---
license: cc-by-nc-nd-4.0
library_name: onnxruntime
tags:
  - climate
  - onnx
  - seasonal
  - subseasonal
  - weather
---

# FuXi-S2S ONNX checkpoint mirror

This repository is an unchanged convenience mirror of the official FuXi-S2S
ONNX checkpoint published by the FuXi Team. The authoritative release remains
the [Zenodo record](https://zenodo.org/records/15718402), and model usage is
documented in the [official FuXi-S2S repository](https://github.com/tpys/FuXi-S2S).

The mirror exists so downstream integrations such as
[NVIDIA Earth2Studio](https://github.com/NVIDIA/earth2studio) can retrieve the
ONNX graph and its large external-data file reliably and pin an immutable
revision.

## Files and provenance

The two files were extracted without modification from `model-1.0.tar` in the
official Zenodo release. They must remain in the same directory because the
ONNX graph refers to its external weights by the sibling filename
`fuxi_s2s`.

| File | Size (bytes) | SHA-256 |
| --- | ---: | --- |
| `fuxi_s2s.onnx` | 1,588,847 | `bcc3419ef698cb894b17fae2c5e40ded05d87ed8d4ba9f6342c9f0afff3c70f5` |
| `fuxi_s2s` | 2,078,823,204 | `628d86076810d3d5c53a7a27f10e67b08051e2d2e5c9224a2df96cba8ecd8ea8` |

The source archive is 2,080,419,840 bytes with MD5 checksum
`17790f6bb7c3f83ce846d36f2339c4b3`, matching the checksum published by
Zenodo. The separate `data.zip` validation sample is intentionally not
mirrored here and remains available from the authoritative record.

## License and use restrictions

The checkpoint is licensed under
[CC BY-NC-ND 4.0](https://creativecommons.org/licenses/by-nc-nd/4.0/).
The official record states that it is intended for research use and that
commercial or competition use requires prior permission from the authors.
Those checkpoint terms apply independently of the license of any downstream
integration.

## Citation

If you use this checkpoint, cite the original work:

```bibtex
@article{chen2024machine,
  title = {A machine learning model that outperforms conventional global subseasonal forecast models},
  author = {Chen, Lei and Zhong, Xiaohui and Li, Hao and Wu, Jie and Lu, Bo and Chen, Deliang and Xie, Shang-Ping and Wu, Libo and Chao, Qingchen and Lin, Chensen and others},
  journal = {Nature Communications},
  volume = {15},
  number = {6425},
  pages = {1--14},
  year = {2024},
  publisher = {Nature Publishing Group UK London},
  doi = {10.1038/s41467-024-50714-1}
}
```