method · first seen 4 Oct 2026
Nash learning from human feedback
Also: NLHF
Preference alignment that learns a lottery policy converging to a Nash equilibrium of pairwise preferences.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Groupwise Distortion Guarantees for Preference-Based AlignmentPapers · 4 Oct 2026 · Jacob Brodkey, Roberto Tamez, Aaron Roth
Latest sources
- Groupwise Distortion Guarantees for Preference-Based AlignmentPapers · 4 Oct 2026 · Jacob Brodkey, Roberto Tamez, Aaron Roth