trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.995610184372256,
  "global_step": 2845,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.999297012302285e-05,
      "loss": 2.1598,
      "step": 1
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.648506151142355e-05,
      "loss": 1.9142,
      "step": 500
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.29701230228471e-05,
      "loss": 1.3053,
      "step": 1000
    },
    {
      "epoch": 2.63,
      "learning_rate": 9.455184534270651e-06,
      "loss": 0.7065,
      "step": 1500
    },
    {
      "epoch": 3.51,
      "learning_rate": 5.9402460456942015e-06,
      "loss": 0.3021,
      "step": 2000
    },
    {
      "epoch": 4.39,
      "learning_rate": 2.425307557117751e-06,
      "loss": 0.1154,
      "step": 2500
    },
    {
      "epoch": 5.0,
      "step": 2845,
      "total_flos": 8.057225483177165e+17,
      "train_loss": 0.7717322093950214,
      "train_runtime": 11631.3607,
      "train_samples_per_second": 1.958,
      "train_steps_per_second": 0.245
    }
  ],
  "max_steps": 2845,
  "num_train_epochs": 5,
  "total_flos": 8.057225483177165e+17,
  "trial_name": null,
  "trial_params": null
}