MXNet uses '0' as placeholder for unknown dimension, Keras uses 'None'

awslabs / keras-apache-mxnet

[DEPRECATED] Amazon Deep Learning's Keras with Apache MXNet support

https://github.com/awslabs/keras-apache-mxnet/wiki

Other

290 stars 65 forks source link

MXNet uses '0' as placeholder for unknown dimension, Keras uses 'None' #30

Open sandeep-krishnamurthy opened 6 years ago

sandeep-krishnamurthy commented 6 years ago

MXNet requires '0' as a placeholder for unknown dimension. Keras uses 'None' as a placeholder for unknown dimension.

This leads to few issues. To be fixed. Example test case to reproduce the error - tests/keras/test_sequential_model => test_clone_functional_model

andre-bergner commented 6 years ago

For me this a quite crucial bug. It also prevents to use certain keras layers. For instance, GaussianNoise does not work due to this. I see it as the task of the keras-mxnet backend to convert accordingly.