trainer_state.json

{
  "best_global_step": 30,
  "best_metric": 0.5896283388137817,
  "best_model_checkpoint": "tiny_bert_rand_10_v1_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": 0.7444829940795898,
      "learning_rate": 4.9e-05,
      "loss": 0.635,
      "step": 15
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6862745098039216,
      "eval_combined_score": 0.7437583729143831,
      "eval_f1": 0.8012422360248447,
      "eval_loss": 0.6054209470748901,
      "eval_runtime": 0.1431,
      "eval_samples_per_second": 2851.932,
      "eval_steps_per_second": 13.98,
      "step": 15
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3274710178375244,
      "learning_rate": 4.8e-05,
      "loss": 0.5924,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6936274509803921,
      "eval_combined_score": 0.7490046488078799,
      "eval_f1": 0.8043818466353677,
      "eval_loss": 0.5896283388137817,
      "eval_runtime": 0.1421,
      "eval_samples_per_second": 2871.823,
      "eval_steps_per_second": 14.078,
      "step": 30
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9758467674255371,
      "learning_rate": 4.7e-05,
      "loss": 0.5573,
      "step": 45
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.678921568627451,
      "eval_combined_score": 0.7375945323697131,
      "eval_f1": 0.7962674961119751,
      "eval_loss": 0.6040505766868591,
      "eval_runtime": 0.1409,
      "eval_samples_per_second": 2896.163,
      "eval_steps_per_second": 14.197,
      "step": 45
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.941855788230896,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.5207,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6862745098039216,
      "eval_combined_score": 0.728029341232896,
      "eval_f1": 0.7697841726618705,
      "eval_loss": 0.6188527941703796,
      "eval_runtime": 0.1392,
      "eval_samples_per_second": 2930.784,
      "eval_steps_per_second": 14.367,
      "step": 60
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.642423391342163,
      "learning_rate": 4.5e-05,
      "loss": 0.4458,
      "step": 75
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6642156862745098,
      "eval_combined_score": 0.7021268184693232,
      "eval_f1": 0.7400379506641366,
      "eval_loss": 0.6644131541252136,
      "eval_runtime": 0.1409,
      "eval_samples_per_second": 2895.354,
      "eval_steps_per_second": 14.193,
      "step": 75
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.2749080657958984,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.3428,
      "step": 90
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6519607843137255,
      "eval_combined_score": 0.6925217455403214,
      "eval_f1": 0.7330827067669173,
      "eval_loss": 0.7664430141448975,
      "eval_runtime": 0.1571,
      "eval_samples_per_second": 2596.663,
      "eval_steps_per_second": 12.729,
      "step": 90
    },
    {
      "epoch": 7.0,
      "grad_norm": 5.924878120422363,
      "learning_rate": 4.3e-05,
      "loss": 0.2562,
      "step": 105
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6446078431372549,
      "eval_combined_score": 0.6847327640733714,
      "eval_f1": 0.7248576850094877,
      "eval_loss": 0.8936505913734436,
      "eval_runtime": 0.1519,
      "eval_samples_per_second": 2685.479,
      "eval_steps_per_second": 13.164,
      "step": 105
    },
    {
      "epoch": 7.0,
      "step": 105,
      "total_flos": 673316591603712.0,
      "train_loss": 0.47859985941932315,
      "train_runtime": 20.3015,
      "train_samples_per_second": 9033.83,
      "train_steps_per_second": 36.943
    }
  ],
  "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
}