{
  "best_global_step": 68838,
  "best_metric": 0.8178100047087637,
  "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": 54.483943939208984,
      "learning_rate": 1.985502193555885e-05,
      "loss": 1.0336,
      "step": 500
    },
    {
      "epoch": 0.043580580493332174,
      "grad_norm": 25.309688568115234,
      "learning_rate": 1.9709753333914408e-05,
      "loss": 0.8222,
      "step": 1000
    },
    {
      "epoch": 0.06537087073999825,
      "grad_norm": 20.759492874145508,
      "learning_rate": 1.956448473226997e-05,
      "loss": 0.8082,
      "step": 1500
    },
    {
      "epoch": 0.08716116098666435,
      "grad_norm": 48.492088317871094,
      "learning_rate": 1.9419216130625527e-05,
      "loss": 0.7739,
      "step": 2000
    },
    {
      "epoch": 0.10895145123333043,
      "grad_norm": 14.695459365844727,
      "learning_rate": 1.927394752898109e-05,
      "loss": 0.7577,
      "step": 2500
    },
    {
      "epoch": 0.1307417414799965,
      "grad_norm": 45.78233337402344,
      "learning_rate": 1.912867892733665e-05,
      "loss": 0.7493,
      "step": 3000
    },
    {
      "epoch": 0.1525320317266626,
      "grad_norm": 65.063720703125,
      "learning_rate": 1.8983410325692207e-05,
      "loss": 0.7372,
      "step": 3500
    },
    {
      "epoch": 0.1743223219733287,
      "grad_norm": 43.56230163574219,
      "learning_rate": 1.8838141724047765e-05,
      "loss": 0.6989,
      "step": 4000
    },
    {
      "epoch": 0.19611261221999476,
      "grad_norm": 12.578514099121094,
      "learning_rate": 1.8692873122403327e-05,
      "loss": 0.6864,
      "step": 4500
    },
    {
      "epoch": 0.21790290246666086,
      "grad_norm": 32.62320327758789,
      "learning_rate": 1.8547604520758884e-05,
      "loss": 0.6947,
      "step": 5000
    },
    {
      "epoch": 0.23969319271332695,
      "grad_norm": 30.916015625,
      "learning_rate": 1.8402335919114446e-05,
      "loss": 0.6831,
      "step": 5500
    },
    {
      "epoch": 0.261483482959993,
      "grad_norm": 28.278722763061523,
      "learning_rate": 1.8257067317470003e-05,
      "loss": 0.6766,
      "step": 6000
    },
    {
      "epoch": 0.2832737732066591,
      "grad_norm": 31.46723747253418,
      "learning_rate": 1.8111798715825565e-05,
      "loss": 0.6939,
      "step": 6500
    },
    {
      "epoch": 0.3050640634533252,
      "grad_norm": 36.962215423583984,
      "learning_rate": 1.7966530114181122e-05,
      "loss": 0.6611,
      "step": 7000
    },
    {
      "epoch": 0.3268543536999913,
      "grad_norm": 33.17079162597656,
      "learning_rate": 1.7821261512536684e-05,
      "loss": 0.6683,
      "step": 7500
    },
    {
      "epoch": 0.3486446439466574,
      "grad_norm": 39.9603271484375,
      "learning_rate": 1.767599291089224e-05,
      "loss": 0.6626,
      "step": 8000
    },
    {
      "epoch": 0.37043493419332346,
      "grad_norm": 35.70758819580078,
      "learning_rate": 1.75307243092478e-05,
      "loss": 0.635,
      "step": 8500
    },
    {
      "epoch": 0.39222522443998953,
      "grad_norm": 36.70634841918945,
      "learning_rate": 1.738545570760336e-05,
      "loss": 0.6663,
      "step": 9000
    },
    {
      "epoch": 0.41401551468665565,
      "grad_norm": 14.346757888793945,
      "learning_rate": 1.7240187105958918e-05,
      "loss": 0.6436,
      "step": 9500
    },
    {
      "epoch": 0.4358058049333217,
      "grad_norm": 20.595735549926758,
      "learning_rate": 1.709491850431448e-05,
      "loss": 0.6469,
      "step": 10000
    },
    {
      "epoch": 0.4575960951799878,
      "grad_norm": 20.74449920654297,
      "learning_rate": 1.6949649902670037e-05,
      "loss": 0.6503,
      "step": 10500
    },
    {
      "epoch": 0.4793863854266539,
      "grad_norm": 13.161200523376465,
      "learning_rate": 1.68043813010256e-05,
      "loss": 0.6214,
      "step": 11000
    },
    {
      "epoch": 0.5011766756733199,
      "grad_norm": 37.41954040527344,
      "learning_rate": 1.6659112699381156e-05,
      "loss": 0.6495,
      "step": 11500
    },
    {
      "epoch": 0.522966965919986,
      "grad_norm": 22.0927791595459,
      "learning_rate": 1.6513844097736717e-05,
      "loss": 0.62,
      "step": 12000
    },
    {
      "epoch": 0.5447572561666522,
      "grad_norm": 37.23984909057617,
      "learning_rate": 1.6368575496092275e-05,
      "loss": 0.6087,
      "step": 12500
    },
    {
      "epoch": 0.5665475464133182,
      "grad_norm": 12.211308479309082,
      "learning_rate": 1.6223306894447833e-05,
      "loss": 0.6237,
      "step": 13000
    },
    {
      "epoch": 0.5883378366599843,
      "grad_norm": 28.417964935302734,
      "learning_rate": 1.6078038292803394e-05,
      "loss": 0.6411,
      "step": 13500
    },
    {
      "epoch": 0.6101281269066504,
      "grad_norm": 25.028160095214844,
      "learning_rate": 1.5932769691158952e-05,
      "loss": 0.6029,
      "step": 14000
    },
    {
      "epoch": 0.6319184171533165,
      "grad_norm": 10.462093353271484,
      "learning_rate": 1.5787501089514513e-05,
      "loss": 0.5853,
      "step": 14500
    },
    {
      "epoch": 0.6537087073999825,
      "grad_norm": 15.21939468383789,
      "learning_rate": 1.564223248787007e-05,
      "loss": 0.5997,
      "step": 15000
    },
    {
      "epoch": 0.6754989976466487,
      "grad_norm": 35.362239837646484,
      "learning_rate": 1.5496963886225632e-05,
      "loss": 0.6184,
      "step": 15500
    },
    {
      "epoch": 0.6972892878933148,
      "grad_norm": 18.156673431396484,
      "learning_rate": 1.535169528458119e-05,
      "loss": 0.6039,
      "step": 16000
    },
    {
      "epoch": 0.7190795781399808,
      "grad_norm": 24.839574813842773,
      "learning_rate": 1.5206426682936753e-05,
      "loss": 0.6252,
      "step": 16500
    },
    {
      "epoch": 0.7408698683866469,
      "grad_norm": 17.548309326171875,
      "learning_rate": 1.506115808129231e-05,
      "loss": 0.5994,
      "step": 17000
    },
    {
      "epoch": 0.762660158633313,
      "grad_norm": 37.97673416137695,
      "learning_rate": 1.4915889479647869e-05,
      "loss": 0.604,
      "step": 17500
    },
    {
      "epoch": 0.7844504488799791,
      "grad_norm": 26.855255126953125,
      "learning_rate": 1.477062087800343e-05,
      "loss": 0.6024,
      "step": 18000
    },
    {
      "epoch": 0.8062407391266452,
      "grad_norm": 29.445913314819336,
      "learning_rate": 1.4625352276358988e-05,
      "loss": 0.5841,
      "step": 18500
    },
    {
      "epoch": 0.8280310293733113,
      "grad_norm": 27.386213302612305,
      "learning_rate": 1.4480083674714549e-05,
      "loss": 0.5867,
      "step": 19000
    },
    {
      "epoch": 0.8498213196199773,
      "grad_norm": 39.60457229614258,
      "learning_rate": 1.4334815073070107e-05,
      "loss": 0.5929,
      "step": 19500
    },
    {
      "epoch": 0.8716116098666434,
      "grad_norm": 23.005340576171875,
      "learning_rate": 1.4189546471425668e-05,
      "loss": 0.5857,
      "step": 20000
    },
    {
      "epoch": 0.8934019001133096,
      "grad_norm": 32.861236572265625,
      "learning_rate": 1.4044277869781226e-05,
      "loss": 0.5842,
      "step": 20500
    },
    {
      "epoch": 0.9151921903599756,
      "grad_norm": 25.834049224853516,
      "learning_rate": 1.3899009268136785e-05,
      "loss": 0.5772,
      "step": 21000
    },
    {
      "epoch": 0.9369824806066417,
      "grad_norm": 18.09916114807129,
      "learning_rate": 1.3753740666492346e-05,
      "loss": 0.5745,
      "step": 21500
    },
    {
      "epoch": 0.9587727708533078,
      "grad_norm": 38.919830322265625,
      "learning_rate": 1.3608472064847904e-05,
      "loss": 0.5673,
      "step": 22000
    },
    {
      "epoch": 0.9805630610999738,
      "grad_norm": 44.96213150024414,
      "learning_rate": 1.3463203463203465e-05,
      "loss": 0.5712,
      "step": 22500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8107525007082615,
      "eval_f1": 0.8068179660475118,
      "eval_loss": 0.5534330010414124,
      "eval_runtime": 103.9311,
      "eval_samples_per_second": 441.514,
      "eval_steps_per_second": 27.595,
      "step": 22946
    },
    {
      "epoch": 1.0023533513466398,
      "grad_norm": 10.949334144592285,
      "learning_rate": 1.3317934861559023e-05,
      "loss": 0.5614,
      "step": 23000
    },
    {
      "epoch": 1.024143641593306,
      "grad_norm": 9.04629135131836,
      "learning_rate": 1.3172666259914584e-05,
      "loss": 0.4356,
      "step": 23500
    },
    {
      "epoch": 1.045933931839972,
      "grad_norm": 20.525428771972656,
      "learning_rate": 1.3027397658270142e-05,
      "loss": 0.4322,
      "step": 24000
    },
    {
      "epoch": 1.067724222086638,
      "grad_norm": 13.471968650817871,
      "learning_rate": 1.2882129056625703e-05,
      "loss": 0.4009,
      "step": 24500
    },
    {
      "epoch": 1.0895145123333043,
      "grad_norm": 22.500341415405273,
      "learning_rate": 1.2736860454981261e-05,
      "loss": 0.4234,
      "step": 25000
    },
    {
      "epoch": 1.1113048025799703,
      "grad_norm": 27.607267379760742,
      "learning_rate": 1.2591591853336819e-05,
      "loss": 0.4179,
      "step": 25500
    },
    {
      "epoch": 1.1330950928266366,
      "grad_norm": 43.847557067871094,
      "learning_rate": 1.244632325169238e-05,
      "loss": 0.4245,
      "step": 26000
    },
    {
      "epoch": 1.1548853830733026,
      "grad_norm": 15.113216400146484,
      "learning_rate": 1.2301054650047938e-05,
      "loss": 0.4072,
      "step": 26500
    },
    {
      "epoch": 1.1766756733199686,
      "grad_norm": 23.319063186645508,
      "learning_rate": 1.21557860484035e-05,
      "loss": 0.4099,
      "step": 27000
    },
    {
      "epoch": 1.1984659635666346,
      "grad_norm": 19.947755813598633,
      "learning_rate": 1.2010517446759057e-05,
      "loss": 0.3983,
      "step": 27500
    },
    {
      "epoch": 1.2202562538133008,
      "grad_norm": 39.462379455566406,
      "learning_rate": 1.1865248845114618e-05,
      "loss": 0.4347,
      "step": 28000
    },
    {
      "epoch": 1.2420465440599668,
      "grad_norm": 9.28600788116455,
      "learning_rate": 1.1719980243470178e-05,
      "loss": 0.4239,
      "step": 28500
    },
    {
      "epoch": 1.263836834306633,
      "grad_norm": 26.335203170776367,
      "learning_rate": 1.1574711641825737e-05,
      "loss": 0.406,
      "step": 29000
    },
    {
      "epoch": 1.285627124553299,
      "grad_norm": 16.803382873535156,
      "learning_rate": 1.1429443040181297e-05,
      "loss": 0.4323,
      "step": 29500
    },
    {
      "epoch": 1.307417414799965,
      "grad_norm": 19.966100692749023,
      "learning_rate": 1.1284174438536855e-05,
      "loss": 0.4071,
      "step": 30000
    },
    {
      "epoch": 1.329207705046631,
      "grad_norm": 11.231242179870605,
      "learning_rate": 1.1138905836892416e-05,
      "loss": 0.4185,
      "step": 30500
    },
    {
      "epoch": 1.3509979952932973,
      "grad_norm": 5.798637866973877,
      "learning_rate": 1.0993637235247974e-05,
      "loss": 0.4094,
      "step": 31000
    },
    {
      "epoch": 1.3727882855399633,
      "grad_norm": 45.0400390625,
      "learning_rate": 1.0848368633603535e-05,
      "loss": 0.4182,
      "step": 31500
    },
    {
      "epoch": 1.3945785757866296,
      "grad_norm": 13.42186450958252,
      "learning_rate": 1.0703100031959093e-05,
      "loss": 0.4138,
      "step": 32000
    },
    {
      "epoch": 1.4163688660332956,
      "grad_norm": 4.720645427703857,
      "learning_rate": 1.0557831430314654e-05,
      "loss": 0.4189,
      "step": 32500
    },
    {
      "epoch": 1.4381591562799616,
      "grad_norm": 15.046038627624512,
      "learning_rate": 1.0412562828670212e-05,
      "loss": 0.4248,
      "step": 33000
    },
    {
      "epoch": 1.4599494465266276,
      "grad_norm": 26.371877670288086,
      "learning_rate": 1.0267294227025773e-05,
      "loss": 0.4203,
      "step": 33500
    },
    {
      "epoch": 1.4817397367732938,
      "grad_norm": 31.495922088623047,
      "learning_rate": 1.012202562538133e-05,
      "loss": 0.4346,
      "step": 34000
    },
    {
      "epoch": 1.5035300270199599,
      "grad_norm": 91.96415710449219,
      "learning_rate": 9.97675702373689e-06,
      "loss": 0.4121,
      "step": 34500
    },
    {
      "epoch": 1.525320317266626,
      "grad_norm": 15.824524879455566,
      "learning_rate": 9.83148842209245e-06,
      "loss": 0.4175,
      "step": 35000
    },
    {
      "epoch": 1.547110607513292,
      "grad_norm": 32.88704299926758,
      "learning_rate": 9.68621982044801e-06,
      "loss": 0.4134,
      "step": 35500
    },
    {
      "epoch": 1.5689008977599581,
      "grad_norm": 13.783001899719238,
      "learning_rate": 9.540951218803569e-06,
      "loss": 0.4131,
      "step": 36000
    },
    {
      "epoch": 1.5906911880066241,
      "grad_norm": 26.346506118774414,
      "learning_rate": 9.395682617159128e-06,
      "loss": 0.4208,
      "step": 36500
    },
    {
      "epoch": 1.6124814782532904,
      "grad_norm": 6.159854888916016,
      "learning_rate": 9.250414015514688e-06,
      "loss": 0.4132,
      "step": 37000
    },
    {
      "epoch": 1.6342717684999564,
      "grad_norm": 53.30848693847656,
      "learning_rate": 9.105145413870247e-06,
      "loss": 0.4167,
      "step": 37500
    },
    {
      "epoch": 1.6560620587466226,
      "grad_norm": 17.382976531982422,
      "learning_rate": 8.959876812225807e-06,
      "loss": 0.4213,
      "step": 38000
    },
    {
      "epoch": 1.6778523489932886,
      "grad_norm": 30.866016387939453,
      "learning_rate": 8.814608210581366e-06,
      "loss": 0.4295,
      "step": 38500
    },
    {
      "epoch": 1.6996426392399546,
      "grad_norm": 12.70687198638916,
      "learning_rate": 8.669339608936926e-06,
      "loss": 0.4101,
      "step": 39000
    },
    {
      "epoch": 1.7214329294866206,
      "grad_norm": 17.50055694580078,
      "learning_rate": 8.524071007292485e-06,
      "loss": 0.4142,
      "step": 39500
    },
    {
      "epoch": 1.7432232197332869,
      "grad_norm": 37.7745361328125,
      "learning_rate": 8.378802405648045e-06,
      "loss": 0.4008,
      "step": 40000
    },
    {
      "epoch": 1.765013509979953,
      "grad_norm": 31.711212158203125,
      "learning_rate": 8.233533804003603e-06,
      "loss": 0.4162,
      "step": 40500
    },
    {
      "epoch": 1.786803800226619,
      "grad_norm": 36.37419509887695,
      "learning_rate": 8.088265202359162e-06,
      "loss": 0.4004,
      "step": 41000
    },
    {
      "epoch": 1.8085940904732851,
      "grad_norm": 1.8152663707733154,
      "learning_rate": 7.942996600714722e-06,
      "loss": 0.3669,
      "step": 41500
    },
    {
      "epoch": 1.8303843807199511,
      "grad_norm": 32.72704315185547,
      "learning_rate": 7.797727999070281e-06,
      "loss": 0.4091,
      "step": 42000
    },
    {
      "epoch": 1.8521746709666171,
      "grad_norm": 25.534046173095703,
      "learning_rate": 7.65245939742584e-06,
      "loss": 0.4194,
      "step": 42500
    },
    {
      "epoch": 1.8739649612132834,
      "grad_norm": 64.45800018310547,
      "learning_rate": 7.5071907957814e-06,
      "loss": 0.406,
      "step": 43000
    },
    {
      "epoch": 1.8957552514599496,
      "grad_norm": 72.69434356689453,
      "learning_rate": 7.36192219413696e-06,
      "loss": 0.4078,
      "step": 43500
    },
    {
      "epoch": 1.9175455417066156,
      "grad_norm": 34.615509033203125,
      "learning_rate": 7.216653592492519e-06,
      "loss": 0.4035,
      "step": 44000
    },
    {
      "epoch": 1.9393358319532816,
      "grad_norm": 31.702251434326172,
      "learning_rate": 7.071384990848079e-06,
      "loss": 0.4107,
      "step": 44500
    },
    {
      "epoch": 1.9611261221999476,
      "grad_norm": 39.912506103515625,
      "learning_rate": 6.926116389203637e-06,
      "loss": 0.3938,
      "step": 45000
    },
    {
      "epoch": 1.9829164124466137,
      "grad_norm": 9.400127410888672,
      "learning_rate": 6.780847787559197e-06,
      "loss": 0.3948,
      "step": 45500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.818096628674788,
      "eval_f1": 0.8153729521200931,
      "eval_loss": 0.5731034278869629,
      "eval_runtime": 103.9443,
      "eval_samples_per_second": 441.458,
      "eval_steps_per_second": 27.592,
      "step": 45892
    },
    {
      "epoch": 2.0047067026932797,
      "grad_norm": 5.7501606941223145,
      "learning_rate": 6.6355791859147564e-06,
      "loss": 0.3754,
      "step": 46000
    },
    {
      "epoch": 2.026496992939946,
      "grad_norm": 15.377955436706543,
      "learning_rate": 6.490310584270316e-06,
      "loss": 0.2614,
      "step": 46500
    },
    {
      "epoch": 2.048287283186612,
      "grad_norm": 26.223241806030273,
      "learning_rate": 6.3450419826258755e-06,
      "loss": 0.266,
      "step": 47000
    },
    {
      "epoch": 2.070077573433278,
      "grad_norm": 14.321675300598145,
      "learning_rate": 6.199773380981435e-06,
      "loss": 0.2725,
      "step": 47500
    },
    {
      "epoch": 2.091867863679944,
      "grad_norm": 63.887062072753906,
      "learning_rate": 6.0545047793369945e-06,
      "loss": 0.2565,
      "step": 48000
    },
    {
      "epoch": 2.11365815392661,
      "grad_norm": 8.824002265930176,
      "learning_rate": 5.909236177692555e-06,
      "loss": 0.2645,
      "step": 48500
    },
    {
      "epoch": 2.135448444173276,
      "grad_norm": 58.92592239379883,
      "learning_rate": 5.763967576048114e-06,
      "loss": 0.27,
      "step": 49000
    },
    {
      "epoch": 2.1572387344199426,
      "grad_norm": 15.57720947265625,
      "learning_rate": 5.618698974403672e-06,
      "loss": 0.2563,
      "step": 49500
    },
    {
      "epoch": 2.1790290246666086,
      "grad_norm": 0.5324369072914124,
      "learning_rate": 5.473430372759232e-06,
      "loss": 0.2523,
      "step": 50000
    },
    {
      "epoch": 2.2008193149132746,
      "grad_norm": 38.53084945678711,
      "learning_rate": 5.328161771114791e-06,
      "loss": 0.2404,
      "step": 50500
    },
    {
      "epoch": 2.2226096051599407,
      "grad_norm": 5.76258659362793,
      "learning_rate": 5.182893169470351e-06,
      "loss": 0.2515,
      "step": 51000
    },
    {
      "epoch": 2.2443998954066067,
      "grad_norm": 33.10593032836914,
      "learning_rate": 5.03762456782591e-06,
      "loss": 0.2526,
      "step": 51500
    },
    {
      "epoch": 2.266190185653273,
      "grad_norm": 23.81480598449707,
      "learning_rate": 4.8923559661814705e-06,
      "loss": 0.2774,
      "step": 52000
    },
    {
      "epoch": 2.287980475899939,
      "grad_norm": 15.034322738647461,
      "learning_rate": 4.747087364537029e-06,
      "loss": 0.2488,
      "step": 52500
    },
    {
      "epoch": 2.309770766146605,
      "grad_norm": 32.9444694519043,
      "learning_rate": 4.601818762892589e-06,
      "loss": 0.241,
      "step": 53000
    },
    {
      "epoch": 2.331561056393271,
      "grad_norm": 17.02962303161621,
      "learning_rate": 4.456550161248148e-06,
      "loss": 0.263,
      "step": 53500
    },
    {
      "epoch": 2.353351346639937,
      "grad_norm": 31.476587295532227,
      "learning_rate": 4.311281559603708e-06,
      "loss": 0.2567,
      "step": 54000
    },
    {
      "epoch": 2.375141636886603,
      "grad_norm": 34.839778900146484,
      "learning_rate": 4.166012957959267e-06,
      "loss": 0.2468,
      "step": 54500
    },
    {
      "epoch": 2.396931927133269,
      "grad_norm": 0.312466561794281,
      "learning_rate": 4.020744356314826e-06,
      "loss": 0.2591,
      "step": 55000
    },
    {
      "epoch": 2.4187222173799356,
      "grad_norm": 10.837553024291992,
      "learning_rate": 3.875475754670385e-06,
      "loss": 0.2552,
      "step": 55500
    },
    {
      "epoch": 2.4405125076266017,
      "grad_norm": 2.8537871837615967,
      "learning_rate": 3.7302071530259454e-06,
      "loss": 0.2675,
      "step": 56000
    },
    {
      "epoch": 2.4623027978732677,
      "grad_norm": 17.37003517150879,
      "learning_rate": 3.584938551381505e-06,
      "loss": 0.255,
      "step": 56500
    },
    {
      "epoch": 2.4840930881199337,
      "grad_norm": 14.81467056274414,
      "learning_rate": 3.439669949737064e-06,
      "loss": 0.2538,
      "step": 57000
    },
    {
      "epoch": 2.5058833783665997,
      "grad_norm": 22.952491760253906,
      "learning_rate": 3.2944013480926235e-06,
      "loss": 0.259,
      "step": 57500
    },
    {
      "epoch": 2.527673668613266,
      "grad_norm": 51.379600524902344,
      "learning_rate": 3.149132746448183e-06,
      "loss": 0.2433,
      "step": 58000
    },
    {
      "epoch": 2.549463958859932,
      "grad_norm": 50.78517150878906,
      "learning_rate": 3.0038641448037425e-06,
      "loss": 0.245,
      "step": 58500
    },
    {
      "epoch": 2.571254249106598,
      "grad_norm": 70.40054321289062,
      "learning_rate": 2.8585955431593016e-06,
      "loss": 0.2578,
      "step": 59000
    },
    {
      "epoch": 2.593044539353264,
      "grad_norm": 13.180349349975586,
      "learning_rate": 2.713326941514861e-06,
      "loss": 0.2569,
      "step": 59500
    },
    {
      "epoch": 2.61483482959993,
      "grad_norm": 29.106149673461914,
      "learning_rate": 2.5680583398704206e-06,
      "loss": 0.2664,
      "step": 60000
    },
    {
      "epoch": 2.636625119846596,
      "grad_norm": 48.964847564697266,
      "learning_rate": 2.42278973822598e-06,
      "loss": 0.2542,
      "step": 60500
    },
    {
      "epoch": 2.658415410093262,
      "grad_norm": 18.914636611938477,
      "learning_rate": 2.2775211365815396e-06,
      "loss": 0.2515,
      "step": 61000
    },
    {
      "epoch": 2.6802057003399287,
      "grad_norm": 44.629173278808594,
      "learning_rate": 2.132252534937099e-06,
      "loss": 0.2576,
      "step": 61500
    },
    {
      "epoch": 2.7019959905865947,
      "grad_norm": 13.741023063659668,
      "learning_rate": 1.986983933292658e-06,
      "loss": 0.2532,
      "step": 62000
    },
    {
      "epoch": 2.7237862808332607,
      "grad_norm": 30.310876846313477,
      "learning_rate": 1.841715331648218e-06,
      "loss": 0.2597,
      "step": 62500
    },
    {
      "epoch": 2.7455765710799267,
      "grad_norm": 21.90140724182129,
      "learning_rate": 1.696446730003777e-06,
      "loss": 0.2435,
      "step": 63000
    },
    {
      "epoch": 2.7673668613265927,
      "grad_norm": 27.499929428100586,
      "learning_rate": 1.5511781283593367e-06,
      "loss": 0.2577,
      "step": 63500
    },
    {
      "epoch": 2.789157151573259,
      "grad_norm": 33.468074798583984,
      "learning_rate": 1.4059095267148958e-06,
      "loss": 0.2532,
      "step": 64000
    },
    {
      "epoch": 2.810947441819925,
      "grad_norm": 14.590469360351562,
      "learning_rate": 1.2606409250704555e-06,
      "loss": 0.2548,
      "step": 64500
    },
    {
      "epoch": 2.832737732066591,
      "grad_norm": 36.13336181640625,
      "learning_rate": 1.1153723234260148e-06,
      "loss": 0.2408,
      "step": 65000
    },
    {
      "epoch": 2.854528022313257,
      "grad_norm": 20.17421531677246,
      "learning_rate": 9.701037217815743e-07,
      "loss": 0.2467,
      "step": 65500
    },
    {
      "epoch": 2.876318312559923,
      "grad_norm": 129.63551330566406,
      "learning_rate": 8.248351201371336e-07,
      "loss": 0.2555,
      "step": 66000
    },
    {
      "epoch": 2.898108602806589,
      "grad_norm": 2.310012102127075,
      "learning_rate": 6.79566518492693e-07,
      "loss": 0.2478,
      "step": 66500
    },
    {
      "epoch": 2.9198988930532552,
      "grad_norm": 60.997982025146484,
      "learning_rate": 5.342979168482524e-07,
      "loss": 0.2468,
      "step": 67000
    },
    {
      "epoch": 2.9416891832999217,
      "grad_norm": 2.189000129699707,
      "learning_rate": 3.890293152038119e-07,
      "loss": 0.2387,
      "step": 67500
    },
    {
      "epoch": 2.9634794735465877,
      "grad_norm": 79.08841705322266,
      "learning_rate": 2.437607135593713e-07,
      "loss": 0.236,
      "step": 68000
    },
    {
      "epoch": 2.9852697637932537,
      "grad_norm": 4.530211925506592,
      "learning_rate": 9.849211191493072e-08,
      "loss": 0.2454,
      "step": 68500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.819404188550134,
      "eval_f1": 0.8178100047087637,
      "eval_loss": 0.7280585169792175,
      "eval_runtime": 106.044,
      "eval_samples_per_second": 432.717,
      "eval_steps_per_second": 27.045,
      "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": 7.244994585504614e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}