Migrate tests to pytest + add GA (#127)

Co-authored-by: Eitan Turok <[email protected]>

.github/workflows/pr-gpu.yaml

@@ -0,0 +1,45 @@
+name: PR GPU tests
+on:
+  push:
+    branches:
+    - main
+    - release/*
+  pull_request_target:
+    branches:
+    - main
+    - release/**
+  workflow_dispatch:
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main'}}
+jobs:
+  pytest-gpu:
+    name: ${{ matrix.name }}
+    runs-on: linux-ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+        - name: "python3.11-pytorch2.3.1-gpus1"
+          gpu_num: 1
+          python_version: 3.11
+          container: mosaicml/pytorch:2.3.1_cu121-python3.11-ubuntu20.04
+        - name: "python3.11-pytorch2.3.1-gpus2"
+          gpu_num: 2
+          python_version: 3.11
+          container: mosaicml/pytorch:2.3.1_cu121-python3.11-ubuntu20.04
+    steps:
+    - name: Run PR GPU tests
+      uses: mosaicml/ci-testing/.github/actions/[email protected]
+      with:
+        name: ${{ matrix.name }}
+        container: ${{ matrix.container }}
+        python_version: ${{ matrix.python_version }}
+        gpu_num: ${{ matrix.gpu_num }}
+        git_repo: databricks/megablocks
+        pip_deps: "[all,testing]"
+        pytest_command: "coverage run -m pytest tests" # todo: remove tests from pytest tests when we delete all tests outside of MegaBlocks repo
+        pytest_markers: "gpu"
+        composer_package_name: mosaicml  # Required as Composer is built from source
+        mcloud_timeout: 3600
+        mcloud_api_key: ${{ secrets.MCLOUD_API_KEY }}

.gitignore

@@ -1,8 +1,165 @@
-*~
-gpt2-merges.txt
-gpt2-vocab.json
-core
-__pycache__
-*.egg-info
-dist
-build
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+pip-wheel-metadata/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+docs/path/
+docs/source/_build/
+docs/source/api_reference/
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+.python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# Virtual Env
+venv/
+
+# WandB
+wandb/
+
+# Neptune
+.neptune/
+
+# Spacemacs
+._#*
+.#*
+.vscode/
+
+# Vim
+*.swp
+
+# README backups
+README.md.bkp
+
+# OS X
+.DS_Store
+
+# Sphinx
+api_reference/
+
+# Node (for pyright)
+node_modules/
+package.json
+package-lock.json
+
+# pycharm
+.idea/

Makefile

@@ -1,11 +1,23 @@
-clean:
-	rm -rf dist/*
+# several pytest settings
+WORLD_SIZE ?= 1  # world size for launcher tests
+MASTER_PORT ?= 26000 # port for distributed tests
+PYTHON ?= python3  # Python command
+PYTEST ?= pytest  # Pytest command
+PYRIGHT ?= pyright  # Pyright command. Pyright must be installed seperately -- e.g. `node install -g pyright`
+EXTRA_ARGS ?=  # extra arguments for pytest
+EXTRA_LAUNCHER_ARGS ?= # extra arguments for the composer cli launcher
 
-dist: clean
-	python3 setup.py sdist
+test:
+	LOCAL_WORLD_SIZE=1 $(PYTHON) -m $(PYTEST) $(EXTRA_ARGS)
 
-upload: dist
-	twine upload dist/*
+test-gpu:
+	LOCAL_WORLD_SIZE=1 $(PYTHON) -m $(PYTEST) -m gpu $(EXTRA_ARGS)
 
-upload-test: dist
-	twine upload --repository-url https://test.pypi.org/legacy/ dist/*
+# runs tests with the launcher
+test-dist:
+	$(PYTHON) -m composer.cli.launcher -n $(WORLD_SIZE) --master_port $(MASTER_PORT) $(EXTRA_LAUNCHER_ARGS) -m $(PYTEST) $(EXTRA_ARGS)
+
+test-dist-gpu:
+	$(PYTHON) -m composer.cli.launcher -n $(WORLD_SIZE) --master_port $(MASTER_PORT) $(EXTRA_LAUNCHER_ARGS) -m $(PYTEST) -m gpu $(EXTRA_ARGS)
+
+.PHONY: test test-gpu test-dist test-dist-gpu

README.md

@@ -22,6 +22,8 @@ NOTE: This assumes you have `numpy` and `torch` installed.
 
 Installing `megablocks[gg]` enables dMoE computation with grouped GEMM. This feature is enabled by setting the `mlp_impl` argument to `grouped`. This is currently our recommended path for Hopper-generation GPUs.
 
+Installing `megablocks[dev]` allows you to contribute to MegaBlocks and test locally. Installing `megablocks[testing]` allows you to test via Github Actions.
+
 MegaBlocks can be installed with all dependencies via the `megablocks[all]` package.
 
 # :steam_locomotive: Usage