Main working repository for Navy ML SBIR Phase I feasibility study
The objective of this repository is to link multiple repositories from both inside and outside of Sofar in order to organize all of the working components that will go into the construction of a hybrid AI/ML-NWP forecast system.
The main objectives of the Navy SBIR are to:
Our execution strategy will be to develop tools to incorporate the AI/ML models into our existing ESM workflow, so as to not require any new significant infrastructure. This will also help us benefit from developments in this project towards our general ESM forecast system goals (e.g. adding new diagnostics and metrics to esm-assessment).
(optional, set up python virtual environment) With default python and venv:
python3 -m venv ./.venv
Or with penv: https://dev.to/mattcale/pyenv-poetry-bffs-20k6
pyenv exec python -m venv .venv
Activate:
source ./.venv/bin/activate
For forpy, need to add:
export PYTHONPATH=$PWD/.venv/lib/python3.10/site-packages
locally in your virtual environment for this project
(the way to go if you don't want to use poetry globally and only for this project)
python3 -m pip install poetry
(optional) Enable poetry completions:
poetry completions bash >> ~/.bash_completion
poetry install
poetry install --with dev,cloud,neuralgcm
poetry config --list
poetry show
poetry run jupyter lab &
(optional) execute pre-commit install to install git hooks in your .git/ directory
this will allow linting and formatting on commit
pre-commit install
To build the dockerfile:
docker build -f ./Dockerfile.neuralgcm -t sofarocean/mlwp-neuralgcm:0.0.0 .
Content type
Image
Digest
sha256:af5255260…
Size
1.3 GB
Last updated
almost 2 years ago
docker pull sofarocean/mlwp-neuralgcm