trainer_state.json

{
  "best_global_step": 3,
  "best_metric": 0.721610963344574,
  "best_model_checkpoint": "tiny_bert_km_50_v2_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": 1.0756258964538574,
      "learning_rate": 4.9e-05,
      "loss": 0.7082,
      "step": 3
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.39436619718309857,
      "eval_loss": 0.721610963344574,
      "eval_runtime": 0.0506,
      "eval_samples_per_second": 1403.901,
      "eval_steps_per_second": 19.773,
      "step": 3
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.6615771651268005,
      "learning_rate": 4.8e-05,
      "loss": 0.6925,
      "step": 6
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.36619718309859156,
      "eval_loss": 0.7240867018699646,
      "eval_runtime": 0.0309,
      "eval_samples_per_second": 2294.972,
      "eval_steps_per_second": 32.324,
      "step": 6
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.686205267906189,
      "learning_rate": 4.7e-05,
      "loss": 0.6924,
      "step": 9
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.39436619718309857,
      "eval_loss": 0.7375109791755676,
      "eval_runtime": 0.0326,
      "eval_samples_per_second": 2178.062,
      "eval_steps_per_second": 30.677,
      "step": 9
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.6912921667098999,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.6921,
      "step": 12
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.352112676056338,
      "eval_loss": 0.7405921220779419,
      "eval_runtime": 0.0334,
      "eval_samples_per_second": 2127.75,
      "eval_steps_per_second": 29.968,
      "step": 12
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.8052303194999695,
      "learning_rate": 4.5e-05,
      "loss": 0.6919,
      "step": 15
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.3380281690140845,
      "eval_loss": 0.7427927851676941,
      "eval_runtime": 0.0602,
      "eval_samples_per_second": 1179.786,
      "eval_steps_per_second": 16.617,
      "step": 15
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.7383120656013489,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.6904,
      "step": 18
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.323943661971831,
      "eval_loss": 0.7542914152145386,
      "eval_runtime": 0.0308,
      "eval_samples_per_second": 2308.475,
      "eval_steps_per_second": 32.514,
      "step": 18
    },
    {
      "epoch": 6.0,
      "step": 18,
      "total_flos": 99911832606720.0,
      "train_loss": 0.6945984231101142,
      "train_runtime": 6.7439,
      "train_samples_per_second": 4707.943,
      "train_steps_per_second": 22.242
    }
  ],
  "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
}