Open m4ttr4ymond opened 3 years ago
Any update on this? Get the same error. Here is the simplest code that triggers the error.
quat = tf.constant([[0., 0., 0., 1.]], dtype=tf.float64)
euler = tfg.geometry.transformation.euler.from_quaternion(quat)
print(euler)
@tf.function
def rot(quat):
euler = tfg.geometry.transformation.euler.from_quaternion(quat)
print(rot(quat))
This triggers the same error as @m4ttr4ymond
I was writing a data augmentation layer for a PointNet implementation and ran into what appears to be a bug in
tensorflow_graphics.shape.check_static(...)
, as seen on this line.Offending layer:
Error message:
It appears that
check_static
is expecting each element from.shape
to be a tensor, but in TF2 they're just ints. If I comment outcheck_static
fromfrom_euler
, the function works fine. Strangely enough, it seems to work fine for tensors in eager execution, and only seems to throw errors when usingDataset
objects with graph compilation.