enable test_offloaded_cache_implementation on XPU (#37514)

Signed-off-by: YAO Matrix <matrix.yao@intel.com>
Co-authored-by: Yih-Dar <2521628+ydshieh@users.noreply.github.com>

tests/generation/test_utils.py

@@ -1947,7 +1947,7 @@ class GenerationTesterMixin:
                     )
 
     @parameterized.expand([("offloaded",)])  # ("offloaded_static",) TODO: @raushan fixme in some models (eg T5)
-    @require_torch_gpu
+    @require_torch_accelerator
     @pytest.mark.generate
     def test_offloaded_cache_implementation(self, cache_implementation):
         """Tests we can generate by indicating `cache_implementation` for each possible cache class"""