{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 31910,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.16,
      "learning_rate": 4.9216546537135696e-05,
      "loss": 3.1564,
      "step": 500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.843309307427139e-05,
      "loss": 2.8629,
      "step": 1000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.764963961140708e-05,
      "loss": 2.7321,
      "step": 1500
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.6866186148542776e-05,
      "loss": 2.657,
      "step": 2000
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.608273268567847e-05,
      "loss": 2.6045,
      "step": 2500
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.529927922281417e-05,
      "loss": 2.5492,
      "step": 3000
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.4515825759949856e-05,
      "loss": 2.5043,
      "step": 3500
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.3732372297085556e-05,
      "loss": 2.4617,
      "step": 4000
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.294891883422125e-05,
      "loss": 2.4395,
      "step": 4500
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.216546537135694e-05,
      "loss": 2.4319,
      "step": 5000
    },
    {
      "epoch": 1.72,
      "learning_rate": 4.1382011908492636e-05,
      "loss": 2.3865,
      "step": 5500
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.059855844562833e-05,
      "loss": 2.3805,
      "step": 6000
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.981510498276402e-05,
      "loss": 2.3441,
      "step": 6500
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.903165151989972e-05,
      "loss": 2.3244,
      "step": 7000
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.824819805703541e-05,
      "loss": 2.3225,
      "step": 7500
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.746474459417111e-05,
      "loss": 2.2977,
      "step": 8000
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.66812911313068e-05,
      "loss": 2.2783,
      "step": 8500
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.5897837668442496e-05,
      "loss": 2.2706,
      "step": 9000
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.511438420557819e-05,
      "loss": 2.26,
      "step": 9500
    },
    {
      "epoch": 3.13,
      "learning_rate": 3.433093074271388e-05,
      "loss": 2.2494,
      "step": 10000
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.3547477279849576e-05,
      "loss": 2.2223,
      "step": 10500
    },
    {
      "epoch": 3.45,
      "learning_rate": 3.2764023816985276e-05,
      "loss": 2.2247,
      "step": 11000
    },
    {
      "epoch": 3.6,
      "learning_rate": 3.198057035412096e-05,
      "loss": 2.223,
      "step": 11500
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.119711689125666e-05,
      "loss": 2.1946,
      "step": 12000
    },
    {
      "epoch": 3.92,
      "learning_rate": 3.0413663428392353e-05,
      "loss": 2.2036,
      "step": 12500
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.963020996552805e-05,
      "loss": 2.1822,
      "step": 13000
    },
    {
      "epoch": 4.23,
      "learning_rate": 2.884675650266374e-05,
      "loss": 2.1706,
      "step": 13500
    },
    {
      "epoch": 4.39,
      "learning_rate": 2.8063303039799436e-05,
      "loss": 2.1655,
      "step": 14000
    },
    {
      "epoch": 4.54,
      "learning_rate": 2.727984957693513e-05,
      "loss": 2.1517,
      "step": 14500
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.6496396114070827e-05,
      "loss": 2.1619,
      "step": 15000
    },
    {
      "epoch": 4.86,
      "learning_rate": 2.5712942651206517e-05,
      "loss": 2.1518,
      "step": 15500
    },
    {
      "epoch": 5.01,
      "learning_rate": 2.4929489188342213e-05,
      "loss": 2.1542,
      "step": 16000
    },
    {
      "epoch": 5.17,
      "learning_rate": 2.414603572547791e-05,
      "loss": 2.1407,
      "step": 16500
    },
    {
      "epoch": 5.33,
      "learning_rate": 2.3362582262613603e-05,
      "loss": 2.1297,
      "step": 17000
    },
    {
      "epoch": 5.48,
      "learning_rate": 2.2579128799749297e-05,
      "loss": 2.1235,
      "step": 17500
    },
    {
      "epoch": 5.64,
      "learning_rate": 2.179567533688499e-05,
      "loss": 2.1185,
      "step": 18000
    },
    {
      "epoch": 5.8,
      "learning_rate": 2.1012221874020683e-05,
      "loss": 2.1085,
      "step": 18500
    },
    {
      "epoch": 5.95,
      "learning_rate": 2.022876841115638e-05,
      "loss": 2.1073,
      "step": 19000
    },
    {
      "epoch": 6.11,
      "learning_rate": 1.9445314948292073e-05,
      "loss": 2.0961,
      "step": 19500
    },
    {
      "epoch": 6.27,
      "learning_rate": 1.8661861485427767e-05,
      "loss": 2.0912,
      "step": 20000
    },
    {
      "epoch": 6.42,
      "learning_rate": 1.787840802256346e-05,
      "loss": 2.0957,
      "step": 20500
    },
    {
      "epoch": 6.58,
      "learning_rate": 1.7094954559699157e-05,
      "loss": 2.0874,
      "step": 21000
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.631150109683485e-05,
      "loss": 2.0912,
      "step": 21500
    },
    {
      "epoch": 6.89,
      "learning_rate": 1.5528047633970544e-05,
      "loss": 2.0849,
      "step": 22000
    },
    {
      "epoch": 7.05,
      "learning_rate": 1.4744594171106239e-05,
      "loss": 2.0867,
      "step": 22500
    },
    {
      "epoch": 7.21,
      "learning_rate": 1.3961140708241932e-05,
      "loss": 2.0729,
      "step": 23000
    },
    {
      "epoch": 7.36,
      "learning_rate": 1.3177687245377627e-05,
      "loss": 2.0824,
      "step": 23500
    },
    {
      "epoch": 7.52,
      "learning_rate": 1.2394233782513319e-05,
      "loss": 2.0678,
      "step": 24000
    },
    {
      "epoch": 7.68,
      "learning_rate": 1.1610780319649014e-05,
      "loss": 2.0553,
      "step": 24500
    },
    {
      "epoch": 7.83,
      "learning_rate": 1.0827326856784707e-05,
      "loss": 2.0678,
      "step": 25000
    },
    {
      "epoch": 7.99,
      "learning_rate": 1.0043873393920402e-05,
      "loss": 2.0624,
      "step": 25500
    },
    {
      "epoch": 8.15,
      "learning_rate": 9.260419931056095e-06,
      "loss": 2.0681,
      "step": 26000
    },
    {
      "epoch": 8.3,
      "learning_rate": 8.476966468191789e-06,
      "loss": 2.0587,
      "step": 26500
    },
    {
      "epoch": 8.46,
      "learning_rate": 7.693513005327484e-06,
      "loss": 2.0504,
      "step": 27000
    },
    {
      "epoch": 8.62,
      "learning_rate": 6.910059542463178e-06,
      "loss": 2.0483,
      "step": 27500
    },
    {
      "epoch": 8.77,
      "learning_rate": 6.126606079598872e-06,
      "loss": 2.0504,
      "step": 28000
    },
    {
      "epoch": 8.93,
      "learning_rate": 5.3431526167345665e-06,
      "loss": 2.0539,
      "step": 28500
    },
    {
      "epoch": 9.09,
      "learning_rate": 4.55969915387026e-06,
      "loss": 2.0507,
      "step": 29000
    },
    {
      "epoch": 9.24,
      "learning_rate": 3.7762456910059544e-06,
      "loss": 2.0396,
      "step": 29500
    },
    {
      "epoch": 9.4,
      "learning_rate": 2.9927922281416486e-06,
      "loss": 2.0538,
      "step": 30000
    },
    {
      "epoch": 9.56,
      "learning_rate": 2.2093387652773424e-06,
      "loss": 2.0514,
      "step": 30500
    },
    {
      "epoch": 9.71,
      "learning_rate": 1.4258853024130368e-06,
      "loss": 2.0394,
      "step": 31000
    },
    {
      "epoch": 9.87,
      "learning_rate": 6.424318395487308e-07,
      "loss": 2.048,
      "step": 31500
    },
    {
      "epoch": 10.0,
      "step": 31910,
      "total_flos": 1.667595862278144e+16,
      "train_loss": 2.2214493739870194,
      "train_runtime": 13454.7366,
      "train_samples_per_second": 37.944,
      "train_steps_per_second": 2.372
    }
  ],
  "max_steps": 31910,
  "num_train_epochs": 10,
  "total_flos": 1.667595862278144e+16,
  "trial_name": null,
  "trial_params": null
}