Changing Batch Size Causes Runtime Error

[ik362]

Hi there,

I am trying to train a CNN on some medical images. When I set batch_size = 1, everything works fine. However, when set batch_size > 1, the validation sanity check runs, and then on the second epoch I get the following error:

I really cannot work out this error as even when I explicitly change data type (e.g. y = y.type(torch.int) similar to https://discuss.pytorch.org/t/nn-bcewithlogitsloss-cant-accept-one-hot-target/59980 ) the error still remains.

Any help would be greatly appreciated!
Thanks!