Member Junction
    Preparing search index...

    Interface DuplicateDecisionResult

    The outcome of one decision call over a matched set.

    interface DuplicateDecisionResult {
        AIPromptRunID: string;
        Candidates: DuplicateCandidateProbability[];
        ErrorMessage?: string;
        Success: boolean;
        UncalibratedModel?: string;
    }
    Index

    Properties

    AIPromptRunID Candidates ErrorMessage? Success UncalibratedModel?

    Properties

    AIPromptRunID: string

    The decision's MJ: AI Prompt Runs row id, when the runner wrote one.

    Candidates: DuplicateCandidateProbability[]

    One entry per input candidate, in input order. Empty when the call failed.

    ErrorMessage?: string

    Why the call failed, when Success is false.

    Success: boolean

    Whether the decision call succeeded.

    UncalibratedModel?: string

    Set when the exact model that answered has no calibration: the model, as DescribeAnsweringModel names it (Jev (typesafe/jev-1.14-...)). Every candidate's Probability is then null and its RawProbability holds the model's own answer. The entry check treats such a result like a failed decision and flags nothing.

    Settings

    Member Visibility

    On This Page

    Properties