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]