The code used six, but does not import it. And in usage it still says "translator.py [<args>] [-h | --help]".
It seems that this script prints a score for every parallel sentence pair. But what is the score for?
We fixed the bugs. Thank you very much.
The scores can be used to compute sentence level cross-entropy loss, that is to say, evaluate in loss style rather than BLEU style.
The code used
six
, but does not import it. And inusage
it still says"translator.py [<args>] [-h | --help]"
. It seems that this script prints a score for every parallel sentence pair. But what is the score for?