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

Fine tuning AI (alien.top)

submitted 9 months ago by silvio194@alien.top to c/localllama@poweruser.forum

7 comments fedilink hide all child comments

Hi, I would like to better understand if it is possible to create a chatbot for free trained on a dataset (which I still have to create) on my PC.I was seeing that llama-2 could be a solution but you have to host it.So I understood that an external machine with GPU is needed, but apart from that, is it possible to make a totally private chatbot?I would like to make a small MVP, if so please share with me anything that might be useful .I'm a web dev and I know also python but not the theory below AI

you are viewing a single comment's thread
view the rest of the comments

[–] tgredditfc@alien.top 1 points 9 months ago (6 children)

You can start with reading Oobabooga’s wiki, I think it’s one of most beginner friendly tools. https://github.com/oobabooga/text-generation-webui/wiki/05-%E2%80%90-Training-Tab

[–] silvio194@alien.top 1 points 9 months ago (5 children)

But do you think an AI chatbot can run on my PC?

[–] tgredditfc@alien.top 1 points 9 months ago (2 children)

You can use oobabooga API to do that. I haven’t done it myself, can’t say much about it.

[–] silvio194@alien.top 1 points 9 months ago (1 children)

Mmmh okok.... But is there a guide from scratch on this?

[–] tgredditfc@alien.top 1 points 9 months ago

Maybe. I have not done it yet so I don’t know. You can google around.

load more comments (2 replies)