Karol Grzegorczyk
Results
2
comments of
Karol Grzegorczyk
> Hey @kagrze This is a frequent complaint indeed. Thanks for contributing! Do you plan on adding some example functions (e.g. translation + BLEU computation) as well as a bit...
@francoishernandez this should work: ``` def detokenize_batch(batch: Tensor, field: Field) -> T.List[str]: sentences = [] for i in range(batch.shape[1]): tokens = "" for t in range(batch.shape[0]): token = field.vocab.itos[batch[t, i,...