{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9328575759695888,
  "eval_steps": 500,
  "global_step": 20000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "grad_norm": 0.3762490749359131,
      "learning_rate": 0.00046641791044776124,
      "loss": 8.4353,
      "step": 500
    },
    {
      "epoch": 0.02,
      "eval_loss": 6.534626483917236,
      "eval_runtime": 213.0459,
      "eval_samples_per_second": 264.877,
      "eval_steps_per_second": 8.28,
      "step": 500
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.518302321434021,
      "learning_rate": 0.0009328358208955225,
      "loss": 5.5553,
      "step": 1000
    },
    {
      "epoch": 0.05,
      "eval_loss": 4.976442813873291,
      "eval_runtime": 215.6331,
      "eval_samples_per_second": 261.699,
      "eval_steps_per_second": 8.181,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8457584381103516,
      "learning_rate": 0.0009789877154220063,
      "loss": 4.5996,
      "step": 1500
    },
    {
      "epoch": 0.07,
      "eval_loss": 4.408080101013184,
      "eval_runtime": 215.4944,
      "eval_samples_per_second": 261.868,
      "eval_steps_per_second": 8.186,
      "step": 1500
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3413056135177612,
      "learning_rate": 0.0009544406539991162,
      "loss": 4.1961,
      "step": 2000
    },
    {
      "epoch": 0.09,
      "eval_loss": 4.121068954467773,
      "eval_runtime": 216.0979,
      "eval_samples_per_second": 261.136,
      "eval_steps_per_second": 8.163,
      "step": 2000
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2387056350708008,
      "learning_rate": 0.0009298935925762261,
      "loss": 3.9584,
      "step": 2500
    },
    {
      "epoch": 0.12,
      "eval_loss": 3.929608106613159,
      "eval_runtime": 216.438,
      "eval_samples_per_second": 260.726,
      "eval_steps_per_second": 8.15,
      "step": 2500
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3727014064788818,
      "learning_rate": 0.0009053465311533362,
      "loss": 3.7833,
      "step": 3000
    },
    {
      "epoch": 0.14,
      "eval_loss": 3.787693738937378,
      "eval_runtime": 218.2815,
      "eval_samples_per_second": 258.524,
      "eval_steps_per_second": 8.081,
      "step": 3000
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5133860111236572,
      "learning_rate": 0.0008807994697304462,
      "loss": 3.6616,
      "step": 3500
    },
    {
      "epoch": 0.16,
      "eval_loss": 3.683837652206421,
      "eval_runtime": 217.4429,
      "eval_samples_per_second": 259.521,
      "eval_steps_per_second": 8.112,
      "step": 3500
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3134872913360596,
      "learning_rate": 0.0008562524083075562,
      "loss": 3.5679,
      "step": 4000
    },
    {
      "epoch": 0.19,
      "eval_loss": 3.6060750484466553,
      "eval_runtime": 216.1087,
      "eval_samples_per_second": 261.123,
      "eval_steps_per_second": 8.163,
      "step": 4000
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3464657068252563,
      "learning_rate": 0.0008317053468846664,
      "loss": 3.4964,
      "step": 4500
    },
    {
      "epoch": 0.21,
      "eval_loss": 3.54757022857666,
      "eval_runtime": 216.7125,
      "eval_samples_per_second": 260.396,
      "eval_steps_per_second": 8.14,
      "step": 4500
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3200907707214355,
      "learning_rate": 0.0008071582854617764,
      "loss": 3.4488,
      "step": 5000
    },
    {
      "epoch": 0.23,
      "eval_loss": 3.4986045360565186,
      "eval_runtime": 217.9581,
      "eval_samples_per_second": 258.908,
      "eval_steps_per_second": 8.093,
      "step": 5000
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.514103651046753,
      "learning_rate": 0.0007826112240388863,
      "loss": 3.3973,
      "step": 5500
    },
    {
      "epoch": 0.26,
      "eval_loss": 3.4664642810821533,
      "eval_runtime": 218.2073,
      "eval_samples_per_second": 258.612,
      "eval_steps_per_second": 8.084,
      "step": 5500
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4389874935150146,
      "learning_rate": 0.0007580641626159963,
      "loss": 3.3587,
      "step": 6000
    },
    {
      "epoch": 0.28,
      "eval_loss": 3.4231605529785156,
      "eval_runtime": 218.4087,
      "eval_samples_per_second": 258.373,
      "eval_steps_per_second": 8.077,
      "step": 6000
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4937726259231567,
      "learning_rate": 0.0007335171011931065,
      "loss": 3.3231,
      "step": 6500
    },
    {
      "epoch": 0.3,
      "eval_loss": 3.390798330307007,
      "eval_runtime": 216.9327,
      "eval_samples_per_second": 260.131,
      "eval_steps_per_second": 8.132,
      "step": 6500
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5942374467849731,
      "learning_rate": 0.0007089700397702165,
      "loss": 3.2971,
      "step": 7000
    },
    {
      "epoch": 0.33,
      "eval_loss": 3.361341714859009,
      "eval_runtime": 217.461,
      "eval_samples_per_second": 259.499,
      "eval_steps_per_second": 8.112,
      "step": 7000
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6310980319976807,
      "learning_rate": 0.0006844229783473265,
      "loss": 3.2679,
      "step": 7500
    },
    {
      "epoch": 0.35,
      "eval_loss": 3.33766770362854,
      "eval_runtime": 217.3952,
      "eval_samples_per_second": 259.578,
      "eval_steps_per_second": 8.114,
      "step": 7500
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6001648902893066,
      "learning_rate": 0.0006598759169244364,
      "loss": 3.2436,
      "step": 8000
    },
    {
      "epoch": 0.37,
      "eval_loss": 3.3167307376861572,
      "eval_runtime": 217.5617,
      "eval_samples_per_second": 259.379,
      "eval_steps_per_second": 8.108,
      "step": 8000
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.590154767036438,
      "learning_rate": 0.0006353288555015467,
      "loss": 3.2322,
      "step": 8500
    },
    {
      "epoch": 0.4,
      "eval_loss": 3.2952842712402344,
      "eval_runtime": 217.0219,
      "eval_samples_per_second": 260.024,
      "eval_steps_per_second": 8.128,
      "step": 8500
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7063907384872437,
      "learning_rate": 0.0006107817940786566,
      "loss": 3.208,
      "step": 9000
    },
    {
      "epoch": 0.42,
      "eval_loss": 3.2771995067596436,
      "eval_runtime": 218.9112,
      "eval_samples_per_second": 257.78,
      "eval_steps_per_second": 8.058,
      "step": 9000
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.544815182685852,
      "learning_rate": 0.0005862347326557666,
      "loss": 3.1923,
      "step": 9500
    },
    {
      "epoch": 0.44,
      "eval_loss": 3.263777017593384,
      "eval_runtime": 218.0114,
      "eval_samples_per_second": 258.844,
      "eval_steps_per_second": 8.091,
      "step": 9500
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.615902304649353,
      "learning_rate": 0.0005616876712328766,
      "loss": 3.1683,
      "step": 10000
    },
    {
      "epoch": 0.47,
      "eval_loss": 3.2488255500793457,
      "eval_runtime": 217.5444,
      "eval_samples_per_second": 259.4,
      "eval_steps_per_second": 8.109,
      "step": 10000
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.533378005027771,
      "learning_rate": 0.0005371897039328324,
      "loss": 3.158,
      "step": 10500
    },
    {
      "epoch": 0.49,
      "eval_loss": 3.2329940795898438,
      "eval_runtime": 214.5885,
      "eval_samples_per_second": 262.973,
      "eval_steps_per_second": 8.22,
      "step": 10500
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9465535879135132,
      "learning_rate": 0.0005126426425099425,
      "loss": 3.1439,
      "step": 11000
    },
    {
      "epoch": 0.51,
      "eval_loss": 3.2245969772338867,
      "eval_runtime": 216.8559,
      "eval_samples_per_second": 260.224,
      "eval_steps_per_second": 8.134,
      "step": 11000
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7405271530151367,
      "learning_rate": 0.0004880955810870526,
      "loss": 3.1374,
      "step": 11500
    },
    {
      "epoch": 0.54,
      "eval_loss": 3.208115339279175,
      "eval_runtime": 217.1863,
      "eval_samples_per_second": 259.828,
      "eval_steps_per_second": 8.122,
      "step": 11500
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6891497373580933,
      "learning_rate": 0.0004635485196641626,
      "loss": 3.1213,
      "step": 12000
    },
    {
      "epoch": 0.56,
      "eval_loss": 3.200620651245117,
      "eval_runtime": 218.4169,
      "eval_samples_per_second": 258.364,
      "eval_steps_per_second": 8.076,
      "step": 12000
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.73305082321167,
      "learning_rate": 0.00043909964648696415,
      "loss": 3.1144,
      "step": 12500
    },
    {
      "epoch": 0.58,
      "eval_loss": 3.188239812850952,
      "eval_runtime": 215.556,
      "eval_samples_per_second": 261.793,
      "eval_steps_per_second": 8.183,
      "step": 12500
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7891405820846558,
      "learning_rate": 0.0004145525850640742,
      "loss": 3.0963,
      "step": 13000
    },
    {
      "epoch": 0.61,
      "eval_loss": 3.174497365951538,
      "eval_runtime": 214.2396,
      "eval_samples_per_second": 263.401,
      "eval_steps_per_second": 8.234,
      "step": 13000
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6677337884902954,
      "learning_rate": 0.00039000552364118425,
      "loss": 3.097,
      "step": 13500
    },
    {
      "epoch": 0.63,
      "eval_loss": 3.165832042694092,
      "eval_runtime": 214.1769,
      "eval_samples_per_second": 263.479,
      "eval_steps_per_second": 8.236,
      "step": 13500
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7241294384002686,
      "learning_rate": 0.00036545846221829433,
      "loss": 3.076,
      "step": 14000
    },
    {
      "epoch": 0.65,
      "eval_loss": 3.1584606170654297,
      "eval_runtime": 264.545,
      "eval_samples_per_second": 213.313,
      "eval_steps_per_second": 6.668,
      "step": 14000
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.835271954536438,
      "learning_rate": 0.00034091140079540435,
      "loss": 3.0666,
      "step": 14500
    },
    {
      "epoch": 0.68,
      "eval_loss": 3.151036262512207,
      "eval_runtime": 218.4608,
      "eval_samples_per_second": 258.312,
      "eval_steps_per_second": 8.075,
      "step": 14500
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7558091878890991,
      "learning_rate": 0.00031636433937251427,
      "loss": 3.0709,
      "step": 15000
    },
    {
      "epoch": 0.7,
      "eval_loss": 3.139883279800415,
      "eval_runtime": 11111.1382,
      "eval_samples_per_second": 5.079,
      "eval_steps_per_second": 0.159,
      "step": 15000
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6270012855529785,
      "learning_rate": 0.00029186637207247017,
      "loss": 3.0609,
      "step": 15500
    },
    {
      "epoch": 0.72,
      "eval_loss": 3.1347908973693848,
      "eval_runtime": 258.6315,
      "eval_samples_per_second": 218.191,
      "eval_steps_per_second": 6.821,
      "step": 15500
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.971229910850525,
      "learning_rate": 0.0002673193106495802,
      "loss": 3.0527,
      "step": 16000
    },
    {
      "epoch": 0.75,
      "eval_loss": 3.127486228942871,
      "eval_runtime": 243.3689,
      "eval_samples_per_second": 231.874,
      "eval_steps_per_second": 7.248,
      "step": 16000
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7010846138000488,
      "learning_rate": 0.00024277224922669027,
      "loss": 3.0449,
      "step": 16500
    },
    {
      "epoch": 0.77,
      "eval_loss": 3.122706890106201,
      "eval_runtime": 259.8928,
      "eval_samples_per_second": 217.132,
      "eval_steps_per_second": 6.787,
      "step": 16500
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0105388164520264,
      "learning_rate": 0.00021822518780380032,
      "loss": 3.0407,
      "step": 17000
    },
    {
      "epoch": 0.79,
      "eval_loss": 3.1165504455566406,
      "eval_runtime": 264.3004,
      "eval_samples_per_second": 213.511,
      "eval_steps_per_second": 6.674,
      "step": 17000
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7834789752960205,
      "learning_rate": 0.00019372722050375605,
      "loss": 3.0376,
      "step": 17500
    },
    {
      "epoch": 0.82,
      "eval_loss": 3.10659122467041,
      "eval_runtime": 265.7435,
      "eval_samples_per_second": 212.351,
      "eval_steps_per_second": 6.638,
      "step": 17500
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8829165697097778,
      "learning_rate": 0.0001691801590808661,
      "loss": 3.0251,
      "step": 18000
    },
    {
      "epoch": 0.84,
      "eval_loss": 3.1006741523742676,
      "eval_runtime": 267.8646,
      "eval_samples_per_second": 210.67,
      "eval_steps_per_second": 6.585,
      "step": 18000
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7196826934814453,
      "learning_rate": 0.00014463309765797616,
      "loss": 3.0226,
      "step": 18500
    },
    {
      "epoch": 0.86,
      "eval_loss": 3.0954883098602295,
      "eval_runtime": 263.6733,
      "eval_samples_per_second": 214.019,
      "eval_steps_per_second": 6.69,
      "step": 18500
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7544931173324585,
      "learning_rate": 0.00012008603623508621,
      "loss": 3.0203,
      "step": 19000
    },
    {
      "epoch": 0.89,
      "eval_loss": 3.0885541439056396,
      "eval_runtime": 225.0692,
      "eval_samples_per_second": 250.727,
      "eval_steps_per_second": 7.838,
      "step": 19000
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.659536361694336,
      "learning_rate": 9.558806893504194e-05,
      "loss": 3.0151,
      "step": 19500
    },
    {
      "epoch": 0.91,
      "eval_loss": 3.083906888961792,
      "eval_runtime": 257.4684,
      "eval_samples_per_second": 219.176,
      "eval_steps_per_second": 6.851,
      "step": 19500
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6146643161773682,
      "learning_rate": 7.1041007512152e-05,
      "loss": 3.0018,
      "step": 20000
    },
    {
      "epoch": 0.93,
      "eval_loss": 3.0793216228485107,
      "eval_runtime": 231.3706,
      "eval_samples_per_second": 243.899,
      "eval_steps_per_second": 7.624,
      "step": 20000
    }
  ],
  "logging_steps": 500,
  "max_steps": 21439,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 10000,
  "total_flos": 196340613120000.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}