Progressive knowledge distillation of stable diffusion xl using layer level loss

D Zhang, S Li, C Chen, Q Xie, H Lu - arXiv preprint arXiv:2404.11098, 2024 - arxiv.org

In the era of AIGC, the demand for low-budget or even on-device applications of diffusion
models emerged. In terms of compressing the Stable Diffusion models (SDMs), several …

被引用次数：6 相关文章所有 2 个版本

[HTML] ennetix.cloud

[HTML][HTML] 'Misc: Ultra-low bitrate image semantic compression driven by large multimodal model

C Li, G Lu, D Feng, H Wu, Z Zhang, X Liu… - arXiv preprint arXiv …, 2024 - ennetix.cloud

With the evolution of storage and communication protocols, ultra-low bitrate image
compression has become a highly demanding topic. However, all existing compression …

被引用次数：5 相关文章所有 3 个版本

[PDF] arxiv.org

EdgeFusion: On-Device Text-to-Image Generation

T Castells, HK Song, T Piao, S Choi, BK Kim… - arXiv preprint arXiv …, 2024 - arxiv.org

The intensive computational burden of Stable Diffusion (SD) for text-to-image generation
poses a significant hurdle for its practical application. To tackle this challenge, recent …

被引用次数：3 相关文章所有 2 个版本

[PDF] arxiv.org

SDXS: Real-Time One-Step Latent Diffusion Models with Image Conditions

Y Song, Z Sun, X Yin - arXiv preprint arXiv:2403.16627, 2024 - arxiv.org

Recent advancements in diffusion models have positioned them at the forefront of image
generation. Despite their superior performance, diffusion models are not without drawbacks; …

被引用次数：5 相关文章所有 2 个版本

[PDF] arxiv.org

ComfyGen: Prompt-Adaptive Workflows for Text-to-Image Generation

R Gal, A Haviv, Y Alaluf, AH Bermano… - arXiv preprint arXiv …, 2024 - arxiv.org

The practical use of text-to-image generation has evolved from simple, monolithic models to
complex workflows that combine multiple specialized components. While workflow-based …

DDIL: Improved Diffusion Distillation With Imitation Learning

R Garrepalli, S Mahajan, M Hayat, F Porikli - arXiv preprint arXiv …, 2024 - arxiv.org

Diffusion models excel at generative modeling (eg, text-to-image) but sampling requires
multiple denoising network passes, limiting practicality. Efforts such as progressive …