AdaTSQ: Pushing the Pareto Frontier of Diffusion Transformers via Temporal-Sensitivity Quantization
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914319431630848 |
|---|---|
| author | Zhang, Shaoqiu Ding, Zizhong Yang, Kaicheng Wu, Junyi Yan, Xianglong Li, Xi Duan, Bingnan Fang, Jianping Zhang, Yulun |
| author_facet | Zhang, Shaoqiu Ding, Zizhong Yang, Kaicheng Wu, Junyi Yan, Xianglong Li, Xi Duan, Bingnan Fang, Jianping Zhang, Yulun |
| contents | Diffusion Transformers (DiTs) have emerged as the state-of-the-art backbone for high-fidelity image and video generation. However, their massive computational cost and memory footprint hinder deployment on edge devices. While post-training quantization (PTQ) has proven effective for large language models (LLMs), directly applying existing methods to DiTs yields suboptimal results due to the neglect of the unique temporal dynamics inherent in diffusion processes. In this paper, we propose AdaTSQ, a novel PTQ framework that pushes the Pareto frontier of efficiency and quality by exploiting the temporal sensitivity of DiTs. First, we propose a Pareto-aware timestep-dynamic bit-width allocation strategy. We model the quantization policy search as a constrained pathfinding problem. We utilize a beam search algorithm guided by end-to-end reconstruction error to dynamically assign layer-wise bit-widths across different timesteps. Second, we propose a Fisher-guided temporal calibration mechanism. It leverages temporal Fisher information to prioritize calibration data from highly sensitive timesteps, seamlessly integrating with Hessian-based weight optimization. Extensive experiments on four advanced DiTs (e.g., Flux-Dev, Flux-Schnell, Z-Image, and Wan2.1) demonstrate that AdaTSQ significantly outperforms state-of-the-art methods like SVDQuant and ViDiT-Q. Our code will be released at https://github.com/Qiushao-E/AdaTSQ. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_09883 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | AdaTSQ: Pushing the Pareto Frontier of Diffusion Transformers via Temporal-Sensitivity Quantization Zhang, Shaoqiu Ding, Zizhong Yang, Kaicheng Wu, Junyi Yan, Xianglong Li, Xi Duan, Bingnan Fang, Jianping Zhang, Yulun Computer Vision and Pattern Recognition Diffusion Transformers (DiTs) have emerged as the state-of-the-art backbone for high-fidelity image and video generation. However, their massive computational cost and memory footprint hinder deployment on edge devices. While post-training quantization (PTQ) has proven effective for large language models (LLMs), directly applying existing methods to DiTs yields suboptimal results due to the neglect of the unique temporal dynamics inherent in diffusion processes. In this paper, we propose AdaTSQ, a novel PTQ framework that pushes the Pareto frontier of efficiency and quality by exploiting the temporal sensitivity of DiTs. First, we propose a Pareto-aware timestep-dynamic bit-width allocation strategy. We model the quantization policy search as a constrained pathfinding problem. We utilize a beam search algorithm guided by end-to-end reconstruction error to dynamically assign layer-wise bit-widths across different timesteps. Second, we propose a Fisher-guided temporal calibration mechanism. It leverages temporal Fisher information to prioritize calibration data from highly sensitive timesteps, seamlessly integrating with Hessian-based weight optimization. Extensive experiments on four advanced DiTs (e.g., Flux-Dev, Flux-Schnell, Z-Image, and Wan2.1) demonstrate that AdaTSQ significantly outperforms state-of-the-art methods like SVDQuant and ViDiT-Q. Our code will be released at https://github.com/Qiushao-E/AdaTSQ. |
| title | AdaTSQ: Pushing the Pareto Frontier of Diffusion Transformers via Temporal-Sensitivity Quantization |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2602.09883 |