30 papers
EmoRSS: Mitigating Emotion-Induced Over-Refusal in Large Language Models
Shuyi Miao, Yaojin Ma, Chenhang Cui +5
Emotional expression can influence the safety decisions of large language models (LLMs), offering a potential avenue for improving safety alignment. Existing studies have mainly fo…
ContractLens: Latent Security Knowledge for Malicious Smart Contract Detection
Shuyi Miao, Xinyi Huang, Wangjie Qiu +4
Detecting malicious smart contracts is essential to safeguarding the Web3.0 ecosystem. However, existing auditing methods based on large language models (LLMs) largely rely on prom…
DV-Lens: Revealing the Functional Organization of Language Model Parameters
Chenhang Cui, Jian Yu, Shuyi Miao +5
Understanding parameter functions helps elucidate the internal mechanisms of large language models (LLMs). However, how to connect parameters from different modules to verifiable o…
ACTR: Aligning Thoughts and Responses for Multilingual Safety in Reasoning LLMs
Xianhui Zhang, Jian Yu, Chengyu Xie +6
Ensuring the safety of reasoning large language models (LLMs) across languages is essential for their reliable deployment. However, when exposed to jailbreak attacks in non-high-re…
What Does Privileged Information Add to On-Policy Self-Distillation?
XiuYu Zhang, Wei Chow, Junfeng Fang +3
On-policy self-distillation (OPSD) lets a language model learn from a frozen copy of itself that sees an answer or a worked solution. Giving the teacher this extra information seem…
Mechanist: AI as a Scientific Instrument for Discovering the Mechanisms of Intelligence
Mengru Wang, Junfeng Fang, Shuofei Qiao +17
AI models are increasingly used in scientific discovery and human decision-making. Yet how AI models work and what risks they pose remain poorly understood. As AI development becom…
LMSM: LLM Security Framework Inspired by Linux Security Modules
XiuYu Zhang, Bonan Ruan, Junfeng Fang +3
Large language models (LLMs) are increasingly deployed with layered defenses, yet malicious prompts can still bypass them. Interpretability methods can expose model-internal signal…
MAE I Trust Myself? Self-Evaluating VLA Action Generation with Markov Attention Entropy
Aniri, Chen Yilin, Jinhe Bi +10
Vision-Language-Action models (VLAs) integrate visual perception, language instruction, and action generation into end-to-end policies across heterogeneous architectures. However,…
Continual Learning in Transition
Zhiyan Hou, Dan Zhang, Tao Feng +11
Classical continual learning (CL) has primarily focused on enabling models to update and retain knowledge through parameter-centric mechanisms, e.g., training strategies, architect…
SafeNexus: Discovering and Steering Modality-Universal Safety Neurons in MLLMs
Jian Yu, Fei Shen, Cong Wang +5
Although Large Language Models (LLMs) have demonstrated promising safety performance, extending them to Multimodal Large Language Models (MLLMs) exposes a significant gap between e…
Who Bridges Safety? Identifying and Targeting Cross-Lingual Shared Safety Pathways
Shuyi Miao, Wangjie Qiu, Pengyang Shao +4
Uncovering the internal mechanisms underlying the safety capabilities of large language models (LLMs) is crucial for developing trustworthy artificial intelligence. Currently, mech…
OPD-V: Visual On-Policy Self-Distillation with Modality Balance
Aniri, Jinhe Bi, Peng Liao +5
On-Policy Self-Distillation (OPSD) has become a standard post-training approach for improving visual reasoning in multimodal large language models (MLLMs). Existing methods draw pr…
ReflectRL: Learning from Golden Negative Trajectories via Reflective-to-Direct Reasoning
Jinhe Bi, Chennan Zhou, Zengjie Jin +10
On-policy training has emerged as a powerful post-training paradigm for improving the reasoning capabilities of large language models, and is often enhanced by golden trajectories…
One Anchor for All: Unified Multilingual and Multimodal Safety Alignment for LVLMs
Enyi Shi, Fei Shen, Chuancheng Shi +4
As large vision-language models (LVLMs) are deployed globally, the combination of multilingual instructions and visual information makes malicious attacks more covert and sophistic…
Suppressing Forgery-Specific Shortcuts for Generalizable Deepfake Detection
Yihui Wang, Yonghui Yang, Jilong Liu +3
Deepfake detection suffers from poor generalization across forgery methods, as existing models tend to rely on spurious method-specific shortcuts that fail to transfer to unseen ma…
Preference-Aware Rubric Learning for Personalized Evaluation
Yilun Qiu, Xiaoyan Zhao, Yang Zhang +7
As Large Language Models (LLMs) evolve from general-purpose assistants to user-centric agents, personalization has become central to aligning model behavior with individual prefere…
Tiny Brains, Giant Impact: Uncovering the Keystone Neurons of LLM with Just a Few Prompts
Xiangtian Ji, Yuxin Chen, Zhengzhou Cai +3
Large language models (LLMs) display strong comprehensive abilities, yet the internal mechanisms that support these behaviors remain insufficiently understood. In this work, we sho…
Know You Before You Speak: User-State Modeling for LLM Personalization in Multi-Turn Conversation
Jiani Luo, Xiaoyan Zhao, Yang Zhang +4
Personalized dialogue requires more than recalling explicit user histories: systems also need to infer hidden user states that evolve through interaction and shape appropriate resp…
Rubric-based On-policy Distillation
Junfeng Fang, Zhepei Hong, Mao Zheng +7
On-policy distillation (OPD) is a powerful paradigm for model alignment, yet its reliance on teacher logits restricts its application to white-box scenarios. We contend that struct…
On the Implicit Reward Overfitting and the Low-rank Dynamics in RLVR
Hao Ye, Jisheng Dang, Junfeng Fang +7
Recent extensive research has demonstrated that the enhanced reasoning capabilities acquired by models through Reinforcement Learning with Verifiable Rewards (RLVR) are primarily c…
UniDetect: LLM-Driven Universal Fraud Detection across Heterogeneous Blockchains
Shuyi Miao, Wangjie Qiu, Shengda Zhuo +5
As cross-chain interoperability advances, decentralized finance (DeFi) protocols enable illicit funds to be reorganized into uniform liquid assets that flow throughout the cryptocu…
Unifying Group-Relative and Self-Distillation Policy Optimization via Sample Routing
Gengsheng Li, Tianyu Yang, Junfeng Fang +6
Reinforcement learning with verifiable rewards (RLVR) has become a standard paradigm for post-training large language models. While Group Relative Policy Optimization (GRPO) is wid…
Targeted Interpretable Safety Neuron Enhancement for Multilingual Vision-Language Large Models
Enyi Shi, Fei Shen, Chuancheng Shi +5
With the widespread deployment of vision-language large models (VLLMs), their safety alignment faces dual challenges across languages and modalities. Existing methods model multili…
DRAFT: Task Decoupled Latent Reasoning for Agent Safety
Lin Wang, Junfeng Fang, Dan Zhang +3
The advent of tool-using LLM agents shifts safety monitoring from output moderation to auditing long, noisy interaction trajectories, where risk-critical evidence is sparse-making…