Paper RAG

A tool for querying research papers using Retrieval-Augmented Generation, with a web interface wrapper for nicer conversations.

How to Use

1. Run python app.py and open http://localhost:5000
2. Create a project and upload your PDF papers
3. Click "Index Papers" to process them
4. Ask questions in natural language and get AI-generated answers based on the content

Configuration

Edit config.py to change:

• EMBEDDING_MODEL: Model for semantic search
• CHUNK_SIZE / CHUNK_OVERLAP: How papers are split
• HF_MODEL: LLM for generating answers
• k: Number of chunks to retrieve per query

Planned Features

• Run flask app on a raspberry pi or online service
• Add option to do retrieval on demand to access entire databases
  • utilize a cache to only search when necessary
  • add ability to do projects within PubMed search