Tencent's RewardVerse uses rubric-guided optimization to fix video reward model drift
tencent · hf · 2026-09-24
Tencent researchers introduce RewardVerse, a rubric-based video reward framework that tackles scalar drift in reward models for video-generation RL. A dynamic rubric serves as an intermediate representation between query and scorer, while the two-stage RGPO algorithm jointly optimizes rubric generation and scorer alignment. Experiments on the 16-dimensional EvalVerse benchmark show state-of-the-art pointwise and pairwise evaluation with a stable, interpretable reward signal.
More from Multimodal
- MiniMax H3-Image image model debuts on Tencent's Miora, free to try Nov 11–17 — aziz4ai · 2026-11-11
- MiniMax H3 hits 100 days open-source: 1,472 derivative models, 148M+ downloads — linoy_tsaban · 2026-11-11
- MiniMax H3-Image explodes an espresso machine photo into 15 labeled parts — Charlietest123 · 2026-11-11
- Indie dev built a tool that clips best moments from 4-hour streams in 30 minutes — kissaadams · 2026-11-11
- Non-Blender user makes a video in 30 minutes with Claude and Seedance — FinanceYF5 · 2026-11-11
- Whimsical Woodland: A Watercolor AI Art Prompt Template for Dreamy Forest Scenes — LudovicCreator · 2026-11-11