trainer_state.json

{
  "best_global_step": 3,
  "best_metric": 0.7231789231300354,
  "best_model_checkpoint": "tiny_bert_km_5_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": 0.9462671875953674,
      "learning_rate": 4.9e-05,
      "loss": 0.7063,
      "step": 3
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.39436619718309857,
      "eval_loss": 0.7231789231300354,
      "eval_runtime": 0.0292,
      "eval_samples_per_second": 2435.099,
      "eval_steps_per_second": 34.297,
      "step": 3
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0501408576965332,
      "learning_rate": 4.8e-05,
      "loss": 0.702,
      "step": 6
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.4647887323943662,
      "eval_loss": 0.7265077233314514,
      "eval_runtime": 0.0349,
      "eval_samples_per_second": 2033.067,
      "eval_steps_per_second": 28.635,
      "step": 6
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8395664095878601,
      "learning_rate": 4.7e-05,
      "loss": 0.6959,
      "step": 9
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.29577464788732394,
      "eval_loss": 0.7417474985122681,
      "eval_runtime": 0.0347,
      "eval_samples_per_second": 2048.268,
      "eval_steps_per_second": 28.849,
      "step": 9
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.6460738182067871,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.6927,
      "step": 12
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.2535211267605634,
      "eval_loss": 0.7537961006164551,
      "eval_runtime": 0.0325,
      "eval_samples_per_second": 2181.573,
      "eval_steps_per_second": 30.726,
      "step": 12
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.7992274761199951,
      "learning_rate": 4.5e-05,
      "loss": 0.6946,
      "step": 15
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.22535211267605634,
      "eval_loss": 0.760233461856842,
      "eval_runtime": 0.0292,
      "eval_samples_per_second": 2427.497,
      "eval_steps_per_second": 34.19,
      "step": 15
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.7142523527145386,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.6829,
      "step": 18
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.2112676056338028,
      "eval_loss": 0.7708516120910645,
      "eval_runtime": 0.0295,
      "eval_samples_per_second": 2410.285,
      "eval_steps_per_second": 33.948,
      "step": 18
    },
    {
      "epoch": 6.0,
      "step": 18,
      "total_flos": 99911832606720.0,
      "train_loss": 0.695737878481547,
      "train_runtime": 6.5593,
      "train_samples_per_second": 4840.435,
      "train_steps_per_second": 22.868
    }
  ],
  "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
}