Shift-Net_pytorch
Shift-Net_pytorch copied to clipboard
Why are the size of inpainting models so big?
Hi, looks like most of inpainting models are more bigger than 200MB. Is it possible to reduce the model size to less than 100MB while keep current performance? thank you~
Change here https://github.com/Zhaoyi-Yan/Shift-Net_pytorch/blob/master/models/modules/shift_unet.py#L41. Relapce some of them with "residual blocks", it should be weight-lighter than unet-skip blocks, as unet-skip block is feature-concatenating while residual-block is feature-adding.
woo... thank you, I will try this~