This repository serves as a central hub for all my Retrieval-Augmented Generation (RAG) experiments and implementations.
Each project explores a different technique, enhancement, or research direction within the RAG ecosystem — ranging from simple baselines to advanced agentic and self-refining pipelines.
| # | Name | Description | Link |
|---|---|---|---|
| 1 | Agentic_RAG | Agent-powered RAG system integrating planning, tool use, and dynamic multi-step reasoning. | https://github.com/supraja777/Agentic_RAG |
| 2 | Self-RAG | Implements self-reflective refinement where the model evaluates and improves its own retrieved context and responses. | https://github.com/supraja777/Self-RAG |
| 3 | CRAG | Corrective RAG: applies post-retrieval validation and filtering to improve factual grounding and reduce noisy context. | https://github.com/supraja777/CRAG |
| 4 | RAG-Reranking | Enhances retrieval via cross-encoder–based re-ranking to deliver higher-quality context before generation. | https://github.com/supraja777/RAG-Reranking |
| 5 | Reliable-RAG | Focuses on stability and consistency — introducing checks to reduce hallucination and improve response robustness. | https://github.com/supraja777/Reliable-RAG |
| 6 | Hypothetical-Document-Embedding | Implements HDE-based retrieval using synthetic text embeddings to improve semantic similarity search. | https://github.com/supraja777/Hypothetical-Document-Embedding |
| 7 | Simple-RAG | A clean and minimal baseline RAG pipeline implementing standard retrieval + generation. | https://github.com/supraja777/Simple-RAG |
| 8 | Query-Transformations-RAG | Demonstrates query rewriting, step-back prompting, and sub-query decomposition to enhance retrieval in RAG pipelines. | https://github.com/supraja777/Query-Transformations-RAG |