search

Andy-zhujunwen / danet-pytorch

[pytorch] DANet: Dual Attention Network for Scene Segmentation
70 stars 24 forks source link

torch.max(attention, dim=-1, keepdim=True)[0].expand_as(attention) - attention #6

Open MorrisXu-Driving opened 3 years ago

MorrisXu-Driving commented 3 years ago

This line may reverse the weight. when MAX - Attention, the positions with max attention weight becomes zero. I also did not find relevant information in the paper. Why add this line?

nuaawh commented 1 year ago

Did you figure out the issue?

MorrisXu-Driving commented 1 year ago

I just deleted this line and do my own experiments with no anomaly spotted.