Metric Interface

The Metric Interface provides functionality for implementing evaluation metrics.

Overview

Metrics are used to evaluate model performance. The MetricTemplate class supports:

Custom metric definitions
State management across batches
Reset and update operations
Integration with training loops

Core MetricTemplate Class

File Location

Cython Implementation: src/AnalysisG/core/metric_template.pyx
Cython Header: src/AnalysisG/core/metric_template.pxd

Methods to Override

compute(): Calculate the metric value.

update(): Update metric state with new predictions.

reset(): Reset metric state.

See Also

../core/metric_template: Core MetricTemplate implementation
../metrics/overview: Concrete metric implementations