search

AILab-CVC / UniRepLKNet

[CVPR'24] UniRepLKNet: A Universal Perception Large-Kernel ConvNet for Audio, Video, Point Cloud, Time-Series and Image Recognition
https://arxiv.org/abs/2311.15599
Apache License 2.0
863 stars 52 forks source link

A small bug in reparameterize() #8

Closed elokuuM closed 6 months ago

elokuuM commented 6 months ago

Problem with parameter in unireplknet.py line 338. conv = nn.Conv2d(self.dwconv.in_channels, self.dwconv.out_channels, self.dwconv.kernel_size, self.dwconv.padding, self.dwconv.groups, bias=True)

-> conv = nn.Conv2d(self.dwconv.in_channels, self.dwconv.out_channels, self.dwconv.kernel_size, padding=self.dwconv.padding, groups=self.dwconv.groups, bias=True)

elokuuM commented 6 months ago

And, in unireplknet.py line 149. identity_kernel = torch.ones((1, 1, 1, 1))

-> identity_kernel = torch.ones((1, 1, 1, 1)).to(kernel.device)

for reparameterize on GPU

DingXiaoH commented 6 months ago

Thank you for the valuable feedback! Fixed that!