A clean error message when the build failed; else null.
OptionalGateThe training run's stored gate decision; absent when the build failed before a run stored one.
When not published, the plain reason (leakage, a calibration hold, the trust verdict, auto-publish off, or the promotion gate's refusal); else null.
OptionalleaderboardThe leaderboard of the build's one training run.
Whether training flagged a leakage warning (held for analyst review).
OptionalmodelThe trained model entity (present on successful build).
OptionalmodelThe trained MJ: ML Models id (present once training completes).
OptionalpipelineThe created pipeline entity (present once pipeline is created).
OptionalpipelineThe created MJ: ML Training Pipelines id (present once the pipeline is created).
Whether the model was published into the business catalog.
True when the model was not published only because auto-publish was off: it shipped under its protocol, training flagged no leakage and it holds no calibration hold, so the promotion gate may be asked to publish it.
Whether the build (pipeline create + train) succeeded.
OptionaltrustThe plain-language trust verdict: the training run's stored gate decision rendered as copy.
OptionalwarningsStructured warnings emitted during plan translation or training (e.g. dropped candidate features).
The outcome of building a prediction from a plan.