Release Sphere Encoder checkpoints on Hugging Face

Hi @kaiyuyue 🤗

I'm Niels and work as part of the open-source team at Hugging Face. I discovered your work through Hugging Face's daily papers as yours got featured: https://huggingface.co/papers/2602.15030.

The paper page lets people discuss your research and find related artifacts. I noticed that you've already shared the FID evaluation artifacts on the Hub (https://huggingface.co/datasets/kaiyuyue/sphere-encoder-fid-artifacts), which is great for reproducibility!

Would you also like to host the pre-trained model checkpoints (such as `sphere-xl-imagenet`, `sphere-l-af`, and `sphere-l-of`) on https://huggingface.co/models?

Hosting the weights on Hugging Face will provide much better visibility and discoverability for your work. We can add metadata tags to the model cards so people can easily find and cite them, and link them directly to the paper page.

If you're interested, you can find a guide for uploading [here](https://huggingface.co/docs/hub/models-uploading). Since these are custom PyTorch models, you can also use the [PyTorchModelHubMixin](https://huggingface.co/docs/huggingface_hub/package_reference/mixins#huggingface_hub.PyTorchModelHubMixin) class to add `from_pretrained` and `push_to_hub` methods, allowing the community to download and use the models instantly.

We could also help you set up a Gradio demo on [Spaces](https://huggingface.co/spaces) to showcase the single-pass generation, and we can provide a ZeroGPU grant for it.

Let me know if you're interested or need any guidance!

Kind regards,

Niels