fix: when on gpu it is relevant to import tensorflow before torch

tests/wrappers/test_pytorch_wrapper.py

@@ -1,7 +1,7 @@
 import numpy as np
 
-import torch.nn as nn
 import tensorflow as tf
+import torch.nn as nn
 
 import pytest
 

xplique/wrappers/pytorch.py

@@ -27,7 +27,12 @@ def __init__(self, torch_model: "nn.Module", device: Union["torch.device", str],
                  is_channel_first: Optional[bool] = None
                  ): # pylint: disable=C0415,C0103
 
-        super().__init__()
+        try:
+            super().__init__()
+        except tf.errors.InternalError as error:
+            raise Exception("If you have a tensorflow InternalError with cudaGetDevice() here, \
+            it is possible that importing tensorflow before torch might resolve the issue."
+            ) from error
 
         try:
             # use PyTorch functionality