EchoPress：查询无关的 KV Cache 剪枝方法

1. 12:00
   arXiv · Machine Learning Theory

   EchoPress：通过虚拟上下文重建实现查询无关的 KV Cache 剪枝

   EchoPress 是一种无需训练的 KV cache 剪枝方法，通过标准 prefill 阶段的 queries 和 keys 近似重建注意力，仅对第一个 chunk 重建以校准剩余上下文的重要性分数。

Current heat 9·Comparable peak 10（Oct 2）·Comparable change over 24 hours –

The trend compares only the same participants observed continuously; its range may be smaller than the current heat count. Move or click on the chart to inspect hourly heat; use the left and right arrow keys to switch.