Picture for Bochuan Cao

Bochuan Cao

On the Intrinsic Self-Correction Capability of LLMs: Uncertainty and Latent Concept

Jun 04, 2024
Viaarxiv icon

XPrompt:Explaining Large Language Model's Generation via Joint Prompt Attribution

May 30, 2024
Viaarxiv icon

Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimization

May 28, 2024
Viaarxiv icon

WordGame: Efficient & Effective LLM Jailbreak via Simultaneous Obfuscation in Query and Response

Add code
May 22, 2024
Viaarxiv icon

On the Difficulty of Defending Contrastive Learning against Backdoor Attacks

Dec 14, 2023
Viaarxiv icon

Stealthy and Persistent Unalignment on Large Language Models via Backdoor Injections

Nov 15, 2023
Viaarxiv icon

IMPRESS: Evaluating the Resilience of Imperceptible Perturbations Against Unauthorized Data Usage in Diffusion-Based Generative AI

Add code
Oct 30, 2023
Viaarxiv icon

On the Safety of Open-Sourced Large Language Models: Does Alignment Really Prevent Them From Being Misused?

Oct 02, 2023
Viaarxiv icon

Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM

Sep 18, 2023
Figure 1 for Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM
Figure 2 for Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM
Figure 3 for Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM
Figure 4 for Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM
Viaarxiv icon

Wild-Time: A Benchmark of in-the-Wild Distribution Shift over Time

Add code
Nov 25, 2022
Figure 1 for Wild-Time: A Benchmark of in-the-Wild Distribution Shift over Time
Figure 2 for Wild-Time: A Benchmark of in-the-Wild Distribution Shift over Time
Figure 3 for Wild-Time: A Benchmark of in-the-Wild Distribution Shift over Time
Figure 4 for Wild-Time: A Benchmark of in-the-Wild Distribution Shift over Time
Viaarxiv icon