Jing Zhou
Jing Zhou is an AI researcher whose work spans few-shot language understanding, fine-tuning data, and Alibaba's Qwen models. In 2022, he coauthored FewNLU and FlipDA, two papers on few-shot language understanding. At Tsinghua's Institute for Interdisciplinary Information Sciences, he earned a PhD in 2023; his research interests span machine learning, NLP and large language models, and Baidu was his first job. He coauthored a discriminator-based zero-shot paper in 2023, a web-data fine-tuning study in 2024, and technical reports for Qwen3 and Qwen3-VL in 2025. As of October 2026, he works on the Qwen team at Alibaba Cloud.
Timeline
-
2022
FewNLU's evaluation framework
In 2022, he coauthors FewNLU, an evaluation framework for few-shot natural-language understanding that finds no single method dominates most tasks consistently.
-
2022
FlipDA for few-shot learning
In 2022, he coauthors FlipDA, which pairs a generative model with a classifier to create label-flipped examples for few-shot language understanding.
-
2023
A Tsinghua PhD
He earns a PhD at Tsinghua’s Institute for Interdisciplinary Information Sciences in 2023. His research interests span machine learning, NLP and large language models; Baidu is his first job.
-
2023
A universal discriminator
In 2023, he coauthors A Universal Discriminator for Zero-Shot Generalization, which trains a single discriminator to judge whether text comes from the true data distribution for zero-shot NLP tasks.
-
2024
Fine-tuning with web data
In 2024, he coauthors Leveraging Web-Crawled Data for High-Quality Fine-Tuning. It pairs web-crawled examples with a smaller high-quality data set, and the model-transformed data improves average scores on Chinese math problems by 9.4%.
-
May 2025
The Qwen3 report
In May 2025, he coauthors the Qwen3 Technical Report, which describes dense and mixture-of-experts models ranging from 0.6 billion to 235 billion parameters.
-
November 2025
The Qwen3-VL report
In November 2025, he coauthors the Qwen3-VL Technical Report, which introduces a vision-language model that accepts interleaved text, image and video contexts up to 256K tokens.
-
As of October 2026
On the Qwen team
As of October 2026, he works on the Qwen team at Alibaba Cloud.
Career
- Since 2024Alibaba
- 2023–2025Xi'an University of Science and Technology
- 2020–2024Zhejiang University
- 2019–2022Jiangsu University
- 2018–2023University of Missouri
- 2017–2020Office of Nuclear Energy
- 2017–2025Central South University
- 2014–2018Nanjing Agricultural University
- 2013–2015University of Technology Sydney
- 2012–2016Tongji University
Education
- 2023Tsinghua UniversityPhD
Most-cited works
- Effects of water content on fracture and mechanical behavior of sandstone with a low clay mineral content
- Qwen3 Technical ReportWith Baosong Yang, Chengen Huang, Chenxu Lv, Chujie Zheng, Fei Huang and Haoran Wei
- Streamwise feature selectionWith Lyle Ungar
- Classification of soybean leaf wilting due to drought stress using UAV-based imagery
- Real-Time Detection and Location of Potted Flowers Based on a ZED Camera and a YOLO V4-Tiny Deep Learning Algorithm
Colleagues
Sources and credits
Photo: We have no published photo of Jing Zhou yet. Send us a free-licensed photo.
- FewNLU: Benchmarking State-of-the-Art Methods for Few-Shot Natural Language Understanding
- FlipDA: Effective and Robust Data Augmentation for Few-Shot Learning
- ADL group @ IIIS
- A Universal Discriminator for Zero-Shot Generalization
- Leveraging Web-Crawled Data for High-Quality Fine-Tuning
- Leveraging Web-Crawled Data for High-Quality Fine-Tuning
- Qwen3 Technical Report
- Qwen3-VL Technical Report
- GitHub - QwenLM/Qwen3: Qwen3 is the large language model series developed by Qwen team, Alibaba Cloud.