README.md

---
base_model:
- XCLiu/instaflow_0_9B_from_sd_1_5
---
<div align="center">
<h1>Official PyTorch models of "Self-Corrected Flow Distillation for Consistent One-Step and Few-Step Text-to-Image Generation" <a href="https://arxiv.org/abs/2412.16906">(AAAI 2025)</a></h1>
</div>

<div align="center">
  <a href="https://quandao10.github.io/" target="_blank">Quan&nbsp;Dao</a><sup>*12†</sup> &emsp; <b>&middot;</b> &emsp;
  <a href="https://hao-pt.github.io/" target="_blank">Hao&nbsp;Phung</a><sup>*13†</sup> &emsp; <b>&middot;</b> &emsp;
  <a href="https://trung-dt.com/" target="_blank">Trung&nbsp;Dao</a><sup>1</sup> &emsp; <b>&middot;</b> &emsp;
  <a href="https://people.cs.rutgers.edu/~dnm/" target="_blank">Dimitris&nbsp;N. Metaxas</a><sup>2</sup> &emsp; <b>&middot;</b> &emsp;
  <a href="https://sites.google.com/site/anhttranusc/" target="_blank">Anh&nbsp;Tran</a><sup>1</sup>
  <br> <br>
  <sup>1</sup>VinAI Research &emsp;
  <sup>2</sup>Rutgers University &emsp;
  <sup>3</sup>Cornell University
  <br> <br>
  <a href="https://arxiv.org/abs/2412.16906">[Paper]</a> &emsp;&emsp;
  <a href="https://github.com/hao-pt/SCFlow.git">[Code]</a> 
  <br> <br>
  <emp><sup>*</sup>Equal contribution</emp> &emsp;
  <emp><sup>†</sup>Work done while at VinAI Research</emp>
</div>

## Model details 
We present a distilled Text-to-Image (T2I) model that supports both few-step and single-step generation. Distilled from XCLiu/instaflow_0_9B_from_sd_1_5, our model achieves an FID of 11.91 for 1-NFE generation on the COCO2014 benchmark.


**Please CITE** our paper and give us a :star: whenever this repository is used to help produce published results or incorporated into other software.

```bibtex
@inproceedings{dao2025scflow,
  title     = {Self-Corrected Flow Distillation for Consistent One-Step and Few-Step Text-to-Image Generation},
  author    = {Quan Dao and Hao Phung and Trung Dao and Dimitris Metaxas and Anh Tran},
  booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year      = {2025}
}
```