Hi, nice to e-meet you! I am a first-year PhD student at Ohio State University(OSU), advised by Prof. Huan Sun and working closely with Prof. Yu Su. I obtained my master’s degree from Nanjing University. Previously, I was also a research intern at Blender NLP, UIUC, working with Prof. Heng Ji.

My research interests primarily lie in agents and reasoning. Specifically, I am interested in improving and developing powerful web agents that can interact with online environments to effectively solve various real-world tasks, as well as enhancing reasoning abilities such as R1-style reasoning and long Chain-of-Thought.

🔥 News

• 2025.3: 🎉 Check out our new work Online-Mind2Web
• 2024.5: Our paper “PACIT: Unlocking the Power of Examples for Better In-Context Instruction Tuning” is accepted by ACL 2024 Findings
• 2024.4: Joining the OSU NLP Group!

📝 Publications

arXiv preprint

An Illusion of Progress? Assessing the Current State of WebAgents
Tianci Xue, Weijian Qi, Tianneng Shi, Chan Hee Song, Boyu Gou, Dawn Song, Huan Sun, Yu Su

• Online-Mind2Web, an online evaluation benchmark consisting of 300 diverse and realistic tasks spanning 136 websites. We conduct the first comprehensive comparative analysis of current web agents under a setting that approximates how real users use these agents.

📃 Blog | 🏆 Leaderboard | 🤗 Data | 🕊️ Twitter

ACL

PACIT: Unlocking the Power of Examples for Better In-Context Instruction Tuning
Tianci Xue, Ziqi Wang, Yixia Li, Yun Chen, Guanhua Chen

• PACIT: a simple and effective in-context instruction tuning method, first verifies the provided example’s correctness, then uses this to generate a better response.

arXiv preprint

Parameter-Efficient Tuning Helps Language Model Alignment
Tianci Xue, Ziqi Wang, Heng Ji

• MEET, a novel approach that incorporates parameter-efficient tuning to better optimize control tokens, thus benefitting controllable generation.

arXiv preprint

RCOT: Detecting and Rectifying Factual Inconsistency in Reasoning by Reversing Chain-of-Thought
Tianci Xue, Ziqi Wang, Zhenhailong Wang, Chi Han, Pengfei Yu, Heng Ji

• RCoT, a novel method to improve LLMs’ reasoning abilities by automatically detecting and rectifying factual inconsistency in LLMs’ generated solutions.

🎖 Honors and Awards

• 26/2654, Top 1%, in Kaggle Feedback Prize - English Language Learning 2022.12
• The Second Prize, Top 2%, in China Undergraduate Mathematical Contest in Modeling 2021.11
• The Second Prize, Top 5% in The National University Students Intelligent Car Race 2021.7
• The Third Prize in Chinese Collegiate Computing Competition 2021.6
• The First Prize Scholarship of Hubei University of Technology 2022
• Outstanding Graduates of the School of Computer Science, Hubei University of Technology 2022

📖 Educations

• 2024.09 - Present, Ph.D., The Ohio State University(OSU).
• 2022.09 - 2024.06, Master, Nanjing University. Rank: 29/231
• 2018.09 - 2022.06, Undergraduate, Hubei University of Technology. Rank: 1/178

📌 Service

• Review Assistant: Neurips 2023

💻 Internships

• 2023.03 - Present, Intern at BLENDER NLP, UIUC.
• 2023.07 - Present, Research Assistant at SUSTECH.