{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 32907,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "learning_rate": 1.9699152156076217e-05,
      "loss": 2.2617,
      "step": 500
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.939526544504209e-05,
      "loss": 1.242,
      "step": 1000
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9091378734007962e-05,
      "loss": 1.1706,
      "step": 1500
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8787492022973838e-05,
      "loss": 1.123,
      "step": 2000
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8484820858783847e-05,
      "loss": 1.1132,
      "step": 2500
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.818093414774972e-05,
      "loss": 1.0418,
      "step": 3000
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.7877655210137663e-05,
      "loss": 1.0142,
      "step": 3500
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.7573768499103536e-05,
      "loss": 1.0665,
      "step": 4000
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.726988178806941e-05,
      "loss": 0.9765,
      "step": 4500
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.696599507703528e-05,
      "loss": 0.9874,
      "step": 5000
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.6662108366001154e-05,
      "loss": 0.9705,
      "step": 5500
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.635822165496703e-05,
      "loss": 0.9824,
      "step": 6000
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.6054334943932903e-05,
      "loss": 0.939,
      "step": 6500
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.5750448232898775e-05,
      "loss": 0.9501,
      "step": 7000
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.5446561521864648e-05,
      "loss": 0.9427,
      "step": 7500
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.5142674810830524e-05,
      "loss": 0.9215,
      "step": 8000
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.4838788099796397e-05,
      "loss": 0.9106,
      "step": 8500
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.4534901388762271e-05,
      "loss": 0.9276,
      "step": 9000
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.4231014677728144e-05,
      "loss": 0.9249,
      "step": 9500
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.3927127966694018e-05,
      "loss": 0.9114,
      "step": 10000
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.3623849029081959e-05,
      "loss": 0.888,
      "step": 10500
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.3319962318047831e-05,
      "loss": 0.8897,
      "step": 11000
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.3016075607013707e-05,
      "loss": 0.765,
      "step": 11500
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.271218889597958e-05,
      "loss": 0.7409,
      "step": 12000
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.2408302184945453e-05,
      "loss": 0.7642,
      "step": 12500
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.2105023247333397e-05,
      "loss": 0.7327,
      "step": 13000
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.180113653629927e-05,
      "loss": 0.7358,
      "step": 13500
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.1497249825265142e-05,
      "loss": 0.7657,
      "step": 14000
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.1193363114231015e-05,
      "loss": 0.7393,
      "step": 14500
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.0889476403196887e-05,
      "loss": 0.7338,
      "step": 15000
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.0586197465584831e-05,
      "loss": 0.7693,
      "step": 15500
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0282310754550704e-05,
      "loss": 0.7357,
      "step": 16000
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.978424043516578e-06,
      "loss": 0.7303,
      "step": 16500
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.674537332482451e-06,
      "loss": 0.7295,
      "step": 17000
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.371258394870393e-06,
      "loss": 0.7393,
      "step": 17500
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.067371683836268e-06,
      "loss": 0.7345,
      "step": 18000
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.76348497280214e-06,
      "loss": 0.7507,
      "step": 18500
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.459598261768015e-06,
      "loss": 0.7195,
      "step": 19000
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.155711550733887e-06,
      "loss": 0.7368,
      "step": 19500
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.852432613121828e-06,
      "loss": 0.7317,
      "step": 20000
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.548545902087703e-06,
      "loss": 0.7191,
      "step": 20500
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.244659191053576e-06,
      "loss": 0.7102,
      "step": 21000
    },
    {
      "epoch": 1.96,
      "learning_rate": 6.94077248001945e-06,
      "loss": 0.7145,
      "step": 21500
    },
    {
      "epoch": 2.01,
      "learning_rate": 6.636885768985323e-06,
      "loss": 0.7201,
      "step": 22000
    },
    {
      "epoch": 2.05,
      "learning_rate": 6.3329990579511964e-06,
      "loss": 0.5908,
      "step": 22500
    },
    {
      "epoch": 2.1,
      "learning_rate": 6.029720120339138e-06,
      "loss": 0.5951,
      "step": 23000
    },
    {
      "epoch": 2.14,
      "learning_rate": 5.725833409305011e-06,
      "loss": 0.5854,
      "step": 23500
    },
    {
      "epoch": 2.19,
      "learning_rate": 5.421946698270885e-06,
      "loss": 0.5901,
      "step": 24000
    },
    {
      "epoch": 2.23,
      "learning_rate": 5.118059987236758e-06,
      "loss": 0.599,
      "step": 24500
    },
    {
      "epoch": 2.28,
      "learning_rate": 4.814173276202632e-06,
      "loss": 0.5736,
      "step": 25000
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.510894338590574e-06,
      "loss": 0.6043,
      "step": 25500
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.207007627556448e-06,
      "loss": 0.5786,
      "step": 26000
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.903728689944389e-06,
      "loss": 0.5959,
      "step": 26500
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.6004497523323308e-06,
      "loss": 0.5894,
      "step": 27000
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.2965630412982043e-06,
      "loss": 0.5782,
      "step": 27500
    },
    {
      "epoch": 2.55,
      "learning_rate": 2.992676330264078e-06,
      "loss": 0.5738,
      "step": 28000
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.6887896192299514e-06,
      "loss": 0.5848,
      "step": 28500
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.384902908195825e-06,
      "loss": 0.5807,
      "step": 29000
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.0810161971616984e-06,
      "loss": 0.5709,
      "step": 29500
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.7771294861275718e-06,
      "loss": 0.5729,
      "step": 30000
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.4738505485155135e-06,
      "loss": 0.5747,
      "step": 30500
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.169963837481387e-06,
      "loss": 0.5834,
      "step": 31000
    },
    {
      "epoch": 2.87,
      "learning_rate": 8.660771264472605e-07,
      "loss": 0.5872,
      "step": 31500
    },
    {
      "epoch": 2.92,
      "learning_rate": 5.62190415413134e-07,
      "loss": 0.5577,
      "step": 32000
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.5830370437900756e-07,
      "loss": 0.5637,
      "step": 32500
    }
  ],
  "max_steps": 32907,
  "num_train_epochs": 3,
  "total_flos": 6.878245175940096e+16,
  "trial_name": null,
  "trial_params": null
}