LocalLLaMA

1 readers

1 users here now

Community to discuss about Llama, the family of large language models created by Meta AI.

founded 10 months ago

MODERATORS

communick@poweruser.forum

What is the tiniest GPT model one can fine tune on home hardware? (alien.top)

submitted 10 months ago by herozorro@alien.top to c/localllama@poweruser.forum

3 comments fedilink hide all child comments

is there a way to get a zero knowledge model that only knows how to chat. and from there fine tune it with specialized knowledge? and do this on consumer hardware (mac M1/16 gig) or free colab hardware?

i want to do this so as to prevent the model from hallucinating outside of the domain knowledge it is fed....like passing in a textbook and it only knows how to answer questions from it

top 3 comments

sorted by: hot top controversial new old

[–] troposfer@alien.top 1 points 10 months ago

Also what can you do with the latest m3 max with 128gb ram , can anyone put it in to context by comparing it

[–] DaltonSC2@alien.top 1 points 10 months ago

Checkout NanoGPT, it's very educational. You can even load GPT2 weights and finetune them if you want: https://github.com/karpathy/nanoGPT (The accompanying YouTube video is also very nice)

[–] Ok_Post_149@alien.top 1 points 10 months ago

Is home hardware a requirement for this project? I guess I'm a little confused what that has to do with model hallucinations.