Xinghao Chen

I am Xinghao Chen, a joint Ph.D. candidate in Computer Science at The Hong Kong Polytechnic University and Eastern Institute of Technology, Ningbo. I am fortunate to be advised by Prof. Wenjie Li and Prof. Xiaoyu Shen. I received my B.S. in Intelligent Science and Technology from Nankai University in 2023. I am currently a research intern at Tencent Youtu Lab, working with Dr. Junnan Dong. For more details, please see my CV.

My research explores how large language models can reason more efficiently through compressed reasoning and abstract latent representations. Inspired by how humans can think without verbalizing every step, I have recently been focusing on the following directions:

• Thought Compression & Distillation: effective chain-of-thought distillation ([ACL’25]), condition-aware reasoning compression ([EMNLP’26]), reasoning efficiency, etc.
• Latent-Space Reasoning: latent reasoning landscape and taxonomy ([EMNLP’26]), effective supervision for latent chain-of-thought ([ICML’26]), continuous thought representations, etc.
• Applications (Efficiency): visual layer selection for multimodal LLMs ([EMNLP’25]), visual token pruning ([CVPR’26]), memory-augmented agents, KV-cache compression and reuse, etc.

News

Selected Publications

(*) Equal Contribution. (†) Corresponding Author.