API Reference - Others - RandomNetworkDistillation

RandomNetworkDistillation is a neural network for producing internal rewards to encourage exploration. Requires neural network as your model.

Constructors

new()

Create new model object. If any of the arguments are nil, default argument values for that argument will be used.

RandomNetworkDistillation.new(): RandomNetworkDistillationObject
  • RandomNetworkDistillationObject: The generated RandomNetworkDistillation object.

Functions

generate() 

RandomNetworkDistillation:generate(featureMatrix: matrix): matrix

Parameters:

  • featureMatrix: The matrix containing all the features.

Returns:

  • outputMatrix: The matrix generated by the model from the given feature matrix.

setModel() 

RandomNetworkDistillation:setModel(Model: ModelObject)

Parameters

  • Model: The model to be used by the RandomNetworkDistillation object.

getModel() 

RandomNetworkDistillation:setModel(): ModelObject

Returns

  • Model: The model that is used by the RandomNetworkDistillation object.

getTargetModelParameters()

Gets the target model parameters from the network.

RandomNetworkDistillation:getTargetModelParameters(doNotDeepCopy: boolean): ModelParameters

Parameters

  • doNotDeepCopy: Set whether or not to deep copy the model parameters.

Returns

  • TargetModelParameters: Target network model parameters to be used for predictor network training.

getPredictorModelParameters()

Gets the predictor model parameters from the network.

RandomNetworkDistillation:getPredictorModelParameters(doNotDeepCopy: boolean): ModelParameters

Parameters

  • doNotDeepCopy: Set whether or not to deep copy the model parameters.

Returns

  • PredictorModelParameters: Target network model parameters to be used for predictor network training.

setTargetModelParameters()

Set the target model parameters to the network

RandomNetworkDistillation:setTargetModelParameters(TargetModelParameters: ModelParameters, doNotDeepCopy: boolean)

Parameters

  • TargetModelParameters: Target network model parameters to be used for predictor network training.

  • doNotDeepCopy: Set whether or not to deep copy the model parameters.

setPredictorModelParameters()

Set the predictor model parameters to the network

RandomNetworkDistillation:setPredictorModelParameters(PredictorModelParameters: ModelParameters, doNotDeepCopy: boolean)

Parameters

  • PredictorModelParameters: Predictor network model parameters to be trained so that it tries to match up with target network model parameters.

  • doNotDeepCopy: Set whether or not to deep copy the model parameters.

References