The translation model is probably poor so it would make sense to just use it for speech recognition and then translate it with DeepL

For fuck sake, why?

>Torch not compiled with CUDA enabled

@matrix kek they probably regex replaced it from the training data
Follow

@e I think it's the models themselves since it seems to do it for all languages

Sign in to participate in the conversation
Game Liberty Mastodon

Mainly gaming/nerd instance for people who value free speech. Everyone is welcome.