{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 16845,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0029682398337785693,
      "grad_norm": 0.48303863406181335,
      "learning_rate": 4.997031760166221e-05,
      "loss": 1.3426,
      "step": 10
    },
    {
      "epoch": 0.005936479667557139,
      "grad_norm": 0.5679779052734375,
      "learning_rate": 4.9940635203324435e-05,
      "loss": 1.2914,
      "step": 20
    },
    {
      "epoch": 0.008904719501335707,
      "grad_norm": 0.565687358379364,
      "learning_rate": 4.9910952804986644e-05,
      "loss": 1.2217,
      "step": 30
    },
    {
      "epoch": 0.011872959335114277,
      "grad_norm": 0.5573933720588684,
      "learning_rate": 4.988127040664886e-05,
      "loss": 1.1249,
      "step": 40
    },
    {
      "epoch": 0.014841199168892847,
      "grad_norm": 0.6410918831825256,
      "learning_rate": 4.9851588008311077e-05,
      "loss": 1.1554,
      "step": 50
    },
    {
      "epoch": 0.017809439002671415,
      "grad_norm": 0.6899677515029907,
      "learning_rate": 4.982190560997329e-05,
      "loss": 1.1577,
      "step": 60
    },
    {
      "epoch": 0.020777678836449986,
      "grad_norm": 0.6809150576591492,
      "learning_rate": 4.97922232116355e-05,
      "loss": 1.107,
      "step": 70
    },
    {
      "epoch": 0.023745918670228554,
      "grad_norm": 0.7853492498397827,
      "learning_rate": 4.976254081329771e-05,
      "loss": 1.1337,
      "step": 80
    },
    {
      "epoch": 0.026714158504007122,
      "grad_norm": 0.6834965944290161,
      "learning_rate": 4.9732858414959934e-05,
      "loss": 1.1123,
      "step": 90
    },
    {
      "epoch": 0.029682398337785694,
      "grad_norm": 0.6894986629486084,
      "learning_rate": 4.9703176016622144e-05,
      "loss": 1.1234,
      "step": 100
    },
    {
      "epoch": 0.032650638171564265,
      "grad_norm": 0.662402868270874,
      "learning_rate": 4.967349361828436e-05,
      "loss": 1.1082,
      "step": 110
    },
    {
      "epoch": 0.03561887800534283,
      "grad_norm": 0.7717655897140503,
      "learning_rate": 4.9643811219946576e-05,
      "loss": 1.1124,
      "step": 120
    },
    {
      "epoch": 0.0385871178391214,
      "grad_norm": 0.8598708510398865,
      "learning_rate": 4.961412882160879e-05,
      "loss": 1.0867,
      "step": 130
    },
    {
      "epoch": 0.04155535767289997,
      "grad_norm": 0.7680957317352295,
      "learning_rate": 4.9584446423271e-05,
      "loss": 1.096,
      "step": 140
    },
    {
      "epoch": 0.04452359750667854,
      "grad_norm": 0.6783134937286377,
      "learning_rate": 4.955476402493322e-05,
      "loss": 1.0742,
      "step": 150
    },
    {
      "epoch": 0.04749183734045711,
      "grad_norm": 0.8696405291557312,
      "learning_rate": 4.9525081626595434e-05,
      "loss": 1.0774,
      "step": 160
    },
    {
      "epoch": 0.05046007717423568,
      "grad_norm": 0.8478761315345764,
      "learning_rate": 4.9495399228257644e-05,
      "loss": 1.1037,
      "step": 170
    },
    {
      "epoch": 0.053428317008014245,
      "grad_norm": 0.7295342087745667,
      "learning_rate": 4.946571682991986e-05,
      "loss": 1.0566,
      "step": 180
    },
    {
      "epoch": 0.056396556841792816,
      "grad_norm": 0.7539967894554138,
      "learning_rate": 4.9436034431582076e-05,
      "loss": 1.1014,
      "step": 190
    },
    {
      "epoch": 0.05936479667557139,
      "grad_norm": 0.7601701021194458,
      "learning_rate": 4.940635203324429e-05,
      "loss": 1.0948,
      "step": 200
    },
    {
      "epoch": 0.06233303650934995,
      "grad_norm": 0.7360059022903442,
      "learning_rate": 4.93766696349065e-05,
      "loss": 1.0612,
      "step": 210
    },
    {
      "epoch": 0.06530127634312853,
      "grad_norm": 0.758794367313385,
      "learning_rate": 4.934698723656872e-05,
      "loss": 1.0671,
      "step": 220
    },
    {
      "epoch": 0.06826951617690709,
      "grad_norm": 0.7825369834899902,
      "learning_rate": 4.9317304838230934e-05,
      "loss": 1.1034,
      "step": 230
    },
    {
      "epoch": 0.07123775601068566,
      "grad_norm": 0.858249306678772,
      "learning_rate": 4.928762243989314e-05,
      "loss": 1.0904,
      "step": 240
    },
    {
      "epoch": 0.07420599584446423,
      "grad_norm": 0.8293972015380859,
      "learning_rate": 4.925794004155536e-05,
      "loss": 1.0787,
      "step": 250
    },
    {
      "epoch": 0.0771742356782428,
      "grad_norm": 0.8243985176086426,
      "learning_rate": 4.9228257643217576e-05,
      "loss": 1.0855,
      "step": 260
    },
    {
      "epoch": 0.08014247551202137,
      "grad_norm": 0.7798173427581787,
      "learning_rate": 4.919857524487979e-05,
      "loss": 1.0975,
      "step": 270
    },
    {
      "epoch": 0.08311071534579995,
      "grad_norm": 0.7953142523765564,
      "learning_rate": 4.9168892846542e-05,
      "loss": 1.0904,
      "step": 280
    },
    {
      "epoch": 0.0860789551795785,
      "grad_norm": 0.7199780941009521,
      "learning_rate": 4.913921044820422e-05,
      "loss": 1.0577,
      "step": 290
    },
    {
      "epoch": 0.08904719501335707,
      "grad_norm": 0.7431743741035461,
      "learning_rate": 4.9109528049866433e-05,
      "loss": 1.1162,
      "step": 300
    },
    {
      "epoch": 0.09201543484713565,
      "grad_norm": 0.7822330594062805,
      "learning_rate": 4.907984565152864e-05,
      "loss": 1.0389,
      "step": 310
    },
    {
      "epoch": 0.09498367468091422,
      "grad_norm": 0.8207771182060242,
      "learning_rate": 4.905016325319086e-05,
      "loss": 1.0779,
      "step": 320
    },
    {
      "epoch": 0.09795191451469279,
      "grad_norm": 0.7167501449584961,
      "learning_rate": 4.9020480854853075e-05,
      "loss": 1.0698,
      "step": 330
    },
    {
      "epoch": 0.10092015434847136,
      "grad_norm": 0.7531447410583496,
      "learning_rate": 4.899079845651529e-05,
      "loss": 1.0787,
      "step": 340
    },
    {
      "epoch": 0.10388839418224993,
      "grad_norm": 0.8100646734237671,
      "learning_rate": 4.89611160581775e-05,
      "loss": 1.0529,
      "step": 350
    },
    {
      "epoch": 0.10685663401602849,
      "grad_norm": 0.7299244403839111,
      "learning_rate": 4.893143365983972e-05,
      "loss": 1.0262,
      "step": 360
    },
    {
      "epoch": 0.10982487384980706,
      "grad_norm": 0.845340371131897,
      "learning_rate": 4.890175126150193e-05,
      "loss": 1.1234,
      "step": 370
    },
    {
      "epoch": 0.11279311368358563,
      "grad_norm": 0.7792521119117737,
      "learning_rate": 4.887206886316414e-05,
      "loss": 1.0919,
      "step": 380
    },
    {
      "epoch": 0.1157613535173642,
      "grad_norm": 0.6663903594017029,
      "learning_rate": 4.884238646482636e-05,
      "loss": 1.0471,
      "step": 390
    },
    {
      "epoch": 0.11872959335114278,
      "grad_norm": 0.822321355342865,
      "learning_rate": 4.8812704066488575e-05,
      "loss": 1.0181,
      "step": 400
    },
    {
      "epoch": 0.12169783318492135,
      "grad_norm": 0.837409257888794,
      "learning_rate": 4.878302166815079e-05,
      "loss": 1.102,
      "step": 410
    },
    {
      "epoch": 0.1246660730186999,
      "grad_norm": 0.7892317771911621,
      "learning_rate": 4.8753339269813e-05,
      "loss": 1.0608,
      "step": 420
    },
    {
      "epoch": 0.12763431285247848,
      "grad_norm": 0.7766367197036743,
      "learning_rate": 4.872365687147522e-05,
      "loss": 1.0525,
      "step": 430
    },
    {
      "epoch": 0.13060255268625706,
      "grad_norm": 0.6542902588844299,
      "learning_rate": 4.869397447313743e-05,
      "loss": 1.0306,
      "step": 440
    },
    {
      "epoch": 0.13357079252003562,
      "grad_norm": 0.8308093547821045,
      "learning_rate": 4.866429207479964e-05,
      "loss": 1.0462,
      "step": 450
    },
    {
      "epoch": 0.13653903235381418,
      "grad_norm": 0.7098076343536377,
      "learning_rate": 4.8634609676461865e-05,
      "loss": 1.0403,
      "step": 460
    },
    {
      "epoch": 0.13950727218759276,
      "grad_norm": 0.6637741923332214,
      "learning_rate": 4.8604927278124075e-05,
      "loss": 1.0298,
      "step": 470
    },
    {
      "epoch": 0.14247551202137132,
      "grad_norm": 0.7593216896057129,
      "learning_rate": 4.857524487978629e-05,
      "loss": 1.006,
      "step": 480
    },
    {
      "epoch": 0.1454437518551499,
      "grad_norm": 0.7959597706794739,
      "learning_rate": 4.85455624814485e-05,
      "loss": 1.0231,
      "step": 490
    },
    {
      "epoch": 0.14841199168892846,
      "grad_norm": 0.7332357168197632,
      "learning_rate": 4.8515880083110716e-05,
      "loss": 1.0455,
      "step": 500
    },
    {
      "epoch": 0.15138023152270705,
      "grad_norm": 0.6890411972999573,
      "learning_rate": 4.848619768477293e-05,
      "loss": 1.0561,
      "step": 510
    },
    {
      "epoch": 0.1543484713564856,
      "grad_norm": 0.7440224885940552,
      "learning_rate": 4.845651528643514e-05,
      "loss": 1.0515,
      "step": 520
    },
    {
      "epoch": 0.15731671119026416,
      "grad_norm": 0.832664966583252,
      "learning_rate": 4.8426832888097365e-05,
      "loss": 1.0584,
      "step": 530
    },
    {
      "epoch": 0.16028495102404275,
      "grad_norm": 0.6851795315742493,
      "learning_rate": 4.8397150489759574e-05,
      "loss": 1.0327,
      "step": 540
    },
    {
      "epoch": 0.1632531908578213,
      "grad_norm": 0.9479386806488037,
      "learning_rate": 4.836746809142179e-05,
      "loss": 1.0745,
      "step": 550
    },
    {
      "epoch": 0.1662214306915999,
      "grad_norm": 0.7720438241958618,
      "learning_rate": 4.8337785693084007e-05,
      "loss": 1.0472,
      "step": 560
    },
    {
      "epoch": 0.16918967052537845,
      "grad_norm": 0.7193672060966492,
      "learning_rate": 4.8308103294746216e-05,
      "loss": 1.0452,
      "step": 570
    },
    {
      "epoch": 0.172157910359157,
      "grad_norm": 0.7027491331100464,
      "learning_rate": 4.827842089640843e-05,
      "loss": 1.0444,
      "step": 580
    },
    {
      "epoch": 0.1751261501929356,
      "grad_norm": 0.7128656506538391,
      "learning_rate": 4.824873849807064e-05,
      "loss": 1.0367,
      "step": 590
    },
    {
      "epoch": 0.17809439002671415,
      "grad_norm": 0.7210800051689148,
      "learning_rate": 4.8219056099732865e-05,
      "loss": 1.0319,
      "step": 600
    },
    {
      "epoch": 0.18106262986049274,
      "grad_norm": 0.7800968885421753,
      "learning_rate": 4.8189373701395074e-05,
      "loss": 1.0222,
      "step": 610
    },
    {
      "epoch": 0.1840308696942713,
      "grad_norm": 0.7994186878204346,
      "learning_rate": 4.815969130305729e-05,
      "loss": 1.0582,
      "step": 620
    },
    {
      "epoch": 0.18699910952804988,
      "grad_norm": 0.723895788192749,
      "learning_rate": 4.8130008904719506e-05,
      "loss": 1.0431,
      "step": 630
    },
    {
      "epoch": 0.18996734936182844,
      "grad_norm": 0.6798288226127625,
      "learning_rate": 4.8100326506381716e-05,
      "loss": 1.0365,
      "step": 640
    },
    {
      "epoch": 0.192935589195607,
      "grad_norm": 0.7264121174812317,
      "learning_rate": 4.807064410804393e-05,
      "loss": 1.0839,
      "step": 650
    },
    {
      "epoch": 0.19590382902938558,
      "grad_norm": 0.7775474786758423,
      "learning_rate": 4.804096170970615e-05,
      "loss": 1.0839,
      "step": 660
    },
    {
      "epoch": 0.19887206886316414,
      "grad_norm": 0.8966870903968811,
      "learning_rate": 4.8011279311368364e-05,
      "loss": 1.075,
      "step": 670
    },
    {
      "epoch": 0.20184030869694272,
      "grad_norm": 0.7206711173057556,
      "learning_rate": 4.7981596913030574e-05,
      "loss": 1.0616,
      "step": 680
    },
    {
      "epoch": 0.20480854853072128,
      "grad_norm": 0.6880177855491638,
      "learning_rate": 4.795191451469279e-05,
      "loss": 1.0684,
      "step": 690
    },
    {
      "epoch": 0.20777678836449986,
      "grad_norm": 0.6823096871376038,
      "learning_rate": 4.7922232116355006e-05,
      "loss": 1.0385,
      "step": 700
    },
    {
      "epoch": 0.21074502819827842,
      "grad_norm": 0.7464755773544312,
      "learning_rate": 4.7892549718017215e-05,
      "loss": 1.052,
      "step": 710
    },
    {
      "epoch": 0.21371326803205698,
      "grad_norm": 0.7194072604179382,
      "learning_rate": 4.786286731967943e-05,
      "loss": 1.0424,
      "step": 720
    },
    {
      "epoch": 0.21668150786583557,
      "grad_norm": 0.6561169028282166,
      "learning_rate": 4.783318492134165e-05,
      "loss": 1.0605,
      "step": 730
    },
    {
      "epoch": 0.21964974769961412,
      "grad_norm": 0.7468820214271545,
      "learning_rate": 4.7803502523003864e-05,
      "loss": 1.0656,
      "step": 740
    },
    {
      "epoch": 0.2226179875333927,
      "grad_norm": 0.784570038318634,
      "learning_rate": 4.777382012466607e-05,
      "loss": 1.0684,
      "step": 750
    },
    {
      "epoch": 0.22558622736717127,
      "grad_norm": 0.8690618276596069,
      "learning_rate": 4.774413772632829e-05,
      "loss": 1.0058,
      "step": 760
    },
    {
      "epoch": 0.22855446720094982,
      "grad_norm": 0.7325943112373352,
      "learning_rate": 4.7714455327990506e-05,
      "loss": 1.0016,
      "step": 770
    },
    {
      "epoch": 0.2315227070347284,
      "grad_norm": 0.8618629574775696,
      "learning_rate": 4.7684772929652715e-05,
      "loss": 1.0219,
      "step": 780
    },
    {
      "epoch": 0.23449094686850697,
      "grad_norm": 0.6887000203132629,
      "learning_rate": 4.765509053131493e-05,
      "loss": 1.054,
      "step": 790
    },
    {
      "epoch": 0.23745918670228555,
      "grad_norm": 0.6989481449127197,
      "learning_rate": 4.762540813297715e-05,
      "loss": 1.0484,
      "step": 800
    },
    {
      "epoch": 0.2404274265360641,
      "grad_norm": 0.7582924365997314,
      "learning_rate": 4.7595725734639364e-05,
      "loss": 1.054,
      "step": 810
    },
    {
      "epoch": 0.2433956663698427,
      "grad_norm": 0.6527506709098816,
      "learning_rate": 4.756604333630157e-05,
      "loss": 1.0076,
      "step": 820
    },
    {
      "epoch": 0.24636390620362125,
      "grad_norm": 0.6977763175964355,
      "learning_rate": 4.753636093796379e-05,
      "loss": 1.0274,
      "step": 830
    },
    {
      "epoch": 0.2493321460373998,
      "grad_norm": 0.6364858746528625,
      "learning_rate": 4.7506678539626005e-05,
      "loss": 1.0389,
      "step": 840
    },
    {
      "epoch": 0.25230038587117837,
      "grad_norm": 0.7233136296272278,
      "learning_rate": 4.7476996141288215e-05,
      "loss": 1.0177,
      "step": 850
    },
    {
      "epoch": 0.25526862570495695,
      "grad_norm": 0.759456217288971,
      "learning_rate": 4.744731374295043e-05,
      "loss": 1.0332,
      "step": 860
    },
    {
      "epoch": 0.25823686553873554,
      "grad_norm": 0.7150506973266602,
      "learning_rate": 4.741763134461265e-05,
      "loss": 1.0146,
      "step": 870
    },
    {
      "epoch": 0.2612051053725141,
      "grad_norm": 0.7771095633506775,
      "learning_rate": 4.738794894627486e-05,
      "loss": 1.0138,
      "step": 880
    },
    {
      "epoch": 0.26417334520629265,
      "grad_norm": 0.6592842936515808,
      "learning_rate": 4.735826654793707e-05,
      "loss": 1.0729,
      "step": 890
    },
    {
      "epoch": 0.26714158504007124,
      "grad_norm": 0.7347294092178345,
      "learning_rate": 4.732858414959929e-05,
      "loss": 1.073,
      "step": 900
    },
    {
      "epoch": 0.2701098248738498,
      "grad_norm": 0.7552283406257629,
      "learning_rate": 4.7298901751261505e-05,
      "loss": 1.0434,
      "step": 910
    },
    {
      "epoch": 0.27307806470762835,
      "grad_norm": 0.7497857213020325,
      "learning_rate": 4.7269219352923714e-05,
      "loss": 1.0236,
      "step": 920
    },
    {
      "epoch": 0.27604630454140694,
      "grad_norm": 0.6930972933769226,
      "learning_rate": 4.723953695458593e-05,
      "loss": 1.0303,
      "step": 930
    },
    {
      "epoch": 0.2790145443751855,
      "grad_norm": 0.674025297164917,
      "learning_rate": 4.720985455624815e-05,
      "loss": 1.0226,
      "step": 940
    },
    {
      "epoch": 0.2819827842089641,
      "grad_norm": 0.7063450217247009,
      "learning_rate": 4.718017215791036e-05,
      "loss": 1.0147,
      "step": 950
    },
    {
      "epoch": 0.28495102404274264,
      "grad_norm": 0.7937106490135193,
      "learning_rate": 4.715048975957257e-05,
      "loss": 1.0197,
      "step": 960
    },
    {
      "epoch": 0.2879192638765212,
      "grad_norm": 0.7385783195495605,
      "learning_rate": 4.7120807361234795e-05,
      "loss": 1.0337,
      "step": 970
    },
    {
      "epoch": 0.2908875037102998,
      "grad_norm": 0.7633090019226074,
      "learning_rate": 4.7091124962897005e-05,
      "loss": 1.0096,
      "step": 980
    },
    {
      "epoch": 0.29385574354407834,
      "grad_norm": 0.8703083992004395,
      "learning_rate": 4.7061442564559214e-05,
      "loss": 1.0368,
      "step": 990
    },
    {
      "epoch": 0.2968239833778569,
      "grad_norm": 0.7346276640892029,
      "learning_rate": 4.703176016622143e-05,
      "loss": 1.0115,
      "step": 1000
    },
    {
      "epoch": 0.2997922232116355,
      "grad_norm": 0.7063167691230774,
      "learning_rate": 4.7002077767883646e-05,
      "loss": 1.0252,
      "step": 1010
    },
    {
      "epoch": 0.3027604630454141,
      "grad_norm": 0.7117500305175781,
      "learning_rate": 4.697239536954586e-05,
      "loss": 1.0244,
      "step": 1020
    },
    {
      "epoch": 0.3057287028791926,
      "grad_norm": 0.6929411292076111,
      "learning_rate": 4.694271297120807e-05,
      "loss": 1.003,
      "step": 1030
    },
    {
      "epoch": 0.3086969427129712,
      "grad_norm": 0.6956213712692261,
      "learning_rate": 4.6913030572870295e-05,
      "loss": 1.0508,
      "step": 1040
    },
    {
      "epoch": 0.3116651825467498,
      "grad_norm": 0.6457957029342651,
      "learning_rate": 4.6883348174532504e-05,
      "loss": 1.0415,
      "step": 1050
    },
    {
      "epoch": 0.3146334223805283,
      "grad_norm": 0.7128310203552246,
      "learning_rate": 4.6853665776194714e-05,
      "loss": 1.0391,
      "step": 1060
    },
    {
      "epoch": 0.3176016622143069,
      "grad_norm": 0.8085856437683105,
      "learning_rate": 4.682398337785694e-05,
      "loss": 1.0181,
      "step": 1070
    },
    {
      "epoch": 0.3205699020480855,
      "grad_norm": 0.6627590656280518,
      "learning_rate": 4.6794300979519146e-05,
      "loss": 0.9902,
      "step": 1080
    },
    {
      "epoch": 0.323538141881864,
      "grad_norm": 0.7251258492469788,
      "learning_rate": 4.676461858118136e-05,
      "loss": 0.9954,
      "step": 1090
    },
    {
      "epoch": 0.3265063817156426,
      "grad_norm": 0.6958507299423218,
      "learning_rate": 4.673493618284357e-05,
      "loss": 1.0152,
      "step": 1100
    },
    {
      "epoch": 0.3294746215494212,
      "grad_norm": 0.714284360408783,
      "learning_rate": 4.6705253784505795e-05,
      "loss": 1.0322,
      "step": 1110
    },
    {
      "epoch": 0.3324428613831998,
      "grad_norm": 0.6643791198730469,
      "learning_rate": 4.6675571386168004e-05,
      "loss": 0.9957,
      "step": 1120
    },
    {
      "epoch": 0.3354111012169783,
      "grad_norm": 0.7135879397392273,
      "learning_rate": 4.6645888987830213e-05,
      "loss": 1.0158,
      "step": 1130
    },
    {
      "epoch": 0.3383793410507569,
      "grad_norm": 0.7087671160697937,
      "learning_rate": 4.6616206589492436e-05,
      "loss": 1.0146,
      "step": 1140
    },
    {
      "epoch": 0.3413475808845355,
      "grad_norm": 0.6507649421691895,
      "learning_rate": 4.6586524191154646e-05,
      "loss": 1.0216,
      "step": 1150
    },
    {
      "epoch": 0.344315820718314,
      "grad_norm": 0.7352689504623413,
      "learning_rate": 4.655684179281686e-05,
      "loss": 1.064,
      "step": 1160
    },
    {
      "epoch": 0.3472840605520926,
      "grad_norm": 0.7079006433486938,
      "learning_rate": 4.652715939447908e-05,
      "loss": 1.0028,
      "step": 1170
    },
    {
      "epoch": 0.3502523003858712,
      "grad_norm": 0.6333305239677429,
      "learning_rate": 4.6497476996141294e-05,
      "loss": 1.0121,
      "step": 1180
    },
    {
      "epoch": 0.35322054021964977,
      "grad_norm": 0.6222066879272461,
      "learning_rate": 4.6467794597803504e-05,
      "loss": 1.0151,
      "step": 1190
    },
    {
      "epoch": 0.3561887800534283,
      "grad_norm": 0.6540125608444214,
      "learning_rate": 4.643811219946571e-05,
      "loss": 0.9984,
      "step": 1200
    },
    {
      "epoch": 0.3591570198872069,
      "grad_norm": 0.7811010479927063,
      "learning_rate": 4.6408429801127936e-05,
      "loss": 1.035,
      "step": 1210
    },
    {
      "epoch": 0.36212525972098547,
      "grad_norm": 0.7098474502563477,
      "learning_rate": 4.6378747402790145e-05,
      "loss": 1.0029,
      "step": 1220
    },
    {
      "epoch": 0.365093499554764,
      "grad_norm": 0.7310671210289001,
      "learning_rate": 4.634906500445236e-05,
      "loss": 0.9954,
      "step": 1230
    },
    {
      "epoch": 0.3680617393885426,
      "grad_norm": 0.8500426411628723,
      "learning_rate": 4.631938260611458e-05,
      "loss": 0.9771,
      "step": 1240
    },
    {
      "epoch": 0.37102997922232117,
      "grad_norm": 0.6685720682144165,
      "learning_rate": 4.6289700207776794e-05,
      "loss": 1.007,
      "step": 1250
    },
    {
      "epoch": 0.37399821905609976,
      "grad_norm": 0.6676350831985474,
      "learning_rate": 4.6260017809439003e-05,
      "loss": 1.0382,
      "step": 1260
    },
    {
      "epoch": 0.3769664588898783,
      "grad_norm": 0.6712237000465393,
      "learning_rate": 4.623033541110122e-05,
      "loss": 1.0302,
      "step": 1270
    },
    {
      "epoch": 0.37993469872365687,
      "grad_norm": 0.6046270132064819,
      "learning_rate": 4.6200653012763436e-05,
      "loss": 1.0235,
      "step": 1280
    },
    {
      "epoch": 0.38290293855743546,
      "grad_norm": 0.6848681569099426,
      "learning_rate": 4.6170970614425645e-05,
      "loss": 1.0059,
      "step": 1290
    },
    {
      "epoch": 0.385871178391214,
      "grad_norm": 0.7414882183074951,
      "learning_rate": 4.614128821608786e-05,
      "loss": 1.0209,
      "step": 1300
    },
    {
      "epoch": 0.3888394182249926,
      "grad_norm": 0.7111567258834839,
      "learning_rate": 4.611160581775008e-05,
      "loss": 1.0614,
      "step": 1310
    },
    {
      "epoch": 0.39180765805877116,
      "grad_norm": 0.6949933767318726,
      "learning_rate": 4.6081923419412294e-05,
      "loss": 1.0055,
      "step": 1320
    },
    {
      "epoch": 0.39477589789254974,
      "grad_norm": 0.7095695734024048,
      "learning_rate": 4.60522410210745e-05,
      "loss": 0.9842,
      "step": 1330
    },
    {
      "epoch": 0.3977441377263283,
      "grad_norm": 0.8123998045921326,
      "learning_rate": 4.602255862273672e-05,
      "loss": 1.0609,
      "step": 1340
    },
    {
      "epoch": 0.40071237756010686,
      "grad_norm": 0.7480998039245605,
      "learning_rate": 4.5992876224398935e-05,
      "loss": 1.0,
      "step": 1350
    },
    {
      "epoch": 0.40368061739388544,
      "grad_norm": 0.7112509608268738,
      "learning_rate": 4.5963193826061145e-05,
      "loss": 1.005,
      "step": 1360
    },
    {
      "epoch": 0.406648857227664,
      "grad_norm": 0.7034320831298828,
      "learning_rate": 4.593351142772336e-05,
      "loss": 0.9875,
      "step": 1370
    },
    {
      "epoch": 0.40961709706144256,
      "grad_norm": 0.6642884612083435,
      "learning_rate": 4.590382902938558e-05,
      "loss": 1.0013,
      "step": 1380
    },
    {
      "epoch": 0.41258533689522114,
      "grad_norm": 0.6435369849205017,
      "learning_rate": 4.587414663104779e-05,
      "loss": 1.0212,
      "step": 1390
    },
    {
      "epoch": 0.41555357672899973,
      "grad_norm": 0.7438673377037048,
      "learning_rate": 4.584446423271e-05,
      "loss": 1.0401,
      "step": 1400
    },
    {
      "epoch": 0.41852181656277826,
      "grad_norm": 0.6593348979949951,
      "learning_rate": 4.581478183437222e-05,
      "loss": 1.0217,
      "step": 1410
    },
    {
      "epoch": 0.42149005639655684,
      "grad_norm": 0.7108020782470703,
      "learning_rate": 4.5785099436034435e-05,
      "loss": 0.9836,
      "step": 1420
    },
    {
      "epoch": 0.42445829623033543,
      "grad_norm": 0.6754259467124939,
      "learning_rate": 4.5755417037696645e-05,
      "loss": 1.0715,
      "step": 1430
    },
    {
      "epoch": 0.42742653606411396,
      "grad_norm": 0.6980791091918945,
      "learning_rate": 4.572573463935886e-05,
      "loss": 1.0257,
      "step": 1440
    },
    {
      "epoch": 0.43039477589789255,
      "grad_norm": 0.7558574080467224,
      "learning_rate": 4.569605224102108e-05,
      "loss": 0.9879,
      "step": 1450
    },
    {
      "epoch": 0.43336301573167113,
      "grad_norm": 0.6725367307662964,
      "learning_rate": 4.566636984268329e-05,
      "loss": 1.0153,
      "step": 1460
    },
    {
      "epoch": 0.4363312555654497,
      "grad_norm": 0.8197546005249023,
      "learning_rate": 4.56366874443455e-05,
      "loss": 0.977,
      "step": 1470
    },
    {
      "epoch": 0.43929949539922825,
      "grad_norm": 0.6687082052230835,
      "learning_rate": 4.560700504600772e-05,
      "loss": 1.0105,
      "step": 1480
    },
    {
      "epoch": 0.44226773523300683,
      "grad_norm": 0.7228406667709351,
      "learning_rate": 4.5577322647669935e-05,
      "loss": 0.9827,
      "step": 1490
    },
    {
      "epoch": 0.4452359750667854,
      "grad_norm": 0.7459806799888611,
      "learning_rate": 4.5547640249332144e-05,
      "loss": 1.0305,
      "step": 1500
    },
    {
      "epoch": 0.44820421490056395,
      "grad_norm": 0.6408952474594116,
      "learning_rate": 4.551795785099437e-05,
      "loss": 0.9861,
      "step": 1510
    },
    {
      "epoch": 0.45117245473434253,
      "grad_norm": 0.7314732074737549,
      "learning_rate": 4.5488275452656577e-05,
      "loss": 0.9961,
      "step": 1520
    },
    {
      "epoch": 0.4541406945681211,
      "grad_norm": 0.7654784917831421,
      "learning_rate": 4.545859305431879e-05,
      "loss": 1.019,
      "step": 1530
    },
    {
      "epoch": 0.45710893440189965,
      "grad_norm": 0.7305381298065186,
      "learning_rate": 4.5428910655981e-05,
      "loss": 0.9938,
      "step": 1540
    },
    {
      "epoch": 0.46007717423567823,
      "grad_norm": 0.6413453221321106,
      "learning_rate": 4.539922825764322e-05,
      "loss": 0.994,
      "step": 1550
    },
    {
      "epoch": 0.4630454140694568,
      "grad_norm": 0.6420827507972717,
      "learning_rate": 4.5369545859305434e-05,
      "loss": 1.0237,
      "step": 1560
    },
    {
      "epoch": 0.4660136539032354,
      "grad_norm": 0.6597052216529846,
      "learning_rate": 4.5339863460967644e-05,
      "loss": 1.0142,
      "step": 1570
    },
    {
      "epoch": 0.46898189373701393,
      "grad_norm": 0.727504312992096,
      "learning_rate": 4.531018106262987e-05,
      "loss": 1.0349,
      "step": 1580
    },
    {
      "epoch": 0.4719501335707925,
      "grad_norm": 0.6038479208946228,
      "learning_rate": 4.5280498664292076e-05,
      "loss": 0.9878,
      "step": 1590
    },
    {
      "epoch": 0.4749183734045711,
      "grad_norm": 0.6358582973480225,
      "learning_rate": 4.525081626595429e-05,
      "loss": 0.9822,
      "step": 1600
    },
    {
      "epoch": 0.47788661323834963,
      "grad_norm": 0.6879693865776062,
      "learning_rate": 4.522113386761651e-05,
      "loss": 1.022,
      "step": 1610
    },
    {
      "epoch": 0.4808548530721282,
      "grad_norm": 0.6969800591468811,
      "learning_rate": 4.519145146927872e-05,
      "loss": 0.999,
      "step": 1620
    },
    {
      "epoch": 0.4838230929059068,
      "grad_norm": 0.632183849811554,
      "learning_rate": 4.5161769070940934e-05,
      "loss": 0.9845,
      "step": 1630
    },
    {
      "epoch": 0.4867913327396854,
      "grad_norm": 0.6474229097366333,
      "learning_rate": 4.5132086672603144e-05,
      "loss": 0.9895,
      "step": 1640
    },
    {
      "epoch": 0.4897595725734639,
      "grad_norm": 0.7708784937858582,
      "learning_rate": 4.5102404274265367e-05,
      "loss": 1.0074,
      "step": 1650
    },
    {
      "epoch": 0.4927278124072425,
      "grad_norm": 0.7153667211532593,
      "learning_rate": 4.5072721875927576e-05,
      "loss": 0.9792,
      "step": 1660
    },
    {
      "epoch": 0.4956960522410211,
      "grad_norm": 0.7407136559486389,
      "learning_rate": 4.504303947758979e-05,
      "loss": 1.008,
      "step": 1670
    },
    {
      "epoch": 0.4986642920747996,
      "grad_norm": 0.7044761776924133,
      "learning_rate": 4.501335707925201e-05,
      "loss": 0.9908,
      "step": 1680
    },
    {
      "epoch": 0.5016325319085783,
      "grad_norm": 0.7155528664588928,
      "learning_rate": 4.498367468091422e-05,
      "loss": 0.9659,
      "step": 1690
    },
    {
      "epoch": 0.5046007717423567,
      "grad_norm": 0.7560014128684998,
      "learning_rate": 4.4953992282576434e-05,
      "loss": 1.0177,
      "step": 1700
    },
    {
      "epoch": 0.5075690115761353,
      "grad_norm": 0.6918985247612,
      "learning_rate": 4.492430988423865e-05,
      "loss": 0.9929,
      "step": 1710
    },
    {
      "epoch": 0.5105372514099139,
      "grad_norm": 0.6356712579727173,
      "learning_rate": 4.4894627485900866e-05,
      "loss": 0.9851,
      "step": 1720
    },
    {
      "epoch": 0.5135054912436925,
      "grad_norm": 0.6340317130088806,
      "learning_rate": 4.4864945087563076e-05,
      "loss": 0.9821,
      "step": 1730
    },
    {
      "epoch": 0.5164737310774711,
      "grad_norm": 0.654170572757721,
      "learning_rate": 4.483526268922529e-05,
      "loss": 1.0106,
      "step": 1740
    },
    {
      "epoch": 0.5194419709112497,
      "grad_norm": 0.6271511912345886,
      "learning_rate": 4.480558029088751e-05,
      "loss": 0.9998,
      "step": 1750
    },
    {
      "epoch": 0.5224102107450282,
      "grad_norm": 0.6410935521125793,
      "learning_rate": 4.477589789254972e-05,
      "loss": 1.0079,
      "step": 1760
    },
    {
      "epoch": 0.5253784505788067,
      "grad_norm": 0.6426156759262085,
      "learning_rate": 4.4746215494211934e-05,
      "loss": 0.9827,
      "step": 1770
    },
    {
      "epoch": 0.5283466904125853,
      "grad_norm": 0.6262437701225281,
      "learning_rate": 4.471653309587415e-05,
      "loss": 1.0245,
      "step": 1780
    },
    {
      "epoch": 0.5313149302463639,
      "grad_norm": 0.6619899868965149,
      "learning_rate": 4.4686850697536366e-05,
      "loss": 0.9612,
      "step": 1790
    },
    {
      "epoch": 0.5342831700801425,
      "grad_norm": 0.7522478699684143,
      "learning_rate": 4.4657168299198575e-05,
      "loss": 0.9784,
      "step": 1800
    },
    {
      "epoch": 0.5372514099139211,
      "grad_norm": 0.69633948802948,
      "learning_rate": 4.462748590086079e-05,
      "loss": 0.9925,
      "step": 1810
    },
    {
      "epoch": 0.5402196497476996,
      "grad_norm": 0.6923314929008484,
      "learning_rate": 4.459780350252301e-05,
      "loss": 1.0409,
      "step": 1820
    },
    {
      "epoch": 0.5431878895814782,
      "grad_norm": 0.6425520181655884,
      "learning_rate": 4.456812110418522e-05,
      "loss": 1.0117,
      "step": 1830
    },
    {
      "epoch": 0.5461561294152567,
      "grad_norm": 0.6298788189888,
      "learning_rate": 4.453843870584743e-05,
      "loss": 1.0165,
      "step": 1840
    },
    {
      "epoch": 0.5491243692490353,
      "grad_norm": 0.6069309711456299,
      "learning_rate": 4.450875630750965e-05,
      "loss": 0.9799,
      "step": 1850
    },
    {
      "epoch": 0.5520926090828139,
      "grad_norm": 0.6480023860931396,
      "learning_rate": 4.4479073909171866e-05,
      "loss": 0.988,
      "step": 1860
    },
    {
      "epoch": 0.5550608489165925,
      "grad_norm": 0.607940673828125,
      "learning_rate": 4.4449391510834075e-05,
      "loss": 0.9802,
      "step": 1870
    },
    {
      "epoch": 0.558029088750371,
      "grad_norm": 0.6470855474472046,
      "learning_rate": 4.441970911249629e-05,
      "loss": 1.0095,
      "step": 1880
    },
    {
      "epoch": 0.5609973285841496,
      "grad_norm": 0.6894873976707458,
      "learning_rate": 4.439002671415851e-05,
      "loss": 0.977,
      "step": 1890
    },
    {
      "epoch": 0.5639655684179282,
      "grad_norm": 0.6529555320739746,
      "learning_rate": 4.436034431582072e-05,
      "loss": 1.0071,
      "step": 1900
    },
    {
      "epoch": 0.5669338082517067,
      "grad_norm": 0.743811845779419,
      "learning_rate": 4.433066191748293e-05,
      "loss": 0.9989,
      "step": 1910
    },
    {
      "epoch": 0.5699020480854853,
      "grad_norm": 0.8114213347434998,
      "learning_rate": 4.430097951914515e-05,
      "loss": 0.9627,
      "step": 1920
    },
    {
      "epoch": 0.5728702879192639,
      "grad_norm": 0.6733983159065247,
      "learning_rate": 4.4271297120807365e-05,
      "loss": 0.9975,
      "step": 1930
    },
    {
      "epoch": 0.5758385277530425,
      "grad_norm": 0.692699134349823,
      "learning_rate": 4.4241614722469575e-05,
      "loss": 0.9982,
      "step": 1940
    },
    {
      "epoch": 0.578806767586821,
      "grad_norm": 0.6752103567123413,
      "learning_rate": 4.421193232413179e-05,
      "loss": 0.9963,
      "step": 1950
    },
    {
      "epoch": 0.5817750074205996,
      "grad_norm": 0.7230833768844604,
      "learning_rate": 4.418224992579401e-05,
      "loss": 0.9941,
      "step": 1960
    },
    {
      "epoch": 0.5847432472543782,
      "grad_norm": 0.7616879343986511,
      "learning_rate": 4.4152567527456216e-05,
      "loss": 0.9763,
      "step": 1970
    },
    {
      "epoch": 0.5877114870881567,
      "grad_norm": 0.6616672873497009,
      "learning_rate": 4.412288512911843e-05,
      "loss": 0.9777,
      "step": 1980
    },
    {
      "epoch": 0.5906797269219353,
      "grad_norm": 0.7364579439163208,
      "learning_rate": 4.409320273078065e-05,
      "loss": 0.9786,
      "step": 1990
    },
    {
      "epoch": 0.5936479667557139,
      "grad_norm": 0.6607961654663086,
      "learning_rate": 4.4063520332442865e-05,
      "loss": 1.0312,
      "step": 2000
    },
    {
      "epoch": 0.5966162065894924,
      "grad_norm": 0.6173883080482483,
      "learning_rate": 4.4033837934105074e-05,
      "loss": 0.9894,
      "step": 2010
    },
    {
      "epoch": 0.599584446423271,
      "grad_norm": 0.8147071599960327,
      "learning_rate": 4.40041555357673e-05,
      "loss": 0.9706,
      "step": 2020
    },
    {
      "epoch": 0.6025526862570496,
      "grad_norm": 0.6534470319747925,
      "learning_rate": 4.397447313742951e-05,
      "loss": 0.9689,
      "step": 2030
    },
    {
      "epoch": 0.6055209260908282,
      "grad_norm": 0.6868370771408081,
      "learning_rate": 4.3944790739091716e-05,
      "loss": 0.9884,
      "step": 2040
    },
    {
      "epoch": 0.6084891659246067,
      "grad_norm": 0.6805328130722046,
      "learning_rate": 4.391510834075393e-05,
      "loss": 0.9819,
      "step": 2050
    },
    {
      "epoch": 0.6114574057583853,
      "grad_norm": 0.6883294582366943,
      "learning_rate": 4.388542594241615e-05,
      "loss": 1.0165,
      "step": 2060
    },
    {
      "epoch": 0.6144256455921638,
      "grad_norm": 0.635663628578186,
      "learning_rate": 4.3855743544078365e-05,
      "loss": 0.9839,
      "step": 2070
    },
    {
      "epoch": 0.6173938854259424,
      "grad_norm": 0.6705318689346313,
      "learning_rate": 4.3826061145740574e-05,
      "loss": 0.9674,
      "step": 2080
    },
    {
      "epoch": 0.620362125259721,
      "grad_norm": 0.7084488868713379,
      "learning_rate": 4.37963787474028e-05,
      "loss": 0.9787,
      "step": 2090
    },
    {
      "epoch": 0.6233303650934996,
      "grad_norm": 0.6482323408126831,
      "learning_rate": 4.3766696349065006e-05,
      "loss": 0.998,
      "step": 2100
    },
    {
      "epoch": 0.6262986049272781,
      "grad_norm": 0.6266573071479797,
      "learning_rate": 4.373701395072722e-05,
      "loss": 0.9931,
      "step": 2110
    },
    {
      "epoch": 0.6292668447610567,
      "grad_norm": 0.7121156454086304,
      "learning_rate": 4.370733155238944e-05,
      "loss": 1.0117,
      "step": 2120
    },
    {
      "epoch": 0.6322350845948352,
      "grad_norm": 0.6650971174240112,
      "learning_rate": 4.367764915405165e-05,
      "loss": 0.9827,
      "step": 2130
    },
    {
      "epoch": 0.6352033244286138,
      "grad_norm": 0.6248673796653748,
      "learning_rate": 4.3647966755713864e-05,
      "loss": 0.9938,
      "step": 2140
    },
    {
      "epoch": 0.6381715642623924,
      "grad_norm": 0.5886913537979126,
      "learning_rate": 4.3618284357376074e-05,
      "loss": 0.9722,
      "step": 2150
    },
    {
      "epoch": 0.641139804096171,
      "grad_norm": 0.6873770356178284,
      "learning_rate": 4.35886019590383e-05,
      "loss": 1.0117,
      "step": 2160
    },
    {
      "epoch": 0.6441080439299496,
      "grad_norm": 0.5833357572555542,
      "learning_rate": 4.3558919560700506e-05,
      "loss": 0.9733,
      "step": 2170
    },
    {
      "epoch": 0.647076283763728,
      "grad_norm": 0.6192591190338135,
      "learning_rate": 4.352923716236272e-05,
      "loss": 1.0046,
      "step": 2180
    },
    {
      "epoch": 0.6500445235975066,
      "grad_norm": 0.657365083694458,
      "learning_rate": 4.349955476402494e-05,
      "loss": 0.9808,
      "step": 2190
    },
    {
      "epoch": 0.6530127634312852,
      "grad_norm": 0.6927903890609741,
      "learning_rate": 4.346987236568715e-05,
      "loss": 1.0062,
      "step": 2200
    },
    {
      "epoch": 0.6559810032650638,
      "grad_norm": 0.6259236931800842,
      "learning_rate": 4.3440189967349364e-05,
      "loss": 0.99,
      "step": 2210
    },
    {
      "epoch": 0.6589492430988424,
      "grad_norm": 0.659789502620697,
      "learning_rate": 4.341050756901158e-05,
      "loss": 0.952,
      "step": 2220
    },
    {
      "epoch": 0.661917482932621,
      "grad_norm": 0.6317498087882996,
      "learning_rate": 4.3380825170673796e-05,
      "loss": 0.9938,
      "step": 2230
    },
    {
      "epoch": 0.6648857227663996,
      "grad_norm": 0.6758004426956177,
      "learning_rate": 4.3351142772336006e-05,
      "loss": 0.9761,
      "step": 2240
    },
    {
      "epoch": 0.667853962600178,
      "grad_norm": 0.707929790019989,
      "learning_rate": 4.332146037399822e-05,
      "loss": 0.9839,
      "step": 2250
    },
    {
      "epoch": 0.6708222024339566,
      "grad_norm": 0.6813105344772339,
      "learning_rate": 4.329177797566044e-05,
      "loss": 0.9933,
      "step": 2260
    },
    {
      "epoch": 0.6737904422677352,
      "grad_norm": 0.7361823320388794,
      "learning_rate": 4.326209557732265e-05,
      "loss": 1.0212,
      "step": 2270
    },
    {
      "epoch": 0.6767586821015138,
      "grad_norm": 0.670718252658844,
      "learning_rate": 4.3232413178984864e-05,
      "loss": 0.9638,
      "step": 2280
    },
    {
      "epoch": 0.6797269219352924,
      "grad_norm": 0.6154988408088684,
      "learning_rate": 4.320273078064708e-05,
      "loss": 0.9738,
      "step": 2290
    },
    {
      "epoch": 0.682695161769071,
      "grad_norm": 0.6230899095535278,
      "learning_rate": 4.3173048382309296e-05,
      "loss": 0.9931,
      "step": 2300
    },
    {
      "epoch": 0.6856634016028496,
      "grad_norm": 0.6064435839653015,
      "learning_rate": 4.3143365983971505e-05,
      "loss": 1.0204,
      "step": 2310
    },
    {
      "epoch": 0.688631641436628,
      "grad_norm": 0.635154128074646,
      "learning_rate": 4.311368358563372e-05,
      "loss": 0.9353,
      "step": 2320
    },
    {
      "epoch": 0.6915998812704066,
      "grad_norm": 0.6036397814750671,
      "learning_rate": 4.308400118729594e-05,
      "loss": 1.0105,
      "step": 2330
    },
    {
      "epoch": 0.6945681211041852,
      "grad_norm": 0.6203991770744324,
      "learning_rate": 4.305431878895815e-05,
      "loss": 0.9804,
      "step": 2340
    },
    {
      "epoch": 0.6975363609379638,
      "grad_norm": 0.6208158135414124,
      "learning_rate": 4.302463639062036e-05,
      "loss": 0.9271,
      "step": 2350
    },
    {
      "epoch": 0.7005046007717424,
      "grad_norm": 0.681175708770752,
      "learning_rate": 4.299495399228258e-05,
      "loss": 0.9734,
      "step": 2360
    },
    {
      "epoch": 0.703472840605521,
      "grad_norm": 0.6127511262893677,
      "learning_rate": 4.2965271593944796e-05,
      "loss": 0.9544,
      "step": 2370
    },
    {
      "epoch": 0.7064410804392995,
      "grad_norm": 0.6363290548324585,
      "learning_rate": 4.2935589195607005e-05,
      "loss": 0.9932,
      "step": 2380
    },
    {
      "epoch": 0.709409320273078,
      "grad_norm": 0.656937837600708,
      "learning_rate": 4.290590679726922e-05,
      "loss": 0.9682,
      "step": 2390
    },
    {
      "epoch": 0.7123775601068566,
      "grad_norm": 0.6660462021827698,
      "learning_rate": 4.287622439893144e-05,
      "loss": 0.983,
      "step": 2400
    },
    {
      "epoch": 0.7153457999406352,
      "grad_norm": 0.6178123354911804,
      "learning_rate": 4.284654200059365e-05,
      "loss": 0.9757,
      "step": 2410
    },
    {
      "epoch": 0.7183140397744138,
      "grad_norm": 0.7197669744491577,
      "learning_rate": 4.281685960225586e-05,
      "loss": 0.9877,
      "step": 2420
    },
    {
      "epoch": 0.7212822796081924,
      "grad_norm": 0.6619410514831543,
      "learning_rate": 4.278717720391808e-05,
      "loss": 1.0249,
      "step": 2430
    },
    {
      "epoch": 0.7242505194419709,
      "grad_norm": 0.5938183665275574,
      "learning_rate": 4.2757494805580295e-05,
      "loss": 0.9593,
      "step": 2440
    },
    {
      "epoch": 0.7272187592757495,
      "grad_norm": 0.6519368886947632,
      "learning_rate": 4.2727812407242505e-05,
      "loss": 0.9552,
      "step": 2450
    },
    {
      "epoch": 0.730186999109528,
      "grad_norm": 0.6326847672462463,
      "learning_rate": 4.269813000890472e-05,
      "loss": 0.9283,
      "step": 2460
    },
    {
      "epoch": 0.7331552389433066,
      "grad_norm": 0.6411892175674438,
      "learning_rate": 4.266844761056694e-05,
      "loss": 1.004,
      "step": 2470
    },
    {
      "epoch": 0.7361234787770852,
      "grad_norm": 0.7936878204345703,
      "learning_rate": 4.2638765212229146e-05,
      "loss": 0.9515,
      "step": 2480
    },
    {
      "epoch": 0.7390917186108638,
      "grad_norm": 0.7002474069595337,
      "learning_rate": 4.260908281389136e-05,
      "loss": 0.9904,
      "step": 2490
    },
    {
      "epoch": 0.7420599584446423,
      "grad_norm": 0.5938469171524048,
      "learning_rate": 4.257940041555358e-05,
      "loss": 0.9709,
      "step": 2500
    },
    {
      "epoch": 0.7450281982784209,
      "grad_norm": 0.6611453890800476,
      "learning_rate": 4.2549718017215795e-05,
      "loss": 0.9774,
      "step": 2510
    },
    {
      "epoch": 0.7479964381121995,
      "grad_norm": 0.6640607714653015,
      "learning_rate": 4.2520035618878004e-05,
      "loss": 0.972,
      "step": 2520
    },
    {
      "epoch": 0.750964677945978,
      "grad_norm": 0.6462842226028442,
      "learning_rate": 4.249035322054023e-05,
      "loss": 0.9776,
      "step": 2530
    },
    {
      "epoch": 0.7539329177797566,
      "grad_norm": 0.5849109292030334,
      "learning_rate": 4.246067082220244e-05,
      "loss": 1.0179,
      "step": 2540
    },
    {
      "epoch": 0.7569011576135352,
      "grad_norm": 0.6768197417259216,
      "learning_rate": 4.2430988423864646e-05,
      "loss": 0.9714,
      "step": 2550
    },
    {
      "epoch": 0.7598693974473137,
      "grad_norm": 0.646824300289154,
      "learning_rate": 4.240130602552686e-05,
      "loss": 1.023,
      "step": 2560
    },
    {
      "epoch": 0.7628376372810923,
      "grad_norm": 0.7211414575576782,
      "learning_rate": 4.237162362718908e-05,
      "loss": 0.972,
      "step": 2570
    },
    {
      "epoch": 0.7658058771148709,
      "grad_norm": 0.6269819736480713,
      "learning_rate": 4.2341941228851295e-05,
      "loss": 0.9738,
      "step": 2580
    },
    {
      "epoch": 0.7687741169486495,
      "grad_norm": 0.6858998537063599,
      "learning_rate": 4.2312258830513504e-05,
      "loss": 1.0041,
      "step": 2590
    },
    {
      "epoch": 0.771742356782428,
      "grad_norm": 0.5189050436019897,
      "learning_rate": 4.228257643217573e-05,
      "loss": 0.9643,
      "step": 2600
    },
    {
      "epoch": 0.7747105966162066,
      "grad_norm": 0.6885724067687988,
      "learning_rate": 4.2252894033837936e-05,
      "loss": 0.943,
      "step": 2610
    },
    {
      "epoch": 0.7776788364499851,
      "grad_norm": 0.6750602126121521,
      "learning_rate": 4.2223211635500146e-05,
      "loss": 0.9982,
      "step": 2620
    },
    {
      "epoch": 0.7806470762837637,
      "grad_norm": 0.659366250038147,
      "learning_rate": 4.219352923716237e-05,
      "loss": 1.0041,
      "step": 2630
    },
    {
      "epoch": 0.7836153161175423,
      "grad_norm": 0.6575495004653931,
      "learning_rate": 4.216384683882458e-05,
      "loss": 0.9499,
      "step": 2640
    },
    {
      "epoch": 0.7865835559513209,
      "grad_norm": 0.6087556481361389,
      "learning_rate": 4.2134164440486794e-05,
      "loss": 1.0186,
      "step": 2650
    },
    {
      "epoch": 0.7895517957850995,
      "grad_norm": 0.5324801206588745,
      "learning_rate": 4.2104482042149004e-05,
      "loss": 1.0128,
      "step": 2660
    },
    {
      "epoch": 0.792520035618878,
      "grad_norm": 0.6805785298347473,
      "learning_rate": 4.207479964381123e-05,
      "loss": 0.9591,
      "step": 2670
    },
    {
      "epoch": 0.7954882754526565,
      "grad_norm": 0.624666690826416,
      "learning_rate": 4.2045117245473436e-05,
      "loss": 0.9764,
      "step": 2680
    },
    {
      "epoch": 0.7984565152864351,
      "grad_norm": 0.6150093078613281,
      "learning_rate": 4.2015434847135646e-05,
      "loss": 1.0,
      "step": 2690
    },
    {
      "epoch": 0.8014247551202137,
      "grad_norm": 0.6825082898139954,
      "learning_rate": 4.198575244879787e-05,
      "loss": 0.98,
      "step": 2700
    },
    {
      "epoch": 0.8043929949539923,
      "grad_norm": 0.745524525642395,
      "learning_rate": 4.195607005046008e-05,
      "loss": 0.9782,
      "step": 2710
    },
    {
      "epoch": 0.8073612347877709,
      "grad_norm": 0.6018673777580261,
      "learning_rate": 4.1926387652122294e-05,
      "loss": 1.0179,
      "step": 2720
    },
    {
      "epoch": 0.8103294746215495,
      "grad_norm": 0.5541884303092957,
      "learning_rate": 4.189670525378451e-05,
      "loss": 0.9905,
      "step": 2730
    },
    {
      "epoch": 0.813297714455328,
      "grad_norm": 0.6213927865028381,
      "learning_rate": 4.1867022855446726e-05,
      "loss": 1.0215,
      "step": 2740
    },
    {
      "epoch": 0.8162659542891065,
      "grad_norm": 0.6007673740386963,
      "learning_rate": 4.1837340457108936e-05,
      "loss": 0.9846,
      "step": 2750
    },
    {
      "epoch": 0.8192341941228851,
      "grad_norm": 0.678551197052002,
      "learning_rate": 4.180765805877115e-05,
      "loss": 0.9933,
      "step": 2760
    },
    {
      "epoch": 0.8222024339566637,
      "grad_norm": 0.5489323735237122,
      "learning_rate": 4.177797566043337e-05,
      "loss": 0.9805,
      "step": 2770
    },
    {
      "epoch": 0.8251706737904423,
      "grad_norm": 0.6029487252235413,
      "learning_rate": 4.174829326209558e-05,
      "loss": 0.9983,
      "step": 2780
    },
    {
      "epoch": 0.8281389136242209,
      "grad_norm": 0.5949262976646423,
      "learning_rate": 4.1718610863757794e-05,
      "loss": 0.9883,
      "step": 2790
    },
    {
      "epoch": 0.8311071534579995,
      "grad_norm": 0.5981915593147278,
      "learning_rate": 4.168892846542001e-05,
      "loss": 0.9652,
      "step": 2800
    },
    {
      "epoch": 0.8340753932917779,
      "grad_norm": 0.6043820977210999,
      "learning_rate": 4.1659246067082226e-05,
      "loss": 0.9893,
      "step": 2810
    },
    {
      "epoch": 0.8370436331255565,
      "grad_norm": 0.6542451977729797,
      "learning_rate": 4.1629563668744435e-05,
      "loss": 1.0184,
      "step": 2820
    },
    {
      "epoch": 0.8400118729593351,
      "grad_norm": 0.5959357619285583,
      "learning_rate": 4.159988127040665e-05,
      "loss": 1.008,
      "step": 2830
    },
    {
      "epoch": 0.8429801127931137,
      "grad_norm": 0.658277153968811,
      "learning_rate": 4.157019887206887e-05,
      "loss": 0.9579,
      "step": 2840
    },
    {
      "epoch": 0.8459483526268923,
      "grad_norm": 0.6425391435623169,
      "learning_rate": 4.154051647373108e-05,
      "loss": 0.9705,
      "step": 2850
    },
    {
      "epoch": 0.8489165924606709,
      "grad_norm": 0.49258121848106384,
      "learning_rate": 4.1510834075393293e-05,
      "loss": 0.9488,
      "step": 2860
    },
    {
      "epoch": 0.8518848322944494,
      "grad_norm": 0.6020271182060242,
      "learning_rate": 4.148115167705551e-05,
      "loss": 0.9847,
      "step": 2870
    },
    {
      "epoch": 0.8548530721282279,
      "grad_norm": 0.659537136554718,
      "learning_rate": 4.1451469278717726e-05,
      "loss": 0.9962,
      "step": 2880
    },
    {
      "epoch": 0.8578213119620065,
      "grad_norm": 0.6259986758232117,
      "learning_rate": 4.1421786880379935e-05,
      "loss": 0.9601,
      "step": 2890
    },
    {
      "epoch": 0.8607895517957851,
      "grad_norm": 0.6433762907981873,
      "learning_rate": 4.139210448204215e-05,
      "loss": 1.0173,
      "step": 2900
    },
    {
      "epoch": 0.8637577916295637,
      "grad_norm": 0.6561344861984253,
      "learning_rate": 4.136242208370437e-05,
      "loss": 0.987,
      "step": 2910
    },
    {
      "epoch": 0.8667260314633423,
      "grad_norm": 0.6170705556869507,
      "learning_rate": 4.133273968536658e-05,
      "loss": 0.9751,
      "step": 2920
    },
    {
      "epoch": 0.8696942712971208,
      "grad_norm": 0.6285633444786072,
      "learning_rate": 4.130305728702879e-05,
      "loss": 1.0202,
      "step": 2930
    },
    {
      "epoch": 0.8726625111308994,
      "grad_norm": 0.5784567594528198,
      "learning_rate": 4.127337488869101e-05,
      "loss": 0.984,
      "step": 2940
    },
    {
      "epoch": 0.8756307509646779,
      "grad_norm": 0.6021987795829773,
      "learning_rate": 4.1243692490353225e-05,
      "loss": 1.0028,
      "step": 2950
    },
    {
      "epoch": 0.8785989907984565,
      "grad_norm": 0.7186152338981628,
      "learning_rate": 4.1214010092015435e-05,
      "loss": 0.9872,
      "step": 2960
    },
    {
      "epoch": 0.8815672306322351,
      "grad_norm": 0.6320556402206421,
      "learning_rate": 4.118432769367765e-05,
      "loss": 0.958,
      "step": 2970
    },
    {
      "epoch": 0.8845354704660137,
      "grad_norm": 0.6496281623840332,
      "learning_rate": 4.115464529533987e-05,
      "loss": 0.9957,
      "step": 2980
    },
    {
      "epoch": 0.8875037102997922,
      "grad_norm": 0.7143193483352661,
      "learning_rate": 4.1124962897002077e-05,
      "loss": 0.9968,
      "step": 2990
    },
    {
      "epoch": 0.8904719501335708,
      "grad_norm": 0.5800313949584961,
      "learning_rate": 4.109528049866429e-05,
      "loss": 0.9762,
      "step": 3000
    },
    {
      "epoch": 0.8934401899673493,
      "grad_norm": 0.6602805852890015,
      "learning_rate": 4.106559810032651e-05,
      "loss": 1.0245,
      "step": 3010
    },
    {
      "epoch": 0.8964084298011279,
      "grad_norm": 0.6473674774169922,
      "learning_rate": 4.1035915701988725e-05,
      "loss": 1.0021,
      "step": 3020
    },
    {
      "epoch": 0.8993766696349065,
      "grad_norm": 0.6156338453292847,
      "learning_rate": 4.1006233303650935e-05,
      "loss": 0.9532,
      "step": 3030
    },
    {
      "epoch": 0.9023449094686851,
      "grad_norm": 0.5925867557525635,
      "learning_rate": 4.097655090531315e-05,
      "loss": 0.9177,
      "step": 3040
    },
    {
      "epoch": 0.9053131493024636,
      "grad_norm": 0.6322952508926392,
      "learning_rate": 4.094686850697537e-05,
      "loss": 0.9607,
      "step": 3050
    },
    {
      "epoch": 0.9082813891362422,
      "grad_norm": 0.6165744662284851,
      "learning_rate": 4.0917186108637576e-05,
      "loss": 0.9635,
      "step": 3060
    },
    {
      "epoch": 0.9112496289700208,
      "grad_norm": 0.58901447057724,
      "learning_rate": 4.08875037102998e-05,
      "loss": 0.9799,
      "step": 3070
    },
    {
      "epoch": 0.9142178688037993,
      "grad_norm": 0.6456793546676636,
      "learning_rate": 4.085782131196201e-05,
      "loss": 0.9751,
      "step": 3080
    },
    {
      "epoch": 0.9171861086375779,
      "grad_norm": 0.5987860560417175,
      "learning_rate": 4.0828138913624225e-05,
      "loss": 0.9327,
      "step": 3090
    },
    {
      "epoch": 0.9201543484713565,
      "grad_norm": 0.5760492086410522,
      "learning_rate": 4.0798456515286434e-05,
      "loss": 0.982,
      "step": 3100
    },
    {
      "epoch": 0.923122588305135,
      "grad_norm": 0.5955607295036316,
      "learning_rate": 4.076877411694865e-05,
      "loss": 0.9669,
      "step": 3110
    },
    {
      "epoch": 0.9260908281389136,
      "grad_norm": 0.6458853483200073,
      "learning_rate": 4.0739091718610867e-05,
      "loss": 0.9424,
      "step": 3120
    },
    {
      "epoch": 0.9290590679726922,
      "grad_norm": 0.6266711354255676,
      "learning_rate": 4.0709409320273076e-05,
      "loss": 0.9545,
      "step": 3130
    },
    {
      "epoch": 0.9320273078064708,
      "grad_norm": 0.5927236080169678,
      "learning_rate": 4.06797269219353e-05,
      "loss": 0.9936,
      "step": 3140
    },
    {
      "epoch": 0.9349955476402493,
      "grad_norm": 0.5582984685897827,
      "learning_rate": 4.065004452359751e-05,
      "loss": 0.9653,
      "step": 3150
    },
    {
      "epoch": 0.9379637874740279,
      "grad_norm": 0.6802617907524109,
      "learning_rate": 4.0620362125259724e-05,
      "loss": 0.9792,
      "step": 3160
    },
    {
      "epoch": 0.9409320273078065,
      "grad_norm": 0.6594977974891663,
      "learning_rate": 4.059067972692194e-05,
      "loss": 0.9381,
      "step": 3170
    },
    {
      "epoch": 0.943900267141585,
      "grad_norm": 0.7159130573272705,
      "learning_rate": 4.056099732858415e-05,
      "loss": 0.9884,
      "step": 3180
    },
    {
      "epoch": 0.9468685069753636,
      "grad_norm": 0.7334505915641785,
      "learning_rate": 4.0531314930246366e-05,
      "loss": 0.9893,
      "step": 3190
    },
    {
      "epoch": 0.9498367468091422,
      "grad_norm": 0.6196255683898926,
      "learning_rate": 4.0501632531908576e-05,
      "loss": 1.0248,
      "step": 3200
    },
    {
      "epoch": 0.9528049866429208,
      "grad_norm": 0.5950149297714233,
      "learning_rate": 4.04719501335708e-05,
      "loss": 0.9723,
      "step": 3210
    },
    {
      "epoch": 0.9557732264766993,
      "grad_norm": 0.6434257626533508,
      "learning_rate": 4.044226773523301e-05,
      "loss": 0.979,
      "step": 3220
    },
    {
      "epoch": 0.9587414663104779,
      "grad_norm": 0.6103981137275696,
      "learning_rate": 4.0412585336895224e-05,
      "loss": 0.9667,
      "step": 3230
    },
    {
      "epoch": 0.9617097061442564,
      "grad_norm": 0.5639011859893799,
      "learning_rate": 4.038290293855744e-05,
      "loss": 0.9826,
      "step": 3240
    },
    {
      "epoch": 0.964677945978035,
      "grad_norm": 0.723637044429779,
      "learning_rate": 4.035322054021965e-05,
      "loss": 0.951,
      "step": 3250
    },
    {
      "epoch": 0.9676461858118136,
      "grad_norm": 0.6385572552680969,
      "learning_rate": 4.0323538141881866e-05,
      "loss": 0.9773,
      "step": 3260
    },
    {
      "epoch": 0.9706144256455922,
      "grad_norm": 0.5733060836791992,
      "learning_rate": 4.029385574354408e-05,
      "loss": 0.9407,
      "step": 3270
    },
    {
      "epoch": 0.9735826654793708,
      "grad_norm": 0.6948421001434326,
      "learning_rate": 4.02641733452063e-05,
      "loss": 0.9977,
      "step": 3280
    },
    {
      "epoch": 0.9765509053131493,
      "grad_norm": 0.6377859711647034,
      "learning_rate": 4.023449094686851e-05,
      "loss": 0.9441,
      "step": 3290
    },
    {
      "epoch": 0.9795191451469278,
      "grad_norm": 0.704069972038269,
      "learning_rate": 4.0204808548530724e-05,
      "loss": 1.0352,
      "step": 3300
    },
    {
      "epoch": 0.9824873849807064,
      "grad_norm": 0.6204991340637207,
      "learning_rate": 4.017512615019294e-05,
      "loss": 0.9821,
      "step": 3310
    },
    {
      "epoch": 0.985455624814485,
      "grad_norm": 0.6215436458587646,
      "learning_rate": 4.014544375185515e-05,
      "loss": 0.9572,
      "step": 3320
    },
    {
      "epoch": 0.9884238646482636,
      "grad_norm": 0.6409376859664917,
      "learning_rate": 4.0115761353517366e-05,
      "loss": 0.9914,
      "step": 3330
    },
    {
      "epoch": 0.9913921044820422,
      "grad_norm": 0.6437116861343384,
      "learning_rate": 4.008607895517958e-05,
      "loss": 0.9499,
      "step": 3340
    },
    {
      "epoch": 0.9943603443158208,
      "grad_norm": 0.5800116658210754,
      "learning_rate": 4.00563965568418e-05,
      "loss": 0.974,
      "step": 3350
    },
    {
      "epoch": 0.9973285841495992,
      "grad_norm": 0.6123467683792114,
      "learning_rate": 4.002671415850401e-05,
      "loss": 0.9675,
      "step": 3360
    },
    {
      "epoch": 1.0002968239833778,
      "grad_norm": 0.6109050512313843,
      "learning_rate": 3.9997031760166224e-05,
      "loss": 1.0152,
      "step": 3370
    },
    {
      "epoch": 1.0032650638171565,
      "grad_norm": 0.6630458235740662,
      "learning_rate": 3.996734936182844e-05,
      "loss": 0.9341,
      "step": 3380
    },
    {
      "epoch": 1.006233303650935,
      "grad_norm": 0.6218925714492798,
      "learning_rate": 3.993766696349065e-05,
      "loss": 0.9263,
      "step": 3390
    },
    {
      "epoch": 1.0092015434847135,
      "grad_norm": 0.6371595859527588,
      "learning_rate": 3.9907984565152865e-05,
      "loss": 0.8966,
      "step": 3400
    },
    {
      "epoch": 1.0121697833184922,
      "grad_norm": 0.6136346459388733,
      "learning_rate": 3.987830216681508e-05,
      "loss": 0.9506,
      "step": 3410
    },
    {
      "epoch": 1.0151380231522706,
      "grad_norm": 0.647017776966095,
      "learning_rate": 3.98486197684773e-05,
      "loss": 0.9079,
      "step": 3420
    },
    {
      "epoch": 1.0181062629860493,
      "grad_norm": 0.6583853960037231,
      "learning_rate": 3.981893737013951e-05,
      "loss": 0.9518,
      "step": 3430
    },
    {
      "epoch": 1.0210745028198278,
      "grad_norm": 0.6366351842880249,
      "learning_rate": 3.978925497180172e-05,
      "loss": 0.8948,
      "step": 3440
    },
    {
      "epoch": 1.0240427426536065,
      "grad_norm": 0.6189222931861877,
      "learning_rate": 3.975957257346394e-05,
      "loss": 0.8937,
      "step": 3450
    },
    {
      "epoch": 1.027010982487385,
      "grad_norm": 0.677386999130249,
      "learning_rate": 3.972989017512615e-05,
      "loss": 0.9326,
      "step": 3460
    },
    {
      "epoch": 1.0299792223211635,
      "grad_norm": 0.7184809446334839,
      "learning_rate": 3.9700207776788365e-05,
      "loss": 0.9174,
      "step": 3470
    },
    {
      "epoch": 1.0329474621549422,
      "grad_norm": 0.7060807943344116,
      "learning_rate": 3.967052537845058e-05,
      "loss": 0.9,
      "step": 3480
    },
    {
      "epoch": 1.0359157019887206,
      "grad_norm": 0.6315202713012695,
      "learning_rate": 3.96408429801128e-05,
      "loss": 0.925,
      "step": 3490
    },
    {
      "epoch": 1.0388839418224993,
      "grad_norm": 0.6721168756484985,
      "learning_rate": 3.961116058177501e-05,
      "loss": 0.911,
      "step": 3500
    },
    {
      "epoch": 1.0418521816562778,
      "grad_norm": 0.7183712720870972,
      "learning_rate": 3.958147818343722e-05,
      "loss": 0.9068,
      "step": 3510
    },
    {
      "epoch": 1.0448204214900565,
      "grad_norm": 0.6344295144081116,
      "learning_rate": 3.955179578509944e-05,
      "loss": 0.9098,
      "step": 3520
    },
    {
      "epoch": 1.047788661323835,
      "grad_norm": 0.6115924119949341,
      "learning_rate": 3.952211338676165e-05,
      "loss": 0.9236,
      "step": 3530
    },
    {
      "epoch": 1.0507569011576134,
      "grad_norm": 0.6145617961883545,
      "learning_rate": 3.9492430988423865e-05,
      "loss": 0.9374,
      "step": 3540
    },
    {
      "epoch": 1.0537251409913921,
      "grad_norm": 0.7246147990226746,
      "learning_rate": 3.946274859008608e-05,
      "loss": 0.9252,
      "step": 3550
    },
    {
      "epoch": 1.0566933808251706,
      "grad_norm": 0.7344697117805481,
      "learning_rate": 3.94330661917483e-05,
      "loss": 0.9314,
      "step": 3560
    },
    {
      "epoch": 1.0596616206589493,
      "grad_norm": 0.677960216999054,
      "learning_rate": 3.9403383793410506e-05,
      "loss": 0.9225,
      "step": 3570
    },
    {
      "epoch": 1.0626298604927278,
      "grad_norm": 0.6257244944572449,
      "learning_rate": 3.937370139507273e-05,
      "loss": 0.8904,
      "step": 3580
    },
    {
      "epoch": 1.0655981003265065,
      "grad_norm": 0.6566025614738464,
      "learning_rate": 3.934401899673494e-05,
      "loss": 0.9171,
      "step": 3590
    },
    {
      "epoch": 1.068566340160285,
      "grad_norm": 0.6468666791915894,
      "learning_rate": 3.931433659839715e-05,
      "loss": 0.9198,
      "step": 3600
    },
    {
      "epoch": 1.0715345799940634,
      "grad_norm": 0.602260410785675,
      "learning_rate": 3.9284654200059364e-05,
      "loss": 0.9406,
      "step": 3610
    },
    {
      "epoch": 1.0745028198278421,
      "grad_norm": 0.6441718339920044,
      "learning_rate": 3.925497180172158e-05,
      "loss": 0.9291,
      "step": 3620
    },
    {
      "epoch": 1.0774710596616206,
      "grad_norm": 0.6680988073348999,
      "learning_rate": 3.92252894033838e-05,
      "loss": 0.8933,
      "step": 3630
    },
    {
      "epoch": 1.0804392994953993,
      "grad_norm": 0.7151992321014404,
      "learning_rate": 3.9195607005046006e-05,
      "loss": 0.9326,
      "step": 3640
    },
    {
      "epoch": 1.0834075393291778,
      "grad_norm": 0.599533200263977,
      "learning_rate": 3.916592460670823e-05,
      "loss": 0.9201,
      "step": 3650
    },
    {
      "epoch": 1.0863757791629565,
      "grad_norm": 0.6293023228645325,
      "learning_rate": 3.913624220837044e-05,
      "loss": 0.9241,
      "step": 3660
    },
    {
      "epoch": 1.089344018996735,
      "grad_norm": 0.6524146795272827,
      "learning_rate": 3.910655981003265e-05,
      "loss": 0.9724,
      "step": 3670
    },
    {
      "epoch": 1.0923122588305134,
      "grad_norm": 0.7468433976173401,
      "learning_rate": 3.907687741169487e-05,
      "loss": 0.9371,
      "step": 3680
    },
    {
      "epoch": 1.0952804986642921,
      "grad_norm": 0.640473484992981,
      "learning_rate": 3.904719501335708e-05,
      "loss": 0.9229,
      "step": 3690
    },
    {
      "epoch": 1.0982487384980706,
      "grad_norm": 0.6132630109786987,
      "learning_rate": 3.9017512615019296e-05,
      "loss": 0.9392,
      "step": 3700
    },
    {
      "epoch": 1.1012169783318493,
      "grad_norm": 0.7036377787590027,
      "learning_rate": 3.8987830216681506e-05,
      "loss": 0.9015,
      "step": 3710
    },
    {
      "epoch": 1.1041852181656278,
      "grad_norm": 0.7262855768203735,
      "learning_rate": 3.895814781834373e-05,
      "loss": 0.9351,
      "step": 3720
    },
    {
      "epoch": 1.1071534579994065,
      "grad_norm": 0.665530264377594,
      "learning_rate": 3.892846542000594e-05,
      "loss": 0.9195,
      "step": 3730
    },
    {
      "epoch": 1.110121697833185,
      "grad_norm": 0.6587583422660828,
      "learning_rate": 3.889878302166815e-05,
      "loss": 0.898,
      "step": 3740
    },
    {
      "epoch": 1.1130899376669634,
      "grad_norm": 0.6382483243942261,
      "learning_rate": 3.886910062333037e-05,
      "loss": 0.9042,
      "step": 3750
    },
    {
      "epoch": 1.116058177500742,
      "grad_norm": 0.678569495677948,
      "learning_rate": 3.883941822499258e-05,
      "loss": 0.9695,
      "step": 3760
    },
    {
      "epoch": 1.1190264173345206,
      "grad_norm": 0.6646463871002197,
      "learning_rate": 3.8809735826654796e-05,
      "loss": 0.8859,
      "step": 3770
    },
    {
      "epoch": 1.1219946571682993,
      "grad_norm": 0.6731215119361877,
      "learning_rate": 3.878005342831701e-05,
      "loss": 0.9431,
      "step": 3780
    },
    {
      "epoch": 1.1249628970020777,
      "grad_norm": 0.6394199728965759,
      "learning_rate": 3.875037102997923e-05,
      "loss": 0.8829,
      "step": 3790
    },
    {
      "epoch": 1.1279311368358562,
      "grad_norm": 0.6833072304725647,
      "learning_rate": 3.872068863164144e-05,
      "loss": 0.8973,
      "step": 3800
    },
    {
      "epoch": 1.130899376669635,
      "grad_norm": 0.7632230520248413,
      "learning_rate": 3.869100623330365e-05,
      "loss": 0.8842,
      "step": 3810
    },
    {
      "epoch": 1.1338676165034134,
      "grad_norm": 0.684288740158081,
      "learning_rate": 3.866132383496587e-05,
      "loss": 0.9413,
      "step": 3820
    },
    {
      "epoch": 1.136835856337192,
      "grad_norm": 0.712907075881958,
      "learning_rate": 3.863164143662808e-05,
      "loss": 0.9003,
      "step": 3830
    },
    {
      "epoch": 1.1398040961709706,
      "grad_norm": 0.7701553702354431,
      "learning_rate": 3.8601959038290296e-05,
      "loss": 0.9691,
      "step": 3840
    },
    {
      "epoch": 1.1427723360047493,
      "grad_norm": 0.715511679649353,
      "learning_rate": 3.857227663995251e-05,
      "loss": 0.9501,
      "step": 3850
    },
    {
      "epoch": 1.1457405758385277,
      "grad_norm": 0.7037255167961121,
      "learning_rate": 3.854259424161473e-05,
      "loss": 0.9306,
      "step": 3860
    },
    {
      "epoch": 1.1487088156723062,
      "grad_norm": 0.7638218998908997,
      "learning_rate": 3.851291184327694e-05,
      "loss": 0.9524,
      "step": 3870
    },
    {
      "epoch": 1.151677055506085,
      "grad_norm": 0.6505693793296814,
      "learning_rate": 3.8483229444939154e-05,
      "loss": 0.9191,
      "step": 3880
    },
    {
      "epoch": 1.1546452953398634,
      "grad_norm": 0.6183308959007263,
      "learning_rate": 3.845354704660137e-05,
      "loss": 0.8744,
      "step": 3890
    },
    {
      "epoch": 1.157613535173642,
      "grad_norm": 0.6301867961883545,
      "learning_rate": 3.842386464826358e-05,
      "loss": 0.9137,
      "step": 3900
    },
    {
      "epoch": 1.1605817750074205,
      "grad_norm": 0.7025741934776306,
      "learning_rate": 3.8394182249925795e-05,
      "loss": 0.9305,
      "step": 3910
    },
    {
      "epoch": 1.1635500148411992,
      "grad_norm": 0.6408293843269348,
      "learning_rate": 3.836449985158801e-05,
      "loss": 0.9075,
      "step": 3920
    },
    {
      "epoch": 1.1665182546749777,
      "grad_norm": 0.6138466000556946,
      "learning_rate": 3.833481745325023e-05,
      "loss": 0.9357,
      "step": 3930
    },
    {
      "epoch": 1.1694864945087562,
      "grad_norm": 0.6550092101097107,
      "learning_rate": 3.830513505491244e-05,
      "loss": 0.9528,
      "step": 3940
    },
    {
      "epoch": 1.1724547343425349,
      "grad_norm": 0.6522756218910217,
      "learning_rate": 3.827545265657465e-05,
      "loss": 0.9317,
      "step": 3950
    },
    {
      "epoch": 1.1754229741763134,
      "grad_norm": 0.7218632102012634,
      "learning_rate": 3.824577025823687e-05,
      "loss": 0.9292,
      "step": 3960
    },
    {
      "epoch": 1.178391214010092,
      "grad_norm": 0.6621205806732178,
      "learning_rate": 3.821608785989908e-05,
      "loss": 0.9204,
      "step": 3970
    },
    {
      "epoch": 1.1813594538438705,
      "grad_norm": 0.6475970149040222,
      "learning_rate": 3.8186405461561295e-05,
      "loss": 0.9097,
      "step": 3980
    },
    {
      "epoch": 1.1843276936776492,
      "grad_norm": 0.6492740511894226,
      "learning_rate": 3.815672306322351e-05,
      "loss": 0.9261,
      "step": 3990
    },
    {
      "epoch": 1.1872959335114277,
      "grad_norm": 0.6680456399917603,
      "learning_rate": 3.812704066488573e-05,
      "loss": 0.929,
      "step": 4000
    },
    {
      "epoch": 1.1902641733452062,
      "grad_norm": 0.7199456095695496,
      "learning_rate": 3.809735826654794e-05,
      "loss": 0.921,
      "step": 4010
    },
    {
      "epoch": 1.1932324131789849,
      "grad_norm": 0.7005553245544434,
      "learning_rate": 3.806767586821015e-05,
      "loss": 0.9489,
      "step": 4020
    },
    {
      "epoch": 1.1962006530127633,
      "grad_norm": 0.735540509223938,
      "learning_rate": 3.803799346987237e-05,
      "loss": 0.9206,
      "step": 4030
    },
    {
      "epoch": 1.199168892846542,
      "grad_norm": 0.7136026620864868,
      "learning_rate": 3.800831107153458e-05,
      "loss": 0.9413,
      "step": 4040
    },
    {
      "epoch": 1.2021371326803205,
      "grad_norm": 0.6640150547027588,
      "learning_rate": 3.7978628673196795e-05,
      "loss": 0.9293,
      "step": 4050
    },
    {
      "epoch": 1.2051053725140992,
      "grad_norm": 0.676105797290802,
      "learning_rate": 3.794894627485901e-05,
      "loss": 0.9007,
      "step": 4060
    },
    {
      "epoch": 1.2080736123478777,
      "grad_norm": 0.6336154937744141,
      "learning_rate": 3.791926387652123e-05,
      "loss": 0.9198,
      "step": 4070
    },
    {
      "epoch": 1.2110418521816562,
      "grad_norm": 0.6762709021568298,
      "learning_rate": 3.7889581478183437e-05,
      "loss": 0.9211,
      "step": 4080
    },
    {
      "epoch": 1.2140100920154349,
      "grad_norm": 0.6613302826881409,
      "learning_rate": 3.785989907984565e-05,
      "loss": 0.8855,
      "step": 4090
    },
    {
      "epoch": 1.2169783318492133,
      "grad_norm": 0.7078067660331726,
      "learning_rate": 3.783021668150787e-05,
      "loss": 0.9235,
      "step": 4100
    },
    {
      "epoch": 1.219946571682992,
      "grad_norm": 0.7373542189598083,
      "learning_rate": 3.780053428317008e-05,
      "loss": 0.893,
      "step": 4110
    },
    {
      "epoch": 1.2229148115167705,
      "grad_norm": 0.6745258569717407,
      "learning_rate": 3.77708518848323e-05,
      "loss": 0.9423,
      "step": 4120
    },
    {
      "epoch": 1.2258830513505492,
      "grad_norm": 0.6126631498336792,
      "learning_rate": 3.774116948649451e-05,
      "loss": 0.9201,
      "step": 4130
    },
    {
      "epoch": 1.2288512911843277,
      "grad_norm": 0.6054243445396423,
      "learning_rate": 3.771148708815673e-05,
      "loss": 0.9148,
      "step": 4140
    },
    {
      "epoch": 1.2318195310181062,
      "grad_norm": 0.6797349452972412,
      "learning_rate": 3.7681804689818936e-05,
      "loss": 0.9171,
      "step": 4150
    },
    {
      "epoch": 1.2347877708518848,
      "grad_norm": 0.6671416163444519,
      "learning_rate": 3.765212229148115e-05,
      "loss": 0.9257,
      "step": 4160
    },
    {
      "epoch": 1.2377560106856633,
      "grad_norm": 0.7235540151596069,
      "learning_rate": 3.762243989314337e-05,
      "loss": 0.9254,
      "step": 4170
    },
    {
      "epoch": 1.240724250519442,
      "grad_norm": 0.7316147089004517,
      "learning_rate": 3.759275749480558e-05,
      "loss": 0.9349,
      "step": 4180
    },
    {
      "epoch": 1.2436924903532205,
      "grad_norm": 0.6851880550384521,
      "learning_rate": 3.75630750964678e-05,
      "loss": 0.9329,
      "step": 4190
    },
    {
      "epoch": 1.2466607301869992,
      "grad_norm": 0.6671837568283081,
      "learning_rate": 3.753339269813001e-05,
      "loss": 0.8949,
      "step": 4200
    },
    {
      "epoch": 1.2496289700207777,
      "grad_norm": 0.8188835978507996,
      "learning_rate": 3.7503710299792226e-05,
      "loss": 0.8751,
      "step": 4210
    },
    {
      "epoch": 1.2525972098545561,
      "grad_norm": 0.635193943977356,
      "learning_rate": 3.747402790145444e-05,
      "loss": 0.933,
      "step": 4220
    },
    {
      "epoch": 1.2555654496883348,
      "grad_norm": 0.7244166731834412,
      "learning_rate": 3.744434550311665e-05,
      "loss": 0.9009,
      "step": 4230
    },
    {
      "epoch": 1.2585336895221133,
      "grad_norm": 0.6825953722000122,
      "learning_rate": 3.741466310477887e-05,
      "loss": 0.9444,
      "step": 4240
    },
    {
      "epoch": 1.261501929355892,
      "grad_norm": 0.7705867290496826,
      "learning_rate": 3.738498070644108e-05,
      "loss": 0.9307,
      "step": 4250
    },
    {
      "epoch": 1.2644701691896705,
      "grad_norm": 0.6319135427474976,
      "learning_rate": 3.73552983081033e-05,
      "loss": 0.9332,
      "step": 4260
    },
    {
      "epoch": 1.2674384090234492,
      "grad_norm": 0.64055335521698,
      "learning_rate": 3.732561590976551e-05,
      "loss": 0.9062,
      "step": 4270
    },
    {
      "epoch": 1.2704066488572276,
      "grad_norm": 0.7409882545471191,
      "learning_rate": 3.7295933511427726e-05,
      "loss": 0.9713,
      "step": 4280
    },
    {
      "epoch": 1.2733748886910061,
      "grad_norm": 0.6357461214065552,
      "learning_rate": 3.726625111308994e-05,
      "loss": 0.9196,
      "step": 4290
    },
    {
      "epoch": 1.2763431285247848,
      "grad_norm": 0.6302955150604248,
      "learning_rate": 3.723656871475215e-05,
      "loss": 0.8913,
      "step": 4300
    },
    {
      "epoch": 1.2793113683585633,
      "grad_norm": 0.6616325378417969,
      "learning_rate": 3.720688631641437e-05,
      "loss": 0.9097,
      "step": 4310
    },
    {
      "epoch": 1.282279608192342,
      "grad_norm": 0.8014053106307983,
      "learning_rate": 3.7177203918076584e-05,
      "loss": 0.9446,
      "step": 4320
    },
    {
      "epoch": 1.2852478480261205,
      "grad_norm": 0.8350517153739929,
      "learning_rate": 3.71475215197388e-05,
      "loss": 0.9156,
      "step": 4330
    },
    {
      "epoch": 1.2882160878598992,
      "grad_norm": 0.6596609950065613,
      "learning_rate": 3.711783912140101e-05,
      "loss": 0.9923,
      "step": 4340
    },
    {
      "epoch": 1.2911843276936776,
      "grad_norm": 0.6925686001777649,
      "learning_rate": 3.7088156723063226e-05,
      "loss": 0.9401,
      "step": 4350
    },
    {
      "epoch": 1.294152567527456,
      "grad_norm": 0.6993871927261353,
      "learning_rate": 3.705847432472544e-05,
      "loss": 0.9515,
      "step": 4360
    },
    {
      "epoch": 1.2971208073612348,
      "grad_norm": 0.8117514848709106,
      "learning_rate": 3.702879192638765e-05,
      "loss": 0.8988,
      "step": 4370
    },
    {
      "epoch": 1.3000890471950133,
      "grad_norm": 0.6895804405212402,
      "learning_rate": 3.699910952804987e-05,
      "loss": 0.9153,
      "step": 4380
    },
    {
      "epoch": 1.303057287028792,
      "grad_norm": 0.6821000576019287,
      "learning_rate": 3.6969427129712084e-05,
      "loss": 0.8968,
      "step": 4390
    },
    {
      "epoch": 1.3060255268625705,
      "grad_norm": 0.6583243608474731,
      "learning_rate": 3.69397447313743e-05,
      "loss": 0.9378,
      "step": 4400
    },
    {
      "epoch": 1.3089937666963491,
      "grad_norm": 0.7524015307426453,
      "learning_rate": 3.691006233303651e-05,
      "loss": 0.9465,
      "step": 4410
    },
    {
      "epoch": 1.3119620065301276,
      "grad_norm": 0.6931015849113464,
      "learning_rate": 3.6880379934698726e-05,
      "loss": 0.901,
      "step": 4420
    },
    {
      "epoch": 1.314930246363906,
      "grad_norm": 0.7283458113670349,
      "learning_rate": 3.685069753636094e-05,
      "loss": 0.8931,
      "step": 4430
    },
    {
      "epoch": 1.3178984861976848,
      "grad_norm": 0.6719711422920227,
      "learning_rate": 3.682101513802315e-05,
      "loss": 0.9117,
      "step": 4440
    },
    {
      "epoch": 1.3208667260314633,
      "grad_norm": 0.794035017490387,
      "learning_rate": 3.679133273968537e-05,
      "loss": 0.9064,
      "step": 4450
    },
    {
      "epoch": 1.323834965865242,
      "grad_norm": 0.8168098330497742,
      "learning_rate": 3.6761650341347583e-05,
      "loss": 0.9362,
      "step": 4460
    },
    {
      "epoch": 1.3268032056990204,
      "grad_norm": 0.7201861143112183,
      "learning_rate": 3.67319679430098e-05,
      "loss": 0.9218,
      "step": 4470
    },
    {
      "epoch": 1.3297714455327991,
      "grad_norm": 0.7084200978279114,
      "learning_rate": 3.670228554467201e-05,
      "loss": 0.9243,
      "step": 4480
    },
    {
      "epoch": 1.3327396853665776,
      "grad_norm": 0.6903102993965149,
      "learning_rate": 3.6672603146334225e-05,
      "loss": 0.9403,
      "step": 4490
    },
    {
      "epoch": 1.335707925200356,
      "grad_norm": 0.6698452830314636,
      "learning_rate": 3.664292074799644e-05,
      "loss": 0.9108,
      "step": 4500
    },
    {
      "epoch": 1.3386761650341348,
      "grad_norm": 0.6291380524635315,
      "learning_rate": 3.661323834965865e-05,
      "loss": 0.9283,
      "step": 4510
    },
    {
      "epoch": 1.3416444048679133,
      "grad_norm": 0.7101170420646667,
      "learning_rate": 3.658355595132087e-05,
      "loss": 0.9198,
      "step": 4520
    },
    {
      "epoch": 1.344612644701692,
      "grad_norm": 0.6618587970733643,
      "learning_rate": 3.655387355298308e-05,
      "loss": 0.8878,
      "step": 4530
    },
    {
      "epoch": 1.3475808845354704,
      "grad_norm": 0.7191241383552551,
      "learning_rate": 3.65241911546453e-05,
      "loss": 0.882,
      "step": 4540
    },
    {
      "epoch": 1.3505491243692491,
      "grad_norm": 0.7250847220420837,
      "learning_rate": 3.649450875630751e-05,
      "loss": 0.8646,
      "step": 4550
    },
    {
      "epoch": 1.3535173642030276,
      "grad_norm": 0.7301775813102722,
      "learning_rate": 3.6464826357969725e-05,
      "loss": 0.8714,
      "step": 4560
    },
    {
      "epoch": 1.356485604036806,
      "grad_norm": 0.7399090528488159,
      "learning_rate": 3.643514395963194e-05,
      "loss": 0.9146,
      "step": 4570
    },
    {
      "epoch": 1.3594538438705848,
      "grad_norm": 0.7338899970054626,
      "learning_rate": 3.640546156129415e-05,
      "loss": 0.9218,
      "step": 4580
    },
    {
      "epoch": 1.3624220837043632,
      "grad_norm": 0.6969752311706543,
      "learning_rate": 3.637577916295637e-05,
      "loss": 0.9117,
      "step": 4590
    },
    {
      "epoch": 1.365390323538142,
      "grad_norm": 0.7742125391960144,
      "learning_rate": 3.634609676461858e-05,
      "loss": 0.9445,
      "step": 4600
    },
    {
      "epoch": 1.3683585633719204,
      "grad_norm": 0.837775468826294,
      "learning_rate": 3.63164143662808e-05,
      "loss": 0.9356,
      "step": 4610
    },
    {
      "epoch": 1.371326803205699,
      "grad_norm": 0.7069649696350098,
      "learning_rate": 3.628673196794301e-05,
      "loss": 0.8942,
      "step": 4620
    },
    {
      "epoch": 1.3742950430394776,
      "grad_norm": 0.6960747241973877,
      "learning_rate": 3.625704956960523e-05,
      "loss": 0.9484,
      "step": 4630
    },
    {
      "epoch": 1.377263282873256,
      "grad_norm": 0.6509305238723755,
      "learning_rate": 3.622736717126744e-05,
      "loss": 0.9079,
      "step": 4640
    },
    {
      "epoch": 1.3802315227070348,
      "grad_norm": 0.7066072821617126,
      "learning_rate": 3.619768477292965e-05,
      "loss": 0.9112,
      "step": 4650
    },
    {
      "epoch": 1.3831997625408132,
      "grad_norm": 0.7418903112411499,
      "learning_rate": 3.6168002374591866e-05,
      "loss": 0.9262,
      "step": 4660
    },
    {
      "epoch": 1.386168002374592,
      "grad_norm": 0.6625681519508362,
      "learning_rate": 3.613831997625408e-05,
      "loss": 0.9116,
      "step": 4670
    },
    {
      "epoch": 1.3891362422083704,
      "grad_norm": 0.766903281211853,
      "learning_rate": 3.61086375779163e-05,
      "loss": 0.9074,
      "step": 4680
    },
    {
      "epoch": 1.392104482042149,
      "grad_norm": 0.617678165435791,
      "learning_rate": 3.607895517957851e-05,
      "loss": 0.9324,
      "step": 4690
    },
    {
      "epoch": 1.3950727218759276,
      "grad_norm": 0.6659789085388184,
      "learning_rate": 3.604927278124073e-05,
      "loss": 0.9099,
      "step": 4700
    },
    {
      "epoch": 1.398040961709706,
      "grad_norm": 0.7493019700050354,
      "learning_rate": 3.601959038290294e-05,
      "loss": 0.9057,
      "step": 4710
    },
    {
      "epoch": 1.4010092015434847,
      "grad_norm": 0.6639522314071655,
      "learning_rate": 3.598990798456515e-05,
      "loss": 0.8924,
      "step": 4720
    },
    {
      "epoch": 1.4039774413772632,
      "grad_norm": 0.7125399708747864,
      "learning_rate": 3.596022558622737e-05,
      "loss": 0.9364,
      "step": 4730
    },
    {
      "epoch": 1.406945681211042,
      "grad_norm": 0.6880870461463928,
      "learning_rate": 3.593054318788958e-05,
      "loss": 0.94,
      "step": 4740
    },
    {
      "epoch": 1.4099139210448204,
      "grad_norm": 0.765393078327179,
      "learning_rate": 3.59008607895518e-05,
      "loss": 0.9592,
      "step": 4750
    },
    {
      "epoch": 1.412882160878599,
      "grad_norm": 0.7591105103492737,
      "learning_rate": 3.587117839121401e-05,
      "loss": 0.8971,
      "step": 4760
    },
    {
      "epoch": 1.4158504007123776,
      "grad_norm": 0.6672634482383728,
      "learning_rate": 3.584149599287623e-05,
      "loss": 0.8898,
      "step": 4770
    },
    {
      "epoch": 1.418818640546156,
      "grad_norm": 0.6270623803138733,
      "learning_rate": 3.581181359453844e-05,
      "loss": 0.8982,
      "step": 4780
    },
    {
      "epoch": 1.4217868803799347,
      "grad_norm": 0.7162853479385376,
      "learning_rate": 3.578213119620065e-05,
      "loss": 0.9368,
      "step": 4790
    },
    {
      "epoch": 1.4247551202137132,
      "grad_norm": 0.7325291037559509,
      "learning_rate": 3.575244879786287e-05,
      "loss": 0.904,
      "step": 4800
    },
    {
      "epoch": 1.427723360047492,
      "grad_norm": 0.7608901858329773,
      "learning_rate": 3.572276639952508e-05,
      "loss": 0.8736,
      "step": 4810
    },
    {
      "epoch": 1.4306915998812704,
      "grad_norm": 0.7135578989982605,
      "learning_rate": 3.56930840011873e-05,
      "loss": 0.926,
      "step": 4820
    },
    {
      "epoch": 1.433659839715049,
      "grad_norm": 0.7218388915061951,
      "learning_rate": 3.5663401602849514e-05,
      "loss": 0.961,
      "step": 4830
    },
    {
      "epoch": 1.4366280795488275,
      "grad_norm": 0.7087059020996094,
      "learning_rate": 3.563371920451173e-05,
      "loss": 0.9115,
      "step": 4840
    },
    {
      "epoch": 1.439596319382606,
      "grad_norm": 0.7503490447998047,
      "learning_rate": 3.560403680617394e-05,
      "loss": 0.9211,
      "step": 4850
    },
    {
      "epoch": 1.4425645592163847,
      "grad_norm": 0.7498166561126709,
      "learning_rate": 3.557435440783615e-05,
      "loss": 0.9059,
      "step": 4860
    },
    {
      "epoch": 1.4455327990501632,
      "grad_norm": 0.7072891592979431,
      "learning_rate": 3.554467200949837e-05,
      "loss": 0.9282,
      "step": 4870
    },
    {
      "epoch": 1.4485010388839419,
      "grad_norm": 0.6605234146118164,
      "learning_rate": 3.551498961116058e-05,
      "loss": 0.9261,
      "step": 4880
    },
    {
      "epoch": 1.4514692787177204,
      "grad_norm": 0.7122581005096436,
      "learning_rate": 3.54853072128228e-05,
      "loss": 0.9446,
      "step": 4890
    },
    {
      "epoch": 1.454437518551499,
      "grad_norm": 0.6746944189071655,
      "learning_rate": 3.5455624814485014e-05,
      "loss": 0.9515,
      "step": 4900
    },
    {
      "epoch": 1.4574057583852775,
      "grad_norm": 0.6817790865898132,
      "learning_rate": 3.542594241614723e-05,
      "loss": 0.9416,
      "step": 4910
    },
    {
      "epoch": 1.460373998219056,
      "grad_norm": 0.6961323022842407,
      "learning_rate": 3.539626001780944e-05,
      "loss": 0.9217,
      "step": 4920
    },
    {
      "epoch": 1.4633422380528347,
      "grad_norm": 0.6741704344749451,
      "learning_rate": 3.5366577619471656e-05,
      "loss": 0.9377,
      "step": 4930
    },
    {
      "epoch": 1.4663104778866132,
      "grad_norm": 0.6607795357704163,
      "learning_rate": 3.533689522113387e-05,
      "loss": 0.8585,
      "step": 4940
    },
    {
      "epoch": 1.4692787177203919,
      "grad_norm": 0.6922871470451355,
      "learning_rate": 3.530721282279608e-05,
      "loss": 0.9089,
      "step": 4950
    },
    {
      "epoch": 1.4722469575541703,
      "grad_norm": 0.6661325097084045,
      "learning_rate": 3.52775304244583e-05,
      "loss": 0.9016,
      "step": 4960
    },
    {
      "epoch": 1.475215197387949,
      "grad_norm": 0.7098068594932556,
      "learning_rate": 3.5247848026120514e-05,
      "loss": 0.9075,
      "step": 4970
    },
    {
      "epoch": 1.4781834372217275,
      "grad_norm": 0.7736371755599976,
      "learning_rate": 3.521816562778273e-05,
      "loss": 0.8992,
      "step": 4980
    },
    {
      "epoch": 1.481151677055506,
      "grad_norm": 0.6673690676689148,
      "learning_rate": 3.518848322944494e-05,
      "loss": 0.9198,
      "step": 4990
    },
    {
      "epoch": 1.4841199168892847,
      "grad_norm": 0.6655102372169495,
      "learning_rate": 3.5158800831107155e-05,
      "loss": 0.8794,
      "step": 5000
    },
    {
      "epoch": 1.4870881567230632,
      "grad_norm": 0.7615334987640381,
      "learning_rate": 3.512911843276937e-05,
      "loss": 0.9617,
      "step": 5010
    },
    {
      "epoch": 1.4900563965568419,
      "grad_norm": 0.7400135397911072,
      "learning_rate": 3.509943603443158e-05,
      "loss": 0.9142,
      "step": 5020
    },
    {
      "epoch": 1.4930246363906203,
      "grad_norm": 0.6787492632865906,
      "learning_rate": 3.50697536360938e-05,
      "loss": 0.9513,
      "step": 5030
    },
    {
      "epoch": 1.495992876224399,
      "grad_norm": 0.6557864546775818,
      "learning_rate": 3.504007123775601e-05,
      "loss": 0.9381,
      "step": 5040
    },
    {
      "epoch": 1.4989611160581775,
      "grad_norm": 0.6889845728874207,
      "learning_rate": 3.501038883941823e-05,
      "loss": 0.8919,
      "step": 5050
    },
    {
      "epoch": 1.501929355891956,
      "grad_norm": 0.6667259335517883,
      "learning_rate": 3.498070644108044e-05,
      "loss": 0.9388,
      "step": 5060
    },
    {
      "epoch": 1.5048975957257347,
      "grad_norm": 0.7543666958808899,
      "learning_rate": 3.4951024042742655e-05,
      "loss": 0.9045,
      "step": 5070
    },
    {
      "epoch": 1.5078658355595134,
      "grad_norm": 0.7088416218757629,
      "learning_rate": 3.492134164440487e-05,
      "loss": 0.9122,
      "step": 5080
    },
    {
      "epoch": 1.5108340753932916,
      "grad_norm": 0.6750200390815735,
      "learning_rate": 3.489165924606708e-05,
      "loss": 0.9358,
      "step": 5090
    },
    {
      "epoch": 1.5138023152270703,
      "grad_norm": 0.6742722988128662,
      "learning_rate": 3.48619768477293e-05,
      "loss": 0.9275,
      "step": 5100
    },
    {
      "epoch": 1.516770555060849,
      "grad_norm": 0.6562049984931946,
      "learning_rate": 3.483229444939151e-05,
      "loss": 0.9,
      "step": 5110
    },
    {
      "epoch": 1.5197387948946275,
      "grad_norm": 0.6847245097160339,
      "learning_rate": 3.480261205105373e-05,
      "loss": 0.897,
      "step": 5120
    },
    {
      "epoch": 1.522707034728406,
      "grad_norm": 0.7098760604858398,
      "learning_rate": 3.477292965271594e-05,
      "loss": 0.9489,
      "step": 5130
    },
    {
      "epoch": 1.5256752745621847,
      "grad_norm": 0.6374083161354065,
      "learning_rate": 3.4743247254378155e-05,
      "loss": 0.895,
      "step": 5140
    },
    {
      "epoch": 1.5286435143959634,
      "grad_norm": 0.7240116596221924,
      "learning_rate": 3.471356485604037e-05,
      "loss": 0.8889,
      "step": 5150
    },
    {
      "epoch": 1.5316117542297416,
      "grad_norm": 0.676428496837616,
      "learning_rate": 3.468388245770258e-05,
      "loss": 0.9342,
      "step": 5160
    },
    {
      "epoch": 1.5345799940635203,
      "grad_norm": 0.6966356635093689,
      "learning_rate": 3.4654200059364796e-05,
      "loss": 0.895,
      "step": 5170
    },
    {
      "epoch": 1.537548233897299,
      "grad_norm": 0.6685094237327576,
      "learning_rate": 3.462451766102701e-05,
      "loss": 0.8986,
      "step": 5180
    },
    {
      "epoch": 1.5405164737310775,
      "grad_norm": 0.7533665299415588,
      "learning_rate": 3.459483526268923e-05,
      "loss": 0.9388,
      "step": 5190
    },
    {
      "epoch": 1.543484713564856,
      "grad_norm": 0.680923581123352,
      "learning_rate": 3.456515286435144e-05,
      "loss": 0.9199,
      "step": 5200
    },
    {
      "epoch": 1.5464529533986346,
      "grad_norm": 0.7609050869941711,
      "learning_rate": 3.4535470466013654e-05,
      "loss": 0.8998,
      "step": 5210
    },
    {
      "epoch": 1.5494211932324133,
      "grad_norm": 0.6809772253036499,
      "learning_rate": 3.450578806767587e-05,
      "loss": 0.9037,
      "step": 5220
    },
    {
      "epoch": 1.5523894330661916,
      "grad_norm": 0.717606246471405,
      "learning_rate": 3.447610566933808e-05,
      "loss": 0.9231,
      "step": 5230
    },
    {
      "epoch": 1.5553576728999703,
      "grad_norm": 0.7689083218574524,
      "learning_rate": 3.44464232710003e-05,
      "loss": 0.9056,
      "step": 5240
    },
    {
      "epoch": 1.558325912733749,
      "grad_norm": 0.6759158372879028,
      "learning_rate": 3.441674087266251e-05,
      "loss": 0.9278,
      "step": 5250
    },
    {
      "epoch": 1.5612941525675275,
      "grad_norm": 0.6883628964424133,
      "learning_rate": 3.438705847432473e-05,
      "loss": 0.9289,
      "step": 5260
    },
    {
      "epoch": 1.564262392401306,
      "grad_norm": 0.7132298350334167,
      "learning_rate": 3.435737607598694e-05,
      "loss": 0.8762,
      "step": 5270
    },
    {
      "epoch": 1.5672306322350846,
      "grad_norm": 0.6003124713897705,
      "learning_rate": 3.4327693677649154e-05,
      "loss": 0.9226,
      "step": 5280
    },
    {
      "epoch": 1.5701988720688633,
      "grad_norm": 0.7471172213554382,
      "learning_rate": 3.429801127931137e-05,
      "loss": 0.9123,
      "step": 5290
    },
    {
      "epoch": 1.5731671119026416,
      "grad_norm": 0.7321667075157166,
      "learning_rate": 3.426832888097358e-05,
      "loss": 0.8938,
      "step": 5300
    },
    {
      "epoch": 1.5761353517364203,
      "grad_norm": 0.7397316098213196,
      "learning_rate": 3.42386464826358e-05,
      "loss": 0.8929,
      "step": 5310
    },
    {
      "epoch": 1.579103591570199,
      "grad_norm": 0.7189614176750183,
      "learning_rate": 3.420896408429801e-05,
      "loss": 0.9306,
      "step": 5320
    },
    {
      "epoch": 1.5820718314039774,
      "grad_norm": 0.6808568835258484,
      "learning_rate": 3.417928168596023e-05,
      "loss": 0.928,
      "step": 5330
    },
    {
      "epoch": 1.585040071237756,
      "grad_norm": 0.706499457359314,
      "learning_rate": 3.4149599287622444e-05,
      "loss": 0.9032,
      "step": 5340
    },
    {
      "epoch": 1.5880083110715346,
      "grad_norm": 0.6810677647590637,
      "learning_rate": 3.4119916889284654e-05,
      "loss": 0.9114,
      "step": 5350
    },
    {
      "epoch": 1.5909765509053133,
      "grad_norm": 0.8776708841323853,
      "learning_rate": 3.409023449094687e-05,
      "loss": 0.9452,
      "step": 5360
    },
    {
      "epoch": 1.5939447907390916,
      "grad_norm": 0.772865891456604,
      "learning_rate": 3.406055209260908e-05,
      "loss": 0.9165,
      "step": 5370
    },
    {
      "epoch": 1.5969130305728703,
      "grad_norm": 0.6921769976615906,
      "learning_rate": 3.40308696942713e-05,
      "loss": 0.9007,
      "step": 5380
    },
    {
      "epoch": 1.599881270406649,
      "grad_norm": 0.7539716362953186,
      "learning_rate": 3.400118729593351e-05,
      "loss": 0.9055,
      "step": 5390
    },
    {
      "epoch": 1.6028495102404274,
      "grad_norm": 0.7750148773193359,
      "learning_rate": 3.397150489759573e-05,
      "loss": 0.9189,
      "step": 5400
    },
    {
      "epoch": 1.605817750074206,
      "grad_norm": 0.7431687116622925,
      "learning_rate": 3.3941822499257944e-05,
      "loss": 0.9346,
      "step": 5410
    },
    {
      "epoch": 1.6087859899079846,
      "grad_norm": 0.6506355404853821,
      "learning_rate": 3.391214010092015e-05,
      "loss": 0.9065,
      "step": 5420
    },
    {
      "epoch": 1.6117542297417633,
      "grad_norm": 0.7202748656272888,
      "learning_rate": 3.388245770258237e-05,
      "loss": 0.914,
      "step": 5430
    },
    {
      "epoch": 1.6147224695755416,
      "grad_norm": 0.7080637216567993,
      "learning_rate": 3.3852775304244586e-05,
      "loss": 0.9121,
      "step": 5440
    },
    {
      "epoch": 1.6176907094093202,
      "grad_norm": 0.694549024105072,
      "learning_rate": 3.38230929059068e-05,
      "loss": 0.886,
      "step": 5450
    },
    {
      "epoch": 1.620658949243099,
      "grad_norm": 0.6965795755386353,
      "learning_rate": 3.379341050756901e-05,
      "loss": 0.9247,
      "step": 5460
    },
    {
      "epoch": 1.6236271890768774,
      "grad_norm": 0.7468060255050659,
      "learning_rate": 3.376372810923123e-05,
      "loss": 0.8834,
      "step": 5470
    },
    {
      "epoch": 1.626595428910656,
      "grad_norm": 0.6878003478050232,
      "learning_rate": 3.3734045710893444e-05,
      "loss": 0.9179,
      "step": 5480
    },
    {
      "epoch": 1.6295636687444346,
      "grad_norm": 0.7024316787719727,
      "learning_rate": 3.370436331255565e-05,
      "loss": 0.9037,
      "step": 5490
    },
    {
      "epoch": 1.632531908578213,
      "grad_norm": 0.6189635396003723,
      "learning_rate": 3.367468091421787e-05,
      "loss": 0.9578,
      "step": 5500
    },
    {
      "epoch": 1.6355001484119915,
      "grad_norm": 0.6469042897224426,
      "learning_rate": 3.3644998515880085e-05,
      "loss": 0.8899,
      "step": 5510
    },
    {
      "epoch": 1.6384683882457702,
      "grad_norm": 0.6131319403648376,
      "learning_rate": 3.36153161175423e-05,
      "loss": 0.8727,
      "step": 5520
    },
    {
      "epoch": 1.641436628079549,
      "grad_norm": 0.6959238052368164,
      "learning_rate": 3.358563371920451e-05,
      "loss": 0.9011,
      "step": 5530
    },
    {
      "epoch": 1.6444048679133274,
      "grad_norm": 0.6863197088241577,
      "learning_rate": 3.355595132086673e-05,
      "loss": 0.9627,
      "step": 5540
    },
    {
      "epoch": 1.6473731077471059,
      "grad_norm": 0.6819227933883667,
      "learning_rate": 3.352626892252894e-05,
      "loss": 0.9205,
      "step": 5550
    },
    {
      "epoch": 1.6503413475808846,
      "grad_norm": 0.7224652171134949,
      "learning_rate": 3.349658652419115e-05,
      "loss": 0.9183,
      "step": 5560
    },
    {
      "epoch": 1.653309587414663,
      "grad_norm": 0.7046286463737488,
      "learning_rate": 3.346690412585337e-05,
      "loss": 0.8784,
      "step": 5570
    },
    {
      "epoch": 1.6562778272484415,
      "grad_norm": 0.6479465961456299,
      "learning_rate": 3.3437221727515585e-05,
      "loss": 0.8717,
      "step": 5580
    },
    {
      "epoch": 1.6592460670822202,
      "grad_norm": 0.7227984070777893,
      "learning_rate": 3.34075393291778e-05,
      "loss": 0.9064,
      "step": 5590
    },
    {
      "epoch": 1.662214306915999,
      "grad_norm": 0.729101836681366,
      "learning_rate": 3.337785693084001e-05,
      "loss": 0.9287,
      "step": 5600
    },
    {
      "epoch": 1.6651825467497774,
      "grad_norm": 0.7223145961761475,
      "learning_rate": 3.334817453250223e-05,
      "loss": 0.9138,
      "step": 5610
    },
    {
      "epoch": 1.6681507865835559,
      "grad_norm": 0.8971875905990601,
      "learning_rate": 3.331849213416444e-05,
      "loss": 0.903,
      "step": 5620
    },
    {
      "epoch": 1.6711190264173346,
      "grad_norm": 0.6369121074676514,
      "learning_rate": 3.328880973582665e-05,
      "loss": 0.9082,
      "step": 5630
    },
    {
      "epoch": 1.674087266251113,
      "grad_norm": 0.689054548740387,
      "learning_rate": 3.325912733748887e-05,
      "loss": 0.9103,
      "step": 5640
    },
    {
      "epoch": 1.6770555060848915,
      "grad_norm": 0.7047765254974365,
      "learning_rate": 3.3229444939151085e-05,
      "loss": 0.9451,
      "step": 5650
    },
    {
      "epoch": 1.6800237459186702,
      "grad_norm": 0.7887316346168518,
      "learning_rate": 3.31997625408133e-05,
      "loss": 0.9225,
      "step": 5660
    },
    {
      "epoch": 1.682991985752449,
      "grad_norm": 0.7298426032066345,
      "learning_rate": 3.317008014247551e-05,
      "loss": 0.8988,
      "step": 5670
    },
    {
      "epoch": 1.6859602255862274,
      "grad_norm": 0.7398642897605896,
      "learning_rate": 3.314039774413773e-05,
      "loss": 0.9101,
      "step": 5680
    },
    {
      "epoch": 1.6889284654200059,
      "grad_norm": 0.6913554072380066,
      "learning_rate": 3.311071534579994e-05,
      "loss": 0.9213,
      "step": 5690
    },
    {
      "epoch": 1.6918967052537845,
      "grad_norm": 0.7257855534553528,
      "learning_rate": 3.308103294746215e-05,
      "loss": 0.9297,
      "step": 5700
    },
    {
      "epoch": 1.694864945087563,
      "grad_norm": 0.7400197982788086,
      "learning_rate": 3.305135054912437e-05,
      "loss": 0.8795,
      "step": 5710
    },
    {
      "epoch": 1.6978331849213415,
      "grad_norm": 0.7153133749961853,
      "learning_rate": 3.3021668150786584e-05,
      "loss": 0.8905,
      "step": 5720
    },
    {
      "epoch": 1.7008014247551202,
      "grad_norm": 0.6668074727058411,
      "learning_rate": 3.29919857524488e-05,
      "loss": 0.9542,
      "step": 5730
    },
    {
      "epoch": 1.703769664588899,
      "grad_norm": 0.7409269213676453,
      "learning_rate": 3.296230335411101e-05,
      "loss": 0.902,
      "step": 5740
    },
    {
      "epoch": 1.7067379044226774,
      "grad_norm": 0.6369599103927612,
      "learning_rate": 3.293262095577323e-05,
      "loss": 0.9278,
      "step": 5750
    },
    {
      "epoch": 1.7097061442564558,
      "grad_norm": 0.74519944190979,
      "learning_rate": 3.290293855743544e-05,
      "loss": 0.9486,
      "step": 5760
    },
    {
      "epoch": 1.7126743840902345,
      "grad_norm": 0.7439436912536621,
      "learning_rate": 3.287325615909765e-05,
      "loss": 0.9283,
      "step": 5770
    },
    {
      "epoch": 1.715642623924013,
      "grad_norm": 0.6844840049743652,
      "learning_rate": 3.2843573760759875e-05,
      "loss": 0.9643,
      "step": 5780
    },
    {
      "epoch": 1.7186108637577915,
      "grad_norm": 0.6939666271209717,
      "learning_rate": 3.2813891362422084e-05,
      "loss": 0.9393,
      "step": 5790
    },
    {
      "epoch": 1.7215791035915702,
      "grad_norm": 0.739400327205658,
      "learning_rate": 3.27842089640843e-05,
      "loss": 0.8803,
      "step": 5800
    },
    {
      "epoch": 1.7245473434253489,
      "grad_norm": 0.7088194489479065,
      "learning_rate": 3.275452656574651e-05,
      "loss": 0.8939,
      "step": 5810
    },
    {
      "epoch": 1.7275155832591274,
      "grad_norm": 0.7284060120582581,
      "learning_rate": 3.272484416740873e-05,
      "loss": 0.916,
      "step": 5820
    },
    {
      "epoch": 1.7304838230929058,
      "grad_norm": 0.6898395419120789,
      "learning_rate": 3.269516176907094e-05,
      "loss": 0.9158,
      "step": 5830
    },
    {
      "epoch": 1.7334520629266845,
      "grad_norm": 0.7644175887107849,
      "learning_rate": 3.266547937073315e-05,
      "loss": 0.8764,
      "step": 5840
    },
    {
      "epoch": 1.736420302760463,
      "grad_norm": 0.7546972036361694,
      "learning_rate": 3.2635796972395374e-05,
      "loss": 0.9211,
      "step": 5850
    },
    {
      "epoch": 1.7393885425942415,
      "grad_norm": 0.6879614591598511,
      "learning_rate": 3.2606114574057584e-05,
      "loss": 0.8859,
      "step": 5860
    },
    {
      "epoch": 1.7423567824280202,
      "grad_norm": 0.760108232498169,
      "learning_rate": 3.25764321757198e-05,
      "loss": 0.8904,
      "step": 5870
    },
    {
      "epoch": 1.7453250222617989,
      "grad_norm": 0.6350589990615845,
      "learning_rate": 3.2546749777382016e-05,
      "loss": 0.9096,
      "step": 5880
    },
    {
      "epoch": 1.7482932620955773,
      "grad_norm": 0.7435024976730347,
      "learning_rate": 3.251706737904423e-05,
      "loss": 0.891,
      "step": 5890
    },
    {
      "epoch": 1.7512615019293558,
      "grad_norm": 0.7279981970787048,
      "learning_rate": 3.248738498070644e-05,
      "loss": 0.8982,
      "step": 5900
    },
    {
      "epoch": 1.7542297417631345,
      "grad_norm": 0.7936573624610901,
      "learning_rate": 3.245770258236865e-05,
      "loss": 0.9058,
      "step": 5910
    },
    {
      "epoch": 1.757197981596913,
      "grad_norm": 0.6942318081855774,
      "learning_rate": 3.2428020184030874e-05,
      "loss": 0.9284,
      "step": 5920
    },
    {
      "epoch": 1.7601662214306915,
      "grad_norm": 0.7330809235572815,
      "learning_rate": 3.2398337785693083e-05,
      "loss": 0.9585,
      "step": 5930
    },
    {
      "epoch": 1.7631344612644702,
      "grad_norm": 0.7430150508880615,
      "learning_rate": 3.23686553873553e-05,
      "loss": 0.898,
      "step": 5940
    },
    {
      "epoch": 1.7661027010982489,
      "grad_norm": 0.6925823092460632,
      "learning_rate": 3.2338972989017516e-05,
      "loss": 0.9267,
      "step": 5950
    },
    {
      "epoch": 1.7690709409320273,
      "grad_norm": 0.6921890377998352,
      "learning_rate": 3.230929059067973e-05,
      "loss": 0.9102,
      "step": 5960
    },
    {
      "epoch": 1.7720391807658058,
      "grad_norm": 0.7293705344200134,
      "learning_rate": 3.227960819234194e-05,
      "loss": 0.9114,
      "step": 5970
    },
    {
      "epoch": 1.7750074205995845,
      "grad_norm": 0.6864549517631531,
      "learning_rate": 3.224992579400416e-05,
      "loss": 0.9243,
      "step": 5980
    },
    {
      "epoch": 1.777975660433363,
      "grad_norm": 0.7275557518005371,
      "learning_rate": 3.2220243395666374e-05,
      "loss": 0.8826,
      "step": 5990
    },
    {
      "epoch": 1.7809439002671414,
      "grad_norm": 0.7278900742530823,
      "learning_rate": 3.219056099732858e-05,
      "loss": 0.8902,
      "step": 6000
    },
    {
      "epoch": 1.7839121401009201,
      "grad_norm": 0.6495439410209656,
      "learning_rate": 3.21608785989908e-05,
      "loss": 0.9048,
      "step": 6010
    },
    {
      "epoch": 1.7868803799346988,
      "grad_norm": 0.6814368367195129,
      "learning_rate": 3.2131196200653016e-05,
      "loss": 0.9232,
      "step": 6020
    },
    {
      "epoch": 1.7898486197684773,
      "grad_norm": 0.756121039390564,
      "learning_rate": 3.210151380231523e-05,
      "loss": 0.9336,
      "step": 6030
    },
    {
      "epoch": 1.7928168596022558,
      "grad_norm": 0.7354190945625305,
      "learning_rate": 3.207183140397744e-05,
      "loss": 0.9535,
      "step": 6040
    },
    {
      "epoch": 1.7957850994360345,
      "grad_norm": 0.6970518827438354,
      "learning_rate": 3.204214900563966e-05,
      "loss": 0.9115,
      "step": 6050
    },
    {
      "epoch": 1.798753339269813,
      "grad_norm": 0.7166738510131836,
      "learning_rate": 3.2012466607301873e-05,
      "loss": 0.9179,
      "step": 6060
    },
    {
      "epoch": 1.8017215791035914,
      "grad_norm": 0.7010670304298401,
      "learning_rate": 3.198278420896408e-05,
      "loss": 0.9037,
      "step": 6070
    },
    {
      "epoch": 1.8046898189373701,
      "grad_norm": 0.7266472578048706,
      "learning_rate": 3.19531018106263e-05,
      "loss": 0.9261,
      "step": 6080
    },
    {
      "epoch": 1.8076580587711488,
      "grad_norm": 0.7845122814178467,
      "learning_rate": 3.1923419412288515e-05,
      "loss": 0.9096,
      "step": 6090
    },
    {
      "epoch": 1.8106262986049273,
      "grad_norm": 0.7793822288513184,
      "learning_rate": 3.189373701395073e-05,
      "loss": 0.9038,
      "step": 6100
    },
    {
      "epoch": 1.8135945384387058,
      "grad_norm": 0.6785141229629517,
      "learning_rate": 3.186405461561294e-05,
      "loss": 0.9126,
      "step": 6110
    },
    {
      "epoch": 1.8165627782724845,
      "grad_norm": 0.7554550766944885,
      "learning_rate": 3.183437221727516e-05,
      "loss": 0.9203,
      "step": 6120
    },
    {
      "epoch": 1.819531018106263,
      "grad_norm": 0.7003015279769897,
      "learning_rate": 3.180468981893737e-05,
      "loss": 0.9211,
      "step": 6130
    },
    {
      "epoch": 1.8224992579400414,
      "grad_norm": 0.6762093901634216,
      "learning_rate": 3.177500742059958e-05,
      "loss": 0.9299,
      "step": 6140
    },
    {
      "epoch": 1.8254674977738201,
      "grad_norm": 0.6585193872451782,
      "learning_rate": 3.17453250222618e-05,
      "loss": 0.9087,
      "step": 6150
    },
    {
      "epoch": 1.8284357376075988,
      "grad_norm": 0.7564946413040161,
      "learning_rate": 3.1715642623924015e-05,
      "loss": 0.9609,
      "step": 6160
    },
    {
      "epoch": 1.8314039774413773,
      "grad_norm": 0.7777302265167236,
      "learning_rate": 3.168596022558623e-05,
      "loss": 0.9081,
      "step": 6170
    },
    {
      "epoch": 1.8343722172751558,
      "grad_norm": 0.6798081994056702,
      "learning_rate": 3.165627782724844e-05,
      "loss": 0.9173,
      "step": 6180
    },
    {
      "epoch": 1.8373404571089345,
      "grad_norm": 0.7171394228935242,
      "learning_rate": 3.162659542891066e-05,
      "loss": 0.9549,
      "step": 6190
    },
    {
      "epoch": 1.840308696942713,
      "grad_norm": 0.621942937374115,
      "learning_rate": 3.159691303057287e-05,
      "loss": 0.938,
      "step": 6200
    },
    {
      "epoch": 1.8432769367764914,
      "grad_norm": 0.6758529543876648,
      "learning_rate": 3.156723063223508e-05,
      "loss": 0.9292,
      "step": 6210
    },
    {
      "epoch": 1.84624517661027,
      "grad_norm": 0.7157332897186279,
      "learning_rate": 3.15375482338973e-05,
      "loss": 0.9095,
      "step": 6220
    },
    {
      "epoch": 1.8492134164440488,
      "grad_norm": 0.6590803265571594,
      "learning_rate": 3.1507865835559515e-05,
      "loss": 0.9146,
      "step": 6230
    },
    {
      "epoch": 1.8521816562778273,
      "grad_norm": 0.6611948609352112,
      "learning_rate": 3.147818343722173e-05,
      "loss": 0.8839,
      "step": 6240
    },
    {
      "epoch": 1.8551498961116057,
      "grad_norm": 0.7113106846809387,
      "learning_rate": 3.144850103888394e-05,
      "loss": 0.9448,
      "step": 6250
    },
    {
      "epoch": 1.8581181359453844,
      "grad_norm": 0.6739339828491211,
      "learning_rate": 3.1418818640546156e-05,
      "loss": 0.8689,
      "step": 6260
    },
    {
      "epoch": 1.861086375779163,
      "grad_norm": 0.7268794178962708,
      "learning_rate": 3.138913624220837e-05,
      "loss": 0.9046,
      "step": 6270
    },
    {
      "epoch": 1.8640546156129414,
      "grad_norm": 0.6519601345062256,
      "learning_rate": 3.135945384387058e-05,
      "loss": 0.95,
      "step": 6280
    },
    {
      "epoch": 1.86702285544672,
      "grad_norm": 0.7493473291397095,
      "learning_rate": 3.1329771445532805e-05,
      "loss": 0.9424,
      "step": 6290
    },
    {
      "epoch": 1.8699910952804988,
      "grad_norm": 0.7745805382728577,
      "learning_rate": 3.1300089047195014e-05,
      "loss": 0.8366,
      "step": 6300
    },
    {
      "epoch": 1.8729593351142773,
      "grad_norm": 0.7086494565010071,
      "learning_rate": 3.127040664885723e-05,
      "loss": 0.9233,
      "step": 6310
    },
    {
      "epoch": 1.8759275749480557,
      "grad_norm": 0.6021324992179871,
      "learning_rate": 3.124072425051944e-05,
      "loss": 0.8949,
      "step": 6320
    },
    {
      "epoch": 1.8788958147818344,
      "grad_norm": 0.7947562336921692,
      "learning_rate": 3.121104185218166e-05,
      "loss": 0.8961,
      "step": 6330
    },
    {
      "epoch": 1.881864054615613,
      "grad_norm": 0.6500969529151917,
      "learning_rate": 3.118135945384387e-05,
      "loss": 0.9317,
      "step": 6340
    },
    {
      "epoch": 1.8848322944493914,
      "grad_norm": 0.7063040137290955,
      "learning_rate": 3.115167705550608e-05,
      "loss": 0.8962,
      "step": 6350
    },
    {
      "epoch": 1.88780053428317,
      "grad_norm": 0.7632299661636353,
      "learning_rate": 3.1121994657168305e-05,
      "loss": 0.891,
      "step": 6360
    },
    {
      "epoch": 1.8907687741169488,
      "grad_norm": 0.7408263087272644,
      "learning_rate": 3.1092312258830514e-05,
      "loss": 0.9047,
      "step": 6370
    },
    {
      "epoch": 1.8937370139507272,
      "grad_norm": 0.7138951420783997,
      "learning_rate": 3.106262986049273e-05,
      "loss": 0.8845,
      "step": 6380
    },
    {
      "epoch": 1.8967052537845057,
      "grad_norm": 0.7365957498550415,
      "learning_rate": 3.1032947462154946e-05,
      "loss": 0.8832,
      "step": 6390
    },
    {
      "epoch": 1.8996734936182844,
      "grad_norm": 0.6995649337768555,
      "learning_rate": 3.100326506381716e-05,
      "loss": 0.9068,
      "step": 6400
    },
    {
      "epoch": 1.9026417334520629,
      "grad_norm": 0.6828333735466003,
      "learning_rate": 3.097358266547937e-05,
      "loss": 0.9118,
      "step": 6410
    },
    {
      "epoch": 1.9056099732858414,
      "grad_norm": 0.6359698176383972,
      "learning_rate": 3.094390026714158e-05,
      "loss": 0.8868,
      "step": 6420
    },
    {
      "epoch": 1.90857821311962,
      "grad_norm": 0.6921895146369934,
      "learning_rate": 3.0914217868803804e-05,
      "loss": 0.8764,
      "step": 6430
    },
    {
      "epoch": 1.9115464529533988,
      "grad_norm": 0.7117785215377808,
      "learning_rate": 3.0884535470466014e-05,
      "loss": 0.923,
      "step": 6440
    },
    {
      "epoch": 1.9145146927871772,
      "grad_norm": 0.6908830404281616,
      "learning_rate": 3.085485307212823e-05,
      "loss": 0.8842,
      "step": 6450
    },
    {
      "epoch": 1.9174829326209557,
      "grad_norm": 0.651472806930542,
      "learning_rate": 3.0825170673790446e-05,
      "loss": 0.9351,
      "step": 6460
    },
    {
      "epoch": 1.9204511724547344,
      "grad_norm": 0.7545375823974609,
      "learning_rate": 3.079548827545266e-05,
      "loss": 0.9037,
      "step": 6470
    },
    {
      "epoch": 1.9234194122885129,
      "grad_norm": 0.6635625958442688,
      "learning_rate": 3.076580587711487e-05,
      "loss": 0.9218,
      "step": 6480
    },
    {
      "epoch": 1.9263876521222913,
      "grad_norm": 0.642776608467102,
      "learning_rate": 3.073612347877709e-05,
      "loss": 0.9211,
      "step": 6490
    },
    {
      "epoch": 1.92935589195607,
      "grad_norm": 0.605273425579071,
      "learning_rate": 3.0706441080439304e-05,
      "loss": 0.8955,
      "step": 6500
    },
    {
      "epoch": 1.9323241317898487,
      "grad_norm": 0.7083524465560913,
      "learning_rate": 3.067675868210151e-05,
      "loss": 0.9138,
      "step": 6510
    },
    {
      "epoch": 1.9352923716236272,
      "grad_norm": 0.6784158945083618,
      "learning_rate": 3.064707628376373e-05,
      "loss": 0.9245,
      "step": 6520
    },
    {
      "epoch": 1.9382606114574057,
      "grad_norm": 0.6602509021759033,
      "learning_rate": 3.0617393885425946e-05,
      "loss": 0.9042,
      "step": 6530
    },
    {
      "epoch": 1.9412288512911844,
      "grad_norm": 0.8108129501342773,
      "learning_rate": 3.058771148708816e-05,
      "loss": 0.873,
      "step": 6540
    },
    {
      "epoch": 1.9441970911249629,
      "grad_norm": 0.7119005918502808,
      "learning_rate": 3.055802908875037e-05,
      "loss": 0.8816,
      "step": 6550
    },
    {
      "epoch": 1.9471653309587413,
      "grad_norm": 0.6957710981369019,
      "learning_rate": 3.052834669041259e-05,
      "loss": 0.8922,
      "step": 6560
    },
    {
      "epoch": 1.95013357079252,
      "grad_norm": 0.6639032363891602,
      "learning_rate": 3.0498664292074804e-05,
      "loss": 0.952,
      "step": 6570
    },
    {
      "epoch": 1.9531018106262987,
      "grad_norm": 0.7830907106399536,
      "learning_rate": 3.0468981893737013e-05,
      "loss": 0.8939,
      "step": 6580
    },
    {
      "epoch": 1.9560700504600772,
      "grad_norm": 0.7688878774642944,
      "learning_rate": 3.0439299495399233e-05,
      "loss": 0.8802,
      "step": 6590
    },
    {
      "epoch": 1.9590382902938557,
      "grad_norm": 0.710884690284729,
      "learning_rate": 3.0409617097061442e-05,
      "loss": 0.9142,
      "step": 6600
    },
    {
      "epoch": 1.9620065301276344,
      "grad_norm": 0.7745290994644165,
      "learning_rate": 3.037993469872366e-05,
      "loss": 0.9117,
      "step": 6610
    },
    {
      "epoch": 1.9649747699614128,
      "grad_norm": 0.8341137766838074,
      "learning_rate": 3.0350252300385874e-05,
      "loss": 0.8686,
      "step": 6620
    },
    {
      "epoch": 1.9679430097951913,
      "grad_norm": 0.7225478887557983,
      "learning_rate": 3.0320569902048084e-05,
      "loss": 0.9018,
      "step": 6630
    },
    {
      "epoch": 1.97091124962897,
      "grad_norm": 0.7451010942459106,
      "learning_rate": 3.0290887503710303e-05,
      "loss": 0.8903,
      "step": 6640
    },
    {
      "epoch": 1.9738794894627487,
      "grad_norm": 0.6880558133125305,
      "learning_rate": 3.0261205105372513e-05,
      "loss": 0.9173,
      "step": 6650
    },
    {
      "epoch": 1.9768477292965272,
      "grad_norm": 0.6597363352775574,
      "learning_rate": 3.0231522707034732e-05,
      "loss": 0.9326,
      "step": 6660
    },
    {
      "epoch": 1.9798159691303057,
      "grad_norm": 0.6880154013633728,
      "learning_rate": 3.0201840308696945e-05,
      "loss": 0.8933,
      "step": 6670
    },
    {
      "epoch": 1.9827842089640844,
      "grad_norm": 0.755906343460083,
      "learning_rate": 3.017215791035916e-05,
      "loss": 0.8941,
      "step": 6680
    },
    {
      "epoch": 1.9857524487978628,
      "grad_norm": 0.7369247078895569,
      "learning_rate": 3.0142475512021374e-05,
      "loss": 0.8862,
      "step": 6690
    },
    {
      "epoch": 1.9887206886316413,
      "grad_norm": 0.7142388820648193,
      "learning_rate": 3.0112793113683583e-05,
      "loss": 0.8947,
      "step": 6700
    },
    {
      "epoch": 1.99168892846542,
      "grad_norm": 0.7839199900627136,
      "learning_rate": 3.0083110715345803e-05,
      "loss": 0.9156,
      "step": 6710
    },
    {
      "epoch": 1.9946571682991987,
      "grad_norm": 0.7071635723114014,
      "learning_rate": 3.0053428317008016e-05,
      "loss": 0.9129,
      "step": 6720
    },
    {
      "epoch": 1.9976254081329772,
      "grad_norm": 0.6864856481552124,
      "learning_rate": 3.0023745918670232e-05,
      "loss": 0.8744,
      "step": 6730
    },
    {
      "epoch": 2.0005936479667557,
      "grad_norm": 0.7013072371482849,
      "learning_rate": 2.9994063520332445e-05,
      "loss": 0.8763,
      "step": 6740
    },
    {
      "epoch": 2.0035618878005343,
      "grad_norm": 0.7171157002449036,
      "learning_rate": 2.996438112199466e-05,
      "loss": 0.8503,
      "step": 6750
    },
    {
      "epoch": 2.006530127634313,
      "grad_norm": 0.7891656756401062,
      "learning_rate": 2.9934698723656874e-05,
      "loss": 0.8224,
      "step": 6760
    },
    {
      "epoch": 2.0094983674680913,
      "grad_norm": 0.7087323069572449,
      "learning_rate": 2.9905016325319086e-05,
      "loss": 0.8945,
      "step": 6770
    },
    {
      "epoch": 2.01246660730187,
      "grad_norm": 0.750305712223053,
      "learning_rate": 2.9875333926981303e-05,
      "loss": 0.8414,
      "step": 6780
    },
    {
      "epoch": 2.0154348471356487,
      "grad_norm": 0.6855888366699219,
      "learning_rate": 2.9845651528643515e-05,
      "loss": 0.8437,
      "step": 6790
    },
    {
      "epoch": 2.018403086969427,
      "grad_norm": 0.7236269116401672,
      "learning_rate": 2.981596913030573e-05,
      "loss": 0.8373,
      "step": 6800
    },
    {
      "epoch": 2.0213713268032056,
      "grad_norm": 0.7629292011260986,
      "learning_rate": 2.9786286731967944e-05,
      "loss": 0.8659,
      "step": 6810
    },
    {
      "epoch": 2.0243395666369843,
      "grad_norm": 0.8071072697639465,
      "learning_rate": 2.975660433363016e-05,
      "loss": 0.9247,
      "step": 6820
    },
    {
      "epoch": 2.027307806470763,
      "grad_norm": 0.7631831765174866,
      "learning_rate": 2.9726921935292373e-05,
      "loss": 0.8482,
      "step": 6830
    },
    {
      "epoch": 2.0302760463045413,
      "grad_norm": 0.681796669960022,
      "learning_rate": 2.9697239536954586e-05,
      "loss": 0.8714,
      "step": 6840
    },
    {
      "epoch": 2.03324428613832,
      "grad_norm": 0.7427048087120056,
      "learning_rate": 2.9667557138616802e-05,
      "loss": 0.8349,
      "step": 6850
    },
    {
      "epoch": 2.0362125259720987,
      "grad_norm": 0.7295932769775391,
      "learning_rate": 2.9637874740279015e-05,
      "loss": 0.829,
      "step": 6860
    },
    {
      "epoch": 2.039180765805877,
      "grad_norm": 0.6897096633911133,
      "learning_rate": 2.960819234194123e-05,
      "loss": 0.8319,
      "step": 6870
    },
    {
      "epoch": 2.0421490056396556,
      "grad_norm": 0.7505813837051392,
      "learning_rate": 2.9578509943603444e-05,
      "loss": 0.869,
      "step": 6880
    },
    {
      "epoch": 2.0451172454734343,
      "grad_norm": 0.7533600330352783,
      "learning_rate": 2.954882754526566e-05,
      "loss": 0.8683,
      "step": 6890
    },
    {
      "epoch": 2.048085485307213,
      "grad_norm": 0.7831742167472839,
      "learning_rate": 2.9519145146927873e-05,
      "loss": 0.8161,
      "step": 6900
    },
    {
      "epoch": 2.0510537251409913,
      "grad_norm": 0.7394450306892395,
      "learning_rate": 2.9489462748590086e-05,
      "loss": 0.8834,
      "step": 6910
    },
    {
      "epoch": 2.05402196497477,
      "grad_norm": 0.7655673623085022,
      "learning_rate": 2.9459780350252302e-05,
      "loss": 0.8265,
      "step": 6920
    },
    {
      "epoch": 2.0569902048085487,
      "grad_norm": 0.7019949555397034,
      "learning_rate": 2.9430097951914515e-05,
      "loss": 0.8573,
      "step": 6930
    },
    {
      "epoch": 2.059958444642327,
      "grad_norm": 0.6387875080108643,
      "learning_rate": 2.940041555357673e-05,
      "loss": 0.8259,
      "step": 6940
    },
    {
      "epoch": 2.0629266844761056,
      "grad_norm": 0.7614507079124451,
      "learning_rate": 2.9370733155238944e-05,
      "loss": 0.8758,
      "step": 6950
    },
    {
      "epoch": 2.0658949243098843,
      "grad_norm": 0.7810057401657104,
      "learning_rate": 2.9341050756901163e-05,
      "loss": 0.824,
      "step": 6960
    },
    {
      "epoch": 2.068863164143663,
      "grad_norm": 0.7228024005889893,
      "learning_rate": 2.9311368358563373e-05,
      "loss": 0.8345,
      "step": 6970
    },
    {
      "epoch": 2.0718314039774413,
      "grad_norm": 0.6962882876396179,
      "learning_rate": 2.9281685960225585e-05,
      "loss": 0.8099,
      "step": 6980
    },
    {
      "epoch": 2.07479964381122,
      "grad_norm": 0.8062400221824646,
      "learning_rate": 2.92520035618878e-05,
      "loss": 0.8453,
      "step": 6990
    },
    {
      "epoch": 2.0777678836449986,
      "grad_norm": 0.787611186504364,
      "learning_rate": 2.9222321163550014e-05,
      "loss": 0.8787,
      "step": 7000
    },
    {
      "epoch": 2.080736123478777,
      "grad_norm": 0.7147652506828308,
      "learning_rate": 2.9192638765212234e-05,
      "loss": 0.8768,
      "step": 7010
    },
    {
      "epoch": 2.0837043633125556,
      "grad_norm": 0.6608722805976868,
      "learning_rate": 2.9162956366874443e-05,
      "loss": 0.8247,
      "step": 7020
    },
    {
      "epoch": 2.0866726031463343,
      "grad_norm": 0.752377450466156,
      "learning_rate": 2.9133273968536663e-05,
      "loss": 0.8037,
      "step": 7030
    },
    {
      "epoch": 2.089640842980113,
      "grad_norm": 0.8308156728744507,
      "learning_rate": 2.9103591570198872e-05,
      "loss": 0.8878,
      "step": 7040
    },
    {
      "epoch": 2.0926090828138912,
      "grad_norm": 0.6480581164360046,
      "learning_rate": 2.9073909171861085e-05,
      "loss": 0.8472,
      "step": 7050
    },
    {
      "epoch": 2.09557732264767,
      "grad_norm": 0.7647300362586975,
      "learning_rate": 2.9044226773523305e-05,
      "loss": 0.864,
      "step": 7060
    },
    {
      "epoch": 2.0985455624814486,
      "grad_norm": 0.7759458422660828,
      "learning_rate": 2.9014544375185514e-05,
      "loss": 0.8124,
      "step": 7070
    },
    {
      "epoch": 2.101513802315227,
      "grad_norm": 0.7516981363296509,
      "learning_rate": 2.8984861976847734e-05,
      "loss": 0.8517,
      "step": 7080
    },
    {
      "epoch": 2.1044820421490056,
      "grad_norm": 0.7401167750358582,
      "learning_rate": 2.8955179578509943e-05,
      "loss": 0.8302,
      "step": 7090
    },
    {
      "epoch": 2.1074502819827843,
      "grad_norm": 0.7449563145637512,
      "learning_rate": 2.8925497180172163e-05,
      "loss": 0.873,
      "step": 7100
    },
    {
      "epoch": 2.110418521816563,
      "grad_norm": 0.8492381572723389,
      "learning_rate": 2.8895814781834375e-05,
      "loss": 0.822,
      "step": 7110
    },
    {
      "epoch": 2.1133867616503412,
      "grad_norm": 0.7222036719322205,
      "learning_rate": 2.8866132383496585e-05,
      "loss": 0.818,
      "step": 7120
    },
    {
      "epoch": 2.11635500148412,
      "grad_norm": 0.7867766618728638,
      "learning_rate": 2.8836449985158804e-05,
      "loss": 0.8539,
      "step": 7130
    },
    {
      "epoch": 2.1193232413178986,
      "grad_norm": 0.7673478126525879,
      "learning_rate": 2.8806767586821014e-05,
      "loss": 0.8221,
      "step": 7140
    },
    {
      "epoch": 2.122291481151677,
      "grad_norm": 0.7407607436180115,
      "learning_rate": 2.8777085188483233e-05,
      "loss": 0.8626,
      "step": 7150
    },
    {
      "epoch": 2.1252597209854556,
      "grad_norm": 0.6958709955215454,
      "learning_rate": 2.8747402790145446e-05,
      "loss": 0.8333,
      "step": 7160
    },
    {
      "epoch": 2.1282279608192343,
      "grad_norm": 0.7231135964393616,
      "learning_rate": 2.8717720391807662e-05,
      "loss": 0.8548,
      "step": 7170
    },
    {
      "epoch": 2.131196200653013,
      "grad_norm": 0.9101832509040833,
      "learning_rate": 2.8688037993469875e-05,
      "loss": 0.8288,
      "step": 7180
    },
    {
      "epoch": 2.134164440486791,
      "grad_norm": 0.7785500884056091,
      "learning_rate": 2.8658355595132085e-05,
      "loss": 0.8675,
      "step": 7190
    },
    {
      "epoch": 2.13713268032057,
      "grad_norm": 0.7221528887748718,
      "learning_rate": 2.8628673196794304e-05,
      "loss": 0.8963,
      "step": 7200
    },
    {
      "epoch": 2.1401009201543486,
      "grad_norm": 0.7630613446235657,
      "learning_rate": 2.8598990798456517e-05,
      "loss": 0.8525,
      "step": 7210
    },
    {
      "epoch": 2.143069159988127,
      "grad_norm": 0.7581575512886047,
      "learning_rate": 2.8569308400118733e-05,
      "loss": 0.8182,
      "step": 7220
    },
    {
      "epoch": 2.1460373998219056,
      "grad_norm": 0.9540184140205383,
      "learning_rate": 2.8539626001780946e-05,
      "loss": 0.8806,
      "step": 7230
    },
    {
      "epoch": 2.1490056396556843,
      "grad_norm": 0.7357391715049744,
      "learning_rate": 2.8509943603443162e-05,
      "loss": 0.8415,
      "step": 7240
    },
    {
      "epoch": 2.151973879489463,
      "grad_norm": 0.9968686103820801,
      "learning_rate": 2.8480261205105375e-05,
      "loss": 0.8193,
      "step": 7250
    },
    {
      "epoch": 2.154942119323241,
      "grad_norm": 0.8082394003868103,
      "learning_rate": 2.8450578806767588e-05,
      "loss": 0.8351,
      "step": 7260
    },
    {
      "epoch": 2.15791035915702,
      "grad_norm": 0.757830023765564,
      "learning_rate": 2.8420896408429804e-05,
      "loss": 0.8291,
      "step": 7270
    },
    {
      "epoch": 2.1608785989907986,
      "grad_norm": 0.8347230553627014,
      "learning_rate": 2.8391214010092017e-05,
      "loss": 0.8315,
      "step": 7280
    },
    {
      "epoch": 2.163846838824577,
      "grad_norm": 0.6753961443901062,
      "learning_rate": 2.8361531611754233e-05,
      "loss": 0.8755,
      "step": 7290
    },
    {
      "epoch": 2.1668150786583555,
      "grad_norm": 0.8221672773361206,
      "learning_rate": 2.8331849213416445e-05,
      "loss": 0.8614,
      "step": 7300
    },
    {
      "epoch": 2.1697833184921342,
      "grad_norm": 0.7886311411857605,
      "learning_rate": 2.830216681507866e-05,
      "loss": 0.8719,
      "step": 7310
    },
    {
      "epoch": 2.172751558325913,
      "grad_norm": 0.7683786749839783,
      "learning_rate": 2.8272484416740874e-05,
      "loss": 0.863,
      "step": 7320
    },
    {
      "epoch": 2.175719798159691,
      "grad_norm": 0.8968241810798645,
      "learning_rate": 2.8242802018403087e-05,
      "loss": 0.873,
      "step": 7330
    },
    {
      "epoch": 2.17868803799347,
      "grad_norm": 0.819404125213623,
      "learning_rate": 2.8213119620065303e-05,
      "loss": 0.8907,
      "step": 7340
    },
    {
      "epoch": 2.1816562778272486,
      "grad_norm": 0.7417108416557312,
      "learning_rate": 2.8183437221727516e-05,
      "loss": 0.8809,
      "step": 7350
    },
    {
      "epoch": 2.184624517661027,
      "grad_norm": 0.8411499261856079,
      "learning_rate": 2.8153754823389732e-05,
      "loss": 0.9122,
      "step": 7360
    },
    {
      "epoch": 2.1875927574948055,
      "grad_norm": 0.776980996131897,
      "learning_rate": 2.8124072425051945e-05,
      "loss": 0.854,
      "step": 7370
    },
    {
      "epoch": 2.1905609973285842,
      "grad_norm": 0.7408774495124817,
      "learning_rate": 2.809439002671416e-05,
      "loss": 0.8633,
      "step": 7380
    },
    {
      "epoch": 2.193529237162363,
      "grad_norm": 0.8015629649162292,
      "learning_rate": 2.8064707628376374e-05,
      "loss": 0.8479,
      "step": 7390
    },
    {
      "epoch": 2.196497476996141,
      "grad_norm": 0.833080530166626,
      "learning_rate": 2.8035025230038587e-05,
      "loss": 0.8482,
      "step": 7400
    },
    {
      "epoch": 2.19946571682992,
      "grad_norm": 0.720993161201477,
      "learning_rate": 2.8005342831700803e-05,
      "loss": 0.8626,
      "step": 7410
    },
    {
      "epoch": 2.2024339566636986,
      "grad_norm": 0.7985531091690063,
      "learning_rate": 2.7975660433363016e-05,
      "loss": 0.8456,
      "step": 7420
    },
    {
      "epoch": 2.205402196497477,
      "grad_norm": 0.7086364030838013,
      "learning_rate": 2.7945978035025232e-05,
      "loss": 0.8585,
      "step": 7430
    },
    {
      "epoch": 2.2083704363312555,
      "grad_norm": 0.8406996130943298,
      "learning_rate": 2.7916295636687445e-05,
      "loss": 0.8614,
      "step": 7440
    },
    {
      "epoch": 2.211338676165034,
      "grad_norm": 0.8609254956245422,
      "learning_rate": 2.788661323834966e-05,
      "loss": 0.8563,
      "step": 7450
    },
    {
      "epoch": 2.214306915998813,
      "grad_norm": 0.7881911396980286,
      "learning_rate": 2.7856930840011874e-05,
      "loss": 0.8697,
      "step": 7460
    },
    {
      "epoch": 2.217275155832591,
      "grad_norm": 0.6949644088745117,
      "learning_rate": 2.7827248441674087e-05,
      "loss": 0.868,
      "step": 7470
    },
    {
      "epoch": 2.22024339566637,
      "grad_norm": 0.786999523639679,
      "learning_rate": 2.7797566043336303e-05,
      "loss": 0.8641,
      "step": 7480
    },
    {
      "epoch": 2.2232116355001486,
      "grad_norm": 0.7273027300834656,
      "learning_rate": 2.7767883644998516e-05,
      "loss": 0.8403,
      "step": 7490
    },
    {
      "epoch": 2.226179875333927,
      "grad_norm": 0.8558393120765686,
      "learning_rate": 2.7738201246660732e-05,
      "loss": 0.8597,
      "step": 7500
    },
    {
      "epoch": 2.2291481151677055,
      "grad_norm": 0.7229394912719727,
      "learning_rate": 2.7708518848322945e-05,
      "loss": 0.8706,
      "step": 7510
    },
    {
      "epoch": 2.232116355001484,
      "grad_norm": 0.825646698474884,
      "learning_rate": 2.7678836449985164e-05,
      "loss": 0.9021,
      "step": 7520
    },
    {
      "epoch": 2.235084594835263,
      "grad_norm": 0.8738429546356201,
      "learning_rate": 2.7649154051647374e-05,
      "loss": 0.8346,
      "step": 7530
    },
    {
      "epoch": 2.238052834669041,
      "grad_norm": 0.7595973014831543,
      "learning_rate": 2.7619471653309586e-05,
      "loss": 0.8598,
      "step": 7540
    },
    {
      "epoch": 2.24102107450282,
      "grad_norm": 0.8204748630523682,
      "learning_rate": 2.7589789254971802e-05,
      "loss": 0.8509,
      "step": 7550
    },
    {
      "epoch": 2.2439893143365985,
      "grad_norm": 0.7805768847465515,
      "learning_rate": 2.7560106856634015e-05,
      "loss": 0.8247,
      "step": 7560
    },
    {
      "epoch": 2.246957554170377,
      "grad_norm": 0.8872819542884827,
      "learning_rate": 2.7530424458296235e-05,
      "loss": 0.8543,
      "step": 7570
    },
    {
      "epoch": 2.2499257940041555,
      "grad_norm": 0.789537250995636,
      "learning_rate": 2.7500742059958444e-05,
      "loss": 0.8641,
      "step": 7580
    },
    {
      "epoch": 2.252894033837934,
      "grad_norm": 0.7404807806015015,
      "learning_rate": 2.7471059661620664e-05,
      "loss": 0.8518,
      "step": 7590
    },
    {
      "epoch": 2.2558622736717124,
      "grad_norm": 0.7324669361114502,
      "learning_rate": 2.7441377263282873e-05,
      "loss": 0.8457,
      "step": 7600
    },
    {
      "epoch": 2.258830513505491,
      "grad_norm": 0.7856626510620117,
      "learning_rate": 2.7411694864945086e-05,
      "loss": 0.8568,
      "step": 7610
    },
    {
      "epoch": 2.26179875333927,
      "grad_norm": 0.7492886781692505,
      "learning_rate": 2.7382012466607306e-05,
      "loss": 0.8075,
      "step": 7620
    },
    {
      "epoch": 2.2647669931730485,
      "grad_norm": 0.7941344976425171,
      "learning_rate": 2.7352330068269515e-05,
      "loss": 0.8431,
      "step": 7630
    },
    {
      "epoch": 2.267735233006827,
      "grad_norm": 0.7928834557533264,
      "learning_rate": 2.7322647669931734e-05,
      "loss": 0.8743,
      "step": 7640
    },
    {
      "epoch": 2.2707034728406055,
      "grad_norm": 0.7476490139961243,
      "learning_rate": 2.7292965271593944e-05,
      "loss": 0.8079,
      "step": 7650
    },
    {
      "epoch": 2.273671712674384,
      "grad_norm": 0.7902363538742065,
      "learning_rate": 2.7263282873256163e-05,
      "loss": 0.8351,
      "step": 7660
    },
    {
      "epoch": 2.2766399525081624,
      "grad_norm": 0.7967187166213989,
      "learning_rate": 2.7233600474918376e-05,
      "loss": 0.8835,
      "step": 7670
    },
    {
      "epoch": 2.279608192341941,
      "grad_norm": 0.8231721520423889,
      "learning_rate": 2.7203918076580586e-05,
      "loss": 0.8446,
      "step": 7680
    },
    {
      "epoch": 2.28257643217572,
      "grad_norm": 0.7556421160697937,
      "learning_rate": 2.7174235678242805e-05,
      "loss": 0.8458,
      "step": 7690
    },
    {
      "epoch": 2.2855446720094985,
      "grad_norm": 0.7829645872116089,
      "learning_rate": 2.7144553279905015e-05,
      "loss": 0.8547,
      "step": 7700
    },
    {
      "epoch": 2.2885129118432768,
      "grad_norm": 0.7063056230545044,
      "learning_rate": 2.7114870881567234e-05,
      "loss": 0.869,
      "step": 7710
    },
    {
      "epoch": 2.2914811516770555,
      "grad_norm": 0.7241949439048767,
      "learning_rate": 2.7085188483229447e-05,
      "loss": 0.8424,
      "step": 7720
    },
    {
      "epoch": 2.294449391510834,
      "grad_norm": 0.7578948140144348,
      "learning_rate": 2.7055506084891663e-05,
      "loss": 0.8515,
      "step": 7730
    },
    {
      "epoch": 2.2974176313446124,
      "grad_norm": 0.835233211517334,
      "learning_rate": 2.7025823686553876e-05,
      "loss": 0.8605,
      "step": 7740
    },
    {
      "epoch": 2.300385871178391,
      "grad_norm": 0.7147203087806702,
      "learning_rate": 2.6996141288216085e-05,
      "loss": 0.8272,
      "step": 7750
    },
    {
      "epoch": 2.30335411101217,
      "grad_norm": 0.7513293027877808,
      "learning_rate": 2.6966458889878305e-05,
      "loss": 0.8783,
      "step": 7760
    },
    {
      "epoch": 2.3063223508459485,
      "grad_norm": 0.8606356978416443,
      "learning_rate": 2.6936776491540518e-05,
      "loss": 0.8841,
      "step": 7770
    },
    {
      "epoch": 2.3092905906797268,
      "grad_norm": 0.8113495707511902,
      "learning_rate": 2.6907094093202734e-05,
      "loss": 0.8665,
      "step": 7780
    },
    {
      "epoch": 2.3122588305135054,
      "grad_norm": 0.8095558285713196,
      "learning_rate": 2.6877411694864947e-05,
      "loss": 0.8226,
      "step": 7790
    },
    {
      "epoch": 2.315227070347284,
      "grad_norm": 0.73026442527771,
      "learning_rate": 2.6847729296527163e-05,
      "loss": 0.8518,
      "step": 7800
    },
    {
      "epoch": 2.3181953101810624,
      "grad_norm": 0.8076465129852295,
      "learning_rate": 2.6818046898189376e-05,
      "loss": 0.8224,
      "step": 7810
    },
    {
      "epoch": 2.321163550014841,
      "grad_norm": 0.7851926684379578,
      "learning_rate": 2.678836449985159e-05,
      "loss": 0.8679,
      "step": 7820
    },
    {
      "epoch": 2.32413178984862,
      "grad_norm": 0.7606148719787598,
      "learning_rate": 2.6758682101513805e-05,
      "loss": 0.876,
      "step": 7830
    },
    {
      "epoch": 2.3271000296823985,
      "grad_norm": 0.79587322473526,
      "learning_rate": 2.6728999703176017e-05,
      "loss": 0.8056,
      "step": 7840
    },
    {
      "epoch": 2.3300682695161767,
      "grad_norm": 0.6922584176063538,
      "learning_rate": 2.6699317304838234e-05,
      "loss": 0.836,
      "step": 7850
    },
    {
      "epoch": 2.3330365093499554,
      "grad_norm": 0.7687097787857056,
      "learning_rate": 2.6669634906500446e-05,
      "loss": 0.8151,
      "step": 7860
    },
    {
      "epoch": 2.336004749183734,
      "grad_norm": 0.8245091438293457,
      "learning_rate": 2.6639952508162663e-05,
      "loss": 0.7964,
      "step": 7870
    },
    {
      "epoch": 2.3389729890175124,
      "grad_norm": 0.7584673762321472,
      "learning_rate": 2.6610270109824875e-05,
      "loss": 0.8807,
      "step": 7880
    },
    {
      "epoch": 2.341941228851291,
      "grad_norm": 0.8042759895324707,
      "learning_rate": 2.6580587711487088e-05,
      "loss": 0.8211,
      "step": 7890
    },
    {
      "epoch": 2.3449094686850698,
      "grad_norm": 0.7295954823493958,
      "learning_rate": 2.6550905313149304e-05,
      "loss": 0.8415,
      "step": 7900
    },
    {
      "epoch": 2.3478777085188485,
      "grad_norm": 0.688973069190979,
      "learning_rate": 2.6521222914811517e-05,
      "loss": 0.8405,
      "step": 7910
    },
    {
      "epoch": 2.3508459483526267,
      "grad_norm": 0.7817041873931885,
      "learning_rate": 2.6491540516473733e-05,
      "loss": 0.8789,
      "step": 7920
    },
    {
      "epoch": 2.3538141881864054,
      "grad_norm": 0.7820131182670593,
      "learning_rate": 2.6461858118135946e-05,
      "loss": 0.8383,
      "step": 7930
    },
    {
      "epoch": 2.356782428020184,
      "grad_norm": 0.728888750076294,
      "learning_rate": 2.6432175719798162e-05,
      "loss": 0.8294,
      "step": 7940
    },
    {
      "epoch": 2.3597506678539624,
      "grad_norm": 0.7914735078811646,
      "learning_rate": 2.6402493321460375e-05,
      "loss": 0.8282,
      "step": 7950
    },
    {
      "epoch": 2.362718907687741,
      "grad_norm": 0.7743697762489319,
      "learning_rate": 2.6372810923122588e-05,
      "loss": 0.8325,
      "step": 7960
    },
    {
      "epoch": 2.3656871475215198,
      "grad_norm": 0.7644380927085876,
      "learning_rate": 2.6343128524784804e-05,
      "loss": 0.8519,
      "step": 7970
    },
    {
      "epoch": 2.3686553873552985,
      "grad_norm": 0.7865247130393982,
      "learning_rate": 2.6313446126447017e-05,
      "loss": 0.8961,
      "step": 7980
    },
    {
      "epoch": 2.3716236271890767,
      "grad_norm": 0.7742238640785217,
      "learning_rate": 2.6283763728109233e-05,
      "loss": 0.8527,
      "step": 7990
    },
    {
      "epoch": 2.3745918670228554,
      "grad_norm": 0.7000154256820679,
      "learning_rate": 2.6254081329771446e-05,
      "loss": 0.8694,
      "step": 8000
    },
    {
      "epoch": 2.377560106856634,
      "grad_norm": 0.7216348648071289,
      "learning_rate": 2.6224398931433662e-05,
      "loss": 0.8619,
      "step": 8010
    },
    {
      "epoch": 2.3805283466904124,
      "grad_norm": 0.7548758387565613,
      "learning_rate": 2.6194716533095875e-05,
      "loss": 0.8852,
      "step": 8020
    },
    {
      "epoch": 2.383496586524191,
      "grad_norm": 0.7663046717643738,
      "learning_rate": 2.6165034134758087e-05,
      "loss": 0.8495,
      "step": 8030
    },
    {
      "epoch": 2.3864648263579697,
      "grad_norm": 0.8307507038116455,
      "learning_rate": 2.6135351736420304e-05,
      "loss": 0.8275,
      "step": 8040
    },
    {
      "epoch": 2.3894330661917484,
      "grad_norm": 0.8365349173545837,
      "learning_rate": 2.6105669338082516e-05,
      "loss": 0.873,
      "step": 8050
    },
    {
      "epoch": 2.3924013060255267,
      "grad_norm": 0.8038210272789001,
      "learning_rate": 2.6075986939744736e-05,
      "loss": 0.8658,
      "step": 8060
    },
    {
      "epoch": 2.3953695458593054,
      "grad_norm": 0.7396426200866699,
      "learning_rate": 2.6046304541406945e-05,
      "loss": 0.8331,
      "step": 8070
    },
    {
      "epoch": 2.398337785693084,
      "grad_norm": 0.8063576221466064,
      "learning_rate": 2.6016622143069165e-05,
      "loss": 0.8453,
      "step": 8080
    },
    {
      "epoch": 2.4013060255268623,
      "grad_norm": 0.8973240256309509,
      "learning_rate": 2.5986939744731374e-05,
      "loss": 0.8578,
      "step": 8090
    },
    {
      "epoch": 2.404274265360641,
      "grad_norm": 0.8081666827201843,
      "learning_rate": 2.5957257346393587e-05,
      "loss": 0.8546,
      "step": 8100
    },
    {
      "epoch": 2.4072425051944197,
      "grad_norm": 0.7552733421325684,
      "learning_rate": 2.5927574948055807e-05,
      "loss": 0.7966,
      "step": 8110
    },
    {
      "epoch": 2.4102107450281984,
      "grad_norm": 0.7936916947364807,
      "learning_rate": 2.5897892549718016e-05,
      "loss": 0.8343,
      "step": 8120
    },
    {
      "epoch": 2.4131789848619767,
      "grad_norm": 0.7276195287704468,
      "learning_rate": 2.5868210151380236e-05,
      "loss": 0.8703,
      "step": 8130
    },
    {
      "epoch": 2.4161472246957554,
      "grad_norm": 0.8015003204345703,
      "learning_rate": 2.5838527753042445e-05,
      "loss": 0.877,
      "step": 8140
    },
    {
      "epoch": 2.419115464529534,
      "grad_norm": 0.8002310991287231,
      "learning_rate": 2.5808845354704665e-05,
      "loss": 0.8319,
      "step": 8150
    },
    {
      "epoch": 2.4220837043633123,
      "grad_norm": 0.8283134698867798,
      "learning_rate": 2.5779162956366877e-05,
      "loss": 0.857,
      "step": 8160
    },
    {
      "epoch": 2.425051944197091,
      "grad_norm": 0.805871844291687,
      "learning_rate": 2.5749480558029087e-05,
      "loss": 0.8277,
      "step": 8170
    },
    {
      "epoch": 2.4280201840308697,
      "grad_norm": 0.7265508770942688,
      "learning_rate": 2.5719798159691306e-05,
      "loss": 0.8733,
      "step": 8180
    },
    {
      "epoch": 2.4309884238646484,
      "grad_norm": 0.850188672542572,
      "learning_rate": 2.5690115761353516e-05,
      "loss": 0.8698,
      "step": 8190
    },
    {
      "epoch": 2.4339566636984267,
      "grad_norm": 0.7770668864250183,
      "learning_rate": 2.5660433363015735e-05,
      "loss": 0.8177,
      "step": 8200
    },
    {
      "epoch": 2.4369249035322054,
      "grad_norm": 0.7985296249389648,
      "learning_rate": 2.5630750964677948e-05,
      "loss": 0.849,
      "step": 8210
    },
    {
      "epoch": 2.439893143365984,
      "grad_norm": 0.8427296876907349,
      "learning_rate": 2.5601068566340164e-05,
      "loss": 0.8884,
      "step": 8220
    },
    {
      "epoch": 2.4428613831997623,
      "grad_norm": 0.7303191423416138,
      "learning_rate": 2.5571386168002377e-05,
      "loss": 0.9021,
      "step": 8230
    },
    {
      "epoch": 2.445829623033541,
      "grad_norm": 0.9019272923469543,
      "learning_rate": 2.5541703769664586e-05,
      "loss": 0.8816,
      "step": 8240
    },
    {
      "epoch": 2.4487978628673197,
      "grad_norm": 0.7612438797950745,
      "learning_rate": 2.5512021371326806e-05,
      "loss": 0.87,
      "step": 8250
    },
    {
      "epoch": 2.4517661027010984,
      "grad_norm": 0.7570661902427673,
      "learning_rate": 2.548233897298902e-05,
      "loss": 0.8375,
      "step": 8260
    },
    {
      "epoch": 2.4547343425348767,
      "grad_norm": 0.7646659016609192,
      "learning_rate": 2.5452656574651235e-05,
      "loss": 0.8595,
      "step": 8270
    },
    {
      "epoch": 2.4577025823686554,
      "grad_norm": 0.7863264679908752,
      "learning_rate": 2.5422974176313448e-05,
      "loss": 0.8535,
      "step": 8280
    },
    {
      "epoch": 2.460670822202434,
      "grad_norm": 0.6919600963592529,
      "learning_rate": 2.5393291777975664e-05,
      "loss": 0.8125,
      "step": 8290
    },
    {
      "epoch": 2.4636390620362123,
      "grad_norm": 0.8025588393211365,
      "learning_rate": 2.5363609379637877e-05,
      "loss": 0.8589,
      "step": 8300
    },
    {
      "epoch": 2.466607301869991,
      "grad_norm": 0.7756131291389465,
      "learning_rate": 2.533392698130009e-05,
      "loss": 0.88,
      "step": 8310
    },
    {
      "epoch": 2.4695755417037697,
      "grad_norm": 0.6762662529945374,
      "learning_rate": 2.5304244582962306e-05,
      "loss": 0.8455,
      "step": 8320
    },
    {
      "epoch": 2.4725437815375484,
      "grad_norm": 0.8348343968391418,
      "learning_rate": 2.527456218462452e-05,
      "loss": 0.8593,
      "step": 8330
    },
    {
      "epoch": 2.4755120213713266,
      "grad_norm": 0.8249140381813049,
      "learning_rate": 2.5244879786286735e-05,
      "loss": 0.8238,
      "step": 8340
    },
    {
      "epoch": 2.4784802612051053,
      "grad_norm": 0.8179166316986084,
      "learning_rate": 2.5215197387948947e-05,
      "loss": 0.8501,
      "step": 8350
    },
    {
      "epoch": 2.481448501038884,
      "grad_norm": 0.8068183064460754,
      "learning_rate": 2.5185514989611164e-05,
      "loss": 0.8611,
      "step": 8360
    },
    {
      "epoch": 2.4844167408726623,
      "grad_norm": 0.8538073301315308,
      "learning_rate": 2.5155832591273376e-05,
      "loss": 0.8695,
      "step": 8370
    },
    {
      "epoch": 2.487384980706441,
      "grad_norm": 0.8070683479309082,
      "learning_rate": 2.512615019293559e-05,
      "loss": 0.8575,
      "step": 8380
    },
    {
      "epoch": 2.4903532205402197,
      "grad_norm": 0.7846097946166992,
      "learning_rate": 2.5096467794597805e-05,
      "loss": 0.8764,
      "step": 8390
    },
    {
      "epoch": 2.4933214603739984,
      "grad_norm": 0.794704020023346,
      "learning_rate": 2.5066785396260018e-05,
      "loss": 0.8097,
      "step": 8400
    },
    {
      "epoch": 2.4962897002077766,
      "grad_norm": 0.7580412030220032,
      "learning_rate": 2.5037102997922234e-05,
      "loss": 0.8092,
      "step": 8410
    },
    {
      "epoch": 2.4992579400415553,
      "grad_norm": 0.7811313271522522,
      "learning_rate": 2.5007420599584447e-05,
      "loss": 0.8548,
      "step": 8420
    },
    {
      "epoch": 2.502226179875334,
      "grad_norm": 0.7651128172874451,
      "learning_rate": 2.497773820124666e-05,
      "loss": 0.8582,
      "step": 8430
    },
    {
      "epoch": 2.5051944197091123,
      "grad_norm": 0.7565361857414246,
      "learning_rate": 2.4948055802908876e-05,
      "loss": 0.8397,
      "step": 8440
    },
    {
      "epoch": 2.508162659542891,
      "grad_norm": 0.8782424330711365,
      "learning_rate": 2.4918373404571092e-05,
      "loss": 0.8598,
      "step": 8450
    },
    {
      "epoch": 2.5111308993766697,
      "grad_norm": 0.8438358306884766,
      "learning_rate": 2.4888691006233305e-05,
      "loss": 0.8433,
      "step": 8460
    },
    {
      "epoch": 2.5140991392104484,
      "grad_norm": 0.839470386505127,
      "learning_rate": 2.485900860789552e-05,
      "loss": 0.8754,
      "step": 8470
    },
    {
      "epoch": 2.5170673790442266,
      "grad_norm": 0.7891715168952942,
      "learning_rate": 2.482932620955773e-05,
      "loss": 0.8744,
      "step": 8480
    },
    {
      "epoch": 2.5200356188780053,
      "grad_norm": 0.8311156034469604,
      "learning_rate": 2.4799643811219947e-05,
      "loss": 0.8741,
      "step": 8490
    },
    {
      "epoch": 2.523003858711784,
      "grad_norm": 0.7338982224464417,
      "learning_rate": 2.4769961412882163e-05,
      "loss": 0.8549,
      "step": 8500
    },
    {
      "epoch": 2.5259720985455623,
      "grad_norm": 0.739483118057251,
      "learning_rate": 2.4740279014544376e-05,
      "loss": 0.8458,
      "step": 8510
    },
    {
      "epoch": 2.528940338379341,
      "grad_norm": 0.7444939613342285,
      "learning_rate": 2.4710596616206592e-05,
      "loss": 0.8497,
      "step": 8520
    },
    {
      "epoch": 2.5319085782131197,
      "grad_norm": 0.8406269550323486,
      "learning_rate": 2.4680914217868805e-05,
      "loss": 0.8514,
      "step": 8530
    },
    {
      "epoch": 2.5348768180468984,
      "grad_norm": 0.7409178614616394,
      "learning_rate": 2.465123181953102e-05,
      "loss": 0.8532,
      "step": 8540
    },
    {
      "epoch": 2.5378450578806766,
      "grad_norm": 0.8377125263214111,
      "learning_rate": 2.4621549421193234e-05,
      "loss": 0.8559,
      "step": 8550
    },
    {
      "epoch": 2.5408132977144553,
      "grad_norm": 0.9167962670326233,
      "learning_rate": 2.4591867022855447e-05,
      "loss": 0.8972,
      "step": 8560
    },
    {
      "epoch": 2.543781537548234,
      "grad_norm": 0.6903221607208252,
      "learning_rate": 2.4562184624517663e-05,
      "loss": 0.8719,
      "step": 8570
    },
    {
      "epoch": 2.5467497773820122,
      "grad_norm": 0.7576964497566223,
      "learning_rate": 2.4532502226179875e-05,
      "loss": 0.8451,
      "step": 8580
    },
    {
      "epoch": 2.549718017215791,
      "grad_norm": 0.736854612827301,
      "learning_rate": 2.450281982784209e-05,
      "loss": 0.8311,
      "step": 8590
    },
    {
      "epoch": 2.5526862570495696,
      "grad_norm": 0.7180027365684509,
      "learning_rate": 2.4473137429504304e-05,
      "loss": 0.8302,
      "step": 8600
    },
    {
      "epoch": 2.5556544968833483,
      "grad_norm": 0.7868888974189758,
      "learning_rate": 2.444345503116652e-05,
      "loss": 0.8455,
      "step": 8610
    },
    {
      "epoch": 2.5586227367171266,
      "grad_norm": 0.8172736167907715,
      "learning_rate": 2.4413772632828733e-05,
      "loss": 0.8489,
      "step": 8620
    },
    {
      "epoch": 2.5615909765509053,
      "grad_norm": 0.8119463920593262,
      "learning_rate": 2.4384090234490946e-05,
      "loss": 0.8814,
      "step": 8630
    },
    {
      "epoch": 2.564559216384684,
      "grad_norm": 0.7911559343338013,
      "learning_rate": 2.4354407836153162e-05,
      "loss": 0.8612,
      "step": 8640
    },
    {
      "epoch": 2.5675274562184622,
      "grad_norm": 0.8564543128013611,
      "learning_rate": 2.4324725437815375e-05,
      "loss": 0.8515,
      "step": 8650
    },
    {
      "epoch": 2.570495696052241,
      "grad_norm": 0.7224608063697815,
      "learning_rate": 2.429504303947759e-05,
      "loss": 0.8362,
      "step": 8660
    },
    {
      "epoch": 2.5734639358860196,
      "grad_norm": 0.8463042974472046,
      "learning_rate": 2.4265360641139808e-05,
      "loss": 0.8347,
      "step": 8670
    },
    {
      "epoch": 2.5764321757197983,
      "grad_norm": 0.7602813243865967,
      "learning_rate": 2.423567824280202e-05,
      "loss": 0.8597,
      "step": 8680
    },
    {
      "epoch": 2.5794004155535766,
      "grad_norm": 0.7302565574645996,
      "learning_rate": 2.4205995844464233e-05,
      "loss": 0.8879,
      "step": 8690
    },
    {
      "epoch": 2.5823686553873553,
      "grad_norm": 0.7882288098335266,
      "learning_rate": 2.4176313446126446e-05,
      "loss": 0.8381,
      "step": 8700
    },
    {
      "epoch": 2.585336895221134,
      "grad_norm": 0.7467893362045288,
      "learning_rate": 2.4146631047788662e-05,
      "loss": 0.8754,
      "step": 8710
    },
    {
      "epoch": 2.588305135054912,
      "grad_norm": 0.8261969685554504,
      "learning_rate": 2.4116948649450878e-05,
      "loss": 0.8302,
      "step": 8720
    },
    {
      "epoch": 2.591273374888691,
      "grad_norm": 0.7031750082969666,
      "learning_rate": 2.408726625111309e-05,
      "loss": 0.88,
      "step": 8730
    },
    {
      "epoch": 2.5942416147224696,
      "grad_norm": 0.9632667303085327,
      "learning_rate": 2.4057583852775307e-05,
      "loss": 0.8479,
      "step": 8740
    },
    {
      "epoch": 2.5972098545562483,
      "grad_norm": 0.737890899181366,
      "learning_rate": 2.402790145443752e-05,
      "loss": 0.843,
      "step": 8750
    },
    {
      "epoch": 2.6001780943900266,
      "grad_norm": 0.7835996747016907,
      "learning_rate": 2.3998219056099733e-05,
      "loss": 0.8806,
      "step": 8760
    },
    {
      "epoch": 2.6031463342238053,
      "grad_norm": 0.7798622250556946,
      "learning_rate": 2.396853665776195e-05,
      "loss": 0.8305,
      "step": 8770
    },
    {
      "epoch": 2.606114574057584,
      "grad_norm": 0.7724559903144836,
      "learning_rate": 2.3938854259424162e-05,
      "loss": 0.8446,
      "step": 8780
    },
    {
      "epoch": 2.609082813891362,
      "grad_norm": 0.8460879325866699,
      "learning_rate": 2.3909171861086378e-05,
      "loss": 0.8879,
      "step": 8790
    },
    {
      "epoch": 2.612051053725141,
      "grad_norm": 0.8908317685127258,
      "learning_rate": 2.387948946274859e-05,
      "loss": 0.8368,
      "step": 8800
    },
    {
      "epoch": 2.6150192935589196,
      "grad_norm": 0.8732563853263855,
      "learning_rate": 2.3849807064410807e-05,
      "loss": 0.8699,
      "step": 8810
    },
    {
      "epoch": 2.6179875333926983,
      "grad_norm": 0.8612620830535889,
      "learning_rate": 2.382012466607302e-05,
      "loss": 0.8259,
      "step": 8820
    },
    {
      "epoch": 2.6209557732264765,
      "grad_norm": 0.7714670896530151,
      "learning_rate": 2.3790442267735232e-05,
      "loss": 0.8301,
      "step": 8830
    },
    {
      "epoch": 2.6239240130602552,
      "grad_norm": 0.7954014539718628,
      "learning_rate": 2.376075986939745e-05,
      "loss": 0.8464,
      "step": 8840
    },
    {
      "epoch": 2.626892252894034,
      "grad_norm": 0.7666088342666626,
      "learning_rate": 2.373107747105966e-05,
      "loss": 0.8164,
      "step": 8850
    },
    {
      "epoch": 2.629860492727812,
      "grad_norm": 0.7557575702667236,
      "learning_rate": 2.3701395072721878e-05,
      "loss": 0.8564,
      "step": 8860
    },
    {
      "epoch": 2.632828732561591,
      "grad_norm": 0.9145186543464661,
      "learning_rate": 2.367171267438409e-05,
      "loss": 0.8065,
      "step": 8870
    },
    {
      "epoch": 2.6357969723953696,
      "grad_norm": 0.8163389563560486,
      "learning_rate": 2.3642030276046307e-05,
      "loss": 0.8479,
      "step": 8880
    },
    {
      "epoch": 2.6387652122291483,
      "grad_norm": 0.8326930403709412,
      "learning_rate": 2.3612347877708523e-05,
      "loss": 0.8564,
      "step": 8890
    },
    {
      "epoch": 2.6417334520629265,
      "grad_norm": 0.824146568775177,
      "learning_rate": 2.3582665479370732e-05,
      "loss": 0.8134,
      "step": 8900
    },
    {
      "epoch": 2.6447016918967052,
      "grad_norm": 0.7646833658218384,
      "learning_rate": 2.3552983081032948e-05,
      "loss": 0.8632,
      "step": 8910
    },
    {
      "epoch": 2.647669931730484,
      "grad_norm": 0.7801716923713684,
      "learning_rate": 2.352330068269516e-05,
      "loss": 0.8241,
      "step": 8920
    },
    {
      "epoch": 2.650638171564262,
      "grad_norm": 0.817964494228363,
      "learning_rate": 2.3493618284357377e-05,
      "loss": 0.8479,
      "step": 8930
    },
    {
      "epoch": 2.653606411398041,
      "grad_norm": 0.7799892425537109,
      "learning_rate": 2.3463935886019593e-05,
      "loss": 0.8611,
      "step": 8940
    },
    {
      "epoch": 2.6565746512318196,
      "grad_norm": 0.9056032299995422,
      "learning_rate": 2.3434253487681806e-05,
      "loss": 0.8473,
      "step": 8950
    },
    {
      "epoch": 2.6595428910655983,
      "grad_norm": 0.8326584696769714,
      "learning_rate": 2.3404571089344022e-05,
      "loss": 0.8504,
      "step": 8960
    },
    {
      "epoch": 2.6625111308993765,
      "grad_norm": 0.9176719188690186,
      "learning_rate": 2.3374888691006232e-05,
      "loss": 0.8573,
      "step": 8970
    },
    {
      "epoch": 2.665479370733155,
      "grad_norm": 0.9030516147613525,
      "learning_rate": 2.3345206292668448e-05,
      "loss": 0.8399,
      "step": 8980
    },
    {
      "epoch": 2.668447610566934,
      "grad_norm": 0.7815122008323669,
      "learning_rate": 2.3315523894330664e-05,
      "loss": 0.814,
      "step": 8990
    },
    {
      "epoch": 2.671415850400712,
      "grad_norm": 0.8600183129310608,
      "learning_rate": 2.3285841495992877e-05,
      "loss": 0.8664,
      "step": 9000
    },
    {
      "epoch": 2.674384090234491,
      "grad_norm": 0.7203440070152283,
      "learning_rate": 2.3256159097655093e-05,
      "loss": 0.8681,
      "step": 9010
    },
    {
      "epoch": 2.6773523300682696,
      "grad_norm": 0.8177782893180847,
      "learning_rate": 2.3226476699317306e-05,
      "loss": 0.8515,
      "step": 9020
    },
    {
      "epoch": 2.6803205699020483,
      "grad_norm": 0.8409428596496582,
      "learning_rate": 2.3196794300979522e-05,
      "loss": 0.8566,
      "step": 9030
    },
    {
      "epoch": 2.6832888097358265,
      "grad_norm": 0.8034857511520386,
      "learning_rate": 2.3167111902641735e-05,
      "loss": 0.8454,
      "step": 9040
    },
    {
      "epoch": 2.686257049569605,
      "grad_norm": 0.773566484451294,
      "learning_rate": 2.3137429504303948e-05,
      "loss": 0.8239,
      "step": 9050
    },
    {
      "epoch": 2.689225289403384,
      "grad_norm": 0.8261438608169556,
      "learning_rate": 2.3107747105966164e-05,
      "loss": 0.839,
      "step": 9060
    },
    {
      "epoch": 2.692193529237162,
      "grad_norm": 0.8331723809242249,
      "learning_rate": 2.3078064707628377e-05,
      "loss": 0.8509,
      "step": 9070
    },
    {
      "epoch": 2.695161769070941,
      "grad_norm": 0.8286955952644348,
      "learning_rate": 2.3048382309290593e-05,
      "loss": 0.8733,
      "step": 9080
    },
    {
      "epoch": 2.6981300089047195,
      "grad_norm": 0.8668794631958008,
      "learning_rate": 2.3018699910952806e-05,
      "loss": 0.8813,
      "step": 9090
    },
    {
      "epoch": 2.7010982487384982,
      "grad_norm": 0.8242696523666382,
      "learning_rate": 2.2989017512615022e-05,
      "loss": 0.8119,
      "step": 9100
    },
    {
      "epoch": 2.7040664885722765,
      "grad_norm": 0.7457850575447083,
      "learning_rate": 2.2959335114277235e-05,
      "loss": 0.8753,
      "step": 9110
    },
    {
      "epoch": 2.707034728406055,
      "grad_norm": 0.8100054860115051,
      "learning_rate": 2.2929652715939447e-05,
      "loss": 0.8304,
      "step": 9120
    },
    {
      "epoch": 2.710002968239834,
      "grad_norm": 0.8167350888252258,
      "learning_rate": 2.2899970317601664e-05,
      "loss": 0.8306,
      "step": 9130
    },
    {
      "epoch": 2.712971208073612,
      "grad_norm": 0.810926079750061,
      "learning_rate": 2.2870287919263876e-05,
      "loss": 0.8541,
      "step": 9140
    },
    {
      "epoch": 2.715939447907391,
      "grad_norm": 0.8000773787498474,
      "learning_rate": 2.2840605520926092e-05,
      "loss": 0.8577,
      "step": 9150
    },
    {
      "epoch": 2.7189076877411695,
      "grad_norm": 0.9856551289558411,
      "learning_rate": 2.2810923122588305e-05,
      "loss": 0.8106,
      "step": 9160
    },
    {
      "epoch": 2.7218759275749482,
      "grad_norm": 0.7883819341659546,
      "learning_rate": 2.278124072425052e-05,
      "loss": 0.8269,
      "step": 9170
    },
    {
      "epoch": 2.7248441674087265,
      "grad_norm": 0.7424699068069458,
      "learning_rate": 2.2751558325912734e-05,
      "loss": 0.8377,
      "step": 9180
    },
    {
      "epoch": 2.727812407242505,
      "grad_norm": 0.8047650456428528,
      "learning_rate": 2.2721875927574947e-05,
      "loss": 0.8431,
      "step": 9190
    },
    {
      "epoch": 2.730780647076284,
      "grad_norm": 0.841957688331604,
      "learning_rate": 2.2692193529237163e-05,
      "loss": 0.9311,
      "step": 9200
    },
    {
      "epoch": 2.733748886910062,
      "grad_norm": 0.7910162806510925,
      "learning_rate": 2.2662511130899376e-05,
      "loss": 0.817,
      "step": 9210
    },
    {
      "epoch": 2.736717126743841,
      "grad_norm": 0.8855330944061279,
      "learning_rate": 2.2632828732561592e-05,
      "loss": 0.8375,
      "step": 9220
    },
    {
      "epoch": 2.7396853665776195,
      "grad_norm": 0.8575757145881653,
      "learning_rate": 2.260314633422381e-05,
      "loss": 0.8518,
      "step": 9230
    },
    {
      "epoch": 2.742653606411398,
      "grad_norm": 0.7841692566871643,
      "learning_rate": 2.257346393588602e-05,
      "loss": 0.8336,
      "step": 9240
    },
    {
      "epoch": 2.7456218462451765,
      "grad_norm": 0.807435154914856,
      "learning_rate": 2.2543781537548234e-05,
      "loss": 0.8165,
      "step": 9250
    },
    {
      "epoch": 2.748590086078955,
      "grad_norm": 0.8038501739501953,
      "learning_rate": 2.2514099139210447e-05,
      "loss": 0.8451,
      "step": 9260
    },
    {
      "epoch": 2.751558325912734,
      "grad_norm": 0.7744222283363342,
      "learning_rate": 2.2484416740872663e-05,
      "loss": 0.899,
      "step": 9270
    },
    {
      "epoch": 2.754526565746512,
      "grad_norm": 0.9123599529266357,
      "learning_rate": 2.245473434253488e-05,
      "loss": 0.8348,
      "step": 9280
    },
    {
      "epoch": 2.757494805580291,
      "grad_norm": 0.7251855134963989,
      "learning_rate": 2.2425051944197092e-05,
      "loss": 0.8503,
      "step": 9290
    },
    {
      "epoch": 2.7604630454140695,
      "grad_norm": 0.7562434673309326,
      "learning_rate": 2.2395369545859308e-05,
      "loss": 0.8369,
      "step": 9300
    },
    {
      "epoch": 2.763431285247848,
      "grad_norm": 0.853834867477417,
      "learning_rate": 2.236568714752152e-05,
      "loss": 0.8664,
      "step": 9310
    },
    {
      "epoch": 2.7663995250816265,
      "grad_norm": 0.7551343441009521,
      "learning_rate": 2.2336004749183734e-05,
      "loss": 0.8375,
      "step": 9320
    },
    {
      "epoch": 2.769367764915405,
      "grad_norm": 0.8466640710830688,
      "learning_rate": 2.2309290590679726e-05,
      "loss": 0.8273,
      "step": 9330
    },
    {
      "epoch": 2.772336004749184,
      "grad_norm": 0.7854313850402832,
      "learning_rate": 2.2279608192341942e-05,
      "loss": 0.8448,
      "step": 9340
    },
    {
      "epoch": 2.775304244582962,
      "grad_norm": 0.7493451237678528,
      "learning_rate": 2.2249925794004155e-05,
      "loss": 0.85,
      "step": 9350
    },
    {
      "epoch": 2.778272484416741,
      "grad_norm": 0.8265935778617859,
      "learning_rate": 2.222024339566637e-05,
      "loss": 0.8185,
      "step": 9360
    },
    {
      "epoch": 2.7812407242505195,
      "grad_norm": 0.8303396701812744,
      "learning_rate": 2.2190560997328584e-05,
      "loss": 0.8358,
      "step": 9370
    },
    {
      "epoch": 2.784208964084298,
      "grad_norm": 0.8008495569229126,
      "learning_rate": 2.21608785989908e-05,
      "loss": 0.8333,
      "step": 9380
    },
    {
      "epoch": 2.7871772039180764,
      "grad_norm": 0.7543458938598633,
      "learning_rate": 2.2131196200653016e-05,
      "loss": 0.84,
      "step": 9390
    },
    {
      "epoch": 2.790145443751855,
      "grad_norm": 0.7685660719871521,
      "learning_rate": 2.2101513802315226e-05,
      "loss": 0.8759,
      "step": 9400
    },
    {
      "epoch": 2.793113683585634,
      "grad_norm": 0.772219717502594,
      "learning_rate": 2.2071831403977442e-05,
      "loss": 0.8415,
      "step": 9410
    },
    {
      "epoch": 2.796081923419412,
      "grad_norm": 0.8281093239784241,
      "learning_rate": 2.2042149005639655e-05,
      "loss": 0.8479,
      "step": 9420
    },
    {
      "epoch": 2.799050163253191,
      "grad_norm": 0.8043393492698669,
      "learning_rate": 2.201246660730187e-05,
      "loss": 0.8703,
      "step": 9430
    },
    {
      "epoch": 2.8020184030869695,
      "grad_norm": 0.7823016047477722,
      "learning_rate": 2.1982784208964087e-05,
      "loss": 0.8415,
      "step": 9440
    },
    {
      "epoch": 2.804986642920748,
      "grad_norm": 0.9165988564491272,
      "learning_rate": 2.19531018106263e-05,
      "loss": 0.9038,
      "step": 9450
    },
    {
      "epoch": 2.8079548827545264,
      "grad_norm": 0.8192057609558105,
      "learning_rate": 2.1923419412288516e-05,
      "loss": 0.859,
      "step": 9460
    },
    {
      "epoch": 2.810923122588305,
      "grad_norm": 0.767427384853363,
      "learning_rate": 2.1893737013950725e-05,
      "loss": 0.8932,
      "step": 9470
    },
    {
      "epoch": 2.813891362422084,
      "grad_norm": 0.7586916089057922,
      "learning_rate": 2.186405461561294e-05,
      "loss": 0.8523,
      "step": 9480
    },
    {
      "epoch": 2.816859602255862,
      "grad_norm": 0.7692698836326599,
      "learning_rate": 2.1834372217275158e-05,
      "loss": 0.8701,
      "step": 9490
    },
    {
      "epoch": 2.8198278420896408,
      "grad_norm": 0.8118112087249756,
      "learning_rate": 2.180468981893737e-05,
      "loss": 0.8878,
      "step": 9500
    },
    {
      "epoch": 2.8227960819234195,
      "grad_norm": 0.8335282206535339,
      "learning_rate": 2.1775007420599587e-05,
      "loss": 0.8679,
      "step": 9510
    },
    {
      "epoch": 2.825764321757198,
      "grad_norm": 0.8098772168159485,
      "learning_rate": 2.17453250222618e-05,
      "loss": 0.8468,
      "step": 9520
    },
    {
      "epoch": 2.8287325615909764,
      "grad_norm": 0.7913337349891663,
      "learning_rate": 2.1715642623924016e-05,
      "loss": 0.8563,
      "step": 9530
    },
    {
      "epoch": 2.831700801424755,
      "grad_norm": 0.7871443033218384,
      "learning_rate": 2.168596022558623e-05,
      "loss": 0.8474,
      "step": 9540
    },
    {
      "epoch": 2.834669041258534,
      "grad_norm": 0.7683349251747131,
      "learning_rate": 2.165627782724844e-05,
      "loss": 0.8429,
      "step": 9550
    },
    {
      "epoch": 2.837637281092312,
      "grad_norm": 0.761978030204773,
      "learning_rate": 2.1626595428910658e-05,
      "loss": 0.8989,
      "step": 9560
    },
    {
      "epoch": 2.8406055209260908,
      "grad_norm": 0.7867503762245178,
      "learning_rate": 2.159691303057287e-05,
      "loss": 0.8187,
      "step": 9570
    },
    {
      "epoch": 2.8435737607598695,
      "grad_norm": 0.8149718046188354,
      "learning_rate": 2.1567230632235086e-05,
      "loss": 0.837,
      "step": 9580
    },
    {
      "epoch": 2.846542000593648,
      "grad_norm": 0.7846998572349548,
      "learning_rate": 2.15375482338973e-05,
      "loss": 0.8346,
      "step": 9590
    },
    {
      "epoch": 2.8495102404274264,
      "grad_norm": 0.7807599902153015,
      "learning_rate": 2.1507865835559515e-05,
      "loss": 0.8718,
      "step": 9600
    },
    {
      "epoch": 2.852478480261205,
      "grad_norm": 0.7701714634895325,
      "learning_rate": 2.1478183437221728e-05,
      "loss": 0.8343,
      "step": 9610
    },
    {
      "epoch": 2.855446720094984,
      "grad_norm": 0.9326925873756409,
      "learning_rate": 2.144850103888394e-05,
      "loss": 0.8588,
      "step": 9620
    },
    {
      "epoch": 2.858414959928762,
      "grad_norm": 0.8950006365776062,
      "learning_rate": 2.1418818640546157e-05,
      "loss": 0.8679,
      "step": 9630
    },
    {
      "epoch": 2.8613831997625407,
      "grad_norm": 0.8486356735229492,
      "learning_rate": 2.138913624220837e-05,
      "loss": 0.864,
      "step": 9640
    },
    {
      "epoch": 2.8643514395963194,
      "grad_norm": 0.8299232125282288,
      "learning_rate": 2.1359453843870586e-05,
      "loss": 0.8436,
      "step": 9650
    },
    {
      "epoch": 2.867319679430098,
      "grad_norm": 0.7777180671691895,
      "learning_rate": 2.1329771445532802e-05,
      "loss": 0.8644,
      "step": 9660
    },
    {
      "epoch": 2.8702879192638764,
      "grad_norm": 0.8722212910652161,
      "learning_rate": 2.1300089047195015e-05,
      "loss": 0.8849,
      "step": 9670
    },
    {
      "epoch": 2.873256159097655,
      "grad_norm": 0.7691166996955872,
      "learning_rate": 2.1270406648857228e-05,
      "loss": 0.9117,
      "step": 9680
    },
    {
      "epoch": 2.876224398931434,
      "grad_norm": 0.8059368133544922,
      "learning_rate": 2.124072425051944e-05,
      "loss": 0.8446,
      "step": 9690
    },
    {
      "epoch": 2.879192638765212,
      "grad_norm": 0.8385917544364929,
      "learning_rate": 2.1211041852181657e-05,
      "loss": 0.8455,
      "step": 9700
    },
    {
      "epoch": 2.8821608785989907,
      "grad_norm": 0.9432308077812195,
      "learning_rate": 2.1181359453843873e-05,
      "loss": 0.8609,
      "step": 9710
    },
    {
      "epoch": 2.8851291184327694,
      "grad_norm": 0.7637094259262085,
      "learning_rate": 2.1151677055506086e-05,
      "loss": 0.8369,
      "step": 9720
    },
    {
      "epoch": 2.888097358266548,
      "grad_norm": 0.8425596952438354,
      "learning_rate": 2.1121994657168302e-05,
      "loss": 0.8171,
      "step": 9730
    },
    {
      "epoch": 2.8910655981003264,
      "grad_norm": 0.7895944714546204,
      "learning_rate": 2.1092312258830515e-05,
      "loss": 0.8657,
      "step": 9740
    },
    {
      "epoch": 2.894033837934105,
      "grad_norm": 0.9404763579368591,
      "learning_rate": 2.1062629860492728e-05,
      "loss": 0.8219,
      "step": 9750
    },
    {
      "epoch": 2.8970020777678838,
      "grad_norm": 0.7366514801979065,
      "learning_rate": 2.1032947462154944e-05,
      "loss": 0.8677,
      "step": 9760
    },
    {
      "epoch": 2.899970317601662,
      "grad_norm": 0.8323311805725098,
      "learning_rate": 2.1003265063817157e-05,
      "loss": 0.841,
      "step": 9770
    },
    {
      "epoch": 2.9029385574354407,
      "grad_norm": 0.8057579398155212,
      "learning_rate": 2.0973582665479373e-05,
      "loss": 0.8571,
      "step": 9780
    },
    {
      "epoch": 2.9059067972692194,
      "grad_norm": 0.7631582617759705,
      "learning_rate": 2.0943900267141586e-05,
      "loss": 0.7837,
      "step": 9790
    },
    {
      "epoch": 2.908875037102998,
      "grad_norm": 0.95359206199646,
      "learning_rate": 2.09142178688038e-05,
      "loss": 0.8971,
      "step": 9800
    },
    {
      "epoch": 2.9118432769367764,
      "grad_norm": 0.746528685092926,
      "learning_rate": 2.0884535470466014e-05,
      "loss": 0.8293,
      "step": 9810
    },
    {
      "epoch": 2.914811516770555,
      "grad_norm": 0.787821888923645,
      "learning_rate": 2.0854853072128227e-05,
      "loss": 0.8613,
      "step": 9820
    },
    {
      "epoch": 2.9177797566043338,
      "grad_norm": 0.7103245854377747,
      "learning_rate": 2.0825170673790443e-05,
      "loss": 0.8385,
      "step": 9830
    },
    {
      "epoch": 2.920747996438112,
      "grad_norm": 0.8258326053619385,
      "learning_rate": 2.0795488275452656e-05,
      "loss": 0.8705,
      "step": 9840
    },
    {
      "epoch": 2.9237162362718907,
      "grad_norm": 0.8143662214279175,
      "learning_rate": 2.0765805877114872e-05,
      "loss": 0.8274,
      "step": 9850
    },
    {
      "epoch": 2.9266844761056694,
      "grad_norm": 0.8519349694252014,
      "learning_rate": 2.0736123478777085e-05,
      "loss": 0.8661,
      "step": 9860
    },
    {
      "epoch": 2.929652715939448,
      "grad_norm": 0.8383392095565796,
      "learning_rate": 2.07064410804393e-05,
      "loss": 0.8704,
      "step": 9870
    },
    {
      "epoch": 2.9326209557732263,
      "grad_norm": 0.8166011571884155,
      "learning_rate": 2.0676758682101514e-05,
      "loss": 0.8754,
      "step": 9880
    },
    {
      "epoch": 2.935589195607005,
      "grad_norm": 0.7948212623596191,
      "learning_rate": 2.0647076283763727e-05,
      "loss": 0.8269,
      "step": 9890
    },
    {
      "epoch": 2.9385574354407837,
      "grad_norm": 0.8796318769454956,
      "learning_rate": 2.0617393885425943e-05,
      "loss": 0.8424,
      "step": 9900
    },
    {
      "epoch": 2.941525675274562,
      "grad_norm": 0.8400830030441284,
      "learning_rate": 2.0587711487088156e-05,
      "loss": 0.8153,
      "step": 9910
    },
    {
      "epoch": 2.9444939151083407,
      "grad_norm": 0.8363904356956482,
      "learning_rate": 2.0558029088750372e-05,
      "loss": 0.8343,
      "step": 9920
    },
    {
      "epoch": 2.9474621549421194,
      "grad_norm": 0.7671076655387878,
      "learning_rate": 2.0528346690412585e-05,
      "loss": 0.8258,
      "step": 9930
    },
    {
      "epoch": 2.950430394775898,
      "grad_norm": 0.7828432321548462,
      "learning_rate": 2.04986642920748e-05,
      "loss": 0.8739,
      "step": 9940
    },
    {
      "epoch": 2.9533986346096763,
      "grad_norm": 0.8337661623954773,
      "learning_rate": 2.0468981893737017e-05,
      "loss": 0.8387,
      "step": 9950
    },
    {
      "epoch": 2.956366874443455,
      "grad_norm": 0.9032126665115356,
      "learning_rate": 2.043929949539923e-05,
      "loss": 0.8669,
      "step": 9960
    },
    {
      "epoch": 2.9593351142772337,
      "grad_norm": 0.813811719417572,
      "learning_rate": 2.0409617097061443e-05,
      "loss": 0.8469,
      "step": 9970
    },
    {
      "epoch": 2.962303354111012,
      "grad_norm": 0.816363513469696,
      "learning_rate": 2.0379934698723656e-05,
      "loss": 0.8226,
      "step": 9980
    },
    {
      "epoch": 2.9652715939447907,
      "grad_norm": 0.7742164134979248,
      "learning_rate": 2.0350252300385872e-05,
      "loss": 0.8795,
      "step": 9990
    },
    {
      "epoch": 2.9682398337785694,
      "grad_norm": 0.7590778470039368,
      "learning_rate": 2.0320569902048088e-05,
      "loss": 0.8313,
      "step": 10000
    },
    {
      "epoch": 2.971208073612348,
      "grad_norm": 0.8371145129203796,
      "learning_rate": 2.02908875037103e-05,
      "loss": 0.8143,
      "step": 10010
    },
    {
      "epoch": 2.9741763134461263,
      "grad_norm": 0.7885297536849976,
      "learning_rate": 2.0261205105372517e-05,
      "loss": 0.8539,
      "step": 10020
    },
    {
      "epoch": 2.977144553279905,
      "grad_norm": 0.8193269968032837,
      "learning_rate": 2.023152270703473e-05,
      "loss": 0.8781,
      "step": 10030
    },
    {
      "epoch": 2.9801127931136837,
      "grad_norm": 0.8111793398857117,
      "learning_rate": 2.0201840308696942e-05,
      "loss": 0.8442,
      "step": 10040
    },
    {
      "epoch": 2.983081032947462,
      "grad_norm": 0.820756196975708,
      "learning_rate": 2.017215791035916e-05,
      "loss": 0.8498,
      "step": 10050
    },
    {
      "epoch": 2.9860492727812407,
      "grad_norm": 0.8105862736701965,
      "learning_rate": 2.014247551202137e-05,
      "loss": 0.8619,
      "step": 10060
    },
    {
      "epoch": 2.9890175126150194,
      "grad_norm": 0.8605939149856567,
      "learning_rate": 2.0112793113683588e-05,
      "loss": 0.8739,
      "step": 10070
    },
    {
      "epoch": 2.991985752448798,
      "grad_norm": 0.7815289497375488,
      "learning_rate": 2.00831107153458e-05,
      "loss": 0.811,
      "step": 10080
    },
    {
      "epoch": 2.9949539922825763,
      "grad_norm": 0.8077415227890015,
      "learning_rate": 2.0053428317008017e-05,
      "loss": 0.8672,
      "step": 10090
    },
    {
      "epoch": 2.997922232116355,
      "grad_norm": 0.8030134439468384,
      "learning_rate": 2.002374591867023e-05,
      "loss": 0.8597,
      "step": 10100
    },
    {
      "epoch": 3.0008904719501337,
      "grad_norm": 0.748300313949585,
      "learning_rate": 1.9994063520332442e-05,
      "loss": 0.8273,
      "step": 10110
    },
    {
      "epoch": 3.003858711783912,
      "grad_norm": 0.8023651838302612,
      "learning_rate": 1.996438112199466e-05,
      "loss": 0.7977,
      "step": 10120
    },
    {
      "epoch": 3.0068269516176906,
      "grad_norm": 0.7826663851737976,
      "learning_rate": 1.993469872365687e-05,
      "loss": 0.7907,
      "step": 10130
    },
    {
      "epoch": 3.0097951914514693,
      "grad_norm": 0.8399567008018494,
      "learning_rate": 1.9905016325319087e-05,
      "loss": 0.8036,
      "step": 10140
    },
    {
      "epoch": 3.012763431285248,
      "grad_norm": 0.9962929487228394,
      "learning_rate": 1.98753339269813e-05,
      "loss": 0.7947,
      "step": 10150
    },
    {
      "epoch": 3.0157316711190263,
      "grad_norm": 0.7716733813285828,
      "learning_rate": 1.9845651528643516e-05,
      "loss": 0.8097,
      "step": 10160
    },
    {
      "epoch": 3.018699910952805,
      "grad_norm": 0.854088306427002,
      "learning_rate": 1.9815969130305732e-05,
      "loss": 0.8075,
      "step": 10170
    },
    {
      "epoch": 3.0216681507865837,
      "grad_norm": 0.8283464312553406,
      "learning_rate": 1.9786286731967942e-05,
      "loss": 0.7759,
      "step": 10180
    },
    {
      "epoch": 3.024636390620362,
      "grad_norm": 0.8063070774078369,
      "learning_rate": 1.9756604333630158e-05,
      "loss": 0.807,
      "step": 10190
    },
    {
      "epoch": 3.0276046304541406,
      "grad_norm": 0.8599423170089722,
      "learning_rate": 1.972692193529237e-05,
      "loss": 0.7849,
      "step": 10200
    },
    {
      "epoch": 3.0305728702879193,
      "grad_norm": 0.8788694739341736,
      "learning_rate": 1.9697239536954587e-05,
      "loss": 0.8065,
      "step": 10210
    },
    {
      "epoch": 3.033541110121698,
      "grad_norm": 0.7816437482833862,
      "learning_rate": 1.9667557138616803e-05,
      "loss": 0.8264,
      "step": 10220
    },
    {
      "epoch": 3.0365093499554763,
      "grad_norm": 0.9156720638275146,
      "learning_rate": 1.9637874740279016e-05,
      "loss": 0.8212,
      "step": 10230
    },
    {
      "epoch": 3.039477589789255,
      "grad_norm": 0.8273279070854187,
      "learning_rate": 1.9608192341941232e-05,
      "loss": 0.798,
      "step": 10240
    },
    {
      "epoch": 3.0424458296230337,
      "grad_norm": 0.8687625527381897,
      "learning_rate": 1.957850994360344e-05,
      "loss": 0.7973,
      "step": 10250
    },
    {
      "epoch": 3.045414069456812,
      "grad_norm": 0.8279813528060913,
      "learning_rate": 1.9548827545265658e-05,
      "loss": 0.7799,
      "step": 10260
    },
    {
      "epoch": 3.0483823092905906,
      "grad_norm": 0.8048006892204285,
      "learning_rate": 1.9519145146927874e-05,
      "loss": 0.8009,
      "step": 10270
    },
    {
      "epoch": 3.0513505491243693,
      "grad_norm": 0.8323532938957214,
      "learning_rate": 1.9489462748590087e-05,
      "loss": 0.8188,
      "step": 10280
    },
    {
      "epoch": 3.054318788958148,
      "grad_norm": 0.8161619901657104,
      "learning_rate": 1.9459780350252303e-05,
      "loss": 0.7741,
      "step": 10290
    },
    {
      "epoch": 3.0572870287919263,
      "grad_norm": 0.8610167503356934,
      "learning_rate": 1.9430097951914516e-05,
      "loss": 0.8311,
      "step": 10300
    },
    {
      "epoch": 3.060255268625705,
      "grad_norm": 0.7956073880195618,
      "learning_rate": 1.9400415553576732e-05,
      "loss": 0.7758,
      "step": 10310
    },
    {
      "epoch": 3.0632235084594837,
      "grad_norm": 0.8731258511543274,
      "learning_rate": 1.9370733155238945e-05,
      "loss": 0.8096,
      "step": 10320
    },
    {
      "epoch": 3.066191748293262,
      "grad_norm": 0.8595923781394958,
      "learning_rate": 1.9341050756901157e-05,
      "loss": 0.8198,
      "step": 10330
    },
    {
      "epoch": 3.0691599881270406,
      "grad_norm": 0.8547385931015015,
      "learning_rate": 1.9311368358563374e-05,
      "loss": 0.7835,
      "step": 10340
    },
    {
      "epoch": 3.0721282279608193,
      "grad_norm": 0.8384952545166016,
      "learning_rate": 1.9281685960225586e-05,
      "loss": 0.788,
      "step": 10350
    },
    {
      "epoch": 3.075096467794598,
      "grad_norm": 0.8666067123413086,
      "learning_rate": 1.9252003561887803e-05,
      "loss": 0.8134,
      "step": 10360
    },
    {
      "epoch": 3.0780647076283763,
      "grad_norm": 0.8557712435722351,
      "learning_rate": 1.9222321163550015e-05,
      "loss": 0.8016,
      "step": 10370
    },
    {
      "epoch": 3.081032947462155,
      "grad_norm": 0.889104425907135,
      "learning_rate": 1.919263876521223e-05,
      "loss": 0.8162,
      "step": 10380
    },
    {
      "epoch": 3.0840011872959336,
      "grad_norm": 0.9883944988250732,
      "learning_rate": 1.9162956366874444e-05,
      "loss": 0.7604,
      "step": 10390
    },
    {
      "epoch": 3.086969427129712,
      "grad_norm": 1.0874537229537964,
      "learning_rate": 1.9133273968536657e-05,
      "loss": 0.8122,
      "step": 10400
    },
    {
      "epoch": 3.0899376669634906,
      "grad_norm": 0.8347809910774231,
      "learning_rate": 1.9103591570198873e-05,
      "loss": 0.775,
      "step": 10410
    },
    {
      "epoch": 3.0929059067972693,
      "grad_norm": 0.8585425615310669,
      "learning_rate": 1.9073909171861086e-05,
      "loss": 0.8108,
      "step": 10420
    },
    {
      "epoch": 3.095874146631048,
      "grad_norm": 0.9726991057395935,
      "learning_rate": 1.9044226773523302e-05,
      "loss": 0.7969,
      "step": 10430
    },
    {
      "epoch": 3.0988423864648262,
      "grad_norm": 0.816733181476593,
      "learning_rate": 1.901454437518552e-05,
      "loss": 0.7843,
      "step": 10440
    },
    {
      "epoch": 3.101810626298605,
      "grad_norm": 0.9138243198394775,
      "learning_rate": 1.898486197684773e-05,
      "loss": 0.7987,
      "step": 10450
    },
    {
      "epoch": 3.1047788661323836,
      "grad_norm": 0.8893081545829773,
      "learning_rate": 1.8955179578509944e-05,
      "loss": 0.8214,
      "step": 10460
    },
    {
      "epoch": 3.107747105966162,
      "grad_norm": 0.87662672996521,
      "learning_rate": 1.8925497180172157e-05,
      "loss": 0.8086,
      "step": 10470
    },
    {
      "epoch": 3.1107153457999406,
      "grad_norm": 0.8176641464233398,
      "learning_rate": 1.8895814781834373e-05,
      "loss": 0.7727,
      "step": 10480
    },
    {
      "epoch": 3.1136835856337193,
      "grad_norm": 0.8557479977607727,
      "learning_rate": 1.886613238349659e-05,
      "loss": 0.7794,
      "step": 10490
    },
    {
      "epoch": 3.116651825467498,
      "grad_norm": 0.8954534530639648,
      "learning_rate": 1.8836449985158802e-05,
      "loss": 0.8434,
      "step": 10500
    },
    {
      "epoch": 3.1196200653012762,
      "grad_norm": 0.8101194500923157,
      "learning_rate": 1.8806767586821018e-05,
      "loss": 0.7627,
      "step": 10510
    },
    {
      "epoch": 3.122588305135055,
      "grad_norm": 0.8535807728767395,
      "learning_rate": 1.877708518848323e-05,
      "loss": 0.7877,
      "step": 10520
    },
    {
      "epoch": 3.1255565449688336,
      "grad_norm": 0.8712933659553528,
      "learning_rate": 1.8747402790145444e-05,
      "loss": 0.8046,
      "step": 10530
    },
    {
      "epoch": 3.128524784802612,
      "grad_norm": 0.825019121170044,
      "learning_rate": 1.871772039180766e-05,
      "loss": 0.7714,
      "step": 10540
    },
    {
      "epoch": 3.1314930246363906,
      "grad_norm": 0.9397444725036621,
      "learning_rate": 1.8688037993469873e-05,
      "loss": 0.8038,
      "step": 10550
    },
    {
      "epoch": 3.1344612644701693,
      "grad_norm": 0.8636755347251892,
      "learning_rate": 1.865835559513209e-05,
      "loss": 0.7902,
      "step": 10560
    },
    {
      "epoch": 3.137429504303948,
      "grad_norm": 0.8545663356781006,
      "learning_rate": 1.86286731967943e-05,
      "loss": 0.7908,
      "step": 10570
    },
    {
      "epoch": 3.140397744137726,
      "grad_norm": 0.9643442034721375,
      "learning_rate": 1.8598990798456518e-05,
      "loss": 0.8231,
      "step": 10580
    },
    {
      "epoch": 3.143365983971505,
      "grad_norm": 0.9168353080749512,
      "learning_rate": 1.856930840011873e-05,
      "loss": 0.7945,
      "step": 10590
    },
    {
      "epoch": 3.1463342238052836,
      "grad_norm": 0.8996456861495972,
      "learning_rate": 1.8539626001780943e-05,
      "loss": 0.8124,
      "step": 10600
    },
    {
      "epoch": 3.149302463639062,
      "grad_norm": 0.86201411485672,
      "learning_rate": 1.850994360344316e-05,
      "loss": 0.7869,
      "step": 10610
    },
    {
      "epoch": 3.1522707034728406,
      "grad_norm": 0.858235239982605,
      "learning_rate": 1.8480261205105372e-05,
      "loss": 0.7738,
      "step": 10620
    },
    {
      "epoch": 3.1552389433066192,
      "grad_norm": 0.8770175576210022,
      "learning_rate": 1.845057880676759e-05,
      "loss": 0.7841,
      "step": 10630
    },
    {
      "epoch": 3.158207183140398,
      "grad_norm": 0.80247563123703,
      "learning_rate": 1.84208964084298e-05,
      "loss": 0.8175,
      "step": 10640
    },
    {
      "epoch": 3.161175422974176,
      "grad_norm": 0.8442670702934265,
      "learning_rate": 1.8391214010092017e-05,
      "loss": 0.7908,
      "step": 10650
    },
    {
      "epoch": 3.164143662807955,
      "grad_norm": 0.9458386301994324,
      "learning_rate": 1.836153161175423e-05,
      "loss": 0.8087,
      "step": 10660
    },
    {
      "epoch": 3.1671119026417336,
      "grad_norm": 0.87587571144104,
      "learning_rate": 1.8331849213416443e-05,
      "loss": 0.8114,
      "step": 10670
    },
    {
      "epoch": 3.170080142475512,
      "grad_norm": 0.8657705187797546,
      "learning_rate": 1.830216681507866e-05,
      "loss": 0.7774,
      "step": 10680
    },
    {
      "epoch": 3.1730483823092905,
      "grad_norm": 0.8687613606452942,
      "learning_rate": 1.8272484416740872e-05,
      "loss": 0.7866,
      "step": 10690
    },
    {
      "epoch": 3.1760166221430692,
      "grad_norm": 0.8471196889877319,
      "learning_rate": 1.8242802018403088e-05,
      "loss": 0.7761,
      "step": 10700
    },
    {
      "epoch": 3.178984861976848,
      "grad_norm": 0.9914931654930115,
      "learning_rate": 1.82131196200653e-05,
      "loss": 0.7994,
      "step": 10710
    },
    {
      "epoch": 3.181953101810626,
      "grad_norm": 0.8881893157958984,
      "learning_rate": 1.8183437221727517e-05,
      "loss": 0.7688,
      "step": 10720
    },
    {
      "epoch": 3.184921341644405,
      "grad_norm": 0.934282124042511,
      "learning_rate": 1.8153754823389733e-05,
      "loss": 0.8089,
      "step": 10730
    },
    {
      "epoch": 3.1878895814781836,
      "grad_norm": 0.9608398079872131,
      "learning_rate": 1.8124072425051943e-05,
      "loss": 0.7777,
      "step": 10740
    },
    {
      "epoch": 3.190857821311962,
      "grad_norm": 0.8693539500236511,
      "learning_rate": 1.809439002671416e-05,
      "loss": 0.7865,
      "step": 10750
    },
    {
      "epoch": 3.1938260611457405,
      "grad_norm": 0.8089728355407715,
      "learning_rate": 1.806470762837637e-05,
      "loss": 0.8126,
      "step": 10760
    },
    {
      "epoch": 3.196794300979519,
      "grad_norm": 0.839136004447937,
      "learning_rate": 1.8035025230038588e-05,
      "loss": 0.8116,
      "step": 10770
    },
    {
      "epoch": 3.199762540813298,
      "grad_norm": 0.9516549706459045,
      "learning_rate": 1.8005342831700804e-05,
      "loss": 0.7901,
      "step": 10780
    },
    {
      "epoch": 3.202730780647076,
      "grad_norm": 0.9636576175689697,
      "learning_rate": 1.7975660433363017e-05,
      "loss": 0.7876,
      "step": 10790
    },
    {
      "epoch": 3.205699020480855,
      "grad_norm": 0.7846948504447937,
      "learning_rate": 1.7945978035025233e-05,
      "loss": 0.7793,
      "step": 10800
    },
    {
      "epoch": 3.2086672603146336,
      "grad_norm": 0.9290656447410583,
      "learning_rate": 1.7916295636687442e-05,
      "loss": 0.7998,
      "step": 10810
    },
    {
      "epoch": 3.211635500148412,
      "grad_norm": 0.8134689331054688,
      "learning_rate": 1.788661323834966e-05,
      "loss": 0.7932,
      "step": 10820
    },
    {
      "epoch": 3.2146037399821905,
      "grad_norm": 0.8373258709907532,
      "learning_rate": 1.7856930840011875e-05,
      "loss": 0.7989,
      "step": 10830
    },
    {
      "epoch": 3.217571979815969,
      "grad_norm": 0.8180834054946899,
      "learning_rate": 1.7827248441674087e-05,
      "loss": 0.8083,
      "step": 10840
    },
    {
      "epoch": 3.220540219649748,
      "grad_norm": 0.947187066078186,
      "learning_rate": 1.7797566043336304e-05,
      "loss": 0.7798,
      "step": 10850
    },
    {
      "epoch": 3.223508459483526,
      "grad_norm": 1.0117039680480957,
      "learning_rate": 1.7767883644998516e-05,
      "loss": 0.8054,
      "step": 10860
    },
    {
      "epoch": 3.226476699317305,
      "grad_norm": 0.9373120069503784,
      "learning_rate": 1.7738201246660733e-05,
      "loss": 0.8004,
      "step": 10870
    },
    {
      "epoch": 3.2294449391510835,
      "grad_norm": 0.8610514998435974,
      "learning_rate": 1.7708518848322945e-05,
      "loss": 0.7677,
      "step": 10880
    },
    {
      "epoch": 3.232413178984862,
      "grad_norm": 0.9003437161445618,
      "learning_rate": 1.7678836449985158e-05,
      "loss": 0.8095,
      "step": 10890
    },
    {
      "epoch": 3.2353814188186405,
      "grad_norm": 0.9555358290672302,
      "learning_rate": 1.7649154051647374e-05,
      "loss": 0.8184,
      "step": 10900
    },
    {
      "epoch": 3.238349658652419,
      "grad_norm": 0.9005022048950195,
      "learning_rate": 1.7619471653309587e-05,
      "loss": 0.7944,
      "step": 10910
    },
    {
      "epoch": 3.241317898486198,
      "grad_norm": 0.8886909484863281,
      "learning_rate": 1.7589789254971803e-05,
      "loss": 0.7745,
      "step": 10920
    },
    {
      "epoch": 3.244286138319976,
      "grad_norm": 0.8552870154380798,
      "learning_rate": 1.7560106856634016e-05,
      "loss": 0.7807,
      "step": 10930
    },
    {
      "epoch": 3.247254378153755,
      "grad_norm": 1.0251131057739258,
      "learning_rate": 1.7530424458296232e-05,
      "loss": 0.79,
      "step": 10940
    },
    {
      "epoch": 3.2502226179875335,
      "grad_norm": 0.8612068891525269,
      "learning_rate": 1.7500742059958445e-05,
      "loss": 0.8086,
      "step": 10950
    },
    {
      "epoch": 3.253190857821312,
      "grad_norm": 0.8735592365264893,
      "learning_rate": 1.7471059661620658e-05,
      "loss": 0.8288,
      "step": 10960
    },
    {
      "epoch": 3.2561590976550905,
      "grad_norm": 1.012428641319275,
      "learning_rate": 1.7441377263282874e-05,
      "loss": 0.8143,
      "step": 10970
    },
    {
      "epoch": 3.259127337488869,
      "grad_norm": 0.8430175185203552,
      "learning_rate": 1.7411694864945087e-05,
      "loss": 0.7853,
      "step": 10980
    },
    {
      "epoch": 3.262095577322648,
      "grad_norm": 0.9051916599273682,
      "learning_rate": 1.7382012466607303e-05,
      "loss": 0.7617,
      "step": 10990
    },
    {
      "epoch": 3.265063817156426,
      "grad_norm": 0.8995745778083801,
      "learning_rate": 1.735233006826952e-05,
      "loss": 0.8128,
      "step": 11000
    },
    {
      "epoch": 3.268032056990205,
      "grad_norm": 0.893775224685669,
      "learning_rate": 1.7322647669931732e-05,
      "loss": 0.8046,
      "step": 11010
    },
    {
      "epoch": 3.2710002968239835,
      "grad_norm": 0.8447341322898865,
      "learning_rate": 1.7292965271593945e-05,
      "loss": 0.7941,
      "step": 11020
    },
    {
      "epoch": 3.2739685366577618,
      "grad_norm": 0.8254780769348145,
      "learning_rate": 1.7263282873256158e-05,
      "loss": 0.7854,
      "step": 11030
    },
    {
      "epoch": 3.2769367764915405,
      "grad_norm": 0.8724144697189331,
      "learning_rate": 1.7233600474918374e-05,
      "loss": 0.8251,
      "step": 11040
    },
    {
      "epoch": 3.279905016325319,
      "grad_norm": 0.848600447177887,
      "learning_rate": 1.720391807658059e-05,
      "loss": 0.7821,
      "step": 11050
    },
    {
      "epoch": 3.282873256159098,
      "grad_norm": 0.9910843372344971,
      "learning_rate": 1.7174235678242803e-05,
      "loss": 0.7618,
      "step": 11060
    },
    {
      "epoch": 3.285841495992876,
      "grad_norm": 0.8709274530410767,
      "learning_rate": 1.714455327990502e-05,
      "loss": 0.7925,
      "step": 11070
    },
    {
      "epoch": 3.288809735826655,
      "grad_norm": 0.8491528630256653,
      "learning_rate": 1.711487088156723e-05,
      "loss": 0.7978,
      "step": 11080
    },
    {
      "epoch": 3.2917779756604335,
      "grad_norm": 0.9000929594039917,
      "learning_rate": 1.7085188483229444e-05,
      "loss": 0.8122,
      "step": 11090
    },
    {
      "epoch": 3.2947462154942118,
      "grad_norm": 0.8909440636634827,
      "learning_rate": 1.705550608489166e-05,
      "loss": 0.7464,
      "step": 11100
    },
    {
      "epoch": 3.2977144553279905,
      "grad_norm": 0.9142547249794006,
      "learning_rate": 1.7025823686553873e-05,
      "loss": 0.8009,
      "step": 11110
    },
    {
      "epoch": 3.300682695161769,
      "grad_norm": 0.8039944171905518,
      "learning_rate": 1.699614128821609e-05,
      "loss": 0.7625,
      "step": 11120
    },
    {
      "epoch": 3.303650934995548,
      "grad_norm": 0.9256371855735779,
      "learning_rate": 1.6966458889878302e-05,
      "loss": 0.8064,
      "step": 11130
    },
    {
      "epoch": 3.306619174829326,
      "grad_norm": 0.8562237024307251,
      "learning_rate": 1.693677649154052e-05,
      "loss": 0.8068,
      "step": 11140
    },
    {
      "epoch": 3.309587414663105,
      "grad_norm": 0.7977147102355957,
      "learning_rate": 1.690709409320273e-05,
      "loss": 0.7862,
      "step": 11150
    },
    {
      "epoch": 3.3125556544968835,
      "grad_norm": 0.8706709742546082,
      "learning_rate": 1.6877411694864944e-05,
      "loss": 0.7739,
      "step": 11160
    },
    {
      "epoch": 3.3155238943306617,
      "grad_norm": 0.8891844749450684,
      "learning_rate": 1.684772929652716e-05,
      "loss": 0.8359,
      "step": 11170
    },
    {
      "epoch": 3.3184921341644404,
      "grad_norm": 0.9938377141952515,
      "learning_rate": 1.6818046898189373e-05,
      "loss": 0.7695,
      "step": 11180
    },
    {
      "epoch": 3.321460373998219,
      "grad_norm": 0.888454794883728,
      "learning_rate": 1.678836449985159e-05,
      "loss": 0.7916,
      "step": 11190
    },
    {
      "epoch": 3.324428613831998,
      "grad_norm": 0.8678205013275146,
      "learning_rate": 1.6758682101513802e-05,
      "loss": 0.7789,
      "step": 11200
    },
    {
      "epoch": 3.327396853665776,
      "grad_norm": 0.842923641204834,
      "learning_rate": 1.6728999703176018e-05,
      "loss": 0.7936,
      "step": 11210
    },
    {
      "epoch": 3.330365093499555,
      "grad_norm": 0.923405647277832,
      "learning_rate": 1.6699317304838234e-05,
      "loss": 0.7737,
      "step": 11220
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 0.9673542380332947,
      "learning_rate": 1.6669634906500444e-05,
      "loss": 0.8157,
      "step": 11230
    },
    {
      "epoch": 3.3363015731671117,
      "grad_norm": 0.8497329354286194,
      "learning_rate": 1.663995250816266e-05,
      "loss": 0.8217,
      "step": 11240
    },
    {
      "epoch": 3.3392698130008904,
      "grad_norm": 0.8752831220626831,
      "learning_rate": 1.6610270109824873e-05,
      "loss": 0.8195,
      "step": 11250
    },
    {
      "epoch": 3.342238052834669,
      "grad_norm": 0.9098172783851624,
      "learning_rate": 1.658058771148709e-05,
      "loss": 0.7981,
      "step": 11260
    },
    {
      "epoch": 3.345206292668448,
      "grad_norm": 0.9954054951667786,
      "learning_rate": 1.6550905313149305e-05,
      "loss": 0.8293,
      "step": 11270
    },
    {
      "epoch": 3.348174532502226,
      "grad_norm": 0.912259578704834,
      "learning_rate": 1.6521222914811518e-05,
      "loss": 0.8394,
      "step": 11280
    },
    {
      "epoch": 3.3511427723360048,
      "grad_norm": 0.8457381725311279,
      "learning_rate": 1.6491540516473734e-05,
      "loss": 0.7682,
      "step": 11290
    },
    {
      "epoch": 3.3541110121697835,
      "grad_norm": 0.8962229490280151,
      "learning_rate": 1.6461858118135943e-05,
      "loss": 0.7953,
      "step": 11300
    },
    {
      "epoch": 3.3570792520035617,
      "grad_norm": 0.843609094619751,
      "learning_rate": 1.643217571979816e-05,
      "loss": 0.7959,
      "step": 11310
    },
    {
      "epoch": 3.3600474918373404,
      "grad_norm": 0.9469923377037048,
      "learning_rate": 1.6402493321460376e-05,
      "loss": 0.7879,
      "step": 11320
    },
    {
      "epoch": 3.363015731671119,
      "grad_norm": 0.897754430770874,
      "learning_rate": 1.637281092312259e-05,
      "loss": 0.8087,
      "step": 11330
    },
    {
      "epoch": 3.365983971504898,
      "grad_norm": 0.895817220211029,
      "learning_rate": 1.6343128524784805e-05,
      "loss": 0.801,
      "step": 11340
    },
    {
      "epoch": 3.368952211338676,
      "grad_norm": 0.8911547660827637,
      "learning_rate": 1.6313446126447018e-05,
      "loss": 0.7607,
      "step": 11350
    },
    {
      "epoch": 3.3719204511724548,
      "grad_norm": 0.9019851088523865,
      "learning_rate": 1.6283763728109234e-05,
      "loss": 0.8001,
      "step": 11360
    },
    {
      "epoch": 3.3748886910062335,
      "grad_norm": 0.9182178974151611,
      "learning_rate": 1.6254081329771447e-05,
      "loss": 0.7948,
      "step": 11370
    },
    {
      "epoch": 3.3778569308400117,
      "grad_norm": 0.8582599759101868,
      "learning_rate": 1.622439893143366e-05,
      "loss": 0.7925,
      "step": 11380
    },
    {
      "epoch": 3.3808251706737904,
      "grad_norm": 0.916682243347168,
      "learning_rate": 1.6194716533095876e-05,
      "loss": 0.7885,
      "step": 11390
    },
    {
      "epoch": 3.383793410507569,
      "grad_norm": 0.9534935355186462,
      "learning_rate": 1.6168002374591868e-05,
      "loss": 0.7961,
      "step": 11400
    },
    {
      "epoch": 3.386761650341348,
      "grad_norm": 0.9537315964698792,
      "learning_rate": 1.613831997625408e-05,
      "loss": 0.8117,
      "step": 11410
    },
    {
      "epoch": 3.389729890175126,
      "grad_norm": 0.8636172413825989,
      "learning_rate": 1.6108637577916297e-05,
      "loss": 0.7971,
      "step": 11420
    },
    {
      "epoch": 3.3926981300089047,
      "grad_norm": 0.8848598003387451,
      "learning_rate": 1.607895517957851e-05,
      "loss": 0.8086,
      "step": 11430
    },
    {
      "epoch": 3.3956663698426834,
      "grad_norm": 0.8908330202102661,
      "learning_rate": 1.6049272781240726e-05,
      "loss": 0.7773,
      "step": 11440
    },
    {
      "epoch": 3.3986346096764617,
      "grad_norm": 0.8260102272033691,
      "learning_rate": 1.601959038290294e-05,
      "loss": 0.8284,
      "step": 11450
    },
    {
      "epoch": 3.4016028495102404,
      "grad_norm": 0.9572765827178955,
      "learning_rate": 1.598990798456515e-05,
      "loss": 0.8103,
      "step": 11460
    },
    {
      "epoch": 3.404571089344019,
      "grad_norm": 0.8650007843971252,
      "learning_rate": 1.5960225586227368e-05,
      "loss": 0.8,
      "step": 11470
    },
    {
      "epoch": 3.407539329177798,
      "grad_norm": 0.8484728932380676,
      "learning_rate": 1.5930543187889584e-05,
      "loss": 0.825,
      "step": 11480
    },
    {
      "epoch": 3.410507569011576,
      "grad_norm": 1.0106016397476196,
      "learning_rate": 1.5900860789551797e-05,
      "loss": 0.8011,
      "step": 11490
    },
    {
      "epoch": 3.4134758088453547,
      "grad_norm": 0.8704240918159485,
      "learning_rate": 1.5871178391214013e-05,
      "loss": 0.8035,
      "step": 11500
    },
    {
      "epoch": 3.4164440486791334,
      "grad_norm": 0.9683484435081482,
      "learning_rate": 1.5841495992876226e-05,
      "loss": 0.8223,
      "step": 11510
    },
    {
      "epoch": 3.4194122885129117,
      "grad_norm": 0.9670411944389343,
      "learning_rate": 1.581181359453844e-05,
      "loss": 0.806,
      "step": 11520
    },
    {
      "epoch": 3.4223805283466904,
      "grad_norm": 0.8676766157150269,
      "learning_rate": 1.5782131196200655e-05,
      "loss": 0.8137,
      "step": 11530
    },
    {
      "epoch": 3.425348768180469,
      "grad_norm": 0.940537691116333,
      "learning_rate": 1.5752448797862867e-05,
      "loss": 0.7878,
      "step": 11540
    },
    {
      "epoch": 3.4283170080142478,
      "grad_norm": 0.9217203855514526,
      "learning_rate": 1.5722766399525084e-05,
      "loss": 0.8388,
      "step": 11550
    },
    {
      "epoch": 3.431285247848026,
      "grad_norm": 0.8722837567329407,
      "learning_rate": 1.5693084001187296e-05,
      "loss": 0.7985,
      "step": 11560
    },
    {
      "epoch": 3.4342534876818047,
      "grad_norm": 0.8811752200126648,
      "learning_rate": 1.5663401602849513e-05,
      "loss": 0.7873,
      "step": 11570
    },
    {
      "epoch": 3.4372217275155834,
      "grad_norm": 0.9148281812667847,
      "learning_rate": 1.5633719204511725e-05,
      "loss": 0.8035,
      "step": 11580
    },
    {
      "epoch": 3.4401899673493617,
      "grad_norm": 0.8932378888130188,
      "learning_rate": 1.5604036806173938e-05,
      "loss": 0.832,
      "step": 11590
    },
    {
      "epoch": 3.4431582071831404,
      "grad_norm": 0.9634125232696533,
      "learning_rate": 1.5574354407836154e-05,
      "loss": 0.7701,
      "step": 11600
    },
    {
      "epoch": 3.446126447016919,
      "grad_norm": 0.9190008044242859,
      "learning_rate": 1.5544672009498367e-05,
      "loss": 0.8277,
      "step": 11610
    },
    {
      "epoch": 3.4490946868506978,
      "grad_norm": 0.8262824416160583,
      "learning_rate": 1.5514989611160583e-05,
      "loss": 0.7593,
      "step": 11620
    },
    {
      "epoch": 3.452062926684476,
      "grad_norm": 0.9550215601921082,
      "learning_rate": 1.5485307212822796e-05,
      "loss": 0.7881,
      "step": 11630
    },
    {
      "epoch": 3.4550311665182547,
      "grad_norm": 0.8827412724494934,
      "learning_rate": 1.5455624814485012e-05,
      "loss": 0.7965,
      "step": 11640
    },
    {
      "epoch": 3.4579994063520334,
      "grad_norm": 0.8873876929283142,
      "learning_rate": 1.5425942416147225e-05,
      "loss": 0.8088,
      "step": 11650
    },
    {
      "epoch": 3.4609676461858117,
      "grad_norm": 0.9476723670959473,
      "learning_rate": 1.5396260017809438e-05,
      "loss": 0.8102,
      "step": 11660
    },
    {
      "epoch": 3.4639358860195903,
      "grad_norm": 0.9086580276489258,
      "learning_rate": 1.5366577619471654e-05,
      "loss": 0.8025,
      "step": 11670
    },
    {
      "epoch": 3.466904125853369,
      "grad_norm": 0.9368143081665039,
      "learning_rate": 1.5336895221133867e-05,
      "loss": 0.8186,
      "step": 11680
    },
    {
      "epoch": 3.4698723656871477,
      "grad_norm": 0.9438993334770203,
      "learning_rate": 1.5307212822796083e-05,
      "loss": 0.7736,
      "step": 11690
    },
    {
      "epoch": 3.472840605520926,
      "grad_norm": 0.9083358645439148,
      "learning_rate": 1.5277530424458296e-05,
      "loss": 0.8278,
      "step": 11700
    },
    {
      "epoch": 3.4758088453547047,
      "grad_norm": 0.8738546371459961,
      "learning_rate": 1.5247848026120512e-05,
      "loss": 0.7832,
      "step": 11710
    },
    {
      "epoch": 3.4787770851884834,
      "grad_norm": 0.9554048180580139,
      "learning_rate": 1.5218165627782726e-05,
      "loss": 0.7921,
      "step": 11720
    },
    {
      "epoch": 3.4817453250222616,
      "grad_norm": 0.8620092868804932,
      "learning_rate": 1.518848322944494e-05,
      "loss": 0.7969,
      "step": 11730
    },
    {
      "epoch": 3.4847135648560403,
      "grad_norm": 0.8661124110221863,
      "learning_rate": 1.5158800831107154e-05,
      "loss": 0.7884,
      "step": 11740
    },
    {
      "epoch": 3.487681804689819,
      "grad_norm": 0.8738637566566467,
      "learning_rate": 1.5129118432769368e-05,
      "loss": 0.7685,
      "step": 11750
    },
    {
      "epoch": 3.4906500445235977,
      "grad_norm": 0.8339844942092896,
      "learning_rate": 1.5099436034431583e-05,
      "loss": 0.7816,
      "step": 11760
    },
    {
      "epoch": 3.493618284357376,
      "grad_norm": 0.9325805902481079,
      "learning_rate": 1.5069753636093797e-05,
      "loss": 0.7474,
      "step": 11770
    },
    {
      "epoch": 3.4965865241911547,
      "grad_norm": 0.798915445804596,
      "learning_rate": 1.5040071237756012e-05,
      "loss": 0.8013,
      "step": 11780
    },
    {
      "epoch": 3.4995547640249334,
      "grad_norm": 0.9759164452552795,
      "learning_rate": 1.5010388839418226e-05,
      "loss": 0.7933,
      "step": 11790
    },
    {
      "epoch": 3.5025230038587116,
      "grad_norm": 0.9221082329750061,
      "learning_rate": 1.4980706441080439e-05,
      "loss": 0.7591,
      "step": 11800
    },
    {
      "epoch": 3.5054912436924903,
      "grad_norm": 0.9643858075141907,
      "learning_rate": 1.4951024042742653e-05,
      "loss": 0.8277,
      "step": 11810
    },
    {
      "epoch": 3.508459483526269,
      "grad_norm": 0.9068611860275269,
      "learning_rate": 1.4921341644404868e-05,
      "loss": 0.8064,
      "step": 11820
    },
    {
      "epoch": 3.5114277233600477,
      "grad_norm": 0.9101418852806091,
      "learning_rate": 1.4891659246067082e-05,
      "loss": 0.8088,
      "step": 11830
    },
    {
      "epoch": 3.514395963193826,
      "grad_norm": 0.832104504108429,
      "learning_rate": 1.4861976847729297e-05,
      "loss": 0.8312,
      "step": 11840
    },
    {
      "epoch": 3.5173642030276047,
      "grad_norm": 0.9002504944801331,
      "learning_rate": 1.4832294449391513e-05,
      "loss": 0.7986,
      "step": 11850
    },
    {
      "epoch": 3.5203324428613834,
      "grad_norm": 1.0496249198913574,
      "learning_rate": 1.4802612051053727e-05,
      "loss": 0.8074,
      "step": 11860
    },
    {
      "epoch": 3.5233006826951616,
      "grad_norm": 0.9258970022201538,
      "learning_rate": 1.4772929652715939e-05,
      "loss": 0.8434,
      "step": 11870
    },
    {
      "epoch": 3.5262689225289403,
      "grad_norm": 0.8295080661773682,
      "learning_rate": 1.4743247254378153e-05,
      "loss": 0.7798,
      "step": 11880
    },
    {
      "epoch": 3.529237162362719,
      "grad_norm": 0.990430474281311,
      "learning_rate": 1.4713564856040368e-05,
      "loss": 0.8251,
      "step": 11890
    },
    {
      "epoch": 3.5322054021964977,
      "grad_norm": 0.9134912490844727,
      "learning_rate": 1.4683882457702584e-05,
      "loss": 0.7745,
      "step": 11900
    },
    {
      "epoch": 3.535173642030276,
      "grad_norm": 0.9166617393493652,
      "learning_rate": 1.4654200059364798e-05,
      "loss": 0.8286,
      "step": 11910
    },
    {
      "epoch": 3.5381418818640546,
      "grad_norm": 0.9793247580528259,
      "learning_rate": 1.4624517661027013e-05,
      "loss": 0.7713,
      "step": 11920
    },
    {
      "epoch": 3.5411101216978333,
      "grad_norm": 0.982891857624054,
      "learning_rate": 1.4594835262689227e-05,
      "loss": 0.7955,
      "step": 11930
    },
    {
      "epoch": 3.5440783615316116,
      "grad_norm": 0.9041152596473694,
      "learning_rate": 1.4565152864351438e-05,
      "loss": 0.8356,
      "step": 11940
    },
    {
      "epoch": 3.5470466013653903,
      "grad_norm": 0.9441630840301514,
      "learning_rate": 1.4535470466013654e-05,
      "loss": 0.7628,
      "step": 11950
    },
    {
      "epoch": 3.550014841199169,
      "grad_norm": 0.8983331322669983,
      "learning_rate": 1.4505788067675869e-05,
      "loss": 0.7795,
      "step": 11960
    },
    {
      "epoch": 3.5529830810329477,
      "grad_norm": 0.9247541427612305,
      "learning_rate": 1.4476105669338083e-05,
      "loss": 0.7868,
      "step": 11970
    },
    {
      "epoch": 3.555951320866726,
      "grad_norm": 1.140688180923462,
      "learning_rate": 1.4446423271000298e-05,
      "loss": 0.7919,
      "step": 11980
    },
    {
      "epoch": 3.5589195607005046,
      "grad_norm": 0.8730014562606812,
      "learning_rate": 1.4416740872662512e-05,
      "loss": 0.8075,
      "step": 11990
    },
    {
      "epoch": 3.5618878005342833,
      "grad_norm": 0.9086686968803406,
      "learning_rate": 1.4387058474324727e-05,
      "loss": 0.7802,
      "step": 12000
    },
    {
      "epoch": 3.5648560403680616,
      "grad_norm": 0.8341102004051208,
      "learning_rate": 1.435737607598694e-05,
      "loss": 0.7868,
      "step": 12010
    },
    {
      "epoch": 3.5678242802018403,
      "grad_norm": 1.0596048831939697,
      "learning_rate": 1.4327693677649154e-05,
      "loss": 0.7901,
      "step": 12020
    },
    {
      "epoch": 3.570792520035619,
      "grad_norm": 0.8686931729316711,
      "learning_rate": 1.4298011279311369e-05,
      "loss": 0.8188,
      "step": 12030
    },
    {
      "epoch": 3.5737607598693977,
      "grad_norm": 0.8936414122581482,
      "learning_rate": 1.4268328880973583e-05,
      "loss": 0.7681,
      "step": 12040
    },
    {
      "epoch": 3.576728999703176,
      "grad_norm": 0.8683655261993408,
      "learning_rate": 1.4238646482635798e-05,
      "loss": 0.8043,
      "step": 12050
    },
    {
      "epoch": 3.5796972395369546,
      "grad_norm": 0.9319801330566406,
      "learning_rate": 1.4208964084298012e-05,
      "loss": 0.7831,
      "step": 12060
    },
    {
      "epoch": 3.5826654793707333,
      "grad_norm": 0.8970152735710144,
      "learning_rate": 1.4179281685960226e-05,
      "loss": 0.7626,
      "step": 12070
    },
    {
      "epoch": 3.5856337192045116,
      "grad_norm": 0.9043969511985779,
      "learning_rate": 1.414959928762244e-05,
      "loss": 0.7909,
      "step": 12080
    },
    {
      "epoch": 3.5886019590382903,
      "grad_norm": 0.9323931336402893,
      "learning_rate": 1.4119916889284654e-05,
      "loss": 0.7831,
      "step": 12090
    },
    {
      "epoch": 3.591570198872069,
      "grad_norm": 0.9297446608543396,
      "learning_rate": 1.4090234490946868e-05,
      "loss": 0.8151,
      "step": 12100
    },
    {
      "epoch": 3.5945384387058477,
      "grad_norm": 0.9543979167938232,
      "learning_rate": 1.4060552092609083e-05,
      "loss": 0.7986,
      "step": 12110
    },
    {
      "epoch": 3.597506678539626,
      "grad_norm": 0.951012134552002,
      "learning_rate": 1.4030869694271297e-05,
      "loss": 0.8287,
      "step": 12120
    },
    {
      "epoch": 3.6004749183734046,
      "grad_norm": 0.9943885207176208,
      "learning_rate": 1.4001187295933513e-05,
      "loss": 0.7854,
      "step": 12130
    },
    {
      "epoch": 3.6034431582071833,
      "grad_norm": 0.8945292234420776,
      "learning_rate": 1.3971504897595728e-05,
      "loss": 0.7545,
      "step": 12140
    },
    {
      "epoch": 3.6064113980409616,
      "grad_norm": 0.8595576286315918,
      "learning_rate": 1.3941822499257939e-05,
      "loss": 0.7712,
      "step": 12150
    },
    {
      "epoch": 3.6093796378747403,
      "grad_norm": 0.9131389260292053,
      "learning_rate": 1.3912140100920153e-05,
      "loss": 0.7732,
      "step": 12160
    },
    {
      "epoch": 3.612347877708519,
      "grad_norm": 0.8835858106613159,
      "learning_rate": 1.388245770258237e-05,
      "loss": 0.805,
      "step": 12170
    },
    {
      "epoch": 3.6153161175422976,
      "grad_norm": 0.8401873707771301,
      "learning_rate": 1.3852775304244584e-05,
      "loss": 0.7814,
      "step": 12180
    },
    {
      "epoch": 3.618284357376076,
      "grad_norm": 0.8316750526428223,
      "learning_rate": 1.3823092905906799e-05,
      "loss": 0.8215,
      "step": 12190
    },
    {
      "epoch": 3.6212525972098546,
      "grad_norm": 0.9151681661605835,
      "learning_rate": 1.3793410507569013e-05,
      "loss": 0.8424,
      "step": 12200
    },
    {
      "epoch": 3.6242208370436333,
      "grad_norm": 0.896335780620575,
      "learning_rate": 1.3763728109231228e-05,
      "loss": 0.788,
      "step": 12210
    },
    {
      "epoch": 3.6271890768774115,
      "grad_norm": 1.0074635744094849,
      "learning_rate": 1.373404571089344e-05,
      "loss": 0.8132,
      "step": 12220
    },
    {
      "epoch": 3.6301573167111902,
      "grad_norm": 0.9395803809165955,
      "learning_rate": 1.3704363312555655e-05,
      "loss": 0.823,
      "step": 12230
    },
    {
      "epoch": 3.633125556544969,
      "grad_norm": 0.8941648006439209,
      "learning_rate": 1.367468091421787e-05,
      "loss": 0.7965,
      "step": 12240
    },
    {
      "epoch": 3.6360937963787476,
      "grad_norm": 0.9906322956085205,
      "learning_rate": 1.3644998515880084e-05,
      "loss": 0.7901,
      "step": 12250
    },
    {
      "epoch": 3.639062036212526,
      "grad_norm": 0.9607106447219849,
      "learning_rate": 1.3615316117542298e-05,
      "loss": 0.7654,
      "step": 12260
    },
    {
      "epoch": 3.6420302760463046,
      "grad_norm": 0.9346127510070801,
      "learning_rate": 1.3585633719204513e-05,
      "loss": 0.785,
      "step": 12270
    },
    {
      "epoch": 3.644998515880083,
      "grad_norm": 0.9905751943588257,
      "learning_rate": 1.3555951320866727e-05,
      "loss": 0.7745,
      "step": 12280
    },
    {
      "epoch": 3.6479667557138615,
      "grad_norm": 0.951194703578949,
      "learning_rate": 1.352626892252894e-05,
      "loss": 0.8131,
      "step": 12290
    },
    {
      "epoch": 3.6509349955476402,
      "grad_norm": 0.9166809320449829,
      "learning_rate": 1.3496586524191154e-05,
      "loss": 0.7626,
      "step": 12300
    },
    {
      "epoch": 3.653903235381419,
      "grad_norm": 0.8549669981002808,
      "learning_rate": 1.3466904125853369e-05,
      "loss": 0.7834,
      "step": 12310
    },
    {
      "epoch": 3.6568714752151976,
      "grad_norm": 0.8752672672271729,
      "learning_rate": 1.3437221727515583e-05,
      "loss": 0.8363,
      "step": 12320
    },
    {
      "epoch": 3.659839715048976,
      "grad_norm": 0.8863996267318726,
      "learning_rate": 1.3407539329177798e-05,
      "loss": 0.7923,
      "step": 12330
    },
    {
      "epoch": 3.6628079548827546,
      "grad_norm": 0.9465095400810242,
      "learning_rate": 1.3377856930840012e-05,
      "loss": 0.8218,
      "step": 12340
    },
    {
      "epoch": 3.665776194716533,
      "grad_norm": 0.9206190705299377,
      "learning_rate": 1.3348174532502229e-05,
      "loss": 0.7686,
      "step": 12350
    },
    {
      "epoch": 3.6687444345503115,
      "grad_norm": 0.9813005924224854,
      "learning_rate": 1.331849213416444e-05,
      "loss": 0.7735,
      "step": 12360
    },
    {
      "epoch": 3.67171267438409,
      "grad_norm": 0.9265437722206116,
      "learning_rate": 1.3288809735826654e-05,
      "loss": 0.7836,
      "step": 12370
    },
    {
      "epoch": 3.674680914217869,
      "grad_norm": 1.0839077234268188,
      "learning_rate": 1.3259127337488869e-05,
      "loss": 0.7987,
      "step": 12380
    },
    {
      "epoch": 3.6776491540516476,
      "grad_norm": 1.095517873764038,
      "learning_rate": 1.3229444939151083e-05,
      "loss": 0.8085,
      "step": 12390
    },
    {
      "epoch": 3.680617393885426,
      "grad_norm": 1.037738561630249,
      "learning_rate": 1.31997625408133e-05,
      "loss": 0.7919,
      "step": 12400
    },
    {
      "epoch": 3.6835856337192046,
      "grad_norm": 0.8873872756958008,
      "learning_rate": 1.3170080142475514e-05,
      "loss": 0.819,
      "step": 12410
    },
    {
      "epoch": 3.686553873552983,
      "grad_norm": 0.8964414596557617,
      "learning_rate": 1.3140397744137728e-05,
      "loss": 0.7901,
      "step": 12420
    },
    {
      "epoch": 3.6895221133867615,
      "grad_norm": 0.8763006329536438,
      "learning_rate": 1.3110715345799943e-05,
      "loss": 0.8127,
      "step": 12430
    },
    {
      "epoch": 3.69249035322054,
      "grad_norm": 0.9101526141166687,
      "learning_rate": 1.3081032947462154e-05,
      "loss": 0.8038,
      "step": 12440
    },
    {
      "epoch": 3.695458593054319,
      "grad_norm": 0.8191701173782349,
      "learning_rate": 1.305135054912437e-05,
      "loss": 0.806,
      "step": 12450
    },
    {
      "epoch": 3.6984268328880976,
      "grad_norm": 0.8987050652503967,
      "learning_rate": 1.3021668150786585e-05,
      "loss": 0.7785,
      "step": 12460
    },
    {
      "epoch": 3.701395072721876,
      "grad_norm": 0.8950318694114685,
      "learning_rate": 1.2991985752448799e-05,
      "loss": 0.7813,
      "step": 12470
    },
    {
      "epoch": 3.7043633125556545,
      "grad_norm": 0.9262118339538574,
      "learning_rate": 1.2962303354111013e-05,
      "loss": 0.8066,
      "step": 12480
    },
    {
      "epoch": 3.707331552389433,
      "grad_norm": 0.9165562391281128,
      "learning_rate": 1.2932620955773228e-05,
      "loss": 0.7721,
      "step": 12490
    },
    {
      "epoch": 3.7102997922232115,
      "grad_norm": 0.8464228510856628,
      "learning_rate": 1.2902938557435442e-05,
      "loss": 0.8219,
      "step": 12500
    },
    {
      "epoch": 3.71326803205699,
      "grad_norm": 0.8785159587860107,
      "learning_rate": 1.2873256159097655e-05,
      "loss": 0.8082,
      "step": 12510
    },
    {
      "epoch": 3.716236271890769,
      "grad_norm": 0.8984147906303406,
      "learning_rate": 1.284357376075987e-05,
      "loss": 0.7928,
      "step": 12520
    },
    {
      "epoch": 3.7192045117245476,
      "grad_norm": 0.9415210485458374,
      "learning_rate": 1.2813891362422084e-05,
      "loss": 0.7921,
      "step": 12530
    },
    {
      "epoch": 3.722172751558326,
      "grad_norm": 1.1073685884475708,
      "learning_rate": 1.2784208964084299e-05,
      "loss": 0.8225,
      "step": 12540
    },
    {
      "epoch": 3.7251409913921045,
      "grad_norm": 1.000339388847351,
      "learning_rate": 1.2754526565746513e-05,
      "loss": 0.8255,
      "step": 12550
    },
    {
      "epoch": 3.728109231225883,
      "grad_norm": 0.9128279685974121,
      "learning_rate": 1.2724844167408728e-05,
      "loss": 0.7989,
      "step": 12560
    },
    {
      "epoch": 3.7310774710596615,
      "grad_norm": 0.8923486471176147,
      "learning_rate": 1.2695161769070942e-05,
      "loss": 0.812,
      "step": 12570
    },
    {
      "epoch": 3.73404571089344,
      "grad_norm": 1.0165687799453735,
      "learning_rate": 1.2665479370733155e-05,
      "loss": 0.8055,
      "step": 12580
    },
    {
      "epoch": 3.737013950727219,
      "grad_norm": 0.964148998260498,
      "learning_rate": 1.263579697239537e-05,
      "loss": 0.8016,
      "step": 12590
    },
    {
      "epoch": 3.7399821905609976,
      "grad_norm": 0.9706274271011353,
      "learning_rate": 1.2606114574057584e-05,
      "loss": 0.7763,
      "step": 12600
    },
    {
      "epoch": 3.742950430394776,
      "grad_norm": 1.0406314134597778,
      "learning_rate": 1.2576432175719798e-05,
      "loss": 0.7821,
      "step": 12610
    },
    {
      "epoch": 3.7459186702285545,
      "grad_norm": 1.0151352882385254,
      "learning_rate": 1.2546749777382013e-05,
      "loss": 0.7774,
      "step": 12620
    },
    {
      "epoch": 3.7488869100623328,
      "grad_norm": 0.9649484753608704,
      "learning_rate": 1.2517067379044229e-05,
      "loss": 0.762,
      "step": 12630
    },
    {
      "epoch": 3.7518551498961115,
      "grad_norm": 0.9047630429267883,
      "learning_rate": 1.2487384980706442e-05,
      "loss": 0.7759,
      "step": 12640
    },
    {
      "epoch": 3.75482338972989,
      "grad_norm": 0.9317800402641296,
      "learning_rate": 1.2457702582368656e-05,
      "loss": 0.8132,
      "step": 12650
    },
    {
      "epoch": 3.757791629563669,
      "grad_norm": 1.1614782810211182,
      "learning_rate": 1.2428020184030869e-05,
      "loss": 0.7627,
      "step": 12660
    },
    {
      "epoch": 3.7607598693974476,
      "grad_norm": 0.9428204298019409,
      "learning_rate": 1.2398337785693084e-05,
      "loss": 0.7913,
      "step": 12670
    },
    {
      "epoch": 3.763728109231226,
      "grad_norm": 0.8790725469589233,
      "learning_rate": 1.23686553873553e-05,
      "loss": 0.8273,
      "step": 12680
    },
    {
      "epoch": 3.7666963490650045,
      "grad_norm": 0.8796747922897339,
      "learning_rate": 1.2338972989017514e-05,
      "loss": 0.8031,
      "step": 12690
    },
    {
      "epoch": 3.7696645888987828,
      "grad_norm": 0.8477352857589722,
      "learning_rate": 1.2309290590679727e-05,
      "loss": 0.7771,
      "step": 12700
    },
    {
      "epoch": 3.7726328287325614,
      "grad_norm": 0.9006447196006775,
      "learning_rate": 1.2279608192341941e-05,
      "loss": 0.8053,
      "step": 12710
    },
    {
      "epoch": 3.77560106856634,
      "grad_norm": 0.8733387589454651,
      "learning_rate": 1.2249925794004156e-05,
      "loss": 0.8107,
      "step": 12720
    },
    {
      "epoch": 3.778569308400119,
      "grad_norm": 0.8696321845054626,
      "learning_rate": 1.222024339566637e-05,
      "loss": 0.8131,
      "step": 12730
    },
    {
      "epoch": 3.7815375482338975,
      "grad_norm": 0.8872879147529602,
      "learning_rate": 1.2190560997328585e-05,
      "loss": 0.7902,
      "step": 12740
    },
    {
      "epoch": 3.784505788067676,
      "grad_norm": 0.9091749787330627,
      "learning_rate": 1.21608785989908e-05,
      "loss": 0.796,
      "step": 12750
    },
    {
      "epoch": 3.7874740279014545,
      "grad_norm": 0.8892313838005066,
      "learning_rate": 1.2131196200653014e-05,
      "loss": 0.8292,
      "step": 12760
    },
    {
      "epoch": 3.7904422677352327,
      "grad_norm": 0.9339777827262878,
      "learning_rate": 1.2101513802315227e-05,
      "loss": 0.8062,
      "step": 12770
    },
    {
      "epoch": 3.7934105075690114,
      "grad_norm": 0.9412952065467834,
      "learning_rate": 1.2071831403977441e-05,
      "loss": 0.8165,
      "step": 12780
    },
    {
      "epoch": 3.79637874740279,
      "grad_norm": 0.9553532004356384,
      "learning_rate": 1.2042149005639657e-05,
      "loss": 0.8031,
      "step": 12790
    },
    {
      "epoch": 3.799346987236569,
      "grad_norm": 1.0988775491714478,
      "learning_rate": 1.201246660730187e-05,
      "loss": 0.7942,
      "step": 12800
    },
    {
      "epoch": 3.8023152270703475,
      "grad_norm": 0.9005224704742432,
      "learning_rate": 1.1982784208964085e-05,
      "loss": 0.8045,
      "step": 12810
    },
    {
      "epoch": 3.8052834669041258,
      "grad_norm": 0.833794116973877,
      "learning_rate": 1.1953101810626299e-05,
      "loss": 0.7853,
      "step": 12820
    },
    {
      "epoch": 3.8082517067379045,
      "grad_norm": 0.8762853145599365,
      "learning_rate": 1.1923419412288514e-05,
      "loss": 0.8235,
      "step": 12830
    },
    {
      "epoch": 3.8112199465716827,
      "grad_norm": 0.9304106831550598,
      "learning_rate": 1.1893737013950728e-05,
      "loss": 0.7997,
      "step": 12840
    },
    {
      "epoch": 3.8141881864054614,
      "grad_norm": 0.8896927237510681,
      "learning_rate": 1.1864054615612943e-05,
      "loss": 0.7598,
      "step": 12850
    },
    {
      "epoch": 3.81715642623924,
      "grad_norm": 0.9807896614074707,
      "learning_rate": 1.1834372217275157e-05,
      "loss": 0.8045,
      "step": 12860
    },
    {
      "epoch": 3.820124666073019,
      "grad_norm": 0.9360542297363281,
      "learning_rate": 1.180468981893737e-05,
      "loss": 0.8234,
      "step": 12870
    },
    {
      "epoch": 3.8230929059067975,
      "grad_norm": 1.0262748003005981,
      "learning_rate": 1.1775007420599584e-05,
      "loss": 0.8039,
      "step": 12880
    },
    {
      "epoch": 3.8260611457405758,
      "grad_norm": 0.862922191619873,
      "learning_rate": 1.1745325022261799e-05,
      "loss": 0.8063,
      "step": 12890
    },
    {
      "epoch": 3.8290293855743545,
      "grad_norm": 0.9350073337554932,
      "learning_rate": 1.1715642623924015e-05,
      "loss": 0.7975,
      "step": 12900
    },
    {
      "epoch": 3.8319976254081327,
      "grad_norm": 1.0467214584350586,
      "learning_rate": 1.1685960225586228e-05,
      "loss": 0.7598,
      "step": 12910
    },
    {
      "epoch": 3.8349658652419114,
      "grad_norm": 0.9889064431190491,
      "learning_rate": 1.1656277827248442e-05,
      "loss": 0.7603,
      "step": 12920
    },
    {
      "epoch": 3.83793410507569,
      "grad_norm": 0.9055907130241394,
      "learning_rate": 1.1626595428910657e-05,
      "loss": 0.8148,
      "step": 12930
    },
    {
      "epoch": 3.840902344909469,
      "grad_norm": 0.8573602437973022,
      "learning_rate": 1.159691303057287e-05,
      "loss": 0.8019,
      "step": 12940
    },
    {
      "epoch": 3.8438705847432475,
      "grad_norm": 0.9720578193664551,
      "learning_rate": 1.1567230632235086e-05,
      "loss": 0.8158,
      "step": 12950
    },
    {
      "epoch": 3.8468388245770258,
      "grad_norm": 0.9214889407157898,
      "learning_rate": 1.15375482338973e-05,
      "loss": 0.8281,
      "step": 12960
    },
    {
      "epoch": 3.8498070644108044,
      "grad_norm": 0.9132649302482605,
      "learning_rate": 1.1507865835559515e-05,
      "loss": 0.7465,
      "step": 12970
    },
    {
      "epoch": 3.8527753042445827,
      "grad_norm": 0.9985355734825134,
      "learning_rate": 1.1478183437221727e-05,
      "loss": 0.8046,
      "step": 12980
    },
    {
      "epoch": 3.8557435440783614,
      "grad_norm": 0.8660067915916443,
      "learning_rate": 1.1448501038883942e-05,
      "loss": 0.7853,
      "step": 12990
    },
    {
      "epoch": 3.85871178391214,
      "grad_norm": 0.9778634309768677,
      "learning_rate": 1.1418818640546156e-05,
      "loss": 0.761,
      "step": 13000
    },
    {
      "epoch": 3.861680023745919,
      "grad_norm": 1.0151288509368896,
      "learning_rate": 1.1389136242208371e-05,
      "loss": 0.7662,
      "step": 13010
    },
    {
      "epoch": 3.8646482635796975,
      "grad_norm": 0.975917398929596,
      "learning_rate": 1.1359453843870585e-05,
      "loss": 0.7647,
      "step": 13020
    },
    {
      "epoch": 3.8676165034134757,
      "grad_norm": 0.9875814914703369,
      "learning_rate": 1.13297714455328e-05,
      "loss": 0.8328,
      "step": 13030
    },
    {
      "epoch": 3.8705847432472544,
      "grad_norm": 1.0790032148361206,
      "learning_rate": 1.1300089047195014e-05,
      "loss": 0.7757,
      "step": 13040
    },
    {
      "epoch": 3.8735529830810327,
      "grad_norm": 0.8688693642616272,
      "learning_rate": 1.1270406648857227e-05,
      "loss": 0.8035,
      "step": 13050
    },
    {
      "epoch": 3.8765212229148114,
      "grad_norm": 0.8801321983337402,
      "learning_rate": 1.1240724250519442e-05,
      "loss": 0.8028,
      "step": 13060
    },
    {
      "epoch": 3.87948946274859,
      "grad_norm": 0.9764793515205383,
      "learning_rate": 1.1211041852181658e-05,
      "loss": 0.8054,
      "step": 13070
    },
    {
      "epoch": 3.8824577025823688,
      "grad_norm": 0.9757158160209656,
      "learning_rate": 1.1181359453843872e-05,
      "loss": 0.7634,
      "step": 13080
    },
    {
      "epoch": 3.8854259424161475,
      "grad_norm": 1.0081627368927002,
      "learning_rate": 1.1151677055506085e-05,
      "loss": 0.7735,
      "step": 13090
    },
    {
      "epoch": 3.8883941822499257,
      "grad_norm": 0.9482357501983643,
      "learning_rate": 1.11219946571683e-05,
      "loss": 0.8046,
      "step": 13100
    },
    {
      "epoch": 3.8913624220837044,
      "grad_norm": 0.9332176446914673,
      "learning_rate": 1.1092312258830514e-05,
      "loss": 0.7904,
      "step": 13110
    },
    {
      "epoch": 3.8943306619174827,
      "grad_norm": 0.8813284635543823,
      "learning_rate": 1.1062629860492728e-05,
      "loss": 0.8054,
      "step": 13120
    },
    {
      "epoch": 3.8972989017512614,
      "grad_norm": 1.1398965120315552,
      "learning_rate": 1.1032947462154943e-05,
      "loss": 0.7948,
      "step": 13130
    },
    {
      "epoch": 3.90026714158504,
      "grad_norm": 0.9105324745178223,
      "learning_rate": 1.1003265063817157e-05,
      "loss": 0.7871,
      "step": 13140
    },
    {
      "epoch": 3.9032353814188188,
      "grad_norm": 0.94728684425354,
      "learning_rate": 1.0973582665479372e-05,
      "loss": 0.8171,
      "step": 13150
    },
    {
      "epoch": 3.9062036212525975,
      "grad_norm": 0.9088504314422607,
      "learning_rate": 1.0943900267141585e-05,
      "loss": 0.7787,
      "step": 13160
    },
    {
      "epoch": 3.9091718610863757,
      "grad_norm": 0.9430853724479675,
      "learning_rate": 1.09142178688038e-05,
      "loss": 0.8105,
      "step": 13170
    },
    {
      "epoch": 3.9121401009201544,
      "grad_norm": 1.0075762271881104,
      "learning_rate": 1.0884535470466015e-05,
      "loss": 0.8273,
      "step": 13180
    },
    {
      "epoch": 3.9151083407539327,
      "grad_norm": 0.9297595024108887,
      "learning_rate": 1.0854853072128228e-05,
      "loss": 0.7725,
      "step": 13190
    },
    {
      "epoch": 3.9180765805877114,
      "grad_norm": 0.9361093640327454,
      "learning_rate": 1.0825170673790443e-05,
      "loss": 0.7916,
      "step": 13200
    },
    {
      "epoch": 3.92104482042149,
      "grad_norm": 0.8753396272659302,
      "learning_rate": 1.0795488275452657e-05,
      "loss": 0.8275,
      "step": 13210
    },
    {
      "epoch": 3.9240130602552687,
      "grad_norm": 0.9651039242744446,
      "learning_rate": 1.0765805877114872e-05,
      "loss": 0.815,
      "step": 13220
    },
    {
      "epoch": 3.9269813000890474,
      "grad_norm": 0.9417471289634705,
      "learning_rate": 1.0736123478777086e-05,
      "loss": 0.7957,
      "step": 13230
    },
    {
      "epoch": 3.9299495399228257,
      "grad_norm": 1.0615134239196777,
      "learning_rate": 1.07064410804393e-05,
      "loss": 0.7787,
      "step": 13240
    },
    {
      "epoch": 3.9329177797566044,
      "grad_norm": 0.969096839427948,
      "learning_rate": 1.0676758682101515e-05,
      "loss": 0.7804,
      "step": 13250
    },
    {
      "epoch": 3.9358860195903826,
      "grad_norm": 0.8581018447875977,
      "learning_rate": 1.0647076283763728e-05,
      "loss": 0.7726,
      "step": 13260
    },
    {
      "epoch": 3.9388542594241613,
      "grad_norm": 0.934791088104248,
      "learning_rate": 1.0617393885425942e-05,
      "loss": 0.8016,
      "step": 13270
    },
    {
      "epoch": 3.94182249925794,
      "grad_norm": 0.9198674559593201,
      "learning_rate": 1.0587711487088157e-05,
      "loss": 0.7846,
      "step": 13280
    },
    {
      "epoch": 3.9447907390917187,
      "grad_norm": 0.9727395176887512,
      "learning_rate": 1.0558029088750373e-05,
      "loss": 0.8088,
      "step": 13290
    },
    {
      "epoch": 3.9477589789254974,
      "grad_norm": 0.8742444515228271,
      "learning_rate": 1.0528346690412586e-05,
      "loss": 0.8278,
      "step": 13300
    },
    {
      "epoch": 3.9507272187592757,
      "grad_norm": 1.0352386236190796,
      "learning_rate": 1.04986642920748e-05,
      "loss": 0.8211,
      "step": 13310
    },
    {
      "epoch": 3.9536954585930544,
      "grad_norm": 0.9503917694091797,
      "learning_rate": 1.0468981893737015e-05,
      "loss": 0.8355,
      "step": 13320
    },
    {
      "epoch": 3.9566636984268326,
      "grad_norm": 1.0704954862594604,
      "learning_rate": 1.0439299495399228e-05,
      "loss": 0.803,
      "step": 13330
    },
    {
      "epoch": 3.9596319382606113,
      "grad_norm": 1.072367548942566,
      "learning_rate": 1.0409617097061444e-05,
      "loss": 0.8187,
      "step": 13340
    },
    {
      "epoch": 3.96260017809439,
      "grad_norm": 0.9495874643325806,
      "learning_rate": 1.0379934698723658e-05,
      "loss": 0.7955,
      "step": 13350
    },
    {
      "epoch": 3.9655684179281687,
      "grad_norm": 0.9812706112861633,
      "learning_rate": 1.0350252300385873e-05,
      "loss": 0.8062,
      "step": 13360
    },
    {
      "epoch": 3.9685366577619474,
      "grad_norm": 0.8435708284378052,
      "learning_rate": 1.0320569902048085e-05,
      "loss": 0.8347,
      "step": 13370
    },
    {
      "epoch": 3.9715048975957257,
      "grad_norm": 0.9395439624786377,
      "learning_rate": 1.02908875037103e-05,
      "loss": 0.7698,
      "step": 13380
    },
    {
      "epoch": 3.9744731374295044,
      "grad_norm": 1.0042177438735962,
      "learning_rate": 1.0261205105372514e-05,
      "loss": 0.8013,
      "step": 13390
    },
    {
      "epoch": 3.9774413772632826,
      "grad_norm": 0.8969469666481018,
      "learning_rate": 1.0231522707034729e-05,
      "loss": 0.7825,
      "step": 13400
    },
    {
      "epoch": 3.9804096170970613,
      "grad_norm": 0.8615506887435913,
      "learning_rate": 1.0201840308696943e-05,
      "loss": 0.8354,
      "step": 13410
    },
    {
      "epoch": 3.98337785693084,
      "grad_norm": 0.893427312374115,
      "learning_rate": 1.0172157910359158e-05,
      "loss": 0.7794,
      "step": 13420
    },
    {
      "epoch": 3.9863460967646187,
      "grad_norm": 0.9048050045967102,
      "learning_rate": 1.0142475512021372e-05,
      "loss": 0.784,
      "step": 13430
    },
    {
      "epoch": 3.9893143365983974,
      "grad_norm": 0.9938015937805176,
      "learning_rate": 1.0112793113683585e-05,
      "loss": 0.8001,
      "step": 13440
    },
    {
      "epoch": 3.9922825764321757,
      "grad_norm": 0.8949004411697388,
      "learning_rate": 1.00831107153458e-05,
      "loss": 0.7774,
      "step": 13450
    },
    {
      "epoch": 3.9952508162659544,
      "grad_norm": 0.9906232953071594,
      "learning_rate": 1.0056396556841794e-05,
      "loss": 0.7881,
      "step": 13460
    },
    {
      "epoch": 3.9982190560997326,
      "grad_norm": 0.8567777276039124,
      "learning_rate": 1.0026714158504008e-05,
      "loss": 0.803,
      "step": 13470
    },
    {
      "epoch": 4.001187295933511,
      "grad_norm": 0.8931456208229065,
      "learning_rate": 9.997031760166221e-06,
      "loss": 0.7788,
      "step": 13480
    },
    {
      "epoch": 4.00415553576729,
      "grad_norm": 0.9593549370765686,
      "learning_rate": 9.967349361828436e-06,
      "loss": 0.7651,
      "step": 13490
    },
    {
      "epoch": 4.007123775601069,
      "grad_norm": 0.9598982930183411,
      "learning_rate": 9.93766696349065e-06,
      "loss": 0.7464,
      "step": 13500
    },
    {
      "epoch": 4.010092015434847,
      "grad_norm": 0.9251787662506104,
      "learning_rate": 9.907984565152866e-06,
      "loss": 0.7754,
      "step": 13510
    },
    {
      "epoch": 4.013060255268626,
      "grad_norm": 0.9092532992362976,
      "learning_rate": 9.878302166815079e-06,
      "loss": 0.766,
      "step": 13520
    },
    {
      "epoch": 4.016028495102404,
      "grad_norm": 1.02048921585083,
      "learning_rate": 9.848619768477293e-06,
      "loss": 0.741,
      "step": 13530
    },
    {
      "epoch": 4.018996734936183,
      "grad_norm": 0.9862306118011475,
      "learning_rate": 9.818937370139508e-06,
      "loss": 0.686,
      "step": 13540
    },
    {
      "epoch": 4.021964974769961,
      "grad_norm": 0.9413052797317505,
      "learning_rate": 9.78925497180172e-06,
      "loss": 0.7775,
      "step": 13550
    },
    {
      "epoch": 4.02493321460374,
      "grad_norm": 0.9390004873275757,
      "learning_rate": 9.759572573463937e-06,
      "loss": 0.7452,
      "step": 13560
    },
    {
      "epoch": 4.027901454437519,
      "grad_norm": 0.9862014055252075,
      "learning_rate": 9.729890175126151e-06,
      "loss": 0.741,
      "step": 13570
    },
    {
      "epoch": 4.030869694271297,
      "grad_norm": 0.9327501654624939,
      "learning_rate": 9.700207776788366e-06,
      "loss": 0.778,
      "step": 13580
    },
    {
      "epoch": 4.033837934105076,
      "grad_norm": 0.9384015798568726,
      "learning_rate": 9.670525378450579e-06,
      "loss": 0.784,
      "step": 13590
    },
    {
      "epoch": 4.036806173938854,
      "grad_norm": 0.8979681134223938,
      "learning_rate": 9.640842980112793e-06,
      "loss": 0.7673,
      "step": 13600
    },
    {
      "epoch": 4.039774413772633,
      "grad_norm": 0.9125372171401978,
      "learning_rate": 9.611160581775008e-06,
      "loss": 0.7654,
      "step": 13610
    },
    {
      "epoch": 4.042742653606411,
      "grad_norm": 0.9703123569488525,
      "learning_rate": 9.581478183437222e-06,
      "loss": 0.7805,
      "step": 13620
    },
    {
      "epoch": 4.04571089344019,
      "grad_norm": 1.0698387622833252,
      "learning_rate": 9.551795785099437e-06,
      "loss": 0.7656,
      "step": 13630
    },
    {
      "epoch": 4.048679133273969,
      "grad_norm": 0.8759061098098755,
      "learning_rate": 9.522113386761651e-06,
      "loss": 0.7672,
      "step": 13640
    },
    {
      "epoch": 4.051647373107747,
      "grad_norm": 1.0125908851623535,
      "learning_rate": 9.492430988423866e-06,
      "loss": 0.773,
      "step": 13650
    },
    {
      "epoch": 4.054615612941526,
      "grad_norm": 1.0893434286117554,
      "learning_rate": 9.462748590086078e-06,
      "loss": 0.7741,
      "step": 13660
    },
    {
      "epoch": 4.057583852775304,
      "grad_norm": 0.9895984530448914,
      "learning_rate": 9.433066191748295e-06,
      "loss": 0.723,
      "step": 13670
    },
    {
      "epoch": 4.060552092609083,
      "grad_norm": 1.0161770582199097,
      "learning_rate": 9.403383793410509e-06,
      "loss": 0.7353,
      "step": 13680
    },
    {
      "epoch": 4.063520332442861,
      "grad_norm": 0.9664089679718018,
      "learning_rate": 9.373701395072722e-06,
      "loss": 0.7777,
      "step": 13690
    },
    {
      "epoch": 4.06648857227664,
      "grad_norm": 1.0124640464782715,
      "learning_rate": 9.344018996734936e-06,
      "loss": 0.7382,
      "step": 13700
    },
    {
      "epoch": 4.069456812110419,
      "grad_norm": 1.000227689743042,
      "learning_rate": 9.31433659839715e-06,
      "loss": 0.7753,
      "step": 13710
    },
    {
      "epoch": 4.072425051944197,
      "grad_norm": 1.014361023902893,
      "learning_rate": 9.284654200059365e-06,
      "loss": 0.7517,
      "step": 13720
    },
    {
      "epoch": 4.075393291777976,
      "grad_norm": 0.9377306699752808,
      "learning_rate": 9.25497180172158e-06,
      "loss": 0.7417,
      "step": 13730
    },
    {
      "epoch": 4.078361531611754,
      "grad_norm": 1.0277228355407715,
      "learning_rate": 9.225289403383794e-06,
      "loss": 0.7262,
      "step": 13740
    },
    {
      "epoch": 4.0813297714455326,
      "grad_norm": 0.9528384804725647,
      "learning_rate": 9.195607005046009e-06,
      "loss": 0.7734,
      "step": 13750
    },
    {
      "epoch": 4.084298011279311,
      "grad_norm": 1.0263906717300415,
      "learning_rate": 9.165924606708221e-06,
      "loss": 0.787,
      "step": 13760
    },
    {
      "epoch": 4.08726625111309,
      "grad_norm": 1.0394752025604248,
      "learning_rate": 9.136242208370436e-06,
      "loss": 0.7997,
      "step": 13770
    },
    {
      "epoch": 4.090234490946869,
      "grad_norm": 1.042912244796753,
      "learning_rate": 9.10655981003265e-06,
      "loss": 0.7714,
      "step": 13780
    },
    {
      "epoch": 4.093202730780647,
      "grad_norm": 1.1661458015441895,
      "learning_rate": 9.076877411694867e-06,
      "loss": 0.7696,
      "step": 13790
    },
    {
      "epoch": 4.096170970614426,
      "grad_norm": 0.9394704699516296,
      "learning_rate": 9.04719501335708e-06,
      "loss": 0.7691,
      "step": 13800
    },
    {
      "epoch": 4.099139210448204,
      "grad_norm": 0.9088842272758484,
      "learning_rate": 9.017512615019294e-06,
      "loss": 0.7794,
      "step": 13810
    },
    {
      "epoch": 4.1021074502819825,
      "grad_norm": 0.8951979279518127,
      "learning_rate": 8.987830216681508e-06,
      "loss": 0.74,
      "step": 13820
    },
    {
      "epoch": 4.105075690115761,
      "grad_norm": 0.9268091917037964,
      "learning_rate": 8.958147818343721e-06,
      "loss": 0.7488,
      "step": 13830
    },
    {
      "epoch": 4.10804392994954,
      "grad_norm": 0.9783716797828674,
      "learning_rate": 8.928465420005937e-06,
      "loss": 0.7325,
      "step": 13840
    },
    {
      "epoch": 4.111012169783319,
      "grad_norm": 1.072424292564392,
      "learning_rate": 8.898783021668152e-06,
      "loss": 0.7322,
      "step": 13850
    },
    {
      "epoch": 4.113980409617097,
      "grad_norm": 0.9002495408058167,
      "learning_rate": 8.869100623330366e-06,
      "loss": 0.7884,
      "step": 13860
    },
    {
      "epoch": 4.116948649450876,
      "grad_norm": 0.8875764012336731,
      "learning_rate": 8.839418224992579e-06,
      "loss": 0.7118,
      "step": 13870
    },
    {
      "epoch": 4.119916889284654,
      "grad_norm": 1.01841402053833,
      "learning_rate": 8.809735826654794e-06,
      "loss": 0.7226,
      "step": 13880
    },
    {
      "epoch": 4.1228851291184325,
      "grad_norm": 0.9352704286575317,
      "learning_rate": 8.780053428317008e-06,
      "loss": 0.7599,
      "step": 13890
    },
    {
      "epoch": 4.125853368952211,
      "grad_norm": 0.9383290410041809,
      "learning_rate": 8.750371029979223e-06,
      "loss": 0.7636,
      "step": 13900
    },
    {
      "epoch": 4.12882160878599,
      "grad_norm": 0.9522103071212769,
      "learning_rate": 8.720688631641437e-06,
      "loss": 0.7806,
      "step": 13910
    },
    {
      "epoch": 4.131789848619769,
      "grad_norm": 0.9566015601158142,
      "learning_rate": 8.691006233303652e-06,
      "loss": 0.729,
      "step": 13920
    },
    {
      "epoch": 4.134758088453547,
      "grad_norm": 0.9936237335205078,
      "learning_rate": 8.661323834965866e-06,
      "loss": 0.786,
      "step": 13930
    },
    {
      "epoch": 4.137726328287326,
      "grad_norm": 1.0021575689315796,
      "learning_rate": 8.631641436628079e-06,
      "loss": 0.7579,
      "step": 13940
    },
    {
      "epoch": 4.140694568121104,
      "grad_norm": 0.9321087002754211,
      "learning_rate": 8.601959038290295e-06,
      "loss": 0.7369,
      "step": 13950
    },
    {
      "epoch": 4.1436628079548825,
      "grad_norm": 0.9588301777839661,
      "learning_rate": 8.57227663995251e-06,
      "loss": 0.7365,
      "step": 13960
    },
    {
      "epoch": 4.146631047788661,
      "grad_norm": 1.0697441101074219,
      "learning_rate": 8.542594241614722e-06,
      "loss": 0.7313,
      "step": 13970
    },
    {
      "epoch": 4.14959928762244,
      "grad_norm": 0.9477381110191345,
      "learning_rate": 8.512911843276937e-06,
      "loss": 0.7815,
      "step": 13980
    },
    {
      "epoch": 4.152567527456219,
      "grad_norm": 0.915964663028717,
      "learning_rate": 8.483229444939151e-06,
      "loss": 0.7679,
      "step": 13990
    },
    {
      "epoch": 4.155535767289997,
      "grad_norm": 0.9938797354698181,
      "learning_rate": 8.453547046601366e-06,
      "loss": 0.7535,
      "step": 14000
    },
    {
      "epoch": 4.158504007123776,
      "grad_norm": 0.9842844009399414,
      "learning_rate": 8.42386464826358e-06,
      "loss": 0.732,
      "step": 14010
    },
    {
      "epoch": 4.161472246957554,
      "grad_norm": 0.9590873122215271,
      "learning_rate": 8.394182249925795e-06,
      "loss": 0.7565,
      "step": 14020
    },
    {
      "epoch": 4.1644404867913325,
      "grad_norm": 0.9819791316986084,
      "learning_rate": 8.364499851588009e-06,
      "loss": 0.7559,
      "step": 14030
    },
    {
      "epoch": 4.167408726625111,
      "grad_norm": 1.014815092086792,
      "learning_rate": 8.334817453250222e-06,
      "loss": 0.794,
      "step": 14040
    },
    {
      "epoch": 4.17037696645889,
      "grad_norm": 1.020995855331421,
      "learning_rate": 8.305135054912436e-06,
      "loss": 0.736,
      "step": 14050
    },
    {
      "epoch": 4.173345206292669,
      "grad_norm": 1.0287991762161255,
      "learning_rate": 8.275452656574653e-06,
      "loss": 0.7423,
      "step": 14060
    },
    {
      "epoch": 4.176313446126447,
      "grad_norm": 0.9976941347122192,
      "learning_rate": 8.245770258236867e-06,
      "loss": 0.7581,
      "step": 14070
    },
    {
      "epoch": 4.179281685960226,
      "grad_norm": 0.9983686804771423,
      "learning_rate": 8.21608785989908e-06,
      "loss": 0.7418,
      "step": 14080
    },
    {
      "epoch": 4.182249925794004,
      "grad_norm": 0.8948219418525696,
      "learning_rate": 8.186405461561294e-06,
      "loss": 0.7374,
      "step": 14090
    },
    {
      "epoch": 4.1852181656277825,
      "grad_norm": 0.9339085221290588,
      "learning_rate": 8.156723063223509e-06,
      "loss": 0.7389,
      "step": 14100
    },
    {
      "epoch": 4.188186405461561,
      "grad_norm": 0.9872913956642151,
      "learning_rate": 8.127040664885723e-06,
      "loss": 0.7815,
      "step": 14110
    },
    {
      "epoch": 4.19115464529534,
      "grad_norm": 1.003295660018921,
      "learning_rate": 8.097358266547938e-06,
      "loss": 0.7535,
      "step": 14120
    },
    {
      "epoch": 4.194122885129119,
      "grad_norm": 0.9385514259338379,
      "learning_rate": 8.067675868210152e-06,
      "loss": 0.7293,
      "step": 14130
    },
    {
      "epoch": 4.197091124962897,
      "grad_norm": 1.1677844524383545,
      "learning_rate": 8.037993469872367e-06,
      "loss": 0.7673,
      "step": 14140
    },
    {
      "epoch": 4.200059364796676,
      "grad_norm": 1.0024638175964355,
      "learning_rate": 8.00831107153458e-06,
      "loss": 0.7371,
      "step": 14150
    },
    {
      "epoch": 4.203027604630454,
      "grad_norm": 1.0117018222808838,
      "learning_rate": 7.978628673196794e-06,
      "loss": 0.7823,
      "step": 14160
    },
    {
      "epoch": 4.2059958444642325,
      "grad_norm": 0.9616550803184509,
      "learning_rate": 7.94894627485901e-06,
      "loss": 0.7307,
      "step": 14170
    },
    {
      "epoch": 4.208964084298011,
      "grad_norm": 0.9083321690559387,
      "learning_rate": 7.919263876521223e-06,
      "loss": 0.7666,
      "step": 14180
    },
    {
      "epoch": 4.21193232413179,
      "grad_norm": 1.014846682548523,
      "learning_rate": 7.889581478183437e-06,
      "loss": 0.7621,
      "step": 14190
    },
    {
      "epoch": 4.214900563965569,
      "grad_norm": 1.0015004873275757,
      "learning_rate": 7.859899079845652e-06,
      "loss": 0.7747,
      "step": 14200
    },
    {
      "epoch": 4.217868803799347,
      "grad_norm": 1.0885722637176514,
      "learning_rate": 7.830216681507866e-06,
      "loss": 0.7743,
      "step": 14210
    },
    {
      "epoch": 4.220837043633126,
      "grad_norm": 1.0359748601913452,
      "learning_rate": 7.800534283170081e-06,
      "loss": 0.7793,
      "step": 14220
    },
    {
      "epoch": 4.223805283466904,
      "grad_norm": 0.9756050109863281,
      "learning_rate": 7.770851884832295e-06,
      "loss": 0.7608,
      "step": 14230
    },
    {
      "epoch": 4.2267735233006825,
      "grad_norm": 1.0150599479675293,
      "learning_rate": 7.74116948649451e-06,
      "loss": 0.7263,
      "step": 14240
    },
    {
      "epoch": 4.229741763134461,
      "grad_norm": 0.9713482856750488,
      "learning_rate": 7.711487088156723e-06,
      "loss": 0.7623,
      "step": 14250
    },
    {
      "epoch": 4.23271000296824,
      "grad_norm": 0.9168558120727539,
      "learning_rate": 7.681804689818937e-06,
      "loss": 0.7639,
      "step": 14260
    },
    {
      "epoch": 4.2356782428020185,
      "grad_norm": 0.9601516723632812,
      "learning_rate": 7.652122291481152e-06,
      "loss": 0.7779,
      "step": 14270
    },
    {
      "epoch": 4.238646482635797,
      "grad_norm": 0.9309169054031372,
      "learning_rate": 7.622439893143367e-06,
      "loss": 0.763,
      "step": 14280
    },
    {
      "epoch": 4.241614722469576,
      "grad_norm": 1.023237943649292,
      "learning_rate": 7.5927574948055806e-06,
      "loss": 0.7467,
      "step": 14290
    },
    {
      "epoch": 4.244582962303354,
      "grad_norm": 1.0153001546859741,
      "learning_rate": 7.563075096467795e-06,
      "loss": 0.7316,
      "step": 14300
    },
    {
      "epoch": 4.247551202137132,
      "grad_norm": 1.017930269241333,
      "learning_rate": 7.5333926981300095e-06,
      "loss": 0.761,
      "step": 14310
    },
    {
      "epoch": 4.250519441970911,
      "grad_norm": 1.10524582862854,
      "learning_rate": 7.503710299792223e-06,
      "loss": 0.7484,
      "step": 14320
    },
    {
      "epoch": 4.25348768180469,
      "grad_norm": 1.083328366279602,
      "learning_rate": 7.474027901454438e-06,
      "loss": 0.7791,
      "step": 14330
    },
    {
      "epoch": 4.2564559216384685,
      "grad_norm": 0.9794231057167053,
      "learning_rate": 7.444345503116652e-06,
      "loss": 0.7829,
      "step": 14340
    },
    {
      "epoch": 4.259424161472247,
      "grad_norm": 1.0356334447860718,
      "learning_rate": 7.414663104778867e-06,
      "loss": 0.7279,
      "step": 14350
    },
    {
      "epoch": 4.262392401306026,
      "grad_norm": 0.8205145001411438,
      "learning_rate": 7.38498070644108e-06,
      "loss": 0.7455,
      "step": 14360
    },
    {
      "epoch": 4.265360641139804,
      "grad_norm": 1.0228333473205566,
      "learning_rate": 7.355298308103295e-06,
      "loss": 0.7565,
      "step": 14370
    },
    {
      "epoch": 4.268328880973582,
      "grad_norm": 1.025168538093567,
      "learning_rate": 7.32561590976551e-06,
      "loss": 0.7615,
      "step": 14380
    },
    {
      "epoch": 4.271297120807361,
      "grad_norm": 1.0346413850784302,
      "learning_rate": 7.295933511427723e-06,
      "loss": 0.7875,
      "step": 14390
    },
    {
      "epoch": 4.27426536064114,
      "grad_norm": 0.9277398586273193,
      "learning_rate": 7.266251113089937e-06,
      "loss": 0.7342,
      "step": 14400
    },
    {
      "epoch": 4.2772336004749185,
      "grad_norm": 1.0324021577835083,
      "learning_rate": 7.236568714752153e-06,
      "loss": 0.7162,
      "step": 14410
    },
    {
      "epoch": 4.280201840308697,
      "grad_norm": 0.9764350652694702,
      "learning_rate": 7.206886316414367e-06,
      "loss": 0.7526,
      "step": 14420
    },
    {
      "epoch": 4.283170080142476,
      "grad_norm": 0.994429886341095,
      "learning_rate": 7.177203918076581e-06,
      "loss": 0.7379,
      "step": 14430
    },
    {
      "epoch": 4.286138319976254,
      "grad_norm": 0.9952360987663269,
      "learning_rate": 7.147521519738795e-06,
      "loss": 0.7566,
      "step": 14440
    },
    {
      "epoch": 4.289106559810032,
      "grad_norm": 1.0014885663986206,
      "learning_rate": 7.11783912140101e-06,
      "loss": 0.7761,
      "step": 14450
    },
    {
      "epoch": 4.292074799643811,
      "grad_norm": 0.9838289618492126,
      "learning_rate": 7.088156723063224e-06,
      "loss": 0.7375,
      "step": 14460
    },
    {
      "epoch": 4.29504303947759,
      "grad_norm": 0.907821536064148,
      "learning_rate": 7.058474324725438e-06,
      "loss": 0.7259,
      "step": 14470
    },
    {
      "epoch": 4.2980112793113685,
      "grad_norm": 1.0343751907348633,
      "learning_rate": 7.028791926387652e-06,
      "loss": 0.7805,
      "step": 14480
    },
    {
      "epoch": 4.300979519145147,
      "grad_norm": 0.8860198259353638,
      "learning_rate": 6.999109528049868e-06,
      "loss": 0.7796,
      "step": 14490
    },
    {
      "epoch": 4.303947758978926,
      "grad_norm": 1.0006810426712036,
      "learning_rate": 6.9694271297120805e-06,
      "loss": 0.7631,
      "step": 14500
    },
    {
      "epoch": 4.306915998812704,
      "grad_norm": 0.976875364780426,
      "learning_rate": 6.939744731374295e-06,
      "loss": 0.7539,
      "step": 14510
    },
    {
      "epoch": 4.309884238646482,
      "grad_norm": 0.9557456374168396,
      "learning_rate": 6.91006233303651e-06,
      "loss": 0.7961,
      "step": 14520
    },
    {
      "epoch": 4.312852478480261,
      "grad_norm": 0.9540473818778992,
      "learning_rate": 6.880379934698725e-06,
      "loss": 0.7522,
      "step": 14530
    },
    {
      "epoch": 4.31582071831404,
      "grad_norm": 1.0581246614456177,
      "learning_rate": 6.850697536360938e-06,
      "loss": 0.7724,
      "step": 14540
    },
    {
      "epoch": 4.3187889581478185,
      "grad_norm": 1.0394785404205322,
      "learning_rate": 6.821015138023153e-06,
      "loss": 0.7827,
      "step": 14550
    },
    {
      "epoch": 4.321757197981597,
      "grad_norm": 0.9216113686561584,
      "learning_rate": 6.791332739685367e-06,
      "loss": 0.7613,
      "step": 14560
    },
    {
      "epoch": 4.324725437815376,
      "grad_norm": 0.980540931224823,
      "learning_rate": 6.761650341347581e-06,
      "loss": 0.7943,
      "step": 14570
    },
    {
      "epoch": 4.327693677649154,
      "grad_norm": 0.999685525894165,
      "learning_rate": 6.7319679430097955e-06,
      "loss": 0.7401,
      "step": 14580
    },
    {
      "epoch": 4.330661917482932,
      "grad_norm": 0.9493926167488098,
      "learning_rate": 6.70228554467201e-06,
      "loss": 0.7254,
      "step": 14590
    },
    {
      "epoch": 4.333630157316711,
      "grad_norm": 0.9986134767532349,
      "learning_rate": 6.6726031463342244e-06,
      "loss": 0.7624,
      "step": 14600
    },
    {
      "epoch": 4.33659839715049,
      "grad_norm": 0.9402568340301514,
      "learning_rate": 6.642920747996438e-06,
      "loss": 0.7523,
      "step": 14610
    },
    {
      "epoch": 4.3395666369842685,
      "grad_norm": 1.1779499053955078,
      "learning_rate": 6.6132383496586525e-06,
      "loss": 0.7676,
      "step": 14620
    },
    {
      "epoch": 4.342534876818047,
      "grad_norm": 0.9890404343605042,
      "learning_rate": 6.583555951320868e-06,
      "loss": 0.7462,
      "step": 14630
    },
    {
      "epoch": 4.345503116651826,
      "grad_norm": 1.0612539052963257,
      "learning_rate": 6.553873552983081e-06,
      "loss": 0.7368,
      "step": 14640
    },
    {
      "epoch": 4.348471356485604,
      "grad_norm": 0.9921562671661377,
      "learning_rate": 6.524191154645295e-06,
      "loss": 0.7644,
      "step": 14650
    },
    {
      "epoch": 4.351439596319382,
      "grad_norm": 0.9459686279296875,
      "learning_rate": 6.4945087563075105e-06,
      "loss": 0.7567,
      "step": 14660
    },
    {
      "epoch": 4.354407836153161,
      "grad_norm": 0.941131055355072,
      "learning_rate": 6.464826357969725e-06,
      "loss": 0.7603,
      "step": 14670
    },
    {
      "epoch": 4.35737607598694,
      "grad_norm": 0.8995440006256104,
      "learning_rate": 6.435143959631939e-06,
      "loss": 0.767,
      "step": 14680
    },
    {
      "epoch": 4.3603443158207185,
      "grad_norm": 0.9734925031661987,
      "learning_rate": 6.405461561294153e-06,
      "loss": 0.7548,
      "step": 14690
    },
    {
      "epoch": 4.363312555654497,
      "grad_norm": 0.9570586681365967,
      "learning_rate": 6.3757791629563676e-06,
      "loss": 0.7515,
      "step": 14700
    },
    {
      "epoch": 4.366280795488276,
      "grad_norm": 0.9729474782943726,
      "learning_rate": 6.346096764618581e-06,
      "loss": 0.7688,
      "step": 14710
    },
    {
      "epoch": 4.369249035322054,
      "grad_norm": 0.9383680820465088,
      "learning_rate": 6.316414366280796e-06,
      "loss": 0.7651,
      "step": 14720
    },
    {
      "epoch": 4.372217275155832,
      "grad_norm": 1.0096242427825928,
      "learning_rate": 6.28673196794301e-06,
      "loss": 0.7359,
      "step": 14730
    },
    {
      "epoch": 4.375185514989611,
      "grad_norm": 0.9830286502838135,
      "learning_rate": 6.257049569605225e-06,
      "loss": 0.7314,
      "step": 14740
    },
    {
      "epoch": 4.37815375482339,
      "grad_norm": 0.9673001170158386,
      "learning_rate": 6.227367171267439e-06,
      "loss": 0.7388,
      "step": 14750
    },
    {
      "epoch": 4.3811219946571685,
      "grad_norm": 1.0183544158935547,
      "learning_rate": 6.197684772929653e-06,
      "loss": 0.7747,
      "step": 14760
    },
    {
      "epoch": 4.384090234490947,
      "grad_norm": 1.0080556869506836,
      "learning_rate": 6.168002374591867e-06,
      "loss": 0.7772,
      "step": 14770
    },
    {
      "epoch": 4.387058474324726,
      "grad_norm": 1.1509473323822021,
      "learning_rate": 6.138319976254082e-06,
      "loss": 0.7222,
      "step": 14780
    },
    {
      "epoch": 4.390026714158504,
      "grad_norm": 0.9294554591178894,
      "learning_rate": 6.108637577916295e-06,
      "loss": 0.7475,
      "step": 14790
    },
    {
      "epoch": 4.392994953992282,
      "grad_norm": 0.8813176155090332,
      "learning_rate": 6.078955179578511e-06,
      "loss": 0.7427,
      "step": 14800
    },
    {
      "epoch": 4.395963193826061,
      "grad_norm": 1.0649441480636597,
      "learning_rate": 6.049272781240724e-06,
      "loss": 0.7343,
      "step": 14810
    },
    {
      "epoch": 4.39893143365984,
      "grad_norm": 1.050569772720337,
      "learning_rate": 6.019590382902939e-06,
      "loss": 0.7447,
      "step": 14820
    },
    {
      "epoch": 4.401899673493618,
      "grad_norm": 1.0193036794662476,
      "learning_rate": 5.989907984565153e-06,
      "loss": 0.7241,
      "step": 14830
    },
    {
      "epoch": 4.404867913327397,
      "grad_norm": 1.064293622970581,
      "learning_rate": 5.960225586227367e-06,
      "loss": 0.7298,
      "step": 14840
    },
    {
      "epoch": 4.407836153161176,
      "grad_norm": 0.9010776281356812,
      "learning_rate": 5.930543187889582e-06,
      "loss": 0.7411,
      "step": 14850
    },
    {
      "epoch": 4.410804392994954,
      "grad_norm": 0.9958186745643616,
      "learning_rate": 5.900860789551796e-06,
      "loss": 0.7527,
      "step": 14860
    },
    {
      "epoch": 4.413772632828732,
      "grad_norm": 1.0218920707702637,
      "learning_rate": 5.87117839121401e-06,
      "loss": 0.7765,
      "step": 14870
    },
    {
      "epoch": 4.416740872662511,
      "grad_norm": 0.9252408742904663,
      "learning_rate": 5.841495992876225e-06,
      "loss": 0.7227,
      "step": 14880
    },
    {
      "epoch": 4.41970911249629,
      "grad_norm": 1.0993282794952393,
      "learning_rate": 5.811813594538439e-06,
      "loss": 0.7219,
      "step": 14890
    },
    {
      "epoch": 4.422677352330068,
      "grad_norm": 0.9480684399604797,
      "learning_rate": 5.782131196200653e-06,
      "loss": 0.7533,
      "step": 14900
    },
    {
      "epoch": 4.425645592163847,
      "grad_norm": 1.0021878480911255,
      "learning_rate": 5.7524487978628674e-06,
      "loss": 0.7884,
      "step": 14910
    },
    {
      "epoch": 4.428613831997626,
      "grad_norm": 0.9122134447097778,
      "learning_rate": 5.722766399525082e-06,
      "loss": 0.741,
      "step": 14920
    },
    {
      "epoch": 4.431582071831404,
      "grad_norm": 0.956417441368103,
      "learning_rate": 5.693084001187296e-06,
      "loss": 0.7292,
      "step": 14930
    },
    {
      "epoch": 4.434550311665182,
      "grad_norm": 0.9715415835380554,
      "learning_rate": 5.663401602849511e-06,
      "loss": 0.7488,
      "step": 14940
    },
    {
      "epoch": 4.437518551498961,
      "grad_norm": 1.014623999595642,
      "learning_rate": 5.6337192045117245e-06,
      "loss": 0.7594,
      "step": 14950
    },
    {
      "epoch": 4.44048679133274,
      "grad_norm": 0.9471456408500671,
      "learning_rate": 5.604036806173939e-06,
      "loss": 0.7512,
      "step": 14960
    },
    {
      "epoch": 4.443455031166518,
      "grad_norm": 1.072137713432312,
      "learning_rate": 5.5743544078361535e-06,
      "loss": 0.7842,
      "step": 14970
    },
    {
      "epoch": 4.446423271000297,
      "grad_norm": 1.037337303161621,
      "learning_rate": 5.544672009498367e-06,
      "loss": 0.7904,
      "step": 14980
    },
    {
      "epoch": 4.449391510834076,
      "grad_norm": 0.995496928691864,
      "learning_rate": 5.5149896111605824e-06,
      "loss": 0.744,
      "step": 14990
    },
    {
      "epoch": 4.452359750667854,
      "grad_norm": 1.038312315940857,
      "learning_rate": 5.485307212822796e-06,
      "loss": 0.7474,
      "step": 15000
    },
    {
      "epoch": 4.455327990501632,
      "grad_norm": 1.0620917081832886,
      "learning_rate": 5.4556248144850106e-06,
      "loss": 0.753,
      "step": 15010
    },
    {
      "epoch": 4.458296230335411,
      "grad_norm": 1.2397912740707397,
      "learning_rate": 5.425942416147225e-06,
      "loss": 0.767,
      "step": 15020
    },
    {
      "epoch": 4.46126447016919,
      "grad_norm": 0.911263644695282,
      "learning_rate": 5.3962600178094395e-06,
      "loss": 0.717,
      "step": 15030
    },
    {
      "epoch": 4.464232710002968,
      "grad_norm": 1.0086246728897095,
      "learning_rate": 5.366577619471653e-06,
      "loss": 0.7645,
      "step": 15040
    },
    {
      "epoch": 4.467200949836747,
      "grad_norm": 1.0217297077178955,
      "learning_rate": 5.336895221133868e-06,
      "loss": 0.7424,
      "step": 15050
    },
    {
      "epoch": 4.470169189670526,
      "grad_norm": 1.0628533363342285,
      "learning_rate": 5.307212822796082e-06,
      "loss": 0.7538,
      "step": 15060
    },
    {
      "epoch": 4.473137429504304,
      "grad_norm": 0.9455434083938599,
      "learning_rate": 5.277530424458297e-06,
      "loss": 0.738,
      "step": 15070
    },
    {
      "epoch": 4.476105669338082,
      "grad_norm": 0.9059712290763855,
      "learning_rate": 5.247848026120511e-06,
      "loss": 0.7403,
      "step": 15080
    },
    {
      "epoch": 4.479073909171861,
      "grad_norm": 1.0143680572509766,
      "learning_rate": 5.218165627782725e-06,
      "loss": 0.7271,
      "step": 15090
    },
    {
      "epoch": 4.48204214900564,
      "grad_norm": 1.046194314956665,
      "learning_rate": 5.18848322944494e-06,
      "loss": 0.8082,
      "step": 15100
    },
    {
      "epoch": 4.485010388839418,
      "grad_norm": 0.9987038969993591,
      "learning_rate": 5.158800831107154e-06,
      "loss": 0.7505,
      "step": 15110
    },
    {
      "epoch": 4.487978628673197,
      "grad_norm": 0.983883798122406,
      "learning_rate": 5.129118432769367e-06,
      "loss": 0.7461,
      "step": 15120
    },
    {
      "epoch": 4.490946868506976,
      "grad_norm": 1.0110454559326172,
      "learning_rate": 5.099436034431583e-06,
      "loss": 0.7858,
      "step": 15130
    },
    {
      "epoch": 4.493915108340754,
      "grad_norm": 1.2014484405517578,
      "learning_rate": 5.069753636093796e-06,
      "loss": 0.7998,
      "step": 15140
    },
    {
      "epoch": 4.496883348174532,
      "grad_norm": 0.9484744071960449,
      "learning_rate": 5.040071237756011e-06,
      "loss": 0.7701,
      "step": 15150
    },
    {
      "epoch": 4.499851588008311,
      "grad_norm": 1.0080126523971558,
      "learning_rate": 5.010388839418225e-06,
      "loss": 0.7773,
      "step": 15160
    },
    {
      "epoch": 4.50281982784209,
      "grad_norm": 1.0008039474487305,
      "learning_rate": 4.98070644108044e-06,
      "loss": 0.7289,
      "step": 15170
    },
    {
      "epoch": 4.505788067675868,
      "grad_norm": 1.0211683511734009,
      "learning_rate": 4.951024042742654e-06,
      "loss": 0.7661,
      "step": 15180
    },
    {
      "epoch": 4.508756307509647,
      "grad_norm": 0.9978165626525879,
      "learning_rate": 4.921341644404868e-06,
      "loss": 0.7935,
      "step": 15190
    },
    {
      "epoch": 4.511724547343425,
      "grad_norm": 1.0135602951049805,
      "learning_rate": 4.891659246067082e-06,
      "loss": 0.7471,
      "step": 15200
    },
    {
      "epoch": 4.514692787177204,
      "grad_norm": 0.928688108921051,
      "learning_rate": 4.861976847729297e-06,
      "loss": 0.7576,
      "step": 15210
    },
    {
      "epoch": 4.517661027010982,
      "grad_norm": 0.9365946054458618,
      "learning_rate": 4.832294449391511e-06,
      "loss": 0.7482,
      "step": 15220
    },
    {
      "epoch": 4.520629266844761,
      "grad_norm": 0.9219352602958679,
      "learning_rate": 4.802612051053725e-06,
      "loss": 0.7675,
      "step": 15230
    },
    {
      "epoch": 4.52359750667854,
      "grad_norm": 1.0753960609436035,
      "learning_rate": 4.77292965271594e-06,
      "loss": 0.7668,
      "step": 15240
    },
    {
      "epoch": 4.526565746512318,
      "grad_norm": 0.9870726466178894,
      "learning_rate": 4.743247254378154e-06,
      "loss": 0.7696,
      "step": 15250
    },
    {
      "epoch": 4.529533986346097,
      "grad_norm": 0.9948970675468445,
      "learning_rate": 4.713564856040368e-06,
      "loss": 0.7636,
      "step": 15260
    },
    {
      "epoch": 4.532502226179876,
      "grad_norm": 0.9187266230583191,
      "learning_rate": 4.683882457702583e-06,
      "loss": 0.7398,
      "step": 15270
    },
    {
      "epoch": 4.535470466013654,
      "grad_norm": 0.8769642114639282,
      "learning_rate": 4.6542000593647965e-06,
      "loss": 0.7604,
      "step": 15280
    },
    {
      "epoch": 4.538438705847432,
      "grad_norm": 0.8811549544334412,
      "learning_rate": 4.624517661027011e-06,
      "loss": 0.7802,
      "step": 15290
    },
    {
      "epoch": 4.541406945681211,
      "grad_norm": 1.0554735660552979,
      "learning_rate": 4.5948352626892255e-06,
      "loss": 0.7527,
      "step": 15300
    },
    {
      "epoch": 4.54437518551499,
      "grad_norm": 1.1264728307724,
      "learning_rate": 4.56515286435144e-06,
      "loss": 0.7845,
      "step": 15310
    },
    {
      "epoch": 4.547343425348768,
      "grad_norm": 1.0017541646957397,
      "learning_rate": 4.5354704660136544e-06,
      "loss": 0.7548,
      "step": 15320
    },
    {
      "epoch": 4.550311665182547,
      "grad_norm": 0.9614369869232178,
      "learning_rate": 4.505788067675869e-06,
      "loss": 0.7319,
      "step": 15330
    },
    {
      "epoch": 4.553279905016325,
      "grad_norm": 1.0290606021881104,
      "learning_rate": 4.4761056693380825e-06,
      "loss": 0.7621,
      "step": 15340
    },
    {
      "epoch": 4.5562481448501035,
      "grad_norm": 1.0442272424697876,
      "learning_rate": 4.446423271000297e-06,
      "loss": 0.7277,
      "step": 15350
    },
    {
      "epoch": 4.559216384683882,
      "grad_norm": 1.0785348415374756,
      "learning_rate": 4.4167408726625115e-06,
      "loss": 0.7557,
      "step": 15360
    },
    {
      "epoch": 4.562184624517661,
      "grad_norm": 0.9761595129966736,
      "learning_rate": 4.387058474324725e-06,
      "loss": 0.7443,
      "step": 15370
    },
    {
      "epoch": 4.56515286435144,
      "grad_norm": 1.0198564529418945,
      "learning_rate": 4.3573760759869405e-06,
      "loss": 0.7454,
      "step": 15380
    },
    {
      "epoch": 4.568121104185218,
      "grad_norm": 0.9881998300552368,
      "learning_rate": 4.327693677649154e-06,
      "loss": 0.79,
      "step": 15390
    },
    {
      "epoch": 4.571089344018997,
      "grad_norm": 1.0645211935043335,
      "learning_rate": 4.298011279311369e-06,
      "loss": 0.759,
      "step": 15400
    },
    {
      "epoch": 4.574057583852776,
      "grad_norm": 0.9828948974609375,
      "learning_rate": 4.268328880973583e-06,
      "loss": 0.7436,
      "step": 15410
    },
    {
      "epoch": 4.5770258236865535,
      "grad_norm": 0.9696997404098511,
      "learning_rate": 4.238646482635797e-06,
      "loss": 0.7422,
      "step": 15420
    },
    {
      "epoch": 4.579994063520332,
      "grad_norm": 1.0432863235473633,
      "learning_rate": 4.208964084298011e-06,
      "loss": 0.7918,
      "step": 15430
    },
    {
      "epoch": 4.582962303354111,
      "grad_norm": 1.055452823638916,
      "learning_rate": 4.179281685960226e-06,
      "loss": 0.7642,
      "step": 15440
    },
    {
      "epoch": 4.58593054318789,
      "grad_norm": 0.9821627736091614,
      "learning_rate": 4.14959928762244e-06,
      "loss": 0.7921,
      "step": 15450
    },
    {
      "epoch": 4.588898783021668,
      "grad_norm": 1.0450557470321655,
      "learning_rate": 4.119916889284655e-06,
      "loss": 0.7618,
      "step": 15460
    },
    {
      "epoch": 4.591867022855447,
      "grad_norm": 1.0069783926010132,
      "learning_rate": 4.090234490946869e-06,
      "loss": 0.7691,
      "step": 15470
    },
    {
      "epoch": 4.594835262689225,
      "grad_norm": 1.1639186143875122,
      "learning_rate": 4.060552092609083e-06,
      "loss": 0.7418,
      "step": 15480
    },
    {
      "epoch": 4.5978035025230035,
      "grad_norm": 0.9840888381004333,
      "learning_rate": 4.030869694271297e-06,
      "loss": 0.7594,
      "step": 15490
    },
    {
      "epoch": 4.600771742356782,
      "grad_norm": 1.1109092235565186,
      "learning_rate": 4.001187295933512e-06,
      "loss": 0.7433,
      "step": 15500
    },
    {
      "epoch": 4.603739982190561,
      "grad_norm": 0.9764207601547241,
      "learning_rate": 3.971504897595725e-06,
      "loss": 0.7388,
      "step": 15510
    },
    {
      "epoch": 4.60670822202434,
      "grad_norm": 1.190716028213501,
      "learning_rate": 3.941822499257941e-06,
      "loss": 0.7682,
      "step": 15520
    },
    {
      "epoch": 4.609676461858118,
      "grad_norm": 1.028148889541626,
      "learning_rate": 3.912140100920154e-06,
      "loss": 0.7526,
      "step": 15530
    },
    {
      "epoch": 4.612644701691897,
      "grad_norm": 1.0454537868499756,
      "learning_rate": 3.882457702582369e-06,
      "loss": 0.7571,
      "step": 15540
    },
    {
      "epoch": 4.615612941525676,
      "grad_norm": 0.9647070169448853,
      "learning_rate": 3.852775304244583e-06,
      "loss": 0.7305,
      "step": 15550
    },
    {
      "epoch": 4.6185811813594535,
      "grad_norm": 1.0753306150436401,
      "learning_rate": 3.823092905906797e-06,
      "loss": 0.7503,
      "step": 15560
    },
    {
      "epoch": 4.621549421193232,
      "grad_norm": 0.9869060516357422,
      "learning_rate": 3.793410507569012e-06,
      "loss": 0.7693,
      "step": 15570
    },
    {
      "epoch": 4.624517661027011,
      "grad_norm": 0.9355076551437378,
      "learning_rate": 3.763728109231226e-06,
      "loss": 0.7838,
      "step": 15580
    },
    {
      "epoch": 4.62748590086079,
      "grad_norm": 1.0376447439193726,
      "learning_rate": 3.7340457108934404e-06,
      "loss": 0.7731,
      "step": 15590
    },
    {
      "epoch": 4.630454140694568,
      "grad_norm": 0.9946711659431458,
      "learning_rate": 3.7043633125556544e-06,
      "loss": 0.7275,
      "step": 15600
    },
    {
      "epoch": 4.633422380528347,
      "grad_norm": 1.0657856464385986,
      "learning_rate": 3.6746809142178693e-06,
      "loss": 0.7254,
      "step": 15610
    },
    {
      "epoch": 4.636390620362125,
      "grad_norm": 1.0746755599975586,
      "learning_rate": 3.6449985158800834e-06,
      "loss": 0.7705,
      "step": 15620
    },
    {
      "epoch": 4.6393588601959035,
      "grad_norm": 1.1331027746200562,
      "learning_rate": 3.6182843573760763e-06,
      "loss": 0.7782,
      "step": 15630
    },
    {
      "epoch": 4.642327100029682,
      "grad_norm": 0.9259887337684631,
      "learning_rate": 3.5886019590382904e-06,
      "loss": 0.7686,
      "step": 15640
    },
    {
      "epoch": 4.645295339863461,
      "grad_norm": 0.9392924904823303,
      "learning_rate": 3.558919560700505e-06,
      "loss": 0.7045,
      "step": 15650
    },
    {
      "epoch": 4.64826357969724,
      "grad_norm": 0.9543144702911377,
      "learning_rate": 3.529237162362719e-06,
      "loss": 0.7865,
      "step": 15660
    },
    {
      "epoch": 4.651231819531018,
      "grad_norm": 1.050999641418457,
      "learning_rate": 3.499554764024934e-06,
      "loss": 0.7581,
      "step": 15670
    },
    {
      "epoch": 4.654200059364797,
      "grad_norm": 0.9569733738899231,
      "learning_rate": 3.4698723656871475e-06,
      "loss": 0.7632,
      "step": 15680
    },
    {
      "epoch": 4.657168299198576,
      "grad_norm": 1.073243498802185,
      "learning_rate": 3.4401899673493624e-06,
      "loss": 0.7074,
      "step": 15690
    },
    {
      "epoch": 4.6601365390323535,
      "grad_norm": 1.131837010383606,
      "learning_rate": 3.4105075690115764e-06,
      "loss": 0.7503,
      "step": 15700
    },
    {
      "epoch": 4.663104778866132,
      "grad_norm": 1.0623327493667603,
      "learning_rate": 3.3808251706737905e-06,
      "loss": 0.7705,
      "step": 15710
    },
    {
      "epoch": 4.666073018699911,
      "grad_norm": 1.076809048652649,
      "learning_rate": 3.351142772336005e-06,
      "loss": 0.7448,
      "step": 15720
    },
    {
      "epoch": 4.66904125853369,
      "grad_norm": 0.9963298439979553,
      "learning_rate": 3.321460373998219e-06,
      "loss": 0.7125,
      "step": 15730
    },
    {
      "epoch": 4.672009498367468,
      "grad_norm": 1.0451970100402832,
      "learning_rate": 3.291777975660434e-06,
      "loss": 0.79,
      "step": 15740
    },
    {
      "epoch": 4.674977738201247,
      "grad_norm": 0.9352405071258545,
      "learning_rate": 3.2620955773226476e-06,
      "loss": 0.7646,
      "step": 15750
    },
    {
      "epoch": 4.677945978035025,
      "grad_norm": 1.0493252277374268,
      "learning_rate": 3.2324131789848625e-06,
      "loss": 0.7355,
      "step": 15760
    },
    {
      "epoch": 4.6809142178688035,
      "grad_norm": 0.9438860416412354,
      "learning_rate": 3.2027307806470765e-06,
      "loss": 0.7492,
      "step": 15770
    },
    {
      "epoch": 4.683882457702582,
      "grad_norm": 0.8936295509338379,
      "learning_rate": 3.1730483823092906e-06,
      "loss": 0.745,
      "step": 15780
    },
    {
      "epoch": 4.686850697536361,
      "grad_norm": 1.0204068422317505,
      "learning_rate": 3.143365983971505e-06,
      "loss": 0.7576,
      "step": 15790
    },
    {
      "epoch": 4.6898189373701396,
      "grad_norm": 0.9555333256721497,
      "learning_rate": 3.1136835856337196e-06,
      "loss": 0.7595,
      "step": 15800
    },
    {
      "epoch": 4.692787177203918,
      "grad_norm": 0.9514308571815491,
      "learning_rate": 3.0840011872959336e-06,
      "loss": 0.7565,
      "step": 15810
    },
    {
      "epoch": 4.695755417037697,
      "grad_norm": 1.0169340372085571,
      "learning_rate": 3.0543187889581477e-06,
      "loss": 0.7104,
      "step": 15820
    },
    {
      "epoch": 4.698723656871476,
      "grad_norm": 0.9857751131057739,
      "learning_rate": 3.024636390620362e-06,
      "loss": 0.7475,
      "step": 15830
    },
    {
      "epoch": 4.7016918967052534,
      "grad_norm": 0.9653582572937012,
      "learning_rate": 2.9949539922825766e-06,
      "loss": 0.76,
      "step": 15840
    },
    {
      "epoch": 4.704660136539032,
      "grad_norm": 1.0300312042236328,
      "learning_rate": 2.965271593944791e-06,
      "loss": 0.7327,
      "step": 15850
    },
    {
      "epoch": 4.707628376372811,
      "grad_norm": 1.1477298736572266,
      "learning_rate": 2.935589195607005e-06,
      "loss": 0.773,
      "step": 15860
    },
    {
      "epoch": 4.7105966162065895,
      "grad_norm": 0.9286091327667236,
      "learning_rate": 2.9059067972692197e-06,
      "loss": 0.7152,
      "step": 15870
    },
    {
      "epoch": 4.713564856040368,
      "grad_norm": 1.0865973234176636,
      "learning_rate": 2.8762243989314337e-06,
      "loss": 0.7692,
      "step": 15880
    },
    {
      "epoch": 4.716533095874147,
      "grad_norm": 1.0828102827072144,
      "learning_rate": 2.846542000593648e-06,
      "loss": 0.7591,
      "step": 15890
    },
    {
      "epoch": 4.719501335707925,
      "grad_norm": 0.9569112062454224,
      "learning_rate": 2.8168596022558623e-06,
      "loss": 0.7642,
      "step": 15900
    },
    {
      "epoch": 4.722469575541703,
      "grad_norm": 0.9393323659896851,
      "learning_rate": 2.7871772039180767e-06,
      "loss": 0.7994,
      "step": 15910
    },
    {
      "epoch": 4.725437815375482,
      "grad_norm": 0.9811632633209229,
      "learning_rate": 2.7574948055802912e-06,
      "loss": 0.7991,
      "step": 15920
    },
    {
      "epoch": 4.728406055209261,
      "grad_norm": 1.021985411643982,
      "learning_rate": 2.7278124072425053e-06,
      "loss": 0.7422,
      "step": 15930
    },
    {
      "epoch": 4.7313742950430395,
      "grad_norm": 1.0036324262619019,
      "learning_rate": 2.6981300089047198e-06,
      "loss": 0.726,
      "step": 15940
    },
    {
      "epoch": 4.734342534876818,
      "grad_norm": 1.1031821966171265,
      "learning_rate": 2.668447610566934e-06,
      "loss": 0.752,
      "step": 15950
    },
    {
      "epoch": 4.737310774710597,
      "grad_norm": 1.0434112548828125,
      "learning_rate": 2.6387652122291483e-06,
      "loss": 0.7227,
      "step": 15960
    },
    {
      "epoch": 4.740279014544376,
      "grad_norm": 1.106284737586975,
      "learning_rate": 2.6090828138913624e-06,
      "loss": 0.7613,
      "step": 15970
    },
    {
      "epoch": 4.743247254378153,
      "grad_norm": 1.0312436819076538,
      "learning_rate": 2.579400415553577e-06,
      "loss": 0.7511,
      "step": 15980
    },
    {
      "epoch": 4.746215494211932,
      "grad_norm": 1.014859914779663,
      "learning_rate": 2.5497180172157913e-06,
      "loss": 0.8043,
      "step": 15990
    },
    {
      "epoch": 4.749183734045711,
      "grad_norm": 0.9922273755073547,
      "learning_rate": 2.5200356188780054e-06,
      "loss": 0.7598,
      "step": 16000
    },
    {
      "epoch": 4.7521519738794895,
      "grad_norm": 0.9891121983528137,
      "learning_rate": 2.49035322054022e-06,
      "loss": 0.762,
      "step": 16010
    },
    {
      "epoch": 4.755120213713268,
      "grad_norm": 1.12299382686615,
      "learning_rate": 2.460670822202434e-06,
      "loss": 0.8067,
      "step": 16020
    },
    {
      "epoch": 4.758088453547047,
      "grad_norm": 1.0930094718933105,
      "learning_rate": 2.4309884238646484e-06,
      "loss": 0.7606,
      "step": 16030
    },
    {
      "epoch": 4.761056693380825,
      "grad_norm": 1.0453695058822632,
      "learning_rate": 2.4013060255268625e-06,
      "loss": 0.7362,
      "step": 16040
    },
    {
      "epoch": 4.764024933214603,
      "grad_norm": 1.031219720840454,
      "learning_rate": 2.371623627189077e-06,
      "loss": 0.7693,
      "step": 16050
    },
    {
      "epoch": 4.766993173048382,
      "grad_norm": 1.1497999429702759,
      "learning_rate": 2.3419412288512914e-06,
      "loss": 0.7343,
      "step": 16060
    },
    {
      "epoch": 4.769961412882161,
      "grad_norm": 0.9976547360420227,
      "learning_rate": 2.3122588305135055e-06,
      "loss": 0.7893,
      "step": 16070
    },
    {
      "epoch": 4.7729296527159395,
      "grad_norm": 0.9629161357879639,
      "learning_rate": 2.28257643217572e-06,
      "loss": 0.7583,
      "step": 16080
    },
    {
      "epoch": 4.775897892549718,
      "grad_norm": 1.0763802528381348,
      "learning_rate": 2.2528940338379345e-06,
      "loss": 0.7597,
      "step": 16090
    },
    {
      "epoch": 4.778866132383497,
      "grad_norm": 1.108970046043396,
      "learning_rate": 2.2232116355001485e-06,
      "loss": 0.7516,
      "step": 16100
    },
    {
      "epoch": 4.781834372217276,
      "grad_norm": 0.9933566451072693,
      "learning_rate": 2.1935292371623626e-06,
      "loss": 0.7305,
      "step": 16110
    },
    {
      "epoch": 4.784802612051053,
      "grad_norm": 1.0121699571609497,
      "learning_rate": 2.163846838824577e-06,
      "loss": 0.7883,
      "step": 16120
    },
    {
      "epoch": 4.787770851884832,
      "grad_norm": 1.1284277439117432,
      "learning_rate": 2.1341644404867915e-06,
      "loss": 0.7378,
      "step": 16130
    },
    {
      "epoch": 4.790739091718611,
      "grad_norm": 1.1266158819198608,
      "learning_rate": 2.1044820421490056e-06,
      "loss": 0.7737,
      "step": 16140
    },
    {
      "epoch": 4.7937073315523895,
      "grad_norm": 0.989107072353363,
      "learning_rate": 2.07479964381122e-06,
      "loss": 0.7542,
      "step": 16150
    },
    {
      "epoch": 4.796675571386168,
      "grad_norm": 1.0217342376708984,
      "learning_rate": 2.0451172454734346e-06,
      "loss": 0.7697,
      "step": 16160
    },
    {
      "epoch": 4.799643811219947,
      "grad_norm": 0.9917885661125183,
      "learning_rate": 2.0154348471356486e-06,
      "loss": 0.7467,
      "step": 16170
    },
    {
      "epoch": 4.802612051053725,
      "grad_norm": 0.9344885349273682,
      "learning_rate": 1.9857524487978627e-06,
      "loss": 0.7926,
      "step": 16180
    },
    {
      "epoch": 4.805580290887503,
      "grad_norm": 1.0406873226165771,
      "learning_rate": 1.956070050460077e-06,
      "loss": 0.7564,
      "step": 16190
    },
    {
      "epoch": 4.808548530721282,
      "grad_norm": 0.9802284836769104,
      "learning_rate": 1.9263876521222916e-06,
      "loss": 0.7383,
      "step": 16200
    },
    {
      "epoch": 4.811516770555061,
      "grad_norm": 1.1220468282699585,
      "learning_rate": 1.896705253784506e-06,
      "loss": 0.7338,
      "step": 16210
    },
    {
      "epoch": 4.8144850103888395,
      "grad_norm": 1.0869120359420776,
      "learning_rate": 1.8670228554467202e-06,
      "loss": 0.7635,
      "step": 16220
    },
    {
      "epoch": 4.817453250222618,
      "grad_norm": 0.9541686177253723,
      "learning_rate": 1.8373404571089347e-06,
      "loss": 0.7626,
      "step": 16230
    },
    {
      "epoch": 4.820421490056397,
      "grad_norm": 1.0052292346954346,
      "learning_rate": 1.8076580587711487e-06,
      "loss": 0.7526,
      "step": 16240
    },
    {
      "epoch": 4.823389729890176,
      "grad_norm": 1.0642937421798706,
      "learning_rate": 1.777975660433363e-06,
      "loss": 0.731,
      "step": 16250
    },
    {
      "epoch": 4.826357969723953,
      "grad_norm": 1.0232363939285278,
      "learning_rate": 1.7482932620955773e-06,
      "loss": 0.8001,
      "step": 16260
    },
    {
      "epoch": 4.829326209557732,
      "grad_norm": 0.9778085947036743,
      "learning_rate": 1.7186108637577917e-06,
      "loss": 0.7323,
      "step": 16270
    },
    {
      "epoch": 4.832294449391511,
      "grad_norm": 1.0876951217651367,
      "learning_rate": 1.688928465420006e-06,
      "loss": 0.7394,
      "step": 16280
    },
    {
      "epoch": 4.8352626892252895,
      "grad_norm": 1.0854240655899048,
      "learning_rate": 1.6592460670822205e-06,
      "loss": 0.7493,
      "step": 16290
    },
    {
      "epoch": 4.838230929059068,
      "grad_norm": 1.0767631530761719,
      "learning_rate": 1.6295636687444348e-06,
      "loss": 0.7355,
      "step": 16300
    },
    {
      "epoch": 4.841199168892847,
      "grad_norm": 0.9216200113296509,
      "learning_rate": 1.599881270406649e-06,
      "loss": 0.7816,
      "step": 16310
    },
    {
      "epoch": 4.844167408726625,
      "grad_norm": 0.9959633350372314,
      "learning_rate": 1.570198872068863e-06,
      "loss": 0.7825,
      "step": 16320
    },
    {
      "epoch": 4.847135648560403,
      "grad_norm": 1.0585358142852783,
      "learning_rate": 1.5405164737310774e-06,
      "loss": 0.7299,
      "step": 16330
    },
    {
      "epoch": 4.850103888394182,
      "grad_norm": 0.9767472743988037,
      "learning_rate": 1.5108340753932918e-06,
      "loss": 0.7335,
      "step": 16340
    },
    {
      "epoch": 4.853072128227961,
      "grad_norm": 0.9296145439147949,
      "learning_rate": 1.4811516770555061e-06,
      "loss": 0.7488,
      "step": 16350
    },
    {
      "epoch": 4.856040368061739,
      "grad_norm": 0.9834486842155457,
      "learning_rate": 1.4514692787177206e-06,
      "loss": 0.7393,
      "step": 16360
    },
    {
      "epoch": 4.859008607895518,
      "grad_norm": 1.0462241172790527,
      "learning_rate": 1.4217868803799347e-06,
      "loss": 0.7789,
      "step": 16370
    },
    {
      "epoch": 4.861976847729297,
      "grad_norm": 1.0443044900894165,
      "learning_rate": 1.3921044820421491e-06,
      "loss": 0.7459,
      "step": 16380
    },
    {
      "epoch": 4.8649450875630755,
      "grad_norm": 1.0716089010238647,
      "learning_rate": 1.3624220837043634e-06,
      "loss": 0.7614,
      "step": 16390
    },
    {
      "epoch": 4.867913327396853,
      "grad_norm": 0.9359921216964722,
      "learning_rate": 1.3327396853665777e-06,
      "loss": 0.751,
      "step": 16400
    },
    {
      "epoch": 4.870881567230632,
      "grad_norm": 0.9718852043151855,
      "learning_rate": 1.303057287028792e-06,
      "loss": 0.7661,
      "step": 16410
    },
    {
      "epoch": 4.873849807064411,
      "grad_norm": 1.0819077491760254,
      "learning_rate": 1.2733748886910062e-06,
      "loss": 0.7634,
      "step": 16420
    },
    {
      "epoch": 4.876818046898189,
      "grad_norm": 1.0779821872711182,
      "learning_rate": 1.2436924903532207e-06,
      "loss": 0.7171,
      "step": 16430
    },
    {
      "epoch": 4.879786286731968,
      "grad_norm": 1.0170753002166748,
      "learning_rate": 1.2140100920154348e-06,
      "loss": 0.7971,
      "step": 16440
    },
    {
      "epoch": 4.882754526565747,
      "grad_norm": 0.9914931058883667,
      "learning_rate": 1.1843276936776492e-06,
      "loss": 0.7852,
      "step": 16450
    },
    {
      "epoch": 4.885722766399525,
      "grad_norm": 1.0433253049850464,
      "learning_rate": 1.1546452953398635e-06,
      "loss": 0.7115,
      "step": 16460
    },
    {
      "epoch": 4.888691006233303,
      "grad_norm": 1.0355966091156006,
      "learning_rate": 1.124962897002078e-06,
      "loss": 0.7221,
      "step": 16470
    },
    {
      "epoch": 4.891659246067082,
      "grad_norm": 1.1262221336364746,
      "learning_rate": 1.095280498664292e-06,
      "loss": 0.7751,
      "step": 16480
    },
    {
      "epoch": 4.894627485900861,
      "grad_norm": 1.0155718326568604,
      "learning_rate": 1.0655981003265063e-06,
      "loss": 0.7433,
      "step": 16490
    },
    {
      "epoch": 4.897595725734639,
      "grad_norm": 1.0135166645050049,
      "learning_rate": 1.0359157019887208e-06,
      "loss": 0.7146,
      "step": 16500
    },
    {
      "epoch": 4.900563965568418,
      "grad_norm": 1.0288039445877075,
      "learning_rate": 1.006233303650935e-06,
      "loss": 0.7492,
      "step": 16510
    },
    {
      "epoch": 4.903532205402197,
      "grad_norm": 1.0542793273925781,
      "learning_rate": 9.765509053131493e-07,
      "loss": 0.7519,
      "step": 16520
    },
    {
      "epoch": 4.9065004452359755,
      "grad_norm": 1.0649762153625488,
      "learning_rate": 9.468685069753636e-07,
      "loss": 0.736,
      "step": 16530
    },
    {
      "epoch": 4.909468685069753,
      "grad_norm": 0.9523289203643799,
      "learning_rate": 9.17186108637578e-07,
      "loss": 0.7419,
      "step": 16540
    },
    {
      "epoch": 4.912436924903532,
      "grad_norm": 1.090806007385254,
      "learning_rate": 8.875037102997922e-07,
      "loss": 0.759,
      "step": 16550
    },
    {
      "epoch": 4.915405164737311,
      "grad_norm": 1.0011529922485352,
      "learning_rate": 8.578213119620065e-07,
      "loss": 0.7418,
      "step": 16560
    },
    {
      "epoch": 4.918373404571089,
      "grad_norm": 1.0001546144485474,
      "learning_rate": 8.281389136242209e-07,
      "loss": 0.756,
      "step": 16570
    },
    {
      "epoch": 4.921341644404868,
      "grad_norm": 0.924248218536377,
      "learning_rate": 7.984565152864353e-07,
      "loss": 0.737,
      "step": 16580
    },
    {
      "epoch": 4.924309884238647,
      "grad_norm": 0.961621105670929,
      "learning_rate": 7.687741169486496e-07,
      "loss": 0.7688,
      "step": 16590
    },
    {
      "epoch": 4.927278124072425,
      "grad_norm": 0.9049778580665588,
      "learning_rate": 7.390917186108638e-07,
      "loss": 0.7132,
      "step": 16600
    },
    {
      "epoch": 4.930246363906203,
      "grad_norm": 1.124666690826416,
      "learning_rate": 7.094093202730781e-07,
      "loss": 0.7389,
      "step": 16610
    },
    {
      "epoch": 4.933214603739982,
      "grad_norm": 1.035509467124939,
      "learning_rate": 6.797269219352924e-07,
      "loss": 0.7779,
      "step": 16620
    },
    {
      "epoch": 4.936182843573761,
      "grad_norm": 1.0520305633544922,
      "learning_rate": 6.500445235975066e-07,
      "loss": 0.7665,
      "step": 16630
    },
    {
      "epoch": 4.939151083407539,
      "grad_norm": 0.9276071786880493,
      "learning_rate": 6.20362125259721e-07,
      "loss": 0.7571,
      "step": 16640
    },
    {
      "epoch": 4.942119323241318,
      "grad_norm": 1.0482933521270752,
      "learning_rate": 5.906797269219353e-07,
      "loss": 0.7781,
      "step": 16650
    },
    {
      "epoch": 4.945087563075097,
      "grad_norm": 0.9943667650222778,
      "learning_rate": 5.609973285841497e-07,
      "loss": 0.7629,
      "step": 16660
    },
    {
      "epoch": 4.9480558029088755,
      "grad_norm": 1.0370135307312012,
      "learning_rate": 5.313149302463639e-07,
      "loss": 0.7767,
      "step": 16670
    },
    {
      "epoch": 4.951024042742653,
      "grad_norm": 0.9482097625732422,
      "learning_rate": 5.016325319085783e-07,
      "loss": 0.7694,
      "step": 16680
    },
    {
      "epoch": 4.953992282576432,
      "grad_norm": 0.9172430634498596,
      "learning_rate": 4.719501335707925e-07,
      "loss": 0.7461,
      "step": 16690
    },
    {
      "epoch": 4.956960522410211,
      "grad_norm": 1.109588623046875,
      "learning_rate": 4.422677352330069e-07,
      "loss": 0.7463,
      "step": 16700
    },
    {
      "epoch": 4.959928762243989,
      "grad_norm": 0.920391857624054,
      "learning_rate": 4.1258533689522117e-07,
      "loss": 0.7494,
      "step": 16710
    },
    {
      "epoch": 4.962897002077768,
      "grad_norm": 1.0499415397644043,
      "learning_rate": 3.8290293855743544e-07,
      "loss": 0.7568,
      "step": 16720
    },
    {
      "epoch": 4.965865241911547,
      "grad_norm": 0.9439771771430969,
      "learning_rate": 3.5322054021964976e-07,
      "loss": 0.7348,
      "step": 16730
    },
    {
      "epoch": 4.968833481745325,
      "grad_norm": 1.0074434280395508,
      "learning_rate": 3.235381418818641e-07,
      "loss": 0.7173,
      "step": 16740
    },
    {
      "epoch": 4.971801721579103,
      "grad_norm": 0.9283884763717651,
      "learning_rate": 2.938557435440784e-07,
      "loss": 0.7524,
      "step": 16750
    },
    {
      "epoch": 4.974769961412882,
      "grad_norm": 1.03079354763031,
      "learning_rate": 2.641733452062927e-07,
      "loss": 0.7605,
      "step": 16760
    },
    {
      "epoch": 4.977738201246661,
      "grad_norm": 1.0735660791397095,
      "learning_rate": 2.3449094686850697e-07,
      "loss": 0.7535,
      "step": 16770
    },
    {
      "epoch": 4.980706441080439,
      "grad_norm": 0.9746400117874146,
      "learning_rate": 2.0480854853072127e-07,
      "loss": 0.7934,
      "step": 16780
    },
    {
      "epoch": 4.983674680914218,
      "grad_norm": 0.9652214646339417,
      "learning_rate": 1.751261501929356e-07,
      "loss": 0.7396,
      "step": 16790
    },
    {
      "epoch": 4.986642920747997,
      "grad_norm": 0.9589136242866516,
      "learning_rate": 1.454437518551499e-07,
      "loss": 0.7564,
      "step": 16800
    },
    {
      "epoch": 4.9896111605817755,
      "grad_norm": 1.137319803237915,
      "learning_rate": 1.157613535173642e-07,
      "loss": 0.742,
      "step": 16810
    },
    {
      "epoch": 4.992579400415553,
      "grad_norm": 1.0568934679031372,
      "learning_rate": 8.607895517957851e-08,
      "loss": 0.7621,
      "step": 16820
    },
    {
      "epoch": 4.995547640249332,
      "grad_norm": 0.9926442503929138,
      "learning_rate": 5.639655684179283e-08,
      "loss": 0.7375,
      "step": 16830
    },
    {
      "epoch": 4.998515880083111,
      "grad_norm": 0.9872409105300903,
      "learning_rate": 2.6714158504007124e-08,
      "loss": 0.74,
      "step": 16840
    }
  ],
  "logging_steps": 10,
  "max_steps": 16845,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.960972002956411e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}