Use `LogSoftmax` instead (it's faster and has better numerical properties). Examples:: >>> m = nn.Softmax(dim=1) >>> input = torch.randn(2, ...
確定! 回上一頁