Weizhu Chen
As of October 2026, Weizhu Chen is a Technical Fellow and corporate vice president at Microsoft AI, where he leads a modeling team for large-scale model training. He contributed to DeBERTa in 2020 and to LoRA in 2021. In 2022, his team owned model quality and helped launch GitHub Copilot's general availability. He contributed to Phi, and the Phi-3 technical report was submitted to arXiv in April 2024. In 2024, NeurIPS named Not All Tokens Are What You Need for Pretraining a Best Paper Runner-up.
Timeline
-
2020
Disentangled attention in DeBERTa
He contributes to DeBERTa: Decoding-enhanced BERT with Disentangled Attention, and the arXiv preprint is submitted in June 2020.
-
2021
The LoRA adaptation method
He contributes to LoRA: Low-Rank Adaptation of Large Language Models, which is submitted to arXiv in June 2021.
-
2022
Model quality for Copilot
His team owns model quality and helps launch GitHub Copilot's general availability in 2022.
-
April 2024
Phi-3 technical report
He contributes to Phi, and the Phi-3 technical report is submitted to arXiv in April 2024.
-
2024
A runner-up best paper
He co-writes Rho-1: Not All Tokens Are What You Need. NeurIPS names Not All Tokens Are What You Need for Pretraining a Best Paper Runner-up in 2024.
-
As of October 2026
Technical Fellow at Microsoft AI
He is a Technical Fellow and corporate vice president at Microsoft AI, where he leads a modeling team for large-scale model training.
Sources and credits
Photo: We have no published photo of Weizhu Chen yet. Send us a free-licensed photo.
- Weizhu Chen | Microsoft Azure Blog | Microsoft Azure
- [2006.03654] DeBERTa: Decoding-enhanced BERT with Disentangled Attention
- [2106.09685] LoRA: Low-Rank Adaptation of Large Language Models
- [2404.14219] Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone
- [2404.07965] Rho-1: Not All Tokens Are What You Need
- NeurIPS 2024 Awards
- Weizhu Chen in Microsoft Research