17 papers
Mixture-of-Depths Attention
Lianghui Zhu, Yuxin Fang, Bencheng Liao +10
Scaling depth is a key driver for large language models (LLMs). Yet, as LLMs become deeper, they often suffer from signal degradation: informative features formed in shallow layers…
Stepsize anything: A unified learning rate schedule for budgeted-iteration training
Anda Tang, Yiming Dong, Yutao Zeng +2
The expanding computational costs and limited resources underscore the critical need for budgeted-iteration training, which aims to achieve optimal learning within predetermined it…
HybridNorm: Towards Stable and Efficient Transformer Training via Hybrid Normalization
Zhijian Zhuo, Yutao Zeng, Ya Wang +5
Transformers have become the de facto architecture for a wide range of machine learning tasks, particularly in large language models (LLMs). Despite their remarkable performance, m…
Virtual Width Networks
Seed, Baisheng Li, Banggu Wu +115
We introduce Virtual Width Networks (VWN), a framework that delivers the benefits of wider representations without incurring the quadratic cost of increasing the hidden size. VWN d…
GatePro: Parameter-Free Expert Selection Optimization for Mixture-of-Experts Models
Chen Zheng, Yuhang Cai, Deyi Liu +7
Modern large language models leverage Mixture-of-Experts (MoE) architectures for efficient scaling, but face a critical challenge: functionally similar experts are often selected s…
UltraMemV2: Memory Networks Scaling to 120B Parameters with Superior Long-Context Learning
Zihao Huang, Yu Bao, Qiyang Min +8
While Mixture of Experts (MoE) models achieve remarkable efficiency by activating only subsets of parameters, they suffer from high memory access costs during inference. Memory-lay…