method · first seen 21 Oct 2026
reinforcement learning from self-judged tournaments
Training a policy with rewards assigned by its own pairwise comparisons of generated outputs in a tournament.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Consistency Is Not Validity: Self-Rewarded Moral Argumentation Makes Language Models Decisive but PliablePapers · 21 Oct 2026 · Marius De Kuthy Meurers, Aaron Parisi, Bernd Bohnet
Latest sources
- Consistency Is Not Validity: Self-Rewarded Moral Argumentation Makes Language Models Decisive but PliablePapers · 21 Oct 2026 · Marius De Kuthy Meurers, Aaron Parisi, Bernd Bohnet