Publications

2026

1. Preprint

   Data-driven Circuit Discovery for Interpretability of Language Models

   Daking Rai*, Mor Geva, and Ziyu Yao

   arXiv preprint arXiv:2605.09129, 2026

   PDF X
2. TOSEM’26

   Guiding AI to Fix Its Own Flaws: An Empirical Study on LLM-Driven Secure Code Generation

   Hao Yan*, Swapneel Suhas Vaidya, Xiaokuan Zhang, and Ziyu Yao

   ACM Trans. Softw. Eng. Methodol., Sep 2026

   PDF
3. EMNLP’26

   Interpreting and Steering for Safe and Correct Code Generation

   Hao Yan, and Ziyu Yao

   arXiv preprint arXiv:2608.30025 (To appear at EMNLP Main), Sep 2026

   PDF Code Website
4. EMNLP’26 Findings

   Can Language Model Agents be Helpful Circuit Explainers in Mechanistic Interpretability?

   Ayan Antik Khan*, Harsh Kohli, Yuekun Yao, Huan Sun, and Ziyu Yao

   arXiv preprint arXiv:2606.24026 (To appear at EMNLP Findings), Sep 2026

   PDF X
5. TIST’26

   A Practical Review of Mechanistic Interpretability for Transformer-Based Language Models

   Daking Rai*, Yilun Zhou, Shi Feng, Abulhair Saparov, and Ziyu Yao

   To appear at ACM Transactions on Intelligent Systems and Technology, Sep 2026

   PDF Website
6. TMLR’26

   Lens: A Knowledge-Guided Foundation Model for Network Traffic

   Xiaochang Li, Chen Qian, Qineng Wang, Jiangtao Kong, Yuchen Wang, Ziyu Yao, and 4 more authors

   Transactions on Machine Learning Research, Sep 2026

   PDF
7. CVPR’26

   Inside-Out: Measuring Generalization in Vision Transformers Through Inner Workings

   (CVPR 2026 Highlight (Top 3%))

   Yunxiang Peng, Mengmeng Ma, Ziyu Yao, and Xi Peng

   IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Sep 2026

   PDF X
8. ACL’26

   Why Do LLM-based Web Agents Fail? A Hierarchical Planning Perspective

   Mohamed Aghzal*, Gregory J Stein, and Ziyu Yao

   arXiv preprint arXiv:2603.14248 (To appear at ACL 2026), Sep 2026

   PDF X
9. CHI’26

   Designing AI Peers for Collaborative Mathematical Problem Solving with Middle School Students: A Participatory Design Study

   Wenhan Lyu, Yimeng Wang, Murong Yue*, Yifan Sun, Jennifer Suh, Meredith Kier, and 2 more authors

   arXiv preprint arXiv:2601.17962 (To appear at ACM CHI 2026), Sep 2026

   PDF
10. TOSEM’26

    Reassessing Code Authorship Attribution in the Era of Language Models

    Atish Kumar Dipongkor, Ziyu Yao, and Kevin Moran

    Accepted to ACM TOSEM, Sep 2026

    PDF
11. ACL-BEA’26

    PeerMathDial: A Middle School Dialogue Dataset for Student Collaborative Math Problem Solving

    Accepted to the 21st Workshop on Innovative Use of NLP for Building Educational Applications (BEA) at ACL, Sep 2026

    PDF

2025

1. Preprint

   Revisiting Prompt Optimization with Large Reasoning Models-A Case Study on Event Extraction

   Saurabh Srivastava*, and Ziyu Yao

   arXiv preprint arXiv:2504.07357, Sep 2025

   PDF
2. Preprint

   A Survey on Large Language Models for Automated Planning

   Mohamed Aghzal*, Erion Plaku, Gregory J Stein, and Ziyu Yao

   arXiv preprint arXiv:2502.12435, Sep 2025

   PDF
3. ICDM’25W

   Evaluating the Effectiveness of Persona Simulation in Opinion Prediction with GPT-4.1

   Sarah Li*, and Ziyu Yao

   In 2025 IEEE International Conference on Data Mining Workshops (ICDMW - Undergraduate and Graduate Honor Symposium), Sep 2025

   PDF
4. NeurIPS’25

   Failure by Interference: Language Models Make Balanced Parentheses Errors When Faulty Mechanisms Overshadow Sound Ones

   Daking Rai*, Samuel Miller*, Kevin Moran, and Ziyu Yao

   arXiv preprint arXiv:2507.00322 (To Appear at NeurIPS 2025), Sep 2025

   PDF X
5. EMNLP’25

   Feature Extraction and Steering for Enhanced Chain-of-Thought Reasoning in Language Models

   Zihao Li, Xu Wang, Yuzhe Yang, Ziyu Yao, Haoyi Xiong, and Mengnan Du

   arXiv preprint arXiv:2505.15634 (to appear at EMNLP 2025 Main), Sep 2025

   PDF
6. EMNLP’25

   All for One: LLMs Solve Mental Math at the Last Token With Information Transferred From Other Tokens

   (Covered by Rohan Paul (AI Influencer at X) and “AI: post transformers” Podcast)

   Siddarth Mamidanna*, Daking Rai*, Ziyu Yao, and Yilun Zhou

   To appear at EMNLP 2025 Main, Sep 2025

   PDF
7. EMNLP’25 Findings

   A survey on sparse autoencoders: Interpreting the internal mechanisms of large language models

   Dong Shu, Xuansheng Wu, Haiyan Zhao, Daking Rai*, Ziyu Yao, Ninghao Liu, and 1 more author

   arXiv preprint arXiv:2503.05613 (to appear at EMNLP 2025 Findings), Sep 2025

   PDF
8. EMNLP’25 Findings

   Beneath the Surface: How Large Language Models Reflect Hidden Bias

   Jinhao Pan, Chahat Raj, Ziyu Yao, and Ziwei Zhu

   arXiv preprint arXiv:2502.19749 (to appear at EMNLP 2025 Findings), Sep 2025

   PDF
9. COLM’25W

   Can LLMs Simulate Personas with Reversed Performance? A Benchmark for Counterfactual Instruction Following

   Sai Adith Senthil Kumar^*, Hao Yan^*, Saipavan Perepa*, Murong Yue*, and Ziyu Yao

   COLM Workshop on Social Simulation with LLMs, Sep 2025

   PDF
10. ACL’25 Findings

    Efficient but Vulnerable: Benchmarking and Defending LLM Batch Prompting Attack

    Murong Yue*, and Ziyu Yao

    Findings of ACL 2025, Sep 2025

    PDF Code
11. ACL’25 Findings

    Instruction-Tuning LLMs for Event Extraction with Annotation Guidelines

    Saurabh Srivastava^*, Sweta Pati^*, and Ziyu Yao

    Findings of ACL 2025, Sep 2025

    PDF Code
12. IROS’25

    Autospatial: Visual-language reasoning for social robot navigation through efficient spatial reasoning learning

    Yangzhe Kong, Daeun Song, Jing Liang, Dinesh Manocha, Ziyu Yao, and Xuesu Xiao

    arXiv preprint arXiv:2503.07557 (to appear at IROS 2025), Sep 2025

    PDF
13. AAAI’25W

    Mechanistic Understanding of Language Models in Syntactic Code Completion

    Samuel Miller^*, Daking Rai^*, and Ziyu Yao

    AAAI Workshop on Towards Knowledgeable Foundation Models, Sep 2025

    PDF
14. CVPR’25

    Evaluating Vision-Language Models as Evaluators in Path Planning

    Mohamed Aghzal*, Xiang Yue, Erion Plaku, and Ziyu Yao

    IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Sep 2025

    PDF Code
15. ICLR’25

    DOTS: Learning to Reason Dynamically in LLMs via Optimal Reasoning Trajectories Search

    (Covered by MIT Tech Review China)

    Murong Yue*, Wenlin Yao, Haitao Mi, Dian Yu, Ziyu Yao, and Dong Yu

    The Thirteenth International Conference on Learning Representations, Sep 2025

    PDF
16. AAAI’25W

    MathVC: An LLM-Simulated Multi-Character Virtual Classroom for Mathematics Education

    (Invited Presentation at Wolfram Research LLM Agent Colloquium; media coverage by GovTech and The George)

    Murong Yue^*, Wenhan Lyu^, Wijdane Mifdal*, Yixuan Zhang, Jennifer Suh, and Ziyu Yao

    AAAI AI4Edu Workshop, Sep 2025

    PDF Website

2024

1. Preprint

   Understanding the Effect of Algorithm Transparency of Model Explanations in Text-to-SQL Semantic Parsing

   Daking Rai*, Rydia R Weiland, Kayla Margaret Gabriella Herrera, Tyler H Shaw, and Ziyu Yao

   arXiv preprint arXiv:2410.16283, Sep 2024

   PDF
2. Preprint

   IntelliExplain: Enhancing Conversational Code Generation for Non-Professional Programmers

   Hao Yan*, Thomas D. Latoza, and Ziyu Yao

   arXiv Preprint, Sep 2024

   PDF Code Website
3. EMNLP’24

   Navigating the Shortcut Maze: A Comprehensive Analysis of Shortcut Learning in Text Classification by Language Models

   Yuqing Zhou, Ruixiang Tang, Ziyu Yao, and Ziwei Zhu

   Findings of EMNLP, Sep 2024

   PDF Code
4. CASE’24

   Look Further Ahead: Testing the Limits of GPT-4 in Path Planning

   Mohamed Aghzal*, Erion Plaku, and Ziyu Yao

   IEEE CASE 2024 (also present at AAAI 2025 LM4Plan Workshop), Sep 2024

   PDF Code
5. ACL’24

   An Investigation of Neuron Activation as a Unified Lens to Explain Chain-of-Thought Eliciting Arithmetic Reasoning of LLMs

   (Covered by MIT Technology Review China [English Translate])

   Daking Rai*, and Ziyu Yao

   ACL, Sep 2024

   PDF Code Website
6. ACL’24

   Instances Need More Care: Rewriting Prompts for Instances with LLMs in the Loop Yields Better Zero-Shot Performance

   Saurabh Srivastava*, Chengyue Huang, Weiguo Fan, and Ziyu Yao

   Findings of ACL, Sep 2024

   PDF Code
7. ICLR’24W

   Can Large Language Models be Good Path Planners? A Benchmark and Investigation on Spatial-temporal Reasoning

   Mohamed Aghzal*, Erion Plaku, and Ziyu Yao

   ICLR Workshop on LLM Agents, Sep 2024

   PDF Code
8. Preprint

   Lens: A Foundation Model for Network Traffic

   Qineng Wang, Chen Qian, Xiaochang Li, Ziyu Yao, and Huajie Shao

   arXiv Preprint, Sep 2024

   PDF
9. ICLR’24

   Large Language Model Cascades with Mixture of Thoughts Representations for Cost-efficient Reasoning

   (Featured in Hugging Face Daily Papers and AWS Builder Blog)

   Murong Yue*, Jie Zhao, Min Zhang, Liang Du, and Ziyu Yao

   The Twelfth International Conference on Learning Representations (also at ICLR Workshop on Reliable and Responsible Foundation Models), Sep 2024

   PDF Code

2023

1. EMNLP’23 Demo

   Gentopia: A Collaborative Platform for Tool-Augmented LLMs

   (An open-source planform for creating, evaluating, and community-sharing Augmented Language Model (ALM)-based Agents)

   Binfeng Xu, Xukun Liu, Hua Shen, Zeyu Han, Yuhan Li, Murong Yue*, and 4 more authors

   EMNLP’23 System Demonstration, Sep 2023

   PDF Code
2. EMNLP’23

   MAILEX: Email Event and Argument Extraction

   Saurabh Srivastava*, Gaurav Singh*, Shou Matsumoto, Ali Raz, Paulo Costa, Joshua Poore, and 1 more author

   EMNLP’23, Sep 2023

   PDF Code
3. ACL’23

   Improving Generalization in Language Model-based Text-to-SQL Semantic Parsing: Two Simple Semantic Boundary-based Techniques

   (Rank#8 on Spider leaderboard as of Aug 2023)

   Daking Rai*, Bailin Wang, Yilun Zhou, and Ziyu Yao

   In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), Sep 2023

   PDF Code
4. ACL’23

   Learning to Simulate Natural Language Feedback for Interactive Semantic Parsing

   Hao Yan*, Saurabh Srivastava*, Yintao Tai*, Sida I. Wang, Wen-tau Yih, and Ziyu Yao

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

   PDF Code
5. AAAI’23 SA

   Explaining Large Language Model-Based Neural Semantic Parsers (Student Abstract)

   Daking Rai*, Yilun Zhou, Bailin Wang, and Ziyu Yao

   AAAI Student Abstract, Sep 2023

   PDF
6. JBP’23

   A paradigm shift from “human writing” to “machine generation” in personality test development: An application of state-of-the-art natural language processing

   (Editor Commendation, one of 13 out of 1,000+ submissions in 2022)

   Philseok Lee, Shea Fyffe, Mina Son, Zihao Jia, and Ziyu Yao

   Journal of Business and Psychology, Sep 2023

   PDF

2022

1. EMNLP’22

   UnifiedSKG: Unifying and Multi-Tasking Structured Knowledge Grounding with Text-to-Text Language Models

   Tianbao Xie^, Chen Henry Wu^, Peng Shi, Ruiqi Zhong, Torsten Scholak, Michihiro Yasunaga, and 17 more authors

   In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, Sep 2022

   PDF Code Website
2. ICLR’22 DL4Code

   Code Editing from Few Exemplars by Adaptive Multi-Extent Composition

   Peizhao Li, Xuchao Zhang, Ziyu Yao, Wei Cheng, Haifeng Chen, and Hongfu Liu

   In Deep Learning for Code Workshop at International Conference on Learning Representations, Sep 2022

   PDF
3. ACL’22

   Synthetic Question Value Estimation for Domain Adaptation of Question Answering

   Xiang Yue, Ziyu Yao, and Huan Sun

   In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Sep 2022

   PDF Code

2021

1. Dissertation

   On Advancing Natural Language Interfaces: Data Collection, Model Development, and User Interaction

   Ziyu Yao

   Sep 2021

   PDF
2. ICLR’21

   Learning Structural Edits via Incremental Tree Transformations

   Ziyu Yao, Frank F. Xu, Pengcheng Yin, Huan Sun, and Graham Neubig

   In International Conference on Learning Representations, Sep 2021

   PDF Code
3. BIBM’21

   Cliniqg4qa: Generating diverse questions for domain adaptation of clinical question answering

   (Best Paper Award)

   Xiang Yue, Xinliang Frederick Zhang, Ziyu Yao, Simon Lin, and Huan Sun

   In 2021 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), Sep 2021

   PDF Code

2020

1. EMNLP’20

   An Imitation Game for Learning Semantic Parsers from User Interaction

   Ziyu Yao, Yiqi Tang, Wen-tau Yih, Huan Sun, and Yu Su

   In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), Sep 2020

   PDF Code Slides

2019

1. EMNLP-IJCNLP’19

   Model-based Interactive Semantic Parsing: A Unified Framework and A Text-to-SQL Case Study

   Ziyu Yao, Yu Su, Huan Sun, and Wen-tau Yih

   In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), Sep 2019

   PDF Code
2. ACL’19

   Reinforced Dynamic Reasoning for Conversational Question Generation

   Boyuan Pan, Hao Li, Ziyu Yao, Deng Cai, and Huan Sun

   In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, Sep 2019

   PDF Code
3. WWW’19

   CoaCor: Code annotation for code retrieval with reinforcement learning

   Ziyu Yao, Jayavardhan Reddy Peddamail, and Huan Sun

   In The World Wide Web Conference, Sep 2019

   PDF Code
4. AAAI’19

   Interactive semantic parsing for if-then recipes via hierarchical reinforcement learning

   Ziyu Yao, Xiujun Li, Jianfeng Gao, Brian Sadler, and Huan Sun

   In Proceedings of the AAAI Conference on Artificial Intelligence, Sep 2019

   PDF Code Poster Slides

2018

1. KDD’18 DL Day

   A comprehensive study of staqc for deep code summarization

   Jayavardhan Reddy Peddamail, Ziyu Yao, Zhen Wang, and Huan Sun

   In Deep Learning Day at KDD, Sep 2018

   PDF
2. WWW’18

   Staqc: A systematically mined question-code dataset from stack overflow

   Ziyu Yao, Daniel S Weld, Wei-Peng Chen, and Huan Sun

   In Proceedings of the 2018 World Wide Web Conference, Sep 2018

   PDF Code Slides

2016

1. AAAI’16

   Semi-supervised multinomial naive bayes for text classification by leveraging word-level statistical constraint

   Li Zhao, Minlie Huang, Ziyu Yao, Rongwei Su, Yingying Jiang, and Xiaoyan Zhu

   In Proceedings of the AAAI Conference on Artificial Intelligence, Sep 2016

   PDF