`interpret` support for model predictions with response levels #1109

	name: Run tests

	on:
	push:
	branches:
	- main
	pull_request:

	jobs:
	test:
	runs-on: ubuntu-latest
	strategy:
	matrix:
	python-version: ["3.9", "3.10", "3.11"]

	name: Set up Python ${{ matrix.python-version }}
	steps:
	- name: Checkout source
	uses: actions/checkout@v2
	with:
	fetch-depth: 0

	- name: Set up Python ${{ matrix.python-version }}
	uses: conda-incubator/setup-miniconda@v2
	with:
	channels: conda-forge, defaults
	channel-priority: true
	python-version: ${{ matrix.python-version }}
	auto-update-conda: true

	- shell: bash -l {0}
	run: \|
	conda info
	conda list

	- name: Install Bambi and all its dependencies
	shell: bash -l {0}
	run: \|
	conda install -c conda-forge python-graphviz
	conda install pip
	pip install .
	pip install .[dev]
	pip install .[jax]

	python --version
	conda list
	pip freeze

	- name: Run linters
	shell: bash -l {0}
	run: \|
	echo "Running black..."
	black bambi --check --diff --color

	echo "Checking code style with pylint..."
	pylint bambi

	- name: Run tests
	shell: bash -l {0}
	run: python -m pytest -vv --cov=bambi --cov-report=term --cov-report=xml tests
	env:
	PYTHON_VERSION: ${{ matrix.python-version }}

	- name: Upload coverage to Codecov
	uses: codecov/codecov-action@v1
	with:
	env_vars: OS,PYTHON
	name: codecov-umbrella
	fail_ci_if_error: false

Provide feedback