Mini Batches respecting class proportion/frequency

[Fra (Francesco)]

Hi All,

I would like to know if there is an efficient way to create mini-batches which respect class proportion ( that is, if my training is formed by 95% samples in class 1 and 5% class2, I would like to have mini-batches respecting these frequencies).

Thank you!

[Fra (Francesco)]

I have currently found the answer here:

How to enable the dataloader to sample from each class with equal probability

The dataloader utility in torch (courtesy of Soumith Chintala) allowed one to sample from each class with equal probability. I was wondering, if there is a straightforward approach to enable the same in pytorch dataloaders.