Hi @XinDiW 🤗
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/2602.16704.
The paper page lets people discuss about your paper and lets them find artifacts about it (your models for instance),
you can also claim the paper as yours which will show up on your public profile at HF, and add Github and project page URLs.
I saw that you have already released the implementation code for ReFINE and linked to the baseline LaCT and DeltaNet checkpoints. It is great to see these architectures being explored!
Would you like to host the ReFINE-tuned model checkpoints (the NSP-trained versions of LaCT-760M and DeltaNet-1.3B) on https://huggingface.co/models?
Given the significant performance gains you've demonstrated across LongBench and RULER, having these specific weights available would provide great value to the community. Hosting on Hugging Face will give your work more visibility and enable better discoverability through metadata tags and paper linking.
If you're down, I'm leaving a guide here. For custom architectures, you can use the PyTorchModelHubMixin class which makes it very easy for users to load the models.
After they are uploaded, we can also link the models to the paper page so people can discover and use them right away.
Let me know if you're interested or need any guidance!
Kind regards,
Niels
Hi @XinDiW 🤗
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/2602.16704.
The paper page lets people discuss about your paper and lets them find artifacts about it (your models for instance),
you can also claim the paper as yours which will show up on your public profile at HF, and add Github and project page URLs.
I saw that you have already released the implementation code for ReFINE and linked to the baseline LaCT and DeltaNet checkpoints. It is great to see these architectures being explored!
Would you like to host the ReFINE-tuned model checkpoints (the NSP-trained versions of LaCT-760M and DeltaNet-1.3B) on https://huggingface.co/models?
Given the significant performance gains you've demonstrated across LongBench and RULER, having these specific weights available would provide great value to the community. Hosting on Hugging Face will give your work more visibility and enable better discoverability through metadata tags and paper linking.
If you're down, I'm leaving a guide here. For custom architectures, you can use the PyTorchModelHubMixin class which makes it very easy for users to load the models.
After they are uploaded, we can also link the models to the paper page so people can discover and use them right away.
Let me know if you're interested or need any guidance!
Kind regards,
Niels