9 citations · 28 across the 19 of their papers we have counts for
20 papers
TRACE: Trajectory-Based Safety Patch Learning for LLM Post-Training Realignment
Changyue Li, Jiaming He, Youliang Yuan +4
Fine-Tuning-as-a-Service (FTaaS) platforms let users perform supervised fine-tuning (SFT) on customized data, but this pipeline can erode model safety alignment. To recover safety…
SHAPE: Unifying Safety, Helpfulness and Pedagogy for Educational LLMs
Sihang Zhao, Kangrui Yu, Youliang Yuan +2
Large Language Models (LLMs) have been widely explored in educational scenarios. We identify a critical vulnerability in current educational LLMs, pedagogical jailbreaks, where stu…
Semantic Router: On the Feasibility of Hijacking MLLMs via a Single Adversarial Perturbation
Changyue Li, Jiaying Li, Youliang Yuan +3
Multimodal Large Language Models (MLLMs) are increasingly deployed in stateless systems, such as autonomous driving and robotics. This paper investigates a novel threat: Semantic-A…
Curing Miracle Steps in LLM Mathematical Reasoning with Rubric Rewards
Youliang Yuan, Qiuyang Mang, Jingbang Chen +7
In this paper, we observe that current models are susceptible to reward hacking, leading to a substantial overestimation of a model's reasoning ability. This is evidenced by a high…
Towards Evaluating Proactive Risk Awareness of Multimodal Language Models
Youliang Yuan, Wenxiang Jiao, Yuejin Xie +5
Human safety awareness gaps often prevent the timely recognition of everyday risks. In solving this problem, a proactive safety artificial intelligence (AI) system would work bette…
VisBias: Measuring Explicit and Implicit Social Biases in Vision Language Models
Jen-tse Huang, Jiantong Qin, Jianping Zhang +3
This research investigates both explicit and implicit social biases exhibited by Vision-Language Models (VLMs). The key distinction between these bias types lies in the level of aw…
Human Cognitive Benchmarks Reveal Foundational Visual Gaps in MLLMs
Jen-Tse Huang, Dasen Dai, Jen-Yuan Huang +7
Humans develop perception through a bottom-up hierarchy: from basic primitives and Gestalt principles to high-level semantics. Current Multimodal Large Language Models (MLLMs) are…
Can't See the Forest for the Trees: Benchmarking Multimodal Safety Awareness for Multimodal LLMs
Wenxuan Wang, Xiaoyuan Liu, Kuiyi Gao +5
Multimodal Large Language Models (MLLMs) have expanded the capabilities of traditional language models by enabling interaction through both text and images. However, ensuring the s…
Difficult Task Yes but Simple Task No: Unveiling the Laziness in Multimodal LLMs
Sihang Zhao, Youliang Yuan, Xiaoying Tang +1
Multimodal Large Language Models (MLLMs) demonstrate a strong understanding of the real world and can even handle complex tasks. However, they still fail on some straightforward vi…
Insight Over Sight: Exploring the Vision-Knowledge Conflicts in Multimodal LLMs
Xiaoyuan Liu, Wenxuan Wang, Youliang Yuan +4
This paper explores the problem of commonsense level vision-knowledge conflict in Multimodal Large Language Models (MLLMs), where visual information contradicts model's internal co…
Chain-of-Jailbreak Attack for Image Generation Models via Editing Step by Step
Wenxuan Wang, Kuiyi Gao, Youliang Yuan +5
Text-based image generation models, such as Stable Diffusion and DALL-E 3, hold significant potential in content creation and publishing workflows, making them the focus in recent…
Learning to Ask: When LLM Agents Meet Unclear Instruction
Wenxuan Wang, Juluan Shi, Zixuan Ling +7
Equipped with the capability to call functions, modern large language models (LLMs) can leverage external tools for addressing a range of tasks unattainable through language skills…
On the Resilience of LLM-Based Multi-Agent Collaboration with Faulty Agents
Jen-tse Huang, Jiaxu Zhou, Tailin Jin +6
Large language model-based multi-agent systems have shown great abilities across various tasks due to the collaboration of expert agents, each focusing on a specific domain. Howeve…
Refuse Whenever You Feel Unsafe: Improving Safety in LLMs via Decoupled Refusal Training
Youliang Yuan, Wenxiang Jiao, Wenxuan Wang +5
This study addresses a critical gap in safety tuning practices for Large Language Models (LLMs) by identifying and tackling a refusal position bias within safety tuning data, which…
How Far Are We on the Decision-Making of LLMs? Evaluating LLMs' Gaming Ability in Multi-Agent Environments
Jen-tse Huang, Eric John Li, Man Ho Lam +7
Decision-making is a complex process requiring diverse abilities, making it an excellent framework for evaluating Large Language Models (LLMs). Researchers have examined LLMs' deci…
New Job, New Gender? Measuring the Social Bias in Image Generation Models
Wenxuan Wang, Haonan Bai, Jen-tse Huang +5
Image generation models can generate or edit images from a given text. Recent advancements in image generation technology, exemplified by DALL-E and Midjourney, have been groundbre…
Identifying the Achilles' Heel: An Iterative Method for Dynamically Uncovering Factual Errors in Large Language Models
Wenxuan Wang, Yuk-Kit Chan, Zixuan Ling +7
Large Language Models (LLMs) like ChatGPT are foundational in various applications due to their extensive knowledge from pre-training and fine-tuning. Despite this, they are prone…
LogicAsker: Evaluating and Improving the Logical Reasoning Ability of Large Language Models
Yuxuan Wan, Wenxuan Wang, Yiliu Yang +5
We introduce LogicAsker, a novel approach for evaluating and enhancing the logical reasoning capabilities of large language models (LLMs) such as ChatGPT and GPT-4. Despite LLMs' p…
Who is ChatGPT? Benchmarking LLMs' Psychological Portrayal Using PsychoBench
Jen-tse Huang, Wenxuan Wang, Eric John Li +6
Large Language Models (LLMs) have recently showcased their remarkable capacities, not only in natural language processing tasks but also across diverse domains such as clinical med…
All Languages Matter: On the Multilingual Safety of Large Language Models
Wenxuan Wang, Zhaopeng Tu, Chang Chen +4
Safety lies at the core of developing and deploying large language models (LLMs). However, previous safety benchmarks only concern the safety in one language, e.g. the majority lan…