Hi @d7chong 馃
I'm Niels and work as part of the open-source team at Hugging Face. I discovered your work through Papers with Code as yours got featured: https://paperswithcode.co/paper/2602.06694.
The paper page lets people discuss your paper and 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, add Github and project page URLs.
Your results on sub-1-bit quantization are very impressive, especially getting a 70B model to run on an 8GB consumer GPU! I noticed that while the code is available, the pre-quantized checkpoints for models like Qwen or Llama-3 aren't hosted yet. I also saw some interest in the community comments on the paper page regarding this.
Would you like to host the quantized checkpoints you've produced on https://huggingface.co/models?
Hosting on Hugging Face will give your work more visibility and enable better discoverability. We can add tags in the model cards so that people find the models easier, link them to the paper page, and allow users to download them directly.
If you're down, leaving a guide here. Since these involve custom kernels, you might also be interested in the PyTorchModelHubMixin class which adds from_pretrained and push_to_hub functionality to your model classes.
After uploaded, we can also link the models to the paper page (read here) so people can discover your work.
Let me know if you're interested or need any guidance!
Kind regards,
Niels
Hi @d7chong 馃
I'm Niels and work as part of the open-source team at Hugging Face. I discovered your work through Papers with Code as yours got featured: https://paperswithcode.co/paper/2602.06694.
The paper page lets people discuss your paper and 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, add Github and project page URLs.
Your results on sub-1-bit quantization are very impressive, especially getting a 70B model to run on an 8GB consumer GPU! I noticed that while the code is available, the pre-quantized checkpoints for models like Qwen or Llama-3 aren't hosted yet. I also saw some interest in the community comments on the paper page regarding this.
Would you like to host the quantized checkpoints you've produced on https://huggingface.co/models?
Hosting on Hugging Face will give your work more visibility and enable better discoverability. We can add tags in the model cards so that people find the models easier, link them to the paper page, and allow users to download them directly.
If you're down, leaving a guide here. Since these involve custom kernels, you might also be interested in the PyTorchModelHubMixin class which adds
from_pretrainedandpush_to_hubfunctionality to your model classes.After uploaded, we can also link the models to the paper page (read here) so people can discover your work.
Let me know if you're interested or need any guidance!
Kind regards,
Niels