When applying OTDD on a small dataset, it might happen that the last item of the labels might be of the form tensor(k) instead of tensor([k]) with k a torch.int64. In such case, it would produce the following error:
RuntimeError: zero-dimensional tensor (at position 1) cannot be concatenated.
To avoid that, I created a dumb dimension for the last item.
When applying OTDD on a small dataset, it might happen that the last item of the labels might be of the form
tensor(k)instead oftensor([k])withkatorch.int64. In such case, it would produce the following error:RuntimeError: zero-dimensional tensor (at position 1) cannot be concatenated.To avoid that, I created a dumb dimension for the last item.