analiza aiqbench
This research introduces a temporal difference objective that enforces cross-time consistency in diffusion models by framing denoising as a reinforcement learning policy evaluation problem. The approach significantly improves sample quality for few-step samplers and offers practical benefits under low-computation budgets without requiring architectural changes to existing generative models. By unifying discrete- and continuous-time formulations, this method provides a robust training strategy that addresses inherent inconsistencies in standard local denoising targets.
Temporal Difference Learning for Diffusion Models
arXiv:2606.15048v1 Announce Type: new Abstract: Diffusion models are typically trained with objectives that focus on local denoising targets at individual time steps (or adjacent pairs), which do not enforce consistency between predictions along the denoising trajectory. This lack of cross-time consistency can degrade performance, especially for few-step samplers. We introduce a temporal difference (TD) objective that penalizes inconsistency of the model's multi-step progress along the denoising path. By reformulating the diffusion process as a Markov reward process and casting denoising as a policy evaluation problem in reinforcement learning, we derive a unified TD approach that applies to both discrete- and continuous-time diffusion formulations. We further propose a principled sample-based reweighting method that stabilizes training. Empirically, we show that using our TD training can significantly improve sample quality measured by FID, with stronger advantages when the number of sampling steps is small, highlighting its practical utility under low-computation-budget scenarios. We provide ablation studies to justify our design choices, including pairwise loss reweighting, regularization weight, and one-step stride. Overall, our TD approach can be a general drop-in that enforces cross-time consistency and improves generation quality across different diffusion generative models.
Abstract:Diffusion models are typically trained with objectives that focus on local denoising targets at individual time steps (or adjacent pairs), which do not enforce consistency between predictions along the denoising trajectory. This lack of cross-time consistency can degrade performance, especially for few-step samplers. We introduce a temporal difference (TD) objective that penalizes inconsistency of the model's multi-step progress along the denoising path. By reformulating the diffusion process as a Markov reward process and casting denoising as a policy evaluation problem in reinforcement learning, we derive a unified TD approach that applies to both discrete- and continuous-time diffusion formulations. We further propose a principled sample-based reweighting method that stabilizes training. Empirically, we show that using our TD training can significantly improve sample quality measured by FID, with stronger advantages when the number of sampling steps is small, highlighting its practical utility under low-computation-budget scenarios. We provide ablation studies to justify our design choices, including pairwise loss reweighting, regularization weight, and one-step stride. Overall, our TD approach can be a general drop-in that enforces cross-time consistency and improves generation quality across different diffusion generative models.
15 pages, 4 figures. Accepted at ICML 2026
Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2606.15048 [cs.LG]
(or arXiv:2606.15048v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2606.15048 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 291 cuvinte
Subiecte conexe
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · 25 sept. 2026
Generalizable Multi-Agent Planning from Signal Temporal Logic Specifications via Diffusion
arXiv · 2 sept. 2026
Visual Representation Matters: Exploiting Temporal Differences in Video-to-Audio Generation
arXiv · 6 aug. 2026
STaR-Quant: State-Time Consistent Post-Training Quantization for Diffusion Large Language Models
arXiv · 9 iun. 2026