CoReDiT: Spatial Coherence-Guided Token Pruning and Reconstruction for Efficient Diffusion Transformers
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_ | 1866918501026889728 |
|---|---|
| author | Li, Zhuojin Cheng, Hsin-Pai Cai, Hong Han, Shizhong Porikli, Fatih |
| author_facet | Li, Zhuojin Cheng, Hsin-Pai Cai, Hong Han, Shizhong Porikli, Fatih |
| contents | Diffusion Transformers (DiTs) deliver remarkable image and video generation quality but incur high computational cost, limiting scalability and on-device deployment. We introduce CoReDiT, a structured token pruning framework for DiTs across vision tasks. CoReDiT uses a linear-time spatial coherence score to estimate local redundancy in the latent token lattice and skips high coherence (redundant) tokens in self-attention. To maintain a dense representation and avoid visual discontinuities, we reconstruct skipped attention outputs via coherence-guided aggregation of spatially neighboring retained tokens. We further introduce a progressive, block-adaptive pruning schedule that increases pruning gradually and allocates larger budgets to blocks and denoising steps with higher redundancy. Across state-of-the-art diffusion backbones including PixArt-α and MagicDrive-V2, CoReDiT achieves up to 55% self-attention FLOPs reduction and inference speedups of 1.33x on cloud GPUs and 1.72x on mobile NPUs, while maintaining high visual quality. Notably, CoReDiT also increases on-device memory head-room, enabling higher-resolution generation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_14191 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | CoReDiT: Spatial Coherence-Guided Token Pruning and Reconstruction for Efficient Diffusion Transformers Li, Zhuojin Cheng, Hsin-Pai Cai, Hong Han, Shizhong Porikli, Fatih Computer Vision and Pattern Recognition Diffusion Transformers (DiTs) deliver remarkable image and video generation quality but incur high computational cost, limiting scalability and on-device deployment. We introduce CoReDiT, a structured token pruning framework for DiTs across vision tasks. CoReDiT uses a linear-time spatial coherence score to estimate local redundancy in the latent token lattice and skips high coherence (redundant) tokens in self-attention. To maintain a dense representation and avoid visual discontinuities, we reconstruct skipped attention outputs via coherence-guided aggregation of spatially neighboring retained tokens. We further introduce a progressive, block-adaptive pruning schedule that increases pruning gradually and allocates larger budgets to blocks and denoising steps with higher redundancy. Across state-of-the-art diffusion backbones including PixArt-α and MagicDrive-V2, CoReDiT achieves up to 55% self-attention FLOPs reduction and inference speedups of 1.33x on cloud GPUs and 1.72x on mobile NPUs, while maintaining high visual quality. Notably, CoReDiT also increases on-device memory head-room, enabling higher-resolution generation. |
| title | CoReDiT: Spatial Coherence-Guided Token Pruning and Reconstruction for Efficient Diffusion Transformers |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.14191 |