# DDT: Decoupled Diffusion Transformer
<div style="text-align: center;">
  <a href="https://arxiv.org/abs/2504.05741"><img src="https://img.shields.io/badge/arXiv-2504.05741-b31b1b.svg" alt="arXiv"></a>
    <a href="https://huggingface.co/spaces/MCG-NJU/DDT"><img src="https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Online_Demo-green" alt="arXiv"></a>  
</div>

[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/ddt-decoupled-diffusion-transformer-1/image-generation-on-imagenet-512x512)](https://paperswithcode.com/sota/image-generation-on-imagenet-512x512?p=ddt-decoupled-diffusion-transformer-1)

[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/ddt-decoupled-diffusion-transformer-1/image-generation-on-imagenet-256x256)](https://paperswithcode.com/sota/image-generation-on-imagenet-256x256?p=ddt-decoupled-diffusion-transformer-1)
## Introduction
We decouple diffusion transformer into encoder-decoder design, and surprisingly that a **more substantial encoder yields performance improvements as model size increases**.
![](./figs/main.png)
* We achieves **1.26 FID** on ImageNet256x256 Benchmark with DDT-XL/2(22en6de).
* We achieves **1.28 FID** on ImageNet512x512 Benchmark with DDT-XL/2(22en6de).
* As a byproduct, our DDT can reuse encoder among adjacent steps to accelerate inference.
## Update 5/6/2025
* PixelDDT-XXL/16-R1024-T2I achieves **66.7** without prompt rewriting and **71.2** with prompt rewriting on GenEval benchmark
* Pixel space Text-to-image models(PixelDDT-XXL/16) will be released soon.
## Visualizations
![](./figs/teaser.png)
## Checkpoints
We take the off-shelf [VAE](https://huggingface.co/stabilityai/sd-vae-ft-ema) to encode image into latent space, and train the decoder with DDT.

| Dataset     | Model             | Params    | FID  | HuggingFace                                              |
|-------------|-------------------|-----------|------|----------------------------------------------------------|
| ImageNet256 | DDT-XL/2(22en6de) | 675M | 1.26 | [🤗](https://huggingface.co/MCG-NJU/DDT-XL-22en6de-R256) |
| ImageNet512 | DDT-XL/2(22en6de) | 675M | 1.28 | [🤗](https://huggingface.co/MCG-NJU/DDT-XL-22en6de-R512) |
## Online Demos
We provide online demos for DDT-XL/2(22en6de) on HuggingFace Spaces.

HF spaces: [https://huggingface.co/spaces/MCG-NJU/DDT](https://huggingface.co/spaces/MCG-NJU/DDT)

To host the local gradio demo(default 512 resolution), run the following command:
```bash
# default 512 resolution
python app.py --config configs/repa_improved_ddt_xlen22de6_512.yaml --resolution 512 --ckpt_path=XXX512.ckpt
# for 256 resolution 
python app.py --config configs/repa_improved_ddt_xlen22de6_256.yaml --resolution 256 --ckpt_path=XXX256.ckpt
```

## Usages
We use ADM evaluation suite to report FID.
```bash
# for installation
pip install -r requirements.txt
```
By default, the `main.py` will use all available GPUs. You can specify the GPU(s) to use with `CUDA_VISIBLE_DEVICES`.
or specify the number of GPUs to use with as :
```yaml
# in configs/repa_improved_ddt_xlen22de6_256.yaml
trainer:
  default_root_dir: universal_flow_workdirs
  accelerator: auto
  strategy: auto
  devices: auto
  # devices: 0,
  # devices: 0,1
  num_nodes: 1
```
By default, the `save_image_callbacks` will only save the first 100 images and npz file(to calculate FID with ADM suite). You can change the number of images to save with as :  
```yaml
# in configs/repa_improved_ddt_xlen22de6_256.yaml
callbacks:
- class_path: src.callbacks.model_checkpoint.CheckpointHook
  init_args:
    every_n_train_steps: 10000
    save_top_k: -1
    save_last: true
- class_path: src.callbacks.save_images.SaveImagesHook
  init_args:
     save_dir: val
     max_save_num: 0
     # max_save_num: 100
```
By default, we infer 50K images with batch size 64. You can change the number of images and classes to infer with as :
```yaml
# in configs/repa_improved_ddt_xlen22de6_256.yaml
data:
  train_dataset: imagenet256
  train_root: /mnt/bn/wangshuai6/data/ImageNet/train
  train_image_size: 256
  train_batch_size: 16
  eval_max_num_instances: 50000
  pred_batch_size: 64
  # pred_batch_size: 16
  pred_num_workers: 4
  pred_seeds: null
  # pred_seeds: 1,2,3,4
  pred_selected_classes: null
  # pred_selected_classes:
  # - 0
  # - 1
  num_classes: 1000
  latent_shape:
  - 4
  - 32
  - 32
```
`pred_selected_classes` is a list of class indices to infer, if `pred_selected_classes` is null, all classes will be inferred.
`pred_seeds` is string of seeds for every class seprated with `,` to infer
```bash
# for inference
python main.py predict -c configs/repa_improved_ddt_xlen22de6_256.yaml --ckpt_path=XXX.ckpt
# # or specify the GPU(s) to use with as :
CUDA_VISIBLE_DEVICES=0,1, python main.py predict -c configs/repa_improved_ddt_xlen22de6_256.yaml --ckpt_path=XXX.ckpt
```

```bash
# for training
# extract image latent (optional)
python3 tools/cache_imlatent4.py
# train
python main.py fit -c configs/repa_improved_ddt_xlen22de6_256.yaml
```


## Reference
```bibtex
@article{wang2025ddt,
  title={DDT: Decoupled Diffusion Transformer},
  author={Wang, Shuai and Tian, Zhi and Huang, Weilin and Wang, Limin},
  journal={arXiv preprint arXiv:2504.05741},
  year={2025}
}
```

## Acknowledgement
The code is mainly built upon [FlowDCN](https://github.com/MCG-NJU/FlowDCN), we also borrow ideas from the [REPA](https://github.com/sihyun-yu/REPA), [MAR](https://github.com/LTH14/mar) and [SiT](https://github.com/willisma/SiT).