DiCo: Revitalizing ConvNets for Scalable and Efficient Diffusion Modeling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ai, Yuang, Fan, Qihang, Hu, Xuefeng, Yang, Zhenheng, He, Ran, Huang, Huaibo
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918145286995968
author Ai, Yuang
Fan, Qihang
Hu, Xuefeng
Yang, Zhenheng
He, Ran
Huang, Huaibo
author_facet Ai, Yuang
Fan, Qihang
Hu, Xuefeng
Yang, Zhenheng
He, Ran
Huang, Huaibo
contents Diffusion Transformer (DiT), a promising diffusion model for visual generation, demonstrates impressive performance but incurs significant computational overhead. Intriguingly, analysis of pre-trained DiT models reveals that global self-attention is often redundant, predominantly capturing local patterns-highlighting the potential for more efficient alternatives. In this paper, we revisit convolution as an alternative building block for constructing efficient and expressive diffusion models. However, naively replacing self-attention with convolution typically results in degraded performance. Our investigations attribute this performance gap to the higher channel redundancy in ConvNets compared to Transformers. To resolve this, we introduce a compact channel attention mechanism that promotes the activation of more diverse channels, thereby enhancing feature diversity. This leads to Diffusion ConvNet (DiCo), a family of diffusion models built entirely from standard ConvNet modules, offering strong generative performance with significant efficiency gains. On class-conditional ImageNet generation benchmarks, DiCo-XL achieves an FID of 2.05 at 256x256 resolution and 2.53 at 512x512, with a 2.7x and 3.1x speedup over DiT-XL/2, respectively. Furthermore, experimental results on MS-COCO demonstrate that the purely convolutional DiCo exhibits strong potential for text-to-image generation. Code: https://github.com/shallowdream204/DiCo.
format Preprint
id arxiv_https___arxiv_org_abs_2505_11196
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DiCo: Revitalizing ConvNets for Scalable and Efficient Diffusion Modeling
Ai, Yuang
Fan, Qihang
Hu, Xuefeng
Yang, Zhenheng
He, Ran
Huang, Huaibo
Computer Vision and Pattern Recognition
Diffusion Transformer (DiT), a promising diffusion model for visual generation, demonstrates impressive performance but incurs significant computational overhead. Intriguingly, analysis of pre-trained DiT models reveals that global self-attention is often redundant, predominantly capturing local patterns-highlighting the potential for more efficient alternatives. In this paper, we revisit convolution as an alternative building block for constructing efficient and expressive diffusion models. However, naively replacing self-attention with convolution typically results in degraded performance. Our investigations attribute this performance gap to the higher channel redundancy in ConvNets compared to Transformers. To resolve this, we introduce a compact channel attention mechanism that promotes the activation of more diverse channels, thereby enhancing feature diversity. This leads to Diffusion ConvNet (DiCo), a family of diffusion models built entirely from standard ConvNet modules, offering strong generative performance with significant efficiency gains. On class-conditional ImageNet generation benchmarks, DiCo-XL achieves an FID of 2.05 at 256x256 resolution and 2.53 at 512x512, with a 2.7x and 3.1x speedup over DiT-XL/2, respectively. Furthermore, experimental results on MS-COCO demonstrate that the purely convolutional DiCo exhibits strong potential for text-to-image generation. Code: https://github.com/shallowdream204/DiCo.
title DiCo: Revitalizing ConvNets for Scalable and Efficient Diffusion Modeling
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2505.11196