Hi @uanu and the @SpectrAI-Initiative team 馃
Niels here from the open-source team at Hugging Face. I discovered your amazing work on SciReasoner through your Hugging Face paper page: https://huggingface.co/papers/2607.07708.
The paper page lets people discuss your work and discover related artifacts (like models, datasets, or demos). You can also claim this paper as yours to show it on your public profiles at HF, and link your GitHub and project page URLs.
I noticed on your GitHub repository (https://github.com/SpectrAI-Initiative/SciReasoner) that the model checkpoints and demo are marked as "Coming soon". When you are ready for the public release, would you be interested in hosting the pre-trained checkpoints and any associated evaluation datasets on the Hugging Face Hub?
Hosting on Hugging Face will significantly improve discoverability and integration into the scientific ML community. We can add custom metadata tags to the model and dataset cards so people can easily find them when filtering for biology, chemistry, or materials science.
Uploading models
See here for a guide: https://huggingface.co/docs/hub/models-uploading.
Since SciReasoner is initialized from Qwen3-14B, it can easily be loaded and shared. If you are using custom PyTorch modules, you can 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 work. We can then link the checkpoints directly to your paper page.
Uploading dataset
Would be awesome to make the processed benchmark configs and evaluation datasets available on 馃, so that people can easily load them:
from datasets import load_dataset
dataset = load_dataset("your-hf-org-or-username/your-dataset")See here for a guide: https://huggingface.co/docs/datasets/loading.
In addition, we would love to help you host your upcoming demo on Hugging Face Spaces. We can provide you with a ZeroGPU grant to give you free GPU-backed compute for your demo space!
Let me know if you're interested or if you need any guidance when you are ready to release!
Best regards,
Niels
ML Engineer @ HF 馃
Hi @uanu and the @SpectrAI-Initiative team 馃
Niels here from the open-source team at Hugging Face. I discovered your amazing work on SciReasoner through your Hugging Face paper page: https://huggingface.co/papers/2607.07708.
The paper page lets people discuss your work and discover related artifacts (like models, datasets, or demos). You can also claim this paper as yours to show it on your public profiles at HF, and link your GitHub and project page URLs.
I noticed on your GitHub repository (https://github.com/SpectrAI-Initiative/SciReasoner) that the model checkpoints and demo are marked as "Coming soon". When you are ready for the public release, would you be interested in hosting the pre-trained checkpoints and any associated evaluation datasets on the Hugging Face Hub?
Hosting on Hugging Face will significantly improve discoverability and integration into the scientific ML community. We can add custom metadata tags to the model and dataset cards so people can easily find them when filtering for biology, chemistry, or materials science.
Uploading models
See here for a guide: https://huggingface.co/docs/hub/models-uploading.
Since SciReasoner is initialized from Qwen3-14B, it can easily be loaded and shared. If you are using custom PyTorch modules, you can leverage the PyTorchModelHubMixin class which adds
from_pretrainedandpush_to_hubto any customnn.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 work. We can then link the checkpoints directly to your paper page.
Uploading dataset
Would be awesome to make the processed benchmark configs and evaluation datasets available on 馃, so that people can easily load them:
See here for a guide: https://huggingface.co/docs/datasets/loading.
In addition, we would love to help you host your upcoming demo on Hugging Face Spaces. We can provide you with a ZeroGPU grant to give you free GPU-backed compute for your demo space!
Let me know if you're interested or if you need any guidance when you are ready to release!
Best regards,
Niels
ML Engineer @ HF 馃