ECMWF Anemoi

Star 1
Version 1.0.1
ECMWF Anemoi
Maintainers
support[at]europeanweather.cloud
Support level
EWC
Category
Machine Learning & AI GPU-accelerated
Technology
Ansible Playbook

ECMWF Anemoi

Create a conda environment with the Anemoi framework for weather forecasting based on machine learning and associated software stack.

Requirements

This ansible role depends on ewc-ansible-role-conda and needs to be applied on a GPU-powered instance if training or inference are used.

Role Variables

  • anemoi_env_wipe: Boolean to decide whether to wipe the environment if exists prior to a reinstallation. Default: no
  • anemoi_env_name: Name of the environment containing the software stack. Default: aifs-single-mse
  • anemoi_env_path: Installation path for the environment. Default: "{{ conda_prefix }}/envs/{{ anemoi_env_name }}"
  • anemoi_create_ipykernel: Create the jupyter kernel for this environment. Default: yes
  • conda_prefix: Prefix where conda is installed. Default: /opt/conda
  • conda_user: User owning the conda installation. Default: root

Example Playbook

- hosts: all
  roles:
    -  ewc-ansible-role-anemoi

License

Apache 2.0.

Author Information

ECMWF for the European Weather Cloud

ewc logo

Other
Deployable
EWCCLI-compatible