Correct the model card: dataset, conditioning, FID, usage and licence

README.md

@@ -1,59 +1,65 @@
 ---
+license: cc-by-nc-4.0
+tags:
+- gan
+- image-generation
+- pytorch
 datasets:
 - ILSVRC/imagenet-1k
-pipeline_tag: unconditional-image-generation
 ---
 
-# Model Card for ImageNet 32x32 R3GAN Model
+# R3GAN: ImageNet 32×32
 
-This model card provides details about the R3GAN model trained on the ImageNet dataset found in the NeurIPS 2024 paper: https://arxiv.org/abs/2501.05441
+The class-conditional (1,000 classes) R3GAN generator for ImageNet at 32×32, from the NeurIPS 2024 paper
+[The GAN is dead; long live the GAN! A Modern GAN Baseline](https://arxiv.org/abs/2501.05441)
+(the proposed configuration, config E).
 
-## Model Details
+| | |
+|---|---|
+| Dataset | ImageNet-1k (ILSVRC 2012) |
+| Resolution | 32×32 |
+| Conditioning | class-conditional (1,000 classes) |
+| FID (50k samples) | 1.27, as reported in the paper |
+| File | `network-snapshot-final.pkl`: the final training snapshot; its `G_ema` network is the generator |
+| Code | https://github.com/brownvc/R3GAN |
+| Licence | [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/) |
 
-The model achieves 1.27 Frechet Inception Distance-50k on ImageNet64x64 class conditional ImgNet generation.
+## Generating images
 
-### Model Description
+```
+git clone https://github.com/brownvc/R3GAN && cd R3GAN
+python gen_images.py --seeds=0-7 --outdir=out --class=0 \
+    --network=https://huggingface.co/brownvc/R3GAN-ImgNet-32x32/resolve/main/network-snapshot-final.pkl
+```
 
-This model is a generative adversarial network (GAN) based on the R3GAN architecture, specifically trained to synthesize high-quality and realistic images from the ImageNet dataset.
+The model is class-conditional, so `--class` is required (class 0 is tench).
+The code requires the same packages as the official StyleGAN3 repository.
 
-- **Developed by:** Brown University and Cornell University
-- **Funded by:** National Science Foundation and National Institute of Health (See paper for funding details)
-- **Shared by:** [Optional: Specify sharer if different from developer]
-- **Model type:** Generative Adversarial Network
-- **Language(s) (NLP):** N/A
-- **License:** [Specify License, e.g., MIT, Apache 2.0, or a custom license]
-- **Finetuned from model:** N/A
+## Intended use and limitations
 
-### Model Sources
+The model is intended for research on generative models. It reproduces the statistics of its training data,
+ImageNet-1k (1.28 million web images in 1,000 classes), including the known biases of a web-scraped dataset. Do not use it to create deceptive content or content that could cause harm.
 
-- **Repository:** https://github.com/brownvc/R3GAN/
-- **Paper:** https://openreview.net/forum?id=OrtN9hPP7V
-- **Demo:** [Optional: Provide a link to a demo or example usage]
+## Licence
 
-## Uses
+The weights are released under [CC BY-NC 4.0](https://creativecommons.org/licenses/by-nc/4.0/). The code that runs them is a separate work: the
+`R3GAN/` directory of the GitHub repository is under the MIT licence, and the rest of the repository, which derives
+from StyleGAN3, is under NVIDIA's Source Code License.
 
-### Direct Use
+## Citation
 
-This model can be used to generate high-resolution images similar to those in the ImageNet dataset. Its primary application includes research in generative models and image synthesis.
+```bibtex
+@inproceedings{huang2024the,
+  title={The {GAN} is dead; long live the {GAN}! A Modern {GAN} Baseline},
+  author={Nick Huang and Aaron Gokaslan and Volodymyr Kuleshov and James Tompkin},
+  booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems},
+  year={2024},
+  url={https://openreview.net/forum?id=OrtN9hPP7V}
+}
+```
 
-### Downstream Use
+## Acknowledgements
 
-The model can be fine-tuned for specific subsets of the ImageNet dataset or other similar datasets for domain-specific image generation tasks.
-
-### Out-of-Scope Use
-
-The model should not be used for generating deceptive or misleading content, malicious purposes, or tasks where realistic image synthesis could cause harm.
-
-## Bias, Risks, and Limitations
-
-The model inherits biases present in the ImageNet dataset, including potential overrepresentation or underrepresentation of certain classes. Users should critically evaluate and mitigate biases before deploying the model.
-
-### Recommendations
-
-- Avoid using the model for sensitive applications without thorough bias evaluation.
-- Ensure appropriate credit is given when publishing or sharing generated images.
-
-## How to Get Started with the Model
-
-Below is an example of how to use the model for image generation:
-- Will add later
+Brown University and Cornell University. Nick Huang was supported by a Brown University Division of Research Seed
+Award, James Tompkin by NSF CAREER 2144956, and Volodymyr Kuleshov by NSF CAREER 2145577 and NIH MIRA 1R35GM15124301.
+Compute was provided by Databricks Mosaic Research.