trainer_state.json

{
  "best_global_step": 3,
  "best_metric": 0.7130281925201416,
  "best_model_checkpoint": "tiny_bert_km_20_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": 0.9930610060691833,
      "learning_rate": 4.9e-05,
      "loss": 0.7045,
      "step": 3
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.39436619718309857,
      "eval_loss": 0.7130281925201416,
      "eval_runtime": 0.0425,
      "eval_samples_per_second": 1672.107,
      "eval_steps_per_second": 23.551,
      "step": 3
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7681521773338318,
      "learning_rate": 4.8e-05,
      "loss": 0.6946,
      "step": 6
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.4225352112676056,
      "eval_loss": 0.7229864597320557,
      "eval_runtime": 0.038,
      "eval_samples_per_second": 1867.42,
      "eval_steps_per_second": 26.302,
      "step": 6
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.7131951451301575,
      "learning_rate": 4.7e-05,
      "loss": 0.6906,
      "step": 9
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.2535211267605634,
      "eval_loss": 0.7394914627075195,
      "eval_runtime": 0.0354,
      "eval_samples_per_second": 2006.29,
      "eval_steps_per_second": 28.258,
      "step": 9
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.7863022089004517,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.6921,
      "step": 12
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.2112676056338028,
      "eval_loss": 0.746753990650177,
      "eval_runtime": 0.03,
      "eval_samples_per_second": 2364.452,
      "eval_steps_per_second": 33.302,
      "step": 12
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.774154007434845,
      "learning_rate": 4.5e-05,
      "loss": 0.6892,
      "step": 15
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.2535211267605634,
      "eval_loss": 0.7496422529220581,
      "eval_runtime": 0.0409,
      "eval_samples_per_second": 1737.755,
      "eval_steps_per_second": 24.475,
      "step": 15
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.7761738896369934,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.6906,
      "step": 18
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.2112676056338028,
      "eval_loss": 0.7533835768699646,
      "eval_runtime": 0.0389,
      "eval_samples_per_second": 1826.731,
      "eval_steps_per_second": 25.729,
      "step": 18
    },
    {
      "epoch": 6.0,
      "step": 18,
      "total_flos": 99911832606720.0,
      "train_loss": 0.6935964028040568,
      "train_runtime": 6.3381,
      "train_samples_per_second": 5009.357,
      "train_steps_per_second": 23.666
    }
  ],
  "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
}