Pengxiao Lin
- Initialization is Critical to Whether Transformers Fit Composite Functions by Reasoning or Memorizing
2024/05/08 by Zhongwang Zhang, Pengxiao Lin, Zhang, Zhongwang +7 · 6 citations
Computer Science · #FOS: Computer and information sciences #Machine Learning (cs.LG) #Neural Networks and Applications
- Complexity Control Facilitates Reasoning-Based Compositional Generalization in Transformers
2025/01/15 by Zhongwang Zhang, Pengxiao Lin, Zhang, Zhongwang +7 · 3 citations
Computer Science · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Neural Networks and Applications