{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.2,
  "eval_steps": 100,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 0.7213201837614178,
      "epoch": 0.016,
      "grad_norm": 127.0,
      "learning_rate": 6.000000000000001e-07,
      "loss": 33.4275,
      "mean_token_accuracy": 0.6860143840312958,
      "num_tokens": 280941.0,
      "step": 10
    },
    {
      "entropy": 0.75641950070858,
      "epoch": 0.032,
      "grad_norm": 118.0,
      "learning_rate": 1.2666666666666669e-06,
      "loss": 33.8085,
      "mean_token_accuracy": 0.6810433834791183,
      "num_tokens": 558056.0,
      "step": 20
    },
    {
      "entropy": 0.7386618180200457,
      "epoch": 0.048,
      "grad_norm": 105.5,
      "learning_rate": 1.9333333333333336e-06,
      "loss": 33.0823,
      "mean_token_accuracy": 0.686032148450613,
      "num_tokens": 836753.0,
      "step": 30
    },
    {
      "entropy": 0.7653292674571276,
      "epoch": 0.064,
      "grad_norm": 106.0,
      "learning_rate": 2.6e-06,
      "loss": 33.3421,
      "mean_token_accuracy": 0.678343191742897,
      "num_tokens": 1126446.0,
      "step": 40
    },
    {
      "entropy": 0.7746079994365573,
      "epoch": 0.08,
      "grad_norm": 112.0,
      "learning_rate": 3.266666666666667e-06,
      "loss": 31.8715,
      "mean_token_accuracy": 0.6860048722475767,
      "num_tokens": 1413596.0,
      "step": 50
    },
    {
      "entropy": 0.7967683909460902,
      "epoch": 0.096,
      "grad_norm": 87.5,
      "learning_rate": 3.9333333333333335e-06,
      "loss": 30.4245,
      "mean_token_accuracy": 0.6932173065841198,
      "num_tokens": 1701193.0,
      "step": 60
    },
    {
      "entropy": 0.8636196419596672,
      "epoch": 0.112,
      "grad_norm": 63.5,
      "learning_rate": 4.600000000000001e-06,
      "loss": 28.963,
      "mean_token_accuracy": 0.6949520386755467,
      "num_tokens": 1979232.0,
      "step": 70
    },
    {
      "entropy": 0.986144457757473,
      "epoch": 0.128,
      "grad_norm": 45.25,
      "learning_rate": 5.2666666666666665e-06,
      "loss": 27.4106,
      "mean_token_accuracy": 0.7027070496231318,
      "num_tokens": 2274177.0,
      "step": 80
    },
    {
      "entropy": 1.0647853959351778,
      "epoch": 0.144,
      "grad_norm": 34.5,
      "learning_rate": 5.933333333333335e-06,
      "loss": 25.7084,
      "mean_token_accuracy": 0.7127731442451477,
      "num_tokens": 2548445.0,
      "step": 90
    },
    {
      "entropy": 1.129902769997716,
      "epoch": 0.16,
      "grad_norm": 29.625,
      "learning_rate": 6.600000000000001e-06,
      "loss": 23.6215,
      "mean_token_accuracy": 0.724789521843195,
      "num_tokens": 2824418.0,
      "step": 100
    },
    {
      "epoch": 0.16,
      "eval_biology_entropy": 3.7158153285980227,
      "eval_biology_loss": 4.645914554595947,
      "eval_biology_mean_token_accuracy": 0.3934867687225342,
      "eval_biology_num_tokens": 2824418.0,
      "eval_biology_runtime": 58.324,
      "eval_biology_samples_per_second": 8.573,
      "eval_biology_steps_per_second": 2.143,
      "step": 100
    },
    {
      "epoch": 0.16,
      "eval_chemistry_entropy": 2.4825774974823,
      "eval_chemistry_loss": 2.797250509262085,
      "eval_chemistry_mean_token_accuracy": 0.5806102266311646,
      "eval_chemistry_num_tokens": 2824418.0,
      "eval_chemistry_runtime": 72.1529,
      "eval_chemistry_samples_per_second": 6.93,
      "eval_chemistry_steps_per_second": 1.732,
      "step": 100
    },
    {
      "epoch": 0.16,
      "eval_math_entropy": 1.2174424457550048,
      "eval_math_loss": 1.4766321182250977,
      "eval_math_mean_token_accuracy": 0.7193697824478149,
      "eval_math_num_tokens": 2824418.0,
      "eval_math_runtime": 75.2312,
      "eval_math_samples_per_second": 6.646,
      "eval_math_steps_per_second": 1.662,
      "step": 100
    },
    {
      "epoch": 0.16,
      "eval_physics_entropy": 2.0236835989952087,
      "eval_physics_loss": 2.248775005340576,
      "eval_physics_mean_token_accuracy": 0.6399151833057404,
      "eval_physics_num_tokens": 2824418.0,
      "eval_physics_runtime": 85.7293,
      "eval_physics_samples_per_second": 5.832,
      "eval_physics_steps_per_second": 1.458,
      "step": 100
    },
    {
      "entropy": 1.1890651997178794,
      "epoch": 0.176,
      "grad_norm": 24.5,
      "learning_rate": 7.266666666666668e-06,
      "loss": 22.5161,
      "mean_token_accuracy": 0.7291781000792981,
      "num_tokens": 3110313.0,
      "step": 110
    },
    {
      "entropy": 1.2231741398572922,
      "epoch": 0.192,
      "grad_norm": 19.375,
      "learning_rate": 7.933333333333334e-06,
      "loss": 22.3058,
      "mean_token_accuracy": 0.7288031376898289,
      "num_tokens": 3394170.0,
      "step": 120
    },
    {
      "entropy": 1.1990749202668667,
      "epoch": 0.208,
      "grad_norm": 17.125,
      "learning_rate": 8.6e-06,
      "loss": 21.6321,
      "mean_token_accuracy": 0.7331194914877415,
      "num_tokens": 3673600.0,
      "step": 130
    },
    {
      "entropy": 1.1935963843017816,
      "epoch": 0.224,
      "grad_norm": 18.5,
      "learning_rate": 9.266666666666667e-06,
      "loss": 20.9973,
      "mean_token_accuracy": 0.7374927394092083,
      "num_tokens": 3953732.0,
      "step": 140
    },
    {
      "entropy": 1.1516071777790784,
      "epoch": 0.24,
      "grad_norm": 15.6875,
      "learning_rate": 9.933333333333334e-06,
      "loss": 20.0443,
      "mean_token_accuracy": 0.7471608892083168,
      "num_tokens": 4243655.0,
      "step": 150
    },
    {
      "entropy": 1.1274361703544855,
      "epoch": 0.256,
      "grad_norm": 17.125,
      "learning_rate": 1.0600000000000002e-05,
      "loss": 19.5101,
      "mean_token_accuracy": 0.7516594093292952,
      "num_tokens": 4531471.0,
      "step": 160
    },
    {
      "entropy": 1.140014799311757,
      "epoch": 0.272,
      "grad_norm": 14.0,
      "learning_rate": 1.1266666666666668e-05,
      "loss": 19.5844,
      "mean_token_accuracy": 0.7498023040592671,
      "num_tokens": 4810284.0,
      "step": 170
    },
    {
      "entropy": 1.1203772608190774,
      "epoch": 0.288,
      "grad_norm": 12.5625,
      "learning_rate": 1.1933333333333335e-05,
      "loss": 19.1823,
      "mean_token_accuracy": 0.753832882642746,
      "num_tokens": 5095104.0,
      "step": 180
    },
    {
      "entropy": 1.129030792787671,
      "epoch": 0.304,
      "grad_norm": 12.9375,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 19.1862,
      "mean_token_accuracy": 0.7524638257920742,
      "num_tokens": 5383732.0,
      "step": 190
    },
    {
      "entropy": 1.090552070736885,
      "epoch": 0.32,
      "grad_norm": 15.0625,
      "learning_rate": 1.3266666666666668e-05,
      "loss": 18.5807,
      "mean_token_accuracy": 0.7593985345214606,
      "num_tokens": 5676334.0,
      "step": 200
    },
    {
      "epoch": 0.32,
      "eval_biology_entropy": 3.458996726989746,
      "eval_biology_loss": 4.269806385040283,
      "eval_biology_mean_token_accuracy": 0.4222190487384796,
      "eval_biology_num_tokens": 5676334.0,
      "eval_biology_runtime": 58.3189,
      "eval_biology_samples_per_second": 8.574,
      "eval_biology_steps_per_second": 2.143,
      "step": 200
    },
    {
      "epoch": 0.32,
      "eval_chemistry_entropy": 2.180737380027771,
      "eval_chemistry_loss": 2.439823865890503,
      "eval_chemistry_mean_token_accuracy": 0.6122945117950439,
      "eval_chemistry_num_tokens": 5676334.0,
      "eval_chemistry_runtime": 72.4241,
      "eval_chemistry_samples_per_second": 6.904,
      "eval_chemistry_steps_per_second": 1.726,
      "step": 200
    },
    {
      "epoch": 0.32,
      "eval_math_entropy": 1.1109818544387817,
      "eval_math_loss": 1.1917932033538818,
      "eval_math_mean_token_accuracy": 0.752472719669342,
      "eval_math_num_tokens": 5676334.0,
      "eval_math_runtime": 75.8098,
      "eval_math_samples_per_second": 6.595,
      "eval_math_steps_per_second": 1.649,
      "step": 200
    },
    {
      "epoch": 0.32,
      "eval_physics_entropy": 1.7737830457687378,
      "eval_physics_loss": 1.935411810874939,
      "eval_physics_mean_token_accuracy": 0.670736522436142,
      "eval_physics_num_tokens": 5676334.0,
      "eval_physics_runtime": 86.0358,
      "eval_physics_samples_per_second": 5.812,
      "eval_physics_steps_per_second": 1.453,
      "step": 200
    },
    {
      "entropy": 1.1083258152008058,
      "epoch": 0.336,
      "grad_norm": 14.875,
      "learning_rate": 1.3933333333333334e-05,
      "loss": 18.9806,
      "mean_token_accuracy": 0.7541178163141012,
      "num_tokens": 5958480.0,
      "step": 210
    },
    {
      "entropy": 1.0838440161198377,
      "epoch": 0.352,
      "grad_norm": 13.625,
      "learning_rate": 1.46e-05,
      "loss": 18.2469,
      "mean_token_accuracy": 0.7600458469241858,
      "num_tokens": 6242161.0,
      "step": 220
    },
    {
      "entropy": 1.0777738582342864,
      "epoch": 0.368,
      "grad_norm": 14.6875,
      "learning_rate": 1.5266666666666667e-05,
      "loss": 18.3345,
      "mean_token_accuracy": 0.759134492278099,
      "num_tokens": 6523679.0,
      "step": 230
    },
    {
      "entropy": 1.0698408473283052,
      "epoch": 0.384,
      "grad_norm": 14.0,
      "learning_rate": 1.5933333333333336e-05,
      "loss": 17.9275,
      "mean_token_accuracy": 0.7646602354943752,
      "num_tokens": 6810978.0,
      "step": 240
    },
    {
      "entropy": 1.0921947304159403,
      "epoch": 0.4,
      "grad_norm": 13.8125,
      "learning_rate": 1.66e-05,
      "loss": 18.5932,
      "mean_token_accuracy": 0.757987079769373,
      "num_tokens": 7096903.0,
      "step": 250
    },
    {
      "entropy": 1.0657698467373848,
      "epoch": 0.416,
      "grad_norm": 12.9375,
      "learning_rate": 1.726666666666667e-05,
      "loss": 17.8874,
      "mean_token_accuracy": 0.7630033176392317,
      "num_tokens": 7377181.0,
      "step": 260
    },
    {
      "entropy": 1.0416713248938323,
      "epoch": 0.432,
      "grad_norm": 15.1875,
      "learning_rate": 1.7933333333333333e-05,
      "loss": 17.5118,
      "mean_token_accuracy": 0.7672814972698688,
      "num_tokens": 7650523.0,
      "step": 270
    },
    {
      "entropy": 1.0186962354928255,
      "epoch": 0.448,
      "grad_norm": 15.3125,
      "learning_rate": 1.86e-05,
      "loss": 17.3118,
      "mean_token_accuracy": 0.7682009607553482,
      "num_tokens": 7936788.0,
      "step": 280
    },
    {
      "entropy": 1.036542361229658,
      "epoch": 0.464,
      "grad_norm": 15.8125,
      "learning_rate": 1.926666666666667e-05,
      "loss": 17.3748,
      "mean_token_accuracy": 0.7694676879793405,
      "num_tokens": 8223147.0,
      "step": 290
    },
    {
      "entropy": 1.0569938763976097,
      "epoch": 0.48,
      "grad_norm": 14.5,
      "learning_rate": 1.9933333333333334e-05,
      "loss": 17.6393,
      "mean_token_accuracy": 0.7655092909932136,
      "num_tokens": 8506030.0,
      "step": 300
    },
    {
      "epoch": 0.48,
      "eval_biology_entropy": 3.567513717651367,
      "eval_biology_loss": 4.103325366973877,
      "eval_biology_mean_token_accuracy": 0.43289893221855164,
      "eval_biology_num_tokens": 8506030.0,
      "eval_biology_runtime": 58.2179,
      "eval_biology_samples_per_second": 8.588,
      "eval_biology_steps_per_second": 2.147,
      "step": 300
    },
    {
      "epoch": 0.48,
      "eval_chemistry_entropy": 2.159853268146515,
      "eval_chemistry_loss": 2.3111026287078857,
      "eval_chemistry_mean_token_accuracy": 0.6211932191848755,
      "eval_chemistry_num_tokens": 8506030.0,
      "eval_chemistry_runtime": 72.4538,
      "eval_chemistry_samples_per_second": 6.901,
      "eval_chemistry_steps_per_second": 1.725,
      "step": 300
    },
    {
      "epoch": 0.48,
      "eval_math_entropy": 1.055273208618164,
      "eval_math_loss": 1.103938341140747,
      "eval_math_mean_token_accuracy": 0.7635622372627259,
      "eval_math_num_tokens": 8506030.0,
      "eval_math_runtime": 75.7526,
      "eval_math_samples_per_second": 6.6,
      "eval_math_steps_per_second": 1.65,
      "step": 300
    },
    {
      "epoch": 0.48,
      "eval_physics_entropy": 1.7374450120925904,
      "eval_physics_loss": 1.8264347314834595,
      "eval_physics_mean_token_accuracy": 0.6800054814815522,
      "eval_physics_num_tokens": 8506030.0,
      "eval_physics_runtime": 85.6984,
      "eval_physics_samples_per_second": 5.834,
      "eval_physics_steps_per_second": 1.459,
      "step": 300
    },
    {
      "entropy": 1.0048395186662673,
      "epoch": 0.496,
      "grad_norm": 20.75,
      "learning_rate": 1.9933333333333334e-05,
      "loss": 17.0439,
      "mean_token_accuracy": 0.7706491403281689,
      "num_tokens": 8788726.0,
      "step": 310
    },
    {
      "entropy": 1.0206645488739015,
      "epoch": 0.512,
      "grad_norm": 14.125,
      "learning_rate": 1.985925925925926e-05,
      "loss": 17.0396,
      "mean_token_accuracy": 0.7719013765454292,
      "num_tokens": 9078039.0,
      "step": 320
    },
    {
      "entropy": 0.9999576449394226,
      "epoch": 0.528,
      "grad_norm": 15.0625,
      "learning_rate": 1.9785185185185187e-05,
      "loss": 16.7994,
      "mean_token_accuracy": 0.7749712273478508,
      "num_tokens": 9373860.0,
      "step": 330
    },
    {
      "entropy": 1.0083586767315864,
      "epoch": 0.544,
      "grad_norm": 15.0625,
      "learning_rate": 1.971111111111111e-05,
      "loss": 16.9209,
      "mean_token_accuracy": 0.7731953993439674,
      "num_tokens": 9660940.0,
      "step": 340
    },
    {
      "entropy": 1.0532048616558314,
      "epoch": 0.56,
      "grad_norm": 18.75,
      "learning_rate": 1.963703703703704e-05,
      "loss": 17.638,
      "mean_token_accuracy": 0.7637613207101822,
      "num_tokens": 9932302.0,
      "step": 350
    },
    {
      "entropy": 1.0087582137435676,
      "epoch": 0.576,
      "grad_norm": 17.75,
      "learning_rate": 1.9562962962962964e-05,
      "loss": 17.0002,
      "mean_token_accuracy": 0.7717114143073559,
      "num_tokens": 10215462.0,
      "step": 360
    },
    {
      "entropy": 1.0042562756687403,
      "epoch": 0.592,
      "grad_norm": 15.0,
      "learning_rate": 1.948888888888889e-05,
      "loss": 16.7266,
      "mean_token_accuracy": 0.7740717619657517,
      "num_tokens": 10504396.0,
      "step": 370
    },
    {
      "entropy": 0.9980081049725413,
      "epoch": 0.608,
      "grad_norm": 16.125,
      "learning_rate": 1.9414814814814817e-05,
      "loss": 16.8411,
      "mean_token_accuracy": 0.7726139184087515,
      "num_tokens": 10793126.0,
      "step": 380
    },
    {
      "entropy": 0.9865465115755796,
      "epoch": 0.624,
      "grad_norm": 14.5,
      "learning_rate": 1.9340740740740743e-05,
      "loss": 16.5077,
      "mean_token_accuracy": 0.7759908959269524,
      "num_tokens": 11081768.0,
      "step": 390
    },
    {
      "entropy": 0.9753864344209433,
      "epoch": 0.64,
      "grad_norm": 13.0625,
      "learning_rate": 1.926666666666667e-05,
      "loss": 16.3328,
      "mean_token_accuracy": 0.777353797480464,
      "num_tokens": 11370320.0,
      "step": 400
    },
    {
      "epoch": 0.64,
      "eval_biology_entropy": 3.4534600849151613,
      "eval_biology_loss": 4.02693510055542,
      "eval_biology_mean_token_accuracy": 0.4380790627002716,
      "eval_biology_num_tokens": 11370320.0,
      "eval_biology_runtime": 58.2953,
      "eval_biology_samples_per_second": 8.577,
      "eval_biology_steps_per_second": 2.144,
      "step": 400
    },
    {
      "epoch": 0.64,
      "eval_chemistry_entropy": 2.068386361122131,
      "eval_chemistry_loss": 2.252868175506592,
      "eval_chemistry_mean_token_accuracy": 0.6284125361442566,
      "eval_chemistry_num_tokens": 11370320.0,
      "eval_chemistry_runtime": 72.4168,
      "eval_chemistry_samples_per_second": 6.904,
      "eval_chemistry_steps_per_second": 1.726,
      "step": 400
    },
    {
      "epoch": 0.64,
      "eval_math_entropy": 1.008522126674652,
      "eval_math_loss": 1.0611059665679932,
      "eval_math_mean_token_accuracy": 0.7697986388206481,
      "eval_math_num_tokens": 11370320.0,
      "eval_math_runtime": 75.8656,
      "eval_math_samples_per_second": 6.591,
      "eval_math_steps_per_second": 1.648,
      "step": 400
    },
    {
      "epoch": 0.64,
      "eval_physics_entropy": 1.6607081370353698,
      "eval_physics_loss": 1.7777302265167236,
      "eval_physics_mean_token_accuracy": 0.685634191274643,
      "eval_physics_num_tokens": 11370320.0,
      "eval_physics_runtime": 86.0729,
      "eval_physics_samples_per_second": 5.809,
      "eval_physics_steps_per_second": 1.452,
      "step": 400
    },
    {
      "entropy": 0.9546368110924959,
      "epoch": 0.656,
      "grad_norm": 13.3125,
      "learning_rate": 1.9192592592592593e-05,
      "loss": 16.1585,
      "mean_token_accuracy": 0.7791520744562149,
      "num_tokens": 11657835.0,
      "step": 410
    },
    {
      "entropy": 1.0121946074068546,
      "epoch": 0.672,
      "grad_norm": 14.9375,
      "learning_rate": 1.911851851851852e-05,
      "loss": 16.7638,
      "mean_token_accuracy": 0.7723276257514954,
      "num_tokens": 11949262.0,
      "step": 420
    },
    {
      "entropy": 0.9564321564510465,
      "epoch": 0.688,
      "grad_norm": 12.875,
      "learning_rate": 1.9044444444444446e-05,
      "loss": 16.2106,
      "mean_token_accuracy": 0.7782567508518696,
      "num_tokens": 12227640.0,
      "step": 430
    },
    {
      "entropy": 0.9711398147046566,
      "epoch": 0.704,
      "grad_norm": 12.875,
      "learning_rate": 1.8970370370370372e-05,
      "loss": 16.2595,
      "mean_token_accuracy": 0.7794518828392029,
      "num_tokens": 12516641.0,
      "step": 440
    },
    {
      "entropy": 0.990043681114912,
      "epoch": 0.72,
      "grad_norm": 16.0,
      "learning_rate": 1.8896296296296295e-05,
      "loss": 16.6764,
      "mean_token_accuracy": 0.7741847857832909,
      "num_tokens": 12793343.0,
      "step": 450
    },
    {
      "entropy": 0.9794478964060545,
      "epoch": 0.736,
      "grad_norm": 13.5625,
      "learning_rate": 1.8822222222222225e-05,
      "loss": 16.3042,
      "mean_token_accuracy": 0.7780686043202877,
      "num_tokens": 13077981.0,
      "step": 460
    },
    {
      "entropy": 0.9613467682152986,
      "epoch": 0.752,
      "grad_norm": 14.6875,
      "learning_rate": 1.874814814814815e-05,
      "loss": 16.0345,
      "mean_token_accuracy": 0.7808727387338876,
      "num_tokens": 13358957.0,
      "step": 470
    },
    {
      "entropy": 0.9237693391740323,
      "epoch": 0.768,
      "grad_norm": 12.0,
      "learning_rate": 1.8674074074074075e-05,
      "loss": 15.6444,
      "mean_token_accuracy": 0.7876643743366003,
      "num_tokens": 13653412.0,
      "step": 480
    },
    {
      "entropy": 1.0019762709736824,
      "epoch": 0.784,
      "grad_norm": 15.6875,
      "learning_rate": 1.86e-05,
      "loss": 16.7034,
      "mean_token_accuracy": 0.7748573537915945,
      "num_tokens": 13940856.0,
      "step": 490
    },
    {
      "entropy": 0.9975844457745552,
      "epoch": 0.8,
      "grad_norm": 17.0,
      "learning_rate": 1.8525925925925928e-05,
      "loss": 16.7128,
      "mean_token_accuracy": 0.77371263243258,
      "num_tokens": 14230754.0,
      "step": 500
    },
    {
      "epoch": 0.8,
      "eval_biology_entropy": 3.498454050064087,
      "eval_biology_loss": 3.949568271636963,
      "eval_biology_mean_token_accuracy": 0.44270548820495603,
      "eval_biology_num_tokens": 14230754.0,
      "eval_biology_runtime": 58.3602,
      "eval_biology_samples_per_second": 8.567,
      "eval_biology_steps_per_second": 2.142,
      "step": 500
    },
    {
      "epoch": 0.8,
      "eval_chemistry_entropy": 2.073633497238159,
      "eval_chemistry_loss": 2.1966874599456787,
      "eval_chemistry_mean_token_accuracy": 0.6324296245574951,
      "eval_chemistry_num_tokens": 14230754.0,
      "eval_chemistry_runtime": 71.3745,
      "eval_chemistry_samples_per_second": 7.005,
      "eval_chemistry_steps_per_second": 1.751,
      "step": 500
    },
    {
      "epoch": 0.8,
      "eval_math_entropy": 0.9948915004730224,
      "eval_math_loss": 1.0365564823150635,
      "eval_math_mean_token_accuracy": 0.7735790548324585,
      "eval_math_num_tokens": 14230754.0,
      "eval_math_runtime": 75.2012,
      "eval_math_samples_per_second": 6.649,
      "eval_math_steps_per_second": 1.662,
      "step": 500
    },
    {
      "epoch": 0.8,
      "eval_physics_entropy": 1.6648971457481385,
      "eval_physics_loss": 1.736081838607788,
      "eval_physics_mean_token_accuracy": 0.6900339169502259,
      "eval_physics_num_tokens": 14230754.0,
      "eval_physics_runtime": 84.9467,
      "eval_physics_samples_per_second": 5.886,
      "eval_physics_steps_per_second": 1.472,
      "step": 500
    },
    {
      "entropy": 0.9665802534669637,
      "epoch": 0.816,
      "grad_norm": 15.5,
      "learning_rate": 1.8451851851851855e-05,
      "loss": 16.0969,
      "mean_token_accuracy": 0.77869370020926,
      "num_tokens": 14519893.0,
      "step": 510
    },
    {
      "entropy": 0.9573450323194266,
      "epoch": 0.832,
      "grad_norm": 14.625,
      "learning_rate": 1.8377777777777778e-05,
      "loss": 16.1991,
      "mean_token_accuracy": 0.7803513094782829,
      "num_tokens": 14805088.0,
      "step": 520
    },
    {
      "entropy": 0.9879640638828278,
      "epoch": 0.848,
      "grad_norm": 14.0625,
      "learning_rate": 1.8303703703703704e-05,
      "loss": 16.5646,
      "mean_token_accuracy": 0.7745878294110298,
      "num_tokens": 15086306.0,
      "step": 530
    },
    {
      "entropy": 0.974746185541153,
      "epoch": 0.864,
      "grad_norm": 13.6875,
      "learning_rate": 1.822962962962963e-05,
      "loss": 16.3173,
      "mean_token_accuracy": 0.7770581971853971,
      "num_tokens": 15370985.0,
      "step": 540
    },
    {
      "entropy": 0.9910150848329067,
      "epoch": 0.88,
      "grad_norm": 17.0,
      "learning_rate": 1.8155555555555557e-05,
      "loss": 16.5347,
      "mean_token_accuracy": 0.7754124801605939,
      "num_tokens": 15653836.0,
      "step": 550
    },
    {
      "entropy": 0.9755746092647314,
      "epoch": 0.896,
      "grad_norm": 14.3125,
      "learning_rate": 1.8081481481481484e-05,
      "loss": 16.3696,
      "mean_token_accuracy": 0.7775021918118,
      "num_tokens": 15932179.0,
      "step": 560
    },
    {
      "entropy": 0.948325376585126,
      "epoch": 0.912,
      "grad_norm": 14.8125,
      "learning_rate": 1.800740740740741e-05,
      "loss": 15.8592,
      "mean_token_accuracy": 0.7806785147637129,
      "num_tokens": 16219842.0,
      "step": 570
    },
    {
      "entropy": 0.9481973759829998,
      "epoch": 0.928,
      "grad_norm": 14.8125,
      "learning_rate": 1.7933333333333333e-05,
      "loss": 15.8814,
      "mean_token_accuracy": 0.782349206879735,
      "num_tokens": 16499842.0,
      "step": 580
    },
    {
      "entropy": 0.9722379047423602,
      "epoch": 0.944,
      "grad_norm": 15.375,
      "learning_rate": 1.785925925925926e-05,
      "loss": 16.4337,
      "mean_token_accuracy": 0.776747428253293,
      "num_tokens": 16781882.0,
      "step": 590
    },
    {
      "entropy": 0.9909376438707114,
      "epoch": 0.96,
      "grad_norm": 14.6875,
      "learning_rate": 1.7785185185185186e-05,
      "loss": 16.4215,
      "mean_token_accuracy": 0.7778498761355876,
      "num_tokens": 17067407.0,
      "step": 600
    },
    {
      "epoch": 0.96,
      "eval_biology_entropy": 3.4036583194732666,
      "eval_biology_loss": 3.9303975105285645,
      "eval_biology_mean_token_accuracy": 0.4453940885066986,
      "eval_biology_num_tokens": 17067407.0,
      "eval_biology_runtime": 58.2933,
      "eval_biology_samples_per_second": 8.577,
      "eval_biology_steps_per_second": 2.144,
      "step": 600
    },
    {
      "epoch": 0.96,
      "eval_chemistry_entropy": 2.0066872639656066,
      "eval_chemistry_loss": 2.172046422958374,
      "eval_chemistry_mean_token_accuracy": 0.635353732585907,
      "eval_chemistry_num_tokens": 17067407.0,
      "eval_chemistry_runtime": 72.0916,
      "eval_chemistry_samples_per_second": 6.936,
      "eval_chemistry_steps_per_second": 1.734,
      "step": 600
    },
    {
      "epoch": 0.96,
      "eval_math_entropy": 0.9714911403656006,
      "eval_math_loss": 1.020026445388794,
      "eval_math_mean_token_accuracy": 0.7759292826652527,
      "eval_math_num_tokens": 17067407.0,
      "eval_math_runtime": 75.4884,
      "eval_math_samples_per_second": 6.624,
      "eval_math_steps_per_second": 1.656,
      "step": 600
    },
    {
      "epoch": 0.96,
      "eval_physics_entropy": 1.613628080844879,
      "eval_physics_loss": 1.7234340906143188,
      "eval_physics_mean_token_accuracy": 0.6920417103767394,
      "eval_physics_num_tokens": 17067407.0,
      "eval_physics_runtime": 85.693,
      "eval_physics_samples_per_second": 5.835,
      "eval_physics_steps_per_second": 1.459,
      "step": 600
    },
    {
      "entropy": 0.9597863841801881,
      "epoch": 0.976,
      "grad_norm": 20.375,
      "learning_rate": 1.7711111111111113e-05,
      "loss": 16.0384,
      "mean_token_accuracy": 0.7816183552145958,
      "num_tokens": 17350994.0,
      "step": 610
    },
    {
      "entropy": 0.973047361895442,
      "epoch": 0.992,
      "grad_norm": 13.9375,
      "learning_rate": 1.763703703703704e-05,
      "loss": 16.3168,
      "mean_token_accuracy": 0.7756219923496246,
      "num_tokens": 17637514.0,
      "step": 620
    },
    {
      "entropy": 0.9593736134469509,
      "epoch": 1.008,
      "grad_norm": 12.75,
      "learning_rate": 1.7562962962962962e-05,
      "loss": 15.9379,
      "mean_token_accuracy": 0.7799578707665205,
      "num_tokens": 17926570.0,
      "step": 630
    },
    {
      "entropy": 0.9494194338098169,
      "epoch": 1.024,
      "grad_norm": 14.125,
      "learning_rate": 1.7488888888888892e-05,
      "loss": 15.8718,
      "mean_token_accuracy": 0.7815670046955348,
      "num_tokens": 18207652.0,
      "step": 640
    },
    {
      "entropy": 0.9487265404313803,
      "epoch": 1.04,
      "grad_norm": 16.125,
      "learning_rate": 1.7414814814814815e-05,
      "loss": 15.8653,
      "mean_token_accuracy": 0.783008698746562,
      "num_tokens": 18484931.0,
      "step": 650
    },
    {
      "entropy": 0.9464438438415528,
      "epoch": 1.056,
      "grad_norm": 19.875,
      "learning_rate": 1.7340740740740742e-05,
      "loss": 15.9465,
      "mean_token_accuracy": 0.7813190460205078,
      "num_tokens": 18773457.0,
      "step": 660
    },
    {
      "entropy": 0.9594576228410006,
      "epoch": 1.072,
      "grad_norm": 15.0625,
      "learning_rate": 1.726666666666667e-05,
      "loss": 15.9518,
      "mean_token_accuracy": 0.779586423188448,
      "num_tokens": 19055365.0,
      "step": 670
    },
    {
      "entropy": 0.9468420147895813,
      "epoch": 1.088,
      "grad_norm": 12.8125,
      "learning_rate": 1.7192592592592595e-05,
      "loss": 15.7798,
      "mean_token_accuracy": 0.782842144370079,
      "num_tokens": 19345730.0,
      "step": 680
    },
    {
      "entropy": 0.9552154852077365,
      "epoch": 1.104,
      "grad_norm": 12.0625,
      "learning_rate": 1.711851851851852e-05,
      "loss": 15.8369,
      "mean_token_accuracy": 0.7829934328794479,
      "num_tokens": 19637390.0,
      "step": 690
    },
    {
      "entropy": 0.9152646586298943,
      "epoch": 1.12,
      "grad_norm": 16.125,
      "learning_rate": 1.7044444444444445e-05,
      "loss": 15.427,
      "mean_token_accuracy": 0.7873255901038647,
      "num_tokens": 19923914.0,
      "step": 700
    },
    {
      "epoch": 1.12,
      "eval_biology_entropy": 3.324676399230957,
      "eval_biology_loss": 3.911534070968628,
      "eval_biology_mean_token_accuracy": 0.44601045060157773,
      "eval_biology_num_tokens": 19923914.0,
      "eval_biology_runtime": 58.5206,
      "eval_biology_samples_per_second": 8.544,
      "eval_biology_steps_per_second": 2.136,
      "step": 700
    },
    {
      "epoch": 1.12,
      "eval_chemistry_entropy": 1.9495020627975463,
      "eval_chemistry_loss": 2.1536266803741455,
      "eval_chemistry_mean_token_accuracy": 0.6368214721679688,
      "eval_chemistry_num_tokens": 19923914.0,
      "eval_chemistry_runtime": 72.1357,
      "eval_chemistry_samples_per_second": 6.931,
      "eval_chemistry_steps_per_second": 1.733,
      "step": 700
    },
    {
      "epoch": 1.12,
      "eval_math_entropy": 0.9499740815162658,
      "eval_math_loss": 1.0100210905075073,
      "eval_math_mean_token_accuracy": 0.7775792751312256,
      "eval_math_num_tokens": 19923914.0,
      "eval_math_runtime": 75.4806,
      "eval_math_samples_per_second": 6.624,
      "eval_math_steps_per_second": 1.656,
      "step": 700
    },
    {
      "epoch": 1.12,
      "eval_physics_entropy": 1.5690358576774597,
      "eval_physics_loss": 1.7045847177505493,
      "eval_physics_mean_token_accuracy": 0.6933669624328613,
      "eval_physics_num_tokens": 19923914.0,
      "eval_physics_runtime": 85.7866,
      "eval_physics_samples_per_second": 5.828,
      "eval_physics_steps_per_second": 1.457,
      "step": 700
    },
    {
      "entropy": 0.9291105344891548,
      "epoch": 1.1360000000000001,
      "grad_norm": 14.4375,
      "learning_rate": 1.697037037037037e-05,
      "loss": 15.4986,
      "mean_token_accuracy": 0.7853703975677491,
      "num_tokens": 20201892.0,
      "step": 710
    },
    {
      "entropy": 0.9415947886183857,
      "epoch": 1.152,
      "grad_norm": 17.375,
      "learning_rate": 1.6896296296296298e-05,
      "loss": 15.7175,
      "mean_token_accuracy": 0.7841294795274735,
      "num_tokens": 20490282.0,
      "step": 720
    },
    {
      "entropy": 0.9340157110244036,
      "epoch": 1.168,
      "grad_norm": 13.625,
      "learning_rate": 1.6822222222222224e-05,
      "loss": 15.6449,
      "mean_token_accuracy": 0.7852166913449764,
      "num_tokens": 20785786.0,
      "step": 730
    },
    {
      "entropy": 0.9397943781688809,
      "epoch": 1.184,
      "grad_norm": 15.75,
      "learning_rate": 1.6748148148148147e-05,
      "loss": 15.6534,
      "mean_token_accuracy": 0.7839406657963991,
      "num_tokens": 21074205.0,
      "step": 740
    },
    {
      "entropy": 0.9129594217985868,
      "epoch": 1.2,
      "grad_norm": 17.375,
      "learning_rate": 1.6674074074074077e-05,
      "loss": 15.3952,
      "mean_token_accuracy": 0.7864840351045131,
      "num_tokens": 21369159.0,
      "step": 750
    },
    {
      "entropy": 0.9617549370974302,
      "epoch": 1.216,
      "grad_norm": 16.0,
      "learning_rate": 1.66e-05,
      "loss": 15.9238,
      "mean_token_accuracy": 0.7814570065587759,
      "num_tokens": 21649178.0,
      "step": 760
    },
    {
      "entropy": 0.916939202696085,
      "epoch": 1.232,
      "grad_norm": 15.5,
      "learning_rate": 1.6525925925925927e-05,
      "loss": 15.3019,
      "mean_token_accuracy": 0.788709907606244,
      "num_tokens": 21930239.0,
      "step": 770
    },
    {
      "entropy": 0.9114697994664311,
      "epoch": 1.248,
      "grad_norm": 14.6875,
      "learning_rate": 1.6451851851851853e-05,
      "loss": 15.1832,
      "mean_token_accuracy": 0.7884270831942558,
      "num_tokens": 22216387.0,
      "step": 780
    },
    {
      "entropy": 0.9520778905600309,
      "epoch": 1.264,
      "grad_norm": 14.9375,
      "learning_rate": 1.637777777777778e-05,
      "loss": 15.9046,
      "mean_token_accuracy": 0.7815822076052428,
      "num_tokens": 22501002.0,
      "step": 790
    },
    {
      "entropy": 0.9420083146542311,
      "epoch": 1.28,
      "grad_norm": 20.375,
      "learning_rate": 1.6303703703703706e-05,
      "loss": 15.8436,
      "mean_token_accuracy": 0.7832419756799937,
      "num_tokens": 22779682.0,
      "step": 800
    },
    {
      "epoch": 1.28,
      "eval_biology_entropy": 3.384992992401123,
      "eval_biology_loss": 3.905174493789673,
      "eval_biology_mean_token_accuracy": 0.447558299779892,
      "eval_biology_num_tokens": 22779682.0,
      "eval_biology_runtime": 58.5193,
      "eval_biology_samples_per_second": 8.544,
      "eval_biology_steps_per_second": 2.136,
      "step": 800
    },
    {
      "epoch": 1.28,
      "eval_chemistry_entropy": 1.9894180879592895,
      "eval_chemistry_loss": 2.1436502933502197,
      "eval_chemistry_mean_token_accuracy": 0.6379990315437317,
      "eval_chemistry_num_tokens": 22779682.0,
      "eval_chemistry_runtime": 72.08,
      "eval_chemistry_samples_per_second": 6.937,
      "eval_chemistry_steps_per_second": 1.734,
      "step": 800
    },
    {
      "epoch": 1.28,
      "eval_math_entropy": 0.9536066613197327,
      "eval_math_loss": 1.0001386404037476,
      "eval_math_mean_token_accuracy": 0.7793465681076049,
      "eval_math_num_tokens": 22779682.0,
      "eval_math_runtime": 75.2623,
      "eval_math_samples_per_second": 6.643,
      "eval_math_steps_per_second": 1.661,
      "step": 800
    },
    {
      "epoch": 1.28,
      "eval_physics_entropy": 1.6050884528160094,
      "eval_physics_loss": 1.695229172706604,
      "eval_physics_mean_token_accuracy": 0.6948762707710266,
      "eval_physics_num_tokens": 22779682.0,
      "eval_physics_runtime": 85.7894,
      "eval_physics_samples_per_second": 5.828,
      "eval_physics_steps_per_second": 1.457,
      "step": 800
    },
    {
      "entropy": 0.9270045252516865,
      "epoch": 1.296,
      "grad_norm": 15.0625,
      "learning_rate": 1.622962962962963e-05,
      "loss": 15.5761,
      "mean_token_accuracy": 0.7847258739173413,
      "num_tokens": 23057744.0,
      "step": 810
    },
    {
      "entropy": 0.9379040151834488,
      "epoch": 1.312,
      "grad_norm": 16.0,
      "learning_rate": 1.6155555555555556e-05,
      "loss": 15.6785,
      "mean_token_accuracy": 0.7841929640620947,
      "num_tokens": 23344644.0,
      "step": 820
    },
    {
      "entropy": 0.9579077050089836,
      "epoch": 1.328,
      "grad_norm": 18.75,
      "learning_rate": 1.6081481481481482e-05,
      "loss": 15.9331,
      "mean_token_accuracy": 0.7809579152613878,
      "num_tokens": 23622405.0,
      "step": 830
    },
    {
      "entropy": 0.9214272113516927,
      "epoch": 1.3439999999999999,
      "grad_norm": 14.3125,
      "learning_rate": 1.600740740740741e-05,
      "loss": 15.4675,
      "mean_token_accuracy": 0.7852523807436228,
      "num_tokens": 23899771.0,
      "step": 840
    },
    {
      "entropy": 0.9227435002103448,
      "epoch": 1.3599999999999999,
      "grad_norm": 16.5,
      "learning_rate": 1.5933333333333336e-05,
      "loss": 15.4552,
      "mean_token_accuracy": 0.7862110409885645,
      "num_tokens": 24187023.0,
      "step": 850
    },
    {
      "entropy": 0.9958576496690512,
      "epoch": 1.376,
      "grad_norm": 18.75,
      "learning_rate": 1.5859259259259262e-05,
      "loss": 16.5571,
      "mean_token_accuracy": 0.7727442421019077,
      "num_tokens": 24466132.0,
      "step": 860
    },
    {
      "entropy": 0.9378596719354391,
      "epoch": 1.392,
      "grad_norm": 18.875,
      "learning_rate": 1.5785185185185185e-05,
      "loss": 15.5497,
      "mean_token_accuracy": 0.783870642259717,
      "num_tokens": 24748043.0,
      "step": 870
    },
    {
      "entropy": 0.902330132946372,
      "epoch": 1.408,
      "grad_norm": 17.0,
      "learning_rate": 1.571111111111111e-05,
      "loss": 15.2254,
      "mean_token_accuracy": 0.7886204585433007,
      "num_tokens": 25036674.0,
      "step": 880
    },
    {
      "entropy": 0.9116618972271681,
      "epoch": 1.424,
      "grad_norm": 12.125,
      "learning_rate": 1.5637037037037038e-05,
      "loss": 15.1439,
      "mean_token_accuracy": 0.789288330078125,
      "num_tokens": 25324579.0,
      "step": 890
    },
    {
      "entropy": 0.9361150417476892,
      "epoch": 1.44,
      "grad_norm": 15.3125,
      "learning_rate": 1.5562962962962965e-05,
      "loss": 15.6872,
      "mean_token_accuracy": 0.7842213969677687,
      "num_tokens": 25606824.0,
      "step": 900
    },
    {
      "epoch": 1.44,
      "eval_biology_entropy": 3.413121500015259,
      "eval_biology_loss": 3.890507221221924,
      "eval_biology_mean_token_accuracy": 0.44829230070114134,
      "eval_biology_num_tokens": 25606824.0,
      "eval_biology_runtime": 58.6058,
      "eval_biology_samples_per_second": 8.532,
      "eval_biology_steps_per_second": 2.133,
      "step": 900
    },
    {
      "epoch": 1.44,
      "eval_chemistry_entropy": 1.990509286403656,
      "eval_chemistry_loss": 2.1352083683013916,
      "eval_chemistry_mean_token_accuracy": 0.6389702727794647,
      "eval_chemistry_num_tokens": 25606824.0,
      "eval_chemistry_runtime": 72.5787,
      "eval_chemistry_samples_per_second": 6.889,
      "eval_chemistry_steps_per_second": 1.722,
      "step": 900
    },
    {
      "epoch": 1.44,
      "eval_math_entropy": 0.9449980363845826,
      "eval_math_loss": 0.9944726228713989,
      "eval_math_mean_token_accuracy": 0.7802386074066162,
      "eval_math_num_tokens": 25606824.0,
      "eval_math_runtime": 75.5905,
      "eval_math_samples_per_second": 6.615,
      "eval_math_steps_per_second": 1.654,
      "step": 900
    },
    {
      "epoch": 1.44,
      "eval_physics_entropy": 1.601661796092987,
      "eval_physics_loss": 1.6853703260421753,
      "eval_physics_mean_token_accuracy": 0.6957120170593262,
      "eval_physics_num_tokens": 25606824.0,
      "eval_physics_runtime": 85.8807,
      "eval_physics_samples_per_second": 5.822,
      "eval_physics_steps_per_second": 1.456,
      "step": 900
    },
    {
      "entropy": 0.9202836632728577,
      "epoch": 1.456,
      "grad_norm": 13.5625,
      "learning_rate": 1.548888888888889e-05,
      "loss": 15.2958,
      "mean_token_accuracy": 0.7856416959315539,
      "num_tokens": 25886396.0,
      "step": 910
    },
    {
      "entropy": 0.9233658915385604,
      "epoch": 1.472,
      "grad_norm": 13.25,
      "learning_rate": 1.5414814814814814e-05,
      "loss": 15.4872,
      "mean_token_accuracy": 0.784604623541236,
      "num_tokens": 26163618.0,
      "step": 920
    },
    {
      "entropy": 0.9581601906567812,
      "epoch": 1.488,
      "grad_norm": 13.8125,
      "learning_rate": 1.5340740740740744e-05,
      "loss": 16.0305,
      "mean_token_accuracy": 0.7800706576555967,
      "num_tokens": 26438338.0,
      "step": 930
    },
    {
      "entropy": 0.9096721589565278,
      "epoch": 1.504,
      "grad_norm": 17.375,
      "learning_rate": 1.5266666666666667e-05,
      "loss": 15.1674,
      "mean_token_accuracy": 0.7894522257149219,
      "num_tokens": 26729255.0,
      "step": 940
    },
    {
      "entropy": 0.9369884114712477,
      "epoch": 1.52,
      "grad_norm": 13.5625,
      "learning_rate": 1.5192592592592594e-05,
      "loss": 15.7623,
      "mean_token_accuracy": 0.783218951150775,
      "num_tokens": 27017935.0,
      "step": 950
    },
    {
      "entropy": 0.9323047822341323,
      "epoch": 1.536,
      "grad_norm": 13.75,
      "learning_rate": 1.5118518518518519e-05,
      "loss": 15.4618,
      "mean_token_accuracy": 0.7854165881872177,
      "num_tokens": 27306339.0,
      "step": 960
    },
    {
      "entropy": 0.9303993556648493,
      "epoch": 1.552,
      "grad_norm": 16.0,
      "learning_rate": 1.5044444444444445e-05,
      "loss": 15.4762,
      "mean_token_accuracy": 0.7855500593781471,
      "num_tokens": 27591959.0,
      "step": 970
    },
    {
      "entropy": 0.8964169017970562,
      "epoch": 1.568,
      "grad_norm": 12.9375,
      "learning_rate": 1.497037037037037e-05,
      "loss": 15.0334,
      "mean_token_accuracy": 0.7918794117867947,
      "num_tokens": 27884398.0,
      "step": 980
    },
    {
      "entropy": 0.9186629202216864,
      "epoch": 1.584,
      "grad_norm": 16.375,
      "learning_rate": 1.4896296296296298e-05,
      "loss": 15.3975,
      "mean_token_accuracy": 0.7869536556303501,
      "num_tokens": 28171274.0,
      "step": 990
    },
    {
      "entropy": 0.9073497140780091,
      "epoch": 1.6,
      "grad_norm": 18.5,
      "learning_rate": 1.4822222222222225e-05,
      "loss": 15.2016,
      "mean_token_accuracy": 0.7862915009260177,
      "num_tokens": 28457624.0,
      "step": 1000
    },
    {
      "epoch": 1.6,
      "eval_biology_entropy": 3.321074670791626,
      "eval_biology_loss": 3.887943983078003,
      "eval_biology_mean_token_accuracy": 0.44904091477394104,
      "eval_biology_num_tokens": 28457624.0,
      "eval_biology_runtime": 58.6475,
      "eval_biology_samples_per_second": 8.526,
      "eval_biology_steps_per_second": 2.131,
      "step": 1000
    },
    {
      "epoch": 1.6,
      "eval_chemistry_entropy": 1.93967240858078,
      "eval_chemistry_loss": 2.1321117877960205,
      "eval_chemistry_mean_token_accuracy": 0.6390365164279938,
      "eval_chemistry_num_tokens": 28457624.0,
      "eval_chemistry_runtime": 71.9495,
      "eval_chemistry_samples_per_second": 6.949,
      "eval_chemistry_steps_per_second": 1.737,
      "step": 1000
    },
    {
      "epoch": 1.6,
      "eval_math_entropy": 0.9376819987297058,
      "eval_math_loss": 0.9886224269866943,
      "eval_math_mean_token_accuracy": 0.7808662033081055,
      "eval_math_num_tokens": 28457624.0,
      "eval_math_runtime": 74.8917,
      "eval_math_samples_per_second": 6.676,
      "eval_math_steps_per_second": 1.669,
      "step": 1000
    },
    {
      "epoch": 1.6,
      "eval_physics_entropy": 1.566874056339264,
      "eval_physics_loss": 1.6789716482162476,
      "eval_physics_mean_token_accuracy": 0.696774085521698,
      "eval_physics_num_tokens": 28457624.0,
      "eval_physics_runtime": 85.493,
      "eval_physics_samples_per_second": 5.848,
      "eval_physics_steps_per_second": 1.462,
      "step": 1000
    },
    {
      "entropy": 0.9235946271568537,
      "epoch": 1.616,
      "grad_norm": 14.4375,
      "learning_rate": 1.474814814814815e-05,
      "loss": 15.3986,
      "mean_token_accuracy": 0.786095867305994,
      "num_tokens": 28743099.0,
      "step": 1010
    },
    {
      "entropy": 0.9303552206605673,
      "epoch": 1.6320000000000001,
      "grad_norm": 15.5625,
      "learning_rate": 1.4674074074074076e-05,
      "loss": 15.5645,
      "mean_token_accuracy": 0.7829754035919905,
      "num_tokens": 29017297.0,
      "step": 1020
    },
    {
      "entropy": 0.9154788227751851,
      "epoch": 1.6480000000000001,
      "grad_norm": 15.4375,
      "learning_rate": 1.46e-05,
      "loss": 15.2985,
      "mean_token_accuracy": 0.7891919452697038,
      "num_tokens": 29303707.0,
      "step": 1030
    },
    {
      "entropy": 0.9193036049604416,
      "epoch": 1.6640000000000001,
      "grad_norm": 15.875,
      "learning_rate": 1.4525925925925927e-05,
      "loss": 15.235,
      "mean_token_accuracy": 0.7860149636864662,
      "num_tokens": 29597156.0,
      "step": 1040
    },
    {
      "entropy": 0.9248684361577034,
      "epoch": 1.6800000000000002,
      "grad_norm": 13.5,
      "learning_rate": 1.4451851851851852e-05,
      "loss": 15.5595,
      "mean_token_accuracy": 0.7858113132417202,
      "num_tokens": 29883879.0,
      "step": 1050
    },
    {
      "entropy": 0.9327945580706001,
      "epoch": 1.696,
      "grad_norm": 14.1875,
      "learning_rate": 1.4377777777777779e-05,
      "loss": 15.5428,
      "mean_token_accuracy": 0.7853838302195072,
      "num_tokens": 30165760.0,
      "step": 1060
    },
    {
      "entropy": 0.9061191271990537,
      "epoch": 1.712,
      "grad_norm": 15.5625,
      "learning_rate": 1.4303703703703703e-05,
      "loss": 15.1161,
      "mean_token_accuracy": 0.7896213702857494,
      "num_tokens": 30460367.0,
      "step": 1070
    },
    {
      "entropy": 0.9054378770291805,
      "epoch": 1.728,
      "grad_norm": 19.625,
      "learning_rate": 1.4229629629629632e-05,
      "loss": 15.098,
      "mean_token_accuracy": 0.7892790406942367,
      "num_tokens": 30739137.0,
      "step": 1080
    },
    {
      "entropy": 0.9035517876967788,
      "epoch": 1.744,
      "grad_norm": 16.25,
      "learning_rate": 1.4155555555555556e-05,
      "loss": 15.2616,
      "mean_token_accuracy": 0.7869241286069155,
      "num_tokens": 31022663.0,
      "step": 1090
    },
    {
      "entropy": 0.9091449549421668,
      "epoch": 1.76,
      "grad_norm": 13.75,
      "learning_rate": 1.4081481481481483e-05,
      "loss": 15.1422,
      "mean_token_accuracy": 0.7880128789693117,
      "num_tokens": 31306494.0,
      "step": 1100
    },
    {
      "epoch": 1.76,
      "eval_biology_entropy": 3.356640110015869,
      "eval_biology_loss": 3.8688619136810303,
      "eval_biology_mean_token_accuracy": 0.45002903842926023,
      "eval_biology_num_tokens": 31306494.0,
      "eval_biology_runtime": 58.9193,
      "eval_biology_samples_per_second": 8.486,
      "eval_biology_steps_per_second": 2.122,
      "step": 1100
    },
    {
      "epoch": 1.76,
      "eval_chemistry_entropy": 1.9536669540405274,
      "eval_chemistry_loss": 2.1165969371795654,
      "eval_chemistry_mean_token_accuracy": 0.6407591633796692,
      "eval_chemistry_num_tokens": 31306494.0,
      "eval_chemistry_runtime": 71.8391,
      "eval_chemistry_samples_per_second": 6.96,
      "eval_chemistry_steps_per_second": 1.74,
      "step": 1100
    },
    {
      "epoch": 1.76,
      "eval_math_entropy": 0.9424327955245971,
      "eval_math_loss": 0.9843871593475342,
      "eval_math_mean_token_accuracy": 0.7814709210395813,
      "eval_math_num_tokens": 31306494.0,
      "eval_math_runtime": 74.8286,
      "eval_math_samples_per_second": 6.682,
      "eval_math_steps_per_second": 1.67,
      "step": 1100
    },
    {
      "epoch": 1.76,
      "eval_physics_entropy": 1.5812338199615479,
      "eval_physics_loss": 1.6700222492218018,
      "eval_physics_mean_token_accuracy": 0.697830177783966,
      "eval_physics_num_tokens": 31306494.0,
      "eval_physics_runtime": 85.7072,
      "eval_physics_samples_per_second": 5.834,
      "eval_physics_steps_per_second": 1.458,
      "step": 1100
    },
    {
      "entropy": 0.924960282817483,
      "epoch": 1.776,
      "grad_norm": 16.5,
      "learning_rate": 1.400740740740741e-05,
      "loss": 15.3105,
      "mean_token_accuracy": 0.7877573404461146,
      "num_tokens": 31595542.0,
      "step": 1110
    },
    {
      "entropy": 0.9376434125006199,
      "epoch": 1.792,
      "grad_norm": 14.9375,
      "learning_rate": 1.3933333333333334e-05,
      "loss": 15.591,
      "mean_token_accuracy": 0.7839235190302134,
      "num_tokens": 31881189.0,
      "step": 1120
    },
    {
      "entropy": 0.9261387910693883,
      "epoch": 1.808,
      "grad_norm": 14.5,
      "learning_rate": 1.385925925925926e-05,
      "loss": 15.4967,
      "mean_token_accuracy": 0.7858365088701248,
      "num_tokens": 32164196.0,
      "step": 1130
    },
    {
      "entropy": 0.881668720766902,
      "epoch": 1.8239999999999998,
      "grad_norm": 13.3125,
      "learning_rate": 1.3785185185185186e-05,
      "loss": 14.7294,
      "mean_token_accuracy": 0.793937573581934,
      "num_tokens": 32441530.0,
      "step": 1140
    },
    {
      "entropy": 0.9093685988336802,
      "epoch": 1.8399999999999999,
      "grad_norm": 15.1875,
      "learning_rate": 1.3711111111111112e-05,
      "loss": 15.1805,
      "mean_token_accuracy": 0.7902645103633403,
      "num_tokens": 32723145.0,
      "step": 1150
    },
    {
      "entropy": 0.9354806829243898,
      "epoch": 1.8559999999999999,
      "grad_norm": 16.0,
      "learning_rate": 1.3637037037037037e-05,
      "loss": 15.5881,
      "mean_token_accuracy": 0.7842747800052166,
      "num_tokens": 33011263.0,
      "step": 1160
    },
    {
      "entropy": 0.9051053501665592,
      "epoch": 1.8719999999999999,
      "grad_norm": 14.1875,
      "learning_rate": 1.3562962962962965e-05,
      "loss": 15.1055,
      "mean_token_accuracy": 0.7910506792366505,
      "num_tokens": 33298921.0,
      "step": 1170
    },
    {
      "entropy": 0.9241808012127877,
      "epoch": 1.888,
      "grad_norm": 17.25,
      "learning_rate": 1.3488888888888888e-05,
      "loss": 15.4978,
      "mean_token_accuracy": 0.7850798547267914,
      "num_tokens": 33576243.0,
      "step": 1180
    },
    {
      "entropy": 0.9191399356350303,
      "epoch": 1.904,
      "grad_norm": 13.5,
      "learning_rate": 1.3414814814814817e-05,
      "loss": 15.3401,
      "mean_token_accuracy": 0.7869909830391407,
      "num_tokens": 33850968.0,
      "step": 1190
    },
    {
      "entropy": 0.8988351378589868,
      "epoch": 1.92,
      "grad_norm": 14.3125,
      "learning_rate": 1.3340740740740741e-05,
      "loss": 15.0702,
      "mean_token_accuracy": 0.7889779832214117,
      "num_tokens": 34128558.0,
      "step": 1200
    },
    {
      "epoch": 1.92,
      "eval_biology_entropy": 3.342221477508545,
      "eval_biology_loss": 3.8490023612976074,
      "eval_biology_mean_token_accuracy": 0.45095433259010315,
      "eval_biology_num_tokens": 34128558.0,
      "eval_biology_runtime": 58.5045,
      "eval_biology_samples_per_second": 8.546,
      "eval_biology_steps_per_second": 2.137,
      "step": 1200
    },
    {
      "epoch": 1.92,
      "eval_chemistry_entropy": 1.9384882287979126,
      "eval_chemistry_loss": 2.1126294136047363,
      "eval_chemistry_mean_token_accuracy": 0.6409776957035065,
      "eval_chemistry_num_tokens": 34128558.0,
      "eval_chemistry_runtime": 72.4461,
      "eval_chemistry_samples_per_second": 6.902,
      "eval_chemistry_steps_per_second": 1.725,
      "step": 1200
    },
    {
      "epoch": 1.92,
      "eval_math_entropy": 0.9290759258270264,
      "eval_math_loss": 0.979581892490387,
      "eval_math_mean_token_accuracy": 0.7822537088394165,
      "eval_math_num_tokens": 34128558.0,
      "eval_math_runtime": 75.8783,
      "eval_math_samples_per_second": 6.589,
      "eval_math_steps_per_second": 1.647,
      "step": 1200
    },
    {
      "epoch": 1.92,
      "eval_physics_entropy": 1.5694564418792725,
      "eval_physics_loss": 1.672090768814087,
      "eval_physics_mean_token_accuracy": 0.6977639694213867,
      "eval_physics_num_tokens": 34128558.0,
      "eval_physics_runtime": 85.8572,
      "eval_physics_samples_per_second": 5.824,
      "eval_physics_steps_per_second": 1.456,
      "step": 1200
    },
    {
      "entropy": 0.9215047180652618,
      "epoch": 1.936,
      "grad_norm": 18.0,
      "learning_rate": 1.3266666666666668e-05,
      "loss": 15.3559,
      "mean_token_accuracy": 0.7858322024345398,
      "num_tokens": 34408056.0,
      "step": 1210
    },
    {
      "entropy": 0.9102119285613298,
      "epoch": 1.952,
      "grad_norm": 15.5,
      "learning_rate": 1.3192592592592594e-05,
      "loss": 15.183,
      "mean_token_accuracy": 0.7886962842196226,
      "num_tokens": 34684679.0,
      "step": 1220
    },
    {
      "entropy": 0.9319753609597683,
      "epoch": 1.968,
      "grad_norm": 16.25,
      "learning_rate": 1.311851851851852e-05,
      "loss": 15.6225,
      "mean_token_accuracy": 0.7829237889498473,
      "num_tokens": 34971038.0,
      "step": 1230
    },
    {
      "entropy": 0.9107826724648476,
      "epoch": 1.984,
      "grad_norm": 15.3125,
      "learning_rate": 1.3044444444444446e-05,
      "loss": 15.0632,
      "mean_token_accuracy": 0.7896871905773878,
      "num_tokens": 35262281.0,
      "step": 1240
    },
    {
      "entropy": 0.8834621507674456,
      "epoch": 2.0,
      "grad_norm": 15.8125,
      "learning_rate": 1.297037037037037e-05,
      "loss": 14.8785,
      "mean_token_accuracy": 0.7918516155332327,
      "num_tokens": 35560864.0,
      "step": 1250
    },
    {
      "entropy": 0.9305843397974968,
      "epoch": 2.016,
      "grad_norm": 13.0,
      "learning_rate": 1.2896296296296299e-05,
      "loss": 15.4293,
      "mean_token_accuracy": 0.786306445300579,
      "num_tokens": 35846704.0,
      "step": 1260
    },
    {
      "entropy": 0.9240877356380224,
      "epoch": 2.032,
      "grad_norm": 18.375,
      "learning_rate": 1.2822222222222222e-05,
      "loss": 15.3222,
      "mean_token_accuracy": 0.7866721350699664,
      "num_tokens": 36128775.0,
      "step": 1270
    },
    {
      "entropy": 0.88644311837852,
      "epoch": 2.048,
      "grad_norm": 14.9375,
      "learning_rate": 1.274814814814815e-05,
      "loss": 14.7864,
      "mean_token_accuracy": 0.7933717489242553,
      "num_tokens": 36419504.0,
      "step": 1280
    },
    {
      "entropy": 0.8904413890093565,
      "epoch": 2.064,
      "grad_norm": 13.8125,
      "learning_rate": 1.2674074074074075e-05,
      "loss": 14.9247,
      "mean_token_accuracy": 0.7927331611514091,
      "num_tokens": 36706816.0,
      "step": 1290
    },
    {
      "entropy": 0.9007925825193525,
      "epoch": 2.08,
      "grad_norm": 15.25,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 14.9352,
      "mean_token_accuracy": 0.7917296580970288,
      "num_tokens": 36988475.0,
      "step": 1300
    },
    {
      "epoch": 2.08,
      "eval_biology_entropy": 3.2783792409896853,
      "eval_biology_loss": 3.845946788787842,
      "eval_biology_mean_token_accuracy": 0.4511831784248352,
      "eval_biology_num_tokens": 36988475.0,
      "eval_biology_runtime": 58.3894,
      "eval_biology_samples_per_second": 8.563,
      "eval_biology_steps_per_second": 2.141,
      "step": 1300
    },
    {
      "epoch": 2.08,
      "eval_chemistry_entropy": 1.9022014627456665,
      "eval_chemistry_loss": 2.1077961921691895,
      "eval_chemistry_mean_token_accuracy": 0.641357579946518,
      "eval_chemistry_num_tokens": 36988475.0,
      "eval_chemistry_runtime": 71.9668,
      "eval_chemistry_samples_per_second": 6.948,
      "eval_chemistry_steps_per_second": 1.737,
      "step": 1300
    },
    {
      "epoch": 2.08,
      "eval_math_entropy": 0.9213895635604858,
      "eval_math_loss": 0.9778892993927002,
      "eval_math_mean_token_accuracy": 0.7825007963180542,
      "eval_math_num_tokens": 36988475.0,
      "eval_math_runtime": 75.2485,
      "eval_math_samples_per_second": 6.645,
      "eval_math_steps_per_second": 1.661,
      "step": 1300
    },
    {
      "epoch": 2.08,
      "eval_physics_entropy": 1.5423394293785095,
      "eval_physics_loss": 1.6658241748809814,
      "eval_physics_mean_token_accuracy": 0.6982630844116211,
      "eval_physics_num_tokens": 36988475.0,
      "eval_physics_runtime": 85.7778,
      "eval_physics_samples_per_second": 5.829,
      "eval_physics_steps_per_second": 1.457,
      "step": 1300
    },
    {
      "entropy": 0.9216024700552226,
      "epoch": 2.096,
      "grad_norm": 15.375,
      "learning_rate": 1.2525925925925928e-05,
      "loss": 15.3605,
      "mean_token_accuracy": 0.7878219708800316,
      "num_tokens": 37270131.0,
      "step": 1310
    },
    {
      "entropy": 0.8836073707789183,
      "epoch": 2.112,
      "grad_norm": 13.875,
      "learning_rate": 1.2451851851851853e-05,
      "loss": 14.7465,
      "mean_token_accuracy": 0.7925711952149868,
      "num_tokens": 37563537.0,
      "step": 1320
    },
    {
      "entropy": 0.8804319033399224,
      "epoch": 2.128,
      "grad_norm": 15.5625,
      "learning_rate": 1.237777777777778e-05,
      "loss": 14.7508,
      "mean_token_accuracy": 0.7928811598569154,
      "num_tokens": 37843959.0,
      "step": 1330
    },
    {
      "entropy": 0.8922224432229996,
      "epoch": 2.144,
      "grad_norm": 16.125,
      "learning_rate": 1.2303703703703704e-05,
      "loss": 14.9258,
      "mean_token_accuracy": 0.7923558741807938,
      "num_tokens": 38133092.0,
      "step": 1340
    },
    {
      "entropy": 0.8934669015929103,
      "epoch": 2.16,
      "grad_norm": 15.0,
      "learning_rate": 1.222962962962963e-05,
      "loss": 14.9419,
      "mean_token_accuracy": 0.7924261353909969,
      "num_tokens": 38421229.0,
      "step": 1350
    },
    {
      "entropy": 0.8715774483978749,
      "epoch": 2.176,
      "grad_norm": 17.75,
      "learning_rate": 1.2155555555555555e-05,
      "loss": 14.4375,
      "mean_token_accuracy": 0.7976760245859623,
      "num_tokens": 38708043.0,
      "step": 1360
    },
    {
      "entropy": 0.8904842108488082,
      "epoch": 2.192,
      "grad_norm": 14.1875,
      "learning_rate": 1.2081481481481484e-05,
      "loss": 14.9715,
      "mean_token_accuracy": 0.7901181749999523,
      "num_tokens": 38996930.0,
      "step": 1370
    },
    {
      "entropy": 0.9079100303351879,
      "epoch": 2.208,
      "grad_norm": 17.375,
      "learning_rate": 1.2007407407407408e-05,
      "loss": 15.1465,
      "mean_token_accuracy": 0.7897144511342049,
      "num_tokens": 39279481.0,
      "step": 1380
    },
    {
      "entropy": 0.9049843695014715,
      "epoch": 2.224,
      "grad_norm": 16.875,
      "learning_rate": 1.1933333333333335e-05,
      "loss": 15.0122,
      "mean_token_accuracy": 0.7912030033767223,
      "num_tokens": 39569030.0,
      "step": 1390
    },
    {
      "entropy": 0.8973436402156949,
      "epoch": 2.24,
      "grad_norm": 14.9375,
      "learning_rate": 1.185925925925926e-05,
      "loss": 14.8965,
      "mean_token_accuracy": 0.7904580168426036,
      "num_tokens": 39854873.0,
      "step": 1400
    },
    {
      "epoch": 2.24,
      "eval_biology_entropy": 3.335082673072815,
      "eval_biology_loss": 3.848658800125122,
      "eval_biology_mean_token_accuracy": 0.4510736925601959,
      "eval_biology_num_tokens": 39854873.0,
      "eval_biology_runtime": 58.072,
      "eval_biology_samples_per_second": 8.61,
      "eval_biology_steps_per_second": 2.153,
      "step": 1400
    },
    {
      "epoch": 2.24,
      "eval_chemistry_entropy": 1.9277294964790344,
      "eval_chemistry_loss": 2.105435371398926,
      "eval_chemistry_mean_token_accuracy": 0.6416595306396484,
      "eval_chemistry_num_tokens": 39854873.0,
      "eval_chemistry_runtime": 71.9974,
      "eval_chemistry_samples_per_second": 6.945,
      "eval_chemistry_steps_per_second": 1.736,
      "step": 1400
    },
    {
      "epoch": 2.24,
      "eval_math_entropy": 0.923270836353302,
      "eval_math_loss": 0.9751716256141663,
      "eval_math_mean_token_accuracy": 0.7829953188896179,
      "eval_math_num_tokens": 39854873.0,
      "eval_math_runtime": 75.3644,
      "eval_math_samples_per_second": 6.634,
      "eval_math_steps_per_second": 1.659,
      "step": 1400
    },
    {
      "epoch": 2.24,
      "eval_physics_entropy": 1.5597347102165222,
      "eval_physics_loss": 1.6624547243118286,
      "eval_physics_mean_token_accuracy": 0.6985624618530274,
      "eval_physics_num_tokens": 39854873.0,
      "eval_physics_runtime": 85.8375,
      "eval_physics_samples_per_second": 5.825,
      "eval_physics_steps_per_second": 1.456,
      "step": 1400
    },
    {
      "entropy": 0.8929722145199775,
      "epoch": 2.2560000000000002,
      "grad_norm": 14.5,
      "learning_rate": 1.1785185185185186e-05,
      "loss": 14.8853,
      "mean_token_accuracy": 0.7903232526034116,
      "num_tokens": 40141190.0,
      "step": 1410
    },
    {
      "entropy": 0.9134282995015383,
      "epoch": 2.2720000000000002,
      "grad_norm": 17.375,
      "learning_rate": 1.1711111111111113e-05,
      "loss": 15.268,
      "mean_token_accuracy": 0.7882483631372452,
      "num_tokens": 40415203.0,
      "step": 1420
    },
    {
      "entropy": 0.893715601041913,
      "epoch": 2.288,
      "grad_norm": 15.3125,
      "learning_rate": 1.1637037037037037e-05,
      "loss": 15.0269,
      "mean_token_accuracy": 0.7883617911487818,
      "num_tokens": 40702393.0,
      "step": 1430
    },
    {
      "entropy": 0.9125330992043018,
      "epoch": 2.304,
      "grad_norm": 16.5,
      "learning_rate": 1.1562962962962964e-05,
      "loss": 15.0516,
      "mean_token_accuracy": 0.7916629381477833,
      "num_tokens": 40982775.0,
      "step": 1440
    },
    {
      "entropy": 0.9030418084934354,
      "epoch": 2.32,
      "grad_norm": 16.625,
      "learning_rate": 1.1488888888888889e-05,
      "loss": 15.0525,
      "mean_token_accuracy": 0.7892082892358303,
      "num_tokens": 41263356.0,
      "step": 1450
    },
    {
      "entropy": 0.9153900254517794,
      "epoch": 2.336,
      "grad_norm": 17.0,
      "learning_rate": 1.1414814814814817e-05,
      "loss": 15.3005,
      "mean_token_accuracy": 0.787101947888732,
      "num_tokens": 41545235.0,
      "step": 1460
    },
    {
      "entropy": 0.9137625804170966,
      "epoch": 2.352,
      "grad_norm": 15.75,
      "learning_rate": 1.1340740740740742e-05,
      "loss": 15.1855,
      "mean_token_accuracy": 0.7869684521108866,
      "num_tokens": 41833417.0,
      "step": 1470
    },
    {
      "entropy": 0.8969733513891697,
      "epoch": 2.368,
      "grad_norm": 16.875,
      "learning_rate": 1.1266666666666668e-05,
      "loss": 15.0008,
      "mean_token_accuracy": 0.7912497945129872,
      "num_tokens": 42117030.0,
      "step": 1480
    },
    {
      "entropy": 0.9257304556667805,
      "epoch": 2.384,
      "grad_norm": 16.125,
      "learning_rate": 1.1192592592592593e-05,
      "loss": 15.3741,
      "mean_token_accuracy": 0.7869575574994088,
      "num_tokens": 42410990.0,
      "step": 1490
    },
    {
      "entropy": 0.9102526713162661,
      "epoch": 2.4,
      "grad_norm": 26.625,
      "learning_rate": 1.111851851851852e-05,
      "loss": 15.2415,
      "mean_token_accuracy": 0.7886898033320904,
      "num_tokens": 42691890.0,
      "step": 1500
    },
    {
      "epoch": 2.4,
      "eval_biology_entropy": 3.2777514753341674,
      "eval_biology_loss": 3.851931095123291,
      "eval_biology_mean_token_accuracy": 0.4516991958618164,
      "eval_biology_num_tokens": 42691890.0,
      "eval_biology_runtime": 58.0868,
      "eval_biology_samples_per_second": 8.608,
      "eval_biology_steps_per_second": 2.152,
      "step": 1500
    },
    {
      "epoch": 2.4,
      "eval_chemistry_entropy": 1.8954128623008728,
      "eval_chemistry_loss": 2.109870672225952,
      "eval_chemistry_mean_token_accuracy": 0.6416130490303039,
      "eval_chemistry_num_tokens": 42691890.0,
      "eval_chemistry_runtime": 71.6131,
      "eval_chemistry_samples_per_second": 6.982,
      "eval_chemistry_steps_per_second": 1.745,
      "step": 1500
    },
    {
      "epoch": 2.4,
      "eval_math_entropy": 0.9151335005760193,
      "eval_math_loss": 0.9732398390769958,
      "eval_math_mean_token_accuracy": 0.78314342212677,
      "eval_math_num_tokens": 42691890.0,
      "eval_math_runtime": 75.1319,
      "eval_math_samples_per_second": 6.655,
      "eval_math_steps_per_second": 1.664,
      "step": 1500
    },
    {
      "epoch": 2.4,
      "eval_physics_entropy": 1.5351521530151366,
      "eval_physics_loss": 1.664096474647522,
      "eval_physics_mean_token_accuracy": 0.6988328862190246,
      "eval_physics_num_tokens": 42691890.0,
      "eval_physics_runtime": 85.6903,
      "eval_physics_samples_per_second": 5.835,
      "eval_physics_steps_per_second": 1.459,
      "step": 1500
    },
    {
      "entropy": 0.9180217150598764,
      "epoch": 2.416,
      "grad_norm": 16.125,
      "learning_rate": 1.1044444444444444e-05,
      "loss": 15.3747,
      "mean_token_accuracy": 0.7853360865265131,
      "num_tokens": 42971588.0,
      "step": 1510
    },
    {
      "entropy": 0.8968389870598912,
      "epoch": 2.432,
      "grad_norm": 14.75,
      "learning_rate": 1.0970370370370371e-05,
      "loss": 14.9289,
      "mean_token_accuracy": 0.7917619660496712,
      "num_tokens": 43253821.0,
      "step": 1520
    },
    {
      "entropy": 0.8765170518308878,
      "epoch": 2.448,
      "grad_norm": 14.6875,
      "learning_rate": 1.0896296296296298e-05,
      "loss": 14.6694,
      "mean_token_accuracy": 0.795002979785204,
      "num_tokens": 43550902.0,
      "step": 1530
    },
    {
      "entropy": 0.9039445007219911,
      "epoch": 2.464,
      "grad_norm": 15.625,
      "learning_rate": 1.0822222222222222e-05,
      "loss": 14.9612,
      "mean_token_accuracy": 0.7890904504805804,
      "num_tokens": 43844259.0,
      "step": 1540
    },
    {
      "entropy": 0.958719066157937,
      "epoch": 2.48,
      "grad_norm": 20.0,
      "learning_rate": 1.074814814814815e-05,
      "loss": 15.9159,
      "mean_token_accuracy": 0.7804724108427763,
      "num_tokens": 44115701.0,
      "step": 1550
    },
    {
      "entropy": 0.9251472631469368,
      "epoch": 2.496,
      "grad_norm": 14.9375,
      "learning_rate": 1.0674074074074074e-05,
      "loss": 15.5099,
      "mean_token_accuracy": 0.785043653100729,
      "num_tokens": 44405520.0,
      "step": 1560
    },
    {
      "entropy": 0.9019034523516893,
      "epoch": 2.512,
      "grad_norm": 14.4375,
      "learning_rate": 1.0600000000000002e-05,
      "loss": 15.063,
      "mean_token_accuracy": 0.78865717202425,
      "num_tokens": 44686477.0,
      "step": 1570
    },
    {
      "entropy": 0.9118882043287158,
      "epoch": 2.528,
      "grad_norm": 15.25,
      "learning_rate": 1.0525925925925927e-05,
      "loss": 15.1807,
      "mean_token_accuracy": 0.7881377603858709,
      "num_tokens": 44969760.0,
      "step": 1580
    },
    {
      "entropy": 0.9028380002826453,
      "epoch": 2.544,
      "grad_norm": 15.8125,
      "learning_rate": 1.0451851851851853e-05,
      "loss": 15.036,
      "mean_token_accuracy": 0.7901853829622268,
      "num_tokens": 45255326.0,
      "step": 1590
    },
    {
      "entropy": 0.9037538858130574,
      "epoch": 2.56,
      "grad_norm": 14.125,
      "learning_rate": 1.0377777777777778e-05,
      "loss": 14.9376,
      "mean_token_accuracy": 0.7912348363548517,
      "num_tokens": 45532525.0,
      "step": 1600
    },
    {
      "epoch": 2.56,
      "eval_biology_entropy": 3.296840294837952,
      "eval_biology_loss": 3.837435722351074,
      "eval_biology_mean_token_accuracy": 0.4518266432285309,
      "eval_biology_num_tokens": 45532525.0,
      "eval_biology_runtime": 58.1201,
      "eval_biology_samples_per_second": 8.603,
      "eval_biology_steps_per_second": 2.151,
      "step": 1600
    },
    {
      "epoch": 2.56,
      "eval_chemistry_entropy": 1.9017902750968934,
      "eval_chemistry_loss": 2.1050920486450195,
      "eval_chemistry_mean_token_accuracy": 0.6415406594276428,
      "eval_chemistry_num_tokens": 45532525.0,
      "eval_chemistry_runtime": 71.7598,
      "eval_chemistry_samples_per_second": 6.968,
      "eval_chemistry_steps_per_second": 1.742,
      "step": 1600
    },
    {
      "epoch": 2.56,
      "eval_math_entropy": 0.9108107857704163,
      "eval_math_loss": 0.9712440967559814,
      "eval_math_mean_token_accuracy": 0.7836082301139832,
      "eval_math_num_tokens": 45532525.0,
      "eval_math_runtime": 74.8757,
      "eval_math_samples_per_second": 6.678,
      "eval_math_steps_per_second": 1.669,
      "step": 1600
    },
    {
      "epoch": 2.56,
      "eval_physics_entropy": 1.5397200474739074,
      "eval_physics_loss": 1.6607189178466797,
      "eval_physics_mean_token_accuracy": 0.6987190752029419,
      "eval_physics_num_tokens": 45532525.0,
      "eval_physics_runtime": 85.3015,
      "eval_physics_samples_per_second": 5.862,
      "eval_physics_steps_per_second": 1.465,
      "step": 1600
    },
    {
      "entropy": 0.8905645634979009,
      "epoch": 2.576,
      "grad_norm": 14.8125,
      "learning_rate": 1.0303703703703705e-05,
      "loss": 15.0284,
      "mean_token_accuracy": 0.790876742824912,
      "num_tokens": 45817478.0,
      "step": 1610
    },
    {
      "entropy": 0.8786801369860768,
      "epoch": 2.592,
      "grad_norm": 15.8125,
      "learning_rate": 1.0229629629629631e-05,
      "loss": 14.7209,
      "mean_token_accuracy": 0.793192445859313,
      "num_tokens": 46109575.0,
      "step": 1620
    },
    {
      "entropy": 0.9261208612471818,
      "epoch": 2.608,
      "grad_norm": 17.0,
      "learning_rate": 1.0155555555555556e-05,
      "loss": 15.3656,
      "mean_token_accuracy": 0.7853963386267424,
      "num_tokens": 46391461.0,
      "step": 1630
    },
    {
      "entropy": 0.8752277784049511,
      "epoch": 2.624,
      "grad_norm": 14.5,
      "learning_rate": 1.0081481481481484e-05,
      "loss": 14.5987,
      "mean_token_accuracy": 0.7973049703985453,
      "num_tokens": 46683117.0,
      "step": 1640
    },
    {
      "entropy": 0.9202517494559288,
      "epoch": 2.64,
      "grad_norm": 14.5,
      "learning_rate": 1.0007407407407407e-05,
      "loss": 15.2219,
      "mean_token_accuracy": 0.7870473835617304,
      "num_tokens": 46965534.0,
      "step": 1650
    },
    {
      "entropy": 0.8683292876929045,
      "epoch": 2.656,
      "grad_norm": 14.9375,
      "learning_rate": 9.933333333333334e-06,
      "loss": 14.5546,
      "mean_token_accuracy": 0.7934495940804481,
      "num_tokens": 47245852.0,
      "step": 1660
    },
    {
      "entropy": 0.9163068011403084,
      "epoch": 2.672,
      "grad_norm": 19.25,
      "learning_rate": 9.85925925925926e-06,
      "loss": 15.3249,
      "mean_token_accuracy": 0.7855281837284565,
      "num_tokens": 47524916.0,
      "step": 1670
    },
    {
      "entropy": 0.8939725562930108,
      "epoch": 2.6879999999999997,
      "grad_norm": 14.0625,
      "learning_rate": 9.785185185185187e-06,
      "loss": 14.9286,
      "mean_token_accuracy": 0.7903482187539339,
      "num_tokens": 47807131.0,
      "step": 1680
    },
    {
      "entropy": 0.9146713107824326,
      "epoch": 2.7039999999999997,
      "grad_norm": 17.125,
      "learning_rate": 9.711111111111111e-06,
      "loss": 15.1649,
      "mean_token_accuracy": 0.7876297283917666,
      "num_tokens": 48099654.0,
      "step": 1690
    },
    {
      "entropy": 0.9037863921374083,
      "epoch": 2.7199999999999998,
      "grad_norm": 16.75,
      "learning_rate": 9.637037037037038e-06,
      "loss": 15.0923,
      "mean_token_accuracy": 0.7894866347312928,
      "num_tokens": 48375019.0,
      "step": 1700
    },
    {
      "epoch": 2.7199999999999998,
      "eval_biology_entropy": 3.329032316207886,
      "eval_biology_loss": 3.840982437133789,
      "eval_biology_mean_token_accuracy": 0.45156898140907287,
      "eval_biology_num_tokens": 48375019.0,
      "eval_biology_runtime": 58.7154,
      "eval_biology_samples_per_second": 8.516,
      "eval_biology_steps_per_second": 2.129,
      "step": 1700
    },
    {
      "epoch": 2.7199999999999998,
      "eval_chemistry_entropy": 1.9188699774742126,
      "eval_chemistry_loss": 2.10003662109375,
      "eval_chemistry_mean_token_accuracy": 0.6422763855457306,
      "eval_chemistry_num_tokens": 48375019.0,
      "eval_chemistry_runtime": 72.0001,
      "eval_chemistry_samples_per_second": 6.944,
      "eval_chemistry_steps_per_second": 1.736,
      "step": 1700
    },
    {
      "epoch": 2.7199999999999998,
      "eval_math_entropy": 0.9202389974594116,
      "eval_math_loss": 0.9695459008216858,
      "eval_math_mean_token_accuracy": 0.7839204125404358,
      "eval_math_num_tokens": 48375019.0,
      "eval_math_runtime": 75.274,
      "eval_math_samples_per_second": 6.642,
      "eval_math_steps_per_second": 1.661,
      "step": 1700
    },
    {
      "epoch": 2.7199999999999998,
      "eval_physics_entropy": 1.5536430368423462,
      "eval_physics_loss": 1.657245397567749,
      "eval_physics_mean_token_accuracy": 0.6994028887748718,
      "eval_physics_num_tokens": 48375019.0,
      "eval_physics_runtime": 85.7301,
      "eval_physics_samples_per_second": 5.832,
      "eval_physics_steps_per_second": 1.458,
      "step": 1700
    },
    {
      "entropy": 0.9091611053794623,
      "epoch": 2.7359999999999998,
      "grad_norm": 15.875,
      "learning_rate": 9.562962962962965e-06,
      "loss": 15.0584,
      "mean_token_accuracy": 0.7888620838522911,
      "num_tokens": 48659284.0,
      "step": 1710
    },
    {
      "entropy": 0.8906566344201565,
      "epoch": 2.752,
      "grad_norm": 15.0,
      "learning_rate": 9.48888888888889e-06,
      "loss": 14.9794,
      "mean_token_accuracy": 0.7898790411651134,
      "num_tokens": 48943804.0,
      "step": 1720
    },
    {
      "entropy": 0.9037273216992616,
      "epoch": 2.768,
      "grad_norm": 16.25,
      "learning_rate": 9.414814814814816e-06,
      "loss": 15.0183,
      "mean_token_accuracy": 0.7900580707937479,
      "num_tokens": 49230939.0,
      "step": 1730
    },
    {
      "entropy": 0.9290654907003045,
      "epoch": 2.784,
      "grad_norm": 14.75,
      "learning_rate": 9.34074074074074e-06,
      "loss": 15.437,
      "mean_token_accuracy": 0.7857069440186024,
      "num_tokens": 49504425.0,
      "step": 1740
    },
    {
      "entropy": 0.8954351406544447,
      "epoch": 2.8,
      "grad_norm": 11.75,
      "learning_rate": 9.266666666666667e-06,
      "loss": 15.0599,
      "mean_token_accuracy": 0.7904092535376549,
      "num_tokens": 49782661.0,
      "step": 1750
    },
    {
      "entropy": 0.9036591846495867,
      "epoch": 2.816,
      "grad_norm": 17.25,
      "learning_rate": 9.192592592592594e-06,
      "loss": 15.1346,
      "mean_token_accuracy": 0.7884920053184032,
      "num_tokens": 50073632.0,
      "step": 1760
    },
    {
      "entropy": 0.8982403434813022,
      "epoch": 2.832,
      "grad_norm": 16.125,
      "learning_rate": 9.118518518518518e-06,
      "loss": 14.9059,
      "mean_token_accuracy": 0.7910256557166576,
      "num_tokens": 50356964.0,
      "step": 1770
    },
    {
      "entropy": 0.9227270413190126,
      "epoch": 2.848,
      "grad_norm": 14.75,
      "learning_rate": 9.044444444444445e-06,
      "loss": 15.421,
      "mean_token_accuracy": 0.7865616176277399,
      "num_tokens": 50644535.0,
      "step": 1780
    },
    {
      "entropy": 0.9126858733594417,
      "epoch": 2.864,
      "grad_norm": 16.25,
      "learning_rate": 8.970370370370372e-06,
      "loss": 15.2386,
      "mean_token_accuracy": 0.7880401182919741,
      "num_tokens": 50925653.0,
      "step": 1790
    },
    {
      "entropy": 0.9097051553428173,
      "epoch": 2.88,
      "grad_norm": 15.375,
      "learning_rate": 8.896296296296298e-06,
      "loss": 15.0677,
      "mean_token_accuracy": 0.7898297529667616,
      "num_tokens": 51204374.0,
      "step": 1800
    },
    {
      "epoch": 2.88,
      "eval_biology_entropy": 3.3198141384124757,
      "eval_biology_loss": 3.837639093399048,
      "eval_biology_mean_token_accuracy": 0.45243658185005187,
      "eval_biology_num_tokens": 51204374.0,
      "eval_biology_runtime": 58.3306,
      "eval_biology_samples_per_second": 8.572,
      "eval_biology_steps_per_second": 2.143,
      "step": 1800
    },
    {
      "epoch": 2.88,
      "eval_chemistry_entropy": 1.913723960876465,
      "eval_chemistry_loss": 2.0984654426574707,
      "eval_chemistry_mean_token_accuracy": 0.6421547355651855,
      "eval_chemistry_num_tokens": 51204374.0,
      "eval_chemistry_runtime": 72.008,
      "eval_chemistry_samples_per_second": 6.944,
      "eval_chemistry_steps_per_second": 1.736,
      "step": 1800
    },
    {
      "epoch": 2.88,
      "eval_math_entropy": 0.9192559933662414,
      "eval_math_loss": 0.9683922529220581,
      "eval_math_mean_token_accuracy": 0.7842840232849121,
      "eval_math_num_tokens": 51204374.0,
      "eval_math_runtime": 75.2308,
      "eval_math_samples_per_second": 6.646,
      "eval_math_steps_per_second": 1.662,
      "step": 1800
    },
    {
      "epoch": 2.88,
      "eval_physics_entropy": 1.5504438242912293,
      "eval_physics_loss": 1.6569702625274658,
      "eval_physics_mean_token_accuracy": 0.6993862442970276,
      "eval_physics_num_tokens": 51204374.0,
      "eval_physics_runtime": 85.7348,
      "eval_physics_samples_per_second": 5.832,
      "eval_physics_steps_per_second": 1.458,
      "step": 1800
    },
    {
      "entropy": 0.8897504400461912,
      "epoch": 2.896,
      "grad_norm": 18.5,
      "learning_rate": 8.822222222222223e-06,
      "loss": 14.82,
      "mean_token_accuracy": 0.7926490116864443,
      "num_tokens": 51483944.0,
      "step": 1810
    },
    {
      "entropy": 0.9077999938279391,
      "epoch": 2.912,
      "grad_norm": 15.875,
      "learning_rate": 8.74814814814815e-06,
      "loss": 15.1766,
      "mean_token_accuracy": 0.789422607794404,
      "num_tokens": 51765755.0,
      "step": 1820
    },
    {
      "entropy": 0.8962426841259002,
      "epoch": 2.928,
      "grad_norm": 19.25,
      "learning_rate": 8.674074074074074e-06,
      "loss": 14.9265,
      "mean_token_accuracy": 0.7920733671635389,
      "num_tokens": 52056379.0,
      "step": 1830
    },
    {
      "entropy": 0.8828163981437683,
      "epoch": 2.944,
      "grad_norm": 13.5,
      "learning_rate": 8.6e-06,
      "loss": 14.663,
      "mean_token_accuracy": 0.7927168175578118,
      "num_tokens": 52346232.0,
      "step": 1840
    },
    {
      "entropy": 0.9042975509539246,
      "epoch": 2.96,
      "grad_norm": 13.4375,
      "learning_rate": 8.525925925925927e-06,
      "loss": 15.0837,
      "mean_token_accuracy": 0.7888835649937391,
      "num_tokens": 52633789.0,
      "step": 1850
    },
    {
      "entropy": 0.8866699416190386,
      "epoch": 2.976,
      "grad_norm": 14.3125,
      "learning_rate": 8.451851851851852e-06,
      "loss": 14.8771,
      "mean_token_accuracy": 0.7902060553431511,
      "num_tokens": 52911755.0,
      "step": 1860
    },
    {
      "entropy": 0.8879684330895543,
      "epoch": 2.992,
      "grad_norm": 14.5625,
      "learning_rate": 8.377777777777779e-06,
      "loss": 14.9035,
      "mean_token_accuracy": 0.7912377085536718,
      "num_tokens": 53198176.0,
      "step": 1870
    },
    {
      "entropy": 0.9142865307629109,
      "epoch": 3.008,
      "grad_norm": 14.1875,
      "learning_rate": 8.303703703703705e-06,
      "loss": 15.1842,
      "mean_token_accuracy": 0.7873686749488116,
      "num_tokens": 53481893.0,
      "step": 1880
    },
    {
      "entropy": 0.9048544511198997,
      "epoch": 3.024,
      "grad_norm": 14.625,
      "learning_rate": 8.229629629629632e-06,
      "loss": 15.0856,
      "mean_token_accuracy": 0.7889429580420255,
      "num_tokens": 53771717.0,
      "step": 1890
    },
    {
      "entropy": 0.9106401093304157,
      "epoch": 3.04,
      "grad_norm": 15.5625,
      "learning_rate": 8.155555555555556e-06,
      "loss": 15.1222,
      "mean_token_accuracy": 0.7894735060632229,
      "num_tokens": 54058045.0,
      "step": 1900
    },
    {
      "epoch": 3.04,
      "eval_biology_entropy": 3.2830596961975096,
      "eval_biology_loss": 3.833789587020874,
      "eval_biology_mean_token_accuracy": 0.45244618344306947,
      "eval_biology_num_tokens": 54058045.0,
      "eval_biology_runtime": 58.7003,
      "eval_biology_samples_per_second": 8.518,
      "eval_biology_steps_per_second": 2.129,
      "step": 1900
    },
    {
      "epoch": 3.04,
      "eval_chemistry_entropy": 1.8965351676940918,
      "eval_chemistry_loss": 2.096755027770996,
      "eval_chemistry_mean_token_accuracy": 0.6419810440540313,
      "eval_chemistry_num_tokens": 54058045.0,
      "eval_chemistry_runtime": 72.1508,
      "eval_chemistry_samples_per_second": 6.93,
      "eval_chemistry_steps_per_second": 1.732,
      "step": 1900
    },
    {
      "epoch": 3.04,
      "eval_math_entropy": 0.9148286261558533,
      "eval_math_loss": 0.967832624912262,
      "eval_math_mean_token_accuracy": 0.7839636888504028,
      "eval_math_num_tokens": 54058045.0,
      "eval_math_runtime": 75.4525,
      "eval_math_samples_per_second": 6.627,
      "eval_math_steps_per_second": 1.657,
      "step": 1900
    },
    {
      "epoch": 3.04,
      "eval_physics_entropy": 1.5378339309692384,
      "eval_physics_loss": 1.6525893211364746,
      "eval_physics_mean_token_accuracy": 0.6997761182785034,
      "eval_physics_num_tokens": 54058045.0,
      "eval_physics_runtime": 85.889,
      "eval_physics_samples_per_second": 5.821,
      "eval_physics_steps_per_second": 1.455,
      "step": 1900
    },
    {
      "entropy": 0.9121185712516308,
      "epoch": 3.056,
      "grad_norm": 13.75,
      "learning_rate": 8.081481481481483e-06,
      "loss": 15.301,
      "mean_token_accuracy": 0.7887876890599728,
      "num_tokens": 54334332.0,
      "step": 1910
    },
    {
      "entropy": 0.8813980983570218,
      "epoch": 3.072,
      "grad_norm": 15.875,
      "learning_rate": 8.007407407407408e-06,
      "loss": 14.6814,
      "mean_token_accuracy": 0.7934495750814676,
      "num_tokens": 54624543.0,
      "step": 1920
    },
    {
      "entropy": 0.887355587631464,
      "epoch": 3.088,
      "grad_norm": 17.25,
      "learning_rate": 7.933333333333334e-06,
      "loss": 14.7427,
      "mean_token_accuracy": 0.7948705721646547,
      "num_tokens": 54907550.0,
      "step": 1930
    },
    {
      "entropy": 0.8912877222523093,
      "epoch": 3.104,
      "grad_norm": 14.125,
      "learning_rate": 7.859259259259259e-06,
      "loss": 14.7452,
      "mean_token_accuracy": 0.7931797504425049,
      "num_tokens": 55190959.0,
      "step": 1940
    },
    {
      "entropy": 0.8836303358897567,
      "epoch": 3.12,
      "grad_norm": 15.75,
      "learning_rate": 7.785185185185185e-06,
      "loss": 14.5563,
      "mean_token_accuracy": 0.7959811810404063,
      "num_tokens": 55481635.0,
      "step": 1950
    },
    {
      "entropy": 0.8748507333919406,
      "epoch": 3.136,
      "grad_norm": 16.125,
      "learning_rate": 7.711111111111112e-06,
      "loss": 14.7034,
      "mean_token_accuracy": 0.7934134747833014,
      "num_tokens": 55769010.0,
      "step": 1960
    },
    {
      "entropy": 0.8872682383283973,
      "epoch": 3.152,
      "grad_norm": 15.375,
      "learning_rate": 7.637037037037037e-06,
      "loss": 14.9389,
      "mean_token_accuracy": 0.7926393784582615,
      "num_tokens": 56053160.0,
      "step": 1970
    },
    {
      "entropy": 0.9089289855211973,
      "epoch": 3.168,
      "grad_norm": 13.25,
      "learning_rate": 7.562962962962963e-06,
      "loss": 15.036,
      "mean_token_accuracy": 0.7902255799621344,
      "num_tokens": 56337066.0,
      "step": 1980
    },
    {
      "entropy": 0.8836135413497687,
      "epoch": 3.184,
      "grad_norm": 16.5,
      "learning_rate": 7.48888888888889e-06,
      "loss": 14.723,
      "mean_token_accuracy": 0.7914205066859722,
      "num_tokens": 56618899.0,
      "step": 1990
    },
    {
      "entropy": 0.8735335521399975,
      "epoch": 3.2,
      "grad_norm": 13.0,
      "learning_rate": 7.4148148148148155e-06,
      "loss": 14.4853,
      "mean_token_accuracy": 0.7956909481436014,
      "num_tokens": 56910071.0,
      "step": 2000
    },
    {
      "epoch": 3.2,
      "eval_biology_entropy": 3.2912813301086428,
      "eval_biology_loss": 3.8430533409118652,
      "eval_biology_mean_token_accuracy": 0.45210277223587036,
      "eval_biology_num_tokens": 56910071.0,
      "eval_biology_runtime": 58.6889,
      "eval_biology_samples_per_second": 8.519,
      "eval_biology_steps_per_second": 2.13,
      "step": 2000
    },
    {
      "epoch": 3.2,
      "eval_chemistry_entropy": 1.8982752962112426,
      "eval_chemistry_loss": 2.09956955909729,
      "eval_chemistry_mean_token_accuracy": 0.642627259016037,
      "eval_chemistry_num_tokens": 56910071.0,
      "eval_chemistry_runtime": 72.2818,
      "eval_chemistry_samples_per_second": 6.917,
      "eval_chemistry_steps_per_second": 1.729,
      "step": 2000
    },
    {
      "epoch": 3.2,
      "eval_math_entropy": 0.9123949198722839,
      "eval_math_loss": 0.9670660495758057,
      "eval_math_mean_token_accuracy": 0.7842952456474305,
      "eval_math_num_tokens": 56910071.0,
      "eval_math_runtime": 75.1235,
      "eval_math_samples_per_second": 6.656,
      "eval_math_steps_per_second": 1.664,
      "step": 2000
    },
    {
      "epoch": 3.2,
      "eval_physics_entropy": 1.5381169476509093,
      "eval_physics_loss": 1.6558867692947388,
      "eval_physics_mean_token_accuracy": 0.6997131333351135,
      "eval_physics_num_tokens": 56910071.0,
      "eval_physics_runtime": 85.7515,
      "eval_physics_samples_per_second": 5.831,
      "eval_physics_steps_per_second": 1.458,
      "step": 2000
    }
  ],
  "logging_steps": 10,
  "max_steps": 3000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "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": 5.791878293573609e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}