Skip to content
Hot eventLive

循环Transformer共享记忆提升质量并降低内存

1 reports1 sources16 hr ago updated

Get the story

AI overview

研究者提出在循环(Looped)Transformer 中使用共享记忆:预训练共享内存的循环语言模型,仅在第一次递归时写入 KV cache,后续递归读取该缓存并保留短窗口。该设计旨在提升模型质量并降低内存占用。报道来自 arXiv Machine Learning Theory(一手),时间为 2026-10-05 12:00。目前仅见此一篇报道,未见与其他报道在数字、时间或说法上的矛盾,也无更早报道可对比。

Generated from reports · updated 15 hr ago

Timeline

Follow the coverage from different angles.

Oct 5, 2026
  1. arXiv · Machine Learning Theory
    共享内存在 Looped Transformer 中的惊人效果

    研究者预训练了共享内存的循环语言模型:仅第一次递归写入 KV cache,后续递归读取它并保留短窗口。

Heat trend

Current heat 7·Comparable peak 10(Oct 5)·Comparable change over 24 hours –

02.557.510Oct5Oct5Oct5Oct6

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.