paper-with-me

홈 › Papers

SafeDiffusion-R1: Online Reward Steering for Safe Diffusion Post-Training

2026-05-18 · Komal Kumar, Ankan Deria, Abhishek Basu, Fahad Shamshad, Hisham Cholakkal, Karthik Nandakumar arxiv

Diffusion models have been widely studied for removing unsafe content learned during pre-training. Existing methods require expensive supervised data, either unsafe-text paired with safe-image groundtruth or negative/positive image pairs, making them impractical to scale. Furthermore, offline reinforcement learning and supervised fine-tuning approaches that generate synthetic data offline suffer from catastrophic forgetting, degrading generation quality. We propose a novel online reinforcement learning framework that addresses both data scarcity and model degradation through post-training with Group Relative Policy Optimization (GRPO) on both negative and positive text prompts. To eliminate the need for fine-tuning specialized safe/unsafe reward models, we introduce a \textit{steering reward mechanism} that exploits an inherent property of CLIP embeddings: steering text representations toward positive safety directions and away from negative ones in the embedding space. Our online-policy approach enables the model to learn from diverse prompts, including explicit unsafe content, without catastrophic forgetting. Extensive experiments demonstrate that our method reduces inappropriate content to 18.07\% (vs. 48.9\% for SD v1.4) and nudity detections to 15 (vs. 646 baseline) while improving compositional generation quality from 42.08\% to 47.83\% on GenEval. Remarkably, these safety gains generalize to out-of-domain unsafe prompts across seven harm categories, achieving state-of-the-art performance without supervised paired data or reward tuning. Github: https://github.com/MAXNORM8650/SafeDiffusion-R1.

📄 PDF Abstract BibTeX arXiv:2605.18719

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

A General Framework for Inference-time Scaling and Steering of Diffusion Models

2025-01-12 · Raghav Singhal, Zachary Horvitz, Ryan Teehan, Mengye Ren 외

Diffusion models produce impressive results in modalities ranging from images and video to protein design and text. However, generating samples with user-specified properties remains a challenge. Recent research proposes…

Protein Design

Explicit Critic Guidance for Aligning Diffusion Models

2026-05-26 · Zhengyang Liang, Qihang Zhang, Ceyuan Yang arxiv

Online reinforcement learning is becoming increasingly important for aligning diffusion models with non-differentiable objectives. However, existing methods still face limitations in assigning fine-grained credit along d…

Reinforcement Learning

Decoupled Guidance Diffusion for Adaptive Offline Safe Reinforcement Learning

2026-05-04 · Rufeng Chen, Zhaofan Zhang, Zhejiang Yang, Hechang Chen 외 arxiv

Offline safe reinforcement learning often requires policies to adapt at deployment time to safety budgets that vary across episodes or change within a single episode. While diffusion-based planners enable flexible trajec…

Reinforcement Learning

Pulling The REINS: Training-Free Safety Alignment of Video Diffusion Models via Representation Steering

2026-06-15 · Rohit Kundu, Arindam Dutta, Sarosij Bose, Athula Balachandran 외 arxiv

Open-weight video diffusion models can generate photorealistic unsafe content, from violence to misinformation, yet existing defenses either require expensive safety fine-tuning that degrades general capability, or apply…

Video Generation

Steering Masked Discrete Diffusion Models via Discrete Denoising Posterior Prediction

2024-10-10 · Jarrid Rector-Brooks, Mohsin Hasan, Zhangzhi Peng, Zachary Quinn 외

Generative modeling of discrete data underlies important applications spanning text-based agents like ChatGPT to the design of the very building blocks of life in protein sequences. However, application domains need to e…

Denoising