Multi-agent RL recipe turns pass@k into pass@1 on hard chess puzzles
shizhediao · x · 2026-11-20
The Inkling team showed that a simple multi-agent RL recipe can train a team of Inkling 2 Small agents to cooperate, communicate, and delegate work on hard chess puzzles without predefined roles — converting pass@k exploration into reliable pass@1 performance. The authors are pushing multi-agent RL as an open research direction.
Related event: Multi-Agent RL Recipe Cracks Chess Puzzles Without Predefined Roles(2 posts)→
More from coding & agent
- Ex-Googler open-sources a CLI that lets Claude Code and Codex chat in one room — thechrisperry · 2026-11-20
- Vibe-coding day 17: a WebGPU kelp lantern built with Gemini in Antigravity — AI_Andrew · 2026-11-20
- Beaker autonomously finds 100+ grading errors across five popular AI benchmarks — VictorSanh · 2026-11-20
- Burkov spots Codex quirk: suggested follow-up knows the agent screwed up, but the agent doesn't — burkov · 2026-11-20
- QuixiInfer author: tell Claude 'defeat is not an option' to make it write kernels for your GPU — QuixiAI · 2026-11-20
- OpenAI's Codex repo has 27,802 open issues: agent harness maintenance compared — SuaveSteve · 2026-11-20