# BEEAR: Embedding-based Adversarial Removal of Safety Backdoors in Instruction-tuned Language Models

> Research article (Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, 2024) · cited 10× · AI/ML

**Wikidata**: [openalex:W4404782315](https://www.wikidata.org/wiki/openalex:W4404782315)  
**Source**: https://4ort.xyz/entity/beear-embedding-based-adversarial-removal-of-safety-backdoors-in-instruction-tuned-language-models