diff --git a/vllm/distributed/kv_transfer/kv_connector/v1/example_connector.py b/vllm/distributed/kv_transfer/kv_connector/v1/example_connector.py
index a0279120ef0..8ed18894e8a 100644
--- a/vllm/distributed/kv_transfer/kv_connector/v1/example_connector.py
+++ b/vllm/distributed/kv_transfer/kv_connector/v1/example_connector.py
@@ -188,7 +188,9 @@ class ExampleConnector(KVConnectorBase_V1):
                 filename = self._generate_filename_debug(
                     layer_name, request.token_ids, request.mm_hashes
                 )
-                kv_cache = safetensors.torch.load_file(filename)["kv_cache"].cuda()
+                kv_cache = safetensors.torch.load_file(
+                    filename, device=str(kv_cache_layer.device)
+                )["kv_cache"]
                 if isinstance(attn_metadata, dict):
                     inject_kv_into_layer(
                         kv_cache_layer,