search

Helsinki-NLP / Opus-MT

Open neural machine translation models and web services
MIT License
604 stars 71 forks source link

opus-mt-he-en unable to translate when text has latin digits #82

Closed hdeval1 closed 1 year ago

hdeval1 commented 1 year ago

Whenever a hebrew sentence that contains a latin is being translated, it fails to complete the translation and returns the entire text as a transliteration. Is there a built in filter/feature to get around this? Maybe converting them to their latin equivalent? I don't want to implement something that already exists and risk loosing efficiency and performance.

hdeval1 commented 1 year ago

I was using the wrong tokenizer and model from huggingface...disregard this ticket