Convert keras tensor to numpy

TensorFlow General Discussion
,
1

Hi,

I am new to this forum and this is my first post.

I am trying to create a generator which will take a random vector as input from a RNN. I would like this random vector to follow standard normal distribution.

The way I thought about implementing this was for the RNN to output a random seed and a custom layer which will take this random seed and create a normal distribution.

However, I am having difficulty implementing this as the RNN outputs a keras tensor and I having difficulty converting this within the custom layer to an integer seed.

Any thoughts from the community on how to implement this?

Thanks,

2

Hi @Saluja_Vijayan. Welcome to the Tensorflow forum.
It is not clear to me why you would need a RNN to generate tensors containing normally distributed data. Is it really what you are willing to achieve?
BTW what have you tried out so far and where are you stuck exactly?
Thank you.

1 Like