Search arXiv⌕ Search

EXPLORE CONNECTIONS

Surgery: Mitigating Harmful Fine-Tuning for Large Language Models via Attention Sink

Follow the relationships that help you find your next source.

Connections are not prepared for this record yet, or its source metadata has changed. The original record remains available while background snapshots are rebuilt.