public protocol Optimizer

A machine learning optimizer.

Optimizers apply an optimization algorithm to update the differentiable variables of a machine learning model.

  • The type of the model whose parameters are optimized.

    Declaration

    associatedtype Model : Differentiable

  • The scalar parameter type.

    Declaration

    associatedtype Scalar : FloatingPoint

  • The learning rate.

    Declaration

    var learningRate: Scalar { get set }

  • Updates the specified differentiable variables along the specified direction.

    Declaration

    mutating func update(_ variables: inout Model.AllDifferentiableVariables, along direction: Model.CotangentVector)