How do i tran LLM bot on my own knowledge for my next.js project?
Unanswered
American Crocodile posted this in #help-forum
American CrocodileOP
hi guys so i've never used ai models in my projects before but i'm trying to create an app for a niche exercise market (post injury specific exercises) and i'm trying to do research on how i can "train" bots on my own data/knowledge.
so my question is, how do i go about doing that or doing research on it, because i dont know what i don't know and there seems to be a ton of these "solutions" online which none of them do what they advertise to a good degree so basically i wanna do everything myself. so what terminology do i need to know to do correct and effecient research on training a bot on my own data? thanks for reading haha
so my question is, how do i go about doing that or doing research on it, because i dont know what i don't know and there seems to be a ton of these "solutions" online which none of them do what they advertise to a good degree so basically i wanna do everything myself. so what terminology do i need to know to do correct and effecient research on training a bot on my own data? thanks for reading haha
11 Replies
i am afraid that this is not right place for LLM and RAG. You might want to check vector DBs and its communities an such as Weaviate.
American Pipit
look up training chatGPT by splitting up text into chunks, embedding them and then training chatgpt with that data
it will take some time, but you can do it without much issues
and you can use nextjs documentation as a reference to upload as .pdf
you'll have to get creative
you can also check out supabase docs for vector DBs etc
hope this helps
Python discord has channels related to machine learning
American CrocodileOP
Thank you so much everybody
i found you Supabase but Weaviate would be better. i'll take care of you there. it's Slack. I have experience in Weaviate, Pinecone and Qdrant.