Fatescript

I currently work at Moonshot AI. My interests center on large language models, training and inference systems, and the infrastructure that turns models into reliable products.

Before Moonshot, I worked at Megvii on computer vision libraries and deep learning infrastructure. I was one of the contributors to YOLOX, an anchor-free object detector, and helped redesign the library before its public release. I also maintained cvpods, and worked on MegEngine-based classification and detection frameworks such as basecls and basedet.

This blog is mostly where I write down engineering details, model-system observations, and long-term reflections. I care about explanations grounded in practice: code paths, numerical behavior, distributed training details, and the small decisions that decide whether a system actually works.

Please feel free to reach out if you want to discuss deep learning systems, LLM infrastructure, computer vision engineering, or anything you find interesting here.

selected publications

arXiv

Kimi K2.5: Visual Agentic Intelligence

Kimi Team, and Feng Wang

arXiv preprint arXiv:2602.02276, 2026

arXiv Code Website
arXiv

Kimi K2: Open Agentic Intelligence

Kimi Team, and Feng Wang

arXiv preprint arXiv:2507.20534, 2026

arXiv Code Website

arXiv

Baichuan 2: Open Large-scale Language Models

Aiyuan Yang, Bin Xiao, Bingning Wang, and 52 more authors

arXiv preprint arXiv:2309.10305, 2023

arXiv Bib Website

@article{yang2023baichuan,
  title = {Baichuan 2: Open Large-scale Language Models},
  author = {Yang, Aiyuan and Xiao, Bin and Wang, Bingning and Zhang, Borong and Bian, Ce and Yin, Chao and Lv, Chenxu and Pan, Da and Wang, Dian and Yan, Dong and Yang, Fan and Deng, Fei and Wang, Feng and Liu, Feng and Ai, Guangwei and Dong, Guosheng and Zhao, Haizhou and Xu, Hang and Sun, Haoze and Zhang, Hongda and Liu, Hui and Ji, Jiaming and Xie, Jian and Dai, JunTao and Fang, Kun and Su, Lei and Song, Liang and Liu, Lifeng and Ru, Liyun and Ma, Luyao and Wang, Mang and Liu, Mickel and Lin, MingAn and Nie, Nuolan and Guo, Peidong and Sun, Ruiyang and Zhang, Tao and Li, Tianpeng and Li, Tianyu and Cheng, Wei and Chen, Weipeng and Zeng, Xiangrong and Wang, Xiaochuan and Chen, Xiaoxi and Men, Xin and Yu, Xin and Pan, Xuehai and Shen, Yanjun and Wang, Yiding and Li, Yiyu and Jiang, Youxin and Gao, Yuchen and Zhang, Yupeng and Zhou, Zenan and Wu, Zhiying},
  journal = {arXiv preprint arXiv:2309.10305},
  year = {2023},
}

YOLOX: Exceeding YOLO Series in 2021

Zheng Ge, Songtao Liu, Feng Wang, and 2 more authors

arXiv preprint arXiv:2107.08430, 2021

arXiv Bib Code

@article{yolox2021,
  title = {YOLOX: Exceeding YOLO Series in 2021},
  author = {Ge, Zheng and Liu, Songtao and Wang, Feng and Li, Zeming and Sun, Jian},
  journal = {arXiv preprint arXiv:2107.08430},
  year = {2021},
}

cvpods: All-in-one Toolbox for Computer Vision Research

Benjin Zhu*, Feng Wang*, Jianfeng Wang, and 3 more authors

2020

Bib Code

@misc{zhu2020cvpods,
  title = {cvpods: All-in-one Toolbox for Computer Vision Research},
  author = {Zhu*, Benjin and Wang*, Feng and Wang, Jianfeng and Yang, Siwei and Chen, Jianhu and Li, Zeming},
  year = {2020},
}