Shypd.aiThe largest open index of AI toolsHippoRAG is an open-source RAG framework that enhances LLMs' ability to continuously integrate knowledge across external documents. It is inspired by human long-term memory and utilizes knowledge graphs and personalized PageRank.
Trending
About
HippoRAG is a novel open-source RAG (Retrieval Augmented Generation) framework designed to equip Large Language Models (LLMs) with continuous knowledge integration capabilities, drawing inspiration from human long-term memory. It significantly improves LLMs' capacity to recognize and leverage connections within new information. The framework incorporates advanced techniques such as knowledge graphs and personalized PageRank to achieve superior performance in associativity (multi-hop retrieval) and sense-making, which involves integrating large and complex contexts. HippoRAG 2, an enhanced version, demonstrates improvements over existing RAG systems, including graph-based solutions like GraphRAG, RAPTOR, and LightRAG, while maintaining cost and latency efficiency for online processes and requiring fewer resources for offline indexing. It supports integration with OpenAI models and local deployments via vLLM, offering flexibility for various use cases.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
