[Known Issue] nn.Embedding layer TypeError: _renorm() takes exactly 5 arguments (4 given)

1

I defined my nn.Embedding layer as

self.word_embedding = nn.Embedding(vec.shape[0], vec.shape[1], vocab_size, max_norm = 1, norm_type = 1)

When forwarding it by self.word_embedding(input), I came across the following error

TypeError: _renorm() takes exactly 5 arguments (4 given)

Is there anything wrong with max_norm and norm_type?

2

It’s a known issue and will be fixed in a upcoming release.

1 Like
3

Thank you for this information.