trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.6856385469436646,
  "best_model_checkpoint": "tiny_bert_km_5_v2_rte/checkpoint-30",
  "epoch": 8.0,
  "eval_steps": 500,
  "global_step": 80,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.6291049718856812,
      "learning_rate": 4.9e-05,
      "loss": 0.707,
      "step": 10
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5054151624548736,
      "eval_loss": 0.6930843591690063,
      "eval_runtime": 0.0996,
      "eval_samples_per_second": 2781.269,
      "eval_steps_per_second": 20.081,
      "step": 10
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7903363704681396,
      "learning_rate": 4.8e-05,
      "loss": 0.6895,
      "step": 20
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.5126353790613718,
      "eval_loss": 0.6901511549949646,
      "eval_runtime": 0.096,
      "eval_samples_per_second": 2885.705,
      "eval_steps_per_second": 20.835,
      "step": 20
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6098823547363281,
      "learning_rate": 4.7e-05,
      "loss": 0.6776,
      "step": 30
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.5776173285198556,
      "eval_loss": 0.6856385469436646,
      "eval_runtime": 0.1017,
      "eval_samples_per_second": 2724.5,
      "eval_steps_per_second": 19.671,
      "step": 30
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.7681831121444702,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.6656,
      "step": 40
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.5667870036101083,
      "eval_loss": 0.693310022354126,
      "eval_runtime": 0.1086,
      "eval_samples_per_second": 2551.7,
      "eval_steps_per_second": 18.424,
      "step": 40
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.828361988067627,
      "learning_rate": 4.5e-05,
      "loss": 0.6465,
      "step": 50
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.5451263537906137,
      "eval_loss": 0.6963560581207275,
      "eval_runtime": 0.0984,
      "eval_samples_per_second": 2815.331,
      "eval_steps_per_second": 20.327,
      "step": 50
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.966827929019928,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.6203,
      "step": 60
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.5487364620938628,
      "eval_loss": 0.7073309421539307,
      "eval_runtime": 0.1016,
      "eval_samples_per_second": 2727.282,
      "eval_steps_per_second": 19.692,
      "step": 60
    },
    {
      "epoch": 7.0,
      "grad_norm": 1.1733816862106323,
      "learning_rate": 4.3e-05,
      "loss": 0.5884,
      "step": 70
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.5776173285198556,
      "eval_loss": 0.7212883830070496,
      "eval_runtime": 0.1016,
      "eval_samples_per_second": 2726.226,
      "eval_steps_per_second": 19.684,
      "step": 70
    },
    {
      "epoch": 8.0,
      "grad_norm": 1.8930528163909912,
      "learning_rate": 4.2e-05,
      "loss": 0.5459,
      "step": 80
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.5415162454873647,
      "eval_loss": 0.7846221327781677,
      "eval_runtime": 0.1072,
      "eval_samples_per_second": 2583.906,
      "eval_steps_per_second": 18.656,
      "step": 80
    },
    {
      "epoch": 8.0,
      "step": 80,
      "total_flos": 522373675991040.0,
      "train_loss": 0.6426122128963471,
      "train_runtime": 16.5209,
      "train_samples_per_second": 7535.914,
      "train_steps_per_second": 30.265
    }
  ],
  "logging_steps": 1,
  "max_steps": 500,
  "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": 522373675991040.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}