ml-wrappers

• fix error in ml-wrappers pypi release script by @imatiach-msft in https://github.com/microsoft/ml-wrappers/pull/51
• Add issues template for bugs and feature requests by @imatiach-msft in https://github.com/microsoft/ml-wrappers/pull/52
• Update requirements-dev.txt to pin transformers to less than 4.20.0 by @gaugup in https://github.com/microsoft/ml-wrappers/pull/54
• Update CI-python.yml to add python 3.10 tests by @gaugup in https://github.com/microsoft/ml-wrappers/pull/53
• add catboost framework support to ml-wrappers by @imatiach-msft in https://github.com/microsoft/ml-wrappers/pull/55
• prevent build error via lower numpy version from conda to prevent re-install from pip by @imatiach-msft in https://github.com/microsoft/ml-wrappers/pull/57
• release ml-wrappers v0.2.1 by @imatiach-msft in https://github.com/microsoft/ml-wrappers/pull/56

Full Changelog: https://github.com/microsoft/ml-wrappers/compare/v0.2.0...v0.2.1

• Update requirements-linting.txt to remove flake8-nb by @gaugup in https://github.com/microsoft/ml-wrappers/pull/45
• Update github action for checkout, upload-artifact and setup-python from version 2 to 3 by @gaugup in https://github.com/microsoft/ml-wrappers/pull/44
• add pypi prod and test release workflow github action to ml-wrappers repository by @imatiach-msft in https://github.com/microsoft/ml-wrappers/pull/46
• add a wrapper for question answering models in ml-wrappers repository by @imatiach-msft in https://github.com/microsoft/ml-wrappers/pull/47
• add image model wrapper by @imatiach-msft in https://github.com/microsoft/ml-wrappers/pull/48
• update CustomTimestampFeaturizer to conform to the scikit-learn custom transformer specifications by @imatiach-msft in https://github.com/microsoft/ml-wrappers/pull/49

Full Changelog: https://github.com/microsoft/ml-wrappers/compare/v0.1.0...v0.2.0

• add tests for wrap_model function
• Correctly annotate python warnings
• Replace numpy.array with numpy.ndarray in docstrings
• make datasetwrapper optional when calling wrap_model and improve README
• add wrapper for text-based models
• add tensorflow model wrapper and refactor pytorch model wrapper
• Add flake8-pytest-style to linting
• add support for tensorflow BatchDataset to DatasetWrapper

• add code scanning pipeline using CodeQL
• update gitignore file for python
• add code coverage to ml-wrappers repository
• fix namespaces on doc strings for DatasetWrapper
• fix imports in model init file to include WrappedClassificationModel and WrappedRegressionModel

• fix dataset wrapper to support more input types

• continuous integration setup
• move tests to top-level folder
• update main readme and add specification docs
• add release process doc for ml-wrappers repository
• fix issues with circular dependencies in ml-wrappers
• rename test directory to tests
• add python 3.9 tests
• add supported type check in DatasetWrapper
• separate out python linting into separate workflow
• move test_dataset_wrapper.py from test/ to tests/
• suppress tensorflow warnings that may sometimes occur on import

• refactor timestamp featurizer and fix model init file

• fix dataset wrapper folder init file

Initial release of ml-wrappers package

The Machine Learning Wrappers SDK provides a unified wrapper for various ML frameworks - to have one uniform scikit-learn format predict and predict_proba functions.

Highlights of the package include:

A dataset wrapper to handle scipy sparse, pandas and numpy datasets in a uniform manner.
A model wrapper to handle models from various frameworks uniformly, including scikit-learn, tensorflow, pytorch, lightgbm and xgboost
Please see the github website for the documentation and sample notebooks:
https://github.com/microsoft/ml-wrappers