Machine learningTime-series forecasting
Reformer：长序列的高效Transformer
Reformer 是 Kitaev、Kaiser 和 Levskaya 于 ICLR 2020 提出的 Transformer 架构的一个高效变体。它解决了标准 Transformer 对于长序列而言 O(L²) 的内存和计算成本过高的问题。其关键创新是局部敏感哈希 (LSH) 注意力，它能在 O(L log L) 时间内近似全注意力，以及可逆残差层，它在训练期间显著降低了激活内存。
阅读完整方法
仅限会员
登录使用免费账户登录即可阅读本节。
Method map
The neighbourhood of related methods — select a node to explore.
来源
- Kitaev, N., Kaiser, Ł., & Levskaya, A. (2020). Reformer: The efficient transformer. ICLR. link ↗
如何引用本页
ScholarGate. (2026, June 2). Reformer (The Efficient Transformer). ScholarGate. https://scholargate.app/zh/deep-learning/reformer
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
Compare side by side →