checkpoint-978/trainer_state.json

{
  "best_global_step": 978,
  "best_metric": 0.03821074962615967,
  "best_model_checkpoint": "siglip2-finetune-full/checkpoint-978",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 978,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9210745123121202,
      "eval_loss": 0.20305734872817993,
      "eval_model_preparation_time": 0.0043,
      "eval_runtime": 197.3887,
      "eval_samples_per_second": 79.209,
      "eval_steps_per_second": 9.904,
      "step": 489
    },
    {
      "epoch": 1.0224948875255624,
      "grad_norm": 2.191868782043457,
      "learning_rate": 0.00010301724137931034,
      "loss": 0.3536,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9861848417013112,
      "eval_loss": 0.03821074962615967,
      "eval_model_preparation_time": 0.0043,
      "eval_runtime": 197.2642,
      "eval_samples_per_second": 79.259,
      "eval_steps_per_second": 9.911,
      "step": 978
    }
  ],
  "logging_steps": 500,
  "max_steps": 978,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 2.619034354332426e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}