paper-with-me

홈 › Papers

UniGRPO: Unified Policy Optimization for Reasoning-Driven Visual Generation

2026-03-24 · Jie Liu, Zilyu Ye, Linxiao Yuan, Shenhan Zhu, Yu Gao, Jie Wu, Kunchang Li, Xionghui Wang, Xiaonan Nie, Weilin Huang, Wanli Ouyang arxiv

Unified models capable of interleaved generation have emerged as a promising paradigm, with the community increasingly converging on autoregressive modeling for text and flow matching for image generation. To advance this direction, we propose a unified reinforcement learning framework tailored for interleaved generation. We validate our approach on its fundamental unit: a single round of reasoning-driven image generation, where the model first expands the user prompt through reasoning, followed by image synthesis. Formulating this multimodal generation process as a Markov Decision Process with sparse terminal rewards, we introduce UniGRPO to jointly optimize text and image generation policies using GRPO. Adopting a minimalist methodology to avoid over-design, we leverage established training recipes for both modalities by seamlessly integrating standard GRPO for reasoning and FlowGRPO for visual synthesis. To ensure scalability to multi-round interleaved generation, we introduce two critical modifications to the original FlowGRPO: (1) eliminating classifier-free guidance to maintain linear, unbranched rollouts, which is essential for scaling to complex scenarios involving multi-turn interactions and multi-condition generation (e.g., editing); and (2) replacing the standard latent KL penalty with an MSE penalty directly on the velocity fields, providing a more robust and direct regularization signal to mitigate reward hacking effectively. Our experiments demonstrate that this unified training recipe significantly enhances image generation quality through reasoning, providing a robust and scalable baseline for the future post-training of fully interleaved models.

📄 PDF Abstract BibTeX arXiv:2603.23500

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learningmultimodal generationImage Generation

Similar Papers 제목 키워드 기반

MMaDA: Multimodal Large Diffusion Language Models

2025-05-21 · Ling Yang, Ye Tian, Bowen Li, Xinchen Zhang 외

We introduce MMaDA, a novel class of multimodal diffusion foundation models designed to achieve superior performance across diverse domains such as textual reasoning, multimodal understanding, and text-to-image generatio…

Image GenerationReinforcement Learning (RL)Text to Image GenerationText-to-Image Generation

FLAG-Trader: Fusion LLM-Agent with Gradient-based Reinforcement Learning for Financial Trading

2025-02-17 · Guojun Xiong, Zhiyang Deng, Keyi Wang, Yupeng Cao 외

Large language models (LLMs) fine-tuned on multimodal financial data have demonstrated impressive reasoning capabilities in various financial tasks. However, they often struggle with multi-step, goal-oriented scenarios i…

Decision Makingparameter-efficient fine-tuningReinforcement Learning (RL)

Beyond Token-Level Policy Gradients for Complex Reasoning with Large Language Models

2026-02-16 · Mufan Xu, Kehai Chen, Xuefeng Bai, Zhengyu Niu 외 arxiv

Existing policy-gradient methods for auto-regressive language models typically select subsequent tokens one at a time as actions in the policy. While effective for many generation tasks, such an approach may not fully ca…

Mathematical Reasoning

The Optimizer Is the Agent: Reasoning-Driven Search across Prompts, Programs, and ML Workflows

2026-08-07 · Junbo Li, Boyi Liu, Canwen Xu, Yite Wang 외 hf

Recent systems for optimizing prompts, programs, and ML workflows typically rely on explicit outer-loop controllers such as evolutionary search, bandits, or textual-gradient methods. We ask a fundamentally different ques…

Bottom-up Policy Optimization: Your Language Model Policy Secretly Contains Internal Policies

2025-12-22 · Yuqiao Tan, Minzheng Wang, Shizhu He, Huanxuan Liao 외 arxiv

Existing reinforcement learning (RL) approaches treat large language models (LLMs) as a unified policy, overlooking their internal mechanisms. In this paper, we decompose the LLM-based policy into Internal Layer Policies…

Reinforcement Learning