trainer_state.json

{
  "best_global_step": 3,
  "best_metric": 0.7400694489479065,
  "best_model_checkpoint": "bert_base_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": 1.8408938646316528,
      "learning_rate": 4.9e-05,
      "loss": 0.7427,
      "step": 3
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.323943661971831,
      "eval_loss": 0.7400694489479065,
      "eval_runtime": 0.0612,
      "eval_samples_per_second": 1159.929,
      "eval_steps_per_second": 16.337,
      "step": 3
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0740141868591309,
      "learning_rate": 4.8e-05,
      "loss": 0.7038,
      "step": 6
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.30985915492957744,
      "eval_loss": 0.7573173642158508,
      "eval_runtime": 0.0579,
      "eval_samples_per_second": 1226.521,
      "eval_steps_per_second": 17.275,
      "step": 6
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.169411540031433,
      "learning_rate": 4.7e-05,
      "loss": 0.6955,
      "step": 9
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.23943661971830985,
      "eval_loss": 0.7737677097320557,
      "eval_runtime": 0.0599,
      "eval_samples_per_second": 1185.884,
      "eval_steps_per_second": 16.703,
      "step": 9
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.5837174654006958,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.6899,
      "step": 12
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.352112676056338,
      "eval_loss": 0.7934638857841492,
      "eval_runtime": 0.0581,
      "eval_samples_per_second": 1221.395,
      "eval_steps_per_second": 17.203,
      "step": 12
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.229299545288086,
      "learning_rate": 4.5e-05,
      "loss": 0.6767,
      "step": 15
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.14084507042253522,
      "eval_loss": 0.8292252421379089,
      "eval_runtime": 0.0603,
      "eval_samples_per_second": 1177.663,
      "eval_steps_per_second": 16.587,
      "step": 15
    },
    {
      "epoch": 6.0,
      "grad_norm": 1.406445860862732,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.6828,
      "step": 18
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.2112676056338028,
      "eval_loss": 0.8702685236930847,
      "eval_runtime": 0.0555,
      "eval_samples_per_second": 1278.592,
      "eval_steps_per_second": 18.008,
      "step": 18
    },
    {
      "epoch": 6.0,
      "step": 18,
      "total_flos": 501226560460800.0,
      "train_loss": 0.6985566218694051,
      "train_runtime": 19.6732,
      "train_samples_per_second": 1613.874,
      "train_steps_per_second": 7.625
    }
  ],
  "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": 501226560460800.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}