vraven.decompiler · function
ground_concepts_from_input_groups
ground_concepts_from_input_groups(atlas: 'ConceptAtlas', inputs: 'np.ndarray | Tensor', groups: 'Mapping[str, Sequence[int]]', *, permutations: 'int' = 200, seed: 'int' = 17, minimum_absolute_association: 'float' = 0.25, promote_labels: 'bool' = False, validated_labels: 'Sequence[str] | None' = None, validation_alpha: 'float' = 0.05) -> 'ConceptAtlas'Attach conservative input-association evidence to discovered concepts.
This function deliberately calls the result an association. ``promote_labels`` is
opt-in and still records that the label has not been independently validated.
``validated_labels`` is intended only for externally defined concepts whose semantic
identity is known independently, such as planted benchmark concepts or a reviewed
domain ontology. It must not be used to upgrade a convenient post-hoc name.
Usage
from vraven.decompiler import ground_concepts_from_input_groups
result = ground_concepts_from_input_groups(... )Source of truthThis page is generated from the checked-out public symbol, signature, and docstring. Pin the package version for reproducible work.