trainer_state.json

{
  "best_global_step": 3,
  "best_metric": 0.7005943655967712,
  "best_model_checkpoint": "tiny_bert_km_20_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": 1.0073267221450806,
      "learning_rate": 4.9e-05,
      "loss": 0.7048,
      "step": 3
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.49295774647887325,
      "eval_loss": 0.7005943655967712,
      "eval_runtime": 0.0315,
      "eval_samples_per_second": 2256.386,
      "eval_steps_per_second": 31.78,
      "step": 3
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7369723320007324,
      "learning_rate": 4.8e-05,
      "loss": 0.6988,
      "step": 6
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.5211267605633803,
      "eval_loss": 0.702409565448761,
      "eval_runtime": 0.0376,
      "eval_samples_per_second": 1889.338,
      "eval_steps_per_second": 26.61,
      "step": 6
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6635440587997437,
      "learning_rate": 4.7e-05,
      "loss": 0.6961,
      "step": 9
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.29577464788732394,
      "eval_loss": 0.7241968512535095,
      "eval_runtime": 0.0661,
      "eval_samples_per_second": 1074.38,
      "eval_steps_per_second": 15.132,
      "step": 9
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.6906999945640564,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.6903,
      "step": 12
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.30985915492957744,
      "eval_loss": 0.731184184551239,
      "eval_runtime": 0.0413,
      "eval_samples_per_second": 1720.884,
      "eval_steps_per_second": 24.238,
      "step": 12
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.9729248285293579,
      "learning_rate": 4.5e-05,
      "loss": 0.692,
      "step": 15
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.3380281690140845,
      "eval_loss": 0.7305235266685486,
      "eval_runtime": 0.0312,
      "eval_samples_per_second": 2274.238,
      "eval_steps_per_second": 32.032,
      "step": 15
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.7727306485176086,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.6949,
      "step": 18
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.2676056338028169,
      "eval_loss": 0.7371810078620911,
      "eval_runtime": 0.0301,
      "eval_samples_per_second": 2357.955,
      "eval_steps_per_second": 33.211,
      "step": 18
    },
    {
      "epoch": 6.0,
      "step": 18,
      "total_flos": 99911832606720.0,
      "train_loss": 0.696158250172933,
      "train_runtime": 6.7636,
      "train_samples_per_second": 4694.268,
      "train_steps_per_second": 22.178
    }
  ],
  "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
}