Replying to nobody

Gonna try my own local GPT-4 Alpaca model that is completely uncensored (no more... As an AI language model etc...) when I get a chance.

Time for personal but local AI assistant too 😅 with Vicuna chat lol 🤺
Avatar
average_bitcoiner 2y ago

Link to model? And hopefully guide on how to use it?

Reply to this note

Please Login to reply.

Discussion

89
nobody 2y ago

I still have to research more but there is a package for running models optimized for CPU instead called Llama cpp library (and then go from there....). I'm gonna throw that on Mac in this case. Just personal needs kinda thing see the heck is up with all this

https://github.com/ggerganov/llama.cpp

Avatar
average_bitcoiner 2y ago

Nice. I’ve seen this before. Just didn’t realize a GPT-4 level model was available.

89
nobody 2y ago

I prefer to think and not let AI do everything but I'm curious and you know digital world 😅

Avatar
average_bitcoiner 2y ago

I’m more interested in using things like LangChain to generate context specific LLM bots to do Q&A. At work I’ve got a large corpus of cybersecurity data I’m trying to contextualize around an LLM. Would be dope for bitcoin content too.

89
nobody 2y ago

Interesting. That is one frontier I need to explore still..... the LLMs and AI in general. Would be cool if you document anything you may run into 🤙

89
nobody 2y ago

March older news but mustve gotten better

https://lmsys.org/blog/2023-03-30-vicuna/