I am using the following code:
def ragged_from_3Dlist(list3d):
values = [item for list2d in list3d for list1d in list2d for item in list1d]
lens2d = [len(list1d) for list2d in list3d for list1d in list2d]
lens1d = [len(list2d) for list2d in list3d]
return tf.RaggedTensor.from_nested_row_lengths(values, [lens1d, lens2d])
Is there any better approach ?
