run-llama / rags

Build ChatGPT over your data, all with natural language

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

How to upload files?

Lauorie opened this issue · comments

I turn to page RAG Config, but it shows "File/URL paths (not editable)".So where can I upload PDFs?

Hi ,

I've got a chance to set filepath during this conversation at Home page :

image
image

But bot at Generated RAG Agent still unable to answer questions about the PDF :

image

@snassimr What if you ask a question whose answer is in the text instead of a figure? It is answering correctly?

@alexfilothodoros and @snassimr I tried creating a bot on llama2 paper and it did answer questions related to text though it was failing to answer queries on figure 2. It might be due to the issue with the pdf parser.

Maybe you need the GPT-4V model.

**EDIT

on second thought that might also not work. I forgot that GPT-4V doesn't accept pdf files.

I turn to page RAG Config, but it shows "File/URL paths (not editable)".So where can I upload PDFs?

I am not quite sure what's the point

Hi ,

Sorry for late answer . The issue was with pdf file name . Linux doesn't suffer blanks and file actually wasn't loaded at home page.
After @jerryjliu point me on my mistake the PDF was loaded at home page and at "Generated RAG Agent" I get nice answer from Agent :

image

I think some Exception handling (in agent_utils.py -> load_data) and returning an appropriate answer to chat is favorable here .

Another issue (might be parser problem) that asking about Figure 2 returning answer about Figure 30 , but it;s another story . I guess we can benefit from this repo which primarily benefit to use agent to built agent and retriever/qa performance can be improved more and more ...

Regarding setting filepath - use bot at home page . Here the example for first filepath setting 👍

image

And continue with the bot until you get :

image