truelearn.models.LearnerMetaWeights

class truelearn.models.LearnerMetaWeights(novelty_weights: Weights = Weights(mean=0.5, variance=0.5), interest_weights: Weights = Weights(mean=0.5, variance=0.5), bias_weights: Weights = Weights(mean=0.0, variance=0.5))

Bases: object

Store the weights used in meta training.

Examples

>>> from truelearn.models import LearnerMetaWeights
>>> # construct an empty learner meta model
>>> LearnerMetaWeights()  
LearnerMetaWeights(novelty_weights=Weights(mean=0.5, variance=0.5)...)
>>> # construct a learner meta model with custom weights
>>> bias_weights = LearnerMetaWeights.Weights(mean=1.0, variance=2.0)
>>> LearnerMetaWeights(bias_weights=bias_weights)
LearnerMetaWeights(...bias_weights=Weights(mean=1.0, variance=2.0))

Attributes

bias_weights: Weights = Weights(mean=0.0, variance=0.5)

A dict that stores the “mean” and “variance” of a bias variable.

interest_weights: Weights = Weights(mean=0.5, variance=0.5)

“A dict that stores the “mean” and “variance” of the learner’s interest weights.

novelty_weights: Weights = Weights(mean=0.5, variance=0.5)

A dict that stores the “mean” and “variance” of the learner’s knowledge/novelty weights.