Thanks :)
I based the implementation in my little "push to talk" project on your work! Real time approach was exactly what I needed.
You deserved a special mention <3
Btw, this model is a beast! Even on CPU that's crazy! No more issues to keep both LLM and ASR models loaded at the same time!
I based the implementation in my little "push to talk" project on your work! Real time approach was exactly what I needed.
You deserved a special mention <3Btw, this model is a beast! Even on CPU that's crazy! No more issues to keep both LLM and ASR models loaded at the same time!
Man, that's AMAZING!
Lets connect via direct messages if you want.
Didn't expected, that someone will really make his own project, using this repo - for real, KEEP GOING!
//Wojciech
support gpu?only use device="cuda"?