Iām a 4th-year undergraduate student majoring in Information and Computing Science at Beijing Jiaotong University. Now I am working as a research intern at Microsoft under the guidance of Principal Researcher Dr. Justin Ding. Previously, I was working at Peking University under the supervision of Prof. Xiaojun Wan.
As a student with a multidisciplinary background in Mathematics and Artificial Intelligence, strong foundations in research and leadership roles, I am keen on exploring opportunities that can deepen my understanding of Generative AI.
š„ News
-2024.08: Ā šš I will join Microsoft as Researcher Intern in August under the guidance of Principal Researcher Justin Ding.
š Publications
![sym](images/Example_page-0001.jpg)
Jinxiang Xie, Yilin Li, Xunjian Yin, Xiaojun Wan
- We introduce new sub-metrics for GEC evaluation, diverging from previous categorical approaches.
- We propose a novel dynamic weighting-based GEC evaluation method, DSGram, which integrates the Analytic Hierarchy Process (Ana 1987) with large language models to ascertain the relative importance of different evaluation criteria.
- We present two datasets: DSGram-Eval, created through human scoring, and DSGram-LLMs, a larger dataset designed to simulate human scoring for fine-tuning.
š Honors and Awards
- 2024.05 Honorable Mention of 2024 Mathematical Contest In Modeling
- 2023.04 First Prize of 2023 Beijing Jiaotong University Mathematical Modeling Competition
š Educations
- 2021.09 - 2025.06, Bachelor of Science in Information and Computing Science, Beijing Jiaotong University.
š» Internships
- Summer Workshop Student, 2023.05 - 2023.07, School of Computer, National University of Singapore.
- Research Intern, 2023.11 - 2024.08, Wangxuan Institute of Computer Technology, Peking University.
- Research Intern, 2024.08 - Present, Data, Knowledge and Intelligence (DKI) Group, Microsoft.
š Blogs
![sym](images/pic02.jpg)
LLMs: Cutting-Edge Technology and Future Applications
My notes from a presentation on LLMs at the Gaoling Schoolof Artificial Intelligence, Renmin University of China, which featured people from top research institutions.
Prompt Engineering: How To Ask LLMS Better
Introduce a number of methods for optimizing the output of large language models and reducing the probability of irrelevant or incorrect responses.