Gintare Karolina Dziugaite

Unlearning with asymmetric sources: improved unlearning-utility trade-off with public data

Less is More: Undertraining Experts Improves Model Upcycling

REPO: Detoxifying LLMs via Representation Erasure-based Preference Optimization

On Traceability in $\ell_p$ Stochastic Convex Optimization

From Dormant to Deleted: Tamper-Resistant Unlearning Through Weight-Space Regularization

The Journey Matters: Average Parameter Count over Pre-training Unifies Sparse and Dense Scaling Laws

Selective Unlearning via Representation Erasure Using Domain Adversarial Training

Mechanistic Unlearning: Robust Knowledge Unlearning and Editing via Mechanistic Localization

Mechanistic Unlearning: Robust Knowledge Unlearning and Editing via Mechanistic Localization

Leveraging Per-Instance Privacy for Machine Unlearning

Improved Localized Machine Unlearning Through the Lens of Memorization

Torque-Aware Momentum

Soup to go: mitigating forgetting during continual learning with model averaging