trainer_state.json

{
  "best_global_step": 410,
  "best_metric": 0.6520935893058777,
  "best_model_checkpoint": "tiny_bert_rand_50_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": 1.406223177909851,
      "learning_rate": 4.9e-05,
      "loss": 0.6655,
      "step": 410
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6124839831594362,
      "eval_loss": 0.6520935893058777,
      "eval_runtime": 1.6245,
      "eval_samples_per_second": 3362.889,
      "eval_steps_per_second": 13.543,
      "step": 410
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5263471603393555,
      "learning_rate": 4.8e-05,
      "loss": 0.6359,
      "step": 820
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6188907193849533,
      "eval_loss": 0.6523372530937195,
      "eval_runtime": 1.5602,
      "eval_samples_per_second": 3501.394,
      "eval_steps_per_second": 14.1,
      "step": 820
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.781379461288452,
      "learning_rate": 4.7e-05,
      "loss": 0.5935,
      "step": 1230
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6203551162365001,
      "eval_loss": 0.6675785779953003,
      "eval_runtime": 1.5826,
      "eval_samples_per_second": 3451.962,
      "eval_steps_per_second": 13.901,
      "step": 1230
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.04308819770813,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.5338,
      "step": 1640
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6214534138751602,
      "eval_loss": 0.7060788869857788,
      "eval_runtime": 1.5593,
      "eval_samples_per_second": 3503.427,
      "eval_steps_per_second": 14.109,
      "step": 1640
    },
    {
      "epoch": 5.0,
      "grad_norm": 5.421135425567627,
      "learning_rate": 4.5e-05,
      "loss": 0.4667,
      "step": 2050
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6157788760754165,
      "eval_loss": 0.7881438732147217,
      "eval_runtime": 1.5868,
      "eval_samples_per_second": 3442.677,
      "eval_steps_per_second": 13.864,
      "step": 2050
    },
    {
      "epoch": 6.0,
      "grad_norm": 4.98477029800415,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.3973,
      "step": 2460
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6110195863078894,
      "eval_loss": 0.9105799794197083,
      "eval_runtime": 1.5608,
      "eval_samples_per_second": 3500.038,
      "eval_steps_per_second": 14.095,
      "step": 2460
    },
    {
      "epoch": 6.0,
      "step": 2460,
      "total_flos": 1.6480417453111296e+16,
      "train_loss": 0.5487746634134432,
      "train_runtime": 343.7213,
      "train_samples_per_second": 15236.618,
      "train_steps_per_second": 59.641
    }
  ],
  "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
}