BPL-RAG-Fall-2025 / README.md
nc0926's picture
Update README.md
7a74b56 verified

A newer version of the Streamlit SDK is available: 1.56.0

Upgrade
metadata
title: BPL RAG Fall 2025
emoji: ๐Ÿ 
colorFrom: yellow
colorTo: red
sdk: streamlit
python_version: '3.11'
app_file: src/app.py
tags:
  - streamlit
pinned: false
short_description: A metadata-aware RAG system built on Digital Commonwealth
license: gpl-3.0
sdk_version: 1.52.1

Welcome to Our Hugging Face Deployment

This Space hosts the BPL RAG System demo for DS 549. To see the full codebase, documentation, and development workflow, visit the GitHub repository:

๐Ÿ‘‰ https://github.com/BU-Spark/ml-bpl-rag


Known Issues & Notes

1. Streamlit SDK Deprecation on Hugging Face

Hugging Face has deprecated the native Streamlit SDK, and Spaces now deploy using Docker under the hood. Even if the Space is configured as "Streamlit," it still builds and runs inside a Docker container.

2. Docker Deployment Stuck on "Restarting"

Switching the Space to the explicit Docker SDK causes the container to get stuck in a perpetual "Restarting" state. This appears to be a Spaces-level issue rather than something wrong with the app.

3. Docker Blocks Outgoing Calls to OpenAI

Spaces running via Docker cannot make outbound API calls to OpenAI. This breaks any RAG workflow that relies on openai as the LLM provider.

โœ” Workaround

To bypass this limitation, we switched the LLM provider to OpenRouter.ai, which does allow outbound API calls from Docker-based Spaces.


Feel free to explore the app and report any issues through GitHub!