Publication

• Xingcheng Xu, Zibo Zhao, Haipeng Zhang, and Yanqing Yang, (2025). "Principled Understanding of Generalization for Generative Transformer Models in Arithmetic Reasoning Tasks.". In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 4721–4747, Vienna, Austria. Association for Computational Linguistics.

  • Preprint available at arxiv.

  • available at aclanthology.

  • Press Coverage: MIT Tech Review China, DeepTech.

• Peiwen Jiang, Xinbo Lin, Zibo Zhao, Ruhui Ma, Yvonne Jie Chen, and Jinhua Cheng, (2024). "TKGT: Redefinition and A New Way of Text-to-Table Tasks Based on Real World Demands and Knowledge Graphs Augmented LLMs". Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing

  • available at aclanthology.

Working Paper

• "The Two-Stage Decision-Sampling Hypothesis: Understanding the Emergence of Self-Reflection in RL-Trained LLM" with Haipeng Zhang, Yuanting Zha, and Xingcheng Xu, Working Paper, Jan. 2026.

• "Better Technology, Worse Motivation: GenAI's Mediocrity Trap" with Yvonne Jie Chen, Jie Gong, and Jin Li, Working Paper, Mar.2025. R&R at Management Science.

  • Press Coverage: The Economist 

• "Measuring Labor Market Match Quality with Large Language Models" with Yi Chen, Hanming Fang, and Yi Zhao, Working Paper, Nov. 2024.

  • Previous NBER WP version at here. 

Selected Working In Progress

• "Skill Accumulation and Wage Dynamics". (with Yi Chen, Hanming Fang, and Yi Zhao)

• "How Do Go Players Make Better Decision After AI-existence: The Role of Learning". (with Yvonne Jie Chen, Xingcheng Xu, Yanqing Yang, and Songfa Zhong)

  • Presented at CCF Computing Economics Annual Conference (Professional Session) 2024.

• "Gender Composition and The Common Information Trap: How All-Female Teams Are More Prone to Decision-Making Biases". (with Yixiao Shen, and Wenxin Xie)

  • presented at AOM2025 symposium “Advancing Gender Diversity and Inclusion in Organizations”