According to this link, we need to give the input to the layer in the format of (Batch x the_size)
. But what I have worked with Pytorch, it is always able to recognize the difference between Batched-Input
and Non-Batched Input
(also as you said).
- So why it this point we need to define an extra dimension for batch before passing to the layer.
- If it can already create the distinct output for
Batched(4D Input)
andNon-Batched(3D Input)
, then why it is throwing aValue Error
. - If it can process both the
4D Input
and3D Input
,why it expecting 4D Input
.
The specific Layer I was getting Issue with was Batch Norm 2d
Sorry for replying Late : )