Head-dimension compression can reduce the memory required by the key-value (KV) cache, but existing methods often reconstruct full-dimensional representation…
机构:MIT
来源:arXiv 2610.28385 | AI4Papers 论文推荐平台