{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 16.515276630883566,
  "eval_steps": 2000,
  "global_step": 120000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "eval_accuracy": 1.2790688050781904e-05,
      "eval_loss": 115.5,
      "eval_runtime": 200.9168,
      "eval_samples_per_second": 8229.121,
      "eval_steps_per_second": 16.076,
      "step": 0
    },
    {
      "epoch": 0.13762730525736305,
      "grad_norm": 26.74563980102539,
      "learning_rate": 2.9639999999999997e-05,
      "loss": 141.26440625,
      "step": 1000
    },
    {
      "epoch": 0.2752546105147261,
      "grad_norm": 22.885421752929688,
      "learning_rate": 5.964e-05,
      "loss": 59.12350390625,
      "step": 2000
    },
    {
      "epoch": 0.2752546105147261,
      "eval_accuracy": 0.676248915563685,
      "eval_loss": 12.453125,
      "eval_runtime": 168.1994,
      "eval_samples_per_second": 9829.817,
      "eval_steps_per_second": 19.203,
      "step": 2000
    },
    {
      "epoch": 0.41288191577208916,
      "grad_norm": 21.04855728149414,
      "learning_rate": 8.957999999999998e-05,
      "loss": 50.17753125,
      "step": 3000
    },
    {
      "epoch": 0.5505092210294522,
      "grad_norm": 20.75350570678711,
      "learning_rate": 0.00011957999999999999,
      "loss": 46.33467578125,
      "step": 4000
    },
    {
      "epoch": 0.5505092210294522,
      "eval_accuracy": 0.7075093226563655,
      "eval_loss": 10.6796875,
      "eval_runtime": 167.9148,
      "eval_samples_per_second": 9846.476,
      "eval_steps_per_second": 19.236,
      "step": 4000
    },
    {
      "epoch": 0.6881365262868153,
      "grad_norm": 19.514423370361328,
      "learning_rate": 0.00014948999999999998,
      "loss": 44.1445078125,
      "step": 5000
    },
    {
      "epoch": 0.8257638315441783,
      "grad_norm": 18.586366653442383,
      "learning_rate": 0.00017949,
      "loss": 42.90046875,
      "step": 6000
    },
    {
      "epoch": 0.8257638315441783,
      "eval_accuracy": 0.7171687154928088,
      "eval_loss": 10.140625,
      "eval_runtime": 169.3136,
      "eval_samples_per_second": 9765.126,
      "eval_steps_per_second": 19.077,
      "step": 6000
    },
    {
      "epoch": 0.9633911368015414,
      "grad_norm": 17.66230010986328,
      "learning_rate": 0.00020946,
      "loss": 41.96175390625,
      "step": 7000
    },
    {
      "epoch": 1.1010184420589044,
      "grad_norm": 19.431285858154297,
      "learning_rate": 0.0002394,
      "loss": 41.42766015625,
      "step": 8000
    },
    {
      "epoch": 1.1010184420589044,
      "eval_accuracy": 0.7211219507641113,
      "eval_loss": 9.921875,
      "eval_runtime": 170.601,
      "eval_samples_per_second": 9691.437,
      "eval_steps_per_second": 18.933,
      "step": 8000
    },
    {
      "epoch": 1.2386457473162675,
      "grad_norm": 19.449247360229492,
      "learning_rate": 0.00026933999999999997,
      "loss": 41.1454921875,
      "step": 9000
    },
    {
      "epoch": 1.3762730525736306,
      "grad_norm": 21.419654846191406,
      "learning_rate": 0.00029934,
      "loss": 40.98086328125,
      "step": 10000
    },
    {
      "epoch": 1.3762730525736306,
      "eval_accuracy": 0.7225334877311377,
      "eval_loss": 9.8359375,
      "eval_runtime": 167.923,
      "eval_samples_per_second": 9845.998,
      "eval_steps_per_second": 19.235,
      "step": 10000
    },
    {
      "epoch": 1.5139003578309937,
      "grad_norm": 17.391115188598633,
      "learning_rate": 0.0003,
      "loss": 40.79280078125,
      "step": 11000
    },
    {
      "epoch": 1.6515276630883569,
      "grad_norm": 29.895591735839844,
      "learning_rate": 0.0003,
      "loss": 40.4991640625,
      "step": 12000
    },
    {
      "epoch": 1.6515276630883569,
      "eval_accuracy": 0.7232843609541934,
      "eval_loss": 9.78125,
      "eval_runtime": 170.7512,
      "eval_samples_per_second": 9682.91,
      "eval_steps_per_second": 18.916,
      "step": 12000
    },
    {
      "epoch": 1.7891549683457197,
      "grad_norm": 16.52672576904297,
      "learning_rate": 0.0003,
      "loss": 40.5724921875,
      "step": 13000
    },
    {
      "epoch": 1.9267822736030829,
      "grad_norm": 21.801998138427734,
      "learning_rate": 0.0003,
      "loss": 40.29026953125,
      "step": 14000
    },
    {
      "epoch": 1.9267822736030829,
      "eval_accuracy": 0.7247579731972368,
      "eval_loss": 9.7109375,
      "eval_runtime": 169.376,
      "eval_samples_per_second": 9761.53,
      "eval_steps_per_second": 19.07,
      "step": 14000
    },
    {
      "epoch": 2.0644095788604457,
      "grad_norm": 16.79710578918457,
      "learning_rate": 0.0003,
      "loss": 40.192109375,
      "step": 15000
    },
    {
      "epoch": 2.202036884117809,
      "grad_norm": 19.166486740112305,
      "learning_rate": 0.0003,
      "loss": 39.6695,
      "step": 16000
    },
    {
      "epoch": 2.202036884117809,
      "eval_accuracy": 0.726963905479327,
      "eval_loss": 9.6171875,
      "eval_runtime": 170.9384,
      "eval_samples_per_second": 9672.308,
      "eval_steps_per_second": 18.896,
      "step": 16000
    },
    {
      "epoch": 2.339664189375172,
      "grad_norm": 15.058335304260254,
      "learning_rate": 0.0003,
      "loss": 39.4829609375,
      "step": 17000
    },
    {
      "epoch": 2.477291494632535,
      "grad_norm": 15.027190208435059,
      "learning_rate": 0.0003,
      "loss": 39.3190859375,
      "step": 18000
    },
    {
      "epoch": 2.477291494632535,
      "eval_accuracy": 0.7287604009112041,
      "eval_loss": 9.546875,
      "eval_runtime": 170.8414,
      "eval_samples_per_second": 9677.8,
      "eval_steps_per_second": 18.906,
      "step": 18000
    },
    {
      "epoch": 2.614918799889898,
      "grad_norm": 17.88246726989746,
      "learning_rate": 0.0003,
      "loss": 39.1439296875,
      "step": 19000
    },
    {
      "epoch": 2.7525461051472613,
      "grad_norm": 63.608863830566406,
      "learning_rate": 0.0003,
      "loss": 39.0614765625,
      "step": 20000
    },
    {
      "epoch": 2.7525461051472613,
      "eval_accuracy": 0.7304020938141926,
      "eval_loss": 9.46875,
      "eval_runtime": 169.4479,
      "eval_samples_per_second": 9757.391,
      "eval_steps_per_second": 19.062,
      "step": 20000
    },
    {
      "epoch": 2.8901734104046244,
      "grad_norm": 16.970745086669922,
      "learning_rate": 0.0003,
      "loss": 38.933671875,
      "step": 21000
    },
    {
      "epoch": 3.0278007156619875,
      "grad_norm": 19.159212112426758,
      "learning_rate": 0.0003,
      "loss": 39.2828828125,
      "step": 22000
    },
    {
      "epoch": 3.0278007156619875,
      "eval_accuracy": 0.7285464859640399,
      "eval_loss": 9.5390625,
      "eval_runtime": 168.2153,
      "eval_samples_per_second": 9828.886,
      "eval_steps_per_second": 19.202,
      "step": 22000
    },
    {
      "epoch": 3.1654280209193506,
      "grad_norm": 28.30343246459961,
      "learning_rate": 0.0003,
      "loss": 39.1001796875,
      "step": 23000
    },
    {
      "epoch": 3.3030553261767133,
      "grad_norm": 14.642151832580566,
      "learning_rate": 0.0003,
      "loss": 38.792859375,
      "step": 24000
    },
    {
      "epoch": 3.3030553261767133,
      "eval_accuracy": 0.7310024378583697,
      "eval_loss": 9.421875,
      "eval_runtime": 168.799,
      "eval_samples_per_second": 9794.899,
      "eval_steps_per_second": 19.135,
      "step": 24000
    },
    {
      "epoch": 3.4406826314340764,
      "grad_norm": 32.9803352355957,
      "learning_rate": 0.0003,
      "loss": 38.72437890625,
      "step": 25000
    },
    {
      "epoch": 3.5783099366914395,
      "grad_norm": 15.326594352722168,
      "learning_rate": 0.0003,
      "loss": 38.7340703125,
      "step": 26000
    },
    {
      "epoch": 3.5783099366914395,
      "eval_accuracy": 0.7314286407398777,
      "eval_loss": 9.390625,
      "eval_runtime": 168.4702,
      "eval_samples_per_second": 9814.015,
      "eval_steps_per_second": 19.173,
      "step": 26000
    },
    {
      "epoch": 3.7159372419488026,
      "grad_norm": 17.456626892089844,
      "learning_rate": 0.0003,
      "loss": 38.50905078125,
      "step": 27000
    },
    {
      "epoch": 3.8535645472061657,
      "grad_norm": 19.74641990661621,
      "learning_rate": 0.0003,
      "loss": 38.4493125,
      "step": 28000
    },
    {
      "epoch": 3.8535645472061657,
      "eval_accuracy": 0.7326063700018453,
      "eval_loss": 9.3359375,
      "eval_runtime": 167.8973,
      "eval_samples_per_second": 9847.504,
      "eval_steps_per_second": 19.238,
      "step": 28000
    },
    {
      "epoch": 3.991191852463529,
      "grad_norm": 16.116609573364258,
      "learning_rate": 0.0003,
      "loss": 38.4295859375,
      "step": 29000
    },
    {
      "epoch": 4.1288191577208915,
      "grad_norm": 14.791277885437012,
      "learning_rate": 0.0003,
      "loss": 38.2604921875,
      "step": 30000
    },
    {
      "epoch": 4.1288191577208915,
      "eval_accuracy": 0.73262638229311,
      "eval_loss": 9.34375,
      "eval_runtime": 169.3505,
      "eval_samples_per_second": 9763.003,
      "eval_steps_per_second": 19.073,
      "step": 30000
    },
    {
      "epoch": 4.266446462978255,
      "grad_norm": 14.701274871826172,
      "learning_rate": 0.0003,
      "loss": 38.178046875,
      "step": 31000
    },
    {
      "epoch": 4.404073768235618,
      "grad_norm": 32.28364181518555,
      "learning_rate": 0.0003,
      "loss": 38.118671875,
      "step": 32000
    },
    {
      "epoch": 4.404073768235618,
      "eval_accuracy": 0.7330061560700385,
      "eval_loss": 9.328125,
      "eval_runtime": 169.1911,
      "eval_samples_per_second": 9772.196,
      "eval_steps_per_second": 19.091,
      "step": 32000
    },
    {
      "epoch": 4.541701073492981,
      "grad_norm": 14.65267562866211,
      "learning_rate": 0.0003,
      "loss": 38.04671484375,
      "step": 33000
    },
    {
      "epoch": 4.679328378750344,
      "grad_norm": 15.814043045043945,
      "learning_rate": 0.0003,
      "loss": 38.03919140625,
      "step": 34000
    },
    {
      "epoch": 4.679328378750344,
      "eval_accuracy": 0.7350299683164979,
      "eval_loss": 9.234375,
      "eval_runtime": 170.3006,
      "eval_samples_per_second": 9708.532,
      "eval_steps_per_second": 18.966,
      "step": 34000
    },
    {
      "epoch": 4.8169556840077075,
      "grad_norm": 22.082040786743164,
      "learning_rate": 0.0003,
      "loss": 38.00453125,
      "step": 35000
    },
    {
      "epoch": 4.95458298926507,
      "grad_norm": 20.287931442260742,
      "learning_rate": 0.0003,
      "loss": 37.977375,
      "step": 36000
    },
    {
      "epoch": 4.95458298926507,
      "eval_accuracy": 0.7339120258458323,
      "eval_loss": 9.28125,
      "eval_runtime": 171.801,
      "eval_samples_per_second": 9623.744,
      "eval_steps_per_second": 18.801,
      "step": 36000
    },
    {
      "epoch": 5.092210294522434,
      "grad_norm": 30.58030128479004,
      "learning_rate": 0.0003,
      "loss": 37.9320703125,
      "step": 37000
    },
    {
      "epoch": 5.229837599779796,
      "grad_norm": 19.570669174194336,
      "learning_rate": 0.0003,
      "loss": 37.9665234375,
      "step": 38000
    },
    {
      "epoch": 5.229837599779796,
      "eval_accuracy": 0.7340658881695064,
      "eval_loss": 9.28125,
      "eval_runtime": 167.7847,
      "eval_samples_per_second": 9854.11,
      "eval_steps_per_second": 19.251,
      "step": 38000
    },
    {
      "epoch": 5.367464905037159,
      "grad_norm": 17.29003143310547,
      "learning_rate": 0.0003,
      "loss": 37.9463984375,
      "step": 39000
    },
    {
      "epoch": 5.505092210294523,
      "grad_norm": 16.568086624145508,
      "learning_rate": 0.0003,
      "loss": 37.7375625,
      "step": 40000
    },
    {
      "epoch": 5.505092210294523,
      "eval_accuracy": 0.735108947509515,
      "eval_loss": 9.2265625,
      "eval_runtime": 170.3335,
      "eval_samples_per_second": 9706.659,
      "eval_steps_per_second": 18.963,
      "step": 40000
    },
    {
      "epoch": 5.642719515551885,
      "grad_norm": 110.49578094482422,
      "learning_rate": 0.0003,
      "loss": 37.6945,
      "step": 41000
    },
    {
      "epoch": 5.780346820809249,
      "grad_norm": 24.624027252197266,
      "learning_rate": 0.0003,
      "loss": 37.72793359375,
      "step": 42000
    },
    {
      "epoch": 5.780346820809249,
      "eval_accuracy": 0.7357611562920721,
      "eval_loss": 9.203125,
      "eval_runtime": 168.0213,
      "eval_samples_per_second": 9840.236,
      "eval_steps_per_second": 19.224,
      "step": 42000
    },
    {
      "epoch": 5.917974126066611,
      "grad_norm": 17.721742630004883,
      "learning_rate": 0.0003,
      "loss": 37.635171875,
      "step": 43000
    },
    {
      "epoch": 6.055601431323975,
      "grad_norm": 14.108975410461426,
      "learning_rate": 0.0003,
      "loss": 37.572765625,
      "step": 44000
    },
    {
      "epoch": 6.055601431323975,
      "eval_accuracy": 0.7358946142236276,
      "eval_loss": 9.1953125,
      "eval_runtime": 168.7352,
      "eval_samples_per_second": 9798.601,
      "eval_steps_per_second": 19.142,
      "step": 44000
    },
    {
      "epoch": 6.193228736581338,
      "grad_norm": 22.010786056518555,
      "learning_rate": 0.0003,
      "loss": 37.4784453125,
      "step": 45000
    },
    {
      "epoch": 6.330856041838701,
      "grad_norm": 28.03326988220215,
      "learning_rate": 0.0003,
      "loss": 37.5036328125,
      "step": 46000
    },
    {
      "epoch": 6.330856041838701,
      "eval_accuracy": 0.7368965623776919,
      "eval_loss": 9.171875,
      "eval_runtime": 166.8883,
      "eval_samples_per_second": 9907.04,
      "eval_steps_per_second": 19.354,
      "step": 46000
    },
    {
      "epoch": 6.468483347096064,
      "grad_norm": 17.36493492126465,
      "learning_rate": 0.0003,
      "loss": 37.4494453125,
      "step": 47000
    },
    {
      "epoch": 6.6061106523534265,
      "grad_norm": 19.51688003540039,
      "learning_rate": 0.0003,
      "loss": 37.435234375,
      "step": 48000
    },
    {
      "epoch": 6.6061106523534265,
      "eval_accuracy": 0.7361287527698055,
      "eval_loss": 9.1796875,
      "eval_runtime": 169.5516,
      "eval_samples_per_second": 9751.421,
      "eval_steps_per_second": 19.05,
      "step": 48000
    },
    {
      "epoch": 6.74373795761079,
      "grad_norm": 15.679317474365234,
      "learning_rate": 0.0003,
      "loss": 37.3981484375,
      "step": 49000
    },
    {
      "epoch": 6.881365262868153,
      "grad_norm": 42.613555908203125,
      "learning_rate": 0.0003,
      "loss": 37.51515625,
      "step": 50000
    },
    {
      "epoch": 6.881365262868153,
      "eval_accuracy": 0.7378156733015306,
      "eval_loss": 9.125,
      "eval_runtime": 170.8004,
      "eval_samples_per_second": 9680.126,
      "eval_steps_per_second": 18.911,
      "step": 50000
    },
    {
      "epoch": 7.018992568125516,
      "grad_norm": 15.565272331237793,
      "learning_rate": 0.0003,
      "loss": 37.4844375,
      "step": 51000
    },
    {
      "epoch": 7.156619873382879,
      "grad_norm": 15.575469017028809,
      "learning_rate": 0.0003,
      "loss": 37.2694375,
      "step": 52000
    },
    {
      "epoch": 7.156619873382879,
      "eval_accuracy": 0.7383814207101249,
      "eval_loss": 9.09375,
      "eval_runtime": 167.1627,
      "eval_samples_per_second": 9890.776,
      "eval_steps_per_second": 19.322,
      "step": 52000
    },
    {
      "epoch": 7.2942471786402425,
      "grad_norm": 15.146500587463379,
      "learning_rate": 0.0003,
      "loss": 37.243421875,
      "step": 53000
    },
    {
      "epoch": 7.431874483897605,
      "grad_norm": 17.28059959411621,
      "learning_rate": 0.0003,
      "loss": 37.23215625,
      "step": 54000
    },
    {
      "epoch": 7.431874483897605,
      "eval_accuracy": 0.7379147365824262,
      "eval_loss": 9.1015625,
      "eval_runtime": 169.7822,
      "eval_samples_per_second": 9738.179,
      "eval_steps_per_second": 19.024,
      "step": 54000
    },
    {
      "epoch": 7.569501789154968,
      "grad_norm": 29.069555282592773,
      "learning_rate": 0.0003,
      "loss": 37.22365625,
      "step": 55000
    },
    {
      "epoch": 7.707129094412331,
      "grad_norm": 26.9329833984375,
      "learning_rate": 0.0003,
      "loss": 37.22718359375,
      "step": 56000
    },
    {
      "epoch": 7.707129094412331,
      "eval_accuracy": 0.7364887765974263,
      "eval_loss": 9.171875,
      "eval_runtime": 170.0511,
      "eval_samples_per_second": 9722.778,
      "eval_steps_per_second": 18.994,
      "step": 56000
    },
    {
      "epoch": 7.844756399669695,
      "grad_norm": 97.25556182861328,
      "learning_rate": 0.0003,
      "loss": 40.3630625,
      "step": 57000
    },
    {
      "epoch": 7.982383704927058,
      "grad_norm": 19.714794158935547,
      "learning_rate": 0.0003,
      "loss": 37.60580859375,
      "step": 58000
    },
    {
      "epoch": 7.982383704927058,
      "eval_accuracy": 0.7352168040424968,
      "eval_loss": 9.234375,
      "eval_runtime": 169.0419,
      "eval_samples_per_second": 9780.822,
      "eval_steps_per_second": 19.108,
      "step": 58000
    },
    {
      "epoch": 8.12001101018442,
      "grad_norm": 14.792304039001465,
      "learning_rate": 0.0003,
      "loss": 37.54733984375,
      "step": 59000
    },
    {
      "epoch": 8.257638315441783,
      "grad_norm": 15.469705581665039,
      "learning_rate": 0.0003,
      "loss": 37.5459453125,
      "step": 60000
    },
    {
      "epoch": 8.257638315441783,
      "eval_accuracy": 0.7371664310551599,
      "eval_loss": 9.1328125,
      "eval_runtime": 166.3796,
      "eval_samples_per_second": 9937.33,
      "eval_steps_per_second": 19.413,
      "step": 60000
    },
    {
      "epoch": 8.395265620699147,
      "grad_norm": 15.161516189575195,
      "learning_rate": 0.0003,
      "loss": 37.47540625,
      "step": 61000
    },
    {
      "epoch": 8.53289292595651,
      "grad_norm": 14.802396774291992,
      "learning_rate": 0.0003,
      "loss": 37.30961328125,
      "step": 62000
    },
    {
      "epoch": 8.53289292595651,
      "eval_accuracy": 0.7381438951141913,
      "eval_loss": 9.09375,
      "eval_runtime": 168.1235,
      "eval_samples_per_second": 9834.251,
      "eval_steps_per_second": 19.212,
      "step": 62000
    },
    {
      "epoch": 8.670520231213873,
      "grad_norm": 20.38652229309082,
      "learning_rate": 0.0003,
      "loss": 37.20555859375,
      "step": 63000
    },
    {
      "epoch": 8.808147536471235,
      "grad_norm": 15.676960945129395,
      "learning_rate": 0.0003,
      "loss": 37.14574609375,
      "step": 64000
    },
    {
      "epoch": 8.808147536471235,
      "eval_accuracy": 0.7382565867319608,
      "eval_loss": 9.1015625,
      "eval_runtime": 171.8616,
      "eval_samples_per_second": 9620.35,
      "eval_steps_per_second": 18.794,
      "step": 64000
    },
    {
      "epoch": 8.9457748417286,
      "grad_norm": 16.668174743652344,
      "learning_rate": 0.0003,
      "loss": 37.125671875,
      "step": 65000
    },
    {
      "epoch": 9.083402146985962,
      "grad_norm": 16.503215789794922,
      "learning_rate": 0.0003,
      "loss": 37.01801171875,
      "step": 66000
    },
    {
      "epoch": 9.083402146985962,
      "eval_accuracy": 0.7393124677323514,
      "eval_loss": 9.046875,
      "eval_runtime": 168.277,
      "eval_samples_per_second": 9825.282,
      "eval_steps_per_second": 19.195,
      "step": 66000
    },
    {
      "epoch": 9.221029452243325,
      "grad_norm": 19.62793731689453,
      "learning_rate": 0.0003,
      "loss": 36.9910703125,
      "step": 67000
    },
    {
      "epoch": 9.358656757500688,
      "grad_norm": 18.604116439819336,
      "learning_rate": 0.0003,
      "loss": 36.9631328125,
      "step": 68000
    },
    {
      "epoch": 9.358656757500688,
      "eval_accuracy": 0.7395416473430355,
      "eval_loss": 9.03125,
      "eval_runtime": 166.5578,
      "eval_samples_per_second": 9926.697,
      "eval_steps_per_second": 19.393,
      "step": 68000
    },
    {
      "epoch": 9.49628406275805,
      "grad_norm": 20.98838233947754,
      "learning_rate": 0.0003,
      "loss": 36.963703125,
      "step": 69000
    },
    {
      "epoch": 9.633911368015415,
      "grad_norm": 17.784839630126953,
      "learning_rate": 0.0003,
      "loss": 36.96887109375,
      "step": 70000
    },
    {
      "epoch": 9.633911368015415,
      "eval_accuracy": 0.7390765595619299,
      "eval_loss": 9.0625,
      "eval_runtime": 166.5724,
      "eval_samples_per_second": 9925.827,
      "eval_steps_per_second": 19.391,
      "step": 70000
    },
    {
      "epoch": 9.771538673272778,
      "grad_norm": 27.863100051879883,
      "learning_rate": 0.0003,
      "loss": 37.0048671875,
      "step": 71000
    },
    {
      "epoch": 9.90916597853014,
      "grad_norm": 16.188459396362305,
      "learning_rate": 0.0003,
      "loss": 36.9179296875,
      "step": 72000
    },
    {
      "epoch": 9.90916597853014,
      "eval_accuracy": 0.7394581492526459,
      "eval_loss": 9.03125,
      "eval_runtime": 170.082,
      "eval_samples_per_second": 9721.013,
      "eval_steps_per_second": 18.991,
      "step": 72000
    },
    {
      "epoch": 10.046793283787503,
      "grad_norm": 14.628756523132324,
      "learning_rate": 0.0003,
      "loss": 36.888484375,
      "step": 73000
    },
    {
      "epoch": 10.184420589044867,
      "grad_norm": 14.934839248657227,
      "learning_rate": 0.0003,
      "loss": 36.7855625,
      "step": 74000
    },
    {
      "epoch": 10.184420589044867,
      "eval_accuracy": 0.7404586215139355,
      "eval_loss": 8.9921875,
      "eval_runtime": 168.2373,
      "eval_samples_per_second": 9827.6,
      "eval_steps_per_second": 19.199,
      "step": 74000
    },
    {
      "epoch": 10.32204789430223,
      "grad_norm": 15.220124244689941,
      "learning_rate": 0.0003,
      "loss": 36.9548515625,
      "step": 75000
    },
    {
      "epoch": 10.459675199559593,
      "grad_norm": 15.933542251586914,
      "learning_rate": 0.0003,
      "loss": 36.85228125,
      "step": 76000
    },
    {
      "epoch": 10.459675199559593,
      "eval_accuracy": 0.7398413045299678,
      "eval_loss": 9.0234375,
      "eval_runtime": 169.5447,
      "eval_samples_per_second": 9751.82,
      "eval_steps_per_second": 19.051,
      "step": 76000
    },
    {
      "epoch": 10.597302504816955,
      "grad_norm": 15.21678638458252,
      "learning_rate": 0.0003,
      "loss": 36.8666171875,
      "step": 77000
    },
    {
      "epoch": 10.734929810074318,
      "grad_norm": 16.906696319580078,
      "learning_rate": 0.0003,
      "loss": 36.78946875,
      "step": 78000
    },
    {
      "epoch": 10.734929810074318,
      "eval_accuracy": 0.7396672771453636,
      "eval_loss": 9.015625,
      "eval_runtime": 168.0605,
      "eval_samples_per_second": 9837.938,
      "eval_steps_per_second": 19.219,
      "step": 78000
    },
    {
      "epoch": 10.872557115331682,
      "grad_norm": 15.474593162536621,
      "learning_rate": 0.0003,
      "loss": 36.7689296875,
      "step": 79000
    },
    {
      "epoch": 11.010184420589045,
      "grad_norm": 14.83968448638916,
      "learning_rate": 0.0003,
      "loss": 36.761296875,
      "step": 80000
    },
    {
      "epoch": 11.010184420589045,
      "eval_accuracy": 0.7401702144442798,
      "eval_loss": 9.0,
      "eval_runtime": 167.1731,
      "eval_samples_per_second": 9890.16,
      "eval_steps_per_second": 19.321,
      "step": 80000
    },
    {
      "epoch": 11.147811725846408,
      "grad_norm": 16.945621490478516,
      "learning_rate": 0.0003,
      "loss": 36.7302734375,
      "step": 81000
    },
    {
      "epoch": 11.28543903110377,
      "grad_norm": 20.112407684326172,
      "learning_rate": 0.0003,
      "loss": 36.74803125,
      "step": 82000
    },
    {
      "epoch": 11.28543903110377,
      "eval_accuracy": 0.7397283390276018,
      "eval_loss": 9.046875,
      "eval_runtime": 167.1017,
      "eval_samples_per_second": 9894.387,
      "eval_steps_per_second": 19.33,
      "step": 82000
    },
    {
      "epoch": 11.423066336361135,
      "grad_norm": 15.264861106872559,
      "learning_rate": 0.0003,
      "loss": 36.7436875,
      "step": 83000
    },
    {
      "epoch": 11.560693641618498,
      "grad_norm": 35.09538269042969,
      "learning_rate": 0.0003,
      "loss": 36.68840625,
      "step": 84000
    },
    {
      "epoch": 11.560693641618498,
      "eval_accuracy": 0.7408698045825953,
      "eval_loss": 8.96875,
      "eval_runtime": 168.7302,
      "eval_samples_per_second": 9798.89,
      "eval_steps_per_second": 19.143,
      "step": 84000
    },
    {
      "epoch": 11.69832094687586,
      "grad_norm": 19.97286605834961,
      "learning_rate": 0.0003,
      "loss": 36.69001171875,
      "step": 85000
    },
    {
      "epoch": 11.835948252133223,
      "grad_norm": 17.521934509277344,
      "learning_rate": 0.0003,
      "loss": 36.67117578125,
      "step": 86000
    },
    {
      "epoch": 11.835948252133223,
      "eval_accuracy": 0.7407812017839025,
      "eval_loss": 8.9765625,
      "eval_runtime": 169.2449,
      "eval_samples_per_second": 9769.092,
      "eval_steps_per_second": 19.085,
      "step": 86000
    },
    {
      "epoch": 11.973575557390586,
      "grad_norm": 15.003190040588379,
      "learning_rate": 0.0003,
      "loss": 36.64848046875,
      "step": 87000
    },
    {
      "epoch": 12.11120286264795,
      "grad_norm": 20.016794204711914,
      "learning_rate": 0.0003,
      "loss": 36.72609765625,
      "step": 88000
    },
    {
      "epoch": 12.11120286264795,
      "eval_accuracy": 0.7402567288458476,
      "eval_loss": 9.0,
      "eval_runtime": 168.1699,
      "eval_samples_per_second": 9831.539,
      "eval_steps_per_second": 19.207,
      "step": 88000
    },
    {
      "epoch": 12.248830167905313,
      "grad_norm": 16.020404815673828,
      "learning_rate": 0.0003,
      "loss": 36.8169140625,
      "step": 89000
    },
    {
      "epoch": 12.386457473162675,
      "grad_norm": 16.005430221557617,
      "learning_rate": 0.0003,
      "loss": 36.652015625,
      "step": 90000
    },
    {
      "epoch": 12.386457473162675,
      "eval_accuracy": 0.740178315482875,
      "eval_loss": 9.0,
      "eval_runtime": 169.3304,
      "eval_samples_per_second": 9764.16,
      "eval_steps_per_second": 19.075,
      "step": 90000
    },
    {
      "epoch": 12.524084778420038,
      "grad_norm": 14.788511276245117,
      "learning_rate": 0.0003,
      "loss": 36.70846875,
      "step": 91000
    },
    {
      "epoch": 12.661712083677402,
      "grad_norm": 15.171088218688965,
      "learning_rate": 0.0003,
      "loss": 36.9015078125,
      "step": 92000
    },
    {
      "epoch": 12.661712083677402,
      "eval_accuracy": 0.7409616192832641,
      "eval_loss": 8.9765625,
      "eval_runtime": 167.4567,
      "eval_samples_per_second": 9873.414,
      "eval_steps_per_second": 19.289,
      "step": 92000
    },
    {
      "epoch": 12.799339388934765,
      "grad_norm": 15.66518497467041,
      "learning_rate": 0.0003,
      "loss": 36.66136328125,
      "step": 93000
    },
    {
      "epoch": 12.936966694192128,
      "grad_norm": 17.391189575195312,
      "learning_rate": 0.0003,
      "loss": 36.62637109375,
      "step": 94000
    },
    {
      "epoch": 12.936966694192128,
      "eval_accuracy": 0.7408841425444469,
      "eval_loss": 8.96875,
      "eval_runtime": 168.9204,
      "eval_samples_per_second": 9787.86,
      "eval_steps_per_second": 19.121,
      "step": 94000
    },
    {
      "epoch": 13.07459399944949,
      "grad_norm": 14.819647789001465,
      "learning_rate": 0.0003,
      "loss": 36.554375,
      "step": 95000
    },
    {
      "epoch": 13.212221304706853,
      "grad_norm": 15.322741508483887,
      "learning_rate": 0.0003,
      "loss": 36.5329375,
      "step": 96000
    },
    {
      "epoch": 13.212221304706853,
      "eval_accuracy": 0.7418048186655257,
      "eval_loss": 8.9296875,
      "eval_runtime": 168.0129,
      "eval_samples_per_second": 9840.727,
      "eval_steps_per_second": 19.225,
      "step": 96000
    },
    {
      "epoch": 13.349848609964218,
      "grad_norm": 15.646768569946289,
      "learning_rate": 0.0003,
      "loss": 36.5148125,
      "step": 97000
    },
    {
      "epoch": 13.48747591522158,
      "grad_norm": 22.885791778564453,
      "learning_rate": 0.0003,
      "loss": 36.550625,
      "step": 98000
    },
    {
      "epoch": 13.48747591522158,
      "eval_accuracy": 0.7403524509758528,
      "eval_loss": 9.0078125,
      "eval_runtime": 169.9032,
      "eval_samples_per_second": 9731.239,
      "eval_steps_per_second": 19.011,
      "step": 98000
    },
    {
      "epoch": 13.625103220478943,
      "grad_norm": 23.617691040039062,
      "learning_rate": 0.0003,
      "loss": 36.64653515625,
      "step": 99000
    },
    {
      "epoch": 13.762730525736306,
      "grad_norm": 16.434221267700195,
      "learning_rate": 0.0003,
      "loss": 36.5250390625,
      "step": 100000
    },
    {
      "epoch": 13.762730525736306,
      "eval_accuracy": 0.741298813475031,
      "eval_loss": 8.9453125,
      "eval_runtime": 170.5631,
      "eval_samples_per_second": 9693.591,
      "eval_steps_per_second": 18.937,
      "step": 100000
    },
    {
      "epoch": 13.90035783099367,
      "grad_norm": 15.552685737609863,
      "learning_rate": 0.0003,
      "loss": 36.50655078125,
      "step": 101000
    },
    {
      "epoch": 14.037985136251033,
      "grad_norm": 15.682257652282715,
      "learning_rate": 0.0003,
      "loss": 36.52517578125,
      "step": 102000
    },
    {
      "epoch": 14.037985136251033,
      "eval_accuracy": 0.7417503019794978,
      "eval_loss": 8.9296875,
      "eval_runtime": 167.2274,
      "eval_samples_per_second": 9886.95,
      "eval_steps_per_second": 19.315,
      "step": 102000
    },
    {
      "epoch": 14.175612441508395,
      "grad_norm": 16.582626342773438,
      "learning_rate": 0.0003,
      "loss": 36.4439453125,
      "step": 103000
    },
    {
      "epoch": 14.313239746765758,
      "grad_norm": 15.228682518005371,
      "learning_rate": 0.0003,
      "loss": 36.7357578125,
      "step": 104000
    },
    {
      "epoch": 14.313239746765758,
      "eval_accuracy": 0.7409231190118309,
      "eval_loss": 8.984375,
      "eval_runtime": 168.495,
      "eval_samples_per_second": 9812.569,
      "eval_steps_per_second": 19.17,
      "step": 104000
    },
    {
      "epoch": 14.45086705202312,
      "grad_norm": 18.01194190979004,
      "learning_rate": 0.0003,
      "loss": 36.506953125,
      "step": 105000
    },
    {
      "epoch": 14.588494357280485,
      "grad_norm": 16.423683166503906,
      "learning_rate": 0.0003,
      "loss": 36.4571796875,
      "step": 106000
    },
    {
      "epoch": 14.588494357280485,
      "eval_accuracy": 0.7415011536669796,
      "eval_loss": 8.9453125,
      "eval_runtime": 170.7059,
      "eval_samples_per_second": 9685.482,
      "eval_steps_per_second": 18.921,
      "step": 106000
    },
    {
      "epoch": 14.726121662537848,
      "grad_norm": 15.935735702514648,
      "learning_rate": 0.0003,
      "loss": 36.4399453125,
      "step": 107000
    },
    {
      "epoch": 14.86374896779521,
      "grad_norm": 34.532562255859375,
      "learning_rate": 0.0003,
      "loss": 36.496796875,
      "step": 108000
    },
    {
      "epoch": 14.86374896779521,
      "eval_accuracy": 0.740452263067793,
      "eval_loss": 9.0,
      "eval_runtime": 169.5593,
      "eval_samples_per_second": 9750.98,
      "eval_steps_per_second": 19.049,
      "step": 108000
    },
    {
      "epoch": 15.001376273052573,
      "grad_norm": 17.5130615234375,
      "learning_rate": 0.0003,
      "loss": 36.44953125,
      "step": 109000
    },
    {
      "epoch": 15.139003578309937,
      "grad_norm": 17.350488662719727,
      "learning_rate": 0.0003,
      "loss": 36.5754375,
      "step": 110000
    },
    {
      "epoch": 15.139003578309937,
      "eval_accuracy": 0.7407140399047041,
      "eval_loss": 8.984375,
      "eval_runtime": 169.3269,
      "eval_samples_per_second": 9764.36,
      "eval_steps_per_second": 19.076,
      "step": 110000
    },
    {
      "epoch": 15.2766308835673,
      "grad_norm": 16.463069915771484,
      "learning_rate": 0.0003,
      "loss": 36.57256640625,
      "step": 111000
    },
    {
      "epoch": 15.414258188824663,
      "grad_norm": 15.29598331451416,
      "learning_rate": 0.0003,
      "loss": 36.57964453125,
      "step": 112000
    },
    {
      "epoch": 15.414258188824663,
      "eval_accuracy": 0.7415344808420624,
      "eval_loss": 8.9375,
      "eval_runtime": 169.2422,
      "eval_samples_per_second": 9769.248,
      "eval_steps_per_second": 19.085,
      "step": 112000
    },
    {
      "epoch": 15.551885494082025,
      "grad_norm": 26.094396591186523,
      "learning_rate": 0.0003,
      "loss": 36.45933203125,
      "step": 113000
    },
    {
      "epoch": 15.689512799339388,
      "grad_norm": 15.42809009552002,
      "learning_rate": 0.0003,
      "loss": 36.4115546875,
      "step": 114000
    },
    {
      "epoch": 15.689512799339388,
      "eval_accuracy": 0.7421097237159107,
      "eval_loss": 8.921875,
      "eval_runtime": 168.9547,
      "eval_samples_per_second": 9785.869,
      "eval_steps_per_second": 19.118,
      "step": 114000
    },
    {
      "epoch": 15.827140104596753,
      "grad_norm": 18.432769775390625,
      "learning_rate": 0.0003,
      "loss": 36.36869921875,
      "step": 115000
    },
    {
      "epoch": 15.964767409854115,
      "grad_norm": 16.479217529296875,
      "learning_rate": 0.0003,
      "loss": 36.33540625,
      "step": 116000
    },
    {
      "epoch": 15.964767409854115,
      "eval_accuracy": 0.742688496326886,
      "eval_loss": 8.9140625,
      "eval_runtime": 171.1574,
      "eval_samples_per_second": 9659.935,
      "eval_steps_per_second": 18.872,
      "step": 116000
    },
    {
      "epoch": 16.10239471511148,
      "grad_norm": 16.778072357177734,
      "learning_rate": 0.0003,
      "loss": 36.3029140625,
      "step": 117000
    },
    {
      "epoch": 16.24002202036884,
      "grad_norm": 17.59860610961914,
      "learning_rate": 0.0003,
      "loss": 36.37992578125,
      "step": 118000
    },
    {
      "epoch": 16.24002202036884,
      "eval_accuracy": 0.7424896095870896,
      "eval_loss": 8.9140625,
      "eval_runtime": 169.4524,
      "eval_samples_per_second": 9757.132,
      "eval_steps_per_second": 19.061,
      "step": 118000
    },
    {
      "epoch": 16.377649325626205,
      "grad_norm": 19.460819244384766,
      "learning_rate": 0.0003,
      "loss": 36.3964765625,
      "step": 119000
    },
    {
      "epoch": 16.515276630883566,
      "grad_norm": 47.38739776611328,
      "learning_rate": 0.0003,
      "loss": 36.399671875,
      "step": 120000
    },
    {
      "epoch": 16.515276630883566,
      "eval_accuracy": 0.7421772826656658,
      "eval_loss": 8.90625,
      "eval_runtime": 168.2033,
      "eval_samples_per_second": 9829.585,
      "eval_steps_per_second": 19.203,
      "step": 120000
    },
    {
      "epoch": 16.515276630883566,
      "step": 120000,
      "total_flos": 1.617130434771026e+19,
      "train_loss": 38.9186564453125,
      "train_runtime": 44793.9304,
      "train_samples_per_second": 5486.458,
      "train_steps_per_second": 2.679
    }
  ],
  "logging_steps": 1000,
  "max_steps": 120000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 17,
  "save_steps": 10000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.617130434771026e+19,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}