seq2seq_torch maybe have a small mistake

[wmathor]

# output : [max_len+1, batch_size, num_directions(=1) * n_hidden]
    output = output.transpose(0, 1) # [batch_size, max_len+1(=6), num_directions(=1) * n_hidden]

to

# output : [max_len+1, batch_size, n_class]
    output = output.transpose(0, 1) # [batch_size, max_len+1(=6), n_class]

[Angry-Echo]

Nope~ I guess you forget an important thing, the last layer of the network is Fully Connected layer , it trans the dim from hidden_size to input_size(n_class) However, i konw you are a amaing coder, if you see me, can we make a friend

[wmathor]

Nope~ I guess you forget an important thing, the last layer of the network is Fully Connected layer , it trans the dim from hidden_size to input_size(n_class) However, i konw you are a amaing coder, if you see me, can we make a friend

Oh my god, I can't believe someone actually responded to my stupid question after all this time. Anyway, thank you for your answer.