Zekai Zhang

Shot from Shenandoah National Park, while waiting to witness the Perseid meteor shower. 🥳

I am a fourth-year undergraduate student in Turing class, School of Electronics Engineering and Computer Science, Peking University. In summer 2023 , I had a research internship in CMU hosted by Prof. Beidi Chen. Currently, I am a research intern at Microsoft Research Asia NLC Group, working under the guidance of Dr. Chenfei Wu. I am privileged to have been accepted into the NLP PhD program at Peking University, where I will be advised by Prof. Dongyan Zhao and Prof. Huishuai Zhang.

My research interests lie in demystifying alignment and leveraging LLMs for real-world tasks. Here are the core problems I’m thinking about:

What specific knowledge/ability do LLMs acquire through alignment?
Is superalignment applicable? How can we achieve superalignment?
What are the ability boundarieas for LLMs in real-world scenarios?

Email / Github / CV

Recent News

Dec 18, 2024	Publish a blog on pretrain experience, summarizing my internship @ StepFun: Large-Scale Pretraining Blog
May 16, 2024	Our paper on Task Completion (PPTC) accepted at ACL2024 Findings!
May 2, 2024	Our paper on LLMs for SVG (StrokeNUWA) accepted at ICML2024!
Apr 15, 2024	Becoming an intern @ StepFun, working on Large Video Models with 10 member & 3k GPU (estimated).
Mar 18, 2024	Our paper on LLMs for Stylized Dialogue (StyleChat) now available on arXiv!
Mar 6, 2024	Our paper on Robustness of Task Completion (PPTC-R) now available on arXiv!
Jan 30, 2024	Our paper on LLMs for SVG (StrokeNUWA) now available on arXiv!
Nov 7, 2023	Our paper on Task Completion (PPTC) now available on arXiv!
Oct 24, 2023	Becoming an intern @ MSRA NLC group!
Oct 6, 2023	Our paper on Stylized Dialogue (KASDG) accepted by EMNLP2023 Findings!

Selected Publications

Under Review

PPTC-R benchmark: Towards Evaluating the Robustness of Large Language Models for PowerPoint Task Completion

Zekai Zhang*, Yiduo Guo*, Yaobo Liang, Dongyan Zhao, and 1 more author

arXiv, 2024

Bib PDF Code

@article{PPTC-R,
  title = {PPTC-R benchmark: Towards Evaluating the Robustness of Large Language Models for PowerPoint Task Completion},
  author = {Zhang*, Zekai and Guo*, Yiduo and Liang, Yaobo and Zhao, Dongyan and Nan, Duan},
  journal = {arXiv},
  year = {2024}
}

Under Review

StyleChat: Learning Recitation-Augmented Memory in LLMs for Stylized Dialogue Generation

Zekai Zhang*, Jinpeng Li*, Quan Tu, Xin Cheng, and 2 more authors

arXiv, 2024

Bib PDF

@article{StyleChat,
  title = {StyleChat: Learning Recitation-Augmented Memory in LLMs for Stylized Dialogue Generation},
  author = {Zhang*, Zekai and Li*, Jinpeng and Tu, Quan and Cheng, Xin and Zhao, Dongyan and Yan, Rui},
  journal = {arXiv},
  year = {2024}
}

Under Review

StrokeNUWA: Tokenizing Strokes for Vector Graphic Synthesis

Zecheng Tang, Chenfei Wu, Zekai Zhang, Mingheng Ni, and 6 more authors

arXiv, 2024

Bib PDF

@article{StrokeNUWA,
  title = {StrokeNUWA: Tokenizing Strokes for Vector Graphic Synthesis},
  author = {Tang, Zecheng and Wu, Chenfei and Zhang, Zekai and Ni, Mingheng and Yin, Shengming and Liu, Yu and Yang, Zhengyuan and Wang, Lijuan and Liu, Zicheng and Li, Juntao},
  journal = {arXiv},
  year = {2024}
}

Under Review

PPTC benchmark: Evaluating large language models for powerpoint task completion

Zekai Zhang*, Yiduo Guo*, Yaobo Liang, Dongyan Zhao, and 1 more author

arXiv, 2023

Bib PDF Code

@article{PPTC,
  title = {PPTC benchmark: Evaluating large language models for powerpoint task completion},
  author = {Zhang*, Zekai and Guo*, Yiduo and Liang, Yaobo and Zhao, Dongyan and Nan, Duan},
  journal = {arXiv},
  year = {2023}
}

EMNLP Findings

Stylized Dialogue Generation with Feature-Guided Knowledge Augmentation

Zekai Zhang*, Jinpeng Li*, Xiuying Chen, Dongyan Zhao, and 1 more author

In Findings of the Association for Computational Linguistics: EMNLP 2023, 2023

Bib PDF Code

@inproceedings{KASDG,
  title = {Stylized Dialogue Generation with Feature-Guided Knowledge Augmentation},
  author = {Zhang*, Zekai and Li*, Jinpeng and Chen, Xiuying and Zhao, Dongyan and Yan, Rui},
  booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2023},
  year = {2023}
}

Selected Honors

John Hopcroft Scholarship, 2023
Award for Scientific Research Excellents, 2022
John Hopcroft Scholarship, 2022
Peking University Dean's Scholarship (Second Prize), 2022
John Hopcroft Scholarship, 2021
Peking University Freshman Scholarship (Second Prize), 2020
Second Highest in National College Entrance Examination, ranked 4/50000+, 2020