trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 3130,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.6,
      "learning_rate": 1.6805111821086264e-05,
      "loss": 0.1306,
      "step": 500
    },
    {
      "epoch": 3.19,
      "learning_rate": 1.3610223642172523e-05,
      "loss": 0.0555,
      "step": 1000
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.0415335463258786e-05,
      "loss": 0.0359,
      "step": 1500
    },
    {
      "epoch": 6.39,
      "learning_rate": 7.220447284345049e-06,
      "loss": 0.0242,
      "step": 2000
    },
    {
      "epoch": 7.99,
      "learning_rate": 4.02555910543131e-06,
      "loss": 0.0182,
      "step": 2500
    },
    {
      "epoch": 9.58,
      "learning_rate": 8.306709265175719e-07,
      "loss": 0.0153,
      "step": 3000
    },
    {
      "epoch": 10.0,
      "step": 3130,
      "total_flos": 2.63115773952e+16,
      "train_loss": 0.0452460962743424,
      "train_runtime": 2237.7872,
      "train_samples_per_second": 178.748,
      "train_steps_per_second": 1.399
    }
  ],
  "logging_steps": 500,
  "max_steps": 3130,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 2.63115773952e+16,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}