Self-attention plays a key role in long-sequence modeling, but its quadratic time complexity fundamentally limits the scaling of Large Language Models. Recen…
机构:DeepMind
来源:arXiv 2610.37645 | AI4Papers 论文推荐平台