trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.0,
  "global_step": 3486,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "learning_rate": 4.2828456683878374e-05,
      "loss": 2.2005,
      "step": 500
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.5656913367756745e-05,
      "loss": 1.8288,
      "step": 1000
    },
    {
      "epoch": 3.01,
      "learning_rate": 2.8485370051635113e-05,
      "loss": 1.6132,
      "step": 1500
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.1313826735513484e-05,
      "loss": 1.4468,
      "step": 2000
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.4142283419391852e-05,
      "loss": 1.3287,
      "step": 2500
    },
    {
      "epoch": 6.02,
      "learning_rate": 6.970740103270223e-06,
      "loss": 1.2313,
      "step": 3000
    },
    {
      "epoch": 7.0,
      "step": 3486,
      "total_flos": 4463221452318720.0,
      "train_loss": 1.5467751831825067,
      "train_runtime": 653.0728,
      "train_samples_per_second": 170.757,
      "train_steps_per_second": 5.338
    }
  ],
  "max_steps": 3486,
  "num_train_epochs": 7,
  "total_flos": 4463221452318720.0,
  "trial_name": null,
  "trial_params": null
}