Downloads · 30 days
5
9% of all-time downloads
dimsum-L2-imagenet256
haopt/dimsum-L2-imagenet256
dimsum-L2-imagenet256 is a machine learning model from haopt. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as bsd-3-clause.
<div align="center" <h1Official PyTorch models of "DiMSUM: Diffusion Mamba - A Scalable and Unified Spatial-Frequency Method for Image Generation" <a href="https://arxiv.org/abs/2411.04168" (NeurIPS'24)</a</h1 </div
All-time downloads
53
Public
Repo size
1.8 GB
Likes
2
Public
Repo makeup
Click a slice to open those files.
.bin1.8 GB · 100%
At a glance
- License
- bsd-3-clause
- Model type
- dimsum
- Access
- Public
- Created
- Feb 18, 2025
- Updated
- Jun 1, 2026
- SHA
- 53d76c20
Datasets
Model card
From the Hugging Face model README
<div align="center">
<h1>Official PyTorch models of "DiMSUM: Diffusion Mamba - A Scalable and Unified
Spatial-Frequency Method for Image Generation" <a href="https://arxiv.org/abs/2411.04168"> (NeurIPS'24)</a></h1>
</div>
<div align="center">
<a href="https://hao-pt.github.io/" target="_blank">Hao Phung</a><sup>*13†</sup>   <b>·</b>  
<a href="https://quandao10.github.io/" target="_blank">Quan Dao</a><sup>*12†</sup>   <b>·</b>  
<a href="https://termanteus.com/" target="_blank">Trung Dao</a><sup>1</sup>
<br> <br>
<a href="https://viethoang1512.github.io/" target="_blank">Hoang Phan</a><sup>4</sup>   <b>·</b>  
<a href="https://people.cs.rutgers.edu/~dnm/" target="_blank"> Dimitris N. Metaxas</a><sup>2</sup>   <b>·</b>  
<a href="https://sites.google.com/site/anhttranusc/" target="_blank">Anh Tran</a><sup>1</sup>
<br> <br>
<sup>1</sup>VinAI Research  
<sup>2</sup>Rutgers University  
<sup>3</sup>Cornell University  
<sup>4</sup>New York University
<br> <br>
<a href="https://vinairesearch.github.io/DiMSUM/">[Page]</a>   
<a href="https://arxiv.org/abs/2411.04168">[Paper]</a>   
<br> <br>
<emp><sup>*</sup>Equal contribution</emp>  
<emp><sup>†</sup>Work done while at VinAI Research</emp>
</div>
Model details
Our model is a hydrid Mamba-Transformer architecture for class-to-image generation. This method is trained with flow matching objective. The model has 460M parameters and achieves an FID score of 2.11 on ImageNet-1K 256 dataset. Our codebase is hosted at https://github.com/VinAIResearch/DiMSUM.git.
To use DiMSUM pre trained model:
from huggingface_hub import hf_hub_download
# Assume model is already initiated
ckpt_path = hf_hub_download("haopt/dimsum-L2-imagenet256")
state_dict = torch.load(ckpt_path)
model.load_state_dict(state_dict)
model.eval()
Please CITE our paper and give us a :star: whenever this repository is used to help produce published results or incorporated into other software.
@inproceedings{phung2024dimsum,
title={DiMSUM: Diffusion Mamba - A Scalable and Unified Spatial-Frequency Method for Image Generation},
author={Phung, Hao and Dao, Quan and Dao, Trung and Phan, Hoang and Metaxas, Dimitris and Tran, Anh},
booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems},
year= {2024},
}
- Type
- dimsum
- License
- bsd-3-clause
- Created
- Feb 18, 2025
- Updated
- Jun 1, 2026