function

policy

function

layer

source

source

source

AplusBXtoN Tanh

DifferentiableField

DifferentiableField TrainableState

DifferentiableField AplusBXtoN Bias CrossEntropy CrossEntropySoftmax CrossEntropySoftmaxes Downsample FullyConnected MaxPooling ReLU RightCrop RightPad Softmax SpatialConvolution StackFeatures Subspace SumOfSquares Tanh TrainableState ZeroPad ByteDataSource ByteLabelSource RandomSource

RandomSource

SpatialConvolution