4 ms·
Yeah, I had some promising results in a project that split markdown-based docs by second-level headers, embedding them all, and then doing basic RAG with GPT-4
by phillipcarter 2y ago
Yeah, I had some promising results in a project that split markdown-based docs by second-level headers, embedding them all, and then doing basic RAG with GPT-4 serving a response. It was too slow at the time (June last year) but I'll probably pick it back up again this year.
The main things I took away were (1) if the information archictecture isn't very splittable, this gets too hard, and (2) always link back to source information.
- dcre 2y agoAgreed on both counts. I do the same thing with headings and I use the results of the retrieval step to display a list of relevant docs while the answer is generating. The latest models are way better and faster than GPT-4 was. You’ll probably be happy when you get back into it.