ScoringResult
The combined result of running scorers against a single example.
Returned by Evaluation.run(). Inspect scorers_data for per-scorer
values, errors, and metadata.
results = evaluation.run(
examples=examples,
scorers=["faithfulness", "answer_relevancy"],
eval_run_name="nightly",
)
for result in results:
for scorer in result.scorers_data:
print(f"{scorer.name}: {scorer.value}")Attributes
PropType
scorers_data?List[ScorerData]
Per-scorer results (see ScorerData).
Type
List[ScorerData]data_object?Union[TraceSpan, Example]
The Example or TraceSpan that was scored.
Type
Union[TraceSpan, Example]name?Optional[str]
The evaluation run name.
Type
Optional[str]Default
Nonetrace_id?Optional[str]
Associated trace ID, if applicable.
Type
Optional[str]Default
Nonerun_duration?Optional[float]
How long the evaluation took (seconds).
Type
Optional[float]Default
Noneevaluation_cost?Optional[float]
Total cost in USD.
Type
Optional[float]Default
Noneto_dict()
def to_dict() -> APIScoringResult:Returns
APIScoringResult