Sotabase

Career

· PhD in Computer Science, INRIA (Institut National de Recherche en Informatique et en Automatique)2023–
· Postdoctoral Research Associate (PDRA) in AI for autonomous systems, Oxford Robotics Institute (ORI) and United Kingdom Atomic Energy Authority (UKAEA)2023–
· Presidential Young Professor, National University of Singapore2020–
· Intern, Intel Asia-Pacific Research Center2015–2016
· MS in Computer Science, Tsinghua University2015–
· PhD in Computer Science, University of California, Berkeley2015–2020
· PhD Student, University of California, Berkeley2015–2020
· BS in Computer Science, China Agricultural University2009–2012
· Undergraduate Studies, China Agricultural University
· Postdoctoral Researcher, Stanford University

Publications (127)

Large Batch Optimization for Deep Learning: Training BERT in 76 minutes
International Conference on Learning Representations · 2019
1,119
cited
Open-Sora: Democratizing Efficient Video Production for All
arXiv.org · 2024
490
cited
ImageNet Training in Minutes
International Conference on Parallel Processing · 2017
464
cited
Dense high-entropy boride ceramics with ultra-high hardness
Scripta Materialia · 2019
250
cited
OpenMoE: An Early Effort on Open Mixture-of-Experts Language Models
International Conference on Machine Learning · 2024
160
cited
Reducing BERT Pre-Training Time from 3 Days to 76 Minutes
arXiv.org · 2019
125
cited
Investigation on the microstructure and toughness of coarse grained heat affected zone in X-100 multi-phase pipeline steel with high Nb content
2012
122
cited
To Repeat or Not To Repeat: Insights from Scaling LLM under Token-Crisis
Neural Information Processing Systems · 2023
122
cited
Preventing Zero-Shot Transfer Degradation in Continual Learning of Vision-Language Models
IEEE International Conference on Computer Vision · 2023
113
cited
Response Length Perception and Sequence Scheduling: An LLM-Empowered LLM Inference Pipeline
Neural Information Processing Systems · 2023
105
cited
Large-Batch Training for LSTM and Beyond
International Conference for High Performance Computing, Networking, Storage and Analysis · 2019
96
cited
Magnetic cobalt ferrite biochar composite as peroxymonosulfate activator for removal of lomefloxacin hydrochloride
2021
95
cited
Open-Sora 2.0: Training a Commercial-Level Video Generation Model in $200k
arXiv.org · 2025
85
cited
Scaling Deep Learning on GPU and Knights Landing clusters
International Conference for High Performance Computing, Networking, Storage and Analysis · 2017
80
cited
InfoBatch: Lossless Training Speed Up by Unbiased Dynamic Data Pruning
International Conference on Learning Representations · 2023
79
cited
Improved densification and hardness of high-entropy diboride ceramics from fine powders synthesized via borothermal reduction process
Ceramics International · 2020
75
cited
Investigation on the crystallography of the transformation products of reverted austenite in intercritically reheated coarse grained heat affected zone
2013
75
cited
Fabrication of textured (Hf0.2Zr0.2Ta0.2Cr0.2Ti0.2)B2 high-entropy ceramics
2021
62
cited
Prompt Vision Transformer for Domain Generalization
arXiv.org · 2022
61
cited
100-epoch ImageNet Training with AlexNet in 24 Minutes
arXiv.org · 2017
60
cited
Sotabase