s³ paper: cooperative multi-agent RL post-training beats best-of-N by up to 166%
StellaLisy · x · 2026-11-07
Natasha Jaques' team questions today's swarm hype: most work builds prescribed coordination harnesses. Their new s³ paper instead lets agents learn emergent coordination and communication protocols purely from task reward.
- First work showing cooperative multi-agent RL post-training of LLM swarms outperforming single-agent best-of-N sampling
- Asynchronous training with no leader, no roles, no turn order — one shared reward
- Up to 166% improvement in token efficiency and success rate over compute-balanced baselines
- Generalizes to swarms up to 3x larger than the training size
The authors frame this as opening a research agenda on emergent multi-agent communication and coordination.
Related event: UW and Ai2 Propose s³: Multi-agent RL Enables Self-organizing LLM Swarms(4 posts)→
More from coding & agent
- Open DCR on remote MCP servers invites auth-code hijacking: should you allowlist redirect URIs? — After_Pumpkin3803 · 2026-11-07
- Building an incident-response agent: why memory alone can't be trusted for operational warnings — noufal_rizwan · 2026-11-07
- Letting AI stay on the line after customers hang up turned post-call notes into actions — gsuiteautomations · 2026-11-07
- Tsinghua's Coca-Cola compresses agent tool outputs, cutting tokens up to 89% while boosting accuracy — Tsinghua · 2026-11-07
- ComfyUI-freefield puts Freefield's 117 workflows in a sidebar with one-click model downloads — neuregex · 2026-11-07
- How do you stop micromanaging a custom agent harness? A Reddit dev's research-task dilemma — dangerous_inference · 2026-11-07