random.sample in siamese CNN tutorial don't work (anymore?)

[MarcinMisiurewicz]

picking random sample from dataset in next_pair with random.sample(foo,1) will produce error now. It works with random.choice(foo) though - lines 53 and 60.

Also getting loss from loss_contrastive should now look like this (line 142):
loss_log.append(loss_contrastive.item())

Thanks for the tutorial!