trainer_state.json

{
  "best_global_step": 3,
  "best_metric": 0.7153939008712769,
  "best_model_checkpoint": "tiny_bert_km_10_v1_wnli/checkpoint-3",
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 18,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.872547447681427,
      "learning_rate": 4.9e-05,
      "loss": 0.7073,
      "step": 3
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.39436619718309857,
      "eval_loss": 0.7153939008712769,
      "eval_runtime": 0.0302,
      "eval_samples_per_second": 2354.692,
      "eval_steps_per_second": 33.165,
      "step": 3
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.6551445722579956,
      "learning_rate": 4.8e-05,
      "loss": 0.6928,
      "step": 6
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.38028169014084506,
      "eval_loss": 0.7218859791755676,
      "eval_runtime": 0.0301,
      "eval_samples_per_second": 2361.433,
      "eval_steps_per_second": 33.26,
      "step": 6
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6504555940628052,
      "learning_rate": 4.7e-05,
      "loss": 0.6926,
      "step": 9
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.28169014084507044,
      "eval_loss": 0.7309087514877319,
      "eval_runtime": 0.0425,
      "eval_samples_per_second": 1668.874,
      "eval_steps_per_second": 23.505,
      "step": 9
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.662017285823822,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.69,
      "step": 12
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.22535211267605634,
      "eval_loss": 0.7406468987464905,
      "eval_runtime": 0.041,
      "eval_samples_per_second": 1731.652,
      "eval_steps_per_second": 24.389,
      "step": 12
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.8602695465087891,
      "learning_rate": 4.5e-05,
      "loss": 0.688,
      "step": 15
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.2112676056338028,
      "eval_loss": 0.7504951357841492,
      "eval_runtime": 0.0354,
      "eval_samples_per_second": 2003.657,
      "eval_steps_per_second": 28.221,
      "step": 15
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.7279769778251648,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.6865,
      "step": 18
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.18309859154929578,
      "eval_loss": 0.7618288993835449,
      "eval_runtime": 0.0312,
      "eval_samples_per_second": 2278.937,
      "eval_steps_per_second": 32.098,
      "step": 18
    },
    {
      "epoch": 6.0,
      "step": 18,
      "total_flos": 99911832606720.0,
      "train_loss": 0.692838642332289,
      "train_runtime": 6.8468,
      "train_samples_per_second": 4637.207,
      "train_steps_per_second": 21.908
    }
  ],
  "logging_steps": 1,
  "max_steps": 150,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 50,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 5
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 99911832606720.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}