Yuanzhi Li
Yuanzhi Li is a researcher whose main field is deep learning theory, and his papers range from LoRA to small language models. He earned a Princeton PhD in 2018 under Sanjeev Arora, then worked as an assistant professor at CMU. In 2023 he received a Sloan Research Fellowship, and his paper on ensembles and self-distillation earned an ICLR honorable mention. In May 2023 he and Ronen Eldan posted TinyStories on arXiv, a study of how small a language model can be and still speak coherent English. He co-authored Microsoft Research's Phi-4 technical report, which describes a 14-billion parameter language model, and he has been a research scientist at Meta since 2025.
Timeline
-
2018
A Princeton PhD under Sanjeev Arora
He completes a Princeton PhD in 2018 under Sanjeev Arora. His dissertation is titled On The Ability Of Gradient Descent To Learn Neural Networks.
-
2021
Co-authoring the LoRA method
He co-authors LoRA, a low-rank adaptation method for large language models. The arXiv preprint appears in June 2021.
-
By 2022
An assistant professorship at CMU
He works as an assistant professor in CMU's Machine Learning Department and as a visiting researcher at Microsoft. His main research area is deep learning theory.
-
2023
A 2023 Sloan Research Fellowship
He and Rashmi Vinayak earn 2023 Sloan Research Fellowships in recognition of their research accomplishments.
-
2023
An ICLR 2023 honorable mention
He and Zeyuan Allen-Zhu co-write a paper on ensembles and self-distillation, which earns an ICLR 2023 honorable mention.
-
May 2023
TinyStories with Ronen Eldan
He and Ronen Eldan submit TinyStories to arXiv in May 2023, asking how small a language model can be and still speak coherent English.
-
December 2024
Phi-4 at Microsoft Research
He co-authors Microsoft Research's Phi-4 technical report, which appears on arXiv in December 2024. The report describes a 14-billion parameter language model.
-
Since 2025
A research scientist at Meta
He works as a research scientist at Meta, after an assistant professorship at Carnegie Mellon.
Sources and credits
Photo: We have no published photo of Yuanzhi Li yet. Send us a free-licensed photo.
- Yuanzhi Li (CMU)
- Yuanzhi Li - The Mathematics Genealogy Project
- [2106.09685] LoRA: Low-Rank Adaptation of Large Language Models
- SCS Faculty Earn 2023 Sloan Research Fellowships
- ICLR 2023 Awards
- [2012.09816] Towards Understanding Ensemble, Knowledge Distillation and Self-Distillation in Deep Learning
- [2305.07759] TinyStories: How Small Can Language Models Be and Still Speak Coherent English?
- [2412.08905] Phi-4 Technical Report
- Phi-4 Technical Report - Microsoft Research
- Meta百亿抢人内幕，清华学霸转行AGI拿千万年薪，教授校友看呆了-36氪