|
--- |
|
{} |
|
--- |
|
# VAE Features of ImageNet Dataset |
|
|
|
<p align="left"> |
|
<a href="https://arxiv.org/abs/2405.02730" alt="arXiv"> |
|
<img src="https://img.shields.io/badge/arXiv-2405.02730-b31b1b.svg?style=flat" /></a> |
|
<a href="https://github.com/YuchuanTian/U-DiT" alt="arXiv"> |
|
<img src="https://img.shields.io/badge/GitHub-100000?style=for-the-badge&logo=github&logoColor=white" /></a> |
|
</p> |
|
This is the ImageNet dataset features processed by sd-vae-ft-ema encoder. The dataset is tailored for DiT training purposes. |
|
|
|
File structure: |
|
|
|
``` |
|
imagenet_feature/ |
|
├── imagenet256_features/ # VAE features |
|
└── imagenet256_labels/ # labels |
|
``` |
