trainer_state.json

{
  "best_global_step": 3068,
  "best_metric": 1.0970898866653442,
  "best_model_checkpoint": "bert_base_train_book_ent_15p_s_init_mnli/checkpoint-3068",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 10738,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.9240047335624695,
      "learning_rate": 4.90006518904824e-05,
      "loss": 1.1034,
      "step": 1534
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.31818644931227713,
      "eval_loss": 1.1021188497543335,
      "eval_runtime": 8.5318,
      "eval_samples_per_second": 1150.405,
      "eval_steps_per_second": 4.571,
      "step": 1534
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.9312124252319336,
      "learning_rate": 4.80006518904824e-05,
      "loss": 1.1019,
      "step": 3068
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.3544574630667346,
      "eval_loss": 1.0970898866653442,
      "eval_runtime": 8.3075,
      "eval_samples_per_second": 1181.46,
      "eval_steps_per_second": 4.695,
      "step": 3068
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6418417692184448,
      "learning_rate": 4.70006518904824e-05,
      "loss": 1.1011,
      "step": 4602
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.3273560876209883,
      "eval_loss": 1.0990735292434692,
      "eval_runtime": 8.3858,
      "eval_samples_per_second": 1170.426,
      "eval_steps_per_second": 4.651,
      "step": 4602
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.7767936587333679,
      "learning_rate": 4.60006518904824e-05,
      "loss": 1.1007,
      "step": 6136
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.3544574630667346,
      "eval_loss": 1.0989652872085571,
      "eval_runtime": 8.4399,
      "eval_samples_per_second": 1162.935,
      "eval_steps_per_second": 4.621,
      "step": 6136
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.0293899774551392,
      "learning_rate": 4.50006518904824e-05,
      "loss": 1.1004,
      "step": 7670
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.3534386143657667,
      "eval_loss": 1.0988004207611084,
      "eval_runtime": 8.324,
      "eval_samples_per_second": 1179.119,
      "eval_steps_per_second": 4.685,
      "step": 7670
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.4480030834674835,
      "learning_rate": 4.40006518904824e-05,
      "loss": 1.1,
      "step": 9204
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.31818644931227713,
      "eval_loss": 1.0991355180740356,
      "eval_runtime": 8.3221,
      "eval_samples_per_second": 1179.385,
      "eval_steps_per_second": 4.686,
      "step": 9204
    },
    {
      "epoch": 7.0,
      "grad_norm": 1.103432059288025,
      "learning_rate": 4.30006518904824e-05,
      "loss": 1.0997,
      "step": 10738
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.31818644931227713,
      "eval_loss": 1.1003295183181763,
      "eval_runtime": 8.2999,
      "eval_samples_per_second": 1182.549,
      "eval_steps_per_second": 4.699,
      "step": 10738
    },
    {
      "epoch": 7.0,
      "step": 10738,
      "total_flos": 3.616380787901737e+17,
      "train_loss": 1.1010421607986822,
      "train_runtime": 4910.0003,
      "train_samples_per_second": 3999.002,
      "train_steps_per_second": 15.621
    }
  ],
  "logging_steps": 1,
  "max_steps": 76700,
  "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": 3.616380787901737e+17,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}