paper-with-me

Papers

Principled RL for Flow Matching Emerges from the Chunk-level Policy Optimization

2025-10-24 · Yifu Luo, Haoyuan Sun, Xinhao Hu, Penghui Du, Keyu Fan, Bo Li, Sinan Du, Xu Wan, Zhiyu Chen, Bo Xia, Yongzhe Chang, Changqian Yu, Kun Gai, Tiantian Zhang, Xueqian Wang arxiv

Recent Progress in post-training flow matching for text-to-image (T2I) generation with Group Relative Policy Optimization (GRPO) has demonstrated strong potential. However, it is hindered by a critical limitation: inaccurate advantage attribution. In this work, we argue that aggregating consecutive steps into a coherent 'chunk' and shifting the policy optimization paradigm from GRPO's step level to the chunk level can effectively mitigate the negative impact of this issue. Building on this insight, we propose Group Chunking Policy Optimization (GCPO), the first chunk-level reinforcement learning approach for post-training flow matching. Extensive experiments demonstrate that GCPO achieves superior performance on both standard T2I benchmarks and preference alignment, with up to 43% relative gains over GRPO, highlighting the promise of chunk-level policy optimization. The code is available on https://github.com/xingzhejun/GCPO.

📄 PDF Abstract BibTeX arXiv:2510.21583

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

Next Tokens Denoising for Speech Synthesis

2025-07-30 · Yanqing Liu, Ruiqing Xue, Chong Zhang, Yufei Liu 외 arxiv

While diffusion and autoregressive (AR) models have significantly advanced generative modeling, they each present distinct limitations. AR models, which rely on causal attention, cannot exploit future context and suffer …

Speech Synthesis

FocalPolicy: Frequency-Optimized Chunking and Locally Anchored Flow Matching for Coherent Visuomotor Policy

2026-05-15 · Qian He, Zhenshuo Yang, Wenqi Liang, Chunhui Hao 외 arxiv

Visuomotor policies aim to learn complex manipulation tasks from expert demonstrations. However, generating smooth and coherent trajectories remains challenging, as it requires balancing proximal precision with distal fo…

EverAnimate: Minute-Scale Human Animation via Latent Flow Restoration

2026-05-14 · Wuyang Li, Yang Gao, Mariam Hassan, Lan Feng 외 arxiv

We propose EverAnimate, an efficient post-training method for long-horizon animated video generation that preserves visual quality and character identity. Long-form animation remains challenging because highly dynamic hu…

Video Generation

SEAM: Smooth Execution of Action-Chunked Motion for Vision-Language-Action Policies

2026-07-06 · Dijia Zhan, Xuemiao Xu, Jinyi Li, Jie Tang arxiv

Vision-Language-Action (VLA) policies that execute fixed-length action chunks can exhibit multimodal bifurcation: a cross-chunk inconsistency in which adjacent chunks generated from independent Gaussian latents can conve…

Guided Action Flow: Q-Guided Inference for Flow-Matching Vision-Language-Action Policies

2026-07-02 · Liuhaichen Yang, Zhuang Jiang, Chenchao Sheng, Zezhi Tang arxiv

Flow-matching vision-language-action policies generate robot action chunks through an iterative transport process, creating an opportunity for test-time guidance without retraining the base policy. We study this opportun…