SOURCE-LINKED INTELLIGENCE
GraftOOD: Grafting Out-of-Distribution Exemplars into Advantage-Weighted Rectification for Video Reward Optimization
Video reward optimization can increase proxy rewards while reducing video dynamics, a mismatch we identify as \emph{motion collapse}, including under a video-level motion-quality reward. We propose \textbf{GraftOOD}, which replaces part of the online rollouts in selected prompt groups with out-of-distribution video exemplars. A shared velocity-regression objective combines group-relative reward rectification with sequence-level exemplar supervision, requiring neither exemplar generation trajectories nor an additional supervised-loss coefficient. We characterize its pointwise optimal field as a
Read original source ↗ Open in workspace
- recordType
- paper
- region
- Global
Evidence & attribution
- arXiv · AI, language, vision and robotics · 2026-10-09T12:06:18.000Z
First collected: 2026-10-13T16:12:24.076Z. This is not the publication date.