trainer_state.json

{
  "best_metric": 0.5116574764251709,
  "best_model_checkpoint": "/datastor1/fcyin/finetuned_checkpoints/mquake/llama2_7B_mquake_lofit_seed42_96heads/checkpoint-51",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 85,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.8741050362586975,
      "learning_rate": 0.004,
      "loss": 0.6931,
      "step": 17
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.5563876628875732,
      "eval_runtime": 4.2921,
      "eval_samples_per_second": 22.134,
      "eval_steps_per_second": 0.699,
      "step": 17
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8767452836036682,
      "learning_rate": 0.003,
      "loss": 0.1286,
      "step": 34
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.5252991318702698,
      "eval_runtime": 4.3599,
      "eval_samples_per_second": 21.789,
      "eval_steps_per_second": 0.688,
      "step": 34
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.04879095405340195,
      "learning_rate": 0.002,
      "loss": 0.0505,
      "step": 51
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.5116574764251709,
      "eval_runtime": 4.368,
      "eval_samples_per_second": 21.749,
      "eval_steps_per_second": 0.687,
      "step": 51
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.08225993812084198,
      "learning_rate": 0.001,
      "loss": 0.0177,
      "step": 68
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.5452096462249756,
      "eval_runtime": 4.3007,
      "eval_samples_per_second": 22.089,
      "eval_steps_per_second": 0.698,
      "step": 68
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.09921150654554367,
      "learning_rate": 0.0,
      "loss": 0.0107,
      "step": 85
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.5571908354759216,
      "eval_runtime": 4.3773,
      "eval_samples_per_second": 21.703,
      "eval_steps_per_second": 0.685,
      "step": 85
    }
  ],
  "logging_steps": 500,
  "max_steps": 85,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1420900423434240.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}