/// <summary>
/// Writes a trainer parameters to event log.
/// </summary>
/// <param name="logFile">The log file to write to.</param>
/// <param name="trainer">The trainer which parameters to write.</param>
/// <param name="algorithm">The training algorithm.</param>
/// <param name="loss">The loss function.</param>
private void WriteTrainerParameters(
StreamWriter logFile,
ClassificationNetworkTrainer trainer,
ITrainingAlgorithm algorithm,
ILoss <int[]> loss)
{
this.WriteLine(logFile, "Trainer parameters:");
this.WriteLine(logFile, " Batch Size: {0}", trainer.BatchSize);
this.WriteLine(logFile, " L1 Rate: {0}", trainer.RateL1);
this.WriteLine(logFile, " L2 Rate: {0}", trainer.RateL2);
this.WriteLine(logFile, " Clip Value: {0}", trainer.ClipValue);
this.WriteLine(logFile, "Algorithm parameters:");
this.WriteLine(logFile, " Algorithm: {0}", algorithm.GetType().Name);
if (algorithm is Adadelta adadelta)
{
this.WriteLine(logFile, " Learning Rate: {0}", adadelta.LearningRate);
this.WriteLine(logFile, " Decay: {0}", adadelta.Decay);
this.WriteLine(logFile, " Rho: {0}", adadelta.Rho);
this.WriteLine(logFile, " Eps: {0}", adadelta.Eps);
}
if (algorithm is Adagrad adagrad)
{
this.WriteLine(logFile, " Learning Rate: {0}", adagrad.LearningRate);
this.WriteLine(logFile, " Eps: {0}", adagrad.Eps);
}
if (algorithm is Adam adam)
{
this.WriteLine(logFile, " Learning Rate: {0}", adam.LearningRate);
this.WriteLine(logFile, " Beta1: {0}", adam.Beta1);
this.WriteLine(logFile, " Beta2: {0}", adam.Beta2);
this.WriteLine(logFile, " Eps: {0}", adam.Eps);
}
if (algorithm is RMSProp rmsProp)
{
this.WriteLine(logFile, " Learning Rate: {0}", rmsProp.LearningRate);
this.WriteLine(logFile, " Rho: {0}", rmsProp.Rho);
this.WriteLine(logFile, " Eps: {0}", rmsProp.Eps);
}
if (algorithm is SGD sgd)
{
this.WriteLine(logFile, " Learning Rate: {0}", sgd.LearningRate);
this.WriteLine(logFile, " Decay: {0}", sgd.Decay);
this.WriteLine(logFile, " Momentum: {0}", sgd.Momentum);
this.WriteLine(logFile, " Nesterov: {0}", sgd.Nesterov);
}
this.WriteLine(logFile, "Loss parameters:");
this.WriteLine(logFile, " Loss: {0}", loss.GetType().Name);
if (loss is LogLikelihoodLoss logLikelihoodLoss)
{
this.WriteLine(logFile, " LSR: {0}", logLikelihoodLoss.LSR);
}
}
private ClassificationNetworkTrainer CreateTrainer()
{
ClassificationNetworkTrainer trainer = new ClassificationNetworkTrainer();
JsonSerializer jsonSerializer = new JsonSerializer();
using (JTokenReader jtokenReader = new JTokenReader(this.TaskParameters.TrainerParameters))
{
jsonSerializer.Populate(jtokenReader, trainer);
}
return(trainer);
}
public void ForwardVolumes()
{
Random random = new Random(0);
ClassificationNetworkTrainer trainer = new ClassificationNetworkTrainer();
SGD sgd = new SGD()
{
LearningRate = 0.0001f,
Momentum = 0.0f
};
ClassificationNetwork network = ClassificationNetwork.FromArchitecture("1x1x2~5N~5N~3N", this.classes);
// lets test 100 random point and label settings
// note that this should work since l2 and l1 regularization are off
// an issue is that if step size is too high, this could technically fail...
for (int k = 0; k < 100; k++)
{
int gti = (int)Math.Floor(random.NextDouble() * 3);
Tensor x = new Tensor(null, new Shape(Shape.BWHC, 1, 1, 1, 2));
x.Set(new float[] { ((float)random.NextDouble() * 2) - 1, ((float)random.NextDouble() * 2) - 1 });
Tensor pv = network.Forward(null, x).Clone() as Tensor;
trainer.RunEpoch(
k,
network,
Enumerable.Repeat((x, new string[] { this.classes[gti] }), 1),
sgd,
new LogLikelihoodLoss(),
CancellationToken.None);
Tensor pv2 = network.Forward(null, x).Clone() as Tensor;
Assert.IsTrue(pv2.Weights[gti] > pv.Weights[gti], "k: {0}, gti: {1}, pv2[gti]: {2}, pv[gti]: {3}", k, gti, pv2.Weights[gti], pv.Weights[gti]);
}
}