[HN Gopher] Infinite Retrieval: Attention enhanced LLMs in long-...
___________________________________________________________________
Infinite Retrieval: Attention enhanced LLMs in long-context
processing
Author : TaurenHunter
Score : 29 points
Date : 2025-03-01 19:37 UTC (3 hours ago)
(HTM) web link (arxiv.org)
(TXT) w3m dump (arxiv.org)
| smallnix wrote:
| Do I understand right this requires access to internals of the
| LLM and can not be used with todays models behind an API like
| ChatGPT or Claude?
| mirekrusin wrote:
| Innovation that would be applicable to open weight models
| running locally only would be awesome.
| briancleland wrote:
| This paper highlights something that should have been obvious:
| prediction and retrieval are two sides of the same coin. To
| predict effectively, you must first identify what's relevant.
| What's remarkable is that a 0.5B parameter model can perform
| perfect retrieval over 1M tokens when its natural attention
| patterns are leveraged properly.
|
| It raises an interesting question: what if we designed
| architectures explicitly around retrieval capabilities?
| Transformer architectures were designed for prediction, and
| retrieval emerged as a byproduct. What would an architecture
| optimized specfically for retrieval look like?
|
| A lot of money has been spent on building out large-scale RAG
| systems. If the performance improvements promised by the paper
| are real, the ramifications will be huge. Exciting to see that
| the authors are promising to release their code - it will be fun
| to how this model performs on consumer hardware.
| mirekrusin wrote:
| I think this could be expanded further. You can convert
| attention traces to knowledge graph with arbitrary and/or
| dynamic density. Traversing it can also be exotic - zooming
| in/expanding details at arbitrary points during traversal. With
| common format you can create topic/knowledge trace packs that
| could be shared, merged (subtracted?) etc.
___________________________________________________________________
(page generated 2025-03-01 23:01 UTC)