Junyang Lin

Research Interests

Areas of current research focus

Scaling laws, architecture design, and efficient training methods for large language models.

Autonomous agents capable of multi-step reasoning and tool usage.

Unified models across text, vision, and other modalities.

Enabling models for extended sequential planning and execution.

Unifying vision-based understanding and generation for comprehensive world modeling.

Reinforcement learning algorithms co-designed with scalable infrastructure.

Selected academic work

Qwen3 Technical Report

A. Yang, A. Li, B. Yang, B. Zhang, ... J. Lin, ...

arXiv preprint arXiv:2505.09388, 2025

arXiv Code

Qwen3-VL Technical Report

S. Bai, K. Chen, X. Liu, J. Wang, ... J. Lin, ...

arXiv preprint arXiv:2502.13923, 2025

arXiv Code

Qwen3-Coder: Agentic Coding in the World

S. Bai, P. Wang, W. Ge, X. Liu, ... J. Lin, ...

Qwen Blog, 2025

Blog Code

Group Sequence Policy Optimization

C. Zheng, S. Liu, M. Li, X. Chen, B. Yu, C. Gao, K. Dang, Y. Liu, R. Men, A. Yang, J. Lin, J. Zhou

arXiv preprint arXiv:2507.18071, 2025

arXiv Code

Gated Attention for Large Language Models: Non-linearity, Sparsity, and Attention-Sink-Free

Z. Qiu, Z. Wang, B. Zheng, Z. Huang, K. Wen, S. Yang, R. Men, L. Yu, F. Huang, S. Huang, D. Liu, J. Zhou, J. Lin

NeurIPS 2025

arXiv Code

Qwen-Image Technical Report

Y. Chu, Z. Cui, K. Dang, X. Deng, F. Huang, ... J. Lin, ...

arXiv preprint arXiv:2505.10338, 2025

arXiv Code

Reach out for collaborations or inquiries

justinlin930319@gmail.com