STC2
STC2 copied to clipboard
jacoxu
Demo code for the paper STC2 which released three short text datasets for clustering and classification
Results
1
STC2 issues
Sort by
recently updated
recently updated
newest added
我在论文中看到：The network defines a trasnformation  which transforms an raw input text **x** to r-dimensional deep representation **h**. 这是不是就意味着最后的h是一个r维的向量呢？ 因为注意到文章在超参数一节中写道CNN的两层中共使用了两次Folding，我疑惑的是，48行的矩阵输入经过两次folding是不是就应该变成12行矩阵呢，想请问最后是怎么成为h向量的呢？
