{
  "best_global_step": 68838,
  "best_metric": 0.8680973729027178,
  "best_model_checkpoint": "./results/checkpoint-68838",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 68838,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.021790290246666087,
      "grad_norm": 21.07741355895996,
      "learning_rate": 1.985502193555885e-05,
      "loss": 0.656,
      "step": 500
    },
    {
      "epoch": 0.043580580493332174,
      "grad_norm": 24.045154571533203,
      "learning_rate": 1.9709753333914408e-05,
      "loss": 0.5404,
      "step": 1000
    },
    {
      "epoch": 0.06537087073999825,
      "grad_norm": 11.716601371765137,
      "learning_rate": 1.956448473226997e-05,
      "loss": 0.5397,
      "step": 1500
    },
    {
      "epoch": 0.08716116098666435,
      "grad_norm": 12.603126525878906,
      "learning_rate": 1.9419216130625527e-05,
      "loss": 0.5234,
      "step": 2000
    },
    {
      "epoch": 0.10895145123333043,
      "grad_norm": 7.865340232849121,
      "learning_rate": 1.927394752898109e-05,
      "loss": 0.5091,
      "step": 2500
    },
    {
      "epoch": 0.1307417414799965,
      "grad_norm": 48.59724807739258,
      "learning_rate": 1.912867892733665e-05,
      "loss": 0.5202,
      "step": 3000
    },
    {
      "epoch": 0.1525320317266626,
      "grad_norm": 40.003543853759766,
      "learning_rate": 1.8983410325692207e-05,
      "loss": 0.4991,
      "step": 3500
    },
    {
      "epoch": 0.1743223219733287,
      "grad_norm": 36.06184768676758,
      "learning_rate": 1.8838141724047765e-05,
      "loss": 0.4751,
      "step": 4000
    },
    {
      "epoch": 0.19611261221999476,
      "grad_norm": 8.7717924118042,
      "learning_rate": 1.8692873122403327e-05,
      "loss": 0.4807,
      "step": 4500
    },
    {
      "epoch": 0.21790290246666086,
      "grad_norm": 31.210308074951172,
      "learning_rate": 1.8547604520758884e-05,
      "loss": 0.4816,
      "step": 5000
    },
    {
      "epoch": 0.23969319271332695,
      "grad_norm": 32.92966079711914,
      "learning_rate": 1.8402335919114446e-05,
      "loss": 0.4734,
      "step": 5500
    },
    {
      "epoch": 0.261483482959993,
      "grad_norm": 16.130868911743164,
      "learning_rate": 1.8257067317470003e-05,
      "loss": 0.4671,
      "step": 6000
    },
    {
      "epoch": 0.2832737732066591,
      "grad_norm": 24.621110916137695,
      "learning_rate": 1.8111798715825565e-05,
      "loss": 0.4803,
      "step": 6500
    },
    {
      "epoch": 0.3050640634533252,
      "grad_norm": 38.10552978515625,
      "learning_rate": 1.7966530114181122e-05,
      "loss": 0.4649,
      "step": 7000
    },
    {
      "epoch": 0.3268543536999913,
      "grad_norm": 38.748111724853516,
      "learning_rate": 1.7821261512536684e-05,
      "loss": 0.4609,
      "step": 7500
    },
    {
      "epoch": 0.3486446439466574,
      "grad_norm": 22.89567756652832,
      "learning_rate": 1.767599291089224e-05,
      "loss": 0.4446,
      "step": 8000
    },
    {
      "epoch": 0.37043493419332346,
      "grad_norm": 19.919641494750977,
      "learning_rate": 1.75307243092478e-05,
      "loss": 0.4438,
      "step": 8500
    },
    {
      "epoch": 0.39222522443998953,
      "grad_norm": 7.110844135284424,
      "learning_rate": 1.738545570760336e-05,
      "loss": 0.4525,
      "step": 9000
    },
    {
      "epoch": 0.41401551468665565,
      "grad_norm": 11.265097618103027,
      "learning_rate": 1.7240187105958918e-05,
      "loss": 0.4518,
      "step": 9500
    },
    {
      "epoch": 0.4358058049333217,
      "grad_norm": 16.030864715576172,
      "learning_rate": 1.709491850431448e-05,
      "loss": 0.4461,
      "step": 10000
    },
    {
      "epoch": 0.4575960951799878,
      "grad_norm": 20.574634552001953,
      "learning_rate": 1.6949649902670037e-05,
      "loss": 0.4485,
      "step": 10500
    },
    {
      "epoch": 0.4793863854266539,
      "grad_norm": 12.368102073669434,
      "learning_rate": 1.68043813010256e-05,
      "loss": 0.4266,
      "step": 11000
    },
    {
      "epoch": 0.5011766756733199,
      "grad_norm": 59.58558654785156,
      "learning_rate": 1.6659112699381156e-05,
      "loss": 0.4512,
      "step": 11500
    },
    {
      "epoch": 0.522966965919986,
      "grad_norm": 22.04444694519043,
      "learning_rate": 1.6513844097736717e-05,
      "loss": 0.4362,
      "step": 12000
    },
    {
      "epoch": 0.5447572561666522,
      "grad_norm": 27.14982032775879,
      "learning_rate": 1.6368575496092275e-05,
      "loss": 0.4315,
      "step": 12500
    },
    {
      "epoch": 0.5665475464133182,
      "grad_norm": 12.58919620513916,
      "learning_rate": 1.6223306894447833e-05,
      "loss": 0.4328,
      "step": 13000
    },
    {
      "epoch": 0.5883378366599843,
      "grad_norm": 39.1475944519043,
      "learning_rate": 1.6078038292803394e-05,
      "loss": 0.4366,
      "step": 13500
    },
    {
      "epoch": 0.6101281269066504,
      "grad_norm": 17.51034927368164,
      "learning_rate": 1.5932769691158952e-05,
      "loss": 0.4175,
      "step": 14000
    },
    {
      "epoch": 0.6319184171533165,
      "grad_norm": 7.703125476837158,
      "learning_rate": 1.5787501089514513e-05,
      "loss": 0.4039,
      "step": 14500
    },
    {
      "epoch": 0.6537087073999825,
      "grad_norm": 7.714233875274658,
      "learning_rate": 1.564223248787007e-05,
      "loss": 0.4153,
      "step": 15000
    },
    {
      "epoch": 0.6754989976466487,
      "grad_norm": 15.32762336730957,
      "learning_rate": 1.5496963886225632e-05,
      "loss": 0.4222,
      "step": 15500
    },
    {
      "epoch": 0.6972892878933148,
      "grad_norm": 11.171319007873535,
      "learning_rate": 1.535169528458119e-05,
      "loss": 0.4087,
      "step": 16000
    },
    {
      "epoch": 0.7190795781399808,
      "grad_norm": 17.960086822509766,
      "learning_rate": 1.5206426682936753e-05,
      "loss": 0.4327,
      "step": 16500
    },
    {
      "epoch": 0.7408698683866469,
      "grad_norm": 15.726662635803223,
      "learning_rate": 1.506115808129231e-05,
      "loss": 0.4146,
      "step": 17000
    },
    {
      "epoch": 0.762660158633313,
      "grad_norm": 19.11526107788086,
      "learning_rate": 1.4915889479647869e-05,
      "loss": 0.4212,
      "step": 17500
    },
    {
      "epoch": 0.7844504488799791,
      "grad_norm": 13.394328117370605,
      "learning_rate": 1.477062087800343e-05,
      "loss": 0.4239,
      "step": 18000
    },
    {
      "epoch": 0.8062407391266452,
      "grad_norm": 23.153011322021484,
      "learning_rate": 1.4625352276358988e-05,
      "loss": 0.4166,
      "step": 18500
    },
    {
      "epoch": 0.8280310293733113,
      "grad_norm": 35.252017974853516,
      "learning_rate": 1.4480083674714549e-05,
      "loss": 0.4133,
      "step": 19000
    },
    {
      "epoch": 0.8498213196199773,
      "grad_norm": 26.121305465698242,
      "learning_rate": 1.4334815073070107e-05,
      "loss": 0.4154,
      "step": 19500
    },
    {
      "epoch": 0.8716116098666434,
      "grad_norm": 15.44404411315918,
      "learning_rate": 1.4189546471425668e-05,
      "loss": 0.4053,
      "step": 20000
    },
    {
      "epoch": 0.8934019001133096,
      "grad_norm": 33.22347640991211,
      "learning_rate": 1.4044277869781226e-05,
      "loss": 0.404,
      "step": 20500
    },
    {
      "epoch": 0.9151921903599756,
      "grad_norm": 8.3173828125,
      "learning_rate": 1.3899009268136785e-05,
      "loss": 0.3904,
      "step": 21000
    },
    {
      "epoch": 0.9369824806066417,
      "grad_norm": 17.548837661743164,
      "learning_rate": 1.3753740666492346e-05,
      "loss": 0.3899,
      "step": 21500
    },
    {
      "epoch": 0.9587727708533078,
      "grad_norm": 63.516632080078125,
      "learning_rate": 1.3608472064847904e-05,
      "loss": 0.3922,
      "step": 22000
    },
    {
      "epoch": 0.9805630610999738,
      "grad_norm": 15.121821403503418,
      "learning_rate": 1.3463203463203465e-05,
      "loss": 0.3987,
      "step": 22500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8653213328393663,
      "eval_f1": 0.8643315347438766,
      "eval_loss": 0.38306593894958496,
      "eval_runtime": 101.9836,
      "eval_samples_per_second": 449.945,
      "eval_steps_per_second": 28.122,
      "step": 22946
    },
    {
      "epoch": 1.0023533513466398,
      "grad_norm": 15.860408782958984,
      "learning_rate": 1.3317934861559023e-05,
      "loss": 0.3836,
      "step": 23000
    },
    {
      "epoch": 1.024143641593306,
      "grad_norm": 5.384024143218994,
      "learning_rate": 1.3172666259914584e-05,
      "loss": 0.3025,
      "step": 23500
    },
    {
      "epoch": 1.045933931839972,
      "grad_norm": 16.427749633789062,
      "learning_rate": 1.3027397658270142e-05,
      "loss": 0.2983,
      "step": 24000
    },
    {
      "epoch": 1.067724222086638,
      "grad_norm": 5.875234603881836,
      "learning_rate": 1.2882129056625703e-05,
      "loss": 0.2798,
      "step": 24500
    },
    {
      "epoch": 1.0895145123333043,
      "grad_norm": 25.95697021484375,
      "learning_rate": 1.2736860454981261e-05,
      "loss": 0.2847,
      "step": 25000
    },
    {
      "epoch": 1.1113048025799703,
      "grad_norm": 32.1990852355957,
      "learning_rate": 1.2591591853336819e-05,
      "loss": 0.289,
      "step": 25500
    },
    {
      "epoch": 1.1330950928266366,
      "grad_norm": 54.05229949951172,
      "learning_rate": 1.244632325169238e-05,
      "loss": 0.3002,
      "step": 26000
    },
    {
      "epoch": 1.1548853830733026,
      "grad_norm": 19.801795959472656,
      "learning_rate": 1.2301054650047938e-05,
      "loss": 0.2849,
      "step": 26500
    },
    {
      "epoch": 1.1766756733199686,
      "grad_norm": 25.528541564941406,
      "learning_rate": 1.21557860484035e-05,
      "loss": 0.2865,
      "step": 27000
    },
    {
      "epoch": 1.1984659635666346,
      "grad_norm": 16.505584716796875,
      "learning_rate": 1.2010517446759057e-05,
      "loss": 0.2843,
      "step": 27500
    },
    {
      "epoch": 1.2202562538133008,
      "grad_norm": 27.86761474609375,
      "learning_rate": 1.1865248845114618e-05,
      "loss": 0.3044,
      "step": 28000
    },
    {
      "epoch": 1.2420465440599668,
      "grad_norm": 14.01541519165039,
      "learning_rate": 1.1719980243470178e-05,
      "loss": 0.3004,
      "step": 28500
    },
    {
      "epoch": 1.263836834306633,
      "grad_norm": 15.49959659576416,
      "learning_rate": 1.1574711641825737e-05,
      "loss": 0.283,
      "step": 29000
    },
    {
      "epoch": 1.285627124553299,
      "grad_norm": 13.934378623962402,
      "learning_rate": 1.1429443040181297e-05,
      "loss": 0.2925,
      "step": 29500
    },
    {
      "epoch": 1.307417414799965,
      "grad_norm": 34.908287048339844,
      "learning_rate": 1.1284174438536855e-05,
      "loss": 0.2888,
      "step": 30000
    },
    {
      "epoch": 1.329207705046631,
      "grad_norm": 1.3758283853530884,
      "learning_rate": 1.1138905836892416e-05,
      "loss": 0.291,
      "step": 30500
    },
    {
      "epoch": 1.3509979952932973,
      "grad_norm": 42.13234329223633,
      "learning_rate": 1.0993637235247974e-05,
      "loss": 0.2792,
      "step": 31000
    },
    {
      "epoch": 1.3727882855399633,
      "grad_norm": 17.52948570251465,
      "learning_rate": 1.0848368633603535e-05,
      "loss": 0.2941,
      "step": 31500
    },
    {
      "epoch": 1.3945785757866296,
      "grad_norm": 17.316434860229492,
      "learning_rate": 1.0703100031959093e-05,
      "loss": 0.2854,
      "step": 32000
    },
    {
      "epoch": 1.4163688660332956,
      "grad_norm": 5.664457321166992,
      "learning_rate": 1.0557831430314654e-05,
      "loss": 0.2919,
      "step": 32500
    },
    {
      "epoch": 1.4381591562799616,
      "grad_norm": 4.736696720123291,
      "learning_rate": 1.0412562828670212e-05,
      "loss": 0.3005,
      "step": 33000
    },
    {
      "epoch": 1.4599494465266276,
      "grad_norm": 32.045555114746094,
      "learning_rate": 1.0267294227025773e-05,
      "loss": 0.2932,
      "step": 33500
    },
    {
      "epoch": 1.4817397367732938,
      "grad_norm": 12.678611755371094,
      "learning_rate": 1.012202562538133e-05,
      "loss": 0.3073,
      "step": 34000
    },
    {
      "epoch": 1.5035300270199599,
      "grad_norm": 73.06298828125,
      "learning_rate": 9.97675702373689e-06,
      "loss": 0.2799,
      "step": 34500
    },
    {
      "epoch": 1.525320317266626,
      "grad_norm": 9.1071138381958,
      "learning_rate": 9.83148842209245e-06,
      "loss": 0.2823,
      "step": 35000
    },
    {
      "epoch": 1.547110607513292,
      "grad_norm": 36.0024299621582,
      "learning_rate": 9.68621982044801e-06,
      "loss": 0.2923,
      "step": 35500
    },
    {
      "epoch": 1.5689008977599581,
      "grad_norm": 17.55816650390625,
      "learning_rate": 9.540951218803569e-06,
      "loss": 0.2886,
      "step": 36000
    },
    {
      "epoch": 1.5906911880066241,
      "grad_norm": 30.378305435180664,
      "learning_rate": 9.395682617159128e-06,
      "loss": 0.2917,
      "step": 36500
    },
    {
      "epoch": 1.6124814782532904,
      "grad_norm": 10.106338500976562,
      "learning_rate": 9.250414015514688e-06,
      "loss": 0.2907,
      "step": 37000
    },
    {
      "epoch": 1.6342717684999564,
      "grad_norm": 20.659482955932617,
      "learning_rate": 9.105145413870247e-06,
      "loss": 0.2842,
      "step": 37500
    },
    {
      "epoch": 1.6560620587466226,
      "grad_norm": 18.197683334350586,
      "learning_rate": 8.959876812225807e-06,
      "loss": 0.2912,
      "step": 38000
    },
    {
      "epoch": 1.6778523489932886,
      "grad_norm": 31.622310638427734,
      "learning_rate": 8.814608210581366e-06,
      "loss": 0.302,
      "step": 38500
    },
    {
      "epoch": 1.6996426392399546,
      "grad_norm": 7.478927135467529,
      "learning_rate": 8.669339608936926e-06,
      "loss": 0.2962,
      "step": 39000
    },
    {
      "epoch": 1.7214329294866206,
      "grad_norm": 1.150900959968567,
      "learning_rate": 8.524071007292485e-06,
      "loss": 0.2905,
      "step": 39500
    },
    {
      "epoch": 1.7432232197332869,
      "grad_norm": 24.470115661621094,
      "learning_rate": 8.378802405648045e-06,
      "loss": 0.2814,
      "step": 40000
    },
    {
      "epoch": 1.765013509979953,
      "grad_norm": 14.964280128479004,
      "learning_rate": 8.233533804003603e-06,
      "loss": 0.2935,
      "step": 40500
    },
    {
      "epoch": 1.786803800226619,
      "grad_norm": 22.0378360748291,
      "learning_rate": 8.088265202359162e-06,
      "loss": 0.2777,
      "step": 41000
    },
    {
      "epoch": 1.8085940904732851,
      "grad_norm": 1.5936992168426514,
      "learning_rate": 7.942996600714722e-06,
      "loss": 0.2557,
      "step": 41500
    },
    {
      "epoch": 1.8303843807199511,
      "grad_norm": 38.885902404785156,
      "learning_rate": 7.797727999070281e-06,
      "loss": 0.2877,
      "step": 42000
    },
    {
      "epoch": 1.8521746709666171,
      "grad_norm": 21.789640426635742,
      "learning_rate": 7.65245939742584e-06,
      "loss": 0.2938,
      "step": 42500
    },
    {
      "epoch": 1.8739649612132834,
      "grad_norm": 31.65484046936035,
      "learning_rate": 7.5071907957814e-06,
      "loss": 0.281,
      "step": 43000
    },
    {
      "epoch": 1.8957552514599496,
      "grad_norm": 41.5484504699707,
      "learning_rate": 7.36192219413696e-06,
      "loss": 0.2819,
      "step": 43500
    },
    {
      "epoch": 1.9175455417066156,
      "grad_norm": 44.10054016113281,
      "learning_rate": 7.216653592492519e-06,
      "loss": 0.2727,
      "step": 44000
    },
    {
      "epoch": 1.9393358319532816,
      "grad_norm": 19.873380661010742,
      "learning_rate": 7.071384990848079e-06,
      "loss": 0.2861,
      "step": 44500
    },
    {
      "epoch": 1.9611261221999476,
      "grad_norm": 39.448081970214844,
      "learning_rate": 6.926116389203637e-06,
      "loss": 0.2726,
      "step": 45000
    },
    {
      "epoch": 1.9829164124466137,
      "grad_norm": 4.273163318634033,
      "learning_rate": 6.780847787559197e-06,
      "loss": 0.2792,
      "step": 45500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8681107939067709,
      "eval_f1": 0.8675808818446633,
      "eval_loss": 0.40907520055770874,
      "eval_runtime": 101.153,
      "eval_samples_per_second": 453.639,
      "eval_steps_per_second": 28.353,
      "step": 45892
    },
    {
      "epoch": 2.0047067026932797,
      "grad_norm": 12.477510452270508,
      "learning_rate": 6.6355791859147564e-06,
      "loss": 0.2583,
      "step": 46000
    },
    {
      "epoch": 2.026496992939946,
      "grad_norm": 0.07342194020748138,
      "learning_rate": 6.490310584270316e-06,
      "loss": 0.1871,
      "step": 46500
    },
    {
      "epoch": 2.048287283186612,
      "grad_norm": 69.57736206054688,
      "learning_rate": 6.3450419826258755e-06,
      "loss": 0.1979,
      "step": 47000
    },
    {
      "epoch": 2.070077573433278,
      "grad_norm": 7.542209148406982,
      "learning_rate": 6.199773380981435e-06,
      "loss": 0.193,
      "step": 47500
    },
    {
      "epoch": 2.091867863679944,
      "grad_norm": 0.9572351574897766,
      "learning_rate": 6.0545047793369945e-06,
      "loss": 0.1914,
      "step": 48000
    },
    {
      "epoch": 2.11365815392661,
      "grad_norm": 77.32627868652344,
      "learning_rate": 5.909236177692555e-06,
      "loss": 0.1896,
      "step": 48500
    },
    {
      "epoch": 2.135448444173276,
      "grad_norm": 2.729917049407959,
      "learning_rate": 5.763967576048114e-06,
      "loss": 0.1963,
      "step": 49000
    },
    {
      "epoch": 2.1572387344199426,
      "grad_norm": 2.4020707607269287,
      "learning_rate": 5.618698974403672e-06,
      "loss": 0.1812,
      "step": 49500
    },
    {
      "epoch": 2.1790290246666086,
      "grad_norm": 1.194051742553711,
      "learning_rate": 5.473430372759232e-06,
      "loss": 0.1933,
      "step": 50000
    },
    {
      "epoch": 2.2008193149132746,
      "grad_norm": 3.58627986907959,
      "learning_rate": 5.328161771114791e-06,
      "loss": 0.1738,
      "step": 50500
    },
    {
      "epoch": 2.2226096051599407,
      "grad_norm": 69.77458190917969,
      "learning_rate": 5.182893169470351e-06,
      "loss": 0.1827,
      "step": 51000
    },
    {
      "epoch": 2.2443998954066067,
      "grad_norm": 12.781591415405273,
      "learning_rate": 5.03762456782591e-06,
      "loss": 0.1799,
      "step": 51500
    },
    {
      "epoch": 2.266190185653273,
      "grad_norm": 23.71147346496582,
      "learning_rate": 4.8923559661814705e-06,
      "loss": 0.2041,
      "step": 52000
    },
    {
      "epoch": 2.287980475899939,
      "grad_norm": 24.240156173706055,
      "learning_rate": 4.747087364537029e-06,
      "loss": 0.1913,
      "step": 52500
    },
    {
      "epoch": 2.309770766146605,
      "grad_norm": 27.50016975402832,
      "learning_rate": 4.601818762892589e-06,
      "loss": 0.1809,
      "step": 53000
    },
    {
      "epoch": 2.331561056393271,
      "grad_norm": 0.6681903004646301,
      "learning_rate": 4.456550161248148e-06,
      "loss": 0.1896,
      "step": 53500
    },
    {
      "epoch": 2.353351346639937,
      "grad_norm": 4.321633338928223,
      "learning_rate": 4.311281559603708e-06,
      "loss": 0.1975,
      "step": 54000
    },
    {
      "epoch": 2.375141636886603,
      "grad_norm": 57.36803436279297,
      "learning_rate": 4.166012957959267e-06,
      "loss": 0.1911,
      "step": 54500
    },
    {
      "epoch": 2.396931927133269,
      "grad_norm": 0.14693371951580048,
      "learning_rate": 4.020744356314826e-06,
      "loss": 0.1979,
      "step": 55000
    },
    {
      "epoch": 2.4187222173799356,
      "grad_norm": 66.72421264648438,
      "learning_rate": 3.875475754670385e-06,
      "loss": 0.1894,
      "step": 55500
    },
    {
      "epoch": 2.4405125076266017,
      "grad_norm": 40.616798400878906,
      "learning_rate": 3.7302071530259454e-06,
      "loss": 0.1932,
      "step": 56000
    },
    {
      "epoch": 2.4623027978732677,
      "grad_norm": 14.315664291381836,
      "learning_rate": 3.584938551381505e-06,
      "loss": 0.1989,
      "step": 56500
    },
    {
      "epoch": 2.4840930881199337,
      "grad_norm": 3.2014641761779785,
      "learning_rate": 3.439669949737064e-06,
      "loss": 0.1854,
      "step": 57000
    },
    {
      "epoch": 2.5058833783665997,
      "grad_norm": 43.74154281616211,
      "learning_rate": 3.2944013480926235e-06,
      "loss": 0.191,
      "step": 57500
    },
    {
      "epoch": 2.527673668613266,
      "grad_norm": 12.705315589904785,
      "learning_rate": 3.149132746448183e-06,
      "loss": 0.1838,
      "step": 58000
    },
    {
      "epoch": 2.549463958859932,
      "grad_norm": 25.88142967224121,
      "learning_rate": 3.0038641448037425e-06,
      "loss": 0.1824,
      "step": 58500
    },
    {
      "epoch": 2.571254249106598,
      "grad_norm": 2.00431752204895,
      "learning_rate": 2.8585955431593016e-06,
      "loss": 0.2008,
      "step": 59000
    },
    {
      "epoch": 2.593044539353264,
      "grad_norm": 14.074618339538574,
      "learning_rate": 2.713326941514861e-06,
      "loss": 0.1771,
      "step": 59500
    },
    {
      "epoch": 2.61483482959993,
      "grad_norm": 24.28592300415039,
      "learning_rate": 2.5680583398704206e-06,
      "loss": 0.1867,
      "step": 60000
    },
    {
      "epoch": 2.636625119846596,
      "grad_norm": 46.3328742980957,
      "learning_rate": 2.42278973822598e-06,
      "loss": 0.186,
      "step": 60500
    },
    {
      "epoch": 2.658415410093262,
      "grad_norm": 15.886663436889648,
      "learning_rate": 2.2775211365815396e-06,
      "loss": 0.1874,
      "step": 61000
    },
    {
      "epoch": 2.6802057003399287,
      "grad_norm": 43.348026275634766,
      "learning_rate": 2.132252534937099e-06,
      "loss": 0.2014,
      "step": 61500
    },
    {
      "epoch": 2.7019959905865947,
      "grad_norm": 4.417407035827637,
      "learning_rate": 1.986983933292658e-06,
      "loss": 0.1874,
      "step": 62000
    },
    {
      "epoch": 2.7237862808332607,
      "grad_norm": 13.068084716796875,
      "learning_rate": 1.841715331648218e-06,
      "loss": 0.1954,
      "step": 62500
    },
    {
      "epoch": 2.7455765710799267,
      "grad_norm": 17.85508155822754,
      "learning_rate": 1.696446730003777e-06,
      "loss": 0.1803,
      "step": 63000
    },
    {
      "epoch": 2.7673668613265927,
      "grad_norm": 80.31468963623047,
      "learning_rate": 1.5511781283593367e-06,
      "loss": 0.1851,
      "step": 63500
    },
    {
      "epoch": 2.789157151573259,
      "grad_norm": 28.26786994934082,
      "learning_rate": 1.4059095267148958e-06,
      "loss": 0.1861,
      "step": 64000
    },
    {
      "epoch": 2.810947441819925,
      "grad_norm": 10.325495719909668,
      "learning_rate": 1.2606409250704555e-06,
      "loss": 0.187,
      "step": 64500
    },
    {
      "epoch": 2.832737732066591,
      "grad_norm": 8.102193832397461,
      "learning_rate": 1.1153723234260148e-06,
      "loss": 0.1712,
      "step": 65000
    },
    {
      "epoch": 2.854528022313257,
      "grad_norm": 36.713531494140625,
      "learning_rate": 9.701037217815743e-07,
      "loss": 0.1864,
      "step": 65500
    },
    {
      "epoch": 2.876318312559923,
      "grad_norm": 6.2764105796813965,
      "learning_rate": 8.248351201371336e-07,
      "loss": 0.1929,
      "step": 66000
    },
    {
      "epoch": 2.898108602806589,
      "grad_norm": 10.849760055541992,
      "learning_rate": 6.79566518492693e-07,
      "loss": 0.1925,
      "step": 66500
    },
    {
      "epoch": 2.9198988930532552,
      "grad_norm": 30.166927337646484,
      "learning_rate": 5.342979168482524e-07,
      "loss": 0.1923,
      "step": 67000
    },
    {
      "epoch": 2.9416891832999217,
      "grad_norm": 0.4179909825325012,
      "learning_rate": 3.890293152038119e-07,
      "loss": 0.1784,
      "step": 67500
    },
    {
      "epoch": 2.9634794735465877,
      "grad_norm": 1.3166455030441284,
      "learning_rate": 2.437607135593713e-07,
      "loss": 0.1518,
      "step": 68000
    },
    {
      "epoch": 2.9852697637932537,
      "grad_norm": 82.50801086425781,
      "learning_rate": 9.849211191493072e-08,
      "loss": 0.1817,
      "step": 68500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8686120251923203,
      "eval_f1": 0.8680973729027178,
      "eval_loss": 0.5957307815551758,
      "eval_runtime": 101.1287,
      "eval_samples_per_second": 453.749,
      "eval_steps_per_second": 28.36,
      "step": 68838
    }
  ],
  "logging_steps": 500,
  "max_steps": 68838,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.244734402325146e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}