Member Junction
    Preparing search index...

    POST /train request body. MJ assembles the matrix via the FeatureAssembly executor (plan §6) and sends it here. Either TrainRequest.data (inline) or TrainRequest.data_ref (shared-storage handle) is provided.

    interface TrainRequest {
        algorithm: string;
        data?: MatrixData;
        data_ref?: string;
        feature_schema: FeatureSchemaEntry[];
        holdout?: MatrixData;
        hyperparameters: Record<string, unknown>;
        preprocessing: PreprocessingOp[];
        problem_type: ProblemType;
        target: string;
        validation: ValidationConfig;
    }
    Index

    Properties

    algorithm: string

    Sidecar algorithm driver key (e.g. xgboost, lightgbm, logistic_regression).

    data?: MatrixData

    Inline matrix data (mutually exclusive with data_ref).

    data_ref?: string

    Shared-storage handle to the matrix (Parquet/Arrow), used for very large sets.

    feature_schema: FeatureSchemaEntry[]

    Ordered feature schema describing each input column.

    holdout?: MatrixData

    The locked holdout matrix (plan §8.2), carved off the assembled data by the orchestrator before any train/test split and never present in TrainRequest.data. Same columns as data (it includes the target column). When supplied, the sidecar scores these rows exactly once using the preprocessing fitted on the training data (frozen fitted transform, never re-fit — the anti-skew guarantee, plan §6.2) and returns the result as TrainResponse.holdout_metrics.

    This is the honest, deterministic holdout: the orchestrator carves the exact rows in TypeScript (auditable) and forwards them here, rather than asking the sidecar to re-derive a holdout via ValidationConfig.holdout_size. When both holdout and validation.holdout_size are set, holdout wins.

    hyperparameters: Record<string, unknown>

    Algorithm-specific hyperparameters (validated against the catalog schema).

    preprocessing: PreprocessingOp[]

    Ordered preprocessing ops to fit + apply at train time.

    problem_type: ProblemType

    Classification or regression.

    target: string

    Target/label column name.

    validation: ValidationConfig

    How to validate during training.