M*******d 发帖数: 176 | 1 非常诡异
Traning set 600-ish time series data of shape (600,6)
It has 12 mutually exclusive classes.
First iteration has training accuracy only around 0.1
不管用什么loss function, 怎么 tune, condense the time step of the data,
training accuracy is always like under 0.1, and test accuracy is even lower
Anyone has any clues? Thanks a lot |
|