Research Paper Analysis System

This application analyzes research papers by extracting hypotheses and identifying limitations of a research paper using a fine-tuned Llama 3.2 model. This project was developed for the course Natural Language Processing (UE22CS342AB)

Features

• PDF Processing: Upload and extract text from research papers in PDF format
• RAG Pipeline: Retrieval-Augmented Generation for context-aware analysis
• Multi-Agent System:
  • Agent 1: Generates null and alternate hypotheses from the abstract
  • Agent 2: Identifies limitations and weaknesses in the research methodology for the paper given as input
• User-friendly Streamlit Interface: Easy-to-use GUI for interacting with the system

Installation

1. Clone this repository
2. Install dependencies:

   pip install -r requirements.txt
   

3. Set up your Hugging Face token:
   • Create a .streamlit/secrets.toml file with content:

     HF_TOKEN = "your_huggingface_token_here"
     

   • Or set it as an environment variable:

     export HF_TOKEN="your_huggingface_token_here"
     

Note: The fine tuned model can be accessed at: https://huggingface.co/NLP-team-6/llama-3-hypothesis-qlora

Usage

1. Run the Streamlit app:

   streamlit run app.py
   

2. Access the web interface at http://localhost:8501
3. Load the model using the sidebar button
4. Upload a research paper PDF
5. Process the PDF and analyze it using the provided buttons

Required Resources

• At least 8GB GPU VRAM (for 4-bit quantized model)
• 16GB+ RAM recommended
• Hugging Face account with access to Llama 3.2 models