paper-with-me

홈 › Papers

When Less Latent Leads to Better Relay: Information-Preserving Compression for Latent Multi-Agent LLM Collaboration

2026-04-14 · Yiping Li, Zhiyu An, Wan Du arxiv

Communication in Large Language Model (LLM)-based multi-agent systems is moving beyond discrete tokens to preserve richer context. Recent work such as LatentMAS enables agents to exchange latent messages through full key-value (KV) caches. However, full KV relay incurs high memory and communication cost. We adapt KV-cache eviction methods to this setting and introduce \textbf{Orthogonal BackFill (OBF)} to mitigate information loss from hard eviction. OBF injects a low-rank orthogonal residual from discarded KV states into the retained KV states. We evaluate OBF against full KV relay on nine benchmarks spanning mathematical reasoning, expert and commonsense QA, and coding. With only 9.9%-20.2% of the prompt KV states retained, H-OBF delivers between $97%$ and $120%$ of full KV relay's per-benchmark accuracy across the nine benchmarks. This suggests that more information does not necessarily lead to better communication; preserving the most useful information matters more. Our codebase is included in the supplementary material. Our codebase is publicly available on https://github.com/markli404/When-Less-Latent-Leads-to-Better-Relay.

📄 PDF Abstract BibTeX arXiv:2604.13349

Code (0)

등록된 구현이 없습니다.

Tasks

Mathematical Reasoning

Similar Papers 제목 키워드 기반

Learned Relay Representations for Forward-Thinking Discrete Diffusion Models

2026-05-21 · Benjamin Rozonoyer, Jacopo Minniti, Dhruvesh Patel, Neil Band 외 arxiv

When Masked Diffusion Models (MDMs) generate sequences through iterative refinement, the rich internal computation over masked positions is discarded, forcing every subsequent refinement step to recompute the valuable in…

On the design of multiple-relay cooperative MIMO networks with partial channel state information

2020-03-03 · Donatella Darsena, Giacinto Gelli, Francesco Verde

This paper deals with the problem of jointly designing the source precoder, the relaying matrices, and the destination equalizer in a multiple-relay amplify-and-forward (AF) cooperative multiple-input multiple-output (MI…

Delay Adaptation Method for Relay Assisted Optical Wireless Systems

2020-04-30 · Yu Zeng, Sanaa Hamid Mohamed, T. E. H. El-Gorashi, Jaafar M. H. Elmirghani

In this paper, we investigate optical wireless repeaters as relay terminals between a transmitter and a user in an Infrared Optical Wireless Communication (IROWC) system. A delay adaptation method is introduced to solve …

When Does Latent Communication Pay? A Causal Audit of Relayed KV Caches in Multi-Agent LLMs

2026-08-05 · Jiaming Cheng, Subhransu Das, Rajiv Ramnath arxiv

Multi-agent LLM systems relay key-value caches instead of text and credit their gains to exchanged "latent thoughts". That credit is a claim about which example's cache is relayed, not merely that one is. We audit it cau…

Can a Multi-Hop Link Relying on Untrusted Amplify-and-Forward Relays Render Security?

2019-05-17 · Milad Tatar Mamaghani, Ali Kuhestani, Hamid Behroozi

Cooperative relaying is utilized as an efficient method for data communication in wireless sensor networks and the Internet of Things (IoT). However, sometimes due to the necessity of multi-hop relaying in such communica…