The prototype of ML models.
The project is implemented based on Torch Lightning.
Install the dependencies with uv:
uv syncuv manages dependencies declared in pyproject.toml and keeps the
uv.lock file in sync.
-
Add a runtime package
uv add <package>
-
Add a development package
uv add --dev <package>
-
Remove a package
uv remove <package>
-
Install dev dependencies and run tests
uv sync pytest
The config file is in ml_prototype/config/transformer_lm.yaml.
Run the training job with the following command:
python ml_prototype/cli.py fit --config ml_prototype/config/transformer_lm.yaml