推荐jsenn · 2025-01
“You might be interested in HippoRAG [1] which takes a graph-based approach similar to what you’re suggesting here.”
推荐 HippoRAG,其基于图的方法与对方设想类似
仿海马体索引理论,结合知识图谱与 PageRank 做检索增强
提出受人类长期记忆海马体索引理论启发的检索框架 HippoRAG,协同使用 LLM、知识图谱和个性化 PageRank 算法,模拟新皮层与海马体的不同作用,以更深入高效地整合新经验。在多跳问答上比现有 RAG 方法最多提升 20%;单步检索即可达到与 IRCoT 等迭代检索相当或更好的效果,同时便宜 10–30 倍、快 6–13 倍。
HN 上的人在讨论为 LLM 构建长期记忆时推荐它,认为它是基于图的做法,并提到它用 GPT-3.5 Turbo 抽取实体来构建知识图谱。
“You might be interested in HippoRAG [1] which takes a graph-based approach similar to what you’re suggesting here.”
“They use GPT-3.5 Turbo[0] for entity extraction when populating the knowledge graph.”
“seems odd to miss out this exactly similarly named paper with related techniques.”