About me
I am currently a Research Scientist at ByteDance, working on large language models for code generation.
Previously, I was a Ph.D. student working on Deep Learning at University of Illinois at Urbana-Chamapign. I was advised by Prof. Ruoyu Sun and co-advised by Prof. Naira Hovakimyan. During the first year of my Ph.D., I worked with Prof. Justin Sirignano on deep learning for computational fluid mechanics (CFD). During my master’s degree, I worked with Prof. Kyle Smith on computational methods for energy storage systems.
Research interest: Efficient Deep Learning; Natural Language Processing; Computer Vision; Generative Models; Graph Neural Networks (GNN); Numerical Methods.
News
[May. 2024] I have joined ByteDance as a Research Scientist.
[Mar. 2024] I have successfully defended my Ph.D. thesis!
[Jan. 2024] Our paper on model expansion is accepted to ICLR’2024!
[Oct. 2023] I have received the NeurIPS 2023 Scholar Award!
[Sept. 2023] Our paper about dynamic sparse training for GANs is accepted to NeurIPS’2023!
[Jan. 2023] Our paper about foresight pruning is accepted to ICLR’2023!
[March. 2022] Our paper about neural architecture search for meta learning is accpeted to CVPR’2022!
Publications
( $\dagger$ denotes to equal contribution)
LEMON: Lossless model expansion [arXiv, code]
Yite Wang, Jiahao Su, Hanlin Lu, Cong Xie, Tianyi Liu, Jianbo Yuan, Haibin Lin, Ruoyu Sun, Hongxia Yang
ICLR’2024：International Conference on Learning Representations
Balanced Training for Sparse GANs [arXiv, page, code]
Yite Wang$\dagger$, Jing Wu$\dagger$, Naira Hovakimyan, Ruoyu Sun
NeurIPS’2023: Neural Information Processing Systems
NTK-SAP: Improving neural network pruning by aligning training dynamics [arXiv, page, code]
Yite Wang, Dawei Li, Ruoyu Sun
ICLR’2023: International Conference on Learning Representations
Global Convergence of MAML and Theory-Inspired Neural Architecture Search for Few-Shot Learning [arXiv, code]
Haoxiang Wang$\dagger$, Yite Wang$\dagger$, Ruoyu Sun, Bo Li
CVPR’2022: The IEEE/CVF Computer Vision and Pattern Recognition Conference
Numerical investigation of convective transport in redox flow battery tanks: Using baffles to increase utilization [paper]
Yite Wang, Kyle Smith
Journal of Energy Storage