trainer_state.json

{
  "best_global_step": 410,
  "best_metric": 0.6490322351455688,
  "best_model_checkpoint": "tiny_bert_rand_20_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": 1.4358452558517456,
      "learning_rate": 4.9e-05,
      "loss": 0.6658,
      "step": 410
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6095551894563427,
      "eval_loss": 0.6490322351455688,
      "eval_runtime": 1.5996,
      "eval_samples_per_second": 3415.311,
      "eval_steps_per_second": 13.754,
      "step": 410
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6815091371536255,
      "learning_rate": 4.8e-05,
      "loss": 0.6359,
      "step": 820
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6220025626944902,
      "eval_loss": 0.652955174446106,
      "eval_runtime": 1.5565,
      "eval_samples_per_second": 3509.703,
      "eval_steps_per_second": 14.134,
      "step": 820
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.6009111404418945,
      "learning_rate": 4.7e-05,
      "loss": 0.5943,
      "step": 1230
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6337177375068643,
      "eval_loss": 0.652291476726532,
      "eval_runtime": 1.5671,
      "eval_samples_per_second": 3486.157,
      "eval_steps_per_second": 14.039,
      "step": 1230
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.9130334854125977,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.5397,
      "step": 1640
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6273110012813472,
      "eval_loss": 0.6960062384605408,
      "eval_runtime": 1.5872,
      "eval_samples_per_second": 3441.862,
      "eval_steps_per_second": 13.861,
      "step": 1640
    },
    {
      "epoch": 5.0,
      "grad_norm": 4.698758602142334,
      "learning_rate": 4.5e-05,
      "loss": 0.4759,
      "step": 2050
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6236500091524804,
      "eval_loss": 0.7842954397201538,
      "eval_runtime": 1.5593,
      "eval_samples_per_second": 3503.478,
      "eval_steps_per_second": 14.109,
      "step": 2050
    },
    {
      "epoch": 6.0,
      "grad_norm": 5.416735649108887,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.4074,
      "step": 2460
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6183415705656233,
      "eval_loss": 0.8884106874465942,
      "eval_runtime": 1.5671,
      "eval_samples_per_second": 3486.166,
      "eval_steps_per_second": 14.039,
      "step": 2460
    },
    {
      "epoch": 6.0,
      "step": 2460,
      "total_flos": 1.6480417453111296e+16,
      "train_loss": 0.5531651256530266,
      "train_runtime": 341.2075,
      "train_samples_per_second": 15348.873,
      "train_steps_per_second": 60.081
    }
  ],
  "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
}