I am a PhD student in computer science at Beijing Institute of Technology. I previously majored in finance and systems science, thus holding a Bachelor’s degree in Economics and a Master’s degree in Science.

My main focus is on cutting-edge artificial intelligence technologies such as large language models. My research areas include model distillation, content saferty, and interpretability. In the future, I am inclined to study post-training of models and reinforcement learning training techniques.

Publication

📄 First Author

• Probing the Difficulty Perception Mechanism of Large Language Models
  • arXiv preprint arXiv:2510.05969
  • Sunbowen Lee, Qingyu Yin, Chak Tou Leong, Jialiang Zhang, Yicheng Gong, Shiwen Ni, Min Yang, Xiaoyu Shen
  • 2025
• Quantification of large language model distillation
  • Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
  • Sunbowen Lee, Junting Zhou, Chang Ao, Kaige Li, Xeron Du, Sirui He, Haihong Wu, Tianci Liu, Jiaheng Liu, Hamid Alinejad-Rokny, et al.
  • 2025
• xjailbreak: Representation space guided reinforcement learning for interpretable LLMs jailbreaking
  • arXiv preprint arXiv:2501.16727
  • Sunbowen Lee, Shiwen Ni, Chi Wei, Shuaimin Li, Liyang Fan, Ahmadreza Argha, Hamid Alinejad-Rokny, Ruifeng Xu, Yicheng Gong, Min Yang
  • 2025
• Counterfactual experience augmented off-policy reinforcement learning
  • Neurocomputing
  • Sunbowen Lee, Yicheng Gong, Chao Deng
  • 2025
• MacLight: Multi-scene Aggregation Convolutional Learning for Traffic Signal Control
  • AAMAS ‘25: Proceedings of the 24th International Conference on Autonomous Agents and Multiagent Systems
  • Sunbowen Lee, Hongqin Lyu, Yicheng Gong, Yingying Sun, Chao Deng
  • 2024

👥 Collaboration

• Evaluating Parameter Efficient Methods for RLVR
  • arXiv preprint arXiv:2512.23165
  • Qingyu Yin, Yulun Wu, Zhennan Shen, Sunbowen Lee, Zhilin Wang, Yanshu Li, Chak Tou Leong, Jiale Kang, Jinjin Gu
  • 2025
• PoolingVQ: A VQVAE Variant for Reducing Audio Redundancy and Boosting Multi-Modal Fusion in Music Emotion Analysis
  • arXiv preprint arXiv:2509.11976
  • Dinghao Zou, Yicheng Gong, Xiaokang Li, Xin Cao, Sunbowen Lee
  • 2025
• Emoanti: audio anti-deepfake with refined emotion-guided representations
  • arXiv preprint arXiv:2509.10781
  • Xiaokang Li, Yicheng Gong, Dinghao Zou, Xin Cao, Sunbowen Lee
  • 2025

🔬 Internship Experience

• Eastern Institute of Technology (EIT) | Large Language Model Research Intern (Sept 2025 – Nov 2025)
  • Research Focus: Efficient inference for LLMs under Prof. Xiaoyu Shen.
  • Achievement: First author of a research paper on interpretability of reasoning.
• Gongdao Network Technology | LLM Algorithm Intern (Apr 2025 – Sept 2025)
  • Interpretability: Studied spoken text representation patterns across BERT and Decoder-only architectures.
  • Audio Engineering: Built automated audio processing pipelines and developed pitch recognition for TTS error correction.
  • NLP Optimization: Fine-tuned BGE embedding models for RAG and implemented model distillation engineering.
• Shenzhen Institute of Advanced Technology (SIAT), CAS | Research Intern (Sept 2024 – Apr 2025)
  • Collaboration: Contributed to multiple LLM research initiatives within Prof. Min Yang’s team.
  • AI Safety: Researched Red Teaming and jailbreak attacks for LLMs and Multimodal models (ARR under review).
  • Key Publication: First Author of a paper on LLM distillation detection, accepted by ACL 2025 (Main Conference).