| title | RAVEN: Real-time Autoregressive Video Extrapolation with Consistency-model GRPO |
|---|---|
| authors | Yanzuo Lu, Ronglai Zuo, Jiankang Deng |
| date | 2026-05-14 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2605.15190 |
| affiliation | MVP Lab |
| upvotes | 14 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf code |
| abstract | Causal autoregressive video diffusion models support real-time streaming generation by extrapolating future chunks from previously generated content. Distilling such generators from high-fidelity bidirectional teachers yields competitive few-step models, yet a persistent gap between the history distributions encountered during training and those arising at inference constrains generation quality over long horizons. We introduce the Real-time Autoregressive Video Extrapolation Network (RAVEN), a training-time test framework that repacks each self rollout into an interleaved sequence of clean historical endpoints and noisy denoising states. This formulation aligns training attention with inference-time extrapolation and allows downstream chunk losses to supervise the history representations on which future predictions depend. We further propose Consistency-model Group Relative Policy Optimization (CM-GRPO), which reformulates a consistency sampling step as a conditional Gaussian transition and applies online Reinforcement Learning (RL) directly to this kernel, avoiding the Euler-Maruyama auxiliary process adopted in prior flow-model RL formulations. Experiments demonstrate that RAVEN surpasses recent causal video distillation baselines across quality, semantic, and dynamic degree evaluations, and that CM-GRPO provides further gains when combined with RAVEN. |
cite
@misc{lu2026raven,
title={RAVEN: Real-time Autoregressive Video Extrapolation with Consistency-model GRPO},
author={Yanzuo Lu and Ronglai Zuo and Jiankang Deng},
year={2026},
eprint={2605.15190},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2605.15190}
}