diff --git a/tests/pipelines/test_pipelines_summarization.py b/tests/pipelines/test_pipelines_summarization.py
index 465dba9743c..613b9dca8e1 100644
--- a/tests/pipelines/test_pipelines_summarization.py
+++ b/tests/pipelines/test_pipelines_summarization.py
@@ -85,8 +85,9 @@ class SummarizationPipelineTests(unittest.TestCase):
                 and len(summarizer.model.trainable_weights) > 0
                 and "GPU" in summarizer.model.trainable_weights[0].device
             ):
-                with self.assertRaises(Exception):
-                    outputs = summarizer("This " * 1000)
+                if str(summarizer.device) == "cpu":
+                    with self.assertRaises(Exception):
+                        outputs = summarizer("This " * 1000)
         outputs = summarizer("This " * 1000, truncation=TruncationStrategy.ONLY_FIRST)
 
     @require_torch
diff --git a/tests/pipelines/test_pipelines_text_generation.py b/tests/pipelines/test_pipelines_text_generation.py
index 277c870b4d1..51f3cae5e31 100644
--- a/tests/pipelines/test_pipelines_text_generation.py
+++ b/tests/pipelines/test_pipelines_text_generation.py
@@ -493,17 +493,19 @@ class TextGenerationPipelineTests(unittest.TestCase):
             and text_generator.model.__class__.__name__ not in EXTRA_MODELS_CAN_HANDLE_LONG_INPUTS
         ):
             # Handling of large generations
-            with self.assertRaises((RuntimeError, IndexError, ValueError, AssertionError)):
-                text_generator("This is a test" * 500, max_new_tokens=20)
+            if str(text_generator.device) == "cpu":
+                with self.assertRaises((RuntimeError, IndexError, ValueError, AssertionError)):
+                    text_generator("This is a test" * 500, max_new_tokens=20)
 
             outputs = text_generator("This is a test" * 500, handle_long_generation="hole", max_new_tokens=20)
             # Hole strategy cannot work
-            with self.assertRaises(ValueError):
-                text_generator(
-                    "This is a test" * 500,
-                    handle_long_generation="hole",
-                    max_new_tokens=tokenizer.model_max_length + 10,
-                )
+            if str(text_generator.device) == "cpu":
+                with self.assertRaises(ValueError):
+                    text_generator(
+                        "This is a test" * 500,
+                        handle_long_generation="hole",
+                        max_new_tokens=tokenizer.model_max_length + 10,
+                    )
 
     @require_torch
     @require_accelerate