The single scoring-execution seam this action depends on — "score this scope
with this model, give me the counts + predictions / write-back confirmation".
The production implementation (ProductionScoreRecordSetRunner) resolves
the scope into records and drives the MLModelInferenceProcessor; tests
inject a deterministic fake.
The single scoring-execution seam this action depends on — "score this scope with this model, give me the counts + predictions / write-back confirmation". The production implementation (ProductionScoreRecordSetRunner) resolves the scope into records and drives the MLModelInferenceProcessor; tests inject a deterministic fake.