Kevin Hu
Kevin Hu
https://github.com/infiniflow/ragflow/blob/main/rag/prompts/keyword_prompt.md
For now, there's only one tenant will be returned.
Always `RAG` instead of ` fine-tuning`. `8B` is far beyond feasible.
For LLM, the larger the better.
Appreciations! Do you have screen recording or something?
Appreciations! Could you share us with screenshot GIF or something?
The screenshot GIF is the most efficient way to demostrate what is this PR all about.
You could utilize `Retrieval Test` to ensure that if the related chunks are retrieved. To improve retrieval: 1. Increase the chunk size. 2. Give it a try: auto-keywords or auto-questions
No limitation. In term of data processing, you'd better categorize your documents into different KBs based on users' possible queries. **Every big data set needs catgorization.**
Register a new user with the old password. Compare the password in MySQL: rag_flow.user. Reset the old password to the new one using update statement.