agrag.ingestion.resolve.batch_validation.validate_batch_verdicts

validate_batch_verdicts(pair_ids:Iterable[str], results:Iterable[object]) -> dict[str, ComparisonResult]

Return fail-safe verdicts keyed by requested candidate pair identifiers.

Unknown, duplicate, missing, and malformed results resolve to NO_MATCH. This avoids assigning a valid LLM response to a different candidate pair.