Scholar
Xunjian Yin
Google Scholar ID: PociQ5EAAAAJ
Peking University
Citations & Impact
Citations
566
H-index
10
i10-index
11
Publications
20
Co-authors
12
list available
Publications
Resume (English only)
Academic Achievements
- - Preprints:
- - The Geometry of Reasoning: Flowing Logics in Representation Space
- - LEDOM: An Open and Fundamental Reverse Language Model
- - ContraSolver: Self-Alignment of Language Models by Resolving Internal Preference Contradictions
- - Selected Publications:
- - DAMON: A Dialogue-Aware MCTS Framework for Jailbreaking Large Language Models
- - Gödel Agent: A Self-Referential Agent Framework for Recursive Self-Improvement
- - Themis: A Reference-free NLG Evaluation Language Model with Flexibility and Interpretability
- - Benchmarking Knowledge Boundary for Large Language Model: A Different Perspective on Model Evaluation
- - History Matters: Temporal Knowledge Editing in Large Language Model
- - ALCUNA: Large Language Models Meet New Knowledge
- - How Do Seq2Seq Models Perform on End-to-End Data-to-Text Generation?
- - Selected Projects:
- - Overleaf-Bib-Helper
- - Gödel Agent
- - Music Letter
Research Experience
- - June 2025 - Aug. 2025, Shanghai AI Lab, Mentor: Dr. Jie Fu, Reinforcement Learning
- - June 2024 - February 2025, UCSB NLP Group, Mentor: Prof. William Yang Wang, One paper about Self-Referential Agent Framework; Ongoing work on reverse language model pre-training
- - Feb. 2022 ~ Aug. 2022, Microsoft Research Asia (NLC Group), Mentors: Dr. Kai Chen and Dr. Shuming Ma, Pre-training with Curriculum Learning and OCR
- - Oct. 2020 - June 2022, Wangxuan Institute of Computer Technology, Peking University, Mentor: Prof. Xiaojun Wan, One paper at ACL 2022 on data-to-text generation
- - Apr. 2020 - Nov. 2021, Institute of Computational Linguistics, Peking University, Mentor: Prof. Yunfang Wu, Multi-Task Learning for Grammar Error Correction
- - July 2019 - Dec. 2019, Institute of Computational Linguistics, Peking University, Incomplete information
Education
- - Duke University, CS PhD student, Advisor: Prof. Shuyan Zhou
- - Peking University, Master's and Bachelor's degree, Advisor: Prof. Xiaojun Wan
Background
- - Research Interests: Making models more generalizable across real-world tasks, believes in recursive self-improvement as the path to general intelligence
- - Current Research Focus: Self-referential agents and world models
- - Believes in using methods that are simple, scalable, generalizable, and graceful
Miscellany
- Personal Interests: Photography
