claim
active
claim:the-phase-diagram-explains-why-neurons-are-sometimes-monosemantic-and-sometimes-polysemanticThe phase diagram explains why neurons are sometimes monosemantic and sometimes polysemantic
Theory contribution.
Neighborhood — ranked by edge-count
Papers (1)
paper
- Toy Models of Superpositionmentions
Findings (2)
finding
- Both empirical and theoretical phase diagrams.
- Neuron-level phase change.
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.
- Precise characterization of why polysemanticity poses a combinatorial obstacle to circuit analysis
- We hypothesize that polysemantic neurons may be resolvable by unfolding networks or training to avoid polysemanticity.hypothesis0.809Forward-looking proposal for how the polysemanticity challenge to circuits research might be overcome
- Identified gap linking polysemanticity challenge to disentangled representations literature
- Neurons responding to a single feature.
- Interpretation of the cars-in-superposition circuit finding as an intentional representational strategy
- Counter-example disproving that architectural sparsity alone can prevent polysemanticity
- Fundamental theoretical claim motivating DAS, attributed to Smolensky/Rumelhart/McClelland.
- Theoretical point.