{
  "best_global_step": 1875,
  "best_metric": 0.2799457609653473,
  "best_model_checkpoint": "siglip2-finetune-full/checkpoint-1875",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1875,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.8,
      "grad_norm": 12.590618133544922,
      "learning_rate": 1.63265306122449e-05,
      "loss": 0.5408,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7876,
      "eval_loss": 0.4367133677005768,
      "eval_model_preparation_time": 0.0039,
      "eval_runtime": 439.5267,
      "eval_samples_per_second": 45.503,
      "eval_steps_per_second": 5.688,
      "step": 625
    },
    {
      "epoch": 1.6,
      "grad_norm": 32.36686706542969,
      "learning_rate": 1.2244897959183674e-05,
      "loss": 0.415,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8439,
      "eval_loss": 0.34147578477859497,
      "eval_model_preparation_time": 0.0039,
      "eval_runtime": 436.8014,
      "eval_samples_per_second": 45.787,
      "eval_steps_per_second": 5.723,
      "step": 1250
    },
    {
      "epoch": 2.4,
      "grad_norm": 10.255178451538086,
      "learning_rate": 8.16326530612245e-06,
      "loss": 0.3582,
      "step": 1500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8738,
      "eval_loss": 0.2799457609653473,
      "eval_model_preparation_time": 0.0039,
      "eval_runtime": 441.0603,
      "eval_samples_per_second": 45.345,
      "eval_steps_per_second": 5.668,
      "step": 1875
    }
  ],
  "logging_steps": 500,
  "max_steps": 2500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5.02532974927872e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}