[CI/Build][Bugfix] Fix Qwen VL tests on CPU (#23818)

Signed-off-by: jiang1.li <jiang1.li@intel.com>
2025-08-28 19:57:05 +08:00
parent d99c3a4f7b
commit 67cee40da0
2 changed files with 14 additions and 14 deletions
--- a/vllm/model_executor/models/utils.py
+++ b/vllm/model_executor/models/utils.py
@@ -507,10 +507,10 @@ def merge_multimodal_embeddings(
        This updates ``inputs_embeds`` in place.
    """
    if isinstance(placeholder_token_id, list):
-        placeholder_token_id = torch.tensor(placeholder_token_id,
-                                            pin_memory=True).to(
-                                                device=input_ids.device,
-                                                non_blocking=True)
+        placeholder_token_id = torch.tensor(
+            placeholder_token_id,
+            pin_memory=is_pin_memory_available()).to(device=input_ids.device,
+                                                     non_blocking=True)
        return _merge_multimodal_embeddings(
            inputs_embeds,
            torch.isin(input_ids, placeholder_token_id),