Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Adnan, Muhammad, Kurella, Nithesh, Arunkumar, Akhil, Nair, Prashant J.
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2506.00329
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916961239171072
author Adnan, Muhammad
Kurella, Nithesh
Arunkumar, Akhil
Nair, Prashant J.
author_facet Adnan, Muhammad
Kurella, Nithesh
Arunkumar, Akhil
Nair, Prashant J.
contents Diffusion Transformers (DiTs) achieve state-of-the-art results in text-to-image, text-to-video generation, and editing. However, their large model size and the quadratic cost of spatial-temporal attention over multiple denoising steps make video generation computationally expensive. Static caching mitigates this by reusing features across fixed steps but fails to adapt to generation dynamics, leading to suboptimal trade-offs between speed and quality. We propose Foresight, an adaptive layer-reuse technique that reduces computational redundancy across denoising steps while preserving baseline performance. Foresight dynamically identifies and reuses DiT block outputs for all layers across steps, adapting to generation parameters such as resolution and denoising schedules to optimize efficiency. Applied to OpenSora, Latte, and CogVideoX, Foresight achieves up to \latencyimprv end-to-end speedup, while maintaining video quality. The source code of Foresight is available at \href{https://github.com/STAR-Laboratory/foresight}{https://github.com/STAR-Laboratory/foresight}.
format Preprint
id arxiv_https___arxiv_org_abs_2506_00329
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Foresight: Adaptive Layer Reuse for Accelerated and High-Quality Text-to-Video Generation
Adnan, Muhammad
Kurella, Nithesh
Arunkumar, Akhil
Nair, Prashant J.
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
Diffusion Transformers (DiTs) achieve state-of-the-art results in text-to-image, text-to-video generation, and editing. However, their large model size and the quadratic cost of spatial-temporal attention over multiple denoising steps make video generation computationally expensive. Static caching mitigates this by reusing features across fixed steps but fails to adapt to generation dynamics, leading to suboptimal trade-offs between speed and quality. We propose Foresight, an adaptive layer-reuse technique that reduces computational redundancy across denoising steps while preserving baseline performance. Foresight dynamically identifies and reuses DiT block outputs for all layers across steps, adapting to generation parameters such as resolution and denoising schedules to optimize efficiency. Applied to OpenSora, Latte, and CogVideoX, Foresight achieves up to \latencyimprv end-to-end speedup, while maintaining video quality. The source code of Foresight is available at \href{https://github.com/STAR-Laboratory/foresight}{https://github.com/STAR-Laboratory/foresight}.
title Foresight: Adaptive Layer Reuse for Accelerated and High-Quality Text-to-Video Generation
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2506.00329