trainer_state.json

{
  "best_metric": 1.1405508518218994,
  "best_model_checkpoint": "saves/Mistral-7B-v0.2-Chat/lora/emma_assistant_v0.3/checkpoint-200",
  "epoch": 4.968152866242038,
  "eval_steps": 200,
  "global_step": 390,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.55,
      "grad_norm": 1.330633521080017,
      "learning_rate": 0.0004798670299452926,
      "loss": 0.9342,
      "step": 200
    },
    {
      "epoch": 2.55,
      "eval_loss": 1.1405508518218994,
      "eval_runtime": 10.6843,
      "eval_samples_per_second": 13.103,
      "eval_steps_per_second": 6.552,
      "step": 200
    },
    {
      "epoch": 4.97,
      "step": 390,
      "total_flos": 1.6822812952815206e+17,
      "train_loss": 0.6901797954852764,
      "train_runtime": 1500.9342,
      "train_samples_per_second": 4.181,
      "train_steps_per_second": 0.26
    }
  ],
  "logging_steps": 200,
  "max_steps": 390,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 200,
  "total_flos": 1.6822812952815206e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}