fix(imperative/utils): module parameters' name do not have scope after dumping

GitOrigin-RevId: 1497272294b125192d2356695259f622eb8d0bc5

imperative/python/megengine/jit/tracing.py

@@ -293,7 +293,9 @@ class trace:
             h = getattr(x, "_mixin_handle", -1)
             if h < 0 or (not self._capture_as_const and self._tinfo[h].exported):
                 h, info = self._new_handle()
-                name = auto_naming.get_scope() + "." + x.c_name if x.c_name else x._name
+                name = (
+                    auto_naming.get_scope() + "." + (x.c_name if x.c_name else x._name)
+                )
                 info.name = name
                 info.external = True
                 info.device = x.device
@@ -1123,11 +1125,11 @@ def apply_symbolic_mode(op: OpDef, *args: RawTensor):
     return outputs
 
 
-def apply_const_symbolic_mode(value, dtype, device):
+def apply_const_symbolic_mode(value, dtype, device, name):
     graph = active_trace._lazy_eval_graph
     # don't need to unset tracing
     # because varnode construction will ignore tracing flag
-    ret = RawTensor(graph.make_const(value, dtype=dtype, device=device))
+    ret = RawTensor(graph.make_const(value, dtype=dtype, device=device, name=name))
     if np.array(value).ndim == 0:
         setscalar(ret)
     return (ret,)
@@ -1175,7 +1177,7 @@ def apply_with_tracing(op: OpDef, *args: RawTensor):
 
 def apply_const_with_tracing(value, dtype, device, is_const, no_cache, name):
     if active_trace._symbolic:
-        outputs = apply_const_symbolic_mode(value, dtype, device)
+        outputs = apply_const_symbolic_mode(value, dtype, device, name)
     else:
         unset_tracing()
         outputs = (RawTensor(value, dtype, device, False, name),)

imperative/python/megengine/tensor.py

@@ -33,7 +33,7 @@ class Tensor(_Tensor, ArrayMethodMixin):
     _q_dict = None
 
     def __new__(
-        cls, data, dtype=None, device=None, is_const=False, no_cache=False, name=""
+        cls, data, dtype=None, device=None, is_const=False, no_cache=False, name=None
     ):
         if device is None:
             cn = get_default_device()

imperative/python/src/tensor.cpp

@@ -234,7 +234,8 @@ TensorWrapper::TensorWrapper(PyObject* args, PyObject* kwargs) {
             CompNode cn = tup[2].cast<CompNode>();
             bool is_const = tup[3].cast<bool>();
             bool no_cache = nargs == 6 ? tup[4].cast<bool>() : false;
-            std::string name = tup[nargs - 1].cast<std::string>();
+            std::string name;
+            if (tup[nargs - 1].ptr() != Py_None) name = tup[nargs - 1].cast<std::string>();
 
             // const op
             if (is_const && is_tracing) {

imperative/python/test/unit/functional/test_tensor.py

@@ -408,6 +408,15 @@ def test_copy_d2d():
     copy_test("gpu0:0", "gpu0:1")
 
 
+def test_name():
+    x = tensor(0)
+    assert x.name == ""
+    x.name = "x"
+    assert x.name == "x"
+    x = tensor(0, name="x")
+    assert x.name == "x"
+
+
 def test_q_dict():
     x = tensor(1)
     assert x.q_dict["scale"] is None