nirs4all.controllers.data.rep_fusion module

Controller for the rep_fusion relational materialisation step.

class nirs4all.controllers.data.rep_fusion.RepFusionController[source]

Bases: OperatorController

Materialise raw multi-source staging via a replayable RepresentationPlan.

execute(step_info: ParsedStep, dataset: Any, context: ExecutionContext, runtime_context: RuntimeContext, source: int = -1, mode: str = 'train', loaded_binaries: list[tuple[str, Any]] | None = None, prediction_store: Any | None = None) → tuple[ExecutionContext, StepOutput][source]: Execute rep_fusion on a RawMultiSourceDataset.

classmethod matches(step: Any, operator: Any, keyword: str) → bool[source]: Return True for the rep_fusion pipeline keyword.

priority: int = 3

classmethod supports_prediction_mode() → bool[source]: The same representation plan is replayable in prediction mode.

classmethod use_multi_source() → bool[source]: The step consumes the whole relation-aware staging object.