{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.04848151844115758,
  "eval_steps": 500,
  "global_step": 12000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 4.040126536763131e-05,
      "grad_norm": 13786306.0,
      "learning_rate": 2e-08,
      "loss": 1195071.7,
      "step": 10
    },
    {
      "epoch": 8.080253073526263e-05,
      "grad_norm": 3502914.5,
      "learning_rate": 4e-08,
      "loss": 1488604.5,
      "step": 20
    },
    {
      "epoch": 0.00012120379610289395,
      "grad_norm": 5249533.5,
      "learning_rate": 6e-08,
      "loss": 1363003.1,
      "step": 30
    },
    {
      "epoch": 0.00016160506147052525,
      "grad_norm": 17823226.0,
      "learning_rate": 8e-08,
      "loss": 1539778.9,
      "step": 40
    },
    {
      "epoch": 0.00020200632683815657,
      "grad_norm": 14010826.0,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 1378274.9,
      "step": 50
    },
    {
      "epoch": 0.0002424075922057879,
      "grad_norm": 23003992.0,
      "learning_rate": 1.2e-07,
      "loss": 1250654.1,
      "step": 60
    },
    {
      "epoch": 0.0002828088575734192,
      "grad_norm": 7809646.0,
      "learning_rate": 1.4e-07,
      "loss": 1819393.6,
      "step": 70
    },
    {
      "epoch": 0.0003232101229410505,
      "grad_norm": 43233020.0,
      "learning_rate": 1.6e-07,
      "loss": 1770900.4,
      "step": 80
    },
    {
      "epoch": 0.0003636113883086818,
      "grad_norm": 6094604.5,
      "learning_rate": 1.8e-07,
      "loss": 1529363.0,
      "step": 90
    },
    {
      "epoch": 0.00040401265367631315,
      "grad_norm": 8696879.0,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 1028300.1,
      "step": 100
    },
    {
      "epoch": 0.00044441391904394446,
      "grad_norm": 7774881.5,
      "learning_rate": 2.2e-07,
      "loss": 1786590.6,
      "step": 110
    },
    {
      "epoch": 0.0004848151844115758,
      "grad_norm": 2879459.0,
      "learning_rate": 2.4e-07,
      "loss": 1377895.4,
      "step": 120
    },
    {
      "epoch": 0.000525216449779207,
      "grad_norm": 19783298.0,
      "learning_rate": 2.6e-07,
      "loss": 1599422.0,
      "step": 130
    },
    {
      "epoch": 0.0005656177151468384,
      "grad_norm": 2185791.0,
      "learning_rate": 2.8e-07,
      "loss": 1243788.4,
      "step": 140
    },
    {
      "epoch": 0.0006060189805144697,
      "grad_norm": 9816560.0,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 1245093.3,
      "step": 150
    },
    {
      "epoch": 0.000646420245882101,
      "grad_norm": 9404809.0,
      "learning_rate": 3.2e-07,
      "loss": 1288713.5,
      "step": 160
    },
    {
      "epoch": 0.0006868215112497323,
      "grad_norm": 8446407.0,
      "learning_rate": 3.4e-07,
      "loss": 1147532.8,
      "step": 170
    },
    {
      "epoch": 0.0007272227766173637,
      "grad_norm": 3988919.0,
      "learning_rate": 3.6e-07,
      "loss": 704286.9,
      "step": 180
    },
    {
      "epoch": 0.000767624041984995,
      "grad_norm": 2574691.5,
      "learning_rate": 3.8e-07,
      "loss": 545635.6,
      "step": 190
    },
    {
      "epoch": 0.0008080253073526263,
      "grad_norm": 21092594.0,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 728793.05,
      "step": 200
    },
    {
      "epoch": 0.0008484265727202576,
      "grad_norm": 2704460.5,
      "learning_rate": 4.2e-07,
      "loss": 467207.7,
      "step": 210
    },
    {
      "epoch": 0.0008888278380878889,
      "grad_norm": 4876784.0,
      "learning_rate": 4.4e-07,
      "loss": 369614.55,
      "step": 220
    },
    {
      "epoch": 0.0009292291034555202,
      "grad_norm": 10803276.0,
      "learning_rate": 4.6e-07,
      "loss": 428869.6,
      "step": 230
    },
    {
      "epoch": 0.0009696303688231516,
      "grad_norm": 5475451.0,
      "learning_rate": 4.8e-07,
      "loss": 404623.45,
      "step": 240
    },
    {
      "epoch": 0.0010100316341907828,
      "grad_norm": 3401077.25,
      "learning_rate": 5.000000000000001e-07,
      "loss": 606814.6,
      "step": 250
    },
    {
      "epoch": 0.001050432899558414,
      "grad_norm": 1785725.0,
      "learning_rate": 5.2e-07,
      "loss": 230520.125,
      "step": 260
    },
    {
      "epoch": 0.0010908341649260454,
      "grad_norm": 26237066.0,
      "learning_rate": 5.4e-07,
      "loss": 251476.375,
      "step": 270
    },
    {
      "epoch": 0.0011312354302936767,
      "grad_norm": 3433710.25,
      "learning_rate": 5.6e-07,
      "loss": 378523.225,
      "step": 280
    },
    {
      "epoch": 0.001171636695661308,
      "grad_norm": 3058920.5,
      "learning_rate": 5.8e-07,
      "loss": 118036.575,
      "step": 290
    },
    {
      "epoch": 0.0012120379610289394,
      "grad_norm": 2775939.25,
      "learning_rate": 6.000000000000001e-07,
      "loss": 128199.575,
      "step": 300
    },
    {
      "epoch": 0.0012524392263965707,
      "grad_norm": 2798658.75,
      "learning_rate": 6.2e-07,
      "loss": 145799.6125,
      "step": 310
    },
    {
      "epoch": 0.001292840491764202,
      "grad_norm": 3050425.75,
      "learning_rate": 6.4e-07,
      "loss": 170385.65,
      "step": 320
    },
    {
      "epoch": 0.0013332417571318333,
      "grad_norm": 483486.6875,
      "learning_rate": 6.6e-07,
      "loss": 79056.7,
      "step": 330
    },
    {
      "epoch": 0.0013736430224994647,
      "grad_norm": 903465.9375,
      "learning_rate": 6.8e-07,
      "loss": 69717.8562,
      "step": 340
    },
    {
      "epoch": 0.001414044287867096,
      "grad_norm": 166525.765625,
      "learning_rate": 7.000000000000001e-07,
      "loss": 29574.1,
      "step": 350
    },
    {
      "epoch": 0.0014544455532347273,
      "grad_norm": 83730.7109375,
      "learning_rate": 7.2e-07,
      "loss": 9268.3289,
      "step": 360
    },
    {
      "epoch": 0.0014948468186023586,
      "grad_norm": 13107.71484375,
      "learning_rate": 7.400000000000001e-07,
      "loss": 16814.9484,
      "step": 370
    },
    {
      "epoch": 0.00153524808396999,
      "grad_norm": 199187.046875,
      "learning_rate": 7.6e-07,
      "loss": 4728.9383,
      "step": 380
    },
    {
      "epoch": 0.0015756493493376213,
      "grad_norm": 209663.171875,
      "learning_rate": 7.8e-07,
      "loss": 3811.3047,
      "step": 390
    },
    {
      "epoch": 0.0016160506147052526,
      "grad_norm": 261596.96875,
      "learning_rate": 8.000000000000001e-07,
      "loss": 1873.8627,
      "step": 400
    },
    {
      "epoch": 0.001656451880072884,
      "grad_norm": 9995.125,
      "learning_rate": 8.200000000000001e-07,
      "loss": 1428.6535,
      "step": 410
    },
    {
      "epoch": 0.0016968531454405152,
      "grad_norm": 2573.125732421875,
      "learning_rate": 8.4e-07,
      "loss": 533.159,
      "step": 420
    },
    {
      "epoch": 0.0017372544108081465,
      "grad_norm": 547.4088134765625,
      "learning_rate": 8.6e-07,
      "loss": 702.8197,
      "step": 430
    },
    {
      "epoch": 0.0017776556761757779,
      "grad_norm": 734.5255126953125,
      "learning_rate": 8.8e-07,
      "loss": 485.3759,
      "step": 440
    },
    {
      "epoch": 0.0018180569415434092,
      "grad_norm": 5114.30615234375,
      "learning_rate": 9e-07,
      "loss": 545.9343,
      "step": 450
    },
    {
      "epoch": 0.0018584582069110405,
      "grad_norm": 1266.8800048828125,
      "learning_rate": 9.2e-07,
      "loss": 482.805,
      "step": 460
    },
    {
      "epoch": 0.0018988594722786718,
      "grad_norm": 5457.6689453125,
      "learning_rate": 9.400000000000001e-07,
      "loss": 550.8278,
      "step": 470
    },
    {
      "epoch": 0.0019392607376463031,
      "grad_norm": 1092.0926513671875,
      "learning_rate": 9.6e-07,
      "loss": 572.7061,
      "step": 480
    },
    {
      "epoch": 0.0019796620030139342,
      "grad_norm": 1708.60546875,
      "learning_rate": 9.8e-07,
      "loss": 447.9137,
      "step": 490
    },
    {
      "epoch": 0.0020200632683815656,
      "grad_norm": 833.2102661132812,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 452.3768,
      "step": 500
    },
    {
      "epoch": 0.002060464533749197,
      "grad_norm": 854.2012329101562,
      "learning_rate": 1.0200000000000002e-06,
      "loss": 453.9597,
      "step": 510
    },
    {
      "epoch": 0.002100865799116828,
      "grad_norm": 883.29052734375,
      "learning_rate": 1.04e-06,
      "loss": 399.3994,
      "step": 520
    },
    {
      "epoch": 0.0021412670644844595,
      "grad_norm": 876.2081909179688,
      "learning_rate": 1.06e-06,
      "loss": 508.956,
      "step": 530
    },
    {
      "epoch": 0.002181668329852091,
      "grad_norm": 1216.003662109375,
      "learning_rate": 1.08e-06,
      "loss": 279.6047,
      "step": 540
    },
    {
      "epoch": 0.002222069595219722,
      "grad_norm": 20951.376953125,
      "learning_rate": 1.1e-06,
      "loss": 523.4955,
      "step": 550
    },
    {
      "epoch": 0.0022624708605873535,
      "grad_norm": 6249.7998046875,
      "learning_rate": 1.12e-06,
      "loss": 373.9713,
      "step": 560
    },
    {
      "epoch": 0.002302872125954985,
      "grad_norm": 8211.34375,
      "learning_rate": 1.14e-06,
      "loss": 322.8237,
      "step": 570
    },
    {
      "epoch": 0.002343273391322616,
      "grad_norm": 1029.943359375,
      "learning_rate": 1.16e-06,
      "loss": 443.8321,
      "step": 580
    },
    {
      "epoch": 0.0023836746566902474,
      "grad_norm": 5415.279296875,
      "learning_rate": 1.18e-06,
      "loss": 580.9646,
      "step": 590
    },
    {
      "epoch": 0.0024240759220578788,
      "grad_norm": 683.0673217773438,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 322.4693,
      "step": 600
    },
    {
      "epoch": 0.00246447718742551,
      "grad_norm": 3990.54931640625,
      "learning_rate": 1.2200000000000002e-06,
      "loss": 543.5786,
      "step": 610
    },
    {
      "epoch": 0.0025048784527931414,
      "grad_norm": 771.86767578125,
      "learning_rate": 1.24e-06,
      "loss": 464.2772,
      "step": 620
    },
    {
      "epoch": 0.0025452797181607727,
      "grad_norm": 830.4238891601562,
      "learning_rate": 1.26e-06,
      "loss": 502.5308,
      "step": 630
    },
    {
      "epoch": 0.002585680983528404,
      "grad_norm": 4528.138671875,
      "learning_rate": 1.28e-06,
      "loss": 368.3498,
      "step": 640
    },
    {
      "epoch": 0.0026260822488960354,
      "grad_norm": 701.6304321289062,
      "learning_rate": 1.3e-06,
      "loss": 433.9283,
      "step": 650
    },
    {
      "epoch": 0.0026664835142636667,
      "grad_norm": 593.0372924804688,
      "learning_rate": 1.32e-06,
      "loss": 327.8368,
      "step": 660
    },
    {
      "epoch": 0.002706884779631298,
      "grad_norm": 718.9718627929688,
      "learning_rate": 1.34e-06,
      "loss": 363.8515,
      "step": 670
    },
    {
      "epoch": 0.0027472860449989293,
      "grad_norm": 1276.8353271484375,
      "learning_rate": 1.36e-06,
      "loss": 457.8593,
      "step": 680
    },
    {
      "epoch": 0.0027876873103665606,
      "grad_norm": 9622.8271484375,
      "learning_rate": 1.3800000000000001e-06,
      "loss": 448.6425,
      "step": 690
    },
    {
      "epoch": 0.002828088575734192,
      "grad_norm": 795.946044921875,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 370.567,
      "step": 700
    },
    {
      "epoch": 0.0028684898411018233,
      "grad_norm": 2605.2119140625,
      "learning_rate": 1.4200000000000002e-06,
      "loss": 399.8033,
      "step": 710
    },
    {
      "epoch": 0.0029088911064694546,
      "grad_norm": 3644.47607421875,
      "learning_rate": 1.44e-06,
      "loss": 519.6482,
      "step": 720
    },
    {
      "epoch": 0.002949292371837086,
      "grad_norm": 859.5913696289062,
      "learning_rate": 1.46e-06,
      "loss": 366.024,
      "step": 730
    },
    {
      "epoch": 0.0029896936372047172,
      "grad_norm": 9709.5673828125,
      "learning_rate": 1.4800000000000002e-06,
      "loss": 387.851,
      "step": 740
    },
    {
      "epoch": 0.0030300949025723486,
      "grad_norm": 1355.1983642578125,
      "learning_rate": 1.5e-06,
      "loss": 287.8213,
      "step": 750
    },
    {
      "epoch": 0.00307049616793998,
      "grad_norm": 1910.957275390625,
      "learning_rate": 1.52e-06,
      "loss": 263.4618,
      "step": 760
    },
    {
      "epoch": 0.003110897433307611,
      "grad_norm": 1562.11669921875,
      "learning_rate": 1.54e-06,
      "loss": 514.2329,
      "step": 770
    },
    {
      "epoch": 0.0031512986986752425,
      "grad_norm": 765.830078125,
      "learning_rate": 1.56e-06,
      "loss": 332.8841,
      "step": 780
    },
    {
      "epoch": 0.003191699964042874,
      "grad_norm": 13385.1689453125,
      "learning_rate": 1.5800000000000003e-06,
      "loss": 456.157,
      "step": 790
    },
    {
      "epoch": 0.003232101229410505,
      "grad_norm": 969.9119262695312,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 300.1112,
      "step": 800
    },
    {
      "epoch": 0.0032725024947781365,
      "grad_norm": 1840.109619140625,
      "learning_rate": 1.62e-06,
      "loss": 384.0655,
      "step": 810
    },
    {
      "epoch": 0.003312903760145768,
      "grad_norm": 1270.9483642578125,
      "learning_rate": 1.6400000000000002e-06,
      "loss": 355.3219,
      "step": 820
    },
    {
      "epoch": 0.003353305025513399,
      "grad_norm": 668.4849243164062,
      "learning_rate": 1.6600000000000002e-06,
      "loss": 247.1219,
      "step": 830
    },
    {
      "epoch": 0.0033937062908810304,
      "grad_norm": 1676.130615234375,
      "learning_rate": 1.68e-06,
      "loss": 312.9415,
      "step": 840
    },
    {
      "epoch": 0.0034341075562486618,
      "grad_norm": 480.2029724121094,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 388.1627,
      "step": 850
    },
    {
      "epoch": 0.003474508821616293,
      "grad_norm": 579.8500366210938,
      "learning_rate": 1.72e-06,
      "loss": 423.7644,
      "step": 860
    },
    {
      "epoch": 0.0035149100869839244,
      "grad_norm": 780.9789428710938,
      "learning_rate": 1.7399999999999999e-06,
      "loss": 335.4401,
      "step": 870
    },
    {
      "epoch": 0.0035553113523515557,
      "grad_norm": 500.0970153808594,
      "learning_rate": 1.76e-06,
      "loss": 375.4284,
      "step": 880
    },
    {
      "epoch": 0.003595712617719187,
      "grad_norm": 663.9608154296875,
      "learning_rate": 1.7800000000000001e-06,
      "loss": 432.7744,
      "step": 890
    },
    {
      "epoch": 0.0036361138830868184,
      "grad_norm": 738.2296142578125,
      "learning_rate": 1.8e-06,
      "loss": 330.4268,
      "step": 900
    },
    {
      "epoch": 0.0036765151484544497,
      "grad_norm": 888.3779907226562,
      "learning_rate": 1.8200000000000002e-06,
      "loss": 328.1051,
      "step": 910
    },
    {
      "epoch": 0.003716916413822081,
      "grad_norm": 681.369873046875,
      "learning_rate": 1.84e-06,
      "loss": 455.4454,
      "step": 920
    },
    {
      "epoch": 0.0037573176791897123,
      "grad_norm": 719.9152221679688,
      "learning_rate": 1.86e-06,
      "loss": 291.1318,
      "step": 930
    },
    {
      "epoch": 0.0037977189445573436,
      "grad_norm": 765.2400512695312,
      "learning_rate": 1.8800000000000002e-06,
      "loss": 453.5919,
      "step": 940
    },
    {
      "epoch": 0.003838120209924975,
      "grad_norm": 586.1494140625,
      "learning_rate": 1.9e-06,
      "loss": 338.6631,
      "step": 950
    },
    {
      "epoch": 0.0038785214752926063,
      "grad_norm": 559.566650390625,
      "learning_rate": 1.92e-06,
      "loss": 309.9115,
      "step": 960
    },
    {
      "epoch": 0.003918922740660237,
      "grad_norm": 2274.710205078125,
      "learning_rate": 1.94e-06,
      "loss": 366.6058,
      "step": 970
    },
    {
      "epoch": 0.0039593240060278685,
      "grad_norm": 2413.857666015625,
      "learning_rate": 1.96e-06,
      "loss": 323.5407,
      "step": 980
    },
    {
      "epoch": 0.0039997252713955,
      "grad_norm": 466.5931396484375,
      "learning_rate": 1.98e-06,
      "loss": 332.9158,
      "step": 990
    },
    {
      "epoch": 0.004040126536763131,
      "grad_norm": 1368.92626953125,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 309.8214,
      "step": 1000
    },
    {
      "epoch": 0.0040805278021307624,
      "grad_norm": 578.7489624023438,
      "learning_rate": 2.02e-06,
      "loss": 293.1464,
      "step": 1010
    },
    {
      "epoch": 0.004120929067498394,
      "grad_norm": 932.1323852539062,
      "learning_rate": 2.0400000000000004e-06,
      "loss": 382.6469,
      "step": 1020
    },
    {
      "epoch": 0.004161330332866025,
      "grad_norm": 861.5737915039062,
      "learning_rate": 2.06e-06,
      "loss": 394.3037,
      "step": 1030
    },
    {
      "epoch": 0.004201731598233656,
      "grad_norm": 540.3167114257812,
      "learning_rate": 2.08e-06,
      "loss": 356.1217,
      "step": 1040
    },
    {
      "epoch": 0.004242132863601288,
      "grad_norm": 645.2182006835938,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 375.9719,
      "step": 1050
    },
    {
      "epoch": 0.004282534128968919,
      "grad_norm": 3072.72607421875,
      "learning_rate": 2.12e-06,
      "loss": 385.9836,
      "step": 1060
    },
    {
      "epoch": 0.00432293539433655,
      "grad_norm": 1441.8131103515625,
      "learning_rate": 2.14e-06,
      "loss": 447.29,
      "step": 1070
    },
    {
      "epoch": 0.004363336659704182,
      "grad_norm": 531.906494140625,
      "learning_rate": 2.16e-06,
      "loss": 307.2454,
      "step": 1080
    },
    {
      "epoch": 0.004403737925071813,
      "grad_norm": 4712.17236328125,
      "learning_rate": 2.1800000000000003e-06,
      "loss": 599.3445,
      "step": 1090
    },
    {
      "epoch": 0.004444139190439444,
      "grad_norm": 543.140380859375,
      "learning_rate": 2.2e-06,
      "loss": 236.8443,
      "step": 1100
    },
    {
      "epoch": 0.004484540455807076,
      "grad_norm": 724.6696166992188,
      "learning_rate": 2.2200000000000003e-06,
      "loss": 402.6835,
      "step": 1110
    },
    {
      "epoch": 0.004524941721174707,
      "grad_norm": 890.041015625,
      "learning_rate": 2.24e-06,
      "loss": 335.9333,
      "step": 1120
    },
    {
      "epoch": 0.004565342986542338,
      "grad_norm": 517.4255981445312,
      "learning_rate": 2.26e-06,
      "loss": 377.6742,
      "step": 1130
    },
    {
      "epoch": 0.00460574425190997,
      "grad_norm": 1773.51025390625,
      "learning_rate": 2.28e-06,
      "loss": 387.5232,
      "step": 1140
    },
    {
      "epoch": 0.004646145517277601,
      "grad_norm": 1350.251220703125,
      "learning_rate": 2.3e-06,
      "loss": 430.9959,
      "step": 1150
    },
    {
      "epoch": 0.004686546782645232,
      "grad_norm": 1267.9136962890625,
      "learning_rate": 2.32e-06,
      "loss": 417.8642,
      "step": 1160
    },
    {
      "epoch": 0.0047269480480128636,
      "grad_norm": 1760.7581787109375,
      "learning_rate": 2.34e-06,
      "loss": 335.7049,
      "step": 1170
    },
    {
      "epoch": 0.004767349313380495,
      "grad_norm": 471.7587890625,
      "learning_rate": 2.36e-06,
      "loss": 321.8847,
      "step": 1180
    },
    {
      "epoch": 0.004807750578748126,
      "grad_norm": 1277.0931396484375,
      "learning_rate": 2.38e-06,
      "loss": 339.8787,
      "step": 1190
    },
    {
      "epoch": 0.0048481518441157575,
      "grad_norm": 766.57568359375,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 343.6708,
      "step": 1200
    },
    {
      "epoch": 0.004888553109483389,
      "grad_norm": 733.90234375,
      "learning_rate": 2.42e-06,
      "loss": 284.1223,
      "step": 1210
    },
    {
      "epoch": 0.00492895437485102,
      "grad_norm": 705.3512573242188,
      "learning_rate": 2.4400000000000004e-06,
      "loss": 290.2316,
      "step": 1220
    },
    {
      "epoch": 0.0049693556402186515,
      "grad_norm": 1195.5079345703125,
      "learning_rate": 2.46e-06,
      "loss": 351.2376,
      "step": 1230
    },
    {
      "epoch": 0.005009756905586283,
      "grad_norm": 1140.5634765625,
      "learning_rate": 2.48e-06,
      "loss": 364.3263,
      "step": 1240
    },
    {
      "epoch": 0.005050158170953914,
      "grad_norm": 1419.2325439453125,
      "learning_rate": 2.5e-06,
      "loss": 420.2925,
      "step": 1250
    },
    {
      "epoch": 0.0050905594363215454,
      "grad_norm": 945.1781616210938,
      "learning_rate": 2.52e-06,
      "loss": 377.5975,
      "step": 1260
    },
    {
      "epoch": 0.005130960701689177,
      "grad_norm": 1156.38818359375,
      "learning_rate": 2.54e-06,
      "loss": 438.8611,
      "step": 1270
    },
    {
      "epoch": 0.005171361967056808,
      "grad_norm": 803.0604858398438,
      "learning_rate": 2.56e-06,
      "loss": 275.1979,
      "step": 1280
    },
    {
      "epoch": 0.005211763232424439,
      "grad_norm": 4027.070556640625,
      "learning_rate": 2.5800000000000003e-06,
      "loss": 417.5213,
      "step": 1290
    },
    {
      "epoch": 0.005252164497792071,
      "grad_norm": 762.65771484375,
      "learning_rate": 2.6e-06,
      "loss": 338.176,
      "step": 1300
    },
    {
      "epoch": 0.005292565763159702,
      "grad_norm": 514.9345092773438,
      "learning_rate": 2.6200000000000003e-06,
      "loss": 360.1529,
      "step": 1310
    },
    {
      "epoch": 0.005332967028527333,
      "grad_norm": 0.0,
      "learning_rate": 2.64e-06,
      "loss": 437.6661,
      "step": 1320
    },
    {
      "epoch": 0.005373368293894965,
      "grad_norm": 964.8443603515625,
      "learning_rate": 2.66e-06,
      "loss": 348.5103,
      "step": 1330
    },
    {
      "epoch": 0.005413769559262596,
      "grad_norm": 1033.2138671875,
      "learning_rate": 2.68e-06,
      "loss": 351.3489,
      "step": 1340
    },
    {
      "epoch": 0.005454170824630227,
      "grad_norm": 621.0077514648438,
      "learning_rate": 2.7e-06,
      "loss": 297.7263,
      "step": 1350
    },
    {
      "epoch": 0.005494572089997859,
      "grad_norm": 670.5285034179688,
      "learning_rate": 2.72e-06,
      "loss": 266.751,
      "step": 1360
    },
    {
      "epoch": 0.00553497335536549,
      "grad_norm": 1063.011962890625,
      "learning_rate": 2.74e-06,
      "loss": 388.5708,
      "step": 1370
    },
    {
      "epoch": 0.005575374620733121,
      "grad_norm": 1352.453369140625,
      "learning_rate": 2.7600000000000003e-06,
      "loss": 413.9111,
      "step": 1380
    },
    {
      "epoch": 0.005615775886100753,
      "grad_norm": 1039.40966796875,
      "learning_rate": 2.78e-06,
      "loss": 425.6994,
      "step": 1390
    },
    {
      "epoch": 0.005656177151468384,
      "grad_norm": 852.0213012695312,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 379.7008,
      "step": 1400
    },
    {
      "epoch": 0.005696578416836015,
      "grad_norm": 560.5407104492188,
      "learning_rate": 2.82e-06,
      "loss": 373.5774,
      "step": 1410
    },
    {
      "epoch": 0.0057369796822036466,
      "grad_norm": 2221.54248046875,
      "learning_rate": 2.8400000000000003e-06,
      "loss": 377.4932,
      "step": 1420
    },
    {
      "epoch": 0.005777380947571278,
      "grad_norm": 986.341064453125,
      "learning_rate": 2.86e-06,
      "loss": 381.8052,
      "step": 1430
    },
    {
      "epoch": 0.005817782212938909,
      "grad_norm": 2288.12060546875,
      "learning_rate": 2.88e-06,
      "loss": 396.2358,
      "step": 1440
    },
    {
      "epoch": 0.0058581834783065405,
      "grad_norm": 506.0441589355469,
      "learning_rate": 2.9e-06,
      "loss": 306.8475,
      "step": 1450
    },
    {
      "epoch": 0.005898584743674172,
      "grad_norm": 1451.433837890625,
      "learning_rate": 2.92e-06,
      "loss": 421.7085,
      "step": 1460
    },
    {
      "epoch": 0.005938986009041803,
      "grad_norm": 410.0963439941406,
      "learning_rate": 2.9400000000000002e-06,
      "loss": 245.0907,
      "step": 1470
    },
    {
      "epoch": 0.0059793872744094345,
      "grad_norm": 644.2957153320312,
      "learning_rate": 2.9600000000000005e-06,
      "loss": 256.7748,
      "step": 1480
    },
    {
      "epoch": 0.006019788539777066,
      "grad_norm": 587.813232421875,
      "learning_rate": 2.9800000000000003e-06,
      "loss": 307.0451,
      "step": 1490
    },
    {
      "epoch": 0.006060189805144697,
      "grad_norm": 661.851806640625,
      "learning_rate": 3e-06,
      "loss": 240.7283,
      "step": 1500
    },
    {
      "epoch": 0.0061005910705123284,
      "grad_norm": 9922.1162109375,
      "learning_rate": 3.0200000000000003e-06,
      "loss": 314.2891,
      "step": 1510
    },
    {
      "epoch": 0.00614099233587996,
      "grad_norm": 681.8471069335938,
      "learning_rate": 3.04e-06,
      "loss": 304.0082,
      "step": 1520
    },
    {
      "epoch": 0.006181393601247591,
      "grad_norm": 1679.9188232421875,
      "learning_rate": 3.06e-06,
      "loss": 316.8349,
      "step": 1530
    },
    {
      "epoch": 0.006221794866615222,
      "grad_norm": 0.0,
      "learning_rate": 3.08e-06,
      "loss": 273.9679,
      "step": 1540
    },
    {
      "epoch": 0.006262196131982854,
      "grad_norm": 934.2025756835938,
      "learning_rate": 3.1e-06,
      "loss": 281.7329,
      "step": 1550
    },
    {
      "epoch": 0.006302597397350485,
      "grad_norm": 1820.7791748046875,
      "learning_rate": 3.12e-06,
      "loss": 292.2456,
      "step": 1560
    },
    {
      "epoch": 0.006342998662718116,
      "grad_norm": 1122.6767578125,
      "learning_rate": 3.14e-06,
      "loss": 327.0943,
      "step": 1570
    },
    {
      "epoch": 0.006383399928085748,
      "grad_norm": 3407.740234375,
      "learning_rate": 3.1600000000000007e-06,
      "loss": 282.118,
      "step": 1580
    },
    {
      "epoch": 0.006423801193453379,
      "grad_norm": 556.7740478515625,
      "learning_rate": 3.1800000000000005e-06,
      "loss": 332.0084,
      "step": 1590
    },
    {
      "epoch": 0.00646420245882101,
      "grad_norm": 1229.8035888671875,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 344.9013,
      "step": 1600
    },
    {
      "epoch": 0.006504603724188642,
      "grad_norm": 949.2359008789062,
      "learning_rate": 3.22e-06,
      "loss": 371.0438,
      "step": 1610
    },
    {
      "epoch": 0.006545004989556273,
      "grad_norm": 1869.7620849609375,
      "learning_rate": 3.24e-06,
      "loss": 388.4585,
      "step": 1620
    },
    {
      "epoch": 0.006585406254923904,
      "grad_norm": 527.3277587890625,
      "learning_rate": 3.2599999999999997e-06,
      "loss": 372.7793,
      "step": 1630
    },
    {
      "epoch": 0.006625807520291536,
      "grad_norm": 1427.3751220703125,
      "learning_rate": 3.2800000000000004e-06,
      "loss": 386.9533,
      "step": 1640
    },
    {
      "epoch": 0.006666208785659167,
      "grad_norm": 1839.386474609375,
      "learning_rate": 3.3e-06,
      "loss": 414.8817,
      "step": 1650
    },
    {
      "epoch": 0.006706610051026798,
      "grad_norm": 767.32373046875,
      "learning_rate": 3.3200000000000004e-06,
      "loss": 262.8998,
      "step": 1660
    },
    {
      "epoch": 0.0067470113163944296,
      "grad_norm": 487.2662353515625,
      "learning_rate": 3.34e-06,
      "loss": 349.8497,
      "step": 1670
    },
    {
      "epoch": 0.006787412581762061,
      "grad_norm": 1609.6903076171875,
      "learning_rate": 3.36e-06,
      "loss": 328.737,
      "step": 1680
    },
    {
      "epoch": 0.006827813847129692,
      "grad_norm": 824.4249267578125,
      "learning_rate": 3.38e-06,
      "loss": 285.9903,
      "step": 1690
    },
    {
      "epoch": 0.0068682151124973235,
      "grad_norm": 3939.00244140625,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 360.8893,
      "step": 1700
    },
    {
      "epoch": 0.006908616377864955,
      "grad_norm": 611.8134155273438,
      "learning_rate": 3.4200000000000003e-06,
      "loss": 316.7729,
      "step": 1710
    },
    {
      "epoch": 0.006949017643232586,
      "grad_norm": 3673.492919921875,
      "learning_rate": 3.44e-06,
      "loss": 446.8052,
      "step": 1720
    },
    {
      "epoch": 0.0069894189086002175,
      "grad_norm": 580.3406982421875,
      "learning_rate": 3.46e-06,
      "loss": 306.8866,
      "step": 1730
    },
    {
      "epoch": 0.007029820173967849,
      "grad_norm": 797.6139526367188,
      "learning_rate": 3.4799999999999997e-06,
      "loss": 360.2713,
      "step": 1740
    },
    {
      "epoch": 0.00707022143933548,
      "grad_norm": 1184.318115234375,
      "learning_rate": 3.5000000000000004e-06,
      "loss": 360.695,
      "step": 1750
    },
    {
      "epoch": 0.0071106227047031114,
      "grad_norm": 659.1159057617188,
      "learning_rate": 3.52e-06,
      "loss": 393.7832,
      "step": 1760
    },
    {
      "epoch": 0.007151023970070743,
      "grad_norm": 611.4116821289062,
      "learning_rate": 3.5400000000000004e-06,
      "loss": 302.0126,
      "step": 1770
    },
    {
      "epoch": 0.007191425235438374,
      "grad_norm": 2148.29541015625,
      "learning_rate": 3.5600000000000002e-06,
      "loss": 411.9108,
      "step": 1780
    },
    {
      "epoch": 0.007231826500806005,
      "grad_norm": 635.7216796875,
      "learning_rate": 3.58e-06,
      "loss": 236.3176,
      "step": 1790
    },
    {
      "epoch": 0.007272227766173637,
      "grad_norm": 367.6564025878906,
      "learning_rate": 3.6e-06,
      "loss": 293.0558,
      "step": 1800
    },
    {
      "epoch": 0.007312629031541268,
      "grad_norm": 712.876220703125,
      "learning_rate": 3.6200000000000005e-06,
      "loss": 349.5894,
      "step": 1810
    },
    {
      "epoch": 0.007353030296908899,
      "grad_norm": 505.4992980957031,
      "learning_rate": 3.6400000000000003e-06,
      "loss": 300.7681,
      "step": 1820
    },
    {
      "epoch": 0.007393431562276531,
      "grad_norm": 852.1588745117188,
      "learning_rate": 3.66e-06,
      "loss": 408.5499,
      "step": 1830
    },
    {
      "epoch": 0.007433832827644162,
      "grad_norm": 1410.185791015625,
      "learning_rate": 3.68e-06,
      "loss": 253.6884,
      "step": 1840
    },
    {
      "epoch": 0.007474234093011793,
      "grad_norm": 889.8245849609375,
      "learning_rate": 3.7e-06,
      "loss": 408.2236,
      "step": 1850
    },
    {
      "epoch": 0.007514635358379425,
      "grad_norm": 906.4953002929688,
      "learning_rate": 3.72e-06,
      "loss": 397.8984,
      "step": 1860
    },
    {
      "epoch": 0.007555036623747056,
      "grad_norm": 0.0,
      "learning_rate": 3.7400000000000006e-06,
      "loss": 238.5037,
      "step": 1870
    },
    {
      "epoch": 0.007595437889114687,
      "grad_norm": 13512.4267578125,
      "learning_rate": 3.7600000000000004e-06,
      "loss": 348.5812,
      "step": 1880
    },
    {
      "epoch": 0.007635839154482319,
      "grad_norm": 4588.23193359375,
      "learning_rate": 3.7800000000000002e-06,
      "loss": 364.4923,
      "step": 1890
    },
    {
      "epoch": 0.00767624041984995,
      "grad_norm": 803.103759765625,
      "learning_rate": 3.8e-06,
      "loss": 372.9157,
      "step": 1900
    },
    {
      "epoch": 0.007716641685217581,
      "grad_norm": 1621.907958984375,
      "learning_rate": 3.82e-06,
      "loss": 360.8791,
      "step": 1910
    },
    {
      "epoch": 0.0077570429505852126,
      "grad_norm": 592.7820434570312,
      "learning_rate": 3.84e-06,
      "loss": 375.443,
      "step": 1920
    },
    {
      "epoch": 0.007797444215952844,
      "grad_norm": 2514.99462890625,
      "learning_rate": 3.86e-06,
      "loss": 332.9206,
      "step": 1930
    },
    {
      "epoch": 0.007837845481320474,
      "grad_norm": 1829.7589111328125,
      "learning_rate": 3.88e-06,
      "loss": 442.6601,
      "step": 1940
    },
    {
      "epoch": 0.007878246746688106,
      "grad_norm": 897.420166015625,
      "learning_rate": 3.9e-06,
      "loss": 288.4365,
      "step": 1950
    },
    {
      "epoch": 0.007918648012055737,
      "grad_norm": 1314.1402587890625,
      "learning_rate": 3.92e-06,
      "loss": 334.6603,
      "step": 1960
    },
    {
      "epoch": 0.007959049277423368,
      "grad_norm": 2495.09130859375,
      "learning_rate": 3.9399999999999995e-06,
      "loss": 434.7513,
      "step": 1970
    },
    {
      "epoch": 0.007999450542791,
      "grad_norm": 402.0351867675781,
      "learning_rate": 3.96e-06,
      "loss": 307.8936,
      "step": 1980
    },
    {
      "epoch": 0.008039851808158631,
      "grad_norm": 751.7537841796875,
      "learning_rate": 3.98e-06,
      "loss": 308.6638,
      "step": 1990
    },
    {
      "epoch": 0.008080253073526262,
      "grad_norm": 604.2171630859375,
      "learning_rate": 4.000000000000001e-06,
      "loss": 392.2409,
      "step": 2000
    },
    {
      "epoch": 0.008120654338893894,
      "grad_norm": 1126.428955078125,
      "learning_rate": 4.0200000000000005e-06,
      "loss": 318.5683,
      "step": 2010
    },
    {
      "epoch": 0.008161055604261525,
      "grad_norm": 4945.88037109375,
      "learning_rate": 4.04e-06,
      "loss": 311.324,
      "step": 2020
    },
    {
      "epoch": 0.008201456869629156,
      "grad_norm": 745.3814697265625,
      "learning_rate": 4.06e-06,
      "loss": 381.683,
      "step": 2030
    },
    {
      "epoch": 0.008241858134996788,
      "grad_norm": 556.3217163085938,
      "learning_rate": 4.080000000000001e-06,
      "loss": 335.636,
      "step": 2040
    },
    {
      "epoch": 0.008282259400364419,
      "grad_norm": 1249.5528564453125,
      "learning_rate": 4.1000000000000006e-06,
      "loss": 216.6521,
      "step": 2050
    },
    {
      "epoch": 0.00832266066573205,
      "grad_norm": 4082.633544921875,
      "learning_rate": 4.12e-06,
      "loss": 422.9517,
      "step": 2060
    },
    {
      "epoch": 0.008363061931099681,
      "grad_norm": 1134.3583984375,
      "learning_rate": 4.14e-06,
      "loss": 368.1808,
      "step": 2070
    },
    {
      "epoch": 0.008403463196467313,
      "grad_norm": 473.27655029296875,
      "learning_rate": 4.16e-06,
      "loss": 323.5618,
      "step": 2080
    },
    {
      "epoch": 0.008443864461834944,
      "grad_norm": 1147.6612548828125,
      "learning_rate": 4.18e-06,
      "loss": 358.5086,
      "step": 2090
    },
    {
      "epoch": 0.008484265727202575,
      "grad_norm": 716.1121826171875,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 289.7142,
      "step": 2100
    },
    {
      "epoch": 0.008524666992570207,
      "grad_norm": 595.6019287109375,
      "learning_rate": 4.22e-06,
      "loss": 249.1531,
      "step": 2110
    },
    {
      "epoch": 0.008565068257937838,
      "grad_norm": 1831.8653564453125,
      "learning_rate": 4.24e-06,
      "loss": 336.2978,
      "step": 2120
    },
    {
      "epoch": 0.00860546952330547,
      "grad_norm": 549.8616943359375,
      "learning_rate": 4.26e-06,
      "loss": 310.2739,
      "step": 2130
    },
    {
      "epoch": 0.0086458707886731,
      "grad_norm": 888.9124755859375,
      "learning_rate": 4.28e-06,
      "loss": 372.0684,
      "step": 2140
    },
    {
      "epoch": 0.008686272054040732,
      "grad_norm": 654.9121704101562,
      "learning_rate": 4.2999999999999995e-06,
      "loss": 218.5617,
      "step": 2150
    },
    {
      "epoch": 0.008726673319408363,
      "grad_norm": 1140.73486328125,
      "learning_rate": 4.32e-06,
      "loss": 396.7231,
      "step": 2160
    },
    {
      "epoch": 0.008767074584775995,
      "grad_norm": 818.670166015625,
      "learning_rate": 4.34e-06,
      "loss": 335.7865,
      "step": 2170
    },
    {
      "epoch": 0.008807475850143626,
      "grad_norm": 792.7094116210938,
      "learning_rate": 4.360000000000001e-06,
      "loss": 321.234,
      "step": 2180
    },
    {
      "epoch": 0.008847877115511257,
      "grad_norm": 2842.64697265625,
      "learning_rate": 4.38e-06,
      "loss": 247.4936,
      "step": 2190
    },
    {
      "epoch": 0.008888278380878889,
      "grad_norm": 756.8375244140625,
      "learning_rate": 4.4e-06,
      "loss": 358.118,
      "step": 2200
    },
    {
      "epoch": 0.00892867964624652,
      "grad_norm": 447.1920166015625,
      "learning_rate": 4.420000000000001e-06,
      "loss": 243.4523,
      "step": 2210
    },
    {
      "epoch": 0.008969080911614151,
      "grad_norm": 598.7188110351562,
      "learning_rate": 4.440000000000001e-06,
      "loss": 317.7062,
      "step": 2220
    },
    {
      "epoch": 0.009009482176981783,
      "grad_norm": 782.6261596679688,
      "learning_rate": 4.4600000000000005e-06,
      "loss": 423.8147,
      "step": 2230
    },
    {
      "epoch": 0.009049883442349414,
      "grad_norm": 650.9381103515625,
      "learning_rate": 4.48e-06,
      "loss": 372.274,
      "step": 2240
    },
    {
      "epoch": 0.009090284707717045,
      "grad_norm": 911.272216796875,
      "learning_rate": 4.5e-06,
      "loss": 309.0599,
      "step": 2250
    },
    {
      "epoch": 0.009130685973084677,
      "grad_norm": 1821.9324951171875,
      "learning_rate": 4.52e-06,
      "loss": 291.9244,
      "step": 2260
    },
    {
      "epoch": 0.009171087238452308,
      "grad_norm": 1428.2496337890625,
      "learning_rate": 4.540000000000001e-06,
      "loss": 299.4853,
      "step": 2270
    },
    {
      "epoch": 0.00921148850381994,
      "grad_norm": 474.319580078125,
      "learning_rate": 4.56e-06,
      "loss": 284.1177,
      "step": 2280
    },
    {
      "epoch": 0.00925188976918757,
      "grad_norm": 521.900146484375,
      "learning_rate": 4.58e-06,
      "loss": 305.8711,
      "step": 2290
    },
    {
      "epoch": 0.009292291034555202,
      "grad_norm": 742.1409301757812,
      "learning_rate": 4.6e-06,
      "loss": 233.3259,
      "step": 2300
    },
    {
      "epoch": 0.009332692299922833,
      "grad_norm": 702.3360595703125,
      "learning_rate": 4.62e-06,
      "loss": 351.8805,
      "step": 2310
    },
    {
      "epoch": 0.009373093565290464,
      "grad_norm": 969.7785034179688,
      "learning_rate": 4.64e-06,
      "loss": 273.1551,
      "step": 2320
    },
    {
      "epoch": 0.009413494830658096,
      "grad_norm": 7805.5947265625,
      "learning_rate": 4.66e-06,
      "loss": 303.4039,
      "step": 2330
    },
    {
      "epoch": 0.009453896096025727,
      "grad_norm": 683.2047119140625,
      "learning_rate": 4.68e-06,
      "loss": 299.0721,
      "step": 2340
    },
    {
      "epoch": 0.009494297361393358,
      "grad_norm": 433.588134765625,
      "learning_rate": 4.7e-06,
      "loss": 275.1868,
      "step": 2350
    },
    {
      "epoch": 0.00953469862676099,
      "grad_norm": 970.4158325195312,
      "learning_rate": 4.72e-06,
      "loss": 508.6281,
      "step": 2360
    },
    {
      "epoch": 0.009575099892128621,
      "grad_norm": 1364.8466796875,
      "learning_rate": 4.74e-06,
      "loss": 333.4644,
      "step": 2370
    },
    {
      "epoch": 0.009615501157496252,
      "grad_norm": 1310.9354248046875,
      "learning_rate": 4.76e-06,
      "loss": 271.3296,
      "step": 2380
    },
    {
      "epoch": 0.009655902422863884,
      "grad_norm": 551.7647094726562,
      "learning_rate": 4.780000000000001e-06,
      "loss": 339.265,
      "step": 2390
    },
    {
      "epoch": 0.009696303688231515,
      "grad_norm": 542.0902709960938,
      "learning_rate": 4.800000000000001e-06,
      "loss": 375.706,
      "step": 2400
    },
    {
      "epoch": 0.009736704953599146,
      "grad_norm": 517.1104736328125,
      "learning_rate": 4.8200000000000004e-06,
      "loss": 183.125,
      "step": 2410
    },
    {
      "epoch": 0.009777106218966778,
      "grad_norm": 1165.0673828125,
      "learning_rate": 4.84e-06,
      "loss": 303.7644,
      "step": 2420
    },
    {
      "epoch": 0.009817507484334409,
      "grad_norm": 4149.34130859375,
      "learning_rate": 4.86e-06,
      "loss": 361.1945,
      "step": 2430
    },
    {
      "epoch": 0.00985790874970204,
      "grad_norm": 957.4842529296875,
      "learning_rate": 4.880000000000001e-06,
      "loss": 238.9853,
      "step": 2440
    },
    {
      "epoch": 0.009898310015069672,
      "grad_norm": 493.18341064453125,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 253.369,
      "step": 2450
    },
    {
      "epoch": 0.009938711280437303,
      "grad_norm": 913.3916015625,
      "learning_rate": 4.92e-06,
      "loss": 197.3718,
      "step": 2460
    },
    {
      "epoch": 0.009979112545804934,
      "grad_norm": 641.3722534179688,
      "learning_rate": 4.94e-06,
      "loss": 359.7244,
      "step": 2470
    },
    {
      "epoch": 0.010019513811172566,
      "grad_norm": 805.710693359375,
      "learning_rate": 4.96e-06,
      "loss": 314.5141,
      "step": 2480
    },
    {
      "epoch": 0.010059915076540197,
      "grad_norm": 979.5321655273438,
      "learning_rate": 4.98e-06,
      "loss": 402.7761,
      "step": 2490
    },
    {
      "epoch": 0.010100316341907828,
      "grad_norm": 434.0482177734375,
      "learning_rate": 5e-06,
      "loss": 190.8965,
      "step": 2500
    },
    {
      "epoch": 0.01014071760727546,
      "grad_norm": 907.0119018554688,
      "learning_rate": 5.02e-06,
      "loss": 269.7839,
      "step": 2510
    },
    {
      "epoch": 0.010181118872643091,
      "grad_norm": 363.0292053222656,
      "learning_rate": 5.04e-06,
      "loss": 258.706,
      "step": 2520
    },
    {
      "epoch": 0.010221520138010722,
      "grad_norm": 794.2672729492188,
      "learning_rate": 5.06e-06,
      "loss": 220.4875,
      "step": 2530
    },
    {
      "epoch": 0.010261921403378354,
      "grad_norm": 956.221923828125,
      "learning_rate": 5.08e-06,
      "loss": 283.6568,
      "step": 2540
    },
    {
      "epoch": 0.010302322668745985,
      "grad_norm": 779.488037109375,
      "learning_rate": 5.1e-06,
      "loss": 356.618,
      "step": 2550
    },
    {
      "epoch": 0.010342723934113616,
      "grad_norm": 505.3084716796875,
      "learning_rate": 5.12e-06,
      "loss": 195.6782,
      "step": 2560
    },
    {
      "epoch": 0.010383125199481247,
      "grad_norm": 1260.047119140625,
      "learning_rate": 5.140000000000001e-06,
      "loss": 321.1118,
      "step": 2570
    },
    {
      "epoch": 0.010423526464848879,
      "grad_norm": 776.1239624023438,
      "learning_rate": 5.1600000000000006e-06,
      "loss": 303.4421,
      "step": 2580
    },
    {
      "epoch": 0.01046392773021651,
      "grad_norm": 899.60009765625,
      "learning_rate": 5.18e-06,
      "loss": 267.6064,
      "step": 2590
    },
    {
      "epoch": 0.010504328995584141,
      "grad_norm": 1198.875732421875,
      "learning_rate": 5.2e-06,
      "loss": 311.9983,
      "step": 2600
    },
    {
      "epoch": 0.010544730260951773,
      "grad_norm": 2578.52734375,
      "learning_rate": 5.220000000000001e-06,
      "loss": 285.3186,
      "step": 2610
    },
    {
      "epoch": 0.010585131526319404,
      "grad_norm": 613.3502807617188,
      "learning_rate": 5.240000000000001e-06,
      "loss": 407.8097,
      "step": 2620
    },
    {
      "epoch": 0.010625532791687035,
      "grad_norm": 724.2944946289062,
      "learning_rate": 5.2600000000000005e-06,
      "loss": 375.3231,
      "step": 2630
    },
    {
      "epoch": 0.010665934057054667,
      "grad_norm": 625.8546752929688,
      "learning_rate": 5.28e-06,
      "loss": 337.0123,
      "step": 2640
    },
    {
      "epoch": 0.010706335322422298,
      "grad_norm": 2968.8515625,
      "learning_rate": 5.3e-06,
      "loss": 279.1451,
      "step": 2650
    },
    {
      "epoch": 0.01074673658778993,
      "grad_norm": 507.51885986328125,
      "learning_rate": 5.32e-06,
      "loss": 214.3604,
      "step": 2660
    },
    {
      "epoch": 0.01078713785315756,
      "grad_norm": 738.8612060546875,
      "learning_rate": 5.3400000000000005e-06,
      "loss": 332.2985,
      "step": 2670
    },
    {
      "epoch": 0.010827539118525192,
      "grad_norm": 684.1400146484375,
      "learning_rate": 5.36e-06,
      "loss": 330.1413,
      "step": 2680
    },
    {
      "epoch": 0.010867940383892823,
      "grad_norm": 6400.771484375,
      "learning_rate": 5.38e-06,
      "loss": 344.3884,
      "step": 2690
    },
    {
      "epoch": 0.010908341649260455,
      "grad_norm": 986.3028564453125,
      "learning_rate": 5.4e-06,
      "loss": 354.2751,
      "step": 2700
    },
    {
      "epoch": 0.010948742914628086,
      "grad_norm": 933.0596313476562,
      "learning_rate": 5.42e-06,
      "loss": 367.7645,
      "step": 2710
    },
    {
      "epoch": 0.010989144179995717,
      "grad_norm": 956.349365234375,
      "learning_rate": 5.44e-06,
      "loss": 293.7472,
      "step": 2720
    },
    {
      "epoch": 0.011029545445363349,
      "grad_norm": 473.7074890136719,
      "learning_rate": 5.46e-06,
      "loss": 281.8573,
      "step": 2730
    },
    {
      "epoch": 0.01106994671073098,
      "grad_norm": 1302.126953125,
      "learning_rate": 5.48e-06,
      "loss": 275.9536,
      "step": 2740
    },
    {
      "epoch": 0.011110347976098611,
      "grad_norm": 557.58203125,
      "learning_rate": 5.500000000000001e-06,
      "loss": 205.2503,
      "step": 2750
    },
    {
      "epoch": 0.011150749241466243,
      "grad_norm": 3368.84228515625,
      "learning_rate": 5.5200000000000005e-06,
      "loss": 328.5797,
      "step": 2760
    },
    {
      "epoch": 0.011191150506833874,
      "grad_norm": 1747.6932373046875,
      "learning_rate": 5.54e-06,
      "loss": 246.7141,
      "step": 2770
    },
    {
      "epoch": 0.011231551772201505,
      "grad_norm": 871.47802734375,
      "learning_rate": 5.56e-06,
      "loss": 261.7437,
      "step": 2780
    },
    {
      "epoch": 0.011271953037569137,
      "grad_norm": 1074.76416015625,
      "learning_rate": 5.580000000000001e-06,
      "loss": 292.0746,
      "step": 2790
    },
    {
      "epoch": 0.011312354302936768,
      "grad_norm": 575.856689453125,
      "learning_rate": 5.600000000000001e-06,
      "loss": 247.1602,
      "step": 2800
    },
    {
      "epoch": 0.0113527555683044,
      "grad_norm": 847.0436401367188,
      "learning_rate": 5.62e-06,
      "loss": 244.562,
      "step": 2810
    },
    {
      "epoch": 0.01139315683367203,
      "grad_norm": 479.6651306152344,
      "learning_rate": 5.64e-06,
      "loss": 196.8626,
      "step": 2820
    },
    {
      "epoch": 0.011433558099039662,
      "grad_norm": 1245.398681640625,
      "learning_rate": 5.66e-06,
      "loss": 338.9747,
      "step": 2830
    },
    {
      "epoch": 0.011473959364407293,
      "grad_norm": 502.24951171875,
      "learning_rate": 5.680000000000001e-06,
      "loss": 243.1993,
      "step": 2840
    },
    {
      "epoch": 0.011514360629774924,
      "grad_norm": 790.6267700195312,
      "learning_rate": 5.7000000000000005e-06,
      "loss": 289.827,
      "step": 2850
    },
    {
      "epoch": 0.011554761895142556,
      "grad_norm": 996.7022094726562,
      "learning_rate": 5.72e-06,
      "loss": 285.2487,
      "step": 2860
    },
    {
      "epoch": 0.011595163160510187,
      "grad_norm": 1381.54541015625,
      "learning_rate": 5.74e-06,
      "loss": 299.6274,
      "step": 2870
    },
    {
      "epoch": 0.011635564425877818,
      "grad_norm": 674.7173461914062,
      "learning_rate": 5.76e-06,
      "loss": 212.4647,
      "step": 2880
    },
    {
      "epoch": 0.01167596569124545,
      "grad_norm": 1115.6590576171875,
      "learning_rate": 5.78e-06,
      "loss": 350.1185,
      "step": 2890
    },
    {
      "epoch": 0.011716366956613081,
      "grad_norm": 967.9352416992188,
      "learning_rate": 5.8e-06,
      "loss": 303.8514,
      "step": 2900
    },
    {
      "epoch": 0.011756768221980712,
      "grad_norm": 826.4132690429688,
      "learning_rate": 5.82e-06,
      "loss": 390.5727,
      "step": 2910
    },
    {
      "epoch": 0.011797169487348344,
      "grad_norm": 554.1842651367188,
      "learning_rate": 5.84e-06,
      "loss": 228.6073,
      "step": 2920
    },
    {
      "epoch": 0.011837570752715975,
      "grad_norm": 555.0242919921875,
      "learning_rate": 5.86e-06,
      "loss": 237.2674,
      "step": 2930
    },
    {
      "epoch": 0.011877972018083606,
      "grad_norm": 554.800048828125,
      "learning_rate": 5.8800000000000005e-06,
      "loss": 307.6302,
      "step": 2940
    },
    {
      "epoch": 0.011918373283451238,
      "grad_norm": 7435.06005859375,
      "learning_rate": 5.9e-06,
      "loss": 363.9746,
      "step": 2950
    },
    {
      "epoch": 0.011958774548818869,
      "grad_norm": 3039.5869140625,
      "learning_rate": 5.920000000000001e-06,
      "loss": 294.7946,
      "step": 2960
    },
    {
      "epoch": 0.0119991758141865,
      "grad_norm": 9456.8857421875,
      "learning_rate": 5.940000000000001e-06,
      "loss": 348.9202,
      "step": 2970
    },
    {
      "epoch": 0.012039577079554132,
      "grad_norm": 833.540771484375,
      "learning_rate": 5.9600000000000005e-06,
      "loss": 302.1644,
      "step": 2980
    },
    {
      "epoch": 0.012079978344921763,
      "grad_norm": 1317.080810546875,
      "learning_rate": 5.98e-06,
      "loss": 346.3184,
      "step": 2990
    },
    {
      "epoch": 0.012120379610289394,
      "grad_norm": 546.1310424804688,
      "learning_rate": 6e-06,
      "loss": 185.3799,
      "step": 3000
    },
    {
      "epoch": 0.012160780875657026,
      "grad_norm": 1277.5128173828125,
      "learning_rate": 6.02e-06,
      "loss": 271.2112,
      "step": 3010
    },
    {
      "epoch": 0.012201182141024657,
      "grad_norm": 651.244384765625,
      "learning_rate": 6.040000000000001e-06,
      "loss": 206.7404,
      "step": 3020
    },
    {
      "epoch": 0.012241583406392288,
      "grad_norm": 886.7020874023438,
      "learning_rate": 6.0600000000000004e-06,
      "loss": 309.8574,
      "step": 3030
    },
    {
      "epoch": 0.01228198467175992,
      "grad_norm": 765.9307861328125,
      "learning_rate": 6.08e-06,
      "loss": 412.906,
      "step": 3040
    },
    {
      "epoch": 0.01232238593712755,
      "grad_norm": 1054.11669921875,
      "learning_rate": 6.1e-06,
      "loss": 356.0523,
      "step": 3050
    },
    {
      "epoch": 0.012362787202495182,
      "grad_norm": 786.2109375,
      "learning_rate": 6.12e-06,
      "loss": 260.9853,
      "step": 3060
    },
    {
      "epoch": 0.012403188467862813,
      "grad_norm": 1215.10400390625,
      "learning_rate": 6.1400000000000005e-06,
      "loss": 259.1632,
      "step": 3070
    },
    {
      "epoch": 0.012443589733230445,
      "grad_norm": 943.6842651367188,
      "learning_rate": 6.16e-06,
      "loss": 216.0937,
      "step": 3080
    },
    {
      "epoch": 0.012483990998598076,
      "grad_norm": 1460.5831298828125,
      "learning_rate": 6.18e-06,
      "loss": 337.1417,
      "step": 3090
    },
    {
      "epoch": 0.012524392263965707,
      "grad_norm": 3781.843017578125,
      "learning_rate": 6.2e-06,
      "loss": 374.3547,
      "step": 3100
    },
    {
      "epoch": 0.012564793529333339,
      "grad_norm": 676.937744140625,
      "learning_rate": 6.22e-06,
      "loss": 350.8003,
      "step": 3110
    },
    {
      "epoch": 0.01260519479470097,
      "grad_norm": 1300.2470703125,
      "learning_rate": 6.24e-06,
      "loss": 354.9542,
      "step": 3120
    },
    {
      "epoch": 0.012645596060068601,
      "grad_norm": 715.6893310546875,
      "learning_rate": 6.26e-06,
      "loss": 225.5772,
      "step": 3130
    },
    {
      "epoch": 0.012685997325436233,
      "grad_norm": 438.3706359863281,
      "learning_rate": 6.28e-06,
      "loss": 262.8397,
      "step": 3140
    },
    {
      "epoch": 0.012726398590803864,
      "grad_norm": 5010.02392578125,
      "learning_rate": 6.300000000000001e-06,
      "loss": 282.8991,
      "step": 3150
    },
    {
      "epoch": 0.012766799856171495,
      "grad_norm": 1642.5548095703125,
      "learning_rate": 6.320000000000001e-06,
      "loss": 273.512,
      "step": 3160
    },
    {
      "epoch": 0.012807201121539127,
      "grad_norm": 484.15069580078125,
      "learning_rate": 6.34e-06,
      "loss": 190.9169,
      "step": 3170
    },
    {
      "epoch": 0.012847602386906758,
      "grad_norm": 1157.349365234375,
      "learning_rate": 6.360000000000001e-06,
      "loss": 378.3686,
      "step": 3180
    },
    {
      "epoch": 0.01288800365227439,
      "grad_norm": 547.7994384765625,
      "learning_rate": 6.38e-06,
      "loss": 241.6315,
      "step": 3190
    },
    {
      "epoch": 0.01292840491764202,
      "grad_norm": 0.0,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 233.8554,
      "step": 3200
    },
    {
      "epoch": 0.012968806183009652,
      "grad_norm": 953.7396850585938,
      "learning_rate": 6.4199999999999995e-06,
      "loss": 316.3731,
      "step": 3210
    },
    {
      "epoch": 0.013009207448377283,
      "grad_norm": 539.7330932617188,
      "learning_rate": 6.44e-06,
      "loss": 234.0129,
      "step": 3220
    },
    {
      "epoch": 0.013049608713744915,
      "grad_norm": 736.7625732421875,
      "learning_rate": 6.460000000000001e-06,
      "loss": 335.0041,
      "step": 3230
    },
    {
      "epoch": 0.013090009979112546,
      "grad_norm": 1194.715576171875,
      "learning_rate": 6.48e-06,
      "loss": 276.9117,
      "step": 3240
    },
    {
      "epoch": 0.013130411244480177,
      "grad_norm": 8021.32763671875,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 374.7519,
      "step": 3250
    },
    {
      "epoch": 0.013170812509847809,
      "grad_norm": 931.0237426757812,
      "learning_rate": 6.519999999999999e-06,
      "loss": 278.108,
      "step": 3260
    },
    {
      "epoch": 0.01321121377521544,
      "grad_norm": 953.587890625,
      "learning_rate": 6.54e-06,
      "loss": 368.7168,
      "step": 3270
    },
    {
      "epoch": 0.013251615040583071,
      "grad_norm": 1098.186767578125,
      "learning_rate": 6.560000000000001e-06,
      "loss": 276.9152,
      "step": 3280
    },
    {
      "epoch": 0.013292016305950703,
      "grad_norm": 1590.1026611328125,
      "learning_rate": 6.58e-06,
      "loss": 422.8925,
      "step": 3290
    },
    {
      "epoch": 0.013332417571318334,
      "grad_norm": 857.4612426757812,
      "learning_rate": 6.6e-06,
      "loss": 343.3155,
      "step": 3300
    },
    {
      "epoch": 0.013372818836685965,
      "grad_norm": 941.4984130859375,
      "learning_rate": 6.62e-06,
      "loss": 187.542,
      "step": 3310
    },
    {
      "epoch": 0.013413220102053596,
      "grad_norm": 938.3494262695312,
      "learning_rate": 6.640000000000001e-06,
      "loss": 237.5327,
      "step": 3320
    },
    {
      "epoch": 0.013453621367421228,
      "grad_norm": 936.9948120117188,
      "learning_rate": 6.660000000000001e-06,
      "loss": 275.3572,
      "step": 3330
    },
    {
      "epoch": 0.013494022632788859,
      "grad_norm": 1025.8851318359375,
      "learning_rate": 6.68e-06,
      "loss": 346.2491,
      "step": 3340
    },
    {
      "epoch": 0.01353442389815649,
      "grad_norm": 2196.15478515625,
      "learning_rate": 6.700000000000001e-06,
      "loss": 242.7346,
      "step": 3350
    },
    {
      "epoch": 0.013574825163524122,
      "grad_norm": 843.5136108398438,
      "learning_rate": 6.72e-06,
      "loss": 188.1487,
      "step": 3360
    },
    {
      "epoch": 0.013615226428891753,
      "grad_norm": 714.3571166992188,
      "learning_rate": 6.740000000000001e-06,
      "loss": 358.2911,
      "step": 3370
    },
    {
      "epoch": 0.013655627694259384,
      "grad_norm": 689.7639770507812,
      "learning_rate": 6.76e-06,
      "loss": 150.4375,
      "step": 3380
    },
    {
      "epoch": 0.013696028959627016,
      "grad_norm": 1244.08837890625,
      "learning_rate": 6.78e-06,
      "loss": 300.2315,
      "step": 3390
    },
    {
      "epoch": 0.013736430224994647,
      "grad_norm": 1171.05419921875,
      "learning_rate": 6.800000000000001e-06,
      "loss": 241.5314,
      "step": 3400
    },
    {
      "epoch": 0.013776831490362278,
      "grad_norm": 1291.3603515625,
      "learning_rate": 6.82e-06,
      "loss": 449.0854,
      "step": 3410
    },
    {
      "epoch": 0.01381723275572991,
      "grad_norm": 1006.1640014648438,
      "learning_rate": 6.840000000000001e-06,
      "loss": 346.3529,
      "step": 3420
    },
    {
      "epoch": 0.013857634021097541,
      "grad_norm": 6614.48583984375,
      "learning_rate": 6.8599999999999995e-06,
      "loss": 289.8388,
      "step": 3430
    },
    {
      "epoch": 0.013898035286465172,
      "grad_norm": 574.2523803710938,
      "learning_rate": 6.88e-06,
      "loss": 187.4786,
      "step": 3440
    },
    {
      "epoch": 0.013938436551832804,
      "grad_norm": 577.289794921875,
      "learning_rate": 6.900000000000001e-06,
      "loss": 280.2674,
      "step": 3450
    },
    {
      "epoch": 0.013978837817200435,
      "grad_norm": 1416.127197265625,
      "learning_rate": 6.92e-06,
      "loss": 330.4,
      "step": 3460
    },
    {
      "epoch": 0.014019239082568066,
      "grad_norm": 973.31689453125,
      "learning_rate": 6.9400000000000005e-06,
      "loss": 251.0885,
      "step": 3470
    },
    {
      "epoch": 0.014059640347935698,
      "grad_norm": 2889.614013671875,
      "learning_rate": 6.9599999999999994e-06,
      "loss": 288.5922,
      "step": 3480
    },
    {
      "epoch": 0.014100041613303329,
      "grad_norm": 734.4885864257812,
      "learning_rate": 6.98e-06,
      "loss": 247.6378,
      "step": 3490
    },
    {
      "epoch": 0.01414044287867096,
      "grad_norm": 1920.7076416015625,
      "learning_rate": 7.000000000000001e-06,
      "loss": 175.3216,
      "step": 3500
    },
    {
      "epoch": 0.014180844144038592,
      "grad_norm": 676.3031005859375,
      "learning_rate": 7.0200000000000006e-06,
      "loss": 209.18,
      "step": 3510
    },
    {
      "epoch": 0.014221245409406223,
      "grad_norm": 1840.922119140625,
      "learning_rate": 7.04e-06,
      "loss": 241.9013,
      "step": 3520
    },
    {
      "epoch": 0.014261646674773854,
      "grad_norm": 1004.5390625,
      "learning_rate": 7.06e-06,
      "loss": 219.0837,
      "step": 3530
    },
    {
      "epoch": 0.014302047940141486,
      "grad_norm": 2489.928955078125,
      "learning_rate": 7.080000000000001e-06,
      "loss": 207.0227,
      "step": 3540
    },
    {
      "epoch": 0.014342449205509117,
      "grad_norm": 713.509033203125,
      "learning_rate": 7.1e-06,
      "loss": 208.6873,
      "step": 3550
    },
    {
      "epoch": 0.014382850470876748,
      "grad_norm": 453.5292053222656,
      "learning_rate": 7.1200000000000004e-06,
      "loss": 281.9877,
      "step": 3560
    },
    {
      "epoch": 0.01442325173624438,
      "grad_norm": 900.6409912109375,
      "learning_rate": 7.140000000000001e-06,
      "loss": 235.5468,
      "step": 3570
    },
    {
      "epoch": 0.01446365300161201,
      "grad_norm": 438.5726623535156,
      "learning_rate": 7.16e-06,
      "loss": 162.5427,
      "step": 3580
    },
    {
      "epoch": 0.014504054266979642,
      "grad_norm": 861.850341796875,
      "learning_rate": 7.180000000000001e-06,
      "loss": 236.0929,
      "step": 3590
    },
    {
      "epoch": 0.014544455532347273,
      "grad_norm": 892.953369140625,
      "learning_rate": 7.2e-06,
      "loss": 249.5097,
      "step": 3600
    },
    {
      "epoch": 0.014584856797714905,
      "grad_norm": 1040.248046875,
      "learning_rate": 7.22e-06,
      "loss": 268.9143,
      "step": 3610
    },
    {
      "epoch": 0.014625258063082536,
      "grad_norm": 774.4244384765625,
      "learning_rate": 7.240000000000001e-06,
      "loss": 235.6966,
      "step": 3620
    },
    {
      "epoch": 0.014665659328450167,
      "grad_norm": 540.7510986328125,
      "learning_rate": 7.26e-06,
      "loss": 222.3721,
      "step": 3630
    },
    {
      "epoch": 0.014706060593817799,
      "grad_norm": 3457.84912109375,
      "learning_rate": 7.280000000000001e-06,
      "loss": 262.0487,
      "step": 3640
    },
    {
      "epoch": 0.01474646185918543,
      "grad_norm": 802.951416015625,
      "learning_rate": 7.2999999999999996e-06,
      "loss": 277.8951,
      "step": 3650
    },
    {
      "epoch": 0.014786863124553061,
      "grad_norm": 1426.4849853515625,
      "learning_rate": 7.32e-06,
      "loss": 273.6454,
      "step": 3660
    },
    {
      "epoch": 0.014827264389920693,
      "grad_norm": 763.5048828125,
      "learning_rate": 7.340000000000001e-06,
      "loss": 324.4447,
      "step": 3670
    },
    {
      "epoch": 0.014867665655288324,
      "grad_norm": 2634.498046875,
      "learning_rate": 7.36e-06,
      "loss": 285.5299,
      "step": 3680
    },
    {
      "epoch": 0.014908066920655955,
      "grad_norm": 1109.640380859375,
      "learning_rate": 7.3800000000000005e-06,
      "loss": 421.3942,
      "step": 3690
    },
    {
      "epoch": 0.014948468186023587,
      "grad_norm": 1876.8143310546875,
      "learning_rate": 7.4e-06,
      "loss": 345.6086,
      "step": 3700
    },
    {
      "epoch": 0.014988869451391218,
      "grad_norm": 820.827880859375,
      "learning_rate": 7.420000000000001e-06,
      "loss": 241.7691,
      "step": 3710
    },
    {
      "epoch": 0.01502927071675885,
      "grad_norm": 1500.841552734375,
      "learning_rate": 7.44e-06,
      "loss": 247.7824,
      "step": 3720
    },
    {
      "epoch": 0.01506967198212648,
      "grad_norm": 1121.291259765625,
      "learning_rate": 7.4600000000000006e-06,
      "loss": 302.3958,
      "step": 3730
    },
    {
      "epoch": 0.015110073247494112,
      "grad_norm": 1057.3134765625,
      "learning_rate": 7.480000000000001e-06,
      "loss": 271.7261,
      "step": 3740
    },
    {
      "epoch": 0.015150474512861743,
      "grad_norm": 1772.322509765625,
      "learning_rate": 7.5e-06,
      "loss": 404.0533,
      "step": 3750
    },
    {
      "epoch": 0.015190875778229375,
      "grad_norm": 547.7960815429688,
      "learning_rate": 7.520000000000001e-06,
      "loss": 255.6221,
      "step": 3760
    },
    {
      "epoch": 0.015231277043597006,
      "grad_norm": 980.1959228515625,
      "learning_rate": 7.54e-06,
      "loss": 310.8835,
      "step": 3770
    },
    {
      "epoch": 0.015271678308964637,
      "grad_norm": 1686.9091796875,
      "learning_rate": 7.5600000000000005e-06,
      "loss": 246.4056,
      "step": 3780
    },
    {
      "epoch": 0.015312079574332269,
      "grad_norm": 867.049072265625,
      "learning_rate": 7.580000000000001e-06,
      "loss": 246.8477,
      "step": 3790
    },
    {
      "epoch": 0.0153524808396999,
      "grad_norm": 1169.5635986328125,
      "learning_rate": 7.6e-06,
      "loss": 184.4648,
      "step": 3800
    },
    {
      "epoch": 0.015392882105067531,
      "grad_norm": 689.9214477539062,
      "learning_rate": 7.620000000000001e-06,
      "loss": 323.659,
      "step": 3810
    },
    {
      "epoch": 0.015433283370435162,
      "grad_norm": 519.1124267578125,
      "learning_rate": 7.64e-06,
      "loss": 225.6972,
      "step": 3820
    },
    {
      "epoch": 0.015473684635802794,
      "grad_norm": 1257.6265869140625,
      "learning_rate": 7.660000000000001e-06,
      "loss": 337.094,
      "step": 3830
    },
    {
      "epoch": 0.015514085901170425,
      "grad_norm": 5566.001953125,
      "learning_rate": 7.68e-06,
      "loss": 191.7031,
      "step": 3840
    },
    {
      "epoch": 0.015554487166538056,
      "grad_norm": 734.2090454101562,
      "learning_rate": 7.7e-06,
      "loss": 221.7874,
      "step": 3850
    },
    {
      "epoch": 0.015594888431905688,
      "grad_norm": 1421.68359375,
      "learning_rate": 7.72e-06,
      "loss": 266.4913,
      "step": 3860
    },
    {
      "epoch": 0.01563528969727332,
      "grad_norm": 2573.87353515625,
      "learning_rate": 7.74e-06,
      "loss": 301.2037,
      "step": 3870
    },
    {
      "epoch": 0.01567569096264095,
      "grad_norm": 492.0390319824219,
      "learning_rate": 7.76e-06,
      "loss": 234.7707,
      "step": 3880
    },
    {
      "epoch": 0.01571609222800858,
      "grad_norm": 600.4080200195312,
      "learning_rate": 7.78e-06,
      "loss": 267.7107,
      "step": 3890
    },
    {
      "epoch": 0.01575649349337621,
      "grad_norm": 1099.673828125,
      "learning_rate": 7.8e-06,
      "loss": 292.7474,
      "step": 3900
    },
    {
      "epoch": 0.015796894758743844,
      "grad_norm": 541.96875,
      "learning_rate": 7.820000000000001e-06,
      "loss": 229.897,
      "step": 3910
    },
    {
      "epoch": 0.015837296024111474,
      "grad_norm": 3221.322509765625,
      "learning_rate": 7.84e-06,
      "loss": 200.1891,
      "step": 3920
    },
    {
      "epoch": 0.015877697289479107,
      "grad_norm": 2533.82177734375,
      "learning_rate": 7.860000000000001e-06,
      "loss": 240.3558,
      "step": 3930
    },
    {
      "epoch": 0.015918098554846737,
      "grad_norm": 909.9031372070312,
      "learning_rate": 7.879999999999999e-06,
      "loss": 231.9597,
      "step": 3940
    },
    {
      "epoch": 0.01595849982021437,
      "grad_norm": 669.5313110351562,
      "learning_rate": 7.9e-06,
      "loss": 316.9614,
      "step": 3950
    },
    {
      "epoch": 0.015998901085582,
      "grad_norm": 694.2930297851562,
      "learning_rate": 7.92e-06,
      "loss": 272.3228,
      "step": 3960
    },
    {
      "epoch": 0.016039302350949632,
      "grad_norm": 528.3792114257812,
      "learning_rate": 7.94e-06,
      "loss": 329.1032,
      "step": 3970
    },
    {
      "epoch": 0.016079703616317262,
      "grad_norm": 579.0252075195312,
      "learning_rate": 7.96e-06,
      "loss": 258.9907,
      "step": 3980
    },
    {
      "epoch": 0.016120104881684895,
      "grad_norm": 645.7807006835938,
      "learning_rate": 7.98e-06,
      "loss": 242.0821,
      "step": 3990
    },
    {
      "epoch": 0.016160506147052525,
      "grad_norm": 543.9231567382812,
      "learning_rate": 8.000000000000001e-06,
      "loss": 284.7925,
      "step": 4000
    },
    {
      "epoch": 0.016200907412420158,
      "grad_norm": 610.683349609375,
      "learning_rate": 8.02e-06,
      "loss": 220.6932,
      "step": 4010
    },
    {
      "epoch": 0.016241308677787787,
      "grad_norm": 0.0,
      "learning_rate": 8.040000000000001e-06,
      "loss": 230.9501,
      "step": 4020
    },
    {
      "epoch": 0.01628170994315542,
      "grad_norm": 1050.20947265625,
      "learning_rate": 8.06e-06,
      "loss": 292.7622,
      "step": 4030
    },
    {
      "epoch": 0.01632211120852305,
      "grad_norm": 1064.6005859375,
      "learning_rate": 8.08e-06,
      "loss": 346.057,
      "step": 4040
    },
    {
      "epoch": 0.016362512473890683,
      "grad_norm": 1817.19482421875,
      "learning_rate": 8.1e-06,
      "loss": 229.4932,
      "step": 4050
    },
    {
      "epoch": 0.016402913739258312,
      "grad_norm": 1169.279052734375,
      "learning_rate": 8.12e-06,
      "loss": 346.2139,
      "step": 4060
    },
    {
      "epoch": 0.016443315004625945,
      "grad_norm": 1004.5889282226562,
      "learning_rate": 8.14e-06,
      "loss": 257.1552,
      "step": 4070
    },
    {
      "epoch": 0.016483716269993575,
      "grad_norm": 660.7535400390625,
      "learning_rate": 8.160000000000001e-06,
      "loss": 247.7336,
      "step": 4080
    },
    {
      "epoch": 0.016524117535361208,
      "grad_norm": 638.9656372070312,
      "learning_rate": 8.18e-06,
      "loss": 360.9811,
      "step": 4090
    },
    {
      "epoch": 0.016564518800728838,
      "grad_norm": 865.1559448242188,
      "learning_rate": 8.200000000000001e-06,
      "loss": 267.7802,
      "step": 4100
    },
    {
      "epoch": 0.01660492006609647,
      "grad_norm": 528.1417236328125,
      "learning_rate": 8.22e-06,
      "loss": 211.1934,
      "step": 4110
    },
    {
      "epoch": 0.0166453213314641,
      "grad_norm": 602.795166015625,
      "learning_rate": 8.24e-06,
      "loss": 342.2845,
      "step": 4120
    },
    {
      "epoch": 0.016685722596831733,
      "grad_norm": 537.5540161132812,
      "learning_rate": 8.26e-06,
      "loss": 222.6927,
      "step": 4130
    },
    {
      "epoch": 0.016726123862199363,
      "grad_norm": 1009.3809814453125,
      "learning_rate": 8.28e-06,
      "loss": 278.8901,
      "step": 4140
    },
    {
      "epoch": 0.016766525127566996,
      "grad_norm": 1916.926513671875,
      "learning_rate": 8.3e-06,
      "loss": 335.051,
      "step": 4150
    },
    {
      "epoch": 0.016806926392934626,
      "grad_norm": 759.0120849609375,
      "learning_rate": 8.32e-06,
      "loss": 292.8282,
      "step": 4160
    },
    {
      "epoch": 0.01684732765830226,
      "grad_norm": 1126.7659912109375,
      "learning_rate": 8.34e-06,
      "loss": 356.8466,
      "step": 4170
    },
    {
      "epoch": 0.016887728923669888,
      "grad_norm": 4519.04833984375,
      "learning_rate": 8.36e-06,
      "loss": 254.2794,
      "step": 4180
    },
    {
      "epoch": 0.01692813018903752,
      "grad_norm": 625.6685791015625,
      "learning_rate": 8.380000000000001e-06,
      "loss": 256.8828,
      "step": 4190
    },
    {
      "epoch": 0.01696853145440515,
      "grad_norm": 901.2313842773438,
      "learning_rate": 8.400000000000001e-06,
      "loss": 173.8549,
      "step": 4200
    },
    {
      "epoch": 0.017008932719772784,
      "grad_norm": 612.3013916015625,
      "learning_rate": 8.42e-06,
      "loss": 278.5784,
      "step": 4210
    },
    {
      "epoch": 0.017049333985140414,
      "grad_norm": 500.49169921875,
      "learning_rate": 8.44e-06,
      "loss": 197.3738,
      "step": 4220
    },
    {
      "epoch": 0.017089735250508047,
      "grad_norm": 1060.2108154296875,
      "learning_rate": 8.46e-06,
      "loss": 298.4693,
      "step": 4230
    },
    {
      "epoch": 0.017130136515875676,
      "grad_norm": 612.9854736328125,
      "learning_rate": 8.48e-06,
      "loss": 329.1869,
      "step": 4240
    },
    {
      "epoch": 0.01717053778124331,
      "grad_norm": 1364.545654296875,
      "learning_rate": 8.500000000000002e-06,
      "loss": 334.3979,
      "step": 4250
    },
    {
      "epoch": 0.01721093904661094,
      "grad_norm": 696.9126586914062,
      "learning_rate": 8.52e-06,
      "loss": 271.8317,
      "step": 4260
    },
    {
      "epoch": 0.017251340311978572,
      "grad_norm": 688.8264770507812,
      "learning_rate": 8.540000000000001e-06,
      "loss": 203.2293,
      "step": 4270
    },
    {
      "epoch": 0.0172917415773462,
      "grad_norm": 1147.045654296875,
      "learning_rate": 8.56e-06,
      "loss": 349.6926,
      "step": 4280
    },
    {
      "epoch": 0.017332142842713835,
      "grad_norm": 477.3202209472656,
      "learning_rate": 8.580000000000001e-06,
      "loss": 146.8848,
      "step": 4290
    },
    {
      "epoch": 0.017372544108081464,
      "grad_norm": 1310.89599609375,
      "learning_rate": 8.599999999999999e-06,
      "loss": 247.6962,
      "step": 4300
    },
    {
      "epoch": 0.017412945373449097,
      "grad_norm": 1124.722900390625,
      "learning_rate": 8.62e-06,
      "loss": 347.9728,
      "step": 4310
    },
    {
      "epoch": 0.017453346638816727,
      "grad_norm": 1394.403564453125,
      "learning_rate": 8.64e-06,
      "loss": 261.2298,
      "step": 4320
    },
    {
      "epoch": 0.01749374790418436,
      "grad_norm": 889.3135375976562,
      "learning_rate": 8.66e-06,
      "loss": 176.3773,
      "step": 4330
    },
    {
      "epoch": 0.01753414916955199,
      "grad_norm": 1988.6873779296875,
      "learning_rate": 8.68e-06,
      "loss": 260.4097,
      "step": 4340
    },
    {
      "epoch": 0.017574550434919622,
      "grad_norm": 1403.53955078125,
      "learning_rate": 8.7e-06,
      "loss": 222.8027,
      "step": 4350
    },
    {
      "epoch": 0.017614951700287252,
      "grad_norm": 1020.5213623046875,
      "learning_rate": 8.720000000000001e-06,
      "loss": 202.7748,
      "step": 4360
    },
    {
      "epoch": 0.017655352965654885,
      "grad_norm": 4106.30126953125,
      "learning_rate": 8.740000000000001e-06,
      "loss": 257.8873,
      "step": 4370
    },
    {
      "epoch": 0.017695754231022515,
      "grad_norm": 458.9610290527344,
      "learning_rate": 8.76e-06,
      "loss": 274.0975,
      "step": 4380
    },
    {
      "epoch": 0.017736155496390148,
      "grad_norm": 1450.5250244140625,
      "learning_rate": 8.78e-06,
      "loss": 212.7885,
      "step": 4390
    },
    {
      "epoch": 0.017776556761757777,
      "grad_norm": 658.0136108398438,
      "learning_rate": 8.8e-06,
      "loss": 313.3228,
      "step": 4400
    },
    {
      "epoch": 0.01781695802712541,
      "grad_norm": 607.8909301757812,
      "learning_rate": 8.82e-06,
      "loss": 220.5824,
      "step": 4410
    },
    {
      "epoch": 0.01785735929249304,
      "grad_norm": 964.3424682617188,
      "learning_rate": 8.840000000000002e-06,
      "loss": 214.6224,
      "step": 4420
    },
    {
      "epoch": 0.017897760557860673,
      "grad_norm": 1267.5516357421875,
      "learning_rate": 8.86e-06,
      "loss": 231.4749,
      "step": 4430
    },
    {
      "epoch": 0.017938161823228303,
      "grad_norm": 852.99462890625,
      "learning_rate": 8.880000000000001e-06,
      "loss": 209.5046,
      "step": 4440
    },
    {
      "epoch": 0.017978563088595936,
      "grad_norm": 895.4765014648438,
      "learning_rate": 8.9e-06,
      "loss": 227.3372,
      "step": 4450
    },
    {
      "epoch": 0.018018964353963565,
      "grad_norm": 658.2288818359375,
      "learning_rate": 8.920000000000001e-06,
      "loss": 235.962,
      "step": 4460
    },
    {
      "epoch": 0.0180593656193312,
      "grad_norm": 1219.0494384765625,
      "learning_rate": 8.939999999999999e-06,
      "loss": 222.9912,
      "step": 4470
    },
    {
      "epoch": 0.018099766884698828,
      "grad_norm": 976.2411499023438,
      "learning_rate": 8.96e-06,
      "loss": 251.0392,
      "step": 4480
    },
    {
      "epoch": 0.01814016815006646,
      "grad_norm": 1230.3253173828125,
      "learning_rate": 8.98e-06,
      "loss": 208.2115,
      "step": 4490
    },
    {
      "epoch": 0.01818056941543409,
      "grad_norm": 738.1622314453125,
      "learning_rate": 9e-06,
      "loss": 248.237,
      "step": 4500
    },
    {
      "epoch": 0.018220970680801724,
      "grad_norm": 825.9674072265625,
      "learning_rate": 9.02e-06,
      "loss": 231.5556,
      "step": 4510
    },
    {
      "epoch": 0.018261371946169353,
      "grad_norm": 782.737060546875,
      "learning_rate": 9.04e-06,
      "loss": 213.6435,
      "step": 4520
    },
    {
      "epoch": 0.018301773211536986,
      "grad_norm": 712.6553344726562,
      "learning_rate": 9.06e-06,
      "loss": 234.0471,
      "step": 4530
    },
    {
      "epoch": 0.018342174476904616,
      "grad_norm": 1101.6629638671875,
      "learning_rate": 9.080000000000001e-06,
      "loss": 198.3389,
      "step": 4540
    },
    {
      "epoch": 0.01838257574227225,
      "grad_norm": 1747.929443359375,
      "learning_rate": 9.100000000000001e-06,
      "loss": 332.4391,
      "step": 4550
    },
    {
      "epoch": 0.01842297700763988,
      "grad_norm": 2980.9208984375,
      "learning_rate": 9.12e-06,
      "loss": 337.9396,
      "step": 4560
    },
    {
      "epoch": 0.01846337827300751,
      "grad_norm": 1282.68115234375,
      "learning_rate": 9.14e-06,
      "loss": 237.8889,
      "step": 4570
    },
    {
      "epoch": 0.01850377953837514,
      "grad_norm": 948.6932983398438,
      "learning_rate": 9.16e-06,
      "loss": 288.0977,
      "step": 4580
    },
    {
      "epoch": 0.018544180803742774,
      "grad_norm": 501.8581237792969,
      "learning_rate": 9.180000000000002e-06,
      "loss": 329.9324,
      "step": 4590
    },
    {
      "epoch": 0.018584582069110404,
      "grad_norm": 1850.9791259765625,
      "learning_rate": 9.2e-06,
      "loss": 319.7892,
      "step": 4600
    },
    {
      "epoch": 0.018624983334478037,
      "grad_norm": 730.577392578125,
      "learning_rate": 9.220000000000002e-06,
      "loss": 217.722,
      "step": 4610
    },
    {
      "epoch": 0.018665384599845666,
      "grad_norm": 408.53619384765625,
      "learning_rate": 9.24e-06,
      "loss": 190.9173,
      "step": 4620
    },
    {
      "epoch": 0.0187057858652133,
      "grad_norm": 884.1920776367188,
      "learning_rate": 9.260000000000001e-06,
      "loss": 193.8064,
      "step": 4630
    },
    {
      "epoch": 0.01874618713058093,
      "grad_norm": 1684.18408203125,
      "learning_rate": 9.28e-06,
      "loss": 197.2281,
      "step": 4640
    },
    {
      "epoch": 0.018786588395948562,
      "grad_norm": 1591.9307861328125,
      "learning_rate": 9.3e-06,
      "loss": 282.6273,
      "step": 4650
    },
    {
      "epoch": 0.01882698966131619,
      "grad_norm": 1377.8363037109375,
      "learning_rate": 9.32e-06,
      "loss": 306.7148,
      "step": 4660
    },
    {
      "epoch": 0.018867390926683825,
      "grad_norm": 600.4591674804688,
      "learning_rate": 9.34e-06,
      "loss": 290.5179,
      "step": 4670
    },
    {
      "epoch": 0.018907792192051454,
      "grad_norm": 1007.29296875,
      "learning_rate": 9.36e-06,
      "loss": 211.2058,
      "step": 4680
    },
    {
      "epoch": 0.018948193457419087,
      "grad_norm": 1834.026611328125,
      "learning_rate": 9.38e-06,
      "loss": 214.4897,
      "step": 4690
    },
    {
      "epoch": 0.018988594722786717,
      "grad_norm": 895.9391479492188,
      "learning_rate": 9.4e-06,
      "loss": 241.2341,
      "step": 4700
    },
    {
      "epoch": 0.01902899598815435,
      "grad_norm": 940.978271484375,
      "learning_rate": 9.420000000000001e-06,
      "loss": 255.7137,
      "step": 4710
    },
    {
      "epoch": 0.01906939725352198,
      "grad_norm": 0.0,
      "learning_rate": 9.44e-06,
      "loss": 202.176,
      "step": 4720
    },
    {
      "epoch": 0.019109798518889613,
      "grad_norm": 1429.115234375,
      "learning_rate": 9.460000000000001e-06,
      "loss": 199.2825,
      "step": 4730
    },
    {
      "epoch": 0.019150199784257242,
      "grad_norm": 1368.30322265625,
      "learning_rate": 9.48e-06,
      "loss": 265.0907,
      "step": 4740
    },
    {
      "epoch": 0.019190601049624875,
      "grad_norm": 773.6226196289062,
      "learning_rate": 9.5e-06,
      "loss": 272.2148,
      "step": 4750
    },
    {
      "epoch": 0.019231002314992505,
      "grad_norm": 799.7369384765625,
      "learning_rate": 9.52e-06,
      "loss": 259.8739,
      "step": 4760
    },
    {
      "epoch": 0.019271403580360138,
      "grad_norm": 718.1935424804688,
      "learning_rate": 9.54e-06,
      "loss": 255.6761,
      "step": 4770
    },
    {
      "epoch": 0.019311804845727767,
      "grad_norm": 1052.681396484375,
      "learning_rate": 9.560000000000002e-06,
      "loss": 240.1528,
      "step": 4780
    },
    {
      "epoch": 0.0193522061110954,
      "grad_norm": 351.5966491699219,
      "learning_rate": 9.58e-06,
      "loss": 170.535,
      "step": 4790
    },
    {
      "epoch": 0.01939260737646303,
      "grad_norm": 759.4265747070312,
      "learning_rate": 9.600000000000001e-06,
      "loss": 185.2903,
      "step": 4800
    },
    {
      "epoch": 0.019433008641830663,
      "grad_norm": 763.7293701171875,
      "learning_rate": 9.62e-06,
      "loss": 233.2488,
      "step": 4810
    },
    {
      "epoch": 0.019473409907198293,
      "grad_norm": 1342.382568359375,
      "learning_rate": 9.640000000000001e-06,
      "loss": 275.3923,
      "step": 4820
    },
    {
      "epoch": 0.019513811172565926,
      "grad_norm": 3298.531005859375,
      "learning_rate": 9.66e-06,
      "loss": 226.5765,
      "step": 4830
    },
    {
      "epoch": 0.019554212437933555,
      "grad_norm": 1122.7933349609375,
      "learning_rate": 9.68e-06,
      "loss": 178.3835,
      "step": 4840
    },
    {
      "epoch": 0.01959461370330119,
      "grad_norm": 1547.0048828125,
      "learning_rate": 9.7e-06,
      "loss": 305.318,
      "step": 4850
    },
    {
      "epoch": 0.019635014968668818,
      "grad_norm": 364.29541015625,
      "learning_rate": 9.72e-06,
      "loss": 170.9627,
      "step": 4860
    },
    {
      "epoch": 0.01967541623403645,
      "grad_norm": 1955.25634765625,
      "learning_rate": 9.74e-06,
      "loss": 302.9251,
      "step": 4870
    },
    {
      "epoch": 0.01971581749940408,
      "grad_norm": 2048.748291015625,
      "learning_rate": 9.760000000000001e-06,
      "loss": 202.4392,
      "step": 4880
    },
    {
      "epoch": 0.019756218764771714,
      "grad_norm": 660.9871215820312,
      "learning_rate": 9.78e-06,
      "loss": 187.9075,
      "step": 4890
    },
    {
      "epoch": 0.019796620030139343,
      "grad_norm": 689.9225463867188,
      "learning_rate": 9.800000000000001e-06,
      "loss": 311.5491,
      "step": 4900
    },
    {
      "epoch": 0.019837021295506976,
      "grad_norm": 953.7089233398438,
      "learning_rate": 9.820000000000001e-06,
      "loss": 299.3845,
      "step": 4910
    },
    {
      "epoch": 0.019877422560874606,
      "grad_norm": 614.9231567382812,
      "learning_rate": 9.84e-06,
      "loss": 168.2812,
      "step": 4920
    },
    {
      "epoch": 0.01991782382624224,
      "grad_norm": 1575.7044677734375,
      "learning_rate": 9.86e-06,
      "loss": 273.8306,
      "step": 4930
    },
    {
      "epoch": 0.01995822509160987,
      "grad_norm": 826.9859619140625,
      "learning_rate": 9.88e-06,
      "loss": 277.1096,
      "step": 4940
    },
    {
      "epoch": 0.0199986263569775,
      "grad_norm": 934.8515625,
      "learning_rate": 9.900000000000002e-06,
      "loss": 230.0742,
      "step": 4950
    },
    {
      "epoch": 0.02003902762234513,
      "grad_norm": 2514.587646484375,
      "learning_rate": 9.92e-06,
      "loss": 143.4857,
      "step": 4960
    },
    {
      "epoch": 0.020079428887712764,
      "grad_norm": 0.0,
      "learning_rate": 9.940000000000001e-06,
      "loss": 292.2021,
      "step": 4970
    },
    {
      "epoch": 0.020119830153080394,
      "grad_norm": 1051.95068359375,
      "learning_rate": 9.96e-06,
      "loss": 243.0733,
      "step": 4980
    },
    {
      "epoch": 0.020160231418448027,
      "grad_norm": 522.6783447265625,
      "learning_rate": 9.980000000000001e-06,
      "loss": 255.3195,
      "step": 4990
    },
    {
      "epoch": 0.020200632683815656,
      "grad_norm": 1430.8314208984375,
      "learning_rate": 1e-05,
      "loss": 181.9884,
      "step": 5000
    },
    {
      "epoch": 0.02024103394918329,
      "grad_norm": 2610.226806640625,
      "learning_rate": 1.002e-05,
      "loss": 240.1806,
      "step": 5010
    },
    {
      "epoch": 0.02028143521455092,
      "grad_norm": 706.251220703125,
      "learning_rate": 1.004e-05,
      "loss": 154.691,
      "step": 5020
    },
    {
      "epoch": 0.020321836479918552,
      "grad_norm": 1110.47021484375,
      "learning_rate": 1.006e-05,
      "loss": 270.7392,
      "step": 5030
    },
    {
      "epoch": 0.020362237745286182,
      "grad_norm": 1037.9814453125,
      "learning_rate": 1.008e-05,
      "loss": 260.6087,
      "step": 5040
    },
    {
      "epoch": 0.020402639010653815,
      "grad_norm": 988.474609375,
      "learning_rate": 1.0100000000000002e-05,
      "loss": 195.7535,
      "step": 5050
    },
    {
      "epoch": 0.020443040276021444,
      "grad_norm": 0.0,
      "learning_rate": 1.012e-05,
      "loss": 292.6674,
      "step": 5060
    },
    {
      "epoch": 0.020483441541389077,
      "grad_norm": 1622.9769287109375,
      "learning_rate": 1.0140000000000001e-05,
      "loss": 226.0767,
      "step": 5070
    },
    {
      "epoch": 0.020523842806756707,
      "grad_norm": 2696.41845703125,
      "learning_rate": 1.016e-05,
      "loss": 242.0753,
      "step": 5080
    },
    {
      "epoch": 0.02056424407212434,
      "grad_norm": 2494.242431640625,
      "learning_rate": 1.018e-05,
      "loss": 310.918,
      "step": 5090
    },
    {
      "epoch": 0.02060464533749197,
      "grad_norm": 549.554443359375,
      "learning_rate": 1.02e-05,
      "loss": 238.1347,
      "step": 5100
    },
    {
      "epoch": 0.020645046602859603,
      "grad_norm": 753.8233642578125,
      "learning_rate": 1.022e-05,
      "loss": 271.2417,
      "step": 5110
    },
    {
      "epoch": 0.020685447868227232,
      "grad_norm": 905.9545288085938,
      "learning_rate": 1.024e-05,
      "loss": 204.41,
      "step": 5120
    },
    {
      "epoch": 0.020725849133594865,
      "grad_norm": 1900.184326171875,
      "learning_rate": 1.026e-05,
      "loss": 203.9288,
      "step": 5130
    },
    {
      "epoch": 0.020766250398962495,
      "grad_norm": 533.8387451171875,
      "learning_rate": 1.0280000000000002e-05,
      "loss": 186.7773,
      "step": 5140
    },
    {
      "epoch": 0.020806651664330128,
      "grad_norm": 671.30322265625,
      "learning_rate": 1.03e-05,
      "loss": 164.2021,
      "step": 5150
    },
    {
      "epoch": 0.020847052929697758,
      "grad_norm": 764.4326171875,
      "learning_rate": 1.0320000000000001e-05,
      "loss": 234.4929,
      "step": 5160
    },
    {
      "epoch": 0.02088745419506539,
      "grad_norm": 921.5126953125,
      "learning_rate": 1.0340000000000001e-05,
      "loss": 257.7942,
      "step": 5170
    },
    {
      "epoch": 0.02092785546043302,
      "grad_norm": 1028.72412109375,
      "learning_rate": 1.036e-05,
      "loss": 219.407,
      "step": 5180
    },
    {
      "epoch": 0.020968256725800653,
      "grad_norm": 857.1285400390625,
      "learning_rate": 1.038e-05,
      "loss": 267.7374,
      "step": 5190
    },
    {
      "epoch": 0.021008657991168283,
      "grad_norm": 733.449951171875,
      "learning_rate": 1.04e-05,
      "loss": 259.0882,
      "step": 5200
    },
    {
      "epoch": 0.021049059256535916,
      "grad_norm": 853.8319091796875,
      "learning_rate": 1.042e-05,
      "loss": 236.0859,
      "step": 5210
    },
    {
      "epoch": 0.021089460521903546,
      "grad_norm": 780.4057006835938,
      "learning_rate": 1.0440000000000002e-05,
      "loss": 235.1703,
      "step": 5220
    },
    {
      "epoch": 0.02112986178727118,
      "grad_norm": 1369.0264892578125,
      "learning_rate": 1.046e-05,
      "loss": 262.7009,
      "step": 5230
    },
    {
      "epoch": 0.021170263052638808,
      "grad_norm": 802.2042236328125,
      "learning_rate": 1.0480000000000001e-05,
      "loss": 317.5121,
      "step": 5240
    },
    {
      "epoch": 0.02121066431800644,
      "grad_norm": 3069.8564453125,
      "learning_rate": 1.05e-05,
      "loss": 266.7954,
      "step": 5250
    },
    {
      "epoch": 0.02125106558337407,
      "grad_norm": 1437.6041259765625,
      "learning_rate": 1.0520000000000001e-05,
      "loss": 290.8906,
      "step": 5260
    },
    {
      "epoch": 0.021291466848741704,
      "grad_norm": 1116.5682373046875,
      "learning_rate": 1.0539999999999999e-05,
      "loss": 217.5333,
      "step": 5270
    },
    {
      "epoch": 0.021331868114109333,
      "grad_norm": 674.6525268554688,
      "learning_rate": 1.056e-05,
      "loss": 262.4274,
      "step": 5280
    },
    {
      "epoch": 0.021372269379476967,
      "grad_norm": 980.0327758789062,
      "learning_rate": 1.058e-05,
      "loss": 253.3151,
      "step": 5290
    },
    {
      "epoch": 0.021412670644844596,
      "grad_norm": 1546.599609375,
      "learning_rate": 1.06e-05,
      "loss": 284.0318,
      "step": 5300
    },
    {
      "epoch": 0.02145307191021223,
      "grad_norm": 4072.65478515625,
      "learning_rate": 1.062e-05,
      "loss": 299.3221,
      "step": 5310
    },
    {
      "epoch": 0.02149347317557986,
      "grad_norm": 835.0826416015625,
      "learning_rate": 1.064e-05,
      "loss": 217.7071,
      "step": 5320
    },
    {
      "epoch": 0.021533874440947492,
      "grad_norm": 825.3660278320312,
      "learning_rate": 1.0660000000000001e-05,
      "loss": 250.3456,
      "step": 5330
    },
    {
      "epoch": 0.02157427570631512,
      "grad_norm": 1196.41455078125,
      "learning_rate": 1.0680000000000001e-05,
      "loss": 217.1434,
      "step": 5340
    },
    {
      "epoch": 0.021614676971682754,
      "grad_norm": 556.876220703125,
      "learning_rate": 1.0700000000000001e-05,
      "loss": 239.3144,
      "step": 5350
    },
    {
      "epoch": 0.021655078237050384,
      "grad_norm": 602.3602905273438,
      "learning_rate": 1.072e-05,
      "loss": 308.0275,
      "step": 5360
    },
    {
      "epoch": 0.021695479502418017,
      "grad_norm": 747.0403442382812,
      "learning_rate": 1.074e-05,
      "loss": 249.5218,
      "step": 5370
    },
    {
      "epoch": 0.021735880767785647,
      "grad_norm": 2065.047607421875,
      "learning_rate": 1.076e-05,
      "loss": 289.0896,
      "step": 5380
    },
    {
      "epoch": 0.02177628203315328,
      "grad_norm": 1931.2476806640625,
      "learning_rate": 1.0780000000000002e-05,
      "loss": 270.5146,
      "step": 5390
    },
    {
      "epoch": 0.02181668329852091,
      "grad_norm": 5352.14013671875,
      "learning_rate": 1.08e-05,
      "loss": 330.6276,
      "step": 5400
    },
    {
      "epoch": 0.021857084563888542,
      "grad_norm": 750.8521728515625,
      "learning_rate": 1.0820000000000001e-05,
      "loss": 260.1782,
      "step": 5410
    },
    {
      "epoch": 0.021897485829256172,
      "grad_norm": 1566.5867919921875,
      "learning_rate": 1.084e-05,
      "loss": 277.7564,
      "step": 5420
    },
    {
      "epoch": 0.021937887094623805,
      "grad_norm": 4719.2412109375,
      "learning_rate": 1.0860000000000001e-05,
      "loss": 270.119,
      "step": 5430
    },
    {
      "epoch": 0.021978288359991435,
      "grad_norm": 886.023681640625,
      "learning_rate": 1.088e-05,
      "loss": 246.386,
      "step": 5440
    },
    {
      "epoch": 0.022018689625359068,
      "grad_norm": 15289.64453125,
      "learning_rate": 1.09e-05,
      "loss": 360.9848,
      "step": 5450
    },
    {
      "epoch": 0.022059090890726697,
      "grad_norm": 854.2662353515625,
      "learning_rate": 1.092e-05,
      "loss": 234.6689,
      "step": 5460
    },
    {
      "epoch": 0.02209949215609433,
      "grad_norm": 1195.251953125,
      "learning_rate": 1.094e-05,
      "loss": 233.2883,
      "step": 5470
    },
    {
      "epoch": 0.02213989342146196,
      "grad_norm": 2974.2724609375,
      "learning_rate": 1.096e-05,
      "loss": 256.327,
      "step": 5480
    },
    {
      "epoch": 0.022180294686829593,
      "grad_norm": 1651.259765625,
      "learning_rate": 1.098e-05,
      "loss": 258.9452,
      "step": 5490
    },
    {
      "epoch": 0.022220695952197222,
      "grad_norm": 1866.4769287109375,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 277.225,
      "step": 5500
    },
    {
      "epoch": 0.022261097217564856,
      "grad_norm": 2012.682861328125,
      "learning_rate": 1.1020000000000001e-05,
      "loss": 213.3877,
      "step": 5510
    },
    {
      "epoch": 0.022301498482932485,
      "grad_norm": 787.662353515625,
      "learning_rate": 1.1040000000000001e-05,
      "loss": 223.892,
      "step": 5520
    },
    {
      "epoch": 0.022341899748300118,
      "grad_norm": 9214.21484375,
      "learning_rate": 1.106e-05,
      "loss": 217.7184,
      "step": 5530
    },
    {
      "epoch": 0.022382301013667748,
      "grad_norm": 1571.4190673828125,
      "learning_rate": 1.108e-05,
      "loss": 268.1608,
      "step": 5540
    },
    {
      "epoch": 0.02242270227903538,
      "grad_norm": 1564.1744384765625,
      "learning_rate": 1.11e-05,
      "loss": 259.3926,
      "step": 5550
    },
    {
      "epoch": 0.02246310354440301,
      "grad_norm": 676.7450561523438,
      "learning_rate": 1.112e-05,
      "loss": 205.3203,
      "step": 5560
    },
    {
      "epoch": 0.022503504809770643,
      "grad_norm": 1126.58837890625,
      "learning_rate": 1.114e-05,
      "loss": 127.3151,
      "step": 5570
    },
    {
      "epoch": 0.022543906075138273,
      "grad_norm": 3336.22607421875,
      "learning_rate": 1.1160000000000002e-05,
      "loss": 226.2071,
      "step": 5580
    },
    {
      "epoch": 0.022584307340505906,
      "grad_norm": 416.8356018066406,
      "learning_rate": 1.118e-05,
      "loss": 257.1062,
      "step": 5590
    },
    {
      "epoch": 0.022624708605873536,
      "grad_norm": 653.0436401367188,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 198.1556,
      "step": 5600
    },
    {
      "epoch": 0.02266510987124117,
      "grad_norm": 5374.44091796875,
      "learning_rate": 1.122e-05,
      "loss": 230.6045,
      "step": 5610
    },
    {
      "epoch": 0.0227055111366088,
      "grad_norm": 1517.1585693359375,
      "learning_rate": 1.124e-05,
      "loss": 217.2467,
      "step": 5620
    },
    {
      "epoch": 0.02274591240197643,
      "grad_norm": 1149.62646484375,
      "learning_rate": 1.126e-05,
      "loss": 294.9917,
      "step": 5630
    },
    {
      "epoch": 0.02278631366734406,
      "grad_norm": 1075.8157958984375,
      "learning_rate": 1.128e-05,
      "loss": 273.2365,
      "step": 5640
    },
    {
      "epoch": 0.022826714932711694,
      "grad_norm": 1706.23388671875,
      "learning_rate": 1.13e-05,
      "loss": 239.202,
      "step": 5650
    },
    {
      "epoch": 0.022867116198079324,
      "grad_norm": 939.1124877929688,
      "learning_rate": 1.132e-05,
      "loss": 262.5748,
      "step": 5660
    },
    {
      "epoch": 0.022907517463446957,
      "grad_norm": 1178.756591796875,
      "learning_rate": 1.134e-05,
      "loss": 205.6744,
      "step": 5670
    },
    {
      "epoch": 0.022947918728814586,
      "grad_norm": 402.2773132324219,
      "learning_rate": 1.1360000000000001e-05,
      "loss": 192.2235,
      "step": 5680
    },
    {
      "epoch": 0.02298831999418222,
      "grad_norm": 0.0,
      "learning_rate": 1.1380000000000001e-05,
      "loss": 258.3543,
      "step": 5690
    },
    {
      "epoch": 0.02302872125954985,
      "grad_norm": 1016.2149658203125,
      "learning_rate": 1.1400000000000001e-05,
      "loss": 217.5345,
      "step": 5700
    },
    {
      "epoch": 0.023069122524917482,
      "grad_norm": 1061.48828125,
      "learning_rate": 1.142e-05,
      "loss": 227.8594,
      "step": 5710
    },
    {
      "epoch": 0.02310952379028511,
      "grad_norm": 481.3313293457031,
      "learning_rate": 1.144e-05,
      "loss": 194.8761,
      "step": 5720
    },
    {
      "epoch": 0.023149925055652745,
      "grad_norm": 782.8972778320312,
      "learning_rate": 1.146e-05,
      "loss": 173.4156,
      "step": 5730
    },
    {
      "epoch": 0.023190326321020374,
      "grad_norm": 672.4293212890625,
      "learning_rate": 1.148e-05,
      "loss": 206.9097,
      "step": 5740
    },
    {
      "epoch": 0.023230727586388007,
      "grad_norm": 2217.28369140625,
      "learning_rate": 1.1500000000000002e-05,
      "loss": 238.6014,
      "step": 5750
    },
    {
      "epoch": 0.023271128851755637,
      "grad_norm": 3414.3115234375,
      "learning_rate": 1.152e-05,
      "loss": 324.1321,
      "step": 5760
    },
    {
      "epoch": 0.02331153011712327,
      "grad_norm": 864.6550903320312,
      "learning_rate": 1.1540000000000001e-05,
      "loss": 213.1406,
      "step": 5770
    },
    {
      "epoch": 0.0233519313824909,
      "grad_norm": 752.6487426757812,
      "learning_rate": 1.156e-05,
      "loss": 136.1037,
      "step": 5780
    },
    {
      "epoch": 0.023392332647858533,
      "grad_norm": 550.0790405273438,
      "learning_rate": 1.1580000000000001e-05,
      "loss": 295.7111,
      "step": 5790
    },
    {
      "epoch": 0.023432733913226162,
      "grad_norm": 918.3283081054688,
      "learning_rate": 1.16e-05,
      "loss": 234.3793,
      "step": 5800
    },
    {
      "epoch": 0.023473135178593795,
      "grad_norm": 1468.6207275390625,
      "learning_rate": 1.162e-05,
      "loss": 191.9201,
      "step": 5810
    },
    {
      "epoch": 0.023513536443961425,
      "grad_norm": 1899.0367431640625,
      "learning_rate": 1.164e-05,
      "loss": 234.0571,
      "step": 5820
    },
    {
      "epoch": 0.023553937709329058,
      "grad_norm": 684.876953125,
      "learning_rate": 1.166e-05,
      "loss": 180.286,
      "step": 5830
    },
    {
      "epoch": 0.023594338974696687,
      "grad_norm": 1083.017578125,
      "learning_rate": 1.168e-05,
      "loss": 300.8844,
      "step": 5840
    },
    {
      "epoch": 0.02363474024006432,
      "grad_norm": 1754.2984619140625,
      "learning_rate": 1.1700000000000001e-05,
      "loss": 391.4959,
      "step": 5850
    },
    {
      "epoch": 0.02367514150543195,
      "grad_norm": 945.9032592773438,
      "learning_rate": 1.172e-05,
      "loss": 272.7161,
      "step": 5860
    },
    {
      "epoch": 0.023715542770799583,
      "grad_norm": 1564.7225341796875,
      "learning_rate": 1.1740000000000001e-05,
      "loss": 233.5665,
      "step": 5870
    },
    {
      "epoch": 0.023755944036167213,
      "grad_norm": 523.8766479492188,
      "learning_rate": 1.1760000000000001e-05,
      "loss": 204.2223,
      "step": 5880
    },
    {
      "epoch": 0.023796345301534846,
      "grad_norm": 1801.8782958984375,
      "learning_rate": 1.178e-05,
      "loss": 215.9124,
      "step": 5890
    },
    {
      "epoch": 0.023836746566902475,
      "grad_norm": 2132.8076171875,
      "learning_rate": 1.18e-05,
      "loss": 328.9819,
      "step": 5900
    },
    {
      "epoch": 0.02387714783227011,
      "grad_norm": 1254.7607421875,
      "learning_rate": 1.182e-05,
      "loss": 249.0257,
      "step": 5910
    },
    {
      "epoch": 0.023917549097637738,
      "grad_norm": 873.1114501953125,
      "learning_rate": 1.1840000000000002e-05,
      "loss": 156.6504,
      "step": 5920
    },
    {
      "epoch": 0.02395795036300537,
      "grad_norm": 2675.64892578125,
      "learning_rate": 1.186e-05,
      "loss": 213.5535,
      "step": 5930
    },
    {
      "epoch": 0.023998351628373,
      "grad_norm": 744.88671875,
      "learning_rate": 1.1880000000000001e-05,
      "loss": 189.1653,
      "step": 5940
    },
    {
      "epoch": 0.024038752893740634,
      "grad_norm": 979.5391845703125,
      "learning_rate": 1.19e-05,
      "loss": 223.2056,
      "step": 5950
    },
    {
      "epoch": 0.024079154159108263,
      "grad_norm": 872.666015625,
      "learning_rate": 1.1920000000000001e-05,
      "loss": 194.5632,
      "step": 5960
    },
    {
      "epoch": 0.024119555424475896,
      "grad_norm": 1239.266845703125,
      "learning_rate": 1.1940000000000001e-05,
      "loss": 221.0386,
      "step": 5970
    },
    {
      "epoch": 0.024159956689843526,
      "grad_norm": 1718.065673828125,
      "learning_rate": 1.196e-05,
      "loss": 237.8773,
      "step": 5980
    },
    {
      "epoch": 0.02420035795521116,
      "grad_norm": 1780.784423828125,
      "learning_rate": 1.198e-05,
      "loss": 119.0116,
      "step": 5990
    },
    {
      "epoch": 0.02424075922057879,
      "grad_norm": 921.1470947265625,
      "learning_rate": 1.2e-05,
      "loss": 225.7212,
      "step": 6000
    },
    {
      "epoch": 0.02428116048594642,
      "grad_norm": 2161.641845703125,
      "learning_rate": 1.202e-05,
      "loss": 181.6732,
      "step": 6010
    },
    {
      "epoch": 0.02432156175131405,
      "grad_norm": 1104.2542724609375,
      "learning_rate": 1.204e-05,
      "loss": 187.8781,
      "step": 6020
    },
    {
      "epoch": 0.024361963016681684,
      "grad_norm": 1214.7061767578125,
      "learning_rate": 1.206e-05,
      "loss": 221.4846,
      "step": 6030
    },
    {
      "epoch": 0.024402364282049314,
      "grad_norm": 1222.162841796875,
      "learning_rate": 1.2080000000000001e-05,
      "loss": 172.2742,
      "step": 6040
    },
    {
      "epoch": 0.024442765547416947,
      "grad_norm": 828.8873291015625,
      "learning_rate": 1.2100000000000001e-05,
      "loss": 227.2261,
      "step": 6050
    },
    {
      "epoch": 0.024483166812784576,
      "grad_norm": 621.658447265625,
      "learning_rate": 1.2120000000000001e-05,
      "loss": 236.3511,
      "step": 6060
    },
    {
      "epoch": 0.02452356807815221,
      "grad_norm": 1036.8626708984375,
      "learning_rate": 1.214e-05,
      "loss": 233.8868,
      "step": 6070
    },
    {
      "epoch": 0.02456396934351984,
      "grad_norm": 1811.7554931640625,
      "learning_rate": 1.216e-05,
      "loss": 238.5441,
      "step": 6080
    },
    {
      "epoch": 0.02460437060888747,
      "grad_norm": 1012.0717163085938,
      "learning_rate": 1.2180000000000002e-05,
      "loss": 234.7135,
      "step": 6090
    },
    {
      "epoch": 0.0246447718742551,
      "grad_norm": 1480.242919921875,
      "learning_rate": 1.22e-05,
      "loss": 275.5092,
      "step": 6100
    },
    {
      "epoch": 0.02468517313962273,
      "grad_norm": 1884.02001953125,
      "learning_rate": 1.2220000000000002e-05,
      "loss": 242.7474,
      "step": 6110
    },
    {
      "epoch": 0.024725574404990364,
      "grad_norm": 1144.38916015625,
      "learning_rate": 1.224e-05,
      "loss": 368.5085,
      "step": 6120
    },
    {
      "epoch": 0.024765975670357994,
      "grad_norm": 1017.46875,
      "learning_rate": 1.2260000000000001e-05,
      "loss": 224.2549,
      "step": 6130
    },
    {
      "epoch": 0.024806376935725627,
      "grad_norm": 1646.821044921875,
      "learning_rate": 1.2280000000000001e-05,
      "loss": 221.9995,
      "step": 6140
    },
    {
      "epoch": 0.024846778201093257,
      "grad_norm": 719.86376953125,
      "learning_rate": 1.23e-05,
      "loss": 221.8839,
      "step": 6150
    },
    {
      "epoch": 0.02488717946646089,
      "grad_norm": 865.7079467773438,
      "learning_rate": 1.232e-05,
      "loss": 216.4899,
      "step": 6160
    },
    {
      "epoch": 0.02492758073182852,
      "grad_norm": 837.6893310546875,
      "learning_rate": 1.234e-05,
      "loss": 212.6082,
      "step": 6170
    },
    {
      "epoch": 0.024967981997196152,
      "grad_norm": 619.9283447265625,
      "learning_rate": 1.236e-05,
      "loss": 231.417,
      "step": 6180
    },
    {
      "epoch": 0.025008383262563782,
      "grad_norm": 938.736328125,
      "learning_rate": 1.238e-05,
      "loss": 222.0059,
      "step": 6190
    },
    {
      "epoch": 0.025048784527931415,
      "grad_norm": 768.3204956054688,
      "learning_rate": 1.24e-05,
      "loss": 269.6617,
      "step": 6200
    },
    {
      "epoch": 0.025089185793299044,
      "grad_norm": 25211.0859375,
      "learning_rate": 1.2420000000000001e-05,
      "loss": 283.9776,
      "step": 6210
    },
    {
      "epoch": 0.025129587058666678,
      "grad_norm": 1000.73046875,
      "learning_rate": 1.244e-05,
      "loss": 237.4733,
      "step": 6220
    },
    {
      "epoch": 0.025169988324034307,
      "grad_norm": 1383.859375,
      "learning_rate": 1.2460000000000001e-05,
      "loss": 207.5625,
      "step": 6230
    },
    {
      "epoch": 0.02521038958940194,
      "grad_norm": 1157.95654296875,
      "learning_rate": 1.248e-05,
      "loss": 188.4448,
      "step": 6240
    },
    {
      "epoch": 0.02525079085476957,
      "grad_norm": 437.2555847167969,
      "learning_rate": 1.25e-05,
      "loss": 253.4308,
      "step": 6250
    },
    {
      "epoch": 0.025291192120137203,
      "grad_norm": 735.6442260742188,
      "learning_rate": 1.252e-05,
      "loss": 239.4786,
      "step": 6260
    },
    {
      "epoch": 0.025331593385504832,
      "grad_norm": 3905.33935546875,
      "learning_rate": 1.2540000000000002e-05,
      "loss": 340.7768,
      "step": 6270
    },
    {
      "epoch": 0.025371994650872465,
      "grad_norm": 1644.240234375,
      "learning_rate": 1.256e-05,
      "loss": 221.2143,
      "step": 6280
    },
    {
      "epoch": 0.025412395916240095,
      "grad_norm": 745.5984497070312,
      "learning_rate": 1.258e-05,
      "loss": 154.9017,
      "step": 6290
    },
    {
      "epoch": 0.025452797181607728,
      "grad_norm": 2822.624267578125,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 193.9914,
      "step": 6300
    },
    {
      "epoch": 0.025493198446975358,
      "grad_norm": 415.230712890625,
      "learning_rate": 1.2620000000000001e-05,
      "loss": 150.345,
      "step": 6310
    },
    {
      "epoch": 0.02553359971234299,
      "grad_norm": 2347.072021484375,
      "learning_rate": 1.2640000000000003e-05,
      "loss": 239.6525,
      "step": 6320
    },
    {
      "epoch": 0.02557400097771062,
      "grad_norm": 1561.8658447265625,
      "learning_rate": 1.2659999999999999e-05,
      "loss": 288.6605,
      "step": 6330
    },
    {
      "epoch": 0.025614402243078253,
      "grad_norm": 651.1387329101562,
      "learning_rate": 1.268e-05,
      "loss": 266.1772,
      "step": 6340
    },
    {
      "epoch": 0.025654803508445883,
      "grad_norm": 544.3804931640625,
      "learning_rate": 1.27e-05,
      "loss": 273.7383,
      "step": 6350
    },
    {
      "epoch": 0.025695204773813516,
      "grad_norm": 1525.947021484375,
      "learning_rate": 1.2720000000000002e-05,
      "loss": 251.1667,
      "step": 6360
    },
    {
      "epoch": 0.025735606039181146,
      "grad_norm": 1018.3317260742188,
      "learning_rate": 1.2740000000000002e-05,
      "loss": 210.4547,
      "step": 6370
    },
    {
      "epoch": 0.02577600730454878,
      "grad_norm": 911.9342651367188,
      "learning_rate": 1.276e-05,
      "loss": 278.7565,
      "step": 6380
    },
    {
      "epoch": 0.025816408569916408,
      "grad_norm": 3710.921630859375,
      "learning_rate": 1.278e-05,
      "loss": 229.9252,
      "step": 6390
    },
    {
      "epoch": 0.02585680983528404,
      "grad_norm": 709.6052856445312,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 197.896,
      "step": 6400
    },
    {
      "epoch": 0.02589721110065167,
      "grad_norm": 688.3367919921875,
      "learning_rate": 1.2820000000000001e-05,
      "loss": 205.0635,
      "step": 6410
    },
    {
      "epoch": 0.025937612366019304,
      "grad_norm": 1232.222412109375,
      "learning_rate": 1.2839999999999999e-05,
      "loss": 196.4835,
      "step": 6420
    },
    {
      "epoch": 0.025978013631386934,
      "grad_norm": 665.0555419921875,
      "learning_rate": 1.286e-05,
      "loss": 211.4123,
      "step": 6430
    },
    {
      "epoch": 0.026018414896754567,
      "grad_norm": 2699.49755859375,
      "learning_rate": 1.288e-05,
      "loss": 266.4242,
      "step": 6440
    },
    {
      "epoch": 0.026058816162122196,
      "grad_norm": 1464.1005859375,
      "learning_rate": 1.29e-05,
      "loss": 271.645,
      "step": 6450
    },
    {
      "epoch": 0.02609921742748983,
      "grad_norm": 605.3104858398438,
      "learning_rate": 1.2920000000000002e-05,
      "loss": 158.6754,
      "step": 6460
    },
    {
      "epoch": 0.02613961869285746,
      "grad_norm": 643.6430053710938,
      "learning_rate": 1.294e-05,
      "loss": 246.8146,
      "step": 6470
    },
    {
      "epoch": 0.026180019958225092,
      "grad_norm": 780.9293823242188,
      "learning_rate": 1.296e-05,
      "loss": 270.3225,
      "step": 6480
    },
    {
      "epoch": 0.02622042122359272,
      "grad_norm": 2428.1328125,
      "learning_rate": 1.2980000000000001e-05,
      "loss": 218.6574,
      "step": 6490
    },
    {
      "epoch": 0.026260822488960354,
      "grad_norm": 737.8549194335938,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 241.2201,
      "step": 6500
    },
    {
      "epoch": 0.026301223754327984,
      "grad_norm": 690.761474609375,
      "learning_rate": 1.3020000000000002e-05,
      "loss": 215.8646,
      "step": 6510
    },
    {
      "epoch": 0.026341625019695617,
      "grad_norm": 693.7470092773438,
      "learning_rate": 1.3039999999999999e-05,
      "loss": 226.2056,
      "step": 6520
    },
    {
      "epoch": 0.026382026285063247,
      "grad_norm": 1125.502197265625,
      "learning_rate": 1.306e-05,
      "loss": 250.0323,
      "step": 6530
    },
    {
      "epoch": 0.02642242755043088,
      "grad_norm": 3172.219482421875,
      "learning_rate": 1.308e-05,
      "loss": 247.7785,
      "step": 6540
    },
    {
      "epoch": 0.02646282881579851,
      "grad_norm": 1940.4234619140625,
      "learning_rate": 1.3100000000000002e-05,
      "loss": 178.4448,
      "step": 6550
    },
    {
      "epoch": 0.026503230081166142,
      "grad_norm": 632.4578857421875,
      "learning_rate": 1.3120000000000001e-05,
      "loss": 206.7619,
      "step": 6560
    },
    {
      "epoch": 0.026543631346533772,
      "grad_norm": 1491.89306640625,
      "learning_rate": 1.314e-05,
      "loss": 164.1459,
      "step": 6570
    },
    {
      "epoch": 0.026584032611901405,
      "grad_norm": 831.048828125,
      "learning_rate": 1.316e-05,
      "loss": 202.0544,
      "step": 6580
    },
    {
      "epoch": 0.026624433877269035,
      "grad_norm": 812.6109619140625,
      "learning_rate": 1.3180000000000001e-05,
      "loss": 127.2249,
      "step": 6590
    },
    {
      "epoch": 0.026664835142636668,
      "grad_norm": 678.7022094726562,
      "learning_rate": 1.32e-05,
      "loss": 262.2538,
      "step": 6600
    },
    {
      "epoch": 0.026705236408004297,
      "grad_norm": 1193.4014892578125,
      "learning_rate": 1.3220000000000002e-05,
      "loss": 258.1862,
      "step": 6610
    },
    {
      "epoch": 0.02674563767337193,
      "grad_norm": 881.2777099609375,
      "learning_rate": 1.324e-05,
      "loss": 181.1106,
      "step": 6620
    },
    {
      "epoch": 0.02678603893873956,
      "grad_norm": 751.0634765625,
      "learning_rate": 1.326e-05,
      "loss": 328.8563,
      "step": 6630
    },
    {
      "epoch": 0.026826440204107193,
      "grad_norm": 810.8790893554688,
      "learning_rate": 1.3280000000000002e-05,
      "loss": 175.6053,
      "step": 6640
    },
    {
      "epoch": 0.026866841469474823,
      "grad_norm": 544.2944946289062,
      "learning_rate": 1.3300000000000001e-05,
      "loss": 172.9304,
      "step": 6650
    },
    {
      "epoch": 0.026907242734842456,
      "grad_norm": 1003.13818359375,
      "learning_rate": 1.3320000000000001e-05,
      "loss": 180.5678,
      "step": 6660
    },
    {
      "epoch": 0.026947644000210085,
      "grad_norm": 1979.7833251953125,
      "learning_rate": 1.334e-05,
      "loss": 212.5934,
      "step": 6670
    },
    {
      "epoch": 0.026988045265577718,
      "grad_norm": 660.824462890625,
      "learning_rate": 1.336e-05,
      "loss": 170.068,
      "step": 6680
    },
    {
      "epoch": 0.027028446530945348,
      "grad_norm": 753.3848876953125,
      "learning_rate": 1.338e-05,
      "loss": 226.8918,
      "step": 6690
    },
    {
      "epoch": 0.02706884779631298,
      "grad_norm": 1572.1107177734375,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 285.1505,
      "step": 6700
    },
    {
      "epoch": 0.02710924906168061,
      "grad_norm": 727.326416015625,
      "learning_rate": 1.3420000000000002e-05,
      "loss": 259.9044,
      "step": 6710
    },
    {
      "epoch": 0.027149650327048244,
      "grad_norm": 1423.973388671875,
      "learning_rate": 1.344e-05,
      "loss": 208.3222,
      "step": 6720
    },
    {
      "epoch": 0.027190051592415873,
      "grad_norm": 3651.573974609375,
      "learning_rate": 1.346e-05,
      "loss": 276.9352,
      "step": 6730
    },
    {
      "epoch": 0.027230452857783506,
      "grad_norm": 986.3583984375,
      "learning_rate": 1.3480000000000001e-05,
      "loss": 233.0887,
      "step": 6740
    },
    {
      "epoch": 0.027270854123151136,
      "grad_norm": 1490.3424072265625,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 308.2051,
      "step": 6750
    },
    {
      "epoch": 0.02731125538851877,
      "grad_norm": 1850.4970703125,
      "learning_rate": 1.352e-05,
      "loss": 220.4273,
      "step": 6760
    },
    {
      "epoch": 0.0273516566538864,
      "grad_norm": 558.9097900390625,
      "learning_rate": 1.3539999999999999e-05,
      "loss": 192.5242,
      "step": 6770
    },
    {
      "epoch": 0.02739205791925403,
      "grad_norm": 1742.455322265625,
      "learning_rate": 1.356e-05,
      "loss": 228.9638,
      "step": 6780
    },
    {
      "epoch": 0.02743245918462166,
      "grad_norm": 6109.70947265625,
      "learning_rate": 1.358e-05,
      "loss": 218.1945,
      "step": 6790
    },
    {
      "epoch": 0.027472860449989294,
      "grad_norm": 2294.8359375,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 259.6226,
      "step": 6800
    },
    {
      "epoch": 0.027513261715356924,
      "grad_norm": 1345.371826171875,
      "learning_rate": 1.362e-05,
      "loss": 208.2849,
      "step": 6810
    },
    {
      "epoch": 0.027553662980724557,
      "grad_norm": 631.337158203125,
      "learning_rate": 1.364e-05,
      "loss": 174.1924,
      "step": 6820
    },
    {
      "epoch": 0.027594064246092186,
      "grad_norm": 2324.395751953125,
      "learning_rate": 1.3660000000000001e-05,
      "loss": 217.651,
      "step": 6830
    },
    {
      "epoch": 0.02763446551145982,
      "grad_norm": 1169.88916015625,
      "learning_rate": 1.3680000000000001e-05,
      "loss": 152.5202,
      "step": 6840
    },
    {
      "epoch": 0.02767486677682745,
      "grad_norm": 580.0879516601562,
      "learning_rate": 1.3700000000000001e-05,
      "loss": 260.7849,
      "step": 6850
    },
    {
      "epoch": 0.027715268042195082,
      "grad_norm": 1336.279052734375,
      "learning_rate": 1.3719999999999999e-05,
      "loss": 184.8877,
      "step": 6860
    },
    {
      "epoch": 0.02775566930756271,
      "grad_norm": 1506.610107421875,
      "learning_rate": 1.374e-05,
      "loss": 236.7283,
      "step": 6870
    },
    {
      "epoch": 0.027796070572930345,
      "grad_norm": 1100.15234375,
      "learning_rate": 1.376e-05,
      "loss": 263.8396,
      "step": 6880
    },
    {
      "epoch": 0.027836471838297974,
      "grad_norm": 794.3699340820312,
      "learning_rate": 1.3780000000000002e-05,
      "loss": 290.786,
      "step": 6890
    },
    {
      "epoch": 0.027876873103665607,
      "grad_norm": 2009.521240234375,
      "learning_rate": 1.3800000000000002e-05,
      "loss": 173.6581,
      "step": 6900
    },
    {
      "epoch": 0.027917274369033237,
      "grad_norm": 1658.059814453125,
      "learning_rate": 1.382e-05,
      "loss": 210.0458,
      "step": 6910
    },
    {
      "epoch": 0.02795767563440087,
      "grad_norm": 632.90869140625,
      "learning_rate": 1.384e-05,
      "loss": 179.4226,
      "step": 6920
    },
    {
      "epoch": 0.0279980768997685,
      "grad_norm": 1309.6959228515625,
      "learning_rate": 1.3860000000000001e-05,
      "loss": 168.1927,
      "step": 6930
    },
    {
      "epoch": 0.028038478165136133,
      "grad_norm": 1135.08935546875,
      "learning_rate": 1.3880000000000001e-05,
      "loss": 189.0599,
      "step": 6940
    },
    {
      "epoch": 0.028078879430503762,
      "grad_norm": 642.1088256835938,
      "learning_rate": 1.3900000000000002e-05,
      "loss": 191.8304,
      "step": 6950
    },
    {
      "epoch": 0.028119280695871395,
      "grad_norm": 1067.7398681640625,
      "learning_rate": 1.3919999999999999e-05,
      "loss": 153.804,
      "step": 6960
    },
    {
      "epoch": 0.028159681961239025,
      "grad_norm": 0.0,
      "learning_rate": 1.394e-05,
      "loss": 131.2499,
      "step": 6970
    },
    {
      "epoch": 0.028200083226606658,
      "grad_norm": 4239.3876953125,
      "learning_rate": 1.396e-05,
      "loss": 219.0053,
      "step": 6980
    },
    {
      "epoch": 0.028240484491974287,
      "grad_norm": 1219.2093505859375,
      "learning_rate": 1.3980000000000002e-05,
      "loss": 216.1493,
      "step": 6990
    },
    {
      "epoch": 0.02828088575734192,
      "grad_norm": 829.5226440429688,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 197.1629,
      "step": 7000
    },
    {
      "epoch": 0.02832128702270955,
      "grad_norm": 783.1107177734375,
      "learning_rate": 1.402e-05,
      "loss": 197.8331,
      "step": 7010
    },
    {
      "epoch": 0.028361688288077183,
      "grad_norm": 1040.990966796875,
      "learning_rate": 1.4040000000000001e-05,
      "loss": 234.4111,
      "step": 7020
    },
    {
      "epoch": 0.028402089553444813,
      "grad_norm": 952.650390625,
      "learning_rate": 1.4060000000000001e-05,
      "loss": 138.1043,
      "step": 7030
    },
    {
      "epoch": 0.028442490818812446,
      "grad_norm": 2120.21337890625,
      "learning_rate": 1.408e-05,
      "loss": 190.9683,
      "step": 7040
    },
    {
      "epoch": 0.028482892084180075,
      "grad_norm": 1032.8970947265625,
      "learning_rate": 1.4099999999999999e-05,
      "loss": 211.108,
      "step": 7050
    },
    {
      "epoch": 0.02852329334954771,
      "grad_norm": 845.619384765625,
      "learning_rate": 1.412e-05,
      "loss": 249.2893,
      "step": 7060
    },
    {
      "epoch": 0.028563694614915338,
      "grad_norm": 2072.714599609375,
      "learning_rate": 1.414e-05,
      "loss": 148.9613,
      "step": 7070
    },
    {
      "epoch": 0.02860409588028297,
      "grad_norm": 1205.255615234375,
      "learning_rate": 1.4160000000000002e-05,
      "loss": 208.2003,
      "step": 7080
    },
    {
      "epoch": 0.0286444971456506,
      "grad_norm": 1514.6177978515625,
      "learning_rate": 1.4180000000000001e-05,
      "loss": 211.4245,
      "step": 7090
    },
    {
      "epoch": 0.028684898411018234,
      "grad_norm": 599.4893188476562,
      "learning_rate": 1.42e-05,
      "loss": 270.9207,
      "step": 7100
    },
    {
      "epoch": 0.028725299676385863,
      "grad_norm": 622.9778442382812,
      "learning_rate": 1.422e-05,
      "loss": 248.9024,
      "step": 7110
    },
    {
      "epoch": 0.028765700941753496,
      "grad_norm": 827.0371704101562,
      "learning_rate": 1.4240000000000001e-05,
      "loss": 228.1889,
      "step": 7120
    },
    {
      "epoch": 0.028806102207121126,
      "grad_norm": 1157.1502685546875,
      "learning_rate": 1.426e-05,
      "loss": 181.2518,
      "step": 7130
    },
    {
      "epoch": 0.02884650347248876,
      "grad_norm": 1035.8681640625,
      "learning_rate": 1.4280000000000002e-05,
      "loss": 223.2644,
      "step": 7140
    },
    {
      "epoch": 0.02888690473785639,
      "grad_norm": 1198.443359375,
      "learning_rate": 1.43e-05,
      "loss": 193.8474,
      "step": 7150
    },
    {
      "epoch": 0.02892730600322402,
      "grad_norm": 1005.276611328125,
      "learning_rate": 1.432e-05,
      "loss": 250.6531,
      "step": 7160
    },
    {
      "epoch": 0.02896770726859165,
      "grad_norm": 2052.37158203125,
      "learning_rate": 1.434e-05,
      "loss": 184.0699,
      "step": 7170
    },
    {
      "epoch": 0.029008108533959284,
      "grad_norm": 22271.3046875,
      "learning_rate": 1.4360000000000001e-05,
      "loss": 298.0969,
      "step": 7180
    },
    {
      "epoch": 0.029048509799326914,
      "grad_norm": 1663.1221923828125,
      "learning_rate": 1.4380000000000001e-05,
      "loss": 256.0283,
      "step": 7190
    },
    {
      "epoch": 0.029088911064694547,
      "grad_norm": 5215.501953125,
      "learning_rate": 1.44e-05,
      "loss": 208.372,
      "step": 7200
    },
    {
      "epoch": 0.029129312330062176,
      "grad_norm": 1427.092529296875,
      "learning_rate": 1.4420000000000001e-05,
      "loss": 196.7174,
      "step": 7210
    },
    {
      "epoch": 0.02916971359542981,
      "grad_norm": 755.31494140625,
      "learning_rate": 1.444e-05,
      "loss": 207.5049,
      "step": 7220
    },
    {
      "epoch": 0.02921011486079744,
      "grad_norm": 1840.75927734375,
      "learning_rate": 1.4460000000000002e-05,
      "loss": 246.7446,
      "step": 7230
    },
    {
      "epoch": 0.029250516126165072,
      "grad_norm": 1352.86767578125,
      "learning_rate": 1.4480000000000002e-05,
      "loss": 193.7286,
      "step": 7240
    },
    {
      "epoch": 0.029290917391532702,
      "grad_norm": 1088.9520263671875,
      "learning_rate": 1.45e-05,
      "loss": 213.7931,
      "step": 7250
    },
    {
      "epoch": 0.029331318656900335,
      "grad_norm": 2126.0322265625,
      "learning_rate": 1.452e-05,
      "loss": 227.5255,
      "step": 7260
    },
    {
      "epoch": 0.029371719922267964,
      "grad_norm": 0.0,
      "learning_rate": 1.4540000000000001e-05,
      "loss": 149.3132,
      "step": 7270
    },
    {
      "epoch": 0.029412121187635597,
      "grad_norm": 582.3840942382812,
      "learning_rate": 1.4560000000000001e-05,
      "loss": 219.7635,
      "step": 7280
    },
    {
      "epoch": 0.029452522453003227,
      "grad_norm": 935.360107421875,
      "learning_rate": 1.4580000000000003e-05,
      "loss": 219.2881,
      "step": 7290
    },
    {
      "epoch": 0.02949292371837086,
      "grad_norm": 2865.860595703125,
      "learning_rate": 1.4599999999999999e-05,
      "loss": 170.7084,
      "step": 7300
    },
    {
      "epoch": 0.02953332498373849,
      "grad_norm": 1290.148681640625,
      "learning_rate": 1.462e-05,
      "loss": 273.1594,
      "step": 7310
    },
    {
      "epoch": 0.029573726249106123,
      "grad_norm": 1073.5101318359375,
      "learning_rate": 1.464e-05,
      "loss": 276.2752,
      "step": 7320
    },
    {
      "epoch": 0.029614127514473752,
      "grad_norm": 2523.898681640625,
      "learning_rate": 1.4660000000000002e-05,
      "loss": 229.5607,
      "step": 7330
    },
    {
      "epoch": 0.029654528779841385,
      "grad_norm": 2364.947021484375,
      "learning_rate": 1.4680000000000002e-05,
      "loss": 194.6274,
      "step": 7340
    },
    {
      "epoch": 0.029694930045209015,
      "grad_norm": 1139.0018310546875,
      "learning_rate": 1.47e-05,
      "loss": 143.1356,
      "step": 7350
    },
    {
      "epoch": 0.029735331310576648,
      "grad_norm": 1392.792236328125,
      "learning_rate": 1.472e-05,
      "loss": 220.7791,
      "step": 7360
    },
    {
      "epoch": 0.029775732575944278,
      "grad_norm": 1605.3038330078125,
      "learning_rate": 1.4740000000000001e-05,
      "loss": 159.6861,
      "step": 7370
    },
    {
      "epoch": 0.02981613384131191,
      "grad_norm": 1095.6702880859375,
      "learning_rate": 1.4760000000000001e-05,
      "loss": 190.0395,
      "step": 7380
    },
    {
      "epoch": 0.02985653510667954,
      "grad_norm": 3128.165283203125,
      "learning_rate": 1.4779999999999999e-05,
      "loss": 217.6996,
      "step": 7390
    },
    {
      "epoch": 0.029896936372047173,
      "grad_norm": 868.65966796875,
      "learning_rate": 1.48e-05,
      "loss": 303.5258,
      "step": 7400
    },
    {
      "epoch": 0.029937337637414803,
      "grad_norm": 1097.5875244140625,
      "learning_rate": 1.482e-05,
      "loss": 208.767,
      "step": 7410
    },
    {
      "epoch": 0.029977738902782436,
      "grad_norm": 1184.8807373046875,
      "learning_rate": 1.4840000000000002e-05,
      "loss": 212.3098,
      "step": 7420
    },
    {
      "epoch": 0.030018140168150065,
      "grad_norm": 1545.9532470703125,
      "learning_rate": 1.4860000000000002e-05,
      "loss": 142.4259,
      "step": 7430
    },
    {
      "epoch": 0.0300585414335177,
      "grad_norm": 1245.4107666015625,
      "learning_rate": 1.488e-05,
      "loss": 164.8042,
      "step": 7440
    },
    {
      "epoch": 0.030098942698885328,
      "grad_norm": 577.9031982421875,
      "learning_rate": 1.49e-05,
      "loss": 216.6332,
      "step": 7450
    },
    {
      "epoch": 0.03013934396425296,
      "grad_norm": 1043.52392578125,
      "learning_rate": 1.4920000000000001e-05,
      "loss": 247.0811,
      "step": 7460
    },
    {
      "epoch": 0.03017974522962059,
      "grad_norm": 1351.240966796875,
      "learning_rate": 1.4940000000000001e-05,
      "loss": 198.2244,
      "step": 7470
    },
    {
      "epoch": 0.030220146494988224,
      "grad_norm": 1484.0284423828125,
      "learning_rate": 1.4960000000000002e-05,
      "loss": 197.885,
      "step": 7480
    },
    {
      "epoch": 0.030260547760355853,
      "grad_norm": 1378.754638671875,
      "learning_rate": 1.4979999999999999e-05,
      "loss": 201.7562,
      "step": 7490
    },
    {
      "epoch": 0.030300949025723486,
      "grad_norm": 7344.61767578125,
      "learning_rate": 1.5e-05,
      "loss": 231.1991,
      "step": 7500
    },
    {
      "epoch": 0.030341350291091116,
      "grad_norm": 3804.617919921875,
      "learning_rate": 1.502e-05,
      "loss": 310.0097,
      "step": 7510
    },
    {
      "epoch": 0.03038175155645875,
      "grad_norm": 1033.6324462890625,
      "learning_rate": 1.5040000000000002e-05,
      "loss": 198.1659,
      "step": 7520
    },
    {
      "epoch": 0.03042215282182638,
      "grad_norm": 653.5598754882812,
      "learning_rate": 1.5060000000000001e-05,
      "loss": 178.0301,
      "step": 7530
    },
    {
      "epoch": 0.030462554087194012,
      "grad_norm": 1495.2060546875,
      "learning_rate": 1.508e-05,
      "loss": 250.4168,
      "step": 7540
    },
    {
      "epoch": 0.03050295535256164,
      "grad_norm": 3768.185791015625,
      "learning_rate": 1.51e-05,
      "loss": 161.5591,
      "step": 7550
    },
    {
      "epoch": 0.030543356617929274,
      "grad_norm": 966.1246337890625,
      "learning_rate": 1.5120000000000001e-05,
      "loss": 173.9351,
      "step": 7560
    },
    {
      "epoch": 0.030583757883296904,
      "grad_norm": 1045.052490234375,
      "learning_rate": 1.514e-05,
      "loss": 140.5521,
      "step": 7570
    },
    {
      "epoch": 0.030624159148664537,
      "grad_norm": 1348.0595703125,
      "learning_rate": 1.5160000000000002e-05,
      "loss": 318.5935,
      "step": 7580
    },
    {
      "epoch": 0.030664560414032167,
      "grad_norm": 798.9227294921875,
      "learning_rate": 1.518e-05,
      "loss": 269.7896,
      "step": 7590
    },
    {
      "epoch": 0.0307049616793998,
      "grad_norm": 1470.73583984375,
      "learning_rate": 1.52e-05,
      "loss": 282.1007,
      "step": 7600
    },
    {
      "epoch": 0.03074536294476743,
      "grad_norm": 770.0698852539062,
      "learning_rate": 1.5220000000000002e-05,
      "loss": 262.2653,
      "step": 7610
    },
    {
      "epoch": 0.030785764210135062,
      "grad_norm": 571.9855346679688,
      "learning_rate": 1.5240000000000001e-05,
      "loss": 192.3983,
      "step": 7620
    },
    {
      "epoch": 0.030826165475502692,
      "grad_norm": 1054.919921875,
      "learning_rate": 1.5260000000000003e-05,
      "loss": 219.8108,
      "step": 7630
    },
    {
      "epoch": 0.030866566740870325,
      "grad_norm": 556.494873046875,
      "learning_rate": 1.528e-05,
      "loss": 165.5094,
      "step": 7640
    },
    {
      "epoch": 0.030906968006237955,
      "grad_norm": 1616.6512451171875,
      "learning_rate": 1.53e-05,
      "loss": 183.0181,
      "step": 7650
    },
    {
      "epoch": 0.030947369271605588,
      "grad_norm": 678.8358764648438,
      "learning_rate": 1.5320000000000002e-05,
      "loss": 138.8231,
      "step": 7660
    },
    {
      "epoch": 0.030987770536973217,
      "grad_norm": 817.9168090820312,
      "learning_rate": 1.5340000000000002e-05,
      "loss": 166.3831,
      "step": 7670
    },
    {
      "epoch": 0.03102817180234085,
      "grad_norm": 1391.115966796875,
      "learning_rate": 1.536e-05,
      "loss": 265.5596,
      "step": 7680
    },
    {
      "epoch": 0.03106857306770848,
      "grad_norm": 2713.052490234375,
      "learning_rate": 1.538e-05,
      "loss": 221.1131,
      "step": 7690
    },
    {
      "epoch": 0.031108974333076113,
      "grad_norm": 1156.0615234375,
      "learning_rate": 1.54e-05,
      "loss": 188.7715,
      "step": 7700
    },
    {
      "epoch": 0.031149375598443742,
      "grad_norm": 1744.91845703125,
      "learning_rate": 1.542e-05,
      "loss": 203.3742,
      "step": 7710
    },
    {
      "epoch": 0.031189776863811376,
      "grad_norm": 1208.4666748046875,
      "learning_rate": 1.544e-05,
      "loss": 146.0077,
      "step": 7720
    },
    {
      "epoch": 0.031230178129179005,
      "grad_norm": 643.0628662109375,
      "learning_rate": 1.546e-05,
      "loss": 195.1551,
      "step": 7730
    },
    {
      "epoch": 0.03127057939454664,
      "grad_norm": 1256.745849609375,
      "learning_rate": 1.548e-05,
      "loss": 128.5509,
      "step": 7740
    },
    {
      "epoch": 0.03131098065991427,
      "grad_norm": 1408.4251708984375,
      "learning_rate": 1.55e-05,
      "loss": 202.1482,
      "step": 7750
    },
    {
      "epoch": 0.0313513819252819,
      "grad_norm": 2253.969970703125,
      "learning_rate": 1.552e-05,
      "loss": 258.4842,
      "step": 7760
    },
    {
      "epoch": 0.03139178319064953,
      "grad_norm": 1037.827392578125,
      "learning_rate": 1.554e-05,
      "loss": 167.7397,
      "step": 7770
    },
    {
      "epoch": 0.03143218445601716,
      "grad_norm": 1066.5501708984375,
      "learning_rate": 1.556e-05,
      "loss": 218.4733,
      "step": 7780
    },
    {
      "epoch": 0.031472585721384796,
      "grad_norm": 1018.4628295898438,
      "learning_rate": 1.558e-05,
      "loss": 196.4911,
      "step": 7790
    },
    {
      "epoch": 0.03151298698675242,
      "grad_norm": 1526.3201904296875,
      "learning_rate": 1.56e-05,
      "loss": 214.0186,
      "step": 7800
    },
    {
      "epoch": 0.031553388252120056,
      "grad_norm": 829.8909301757812,
      "learning_rate": 1.5620000000000003e-05,
      "loss": 150.7465,
      "step": 7810
    },
    {
      "epoch": 0.03159378951748769,
      "grad_norm": 1091.463623046875,
      "learning_rate": 1.5640000000000003e-05,
      "loss": 240.4906,
      "step": 7820
    },
    {
      "epoch": 0.03163419078285532,
      "grad_norm": 470.00250244140625,
      "learning_rate": 1.566e-05,
      "loss": 159.8959,
      "step": 7830
    },
    {
      "epoch": 0.03167459204822295,
      "grad_norm": 1311.1846923828125,
      "learning_rate": 1.568e-05,
      "loss": 205.6226,
      "step": 7840
    },
    {
      "epoch": 0.03171499331359058,
      "grad_norm": 1698.6754150390625,
      "learning_rate": 1.5700000000000002e-05,
      "loss": 217.6704,
      "step": 7850
    },
    {
      "epoch": 0.031755394578958214,
      "grad_norm": 1376.8931884765625,
      "learning_rate": 1.5720000000000002e-05,
      "loss": 172.5748,
      "step": 7860
    },
    {
      "epoch": 0.03179579584432585,
      "grad_norm": 1433.7232666015625,
      "learning_rate": 1.5740000000000002e-05,
      "loss": 185.2836,
      "step": 7870
    },
    {
      "epoch": 0.03183619710969347,
      "grad_norm": 4025.76220703125,
      "learning_rate": 1.5759999999999998e-05,
      "loss": 315.4204,
      "step": 7880
    },
    {
      "epoch": 0.031876598375061106,
      "grad_norm": 853.8228759765625,
      "learning_rate": 1.578e-05,
      "loss": 153.3957,
      "step": 7890
    },
    {
      "epoch": 0.03191699964042874,
      "grad_norm": 944.7000732421875,
      "learning_rate": 1.58e-05,
      "loss": 128.8845,
      "step": 7900
    },
    {
      "epoch": 0.03195740090579637,
      "grad_norm": 643.9895629882812,
      "learning_rate": 1.582e-05,
      "loss": 223.8146,
      "step": 7910
    },
    {
      "epoch": 0.031997802171164,
      "grad_norm": 793.1214599609375,
      "learning_rate": 1.584e-05,
      "loss": 231.9548,
      "step": 7920
    },
    {
      "epoch": 0.03203820343653163,
      "grad_norm": 1593.3084716796875,
      "learning_rate": 1.586e-05,
      "loss": 222.1474,
      "step": 7930
    },
    {
      "epoch": 0.032078604701899265,
      "grad_norm": 0.0,
      "learning_rate": 1.588e-05,
      "loss": 156.401,
      "step": 7940
    },
    {
      "epoch": 0.0321190059672669,
      "grad_norm": 941.607177734375,
      "learning_rate": 1.59e-05,
      "loss": 239.5231,
      "step": 7950
    },
    {
      "epoch": 0.032159407232634524,
      "grad_norm": 1561.1961669921875,
      "learning_rate": 1.592e-05,
      "loss": 269.0299,
      "step": 7960
    },
    {
      "epoch": 0.03219980849800216,
      "grad_norm": 4084.514892578125,
      "learning_rate": 1.594e-05,
      "loss": 269.6239,
      "step": 7970
    },
    {
      "epoch": 0.03224020976336979,
      "grad_norm": 1185.47314453125,
      "learning_rate": 1.596e-05,
      "loss": 217.7421,
      "step": 7980
    },
    {
      "epoch": 0.03228061102873742,
      "grad_norm": 1212.916748046875,
      "learning_rate": 1.598e-05,
      "loss": 187.7411,
      "step": 7990
    },
    {
      "epoch": 0.03232101229410505,
      "grad_norm": 658.2976684570312,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 230.9428,
      "step": 8000
    },
    {
      "epoch": 0.03236141355947268,
      "grad_norm": 1249.109130859375,
      "learning_rate": 1.6020000000000002e-05,
      "loss": 222.3688,
      "step": 8010
    },
    {
      "epoch": 0.032401814824840315,
      "grad_norm": 3268.414306640625,
      "learning_rate": 1.604e-05,
      "loss": 330.1671,
      "step": 8020
    },
    {
      "epoch": 0.03244221609020795,
      "grad_norm": 2235.10595703125,
      "learning_rate": 1.606e-05,
      "loss": 239.3141,
      "step": 8030
    },
    {
      "epoch": 0.032482617355575574,
      "grad_norm": 461.5077819824219,
      "learning_rate": 1.6080000000000002e-05,
      "loss": 210.1967,
      "step": 8040
    },
    {
      "epoch": 0.03252301862094321,
      "grad_norm": 754.9512939453125,
      "learning_rate": 1.6100000000000002e-05,
      "loss": 196.5837,
      "step": 8050
    },
    {
      "epoch": 0.03256341988631084,
      "grad_norm": 1378.9425048828125,
      "learning_rate": 1.612e-05,
      "loss": 261.1862,
      "step": 8060
    },
    {
      "epoch": 0.03260382115167847,
      "grad_norm": 12839.7548828125,
      "learning_rate": 1.6139999999999998e-05,
      "loss": 188.6295,
      "step": 8070
    },
    {
      "epoch": 0.0326442224170461,
      "grad_norm": 1415.6925048828125,
      "learning_rate": 1.616e-05,
      "loss": 212.514,
      "step": 8080
    },
    {
      "epoch": 0.03268462368241373,
      "grad_norm": 633.0122680664062,
      "learning_rate": 1.618e-05,
      "loss": 167.5946,
      "step": 8090
    },
    {
      "epoch": 0.032725024947781366,
      "grad_norm": 1421.470947265625,
      "learning_rate": 1.62e-05,
      "loss": 141.4309,
      "step": 8100
    },
    {
      "epoch": 0.032765426213149,
      "grad_norm": 7646.71875,
      "learning_rate": 1.622e-05,
      "loss": 228.34,
      "step": 8110
    },
    {
      "epoch": 0.032805827478516625,
      "grad_norm": 1138.2156982421875,
      "learning_rate": 1.624e-05,
      "loss": 209.1626,
      "step": 8120
    },
    {
      "epoch": 0.03284622874388426,
      "grad_norm": 649.6055908203125,
      "learning_rate": 1.626e-05,
      "loss": 204.4883,
      "step": 8130
    },
    {
      "epoch": 0.03288663000925189,
      "grad_norm": 646.7942504882812,
      "learning_rate": 1.628e-05,
      "loss": 155.1816,
      "step": 8140
    },
    {
      "epoch": 0.032927031274619524,
      "grad_norm": 632.5813598632812,
      "learning_rate": 1.63e-05,
      "loss": 144.9363,
      "step": 8150
    },
    {
      "epoch": 0.03296743253998715,
      "grad_norm": 859.72900390625,
      "learning_rate": 1.6320000000000003e-05,
      "loss": 207.8656,
      "step": 8160
    },
    {
      "epoch": 0.03300783380535478,
      "grad_norm": 661.2268676757812,
      "learning_rate": 1.634e-05,
      "loss": 156.5535,
      "step": 8170
    },
    {
      "epoch": 0.033048235070722416,
      "grad_norm": 1062.429931640625,
      "learning_rate": 1.636e-05,
      "loss": 206.9411,
      "step": 8180
    },
    {
      "epoch": 0.03308863633609005,
      "grad_norm": 1048.27294921875,
      "learning_rate": 1.6380000000000002e-05,
      "loss": 186.6454,
      "step": 8190
    },
    {
      "epoch": 0.033129037601457675,
      "grad_norm": 1131.9554443359375,
      "learning_rate": 1.6400000000000002e-05,
      "loss": 255.3936,
      "step": 8200
    },
    {
      "epoch": 0.03316943886682531,
      "grad_norm": 1163.3974609375,
      "learning_rate": 1.6420000000000002e-05,
      "loss": 188.8816,
      "step": 8210
    },
    {
      "epoch": 0.03320984013219294,
      "grad_norm": 555.4834594726562,
      "learning_rate": 1.644e-05,
      "loss": 134.8741,
      "step": 8220
    },
    {
      "epoch": 0.033250241397560575,
      "grad_norm": 837.699951171875,
      "learning_rate": 1.646e-05,
      "loss": 183.4674,
      "step": 8230
    },
    {
      "epoch": 0.0332906426629282,
      "grad_norm": 731.1207275390625,
      "learning_rate": 1.648e-05,
      "loss": 117.9002,
      "step": 8240
    },
    {
      "epoch": 0.033331043928295834,
      "grad_norm": 747.77880859375,
      "learning_rate": 1.65e-05,
      "loss": 222.1833,
      "step": 8250
    },
    {
      "epoch": 0.03337144519366347,
      "grad_norm": 1067.3509521484375,
      "learning_rate": 1.652e-05,
      "loss": 186.3794,
      "step": 8260
    },
    {
      "epoch": 0.0334118464590311,
      "grad_norm": 1084.0224609375,
      "learning_rate": 1.654e-05,
      "loss": 236.8639,
      "step": 8270
    },
    {
      "epoch": 0.033452247724398726,
      "grad_norm": 1279.375732421875,
      "learning_rate": 1.656e-05,
      "loss": 259.652,
      "step": 8280
    },
    {
      "epoch": 0.03349264898976636,
      "grad_norm": 947.9122314453125,
      "learning_rate": 1.658e-05,
      "loss": 233.7726,
      "step": 8290
    },
    {
      "epoch": 0.03353305025513399,
      "grad_norm": 439.6305236816406,
      "learning_rate": 1.66e-05,
      "loss": 177.2815,
      "step": 8300
    },
    {
      "epoch": 0.033573451520501625,
      "grad_norm": 815.047119140625,
      "learning_rate": 1.662e-05,
      "loss": 176.6726,
      "step": 8310
    },
    {
      "epoch": 0.03361385278586925,
      "grad_norm": 6470.5361328125,
      "learning_rate": 1.664e-05,
      "loss": 222.3292,
      "step": 8320
    },
    {
      "epoch": 0.033654254051236884,
      "grad_norm": 790.7086791992188,
      "learning_rate": 1.666e-05,
      "loss": 194.9121,
      "step": 8330
    },
    {
      "epoch": 0.03369465531660452,
      "grad_norm": 794.2681884765625,
      "learning_rate": 1.668e-05,
      "loss": 175.7233,
      "step": 8340
    },
    {
      "epoch": 0.03373505658197215,
      "grad_norm": 1229.4151611328125,
      "learning_rate": 1.6700000000000003e-05,
      "loss": 227.8069,
      "step": 8350
    },
    {
      "epoch": 0.033775457847339777,
      "grad_norm": 792.4901123046875,
      "learning_rate": 1.672e-05,
      "loss": 198.4519,
      "step": 8360
    },
    {
      "epoch": 0.03381585911270741,
      "grad_norm": 1083.63623046875,
      "learning_rate": 1.674e-05,
      "loss": 164.7306,
      "step": 8370
    },
    {
      "epoch": 0.03385626037807504,
      "grad_norm": 1077.84765625,
      "learning_rate": 1.6760000000000002e-05,
      "loss": 154.6505,
      "step": 8380
    },
    {
      "epoch": 0.033896661643442676,
      "grad_norm": 2413.804931640625,
      "learning_rate": 1.6780000000000002e-05,
      "loss": 247.3373,
      "step": 8390
    },
    {
      "epoch": 0.0339370629088103,
      "grad_norm": 2946.898681640625,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 161.9533,
      "step": 8400
    },
    {
      "epoch": 0.033977464174177935,
      "grad_norm": 1092.52392578125,
      "learning_rate": 1.6819999999999998e-05,
      "loss": 216.0875,
      "step": 8410
    },
    {
      "epoch": 0.03401786543954557,
      "grad_norm": 930.1837158203125,
      "learning_rate": 1.684e-05,
      "loss": 169.9454,
      "step": 8420
    },
    {
      "epoch": 0.0340582667049132,
      "grad_norm": 514.6612548828125,
      "learning_rate": 1.686e-05,
      "loss": 202.2469,
      "step": 8430
    },
    {
      "epoch": 0.03409866797028083,
      "grad_norm": 747.2769775390625,
      "learning_rate": 1.688e-05,
      "loss": 176.8761,
      "step": 8440
    },
    {
      "epoch": 0.03413906923564846,
      "grad_norm": 1148.2232666015625,
      "learning_rate": 1.69e-05,
      "loss": 124.8942,
      "step": 8450
    },
    {
      "epoch": 0.03417947050101609,
      "grad_norm": 1272.0814208984375,
      "learning_rate": 1.692e-05,
      "loss": 201.3207,
      "step": 8460
    },
    {
      "epoch": 0.034219871766383726,
      "grad_norm": 658.9476318359375,
      "learning_rate": 1.694e-05,
      "loss": 193.7989,
      "step": 8470
    },
    {
      "epoch": 0.03426027303175135,
      "grad_norm": 1220.1839599609375,
      "learning_rate": 1.696e-05,
      "loss": 204.7586,
      "step": 8480
    },
    {
      "epoch": 0.034300674297118985,
      "grad_norm": 650.1453857421875,
      "learning_rate": 1.698e-05,
      "loss": 243.6956,
      "step": 8490
    },
    {
      "epoch": 0.03434107556248662,
      "grad_norm": 654.3552856445312,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 200.6542,
      "step": 8500
    },
    {
      "epoch": 0.03438147682785425,
      "grad_norm": 831.0885009765625,
      "learning_rate": 1.702e-05,
      "loss": 245.5201,
      "step": 8510
    },
    {
      "epoch": 0.03442187809322188,
      "grad_norm": 945.2589111328125,
      "learning_rate": 1.704e-05,
      "loss": 192.208,
      "step": 8520
    },
    {
      "epoch": 0.03446227935858951,
      "grad_norm": 929.2937622070312,
      "learning_rate": 1.706e-05,
      "loss": 195.5855,
      "step": 8530
    },
    {
      "epoch": 0.034502680623957144,
      "grad_norm": 1585.8245849609375,
      "learning_rate": 1.7080000000000002e-05,
      "loss": 161.8802,
      "step": 8540
    },
    {
      "epoch": 0.03454308188932478,
      "grad_norm": 1783.3887939453125,
      "learning_rate": 1.7100000000000002e-05,
      "loss": 259.6359,
      "step": 8550
    },
    {
      "epoch": 0.0345834831546924,
      "grad_norm": 1169.6343994140625,
      "learning_rate": 1.712e-05,
      "loss": 126.7247,
      "step": 8560
    },
    {
      "epoch": 0.034623884420060036,
      "grad_norm": 967.0234375,
      "learning_rate": 1.7140000000000002e-05,
      "loss": 228.0373,
      "step": 8570
    },
    {
      "epoch": 0.03466428568542767,
      "grad_norm": 1204.86865234375,
      "learning_rate": 1.7160000000000002e-05,
      "loss": 183.3604,
      "step": 8580
    },
    {
      "epoch": 0.0347046869507953,
      "grad_norm": 3491.818603515625,
      "learning_rate": 1.718e-05,
      "loss": 151.8515,
      "step": 8590
    },
    {
      "epoch": 0.03474508821616293,
      "grad_norm": 698.922607421875,
      "learning_rate": 1.7199999999999998e-05,
      "loss": 364.246,
      "step": 8600
    },
    {
      "epoch": 0.03478548948153056,
      "grad_norm": 754.37646484375,
      "learning_rate": 1.722e-05,
      "loss": 214.1979,
      "step": 8610
    },
    {
      "epoch": 0.034825890746898194,
      "grad_norm": 1443.7174072265625,
      "learning_rate": 1.724e-05,
      "loss": 146.2507,
      "step": 8620
    },
    {
      "epoch": 0.03486629201226583,
      "grad_norm": 1486.6044921875,
      "learning_rate": 1.726e-05,
      "loss": 209.7959,
      "step": 8630
    },
    {
      "epoch": 0.034906693277633453,
      "grad_norm": 1869.3212890625,
      "learning_rate": 1.728e-05,
      "loss": 216.6505,
      "step": 8640
    },
    {
      "epoch": 0.03494709454300109,
      "grad_norm": 1274.484130859375,
      "learning_rate": 1.73e-05,
      "loss": 175.598,
      "step": 8650
    },
    {
      "epoch": 0.03498749580836872,
      "grad_norm": 906.1784057617188,
      "learning_rate": 1.732e-05,
      "loss": 194.2832,
      "step": 8660
    },
    {
      "epoch": 0.03502789707373635,
      "grad_norm": 897.7321166992188,
      "learning_rate": 1.734e-05,
      "loss": 167.8126,
      "step": 8670
    },
    {
      "epoch": 0.03506829833910398,
      "grad_norm": 1329.6224365234375,
      "learning_rate": 1.736e-05,
      "loss": 100.7456,
      "step": 8680
    },
    {
      "epoch": 0.03510869960447161,
      "grad_norm": 2856.0009765625,
      "learning_rate": 1.7380000000000003e-05,
      "loss": 136.0745,
      "step": 8690
    },
    {
      "epoch": 0.035149100869839245,
      "grad_norm": 851.291748046875,
      "learning_rate": 1.74e-05,
      "loss": 276.2555,
      "step": 8700
    },
    {
      "epoch": 0.03518950213520688,
      "grad_norm": 1677.5799560546875,
      "learning_rate": 1.742e-05,
      "loss": 209.6706,
      "step": 8710
    },
    {
      "epoch": 0.035229903400574504,
      "grad_norm": 640.1288452148438,
      "learning_rate": 1.7440000000000002e-05,
      "loss": 223.2772,
      "step": 8720
    },
    {
      "epoch": 0.03527030466594214,
      "grad_norm": 4403.36962890625,
      "learning_rate": 1.7460000000000002e-05,
      "loss": 219.3949,
      "step": 8730
    },
    {
      "epoch": 0.03531070593130977,
      "grad_norm": 2118.526123046875,
      "learning_rate": 1.7480000000000002e-05,
      "loss": 264.3927,
      "step": 8740
    },
    {
      "epoch": 0.0353511071966774,
      "grad_norm": 1355.910400390625,
      "learning_rate": 1.75e-05,
      "loss": 172.0311,
      "step": 8750
    },
    {
      "epoch": 0.03539150846204503,
      "grad_norm": 1693.513671875,
      "learning_rate": 1.752e-05,
      "loss": 185.7292,
      "step": 8760
    },
    {
      "epoch": 0.03543190972741266,
      "grad_norm": 1187.20263671875,
      "learning_rate": 1.754e-05,
      "loss": 157.557,
      "step": 8770
    },
    {
      "epoch": 0.035472310992780295,
      "grad_norm": 893.8792114257812,
      "learning_rate": 1.756e-05,
      "loss": 191.817,
      "step": 8780
    },
    {
      "epoch": 0.03551271225814793,
      "grad_norm": 1077.1021728515625,
      "learning_rate": 1.758e-05,
      "loss": 189.9844,
      "step": 8790
    },
    {
      "epoch": 0.035553113523515555,
      "grad_norm": 929.8483276367188,
      "learning_rate": 1.76e-05,
      "loss": 197.5466,
      "step": 8800
    },
    {
      "epoch": 0.03559351478888319,
      "grad_norm": 2271.410400390625,
      "learning_rate": 1.762e-05,
      "loss": 228.7887,
      "step": 8810
    },
    {
      "epoch": 0.03563391605425082,
      "grad_norm": 2570.052001953125,
      "learning_rate": 1.764e-05,
      "loss": 155.6589,
      "step": 8820
    },
    {
      "epoch": 0.035674317319618454,
      "grad_norm": 657.0897216796875,
      "learning_rate": 1.766e-05,
      "loss": 225.2949,
      "step": 8830
    },
    {
      "epoch": 0.03571471858498608,
      "grad_norm": 1940.9866943359375,
      "learning_rate": 1.7680000000000004e-05,
      "loss": 162.1829,
      "step": 8840
    },
    {
      "epoch": 0.03575511985035371,
      "grad_norm": 1256.1148681640625,
      "learning_rate": 1.77e-05,
      "loss": 174.6637,
      "step": 8850
    },
    {
      "epoch": 0.035795521115721346,
      "grad_norm": 394.8336181640625,
      "learning_rate": 1.772e-05,
      "loss": 173.8255,
      "step": 8860
    },
    {
      "epoch": 0.03583592238108898,
      "grad_norm": 2153.534423828125,
      "learning_rate": 1.774e-05,
      "loss": 171.3043,
      "step": 8870
    },
    {
      "epoch": 0.035876323646456605,
      "grad_norm": 760.2359619140625,
      "learning_rate": 1.7760000000000003e-05,
      "loss": 200.2851,
      "step": 8880
    },
    {
      "epoch": 0.03591672491182424,
      "grad_norm": 1478.953369140625,
      "learning_rate": 1.7780000000000003e-05,
      "loss": 225.8184,
      "step": 8890
    },
    {
      "epoch": 0.03595712617719187,
      "grad_norm": 1087.418212890625,
      "learning_rate": 1.78e-05,
      "loss": 185.8715,
      "step": 8900
    },
    {
      "epoch": 0.035997527442559504,
      "grad_norm": 2434.63623046875,
      "learning_rate": 1.7820000000000002e-05,
      "loss": 178.0469,
      "step": 8910
    },
    {
      "epoch": 0.03603792870792713,
      "grad_norm": 1043.3377685546875,
      "learning_rate": 1.7840000000000002e-05,
      "loss": 179.4526,
      "step": 8920
    },
    {
      "epoch": 0.036078329973294763,
      "grad_norm": 1111.5899658203125,
      "learning_rate": 1.7860000000000002e-05,
      "loss": 210.2127,
      "step": 8930
    },
    {
      "epoch": 0.0361187312386624,
      "grad_norm": 688.8521728515625,
      "learning_rate": 1.7879999999999998e-05,
      "loss": 126.2401,
      "step": 8940
    },
    {
      "epoch": 0.03615913250403003,
      "grad_norm": 958.8832397460938,
      "learning_rate": 1.79e-05,
      "loss": 221.2132,
      "step": 8950
    },
    {
      "epoch": 0.036199533769397656,
      "grad_norm": 1401.521728515625,
      "learning_rate": 1.792e-05,
      "loss": 236.2084,
      "step": 8960
    },
    {
      "epoch": 0.03623993503476529,
      "grad_norm": 10308.7021484375,
      "learning_rate": 1.794e-05,
      "loss": 186.4272,
      "step": 8970
    },
    {
      "epoch": 0.03628033630013292,
      "grad_norm": 1097.2364501953125,
      "learning_rate": 1.796e-05,
      "loss": 178.1187,
      "step": 8980
    },
    {
      "epoch": 0.036320737565500555,
      "grad_norm": 1978.429931640625,
      "learning_rate": 1.798e-05,
      "loss": 196.0737,
      "step": 8990
    },
    {
      "epoch": 0.03636113883086818,
      "grad_norm": 3581.975341796875,
      "learning_rate": 1.8e-05,
      "loss": 209.1971,
      "step": 9000
    },
    {
      "epoch": 0.036401540096235814,
      "grad_norm": 1180.078125,
      "learning_rate": 1.802e-05,
      "loss": 220.5676,
      "step": 9010
    },
    {
      "epoch": 0.03644194136160345,
      "grad_norm": 1235.55859375,
      "learning_rate": 1.804e-05,
      "loss": 124.504,
      "step": 9020
    },
    {
      "epoch": 0.03648234262697108,
      "grad_norm": 1020.11865234375,
      "learning_rate": 1.8060000000000003e-05,
      "loss": 99.5129,
      "step": 9030
    },
    {
      "epoch": 0.036522743892338706,
      "grad_norm": 962.7813110351562,
      "learning_rate": 1.808e-05,
      "loss": 178.8402,
      "step": 9040
    },
    {
      "epoch": 0.03656314515770634,
      "grad_norm": 1318.480224609375,
      "learning_rate": 1.81e-05,
      "loss": 180.0119,
      "step": 9050
    },
    {
      "epoch": 0.03660354642307397,
      "grad_norm": 2261.136962890625,
      "learning_rate": 1.812e-05,
      "loss": 329.5452,
      "step": 9060
    },
    {
      "epoch": 0.036643947688441605,
      "grad_norm": 3996.27197265625,
      "learning_rate": 1.8140000000000003e-05,
      "loss": 224.336,
      "step": 9070
    },
    {
      "epoch": 0.03668434895380923,
      "grad_norm": 724.8762817382812,
      "learning_rate": 1.8160000000000002e-05,
      "loss": 148.4422,
      "step": 9080
    },
    {
      "epoch": 0.036724750219176865,
      "grad_norm": 838.4283447265625,
      "learning_rate": 1.818e-05,
      "loss": 160.2637,
      "step": 9090
    },
    {
      "epoch": 0.0367651514845445,
      "grad_norm": 1999.3927001953125,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 252.0707,
      "step": 9100
    },
    {
      "epoch": 0.03680555274991213,
      "grad_norm": 1008.3331298828125,
      "learning_rate": 1.8220000000000002e-05,
      "loss": 112.7752,
      "step": 9110
    },
    {
      "epoch": 0.03684595401527976,
      "grad_norm": 1373.5531005859375,
      "learning_rate": 1.824e-05,
      "loss": 140.497,
      "step": 9120
    },
    {
      "epoch": 0.03688635528064739,
      "grad_norm": 698.5148315429688,
      "learning_rate": 1.826e-05,
      "loss": 132.642,
      "step": 9130
    },
    {
      "epoch": 0.03692675654601502,
      "grad_norm": 1003.8239135742188,
      "learning_rate": 1.828e-05,
      "loss": 172.7703,
      "step": 9140
    },
    {
      "epoch": 0.03696715781138265,
      "grad_norm": 695.0194702148438,
      "learning_rate": 1.83e-05,
      "loss": 188.6919,
      "step": 9150
    },
    {
      "epoch": 0.03700755907675028,
      "grad_norm": 844.4266967773438,
      "learning_rate": 1.832e-05,
      "loss": 253.677,
      "step": 9160
    },
    {
      "epoch": 0.037047960342117915,
      "grad_norm": 1122.217041015625,
      "learning_rate": 1.834e-05,
      "loss": 197.4021,
      "step": 9170
    },
    {
      "epoch": 0.03708836160748555,
      "grad_norm": 1105.47314453125,
      "learning_rate": 1.8360000000000004e-05,
      "loss": 146.558,
      "step": 9180
    },
    {
      "epoch": 0.037128762872853174,
      "grad_norm": 1299.706298828125,
      "learning_rate": 1.838e-05,
      "loss": 196.9577,
      "step": 9190
    },
    {
      "epoch": 0.03716916413822081,
      "grad_norm": 4088.533447265625,
      "learning_rate": 1.84e-05,
      "loss": 178.9004,
      "step": 9200
    },
    {
      "epoch": 0.03720956540358844,
      "grad_norm": 2833.98388671875,
      "learning_rate": 1.842e-05,
      "loss": 220.8664,
      "step": 9210
    },
    {
      "epoch": 0.037249966668956074,
      "grad_norm": 649.87548828125,
      "learning_rate": 1.8440000000000003e-05,
      "loss": 163.9712,
      "step": 9220
    },
    {
      "epoch": 0.0372903679343237,
      "grad_norm": 1362.6846923828125,
      "learning_rate": 1.846e-05,
      "loss": 186.3454,
      "step": 9230
    },
    {
      "epoch": 0.03733076919969133,
      "grad_norm": 4132.52099609375,
      "learning_rate": 1.848e-05,
      "loss": 205.0476,
      "step": 9240
    },
    {
      "epoch": 0.037371170465058966,
      "grad_norm": 1643.118896484375,
      "learning_rate": 1.85e-05,
      "loss": 209.0955,
      "step": 9250
    },
    {
      "epoch": 0.0374115717304266,
      "grad_norm": 5062.59716796875,
      "learning_rate": 1.8520000000000002e-05,
      "loss": 145.1507,
      "step": 9260
    },
    {
      "epoch": 0.037451972995794225,
      "grad_norm": 390.54925537109375,
      "learning_rate": 1.8540000000000002e-05,
      "loss": 221.6778,
      "step": 9270
    },
    {
      "epoch": 0.03749237426116186,
      "grad_norm": 593.5599975585938,
      "learning_rate": 1.856e-05,
      "loss": 188.7002,
      "step": 9280
    },
    {
      "epoch": 0.03753277552652949,
      "grad_norm": 1883.616455078125,
      "learning_rate": 1.858e-05,
      "loss": 189.2378,
      "step": 9290
    },
    {
      "epoch": 0.037573176791897124,
      "grad_norm": 1161.8848876953125,
      "learning_rate": 1.86e-05,
      "loss": 197.2501,
      "step": 9300
    },
    {
      "epoch": 0.03761357805726475,
      "grad_norm": 692.7471313476562,
      "learning_rate": 1.862e-05,
      "loss": 171.7747,
      "step": 9310
    },
    {
      "epoch": 0.03765397932263238,
      "grad_norm": 644.8750610351562,
      "learning_rate": 1.864e-05,
      "loss": 243.3304,
      "step": 9320
    },
    {
      "epoch": 0.037694380588000016,
      "grad_norm": 1095.04443359375,
      "learning_rate": 1.866e-05,
      "loss": 262.2983,
      "step": 9330
    },
    {
      "epoch": 0.03773478185336765,
      "grad_norm": 1455.4674072265625,
      "learning_rate": 1.868e-05,
      "loss": 179.6782,
      "step": 9340
    },
    {
      "epoch": 0.037775183118735275,
      "grad_norm": 788.5756225585938,
      "learning_rate": 1.87e-05,
      "loss": 108.4792,
      "step": 9350
    },
    {
      "epoch": 0.03781558438410291,
      "grad_norm": 1122.2181396484375,
      "learning_rate": 1.872e-05,
      "loss": 261.3737,
      "step": 9360
    },
    {
      "epoch": 0.03785598564947054,
      "grad_norm": 1429.47607421875,
      "learning_rate": 1.8740000000000004e-05,
      "loss": 143.9239,
      "step": 9370
    },
    {
      "epoch": 0.037896386914838175,
      "grad_norm": 1072.0924072265625,
      "learning_rate": 1.876e-05,
      "loss": 217.753,
      "step": 9380
    },
    {
      "epoch": 0.0379367881802058,
      "grad_norm": 2434.5927734375,
      "learning_rate": 1.878e-05,
      "loss": 144.4152,
      "step": 9390
    },
    {
      "epoch": 0.037977189445573434,
      "grad_norm": 803.2135620117188,
      "learning_rate": 1.88e-05,
      "loss": 167.2345,
      "step": 9400
    },
    {
      "epoch": 0.03801759071094107,
      "grad_norm": 1207.8824462890625,
      "learning_rate": 1.8820000000000003e-05,
      "loss": 211.6839,
      "step": 9410
    },
    {
      "epoch": 0.0380579919763087,
      "grad_norm": 1132.8394775390625,
      "learning_rate": 1.8840000000000003e-05,
      "loss": 147.7339,
      "step": 9420
    },
    {
      "epoch": 0.038098393241676326,
      "grad_norm": 802.6298828125,
      "learning_rate": 1.886e-05,
      "loss": 139.8525,
      "step": 9430
    },
    {
      "epoch": 0.03813879450704396,
      "grad_norm": 3992.421875,
      "learning_rate": 1.888e-05,
      "loss": 199.5186,
      "step": 9440
    },
    {
      "epoch": 0.03817919577241159,
      "grad_norm": 1528.7784423828125,
      "learning_rate": 1.8900000000000002e-05,
      "loss": 186.0713,
      "step": 9450
    },
    {
      "epoch": 0.038219597037779225,
      "grad_norm": 853.1768798828125,
      "learning_rate": 1.8920000000000002e-05,
      "loss": 222.1405,
      "step": 9460
    },
    {
      "epoch": 0.03825999830314685,
      "grad_norm": 991.3099975585938,
      "learning_rate": 1.894e-05,
      "loss": 219.1491,
      "step": 9470
    },
    {
      "epoch": 0.038300399568514484,
      "grad_norm": 2823.091064453125,
      "learning_rate": 1.896e-05,
      "loss": 216.1606,
      "step": 9480
    },
    {
      "epoch": 0.03834080083388212,
      "grad_norm": 792.9014892578125,
      "learning_rate": 1.898e-05,
      "loss": 185.3038,
      "step": 9490
    },
    {
      "epoch": 0.03838120209924975,
      "grad_norm": 760.1093139648438,
      "learning_rate": 1.9e-05,
      "loss": 213.5252,
      "step": 9500
    },
    {
      "epoch": 0.03842160336461738,
      "grad_norm": 1229.9395751953125,
      "learning_rate": 1.902e-05,
      "loss": 99.1262,
      "step": 9510
    },
    {
      "epoch": 0.03846200462998501,
      "grad_norm": 1547.040771484375,
      "learning_rate": 1.904e-05,
      "loss": 193.0997,
      "step": 9520
    },
    {
      "epoch": 0.03850240589535264,
      "grad_norm": 5546.650390625,
      "learning_rate": 1.906e-05,
      "loss": 191.2586,
      "step": 9530
    },
    {
      "epoch": 0.038542807160720276,
      "grad_norm": 1049.4068603515625,
      "learning_rate": 1.908e-05,
      "loss": 126.0632,
      "step": 9540
    },
    {
      "epoch": 0.0385832084260879,
      "grad_norm": 670.5220947265625,
      "learning_rate": 1.91e-05,
      "loss": 152.6221,
      "step": 9550
    },
    {
      "epoch": 0.038623609691455535,
      "grad_norm": 1974.0662841796875,
      "learning_rate": 1.9120000000000003e-05,
      "loss": 175.8063,
      "step": 9560
    },
    {
      "epoch": 0.03866401095682317,
      "grad_norm": 4551.0517578125,
      "learning_rate": 1.914e-05,
      "loss": 188.1946,
      "step": 9570
    },
    {
      "epoch": 0.0387044122221908,
      "grad_norm": 806.855712890625,
      "learning_rate": 1.916e-05,
      "loss": 166.843,
      "step": 9580
    },
    {
      "epoch": 0.03874481348755843,
      "grad_norm": 531.7847290039062,
      "learning_rate": 1.918e-05,
      "loss": 141.4506,
      "step": 9590
    },
    {
      "epoch": 0.03878521475292606,
      "grad_norm": 931.3214111328125,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 184.4449,
      "step": 9600
    },
    {
      "epoch": 0.03882561601829369,
      "grad_norm": 728.2243041992188,
      "learning_rate": 1.9220000000000002e-05,
      "loss": 144.1032,
      "step": 9610
    },
    {
      "epoch": 0.038866017283661326,
      "grad_norm": 1086.5455322265625,
      "learning_rate": 1.924e-05,
      "loss": 166.8239,
      "step": 9620
    },
    {
      "epoch": 0.03890641854902895,
      "grad_norm": 0.0,
      "learning_rate": 1.9260000000000002e-05,
      "loss": 195.737,
      "step": 9630
    },
    {
      "epoch": 0.038946819814396585,
      "grad_norm": 1560.2904052734375,
      "learning_rate": 1.9280000000000002e-05,
      "loss": 165.3053,
      "step": 9640
    },
    {
      "epoch": 0.03898722107976422,
      "grad_norm": 1141.6810302734375,
      "learning_rate": 1.93e-05,
      "loss": 170.9885,
      "step": 9650
    },
    {
      "epoch": 0.03902762234513185,
      "grad_norm": 0.0,
      "learning_rate": 1.932e-05,
      "loss": 154.8165,
      "step": 9660
    },
    {
      "epoch": 0.03906802361049948,
      "grad_norm": 1502.9287109375,
      "learning_rate": 1.934e-05,
      "loss": 161.04,
      "step": 9670
    },
    {
      "epoch": 0.03910842487586711,
      "grad_norm": 1250.09619140625,
      "learning_rate": 1.936e-05,
      "loss": 140.4067,
      "step": 9680
    },
    {
      "epoch": 0.039148826141234744,
      "grad_norm": 1319.486328125,
      "learning_rate": 1.938e-05,
      "loss": 183.2919,
      "step": 9690
    },
    {
      "epoch": 0.03918922740660238,
      "grad_norm": 1285.4862060546875,
      "learning_rate": 1.94e-05,
      "loss": 200.0876,
      "step": 9700
    },
    {
      "epoch": 0.03922962867197,
      "grad_norm": 935.0762939453125,
      "learning_rate": 1.942e-05,
      "loss": 135.3171,
      "step": 9710
    },
    {
      "epoch": 0.039270029937337636,
      "grad_norm": 1619.843017578125,
      "learning_rate": 1.944e-05,
      "loss": 193.6001,
      "step": 9720
    },
    {
      "epoch": 0.03931043120270527,
      "grad_norm": 1271.864990234375,
      "learning_rate": 1.946e-05,
      "loss": 115.0868,
      "step": 9730
    },
    {
      "epoch": 0.0393508324680729,
      "grad_norm": 724.2216186523438,
      "learning_rate": 1.948e-05,
      "loss": 184.4641,
      "step": 9740
    },
    {
      "epoch": 0.03939123373344053,
      "grad_norm": 1635.4261474609375,
      "learning_rate": 1.9500000000000003e-05,
      "loss": 208.8505,
      "step": 9750
    },
    {
      "epoch": 0.03943163499880816,
      "grad_norm": 2965.97265625,
      "learning_rate": 1.9520000000000003e-05,
      "loss": 172.8497,
      "step": 9760
    },
    {
      "epoch": 0.039472036264175794,
      "grad_norm": 2287.69287109375,
      "learning_rate": 1.954e-05,
      "loss": 182.5987,
      "step": 9770
    },
    {
      "epoch": 0.03951243752954343,
      "grad_norm": 812.2120971679688,
      "learning_rate": 1.956e-05,
      "loss": 125.8975,
      "step": 9780
    },
    {
      "epoch": 0.039552838794911054,
      "grad_norm": 2255.82275390625,
      "learning_rate": 1.9580000000000002e-05,
      "loss": 205.0292,
      "step": 9790
    },
    {
      "epoch": 0.03959324006027869,
      "grad_norm": 794.7918701171875,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 140.7012,
      "step": 9800
    },
    {
      "epoch": 0.03963364132564632,
      "grad_norm": 1928.4591064453125,
      "learning_rate": 1.9620000000000002e-05,
      "loss": 230.6109,
      "step": 9810
    },
    {
      "epoch": 0.03967404259101395,
      "grad_norm": 497.63433837890625,
      "learning_rate": 1.9640000000000002e-05,
      "loss": 175.5,
      "step": 9820
    },
    {
      "epoch": 0.03971444385638158,
      "grad_norm": 605.4525146484375,
      "learning_rate": 1.966e-05,
      "loss": 146.0165,
      "step": 9830
    },
    {
      "epoch": 0.03975484512174921,
      "grad_norm": 12817.353515625,
      "learning_rate": 1.968e-05,
      "loss": 223.3047,
      "step": 9840
    },
    {
      "epoch": 0.039795246387116845,
      "grad_norm": 1107.6016845703125,
      "learning_rate": 1.97e-05,
      "loss": 164.6401,
      "step": 9850
    },
    {
      "epoch": 0.03983564765248448,
      "grad_norm": 1794.5758056640625,
      "learning_rate": 1.972e-05,
      "loss": 211.4972,
      "step": 9860
    },
    {
      "epoch": 0.039876048917852104,
      "grad_norm": 1120.5679931640625,
      "learning_rate": 1.974e-05,
      "loss": 198.396,
      "step": 9870
    },
    {
      "epoch": 0.03991645018321974,
      "grad_norm": 627.087646484375,
      "learning_rate": 1.976e-05,
      "loss": 142.4417,
      "step": 9880
    },
    {
      "epoch": 0.03995685144858737,
      "grad_norm": 415.39337158203125,
      "learning_rate": 1.978e-05,
      "loss": 109.121,
      "step": 9890
    },
    {
      "epoch": 0.039997252713955,
      "grad_norm": 1138.5850830078125,
      "learning_rate": 1.9800000000000004e-05,
      "loss": 190.1185,
      "step": 9900
    },
    {
      "epoch": 0.04003765397932263,
      "grad_norm": 506.6181945800781,
      "learning_rate": 1.982e-05,
      "loss": 152.6702,
      "step": 9910
    },
    {
      "epoch": 0.04007805524469026,
      "grad_norm": 1317.2239990234375,
      "learning_rate": 1.984e-05,
      "loss": 210.7801,
      "step": 9920
    },
    {
      "epoch": 0.040118456510057895,
      "grad_norm": 1474.5733642578125,
      "learning_rate": 1.986e-05,
      "loss": 224.4348,
      "step": 9930
    },
    {
      "epoch": 0.04015885777542553,
      "grad_norm": 1309.1876220703125,
      "learning_rate": 1.9880000000000003e-05,
      "loss": 87.4433,
      "step": 9940
    },
    {
      "epoch": 0.040199259040793155,
      "grad_norm": 861.6743774414062,
      "learning_rate": 1.9900000000000003e-05,
      "loss": 136.5576,
      "step": 9950
    },
    {
      "epoch": 0.04023966030616079,
      "grad_norm": 847.05517578125,
      "learning_rate": 1.992e-05,
      "loss": 182.9425,
      "step": 9960
    },
    {
      "epoch": 0.04028006157152842,
      "grad_norm": 1133.978515625,
      "learning_rate": 1.994e-05,
      "loss": 205.3185,
      "step": 9970
    },
    {
      "epoch": 0.040320462836896054,
      "grad_norm": 670.1548461914062,
      "learning_rate": 1.9960000000000002e-05,
      "loss": 190.8365,
      "step": 9980
    },
    {
      "epoch": 0.04036086410226368,
      "grad_norm": 887.6602783203125,
      "learning_rate": 1.9980000000000002e-05,
      "loss": 180.067,
      "step": 9990
    },
    {
      "epoch": 0.04040126536763131,
      "grad_norm": 687.7095336914062,
      "learning_rate": 2e-05,
      "loss": 127.6409,
      "step": 10000
    },
    {
      "epoch": 0.040441666632998946,
      "grad_norm": 385.455810546875,
      "learning_rate": 2.002e-05,
      "loss": 161.4435,
      "step": 10010
    },
    {
      "epoch": 0.04048206789836658,
      "grad_norm": 1409.2718505859375,
      "learning_rate": 2.004e-05,
      "loss": 182.3646,
      "step": 10020
    },
    {
      "epoch": 0.040522469163734205,
      "grad_norm": 803.884765625,
      "learning_rate": 2.006e-05,
      "loss": 210.5829,
      "step": 10030
    },
    {
      "epoch": 0.04056287042910184,
      "grad_norm": 1560.3782958984375,
      "learning_rate": 2.008e-05,
      "loss": 164.0962,
      "step": 10040
    },
    {
      "epoch": 0.04060327169446947,
      "grad_norm": 2029.135009765625,
      "learning_rate": 2.01e-05,
      "loss": 140.0395,
      "step": 10050
    },
    {
      "epoch": 0.040643672959837104,
      "grad_norm": 0.0,
      "learning_rate": 2.012e-05,
      "loss": 144.4245,
      "step": 10060
    },
    {
      "epoch": 0.04068407422520473,
      "grad_norm": 1140.849853515625,
      "learning_rate": 2.014e-05,
      "loss": 179.6542,
      "step": 10070
    },
    {
      "epoch": 0.040724475490572364,
      "grad_norm": 1746.8056640625,
      "learning_rate": 2.016e-05,
      "loss": 222.9251,
      "step": 10080
    },
    {
      "epoch": 0.04076487675594,
      "grad_norm": 722.617919921875,
      "learning_rate": 2.0180000000000003e-05,
      "loss": 175.7524,
      "step": 10090
    },
    {
      "epoch": 0.04080527802130763,
      "grad_norm": 837.6021118164062,
      "learning_rate": 2.0200000000000003e-05,
      "loss": 180.1091,
      "step": 10100
    },
    {
      "epoch": 0.040845679286675256,
      "grad_norm": 1460.4134521484375,
      "learning_rate": 2.022e-05,
      "loss": 188.8666,
      "step": 10110
    },
    {
      "epoch": 0.04088608055204289,
      "grad_norm": 770.96484375,
      "learning_rate": 2.024e-05,
      "loss": 164.6192,
      "step": 10120
    },
    {
      "epoch": 0.04092648181741052,
      "grad_norm": 887.3283081054688,
      "learning_rate": 2.0260000000000003e-05,
      "loss": 181.0503,
      "step": 10130
    },
    {
      "epoch": 0.040966883082778155,
      "grad_norm": 1267.9124755859375,
      "learning_rate": 2.0280000000000002e-05,
      "loss": 191.1377,
      "step": 10140
    },
    {
      "epoch": 0.04100728434814578,
      "grad_norm": 785.875244140625,
      "learning_rate": 2.0300000000000002e-05,
      "loss": 175.139,
      "step": 10150
    },
    {
      "epoch": 0.041047685613513414,
      "grad_norm": 3146.860107421875,
      "learning_rate": 2.032e-05,
      "loss": 189.9724,
      "step": 10160
    },
    {
      "epoch": 0.04108808687888105,
      "grad_norm": 1048.8988037109375,
      "learning_rate": 2.0340000000000002e-05,
      "loss": 219.648,
      "step": 10170
    },
    {
      "epoch": 0.04112848814424868,
      "grad_norm": 534.6448364257812,
      "learning_rate": 2.036e-05,
      "loss": 165.6375,
      "step": 10180
    },
    {
      "epoch": 0.041168889409616306,
      "grad_norm": 2196.583984375,
      "learning_rate": 2.038e-05,
      "loss": 219.21,
      "step": 10190
    },
    {
      "epoch": 0.04120929067498394,
      "grad_norm": 868.2059326171875,
      "learning_rate": 2.04e-05,
      "loss": 186.5126,
      "step": 10200
    },
    {
      "epoch": 0.04124969194035157,
      "grad_norm": 1177.8323974609375,
      "learning_rate": 2.042e-05,
      "loss": 173.2809,
      "step": 10210
    },
    {
      "epoch": 0.041290093205719205,
      "grad_norm": 1213.236572265625,
      "learning_rate": 2.044e-05,
      "loss": 195.4992,
      "step": 10220
    },
    {
      "epoch": 0.04133049447108683,
      "grad_norm": 712.7450561523438,
      "learning_rate": 2.046e-05,
      "loss": 123.3338,
      "step": 10230
    },
    {
      "epoch": 0.041370895736454465,
      "grad_norm": 1610.07373046875,
      "learning_rate": 2.048e-05,
      "loss": 182.489,
      "step": 10240
    },
    {
      "epoch": 0.0414112970018221,
      "grad_norm": 1756.5374755859375,
      "learning_rate": 2.05e-05,
      "loss": 271.8728,
      "step": 10250
    },
    {
      "epoch": 0.04145169826718973,
      "grad_norm": 1240.2843017578125,
      "learning_rate": 2.052e-05,
      "loss": 166.9299,
      "step": 10260
    },
    {
      "epoch": 0.04149209953255736,
      "grad_norm": 803.8770141601562,
      "learning_rate": 2.054e-05,
      "loss": 155.5844,
      "step": 10270
    },
    {
      "epoch": 0.04153250079792499,
      "grad_norm": 954.6376953125,
      "learning_rate": 2.0560000000000003e-05,
      "loss": 199.4898,
      "step": 10280
    },
    {
      "epoch": 0.04157290206329262,
      "grad_norm": 667.2396240234375,
      "learning_rate": 2.0580000000000003e-05,
      "loss": 112.1381,
      "step": 10290
    },
    {
      "epoch": 0.041613303328660256,
      "grad_norm": 1343.771240234375,
      "learning_rate": 2.06e-05,
      "loss": 200.9802,
      "step": 10300
    },
    {
      "epoch": 0.04165370459402788,
      "grad_norm": 1224.6522216796875,
      "learning_rate": 2.062e-05,
      "loss": 128.1033,
      "step": 10310
    },
    {
      "epoch": 0.041694105859395515,
      "grad_norm": 845.0104370117188,
      "learning_rate": 2.0640000000000002e-05,
      "loss": 165.8742,
      "step": 10320
    },
    {
      "epoch": 0.04173450712476315,
      "grad_norm": 532.292236328125,
      "learning_rate": 2.0660000000000002e-05,
      "loss": 131.5288,
      "step": 10330
    },
    {
      "epoch": 0.04177490839013078,
      "grad_norm": 876.7532958984375,
      "learning_rate": 2.0680000000000002e-05,
      "loss": 136.8921,
      "step": 10340
    },
    {
      "epoch": 0.04181530965549841,
      "grad_norm": 2460.5791015625,
      "learning_rate": 2.07e-05,
      "loss": 152.2601,
      "step": 10350
    },
    {
      "epoch": 0.04185571092086604,
      "grad_norm": 1794.477783203125,
      "learning_rate": 2.072e-05,
      "loss": 201.8371,
      "step": 10360
    },
    {
      "epoch": 0.041896112186233674,
      "grad_norm": 1977.7452392578125,
      "learning_rate": 2.074e-05,
      "loss": 247.7894,
      "step": 10370
    },
    {
      "epoch": 0.04193651345160131,
      "grad_norm": 667.1467895507812,
      "learning_rate": 2.076e-05,
      "loss": 222.6323,
      "step": 10380
    },
    {
      "epoch": 0.04197691471696893,
      "grad_norm": 351.2049865722656,
      "learning_rate": 2.078e-05,
      "loss": 138.2745,
      "step": 10390
    },
    {
      "epoch": 0.042017315982336566,
      "grad_norm": 944.7986450195312,
      "learning_rate": 2.08e-05,
      "loss": 196.8081,
      "step": 10400
    },
    {
      "epoch": 0.0420577172477042,
      "grad_norm": 1119.5870361328125,
      "learning_rate": 2.082e-05,
      "loss": 163.2965,
      "step": 10410
    },
    {
      "epoch": 0.04209811851307183,
      "grad_norm": 684.2236938476562,
      "learning_rate": 2.084e-05,
      "loss": 154.9308,
      "step": 10420
    },
    {
      "epoch": 0.04213851977843946,
      "grad_norm": 1214.9677734375,
      "learning_rate": 2.086e-05,
      "loss": 181.0213,
      "step": 10430
    },
    {
      "epoch": 0.04217892104380709,
      "grad_norm": 979.2085571289062,
      "learning_rate": 2.0880000000000003e-05,
      "loss": 214.0994,
      "step": 10440
    },
    {
      "epoch": 0.042219322309174724,
      "grad_norm": 1195.45947265625,
      "learning_rate": 2.09e-05,
      "loss": 184.8288,
      "step": 10450
    },
    {
      "epoch": 0.04225972357454236,
      "grad_norm": 3266.998291015625,
      "learning_rate": 2.092e-05,
      "loss": 154.9474,
      "step": 10460
    },
    {
      "epoch": 0.04230012483990998,
      "grad_norm": 1256.9425048828125,
      "learning_rate": 2.0940000000000003e-05,
      "loss": 119.0928,
      "step": 10470
    },
    {
      "epoch": 0.042340526105277616,
      "grad_norm": 816.3329467773438,
      "learning_rate": 2.0960000000000003e-05,
      "loss": 138.4739,
      "step": 10480
    },
    {
      "epoch": 0.04238092737064525,
      "grad_norm": 1145.7486572265625,
      "learning_rate": 2.098e-05,
      "loss": 179.8457,
      "step": 10490
    },
    {
      "epoch": 0.04242132863601288,
      "grad_norm": 4142.47998046875,
      "learning_rate": 2.1e-05,
      "loss": 190.1194,
      "step": 10500
    },
    {
      "epoch": 0.04246172990138051,
      "grad_norm": 748.7318115234375,
      "learning_rate": 2.1020000000000002e-05,
      "loss": 156.9302,
      "step": 10510
    },
    {
      "epoch": 0.04250213116674814,
      "grad_norm": 1273.5777587890625,
      "learning_rate": 2.1040000000000002e-05,
      "loss": 175.272,
      "step": 10520
    },
    {
      "epoch": 0.042542532432115775,
      "grad_norm": 551.0802612304688,
      "learning_rate": 2.106e-05,
      "loss": 167.7439,
      "step": 10530
    },
    {
      "epoch": 0.04258293369748341,
      "grad_norm": 604.2559814453125,
      "learning_rate": 2.1079999999999998e-05,
      "loss": 153.7686,
      "step": 10540
    },
    {
      "epoch": 0.042623334962851034,
      "grad_norm": 460.89154052734375,
      "learning_rate": 2.11e-05,
      "loss": 117.5873,
      "step": 10550
    },
    {
      "epoch": 0.04266373622821867,
      "grad_norm": 1551.710693359375,
      "learning_rate": 2.112e-05,
      "loss": 165.0876,
      "step": 10560
    },
    {
      "epoch": 0.0427041374935863,
      "grad_norm": 934.1738891601562,
      "learning_rate": 2.114e-05,
      "loss": 143.8318,
      "step": 10570
    },
    {
      "epoch": 0.04274453875895393,
      "grad_norm": 1267.439453125,
      "learning_rate": 2.116e-05,
      "loss": 187.996,
      "step": 10580
    },
    {
      "epoch": 0.04278494002432156,
      "grad_norm": 1843.2279052734375,
      "learning_rate": 2.118e-05,
      "loss": 181.2761,
      "step": 10590
    },
    {
      "epoch": 0.04282534128968919,
      "grad_norm": 2143.365966796875,
      "learning_rate": 2.12e-05,
      "loss": 158.841,
      "step": 10600
    },
    {
      "epoch": 0.042865742555056825,
      "grad_norm": 925.6973266601562,
      "learning_rate": 2.122e-05,
      "loss": 210.331,
      "step": 10610
    },
    {
      "epoch": 0.04290614382042446,
      "grad_norm": 2171.398681640625,
      "learning_rate": 2.124e-05,
      "loss": 225.7547,
      "step": 10620
    },
    {
      "epoch": 0.042946545085792084,
      "grad_norm": 1160.7347412109375,
      "learning_rate": 2.1260000000000003e-05,
      "loss": 153.8757,
      "step": 10630
    },
    {
      "epoch": 0.04298694635115972,
      "grad_norm": 1135.9134521484375,
      "learning_rate": 2.128e-05,
      "loss": 188.1846,
      "step": 10640
    },
    {
      "epoch": 0.04302734761652735,
      "grad_norm": 2029.5738525390625,
      "learning_rate": 2.13e-05,
      "loss": 201.2368,
      "step": 10650
    },
    {
      "epoch": 0.043067748881894984,
      "grad_norm": 814.3689575195312,
      "learning_rate": 2.1320000000000003e-05,
      "loss": 206.1595,
      "step": 10660
    },
    {
      "epoch": 0.04310815014726261,
      "grad_norm": 422.2183837890625,
      "learning_rate": 2.1340000000000002e-05,
      "loss": 157.6857,
      "step": 10670
    },
    {
      "epoch": 0.04314855141263024,
      "grad_norm": 880.4464111328125,
      "learning_rate": 2.1360000000000002e-05,
      "loss": 163.0372,
      "step": 10680
    },
    {
      "epoch": 0.043188952677997876,
      "grad_norm": 484.959228515625,
      "learning_rate": 2.138e-05,
      "loss": 144.4623,
      "step": 10690
    },
    {
      "epoch": 0.04322935394336551,
      "grad_norm": 1225.7841796875,
      "learning_rate": 2.1400000000000002e-05,
      "loss": 226.4971,
      "step": 10700
    },
    {
      "epoch": 0.043269755208733135,
      "grad_norm": 569.0413208007812,
      "learning_rate": 2.142e-05,
      "loss": 186.6787,
      "step": 10710
    },
    {
      "epoch": 0.04331015647410077,
      "grad_norm": 1226.3690185546875,
      "learning_rate": 2.144e-05,
      "loss": 219.7577,
      "step": 10720
    },
    {
      "epoch": 0.0433505577394684,
      "grad_norm": 1915.441650390625,
      "learning_rate": 2.146e-05,
      "loss": 194.0689,
      "step": 10730
    },
    {
      "epoch": 0.043390959004836034,
      "grad_norm": 1193.3553466796875,
      "learning_rate": 2.148e-05,
      "loss": 149.6157,
      "step": 10740
    },
    {
      "epoch": 0.04343136027020366,
      "grad_norm": 562.625,
      "learning_rate": 2.15e-05,
      "loss": 204.4145,
      "step": 10750
    },
    {
      "epoch": 0.04347176153557129,
      "grad_norm": 12496.0029296875,
      "learning_rate": 2.152e-05,
      "loss": 248.7909,
      "step": 10760
    },
    {
      "epoch": 0.043512162800938926,
      "grad_norm": 1115.0479736328125,
      "learning_rate": 2.154e-05,
      "loss": 183.6795,
      "step": 10770
    },
    {
      "epoch": 0.04355256406630656,
      "grad_norm": 1180.6798095703125,
      "learning_rate": 2.1560000000000004e-05,
      "loss": 161.4541,
      "step": 10780
    },
    {
      "epoch": 0.043592965331674186,
      "grad_norm": 2096.56298828125,
      "learning_rate": 2.158e-05,
      "loss": 177.0552,
      "step": 10790
    },
    {
      "epoch": 0.04363336659704182,
      "grad_norm": 1031.798828125,
      "learning_rate": 2.16e-05,
      "loss": 160.7948,
      "step": 10800
    },
    {
      "epoch": 0.04367376786240945,
      "grad_norm": 1001.6660766601562,
      "learning_rate": 2.162e-05,
      "loss": 279.4042,
      "step": 10810
    },
    {
      "epoch": 0.043714169127777085,
      "grad_norm": 5103.36669921875,
      "learning_rate": 2.1640000000000003e-05,
      "loss": 198.244,
      "step": 10820
    },
    {
      "epoch": 0.04375457039314471,
      "grad_norm": 813.843017578125,
      "learning_rate": 2.166e-05,
      "loss": 204.9555,
      "step": 10830
    },
    {
      "epoch": 0.043794971658512344,
      "grad_norm": 1556.0152587890625,
      "learning_rate": 2.168e-05,
      "loss": 178.9818,
      "step": 10840
    },
    {
      "epoch": 0.04383537292387998,
      "grad_norm": 1136.54443359375,
      "learning_rate": 2.1700000000000002e-05,
      "loss": 214.708,
      "step": 10850
    },
    {
      "epoch": 0.04387577418924761,
      "grad_norm": 913.5957641601562,
      "learning_rate": 2.1720000000000002e-05,
      "loss": 162.8733,
      "step": 10860
    },
    {
      "epoch": 0.043916175454615236,
      "grad_norm": 1809.7088623046875,
      "learning_rate": 2.1740000000000002e-05,
      "loss": 202.5214,
      "step": 10870
    },
    {
      "epoch": 0.04395657671998287,
      "grad_norm": 1365.0262451171875,
      "learning_rate": 2.176e-05,
      "loss": 197.6029,
      "step": 10880
    },
    {
      "epoch": 0.0439969779853505,
      "grad_norm": 1124.8267822265625,
      "learning_rate": 2.178e-05,
      "loss": 195.3639,
      "step": 10890
    },
    {
      "epoch": 0.044037379250718135,
      "grad_norm": 933.0280151367188,
      "learning_rate": 2.18e-05,
      "loss": 166.1828,
      "step": 10900
    },
    {
      "epoch": 0.04407778051608576,
      "grad_norm": 893.4417724609375,
      "learning_rate": 2.182e-05,
      "loss": 125.791,
      "step": 10910
    },
    {
      "epoch": 0.044118181781453394,
      "grad_norm": 1642.8094482421875,
      "learning_rate": 2.184e-05,
      "loss": 144.994,
      "step": 10920
    },
    {
      "epoch": 0.04415858304682103,
      "grad_norm": 1519.8751220703125,
      "learning_rate": 2.186e-05,
      "loss": 158.417,
      "step": 10930
    },
    {
      "epoch": 0.04419898431218866,
      "grad_norm": 1085.8243408203125,
      "learning_rate": 2.188e-05,
      "loss": 189.8353,
      "step": 10940
    },
    {
      "epoch": 0.04423938557755629,
      "grad_norm": 907.6950073242188,
      "learning_rate": 2.19e-05,
      "loss": 131.8621,
      "step": 10950
    },
    {
      "epoch": 0.04427978684292392,
      "grad_norm": 988.0367431640625,
      "learning_rate": 2.192e-05,
      "loss": 110.0259,
      "step": 10960
    },
    {
      "epoch": 0.04432018810829155,
      "grad_norm": 358.17889404296875,
      "learning_rate": 2.1940000000000003e-05,
      "loss": 144.1581,
      "step": 10970
    },
    {
      "epoch": 0.044360589373659186,
      "grad_norm": 3126.972900390625,
      "learning_rate": 2.196e-05,
      "loss": 242.1895,
      "step": 10980
    },
    {
      "epoch": 0.04440099063902681,
      "grad_norm": 660.0758056640625,
      "learning_rate": 2.198e-05,
      "loss": 307.6868,
      "step": 10990
    },
    {
      "epoch": 0.044441391904394445,
      "grad_norm": 1388.6563720703125,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 191.1774,
      "step": 11000
    },
    {
      "epoch": 0.04448179316976208,
      "grad_norm": 0.0,
      "learning_rate": 2.2020000000000003e-05,
      "loss": 182.281,
      "step": 11010
    },
    {
      "epoch": 0.04452219443512971,
      "grad_norm": 1133.305419921875,
      "learning_rate": 2.2040000000000002e-05,
      "loss": 141.6396,
      "step": 11020
    },
    {
      "epoch": 0.04456259570049734,
      "grad_norm": 760.71728515625,
      "learning_rate": 2.206e-05,
      "loss": 184.8224,
      "step": 11030
    },
    {
      "epoch": 0.04460299696586497,
      "grad_norm": 700.0611572265625,
      "learning_rate": 2.2080000000000002e-05,
      "loss": 174.1201,
      "step": 11040
    },
    {
      "epoch": 0.0446433982312326,
      "grad_norm": 464.8121643066406,
      "learning_rate": 2.2100000000000002e-05,
      "loss": 147.1738,
      "step": 11050
    },
    {
      "epoch": 0.044683799496600236,
      "grad_norm": 659.0347900390625,
      "learning_rate": 2.212e-05,
      "loss": 121.4146,
      "step": 11060
    },
    {
      "epoch": 0.04472420076196786,
      "grad_norm": 495.3475646972656,
      "learning_rate": 2.214e-05,
      "loss": 122.6828,
      "step": 11070
    },
    {
      "epoch": 0.044764602027335496,
      "grad_norm": 899.0453491210938,
      "learning_rate": 2.216e-05,
      "loss": 171.483,
      "step": 11080
    },
    {
      "epoch": 0.04480500329270313,
      "grad_norm": 1652.7977294921875,
      "learning_rate": 2.218e-05,
      "loss": 136.9625,
      "step": 11090
    },
    {
      "epoch": 0.04484540455807076,
      "grad_norm": 1164.9923095703125,
      "learning_rate": 2.22e-05,
      "loss": 218.6428,
      "step": 11100
    },
    {
      "epoch": 0.04488580582343839,
      "grad_norm": 1860.8387451171875,
      "learning_rate": 2.222e-05,
      "loss": 128.8451,
      "step": 11110
    },
    {
      "epoch": 0.04492620708880602,
      "grad_norm": 1793.767578125,
      "learning_rate": 2.224e-05,
      "loss": 127.518,
      "step": 11120
    },
    {
      "epoch": 0.044966608354173654,
      "grad_norm": 910.3096923828125,
      "learning_rate": 2.226e-05,
      "loss": 102.4282,
      "step": 11130
    },
    {
      "epoch": 0.04500700961954129,
      "grad_norm": 3313.176025390625,
      "learning_rate": 2.228e-05,
      "loss": 205.4728,
      "step": 11140
    },
    {
      "epoch": 0.04504741088490891,
      "grad_norm": 959.5852661132812,
      "learning_rate": 2.23e-05,
      "loss": 181.3004,
      "step": 11150
    },
    {
      "epoch": 0.045087812150276546,
      "grad_norm": 3117.2705078125,
      "learning_rate": 2.2320000000000003e-05,
      "loss": 159.9985,
      "step": 11160
    },
    {
      "epoch": 0.04512821341564418,
      "grad_norm": 4501.052734375,
      "learning_rate": 2.234e-05,
      "loss": 150.3723,
      "step": 11170
    },
    {
      "epoch": 0.04516861468101181,
      "grad_norm": 981.978515625,
      "learning_rate": 2.236e-05,
      "loss": 142.0348,
      "step": 11180
    },
    {
      "epoch": 0.04520901594637944,
      "grad_norm": 1109.785400390625,
      "learning_rate": 2.2380000000000003e-05,
      "loss": 163.3822,
      "step": 11190
    },
    {
      "epoch": 0.04524941721174707,
      "grad_norm": 1274.3033447265625,
      "learning_rate": 2.2400000000000002e-05,
      "loss": 268.0728,
      "step": 11200
    },
    {
      "epoch": 0.045289818477114704,
      "grad_norm": 750.956298828125,
      "learning_rate": 2.2420000000000002e-05,
      "loss": 164.3101,
      "step": 11210
    },
    {
      "epoch": 0.04533021974248234,
      "grad_norm": 908.2940063476562,
      "learning_rate": 2.244e-05,
      "loss": 205.5986,
      "step": 11220
    },
    {
      "epoch": 0.045370621007849964,
      "grad_norm": 1060.370849609375,
      "learning_rate": 2.2460000000000002e-05,
      "loss": 152.6007,
      "step": 11230
    },
    {
      "epoch": 0.0454110222732176,
      "grad_norm": 1556.2314453125,
      "learning_rate": 2.248e-05,
      "loss": 204.8753,
      "step": 11240
    },
    {
      "epoch": 0.04545142353858523,
      "grad_norm": 683.9859619140625,
      "learning_rate": 2.25e-05,
      "loss": 142.627,
      "step": 11250
    },
    {
      "epoch": 0.04549182480395286,
      "grad_norm": 1313.3026123046875,
      "learning_rate": 2.252e-05,
      "loss": 183.1745,
      "step": 11260
    },
    {
      "epoch": 0.04553222606932049,
      "grad_norm": 1381.1295166015625,
      "learning_rate": 2.254e-05,
      "loss": 175.9728,
      "step": 11270
    },
    {
      "epoch": 0.04557262733468812,
      "grad_norm": 509.716796875,
      "learning_rate": 2.256e-05,
      "loss": 210.1509,
      "step": 11280
    },
    {
      "epoch": 0.045613028600055755,
      "grad_norm": 2492.5234375,
      "learning_rate": 2.258e-05,
      "loss": 231.53,
      "step": 11290
    },
    {
      "epoch": 0.04565342986542339,
      "grad_norm": 1633.21630859375,
      "learning_rate": 2.26e-05,
      "loss": 152.2005,
      "step": 11300
    },
    {
      "epoch": 0.045693831130791014,
      "grad_norm": 768.939208984375,
      "learning_rate": 2.2620000000000004e-05,
      "loss": 185.2856,
      "step": 11310
    },
    {
      "epoch": 0.04573423239615865,
      "grad_norm": 799.730712890625,
      "learning_rate": 2.264e-05,
      "loss": 194.0909,
      "step": 11320
    },
    {
      "epoch": 0.04577463366152628,
      "grad_norm": 3679.43310546875,
      "learning_rate": 2.266e-05,
      "loss": 354.1977,
      "step": 11330
    },
    {
      "epoch": 0.04581503492689391,
      "grad_norm": 1576.691650390625,
      "learning_rate": 2.268e-05,
      "loss": 181.5113,
      "step": 11340
    },
    {
      "epoch": 0.04585543619226154,
      "grad_norm": 1108.6512451171875,
      "learning_rate": 2.2700000000000003e-05,
      "loss": 184.2569,
      "step": 11350
    },
    {
      "epoch": 0.04589583745762917,
      "grad_norm": 1905.447998046875,
      "learning_rate": 2.2720000000000003e-05,
      "loss": 217.2384,
      "step": 11360
    },
    {
      "epoch": 0.045936238722996806,
      "grad_norm": 884.3139038085938,
      "learning_rate": 2.274e-05,
      "loss": 141.6524,
      "step": 11370
    },
    {
      "epoch": 0.04597663998836444,
      "grad_norm": 617.2454223632812,
      "learning_rate": 2.2760000000000002e-05,
      "loss": 148.7823,
      "step": 11380
    },
    {
      "epoch": 0.046017041253732065,
      "grad_norm": 551.2994384765625,
      "learning_rate": 2.2780000000000002e-05,
      "loss": 133.9882,
      "step": 11390
    },
    {
      "epoch": 0.0460574425190997,
      "grad_norm": 700.8192138671875,
      "learning_rate": 2.2800000000000002e-05,
      "loss": 154.8847,
      "step": 11400
    },
    {
      "epoch": 0.04609784378446733,
      "grad_norm": 1914.3916015625,
      "learning_rate": 2.282e-05,
      "loss": 145.8075,
      "step": 11410
    },
    {
      "epoch": 0.046138245049834964,
      "grad_norm": 2445.005859375,
      "learning_rate": 2.284e-05,
      "loss": 199.9154,
      "step": 11420
    },
    {
      "epoch": 0.04617864631520259,
      "grad_norm": 1541.644775390625,
      "learning_rate": 2.286e-05,
      "loss": 168.6186,
      "step": 11430
    },
    {
      "epoch": 0.04621904758057022,
      "grad_norm": 816.3759765625,
      "learning_rate": 2.288e-05,
      "loss": 181.8408,
      "step": 11440
    },
    {
      "epoch": 0.046259448845937856,
      "grad_norm": 1267.0755615234375,
      "learning_rate": 2.29e-05,
      "loss": 172.8114,
      "step": 11450
    },
    {
      "epoch": 0.04629985011130549,
      "grad_norm": 855.4586791992188,
      "learning_rate": 2.292e-05,
      "loss": 197.7767,
      "step": 11460
    },
    {
      "epoch": 0.046340251376673115,
      "grad_norm": 1094.00634765625,
      "learning_rate": 2.294e-05,
      "loss": 171.698,
      "step": 11470
    },
    {
      "epoch": 0.04638065264204075,
      "grad_norm": 1205.0,
      "learning_rate": 2.296e-05,
      "loss": 215.9526,
      "step": 11480
    },
    {
      "epoch": 0.04642105390740838,
      "grad_norm": 1176.0413818359375,
      "learning_rate": 2.298e-05,
      "loss": 163.7831,
      "step": 11490
    },
    {
      "epoch": 0.046461455172776014,
      "grad_norm": 786.1920776367188,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 167.4065,
      "step": 11500
    },
    {
      "epoch": 0.04650185643814364,
      "grad_norm": 516.4776000976562,
      "learning_rate": 2.302e-05,
      "loss": 165.6811,
      "step": 11510
    },
    {
      "epoch": 0.046542257703511274,
      "grad_norm": 396.16497802734375,
      "learning_rate": 2.304e-05,
      "loss": 168.2221,
      "step": 11520
    },
    {
      "epoch": 0.04658265896887891,
      "grad_norm": 882.178466796875,
      "learning_rate": 2.306e-05,
      "loss": 140.923,
      "step": 11530
    },
    {
      "epoch": 0.04662306023424654,
      "grad_norm": 934.631591796875,
      "learning_rate": 2.3080000000000003e-05,
      "loss": 186.3469,
      "step": 11540
    },
    {
      "epoch": 0.046663461499614166,
      "grad_norm": 1052.7349853515625,
      "learning_rate": 2.3100000000000002e-05,
      "loss": 201.8761,
      "step": 11550
    },
    {
      "epoch": 0.0467038627649818,
      "grad_norm": 1325.5751953125,
      "learning_rate": 2.312e-05,
      "loss": 255.0451,
      "step": 11560
    },
    {
      "epoch": 0.04674426403034943,
      "grad_norm": 827.1654663085938,
      "learning_rate": 2.3140000000000002e-05,
      "loss": 162.8754,
      "step": 11570
    },
    {
      "epoch": 0.046784665295717065,
      "grad_norm": 506.81097412109375,
      "learning_rate": 2.3160000000000002e-05,
      "loss": 99.88,
      "step": 11580
    },
    {
      "epoch": 0.04682506656108469,
      "grad_norm": 1403.5689697265625,
      "learning_rate": 2.318e-05,
      "loss": 157.3405,
      "step": 11590
    },
    {
      "epoch": 0.046865467826452324,
      "grad_norm": 1264.109375,
      "learning_rate": 2.32e-05,
      "loss": 135.8957,
      "step": 11600
    },
    {
      "epoch": 0.04690586909181996,
      "grad_norm": 851.1090698242188,
      "learning_rate": 2.322e-05,
      "loss": 98.8174,
      "step": 11610
    },
    {
      "epoch": 0.04694627035718759,
      "grad_norm": 1114.6529541015625,
      "learning_rate": 2.324e-05,
      "loss": 174.8662,
      "step": 11620
    },
    {
      "epoch": 0.046986671622555216,
      "grad_norm": 997.0178833007812,
      "learning_rate": 2.326e-05,
      "loss": 202.974,
      "step": 11630
    },
    {
      "epoch": 0.04702707288792285,
      "grad_norm": 2686.275146484375,
      "learning_rate": 2.328e-05,
      "loss": 180.5854,
      "step": 11640
    },
    {
      "epoch": 0.04706747415329048,
      "grad_norm": 1341.779541015625,
      "learning_rate": 2.3300000000000004e-05,
      "loss": 188.3465,
      "step": 11650
    },
    {
      "epoch": 0.047107875418658116,
      "grad_norm": 1174.360595703125,
      "learning_rate": 2.332e-05,
      "loss": 100.4389,
      "step": 11660
    },
    {
      "epoch": 0.04714827668402574,
      "grad_norm": 485.88134765625,
      "learning_rate": 2.334e-05,
      "loss": 84.7096,
      "step": 11670
    },
    {
      "epoch": 0.047188677949393375,
      "grad_norm": 700.1830444335938,
      "learning_rate": 2.336e-05,
      "loss": 252.0006,
      "step": 11680
    },
    {
      "epoch": 0.04722907921476101,
      "grad_norm": 1874.8160400390625,
      "learning_rate": 2.3380000000000003e-05,
      "loss": 271.8778,
      "step": 11690
    },
    {
      "epoch": 0.04726948048012864,
      "grad_norm": 780.0234985351562,
      "learning_rate": 2.3400000000000003e-05,
      "loss": 156.8313,
      "step": 11700
    },
    {
      "epoch": 0.04730988174549627,
      "grad_norm": 1217.7091064453125,
      "learning_rate": 2.342e-05,
      "loss": 139.3272,
      "step": 11710
    },
    {
      "epoch": 0.0473502830108639,
      "grad_norm": 1002.0595092773438,
      "learning_rate": 2.344e-05,
      "loss": 155.965,
      "step": 11720
    },
    {
      "epoch": 0.04739068427623153,
      "grad_norm": 597.17529296875,
      "learning_rate": 2.3460000000000002e-05,
      "loss": 172.1847,
      "step": 11730
    },
    {
      "epoch": 0.047431085541599166,
      "grad_norm": 910.2625732421875,
      "learning_rate": 2.3480000000000002e-05,
      "loss": 148.3666,
      "step": 11740
    },
    {
      "epoch": 0.04747148680696679,
      "grad_norm": 1215.2615966796875,
      "learning_rate": 2.35e-05,
      "loss": 194.8361,
      "step": 11750
    },
    {
      "epoch": 0.047511888072334425,
      "grad_norm": 762.500244140625,
      "learning_rate": 2.3520000000000002e-05,
      "loss": 113.7633,
      "step": 11760
    },
    {
      "epoch": 0.04755228933770206,
      "grad_norm": 3147.351318359375,
      "learning_rate": 2.354e-05,
      "loss": 194.7475,
      "step": 11770
    },
    {
      "epoch": 0.04759269060306969,
      "grad_norm": 1164.023681640625,
      "learning_rate": 2.356e-05,
      "loss": 108.1381,
      "step": 11780
    },
    {
      "epoch": 0.04763309186843732,
      "grad_norm": 2093.1826171875,
      "learning_rate": 2.358e-05,
      "loss": 165.2831,
      "step": 11790
    },
    {
      "epoch": 0.04767349313380495,
      "grad_norm": 628.909423828125,
      "learning_rate": 2.36e-05,
      "loss": 172.9029,
      "step": 11800
    },
    {
      "epoch": 0.047713894399172584,
      "grad_norm": 400.00811767578125,
      "learning_rate": 2.362e-05,
      "loss": 181.3741,
      "step": 11810
    },
    {
      "epoch": 0.04775429566454022,
      "grad_norm": 1536.272216796875,
      "learning_rate": 2.364e-05,
      "loss": 154.7509,
      "step": 11820
    },
    {
      "epoch": 0.04779469692990784,
      "grad_norm": 1407.3763427734375,
      "learning_rate": 2.366e-05,
      "loss": 174.5441,
      "step": 11830
    },
    {
      "epoch": 0.047835098195275476,
      "grad_norm": 726.475830078125,
      "learning_rate": 2.3680000000000004e-05,
      "loss": 171.3695,
      "step": 11840
    },
    {
      "epoch": 0.04787549946064311,
      "grad_norm": 1060.8369140625,
      "learning_rate": 2.37e-05,
      "loss": 150.6572,
      "step": 11850
    },
    {
      "epoch": 0.04791590072601074,
      "grad_norm": 2245.559814453125,
      "learning_rate": 2.372e-05,
      "loss": 172.8109,
      "step": 11860
    },
    {
      "epoch": 0.04795630199137837,
      "grad_norm": 2129.251708984375,
      "learning_rate": 2.374e-05,
      "loss": 191.5867,
      "step": 11870
    },
    {
      "epoch": 0.047996703256746,
      "grad_norm": 1103.0263671875,
      "learning_rate": 2.3760000000000003e-05,
      "loss": 213.8663,
      "step": 11880
    },
    {
      "epoch": 0.048037104522113634,
      "grad_norm": 1063.9124755859375,
      "learning_rate": 2.3780000000000003e-05,
      "loss": 194.2018,
      "step": 11890
    },
    {
      "epoch": 0.04807750578748127,
      "grad_norm": 7122.65185546875,
      "learning_rate": 2.38e-05,
      "loss": 229.5696,
      "step": 11900
    },
    {
      "epoch": 0.04811790705284889,
      "grad_norm": 1017.9201049804688,
      "learning_rate": 2.3820000000000002e-05,
      "loss": 146.0323,
      "step": 11910
    },
    {
      "epoch": 0.048158308318216526,
      "grad_norm": 1932.71435546875,
      "learning_rate": 2.3840000000000002e-05,
      "loss": 143.6282,
      "step": 11920
    },
    {
      "epoch": 0.04819870958358416,
      "grad_norm": 753.262939453125,
      "learning_rate": 2.3860000000000002e-05,
      "loss": 151.7239,
      "step": 11930
    },
    {
      "epoch": 0.04823911084895179,
      "grad_norm": 1837.5714111328125,
      "learning_rate": 2.3880000000000002e-05,
      "loss": 192.4802,
      "step": 11940
    },
    {
      "epoch": 0.04827951211431942,
      "grad_norm": 1542.8756103515625,
      "learning_rate": 2.39e-05,
      "loss": 239.2184,
      "step": 11950
    },
    {
      "epoch": 0.04831991337968705,
      "grad_norm": 1126.86572265625,
      "learning_rate": 2.392e-05,
      "loss": 132.8466,
      "step": 11960
    },
    {
      "epoch": 0.048360314645054685,
      "grad_norm": 650.4667358398438,
      "learning_rate": 2.394e-05,
      "loss": 121.4513,
      "step": 11970
    },
    {
      "epoch": 0.04840071591042232,
      "grad_norm": 708.6920776367188,
      "learning_rate": 2.396e-05,
      "loss": 172.2358,
      "step": 11980
    },
    {
      "epoch": 0.048441117175789944,
      "grad_norm": 1389.237548828125,
      "learning_rate": 2.398e-05,
      "loss": 220.8339,
      "step": 11990
    },
    {
      "epoch": 0.04848151844115758,
      "grad_norm": 580.6951293945312,
      "learning_rate": 2.4e-05,
      "loss": 178.8768,
      "step": 12000
    }
  ],
  "logging_steps": 10,
  "max_steps": 250000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}