Member Junction
    Preparing search index...

    The result of training one experiment — the model produced, the training-run leaf (linked to the iteration), the normalized leaderboard score, and the compute/token cost attributed to the attempt (for budget enforcement, §8.4).

    interface TrainExperimentResult {
        ComputeCost?: number;
        ComputeSeconds?: number;
        model: MJMLModelEntity;
        run: MJMLTrainingRunEntity;
        Score: number;
        ScoreMetric: string;
        TokensUsed?: number;
    }
    Index

    Properties

    ComputeCost?: number

    Compute cost attributed to this attempt, in USD (defaults to 0 when unknown).

    ComputeSeconds?: number

    Compute seconds the attempt spent (defaults to 0 when unknown); the session's compute-seconds bound sums these.

    The immutable MJ: ML Models row produced (Draft); null when the run's gate decision shipped no artifact.

    The MJ: ML Training Runs leaf recording the attempt (linked to the iteration).

    Score: number

    Normalized holdout score (higher is better) — drives the leaderboard: on the plan's SuccessMetric, or on the model's primary proper score when its holdout gave no score on the success metric; null when neither gives one (or no model shipped).

    ScoreMetric: string

    The holdout metric Score was read from (ExperimentScore.Metric); null exactly when Score is null.

    TokensUsed?: number

    LLM tokens used by this attempt (defaults to 0 when unknown).