public struct BasicRNNCell<Scalar> : RecurrentLayerCell where Scalar : TensorFlowFloatingPointA basic RNN cell.
-
Declaration
public var weight: Tensor<Scalar> -
Declaration
public var bias: Tensor<Scalar> -
Declaration
public typealias State = Tensor<Scalar> -
Declaration
public typealias TimeStepInput = Tensor<Scalar> -
Declaration
public typealias TimeStepOutput = State -
Declaration
public typealias Input = RNNCellInput<TimeStepInput, State> -
Declaration
public typealias Output = RNNCellOutput<TimeStepOutput, State> -
Creates a
SimpleRNNCellwith the specified input size and hidden state size.Declaration
public init(inputSize: Int, hiddenSize: Int, seed: TensorFlowSeed = Context.local.randomSeed)Parameters
inputSizeThe number of features in 2-D input tensors.
hiddenSizeThe number of features in 2-D hidden states.
seedThe random seed for initialization. The default value is random.