Orchestrate model training

A Machine Learning Engineer must log the numeric validation dictionary to an active MLflow run with one dictionary-metric logging call, rather than separate per-key metric calls. The tracking server is configured. Which replacement implements the missing line?

import mlflow
metrics = {"f1_score": 0.89, "precision": 0.91, "recall": 0.87}
with mlflow.start_run():
    # Insert one dictionary-metric logging call here
    pass