Add metadata and improve model card (#1)

- Add metadata and improve model card (2551b5f1331f281f1b96d33b553539e5d4b1cd16)


Co-authored-by: Niels Rogge <nielsr@users.noreply.huggingface.co>

README.md

@@ -1,9 +1,24 @@
 ---
-{}
+pipeline_tag: unconditional-image-generation
 ---
-## Sampling with the MDLM / Duo Checkpoints
 
-To sample from the pre-trained MDLM & Duo models, you can either play with our [Colab notebook](https://colab.research.google.com/drive/1uFSzrfG0KXhGcohRIfWIM2Y7V9Q7cQNA), or download the raw checkpoints [here](https://huggingface.co/jdeschena/duo2-cifar10/tree/main), clone our [GitHub repo](https://github.com/s-sahoo/duo), and run the following command (see the GitHub repo for more examples): 
+# Duo (Image Modeling) - CIFAR-10
+
+This repository contains pre-trained checkpoints for image modeling on CIFAR-10, as presented in the paper [The Diffusion Duality, Chapter II: $\Psi$-Samplers and Efficient Curriculum](https://huggingface.co/papers/2602.21185).
+
+- **Paper:** [The Diffusion Duality, Chapter II: $\Psi$-Samplers and Efficient Curriculum](https://huggingface.co/papers/2602.21185)
+- **Project Page:** [s-sahoo.com/duo-ch2](https://s-sahoo.com/duo-ch2)
+- **GitHub Repository:** [s-sahoo/duo](https://github.com/s-sahoo/duo)
+
+## Model Description
+
+Uniform-state discrete diffusion models excel at few-step generation and guidance due to their ability to self-correct. This checkpoint is part of the Duo series, which introduces a family of Predictor-Corrector (PC) samplers called $\Psi$-samplers. Unlike conventional samplers, these methods continue to improve quality as the number of sampling steps increases.
+
+The CIFAR-10 models are trained for 1.5M steps and have approximately 35M parameters. The architecture is the same as in [D3PM](https://arxiv.org/abs/2107.03006).
+
+## Sampling with the Duo Checkpoints
+
+To sample from the pre-trained MDLM & Duo models, you can either play with our [Colab notebook](https://colab.research.google.com/drive/1uFSzrfG0KXhGcohRIfWIM2Y7V9Q7cQNA), or download the raw checkpoints from this repository, clone our [GitHub repo](https://github.com/s-sahoo/duo), and run the following command: 
 
 ```bash
 TORCH_FORCE_NO_WEIGHTS_ONLY_LOAD=1  # Depending on your PyTorch version, this might be needed to load the checkpoint
@@ -21,17 +36,13 @@ python -u -m main \
     eval.checkpoint_path=<PATH-TO-THE-DUO-CHECKPOINT>
 ```
 
-The models are trained for 1.5M steps on CIFAR-10, and have approximately 35M parameters. The architecture is the same as in [D3PM](https://arxiv.org/abs/2107.03006).
-
-Find the text checkpoints in this [here](https://huggingface.co/s-sahoo/duo).
+Find the text checkpoints [here](https://huggingface.co/s-sahoo/duo).
 
 ### Citation
 
-Please cite our work using the bibtex below:
-
-**BibTeX**: 
+If you use this work, please cite the following:
 
-```
+```bibtex
 @inproceedings{
     deschenaux2026the,
     title={The Diffusion Duality, Chapter {II}: \${\textbackslash}Psi\$-Samplers and Efficient Curriculum},
@@ -39,5 +50,14 @@ Please cite our work using the bibtex below:
     booktitle={The Fourteenth International Conference on Learning Representations},
     year={2026},
     url={https://openreview.net/forum?id=RSIoYWIzaP}
-  }
+}
+
+@inproceedings{
+    sahoo2025the,
+    title={The Diffusion Duality},
+    author={Subham Sekhar Sahoo and Justin Deschenaux and Aaron Gokaslan and Guanghan Wang and Justin T Chiu and Volodymyr Kuleshov},
+    booktitle={Forty-second International Conference on Machine Learning},
+    year={2025},
+    url={https://openreview.net/forum?id=9P9Y8FOSOk}
+}
 ```