{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 29745,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "learning_rate": 4.9159522608841824e-05,
      "loss": 1.6853,
      "step": 500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8319045217683645e-05,
      "loss": 1.2089,
      "step": 1000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7478567826525466e-05,
      "loss": 1.0874,
      "step": 1500
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.6638090435367294e-05,
      "loss": 1.0041,
      "step": 2000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.5797613044209115e-05,
      "loss": 0.9324,
      "step": 2500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.4957135653050936e-05,
      "loss": 0.901,
      "step": 3000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.411665826189276e-05,
      "loss": 0.8671,
      "step": 3500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.327618087073458e-05,
      "loss": 0.8286,
      "step": 4000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.24357034795764e-05,
      "loss": 0.8044,
      "step": 4500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.159522608841822e-05,
      "loss": 0.7759,
      "step": 5000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.075474869726005e-05,
      "loss": 0.7616,
      "step": 5500
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.991427130610187e-05,
      "loss": 0.7501,
      "step": 6000
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.907379391494369e-05,
      "loss": 0.7303,
      "step": 6500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.823331652378551e-05,
      "loss": 0.7281,
      "step": 7000
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.7392839132627334e-05,
      "loss": 0.6958,
      "step": 7500
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.6552361741469155e-05,
      "loss": 0.7032,
      "step": 8000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.5711884350310976e-05,
      "loss": 0.6656,
      "step": 8500
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.4871406959152804e-05,
      "loss": 0.6875,
      "step": 9000
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.4030929567994625e-05,
      "loss": 0.655,
      "step": 9500
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3190452176836446e-05,
      "loss": 0.6567,
      "step": 10000
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.234997478567827e-05,
      "loss": 0.6277,
      "step": 10500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.150949739452009e-05,
      "loss": 0.6331,
      "step": 11000
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.066902000336191e-05,
      "loss": 0.6342,
      "step": 11500
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.9828542612203734e-05,
      "loss": 0.6065,
      "step": 12000
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.898806522104556e-05,
      "loss": 0.6116,
      "step": 12500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.814758782988738e-05,
      "loss": 0.611,
      "step": 13000
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.73071104387292e-05,
      "loss": 0.6014,
      "step": 13500
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6466633047571022e-05,
      "loss": 0.5935,
      "step": 14000
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5626155656412843e-05,
      "loss": 0.5846,
      "step": 14500
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4785678265254668e-05,
      "loss": 0.5844,
      "step": 15000
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.394520087409649e-05,
      "loss": 0.5863,
      "step": 15500
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.310472348293831e-05,
      "loss": 0.5635,
      "step": 16000
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.226424609178013e-05,
      "loss": 0.5627,
      "step": 16500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1423768700621956e-05,
      "loss": 0.568,
      "step": 17000
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.0583291309463777e-05,
      "loss": 0.5658,
      "step": 17500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.97428139183056e-05,
      "loss": 0.5505,
      "step": 18000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.890233652714742e-05,
      "loss": 0.5476,
      "step": 18500
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8061859135989244e-05,
      "loss": 0.5423,
      "step": 19000
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.7221381744831065e-05,
      "loss": 0.5451,
      "step": 19500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.6380904353672887e-05,
      "loss": 0.5473,
      "step": 20000
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.5540426962514708e-05,
      "loss": 0.5387,
      "step": 20500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.469994957135653e-05,
      "loss": 0.5282,
      "step": 21000
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3859472180198355e-05,
      "loss": 0.5138,
      "step": 21500
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.3018994789040176e-05,
      "loss": 0.5181,
      "step": 22000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2178517397881997e-05,
      "loss": 0.5124,
      "step": 22500
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.133804000672382e-05,
      "loss": 0.507,
      "step": 23000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0497562615565642e-05,
      "loss": 0.5033,
      "step": 23500
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.657085224407464e-06,
      "loss": 0.5061,
      "step": 24000
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.816607833249287e-06,
      "loss": 0.4931,
      "step": 24500
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.976130442091108e-06,
      "loss": 0.4979,
      "step": 25000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.1356530509329304e-06,
      "loss": 0.4922,
      "step": 25500
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.295175659774752e-06,
      "loss": 0.4972,
      "step": 26000
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.4546982686165745e-06,
      "loss": 0.4821,
      "step": 26500
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.6142208774583965e-06,
      "loss": 0.488,
      "step": 27000
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.7737434863002185e-06,
      "loss": 0.4838,
      "step": 27500
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.933266095142041e-06,
      "loss": 0.4894,
      "step": 28000
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.092788703983863e-06,
      "loss": 0.4894,
      "step": 28500
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.252311312825685e-06,
      "loss": 0.486,
      "step": 29000
    },
    {
      "epoch": 2.98,
      "learning_rate": 4.1183392166750716e-07,
      "loss": 0.4803,
      "step": 29500
    },
    {
      "epoch": 3.0,
      "step": 29745,
      "total_flos": 2.7743143438424474e+17,
      "train_loss": 0.6477808796030912,
      "train_runtime": 32967.4964,
      "train_samples_per_second": 9.023,
      "train_steps_per_second": 0.902
    }
  ],
  "max_steps": 29745,
  "num_train_epochs": 3,
  "total_flos": 2.7743143438424474e+17,
  "trial_name": null,
  "trial_params": null
}