Optimize RAG pipelines with HyDE, reranking, and hybrid search.
npx skills add https://github.com/JNZader/javi-ai --skill rag-advanced
Please help me install this Agent Skill. Skill: rag-advanced Source: https://github.com/JNZader/javi-ai/tree/main/own/skills/rag-advanced Command: npx skills add https://github.com/JNZader/javi-ai --skill rag-advanced
Advanced RAG patterns address the limitations of basic retrieval by combining hypothetical document embeddings, robust reranking, and hybrid search to significantly improve answer quality in production settings.
Configure a production RAG workflow that enables HyDE, multi-query retrieval, contextual compression, and hybrid search to begin improving retrieval performance.
High-intent search queries and answers about installing and using this skill.
HyDE, multi-query retrieval, and contextual compression improve retrieval-augmented generation accuracy within LangChain pipelines by refining dense and sparse signal gathering to significantly boost answer quality.
HyDE, or hypothetical document embeddings, enhances vector search recall by generating synthetic documents to augment retrieval. This technique boosts context gathering before applying cross-encoder reranking to improve result quality.
Combine dense and sparse search signals using reciprocal rank fusion (RRF) within a hybrid search configuration. This approach maximizes retrieval performance across large document collections in production RAG pipelines.
Use cross-encoder reranking in a production RAG pipeline when retrieval quality and answer accuracy across large document collections are critical. It applies contextual compression to filter and prioritize the most relevant chunks.
You can configure flexible document chunking strategies to balance latency and context within hybrid search pipelines. This setup supports parent document retrieval and multi-query operations for robust context gathering.
Basic retrieval falls short for large document collections by lacking robust context gathering. Applying advanced patterns like HyDE, cross-encoder reranking, and hybrid search overcomes these limitations to ensure high answer accuracy.
Alex1980AlexMigrate, configure, and optimize Qdrant vector stores and collections.
Miosa-osaEvaluate RAG pipelines with retrieval and generation metrics plus bottleneck analysis.
floflo777Generate RAG project scaffolds for LangChain, LlamaIndex, and vector stores.
yonatangrossImplement RAG patterns with hybrid search, reranking, and PGVector integration.
