Transformer circuits show free evolution during SFT, rendering static mechanistic localization inadequate for future parameter updates due to inherent temporal latency.
Negative preference optimization: From catastrophic collapse to effective unlearning
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
ROSU derives a closed-form retain-neutral perturbation for min-max unlearning that bounds retain damage via curvature and improves performance when gradients are aligned.
citing papers explorer
-
Navigating by Old Maps: The Pitfalls of Static Mechanistic Localization in LLM Post-Training
Transformer circuits show free evolution during SFT, rendering static mechanistic localization inadequate for future parameter updates due to inherent temporal latency.
-
Retain-Neutral Surrogates for Min-Max Unlearning
ROSU derives a closed-form retain-neutral perturbation for min-max unlearning that bounds retain damage via curvature and improves performance when gradients are aligned.