Measuring the Depth of LLM Unlearning via Activation Patching

3 points | by jaeunglee 9 hours ago

1 comments