Generally if you are looking to provide additional knowledge to GPT the best way is through embeddings, not fine-tuning (which is better at classification/structure changes). LangChain project is probably the easiest way to get a proof-of-concept running so you can evaluate the results.
I’m actually wanting to teach AI to answer the main questions that are asked in my business.
From the information I read, it said to follow these steps:
Create a basic document with contextualization and as many questions and answers as possible.
Transform it into a language that makes it easier for AI to better understand and absorb the knowledge case.
Raise this base as a fine tuned base.
And continue training.
At this point I need to standardize the file. I’m sure someone must have already developed a way to transform each paragraph into a json compatible with the data model for fine tuning.
And in the sequel to know how this base using an https node from n8n through the api.
And excuse my complete ignorance, but I have no idea what langchain is…
The fine tuning guide I linked has all the information on that format.
Again, the consensus is not to use fine tuning for knowledgebase Q&A chatbot style applications. Here’s a relevant LangChain tutorial. You can search these forums for fine-tuning vs. embeddings for more discussion.