[Tests] fix things after #7013 (#7899)

* debugging

* save the resulting image

* check if order reversing works.

* checking values.

* up

* okay

* checking

* fix

* remove print

src/diffusers/loaders/lora.py

@@ -1294,7 +1294,7 @@ class LoraLoaderMixin:
                         text_encoder_module.lora_B[adapter_name].to(device)
                         # this is a param, not a module, so device placement is not in-place -> re-assign
                         if (
-                            hasattr(text_encoder, "lora_magnitude_vector")
+                            hasattr(text_encoder_module, "lora_magnitude_vector")
                             and text_encoder_module.lora_magnitude_vector is not None
                         ):
                             text_encoder_module.lora_magnitude_vector[

tests/lora/test_lora_layers_sdxl.py

@@ -194,7 +194,7 @@ class LoraSDXLIntegrationTests(unittest.TestCase):
         ).images
 
         images = images[0, -3:, -3:, -1].flatten()
-        expected = np.array([0.4468, 0.4087, 0.4134, 0.366, 0.3202, 0.3505, 0.3786, 0.387, 0.3535])
+        expected = np.array([0.4468, 0.4061, 0.4134, 0.3637, 0.3202, 0.365, 0.3786, 0.3725, 0.3535])
 
         max_diff = numpy_cosine_similarity_distance(expected, images)
         assert max_diff < 1e-4
@@ -283,7 +283,7 @@ class LoraSDXLIntegrationTests(unittest.TestCase):
 
         images = images[0, -3:, -3:, -1].flatten()
         # This way we also test equivalence between LoRA fusion and the non-fusion behaviour.
-        expected = np.array([0.4468, 0.4087, 0.4134, 0.366, 0.3202, 0.3505, 0.3786, 0.387, 0.3535])
+        expected = np.array([0.4468, 0.4061, 0.4134, 0.3637, 0.3202, 0.365, 0.3786, 0.3725, 0.3535])
 
         max_diff = numpy_cosine_similarity_distance(expected, images)
         assert max_diff < 1e-4