I read in appendix E that "We employ PyTorch version 1.10 to use the latest “Channel Last” memory layout for further speedup". However, I have not seen that in the code. Is there a specific reason for not releasing the code with channel last memory layout?
I am just wondering because I wanted to see how you implemented it.
Hi
I read in appendix E that "We employ PyTorch version 1.10 to use the latest “Channel Last” memory layout for further speedup". However, I have not seen that in the code. Is there a specific reason for not releasing the code with channel last memory layout? I am just wondering because I wanted to see how you implemented it.
Cheers