Release ImpScore and its training data on Hugging Face

[NielsRogge]

Hi @audreycs 🤗

I'm Niels from the open-source team at Hugging Face. I came across your work on ImpScore via the Hugging Face Papers page: https://huggingface.co/papers/2411.05172. It's impressive work!

We noticed that ImpScore and its training data are currently available via pip and your GitHub repo. To maximize their impact and facilitate easier access for the research community, we would love to help you host both the ImpScore model and the training dataset on the Hugging Face Hub.

Hosting on Hugging Face provides numerous benefits, including:

• Increased visibility: Researchers actively searching for resources related to implicit language understanding will easily discover your work.
• Improved discoverability: You can add rich metadata (tags, descriptions, etc.) to the model and dataset cards, making your work easier to find.
• Simplified access: Researchers can directly download and use your artifacts with Hugging Face's tools, such as transformers and datasets.

For the ImpScore model, we can help you create a model card with relevant details and metadata. For the dataset, we can guide you through the upload process and help create a dataset card that highlights the key features of your dataset.

Are you interested in exploring this option? Let me know if you'd like assistance with the upload process or have any questions.

Best Regards,

Niels
ML Engineer @ Hugging Face 🤗