Repository files navigation

This repository contains the implementation of aRetrieval-Augmented Generation (RAG) pipeline designed to provide scientifically grounded justifications for drug efficacy claims. By leveragingLarge Language Models (LLMs) and integrating verified biomedical data sources such asPubMed andDrugBank, this project aims to enhance the reliability of AI-driven justifications in biomedical research.

• RAG-based Justification: Retrieves task-specific, verified information to support drug-disease relationships.
• Multiple LLM Testing: Evaluates different LLMs with various RAG techniques to determine the optimal model.
• Expert-Guided Evaluation: Utilizes expert-curated ground truth for performance assessment.
• Role-Play Reasoning: Employs scenario-based reasoning to enhance the logical consistency of generated justifications.

• Python 3.8+
• CUDA-enabled GPU (optional but recommended)
• Dependencies listed inrequirements.txt

git clone https://github.com/your-username/your-repository.gitcd your-repository