A causal audit via neuron-row zeroing shows attribution methods (LRP, IG) faithfully identify dispensable neurons and can install refusal behavior, while rank-stability proxies systematically miss selector failures.
Interpretability as alignment: Making internal understanding a design principle.CoRR, abs/2509.08592, 2025
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Faithfulness to Refusal: A Causal Audit of Neuron Selectors
A causal audit via neuron-row zeroing shows attribution methods (LRP, IG) faithfully identify dispensable neurons and can install refusal behavior, while rank-stability proxies systematically miss selector failures.