Manual model editing through parameter manipulation

Application enabled by VPD: direct manipulation of weight matrices for interpretable model modification.

Neighborhood — ranked by edge-count

concept

Manual model editing
related_tosame_as
Ability to surgically alter model behavior through direct parameter changes rather than activation interventions.
Model Editing
related_to
Technique for modifying model knowledge or behavior via targeted interventions, e.g., ROME by Meng et al.
VPD (adVersarial Parameter Decomposition)
about
Core methodological framework introduced in this paper; decomposes weight matrices into rank-one interpretable subcomponents using adversarial ablations.

cosine ≥ 0.65 · no typed edge

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

Parameter manipulationconcept0.831
Direct editing of model parameters, enabled by VPD's decomposition, for manual model editing.
VPD enables manual model editing through direct parameter manipulation.claim0.816
Applied capability claim: VPD enables surgical changes to model behaviour at the parameter level.
Model changeconcept0.786
When a single conversation involves different models, e.g., routing between standard and reasoning models
Model editing via direct subcomponent overwritemethod0.763
Technique to alter model behavior by directly editing a parameter subcomponent without training, demonstrated by changing an emoticon eye subcomponent.
model selectionconcept0.742
Comparing models using log-evidence approximated by free energy.
Model Steeringconcept0.740
Using interventions to guide model generation behavior, e.g., adding sentiment vectors at inference time
Language Modelsconcept0.739
Primary substrate for manifold steering experiments; demonstrates method on reasoning and in-context tasks.
How Do Systems Modify The Parameters Of Theirquestion0.738