8 papers
Improving Medical Calculation of LLMs with Embedded Coding
Tianshi Ming, Yingying Zhang, Xian Wu
Large Language Models (LLMs) perform well on medical examinations and question-answering benchmarks, but remain unreliable on medical calculation tasks that require exact numerical…
Controllable Affective Generation via Latent Vector Steering
Xixian Yong, Siyuan Chang, Yingying Zhang +2
Large Language Models (LLMs) often produce emotionally flattened responses after alignment, limiting their effectiveness in affect-sensitive applications. In this paper, we propose…
Robust Sequential Experimental Design for A/B Testing
Qianglin Wen, Xiangkun Wu, Chengchun Shi +4
Experimental design has emerged as a powerful approach for improving the sample efficiency of A/B testing, yet existing designs rely critically on correctly specified models. We st…
ECG-R1: Protocol-Guided and Modality-Agnostic MLLM for Reliable ECG Interpretation
Jiarui Jin, Haoyu Wang, Xingliang Wu +9
Electrocardiography (ECG) serves as an indispensable diagnostic tool in clinical practice, yet existing multimodal large language models (MLLMs) remain unreliable for ECG interpret…
Beyond Single Models: Mitigating Multimodal Hallucinations via Adaptive Token Ensemble Decoding
Jinlin Li, Yuran Wang, Yifei Yuan +5
Large Vision-Language Models (LVLMs) have recently achieved impressive results in multimodal tasks such as image captioning and visual question answering. However, they remain pron…
From Retrieval to Generation: Unifying External and Parametric Knowledge for Medical Question Answering
Lei Li, Xiao Zhou, Yingying Zhang +1
Medical question answering (QA) requires extensive access to domain-specific knowledge. A promising direction is to enhance large language models (LLMs) with external knowledge ret…
Inflated Excellence or True Performance? Rethinking Medical Diagnostic Benchmarks with Dynamic Evaluation
Xiangxu Zhang, Lei Li, Yanyun Zhou +3
Medical diagnostics is a high-stakes and complex domain that is critical to patient care. However, current evaluations of large language models (LLMs) remain limited in capturing k…
Copy-Paste to Mitigate Large Language Model Hallucinations
Yongchao Long, Xian Wu, Yingying Zhang +3
While Retrieval-Augmented Generation (RAG) enables large language models (LLMs) to generate contextually grounded responses, contextual faithfulness remains challenging as LLMs may…