Nengo_DL function


#1

I am looking for a nengo_DL function that would take a tensor_layer as input and return the height and width. I have tried tensor.get_shape().as_list() but this is not working in nengo_dl.


#2

TensorNodes need to return a tf.Tensor, whereas tensor.get_shape().as_list() returns a list. What you want is tf.shape(tensor), which will return a tf.Tensor containing the shape.


#3

To be clear, I’m assuming that what you’re trying to do is write a TensorNode (or tensor_layer) that returns the shape of its input. If you just want to know what the shape of a tensor is, then tensor.get_shape() will work as well.


#4

Yes, this is perfect. Thank you!


#5

I tried using the tensor.get_shape(), however it is causing an error that says "function object has no attribute get_shape. What would be causing this? Know any fixes?


#6

That means that your tensor object is not a tf.Tensor. You can check the type of a Python variable using print(type(tensor)).