trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.5925148725509644,
  "best_model_checkpoint": "tiny_bert_rand_10_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": 0.7123093008995056,
      "learning_rate": 4.9e-05,
      "loss": 0.633,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6936274509803921,
      "eval_combined_score": 0.7493098253341899,
      "eval_f1": 0.8049921996879875,
      "eval_loss": 0.6069743037223816,
      "eval_runtime": 0.2253,
      "eval_samples_per_second": 1810.799,
      "eval_steps_per_second": 8.876,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2349159717559814,
      "learning_rate": 4.8e-05,
      "loss": 0.5928,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6936274509803921,
      "eval_combined_score": 0.7493098253341899,
      "eval_f1": 0.8049921996879875,
      "eval_loss": 0.5925148725509644,
      "eval_runtime": 0.1646,
      "eval_samples_per_second": 2479.406,
      "eval_steps_per_second": 12.154,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9224690198898315,
      "learning_rate": 4.7e-05,
      "loss": 0.5569,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6985294117647058,
      "eval_combined_score": 0.7553715761113606,
      "eval_f1": 0.8122137404580153,
      "eval_loss": 0.6044036746025085,
      "eval_runtime": 0.1362,
      "eval_samples_per_second": 2995.695,
      "eval_steps_per_second": 14.685,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.914901852607727,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.5237,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.678921568627451,
      "eval_combined_score": 0.7210159380207779,
      "eval_f1": 0.7631103074141049,
      "eval_loss": 0.6260837316513062,
      "eval_runtime": 0.1534,
      "eval_samples_per_second": 2659.28,
      "eval_steps_per_second": 13.036,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 4.1404900550842285,
      "learning_rate": 4.5e-05,
      "loss": 0.4462,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6544117647058824,
      "eval_combined_score": 0.6964081086980247,
      "eval_f1": 0.738404452690167,
      "eval_loss": 0.6786839962005615,
      "eval_runtime": 0.1384,
      "eval_samples_per_second": 2947.056,
      "eval_steps_per_second": 14.446,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.376439332962036,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.3558,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6544117647058824,
      "eval_combined_score": 0.7001788553259142,
      "eval_f1": 0.745945945945946,
      "eval_loss": 0.7894179224967957,
      "eval_runtime": 0.1647,
      "eval_samples_per_second": 2477.03,
      "eval_steps_per_second": 12.142,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 4.353763103485107,
      "learning_rate": 4.3e-05,
      "loss": 0.26,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6568627450980392,
      "eval_combined_score": 0.7073241061130334,
      "eval_f1": 0.7577854671280276,
      "eval_loss": 0.9015735387802124,
      "eval_runtime": 0.1697,
      "eval_samples_per_second": 2404.916,
      "eval_steps_per_second": 11.789,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 673316591603712.0,
      "train_loss": 0.4811921028863816,
      "train_runtime": 20.9726,
      "train_samples_per_second": 8744.756,
      "train_steps_per_second": 35.761
    }
  ],
  "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
}