Closed wtl-lang closed 2 years ago
Hi @wtl-lang. You are right. This is an error.
Line 453 should be out = self.conv1(out)
as you said.
However, changing this would make pre-trained weights useless; hence I will leave this a known issue for now.
Thank you for reporting this issue :-)
https://github.com/clovaai/aasist/blob/a04c9863f63d44471dde8a6abcb3b082b07cd1d1/models/AASIST.py#L453
Why on the right of this expression you use 'x' instead of 'out'? When you use 'out = self.conv1(x)', the 'if-else' is nonsense.