agrag.eval.extraction.entity_quality_metric
entity_quality_metric(*, threshold:float = 0.0) -> ScoreMetric
Build a metric for entity F1 on one test case.
The case score is the exact F1. Use a new metric for each case, and pass the
measured metrics to micro_scores. The default threshold is 0 because the
gate belongs on the dataset score.
Parameters:
- threshold (
float) – The minimum case score that counts as success.
Returns:
ScoreMetric– A metric that scores exact entity F1 for one case.