SelfSupervisedLearningTrainingHistory<T>

Models & Types · AiDotNet.SelfSupervisedLearning

Training history from SSL pretraining.

Constructors

Constructor Summary
SelfSupervisedLearningTrainingHistory()

Properties

Property Summary
Dictionary<string, List<T>> CustomMetrics { get; set; } Custom metrics per epoch.
List<T> LossHistory { get; set; } Loss values per epoch.
List<T> StdHistory { get; set; } Representation standard deviation per epoch (for collapse detection).
List<double> KNNHistory { get; set; } k-NN accuracy per epoch (if computed).
List<double> LearningRateHistory { get; set; } Learning rate per epoch.
List<double> MomentumHistory { get; set; } Momentum value per epoch (for methods with momentum encoder).

Methods

Method Summary
void AddCustomMetric(string name, T value) Adds a custom metric value.
void AddEpochMetrics(T loss, T std, double knnAcc, double lr, double momentum) Adds metrics from a training step.