About me

I’m Zitong Wang, an undergraduate student at Sun Yat-sen University. I am currently a research intern at the LAI Lab, working with Prof. Chengwei Qin. Previously, I interned at the LUMIA Lab, supervised by Prof. Zhouhan Lin. During my undergraduate studies, I am also mentored by Prof. Weibin Wu.

My research interests lie in LLM pretraining and model architectures, as well as autonomous agents capable of solving long-horizon tasks. I view these directions through a unified lens of scaling intelligence: pretraining and architectural innovations scale a model’s internal capabilities, while agents scale intelligence externally through memory, tool use, interaction, and sustained action in complex environments.

I am particularly interested in bridging these two forms of scaling to build more capable and general intelligent systems.

News 🔥

• [2026-08] 1 paper accepted at EMNLP 2026 (Budapest, Hungary)!
• [2026-06] Started LLM Research Internship at IQuest Research, UBIQUANT , developing more powerful pretraining models!
• [2026-04] 1 paper accepted by ICML 2026 as Spotlight (top 2.2%)！ See you in Seoul！
• [2025-07] 1 paper accepted by ACM Multimedia 2025！

Publication

ACM MM 2025

Detecting Violations of Physical Common Sense in Images: A Challenge Dataset and Effective Model

Weibin Wu, Zitong Wang, Zhengjie Luo, Wenqing Chen, Zibin Zheng

First student author

ACM Multimedia 2025 (CCF-A)

PDF Project

arXiv

Delta-K: Boosting Multi-Instance Generation via Cross-Attention Augmentation

Zitong Wang, Zijun Shen, Haohao Xu, Zhengjie Luo, Weibin Wu

arXiv preprint

PDF arXiv

ICML Spotlight

PonderLM-2: Pretraining LLM with Latent Thoughts in Continuous Space

Boyi Zeng, He Li, Shixiang Song, Yixuan Wang, Zitong Wang, Ziwei He, Xinbing Wang, Zhouhan Lin

ICML 2026 Spotlight (top 2.2%)

PDF

Preprint

Every Step a Thought: Implicit Visual Reasoning in Diffusion Language Models

Zitong Wang, Haohao Xu, Zijun Shen, Weibin Wu, Zibin Zheng

Under review

arXiv

AdaPonderLM: Gated Pondering Language Models with Token-Wise Adaptive Depth

Shixiang Song, He Li, Zitong Wang, Boyi Zeng, Feichen Song, Yixuan Wang, Ziwei He, Zhouhan Lin

arXiv preprint

PDF arXiv

arXiv

Pretraining with Token-Level Adaptive Latent Chain-of-Thought

Boyi Zeng, Yiqin Hao, He Li, Shixiang Song, Feichen Song, Zitong Wang, Siyuan Huang, Yi Xu, Ziwei He, Xinbing Wang, Zhouhan Lin

arXiv preprint

PDF arXiv

Selected Awards and Honors

• National Scholarship (Top 0.2% nationwide)
• First Class Academic Scholarship (top 5% in Sun Yat-sen University)
• National Second Award in the Contemporary Undergraduate Mathematical Contest in Modeling (CUMCM)
• Meritorious Winner in COMAP’s Mathematical Contest in Modeling

Personal Interests

I love cooking great food🥩, listening to jazz music🎵 and discussing the future tech world🏙️ with others. I also have a strong interest in quantitative trading and investing🪙. Feel free to reach out and chat with me about any of these!

↗ field notes