Author

Zihan Qiu

Observed name: Qiu, Zihan

Institutions: Alibaba Group (China), King Abdullah University of Science and Technology

Observed institutions: Alibaba Group (China), King Abdullah University of Science and Technology, University of Edinburgh

OpenAlex ID: A5111322472

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

2025 · Zihan Qiu, Zekun Wang, et al. · neural information processing systems · 20

Demons in the Detail: On Implementing Load Balancing Loss for Training Specialized Mixture-of-Expert Models

2025 · Zihan Qiu, Zeyu Huang, et al. · Annual Meeting of the Association for Computational Linguistics (ACL) · 9

Layerwise Recurrent Router for Mixture-of-Experts

2024 · Qiu, Zihan, Huang, Zeyu, et al. · arXiv (Cornell University) · 2

SlimQwen: Exploring the Pruning and Distillation in Large MoE Model Pre-training

2026 · Tang, Shengkun, Wang, Zekun, et al. · arXiv (Cornell University)

The Cancellation Hypothesis in Critic-Free RL: From Outcome Rewards to Token Credits

2026 · Cheng, Tianhao, Huang, Zeyu, et al. · arXiv (Cornell University)

Demons in the Detail: On Implementing Load Balancing Loss for Training Specialized Mixture-of-Expert Models

2025 · Qiu, Zihan, Huang, Zeyu, et al. · arXiv (Cornell University)

GW-MoE: Resolving Uncertainty in MoE Router with Global Workspace Theory

2024 · Wu, Haoze, Qiu, Zihan, et al. · arXiv (Cornell University)

Works count: 9

Citation count: 31

H-index: 2

I10-index: 1