method
active
method:causal-importance-functionCausal importance function
Small per-subcomponent MLP with hard sigmoid predicting causal importance.
Neighborhood — ranked by edge-count
Papers (1)
paper
- Stochastic Parameter Decompositionintroducesmentions
Frameworks (1)
framework
- Linear parameter decomposition with stochastic ablation.
Concepts (1)
concept
- Causal importanceimplementsA measure of whether a subcomponent is necessary to reproduce model behavior on a specific prompt, predicted by the causal importance network.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Auxiliary model trained alongside VPD to predict which subcomponents are causally important for each prompt, enabling mechanistic isolation of components.
- Causal importance is related but not identical to causal dependence because of self-repairclaim0.787Related work distinction.
- Weighted Spearman correlation that corrects for sampling bias in automated interpretability evaluation
- Prior state-of-the-art baseline using the abliterated model as proposal distribution.
- The ability of an agent to be a driver of subsequent events; a hallmark of cognition that causal emergence quantifies.
- Probability that effect would not occur without cause; counterfactual primitive.
- Function determining the value of a variable based on its causal parents in an acyclic causal model.
- A framework the paper uses alongside feature geometry to deepen mechanistic understanding of LMs