← All threads

Model Unlearning

Methods for selectively removing dangerous or unwanted knowledge from pretrained models while preserving overall performance.

1 papers

Where this stands

The written synthesis of this thread is for subscribers. Subscribe.