{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8,
  "eval_steps": 100,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 0.7975468060001731,
      "epoch": 0.016,
      "grad_norm": 48.0,
      "learning_rate": 1.8e-07,
      "loss": 13.872,
      "mean_token_accuracy": 0.7770325228571892,
      "num_tokens": 338352.0,
      "step": 10
    },
    {
      "entropy": 0.7971246171742677,
      "epoch": 0.032,
      "grad_norm": 38.75,
      "learning_rate": 3.8e-07,
      "loss": 13.8259,
      "mean_token_accuracy": 0.7767761040478944,
      "num_tokens": 671193.0,
      "step": 20
    },
    {
      "entropy": 0.7927551301196217,
      "epoch": 0.048,
      "grad_norm": 36.25,
      "learning_rate": 5.800000000000001e-07,
      "loss": 13.6952,
      "mean_token_accuracy": 0.7782795619219541,
      "num_tokens": 1001008.0,
      "step": 30
    },
    {
      "entropy": 0.8188681453466415,
      "epoch": 0.064,
      "grad_norm": 33.5,
      "learning_rate": 7.8e-07,
      "loss": 13.9837,
      "mean_token_accuracy": 0.7739029213786125,
      "num_tokens": 1331208.0,
      "step": 40
    },
    {
      "entropy": 0.8227416690438986,
      "epoch": 0.08,
      "grad_norm": 32.75,
      "learning_rate": 9.800000000000001e-07,
      "loss": 13.7971,
      "mean_token_accuracy": 0.7759626224637032,
      "num_tokens": 1667532.0,
      "step": 50
    },
    {
      "entropy": 0.809746851399541,
      "epoch": 0.096,
      "grad_norm": 34.0,
      "learning_rate": 1.1800000000000001e-06,
      "loss": 13.4702,
      "mean_token_accuracy": 0.7815078292042017,
      "num_tokens": 2007176.0,
      "step": 60
    },
    {
      "entropy": 0.8254514675587415,
      "epoch": 0.112,
      "grad_norm": 36.75,
      "learning_rate": 1.3800000000000001e-06,
      "loss": 13.7471,
      "mean_token_accuracy": 0.7760862343013286,
      "num_tokens": 2342259.0,
      "step": 70
    },
    {
      "entropy": 0.8527037408202887,
      "epoch": 0.128,
      "grad_norm": 27.125,
      "learning_rate": 1.5800000000000001e-06,
      "loss": 13.9233,
      "mean_token_accuracy": 0.7734280046075582,
      "num_tokens": 2663987.0,
      "step": 80
    },
    {
      "entropy": 0.8274104345589877,
      "epoch": 0.144,
      "grad_norm": 25.75,
      "learning_rate": 1.7800000000000001e-06,
      "loss": 13.653,
      "mean_token_accuracy": 0.7781244270503521,
      "num_tokens": 2997824.0,
      "step": 90
    },
    {
      "entropy": 0.8279923181980848,
      "epoch": 0.16,
      "grad_norm": 26.5,
      "learning_rate": 1.98e-06,
      "loss": 13.4841,
      "mean_token_accuracy": 0.7816651199012995,
      "num_tokens": 3330597.0,
      "step": 100
    },
    {
      "epoch": 0.16,
      "eval_biology_entropy": 1.125685109615326,
      "eval_biology_loss": 1.1623297929763794,
      "eval_biology_mean_token_accuracy": 0.7038317298889161,
      "eval_biology_num_tokens": 3330597.0,
      "eval_biology_runtime": 45.8822,
      "eval_biology_samples_per_second": 10.897,
      "eval_biology_steps_per_second": 2.724,
      "step": 100
    },
    {
      "epoch": 0.16,
      "eval_physics_entropy": 0.8375687637329101,
      "eval_physics_loss": 0.8513404130935669,
      "eval_physics_mean_token_accuracy": 0.7783295741081238,
      "eval_physics_num_tokens": 3330597.0,
      "eval_physics_runtime": 65.9246,
      "eval_physics_samples_per_second": 7.584,
      "eval_physics_steps_per_second": 1.896,
      "step": 100
    },
    {
      "entropy": 0.8265087101608515,
      "epoch": 0.176,
      "grad_norm": 33.0,
      "learning_rate": 2.1800000000000003e-06,
      "loss": 13.4457,
      "mean_token_accuracy": 0.7798256956040859,
      "num_tokens": 3658264.0,
      "step": 110
    },
    {
      "entropy": 0.8278129205107689,
      "epoch": 0.192,
      "grad_norm": 25.75,
      "learning_rate": 2.38e-06,
      "loss": 13.4175,
      "mean_token_accuracy": 0.7793817535042763,
      "num_tokens": 3995568.0,
      "step": 120
    },
    {
      "entropy": 0.818607559427619,
      "epoch": 0.208,
      "grad_norm": 26.125,
      "learning_rate": 2.5800000000000003e-06,
      "loss": 13.2096,
      "mean_token_accuracy": 0.7836348541080952,
      "num_tokens": 4321436.0,
      "step": 130
    },
    {
      "entropy": 0.8120364205911755,
      "epoch": 0.224,
      "grad_norm": 23.625,
      "learning_rate": 2.7800000000000005e-06,
      "loss": 13.0693,
      "mean_token_accuracy": 0.7830110590904951,
      "num_tokens": 4648491.0,
      "step": 140
    },
    {
      "entropy": 0.8133310537785292,
      "epoch": 0.24,
      "grad_norm": 23.75,
      "learning_rate": 2.9800000000000003e-06,
      "loss": 13.1184,
      "mean_token_accuracy": 0.7830653071403504,
      "num_tokens": 4986175.0,
      "step": 150
    },
    {
      "entropy": 0.8032114181667567,
      "epoch": 0.256,
      "grad_norm": 27.375,
      "learning_rate": 3.1800000000000005e-06,
      "loss": 13.0503,
      "mean_token_accuracy": 0.7851274147629738,
      "num_tokens": 5329320.0,
      "step": 160
    },
    {
      "entropy": 0.8205402866005898,
      "epoch": 0.272,
      "grad_norm": 26.125,
      "learning_rate": 3.3800000000000007e-06,
      "loss": 13.1719,
      "mean_token_accuracy": 0.7810409177094698,
      "num_tokens": 5658796.0,
      "step": 170
    },
    {
      "entropy": 0.8173610385507345,
      "epoch": 0.288,
      "grad_norm": 22.25,
      "learning_rate": 3.58e-06,
      "loss": 13.1428,
      "mean_token_accuracy": 0.7840665753930807,
      "num_tokens": 5980201.0,
      "step": 180
    },
    {
      "entropy": 0.7997741607949138,
      "epoch": 0.304,
      "grad_norm": 24.875,
      "learning_rate": 3.7800000000000002e-06,
      "loss": 12.8822,
      "mean_token_accuracy": 0.7862737070769071,
      "num_tokens": 6298858.0,
      "step": 190
    },
    {
      "entropy": 0.8092747095972299,
      "epoch": 0.32,
      "grad_norm": 22.875,
      "learning_rate": 3.980000000000001e-06,
      "loss": 13.0185,
      "mean_token_accuracy": 0.784797790274024,
      "num_tokens": 6622798.0,
      "step": 200
    },
    {
      "epoch": 0.32,
      "eval_biology_entropy": 1.1380381927490235,
      "eval_biology_loss": 1.1702971458435059,
      "eval_biology_mean_token_accuracy": 0.7033402571678161,
      "eval_biology_num_tokens": 6622798.0,
      "eval_biology_runtime": 175.631,
      "eval_biology_samples_per_second": 2.847,
      "eval_biology_steps_per_second": 0.712,
      "step": 200
    },
    {
      "epoch": 0.32,
      "eval_physics_entropy": 0.812020037651062,
      "eval_physics_loss": 0.8106491565704346,
      "eval_physics_mean_token_accuracy": 0.7852507462501526,
      "eval_physics_num_tokens": 6622798.0,
      "eval_physics_runtime": 200.6075,
      "eval_physics_samples_per_second": 2.492,
      "eval_physics_steps_per_second": 0.623,
      "step": 200
    },
    {
      "entropy": 0.797424552589655,
      "epoch": 0.336,
      "grad_norm": 21.5,
      "learning_rate": 4.18e-06,
      "loss": 12.8378,
      "mean_token_accuracy": 0.7871691755950451,
      "num_tokens": 6953611.0,
      "step": 210
    },
    {
      "entropy": 0.7838323166593909,
      "epoch": 0.352,
      "grad_norm": 24.0,
      "learning_rate": 4.38e-06,
      "loss": 12.6327,
      "mean_token_accuracy": 0.7907786477357149,
      "num_tokens": 7290378.0,
      "step": 220
    },
    {
      "entropy": 0.7731631100177765,
      "epoch": 0.368,
      "grad_norm": 23.25,
      "learning_rate": 4.58e-06,
      "loss": 12.3982,
      "mean_token_accuracy": 0.7930114820599556,
      "num_tokens": 7621082.0,
      "step": 230
    },
    {
      "entropy": 0.801326191239059,
      "epoch": 0.384,
      "grad_norm": 22.75,
      "learning_rate": 4.78e-06,
      "loss": 12.8917,
      "mean_token_accuracy": 0.7840922765433789,
      "num_tokens": 7955570.0,
      "step": 240
    },
    {
      "entropy": 0.7741426464170218,
      "epoch": 0.4,
      "grad_norm": 23.875,
      "learning_rate": 4.980000000000001e-06,
      "loss": 12.4671,
      "mean_token_accuracy": 0.791225866228342,
      "num_tokens": 8291049.0,
      "step": 250
    },
    {
      "entropy": 0.7628269851207733,
      "epoch": 0.416,
      "grad_norm": 22.125,
      "learning_rate": 5.18e-06,
      "loss": 12.2906,
      "mean_token_accuracy": 0.7947692718356848,
      "num_tokens": 8627310.0,
      "step": 260
    },
    {
      "entropy": 0.7752459084615111,
      "epoch": 0.432,
      "grad_norm": 23.125,
      "learning_rate": 5.380000000000001e-06,
      "loss": 12.4973,
      "mean_token_accuracy": 0.7905333787202835,
      "num_tokens": 8958371.0,
      "step": 270
    },
    {
      "entropy": 0.7545073958113789,
      "epoch": 0.448,
      "grad_norm": 22.625,
      "learning_rate": 5.580000000000001e-06,
      "loss": 12.1817,
      "mean_token_accuracy": 0.7961019795387984,
      "num_tokens": 9294388.0,
      "step": 280
    },
    {
      "entropy": 0.7482077127322555,
      "epoch": 0.464,
      "grad_norm": 23.375,
      "learning_rate": 5.78e-06,
      "loss": 11.9906,
      "mean_token_accuracy": 0.7969086967408657,
      "num_tokens": 9620366.0,
      "step": 290
    },
    {
      "entropy": 0.7531273065134882,
      "epoch": 0.48,
      "grad_norm": 21.0,
      "learning_rate": 5.98e-06,
      "loss": 12.0906,
      "mean_token_accuracy": 0.7956845626235008,
      "num_tokens": 9955431.0,
      "step": 300
    },
    {
      "epoch": 0.48,
      "eval_biology_entropy": 1.1282176275253295,
      "eval_biology_loss": 1.176755428314209,
      "eval_biology_mean_token_accuracy": 0.7018080081939697,
      "eval_biology_num_tokens": 9955431.0,
      "eval_biology_runtime": 45.9288,
      "eval_biology_samples_per_second": 10.886,
      "eval_biology_steps_per_second": 2.722,
      "step": 300
    },
    {
      "epoch": 0.48,
      "eval_physics_entropy": 0.7810471363067627,
      "eval_physics_loss": 0.7771082520484924,
      "eval_physics_mean_token_accuracy": 0.7920530920028687,
      "eval_physics_num_tokens": 9955431.0,
      "eval_physics_runtime": 66.0018,
      "eval_physics_samples_per_second": 7.576,
      "eval_physics_steps_per_second": 1.894,
      "step": 300
    },
    {
      "entropy": 0.7689633307978511,
      "epoch": 0.496,
      "grad_norm": 20.875,
      "learning_rate": 6.18e-06,
      "loss": 12.3289,
      "mean_token_accuracy": 0.7935276433825493,
      "num_tokens": 10289883.0,
      "step": 310
    },
    {
      "entropy": 0.7578680694103241,
      "epoch": 0.512,
      "grad_norm": 25.0,
      "learning_rate": 6.380000000000001e-06,
      "loss": 12.2227,
      "mean_token_accuracy": 0.7949301645159721,
      "num_tokens": 10619468.0,
      "step": 320
    },
    {
      "entropy": 0.7704175990074873,
      "epoch": 0.528,
      "grad_norm": 20.75,
      "learning_rate": 6.5800000000000005e-06,
      "loss": 12.3943,
      "mean_token_accuracy": 0.7915358774363994,
      "num_tokens": 10944904.0,
      "step": 330
    },
    {
      "entropy": 0.7533261435106396,
      "epoch": 0.544,
      "grad_norm": 23.125,
      "learning_rate": 6.780000000000001e-06,
      "loss": 12.1385,
      "mean_token_accuracy": 0.7953089620918036,
      "num_tokens": 11266409.0,
      "step": 340
    },
    {
      "entropy": 0.7507271701470017,
      "epoch": 0.56,
      "grad_norm": 18.875,
      "learning_rate": 6.98e-06,
      "loss": 12.0437,
      "mean_token_accuracy": 0.79745435975492,
      "num_tokens": 11605544.0,
      "step": 350
    },
    {
      "entropy": 0.7390375791117549,
      "epoch": 0.576,
      "grad_norm": 21.625,
      "learning_rate": 7.180000000000001e-06,
      "loss": 11.8508,
      "mean_token_accuracy": 0.8004897948354482,
      "num_tokens": 11931715.0,
      "step": 360
    },
    {
      "entropy": 0.7514585722237825,
      "epoch": 0.592,
      "grad_norm": 21.5,
      "learning_rate": 7.3800000000000005e-06,
      "loss": 12.0218,
      "mean_token_accuracy": 0.7965340100228786,
      "num_tokens": 12257243.0,
      "step": 370
    },
    {
      "entropy": 0.7325067885220051,
      "epoch": 0.608,
      "grad_norm": 20.75,
      "learning_rate": 7.58e-06,
      "loss": 11.7926,
      "mean_token_accuracy": 0.7993212066590786,
      "num_tokens": 12583154.0,
      "step": 380
    },
    {
      "entropy": 0.7487934850156307,
      "epoch": 0.624,
      "grad_norm": 21.0,
      "learning_rate": 7.78e-06,
      "loss": 12.053,
      "mean_token_accuracy": 0.7958210565149784,
      "num_tokens": 12905392.0,
      "step": 390
    },
    {
      "entropy": 0.7440792236477136,
      "epoch": 0.64,
      "grad_norm": 18.75,
      "learning_rate": 7.980000000000002e-06,
      "loss": 11.9643,
      "mean_token_accuracy": 0.7981198724359274,
      "num_tokens": 13232198.0,
      "step": 400
    },
    {
      "epoch": 0.64,
      "eval_biology_entropy": 1.0903304896354675,
      "eval_biology_loss": 1.182153344154358,
      "eval_biology_mean_token_accuracy": 0.7013925290107728,
      "eval_biology_num_tokens": 13232198.0,
      "eval_biology_runtime": 45.8492,
      "eval_biology_samples_per_second": 10.905,
      "eval_biology_steps_per_second": 2.726,
      "step": 400
    },
    {
      "epoch": 0.64,
      "eval_physics_entropy": 0.7371547894477845,
      "eval_physics_loss": 0.7496699094772339,
      "eval_physics_mean_token_accuracy": 0.7974319715499878,
      "eval_physics_num_tokens": 13232198.0,
      "eval_physics_runtime": 65.8732,
      "eval_physics_samples_per_second": 7.59,
      "eval_physics_steps_per_second": 1.898,
      "step": 400
    },
    {
      "entropy": 0.7319347187876701,
      "epoch": 0.656,
      "grad_norm": 18.375,
      "learning_rate": 8.18e-06,
      "loss": 11.7795,
      "mean_token_accuracy": 0.8000467628240585,
      "num_tokens": 13575902.0,
      "step": 410
    },
    {
      "entropy": 0.71892513576895,
      "epoch": 0.672,
      "grad_norm": 21.25,
      "learning_rate": 8.380000000000001e-06,
      "loss": 11.525,
      "mean_token_accuracy": 0.8044146560132504,
      "num_tokens": 13895997.0,
      "step": 420
    },
    {
      "entropy": 0.7239193903282285,
      "epoch": 0.688,
      "grad_norm": 21.375,
      "learning_rate": 8.580000000000001e-06,
      "loss": 11.6384,
      "mean_token_accuracy": 0.8023904841393232,
      "num_tokens": 14234888.0,
      "step": 430
    },
    {
      "entropy": 0.7392614649608731,
      "epoch": 0.704,
      "grad_norm": 21.875,
      "learning_rate": 8.78e-06,
      "loss": 11.9131,
      "mean_token_accuracy": 0.7982321321964264,
      "num_tokens": 14567908.0,
      "step": 440
    },
    {
      "entropy": 0.7193062495440244,
      "epoch": 0.72,
      "grad_norm": 23.5,
      "learning_rate": 8.98e-06,
      "loss": 11.5749,
      "mean_token_accuracy": 0.8031982038170099,
      "num_tokens": 14882927.0,
      "step": 450
    },
    {
      "entropy": 0.7137783830985427,
      "epoch": 0.736,
      "grad_norm": 22.375,
      "learning_rate": 9.180000000000002e-06,
      "loss": 11.4717,
      "mean_token_accuracy": 0.8033147465437651,
      "num_tokens": 15217342.0,
      "step": 460
    },
    {
      "entropy": 0.7088428331539035,
      "epoch": 0.752,
      "grad_norm": 18.75,
      "learning_rate": 9.38e-06,
      "loss": 11.392,
      "mean_token_accuracy": 0.8049089256674051,
      "num_tokens": 15547710.0,
      "step": 470
    },
    {
      "entropy": 0.7304569650441408,
      "epoch": 0.768,
      "grad_norm": 20.625,
      "learning_rate": 9.58e-06,
      "loss": 11.7448,
      "mean_token_accuracy": 0.7994368057698011,
      "num_tokens": 15877177.0,
      "step": 480
    },
    {
      "entropy": 0.7059216756373644,
      "epoch": 0.784,
      "grad_norm": 18.75,
      "learning_rate": 9.780000000000001e-06,
      "loss": 11.3295,
      "mean_token_accuracy": 0.806659733131528,
      "num_tokens": 16219640.0,
      "step": 490
    },
    {
      "entropy": 0.712453056499362,
      "epoch": 0.8,
      "grad_norm": 19.125,
      "learning_rate": 9.980000000000001e-06,
      "loss": 11.4991,
      "mean_token_accuracy": 0.8039865717291832,
      "num_tokens": 16548261.0,
      "step": 500
    },
    {
      "epoch": 0.8,
      "eval_biology_entropy": 1.1249133114814758,
      "eval_biology_loss": 1.1854252815246582,
      "eval_biology_mean_token_accuracy": 0.7005919570922852,
      "eval_biology_num_tokens": 16548261.0,
      "eval_biology_runtime": 45.6964,
      "eval_biology_samples_per_second": 10.942,
      "eval_biology_steps_per_second": 2.735,
      "step": 500
    },
    {
      "epoch": 0.8,
      "eval_physics_entropy": 0.730017361164093,
      "eval_physics_loss": 0.724478542804718,
      "eval_physics_mean_token_accuracy": 0.8027530283927917,
      "eval_physics_num_tokens": 16548261.0,
      "eval_physics_runtime": 65.8504,
      "eval_physics_samples_per_second": 7.593,
      "eval_physics_steps_per_second": 1.898,
      "step": 500
    }
  ],
  "logging_steps": 10,
  "max_steps": 10000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 16,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.652179086470234e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}