What should you do if you want to perform a causal analysis of your DNN?

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

• What can causal abstraction analyses tell us about how DNNs encode features if the methods themselves rely on encoding assumptions?question0.768
  Circular dependency problem raised in discussion
• How do we establish bidirectional causal relationships between neural systems?question0.754
  Motivates the bidirectional design of MAS over unidirectional model stitching.
• Causal abstraction implicitly relies on strong assumptions about feature encoding in DNNs, and becomes trivial without such assumptionsclaim0.749
  Authors' interpretation connecting their proof to practical interpretability methodology
• causal abstraction implicitly relies on strong assumptions about how features are encoded in deep neural networks (DNNs), and becomes trivial without such assumptionsquote0.743
  Load-bearing formulation of the paper's central argument
• Causal abstraction analysismethod0.733
  The formal method used to establish that the identified circuit causally mediates the model's cyclic reasoning behavior
• Investigating the causal substructure of neural representations is necessary to avoid misidentifying data structures of simpler representations as abstract conceptsclaim0.732
  Motivated by the finding that lexical entailment decomposes into word identities.
• Causal Tracingconcept0.731
  Mechanistic interpretability technique for locating factual associations, mentioned as future work direction.
• Ultimately, we would like to understand neural networks well enough to be able to intentionally design them.quote0.722
  Vision statement in the conclusion.