trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.5912344455718994,
  "best_model_checkpoint": "tiny_bert_rand_20_v2_mrpc/checkpoint-30",
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 105,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 0.7157958149909973,
      "learning_rate": 4.9e-05,
      "loss": 0.6287,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7058823529411765,
      "eval_combined_score": 0.7623067353829749,
      "eval_f1": 0.8187311178247734,
      "eval_loss": 0.6033313870429993,
      "eval_runtime": 0.1374,
      "eval_samples_per_second": 2970.439,
      "eval_steps_per_second": 14.561,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1318848133087158,
      "learning_rate": 4.8e-05,
      "loss": 0.5917,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6936274509803921,
      "eval_combined_score": 0.7499145006839945,
      "eval_f1": 0.8062015503875969,
      "eval_loss": 0.5912344455718994,
      "eval_runtime": 0.1309,
      "eval_samples_per_second": 3117.538,
      "eval_steps_per_second": 15.282,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9104712605476379,
      "learning_rate": 4.7e-05,
      "loss": 0.5567,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6887254901960784,
      "eval_combined_score": 0.744988723188806,
      "eval_f1": 0.8012519561815337,
      "eval_loss": 0.6026718616485596,
      "eval_runtime": 0.1405,
      "eval_samples_per_second": 2904.554,
      "eval_steps_per_second": 14.238,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.6262362003326416,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.5147,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6764705882352942,
      "eval_combined_score": 0.7177973379132675,
      "eval_f1": 0.7591240875912408,
      "eval_loss": 0.6323015093803406,
      "eval_runtime": 0.139,
      "eval_samples_per_second": 2934.845,
      "eval_steps_per_second": 14.386,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.4019546508789062,
      "learning_rate": 4.5e-05,
      "loss": 0.4186,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6691176470588235,
      "eval_combined_score": 0.7120542863243301,
      "eval_f1": 0.7549909255898367,
      "eval_loss": 0.6770791411399841,
      "eval_runtime": 0.142,
      "eval_samples_per_second": 2873.284,
      "eval_steps_per_second": 14.085,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.541828155517578,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.3291,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6715686274509803,
      "eval_combined_score": 0.7121680775631285,
      "eval_f1": 0.7527675276752768,
      "eval_loss": 0.7956589460372925,
      "eval_runtime": 0.1311,
      "eval_samples_per_second": 3113.251,
      "eval_steps_per_second": 15.261,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 4.177097320556641,
      "learning_rate": 4.3e-05,
      "loss": 0.242,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6372549019607843,
      "eval_combined_score": 0.6774060769345907,
      "eval_f1": 0.7175572519083969,
      "eval_loss": 0.9224784970283508,
      "eval_runtime": 0.1516,
      "eval_samples_per_second": 2691.291,
      "eval_steps_per_second": 13.193,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 673316591603712.0,
      "train_loss": 0.4687800475529262,
      "train_runtime": 20.0012,
      "train_samples_per_second": 9169.437,
      "train_steps_per_second": 37.498
    }
  ],
  "logging_steps": 1,
  "max_steps": 750,
  "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": 673316591603712.0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": null
}