UniCP: A Unified Caching and Pruning Framework for Efficient Video Generation

2025-02-06 · Wenzhang Sun, Qirui Hou, Donglin Di, Jiahui Yang, Yongjia Ma, Jianxun Cui

Diffusion Transformers (DiT) excel in video generation but encounter significant computational challenges due to the quadratic complexity of attention. Notably, attention differences between adjacent diffusion steps follow a U-shaped pattern. Current methods leverage this property by caching attention blocks, however, they still struggle with sudden error spikes and large discrepancies. To address these issues, we propose UniCP a unified caching and pruning framework for efficient video generation. UniCP optimizes both temporal and spatial dimensions through. Error Aware Dynamic Cache Window (EDCW): Dynamically adjusts cache window sizes for different blocks at various timesteps, adapting to abrupt error changes. PCA based Slicing (PCAS) and Dynamic Weight Shift (DWS): PCAS prunes redundant attention components, and DWS integrates caching and pruning by enabling dynamic switching between pruned and cached outputs. By adjusting cache windows and pruning redundant components, UniCP enhances computational efficiency and maintains video detail fidelity. Experimental results show that UniCP outperforms existing methods in both performance and efficiency.

📄 PDF Abstract BibTeX arXiv:2502.04393

Code (0)

등록된 구현이 없습니다.

Tasks

Computational Efficiency Video Generation

Methods 이 논문이 사용한 방법론

Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…

Attention 설명 없음

PCA Principle Components Analysis (PCA) is an unsupervised method primary used for dimensionality reduction within machine learning. PCA is calculated via a singular value…

AWARE We propose to theoretically and empirically examine the effect of incorporating weighting schemes into walk-aggregating GNNs. To this end, we propose a simple, interpretable, and…

Diffusion Diffusion models generate samples by gradually removing noise from a signal, and their training objective can be expressed as a reweighted variational lower-bound…

Pruning 설명 없음

SODA: Sensitivity-Oriented Dynamic Acceleration for Diffusion Transformer

2026-03-07 · Tong Shao, Yusen Fu, Guoying Sun, Jingde Kong 외 arxiv

Diffusion Transformers have become a dominant paradigm in visual generation, yet their low inference efficiency remains a key bottleneck hindering further advancement. Among common training-free techniques, caching offer…

CachedSearch: Training-Free Cached Exploration for Test-Time Search in Video Diffusion

2026-07-25 · Shreshth Saini, Neil Birkbeck, Yilin Wang, Balu Adsumilli 외 arxiv

Test-time search lets small video diffusion models rival larger ones, but costs 2-10x more. All candidates are fully denoised, although most are discarded. Training-free caching makes each rollout 2-3x faster at near-los…

QuantCache: Adaptive Importance-Guided Quantization with Hierarchical Latent and Layer Caching for Video Generation

2025-03-09 · Junyi Wu, Zhiteng Li, Zheng Hui, Yulun Zhang 외

Recently, Diffusion Transformers (DiTs) have emerged as a dominant architecture in video generation, surpassing U-Net-based models in terms of performance. However, the enhanced capabilities of DiTs come with significant…

Quantization Video Generation

Frequency-Aware Error-Bounded Caching for Accelerating Diffusion Transformers

2026-03-05 · Guandong Li arxiv

Diffusion Transformers (DiTs) have emerged as the dominant architecture for high-quality image and video generation, yet their iterative denoising process incurs substantial computational cost during inference. Existing …

Video Generation

SenCache: Accelerating Diffusion Model Inference via Sensitivity-Aware Caching

2026-02-27 · Yasaman Haghighi, Alexandre Alahi arxiv

Diffusion models achieve state-of-the-art video generation quality, but their inference remains expensive due to the large number of sequential denoising steps. This has motivated a growing line of research on accelerati…