Learning and Forgetting Unsafe Examples in Large Language Models

Jiachen Zhao, Zhun Deng, David Madras +2

As the number of large language models (LLMs) released to the public grows, there is a pressing need to understand the safety implications associated with these models learning fro…