{
  "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": 1.5182671546936035,
      "learning_rate": 0.00046641791044776124,
      "loss": 7.4823,
      "step": 500
    },
    {
      "epoch": 0.02,
      "eval_loss": 5.233279228210449,
      "eval_runtime": 485.9279,
      "eval_samples_per_second": 116.13,
      "eval_steps_per_second": 3.63,
      "step": 500
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2752304077148438,
      "learning_rate": 0.0009328358208955225,
      "loss": 4.5296,
      "step": 1000
    },
    {
      "epoch": 0.05,
      "eval_loss": 4.0948486328125,
      "eval_runtime": 455.2584,
      "eval_samples_per_second": 123.954,
      "eval_steps_per_second": 3.875,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1236172914505005,
      "learning_rate": 0.0009789877154220063,
      "loss": 3.6901,
      "step": 1500
    },
    {
      "epoch": 0.07,
      "eval_loss": 3.4836795330047607,
      "eval_runtime": 455.9718,
      "eval_samples_per_second": 123.76,
      "eval_steps_per_second": 3.869,
      "step": 1500
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3323962688446045,
      "learning_rate": 0.0009544406539991162,
      "loss": 3.2463,
      "step": 2000
    },
    {
      "epoch": 0.09,
      "eval_loss": 3.187832832336426,
      "eval_runtime": 452.455,
      "eval_samples_per_second": 124.722,
      "eval_steps_per_second": 3.899,
      "step": 2000
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1064891815185547,
      "learning_rate": 0.0009298935925762261,
      "loss": 3.0277,
      "step": 2500
    },
    {
      "epoch": 0.12,
      "eval_loss": 3.018643856048584,
      "eval_runtime": 455.738,
      "eval_samples_per_second": 123.823,
      "eval_steps_per_second": 3.871,
      "step": 2500
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0591437816619873,
      "learning_rate": 0.0009053465311533362,
      "loss": 2.893,
      "step": 3000
    },
    {
      "epoch": 0.14,
      "eval_loss": 2.910371780395508,
      "eval_runtime": 455.3219,
      "eval_samples_per_second": 123.936,
      "eval_steps_per_second": 3.874,
      "step": 3000
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1225671768188477,
      "learning_rate": 0.0008807994697304462,
      "loss": 2.8067,
      "step": 3500
    },
    {
      "epoch": 0.16,
      "eval_loss": 2.8422658443450928,
      "eval_runtime": 456.888,
      "eval_samples_per_second": 123.512,
      "eval_steps_per_second": 3.861,
      "step": 3500
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9899258613586426,
      "learning_rate": 0.0008562524083075562,
      "loss": 2.7388,
      "step": 4000
    },
    {
      "epoch": 0.19,
      "eval_loss": 2.7853689193725586,
      "eval_runtime": 467.1715,
      "eval_samples_per_second": 120.793,
      "eval_steps_per_second": 3.776,
      "step": 4000
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0482165813446045,
      "learning_rate": 0.0008317053468846664,
      "loss": 2.6884,
      "step": 4500
    },
    {
      "epoch": 0.21,
      "eval_loss": 2.744969129562378,
      "eval_runtime": 464.7403,
      "eval_samples_per_second": 121.425,
      "eval_steps_per_second": 3.796,
      "step": 4500
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9657552242279053,
      "learning_rate": 0.0008071582854617764,
      "loss": 2.652,
      "step": 5000
    },
    {
      "epoch": 0.23,
      "eval_loss": 2.704190969467163,
      "eval_runtime": 465.0407,
      "eval_samples_per_second": 121.346,
      "eval_steps_per_second": 3.793,
      "step": 5000
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0542707443237305,
      "learning_rate": 0.0007826112240388863,
      "loss": 2.6152,
      "step": 5500
    },
    {
      "epoch": 0.26,
      "eval_loss": 2.677891969680786,
      "eval_runtime": 462.2086,
      "eval_samples_per_second": 122.09,
      "eval_steps_per_second": 3.816,
      "step": 5500
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1020146608352661,
      "learning_rate": 0.0007580641626159963,
      "loss": 2.5896,
      "step": 6000
    },
    {
      "epoch": 0.28,
      "eval_loss": 2.6452853679656982,
      "eval_runtime": 464.0405,
      "eval_samples_per_second": 121.608,
      "eval_steps_per_second": 3.801,
      "step": 6000
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0760774612426758,
      "learning_rate": 0.0007335171011931065,
      "loss": 2.5639,
      "step": 6500
    },
    {
      "epoch": 0.3,
      "eval_loss": 2.6233344078063965,
      "eval_runtime": 456.558,
      "eval_samples_per_second": 123.601,
      "eval_steps_per_second": 3.864,
      "step": 6500
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0427649021148682,
      "learning_rate": 0.0007089700397702165,
      "loss": 2.5448,
      "step": 7000
    },
    {
      "epoch": 0.33,
      "eval_loss": 2.602931022644043,
      "eval_runtime": 383.1386,
      "eval_samples_per_second": 147.286,
      "eval_steps_per_second": 4.604,
      "step": 7000
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0298473834991455,
      "learning_rate": 0.0006844229783473265,
      "loss": 2.5212,
      "step": 7500
    },
    {
      "epoch": 0.35,
      "eval_loss": 2.583045244216919,
      "eval_runtime": 385.2194,
      "eval_samples_per_second": 146.491,
      "eval_steps_per_second": 4.579,
      "step": 7500
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0979411602020264,
      "learning_rate": 0.0006598759169244364,
      "loss": 2.5057,
      "step": 8000
    },
    {
      "epoch": 0.37,
      "eval_loss": 2.568408727645874,
      "eval_runtime": 382.7875,
      "eval_samples_per_second": 147.421,
      "eval_steps_per_second": 4.608,
      "step": 8000
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9886611700057983,
      "learning_rate": 0.0006353779496243923,
      "loss": 2.4951,
      "step": 8500
    },
    {
      "epoch": 0.4,
      "eval_loss": 2.550605058670044,
      "eval_runtime": 383.6666,
      "eval_samples_per_second": 147.083,
      "eval_steps_per_second": 4.598,
      "step": 8500
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1519851684570312,
      "learning_rate": 0.0006108799823243482,
      "loss": 2.4801,
      "step": 9000
    },
    {
      "epoch": 0.42,
      "eval_loss": 2.5366618633270264,
      "eval_runtime": 383.6303,
      "eval_samples_per_second": 147.097,
      "eval_steps_per_second": 4.598,
      "step": 9000
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.13021719455719,
      "learning_rate": 0.0005863329209014581,
      "loss": 2.4656,
      "step": 9500
    },
    {
      "epoch": 0.44,
      "eval_loss": 2.5247080326080322,
      "eval_runtime": 383.3382,
      "eval_samples_per_second": 147.209,
      "eval_steps_per_second": 4.602,
      "step": 9500
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0263173580169678,
      "learning_rate": 0.0005617858594785683,
      "loss": 2.4445,
      "step": 10000
    },
    {
      "epoch": 0.47,
      "eval_loss": 2.5110859870910645,
      "eval_runtime": 383.0191,
      "eval_samples_per_second": 147.332,
      "eval_steps_per_second": 4.606,
      "step": 10000
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1149547100067139,
      "learning_rate": 0.0005372387980556782,
      "loss": 2.4358,
      "step": 10500
    },
    {
      "epoch": 0.49,
      "eval_loss": 2.501206398010254,
      "eval_runtime": 382.9886,
      "eval_samples_per_second": 147.344,
      "eval_steps_per_second": 4.606,
      "step": 10500
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2581932544708252,
      "learning_rate": 0.0005126917366327883,
      "loss": 2.4232,
      "step": 11000
    },
    {
      "epoch": 0.51,
      "eval_loss": 2.489328145980835,
      "eval_runtime": 383.1241,
      "eval_samples_per_second": 147.292,
      "eval_steps_per_second": 4.604,
      "step": 11000
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1889700889587402,
      "learning_rate": 0.0004881446752098984,
      "loss": 2.4174,
      "step": 11500
    },
    {
      "epoch": 0.54,
      "eval_loss": 2.4806177616119385,
      "eval_runtime": 383.1272,
      "eval_samples_per_second": 147.291,
      "eval_steps_per_second": 4.604,
      "step": 11500
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1834616661071777,
      "learning_rate": 0.0004635976137870084,
      "loss": 2.4025,
      "step": 12000
    },
    {
      "epoch": 0.56,
      "eval_loss": 2.468430757522583,
      "eval_runtime": 383.1022,
      "eval_samples_per_second": 147.3,
      "eval_steps_per_second": 4.605,
      "step": 12000
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.162919282913208,
      "learning_rate": 0.0004390505523641184,
      "loss": 2.3933,
      "step": 12500
    },
    {
      "epoch": 0.58,
      "eval_loss": 2.454052686691284,
      "eval_runtime": 382.9498,
      "eval_samples_per_second": 147.359,
      "eval_steps_per_second": 4.606,
      "step": 12500
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1778837442398071,
      "learning_rate": 0.0004145525850640742,
      "loss": 2.3762,
      "step": 13000
    },
    {
      "epoch": 0.61,
      "eval_loss": 2.447908639907837,
      "eval_runtime": 382.972,
      "eval_samples_per_second": 147.35,
      "eval_steps_per_second": 4.606,
      "step": 13000
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1710721254348755,
      "learning_rate": 0.00039000552364118425,
      "loss": 2.3776,
      "step": 13500
    },
    {
      "epoch": 0.63,
      "eval_loss": 2.434429883956909,
      "eval_runtime": 383.0888,
      "eval_samples_per_second": 147.305,
      "eval_steps_per_second": 4.605,
      "step": 13500
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.201686143875122,
      "learning_rate": 0.00036545846221829433,
      "loss": 2.3547,
      "step": 14000
    },
    {
      "epoch": 0.65,
      "eval_loss": 2.4281551837921143,
      "eval_runtime": 382.9249,
      "eval_samples_per_second": 147.368,
      "eval_steps_per_second": 4.607,
      "step": 14000
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2691400051116943,
      "learning_rate": 0.00034091140079540435,
      "loss": 2.3442,
      "step": 14500
    },
    {
      "epoch": 0.68,
      "eval_loss": 2.420260429382324,
      "eval_runtime": 382.7963,
      "eval_samples_per_second": 147.418,
      "eval_steps_per_second": 4.608,
      "step": 14500
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1737455129623413,
      "learning_rate": 0.0003164134334953601,
      "loss": 2.3457,
      "step": 15000
    },
    {
      "epoch": 0.7,
      "eval_loss": 2.4077823162078857,
      "eval_runtime": 382.9004,
      "eval_samples_per_second": 147.378,
      "eval_steps_per_second": 4.607,
      "step": 15000
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2281179428100586,
      "learning_rate": 0.00029186637207247017,
      "loss": 2.3349,
      "step": 15500
    },
    {
      "epoch": 0.72,
      "eval_loss": 2.402637243270874,
      "eval_runtime": 382.7179,
      "eval_samples_per_second": 147.448,
      "eval_steps_per_second": 4.609,
      "step": 15500
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2782984972000122,
      "learning_rate": 0.0002673193106495802,
      "loss": 2.3277,
      "step": 16000
    },
    {
      "epoch": 0.75,
      "eval_loss": 2.3890154361724854,
      "eval_runtime": 382.9227,
      "eval_samples_per_second": 147.369,
      "eval_steps_per_second": 4.607,
      "step": 16000
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2969355583190918,
      "learning_rate": 0.00024277224922669027,
      "loss": 2.316,
      "step": 16500
    },
    {
      "epoch": 0.77,
      "eval_loss": 2.3841326236724854,
      "eval_runtime": 382.942,
      "eval_samples_per_second": 147.362,
      "eval_steps_per_second": 4.606,
      "step": 16500
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.244382381439209,
      "learning_rate": 0.000218274281926646,
      "loss": 2.306,
      "step": 17000
    },
    {
      "epoch": 0.79,
      "eval_loss": 2.374117851257324,
      "eval_runtime": 382.7874,
      "eval_samples_per_second": 147.421,
      "eval_steps_per_second": 4.608,
      "step": 17000
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1741069555282593,
      "learning_rate": 0.00019372722050375605,
      "loss": 2.3011,
      "step": 17500
    },
    {
      "epoch": 0.82,
      "eval_loss": 2.365194320678711,
      "eval_runtime": 382.8368,
      "eval_samples_per_second": 147.402,
      "eval_steps_per_second": 4.608,
      "step": 17500
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2608463764190674,
      "learning_rate": 0.0001691801590808661,
      "loss": 2.2839,
      "step": 18000
    },
    {
      "epoch": 0.84,
      "eval_loss": 2.3552064895629883,
      "eval_runtime": 382.676,
      "eval_samples_per_second": 147.464,
      "eval_steps_per_second": 4.61,
      "step": 18000
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2342902421951294,
      "learning_rate": 0.00014463309765797616,
      "loss": 2.2846,
      "step": 18500
    },
    {
      "epoch": 0.86,
      "eval_loss": 2.346794843673706,
      "eval_runtime": 382.6366,
      "eval_samples_per_second": 147.479,
      "eval_steps_per_second": 4.61,
      "step": 18500
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2380675077438354,
      "learning_rate": 0.00012008603623508621,
      "loss": 2.2767,
      "step": 19000
    },
    {
      "epoch": 0.89,
      "eval_loss": 2.3396549224853516,
      "eval_runtime": 382.7214,
      "eval_samples_per_second": 147.447,
      "eval_steps_per_second": 4.609,
      "step": 19000
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2454231977462769,
      "learning_rate": 9.553897481219615e-05,
      "loss": 2.2672,
      "step": 19500
    },
    {
      "epoch": 0.91,
      "eval_loss": 2.333063840866089,
      "eval_runtime": 382.7079,
      "eval_samples_per_second": 147.452,
      "eval_steps_per_second": 4.609,
      "step": 19500
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.155750036239624,
      "learning_rate": 7.09919133893062e-05,
      "loss": 2.2536,
      "step": 20000
    },
    {
      "epoch": 0.93,
      "eval_loss": 2.3241796493530273,
      "eval_runtime": 383.212,
      "eval_samples_per_second": 147.258,
      "eval_steps_per_second": 4.603,
      "step": 20000
    }
  ],
  "logging_steps": 500,
  "max_steps": 21439,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 10000,
  "total_flos": 2336536657920000.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}