Penghao Yin

Master's student · Tsinghua University

Penghao Yin

I am a second-year master's student at Tsinghua University, advised by Prof. Xiaoping Zhang. Before joining Tsinghua, I received my bachelor's degree from Harbin Institute of Technology, Shenzhen, in June 2025, ranking 1st out of 172 students by GPA. I am currently an intern at Naive AI. Previously, I was an intern at ByteDance, OpenGVLab (InternVL team), and THUNLP.

My current research focuses on AI agents, especially coding agents, with a broader interest in all topics related to Artificial General Intelligence (AGI).

Publications

MetaReason pipeline for geometry parsing, verification, and interleaved reasoning

MetaReason: Precise Interleaved Multimodal Reasoning via Editing Meta Information for Solving Geometry Problems

Penghao Yin, Haomin Wang, Qihong Tang, Xiaoye Qu, Hongjie Zhang, Xiao-Ping Zhang
ArXiv, 2026.
AutoTrainess workflow connecting a CLI agent, AutoTrain Hub, and a training machine

AutoTrainess: Teaching Language Models to Improve Language Models Autonomously

Zhaojian Yu*, Penghao Yin*, Shuzheng Gao*, Shilin He, Kai Cai, Xiao-Ping Zhang
ArXiv, 2026.
InternVL-U examples of image generation and editing

InternVL-U: Democratizing Unified Multimodal Models for Understanding, Reasoning, Generation and Editing

Changyao Tian*, Danni Yang*, Guanzhou Chen*, Erfei Cui*, Zhaokai Wang*, Yuchen Duan*, Penghao Yin*, Sitao Chen, Ganlin Yang, Mingxin Liu, Zirun Zhu, Ziqian Fan, Leyao Gu, Haomin Wang, Qi Wei, Jinhui Yin, Xue Yang, Zhihang Zhong, Qi Qin, Yi Xin, Bin Fu, Yihao Liu, Jiaye Ge, Qipeng Guo, Gen Luo, Hongsheng Li, Yu Qiao, Kai Chen, Hongjie Zhang
Technical Report, 2026.
GenExam teaser with a graph-drawing task and evaluation criteria

GenExam: A Multidisciplinary Text-to-Image Exam

Zhaokai Wang*, Penghao Yin*, Xiangyu Zhao, Changyao Tian, Yu Qiao, Wenhai Wang, Jifeng Dai, Gen Luo
ICML, 2026.

Experiences

2026.8 – Present

Naive AI

Research Intern

2025.1 – 2025.5

THUNLP

Research Intern

Projects

GenEditEvalKit

The first unified, efficient, and extensible evaluation toolkit for evaluating image generation and editing models across multiple benchmarks.

Honors & Awards