TLDR AI · RESEARCH & INNOVATION
UNLEARNING SENSITIVE CONTENT FROM LLMS
Summary from the TLDR AI newsletter
This paper proposes a model merging technique to selectively forget sensitive content in LLMs while preserving general knowledge.
TLDR AI · RESEARCH & INNOVATION
This paper proposes a model merging technique to selectively forget sensitive content in LLMs while preserving general knowledge.