6 citations · 18 across the 20 of their papers we have counts for
7 papers · 1 filter
HARDE: Optimizing Agent Harnesses for Runtime Risk Detection and Execution Control
Zhuo Liu, Moxin Li, Zhixin Ma +3
Large language model (LLM) agents are vulnerable to safety risks such as injected malicious instructions or misleading information, motivating runtime defenses that prevent unsafe…
OpenCompass: A Universal Evaluation Platform for Large Language Models
Maosong Cao, Kai Chen, Haodong Duan +27
In recent years, the field of artificial intelligence has undergone a paradigm shift from task-specific small-scale models to general-purpose large language models (LLMs). With the…
Plant, Persist, Trigger: Sleeper Attack on Large Language Model Agents
Yongxiang Li, Moxin Li, Zhixin Ma +4
Large Language Model (LLM) agents remain vulnerable to safety threats from the external environment, where attackers inject adversarial content into external observations such as t…
ARES: Automated Rubric Synthesis for Scalable LLM Reinforcement Learning
Xiaoyuan Li, Keqin Bao, Moxin Li +5
Rubric-based rewards offer a promising way to extend reinforcement learning (RL) for large language models beyond tasks with automatically verifiable answers. However, scaling rubr…
SkillGraph: Skill-Augmented Reinforcement Learning for Agents via Evolving Skill Graphs
Xiaoyuan Li, Moxin Li, Keqin Bao +4
Skill libraries enable large language model agents to reuse experience from past interactions, but most existing libraries store skills as isolated entries and retrieve them only b…
SAGE: Scalable Automated Robustness Augmentation for LLM Knowledge Evaluation
Xiaoyuan Li, Yuzhe Wang, Moxin Li +6
Large Language Models (LLMs) achieve strong performance on standard knowledge evaluation benchmarks, yet recent work shows that their knowledge capabilities remain brittle under qu…
On Predicting the Post-training Potential of Pre-trained LLMs
Xiaoyuan Li, Yubo Ma, Kexin Yang +5
The performance of Large Language Models (LLMs) on downstream tasks is fundamentally constrained by the capabilities acquired during pre-training. However, traditional benchmarks l…