Skip to content

tahmidmir/Graph-RAG

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 

Repository files navigation

🔥 Project: Fine-tuning GPT-2 & GraphRAG for Medical Text Processing

This repository includes two Natural Language Processing (NLP) projects designed for medical text analysis and improving information retrieval.


📌 1. Fine-tuning GPT-2 on Skin Cancer Articles

In this section, the GPT-2 model is fine-tuned on specialized skin cancer articles to enhance its ability to understand medical texts.

Features:

✅ Data collection from PubMed and other scientific sources
✅ Preprocessing of medical texts to optimize model learning
✅ Fine-tuning the GPT-2 model using Hugging Face Transformers
✅ Model evaluation and testing for generating medical text responses


🔧 Installation & Requirements

Dependencies

Libraries used in this project:

  • torch
  • transformers
  • numpy
  • pandas
  • matplotlib
  • datasets
  • biopython
  • scikit-learn
  • tqdm

First, install the necessary packages:

pip install torch transformers datasets biopython

Then, run the notebook in Jupyter Notebook.

Run the Notebook

jupyter notebook fine-tuning-gpt-2-on-skin-cancer-articles.ipynb

📌 2. GraphRAG: Combining Graphs and Generative Models for Enhanced Information Retrieval

In this section, the Retrieval-Augmented Generation (RAG) method is combined with a graph-based structure to extract more accurate and relevant information for text generation.

Why is GraphRAG Important?

Intelligent Information Retrieval: Using graphs to optimize knowledge search
Stronger Generative Models: Combining RAG and GPT to improve response accuracy
Processing Complex Data: Suitable for scientific and medical search systems


🔧 Installation & Requirements

Dependencies

Libraries used in this project:

  • torch
  • transformers
  • faiss-cpu
  • networkx
  • numpy
  • scipy
  • pandas
  • matplotlib

First, install the necessary packages. Then, run the notebook in Jupyter Notebook.

Run the Notebook

jupyter notebook graphrag.ipynb

🎯 Applications of These Projects

Medical Applications: Analyzing scientific articles, aiding disease diagnosis
Search Systems: Optimizing scientific and medical search engines
Research Assistance: Gathering more accurate information for medical research
Improving Chatbots: Enhancing medical chatbots for more precise responses


  • If you have suggestions for improving the project, please submit a Pull Request.
  • To report issues, please open an Issue.

About

Fine-tuning GPT-2 on domain-specific articles related to skin cancer, using PubMed data and Hugging Face's Transformers. The model is trained to generate relevant medical responses

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors