TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

HippoRAG: Neurobiologically Inspired Long-Term Memory for LLMs (2024)

65 pointsby veryluckyxyz3 months ago

3 comments

carschno3 months ago
I am not a biologist, but the first sentence seems bold to me:<p>&gt; Millions of years of evolution have led mammalian brains to develop the crucial ability to store large amounts of world knowledge and continuously integrate new experiences without losing previous ones.<p>My impression has always been that humans have been good at selective forgetting, hence keeping relevant memories and dropping others.<p>Edit: it looks like none of the authors has a biological background either. How serious do they mean the &quot;neurobiologically inspired&quot; claim?
评论 #42970375 未加载
jumploops3 months ago
&gt; HippoRAG [..] builds a KG from scratch using LLMs and performs multi-hop retrieval without any supervision<p>They use GPT-3.5 Turbo[0] for entity extraction when populating the knowledge graph.<p>Curious how well GPT-4&#x2F;Claude would do here?<p>[0]<a href="https:&#x2F;&#x2F;arxiv.org&#x2F;html&#x2F;2405.14831v3#A3" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;html&#x2F;2405.14831v3#A3</a>
thawab3 months ago
This was released on 23 May 2024, and the source code is on github[0]. It didn&#x27;t pickup traction like other RAG approaches. I haven&#x27;t tested it so i don&#x27;t have an opinion on it.<p>[0] <a href="https:&#x2F;&#x2F;github.com&#x2F;OSU-NLP-Group&#x2F;HippoRAG">https:&#x2F;&#x2F;github.com&#x2F;OSU-NLP-Group&#x2F;HippoRAG</a>