 296920
    },
    {
      "epoch": 1.9044167871905797,
      "grad_norm": 0.04062240198254585,
      "learning_rate": 6.940610068031484e-08,
      "loss": 0.0032,
      "step": 296930
    },
    {
      "epoch": 1.9044809240843656,
      "grad_norm": 0.07179093360900879,
      "learning_rate": 6.931319763128997e-08,
      "loss": 0.0015,
      "step": 296940
    },
    {
      "epoch": 1.9045450609781518,
      "grad_norm": 0.18170824646949768,
      "learning_rate": 6.922035636729441e-08,
      "loss": 0.0013,
      "step": 296950
    },
    {
      "epoch": 1.9046091978719377,
      "grad_norm": 0.13805805146694183,
      "learning_rate": 6.912757688949167e-08,
      "loss": 0.001,
      "step": 296960
    },
    {
      "epoch": 1.9046733347657239,
      "grad_norm": 0.08730573207139969,
      "learning_rate": 6.903485919904474e-08,
      "loss": 0.0014,
      "step": 296970
    },
    {
      "epoch": 1.90473747165951,
      "grad_norm": 0.11466261744499207,
      "learning_rate": 6.894220329711432e-08,
      "loss": 0.001,
      "step": 296980
    },
    {
      "epoch": 1.9048016085532962,
      "grad_norm": 0.056746866554021835,
      "learning_rate": 6.884960918486339e-08,
      "loss": 0.0007,
      "step": 296990
    },
    {
      "epoch": 1.9048657454470823,
      "grad_norm": 0.06870730966329575,
      "learning_rate": 6.875707686345045e-08,
      "loss": 0.0007,
      "step": 297000
    },
    {
      "epoch": 1.9049298823408685,
      "grad_norm": 0.12024735659360886,
      "learning_rate": 6.866460633403571e-08,
      "loss": 0.0007,
      "step": 297010
    },
    {
      "epoch": 1.9049940192346546,
      "grad_norm": 0.0668744146823883,
      "learning_rate": 6.857219759777767e-08,
      "loss": 0.0011,
      "step": 297020
    },
    {
      "epoch": 1.9050581561284405,
      "grad_norm": 0.15155884623527527,
      "learning_rate": 6.847985065583485e-08,
      "loss": 0.0012,
      "step": 297030
    },
    {
      "epoch": 1.9051222930222267,
      "grad_norm": 0.0582219660282135,
      "learning_rate": 6.83875655093641e-08,
      "loss": 0.0005,
      "step": 297040
    },
    {
      "epoch": 1.9051864299160126,
      "grad_norm": 0.08068463206291199,
      "learning_rate": 6.829534215952116e-08,
      "loss": 0.0008,
      "step": 297050
    },
    {
      "epoch": 1.9052505668097988,
      "grad_norm": 0.08450084924697876,
      "learning_rate": 6.820318060746234e-08,
      "loss": 0.0014,
      "step": 297060
    },
    {
      "epoch": 1.905314703703585,
      "grad_norm": 0.0036552983801811934,
      "learning_rate": 6.811108085434282e-08,
      "loss": 0.0004,
      "step": 297070
    },
    {
      "epoch": 1.905378840597371,
      "grad_norm": 0.02461552433669567,
      "learning_rate": 6.801904290131556e-08,
      "loss": 0.0025,
      "step": 297080
    },
    {
      "epoch": 1.9054429774911572,
      "grad_norm": 0.0689999908208847,
      "learning_rate": 6.792706674953519e-08,
      "loss": 0.0011,
      "step": 297090
    },
    {
      "epoch": 1.9055071143849434,
      "grad_norm": 0.15311597287654877,
      "learning_rate": 6.783515240015304e-08,
      "loss": 0.0014,
      "step": 297100
    },
    {
      "epoch": 1.9055712512787293,
      "grad_norm": 0.05924878641963005,
      "learning_rate": 6.774329985432149e-08,
      "loss": 0.0018,
      "step": 297110
    },
    {
      "epoch": 1.9056353881725154,
      "grad_norm": 0.06728945672512054,
      "learning_rate": 6.765150911319185e-08,
      "loss": 0.001,
      "step": 297120
    },
    {
      "epoch": 1.9056995250663014,
      "grad_norm": 0.04347103834152222,
      "learning_rate": 6.755978017791321e-08,
      "loss": 0.0008,
      "step": 297130
    },
    {
      "epoch": 1.9057636619600875,
      "grad_norm": 0.07253609597682953,
      "learning_rate": 6.74681130496363e-08,
      "loss": 0.0008,
      "step": 297140
    },
    {
      "epoch": 1.9058277988538737,
      "grad_norm": 0.035815589129924774,
      "learning_rate": 6.73765077295091e-08,
      "loss": 0.0009,
      "step": 297150
    },
    {
      "epoch": 1.9058919357476598,
      "grad_norm": 0.11661411821842194,
      "learning_rate": 6.728496421867902e-08,
      "loss": 0.0009,
      "step": 297160
    },
    {
      "epoch": 1.905956072641446,
      "grad_norm": 0.10119163990020752,
      "learning_rate": 6.719348251829405e-08,
      "loss": 0.0007,
      "step": 297170
    },
    {
      "epoch": 1.906020209535232,
      "grad_norm": 0.14704670011997223,
      "learning_rate": 6.710206262950047e-08,
      "loss": 0.0006,
      "step": 297180
    },
    {
      "epoch": 1.9060843464290183,
      "grad_norm": 0.03064277581870556,
      "learning_rate": 6.701070455344294e-08,
      "loss": 0.0008,
      "step": 297190
    },
    {
      "epoch": 1.9061484833228042,
      "grad_norm": 0.01659799925982952,
      "learning_rate": 6.69194082912672e-08,
      "loss": 0.0006,
      "step": 297200
    },
    {
      "epoch": 1.9062126202165903,
      "grad_norm": 0.1433970034122467,
      "learning_rate": 6.682817384411677e-08,
      "loss": 0.0008,
      "step": 297210
    },
    {
      "epoch": 1.9062767571103763,
      "grad_norm": 0.028640659525990486,
      "learning_rate": 6.673700121313464e-08,
      "loss": 0.0008,
      "step": 297220
    },
    {
      "epoch": 1.9063408940041624,
      "grad_norm": 0.060921695083379745,
      "learning_rate": 6.664589039946434e-08,
      "loss": 0.0007,
      "step": 297230
    },
    {
      "epoch": 1.9064050308979485,
      "grad_norm": 0.08914317190647125,
      "learning_rate": 6.655484140424661e-08,
      "loss": 0.0007,
      "step": 297240
    },
    {
      "epoch": 1.9064691677917347,
      "grad_norm": 0.03564678505063057,
      "learning_rate": 6.646385422862223e-08,
      "loss": 0.0004,
      "step": 297250
    },
    {
      "epoch": 1.9065333046855208,
      "grad_norm": 0.05976053699851036,
      "learning_rate": 6.637292887373248e-08,
      "loss": 0.0008,
      "step": 297260
    },
    {
      "epoch": 1.906597441579307,
      "grad_norm": 0.13643500208854675,
      "learning_rate": 6.628206534071535e-08,
      "loss": 0.0016,
      "step": 297270
    },
    {
      "epoch": 1.906661578473093,
      "grad_norm": 0.040523845702409744,
      "learning_rate": 6.619126363071049e-08,
      "loss": 0.001,
      "step": 297280
    },
    {
      "epoch": 1.906725715366879,
      "grad_norm": 0.05647537112236023,
      "learning_rate": 6.61005237448542e-08,
      "loss": 0.001,
      "step": 297290
    },
    {
      "epoch": 1.9067898522606652,
      "grad_norm": 0.002782865660265088,
      "learning_rate": 6.600984568428559e-08,
      "loss": 0.0016,
      "step": 297300
    },
    {
      "epoch": 1.9068539891544511,
      "grad_norm": 0.004738634917885065,
      "learning_rate": 6.591922945013984e-08,
      "loss": 0.0014,
      "step": 297310
    },
    {
      "epoch": 1.9069181260482373,
      "grad_norm": 0.04857382923364639,
      "learning_rate": 6.582867504355217e-08,
      "loss": 0.0014,
      "step": 297320
    },
    {
      "epoch": 1.9069822629420234,
      "grad_norm": 0.00765692163258791,
      "learning_rate": 6.573818246565722e-08,
      "loss": 0.0015,
      "step": 297330
    },
    {
      "epoch": 1.9070463998358096,
      "grad_norm": 0.016255423426628113,
      "learning_rate": 6.564775171758964e-08,
      "loss": 0.0009,
      "step": 297340
    },
    {
      "epoch": 1.9071105367295957,
      "grad_norm": 0.12796856462955475,
      "learning_rate": 6.55573828004824e-08,
      "loss": 0.0009,
      "step": 297350
    },
    {
      "epoch": 1.9071746736233819,
      "grad_norm": 0.01191774196922779,
      "learning_rate": 6.546707571546796e-08,
      "loss": 0.0012,
      "step": 297360
    },
    {
      "epoch": 1.9072388105171678,
      "grad_norm": 0.08716825395822525,
      "learning_rate": 6.537683046367704e-08,
      "loss": 0.0015,
      "step": 297370
    },
    {
      "epoch": 1.907302947410954,
      "grad_norm": 0.18957430124282837,
      "learning_rate": 6.528664704624155e-08,
      "loss": 0.0012,
      "step": 297380
    },
    {
      "epoch": 1.90736708430474,
      "grad_norm": 0.11020193248987198,
      "learning_rate": 6.519652546429167e-08,
      "loss": 0.0011,
      "step": 297390
    },
    {
      "epoch": 1.907431221198526,
      "grad_norm": 0.02139066718518734,
      "learning_rate": 6.51064657189554e-08,
      "loss": 0.0008,
      "step": 297400
    },
    {
      "epoch": 1.9074953580923122,
      "grad_norm": 0.0030964526813477278,
      "learning_rate": 6.50164678113624e-08,
      "loss": 0.0004,
      "step": 297410
    },
    {
      "epoch": 1.9075594949860983,
      "grad_norm": 0.024820616468787193,
      "learning_rate": 6.492653174263951e-08,
      "loss": 0.002,
      "step": 297420
    },
    {
      "epoch": 1.9076236318798845,
      "grad_norm": 0.04223772883415222,
      "learning_rate": 6.483665751391477e-08,
      "loss": 0.0019,
      "step": 297430
    },
    {
      "epoch": 1.9076877687736706,
      "grad_norm": 0.0062083834782242775,
      "learning_rate": 6.47468451263139e-08,
      "loss": 0.0009,
      "step": 297440
    },
    {
      "epoch": 1.9077519056674568,
      "grad_norm": 0.08388002961874008,
      "learning_rate": 6.465709458096214e-08,
      "loss": 0.0023,
      "step": 297450
    },
    {
      "epoch": 1.9078160425612427,
      "grad_norm": 0.037969715893268585,
      "learning_rate": 6.456740587898414e-08,
      "loss": 0.0017,
      "step": 297460
    },
    {
      "epoch": 1.9078801794550289,
      "grad_norm": 0.1530952900648117,
      "learning_rate": 6.447777902150398e-08,
      "loss": 0.0021,
      "step": 297470
    },
    {
      "epoch": 1.9079443163488148,
      "grad_norm": 0.09825599193572998,
      "learning_rate": 6.438821400964412e-08,
      "loss": 0.0009,
      "step": 297480
    },
    {
      "epoch": 1.908008453242601,
      "grad_norm": 0.09959319233894348,
      "learning_rate": 6.429871084452755e-08,
      "loss": 0.001,
      "step": 297490
    },
    {
      "epoch": 1.908072590136387,
      "grad_norm": 0.1308680921792984,
      "learning_rate": 6.420926952727613e-08,
      "loss": 0.0016,
      "step": 297500
    },
    {
      "epoch": 1.9081367270301732,
      "grad_norm": 0.07021002471446991,
      "learning_rate": 6.411989005900954e-08,
      "loss": 0.0008,
      "step": 297510
    },
    {
      "epoch": 1.9082008639239594,
      "grad_norm": 0.0220838263630867,
      "learning_rate": 6.403057244084854e-08,
      "loss": 0.0012,
      "step": 297520
    },
    {
      "epoch": 1.9082650008177455,
      "grad_norm": 0.10355224460363388,
      "learning_rate": 6.394131667391224e-08,
      "loss": 0.0007,
      "step": 297530
    },
    {
      "epoch": 1.9083291377115315,
      "grad_norm": 0.011518482118844986,
      "learning_rate": 6.385212275931862e-08,
      "loss": 0.0009,
      "step": 297540
    },
    {
      "epoch": 1.9083932746053176,
      "grad_norm": 0.1443081796169281,
      "learning_rate": 6.376299069818626e-08,
      "loss": 0.0006,
      "step": 297550
    },
    {
      "epoch": 1.9084574114991035,
      "grad_norm": 0.07212929427623749,
      "learning_rate": 6.36739204916309e-08,
      "loss": 0.0009,
      "step": 297560
    },
    {
      "epoch": 1.9085215483928897,
      "grad_norm": 0.06594572961330414,
      "learning_rate": 6.358491214077e-08,
      "loss": 0.0008,
      "step": 297570
    },
    {
      "epoch": 1.9085856852866758,
      "grad_norm": 0.03384760767221451,
      "learning_rate": 6.349596564671767e-08,
      "loss": 0.0009,
      "step": 297580
    },
    {
      "epoch": 1.908649822180462,
      "grad_norm": 0.030025290325284004,
      "learning_rate": 6.340708101058913e-08,
      "loss": 0.0014,
      "step": 297590
    },
    {
      "epoch": 1.9087139590742481,
      "grad_norm": 0.08829595893621445,
      "learning_rate": 6.331825823349791e-08,
      "loss": 0.001,
      "step": 297600
    },
    {
      "epoch": 1.9087780959680343,
      "grad_norm": 0.05397048965096474,
      "learning_rate": 6.322949731655704e-08,
      "loss": 0.0006,
      "step": 297610
    },
    {
      "epoch": 1.9088422328618204,
      "grad_norm": 0.14397847652435303,
      "learning_rate": 6.314079826087894e-08,
      "loss": 0.001,
      "step": 297620
    },
    {
      "epoch": 1.9089063697556063,
      "grad_norm": 0.2537747621536255,
      "learning_rate": 6.305216106757494e-08,
      "loss": 0.0019,
      "step": 297630
    },
    {
      "epoch": 1.9089705066493925,
      "grad_norm": 0.09246334433555603,
      "learning_rate": 6.296358573775585e-08,
      "loss": 0.0009,
      "step": 297640
    },
    {
      "epoch": 1.9090346435431784,
      "grad_norm": 0.006885716691613197,
      "learning_rate": 6.287507227253131e-08,
      "loss": 0.0003,
      "step": 297650
    },
    {
      "epoch": 1.9090987804369646,
      "grad_norm": 0.06697414070367813,
      "learning_rate": 6.278662067301044e-08,
      "loss": 0.0009,
      "step": 297660
    },
    {
      "epoch": 1.9091629173307507,
      "grad_norm": 0.10258012264966965,
      "learning_rate": 6.269823094030181e-08,
      "loss": 0.0009,
      "step": 297670
    },
    {
      "epoch": 1.9092270542245369,
      "grad_norm": 0.08741557598114014,
      "learning_rate": 6.26099030755134e-08,
      "loss": 0.0012,
      "step": 297680
    },
    {
      "epoch": 1.909291191118323,
      "grad_norm": 0.07768207788467407,
      "learning_rate": 6.252163707975157e-08,
      "loss": 0.0008,
      "step": 297690
    },
    {
      "epoch": 1.9093553280121092,
      "grad_norm": 0.1813802868127823,
      "learning_rate": 6.243343295412207e-08,
      "loss": 0.0013,
      "step": 297700
    },
    {
      "epoch": 1.909419464905895,
      "grad_norm": 0.07702813297510147,
      "learning_rate": 6.234529069973071e-08,
      "loss": 0.0014,
      "step": 297710
    },
    {
      "epoch": 1.9094836017996812,
      "grad_norm": 0.1788177788257599,
      "learning_rate": 6.225721031768162e-08,
      "loss": 0.0014,
      "step": 297720
    },
    {
      "epoch": 1.9095477386934674,
      "grad_norm": 0.03819502145051956,
      "learning_rate": 6.216919180907888e-08,
      "loss": 0.0005,
      "step": 297730
    },
    {
      "epoch": 1.9096118755872533,
      "grad_norm": 0.07616864889860153,
      "learning_rate": 6.208123517502441e-08,
      "loss": 0.0008,
      "step": 297740
    },
    {
      "epoch": 1.9096760124810395,
      "grad_norm": 0.09854179620742798,
      "learning_rate": 6.199334041662175e-08,
      "loss": 0.0008,
      "step": 297750
    },
    {
      "epoch": 1.9097401493748256,
      "grad_norm": 0.1113155409693718,
      "learning_rate": 6.190550753497171e-08,
      "loss": 0.0013,
      "step": 297760
    },
    {
      "epoch": 1.9098042862686118,
      "grad_norm": 0.03208629786968231,
      "learning_rate": 6.181773653117506e-08,
      "loss": 0.0004,
      "step": 297770
    },
    {
      "epoch": 1.909868423162398,
      "grad_norm": 0.03522845730185509,
      "learning_rate": 6.173002740633095e-08,
      "loss": 0.0007,
      "step": 297780
    },
    {
      "epoch": 1.909932560056184,
      "grad_norm": 0.1295401155948639,
      "learning_rate": 6.164238016153901e-08,
      "loss": 0.0005,
      "step": 297790
    },
    {
      "epoch": 1.90999669694997,
      "grad_norm": 0.04841848090291023,
      "learning_rate": 6.155479479789783e-08,
      "loss": 0.0013,
      "step": 297800
    },
    {
      "epoch": 1.9100608338437561,
      "grad_norm": 0.10158253461122513,
      "learning_rate": 6.146727131650432e-08,
      "loss": 0.0014,
      "step": 297810
    },
    {
      "epoch": 1.910124970737542,
      "grad_norm": 0.024595726281404495,
      "learning_rate": 6.137980971845536e-08,
      "loss": 0.0016,
      "step": 297820
    },
    {
      "epoch": 1.9101891076313282,
      "grad_norm": 0.044377777725458145,
      "learning_rate": 6.129241000484676e-08,
      "loss": 0.0013,
      "step": 297830
    },
    {
      "epoch": 1.9102532445251144,
      "grad_norm": 0.13128231465816498,
      "learning_rate": 6.120507217677429e-08,
      "loss": 0.0011,
      "step": 297840
    },
    {
      "epoch": 1.9103173814189005,
      "grad_norm": 0.029137341305613518,
      "learning_rate": 6.111779623533154e-08,
      "loss": 0.0007,
      "step": 297850
    },
    {
      "epoch": 1.9103815183126867,
      "grad_norm": 0.043349724262952805,
      "learning_rate": 6.103058218161262e-08,
      "loss": 0.0014,
      "step": 297860
    },
    {
      "epoch": 1.9104456552064728,
      "grad_norm": 0.0957290530204773,
      "learning_rate": 6.094343001671055e-08,
      "loss": 0.0012,
      "step": 297870
    },
    {
      "epoch": 1.910509792100259,
      "grad_norm": 0.149446502327919,
      "learning_rate": 6.085633974171778e-08,
      "loss": 0.0019,
      "step": 297880
    },
    {
      "epoch": 1.9105739289940449,
      "grad_norm": 0.032342396676540375,
      "learning_rate": 6.076931135772402e-08,
      "loss": 0.0014,
      "step": 297890
    },
    {
      "epoch": 1.910638065887831,
      "grad_norm": 0.04412601888179779,
      "learning_rate": 6.068234486582113e-08,
      "loss": 0.0007,
      "step": 297900
    },
    {
      "epoch": 1.910702202781617,
      "grad_norm": 0.04877545312047005,
      "learning_rate": 6.059544026709885e-08,
      "loss": 0.001,
      "step": 297910
    },
    {
      "epoch": 1.910766339675403,
      "grad_norm": 0.08056735247373581,
      "learning_rate": 6.050859756264571e-08,
      "loss": 0.0007,
      "step": 297920
    },
    {
      "epoch": 1.9108304765691893,
      "grad_norm": 0.043473027646541595,
      "learning_rate": 6.042181675354975e-08,
      "loss": 0.0007,
      "step": 297930
    },
    {
      "epoch": 1.9108946134629754,
      "grad_norm": 0.011658878065645695,
      "learning_rate": 6.033509784089897e-08,
      "loss": 0.0006,
      "step": 297940
    },
    {
      "epoch": 1.9109587503567615,
      "grad_norm": 0.030716655775904655,
      "learning_rate": 6.024844082577919e-08,
      "loss": 0.0021,
      "step": 297950
    },
    {
      "epoch": 1.9110228872505477,
      "grad_norm": 0.002981361001729965,
      "learning_rate": 6.01618457092773e-08,
      "loss": 0.0014,
      "step": 297960
    },
    {
      "epoch": 1.9110870241443336,
      "grad_norm": 0.008524656295776367,
      "learning_rate": 6.007531249247744e-08,
      "loss": 0.0023,
      "step": 297970
    },
    {
      "epoch": 1.9111511610381198,
      "grad_norm": 0.03623645007610321,
      "learning_rate": 5.998884117646542e-08,
      "loss": 0.0014,
      "step": 297980
    },
    {
      "epoch": 1.9112152979319057,
      "grad_norm": 0.13329999148845673,
      "learning_rate": 5.990243176232313e-08,
      "loss": 0.001,
      "step": 297990
    },
    {
      "epoch": 1.9112794348256918,
      "grad_norm": 0.15252041816711426,
      "learning_rate": 5.981608425113416e-08,
      "loss": 0.0011,
      "step": 298000
    },
    {
      "epoch": 1.911343571719478,
      "grad_norm": 0.07557599246501923,
      "learning_rate": 5.972979864397988e-08,
      "loss": 0.0007,
      "step": 298010
    },
    {
      "epoch": 1.9114077086132641,
      "grad_norm": 0.032776590436697006,
      "learning_rate": 5.964357494194273e-08,
      "loss": 0.0012,
      "step": 298020
    },
    {
      "epoch": 1.9114718455070503,
      "grad_norm": 0.16524513065814972,
      "learning_rate": 5.9557413146102415e-08,
      "loss": 0.0033,
      "step": 298030
    },
    {
      "epoch": 1.9115359824008364,
      "grad_norm": 0.13968272507190704,
      "learning_rate": 5.947131325753808e-08,
      "loss": 0.0017,
      "step": 298040
    },
    {
      "epoch": 1.9116001192946226,
      "grad_norm": 0.07123354822397232,
      "learning_rate": 5.9385275277329404e-08,
      "loss": 0.0015,
      "step": 298050
    },
    {
      "epoch": 1.9116642561884085,
      "grad_norm": 0.17939843237400055,
      "learning_rate": 5.929929920655386e-08,
      "loss": 0.001,
      "step": 298060
    },
    {
      "epoch": 1.9117283930821947,
      "grad_norm": 0.09192238003015518,
      "learning_rate": 5.921338504629004e-08,
      "loss": 0.0008,
      "step": 298070
    },
    {
      "epoch": 1.9117925299759806,
      "grad_norm": 0.07744672894477844,
      "learning_rate": 5.912753279761263e-08,
      "loss": 0.0012,
      "step": 298080
    },
    {
      "epoch": 1.9118566668697667,
      "grad_norm": 0.10017392039299011,
      "learning_rate": 5.9041742461599105e-08,
      "loss": 0.001,
      "step": 298090
    },
    {
      "epoch": 1.911920803763553,
      "grad_norm": 0.0025806869380176067,
      "learning_rate": 5.8956014039323604e-08,
      "loss": 0.0007,
      "step": 298100
    },
    {
      "epoch": 1.911984940657339,
      "grad_norm": 0.04378533735871315,
      "learning_rate": 5.887034753186027e-08,
      "loss": 0.0009,
      "step": 298110
    },
    {
      "epoch": 1.9120490775511252,
      "grad_norm": 0.06870028376579285,
      "learning_rate": 5.878474294028269e-08,
      "loss": 0.001,
      "step": 298120
    },
    {
      "epoch": 1.9121132144449113,
      "grad_norm": 0.08500008285045624,
      "learning_rate": 5.8699200265664445e-08,
      "loss": 0.0008,
      "step": 298130
    },
    {
      "epoch": 1.9121773513386973,
      "grad_norm": 0.03247741609811783,
      "learning_rate": 5.861371950907635e-08,
      "loss": 0.0012,
      "step": 298140
    },
    {
      "epoch": 1.9122414882324834,
      "grad_norm": 0.007751672528684139,
      "learning_rate": 5.8528300671590324e-08,
      "loss": 0.0008,
      "step": 298150
    },
    {
      "epoch": 1.9123056251262696,
      "grad_norm": 0.054278917610645294,
      "learning_rate": 5.844294375427551e-08,
      "loss": 0.0006,
      "step": 298160
    },
    {
      "epoch": 1.9123697620200555,
      "grad_norm": 0.056611429899930954,
      "learning_rate": 5.835764875820271e-08,
      "loss": 0.0015,
      "step": 298170
    },
    {
      "epoch": 1.9124338989138416,
      "grad_norm": 0.1559065580368042,
      "learning_rate": 5.827241568444053e-08,
      "loss": 0.0011,
      "step": 298180
    },
    {
      "epoch": 1.9124980358076278,
      "grad_norm": 0.006949694361537695,
      "learning_rate": 5.818724453405644e-08,
      "loss": 0.0008,
      "step": 298190
    },
    {
      "epoch": 1.912562172701414,
      "grad_norm": 0.08792159706354141,
      "learning_rate": 5.810213530811792e-08,
      "loss": 0.0016,
      "step": 298200
    },
    {
      "epoch": 1.9126263095952,
      "grad_norm": 0.13154152035713196,
      "learning_rate": 5.8017088007691904e-08,
      "loss": 0.0006,
      "step": 298210
    },
    {
      "epoch": 1.9126904464889862,
      "grad_norm": 0.1045413538813591,
      "learning_rate": 5.7932102633843633e-08,
      "loss": 0.0017,
      "step": 298220
    },
    {
      "epoch": 1.9127545833827722,
      "grad_norm": 0.15327368676662445,
      "learning_rate": 5.7847179187637825e-08,
      "loss": 0.0008,
      "step": 298230
    },
    {
      "epoch": 1.9128187202765583,
      "grad_norm": 0.015088552609086037,
      "learning_rate": 5.7762317670139734e-08,
      "loss": 0.0006,
      "step": 298240
    },
    {
      "epoch": 1.9128828571703442,
      "grad_norm": 0.08311817795038223,
      "learning_rate": 5.767751808241129e-08,
      "loss": 0.0011,
      "step": 298250
    },
    {
      "epoch": 1.9129469940641304,
      "grad_norm": 0.02838069386780262,
      "learning_rate": 5.759278042551664e-08,
      "loss": 0.0004,
      "step": 298260
    },
    {
      "epoch": 1.9130111309579165,
      "grad_norm": 0.045654039829969406,
      "learning_rate": 5.7508104700515486e-08,
      "loss": 0.0006,
      "step": 298270
    },
    {
      "epoch": 1.9130752678517027,
      "grad_norm": 0.08993151038885117,
      "learning_rate": 5.742349090847088e-08,
      "loss": 0.0005,
      "step": 298280
    },
    {
      "epoch": 1.9131394047454888,
      "grad_norm": 0.013555367477238178,
      "learning_rate": 5.7338939050442524e-08,
      "loss": 0.0007,
      "step": 298290
    },
    {
      "epoch": 1.913203541639275,
      "grad_norm": 0.0012622076319530606,
      "learning_rate": 5.725444912748956e-08,
      "loss": 0.0007,
      "step": 298300
    },
    {
      "epoch": 1.9132676785330611,
      "grad_norm": 0.07890114933252335,
      "learning_rate": 5.717002114067005e-08,
      "loss": 0.0006,
      "step": 298310
    },
    {
      "epoch": 1.913331815426847,
      "grad_norm": 0.09559661149978638,
      "learning_rate": 5.708565509104369e-08,
      "loss": 0.0013,
      "step": 298320
    },
    {
      "epoch": 1.9133959523206332,
      "grad_norm": 0.36704665422439575,
      "learning_rate": 5.700135097966686e-08,
      "loss": 0.0031,
      "step": 298330
    },
    {
      "epoch": 1.9134600892144191,
      "grad_norm": 0.109199658036232,
      "learning_rate": 5.691710880759538e-08,
      "loss": 0.0008,
      "step": 298340
    },
    {
      "epoch": 1.9135242261082053,
      "grad_norm": 0.0839865654706955,
      "learning_rate": 5.683292857588507e-08,
      "loss": 0.0013,
      "step": 298350
    },
    {
      "epoch": 1.9135883630019914,
      "grad_norm": 0.13686271011829376,
      "learning_rate": 5.674881028559121e-08,
      "loss": 0.0008,
      "step": 298360
    },
    {
      "epoch": 1.9136524998957776,
      "grad_norm": 0.17503884434700012,
      "learning_rate": 5.666475393776793e-08,
      "loss": 0.0012,
      "step": 298370
    },
    {
      "epoch": 1.9137166367895637,
      "grad_norm": 0.023506227880716324,
      "learning_rate": 5.658075953346776e-08,
      "loss": 0.0012,
      "step": 298380
    },
    {
      "epoch": 1.9137807736833499,
      "grad_norm": 0.007835052907466888,
      "learning_rate": 5.649682707374371e-08,
      "loss": 0.0024,
      "step": 298390
    },
    {
      "epoch": 1.9138449105771358,
      "grad_norm": 0.017951978370547295,
      "learning_rate": 5.6412956559647734e-08,
      "loss": 0.0008,
      "step": 298400
    },
    {
      "epoch": 1.913909047470922,
      "grad_norm": 0.013206799514591694,
      "learning_rate": 5.632914799223066e-08,
      "loss": 0.001,
      "step": 298410
    },
    {
      "epoch": 1.9139731843647079,
      "grad_norm": 0.09789872169494629,
      "learning_rate": 5.6245401372542193e-08,
      "loss": 0.0016,
      "step": 298420
    },
    {
      "epoch": 1.914037321258494,
      "grad_norm": 0.08239579200744629,
      "learning_rate": 5.616171670163262e-08,
      "loss": 0.0008,
      "step": 298430
    },
    {
      "epoch": 1.9141014581522802,
      "grad_norm": 0.21788769960403442,
      "learning_rate": 5.607809398054998e-08,
      "loss": 0.0007,
      "step": 298440
    },
    {
      "epoch": 1.9141655950460663,
      "grad_norm": 0.05247426778078079,
      "learning_rate": 5.599453321034176e-08,
      "loss": 0.002,
      "step": 298450
    },
    {
      "epoch": 1.9142297319398525,
      "grad_norm": 0.1064145565032959,
      "learning_rate": 5.5911034392055474e-08,
      "loss": 0.0013,
      "step": 298460
    },
    {
      "epoch": 1.9142938688336386,
      "grad_norm": 0.05088728293776512,
      "learning_rate": 5.58275975267375e-08,
      "loss": 0.0008,
      "step": 298470
    },
    {
      "epoch": 1.9143580057274248,
      "grad_norm": 0.04615416377782822,
      "learning_rate": 5.574422261543366e-08,
      "loss": 0.0025,
      "step": 298480
    },
    {
      "epoch": 1.9144221426212107,
      "grad_norm": 0.1170719638466835,
      "learning_rate": 5.566090965918814e-08,
      "loss": 0.0013,
      "step": 298490
    },
    {
      "epoch": 1.9144862795149968,
      "grad_norm": 0.1538655310869217,
      "learning_rate": 5.557765865904452e-08,
      "loss": 0.0007,
      "step": 298500
    },
    {
      "epoch": 1.9145504164087828,
      "grad_norm": 0.02719215862452984,
      "learning_rate": 5.5494469616046984e-08,
      "loss": 0.0006,
      "step": 298510
    },
    {
      "epoch": 1.914614553302569,
      "grad_norm": 0.008994230069220066,
      "learning_rate": 5.541134253123748e-08,
      "loss": 0.0013,
      "step": 298520
    },
    {
      "epoch": 1.914678690196355,
      "grad_norm": 0.013056914322078228,
      "learning_rate": 5.532827740565794e-08,
      "loss": 0.0009,
      "step": 298530
    },
    {
      "epoch": 1.9147428270901412,
      "grad_norm": 0.07613728940486908,
      "learning_rate": 5.524527424034865e-08,
      "loss": 0.0009,
      "step": 298540
    },
    {
      "epoch": 1.9148069639839274,
      "grad_norm": 0.019116273149847984,
      "learning_rate": 5.516233303635044e-08,
      "loss": 0.0008,
      "step": 298550
    },
    {
      "epoch": 1.9148711008777135,
      "grad_norm": 0.12364184856414795,
      "learning_rate": 5.5079453794701924e-08,
      "loss": 0.0012,
      "step": 298560
    },
    {
      "epoch": 1.9149352377714997,
      "grad_norm": 0.05676203966140747,
      "learning_rate": 5.4996636516441715e-08,
      "loss": 0.0006,
      "step": 298570
    },
    {
      "epoch": 1.9149993746652856,
      "grad_norm": 0.09833601117134094,
      "learning_rate": 5.491388120260843e-08,
      "loss": 0.0007,
      "step": 298580
    },
    {
      "epoch": 1.9150635115590717,
      "grad_norm": 0.0450495220720768,
      "learning_rate": 5.483118785423791e-08,
      "loss": 0.001,
      "step": 298590
    },
    {
      "epoch": 1.9151276484528577,
      "grad_norm": 0.050660811364650726,
      "learning_rate": 5.47485564723671e-08,
      "loss": 0.0007,
      "step": 298600
    },
    {
      "epoch": 1.9151917853466438,
      "grad_norm": 0.245155930519104,
      "learning_rate": 5.4665987058030724e-08,
      "loss": 0.0009,
      "step": 298610
    },
    {
      "epoch": 1.91525592224043,
      "grad_norm": 0.07262758165597916,
      "learning_rate": 5.458347961226462e-08,
      "loss": 0.0008,
      "step": 298620
    },
    {
      "epoch": 1.915320059134216,
      "grad_norm": 0.005807352717965841,
      "learning_rate": 5.450103413610186e-08,
      "loss": 0.0008,
      "step": 298630
    },
    {
      "epoch": 1.9153841960280023,
      "grad_norm": 0.0009603967191651464,
      "learning_rate": 5.441865063057494e-08,
      "loss": 0.0007,
      "step": 298640
    },
    {
      "epoch": 1.9154483329217884,
      "grad_norm": 0.06040569022297859,
      "learning_rate": 5.433632909671749e-08,
      "loss": 0.0012,
      "step": 298650
    },
    {
      "epoch": 1.9155124698155743,
      "grad_norm": 0.039267648011446,
      "learning_rate": 5.425406953556034e-08,
      "loss": 0.0012,
      "step": 298660
    },
    {
      "epoch": 1.9155766067093605,
      "grad_norm": 0.09607995301485062,
      "learning_rate": 5.4171871948134335e-08,
      "loss": 0.002,
      "step": 298670
    },
    {
      "epoch": 1.9156407436031464,
      "grad_norm": 0.08390876650810242,
      "learning_rate": 5.408973633546921e-08,
      "loss": 0.0011,
      "step": 298680
    },
    {
      "epoch": 1.9157048804969325,
      "grad_norm": 0.056424278765916824,
      "learning_rate": 5.40076626985947e-08,
      "loss": 0.0008,
      "step": 298690
    },
    {
      "epoch": 1.9157690173907187,
      "grad_norm": 0.048826076090335846,
      "learning_rate": 5.392565103853942e-08,
      "loss": 0.0007,
      "step": 298700
    },
    {
      "epoch": 1.9158331542845048,
      "grad_norm": 0.03236357867717743,
      "learning_rate": 5.3843701356330326e-08,
      "loss": 0.0046,
      "step": 298710
    },
    {
      "epoch": 1.915897291178291,
      "grad_norm": 0.0056189983151853085,
      "learning_rate": 5.3761813652994374e-08,
      "loss": 0.0007,
      "step": 298720
    },
    {
      "epoch": 1.9159614280720771,
      "grad_norm": 0.0127511415630579,
      "learning_rate": 5.367998792955798e-08,
      "loss": 0.0006,
      "step": 298730
    },
    {
      "epoch": 1.9160255649658633,
      "grad_norm": 0.12237073481082916,
      "learning_rate": 5.359822418704641e-08,
      "loss": 0.0006,
      "step": 298740
    },
    {
      "epoch": 1.9160897018596492,
      "grad_norm": 0.06909266859292984,
      "learning_rate": 5.351652242648442e-08,
      "loss": 0.0011,
      "step": 298750
    },
    {
      "epoch": 1.9161538387534354,
      "grad_norm": 0.07474600523710251,
      "learning_rate": 5.3434882648895626e-08,
      "loss": 0.0012,
      "step": 298760
    },
    {
      "epoch": 1.9162179756472213,
      "grad_norm": 0.005450695753097534,
      "learning_rate": 5.33533048553031e-08,
      "loss": 0.0027,
      "step": 298770
    },
    {
      "epoch": 1.9162821125410074,
      "grad_norm": 0.03122709132730961,
      "learning_rate": 5.327178904672881e-08,
      "loss": 0.001,
      "step": 298780
    },
    {
      "epoch": 1.9163462494347936,
      "grad_norm": 0.0031229574233293533,
      "learning_rate": 5.319033522419414e-08,
      "loss": 0.001,
      "step": 298790
    },
    {
      "epoch": 1.9164103863285797,
      "grad_norm": 0.035058699548244476,
      "learning_rate": 5.3108943388720527e-08,
      "loss": 0.0004,
      "step": 298800
    },
    {
      "epoch": 1.9164745232223659,
      "grad_norm": 0.012346319854259491,
      "learning_rate": 5.3027613541327126e-08,
      "loss": 0.0009,
      "step": 298810
    },
    {
      "epoch": 1.916538660116152,
      "grad_norm": 0.155277818441391,
      "learning_rate": 5.294634568303314e-08,
      "loss": 0.0016,
      "step": 298820
    },
    {
      "epoch": 1.916602797009938,
      "grad_norm": 0.14972133934497833,
      "learning_rate": 5.286513981485719e-08,
      "loss": 0.001,
      "step": 298830
    },
    {
      "epoch": 1.9166669339037241,
      "grad_norm": 0.011021796613931656,
      "learning_rate": 5.2783995937816244e-08,
      "loss": 0.0007,
      "step": 298840
    },
    {
      "epoch": 1.9167310707975103,
      "grad_norm": 0.007839307188987732,
      "learning_rate": 5.270291405292838e-08,
      "loss": 0.0012,
      "step": 298850
    },
    {
      "epoch": 1.9167952076912962,
      "grad_norm": 0.027409091591835022,
      "learning_rate": 5.262189416120833e-08,
      "loss": 0.0011,
      "step": 298860
    },
    {
      "epoch": 1.9168593445850823,
      "grad_norm": 0.048824433237314224,
      "learning_rate": 5.254093626367196e-08,
      "loss": 0.001,
      "step": 298870
    },
    {
      "epoch": 1.9169234814788685,
      "grad_norm": 0.10896562039852142,
      "learning_rate": 5.24600403613329e-08,
      "loss": 0.0011,
      "step": 298880
    },
    {
      "epoch": 1.9169876183726546,
      "grad_norm": 0.14132238924503326,
      "learning_rate": 5.2379206455206446e-08,
      "loss": 0.0007,
      "step": 298890
    },
    {
      "epoch": 1.9170517552664408,
      "grad_norm": 0.08344261348247528,
      "learning_rate": 5.229843454630401e-08,
      "loss": 0.0007,
      "step": 298900
    },
    {
      "epoch": 1.917115892160227,
      "grad_norm": 0.08146130293607712,
      "learning_rate": 5.221772463563868e-08,
      "loss": 0.0034,
      "step": 298910
    },
    {
      "epoch": 1.9171800290540129,
      "grad_norm": 0.14258337020874023,
      "learning_rate": 5.213707672422075e-08,
      "loss": 0.0009,
      "step": 298920
    },
    {
      "epoch": 1.917244165947799,
      "grad_norm": 0.04950059950351715,
      "learning_rate": 5.2056490813061636e-08,
      "loss": 0.0003,
      "step": 298930
    },
    {
      "epoch": 1.917308302841585,
      "grad_norm": 0.0863884910941124,
      "learning_rate": 5.197596690317108e-08,
      "loss": 0.0015,
      "step": 298940
    },
    {
      "epoch": 1.917372439735371,
      "grad_norm": 0.12216462194919586,
      "learning_rate": 5.189550499555773e-08,
      "loss": 0.0009,
      "step": 298950
    },
    {
      "epoch": 1.9174365766291572,
      "grad_norm": 0.014564010314643383,
      "learning_rate": 5.181510509123022e-08,
      "loss": 0.0007,
      "step": 298960
    },
    {
      "epoch": 1.9175007135229434,
      "grad_norm": 0.02649235725402832,
      "learning_rate": 5.173476719119608e-08,
      "loss": 0.0009,
      "step": 298970
    },
    {
      "epoch": 1.9175648504167295,
      "grad_norm": 0.14758247137069702,
      "learning_rate": 5.165449129646172e-08,
      "loss": 0.0008,
      "step": 298980
    },
    {
      "epoch": 1.9176289873105157,
      "grad_norm": 0.12455189228057861,
      "learning_rate": 5.157427740803245e-08,
      "loss": 0.0009,
      "step": 298990
    },
    {
      "epoch": 1.9176931242043018,
      "grad_norm": 0.08838582783937454,
      "learning_rate": 5.1494125526914687e-08,
      "loss": 0.0012,
      "step": 299000
    },
    {
      "epoch": 1.9177572610980878,
      "grad_norm": 0.08339110761880875,
      "learning_rate": 5.141403565411207e-08,
      "loss": 0.0009,
      "step": 299010
    },
    {
      "epoch": 1.917821397991874,
      "grad_norm": 0.058458272367715836,
      "learning_rate": 5.1334007790628805e-08,
      "loss": 0.0013,
      "step": 299020
    },
    {
      "epoch": 1.9178855348856598,
      "grad_norm": 0.09558631479740143,
      "learning_rate": 5.12540419374663e-08,
      "loss": 0.001,
      "step": 299030
    },
    {
      "epoch": 1.917949671779446,
      "grad_norm": 0.08977165818214417,
      "learning_rate": 5.1174138095627654e-08,
      "loss": 0.0009,
      "step": 299040
    },
    {
      "epoch": 1.9180138086732321,
      "grad_norm": 0.08206156641244888,
      "learning_rate": 5.109429626611428e-08,
      "loss": 0.0009,
      "step": 299050
    },
    {
      "epoch": 1.9180779455670183,
      "grad_norm": 0.04454299435019493,
      "learning_rate": 5.10145164499265e-08,
      "loss": 0.0014,
      "step": 299060
    },
    {
      "epoch": 1.9181420824608044,
      "grad_norm": 0.09718791395425797,
      "learning_rate": 5.0934798648062944e-08,
      "loss": 0.0012,
      "step": 299070
    },
    {
      "epoch": 1.9182062193545906,
      "grad_norm": 0.13956280052661896,
      "learning_rate": 5.0855142861523934e-08,
      "loss": 0.0013,
      "step": 299080
    },
    {
      "epoch": 1.9182703562483765,
      "grad_norm": 0.03992290794849396,
      "learning_rate": 5.0775549091307554e-08,
      "loss": 0.0006,
      "step": 299090
    },
    {
      "epoch": 1.9183344931421626,
      "grad_norm": 0.13538269698619843,
      "learning_rate": 5.0696017338409675e-08,
      "loss": 0.0011,
      "step": 299100
    },
    {
      "epoch": 1.9183986300359486,
      "grad_norm": 0.03277270868420601,
      "learning_rate": 5.0616547603828394e-08,
      "loss": 0.0017,
      "step": 299110
    },
    {
      "epoch": 1.9184627669297347,
      "grad_norm": 0.049767736345529556,
      "learning_rate": 5.053713988855957e-08,
      "loss": 0.001,
      "step": 299120
    },
    {
      "epoch": 1.9185269038235209,
      "grad_norm": 0.12353898584842682,
      "learning_rate": 5.045779419359742e-08,
      "loss": 0.0013,
      "step": 299130
    },
    {
      "epoch": 1.918591040717307,
      "grad_norm": 0.015955159440636635,
      "learning_rate": 5.0378510519935584e-08,
      "loss": 0.0018,
      "step": 299140
    },
    {
      "epoch": 1.9186551776110932,
      "grad_norm": 0.0956311970949173,
      "learning_rate": 5.029928886856939e-08,
      "loss": 0.0008,
      "step": 299150
    },
    {
      "epoch": 1.9187193145048793,
      "grad_norm": 0.04158175364136696,
      "learning_rate": 5.02201292404908e-08,
      "loss": 0.0005,
      "step": 299160
    },
    {
      "epoch": 1.9187834513986655,
      "grad_norm": 0.05196625366806984,
      "learning_rate": 5.014103163669126e-08,
      "loss": 0.0006,
      "step": 299170
    },
    {
      "epoch": 1.9188475882924514,
      "grad_norm": 0.048744142055511475,
      "learning_rate": 5.0061996058161644e-08,
      "loss": 0.0008,
      "step": 299180
    },
    {
      "epoch": 1.9189117251862375,
      "grad_norm": 0.04593722149729729,
      "learning_rate": 4.998302250589338e-08,
      "loss": 0.0006,
      "step": 299190
    },
    {
      "epoch": 1.9189758620800235,
      "grad_norm": 0.012638093903660774,
      "learning_rate": 4.9904110980875664e-08,
      "loss": 0.0007,
      "step": 299200
    },
    {
      "epoch": 1.9190399989738096,
      "grad_norm": 0.0911950096487999,
      "learning_rate": 4.982526148409716e-08,
      "loss": 0.0006,
      "step": 299210
    },
    {
      "epoch": 1.9191041358675958,
      "grad_norm": 0.005118612200021744,
      "learning_rate": 4.974647401654542e-08,
      "loss": 0.0017,
      "step": 299220
    },
    {
      "epoch": 1.919168272761382,
      "grad_norm": 0.23245181143283844,
      "learning_rate": 4.966774857920909e-08,
      "loss": 0.0017,
      "step": 299230
    },
    {
      "epoch": 1.919232409655168,
      "grad_norm": 0.05644499883055687,
      "learning_rate": 4.958908517307348e-08,
      "loss": 0.0009,
      "step": 299240
    },
    {
      "epoch": 1.9192965465489542,
      "grad_norm": 0.05527875944972038,
      "learning_rate": 4.951048379912449e-08,
      "loss": 0.0018,
      "step": 299250
    },
    {
      "epoch": 1.9193606834427401,
      "grad_norm": 0.055565234273672104,
      "learning_rate": 4.9431944458347426e-08,
      "loss": 0.0006,
      "step": 299260
    },
    {
      "epoch": 1.9194248203365263,
      "grad_norm": 0.18801401555538177,
      "learning_rate": 4.935346715172595e-08,
      "loss": 0.0015,
      "step": 299270
    },
    {
      "epoch": 1.9194889572303124,
      "grad_norm": 0.0394013486802578,
      "learning_rate": 4.927505188024373e-08,
      "loss": 0.0035,
      "step": 299280
    },
    {
      "epoch": 1.9195530941240984,
      "grad_norm": 0.07867538183927536,
      "learning_rate": 4.9196698644883303e-08,
      "loss": 0.0011,
      "step": 299290
    },
    {
      "epoch": 1.9196172310178845,
      "grad_norm": 0.027638576924800873,
      "learning_rate": 4.911840744662666e-08,
      "loss": 0.001,
      "step": 299300
    },
    {
      "epoch": 1.9196813679116707,
      "grad_norm": 0.008723137900233269,
      "learning_rate": 4.9040178286455245e-08,
      "loss": 0.0013,
      "step": 299310
    },
    {
      "epoch": 1.9197455048054568,
      "grad_norm": 0.046434395015239716,
      "learning_rate": 4.896201116534827e-08,
      "loss": 0.0009,
      "step": 299320
    },
    {
      "epoch": 1.919809641699243,
      "grad_norm": 0.27379703521728516,
      "learning_rate": 4.88839060842855e-08,
      "loss": 0.0008,
      "step": 299330
    },
    {
      "epoch": 1.919873778593029,
      "grad_norm": 0.03799886628985405,
      "learning_rate": 4.880586304424617e-08,
      "loss": 0.0012,
      "step": 299340
    },
    {
      "epoch": 1.919937915486815,
      "grad_norm": 0.036054909229278564,
      "learning_rate": 4.872788204620782e-08,
      "loss": 0.0017,
      "step": 299350
    },
    {
      "epoch": 1.9200020523806012,
      "grad_norm": 0.0491146519780159,
      "learning_rate": 4.8649963091148e-08,
      "loss": 0.001,
      "step": 299360
    },
    {
      "epoch": 1.920066189274387,
      "grad_norm": 0.042426567524671555,
      "learning_rate": 4.85721061800426e-08,
      "loss": 0.0007,
      "step": 299370
    },
    {
      "epoch": 1.9201303261681733,
      "grad_norm": 0.0619833804666996,
      "learning_rate": 4.84943113138675e-08,
      "loss": 0.0011,
      "step": 299380
    },
    {
      "epoch": 1.9201944630619594,
      "grad_norm": 0.049811821430921555,
      "learning_rate": 4.8416578493597485e-08,
      "loss": 0.0019,
      "step": 299390
    },
    {
      "epoch": 1.9202585999557455,
      "grad_norm": 0.1061360165476799,
      "learning_rate": 4.833890772020622e-08,
      "loss": 0.0029,
      "step": 299400
    },
    {
      "epoch": 1.9203227368495317,
      "grad_norm": 0.07087648659944534,
      "learning_rate": 4.826129899466792e-08,
      "loss": 0.0005,
      "step": 299410
    },
    {
      "epoch": 1.9203868737433178,
      "grad_norm": 0.04247698560357094,
      "learning_rate": 4.818375231795458e-08,
      "loss": 0.0006,
      "step": 299420
    },
    {
      "epoch": 1.920451010637104,
      "grad_norm": 0.13066044449806213,
      "learning_rate": 4.81062676910371e-08,
      "loss": 0.0013,
      "step": 299430
    },
    {
      "epoch": 1.92051514753089,
      "grad_norm": 0.08296853303909302,
      "learning_rate": 4.802884511488748e-08,
      "loss": 0.0006,
      "step": 299440
    },
    {
      "epoch": 1.920579284424676,
      "grad_norm": 0.03349597379565239,
      "learning_rate": 4.795148459047605e-08,
      "loss": 0.0006,
      "step": 299450
    },
    {
      "epoch": 1.920643421318462,
      "grad_norm": 0.04592498391866684,
      "learning_rate": 4.787418611877093e-08,
      "loss": 0.0004,
      "step": 299460
    },
    {
      "epoch": 1.9207075582122481,
      "grad_norm": 0.045830707997083664,
      "learning_rate": 4.779694970074189e-08,
      "loss": 0.001,
      "step": 299470
    },
    {
      "epoch": 1.9207716951060343,
      "grad_norm": 0.0611383356153965,
      "learning_rate": 4.77197753373565e-08,
      "loss": 0.0015,
      "step": 299480
    },
    {
      "epoch": 1.9208358319998204,
      "grad_norm": 0.03120235539972782,
      "learning_rate": 4.76426630295812e-08,
      "loss": 0.0011,
      "step": 299490
    },
    {
      "epoch": 1.9208999688936066,
      "grad_norm": 0.0034458893351256847,
      "learning_rate": 4.756561277838301e-08,
      "loss": 0.0014,
      "step": 299500
    },
    {
      "epoch": 1.9209641057873927,
      "grad_norm": 0.0789545401930809,
      "learning_rate": 4.748862458472725e-08,
      "loss": 0.0008,
      "step": 299510
    },
    {
      "epoch": 1.9210282426811787,
      "grad_norm": 0.03234486281871796,
      "learning_rate": 4.741169844957816e-08,
      "loss": 0.0013,
      "step": 299520
    },
    {
      "epoch": 1.9210923795749648,
      "grad_norm": 0.059302132576704025,
      "learning_rate": 4.733483437390052e-08,
      "loss": 0.0012,
      "step": 299530
    },
    {
      "epoch": 1.9211565164687507,
      "grad_norm": 0.13612000644207,
      "learning_rate": 4.7258032358656893e-08,
      "loss": 0.0009,
      "step": 299540
    },
    {
      "epoch": 1.921220653362537,
      "grad_norm": 0.07007851451635361,
      "learning_rate": 4.7181292404809286e-08,
      "loss": 0.0011,
      "step": 299550
    },
    {
      "epoch": 1.921284790256323,
      "grad_norm": 0.117681585252285,
      "learning_rate": 4.710461451332027e-08,
      "loss": 0.0009,
      "step": 299560
    },
    {
      "epoch": 1.9213489271501092,
      "grad_norm": 0.05283145606517792,
      "learning_rate": 4.7027998685150176e-08,
      "loss": 0.0012,
      "step": 299570
    },
    {
      "epoch": 1.9214130640438953,
      "grad_norm": 0.11339683830738068,
      "learning_rate": 4.695144492125936e-08,
      "loss": 0.001,
      "step": 299580
    },
    {
      "epoch": 1.9214772009376815,
      "grad_norm": 0.07208563387393951,
      "learning_rate": 4.687495322260649e-08,
      "loss": 0.0012,
      "step": 299590
    },
    {
      "epoch": 1.9215413378314676,
      "grad_norm": 0.16996242105960846,
      "learning_rate": 4.6798523590150246e-08,
      "loss": 0.0017,
      "step": 299600
    },
    {
      "epoch": 1.9216054747252536,
      "grad_norm": 0.03294508531689644,
      "learning_rate": 4.6722156024848754e-08,
      "loss": 0.0007,
      "step": 299610
    },
    {
      "epoch": 1.9216696116190397,
      "grad_norm": 0.007039350923150778,
      "learning_rate": 4.664585052765902e-08,
      "loss": 0.0018,
      "step": 299620
    },
    {
      "epoch": 1.9217337485128256,
      "grad_norm": 0.10350693017244339,
      "learning_rate": 4.6569607099536393e-08,
      "loss": 0.0009,
      "step": 299630
    },
    {
      "epoch": 1.9217978854066118,
      "grad_norm": 0.056322213262319565,
      "learning_rate": 4.649342574143678e-08,
      "loss": 0.0007,
      "step": 299640
    },
    {
      "epoch": 1.921862022300398,
      "grad_norm": 0.009310414083302021,
      "learning_rate": 4.641730645431497e-08,
      "loss": 0.0018,
      "step": 299650
    },
    {
      "epoch": 1.921926159194184,
      "grad_norm": 0.07358931750059128,
      "learning_rate": 4.634124923912464e-08,
      "loss": 0.0008,
      "step": 299660
    },
    {
      "epoch": 1.9219902960879702,
      "grad_norm": 0.0843188539147377,
      "learning_rate": 4.6265254096818367e-08,
      "loss": 0.0035,
      "step": 299670
    },
    {
      "epoch": 1.9220544329817564,
      "grad_norm": 0.023625105619430542,
      "learning_rate": 4.618932102834928e-08,
      "loss": 0.0014,
      "step": 299680
    },
    {
      "epoch": 1.9221185698755423,
      "grad_norm": 0.11312787979841232,
      "learning_rate": 4.611345003466827e-08,
      "loss": 0.0009,
      "step": 299690
    },
    {
      "epoch": 1.9221827067693285,
      "grad_norm": 0.0016061868518590927,
      "learning_rate": 4.603764111672626e-08,
      "loss": 0.0004,
      "step": 299700
    },
    {
      "epoch": 1.9222468436631146,
      "grad_norm": 0.08886443078517914,
      "learning_rate": 4.596189427547304e-08,
      "loss": 0.0011,
      "step": 299710
    },
    {
      "epoch": 1.9223109805569005,
      "grad_norm": 0.0731070414185524,
      "learning_rate": 4.588620951185785e-08,
      "loss": 0.0008,
      "step": 299720
    },
    {
      "epoch": 1.9223751174506867,
      "grad_norm": 0.2453460395336151,
      "learning_rate": 4.581058682682937e-08,
      "loss": 0.0023,
      "step": 299730
    },
    {
      "epoch": 1.9224392543444728,
      "grad_norm": 0.006312183570116758,
      "learning_rate": 4.573502622133463e-08,
      "loss": 0.0017,
      "step": 299740
    },
    {
      "epoch": 1.922503391238259,
      "grad_norm": 0.1199653148651123,
      "learning_rate": 4.565952769632065e-08,
      "loss": 0.0008,
      "step": 299750
    },
    {
      "epoch": 1.9225675281320451,
      "grad_norm": 0.10376963764429092,
      "learning_rate": 4.558409125273444e-08,
      "loss": 0.0023,
      "step": 299760
    },
    {
      "epoch": 1.9226316650258313,
      "grad_norm": 0.2373136430978775,
      "learning_rate": 4.55087168915197e-08,
      "loss": 0.0043,
      "step": 299770
    },
    {
      "epoch": 1.9226958019196172,
      "grad_norm": 0.10240045189857483,
      "learning_rate": 4.543340461362178e-08,
      "loss": 0.0007,
      "step": 299780
    },
    {
      "epoch": 1.9227599388134033,
      "grad_norm": 0.07114317268133163,
      "learning_rate": 4.535815441998437e-08,
      "loss": 0.0009,
      "step": 299790
    },
    {
      "epoch": 1.9228240757071893,
      "grad_norm": 0.13522303104400635,
      "learning_rate": 4.5282966311550045e-08,
      "loss": 0.0007,
      "step": 299800
    },
    {
      "epoch": 1.9228882126009754,
      "grad_norm": 0.05266407132148743,
      "learning_rate": 4.520784028926195e-08,
      "loss": 0.0011,
      "step": 299810
    },
    {
      "epoch": 1.9229523494947616,
      "grad_norm": 0.037769682705402374,
      "learning_rate": 4.5132776354059884e-08,
      "loss": 0.0006,
      "step": 299820
    },
    {
      "epoch": 1.9230164863885477,
      "grad_norm": 0.04296983778476715,
      "learning_rate": 4.505777450688586e-08,
      "loss": 0.0006,
      "step": 299830
    },
    {
      "epoch": 1.9230806232823339,
      "grad_norm": 0.12407558411359787,
      "learning_rate": 4.498283474867915e-08,
      "loss": 0.0011,
      "step": 299840
    },
    {
      "epoch": 1.92314476017612,
      "grad_norm": 0.08229169994592667,
      "learning_rate": 4.490795708037843e-08,
      "loss": 0.0006,
      "step": 299850
    },
    {
      "epoch": 1.9232088970699062,
      "grad_norm": 0.010448667220771313,
      "learning_rate": 4.4833141502922395e-08,
      "loss": 0.0007,
      "step": 299860
    },
    {
      "epoch": 1.923273033963692,
      "grad_norm": 0.0727403461933136,
      "learning_rate": 4.4758388017248636e-08,
      "loss": 0.0008,
      "step": 299870
    },
    {
      "epoch": 1.9233371708574782,
      "grad_norm": 0.09879482537508011,
      "learning_rate": 4.468369662429417e-08,
      "loss": 0.0011,
      "step": 299880
    },
    {
      "epoch": 1.9234013077512642,
      "grad_norm": 0.15423615276813507,
      "learning_rate": 4.460906732499437e-08,
      "loss": 0.0029,
      "step": 299890
    },
    {
      "epoch": 1.9234654446450503,
      "grad_norm": 0.2108854353427887,
      "learning_rate": 4.453450012028404e-08,
      "loss": 0.0024,
      "step": 299900
    },
    {
      "epoch": 1.9235295815388365,
      "grad_norm": 0.08908912539482117,
      "learning_rate": 4.445999501109799e-08,
      "loss": 0.0013,
      "step": 299910
    },
    {
      "epoch": 1.9235937184326226,
      "grad_norm": 0.13990257680416107,
      "learning_rate": 4.438555199837047e-08,
      "loss": 0.0009,
      "step": 299920
    },
    {
      "epoch": 1.9236578553264088,
      "grad_norm": 0.05871908366680145,
      "learning_rate": 4.4311171083032956e-08,
      "loss": 0.0004,
      "step": 299930
    },
    {
      "epoch": 1.923721992220195,
      "grad_norm": 0.05720856785774231,
      "learning_rate": 4.423685226601915e-08,
      "loss": 0.0008,
      "step": 299940
    },
    {
      "epoch": 1.9237861291139808,
      "grad_norm": 0.025936510413885117,
      "learning_rate": 4.416259554825886e-08,
      "loss": 0.001,
      "step": 299950
    },
    {
      "epoch": 1.923850266007767,
      "grad_norm": 0.06441635638475418,
      "learning_rate": 4.408840093068301e-08,
      "loss": 0.0007,
      "step": 299960
    },
    {
      "epoch": 1.923914402901553,
      "grad_norm": 0.17235547304153442,
      "learning_rate": 4.4014268414221404e-08,
      "loss": 0.0021,
      "step": 299970
    },
    {
      "epoch": 1.923978539795339,
      "grad_norm": 0.07071554660797119,
      "learning_rate": 4.3940197999803315e-08,
      "loss": 0.0009,
      "step": 299980
    },
    {
      "epoch": 1.9240426766891252,
      "grad_norm": 0.01034556794911623,
      "learning_rate": 4.386618968835688e-08,
      "loss": 0.0009,
      "step": 299990
    },
    {
      "epoch": 1.9241068135829114,
      "grad_norm": 0.08722493052482605,
      "learning_rate": 4.379224348080913e-08,
      "loss": 0.0008,
      "step": 300000
    },
    {
      "epoch": 1.9241709504766975,
      "grad_norm": 0.07715963572263718,
      "learning_rate": 4.3718359378085994e-08,
      "loss": 0.0007,
      "step": 300010
    },
    {
      "epoch": 1.9242350873704837,
      "grad_norm": 0.08677951246500015,
      "learning_rate": 4.364453738111451e-08,
      "loss": 0.0006,
      "step": 300020
    },
    {
      "epoch": 1.9242992242642698,
      "grad_norm": 0.05860080569982529,
      "learning_rate": 4.357077749081895e-08,
      "loss": 0.0009,
      "step": 300030
    },
    {
      "epoch": 1.9243633611580557,
      "grad_norm": 0.05761454999446869,
      "learning_rate": 4.3497079708124114e-08,
      "loss": 0.0011,
      "step": 300040
    },
    {
      "epoch": 1.9244274980518419,
      "grad_norm": 0.20905107259750366,
      "learning_rate": 4.342344403395316e-08,
      "loss": 0.0008,
      "step": 300050
    },
    {
      "epoch": 1.9244916349456278,
      "grad_norm": 0.01208533812314272,
      "learning_rate": 4.3349870469229226e-08,
      "loss": 0.0012,
      "step": 300060
    },
    {
      "epoch": 1.924555771839414,
      "grad_norm": 0.05330813676118851,
      "learning_rate": 4.327635901487326e-08,
      "loss": 0.0006,
      "step": 300070
    },
    {
      "epoch": 1.9246199087332,
      "grad_norm": 0.0022559245117008686,
      "learning_rate": 4.320290967180729e-08,
      "loss": 0.0013,
      "step": 300080
    },
    {
      "epoch": 1.9246840456269863,
      "grad_norm": 0.06547002494335175,
      "learning_rate": 4.31295224409517e-08,
      "loss": 0.0022,
      "step": 300090
    },
    {
      "epoch": 1.9247481825207724,
      "grad_norm": 0.10205364972352982,
      "learning_rate": 4.305619732322519e-08,
      "loss": 0.0013,
      "step": 300100
    },
    {
      "epoch": 1.9248123194145585,
      "grad_norm": 0.14018011093139648,
      "learning_rate": 4.298293431954759e-08,
      "loss": 0.0009,
      "step": 300110
    },
    {
      "epoch": 1.9248764563083447,
      "grad_norm": 0.0348082073032856,
      "learning_rate": 4.290973343083649e-08,
      "loss": 0.0005,
      "step": 300120
    },
    {
      "epoch": 1.9249405932021306,
      "grad_norm": 0.029344813898205757,
      "learning_rate": 4.283659465800949e-08,
      "loss": 0.0007,
      "step": 300130
    },
    {
      "epoch": 1.9250047300959168,
      "grad_norm": 0.1087023913860321,
      "learning_rate": 4.276351800198253e-08,
      "loss": 0.0007,
      "step": 300140
    },
    {
      "epoch": 1.9250688669897027,
      "grad_norm": 0.01807575672864914,
      "learning_rate": 4.269050346367154e-08,
      "loss": 0.001,
      "step": 300150
    },
    {
      "epoch": 1.9251330038834888,
      "grad_norm": 0.0954509899020195,
      "learning_rate": 4.261755104399135e-08,
      "loss": 0.0011,
      "step": 300160
    },
    {
      "epoch": 1.925197140777275,
      "grad_norm": 0.019217127934098244,
      "learning_rate": 4.254466074385677e-08,
      "loss": 0.0019,
      "step": 300170
    },
    {
      "epoch": 1.9252612776710611,
      "grad_norm": 0.03665965422987938,
      "learning_rate": 4.2471832564180414e-08,
      "loss": 0.0008,
      "step": 300180
    },
    {
      "epoch": 1.9253254145648473,
      "grad_norm": 0.04427497088909149,
      "learning_rate": 4.2399066505874886e-08,
      "loss": 0.0017,
      "step": 300190
    },
    {
      "epoch": 1.9253895514586334,
      "grad_norm": 0.003385307500138879,
      "learning_rate": 4.232636256985167e-08,
      "loss": 0.0005,
      "step": 300200
    },
    {
      "epoch": 1.9254536883524194,
      "grad_norm": 0.08548618108034134,
      "learning_rate": 4.225372075702339e-08,
      "loss": 0.0012,
      "step": 300210
    },
    {
      "epoch": 1.9255178252462055,
      "grad_norm": 0.07868991792201996,
      "learning_rate": 4.2181141068298736e-08,
      "loss": 0.0013,
      "step": 300220
    },
    {
      "epoch": 1.9255819621399914,
      "grad_norm": 0.07562734186649323,
      "learning_rate": 4.210862350458755e-08,
      "loss": 0.0019,
      "step": 300230
    },
    {
      "epoch": 1.9256460990337776,
      "grad_norm": 0.08704736828804016,
      "learning_rate": 4.2036168066798554e-08,
      "loss": 0.0018,
      "step": 300240
    },
    {
      "epoch": 1.9257102359275637,
      "grad_norm": 0.09126976877450943,
      "learning_rate": 4.196377475584046e-08,
      "loss": 0.0009,
      "step": 300250
    },
    {
      "epoch": 1.92577437282135,
      "grad_norm": 0.10753437131643295,
      "learning_rate": 4.189144357261921e-08,
      "loss": 0.0016,
      "step": 300260
    },
    {
      "epoch": 1.925838509715136,
      "grad_norm": 0.11447235196828842,
      "learning_rate": 4.181917451804185e-08,
      "loss": 0.0008,
      "step": 300270
    },
    {
      "epoch": 1.9259026466089222,
      "grad_norm": 0.08692572265863419,
      "learning_rate": 4.174696759301322e-08,
      "loss": 0.0006,
      "step": 300280
    },
    {
      "epoch": 1.9259667835027083,
      "grad_norm": 0.02848992496728897,
      "learning_rate": 4.167482279843926e-08,
      "loss": 0.0008,
      "step": 300290
    },
    {
      "epoch": 1.9260309203964943,
      "grad_norm": 0.012743757106363773,
      "learning_rate": 4.1602740135223697e-08,
      "loss": 0.0011,
      "step": 300300
    },
    {
      "epoch": 1.9260950572902804,
      "grad_norm": 0.04856209456920624,
      "learning_rate": 4.153071960426913e-08,
      "loss": 0.0008,
      "step": 300310
    },
    {
      "epoch": 1.9261591941840663,
      "grad_norm": 0.05277502164244652,
      "learning_rate": 4.1458761206478185e-08,
      "loss": 0.0005,
      "step": 300320
    },
    {
      "epoch": 1.9262233310778525,
      "grad_norm": 0.10218650102615356,
      "learning_rate": 4.1386864942753456e-08,
      "loss": 0.0009,
      "step": 300330
    },
    {
      "epoch": 1.9262874679716386,
      "grad_norm": 0.03546750545501709,
      "learning_rate": 4.13150308139948e-08,
      "loss": 0.0006,
      "step": 300340
    },
    {
      "epoch": 1.9263516048654248,
      "grad_norm": 0.07997418940067291,
      "learning_rate": 4.124325882110203e-08,
      "loss": 0.0005,
      "step": 300350
    },
    {
      "epoch": 1.926415741759211,
      "grad_norm": 0.11265015602111816,
      "learning_rate": 4.1171548964976106e-08,
      "loss": 0.0017,
      "step": 300360
    },
    {
      "epoch": 1.926479878652997,
      "grad_norm": 0.013606592081487179,
      "learning_rate": 4.109990124651464e-08,
      "loss": 0.001,
      "step": 300370
    },
    {
      "epoch": 1.926544015546783,
      "grad_norm": 0.02300187200307846,
      "learning_rate": 4.102831566661525e-08,
      "loss": 0.0012,
      "step": 300380
    },
    {
      "epoch": 1.9266081524405692,
      "grad_norm": 0.02995084412395954,
      "learning_rate": 4.095679222617499e-08,
      "loss": 0.0009,
      "step": 300390
    },
    {
      "epoch": 1.926672289334355,
      "grad_norm": 0.060416627675294876,
      "learning_rate": 4.0885330926090924e-08,
      "loss": 0.0004,
      "step": 300400
    },
    {
      "epoch": 1.9267364262281412,
      "grad_norm": 0.04969345033168793,
      "learning_rate": 4.0813931767257345e-08,
      "loss": 0.0008,
      "step": 300410
    },
    {
      "epoch": 1.9268005631219274,
      "grad_norm": 0.07248201966285706,
      "learning_rate": 4.074259475056963e-08,
      "loss": 0.0013,
      "step": 300420
    },
    {
      "epoch": 1.9268647000157135,
      "grad_norm": 0.05075995996594429,
      "learning_rate": 4.0671319876921524e-08,
      "loss": 0.0009,
      "step": 300430
    },
    {
      "epoch": 1.9269288369094997,
      "grad_norm": 0.02230651304125786,
      "learning_rate": 4.060010714720619e-08,
      "loss": 0.0006,
      "step": 300440
    },
    {
      "epoch": 1.9269929738032858,
      "grad_norm": 0.11635108292102814,
      "learning_rate": 4.052895656231626e-08,
      "loss": 0.0009,
      "step": 300450
    },
    {
      "epoch": 1.927057110697072,
      "grad_norm": 0.04682289436459541,
      "learning_rate": 4.045786812314268e-08,
      "loss": 0.0016,
      "step": 300460
    },
    {
      "epoch": 1.927121247590858,
      "grad_norm": 0.221276193857193,
      "learning_rate": 4.0386841830576396e-08,
      "loss": 0.0007,
      "step": 300470
    },
    {
      "epoch": 1.927185384484644,
      "grad_norm": 0.3462485074996948,
      "learning_rate": 4.031587768550782e-08,
      "loss": 0.0021,
      "step": 300480
    },
    {
      "epoch": 1.92724952137843,
      "grad_norm": 0.02788991667330265,
      "learning_rate": 4.0244975688825685e-08,
      "loss": 0.0004,
      "step": 300490
    },
    {
      "epoch": 1.9273136582722161,
      "grad_norm": 0.05282348394393921,
      "learning_rate": 4.017413584141871e-08,
      "loss": 0.0019,
      "step": 300500
    },
    {
      "epoch": 1.9273777951660023,
      "grad_norm": 0.10648701339960098,
      "learning_rate": 4.0103358144174544e-08,
      "loss": 0.0009,
      "step": 300510
    },
    {
      "epoch": 1.9274419320597884,
      "grad_norm": 0.07527918368577957,
      "learning_rate": 4.003264259798023e-08,
      "loss": 0.0022,
      "step": 300520
    },
    {
      "epoch": 1.9275060689535746,
      "grad_norm": 0.018936704844236374,
      "learning_rate": 3.996198920372174e-08,
      "loss": 0.0015,
      "step": 300530
    },
    {
      "epoch": 1.9275702058473607,
      "grad_norm": 0.09216234087944031,
      "learning_rate": 3.989139796228447e-08,
      "loss": 0.0016,
      "step": 300540
    },
    {
      "epoch": 1.9276343427411469,
      "grad_norm": 0.02676011249423027,
      "learning_rate": 3.982086887455272e-08,
      "loss": 0.0007,
      "step": 300550
    },
    {
      "epoch": 1.9276984796349328,
      "grad_norm": 0.14437265694141388,
      "learning_rate": 3.9750401941410775e-08,
      "loss": 0.0011,
      "step": 300560
    },
    {
      "epoch": 1.927762616528719,
      "grad_norm": 0.008645369671285152,
      "learning_rate": 3.967999716374127e-08,
      "loss": 0.0003,
      "step": 300570
    },
    {
      "epoch": 1.9278267534225049,
      "grad_norm": 0.028107842430472374,
      "learning_rate": 3.9609654542425716e-08,
      "loss": 0.0008,
      "step": 300580
    },
    {
      "epoch": 1.927890890316291,
      "grad_norm": 0.022598829120397568,
      "learning_rate": 3.953937407834729e-08,
      "loss": 0.001,
      "step": 300590
    },
    {
      "epoch": 1.9279550272100772,
      "grad_norm": 0.0549064464867115,
      "learning_rate": 3.9469155772385304e-08,
      "loss": 0.0004,
      "step": 300600
    },
    {
      "epoch": 1.9280191641038633,
      "grad_norm": 0.07589433342218399,
      "learning_rate": 3.9398999625420154e-08,
      "loss": 0.001,
      "step": 300610
    },
    {
      "epoch": 1.9280833009976495,
      "grad_norm": 0.09359247982501984,
      "learning_rate": 3.932890563833058e-08,
      "loss": 0.0008,
      "step": 300620
    },
    {
      "epoch": 1.9281474378914356,
      "grad_norm": 0.10599889606237411,
      "learning_rate": 3.9258873811995334e-08,
      "loss": 0.001,
      "step": 300630
    },
    {
      "epoch": 1.9282115747852215,
      "grad_norm": 0.08819228410720825,
      "learning_rate": 3.918890414729204e-08,
      "loss": 0.0009,
      "step": 300640
    },
    {
      "epoch": 1.9282757116790077,
      "grad_norm": 0.08009287714958191,
      "learning_rate": 3.9118996645096666e-08,
      "loss": 0.0008,
      "step": 300650
    },
    {
      "epoch": 1.9283398485727936,
      "grad_norm": 0.11573512852191925,
      "learning_rate": 3.904915130628628e-08,
      "loss": 0.0012,
      "step": 300660
    },
    {
      "epoch": 1.9284039854665798,
      "grad_norm": 0.0738484263420105,
      "learning_rate": 3.8979368131735196e-08,
      "loss": 0.0011,
      "step": 300670
    },
    {
      "epoch": 1.928468122360366,
      "grad_norm": 0.007015854585915804,
      "learning_rate": 3.890964712231826e-08,
      "loss": 0.0012,
      "step": 300680
    },
    {
      "epoch": 1.928532259254152,
      "grad_norm": 0.19372880458831787,
      "learning_rate": 3.883998827890923e-08,
      "loss": 0.0018,
      "step": 300690
    },
    {
      "epoch": 1.9285963961479382,
      "grad_norm": 0.024199509993195534,
      "learning_rate": 3.877039160238072e-08,
      "loss": 0.0006,
      "step": 300700
    },
    {
      "epoch": 1.9286605330417244,
      "grad_norm": 0.30805060267448425,
      "learning_rate": 3.870085709360538e-08,
      "loss": 0.0013,
      "step": 300710
    },
    {
      "epoch": 1.9287246699355105,
      "grad_norm": 0.01284112874418497,
      "learning_rate": 3.863138475345363e-08,
      "loss": 0.0019,
      "step": 300720
    },
    {
      "epoch": 1.9287888068292964,
      "grad_norm": 0.057999007403850555,
      "learning_rate": 3.8561974582796423e-08,
      "loss": 0.0007,
      "step": 300730
    },
    {
      "epoch": 1.9288529437230826,
      "grad_norm": 0.3170008659362793,
      "learning_rate": 3.849262658250363e-08,
      "loss": 0.0021,
      "step": 300740
    },
    {
      "epoch": 1.9289170806168685,
      "grad_norm": 0.024755142629146576,
      "learning_rate": 3.842334075344401e-08,
      "loss": 0.002,
      "step": 300750
    },
    {
      "epoch": 1.9289812175106547,
      "grad_norm": 0.12757644057273865,
      "learning_rate": 3.8354117096486285e-08,
      "loss": 0.001,
      "step": 300760
    },
    {
      "epoch": 1.9290453544044408,
      "grad_norm": 0.04172433167695999,
      "learning_rate": 3.828495561249757e-08,
      "loss": 0.0009,
      "step": 300770
    },
    {
      "epoch": 1.929109491298227,
      "grad_norm": 0.03098372370004654,
      "learning_rate": 3.8215856302343815e-08,
      "loss": 0.0012,
      "step": 300780
    },
    {
      "epoch": 1.929173628192013,
      "grad_norm": 0.04408879205584526,
      "learning_rate": 3.8146819166892115e-08,
      "loss": 0.0012,
      "step": 300790
    },
    {
      "epoch": 1.9292377650857992,
      "grad_norm": 0.039453618228435516,
      "learning_rate": 3.8077844207006774e-08,
      "loss": 0.0008,
      "step": 300800
    },
    {
      "epoch": 1.9293019019795852,
      "grad_norm": 0.14602237939834595,
      "learning_rate": 3.800893142355211e-08,
      "loss": 0.0009,
      "step": 300810
    },
    {
      "epoch": 1.9293660388733713,
      "grad_norm": 0.09760798513889313,
      "learning_rate": 3.794008081739187e-08,
      "loss": 0.0009,
      "step": 300820
    },
    {
      "epoch": 1.9294301757671575,
      "grad_norm": 0.004577153827995062,
      "learning_rate": 3.787129238938925e-08,
      "loss": 0.002,
      "step": 300830
    },
    {
      "epoch": 1.9294943126609434,
      "grad_norm": 0.0346844345331192,
      "learning_rate": 3.7802566140405225e-08,
      "loss": 0.0011,
      "step": 300840
    },
    {
      "epoch": 1.9295584495547295,
      "grad_norm": 0.2503458857536316,
      "learning_rate": 3.773390207130134e-08,
      "loss": 0.0011,
      "step": 300850
    },
    {
      "epoch": 1.9296225864485157,
      "grad_norm": 0.006880940869450569,
      "learning_rate": 3.7665300182938566e-08,
      "loss": 0.0011,
      "step": 300860
    },
    {
      "epoch": 1.9296867233423018,
      "grad_norm": 0.06815887242555618,
      "learning_rate": 3.759676047617622e-08,
      "loss": 0.0012,
      "step": 300870
    },
    {
      "epoch": 1.929750860236088,
      "grad_norm": 0.2355438768863678,
      "learning_rate": 3.75282829518725e-08,
      "loss": 0.0018,
      "step": 300880
    },
    {
      "epoch": 1.9298149971298741,
      "grad_norm": 0.14180922508239746,
      "learning_rate": 3.745986761088616e-08,
      "loss": 0.0011,
      "step": 300890
    },
    {
      "epoch": 1.92987913402366,
      "grad_norm": 0.01800578273832798,
      "learning_rate": 3.73915144540743e-08,
      "loss": 0.0007,
      "step": 300900
    },
    {
      "epoch": 1.9299432709174462,
      "grad_norm": 0.0636366456747055,
      "learning_rate": 3.7323223482294e-08,
      "loss": 0.0032,
      "step": 300910
    },
    {
      "epoch": 1.9300074078112321,
      "grad_norm": 0.04221118241548538,
      "learning_rate": 3.7254994696399596e-08,
      "loss": 0.0007,
      "step": 300920
    },
    {
      "epoch": 1.9300715447050183,
      "grad_norm": 0.03675359487533569,
      "learning_rate": 3.7186828097247607e-08,
      "loss": 0.0009,
      "step": 300930
    },
    {
      "epoch": 1.9301356815988044,
      "grad_norm": 0.02007216587662697,
      "learning_rate": 3.711872368569125e-08,
      "loss": 0.0011,
      "step": 300940
    },
    {
      "epoch": 1.9301998184925906,
      "grad_norm": 0.12596172094345093,
      "learning_rate": 3.705068146258428e-08,
      "loss": 0.0023,
      "step": 300950
    },
    {
      "epoch": 1.9302639553863767,
      "grad_norm": 0.047864366322755814,
      "learning_rate": 3.6982701428779356e-08,
      "loss": 0.0008,
      "step": 300960
    },
    {
      "epoch": 1.9303280922801629,
      "grad_norm": 0.012640939094126225,
      "learning_rate": 3.6914783585128014e-08,
      "loss": 0.001,
      "step": 300970
    },
    {
      "epoch": 1.930392229173949,
      "grad_norm": 0.0019188645528629422,
      "learning_rate": 3.684692793248179e-08,
      "loss": 0.001,
      "step": 300980
    },
    {
      "epoch": 1.930456366067735,
      "grad_norm": 0.13600589334964752,
      "learning_rate": 3.677913447169057e-08,
      "loss": 0.0008,
      "step": 300990
    },
    {
      "epoch": 1.9305205029615211,
      "grad_norm": 0.041920121759176254,
      "learning_rate": 3.671140320360367e-08,
      "loss": 0.0012,
      "step": 301000
    },
    {
      "epoch": 1.930584639855307,
      "grad_norm": 0.050655558705329895,
      "learning_rate": 3.6643734129070406e-08,
      "loss": 0.0009,
      "step": 301010
    },
    {
      "epoch": 1.9306487767490932,
      "grad_norm": 0.12476855516433716,
      "learning_rate": 3.6576127248938444e-08,
      "loss": 0.001,
      "step": 301020
    },
    {
      "epoch": 1.9307129136428793,
      "grad_norm": 0.1970900446176529,
      "learning_rate": 3.650858256405432e-08,
      "loss": 0.0009,
      "step": 301030
    },
    {
      "epoch": 1.9307770505366655,
      "grad_norm": 0.009478774853050709,
      "learning_rate": 3.644110007526569e-08,
      "loss": 0.0012,
      "step": 301040
    },
    {
      "epoch": 1.9308411874304516,
      "grad_norm": 0.14857624471187592,
      "learning_rate": 3.637367978341688e-08,
      "loss": 0.0007,
      "step": 301050
    },
    {
      "epoch": 1.9309053243242378,
      "grad_norm": 0.0902068167924881,
      "learning_rate": 3.630632168935389e-08,
      "loss": 0.0009,
      "step": 301060
    },
    {
      "epoch": 1.9309694612180237,
      "grad_norm": 0.15382826328277588,
      "learning_rate": 3.6239025793919914e-08,
      "loss": 0.0008,
      "step": 301070
    },
    {
      "epoch": 1.9310335981118099,
      "grad_norm": 0.0057074157521128654,
      "learning_rate": 3.617179209795873e-08,
      "loss": 0.0008,
      "step": 301080
    },
    {
      "epoch": 1.9310977350055958,
      "grad_norm": 0.027745436877012253,
      "learning_rate": 3.610462060231246e-08,
      "loss": 0.0012,
      "step": 301090
    },
    {
      "epoch": 1.931161871899382,
      "grad_norm": 0.09244703501462936,
      "learning_rate": 3.603751130782263e-08,
      "loss": 0.0012,
      "step": 301100
    },
    {
      "epoch": 1.931226008793168,
      "grad_norm": 0.12194220721721649,
      "learning_rate": 3.597046421533079e-08,
      "loss": 0.0006,
      "step": 301110
    },
    {
      "epoch": 1.9312901456869542,
      "grad_norm": 0.07963887602090836,
      "learning_rate": 3.590347932567684e-08,
      "loss": 0.0008,
      "step": 301120
    },
    {
      "epoch": 1.9313542825807404,
      "grad_norm": 0.1852940022945404,
      "learning_rate": 3.5836556639700096e-08,
      "loss": 0.0016,
      "step": 301130
    },
    {
      "epoch": 1.9314184194745265,
      "grad_norm": 0.0025362635497003794,
      "learning_rate": 3.5769696158238775e-08,
      "loss": 0.0006,
      "step": 301140
    },
    {
      "epoch": 1.9314825563683127,
      "grad_norm": 0.1968517005443573,
      "learning_rate": 3.57028978821311e-08,
      "loss": 0.0007,
      "step": 301150
    },
    {
      "epoch": 1.9315466932620986,
      "grad_norm": 0.11529047787189484,
      "learning_rate": 3.5636161812214186e-08,
      "loss": 0.0012,
      "step": 301160
    },
    {
      "epoch": 1.9316108301558848,
      "grad_norm": 0.022898223251104355,
      "learning_rate": 3.5569487949324576e-08,
      "loss": 0.0097,
      "step": 301170
    },
    {
      "epoch": 1.9316749670496707,
      "grad_norm": 0.07718861103057861,
      "learning_rate": 3.5502876294296605e-08,
      "loss": 0.0006,
      "step": 301180
    },
    {
      "epoch": 1.9317391039434568,
      "grad_norm": 0.06445500254631042,
      "learning_rate": 3.543632684796572e-08,
      "loss": 0.0009,
      "step": 301190
    },
    {
      "epoch": 1.931803240837243,
      "grad_norm": 0.045879118144512177,
      "learning_rate": 3.5369839611166244e-08,
      "loss": 0.0008,
      "step": 301200
    },
    {
      "epoch": 1.9318673777310291,
      "grad_norm": 0.12509427964687347,
      "learning_rate": 3.530341458473085e-08,
      "loss": 0.0009,
      "step": 301210
    },
    {
      "epoch": 1.9319315146248153,
      "grad_norm": 0.041130181401968,
      "learning_rate": 3.5237051769491104e-08,
      "loss": 0.0008,
      "step": 301220
    },
    {
      "epoch": 1.9319956515186014,
      "grad_norm": 0.09322613477706909,
      "learning_rate": 3.517075116628021e-08,
      "loss": 0.0006,
      "step": 301230
    },
    {
      "epoch": 1.9320597884123873,
      "grad_norm": 0.15340624749660492,
      "learning_rate": 3.510451277592752e-08,
      "loss": 0.0011,
      "step": 301240
    },
    {
      "epoch": 1.9321239253061735,
      "grad_norm": 0.005170913878828287,
      "learning_rate": 3.503833659926403e-08,
      "loss": 0.0013,
      "step": 301250
    },
    {
      "epoch": 1.9321880621999596,
      "grad_norm": 0.04025174677371979,
      "learning_rate": 3.4972222637118524e-08,
      "loss": 0.0026,
      "step": 301260
    },
    {
      "epoch": 1.9322521990937456,
      "grad_norm": 0.061025556176900864,
      "learning_rate": 3.490617089031978e-08,
      "loss": 0.005,
      "step": 301270
    },
    {
      "epoch": 1.9323163359875317,
      "grad_norm": 0.04001593962311745,
      "learning_rate": 3.484018135969491e-08,
      "loss": 0.0011,
      "step": 301280
    },
    {
      "epoch": 1.9323804728813179,
      "grad_norm": 0.11239849030971527,
      "learning_rate": 3.477425404607104e-08,
      "loss": 0.001,
      "step": 301290
    },
    {
      "epoch": 1.932444609775104,
      "grad_norm": 0.023942064493894577,
      "learning_rate": 3.4708388950274174e-08,
      "loss": 0.0005,
      "step": 301300
    },
    {
      "epoch": 1.9325087466688902,
      "grad_norm": 0.08422784507274628,
      "learning_rate": 3.464258607313031e-08,
      "loss": 0.0019,
      "step": 301310
    },
    {
      "epoch": 1.9325728835626763,
      "grad_norm": 0.10249890387058258,
      "learning_rate": 3.457684541546325e-08,
      "loss": 0.0011,
      "step": 301320
    },
    {
      "epoch": 1.9326370204564622,
      "grad_norm": 0.07031677663326263,
      "learning_rate": 3.451116697809731e-08,
      "loss": 0.0007,
      "step": 301330
    },
    {
      "epoch": 1.9327011573502484,
      "grad_norm": 0.0049631018191576,
      "learning_rate": 3.444555076185463e-08,
      "loss": 0.0012,
      "step": 301340
    },
    {
      "epoch": 1.9327652942440343,
      "grad_norm": 0.0954102948307991,
      "learning_rate": 3.437999676755843e-08,
      "loss": 0.0017,
      "step": 301350
    },
    {
      "epoch": 1.9328294311378205,
      "grad_norm": 0.06614626944065094,
      "learning_rate": 3.431450499602973e-08,
      "loss": 0.0006,
      "step": 301360
    },
    {
      "epoch": 1.9328935680316066,
      "grad_norm": 0.02178000658750534,
      "learning_rate": 3.424907544808953e-08,
      "loss": 0.001,
      "step": 301370
    },
    {
      "epoch": 1.9329577049253928,
      "grad_norm": 0.024998584762215614,
      "learning_rate": 3.418370812455718e-08,
      "loss": 0.0009,
      "step": 301380
    },
    {
      "epoch": 1.933021841819179,
      "grad_norm": 0.03699403628706932,
      "learning_rate": 3.4118403026251464e-08,
      "loss": 0.0008,
      "step": 301390
    },
    {
      "epoch": 1.933085978712965,
      "grad_norm": 0.039904557168483734,
      "learning_rate": 3.405316015399174e-08,
      "loss": 0.0025,
      "step": 301400
    },
    {
      "epoch": 1.9331501156067512,
      "grad_norm": 0.05057976767420769,
      "learning_rate": 3.398797950859511e-08,
      "loss": 0.0009,
      "step": 301410
    },
    {
      "epoch": 1.9332142525005371,
      "grad_norm": 0.003374518360942602,
      "learning_rate": 3.392286109087817e-08,
      "loss": 0.001,
      "step": 301420
    },
    {
      "epoch": 1.9332783893943233,
      "grad_norm": 0.07602231949567795,
      "learning_rate": 3.385780490165691e-08,
      "loss": 0.0019,
      "step": 301430
    },
    {
      "epoch": 1.9333425262881092,
      "grad_norm": 0.08708921074867249,
      "learning_rate": 3.379281094174625e-08,
      "loss": 0.0007,
      "step": 301440
    },
    {
      "epoch": 1.9334066631818954,
      "grad_norm": 0.02393578365445137,
      "learning_rate": 3.372787921196108e-08,
      "loss": 0.0006,
      "step": 301450
    },
    {
      "epoch": 1.9334708000756815,
      "grad_norm": 0.12386545538902283,
      "learning_rate": 3.36630097131152e-08,
      "loss": 0.0006,
      "step": 301460
    },
    {
      "epoch": 1.9335349369694677,
      "grad_norm": 0.07013312727212906,
      "learning_rate": 3.35982024460213e-08,
      "loss": 0.0007,
      "step": 301470
    },
    {
      "epoch": 1.9335990738632538,
      "grad_norm": 0.040222447365522385,
      "learning_rate": 3.35334574114915e-08,
      "loss": 0.0005,
      "step": 301480
    },
    {
      "epoch": 1.93366321075704,
      "grad_norm": 0.030433671548962593,
      "learning_rate": 3.346877461033626e-08,
      "loss": 0.0017,
      "step": 301490
    },
    {
      "epoch": 1.9337273476508259,
      "grad_norm": 0.16435152292251587,
      "learning_rate": 3.340415404336716e-08,
      "loss": 0.0015,
      "step": 301500
    },
    {
      "epoch": 1.933791484544612,
      "grad_norm": 0.056320659816265106,
      "learning_rate": 3.333959571139356e-08,
      "loss": 0.0012,
      "step": 301510
    },
    {
      "epoch": 1.933855621438398,
      "grad_norm": 0.18155111372470856,
      "learning_rate": 3.327509961522479e-08,
      "loss": 0.0015,
      "step": 301520
    },
    {
      "epoch": 1.933919758332184,
      "grad_norm": 0.026386840268969536,
      "learning_rate": 3.3210665755668004e-08,
      "loss": 0.0021,
      "step": 301530
    },
    {
      "epoch": 1.9339838952259703,
      "grad_norm": 0.024649258702993393,
      "learning_rate": 3.3146294133531984e-08,
      "loss": 0.0008,
      "step": 301540
    },
    {
      "epoch": 1.9340480321197564,
      "grad_norm": 0.0012394103687256575,
      "learning_rate": 3.308198474962221e-08,
      "loss": 0.0015,
      "step": 301550
    },
    {
      "epoch": 1.9341121690135425,
      "grad_norm": 0.0007696148240938783,
      "learning_rate": 3.301773760474525e-08,
      "loss": 0.0009,
      "step": 301560
    },
    {
      "epoch": 1.9341763059073287,
      "grad_norm": 0.058741334825754166,
      "learning_rate": 3.295355269970546e-08,
      "loss": 0.0019,
      "step": 301570
    },
    {
      "epoch": 1.9342404428011148,
      "grad_norm": 0.20647259056568146,
      "learning_rate": 3.288943003530831e-08,
      "loss": 0.001,
      "step": 301580
    },
    {
      "epoch": 1.9343045796949008,
      "grad_norm": 0.03462643548846245,
      "learning_rate": 3.282536961235594e-08,
      "loss": 0.0007,
      "step": 301590
    },
    {
      "epoch": 1.934368716588687,
      "grad_norm": 0.0032162105198949575,
      "learning_rate": 3.276137143165159e-08,
      "loss": 0.0011,
      "step": 301600
    },
    {
      "epoch": 1.9344328534824728,
      "grad_norm": 0.059601809829473495,
      "learning_rate": 3.26974354939974e-08,
      "loss": 0.0008,
      "step": 301610
    },
    {
      "epoch": 1.934496990376259,
      "grad_norm": 0.012852217070758343,
      "learning_rate": 3.2633561800194945e-08,
      "loss": 0.0011,
      "step": 301620
    },
    {
      "epoch": 1.9345611272700451,
      "grad_norm": 0.13055984675884247,
      "learning_rate": 3.256975035104304e-08,
      "loss": 0.0006,
      "step": 301630
    },
    {
      "epoch": 1.9346252641638313,
      "grad_norm": 0.14420166611671448,
      "learning_rate": 3.250600114734326e-08,
      "loss": 0.0011,
      "step": 301640
    },
    {
      "epoch": 1.9346894010576174,
      "grad_norm": 0.25341367721557617,
      "learning_rate": 3.24423141898933e-08,
      "loss": 0.0013,
      "step": 301650
    },
    {
      "epoch": 1.9347535379514036,
      "grad_norm": 0.02456929162144661,
      "learning_rate": 3.237868947949141e-08,
      "loss": 0.0008,
      "step": 301660
    },
    {
      "epoch": 1.9348176748451897,
      "grad_norm": 0.10025903582572937,
      "learning_rate": 3.231512701693418e-08,
      "loss": 0.0005,
      "step": 301670
    },
    {
      "epoch": 1.9348818117389757,
      "grad_norm": 0.04823238030076027,
      "learning_rate": 3.22516268030193e-08,
      "loss": 0.0007,
      "step": 301680
    },
    {
      "epoch": 1.9349459486327618,
      "grad_norm": 0.011798826046288013,
      "learning_rate": 3.2188188838542246e-08,
      "loss": 0.0012,
      "step": 301690
    },
    {
      "epoch": 1.9350100855265477,
      "grad_norm": 0.04304208979010582,
      "learning_rate": 3.212481312429738e-08,
      "loss": 0.0008,
      "step": 301700
    },
    {
      "epoch": 1.935074222420334,
      "grad_norm": 0.0785306766629219,
      "learning_rate": 3.2061499661079075e-08,
      "loss": 0.0016,
      "step": 301710
    },
    {
      "epoch": 1.93513835931412,
      "grad_norm": 0.012932732701301575,
      "learning_rate": 3.199824844968058e-08,
      "loss": 0.0005,
      "step": 301720
    },
    {
      "epoch": 1.9352024962079062,
      "grad_norm": 0.0498775988817215,
      "learning_rate": 3.193505949089459e-08,
      "loss": 0.0009,
      "step": 301730
    },
    {
      "epoch": 1.9352666331016923,
      "grad_norm": 0.20959006249904633,
      "learning_rate": 3.187193278551326e-08,
      "loss": 0.0014,
      "step": 301740
    },
    {
      "epoch": 1.9353307699954785,
      "grad_norm": 0.057461485266685486,
      "learning_rate": 3.1808868334327056e-08,
      "loss": 0.0006,
      "step": 301750
    },
    {
      "epoch": 1.9353949068892644,
      "grad_norm": 0.052913159132003784,
      "learning_rate": 3.1745866138126466e-08,
      "loss": 0.0019,
      "step": 301760
    },
    {
      "epoch": 1.9354590437830506,
      "grad_norm": 0.12710267305374146,
      "learning_rate": 3.168292619770086e-08,
      "loss": 0.0013,
      "step": 301770
    },
    {
      "epoch": 1.9355231806768365,
      "grad_norm": 0.1411878615617752,
      "learning_rate": 3.1620048513838485e-08,
      "loss": 0.0011,
      "step": 301780
    },
    {
      "epoch": 1.9355873175706226,
      "grad_norm": 0.14976029098033905,
      "learning_rate": 3.155723308732872e-08,
      "loss": 0.0024,
      "step": 301790
    },
    {
      "epoch": 1.9356514544644088,
      "grad_norm": 0.10860023647546768,
      "learning_rate": 3.1494479918957044e-08,
      "loss": 0.0011,
      "step": 301800
    },
    {
      "epoch": 1.935715591358195,
      "grad_norm": 0.09502911567687988,
      "learning_rate": 3.143178900951061e-08,
      "loss": 0.0012,
      "step": 301810
    },
    {
      "epoch": 1.935779728251981,
      "grad_norm": 0.05954446643590927,
      "learning_rate": 3.136916035977489e-08,
      "loss": 0.0013,
      "step": 301820
    },
    {
      "epoch": 1.9358438651457672,
      "grad_norm": 0.05650576576590538,
      "learning_rate": 3.1306593970534814e-08,
      "loss": 0.0008,
      "step": 301830
    },
    {
      "epoch": 1.9359080020395534,
      "grad_norm": 0.13886955380439758,
      "learning_rate": 3.124408984257421e-08,
      "loss": 0.0015,
      "step": 301840
    },
    {
      "epoch": 1.9359721389333393,
      "grad_norm": 0.0654849037528038,
      "learning_rate": 3.1181647976676334e-08,
      "loss": 0.0012,
      "step": 301850
    },
    {
      "epoch": 1.9360362758271255,
      "grad_norm": 0.02531741000711918,
      "learning_rate": 3.1119268373623336e-08,
      "loss": 0.0008,
      "step": 301860
    },
    {
      "epoch": 1.9361004127209114,
      "grad_norm": 0.03234170749783516,
      "learning_rate": 3.105695103419682e-08,
      "loss": 0.0007,
      "step": 301870
    },
    {
      "epoch": 1.9361645496146975,
      "grad_norm": 0.07435993105173111,
      "learning_rate": 3.099469595917837e-08,
      "loss": 0.0007,
      "step": 301880
    },
    {
      "epoch": 1.9362286865084837,
      "grad_norm": 0.06573089957237244,
      "learning_rate": 3.093250314934737e-08,
      "loss": 0.0011,
      "step": 301890
    },
    {
      "epoch": 1.9362928234022698,
      "grad_norm": 0.0297558456659317,
      "learning_rate": 3.087037260548376e-08,
      "loss": 0.0005,
      "step": 301900
    },
    {
      "epoch": 1.936356960296056,
      "grad_norm": 0.03819414973258972,
      "learning_rate": 3.080830432836579e-08,
      "loss": 0.0006,
      "step": 301910
    },
    {
      "epoch": 1.9364210971898421,
      "grad_norm": 0.07690879702568054,
      "learning_rate": 3.074629831877119e-08,
      "loss": 0.0007,
      "step": 301920
    },
    {
      "epoch": 1.936485234083628,
      "grad_norm": 0.06591645628213882,
      "learning_rate": 3.0684354577476536e-08,
      "loss": 0.001,
      "step": 301930
    },
    {
      "epoch": 1.9365493709774142,
      "grad_norm": 0.08025170862674713,
      "learning_rate": 3.062247310525845e-08,
      "loss": 0.0009,
      "step": 301940
    },
    {
      "epoch": 1.9366135078712001,
      "grad_norm": 0.044352661818265915,
      "learning_rate": 3.056065390289298e-08,
      "loss": 0.0008,
      "step": 301950
    },
    {
      "epoch": 1.9366776447649863,
      "grad_norm": 0.02093079872429371,
      "learning_rate": 3.049889697115393e-08,
      "loss": 0.0016,
      "step": 301960
    },
    {
      "epoch": 1.9367417816587724,
      "grad_norm": 0.0008874760824255645,
      "learning_rate": 3.043720231081515e-08,
      "loss": 0.0017,
      "step": 301970
    },
    {
      "epoch": 1.9368059185525586,
      "grad_norm": 0.03476732224225998,
      "learning_rate": 3.037556992264934e-08,
      "loss": 0.0008,
      "step": 301980
    },
    {
      "epoch": 1.9368700554463447,
      "grad_norm": 0.027935517951846123,
      "learning_rate": 3.0313999807430324e-08,
      "loss": 0.0014,
      "step": 301990
    },
    {
      "epoch": 1.9369341923401309,
      "grad_norm": 0.06478878855705261,
      "learning_rate": 3.0252491965928057e-08,
      "loss": 0.0009,
      "step": 302000
    },
    {
      "epoch": 1.936998329233917,
      "grad_norm": 0.0036114610265940428,
      "learning_rate": 3.019104639891468e-08,
      "loss": 0.0009,
      "step": 302010
    },
    {
      "epoch": 1.937062466127703,
      "grad_norm": 0.007749971468001604,
      "learning_rate": 3.012966310715848e-08,
      "loss": 0.0007,
      "step": 302020
    },
    {
      "epoch": 1.937126603021489,
      "grad_norm": 0.0017212193924933672,
      "learning_rate": 3.0068342091429945e-08,
      "loss": 0.0007,
      "step": 302030
    },
    {
      "epoch": 1.937190739915275,
      "grad_norm": 0.02803710475564003,
      "learning_rate": 3.00070833524968e-08,
      "loss": 0.0013,
      "step": 302040
    },
    {
      "epoch": 1.9372548768090612,
      "grad_norm": 0.11851103603839874,
      "learning_rate": 2.994588689112676e-08,
      "loss": 0.0007,
      "step": 302050
    },
    {
      "epoch": 1.9373190137028473,
      "grad_norm": 0.10424335300922394,
      "learning_rate": 2.988475270808755e-08,
      "loss": 0.0014,
      "step": 302060
    },
    {
      "epoch": 1.9373831505966335,
      "grad_norm": 0.16494068503379822,
      "learning_rate": 2.982368080414411e-08,
      "loss": 0.0007,
      "step": 302070
    },
    {
      "epoch": 1.9374472874904196,
      "grad_norm": 0.032014600932598114,
      "learning_rate": 2.976267118006193e-08,
      "loss": 0.0011,
      "step": 302080
    },
    {
      "epoch": 1.9375114243842058,
      "grad_norm": 0.06066601350903511,
      "learning_rate": 2.970172383660541e-08,
      "loss": 0.0009,
      "step": 302090
    },
    {
      "epoch": 1.937575561277992,
      "grad_norm": 0.0030617748852819204,
      "learning_rate": 2.964083877453894e-08,
      "loss": 0.0005,
      "step": 302100
    },
    {
      "epoch": 1.9376396981717778,
      "grad_norm": 0.1492859423160553,
      "learning_rate": 2.9580015994625234e-08,
      "loss": 0.0009,
      "step": 302110
    },
    {
      "epoch": 1.937703835065564,
      "grad_norm": 0.10125358402729034,
      "learning_rate": 2.951925549762591e-08,
      "loss": 0.0025,
      "step": 302120
    },
    {
      "epoch": 1.93776797195935,
      "grad_norm": 0.11417628824710846,
      "learning_rate": 2.9458557284302027e-08,
      "loss": 0.0013,
      "step": 302130
    },
    {
      "epoch": 1.937832108853136,
      "grad_norm": 0.041539266705513,
      "learning_rate": 2.9397921355415747e-08,
      "loss": 0.0053,
      "step": 302140
    },
    {
      "epoch": 1.9378962457469222,
      "grad_norm": 0.05827486515045166,
      "learning_rate": 2.9337347711725917e-08,
      "loss": 0.0009,
      "step": 302150
    },
    {
      "epoch": 1.9379603826407084,
      "grad_norm": 0.30581918358802795,
      "learning_rate": 2.9276836353991366e-08,
      "loss": 0.002,
      "step": 302160
    },
    {
      "epoch": 1.9380245195344945,
      "grad_norm": 0.02726643532514572,
      "learning_rate": 2.921638728297038e-08,
      "loss": 0.0011,
      "step": 302170
    },
    {
      "epoch": 1.9380886564282807,
      "grad_norm": 0.025319676846265793,
      "learning_rate": 2.915600049942069e-08,
      "loss": 0.0004,
      "step": 302180
    },
    {
      "epoch": 1.9381527933220666,
      "grad_norm": 0.0578426830470562,
      "learning_rate": 2.9095676004098905e-08,
      "loss": 0.0005,
      "step": 302190
    },
    {
      "epoch": 1.9382169302158527,
      "grad_norm": 0.007352834101766348,
      "learning_rate": 2.903541379776109e-08,
      "loss": 0.001,
      "step": 302200
    },
    {
      "epoch": 1.9382810671096387,
      "grad_norm": 0.020987290889024734,
      "learning_rate": 2.8975213881162202e-08,
      "loss": 0.0009,
      "step": 302210
    },
    {
      "epoch": 1.9383452040034248,
      "grad_norm": 0.14074473083019257,
      "learning_rate": 2.8915076255056628e-08,
      "loss": 0.0008,
      "step": 302220
    },
    {
      "epoch": 1.938409340897211,
      "grad_norm": 0.0046698665246367455,
      "learning_rate": 2.8855000920198217e-08,
      "loss": 0.0014,
      "step": 302230
    },
    {
      "epoch": 1.938473477790997,
      "grad_norm": 0.08994993567466736,
      "learning_rate": 2.8794987877338588e-08,
      "loss": 0.0011,
      "step": 302240
    },
    {
      "epoch": 1.9385376146847833,
      "grad_norm": 0.03455111384391785,
      "learning_rate": 2.873503712723158e-08,
      "loss": 0.0012,
      "step": 302250
    },
    {
      "epoch": 1.9386017515785694,
      "grad_norm": 0.014766179956495762,
      "learning_rate": 2.867514867062715e-08,
      "loss": 0.0009,
      "step": 302260
    },
    {
      "epoch": 1.9386658884723555,
      "grad_norm": 0.006934038363397121,
      "learning_rate": 2.861532250827581e-08,
      "loss": 0.0006,
      "step": 302270
    },
    {
      "epoch": 1.9387300253661415,
      "grad_norm": 0.027285199612379074,
      "learning_rate": 2.8555558640927516e-08,
      "loss": 0.0008,
      "step": 302280
    },
    {
      "epoch": 1.9387941622599276,
      "grad_norm": 0.07449705898761749,
      "learning_rate": 2.8495857069331668e-08,
      "loss": 0.0005,
      "step": 302290
    },
    {
      "epoch": 1.9388582991537135,
      "grad_norm": 0.16933131217956543,
      "learning_rate": 2.8436217794235442e-08,
      "loss": 0.0011,
      "step": 302300
    },
    {
      "epoch": 1.9389224360474997,
      "grad_norm": 0.10641875863075256,
      "learning_rate": 2.837664081638658e-08,
      "loss": 0.0012,
      "step": 302310
    },
    {
      "epoch": 1.9389865729412858,
      "grad_norm": 0.13166961073875427,
      "learning_rate": 2.8317126136531148e-08,
      "loss": 0.0023,
      "step": 302320
    },
    {
      "epoch": 1.939050709835072,
      "grad_norm": 0.10253006219863892,
      "learning_rate": 2.825767375541577e-08,
      "loss": 0.001,
      "step": 302330
    },
    {
      "epoch": 1.9391148467288581,
      "grad_norm": 0.06676961481571198,
      "learning_rate": 2.8198283673784855e-08,
      "loss": 0.0006,
      "step": 302340
    },
    {
      "epoch": 1.9391789836226443,
      "grad_norm": 0.020916448906064034,
      "learning_rate": 2.8138955892382804e-08,
      "loss": 0.0009,
      "step": 302350
    },
    {
      "epoch": 1.9392431205164302,
      "grad_norm": 0.03495674952864647,
      "learning_rate": 2.807969041195291e-08,
      "loss": 0.0009,
      "step": 302360
    },
    {
      "epoch": 1.9393072574102164,
      "grad_norm": 0.10187196731567383,
      "learning_rate": 2.8020487233237916e-08,
      "loss": 0.0017,
      "step": 302370
    },
    {
      "epoch": 1.9393713943040025,
      "grad_norm": 0.08674407750368118,
      "learning_rate": 2.796134635698e-08,
      "loss": 0.0004,
      "step": 302380
    },
    {
      "epoch": 1.9394355311977884,
      "grad_norm": 0.01654968410730362,
      "learning_rate": 2.7902267783919136e-08,
      "loss": 0.0011,
      "step": 302390
    },
    {
      "epoch": 1.9394996680915746,
      "grad_norm": 0.42116132378578186,
      "learning_rate": 2.7843251514796943e-08,
      "loss": 0.0007,
      "step": 302400
    },
    {
      "epoch": 1.9395638049853607,
      "grad_norm": 0.17139796912670135,
      "learning_rate": 2.778429755035228e-08,
      "loss": 0.0013,
      "step": 302410
    },
    {
      "epoch": 1.9396279418791469,
      "grad_norm": 0.032461978495121,
      "learning_rate": 2.772540589132344e-08,
      "loss": 0.0008,
      "step": 302420
    },
    {
      "epoch": 1.939692078772933,
      "grad_norm": 0.051237910985946655,
      "learning_rate": 2.7666576538449285e-08,
      "loss": 0.0009,
      "step": 302430
    },
    {
      "epoch": 1.9397562156667192,
      "grad_norm": 0.09664763510227203,
      "learning_rate": 2.7607809492466997e-08,
      "loss": 0.0006,
      "step": 302440
    },
    {
      "epoch": 1.9398203525605051,
      "grad_norm": 0.0451032929122448,
      "learning_rate": 2.75491047541121e-08,
      "loss": 0.0006,
      "step": 302450
    },
    {
      "epoch": 1.9398844894542913,
      "grad_norm": 0.04460052028298378,
      "learning_rate": 2.749046232412067e-08,
      "loss": 0.003,
      "step": 302460
    },
    {
      "epoch": 1.9399486263480772,
      "grad_norm": 0.0648808553814888,
      "learning_rate": 2.7431882203227678e-08,
      "loss": 0.0008,
      "step": 302470
    },
    {
      "epoch": 1.9400127632418633,
      "grad_norm": 0.09854970127344131,
      "learning_rate": 2.737336439216698e-08,
      "loss": 0.0006,
      "step": 302480
    },
    {
      "epoch": 1.9400769001356495,
      "grad_norm": 0.08210846781730652,
      "learning_rate": 2.7314908891671875e-08,
      "loss": 0.0008,
      "step": 302490
    },
    {
      "epoch": 1.9401410370294356,
      "grad_norm": 0.04816485196352005,
      "learning_rate": 2.725651570247456e-08,
      "loss": 0.0005,
      "step": 302500
    },
    {
      "epoch": 1.9402051739232218,
      "grad_norm": 0.05821401998400688,
      "learning_rate": 2.7198184825307782e-08,
      "loss": 0.0013,
      "step": 302510
    },
    {
      "epoch": 1.940269310817008,
      "grad_norm": 0.07066737115383148,
      "learning_rate": 2.713991626090151e-08,
      "loss": 0.0012,
      "step": 302520
    },
    {
      "epoch": 1.940333447710794,
      "grad_norm": 0.048833467066287994,
      "learning_rate": 2.708171000998572e-08,
      "loss": 0.001,
      "step": 302530
    },
    {
      "epoch": 1.94039758460458,
      "grad_norm": 0.19674324989318848,
      "learning_rate": 2.7023566073290374e-08,
      "loss": 0.0008,
      "step": 302540
    },
    {
      "epoch": 1.9404617214983662,
      "grad_norm": 0.15980391204357147,
      "learning_rate": 2.6965484451544343e-08,
      "loss": 0.0011,
      "step": 302550
    },
    {
      "epoch": 1.940525858392152,
      "grad_norm": 0.029844066128134727,
      "learning_rate": 2.690746514547482e-08,
      "loss": 0.0005,
      "step": 302560
    },
    {
      "epoch": 1.9405899952859382,
      "grad_norm": 0.04482239857316017,
      "learning_rate": 2.6849508155808446e-08,
      "loss": 0.0017,
      "step": 302570
    },
    {
      "epoch": 1.9406541321797244,
      "grad_norm": 0.04891893267631531,
      "learning_rate": 2.6791613483272416e-08,
      "loss": 0.0007,
      "step": 302580
    },
    {
      "epoch": 1.9407182690735105,
      "grad_norm": 0.07550440728664398,
      "learning_rate": 2.6733781128591707e-08,
      "loss": 0.0019,
      "step": 302590
    },
    {
      "epoch": 1.9407824059672967,
      "grad_norm": 0.07568196952342987,
      "learning_rate": 2.667601109249074e-08,
      "loss": 0.0011,
      "step": 302600
    },
    {
      "epoch": 1.9408465428610828,
      "grad_norm": 0.05623389407992363,
      "learning_rate": 2.6618303375694487e-08,
      "loss": 0.0011,
      "step": 302610
    },
    {
      "epoch": 1.9409106797548688,
      "grad_norm": 0.07222924381494522,
      "learning_rate": 2.6560657978924598e-08,
      "loss": 0.0013,
      "step": 302620
    },
    {
      "epoch": 1.940974816648655,
      "grad_norm": 0.05307445302605629,
      "learning_rate": 2.650307490290438e-08,
      "loss": 0.0005,
      "step": 302630
    },
    {
      "epoch": 1.9410389535424408,
      "grad_norm": 0.1193118616938591,
      "learning_rate": 2.644555414835548e-08,
      "loss": 0.0013,
      "step": 302640
    },
    {
      "epoch": 1.941103090436227,
      "grad_norm": 0.01188894547522068,
      "learning_rate": 2.6388095715997875e-08,
      "loss": 0.001,
      "step": 302650
    },
    {
      "epoch": 1.9411672273300131,
      "grad_norm": 0.0663500651717186,
      "learning_rate": 2.63306996065521e-08,
      "loss": 0.0022,
      "step": 302660
    },
    {
      "epoch": 1.9412313642237993,
      "grad_norm": 0.10533638298511505,
      "learning_rate": 2.6273365820737028e-08,
      "loss": 0.0005,
      "step": 302670
    },
    {
      "epoch": 1.9412955011175854,
      "grad_norm": 0.061424218118190765,
      "learning_rate": 2.6216094359272083e-08,
      "loss": 0.0007,
      "step": 302680
    },
    {
      "epoch": 1.9413596380113716,
      "grad_norm": 0.058211881667375565,
      "learning_rate": 2.615888522287391e-08,
      "loss": 0.0022,
      "step": 302690
    },
    {
      "epoch": 1.9414237749051577,
      "grad_norm": 0.38756659626960754,
      "learning_rate": 2.6101738412259158e-08,
      "loss": 0.0033,
      "step": 302700
    },
    {
      "epoch": 1.9414879117989436,
      "grad_norm": 0.10443129390478134,
      "learning_rate": 2.6044653928144483e-08,
      "loss": 0.0023,
      "step": 302710
    },
    {
      "epoch": 1.9415520486927298,
      "grad_norm": 0.0632278099656105,
      "learning_rate": 2.598763177124597e-08,
      "loss": 0.0009,
      "step": 302720
    },
    {
      "epoch": 1.9416161855865157,
      "grad_norm": 0.15798984467983246,
      "learning_rate": 2.5930671942276386e-08,
      "loss": 0.0013,
      "step": 302730
    },
    {
      "epoch": 1.9416803224803019,
      "grad_norm": 0.11811117082834244,
      "learning_rate": 2.5873774441950718e-08,
      "loss": 0.0019,
      "step": 302740
    },
    {
      "epoch": 1.941744459374088,
      "grad_norm": 0.11421682685613632,
      "learning_rate": 2.581693927098172e-08,
      "loss": 0.0007,
      "step": 302750
    },
    {
      "epoch": 1.9418085962678742,
      "grad_norm": 0.11522186547517776,
      "learning_rate": 2.576016643008161e-08,
      "loss": 0.0011,
      "step": 302760
    },
    {
      "epoch": 1.9418727331616603,
      "grad_norm": 0.12734942138195038,
      "learning_rate": 2.570345591996093e-08,
      "loss": 0.0007,
      "step": 302770
    },
    {
      "epoch": 1.9419368700554465,
      "grad_norm": 0.16883651912212372,
      "learning_rate": 2.564680774133188e-08,
      "loss": 0.0006,
      "step": 302780
    },
    {
      "epoch": 1.9420010069492324,
      "grad_norm": 0.26879411935806274,
      "learning_rate": 2.5590221894903346e-08,
      "loss": 0.001,
      "step": 302790
    },
    {
      "epoch": 1.9420651438430185,
      "grad_norm": 0.027143623679876328,
      "learning_rate": 2.5533698381384196e-08,
      "loss": 0.0009,
      "step": 302800
    },
    {
      "epoch": 1.9421292807368047,
      "grad_norm": 0.21365833282470703,
      "learning_rate": 2.5477237201482764e-08,
      "loss": 0.0012,
      "step": 302810
    },
    {
      "epoch": 1.9421934176305906,
      "grad_norm": 0.1140991821885109,
      "learning_rate": 2.5420838355907364e-08,
      "loss": 0.001,
      "step": 302820
    },
    {
      "epoch": 1.9422575545243768,
      "grad_norm": 0.09537651389837265,
      "learning_rate": 2.5364501845363543e-08,
      "loss": 0.0012,
      "step": 302830
    },
    {
      "epoch": 1.942321691418163,
      "grad_norm": 0.002741380361840129,
      "learning_rate": 2.5308227670558517e-08,
      "loss": 0.0013,
      "step": 302840
    },
    {
      "epoch": 1.942385828311949,
      "grad_norm": 0.029889706522226334,
      "learning_rate": 2.5252015832196164e-08,
      "loss": 0.0009,
      "step": 302850
    },
    {
      "epoch": 1.9424499652057352,
      "grad_norm": 0.011436429806053638,
      "learning_rate": 2.519586633098148e-08,
      "loss": 0.0024,
      "step": 302860
    },
    {
      "epoch": 1.9425141020995214,
      "grad_norm": 0.09918776899576187,
      "learning_rate": 2.513977916761834e-08,
      "loss": 0.001,
      "step": 302870
    },
    {
      "epoch": 1.9425782389933073,
      "grad_norm": 0.03670860826969147,
      "learning_rate": 2.5083754342808963e-08,
      "loss": 0.0009,
      "step": 302880
    },
    {
      "epoch": 1.9426423758870934,
      "grad_norm": 0.14740601181983948,
      "learning_rate": 2.5027791857255567e-08,
      "loss": 0.0017,
      "step": 302890
    },
    {
      "epoch": 1.9427065127808794,
      "grad_norm": 0.0429193489253521,
      "learning_rate": 2.4971891711659813e-08,
      "loss": 0.0009,
      "step": 302900
    },
    {
      "epoch": 1.9427706496746655,
      "grad_norm": 0.0554153174161911,
      "learning_rate": 2.4916053906722247e-08,
      "loss": 0.0004,
      "step": 302910
    },
    {
      "epoch": 1.9428347865684517,
      "grad_norm": 0.22062762081623077,
      "learning_rate": 2.4860278443141206e-08,
      "loss": 0.0007,
      "step": 302920
    },
    {
      "epoch": 1.9428989234622378,
      "grad_norm": 0.02430759370326996,
      "learning_rate": 2.4804565321617235e-08,
      "loss": 0.0011,
      "step": 302930
    },
    {
      "epoch": 1.942963060356024,
      "grad_norm": 0.059388384222984314,
      "learning_rate": 2.474891454284756e-08,
      "loss": 0.0015,
      "step": 302940
    },
    {
      "epoch": 1.94302719724981,
      "grad_norm": 0.020595358684659004,
      "learning_rate": 2.4693326107529945e-08,
      "loss": 0.001,
      "step": 302950
    },
    {
      "epoch": 1.9430913341435962,
      "grad_norm": 0.10308534651994705,
      "learning_rate": 2.4637800016360512e-08,
      "loss": 0.0016,
      "step": 302960
    },
    {
      "epoch": 1.9431554710373822,
      "grad_norm": 0.1445896029472351,
      "learning_rate": 2.4582336270035367e-08,
      "loss": 0.0012,
      "step": 302970
    },
    {
      "epoch": 1.9432196079311683,
      "grad_norm": 0.1349262297153473,
      "learning_rate": 2.4526934869249507e-08,
      "loss": 0.001,
      "step": 302980
    },
    {
      "epoch": 1.9432837448249543,
      "grad_norm": 0.024352222681045532,
      "learning_rate": 2.4471595814696825e-08,
      "loss": 0.0012,
      "step": 302990
    },
    {
      "epoch": 1.9433478817187404,
      "grad_norm": 0.06732741743326187,
      "learning_rate": 2.4416319107071206e-08,
      "loss": 0.0015,
      "step": 303000
    },
    {
      "epoch": 1.9434120186125265,
      "grad_norm": 0.1403336524963379,
      "learning_rate": 2.436110474706488e-08,
      "loss": 0.0015,
      "step": 303010
    },
    {
      "epoch": 1.9434761555063127,
      "grad_norm": 0.005384480115026236,
      "learning_rate": 2.430595273537062e-08,
      "loss": 0.0008,
      "step": 303020
    },
    {
      "epoch": 1.9435402924000988,
      "grad_norm": 0.032813455909490585,
      "learning_rate": 2.4250863072678434e-08,
      "loss": 0.0015,
      "step": 303030
    },
    {
      "epoch": 1.943604429293885,
      "grad_norm": 0.08630481362342834,
      "learning_rate": 2.419583575967832e-08,
      "loss": 0.0015,
      "step": 303040
    },
    {
      "epoch": 1.943668566187671,
      "grad_norm": 0.045145533978939056,
      "learning_rate": 2.41408707970614e-08,
      "loss": 0.0012,
      "step": 303050
    },
    {
      "epoch": 1.943732703081457,
      "grad_norm": 0.04118989035487175,
      "learning_rate": 2.408596818551545e-08,
      "loss": 0.0013,
      "step": 303060
    },
    {
      "epoch": 1.943796839975243,
      "grad_norm": 0.10800248384475708,
      "learning_rate": 2.403112792572826e-08,
      "loss": 0.0012,
      "step": 303070
    },
    {
      "epoch": 1.9438609768690291,
      "grad_norm": 0.06161636859178543,
      "learning_rate": 2.3976350018387608e-08,
      "loss": 0.0016,
      "step": 303080
    },
    {
      "epoch": 1.9439251137628153,
      "grad_norm": 0.06434313952922821,
      "learning_rate": 2.392163446417961e-08,
      "loss": 0.0005,
      "step": 303090
    },
    {
      "epoch": 1.9439892506566014,
      "grad_norm": 0.028090180829167366,
      "learning_rate": 2.3866981263789835e-08,
      "loss": 0.0007,
      "step": 303100
    },
    {
      "epoch": 1.9440533875503876,
      "grad_norm": 0.01200695801526308,
      "learning_rate": 2.381239041790273e-08,
      "loss": 0.0012,
      "step": 303110
    },
    {
      "epoch": 1.9441175244441737,
      "grad_norm": 0.17337866127490997,
      "learning_rate": 2.3757861927203308e-08,
      "loss": 0.0009,
      "step": 303120
    },
    {
      "epoch": 1.9441816613379599,
      "grad_norm": 0.0266607403755188,
      "learning_rate": 2.3703395792374352e-08,
      "loss": 0.0008,
      "step": 303130
    },
    {
      "epoch": 1.9442457982317458,
      "grad_norm": 0.11554598808288574,
      "learning_rate": 2.3648992014098092e-08,
      "loss": 0.0011,
      "step": 303140
    },
    {
      "epoch": 1.944309935125532,
      "grad_norm": 0.059785395860672,
      "learning_rate": 2.3594650593056767e-08,
      "loss": 0.0012,
      "step": 303150
    },
    {
      "epoch": 1.944374072019318,
      "grad_norm": 0.027033396065235138,
      "learning_rate": 2.354037152993094e-08,
      "loss": 0.0005,
      "step": 303160
    },
    {
      "epoch": 1.944438208913104,
      "grad_norm": 0.04057193920016289,
      "learning_rate": 2.348615482540062e-08,
      "loss": 0.0023,
      "step": 303170
    },
    {
      "epoch": 1.9445023458068902,
      "grad_norm": 0.05944840610027313,
      "learning_rate": 2.3432000480145822e-08,
      "loss": 0.0007,
      "step": 303180
    },
    {
      "epoch": 1.9445664827006763,
      "grad_norm": 0.026654450222849846,
      "learning_rate": 2.3377908494844337e-08,
      "loss": 0.0016,
      "step": 303190
    },
    {
      "epoch": 1.9446306195944625,
      "grad_norm": 0.05117101967334747,
      "learning_rate": 2.3323878870175064e-08,
      "loss": 0.0006,
      "step": 303200
    },
    {
      "epoch": 1.9446947564882486,
      "grad_norm": 0.07699143141508102,
      "learning_rate": 2.3269911606813577e-08,
      "loss": 0.0009,
      "step": 303210
    },
    {
      "epoch": 1.9447588933820348,
      "grad_norm": 0.07043943554162979,
      "learning_rate": 2.321600670543711e-08,
      "loss": 0.0011,
      "step": 303220
    },
    {
      "epoch": 1.9448230302758207,
      "grad_norm": 0.11004768311977386,
      "learning_rate": 2.3162164166721235e-08,
      "loss": 0.0009,
      "step": 303230
    },
    {
      "epoch": 1.9448871671696069,
      "grad_norm": 0.02211976796388626,
      "learning_rate": 2.310838399134041e-08,
      "loss": 0.0015,
      "step": 303240
    },
    {
      "epoch": 1.9449513040633928,
      "grad_norm": 0.11930811405181885,
      "learning_rate": 2.3054666179968544e-08,
      "loss": 0.0007,
      "step": 303250
    },
    {
      "epoch": 1.945015440957179,
      "grad_norm": 0.0007218879763968289,
      "learning_rate": 2.3001010733277873e-08,
      "loss": 0.0005,
      "step": 303260
    },
    {
      "epoch": 1.945079577850965,
      "grad_norm": 0.020383605733513832,
      "learning_rate": 2.2947417651942305e-08,
      "loss": 0.0007,
      "step": 303270
    },
    {
      "epoch": 1.9451437147447512,
      "grad_norm": 0.03223626688122749,
      "learning_rate": 2.2893886936632413e-08,
      "loss": 0.0028,
      "step": 303280
    },
    {
      "epoch": 1.9452078516385374,
      "grad_norm": 0.08068500459194183,
      "learning_rate": 2.284041858801933e-08,
      "loss": 0.0011,
      "step": 303290
    },
    {
      "epoch": 1.9452719885323235,
      "grad_norm": 0.06491301208734512,
      "learning_rate": 2.278701260677252e-08,
      "loss": 0.0007,
      "step": 303300
    },
    {
      "epoch": 1.9453361254261095,
      "grad_norm": 0.07721704989671707,
      "learning_rate": 2.2733668993561998e-08,
      "loss": 0.0011,
      "step": 303310
    },
    {
      "epoch": 1.9454002623198956,
      "grad_norm": 0.03412219136953354,
      "learning_rate": 2.268038774905612e-08,
      "loss": 0.0021,
      "step": 303320
    },
    {
      "epoch": 1.9454643992136815,
      "grad_norm": 0.17845505475997925,
      "learning_rate": 2.262716887392158e-08,
      "loss": 0.0007,
      "step": 303330
    },
    {
      "epoch": 1.9455285361074677,
      "grad_norm": 0.040481481701135635,
      "learning_rate": 2.2574012368825616e-08,
      "loss": 0.0011,
      "step": 303340
    },
    {
      "epoch": 1.9455926730012538,
      "grad_norm": 0.10173839330673218,
      "learning_rate": 2.2520918234435473e-08,
      "loss": 0.0014,
      "step": 303350
    },
    {
      "epoch": 1.94565680989504,
      "grad_norm": 0.003593247616663575,
      "learning_rate": 2.2467886471415068e-08,
      "loss": 0.0018,
      "step": 303360
    },
    {
      "epoch": 1.9457209467888261,
      "grad_norm": 0.017439965158700943,
      "learning_rate": 2.241491708042942e-08,
      "loss": 0.0011,
      "step": 303370
    },
    {
      "epoch": 1.9457850836826123,
      "grad_norm": 0.12718206644058228,
      "learning_rate": 2.2362010062142448e-08,
      "loss": 0.0011,
      "step": 303380
    },
    {
      "epoch": 1.9458492205763984,
      "grad_norm": 0.0425318107008934,
      "learning_rate": 2.230916541721695e-08,
      "loss": 0.0003,
      "step": 303390
    },
    {
      "epoch": 1.9459133574701843,
      "grad_norm": 0.03548678010702133,
      "learning_rate": 2.2256383146315175e-08,
      "loss": 0.0006,
      "step": 303400
    },
    {
      "epoch": 1.9459774943639705,
      "grad_norm": 0.0281289741396904,
      "learning_rate": 2.2203663250098263e-08,
      "loss": 0.0015,
      "step": 303410
    },
    {
      "epoch": 1.9460416312577564,
      "grad_norm": 0.00972016528248787,
      "learning_rate": 2.2151005729226794e-08,
      "loss": 0.0009,
      "step": 303420
    },
    {
      "epoch": 1.9461057681515426,
      "grad_norm": 0.3040933907032013,
      "learning_rate": 2.2098410584361352e-08,
      "loss": 0.0013,
      "step": 303430
    },
    {
      "epoch": 1.9461699050453287,
      "grad_norm": 0.00990355759859085,
      "learning_rate": 2.20458778161603e-08,
      "loss": 0.001,
      "step": 303440
    },
    {
      "epoch": 1.9462340419391149,
      "grad_norm": 0.06964608281850815,
      "learning_rate": 2.1993407425282e-08,
      "loss": 0.0013,
      "step": 303450
    },
    {
      "epoch": 1.946298178832901,
      "grad_norm": 0.013352076523005962,
      "learning_rate": 2.1940999412384258e-08,
      "loss": 0.0007,
      "step": 303460
    },
    {
      "epoch": 1.9463623157266872,
      "grad_norm": 0.09254525601863861,
      "learning_rate": 2.1888653778123214e-08,
      "loss": 0.001,
      "step": 303470
    },
    {
      "epoch": 1.946426452620473,
      "grad_norm": 0.04405030235648155,
      "learning_rate": 2.183637052315557e-08,
      "loss": 0.0009,
      "step": 303480
    },
    {
      "epoch": 1.9464905895142592,
      "grad_norm": 0.014495084062218666,
      "learning_rate": 2.17841496481358e-08,
      "loss": 0.0021,
      "step": 303490
    },
    {
      "epoch": 1.9465547264080452,
      "grad_norm": 0.02775007300078869,
      "learning_rate": 2.173199115371838e-08,
      "loss": 0.0008,
      "step": 303500
    },
    {
      "epoch": 1.9466188633018313,
      "grad_norm": 0.06201065331697464,
      "learning_rate": 2.167989504055723e-08,
      "loss": 0.0005,
      "step": 303510
    },
    {
      "epoch": 1.9466830001956175,
      "grad_norm": 0.032466646283864975,
      "learning_rate": 2.1627861309305164e-08,
      "loss": 0.0016,
      "step": 303520
    },
    {
      "epoch": 1.9467471370894036,
      "grad_norm": 0.10472922772169113,
      "learning_rate": 2.1575889960613327e-08,
      "loss": 0.0011,
      "step": 303530
    },
    {
      "epoch": 1.9468112739831898,
      "grad_norm": 0.21284520626068115,
      "learning_rate": 2.1523980995133976e-08,
      "loss": 0.0008,
      "step": 303540
    },
    {
      "epoch": 1.946875410876976,
      "grad_norm": 0.06618592888116837,
      "learning_rate": 2.1472134413517698e-08,
      "loss": 0.0013,
      "step": 303550
    },
    {
      "epoch": 1.946939547770762,
      "grad_norm": 0.050356101244688034,
      "learning_rate": 2.1420350216412866e-08,
      "loss": 0.0008,
      "step": 303560
    },
    {
      "epoch": 1.947003684664548,
      "grad_norm": 0.03412504494190216,
      "learning_rate": 2.1368628404469517e-08,
      "loss": 0.0012,
      "step": 303570
    },
    {
      "epoch": 1.9470678215583341,
      "grad_norm": 0.08310101926326752,
      "learning_rate": 2.1316968978335463e-08,
      "loss": 0.0012,
      "step": 303580
    },
    {
      "epoch": 1.94713195845212,
      "grad_norm": 0.006315671838819981,
      "learning_rate": 2.1265371938657965e-08,
      "loss": 0.0009,
      "step": 303590
    },
    {
      "epoch": 1.9471960953459062,
      "grad_norm": 0.14996036887168884,
      "learning_rate": 2.121383728608373e-08,
      "loss": 0.0008,
      "step": 303600
    },
    {
      "epoch": 1.9472602322396924,
      "grad_norm": 0.03868402913212776,
      "learning_rate": 2.1162365021258345e-08,
      "loss": 0.0015,
      "step": 303610
    },
    {
      "epoch": 1.9473243691334785,
      "grad_norm": 0.02724253199994564,
      "learning_rate": 2.111095514482686e-08,
      "loss": 0.0007,
      "step": 303620
    },
    {
      "epoch": 1.9473885060272647,
      "grad_norm": 0.08939662575721741,
      "learning_rate": 2.10596076574332e-08,
      "loss": 0.0008,
      "step": 303630
    },
    {
      "epoch": 1.9474526429210508,
      "grad_norm": 0.037689194083213806,
      "learning_rate": 2.1008322559721296e-08,
      "loss": 0.0004,
      "step": 303640
    },
    {
      "epoch": 1.947516779814837,
      "grad_norm": 0.023243412375450134,
      "learning_rate": 2.0957099852333407e-08,
      "loss": 0.0008,
      "step": 303650
    },
    {
      "epoch": 1.9475809167086229,
      "grad_norm": 0.024731189012527466,
      "learning_rate": 2.0905939535911802e-08,
      "loss": 0.0008,
      "step": 303660
    },
    {
      "epoch": 1.947645053602409,
      "grad_norm": 0.014677757397294044,
      "learning_rate": 2.0854841611097078e-08,
      "loss": 0.0007,
      "step": 303670
    },
    {
      "epoch": 1.947709190496195,
      "grad_norm": 0.32674118876457214,
      "learning_rate": 2.0803806078529275e-08,
      "loss": 0.0023,
      "step": 303680
    },
    {
      "epoch": 1.947773327389981,
      "grad_norm": 0.1138594001531601,
      "learning_rate": 2.0752832938849e-08,
      "loss": 0.001,
      "step": 303690
    },
    {
      "epoch": 1.9478374642837673,
      "grad_norm": 0.0010594233172014356,
      "learning_rate": 2.0701922192694067e-08,
      "loss": 0.0012,
      "step": 303700
    },
    {
      "epoch": 1.9479016011775534,
      "grad_norm": 0.010231339372694492,
      "learning_rate": 2.0651073840702862e-08,
      "loss": 0.001,
      "step": 303710
    },
    {
      "epoch": 1.9479657380713395,
      "grad_norm": 0.10267619788646698,
      "learning_rate": 2.0600287883512093e-08,
      "loss": 0.0006,
      "step": 303720
    },
    {
      "epoch": 1.9480298749651257,
      "grad_norm": 0.08820979297161102,
      "learning_rate": 2.054956432175903e-08,
      "loss": 0.0004,
      "step": 303730
    },
    {
      "epoch": 1.9480940118589116,
      "grad_norm": 0.10496065020561218,
      "learning_rate": 2.0498903156078165e-08,
      "loss": 0.0005,
      "step": 303740
    },
    {
      "epoch": 1.9481581487526978,
      "grad_norm": 0.07082303613424301,
      "learning_rate": 2.04483043871051e-08,
      "loss": 0.0012,
      "step": 303750
    },
    {
      "epoch": 1.9482222856464837,
      "grad_norm": 0.03306499868631363,
      "learning_rate": 2.0397768015473218e-08,
      "loss": 0.0009,
      "step": 303760
    },
    {
      "epoch": 1.9482864225402698,
      "grad_norm": 0.0331336110830307,
      "learning_rate": 2.0347294041816456e-08,
      "loss": 0.0005,
      "step": 303770
    },
    {
      "epoch": 1.948350559434056,
      "grad_norm": 0.10384833067655563,
      "learning_rate": 2.0296882466767086e-08,
      "loss": 0.0015,
      "step": 303780
    },
    {
      "epoch": 1.9484146963278421,
      "grad_norm": 0.07057362049818039,
      "learning_rate": 2.0246533290956826e-08,
      "loss": 0.0013,
      "step": 303790
    },
    {
      "epoch": 1.9484788332216283,
      "grad_norm": 0.05630839988589287,
      "learning_rate": 2.019624651501628e-08,
      "loss": 0.001,
      "step": 303800
    },
    {
      "epoch": 1.9485429701154144,
      "grad_norm": 0.055216625332832336,
      "learning_rate": 2.0146022139575506e-08,
      "loss": 0.0006,
      "step": 303810
    },
    {
      "epoch": 1.9486071070092006,
      "grad_norm": 0.05376419052481651,
      "learning_rate": 2.0095860165264547e-08,
      "loss": 0.0017,
      "step": 303820
    },
    {
      "epoch": 1.9486712439029865,
      "grad_norm": 0.02756691165268421,
      "learning_rate": 2.0045760592711238e-08,
      "loss": 0.0009,
      "step": 303830
    },
    {
      "epoch": 1.9487353807967727,
      "grad_norm": 0.056577298790216446,
      "learning_rate": 1.9995723422543968e-08,
      "loss": 0.0008,
      "step": 303840
    },
    {
      "epoch": 1.9487995176905586,
      "grad_norm": 0.0758282020688057,
      "learning_rate": 1.994574865538945e-08,
      "loss": 0.0013,
      "step": 303850
    },
    {
      "epoch": 1.9488636545843447,
      "grad_norm": 0.02690417319536209,
      "learning_rate": 1.989583629187386e-08,
      "loss": 0.0011,
      "step": 303860
    },
    {
      "epoch": 1.9489277914781309,
      "grad_norm": 0.012038026005029678,
      "learning_rate": 1.9845986332622248e-08,
      "loss": 0.0015,
      "step": 303870
    },
    {
      "epoch": 1.948991928371917,
      "grad_norm": 0.06118635833263397,
      "learning_rate": 1.9796198778260222e-08,
      "loss": 0.0006,
      "step": 303880
    },
    {
      "epoch": 1.9490560652657032,
      "grad_norm": 0.10096149891614914,
      "learning_rate": 1.9746473629410624e-08,
      "loss": 0.0007,
      "step": 303890
    },
    {
      "epoch": 1.9491202021594893,
      "grad_norm": 0.13723134994506836,
      "learning_rate": 1.9696810886697392e-08,
      "loss": 0.0019,
      "step": 303900
    },
    {
      "epoch": 1.9491843390532753,
      "grad_norm": 0.05517492815852165,
      "learning_rate": 1.9647210550742256e-08,
      "loss": 0.0006,
      "step": 303910
    },
    {
      "epoch": 1.9492484759470614,
      "grad_norm": 0.03260452672839165,
      "learning_rate": 1.9597672622167497e-08,
      "loss": 0.0012,
      "step": 303920
    },
    {
      "epoch": 1.9493126128408476,
      "grad_norm": 0.05614691600203514,
      "learning_rate": 1.954819710159317e-08,
      "loss": 0.0007,
      "step": 303930
    },
    {
      "epoch": 1.9493767497346335,
      "grad_norm": 0.03515557199716568,
      "learning_rate": 1.949878398963878e-08,
      "loss": 0.0006,
      "step": 303940
    },
    {
      "epoch": 1.9494408866284196,
      "grad_norm": 0.040092095732688904,
      "learning_rate": 1.9449433286924946e-08,
      "loss": 0.0027,
      "step": 303950
    },
    {
      "epoch": 1.9495050235222058,
      "grad_norm": 0.03886708989739418,
      "learning_rate": 1.9400144994068948e-08,
      "loss": 0.0009,
      "step": 303960
    },
    {
      "epoch": 1.949569160415992,
      "grad_norm": 0.13820737600326538,
      "learning_rate": 1.9350919111688626e-08,
      "loss": 0.0005,
      "step": 303970
    },
    {
      "epoch": 1.949633297309778,
      "grad_norm": 0.004038135055452585,
      "learning_rate": 1.930175564040071e-08,
      "loss": 0.0008,
      "step": 303980
    },
    {
      "epoch": 1.9496974342035642,
      "grad_norm": 0.006674329750239849,
      "learning_rate": 1.9252654580821374e-08,
      "loss": 0.001,
      "step": 303990
    },
    {
      "epoch": 1.9497615710973502,
      "grad_norm": 0.01439402624964714,
      "learning_rate": 1.9203615933566787e-08,
      "loss": 0.0009,
      "step": 304000
    },
    {
      "epoch": 1.9498257079911363,
      "grad_norm": 0.004403538070619106,
      "learning_rate": 1.91546396992498e-08,
      "loss": 0.0009,
      "step": 304010
    },
    {
      "epoch": 1.9498898448849222,
      "grad_norm": 0.05232447758316994,
      "learning_rate": 1.9105725878485472e-08,
      "loss": 0.0019,
      "step": 304020
    },
    {
      "epoch": 1.9499539817787084,
      "grad_norm": 0.05275091156363487,
      "learning_rate": 1.9056874471885535e-08,
      "loss": 0.0007,
      "step": 304030
    },
    {
      "epoch": 1.9500181186724945,
      "grad_norm": 0.03211916610598564,
      "learning_rate": 1.900808548006339e-08,
      "loss": 0.0009,
      "step": 304040
    },
    {
      "epoch": 1.9500822555662807,
      "grad_norm": 0.07366874814033508,
      "learning_rate": 1.8959358903629653e-08,
      "loss": 0.0013,
      "step": 304050
    },
    {
      "epoch": 1.9501463924600668,
      "grad_norm": 0.0299956277012825,
      "learning_rate": 1.8910694743194958e-08,
      "loss": 0.0003,
      "step": 304060
    },
    {
      "epoch": 1.950210529353853,
      "grad_norm": 0.034520067274570465,
      "learning_rate": 1.8862092999369364e-08,
      "loss": 0.0007,
      "step": 304070
    },
    {
      "epoch": 1.9502746662476391,
      "grad_norm": 0.1465241014957428,
      "learning_rate": 1.8813553672761274e-08,
      "loss": 0.0008,
      "step": 304080
    },
    {
      "epoch": 1.950338803141425,
      "grad_norm": 0.15555109083652496,
      "learning_rate": 1.876507676397965e-08,
      "loss": 0.001,
      "step": 304090
    },
    {
      "epoch": 1.9504029400352112,
      "grad_norm": 0.05211598053574562,
      "learning_rate": 1.8716662273631224e-08,
      "loss": 0.0012,
      "step": 304100
    },
    {
      "epoch": 1.9504670769289971,
      "grad_norm": 0.03298133611679077,
      "learning_rate": 1.8668310202323847e-08,
      "loss": 0.0009,
      "step": 304110
    },
    {
      "epoch": 1.9505312138227833,
      "grad_norm": 0.009433920495212078,
      "learning_rate": 1.8620020550662588e-08,
      "loss": 0.0022,
      "step": 304120
    },
    {
      "epoch": 1.9505953507165694,
      "grad_norm": 0.1759236752986908,
      "learning_rate": 1.857179331925196e-08,
      "loss": 0.0014,
      "step": 304130
    },
    {
      "epoch": 1.9506594876103556,
      "grad_norm": 0.09502455592155457,
      "learning_rate": 1.85236285086976e-08,
      "loss": 0.0007,
      "step": 304140
    },
    {
      "epoch": 1.9507236245041417,
      "grad_norm": 0.10114434361457825,
      "learning_rate": 1.8475526119601794e-08,
      "loss": 0.0013,
      "step": 304150
    },
    {
      "epoch": 1.9507877613979279,
      "grad_norm": 0.10108424723148346,
      "learning_rate": 1.8427486152568507e-08,
      "loss": 0.0012,
      "step": 304160
    },
    {
      "epoch": 1.9508518982917138,
      "grad_norm": 0.13766522705554962,
      "learning_rate": 1.837950860819837e-08,
      "loss": 0.0011,
      "step": 304170
    },
    {
      "epoch": 1.9509160351855,
      "grad_norm": 0.10113713890314102,
      "learning_rate": 1.833159348709368e-08,
      "loss": 0.0008,
      "step": 304180
    },
    {
      "epoch": 1.9509801720792859,
      "grad_norm": 0.09011459350585938,
      "learning_rate": 1.828374078985451e-08,
      "loss": 0.0007,
      "step": 304190
    },
    {
      "epoch": 1.951044308973072,
      "grad_norm": 0.028076674789190292,
      "learning_rate": 1.823595051708038e-08,
      "loss": 0.0009,
      "step": 304200
    },
    {
      "epoch": 1.9511084458668582,
      "grad_norm": 0.11939436197280884,
      "learning_rate": 1.818822266937026e-08,
      "loss": 0.001,
      "step": 304210
    },
    {
      "epoch": 1.9511725827606443,
      "grad_norm": 0.08247743546962738,
      "learning_rate": 1.8140557247322554e-08,
      "loss": 0.0005,
      "step": 304220
    },
    {
      "epoch": 1.9512367196544305,
      "grad_norm": 0.043421365320682526,
      "learning_rate": 1.8092954251533455e-08,
      "loss": 0.0007,
      "step": 304230
    },
    {
      "epoch": 1.9513008565482166,
      "grad_norm": 0.18763089179992676,
      "learning_rate": 1.804541368260082e-08,
      "loss": 0.0012,
      "step": 304240
    },
    {
      "epoch": 1.9513649934420028,
      "grad_norm": 0.14211469888687134,
      "learning_rate": 1.7997935541119172e-08,
      "loss": 0.0009,
      "step": 304250
    },
    {
      "epoch": 1.9514291303357887,
      "grad_norm": 0.0641964003443718,
      "learning_rate": 1.7950519827684142e-08,
      "loss": 0.001,
      "step": 304260
    },
    {
      "epoch": 1.9514932672295748,
      "grad_norm": 0.091981440782547,
      "learning_rate": 1.7903166542889705e-08,
      "loss": 0.0025,
      "step": 304270
    },
    {
      "epoch": 1.9515574041233608,
      "grad_norm": 0.032483477145433426,
      "learning_rate": 1.785587568732927e-08,
      "loss": 0.0009,
      "step": 304280
    },
    {
      "epoch": 1.951621541017147,
      "grad_norm": 0.3191652297973633,
      "learning_rate": 1.7808647261595148e-08,
      "loss": 0.001,
      "step": 304290
    },
    {
      "epoch": 1.951685677910933,
      "grad_norm": 0.04146173223853111,
      "learning_rate": 1.7761481266279634e-08,
      "loss": 0.0007,
      "step": 304300
    },
    {
      "epoch": 1.9517498148047192,
      "grad_norm": 0.003173446049913764,
      "learning_rate": 1.7714377701973374e-08,
      "loss": 0.0009,
      "step": 304310
    },
    {
      "epoch": 1.9518139516985054,
      "grad_norm": 0.07733031362295151,
      "learning_rate": 1.7667336569267e-08,
      "loss": 0.0008,
      "step": 304320
    },
    {
      "epoch": 1.9518780885922915,
      "grad_norm": 0.06869202107191086,
      "learning_rate": 1.7620357868749495e-08,
      "loss": 0.0009,
      "step": 304330
    },
    {
      "epoch": 1.9519422254860774,
      "grad_norm": 0.12415417283773422,
      "learning_rate": 1.7573441601009822e-08,
      "loss": 0.0015,
      "step": 304340
    },
    {
      "epoch": 1.9520063623798636,
      "grad_norm": 0.06551288068294525,
      "learning_rate": 1.7526587766635848e-08,
      "loss": 0.0012,
      "step": 304350
    },
    {
      "epoch": 1.9520704992736497,
      "grad_norm": 0.09301673620939255,
      "learning_rate": 1.747979636621433e-08,
      "loss": 0.0006,
      "step": 304360
    },
    {
      "epoch": 1.9521346361674357,
      "grad_norm": 0.0703938826918602,
      "learning_rate": 1.743306740033257e-08,
      "loss": 0.0011,
      "step": 304370
    },
    {
      "epoch": 1.9521987730612218,
      "grad_norm": 0.285512775182724,
      "learning_rate": 1.7386400869575104e-08,
      "loss": 0.0022,
      "step": 304380
    },
    {
      "epoch": 1.952262909955008,
      "grad_norm": 0.025058260187506676,
      "learning_rate": 1.733979677452702e-08,
      "loss": 0.0008,
      "step": 304390
    },
    {
      "epoch": 1.952327046848794,
      "grad_norm": 0.04474668204784393,
      "learning_rate": 1.7293255115772288e-08,
      "loss": 0.0005,
      "step": 304400
    },
    {
      "epoch": 1.9523911837425802,
      "grad_norm": 0.03960119187831879,
      "learning_rate": 1.724677589389434e-08,
      "loss": 0.0035,
      "step": 304410
    },
    {
      "epoch": 1.9524553206363664,
      "grad_norm": 0.0017269984818995,
      "learning_rate": 1.7200359109475485e-08,
      "loss": 0.0006,
      "step": 304420
    },
    {
      "epoch": 1.9525194575301523,
      "grad_norm": 0.03660395368933678,
      "learning_rate": 1.715400476309692e-08,
      "loss": 0.0017,
      "step": 304430
    },
    {
      "epoch": 1.9525835944239385,
      "grad_norm": 0.042580440640449524,
      "learning_rate": 1.7107712855339853e-08,
      "loss": 0.0009,
      "step": 304440
    },
    {
      "epoch": 1.9526477313177244,
      "grad_norm": 0.10549118369817734,
      "learning_rate": 1.7061483386784928e-08,
      "loss": 0.0005,
      "step": 304450
    },
    {
      "epoch": 1.9527118682115105,
      "grad_norm": 0.18237760663032532,
      "learning_rate": 1.701531635801057e-08,
      "loss": 0.0017,
      "step": 304460
    },
    {
      "epoch": 1.9527760051052967,
      "grad_norm": 0.01771414466202259,
      "learning_rate": 1.6969211769595206e-08,
      "loss": 0.0022,
      "step": 304470
    },
    {
      "epoch": 1.9528401419990828,
      "grad_norm": 0.10704022645950317,
      "learning_rate": 1.6923169622117264e-08,
      "loss": 0.0008,
      "step": 304480
    },
    {
      "epoch": 1.952904278892869,
      "grad_norm": 0.11638756841421127,
      "learning_rate": 1.6877189916153503e-08,
      "loss": 0.0016,
      "step": 304490
    },
    {
      "epoch": 1.9529684157866551,
      "grad_norm": 0.037044230848550797,
      "learning_rate": 1.6831272652279573e-08,
      "loss": 0.0009,
      "step": 304500
    },
    {
      "epoch": 1.9530325526804413,
      "grad_norm": 0.043244726955890656,
      "learning_rate": 1.6785417831071682e-08,
      "loss": 0.0006,
      "step": 304510
    },
    {
      "epoch": 1.9530966895742272,
      "grad_norm": 0.14400893449783325,
      "learning_rate": 1.6739625453103258e-08,
      "loss": 0.0015,
      "step": 304520
    },
    {
      "epoch": 1.9531608264680134,
      "grad_norm": 0.02449709363281727,
      "learning_rate": 1.66938955189494e-08,
      "loss": 0.0006,
      "step": 304530
    },
    {
      "epoch": 1.9532249633617993,
      "grad_norm": 0.05143209546804428,
      "learning_rate": 1.6648228029182424e-08,
      "loss": 0.0009,
      "step": 304540
    },
    {
      "epoch": 1.9532891002555854,
      "grad_norm": 0.01187801267951727,
      "learning_rate": 1.660262298437465e-08,
      "loss": 0.0007,
      "step": 304550
    },
    {
      "epoch": 1.9533532371493716,
      "grad_norm": 0.14507299661636353,
      "learning_rate": 1.6557080385097845e-08,
      "loss": 0.0011,
      "step": 304560
    },
    {
      "epoch": 1.9534173740431577,
      "grad_norm": 0.20421335101127625,
      "learning_rate": 1.651160023192211e-08,
      "loss": 0.0005,
      "step": 304570
    },
    {
      "epoch": 1.9534815109369439,
      "grad_norm": 0.020057285204529762,
      "learning_rate": 1.646618252541754e-08,
      "loss": 0.0058,
      "step": 304580
    },
    {
      "epoch": 1.95354564783073,
      "grad_norm": 0.12206927686929703,
      "learning_rate": 1.6420827266153683e-08,
      "loss": 0.0009,
      "step": 304590
    },
    {
      "epoch": 1.953609784724516,
      "grad_norm": 0.1060272827744484,
      "learning_rate": 1.6375534454698417e-08,
      "loss": 0.0011,
      "step": 304600
    },
    {
      "epoch": 1.9536739216183021,
      "grad_norm": 0.04592036083340645,
      "learning_rate": 1.633030409161962e-08,
      "loss": 0.0015,
      "step": 304610
    },
    {
      "epoch": 1.953738058512088,
      "grad_norm": 0.10995520651340485,
      "learning_rate": 1.6285136177483505e-08,
      "loss": 0.0007,
      "step": 304620
    },
    {
      "epoch": 1.9538021954058742,
      "grad_norm": 0.31133806705474854,
      "learning_rate": 1.624003071285685e-08,
      "loss": 0.0016,
      "step": 304630
    },
    {
      "epoch": 1.9538663322996603,
      "grad_norm": 0.19678683578968048,
      "learning_rate": 1.619498769830474e-08,
      "loss": 0.0012,
      "step": 304640
    },
    {
      "epoch": 1.9539304691934465,
      "grad_norm": 0.03897224739193916,
      "learning_rate": 1.6150007134390632e-08,
      "loss": 0.0019,
      "step": 304650
    },
    {
      "epoch": 1.9539946060872326,
      "grad_norm": 0.05141619220376015,
      "learning_rate": 1.6105089021679067e-08,
      "loss": 0.0021,
      "step": 304660
    },
    {
      "epoch": 1.9540587429810188,
      "grad_norm": 0.024906743317842484,
      "learning_rate": 1.6060233360732924e-08,
      "loss": 0.0011,
      "step": 304670
    },
    {
      "epoch": 1.954122879874805,
      "grad_norm": 0.05543055385351181,
      "learning_rate": 1.6015440152113425e-08,
      "loss": 0.0007,
      "step": 304680
    },
    {
      "epoch": 1.9541870167685909,
      "grad_norm": 0.010509229265153408,
      "learning_rate": 1.59707093963829e-08,
      "loss": 0.0009,
      "step": 304690
    },
    {
      "epoch": 1.954251153662377,
      "grad_norm": 0.15427754819393158,
      "learning_rate": 1.5926041094101452e-08,
      "loss": 0.0021,
      "step": 304700
    },
    {
      "epoch": 1.954315290556163,
      "grad_norm": 0.1171812117099762,
      "learning_rate": 1.5881435245828636e-08,
      "loss": 0.0015,
      "step": 304710
    },
    {
      "epoch": 1.954379427449949,
      "grad_norm": 0.03251752257347107,
      "learning_rate": 1.583689185212345e-08,
      "loss": 0.0012,
      "step": 304720
    },
    {
      "epoch": 1.9544435643437352,
      "grad_norm": 0.05555243790149689,
      "learning_rate": 1.5792410913544332e-08,
      "loss": 0.0012,
      "step": 304730
    },
    {
      "epoch": 1.9545077012375214,
      "grad_norm": 0.3689599931240082,
      "learning_rate": 1.5747992430648617e-08,
      "loss": 0.0018,
      "step": 304740
    },
    {
      "epoch": 1.9545718381313075,
      "grad_norm": 0.044546645134687424,
      "learning_rate": 1.570363640399253e-08,
      "loss": 0.0012,
      "step": 304750
    },
    {
      "epoch": 1.9546359750250937,
      "grad_norm": 0.025621255859732628,
      "learning_rate": 1.565934283413173e-08,
      "loss": 0.001,
      "step": 304760
    },
    {
      "epoch": 1.9547001119188796,
      "grad_norm": 0.18231767416000366,
      "learning_rate": 1.5615111721621888e-08,
      "loss": 0.0012,
      "step": 304770
    },
    {
      "epoch": 1.9547642488126658,
      "grad_norm": 0.04967685416340828,
      "learning_rate": 1.5570943067017564e-08,
      "loss": 0.0013,
      "step": 304780
    },
    {
      "epoch": 1.954828385706452,
      "grad_norm": 0.026820089668035507,
      "learning_rate": 1.5526836870871086e-08,
      "loss": 0.0014,
      "step": 304790
    },
    {
      "epoch": 1.9548925226002378,
      "grad_norm": 0.11177382618188858,
      "learning_rate": 1.548279313373591e-08,
      "loss": 0.0011,
      "step": 304800
    },
    {
      "epoch": 1.954956659494024,
      "grad_norm": 0.03823992982506752,
      "learning_rate": 1.543881185616325e-08,
      "loss": 0.0006,
      "step": 304810
    },
    {
      "epoch": 1.9550207963878101,
      "grad_norm": 0.2680332362651825,
      "learning_rate": 1.5394893038704895e-08,
      "loss": 0.0015,
      "step": 304820
    },
    {
      "epoch": 1.9550849332815963,
      "grad_norm": 0.0011533283395692706,
      "learning_rate": 1.535103668191096e-08,
      "loss": 0.0009,
      "step": 304830
    },
    {
      "epoch": 1.9551490701753824,
      "grad_norm": 0.07149229943752289,
      "learning_rate": 1.5307242786331556e-08,
      "loss": 0.0006,
      "step": 304840
    },
    {
      "epoch": 1.9552132070691686,
      "grad_norm": 0.07902127504348755,
      "learning_rate": 1.5263511352514026e-08,
      "loss": 0.0011,
      "step": 304850
    },
    {
      "epoch": 1.9552773439629545,
      "grad_norm": 0.065912626683712,
      "learning_rate": 1.5219842381007932e-08,
      "loss": 0.001,
      "step": 304860
    },
    {
      "epoch": 1.9553414808567406,
      "grad_norm": 0.13002678751945496,
      "learning_rate": 1.5176235872359502e-08,
      "loss": 0.0019,
      "step": 304870
    },
    {
      "epoch": 1.9554056177505266,
      "grad_norm": 0.061404649168252945,
      "learning_rate": 1.513269182711552e-08,
      "loss": 0.0014,
      "step": 304880
    },
    {
      "epoch": 1.9554697546443127,
      "grad_norm": 0.027881687507033348,
      "learning_rate": 1.5089210245821105e-08,
      "loss": 0.0034,
      "step": 304890
    },
    {
      "epoch": 1.9555338915380989,
      "grad_norm": 0.02497880719602108,
      "learning_rate": 1.5045791129021935e-08,
      "loss": 0.0008,
      "step": 304900
    },
    {
      "epoch": 1.955598028431885,
      "grad_norm": 0.009090565145015717,
      "learning_rate": 1.5002434477262018e-08,
      "loss": 0.001,
      "step": 304910
    },
    {
      "epoch": 1.9556621653256712,
      "grad_norm": 0.07809104025363922,
      "learning_rate": 1.4959140291083696e-08,
      "loss": 0.0012,
      "step": 304920
    },
    {
      "epoch": 1.9557263022194573,
      "grad_norm": 0.11369671672582626,
      "learning_rate": 1.4915908571030424e-08,
      "loss": 0.0009,
      "step": 304930
    },
    {
      "epoch": 1.9557904391132435,
      "grad_norm": 0.12540864944458008,
      "learning_rate": 1.4872739317643992e-08,
      "loss": 0.0014,
      "step": 304940
    },
    {
      "epoch": 1.9558545760070294,
      "grad_norm": 0.0958738848567009,
      "learning_rate": 1.4829632531464523e-08,
      "loss": 0.0004,
      "step": 304950
    },
    {
      "epoch": 1.9559187129008155,
      "grad_norm": 0.07722891122102737,
      "learning_rate": 1.4786588213032694e-08,
      "loss": 0.0009,
      "step": 304960
    },
    {
      "epoch": 1.9559828497946015,
      "grad_norm": 0.06726028025150299,
      "learning_rate": 1.474360636288752e-08,
      "loss": 0.0026,
      "step": 304970
    },
    {
      "epoch": 1.9560469866883876,
      "grad_norm": 0.03700147196650505,
      "learning_rate": 1.4700686981568014e-08,
      "loss": 0.0008,
      "step": 304980
    },
    {
      "epoch": 1.9561111235821738,
      "grad_norm": 0.14430172741413116,
      "learning_rate": 1.4657830069612078e-08,
      "loss": 0.0015,
      "step": 304990
    },
    {
      "epoch": 1.95617526047596,
      "grad_norm": 0.09119134396314621,
      "learning_rate": 1.4615035627556507e-08,
      "loss": 0.0011,
      "step": 305000
    },
    {
      "epoch": 1.95617526047596,
      "eval_loss": 0.0019545629620552063,
      "eval_runtime": 3.3198,
      "eval_samples_per_second": 60.244,
      "eval_steps_per_second": 15.061,
      "step": 305000
    },
    {
      "epoch": 1.956239397369746,
      "grad_norm": 0.010408638045191765,
      "learning_rate": 1.4572303655936981e-08,
      "loss": 0.0013,
      "step": 305010
    },
    {
      "epoch": 1.9563035342635322,
      "grad_norm": 0.05148407071828842,
      "learning_rate": 1.452963415528974e-08,
      "loss": 0.0008,
      "step": 305020
    },
    {
      "epoch": 1.9563676711573181,
      "grad_norm": 0.025471650063991547,
      "learning_rate": 1.4487027126149356e-08,
      "loss": 0.001,
      "step": 305030
    },
    {
      "epoch": 1.9564318080511043,
      "grad_norm": 0.14106789231300354,
      "learning_rate": 1.4444482569049845e-08,
      "loss": 0.0006,
      "step": 305040
    },
    {
      "epoch": 1.9564959449448902,
      "grad_norm": 0.08223976194858551,
      "learning_rate": 1.4402000484524115e-08,
      "loss": 0.0017,
      "step": 305050
    },
    {
      "epoch": 1.9565600818386764,
      "grad_norm": 0.06003883108496666,
      "learning_rate": 1.4359580873103962e-08,
      "loss": 0.0009,
      "step": 305060
    },
    {
      "epoch": 1.9566242187324625,
      "grad_norm": 0.02777407504618168,
      "learning_rate": 1.4317223735321739e-08,
      "loss": 0.0011,
      "step": 305070
    },
    {
      "epoch": 1.9566883556262487,
      "grad_norm": 0.09857097268104553,
      "learning_rate": 1.4274929071708133e-08,
      "loss": 0.0009,
      "step": 305080
    },
    {
      "epoch": 1.9567524925200348,
      "grad_norm": 0.11872630566358566,
      "learning_rate": 1.4232696882792718e-08,
      "loss": 0.0009,
      "step": 305090
    },
    {
      "epoch": 1.956816629413821,
      "grad_norm": 0.049037232995033264,
      "learning_rate": 1.4190527169105073e-08,
      "loss": 0.002,
      "step": 305100
    },
    {
      "epoch": 1.956880766307607,
      "grad_norm": 0.05172473192214966,
      "learning_rate": 1.4148419931173107e-08,
      "loss": 0.0015,
      "step": 305110
    },
    {
      "epoch": 1.956944903201393,
      "grad_norm": 0.11926617473363876,
      "learning_rate": 1.4106375169524733e-08,
      "loss": 0.0013,
      "step": 305120
    },
    {
      "epoch": 1.9570090400951792,
      "grad_norm": 0.0758260115981102,
      "learning_rate": 1.406439288468675e-08,
      "loss": 0.0009,
      "step": 305130
    },
    {
      "epoch": 1.957073176988965,
      "grad_norm": 0.023133208975195885,
      "learning_rate": 1.402247307718596e-08,
      "loss": 0.0003,
      "step": 305140
    },
    {
      "epoch": 1.9571373138827513,
      "grad_norm": 0.1966695487499237,
      "learning_rate": 1.3980615747546389e-08,
      "loss": 0.0015,
      "step": 305150
    },
    {
      "epoch": 1.9572014507765374,
      "grad_norm": 0.05596424639225006,
      "learning_rate": 1.3938820896293171e-08,
      "loss": 0.0006,
      "step": 305160
    },
    {
      "epoch": 1.9572655876703235,
      "grad_norm": 0.029670780524611473,
      "learning_rate": 1.389708852395033e-08,
      "loss": 0.001,
      "step": 305170
    },
    {
      "epoch": 1.9573297245641097,
      "grad_norm": 0.034810930490493774,
      "learning_rate": 1.3855418631040785e-08,
      "loss": 0.0008,
      "step": 305180
    },
    {
      "epoch": 1.9573938614578958,
      "grad_norm": 0.16377195715904236,
      "learning_rate": 1.3813811218085781e-08,
      "loss": 0.0017,
      "step": 305190
    },
    {
      "epoch": 1.957457998351682,
      "grad_norm": 0.11208045482635498,
      "learning_rate": 1.3772266285607128e-08,
      "loss": 0.0011,
      "step": 305200
    },
    {
      "epoch": 1.957522135245468,
      "grad_norm": 0.031090980395674706,
      "learning_rate": 1.3730783834126072e-08,
      "loss": 0.0006,
      "step": 305210
    },
    {
      "epoch": 1.957586272139254,
      "grad_norm": 0.04496197775006294,
      "learning_rate": 1.3689363864162198e-08,
      "loss": 0.0011,
      "step": 305220
    },
    {
      "epoch": 1.95765040903304,
      "grad_norm": 0.06327894330024719,
      "learning_rate": 1.3648006376233425e-08,
      "loss": 0.0006,
      "step": 305230
    },
    {
      "epoch": 1.9577145459268261,
      "grad_norm": 0.023319434374570847,
      "learning_rate": 1.3606711370859338e-08,
      "loss": 0.0008,
      "step": 305240
    },
    {
      "epoch": 1.9577786828206123,
      "grad_norm": 0.14921852946281433,
      "learning_rate": 1.3565478848556746e-08,
      "loss": 0.0012,
      "step": 305250
    },
    {
      "epoch": 1.9578428197143984,
      "grad_norm": 0.08323876559734344,
      "learning_rate": 1.3524308809843013e-08,
      "loss": 0.0011,
      "step": 305260
    },
    {
      "epoch": 1.9579069566081846,
      "grad_norm": 0.08927945047616959,
      "learning_rate": 1.3483201255232726e-08,
      "loss": 0.003,
      "step": 305270
    },
    {
      "epoch": 1.9579710935019707,
      "grad_norm": 0.043267734348773956,
      "learning_rate": 1.3442156185242139e-08,
      "loss": 0.0006,
      "step": 305280
    },
    {
      "epoch": 1.9580352303957567,
      "grad_norm": 0.019233791157603264,
      "learning_rate": 1.3401173600384731e-08,
      "loss": 0.0008,
      "step": 305290
    },
    {
      "epoch": 1.9580993672895428,
      "grad_norm": 0.10895214974880219,
      "learning_rate": 1.3360253501175091e-08,
      "loss": 0.0013,
      "step": 305300
    },
    {
      "epoch": 1.9581635041833287,
      "grad_norm": 0.2412586361169815,
      "learning_rate": 1.3319395888124476e-08,
      "loss": 0.0009,
      "step": 305310
    },
    {
      "epoch": 1.9582276410771149,
      "grad_norm": 0.20381537079811096,
      "learning_rate": 1.3278600761746364e-08,
      "loss": 0.0007,
      "step": 305320
    },
    {
      "epoch": 1.958291777970901,
      "grad_norm": 0.037070974707603455,
      "learning_rate": 1.3237868122551457e-08,
      "loss": 0.0008,
      "step": 305330
    },
    {
      "epoch": 1.9583559148646872,
      "grad_norm": 0.15732403099536896,
      "learning_rate": 1.3197197971049902e-08,
      "loss": 0.0012,
      "step": 305340
    },
    {
      "epoch": 1.9584200517584733,
      "grad_norm": 0.022160615772008896,
      "learning_rate": 1.3156590307751293e-08,
      "loss": 0.0011,
      "step": 305350
    },
    {
      "epoch": 1.9584841886522595,
      "grad_norm": 0.013370877131819725,
      "learning_rate": 1.3116045133165222e-08,
      "loss": 0.0015,
      "step": 305360
    },
    {
      "epoch": 1.9585483255460456,
      "grad_norm": 0.00940181128680706,
      "learning_rate": 1.3075562447798506e-08,
      "loss": 0.0006,
      "step": 305370
    },
    {
      "epoch": 1.9586124624398316,
      "grad_norm": 0.07063985615968704,
      "learning_rate": 1.3035142252159627e-08,
      "loss": 0.001,
      "step": 305380
    },
    {
      "epoch": 1.9586765993336177,
      "grad_norm": 0.17283165454864502,
      "learning_rate": 1.299478454675429e-08,
      "loss": 0.0012,
      "step": 305390
    },
    {
      "epoch": 1.9587407362274036,
      "grad_norm": 0.05347483977675438,
      "learning_rate": 1.2954489332088206e-08,
      "loss": 0.0014,
      "step": 305400
    },
    {
      "epoch": 1.9588048731211898,
      "grad_norm": 0.10970115661621094,
      "learning_rate": 1.2914256608667076e-08,
      "loss": 0.0032,
      "step": 305410
    },
    {
      "epoch": 1.958869010014976,
      "grad_norm": 0.02583213895559311,
      "learning_rate": 1.2874086376993833e-08,
      "loss": 0.0008,
      "step": 305420
    },
    {
      "epoch": 1.958933146908762,
      "grad_norm": 0.17533306777477264,
      "learning_rate": 1.2833978637573075e-08,
      "loss": 0.0015,
      "step": 305430
    },
    {
      "epoch": 1.9589972838025482,
      "grad_norm": 0.06510546803474426,
      "learning_rate": 1.2793933390907176e-08,
      "loss": 0.0012,
      "step": 305440
    },
    {
      "epoch": 1.9590614206963344,
      "grad_norm": 0.045575156807899475,
      "learning_rate": 1.2753950637496849e-08,
      "loss": 0.0028,
      "step": 305450
    },
    {
      "epoch": 1.9591255575901203,
      "grad_norm": 0.02801566943526268,
      "learning_rate": 1.2714030377843911e-08,
      "loss": 0.0009,
      "step": 305460
    },
    {
      "epoch": 1.9591896944839065,
      "grad_norm": 0.055962517857551575,
      "learning_rate": 1.2674172612449077e-08,
      "loss": 0.0009,
      "step": 305470
    },
    {
      "epoch": 1.9592538313776926,
      "grad_norm": 0.12682779133319855,
      "learning_rate": 1.2634377341810832e-08,
      "loss": 0.0012,
      "step": 305480
    },
    {
      "epoch": 1.9593179682714785,
      "grad_norm": 0.0686289519071579,
      "learning_rate": 1.2594644566428227e-08,
      "loss": 0.002,
      "step": 305490
    },
    {
      "epoch": 1.9593821051652647,
      "grad_norm": 0.026192547753453255,
      "learning_rate": 1.2554974286799748e-08,
      "loss": 0.0016,
      "step": 305500
    },
    {
      "epoch": 1.9594462420590508,
      "grad_norm": 0.002885625697672367,
      "learning_rate": 1.251536650342111e-08,
      "loss": 0.0009,
      "step": 305510
    },
    {
      "epoch": 1.959510378952837,
      "grad_norm": 0.18336519598960876,
      "learning_rate": 1.2475821216789696e-08,
      "loss": 0.0009,
      "step": 305520
    },
    {
      "epoch": 1.9595745158466231,
      "grad_norm": 0.11568191647529602,
      "learning_rate": 1.2436338427401218e-08,
      "loss": 0.0011,
      "step": 305530
    },
    {
      "epoch": 1.9596386527404093,
      "grad_norm": 0.01961652934551239,
      "learning_rate": 1.2396918135749724e-08,
      "loss": 0.0009,
      "step": 305540
    },
    {
      "epoch": 1.9597027896341952,
      "grad_norm": 0.04887218400835991,
      "learning_rate": 1.2357560342329267e-08,
      "loss": 0.0008,
      "step": 305550
    },
    {
      "epoch": 1.9597669265279813,
      "grad_norm": 0.07078026235103607,
      "learning_rate": 1.2318265047633338e-08,
      "loss": 0.0011,
      "step": 305560
    },
    {
      "epoch": 1.9598310634217673,
      "grad_norm": 0.07141014188528061,
      "learning_rate": 1.2279032252153766e-08,
      "loss": 0.0007,
      "step": 305570
    },
    {
      "epoch": 1.9598952003155534,
      "grad_norm": 0.0644538402557373,
      "learning_rate": 1.2239861956382936e-08,
      "loss": 0.0009,
      "step": 305580
    },
    {
      "epoch": 1.9599593372093396,
      "grad_norm": 0.0744011253118515,
      "learning_rate": 1.2200754160811568e-08,
      "loss": 0.0009,
      "step": 305590
    },
    {
      "epoch": 1.9600234741031257,
      "grad_norm": 0.14834116399288177,
      "learning_rate": 1.2161708865929268e-08,
      "loss": 0.0016,
      "step": 305600
    },
    {
      "epoch": 1.9600876109969119,
      "grad_norm": 0.06007117033004761,
      "learning_rate": 1.2122726072225089e-08,
      "loss": 0.002,
      "step": 305610
    },
    {
      "epoch": 1.960151747890698,
      "grad_norm": 0.07412609457969666,
      "learning_rate": 1.208380578018864e-08,
      "loss": 0.0005,
      "step": 305620
    },
    {
      "epoch": 1.9602158847844842,
      "grad_norm": 0.04260360077023506,
      "learning_rate": 1.2044947990306199e-08,
      "loss": 0.0006,
      "step": 305630
    },
    {
      "epoch": 1.96028002167827,
      "grad_norm": 0.11977134644985199,
      "learning_rate": 1.2006152703066265e-08,
      "loss": 0.0011,
      "step": 305640
    },
    {
      "epoch": 1.9603441585720562,
      "grad_norm": 0.1177653968334198,
      "learning_rate": 1.196741991895345e-08,
      "loss": 0.0022,
      "step": 305650
    },
    {
      "epoch": 1.9604082954658422,
      "grad_norm": 0.016976799815893173,
      "learning_rate": 1.1928749638454029e-08,
      "loss": 0.0016,
      "step": 305660
    },
    {
      "epoch": 1.9604724323596283,
      "grad_norm": 0.009372200816869736,
      "learning_rate": 1.1890141862052062e-08,
      "loss": 0.0011,
      "step": 305670
    },
    {
      "epoch": 1.9605365692534145,
      "grad_norm": 0.07452521473169327,
      "learning_rate": 1.1851596590231607e-08,
      "loss": 0.0008,
      "step": 305680
    },
    {
      "epoch": 1.9606007061472006,
      "grad_norm": 0.03782688453793526,
      "learning_rate": 1.1813113823475609e-08,
      "loss": 0.0006,
      "step": 305690
    },
    {
      "epoch": 1.9606648430409868,
      "grad_norm": 0.10139846056699753,
      "learning_rate": 1.177469356226646e-08,
      "loss": 0.0011,
      "step": 305700
    },
    {
      "epoch": 1.960728979934773,
      "grad_norm": 0.15534450113773346,
      "learning_rate": 1.1736335807085442e-08,
      "loss": 0.002,
      "step": 305710
    },
    {
      "epoch": 1.9607931168285588,
      "grad_norm": 0.027061453089118004,
      "learning_rate": 1.1698040558413281e-08,
      "loss": 0.0008,
      "step": 305720
    },
    {
      "epoch": 1.960857253722345,
      "grad_norm": 0.0902622640132904,
      "learning_rate": 1.1659807816729595e-08,
      "loss": 0.0024,
      "step": 305730
    },
    {
      "epoch": 1.960921390616131,
      "grad_norm": 0.21433427929878235,
      "learning_rate": 1.1621637582514e-08,
      "loss": 0.001,
      "step": 305740
    },
    {
      "epoch": 1.960985527509917,
      "grad_norm": 0.045051656663417816,
      "learning_rate": 1.1583529856243891e-08,
      "loss": 0.0007,
      "step": 305750
    },
    {
      "epoch": 1.9610496644037032,
      "grad_norm": 0.08645248413085938,
      "learning_rate": 1.154548463839722e-08,
      "loss": 0.001,
      "step": 305760
    },
    {
      "epoch": 1.9611138012974894,
      "grad_norm": 0.02531195990741253,
      "learning_rate": 1.1507501929451381e-08,
      "loss": 0.0012,
      "step": 305770
    },
    {
      "epoch": 1.9611779381912755,
      "grad_norm": 0.20479781925678253,
      "learning_rate": 1.146958172988155e-08,
      "loss": 0.0014,
      "step": 305780
    },
    {
      "epoch": 1.9612420750850617,
      "grad_norm": 0.10863394290208817,
      "learning_rate": 1.1431724040162907e-08,
      "loss": 0.0016,
      "step": 305790
    },
    {
      "epoch": 1.9613062119788478,
      "grad_norm": 0.44899410009384155,
      "learning_rate": 1.1393928860770064e-08,
      "loss": 0.0013,
      "step": 305800
    },
    {
      "epoch": 1.9613703488726337,
      "grad_norm": 0.029866235330700874,
      "learning_rate": 1.1356196192177094e-08,
      "loss": 0.0011,
      "step": 305810
    },
    {
      "epoch": 1.9614344857664199,
      "grad_norm": 0.0236195120960474,
      "learning_rate": 1.1318526034855837e-08,
      "loss": 0.0007,
      "step": 305820
    },
    {
      "epoch": 1.9614986226602058,
      "grad_norm": 0.058961015194654465,
      "learning_rate": 1.1280918389278694e-08,
      "loss": 0.0012,
      "step": 305830
    },
    {
      "epoch": 1.961562759553992,
      "grad_norm": 0.06082117184996605,
      "learning_rate": 1.1243373255917512e-08,
      "loss": 0.0008,
      "step": 305840
    },
    {
      "epoch": 1.961626896447778,
      "grad_norm": 0.00631625484675169,
      "learning_rate": 1.1205890635241912e-08,
      "loss": 0.0015,
      "step": 305850
    },
    {
      "epoch": 1.9616910333415642,
      "grad_norm": 0.05238247662782669,
      "learning_rate": 1.1168470527722075e-08,
      "loss": 0.0012,
      "step": 305860
    },
    {
      "epoch": 1.9617551702353504,
      "grad_norm": 0.0238849688321352,
      "learning_rate": 1.1131112933826516e-08,
      "loss": 0.0015,
      "step": 305870
    },
    {
      "epoch": 1.9618193071291365,
      "grad_norm": 0.13209733366966248,
      "learning_rate": 1.1093817854023748e-08,
      "loss": 0.0011,
      "step": 305880
    },
    {
      "epoch": 1.9618834440229225,
      "grad_norm": 0.08128155767917633,
      "learning_rate": 1.1056585288781174e-08,
      "loss": 0.0018,
      "step": 305890
    },
    {
      "epoch": 1.9619475809167086,
      "grad_norm": 0.10184819251298904,
      "learning_rate": 1.101941523856509e-08,
      "loss": 0.0008,
      "step": 305900
    },
    {
      "epoch": 1.9620117178104948,
      "grad_norm": 0.012238494120538235,
      "learning_rate": 1.0982307703841232e-08,
      "loss": 0.0008,
      "step": 305910
    },
    {
      "epoch": 1.9620758547042807,
      "grad_norm": 0.1712855100631714,
      "learning_rate": 1.0945262685074787e-08,
      "loss": 0.001,
      "step": 305920
    },
    {
      "epoch": 1.9621399915980668,
      "grad_norm": 0.03880096226930618,
      "learning_rate": 1.090828018272927e-08,
      "loss": 0.0005,
      "step": 305930
    },
    {
      "epoch": 1.962204128491853,
      "grad_norm": 0.18040379881858826,
      "learning_rate": 1.0871360197269309e-08,
      "loss": 0.0007,
      "step": 305940
    },
    {
      "epoch": 1.9622682653856391,
      "grad_norm": 0.09596449881792068,
      "learning_rate": 1.0834502729156205e-08,
      "loss": 0.0009,
      "step": 305950
    },
    {
      "epoch": 1.9623324022794253,
      "grad_norm": 0.10705173760652542,
      "learning_rate": 1.0797707778852917e-08,
      "loss": 0.0007,
      "step": 305960
    },
    {
      "epoch": 1.9623965391732114,
      "grad_norm": 0.05151883885264397,
      "learning_rate": 1.0760975346820191e-08,
      "loss": 0.0015,
      "step": 305970
    },
    {
      "epoch": 1.9624606760669974,
      "grad_norm": 0.0651080310344696,
      "learning_rate": 1.0724305433518212e-08,
      "loss": 0.0015,
      "step": 305980
    },
    {
      "epoch": 1.9625248129607835,
      "grad_norm": 0.041869163513183594,
      "learning_rate": 1.0687698039406058e-08,
      "loss": 0.0012,
      "step": 305990
    },
    {
      "epoch": 1.9625889498545694,
      "grad_norm": 0.014452272094786167,
      "learning_rate": 1.0651153164943362e-08,
      "loss": 0.0009,
      "step": 306000
    },
    {
      "epoch": 1.9626530867483556,
      "grad_norm": 0.07938862591981888,
      "learning_rate": 1.0614670810587535e-08,
      "loss": 0.0008,
      "step": 306010
    },
    {
      "epoch": 1.9627172236421417,
      "grad_norm": 0.005171359051018953,
      "learning_rate": 1.0578250976795434e-08,
      "loss": 0.001,
      "step": 306020
    },
    {
      "epoch": 1.9627813605359279,
      "grad_norm": 0.06828926503658295,
      "learning_rate": 1.0541893664023361e-08,
      "loss": 0.0008,
      "step": 306030
    },
    {
      "epoch": 1.962845497429714,
      "grad_norm": 0.06769850850105286,
      "learning_rate": 1.0505598872728174e-08,
      "loss": 0.001,
      "step": 306040
    },
    {
      "epoch": 1.9629096343235002,
      "grad_norm": 0.03014775551855564,
      "learning_rate": 1.0469366603363396e-08,
      "loss": 0.0009,
      "step": 306050
    },
    {
      "epoch": 1.9629737712172863,
      "grad_norm": 0.16810640692710876,
      "learning_rate": 1.043319685638311e-08,
      "loss": 0.0008,
      "step": 306060
    },
    {
      "epoch": 1.9630379081110723,
      "grad_norm": 0.1911521703004837,
      "learning_rate": 1.0397089632240843e-08,
      "loss": 0.0008,
      "step": 306070
    },
    {
      "epoch": 1.9631020450048584,
      "grad_norm": 0.16602012515068054,
      "learning_rate": 1.0361044931389564e-08,
      "loss": 0.0012,
      "step": 306080
    },
    {
      "epoch": 1.9631661818986443,
      "grad_norm": 0.059331078082323074,
      "learning_rate": 1.0325062754280024e-08,
      "loss": 0.0014,
      "step": 306090
    },
    {
      "epoch": 1.9632303187924305,
      "grad_norm": 0.40836986899375916,
      "learning_rate": 1.028914310136353e-08,
      "loss": 0.0061,
      "step": 306100
    },
    {
      "epoch": 1.9632944556862166,
      "grad_norm": 0.08073293417692184,
      "learning_rate": 1.0253285973090277e-08,
      "loss": 0.0005,
      "step": 306110
    },
    {
      "epoch": 1.9633585925800028,
      "grad_norm": 0.02992640808224678,
      "learning_rate": 1.0217491369909905e-08,
      "loss": 0.0008,
      "step": 306120
    },
    {
      "epoch": 1.963422729473789,
      "grad_norm": 0.004306135233491659,
      "learning_rate": 1.0181759292269833e-08,
      "loss": 0.0011,
      "step": 306130
    },
    {
      "epoch": 1.963486866367575,
      "grad_norm": 0.06630755960941315,
      "learning_rate": 1.0146089740618592e-08,
      "loss": 0.0003,
      "step": 306140
    },
    {
      "epoch": 1.963551003261361,
      "grad_norm": 0.15983599424362183,
      "learning_rate": 1.0110482715403603e-08,
      "loss": 0.0014,
      "step": 306150
    },
    {
      "epoch": 1.9636151401551472,
      "grad_norm": 0.1168171614408493,
      "learning_rate": 1.0074938217070062e-08,
      "loss": 0.0008,
      "step": 306160
    },
    {
      "epoch": 1.963679277048933,
      "grad_norm": 0.10949229449033737,
      "learning_rate": 1.0039456246063727e-08,
      "loss": 0.0012,
      "step": 306170
    },
    {
      "epoch": 1.9637434139427192,
      "grad_norm": 0.0038095058407634497,
      "learning_rate": 1.0004036802829242e-08,
      "loss": 0.0005,
      "step": 306180
    },
    {
      "epoch": 1.9638075508365054,
      "grad_norm": 0.059697799384593964,
      "learning_rate": 9.968679887810695e-09,
      "loss": 0.001,
      "step": 306190
    },
    {
      "epoch": 1.9638716877302915,
      "grad_norm": 0.09330522269010544,
      "learning_rate": 9.933385501451064e-09,
      "loss": 0.0003,
      "step": 306200
    },
    {
      "epoch": 1.9639358246240777,
      "grad_norm": 0.06144918128848076,
      "learning_rate": 9.898153644191666e-09,
      "loss": 0.0009,
      "step": 306210
    },
    {
      "epoch": 1.9639999615178638,
      "grad_norm": 0.03653734177350998,
      "learning_rate": 9.862984316475476e-09,
      "loss": 0.0012,
      "step": 306220
    },
    {
      "epoch": 1.96406409841165,
      "grad_norm": 0.25971123576164246,
      "learning_rate": 9.827877518742145e-09,
      "loss": 0.001,
      "step": 306230
    },
    {
      "epoch": 1.964128235305436,
      "grad_norm": 0.040889885276556015,
      "learning_rate": 9.792833251432431e-09,
      "loss": 0.0022,
      "step": 306240
    },
    {
      "epoch": 1.964192372199222,
      "grad_norm": 0.1783040314912796,
      "learning_rate": 9.757851514984318e-09,
      "loss": 0.0014,
      "step": 306250
    },
    {
      "epoch": 1.964256509093008,
      "grad_norm": 0.039647918194532394,
      "learning_rate": 9.722932309837451e-09,
      "loss": 0.0014,
      "step": 306260
    },
    {
      "epoch": 1.9643206459867941,
      "grad_norm": 0.257514089345932,
      "learning_rate": 9.688075636428152e-09,
      "loss": 0.0012,
      "step": 306270
    },
    {
      "epoch": 1.9643847828805803,
      "grad_norm": 0.17515437304973602,
      "learning_rate": 9.653281495194399e-09,
      "loss": 0.0021,
      "step": 306280
    },
    {
      "epoch": 1.9644489197743664,
      "grad_norm": 0.06372150033712387,
      "learning_rate": 9.618549886570849e-09,
      "loss": 0.0015,
      "step": 306290
    },
    {
      "epoch": 1.9645130566681526,
      "grad_norm": 0.011198348365724087,
      "learning_rate": 9.583880810993818e-09,
      "loss": 0.0007,
      "step": 306300
    },
    {
      "epoch": 1.9645771935619387,
      "grad_norm": 0.08975838124752045,
      "learning_rate": 9.549274268897402e-09,
      "loss": 0.0008,
      "step": 306310
    },
    {
      "epoch": 1.9646413304557246,
      "grad_norm": 0.04905629903078079,
      "learning_rate": 9.514730260715144e-09,
      "loss": 0.0004,
      "step": 306320
    },
    {
      "epoch": 1.9647054673495108,
      "grad_norm": 0.04664885625243187,
      "learning_rate": 9.480248786880031e-09,
      "loss": 0.0013,
      "step": 306330
    },
    {
      "epoch": 1.964769604243297,
      "grad_norm": 0.02026456594467163,
      "learning_rate": 9.44582984782394e-09,
      "loss": 0.0008,
      "step": 306340
    },
    {
      "epoch": 1.9648337411370829,
      "grad_norm": 0.043228600174188614,
      "learning_rate": 9.411473443978747e-09,
      "loss": 0.0008,
      "step": 306350
    },
    {
      "epoch": 1.964897878030869,
      "grad_norm": 0.057951316237449646,
      "learning_rate": 9.377179575773554e-09,
      "loss": 0.0005,
      "step": 306360
    },
    {
      "epoch": 1.9649620149246552,
      "grad_norm": 0.0890708640217781,
      "learning_rate": 9.342948243639683e-09,
      "loss": 0.0007,
      "step": 306370
    },
    {
      "epoch": 1.9650261518184413,
      "grad_norm": 0.004393084440380335,
      "learning_rate": 9.308779448005678e-09,
      "loss": 0.0012,
      "step": 306380
    },
    {
      "epoch": 1.9650902887122275,
      "grad_norm": 0.018562108278274536,
      "learning_rate": 9.274673189298977e-09,
      "loss": 0.0008,
      "step": 306390
    },
    {
      "epoch": 1.9651544256060136,
      "grad_norm": 0.004979966674000025,
      "learning_rate": 9.240629467947571e-09,
      "loss": 0.0005,
      "step": 306400
    },
    {
      "epoch": 1.9652185624997995,
      "grad_norm": 0.05927908048033714,
      "learning_rate": 9.206648284377784e-09,
      "loss": 0.0015,
      "step": 306410
    },
    {
      "epoch": 1.9652826993935857,
      "grad_norm": 0.009664489887654781,
      "learning_rate": 9.172729639015943e-09,
      "loss": 0.0015,
      "step": 306420
    },
    {
      "epoch": 1.9653468362873716,
      "grad_norm": 0.054331354796886444,
      "learning_rate": 9.138873532286707e-09,
      "loss": 0.002,
      "step": 306430
    },
    {
      "epoch": 1.9654109731811578,
      "grad_norm": 0.08084974437952042,
      "learning_rate": 9.105079964613627e-09,
      "loss": 0.0011,
      "step": 306440
    },
    {
      "epoch": 1.965475110074944,
      "grad_norm": 0.13514891266822815,
      "learning_rate": 9.071348936421364e-09,
      "loss": 0.0012,
      "step": 306450
    },
    {
      "epoch": 1.96553924696873,
      "grad_norm": 0.08561267703771591,
      "learning_rate": 9.037680448132358e-09,
      "loss": 0.0018,
      "step": 306460
    },
    {
      "epoch": 1.9656033838625162,
      "grad_norm": 0.017613446339964867,
      "learning_rate": 9.004074500167937e-09,
      "loss": 0.0015,
      "step": 306470
    },
    {
      "epoch": 1.9656675207563024,
      "grad_norm": 0.04687608405947685,
      "learning_rate": 8.970531092948875e-09,
      "loss": 0.0013,
      "step": 306480
    },
    {
      "epoch": 1.9657316576500885,
      "grad_norm": 0.18052297830581665,
      "learning_rate": 8.93705022689706e-09,
      "loss": 0.0009,
      "step": 306490
    },
    {
      "epoch": 1.9657957945438744,
      "grad_norm": 0.0018816557712852955,
      "learning_rate": 8.903631902430487e-09,
      "loss": 0.0008,
      "step": 306500
    },
    {
      "epoch": 1.9658599314376606,
      "grad_norm": 0.0032184049487113953,
      "learning_rate": 8.870276119968823e-09,
      "loss": 0.0014,
      "step": 306510
    },
    {
      "epoch": 1.9659240683314465,
      "grad_norm": 0.040409523993730545,
      "learning_rate": 8.836982879928957e-09,
      "loss": 0.0006,
      "step": 306520
    },
    {
      "epoch": 1.9659882052252327,
      "grad_norm": 0.0843072310090065,
      "learning_rate": 8.803752182729441e-09,
      "loss": 0.0007,
      "step": 306530
    },
    {
      "epoch": 1.9660523421190188,
      "grad_norm": 0.07782573997974396,
      "learning_rate": 8.770584028786055e-09,
      "loss": 0.0006,
      "step": 306540
    },
    {
      "epoch": 1.966116479012805,
      "grad_norm": 0.28871554136276245,
      "learning_rate": 8.737478418514578e-09,
      "loss": 0.002,
      "step": 306550
    },
    {
      "epoch": 1.966180615906591,
      "grad_norm": 0.3268186151981354,
      "learning_rate": 8.704435352329121e-09,
      "loss": 0.0016,
      "step": 306560
    },
    {
      "epoch": 1.9662447528003772,
      "grad_norm": 0.03365792706608772,
      "learning_rate": 8.671454830644909e-09,
      "loss": 0.0008,
      "step": 306570
    },
    {
      "epoch": 1.9663088896941632,
      "grad_norm": 0.13226476311683655,
      "learning_rate": 8.63853685387439e-09,
      "loss": 0.0012,
      "step": 306580
    },
    {
      "epoch": 1.9663730265879493,
      "grad_norm": 0.07006511837244034,
      "learning_rate": 8.605681422430012e-09,
      "loss": 0.001,
      "step": 306590
    },
    {
      "epoch": 1.9664371634817353,
      "grad_norm": 0.2049380987882614,
      "learning_rate": 8.572888536724223e-09,
      "loss": 0.0014,
      "step": 306600
    },
    {
      "epoch": 1.9665013003755214,
      "grad_norm": 0.1545920968055725,
      "learning_rate": 8.540158197167247e-09,
      "loss": 0.001,
      "step": 306610
    },
    {
      "epoch": 1.9665654372693075,
      "grad_norm": 0.06900747120380402,
      "learning_rate": 8.507490404169317e-09,
      "loss": 0.0006,
      "step": 306620
    },
    {
      "epoch": 1.9666295741630937,
      "grad_norm": 0.06036344915628433,
      "learning_rate": 8.474885158140101e-09,
      "loss": 0.0008,
      "step": 306630
    },
    {
      "epoch": 1.9666937110568798,
      "grad_norm": 0.0792551040649414,
      "learning_rate": 8.442342459487607e-09,
      "loss": 0.001,
      "step": 306640
    },
    {
      "epoch": 1.966757847950666,
      "grad_norm": 0.017666907981038094,
      "learning_rate": 8.409862308620398e-09,
      "loss": 0.0008,
      "step": 306650
    },
    {
      "epoch": 1.9668219848444521,
      "grad_norm": 0.08386830240488052,
      "learning_rate": 8.377444705944815e-09,
      "loss": 0.0023,
      "step": 306660
    },
    {
      "epoch": 1.966886121738238,
      "grad_norm": 0.005592831410467625,
      "learning_rate": 8.345089651867199e-09,
      "loss": 0.0025,
      "step": 306670
    },
    {
      "epoch": 1.9669502586320242,
      "grad_norm": 0.07596705853939056,
      "learning_rate": 8.312797146793339e-09,
      "loss": 0.0015,
      "step": 306680
    },
    {
      "epoch": 1.9670143955258101,
      "grad_norm": 0.03802109137177467,
      "learning_rate": 8.280567191127354e-09,
      "loss": 0.0007,
      "step": 306690
    },
    {
      "epoch": 1.9670785324195963,
      "grad_norm": 0.03789745643734932,
      "learning_rate": 8.248399785273919e-09,
      "loss": 0.0005,
      "step": 306700
    },
    {
      "epoch": 1.9671426693133824,
      "grad_norm": 0.08203060179948807,
      "learning_rate": 8.216294929634939e-09,
      "loss": 0.0013,
      "step": 306710
    },
    {
      "epoch": 1.9672068062071686,
      "grad_norm": 0.025898775085806847,
      "learning_rate": 8.184252624613975e-09,
      "loss": 0.0016,
      "step": 306720
    },
    {
      "epoch": 1.9672709431009547,
      "grad_norm": 0.16674844920635223,
      "learning_rate": 8.152272870611266e-09,
      "loss": 0.0009,
      "step": 306730
    },
    {
      "epoch": 1.9673350799947409,
      "grad_norm": 0.08714725822210312,
      "learning_rate": 8.120355668028712e-09,
      "loss": 0.0009,
      "step": 306740
    },
    {
      "epoch": 1.967399216888527,
      "grad_norm": 0.10741503536701202,
      "learning_rate": 8.088501017265438e-09,
      "loss": 0.0016,
      "step": 306750
    },
    {
      "epoch": 1.967463353782313,
      "grad_norm": 0.044649600982666016,
      "learning_rate": 8.056708918721123e-09,
      "loss": 0.0012,
      "step": 306760
    },
    {
      "epoch": 1.9675274906760991,
      "grad_norm": 0.1709849089384079,
      "learning_rate": 8.024979372793784e-09,
      "loss": 0.0012,
      "step": 306770
    },
    {
      "epoch": 1.967591627569885,
      "grad_norm": 0.13717105984687805,
      "learning_rate": 7.993312379881435e-09,
      "loss": 0.0024,
      "step": 306780
    },
    {
      "epoch": 1.9676557644636712,
      "grad_norm": 0.03835640475153923,
      "learning_rate": 7.961707940380425e-09,
      "loss": 0.0019,
      "step": 306790
    },
    {
      "epoch": 1.9677199013574573,
      "grad_norm": 0.06769486516714096,
      "learning_rate": 7.930166054687104e-09,
      "loss": 0.0005,
      "step": 306800
    },
    {
      "epoch": 1.9677840382512435,
      "grad_norm": 0.12234804034233093,
      "learning_rate": 7.898686723196159e-09,
      "loss": 0.001,
      "step": 306810
    },
    {
      "epoch": 1.9678481751450296,
      "grad_norm": 0.09259574115276337,
      "learning_rate": 7.867269946302825e-09,
      "loss": 0.0011,
      "step": 306820
    },
    {
      "epoch": 1.9679123120388158,
      "grad_norm": 0.03439909592270851,
      "learning_rate": 7.835915724400678e-09,
      "loss": 0.0007,
      "step": 306830
    },
    {
      "epoch": 1.9679764489326017,
      "grad_norm": 0.10112472623586655,
      "learning_rate": 7.804624057881626e-09,
      "loss": 0.0005,
      "step": 306840
    },
    {
      "epoch": 1.9680405858263879,
      "grad_norm": 0.05775924399495125,
      "learning_rate": 7.773394947139246e-09,
      "loss": 0.0011,
      "step": 306850
    },
    {
      "epoch": 1.9681047227201738,
      "grad_norm": 0.03136734664440155,
      "learning_rate": 7.742228392563778e-09,
      "loss": 0.0017,
      "step": 306860
    },
    {
      "epoch": 1.96816885961396,
      "grad_norm": 0.004767545498907566,
      "learning_rate": 7.711124394546021e-09,
      "loss": 0.0004,
      "step": 306870
    },
    {
      "epoch": 1.968232996507746,
      "grad_norm": 0.060319263488054276,
      "learning_rate": 7.680082953475665e-09,
      "loss": 0.0015,
      "step": 306880
    },
    {
      "epoch": 1.9682971334015322,
      "grad_norm": 0.016754992306232452,
      "learning_rate": 7.649104069742396e-09,
      "loss": 0.0009,
      "step": 306890
    },
    {
      "epoch": 1.9683612702953184,
      "grad_norm": 0.07807540893554688,
      "learning_rate": 7.61818774373313e-09,
      "loss": 0.0014,
      "step": 306900
    },
    {
      "epoch": 1.9684254071891045,
      "grad_norm": 0.12074722349643707,
      "learning_rate": 7.587333975836442e-09,
      "loss": 0.0018,
      "step": 306910
    },
    {
      "epoch": 1.9684895440828907,
      "grad_norm": 0.15460114181041718,
      "learning_rate": 7.556542766438691e-09,
      "loss": 0.0026,
      "step": 306920
    },
    {
      "epoch": 1.9685536809766766,
      "grad_norm": 0.0063071777112782,
      "learning_rate": 7.525814115925124e-09,
      "loss": 0.0009,
      "step": 306930
    },
    {
      "epoch": 1.9686178178704627,
      "grad_norm": 0.06452561914920807,
      "learning_rate": 7.495148024680988e-09,
      "loss": 0.0008,
      "step": 306940
    },
    {
      "epoch": 1.9686819547642487,
      "grad_norm": 0.1113513708114624,
      "learning_rate": 7.464544493090975e-09,
      "loss": 0.0016,
      "step": 306950
    },
    {
      "epoch": 1.9687460916580348,
      "grad_norm": 0.047639150172472,
      "learning_rate": 7.434003521538668e-09,
      "loss": 0.0014,
      "step": 306960
    },
    {
      "epoch": 1.968810228551821,
      "grad_norm": 0.005297117866575718,
      "learning_rate": 7.403525110405985e-09,
      "loss": 0.0006,
      "step": 306970
    },
    {
      "epoch": 1.9688743654456071,
      "grad_norm": 0.04109714552760124,
      "learning_rate": 7.373109260075395e-09,
      "loss": 0.0015,
      "step": 306980
    },
    {
      "epoch": 1.9689385023393933,
      "grad_norm": 0.06713901460170746,
      "learning_rate": 7.342755970927707e-09,
      "loss": 0.0022,
      "step": 306990
    },
    {
      "epoch": 1.9690026392331794,
      "grad_norm": 0.11273383349180222,
      "learning_rate": 7.3124652433437246e-09,
      "loss": 0.0015,
      "step": 307000
    },
    {
      "epoch": 1.9690667761269653,
      "grad_norm": 0.001334617380052805,
      "learning_rate": 7.282237077702592e-09,
      "loss": 0.0009,
      "step": 307010
    },
    {
      "epoch": 1.9691309130207515,
      "grad_norm": 0.07082372158765793,
      "learning_rate": 7.252071474383449e-09,
      "loss": 0.0009,
      "step": 307020
    },
    {
      "epoch": 1.9691950499145376,
      "grad_norm": 0.03211138769984245,
      "learning_rate": 7.2219684337643255e-09,
      "loss": 0.0005,
      "step": 307030
    },
    {
      "epoch": 1.9692591868083236,
      "grad_norm": 0.03824929520487785,
      "learning_rate": 7.1919279562221446e-09,
      "loss": 0.0009,
      "step": 307040
    },
    {
      "epoch": 1.9693233237021097,
      "grad_norm": 0.03742806240916252,
      "learning_rate": 7.16195004213327e-09,
      "loss": 0.0007,
      "step": 307050
    },
    {
      "epoch": 1.9693874605958959,
      "grad_norm": 0.046398505568504333,
      "learning_rate": 7.132034691873513e-09,
      "loss": 0.001,
      "step": 307060
    },
    {
      "epoch": 1.969451597489682,
      "grad_norm": 0.027226777747273445,
      "learning_rate": 7.102181905817573e-09,
      "loss": 0.0011,
      "step": 307070
    },
    {
      "epoch": 1.9695157343834682,
      "grad_norm": 0.0399356409907341,
      "learning_rate": 7.072391684339597e-09,
      "loss": 0.0011,
      "step": 307080
    },
    {
      "epoch": 1.9695798712772543,
      "grad_norm": 0.07177360355854034,
      "learning_rate": 7.042664027813173e-09,
      "loss": 0.0008,
      "step": 307090
    },
    {
      "epoch": 1.9696440081710402,
      "grad_norm": 0.029129868373274803,
      "learning_rate": 7.012998936610782e-09,
      "loss": 0.0013,
      "step": 307100
    },
    {
      "epoch": 1.9697081450648264,
      "grad_norm": 0.017454329878091812,
      "learning_rate": 6.983396411103793e-09,
      "loss": 0.0008,
      "step": 307110
    },
    {
      "epoch": 1.9697722819586123,
      "grad_norm": 0.07953023165464401,
      "learning_rate": 6.953856451663021e-09,
      "loss": 0.0004,
      "step": 307120
    },
    {
      "epoch": 1.9698364188523985,
      "grad_norm": 0.037367139011621475,
      "learning_rate": 6.924379058658725e-09,
      "loss": 0.0011,
      "step": 307130
    },
    {
      "epoch": 1.9699005557461846,
      "grad_norm": 0.05491355061531067,
      "learning_rate": 6.8949642324611654e-09,
      "loss": 0.0005,
      "step": 307140
    },
    {
      "epoch": 1.9699646926399708,
      "grad_norm": 0.12753793597221375,
      "learning_rate": 6.865611973437825e-09,
      "loss": 0.0012,
      "step": 307150
    },
    {
      "epoch": 1.970028829533757,
      "grad_norm": 0.00912963505834341,
      "learning_rate": 6.836322281956742e-09,
      "loss": 0.0007,
      "step": 307160
    },
    {
      "epoch": 1.970092966427543,
      "grad_norm": 0.1551181524991989,
      "learning_rate": 6.807095158385402e-09,
      "loss": 0.0011,
      "step": 307170
    },
    {
      "epoch": 1.9701571033213292,
      "grad_norm": 0.05082277953624725,
      "learning_rate": 6.777930603089622e-09,
      "loss": 0.001,
      "step": 307180
    },
    {
      "epoch": 1.9702212402151151,
      "grad_norm": 0.03281303867697716,
      "learning_rate": 6.7488286164352215e-09,
      "loss": 0.0007,
      "step": 307190
    },
    {
      "epoch": 1.9702853771089013,
      "grad_norm": 0.015306927263736725,
      "learning_rate": 6.719789198786353e-09,
      "loss": 0.0005,
      "step": 307200
    },
    {
      "epoch": 1.9703495140026872,
      "grad_norm": 0.0468122772872448,
      "learning_rate": 6.69081235050717e-09,
      "loss": 0.0013,
      "step": 307210
    },
    {
      "epoch": 1.9704136508964734,
      "grad_norm": 0.017557993531227112,
      "learning_rate": 6.661898071960715e-09,
      "loss": 0.0006,
      "step": 307220
    },
    {
      "epoch": 1.9704777877902595,
      "grad_norm": 0.07849603146314621,
      "learning_rate": 6.633046363509477e-09,
      "loss": 0.0009,
      "step": 307230
    },
    {
      "epoch": 1.9705419246840457,
      "grad_norm": 0.10207756608724594,
      "learning_rate": 6.604257225514832e-09,
      "loss": 0.0014,
      "step": 307240
    },
    {
      "epoch": 1.9706060615778318,
      "grad_norm": 0.040534086525440216,
      "learning_rate": 6.575530658337603e-09,
      "loss": 0.0008,
      "step": 307250
    },
    {
      "epoch": 1.970670198471618,
      "grad_norm": 0.2911793291568756,
      "learning_rate": 6.5468666623380584e-09,
      "loss": 0.0012,
      "step": 307260
    },
    {
      "epoch": 1.9707343353654039,
      "grad_norm": 0.1544741541147232,
      "learning_rate": 6.518265237874244e-09,
      "loss": 0.001,
      "step": 307270
    },
    {
      "epoch": 1.97079847225919,
      "grad_norm": 0.022588972002267838,
      "learning_rate": 6.489726385305872e-09,
      "loss": 0.0008,
      "step": 307280
    },
    {
      "epoch": 1.970862609152976,
      "grad_norm": 0.2511366903781891,
      "learning_rate": 6.4612501049898804e-09,
      "loss": 0.001,
      "step": 307290
    },
    {
      "epoch": 1.970926746046762,
      "grad_norm": 0.016783207654953003,
      "learning_rate": 6.432836397283759e-09,
      "loss": 0.0006,
      "step": 307300
    },
    {
      "epoch": 1.9709908829405482,
      "grad_norm": 0.025259699672460556,
      "learning_rate": 6.404485262542226e-09,
      "loss": 0.0007,
      "step": 307310
    },
    {
      "epoch": 1.9710550198343344,
      "grad_norm": 0.11235236376523972,
      "learning_rate": 6.376196701122217e-09,
      "loss": 0.0007,
      "step": 307320
    },
    {
      "epoch": 1.9711191567281205,
      "grad_norm": 0.08961598575115204,
      "learning_rate": 6.347970713376783e-09,
      "loss": 0.0017,
      "step": 307330
    },
    {
      "epoch": 1.9711832936219067,
      "grad_norm": 0.1996752917766571,
      "learning_rate": 6.319807299660085e-09,
      "loss": 0.0012,
      "step": 307340
    },
    {
      "epoch": 1.9712474305156928,
      "grad_norm": 0.0376928374171257,
      "learning_rate": 6.2917064603257304e-09,
      "loss": 0.0009,
      "step": 307350
    },
    {
      "epoch": 1.9713115674094788,
      "grad_norm": 0.17062856256961823,
      "learning_rate": 6.263668195724548e-09,
      "loss": 0.0013,
      "step": 307360
    },
    {
      "epoch": 1.971375704303265,
      "grad_norm": 0.04219835624098778,
      "learning_rate": 6.235692506209034e-09,
      "loss": 0.0005,
      "step": 307370
    },
    {
      "epoch": 1.9714398411970508,
      "grad_norm": 0.06409318745136261,
      "learning_rate": 6.207779392129465e-09,
      "loss": 0.0007,
      "step": 307380
    },
    {
      "epoch": 1.971503978090837,
      "grad_norm": 0.043306782841682434,
      "learning_rate": 6.179928853835005e-09,
      "loss": 0.0006,
      "step": 307390
    },
    {
      "epoch": 1.9715681149846231,
      "grad_norm": 0.045628488063812256,
      "learning_rate": 6.152140891674818e-09,
      "loss": 0.0005,
      "step": 307400
    },
    {
      "epoch": 1.9716322518784093,
      "grad_norm": 0.09256916493177414,
      "learning_rate": 6.124415505998071e-09,
      "loss": 0.0021,
      "step": 307410
    },
    {
      "epoch": 1.9716963887721954,
      "grad_norm": 0.08107291162014008,
      "learning_rate": 6.096752697150598e-09,
      "loss": 0.0011,
      "step": 307420
    },
    {
      "epoch": 1.9717605256659816,
      "grad_norm": 0.016026349738240242,
      "learning_rate": 6.069152465480455e-09,
      "loss": 0.0013,
      "step": 307430
    },
    {
      "epoch": 1.9718246625597675,
      "grad_norm": 0.045945633202791214,
      "learning_rate": 6.041614811332919e-09,
      "loss": 0.0005,
      "step": 307440
    },
    {
      "epoch": 1.9718887994535537,
      "grad_norm": 0.06947251409292221,
      "learning_rate": 6.014139735053271e-09,
      "loss": 0.0007,
      "step": 307450
    },
    {
      "epoch": 1.9719529363473398,
      "grad_norm": 0.08861232548952103,
      "learning_rate": 5.986727236985124e-09,
      "loss": 0.0004,
      "step": 307460
    },
    {
      "epoch": 1.9720170732411257,
      "grad_norm": 0.11720695346593857,
      "learning_rate": 5.959377317473203e-09,
      "loss": 0.0006,
      "step": 307470
    },
    {
      "epoch": 1.9720812101349119,
      "grad_norm": 0.04580852761864662,
      "learning_rate": 5.932089976858901e-09,
      "loss": 0.0009,
      "step": 307480
    },
    {
      "epoch": 1.972145347028698,
      "grad_norm": 0.07148537784814835,
      "learning_rate": 5.904865215484723e-09,
      "loss": 0.0006,
      "step": 307490
    },
    {
      "epoch": 1.9722094839224842,
      "grad_norm": 0.018794462084770203,
      "learning_rate": 5.877703033692062e-09,
      "loss": 0.0004,
      "step": 307500
    },
    {
      "epoch": 1.9722736208162703,
      "grad_norm": 0.04394660145044327,
      "learning_rate": 5.8506034318206454e-09,
      "loss": 0.0014,
      "step": 307510
    },
    {
      "epoch": 1.9723377577100565,
      "grad_norm": 0.10869495570659637,
      "learning_rate": 5.823566410210757e-09,
      "loss": 0.0012,
      "step": 307520
    },
    {
      "epoch": 1.9724018946038424,
      "grad_norm": 0.03450937941670418,
      "learning_rate": 5.7965919692004605e-09,
      "loss": 0.0012,
      "step": 307530
    },
    {
      "epoch": 1.9724660314976286,
      "grad_norm": 0.020507413893938065,
      "learning_rate": 5.769680109128373e-09,
      "loss": 0.0003,
      "step": 307540
    },
    {
      "epoch": 1.9725301683914145,
      "grad_norm": 0.05193580687046051,
      "learning_rate": 5.742830830332002e-09,
      "loss": 0.0007,
      "step": 307550
    },
    {
      "epoch": 1.9725943052852006,
      "grad_norm": 0.017926448956131935,
      "learning_rate": 5.7160441331466364e-09,
      "loss": 0.0005,
      "step": 307560
    },
    {
      "epoch": 1.9726584421789868,
      "grad_norm": 0.09575698524713516,
      "learning_rate": 5.689320017908673e-09,
      "loss": 0.0014,
      "step": 307570
    },
    {
      "epoch": 1.972722579072773,
      "grad_norm": 0.06409459561109543,
      "learning_rate": 5.662658484953398e-09,
      "loss": 0.0007,
      "step": 307580
    },
    {
      "epoch": 1.972786715966559,
      "grad_norm": 0.1508537083864212,
      "learning_rate": 5.6360595346138805e-09,
      "loss": 0.0013,
      "step": 307590
    },
    {
      "epoch": 1.9728508528603452,
      "grad_norm": 0.20504428446292877,
      "learning_rate": 5.6095231672242955e-09,
      "loss": 0.0011,
      "step": 307600
    },
    {
      "epoch": 1.9729149897541314,
      "grad_norm": 0.20593594014644623,
      "learning_rate": 5.583049383116601e-09,
      "loss": 0.0013,
      "step": 307610
    },
    {
      "epoch": 1.9729791266479173,
      "grad_norm": 0.15778371691703796,
      "learning_rate": 5.556638182623309e-09,
      "loss": 0.0012,
      "step": 307620
    },
    {
      "epoch": 1.9730432635417035,
      "grad_norm": 0.018617259338498116,
      "learning_rate": 5.530289566074154e-09,
      "loss": 0.0008,
      "step": 307630
    },
    {
      "epoch": 1.9731074004354894,
      "grad_norm": 0.08489036560058594,
      "learning_rate": 5.504003533799984e-09,
      "loss": 0.0027,
      "step": 307640
    },
    {
      "epoch": 1.9731715373292755,
      "grad_norm": 0.17733611166477203,
      "learning_rate": 5.477780086130535e-09,
      "loss": 0.0017,
      "step": 307650
    },
    {
      "epoch": 1.9732356742230617,
      "grad_norm": 0.10488373041152954,
      "learning_rate": 5.451619223393878e-09,
      "loss": 0.0009,
      "step": 307660
    },
    {
      "epoch": 1.9732998111168478,
      "grad_norm": 0.0796809196472168,
      "learning_rate": 5.425520945918083e-09,
      "loss": 0.001,
      "step": 307670
    },
    {
      "epoch": 1.973363948010634,
      "grad_norm": 0.07839123159646988,
      "learning_rate": 5.3994852540301125e-09,
      "loss": 0.0011,
      "step": 307680
    },
    {
      "epoch": 1.9734280849044201,
      "grad_norm": 0.09871714562177658,
      "learning_rate": 5.373512148055815e-09,
      "loss": 0.0004,
      "step": 307690
    },
    {
      "epoch": 1.973492221798206,
      "grad_norm": 0.0920652449131012,
      "learning_rate": 5.347601628321042e-09,
      "loss": 0.0007,
      "step": 307700
    },
    {
      "epoch": 1.9735563586919922,
      "grad_norm": 0.15558601915836334,
      "learning_rate": 5.321753695150533e-09,
      "loss": 0.0017,
      "step": 307710
    },
    {
      "epoch": 1.9736204955857781,
      "grad_norm": 0.027167458087205887,
      "learning_rate": 5.2959683488684746e-09,
      "loss": 0.0041,
      "step": 307720
    },
    {
      "epoch": 1.9736846324795643,
      "grad_norm": 0.03996816277503967,
      "learning_rate": 5.270245589797385e-09,
      "loss": 0.0011,
      "step": 307730
    },
    {
      "epoch": 1.9737487693733504,
      "grad_norm": 0.03641613945364952,
      "learning_rate": 5.244585418259785e-09,
      "loss": 0.0014,
      "step": 307740
    },
    {
      "epoch": 1.9738129062671366,
      "grad_norm": 0.1493997573852539,
      "learning_rate": 5.218987834577083e-09,
      "loss": 0.0005,
      "step": 307750
    },
    {
      "epoch": 1.9738770431609227,
      "grad_norm": 0.11646231263875961,
      "learning_rate": 5.193452839070134e-09,
      "loss": 0.0003,
      "step": 307760
    },
    {
      "epoch": 1.9739411800547089,
      "grad_norm": 0.024753261357545853,
      "learning_rate": 5.167980432059238e-09,
      "loss": 0.0013,
      "step": 307770
    },
    {
      "epoch": 1.974005316948495,
      "grad_norm": 0.09728194773197174,
      "learning_rate": 5.142570613863029e-09,
      "loss": 0.0012,
      "step": 307780
    },
    {
      "epoch": 1.974069453842281,
      "grad_norm": 0.06630918383598328,
      "learning_rate": 5.117223384800141e-09,
      "loss": 0.0013,
      "step": 307790
    },
    {
      "epoch": 1.974133590736067,
      "grad_norm": 0.042899925261735916,
      "learning_rate": 5.0919387451886516e-09,
      "loss": 0.0013,
      "step": 307800
    },
    {
      "epoch": 1.974197727629853,
      "grad_norm": 0.029828933998942375,
      "learning_rate": 5.0667166953444205e-09,
      "loss": 0.0009,
      "step": 307810
    },
    {
      "epoch": 1.9742618645236392,
      "grad_norm": 0.021324610337615013,
      "learning_rate": 5.041557235584416e-09,
      "loss": 0.0008,
      "step": 307820
    },
    {
      "epoch": 1.9743260014174253,
      "grad_norm": 0.041082773357629776,
      "learning_rate": 5.016460366223385e-09,
      "loss": 0.0018,
      "step": 307830
    },
    {
      "epoch": 1.9743901383112115,
      "grad_norm": 0.12822555005550385,
      "learning_rate": 4.991426087576079e-09,
      "loss": 0.0023,
      "step": 307840
    },
    {
      "epoch": 1.9744542752049976,
      "grad_norm": 0.026461130008101463,
      "learning_rate": 4.9664543999561335e-09,
      "loss": 0.0014,
      "step": 307850
    },
    {
      "epoch": 1.9745184120987838,
      "grad_norm": 0.011752900667488575,
      "learning_rate": 4.941545303676076e-09,
      "loss": 0.0014,
      "step": 307860
    },
    {
      "epoch": 1.9745825489925697,
      "grad_norm": 0.037690773606300354,
      "learning_rate": 4.9166987990489915e-09,
      "loss": 0.0029,
      "step": 307870
    },
    {
      "epoch": 1.9746466858863558,
      "grad_norm": 0.11344976723194122,
      "learning_rate": 4.891914886385185e-09,
      "loss": 0.0015,
      "step": 307880
    },
    {
      "epoch": 1.974710822780142,
      "grad_norm": 0.16902725398540497,
      "learning_rate": 4.86719356599552e-09,
      "loss": 0.0006,
      "step": 307890
    },
    {
      "epoch": 1.974774959673928,
      "grad_norm": 0.07160399109125137,
      "learning_rate": 4.842534838190305e-09,
      "loss": 0.0009,
      "step": 307900
    },
    {
      "epoch": 1.974839096567714,
      "grad_norm": 0.1412993222475052,
      "learning_rate": 4.817938703277625e-09,
      "loss": 0.0012,
      "step": 307910
    },
    {
      "epoch": 1.9749032334615002,
      "grad_norm": 0.014993314631283283,
      "learning_rate": 4.793405161566678e-09,
      "loss": 0.0025,
      "step": 307920
    },
    {
      "epoch": 1.9749673703552864,
      "grad_norm": 0.07830905169248581,
      "learning_rate": 4.7689342133638846e-09,
      "loss": 0.0004,
      "step": 307930
    },
    {
      "epoch": 1.9750315072490725,
      "grad_norm": 0.08769722282886505,
      "learning_rate": 4.744525858976778e-09,
      "loss": 0.0006,
      "step": 307940
    },
    {
      "epoch": 1.9750956441428587,
      "grad_norm": 0.167959064245224,
      "learning_rate": 4.720180098710669e-09,
      "loss": 0.0013,
      "step": 307950
    },
    {
      "epoch": 1.9751597810366446,
      "grad_norm": 0.08166962116956711,
      "learning_rate": 4.695896932870314e-09,
      "loss": 0.0008,
      "step": 307960
    },
    {
      "epoch": 1.9752239179304307,
      "grad_norm": 0.11420666426420212,
      "learning_rate": 4.671676361761024e-09,
      "loss": 0.001,
      "step": 307970
    },
    {
      "epoch": 1.9752880548242167,
      "grad_norm": 0.1623649150133133,
      "learning_rate": 4.647518385685889e-09,
      "loss": 0.0008,
      "step": 307980
    },
    {
      "epoch": 1.9753521917180028,
      "grad_norm": 0.022991027683019638,
      "learning_rate": 4.623423004946892e-09,
      "loss": 0.0011,
      "step": 307990
    },
    {
      "epoch": 1.975416328611789,
      "grad_norm": 0.07424769550561905,
      "learning_rate": 4.599390219847122e-09,
      "loss": 0.0008,
      "step": 308000
    },
    {
      "epoch": 1.975480465505575,
      "grad_norm": 0.052277661859989166,
      "learning_rate": 4.575420030686895e-09,
      "loss": 0.001,
      "step": 308010
    },
    {
      "epoch": 1.9755446023993612,
      "grad_norm": 0.016164889559149742,
      "learning_rate": 4.551512437767081e-09,
      "loss": 0.0014,
      "step": 308020
    },
    {
      "epoch": 1.9756087392931474,
      "grad_norm": 0.12415362149477005,
      "learning_rate": 4.52766744138633e-09,
      "loss": 0.0014,
      "step": 308030
    },
    {
      "epoch": 1.9756728761869335,
      "grad_norm": 0.1140543669462204,
      "learning_rate": 4.503885041844403e-09,
      "loss": 0.0007,
      "step": 308040
    },
    {
      "epoch": 1.9757370130807195,
      "grad_norm": 0.11638864874839783,
      "learning_rate": 4.480165239439393e-09,
      "loss": 0.0007,
      "step": 308050
    },
    {
      "epoch": 1.9758011499745056,
      "grad_norm": 0.13007675111293793,
      "learning_rate": 4.456508034467733e-09,
      "loss": 0.001,
      "step": 308060
    },
    {
      "epoch": 1.9758652868682915,
      "grad_norm": 0.05380553379654884,
      "learning_rate": 4.432913427226959e-09,
      "loss": 0.0004,
      "step": 308070
    },
    {
      "epoch": 1.9759294237620777,
      "grad_norm": 0.019417930394411087,
      "learning_rate": 4.409381418011283e-09,
      "loss": 0.0007,
      "step": 308080
    },
    {
      "epoch": 1.9759935606558638,
      "grad_norm": 0.06386806815862656,
      "learning_rate": 4.385912007116577e-09,
      "loss": 0.0013,
      "step": 308090
    },
    {
      "epoch": 1.97605769754965,
      "grad_norm": 0.042818184942007065,
      "learning_rate": 4.362505194837052e-09,
      "loss": 0.0005,
      "step": 308100
    },
    {
      "epoch": 1.9761218344434361,
      "grad_norm": 0.04058590158820152,
      "learning_rate": 4.3391609814652516e-09,
      "loss": 0.0007,
      "step": 308110
    },
    {
      "epoch": 1.9761859713372223,
      "grad_norm": 0.07583338022232056,
      "learning_rate": 4.315879367294274e-09,
      "loss": 0.0005,
      "step": 308120
    },
    {
      "epoch": 1.9762501082310082,
      "grad_norm": 0.11366616189479828,
      "learning_rate": 4.292660352615552e-09,
      "loss": 0.0008,
      "step": 308130
    },
    {
      "epoch": 1.9763142451247944,
      "grad_norm": 0.015062171965837479,
      "learning_rate": 4.2695039377205206e-09,
      "loss": 0.0007,
      "step": 308140
    },
    {
      "epoch": 1.9763783820185803,
      "grad_norm": 0.059372857213020325,
      "learning_rate": 4.246410122898392e-09,
      "loss": 0.0008,
      "step": 308150
    },
    {
      "epoch": 1.9764425189123664,
      "grad_norm": 0.009204063564538956,
      "learning_rate": 4.2233789084400456e-09,
      "loss": 0.0004,
      "step": 308160
    },
    {
      "epoch": 1.9765066558061526,
      "grad_norm": 0.36856114864349365,
      "learning_rate": 4.2004102946324735e-09,
      "loss": 0.0027,
      "step": 308170
    },
    {
      "epoch": 1.9765707926999387,
      "grad_norm": 0.0946122258901596,
      "learning_rate": 4.177504281764333e-09,
      "loss": 0.0009,
      "step": 308180
    },
    {
      "epoch": 1.9766349295937249,
      "grad_norm": 0.09484247863292694,
      "learning_rate": 4.154660870122618e-09,
      "loss": 0.0011,
      "step": 308190
    },
    {
      "epoch": 1.976699066487511,
      "grad_norm": 0.04668445140123367,
      "learning_rate": 4.1318800599932095e-09,
      "loss": 0.0011,
      "step": 308200
    },
    {
      "epoch": 1.9767632033812972,
      "grad_norm": 0.0832228809595108,
      "learning_rate": 4.109161851662546e-09,
      "loss": 0.0005,
      "step": 308210
    },
    {
      "epoch": 1.9768273402750831,
      "grad_norm": 0.06130097806453705,
      "learning_rate": 4.086506245413735e-09,
      "loss": 0.0017,
      "step": 308220
    },
    {
      "epoch": 1.9768914771688693,
      "grad_norm": 0.08028879016637802,
      "learning_rate": 4.063913241532103e-09,
      "loss": 0.0015,
      "step": 308230
    },
    {
      "epoch": 1.9769556140626552,
      "grad_norm": 0.032631564885377884,
      "learning_rate": 4.041382840300201e-09,
      "loss": 0.0009,
      "step": 308240
    },
    {
      "epoch": 1.9770197509564413,
      "grad_norm": 0.37207770347595215,
      "learning_rate": 4.018915042000027e-09,
      "loss": 0.0013,
      "step": 308250
    },
    {
      "epoch": 1.9770838878502275,
      "grad_norm": 0.12970571219921112,
      "learning_rate": 3.996509846913577e-09,
      "loss": 0.0015,
      "step": 308260
    },
    {
      "epoch": 1.9771480247440136,
      "grad_norm": 0.1854458898305893,
      "learning_rate": 3.974167255321182e-09,
      "loss": 0.0011,
      "step": 308270
    },
    {
      "epoch": 1.9772121616377998,
      "grad_norm": 0.14242278039455414,
      "learning_rate": 3.951887267503174e-09,
      "loss": 0.0018,
      "step": 308280
    },
    {
      "epoch": 1.977276298531586,
      "grad_norm": 0.10432162880897522,
      "learning_rate": 3.929669883738774e-09,
      "loss": 0.001,
      "step": 308290
    },
    {
      "epoch": 1.977340435425372,
      "grad_norm": 0.1106819212436676,
      "learning_rate": 3.9075151043066475e-09,
      "loss": 0.0011,
      "step": 308300
    },
    {
      "epoch": 1.977404572319158,
      "grad_norm": 0.08355239033699036,
      "learning_rate": 3.8854229294832405e-09,
      "loss": 0.0015,
      "step": 308310
    },
    {
      "epoch": 1.9774687092129442,
      "grad_norm": 0.03320559114217758,
      "learning_rate": 3.8633933595466636e-09,
      "loss": 0.0011,
      "step": 308320
    },
    {
      "epoch": 1.97753284610673,
      "grad_norm": 0.043652139604091644,
      "learning_rate": 3.841426394772252e-09,
      "loss": 0.001,
      "step": 308330
    },
    {
      "epoch": 1.9775969830005162,
      "grad_norm": 0.08986610174179077,
      "learning_rate": 3.8195220354353416e-09,
      "loss": 0.0019,
      "step": 308340
    },
    {
      "epoch": 1.9776611198943024,
      "grad_norm": 0.034796494990587234,
      "learning_rate": 3.797680281810712e-09,
      "loss": 0.0011,
      "step": 308350
    },
    {
      "epoch": 1.9777252567880885,
      "grad_norm": 0.0075569432228803635,
      "learning_rate": 3.775901134172033e-09,
      "loss": 0.0015,
      "step": 308360
    },
    {
      "epoch": 1.9777893936818747,
      "grad_norm": 0.19257009029388428,
      "learning_rate": 3.754184592791866e-09,
      "loss": 0.0012,
      "step": 308370
    },
    {
      "epoch": 1.9778535305756608,
      "grad_norm": 0.02256534807384014,
      "learning_rate": 3.732530657942768e-09,
      "loss": 0.0008,
      "step": 308380
    },
    {
      "epoch": 1.9779176674694467,
      "grad_norm": 0.03438534960150719,
      "learning_rate": 3.710939329895635e-09,
      "loss": 0.0004,
      "step": 308390
    },
    {
      "epoch": 1.977981804363233,
      "grad_norm": 0.042067401111125946,
      "learning_rate": 3.6894106089213623e-09,
      "loss": 0.0025,
      "step": 308400
    },
    {
      "epoch": 1.9780459412570188,
      "grad_norm": 0.11088132113218307,
      "learning_rate": 3.6679444952891773e-09,
      "loss": 0.0016,
      "step": 308410
    },
    {
      "epoch": 1.978110078150805,
      "grad_norm": 0.12521320581436157,
      "learning_rate": 3.646540989268865e-09,
      "loss": 0.0013,
      "step": 308420
    },
    {
      "epoch": 1.9781742150445911,
      "grad_norm": 0.01596071943640709,
      "learning_rate": 3.6252000911279894e-09,
      "loss": 0.0004,
      "step": 308430
    },
    {
      "epoch": 1.9782383519383773,
      "grad_norm": 0.001190666574984789,
      "learning_rate": 3.6039218011341137e-09,
      "loss": 0.0011,
      "step": 308440
    },
    {
      "epoch": 1.9783024888321634,
      "grad_norm": 0.06425302475690842,
      "learning_rate": 3.582706119554247e-09,
      "loss": 0.0014,
      "step": 308450
    },
    {
      "epoch": 1.9783666257259496,
      "grad_norm": 0.04505036026239395,
      "learning_rate": 3.561553046654287e-09,
      "loss": 0.001,
      "step": 308460
    },
    {
      "epoch": 1.9784307626197357,
      "grad_norm": 0.042695268988609314,
      "learning_rate": 3.5404625826984674e-09,
      "loss": 0.0009,
      "step": 308470
    },
    {
      "epoch": 1.9784948995135216,
      "grad_norm": 0.02438264898955822,
      "learning_rate": 3.519434727951576e-09,
      "loss": 0.0009,
      "step": 308480
    },
    {
      "epoch": 1.9785590364073078,
      "grad_norm": 0.02695961482822895,
      "learning_rate": 3.4984694826772912e-09,
      "loss": 0.0019,
      "step": 308490
    },
    {
      "epoch": 1.9786231733010937,
      "grad_norm": 0.0018918344285339117,
      "learning_rate": 3.4775668471381807e-09,
      "loss": 0.0006,
      "step": 308500
    },
    {
      "epoch": 1.9786873101948799,
      "grad_norm": 0.11057069152593613,
      "learning_rate": 3.4567268215957015e-09,
      "loss": 0.001,
      "step": 308510
    },
    {
      "epoch": 1.978751447088666,
      "grad_norm": 0.00726739689707756,
      "learning_rate": 3.435949406311867e-09,
      "loss": 0.0008,
      "step": 308520
    },
    {
      "epoch": 1.9788155839824522,
      "grad_norm": 0.05602853000164032,
      "learning_rate": 3.415234601546469e-09,
      "loss": 0.0024,
      "step": 308530
    },
    {
      "epoch": 1.9788797208762383,
      "grad_norm": 0.054609403014183044,
      "learning_rate": 3.394582407558744e-09,
      "loss": 0.0009,
      "step": 308540
    },
    {
      "epoch": 1.9789438577700245,
      "grad_norm": 0.1321811079978943,
      "learning_rate": 3.3739928246084854e-09,
      "loss": 0.0009,
      "step": 308550
    },
    {
      "epoch": 1.9790079946638104,
      "grad_norm": 0.09992507100105286,
      "learning_rate": 3.353465852952709e-09,
      "loss": 0.0008,
      "step": 308560
    },
    {
      "epoch": 1.9790721315575965,
      "grad_norm": 0.06722811609506607,
      "learning_rate": 3.3330014928495415e-09,
      "loss": 0.0009,
      "step": 308570
    },
    {
      "epoch": 1.9791362684513827,
      "grad_norm": 0.0741615891456604,
      "learning_rate": 3.3125997445548895e-09,
      "loss": 0.0006,
      "step": 308580
    },
    {
      "epoch": 1.9792004053451686,
      "grad_norm": 0.12428124248981476,
      "learning_rate": 3.2922606083241048e-09,
      "loss": 0.0017,
      "step": 308590
    },
    {
      "epoch": 1.9792645422389548,
      "grad_norm": 0.056774888187646866,
      "learning_rate": 3.2719840844125384e-09,
      "loss": 0.0009,
      "step": 308600
    },
    {
      "epoch": 1.979328679132741,
      "grad_norm": 0.03519848361611366,
      "learning_rate": 3.251770173073876e-09,
      "loss": 0.0016,
      "step": 308610
    },
    {
      "epoch": 1.979392816026527,
      "grad_norm": 0.03318639099597931,
      "learning_rate": 3.2316188745618037e-09,
      "loss": 0.001,
      "step": 308620
    },
    {
      "epoch": 1.9794569529203132,
      "grad_norm": 0.019073570147156715,
      "learning_rate": 3.2115301891288975e-09,
      "loss": 0.0006,
      "step": 308630
    },
    {
      "epoch": 1.9795210898140994,
      "grad_norm": 0.00515739107504487,
      "learning_rate": 3.1915041170266225e-09,
      "loss": 0.0008,
      "step": 308640
    },
    {
      "epoch": 1.9795852267078853,
      "grad_norm": 0.03800279274582863,
      "learning_rate": 3.1715406585058893e-09,
      "loss": 0.001,
      "step": 308650
    },
    {
      "epoch": 1.9796493636016714,
      "grad_norm": 0.075951486825943,
      "learning_rate": 3.1516398138170533e-09,
      "loss": 0.0008,
      "step": 308660
    },
    {
      "epoch": 1.9797135004954574,
      "grad_norm": 0.024031635373830795,
      "learning_rate": 3.1318015832088043e-09,
      "loss": 0.0008,
      "step": 308670
    },
    {
      "epoch": 1.9797776373892435,
      "grad_norm": 0.10507842153310776,
      "learning_rate": 3.112025966930943e-09,
      "loss": 0.0012,
      "step": 308680
    },
    {
      "epoch": 1.9798417742830297,
      "grad_norm": 0.04669702425599098,
      "learning_rate": 3.092312965230493e-09,
      "loss": 0.0015,
      "step": 308690
    },
    {
      "epoch": 1.9799059111768158,
      "grad_norm": 0.11659438163042068,
      "learning_rate": 3.0726625783544796e-09,
      "loss": 0.0007,
      "step": 308700
    },
    {
      "epoch": 1.979970048070602,
      "grad_norm": 0.11450156569480896,
      "learning_rate": 3.053074806548817e-09,
      "loss": 0.0008,
      "step": 308710
    },
    {
      "epoch": 1.980034184964388,
      "grad_norm": 0.10217200964689255,
      "learning_rate": 3.0335496500599748e-09,
      "loss": 0.0005,
      "step": 308720
    },
    {
      "epoch": 1.9800983218581742,
      "grad_norm": 0.12170190364122391,
      "learning_rate": 3.014087109131647e-09,
      "loss": 0.0007,
      "step": 308730
    },
    {
      "epoch": 1.9801624587519602,
      "grad_norm": 0.1131337434053421,
      "learning_rate": 2.9946871840086376e-09,
      "loss": 0.001,
      "step": 308740
    },
    {
      "epoch": 1.9802265956457463,
      "grad_norm": 0.048643749207258224,
      "learning_rate": 2.9753498749329757e-09,
      "loss": 0.0012,
      "step": 308750
    },
    {
      "epoch": 1.9802907325395323,
      "grad_norm": 0.08117853105068207,
      "learning_rate": 2.9560751821477997e-09,
      "loss": 0.0003,
      "step": 308760
    },
    {
      "epoch": 1.9803548694333184,
      "grad_norm": 0.0409931018948555,
      "learning_rate": 2.936863105894583e-09,
      "loss": 0.0015,
      "step": 308770
    },
    {
      "epoch": 1.9804190063271045,
      "grad_norm": 0.040586985647678375,
      "learning_rate": 2.917713646413134e-09,
      "loss": 0.0007,
      "step": 308780
    },
    {
      "epoch": 1.9804831432208907,
      "grad_norm": 0.0733502060174942,
      "learning_rate": 2.8986268039443712e-09,
      "loss": 0.0009,
      "step": 308790
    },
    {
      "epoch": 1.9805472801146768,
      "grad_norm": 0.004943209234625101,
      "learning_rate": 2.8796025787275474e-09,
      "loss": 0.0006,
      "step": 308800
    },
    {
      "epoch": 1.980611417008463,
      "grad_norm": 0.14331646263599396,
      "learning_rate": 2.86064097100025e-09,
      "loss": 0.0014,
      "step": 308810
    },
    {
      "epoch": 1.980675553902249,
      "grad_norm": 0.07906894385814667,
      "learning_rate": 2.8417419810011783e-09,
      "loss": 0.0011,
      "step": 308820
    },
    {
      "epoch": 1.980739690796035,
      "grad_norm": 0.11326241493225098,
      "learning_rate": 2.822905608965698e-09,
      "loss": 0.0014,
      "step": 308830
    },
    {
      "epoch": 1.980803827689821,
      "grad_norm": 0.16474296152591705,
      "learning_rate": 2.8041318551308426e-09,
      "loss": 0.0022,
      "step": 308840
    },
    {
      "epoch": 1.9808679645836071,
      "grad_norm": 0.0755569115281105,
      "learning_rate": 2.7854207197319794e-09,
      "loss": 0.0036,
      "step": 308850
    },
    {
      "epoch": 1.9809321014773933,
      "grad_norm": 0.1894502192735672,
      "learning_rate": 2.7667722030028098e-09,
      "loss": 0.0007,
      "step": 308860
    },
    {
      "epoch": 1.9809962383711794,
      "grad_norm": 0.08864570409059525,
      "learning_rate": 2.7481863051775913e-09,
      "loss": 0.0007,
      "step": 308870
    },
    {
      "epoch": 1.9810603752649656,
      "grad_norm": 0.16819244623184204,
      "learning_rate": 2.7296630264894706e-09,
      "loss": 0.0014,
      "step": 308880
    },
    {
      "epoch": 1.9811245121587517,
      "grad_norm": 0.005292494315654039,
      "learning_rate": 2.7112023671699297e-09,
      "loss": 0.0008,
      "step": 308890
    },
    {
      "epoch": 1.9811886490525379,
      "grad_norm": 0.016621023416519165,
      "learning_rate": 2.692804327450449e-09,
      "loss": 0.0027,
      "step": 308900
    },
    {
      "epoch": 1.9812527859463238,
      "grad_norm": 0.12444616109132767,
      "learning_rate": 2.6744689075614005e-09,
      "loss": 0.001,
      "step": 308910
    },
    {
      "epoch": 1.98131692284011,
      "grad_norm": 0.10248826444149017,
      "learning_rate": 2.6561961077331556e-09,
      "loss": 0.001,
      "step": 308920
    },
    {
      "epoch": 1.9813810597338959,
      "grad_norm": 0.0755588635802269,
      "learning_rate": 2.6379859281944196e-09,
      "loss": 0.0009,
      "step": 308930
    },
    {
      "epoch": 1.981445196627682,
      "grad_norm": 0.08173400908708572,
      "learning_rate": 2.619838369172789e-09,
      "loss": 0.0009,
      "step": 308940
    },
    {
      "epoch": 1.9815093335214682,
      "grad_norm": 0.07572423666715622,
      "learning_rate": 2.601753430896414e-09,
      "loss": 0.0007,
      "step": 308950
    },
    {
      "epoch": 1.9815734704152543,
      "grad_norm": 0.04430937021970749,
      "learning_rate": 2.5837311135917807e-09,
      "loss": 0.0012,
      "step": 308960
    },
    {
      "epoch": 1.9816376073090405,
      "grad_norm": 0.41289564967155457,
      "learning_rate": 2.5657714174848192e-09,
      "loss": 0.0013,
      "step": 308970
    },
    {
      "epoch": 1.9817017442028266,
      "grad_norm": 0.08378173410892487,
      "learning_rate": 2.5478743427997943e-09,
      "loss": 0.0009,
      "step": 308980
    },
    {
      "epoch": 1.9817658810966126,
      "grad_norm": 0.05246363952755928,
      "learning_rate": 2.5300398897620816e-09,
      "loss": 0.0012,
      "step": 308990
    },
    {
      "epoch": 1.9818300179903987,
      "grad_norm": 0.06257522851228714,
      "learning_rate": 2.5122680585942803e-09,
      "loss": 0.0009,
      "step": 309000
    },
    {
      "epoch": 1.9818941548841849,
      "grad_norm": 0.0537213571369648,
      "learning_rate": 2.494558849519546e-09,
      "loss": 0.0007,
      "step": 309010
    },
    {
      "epoch": 1.9819582917779708,
      "grad_norm": 0.24839678406715393,
      "learning_rate": 2.476912262759368e-09,
      "loss": 0.0026,
      "step": 309020
    },
    {
      "epoch": 1.982022428671757,
      "grad_norm": 0.06641262769699097,
      "learning_rate": 2.459328298535235e-09,
      "loss": 0.0013,
      "step": 309030
    },
    {
      "epoch": 1.982086565565543,
      "grad_norm": 0.15526942908763885,
      "learning_rate": 2.4418069570675272e-09,
      "loss": 0.0017,
      "step": 309040
    },
    {
      "epoch": 1.9821507024593292,
      "grad_norm": 0.09031543880701065,
      "learning_rate": 2.4243482385755133e-09,
      "loss": 0.0014,
      "step": 309050
    },
    {
      "epoch": 1.9822148393531154,
      "grad_norm": 0.10646952688694,
      "learning_rate": 2.406952143278463e-09,
      "loss": 0.0008,
      "step": 309060
    },
    {
      "epoch": 1.9822789762469015,
      "grad_norm": 0.04314703121781349,
      "learning_rate": 2.3896186713939783e-09,
      "loss": 0.0005,
      "step": 309070
    },
    {
      "epoch": 1.9823431131406875,
      "grad_norm": 0.03664984926581383,
      "learning_rate": 2.3723478231391097e-09,
      "loss": 0.0011,
      "step": 309080
    },
    {
      "epoch": 1.9824072500344736,
      "grad_norm": 0.002345103770494461,
      "learning_rate": 2.3551395987303494e-09,
      "loss": 0.0007,
      "step": 309090
    },
    {
      "epoch": 1.9824713869282595,
      "grad_norm": 0.07222457975149155,
      "learning_rate": 2.3379939983836363e-09,
      "loss": 0.0006,
      "step": 309100
    },
    {
      "epoch": 1.9825355238220457,
      "grad_norm": 0.06955096870660782,
      "learning_rate": 2.3209110223137987e-09,
      "loss": 0.0008,
      "step": 309110
    },
    {
      "epoch": 1.9825996607158318,
      "grad_norm": 0.016046542674303055,
      "learning_rate": 2.3038906707345545e-09,
      "loss": 0.0015,
      "step": 309120
    },
    {
      "epoch": 1.982663797609618,
      "grad_norm": 0.07722265273332596,
      "learning_rate": 2.286932943859621e-09,
      "loss": 0.0041,
      "step": 309130
    },
    {
      "epoch": 1.9827279345034041,
      "grad_norm": 0.11098852753639221,
      "learning_rate": 2.2700378419004963e-09,
      "loss": 0.0023,
      "step": 309140
    },
    {
      "epoch": 1.9827920713971903,
      "grad_norm": 0.10734561830759048,
      "learning_rate": 2.2532053650703433e-09,
      "loss": 0.0016,
      "step": 309150
    },
    {
      "epoch": 1.9828562082909764,
      "grad_norm": 0.07452143728733063,
      "learning_rate": 2.236435513578994e-09,
      "loss": 0.0004,
      "step": 309160
    },
    {
      "epoch": 1.9829203451847623,
      "grad_norm": 0.11671552807092667,
      "learning_rate": 2.2197282876373904e-09,
      "loss": 0.0011,
      "step": 309170
    },
    {
      "epoch": 1.9829844820785485,
      "grad_norm": 0.04381822794675827,
      "learning_rate": 2.2030836874542548e-09,
      "loss": 0.0014,
      "step": 309180
    },
    {
      "epoch": 1.9830486189723344,
      "grad_norm": 0.047901857644319534,
      "learning_rate": 2.1865017132377543e-09,
      "loss": 0.0014,
      "step": 309190
    },
    {
      "epoch": 1.9831127558661206,
      "grad_norm": 0.08380786329507828,
      "learning_rate": 2.169982365197165e-09,
      "loss": 0.0018,
      "step": 309200
    },
    {
      "epoch": 1.9831768927599067,
      "grad_norm": 0.06376513838768005,
      "learning_rate": 2.1535256435378793e-09,
      "loss": 0.0021,
      "step": 309210
    },
    {
      "epoch": 1.9832410296536929,
      "grad_norm": 0.2591180205345154,
      "learning_rate": 2.1371315484675082e-09,
      "loss": 0.0011,
      "step": 309220
    },
    {
      "epoch": 1.983305166547479,
      "grad_norm": 0.07580866664648056,
      "learning_rate": 2.120800080190333e-09,
      "loss": 0.0006,
      "step": 309230
    },
    {
      "epoch": 1.9833693034412652,
      "grad_norm": 0.1985291689634323,
      "learning_rate": 2.1045312389111893e-09,
      "loss": 0.0012,
      "step": 309240
    },
    {
      "epoch": 1.983433440335051,
      "grad_norm": 0.042491424828767776,
      "learning_rate": 2.0883250248349142e-09,
      "loss": 0.0008,
      "step": 309250
    },
    {
      "epoch": 1.9834975772288372,
      "grad_norm": 0.05674993619322777,
      "learning_rate": 2.0721814381635676e-09,
      "loss": 0.0015,
      "step": 309260
    },
    {
      "epoch": 1.9835617141226232,
      "grad_norm": 0.02484288439154625,
      "learning_rate": 2.0561004790997653e-09,
      "loss": 0.0017,
      "step": 309270
    },
    {
      "epoch": 1.9836258510164093,
      "grad_norm": 0.041392479091882706,
      "learning_rate": 2.0400821478450126e-09,
      "loss": 0.0011,
      "step": 309280
    },
    {
      "epoch": 1.9836899879101955,
      "grad_norm": 0.03724001348018646,
      "learning_rate": 2.0241264445997057e-09,
      "loss": 0.0019,
      "step": 309290
    },
    {
      "epoch": 1.9837541248039816,
      "grad_norm": 0.01675868220627308,
      "learning_rate": 2.0082333695642385e-09,
      "loss": 0.0008,
      "step": 309300
    },
    {
      "epoch": 1.9838182616977678,
      "grad_norm": 0.08815675973892212,
      "learning_rate": 1.992402922937897e-09,
      "loss": 0.0011,
      "step": 309310
    },
    {
      "epoch": 1.983882398591554,
      "grad_norm": 0.08469753712415695,
      "learning_rate": 1.976635104918856e-09,
      "loss": 0.0008,
      "step": 309320
    },
    {
      "epoch": 1.98394653548534,
      "grad_norm": 0.12119343876838684,
      "learning_rate": 1.9609299157041796e-09,
      "loss": 0.0008,
      "step": 309330
    },
    {
      "epoch": 1.984010672379126,
      "grad_norm": 0.07586673647165298,
      "learning_rate": 1.9452873554914875e-09,
      "loss": 0.0015,
      "step": 309340
    },
    {
      "epoch": 1.9840748092729121,
      "grad_norm": 0.06572098284959793,
      "learning_rate": 1.929707424476179e-09,
      "loss": 0.001,
      "step": 309350
    },
    {
      "epoch": 1.984138946166698,
      "grad_norm": 0.10405579209327698,
      "learning_rate": 1.914190122853654e-09,
      "loss": 0.0007,
      "step": 309360
    },
    {
      "epoch": 1.9842030830604842,
      "grad_norm": 0.009541662409901619,
      "learning_rate": 1.898735450818201e-09,
      "loss": 0.0011,
      "step": 309370
    },
    {
      "epoch": 1.9842672199542704,
      "grad_norm": 0.013885400258004665,
      "learning_rate": 1.883343408564109e-09,
      "loss": 0.001,
      "step": 309380
    },
    {
      "epoch": 1.9843313568480565,
      "grad_norm": 0.27208417654037476,
      "learning_rate": 1.8680139962834466e-09,
      "loss": 0.0012,
      "step": 309390
    },
    {
      "epoch": 1.9843954937418427,
      "grad_norm": 0.04294242709875107,
      "learning_rate": 1.8527472141688375e-09,
      "loss": 0.0005,
      "step": 309400
    },
    {
      "epoch": 1.9844596306356288,
      "grad_norm": 0.05179812014102936,
      "learning_rate": 1.8375430624117952e-09,
      "loss": 0.0008,
      "step": 309410
    },
    {
      "epoch": 1.9845237675294147,
      "grad_norm": 0.03251425921916962,
      "learning_rate": 1.822401541202168e-09,
      "loss": 0.0007,
      "step": 309420
    },
    {
      "epoch": 1.9845879044232009,
      "grad_norm": 0.05451219528913498,
      "learning_rate": 1.8073226507298036e-09,
      "loss": 0.0012,
      "step": 309430
    },
    {
      "epoch": 1.984652041316987,
      "grad_norm": 0.14095070958137512,
      "learning_rate": 1.7923063911839956e-09,
      "loss": 0.001,
      "step": 309440
    },
    {
      "epoch": 1.984716178210773,
      "grad_norm": 0.10173560678958893,
      "learning_rate": 1.7773527627529263e-09,
      "loss": 0.0009,
      "step": 309450
    },
    {
      "epoch": 1.984780315104559,
      "grad_norm": 0.036518894135951996,
      "learning_rate": 1.7624617656236686e-09,
      "loss": 0.0009,
      "step": 309460
    },
    {
      "epoch": 1.9848444519983452,
      "grad_norm": 0.12577344477176666,
      "learning_rate": 1.74763339998274e-09,
      "loss": 0.001,
      "step": 309470
    },
    {
      "epoch": 1.9849085888921314,
      "grad_norm": 0.014238179661333561,
      "learning_rate": 1.7328676660166578e-09,
      "loss": 0.0005,
      "step": 309480
    },
    {
      "epoch": 1.9849727257859175,
      "grad_norm": 0.047518134117126465,
      "learning_rate": 1.7181645639097188e-09,
      "loss": 0.0009,
      "step": 309490
    },
    {
      "epoch": 1.9850368626797037,
      "grad_norm": 0.15631861984729767,
      "learning_rate": 1.7035240938462204e-09,
      "loss": 0.0016,
      "step": 309500
    },
    {
      "epoch": 1.9851009995734896,
      "grad_norm": 0.013858428224921227,
      "learning_rate": 1.6889462560099046e-09,
      "loss": 0.0004,
      "step": 309510
    },
    {
      "epoch": 1.9851651364672758,
      "grad_norm": 0.045731060206890106,
      "learning_rate": 1.6744310505834027e-09,
      "loss": 0.0013,
      "step": 309520
    },
    {
      "epoch": 1.9852292733610617,
      "grad_norm": 0.01558227464556694,
      "learning_rate": 1.6599784777482365e-09,
      "loss": 0.001,
      "step": 309530
    },
    {
      "epoch": 1.9852934102548478,
      "grad_norm": 0.2027255892753601,
      "learning_rate": 1.645588537685927e-09,
      "loss": 0.0012,
      "step": 309540
    },
    {
      "epoch": 1.985357547148634,
      "grad_norm": 0.06667507439851761,
      "learning_rate": 1.631261230576886e-09,
      "loss": 0.0011,
      "step": 309550
    },
    {
      "epoch": 1.9854216840424201,
      "grad_norm": 0.05854310467839241,
      "learning_rate": 1.6169965566004141e-09,
      "loss": 0.0007,
      "step": 309560
    },
    {
      "epoch": 1.9854858209362063,
      "grad_norm": 0.09671935439109802,
      "learning_rate": 1.6027945159352575e-09,
      "loss": 0.0007,
      "step": 309570
    },
    {
      "epoch": 1.9855499578299924,
      "grad_norm": 0.08241838961839676,
      "learning_rate": 1.5886551087590518e-09,
      "loss": 0.0018,
      "step": 309580
    },
    {
      "epoch": 1.9856140947237786,
      "grad_norm": 0.10554327070713043,
      "learning_rate": 1.5745783352494327e-09,
      "loss": 0.0006,
      "step": 309590
    },
    {
      "epoch": 1.9856782316175645,
      "grad_norm": 0.05703651160001755,
      "learning_rate": 1.5605641955829254e-09,
      "loss": 0.0003,
      "step": 309600
    },
    {
      "epoch": 1.9857423685113507,
      "grad_norm": 0.10870242118835449,
      "learning_rate": 1.5466126899349454e-09,
      "loss": 0.0012,
      "step": 309610
    },
    {
      "epoch": 1.9858065054051366,
      "grad_norm": 0.11203661561012268,
      "learning_rate": 1.5327238184797977e-09,
      "loss": 0.0009,
      "step": 309620
    },
    {
      "epoch": 1.9858706422989227,
      "grad_norm": 0.11214514821767807,
      "learning_rate": 1.5188975813923424e-09,
      "loss": 0.001,
      "step": 309630
    },
    {
      "epoch": 1.9859347791927089,
      "grad_norm": 0.04769856482744217,
      "learning_rate": 1.5051339788452192e-09,
      "loss": 0.0005,
      "step": 309640
    },
    {
      "epoch": 1.985998916086495,
      "grad_norm": 0.10414264351129532,
      "learning_rate": 1.4914330110110675e-09,
      "loss": 0.0021,
      "step": 309650
    },
    {
      "epoch": 1.9860630529802812,
      "grad_norm": 0.059631455689668655,
      "learning_rate": 1.4777946780619724e-09,
      "loss": 0.0006,
      "step": 309660
    },
    {
      "epoch": 1.9861271898740673,
      "grad_norm": 0.09064489603042603,
      "learning_rate": 1.4642189801683527e-09,
      "loss": 0.0005,
      "step": 309670
    },
    {
      "epoch": 1.9861913267678533,
      "grad_norm": 0.3767741620540619,
      "learning_rate": 1.4507059175006277e-09,
      "loss": 0.0023,
      "step": 309680
    },
    {
      "epoch": 1.9862554636616394,
      "grad_norm": 0.05923156067728996,
      "learning_rate": 1.4372554902275514e-09,
      "loss": 0.0007,
      "step": 309690
    },
    {
      "epoch": 1.9863196005554253,
      "grad_norm": 0.07710826396942139,
      "learning_rate": 1.423867698518433e-09,
      "loss": 0.0012,
      "step": 309700
    },
    {
      "epoch": 1.9863837374492115,
      "grad_norm": 0.08025619387626648,
      "learning_rate": 1.4105425425403607e-09,
      "loss": 0.0008,
      "step": 309710
    },
    {
      "epoch": 1.9864478743429976,
      "grad_norm": 0.007851624861359596,
      "learning_rate": 1.3972800224609784e-09,
      "loss": 0.0007,
      "step": 309720
    },
    {
      "epoch": 1.9865120112367838,
      "grad_norm": 0.14910300076007843,
      "learning_rate": 1.3840801384462643e-09,
      "loss": 0.0006,
      "step": 309730
    },
    {
      "epoch": 1.98657614813057,
      "grad_norm": 0.05662422627210617,
      "learning_rate": 1.3709428906610868e-09,
      "loss": 0.0011,
      "step": 309740
    },
    {
      "epoch": 1.986640285024356,
      "grad_norm": 0.025845753028988838,
      "learning_rate": 1.3578682792703136e-09,
      "loss": 0.001,
      "step": 309750
    },
    {
      "epoch": 1.9867044219181422,
      "grad_norm": 0.011817049235105515,
      "learning_rate": 1.344856304438258e-09,
      "loss": 0.0008,
      "step": 309760
    },
    {
      "epoch": 1.9867685588119282,
      "grad_norm": 0.05935809388756752,
      "learning_rate": 1.3319069663275674e-09,
      "loss": 0.0006,
      "step": 309770
    },
    {
      "epoch": 1.9868326957057143,
      "grad_norm": 0.06123574078083038,
      "learning_rate": 1.3190202651008899e-09,
      "loss": 0.0012,
      "step": 309780
    },
    {
      "epoch": 1.9868968325995002,
      "grad_norm": 0.0641397163271904,
      "learning_rate": 1.3061962009192075e-09,
      "loss": 0.002,
      "step": 309790
    },
    {
      "epoch": 1.9869609694932864,
      "grad_norm": 0.05723587051033974,
      "learning_rate": 1.2934347739429476e-09,
      "loss": 0.0002,
      "step": 309800
    },
    {
      "epoch": 1.9870251063870725,
      "grad_norm": 0.004579669795930386,
      "learning_rate": 1.2807359843330925e-09,
      "loss": 0.0007,
      "step": 309810
    },
    {
      "epoch": 1.9870892432808587,
      "grad_norm": 0.022011714056134224,
      "learning_rate": 1.268099832247849e-09,
      "loss": 0.0008,
      "step": 309820
    },
    {
      "epoch": 1.9871533801746448,
      "grad_norm": 0.10788783431053162,
      "learning_rate": 1.2555263178454235e-09,
      "loss": 0.001,
      "step": 309830
    },
    {
      "epoch": 1.987217517068431,
      "grad_norm": 0.061985645443201065,
      "learning_rate": 1.2430154412845785e-09,
      "loss": 0.002,
      "step": 309840
    },
    {
      "epoch": 1.9872816539622171,
      "grad_norm": 0.04927527531981468,
      "learning_rate": 1.2305672027207448e-09,
      "loss": 0.0013,
      "step": 309850
    },
    {
      "epoch": 1.987345790856003,
      "grad_norm": 0.21452468633651733,
      "learning_rate": 1.2181816023110193e-09,
      "loss": 0.0012,
      "step": 309860
    },
    {
      "epoch": 1.9874099277497892,
      "grad_norm": 0.012229650281369686,
      "learning_rate": 1.2058586402097227e-09,
      "loss": 0.0018,
      "step": 309870
    },
    {
      "epoch": 1.9874740646435751,
      "grad_norm": 0.026618018746376038,
      "learning_rate": 1.1935983165711762e-09,
      "loss": 0.0026,
      "step": 309880
    },
    {
      "epoch": 1.9875382015373613,
      "grad_norm": 0.030633946880698204,
      "learning_rate": 1.1814006315502558e-09,
      "loss": 0.0005,
      "step": 309890
    },
    {
      "epoch": 1.9876023384311474,
      "grad_norm": 0.005378763657063246,
      "learning_rate": 1.1692655852985069e-09,
      "loss": 0.0009,
      "step": 309900
    },
    {
      "epoch": 1.9876664753249336,
      "grad_norm": 0.022457418963313103,
      "learning_rate": 1.1571931779685852e-09,
      "loss": 0.0021,
      "step": 309910
    },
    {
      "epoch": 1.9877306122187197,
      "grad_norm": 0.024466663599014282,
      "learning_rate": 1.1451834097114811e-09,
      "loss": 0.0005,
      "step": 309920
    },
    {
      "epoch": 1.9877947491125059,
      "grad_norm": 0.06743562966585159,
      "learning_rate": 1.133236280678185e-09,
      "loss": 0.0009,
      "step": 309930
    },
    {
      "epoch": 1.9878588860062918,
      "grad_norm": 0.047757603228092194,
      "learning_rate": 1.1213517910180217e-09,
      "loss": 0.0009,
      "step": 309940
    },
    {
      "epoch": 1.987923022900078,
      "grad_norm": 0.0661270022392273,
      "learning_rate": 1.1095299408797611e-09,
      "loss": 0.0021,
      "step": 309950
    },
    {
      "epoch": 1.9879871597938639,
      "grad_norm": 0.011862175539135933,
      "learning_rate": 1.097770730411618e-09,
      "loss": 0.0007,
      "step": 309960
    },
    {
      "epoch": 1.98805129668765,
      "grad_norm": 0.023457394912838936,
      "learning_rate": 1.0860741597612524e-09,
      "loss": 0.0005,
      "step": 309970
    },
    {
      "epoch": 1.9881154335814362,
      "grad_norm": 0.06609214842319489,
      "learning_rate": 1.0744402290752132e-09,
      "loss": 0.0029,
      "step": 309980
    },
    {
      "epoch": 1.9881795704752223,
      "grad_norm": 0.10954923182725906,
      "learning_rate": 1.0628689384989399e-09,
      "loss": 0.0006,
      "step": 309990
    },
    {
      "epoch": 1.9882437073690085,
      "grad_norm": 0.09938246011734009,
      "learning_rate": 1.0513602881773166e-09,
      "loss": 0.0012,
      "step": 310000
    },
    {
      "epoch": 1.9883078442627946,
      "grad_norm": 0.056122008711099625,
      "learning_rate": 1.0399142782552273e-09,
      "loss": 0.0007,
      "step": 310010
    },
    {
      "epoch": 1.9883719811565808,
      "grad_norm": 0.06735320389270782,
      "learning_rate": 1.028530908875891e-09,
      "loss": 0.0004,
      "step": 310020
    },
    {
      "epoch": 1.9884361180503667,
      "grad_norm": 0.03025159053504467,
      "learning_rate": 1.017210180181416e-09,
      "loss": 0.0007,
      "step": 310030
    },
    {
      "epoch": 1.9885002549441528,
      "grad_norm": 0.26144954562187195,
      "learning_rate": 1.005952092313911e-09,
      "loss": 0.0016,
      "step": 310040
    },
    {
      "epoch": 1.9885643918379388,
      "grad_norm": 0.05625905096530914,
      "learning_rate": 9.947566454149293e-10,
      "loss": 0.0017,
      "step": 310050
    },
    {
      "epoch": 1.988628528731725,
      "grad_norm": 0.05733387544751167,
      "learning_rate": 9.836238396243592e-10,
      "loss": 0.0008,
      "step": 310060
    },
    {
      "epoch": 1.988692665625511,
      "grad_norm": 0.08595871925354004,
      "learning_rate": 9.725536750815334e-10,
      "loss": 0.0017,
      "step": 310070
    },
    {
      "epoch": 1.9887568025192972,
      "grad_norm": 0.019021539017558098,
      "learning_rate": 9.6154615192523e-10,
      "loss": 0.0003,
      "step": 310080
    },
    {
      "epoch": 1.9888209394130834,
      "grad_norm": 0.08416017144918442,
      "learning_rate": 9.50601270294227e-10,
      "loss": 0.0018,
      "step": 310090
    },
    {
      "epoch": 1.9888850763068695,
      "grad_norm": 0.04324590787291527,
      "learning_rate": 9.397190303239712e-10,
      "loss": 0.0005,
      "step": 310100
    },
    {
      "epoch": 1.9889492132006554,
      "grad_norm": 0.03690469264984131,
      "learning_rate": 9.288994321526857e-10,
      "loss": 0.0008,
      "step": 310110
    },
    {
      "epoch": 1.9890133500944416,
      "grad_norm": 0.014062805101275444,
      "learning_rate": 9.181424759141522e-10,
      "loss": 0.001,
      "step": 310120
    },
    {
      "epoch": 1.9890774869882277,
      "grad_norm": 0.059747498482465744,
      "learning_rate": 9.074481617449282e-10,
      "loss": 0.0016,
      "step": 310130
    },
    {
      "epoch": 1.9891416238820137,
      "grad_norm": 0.02084960974752903,
      "learning_rate": 8.968164897776855e-10,
      "loss": 0.0048,
      "step": 310140
    },
    {
      "epoch": 1.9892057607757998,
      "grad_norm": 0.07610044628381729,
      "learning_rate": 8.862474601462057e-10,
      "loss": 0.0016,
      "step": 310150
    },
    {
      "epoch": 1.989269897669586,
      "grad_norm": 0.21877041459083557,
      "learning_rate": 8.757410729831606e-10,
      "loss": 0.0024,
      "step": 310160
    },
    {
      "epoch": 1.989334034563372,
      "grad_norm": 0.010066531598567963,
      "learning_rate": 8.652973284195565e-10,
      "loss": 0.001,
      "step": 310170
    },
    {
      "epoch": 1.9893981714571582,
      "grad_norm": 0.04835638031363487,
      "learning_rate": 8.549162265869548e-10,
      "loss": 0.001,
      "step": 310180
    },
    {
      "epoch": 1.9894623083509444,
      "grad_norm": 0.03072960674762726,
      "learning_rate": 8.445977676146965e-10,
      "loss": 0.0015,
      "step": 310190
    },
    {
      "epoch": 1.9895264452447303,
      "grad_norm": 0.02436680905520916,
      "learning_rate": 8.343419516326779e-10,
      "loss": 0.001,
      "step": 310200
    },
    {
      "epoch": 1.9895905821385165,
      "grad_norm": 0.028769398108124733,
      "learning_rate": 8.241487787691293e-10,
      "loss": 0.0008,
      "step": 310210
    },
    {
      "epoch": 1.9896547190323024,
      "grad_norm": 0.01909705065190792,
      "learning_rate": 8.140182491517268e-10,
      "loss": 0.0013,
      "step": 310220
    },
    {
      "epoch": 1.9897188559260885,
      "grad_norm": 0.025863518938422203,
      "learning_rate": 8.039503629075907e-10,
      "loss": 0.001,
      "step": 310230
    },
    {
      "epoch": 1.9897829928198747,
      "grad_norm": 0.0008688007947057486,
      "learning_rate": 7.939451201632864e-10,
      "loss": 0.0016,
      "step": 310240
    },
    {
      "epoch": 1.9898471297136608,
      "grad_norm": 0.5137998461723328,
      "learning_rate": 7.84002521043159e-10,
      "loss": 0.0014,
      "step": 310250
    },
    {
      "epoch": 1.989911266607447,
      "grad_norm": 0.10145371407270432,
      "learning_rate": 7.741225656726636e-10,
      "loss": 0.0011,
      "step": 310260
    },
    {
      "epoch": 1.9899754035012331,
      "grad_norm": 0.013956460170447826,
      "learning_rate": 7.643052541750351e-10,
      "loss": 0.0007,
      "step": 310270
    },
    {
      "epoch": 1.9900395403950193,
      "grad_norm": 0.05402513965964317,
      "learning_rate": 7.545505866740632e-10,
      "loss": 0.0014,
      "step": 310280
    },
    {
      "epoch": 1.9901036772888052,
      "grad_norm": 0.030353525653481483,
      "learning_rate": 7.448585632913175e-10,
      "loss": 0.001,
      "step": 310290
    },
    {
      "epoch": 1.9901678141825914,
      "grad_norm": 0.10931088775396347,
      "learning_rate": 7.352291841478121e-10,
      "loss": 0.002,
      "step": 310300
    },
    {
      "epoch": 1.9902319510763773,
      "grad_norm": 0.016313182190060616,
      "learning_rate": 7.256624493656717e-10,
      "loss": 0.001,
      "step": 310310
    },
    {
      "epoch": 1.9902960879701634,
      "grad_norm": 0.12528997659683228,
      "learning_rate": 7.161583590631349e-10,
      "loss": 0.0013,
      "step": 310320
    },
    {
      "epoch": 1.9903602248639496,
      "grad_norm": 0.002206750214099884,
      "learning_rate": 7.06716913360661e-10,
      "loss": 0.0009,
      "step": 310330
    },
    {
      "epoch": 1.9904243617577357,
      "grad_norm": 0.08335158228874207,
      "learning_rate": 6.973381123759338e-10,
      "loss": 0.0019,
      "step": 310340
    },
    {
      "epoch": 1.9904884986515219,
      "grad_norm": 0.11780422180891037,
      "learning_rate": 6.880219562260815e-10,
      "loss": 0.0005,
      "step": 310350
    },
    {
      "epoch": 1.990552635545308,
      "grad_norm": 0.0865696370601654,
      "learning_rate": 6.787684450282328e-10,
      "loss": 0.0009,
      "step": 310360
    },
    {
      "epoch": 1.990616772439094,
      "grad_norm": 0.022653456777334213,
      "learning_rate": 6.695775788989611e-10,
      "loss": 0.0017,
      "step": 310370
    },
    {
      "epoch": 1.99068090933288,
      "grad_norm": 0.03010975569486618,
      "learning_rate": 6.604493579526195e-10,
      "loss": 0.0009,
      "step": 310380
    },
    {
      "epoch": 1.990745046226666,
      "grad_norm": 0.07456130534410477,
      "learning_rate": 6.513837823035607e-10,
      "loss": 0.0007,
      "step": 310390
    },
    {
      "epoch": 1.9908091831204522,
      "grad_norm": 0.3172712028026581,
      "learning_rate": 6.423808520655828e-10,
      "loss": 0.001,
      "step": 310400
    },
    {
      "epoch": 1.9908733200142383,
      "grad_norm": 0.049600999802351,
      "learning_rate": 6.334405673519284e-10,
      "loss": 0.0012,
      "step": 310410
    },
    {
      "epoch": 1.9909374569080245,
      "grad_norm": 0.03408721834421158,
      "learning_rate": 6.24562928274175e-10,
      "loss": 0.0014,
      "step": 310420
    },
    {
      "epoch": 1.9910015938018106,
      "grad_norm": 0.13581745326519012,
      "learning_rate": 6.157479349433448e-10,
      "loss": 0.0005,
      "step": 310430
    },
    {
      "epoch": 1.9910657306955968,
      "grad_norm": 0.030747568234801292,
      "learning_rate": 6.069955874704603e-10,
      "loss": 0.0005,
      "step": 310440
    },
    {
      "epoch": 1.991129867589383,
      "grad_norm": 0.08988082408905029,
      "learning_rate": 5.983058859648783e-10,
      "loss": 0.0012,
      "step": 310450
    },
    {
      "epoch": 1.9911940044831689,
      "grad_norm": 0.07408042997121811,
      "learning_rate": 5.896788305359558e-10,
      "loss": 0.0011,
      "step": 310460
    },
    {
      "epoch": 1.991258141376955,
      "grad_norm": 0.08488886058330536,
      "learning_rate": 5.811144212908293e-10,
      "loss": 0.0017,
      "step": 310470
    },
    {
      "epoch": 1.991322278270741,
      "grad_norm": 0.017424480989575386,
      "learning_rate": 5.726126583377456e-10,
      "loss": 0.0006,
      "step": 310480
    },
    {
      "epoch": 1.991386415164527,
      "grad_norm": 0.004001277964562178,
      "learning_rate": 5.641735417827309e-10,
      "loss": 0.001,
      "step": 310490
    },
    {
      "epoch": 1.9914505520583132,
      "grad_norm": 0.027292709797620773,
      "learning_rate": 5.557970717312566e-10,
      "loss": 0.0009,
      "step": 310500
    },
    {
      "epoch": 1.9915146889520994,
      "grad_norm": 0.10521198809146881,
      "learning_rate": 5.474832482893489e-10,
      "loss": 0.0005,
      "step": 310510
    },
    {
      "epoch": 1.9915788258458855,
      "grad_norm": 0.11118293553590775,
      "learning_rate": 5.392320715602584e-10,
      "loss": 0.0022,
      "step": 310520
    },
    {
      "epoch": 1.9916429627396717,
      "grad_norm": 0.01775343529880047,
      "learning_rate": 5.310435416472359e-10,
      "loss": 0.0015,
      "step": 310530
    },
    {
      "epoch": 1.9917070996334576,
      "grad_norm": 0.05329453945159912,
      "learning_rate": 5.229176586540874e-10,
      "loss": 0.0007,
      "step": 310540
    },
    {
      "epoch": 1.9917712365272437,
      "grad_norm": 0.08876674622297287,
      "learning_rate": 5.148544226812879e-10,
      "loss": 0.0009,
      "step": 310550
    },
    {
      "epoch": 1.99183537342103,
      "grad_norm": 0.055429551750421524,
      "learning_rate": 5.06853833830423e-10,
      "loss": 0.0008,
      "step": 310560
    },
    {
      "epoch": 1.9918995103148158,
      "grad_norm": 0.12451827526092529,
      "learning_rate": 4.989158922019677e-10,
      "loss": 0.0009,
      "step": 310570
    },
    {
      "epoch": 1.991963647208602,
      "grad_norm": 0.13710607588291168,
      "learning_rate": 4.91040597895287e-10,
      "loss": 0.0008,
      "step": 310580
    },
    {
      "epoch": 1.9920277841023881,
      "grad_norm": 0.0039566149935126305,
      "learning_rate": 4.832279510086358e-10,
      "loss": 0.0018,
      "step": 310590
    },
    {
      "epoch": 1.9920919209961743,
      "grad_norm": 0.11361566185951233,
      "learning_rate": 4.754779516402685e-10,
      "loss": 0.0011,
      "step": 310600
    },
    {
      "epoch": 1.9921560578899604,
      "grad_norm": 0.3491664230823517,
      "learning_rate": 4.677905998878852e-10,
      "loss": 0.006,
      "step": 310610
    },
    {
      "epoch": 1.9922201947837466,
      "grad_norm": 0.014886301942169666,
      "learning_rate": 4.6016589584640945e-10,
      "loss": 0.0007,
      "step": 310620
    },
    {
      "epoch": 1.9922843316775325,
      "grad_norm": 0.09547198563814163,
      "learning_rate": 4.5260383961243105e-10,
      "loss": 0.001,
      "step": 310630
    },
    {
      "epoch": 1.9923484685713186,
      "grad_norm": 0.07460545003414154,
      "learning_rate": 4.451044312808739e-10,
      "loss": 0.0011,
      "step": 310640
    },
    {
      "epoch": 1.9924126054651046,
      "grad_norm": 0.1180550679564476,
      "learning_rate": 4.3766767094499673e-10,
      "loss": 0.001,
      "step": 310650
    },
    {
      "epoch": 1.9924767423588907,
      "grad_norm": 0.09302978962659836,
      "learning_rate": 4.3029355869805836e-10,
      "loss": 0.0013,
      "step": 310660
    },
    {
      "epoch": 1.9925408792526769,
      "grad_norm": 0.06479623913764954,
      "learning_rate": 4.229820946327623e-10,
      "loss": 0.0011,
      "step": 310670
    },
    {
      "epoch": 1.992605016146463,
      "grad_norm": 0.027378110215067863,
      "learning_rate": 4.1573327884070203e-10,
      "loss": 0.0008,
      "step": 310680
    },
    {
      "epoch": 1.9926691530402492,
      "grad_norm": 0.01163564994931221,
      "learning_rate": 4.085471114129158e-10,
      "loss": 0.0008,
      "step": 310690
    },
    {
      "epoch": 1.9927332899340353,
      "grad_norm": 0.07593715935945511,
      "learning_rate": 4.014235924387766e-10,
      "loss": 0.0008,
      "step": 310700
    },
    {
      "epoch": 1.9927974268278215,
      "grad_norm": 0.06961977481842041,
      "learning_rate": 3.943627220082125e-10,
      "loss": 0.0013,
      "step": 310710
    },
    {
      "epoch": 1.9928615637216074,
      "grad_norm": 0.04441741853952408,
      "learning_rate": 3.873645002094861e-10,
      "loss": 0.0012,
      "step": 310720
    },
    {
      "epoch": 1.9929257006153935,
      "grad_norm": 0.05210668221116066,
      "learning_rate": 3.804289271297501e-10,
      "loss": 0.001,
      "step": 310730
    },
    {
      "epoch": 1.9929898375091795,
      "grad_norm": 0.05263422802090645,
      "learning_rate": 3.73556002856712e-10,
      "loss": 0.0011,
      "step": 310740
    },
    {
      "epoch": 1.9930539744029656,
      "grad_norm": 0.1447191685438156,
      "learning_rate": 3.6674572747641413e-10,
      "loss": 0.001,
      "step": 310750
    },
    {
      "epoch": 1.9931181112967518,
      "grad_norm": 0.06727171689271927,
      "learning_rate": 3.599981010737885e-10,
      "loss": 0.0014,
      "step": 310760
    },
    {
      "epoch": 1.993182248190538,
      "grad_norm": 0.05539167672395706,
      "learning_rate": 3.5331312373376723e-10,
      "loss": 0.001,
      "step": 310770
    },
    {
      "epoch": 1.993246385084324,
      "grad_norm": 0.058031290769577026,
      "learning_rate": 3.4669079553961706e-10,
      "loss": 0.0012,
      "step": 310780
    },
    {
      "epoch": 1.9933105219781102,
      "grad_norm": 0.0378909595310688,
      "learning_rate": 3.401311165751597e-10,
      "loss": 0.001,
      "step": 310790
    },
    {
      "epoch": 1.9933746588718961,
      "grad_norm": 0.09321986138820648,
      "learning_rate": 3.3363408692144163e-10,
      "loss": 0.0005,
      "step": 310800
    },
    {
      "epoch": 1.9934387957656823,
      "grad_norm": 0.07231327146291733,
      "learning_rate": 3.2719970666061915e-10,
      "loss": 0.002,
      "step": 310810
    },
    {
      "epoch": 1.9935029326594682,
      "grad_norm": 0.05760103464126587,
      "learning_rate": 3.208279758737387e-10,
      "loss": 0.0009,
      "step": 310820
    },
    {
      "epoch": 1.9935670695532544,
      "grad_norm": 0.12361236661672592,
      "learning_rate": 3.1451889463962605e-10,
      "loss": 0.0018,
      "step": 310830
    },
    {
      "epoch": 1.9936312064470405,
      "grad_norm": 0.006827784236520529,
      "learning_rate": 3.082724630376621e-10,
      "loss": 0.0014,
      "step": 310840
    },
    {
      "epoch": 1.9936953433408267,
      "grad_norm": 0.2531537413597107,
      "learning_rate": 3.0208868114667277e-10,
      "loss": 0.0007,
      "step": 310850
    },
    {
      "epoch": 1.9937594802346128,
      "grad_norm": 0.0022021911572664976,
      "learning_rate": 2.959675490432634e-10,
      "loss": 0.0005,
      "step": 310860
    },
    {
      "epoch": 1.993823617128399,
      "grad_norm": 0.0008656713180243969,
      "learning_rate": 2.8990906680514964e-10,
      "loss": 0.001,
      "step": 310870
    },
    {
      "epoch": 1.993887754022185,
      "grad_norm": 0.09890484809875488,
      "learning_rate": 2.839132345072715e-10,
      "loss": 0.0023,
      "step": 310880
    },
    {
      "epoch": 1.993951890915971,
      "grad_norm": 0.10886561870574951,
      "learning_rate": 2.7798005222567923e-10,
      "loss": 0.0009,
      "step": 310890
    },
    {
      "epoch": 1.9940160278097572,
      "grad_norm": 0.04242394119501114,
      "learning_rate": 2.721095200336477e-10,
      "loss": 0.001,
      "step": 310900
    },
    {
      "epoch": 1.994080164703543,
      "grad_norm": 0.06635317206382751,
      "learning_rate": 2.6630163800556163e-10,
      "loss": 0.001,
      "step": 310910
    },
    {
      "epoch": 1.9941443015973292,
      "grad_norm": 0.10035008192062378,
      "learning_rate": 2.605564062141408e-10,
      "loss": 0.0007,
      "step": 310920
    },
    {
      "epoch": 1.9942084384911154,
      "grad_norm": 0.3456178307533264,
      "learning_rate": 2.5487382473154965e-10,
      "loss": 0.0026,
      "step": 310930
    },
    {
      "epoch": 1.9942725753849015,
      "grad_norm": 0.025355815887451172,
      "learning_rate": 2.4925389362828734e-10,
      "loss": 0.0008,
      "step": 310940
    },
    {
      "epoch": 1.9943367122786877,
      "grad_norm": 0.176799014210701,
      "learning_rate": 2.43696612974853e-10,
      "loss": 0.0015,
      "step": 310950
    },
    {
      "epoch": 1.9944008491724738,
      "grad_norm": 0.07181256264448166,
      "learning_rate": 2.3820198284119076e-10,
      "loss": 0.0006,
      "step": 310960
    },
    {
      "epoch": 1.9944649860662598,
      "grad_norm": 0.033693600445985794,
      "learning_rate": 2.3277000329668952e-10,
      "loss": 0.0005,
      "step": 310970
    },
    {
      "epoch": 1.994529122960046,
      "grad_norm": 0.07449740916490555,
      "learning_rate": 2.2740067440851778e-10,
      "loss": 0.0016,
      "step": 310980
    },
    {
      "epoch": 1.994593259853832,
      "grad_norm": 0.014070970937609673,
      "learning_rate": 2.2209399624439908e-10,
      "loss": 0.001,
      "step": 310990
    },
    {
      "epoch": 1.994657396747618,
      "grad_norm": 0.07527777552604675,
      "learning_rate": 2.168499688703918e-10,
      "loss": 0.0009,
      "step": 311000
    },
    {
      "epoch": 1.9947215336414041,
      "grad_norm": 0.060144905000925064,
      "learning_rate": 2.1166859235255412e-10,
      "loss": 0.0017,
      "step": 311010
    },
    {
      "epoch": 1.9947856705351903,
      "grad_norm": 0.06478618830442429,
      "learning_rate": 2.0654986675638922e-10,
      "loss": 0.0005,
      "step": 311020
    },
    {
      "epoch": 1.9948498074289764,
      "grad_norm": 0.114219069480896,
      "learning_rate": 2.0149379214517982e-10,
      "loss": 0.0014,
      "step": 311030
    },
    {
      "epoch": 1.9949139443227626,
      "grad_norm": 0.05162247642874718,
      "learning_rate": 1.9650036858220867e-10,
      "loss": 0.0024,
      "step": 311040
    },
    {
      "epoch": 1.9949780812165487,
      "grad_norm": 0.2268955409526825,
      "learning_rate": 1.9156959613075843e-10,
      "loss": 0.0011,
      "step": 311050
    },
    {
      "epoch": 1.9950422181103347,
      "grad_norm": 0.093012735247612,
      "learning_rate": 1.867014748518914e-10,
      "loss": 0.0007,
      "step": 311060
    },
    {
      "epoch": 1.9951063550041208,
      "grad_norm": 0.06731496751308441,
      "learning_rate": 1.8189600480778003e-10,
      "loss": 0.0008,
      "step": 311070
    },
    {
      "epoch": 1.9951704918979067,
      "grad_norm": 0.1012783870100975,
      "learning_rate": 1.7715318605726617e-10,
      "loss": 0.001,
      "step": 311080
    },
    {
      "epoch": 1.9952346287916929,
      "grad_norm": 0.07619098573923111,
      "learning_rate": 1.7247301866030186e-10,
      "loss": 0.0012,
      "step": 311090
    },
    {
      "epoch": 1.995298765685479,
      "grad_norm": 0.09405265748500824,
      "learning_rate": 1.678555026751738e-10,
      "loss": 0.0008,
      "step": 311100
    },
    {
      "epoch": 1.9953629025792652,
      "grad_norm": 0.07160182297229767,
      "learning_rate": 1.6330063816072384e-10,
      "loss": 0.0008,
      "step": 311110
    },
    {
      "epoch": 1.9954270394730513,
      "grad_norm": 0.12579509615898132,
      "learning_rate": 1.588084251735733e-10,
      "loss": 0.0009,
      "step": 311120
    },
    {
      "epoch": 1.9954911763668375,
      "grad_norm": 0.00943692959845066,
      "learning_rate": 1.543788637692334e-10,
      "loss": 0.0008,
      "step": 311130
    },
    {
      "epoch": 1.9955553132606236,
      "grad_norm": 0.05778767913579941,
      "learning_rate": 1.5001195400432544e-10,
      "loss": 0.001,
      "step": 311140
    },
    {
      "epoch": 1.9956194501544096,
      "grad_norm": 0.019559109583497047,
      "learning_rate": 1.457076959332504e-10,
      "loss": 0.0007,
      "step": 311150
    },
    {
      "epoch": 1.9956835870481957,
      "grad_norm": 0.0378846637904644,
      "learning_rate": 1.414660896098541e-10,
      "loss": 0.0016,
      "step": 311160
    },
    {
      "epoch": 1.9957477239419816,
      "grad_norm": 0.03069428727030754,
      "learning_rate": 1.372871350868721e-10,
      "loss": 0.0038,
      "step": 311170
    },
    {
      "epoch": 1.9958118608357678,
      "grad_norm": 0.05695541203022003,
      "learning_rate": 1.3317083241704e-10,
      "loss": 0.0006,
      "step": 311180
    },
    {
      "epoch": 1.995875997729554,
      "grad_norm": 0.08141503483057022,
      "learning_rate": 1.2911718165198317e-10,
      "loss": 0.0007,
      "step": 311190
    },
    {
      "epoch": 1.99594013462334,
      "grad_norm": 0.08701863139867783,
      "learning_rate": 1.2512618284221678e-10,
      "loss": 0.002,
      "step": 311200
    },
    {
      "epoch": 1.9960042715171262,
      "grad_norm": 0.02023044228553772,
      "learning_rate": 1.2119783603825596e-10,
      "loss": 0.0008,
      "step": 311210
    },
    {
      "epoch": 1.9960684084109124,
      "grad_norm": 0.11604054272174835,
      "learning_rate": 1.1733214128950565e-10,
      "loss": 0.0011,
      "step": 311220
    },
    {
      "epoch": 1.9961325453046983,
      "grad_norm": 0.028305407613515854,
      "learning_rate": 1.1352909864315031e-10,
      "loss": 0.0007,
      "step": 311230
    },
    {
      "epoch": 1.9961966821984845,
      "grad_norm": 0.04502418637275696,
      "learning_rate": 1.0978870814803977e-10,
      "loss": 0.0007,
      "step": 311240
    },
    {
      "epoch": 1.9962608190922704,
      "grad_norm": 0.002515362109988928,
      "learning_rate": 1.0611096985024827e-10,
      "loss": 0.0008,
      "step": 311250
    },
    {
      "epoch": 1.9963249559860565,
      "grad_norm": 0.0246027410030365,
      "learning_rate": 1.024958837969603e-10,
      "loss": 0.0009,
      "step": 311260
    },
    {
      "epoch": 1.9963890928798427,
      "grad_norm": 0.005193647928535938,
      "learning_rate": 9.894345003202966e-11,
      "loss": 0.0003,
      "step": 311270
    },
    {
      "epoch": 1.9964532297736288,
      "grad_norm": 0.06985964626073837,
      "learning_rate": 9.54536686015306e-11,
      "loss": 0.0005,
      "step": 311280
    },
    {
      "epoch": 1.996517366667415,
      "grad_norm": 0.06013587862253189,
      "learning_rate": 9.202653954765161e-11,
      "loss": 0.0011,
      "step": 311290
    },
    {
      "epoch": 1.9965815035612011,
      "grad_norm": 0.04733504727482796,
      "learning_rate": 8.86620629142465e-11,
      "loss": 0.0007,
      "step": 311300
    },
    {
      "epoch": 1.9966456404549873,
      "grad_norm": 0.04047602415084839,
      "learning_rate": 8.536023874350374e-11,
      "loss": 0.0008,
      "step": 311310
    },
    {
      "epoch": 1.9967097773487732,
      "grad_norm": 0.03270931541919708,
      "learning_rate": 8.212106707650158e-11,
      "loss": 0.0005,
      "step": 311320
    },
    {
      "epoch": 1.9967739142425593,
      "grad_norm": 0.01887642592191696,
      "learning_rate": 7.894454795376317e-11,
      "loss": 0.0009,
      "step": 311330
    },
    {
      "epoch": 1.9968380511363453,
      "grad_norm": 0.3225075602531433,
      "learning_rate": 7.583068141525652e-11,
      "loss": 0.0013,
      "step": 311340
    },
    {
      "epoch": 1.9969021880301314,
      "grad_norm": 0.011128585785627365,
      "learning_rate": 7.277946749983944e-11,
      "loss": 0.001,
      "step": 311350
    },
    {
      "epoch": 1.9969663249239176,
      "grad_norm": 0.03264245018362999,
      "learning_rate": 6.979090624636975e-11,
      "loss": 0.0008,
      "step": 311360
    },
    {
      "epoch": 1.9970304618177037,
      "grad_norm": 0.0062835942953825,
      "learning_rate": 6.68649976914848e-11,
      "loss": 0.0009,
      "step": 311370
    },
    {
      "epoch": 1.9970945987114899,
      "grad_norm": 0.006522227544337511,
      "learning_rate": 6.400174187237707e-11,
      "loss": 0.0025,
      "step": 311380
    },
    {
      "epoch": 1.997158735605276,
      "grad_norm": 0.06396842002868652,
      "learning_rate": 6.120113882457368e-11,
      "loss": 0.0007,
      "step": 311390
    },
    {
      "epoch": 1.9972228724990622,
      "grad_norm": 0.08676396310329437,
      "learning_rate": 5.846318858304667e-11,
      "loss": 0.0018,
      "step": 311400
    },
    {
      "epoch": 1.997287009392848,
      "grad_norm": 0.1269548535346985,
      "learning_rate": 5.578789118276806e-11,
      "loss": 0.0014,
      "step": 311410
    },
    {
      "epoch": 1.9973511462866342,
      "grad_norm": 0.0786188468337059,
      "learning_rate": 5.3175246657044546e-11,
      "loss": 0.0012,
      "step": 311420
    },
    {
      "epoch": 1.9974152831804202,
      "grad_norm": 0.11513978242874146,
      "learning_rate": 5.062525503807258e-11,
      "loss": 0.0017,
      "step": 311430
    },
    {
      "epoch": 1.9974794200742063,
      "grad_norm": 0.0169609896838665,
      "learning_rate": 4.813791635804865e-11,
      "loss": 0.0009,
      "step": 311440
    },
    {
      "epoch": 1.9975435569679925,
      "grad_norm": 0.09304054081439972,
      "learning_rate": 4.57132306486141e-11,
      "loss": 0.0023,
      "step": 311450
    },
    {
      "epoch": 1.9976076938617786,
      "grad_norm": 0.007657730020582676,
      "learning_rate": 4.335119793974496e-11,
      "loss": 0.0009,
      "step": 311460
    },
    {
      "epoch": 1.9976718307555648,
      "grad_norm": 0.04164130985736847,
      "learning_rate": 4.105181826086213e-11,
      "loss": 0.0014,
      "step": 311470
    },
    {
      "epoch": 1.997735967649351,
      "grad_norm": 0.07427259534597397,
      "learning_rate": 3.8815091641386526e-11,
      "loss": 0.001,
      "step": 311480
    },
    {
      "epoch": 1.9978001045431368,
      "grad_norm": 0.010587507858872414,
      "learning_rate": 3.664101810851861e-11,
      "loss": 0.0012,
      "step": 311490
    },
    {
      "epoch": 1.997864241436923,
      "grad_norm": 0.0805448368191719,
      "learning_rate": 3.4529597690569074e-11,
      "loss": 0.0012,
      "step": 311500
    },
    {
      "epoch": 1.997928378330709,
      "grad_norm": 0.11400971561670303,
      "learning_rate": 3.248083041307304e-11,
      "loss": 0.0014,
      "step": 311510
    },
    {
      "epoch": 1.997992515224495,
      "grad_norm": 0.04872676730155945,
      "learning_rate": 3.049471630212075e-11,
      "loss": 0.0006,
      "step": 311520
    },
    {
      "epoch": 1.9980566521182812,
      "grad_norm": 0.024912770837545395,
      "learning_rate": 2.857125538213712e-11,
      "loss": 0.0023,
      "step": 311530
    },
    {
      "epoch": 1.9981207890120674,
      "grad_norm": 0.07626421749591827,
      "learning_rate": 2.6710447677547046e-11,
      "loss": 0.0008,
      "step": 311540
    },
    {
      "epoch": 1.9981849259058535,
      "grad_norm": 0.04140548035502434,
      "learning_rate": 2.491229321222033e-11,
      "loss": 0.0007,
      "step": 311550
    },
    {
      "epoch": 1.9982490627996397,
      "grad_norm": 0.036363955587148666,
      "learning_rate": 2.3176792007806314e-11,
      "loss": 0.0006,
      "step": 311560
    },
    {
      "epoch": 1.9983131996934258,
      "grad_norm": 0.04990491643548012,
      "learning_rate": 2.1503944086509465e-11,
      "loss": 0.0006,
      "step": 311570
    },
    {
      "epoch": 1.9983773365872117,
      "grad_norm": 0.07418892532587051,
      "learning_rate": 1.9893749469424017e-11,
      "loss": 0.0062,
      "step": 311580
    },
    {
      "epoch": 1.9984414734809979,
      "grad_norm": 0.001999436877667904,
      "learning_rate": 1.8346208175978874e-11,
      "loss": 0.0016,
      "step": 311590
    },
    {
      "epoch": 1.9985056103747838,
      "grad_norm": 0.09791887551546097,
      "learning_rate": 1.686132022671316e-11,
      "loss": 0.0015,
      "step": 311600
    },
    {
      "epoch": 1.99856974726857,
      "grad_norm": 0.09272914379835129,
      "learning_rate": 1.543908563939045e-11,
      "loss": 0.0011,
      "step": 311610
    },
    {
      "epoch": 1.998633884162356,
      "grad_norm": 0.05816177278757095,
      "learning_rate": 1.4079504431774304e-11,
      "loss": 0.0008,
      "step": 311620
    },
    {
      "epoch": 1.9986980210561422,
      "grad_norm": 0.032101716846227646,
      "learning_rate": 1.2782576621073183e-11,
      "loss": 0.0012,
      "step": 311630
    },
    {
      "epoch": 1.9987621579499284,
      "grad_norm": 0.011775681748986244,
      "learning_rate": 1.1548302223940433e-11,
      "loss": 0.0019,
      "step": 311640
    },
    {
      "epoch": 1.9988262948437145,
      "grad_norm": 0.060431741178035736,
      "learning_rate": 1.0376681255919174e-11,
      "loss": 0.0008,
      "step": 311650
    },
    {
      "epoch": 1.9988904317375005,
      "grad_norm": 0.07768959552049637,
      "learning_rate": 9.267713730887196e-12,
      "loss": 0.0008,
      "step": 311660
    },
    {
      "epoch": 1.9989545686312866,
      "grad_norm": 0.09777266532182693,
      "learning_rate": 8.221399663277396e-12,
      "loss": 0.0029,
      "step": 311670
    },
    {
      "epoch": 1.9990187055250728,
      "grad_norm": 0.04257415980100632,
      "learning_rate": 7.237739065857341e-12,
      "loss": 0.0011,
      "step": 311680
    },
    {
      "epoch": 1.9990828424188587,
      "grad_norm": 0.026949284598231316,
      "learning_rate": 6.316731951394594e-12,
      "loss": 0.0012,
      "step": 311690
    },
    {
      "epoch": 1.9991469793126448,
      "grad_norm": 0.004648712929338217,
      "learning_rate": 5.458378330991387e-12,
      "loss": 0.0007,
      "step": 311700
    },
    {
      "epoch": 1.999211116206431,
      "grad_norm": 0.025176560506224632,
      "learning_rate": 4.662678216305061e-12,
      "loss": 0.0011,
      "step": 311710
    },
    {
      "epoch": 1.9992752531002171,
      "grad_norm": 0.06456632167100906,
      "learning_rate": 3.929631615662288e-12,
      "loss": 0.0025,
      "step": 311720
    },
    {
      "epoch": 1.9993393899940033,
      "grad_norm": 0.030822809785604477,
      "learning_rate": 3.2592385401652992e-12,
      "loss": 0.0008,
      "step": 311730
    },
    {
      "epoch": 1.9994035268877894,
      "grad_norm": 0.0362224243581295,
      "learning_rate": 2.6514989964754323e-12,
      "loss": 0.0013,
      "step": 311740
    },
    {
      "epoch": 1.9994676637815754,
      "grad_norm": 0.02068529650568962,
      "learning_rate": 2.1064129934744713e-12,
      "loss": 0.0014,
      "step": 311750
    },
    {
      "epoch": 1.9995318006753615,
      "grad_norm": 0.1268768012523651,
      "learning_rate": 1.6239805372686435e-12,
      "loss": 0.0009,
      "step": 311760
    },
    {
      "epoch": 1.9995959375691474,
      "grad_norm": 0.04929465427994728,
      "learning_rate": 1.2042016345192864e-12,
      "loss": 0.0022,
      "step": 311770
    },
    {
      "epoch": 1.9996600744629336,
      "grad_norm": 0.0022703553549945354,
      "learning_rate": 8.470762902224039e-13,
      "loss": 0.0011,
      "step": 311780
    },
    {
      "epoch": 1.9997242113567197,
      "grad_norm": 0.09381227195262909,
      "learning_rate": 5.526045082637765e-13,
      "loss": 0.0014,
      "step": 311790
    },
    {
      "epoch": 1.9997883482505059,
      "grad_norm": 0.06136627495288849,
      "learning_rate": 3.2078629308429643e-13,
      "loss": 0.0008,
      "step": 311800
    },
    {
      "epoch": 1.999852485144292,
      "grad_norm": 0.06326216459274292,
      "learning_rate": 1.5162164690440962e-13,
      "loss": 0.0007,
      "step": 311810
    },
    {
      "epoch": 1.9999166220380782,
      "grad_norm": 0.06475763022899628,
      "learning_rate": 4.5110573054785166e-14,
      "loss": 0.0007,
      "step": 311820
    },
    {
      "epoch": 1.9999807589318643,
      "grad_norm": 0.20405817031860352,
      "learning_rate": 1.2530715354230894e-15,
      "loss": 0.0024,
      "step": 311830
    },
    {
      "epoch": 1.9999935863106213,
      "step": 311832,
      "total_flos": 1.0432719577686213e+19,
      "train_loss": 0.004895635656955134,
      "train_runtime": 187864.5289,
      "train_samples_per_second": 13.279,
      "train_steps_per_second": 1.66
    }
  ],
  "logging_steps": 10,
  "max_steps": 311832,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 62000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0432719577686213e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}