Shengyun Si | Technical University of Munich

About Me

I am a master student in Robotics, Cognition, Intelligence at Technical University of Munich. Currently, I’m completing my master thesis about safety alignment of LLMs under the supervision of Prof. Barbara Plank, working with Xinpeng Wang. Previously, I worked as a research assistant in the NLP group at Yale University under the supervision of Prof. Arman Cohan, working with Yilun Zhao. My current research focuses on 1) Safety Alignment of Large Language Models, 2) LLM-based Retrieval Augmented Generation, 3) Multimodal Large Language Models.

I obtained my B.S. degree from Zhejiang University. During this period, I did my thesis about leveraging SVM algorithm to detect two-phase flow.

Research Interests

Natural Language Processing: Safety Alignment of LLMs, Table to Text Generation, Retrieval Augmented Generation

News

[Nov. 2024] Our paper about open-source Table2Text tool is accepted to EMNLP 2024.
[Dec. 2023] Our paper about investigating LLMs in Table2Text task is accepted to EMNLP 2023

Publications

THINK BEFORE REFUSE: Triggering Safety Reflection in LLMs to Mitigate False Refusal Behaviour

Shengyun Si*, Xinpeng Wang*, Guangyao Zhai, Nassir Navab, Barbara Plank

Arxiv

PDF under review

OpenT2T: An Open-Source Toolkit for Table-to-Text Generation

Haowei Zhang*, Shengyun Si*, Yilun Zhao, Lujing Xie, Zhijian Xu, Lyuhao Chen, Linyong Nan, Pengcheng Wang, Xiangru Tang, Arman Cohan

EMNLP, 2024

PDF Demo Track

Investigating table-to-text generation capabilities of large language models in real-world information seeking scenarios

Yilun Zhao*, Haowei Zhang*, Shengyun Si*, Linyong Nan, Xiangru Tang, Arman Cohan

EMNLP, 2023

PDF Industry Track