method
active
method:counterfactual-f-attribution

Counterfactual Δf attribution

Computes feature attribution passing ablated pre-activation through JumpReLU to handle inhibition.

Neighborhood — ranked by edge-count

Methods (1)

method
  • Second-order KL estimate of a weight's effect on outputs using Fisher metric.

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

  • Ability to entertain competing hypotheses within one inference engine; proposed hallmark of mindful inference
  • Counterfactualconcept0.803
    The output value a model produces when an interchange intervention forces certain variables to take values from source inputs.
  • The behavior that would have occurred had the value of a causal variable been different while everything else remained the same; used as training labels in DAS/MAS.
  • Data Attributionconcept0.782
    The task of attributing model behaviors to specific training datapoints.
  • The state a neural network is placed in when its activations are modified via intervention
  • True/false claims that a prompt edit shifts a behavior rate by at least 30pp; basis for all evaluation labels.
  • Encoding possible future or alternative states; bioelectric patterns can represent an alternative morphological target even in an intact animal.