arXiv AI

Steering Away from Memorization: Reachability-Constrained Reinforcement Learning for Text-to-Image Diffusion

The paper introduces Reachability-Aware Diffusion Steering (RADS), an offline-trained, sampling-time control framework designed to reduce memorization in text-to-image diffusion models while preserving image quality and prompt alignment. RADS treats the diffusion denoising process as a dynamical system and learns a policy-conditioned reachability value function that predicts future memorization risk along the denoising trajectory. By formulating a constrained reinforcement learning problem, the policy steers the trajectory away from memorization through perturbations in the caption embedding space, achieving a better trade‑off between generation diversity, quality, and alignment than existing inference‑time baselines.

arXiv Computer Vision
Sep 7

Step Back to Move Forward: Reflection-Aware Preference Optimization for Visual Generation

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 AI
Jul 1

Finite Difference Flow Optimization for RL Post-Training of Text-to-Image Models

arXiv:2603. 12893v2 Announce Type: replace-cross Abstract: Reinforcement learning (RL) has become a standard technique for post-training diffusion-based image synthesis models, as it enables learning from reward signals to explicitly improve desirable aspects such as image quality and prompt alignment.

By David McAllister, Miika Aittala, Tero Karras, Janne Hellsten, Angjoo Kanazawa, Timo Aila, Samuli Laine
arXiv Machine Learning
Oct 23

Efficient Adjoint Matching for Fine-tuning Diffusion Models

Efficient Adjoint Matching (EAM) is a new method for reward fine‑tuning of pretrained diffusion models that reduces computational cost by reformulating the stochastic optimal control problem with a linear base drift and a modified terminal cost. This change eliminates the need for extensive stochastic simulation of full generative trajectories and backward ODE simulation of the adjoint state, allowing training‑time sampling with a few‑step deterministic ODE solver and a closed‑form adjoint solution. On standard text‑to‑image reward fine‑tuning benchmarks, EAM converges up to four times faster than the previous Adjoint Matching approach while matching or surpassing it on metrics such as PickScore, ImageReward, HPSv2.1, CLIPScore, and Aesthetics.

By Jeongwoo Shin, Dongsoo Shin, Yuchen Zhu, Wei Guo, Yongxin Chen, Joonseok Lee, Jaewoong Choi, Jaemoo Choi
arXiv Machine Learning
Aug 31

Amortizing intractable inference in diffusion models for vision, language, and control

The paper introduces a data‑free learning objective called relative trajectory balance for training diffusion models to sample from a posterior defined by a diffusion prior and an arbitrary black‑box constraint or likelihood. It proves asymptotic correctness of this objective and demonstrates its use across vision, language, and multimodal tasks, including classifier guidance, language infilling, and text‑to‑image generation. Additionally, the method is applied to continuous control with a score‑based behavior prior, achieving state‑of‑the‑art results in offline reinforcement learning.

By Siddarth Venkatraman, Moksh Jain, Luca Scimeca, Minsu Kim, Marcin Sendera, Mohsin Hasan, Luke Rowe, Sarthak Mittal, Pablo Lemos, Emmanuel Bengio, Alexandre Adam, Jarrid Rector-Brooks, Yoshua Bengio, Glen Berseth, Esmeralda S. Whitammer
Hugging Face Trending Papers
Jul 26

Learning Sampling Parameters for Diffusion Models

Text-to-image diffusion models expose many inference-time sampling parameters, including prompts, negative prompts, classifier-free guidance scales, and noise schedules. These parameters are typically manually chosen once and then held fixed across prompts and denoising timesteps, even though different prompts and stages of generation can benefit from different parameter values.

Hugging Face Trending Papers
Oct 15

Routing Policy Optimization for Online Diffusion Reinforcement Learning

Routing Policy Optimization (RPO) introduces stochastic exploration within a diffusion visual generator by treating attention routing as a stochastic action. Instead of perturbing the entire reverse denoising trajectory, RPO samples key attention blocks at selected denoising levels, perturbs the transition velocity, and evaluates the perturbation with a score‑function objective. A two‑rollout estimator provides an advantage signal, and after training the model restores dense attention, yielding higher generation quality and nearly four times the training efficiency compared to Flow‑GRPO‑style methods.

arXiv Machine Learning
Oct 3

Learned End-to-End Guidance Schedules for Diffusion Models

The paper introduces Learned End-to-End Guidance Schedules (LEEGS) for diffusion models, which train a time‑dependent guidance schedule to balance data quality and requirement satisfaction while reducing sampling steps. LEEGS minimizes the guidance function over a small set of examples using stochastic gradient descent and employs a gradient approximation to cut training time by a factor of four. Experiments on tasks such as image inpainting, noisy image inverse problems, face‑ID‑guided generation, and PDE problems show that LEEGS outperforms baselines at the same computational budget or matches constant guidance with only 10% of the steps.

By Aneesh Barthakur, Mathias Niepert, Luiz F. O. Chamon