mab2rec 1.3.1

• Updated requirements to use mabwiser>=2.7.4 to reflect change from np.Inf to np.inf in mabwiser.
• Fixed default KMeans n_init parameters in tests instead of using 'auto' used in scikit-learn>=1.4

mab2rec 1.3.0

Major:

• Added optional apply_sigmoid argument to recommend() method, to
  control whether sigmoid transformation is applied to scores or not.
  Minor:
• Fixed bug when recommending single context.

mab2rec 1.2.1

• Replaced NoReturn type hinting with None - thank you @SaraEkmanSVT
• Release extended data

mab2rec 1.2.0

• Removed spock-config dependency and train/test scripts using Spock
• Updated tests and documentation to reflect Python 3.8+ support

mab2rec 1.1.0

Major:

• Updated requirements to use mabwiser>=2.7 to benefit from enhancements,
  including vectorized predict for Linear policies and tracking of arm status.
• Fixed tests due to changes in random seeding for Linear policies.
  Minor:
• Added Diversity metrics to available MAB evaluation metrics.

mab2rec 1.0.3

Minor:

• Fixed bug with inconsistency between scored and eligible items.

mab2rec 1.0.2

Minor:

• Fixed bug with arguments not being passed correctly to load_response_data.

mab2rec 1.0.1

Minor:

• Fix missing top-k recommendations when top messages are excluded - thanks @nateewall!

mab2rec 1.0.0

• Initial public release.