You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Not to be nitpicky, but this could actually be replaced with two "1d" convolutions, one for width and one for height, which would use ~2K operations instead of ~K^2:
ffcv-imagenet/train_imagenet.py
Line 124 in e97289f
Not to be nitpicky, but this could actually be replaced with two "1d" convolutions, one for width and one for height, which would use
~2K
operations instead of~K^2
:The text was updated successfully, but these errors were encountered: