Skip to main content
A SUCCESS item does not imply that every score succeeded: evaluator errors are recorded per score (see ScoreStatus), not on the item. Values:
  • SUCCESS - The task function returned outputs and all evaluators ran on this item, even if some individual scores failed.
  • FAILED - The task function raised an exception on this item, so no evaluators ran; error_reason and error_message are populated.
  • SKIPPED - The item was skipped and produced no result.