I am working on modification on VAE to make it Bayesian VAE. To do so, I follow the structure on tfp.experimental.nn.ConvolutionTransposeVariationalReparameterization | TensorFlow Probability
I have the following code that is part of VAE model:
That is work very well. However, Once I tried to modify it like the following.
I got the following error for the last line. Is there anybody that can help me resolve it:
tensorflow.python.framework.errors_impl.FailedPreconditionError: Could not find variable posterior_bias_loc_1. This could mean that the variable has been deleted. In TF1, it can also mean the variable is uninitialized. Debug info: container=localhost, status error message=Resource localhost/posterior_bias_loc_1/class tensorflow::Var does not exist.