- SoLoPO: Unlocking Long-Context Capabilities in LLMs via Short-to-Long Preference Optimization
- Unveiling and Addressing Pseudo Forgetting in Large Language Models
- PSST: A Benchmark for Evaluation-driven Text Public-Speaking Style Transfer
- MindLLM: Pre-training Lightweight Large Language Model from Scratch, Evaluations and Domain Applications
Research Experience
Interned at Rednote (2022.10-2023.02) and Tongyi, Alibaba (2024.12-2025.10).
Education
Master's student at the School of Computer Science and Technology, Beijing Institute of Technology (2023-present), under the guidance of Prof. Yang Gao; Bachelor's degree in Artificial Intelligence from Beijing Institute of Technology (2019-2023).
Background
Research interests include natural language processing, capabilities of large language models (LLMs), and underlying mechanisms. Currently focused on long-context language modeling, particularly efficient training and inference. Previously worked on continual learning and long-text style modeling. Research aims to enhance models' real-world applicability while making them more controllable and safe.