This is the Linux app named Google DeepMind GraphCast and GenCast whose latest release can be downloaded as Version0.1.1sourcecode.tar.gz. It can be run online in the free hosting provider OnWorks for workstations.
Download and run online this app named Google DeepMind GraphCast and GenCast with OnWorks for free.
이 앱을 실행하려면 다음 지침을 따르세요.
- 1. 이 애플리케이션을 PC에 다운로드했습니다.
- 2. 파일 관리자 https://www.onworks.net/myfiles.php?username=XXXXX에 원하는 사용자 이름을 입력합니다.
- 3. 이러한 파일 관리자에서 이 응용 프로그램을 업로드합니다.
- 4. 이 웹사이트에서 OnWorks Linux 온라인 또는 Windows 온라인 에뮬레이터 또는 MACOS 온라인 에뮬레이터를 시작합니다.
- 5. 방금 시작한 OnWorks Linux OS에서 원하는 사용자 이름으로 파일 관리자 https://www.onworks.net/myfiles.php?username=XXXXX로 이동합니다.
- 6. 응용 프로그램을 다운로드하여 설치하고 실행합니다.
스크린 샷
Ad
Google DeepMind GraphCast 및 GenCast
기술
GraphCast, developed by Google DeepMind, is a research-grade weather forecasting framework that employs graph neural networks (GNNs) to generate medium-range global weather predictions. The repository provides complete example code for running and training both GraphCast and GenCast, two models introduced in DeepMind’s research papers. GraphCast is designed to perform high-resolution atmospheric simulations using the ERA5 dataset from ECMWF, while GenCast extends the approach with diffusion-based ensemble forecasting for probabilistic weather prediction. Both models are built on JAX and integrate advanced neural architectures capable of learning from multi-scale geophysical data represented on icosahedral meshes. The package includes pretrained model weights, normalization statistics, and demonstration notebooks that allow users to replicate and fine-tune weather forecasting experiments in Colab or on Google Cloud TPUs and GPUs.
기능
- Implements GraphCast and GenCast architectures for data-driven weather forecasting
- Pretrained model weights and normalization data available via Google Cloud Bucket
- JAX-based differentiable simulation framework using graph neural networks
- Colab-ready demonstration notebooks for quick experimentation and learning
- Compatible with ERA5 and HRES datasets for historical and operational fine-tuning
- Supports execution on TPUs and GPUs for scalable model training and inference
프로그래밍 언어
Python
카테고리
This is an application that can also be fetched from https://sourceforge.net/projects/g-deepmind-graph-gen.mirror/. It has been hosted in OnWorks in order to be run online in an easiest way from one of our free Operative Systems.
