Speed up translation process

Hello all,

I want to use Libretranslate to translate a larger dataset (1+ Mio) rows from English to German. I setup LibreTranslate locally via Docker and using LibreTranslatepy in my python application. The translations are stored back in the DB.

It all works fine and the translation works so far. Nevertheless, it takes a lot of time to get the translation done and I was wondering if one of you had some similar challenges. I was thinking about using two dockers and run the translations in parallel, but maybe one of you has some great inputs on how to translate larger amounts of data locally and fast.

Thanks in advance for your help.

Best regards,
Oliver

1 Like

Parallelization is probably your best bet. Most of the translation time is CTranslate2 doing the translation inference. You could also try using a GPU.

1 Like