trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 3171,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.47,
      "learning_rate": 4.211605171870073e-05,
      "loss": 0.8922,
      "step": 500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.423210343740145e-05,
      "loss": 0.7122,
      "step": 1000
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.634815515610218e-05,
      "loss": 0.6706,
      "step": 1500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8464206874802904e-05,
      "loss": 0.648,
      "step": 2000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0580258593503627e-05,
      "loss": 0.6318,
      "step": 2500
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.6963103122043523e-06,
      "loss": 0.6272,
      "step": 3000
    },
    {
      "epoch": 3.0,
      "step": 3171,
      "total_flos": 828180051001344.0,
      "train_loss": 0.6928272205209627,
      "train_runtime": 684.2817,
      "train_samples_per_second": 18.527,
      "train_steps_per_second": 4.634
    }
  ],
  "max_steps": 3171,
  "num_train_epochs": 3,
  "total_flos": 828180051001344.0,
  "trial_name": null,
  "trial_params": null
}