Picture for Haomin Zhuang

Haomin Zhuang

UOE: Unlearning One Expert Is Enough For Mixture-of-experts LLMS

Add code
Nov 27, 2024
Viaarxiv icon

Social Science Meets LLMs: How Reliable Are Large Language Models in Social Simulations?

Add code
Oct 30, 2024
Viaarxiv icon

Defending Jailbreak Prompts via In-Context Adversarial Game

Add code
Feb 20, 2024
Figure 1 for Defending Jailbreak Prompts via In-Context Adversarial Game
Figure 2 for Defending Jailbreak Prompts via In-Context Adversarial Game
Figure 3 for Defending Jailbreak Prompts via In-Context Adversarial Game
Figure 4 for Defending Jailbreak Prompts via In-Context Adversarial Game
Viaarxiv icon

Backdoor Federated Learning by Poisoning Backdoor-Critical Layers

Add code
Aug 08, 2023
Figure 1 for Backdoor Federated Learning by Poisoning Backdoor-Critical Layers
Figure 2 for Backdoor Federated Learning by Poisoning Backdoor-Critical Layers
Figure 3 for Backdoor Federated Learning by Poisoning Backdoor-Critical Layers
Figure 4 for Backdoor Federated Learning by Poisoning Backdoor-Critical Layers
Viaarxiv icon

A Comparison of Image Denoising Methods

Add code
Apr 18, 2023
Viaarxiv icon

A Pilot Study of Query-Free Adversarial Attack against Stable Diffusion

Add code
Apr 03, 2023
Viaarxiv icon