{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 60210,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 1.25e-06,
      "loss": 7.7769,
      "step": 500
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.5e-06,
      "loss": 4.9602,
      "step": 1000
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.75e-06,
      "loss": 4.1666,
      "step": 1500
    },
    {
      "epoch": 0.17,
      "learning_rate": 5e-06,
      "loss": 3.688,
      "step": 2000
    },
    {
      "epoch": 0.21,
      "learning_rate": 6.25e-06,
      "loss": 3.4594,
      "step": 2500
    },
    {
      "epoch": 0.25,
      "learning_rate": 7.5e-06,
      "loss": 3.2406,
      "step": 3000
    },
    {
      "epoch": 0.29,
      "learning_rate": 8.75e-06,
      "loss": 3.1113,
      "step": 3500
    },
    {
      "epoch": 0.33,
      "learning_rate": 1e-05,
      "loss": 2.9628,
      "step": 4000
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.125e-05,
      "loss": 2.8324,
      "step": 4500
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.25e-05,
      "loss": 2.7406,
      "step": 5000
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.3750000000000002e-05,
      "loss": 2.6505,
      "step": 5500
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.5e-05,
      "loss": 2.5864,
      "step": 6000
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.6250000000000002e-05,
      "loss": 2.5592,
      "step": 6500
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.75e-05,
      "loss": 2.513,
      "step": 7000
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 2.4164,
      "step": 7500
    },
    {
      "epoch": 0.66,
      "learning_rate": 2e-05,
      "loss": 2.4196,
      "step": 8000
    },
    {
      "epoch": 0.71,
      "learning_rate": 2.125e-05,
      "loss": 2.387,
      "step": 8500
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.25e-05,
      "loss": 2.3328,
      "step": 9000
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.375e-05,
      "loss": 2.307,
      "step": 9500
    },
    {
      "epoch": 0.83,
      "learning_rate": 2.5e-05,
      "loss": 2.2919,
      "step": 10000
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.625e-05,
      "loss": 2.2488,
      "step": 10500
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 2.2451,
      "step": 11000
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.8749999999999997e-05,
      "loss": 2.228,
      "step": 11500
    },
    {
      "epoch": 1.0,
      "learning_rate": 3e-05,
      "loss": 2.2017,
      "step": 12000
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.125e-05,
      "loss": 2.1533,
      "step": 12500
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 2.1322,
      "step": 13000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.375000000000001e-05,
      "loss": 2.1557,
      "step": 13500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.5e-05,
      "loss": 2.1209,
      "step": 14000
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.625e-05,
      "loss": 2.103,
      "step": 14500
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 2.1099,
      "step": 15000
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.875e-05,
      "loss": 2.1024,
      "step": 15500
    },
    {
      "epoch": 1.33,
      "learning_rate": 4e-05,
      "loss": 2.0618,
      "step": 16000
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.125e-05,
      "loss": 2.073,
      "step": 16500
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.25e-05,
      "loss": 2.0616,
      "step": 17000
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.375e-05,
      "loss": 2.0967,
      "step": 17500
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.5e-05,
      "loss": 2.0628,
      "step": 18000
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.6250000000000006e-05,
      "loss": 2.0786,
      "step": 18500
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.75e-05,
      "loss": 2.0524,
      "step": 19000
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.875e-05,
      "loss": 2.0508,
      "step": 19500
    },
    {
      "epoch": 1.66,
      "learning_rate": 5e-05,
      "loss": 2.0352,
      "step": 20000
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.937826411340463e-05,
      "loss": 2.0348,
      "step": 20500
    },
    {
      "epoch": 1.74,
      "learning_rate": 4.8756528226809256e-05,
      "loss": 2.0173,
      "step": 21000
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.813479234021388e-05,
      "loss": 2.0102,
      "step": 21500
    },
    {
      "epoch": 1.83,
      "learning_rate": 4.7513056453618504e-05,
      "loss": 1.9995,
      "step": 22000
    },
    {
      "epoch": 1.87,
      "learning_rate": 4.689132056702313e-05,
      "loss": 2.0089,
      "step": 22500
    },
    {
      "epoch": 1.91,
      "learning_rate": 4.626958468042776e-05,
      "loss": 1.9712,
      "step": 23000
    },
    {
      "epoch": 1.95,
      "learning_rate": 4.5647848793832385e-05,
      "loss": 1.9776,
      "step": 23500
    },
    {
      "epoch": 1.99,
      "learning_rate": 4.5026112907237005e-05,
      "loss": 1.9573,
      "step": 24000
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.440437702064163e-05,
      "loss": 1.877,
      "step": 24500
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.378264113404626e-05,
      "loss": 1.8312,
      "step": 25000
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.3160905247450886e-05,
      "loss": 1.8694,
      "step": 25500
    },
    {
      "epoch": 2.16,
      "learning_rate": 4.253916936085551e-05,
      "loss": 1.8549,
      "step": 26000
    },
    {
      "epoch": 2.2,
      "learning_rate": 4.191743347426014e-05,
      "loss": 1.8445,
      "step": 26500
    },
    {
      "epoch": 2.24,
      "learning_rate": 4.129569758766476e-05,
      "loss": 1.8301,
      "step": 27000
    },
    {
      "epoch": 2.28,
      "learning_rate": 4.067396170106939e-05,
      "loss": 1.8494,
      "step": 27500
    },
    {
      "epoch": 2.33,
      "learning_rate": 4.0052225814474015e-05,
      "loss": 1.8504,
      "step": 28000
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.943048992787864e-05,
      "loss": 1.8388,
      "step": 28500
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.880875404128326e-05,
      "loss": 1.8036,
      "step": 29000
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.8187018154687896e-05,
      "loss": 1.825,
      "step": 29500
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.7565282268092516e-05,
      "loss": 1.8174,
      "step": 30000
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.694354638149714e-05,
      "loss": 1.8188,
      "step": 30500
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.632181049490177e-05,
      "loss": 1.8334,
      "step": 31000
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.57000746083064e-05,
      "loss": 1.8342,
      "step": 31500
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.507833872171102e-05,
      "loss": 1.8119,
      "step": 32000
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.4456602835115645e-05,
      "loss": 1.8034,
      "step": 32500
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.383486694852027e-05,
      "loss": 1.7964,
      "step": 33000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.32131310619249e-05,
      "loss": 1.7865,
      "step": 33500
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.259139517532952e-05,
      "loss": 1.7937,
      "step": 34000
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.196965928873415e-05,
      "loss": 1.7982,
      "step": 34500
    },
    {
      "epoch": 2.91,
      "learning_rate": 3.134792340213877e-05,
      "loss": 1.7969,
      "step": 35000
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.07261875155434e-05,
      "loss": 1.7892,
      "step": 35500
    },
    {
      "epoch": 2.99,
      "learning_rate": 3.010445162894802e-05,
      "loss": 1.769,
      "step": 36000
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.948271574235265e-05,
      "loss": 1.6569,
      "step": 36500
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.8860979855757274e-05,
      "loss": 1.6269,
      "step": 37000
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.8239243969161898e-05,
      "loss": 1.6806,
      "step": 37500
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.761750808256653e-05,
      "loss": 1.6658,
      "step": 38000
    },
    {
      "epoch": 3.2,
      "learning_rate": 2.6995772195971152e-05,
      "loss": 1.654,
      "step": 38500
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.6374036309375776e-05,
      "loss": 1.6672,
      "step": 39000
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.5752300422780406e-05,
      "loss": 1.6427,
      "step": 39500
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.513056453618503e-05,
      "loss": 1.6182,
      "step": 40000
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.4508828649589653e-05,
      "loss": 1.6515,
      "step": 40500
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.388709276299428e-05,
      "loss": 1.6646,
      "step": 41000
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.3265356876398904e-05,
      "loss": 1.644,
      "step": 41500
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.264362098980353e-05,
      "loss": 1.6337,
      "step": 42000
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.2021885103208158e-05,
      "loss": 1.6412,
      "step": 42500
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.1400149216612782e-05,
      "loss": 1.6222,
      "step": 43000
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.077841333001741e-05,
      "loss": 1.6513,
      "step": 43500
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.0156677443422033e-05,
      "loss": 1.6642,
      "step": 44000
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.953494155682666e-05,
      "loss": 1.6336,
      "step": 44500
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.8913205670231287e-05,
      "loss": 1.6307,
      "step": 45000
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.829146978363591e-05,
      "loss": 1.6365,
      "step": 45500
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.7669733897040537e-05,
      "loss": 1.6427,
      "step": 46000
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.704799801044516e-05,
      "loss": 1.6439,
      "step": 46500
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.6426262123849788e-05,
      "loss": 1.6309,
      "step": 47000
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.5804526237254415e-05,
      "loss": 1.6301,
      "step": 47500
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.518279035065904e-05,
      "loss": 1.6172,
      "step": 48000
    },
    {
      "epoch": 4.03,
      "learning_rate": 1.4561054464063667e-05,
      "loss": 1.55,
      "step": 48500
    },
    {
      "epoch": 4.07,
      "learning_rate": 1.3939318577468291e-05,
      "loss": 1.5255,
      "step": 49000
    },
    {
      "epoch": 4.11,
      "learning_rate": 1.3317582690872918e-05,
      "loss": 1.5083,
      "step": 49500
    },
    {
      "epoch": 4.15,
      "learning_rate": 1.2695846804277545e-05,
      "loss": 1.524,
      "step": 50000
    },
    {
      "epoch": 4.19,
      "learning_rate": 1.2074110917682169e-05,
      "loss": 1.5256,
      "step": 50500
    },
    {
      "epoch": 4.24,
      "learning_rate": 1.1452375031086796e-05,
      "loss": 1.5113,
      "step": 51000
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.0830639144491421e-05,
      "loss": 1.5289,
      "step": 51500
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.0208903257896047e-05,
      "loss": 1.5078,
      "step": 52000
    },
    {
      "epoch": 4.36,
      "learning_rate": 9.587167371300672e-06,
      "loss": 1.5072,
      "step": 52500
    },
    {
      "epoch": 4.4,
      "learning_rate": 8.965431484705297e-06,
      "loss": 1.4986,
      "step": 53000
    },
    {
      "epoch": 4.44,
      "learning_rate": 8.343695598109923e-06,
      "loss": 1.5014,
      "step": 53500
    },
    {
      "epoch": 4.48,
      "learning_rate": 7.72195971151455e-06,
      "loss": 1.5196,
      "step": 54000
    },
    {
      "epoch": 4.53,
      "learning_rate": 7.100223824919175e-06,
      "loss": 1.5027,
      "step": 54500
    },
    {
      "epoch": 4.57,
      "learning_rate": 6.4784879383238e-06,
      "loss": 1.4911,
      "step": 55000
    },
    {
      "epoch": 4.61,
      "learning_rate": 5.856752051728426e-06,
      "loss": 1.4912,
      "step": 55500
    },
    {
      "epoch": 4.65,
      "learning_rate": 5.235016165133052e-06,
      "loss": 1.4958,
      "step": 56000
    },
    {
      "epoch": 4.69,
      "learning_rate": 4.613280278537677e-06,
      "loss": 1.5077,
      "step": 56500
    },
    {
      "epoch": 4.73,
      "learning_rate": 3.9915443919423026e-06,
      "loss": 1.4834,
      "step": 57000
    },
    {
      "epoch": 4.77,
      "learning_rate": 3.3698085053469288e-06,
      "loss": 1.4964,
      "step": 57500
    },
    {
      "epoch": 4.82,
      "learning_rate": 2.748072618751554e-06,
      "loss": 1.5195,
      "step": 58000
    },
    {
      "epoch": 4.86,
      "learning_rate": 2.12633673215618e-06,
      "loss": 1.5117,
      "step": 58500
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.5046008455608059e-06,
      "loss": 1.507,
      "step": 59000
    },
    {
      "epoch": 4.94,
      "learning_rate": 8.828649589654314e-07,
      "loss": 1.5086,
      "step": 59500
    },
    {
      "epoch": 4.98,
      "learning_rate": 2.6112907237005724e-07,
      "loss": 1.5051,
      "step": 60000
    },
    {
      "epoch": 5.0,
      "step": 60210,
      "total_flos": 4721678852567040.0,
      "train_loss": 2.007782585069282,
      "train_runtime": 8390.77,
      "train_samples_per_second": 14.351,
      "train_steps_per_second": 7.176
    }
  ],
  "logging_steps": 500,
  "max_steps": 60210,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 4721678852567040.0,
  "trial_name": null,
  "trial_params": null
}