Installation¶
Use uv for local development unless you have a reason to use pip directly.
Core library setup¶
Install the package and its core runtime dependencies:
uv sync
This gives you the library with the dependencies needed to run evaluations.
Development setup¶
Install the package with test, lint, and type-check tooling:
uv sync --extra dev
Run the test suite:
uv run pytest
Example and visualization dependencies¶
Install optional dependencies used by the example scripts and Streamlit demo:
uv sync --extra examples
Example commands:
uv run python examples/test.py
uv run python examples/visualize_results.py
uv run streamlit run examples/streamlit_app.py
Documentation dependencies¶
Install the documentation toolchain:
uv sync --extra docs
Serve the docs site locally:
uv run mkdocs serve
pip alternative¶
If you are not using uv, install the package in editable mode:
pip install -e .
For development tools:
pip install -e ".[dev]"
What to read next¶
- Quickstart for the smallest realistic evaluation
- Choosing a Task Type if you are not sure which evaluator shape matches your data