ci: fix xpu skip condition for test_model_parallel_beam_search

src/transformers/utils/import_utils.py

@@ -865,7 +865,7 @@ def is_ninja_available():
         return True
 
 
-def is_ipex_available():
+def is_ipex_available(min_version: str = ""):
     def get_major_and_minor_from_version(full_version):
         return str(version.parse(full_version).major) + "." + str(version.parse(full_version).minor)
 
@@ -880,6 +880,8 @@ def get_major_and_minor_from_version(full_version):
             f" but PyTorch {_torch_version} is found. Please switch to the matching version and run again."
         )
         return False
+    if min_version:
+        return version.parse(_ipex_version) >= version.parse(min_version)
     return True
 
 

tests/generation/test_utils.py

@@ -24,6 +24,7 @@
 
 import numpy as np
 import pytest
+from packaging import version
 from parameterized import parameterized
 
 from transformers import AutoConfig, is_torch_available, pipeline, set_seed
@@ -44,6 +45,7 @@
     slow,
     torch_device,
 )
+from transformers.utils import is_ipex_available
 
 from ..test_modeling_common import floats_tensor, ids_tensor
 from .test_framework_agnostic import GenerationIntegrationTestsMixin
@@ -675,10 +677,11 @@ def test_beam_search_generate_dict_outputs_use_cache(self):
     @require_torch_multi_accelerator
     @pytest.mark.generate
     def test_model_parallel_beam_search(self):
-        for model_class in self.all_generative_model_classes:
-            if "xpu" in torch_device:
-                return unittest.skip(reason="device_map='auto' does not work with XPU devices")
+        if "xpu" in torch_device:
+            if not (is_ipex_available("2.5") or version.parse(torch.__version__) >= version.parse("2.6")):
+                self.skipTest(reason="device_map='auto' does not work with XPU devices")
 
+        for model_class in self.all_generative_model_classes:
             if model_class._no_split_modules is None:
                 continue