9 papers
ASCII Attack: Recontextualising Harmful Requests as Artistic Critique in Large Language Models
Da Cheng Gu, Yifei Dong, Xinghao Yang +2
Safety alignment trains large language models to refuse harmful requests stated plainly, but that training is applied mostly to surface form. Requests that only recontextualise the…
SkillMaster: Toward Autonomous Skill Mastery in LLM Agents
Min Yang, Jinghua Piao, Xu Xia +4
Skills provide an effective mechanism for improving LLM agents on complex tasks, yet in existing agent frameworks, their creation, refinement, and selection are typically governed…
SCOT: Multi-Source Cross-City Transfer with Optimal-Transport Soft-Correspondence Objective
Yuyao Wang, Min Yang, Meng Chen +3
Cross-city transfer improves prediction in label-scarce cities by leveraging labeled data from other cities, but it becomes challenging when cities adopt incompatible partitions an…
Joint Semantic Token Selection and Prompt Optimization for Interpretable Prompt Learning
Yating Wang, Yaqi Zhao, Yongshun Gong +2
Vision-language models such as CLIP achieve strong visual-textual alignment, but often suffer from overfitting and limited interpretability when adapted through continuous prompt l…
Distributed Multi-Layer Editing for Rule-Level Knowledge in Large Language Models
Yating Wang, Wenting Zhao, Yaqi Zhao +3
Large language models store not only isolated facts but also rules that support reasoning across symbolic expressions, natural language explanations, and concrete instances. Yet mo…
From Text to Forecasts: Bridging Modality Gap with Temporal Evolution Semantic Space
Lehui Li, Yuyao Wang, Jisheng Yan +5
Incorporating textual information into time-series forecasting holds promise for addressing event-driven non-stationarity; however, a fundamental modality gap hinders effective fus…
What Papers Don't Tell You: Recovering Tacit Knowledge for Automated Paper Reproduction
Lehui Li, Ruining Wang, Haochen Song +8
Automated paper reproduction -- generating executable code from academic papers -- is bottlenecked not by information retrieval but by the tacit knowledge that papers inevitably le…
Cross-Entropy Attacks to Language Models via Rare Event Simulation
Mingze Ni, Yongshun Gong, Wei Liu
Black-box textual adversarial attacks are challenging due to the lack of model information and the discrete, non-differentiable nature of text. Existing methods often lack versatil…
Deceiving Question-Answering Models: A Hybrid Word-Level Adversarial Approach
Jiyao Li, Mingze Ni, Yongshun Gong +1
Deep learning underpins most of the currently advanced natural language processing (NLP) tasks such as textual classification, neural machine translation (NMT), abstractive summari…