Improve discoverability of your work on HF

[NielsRogge]

Hi,

Niels here from the open-source team at Hugging Face. Congrats on your work! I found it based on the paper page: https://huggingface.co/papers/2406.03184 (feel free to claim the paper with your HF account so that it appears under your profile).

It would be great to host checkpoints on the hub to improve visibility, here's some guidance.

1. Upload checkpoints to the hub

For custom PyTorch models, we recommend leveraging the PyTorchModelHubMixin class, as it adds push_to_hub and from_pretrained capabilities to any custom nn.Module. It creates a config.json along with safetensors for each model, enforcing downloads to work. Do note that this class assumes that all init arguments are JSON serializable (the class cannot take other nn.Module's in its init).

Also, would be great to tag them with "image-to-3D" in the model card, so that people will find your models when searching https://huggingface.co/models?pipeline_tag=image-to-3d&sort=trending.

2. Make the model Diffusers compatible

In case you want your models to be usable through the Diffusers library, let us know so that we can help you in the process.

Cheers,

Niels
ML Engineer @ HF 🤗