Prompts design and fine tune model output

Recently I have fine tuned LLM model with large set of prompts, when I test fine tune model it is performing worst, even not getting correct answer. How can I resolve this issue.

Can you provide some examples of the training set you used and the test prompts and replies?

2 Likes

I can not it is confidential document and meanwhile if any option to generate prompts from pdf table context. That going to helps me lot.