Metrics

Performance metrics for model evaluation and monitoring during training. Includes accuracy, precision, recall, F1 score, confusion matrices, and running statistics (average, Welford variance). MetricState provides state containers, while MultiMetric enables tracking multiple metrics simultaneously.

MetricState	Wrapper class for Metric Variables.
Metric	Base class for metrics.
AverageMetric	Average metric for computing running mean of values.
WelfordMetric	Welford's algorithm for computing mean and variance of streaming data.
AccuracyMetric	Accuracy metric for classification tasks.
MultiMetric	Container for multiple metrics updated simultaneously.
PrecisionMetric	Precision metric for binary and multi-class classification.
RecallMetric	Recall (sensitivity) metric for binary and multi-class classification.
F1ScoreMetric	F1 score metric for binary and multi-class classification.
ConfusionMatrix	Confusion matrix metric for multi-class classification.