Ziming Liu
Ziming Liu (刘子鸣)
zmliu@tsinghua.edu.cn
liuziming@meta-circle.com
lzmsldmjxm@gmail.com
Hi! I’m Ziming Liu (刘子鸣). I am a tenure-track Assistant Professor in College of AI at Tsinghua University. I am the founder of MetaCircle (元环智能), whose mission is to uncover Science of AI and use this science to achieve AI for AI.
I did my postdoc in AI + Neuroscience at Stanford with Andreas Tolias. I did my PhD in AI + Physics at MIT with Max Tegmark. Before that, I obtained my B.S. in physics from Peking University in 2020. Get my CV here.
I’m broadly interested in the intersection of AI and Science:
- Science of AI: Understanding AI using science. I’m interested in understanding intriguing network phenomena (e.g., grokking, neural scaling laws). For example: How Do Machines ‘Grok’ Data?.
- Science for AI: Advancing AI using science. When the scaling of the current paradigm plateaus, it is time to focus back on fundamental science (for AI). For example: The Physical Process That Powers a New Type of Generative AI.
- AI for Science: Advancing science using AI. I’m excited about inventing accurate/interpretable AI4Science models and curiosity-driven AI scientists. For example: Novel Architecture Makes Neural Networks More Understandable.
latest posts
| Jun 22, 2026 | Discovering 108 tricks to accelerate grokking |
|---|---|
| May 06, 2026 | AI4AI needs its own "world model" |
| May 05, 2026 | How does memorization affect l2? |
selected publications
-
-
Seeing is believing: Brain-inspired modular training for mechanistic interpretabilityEntropy, 2023 -
-
The Clock and the Pizza: Two Stories in Mechanistic Explanation of Neural NetworksIn Thirty-seventh Conference on Neural Information Processing Systems , 2023 -
Omnigrok: Grokking Beyond Algorithmic DataIn The Eleventh International Conference on Learning Representations , 2023 -
-
-