What retrieval strategies are included?

It covers dense retrieval, sparse retrieval (BM25), hybrid search, multi-query variations, and contextual compression.

Does it support different chunking methods?

Yes, it supports recursive character splitting, token-based splitting, markdown header splitting, and advanced semantic chunking.

Can I use this for local AI development?

Yes, it includes specific configurations for local vector stores like Chroma and local embedding models using libraries like Sentence Transformers.

What vector databases are supported by this skill?

The skill provides patterns and configurations for Pinecone, Weaviate, Milvus, Chroma, Qdrant, and FAISS.

How does this RAG skill help reduce LLM hallucinations?

It implements grounded response patterns and prompt engineering that forces the LLM to use retrieved context and provide source citations.

RAG Implementation

Name: RAG Implementation
Author: Kingly-Agency

byKingly-Agency

数据科学与机器学习

Implements advanced Retrieval-Augmented Generation (RAG) workflows to build knowledge-grounded LLM applications and semantic search systems.

关于

This skill provides a comprehensive framework for building robust Retrieval-Augmented Generation (RAG) systems. It empowers developers to integrate external knowledge bases with LLMs through specialized document chunking, vector database management, and advanced retrieval patterns. Whether you are building internal documentation assistants or high-performance semantic search tools, this skill guides you through implementing hybrid search, reranking, and grounded response patterns to ensure accuracy and minimize hallucinations in AI applications.

主要功能

0 GitHub stars
Support for major vector databases like Pinecone, Weaviate, Chroma, and Milvus
Reranking integration using Cross-Encoders and API-based solutions like Cohere
Advanced retrieval patterns including Hybrid Search, Multi-Query, and Contextual Compression
Evaluation frameworks for measuring accuracy, retrieval quality, and groundedness
Sophisticated document chunking including recursive, token-based, and semantic strategies

使用场景

Creating documentation assistants with source citations and factual grounding
Implementing semantic search features for large-scale content repositories
Building proprietary Q&A systems over internal company documentation

关于

主要功能

0 GitHub stars
Support for major vector databases like Pinecone, Weaviate, Chroma, and Milvus
Reranking integration using Cross-Encoders and API-based solutions like Cohere
Advanced retrieval patterns including Hybrid Search, Multi-Query, and Contextual Compression
Evaluation frameworks for measuring accuracy, retrieval quality, and groundedness
Sophisticated document chunking including recursive, token-based, and semantic strategies

使用场景

Creating documentation assistants with source citations and factual grounding
Implementing semantic search features for large-scale content repositories
Building proprietary Q&A systems over internal company documentation