Hi All, is it possible to add a custom knowledge base to a fine-tuned DaVinci model? To be more specific, I am looking to add data in tabular format (ideally in csv). This tabular knowledge base may change once every quarter. The table would contain specific factual details about the business. I would like the fine-tuned model to be able to understand and use this data for text generation along with the details it has already been trained on while fine-tuning. Would love to get your thoughts on this. If there is an alternate option, happy to discuss that as well.
This is new to me. Can you link to any examples?
No, he can’t. I already reported this guy in 3 different topics, he keeps posting crap that is generated by GPT and, obviously, wrong. Bad links, inexistent projects, and now inexistent apis and features.
I already replied to you in several topics, you keep posting junk replies. They ARE bad answers because its made up stuff by chatgpt, which actually misdirects people who are looking for help.
If you don’t know the answer, let others reply, no need to go to chatgpt to make up stuff. Its not helpful at all.
@rajuroyal No, I didn’t misunderstand. And you are harming yourself.
I think what @nunodonato is saying is super simple. You are posting misleading answers and you should stop doing that, that’s it.
@Neural.Navigator, you can try to convert your table into text format and inject it into the prompt. We have tried that and it works.
2 Likes
exactly. it’s very simple and nothing personal @rajuroyal But if you are just creating misinformation, better to keep quiet. No need to leave, stay and learn. If you don’t learn, don’t try to teach what you don’t know.
No, the answers ARE WRONG! That’s the whole point. Why is this so hard to get? Let people be helped by users who know what they are talking about. Everyone knows ChatGPT exists, we don’t need you to go fetch answers for us. People come here to ask questions to other humans.
2 Likes
Bruh you literally talk about yourself in 3rd person. Massive red flag.
1 Like
The key concept that jumps off the page here is “specific factual details about the business” to which I would then ask (before suggesting any specific technical approach) – what is the business requirement you are trying to satisfy?
Suppose the goal is to create an AI experience for the customers of the “business”, and such experience would be ideally provided within the brand context of that business. In that case, I don’t think it makes sense to embellish the model in a fine-tuning approach. I assume you want this experience localized to your brand, eh?