Closed pr509 closed 4 months ago
Post porting a fairseq model to ctranslate2, using the ct2-fairseq-converter
the source and target vocabularies are dumped in txt format and not in json format, so this issue would never arise.
Can you share more details as to what was the command used and double-check what was the model being used ?
when i am trying to convert trained fairseq model to ctranslate i got a source_vocabolary.json and target_vocabalory.json out of these two the target_vocab.json in full with unicodes and because of this the translation is not coming good .could you plz help me in finding solution for this