torch.mean(input, dim=[2, 3], keepdim=True) dim (int) – the dimension to reduce?

znxlwm / UGATIT-pytorch

Official PyTorch implementation of U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation

MIT License

2.51k stars 476 forks source link

torch.mean(input, dim=[2, 3], keepdim=True) dim (int) – the dimension to reduce? #37

Open lmobot opened 4 years ago

lmobot commented 4 years ago

in networks.py 165 torch.mean(input, dim=[2, 3], keepdim=True) , why use list to dim

kouhinn commented 4 years ago

Same question for me.

dxm343488059 commented 4 years ago

Same question for me.

xinzi2018 commented 4 years ago

Same question for me.

BraveAsta commented 3 years ago

I think the function if it may be GAP(Group Average Pooling),which squeeze a feature map into an average value of it.