HI, jisu
I find that there are some errors in your code. For example:
the params of nn.Embedding are int, int...But, they are emb,list in the line 228. You can use torch.index_select(tensor, dim, ...) instead of tf.nn.embedding_lookup.
the second param of torch.split() should be the dim that tensor is splited, rather than the num of split.
HI, jisu
I find that there are some errors in your code. For example:
the params of nn.Embedding are int, int...But, they are emb,list in the line 228. You can use torch.index_select(tensor, dim, ...) instead of tf.nn.embedding_lookup.
the second param of torch.split() should be the dim that tensor is splited, rather than the num of split.