paper-with-me

Papers

Test-time Alignment of Diffusion Models without Reward Over-optimization

2025-01-10 · Sunwoo Kim, Minkyu Kim, Dongmin Park

Diffusion models excel in generative tasks, but aligning them with specific objectives while maintaining their versatility remains challenging. Existing fine-tuning methods often suffer from reward over-optimization, while approximate guidance approaches fail to optimize target rewards effectively. Addressing these limitations, we propose a training-free, test-time method based on Sequential Monte Carlo (SMC) to sample from the reward-aligned target distribution. Our approach, tailored for diffusion sampling and incorporating tempering techniques, achieves comparable or superior target rewards to fine-tuning methods while preserving diversity and cross-reward generalization. We demonstrate its effectiveness in single-reward optimization, multi-objective scenarios, and online black-box optimization. This work offers a robust solution for aligning diffusion models with diverse downstream objectives without compromising their general capabilities. Code is available at https://github.com/krafton-ai/DAS.

📄 PDF Abstract BibTeX arXiv:2501.05803

Code (1)

krafton-ai/das 공식 구현 pytorch

Tasks

Diversity

Methods 이 논문이 사용한 방법론

Diffusion Diffusion models generate samples by gradually removing noise from a signal, and their training objective can be expressed as a reweighted variational lower-bound…

Similar Papers 제목 키워드 기반

Test-Time Alignment of Text-to-Image Diffusion Models via Null-Text Embedding Optimisation

2025-11-25 · Taehoon Kim, Henry Gouk, Timothy Hospedales arxiv

Test-time alignment (TTA) aims to adapt models to specific rewards during inference. However, existing methods tend to either under-optimise or over-optimise (reward hack) the target reward function. We propose Null-Text…

Beyond VLM-Based Rewards: Diffusion-Native Latent Reward Modeling

2026-02-11 · Gongye Liu, Bo Yang, Yida Zhi, Zhizhou Zhong 외 arxiv

Preference optimization for diffusion and flow-matching models relies on reward functions that are both discriminatively robust and computationally efficient. Vision-Language Models (VLMs) have emerged as the primary rew…

HyperAlign: Hypernetwork for Efficient Test-Time Alignment of Diffusion Models

2026-01-22 · Xin Xie, Jiaxian Guo, Dong Gong arxiv

Diffusion model alignment aims to bridge the gap between generated outputs and human preferences by enhancing both semantic consistency with textual prompts and overall visual quality. Existing alignment methods face a c…

Computational Efficiency

Diffusion Alignment as Variational Expectation-Maximization

2025-10-01 · Jaewoo Lee, Minsu Kim, Sanghyeok Choi, Inhyuck Song 외 arxiv

Diffusion alignment aims to optimize diffusion models for the downstream objective. While existing methods based on reinforcement learning or direct backpropagation achieve considerable success in maximizing rewards, the…

Reinforcement Learning

MIRA: Towards Mitigating Reward Hacking in Inference-Time Alignment of T2I Diffusion Models

2025-10-02 · Kevin Zhai, Utsav Singh, Anirudh Thatipelli, Souradip Chakraborty 외 arxiv

Diffusion models excel at generating images conditioned on text prompts, but the resulting images often do not satisfy user-specific criteria measured by scalar rewards such as Aesthetic Scores. This alignment typically …