trainer_state.json

{
  "best_global_step": 2964,
  "best_metric": 0.2924116253852844,
  "best_model_checkpoint": "embeddinggemma-300m-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": 17.098201751708984,
      "learning_rate": 4.438821412505623e-05,
      "loss": 0.5645,
      "num_input_tokens_seen": 512000,
      "step": 500,
      "train_runtime": 47.9743,
      "train_tokens_per_second": 10672.374
    },
    {
      "epoch": 0.6747638326585695,
      "grad_norm": 1.4570860862731934,
      "learning_rate": 3.876518218623482e-05,
      "loss": 0.3907,
      "num_input_tokens_seen": 1024000,
      "step": 1000,
      "train_runtime": 94.2729,
      "train_tokens_per_second": 10862.077
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8627318718381113,
      "eval_loss": 0.30147919058799744,
      "eval_runtime": 7.391,
      "eval_samples_per_second": 401.161,
      "eval_steps_per_second": 50.196,
      "num_input_tokens_seen": 1517568,
      "step": 1482
    },
    {
      "epoch": 1.0121457489878543,
      "grad_norm": 14.889694213867188,
      "learning_rate": 3.3142150247413403e-05,
      "loss": 0.3335,
      "num_input_tokens_seen": 1536000,
      "step": 1500,
      "train_runtime": 151.9538,
      "train_tokens_per_second": 10108.336
    },
    {
      "epoch": 1.349527665317139,
      "grad_norm": 0.4145456552505493,
      "learning_rate": 2.7519118308591997e-05,
      "loss": 0.3069,
      "num_input_tokens_seen": 2048000,
      "step": 2000,
      "train_runtime": 198.9267,
      "train_tokens_per_second": 10295.252
    },
    {
      "epoch": 1.686909581646424,
      "grad_norm": 49.79447555541992,
      "learning_rate": 2.1896086369770583e-05,
      "loss": 0.2847,
      "num_input_tokens_seen": 2560000,
      "step": 2500,
      "train_runtime": 245.6972,
      "train_tokens_per_second": 10419.327
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8978077571669477,
      "eval_loss": 0.2924116253852844,
      "eval_runtime": 7.3471,
      "eval_samples_per_second": 403.563,
      "eval_steps_per_second": 50.496,
      "num_input_tokens_seen": 3035136,
      "step": 2964
    },
    {
      "epoch": 2.0242914979757085,
      "grad_norm": 0.2802899181842804,
      "learning_rate": 1.627305443094917e-05,
      "loss": 0.2542,
      "num_input_tokens_seen": 3072000,
      "step": 3000,
      "train_runtime": 303.072,
      "train_tokens_per_second": 10136.207
    },
    {
      "epoch": 2.361673414304993,
      "grad_norm": 29.810983657836914,
      "learning_rate": 1.0650022492127757e-05,
      "loss": 0.2176,
      "num_input_tokens_seen": 3584000,
      "step": 3500,
      "train_runtime": 350.1674,
      "train_tokens_per_second": 10235.104
    },
    {
      "epoch": 2.699055330634278,
      "grad_norm": 14.638989448547363,
      "learning_rate": 5.026990553306343e-06,
      "loss": 0.2064,
      "num_input_tokens_seen": 4096000,
      "step": 4000,
      "train_runtime": 397.7303,
      "train_tokens_per_second": 10298.436
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8930860033726813,
      "eval_loss": 0.30713844299316406,
      "eval_runtime": 7.2761,
      "eval_samples_per_second": 407.498,
      "eval_steps_per_second": 50.989,
      "num_input_tokens_seen": 4552704,
      "step": 4446
    },
    {
      "epoch": 3.0,
      "num_input_tokens_seen": 4552704,
      "step": 4446,
      "total_flos": 2773636063690752.0,
      "train_loss": 0.30860748496937207,
      "train_runtime": 450.538,
      "train_samples_per_second": 78.946,
      "train_steps_per_second": 9.868
    }
  ],
  "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": 2773636063690752.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}