unlearning
- EraseFlow: Learning Concept Erasure Policies via GFlowNet-Driven Alignment
- Exploring and Leveraging Class Vectors for Classifier Editing
- LLM Unlearning via Neural Activation Redirection
- ModHiFi: Identifying High Fidelity predictive components for Model Modification
- On the creation of narrow AI: hierarchy and nonlocality of neural network skills
- Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning