7 papers
Alignment-Weighted DPO: A principled reasoning approach to improve safety alignment
Mengxuan Hu, Vivek V. Datla, Anoop Kumar +4
Recent advances in alignment techniques such as Supervised Fine-Tuning (SFT), Reinforcement Learning from Human Feedback (RLHF), and Direct Preference Optimization (DPO) have impro…
Enhanced Diagnostic Performance via Large-Resolution Inference Optimization for Pathology Foundation Models
Mengxuan Hu, Zihan Guan, John Kang +2
Despite their prominent performance on tasks such as ROI classification and segmentation, many pathology foundation models remain constrained by a specific input size e.g. 224 x 22…
BalancEdit: Dynamically Balancing the Generality-Locality Trade-off in Multi-modal Model Editing
Dongliang Guo, Mengxuan Hu, Zihan Guan +2
Large multi-modal models inevitably decay over time as facts update and previously learned information becomes outdated. Traditional approaches such as fine-tuning are often imprac…
Benign Samples Matter! Fine-tuning On Outlier Benign Samples Severely Breaks Safety
Zihan Guan, Mengxuan Hu, Ronghang Zhu +2
Recent studies have uncovered a troubling vulnerability in the fine-tuning stage of large language models (LLMs): even fine-tuning on entirely benign datasets can lead to a signifi…
UFID: A Unified Framework for Input-level Backdoor Detection on Diffusion Models
Zihan Guan, Mengxuan Hu, Sheng Li +1
Diffusion models are vulnerable to backdoor attacks, where malicious attackers inject backdoors by poisoning certain training samples during the training stage. This poses a signif…
Backdoor in Seconds: Unlocking Vulnerabilities in Large Pre-trained Models via Model Editing
Dongliang Guo, Mengxuan Hu, Zihan Guan +3
Large pre-trained models have achieved notable success across a range of downstream tasks. However, recent research shows that a type of adversarial attack ( backdo…