阅读论文 《Train RNN as fast as CNN》

Train RNN as fast as CNN @(NLP)[cuDNN-LSTM, SRU] ABSTRACT 作者提出了一个简单循环单元的架构(SRU),能够轻易的实现并行化。SRU像CNN一样快,比优化后的LSTM实现快5到10倍。 In SRU, the majority of computation for each step is independent of the recurre
相关文章
相关标签/搜索