avoid calling gc.collect and cuda.empty_cache (#34514)

* update

* update

* update

* update

* update

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

tests/models/idefics2/test_modeling_idefics2.py

@@ -15,7 +15,6 @@
 """Testing suite for the PyTorch Idefics2 model."""
 
 import copy
-import gc
 import tempfile
 import unittest
 from io import BytesIO
@@ -31,6 +30,7 @@ from transformers import (
     is_vision_available,
 )
 from transformers.testing_utils import (
+    cleanup,
     require_bitsandbytes,
     require_flash_attn,
     require_torch,
@@ -583,8 +583,7 @@ class Idefics2ForConditionalGenerationIntegrationTest(unittest.TestCase):
         )
 
     def tearDown(self):
-        gc.collect()
-        torch.cuda.empty_cache()
+        cleanup(torch_device, gc_collect=True)
 
     @slow
     @require_torch_multi_gpu