trulens.core.utils.evaluator¶

trulens.core.utils.evaluator ¶

enqueue_conversation(
    conversation_id: str, record_ids: List[str]
) -> None

Queue one context-bounded conversation evaluation.

get_last_error() -> Optional[BaseException]

Return the most recent exception observed by the evaluator thread, or None if no error has occurred.

start_evaluator() -> None

Start the evaluator for the app.

stop_evaluator() -> None

Stop the evaluator for the app.

This is only supported for OTEL Tracing.

compute_now(
    record_ids: Optional[List[str]],
    lock: Optional[Lock] = None,
) -> None

Trigger immediate computation.

PARAMETER	DESCRIPTION
`record_ids`	Optional list of record ids to compute feedbacks for. If None, computes feedbacks for all unprocessed records. TYPE: `Optional[List[str]]`
`lock`	Optional lock to use for the computation. If None, will use the default lock. TYPE: `Optional[Lock]` DEFAULT: `None`

compute_conversation_now(record_ids: List[str]) -> None

Compute conversation metrics for one exact recording batch.