❀ Biography
I am now a M.S. student at Shanghai Jiao Tong University, Department of Computer Science and Engineering, under the supervision of Prof. Hai Zhao. Before that, I received a B.S. degree in Computer Science and Technology from Southeast University, in 2023. In my undergraduate stage, I have participated in several research projects to explore my interest in computer science. Foremost among them is a personal scientific research project related to order dispatching which raised my curiosity in reinforcement learning and data mining.
Thinking of the current postgraduate stage, it was literally a serendipity that I began to delve into researches in Large Language Models (LLMs). I worked on several interesting research topics related to LLMs, including hallucination, model compression and representation engineering. Currently, I am working as a research intern at Alibaba, primarily focusing on LLM agents. Here is my CV in English and I’m actively seeking summer internship opportunities related to LLMs for Summer 2025.
📝 Publications
A full publication list is available on my google scholar page.
(* denotes equal contribution)
- [Preprint, 2025] PGPO: Enhancing Agent Reasoning via Pseudocode-style Planning Guided Preference Optimization, Zouying Cao, Runze Wang, Yifei Yang, Xinbei Ma, Xiaoyong Zhu, Bo Zheng, Hai Zhao. [Code]
- [Preprint, 2025] LESA: Learnable LLM Layer Scaling-Up, Yifei Yang, Zouying Cao, Xinbei Ma, Yao Yao, Libo Qin, Zhi Chen, Hai Zhao. [PDF] [Code]
- [Preprint, 2025] Plan-over-Graph: Towards Parallelable LLM Agent Schedule, Shiqi Zhang*, Xinbei Ma*, Zouying Cao, Zhuosheng Zhang, Hai Zhao. [PDF] [Code]
- [Preprint, 2024] KVSharer: Efficient Inference via Layer-Wise Dissimilar KV Cache Sharing, Yifei Yang, Zouying Cao, Qiguang Chen, Libo Qin, Dongjie Yang, Hai Zhao, Zhi Chen. [PDF] [Code]
- [AAAI Oral, 2025] SCANS: Mitigating the Exaggerated Safety for LLMs via Safety-Conscious Activation Steering, Zouying Cao, Yifei Yang, Hai Zhao. [PDF] [Code]
- [EMNLP-Findings, 2024] Head-wise Shareable Attention for Large Language Models, Zouying Cao, Yifei Yang, Hai Zhao. [PDF] [Code] [Slides] [Poster]
- [EMNLP-Findings, 2024] LaCo: Large Language Model Pruning via Layer Collapse, Yifei Yang, Zouying Cao, Hai Zhao. [PDF] [Code]
- [Preprint, 2023] AutoHall: Automated Hallucination Dataset Generation for Large Language Models, Zouying Cao, Yifei Yang, Hai Zhao. [PDF] [Code]
💻 Internship
-
2024.07 - Present, Alibaba’s Taotian Group, Research Intern
Topic: LLM Agent Planning
Contribution: We investigate the effectiveness of pseudocode-style plans in agent reasoning, which are more concise and structured than NL plans. Based on two designed planning-oriented rewards, we further introduce PGPO, a preference optimization method that empowers LLM agents with enhanced reasoning capabilities.
🎓 Education
- 2023.09 - 2026.03 (expected), M.S.@SJTU, Computer Science and Technology, Shanghai, China.
- 2019.09 - 2023.06, B.S.@SEU, Computer Science and Technology, Nanjing, China.
🌲 Service
- Reviewers: ARR
- Student Works
- Volunteers
🏆 Honors and Awards
Below, I list some Honors and Awards that inspire me deeply.
- 2024-11 Huatai Securities Technology Scholarship
- 2024-11 First Prize Graduate Academic Scholarship in SJTU
- 2023-06 Outstanding Graduates of Southeast University
- 2023-06 Outstanding Undergraduate Thesis Award of Southeast University
- 2022-10 Huawei Scholarship
- 2021-12 National Scholarship
- 2021-05 Second Price, National English Competition for College Students(NECCS)
- 2020~2022 Three-good Student for three consecutive years
- 2020-12 Scholarship on Social Works in SEU
- 2020-12 President Scholarship in SEU
- 2020-05 Excellent League Cadres, Award for the Models of the Chinese Youth in SEU
- 2020-05 Third Prize, the 17th Southeast University College Student Programming Competition