trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 100,
  "global_step": 3416,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.117096018735363,
      "eval_accuracy": 0.16272885212172436,
      "eval_loss": 6.373325347900391,
      "eval_runtime": 1.492,
      "eval_samples_per_second": 153.483,
      "eval_steps_per_second": 1.34,
      "step": 100
    },
    {
      "epoch": 0.234192037470726,
      "grad_norm": 0.5273725986480713,
      "learning_rate": 0.00029824050644124145,
      "loss": 6.8543,
      "step": 200
    },
    {
      "epoch": 0.234192037470726,
      "eval_accuracy": 0.19693341358364602,
      "eval_loss": 5.696830749511719,
      "eval_runtime": 1.1659,
      "eval_samples_per_second": 196.414,
      "eval_steps_per_second": 1.715,
      "step": 200
    },
    {
      "epoch": 0.351288056206089,
      "eval_accuracy": 0.21933947162852643,
      "eval_loss": 5.292416095733643,
      "eval_runtime": 1.1779,
      "eval_samples_per_second": 194.406,
      "eval_steps_per_second": 1.698,
      "step": 300
    },
    {
      "epoch": 0.468384074941452,
      "grad_norm": 0.7039262056350708,
      "learning_rate": 0.00029145544445670097,
      "loss": 5.3976,
      "step": 400
    },
    {
      "epoch": 0.468384074941452,
      "eval_accuracy": 0.2335839021287676,
      "eval_loss": 5.054598808288574,
      "eval_runtime": 1.1808,
      "eval_samples_per_second": 193.929,
      "eval_steps_per_second": 1.694,
      "step": 400
    },
    {
      "epoch": 0.585480093676815,
      "eval_accuracy": 0.24265474864150735,
      "eval_loss": 4.891258716583252,
      "eval_runtime": 1.1832,
      "eval_samples_per_second": 193.538,
      "eval_steps_per_second": 1.69,
      "step": 500
    },
    {
      "epoch": 0.702576112412178,
      "grad_norm": 0.7881871461868286,
      "learning_rate": 0.0002797991505064349,
      "loss": 4.9739,
      "step": 600
    },
    {
      "epoch": 0.702576112412178,
      "eval_accuracy": 0.2506498994736775,
      "eval_loss": 4.759438514709473,
      "eval_runtime": 1.1726,
      "eval_samples_per_second": 195.299,
      "eval_steps_per_second": 1.706,
      "step": 600
    },
    {
      "epoch": 0.819672131147541,
      "eval_accuracy": 0.2589993468990511,
      "eval_loss": 4.651967525482178,
      "eval_runtime": 1.1752,
      "eval_samples_per_second": 194.864,
      "eval_steps_per_second": 1.702,
      "step": 700
    },
    {
      "epoch": 0.936768149882904,
      "grad_norm": 0.5959051251411438,
      "learning_rate": 0.00026367302668500736,
      "loss": 4.7436,
      "step": 800
    },
    {
      "epoch": 0.936768149882904,
      "eval_accuracy": 0.2653638796757546,
      "eval_loss": 4.566330432891846,
      "eval_runtime": 1.1821,
      "eval_samples_per_second": 193.727,
      "eval_steps_per_second": 1.692,
      "step": 800
    },
    {
      "epoch": 1.053864168618267,
      "eval_accuracy": 0.270379524218093,
      "eval_loss": 4.497960567474365,
      "eval_runtime": 1.1842,
      "eval_samples_per_second": 193.385,
      "eval_steps_per_second": 1.689,
      "step": 900
    },
    {
      "epoch": 1.17096018735363,
      "grad_norm": 0.6118926405906677,
      "learning_rate": 0.00024363240043240403,
      "loss": 4.5595,
      "step": 1000
    },
    {
      "epoch": 1.17096018735363,
      "eval_accuracy": 0.2755872572748189,
      "eval_loss": 4.44118070602417,
      "eval_runtime": 1.1779,
      "eval_samples_per_second": 194.409,
      "eval_steps_per_second": 1.698,
      "step": 1000
    },
    {
      "epoch": 1.288056206088993,
      "eval_accuracy": 0.2790832682366701,
      "eval_loss": 4.388593673706055,
      "eval_runtime": 1.1719,
      "eval_samples_per_second": 195.408,
      "eval_steps_per_second": 1.707,
      "step": 1100
    },
    {
      "epoch": 1.405152224824356,
      "grad_norm": 0.6060830354690552,
      "learning_rate": 0.00022036740099454973,
      "loss": 4.44,
      "step": 1200
    },
    {
      "epoch": 1.405152224824356,
      "eval_accuracy": 0.28338605095894853,
      "eval_loss": 4.339376449584961,
      "eval_runtime": 1.1734,
      "eval_samples_per_second": 195.157,
      "eval_steps_per_second": 1.704,
      "step": 1200
    },
    {
      "epoch": 1.5222482435597189,
      "eval_accuracy": 0.28774005728506363,
      "eval_loss": 4.292956352233887,
      "eval_runtime": 1.1856,
      "eval_samples_per_second": 193.144,
      "eval_steps_per_second": 1.687,
      "step": 1300
    },
    {
      "epoch": 1.639344262295082,
      "grad_norm": 0.5661224126815796,
      "learning_rate": 0.0001946791937799555,
      "loss": 4.3482,
      "step": 1400
    },
    {
      "epoch": 1.639344262295082,
      "eval_accuracy": 0.2911549642075068,
      "eval_loss": 4.25685977935791,
      "eval_runtime": 1.175,
      "eval_samples_per_second": 194.899,
      "eval_steps_per_second": 1.702,
      "step": 1400
    },
    {
      "epoch": 1.756440281030445,
      "eval_accuracy": 0.29346429501380905,
      "eval_loss": 4.224252700805664,
      "eval_runtime": 1.1755,
      "eval_samples_per_second": 194.807,
      "eval_steps_per_second": 1.701,
      "step": 1500
    },
    {
      "epoch": 1.8735362997658078,
      "grad_norm": 0.5346641540527344,
      "learning_rate": 0.00016745239101833442,
      "loss": 4.2765,
      "step": 1600
    },
    {
      "epoch": 1.8735362997658078,
      "eval_accuracy": 0.29705421591602743,
      "eval_loss": 4.186337947845459,
      "eval_runtime": 1.1767,
      "eval_samples_per_second": 194.606,
      "eval_steps_per_second": 1.7,
      "step": 1600
    },
    {
      "epoch": 1.990632318501171,
      "eval_accuracy": 0.3002471538885118,
      "eval_loss": 4.158440113067627,
      "eval_runtime": 1.1742,
      "eval_samples_per_second": 195.024,
      "eval_steps_per_second": 1.703,
      "step": 1700
    },
    {
      "epoch": 2.107728337236534,
      "grad_norm": 0.5907180309295654,
      "learning_rate": 0.0001396245888018435,
      "loss": 4.1869,
      "step": 1800
    },
    {
      "epoch": 2.107728337236534,
      "eval_accuracy": 0.3025009924573243,
      "eval_loss": 4.138100624084473,
      "eval_runtime": 1.1874,
      "eval_samples_per_second": 192.866,
      "eval_steps_per_second": 1.684,
      "step": 1800
    },
    {
      "epoch": 2.2248243559718968,
      "eval_accuracy": 0.30516461985683,
      "eval_loss": 4.114769458770752,
      "eval_runtime": 1.1787,
      "eval_samples_per_second": 194.278,
      "eval_steps_per_second": 1.697,
      "step": 1900
    },
    {
      "epoch": 2.34192037470726,
      "grad_norm": 0.5956910848617554,
      "learning_rate": 0.00011215407954696412,
      "loss": 4.1154,
      "step": 2000
    },
    {
      "epoch": 2.34192037470726,
      "eval_accuracy": 0.30651777672484815,
      "eval_loss": 4.09398889541626,
      "eval_runtime": 1.1823,
      "eval_samples_per_second": 193.693,
      "eval_steps_per_second": 1.692,
      "step": 2000
    },
    {
      "epoch": 2.459016393442623,
      "eval_accuracy": 0.30912164325321106,
      "eval_loss": 4.074470043182373,
      "eval_runtime": 1.1705,
      "eval_samples_per_second": 195.64,
      "eval_steps_per_second": 1.709,
      "step": 2100
    },
    {
      "epoch": 2.576112412177986,
      "grad_norm": 0.6125728487968445,
      "learning_rate": 8.611292881304631e-05,
      "loss": 4.0822,
      "step": 2200
    },
    {
      "epoch": 2.576112412177986,
      "eval_accuracy": 0.31070530633849414,
      "eval_loss": 4.0601091384887695,
      "eval_runtime": 1.176,
      "eval_samples_per_second": 194.735,
      "eval_steps_per_second": 1.701,
      "step": 2200
    },
    {
      "epoch": 2.693208430913349,
      "eval_accuracy": 0.31145231722777855,
      "eval_loss": 4.047569274902344,
      "eval_runtime": 1.1721,
      "eval_samples_per_second": 195.374,
      "eval_steps_per_second": 1.706,
      "step": 2300
    },
    {
      "epoch": 2.810304449648712,
      "grad_norm": 0.5131784081459045,
      "learning_rate": 6.21369402825085e-05,
      "loss": 4.0496,
      "step": 2400
    },
    {
      "epoch": 2.810304449648712,
      "eval_accuracy": 0.3138342148061827,
      "eval_loss": 4.032547950744629,
      "eval_runtime": 1.1719,
      "eval_samples_per_second": 195.415,
      "eval_steps_per_second": 1.707,
      "step": 2400
    },
    {
      "epoch": 2.927400468384075,
      "eval_accuracy": 0.31531969931744547,
      "eval_loss": 4.0211310386657715,
      "eval_runtime": 1.1766,
      "eval_samples_per_second": 194.63,
      "eval_steps_per_second": 1.7,
      "step": 2500
    },
    {
      "epoch": 3.0444964871194378,
      "grad_norm": 0.5091509819030762,
      "learning_rate": 4.118664897236514e-05,
      "loss": 4.0193,
      "step": 2600
    },
    {
      "epoch": 3.0444964871194378,
      "eval_accuracy": 0.31634844002783147,
      "eval_loss": 4.012126445770264,
      "eval_runtime": 1.1748,
      "eval_samples_per_second": 194.929,
      "eval_steps_per_second": 1.702,
      "step": 2600
    },
    {
      "epoch": 3.161592505854801,
      "eval_accuracy": 0.31667712481911664,
      "eval_loss": 4.007138729095459,
      "eval_runtime": 1.1742,
      "eval_samples_per_second": 195.034,
      "eval_steps_per_second": 1.703,
      "step": 2700
    },
    {
      "epoch": 3.278688524590164,
      "grad_norm": 0.4789488911628723,
      "learning_rate": 2.39835098192799e-05,
      "loss": 3.9665,
      "step": 2800
    },
    {
      "epoch": 3.278688524590164,
      "eval_accuracy": 0.31719362949113616,
      "eval_loss": 4.001045227050781,
      "eval_runtime": 1.1742,
      "eval_samples_per_second": 195.029,
      "eval_steps_per_second": 1.703,
      "step": 2800
    },
    {
      "epoch": 3.3957845433255267,
      "eval_accuracy": 0.31759488105452327,
      "eval_loss": 3.997783899307251,
      "eval_runtime": 1.1723,
      "eval_samples_per_second": 195.344,
      "eval_steps_per_second": 1.706,
      "step": 2900
    },
    {
      "epoch": 3.51288056206089,
      "grad_norm": 0.4739971458911896,
      "learning_rate": 1.1119938913529075e-05,
      "loss": 3.9575,
      "step": 3000
    },
    {
      "epoch": 3.51288056206089,
      "eval_accuracy": 0.3181796838649917,
      "eval_loss": 3.993601083755493,
      "eval_runtime": 1.1865,
      "eval_samples_per_second": 192.998,
      "eval_steps_per_second": 1.686,
      "step": 3000
    },
    {
      "epoch": 3.629976580796253,
      "eval_accuracy": 0.3181924897659508,
      "eval_loss": 3.991323947906494,
      "eval_runtime": 1.1742,
      "eval_samples_per_second": 195.026,
      "eval_steps_per_second": 1.703,
      "step": 3100
    },
    {
      "epoch": 3.747072599531616,
      "grad_norm": 0.47446030378341675,
      "learning_rate": 3.038912755806222e-06,
      "loss": 3.9558,
      "step": 3200
    },
    {
      "epoch": 3.747072599531616,
      "eval_accuracy": 0.3182010270332569,
      "eval_loss": 3.9895448684692383,
      "eval_runtime": 1.1801,
      "eval_samples_per_second": 194.046,
      "eval_steps_per_second": 1.695,
      "step": 3200
    },
    {
      "epoch": 3.8641686182669788,
      "eval_accuracy": 0.3184998313889707,
      "eval_loss": 3.9884603023529053,
      "eval_runtime": 1.1735,
      "eval_samples_per_second": 195.146,
      "eval_steps_per_second": 1.704,
      "step": 3300
    },
    {
      "epoch": 3.981264637002342,
      "grad_norm": 0.4558132290840149,
      "learning_rate": 1.8713692093474952e-08,
      "loss": 3.9531,
      "step": 3400
    },
    {
      "epoch": 3.981264637002342,
      "eval_accuracy": 0.3183333546765016,
      "eval_loss": 3.988285541534424,
      "eval_runtime": 1.1729,
      "eval_samples_per_second": 195.237,
      "eval_steps_per_second": 1.705,
      "step": 3400
    },
    {
      "epoch": 4.0,
      "step": 3416,
      "total_flos": 1.543347550373806e+17,
      "train_loss": 4.46121194826077,
      "train_runtime": 2449.295,
      "train_samples_per_second": 178.438,
      "train_steps_per_second": 1.395
    }
  ],
  "logging_steps": 200,
  "max_steps": 3416,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 5000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.543347550373806e+17,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}