Releases: fidelity/mabwiser
Releases · fidelity/mabwiser
mabwiser 2.7.3
minor:
- Changed return type of functions annotated with NoReturn to None by @LiHRaM
mabwiser 2.7.2
mabwiser 2.7.1
- Implemented LearningPolicyType and NeighborhoodPolicyType to simplify input for MAB.
- Updated tests to accommodate LearningPolicyType and NeighborhoodPolicyType.
mabwiser 2.7.0
- Implement vectorized functions for Linear policies to speed-up prediction for multiple inputs.
- Update tests for LinTS and LinGreedy Policies as a result of using global random seed in vectorized function
mabwiser 2.6.0
- Implement tracking of warm started and partial fitted arms
- Implement tracking of cold arms
- Limit warm start to learning policies
mabwiser 2.5.0
- Update warm start logic to only warm start an arm once
- Implement fix for fitting scalers in Linear policies when variance is too small
mabwiser 2.4.1
- Bug fixes in examples
- Validate tree parameters of TreeBandit to be compatible with sklearn.tree.DecisionTreeRegressor
mabwiser 2.4.0
- Implement vectorized functions for non-contextual policies to speed-up prediction for multiple decisions.
- Change MAB predict and predict_expectations to allow empty contexts to be specified for non-contextual policies.
- Update scaler use in Linear policies so that standard scaler can be fit directly instead of pre-trained scalers.
- Change scaler argument from pre-trained
arm_to_scaler
input to a boolean scale flag.
mabwiser 2.3.0
- New Algorithm: LinGreedy as a learning policy.
mabwiser 2.2.0
- Modified
predict_expectations
, such thatpredict
can usepredict_expectations
directly in all non-contextual learning policies.