KVPO: ODE-Native GRPO for Autoregressive Video Alignment via KV Semantic Exploration

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Ruicheng, Cong, Kaixi, Zhou, Jun, Zhong, Zhizhou, Xu, Zunnan, Mao, Shuiyang, Liu, Wei, Li, Xiu
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917494165340160
author Zhang, Ruicheng
Cong, Kaixi
Zhou, Jun
Zhong, Zhizhou
Xu, Zunnan
Mao, Shuiyang
Liu, Wei
Li, Xiu
author_facet Zhang, Ruicheng
Cong, Kaixi
Zhou, Jun
Zhong, Zhizhou
Xu, Zunnan
Mao, Shuiyang
Liu, Wei
Li, Xiu
contents Aligning streaming autoregressive (AR) video generators with human preferences is challenging. Existing reinforcement learning methods predominantly rely on noise-based exploration and SDE-based surrogate policies that are mismatched to the deterministic ODE dynamics of distilled AR models, and tend to perturb low-level appearance rather than the high-level semantic storyline progression critical for long-horizon coherence. To address these limitations, we present KVPO, an ODE-native online Group Relative Policy Optimization (GRPO) framework for aligning streaming video generators. For diversity exploration, KVPO introduces a causal-semantic exploration paradigm that relocates the source of variation from stochastic noise to the historical KV cache. By stochastically routing historical KV entries, it constructs semantically diverse generation branches that remain strictly on the data manifold. For policy modeling, KVPO introduces a velocity-field surrogate policy based on Trajectory Velocity Energy (TVE), which quantifies branch likelihood in flow-matching velocity space and yields a reward-weighted contrastive objective fully consistent with the native ODE formulation. Experiments on multiple distilled AR video generators demonstrate consistent gains in visual quality, motion quality, and text-video alignment across both single-prompt short-video and multi-prompt long-video settings.
format Preprint
id arxiv_https___arxiv_org_abs_2605_14278
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle KVPO: ODE-Native GRPO for Autoregressive Video Alignment via KV Semantic Exploration
Zhang, Ruicheng
Cong, Kaixi
Zhou, Jun
Zhong, Zhizhou
Xu, Zunnan
Mao, Shuiyang
Liu, Wei
Li, Xiu
Computer Vision and Pattern Recognition
Aligning streaming autoregressive (AR) video generators with human preferences is challenging. Existing reinforcement learning methods predominantly rely on noise-based exploration and SDE-based surrogate policies that are mismatched to the deterministic ODE dynamics of distilled AR models, and tend to perturb low-level appearance rather than the high-level semantic storyline progression critical for long-horizon coherence. To address these limitations, we present KVPO, an ODE-native online Group Relative Policy Optimization (GRPO) framework for aligning streaming video generators. For diversity exploration, KVPO introduces a causal-semantic exploration paradigm that relocates the source of variation from stochastic noise to the historical KV cache. By stochastically routing historical KV entries, it constructs semantically diverse generation branches that remain strictly on the data manifold. For policy modeling, KVPO introduces a velocity-field surrogate policy based on Trajectory Velocity Energy (TVE), which quantifies branch likelihood in flow-matching velocity space and yields a reward-weighted contrastive objective fully consistent with the native ODE formulation. Experiments on multiple distilled AR video generators demonstrate consistent gains in visual quality, motion quality, and text-video alignment across both single-prompt short-video and multi-prompt long-video settings.
title KVPO: ODE-Native GRPO for Autoregressive Video Alignment via KV Semantic Exploration
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.14278