conftrace_

delta parameters

2 papers

Explore in graph

Co-occurring keywords · papers mentioning both

fine tuning (2) dare (1) homologous models (1) parameter fusing (1) safety compensation vector (1) utility (1) safety degradation (1) large language model (1) language model (1)

Papers

Safe Delta: Consistently Preserving Safety when Fine-Tuning LLMs on Diverse Datasets ICML 2025 Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch ICML 2024