today local_bar

Zijing Liu

Researcher, International Digital Economy Academy

1 paper at NeurIPS 2025

Homepage· OpenReview· Semantic Scholar· Google Scholar

Poster Session 5

Friday, December 5, 2025 · 11:00 AM → 2:00 PM

Exhibit Hall C,D,E

GPAS: Accelerating Convergence of LLM Pretraining via Gradient-Preserving Activation Scaling

#4102 · Tianhao Chen, Xin Xu, Zijing Liu, Pengxiang Li, Xinyuan Song, AJAY KUMAR JAISWAL, Fan Zhang, Jishan Hu, Yang Wang, Hao Chen, Shizhe Diao, Shiwei Liu, Yu Li, Lu Yin, Can Yang

We propose GPAS, a simple method that scales activations without scaling gradients to accelerate pretraining convergence of LLMs.