Model Editing Robustness
Investigating the robustness of model editing and unlearning methods against adversarial perturbations such as tokenization changes.
1 papers
Where this stands
The written synthesis of this thread is for subscribers. Subscribe.