concept
active
concept:causal-tracingCausal Tracing
Mechanistic interpretability technique for locating factual associations, mentioned as future work direction.
Neighborhood — ranked by edge-count
Papers (2)
paper
Methods (1)
method
- Activation patchingextendsStandard method in mechanistic interpretability that intervenes on activations; VPD flips this paradigm by patching parameters.
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.
- The structural-realist grounding for self-evidencing after the bounded self is relinquished.
- A framework the paper uses alongside feature geometry to deepen mechanistic understanding of LMs
- The ability of an agent to be a driver of subsequent events; a hallmark of cognition that causal emergence quantifies.
- Method by Chan et al. 2022 for rigorously testing interpretability hypotheses via interventions
- Chvykov and Hoel's geometric extension of causal emergence to continuous systems using Fisher information.
- Whether an internal direction causally controls a target behavior, verified by intervention success
- Framework informing path-specific objectives by identifying causal chains leading to risky behaviors
- Function determining the value of a variable based on its causal parents in an acyclic causal model.