Lei Sha

Professor, Beihang University
I'm always looking for highly motivated students to work with me on broad NLP research. Please feel free to reach out and apply!

I am now a professor in the Artificial Intelligence Institute of Beihang University.

Previously, I was a research associate in University of Oxford under the supervision of Prof. Thomas Lukasiewicz. I obtained my Ph.D. degree from Peking University in 2018. During my Ph.D. period, I was co-advised by Prof. Zhifang Sui, Baobao Chang and Sujian Li.

Research Interests: Trustworthy LLMs, LLM for science

  • Email: shalei[AT]buaa.edu.cn
Group

The Language Computing and Sequential Data Analysis Group (LecSa Group)

Current Students

2023
Hao Wang(Beihang IAI, PhD) Tianrui Gou(Beihang IAI, PhD)
2024
Hao Li(Beihang IAI, Master) Sihan Fan(Beihang Shen Yuan Honors College, Master) Zengqi Xiu(Beihang Shen Yuan Honors College / ZGC Lab, PhD) Jian Liu(Beihang IAI, PhD) Xiuping Liang(Beihang IAI, PhD) Yunhao Mei(Beihang IAI, PhD)
2025
Guanhua Chen(Beihang IAI, PhD) Pengsheng Chen(Beihang ECPKN, Master) Wenxin Wu(Beihang IAI, Master) Yanting Wang(Beihang IAI, PhD)
2026
Wenhan Yu(Beihang IAI, Master) Chen Wang(Beihang IAI, PhD) Yanqiu Zhao(Beihang IAI(GJJZX), Master) Yunbo Zhang(Beihang IAI, PhD, AI4sci) Zeyang Yue(Beihang IAI / ZGC Academy, PhD, AI4sci)

Research Internship

2023
Shanghao Bai(Beihang University, Bachelor; LLM related research; Sep 2025 -) Xi Chen(Beihang University, Bachelor; LLM Safety (ZGC); Sep 2025 -) Dongze Wu(Beihang University, Bachelor; LLM related research; Mar 2025 -) Wendi Feng(Beihang University, Bachelor; LLM related research; May 2025 -) Mozhi Li(Beihang University, Bachelor; LLM related research; Sep 2025 -) Bangxu Li(Beihang University, Bachelor; LLM related research; Sep 2025 -) Jie Li(USTB, Bachelor; LLM related research; May 2025 -) Jia-Ning Zhou(Beihang University, Bachelor; LLM related research; July 2025 -) ZhiXiang Yu(Beihang University, Bachelor; LLM related research; Sep 2025 -)
2024
Dayou Zhou(Beihang University, Bachelor; LLM for 3D generation; Nov 2024 -) Tsan Ouyang(Beihang University, Bachelor; LLM related research; Feb 2025 -) Qiang Qi(Beihang University, Bachelor; LLM related research; May 2025 -) Yueran Pei(Beihang University, Bachelor; LLM related research; May 2025 -) Wenjing Zhi(Beihang University, Bachelor; LLM related research; May 2025 -) Xingzhou Zhang(Beihang University, Bachelor; LLM related research; May 2025 -) Yizhuo Duan(Beihang University, Bachelor; LLM related research; July 2025 -) Shiyun Peng(Beihang University, Bachelor; LLM related research; Sep 2025 -) Hao Jin(Beihang University, Bachelor; LLM related research; Sep 2025 -) Yijie Qiao(Beihang University, Bachelor; LLM related research; Sep 2025 -) Yingcheng Mao(Beihang University, Bachelor; LLM related research; Sep 2025 -) Yucheng Yao(Beihang University, Bachelor; LLM related research; Sep 2025 -) Weilun Zhao(Beihang University, Bachelor; LLM Safety (ZGC); Sep 2025 -) Xinping Liu(Beihang University, Bachelor; LLM Safety (ZGC); Sep 2025 -) Yiheng Li(USTB, Bachelor; LLM Safety (ZGC); Sep 2025 -) Zhipeng Song(Beihang University, Bachelor; LLM Safety (ZGC); Sep 2025 -)

Visiting Scholar

2025
Chun Kang(Beihang University; LLM research; Mar 2025 -) Zhuang Liu(Beihang University; LLM research; Apr 2025 -)

Alumni

2016
Feng Qian
Peking University Bachelor Harvard University Master
SRL Research Oct 2016 - Feb 2017
2017
Chengyue Gong
Peking University Bachelor University of Texas at Austin PhD
ML Research Jun 2017 - Aug 2017
2023
Qianpu Liu
Beihang University Bachelor GaoLing School of AI, RUC PhD
LLM for cfRNA LLM for Theorem Proving May 2023 - Oct 2025
Publications
(#) for corresponding author
  1. MA-RWG: A Multi-Agent Framework for Thematically Structuring and Generation of Related Work
    Zhuang Liu, Jian Liu, Chun Kang, Chenbin Zhang, Rui Li, Fanhu Zeng, Yong Dai, Lei Sha(#)
    In Proceedings of IJCAI 2026 Conference, 2026. (IJCAI 2026).
    [PDF][Arxiv]
  2. StepCodeReasoner: Aligning Code Reasoning with Stepwise Execution Traces via Reinforcement Learning
    Hao Wang, Lei Sha(#), Jie M. Zhang(#)
    In Proceedings of ICML 2026 Conference, 2026. (ICML 2026).
    [PDF][Arxiv]
  3. SafetyMem: Adaptive Jailbreak Defense via Dual-Component Safety Memory
    Hao Wang, Ziyi Ni, Huacan Wang, Pin Lyu, Lei Sha(#)
    In Proceedings of ACL 2026 Conference, 2026. (ACL 2026).
    [PDF][Arxiv]
  4. Be Your Own Red Teamer: Safety Alignment via Self-Play and Reflective Experience Replay
    Hao Wang, Yanting Wang, Hao Li, Rui Li, Lei Sha(#)
    In Findings of ACL 2026 Conference, 2026. (ACL 2026 findings).
    [PDF][Arxiv]
  5. HAUNTATTACK: When Attack Follows Reasoning as a Shadow
    Jingyuan Ma, Rui Li, Zheng Li, Junfeng liu, Heming Xia, Lei Sha, Zhifang Sui
    In Findings of ACL 2026 Conference, 2026. (ACL 2026 findings).
    [PDF][Arxiv]
  6. Rethinking Semantic–Collaborative Integration: Why Alignment Is Not Enough
    Maolin Wang, Dongze Wu, Jianing Zhou, Hongyu Chen, Beining Bao, Yu Jiang, Chenbin Zhang, Chang Wang, Jian Liu, Lei Sha(#)
    In Proceedings of SIGIR 2026 Conference, 2026. (SIGIR 2026).
    [PDF][Arxiv]
  7. Large Language Models Struggle with Unreasonability in Math Problems
    Jingyuan Ma, Damai Dai, Zihang Yuan, Rui Li, Weilin Luo, Bin Wang, Qun Liu, Lei Sha, Zhifang Sui
    In Proceedings of AAAI 2026 Conference, 2026. (AAAI 2026).
    [PDF][Arxiv]
  8. Unlocking Multi-View Insights in Knowledge-Dense Retrieval-Augmented Generation
    Guanhua Chen, Wenhan Yu, Xiao Lu, Xiao Zhang, Erli Meng, Lei Sha(#)
    In IEEE/ACM Transactions on Audio, Speech, and Language Processing, 2025. (TASLP 2025).
    [PDF][Arxiv]
  9. DiffusionAttacker: Diffusion-Driven Prompt Manipulation for LLM Jailbreak
    Hao Wang, Hao Li, Junda Zhu, Xinyuan Wang, Chengwei Pan, Minlie Huang, Lei Sha(#)
    In Proceedings of EMNLP 2025 Conference, 2025. (EMNLP 2025).
    [PDF][Arxiv]
  10. Reasoning-to-Defend: Safety-Aware Reasoning Can Defend Large Language Models from Jailbreaking
    Junda Zhu, Lingyong Yan, Shuaiqiang Wang, Dawei Yin, Lei Sha(#)
    In Proceedings of EMNLP 2025 Conference, 2025. (EMNLP 2025).
    [PDF][Arxiv]
  11. Layer-Aware Representation Filtering: Purifying Finetuning Data to Preserve LLM Safety Alignment
    Hao Li, Lijun Li, Zhenghao Lu, Xianyi Wei, Rui Li, Jing Shao, Lei Sha(#)
    In Proceedings of EMNLP 2025 Conference, 2025. (EMNLP 2025).
    [PDF][Arxiv]
  12. LLMs know their vulnerabilities: Uncover Safety Gaps through Natural Distribution Shifts [outstanding paper]
    Qibing Ren, Hao Li, Dongrui Liu, Zhanxu Xie, Xiaoya Lu, Yu Qiao, Lei Sha, Junchi Yan, Lizhuang Ma, Jing Shao
    In Proceedings of ACL 2025 Conference, 2025. (ACL 2025).
    [PDF][Arxiv]
  13. How Far are LLMs from Being Our Digital Twins? A Benchmark for Persona-Based Behavior Chain Simulation
    Rui Li, Heming Xia, Xinfeng Yuan, Qingxiu Dong, Lei Sha, Wenjie Li, Zhifang Sui
    In Findings of ACL 2025 Conference, 2025. (ACL 2025 Findings).
    [PDF][Arxiv]
  14. Towards Harmonized Uncertainty Estimation for Large Language Models
    Rui Li, Jing Long, Muge Qi, Heming Xia, Lei Sha, Peiyi Wang, Zhifang Sui
    In Proceedings of ACL 2025 Conference, 2025. (ACL 2025).
    [PDF][Arxiv]
  15. Be a Multitude to Itself: A Prompt Evolution Framework for Red Teaming
    Rui Li, Peiyi Wang, Jingyuan Ma, Di Zhang, Lei Sha, Zhifang Sui
    In Proceedings of EMNLP 2024 Conference, 2024. (EMNLP 2024).
    [PDF][Arxiv]
  16. ASETF: A Novel Method for Jailbreak Attack on LLMs through Translate Suffix Embeddings
    Hao Wang, Hao Li, Minlie Huang, Lei Sha(#)
    In Proceedings of EMNLP 2024 Conference, 2024. (EMNLP 2024).
    [PDF][Arxiv]
  17. ATM: Adversarial Tuning Multi-agent System Makes a Robust Retrieval-Augmented Generator
    Junda Zhu, Lingyong Yan, Haibo Shi, Dawei Yin, Lei Sha(#)
    In Proceedings of EMNLP 2024 Conference, 2024. (EMNLP 2024).
    [PDF][Arxiv]
  18. Harnessing the Plug-and-Play Controller by Prompting
    Hao Wang, Lei Sha(#)
    In Proceedings of EMNLP 2023 WorkShop, 2023. (EMNLP-GEM 2023).
    [PDF][Arxiv]
  19. Correcting Flaws in Common Disentanglement Metrics
    Louis Mahon, Lei Sha, Thomas Lukasiewicz
    In Transactions of Machine Learning Research, 2024. (TMLR 2024).
    [PDF][Arxiv]
  20. Text Attribute Control via Closed-Loop Disentanglement
    Lei Sha, Thomas Lukasiewicz
    In Transactions of the Association for Computational Linguistics. (TACL 2024). [IF=17.59]
    [PDF]
  21. Rationalizing Predictions by Adversarial Information Calibration
    Lei Sha, Oana-Maria Camburu, Thomas Lukasiewicz
    In Artificial Intelligence. (AI 2023). [IF=14.05]
    [PDF][Arxiv]
  22. Controlling Text Edition by Changing Answers of Specific Questions
    Lei Sha, Patrick Hohenecker, Thomas Lukasiewicz
    In Finding of ACL 2021 Conference, 2021. (ACL 2021 (Finding)).
    [PDF][Arxiv][Data & Code]
  23. Learning from the Best: Rationalizing Predictions by Adversarial Information Calibration
    Lei Sha, Oana-Maria Camburu, Thomas Lukasiewicz
    In Proceedings of AAAI 2021 Conference, 2021. (AAAI 2021).
    [PDF][Arxiv][Code][BeerAdvocate]
  24. Multi-type Disentanglement without Adversarial Training
    Lei Sha, Thomas Lukasiewicz
    In Proceedings of AAAI 2021 Conference, 2021. (AAAI 2021).
    [PDF][Arxiv][Code]
  25. Gradient-guided Unsupervised Lexically Constrained Text Generation
    Lei Sha
    In Proceedings of EMNLP 2020 Conference, 2020. (EMNLP 2020).
    [PDF]
  26. Estimating Minimum Operation Steps via Memory-based Recurrent Calculation Network
    Lei Sha, Chen Shi, Qi Chen, Lintao Zhang, Houfeng Wang
    In 2020 international joint conference on neural networks. (IJCNN 2020).
    [PDF][Code]
  27. Order-Planning Neural Text Generation from Structured Data
    Lei Sha, Lili Mou, Tianyu Liu, Pascal Poupart, Sujian Li, Baobao Chang, Zhifang Sui
    In Proceedings of AAAI 2018 Conference, 2018. (AAAI 2018).
    [PDF][Arxiv]
  28. A Multi-View Fusion Neural Network for Answer Selection
    Lei Sha, Xiaodong Zhang, Feng Qian, Baobao Chang, Zhifang Sui
    In Proceedings of AAAI 2018 Conference, 2018. (AAAI 2018).
    [PDF]
  29. Jointly Extracting Event Triggers and Arguments by Dependency-Bridge RNN and Tensor-Based Argument Interaction
    Lei Sha, Feng Qian, Baobao Chang, Zhifang Sui
    In Proceedings of AAAI 2018 Conference, 2018. (AAAI 2018).
    [AAAIPDF] [PDF]
  30. Will Repeated Reading Benefit Natural Language Understanding?
    Lei Sha, Zhifang Sui
    In Proceedings of NLPCC 2017 Conference, 2017. (NLPCC 2017).
    [PDF]
  31. Reading and Thinking: Re-read LSTM Unit for Textual Entailment Recognition
    Lei Sha, Baobao Chang, Zhifang Sui, Sujian Li
    In Proceedings of Coling 2016 Conference, 2016. (Coling 2016).
    [PDF]
  32. Capturing Argument Relationship for Chinese Semantic Role Labeling
    Lei Sha, Tingsong Jiang, Sujian Li, Baobao Chang, Zhifang Sui
    In Proceedings of EMNLP 2016 Conference, 2016. (EMNLP 2016).
    [PDF]
  33. RBPB: Regularization-Based Pattern Balancing Method for Event Extraction
    Lei Sha, Jing Liu, Chin-Yew Lin, Sujian Li, Baobao Chang, Zhifang Sui
    In Proceedings of ACL 2016 Conference, 2016. (ACL 2016).
    [PDF]
  34. Joint Learning Templates and Slots for Event Schema Induction
    Lei Sha, Sujian Li, Baobao Chang, Zhifang Sui.
    In Proceedings of NAACL 2016 Conference, 2016. (NAACL 2016).
    [arxiv(including supplement material)][PDF]
  35. Recognizing Textual Entailment Using Probabilistic Inference
    Lei Sha, Sujian Li, Tingsong Jiang, Baobao Chang, Zhifang Sui.
    In Proceedings of the EMNLP 2015 Conference, 2015. (EMNLP 2015).
    [PDF]
  36. RecInDial: A Unified Framework for Conversational Recommendation with Pretrained Language Models.
    Lingzhi Wang, Huang Hu, Lei Sha, Can Xu, Daxin Jiang, Kam-Fai Wong
    In Proceedings of AACL 2022 Conference, 2022. (AACL 2022).
    [PDF]
  37. Small Changes Make Big Differences: Improving Multi-turn Response Selection in Dialogue Systems via Fine-Grained Contrastive Learning.
    Yuntao Li, Can Xu, Huang Hu, Lei Sha, Yan Zhang, Daxin Jiang
    In Proceedings of INTERSPEECH 2022 Conference, 2022. ( INTERSPEECH 2022).
    [PDF]
  38. Associative Memory via Predictive Coding
    Tommaso Salvatori, Yuhang Song, Yujian Hong, Simon Frieder, Lei Sha, Zhenghua Xu, Rafal Bogacz, Thomas Lukasiewicz
    In Proceedings of NeurIPS 2021 Conference, 2021. (NeurIPS 2021).
    [PDF] [Arxiv]
  39. Auto-Dialabel: Labeling Dialogue Data with Unsupervised Learning
    Chen Shi, Qi Chen, Lei Sha, Sujian Li, Xu Sun, Houfeng Wang, Lintao Zhang
    In Proceedings of EMNLP 2018 Conference, 2018. (EMNLP 2018).
    [PDF]
  40. Table-to-text Generation by Structure-aware Seq2seq learning
    Tianyu Liu, Kexiang Wang, Lei Sha, Zhifang Sui, Baobao Chang
    In Proceedings of AAAI 2018 Conference, 2018. (AAAI 2018).
    [PDF]
  41. Topic Medical Concept Embedding: Multi-Sense Representation Learning For Medical Concept
    Feng Qian, Chengyue Gong, Luchen Liu, Lei Sha, Ming Zhang
    In Proceedings of BIBM 2017 conference, 2017. (BIBM 2017).
  42. Syntax Aware LSTM Model for Chinese Semantic Role Labeling
    Feng Qian, Lei Sha, Baobao Chang, Lu-chen Liu, Ming Zhang
    In Proceedings of EMNLP 2017 Workshop, 2017. (EMNLP 2017 (workshop)). [PDF]
  43. A Progressive Learning Approach to Chinese SRL Using Heterogeneous Data
    Qiaolin Xia, Lei Sha, Baobao Chang, Zhifang Sui
    In Proceedings of ACL 2017 Conference, 2017. (ACL 2017). [PDF]
  44. Attentive Interactive Neural Networks for Answer Selection in Community Question Answering
    Xiaodong Zhang, Lei Sha, Sujian Li, Houfeng Wang
    In Proceedings of AAAI 2017 Conference, 2017. (AAAI 2017). [PDF]
  45. Encoding Temporal Information for Time-Aware Link Prediction
    Tingsong Jiang, Tianyu Liu, Tao Ge, Lei Sha, Sujian Li, Baobao Chang, Zhifang Sui.
    In Proceedings of EMNLP 2016 Conference, 2016. (EMNLP 2016). [PDF]
  46. Towards Time-Aware Knowledge Graph Completion
    Tingsong Jiang, Tianyu Liu, Tao Ge, Lei Sha, Sujian Li, Baobao Chang, Zhifang Sui
    In Proceedings of Coling 2016 Conference, 2016. (Coling 2016). [PDF]
  47. Multi-label Text Categorization with Joint Learning Predictions-as-Features Method
    Li Li, Houfeng Wang, Lei Sha, Xu Sun, Baobao Chang, Shi Zhao.
    In Proceedings of EMNLP 2015 Conference, 2015. (EMNLP 2015). [PDF]
  48. Event Schema Induction Based on Relational Co-occurrence over Multiple Documents
    Tingsong Jiang, Lei Sha, Zhifang Sui
    In Proceedings of Natural Language Processing and Chinese Computing, 2014. (NLPCC 2014).
Cooperation

Real-World AI Deployments

Any organization interested in these technologies is welcome to contact me; My group is able to provide bespoke technical development or deliver fully integrated solutions.

Automated Legal Judgment Robot

Experience the next-generation AI legal assistant that delivers rapid, ultra-precise judgments by analyzing millions of legal precedents in seconds. Designed for high-stakes corporate and international law environments, it provides transparent reasoning, risk assessment, and optimized decision support—perfect for enterprises seeking to reduce legal uncertainty and maximize efficiency.

Automated Legal Judgment Robot Overview

AI Medical Record Quality Control

Unlock the future of premium healthcare management with an AI-powered medical record quality control system. Seamlessly validate, standardize, and optimize patient records in real time, ensuring compliance, eliminating errors, and enhancing clinical decision-making. Perfect for luxury hospital chains and elite healthcare providers seeking unparalleled operational precision and patient safety.

AI Medical Record QC Overview

AI4S Intelligent Lab Management

Revolutionize laboratory operations with AI4S, a state-of-the-art intelligent lab management platform. Optimize experiments, dynamically schedule resources, and orchestrate personnel, instruments, and consumables in real time. Designed for elite research institutions and luxury biotech ventures, AI4S ensures peak efficiency, precision-driven experimentation, and unparalleled oversight across your entire scientific workflow.

AI4S Intelligent Lab Management Overview

Scientific Model Library

Discover the ultimate research powerhouse with BioFord™, an integrated platform hosting a vast scientific model library powered by GeneLLM. Accelerate experimental design, execution, and data analysis across hundreds of scientific domains—from CRISPR prediction to 3D RNA modeling and protein structure forecasting. Elite institutions and high-end biotech ventures can leverage AI agents for end-to-end experiment orchestration, ensuring unmatched precision, efficiency, and insight across the full research lifecycle.

BioFord Scientific Model Library Overview