Paper and code released: Multi-LLM Collaborative Alignment via Stackelberg Games
shangbinfeng · x · 2026-10-20
The Stackelberg Alignment paper (arXiv:2609.39076) and code are now public. Co-led by Christina Hahn and Shangbin Feng with Dean Light, Swastik Roy, Hila Gonen, and Yulia Tsvetkov, building on Sparta Alignment. The abstract notes existing methods sample instructions uniformly even as their usefulness shifts with model progress; the framework instead uses an EXP3 bandit to adaptively allocate sampling budget by difficulty and discriminativeness, with Elo-style reputation-weighted peer judgment, achieving the highest macro-average across 3 model pools and 12 benchmarks.
More from Research
- Turing Post's Must-Read Papers of the Week: Cortex, Librarian Agents, Mosaic and 12 More — TheTuringPost · 2026-10-20
- MTSQL-R1 Targets the Short-Horizon Habits That Break Multi-Turn Text-to-SQL — burkov · 2026-10-20
- Robotics take: IK guarantees no precision edge, hardware—not the solver—is the bottleneck — ChongZzZhang · 2026-10-20
- MPC vs RL for robot tracking: the precision bottleneck is hardware, not the solver — ChongZzZhang · 2026-10-20
- Developer Spends 1.6M Tokens Having Codex Benchmark MPC vs RL for Arm Tracking — ChongZzZhang · 2026-10-20
- Terminal Bench 4 audit: 44 of 65 tasks cheatable, Sonnet 5.5 hits 89.2% after fixes — scaling01 · 2026-10-20