vraven.darkside · function
analyse_circuit_multiplicity
analyse_circuit_multiplicity(circuit_masks: 'Mapping[str, object] | object', *, component_labels: 'Sequence[str] | None' = None, confidence: 'float' = 0.95, stable_threshold: 'float' = 0.8) -> 'AnalysisResult'Measure non-uniqueness and empirical stability across discovered circuits.
Parameters
----------
circuit_masks:
Binary matrix with one discovered circuit per row and one aligned model
component per column, or a mapping from run name to one binary mask.
component_labels:
Optional labels in exactly the same component order as the masks.
confidence:
Confidence level for Wilson intervals over inclusion frequency.
stable_threshold:
A component belongs to the empirical stable core only when the lower
Wilson bound is at least this threshold.
Notes
-----
This is an empirical repeated-discovery audit. It is not a formal
verification certificate over a continuous input domain or dataset edit
neighbourhood.
Usage
from vraven.darkside import analyse_circuit_multiplicity
result = analyse_circuit_multiplicity(... )Source of truthThis page is generated from the checked-out public symbol, signature, and docstring. Pin the package version for reproducible work.