efficient_densenet_pytorch icon indicating copy to clipboard operation
efficient_densenet_pytorch copied to clipboard

will the inference memory reduced too?

Open emergencyd opened this issue 2 years ago • 1 comments

As described in the title

emergencyd avatar Aug 18 '22 04:08 emergencyd

No, but that's not necessary. There are no intermediate feature maps stored in memory during inference.

gpleiss avatar Aug 19 '22 18:08 gpleiss