cd /news/large-language-models/compact-memory-llm-agents-via-online… · home topics large-language-models article
[ARTICLE · art-122113] src=machinebrief.com ↗ pub= topic=large-language-models verified=true sentiment=· neutral

Compact-Memory LLM Agents via Online Max-Member Clustering and Atom-Aware Packing

Researchers introduced RSM-full, an online clustered-memory pipeline for long-horizon LLM deployments, achieving 83% of Full-Context quality at 32% of the token cost on the AMA-Bench benchmark at a 4k budget. The method outperformed Online K-Means by +3.5–6.0 percentage points across the ~2.6k–5k token regime and beat Budget-RAG by +0.69 pp on the RealMem benchmark, with gains driven by its cosine-gated max-member merge rule and atom-aware grouped context packer.

read1 min views1 publishedSep 7, 2026

arXiv:2609.04915v1 Announce Type: new Abstract: Many long-horizon LLM deployments face tight prompt budgets: latency, cost, and context limits make full-context prompting impractical as interaction length grows. The key question is then not raw recall alone, but which memory design gives the best quality--token trade-off in the compact-memory regime. We present \textbf{RSM-full}, an online clustered-memory pipeline designed for a strong quality--token Pareto point. RSM-full combines two design choices: a cosine-gated \emph{max-member merge} write rule and an atom-aware grouped context packer. On AMA-Bench, our primary compact-memory benchmark, it reaches $83%$ of Full-Context quality at $32%$ of the token cost at a $4$k budget; under four-seed averaging it beats the closest streaming-clustered baseline (Online K-Means) by $+3.5$--$6.0$,pp ($p{<}.001$) across the whole ${\sim}2.6$k--${\sim}5$k regime. Three-seed ablations show most of this gain comes from the merge rule ($+5.7$,pp over Online K-Means and matched-$\tau$ DP-means) and the grouped packer ($+5.0$,pp over flat concatenation). The pattern reproduces on RealMem, an independent long-horizon persona-memory benchmark: RSM-full improves on Budget-RAG ($+0.69$,pp, $p{=}.006$), is on par with BM25-RAG (paired $\Delta{=}{+}0.27$,pp, $p{=}.47$; we do \emph{not} claim BM25 equivalence in the equivalence-test sense), and significantly outperforms Streaming-Proto ($+2.97$,pp) and the closest reproduced 2025 agentic-memory baseline A-MEM ($+1.65$,pp, $p{<}.001$). Across benchmarks the message is consistent: under tight budgets, compact-memory performance is driven mainly by how streaming memories are merged and how retrieved content is assembled. Overall, RSM-full is most useful when answeroughly $2k$--$5k$ prompt tokens, where itdefines a strong compact-memory Pareto point; higher-token baselines remain stronger outside this regime.

── more in #large-language-models 4 stories · sorted by recency
── more on @rsm-full 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/compact-memory-llm-a…] indexed:0 read:1min 2026-09-07 ·