Open ZhenboZhao77 opened 2 months ago
The input and output of VSSBlock are both [B, C, H, W] (or [B, H, W, C], based on hyper-parameter channel_first). So if the feature map you use is 1024x1024x1 then you can input the tensor with shape 1x1x1024x1024 into it.
Hello author,
How should I use VSSBlock to process a tensor with shape torch.Size([1024, 1024])? Could you please provide some guidance? Thank you!