incerto.data.create_calibration_loaders

incerto.data.create_calibration_loaders#

incerto.data.create_calibration_loaders(train_dataset, test_dataset, calib_split=0.5, batch_size=128, num_workers=4, pin_memory=True, seed=42)[source]#

Create loaders for calibration experiments.

Splits training data into train/calibration sets, plus test set.

Parameters:

  • train_dataset (Dataset) – Training dataset

  • test_dataset (Dataset) – Test dataset

  • calib_split (float) – Fraction of training data to use for calibration

  • batch_size (int) – Batch size

  • num_workers (int) – Number of workers

  • pin_memory (bool) – Whether to pin memory

  • seed (int) – Random seed

Return type:

Tuple[DataLoader, DataLoader, DataLoader]

Returns:

Tuple of (train_loader, calib_loader, test_loader)

Contents