NAG constructor

NAG(
  1. List<Tensor> parameters, {
  2. required double learningRate,
  3. double momentum = 0.9,
})

Creates a NAG optimizer for parameters with the given learningRate and momentum factor.

Implementation

NAG(
  List<Tensor<dynamic>> parameters, {
  required double learningRate,
  this.momentum = 0.9,
}) : super(parameters, learningRate) {
  _v = {};
  for (int p = 0; p < parameters.length; p = p + 1) {
    Tensor<dynamic> param = parameters[p];
    int size = param.data.length;
    List<double> vList = [];
    for (int i = 0; i < size; i = i + 1) {
      vList.add(0.0);
    }
    _v[param.id] = vList;
  }
}