RAG_Type

---

Project Overview

RAG_Type is a Python-based implementation of Retrieval-Augmented Generation (RAG) techniques, designed to enhance information retrieval and generation processes. This project focuses on adaptive retrieval, semantic chunking, hierarchical indexing, explainability, and feedback loops to improve the performance of RAG models.

---

Key Features

1. Adaptive Retrieval
   Dynamically adjusts the retrieval process based on the query requirements.

2. Explainable Retrieval
   Provides insights into why specific chunks or documents were retrieved.

3. Feedback Loop Retrieval
   Incorporates user feedback to iteratively refine retrieval and generation processes.

4. Fusion Retrieval
   Combines multiple retrieval strategies or sources for improved relevance and coverage.

5. Hierarchical RAG
   Uses a multi-level retrieval approach (e.g., sections, paragraphs) to match query depth effectively.

6. Query Transform Retrieval
   Reformulates user queries using paraphrasing or expansion for better document matching.

7. Semantic RAG
   Retrieves documents based on semantic similarity using embeddings rather than keyword matches.

8. Self RAG
   Enables the model to self-assess and refine its own retrieval outputs for enhanced accuracy.

9. Simple RAG
   A basic retrieval-augmented generation method using standard similarity-based retrieval.

---

Repository Structure

File/Folder	Description
adaptive_retrieval.py	Implements adaptive retrieval techniques.
explainable_retrieval.py	Adds explainability to the retrieval process.
semantic_chunking.py	Handles semantic chunking of documents.
hierarchical_indices.py	Builds hierarchical FAISS indices for document retrieval.
fusion_retrieval.py	Implements fusion-based retrieval strategies.
feedback_loop.py	Incorporates user feedback into the system.
Main.py	Main script to run the RAG pipeline.
Eval.py	Evaluation script for measuring model performance.
Excel_save.py	Saves evaluation results in Excel format.
pdf_loader.py	Loads PDF documents for processing.
Understanding_Climate_Change.pdf	Example dataset used for testing purposes.

---

Installation

Step 1: Clone the Repository

git clone https://github.com/vanshksingh/RAG_Type.git
cd RAG_Type

Step 2: Install Dependencies

pip install -r requirements.txt

Step 3: Install FAISS (for efficient indexing)

pip install faiss-cpu

---

Usage Instructions

1. Load your dataset using pdf_loader.py.
2. Configure retrieval settings in the respective scripts (adaptive_retrieval.py, etc.).
3. Run the main pipeline:

   python Main.py

4. Evaluate results:

   python Eval.py

---

Example Dataset

The repository includes an example PDF file, Understanding_Climate_Change.pdf, to test and demonstrate the functionality of RAG_Type.

---

Contributing Guidelines

We welcome contributions! Follow these steps to contribute:

1. Fork this repository.
2. Create a feature branch:

   git checkout -b feature-name

3. Commit your changes and push them to your forked repository.
4. Submit a pull request describing your changes.

---

License

This project is licensed under the MIT License.

---

Contact

For issues or suggestions, feel free to open a GitHub issue in this repository!