{
  "best_metric": 1.3688740730285645,
  "best_model_checkpoint": "saved_model/c2s_dec2024/checkpoint-11928",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 23857,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": NaN,
      "learning_rate": 0.0,
      "loss": 74.8478,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.946953773498535,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 73.9489,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.25737762451172,
      "learning_rate": 7.000000000000001e-06,
      "loss": 73.6478,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.36816692352295,
      "learning_rate": 1.2e-05,
      "loss": 72.7254,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.802995681762695,
      "learning_rate": 1.65e-05,
      "loss": 72.1129,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.857714653015137,
      "learning_rate": 2.15e-05,
      "loss": 67.5212,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.70682716369629,
      "learning_rate": 2.6500000000000004e-05,
      "loss": 63.3518,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 27.735233306884766,
      "learning_rate": 3.15e-05,
      "loss": 52.952,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 25.16404151916504,
      "learning_rate": 3.65e-05,
      "loss": 31.7544,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.33328914642334,
      "learning_rate": 4.15e-05,
      "loss": 13.379,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.520598888397217,
      "learning_rate": 4.6500000000000005e-05,
      "loss": 7.3857,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.755812168121338,
      "learning_rate": 5.1500000000000005e-05,
      "loss": 5.2951,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.882263660430908,
      "learning_rate": 5.65e-05,
      "loss": 4.2907,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.14441442489624,
      "learning_rate": 6.15e-05,
      "loss": 3.9018,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.848437309265137,
      "learning_rate": 6.65e-05,
      "loss": 3.7094,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.878305435180664,
      "learning_rate": 7.15e-05,
      "loss": 3.5207,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.875974655151367,
      "learning_rate": 7.65e-05,
      "loss": 3.5124,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.956191062927246,
      "learning_rate": 8.15e-05,
      "loss": 3.3862,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.560382843017578,
      "learning_rate": 8.65e-05,
      "loss": 3.3764,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.907286167144775,
      "learning_rate": 9.15e-05,
      "loss": 3.2589,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.328891754150391,
      "learning_rate": 9.65e-05,
      "loss": 3.2308,
      "step": 200
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.169933319091797,
      "learning_rate": 9.999748068525362e-05,
      "loss": 3.2297,
      "step": 210
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.502355575561523,
      "learning_rate": 9.998908296943233e-05,
      "loss": 3.1368,
      "step": 220
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.5027313232421875,
      "learning_rate": 9.998068525361102e-05,
      "loss": 3.0277,
      "step": 230
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.679823875427246,
      "learning_rate": 9.997228753778972e-05,
      "loss": 2.9859,
      "step": 240
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.180417537689209,
      "learning_rate": 9.996388982196843e-05,
      "loss": 2.9394,
      "step": 250
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.605576992034912,
      "learning_rate": 9.995549210614713e-05,
      "loss": 2.9062,
      "step": 260
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.151813983917236,
      "learning_rate": 9.994709439032583e-05,
      "loss": 2.9341,
      "step": 270
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.148714065551758,
      "learning_rate": 9.993869667450453e-05,
      "loss": 2.9287,
      "step": 280
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.156353950500488,
      "learning_rate": 9.993029895868325e-05,
      "loss": 2.8849,
      "step": 290
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.806578159332275,
      "learning_rate": 9.992190124286195e-05,
      "loss": 2.9143,
      "step": 300
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.209172248840332,
      "learning_rate": 9.991350352704065e-05,
      "loss": 2.819,
      "step": 310
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.573767185211182,
      "learning_rate": 9.990510581121935e-05,
      "loss": 2.8762,
      "step": 320
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.73787784576416,
      "learning_rate": 9.989670809539807e-05,
      "loss": 2.8593,
      "step": 330
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.8452227115631104,
      "learning_rate": 9.988831037957675e-05,
      "loss": 2.7755,
      "step": 340
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.609687328338623,
      "learning_rate": 9.987991266375545e-05,
      "loss": 2.8353,
      "step": 350
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.1777472496032715,
      "learning_rate": 9.987151494793417e-05,
      "loss": 2.7933,
      "step": 360
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.308431148529053,
      "learning_rate": 9.986311723211287e-05,
      "loss": 2.7714,
      "step": 370
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.116801738739014,
      "learning_rate": 9.985471951629157e-05,
      "loss": 2.8051,
      "step": 380
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.194584369659424,
      "learning_rate": 9.984632180047028e-05,
      "loss": 2.7259,
      "step": 390
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.333553314208984,
      "learning_rate": 9.983792408464898e-05,
      "loss": 2.7394,
      "step": 400
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.2470808029174805,
      "learning_rate": 9.982952636882768e-05,
      "loss": 2.7138,
      "step": 410
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.5597198009490967,
      "learning_rate": 9.982112865300639e-05,
      "loss": 2.6112,
      "step": 420
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.565213918685913,
      "learning_rate": 9.98127309371851e-05,
      "loss": 2.5829,
      "step": 430
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.250683069229126,
      "learning_rate": 9.98043332213638e-05,
      "loss": 2.5245,
      "step": 440
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.06723690032959,
      "learning_rate": 9.979593550554249e-05,
      "loss": 2.4863,
      "step": 450
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.058586120605469,
      "learning_rate": 9.97875377897212e-05,
      "loss": 2.4628,
      "step": 460
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.2529423236846924,
      "learning_rate": 9.97791400738999e-05,
      "loss": 2.4223,
      "step": 470
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.3110713958740234,
      "learning_rate": 9.97707423580786e-05,
      "loss": 2.2794,
      "step": 480
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3825504779815674,
      "learning_rate": 9.976234464225732e-05,
      "loss": 2.2906,
      "step": 490
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7417995929718018,
      "learning_rate": 9.975394692643602e-05,
      "loss": 2.2686,
      "step": 500
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.9438352584838867,
      "learning_rate": 9.974554921061472e-05,
      "loss": 2.1916,
      "step": 510
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3046646118164062,
      "learning_rate": 9.973715149479342e-05,
      "loss": 2.2106,
      "step": 520
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.30639386177063,
      "learning_rate": 9.972875377897213e-05,
      "loss": 2.1224,
      "step": 530
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3796310424804688,
      "learning_rate": 9.972035606315083e-05,
      "loss": 2.0459,
      "step": 540
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4527125358581543,
      "learning_rate": 9.971195834732952e-05,
      "loss": 2.0884,
      "step": 550
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.423755645751953,
      "learning_rate": 9.970356063150824e-05,
      "loss": 2.0769,
      "step": 560
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.461524248123169,
      "learning_rate": 9.969516291568694e-05,
      "loss": 2.0332,
      "step": 570
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.436368942260742,
      "learning_rate": 9.968676519986564e-05,
      "loss": 2.0041,
      "step": 580
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2243542671203613,
      "learning_rate": 9.967836748404434e-05,
      "loss": 1.9403,
      "step": 590
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9886754751205444,
      "learning_rate": 9.966996976822305e-05,
      "loss": 1.886,
      "step": 600
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1096863746643066,
      "learning_rate": 9.966157205240175e-05,
      "loss": 1.9461,
      "step": 610
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2656311988830566,
      "learning_rate": 9.965317433658045e-05,
      "loss": 2.0326,
      "step": 620
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.069763660430908,
      "learning_rate": 9.964477662075917e-05,
      "loss": 1.9488,
      "step": 630
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.215855836868286,
      "learning_rate": 9.963637890493787e-05,
      "loss": 1.9374,
      "step": 640
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1772992610931396,
      "learning_rate": 9.962798118911657e-05,
      "loss": 1.9783,
      "step": 650
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4245526790618896,
      "learning_rate": 9.961958347329527e-05,
      "loss": 1.8929,
      "step": 660
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.761659026145935,
      "learning_rate": 9.961118575747397e-05,
      "loss": 1.9231,
      "step": 670
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8516594171524048,
      "learning_rate": 9.960278804165267e-05,
      "loss": 1.9017,
      "step": 680
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9810080528259277,
      "learning_rate": 9.959439032583137e-05,
      "loss": 1.9036,
      "step": 690
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9358642101287842,
      "learning_rate": 9.958599261001009e-05,
      "loss": 1.8785,
      "step": 700
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.024515390396118,
      "learning_rate": 9.957759489418879e-05,
      "loss": 1.7852,
      "step": 710
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8687939643859863,
      "learning_rate": 9.956919717836749e-05,
      "loss": 1.8843,
      "step": 720
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6494457721710205,
      "learning_rate": 9.95607994625462e-05,
      "loss": 1.879,
      "step": 730
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7731373310089111,
      "learning_rate": 9.95524017467249e-05,
      "loss": 1.7703,
      "step": 740
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8192178010940552,
      "learning_rate": 9.95440040309036e-05,
      "loss": 1.8501,
      "step": 750
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.732409119606018,
      "learning_rate": 9.95356063150823e-05,
      "loss": 1.8356,
      "step": 760
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.898467779159546,
      "learning_rate": 9.9527208599261e-05,
      "loss": 1.8914,
      "step": 770
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9288657903671265,
      "learning_rate": 9.95188108834397e-05,
      "loss": 1.8111,
      "step": 780
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6328600645065308,
      "learning_rate": 9.95104131676184e-05,
      "loss": 1.782,
      "step": 790
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8500330448150635,
      "learning_rate": 9.950201545179712e-05,
      "loss": 1.7623,
      "step": 800
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.769045352935791,
      "learning_rate": 9.949361773597582e-05,
      "loss": 1.7836,
      "step": 810
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2759504318237305,
      "learning_rate": 9.948522002015452e-05,
      "loss": 1.7352,
      "step": 820
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5127414464950562,
      "learning_rate": 9.947682230433322e-05,
      "loss": 1.7412,
      "step": 830
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9331601858139038,
      "learning_rate": 9.946842458851194e-05,
      "loss": 1.7243,
      "step": 840
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8960950374603271,
      "learning_rate": 9.946002687269064e-05,
      "loss": 1.6971,
      "step": 850
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9343823194503784,
      "learning_rate": 9.945162915686934e-05,
      "loss": 1.8488,
      "step": 860
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6667115688323975,
      "learning_rate": 9.944323144104804e-05,
      "loss": 1.7665,
      "step": 870
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4781936407089233,
      "learning_rate": 9.943483372522674e-05,
      "loss": 1.7995,
      "step": 880
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8188730478286743,
      "learning_rate": 9.942643600940544e-05,
      "loss": 1.7878,
      "step": 890
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3943722248077393,
      "learning_rate": 9.941803829358415e-05,
      "loss": 1.7612,
      "step": 900
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.617660641670227,
      "learning_rate": 9.940964057776285e-05,
      "loss": 1.7199,
      "step": 910
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8252322673797607,
      "learning_rate": 9.940124286194156e-05,
      "loss": 1.8101,
      "step": 920
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.598960041999817,
      "learning_rate": 9.939284514612026e-05,
      "loss": 1.7466,
      "step": 930
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8220932483673096,
      "learning_rate": 9.938444743029897e-05,
      "loss": 1.7581,
      "step": 940
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2253363132476807,
      "learning_rate": 9.937604971447767e-05,
      "loss": 1.6965,
      "step": 950
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.05928111076355,
      "learning_rate": 9.936765199865637e-05,
      "loss": 1.6986,
      "step": 960
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8350791931152344,
      "learning_rate": 9.935925428283507e-05,
      "loss": 1.7013,
      "step": 970
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6412466764450073,
      "learning_rate": 9.935085656701377e-05,
      "loss": 1.7267,
      "step": 980
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1339075565338135,
      "learning_rate": 9.934245885119247e-05,
      "loss": 1.7274,
      "step": 990
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5142629146575928,
      "learning_rate": 9.933406113537119e-05,
      "loss": 1.7106,
      "step": 1000
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6646658182144165,
      "learning_rate": 9.932566341954989e-05,
      "loss": 1.7493,
      "step": 1010
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9897111654281616,
      "learning_rate": 9.931726570372859e-05,
      "loss": 1.6104,
      "step": 1020
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6289434432983398,
      "learning_rate": 9.930886798790729e-05,
      "loss": 1.6626,
      "step": 1030
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7638473510742188,
      "learning_rate": 9.9300470272086e-05,
      "loss": 1.7381,
      "step": 1040
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5210635662078857,
      "learning_rate": 9.92920725562647e-05,
      "loss": 1.7418,
      "step": 1050
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6245310306549072,
      "learning_rate": 9.92836748404434e-05,
      "loss": 1.6449,
      "step": 1060
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5606416463851929,
      "learning_rate": 9.92752771246221e-05,
      "loss": 1.6615,
      "step": 1070
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.652024745941162,
      "learning_rate": 9.92668794088008e-05,
      "loss": 1.643,
      "step": 1080
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6315152645111084,
      "learning_rate": 9.925848169297951e-05,
      "loss": 1.7083,
      "step": 1090
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5722026824951172,
      "learning_rate": 9.925008397715821e-05,
      "loss": 1.6538,
      "step": 1100
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8919326066970825,
      "learning_rate": 9.924168626133692e-05,
      "loss": 1.6246,
      "step": 1110
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0076210498809814,
      "learning_rate": 9.923328854551562e-05,
      "loss": 1.6728,
      "step": 1120
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8884912729263306,
      "learning_rate": 9.922489082969432e-05,
      "loss": 1.7246,
      "step": 1130
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6543947458267212,
      "learning_rate": 9.921649311387304e-05,
      "loss": 1.7703,
      "step": 1140
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4063857793807983,
      "learning_rate": 9.920809539805174e-05,
      "loss": 1.7649,
      "step": 1150
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7405593395233154,
      "learning_rate": 9.919969768223044e-05,
      "loss": 1.6634,
      "step": 1160
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5649250745773315,
      "learning_rate": 9.919129996640914e-05,
      "loss": 1.622,
      "step": 1170
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.510793924331665,
      "learning_rate": 9.918290225058785e-05,
      "loss": 1.6798,
      "step": 1180
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4818421602249146,
      "learning_rate": 9.917450453476654e-05,
      "loss": 1.7286,
      "step": 1190
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5227017402648926,
      "learning_rate": 9.916610681894524e-05,
      "loss": 1.6804,
      "step": 1200
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.819848895072937,
      "learning_rate": 9.915770910312396e-05,
      "loss": 1.7091,
      "step": 1210
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6922246217727661,
      "learning_rate": 9.914931138730266e-05,
      "loss": 1.7101,
      "step": 1220
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8223869800567627,
      "learning_rate": 9.914091367148136e-05,
      "loss": 1.6555,
      "step": 1230
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.982619047164917,
      "learning_rate": 9.913251595566007e-05,
      "loss": 1.5837,
      "step": 1240
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7177784442901611,
      "learning_rate": 9.912411823983877e-05,
      "loss": 1.6073,
      "step": 1250
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5832000970840454,
      "learning_rate": 9.911572052401747e-05,
      "loss": 1.5813,
      "step": 1260
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5784751176834106,
      "learning_rate": 9.910732280819617e-05,
      "loss": 1.6822,
      "step": 1270
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7744930982589722,
      "learning_rate": 9.909892509237489e-05,
      "loss": 1.6797,
      "step": 1280
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.056403875350952,
      "learning_rate": 9.909052737655358e-05,
      "loss": 1.6241,
      "step": 1290
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5790748596191406,
      "learning_rate": 9.908212966073228e-05,
      "loss": 1.645,
      "step": 1300
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6673494577407837,
      "learning_rate": 9.907373194491099e-05,
      "loss": 1.625,
      "step": 1310
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.517059326171875,
      "learning_rate": 9.906533422908969e-05,
      "loss": 1.5792,
      "step": 1320
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6006860733032227,
      "learning_rate": 9.905693651326839e-05,
      "loss": 1.6333,
      "step": 1330
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4926496744155884,
      "learning_rate": 9.904853879744709e-05,
      "loss": 1.5963,
      "step": 1340
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5649166107177734,
      "learning_rate": 9.90401410816258e-05,
      "loss": 1.6037,
      "step": 1350
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5477185249328613,
      "learning_rate": 9.903174336580451e-05,
      "loss": 1.5926,
      "step": 1360
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4812482595443726,
      "learning_rate": 9.902334564998321e-05,
      "loss": 1.6019,
      "step": 1370
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6920404434204102,
      "learning_rate": 9.901494793416192e-05,
      "loss": 1.6207,
      "step": 1380
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8390436172485352,
      "learning_rate": 9.900655021834062e-05,
      "loss": 1.6414,
      "step": 1390
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5190553665161133,
      "learning_rate": 9.899815250251931e-05,
      "loss": 1.6214,
      "step": 1400
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5912328958511353,
      "learning_rate": 9.898975478669802e-05,
      "loss": 1.5515,
      "step": 1410
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6428656578063965,
      "learning_rate": 9.898135707087673e-05,
      "loss": 1.5867,
      "step": 1420
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7359733581542969,
      "learning_rate": 9.897295935505543e-05,
      "loss": 1.6235,
      "step": 1430
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4765956401824951,
      "learning_rate": 9.896456163923413e-05,
      "loss": 1.6047,
      "step": 1440
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5506502389907837,
      "learning_rate": 9.895616392341284e-05,
      "loss": 1.6649,
      "step": 1450
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.713420033454895,
      "learning_rate": 9.894776620759154e-05,
      "loss": 1.5648,
      "step": 1460
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3702709674835205,
      "learning_rate": 9.893936849177024e-05,
      "loss": 1.6239,
      "step": 1470
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5771777629852295,
      "learning_rate": 9.893097077594896e-05,
      "loss": 1.5701,
      "step": 1480
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6002380847930908,
      "learning_rate": 9.892257306012766e-05,
      "loss": 1.5596,
      "step": 1490
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6560975313186646,
      "learning_rate": 9.891417534430634e-05,
      "loss": 1.621,
      "step": 1500
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.880449891090393,
      "learning_rate": 9.890577762848506e-05,
      "loss": 1.5582,
      "step": 1510
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.608383297920227,
      "learning_rate": 9.889737991266376e-05,
      "loss": 1.591,
      "step": 1520
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4543036222457886,
      "learning_rate": 9.888898219684246e-05,
      "loss": 1.6353,
      "step": 1530
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5987427234649658,
      "learning_rate": 9.888058448102116e-05,
      "loss": 1.5747,
      "step": 1540
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4784051179885864,
      "learning_rate": 9.887218676519987e-05,
      "loss": 1.589,
      "step": 1550
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4499478340148926,
      "learning_rate": 9.886378904937858e-05,
      "loss": 1.5605,
      "step": 1560
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.133988380432129,
      "learning_rate": 9.885539133355728e-05,
      "loss": 1.5543,
      "step": 1570
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6200871467590332,
      "learning_rate": 9.884699361773598e-05,
      "loss": 1.5787,
      "step": 1580
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5930988788604736,
      "learning_rate": 9.883859590191469e-05,
      "loss": 1.6567,
      "step": 1590
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.498531699180603,
      "learning_rate": 9.883019818609339e-05,
      "loss": 1.5763,
      "step": 1600
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6275144815444946,
      "learning_rate": 9.882180047027209e-05,
      "loss": 1.575,
      "step": 1610
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5897412300109863,
      "learning_rate": 9.881340275445079e-05,
      "loss": 1.5799,
      "step": 1620
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5260472297668457,
      "learning_rate": 9.88050050386295e-05,
      "loss": 1.5347,
      "step": 1630
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5881023406982422,
      "learning_rate": 9.87966073228082e-05,
      "loss": 1.53,
      "step": 1640
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.548144817352295,
      "learning_rate": 9.878820960698691e-05,
      "loss": 1.5536,
      "step": 1650
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6072583198547363,
      "learning_rate": 9.877981189116561e-05,
      "loss": 1.5693,
      "step": 1660
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8414642810821533,
      "learning_rate": 9.877141417534431e-05,
      "loss": 1.5985,
      "step": 1670
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6828004121780396,
      "learning_rate": 9.876301645952301e-05,
      "loss": 1.5622,
      "step": 1680
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4494273662567139,
      "learning_rate": 9.875461874370172e-05,
      "loss": 1.5535,
      "step": 1690
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6359646320343018,
      "learning_rate": 9.874622102788043e-05,
      "loss": 1.5739,
      "step": 1700
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6679375171661377,
      "learning_rate": 9.873782331205913e-05,
      "loss": 1.5461,
      "step": 1710
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.581217646598816,
      "learning_rate": 9.872942559623783e-05,
      "loss": 1.5059,
      "step": 1720
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5814200639724731,
      "learning_rate": 9.872102788041653e-05,
      "loss": 1.5596,
      "step": 1730
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4673880338668823,
      "learning_rate": 9.871263016459523e-05,
      "loss": 1.517,
      "step": 1740
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8205835819244385,
      "learning_rate": 9.870423244877394e-05,
      "loss": 1.5124,
      "step": 1750
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6103378534317017,
      "learning_rate": 9.869583473295264e-05,
      "loss": 1.5258,
      "step": 1760
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6353694200515747,
      "learning_rate": 9.868743701713134e-05,
      "loss": 1.4667,
      "step": 1770
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.530811071395874,
      "learning_rate": 9.867903930131004e-05,
      "loss": 1.4667,
      "step": 1780
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6762795448303223,
      "learning_rate": 9.867064158548876e-05,
      "loss": 1.5848,
      "step": 1790
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.54753839969635,
      "learning_rate": 9.866224386966746e-05,
      "loss": 1.5819,
      "step": 1800
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3980042934417725,
      "learning_rate": 9.865384615384616e-05,
      "loss": 1.5145,
      "step": 1810
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5987871885299683,
      "learning_rate": 9.864544843802486e-05,
      "loss": 1.5615,
      "step": 1820
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.598125696182251,
      "learning_rate": 9.863705072220356e-05,
      "loss": 1.493,
      "step": 1830
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4321807622909546,
      "learning_rate": 9.862865300638226e-05,
      "loss": 1.4781,
      "step": 1840
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6206380128860474,
      "learning_rate": 9.862025529056098e-05,
      "loss": 1.5203,
      "step": 1850
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4826364517211914,
      "learning_rate": 9.861185757473968e-05,
      "loss": 1.5283,
      "step": 1860
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5955564975738525,
      "learning_rate": 9.860345985891838e-05,
      "loss": 1.5282,
      "step": 1870
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.689847469329834,
      "learning_rate": 9.859506214309708e-05,
      "loss": 1.491,
      "step": 1880
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6846811771392822,
      "learning_rate": 9.858666442727579e-05,
      "loss": 1.5077,
      "step": 1890
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6529998779296875,
      "learning_rate": 9.85782667114545e-05,
      "loss": 1.5171,
      "step": 1900
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.607906699180603,
      "learning_rate": 9.85698689956332e-05,
      "loss": 1.4852,
      "step": 1910
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4496583938598633,
      "learning_rate": 9.85614712798119e-05,
      "loss": 1.5379,
      "step": 1920
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5358500480651855,
      "learning_rate": 9.85530735639906e-05,
      "loss": 1.4627,
      "step": 1930
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.600895643234253,
      "learning_rate": 9.85446758481693e-05,
      "loss": 1.4699,
      "step": 1940
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5847537517547607,
      "learning_rate": 9.8536278132348e-05,
      "loss": 1.4595,
      "step": 1950
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5020571947097778,
      "learning_rate": 9.852788041652671e-05,
      "loss": 1.4592,
      "step": 1960
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.583520531654358,
      "learning_rate": 9.851948270070541e-05,
      "loss": 1.4813,
      "step": 1970
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5839787721633911,
      "learning_rate": 9.851108498488411e-05,
      "loss": 1.4936,
      "step": 1980
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5823278427124023,
      "learning_rate": 9.850268726906283e-05,
      "loss": 1.5422,
      "step": 1990
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7043323516845703,
      "learning_rate": 9.849428955324153e-05,
      "loss": 1.4834,
      "step": 2000
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.630943775177002,
      "learning_rate": 9.848589183742023e-05,
      "loss": 1.4653,
      "step": 2010
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5110965967178345,
      "learning_rate": 9.847749412159893e-05,
      "loss": 1.46,
      "step": 2020
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.657193660736084,
      "learning_rate": 9.846909640577763e-05,
      "loss": 1.4669,
      "step": 2030
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6347354650497437,
      "learning_rate": 9.846069868995633e-05,
      "loss": 1.502,
      "step": 2040
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6823369264602661,
      "learning_rate": 9.845230097413503e-05,
      "loss": 1.4591,
      "step": 2050
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7435994148254395,
      "learning_rate": 9.844390325831375e-05,
      "loss": 1.4785,
      "step": 2060
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4445375204086304,
      "learning_rate": 9.843550554249245e-05,
      "loss": 1.4508,
      "step": 2070
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5260076522827148,
      "learning_rate": 9.842710782667115e-05,
      "loss": 1.4983,
      "step": 2080
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6740764379501343,
      "learning_rate": 9.841871011084986e-05,
      "loss": 1.4386,
      "step": 2090
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.517861008644104,
      "learning_rate": 9.841031239502856e-05,
      "loss": 1.4352,
      "step": 2100
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.708284854888916,
      "learning_rate": 9.840191467920726e-05,
      "loss": 1.4434,
      "step": 2110
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4811726808547974,
      "learning_rate": 9.839351696338596e-05,
      "loss": 1.4244,
      "step": 2120
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6109236478805542,
      "learning_rate": 9.838511924756468e-05,
      "loss": 1.4156,
      "step": 2130
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6670873165130615,
      "learning_rate": 9.837672153174336e-05,
      "loss": 1.4482,
      "step": 2140
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3890584707260132,
      "learning_rate": 9.836832381592207e-05,
      "loss": 1.448,
      "step": 2150
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6003090143203735,
      "learning_rate": 9.835992610010078e-05,
      "loss": 1.446,
      "step": 2160
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4101319313049316,
      "learning_rate": 9.835152838427948e-05,
      "loss": 1.4429,
      "step": 2170
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7009485960006714,
      "learning_rate": 9.834313066845818e-05,
      "loss": 1.4242,
      "step": 2180
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4896700382232666,
      "learning_rate": 9.833473295263688e-05,
      "loss": 1.4165,
      "step": 2190
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4147573709487915,
      "learning_rate": 9.83263352368156e-05,
      "loss": 1.4139,
      "step": 2200
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4627562761306763,
      "learning_rate": 9.83179375209943e-05,
      "loss": 1.4183,
      "step": 2210
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4203674793243408,
      "learning_rate": 9.8309539805173e-05,
      "loss": 1.4039,
      "step": 2220
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2355053424835205,
      "learning_rate": 9.830114208935171e-05,
      "loss": 1.4043,
      "step": 2230
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4707199335098267,
      "learning_rate": 9.829274437353041e-05,
      "loss": 1.3782,
      "step": 2240
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3370139598846436,
      "learning_rate": 9.82843466577091e-05,
      "loss": 1.386,
      "step": 2250
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.231668472290039,
      "learning_rate": 9.827594894188781e-05,
      "loss": 1.4013,
      "step": 2260
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7293065786361694,
      "learning_rate": 9.826755122606651e-05,
      "loss": 1.435,
      "step": 2270
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3585022687911987,
      "learning_rate": 9.825915351024521e-05,
      "loss": 1.407,
      "step": 2280
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3481025695800781,
      "learning_rate": 9.825075579442392e-05,
      "loss": 1.3815,
      "step": 2290
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.437150239944458,
      "learning_rate": 9.824235807860263e-05,
      "loss": 1.3801,
      "step": 2300
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.405595302581787,
      "learning_rate": 9.823396036278133e-05,
      "loss": 1.3788,
      "step": 2310
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2758052349090576,
      "learning_rate": 9.822556264696003e-05,
      "loss": 1.3897,
      "step": 2320
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2883045673370361,
      "learning_rate": 9.821716493113875e-05,
      "loss": 1.4058,
      "step": 2330
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2251172065734863,
      "learning_rate": 9.820876721531745e-05,
      "loss": 1.3813,
      "step": 2340
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2864748239517212,
      "learning_rate": 9.820036949949613e-05,
      "loss": 1.4176,
      "step": 2350
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1874711513519287,
      "learning_rate": 9.819197178367485e-05,
      "loss": 1.4,
      "step": 2360
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.185160517692566,
      "learning_rate": 9.818357406785355e-05,
      "loss": 1.3999,
      "step": 2370
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1715202331542969,
      "learning_rate": 9.817517635203225e-05,
      "loss": 1.3976,
      "step": 2380
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2971388101577759,
      "learning_rate": 9.816677863621095e-05,
      "loss": 1.3743,
      "step": 2390
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3223909139633179,
      "learning_rate": 9.815838092038966e-05,
      "loss": 1.4166,
      "step": 2400
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.138414978981018,
      "learning_rate": 9.814998320456836e-05,
      "loss": 1.3848,
      "step": 2410
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0639983415603638,
      "learning_rate": 9.814158548874706e-05,
      "loss": 1.4086,
      "step": 2420
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1832982301712036,
      "learning_rate": 9.813318777292577e-05,
      "loss": 1.3755,
      "step": 2430
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1679545640945435,
      "learning_rate": 9.812479005710448e-05,
      "loss": 1.3657,
      "step": 2440
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3011316061019897,
      "learning_rate": 9.811639234128318e-05,
      "loss": 1.3952,
      "step": 2450
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.189110279083252,
      "learning_rate": 9.810799462546187e-05,
      "loss": 1.3809,
      "step": 2460
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1846518516540527,
      "learning_rate": 9.809959690964058e-05,
      "loss": 1.37,
      "step": 2470
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.091399908065796,
      "learning_rate": 9.809119919381928e-05,
      "loss": 1.3758,
      "step": 2480
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2157222032546997,
      "learning_rate": 9.808280147799798e-05,
      "loss": 1.3915,
      "step": 2490
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0647687911987305,
      "learning_rate": 9.80744037621767e-05,
      "loss": 1.3691,
      "step": 2500
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2293319702148438,
      "learning_rate": 9.80660060463554e-05,
      "loss": 1.3363,
      "step": 2510
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9640077948570251,
      "learning_rate": 9.80576083305341e-05,
      "loss": 1.3572,
      "step": 2520
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1838915348052979,
      "learning_rate": 9.80492106147128e-05,
      "loss": 1.3949,
      "step": 2530
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2546766996383667,
      "learning_rate": 9.804081289889151e-05,
      "loss": 1.3735,
      "step": 2540
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.21607506275177,
      "learning_rate": 9.803241518307021e-05,
      "loss": 1.3972,
      "step": 2550
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1871435642242432,
      "learning_rate": 9.80240174672489e-05,
      "loss": 1.3574,
      "step": 2560
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.048129677772522,
      "learning_rate": 9.801561975142762e-05,
      "loss": 1.3403,
      "step": 2570
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2536293268203735,
      "learning_rate": 9.800722203560632e-05,
      "loss": 1.3687,
      "step": 2580
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1996736526489258,
      "learning_rate": 9.799882431978502e-05,
      "loss": 1.3523,
      "step": 2590
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1699773073196411,
      "learning_rate": 9.799042660396373e-05,
      "loss": 1.33,
      "step": 2600
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3010507822036743,
      "learning_rate": 9.798202888814243e-05,
      "loss": 1.3843,
      "step": 2610
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1527621746063232,
      "learning_rate": 9.797363117232113e-05,
      "loss": 1.3588,
      "step": 2620
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1328866481781006,
      "learning_rate": 9.796523345649983e-05,
      "loss": 1.3437,
      "step": 2630
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.090882658958435,
      "learning_rate": 9.795683574067855e-05,
      "loss": 1.3776,
      "step": 2640
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.256652593612671,
      "learning_rate": 9.794843802485725e-05,
      "loss": 1.3196,
      "step": 2650
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0616453886032104,
      "learning_rate": 9.794004030903595e-05,
      "loss": 1.341,
      "step": 2660
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.085680365562439,
      "learning_rate": 9.793164259321465e-05,
      "loss": 1.3547,
      "step": 2670
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0737241506576538,
      "learning_rate": 9.792324487739335e-05,
      "loss": 1.3835,
      "step": 2680
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.245523452758789,
      "learning_rate": 9.791484716157205e-05,
      "loss": 1.3737,
      "step": 2690
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0836365222930908,
      "learning_rate": 9.790644944575075e-05,
      "loss": 1.3777,
      "step": 2700
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1912739276885986,
      "learning_rate": 9.789805172992947e-05,
      "loss": 1.3563,
      "step": 2710
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2736655473709106,
      "learning_rate": 9.788965401410817e-05,
      "loss": 1.3194,
      "step": 2720
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.104282021522522,
      "learning_rate": 9.788125629828687e-05,
      "loss": 1.3479,
      "step": 2730
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.048940658569336,
      "learning_rate": 9.787285858246558e-05,
      "loss": 1.3532,
      "step": 2740
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0983320474624634,
      "learning_rate": 9.786446086664428e-05,
      "loss": 1.3562,
      "step": 2750
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1402326822280884,
      "learning_rate": 9.785606315082298e-05,
      "loss": 1.354,
      "step": 2760
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2619010210037231,
      "learning_rate": 9.784766543500168e-05,
      "loss": 1.3521,
      "step": 2770
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0591953992843628,
      "learning_rate": 9.783926771918038e-05,
      "loss": 1.3789,
      "step": 2780
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1339819431304932,
      "learning_rate": 9.783087000335909e-05,
      "loss": 1.3256,
      "step": 2790
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1497844457626343,
      "learning_rate": 9.782247228753779e-05,
      "loss": 1.3132,
      "step": 2800
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1759591102600098,
      "learning_rate": 9.78140745717165e-05,
      "loss": 1.3177,
      "step": 2810
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9412186145782471,
      "learning_rate": 9.78056768558952e-05,
      "loss": 1.3229,
      "step": 2820
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9941621422767639,
      "learning_rate": 9.77972791400739e-05,
      "loss": 1.3377,
      "step": 2830
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.157085657119751,
      "learning_rate": 9.778888142425262e-05,
      "loss": 1.3507,
      "step": 2840
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0099543333053589,
      "learning_rate": 9.778048370843132e-05,
      "loss": 1.3161,
      "step": 2850
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0422873497009277,
      "learning_rate": 9.777208599261002e-05,
      "loss": 1.333,
      "step": 2860
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9914569854736328,
      "learning_rate": 9.776368827678872e-05,
      "loss": 1.3464,
      "step": 2870
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.021345853805542,
      "learning_rate": 9.775529056096742e-05,
      "loss": 1.3211,
      "step": 2880
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.109337568283081,
      "learning_rate": 9.774689284514612e-05,
      "loss": 1.3276,
      "step": 2890
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.103621482849121,
      "learning_rate": 9.773849512932482e-05,
      "loss": 1.3461,
      "step": 2900
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9739851355552673,
      "learning_rate": 9.773009741350353e-05,
      "loss": 1.3455,
      "step": 2910
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0525223016738892,
      "learning_rate": 9.772169969768223e-05,
      "loss": 1.3217,
      "step": 2920
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0851151943206787,
      "learning_rate": 9.771330198186094e-05,
      "loss": 1.3086,
      "step": 2930
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9990787506103516,
      "learning_rate": 9.770490426603964e-05,
      "loss": 1.3356,
      "step": 2940
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.95279461145401,
      "learning_rate": 9.769650655021835e-05,
      "loss": 1.3354,
      "step": 2950
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.021781325340271,
      "learning_rate": 9.768810883439705e-05,
      "loss": 1.345,
      "step": 2960
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0377811193466187,
      "learning_rate": 9.767971111857575e-05,
      "loss": 1.336,
      "step": 2970
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1588681936264038,
      "learning_rate": 9.767131340275447e-05,
      "loss": 1.3243,
      "step": 2980
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.020140528678894,
      "learning_rate": 9.766291568693315e-05,
      "loss": 1.3228,
      "step": 2990
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0705044269561768,
      "learning_rate": 9.765451797111185e-05,
      "loss": 1.3296,
      "step": 3000
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0218137502670288,
      "learning_rate": 9.764612025529057e-05,
      "loss": 1.3411,
      "step": 3010
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9504896402359009,
      "learning_rate": 9.763772253946927e-05,
      "loss": 1.3453,
      "step": 3020
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9712496399879456,
      "learning_rate": 9.762932482364797e-05,
      "loss": 1.3374,
      "step": 3030
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0498982667922974,
      "learning_rate": 9.762092710782667e-05,
      "loss": 1.3487,
      "step": 3040
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.003515601158142,
      "learning_rate": 9.761252939200538e-05,
      "loss": 1.3378,
      "step": 3050
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0123205184936523,
      "learning_rate": 9.760413167618409e-05,
      "loss": 1.3169,
      "step": 3060
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0419187545776367,
      "learning_rate": 9.759573396036279e-05,
      "loss": 1.2904,
      "step": 3070
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0047117471694946,
      "learning_rate": 9.75873362445415e-05,
      "loss": 1.3109,
      "step": 3080
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0547492504119873,
      "learning_rate": 9.757893852872019e-05,
      "loss": 1.3664,
      "step": 3090
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0447371006011963,
      "learning_rate": 9.757054081289889e-05,
      "loss": 1.3109,
      "step": 3100
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9989799857139587,
      "learning_rate": 9.75621430970776e-05,
      "loss": 1.3102,
      "step": 3110
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0217671394348145,
      "learning_rate": 9.75537453812563e-05,
      "loss": 1.329,
      "step": 3120
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0475112199783325,
      "learning_rate": 9.7545347665435e-05,
      "loss": 1.3207,
      "step": 3130
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1760674715042114,
      "learning_rate": 9.75369499496137e-05,
      "loss": 1.3253,
      "step": 3140
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0421987771987915,
      "learning_rate": 9.752855223379242e-05,
      "loss": 1.3213,
      "step": 3150
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.071460247039795,
      "learning_rate": 9.752015451797112e-05,
      "loss": 1.3033,
      "step": 3160
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.061184287071228,
      "learning_rate": 9.751175680214982e-05,
      "loss": 1.327,
      "step": 3170
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9795036315917969,
      "learning_rate": 9.750335908632852e-05,
      "loss": 1.3328,
      "step": 3180
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0202151536941528,
      "learning_rate": 9.749496137050723e-05,
      "loss": 1.3191,
      "step": 3190
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9541215896606445,
      "learning_rate": 9.748656365468592e-05,
      "loss": 1.3145,
      "step": 3200
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9995015263557434,
      "learning_rate": 9.747816593886464e-05,
      "loss": 1.3217,
      "step": 3210
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1494383811950684,
      "learning_rate": 9.746976822304334e-05,
      "loss": 1.3004,
      "step": 3220
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0284998416900635,
      "learning_rate": 9.746137050722204e-05,
      "loss": 1.3176,
      "step": 3230
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1794145107269287,
      "learning_rate": 9.745297279140074e-05,
      "loss": 1.3441,
      "step": 3240
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.961778461933136,
      "learning_rate": 9.744457507557945e-05,
      "loss": 1.3103,
      "step": 3250
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9705653786659241,
      "learning_rate": 9.743617735975815e-05,
      "loss": 1.3338,
      "step": 3260
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.991642951965332,
      "learning_rate": 9.742777964393685e-05,
      "loss": 1.3029,
      "step": 3270
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0278716087341309,
      "learning_rate": 9.741938192811555e-05,
      "loss": 1.3048,
      "step": 3280
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9765398502349854,
      "learning_rate": 9.741098421229427e-05,
      "loss": 1.3168,
      "step": 3290
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9255518913269043,
      "learning_rate": 9.740258649647297e-05,
      "loss": 1.3615,
      "step": 3300
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9744580388069153,
      "learning_rate": 9.739418878065166e-05,
      "loss": 1.3185,
      "step": 3310
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.084482192993164,
      "learning_rate": 9.738579106483037e-05,
      "loss": 1.3495,
      "step": 3320
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2292135953903198,
      "learning_rate": 9.737739334900907e-05,
      "loss": 1.3105,
      "step": 3330
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.961018443107605,
      "learning_rate": 9.736899563318777e-05,
      "loss": 1.3035,
      "step": 3340
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.075959324836731,
      "learning_rate": 9.736059791736649e-05,
      "loss": 1.3054,
      "step": 3350
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0361067056655884,
      "learning_rate": 9.735220020154519e-05,
      "loss": 1.2918,
      "step": 3360
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9005823731422424,
      "learning_rate": 9.734380248572389e-05,
      "loss": 1.3137,
      "step": 3370
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9153285026550293,
      "learning_rate": 9.733540476990259e-05,
      "loss": 1.3288,
      "step": 3380
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8635299205780029,
      "learning_rate": 9.73270070540813e-05,
      "loss": 1.3137,
      "step": 3390
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0367728471755981,
      "learning_rate": 9.731860933826e-05,
      "loss": 1.2876,
      "step": 3400
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9176615476608276,
      "learning_rate": 9.731021162243869e-05,
      "loss": 1.3387,
      "step": 3410
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0236858129501343,
      "learning_rate": 9.73018139066174e-05,
      "loss": 1.3155,
      "step": 3420
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9770137071609497,
      "learning_rate": 9.72934161907961e-05,
      "loss": 1.3125,
      "step": 3430
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9246246814727783,
      "learning_rate": 9.72850184749748e-05,
      "loss": 1.336,
      "step": 3440
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9181159138679504,
      "learning_rate": 9.727662075915352e-05,
      "loss": 1.308,
      "step": 3450
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0865639448165894,
      "learning_rate": 9.726822304333222e-05,
      "loss": 1.301,
      "step": 3460
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8808737993240356,
      "learning_rate": 9.725982532751092e-05,
      "loss": 1.2944,
      "step": 3470
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.91978919506073,
      "learning_rate": 9.725142761168962e-05,
      "loss": 1.2873,
      "step": 3480
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9652817845344543,
      "learning_rate": 9.724302989586834e-05,
      "loss": 1.2933,
      "step": 3490
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8826923966407776,
      "learning_rate": 9.723463218004704e-05,
      "loss": 1.3111,
      "step": 3500
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9343136548995972,
      "learning_rate": 9.722623446422574e-05,
      "loss": 1.3041,
      "step": 3510
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9622355699539185,
      "learning_rate": 9.721783674840444e-05,
      "loss": 1.3273,
      "step": 3520
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8856210112571716,
      "learning_rate": 9.720943903258314e-05,
      "loss": 1.308,
      "step": 3530
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0706381797790527,
      "learning_rate": 9.720104131676184e-05,
      "loss": 1.2895,
      "step": 3540
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9392056465148926,
      "learning_rate": 9.719264360094054e-05,
      "loss": 1.3051,
      "step": 3550
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9185264110565186,
      "learning_rate": 9.718424588511926e-05,
      "loss": 1.3149,
      "step": 3560
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9602808356285095,
      "learning_rate": 9.717584816929796e-05,
      "loss": 1.3119,
      "step": 3570
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9972320795059204,
      "learning_rate": 9.716745045347666e-05,
      "loss": 1.2968,
      "step": 3580
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9500466585159302,
      "learning_rate": 9.715905273765537e-05,
      "loss": 1.3258,
      "step": 3590
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.031611680984497,
      "learning_rate": 9.715065502183407e-05,
      "loss": 1.2904,
      "step": 3600
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9221827387809753,
      "learning_rate": 9.714225730601277e-05,
      "loss": 1.2762,
      "step": 3610
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0512107610702515,
      "learning_rate": 9.713385959019147e-05,
      "loss": 1.3046,
      "step": 3620
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9803313612937927,
      "learning_rate": 9.712546187437017e-05,
      "loss": 1.3,
      "step": 3630
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8865204453468323,
      "learning_rate": 9.711706415854887e-05,
      "loss": 1.3094,
      "step": 3640
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0050348043441772,
      "learning_rate": 9.710866644272758e-05,
      "loss": 1.3196,
      "step": 3650
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8442556262016296,
      "learning_rate": 9.710026872690629e-05,
      "loss": 1.3467,
      "step": 3660
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9233782887458801,
      "learning_rate": 9.709187101108499e-05,
      "loss": 1.2706,
      "step": 3670
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9729004502296448,
      "learning_rate": 9.708347329526369e-05,
      "loss": 1.336,
      "step": 3680
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9508803486824036,
      "learning_rate": 9.70750755794424e-05,
      "loss": 1.3274,
      "step": 3690
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9443732500076294,
      "learning_rate": 9.70666778636211e-05,
      "loss": 1.2841,
      "step": 3700
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.868013322353363,
      "learning_rate": 9.70582801477998e-05,
      "loss": 1.2957,
      "step": 3710
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.965334415435791,
      "learning_rate": 9.704988243197851e-05,
      "loss": 1.3304,
      "step": 3720
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9084999561309814,
      "learning_rate": 9.704148471615721e-05,
      "loss": 1.2923,
      "step": 3730
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.919133722782135,
      "learning_rate": 9.703308700033591e-05,
      "loss": 1.29,
      "step": 3740
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9951848983764648,
      "learning_rate": 9.702468928451461e-05,
      "loss": 1.259,
      "step": 3750
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9185649752616882,
      "learning_rate": 9.701629156869332e-05,
      "loss": 1.2836,
      "step": 3760
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.054055094718933,
      "learning_rate": 9.700789385287202e-05,
      "loss": 1.3071,
      "step": 3770
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8889206051826477,
      "learning_rate": 9.699949613705072e-05,
      "loss": 1.288,
      "step": 3780
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0557209253311157,
      "learning_rate": 9.699109842122943e-05,
      "loss": 1.2997,
      "step": 3790
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9376937747001648,
      "learning_rate": 9.698270070540814e-05,
      "loss": 1.2725,
      "step": 3800
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9468439817428589,
      "learning_rate": 9.697430298958684e-05,
      "loss": 1.2995,
      "step": 3810
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9373131394386292,
      "learning_rate": 9.696590527376554e-05,
      "loss": 1.2954,
      "step": 3820
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9124709367752075,
      "learning_rate": 9.695750755794426e-05,
      "loss": 1.2801,
      "step": 3830
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8868261575698853,
      "learning_rate": 9.694910984212294e-05,
      "loss": 1.3157,
      "step": 3840
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9589577317237854,
      "learning_rate": 9.694071212630164e-05,
      "loss": 1.2965,
      "step": 3850
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.692502737045288,
      "learning_rate": 9.693231441048036e-05,
      "loss": 1.2856,
      "step": 3860
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1974432468414307,
      "learning_rate": 9.692391669465906e-05,
      "loss": 1.297,
      "step": 3870
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.018686294555664,
      "learning_rate": 9.691551897883776e-05,
      "loss": 1.2363,
      "step": 3880
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9238420128822327,
      "learning_rate": 9.690712126301646e-05,
      "loss": 1.3077,
      "step": 3890
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0184767246246338,
      "learning_rate": 9.689872354719517e-05,
      "loss": 1.2869,
      "step": 3900
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2680937051773071,
      "learning_rate": 9.689032583137387e-05,
      "loss": 1.2759,
      "step": 3910
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0533655881881714,
      "learning_rate": 9.688192811555257e-05,
      "loss": 1.3035,
      "step": 3920
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1051759719848633,
      "learning_rate": 9.687353039973129e-05,
      "loss": 1.2891,
      "step": 3930
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1200437545776367,
      "learning_rate": 9.686513268390998e-05,
      "loss": 1.2887,
      "step": 3940
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.049575686454773,
      "learning_rate": 9.685673496808868e-05,
      "loss": 1.2664,
      "step": 3950
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9664704203605652,
      "learning_rate": 9.684833725226739e-05,
      "loss": 1.2763,
      "step": 3960
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.984643280506134,
      "learning_rate": 9.683993953644609e-05,
      "loss": 1.3093,
      "step": 3970
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1789389848709106,
      "learning_rate": 9.683154182062479e-05,
      "loss": 1.3181,
      "step": 3980
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9238845109939575,
      "learning_rate": 9.68231441048035e-05,
      "loss": 1.2999,
      "step": 3990
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9249722957611084,
      "learning_rate": 9.681474638898221e-05,
      "loss": 1.278,
      "step": 4000
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9088603258132935,
      "learning_rate": 9.680634867316091e-05,
      "loss": 1.2853,
      "step": 4010
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9020570516586304,
      "learning_rate": 9.679795095733961e-05,
      "loss": 1.289,
      "step": 4020
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.940483570098877,
      "learning_rate": 9.678955324151831e-05,
      "loss": 1.2916,
      "step": 4030
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9913098812103271,
      "learning_rate": 9.678115552569702e-05,
      "loss": 1.2713,
      "step": 4040
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9511467218399048,
      "learning_rate": 9.677275780987571e-05,
      "loss": 1.2768,
      "step": 4050
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9045311212539673,
      "learning_rate": 9.676436009405441e-05,
      "loss": 1.2752,
      "step": 4060
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8729120492935181,
      "learning_rate": 9.675596237823313e-05,
      "loss": 1.2921,
      "step": 4070
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8952471613883972,
      "learning_rate": 9.674756466241183e-05,
      "loss": 1.2502,
      "step": 4080
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.986374020576477,
      "learning_rate": 9.673916694659053e-05,
      "loss": 1.2775,
      "step": 4090
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8978623747825623,
      "learning_rate": 9.673076923076924e-05,
      "loss": 1.3128,
      "step": 4100
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8826311826705933,
      "learning_rate": 9.672237151494794e-05,
      "loss": 1.297,
      "step": 4110
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.96685391664505,
      "learning_rate": 9.671397379912664e-05,
      "loss": 1.2506,
      "step": 4120
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9718626141548157,
      "learning_rate": 9.670557608330534e-05,
      "loss": 1.2876,
      "step": 4130
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8737813830375671,
      "learning_rate": 9.669717836748406e-05,
      "loss": 1.2789,
      "step": 4140
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9341663122177124,
      "learning_rate": 9.668878065166275e-05,
      "loss": 1.2711,
      "step": 4150
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.970474123954773,
      "learning_rate": 9.668038293584145e-05,
      "loss": 1.2909,
      "step": 4160
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8914270401000977,
      "learning_rate": 9.667198522002016e-05,
      "loss": 1.2858,
      "step": 4170
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9272024631500244,
      "learning_rate": 9.666358750419886e-05,
      "loss": 1.2892,
      "step": 4180
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0669282674789429,
      "learning_rate": 9.665518978837756e-05,
      "loss": 1.2738,
      "step": 4190
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.016957402229309,
      "learning_rate": 9.664679207255628e-05,
      "loss": 1.2725,
      "step": 4200
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8794147372245789,
      "learning_rate": 9.663839435673498e-05,
      "loss": 1.3021,
      "step": 4210
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9839082956314087,
      "learning_rate": 9.662999664091368e-05,
      "loss": 1.2664,
      "step": 4220
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9609381556510925,
      "learning_rate": 9.662159892509238e-05,
      "loss": 1.2475,
      "step": 4230
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9309755563735962,
      "learning_rate": 9.661320120927109e-05,
      "loss": 1.291,
      "step": 4240
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9350253343582153,
      "learning_rate": 9.660480349344979e-05,
      "loss": 1.2724,
      "step": 4250
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.842782735824585,
      "learning_rate": 9.659640577762848e-05,
      "loss": 1.2922,
      "step": 4260
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9164971709251404,
      "learning_rate": 9.65880080618072e-05,
      "loss": 1.2648,
      "step": 4270
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8968715071678162,
      "learning_rate": 9.65796103459859e-05,
      "loss": 1.27,
      "step": 4280
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9750767946243286,
      "learning_rate": 9.65712126301646e-05,
      "loss": 1.2772,
      "step": 4290
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8595716953277588,
      "learning_rate": 9.65628149143433e-05,
      "loss": 1.272,
      "step": 4300
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9079431295394897,
      "learning_rate": 9.655441719852201e-05,
      "loss": 1.2988,
      "step": 4310
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8873557448387146,
      "learning_rate": 9.654601948270071e-05,
      "loss": 1.2621,
      "step": 4320
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9525663256645203,
      "learning_rate": 9.653762176687941e-05,
      "loss": 1.304,
      "step": 4330
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9584593176841736,
      "learning_rate": 9.652922405105813e-05,
      "loss": 1.2749,
      "step": 4340
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9332358837127686,
      "learning_rate": 9.652082633523683e-05,
      "loss": 1.2664,
      "step": 4350
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9320657849311829,
      "learning_rate": 9.651242861941551e-05,
      "loss": 1.2689,
      "step": 4360
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9525415301322937,
      "learning_rate": 9.650403090359423e-05,
      "loss": 1.2391,
      "step": 4370
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9248800873756409,
      "learning_rate": 9.649563318777293e-05,
      "loss": 1.2483,
      "step": 4380
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9141398072242737,
      "learning_rate": 9.648723547195163e-05,
      "loss": 1.2667,
      "step": 4390
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0400909185409546,
      "learning_rate": 9.647883775613033e-05,
      "loss": 1.2725,
      "step": 4400
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8996667265892029,
      "learning_rate": 9.647044004030904e-05,
      "loss": 1.2561,
      "step": 4410
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9660255908966064,
      "learning_rate": 9.646204232448774e-05,
      "loss": 1.2951,
      "step": 4420
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8532864451408386,
      "learning_rate": 9.645364460866645e-05,
      "loss": 1.2612,
      "step": 4430
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8848567605018616,
      "learning_rate": 9.644524689284516e-05,
      "loss": 1.2857,
      "step": 4440
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9615711569786072,
      "learning_rate": 9.643684917702386e-05,
      "loss": 1.2464,
      "step": 4450
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8904231786727905,
      "learning_rate": 9.642845146120256e-05,
      "loss": 1.2728,
      "step": 4460
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8821005821228027,
      "learning_rate": 9.642005374538126e-05,
      "loss": 1.2547,
      "step": 4470
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9640454649925232,
      "learning_rate": 9.641165602955996e-05,
      "loss": 1.2678,
      "step": 4480
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9105495810508728,
      "learning_rate": 9.640325831373866e-05,
      "loss": 1.292,
      "step": 4490
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8753247261047363,
      "learning_rate": 9.639486059791736e-05,
      "loss": 1.2948,
      "step": 4500
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9737322926521301,
      "learning_rate": 9.638646288209608e-05,
      "loss": 1.2815,
      "step": 4510
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.990397572517395,
      "learning_rate": 9.637806516627478e-05,
      "loss": 1.2795,
      "step": 4520
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8985497951507568,
      "learning_rate": 9.636966745045348e-05,
      "loss": 1.2681,
      "step": 4530
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9467973709106445,
      "learning_rate": 9.636126973463218e-05,
      "loss": 1.2363,
      "step": 4540
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8291370868682861,
      "learning_rate": 9.63528720188109e-05,
      "loss": 1.2906,
      "step": 4550
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9621056914329529,
      "learning_rate": 9.63444743029896e-05,
      "loss": 1.2666,
      "step": 4560
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9514476656913757,
      "learning_rate": 9.63360765871683e-05,
      "loss": 1.2506,
      "step": 4570
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0280085802078247,
      "learning_rate": 9.6327678871347e-05,
      "loss": 1.2565,
      "step": 4580
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8504949808120728,
      "learning_rate": 9.63192811555257e-05,
      "loss": 1.2685,
      "step": 4590
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6871306896209717,
      "learning_rate": 9.63108834397044e-05,
      "loss": 1.2822,
      "step": 4600
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9096508026123047,
      "learning_rate": 9.630248572388311e-05,
      "loss": 1.2605,
      "step": 4610
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9401909112930298,
      "learning_rate": 9.629408800806181e-05,
      "loss": 1.2963,
      "step": 4620
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9390792846679688,
      "learning_rate": 9.628569029224051e-05,
      "loss": 1.2491,
      "step": 4630
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9321569204330444,
      "learning_rate": 9.627729257641921e-05,
      "loss": 1.2379,
      "step": 4640
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8815140724182129,
      "learning_rate": 9.626889486059793e-05,
      "loss": 1.2647,
      "step": 4650
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9437508583068848,
      "learning_rate": 9.626049714477663e-05,
      "loss": 1.2941,
      "step": 4660
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9065150618553162,
      "learning_rate": 9.625209942895533e-05,
      "loss": 1.2586,
      "step": 4670
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.897009551525116,
      "learning_rate": 9.624370171313403e-05,
      "loss": 1.2517,
      "step": 4680
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9548740983009338,
      "learning_rate": 9.623530399731273e-05,
      "loss": 1.2523,
      "step": 4690
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9641335606575012,
      "learning_rate": 9.622690628149143e-05,
      "loss": 1.2527,
      "step": 4700
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9327057600021362,
      "learning_rate": 9.621850856567015e-05,
      "loss": 1.2544,
      "step": 4710
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8007282018661499,
      "learning_rate": 9.621011084984885e-05,
      "loss": 1.2438,
      "step": 4720
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9581779837608337,
      "learning_rate": 9.620171313402755e-05,
      "loss": 1.2542,
      "step": 4730
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8529838919639587,
      "learning_rate": 9.619331541820625e-05,
      "loss": 1.2661,
      "step": 4740
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8764471411705017,
      "learning_rate": 9.618491770238496e-05,
      "loss": 1.2458,
      "step": 4750
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8999229669570923,
      "learning_rate": 9.617651998656366e-05,
      "loss": 1.2775,
      "step": 4760
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8830211758613586,
      "learning_rate": 9.616812227074236e-05,
      "loss": 1.2525,
      "step": 4770
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8796391487121582,
      "learning_rate": 9.615972455492106e-05,
      "loss": 1.2651,
      "step": 4780
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0258424282073975,
      "learning_rate": 9.615132683909977e-05,
      "loss": 1.2467,
      "step": 4790
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9051372408866882,
      "learning_rate": 9.614292912327847e-05,
      "loss": 1.251,
      "step": 4800
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8889206051826477,
      "learning_rate": 9.613453140745717e-05,
      "loss": 1.2472,
      "step": 4810
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8488065600395203,
      "learning_rate": 9.612613369163588e-05,
      "loss": 1.2444,
      "step": 4820
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8944873213768005,
      "learning_rate": 9.611773597581458e-05,
      "loss": 1.2504,
      "step": 4830
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8767454028129578,
      "learning_rate": 9.610933825999328e-05,
      "loss": 1.2464,
      "step": 4840
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8748544454574585,
      "learning_rate": 9.6100940544172e-05,
      "loss": 1.2676,
      "step": 4850
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9283246397972107,
      "learning_rate": 9.60925428283507e-05,
      "loss": 1.2513,
      "step": 4860
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9286162853240967,
      "learning_rate": 9.60841451125294e-05,
      "loss": 1.2723,
      "step": 4870
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8313989639282227,
      "learning_rate": 9.60757473967081e-05,
      "loss": 1.2681,
      "step": 4880
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9085673093795776,
      "learning_rate": 9.60673496808868e-05,
      "loss": 1.2321,
      "step": 4890
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9357420802116394,
      "learning_rate": 9.60589519650655e-05,
      "loss": 1.2549,
      "step": 4900
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.879021942615509,
      "learning_rate": 9.60505542492442e-05,
      "loss": 1.2443,
      "step": 4910
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8786951303482056,
      "learning_rate": 9.604215653342291e-05,
      "loss": 1.2433,
      "step": 4920
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9177870154380798,
      "learning_rate": 9.603375881760162e-05,
      "loss": 1.2367,
      "step": 4930
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.951474130153656,
      "learning_rate": 9.602536110178032e-05,
      "loss": 1.227,
      "step": 4940
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0126953125,
      "learning_rate": 9.601696338595903e-05,
      "loss": 1.2377,
      "step": 4950
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9180831909179688,
      "learning_rate": 9.600856567013773e-05,
      "loss": 1.2337,
      "step": 4960
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9630677700042725,
      "learning_rate": 9.600016795431643e-05,
      "loss": 1.2728,
      "step": 4970
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.049486517906189,
      "learning_rate": 9.599177023849513e-05,
      "loss": 1.2572,
      "step": 4980
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9184420108795166,
      "learning_rate": 9.598337252267385e-05,
      "loss": 1.2352,
      "step": 4990
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9000878930091858,
      "learning_rate": 9.597497480685253e-05,
      "loss": 1.239,
      "step": 5000
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9162322282791138,
      "learning_rate": 9.596657709103123e-05,
      "loss": 1.2272,
      "step": 5010
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9241054654121399,
      "learning_rate": 9.595817937520995e-05,
      "loss": 1.2541,
      "step": 5020
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9283685088157654,
      "learning_rate": 9.594978165938865e-05,
      "loss": 1.2387,
      "step": 5030
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7603573203086853,
      "learning_rate": 9.594138394356735e-05,
      "loss": 1.2344,
      "step": 5040
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8442432880401611,
      "learning_rate": 9.593298622774606e-05,
      "loss": 1.2368,
      "step": 5050
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8469812273979187,
      "learning_rate": 9.592458851192477e-05,
      "loss": 1.2575,
      "step": 5060
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9024328589439392,
      "learning_rate": 9.591619079610347e-05,
      "loss": 1.2473,
      "step": 5070
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8618769645690918,
      "learning_rate": 9.590779308028217e-05,
      "loss": 1.259,
      "step": 5080
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8754567503929138,
      "learning_rate": 9.589939536446088e-05,
      "loss": 1.2315,
      "step": 5090
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9073194861412048,
      "learning_rate": 9.589099764863958e-05,
      "loss": 1.2162,
      "step": 5100
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8669455647468567,
      "learning_rate": 9.588259993281827e-05,
      "loss": 1.2629,
      "step": 5110
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9005383849143982,
      "learning_rate": 9.587420221699698e-05,
      "loss": 1.2393,
      "step": 5120
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8723963499069214,
      "learning_rate": 9.586580450117568e-05,
      "loss": 1.2485,
      "step": 5130
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9381632208824158,
      "learning_rate": 9.585740678535438e-05,
      "loss": 1.2426,
      "step": 5140
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8653701543807983,
      "learning_rate": 9.584900906953309e-05,
      "loss": 1.2514,
      "step": 5150
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9231656789779663,
      "learning_rate": 9.58406113537118e-05,
      "loss": 1.2246,
      "step": 5160
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9562845230102539,
      "learning_rate": 9.58322136378905e-05,
      "loss": 1.2323,
      "step": 5170
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9132975935935974,
      "learning_rate": 9.58238159220692e-05,
      "loss": 1.2111,
      "step": 5180
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9352735280990601,
      "learning_rate": 9.581541820624791e-05,
      "loss": 1.2388,
      "step": 5190
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9456208348274231,
      "learning_rate": 9.580702049042662e-05,
      "loss": 1.2243,
      "step": 5200
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9143781065940857,
      "learning_rate": 9.57986227746053e-05,
      "loss": 1.2381,
      "step": 5210
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0085742473602295,
      "learning_rate": 9.579022505878402e-05,
      "loss": 1.2352,
      "step": 5220
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8309242725372314,
      "learning_rate": 9.578182734296272e-05,
      "loss": 1.2511,
      "step": 5230
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9421424269676208,
      "learning_rate": 9.577342962714142e-05,
      "loss": 1.247,
      "step": 5240
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8936107158660889,
      "learning_rate": 9.576503191132012e-05,
      "loss": 1.23,
      "step": 5250
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8589064478874207,
      "learning_rate": 9.575663419549883e-05,
      "loss": 1.2556,
      "step": 5260
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9755228757858276,
      "learning_rate": 9.574823647967753e-05,
      "loss": 1.2399,
      "step": 5270
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9112702012062073,
      "learning_rate": 9.573983876385623e-05,
      "loss": 1.2396,
      "step": 5280
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.964478075504303,
      "learning_rate": 9.573144104803495e-05,
      "loss": 1.2749,
      "step": 5290
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8301079869270325,
      "learning_rate": 9.572304333221365e-05,
      "loss": 1.2346,
      "step": 5300
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8552051782608032,
      "learning_rate": 9.571464561639235e-05,
      "loss": 1.2357,
      "step": 5310
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9006170630455017,
      "learning_rate": 9.570624790057105e-05,
      "loss": 1.2424,
      "step": 5320
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8143618702888489,
      "learning_rate": 9.569785018474975e-05,
      "loss": 1.2378,
      "step": 5330
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8069211840629578,
      "learning_rate": 9.568945246892845e-05,
      "loss": 1.228,
      "step": 5340
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9581374526023865,
      "learning_rate": 9.568105475310715e-05,
      "loss": 1.2525,
      "step": 5350
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.02692711353302,
      "learning_rate": 9.567265703728587e-05,
      "loss": 1.2352,
      "step": 5360
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8993972539901733,
      "learning_rate": 9.566425932146457e-05,
      "loss": 1.2289,
      "step": 5370
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8615926504135132,
      "learning_rate": 9.565586160564327e-05,
      "loss": 1.2459,
      "step": 5380
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8434697985649109,
      "learning_rate": 9.564746388982197e-05,
      "loss": 1.2479,
      "step": 5390
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.866491973400116,
      "learning_rate": 9.563906617400068e-05,
      "loss": 1.2404,
      "step": 5400
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8595972061157227,
      "learning_rate": 9.563066845817938e-05,
      "loss": 1.2231,
      "step": 5410
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.918602466583252,
      "learning_rate": 9.562227074235807e-05,
      "loss": 1.2661,
      "step": 5420
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9388463497161865,
      "learning_rate": 9.561387302653679e-05,
      "loss": 1.2321,
      "step": 5430
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9377898573875427,
      "learning_rate": 9.560547531071549e-05,
      "loss": 1.2244,
      "step": 5440
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8881635069847107,
      "learning_rate": 9.559707759489419e-05,
      "loss": 1.2646,
      "step": 5450
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8622443079948425,
      "learning_rate": 9.55886798790729e-05,
      "loss": 1.2117,
      "step": 5460
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9005386829376221,
      "learning_rate": 9.55802821632516e-05,
      "loss": 1.2229,
      "step": 5470
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9114660024642944,
      "learning_rate": 9.55718844474303e-05,
      "loss": 1.2205,
      "step": 5480
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8409573435783386,
      "learning_rate": 9.5563486731609e-05,
      "loss": 1.2194,
      "step": 5490
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9288071990013123,
      "learning_rate": 9.555508901578772e-05,
      "loss": 1.2435,
      "step": 5500
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.891653835773468,
      "learning_rate": 9.554669129996642e-05,
      "loss": 1.2262,
      "step": 5510
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.875237762928009,
      "learning_rate": 9.553829358414512e-05,
      "loss": 1.2303,
      "step": 5520
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.883129894733429,
      "learning_rate": 9.552989586832382e-05,
      "loss": 1.2264,
      "step": 5530
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8907957673072815,
      "learning_rate": 9.552149815250252e-05,
      "loss": 1.2255,
      "step": 5540
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8646265864372253,
      "learning_rate": 9.551310043668122e-05,
      "loss": 1.242,
      "step": 5550
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8700964450836182,
      "learning_rate": 9.550470272085994e-05,
      "loss": 1.2173,
      "step": 5560
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9184081554412842,
      "learning_rate": 9.549630500503864e-05,
      "loss": 1.2357,
      "step": 5570
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8421388268470764,
      "learning_rate": 9.548790728921734e-05,
      "loss": 1.2173,
      "step": 5580
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8952757120132446,
      "learning_rate": 9.547950957339604e-05,
      "loss": 1.2104,
      "step": 5590
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9406217336654663,
      "learning_rate": 9.547111185757475e-05,
      "loss": 1.2228,
      "step": 5600
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8921267986297607,
      "learning_rate": 9.546271414175345e-05,
      "loss": 1.2694,
      "step": 5610
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9374231696128845,
      "learning_rate": 9.545431642593215e-05,
      "loss": 1.2112,
      "step": 5620
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9043653011322021,
      "learning_rate": 9.544591871011085e-05,
      "loss": 1.2187,
      "step": 5630
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9351363182067871,
      "learning_rate": 9.543752099428955e-05,
      "loss": 1.243,
      "step": 5640
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8760493993759155,
      "learning_rate": 9.542912327846825e-05,
      "loss": 1.2347,
      "step": 5650
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8686032891273499,
      "learning_rate": 9.542072556264696e-05,
      "loss": 1.2262,
      "step": 5660
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.897517204284668,
      "learning_rate": 9.541232784682567e-05,
      "loss": 1.2251,
      "step": 5670
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9855635762214661,
      "learning_rate": 9.540393013100437e-05,
      "loss": 1.2458,
      "step": 5680
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8823346495628357,
      "learning_rate": 9.539553241518307e-05,
      "loss": 1.2412,
      "step": 5690
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9292605519294739,
      "learning_rate": 9.538713469936179e-05,
      "loss": 1.2393,
      "step": 5700
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8954950571060181,
      "learning_rate": 9.537873698354049e-05,
      "loss": 1.212,
      "step": 5710
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8325024843215942,
      "learning_rate": 9.537033926771919e-05,
      "loss": 1.2385,
      "step": 5720
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3778843879699707,
      "learning_rate": 9.536194155189789e-05,
      "loss": 1.2217,
      "step": 5730
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.133502721786499,
      "learning_rate": 9.535354383607659e-05,
      "loss": 1.2193,
      "step": 5740
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8802908062934875,
      "learning_rate": 9.534514612025529e-05,
      "loss": 1.2356,
      "step": 5750
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9083417057991028,
      "learning_rate": 9.533674840443399e-05,
      "loss": 1.2235,
      "step": 5760
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0076510906219482,
      "learning_rate": 9.53283506886127e-05,
      "loss": 1.2317,
      "step": 5770
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8874029517173767,
      "learning_rate": 9.53199529727914e-05,
      "loss": 1.2335,
      "step": 5780
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.912630558013916,
      "learning_rate": 9.53115552569701e-05,
      "loss": 1.226,
      "step": 5790
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9173579216003418,
      "learning_rate": 9.530315754114882e-05,
      "loss": 1.2371,
      "step": 5800
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9296732544898987,
      "learning_rate": 9.529475982532752e-05,
      "loss": 1.2151,
      "step": 5810
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.885964572429657,
      "learning_rate": 9.528636210950622e-05,
      "loss": 1.2064,
      "step": 5820
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.880092203617096,
      "learning_rate": 9.527796439368492e-05,
      "loss": 1.1906,
      "step": 5830
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9691383242607117,
      "learning_rate": 9.526956667786364e-05,
      "loss": 1.2388,
      "step": 5840
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8995313048362732,
      "learning_rate": 9.526116896204232e-05,
      "loss": 1.2216,
      "step": 5850
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9122263193130493,
      "learning_rate": 9.525277124622102e-05,
      "loss": 1.209,
      "step": 5860
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9010337591171265,
      "learning_rate": 9.524437353039974e-05,
      "loss": 1.2196,
      "step": 5870
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.842031717300415,
      "learning_rate": 9.523597581457844e-05,
      "loss": 1.2319,
      "step": 5880
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9067339897155762,
      "learning_rate": 9.522757809875714e-05,
      "loss": 1.2401,
      "step": 5890
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8489568829536438,
      "learning_rate": 9.521918038293584e-05,
      "loss": 1.2698,
      "step": 5900
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9444595575332642,
      "learning_rate": 9.521078266711455e-05,
      "loss": 1.2197,
      "step": 5910
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8654924631118774,
      "learning_rate": 9.520238495129325e-05,
      "loss": 1.2156,
      "step": 5920
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9583361744880676,
      "learning_rate": 9.519398723547196e-05,
      "loss": 1.2075,
      "step": 5930
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9642980694770813,
      "learning_rate": 9.518558951965067e-05,
      "loss": 1.2226,
      "step": 5940
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9107359647750854,
      "learning_rate": 9.517719180382936e-05,
      "loss": 1.2078,
      "step": 5950
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8962581157684326,
      "learning_rate": 9.516879408800806e-05,
      "loss": 1.2236,
      "step": 5960
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8890233635902405,
      "learning_rate": 9.516039637218677e-05,
      "loss": 1.2338,
      "step": 5970
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8808488249778748,
      "learning_rate": 9.515199865636547e-05,
      "loss": 1.2019,
      "step": 5980
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.849258303642273,
      "learning_rate": 9.514360094054417e-05,
      "loss": 1.2294,
      "step": 5990
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9754767417907715,
      "learning_rate": 9.513520322472287e-05,
      "loss": 1.2507,
      "step": 6000
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9048311114311218,
      "learning_rate": 9.512680550890159e-05,
      "loss": 1.2346,
      "step": 6010
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8853904008865356,
      "learning_rate": 9.511840779308029e-05,
      "loss": 1.2564,
      "step": 6020
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9003530740737915,
      "learning_rate": 9.511001007725899e-05,
      "loss": 1.2243,
      "step": 6030
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9003586769104004,
      "learning_rate": 9.51016123614377e-05,
      "loss": 1.2255,
      "step": 6040
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9166625142097473,
      "learning_rate": 9.50932146456164e-05,
      "loss": 1.1941,
      "step": 6050
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0153764486312866,
      "learning_rate": 9.508481692979509e-05,
      "loss": 1.2123,
      "step": 6060
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9264450669288635,
      "learning_rate": 9.50764192139738e-05,
      "loss": 1.2143,
      "step": 6070
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9054614901542664,
      "learning_rate": 9.50680214981525e-05,
      "loss": 1.2263,
      "step": 6080
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8641769289970398,
      "learning_rate": 9.505962378233121e-05,
      "loss": 1.248,
      "step": 6090
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8716381788253784,
      "learning_rate": 9.505206583809204e-05,
      "loss": 1.2096,
      "step": 6100
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9181687235832214,
      "learning_rate": 9.504366812227074e-05,
      "loss": 1.2317,
      "step": 6110
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9484494924545288,
      "learning_rate": 9.503527040644946e-05,
      "loss": 1.2132,
      "step": 6120
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9860294461250305,
      "learning_rate": 9.502687269062816e-05,
      "loss": 1.2321,
      "step": 6130
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9444969892501831,
      "learning_rate": 9.501847497480686e-05,
      "loss": 1.2285,
      "step": 6140
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9693177938461304,
      "learning_rate": 9.501007725898556e-05,
      "loss": 1.2148,
      "step": 6150
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2146936655044556,
      "learning_rate": 9.500167954316427e-05,
      "loss": 1.2278,
      "step": 6160
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9109039902687073,
      "learning_rate": 9.499328182734297e-05,
      "loss": 1.2481,
      "step": 6170
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9141036868095398,
      "learning_rate": 9.498488411152166e-05,
      "loss": 1.1948,
      "step": 6180
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9329163432121277,
      "learning_rate": 9.497648639570038e-05,
      "loss": 1.202,
      "step": 6190
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9124045968055725,
      "learning_rate": 9.496808867987908e-05,
      "loss": 1.2195,
      "step": 6200
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8825882077217102,
      "learning_rate": 9.495969096405778e-05,
      "loss": 1.2147,
      "step": 6210
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9746090173721313,
      "learning_rate": 9.495129324823648e-05,
      "loss": 1.213,
      "step": 6220
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.850538432598114,
      "learning_rate": 9.494289553241519e-05,
      "loss": 1.2326,
      "step": 6230
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9482828974723816,
      "learning_rate": 9.493449781659389e-05,
      "loss": 1.2033,
      "step": 6240
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9356690645217896,
      "learning_rate": 9.492610010077259e-05,
      "loss": 1.224,
      "step": 6250
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.794873833656311,
      "learning_rate": 9.491770238495131e-05,
      "loss": 1.2311,
      "step": 6260
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9129030704498291,
      "learning_rate": 9.490930466913001e-05,
      "loss": 1.2166,
      "step": 6270
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8686519861221313,
      "learning_rate": 9.49009069533087e-05,
      "loss": 1.2195,
      "step": 6280
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9123719334602356,
      "learning_rate": 9.489250923748741e-05,
      "loss": 1.2023,
      "step": 6290
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3565350770950317,
      "learning_rate": 9.488411152166611e-05,
      "loss": 1.2282,
      "step": 6300
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8791282773017883,
      "learning_rate": 9.487571380584481e-05,
      "loss": 1.2231,
      "step": 6310
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8691518902778625,
      "learning_rate": 9.486731609002351e-05,
      "loss": 1.221,
      "step": 6320
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9510216116905212,
      "learning_rate": 9.485891837420223e-05,
      "loss": 1.2005,
      "step": 6330
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8951911926269531,
      "learning_rate": 9.485052065838093e-05,
      "loss": 1.1917,
      "step": 6340
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9834571480751038,
      "learning_rate": 9.484212294255963e-05,
      "loss": 1.203,
      "step": 6350
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8950636386871338,
      "learning_rate": 9.483372522673834e-05,
      "loss": 1.202,
      "step": 6360
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9080728888511658,
      "learning_rate": 9.482532751091704e-05,
      "loss": 1.2099,
      "step": 6370
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8755112290382385,
      "learning_rate": 9.481692979509574e-05,
      "loss": 1.2189,
      "step": 6380
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8602133989334106,
      "learning_rate": 9.480853207927444e-05,
      "loss": 1.2055,
      "step": 6390
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0315263271331787,
      "learning_rate": 9.480013436345314e-05,
      "loss": 1.2011,
      "step": 6400
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.82036954164505,
      "learning_rate": 9.479173664763184e-05,
      "loss": 1.1974,
      "step": 6410
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.011440396308899,
      "learning_rate": 9.478333893181055e-05,
      "loss": 1.2138,
      "step": 6420
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9471117258071899,
      "learning_rate": 9.477494121598926e-05,
      "loss": 1.2089,
      "step": 6430
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8859214186668396,
      "learning_rate": 9.476654350016796e-05,
      "loss": 1.2032,
      "step": 6440
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9290270209312439,
      "learning_rate": 9.475814578434666e-05,
      "loss": 1.22,
      "step": 6450
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8193401098251343,
      "learning_rate": 9.474974806852536e-05,
      "loss": 1.2009,
      "step": 6460
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.91757732629776,
      "learning_rate": 9.474135035270408e-05,
      "loss": 1.2186,
      "step": 6470
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.834231436252594,
      "learning_rate": 9.473295263688278e-05,
      "loss": 1.2395,
      "step": 6480
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.912388801574707,
      "learning_rate": 9.472455492106148e-05,
      "loss": 1.211,
      "step": 6490
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8885015249252319,
      "learning_rate": 9.471615720524018e-05,
      "loss": 1.214,
      "step": 6500
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9009397625923157,
      "learning_rate": 9.470775948941888e-05,
      "loss": 1.2304,
      "step": 6510
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.906132161617279,
      "learning_rate": 9.469936177359758e-05,
      "loss": 1.2052,
      "step": 6520
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8399680256843567,
      "learning_rate": 9.46909640577763e-05,
      "loss": 1.205,
      "step": 6530
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8931518197059631,
      "learning_rate": 9.4682566341955e-05,
      "loss": 1.22,
      "step": 6540
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8453435897827148,
      "learning_rate": 9.46741686261337e-05,
      "loss": 1.2252,
      "step": 6550
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.885593831539154,
      "learning_rate": 9.46657709103124e-05,
      "loss": 1.2406,
      "step": 6560
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9239007830619812,
      "learning_rate": 9.465737319449111e-05,
      "loss": 1.2295,
      "step": 6570
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9348042607307434,
      "learning_rate": 9.464897547866981e-05,
      "loss": 1.1996,
      "step": 6580
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8948506116867065,
      "learning_rate": 9.464057776284851e-05,
      "loss": 1.193,
      "step": 6590
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8650892376899719,
      "learning_rate": 9.463218004702721e-05,
      "loss": 1.2212,
      "step": 6600
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8444058895111084,
      "learning_rate": 9.462378233120591e-05,
      "loss": 1.2107,
      "step": 6610
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9454925060272217,
      "learning_rate": 9.461538461538461e-05,
      "loss": 1.2147,
      "step": 6620
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9183284044265747,
      "learning_rate": 9.460698689956333e-05,
      "loss": 1.2043,
      "step": 6630
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.900619626045227,
      "learning_rate": 9.459858918374203e-05,
      "loss": 1.2367,
      "step": 6640
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8767586946487427,
      "learning_rate": 9.459019146792073e-05,
      "loss": 1.1921,
      "step": 6650
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9866166710853577,
      "learning_rate": 9.458179375209943e-05,
      "loss": 1.2297,
      "step": 6660
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9303044676780701,
      "learning_rate": 9.457339603627814e-05,
      "loss": 1.2016,
      "step": 6670
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9829623699188232,
      "learning_rate": 9.456499832045684e-05,
      "loss": 1.2117,
      "step": 6680
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8491721153259277,
      "learning_rate": 9.455660060463555e-05,
      "loss": 1.2002,
      "step": 6690
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9353068470954895,
      "learning_rate": 9.454820288881425e-05,
      "loss": 1.1971,
      "step": 6700
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8836992383003235,
      "learning_rate": 9.453980517299295e-05,
      "loss": 1.2072,
      "step": 6710
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9184627532958984,
      "learning_rate": 9.453140745717165e-05,
      "loss": 1.2088,
      "step": 6720
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9549211859703064,
      "learning_rate": 9.452300974135035e-05,
      "loss": 1.2093,
      "step": 6730
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8630688190460205,
      "learning_rate": 9.451461202552906e-05,
      "loss": 1.2306,
      "step": 6740
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8126599192619324,
      "learning_rate": 9.450621430970776e-05,
      "loss": 1.2013,
      "step": 6750
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.930507242679596,
      "learning_rate": 9.449781659388646e-05,
      "loss": 1.1965,
      "step": 6760
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9245671033859253,
      "learning_rate": 9.448941887806518e-05,
      "loss": 1.2054,
      "step": 6770
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9427335858345032,
      "learning_rate": 9.448102116224388e-05,
      "loss": 1.1885,
      "step": 6780
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8826670050621033,
      "learning_rate": 9.447262344642258e-05,
      "loss": 1.196,
      "step": 6790
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8583161234855652,
      "learning_rate": 9.446422573060128e-05,
      "loss": 1.2018,
      "step": 6800
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8399899005889893,
      "learning_rate": 9.445582801477998e-05,
      "loss": 1.1759,
      "step": 6810
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9219769835472107,
      "learning_rate": 9.444743029895868e-05,
      "loss": 1.1996,
      "step": 6820
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8673033118247986,
      "learning_rate": 9.443903258313738e-05,
      "loss": 1.2203,
      "step": 6830
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8609717488288879,
      "learning_rate": 9.44306348673161e-05,
      "loss": 1.2273,
      "step": 6840
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8942599892616272,
      "learning_rate": 9.44222371514948e-05,
      "loss": 1.2236,
      "step": 6850
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8240846395492554,
      "learning_rate": 9.44138394356735e-05,
      "loss": 1.215,
      "step": 6860
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8023459911346436,
      "learning_rate": 9.440544171985221e-05,
      "loss": 1.2105,
      "step": 6870
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9448633193969727,
      "learning_rate": 9.439704400403091e-05,
      "loss": 1.1919,
      "step": 6880
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8635557293891907,
      "learning_rate": 9.438864628820961e-05,
      "loss": 1.1764,
      "step": 6890
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8735735416412354,
      "learning_rate": 9.438024857238831e-05,
      "loss": 1.225,
      "step": 6900
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8482799530029297,
      "learning_rate": 9.437185085656703e-05,
      "loss": 1.2107,
      "step": 6910
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9306829571723938,
      "learning_rate": 9.436345314074572e-05,
      "loss": 1.2181,
      "step": 6920
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.88686603307724,
      "learning_rate": 9.435505542492442e-05,
      "loss": 1.1963,
      "step": 6930
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8896923065185547,
      "learning_rate": 9.434665770910313e-05,
      "loss": 1.1999,
      "step": 6940
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7939994931221008,
      "learning_rate": 9.433825999328183e-05,
      "loss": 1.1835,
      "step": 6950
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9401127099990845,
      "learning_rate": 9.432986227746053e-05,
      "loss": 1.1855,
      "step": 6960
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9034574031829834,
      "learning_rate": 9.432146456163923e-05,
      "loss": 1.1748,
      "step": 6970
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8599429726600647,
      "learning_rate": 9.431306684581795e-05,
      "loss": 1.1934,
      "step": 6980
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9708506464958191,
      "learning_rate": 9.430466912999665e-05,
      "loss": 1.1799,
      "step": 6990
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9697883129119873,
      "learning_rate": 9.429627141417535e-05,
      "loss": 1.2021,
      "step": 7000
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8883392214775085,
      "learning_rate": 9.428787369835406e-05,
      "loss": 1.1806,
      "step": 7010
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9711785912513733,
      "learning_rate": 9.427947598253276e-05,
      "loss": 1.2261,
      "step": 7020
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9639722108840942,
      "learning_rate": 9.427107826671145e-05,
      "loss": 1.193,
      "step": 7030
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9306802749633789,
      "learning_rate": 9.426268055089016e-05,
      "loss": 1.2203,
      "step": 7040
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8337755799293518,
      "learning_rate": 9.425428283506886e-05,
      "loss": 1.2426,
      "step": 7050
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.83501797914505,
      "learning_rate": 9.424588511924757e-05,
      "loss": 1.2126,
      "step": 7060
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9699921011924744,
      "learning_rate": 9.423748740342627e-05,
      "loss": 1.2122,
      "step": 7070
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9536002278327942,
      "learning_rate": 9.422908968760498e-05,
      "loss": 1.1994,
      "step": 7080
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9766404032707214,
      "learning_rate": 9.422069197178368e-05,
      "loss": 1.2346,
      "step": 7090
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8767348527908325,
      "learning_rate": 9.421229425596238e-05,
      "loss": 1.1839,
      "step": 7100
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.923218309879303,
      "learning_rate": 9.42038965401411e-05,
      "loss": 1.1978,
      "step": 7110
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9360674619674683,
      "learning_rate": 9.41954988243198e-05,
      "loss": 1.2313,
      "step": 7120
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9123576283454895,
      "learning_rate": 9.418710110849848e-05,
      "loss": 1.1958,
      "step": 7130
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8900172114372253,
      "learning_rate": 9.41787033926772e-05,
      "loss": 1.1824,
      "step": 7140
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9195378422737122,
      "learning_rate": 9.41703056768559e-05,
      "loss": 1.2176,
      "step": 7150
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8248353600502014,
      "learning_rate": 9.41619079610346e-05,
      "loss": 1.2058,
      "step": 7160
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.137145757675171,
      "learning_rate": 9.41535102452133e-05,
      "loss": 1.2013,
      "step": 7170
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9338940382003784,
      "learning_rate": 9.414511252939201e-05,
      "loss": 1.2069,
      "step": 7180
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9436649680137634,
      "learning_rate": 9.413671481357072e-05,
      "loss": 1.2138,
      "step": 7190
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.867423951625824,
      "learning_rate": 9.412831709774942e-05,
      "loss": 1.2204,
      "step": 7200
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8983123898506165,
      "learning_rate": 9.411991938192812e-05,
      "loss": 1.1771,
      "step": 7210
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9414335489273071,
      "learning_rate": 9.411152166610683e-05,
      "loss": 1.1896,
      "step": 7220
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8820247650146484,
      "learning_rate": 9.410312395028553e-05,
      "loss": 1.1988,
      "step": 7230
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9329501986503601,
      "learning_rate": 9.409472623446423e-05,
      "loss": 1.1778,
      "step": 7240
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8825410008430481,
      "learning_rate": 9.408632851864293e-05,
      "loss": 1.2103,
      "step": 7250
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8636690974235535,
      "learning_rate": 9.407793080282163e-05,
      "loss": 1.2002,
      "step": 7260
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8834033012390137,
      "learning_rate": 9.406953308700033e-05,
      "loss": 1.1957,
      "step": 7270
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9180720448493958,
      "learning_rate": 9.406113537117905e-05,
      "loss": 1.2097,
      "step": 7280
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8845052719116211,
      "learning_rate": 9.405273765535775e-05,
      "loss": 1.174,
      "step": 7290
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.934151828289032,
      "learning_rate": 9.404433993953645e-05,
      "loss": 1.2053,
      "step": 7300
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9797338843345642,
      "learning_rate": 9.403594222371515e-05,
      "loss": 1.1866,
      "step": 7310
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8769814968109131,
      "learning_rate": 9.402754450789386e-05,
      "loss": 1.1849,
      "step": 7320
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9255028963088989,
      "learning_rate": 9.401914679207257e-05,
      "loss": 1.1953,
      "step": 7330
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8942875862121582,
      "learning_rate": 9.401074907625125e-05,
      "loss": 1.191,
      "step": 7340
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9048584699630737,
      "learning_rate": 9.400235136042997e-05,
      "loss": 1.1908,
      "step": 7350
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8529855012893677,
      "learning_rate": 9.399395364460867e-05,
      "loss": 1.1891,
      "step": 7360
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8380342125892639,
      "learning_rate": 9.398555592878737e-05,
      "loss": 1.1909,
      "step": 7370
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9068658351898193,
      "learning_rate": 9.397715821296608e-05,
      "loss": 1.1929,
      "step": 7380
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9368525147438049,
      "learning_rate": 9.396876049714478e-05,
      "loss": 1.202,
      "step": 7390
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8885350227355957,
      "learning_rate": 9.396036278132348e-05,
      "loss": 1.2079,
      "step": 7400
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9168792963027954,
      "learning_rate": 9.395196506550218e-05,
      "loss": 1.1953,
      "step": 7410
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8806029558181763,
      "learning_rate": 9.39435673496809e-05,
      "loss": 1.1994,
      "step": 7420
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9570105075836182,
      "learning_rate": 9.39351696338596e-05,
      "loss": 1.1827,
      "step": 7430
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.845789909362793,
      "learning_rate": 9.39267719180383e-05,
      "loss": 1.1948,
      "step": 7440
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9345555901527405,
      "learning_rate": 9.3918374202217e-05,
      "loss": 1.207,
      "step": 7450
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.809235692024231,
      "learning_rate": 9.39099764863957e-05,
      "loss": 1.2255,
      "step": 7460
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9185549020767212,
      "learning_rate": 9.39015787705744e-05,
      "loss": 1.2116,
      "step": 7470
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9820988178253174,
      "learning_rate": 9.389318105475312e-05,
      "loss": 1.2023,
      "step": 7480
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8993212580680847,
      "learning_rate": 9.388478333893182e-05,
      "loss": 1.1936,
      "step": 7490
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8562648892402649,
      "learning_rate": 9.387638562311052e-05,
      "loss": 1.1738,
      "step": 7500
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9744245409965515,
      "learning_rate": 9.386798790728922e-05,
      "loss": 1.1874,
      "step": 7510
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8549754619598389,
      "learning_rate": 9.385959019146793e-05,
      "loss": 1.2091,
      "step": 7520
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9932068586349487,
      "learning_rate": 9.385119247564663e-05,
      "loss": 1.1886,
      "step": 7530
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9793296456336975,
      "learning_rate": 9.384279475982533e-05,
      "loss": 1.2022,
      "step": 7540
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9506256580352783,
      "learning_rate": 9.383439704400403e-05,
      "loss": 1.1937,
      "step": 7550
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8991612792015076,
      "learning_rate": 9.382599932818274e-05,
      "loss": 1.1652,
      "step": 7560
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9225620627403259,
      "learning_rate": 9.381760161236144e-05,
      "loss": 1.1848,
      "step": 7570
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8967307209968567,
      "learning_rate": 9.380920389654014e-05,
      "loss": 1.2174,
      "step": 7580
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.936236560344696,
      "learning_rate": 9.380080618071885e-05,
      "loss": 1.1914,
      "step": 7590
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8682725429534912,
      "learning_rate": 9.379240846489755e-05,
      "loss": 1.2043,
      "step": 7600
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9163506031036377,
      "learning_rate": 9.378401074907625e-05,
      "loss": 1.1937,
      "step": 7610
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9057551622390747,
      "learning_rate": 9.377561303325497e-05,
      "loss": 1.2134,
      "step": 7620
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8398305773735046,
      "learning_rate": 9.376721531743367e-05,
      "loss": 1.1748,
      "step": 7630
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9422807097434998,
      "learning_rate": 9.375881760161237e-05,
      "loss": 1.1899,
      "step": 7640
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9004360437393188,
      "learning_rate": 9.375041988579107e-05,
      "loss": 1.171,
      "step": 7650
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0320866107940674,
      "learning_rate": 9.374202216996977e-05,
      "loss": 1.1797,
      "step": 7660
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8690953850746155,
      "learning_rate": 9.373362445414847e-05,
      "loss": 1.1685,
      "step": 7670
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.808520495891571,
      "learning_rate": 9.372522673832717e-05,
      "loss": 1.1722,
      "step": 7680
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9039348363876343,
      "learning_rate": 9.371682902250589e-05,
      "loss": 1.2031,
      "step": 7690
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9256361126899719,
      "learning_rate": 9.370843130668459e-05,
      "loss": 1.1916,
      "step": 7700
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9331594705581665,
      "learning_rate": 9.370003359086329e-05,
      "loss": 1.1875,
      "step": 7710
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0034734010696411,
      "learning_rate": 9.3691635875042e-05,
      "loss": 1.1893,
      "step": 7720
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9952471852302551,
      "learning_rate": 9.36832381592207e-05,
      "loss": 1.1856,
      "step": 7730
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8371817469596863,
      "learning_rate": 9.36748404433994e-05,
      "loss": 1.2,
      "step": 7740
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8598681092262268,
      "learning_rate": 9.36664427275781e-05,
      "loss": 1.201,
      "step": 7750
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8538065552711487,
      "learning_rate": 9.365804501175682e-05,
      "loss": 1.1881,
      "step": 7760
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8651682138442993,
      "learning_rate": 9.36496472959355e-05,
      "loss": 1.1929,
      "step": 7770
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9020833969116211,
      "learning_rate": 9.36412495801142e-05,
      "loss": 1.1998,
      "step": 7780
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8552895188331604,
      "learning_rate": 9.363285186429292e-05,
      "loss": 1.188,
      "step": 7790
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.929107666015625,
      "learning_rate": 9.362445414847162e-05,
      "loss": 1.2101,
      "step": 7800
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8756254315376282,
      "learning_rate": 9.361605643265032e-05,
      "loss": 1.1829,
      "step": 7810
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8464421033859253,
      "learning_rate": 9.360765871682902e-05,
      "loss": 1.1875,
      "step": 7820
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9108934998512268,
      "learning_rate": 9.359926100100774e-05,
      "loss": 1.193,
      "step": 7830
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8196065425872803,
      "learning_rate": 9.359086328518644e-05,
      "loss": 1.1815,
      "step": 7840
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9362756013870239,
      "learning_rate": 9.358246556936514e-05,
      "loss": 1.19,
      "step": 7850
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0048913955688477,
      "learning_rate": 9.357406785354385e-05,
      "loss": 1.1869,
      "step": 7860
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8194365501403809,
      "learning_rate": 9.356567013772254e-05,
      "loss": 1.1872,
      "step": 7870
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9167453050613403,
      "learning_rate": 9.355727242190124e-05,
      "loss": 1.2053,
      "step": 7880
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0007411241531372,
      "learning_rate": 9.354887470607995e-05,
      "loss": 1.2083,
      "step": 7890
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9240213632583618,
      "learning_rate": 9.354047699025865e-05,
      "loss": 1.1961,
      "step": 7900
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8382523059844971,
      "learning_rate": 9.353207927443735e-05,
      "loss": 1.2134,
      "step": 7910
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9030473232269287,
      "learning_rate": 9.352368155861606e-05,
      "loss": 1.1787,
      "step": 7920
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9283421039581299,
      "learning_rate": 9.351528384279477e-05,
      "loss": 1.1609,
      "step": 7930
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9871938228607178,
      "learning_rate": 9.350688612697347e-05,
      "loss": 1.2078,
      "step": 7940
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8591379523277283,
      "learning_rate": 9.349848841115217e-05,
      "loss": 1.1909,
      "step": 7950
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8470222353935242,
      "learning_rate": 9.349009069533089e-05,
      "loss": 1.2113,
      "step": 7960
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8578379154205322,
      "learning_rate": 9.348169297950959e-05,
      "loss": 1.1959,
      "step": 7970
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8939791917800903,
      "learning_rate": 9.347329526368827e-05,
      "loss": 1.1721,
      "step": 7980
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8363282084465027,
      "learning_rate": 9.346489754786699e-05,
      "loss": 1.1505,
      "step": 7990
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9319391846656799,
      "learning_rate": 9.345649983204569e-05,
      "loss": 1.183,
      "step": 8000
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8514727354049683,
      "learning_rate": 9.344810211622439e-05,
      "loss": 1.1785,
      "step": 8010
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8914164304733276,
      "learning_rate": 9.343970440040309e-05,
      "loss": 1.1893,
      "step": 8020
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9012137055397034,
      "learning_rate": 9.34313066845818e-05,
      "loss": 1.1586,
      "step": 8030
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.892142653465271,
      "learning_rate": 9.34229089687605e-05,
      "loss": 1.1858,
      "step": 8040
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.886681079864502,
      "learning_rate": 9.34145112529392e-05,
      "loss": 1.1945,
      "step": 8050
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8880460262298584,
      "learning_rate": 9.34061135371179e-05,
      "loss": 1.186,
      "step": 8060
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9022466540336609,
      "learning_rate": 9.339771582129662e-05,
      "loss": 1.1829,
      "step": 8070
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9207655787467957,
      "learning_rate": 9.338931810547531e-05,
      "loss": 1.1982,
      "step": 8080
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8897063136100769,
      "learning_rate": 9.338092038965401e-05,
      "loss": 1.1957,
      "step": 8090
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8721984624862671,
      "learning_rate": 9.337252267383272e-05,
      "loss": 1.1866,
      "step": 8100
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9680582284927368,
      "learning_rate": 9.336412495801142e-05,
      "loss": 1.1982,
      "step": 8110
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8685674071311951,
      "learning_rate": 9.335572724219012e-05,
      "loss": 1.1627,
      "step": 8120
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8975292444229126,
      "learning_rate": 9.334732952636884e-05,
      "loss": 1.1856,
      "step": 8130
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8757001161575317,
      "learning_rate": 9.333893181054754e-05,
      "loss": 1.1939,
      "step": 8140
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9586355090141296,
      "learning_rate": 9.333053409472624e-05,
      "loss": 1.1728,
      "step": 8150
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8423217535018921,
      "learning_rate": 9.332213637890494e-05,
      "loss": 1.1613,
      "step": 8160
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.884924054145813,
      "learning_rate": 9.331373866308365e-05,
      "loss": 1.1895,
      "step": 8170
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.856117844581604,
      "learning_rate": 9.330534094726235e-05,
      "loss": 1.2027,
      "step": 8180
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8681750893592834,
      "learning_rate": 9.329694323144104e-05,
      "loss": 1.1808,
      "step": 8190
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9215348362922668,
      "learning_rate": 9.328854551561976e-05,
      "loss": 1.1862,
      "step": 8200
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9275267720222473,
      "learning_rate": 9.328014779979846e-05,
      "loss": 1.1984,
      "step": 8210
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9127364158630371,
      "learning_rate": 9.327175008397716e-05,
      "loss": 1.172,
      "step": 8220
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9053615927696228,
      "learning_rate": 9.326335236815587e-05,
      "loss": 1.1817,
      "step": 8230
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9082298278808594,
      "learning_rate": 9.325495465233457e-05,
      "loss": 1.1783,
      "step": 8240
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.844623327255249,
      "learning_rate": 9.324655693651327e-05,
      "loss": 1.2072,
      "step": 8250
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8887671828269958,
      "learning_rate": 9.323815922069197e-05,
      "loss": 1.1798,
      "step": 8260
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9145581722259521,
      "learning_rate": 9.322976150487069e-05,
      "loss": 1.1876,
      "step": 8270
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8859278559684753,
      "learning_rate": 9.322136378904939e-05,
      "loss": 1.1902,
      "step": 8280
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8818907737731934,
      "learning_rate": 9.321296607322809e-05,
      "loss": 1.1694,
      "step": 8290
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0000308752059937,
      "learning_rate": 9.320456835740679e-05,
      "loss": 1.1692,
      "step": 8300
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8591494560241699,
      "learning_rate": 9.319617064158549e-05,
      "loss": 1.1747,
      "step": 8310
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8520761728286743,
      "learning_rate": 9.318777292576419e-05,
      "loss": 1.2012,
      "step": 8320
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9467620253562927,
      "learning_rate": 9.317937520994289e-05,
      "loss": 1.1861,
      "step": 8330
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8405246734619141,
      "learning_rate": 9.31709774941216e-05,
      "loss": 1.1778,
      "step": 8340
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9633544683456421,
      "learning_rate": 9.316257977830031e-05,
      "loss": 1.1955,
      "step": 8350
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0017706155776978,
      "learning_rate": 9.315418206247901e-05,
      "loss": 1.1693,
      "step": 8360
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8827065825462341,
      "learning_rate": 9.314578434665772e-05,
      "loss": 1.1906,
      "step": 8370
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.939825177192688,
      "learning_rate": 9.313738663083642e-05,
      "loss": 1.1933,
      "step": 8380
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9015594720840454,
      "learning_rate": 9.312898891501512e-05,
      "loss": 1.1589,
      "step": 8390
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8750881552696228,
      "learning_rate": 9.312059119919382e-05,
      "loss": 1.1882,
      "step": 8400
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8922222256660461,
      "learning_rate": 9.311219348337252e-05,
      "loss": 1.1742,
      "step": 8410
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8243392705917358,
      "learning_rate": 9.310379576755123e-05,
      "loss": 1.1882,
      "step": 8420
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9192656874656677,
      "learning_rate": 9.309539805172993e-05,
      "loss": 1.2012,
      "step": 8430
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8955969214439392,
      "learning_rate": 9.308700033590864e-05,
      "loss": 1.1702,
      "step": 8440
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9001456499099731,
      "learning_rate": 9.307860262008734e-05,
      "loss": 1.164,
      "step": 8450
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9576339721679688,
      "learning_rate": 9.307020490426604e-05,
      "loss": 1.1681,
      "step": 8460
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8319069743156433,
      "learning_rate": 9.306180718844476e-05,
      "loss": 1.1865,
      "step": 8470
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8679335713386536,
      "learning_rate": 9.305340947262346e-05,
      "loss": 1.1936,
      "step": 8480
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8643022775650024,
      "learning_rate": 9.304501175680216e-05,
      "loss": 1.1775,
      "step": 8490
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9259525537490845,
      "learning_rate": 9.303661404098086e-05,
      "loss": 1.1834,
      "step": 8500
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8675290942192078,
      "learning_rate": 9.302821632515956e-05,
      "loss": 1.2006,
      "step": 8510
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9066543579101562,
      "learning_rate": 9.301981860933826e-05,
      "loss": 1.1676,
      "step": 8520
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8691399097442627,
      "learning_rate": 9.301142089351696e-05,
      "loss": 1.1869,
      "step": 8530
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9211807250976562,
      "learning_rate": 9.300302317769567e-05,
      "loss": 1.158,
      "step": 8540
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8890571594238281,
      "learning_rate": 9.299462546187437e-05,
      "loss": 1.1683,
      "step": 8550
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9606330990791321,
      "learning_rate": 9.298622774605308e-05,
      "loss": 1.1803,
      "step": 8560
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.923599123954773,
      "learning_rate": 9.297783003023178e-05,
      "loss": 1.1786,
      "step": 8570
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8857382535934448,
      "learning_rate": 9.296943231441049e-05,
      "loss": 1.184,
      "step": 8580
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9653752446174622,
      "learning_rate": 9.296103459858919e-05,
      "loss": 1.1642,
      "step": 8590
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9239545464515686,
      "learning_rate": 9.295263688276789e-05,
      "loss": 1.1574,
      "step": 8600
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.933377742767334,
      "learning_rate": 9.294423916694659e-05,
      "loss": 1.179,
      "step": 8610
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9790053963661194,
      "learning_rate": 9.29358414511253e-05,
      "loss": 1.1891,
      "step": 8620
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9635185599327087,
      "learning_rate": 9.2927443735304e-05,
      "loss": 1.173,
      "step": 8630
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8766095638275146,
      "learning_rate": 9.291904601948271e-05,
      "loss": 1.1868,
      "step": 8640
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8605684638023376,
      "learning_rate": 9.291064830366141e-05,
      "loss": 1.1693,
      "step": 8650
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9944674968719482,
      "learning_rate": 9.290225058784011e-05,
      "loss": 1.1793,
      "step": 8660
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9609712958335876,
      "learning_rate": 9.289385287201881e-05,
      "loss": 1.1832,
      "step": 8670
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9557624459266663,
      "learning_rate": 9.288545515619752e-05,
      "loss": 1.1872,
      "step": 8680
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8827246427536011,
      "learning_rate": 9.287705744037623e-05,
      "loss": 1.1808,
      "step": 8690
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0152974128723145,
      "learning_rate": 9.286865972455493e-05,
      "loss": 1.1783,
      "step": 8700
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8559834361076355,
      "learning_rate": 9.286026200873364e-05,
      "loss": 1.2048,
      "step": 8710
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9328780174255371,
      "learning_rate": 9.285186429291233e-05,
      "loss": 1.1695,
      "step": 8720
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8733211755752563,
      "learning_rate": 9.284346657709103e-05,
      "loss": 1.1889,
      "step": 8730
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9180071353912354,
      "learning_rate": 9.283506886126974e-05,
      "loss": 1.1722,
      "step": 8740
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8515558838844299,
      "learning_rate": 9.282667114544844e-05,
      "loss": 1.1453,
      "step": 8750
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9535048604011536,
      "learning_rate": 9.281827342962714e-05,
      "loss": 1.1899,
      "step": 8760
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8584855198860168,
      "learning_rate": 9.280987571380584e-05,
      "loss": 1.1743,
      "step": 8770
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.824908435344696,
      "learning_rate": 9.280147799798456e-05,
      "loss": 1.1732,
      "step": 8780
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9606408476829529,
      "learning_rate": 9.279308028216326e-05,
      "loss": 1.1914,
      "step": 8790
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8435086011886597,
      "learning_rate": 9.278468256634196e-05,
      "loss": 1.1803,
      "step": 8800
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9127106666564941,
      "learning_rate": 9.277628485052066e-05,
      "loss": 1.1815,
      "step": 8810
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8833587169647217,
      "learning_rate": 9.276788713469937e-05,
      "loss": 1.1751,
      "step": 8820
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8695579171180725,
      "learning_rate": 9.275948941887806e-05,
      "loss": 1.1654,
      "step": 8830
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8679482340812683,
      "learning_rate": 9.275109170305676e-05,
      "loss": 1.1682,
      "step": 8840
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9294605255126953,
      "learning_rate": 9.274269398723548e-05,
      "loss": 1.1913,
      "step": 8850
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8430615663528442,
      "learning_rate": 9.273429627141418e-05,
      "loss": 1.1837,
      "step": 8860
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8895713090896606,
      "learning_rate": 9.272589855559288e-05,
      "loss": 1.169,
      "step": 8870
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9212405681610107,
      "learning_rate": 9.271750083977159e-05,
      "loss": 1.1847,
      "step": 8880
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.814430296421051,
      "learning_rate": 9.27091031239503e-05,
      "loss": 1.1888,
      "step": 8890
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8758382797241211,
      "learning_rate": 9.2700705408129e-05,
      "loss": 1.1532,
      "step": 8900
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9425457119941711,
      "learning_rate": 9.26923076923077e-05,
      "loss": 1.1749,
      "step": 8910
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9039488434791565,
      "learning_rate": 9.268390997648641e-05,
      "loss": 1.1956,
      "step": 8920
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8784694075584412,
      "learning_rate": 9.26755122606651e-05,
      "loss": 1.1836,
      "step": 8930
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9849941730499268,
      "learning_rate": 9.26671145448438e-05,
      "loss": 1.1751,
      "step": 8940
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9492660164833069,
      "learning_rate": 9.265871682902251e-05,
      "loss": 1.1705,
      "step": 8950
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9221577048301697,
      "learning_rate": 9.265031911320121e-05,
      "loss": 1.1637,
      "step": 8960
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.910025954246521,
      "learning_rate": 9.264192139737991e-05,
      "loss": 1.2021,
      "step": 8970
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.950025737285614,
      "learning_rate": 9.263352368155863e-05,
      "loss": 1.153,
      "step": 8980
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8990755081176758,
      "learning_rate": 9.262512596573733e-05,
      "loss": 1.1666,
      "step": 8990
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.961871862411499,
      "learning_rate": 9.261672824991603e-05,
      "loss": 1.1778,
      "step": 9000
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9348917603492737,
      "learning_rate": 9.260833053409473e-05,
      "loss": 1.1706,
      "step": 9010
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9270437359809875,
      "learning_rate": 9.259993281827344e-05,
      "loss": 1.1743,
      "step": 9020
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9181728959083557,
      "learning_rate": 9.259153510245214e-05,
      "loss": 1.2038,
      "step": 9030
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3287311792373657,
      "learning_rate": 9.258313738663083e-05,
      "loss": 1.1703,
      "step": 9040
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8790122866630554,
      "learning_rate": 9.257473967080954e-05,
      "loss": 1.1765,
      "step": 9050
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9112759232521057,
      "learning_rate": 9.256634195498825e-05,
      "loss": 1.1792,
      "step": 9060
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9480587244033813,
      "learning_rate": 9.255794423916695e-05,
      "loss": 1.1682,
      "step": 9070
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9825183749198914,
      "learning_rate": 9.254954652334566e-05,
      "loss": 1.1682,
      "step": 9080
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9880927205085754,
      "learning_rate": 9.254114880752436e-05,
      "loss": 1.1512,
      "step": 9090
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9363360404968262,
      "learning_rate": 9.253275109170306e-05,
      "loss": 1.1675,
      "step": 9100
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9263131618499756,
      "learning_rate": 9.252435337588176e-05,
      "loss": 1.1895,
      "step": 9110
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8793935179710388,
      "learning_rate": 9.251595566006048e-05,
      "loss": 1.1815,
      "step": 9120
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9015591740608215,
      "learning_rate": 9.250755794423918e-05,
      "loss": 1.2022,
      "step": 9130
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8961725831031799,
      "learning_rate": 9.249916022841786e-05,
      "loss": 1.1599,
      "step": 9140
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8976597189903259,
      "learning_rate": 9.249076251259658e-05,
      "loss": 1.1735,
      "step": 9150
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8971372246742249,
      "learning_rate": 9.248236479677528e-05,
      "loss": 1.1544,
      "step": 9160
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0064512491226196,
      "learning_rate": 9.247396708095398e-05,
      "loss": 1.1762,
      "step": 9170
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9865056872367859,
      "learning_rate": 9.246556936513268e-05,
      "loss": 1.1849,
      "step": 9180
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8811156153678894,
      "learning_rate": 9.24571716493114e-05,
      "loss": 1.1826,
      "step": 9190
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9140886664390564,
      "learning_rate": 9.24487739334901e-05,
      "loss": 1.1834,
      "step": 9200
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8452087044715881,
      "learning_rate": 9.24403762176688e-05,
      "loss": 1.1668,
      "step": 9210
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9415846467018127,
      "learning_rate": 9.243197850184751e-05,
      "loss": 1.1692,
      "step": 9220
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9672451615333557,
      "learning_rate": 9.242358078602621e-05,
      "loss": 1.1521,
      "step": 9230
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9130426049232483,
      "learning_rate": 9.241518307020491e-05,
      "loss": 1.189,
      "step": 9240
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8777864575386047,
      "learning_rate": 9.240678535438361e-05,
      "loss": 1.1851,
      "step": 9250
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9119729399681091,
      "learning_rate": 9.239838763856231e-05,
      "loss": 1.1579,
      "step": 9260
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.889671802520752,
      "learning_rate": 9.238998992274101e-05,
      "loss": 1.1705,
      "step": 9270
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9760835766792297,
      "learning_rate": 9.238159220691971e-05,
      "loss": 1.1775,
      "step": 9280
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8995108008384705,
      "learning_rate": 9.237319449109843e-05,
      "loss": 1.1662,
      "step": 9290
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.81166011095047,
      "learning_rate": 9.236479677527713e-05,
      "loss": 1.1838,
      "step": 9300
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9276900291442871,
      "learning_rate": 9.235639905945583e-05,
      "loss": 1.1678,
      "step": 9310
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8583055138587952,
      "learning_rate": 9.234800134363454e-05,
      "loss": 1.1623,
      "step": 9320
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9347761869430542,
      "learning_rate": 9.233960362781325e-05,
      "loss": 1.1803,
      "step": 9330
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9358561038970947,
      "learning_rate": 9.233120591199195e-05,
      "loss": 1.1681,
      "step": 9340
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8344528675079346,
      "learning_rate": 9.232280819617065e-05,
      "loss": 1.1495,
      "step": 9350
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8631145358085632,
      "learning_rate": 9.231441048034935e-05,
      "loss": 1.1783,
      "step": 9360
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8783475756645203,
      "learning_rate": 9.230601276452805e-05,
      "loss": 1.1714,
      "step": 9370
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9026092886924744,
      "learning_rate": 9.229761504870675e-05,
      "loss": 1.1732,
      "step": 9380
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9233419895172119,
      "learning_rate": 9.228921733288546e-05,
      "loss": 1.1703,
      "step": 9390
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8974713683128357,
      "learning_rate": 9.228081961706416e-05,
      "loss": 1.1636,
      "step": 9400
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9101700782775879,
      "learning_rate": 9.227242190124286e-05,
      "loss": 1.1935,
      "step": 9410
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9138025045394897,
      "learning_rate": 9.226402418542157e-05,
      "loss": 1.1512,
      "step": 9420
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9470272660255432,
      "learning_rate": 9.225562646960028e-05,
      "loss": 1.1786,
      "step": 9430
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9542033076286316,
      "learning_rate": 9.224722875377898e-05,
      "loss": 1.183,
      "step": 9440
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8865834474563599,
      "learning_rate": 9.223883103795768e-05,
      "loss": 1.1781,
      "step": 9450
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.901871919631958,
      "learning_rate": 9.223043332213638e-05,
      "loss": 1.1771,
      "step": 9460
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9759643077850342,
      "learning_rate": 9.222203560631508e-05,
      "loss": 1.1702,
      "step": 9470
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9939475655555725,
      "learning_rate": 9.221363789049378e-05,
      "loss": 1.1553,
      "step": 9480
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.939246416091919,
      "learning_rate": 9.22052401746725e-05,
      "loss": 1.1614,
      "step": 9490
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9197496175765991,
      "learning_rate": 9.21968424588512e-05,
      "loss": 1.1485,
      "step": 9500
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8700929880142212,
      "learning_rate": 9.21884447430299e-05,
      "loss": 1.1646,
      "step": 9510
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8767945170402527,
      "learning_rate": 9.21800470272086e-05,
      "loss": 1.1616,
      "step": 9520
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8359785079956055,
      "learning_rate": 9.217164931138731e-05,
      "loss": 1.1835,
      "step": 9530
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.865118682384491,
      "learning_rate": 9.216325159556601e-05,
      "loss": 1.1891,
      "step": 9540
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0089035034179688,
      "learning_rate": 9.215485387974471e-05,
      "loss": 1.1802,
      "step": 9550
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9262118339538574,
      "learning_rate": 9.214645616392343e-05,
      "loss": 1.1637,
      "step": 9560
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9462135434150696,
      "learning_rate": 9.213805844810212e-05,
      "loss": 1.1884,
      "step": 9570
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8781193494796753,
      "learning_rate": 9.212966073228082e-05,
      "loss": 1.1789,
      "step": 9580
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8772711753845215,
      "learning_rate": 9.212126301645953e-05,
      "loss": 1.1745,
      "step": 9590
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9201531410217285,
      "learning_rate": 9.211286530063823e-05,
      "loss": 1.1797,
      "step": 9600
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8998798131942749,
      "learning_rate": 9.210446758481693e-05,
      "loss": 1.1627,
      "step": 9610
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8889737725257874,
      "learning_rate": 9.209606986899563e-05,
      "loss": 1.1745,
      "step": 9620
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.959130585193634,
      "learning_rate": 9.208767215317435e-05,
      "loss": 1.1554,
      "step": 9630
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8671838045120239,
      "learning_rate": 9.207927443735305e-05,
      "loss": 1.1606,
      "step": 9640
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8932032585144043,
      "learning_rate": 9.207087672153175e-05,
      "loss": 1.1815,
      "step": 9650
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9338779449462891,
      "learning_rate": 9.206247900571045e-05,
      "loss": 1.1754,
      "step": 9660
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9668774604797363,
      "learning_rate": 9.205408128988915e-05,
      "loss": 1.1867,
      "step": 9670
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8983999490737915,
      "learning_rate": 9.204568357406785e-05,
      "loss": 1.1837,
      "step": 9680
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9296559691429138,
      "learning_rate": 9.203728585824655e-05,
      "loss": 1.1728,
      "step": 9690
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9125829935073853,
      "learning_rate": 9.202888814242527e-05,
      "loss": 1.201,
      "step": 9700
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9010559916496277,
      "learning_rate": 9.202049042660397e-05,
      "loss": 1.1753,
      "step": 9710
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9880275726318359,
      "learning_rate": 9.201209271078267e-05,
      "loss": 1.187,
      "step": 9720
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9273577332496643,
      "learning_rate": 9.200369499496138e-05,
      "loss": 1.1558,
      "step": 9730
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9076495170593262,
      "learning_rate": 9.199529727914008e-05,
      "loss": 1.1535,
      "step": 9740
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9300388693809509,
      "learning_rate": 9.198689956331878e-05,
      "loss": 1.1452,
      "step": 9750
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8782075047492981,
      "learning_rate": 9.197850184749748e-05,
      "loss": 1.1606,
      "step": 9760
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8622875809669495,
      "learning_rate": 9.19701041316762e-05,
      "loss": 1.1505,
      "step": 9770
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9150767922401428,
      "learning_rate": 9.196170641585488e-05,
      "loss": 1.1427,
      "step": 9780
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8404105305671692,
      "learning_rate": 9.195330870003359e-05,
      "loss": 1.1717,
      "step": 9790
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9195795059204102,
      "learning_rate": 9.19449109842123e-05,
      "loss": 1.1623,
      "step": 9800
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9158319234848022,
      "learning_rate": 9.1936513268391e-05,
      "loss": 1.1797,
      "step": 9810
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.859521210193634,
      "learning_rate": 9.19281155525697e-05,
      "loss": 1.1862,
      "step": 9820
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.876270055770874,
      "learning_rate": 9.191971783674842e-05,
      "loss": 1.1652,
      "step": 9830
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8563090562820435,
      "learning_rate": 9.191132012092712e-05,
      "loss": 1.1576,
      "step": 9840
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7962527275085449,
      "learning_rate": 9.190292240510582e-05,
      "loss": 1.1742,
      "step": 9850
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9961697459220886,
      "learning_rate": 9.189452468928452e-05,
      "loss": 1.1736,
      "step": 9860
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9126843810081482,
      "learning_rate": 9.188612697346323e-05,
      "loss": 1.1671,
      "step": 9870
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9100418090820312,
      "learning_rate": 9.187772925764193e-05,
      "loss": 1.1627,
      "step": 9880
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9611273407936096,
      "learning_rate": 9.186933154182062e-05,
      "loss": 1.2006,
      "step": 9890
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8314115405082703,
      "learning_rate": 9.186093382599933e-05,
      "loss": 1.1535,
      "step": 9900
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8938583135604858,
      "learning_rate": 9.185253611017803e-05,
      "loss": 1.1356,
      "step": 9910
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8760045170783997,
      "learning_rate": 9.184413839435674e-05,
      "loss": 1.1729,
      "step": 9920
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9797518253326416,
      "learning_rate": 9.183574067853544e-05,
      "loss": 1.1655,
      "step": 9930
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9288764595985413,
      "learning_rate": 9.182734296271415e-05,
      "loss": 1.1553,
      "step": 9940
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8398875594139099,
      "learning_rate": 9.181894524689285e-05,
      "loss": 1.1733,
      "step": 9950
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9302919507026672,
      "learning_rate": 9.181054753107155e-05,
      "loss": 1.1741,
      "step": 9960
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9887378215789795,
      "learning_rate": 9.180214981525027e-05,
      "loss": 1.1858,
      "step": 9970
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8962682485580444,
      "learning_rate": 9.179375209942897e-05,
      "loss": 1.185,
      "step": 9980
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9205465912818909,
      "learning_rate": 9.178535438360765e-05,
      "loss": 1.1588,
      "step": 9990
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8461117148399353,
      "learning_rate": 9.177695666778637e-05,
      "loss": 1.1647,
      "step": 10000
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8444212675094604,
      "learning_rate": 9.176855895196507e-05,
      "loss": 1.1676,
      "step": 10010
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8764084577560425,
      "learning_rate": 9.176016123614377e-05,
      "loss": 1.1807,
      "step": 10020
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9083003997802734,
      "learning_rate": 9.175176352032247e-05,
      "loss": 1.1883,
      "step": 10030
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9371087551116943,
      "learning_rate": 9.174336580450118e-05,
      "loss": 1.1619,
      "step": 10040
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7800226807594299,
      "learning_rate": 9.173496808867988e-05,
      "loss": 1.1596,
      "step": 10050
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9080156087875366,
      "learning_rate": 9.172657037285859e-05,
      "loss": 1.1919,
      "step": 10060
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9327325224876404,
      "learning_rate": 9.17181726570373e-05,
      "loss": 1.1572,
      "step": 10070
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9094728231430054,
      "learning_rate": 9.1709774941216e-05,
      "loss": 1.1813,
      "step": 10080
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.922223687171936,
      "learning_rate": 9.17013772253947e-05,
      "loss": 1.1788,
      "step": 10090
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9285407662391663,
      "learning_rate": 9.16929795095734e-05,
      "loss": 1.1764,
      "step": 10100
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9567670226097107,
      "learning_rate": 9.16845817937521e-05,
      "loss": 1.1673,
      "step": 10110
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9216251373291016,
      "learning_rate": 9.16761840779308e-05,
      "loss": 1.1586,
      "step": 10120
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8563306927680969,
      "learning_rate": 9.16677863621095e-05,
      "loss": 1.1696,
      "step": 10130
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8764593601226807,
      "learning_rate": 9.165938864628822e-05,
      "loss": 1.1917,
      "step": 10140
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8917428255081177,
      "learning_rate": 9.165099093046692e-05,
      "loss": 1.1826,
      "step": 10150
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8859480619430542,
      "learning_rate": 9.164259321464562e-05,
      "loss": 1.1565,
      "step": 10160
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8752598762512207,
      "learning_rate": 9.163419549882432e-05,
      "loss": 1.1547,
      "step": 10170
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9562612771987915,
      "learning_rate": 9.162579778300303e-05,
      "loss": 1.1731,
      "step": 10180
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9379001259803772,
      "learning_rate": 9.161740006718174e-05,
      "loss": 1.1668,
      "step": 10190
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9116417169570923,
      "learning_rate": 9.160900235136042e-05,
      "loss": 1.1662,
      "step": 10200
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8588632941246033,
      "learning_rate": 9.160060463553914e-05,
      "loss": 1.1424,
      "step": 10210
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9671397805213928,
      "learning_rate": 9.159220691971784e-05,
      "loss": 1.1892,
      "step": 10220
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.845672070980072,
      "learning_rate": 9.158380920389654e-05,
      "loss": 1.1758,
      "step": 10230
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9316462278366089,
      "learning_rate": 9.157541148807525e-05,
      "loss": 1.168,
      "step": 10240
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8582399487495422,
      "learning_rate": 9.156701377225395e-05,
      "loss": 1.1648,
      "step": 10250
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8619750142097473,
      "learning_rate": 9.155861605643265e-05,
      "loss": 1.1883,
      "step": 10260
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.935123085975647,
      "learning_rate": 9.155021834061135e-05,
      "loss": 1.1745,
      "step": 10270
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8686178922653198,
      "learning_rate": 9.154182062479007e-05,
      "loss": 1.1724,
      "step": 10280
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9323543906211853,
      "learning_rate": 9.153342290896877e-05,
      "loss": 1.1667,
      "step": 10290
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9259883761405945,
      "learning_rate": 9.152502519314747e-05,
      "loss": 1.1531,
      "step": 10300
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9095200896263123,
      "learning_rate": 9.151662747732617e-05,
      "loss": 1.1697,
      "step": 10310
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9084714651107788,
      "learning_rate": 9.150822976150487e-05,
      "loss": 1.1824,
      "step": 10320
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9516263008117676,
      "learning_rate": 9.149983204568357e-05,
      "loss": 1.156,
      "step": 10330
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9567800760269165,
      "learning_rate": 9.149143432986229e-05,
      "loss": 1.1637,
      "step": 10340
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9155131578445435,
      "learning_rate": 9.148303661404099e-05,
      "loss": 1.1637,
      "step": 10350
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8505834937095642,
      "learning_rate": 9.147463889821969e-05,
      "loss": 1.1756,
      "step": 10360
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8978041410446167,
      "learning_rate": 9.146624118239839e-05,
      "loss": 1.1489,
      "step": 10370
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9028456807136536,
      "learning_rate": 9.14578434665771e-05,
      "loss": 1.1482,
      "step": 10380
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9012831449508667,
      "learning_rate": 9.14494457507558e-05,
      "loss": 1.184,
      "step": 10390
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9119836091995239,
      "learning_rate": 9.14410480349345e-05,
      "loss": 1.149,
      "step": 10400
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8373876214027405,
      "learning_rate": 9.14326503191132e-05,
      "loss": 1.171,
      "step": 10410
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9856306314468384,
      "learning_rate": 9.14242526032919e-05,
      "loss": 1.1641,
      "step": 10420
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9325682520866394,
      "learning_rate": 9.14158548874706e-05,
      "loss": 1.1481,
      "step": 10430
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9177975654602051,
      "learning_rate": 9.14074571716493e-05,
      "loss": 1.172,
      "step": 10440
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8825708627700806,
      "learning_rate": 9.139905945582802e-05,
      "loss": 1.1624,
      "step": 10450
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8930522203445435,
      "learning_rate": 9.139066174000672e-05,
      "loss": 1.1812,
      "step": 10460
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.957118809223175,
      "learning_rate": 9.138226402418542e-05,
      "loss": 1.165,
      "step": 10470
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9745475649833679,
      "learning_rate": 9.137386630836414e-05,
      "loss": 1.1507,
      "step": 10480
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0125019550323486,
      "learning_rate": 9.136546859254284e-05,
      "loss": 1.1545,
      "step": 10490
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8771555423736572,
      "learning_rate": 9.135707087672154e-05,
      "loss": 1.1521,
      "step": 10500
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8748019933700562,
      "learning_rate": 9.134867316090024e-05,
      "loss": 1.14,
      "step": 10510
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8950918316841125,
      "learning_rate": 9.134027544507894e-05,
      "loss": 1.1486,
      "step": 10520
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9525067210197449,
      "learning_rate": 9.133187772925764e-05,
      "loss": 1.1501,
      "step": 10530
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8265681266784668,
      "learning_rate": 9.132348001343634e-05,
      "loss": 1.1605,
      "step": 10540
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9634080529212952,
      "learning_rate": 9.131508229761505e-05,
      "loss": 1.1561,
      "step": 10550
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8635426759719849,
      "learning_rate": 9.130668458179376e-05,
      "loss": 1.1696,
      "step": 10560
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.994739294052124,
      "learning_rate": 9.129828686597246e-05,
      "loss": 1.1778,
      "step": 10570
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.89042729139328,
      "learning_rate": 9.128988915015117e-05,
      "loss": 1.1667,
      "step": 10580
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.868971586227417,
      "learning_rate": 9.128149143432987e-05,
      "loss": 1.1775,
      "step": 10590
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9349837899208069,
      "learning_rate": 9.127309371850857e-05,
      "loss": 1.1653,
      "step": 10600
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9134425520896912,
      "learning_rate": 9.126469600268727e-05,
      "loss": 1.1641,
      "step": 10610
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9141260385513306,
      "learning_rate": 9.125629828686599e-05,
      "loss": 1.1789,
      "step": 10620
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9114556908607483,
      "learning_rate": 9.124790057104467e-05,
      "loss": 1.1265,
      "step": 10630
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8724734783172607,
      "learning_rate": 9.123950285522337e-05,
      "loss": 1.1798,
      "step": 10640
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8508509993553162,
      "learning_rate": 9.123110513940209e-05,
      "loss": 1.1806,
      "step": 10650
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8755638003349304,
      "learning_rate": 9.122270742358079e-05,
      "loss": 1.1511,
      "step": 10660
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9825658798217773,
      "learning_rate": 9.121430970775949e-05,
      "loss": 1.1432,
      "step": 10670
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8838762044906616,
      "learning_rate": 9.120591199193819e-05,
      "loss": 1.1659,
      "step": 10680
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9895627498626709,
      "learning_rate": 9.11975142761169e-05,
      "loss": 1.1722,
      "step": 10690
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9133514761924744,
      "learning_rate": 9.11891165602956e-05,
      "loss": 1.1597,
      "step": 10700
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9397072196006775,
      "learning_rate": 9.11807188444743e-05,
      "loss": 1.1699,
      "step": 10710
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9063377380371094,
      "learning_rate": 9.117232112865302e-05,
      "loss": 1.1468,
      "step": 10720
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8843746781349182,
      "learning_rate": 9.116392341283171e-05,
      "loss": 1.167,
      "step": 10730
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9365702867507935,
      "learning_rate": 9.115552569701041e-05,
      "loss": 1.1658,
      "step": 10740
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9855867624282837,
      "learning_rate": 9.114712798118912e-05,
      "loss": 1.1822,
      "step": 10750
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8695436120033264,
      "learning_rate": 9.113873026536782e-05,
      "loss": 1.1581,
      "step": 10760
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9232392311096191,
      "learning_rate": 9.113033254954652e-05,
      "loss": 1.154,
      "step": 10770
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9410752654075623,
      "learning_rate": 9.112193483372522e-05,
      "loss": 1.1388,
      "step": 10780
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9676739573478699,
      "learning_rate": 9.111353711790394e-05,
      "loss": 1.1589,
      "step": 10790
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8949344754219055,
      "learning_rate": 9.110513940208264e-05,
      "loss": 1.1707,
      "step": 10800
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9615799188613892,
      "learning_rate": 9.109674168626134e-05,
      "loss": 1.1523,
      "step": 10810
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9283060431480408,
      "learning_rate": 9.108834397044005e-05,
      "loss": 1.1549,
      "step": 10820
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9590465426445007,
      "learning_rate": 9.107994625461876e-05,
      "loss": 1.1803,
      "step": 10830
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.908495306968689,
      "learning_rate": 9.107154853879744e-05,
      "loss": 1.1709,
      "step": 10840
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9319912791252136,
      "learning_rate": 9.106315082297616e-05,
      "loss": 1.1479,
      "step": 10850
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9135957360267639,
      "learning_rate": 9.105475310715486e-05,
      "loss": 1.1666,
      "step": 10860
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8900753259658813,
      "learning_rate": 9.104635539133356e-05,
      "loss": 1.1434,
      "step": 10870
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8928583860397339,
      "learning_rate": 9.103795767551226e-05,
      "loss": 1.151,
      "step": 10880
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9014522433280945,
      "learning_rate": 9.102955995969097e-05,
      "loss": 1.1611,
      "step": 10890
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9392279386520386,
      "learning_rate": 9.102116224386967e-05,
      "loss": 1.1478,
      "step": 10900
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9129185080528259,
      "learning_rate": 9.101276452804837e-05,
      "loss": 1.1241,
      "step": 10910
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9602782726287842,
      "learning_rate": 9.100436681222709e-05,
      "loss": 1.158,
      "step": 10920
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.955150306224823,
      "learning_rate": 9.099596909640579e-05,
      "loss": 1.1292,
      "step": 10930
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9290515184402466,
      "learning_rate": 9.098757138058449e-05,
      "loss": 1.1485,
      "step": 10940
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9749833941459656,
      "learning_rate": 9.097917366476319e-05,
      "loss": 1.1289,
      "step": 10950
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8840381503105164,
      "learning_rate": 9.097077594894189e-05,
      "loss": 1.1734,
      "step": 10960
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.852929413318634,
      "learning_rate": 9.096237823312059e-05,
      "loss": 1.1719,
      "step": 10970
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9287542104721069,
      "learning_rate": 9.095398051729929e-05,
      "loss": 1.1687,
      "step": 10980
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9062138199806213,
      "learning_rate": 9.094558280147801e-05,
      "loss": 1.1606,
      "step": 10990
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8329747319221497,
      "learning_rate": 9.093718508565671e-05,
      "loss": 1.1551,
      "step": 11000
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0348055362701416,
      "learning_rate": 9.092878736983541e-05,
      "loss": 1.162,
      "step": 11010
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.895537257194519,
      "learning_rate": 9.092038965401411e-05,
      "loss": 1.1596,
      "step": 11020
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.895803689956665,
      "learning_rate": 9.091199193819282e-05,
      "loss": 1.1467,
      "step": 11030
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8784132599830627,
      "learning_rate": 9.090443399395366e-05,
      "loss": 1.1938,
      "step": 11040
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8735492825508118,
      "learning_rate": 9.089603627813236e-05,
      "loss": 1.158,
      "step": 11050
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8453689217567444,
      "learning_rate": 9.088763856231105e-05,
      "loss": 1.163,
      "step": 11060
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9577978849411011,
      "learning_rate": 9.087924084648976e-05,
      "loss": 1.1553,
      "step": 11070
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9026578068733215,
      "learning_rate": 9.087084313066846e-05,
      "loss": 1.1663,
      "step": 11080
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0003416538238525,
      "learning_rate": 9.086244541484716e-05,
      "loss": 1.1512,
      "step": 11090
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8842092156410217,
      "learning_rate": 9.085404769902586e-05,
      "loss": 1.1562,
      "step": 11100
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9195184111595154,
      "learning_rate": 9.084564998320458e-05,
      "loss": 1.1726,
      "step": 11110
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8893963694572449,
      "learning_rate": 9.083725226738328e-05,
      "loss": 1.1565,
      "step": 11120
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9130131006240845,
      "learning_rate": 9.082885455156198e-05,
      "loss": 1.1418,
      "step": 11130
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8881024122238159,
      "learning_rate": 9.082045683574069e-05,
      "loss": 1.1546,
      "step": 11140
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9244464039802551,
      "learning_rate": 9.081205911991939e-05,
      "loss": 1.1404,
      "step": 11150
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9388617873191833,
      "learning_rate": 9.08036614040981e-05,
      "loss": 1.1541,
      "step": 11160
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.936011016368866,
      "learning_rate": 9.07952636882768e-05,
      "loss": 1.1812,
      "step": 11170
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8580996990203857,
      "learning_rate": 9.07868659724555e-05,
      "loss": 1.1703,
      "step": 11180
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8237022161483765,
      "learning_rate": 9.07784682566342e-05,
      "loss": 1.1398,
      "step": 11190
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8908625245094299,
      "learning_rate": 9.07700705408129e-05,
      "loss": 1.1771,
      "step": 11200
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8714804649353027,
      "learning_rate": 9.076167282499161e-05,
      "loss": 1.1444,
      "step": 11210
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9208613634109497,
      "learning_rate": 9.075327510917031e-05,
      "loss": 1.141,
      "step": 11220
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9038468599319458,
      "learning_rate": 9.074487739334901e-05,
      "loss": 1.1353,
      "step": 11230
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9819357991218567,
      "learning_rate": 9.073647967752771e-05,
      "loss": 1.1558,
      "step": 11240
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8615383505821228,
      "learning_rate": 9.072808196170643e-05,
      "loss": 1.1697,
      "step": 11250
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8920129537582397,
      "learning_rate": 9.071968424588513e-05,
      "loss": 1.1719,
      "step": 11260
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.90557461977005,
      "learning_rate": 9.071128653006383e-05,
      "loss": 1.1732,
      "step": 11270
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8542124032974243,
      "learning_rate": 9.070288881424253e-05,
      "loss": 1.1391,
      "step": 11280
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9274539947509766,
      "learning_rate": 9.069449109842123e-05,
      "loss": 1.1668,
      "step": 11290
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9204723238945007,
      "learning_rate": 9.068609338259993e-05,
      "loss": 1.1517,
      "step": 11300
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9253469705581665,
      "learning_rate": 9.067769566677864e-05,
      "loss": 1.152,
      "step": 11310
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8748138546943665,
      "learning_rate": 9.066929795095735e-05,
      "loss": 1.1208,
      "step": 11320
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9187504053115845,
      "learning_rate": 9.066090023513605e-05,
      "loss": 1.1434,
      "step": 11330
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8502510786056519,
      "learning_rate": 9.065250251931475e-05,
      "loss": 1.1481,
      "step": 11340
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9351586103439331,
      "learning_rate": 9.064410480349346e-05,
      "loss": 1.1405,
      "step": 11350
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9439583420753479,
      "learning_rate": 9.063570708767216e-05,
      "loss": 1.1625,
      "step": 11360
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9154985547065735,
      "learning_rate": 9.062730937185086e-05,
      "loss": 1.1505,
      "step": 11370
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9449728727340698,
      "learning_rate": 9.061891165602956e-05,
      "loss": 1.1566,
      "step": 11380
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8563029766082764,
      "learning_rate": 9.061051394020826e-05,
      "loss": 1.1659,
      "step": 11390
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8563398122787476,
      "learning_rate": 9.060211622438696e-05,
      "loss": 1.1266,
      "step": 11400
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9950500726699829,
      "learning_rate": 9.059371850856568e-05,
      "loss": 1.1569,
      "step": 11410
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8899383544921875,
      "learning_rate": 9.058532079274438e-05,
      "loss": 1.1406,
      "step": 11420
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9327157139778137,
      "learning_rate": 9.057692307692308e-05,
      "loss": 1.132,
      "step": 11430
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8532604575157166,
      "learning_rate": 9.056852536110178e-05,
      "loss": 1.1256,
      "step": 11440
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0322303771972656,
      "learning_rate": 9.05601276452805e-05,
      "loss": 1.1338,
      "step": 11450
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.937974214553833,
      "learning_rate": 9.05517299294592e-05,
      "loss": 1.1593,
      "step": 11460
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9321600198745728,
      "learning_rate": 9.05433322136379e-05,
      "loss": 1.1329,
      "step": 11470
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9697784781455994,
      "learning_rate": 9.05349344978166e-05,
      "loss": 1.1496,
      "step": 11480
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9312222599983215,
      "learning_rate": 9.05265367819953e-05,
      "loss": 1.1511,
      "step": 11490
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9349722266197205,
      "learning_rate": 9.0518139066174e-05,
      "loss": 1.1425,
      "step": 11500
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9298446774482727,
      "learning_rate": 9.050974135035271e-05,
      "loss": 1.1334,
      "step": 11510
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9233946800231934,
      "learning_rate": 9.050134363453141e-05,
      "loss": 1.145,
      "step": 11520
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8692603707313538,
      "learning_rate": 9.049294591871011e-05,
      "loss": 1.1444,
      "step": 11530
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9367170333862305,
      "learning_rate": 9.048454820288881e-05,
      "loss": 1.1709,
      "step": 11540
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8840388059616089,
      "learning_rate": 9.047615048706753e-05,
      "loss": 1.1488,
      "step": 11550
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8870083093643188,
      "learning_rate": 9.046775277124623e-05,
      "loss": 1.1619,
      "step": 11560
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9165491461753845,
      "learning_rate": 9.045935505542493e-05,
      "loss": 1.1678,
      "step": 11570
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9082239270210266,
      "learning_rate": 9.045095733960363e-05,
      "loss": 1.1483,
      "step": 11580
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8463948965072632,
      "learning_rate": 9.044255962378233e-05,
      "loss": 1.1622,
      "step": 11590
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9144827127456665,
      "learning_rate": 9.043416190796103e-05,
      "loss": 1.1706,
      "step": 11600
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.887029230594635,
      "learning_rate": 9.042576419213973e-05,
      "loss": 1.1497,
      "step": 11610
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8472746014595032,
      "learning_rate": 9.041736647631845e-05,
      "loss": 1.1532,
      "step": 11620
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9237121939659119,
      "learning_rate": 9.040896876049715e-05,
      "loss": 1.1652,
      "step": 11630
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9166734218597412,
      "learning_rate": 9.040057104467585e-05,
      "loss": 1.1908,
      "step": 11640
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.907728910446167,
      "learning_rate": 9.039217332885456e-05,
      "loss": 1.1566,
      "step": 11650
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.887100875377655,
      "learning_rate": 9.038377561303326e-05,
      "loss": 1.1541,
      "step": 11660
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8894091844558716,
      "learning_rate": 9.037537789721196e-05,
      "loss": 1.1457,
      "step": 11670
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9175511002540588,
      "learning_rate": 9.036698018139066e-05,
      "loss": 1.1317,
      "step": 11680
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8548650145530701,
      "learning_rate": 9.035858246556938e-05,
      "loss": 1.1816,
      "step": 11690
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9049464464187622,
      "learning_rate": 9.035018474974807e-05,
      "loss": 1.1413,
      "step": 11700
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8453555703163147,
      "learning_rate": 9.034178703392677e-05,
      "loss": 1.1587,
      "step": 11710
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9028091430664062,
      "learning_rate": 9.033338931810548e-05,
      "loss": 1.1363,
      "step": 11720
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8624139428138733,
      "learning_rate": 9.032499160228418e-05,
      "loss": 1.1436,
      "step": 11730
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9154136180877686,
      "learning_rate": 9.031659388646288e-05,
      "loss": 1.1431,
      "step": 11740
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9075929522514343,
      "learning_rate": 9.03081961706416e-05,
      "loss": 1.1415,
      "step": 11750
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8844103217124939,
      "learning_rate": 9.02997984548203e-05,
      "loss": 1.1394,
      "step": 11760
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8745536208152771,
      "learning_rate": 9.0291400738999e-05,
      "loss": 1.1666,
      "step": 11770
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8478753566741943,
      "learning_rate": 9.02830030231777e-05,
      "loss": 1.1605,
      "step": 11780
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9206047654151917,
      "learning_rate": 9.027460530735641e-05,
      "loss": 1.1279,
      "step": 11790
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8796776533126831,
      "learning_rate": 9.02662075915351e-05,
      "loss": 1.1264,
      "step": 11800
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8907874226570129,
      "learning_rate": 9.02578098757138e-05,
      "loss": 1.1498,
      "step": 11810
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9629140496253967,
      "learning_rate": 9.024941215989252e-05,
      "loss": 1.151,
      "step": 11820
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.897369384765625,
      "learning_rate": 9.024101444407122e-05,
      "loss": 1.1438,
      "step": 11830
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8689546585083008,
      "learning_rate": 9.023261672824992e-05,
      "loss": 1.1573,
      "step": 11840
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8658522963523865,
      "learning_rate": 9.022421901242862e-05,
      "loss": 1.1485,
      "step": 11850
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9156214594841003,
      "learning_rate": 9.021582129660733e-05,
      "loss": 1.1285,
      "step": 11860
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8841491937637329,
      "learning_rate": 9.020742358078603e-05,
      "loss": 1.1536,
      "step": 11870
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9126939177513123,
      "learning_rate": 9.019902586496473e-05,
      "loss": 1.1572,
      "step": 11880
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9111533164978027,
      "learning_rate": 9.019062814914345e-05,
      "loss": 1.164,
      "step": 11890
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8516184687614441,
      "learning_rate": 9.018223043332215e-05,
      "loss": 1.1695,
      "step": 11900
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9268118739128113,
      "learning_rate": 9.017383271750083e-05,
      "loss": 1.1369,
      "step": 11910
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8939505219459534,
      "learning_rate": 9.016543500167955e-05,
      "loss": 1.1642,
      "step": 11920
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.3688740730285645,
      "eval_runtime": 6239.3798,
      "eval_samples_per_second": 266.106,
      "eval_steps_per_second": 4.158,
      "step": 11928
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9582164287567139,
      "learning_rate": 9.015703728585825e-05,
      "loss": 1.1547,
      "step": 11930
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8907540440559387,
      "learning_rate": 9.014863957003695e-05,
      "loss": 1.1259,
      "step": 11940
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1118344068527222,
      "learning_rate": 9.014024185421565e-05,
      "loss": 1.1284,
      "step": 11950
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9015932679176331,
      "learning_rate": 9.013184413839437e-05,
      "loss": 1.1224,
      "step": 11960
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8888578414916992,
      "learning_rate": 9.012344642257307e-05,
      "loss": 1.1172,
      "step": 11970
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8929213881492615,
      "learning_rate": 9.011504870675177e-05,
      "loss": 1.116,
      "step": 11980
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9146283864974976,
      "learning_rate": 9.010665099093048e-05,
      "loss": 1.1452,
      "step": 11990
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9056942462921143,
      "learning_rate": 9.009825327510918e-05,
      "loss": 1.1497,
      "step": 12000
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9360379576683044,
      "learning_rate": 9.008985555928788e-05,
      "loss": 1.1482,
      "step": 12010
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9058497548103333,
      "learning_rate": 9.008145784346658e-05,
      "loss": 1.1476,
      "step": 12020
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9097636938095093,
      "learning_rate": 9.007306012764528e-05,
      "loss": 1.1122,
      "step": 12030
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9412002563476562,
      "learning_rate": 9.006466241182398e-05,
      "loss": 1.1507,
      "step": 12040
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9335237741470337,
      "learning_rate": 9.005626469600269e-05,
      "loss": 1.138,
      "step": 12050
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8122248649597168,
      "learning_rate": 9.00478669801814e-05,
      "loss": 1.1395,
      "step": 12060
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9185567498207092,
      "learning_rate": 9.00394692643601e-05,
      "loss": 1.1466,
      "step": 12070
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8701349496841431,
      "learning_rate": 9.00310715485388e-05,
      "loss": 1.1274,
      "step": 12080
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9429726600646973,
      "learning_rate": 9.00226738327175e-05,
      "loss": 1.1621,
      "step": 12090
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8825283646583557,
      "learning_rate": 9.001427611689622e-05,
      "loss": 1.1305,
      "step": 12100
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.938089907169342,
      "learning_rate": 9.000587840107492e-05,
      "loss": 1.1252,
      "step": 12110
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8835190534591675,
      "learning_rate": 8.99974806852536e-05,
      "loss": 1.1597,
      "step": 12120
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9665217399597168,
      "learning_rate": 8.998908296943232e-05,
      "loss": 1.1147,
      "step": 12130
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9050577282905579,
      "learning_rate": 8.998068525361102e-05,
      "loss": 1.1258,
      "step": 12140
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9124248623847961,
      "learning_rate": 8.997228753778972e-05,
      "loss": 1.1242,
      "step": 12150
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9285717010498047,
      "learning_rate": 8.996388982196843e-05,
      "loss": 1.1339,
      "step": 12160
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9225600361824036,
      "learning_rate": 8.995549210614713e-05,
      "loss": 1.1437,
      "step": 12170
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9179040193557739,
      "learning_rate": 8.994709439032583e-05,
      "loss": 1.142,
      "step": 12180
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9169156551361084,
      "learning_rate": 8.993869667450454e-05,
      "loss": 1.1602,
      "step": 12190
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8557838201522827,
      "learning_rate": 8.993029895868325e-05,
      "loss": 1.1229,
      "step": 12200
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9075002670288086,
      "learning_rate": 8.992190124286195e-05,
      "loss": 1.133,
      "step": 12210
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8852888345718384,
      "learning_rate": 8.991350352704065e-05,
      "loss": 1.1424,
      "step": 12220
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.884982168674469,
      "learning_rate": 8.990510581121935e-05,
      "loss": 1.1372,
      "step": 12230
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8553844690322876,
      "learning_rate": 8.989670809539805e-05,
      "loss": 1.142,
      "step": 12240
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9492677450180054,
      "learning_rate": 8.988831037957675e-05,
      "loss": 1.1527,
      "step": 12250
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9053412079811096,
      "learning_rate": 8.987991266375547e-05,
      "loss": 1.1275,
      "step": 12260
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9139066338539124,
      "learning_rate": 8.987151494793417e-05,
      "loss": 1.1351,
      "step": 12270
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9355369210243225,
      "learning_rate": 8.986311723211287e-05,
      "loss": 1.1199,
      "step": 12280
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9116610884666443,
      "learning_rate": 8.985471951629157e-05,
      "loss": 1.1324,
      "step": 12290
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8947935700416565,
      "learning_rate": 8.984632180047028e-05,
      "loss": 1.1375,
      "step": 12300
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8590942025184631,
      "learning_rate": 8.983792408464898e-05,
      "loss": 1.1369,
      "step": 12310
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.006734848022461,
      "learning_rate": 8.982952636882769e-05,
      "loss": 1.131,
      "step": 12320
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9852463006973267,
      "learning_rate": 8.982112865300639e-05,
      "loss": 1.1401,
      "step": 12330
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9508471488952637,
      "learning_rate": 8.981273093718509e-05,
      "loss": 1.1185,
      "step": 12340
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9054518342018127,
      "learning_rate": 8.980433322136379e-05,
      "loss": 1.1347,
      "step": 12350
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9307007193565369,
      "learning_rate": 8.979593550554249e-05,
      "loss": 1.1169,
      "step": 12360
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9499795436859131,
      "learning_rate": 8.97875377897212e-05,
      "loss": 1.1481,
      "step": 12370
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9338065385818481,
      "learning_rate": 8.97791400738999e-05,
      "loss": 1.1127,
      "step": 12380
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8858123421669006,
      "learning_rate": 8.97707423580786e-05,
      "loss": 1.1279,
      "step": 12390
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9713876843452454,
      "learning_rate": 8.976234464225732e-05,
      "loss": 1.152,
      "step": 12400
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9448899030685425,
      "learning_rate": 8.975394692643602e-05,
      "loss": 1.116,
      "step": 12410
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9086986780166626,
      "learning_rate": 8.974554921061472e-05,
      "loss": 1.1483,
      "step": 12420
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8648169040679932,
      "learning_rate": 8.973715149479342e-05,
      "loss": 1.1272,
      "step": 12430
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0005784034729004,
      "learning_rate": 8.972875377897212e-05,
      "loss": 1.1388,
      "step": 12440
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9356294870376587,
      "learning_rate": 8.972035606315082e-05,
      "loss": 1.1399,
      "step": 12450
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8659923672676086,
      "learning_rate": 8.971195834732952e-05,
      "loss": 1.1269,
      "step": 12460
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8799879550933838,
      "learning_rate": 8.970356063150824e-05,
      "loss": 1.1384,
      "step": 12470
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9933121800422668,
      "learning_rate": 8.969516291568694e-05,
      "loss": 1.1437,
      "step": 12480
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9219959378242493,
      "learning_rate": 8.968676519986564e-05,
      "loss": 1.1333,
      "step": 12490
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8931273818016052,
      "learning_rate": 8.967836748404435e-05,
      "loss": 1.1088,
      "step": 12500
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9934206604957581,
      "learning_rate": 8.966996976822305e-05,
      "loss": 1.1439,
      "step": 12510
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9685350656509399,
      "learning_rate": 8.966157205240175e-05,
      "loss": 1.1238,
      "step": 12520
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9198681116104126,
      "learning_rate": 8.965317433658045e-05,
      "loss": 1.1155,
      "step": 12530
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9107829928398132,
      "learning_rate": 8.964477662075917e-05,
      "loss": 1.1306,
      "step": 12540
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8961829543113708,
      "learning_rate": 8.963637890493786e-05,
      "loss": 1.1327,
      "step": 12550
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8820117712020874,
      "learning_rate": 8.962798118911656e-05,
      "loss": 1.1368,
      "step": 12560
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9899251461029053,
      "learning_rate": 8.961958347329527e-05,
      "loss": 1.1217,
      "step": 12570
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8744221329689026,
      "learning_rate": 8.961118575747397e-05,
      "loss": 1.1484,
      "step": 12580
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8801760077476501,
      "learning_rate": 8.960278804165267e-05,
      "loss": 1.1265,
      "step": 12590
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.888437807559967,
      "learning_rate": 8.959439032583137e-05,
      "loss": 1.1398,
      "step": 12600
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9583972096443176,
      "learning_rate": 8.958599261001009e-05,
      "loss": 1.1228,
      "step": 12610
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8810545206069946,
      "learning_rate": 8.957759489418879e-05,
      "loss": 1.1467,
      "step": 12620
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8706270456314087,
      "learning_rate": 8.956919717836749e-05,
      "loss": 1.1321,
      "step": 12630
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9265187382698059,
      "learning_rate": 8.95607994625462e-05,
      "loss": 1.1348,
      "step": 12640
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8875660300254822,
      "learning_rate": 8.955240174672489e-05,
      "loss": 1.132,
      "step": 12650
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9585201740264893,
      "learning_rate": 8.954400403090359e-05,
      "loss": 1.1255,
      "step": 12660
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9146342277526855,
      "learning_rate": 8.95356063150823e-05,
      "loss": 1.0998,
      "step": 12670
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9078376293182373,
      "learning_rate": 8.9527208599261e-05,
      "loss": 1.1111,
      "step": 12680
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8908431529998779,
      "learning_rate": 8.95188108834397e-05,
      "loss": 1.1231,
      "step": 12690
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9033951163291931,
      "learning_rate": 8.95104131676184e-05,
      "loss": 1.1354,
      "step": 12700
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8694300055503845,
      "learning_rate": 8.950201545179712e-05,
      "loss": 1.1149,
      "step": 12710
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9056485295295715,
      "learning_rate": 8.949361773597582e-05,
      "loss": 1.1206,
      "step": 12720
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8865606188774109,
      "learning_rate": 8.948522002015452e-05,
      "loss": 1.1173,
      "step": 12730
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8753979206085205,
      "learning_rate": 8.947682230433324e-05,
      "loss": 1.1304,
      "step": 12740
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9080657362937927,
      "learning_rate": 8.946842458851194e-05,
      "loss": 1.1218,
      "step": 12750
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9731104373931885,
      "learning_rate": 8.946002687269062e-05,
      "loss": 1.1478,
      "step": 12760
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9440938830375671,
      "learning_rate": 8.945162915686934e-05,
      "loss": 1.1138,
      "step": 12770
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8991868495941162,
      "learning_rate": 8.944323144104804e-05,
      "loss": 1.1316,
      "step": 12780
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9543919563293457,
      "learning_rate": 8.943483372522674e-05,
      "loss": 1.1335,
      "step": 12790
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9015852808952332,
      "learning_rate": 8.942643600940544e-05,
      "loss": 1.1245,
      "step": 12800
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.979584813117981,
      "learning_rate": 8.941803829358415e-05,
      "loss": 1.1523,
      "step": 12810
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9773133993148804,
      "learning_rate": 8.940964057776286e-05,
      "loss": 1.1287,
      "step": 12820
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9675416946411133,
      "learning_rate": 8.940124286194156e-05,
      "loss": 1.1153,
      "step": 12830
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9855402708053589,
      "learning_rate": 8.939284514612026e-05,
      "loss": 1.14,
      "step": 12840
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8832024335861206,
      "learning_rate": 8.938444743029897e-05,
      "loss": 1.1252,
      "step": 12850
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9848988056182861,
      "learning_rate": 8.937604971447766e-05,
      "loss": 1.1218,
      "step": 12860
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9306350350379944,
      "learning_rate": 8.936765199865636e-05,
      "loss": 1.1239,
      "step": 12870
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9815962910652161,
      "learning_rate": 8.935925428283507e-05,
      "loss": 1.1228,
      "step": 12880
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9946402907371521,
      "learning_rate": 8.935085656701377e-05,
      "loss": 1.1246,
      "step": 12890
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8948855996131897,
      "learning_rate": 8.934245885119247e-05,
      "loss": 1.1028,
      "step": 12900
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9454535245895386,
      "learning_rate": 8.933406113537119e-05,
      "loss": 1.1268,
      "step": 12910
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.086484670639038,
      "learning_rate": 8.932566341954989e-05,
      "loss": 1.1287,
      "step": 12920
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9772274494171143,
      "learning_rate": 8.931726570372859e-05,
      "loss": 1.1373,
      "step": 12930
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9615238308906555,
      "learning_rate": 8.930886798790729e-05,
      "loss": 1.1153,
      "step": 12940
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8882322311401367,
      "learning_rate": 8.9300470272086e-05,
      "loss": 1.1171,
      "step": 12950
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9522902965545654,
      "learning_rate": 8.92920725562647e-05,
      "loss": 1.1239,
      "step": 12960
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9135017991065979,
      "learning_rate": 8.928367484044339e-05,
      "loss": 1.1188,
      "step": 12970
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9306880235671997,
      "learning_rate": 8.927527712462211e-05,
      "loss": 1.123,
      "step": 12980
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9436923861503601,
      "learning_rate": 8.926687940880081e-05,
      "loss": 1.1585,
      "step": 12990
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9140714406967163,
      "learning_rate": 8.925848169297951e-05,
      "loss": 1.1146,
      "step": 13000
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9659335613250732,
      "learning_rate": 8.925008397715822e-05,
      "loss": 1.1426,
      "step": 13010
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8587934374809265,
      "learning_rate": 8.924168626133692e-05,
      "loss": 1.1476,
      "step": 13020
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8232706189155579,
      "learning_rate": 8.923328854551562e-05,
      "loss": 1.1218,
      "step": 13030
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9183794856071472,
      "learning_rate": 8.922489082969432e-05,
      "loss": 1.1379,
      "step": 13040
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9124522805213928,
      "learning_rate": 8.921649311387304e-05,
      "loss": 1.1068,
      "step": 13050
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9751561284065247,
      "learning_rate": 8.920809539805174e-05,
      "loss": 1.1305,
      "step": 13060
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8878384232521057,
      "learning_rate": 8.919969768223044e-05,
      "loss": 1.1397,
      "step": 13070
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9667701125144958,
      "learning_rate": 8.919129996640914e-05,
      "loss": 1.1258,
      "step": 13080
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9521215558052063,
      "learning_rate": 8.918290225058784e-05,
      "loss": 1.1199,
      "step": 13090
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9559984803199768,
      "learning_rate": 8.917450453476654e-05,
      "loss": 1.1239,
      "step": 13100
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9714217782020569,
      "learning_rate": 8.916610681894526e-05,
      "loss": 1.1227,
      "step": 13110
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9680772423744202,
      "learning_rate": 8.915770910312396e-05,
      "loss": 1.1287,
      "step": 13120
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8694576621055603,
      "learning_rate": 8.914931138730266e-05,
      "loss": 1.1511,
      "step": 13130
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9465891718864441,
      "learning_rate": 8.914091367148136e-05,
      "loss": 1.1347,
      "step": 13140
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9178768396377563,
      "learning_rate": 8.913251595566007e-05,
      "loss": 1.1349,
      "step": 13150
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9185075759887695,
      "learning_rate": 8.91249580114209e-05,
      "loss": 1.1106,
      "step": 13160
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8623015880584717,
      "learning_rate": 8.911656029559961e-05,
      "loss": 1.1217,
      "step": 13170
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9602338671684265,
      "learning_rate": 8.910816257977831e-05,
      "loss": 1.1426,
      "step": 13180
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9467712640762329,
      "learning_rate": 8.9099764863957e-05,
      "loss": 1.1318,
      "step": 13190
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0699985027313232,
      "learning_rate": 8.909136714813571e-05,
      "loss": 1.1224,
      "step": 13200
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9185535311698914,
      "learning_rate": 8.908296943231441e-05,
      "loss": 1.1364,
      "step": 13210
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8876036405563354,
      "learning_rate": 8.907457171649311e-05,
      "loss": 1.1465,
      "step": 13220
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9188990592956543,
      "learning_rate": 8.906617400067183e-05,
      "loss": 1.1087,
      "step": 13230
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8698427677154541,
      "learning_rate": 8.905777628485053e-05,
      "loss": 1.123,
      "step": 13240
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9527921080589294,
      "learning_rate": 8.904937856902923e-05,
      "loss": 1.1522,
      "step": 13250
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9535458087921143,
      "learning_rate": 8.904098085320793e-05,
      "loss": 1.1181,
      "step": 13260
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9558416604995728,
      "learning_rate": 8.903258313738664e-05,
      "loss": 1.1384,
      "step": 13270
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9938929677009583,
      "learning_rate": 8.902418542156534e-05,
      "loss": 1.1259,
      "step": 13280
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8923820853233337,
      "learning_rate": 8.901578770574404e-05,
      "loss": 1.1163,
      "step": 13290
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8947280049324036,
      "learning_rate": 8.900738998992274e-05,
      "loss": 1.1203,
      "step": 13300
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9547944664955139,
      "learning_rate": 8.899899227410144e-05,
      "loss": 1.1546,
      "step": 13310
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9443290829658508,
      "learning_rate": 8.899059455828015e-05,
      "loss": 1.1091,
      "step": 13320
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8698960542678833,
      "learning_rate": 8.898219684245886e-05,
      "loss": 1.1139,
      "step": 13330
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9820073843002319,
      "learning_rate": 8.897379912663756e-05,
      "loss": 1.0837,
      "step": 13340
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9299082159996033,
      "learning_rate": 8.896540141081626e-05,
      "loss": 1.1372,
      "step": 13350
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.929469108581543,
      "learning_rate": 8.895700369499496e-05,
      "loss": 1.146,
      "step": 13360
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9092618823051453,
      "learning_rate": 8.894860597917368e-05,
      "loss": 1.1207,
      "step": 13370
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9252662062644958,
      "learning_rate": 8.894020826335238e-05,
      "loss": 1.1167,
      "step": 13380
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8673176169395447,
      "learning_rate": 8.893181054753108e-05,
      "loss": 1.1265,
      "step": 13390
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8885186910629272,
      "learning_rate": 8.892341283170978e-05,
      "loss": 1.142,
      "step": 13400
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9046466946601868,
      "learning_rate": 8.891501511588848e-05,
      "loss": 1.1231,
      "step": 13410
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.954371452331543,
      "learning_rate": 8.890661740006718e-05,
      "loss": 1.1062,
      "step": 13420
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9183551669120789,
      "learning_rate": 8.889821968424588e-05,
      "loss": 1.1086,
      "step": 13430
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9714826941490173,
      "learning_rate": 8.88898219684246e-05,
      "loss": 1.1209,
      "step": 13440
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8576006293296814,
      "learning_rate": 8.88814242526033e-05,
      "loss": 1.1001,
      "step": 13450
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8900761008262634,
      "learning_rate": 8.8873026536782e-05,
      "loss": 1.1248,
      "step": 13460
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8779008984565735,
      "learning_rate": 8.886462882096071e-05,
      "loss": 1.105,
      "step": 13470
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8928524851799011,
      "learning_rate": 8.885623110513941e-05,
      "loss": 1.1053,
      "step": 13480
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8737199902534485,
      "learning_rate": 8.884783338931811e-05,
      "loss": 1.1408,
      "step": 13490
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.951984703540802,
      "learning_rate": 8.883943567349681e-05,
      "loss": 1.1152,
      "step": 13500
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9574511647224426,
      "learning_rate": 8.883103795767551e-05,
      "loss": 1.1118,
      "step": 13510
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9811375737190247,
      "learning_rate": 8.882264024185421e-05,
      "loss": 1.159,
      "step": 13520
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.923875093460083,
      "learning_rate": 8.881424252603291e-05,
      "loss": 1.1295,
      "step": 13530
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9695706367492676,
      "learning_rate": 8.880584481021163e-05,
      "loss": 1.132,
      "step": 13540
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0641175508499146,
      "learning_rate": 8.879744709439033e-05,
      "loss": 1.1273,
      "step": 13550
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.870869517326355,
      "learning_rate": 8.878904937856903e-05,
      "loss": 1.1101,
      "step": 13560
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.937880277633667,
      "learning_rate": 8.878065166274774e-05,
      "loss": 1.0993,
      "step": 13570
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8831961154937744,
      "learning_rate": 8.877225394692644e-05,
      "loss": 1.1077,
      "step": 13580
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8785178065299988,
      "learning_rate": 8.876385623110515e-05,
      "loss": 1.1198,
      "step": 13590
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9101758599281311,
      "learning_rate": 8.875545851528385e-05,
      "loss": 1.1316,
      "step": 13600
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9287114143371582,
      "learning_rate": 8.874706079946256e-05,
      "loss": 1.0877,
      "step": 13610
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9042177796363831,
      "learning_rate": 8.873866308364125e-05,
      "loss": 1.1398,
      "step": 13620
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9239535331726074,
      "learning_rate": 8.873026536781995e-05,
      "loss": 1.137,
      "step": 13630
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9701600670814514,
      "learning_rate": 8.872186765199866e-05,
      "loss": 1.1118,
      "step": 13640
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9426175951957703,
      "learning_rate": 8.871346993617736e-05,
      "loss": 1.1129,
      "step": 13650
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8625684380531311,
      "learning_rate": 8.870507222035606e-05,
      "loss": 1.1198,
      "step": 13660
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9156553149223328,
      "learning_rate": 8.869667450453476e-05,
      "loss": 1.0969,
      "step": 13670
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9108346700668335,
      "learning_rate": 8.868827678871348e-05,
      "loss": 1.0853,
      "step": 13680
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9348630309104919,
      "learning_rate": 8.867987907289218e-05,
      "loss": 1.1061,
      "step": 13690
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9263437390327454,
      "learning_rate": 8.867148135707088e-05,
      "loss": 1.0909,
      "step": 13700
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9103613495826721,
      "learning_rate": 8.86630836412496e-05,
      "loss": 1.1466,
      "step": 13710
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8632253408432007,
      "learning_rate": 8.865468592542828e-05,
      "loss": 1.1456,
      "step": 13720
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9509271383285522,
      "learning_rate": 8.864628820960698e-05,
      "loss": 1.1197,
      "step": 13730
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9113433957099915,
      "learning_rate": 8.86378904937857e-05,
      "loss": 1.114,
      "step": 13740
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8760579824447632,
      "learning_rate": 8.86294927779644e-05,
      "loss": 1.1264,
      "step": 13750
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8918175101280212,
      "learning_rate": 8.86210950621431e-05,
      "loss": 1.1495,
      "step": 13760
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8605629205703735,
      "learning_rate": 8.86126973463218e-05,
      "loss": 1.1184,
      "step": 13770
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9115935564041138,
      "learning_rate": 8.860429963050051e-05,
      "loss": 1.1158,
      "step": 13780
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8542095422744751,
      "learning_rate": 8.859590191467921e-05,
      "loss": 1.0958,
      "step": 13790
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9421138763427734,
      "learning_rate": 8.858750419885791e-05,
      "loss": 1.1353,
      "step": 13800
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9435168504714966,
      "learning_rate": 8.857910648303663e-05,
      "loss": 1.1387,
      "step": 13810
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8814151883125305,
      "learning_rate": 8.857070876721533e-05,
      "loss": 1.1259,
      "step": 13820
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9127605557441711,
      "learning_rate": 8.856231105139402e-05,
      "loss": 1.1021,
      "step": 13830
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9513577222824097,
      "learning_rate": 8.855391333557273e-05,
      "loss": 1.1498,
      "step": 13840
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.018366813659668,
      "learning_rate": 8.854551561975143e-05,
      "loss": 1.1185,
      "step": 13850
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0120519399642944,
      "learning_rate": 8.853711790393013e-05,
      "loss": 1.1099,
      "step": 13860
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9408965110778809,
      "learning_rate": 8.852872018810883e-05,
      "loss": 1.13,
      "step": 13870
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8699688911437988,
      "learning_rate": 8.852032247228755e-05,
      "loss": 1.1464,
      "step": 13880
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9382355809211731,
      "learning_rate": 8.851192475646625e-05,
      "loss": 1.126,
      "step": 13890
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9635114669799805,
      "learning_rate": 8.850352704064495e-05,
      "loss": 1.1163,
      "step": 13900
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9410046935081482,
      "learning_rate": 8.849512932482365e-05,
      "loss": 1.1236,
      "step": 13910
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9217486381530762,
      "learning_rate": 8.848673160900236e-05,
      "loss": 1.1144,
      "step": 13920
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.874560534954071,
      "learning_rate": 8.847833389318105e-05,
      "loss": 1.1266,
      "step": 13930
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8844665884971619,
      "learning_rate": 8.846993617735976e-05,
      "loss": 1.1049,
      "step": 13940
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9224509000778198,
      "learning_rate": 8.846153846153847e-05,
      "loss": 1.1361,
      "step": 13950
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9551849961280823,
      "learning_rate": 8.845314074571717e-05,
      "loss": 1.1304,
      "step": 13960
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9024940729141235,
      "learning_rate": 8.844474302989587e-05,
      "loss": 1.0848,
      "step": 13970
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9649220108985901,
      "learning_rate": 8.843634531407458e-05,
      "loss": 1.1057,
      "step": 13980
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.920486330986023,
      "learning_rate": 8.842794759825328e-05,
      "loss": 1.1071,
      "step": 13990
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8891380429267883,
      "learning_rate": 8.841954988243198e-05,
      "loss": 1.11,
      "step": 14000
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9146546125411987,
      "learning_rate": 8.841115216661068e-05,
      "loss": 1.1002,
      "step": 14010
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9723424911499023,
      "learning_rate": 8.84027544507894e-05,
      "loss": 1.1125,
      "step": 14020
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9875556826591492,
      "learning_rate": 8.83943567349681e-05,
      "loss": 1.1088,
      "step": 14030
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9426760673522949,
      "learning_rate": 8.838595901914679e-05,
      "loss": 1.1558,
      "step": 14040
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9075933694839478,
      "learning_rate": 8.83775613033255e-05,
      "loss": 1.0947,
      "step": 14050
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8498238921165466,
      "learning_rate": 8.83691635875042e-05,
      "loss": 1.121,
      "step": 14060
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9222276210784912,
      "learning_rate": 8.83607658716829e-05,
      "loss": 1.1325,
      "step": 14070
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8715246319770813,
      "learning_rate": 8.835236815586161e-05,
      "loss": 1.1274,
      "step": 14080
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8748525381088257,
      "learning_rate": 8.834397044004032e-05,
      "loss": 1.1333,
      "step": 14090
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8961807489395142,
      "learning_rate": 8.833557272421902e-05,
      "loss": 1.1172,
      "step": 14100
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9251819252967834,
      "learning_rate": 8.832717500839772e-05,
      "loss": 1.0832,
      "step": 14110
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8879318833351135,
      "learning_rate": 8.831877729257643e-05,
      "loss": 1.1097,
      "step": 14120
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9097147583961487,
      "learning_rate": 8.831037957675513e-05,
      "loss": 1.1073,
      "step": 14130
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9001836776733398,
      "learning_rate": 8.830198186093383e-05,
      "loss": 1.1066,
      "step": 14140
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8732591867446899,
      "learning_rate": 8.829358414511253e-05,
      "loss": 1.1208,
      "step": 14150
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8660206198692322,
      "learning_rate": 8.828518642929123e-05,
      "loss": 1.1289,
      "step": 14160
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8846068978309631,
      "learning_rate": 8.827678871346993e-05,
      "loss": 1.1122,
      "step": 14170
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9000621438026428,
      "learning_rate": 8.826839099764865e-05,
      "loss": 1.1296,
      "step": 14180
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9542943239212036,
      "learning_rate": 8.825999328182735e-05,
      "loss": 1.1128,
      "step": 14190
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.982465386390686,
      "learning_rate": 8.825159556600605e-05,
      "loss": 1.1104,
      "step": 14200
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8780309557914734,
      "learning_rate": 8.824319785018475e-05,
      "loss": 1.1371,
      "step": 14210
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9983294606208801,
      "learning_rate": 8.823480013436347e-05,
      "loss": 1.1467,
      "step": 14220
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9325399398803711,
      "learning_rate": 8.822640241854217e-05,
      "loss": 1.1222,
      "step": 14230
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9618887901306152,
      "learning_rate": 8.821800470272087e-05,
      "loss": 1.1368,
      "step": 14240
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9248318076133728,
      "learning_rate": 8.820960698689957e-05,
      "loss": 1.117,
      "step": 14250
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9239673614501953,
      "learning_rate": 8.820120927107827e-05,
      "loss": 1.1189,
      "step": 14260
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9307476878166199,
      "learning_rate": 8.819281155525697e-05,
      "loss": 1.1231,
      "step": 14270
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.920782208442688,
      "learning_rate": 8.818441383943567e-05,
      "loss": 1.1253,
      "step": 14280
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.90324467420578,
      "learning_rate": 8.817601612361438e-05,
      "loss": 1.1256,
      "step": 14290
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9153021574020386,
      "learning_rate": 8.816761840779308e-05,
      "loss": 1.1112,
      "step": 14300
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9251894354820251,
      "learning_rate": 8.815922069197178e-05,
      "loss": 1.1134,
      "step": 14310
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9242687821388245,
      "learning_rate": 8.81508229761505e-05,
      "loss": 1.109,
      "step": 14320
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0389292240142822,
      "learning_rate": 8.81424252603292e-05,
      "loss": 1.1053,
      "step": 14330
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8664111495018005,
      "learning_rate": 8.81340275445079e-05,
      "loss": 1.1138,
      "step": 14340
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.934990644454956,
      "learning_rate": 8.81256298286866e-05,
      "loss": 1.1246,
      "step": 14350
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9289098978042603,
      "learning_rate": 8.81172321128653e-05,
      "loss": 1.1209,
      "step": 14360
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9563125371932983,
      "learning_rate": 8.8108834397044e-05,
      "loss": 1.131,
      "step": 14370
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9024096727371216,
      "learning_rate": 8.81004366812227e-05,
      "loss": 1.1249,
      "step": 14380
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.93302983045578,
      "learning_rate": 8.809203896540142e-05,
      "loss": 1.1093,
      "step": 14390
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9166001677513123,
      "learning_rate": 8.808364124958012e-05,
      "loss": 1.1162,
      "step": 14400
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.015295147895813,
      "learning_rate": 8.807524353375882e-05,
      "loss": 1.123,
      "step": 14410
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9155243039131165,
      "learning_rate": 8.806684581793753e-05,
      "loss": 1.1173,
      "step": 14420
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9881671071052551,
      "learning_rate": 8.805844810211623e-05,
      "loss": 1.1129,
      "step": 14430
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9244381189346313,
      "learning_rate": 8.805005038629493e-05,
      "loss": 1.1483,
      "step": 14440
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8698108792304993,
      "learning_rate": 8.804165267047364e-05,
      "loss": 1.1186,
      "step": 14450
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9278691411018372,
      "learning_rate": 8.803325495465234e-05,
      "loss": 1.1311,
      "step": 14460
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9126485586166382,
      "learning_rate": 8.802485723883104e-05,
      "loss": 1.1314,
      "step": 14470
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9807928204536438,
      "learning_rate": 8.801645952300974e-05,
      "loss": 1.1336,
      "step": 14480
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0050467252731323,
      "learning_rate": 8.800806180718845e-05,
      "loss": 1.1156,
      "step": 14490
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8595203757286072,
      "learning_rate": 8.799966409136715e-05,
      "loss": 1.1032,
      "step": 14500
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9575943350791931,
      "learning_rate": 8.799126637554585e-05,
      "loss": 1.1299,
      "step": 14510
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9149112105369568,
      "learning_rate": 8.798286865972455e-05,
      "loss": 1.1207,
      "step": 14520
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9616824388504028,
      "learning_rate": 8.797447094390327e-05,
      "loss": 1.1279,
      "step": 14530
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9843555092811584,
      "learning_rate": 8.796607322808197e-05,
      "loss": 1.1254,
      "step": 14540
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9238868951797485,
      "learning_rate": 8.795767551226067e-05,
      "loss": 1.1162,
      "step": 14550
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9549159407615662,
      "learning_rate": 8.794927779643938e-05,
      "loss": 1.1256,
      "step": 14560
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9398096799850464,
      "learning_rate": 8.794088008061807e-05,
      "loss": 1.1107,
      "step": 14570
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9384233951568604,
      "learning_rate": 8.793248236479677e-05,
      "loss": 1.1136,
      "step": 14580
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9121764898300171,
      "learning_rate": 8.792408464897549e-05,
      "loss": 1.1168,
      "step": 14590
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9835638999938965,
      "learning_rate": 8.791568693315419e-05,
      "loss": 1.1335,
      "step": 14600
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.090331792831421,
      "learning_rate": 8.790728921733289e-05,
      "loss": 1.085,
      "step": 14610
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9265365600585938,
      "learning_rate": 8.789889150151159e-05,
      "loss": 1.0982,
      "step": 14620
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8829748034477234,
      "learning_rate": 8.78904937856903e-05,
      "loss": 1.1142,
      "step": 14630
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9203078746795654,
      "learning_rate": 8.7882096069869e-05,
      "loss": 1.1174,
      "step": 14640
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.954437255859375,
      "learning_rate": 8.78736983540477e-05,
      "loss": 1.1255,
      "step": 14650
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9201489090919495,
      "learning_rate": 8.786530063822642e-05,
      "loss": 1.1278,
      "step": 14660
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9098379015922546,
      "learning_rate": 8.785690292240512e-05,
      "loss": 1.1045,
      "step": 14670
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9808036088943481,
      "learning_rate": 8.78485052065838e-05,
      "loss": 1.1123,
      "step": 14680
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9970177412033081,
      "learning_rate": 8.784010749076252e-05,
      "loss": 1.1244,
      "step": 14690
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9139359593391418,
      "learning_rate": 8.783170977494122e-05,
      "loss": 1.1054,
      "step": 14700
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9273451566696167,
      "learning_rate": 8.782331205911992e-05,
      "loss": 1.0995,
      "step": 14710
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.91801518201828,
      "learning_rate": 8.781491434329862e-05,
      "loss": 1.1142,
      "step": 14720
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9866279363632202,
      "learning_rate": 8.780651662747734e-05,
      "loss": 1.1147,
      "step": 14730
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9117400646209717,
      "learning_rate": 8.779811891165604e-05,
      "loss": 1.1107,
      "step": 14740
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9210174083709717,
      "learning_rate": 8.778972119583474e-05,
      "loss": 1.1188,
      "step": 14750
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9606226086616516,
      "learning_rate": 8.778132348001344e-05,
      "loss": 1.1036,
      "step": 14760
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9500154852867126,
      "learning_rate": 8.777292576419215e-05,
      "loss": 1.0973,
      "step": 14770
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9500105381011963,
      "learning_rate": 8.776452804837084e-05,
      "loss": 1.0983,
      "step": 14780
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8843684792518616,
      "learning_rate": 8.775613033254954e-05,
      "loss": 1.106,
      "step": 14790
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9123238921165466,
      "learning_rate": 8.774773261672825e-05,
      "loss": 1.1233,
      "step": 14800
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9631816744804382,
      "learning_rate": 8.773933490090695e-05,
      "loss": 1.1282,
      "step": 14810
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9595421552658081,
      "learning_rate": 8.773093718508566e-05,
      "loss": 1.1034,
      "step": 14820
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0419737100601196,
      "learning_rate": 8.772253946926437e-05,
      "loss": 1.1325,
      "step": 14830
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9546859860420227,
      "learning_rate": 8.771414175344307e-05,
      "loss": 1.1095,
      "step": 14840
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9662930965423584,
      "learning_rate": 8.770574403762177e-05,
      "loss": 1.1339,
      "step": 14850
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9654275178909302,
      "learning_rate": 8.769734632180047e-05,
      "loss": 1.114,
      "step": 14860
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0001049041748047,
      "learning_rate": 8.768894860597919e-05,
      "loss": 1.1116,
      "step": 14870
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9047713279724121,
      "learning_rate": 8.768055089015789e-05,
      "loss": 1.1283,
      "step": 14880
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8955701589584351,
      "learning_rate": 8.767215317433657e-05,
      "loss": 1.115,
      "step": 14890
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9277672171592712,
      "learning_rate": 8.766375545851529e-05,
      "loss": 1.0798,
      "step": 14900
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8995895981788635,
      "learning_rate": 8.765535774269399e-05,
      "loss": 1.0962,
      "step": 14910
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9734140038490295,
      "learning_rate": 8.764696002687269e-05,
      "loss": 1.1177,
      "step": 14920
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.007621169090271,
      "learning_rate": 8.76385623110514e-05,
      "loss": 1.0963,
      "step": 14930
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9346214532852173,
      "learning_rate": 8.76301645952301e-05,
      "loss": 1.1207,
      "step": 14940
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8714445233345032,
      "learning_rate": 8.76217668794088e-05,
      "loss": 1.1019,
      "step": 14950
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9773920774459839,
      "learning_rate": 8.76133691635875e-05,
      "loss": 1.1345,
      "step": 14960
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.962600588798523,
      "learning_rate": 8.760497144776622e-05,
      "loss": 1.1099,
      "step": 14970
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9157037138938904,
      "learning_rate": 8.759657373194492e-05,
      "loss": 1.1178,
      "step": 14980
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9327508211135864,
      "learning_rate": 8.758817601612361e-05,
      "loss": 1.1125,
      "step": 14990
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9027820825576782,
      "learning_rate": 8.757977830030232e-05,
      "loss": 1.1151,
      "step": 15000
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9330666065216064,
      "learning_rate": 8.757138058448102e-05,
      "loss": 1.1033,
      "step": 15010
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9602764844894409,
      "learning_rate": 8.756298286865972e-05,
      "loss": 1.1159,
      "step": 15020
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9144401550292969,
      "learning_rate": 8.755458515283842e-05,
      "loss": 1.1232,
      "step": 15030
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.928410530090332,
      "learning_rate": 8.754618743701714e-05,
      "loss": 1.1226,
      "step": 15040
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9314349293708801,
      "learning_rate": 8.753778972119584e-05,
      "loss": 1.1214,
      "step": 15050
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.908319890499115,
      "learning_rate": 8.752939200537454e-05,
      "loss": 1.122,
      "step": 15060
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.936560869216919,
      "learning_rate": 8.752099428955325e-05,
      "loss": 1.1281,
      "step": 15070
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0261152982711792,
      "learning_rate": 8.751259657373195e-05,
      "loss": 1.1326,
      "step": 15080
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9165915846824646,
      "learning_rate": 8.750419885791066e-05,
      "loss": 1.115,
      "step": 15090
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9344234466552734,
      "learning_rate": 8.749580114208936e-05,
      "loss": 1.1175,
      "step": 15100
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8795295357704163,
      "learning_rate": 8.748740342626806e-05,
      "loss": 1.1173,
      "step": 15110
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9428615570068359,
      "learning_rate": 8.747900571044676e-05,
      "loss": 1.1058,
      "step": 15120
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9324769377708435,
      "learning_rate": 8.747060799462546e-05,
      "loss": 1.1184,
      "step": 15130
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8626567125320435,
      "learning_rate": 8.746221027880417e-05,
      "loss": 1.0974,
      "step": 15140
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9009515643119812,
      "learning_rate": 8.745381256298287e-05,
      "loss": 1.1054,
      "step": 15150
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9502636194229126,
      "learning_rate": 8.744541484716157e-05,
      "loss": 1.1219,
      "step": 15160
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9786444902420044,
      "learning_rate": 8.743701713134029e-05,
      "loss": 1.1044,
      "step": 15170
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9187127351760864,
      "learning_rate": 8.742861941551899e-05,
      "loss": 1.1318,
      "step": 15180
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.924630880355835,
      "learning_rate": 8.742022169969769e-05,
      "loss": 1.1403,
      "step": 15190
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9214105010032654,
      "learning_rate": 8.741182398387639e-05,
      "loss": 1.1046,
      "step": 15200
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9642462134361267,
      "learning_rate": 8.740342626805509e-05,
      "loss": 1.1336,
      "step": 15210
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9122458696365356,
      "learning_rate": 8.739502855223379e-05,
      "loss": 1.1193,
      "step": 15220
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9484336972236633,
      "learning_rate": 8.738663083641249e-05,
      "loss": 1.1147,
      "step": 15230
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9583941102027893,
      "learning_rate": 8.73782331205912e-05,
      "loss": 1.1261,
      "step": 15240
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9006879329681396,
      "learning_rate": 8.736983540476991e-05,
      "loss": 1.1185,
      "step": 15250
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9025834202766418,
      "learning_rate": 8.736143768894861e-05,
      "loss": 1.1307,
      "step": 15260
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9103566408157349,
      "learning_rate": 8.735303997312731e-05,
      "loss": 1.1085,
      "step": 15270
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9725592732429504,
      "learning_rate": 8.734464225730602e-05,
      "loss": 1.1219,
      "step": 15280
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9697563648223877,
      "learning_rate": 8.733624454148472e-05,
      "loss": 1.1152,
      "step": 15290
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8665038347244263,
      "learning_rate": 8.732784682566342e-05,
      "loss": 1.1247,
      "step": 15300
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8986729383468628,
      "learning_rate": 8.731944910984212e-05,
      "loss": 1.1243,
      "step": 15310
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9079999923706055,
      "learning_rate": 8.731105139402083e-05,
      "loss": 1.1162,
      "step": 15320
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9079917669296265,
      "learning_rate": 8.730265367819953e-05,
      "loss": 1.1044,
      "step": 15330
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9908376336097717,
      "learning_rate": 8.729425596237824e-05,
      "loss": 1.1112,
      "step": 15340
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8806149363517761,
      "learning_rate": 8.728585824655694e-05,
      "loss": 1.112,
      "step": 15350
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9354682564735413,
      "learning_rate": 8.727746053073564e-05,
      "loss": 1.1388,
      "step": 15360
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0563576221466064,
      "learning_rate": 8.726906281491434e-05,
      "loss": 1.1027,
      "step": 15370
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9495493173599243,
      "learning_rate": 8.726066509909306e-05,
      "loss": 1.131,
      "step": 15380
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8986271619796753,
      "learning_rate": 8.725226738327176e-05,
      "loss": 1.114,
      "step": 15390
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.924021303653717,
      "learning_rate": 8.724386966745046e-05,
      "loss": 1.093,
      "step": 15400
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9179587364196777,
      "learning_rate": 8.723547195162917e-05,
      "loss": 1.1203,
      "step": 15410
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.975875198841095,
      "learning_rate": 8.722707423580786e-05,
      "loss": 1.1177,
      "step": 15420
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9055646061897278,
      "learning_rate": 8.721867651998656e-05,
      "loss": 1.1213,
      "step": 15430
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9086652994155884,
      "learning_rate": 8.721027880416527e-05,
      "loss": 1.1207,
      "step": 15440
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9314917325973511,
      "learning_rate": 8.720188108834398e-05,
      "loss": 1.0941,
      "step": 15450
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9791564345359802,
      "learning_rate": 8.719348337252268e-05,
      "loss": 1.1331,
      "step": 15460
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8971198797225952,
      "learning_rate": 8.718508565670138e-05,
      "loss": 1.1099,
      "step": 15470
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8518558740615845,
      "learning_rate": 8.717668794088009e-05,
      "loss": 1.103,
      "step": 15480
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9296644926071167,
      "learning_rate": 8.716829022505879e-05,
      "loss": 1.1074,
      "step": 15490
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.972202479839325,
      "learning_rate": 8.715989250923749e-05,
      "loss": 1.1089,
      "step": 15500
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9175410866737366,
      "learning_rate": 8.715149479341619e-05,
      "loss": 1.1179,
      "step": 15510
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8872547149658203,
      "learning_rate": 8.71430970775949e-05,
      "loss": 1.1153,
      "step": 15520
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9035467505455017,
      "learning_rate": 8.71346993617736e-05,
      "loss": 1.1164,
      "step": 15530
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9036070108413696,
      "learning_rate": 8.712630164595231e-05,
      "loss": 1.103,
      "step": 15540
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9524517059326172,
      "learning_rate": 8.711790393013101e-05,
      "loss": 1.1093,
      "step": 15550
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8619802594184875,
      "learning_rate": 8.710950621430971e-05,
      "loss": 1.1252,
      "step": 15560
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9018606543540955,
      "learning_rate": 8.710110849848841e-05,
      "loss": 1.1197,
      "step": 15570
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9069834351539612,
      "learning_rate": 8.709271078266712e-05,
      "loss": 1.1097,
      "step": 15580
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8745734691619873,
      "learning_rate": 8.708431306684583e-05,
      "loss": 1.1051,
      "step": 15590
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0042290687561035,
      "learning_rate": 8.707591535102453e-05,
      "loss": 1.1062,
      "step": 15600
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9338233470916748,
      "learning_rate": 8.706751763520323e-05,
      "loss": 1.1223,
      "step": 15610
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8764388561248779,
      "learning_rate": 8.705911991938194e-05,
      "loss": 1.1312,
      "step": 15620
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9034644365310669,
      "learning_rate": 8.705072220356063e-05,
      "loss": 1.1199,
      "step": 15630
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9459267854690552,
      "learning_rate": 8.704232448773933e-05,
      "loss": 1.1005,
      "step": 15640
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9199139475822449,
      "learning_rate": 8.703392677191804e-05,
      "loss": 1.1113,
      "step": 15650
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9240623712539673,
      "learning_rate": 8.702552905609674e-05,
      "loss": 1.1226,
      "step": 15660
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9151225686073303,
      "learning_rate": 8.701713134027544e-05,
      "loss": 1.1148,
      "step": 15670
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.002893328666687,
      "learning_rate": 8.700873362445416e-05,
      "loss": 1.0963,
      "step": 15680
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9323188662528992,
      "learning_rate": 8.700033590863286e-05,
      "loss": 1.0792,
      "step": 15690
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9247946739196777,
      "learning_rate": 8.699193819281156e-05,
      "loss": 1.1055,
      "step": 15700
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9568535685539246,
      "learning_rate": 8.698354047699026e-05,
      "loss": 1.0878,
      "step": 15710
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9059244990348816,
      "learning_rate": 8.697514276116898e-05,
      "loss": 1.1345,
      "step": 15720
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9476732611656189,
      "learning_rate": 8.696674504534768e-05,
      "loss": 1.0899,
      "step": 15730
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9337188601493835,
      "learning_rate": 8.695834732952636e-05,
      "loss": 1.1426,
      "step": 15740
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8653301000595093,
      "learning_rate": 8.694994961370508e-05,
      "loss": 1.1234,
      "step": 15750
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8637192845344543,
      "learning_rate": 8.694155189788378e-05,
      "loss": 1.1005,
      "step": 15760
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9008951187133789,
      "learning_rate": 8.693315418206248e-05,
      "loss": 1.1104,
      "step": 15770
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9144231081008911,
      "learning_rate": 8.692475646624119e-05,
      "loss": 1.0905,
      "step": 15780
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.952981173992157,
      "learning_rate": 8.69163587504199e-05,
      "loss": 1.1172,
      "step": 15790
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.938640296459198,
      "learning_rate": 8.69079610345986e-05,
      "loss": 1.1205,
      "step": 15800
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9345549941062927,
      "learning_rate": 8.68995633187773e-05,
      "loss": 1.1082,
      "step": 15810
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8561148643493652,
      "learning_rate": 8.689116560295601e-05,
      "loss": 1.0963,
      "step": 15820
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8445508480072021,
      "learning_rate": 8.688276788713471e-05,
      "loss": 1.1132,
      "step": 15830
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9664802551269531,
      "learning_rate": 8.68743701713134e-05,
      "loss": 1.0823,
      "step": 15840
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9483756422996521,
      "learning_rate": 8.686597245549211e-05,
      "loss": 1.0965,
      "step": 15850
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0393468141555786,
      "learning_rate": 8.685757473967081e-05,
      "loss": 1.0732,
      "step": 15860
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9285652041435242,
      "learning_rate": 8.684917702384951e-05,
      "loss": 1.1245,
      "step": 15870
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.907402515411377,
      "learning_rate": 8.684077930802821e-05,
      "loss": 1.1197,
      "step": 15880
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9397310614585876,
      "learning_rate": 8.683238159220693e-05,
      "loss": 1.106,
      "step": 15890
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9912298321723938,
      "learning_rate": 8.682398387638563e-05,
      "loss": 1.1106,
      "step": 15900
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8773438334465027,
      "learning_rate": 8.681558616056433e-05,
      "loss": 1.1183,
      "step": 15910
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8921714425086975,
      "learning_rate": 8.680718844474304e-05,
      "loss": 1.1019,
      "step": 15920
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9610801339149475,
      "learning_rate": 8.679879072892174e-05,
      "loss": 1.1106,
      "step": 15930
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8875612020492554,
      "learning_rate": 8.679039301310044e-05,
      "loss": 1.1122,
      "step": 15940
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.925846517086029,
      "learning_rate": 8.678199529727915e-05,
      "loss": 1.1097,
      "step": 15950
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9786810278892517,
      "learning_rate": 8.677359758145785e-05,
      "loss": 1.1074,
      "step": 15960
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9146912693977356,
      "learning_rate": 8.676519986563655e-05,
      "loss": 1.1139,
      "step": 15970
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8765195608139038,
      "learning_rate": 8.675680214981525e-05,
      "loss": 1.1031,
      "step": 15980
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9373301267623901,
      "learning_rate": 8.674840443399396e-05,
      "loss": 1.1142,
      "step": 15990
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9163397550582886,
      "learning_rate": 8.674000671817266e-05,
      "loss": 1.1289,
      "step": 16000
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9457612633705139,
      "learning_rate": 8.673160900235136e-05,
      "loss": 1.1219,
      "step": 16010
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9161149859428406,
      "learning_rate": 8.672321128653008e-05,
      "loss": 1.1418,
      "step": 16020
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9269108176231384,
      "learning_rate": 8.671481357070878e-05,
      "loss": 1.1175,
      "step": 16030
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9797216653823853,
      "learning_rate": 8.670641585488748e-05,
      "loss": 1.0838,
      "step": 16040
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9082381725311279,
      "learning_rate": 8.669801813906618e-05,
      "loss": 1.1207,
      "step": 16050
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9488618969917297,
      "learning_rate": 8.668962042324488e-05,
      "loss": 1.0971,
      "step": 16060
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8994138836860657,
      "learning_rate": 8.668122270742358e-05,
      "loss": 1.1055,
      "step": 16070
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9246254563331604,
      "learning_rate": 8.667282499160228e-05,
      "loss": 1.1319,
      "step": 16080
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8879727125167847,
      "learning_rate": 8.6664427275781e-05,
      "loss": 1.1192,
      "step": 16090
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9572038650512695,
      "learning_rate": 8.66560295599597e-05,
      "loss": 1.1088,
      "step": 16100
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9084638357162476,
      "learning_rate": 8.66476318441384e-05,
      "loss": 1.0985,
      "step": 16110
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9260880351066589,
      "learning_rate": 8.66392341283171e-05,
      "loss": 1.0889,
      "step": 16120
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0169014930725098,
      "learning_rate": 8.663083641249581e-05,
      "loss": 1.1212,
      "step": 16130
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9938318729400635,
      "learning_rate": 8.662243869667451e-05,
      "loss": 1.1113,
      "step": 16140
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.01992928981781,
      "learning_rate": 8.661404098085321e-05,
      "loss": 1.1201,
      "step": 16150
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9507731199264526,
      "learning_rate": 8.660564326503191e-05,
      "loss": 1.0985,
      "step": 16160
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.938035249710083,
      "learning_rate": 8.659724554921061e-05,
      "loss": 1.0891,
      "step": 16170
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9544059038162231,
      "learning_rate": 8.658884783338932e-05,
      "loss": 1.1158,
      "step": 16180
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9504319429397583,
      "learning_rate": 8.658045011756803e-05,
      "loss": 1.1165,
      "step": 16190
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9170682430267334,
      "learning_rate": 8.657205240174673e-05,
      "loss": 1.1279,
      "step": 16200
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.878187358379364,
      "learning_rate": 8.656365468592543e-05,
      "loss": 1.1164,
      "step": 16210
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0140724182128906,
      "learning_rate": 8.655525697010413e-05,
      "loss": 1.0925,
      "step": 16220
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9395869970321655,
      "learning_rate": 8.654685925428285e-05,
      "loss": 1.096,
      "step": 16230
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9655680060386658,
      "learning_rate": 8.653846153846155e-05,
      "loss": 1.1219,
      "step": 16240
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9558068513870239,
      "learning_rate": 8.653006382264025e-05,
      "loss": 1.1072,
      "step": 16250
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8891133069992065,
      "learning_rate": 8.652166610681896e-05,
      "loss": 1.1004,
      "step": 16260
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8977444767951965,
      "learning_rate": 8.651326839099765e-05,
      "loss": 1.1175,
      "step": 16270
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9006922245025635,
      "learning_rate": 8.650487067517635e-05,
      "loss": 1.0965,
      "step": 16280
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9141904711723328,
      "learning_rate": 8.649647295935506e-05,
      "loss": 1.1182,
      "step": 16290
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9052436351776123,
      "learning_rate": 8.648807524353376e-05,
      "loss": 1.1037,
      "step": 16300
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8963584303855896,
      "learning_rate": 8.647967752771246e-05,
      "loss": 1.0736,
      "step": 16310
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.895529568195343,
      "learning_rate": 8.647127981189117e-05,
      "loss": 1.1201,
      "step": 16320
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8678609728813171,
      "learning_rate": 8.646288209606988e-05,
      "loss": 1.1195,
      "step": 16330
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9141554832458496,
      "learning_rate": 8.645448438024858e-05,
      "loss": 1.0943,
      "step": 16340
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9165868759155273,
      "learning_rate": 8.644608666442728e-05,
      "loss": 1.1025,
      "step": 16350
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9144678711891174,
      "learning_rate": 8.643768894860598e-05,
      "loss": 1.0986,
      "step": 16360
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8954872488975525,
      "learning_rate": 8.642929123278468e-05,
      "loss": 1.1016,
      "step": 16370
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8864966034889221,
      "learning_rate": 8.642089351696338e-05,
      "loss": 1.1244,
      "step": 16380
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9300432801246643,
      "learning_rate": 8.641249580114208e-05,
      "loss": 1.0974,
      "step": 16390
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9392675757408142,
      "learning_rate": 8.64040980853208e-05,
      "loss": 1.0768,
      "step": 16400
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9090225696563721,
      "learning_rate": 8.63957003694995e-05,
      "loss": 1.1425,
      "step": 16410
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8820337057113647,
      "learning_rate": 8.63873026536782e-05,
      "loss": 1.117,
      "step": 16420
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9691148400306702,
      "learning_rate": 8.637890493785691e-05,
      "loss": 1.1105,
      "step": 16430
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9114964008331299,
      "learning_rate": 8.637050722203561e-05,
      "loss": 1.1094,
      "step": 16440
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9060763716697693,
      "learning_rate": 8.636210950621432e-05,
      "loss": 1.1227,
      "step": 16450
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9717005491256714,
      "learning_rate": 8.635371179039302e-05,
      "loss": 1.1099,
      "step": 16460
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.926828920841217,
      "learning_rate": 8.634531407457173e-05,
      "loss": 1.0955,
      "step": 16470
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.944095253944397,
      "learning_rate": 8.633691635875042e-05,
      "loss": 1.0987,
      "step": 16480
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9036884903907776,
      "learning_rate": 8.632851864292912e-05,
      "loss": 1.0847,
      "step": 16490
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.936955451965332,
      "learning_rate": 8.632012092710783e-05,
      "loss": 1.0932,
      "step": 16500
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9809815287590027,
      "learning_rate": 8.631172321128653e-05,
      "loss": 1.1117,
      "step": 16510
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.871856689453125,
      "learning_rate": 8.630332549546523e-05,
      "loss": 1.0907,
      "step": 16520
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9420427680015564,
      "learning_rate": 8.629492777964395e-05,
      "loss": 1.1084,
      "step": 16530
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9744893312454224,
      "learning_rate": 8.628653006382265e-05,
      "loss": 1.1061,
      "step": 16540
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9527720808982849,
      "learning_rate": 8.627813234800135e-05,
      "loss": 1.102,
      "step": 16550
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9135245084762573,
      "learning_rate": 8.626973463218005e-05,
      "loss": 1.1257,
      "step": 16560
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9639981985092163,
      "learning_rate": 8.626133691635876e-05,
      "loss": 1.0942,
      "step": 16570
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9305285215377808,
      "learning_rate": 8.625293920053745e-05,
      "loss": 1.0832,
      "step": 16580
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9467176198959351,
      "learning_rate": 8.624454148471615e-05,
      "loss": 1.1014,
      "step": 16590
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9170607328414917,
      "learning_rate": 8.623614376889487e-05,
      "loss": 1.1042,
      "step": 16600
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8990239500999451,
      "learning_rate": 8.622774605307357e-05,
      "loss": 1.0938,
      "step": 16610
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.919737696647644,
      "learning_rate": 8.621934833725227e-05,
      "loss": 1.1256,
      "step": 16620
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.927342414855957,
      "learning_rate": 8.621095062143097e-05,
      "loss": 1.0588,
      "step": 16630
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9596610069274902,
      "learning_rate": 8.620255290560968e-05,
      "loss": 1.1139,
      "step": 16640
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9167120456695557,
      "learning_rate": 8.619415518978838e-05,
      "loss": 1.1182,
      "step": 16650
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9366847276687622,
      "learning_rate": 8.618575747396708e-05,
      "loss": 1.1034,
      "step": 16660
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9725133180618286,
      "learning_rate": 8.61773597581458e-05,
      "loss": 1.1178,
      "step": 16670
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9172531366348267,
      "learning_rate": 8.61689620423245e-05,
      "loss": 1.0879,
      "step": 16680
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8928595781326294,
      "learning_rate": 8.616056432650319e-05,
      "loss": 1.0983,
      "step": 16690
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9851216673851013,
      "learning_rate": 8.61521666106819e-05,
      "loss": 1.1089,
      "step": 16700
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9329010844230652,
      "learning_rate": 8.61437688948606e-05,
      "loss": 1.1232,
      "step": 16710
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8544999957084656,
      "learning_rate": 8.61353711790393e-05,
      "loss": 1.0947,
      "step": 16720
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9257146716117859,
      "learning_rate": 8.6126973463218e-05,
      "loss": 1.1086,
      "step": 16730
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8957710266113281,
      "learning_rate": 8.611857574739672e-05,
      "loss": 1.1104,
      "step": 16740
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9224327206611633,
      "learning_rate": 8.611017803157542e-05,
      "loss": 1.1093,
      "step": 16750
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9331311583518982,
      "learning_rate": 8.610178031575412e-05,
      "loss": 1.1207,
      "step": 16760
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9614216685295105,
      "learning_rate": 8.609338259993283e-05,
      "loss": 1.1143,
      "step": 16770
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.950002133846283,
      "learning_rate": 8.608498488411153e-05,
      "loss": 1.0925,
      "step": 16780
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9124014377593994,
      "learning_rate": 8.607658716829023e-05,
      "loss": 1.0883,
      "step": 16790
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9585002660751343,
      "learning_rate": 8.606818945246893e-05,
      "loss": 1.086,
      "step": 16800
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9788587093353271,
      "learning_rate": 8.605979173664763e-05,
      "loss": 1.0986,
      "step": 16810
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9377402067184448,
      "learning_rate": 8.605139402082634e-05,
      "loss": 1.0789,
      "step": 16820
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9702931642532349,
      "learning_rate": 8.604299630500504e-05,
      "loss": 1.0968,
      "step": 16830
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8989161849021912,
      "learning_rate": 8.603459858918375e-05,
      "loss": 1.1002,
      "step": 16840
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0440198183059692,
      "learning_rate": 8.602620087336245e-05,
      "loss": 1.11,
      "step": 16850
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9456074833869934,
      "learning_rate": 8.601780315754115e-05,
      "loss": 1.1072,
      "step": 16860
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.880145251750946,
      "learning_rate": 8.600940544171985e-05,
      "loss": 1.1017,
      "step": 16870
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9445072412490845,
      "learning_rate": 8.600100772589857e-05,
      "loss": 1.1128,
      "step": 16880
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.869269073009491,
      "learning_rate": 8.599261001007727e-05,
      "loss": 1.1128,
      "step": 16890
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0037399530410767,
      "learning_rate": 8.598421229425595e-05,
      "loss": 1.1128,
      "step": 16900
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9507962465286255,
      "learning_rate": 8.597581457843467e-05,
      "loss": 1.1139,
      "step": 16910
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9215889573097229,
      "learning_rate": 8.596741686261337e-05,
      "loss": 1.1184,
      "step": 16920
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0083012580871582,
      "learning_rate": 8.595901914679207e-05,
      "loss": 1.1049,
      "step": 16930
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9698511958122253,
      "learning_rate": 8.595062143097078e-05,
      "loss": 1.0972,
      "step": 16940
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9426410794258118,
      "learning_rate": 8.594222371514949e-05,
      "loss": 1.126,
      "step": 16950
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8693003058433533,
      "learning_rate": 8.593382599932819e-05,
      "loss": 1.1075,
      "step": 16960
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9229259490966797,
      "learning_rate": 8.592542828350689e-05,
      "loss": 1.1006,
      "step": 16970
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9683804512023926,
      "learning_rate": 8.59170305676856e-05,
      "loss": 1.1131,
      "step": 16980
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9072035551071167,
      "learning_rate": 8.59086328518643e-05,
      "loss": 1.1056,
      "step": 16990
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8945325613021851,
      "learning_rate": 8.5900235136043e-05,
      "loss": 1.1063,
      "step": 17000
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9545913934707642,
      "learning_rate": 8.58918374202217e-05,
      "loss": 1.0916,
      "step": 17010
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9185695052146912,
      "learning_rate": 8.58834397044004e-05,
      "loss": 1.1102,
      "step": 17020
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9400401711463928,
      "learning_rate": 8.58750419885791e-05,
      "loss": 1.0953,
      "step": 17030
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9485997557640076,
      "learning_rate": 8.586664427275782e-05,
      "loss": 1.0975,
      "step": 17040
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9165788292884827,
      "learning_rate": 8.585824655693652e-05,
      "loss": 1.113,
      "step": 17050
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9310771822929382,
      "learning_rate": 8.584984884111522e-05,
      "loss": 1.1241,
      "step": 17060
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8730702996253967,
      "learning_rate": 8.584145112529392e-05,
      "loss": 1.1075,
      "step": 17070
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0038843154907227,
      "learning_rate": 8.583305340947263e-05,
      "loss": 1.0916,
      "step": 17080
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9894881844520569,
      "learning_rate": 8.582465569365134e-05,
      "loss": 1.1008,
      "step": 17090
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9492472410202026,
      "learning_rate": 8.581625797783004e-05,
      "loss": 1.1115,
      "step": 17100
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9209385514259338,
      "learning_rate": 8.580786026200874e-05,
      "loss": 1.0971,
      "step": 17110
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9176464676856995,
      "learning_rate": 8.579946254618744e-05,
      "loss": 1.0947,
      "step": 17120
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.956646740436554,
      "learning_rate": 8.579190460194827e-05,
      "loss": 1.1075,
      "step": 17130
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9240680932998657,
      "learning_rate": 8.578350688612697e-05,
      "loss": 1.1004,
      "step": 17140
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9465555548667908,
      "learning_rate": 8.577510917030567e-05,
      "loss": 1.1027,
      "step": 17150
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9345818161964417,
      "learning_rate": 8.576671145448439e-05,
      "loss": 1.1061,
      "step": 17160
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.91121906042099,
      "learning_rate": 8.575831373866309e-05,
      "loss": 1.1047,
      "step": 17170
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9541527032852173,
      "learning_rate": 8.574991602284179e-05,
      "loss": 1.1144,
      "step": 17180
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.001551866531372,
      "learning_rate": 8.574151830702049e-05,
      "loss": 1.099,
      "step": 17190
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9648950695991516,
      "learning_rate": 8.57331205911992e-05,
      "loss": 1.1263,
      "step": 17200
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9161174893379211,
      "learning_rate": 8.57247228753779e-05,
      "loss": 1.1075,
      "step": 17210
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9444253444671631,
      "learning_rate": 8.57163251595566e-05,
      "loss": 1.1183,
      "step": 17220
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8742684125900269,
      "learning_rate": 8.57079274437353e-05,
      "loss": 1.0996,
      "step": 17230
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8935016393661499,
      "learning_rate": 8.569952972791401e-05,
      "loss": 1.0964,
      "step": 17240
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8797896504402161,
      "learning_rate": 8.569113201209271e-05,
      "loss": 1.1144,
      "step": 17250
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9014408588409424,
      "learning_rate": 8.568273429627142e-05,
      "loss": 1.1027,
      "step": 17260
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9957795143127441,
      "learning_rate": 8.567433658045012e-05,
      "loss": 1.0931,
      "step": 17270
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8871063590049744,
      "learning_rate": 8.566593886462882e-05,
      "loss": 1.1194,
      "step": 17280
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8539029359817505,
      "learning_rate": 8.565754114880752e-05,
      "loss": 1.0869,
      "step": 17290
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8829309940338135,
      "learning_rate": 8.564914343298624e-05,
      "loss": 1.1106,
      "step": 17300
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.938969612121582,
      "learning_rate": 8.564074571716494e-05,
      "loss": 1.1191,
      "step": 17310
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9321015477180481,
      "learning_rate": 8.563234800134364e-05,
      "loss": 1.1278,
      "step": 17320
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9374982714653015,
      "learning_rate": 8.562395028552235e-05,
      "loss": 1.0762,
      "step": 17330
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8815956711769104,
      "learning_rate": 8.561555256970104e-05,
      "loss": 1.0932,
      "step": 17340
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9213575720787048,
      "learning_rate": 8.560715485387974e-05,
      "loss": 1.0962,
      "step": 17350
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0526560544967651,
      "learning_rate": 8.559875713805846e-05,
      "loss": 1.0923,
      "step": 17360
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8528130650520325,
      "learning_rate": 8.559035942223716e-05,
      "loss": 1.1294,
      "step": 17370
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9504371285438538,
      "learning_rate": 8.558196170641586e-05,
      "loss": 1.1106,
      "step": 17380
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0213943719863892,
      "learning_rate": 8.557356399059456e-05,
      "loss": 1.1284,
      "step": 17390
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9347829818725586,
      "learning_rate": 8.556516627477327e-05,
      "loss": 1.0845,
      "step": 17400
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9889435768127441,
      "learning_rate": 8.555676855895197e-05,
      "loss": 1.1067,
      "step": 17410
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9114763140678406,
      "learning_rate": 8.554837084313067e-05,
      "loss": 1.0956,
      "step": 17420
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9614638090133667,
      "learning_rate": 8.553997312730937e-05,
      "loss": 1.0971,
      "step": 17430
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.945780873298645,
      "learning_rate": 8.553157541148807e-05,
      "loss": 1.1041,
      "step": 17440
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9689162373542786,
      "learning_rate": 8.552317769566678e-05,
      "loss": 1.1195,
      "step": 17450
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2564977407455444,
      "learning_rate": 8.551477997984548e-05,
      "loss": 1.0891,
      "step": 17460
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9196128845214844,
      "learning_rate": 8.550638226402419e-05,
      "loss": 1.1039,
      "step": 17470
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8786943554878235,
      "learning_rate": 8.549798454820289e-05,
      "loss": 1.0919,
      "step": 17480
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9473636746406555,
      "learning_rate": 8.548958683238159e-05,
      "loss": 1.095,
      "step": 17490
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.948150098323822,
      "learning_rate": 8.54811891165603e-05,
      "loss": 1.1019,
      "step": 17500
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9033188819885254,
      "learning_rate": 8.547279140073901e-05,
      "loss": 1.0691,
      "step": 17510
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8785542249679565,
      "learning_rate": 8.546439368491771e-05,
      "loss": 1.1045,
      "step": 17520
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8864615559577942,
      "learning_rate": 8.545599596909641e-05,
      "loss": 1.1086,
      "step": 17530
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9623948931694031,
      "learning_rate": 8.544759825327512e-05,
      "loss": 1.0822,
      "step": 17540
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9355566501617432,
      "learning_rate": 8.543920053745381e-05,
      "loss": 1.0975,
      "step": 17550
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9936404228210449,
      "learning_rate": 8.543080282163251e-05,
      "loss": 1.0923,
      "step": 17560
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9029406905174255,
      "learning_rate": 8.542240510581122e-05,
      "loss": 1.1152,
      "step": 17570
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8796647191047668,
      "learning_rate": 8.541400738998993e-05,
      "loss": 1.0763,
      "step": 17580
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9394406676292419,
      "learning_rate": 8.540560967416863e-05,
      "loss": 1.1015,
      "step": 17590
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9513610005378723,
      "learning_rate": 8.539721195834734e-05,
      "loss": 1.0954,
      "step": 17600
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9663611650466919,
      "learning_rate": 8.538881424252604e-05,
      "loss": 1.1162,
      "step": 17610
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.923028826713562,
      "learning_rate": 8.538041652670474e-05,
      "loss": 1.1077,
      "step": 17620
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9535107016563416,
      "learning_rate": 8.537201881088344e-05,
      "loss": 1.1055,
      "step": 17630
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9206113219261169,
      "learning_rate": 8.536362109506216e-05,
      "loss": 1.0738,
      "step": 17640
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9430006742477417,
      "learning_rate": 8.535522337924084e-05,
      "loss": 1.1086,
      "step": 17650
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.946961522102356,
      "learning_rate": 8.534682566341954e-05,
      "loss": 1.1203,
      "step": 17660
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9791433811187744,
      "learning_rate": 8.533842794759826e-05,
      "loss": 1.09,
      "step": 17670
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9791099429130554,
      "learning_rate": 8.533003023177696e-05,
      "loss": 1.0989,
      "step": 17680
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9486845135688782,
      "learning_rate": 8.532163251595566e-05,
      "loss": 1.0983,
      "step": 17690
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8685942888259888,
      "learning_rate": 8.531323480013436e-05,
      "loss": 1.1229,
      "step": 17700
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9693712592124939,
      "learning_rate": 8.530483708431307e-05,
      "loss": 1.1066,
      "step": 17710
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9077468514442444,
      "learning_rate": 8.529643936849178e-05,
      "loss": 1.1226,
      "step": 17720
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8657535910606384,
      "learning_rate": 8.528804165267048e-05,
      "loss": 1.0945,
      "step": 17730
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8869027495384216,
      "learning_rate": 8.527964393684919e-05,
      "loss": 1.0969,
      "step": 17740
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9077990651130676,
      "learning_rate": 8.527124622102789e-05,
      "loss": 1.1164,
      "step": 17750
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8592579364776611,
      "learning_rate": 8.526284850520658e-05,
      "loss": 1.1137,
      "step": 17760
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8895130753517151,
      "learning_rate": 8.525445078938529e-05,
      "loss": 1.0995,
      "step": 17770
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8667956590652466,
      "learning_rate": 8.5246053073564e-05,
      "loss": 1.1068,
      "step": 17780
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8754933476448059,
      "learning_rate": 8.52376553577427e-05,
      "loss": 1.107,
      "step": 17790
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9518774747848511,
      "learning_rate": 8.52292576419214e-05,
      "loss": 1.1096,
      "step": 17800
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9864497780799866,
      "learning_rate": 8.522085992610011e-05,
      "loss": 1.0849,
      "step": 17810
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.912307858467102,
      "learning_rate": 8.521246221027881e-05,
      "loss": 1.0872,
      "step": 17820
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9122206568717957,
      "learning_rate": 8.520406449445751e-05,
      "loss": 1.0971,
      "step": 17830
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8828033804893494,
      "learning_rate": 8.519566677863622e-05,
      "loss": 1.1096,
      "step": 17840
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9641870856285095,
      "learning_rate": 8.518726906281493e-05,
      "loss": 1.0712,
      "step": 17850
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9996892809867859,
      "learning_rate": 8.517887134699363e-05,
      "loss": 1.1109,
      "step": 17860
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8540690541267395,
      "learning_rate": 8.517047363117233e-05,
      "loss": 1.1083,
      "step": 17870
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9290221929550171,
      "learning_rate": 8.516207591535103e-05,
      "loss": 1.1033,
      "step": 17880
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9058725237846375,
      "learning_rate": 8.515367819952973e-05,
      "loss": 1.0865,
      "step": 17890
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9277447462081909,
      "learning_rate": 8.514528048370843e-05,
      "loss": 1.113,
      "step": 17900
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.920447051525116,
      "learning_rate": 8.513688276788714e-05,
      "loss": 1.0831,
      "step": 17910
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9684462547302246,
      "learning_rate": 8.512848505206584e-05,
      "loss": 1.0917,
      "step": 17920
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9636054039001465,
      "learning_rate": 8.512008733624454e-05,
      "loss": 1.0996,
      "step": 17930
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8802708983421326,
      "learning_rate": 8.511168962042324e-05,
      "loss": 1.1095,
      "step": 17940
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9438508749008179,
      "learning_rate": 8.510329190460196e-05,
      "loss": 1.0972,
      "step": 17950
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9617946147918701,
      "learning_rate": 8.509489418878066e-05,
      "loss": 1.1107,
      "step": 17960
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9735690355300903,
      "learning_rate": 8.508649647295936e-05,
      "loss": 1.1005,
      "step": 17970
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9920896291732788,
      "learning_rate": 8.507809875713806e-05,
      "loss": 1.0953,
      "step": 17980
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9597938656806946,
      "learning_rate": 8.506970104131676e-05,
      "loss": 1.1043,
      "step": 17990
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8762233853340149,
      "learning_rate": 8.506130332549546e-05,
      "loss": 1.103,
      "step": 18000
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9935160875320435,
      "learning_rate": 8.505290560967418e-05,
      "loss": 1.1178,
      "step": 18010
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9437522888183594,
      "learning_rate": 8.504450789385288e-05,
      "loss": 1.1281,
      "step": 18020
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9870259761810303,
      "learning_rate": 8.503611017803158e-05,
      "loss": 1.0936,
      "step": 18030
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0250309705734253,
      "learning_rate": 8.502771246221028e-05,
      "loss": 1.0999,
      "step": 18040
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9312120676040649,
      "learning_rate": 8.501931474638899e-05,
      "loss": 1.1039,
      "step": 18050
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9383872747421265,
      "learning_rate": 8.50109170305677e-05,
      "loss": 1.0702,
      "step": 18060
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0627061128616333,
      "learning_rate": 8.50025193147464e-05,
      "loss": 1.1119,
      "step": 18070
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9264964461326599,
      "learning_rate": 8.49941215989251e-05,
      "loss": 1.0883,
      "step": 18080
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9627074599266052,
      "learning_rate": 8.49857238831038e-05,
      "loss": 1.0718,
      "step": 18090
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.029201626777649,
      "learning_rate": 8.49773261672825e-05,
      "loss": 1.0991,
      "step": 18100
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8738752007484436,
      "learning_rate": 8.496892845146121e-05,
      "loss": 1.0994,
      "step": 18110
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9824410080909729,
      "learning_rate": 8.496053073563991e-05,
      "loss": 1.1017,
      "step": 18120
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8930276036262512,
      "learning_rate": 8.495213301981861e-05,
      "loss": 1.095,
      "step": 18130
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9773525595664978,
      "learning_rate": 8.494373530399731e-05,
      "loss": 1.0967,
      "step": 18140
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9366147518157959,
      "learning_rate": 8.493533758817603e-05,
      "loss": 1.0693,
      "step": 18150
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9455195665359497,
      "learning_rate": 8.492693987235473e-05,
      "loss": 1.1014,
      "step": 18160
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9755341410636902,
      "learning_rate": 8.491854215653343e-05,
      "loss": 1.0884,
      "step": 18170
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9536694884300232,
      "learning_rate": 8.491014444071213e-05,
      "loss": 1.0861,
      "step": 18180
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8827406764030457,
      "learning_rate": 8.490174672489083e-05,
      "loss": 1.0809,
      "step": 18190
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9611886143684387,
      "learning_rate": 8.489334900906953e-05,
      "loss": 1.0961,
      "step": 18200
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9199916124343872,
      "learning_rate": 8.488495129324824e-05,
      "loss": 1.1018,
      "step": 18210
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9742301106452942,
      "learning_rate": 8.487655357742695e-05,
      "loss": 1.1064,
      "step": 18220
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9249382615089417,
      "learning_rate": 8.486815586160565e-05,
      "loss": 1.1103,
      "step": 18230
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9248377680778503,
      "learning_rate": 8.485975814578435e-05,
      "loss": 1.0903,
      "step": 18240
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.983910083770752,
      "learning_rate": 8.485136042996306e-05,
      "loss": 1.0985,
      "step": 18250
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9560139179229736,
      "learning_rate": 8.484296271414176e-05,
      "loss": 1.0994,
      "step": 18260
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.892369270324707,
      "learning_rate": 8.483456499832046e-05,
      "loss": 1.0989,
      "step": 18270
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9658694267272949,
      "learning_rate": 8.482616728249916e-05,
      "loss": 1.0846,
      "step": 18280
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9675649404525757,
      "learning_rate": 8.481776956667786e-05,
      "loss": 1.0803,
      "step": 18290
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9331943392753601,
      "learning_rate": 8.480937185085656e-05,
      "loss": 1.1074,
      "step": 18300
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9198856949806213,
      "learning_rate": 8.480097413503527e-05,
      "loss": 1.0911,
      "step": 18310
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9248422980308533,
      "learning_rate": 8.479257641921398e-05,
      "loss": 1.1112,
      "step": 18320
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9331924319267273,
      "learning_rate": 8.478417870339268e-05,
      "loss": 1.093,
      "step": 18330
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.924685001373291,
      "learning_rate": 8.477578098757138e-05,
      "loss": 1.0987,
      "step": 18340
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9171777963638306,
      "learning_rate": 8.47673832717501e-05,
      "loss": 1.0892,
      "step": 18350
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9498481750488281,
      "learning_rate": 8.47589855559288e-05,
      "loss": 1.0988,
      "step": 18360
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9432202577590942,
      "learning_rate": 8.47505878401075e-05,
      "loss": 1.099,
      "step": 18370
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9923070073127747,
      "learning_rate": 8.47421901242862e-05,
      "loss": 1.1257,
      "step": 18380
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9701034426689148,
      "learning_rate": 8.473379240846491e-05,
      "loss": 1.094,
      "step": 18390
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9445346593856812,
      "learning_rate": 8.47253946926436e-05,
      "loss": 1.1179,
      "step": 18400
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8754343390464783,
      "learning_rate": 8.47169969768223e-05,
      "loss": 1.1053,
      "step": 18410
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.987666666507721,
      "learning_rate": 8.470859926100101e-05,
      "loss": 1.12,
      "step": 18420
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9782336354255676,
      "learning_rate": 8.470020154517971e-05,
      "loss": 1.0881,
      "step": 18430
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9279545545578003,
      "learning_rate": 8.469180382935841e-05,
      "loss": 1.0923,
      "step": 18440
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8741807341575623,
      "learning_rate": 8.468340611353713e-05,
      "loss": 1.1156,
      "step": 18450
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9295707941055298,
      "learning_rate": 8.467500839771583e-05,
      "loss": 1.1109,
      "step": 18460
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9178563356399536,
      "learning_rate": 8.466661068189453e-05,
      "loss": 1.0844,
      "step": 18470
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9312026500701904,
      "learning_rate": 8.465821296607323e-05,
      "loss": 1.0835,
      "step": 18480
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9479117393493652,
      "learning_rate": 8.464981525025195e-05,
      "loss": 1.1042,
      "step": 18490
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9105421304702759,
      "learning_rate": 8.464141753443063e-05,
      "loss": 1.0962,
      "step": 18500
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9587165117263794,
      "learning_rate": 8.463301981860933e-05,
      "loss": 1.111,
      "step": 18510
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9755902886390686,
      "learning_rate": 8.462462210278805e-05,
      "loss": 1.1001,
      "step": 18520
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9663732647895813,
      "learning_rate": 8.461622438696675e-05,
      "loss": 1.0885,
      "step": 18530
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0019534826278687,
      "learning_rate": 8.460782667114545e-05,
      "loss": 1.1046,
      "step": 18540
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9537042379379272,
      "learning_rate": 8.459942895532415e-05,
      "loss": 1.0861,
      "step": 18550
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9799646735191345,
      "learning_rate": 8.459103123950286e-05,
      "loss": 1.0812,
      "step": 18560
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8722805976867676,
      "learning_rate": 8.458263352368156e-05,
      "loss": 1.0961,
      "step": 18570
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9394450187683105,
      "learning_rate": 8.457423580786027e-05,
      "loss": 1.1177,
      "step": 18580
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8935893774032593,
      "learning_rate": 8.45666778636211e-05,
      "loss": 1.0953,
      "step": 18590
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.930616021156311,
      "learning_rate": 8.45582801477998e-05,
      "loss": 1.1041,
      "step": 18600
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.894257664680481,
      "learning_rate": 8.454988243197851e-05,
      "loss": 1.089,
      "step": 18610
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9630336165428162,
      "learning_rate": 8.45414847161572e-05,
      "loss": 1.0842,
      "step": 18620
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9433919787406921,
      "learning_rate": 8.45330870003359e-05,
      "loss": 1.0839,
      "step": 18630
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9394075274467468,
      "learning_rate": 8.452468928451462e-05,
      "loss": 1.0964,
      "step": 18640
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9664998650550842,
      "learning_rate": 8.451629156869332e-05,
      "loss": 1.1152,
      "step": 18650
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9555561542510986,
      "learning_rate": 8.450789385287202e-05,
      "loss": 1.093,
      "step": 18660
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9212177395820618,
      "learning_rate": 8.449949613705073e-05,
      "loss": 1.1,
      "step": 18670
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9876947402954102,
      "learning_rate": 8.449109842122943e-05,
      "loss": 1.1108,
      "step": 18680
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9156479239463806,
      "learning_rate": 8.448270070540813e-05,
      "loss": 1.0736,
      "step": 18690
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9072255492210388,
      "learning_rate": 8.447430298958683e-05,
      "loss": 1.1047,
      "step": 18700
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9976296424865723,
      "learning_rate": 8.446590527376555e-05,
      "loss": 1.0892,
      "step": 18710
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9257268905639648,
      "learning_rate": 8.445750755794425e-05,
      "loss": 1.1081,
      "step": 18720
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9828330874443054,
      "learning_rate": 8.444910984212294e-05,
      "loss": 1.0786,
      "step": 18730
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9410355091094971,
      "learning_rate": 8.444071212630165e-05,
      "loss": 1.0948,
      "step": 18740
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9980003833770752,
      "learning_rate": 8.443231441048035e-05,
      "loss": 1.0866,
      "step": 18750
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9311619400978088,
      "learning_rate": 8.442391669465905e-05,
      "loss": 1.0606,
      "step": 18760
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9579390287399292,
      "learning_rate": 8.441551897883777e-05,
      "loss": 1.1125,
      "step": 18770
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.892838180065155,
      "learning_rate": 8.440712126301647e-05,
      "loss": 1.0958,
      "step": 18780
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.964271605014801,
      "learning_rate": 8.439872354719517e-05,
      "loss": 1.0817,
      "step": 18790
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9034227132797241,
      "learning_rate": 8.439032583137387e-05,
      "loss": 1.0884,
      "step": 18800
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9671628475189209,
      "learning_rate": 8.438192811555258e-05,
      "loss": 1.1003,
      "step": 18810
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9182506799697876,
      "learning_rate": 8.437353039973128e-05,
      "loss": 1.1334,
      "step": 18820
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9010951519012451,
      "learning_rate": 8.436513268390997e-05,
      "loss": 1.087,
      "step": 18830
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1023180484771729,
      "learning_rate": 8.435673496808868e-05,
      "loss": 1.0912,
      "step": 18840
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9568824768066406,
      "learning_rate": 8.434833725226739e-05,
      "loss": 1.071,
      "step": 18850
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9224495887756348,
      "learning_rate": 8.433993953644609e-05,
      "loss": 1.0681,
      "step": 18860
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9378868341445923,
      "learning_rate": 8.433154182062479e-05,
      "loss": 1.0908,
      "step": 18870
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9482252597808838,
      "learning_rate": 8.43231441048035e-05,
      "loss": 1.0853,
      "step": 18880
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9346802830696106,
      "learning_rate": 8.43147463889822e-05,
      "loss": 1.1016,
      "step": 18890
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9604766368865967,
      "learning_rate": 8.43063486731609e-05,
      "loss": 1.0791,
      "step": 18900
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9479242563247681,
      "learning_rate": 8.429795095733962e-05,
      "loss": 1.0748,
      "step": 18910
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9255791902542114,
      "learning_rate": 8.428955324151832e-05,
      "loss": 1.1082,
      "step": 18920
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9649567604064941,
      "learning_rate": 8.428115552569702e-05,
      "loss": 1.0851,
      "step": 18930
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9030554294586182,
      "learning_rate": 8.427275780987572e-05,
      "loss": 1.0974,
      "step": 18940
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9574862718582153,
      "learning_rate": 8.426436009405442e-05,
      "loss": 1.0973,
      "step": 18950
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9443824887275696,
      "learning_rate": 8.425596237823312e-05,
      "loss": 1.0973,
      "step": 18960
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9072438478469849,
      "learning_rate": 8.424756466241182e-05,
      "loss": 1.094,
      "step": 18970
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0134698152542114,
      "learning_rate": 8.423916694659054e-05,
      "loss": 1.0964,
      "step": 18980
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9087139964103699,
      "learning_rate": 8.423076923076924e-05,
      "loss": 1.1001,
      "step": 18990
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9348397850990295,
      "learning_rate": 8.422237151494794e-05,
      "loss": 1.0988,
      "step": 19000
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9368646740913391,
      "learning_rate": 8.421397379912665e-05,
      "loss": 1.0877,
      "step": 19010
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9744086265563965,
      "learning_rate": 8.420557608330535e-05,
      "loss": 1.114,
      "step": 19020
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8904855847358704,
      "learning_rate": 8.419717836748405e-05,
      "loss": 1.1,
      "step": 19030
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9219388365745544,
      "learning_rate": 8.418878065166275e-05,
      "loss": 1.0974,
      "step": 19040
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9860958456993103,
      "learning_rate": 8.418038293584145e-05,
      "loss": 1.0859,
      "step": 19050
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9642945528030396,
      "learning_rate": 8.417198522002015e-05,
      "loss": 1.0886,
      "step": 19060
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9327720403671265,
      "learning_rate": 8.416358750419885e-05,
      "loss": 1.117,
      "step": 19070
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8988340497016907,
      "learning_rate": 8.415518978837757e-05,
      "loss": 1.1063,
      "step": 19080
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9338961243629456,
      "learning_rate": 8.414679207255627e-05,
      "loss": 1.0951,
      "step": 19090
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8888064026832581,
      "learning_rate": 8.413839435673497e-05,
      "loss": 1.0878,
      "step": 19100
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9350845813751221,
      "learning_rate": 8.412999664091367e-05,
      "loss": 1.0779,
      "step": 19110
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9388766884803772,
      "learning_rate": 8.412159892509239e-05,
      "loss": 1.1079,
      "step": 19120
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9182081818580627,
      "learning_rate": 8.411320120927109e-05,
      "loss": 1.105,
      "step": 19130
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.932131826877594,
      "learning_rate": 8.410480349344979e-05,
      "loss": 1.093,
      "step": 19140
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9025347232818604,
      "learning_rate": 8.409640577762849e-05,
      "loss": 1.0969,
      "step": 19150
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.036621332168579,
      "learning_rate": 8.408800806180719e-05,
      "loss": 1.0982,
      "step": 19160
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9636215567588806,
      "learning_rate": 8.407961034598589e-05,
      "loss": 1.0775,
      "step": 19170
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9591709971427917,
      "learning_rate": 8.40712126301646e-05,
      "loss": 1.099,
      "step": 19180
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9226289391517639,
      "learning_rate": 8.40628149143433e-05,
      "loss": 1.0922,
      "step": 19190
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8729637265205383,
      "learning_rate": 8.4054417198522e-05,
      "loss": 1.0932,
      "step": 19200
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9399529695510864,
      "learning_rate": 8.40460194827007e-05,
      "loss": 1.0707,
      "step": 19210
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.930468738079071,
      "learning_rate": 8.403762176687942e-05,
      "loss": 1.0735,
      "step": 19220
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.01289701461792,
      "learning_rate": 8.402922405105812e-05,
      "loss": 1.0849,
      "step": 19230
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9548333883285522,
      "learning_rate": 8.402082633523682e-05,
      "loss": 1.1105,
      "step": 19240
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8973714113235474,
      "learning_rate": 8.401242861941552e-05,
      "loss": 1.1048,
      "step": 19250
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8910929560661316,
      "learning_rate": 8.400403090359422e-05,
      "loss": 1.0919,
      "step": 19260
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9398332834243774,
      "learning_rate": 8.399563318777292e-05,
      "loss": 1.1018,
      "step": 19270
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9385080337524414,
      "learning_rate": 8.398723547195164e-05,
      "loss": 1.0744,
      "step": 19280
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9517788887023926,
      "learning_rate": 8.397883775613034e-05,
      "loss": 1.0762,
      "step": 19290
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0044435262680054,
      "learning_rate": 8.397044004030904e-05,
      "loss": 1.0799,
      "step": 19300
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9619836807250977,
      "learning_rate": 8.396204232448774e-05,
      "loss": 1.0788,
      "step": 19310
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8785536289215088,
      "learning_rate": 8.395364460866645e-05,
      "loss": 1.0957,
      "step": 19320
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9379634261131287,
      "learning_rate": 8.394524689284515e-05,
      "loss": 1.1004,
      "step": 19330
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9481569528579712,
      "learning_rate": 8.393684917702385e-05,
      "loss": 1.1068,
      "step": 19340
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8910374045372009,
      "learning_rate": 8.392845146120256e-05,
      "loss": 1.0787,
      "step": 19350
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9397358894348145,
      "learning_rate": 8.392005374538126e-05,
      "loss": 1.1065,
      "step": 19360
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9177324771881104,
      "learning_rate": 8.391165602955996e-05,
      "loss": 1.0805,
      "step": 19370
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9107533693313599,
      "learning_rate": 8.390325831373866e-05,
      "loss": 1.0888,
      "step": 19380
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9538021087646484,
      "learning_rate": 8.389486059791737e-05,
      "loss": 1.1125,
      "step": 19390
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9694135785102844,
      "learning_rate": 8.388646288209607e-05,
      "loss": 1.0996,
      "step": 19400
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9632307291030884,
      "learning_rate": 8.387806516627477e-05,
      "loss": 1.0881,
      "step": 19410
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9254166483879089,
      "learning_rate": 8.386966745045349e-05,
      "loss": 1.0994,
      "step": 19420
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9248160719871521,
      "learning_rate": 8.386126973463219e-05,
      "loss": 1.0838,
      "step": 19430
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9138569831848145,
      "learning_rate": 8.385287201881089e-05,
      "loss": 1.1108,
      "step": 19440
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8703060746192932,
      "learning_rate": 8.384447430298959e-05,
      "loss": 1.0866,
      "step": 19450
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9065632224082947,
      "learning_rate": 8.38360765871683e-05,
      "loss": 1.132,
      "step": 19460
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.037733793258667,
      "learning_rate": 8.382767887134699e-05,
      "loss": 1.0943,
      "step": 19470
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9354545474052429,
      "learning_rate": 8.381928115552569e-05,
      "loss": 1.0801,
      "step": 19480
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0639814138412476,
      "learning_rate": 8.38108834397044e-05,
      "loss": 1.0851,
      "step": 19490
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9247221350669861,
      "learning_rate": 8.38024857238831e-05,
      "loss": 1.1058,
      "step": 19500
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8831647634506226,
      "learning_rate": 8.379408800806181e-05,
      "loss": 1.1169,
      "step": 19510
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0084609985351562,
      "learning_rate": 8.378569029224052e-05,
      "loss": 1.0865,
      "step": 19520
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9183603525161743,
      "learning_rate": 8.377729257641922e-05,
      "loss": 1.1007,
      "step": 19530
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9382530450820923,
      "learning_rate": 8.376889486059792e-05,
      "loss": 1.097,
      "step": 19540
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9827193021774292,
      "learning_rate": 8.376049714477662e-05,
      "loss": 1.0903,
      "step": 19550
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9732797741889954,
      "learning_rate": 8.375209942895534e-05,
      "loss": 1.1017,
      "step": 19560
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9216248393058777,
      "learning_rate": 8.374370171313402e-05,
      "loss": 1.0983,
      "step": 19570
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9452435374259949,
      "learning_rate": 8.373530399731273e-05,
      "loss": 1.0843,
      "step": 19580
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.893520176410675,
      "learning_rate": 8.372690628149144e-05,
      "loss": 1.0868,
      "step": 19590
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.952495276927948,
      "learning_rate": 8.371850856567014e-05,
      "loss": 1.1068,
      "step": 19600
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9048551917076111,
      "learning_rate": 8.371011084984884e-05,
      "loss": 1.1087,
      "step": 19610
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9745410084724426,
      "learning_rate": 8.370171313402754e-05,
      "loss": 1.0941,
      "step": 19620
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9055392742156982,
      "learning_rate": 8.369331541820626e-05,
      "loss": 1.0793,
      "step": 19630
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.943188488483429,
      "learning_rate": 8.368491770238496e-05,
      "loss": 1.0741,
      "step": 19640
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9493341445922852,
      "learning_rate": 8.367651998656366e-05,
      "loss": 1.0611,
      "step": 19650
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9519808292388916,
      "learning_rate": 8.366812227074237e-05,
      "loss": 1.0812,
      "step": 19660
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.895848274230957,
      "learning_rate": 8.365972455492107e-05,
      "loss": 1.1028,
      "step": 19670
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9439095854759216,
      "learning_rate": 8.365132683909976e-05,
      "loss": 1.1063,
      "step": 19680
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.933009922504425,
      "learning_rate": 8.364292912327847e-05,
      "loss": 1.0913,
      "step": 19690
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9215378761291504,
      "learning_rate": 8.363453140745717e-05,
      "loss": 1.0949,
      "step": 19700
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9607491493225098,
      "learning_rate": 8.362613369163588e-05,
      "loss": 1.0928,
      "step": 19710
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9107180237770081,
      "learning_rate": 8.361773597581458e-05,
      "loss": 1.0811,
      "step": 19720
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9495106339454651,
      "learning_rate": 8.360933825999329e-05,
      "loss": 1.1017,
      "step": 19730
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9339221119880676,
      "learning_rate": 8.360094054417199e-05,
      "loss": 1.088,
      "step": 19740
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9448457360267639,
      "learning_rate": 8.359254282835069e-05,
      "loss": 1.0972,
      "step": 19750
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9283002018928528,
      "learning_rate": 8.35841451125294e-05,
      "loss": 1.1112,
      "step": 19760
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9877318143844604,
      "learning_rate": 8.35757473967081e-05,
      "loss": 1.0791,
      "step": 19770
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9149442911148071,
      "learning_rate": 8.35673496808868e-05,
      "loss": 1.0863,
      "step": 19780
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9150471091270447,
      "learning_rate": 8.355895196506551e-05,
      "loss": 1.0826,
      "step": 19790
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9501169323921204,
      "learning_rate": 8.355055424924421e-05,
      "loss": 1.0954,
      "step": 19800
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9282243847846985,
      "learning_rate": 8.354215653342291e-05,
      "loss": 1.0757,
      "step": 19810
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.954955518245697,
      "learning_rate": 8.353375881760161e-05,
      "loss": 1.0805,
      "step": 19820
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9169294834136963,
      "learning_rate": 8.352536110178032e-05,
      "loss": 1.1015,
      "step": 19830
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9292359948158264,
      "learning_rate": 8.351696338595902e-05,
      "loss": 1.081,
      "step": 19840
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9779778718948364,
      "learning_rate": 8.350856567013773e-05,
      "loss": 1.0671,
      "step": 19850
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9182485342025757,
      "learning_rate": 8.350016795431643e-05,
      "loss": 1.1031,
      "step": 19860
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9590606689453125,
      "learning_rate": 8.349177023849514e-05,
      "loss": 1.1048,
      "step": 19870
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9228118062019348,
      "learning_rate": 8.348337252267384e-05,
      "loss": 1.0764,
      "step": 19880
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9537436366081238,
      "learning_rate": 8.347497480685253e-05,
      "loss": 1.1226,
      "step": 19890
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9905833601951599,
      "learning_rate": 8.346657709103124e-05,
      "loss": 1.0988,
      "step": 19900
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.859458327293396,
      "learning_rate": 8.345817937520994e-05,
      "loss": 1.0828,
      "step": 19910
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9201727509498596,
      "learning_rate": 8.344978165938864e-05,
      "loss": 1.0924,
      "step": 19920
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9572312831878662,
      "learning_rate": 8.344138394356736e-05,
      "loss": 1.0809,
      "step": 19930
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9225120544433594,
      "learning_rate": 8.343298622774606e-05,
      "loss": 1.0869,
      "step": 19940
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9653365612030029,
      "learning_rate": 8.342458851192476e-05,
      "loss": 1.0666,
      "step": 19950
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.939113438129425,
      "learning_rate": 8.341619079610346e-05,
      "loss": 1.071,
      "step": 19960
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9120537638664246,
      "learning_rate": 8.340779308028217e-05,
      "loss": 1.125,
      "step": 19970
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9475664496421814,
      "learning_rate": 8.339939536446088e-05,
      "loss": 1.0907,
      "step": 19980
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9591881632804871,
      "learning_rate": 8.339099764863958e-05,
      "loss": 1.0754,
      "step": 19990
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0526905059814453,
      "learning_rate": 8.338259993281828e-05,
      "loss": 1.0928,
      "step": 20000
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9447547197341919,
      "learning_rate": 8.337420221699698e-05,
      "loss": 1.1016,
      "step": 20010
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9428248405456543,
      "learning_rate": 8.336580450117568e-05,
      "loss": 1.1052,
      "step": 20020
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9033694267272949,
      "learning_rate": 8.335740678535439e-05,
      "loss": 1.0798,
      "step": 20030
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.012122631072998,
      "learning_rate": 8.334900906953309e-05,
      "loss": 1.0948,
      "step": 20040
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9211704134941101,
      "learning_rate": 8.33406113537118e-05,
      "loss": 1.0769,
      "step": 20050
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0090031623840332,
      "learning_rate": 8.33322136378905e-05,
      "loss": 1.0716,
      "step": 20060
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9996041655540466,
      "learning_rate": 8.332381592206921e-05,
      "loss": 1.1029,
      "step": 20070
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.950567901134491,
      "learning_rate": 8.331541820624791e-05,
      "loss": 1.0849,
      "step": 20080
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9663973450660706,
      "learning_rate": 8.330702049042661e-05,
      "loss": 1.0459,
      "step": 20090
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.012169599533081,
      "learning_rate": 8.329862277460531e-05,
      "loss": 1.1009,
      "step": 20100
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9550958275794983,
      "learning_rate": 8.329022505878401e-05,
      "loss": 1.1168,
      "step": 20110
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9381619095802307,
      "learning_rate": 8.328182734296271e-05,
      "loss": 1.0956,
      "step": 20120
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9636103510856628,
      "learning_rate": 8.327342962714141e-05,
      "loss": 1.1065,
      "step": 20130
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9450669884681702,
      "learning_rate": 8.326503191132013e-05,
      "loss": 1.0884,
      "step": 20140
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.915797233581543,
      "learning_rate": 8.325663419549883e-05,
      "loss": 1.0725,
      "step": 20150
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9706189036369324,
      "learning_rate": 8.324823647967753e-05,
      "loss": 1.0718,
      "step": 20160
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9453221559524536,
      "learning_rate": 8.323983876385624e-05,
      "loss": 1.0777,
      "step": 20170
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.932264506816864,
      "learning_rate": 8.323144104803494e-05,
      "loss": 1.0708,
      "step": 20180
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9162026047706604,
      "learning_rate": 8.322304333221364e-05,
      "loss": 1.0987,
      "step": 20190
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9243186712265015,
      "learning_rate": 8.321464561639234e-05,
      "loss": 1.1059,
      "step": 20200
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9535598754882812,
      "learning_rate": 8.320624790057105e-05,
      "loss": 1.0899,
      "step": 20210
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.974871039390564,
      "learning_rate": 8.319785018474975e-05,
      "loss": 1.1007,
      "step": 20220
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8855296969413757,
      "learning_rate": 8.318945246892845e-05,
      "loss": 1.0642,
      "step": 20230
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8772308230400085,
      "learning_rate": 8.318105475310716e-05,
      "loss": 1.109,
      "step": 20240
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9719114303588867,
      "learning_rate": 8.317265703728586e-05,
      "loss": 1.0832,
      "step": 20250
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9716581106185913,
      "learning_rate": 8.316425932146456e-05,
      "loss": 1.0882,
      "step": 20260
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9847034811973572,
      "learning_rate": 8.315586160564328e-05,
      "loss": 1.0608,
      "step": 20270
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9305758476257324,
      "learning_rate": 8.314746388982198e-05,
      "loss": 1.0842,
      "step": 20280
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9634222388267517,
      "learning_rate": 8.313906617400068e-05,
      "loss": 1.0948,
      "step": 20290
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9088085889816284,
      "learning_rate": 8.313066845817938e-05,
      "loss": 1.0944,
      "step": 20300
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9427503943443298,
      "learning_rate": 8.312227074235808e-05,
      "loss": 1.0859,
      "step": 20310
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9813087582588196,
      "learning_rate": 8.311387302653678e-05,
      "loss": 1.0772,
      "step": 20320
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0378156900405884,
      "learning_rate": 8.310547531071548e-05,
      "loss": 1.085,
      "step": 20330
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9516633749008179,
      "learning_rate": 8.30970775948942e-05,
      "loss": 1.0633,
      "step": 20340
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9724482297897339,
      "learning_rate": 8.30886798790729e-05,
      "loss": 1.0834,
      "step": 20350
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9871209859848022,
      "learning_rate": 8.30802821632516e-05,
      "loss": 1.0966,
      "step": 20360
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8819617629051208,
      "learning_rate": 8.307188444743031e-05,
      "loss": 1.1045,
      "step": 20370
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8651036024093628,
      "learning_rate": 8.306348673160901e-05,
      "loss": 1.0794,
      "step": 20380
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8954698443412781,
      "learning_rate": 8.305508901578771e-05,
      "loss": 1.0852,
      "step": 20390
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9244828224182129,
      "learning_rate": 8.304669129996641e-05,
      "loss": 1.1205,
      "step": 20400
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0268381834030151,
      "learning_rate": 8.303829358414513e-05,
      "loss": 1.0525,
      "step": 20410
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9340755939483643,
      "learning_rate": 8.302989586832381e-05,
      "loss": 1.0769,
      "step": 20420
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.969909131526947,
      "learning_rate": 8.302149815250251e-05,
      "loss": 1.0801,
      "step": 20430
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9364521503448486,
      "learning_rate": 8.301310043668123e-05,
      "loss": 1.0712,
      "step": 20440
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9704520106315613,
      "learning_rate": 8.300470272085993e-05,
      "loss": 1.0939,
      "step": 20450
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9697971940040588,
      "learning_rate": 8.299630500503863e-05,
      "loss": 1.0604,
      "step": 20460
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9544804692268372,
      "learning_rate": 8.298790728921733e-05,
      "loss": 1.1033,
      "step": 20470
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.962824821472168,
      "learning_rate": 8.297950957339605e-05,
      "loss": 1.0785,
      "step": 20480
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0860247611999512,
      "learning_rate": 8.297111185757475e-05,
      "loss": 1.0572,
      "step": 20490
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9392929077148438,
      "learning_rate": 8.296271414175345e-05,
      "loss": 1.0893,
      "step": 20500
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8851145505905151,
      "learning_rate": 8.295431642593216e-05,
      "loss": 1.0796,
      "step": 20510
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9846409559249878,
      "learning_rate": 8.294591871011086e-05,
      "loss": 1.0796,
      "step": 20520
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9698294401168823,
      "learning_rate": 8.293752099428955e-05,
      "loss": 1.0969,
      "step": 20530
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9646922945976257,
      "learning_rate": 8.292912327846826e-05,
      "loss": 1.0767,
      "step": 20540
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9086426496505737,
      "learning_rate": 8.292072556264696e-05,
      "loss": 1.0514,
      "step": 20550
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.943329393863678,
      "learning_rate": 8.291232784682566e-05,
      "loss": 1.0919,
      "step": 20560
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9775128364562988,
      "learning_rate": 8.290393013100436e-05,
      "loss": 1.0855,
      "step": 20570
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9288136959075928,
      "learning_rate": 8.289553241518308e-05,
      "loss": 1.0845,
      "step": 20580
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9252897500991821,
      "learning_rate": 8.288713469936178e-05,
      "loss": 1.1043,
      "step": 20590
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9563115239143372,
      "learning_rate": 8.287873698354048e-05,
      "loss": 1.0949,
      "step": 20600
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9599754214286804,
      "learning_rate": 8.28703392677192e-05,
      "loss": 1.0817,
      "step": 20610
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9107076525688171,
      "learning_rate": 8.28619415518979e-05,
      "loss": 1.085,
      "step": 20620
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9191986322402954,
      "learning_rate": 8.285354383607658e-05,
      "loss": 1.0903,
      "step": 20630
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9460235834121704,
      "learning_rate": 8.28451461202553e-05,
      "loss": 1.0975,
      "step": 20640
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9365357756614685,
      "learning_rate": 8.2836748404434e-05,
      "loss": 1.0927,
      "step": 20650
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9596380591392517,
      "learning_rate": 8.28283506886127e-05,
      "loss": 1.0926,
      "step": 20660
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9388223886489868,
      "learning_rate": 8.28199529727914e-05,
      "loss": 1.0611,
      "step": 20670
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9124281406402588,
      "learning_rate": 8.281155525697011e-05,
      "loss": 1.0949,
      "step": 20680
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9100083708763123,
      "learning_rate": 8.280315754114881e-05,
      "loss": 1.0685,
      "step": 20690
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9451543092727661,
      "learning_rate": 8.279475982532751e-05,
      "loss": 1.0742,
      "step": 20700
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9619319438934326,
      "learning_rate": 8.278636210950622e-05,
      "loss": 1.0732,
      "step": 20710
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9725054502487183,
      "learning_rate": 8.277796439368493e-05,
      "loss": 1.074,
      "step": 20720
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0174696445465088,
      "learning_rate": 8.276956667786363e-05,
      "loss": 1.0804,
      "step": 20730
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9788162112236023,
      "learning_rate": 8.276116896204232e-05,
      "loss": 1.0705,
      "step": 20740
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9715824127197266,
      "learning_rate": 8.275277124622103e-05,
      "loss": 1.0692,
      "step": 20750
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.943411111831665,
      "learning_rate": 8.274437353039973e-05,
      "loss": 1.102,
      "step": 20760
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.929039716720581,
      "learning_rate": 8.273597581457843e-05,
      "loss": 1.0862,
      "step": 20770
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9214218258857727,
      "learning_rate": 8.272757809875715e-05,
      "loss": 1.0552,
      "step": 20780
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9461379051208496,
      "learning_rate": 8.271918038293585e-05,
      "loss": 1.0901,
      "step": 20790
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8952659964561462,
      "learning_rate": 8.271078266711455e-05,
      "loss": 1.1071,
      "step": 20800
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.903317928314209,
      "learning_rate": 8.270238495129325e-05,
      "loss": 1.0806,
      "step": 20810
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.941967248916626,
      "learning_rate": 8.269398723547196e-05,
      "loss": 1.0882,
      "step": 20820
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9013721346855164,
      "learning_rate": 8.268558951965066e-05,
      "loss": 1.0925,
      "step": 20830
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9886294007301331,
      "learning_rate": 8.267719180382935e-05,
      "loss": 1.1032,
      "step": 20840
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9935934543609619,
      "learning_rate": 8.266879408800807e-05,
      "loss": 1.112,
      "step": 20850
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9625208377838135,
      "learning_rate": 8.266039637218677e-05,
      "loss": 1.0961,
      "step": 20860
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9433298707008362,
      "learning_rate": 8.265199865636547e-05,
      "loss": 1.0905,
      "step": 20870
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9611908197402954,
      "learning_rate": 8.264360094054418e-05,
      "loss": 1.0865,
      "step": 20880
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9369805455207825,
      "learning_rate": 8.263520322472288e-05,
      "loss": 1.0922,
      "step": 20890
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9557572603225708,
      "learning_rate": 8.262680550890158e-05,
      "loss": 1.0972,
      "step": 20900
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9097500443458557,
      "learning_rate": 8.261840779308028e-05,
      "loss": 1.0754,
      "step": 20910
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9624474048614502,
      "learning_rate": 8.2610010077259e-05,
      "loss": 1.0619,
      "step": 20920
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9251635074615479,
      "learning_rate": 8.26016123614377e-05,
      "loss": 1.0965,
      "step": 20930
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9400815963745117,
      "learning_rate": 8.25932146456164e-05,
      "loss": 1.0839,
      "step": 20940
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9335811138153076,
      "learning_rate": 8.25848169297951e-05,
      "loss": 1.0913,
      "step": 20950
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0157513618469238,
      "learning_rate": 8.25764192139738e-05,
      "loss": 1.0885,
      "step": 20960
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9043487310409546,
      "learning_rate": 8.25680214981525e-05,
      "loss": 1.0927,
      "step": 20970
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8861150741577148,
      "learning_rate": 8.25596237823312e-05,
      "loss": 1.0791,
      "step": 20980
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9011774063110352,
      "learning_rate": 8.255122606650992e-05,
      "loss": 1.0486,
      "step": 20990
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9446386098861694,
      "learning_rate": 8.254282835068862e-05,
      "loss": 1.1008,
      "step": 21000
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8907005190849304,
      "learning_rate": 8.253443063486732e-05,
      "loss": 1.0767,
      "step": 21010
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0680102109909058,
      "learning_rate": 8.252603291904603e-05,
      "loss": 1.0766,
      "step": 21020
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9224295616149902,
      "learning_rate": 8.251763520322473e-05,
      "loss": 1.093,
      "step": 21030
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9853148460388184,
      "learning_rate": 8.250923748740343e-05,
      "loss": 1.0854,
      "step": 21040
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0135034322738647,
      "learning_rate": 8.250083977158213e-05,
      "loss": 1.06,
      "step": 21050
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8956427574157715,
      "learning_rate": 8.249244205576083e-05,
      "loss": 1.0805,
      "step": 21060
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9784128069877625,
      "learning_rate": 8.248404433993953e-05,
      "loss": 1.0861,
      "step": 21070
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9585116505622864,
      "learning_rate": 8.247564662411824e-05,
      "loss": 1.0792,
      "step": 21080
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.895172119140625,
      "learning_rate": 8.246724890829695e-05,
      "loss": 1.0902,
      "step": 21090
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9525881409645081,
      "learning_rate": 8.245885119247565e-05,
      "loss": 1.0565,
      "step": 21100
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8964269757270813,
      "learning_rate": 8.245045347665435e-05,
      "loss": 1.0909,
      "step": 21110
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9538741111755371,
      "learning_rate": 8.244205576083307e-05,
      "loss": 1.0751,
      "step": 21120
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0493577718734741,
      "learning_rate": 8.243365804501177e-05,
      "loss": 1.0944,
      "step": 21130
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.932163417339325,
      "learning_rate": 8.242526032919047e-05,
      "loss": 1.0669,
      "step": 21140
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9787120819091797,
      "learning_rate": 8.241686261336917e-05,
      "loss": 1.0753,
      "step": 21150
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9854269027709961,
      "learning_rate": 8.240846489754787e-05,
      "loss": 1.0568,
      "step": 21160
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9161325693130493,
      "learning_rate": 8.240006718172657e-05,
      "loss": 1.0643,
      "step": 21170
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9661205410957336,
      "learning_rate": 8.239166946590527e-05,
      "loss": 1.0643,
      "step": 21180
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9399836659431458,
      "learning_rate": 8.238327175008398e-05,
      "loss": 1.0842,
      "step": 21190
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9120281338691711,
      "learning_rate": 8.237487403426268e-05,
      "loss": 1.0735,
      "step": 21200
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.958123505115509,
      "learning_rate": 8.236647631844139e-05,
      "loss": 1.0996,
      "step": 21210
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9809736609458923,
      "learning_rate": 8.235807860262009e-05,
      "loss": 1.092,
      "step": 21220
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9468401670455933,
      "learning_rate": 8.23496808867988e-05,
      "loss": 1.0811,
      "step": 21230
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.90665602684021,
      "learning_rate": 8.23412831709775e-05,
      "loss": 1.0662,
      "step": 21240
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.008862018585205,
      "learning_rate": 8.23328854551562e-05,
      "loss": 1.0935,
      "step": 21250
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9973275661468506,
      "learning_rate": 8.232448773933492e-05,
      "loss": 1.0741,
      "step": 21260
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.007988691329956,
      "learning_rate": 8.23160900235136e-05,
      "loss": 1.1038,
      "step": 21270
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9017188549041748,
      "learning_rate": 8.23076923076923e-05,
      "loss": 1.0901,
      "step": 21280
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9623133540153503,
      "learning_rate": 8.229929459187102e-05,
      "loss": 1.0838,
      "step": 21290
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9186885952949524,
      "learning_rate": 8.229089687604972e-05,
      "loss": 1.0923,
      "step": 21300
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9776137471199036,
      "learning_rate": 8.228249916022842e-05,
      "loss": 1.062,
      "step": 21310
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9220883846282959,
      "learning_rate": 8.227410144440712e-05,
      "loss": 1.0607,
      "step": 21320
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9247794151306152,
      "learning_rate": 8.226570372858583e-05,
      "loss": 1.0754,
      "step": 21330
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9729852676391602,
      "learning_rate": 8.225730601276453e-05,
      "loss": 1.1019,
      "step": 21340
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8800434470176697,
      "learning_rate": 8.224890829694324e-05,
      "loss": 1.0904,
      "step": 21350
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9299606084823608,
      "learning_rate": 8.224051058112195e-05,
      "loss": 1.0743,
      "step": 21360
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9961258769035339,
      "learning_rate": 8.223211286530064e-05,
      "loss": 1.1022,
      "step": 21370
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9743673205375671,
      "learning_rate": 8.222371514947934e-05,
      "loss": 1.0716,
      "step": 21380
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.998546302318573,
      "learning_rate": 8.221531743365805e-05,
      "loss": 1.0825,
      "step": 21390
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9833109974861145,
      "learning_rate": 8.220691971783675e-05,
      "loss": 1.1103,
      "step": 21400
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9474690556526184,
      "learning_rate": 8.219852200201545e-05,
      "loss": 1.0768,
      "step": 21410
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9872686266899109,
      "learning_rate": 8.219012428619415e-05,
      "loss": 1.0924,
      "step": 21420
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.017450213432312,
      "learning_rate": 8.218172657037287e-05,
      "loss": 1.0793,
      "step": 21430
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.972040593624115,
      "learning_rate": 8.217332885455157e-05,
      "loss": 1.073,
      "step": 21440
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9796550869941711,
      "learning_rate": 8.216493113873027e-05,
      "loss": 1.1053,
      "step": 21450
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8813996315002441,
      "learning_rate": 8.215653342290897e-05,
      "loss": 1.077,
      "step": 21460
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8760569095611572,
      "learning_rate": 8.214813570708768e-05,
      "loss": 1.0903,
      "step": 21470
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0394665002822876,
      "learning_rate": 8.213973799126637e-05,
      "loss": 1.0911,
      "step": 21480
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.933469295501709,
      "learning_rate": 8.213134027544507e-05,
      "loss": 1.0826,
      "step": 21490
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9994786381721497,
      "learning_rate": 8.212294255962379e-05,
      "loss": 1.0663,
      "step": 21500
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9118964672088623,
      "learning_rate": 8.211454484380249e-05,
      "loss": 1.1219,
      "step": 21510
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.920733630657196,
      "learning_rate": 8.210614712798119e-05,
      "loss": 1.0899,
      "step": 21520
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9079091548919678,
      "learning_rate": 8.20977494121599e-05,
      "loss": 1.0845,
      "step": 21530
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9618085622787476,
      "learning_rate": 8.20893516963386e-05,
      "loss": 1.085,
      "step": 21540
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9206164479255676,
      "learning_rate": 8.20809539805173e-05,
      "loss": 1.0712,
      "step": 21550
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9424861073493958,
      "learning_rate": 8.2072556264696e-05,
      "loss": 1.0852,
      "step": 21560
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9359887838363647,
      "learning_rate": 8.206415854887472e-05,
      "loss": 1.0682,
      "step": 21570
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9242563247680664,
      "learning_rate": 8.205576083305342e-05,
      "loss": 1.0746,
      "step": 21580
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0069301128387451,
      "learning_rate": 8.20473631172321e-05,
      "loss": 1.1143,
      "step": 21590
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9253529906272888,
      "learning_rate": 8.203896540141082e-05,
      "loss": 1.0594,
      "step": 21600
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9807161688804626,
      "learning_rate": 8.203056768558952e-05,
      "loss": 1.0656,
      "step": 21610
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0018244981765747,
      "learning_rate": 8.202216996976822e-05,
      "loss": 1.1132,
      "step": 21620
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9353353977203369,
      "learning_rate": 8.201377225394694e-05,
      "loss": 1.0787,
      "step": 21630
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8812234997749329,
      "learning_rate": 8.200537453812564e-05,
      "loss": 1.0941,
      "step": 21640
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.923886239528656,
      "learning_rate": 8.199697682230434e-05,
      "loss": 1.0931,
      "step": 21650
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9797891974449158,
      "learning_rate": 8.198857910648304e-05,
      "loss": 1.0864,
      "step": 21660
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9551218152046204,
      "learning_rate": 8.198018139066175e-05,
      "loss": 1.0962,
      "step": 21670
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9550528526306152,
      "learning_rate": 8.197178367484045e-05,
      "loss": 1.0974,
      "step": 21680
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9366137385368347,
      "learning_rate": 8.196338595901914e-05,
      "loss": 1.1015,
      "step": 21690
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0413545370101929,
      "learning_rate": 8.195498824319785e-05,
      "loss": 1.0791,
      "step": 21700
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9629783630371094,
      "learning_rate": 8.194659052737656e-05,
      "loss": 1.0885,
      "step": 21710
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9376969337463379,
      "learning_rate": 8.193819281155526e-05,
      "loss": 1.0595,
      "step": 21720
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.922089159488678,
      "learning_rate": 8.192979509573396e-05,
      "loss": 1.0483,
      "step": 21730
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9900693893432617,
      "learning_rate": 8.192139737991267e-05,
      "loss": 1.0981,
      "step": 21740
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9913142919540405,
      "learning_rate": 8.191299966409137e-05,
      "loss": 1.0775,
      "step": 21750
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0081490278244019,
      "learning_rate": 8.190460194827007e-05,
      "loss": 1.0845,
      "step": 21760
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8816586136817932,
      "learning_rate": 8.189620423244879e-05,
      "loss": 1.0928,
      "step": 21770
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9351156949996948,
      "learning_rate": 8.188780651662749e-05,
      "loss": 1.0962,
      "step": 21780
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9366614818572998,
      "learning_rate": 8.187940880080619e-05,
      "loss": 1.0729,
      "step": 21790
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.930763304233551,
      "learning_rate": 8.187101108498489e-05,
      "loss": 1.0951,
      "step": 21800
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9486774802207947,
      "learning_rate": 8.186261336916359e-05,
      "loss": 1.0833,
      "step": 21810
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9392075538635254,
      "learning_rate": 8.185421565334229e-05,
      "loss": 1.0957,
      "step": 21820
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9175149202346802,
      "learning_rate": 8.184581793752099e-05,
      "loss": 1.0716,
      "step": 21830
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9535057544708252,
      "learning_rate": 8.18374202216997e-05,
      "loss": 1.092,
      "step": 21840
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9135550856590271,
      "learning_rate": 8.18290225058784e-05,
      "loss": 1.0677,
      "step": 21850
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9349638223648071,
      "learning_rate": 8.18206247900571e-05,
      "loss": 1.0694,
      "step": 21860
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9543958902359009,
      "learning_rate": 8.181222707423582e-05,
      "loss": 1.0898,
      "step": 21870
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9582996964454651,
      "learning_rate": 8.180382935841452e-05,
      "loss": 1.0672,
      "step": 21880
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9028475880622864,
      "learning_rate": 8.179543164259322e-05,
      "loss": 1.0957,
      "step": 21890
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9782268404960632,
      "learning_rate": 8.178703392677192e-05,
      "loss": 1.0925,
      "step": 21900
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9743828773498535,
      "learning_rate": 8.177863621095062e-05,
      "loss": 1.0715,
      "step": 21910
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8652592897415161,
      "learning_rate": 8.177023849512932e-05,
      "loss": 1.0864,
      "step": 21920
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9322866201400757,
      "learning_rate": 8.176184077930802e-05,
      "loss": 1.0908,
      "step": 21930
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.929085910320282,
      "learning_rate": 8.175344306348674e-05,
      "loss": 1.0787,
      "step": 21940
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9540700912475586,
      "learning_rate": 8.174504534766544e-05,
      "loss": 1.0783,
      "step": 21950
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0052622556686401,
      "learning_rate": 8.173664763184414e-05,
      "loss": 1.0562,
      "step": 21960
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9070284366607666,
      "learning_rate": 8.172824991602284e-05,
      "loss": 1.0682,
      "step": 21970
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9628702402114868,
      "learning_rate": 8.171985220020156e-05,
      "loss": 1.0817,
      "step": 21980
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9631544947624207,
      "learning_rate": 8.171145448438026e-05,
      "loss": 1.0646,
      "step": 21990
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9579653739929199,
      "learning_rate": 8.170305676855896e-05,
      "loss": 1.0801,
      "step": 22000
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9916050434112549,
      "learning_rate": 8.169465905273766e-05,
      "loss": 1.0699,
      "step": 22010
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8972102403640747,
      "learning_rate": 8.168626133691636e-05,
      "loss": 1.0924,
      "step": 22020
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9655322432518005,
      "learning_rate": 8.167786362109506e-05,
      "loss": 1.0855,
      "step": 22030
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.961615264415741,
      "learning_rate": 8.166946590527377e-05,
      "loss": 1.0665,
      "step": 22040
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9401275515556335,
      "learning_rate": 8.166106818945247e-05,
      "loss": 1.0871,
      "step": 22050
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9355718493461609,
      "learning_rate": 8.165267047363117e-05,
      "loss": 1.0628,
      "step": 22060
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9333938956260681,
      "learning_rate": 8.164427275780987e-05,
      "loss": 1.0686,
      "step": 22070
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9262435436248779,
      "learning_rate": 8.163587504198859e-05,
      "loss": 1.0774,
      "step": 22080
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.881001889705658,
      "learning_rate": 8.162747732616729e-05,
      "loss": 1.0975,
      "step": 22090
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9583486318588257,
      "learning_rate": 8.161907961034599e-05,
      "loss": 1.0693,
      "step": 22100
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9099860191345215,
      "learning_rate": 8.16106818945247e-05,
      "loss": 1.0781,
      "step": 22110
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9885082244873047,
      "learning_rate": 8.160228417870339e-05,
      "loss": 1.081,
      "step": 22120
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9352201223373413,
      "learning_rate": 8.159388646288209e-05,
      "loss": 1.0734,
      "step": 22130
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9802115559577942,
      "learning_rate": 8.158548874706081e-05,
      "loss": 1.0791,
      "step": 22140
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.951797366142273,
      "learning_rate": 8.157709103123951e-05,
      "loss": 1.0886,
      "step": 22150
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8874569535255432,
      "learning_rate": 8.156869331541821e-05,
      "loss": 1.09,
      "step": 22160
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0224614143371582,
      "learning_rate": 8.156029559959691e-05,
      "loss": 1.0689,
      "step": 22170
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9177655577659607,
      "learning_rate": 8.155189788377562e-05,
      "loss": 1.0579,
      "step": 22180
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9853928089141846,
      "learning_rate": 8.154350016795432e-05,
      "loss": 1.0966,
      "step": 22190
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.974591076374054,
      "learning_rate": 8.153510245213302e-05,
      "loss": 1.0844,
      "step": 22200
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9281429052352905,
      "learning_rate": 8.152670473631174e-05,
      "loss": 1.0812,
      "step": 22210
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0357728004455566,
      "learning_rate": 8.151830702049043e-05,
      "loss": 1.1061,
      "step": 22220
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9517058730125427,
      "learning_rate": 8.150990930466913e-05,
      "loss": 1.0812,
      "step": 22230
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9043789505958557,
      "learning_rate": 8.150151158884784e-05,
      "loss": 1.0764,
      "step": 22240
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9185697436332703,
      "learning_rate": 8.149311387302654e-05,
      "loss": 1.0845,
      "step": 22250
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9180626273155212,
      "learning_rate": 8.148471615720524e-05,
      "loss": 1.0975,
      "step": 22260
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.929567813873291,
      "learning_rate": 8.147631844138394e-05,
      "loss": 1.0699,
      "step": 22270
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.914916455745697,
      "learning_rate": 8.146792072556266e-05,
      "loss": 1.0637,
      "step": 22280
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0016752481460571,
      "learning_rate": 8.145952300974136e-05,
      "loss": 1.0874,
      "step": 22290
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9593884944915771,
      "learning_rate": 8.145112529392006e-05,
      "loss": 1.0892,
      "step": 22300
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9632132053375244,
      "learning_rate": 8.144272757809876e-05,
      "loss": 1.0751,
      "step": 22310
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9335569739341736,
      "learning_rate": 8.143432986227747e-05,
      "loss": 1.0691,
      "step": 22320
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9385718703269958,
      "learning_rate": 8.142593214645616e-05,
      "loss": 1.0819,
      "step": 22330
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9883543848991394,
      "learning_rate": 8.141753443063486e-05,
      "loss": 1.1087,
      "step": 22340
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9951726198196411,
      "learning_rate": 8.140913671481358e-05,
      "loss": 1.0783,
      "step": 22350
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9235413074493408,
      "learning_rate": 8.140073899899228e-05,
      "loss": 1.0514,
      "step": 22360
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9246856570243835,
      "learning_rate": 8.139234128317098e-05,
      "loss": 1.0962,
      "step": 22370
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9286307692527771,
      "learning_rate": 8.138394356734969e-05,
      "loss": 1.0609,
      "step": 22380
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.98885178565979,
      "learning_rate": 8.137554585152839e-05,
      "loss": 1.0703,
      "step": 22390
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9366748332977295,
      "learning_rate": 8.136714813570709e-05,
      "loss": 1.062,
      "step": 22400
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8993573784828186,
      "learning_rate": 8.135875041988579e-05,
      "loss": 1.0775,
      "step": 22410
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9327789545059204,
      "learning_rate": 8.135035270406451e-05,
      "loss": 1.0756,
      "step": 22420
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9567099809646606,
      "learning_rate": 8.13419549882432e-05,
      "loss": 1.0785,
      "step": 22430
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.86855149269104,
      "learning_rate": 8.13335572724219e-05,
      "loss": 1.0636,
      "step": 22440
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9225255250930786,
      "learning_rate": 8.132515955660061e-05,
      "loss": 1.0647,
      "step": 22450
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9583046436309814,
      "learning_rate": 8.131676184077931e-05,
      "loss": 1.0691,
      "step": 22460
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9480694532394409,
      "learning_rate": 8.130836412495801e-05,
      "loss": 1.0844,
      "step": 22470
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9560478329658508,
      "learning_rate": 8.129996640913673e-05,
      "loss": 1.0658,
      "step": 22480
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9470974206924438,
      "learning_rate": 8.129156869331543e-05,
      "loss": 1.0805,
      "step": 22490
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9007673263549805,
      "learning_rate": 8.128317097749413e-05,
      "loss": 1.0868,
      "step": 22500
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9324938654899597,
      "learning_rate": 8.127477326167283e-05,
      "loss": 1.0575,
      "step": 22510
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9622287750244141,
      "learning_rate": 8.126637554585154e-05,
      "loss": 1.0618,
      "step": 22520
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8709717988967896,
      "learning_rate": 8.125797783003024e-05,
      "loss": 1.0648,
      "step": 22530
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9371627569198608,
      "learning_rate": 8.124958011420893e-05,
      "loss": 1.0953,
      "step": 22540
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.948164701461792,
      "learning_rate": 8.124118239838764e-05,
      "loss": 1.0774,
      "step": 22550
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9619824290275574,
      "learning_rate": 8.123278468256634e-05,
      "loss": 1.0783,
      "step": 22560
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9640927910804749,
      "learning_rate": 8.122438696674504e-05,
      "loss": 1.0637,
      "step": 22570
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8726696968078613,
      "learning_rate": 8.121598925092375e-05,
      "loss": 1.0622,
      "step": 22580
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9290249943733215,
      "learning_rate": 8.120759153510246e-05,
      "loss": 1.0841,
      "step": 22590
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9070258736610413,
      "learning_rate": 8.119919381928116e-05,
      "loss": 1.0719,
      "step": 22600
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9861231446266174,
      "learning_rate": 8.119079610345986e-05,
      "loss": 1.0827,
      "step": 22610
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9475063681602478,
      "learning_rate": 8.118239838763858e-05,
      "loss": 1.0894,
      "step": 22620
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9697584509849548,
      "learning_rate": 8.117400067181728e-05,
      "loss": 1.0965,
      "step": 22630
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9208489060401917,
      "learning_rate": 8.116560295599598e-05,
      "loss": 1.0746,
      "step": 22640
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9453262686729431,
      "learning_rate": 8.115720524017468e-05,
      "loss": 1.1015,
      "step": 22650
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9898834824562073,
      "learning_rate": 8.114880752435338e-05,
      "loss": 1.0798,
      "step": 22660
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9766094088554382,
      "learning_rate": 8.114040980853208e-05,
      "loss": 1.062,
      "step": 22670
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9808549880981445,
      "learning_rate": 8.113201209271078e-05,
      "loss": 1.1038,
      "step": 22680
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8814826607704163,
      "learning_rate": 8.11236143768895e-05,
      "loss": 1.0757,
      "step": 22690
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9363024830818176,
      "learning_rate": 8.11152166610682e-05,
      "loss": 1.0753,
      "step": 22700
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8872799277305603,
      "learning_rate": 8.11068189452469e-05,
      "loss": 1.0931,
      "step": 22710
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9672569036483765,
      "learning_rate": 8.109842122942561e-05,
      "loss": 1.0803,
      "step": 22720
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9465705156326294,
      "learning_rate": 8.109002351360431e-05,
      "loss": 1.0726,
      "step": 22730
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9397774934768677,
      "learning_rate": 8.108162579778301e-05,
      "loss": 1.0746,
      "step": 22740
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9704985022544861,
      "learning_rate": 8.107322808196171e-05,
      "loss": 1.0571,
      "step": 22750
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.961126446723938,
      "learning_rate": 8.106483036614041e-05,
      "loss": 1.0824,
      "step": 22760
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9272827506065369,
      "learning_rate": 8.105643265031911e-05,
      "loss": 1.0959,
      "step": 22770
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9782728552818298,
      "learning_rate": 8.104803493449781e-05,
      "loss": 1.0653,
      "step": 22780
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9585440158843994,
      "learning_rate": 8.103963721867653e-05,
      "loss": 1.0561,
      "step": 22790
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9587463736534119,
      "learning_rate": 8.103123950285523e-05,
      "loss": 1.0614,
      "step": 22800
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0167219638824463,
      "learning_rate": 8.102284178703393e-05,
      "loss": 1.079,
      "step": 22810
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9722715616226196,
      "learning_rate": 8.101444407121263e-05,
      "loss": 1.0879,
      "step": 22820
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0199732780456543,
      "learning_rate": 8.100604635539134e-05,
      "loss": 1.0716,
      "step": 22830
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9810200929641724,
      "learning_rate": 8.099764863957004e-05,
      "loss": 1.087,
      "step": 22840
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9538648128509521,
      "learning_rate": 8.098925092374875e-05,
      "loss": 1.0954,
      "step": 22850
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.952927827835083,
      "learning_rate": 8.098085320792745e-05,
      "loss": 1.0757,
      "step": 22860
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9318287372589111,
      "learning_rate": 8.097245549210615e-05,
      "loss": 1.0864,
      "step": 22870
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.94853675365448,
      "learning_rate": 8.096405777628485e-05,
      "loss": 1.0744,
      "step": 22880
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9416312575340271,
      "learning_rate": 8.095566006046356e-05,
      "loss": 1.0826,
      "step": 22890
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9539600014686584,
      "learning_rate": 8.094726234464226e-05,
      "loss": 1.0803,
      "step": 22900
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9516565203666687,
      "learning_rate": 8.093886462882096e-05,
      "loss": 1.0729,
      "step": 22910
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9808760285377502,
      "learning_rate": 8.093046691299966e-05,
      "loss": 1.0778,
      "step": 22920
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9635851979255676,
      "learning_rate": 8.092206919717838e-05,
      "loss": 1.1096,
      "step": 22930
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9341956973075867,
      "learning_rate": 8.091367148135708e-05,
      "loss": 1.0787,
      "step": 22940
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9545960426330566,
      "learning_rate": 8.090527376553578e-05,
      "loss": 1.106,
      "step": 22950
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.996005654335022,
      "learning_rate": 8.089687604971448e-05,
      "loss": 1.0722,
      "step": 22960
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9320337176322937,
      "learning_rate": 8.088847833389318e-05,
      "loss": 1.0923,
      "step": 22970
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9646962881088257,
      "learning_rate": 8.088008061807188e-05,
      "loss": 1.088,
      "step": 22980
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9459730386734009,
      "learning_rate": 8.08716829022506e-05,
      "loss": 1.0723,
      "step": 22990
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0047719478607178,
      "learning_rate": 8.08632851864293e-05,
      "loss": 1.0615,
      "step": 23000
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0436536073684692,
      "learning_rate": 8.0854887470608e-05,
      "loss": 1.0672,
      "step": 23010
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9911855459213257,
      "learning_rate": 8.08464897547867e-05,
      "loss": 1.0861,
      "step": 23020
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9437386989593506,
      "learning_rate": 8.083809203896541e-05,
      "loss": 1.0601,
      "step": 23030
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9278973340988159,
      "learning_rate": 8.082969432314411e-05,
      "loss": 1.0567,
      "step": 23040
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.962884783744812,
      "learning_rate": 8.082129660732281e-05,
      "loss": 1.0636,
      "step": 23050
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9525705575942993,
      "learning_rate": 8.081289889150151e-05,
      "loss": 1.0779,
      "step": 23060
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9483520984649658,
      "learning_rate": 8.080450117568021e-05,
      "loss": 1.0593,
      "step": 23070
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9891012907028198,
      "learning_rate": 8.079610345985892e-05,
      "loss": 1.0769,
      "step": 23080
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9191649556159973,
      "learning_rate": 8.078770574403762e-05,
      "loss": 1.0844,
      "step": 23090
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8751499652862549,
      "learning_rate": 8.077930802821633e-05,
      "loss": 1.063,
      "step": 23100
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9491960406303406,
      "learning_rate": 8.077091031239503e-05,
      "loss": 1.0772,
      "step": 23110
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8733270764350891,
      "learning_rate": 8.076251259657373e-05,
      "loss": 1.1049,
      "step": 23120
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9507169723510742,
      "learning_rate": 8.075411488075245e-05,
      "loss": 1.0715,
      "step": 23130
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9581748843193054,
      "learning_rate": 8.074571716493115e-05,
      "loss": 1.0776,
      "step": 23140
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9720478057861328,
      "learning_rate": 8.073731944910985e-05,
      "loss": 1.0703,
      "step": 23150
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9863296151161194,
      "learning_rate": 8.072892173328855e-05,
      "loss": 1.0729,
      "step": 23160
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9228429198265076,
      "learning_rate": 8.072052401746725e-05,
      "loss": 1.0739,
      "step": 23170
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.95717453956604,
      "learning_rate": 8.071212630164595e-05,
      "loss": 1.0908,
      "step": 23180
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.926839292049408,
      "learning_rate": 8.070372858582465e-05,
      "loss": 1.0702,
      "step": 23190
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9715284705162048,
      "learning_rate": 8.069533087000336e-05,
      "loss": 1.1033,
      "step": 23200
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9391289949417114,
      "learning_rate": 8.068693315418207e-05,
      "loss": 1.0439,
      "step": 23210
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9625375270843506,
      "learning_rate": 8.067853543836077e-05,
      "loss": 1.0894,
      "step": 23220
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9507083892822266,
      "learning_rate": 8.067013772253948e-05,
      "loss": 1.0776,
      "step": 23230
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9163172841072083,
      "learning_rate": 8.066174000671818e-05,
      "loss": 1.049,
      "step": 23240
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9412251114845276,
      "learning_rate": 8.065334229089688e-05,
      "loss": 1.0854,
      "step": 23250
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9556521773338318,
      "learning_rate": 8.064494457507558e-05,
      "loss": 1.1001,
      "step": 23260
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9202659130096436,
      "learning_rate": 8.06365468592543e-05,
      "loss": 1.0694,
      "step": 23270
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9295778870582581,
      "learning_rate": 8.062814914343298e-05,
      "loss": 1.0734,
      "step": 23280
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9153093099594116,
      "learning_rate": 8.061975142761168e-05,
      "loss": 1.0742,
      "step": 23290
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9639344811439514,
      "learning_rate": 8.06113537117904e-05,
      "loss": 1.0628,
      "step": 23300
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.92649245262146,
      "learning_rate": 8.06029559959691e-05,
      "loss": 1.0664,
      "step": 23310
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9394589066505432,
      "learning_rate": 8.05945582801478e-05,
      "loss": 1.0517,
      "step": 23320
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.853520393371582,
      "learning_rate": 8.05861605643265e-05,
      "loss": 1.0742,
      "step": 23330
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8790643215179443,
      "learning_rate": 8.057776284850521e-05,
      "loss": 1.066,
      "step": 23340
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9281080365180969,
      "learning_rate": 8.056936513268392e-05,
      "loss": 1.0934,
      "step": 23350
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9617061614990234,
      "learning_rate": 8.056096741686262e-05,
      "loss": 1.0645,
      "step": 23360
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0057429075241089,
      "learning_rate": 8.055256970104133e-05,
      "loss": 1.0467,
      "step": 23370
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.934522271156311,
      "learning_rate": 8.054417198522003e-05,
      "loss": 1.1074,
      "step": 23380
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8520189523696899,
      "learning_rate": 8.053577426939872e-05,
      "loss": 1.0724,
      "step": 23390
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9682310223579407,
      "learning_rate": 8.052737655357743e-05,
      "loss": 1.0724,
      "step": 23400
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9934664368629456,
      "learning_rate": 8.051897883775613e-05,
      "loss": 1.053,
      "step": 23410
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.938259482383728,
      "learning_rate": 8.051058112193483e-05,
      "loss": 1.054,
      "step": 23420
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9081432819366455,
      "learning_rate": 8.050218340611353e-05,
      "loss": 1.0939,
      "step": 23430
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.004728078842163,
      "learning_rate": 8.049378569029225e-05,
      "loss": 1.0719,
      "step": 23440
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.04097580909729,
      "learning_rate": 8.048538797447095e-05,
      "loss": 1.0872,
      "step": 23450
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9404523372650146,
      "learning_rate": 8.047699025864965e-05,
      "loss": 1.0627,
      "step": 23460
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9675320982933044,
      "learning_rate": 8.046859254282836e-05,
      "loss": 1.0611,
      "step": 23470
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0386711359024048,
      "learning_rate": 8.046019482700707e-05,
      "loss": 1.0842,
      "step": 23480
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9149951934814453,
      "learning_rate": 8.045179711118575e-05,
      "loss": 1.0578,
      "step": 23490
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9119037389755249,
      "learning_rate": 8.044339939536447e-05,
      "loss": 1.1014,
      "step": 23500
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9563122987747192,
      "learning_rate": 8.043500167954317e-05,
      "loss": 1.068,
      "step": 23510
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9674993753433228,
      "learning_rate": 8.042660396372187e-05,
      "loss": 1.0653,
      "step": 23520
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9631643891334534,
      "learning_rate": 8.041820624790057e-05,
      "loss": 1.0811,
      "step": 23530
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9869856238365173,
      "learning_rate": 8.040980853207928e-05,
      "loss": 1.0715,
      "step": 23540
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9949793815612793,
      "learning_rate": 8.040141081625798e-05,
      "loss": 1.0768,
      "step": 23550
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0151121616363525,
      "learning_rate": 8.039301310043668e-05,
      "loss": 1.0648,
      "step": 23560
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9796632528305054,
      "learning_rate": 8.038461538461538e-05,
      "loss": 1.0791,
      "step": 23570
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9762797951698303,
      "learning_rate": 8.03762176687941e-05,
      "loss": 1.0559,
      "step": 23580
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.948366641998291,
      "learning_rate": 8.03678199529728e-05,
      "loss": 1.0756,
      "step": 23590
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9642534255981445,
      "learning_rate": 8.03594222371515e-05,
      "loss": 1.0903,
      "step": 23600
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9076772332191467,
      "learning_rate": 8.03510245213302e-05,
      "loss": 1.0742,
      "step": 23610
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9649861454963684,
      "learning_rate": 8.03426268055089e-05,
      "loss": 1.1095,
      "step": 23620
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0080221891403198,
      "learning_rate": 8.03342290896876e-05,
      "loss": 1.0679,
      "step": 23630
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9600275754928589,
      "learning_rate": 8.032583137386632e-05,
      "loss": 1.0607,
      "step": 23640
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.002281904220581,
      "learning_rate": 8.031743365804502e-05,
      "loss": 1.0769,
      "step": 23650
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9854843616485596,
      "learning_rate": 8.030903594222372e-05,
      "loss": 1.0536,
      "step": 23660
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9507874846458435,
      "learning_rate": 8.030063822640242e-05,
      "loss": 1.0959,
      "step": 23670
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9268907904624939,
      "learning_rate": 8.029224051058113e-05,
      "loss": 1.0862,
      "step": 23680
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9411166310310364,
      "learning_rate": 8.028384279475983e-05,
      "loss": 1.0633,
      "step": 23690
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.055272102355957,
      "learning_rate": 8.027544507893852e-05,
      "loss": 1.0693,
      "step": 23700
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9390666484832764,
      "learning_rate": 8.026704736311724e-05,
      "loss": 1.071,
      "step": 23710
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.902016282081604,
      "learning_rate": 8.025864964729594e-05,
      "loss": 1.0802,
      "step": 23720
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9374504089355469,
      "learning_rate": 8.025025193147464e-05,
      "loss": 1.0863,
      "step": 23730
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9502925276756287,
      "learning_rate": 8.024185421565335e-05,
      "loss": 1.0681,
      "step": 23740
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9141817092895508,
      "learning_rate": 8.023345649983205e-05,
      "loss": 1.0916,
      "step": 23750
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9485606551170349,
      "learning_rate": 8.022505878401075e-05,
      "loss": 1.0686,
      "step": 23760
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9022987484931946,
      "learning_rate": 8.021666106818945e-05,
      "loss": 1.0722,
      "step": 23770
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.999480128288269,
      "learning_rate": 8.020826335236817e-05,
      "loss": 1.0986,
      "step": 23780
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.994338870048523,
      "learning_rate": 8.019986563654687e-05,
      "loss": 1.0546,
      "step": 23790
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.007479190826416,
      "learning_rate": 8.019146792072557e-05,
      "loss": 1.0625,
      "step": 23800
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9863058924674988,
      "learning_rate": 8.018307020490427e-05,
      "loss": 1.065,
      "step": 23810
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9091801047325134,
      "learning_rate": 8.017467248908297e-05,
      "loss": 1.0705,
      "step": 23820
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9525049924850464,
      "learning_rate": 8.016627477326167e-05,
      "loss": 1.0779,
      "step": 23830
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.005071997642517,
      "learning_rate": 8.015787705744038e-05,
      "loss": 1.0597,
      "step": 23840
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9436593651771545,
      "learning_rate": 8.014947934161909e-05,
      "loss": 1.078,
      "step": 23850
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.3766697645187378,
      "eval_runtime": 6244.9948,
      "eval_samples_per_second": 265.867,
      "eval_steps_per_second": 4.154,
      "step": 23857
    }
  ],
  "logging_steps": 10,
  "max_steps": 119280,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 1.0174807412263879e+19,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}