In the demonstration of different ways of labelling our dataset using MNIST dataset, the value used in the ‘size’ parameter was different each time. Could anyone please explain?
data = ImageDataBunch.from_name_func(path, fn_paths, ds_tfms=tfms, size=24,
label_func = lambda x: '3' if '/3/' in str(x) else '7')
data = ImageDataBunch.from_csv(path, ds_tfms=tfms, size=28)
data = ImageDataBunch.from_folder(path, ds_tfms=tfms, size=26)
The values 24, 26, 28 were used on the same dataset.