作者位次 #2/6ICML 2026Poster· Deep Learning / Large Language Models
Toward Safe Quantization-Aware Fine-tuning: Understanding and Mitigating Safety Alignment Degradation
Yuning Yang · Guowei Peng · Xiurui Xie · Minrui Jiang · Shuang Liang · Guisong Liu
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Yuning Yang · Guowei Peng · Xiurui Xie · Minrui Jiang · Shuang Liang · Guisong Liu
Jiayu Tang · Guowei Peng · Qiuhao Xie · Yuning Yang · Xiurui Xie · Guisong Liu
The “Fine-Tuning-as-a-Service” paradigm exposes large language models to catastrophic safety degradation from less harmful samples. Alignment-stage defenses address this by proactively injecting adversarial perturbations to bolster the model’s inherent robustness against harmful drift. However, e...