v1.0.2

What's Changed

• Use local environment for pre-commit by @thomasgaudelet in #146
• Readthedoc build fix & single-source of requirements by @thomasgaudelet in #140
• Update pandas requirement from <2.2.1,>=1.3.0 to >=1.3.0,<2.2.2 in /requirements by @dependabot in #141
• Bump pytest from 8.0.0 to 8.1.1 in /requirements by @dependabot in #145
• Update scikit-learn requirement from <1.2.2,>=1.0.2 to >=1.0.2,<1.4.2 in /requirements by @dependabot in #142
• Update version.py by @thomasgaudelet in #149
• Bump black from 24.2.0 to 24.3.0 in /requirements by @dependabot in #143
• Update pandas requirement from <2.2.2,>=1.3.0 to >=1.3.0,<2.2.3 in /requirements by @dependabot in #150
• Update scikit-learn requirement from <1.4.2,>=1.0.2 to >=1.0.2,<1.4.3 in /requirements by @dependabot in #148
• Bump mypy from 1.8 to 1.9.0 in /requirements by @dependabot in #144

Full Changelog: v1.0.1...v1.0.2

v1.0.1

What's Changed

• Fix for compatibility with new versions of pytorch lightning that remove support for the gpus argument by @thomasgaudelet in #112
• Fix bugs in examples and simplify imports to remove need to temper with PYTHONPATH by @thomasgaudelet in #113
• Update lightning requirement from <2.1.0,>=1.8.6 to >=1.8.6,<2.3.0 by @dependabot in #114
• Update numpy requirement from <1.24.3,>=1.22.0 to >=1.22.0,<1.26.5 by @dependabot in #115
• Bump flake8 from 6.0.0 to 7.0.0 by @dependabot in #117
• Update torch requirement from <1.14.0,>=1.9.0 to >=1.9.0,<2.3.0 by @dependabot in #116
• Update pandas requirement from <1.5.4,>=1.3.0 to >=1.3.0,<2.2.1 by @dependabot in #120
• Update gpytorch requirement from <1.10.0,>=1.4.0 to >=1.4.0,<1.12.0 by @dependabot in #118
• Bump black from 22.12.0 to 24.2.0 by @dependabot in #121
• Bump pytest from 7.2.1 to 8.0.0 by @dependabot in #123
• Bump parameterized from 0.8.1 to 0.9.0 by @dependabot in #125
• Bump flake8-bugbear from 23.1.20 to 24.2.6 by @dependabot in #124
• Update torchvision requirement from <0.15.0,>=0.10.0 to >=0.10.0,<0.18.0 by @dependabot in #126

Full Changelog: v1.0.0...v1.0.1

Major version update

In this new release, we introduce significant changes to the overall framework by adding a clear separation between pipeline management and active learning strategy. The latter is now only responsible for query selection, while the former is now handled by a specific Pipeline class responsible for interfacing the now four components of the active learning loop:

• the data manager,
• the model manager,
• the active learning strategy, or simply strategy,
• the oracle

We also add a template class for oracles such that users can design and connect their own to PyRelationAL pipeline. By default, the pipeline will use a BenchmarkOracle which assumes the dataset contains all the labels already.

The new release introduces a few syntactic changes such as renaming of classes and methods, to make the syntax lighter and/or more consistent across the library, as well as a refactor of the unit test suite to improve clarity.

See below for a full list of changes and fixes.

What's Changed

• bug fix due to new scipy versions removing scipy.random by @ojpb in #11
• Shuffling labelled set by @ojpb in #18
• Fixes to badge example by @alicerdv in #15
• Add missing Doc strings by @alicerdv in #20
• Moving the call function of models to a specific device by @alicerdv in #17
• Automate codecov by @RelationTherapeutics in #21
• dependabot update requirements.txt by @thomasgaudelet in #22
• Changing resolution of train/test/validation indices in datamanager. And make data manager mypy compliant by @ojpb in #19
• Refactor unit tests by @thomasgaudelet in #40
• Update torch requirement from <=1.13.0,>=1.9.0 to >=1.9.0,<1.14.0 by @dependabot in #43
• AL-Pipeline by @paulmorio in #33
• Renaming Classes by @a-pouplin in #51
• Updated examples by @paulmorio in #54
• Refactor RelativeDistanceStrategy by @thomasgaudelet in #59
• Refactor how labelling history is stored by @thomasgaudelet in #58
• Readme update by @paulmorio in #64
• Adding mypy to pre-commit by @thomasgaudelet in #65
• Notebooks updated by @a-pouplin in #63
• Refactor regression informativeness/strategies by @thomasgaudelet in #78
• Change regression_greedy_score name to regression_mean_prediction by @alicerdv in #82
• Change module names and change naming to reflect change from model to model manager by @alicerdv in #79
• Current model reset by @thomasgaudelet in #83
• Rename methods by @a-pouplin in #85
• Add label attr input to data manager by @thomasgaudelet in #86
• Documentation update by @alicerdv in #80
• Updated Readme with new image and deprecated older images by @paulmorio in #88

New Contributors

• @ojpb made their first contribution in #11
• @alicerdv made their first contribution in #15
• @a-pouplin made their first contribution in #51

Full Changelog: v0.1.6...v1.0.0

Benchmark datasets addition

What's Changed

• Datasets for benchmarking strategies by @paulmorio in #8

Full Changelog: v0.1.5...v0.1.6

0.1.5

What's Changed

• Adding hit ratio metric by @thomasgaudelet in #7

Full Changelog: v0.1.4...v0.1.5

v0.1.4

What's Changed

Full Changelog: v0.1.3...v0.1.4

patch 0.1.3

What's Changed

• Populated init.py files to enable python autocomplete and shorten import statements length.

Full Changelog: v0.1.2...v0.1.3

pyrelational release

pyrelational unveiling