method · first seen 4 Oct 2026
minimax preference optimization
Also: adversarial preference distribution
Formulating alignment as a minimax game where an adversary reweights preferences the policy serves worst.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Which Preferences to Train On? End-to-End Multi-Objective Alignment with an Adversarial Preference DistributionPapers · 4 Oct 2026 · Minjae Lee, Kyunghyun Cho, Sangdon Park
Latest sources
- Which Preferences to Train On? End-to-End Multi-Objective Alignment with an Adversarial Preference DistributionPapers · 4 Oct 2026 · Minjae Lee, Kyunghyun Cho, Sangdon Park