{"ID":2877328,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2508.21066","arxiv_id":"2508.21066","title":"OneReward: Unified Mask-Guided Image Generation via Multi-Task Human Preference Learning","abstract":"In this paper, we introduce OneReward, a unified reinforcement learning framework that enhances the model's generative capabilities across multiple tasks under different evaluation criteria using only \\textit{One Reward} model. By employing a single vision-language model (VLM) as the generative reward model, which can distinguish the winner and loser for a given task and a given evaluation criterion, it can be effectively applied to multi-task generation models, particularly in contexts with varied data and diverse task objectives. We utilize OneReward for mask-guided image generation, which can be further divided into several sub-tasks such as image fill, image extend, object removal, and text rendering, involving a binary mask as the edit area. Although these domain-specific tasks share same conditioning paradigm, they differ significantly in underlying data distributions and evaluation metrics. Existing methods often rely on task-specific supervised fine-tuning (SFT), which limits generalization and training efficiency. Building on OneReward, we develop Seedream 3.0 Fill, a mask-guided generation model trained via multi-task reinforcement learning directly on a pre-trained base model, eliminating the need for task-specific SFT. Experimental results demonstrate that our unified edit model consistently outperforms both commercial and open-source competitors, such as Ideogram, Adobe Photoshop, and FLUX Fill [Pro], across multiple evaluation dimensions. Code and model are available at: https://one-reward.github.io","short_abstract":"In this paper, we introduce OneReward, a unified reinforcement learning framework that enhances the model's generative capabilities across multiple tasks under different evaluation criteria using only \\textit{One Reward} model. By employing a single vision-language model (VLM) as the generative reward model, which can...","url_abs":"https://arxiv.org/abs/2508.21066","url_pdf":"https://arxiv.org/pdf/2508.21066v1","authors":"[\"Yuan Gong\",\"Xionghui Wang\",\"Jie Wu\",\"Shiyin Wang\",\"Yitong Wang\",\"Xinglong Wu\"]","published":"2025-08-28T17:59:46Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Reinforcement Learning\",\"Language Model\"]","has_code":false}