CoReDiT: Spatial Coherence-Guided Token Pruning and Reconstruction for Efficient Diffusion Transformers

2026-05-13 · Zhuojin Li, Hsin-Pai Cheng, Hong Cai, Shizhong Han, Fatih Porikli arxiv

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.

📄 PDF Abstract BibTeX arXiv:2605.14191

Code (0)

등록된 구현이 없습니다.

Tasks

Video Generation

Geometry-Guided 3D Visual Token Pruning for Video-Language Models

2026-04-20 · Han Li, Zehao Huang, Jiahui Fu, Naiyan Wang 외 arxiv

Multimodal large language models have demonstrated remarkable capabilities in 2D vision, motivating their extension to 3D scene understanding. Recent studies represent 3D scenes as 3D spatial videos composed of image seq…

Scene Understanding

Streaming-dLLM: Accelerating Diffusion LLMs via Suffix Pruning and Dynamic Decoding

2026-01-25 · Zhongyu Xiao, Zhiwei Hao, Jianyuan Guo, Yong Luo 외 arxiv

Diffusion Large Language Models (dLLMs) offer a compelling paradigm for natural language generation, leveraging parallel decoding and bidirectional attention to achieve superior global coherence compared to autoregressiv…

EventPrune: Cascaded Event-Assisted Token Pruning for Efficient First-Person Dynamic Spatial Reasoning

2026-05-19 · Pengtao Ma, Ziliang Zhou, Ciyu Ruan, Haoyang Wang 외 arxiv

First-person dynamic spatial reasoning requires models to track continuous motion and precise geometric structure, but the quadratic attention cost of Transformer-based Video-LLMs makes dense visual tokens computationall…

Spatial Reasoning

Spectral Heat Flow for Conservative Token Condensation in Vision-Language Models

2026-07-12 · Zhaoyang Li, Yanjun Li, Wangkai Li, Yujia Chen 외 arxiv

Vision-Language Models (VLMs) are costly at inference time because they must process long sequences of visual tokens. Existing token pruning methods often degrade under high compression by blindly discarding information,…

CoverPrune: Coverage-Driven Token Pruning for 3D VLMs via Optimal Transport

2026-08-13 · Peng Ling, Yingda Yin, Lingting Zhu, Weikai Chen 외 arxiv

While 3D Vision-Language Models (3D VLMs) have demonstrated remarkable spatial reasoning capabilities, they suffer from massive visual token counts that create severe computational bottlenecks during inference. Existing …