{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9795004547680682,
  "eval_steps": 500,
  "global_step": 21000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "grad_norm": 1.301074743270874,
      "learning_rate": 0.00046641791044776124,
      "loss": 7.611,
      "step": 500
    },
    {
      "epoch": 0.02,
      "eval_loss": 5.238980293273926,
      "eval_runtime": 567.2757,
      "eval_samples_per_second": 99.477,
      "eval_steps_per_second": 3.11,
      "step": 500
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1419352293014526,
      "learning_rate": 0.0009328358208955225,
      "loss": 4.4792,
      "step": 1000
    },
    {
      "epoch": 0.05,
      "eval_loss": 4.0363240242004395,
      "eval_runtime": 568.283,
      "eval_samples_per_second": 99.301,
      "eval_steps_per_second": 3.104,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1277365684509277,
      "learning_rate": 0.0009789877154220063,
      "loss": 3.6503,
      "step": 1500
    },
    {
      "epoch": 0.07,
      "eval_loss": 3.468844175338745,
      "eval_runtime": 568.7397,
      "eval_samples_per_second": 99.221,
      "eval_steps_per_second": 3.102,
      "step": 1500
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1394764184951782,
      "learning_rate": 0.0009544406539991162,
      "loss": 3.2321,
      "step": 2000
    },
    {
      "epoch": 0.09,
      "eval_loss": 3.1853878498077393,
      "eval_runtime": 568.7397,
      "eval_samples_per_second": 99.221,
      "eval_steps_per_second": 3.102,
      "step": 2000
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0943962335586548,
      "learning_rate": 0.0009298935925762261,
      "loss": 3.0165,
      "step": 2500
    },
    {
      "epoch": 0.12,
      "eval_loss": 3.0076937675476074,
      "eval_runtime": 569.7751,
      "eval_samples_per_second": 99.041,
      "eval_steps_per_second": 3.096,
      "step": 2500
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9250391721725464,
      "learning_rate": 0.0009053465311533362,
      "loss": 2.8757,
      "step": 3000
    },
    {
      "epoch": 0.14,
      "eval_loss": 2.8989241123199463,
      "eval_runtime": 569.8016,
      "eval_samples_per_second": 99.036,
      "eval_steps_per_second": 3.096,
      "step": 3000
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9219993948936462,
      "learning_rate": 0.0008807994697304462,
      "loss": 2.7831,
      "step": 3500
    },
    {
      "epoch": 0.16,
      "eval_loss": 2.815277099609375,
      "eval_runtime": 569.7431,
      "eval_samples_per_second": 99.046,
      "eval_steps_per_second": 3.096,
      "step": 3500
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0010333061218262,
      "learning_rate": 0.0008562524083075562,
      "loss": 2.7098,
      "step": 4000
    },
    {
      "epoch": 0.19,
      "eval_loss": 2.7535383701324463,
      "eval_runtime": 564.7218,
      "eval_samples_per_second": 99.927,
      "eval_steps_per_second": 3.124,
      "step": 4000
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9716792702674866,
      "learning_rate": 0.0008317053468846664,
      "loss": 2.6551,
      "step": 4500
    },
    {
      "epoch": 0.21,
      "eval_loss": 2.7090485095977783,
      "eval_runtime": 566.8661,
      "eval_samples_per_second": 99.549,
      "eval_steps_per_second": 3.112,
      "step": 4500
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9213451743125916,
      "learning_rate": 0.0008071582854617764,
      "loss": 2.6152,
      "step": 5000
    },
    {
      "epoch": 0.23,
      "eval_loss": 2.666491746902466,
      "eval_runtime": 565.311,
      "eval_samples_per_second": 99.823,
      "eval_steps_per_second": 3.12,
      "step": 5000
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8985408544540405,
      "learning_rate": 0.0007826112240388863,
      "loss": 2.575,
      "step": 5500
    },
    {
      "epoch": 0.26,
      "eval_loss": 2.6350765228271484,
      "eval_runtime": 564.212,
      "eval_samples_per_second": 100.017,
      "eval_steps_per_second": 3.126,
      "step": 5500
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8632990717887878,
      "learning_rate": 0.0007580641626159963,
      "loss": 2.5464,
      "step": 6000
    },
    {
      "epoch": 0.28,
      "eval_loss": 2.6097421646118164,
      "eval_runtime": 568.485,
      "eval_samples_per_second": 99.266,
      "eval_steps_per_second": 3.103,
      "step": 6000
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8819186687469482,
      "learning_rate": 0.0007335171011931065,
      "loss": 2.5183,
      "step": 6500
    },
    {
      "epoch": 0.3,
      "eval_loss": 2.580737590789795,
      "eval_runtime": 568.2279,
      "eval_samples_per_second": 99.311,
      "eval_steps_per_second": 3.104,
      "step": 6500
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9012317657470703,
      "learning_rate": 0.0007089700397702165,
      "loss": 2.4977,
      "step": 7000
    },
    {
      "epoch": 0.33,
      "eval_loss": 2.5541210174560547,
      "eval_runtime": 568.3683,
      "eval_samples_per_second": 99.286,
      "eval_steps_per_second": 3.104,
      "step": 7000
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9294862151145935,
      "learning_rate": 0.0006844229783473265,
      "loss": 2.4723,
      "step": 7500
    },
    {
      "epoch": 0.35,
      "eval_loss": 2.535729169845581,
      "eval_runtime": 569.135,
      "eval_samples_per_second": 99.152,
      "eval_steps_per_second": 3.099,
      "step": 7500
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9035699367523193,
      "learning_rate": 0.0006598759169244364,
      "loss": 2.4549,
      "step": 8000
    },
    {
      "epoch": 0.37,
      "eval_loss": 2.5194778442382812,
      "eval_runtime": 569.6481,
      "eval_samples_per_second": 99.063,
      "eval_steps_per_second": 3.097,
      "step": 8000
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9459491968154907,
      "learning_rate": 0.0006353779496243923,
      "loss": 2.4428,
      "step": 8500
    },
    {
      "epoch": 0.4,
      "eval_loss": 2.502213716506958,
      "eval_runtime": 565.3877,
      "eval_samples_per_second": 99.809,
      "eval_steps_per_second": 3.12,
      "step": 8500
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8880784511566162,
      "learning_rate": 0.0006108308882015024,
      "loss": 2.4255,
      "step": 9000
    },
    {
      "epoch": 0.42,
      "eval_loss": 2.4834067821502686,
      "eval_runtime": 567.6005,
      "eval_samples_per_second": 99.42,
      "eval_steps_per_second": 3.108,
      "step": 9000
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9531754851341248,
      "learning_rate": 0.0005862838267786123,
      "loss": 2.4092,
      "step": 9500
    },
    {
      "epoch": 0.44,
      "eval_loss": 2.473543643951416,
      "eval_runtime": 561.7685,
      "eval_samples_per_second": 100.452,
      "eval_steps_per_second": 3.14,
      "step": 9500
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9443451762199402,
      "learning_rate": 0.0005617367653557224,
      "loss": 2.3859,
      "step": 10000
    },
    {
      "epoch": 0.47,
      "eval_loss": 2.456810712814331,
      "eval_runtime": 563.5685,
      "eval_samples_per_second": 100.132,
      "eval_steps_per_second": 3.13,
      "step": 10000
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9646140933036804,
      "learning_rate": 0.0005372387980556782,
      "loss": 2.3753,
      "step": 10500
    },
    {
      "epoch": 0.49,
      "eval_loss": 2.44191837310791,
      "eval_runtime": 563.9978,
      "eval_samples_per_second": 100.055,
      "eval_steps_per_second": 3.128,
      "step": 10500
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9498624801635742,
      "learning_rate": 0.0005126917366327883,
      "loss": 2.3625,
      "step": 11000
    },
    {
      "epoch": 0.51,
      "eval_loss": 2.428806781768799,
      "eval_runtime": 563.9618,
      "eval_samples_per_second": 100.062,
      "eval_steps_per_second": 3.128,
      "step": 11000
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9920567870140076,
      "learning_rate": 0.0004881446752098984,
      "loss": 2.3547,
      "step": 11500
    },
    {
      "epoch": 0.54,
      "eval_loss": 2.4165408611297607,
      "eval_runtime": 563.8141,
      "eval_samples_per_second": 100.088,
      "eval_steps_per_second": 3.129,
      "step": 11500
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9328219294548035,
      "learning_rate": 0.0004635976137870084,
      "loss": 2.3387,
      "step": 12000
    },
    {
      "epoch": 0.56,
      "eval_loss": 2.40574312210083,
      "eval_runtime": 563.9451,
      "eval_samples_per_second": 100.065,
      "eval_steps_per_second": 3.128,
      "step": 12000
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9478664994239807,
      "learning_rate": 0.00043909964648696415,
      "loss": 2.3277,
      "step": 12500
    },
    {
      "epoch": 0.58,
      "eval_loss": 2.390106201171875,
      "eval_runtime": 563.4302,
      "eval_samples_per_second": 100.156,
      "eval_steps_per_second": 3.131,
      "step": 12500
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9822357296943665,
      "learning_rate": 0.0004145525850640742,
      "loss": 2.3071,
      "step": 13000
    },
    {
      "epoch": 0.61,
      "eval_loss": 2.380479335784912,
      "eval_runtime": 563.2217,
      "eval_samples_per_second": 100.193,
      "eval_steps_per_second": 3.132,
      "step": 13000
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9824458956718445,
      "learning_rate": 0.00039000552364118425,
      "loss": 2.3087,
      "step": 13500
    },
    {
      "epoch": 0.63,
      "eval_loss": 2.366305112838745,
      "eval_runtime": 563.1953,
      "eval_samples_per_second": 100.198,
      "eval_steps_per_second": 3.132,
      "step": 13500
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0211139917373657,
      "learning_rate": 0.00036545846221829433,
      "loss": 2.283,
      "step": 14000
    },
    {
      "epoch": 0.65,
      "eval_loss": 2.3575291633605957,
      "eval_runtime": 563.2425,
      "eval_samples_per_second": 100.19,
      "eval_steps_per_second": 3.132,
      "step": 14000
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0012933015823364,
      "learning_rate": 0.00034091140079540435,
      "loss": 2.2716,
      "step": 14500
    },
    {
      "epoch": 0.68,
      "eval_loss": 2.3457164764404297,
      "eval_runtime": 562.9008,
      "eval_samples_per_second": 100.25,
      "eval_steps_per_second": 3.134,
      "step": 14500
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0424264669418335,
      "learning_rate": 0.0003164134334953601,
      "loss": 2.2709,
      "step": 15000
    },
    {
      "epoch": 0.7,
      "eval_loss": 2.3343236446380615,
      "eval_runtime": 562.6281,
      "eval_samples_per_second": 100.299,
      "eval_steps_per_second": 3.135,
      "step": 15000
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1129311323165894,
      "learning_rate": 0.00029186637207247017,
      "loss": 2.2584,
      "step": 15500
    },
    {
      "epoch": 0.72,
      "eval_loss": 2.3277084827423096,
      "eval_runtime": 574.0103,
      "eval_samples_per_second": 98.31,
      "eval_steps_per_second": 3.073,
      "step": 15500
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0602174997329712,
      "learning_rate": 0.0002673193106495802,
      "loss": 2.2509,
      "step": 16000
    },
    {
      "epoch": 0.75,
      "eval_loss": 2.3144421577453613,
      "eval_runtime": 566.0389,
      "eval_samples_per_second": 99.695,
      "eval_steps_per_second": 3.116,
      "step": 16000
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1678231954574585,
      "learning_rate": 0.00024277224922669027,
      "loss": 2.237,
      "step": 16500
    },
    {
      "epoch": 0.77,
      "eval_loss": 2.305784225463867,
      "eval_runtime": 563.0281,
      "eval_samples_per_second": 100.228,
      "eval_steps_per_second": 3.133,
      "step": 16500
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1629687547683716,
      "learning_rate": 0.000218274281926646,
      "loss": 2.2249,
      "step": 17000
    },
    {
      "epoch": 0.79,
      "eval_loss": 2.2936198711395264,
      "eval_runtime": 562.9504,
      "eval_samples_per_second": 100.242,
      "eval_steps_per_second": 3.133,
      "step": 17000
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0790985822677612,
      "learning_rate": 0.00019372722050375605,
      "loss": 2.219,
      "step": 17500
    },
    {
      "epoch": 0.82,
      "eval_loss": 2.2830076217651367,
      "eval_runtime": 562.6154,
      "eval_samples_per_second": 100.301,
      "eval_steps_per_second": 3.135,
      "step": 17500
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.157651424407959,
      "learning_rate": 0.0001691801590808661,
      "loss": 2.1999,
      "step": 18000
    },
    {
      "epoch": 0.84,
      "eval_loss": 2.2722585201263428,
      "eval_runtime": 562.9469,
      "eval_samples_per_second": 100.242,
      "eval_steps_per_second": 3.134,
      "step": 18000
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0704936981201172,
      "learning_rate": 0.00014463309765797616,
      "loss": 2.1999,
      "step": 18500
    },
    {
      "epoch": 0.86,
      "eval_loss": 2.2624754905700684,
      "eval_runtime": 562.9358,
      "eval_samples_per_second": 100.244,
      "eval_steps_per_second": 3.134,
      "step": 18500
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1210389137268066,
      "learning_rate": 0.00012008603623508621,
      "loss": 2.1898,
      "step": 19000
    },
    {
      "epoch": 0.89,
      "eval_loss": 2.254652976989746,
      "eval_runtime": 563.3356,
      "eval_samples_per_second": 100.173,
      "eval_steps_per_second": 3.131,
      "step": 19000
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0510447025299072,
      "learning_rate": 9.558806893504194e-05,
      "loss": 2.1796,
      "step": 19500
    },
    {
      "epoch": 0.91,
      "eval_loss": 2.2447702884674072,
      "eval_runtime": 563.3055,
      "eval_samples_per_second": 100.178,
      "eval_steps_per_second": 3.132,
      "step": 19500
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0447243452072144,
      "learning_rate": 7.1041007512152e-05,
      "loss": 2.1649,
      "step": 20000
    },
    {
      "epoch": 0.93,
      "eval_loss": 2.236150026321411,
      "eval_runtime": 563.3424,
      "eval_samples_per_second": 100.172,
      "eval_steps_per_second": 3.131,
      "step": 20000
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.039281964302063,
      "learning_rate": 4.649394608926206e-05,
      "loss": 2.1606,
      "step": 20500
    },
    {
      "epoch": 0.96,
      "eval_loss": 2.228179693222046,
      "eval_runtime": 563.3883,
      "eval_samples_per_second": 100.164,
      "eval_steps_per_second": 3.131,
      "step": 20500
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0719681978225708,
      "learning_rate": 2.1946884666372103e-05,
      "loss": 2.1521,
      "step": 21000
    },
    {
      "epoch": 0.98,
      "eval_loss": 2.221137285232544,
      "eval_runtime": 563.1866,
      "eval_samples_per_second": 100.199,
      "eval_steps_per_second": 3.132,
      "step": 21000
    }
  ],
  "logging_steps": 500,
  "max_steps": 21439,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 7000,
  "total_flos": 4102529679360000.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}