finetune model

[drewskidang]

is it possible to use my own finetune llma with this

[debanjum]

Your fine tune of Llama/Llama 2? Khoj doesn't support swapping the offline models in any neat way.
But if you're fine with a very hacky solution, you could place any GPT4All compatible model at ~/.cache/gpt4all/llama-2-7b-chat.ggmlv3.q4_K_S.bin and see if that works?

[Ellen7ions]

I propose a new PR #402 that supports specifying gpt4all models based on user input. I hope it helps @drewskidang @debanjum

[debanjum]

I'll convert this into a Github discussion, just in case you want to continue this conversation there. Hope that's fine

[Ellen7ions]

#402 might help you 😊

[drewskidang]

Appreciate it!!