search

HKUDS / LightRAG

"LightRAG: Simple and Fast Retrieval-Augmented Generation"
https://arxiv.org/abs/2410.05779
MIT License
9.26k stars 1.14k forks source link

Added GUI (linked), updated README.md #241

Closed aiproductguy closed 1 week ago

aiproductguy commented 1 week ago

I am not sure streamlit is the interface I want to contribute back to the LightRAG project, but willing to share it all under the same MIT license. It seemed you did not like streamlit as well. I am working on taipy or flask next and can contribute that how you suggest.

I figured I would at least share the link and source code in the README.

I would like to understand how you benchmark and want to visualize the investigation/verification workflows for future version.

ASAD-BE18 commented 1 week ago

If you could, please add Ollama support in the streamlit application using API endpoints

LarFii commented 1 week ago

Thanks for your GUI! I'm not very familiar with these visualization methods, so I don't have any specific suggestions. For the evaluation of LightRAG, we use a one-to-one approach similar to GraphRAG. We generate users and tasks for a specified dataset using GPT, then have different RAG systems answer the same query.