AdaPilot introduces a scene-adaptive, cross-generator policy for optimizing text-to-image generation quality. By framing multi-turn image generation as a Markov Decision Process and using reinforcement learning, it decouples the policy from specific generator internals, incorporates scene-aware and process-level rewards, and achieves superior quality and generalization compared to baselines. Experiments demonstrate that a single AdaPilot policy can transfer zero‑shot to unseen generators while consistently improving performance across all evaluated models.
By Wenjin Liu, Fayuan Ke, Yue Lu, Zhe Cui, Anh Tuan Luu, Haoran Luo
The paper introduces Objective-aware Trajectory Credit Assignment (OTCA), a framework that refines reinforcement learning for diffusion-based visual generation. OTCA decomposes credit across denoising steps and allocates multiple reward signals adaptively, addressing the coarse, uniform reward assignment of existing GRPO pipelines. Experiments demonstrate that OTCA consistently enhances image and video generation quality across various metrics.
By Rui Li, Ke Hao, Yuanzhi Liang, Haibin Huang, Chi Zhang, Yun Gu, Xuelong Li
The paper introduces Reflection-Aware GRPO (RA‑GRPO), a reinforcement‑learning framework that aligns diffusion generative models with human preferences. It uses Diffusion Reflection to correct intermediate sampling paths by reversing the diffusion process, and Counterfactual Path Synthesis to embed these corrected trajectories into the policy, avoiding extra inference cost. Experiments on text‑to‑image and text‑to‑video models show RA‑GRPO outperforms existing methods, reducing reward hacking and improving generalization while remaining architecture‑agnostic.
By Junlong Wu, Jiuzhou Lin, Jia Sun, Boheng Zhang, Huaiqing Wang, Dewen Fan, Houde Liu, Qianqian Gan, Fan Yang, Tingting Gao
arXiv:2510. 21583v3 Announce Type: replace-cross Abstract: Recent Progress in post-training flow matching for text-to-image (T2I) generation with Group Relative Policy Optimization (GRPO) has demonstrated strong potential.
By 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:2607. 00535v1 Announce Type: cross Abstract: Few-step flow-map generators, such as consistency models and MeanFlow, accelerate sampling by directly learning long-range transport maps between noise and data.
By Zhiqi Li, Wen Zhang, Bo Zhu
arXiv:2605.26013v2 Announce Type: replace-cross
Abstract: We present AdvantageFlow, a forward-process reinforcement learning (RL) algorithm for rectified flow models. The algorithm minimizes an advan...
By Branislav Kveton, Anup Rao, Subhojyoti Mukherjee, Krishna Kumar Singh, Viet Dac Lai
The paper compares diffusion and rectified flow objectives within the MotionGPT3 framework for text-driven motion generation. Experiments on HumanML3D show that rectified flow converges faster, achieves strong test performance earlier, and matches or exceeds diffusion quality while requiring fewer sampling steps. The study isolates the generative objective’s impact, demonstrating that rectified flow’s benefits transfer to continuous-latent motion generation.
By Jaymin Bhan, JiHong Jeon, SangYeop Jeong
The paper introduces an adaptive step schedule controller for text‑to‑image diffusion models, allowing the number of denoising steps to vary based on the complexity of the input prompt. By mixing step schedules of different sizes and monitoring error discrepancies at each timestep, the method switches schedules to maintain image quality while reducing inference time. Experiments on COCO and DiffusionDB demonstrate that this approach achieves faster generation without sacrificing visual fidelity.
By Kuluhan Binici, Cihan Acar, Shivam Aggarwal, Siying Liu, Tulika Mitra
arXiv:2607. 02291v1 Announce Type: new Abstract: Conventional reinforcement learning strategies for visual generation typically employ sample-wise reward functions, yet this practice frequently results in reward hacking that degrades image diversity and introduces visual anomalies.
By Ruihang Li, Mengde Xu, Shuyang Gu, Leigang Qu, Fuli Feng, Han Hu, Wenjie Wang
The paper introduces a post‑training approach that enables a single inference process to transition from text reasoning to image synthesis, eliminating the need for explicit modality switching. Using the 14B BAGEL model, the authors demonstrate that targeted post‑training data and reward‑weighted training improve multimodal image generation across four independent T2I benchmarks. The study highlights the benefits of joint text‑image generation and strategic data selection for enhancing T2I performance.
By Jiahui Chen, Philippe Hansen-Estruch, Xiaochuang Han, Yushi Hu, Emily Dinan, Amita Kamath, Michal Drozdzal, Reyhane Askari-Hemmat, Luke Zettlemoyer, Marjan Ghazvininejad
arXiv:2604. 27147v3 Announce Type: replace-cross Abstract: In generative modeling, we often wish to produce samples that maximize a user-specified reward such as aesthetic quality or alignment with human preferences, a problem known as \textit{guidance}.
By Jerry Y. Huang, Justin Lin, Sheel Shah, Kartik Nair, Nicholas M. Boffi
arXiv:2606. 14792v1 Announce Type: cross Abstract: RL-based post-training has been widely adopted to enable interleaved visual and textual reasoning in unified multimodal models capable of both text and image generation.
By Yoonjeon Kim, Yuhta Takida, Chieh-Hsin Lai, Eunho Yang, Yuki Mitsufuji