2 papers
cs.LG2026
Softmax Linear Attention: Reclaiming Global Competition
Mingwei Xu, Xuan Lin, Xinnan Guo +2
While linear attention reduces the quadratic complexity of standard Transformers to linear time, it often lags behind in expressivity due to the removal of softmax normalization. T…
cs.CL2025
Homogeneous Keys, Heterogeneous Values: Exploiting Local KV Cache Asymmetry for Long-Context LLMs
Wanyun Cui, Mingwei Xu
Recent advances in Large Language Models (LLMs) have highlighted the critical importance of extending context length, yet the quadratic complexity of attention mechanisms poses sig…