self-disagreement

Tag

Cards List
#self-disagreement

Benchmarks Don't Talk Back: Learning Through Self-Disagreement

arXiv cs.CL ↗ · yesterday Cached

The paper introduces 'dialectical competence' — knowing when to hold or yield a position under counter-reasoning — and proposes Self-Disagreement, an RLVR method where a model generates its own antithesis and is trained on ruling over it, with no second model, reward model, or human annotation.

0 favorites 0 likes
← Back to home

Submit Feedback