trainer_state.json

{
  "best_global_step": 2964,
  "best_metric": 0.21923576295375824,
  "best_model_checkpoint": "multilingual-e5-small-pii-detector/checkpoint-2964",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 4446,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.33738191632928477,
      "grad_norm": 5.012300491333008,
      "learning_rate": 4.438821412505623e-05,
      "loss": 0.3135,
      "num_input_tokens_seen": 512000,
      "step": 500,
      "train_runtime": 14.3457,
      "train_tokens_per_second": 35690.033
    },
    {
      "epoch": 0.6747638326585695,
      "grad_norm": 0.07379312068223953,
      "learning_rate": 3.876518218623482e-05,
      "loss": 0.2333,
      "num_input_tokens_seen": 1024000,
      "step": 1000,
      "train_runtime": 28.4978,
      "train_tokens_per_second": 35932.607
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9038785834738617,
      "eval_loss": 0.25766971707344055,
      "eval_runtime": 1.7391,
      "eval_samples_per_second": 1704.926,
      "eval_steps_per_second": 213.331,
      "num_input_tokens_seen": 1517568,
      "step": 1482
    },
    {
      "epoch": 1.0121457489878543,
      "grad_norm": 1.9765552282333374,
      "learning_rate": 3.3142150247413403e-05,
      "loss": 0.2038,
      "num_input_tokens_seen": 1536000,
      "step": 1500,
      "train_runtime": 45.5628,
      "train_tokens_per_second": 33711.681
    },
    {
      "epoch": 1.349527665317139,
      "grad_norm": 0.05636508762836456,
      "learning_rate": 2.7519118308591997e-05,
      "loss": 0.1658,
      "num_input_tokens_seen": 2048000,
      "step": 2000,
      "train_runtime": 59.6595,
      "train_tokens_per_second": 34328.161
    },
    {
      "epoch": 1.686909581646424,
      "grad_norm": 1.788957118988037,
      "learning_rate": 2.1896086369770583e-05,
      "loss": 0.1719,
      "num_input_tokens_seen": 2560000,
      "step": 2500,
      "train_runtime": 73.8011,
      "train_tokens_per_second": 34687.829
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.92141652613828,
      "eval_loss": 0.21923576295375824,
      "eval_runtime": 1.7041,
      "eval_samples_per_second": 1739.889,
      "eval_steps_per_second": 217.706,
      "num_input_tokens_seen": 3035136,
      "step": 2964
    },
    {
      "epoch": 2.0242914979757085,
      "grad_norm": 0.0168524831533432,
      "learning_rate": 1.627305443094917e-05,
      "loss": 0.1458,
      "num_input_tokens_seen": 3072000,
      "step": 3000,
      "train_runtime": 90.7129,
      "train_tokens_per_second": 33865.09
    },
    {
      "epoch": 2.361673414304993,
      "grad_norm": 0.3309631049633026,
      "learning_rate": 1.0650022492127757e-05,
      "loss": 0.1314,
      "num_input_tokens_seen": 3584000,
      "step": 3500,
      "train_runtime": 104.7568,
      "train_tokens_per_second": 34212.57
    },
    {
      "epoch": 2.699055330634278,
      "grad_norm": 3.9781689643859863,
      "learning_rate": 5.026990553306343e-06,
      "loss": 0.1224,
      "num_input_tokens_seen": 4096000,
      "step": 4000,
      "train_runtime": 118.8877,
      "train_tokens_per_second": 34452.667
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9133220910623946,
      "eval_loss": 0.27373236417770386,
      "eval_runtime": 1.7591,
      "eval_samples_per_second": 1685.527,
      "eval_steps_per_second": 210.904,
      "num_input_tokens_seen": 4552704,
      "step": 4446
    },
    {
      "epoch": 3.0,
      "num_input_tokens_seen": 4552704,
      "step": 4446,
      "total_flos": 585740316155904.0,
      "train_loss": 0.17994267544765705,
      "train_runtime": 134.4715,
      "train_samples_per_second": 264.502,
      "train_steps_per_second": 33.063
    }
  ],
  "logging_steps": 500,
  "max_steps": 4446,
  "num_input_tokens_seen": 4552704,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 585740316155904.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}