Closed mandar2812 closed 6 years ago
@mandar2812 Did you notice that the arguments to that class constructor changed? scaleOutput
was added and I think you may not be accounting for that change where you call the constructor.
@eaplatanios I think this error is happening because of difference in binary version of tensorflow-scala used in my code versus the version which was used to compile DynaML. Closing this issue for now.
Its possible that sometime after commit e0691f8e2aed5f796d642a04922e59c3f857f5af the dropout layer stopped working (tested so far on gpu).