, spatial patches or temporal frames) caused success or failure.
Video RL improves by assigning credit to individual visual tokens
TVRL uses gradient magnitudes from a frozen vision-language model to weight denoising updates per token, outperforming video-level baselines on VBench-2.0.
Academic
Yifan Wang · Gordon Guocheng Qian · Yanyu Li · Anil Kag · Yun Fu
Northeastern University
Research Digest··3 min read
The authors introduce TVRL, a framework that derives token-level credit for video reinforcement learning from the same reward signal.
Why this paper
From Northeastern University
In one line
TVRL uses gradients from a frozen vision-language model to allocate token-level credit, outperforming scalar-reward GRPO on video generation.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§