nhanitvn/TSAE

reconstruct, generate and num_layers

Opened this issue · 0 comments

hello,
I came across a few issues in your implementation.
1- reconstruct function always return zeros. apparently the output of fully connected layer is zeros!
2- I have tried to implement a generate function receiving z and synthesizing fake data but it was very challenging. Any tip would be appreciated.
3- if one increases num_layers, code crashes since some tensors dimension are not matched.
I would be grateful, if you can give me some hint concerning those issues and I would share my contribution for sure.