Packed_sequence and packed_padded_sequence

I want to build a Bi-LSTM for text classification.

For computational efficiency it is a good practice to use PackedSequence. I am not sure if I should use packed_sequence or packed_padded_sequence.

Can anyone provide a short explanation as to which one to prefer and why.

Thanks in advance,

1 Like

I am also curious to understand the difference between when to use pack_sequence and pack_padded_sequence. What are the pros and cons of each?