self-rag

A complete collection of RAG interview questions, answers (1509 questions & 52 RAG types), system design scenarios, architecture patterns, and production-ready concepts.

生产级 3GPP 5G 规范 RAG Agent：自然语言提问，回答带段落级原文引用 + 严格 grounding，覆盖 Rel-18/19 全部 TS。

Evidence-synthesis RAG assistant for TCM practitioners — hybrid vector + knowledge graph retrieval over 17 classical texts, with query classification, self-critique verification, and blind A/B arena evaluation.

Training code for advanced RAG techniques - Adaptive-RAG, Corrective RAG, RQ-RAG, Self-RAG, Agentic RAG, and ReZero. Reproduces paper methodologies to fine-tune LLMs via SFT and GRPO for adaptive retrieval, corrective evaluation, query refinement, self-reflection, and agentic search behaviors.

ATLAS — industrial Agentic RAG platform (LangGraph): routing, query decomposition, Corrective RAG, Self-RAG reflection. FastAPI + SSE, pluggable vector stores, auth, rate limiting, tracing, Docker, CI. Academic webinar kit on archive/academic-labs branch.

Self-Reflective Question Answering for Biomedical Reasoning. GRPO fine-tuning via QLoRA & Unsloth with rewards for correctness, relevance, groundness, utility & XML structure. Structured think → answer → self-reflection with context grading, relevance assessment & groundness evaluation. DeepEval LLM-as-a-Judge (GEval, Faithfulness, Relevancy).

RAG enhances LLMs by retrieving relevant external knowledge before generating responses, improving accuracy and reducing hallucinations.

Self-correcting agentic RAG with LangGraph — combines Adaptive, Corrective, and Self-RAG. Routes queries, grades retrieved documents, and checks answers for hallucinations. Runs fully local on Ollama.

Learn Retrieval-Augmented Generation (RAG) from scratch by manually building chunking, embeddings, retrieval, reranking, answerability, local generation, Graph-RAG, and Self-RAG.

Agentic RAG system with five workflow modes from Naive to Deep, routing each question to the cheapest path that answers it well, fusing dense and BM25 retrieval through Reciprocal Rank Fusion with CrossEncoder reranking, verifying groundedness and citations, and gating risky answers behind a human-in-the-loop approval queue.

基于 LangGraph 的 Agentic RAG 知识库助手，支持混合检索、Self-RAG、引用溯源和 RAGAS 自动评估。

LangGraph implementation of Self-RAG, based on “Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection” (Asai et al., ICLR 2024), implementing adaptive retrieval, evidence evaluation, generation, and self-critique inspired by the paper’s reflection-token architecture.

Self-RAG chatbot with LangGraph that retrieves from internal PDFs, verifies grounding (IsSUP), checks usefulness (IsUSE), and iteratively rewrites queries to improve answer quality.

Production-ready Retrieval-Augmented Generation (RAG) system with hybrid retrieval, Self-RAG agent workflows, cross-encoder reranking, and comprehensive benchmarking.

Backend Of STELLA that run SELF Retrieval-Augmented Generation [Senior Project]

企业级 Agentic RAG 技术文档问答系统

A self-correcting RAG agent in LangGraph — routes questions, grades its own retrieval, falls back to web search, and refuses answers that aren't grounded in its sources.

Most AI chatbots just guess answers based on a simple search. Enterprise RAG Orchestrator is built differently. It is an advanced and a highly secure AI-assistant for engineers. It combines document search with live database querying, protected by a 9-layer security wall and self-correcting logic.

A Streamlit-based Self-RAG-inspired evidence QA assistant with retrieval, evidence critique, answer revision, and reflection.