{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 40.0,
  "global_step": 31880,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.63,
      "learning_rate": 2.976e-05,
      "loss": 7.9128,
      "step": 500
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9525812619502866e-05,
      "loss": 3.5609,
      "step": 1000
    },
    {
      "epoch": 1.88,
      "learning_rate": 2.9048757170172085e-05,
      "loss": 3.3615,
      "step": 1500
    },
    {
      "epoch": 2.51,
      "learning_rate": 2.857074569789675e-05,
      "loss": 2.8318,
      "step": 2000
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.8093690248565966e-05,
      "loss": 1.7371,
      "step": 2500
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.7615678776290633e-05,
      "loss": 1.3703,
      "step": 3000
    },
    {
      "epoch": 4.39,
      "learning_rate": 2.71395793499044e-05,
      "loss": 1.1862,
      "step": 3500
    },
    {
      "epoch": 5.02,
      "learning_rate": 2.6661567877629064e-05,
      "loss": 1.0873,
      "step": 4000
    },
    {
      "epoch": 5.65,
      "learning_rate": 2.618451242829828e-05,
      "loss": 0.981,
      "step": 4500
    },
    {
      "epoch": 6.27,
      "learning_rate": 2.5707456978967497e-05,
      "loss": 0.9185,
      "step": 5000
    },
    {
      "epoch": 6.9,
      "learning_rate": 2.522944550669216e-05,
      "loss": 0.8622,
      "step": 5500
    },
    {
      "epoch": 7.53,
      "learning_rate": 2.4752390057361376e-05,
      "loss": 0.8181,
      "step": 6000
    },
    {
      "epoch": 8.16,
      "learning_rate": 2.4274378585086042e-05,
      "loss": 0.7863,
      "step": 6500
    },
    {
      "epoch": 8.78,
      "learning_rate": 2.3797323135755258e-05,
      "loss": 0.7409,
      "step": 7000
    },
    {
      "epoch": 9.41,
      "learning_rate": 2.3319311663479924e-05,
      "loss": 0.7223,
      "step": 7500
    },
    {
      "epoch": 10.04,
      "learning_rate": 2.284225621414914e-05,
      "loss": 0.7005,
      "step": 8000
    },
    {
      "epoch": 10.66,
      "learning_rate": 2.2365200764818355e-05,
      "loss": 0.6662,
      "step": 8500
    },
    {
      "epoch": 11.29,
      "learning_rate": 2.188718929254302e-05,
      "loss": 0.6458,
      "step": 9000
    },
    {
      "epoch": 11.92,
      "learning_rate": 2.1410133843212236e-05,
      "loss": 0.6246,
      "step": 9500
    },
    {
      "epoch": 12.55,
      "learning_rate": 2.093307839388145e-05,
      "loss": 0.5987,
      "step": 10000
    },
    {
      "epoch": 13.17,
      "learning_rate": 2.0455066921606118e-05,
      "loss": 0.5974,
      "step": 10500
    },
    {
      "epoch": 13.8,
      "learning_rate": 1.9978011472275333e-05,
      "loss": 0.5831,
      "step": 11000
    },
    {
      "epoch": 14.43,
      "learning_rate": 1.9500956022944552e-05,
      "loss": 0.5549,
      "step": 11500
    },
    {
      "epoch": 15.06,
      "learning_rate": 1.902294455066922e-05,
      "loss": 0.561,
      "step": 12000
    },
    {
      "epoch": 15.68,
      "learning_rate": 1.8545889101338434e-05,
      "loss": 0.5462,
      "step": 12500
    },
    {
      "epoch": 16.31,
      "learning_rate": 1.806883365200765e-05,
      "loss": 0.5491,
      "step": 13000
    },
    {
      "epoch": 16.94,
      "learning_rate": 1.7590822179732315e-05,
      "loss": 0.5594,
      "step": 13500
    },
    {
      "epoch": 17.57,
      "learning_rate": 1.7112810707456982e-05,
      "loss": 0.5565,
      "step": 14000
    },
    {
      "epoch": 18.19,
      "learning_rate": 1.6636711281070746e-05,
      "loss": 0.5745,
      "step": 14500
    },
    {
      "epoch": 18.82,
      "learning_rate": 1.6158699808795412e-05,
      "loss": 0.6096,
      "step": 15000
    },
    {
      "epoch": 19.45,
      "learning_rate": 1.568068833652008e-05,
      "loss": 0.67,
      "step": 15500
    },
    {
      "epoch": 20.08,
      "learning_rate": 1.5203632887189294e-05,
      "loss": 0.6566,
      "step": 16000
    },
    {
      "epoch": 20.7,
      "learning_rate": 1.4725621414913959e-05,
      "loss": 0.6627,
      "step": 16500
    },
    {
      "epoch": 21.33,
      "learning_rate": 1.4249521988527725e-05,
      "loss": 0.6647,
      "step": 17000
    },
    {
      "epoch": 21.96,
      "learning_rate": 1.377151051625239e-05,
      "loss": 0.9268,
      "step": 17500
    },
    {
      "epoch": 22.58,
      "learning_rate": 1.3293499043977056e-05,
      "loss": 0.9965,
      "step": 18000
    },
    {
      "epoch": 23.21,
      "learning_rate": 1.2816443594646271e-05,
      "loss": 0.9737,
      "step": 18500
    },
    {
      "epoch": 23.84,
      "learning_rate": 1.2339388145315486e-05,
      "loss": 1.4449,
      "step": 19000
    },
    {
      "epoch": 24.47,
      "learning_rate": 1.1861376673040153e-05,
      "loss": 2.0548,
      "step": 19500
    },
    {
      "epoch": 25.09,
      "learning_rate": 1.138432122370937e-05,
      "loss": 2.4726,
      "step": 20000
    },
    {
      "epoch": 25.72,
      "learning_rate": 1.0906309751434036e-05,
      "loss": 2.7061,
      "step": 20500
    },
    {
      "epoch": 26.35,
      "learning_rate": 1.0430210325047802e-05,
      "loss": 2.9211,
      "step": 21000
    },
    {
      "epoch": 26.98,
      "learning_rate": 9.952198852772467e-06,
      "loss": 2.9367,
      "step": 21500
    },
    {
      "epoch": 27.6,
      "learning_rate": 9.474187380497133e-06,
      "loss": 2.9015,
      "step": 22000
    },
    {
      "epoch": 28.23,
      "learning_rate": 8.997131931166348e-06,
      "loss": 2.9781,
      "step": 22500
    },
    {
      "epoch": 28.86,
      "learning_rate": 8.520076481835564e-06,
      "loss": 3.0789,
      "step": 23000
    },
    {
      "epoch": 29.49,
      "learning_rate": 8.042065009560229e-06,
      "loss": 3.0514,
      "step": 23500
    },
    {
      "epoch": 30.11,
      "learning_rate": 7.565009560229446e-06,
      "loss": 3.0248,
      "step": 24000
    },
    {
      "epoch": 30.74,
      "learning_rate": 7.087954110898662e-06,
      "loss": 2.9983,
      "step": 24500
    },
    {
      "epoch": 31.37,
      "learning_rate": 6.609942638623327e-06,
      "loss": 2.9716,
      "step": 25000
    },
    {
      "epoch": 31.99,
      "learning_rate": 6.1328871892925425e-06,
      "loss": 2.9549,
      "step": 25500
    },
    {
      "epoch": 32.62,
      "learning_rate": 5.6558317399617595e-06,
      "loss": 2.9551,
      "step": 26000
    },
    {
      "epoch": 33.25,
      "learning_rate": 5.177820267686425e-06,
      "loss": 2.9571,
      "step": 26500
    },
    {
      "epoch": 33.88,
      "learning_rate": 4.70076481835564e-06,
      "loss": 2.9589,
      "step": 27000
    },
    {
      "epoch": 34.5,
      "learning_rate": 4.2237093690248565e-06,
      "loss": 2.9573,
      "step": 27500
    },
    {
      "epoch": 35.13,
      "learning_rate": 3.7456978967495225e-06,
      "loss": 2.9556,
      "step": 28000
    },
    {
      "epoch": 35.76,
      "learning_rate": 3.268642447418738e-06,
      "loss": 2.9545,
      "step": 28500
    },
    {
      "epoch": 36.39,
      "learning_rate": 2.7906309751434034e-06,
      "loss": 2.9585,
      "step": 29000
    },
    {
      "epoch": 37.01,
      "learning_rate": 2.3135755258126195e-06,
      "loss": 2.9558,
      "step": 29500
    },
    {
      "epoch": 37.64,
      "learning_rate": 1.8355640535372848e-06,
      "loss": 2.9561,
      "step": 30000
    },
    {
      "epoch": 38.27,
      "learning_rate": 1.358508604206501e-06,
      "loss": 2.9597,
      "step": 30500
    },
    {
      "epoch": 38.9,
      "learning_rate": 8.81453154875717e-07,
      "loss": 2.957,
      "step": 31000
    },
    {
      "epoch": 39.52,
      "learning_rate": 4.0344168260038244e-07,
      "loss": 2.9569,
      "step": 31500
    },
    {
      "epoch": 40.0,
      "step": 31880,
      "total_flos": 2.3797340595067814e+19,
      "train_loss": 1.875457509039636,
      "train_runtime": 14726.9257,
      "train_samples_per_second": 34.625,
      "train_steps_per_second": 2.165
    }
  ],
  "max_steps": 31880,
  "num_train_epochs": 40,
  "total_flos": 2.3797340595067814e+19,
  "trial_name": null,
  "trial_params": null
}