RSM agent memory hits 83% of full-context quality at 32% token cost on a 4k budget
dair_ai · x · 2026-09-09
- dair-ai highlights RSM, a long-horizon agent memory method that separates two things usually collapsed into one: how memories merge on write, and how retrieved content is packed into the prompt. Setting: tight 2k–5k token budgets where full-context prompting is infeasible.
- Method: cosine-gated max-member merge on the write side plus an atom-aware grouped packer on the read side. At a 4k budget it reaches 83% of full-context quality at 32% of the token cost.
- Ablations attribute gains to both halves (+5.7 pts over online k-means for merging, +5.0 pts over flat packing); it reproduces on RealMem, beating Budget-RAG, Streaming-Proto and A-MEM.
Related event: RSM Achieves 83% Memory Quality with 32% of Tokens(3 posts)→
More from coding & agent
- Magic Patterns' first engineer: Cursor in Slack, a Patterson review agent, repo clones over worktrees — Y Combinator Blog · 2026-10-28
- Dev builds AI job-search app for his wife, she lands a job, and he open-sources it — CurieuxExplorer · 2026-10-28
- Meshy partners with MCP for Blender: agents drop textured 3D models into your scene — sidahuj · 2026-10-28
- Supercharging LLM classification with logprobs: soft classifiers you can threshold-tune — JnBrymn · 2026-10-28
- Codex daily build series day 23: skipping iCloud plugins for a reset — dotey · 2026-10-28
- Cloudflare Access adds fully customizable pages with HTML + Liquid templates — dinasaur_404 · 2026-10-28