publications

2026

1. Dr. Zero: Self-Evolving Search Agents without Training Data

   Zhenrui Yue, Kartikeya Upasani, Xianjun Yang, Suyu Ge, Shaoliang Nie, Yuning Mao, and 2 more authors

   arXiv preprint arXiv:2601.07055, 2026
2. Learning Personalized Agents from Human Feedback

   Kaiqu Liang, Julia Kruk, Shengyi Qian, Xianjun Yang, Shengjie Bi, Yuanshun Yao, and 6 more authors

   arXiv preprint arXiv:2602.16173, 2026
3. Verifying Chain-of-Thought Reasoning via Its Computational Graph

   Zheng Zhao, Yeskendir Koishekenov, Xianjun Yang, Naila Murray, and Nicola Cancedda

   ICLR 2026 (Oral), 2026

2025

1. Weak-to-strong jailbreaking on large language models

   X Zhao, Xianjun Yang, T Pang, C Du, L Li, YX Wang, and 1 more author

   In International Conference on Machine Learning (ICML), 2025
2. AILuminate: Introducing v1.0 of the AI Risk and Reliability Benchmark from MLCommons

   Shaona Ghosh, Heather Frase, Adina Williams, Sarah Luger, Paul Röttger, Xianjun Yang, and 1 more author

   arXiv preprint arXiv:2503.05731, 2025
3. MELON: Provable Defense Against Indirect Prompt Injection Attacks in AI Agents

   Kaijie Zhu, Jindong Wang, Wenbo Guo, Xianjun Yang, and William Yang Wang

   arXiv preprint arXiv:2502.05174, 2025
4. Diversity-driven Data Selection for Language Model Tuning through Sparse Autoencoder

   Xianjun Yang, Shaoliang Nie, Lijuan Liu, Suchin Gururangan, Ujjwal Karn, Rui Hou, and 2 more authors

   arXiv preprint arXiv:2502.14050, 2025
5. Many-Turn Jailbreaking

   Liqiang Xiao, Shiyang Li, Faisal Ladhak, Hyokun Yun, Xianjun Yang, Linda Ruth Petzold, and 2 more authors

   arXiv preprint arXiv:2508.06755, 2025
6. Your Thoughts Tell Who You Are: Characterize the Reasoning Patterns of LRMs

   Yida Chen, Yuning Mao, Xianjun Yang, Suyu Ge, Shengjie Bi, Lijuan Liu, and 4 more authors

   arXiv preprint arXiv:2509.24147, 2025
7. Humanity’s last exam

   L Phan, A Gatti, Z Han, N Li, J Hu, H Zhang, and 3 more authors

   arXiv preprint arXiv:2501.14249, 2025
8. Beyond Reasoning Gains: Mitigating General Capabilities Forgetting in Large Reasoning Models

   Hoang Phan, Xianjun Yang, Kevin Yao, Jingyu Zhang, Shengjie Bi, Xiaocheng Tang, and 3 more authors

   arXiv preprint arXiv:2510.21978, 2025

2024

1. A survey on large language models for critical societal domains: Finance, healthcare, and law

   ZZ Chen, J Ma, X Zhang, N Hao, A Yan, A Nourbakhsh, and 2 more authors

   arXiv preprint arXiv:2405.01769, 2024
2. Introducing v0.5 of the ai safety benchmark from mlcommons

   B Vidgen, A Agrawal, AM Ahmed, V Akinwande, N Al-Nuaimi, N Alfaraj, and 1 more author

   arXiv preprint arXiv:2404.12241, 2024
3. A safe harbor for ai evaluation and red teaming

   S Longpre, S Kapoor, K Klyman, A Ramaswami, R Bommasani, and  others

   In International Conference on Machine Learning (ICML), 2024

   Oral
4. Test-time backdoor attacks on multimodal large language models

   D Lu, T Pang, C Du, Q Liu, Xianjun Yang, and M Lin

   arXiv preprint arXiv:2402.08577, 2024
5. TrustAgent: Towards Safe and Trustworthy LLM-based Agents through Agent Constitution

   W Hua, Xianjun Yang, Z Li, C Wei, and Y Zhang

   In Findings of the Association for Computational Linguistics: EMNLP 2024, 2024
6. Navigating the OverKill in Large Language Models

   C Shi, X Wang, Q Ge, S Gao, Xianjun Yang, T Gui, and 4 more authors

   In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL), 2024
7. Pllama: An open-source large language model for plant science

   Xianjun Yang, J Gao, W Xue, and E Alexandersson

   arXiv preprint arXiv:2401.01600, 2024
8. MultiAgent Collaboration Attack: Investigating Adversarial Attacks in Large Language Model Collaborations via Debate

   A Amayuelas, Xianjun Yang, A Antoniades, W Hua, L Pan, and W Wang

   In Findings of the Association for Computational Linguistics: EMNLP 2024, 2024
9. Trustagent: Towards safe and trustworthy llm-based agents

   W Hua, Xianjun Yang, M Jin, Z Li, W Cheng, R Tang, and 1 more author

   arXiv preprint arXiv:2402.01586, 2024
10. A survey on detection of llms-generated content

    Xianjun Yang, L Pan, X Zhao, H Chen, L Petzold, WY Wang, and 1 more author

    In Findings of the Association for Computational Linguistics: EMNLP 2024, 2024
11. Large language models can be good privacy protection learners

    Y Xiao, Y Jin, Y Bai, Y Wu, Xianjun Yang, X Luo, and 5 more authors

    In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), 2024
12. Quokka: An Open-source Large Language Model Chatbot for Materials Science

    Xianjun Yang, Stephen D Wilson, and Linda Petzold

    arXiv preprint arXiv:2401.01089, 2024
13. Unveiling the Misuse Potential of Base Large Language Models via In-Context Learning

    Xiao Wang, Xianjun Yang, Tianze Chen, Qi Zhang, Xun Zhao, and Dahua Lin

    arXiv preprint arXiv:2404.10552, 2024
14. Unveiling the Impact of Coding Data Instruction Fine-Tuning on Large Language Models Reasoning

    Xinlu Zhang, Zhiyu Zoey Chen, Xi Ye, Lichang Chen, Xianjun Yang, William Yang Wang, and 1 more author

    arXiv preprint arXiv:2405.20535, 2024
15. DALD: Improving Logits-based Detector without Logits from Black-box LLMs

    Cong Zeng, Shengkun Tang, Yuanzhou Chen, Yiyou Sun, Zhiqiang Xu, Yao Li, and 4 more authors

    arXiv preprint arXiv:2406.05232, 2024
16. MMSci: A Dataset for Graduate-Level Multi-Discipline Multimodal Scientific Understanding

    Zekun Li, Kyuri Choi, Wanrong Zhu, Ryan Hsieh, HyeonJung Kim, Jin Hyuk Lim, and 8 more authors

    In Proceedings of the 38th Conference on Neural Information Processing Systems (NeurIPS) Datasets and Benchmarks Track, 2024
17. CBT-Bench: Evaluating Large Language Models on Assisting Cognitive Behavior Therapy

    Mian Zhang, Xinlu Zhang, Travis Labrum, Jamie C Chiu, Shaun M Eack, Fei Fang, and 3 more authors

    arXiv preprint arXiv:2410.13218, 2024

2023

1. TRACE: A Comprehensive Benchmark for Continual Learning in Large Language Models

   Xiao Wang, Yuansen Zhang, Tianze Chen, Songyang Gao, Senjie Jin, Xianjun Yang, and 6 more authors

   In , 2023
2. Shadow Alignment: The Ease of Subverting Safely-Aligned Language Models

   Xianjun Yang, Xiao Wang, Qi Zhang, Linda Petzold, William Yang Wang, Xun Zhao, and 1 more author

   In , 2023
3. Zero-Shot Detection of Machine-Generated Codes

   Xianjun Yang, Kexun Zhang, Haifeng Chen, Linda Petzold, William Yang Wang, and Wei Cheng

   arXiv preprint arXiv:2310.05103, 2023
4. Large Language Models Can Be Good Privacy Protection Learners

   Yijia Xiao, Yiqiao Jin, Yushi Bai, Yue Wu, Xianjun Yang, Xiao Luo, and 5 more authors

   arXiv preprint arXiv:2310.02469, 2023
5. DNA-GPT: Divergent N-Gram Analysis for Training-Free Detection of GPT-Generated Text

   Xianjun Yang, Wei Cheng, Linda Petzold, William Yang Wang, and Haifeng Chen

   arXiv preprint arXiv:2305.17359, 2023
6. Enhancing Small Medical Learners with Privacy-preserving Contextual Prompting

   Xinlu Zhang, Shiyang Li, Xianjun Yang, Chenxin Tian, Yao Qin, and Linda Ruth Petzold

   arXiv preprint arXiv:2305.12723, 2023
7. LLMScore: Unveiling the Power of Large Language Models in Text-to-Image Synthesis Evaluation

   Yujie Lu, Xianjun Yang, Xiujun Li, Xin Eric Wang, and William Yang Wang

   arXiv preprint arXiv:2305.11116, 2023
8. Dynamic Prompting: A Unified Framework for Prompt Tuning

   Xianjun Yang, Wei Cheng, Xujiang Zhao, Linda Petzold, and Haifeng Chen

   arXiv preprint arXiv:2303.02909, 2023
9. Exploring the limits of chatgpt for query or aspect-based text summarization

   Xianjun Yang, Yan Li, Xinlu Zhang, Haifeng Chen, and Wei Cheng

   arXiv preprint arXiv:2302.08081, 2023
10. MatKB: Semantic Search for Polycrystalline Materials Synthesis Procedures

    Xianjun Yang, Stephen Wilson, and Linda Petzold

    arXiv preprint arXiv:2302.05597, 2023
11. ReDi: Efficient Learning-Free Diffusion Inference via Trajectory Retrieval

    Kexun Zhang, Xianjun Yang, William Yang Wang, and Lei Li

    arXiv preprint arXiv:2302.02285, 2023
12. OASum: Large-Scale Open Domain Aspect-based Summarization

    Xianjun Yang, Kaiqiang Song, Sangwoo Cho, Xiaoyang Wang, Xiaoman Pan, Linda Petzold, and 1 more author

    In Findings of the Association for Computational Linguistics: ACL 2023, Jul 2023

    Abs

    Aspect or query-based summarization has recently caught more attention, as it can generate differentiated summaries based on users’ interests. However, the current dataset for aspect or query-based summarization either focuses on specific domains, on a relatively small scale, or contains only a few aspect types. Such limitations hinder further explorations in this direction. In this work, we take advantage of crowd-sourcing knowledge on Wikipedia and automatically create a high-quality, large-scale open-domain aspect-based summarization dataset named OASum, which contains more than 3.7 million instances with around 1 million different aspects on 2 million Wikipedia pages. We provide benchmark results on OASum and demonstrate its ability for diverse aspect-based summarization generation. To overcome the data scarcity problem on specific domains, we also perform zero-shot, few-shot, and fine-tuning on seven downstream datasets. Specifically, zero/few-shot and fine-tuning results show that the model pre-trained on our corpus demonstrates a strong aspect or query-focused generation ability compared with the backbone model. Our dataset and pre-trained checkpoints are publicly available.

13. Few-Shot Document-Level Event Argument Extraction

    Xianjun Yang, Yujie Lu, and Linda Petzold

    In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Jul 2023

    Abs

    Event argument extraction (EAE) has been well studied at the sentence level but under-explored at the document level. In this paper, we study to capture event arguments that actually spread across sentences in documents. Prior works usually assume full access to rich document supervision, ignoring the fact that the available argument annotation is limited in production. To fill this gap, we present FewDocAE, a Few-Shot Document-Level Event Argument Extraction benchmark, based on the existing document-level event extraction dataset. We first define the new problem and reconstruct the corpus by a novel N-Way-D-Doc sampling instead of the traditional N-Way-K-Shot strategy. Then we adjust the current document-level neural models into the few-shot setting to provide baseline results under in- and cross-domain settings. Since the argument extraction depends on the context from multiple sentences and the learning process is limited to very few examples, we find this novel task to be very challenging with substantively low performance. Considering FewDocAE is closely related to practical use under low-resource regimes, we hope this benchmark encourages more research in this direction. Our data and codes will be available online.

14. Alpacare: Instruction-tuned large language models for medical application

    X Zhang, C Tian, Xianjun Yang, L Chen, Z Li, and LR Petzold

    arXiv preprint arXiv:2310.14558, Jul 2023

2022

1. PcMSP: A Dataset for Scientific Action Graphs Extraction from Polycrystalline Materials Synthesis Procedure Text

   Xianjun Yang, Ya Zhuo, Julia Zuo, Xinlu Zhang, Stephen Wilson, and Linda Petzold

   In Findings of the Association for Computational Linguistics: EMNLP 2022, Jul 2022

2021

1. An Analysis of Relation Extraction within Sentences from Wet Lab Protocols

   Xianjun Yang, Xinlu Zhang, Julia Zuo, Stephen Wilson, and Linda Petzold

   In 2021 IEEE International Conference on Big Data (Big Data), Jul 2021
2. On explosive boiling of a multicomponent Leidenfrost drop

   Sijia Lyu, Huanshu Tan, Yuki Wakata, Xianjun Yang, Chung K Law, Detlef Lohse, and 1 more author

   Proceedings of the National Academy of Sciences, Jul 2021

2019

1. Convective heat transfer along ratchet surfaces in vertical natural convection

   Hechuan Jiang, Xiaojue Zhu, Varghese Mathai, Xianjun Yang, Roberto Verzicco, Detlef Lohse, and 1 more author

   Journal of fluid mechanics, Jul 2019