conftrace_

safety degradation

3 papers

Explore in graph

Co-occurring keywords · papers mentioning both

large language model (2) fine tuning (2) utility (1) delta parameters (1) alignment safeguards (1) outlier samples (1) self-inf-n (1) safety compensation vector (1) weight merging (1) model alignment (1)

Papers

Benign Samples Matter! Fine-tuning On Outlier Benign Samples Severely Breaks Safety ICML 2025 How Does Vision-Language Adaptation Impact the Safety of Vision Language Models? ICLR 2025 Safe Delta: Consistently Preserving Safety when Fine-Tuning LLMs on Diverse Datasets ICML 2025