VPD achieves a better sparsity-reconstruction tradeoff than transcoders.

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

• VPD achieves better sparsity-reconstruction tradeoff than transcoders on 67M modelfinding0.946
  Empirical result demonstrating VPD's efficiency advantage in parameter decomposition.
• VPD subcomponents are sparse, interpretable, and avoid feature splitting.claim0.790
  Assertion about the qualitative advantages of VPD's rank-one decomposition.
• VPD is a meaningful step toward bottom-up interpretabilityclaim0.758
  Positioning of VPD as advancing the paradigm of explaining computation in the model's terms.
• VPD decomposes parameters, not activations, flipping the standard SAE / activation-patching paradigm.claim0.756
  Core proposition of the paper: a substrate-level critique of existing interpretability methods.
• The VPD-based edit has similarly low off-target effects as uninterpretable fine-tuning methodsfinding0.755
  Performance comparison showing subcomponent editing is comparable to fine-tuning in preserving off-target behavior.
• VPD enables manual model editing through direct parameter manipulation.claim0.746
  Applied capability claim: VPD enables surgical changes to model behaviour at the parameter level.
• probably helps not only with faithful reconstruction but also creates interference patterns that encode nuanced information about the deltas and convergences between states.quote0.746
  Key quote connecting path redundancy to interferometric information encoding.
• The ability to make precise edits demonstrates that VPD identifies real computational machineryclaim0.745
  Claim that editing success validates VPD's decomposition.