Release VectorSynth artifacts (models, dataset) on Hugging Face

[NielsRogge]

Hi @dcher95 🤗

Niels here from the open-source team at Hugging Face. I discovered your work on Arxiv and your paper "VectorSynth: Fine-Grained Satellite Image Synthesis with Structured Semantics" (https://huggingface.co/papers/2511.07744).

I saw in your GitHub repository (https://github.com/mvrl/VectorSynth) that you're planning to release the PyTorch ckpt files for the VectorSynth model and the associated OSM-Polygon Dataset. That's fantastic! The Hugging Face Hub would be a great platform for hosting these artifacts, improving their discoverability and visibility within the AI community. You can also claim your paper on the Hugging Face page, which will link directly to your models and datasets once they're on the Hub.

It'd be great to make the VectorSynth checkpoints and the OSM-Polygon dataset available on the 🤗 Hub. We can add tags so that people find them easily when filtering https://huggingface.co/models and https://huggingface.co/datasets.

Uploading models

For your VectorSynth model, which performs text-to-image generation from satellite imagery, we can add the text-to-image pipeline tag.

See here for a guide: https://huggingface.co/docs/hub/models-uploading.

In this case, we could leverage the PyTorchModelHubMixin class which adds from_pretrained and push_to_hub to any custom nn.Module. Alternatively, one can leverage the hf_hub_download one-liner to download a checkpoint from the Hub.

We encourage researchers to push each model checkpoint to a separate model repository, so that things like download stats also work. We can then also link the checkpoints to the paper page.

Uploading dataset

Would be awesome to make the OSM-Polygon Dataset available on 🤗 . This dataset, consisting of satellite scenes paired with pixel-registered polygon annotations, is ideal for training text-to-image models. We could tag it with text-to-image or image-segmentation task categories. Hosting it would allow people to do:

from datasets import load_dataset

dataset = load_dataset("your-hf-org-or-username/your-osm-polygon-dataset")

See here for a guide: https://huggingface.co/docs/datasets/loading.

Besides that, there's the dataset viewer which allows people to quickly explore the first few rows of the data in the browser.

Please let me know if you're interested or need any guidance once the models and dataset are ready for release! We're here to assist with any help you might need.

Cheers,

Niels
ML Engineer @ HF 🤗