trainer_state.json

{
  "best_global_step": 410,
  "best_metric": 0.6495097875595093,
  "best_model_checkpoint": "tiny_bert_rand_100_v2_qnli/checkpoint-410",
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 2460,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 1.4908751249313354,
      "learning_rate": 4.9e-05,
      "loss": 0.666,
      "step": 410
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6168771737140765,
      "eval_loss": 0.6495097875595093,
      "eval_runtime": 1.6474,
      "eval_samples_per_second": 3316.126,
      "eval_steps_per_second": 13.354,
      "step": 410
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4664387702941895,
      "learning_rate": 4.8e-05,
      "loss": 0.6355,
      "step": 820
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6247483067911404,
      "eval_loss": 0.6537632942199707,
      "eval_runtime": 1.5717,
      "eval_samples_per_second": 3475.962,
      "eval_steps_per_second": 13.998,
      "step": 820
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.392573356628418,
      "learning_rate": 4.7e-05,
      "loss": 0.5919,
      "step": 1230
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6159619256818598,
      "eval_loss": 0.6659488677978516,
      "eval_runtime": 1.554,
      "eval_samples_per_second": 3515.369,
      "eval_steps_per_second": 14.157,
      "step": 1230
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.3592777252197266,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.531,
      "step": 1640
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6207212154493867,
      "eval_loss": 0.7202514410018921,
      "eval_runtime": 1.5784,
      "eval_samples_per_second": 3461.089,
      "eval_steps_per_second": 13.938,
      "step": 1640
    },
    {
      "epoch": 5.0,
      "grad_norm": 4.560001373291016,
      "learning_rate": 4.5e-05,
      "loss": 0.4603,
      "step": 2050
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6132161815852096,
      "eval_loss": 0.7936855554580688,
      "eval_runtime": 1.5551,
      "eval_samples_per_second": 3513.016,
      "eval_steps_per_second": 14.147,
      "step": 2050
    },
    {
      "epoch": 6.0,
      "grad_norm": 6.721935749053955,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.3924,
      "step": 2460
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6068094453596925,
      "eval_loss": 0.9373321533203125,
      "eval_runtime": 1.654,
      "eval_samples_per_second": 3302.864,
      "eval_steps_per_second": 13.301,
      "step": 2460
    },
    {
      "epoch": 6.0,
      "step": 2460,
      "total_flos": 1.6480417453111296e+16,
      "train_loss": 0.5461745967709921,
      "train_runtime": 341.9865,
      "train_samples_per_second": 15313.908,
      "train_steps_per_second": 59.944
    }
  ],
  "logging_steps": 1,
  "max_steps": 20500,
  "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": 1.6480417453111296e+16,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}