diff --git a/mindspore/_extends/builtin_operations.py b/mindspore/_extends/builtin_operations.py
index 494c4e2c38..a38135c295 100644
--- a/mindspore/_extends/builtin_operations.py
+++ b/mindspore/_extends/builtin_operations.py
@@ -157,8 +157,6 @@ def tuple_to_array(x):
 
 def stop_gradient(x):
     """Implement `stop_gradient`."""
-    if isinstance(x, Tensor):
-        return Tensor(x.asnumpy())
     return x
 
 
diff --git a/mindspore/ccsrc/pipeline/pynative/pynative_execute.cc b/mindspore/ccsrc/pipeline/pynative/pynative_execute.cc
index ed28b36a88..edb6620e14 100644
--- a/mindspore/ccsrc/pipeline/pynative/pynative_execute.cc
+++ b/mindspore/ccsrc/pipeline/pynative/pynative_execute.cc
@@ -314,6 +314,18 @@ py::object RunOpInVM(const OpExecInfoPtr &op_exec_info, PynativeStatusCode *stat
     py::tuple err_ret(0);
     return std::move(err_ret);
   }
+  if (op_exec_info->op_name == "stop_gradient" && py::isinstance<tensor::Tensor>(result)) {
+    py::tuple tuple_result(1);
+    auto tensor = py::cast<tensor::TensorPtr>(result);
+    MS_EXCEPTION_IF_NULL(tensor);
+    auto new_tensor = std::make_shared<tensor::Tensor>(tensor->data_type(), tensor->shape(), tensor->data_ptr());
+    new_tensor->set_device_address(tensor->device_address());
+    new_tensor->set_sync_status(tensor->sync_status());
+    tuple_result[0] = new_tensor;
+    *status = PYNATIVE_SUCCESS;
+    MS_LOG(INFO) << "RunOpInVM end";
+    return std::move(tuple_result);
+  }
 
   // execute op
   py::tuple tuple_result = py::make_tuple(result);