Upload models to the Hugging Face Hub

[osanseviero]

Hi!

Very cool work! It would be nice to have the model checkpoints on the Hugging Face Hub rather than a Dropbox link

Some of the benefits of sharing your models through the Hub would be:

• versioning, commit history and diffs
• repos provide useful metadata about their tasks, languages, metrics, etc that make them discoverable
• multiple features from TensorBoard visualizations, PapersWithCode integration, and more
• wider reach of your work to the ecosystem

Creating the repos and adding new models should be a relatively straightforward process if you've used Git before. This is a step-by-step guide explaining the process in case you're interested. Please let us know if you would be interested and if you have any questions.

[yilunliao]

Hi @osanseviero

I see. Thanks for the suggestion!

Let me ask other people to see if that is ok to them. If yes, I will do that this weekend or next week and get back to you.

Please ping me if you do not receive my response after that.

[yilunliao]

Hi @osanseviero

I have uploaded models to my personal Hugging Face Hub.

Just curious if you have some good examples of using the Hub to store models.

The checkpoints of this work would be stored along with other OCP models. However, I would use the Hub to store models in my future works.

[osanseviero]

Hi there! You can explore some of the most popular models here, from Stable Diffusion to StarCoder, bert, Falcon, and many others :) In general, we recommend one model repository per model.

Model cards are recommended to ensure the community is able to understand the model, how to use and which are its limitations or intended use cases. We have docs about them in https://huggingface.co/docs/hub/model-cards.

You can also use the huggingface_hub Python library to access models on the Hub programmatically

[yilunliao]

@osanseviero I see. That looks pretty great. Let me learn more and use it for future works if applicable.

Thanks for the comment.