trainer_state.json

{
  "best_metric": 66.49750868537264,
  "best_model_checkpoint": "/home/jcanete/ft-data/all_results/sqac/albeto_base_4/epochs_3_bs_16_lr_5e-5/checkpoint-2600",
  "epoch": 3.0,
  "global_step": 3114,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.19,
      "eval_exact_match": 22.478540772532188,
      "eval_f1": 41.008595516114795,
      "step": 200
    },
    {
      "epoch": 0.39,
      "eval_exact_match": 34.12017167381974,
      "eval_f1": 53.95894344679839,
      "step": 400
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.20199100834939e-05,
      "loss": 2.7553,
      "step": 500
    },
    {
      "epoch": 0.58,
      "eval_exact_match": 40.28969957081545,
      "eval_f1": 59.0770672275212,
      "step": 600
    },
    {
      "epoch": 0.77,
      "eval_exact_match": 42.38197424892704,
      "eval_f1": 61.63724279749357,
      "step": 800
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.399165061014773e-05,
      "loss": 1.6752,
      "step": 1000
    },
    {
      "epoch": 0.96,
      "eval_exact_match": 44.957081545064376,
      "eval_f1": 64.03757621093638,
      "step": 1000
    },
    {
      "epoch": 1.16,
      "eval_exact_match": 45.43991416309013,
      "eval_f1": 64.49291926425883,
      "step": 1200
    },
    {
      "epoch": 1.35,
      "eval_exact_match": 46.08369098712446,
      "eval_f1": 64.95279156327298,
      "step": 1400
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.596339113680154e-05,
      "loss": 1.0992,
      "step": 1500
    },
    {
      "epoch": 1.54,
      "eval_exact_match": 48.17596566523605,
      "eval_f1": 66.34837628911647,
      "step": 1600
    },
    {
      "epoch": 1.73,
      "eval_exact_match": 47.15665236051502,
      "eval_f1": 66.45366106066392,
      "step": 1800
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7935131663455366e-05,
      "loss": 1.0281,
      "step": 2000
    },
    {
      "epoch": 1.93,
      "eval_exact_match": 47.26394849785408,
      "eval_f1": 66.38564273683892,
      "step": 2000
    },
    {
      "epoch": 2.12,
      "eval_exact_match": 47.58583690987125,
      "eval_f1": 66.37193238617046,
      "step": 2200
    },
    {
      "epoch": 2.31,
      "eval_exact_match": 46.566523605150216,
      "eval_f1": 65.90752587534809,
      "step": 2400
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.922928709055877e-06,
      "loss": 0.6328,
      "step": 2500
    },
    {
      "epoch": 2.5,
      "eval_exact_match": 47.58583690987125,
      "eval_f1": 66.49750868537264,
      "step": 2600
    },
    {
      "epoch": 2.7,
      "eval_exact_match": 46.99570815450644,
      "eval_f1": 66.00471724341504,
      "step": 2800
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.8946692357096983e-06,
      "loss": 0.5553,
      "step": 3000
    },
    {
      "epoch": 2.89,
      "eval_exact_match": 46.62017167381974,
      "eval_f1": 65.72848295022517,
      "step": 3000
    },
    {
      "epoch": 3.0,
      "step": 3114,
      "total_flos": 1075319724912768.0,
      "train_loss": 1.2644996986070656,
      "train_runtime": 276.7945,
      "train_samples_per_second": 179.895,
      "train_steps_per_second": 11.25
    }
  ],
  "max_steps": 3114,
  "num_train_epochs": 3,
  "total_flos": 1075319724912768.0,
  "trial_name": null,
  "trial_params": null
}