trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.6038627028465271,
  "best_model_checkpoint": "tiny_bert_km_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": 0.7767913341522217,
      "learning_rate": 4.9e-05,
      "loss": 0.6326,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6862745098039216,
      "eval_combined_score": 0.7470411588058647,
      "eval_f1": 0.8078078078078078,
      "eval_loss": 0.6154976487159729,
      "eval_runtime": 0.1382,
      "eval_samples_per_second": 2952.385,
      "eval_steps_per_second": 14.472,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0868405103683472,
      "learning_rate": 4.8e-05,
      "loss": 0.5959,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7058823529411765,
      "eval_combined_score": 0.7620320855614974,
      "eval_f1": 0.8181818181818182,
      "eval_loss": 0.6038627028465271,
      "eval_runtime": 0.1291,
      "eval_samples_per_second": 3160.216,
      "eval_steps_per_second": 15.491,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0897839069366455,
      "learning_rate": 4.7e-05,
      "loss": 0.5662,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6936274509803921,
      "eval_combined_score": 0.752828763084181,
      "eval_f1": 0.8120300751879699,
      "eval_loss": 0.6089034676551819,
      "eval_runtime": 0.134,
      "eval_samples_per_second": 3045.322,
      "eval_steps_per_second": 14.928,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.4412758350372314,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.5395,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.7009803921568627,
      "eval_combined_score": 0.7545782463928967,
      "eval_f1": 0.8081761006289309,
      "eval_loss": 0.6144157648086548,
      "eval_runtime": 0.1327,
      "eval_samples_per_second": 3075.119,
      "eval_steps_per_second": 15.074,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.733062505722046,
      "learning_rate": 4.5e-05,
      "loss": 0.4794,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6838235294117647,
      "eval_combined_score": 0.7338715637008573,
      "eval_f1": 0.7839195979899497,
      "eval_loss": 0.6358864903450012,
      "eval_runtime": 0.1488,
      "eval_samples_per_second": 2742.447,
      "eval_steps_per_second": 13.443,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 2.0078938007354736,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.3903,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6593137254901961,
      "eval_combined_score": 0.7087872975277067,
      "eval_f1": 0.7582608695652174,
      "eval_loss": 0.7128481268882751,
      "eval_runtime": 0.1316,
      "eval_samples_per_second": 3099.453,
      "eval_steps_per_second": 15.193,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 3.0901219844818115,
      "learning_rate": 4.3e-05,
      "loss": 0.2926,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6568627450980392,
      "eval_combined_score": 0.7109817081194894,
      "eval_f1": 0.7651006711409396,
      "eval_loss": 0.8221036195755005,
      "eval_runtime": 0.1343,
      "eval_samples_per_second": 3037.732,
      "eval_steps_per_second": 14.891,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 673316591603712.0,
      "train_loss": 0.49947141919817245,
      "train_runtime": 19.5031,
      "train_samples_per_second": 9403.623,
      "train_steps_per_second": 38.455
    }
  ],
  "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
}