ggt.data

Submodules

Classes

FITSDataset

Dataset from FITS files. Pre-caches FITS files as PyTorch tensors to

Functions

get_data_loader(dataset, batch_size, n_workers[, shuffle])

Package Contents

class ggt.data.FITSDataset(data_dir, slug=None, split=None, channels=1, cutout_size=167, label_col='bt_g', normalize=True, transform=None, expand_factor=1, repeat_dims=False, label_scaling=None, scaling_data_dir=None, scaling_slug=None, load_labels=True)

Bases: torch.utils.data.Dataset

Dataset from FITS files. Pre-caches FITS files as PyTorch tensors to improve data load speed.

data_dir

cutout_shape

normalize = True

transform = None

repeat_dims = False

expand_factor = 1

data_info = None

cutouts_path

tensors_path

filenames

observations

__getitem__(index): Magic method to index into the dataset.

__len__(): Return the effective length of the dataset.

static load_fits_as_tensor(filename): Open a FITS file and convert it to a Torch tensor.

ggt.data.get_data_loader(dataset, batch_size, n_workers, shuffle=True)