Training AI for frictional Language?

Well, I was trying to use the Davinci model and try to translate the general word into covenant language (Sangheili language specifically)

Problem is that the language doesn’t officially exist nor does the translator (As the information is kinda limited)

So I’m curious, is it possible to use Fine-tuning for this specific case? (Using AI as translator)

Another thing, is it possible to run the AI locally after the Fine-tunning?

Forgot to mention that for the information, currently, I got two of them

I did try to feed it through via playground, but it didn’t work quite well :sweat_smile:

Hypothetically yeah, any pattern. You would need many thousands of examples though.

No, it will be many years before computers are powerful enough to run GPT-3 locally

Hypothetically?
If that’s true, how do I set up the training system then? (First time doing this, I’m not the programmer)

I forgot to mention that it’s sorta like, local translate library (similar to Google Translate Offline)