AttributeError: module 'sionna' has no attribute 'config'

[MizWong]

Hi,

When I run Simple_MIMO_Simulation, this error occurred, the sionna version is the latest released one by PIP.

[SebastianCa]

Hi @MizWong,

I am unable to reproduce the error with my setup, and it also works fine in Colab.

It seems to be an issue with your local installation. Could you please provide more details about your system setup?

[vishalgoyal316]

I am also facing same issue with my setup.

[SebastianCa]

Hi @vishalgoyal316,

could you please provide more details about your system setup?

[IhabAdarbeh]

Hallo, I also get the same error":(AttributeError: Exception encountered when calling layer 'lmmse_equalizer' (type LMMSEEqualizer).).module 'sionna' has no attribute 'config'" when I try to execute this section:

//simulation parameters
ebno_db = 10
batch_size = 200
// and call the model
b, b_hat = model(batch_size, ebno_db)
ber = sionna.utils.metrics.compute_ber(b, b_hat)
nb_bits = np.size(b.numpy())
print("BER: {:.4} at Eb/No of {} dB and {} simulated bits".format(ber.numpy(), ebno_db, nb_bits))

I am using Anaconda as the development environment, Jupyter Notebook as the compiler, and Python 3.8 with TensorFlow 2.14 und Sionna 0.18.0 version.

[SebastianCa]

Hi @IhabAdarbeh,

It seems like there might be an issue with your local Sionna installation. Here are a few things you can try to troubleshoot:

• Could you please run other notebooks to see if the issue is specific to the Simple MIMO notebook and the lmmse_equalizer?
• Try importing Sionna in an empty notebook (or Python environment) and print the version:

import sionna as sn
print(sn.__version__)

• The tutorial notebooks are designed to automatically install Sionna if it’s not found (to ensure compatibility with Google Colab). This could potentially interfere with your Anaconda virtual environment. For more details, you can refer to issue #471.

[sycskr]

Hello,i also meet this issue. Same problem as in question #123 : I was attempting to run the Sionna Tutorial on "Part 3: Advanced Link-Level Simulations" when I encountered this error in the LMMSE Equalizer. The notebook was taken directly from the site. The Sionna version I am using is 0.18.0. But reinstalling the environment as in #123 didn't solve the problem.

File [~\AppData\Local\Programs\Python\Python310\lib\site-packages\sionna\utils\tensors.py:242](http://localhost:8890/lab/tree/discover%20sionna/~/AppData/Local/Programs/Python/Python310/lib/site-packages/sionna/utils/tensors.py#line=241), in matrix_sqrt_inv(tensor)
    216 def matrix_sqrt_inv(tensor):
    217     r""" Computes the inverse square root of a Hermitian matrix.
    218 
    219     Given a batch of Hermitian positive definite matrices
   (...)
    240         See :py:attr:`~sionna.Config.xla_compat`.
    241     """
--> 242     if sn.config.xla_compat and not tf.executing_eagerly():
    243         s, u = tf.linalg.eigh(tensor)
    245         # Compute 1/sqrt of eigenvalues

AttributeError: Exception encountered when calling layer 'lmmse_equalizer_2' (type LMMSEEqualizer).

module 'sionna' has no attribute 'config'

Call arguments received by layer 'lmmse_equalizer_2' (type LMMSEEqualizer):
  • inputs=['tf.Tensor(shape=(200, 1, 1, 14, 256), dtype=complex64)', 'tf.Tensor(shape=(200, 1, 1, 1, 1, 14, 256), dtype=complex64)', 'tf.Tensor(shape=(1, 1, 1, 1, 1, 14, 256), dtype=float32)', 'tf.Tensor(shape=(), dtype=float32)']

I tried the sample given in question #123 and got the same error result

import sionna
import tensorflow as tf
from sionna.utils import matrix_sqrt_inv
matrix_sqrt_inv(tf.eye(5))

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[34], line 9
      7 import tensorflow as tf
      8 from sionna.utils import matrix_sqrt_inv
----> 9 matrix_sqrt_inv(tf.eye(5))

File [~\AppData\Local\Programs\Python\Python310\lib\site-packages\sionna\utils\tensors.py:242](http://localhost:8890/lab/tree/discover%20sionna/~/AppData/Local/Programs/Python/Python310/lib/site-packages/sionna/utils/tensors.py#line=241), in matrix_sqrt_inv(tensor)
    216 def matrix_sqrt_inv(tensor):
    217     r""" Computes the inverse square root of a Hermitian matrix.
    218 
    219     Given a batch of Hermitian positive definite matrices
   (...)
    240         See :py:attr:`~sionna.Config.xla_compat`.
    241     """
--> 242     if sn.config.xla_compat and not tf.executing_eagerly():
    243         s, u = tf.linalg.eigh(tensor)
    245         # Compute 1/sqrt of eigenvalues

AttributeError: module 'sionna' has no attribute 'config'

[jhoydis]

Could you try this again with the latest release 0.19?

[zylinzzz]

Same issue here with release 0.19.1:

line 231, in training_loop sn.Config.xla_compat = xla AttributeError: module 'sionna' has no attribute 'Config'. Did you mean: 'config'?

[merlinND]

Hello @zylinzzz,

The previous messages in this thread reported the error: AttributeError: module 'sionna' has no attribute 'config'

However you pasted: AttributeError: module 'sionna' has no attribute 'Config'. Did you mean: 'config'? Maybe you simply wrote sn.Config instead of sn.config?

[jhoydis]

Closed due to inactivity.