Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
Wenjie Hu 6e3f934533 | 1 year ago | |
---|---|---|
data_factory | 1 year ago | |
img | 1 year ago | |
model | 1 year ago | |
utils | 1 year ago | |
.gitignore | 1 year ago | |
LICENSE | 1 year ago | |
README.md | 1 year ago | |
README_en.md | 1 year ago | |
train_fourcastnet.py | 1 year ago | |
train_graphcast.py | 1 year ago |
简体中文 | English
本项目是由幻方AI团队复现优化,并开源的全球AI气象预报模型工具库。基于 FourCastNet 和 GraphCast 的论文,我们构建了一个新的全球AI气象预报项目——OpenCastKit,它能够与欧洲中期天气预报中心(ECMWF)的传统物理模型——高分辨率综合预测系统(IFS),进行直接比较。
我们将基于1979年1月到2022年12月的ERA5数据训练出来的模型参数开源到 Hugging Face 仓库中,并上线了一个每日更新的 HF-Earth,展示模型的预测效果。
下面是一些预测案例:
原始数据来自欧洲中期天气预报中心(ECMWF)提供的一个公开可用的综合数据集 ERA5 ,需要通过 data_factory/convert_ear5_hourly.py
脚本提出数据特征,转化为高性能训练样本格式 FFRecord 下的样本数据。
本地运行:
python train_fourcastnet.py --pretrain-epochs 100 --fintune-epochs 40 --batch-size 4
也可以提交任务至幻方萤火集群,使用96张A100进行数据并行训练
hfai python train_fourcastnet.py --pretrain-epochs 100 --fintune-epochs 40 --batch-size 4 -- -n 12 --name train_fourcastnet
本地运行:
python train_graphcast.py --epochs 200 --batch-size 2
也可以提交任务至幻方萤火集群,使用256张A100进行流水线并行训练
hfai python train_graphcast.py --epochs 200 --batch-size 2 -- -n 32 --name train_graphcast
@article{pathak2022fourcastnet,
title={Fourcastnet: A global data-driven high-resolution weather model using adaptive fourier neural operators},
author={Pathak, Jaideep and Subramanian, Shashank and Harrington, Peter and Raja, Sanjeev and Chattopadhyay, Ashesh and Mardani, Morteza and Kurth, Thorsten and Hall, David and Li, Zongyi and Azizzadenesheli, Kamyar and others},
journal={arXiv preprint arXiv:2202.11214},
year={2022}
}
@article{remi2022graphcast,
title={GraphCast: Learning skillful medium-range global weather forecasting},
author={Remi Lam, Alvaro Sanchez-Gonzalez, Matthew Willson, Peter Wirnsberger, Meire Fortunato, Alexander Pritzel, Suman Ravuri, Timo Ewalds, Ferran Alet, Zach Eaton-Rosen, Weihua Hu, Alexander Merose, Stephan Hoyer, George Holland, Jacklynn Stott, Oriol Vinyals, Shakir Mohamed, Peter Battaglia},
journal={arXiv preprint arXiv:2212.12794},
year={2022}
}
No Description
Python Markdown
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》