latent.flows.conversation_scoring_flow

conversation_scoring_flow — Turn-level scoring with aggregation (FR-3.5).

Functions

conversation_scoring_flow

conversation_scoring_flow(conversations: list, judge: Any, aggregation: str = 'mean', gates: dict[str, float | GateSpec] | None = None, n_resamples: int = 10000, confidence_level: float = 0.95, seed: int | None = None) -> dict[str, Any]

Score conversations at the turn level, aggregate, and analyze.

Args: conversations: List of Conversation objects. judge: A Judge[T] instance for scoring turns. aggregation: Strategy for aggregating turn scores ('mean', 'min', 'max', 'last'). gates: Optional quality gates. n_resamples: Bootstrap resamples. confidence_level: CI confidence level. seed: Random seed.

Returns: Dict with: scored_conversations, aggregated_scores, report, markdown.