Junbo Zhao (Jake)

赵俊博（官方主页）

prof_pic.jpg

Motto: 日拱一卒(Advance a little every day like a pawn in chess)

Assistant Professor, Zhejiang University

Director, Ant Group

Bio (2026 Version)

Junbo Zhao is an Assistant Professor at Zhejiang University (100-Young Professor Program) and a Director at Ant Group.

Current Focus (2025+): He focuses on two pivotal frontiers:

Next-Gen Scaling (dLLMs): Spearheading LLaDA 2.0 (100B MoE), a diffusion language model that matches AR performance with doubled inference speed.

The “Second Half” of Data: Moving from quantity to density. He explores high-leverage data strategies (synthesis, distillation, and evaluation benchmarks like HeartBench) to unlock efficiency.

Background: In 2024, he led the TableGPT project (TableGPT2 & R1), achieving over 200k downloads and widespread industry adoption in finance and manufacturing. He earned his Ph.D. from NYU under Turing Award laureate Yann LeCun. Previously, he contributed to foundational frameworks (PyTorch, FAISS) at Meta FAIR and developed the first end-to-end self-driving system at NVIDIA (showcased at GTC).

Impact: His work has garnered over 26,000 citations (as of Jan 2026). He is a recipient of the Zhejiang Province Science and Technology Progress First Prize (2024), the QbitAI 2025 AI Person of the Year, the WAIC Yunfan Award, and was featured on the cover of Forbes 30-under-30.

news

Dec 23, 2025 TableGPT-R1 fully released and opened.
Dec 10, 2025 LLaDA 2.0 fully released and opened.
Dec 06, 2025 HeartBench (Evaluating LLMs’ Anthropomorphicity) is released.
Aug 23, 2025 Six paper were accpeted at EMNLP 2025.

selected stuff

  1. llada2.png
    LLaDA2. 0: Scaling Up Diffusion Language Models to 100B
    Tiwei Bie, Maosong Cao, Kun Chen, Lun Du, Mingliang Gong, Zhuochen Gong, Yanmei Gu, Jiaqi Hu, Zenan Huang, Zhenzhong Lan, and  others
    arXiv preprint arXiv:2512.15745, 2025
    Model Repository
  2. heartbench.png
    HeartBench: Probing Core Dimensions of Anthropomorphic Intelligence in LLMs
    Jiaxin Liu, Peiyi Tu, Wenyu Chen, Yihong Zhuang, Xinxia Ling, Anji Zhou, Chenxi Wang, Zhuo Han, Zhengkai Yang, Junbo Zhao, and  others
    arXiv preprint arXiv:2512.21849, 2025
    Repository
  3. rubicon.png
    Reinforcement learning with rubric anchors
    Zenan Huang, Yihong Zhuang, Guoshan Lu, Zeyu Qin, Haokai Xu, Tianyu Zhao, Ru Peng, Jiaqi Hu, Zhanming Shen, Xiaomeng Hu, and  others
    arXiv preprint arXiv:2508.12790, 2025
    Model
  4. ling1t.png
    Every step evolves: Scaling reinforcement learning for trillion-scale thinking model
    Ling Team, Anqi Shen, Baihui Li, Bin Hu, Bin Jing, Cai Chen, Chao Huang, Chao Zhang, Chaokun Yang, Cheng Lin, and  others
    arXiv preprint arXiv:2510.18855, 2025
    Model Repository
  5. tablegpt2.png
    Tablegpt2: A large multimodal model with tabular data integration
    arXiv preprint arXiv:2411.02059, 2024
    Model Repository
  6. bme2024.png
    Identification of potent antimicrobial peptides via a machine-learning pipeline that mines the entire space of peptide sequences
    Junjie Huang, Yanchao Xu, Yunfan Xue, Yue Huang, Xu Li, Xiaohui Chen, Yao Xu, Dongxiang Zhang, Peng Zhang, Junbo Zhao, and  others
    Nature Biomedical Engineering, 2023