diff --git a/mindspore/ccsrc/debug/trace.cc b/mindspore/ccsrc/debug/trace.cc
index 00abc2b9ad4..5d0ff86fbb7 100644
--- a/mindspore/ccsrc/debug/trace.cc
+++ b/mindspore/ccsrc/debug/trace.cc
@@ -230,11 +230,11 @@ std::string AnalyzedFuncGraphExporter::GetNodeType(const AnfNodePtr &node) {
   auto ctx = node_cfg_->context();
   auto engine = node_cfg_->engine();
   auto cfg = engine->MakeConfig(node, ctx);
-  auto abs = engine->cache().GetValue(cfg);
-  if (abs == nullptr) {
+  auto eval_result = engine->cache().GetValue(cfg);
+  if (eval_result == nullptr || eval_result->abstract() == nullptr) {
     return "Undefined";
   }
-
+  auto abs = eval_result->abstract();
   auto dtype = abs->BuildType();
   auto shape = abs->BuildShape();
   std::ostringstream oss;
diff --git a/mindspore/ccsrc/ir/primitive_base.h b/mindspore/ccsrc/ir/primitive_base.h
index 1b347a30739..78623f8542d 100644
--- a/mindspore/ccsrc/ir/primitive_base.h
+++ b/mindspore/ccsrc/ir/primitive_base.h
@@ -42,7 +42,11 @@ enum PrimType {
 class Primitive : public Named {
  public:
   explicit Primitive(const std::string &name, const bool is_base = true, const PrimType prim_type = kPrimTypeBuiltIn)
-      : Named(name), is_base_(is_base), has_signature_(false), prim_type_(prim_type) {}
+      : Named(name),
+        is_base_(is_base),
+        has_signature_(false),
+        prim_type_(prim_type),
+        record_evaluate_add_attr_(false) {}
 
   Primitive(const Primitive &prim)
       : Named(prim),
@@ -50,14 +54,23 @@ class Primitive : public Named {
         instance_name_(prim.instance_name_),
         is_base_(prim.is_base_),
         has_signature_(prim.has_signature_),
-        prim_type_(prim.prim_type_) {}
+        prim_type_(prim.prim_type_),
+        record_evaluate_add_attr_(false) {}
 
   MS_DECLARE_PARENT(Primitive, Named);
 
   abstract::AbstractBasePtr ToPrimAbstract(const AnfNodePtr &anf_node);
   std::string ToString() const override { return name(); }
+  void BeginRecordAddAttr() {
+    evaluate_added_attrs_.clear();
+    record_evaluate_add_attr_ = true;
+  }
+  void EndRecordAddAttr() { record_evaluate_add_attr_ = false; }
   Primitive &AddAttr(const std::string &name, const ValuePtr &attr) {
     attrs_[name] = attr;
+    if (record_evaluate_add_attr_) {
+      evaluate_added_attrs_[name] = attr;
+    }
     return *this;
   }
 
@@ -80,6 +93,7 @@ class Primitive : public Named {
   py::function hook() const { return hook_; }
 
   const std::unordered_map<std::string, ValuePtr> &attrs() const { return attrs_; }
+  std::unordered_map<std::string, ValuePtr> &evaluate_added_attrs() { return evaluate_added_attrs_; }
 
   // if Primitive has any attribute, for Primitives like scalar_add, return, etc, don't have any attribute.
   bool HasAttr() const { return !attrs_.empty(); }
@@ -106,6 +120,7 @@ class Primitive : public Named {
 
  protected:
   std::unordered_map<std::string, ValuePtr> attrs_;
+  std::unordered_map<std::string, ValuePtr> evaluate_added_attrs_;
 
  private:
   std::string instance_name_;
@@ -113,6 +128,7 @@ class Primitive : public Named {
   bool is_base_;
   bool has_signature_;
   PrimType prim_type_;
+  bool record_evaluate_add_attr_;
 };
 
 inline std::ostream &operator<<(std::ostream &os, const PrimitivePtr &p) {
diff --git a/mindspore/ccsrc/operator/prim_structures.cc b/mindspore/ccsrc/operator/prim_structures.cc
index 31d2bff43d8..7b0bba98a59 100644
--- a/mindspore/ccsrc/operator/prim_structures.cc
+++ b/mindspore/ccsrc/operator/prim_structures.cc
@@ -377,10 +377,10 @@ AbstractBasePtr InferImplListMap(const AnalysisEnginePtr &engine, const Primitiv
     }
     subargs.push_back(AbstractJoin(l_ptr->elements()));
   }
-  AbstractBasePtr engin_exc = engine->Execute(fn, subargs);
+  EvalResultPtr engin_exc = engine->Execute(fn, subargs);
   AbstractBasePtrList result;
   for (std::size_t i = 1; i < args_spec_list.size(); i++) {
-    result.push_back(engin_exc);
+    result.push_back(engin_exc->abstract());
   }
   return std::make_shared<AbstractList>(result);
 }
@@ -398,8 +398,9 @@ AbstractBasePtr InferImplListReduce(const AnalysisEnginePtr &engine, const Primi
   AbstractBasePtr list_type = AbstractJoin(lst->elements());
   auto result1 = engine->Execute(fn, lst->elements());
   auto result2 = engine->Execute(fn, {dflt, list_type});
-  MS_EXCEPTION_IF_NULL(result1);
-  return result1->Join(result2);
+  MS_EXCEPTION_IF_NULL(result1->abstract());
+  MS_EXCEPTION_IF_NULL(result2->abstract());
+  return result1->abstract()->Join(result2->abstract());
 }
 
 AbstractBasePtr InferImplTupleReversed(const AnalysisEnginePtr &, const PrimitivePtr &primitive,
diff --git a/mindspore/ccsrc/pipeline/static_analysis/evaluator.cc b/mindspore/ccsrc/pipeline/static_analysis/evaluator.cc
index fde8558d3dd..542abc62058 100644
--- a/mindspore/ccsrc/pipeline/static_analysis/evaluator.cc
+++ b/mindspore/ccsrc/pipeline/static_analysis/evaluator.cc
@@ -89,7 +89,7 @@ static std::vector<AnfNodePtr> FastShadowSort(const AnfNodePtr &ret_node) {
   return sorted_nodes;
 }
 
-AbstractBasePtr BaseFuncGraphEvaluator::Eval(AnalysisEnginePtr engine, const AbstractBasePtrList &args_spec_list) {
+EvalResultPtr BaseFuncGraphEvaluator::Eval(AnalysisEnginePtr engine, const AbstractBasePtrList &args_spec_list) {
   FuncGraphPtr fg = GetFuncGraph(engine, args_spec_list);
   MS_EXCEPTION_IF_NULL(fg);
   std::size_t nargs = fg->parameters().size();
@@ -106,7 +106,7 @@ AbstractBasePtr BaseFuncGraphEvaluator::Eval(AnalysisEnginePtr engine, const Abs
     const auto &arg = args_spec_list[i];
     const auto &node = parameters[i];
     AnfNodeConfigPtr conf = engine->MakeConfig(node, graph_context_);
-    engine->cache().set_value(conf, arg);
+    engine->cache().set_value(conf, std::make_shared<EvalResult>(arg, nullptr));
   }
   const AnfNodePtr &func_node = fg->get_return();
 
@@ -118,14 +118,14 @@ AbstractBasePtr BaseFuncGraphEvaluator::Eval(AnalysisEnginePtr engine, const Abs
     const auto &node = *it;
     AnfNodeConfigPtr node_conf = engine->MakeConfig(node, graph_context_);
     MS_LOG(DEBUG) << "Analysis node begin, func graph: " << fg->ToString() << ", node_conf: " << node_conf->ToString();
-    ret_base = engine->GetEvaluatedValue(node_conf);
+    ret_base = engine->GetEvaluatedValue(node_conf)->abstract();
     MS_LOG(DEBUG) << "Analysis node end, func graph: " << fg->ToString() << ", node_conf: " << node_conf->ToString()
                   << ", abstract: " << ret_base->ToString();
   }
 
   MS_EXCEPTION_IF_NULL(ret_base);
-  MS_LOG(DEBUG) << "BaseFuncGraph " << fg->ToString() << " Eval end, evaluated abstract: " << ret_base->ToString();
-  return ret_base;
+  MS_LOG(DEBUG) << "BaseFuncGraph " << fg->ToString() << " eval end, evaluated abstract: " << ret_base->ToString();
+  return std::make_shared<EvalResult>(ret_base, nullptr);
 }
 
 AbstractBasePtrList FuncGraphEvaluator::NormalizeArgs(const AbstractBasePtrList &args_spec_list) const {
@@ -236,15 +236,14 @@ FuncGraphPtr MetaFuncGraphEvaluator::GetFuncGraph(AnalysisEnginePtr engine, cons
   return cloned_func_graph;
 }
 
-AbstractBasePtr Evaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
-                               AnfNodeConfigPtr out_conf) {
+EvalResultPtr Evaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf) {
   const std::string &evaluator_name = ToString();
 
   AbstractBasePtrList args_spec_list;
   (void)std::transform(args_conf_list.begin(), args_conf_list.end(), std::back_inserter(args_spec_list),
                        [](const ConfigPtr &conf) -> AbstractBasePtr {
                          MS_EXCEPTION_IF_NULL(conf);
-                         return conf->GetEvaluatedValue();
+                         return conf->GetEvaluatedValue()->abstract();
                        });
   args_spec_list = NormalizeArgs(args_spec_list);
   args_spec_list = BroadenUndeterminedArgs(args_spec_list);
@@ -254,79 +253,79 @@ AbstractBasePtr Evaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &ar
   auto iter = cache_->find(args_spec_list);
   if (iter == cache_->end()) {
     MS_LOG(DEBUG) << evaluator_name << " cache miss, call Eval().";
-    AbstractBasePtr ret = Eval(engine, args_spec_list);
-    if (ret == nullptr) {
+    EvalResultPtr ret = Eval(engine, args_spec_list);
+    if (ret->abstract() == nullptr) {
       EvalFailLogging(shared_from_base<Evaluator>(), args_spec_list, out_conf);
       MS_LOG(EXCEPTION) << "Evaluator " << evaluator_name << " result is nullptr.";
     }
-    MS_EXCEPTION_IF_NULL(ret);
-    MS_LOG(DEBUG) << evaluator_name << " set cache. return: " << ret->ToString() << ".";
+    MS_LOG(DEBUG) << evaluator_name << " set cache. return: " << ret->abstract()->ToString() << ".";
     (*cache_)[args_spec_list] = ret;
     trace::TraceGraphEvalLeave(shared_from_base<Evaluator>());
     return ret;
   } else {
     MS_EXCEPTION_IF_NULL(iter->second);
-    MS_LOG(DEBUG) << evaluator_name << " cache hit. return: " << iter->second->ToString() << ".";
+    MS_EXCEPTION_IF_NULL(iter->second->abstract());
+    MS_LOG(DEBUG) << evaluator_name << " cache hit. return: " << iter->second->abstract()->ToString() << ".";
     trace::TraceGraphEvalLeave(shared_from_base<Evaluator>());
     return iter->second;
   }
 }
 
-AbstractBasePtr TrivialPrimEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
-                                          AnfNodeConfigPtr) {
+EvalResultPtr TrivialPrimEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
+                                        AnfNodeConfigPtr) {
   AbstractBasePtrList args_spec_list;
   (void)std::transform(args_conf_list.begin(), args_conf_list.end(), std::back_inserter(args_spec_list),
                        [](const ConfigPtr &conf) -> AbstractBasePtr {
                          MS_EXCEPTION_IF_NULL(conf);
-                         return conf->GetEvaluatedValue();
+                         return conf->GetEvaluatedValue()->abstract();
                        });
-  AbstractBasePtr ret = EvalPrim(engine, args_spec_list);
+  EvalResultPtr ret = EvalPrim(engine, args_spec_list);
   return ret;
 }
 
-AbstractBasePtr TransitionPrimEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
-                                             AnfNodeConfigPtr out_conf) {
+EvalResultPtr TransitionPrimEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
+                                           AnfNodeConfigPtr out_conf) {
   AbstractBasePtrList args_spec_list;
   (void)std::transform(args_conf_list.begin(), args_conf_list.end(), std::back_inserter(args_spec_list),
                        [](const ConfigPtr &conf) -> AbstractBasePtr {
                          MS_EXCEPTION_IF_NULL(conf);
-                         return conf->GetEvaluatedValue();
+                         return conf->GetEvaluatedValue()->abstract();
                        });
   if (args_conf_list.size() == 0) {
     MS_LOG(EXCEPTION) << "Size should greater than 0";
   }
-  AbstractBasePtr ret = EvalPrim(engine, args_spec_list, args_conf_list[0], out_conf);
+  EvalResultPtr ret = EvalPrim(engine, args_spec_list, args_conf_list[0], out_conf);
   // No need to cache.
   return ret;
 }
 
-AbstractBasePtr SymbolicPrimEvaluator::Run(AnalysisEnginePtr, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr) {
-  AbstractBasePtr ret = EvalPrim(args_conf_list);
+EvalResultPtr SymbolicPrimEvaluator::Run(AnalysisEnginePtr, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr) {
+  EvalResultPtr ret = EvalPrim(args_conf_list);
   return ret;
 }
 
-AbstractBasePtr TrackedEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
-                                      AnfNodeConfigPtr out_conf) {
+EvalResultPtr TrackedEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
+                                    AnfNodeConfigPtr out_conf) {
   AbstractBasePtrList args_spec_list;
   (void)std::transform(args_conf_list.begin(), args_conf_list.end(), std::back_inserter(args_spec_list),
                        [](const ConfigPtr &conf) -> AbstractBasePtr {
                          MS_EXCEPTION_IF_NULL(conf);
-                         return conf->GetEvaluatedValue();
+                         return conf->GetEvaluatedValue()->abstract();
                        });
-  AbstractBasePtr ret = sub_evaluator_->Run(engine, args_conf_list, out_conf);
+  EvalResultPtr ret = sub_evaluator_->Run(engine, args_conf_list, out_conf);
   // Don't lookup from cache, as different out_conf with same node but different context
   // may add different entry to anfnode_config_map_, like getattr primitive.
   (*cache_)[args_spec_list] = ret;
   return ret;
 }
 
-AbstractBasePtr PartialAppEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
-                                         AnfNodeConfigPtr out_conf) {
+EvalResultPtr PartialAppEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
+                                       AnfNodeConfigPtr out_conf) {
   AbstractBasePtrList args_spec_list;
   (void)std::transform(args_conf_list.begin(), args_conf_list.end(), std::back_inserter(args_spec_list),
                        [](const ConfigPtr &conf) -> AbstractBasePtr {
                          MS_EXCEPTION_IF_NULL(conf);
-                         return conf->GetEvaluatedValue();
+                         return conf->GetEvaluatedValue()->abstract();
                        });
   MS_EXCEPTION_IF_NULL(cache_);
   auto iter = cache_->find(args_spec_list);
@@ -341,17 +340,18 @@ AbstractBasePtr PartialAppEvaluator::Run(AnalysisEnginePtr engine, const ConfigP
 
   (void)std::transform(args_spec_list.begin(), args_spec_list.end(), std::back_inserter(partial_args_conf_list),
                        [](const AbstractBasePtr &arg) -> ConfigPtr { return std::make_shared<VirtualConfig>(arg); });
-  AbstractBasePtr ret = evaluator_->Run(engine, partial_args_conf_list, out_conf);
+  EvalResultPtr ret = evaluator_->Run(engine, partial_args_conf_list, out_conf);
+
   (*cache_)[args_spec_list] = ret;
   return ret;
 }
 
-AbstractBasePtr JEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr) {
+EvalResultPtr JEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr) {
   AbstractBasePtrList args_spec_list;
   (void)std::transform(args_conf_list.begin(), args_conf_list.end(), std::back_inserter(args_spec_list),
                        [](const ConfigPtr &conf) -> AbstractBasePtr {
                          MS_EXCEPTION_IF_NULL(conf);
-                         return conf->GetEvaluatedValue();
+                         return conf->GetEvaluatedValue()->abstract();
                        });
   MS_EXCEPTION_IF_NULL(cache_);
   auto iter = cache_->find(args_spec_list);
@@ -360,7 +360,7 @@ AbstractBasePtr JEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &a
   }
 
   // Call the original evaluator, get the result: y = f(x)
-  AbstractBasePtr result = evaluator_->Run(engine, args_conf_list, nullptr);
+  EvalResultPtr result = evaluator_->Run(engine, args_conf_list, nullptr);
   // Build a virtual function: bprop_f which use sense of y as input, return sense of function free variable and input
   // parameters. (sense_f, sense_x, ...)(*bpro_f) (sense_y)
   AbstractBasePtrList bparams;
@@ -369,16 +369,18 @@ AbstractBasePtr JEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &a
     args_spec_list.begin(), args_spec_list.end(), std::back_inserter(bparams),
     [](const AbstractBasePtr &arg_spec) -> AbstractBasePtr { return SensitivityTransform(arg_spec); });
   AbstractBasePtr bparams_final = std::make_shared<AbstractTuple>(bparams);
-  AbstractFunctionPtr bprop = std::make_shared<VirtualAbstractClosure>(SensitivityTransform(result), bparams_final);
+  AbstractFunctionPtr bprop =
+    std::make_shared<VirtualAbstractClosure>(SensitivityTransform(result->abstract()), bparams_final);
 
   // J(f)(J(x)) return a tuple (y, bprop_f)
-  AbstractBasePtrList jargs = {result, bprop};
+  AbstractBasePtrList jargs = {result->abstract(), bprop};
   AbstractBasePtr jtuple = std::make_shared<AbstractTuple>(jargs);
-  (*cache_)[args_spec_list] = jtuple;
-  return jtuple;
+  auto infer_reuslt = std::make_shared<EvalResult>(jtuple, std::make_shared<AttrValueMap>());
+  (*cache_)[args_spec_list] = infer_reuslt;
+  return infer_reuslt;
 }
 
-AbstractBasePtr VirtualEvaluator::Eval(AnalysisEnginePtr, const AbstractBasePtrList &args_spec_list) {
+EvalResultPtr VirtualEvaluator::Eval(AnalysisEnginePtr, const AbstractBasePtrList &args_spec_list) {
   if (args_spec_list.size() != args_spec_list_.size()) {
     MS_LOG(EXCEPTION) << "Arguments mismatch, parameters no: " << args_spec_list_.size()
                       << ", arguments no: " << args_spec_list.size();
@@ -388,7 +390,7 @@ AbstractBasePtr VirtualEvaluator::Eval(AnalysisEnginePtr, const AbstractBasePtrL
     MS_EXCEPTION_IF_NULL(args_spec_list[i]);
     (void)args_spec_list[i]->Join(args_spec_list_[i]);
   }
-  return output_;
+  return std::make_shared<EvalResult>(output_, std::make_shared<AttrValueMap>());
 }
 }  // namespace abstract
 }  // namespace mindspore
diff --git a/mindspore/ccsrc/pipeline/static_analysis/evaluator.h b/mindspore/ccsrc/pipeline/static_analysis/evaluator.h
index 0f7e1fe0520..c7a004ac44e 100644
--- a/mindspore/ccsrc/pipeline/static_analysis/evaluator.h
+++ b/mindspore/ccsrc/pipeline/static_analysis/evaluator.h
@@ -29,21 +29,28 @@
 namespace mindspore {
 namespace abstract {
 using EvaluatorCacheMap =
-  std::unordered_map<AbstractBasePtrList, AbstractBasePtr, AbstractBasePtrListHasher, AbstractBasePtrListEqual>;
+  std::unordered_map<AbstractBasePtrList, EvalResultPtr, AbstractBasePtrListHasher, AbstractBasePtrListEqual>;
 using EvaluatorCacheMapPtr = std::shared_ptr<EvaluatorCacheMap>;
 
+using EvaluatorAttrMap =
+  std::unordered_map<AbstractBasePtrList, AttrValueMapPtr, AbstractBasePtrListHasher, AbstractBasePtrListEqual>;
+using EvaluatorAttrMapPtr = std::shared_ptr<EvaluatorAttrMap>;
+
 class Evaluator : public Base {
  public:
-  explicit Evaluator(const std::string &id) : cache_(std::make_shared<EvaluatorCacheMap>()), identifier_(id) {}
+  explicit Evaluator(const std::string &id)
+      : cache_(std::make_shared<EvaluatorCacheMap>()),
+        attr_cache_(std::make_shared<EvaluatorAttrMap>()),
+        identifier_(id) {}
   ~Evaluator() override = default;
   MS_DECLARE_PARENT(Evaluator, Base);
 
   // difference between Run() and Eval():
   // Run() will be called with ConfigPtrList, but Eval() will be called with AbstractBasePtr.
   // Run() will modify cache_ member, so it cannot marked as const;
-  virtual AbstractBasePtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf);
+  virtual EvalResultPtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf);
 
-  virtual AbstractBasePtr Eval(AnalysisEnginePtr engine, const AbstractBasePtrList &args_spec_list) = 0;
+  virtual EvalResultPtr Eval(AnalysisEnginePtr engine, const AbstractBasePtrList &args_spec_list) = 0;
 
   virtual AbstractBasePtrList NormalizeArgs(const AbstractBasePtrList &args_spec_list) const { return args_spec_list; }
 
@@ -58,9 +65,10 @@ class Evaluator : public Base {
   virtual void set_bound_node(const AnfNodePtr &node) { bound_node_ = AnfNodeWeakPtr(node); }
 
   EvaluatorCacheMapPtr &cache() { return cache_; }
+  EvaluatorAttrMapPtr &attr_cache() { return attr_cache_; }
 
   EvaluatorCacheMapPtr cache_;
-
+  EvaluatorAttrMapPtr attr_cache_;
   std::string identifier_;
 
   AnfNodeWeakPtr bound_node_;
@@ -71,7 +79,7 @@ class PrimEvaluator : public Evaluator {
   explicit PrimEvaluator(const std::string &id) : Evaluator(id) {}
   ~PrimEvaluator() override = default;
   MS_DECLARE_PARENT(PrimEvaluator, Evaluator);
-  AbstractBasePtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) final {
+  EvalResultPtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) final {
     MS_LOG(EXCEPTION) << "Eval() should not be called, Run() method should be called";
   }
 };
@@ -81,8 +89,8 @@ class TrivialPrimEvaluator : public PrimEvaluator {
   explicit TrivialPrimEvaluator(const std::string &id) : PrimEvaluator(id) {}
   ~TrivialPrimEvaluator() override = default;
   MS_DECLARE_PARENT(TrivialPrimEvaluator, PrimEvaluator);
-  AbstractBasePtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf) final;
-  virtual AbstractBasePtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list) = 0;
+  EvalResultPtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf) final;
+  virtual EvalResultPtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list) = 0;
 };
 
 class TransitionPrimEvaluator : public PrimEvaluator {
@@ -90,10 +98,10 @@ class TransitionPrimEvaluator : public PrimEvaluator {
   explicit TransitionPrimEvaluator(const std::string &id) : PrimEvaluator(id) {}
   ~TransitionPrimEvaluator() override = default;
   MS_DECLARE_PARENT(TransitionPrimEvaluator, PrimEvaluator);
-  AbstractBasePtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf) final;
+  EvalResultPtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf) final;
   // Parameter in_conf0 : the first element in args_conf_list;
-  virtual AbstractBasePtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list,
-                                   const ConfigPtr &in_conf0, const AnfNodeConfigPtr &out_conf) = 0;
+  virtual EvalResultPtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list,
+                                 const ConfigPtr &in_conf0, const AnfNodeConfigPtr &out_conf) = 0;
 };
 
 class SymbolicPrimEvaluator : public PrimEvaluator {
@@ -101,8 +109,8 @@ class SymbolicPrimEvaluator : public PrimEvaluator {
   explicit SymbolicPrimEvaluator(const std::string &id) : PrimEvaluator(id) {}
   ~SymbolicPrimEvaluator() override = default;
   MS_DECLARE_PARENT(SymbolicPrimEvaluator, PrimEvaluator);
-  AbstractBasePtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf) final;
-  virtual AbstractBasePtr EvalPrim(const ConfigPtrList &args_conf_list) = 0;
+  EvalResultPtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf) final;
+  virtual EvalResultPtr EvalPrim(const ConfigPtrList &args_conf_list) = 0;
 };
 
 // Evaluator will be stored in AnalysisEngine.constructors_
@@ -113,7 +121,7 @@ class DummyEvaluator : public Evaluator {
   DummyEvaluator() : Evaluator("dummy") {}
   ~DummyEvaluator() override = default;
   MS_DECLARE_PARENT(DummyEvaluator, Evaluator);
-  AbstractBasePtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override { return nullptr; }
+  EvalResultPtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override { return nullptr; }
 };
 
 // Wrap another evaluator to track a subset of uses.
@@ -139,11 +147,10 @@ class TrackedEvaluator : public Evaluator {
     bound_node_ = AnfNodeWeakPtr(node);
   }
 
-  AbstractBasePtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
+  EvalResultPtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
     MS_LOG(EXCEPTION) << "Eval() should not be called, Run() method should be called";
   }
-  AbstractBasePtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
-                      AnfNodeConfigPtr out_conf) override;
+  EvalResultPtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf) override;
   std::string ToString() const override { return identifier_ + "_" + sub_evaluator_->ToString(); }
 
  private:
@@ -158,7 +165,7 @@ class BaseFuncGraphEvaluator : public Evaluator {
   ~BaseFuncGraphEvaluator() override = default;
   MS_DECLARE_PARENT(BaseFuncGraphEvaluator, Evaluator);
 
-  AbstractBasePtr Eval(AnalysisEnginePtr engine, const AbstractBasePtrList &args_spec_list) override;
+  EvalResultPtr Eval(AnalysisEnginePtr engine, const AbstractBasePtrList &args_spec_list) override;
 
   virtual FuncGraphPtr GetFuncGraph(AnalysisEnginePtr engine, const AbstractBasePtrList &args_spec_list) = 0;
 
@@ -238,12 +245,12 @@ class PartialAppEvaluator : public Evaluator {
     }
     bound_node_ = AnfNodeWeakPtr(node);
   }
-  AbstractBasePtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
+
+  EvalResultPtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
     MS_LOG(EXCEPTION) << "Should not be called, Run() method should be called";
   }
 
-  AbstractBasePtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
-                      AnfNodeConfigPtr out_conf) override;
+  EvalResultPtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf) override;
   std::string ToString() const override { return identifier_ + "_" + evaluator_->ToString(); }
 
  private:
@@ -258,7 +265,7 @@ class VirtualEvaluator : public Evaluator {
   ~VirtualEvaluator() override = default;
   MS_DECLARE_PARENT(VirtualEvaluator, Evaluator);
 
-  AbstractBasePtr Eval(AnalysisEnginePtr engine, const AbstractBasePtrList &args_spec_list) override;
+  EvalResultPtr Eval(AnalysisEnginePtr engine, const AbstractBasePtrList &args_spec_list) override;
   std::string ToString() const override { return identifier_; }
 
  private:
@@ -285,11 +292,11 @@ class JEvaluator : public Evaluator {
     }
     bound_node_ = AnfNodeWeakPtr(node);
   }
-  AbstractBasePtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
+
+  EvalResultPtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
     MS_LOG(EXCEPTION) << "Should not be called, Run() method should be called";
   }
-  AbstractBasePtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
-                      AnfNodeConfigPtr out_conf) override;
+  EvalResultPtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list, AnfNodeConfigPtr out_conf) override;
   std::string ToString() const override { return identifier_ + "_" + evaluator_->ToString(); }
 
  private:
diff --git a/mindspore/ccsrc/pipeline/static_analysis/prim.cc b/mindspore/ccsrc/pipeline/static_analysis/prim.cc
index dbd08c5f58a..f8d53c71496 100644
--- a/mindspore/ccsrc/pipeline/static_analysis/prim.cc
+++ b/mindspore/ccsrc/pipeline/static_analysis/prim.cc
@@ -135,13 +135,17 @@ PrimitiveEvalImplMap &GetPrimitiveToEvalImplMap() {
 
 using mindspore::parse::PyObjectWrapper;
 
-AbstractBasePtr StandardPrimEvaluator::EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args) {
+EvalResultPtr StandardPrimEvaluator::EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args) {
+  prim_->BeginRecordAddAttr();
   AbstractBasePtr abs_base = eval_impl_(engine, prim_, args);
-  return abs_base;
+  prim_->EndRecordAddAttr();
+  auto added_attrs = prim_->evaluate_added_attrs();
+  auto infer_result = std::make_shared<EvalResult>(abs_base, std::make_shared<AttrValueMap>(added_attrs));
+  return infer_result;
 }
 
-AbstractBasePtr DoSignatureEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
-                                          AnfNodeConfigPtr out_conf) {
+EvalResultPtr DoSignatureEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
+                                        AnfNodeConfigPtr out_conf) {
   AbstractBasePtrList args_spec_list;
   if (!prim_->isa<prim::DoSignaturePrimitive>()) {
     MS_LOG(EXCEPTION) << "Primitive should be DoSignature, but " << prim_->ToString();
@@ -161,7 +165,7 @@ AbstractBasePtr DoSignatureEvaluator::Run(AnalysisEnginePtr engine, const Config
   AnfNodePtrList args_inputs{out_node_inputs.begin() + 1, out_node_inputs.end()};
 
   (void)std::transform(args_conf_list.begin(), args_conf_list.end(), std::back_inserter(args_spec_list),
-                       [](const ConfigPtr &ref) -> AbstractBasePtr { return ref->GetEvaluatedValue(); });
+                       [](const ConfigPtr &ref) -> AbstractBasePtr { return ref->GetEvaluatedValue()->abstract(); });
 
   ScopePtr scope = kDefaultScope;
   if (out_conf != nullptr) {
@@ -212,8 +216,8 @@ static AbstractBasePtrList GetUnpackGraphSpecArgsList(AbstractBasePtrList args_s
   return graph_specialize_args;
 }
 
-AbstractBasePtr UnpackGraphEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
-                                          AnfNodeConfigPtr out_conf) {
+EvalResultPtr UnpackGraphEvaluator::Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
+                                        AnfNodeConfigPtr out_conf) {
   if (out_conf->node() == nullptr || !out_conf->node()->isa<CNode>()) {
     MS_LOG(EXCEPTION) << "Node of out_conf should be CNode";
   }
@@ -232,7 +236,7 @@ AbstractBasePtr UnpackGraphEvaluator::Run(AnalysisEnginePtr engine, const Config
   AnfNodePtrList args_inputs{out_node_inputs.begin() + 1, out_node_inputs.end()};
   AbstractBasePtrList args_spec_list;
   (void)std::transform(args_conf_list.begin(), args_conf_list.end(), std::back_inserter(args_spec_list),
-                       [](const ConfigPtr &ref) -> AbstractBasePtr { return ref->GetEvaluatedValue(); });
+                       [](const ConfigPtr &ref) -> AbstractBasePtr { return ref->GetEvaluatedValue()->abstract(); });
   // get the forward graph
   MS_EXCEPTION_IF_NULL(args_spec_list[0]);
   AbstractFunctionPtr fn = args_spec_list[0]->cast<AbstractFunctionPtr>();
@@ -411,7 +415,7 @@ AbstractBasePtr PyInferRes2Abstract(const PrimitivePyPtr &prim_py, const py::dic
 }
 }  // end anonymous namespace
 
-AbstractBasePtr PythonPrimEvaluator::EvalPrim(const AnalysisEnginePtr &, const AbstractBasePtrList &args) {
+EvalResultPtr PythonPrimEvaluator::EvalPrim(const AnalysisEnginePtr &, const AbstractBasePtrList &args) {
   MS_LOG(DEBUG) << "Eval for:" << prim_py_->ToString();
 
   const auto &iter = cache_->find(args);
@@ -425,17 +429,20 @@ AbstractBasePtr PythonPrimEvaluator::EvalPrim(const AnalysisEnginePtr &, const A
     MS_LOG(EXCEPTION) << "[" << prim_py_->ToString() << "]: pyobj is empty";
   }
   auto infer_fuc = pyobj.attr("__infer__");
-
+  prim_py_->BeginRecordAddAttr();
   py::dict output = infer_fuc(*py_args);
+  prim_py_->EndRecordAddAttr();
+  auto added_attrs = prim_py_->evaluate_added_attrs();
   MS_LOG(DEBUG) << "Output type is " << (std::string)py::str(output);
   auto res_spec = PyInferRes2Abstract(prim_py_, output);
 
   MS_LOG(DEBUG) << "Python InferTensor result spec: " << res_spec->ToString() << ".";
-  (*cache_)[args] = res_spec;
-  return res_spec;
+  auto infer_result = std::make_shared<EvalResult>(res_spec, std::make_shared<AttrValueMap>(added_attrs));
+  (*cache_)[args] = infer_result;
+  return infer_result;
 }
 
-AbstractBasePtr UniformPrimEvaluator::EvalPrim(const AnalysisEnginePtr &, const AbstractBasePtrList &args) {
+EvalResultPtr UniformPrimEvaluator::EvalPrim(const AnalysisEnginePtr &, const AbstractBasePtrList &args) {
   // if func_desc_.retval type is super class of parameter type, then make the retval type as parameter type.
   if (nargs_ != args.size()) {
     MS_LOG(ERROR) << "UniformPrimEvaluator expect " << nargs_ << " args, but got " << args.size() << " inputs";
@@ -476,7 +483,7 @@ AbstractBasePtr UniformPrimEvaluator::EvalPrim(const AnalysisEnginePtr &, const
   }
 
   AbstractScalarPtr abs_base = std::make_shared<AbstractScalar>(evaluated_value, ret_value_type);
-  return abs_base;
+  return std::make_shared<EvalResult>(abs_base, std::make_shared<AttrValueMap>());
 }
 
 ValuePtr UniformPrimEvaluator::RunImpl(const ValuePtrList &args) const {
@@ -553,8 +560,8 @@ inline void AddToManager(const AnalysisEnginePtr &engine, const FuncGraphPtr fun
   manager->AddFuncGraph(func_graph);
 }
 
-AbstractBasePtr StaticGetterInferred(const ValuePtr &value, const ConfigPtr &data_conf,
-                                     const AnfNodeConfigPtr &old_conf) {
+EvalResultPtr StaticGetterInferred(const ValuePtr &value, const ConfigPtr &data_conf,
+                                   const AnfNodeConfigPtr &old_conf) {
   MS_EXCEPTION_IF_NULL(old_conf);
 
   AbstractBasePtr abs_ptr = ToAbstract(value, AnalysisContext::DummyContext(), old_conf);
@@ -585,9 +592,9 @@ AbstractBasePtr StaticGetterInferred(const ValuePtr &value, const ConfigPtr &dat
   return eng->ForwardConfig(old_conf, fn_conf);
 }
 
-AbstractBasePtr GetEvaluatedValueForNameSpaceString(const AnalysisEnginePtr &engine,
-                                                    const AbstractBasePtrList &args_spec_list,
-                                                    const AnfNodeConfigPtr &out_conf) {
+EvalResultPtr GetEvaluatedValueForNameSpaceString(const AnalysisEnginePtr &engine,
+                                                  const AbstractBasePtrList &args_spec_list,
+                                                  const AnfNodeConfigPtr &out_conf) {
   // args_spec_list: same as StaticGetter
   if (args_spec_list.size() < 2) {
     MS_LOG(EXCEPTION) << "Size of args_spec_list is less than 2";
@@ -627,9 +634,9 @@ AbstractBasePtr GetEvaluatedValueForNameSpaceString(const AnalysisEnginePtr &eng
   return eng->ForwardConfig(out_conf, fn_conf);
 }
 
-AbstractBasePtr GetEvaluatedValueForClassAttrOrMethod(const AnalysisEnginePtr &engine,
-                                                      const AbstractBasePtrList &args_spec_list, const ValuePtr &item_v,
-                                                      const ConfigPtr &data_conf, const AnfNodeConfigPtr &out_conf) {
+EvalResultPtr GetEvaluatedValueForClassAttrOrMethod(const AnalysisEnginePtr &engine,
+                                                    const AbstractBasePtrList &args_spec_list, const ValuePtr &item_v,
+                                                    const ConfigPtr &data_conf, const AnfNodeConfigPtr &out_conf) {
   if (args_spec_list.empty()) {
     MS_LOG(EXCEPTION) << "args_spec_list is empty";
   }
@@ -646,7 +653,7 @@ AbstractBasePtr GetEvaluatedValueForClassAttrOrMethod(const AnalysisEnginePtr &e
 
   AbstractBasePtr attr = cls->GetAttribute(item_name);
   if (attr != nullptr) {
-    return attr;
+    return std::make_shared<EvalResult>(attr, nullptr);
   }
 
   ValuePtr method = cls->GetMethod(item_name);
@@ -660,9 +667,9 @@ AbstractBasePtr GetEvaluatedValueForClassAttrOrMethod(const AnalysisEnginePtr &e
   return StaticGetterInferred(converted_v, data_conf, out_conf);
 }
 
-AbstractBasePtr GetEvaluatedValueForBuiltinTypeMethod(const AnalysisEnginePtr &engine, const ValuePtr &item_v,
-                                                      const TypePtr &data_type, const ConfigPtr &data_conf,
-                                                      const AnfNodeConfigPtr &out_conf) {
+EvalResultPtr GetEvaluatedValueForBuiltinTypeMethod(const AnalysisEnginePtr &engine, const ValuePtr &item_v,
+                                                    const TypePtr &data_type, const ConfigPtr &data_conf,
+                                                    const AnfNodeConfigPtr &out_conf) {
   MS_EXCEPTION_IF_NULL(item_v);
   MS_EXCEPTION_IF_NULL(data_type);
   // The method maybe a Primitive or Composite
@@ -689,8 +696,8 @@ AbstractBasePtr GetEvaluatedValueForBuiltinTypeMethod(const AnalysisEnginePtr &e
   return StaticGetterInferred(converted_v, data_conf, out_conf);
 }
 
-AbstractBasePtr StaticGetter(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list,
-                             const ConfigPtr &data_conf, const AnfNodeConfigPtr &out_conf) {
+EvalResultPtr StaticGetter(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list,
+                           const ConfigPtr &data_conf, const AnfNodeConfigPtr &out_conf) {
   // Inputs: namespace and its static function; or class and its member function
   CheckArgsSize("StaticGetter", args_spec_list, 2);
 
@@ -725,7 +732,7 @@ class EmbedEvaluator : public SymbolicPrimEvaluator {
   EmbedEvaluator() : SymbolicPrimEvaluator("EmbedEvaluator") {}
   ~EmbedEvaluator() override = default;
   MS_DECLARE_PARENT(EmbedEvaluator, SymbolicPrimEvaluator);
-  AbstractBasePtr EvalPrim(const ConfigPtrList &args_conf_list) override {
+  EvalResultPtr EvalPrim(const ConfigPtrList &args_conf_list) override {
     // arg: free variable to be embedded
     if (args_conf_list.size() != 1) {
       MS_LOG(EXCEPTION) << "EmbedEvaluator requires 1 parameter, but got " << args_conf_list.size();
@@ -733,11 +740,11 @@ class EmbedEvaluator : public SymbolicPrimEvaluator {
     AnfNodeConfigPtr node_conf = dyn_cast<AnfNodeConfig>(args_conf_list[0]);
     MS_EXCEPTION_IF_NULL(node_conf);
 
-    AbstractBasePtr x = node_conf->GetEvaluatedValue();
+    AbstractBasePtr x = node_conf->GetEvaluatedValue()->abstract();
     x = SensitivityTransform(x);
     SymbolicKeyInstancePtr key = std::make_shared<SymbolicKeyInstance>(node_conf->node(), x);
     AbstractScalarPtr abs_scalar = std::make_shared<AbstractScalar>(key, std::make_shared<SymbolicKeyType>());
-    return abs_scalar;
+    return std::make_shared<EvalResult>(abs_scalar, std::make_shared<AttrValueMap>());
   }
 };
 
@@ -762,7 +769,7 @@ class RefToEmbedEvaluator : public SymbolicPrimEvaluator {
   RefToEmbedEvaluator() : SymbolicPrimEvaluator("RefToEmbedEvaluator") {}
   ~RefToEmbedEvaluator() override = default;
   MS_DECLARE_PARENT(RefToEmbedEvaluator, SymbolicPrimEvaluator);
-  AbstractBasePtr EvalPrim(const ConfigPtrList &args_conf_list) override {
+  EvalResultPtr EvalPrim(const ConfigPtrList &args_conf_list) override {
     if (args_conf_list.size() != 1) {
       MS_LOG(ERROR) << "Requires 1 parameter, but has: " << args_conf_list.size();
       return nullptr;
@@ -773,7 +780,7 @@ class RefToEmbedEvaluator : public SymbolicPrimEvaluator {
       MS_LOG(ERROR) << "Conf should be AnfNodeConfig";
       return nullptr;
     }
-    AbstractBasePtr abs = node_conf->GetEvaluatedValue();
+    AbstractBasePtr abs = node_conf->GetEvaluatedValue()->abstract();
     AbstractRefPtr ref_abs = abs->cast<AbstractRefPtr>();
     if (ref_abs == nullptr) {
       MS_LOG(ERROR) << "The first parameter of RefToEmbed should be Ref.";
@@ -791,7 +798,7 @@ class RefToEmbedEvaluator : public SymbolicPrimEvaluator {
     }
     auto refkey = key_value->cast<RefKeyPtr>();
     if (refkey == nullptr) {
-      return std::make_shared<AbstractScalar>(type);
+      return std::make_shared<EvalResult>(std::make_shared<AbstractScalar>(type), std::make_shared<AttrValueMap>());
     }
 
     std::string name = refkey->tag();
@@ -805,7 +812,7 @@ class RefToEmbedEvaluator : public SymbolicPrimEvaluator {
     x = SensitivityTransform(x);
     std::shared_ptr<SymbolicKeyInstance> key = std::make_shared<SymbolicKeyInstance>(node, x);
     std::shared_ptr<AbstractScalar> abs_scalar = std::make_shared<AbstractScalar>(key, type);
-    return abs_scalar;
+    return std::make_shared<EvalResult>(abs_scalar, std::make_shared<AttrValueMap>());
   }
 };
 
@@ -814,13 +821,13 @@ class GetAttrEvaluator : public TransitionPrimEvaluator {
   GetAttrEvaluator() : TransitionPrimEvaluator("GetAttrEvaluator") {}
   ~GetAttrEvaluator() override = default;
   MS_DECLARE_PARENT(GetAttrEvaluator, TransitionPrimEvaluator);
-  AbstractBasePtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list,
-                           const ConfigPtr &in_conf0, const AnfNodeConfigPtr &out_conf) override {
+  EvalResultPtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list,
+                         const ConfigPtr &in_conf0, const AnfNodeConfigPtr &out_conf) override {
     // Inputs: data, item
     if (args_spec_list.size() != 2) {
       MS_LOG(EXCEPTION) << "Expected args_spec_list size = 2, but has size:" << args_spec_list.size();
     }
-    AbstractBasePtr ret = nullptr;
+    EvalResultPtr ret = nullptr;
     if (bound_node() != nullptr) {
       TraceManager::DebugTrace(std::make_shared<TraceResolve>(bound_node()->debug_info()));
       ret = StaticGetter(engine, args_spec_list, in_conf0, out_conf);
@@ -840,13 +847,13 @@ class ResolveEvaluator : public TransitionPrimEvaluator {
   ResolveEvaluator() : TransitionPrimEvaluator("ResolveEvaluator") {}
   ~ResolveEvaluator() override = default;
   MS_DECLARE_PARENT(ResolveEvaluator, TransitionPrimEvaluator);
-  AbstractBasePtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list,
-                           const ConfigPtr &in_conf0, const AnfNodeConfigPtr &out_conf) override {
+  EvalResultPtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list,
+                         const ConfigPtr &in_conf0, const AnfNodeConfigPtr &out_conf) override {
     // Inputs: namespace, symbol
     if (args_spec_list.size() != 2) {
       MS_LOG(EXCEPTION) << "Expected args_spec_list size = 2, but has size:" << args_spec_list.size();
     }
-    AbstractBasePtr ret = nullptr;
+    EvalResultPtr ret = nullptr;
     if (bound_node() != nullptr) {
       TraceManager::DebugTrace(std::make_shared<TraceResolve>(bound_node()->debug_info()));
       ret = StaticGetter(engine, args_spec_list, in_conf0, out_conf);
@@ -863,8 +870,8 @@ class CreateInstanceEvaluator : public TransitionPrimEvaluator {
   CreateInstanceEvaluator() : TransitionPrimEvaluator("CreateInstanceEvaluator") {}
   ~CreateInstanceEvaluator() override = default;
   MS_DECLARE_PARENT(CreateInstanceEvaluator, TransitionPrimEvaluator);
-  AbstractBasePtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list,
-                           const ConfigPtr &, const AnfNodeConfigPtr &out_conf) override {
+  EvalResultPtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args_spec_list, const ConfigPtr &,
+                         const AnfNodeConfigPtr &out_conf) override {
     if (args_spec_list.empty()) {
       MS_LOG(EXCEPTION) << "'args_spec_list' should not be empty";
     }
@@ -915,8 +922,9 @@ class CreateInstanceEvaluator : public TransitionPrimEvaluator {
     }
 
     AbstractBasePtr ret = ToAbstract(converted_ret, AnalysisContext::DummyContext(), out_conf);
-    (*cache_)[args_spec_list] = ret;
-    return ret;
+    auto infer_result = std::make_shared<EvalResult>(ret, nullptr);
+    (*cache_)[args_spec_list] = infer_result;
+    return infer_result;
   }
 
   pybind11::tuple GetParameters(const AbstractBasePtrList &args_spec_list) const {
@@ -942,23 +950,24 @@ class PartialEvaluator : public Evaluator {
  public:
   PartialEvaluator() : Evaluator("PartialEvaluator") {}
   ~PartialEvaluator() override = default;
-  AbstractBasePtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
-                      AnfNodeConfigPtr out_conf = nullptr) override {
+  EvalResultPtr Run(AnalysisEnginePtr engine, const ConfigPtrList &args_conf_list,
+                    AnfNodeConfigPtr out_conf = nullptr) override {
     if (args_conf_list.size() == 0) {
       MS_LOG(EXCEPTION) << "Args size should be greater than 0";
     }
+
     MS_EXCEPTION_IF_NULL(out_conf);
     MS_EXCEPTION_IF_NULL(out_conf->node());
-
-    auto arg0_value = args_conf_list[0]->GetEvaluatedValue();
+    auto arg0_value = args_conf_list[0]->GetEvaluatedValue()->abstract();
     AbstractBasePtrList args_spec_list{arg0_value};
     // Func in hypermap(partial(Func, arg0), arg1, arg2) may become Poly Node.
     if (arg0_value->isa<AbstractError>()) {
       auto ret = std::make_shared<AbstractError>(arg0_value->GetValueTrack()->cast<StringImmPtr>(), out_conf->node());
       MS_LOG(DEBUG) << "AbstractError for node: " << out_conf->node()->DebugString()
                     << " as func is: " << arg0_value->ToString();
-      (*cache_)[args_spec_list] = ret;
-      return ret;
+      auto eval_result = std::make_shared<EvalResult>(ret, std::make_shared<AttrValueMap>());
+      (*cache_)[args_spec_list] = eval_result;
+      return eval_result;
     }
     auto func = CheckArg<AbstractFunction>("partial", args_spec_list, 0);
     // Sometimes, node[0] in out_conf becomes phi0;
@@ -970,8 +979,9 @@ class PartialEvaluator : public Evaluator {
       }
     }
 
-    (void)std::transform(args_conf_list.begin() + 1, args_conf_list.end(), std::back_inserter(args_spec_list),
-                         [](const ConfigPtr &config) -> AbstractBasePtr { return config->GetEvaluatedValue(); });
+    (void)std::transform(
+      args_conf_list.begin() + 1, args_conf_list.end(), std::back_inserter(args_spec_list),
+      [](const ConfigPtr &config) -> AbstractBasePtr { return config->GetEvaluatedValue()->abstract(); });
     AbstractBasePtrList args(args_spec_list.begin() + 1, args_spec_list.end());
 
     auto cnode = out_conf->node()->cast<CNodePtr>();
@@ -989,16 +999,17 @@ class PartialEvaluator : public Evaluator {
     func->Visit(build_partial);
 
     auto ret = AbstractFunction::MakeAbstractFunction(partial_funcs_list);
-    (*cache_)[args_spec_list] = ret;
-    return ret;
+    auto infer_result = std::make_shared<EvalResult>(ret, std::make_shared<AttrValueMap>());
+    (*cache_)[args_spec_list] = infer_result;
+    return infer_result;
   }
 
-  AbstractBasePtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
+  EvalResultPtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
     MS_LOG(EXCEPTION) << "Eval() should not be called, Run() method should be called";
   }
 
-  AbstractBasePtr HandleDoSignature(const AnalysisEnginePtr &engine, const ValuePtr &signature_value,
-                                    const AnfNodeConfigPtr &out_conf = nullptr) const {
+  EvalResultPtr HandleDoSignature(const AnalysisEnginePtr &engine, const ValuePtr &signature_value,
+                                  const AnfNodeConfigPtr &out_conf = nullptr) const {
     MS_EXCEPTION_IF_NULL(out_conf);
     MS_EXCEPTION_IF_NULL(out_conf->node());
     auto cnode = out_conf->node()->cast<CNodePtr>();
diff --git a/mindspore/ccsrc/pipeline/static_analysis/prim.h b/mindspore/ccsrc/pipeline/static_analysis/prim.h
index 91b47ddb7f6..22418180f76 100644
--- a/mindspore/ccsrc/pipeline/static_analysis/prim.h
+++ b/mindspore/ccsrc/pipeline/static_analysis/prim.h
@@ -45,7 +45,7 @@ class StandardPrimEvaluator : public TrivialPrimEvaluator {
       : TrivialPrimEvaluator("StandardPrimEvaluator"), prim_(primitive), eval_impl_(eval_impl) {}
   ~StandardPrimEvaluator() override = default;
   MS_DECLARE_PARENT(StandardPrimEvaluator, TrivialPrimEvaluator);
-  AbstractBasePtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args) override;
+  EvalResultPtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args) override;
   PrimitivePtr prim() { return prim_; }
 
   std::string ToString() const override { return identifier_ + prim_->name(); }
@@ -63,7 +63,7 @@ class PythonPrimEvaluator : public TrivialPrimEvaluator {
       : TrivialPrimEvaluator("PythonPrimEvaluator"), prim_py_(primitive) {}
   ~PythonPrimEvaluator() override = default;
   MS_DECLARE_PARENT(PythonPrimEvaluator, TrivialPrimEvaluator);
-  AbstractBasePtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args) override;
+  EvalResultPtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args) override;
   PrimitivePtr prim() { return dyn_cast<Primitive>(prim_py_); }
 
   std::string ToString() const override { return identifier_ + prim_py_->name(); }
@@ -76,10 +76,10 @@ class DoSignatureEvaluator : public Evaluator {
  public:
   explicit DoSignatureEvaluator(const PrimitivePtr primitive) : Evaluator("DoSignatureEvaluator"), prim_(primitive) {}
   ~DoSignatureEvaluator() override = default;
-  AbstractBasePtr Run(AnalysisEnginePtr engine, const ConfigPtrList &argrefs,
-                      AnfNodeConfigPtr out_config = nullptr) override;
+  EvalResultPtr Run(AnalysisEnginePtr engine, const ConfigPtrList &argrefs,
+                    AnfNodeConfigPtr out_config = nullptr) override;
 
-  AbstractBasePtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
+  EvalResultPtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
     MS_LOG(EXCEPTION) << "Eval() should not be called, Run() method should be called";
   }
 
@@ -91,10 +91,10 @@ class UnpackGraphEvaluator : public Evaluator {
  public:
   explicit UnpackGraphEvaluator(const PrimitivePtr primitive) : Evaluator("UnpackGraphEvaluator"), prim_(primitive) {}
   ~UnpackGraphEvaluator() override = default;
-  AbstractBasePtr Run(AnalysisEnginePtr engine, const ConfigPtrList &argrefs,
-                      AnfNodeConfigPtr out_config = nullptr) override;
+  EvalResultPtr Run(AnalysisEnginePtr engine, const ConfigPtrList &argrefs,
+                    AnfNodeConfigPtr out_config = nullptr) override;
 
-  AbstractBasePtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
+  EvalResultPtr Eval(AnalysisEnginePtr, const AbstractBasePtrList &) override {
     MS_LOG(EXCEPTION) << "Eval() should not be called, Run() method should be called";
   }
 
@@ -131,7 +131,7 @@ class UniformPrimEvaluator : public TrivialPrimEvaluator {
   ~UniformPrimEvaluator() override = default;
   MS_DECLARE_PARENT(UniformPrimEvaluator, TrivialPrimEvaluator);
 
-  AbstractBasePtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args) override;
+  EvalResultPtr EvalPrim(const AnalysisEnginePtr &engine, const AbstractBasePtrList &args) override;
   ValuePtr RunImpl(const ValuePtrList &args) const;
 
   // If eval_value_ is False, return broadened arguments.
diff --git a/mindspore/ccsrc/pipeline/static_analysis/program_specialize.cc b/mindspore/ccsrc/pipeline/static_analysis/program_specialize.cc
index 6f67e7348c5..2a03eb6d5c1 100644
--- a/mindspore/ccsrc/pipeline/static_analysis/program_specialize.cc
+++ b/mindspore/ccsrc/pipeline/static_analysis/program_specialize.cc
@@ -36,7 +36,7 @@ inline AbstractBasePtr GetEvaluatedValueWrap(const AnfNodeConfigPtr &conf) {
   if (conf->node()->intermediate_abstract()) {
     return conf->node()->intermediate_abstract();
   }
-  return conf->GetEvaluatedValue();
+  return conf->GetEvaluatedValue()->abstract();
 }
 
 AnfNodePtr BuildValueNode(const ValuePtr &v, const AbstractBasePtr &abs_base) {
@@ -212,7 +212,7 @@ void FuncGraphSpecializer::FirstPass() {
 
 // Specialize CNode in func graphs
 void FuncGraphSpecializer::SecondPass() {
-  for (auto &node : DeepLinkedGraphSearch(specialized_func_graph_->get_return())) {
+  for (auto &node : BroadFirstSearchGraphCNodes(specialized_func_graph_->get_return())) {
     if (node->isa<CNode>()) {
       ProcessCNode(node->cast<CNodePtr>());
     }
@@ -225,7 +225,6 @@ void FuncGraphSpecializer::ProcessNode(const AnfNodePtr &node) {
   AnfNodeConfigPtr conf = MakeConfig(node);
   AnfNodePtr new_node = GetReplicatedNode(node);
   MS_EXCEPTION_IF_NULL(new_node);
-
   if (new_node->func_graph() != specialized_func_graph_) {
     MS_LOG(EXCEPTION) << "Error in specializer [A] node: " << node->DebugString()
                       << ", new_node: " << new_node->DebugString()
@@ -244,6 +243,7 @@ void FuncGraphSpecializer::ProcessNode(const AnfNodePtr &node) {
   MS_LOG(DEBUG) << "Set new_node: " << new_node->ToString() << ", abstract as: " << new_node->abstract()->ToString();
 
   if (node->isa<CNode>()) {
+    auto attrs = conf->GetEvaluatedValue()->attribute();
     auto c_old = node->cast<CNodePtr>();
     auto c_new = new_node->cast<CNodePtr>();
     auto new_inputs = c_new->inputs();
@@ -254,7 +254,7 @@ void FuncGraphSpecializer::ProcessNode(const AnfNodePtr &node) {
       AbstractBasePtr ival = GetEvaluatedValueWrap(iconf);
       // First try to check if node_input can be replaced by a ValueNode. If cannot, then try to check if
       // can be replaced by another CNode from anfnode_config_map, otherwise use the replicated node.
-      AnfNodePtr replace_node = BuildPossibleValueNode(iconf->node(), ival);
+      AnfNodePtr replace_node = BuildPossibleValueNode(iconf->node(), ival, attrs);
       if (replace_node == nullptr) {
         replace_node = BuildReplacedNode(iconf);
         MS_EXCEPTION_IF_NULL(replace_node);
@@ -424,9 +424,10 @@ AnfNodePtr FuncGraphSpecializer::BuildSpecializedParameterNode(const CNodePtr &n
       MS_LOG(EXCEPTION) << "Size of cnode: " << cnode->DebugString()
                         << " is not equal to 2 added to size of args: " << mindspore::ToString(partial_closure->args());
     }
+    auto attrs = std::make_shared<AttrValueMap>();
     for (size_t i = 0; i < partial_closure->args().size(); i++) {
       auto old_node = cnode->input(i + 2);
-      auto possibile_value_node = BuildPossibleValueNode(old_node, partial_closure->args()[i]);
+      auto possibile_value_node = BuildPossibleValueNode(old_node, partial_closure->args()[i], attrs);
       if (possibile_value_node != nullptr) {
         partial_node_list.push_back(possibile_value_node);
       } else {
@@ -455,7 +456,7 @@ std::pair<AbstractBasePtrList, AbstractBasePtr> FuncGraphSpecializer::BuildFromB
   const EvaluatorPtr &eval) {
   MS_EXCEPTION_IF_NULL(eval);
   std::unordered_set<AbstractBasePtrList, AbstractBasePtrListHasher, AbstractBasePtrListEqual> choices;
-  AbstractBasePtr ret = nullptr;
+  EvalResultPtr ret = nullptr;
   AbstractBasePtrList broaded_argvals;
   for (auto &argvals_map : *evalcaches_[eval]) {
     auto argvals = argvals_map.first;
@@ -478,7 +479,7 @@ std::pair<AbstractBasePtrList, AbstractBasePtr> FuncGraphSpecializer::BuildFromB
 
     (*real)[broaded_argvals] = ret;
     evalcaches_[eval] = real;
-    return std::make_pair(broaded_argvals, ret);
+    return std::make_pair(broaded_argvals, ret->abstract());
   } else {
     MS_LOG(DEBUG) << "Choices.size: " << choices.size();
     return std::make_pair(AbstractBasePtrList(), nullptr);
@@ -491,7 +492,6 @@ void FuncGraphSpecializer::ProcessCNode(const CNodePtr &new_node) {
     return;
   }
   specializer_->AddSeen(new_node);
-
   auto new_inputs = new_node->inputs();
   if (new_inputs.empty()) {
     MS_LOG(EXCEPTION) << "Inputs of CNode is empty";
@@ -530,7 +530,13 @@ void FuncGraphSpecializer::ProcessCNode(const CNodePtr &new_node) {
   }
 
   if (CanSpecializeNode(func)) {
-    new_inputs[0] = BuildSpecializedNode(func, fnval, argvals);
+    // for primitive node , we build the primitive node with infered attributes in the first pass
+    // so we do not build replaced node again here in second pass
+    if (IsValueNode<Primitive>(func)) {
+      new_inputs[0] = func;
+    } else {
+      new_inputs[0] = BuildSpecializedNode(func, fnval, argvals);
+    }
   }
 
   for (size_t i = 0; i < argvals.size();) {
@@ -540,7 +546,6 @@ void FuncGraphSpecializer::ProcessCNode(const CNodePtr &new_node) {
     }
     i = next;
   }
-
   new_node->set_inputs(new_inputs);
 }
 
@@ -582,7 +587,7 @@ SpecializeStatusCode FuncGraphSpecializer::FindUniqueArgvals(const AbstractFunct
 
   EvaluatorCacheMap evaluator_cache_map = *eval->cache();
   if (evaluator_cache_map.find(argvals) != evaluator_cache_map.end()) {
-    *result = std::make_pair(argvals, evaluator_cache_map[argvals]);
+    *result = std::make_pair(argvals, evaluator_cache_map[argvals]->abstract());
     return kSpecializeSuccess;
   }
   DumpEvaluatorCache(evaluator_cache_map, argvals);
@@ -591,11 +596,11 @@ SpecializeStatusCode FuncGraphSpecializer::FindUniqueArgvals(const AbstractFunct
   MS_EXCEPTION_IF_NULL(choices);
 
   if (choices->count(argvals)) {
-    *result = std::make_pair(argvals, (*choices)[argvals]);
+    *result = std::make_pair(argvals, (*choices)[argvals]->abstract());
     return kSpecializeSuccess;
   } else if (choices->size() == 1) {
     MS_LOG(DEBUG) << "Evaluator cache has a single item, just use it.";
-    *result = std::make_pair(choices->begin()->first, choices->begin()->second);
+    *result = std::make_pair(choices->begin()->first, choices->begin()->second->abstract());
     return kSpecializeSuccess;
   } else if (choices->empty()) {
     MS_LOG(DEBUG) << "Find DEAD code, it may be optimized in later phase.";
@@ -614,8 +619,43 @@ SpecializeStatusCode FuncGraphSpecializer::FindUniqueArgvals(const AbstractFunct
     return kSpecializeFindUniqueArgvalPoly;
   }
 }
+static PrimitivePtr BuildPrimtiveValueWithAttributes(const PrimitivePtr &prim, const AttrValueMapPtr &attrs) {
+  auto &prim_attrs = prim->attrs();
+  bool is_attr_same = true;
+  for (auto &item : *attrs) {
+    auto itr = prim_attrs.find(item.first);
+    if (itr != prim_attrs.end()) {
+      if (!(*(itr->second) == *(item.second))) {
+        is_attr_same = false;
+        break;
+      }
+    } else {
+      is_attr_same = false;
+      break;
+    }
+  }
+  if (!is_attr_same) {
+    if (prim->isa<PrimitivePy>()) {
+      PrimitivePyPtr prim_py = prim->cast<PrimitivePyPtr>();
+      auto clone_fn = prim_py->GetPyObj().attr("_clone");
+      py::object new_obj = clone_fn();
+      auto cloned_prim = new_obj.cast<PrimitivePyPtr>();
+      for (auto &item : *attrs) {
+        cloned_prim->AddAttr(item.first, item.second);
+      }
+      return cloned_prim;
+    }
+    auto cloned_prim = std::make_shared<Primitive>(*prim);
+    for (auto &item : *attrs) {
+      cloned_prim->AddAttr(item.first, item.second);
+    }
+    return cloned_prim;
+  }
+  return prim;
+}
 
-AnfNodePtr FuncGraphSpecializer::BuildPossibleValueNode(const AnfNodePtr &origin_node, const AbstractBasePtr &ival) {
+AnfNodePtr FuncGraphSpecializer::BuildPossibleValueNode(const AnfNodePtr &origin_node, const AbstractBasePtr &ival,
+                                                        const AttrValueMapPtr &attrs) {
   MS_EXCEPTION_IF_NULL(origin_node);
   MS_EXCEPTION_IF_NULL(ival);
 
@@ -628,7 +668,12 @@ AnfNodePtr FuncGraphSpecializer::BuildPossibleValueNode(const AnfNodePtr &origin
     ValuePtr value = nullptr;
     if (abs->isa<PrimitiveAbstractClosure>()) {
       auto real_fn = dyn_cast<PrimitiveAbstractClosure>(abs);
-      value = real_fn->prim();
+      // for primitive, check if the attribute is the same with cnode infererd attribute ,if not, clone a new one
+      if (attrs != nullptr) {
+        value = BuildPrimtiveValueWithAttributes(real_fn->prim(), attrs);
+      } else {
+        value = real_fn->prim();
+      }
     } else if (abs->isa<MetaFuncGraphAbstractClosure>()) {
       auto real_fn = dyn_cast<MetaFuncGraphAbstractClosure>(abs);
       value = real_fn->meta_func_graph();
diff --git a/mindspore/ccsrc/pipeline/static_analysis/program_specialize.h b/mindspore/ccsrc/pipeline/static_analysis/program_specialize.h
index e3c2027d418..b04978586d9 100644
--- a/mindspore/ccsrc/pipeline/static_analysis/program_specialize.h
+++ b/mindspore/ccsrc/pipeline/static_analysis/program_specialize.h
@@ -110,7 +110,8 @@ class FuncGraphSpecializer : public std::enable_shared_from_this<FuncGraphSpecia
   AnfNodePtr BuildSpecializedParameterNode(const CNodePtr &new_node);
 
   // Build a value node if ival is constant and not any-value
-  AnfNodePtr BuildPossibleValueNode(const AnfNodePtr &origin_node, const AbstractBasePtr &ival);
+  AnfNodePtr BuildPossibleValueNode(const AnfNodePtr &origin_node, const AbstractBasePtr &ival,
+                                    const AttrValueMapPtr &attrs);
   // Build a replacable node for iconf->node; it may be a replicated forwared CNode in static analysis or just a
   // replicated node.
   AnfNodePtr BuildReplacedNode(const AnfNodeConfigPtr &conf);
diff --git a/mindspore/ccsrc/pipeline/static_analysis/static_analysis.cc b/mindspore/ccsrc/pipeline/static_analysis/static_analysis.cc
index 464f8313e92..e024d11b474 100644
--- a/mindspore/ccsrc/pipeline/static_analysis/static_analysis.cc
+++ b/mindspore/ccsrc/pipeline/static_analysis/static_analysis.cc
@@ -55,29 +55,29 @@ AbstractBasePtr IntermediateJoin(const AbstractBasePtr &arg1, const AbstractBase
   return nullptr;
 }
 
-void AnalysisCache::set_value(const AnfNodeConfigPtr &conf, const AbstractBasePtr &arg) {
+void AnalysisCache::set_value(const AnfNodeConfigPtr &conf, const EvalResultPtr &result) {
   MS_LOG(DEBUG) << "AnalysisCache set for NodeConfig: " << conf->node()->DebugString()
-                << ", Context: " << conf->context()->ToString() << ", Value: " << arg->ToString()
-                << ", Pointer: " << arg.get();
-  cache_[conf] = arg;
+                << ", Context: " << conf->context()->ToString() << ", Value: " << result->abstract()->ToString()
+                << ", Pointer: " << result->abstract().get();
+  cache_[conf] = result;
 
   // Set intermediate abstract value.
-  if (IsIntermediateAbstract(arg)) {
+  if (IsIntermediateAbstract(result->abstract())) {
     if (conf->node()->intermediate_abstract() == nullptr) {
-      conf->node()->set_intermediate_abstract(arg);
-      MS_LOG(DEBUG) << "Set intermediate abstract: " << arg->ToString();
+      conf->node()->set_intermediate_abstract(result->abstract());
+      MS_LOG(DEBUG) << "Set intermediate abstract: " << result->abstract()->ToString();
     } else {
       auto old_spec = conf->node()->intermediate_abstract();
-      auto joined_spec = IntermediateJoin(arg, old_spec);
+      auto joined_spec = IntermediateJoin(result->abstract(), old_spec);
       conf->node()->set_intermediate_abstract(joined_spec);
       MS_LOG(DEBUG) << "Set joined intermediate abstract:\nold_spec:\t\t" << old_spec->ToString() << "\nnew_spec:\t\t"
-                    << arg->ToString() << "\njoined_spec:\t"
+                    << result->abstract()->ToString() << "\njoined_spec:\t"
                     << (joined_spec != nullptr ? joined_spec->ToString() : "nullptr");
     }
   }
 }
 
-AbstractBasePtr AnalysisCache::GetValue(const AnfNodeConfigPtr &conf) {
+EvalResultPtr AnalysisCache::GetValue(const AnfNodeConfigPtr &conf) {
   auto value = cache_.find(conf);
   if (value == cache_.end()) {
     return nullptr;
@@ -142,12 +142,12 @@ AnalysisContextPtr AnalysisEngine::Run(const FuncGraphPtr &func_graph, const Ana
   return eval->graph_context();
 }
 
-AbstractBasePtr AnalysisEngine::GetEvaluatedValue(const AnfNodeConfigPtr &conf) {
+EvalResultPtr AnalysisEngine::GetEvaluatedValue(const AnfNodeConfigPtr &conf) {
   MS_EXCEPTION_IF_NULL(conf);
   auto value = cache_.GetValue(conf);
   if (value != nullptr) {
-    MS_LOG(DEBUG) << "Evaluate cache hit for NodeConfig: " << conf->ToString() << ", Value: " << value.get() << ", "
-                  << value->ToString();
+    MS_LOG(DEBUG) << "Evaluate cache hit for NodeConfig: " << conf->ToString() << ", Value: " << value->abstract().get()
+                  << ", " << value->abstract()->ToString();
     return value;
   }
 
@@ -160,10 +160,10 @@ AbstractBasePtr AnalysisEngine::GetEvaluatedValue(const AnfNodeConfigPtr &conf)
   return value;
 }
 
-AbstractBasePtr AnalysisEngine::Eval(const AnfNodeConfigPtr &conf) {
+EvalResultPtr AnalysisEngine::Eval(const AnfNodeConfigPtr &conf) {
   MS_EXCEPTION_IF_NULL(conf);
   AnfNodePtr node = conf->node();
-  AbstractBasePtr ret_abstract = nullptr;
+  EvalResultPtr eval_result = nullptr;
 #ifdef DEBUG
   compute_conf_stack_.push_back(node);
   std::ostringstream buffer;
@@ -177,14 +177,14 @@ AbstractBasePtr AnalysisEngine::Eval(const AnfNodeConfigPtr &conf) {
   MS_EXCEPTION_IF_NULL(node);
   if (node->abstract() != nullptr) {
     MS_LOG(DEBUG) << "Return old abstract: " << node->DebugString();
-    ret_abstract = node->abstract();
+    eval_result = std::make_shared<EvalResult>(node->abstract(), std::make_shared<AttrValueMap>());
   } else if (node->isa<ValueNode>()) {
     auto value_node = node->cast<ValueNodePtr>();
-    ret_abstract = EvalValueNode(value_node, conf);
+    eval_result = std::make_shared<EvalResult>(EvalValueNode(value_node, conf), nullptr);
   } else if (node->isa<CNode>()) {
     auto cnode = node->cast<CNodePtr>();
     trace::TraceEvalCNodeEnter(conf);
-    ret_abstract = EvalCNode(cnode, conf);
+    eval_result = EvalCNode(cnode, conf);
     trace::TraceEvalCNodeLeave();
   } else {
     MS_LOG(EXCEPTION) << "Illegal AnfNode for evaluating, " << node->DebugString()
@@ -193,13 +193,13 @@ AbstractBasePtr AnalysisEngine::Eval(const AnfNodeConfigPtr &conf) {
 
 #ifdef DEBUG
   compute_conf_stack_.pop_back();
-  if (ret_abstract == nullptr) {
+  if (eval_result == nullptr) {
     MS_LOG(EXCEPTION) << "Compute Config failed, node: " << node->DebugString()
                       << " NodeInfo: " << trace::GetDebugInfo(node->debug_info());
   }
 #endif
-  MS_LOG(DEBUG) << "End Eval NodeConfig " << conf->ToString() << ", res: " << ret_abstract->ToString();
-  return ret_abstract;
+  MS_LOG(DEBUG) << "End Eval NodeConfig " << conf->ToString() << ", res: " << eval_result->abstract()->ToString();
+  return eval_result;
 }
 
 AbstractBasePtr AnalysisEngine::EvalValueNode(const ValueNodePtr &value_node, const AnfNodeConfigPtr &conf) {
@@ -208,7 +208,7 @@ AbstractBasePtr AnalysisEngine::EvalValueNode(const ValueNodePtr &value_node, co
   return ToAbstract(value_node->value(), conf->context(), conf);
 }
 
-AbstractBasePtr AnalysisEngine::EvalCNode(const CNodePtr &cnode, const AnfNodeConfigPtr &conf) {
+EvalResultPtr AnalysisEngine::EvalCNode(const CNodePtr &cnode, const AnfNodeConfigPtr &conf) {
   MS_EXCEPTION_IF_NULL(conf);
   MS_EXCEPTION_IF_NULL(cnode);
   auto &inputs = cnode->inputs();
@@ -223,7 +223,7 @@ AbstractBasePtr AnalysisEngine::EvalCNode(const CNodePtr &cnode, const AnfNodeCo
   AnfNodeConfigPtr func_conf = MakeConfig(func_node, context);
   MS_EXCEPTION_IF_NULL(func_conf);
   // Keep it in a local variable, otherwise smart pointer will free it.
-  AbstractBasePtr maybe_func = func_conf->GetEvaluatedValue();
+  AbstractBasePtr maybe_func = func_conf->GetEvaluatedValue()->abstract();
   if (maybe_func == nullptr) {
     MS_LOG(EXCEPTION) << "func_conf.GetEvaluatedValue() return null, func_conf: " << func_conf->ToString()
                       << " NodeInfo: " << trace::GetDebugInfo(cnode->debug_info());
@@ -253,7 +253,7 @@ AbstractBasePtr AnalysisEngine::EvalCNode(const CNodePtr &cnode, const AnfNodeCo
   return ExecuteEvaluators(infs, conf, args_conf_list);
 }
 
-AbstractBasePtr AnalysisEngine::Execute(const AbstractFunctionPtr &func, const AbstractBasePtrList &args_spec_list) {
+EvalResultPtr AnalysisEngine::Execute(const AbstractFunctionPtr &func, const AbstractBasePtrList &args_spec_list) {
   ConfigPtrList args_conf_list;
   (void)std::transform(args_spec_list.begin(), args_spec_list.end(), std::back_inserter(args_conf_list),
                        [](const AbstractBasePtr &arg) -> ConfigPtr { return std::make_shared<VirtualConfig>(arg); });
@@ -454,9 +454,8 @@ EvaluatorPtr AnalysisEngine::GetEvaluatorFor(const AbstractFunctionPtr &func) {
   return tracked_eval;
 }
 
-AbstractBasePtr AnalysisEngine::ExecuteEvaluators(const std::vector<EvaluatorPtr> &evaluators,
-                                                  const AnfNodeConfigPtr &out_conf,
-                                                  const ConfigPtrList &args_conf_list) {
+EvalResultPtr AnalysisEngine::ExecuteEvaluators(const std::vector<EvaluatorPtr> &evaluators,
+                                                const AnfNodeConfigPtr &out_conf, const ConfigPtrList &args_conf_list) {
   if (evaluators.size() == 1) {
     EvaluatorPtr eval = evaluators[0];
     MS_EXCEPTION_IF_NULL(eval);
@@ -465,9 +464,9 @@ AbstractBasePtr AnalysisEngine::ExecuteEvaluators(const std::vector<EvaluatorPtr
   return ExecuteMultipleEvaluators(evaluators, out_conf, args_conf_list);
 }
 
-AbstractBasePtr AnalysisEngine::ExecuteMultipleEvaluators(const std::vector<EvaluatorPtr> &evaluators,
-                                                          const AnfNodeConfigPtr &out_conf,
-                                                          const ConfigPtrList &args_conf_list) {
+EvalResultPtr AnalysisEngine::ExecuteMultipleEvaluators(const std::vector<EvaluatorPtr> &evaluators,
+                                                        const AnfNodeConfigPtr &out_conf,
+                                                        const ConfigPtrList &args_conf_list) {
   AbstractBasePtrList out_specs;
   if (!multi_poss_.count(evaluators[0])) {
     multi_poss_[evaluators[0]] = evaluators[1];
@@ -477,7 +476,7 @@ AbstractBasePtr AnalysisEngine::ExecuteMultipleEvaluators(const std::vector<Eval
   (void)std::transform(args_conf_list.begin(), args_conf_list.end(), std::back_inserter(args_spec_list),
                        [](const ConfigPtr &conf) -> AbstractBasePtr {
                          MS_EXCEPTION_IF_NULL(conf);
-                         return conf->GetEvaluatedValue();
+                         return conf->GetEvaluatedValue()->abstract();
                        });
   for (auto eval : evaluators) {
     auto fg_eval = eval->cast<FuncGraphEvaluatorPtr>();
@@ -502,11 +501,10 @@ AbstractBasePtr AnalysisEngine::ExecuteMultipleEvaluators(const std::vector<Eval
       eval_trace_.push_back(current_inf);
       MS_LOG(DEBUG) << "Trace Evaluator " << eval->ToString() << " ptr: " << eval.get();
       MS_EXCEPTION_IF_NULL(eval);
-      auto out_spec = eval->Run(shared_from_this(), args_conf_list, out_conf);
-      MS_EXCEPTION_IF_NULL(out_spec);
-      MS_LOG(DEBUG) << "Evaluator " << eval->ToString() << " return out_spec: " << out_spec->ToString();
-      out_specs.push_back(out_spec);
-      MS_LOG(DEBUG) << "Pop Evaluator " << eval->ToString();
+      auto eval_result = eval->Run(shared_from_this(), args_conf_list, out_conf);
+      MS_EXCEPTION_IF_NULL(eval_result->abstract());
+      MS_LOG(DEBUG) << "Evaluator " << eval->ToString() << " return out_spec: " << eval_result->abstract()->ToString();
+      out_specs.push_back(eval_result->abstract());
       eval_trace_.pop_back();
       if (eval_trace_.empty()) {
         multi_poss_.clear();
@@ -552,10 +550,11 @@ AbstractBasePtr AnalysisEngine::ExecuteMultipleEvaluators(const std::vector<Eval
       // Try to travel the latest undetermined.
       if (latest_entry != eval_trace_.rbegin()->first) {
         MS_LOG(DEBUG) << "Direct Run Evaluator " << eval->ToString();
-        auto out_spec = latest_entry->Run(shared_from_this(), args_conf_list, out_conf);
-        MS_EXCEPTION_IF_NULL(out_spec);
-        MS_LOG(DEBUG) << "Evaluator " << latest_entry->ToString() << " return out_spec: " << out_spec->ToString();
-        return out_spec;
+        auto eval_result = latest_entry->Run(shared_from_this(), args_conf_list, out_conf);
+        MS_EXCEPTION_IF_NULL(eval_result->abstract());
+        MS_LOG(DEBUG) << "Evaluator " << latest_entry->ToString()
+                      << " return out_spec: " << eval_result->abstract()->ToString();
+        return eval_result;
       }
     }
   }
@@ -566,15 +565,15 @@ AbstractBasePtr AnalysisEngine::ExecuteMultipleEvaluators(const std::vector<Eval
   if (out_specs.size() == 1) {
     MS_EXCEPTION_IF_NULL(out_specs[0]);
     // If only one result derived, then broaden it to avoid wrong constant propagation.
-    return out_specs[0]->Broaden();
+    return std::make_shared<EvalResult>(out_specs[0]->Broaden(), std::make_shared<AttrValueMap>());
   }
   auto joined_spec = AbstractJoin(out_specs);
   MS_EXCEPTION_IF_NULL(joined_spec);
   MS_LOG(DEBUG) << "Multiple evaluators joined: " << joined_spec->ToString();
-  return joined_spec;
+  return std::make_shared<EvalResult>(joined_spec, std::make_shared<AttrValueMap>());
 }
 
-AbstractBasePtr AnfNodeConfig::GetEvaluatedValue() {
+EvalResultPtr AnfNodeConfig::GetEvaluatedValue() {
   AnfNodeConfigPtr self = shared_from_base<AnfNodeConfig>();
   return engine_.lock()->GetEvaluatedValue(self);
 }
@@ -607,7 +606,7 @@ AbstractBasePtr FromValueInside(const ValuePtr &value, bool broaden) {
   return a;
 }
 
-AbstractBasePtr EvalOnePrim(const PrimitivePtr &primitive, const AbstractBasePtrList &arg_specs) {
+EvalResultPtr EvalOnePrim(const PrimitivePtr &primitive, const AbstractBasePtrList &arg_specs) {
   auto evaluator = GetPrimEvaluator(primitive, nullptr);
   MS_EXCEPTION_IF_NULL(evaluator);
   if (!evaluator->isa<TrivialPrimEvaluator>()) {
@@ -615,8 +614,8 @@ AbstractBasePtr EvalOnePrim(const PrimitivePtr &primitive, const AbstractBasePtr
                       << evaluator->ToString();
   }
   auto trivial_evaluator = dyn_cast<TrivialPrimEvaluator>(evaluator);
-  auto res_spec = trivial_evaluator->EvalPrim(nullptr, arg_specs);
-  return res_spec;
+  auto eval_result = trivial_evaluator->EvalPrim(nullptr, arg_specs);
+  return eval_result;
 }
 }  // namespace abstract
 }  // namespace mindspore
diff --git a/mindspore/ccsrc/pipeline/static_analysis/static_analysis.h b/mindspore/ccsrc/pipeline/static_analysis/static_analysis.h
index e8cccd3d824..7c547266028 100644
--- a/mindspore/ccsrc/pipeline/static_analysis/static_analysis.h
+++ b/mindspore/ccsrc/pipeline/static_analysis/static_analysis.h
@@ -40,13 +40,33 @@
 
 namespace mindspore {
 namespace abstract {
+
+// define attribute value map
+using AttrValueMap = std::unordered_map<std::string, ValuePtr>;
+using AttrValueMapPtr = std::shared_ptr<AttrValueMap>;
+
+// the class to save evaluated result: abstract value and modified attribute
+class EvalResult : public Base {
+ public:
+  EvalResult(AbstractBasePtr abs, AttrValueMapPtr attr) : abstract_(abs), attribute_(attr) {}
+  ~EvalResult() override = default;
+  MS_DECLARE_PARENT(EvalResult, Base);
+  AbstractBasePtr abstract() { return abstract_; }
+  AttrValueMapPtr attribute() { return attribute_; }
+
+ private:
+  AbstractBasePtr abstract_;
+  AttrValueMapPtr attribute_;
+};
+
+using EvalResultPtr = std::shared_ptr<EvalResult>;
 // Superclass for AnfNodeConfig and VirtualConfig.
 class Config : public Base {
  public:
   Config() = default;
   ~Config() override = default;
   MS_DECLARE_PARENT(Config, Base);
-  virtual AbstractBasePtr GetEvaluatedValue() = 0;
+  virtual EvalResultPtr GetEvaluatedValue() = 0;
 };
 
 // Config will be stored in AnalysisCache
@@ -74,7 +94,7 @@ class AnfNodeConfig : public Config {
   ~AnfNodeConfig() override = default;
   MS_DECLARE_PARENT(AnfNodeConfig, Config);
 
-  AbstractBasePtr GetEvaluatedValue() override;
+  EvalResultPtr GetEvaluatedValue() override;
 
   AnalysisContextPtr context() const { return context_; }
 
@@ -123,7 +143,9 @@ class VirtualConfig : public Config {
 
   ~VirtualConfig() override = default;
   MS_DECLARE_PARENT(VirtualConfig, Config);
-  AbstractBasePtr GetEvaluatedValue() override { return abstract_; }
+  EvalResultPtr GetEvaluatedValue() override {
+    return std::make_shared<EvalResult>(abstract_, std::make_shared<AttrValueMap>());
+  }
 
  private:
   AbstractBasePtr abstract_;
@@ -135,11 +157,11 @@ class AnalysisCache {
   AnalysisCache() = default;
   ~AnalysisCache() = default;
   void Clear() { cache_.clear(); }
-  void set_value(const AnfNodeConfigPtr &conf, const AbstractBasePtr &arg);
-  AbstractBasePtr GetValue(const AnfNodeConfigPtr &conf);
+  void set_value(const AnfNodeConfigPtr &conf, const EvalResultPtr &arg);
+  EvalResultPtr GetValue(const AnfNodeConfigPtr &conf);
 
  private:
-  std::unordered_map<AnfNodeConfigPtr, AbstractBasePtr, AnfNodeConfigHasher, AnfNodeConfigEqual> cache_;
+  std::unordered_map<AnfNodeConfigPtr, EvalResultPtr, AnfNodeConfigHasher, AnfNodeConfigEqual> cache_;
 };
 
 using PrimEvaluatorMap = std::unordered_map<PrimitivePtr, EvaluatorPtr, PrimitiveHasher, PrimitiveEqual>;
@@ -147,7 +169,7 @@ using AnfNodeConfigMap =
   std::unordered_map<AnfNodeConfigPtr, AnfNodeConfigPtr, AnfNodeConfigHasher, AnfNodeConfigEqual>;
 
 struct AnalysisResult {
-  AbstractBasePtr inferred;
+  EvalResultPtr inferred;
   AnalysisContextPtr context;
 };
 
@@ -160,14 +182,14 @@ class AnalysisEngine : public std::enable_shared_from_this<AnalysisEngine> {
   // func_graph: The func_graph to analyze.
   // args_spec_list: The abstracted arguments for the func_graph. Must be a tuple of AbstractBase.
   AnalysisResult Run(const FuncGraphPtr &func_graph, const AbstractBasePtrList &args_spec_list);
-  AbstractBasePtr GetEvaluatedValue(const AnfNodeConfigPtr &conf);
+  EvalResultPtr GetEvaluatedValue(const AnfNodeConfigPtr &conf);
   // Return the Evaluator for the given function.
   EvaluatorPtr GetEvaluatorFor(const AbstractFunctionPtr &fn);
 
   AbstractBasePtr EvalValueNode(const ValueNodePtr &value_node, const AnfNodeConfigPtr &conf);
-  AbstractBasePtr EvalCNode(const CNodePtr &cnode, const AnfNodeConfigPtr &conf);
+  EvalResultPtr EvalCNode(const CNodePtr &cnode, const AnfNodeConfigPtr &conf);
   // Infer the result of fn(args).
-  AbstractBasePtr Execute(const AbstractFunctionPtr &fn, const AbstractBasePtrList &args_spec_list);
+  EvalResultPtr Execute(const AbstractFunctionPtr &fn, const AbstractBasePtrList &args_spec_list);
   void Clear();
   void ClearEvaluatorCache();
   AnalysisCache &cache() { return cache_; }
@@ -188,7 +210,7 @@ class AnalysisEngine : public std::enable_shared_from_this<AnalysisEngine> {
 
   // Set the analysis result for orig to the result for new.
   // This sets an entry in anfnode_config_map from orig to new.
-  AbstractBasePtr ForwardConfig(const AnfNodeConfigPtr &orig_conf, const AnfNodeConfigPtr new_conf) {
+  EvalResultPtr ForwardConfig(const AnfNodeConfigPtr &orig_conf, const AnfNodeConfigPtr new_conf) {
     // Use anfnode_config_map_[orig_conf] = new_conf will require AnfNodeConfig provide copy constructor.
     (void)anfnode_config_map_.emplace(orig_conf, new_conf);
     MS_LOG(DEBUG) << "Forward orig_conf: " << orig_conf->node()->DebugString()
@@ -211,12 +233,12 @@ class AnalysisEngine : public std::enable_shared_from_this<AnalysisEngine> {
 
   AnalysisContextPtr Run(const FuncGraphPtr &func_graph, const AnalysisContextPtr &context,
                          const ConfigPtrList &args_conf_list);
-  AbstractBasePtr Eval(const AnfNodeConfigPtr &conf);
+  EvalResultPtr Eval(const AnfNodeConfigPtr &conf);
   EvaluatorPtr _GetEvaluatorFor(const AbstractFunctionPtr &fn);
-  AbstractBasePtr ExecuteEvaluators(const std::vector<EvaluatorPtr> &evaluators, const AnfNodeConfigPtr &out_conf,
-                                    const ConfigPtrList &args_conf_list);
-  AbstractBasePtr ExecuteMultipleEvaluators(const std::vector<EvaluatorPtr> &evaluators,
-                                            const AnfNodeConfigPtr &out_conf, const ConfigPtrList &args_conf_list);
+  EvalResultPtr ExecuteEvaluators(const std::vector<EvaluatorPtr> &evaluators, const AnfNodeConfigPtr &out_conf,
+                                  const ConfigPtrList &args_conf_list);
+  EvalResultPtr ExecuteMultipleEvaluators(const std::vector<EvaluatorPtr> &evaluators, const AnfNodeConfigPtr &out_conf,
+                                          const ConfigPtrList &args_conf_list);
 
 #ifdef DEBUG
   std::vector<AnfNodePtr> compute_conf_stack_;
@@ -244,7 +266,7 @@ AbstractBasePtr FromValue(const T &value, bool broaden = false) {
   return FromValueInside(MakeValue(value), broaden);
 }
 
-AbstractBasePtr EvalOnePrim(const PrimitivePtr &p, const AbstractBasePtrList &arg_specs);
+EvalResultPtr EvalOnePrim(const PrimitivePtr &p, const AbstractBasePtrList &arg_specs);
 }  // namespace abstract
 }  // namespace mindspore
 
diff --git a/mindspore/ccsrc/pynative/pynative_execute.cc b/mindspore/ccsrc/pynative/pynative_execute.cc
index 451b7f4b3e0..6f0a4e57905 100644
--- a/mindspore/ccsrc/pynative/pynative_execute.cc
+++ b/mindspore/ccsrc/pynative/pynative_execute.cc
@@ -116,7 +116,7 @@ void PynativeInfer(const PrimitivePyPtr &prim, const py::tuple &py_args, OpExecI
       args_spec_list.emplace_back(abstract::FromValueInside(input_value, false));
     }
   }
-  AbstractBasePtr infer_res = EvalOnePrim(prim, args_spec_list);
+  AbstractBasePtr infer_res = EvalOnePrim(prim, args_spec_list)->abstract();
   op_exec_info->abstract = infer_res;
 }
 
diff --git a/mindspore/ccsrc/utils/graph_utils.cc b/mindspore/ccsrc/utils/graph_utils.cc
index 6a4ee58e307..60733642e78 100644
--- a/mindspore/ccsrc/utils/graph_utils.cc
+++ b/mindspore/ccsrc/utils/graph_utils.cc
@@ -26,6 +26,8 @@
 #include <list>
 #include <string>
 #include <fstream>
+#include <queue>
+#include <set>
 
 #include "ir/visitor.h"
 #include "utils/log_adapter.h"
@@ -223,6 +225,31 @@ std::vector<AnfNodePtr> TopoSort(const AnfNodePtr &root, const SuccFunc &succ, c
   return res;
 }
 
+// search the cnodes inside this graph only
+std::vector<CNodePtr> BroadFirstSearchGraphCNodes(CNodePtr ret) {
+  std::queue<CNodePtr> todo;
+  todo.push(ret);
+  std::vector<CNodePtr> sorted_nodes;
+  auto seen = NewSeenGeneration();
+  while (!todo.empty()) {
+    CNodePtr top = todo.front();
+    todo.pop();
+    sorted_nodes.push_back(top);
+    auto inputs = top->inputs();
+    for (auto &item : inputs) {
+      if (item->seen_ == seen) {
+        continue;
+      }
+
+      if (item->isa<CNode>()) {
+        todo.push(item->cast<CNodePtr>());
+      }
+      item->seen_ = seen;
+    }
+  }
+  return sorted_nodes;
+}
+
 std::vector<AnfNodePtr> SuccDeeper(const AnfNodePtr &node) {
   std::vector<AnfNodePtr> vecs;
   if (node == nullptr) {
diff --git a/mindspore/ccsrc/utils/graph_utils.h b/mindspore/ccsrc/utils/graph_utils.h
index d01335af829..7e8b36fb8e1 100644
--- a/mindspore/ccsrc/utils/graph_utils.h
+++ b/mindspore/ccsrc/utils/graph_utils.h
@@ -57,6 +57,7 @@ std::vector<AnfNodePtr> DeepLinkedGraphSearch(const AnfNodePtr &root, const Incl
 std::vector<AnfNodePtr> TopoSort(const AnfNodePtr &root, const SuccFunc &succ = SuccIncoming,
                                  const IncludeFunc &include = AlwaysInclude);
 
+std::vector<CNodePtr> BroadFirstSearchGraphCNodes(CNodePtr ret);
 class FuncGraphIndex {
  public:
   explicit FuncGraphIndex(const FuncGraphPtr &fg, const SearchFunc &search = DeepScopedGraphSearch,
diff --git a/mindspore/ops/operations/array_ops.py b/mindspore/ops/operations/array_ops.py
index 757335e2931..7fd2314e2b4 100644
--- a/mindspore/ops/operations/array_ops.py
+++ b/mindspore/ops/operations/array_ops.py
@@ -71,7 +71,6 @@ class ExpandDims(PrimitiveWithInfer):
     @prim_attr_register
     def __init__(self):
         """init ExpandDims"""
-        self.__setattr_flag__ = True
         self.init_prim_io_names(inputs=['x', 'axis'], outputs=['output'])
 
     def __infer__(self, x, axis):
@@ -182,7 +181,6 @@ class Cast(PrimitiveWithInfer):
         # if primitive need setattr in __infer__ need add this flag
         """init Cast"""
         self.init_prim_io_names(inputs=['x', 'dst_type'], outputs=['output'])
-        self.__setattr_flag__ = True
 
     def __infer__(self, x, t):
         src_type = x['dtype']
@@ -308,7 +306,6 @@ class Reshape(PrimitiveWithInfer):
     def __init__(self):
         """init Reshape"""
         self.init_prim_io_names(inputs=['tensor', 'shape'], outputs=['output'])
-        self.__setattr_flag__ = True
 
     def __infer__(self, x, shape):
         shape_v = shape['value']
@@ -453,7 +450,6 @@ class Transpose(PrimitiveWithInfer):
     @prim_attr_register
     def __init__(self):
         """init Transpose"""
-        self.__setattr_flag__ = True
         self.init_prim_io_names(inputs=['x', 'perm'], outputs=['output'])
 
     def __infer__(self, x, perm):
@@ -508,7 +504,6 @@ class GatherV2(PrimitiveWithInfer):
     @prim_attr_register
     def __init__(self):
         """init index_select"""
-        self.__setattr_flag__ = True
         self.init_prim_io_names(inputs=['params', 'indices', 'axis'], outputs=['output'])
 
     def __infer__(self, params, indices, axis):
@@ -1402,7 +1397,6 @@ class Concat(PrimitiveWithInfer):
     @prim_attr_register
     def __init__(self, axis=0):
         """init Tile"""
-        self.__setattr_flag__ = True
         validator.check_value_type("axis", axis, [int], self.name)
 
     def __infer__(self, input_x):
@@ -1476,7 +1470,6 @@ class Pack(PrimitiveWithInfer):
     @prim_attr_register
     def __init__(self, axis=0):
         """init Pack"""
-        self.__setattr_flag__ = True
         validator.check_value_type("axis", axis, [int], self.name)
         self.axis = axis
 
@@ -1526,7 +1519,6 @@ class Unpack(PrimitiveWithInfer):
     @prim_attr_register
     def __init__(self, axis=0):
         """init Unpack"""
-        self.__setattr_flag__ = True
         validator.check_value_type("axis", axis, [int], self.name)
         self.axis = axis
 
@@ -1656,7 +1648,6 @@ class Select(PrimitiveWithInfer):
     @prim_attr_register
     def __init__(self):
         """init"""
-        self.__setattr_flag__ = True
 
     def infer_shape(self, cond_shape, x_shape, y_shape):
         if cond_shape != x_shape or x_shape != y_shape:
diff --git a/mindspore/ops/operations/math_ops.py b/mindspore/ops/operations/math_ops.py
index fdea0e3672f..40d01bab09d 100644
--- a/mindspore/ops/operations/math_ops.py
+++ b/mindspore/ops/operations/math_ops.py
@@ -516,7 +516,6 @@ class MatMul(PrimitiveWithInfer):
     @prim_attr_register
     def __init__(self, transpose_a=False, transpose_b=False):
         self.init_prim_io_names(inputs=['x1', 'x2'], outputs=['output'])
-        self.__setattr_flag__ = True
         cls_name = self.name
         validator.check_value_type("transpose_a", transpose_a, [bool], cls_name)
         validator.check_value_type("transpose_b", transpose_b, [bool], cls_name)
@@ -596,7 +595,6 @@ class BatchMatMul(MatMul):
     @prim_attr_register
     def __init__(self, transpose_a=False, transpose_b=False):
         self.init_prim_io_names(inputs=['x1', 'x2'], outputs=['output'])
-        self.__setattr_flag__ = True
         cls_name = self.name
         validator.check_value_type("transpose_a", transpose_a, [bool], cls_name)
         validator.check_value_type("transpose_b", transpose_b, [bool], cls_name)
@@ -682,7 +680,6 @@ class AddN(PrimitiveWithInfer):
 
     @prim_attr_register
     def __init__(self):
-        self.__setattr_flag__ = True
         self.init_prim_io_names(inputs=["inputs"], outputs=["sum"])
 
     def infer_shape(self, inputs):
diff --git a/mindspore/ops/operations/nn_ops.py b/mindspore/ops/operations/nn_ops.py
index a4e998589e1..6552d9f47cd 100644
--- a/mindspore/ops/operations/nn_ops.py
+++ b/mindspore/ops/operations/nn_ops.py
@@ -730,8 +730,8 @@ class Conv2D(PrimitiveWithInfer):
         """init Conv2D"""
         self.init_prim_io_names(inputs=['x', 'w'], outputs=['output'])
         self.kernel_size = _check_positive_int_or_tuple('kernel_size', kernel_size, self.name)
-        self.stride = _check_positive_int_or_tuple('stride', stride, self.name)
-        self.add_prim_attr('stride', (1, 1, self.stride[0], self.stride[1]))
+        self.stride = _check_positive_int_or_tuple('stride', stride, self.name, allow_four=True, ret_four=True)
+        self.add_prim_attr('stride', self.stride)
         self.dilation = _check_positive_int_or_tuple('dilation', dilation, self.name, allow_four=True, ret_four=True)
         self.add_prim_attr('dilation', self.dilation)
         validator.check_value_type('pad', pad, (int,), self.name)
@@ -787,7 +787,6 @@ class Conv2D(PrimitiveWithInfer):
 
         self.pad_list = [pad_top, pad_bottom, pad_left, pad_right]
         self.add_prim_attr('pad_list', (pad_top, pad_bottom, pad_left, pad_right))
-
         out_channel = self.out_channel
         out_shape = [x_shape[0], out_channel, h_out, w_out]
         return out_shape
diff --git a/tests/st/ops/ascend/test_ops_infer.py b/tests/st/ops/ascend/test_ops_infer.py
new file mode 100644
index 00000000000..43d56c865c7
--- /dev/null
+++ b/tests/st/ops/ascend/test_ops_infer.py
@@ -0,0 +1,65 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+""" test nn ops """
+import functools
+import numpy as np
+import mindspore.nn as nn
+import mindspore.context as context
+import mindspore.common.dtype as mstype
+
+from mindspore import Tensor, Parameter
+from mindspore.common.initializer import initializer
+from mindspore.ops import Primitive
+from mindspore.ops import composite as C
+from mindspore.ops import operations as P
+from mindspore.ops import functional as F
+from mindspore.ops import prim_attr_register, PrimitiveWithInfer
+from mindspore.ops.primitive import constexpr
+from mindspore import context
+context.set_context(mode=context.GRAPH_MODE, save_graphs=True)
+
+
+def test_cast_op_attr():
+    class CastNet(nn.Cell):
+        def __init__(self):
+            super(CastNet, self).__init__()
+            self.cast = P.Cast()
+        def construct(self, x, t):
+            return self.cast(x, t)
+    
+    class CastTypeTest(nn.Cell):
+        def __init__(self, net):
+            super(CastTypeTest, self).__init__()
+            self.net = net
+            self.cast = P.Cast()
+        def construct(self, x, y, z):
+            cast_op = self.cast
+            t1 = cast_op(x, mstype.float32)
+            t2 = cast_op(y, mstype.int32)
+            cast_net = self.net
+            t3 = cast_net(x, mstype.float16)
+            t4 = cast_net(y, mstype.int32)
+            t5 = cast_net(z, mstype.float16)
+            return (t1, t2, t3, t4, t5)
+    net = CastTypeTest(CastNet())
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.int32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    t3 = Tensor(np.ones([1,16,1,1918]).astype(np.int32))
+    out = net(t1, t2, t3)
+    assert out[0].asnumpy().dtype == np.float32
+    assert out[1].asnumpy().dtype == np.int32
+    assert out[2].asnumpy().dtype == np.float16
+    assert out[3].asnumpy().dtype == np.int32
+    assert out[4].asnumpy().dtype == np.float16
diff --git a/tests/ut/cpp/operator/composite_test.cc b/tests/ut/cpp/operator/composite_test.cc
index 2c4b9b71468..84e9fda9d2e 100644
--- a/tests/ut/cpp/operator/composite_test.cc
+++ b/tests/ut/cpp/operator/composite_test.cc
@@ -153,7 +153,7 @@ TEST_F(TestComposite, test_TupleSlice_arg_slice) {
   auto slice = std::make_shared<AbstractSlice>(start_index, stop_index, step);
   AbstractBasePtrList args_spec_list = {tuple_tensor, slice};
 
-  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(tupleSliceGraphPtr, args_spec_list).inferred);
+  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(tupleSliceGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract tuple failed.";
   }
@@ -179,7 +179,7 @@ TEST_F(TestComposite, test_TupleSlice_arg_slice_step_none) {
   auto slice = std::make_shared<AbstractSlice>(start_index, stop_index, step);
   AbstractBasePtrList args_spec_list = {tuple_tensor, slice};
 
-  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(tupleSliceGraphPtr, args_spec_list).inferred);
+  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(tupleSliceGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract tuple failed.";
   }
@@ -205,7 +205,7 @@ TEST_F(TestComposite, test_TupleSlice_arg_slice_step_negative) {
   auto slice = std::make_shared<AbstractSlice>(start_index, stop_index, step);
   AbstractBasePtrList args_spec_list = {tuple_tensor, slice};
 
-  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(tupleSliceGraphPtr, args_spec_list).inferred);
+  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(tupleSliceGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract tuple failed.";
   }
@@ -231,7 +231,7 @@ TEST_F(TestComposite, test_TupleSlice_arg_slice_step_positive) {
   auto slice = std::make_shared<AbstractSlice>(start_index, stop_index, step);
   AbstractBasePtrList args_spec_list = {tuple_tensor, slice};
 
-  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(tupleSliceGraphPtr, args_spec_list).inferred);
+  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(tupleSliceGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract tuple failed.";
   }
@@ -253,7 +253,7 @@ TEST_F(TestComposite, test_TensorSliceBySlice) {
   AbstractSlicePtr slice = std::make_shared<AbstractSlice>(start_index, stop_index, step);
   AbstractBasePtrList args_spec_list = {tensor, slice};
 
-  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSlicePtrGraphPtr, args_spec_list).inferred);
+  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSlicePtrGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract array failed.";
   }
@@ -288,7 +288,7 @@ TEST_F(TestComposite, test_TensorSliceBySliceTuple) {
   AbstractTuplePtr slice_tuple = std::make_shared<AbstractTuple>(eles);
   AbstractBasePtrList args_spec_list = {tensor, slice_tuple};
 
-  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSliceGraphPtr, args_spec_list).inferred);
+  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSliceGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract array failed.";
   }
@@ -320,7 +320,7 @@ TEST_F(TestComposite, test_TensorSliceBySliceTupleToReduceDimension) {
   AbstractTuplePtr slice_tuple = std::make_shared<AbstractTuple>(eles);
   AbstractBasePtrList args_spec_list = {tensor, slice_tuple};
 
-  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSliceGraphPtr, args_spec_list).inferred);
+  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSliceGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract array failed.";
   }
@@ -336,7 +336,7 @@ TEST_F(TestComposite, test_TensorSliceByScalar) {
   AbstractScalarPtr start_index = std::make_shared<AbstractScalar>(2);
   AbstractBasePtrList args_spec_list = {tensor, start_index};
 
-  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSliceGraphPtr, args_spec_list).inferred);
+  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSliceGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract array failed.";
   }
@@ -358,7 +358,7 @@ TEST_F(TestComposite, test_TensorSliceByScalarTuple) {
   AbstractTuplePtr slice_tuple = std::make_shared<AbstractTuple>(eles);
   AbstractBasePtrList args_spec_list = {tensor, slice_tuple};
 
-  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSliceGraphPtr, args_spec_list).inferred);
+  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSliceGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract array failed.";
   }
@@ -382,7 +382,7 @@ TEST_F(TestComposite, test_TensorSliceByScalarTupleToScalar) {
   AbstractTuplePtr slice_tuple = std::make_shared<AbstractTuple>(eles);
   AbstractBasePtrList args_spec_list = {tensor, slice_tuple};
 
-  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSliceGraphPtr, args_spec_list).inferred);
+  AbstractTensorPtr ret = dyn_cast<AbstractTensor>(engine_->Run(tensorSliceGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract array failed.";
   }
@@ -408,7 +408,7 @@ TEST_F(TestComposite, test_UnpackCall_3args) {
   abstract::AbstractDictionaryPtr tensor_dict = std::make_shared<abstract::AbstractDictionary>(tensor_map);
 
   AbstractBasePtrList args_spec_list = {fn_arg, tensor_tuple, tensor_dict};
-  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(unPackCallGraphPtr, args_spec_list).inferred);
+  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(unPackCallGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract tuple failed.";
   }
@@ -435,7 +435,7 @@ TEST_F(TestComposite, test_UnpackCall_5args) {
   abstract::AbstractDictionaryPtr tensor_dict = std::make_shared<abstract::AbstractDictionary>(tensor_map);
 
   AbstractBasePtrList args_spec_list = {fn_arg, tensor_dict, tensor_tuple, tensor_dict, tensor_tuple};
-  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(unPackCallGraphPtr, args_spec_list).inferred);
+  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(unPackCallGraphPtr, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract tuple failed.";
   }
@@ -457,7 +457,7 @@ TEST_F(TestComposite, test_ZipOperation) {
   auto tuple = std::make_shared<AbstractTuple>(eles);
   AbstractBasePtrList args_spec_list = {tuple};
 
-  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(zip_op_graph, args_spec_list).inferred);
+  AbstractTuplePtr ret = dyn_cast<AbstractTuple>(engine_->Run(zip_op_graph, args_spec_list).inferred->abstract());
   if (ret == nullptr) {
     FAIL() << "Cast ret to abstract tuple failed.";
   }
diff --git a/tests/ut/cpp/pipeline/static_analysis/evaluator_test.cc b/tests/ut/cpp/pipeline/static_analysis/evaluator_test.cc
index 80acbe6ad57..eebe6c252ba 100644
--- a/tests/ut/cpp/pipeline/static_analysis/evaluator_test.cc
+++ b/tests/ut/cpp/pipeline/static_analysis/evaluator_test.cc
@@ -41,11 +41,11 @@ TEST_F(TestEvaluatorCacheMap, test_evaluator_cache_map) {
   AbstractBasePtr abstract_v2 = FromValue(2, false);
   AbstractBasePtrList args_spec_list = {abstract_v1, abstract_v2};
   AbstractBasePtr abstract_val = FromValue(10, false);
-  cache[args_spec_list] = abstract_val;
+  cache[args_spec_list] = std::make_shared<EvalResult>(abstract_val, std::make_shared<AttrValueMap>());
 
   auto iter = cache.find(args_spec_list);
   ASSERT_TRUE(iter != cache.end());
-  ASSERT_TRUE(iter->second == abstract_val);
+  ASSERT_TRUE(iter->second->abstract() == abstract_val);
 
   AbstractBasePtr abstract_v1_variant1 = FromValue(1, false);
   AbstractBasePtr abstract_v2_variant1 = FromValue(2, false);
@@ -53,7 +53,7 @@ TEST_F(TestEvaluatorCacheMap, test_evaluator_cache_map) {
 
   iter = cache.find(args_spec_list_variant1);
   ASSERT_TRUE(iter != cache.end());
-  ASSERT_TRUE(iter->second == abstract_val);
+  ASSERT_TRUE(iter->second->abstract() == abstract_val);
 
   AbstractBasePtr abstract_v1_variant2 = FromValue(1, false);
   AbstractBasePtr abstract_v2_variant2 = FromValue(3, false);
@@ -111,7 +111,7 @@ TEST_F(TestStandardEvaluator, test_multiple_conv2d) {
   std::vector<int> shape = {2, 2, 6, 6};
   expected->set_shape(std::make_shared<Shape>(shape));
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "result: " << res->ToString();
   MS_LOG(INFO) << "expected: " << expected->ToString();
 
@@ -144,7 +144,7 @@ TEST_F(TestPartialEvaluator, test_infer_dataclass_resolved) {
   AbstractBasePtr abstract_x = FromValue(x, false);
   args_spec_list.push_back(abstract_x);
 
-  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*(abs_base_got->GetTypeTrack()) == *(abstract_x->GetTypeTrack()));
   ASSERT_TRUE(abs_base_got->GetTypeTrack()->type_id() == kNumberTypeFloat32);
 }
@@ -160,7 +160,7 @@ TEST_F(TestPartialEvaluator, test_infer_dataclass_unresolved) {
   AbstractBasePtr abstract_x = FromValue(x, false);
   args_spec_list.push_back(abstract_x);
 
-  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*(abs_base_got->GetTypeTrack()) == *(abstract_x->GetTypeTrack()));
   ASSERT_TRUE(abs_base_got->GetTypeTrack()->type_id() == kNumberTypeFloat32);
 }
@@ -179,7 +179,7 @@ TEST_F(TestPartialEvaluator, test_infer_add_resolved) {
   args_spec_list.push_back(abstract_x);
   args_spec_list.push_back(abstract_y);
 
-  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*(abs_base_got->GetTypeTrack()) == *(abstract_x->GetTypeTrack()));
   ASSERT_TRUE(abs_base_got->GetTypeTrack()->type_id() == kNumberTypeFloat64);
 }
@@ -198,7 +198,7 @@ TEST_F(TestPartialEvaluator, test_infer_sub_unresolved) {
   args_spec_list.push_back(abstract_x);
   args_spec_list.push_back(abstract_y);
 
-  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*(abs_base_got->GetTypeTrack()) == *(abstract_x->GetTypeTrack()));
   ASSERT_TRUE(abs_base_got->GetTypeTrack()->type_id() == kNumberTypeFloat64);
 }
@@ -217,7 +217,7 @@ TEST_F(TestPartialEvaluator, test_infer_net_construct_add_resolved) {
   args_spec_list.push_back(abstract_x);
   args_spec_list.push_back(abstract_y);
 
-  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*(abs_base_got->GetTypeTrack()) == *(abstract_x->GetTypeTrack()));
   ASSERT_TRUE(abs_base_got->GetTypeTrack()->type_id() == kNumberTypeFloat64);
 }
@@ -237,7 +237,7 @@ TEST_F(TestPartialEvaluator, test_infer_construct_sub_unresolved) {
   args_spec_list.push_back(abstract_x);
   args_spec_list.push_back(abstract_y);
 
-  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*(abs_base_got->GetTypeTrack()) == *(abstract_x->GetTypeTrack()));
   ASSERT_TRUE(abs_base_got->GetTypeTrack()->type_id() == kNumberTypeFloat64);
 }
diff --git a/tests/ut/cpp/pipeline/static_analysis/prim_test.cc b/tests/ut/cpp/pipeline/static_analysis/prim_test.cc
index f54961af94f..532666b7a94 100644
--- a/tests/ut/cpp/pipeline/static_analysis/prim_test.cc
+++ b/tests/ut/cpp/pipeline/static_analysis/prim_test.cc
@@ -139,7 +139,7 @@ TEST_F(TestPrim, test_typeof) {
 
   auto prim_typeof = std::make_shared<Primitive>("typeof");
   FuncGraphPtr func_graph = MakeFuncGraph(prim_typeof, 1);
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   res->dump();
   TypePtr res_value = res->GetValueTrack()->cast<TypePtr>();
   res_value->dump();
@@ -164,7 +164,7 @@ TEST_F(TestPrim, test_list_map) {
 
   auto prim_list_map = std::make_shared<Primitive>("list_map");
   FuncGraphPtr func_graph = MakeFuncGraph(prim_list_map, 3);
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   auto expected = std::make_shared<AbstractList>(AbstractBasePtrList({FromValue(3, false), FromValue(3, false)}));
   res->dump();
   MS_LOG(INFO) << "result res: " << res->ToString();
@@ -188,7 +188,7 @@ TEST_F(TestPrim, test_list_reduce) {
 
   auto prim_list_reduce = std::make_shared<Primitive>("list_reduce");
   FuncGraphPtr func_graph = MakeFuncGraph(prim_list_reduce, 3);
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   res->dump();
   TypePtr res_type = res->GetTypeTrack();
   res_type->dump();
@@ -205,7 +205,7 @@ TEST_F(TestPrim, test_scalar_to_array) {
 
   auto prim_scalar_to_array = std::make_shared<Primitive>("scalar_to_array");
   FuncGraphPtr func_graph = MakeFuncGraph(prim_scalar_to_array, 1);
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   res->dump();
   TypePtr res_type = res->BuildType();
   res_type->dump();
@@ -223,7 +223,7 @@ TEST_F(TestPrim, test_array_to_scalar) {
 
   auto prim_array_to_scalar = std::make_shared<Primitive>("array_to_scalar");
   FuncGraphPtr func_graph = MakeFuncGraph(prim_array_to_scalar, 1);
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   res->dump();
   TypePtr res_type = res->BuildType();
   res_type->dump();
@@ -239,7 +239,7 @@ TEST_F(TestPrim, test_J_1) {
 
   auto prim_J = std::make_shared<Primitive>("J");
   FuncGraphPtr func_graph = MakeFuncGraph(prim_J, 1);
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   AbstractJTaggedPtr res_J = dyn_cast<AbstractJTagged>(res);
   ASSERT_TRUE(res_J != nullptr);
   ASSERT_TRUE(*(res_J->element()) == *abstract_v1);
@@ -280,7 +280,7 @@ TEST_F(TestPrim, test_J_2) {
   int v1 = 1;
   AbstractBasePtr abstract_v1 = FromValue(v1, false);
   AbstractBasePtrList args_spec_list = {abstract_v1};
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   res->dump();
   AbstractTuplePtr res_J = dyn_cast<AbstractTuple>(res);
   ASSERT_TRUE(res_J != nullptr);
@@ -302,7 +302,7 @@ TEST_F(TestPrim, test_dot) {
 
   AbstractBasePtrList args_spec_list = {a1, a2};
 
-  AbstractTensorPtr res = dyn_cast<AbstractTensor>(engine_->Run(func_graph, args_spec_list).inferred);
+  AbstractTensorPtr res = dyn_cast<AbstractTensor>(engine_->Run(func_graph, args_spec_list).inferred->abstract());
 
   ASSERT_TRUE(*(dyn_cast<Shape>(res->GetShapeTrack())) == *(dyn_cast<Shape>(expected->GetShapeTrack())));
 }
@@ -317,7 +317,7 @@ TEST_F(TestPrim, test_switch1) {
   AbstractBasePtr arg2 = FromValue(2, false);
   AbstractBasePtrList args_spec_list = {arg0, arg1, arg2};
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*res == *arg1);
 }
 
@@ -330,7 +330,7 @@ TEST_F(TestPrim, test_switch2) {
   AbstractBasePtr arg2 = FromValue(2, false);
   AbstractBasePtrList args_spec_list = {arg0, arg1, arg2};
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "make result res: " << res->ToString();
   MS_LOG(INFO) << "make result arg2: " << arg2->ToString();
   ASSERT_TRUE(*res == *arg2);
@@ -343,7 +343,7 @@ TEST_F(TestPrim, test_identity) {
   AbstractBasePtr abstract_v1 = FromValue(1, false);
   AbstractBasePtrList args_spec_list = {abstract_v1};
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*res == *abstract_v1);
 }
 
@@ -357,7 +357,7 @@ TEST_F(TestPrim, test_broadcast_shape) {
 
   AbstractBasePtrList args_spec_list = {a, b};
 
-  AbstractTuplePtr res = dyn_cast<AbstractTuple>(engine_->Run(func_graph, args_spec_list).inferred);
+  AbstractTuplePtr res = dyn_cast<AbstractTuple>(engine_->Run(func_graph, args_spec_list).inferred->abstract());
 
   auto ret = res->BuildValue()->cast<ValueTuplePtr>()->value();
   std::vector<ValuePtr> element_list = {MakeValue(Shape::SHP_ANY), MakeValue(Shape::SHP_ANY)};
@@ -377,7 +377,7 @@ TEST_F(TestPrim, test_partial) {
   AbstractBasePtr abstract_v2 = FromValue(1, false);
   AbstractBasePtrList args_spec_list = {abstract_add, abstract_v1, abstract_v2};
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   AbstractBasePtrList fn_args_list = {abstract_v1, abstract_v2};
   auto expected = std::make_shared<PartialAbstractClosure>(
     std::make_shared<PrimitiveAbstractClosure>(prim::kPrimScalarAdd), fn_args_list);
@@ -392,7 +392,7 @@ TEST_F(TestPrim, test_env_setitem) {
   FuncGraphPtr graph_embed = MakeFuncGraph(prim::kPrimEmbed, 1);
   AbstractBasePtr abstract_x = FromValue(1, false);
   AbstractBasePtrList args_spec_list = {abstract_x};
-  AbstractBasePtr embed_x = engine_->Run(graph_embed, args_spec_list).inferred;
+  AbstractBasePtr embed_x = engine_->Run(graph_embed, args_spec_list).inferred->abstract();
 
   FuncGraphPtr func_graph = MakeFuncGraph(prim::kPrimEnvSetItem, 3);
 
@@ -400,7 +400,7 @@ TEST_F(TestPrim, test_env_setitem) {
   AbstractBasePtr abstract_y = FromValue(2, false);
   args_spec_list = {abstract_env, embed_x, abstract_y};
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   AbstractBasePtr exp = std::make_shared<AbstractScalar>(kAnyValue, std::make_shared<EnvType>());
   ASSERT_TRUE(*res == *exp);
 }
@@ -412,7 +412,7 @@ TEST_F(TestPrim, test_env_getitem) {
   FuncGraphPtr graph_embed = MakeFuncGraph(prim::kPrimEmbed, 1);
   AbstractBasePtr abstract_x = FromValue(1, false);
   AbstractBasePtrList args_spec_list = {abstract_x};
-  AbstractBasePtr embed_x = engine_->Run(graph_embed, args_spec_list).inferred;
+  AbstractBasePtr embed_x = engine_->Run(graph_embed, args_spec_list).inferred->abstract();
 
   FuncGraphPtr graph_setitem = MakeFuncGraph(prim::kPrimEnvSetItem, 3);
 
@@ -420,7 +420,7 @@ TEST_F(TestPrim, test_env_getitem) {
   AbstractBasePtr abstract_y = FromValue(2, false);
   args_spec_list = {abstract_env, embed_x, abstract_y};
 
-  AbstractBasePtr res = engine_->Run(graph_setitem, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(graph_setitem, args_spec_list).inferred->abstract();
   AbstractBasePtr exp = std::make_shared<AbstractScalar>(kAnyValue, std::make_shared<EnvType>());
   ASSERT_TRUE(*res == *exp);
 
@@ -429,7 +429,7 @@ TEST_F(TestPrim, test_env_getitem) {
   AbstractBasePtr abstract_z = FromValue(3, false);
   args_spec_list = {res, embed_x, abstract_z};
 
-  res = engine_->Run(graph_getitem, args_spec_list).inferred;
+  res = engine_->Run(graph_getitem, args_spec_list).inferred->abstract();
 
   ASSERT_TRUE(*res == *abstract_x);
 }
@@ -442,7 +442,7 @@ TEST_F(TestPrim, test_env_add) {
   FuncGraphPtr graph_embed = MakeFuncGraph(prim::kPrimEmbed, 1);
   AbstractBasePtr abstract_x = FromValue(1, false);
   AbstractBasePtrList args_spec_list = {abstract_x};
-  AbstractBasePtr embed_x = engine_->Run(graph_embed, args_spec_list).inferred;
+  AbstractBasePtr embed_x = engine_->Run(graph_embed, args_spec_list).inferred->abstract();
 
   FuncGraphPtr graph_setitem = MakeFuncGraph(prim::kPrimEnvSetItem, 3);
 
@@ -450,19 +450,19 @@ TEST_F(TestPrim, test_env_add) {
   AbstractBasePtr abstract_y = FromValue(2, false);
   args_spec_list = {abstract_env, embed_x, abstract_y};
 
-  AbstractBasePtr abstract_e1 = engine_->Run(graph_setitem, args_spec_list).inferred;
+  AbstractBasePtr abstract_e1 = engine_->Run(graph_setitem, args_spec_list).inferred->abstract();
   AbstractBasePtr exp = std::make_shared<AbstractScalar>(kAnyValue, std::make_shared<EnvType>());
   ASSERT_TRUE(*abstract_e1 == *exp);
 
   AbstractBasePtr abstract_z = FromValue(3, false);
   args_spec_list = {abstract_env, embed_x, abstract_z};
 
-  AbstractBasePtr abstract_e2 = engine_->Run(graph_setitem, args_spec_list).inferred;
+  AbstractBasePtr abstract_e2 = engine_->Run(graph_setitem, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*abstract_e2 == *exp);
 
   FuncGraphPtr graph_add = MakeFuncGraph(prim::kPrimEnvAdd, 2);
   args_spec_list = {abstract_e1, abstract_e2};
-  AbstractBasePtr res = engine_->Run(graph_add, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(graph_add, args_spec_list).inferred->abstract();
 
   ASSERT_TRUE(*res == *exp);
 }
@@ -475,7 +475,7 @@ TEST_F(TestPrim, test_shape) {
 
   AbstractBasePtrList args_spec_list = {a};
 
-  AbstractTuplePtr res = dyn_cast<AbstractTuple>(engine_->Run(func_graph, args_spec_list).inferred);
+  AbstractTuplePtr res = dyn_cast<AbstractTuple>(engine_->Run(func_graph, args_spec_list).inferred->abstract());
   auto ret = res->BuildValue()->cast<ValueTuplePtr>()->value();
 
   std::vector<ValuePtr> element_list = {MakeValue(2), MakeValue(3)};
@@ -493,7 +493,7 @@ TEST_F(TestPrim, test_relu) {
   AbstractBasePtr expected = UTPrimUtils::ArrayFloat64Of({2, 2, 2, 3});  // NCHW
   AbstractBasePtrList args_spec_list = {expected};
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*res == *expected);
 }
 
@@ -507,7 +507,7 @@ TEST_F(TestPrim, test_relu2) {
   auto expected = ArrayOfTensor(UTPrimUtils::kF32, {3, 4, 5});
 
   AbstractBasePtrList args_spec_list = {arr};
-  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   auto res = dyn_cast<AbstractTensor>(ret);
   ASSERT_TRUE(*(res->GetShapeTrack()) == *(expected->GetShapeTrack()));
 }
@@ -540,7 +540,7 @@ TEST_F(TestPrim, test_conv2d1) {
   std::vector<int> shape = {2, 64, 14, 14};
   expected->set_shape(std::make_shared<Shape>(shape));
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "result: " << res->ToString();
   MS_LOG(INFO) << "expected: " << expected->ToString();
 
@@ -558,7 +558,7 @@ TEST_F(TestPrim, test_conv2d) {
   auto weight = ArrayOfTensor(UTPrimUtils::kF32, {64, 32, 3, 3});
 
   AbstractBasePtrList args_spec_list = {input, weight};
-  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   auto res = dyn_cast<AbstractTensor>(ret);
   auto expected = ArrayOfTensor(UTPrimUtils::kF32, {10, 64, 16, 16});
   MS_LOG(INFO) << "result: " << res->ToString();
@@ -574,7 +574,7 @@ TEST_F(TestPrim, test_conv2d_native) {
   auto weight = ArrayOfTensor(UTPrimUtils::kF64, {3, 32, 3, 3});
 
   AbstractBasePtrList args_spec_list = {input, weight};
-  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   auto res = dyn_cast<AbstractTensor>(ret);
   auto expected = ArrayOfTensor(UTPrimUtils::kF64, {10, 96, 16, 16});
   MS_LOG(INFO) << "result: " << res->ToString();
@@ -590,7 +590,7 @@ TEST_F(TestPrim, test_biasAdd) {
   auto bias = ArrayOfTensor(UTPrimUtils::kF32, {32});
 
   AbstractBasePtrList args_spec_list = {value, bias};
-  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   auto res = dyn_cast<AbstractTensor>(ret);
   auto expected = ArrayOfTensor(UTPrimUtils::kF32, {10, 32, 32, 32});
   MS_LOG(INFO) << "result: " << res->ToString();
@@ -606,7 +606,7 @@ TEST_F(TestPrim, test_softmax_cross_entropy_with_logits) {
   auto labels = ArrayOfTensor(UTPrimUtils::kF32, {64, 10});
 
   AbstractBasePtrList args_spec_list = {logits, labels};
-  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_NE(ret, nullptr);
   auto res = dyn_cast<AbstractTuple>(ret);
   auto loss = ArrayOfTensor(UTPrimUtils::kF32, {64});
@@ -636,7 +636,7 @@ TEST_F(TestPrim, test_tensor_to_scalar_prim) {
   auto labels = ArrayOfTensor(UTPrimUtils::kF64, {64, 10});
 
   AbstractBasePtrList args_spec_list = {logits, labels};
-  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   auto res = dyn_cast<AbstractScalar>(ret);
   AbstractScalarPtr expected = std::make_shared<AbstractScalar>(kAnyValue, kFloat64);
   expected->set_type(UTPrimUtils::kF64);
@@ -690,7 +690,7 @@ TEST_F(TestPrim, test_fused_batch_norm) {
   AbstractBasePtr expected0 = abstract_inputs->Clone();
   AbstractBasePtr expected1 = abstract_scale->Clone();
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "result: " << res->ToString();
   MS_LOG(INFO) << "expected0: " << expected0->ToString();
   MS_LOG(INFO) << "expected1: " << expected1->ToString();
@@ -722,7 +722,7 @@ TEST_F(TestPrim, test_pooling) {
   inputs->set_shape(inputs_dims);
   AbstractBasePtr abstract_input = FromValue(inputs, false);
   AbstractBasePtrList args_spec_list = {abstract_input};
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
 
   AbstractBasePtr expected = abstract_input->Clone()->Broaden();
   std::vector<int> expected_dims = {8, 64, 2, 2};
@@ -747,7 +747,7 @@ TEST_F(TestPrim, test_hastype) {
   auto prim = std::make_shared<Primitive>("hastype");
   FuncGraphPtr func_graph = MakeFuncGraph(prim, 2);
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*res == *expected);
 }
 
@@ -761,7 +761,7 @@ TEST_F(TestPrim, test_array_len) {
   auto prim = std::make_shared<Primitive>("array_len");
   FuncGraphPtr func_graph = MakeFuncGraph(prim, 1);
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*res == *expected);
 }
 
@@ -775,7 +775,7 @@ TEST_F(TestPrim, test_list_len) {
   auto prim = std::make_shared<Primitive>("list_len");
   FuncGraphPtr func_graph = MakeFuncGraph(prim, 1);
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*res == *expected);
 }
 
@@ -789,7 +789,7 @@ TEST_F(TestPrim, test_tuple_len) {
   auto prim = std::make_shared<Primitive>("tuple_len");
   FuncGraphPtr func_graph = MakeFuncGraph(prim, 1);
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*res == *expected);
 }
 
@@ -803,7 +803,7 @@ TEST_F(TestPrim, test_tuple_reversed) {
   auto prim = std::make_shared<Primitive>("tuple_reversed");
   FuncGraphPtr func_graph = MakeFuncGraph(prim, 1);
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "expect=" << expected->ToString();
   ASSERT_TRUE(*res == *expected);
 }
@@ -825,7 +825,7 @@ TEST_F(TestPrim, test_list_getitem) {
   auto prim = std::make_shared<Primitive>("list_getitem");
   FuncGraphPtr func_graph = MakeFuncGraph(prim, 2);
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*res == *elem);
 }
 
@@ -844,7 +844,7 @@ TEST_F(TestPrim, test_list_setitem) {
   auto prim = std::make_shared<Primitive>("list_setitem");
   FuncGraphPtr func_graph = MakeFuncGraph(prim, 3);
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "result: " << res->ToString();
   AbstractBasePtrList elems_exp = {elem1, elem2};
   auto expected = std::make_shared<AbstractList>(elems_exp);
@@ -866,7 +866,7 @@ TEST_F(TestPrim, test_list_append) {
   auto prim = std::make_shared<Primitive>("list_append");
   FuncGraphPtr func_graph = MakeFuncGraph(prim, 2);
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "result: " << res->ToString();
   auto expected = std::make_shared<AbstractList>(AbstractBasePtrList({elem1, elem2}));
   MS_LOG(INFO) << "expected: " << expected->ToString();
@@ -890,7 +890,7 @@ TEST_F(TestPrim, test_tuple_setitem) {
   auto prim = std::make_shared<Primitive>("tuple_setitem");
   FuncGraphPtr func_graph = MakeFuncGraph(prim, 3);
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "result: " << res->ToString();
   AbstractBasePtrList elems_exp = {elem1, elem2};
   auto expected = std::make_shared<AbstractTuple>(elems_exp);
@@ -916,7 +916,7 @@ TEST_F(TestPrim, test_make_list) {
   auto prim = std::make_shared<Primitive>("make_list");
   FuncGraphPtr func_graph = MakeFuncGraph(prim, 2);
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(*res == *expected);
 }
 
@@ -939,7 +939,7 @@ TEST_F(TestPrim, test_make_range) {
   AbstractBasePtrList elem_list({ele1, ele2, ele3});
   AbstractBasePtr expected = std::make_shared<AbstractTuple>(elem_list);
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "res=" << res->ToString();
   MS_LOG(INFO) << "expected=" << expected->ToString();
   ASSERT_TRUE(*res == *expected);
@@ -982,7 +982,7 @@ TEST_F(TestPrim, test_layernorm) {
   AbstractBasePtr expected1 = abstract_mean_var->Clone();
   AbstractBasePtr expected2 = abstract_mean_var->Clone();
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "result: " << res->ToString();
   MS_LOG(INFO) << "expected0: " << expected0->ToString();
   MS_LOG(INFO) << "expected1: " << expected1->ToString();
@@ -1028,7 +1028,7 @@ TEST_F(TestPrim, test_DropoutGenMask) {
   AbstractBasePtr expected = std::make_shared<AbstractTensor>(std::make_shared<AbstractScalar>(kAnyValue, kUInt8),
                                                               std::make_shared<Shape>(std::vector<int>{79}));
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "res=" << res->ToString();
   MS_LOG(INFO) << "expected=" << expected->ToString();
   ASSERT_TRUE(*res == *expected);
@@ -1058,7 +1058,7 @@ TEST_F(TestPrim, test_dropout) {
   std::vector<int> shape = {2, 20, 32, 32};
   expected->set_shape(std::make_shared<Shape>(shape));
 
-  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   MS_LOG(INFO) << "result: " << res->ToString();
   MS_LOG(INFO) << "expected: " << expected->ToString();
 
@@ -1079,7 +1079,7 @@ TEST_F(TestPrim, test_BroadcastGradientArgs_01_dim) {
   auto x_input = std::make_shared<AbstractTuple>(x_arg_list);
   auto y_input = std::make_shared<AbstractTuple>(y_arg_list);
   AbstractBasePtrList args_spec_list = {x_input, y_input};
-  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   auto res = dyn_cast<AbstractTuple>(ret);
   AbstractBasePtrList x_idx_list;
   auto r_x = std::make_shared<AbstractTuple>(x_idx_list);
@@ -1103,7 +1103,7 @@ TEST_F(TestPrim, test_BroadcastGradientArgs_1_dim) {
   auto x_input = std::make_shared<AbstractTuple>(x_arg_list);
   auto y_input = std::make_shared<AbstractTuple>(y_arg_list);
   AbstractBasePtrList args_spec_list = {x_input, y_input};
-  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   auto res = dyn_cast<AbstractTuple>(ret);
   AbstractBasePtrList x_idx_list({abstract::FromValue(1)});
   auto r_x = std::make_shared<AbstractTuple>(x_idx_list);
@@ -1128,7 +1128,7 @@ TEST_F(TestPrim, test_DictGetItem) {
   AbstractBasePtr key = abstract::FromValue("x");
   AbstractBasePtrList args_spec_list = {array_dict, key};
 
-  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   AbstractTensorPtr tensor_ret = dyn_cast<AbstractTensor>(ret);
   AbstractTensorPtr expect = dyn_cast<AbstractTensor>(FromValue(tensor_map[0].second));
 
@@ -1147,7 +1147,7 @@ TEST_F(TestPrim, test_DictGetItem2) {
   AbstractBasePtr key = abstract::FromValue("x");
   AbstractBasePtrList args_spec_list = {array_dict, key};
 
-  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr ret = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   AbstractTensorPtr tensor_ret = dyn_cast<AbstractTensor>(ret);
   AbstractTensorPtr expect = dyn_cast<AbstractTensor>(arr_x);
 
diff --git a/tests/ut/cpp/pipeline/static_analysis/static_analysis_test.cc b/tests/ut/cpp/pipeline/static_analysis/static_analysis_test.cc
index c60ac75f5b2..510651c5f03 100644
--- a/tests/ut/cpp/pipeline/static_analysis/static_analysis_test.cc
+++ b/tests/ut/cpp/pipeline/static_analysis/static_analysis_test.cc
@@ -163,7 +163,7 @@ TEST_F(TestInfer, test_inferred_scalar_add) {
 
   auto prim_scalar_add = std::make_shared<Primitive>("scalar_add");
   FuncGraphPtr func_graph = MakeFuncGraph(prim_scalar_add);
-  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred;
+  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec_list).inferred->abstract();
   ASSERT_TRUE(abs_base_got.get() == abstract_v1.get());
 }
 
@@ -261,7 +261,7 @@ TEST_F(TestInferGraph, test_inferred) {
   MS_LOG(INFO) << "" << graph_f_->get_return()->ToString();
   AbstractBasePtr abstract_v1 = FromValue(1, false);
   args_spec_list.push_back(abstract_v1);
-  AbstractBasePtr abs_base_got = engine_->Run(graph_f_, args_spec_list).inferred;
+  AbstractBasePtr abs_base_got = engine_->Run(graph_f_, args_spec_list).inferred->abstract();
   ASSERT_TRUE(abs_base_got.get() == abstract_v1.get());
 
   // now this test case failed randomly, have to debug.
@@ -272,7 +272,7 @@ TEST_F(TestInferGraph, test_inferred) {
   args_spec_list.clear();
   args_spec_list.push_back(abstract_v1);
   args_spec_list.push_back(abstract_v2);
-  abs_base_got = engine_->Run(graph_alpha_, args_spec_list).inferred;
+  abs_base_got = engine_->Run(graph_alpha_, args_spec_list).inferred->abstract();
   ASSERT_TRUE(abs_base_got.get() == abstract_v1.get());
 }
 
@@ -358,7 +358,7 @@ TEST_F(TestInferMetaGraph, test_inferred) {
   AbstractBasePtr abstract_v2 = FromValue(v1, false);
   args_spec_list.push_back(abstract_v1);
   args_spec_list.push_back(abstract_v2);
-  AbstractBasePtr abs_base_got = engine_->Run(func_graph_, args_spec_list).inferred;
+  AbstractBasePtr abs_base_got = engine_->Run(func_graph_, args_spec_list).inferred->abstract();
   ASSERT_TRUE(abs_base_got.get() == abstract_v1.get());
 }
 
@@ -390,7 +390,7 @@ TEST_F(TestInferUniform, test_inferred_scalar_add) {
 
   auto prim_scalar_add = std::make_shared<Primitive>("scalar_add");
   FuncGraphPtr func_graph = MakeFuncGraph(prim_scalar_add);
-  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec).inferred;
+  AbstractBasePtr abs_base_got = engine_->Run(func_graph, args_spec).inferred->abstract();
   ASSERT_TRUE(*(abs_base_got->GetTypeTrack()) == *(abstract_v1->GetTypeTrack()));
   ASSERT_TRUE(abs_base_got->GetTypeTrack()->type_id() == kNumberTypeInt32);
 }
@@ -418,7 +418,7 @@ TEST_F(TestEvalOnePrim, test_scalar_add) {
   AbstractBasePtr base1 = FromValue(x1, false);
   AbstractBasePtr base2 = FromValue(x2, false);
   AbstractBasePtrList base_list = {base1, base2};
-  auto res = EvalOnePrim(std::make_shared<Primitive>("scalar_add"), base_list);
+  auto res = EvalOnePrim(std::make_shared<Primitive>("scalar_add"), base_list)->abstract();
   MS_LOG(INFO) << "result spec: " << res->ToString();
   AbstractBasePtr exp = FromValue(x3, false);
   MS_LOG(INFO) << "result exp: " << exp->ToString();
@@ -446,7 +446,7 @@ void TestGraphEval::TearDown() {
 TEST_F(TestGraphInfer, test_graph_infer_defaults) {
   FuncGraphPtr graph = getPyFun.CallAndParseRet("test_graph_infer_defaults");
   AbstractBasePtrList args_spec_list = {};
-  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred->abstract();
   AbstractBasePtr expect = FromValue(MakeValue(50), false);
   ASSERT_EQ(*res, *expect);
 }
@@ -454,7 +454,7 @@ TEST_F(TestGraphInfer, test_graph_infer_defaults) {
 TEST_F(TestGraphInfer, test_graph_infer_vararg_0) {
   FuncGraphPtr graph = getPyFun.CallAndParseRet("test_graph_infer_vararg_0");
   AbstractBasePtrList args_spec_list = {};
-  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred->abstract();
   AbstractBasePtr expect = FromValue(MakeValue(1), false);
   ASSERT_EQ(*res, *expect);
 }
@@ -462,7 +462,7 @@ TEST_F(TestGraphInfer, test_graph_infer_vararg_0) {
 TEST_F(TestGraphInfer, test_graph_infer_vararg) {
   FuncGraphPtr graph = getPyFun.CallAndParseRet("test_graph_infer_vararg");
   AbstractBasePtrList args_spec_list = {};
-  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred->abstract();
   AbstractBasePtr expect = FromValue(MakeValue(9), false);
   ASSERT_EQ(*res, *expect);
 }
@@ -470,7 +470,7 @@ TEST_F(TestGraphInfer, test_graph_infer_vararg) {
 TEST_F(TestGraphInfer, test_graph_infer_vararg_kwonlyargs) {
   FuncGraphPtr graph = getPyFun.CallAndParseRet("test_graph_infer_vararg_kwonlyargs");
   AbstractBasePtrList args_spec_list = {};
-  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred->abstract();
   AbstractBasePtr expect = FromValue(MakeValue(48), false);
   ASSERT_EQ(*res, *expect);
 }
@@ -478,7 +478,7 @@ TEST_F(TestGraphInfer, test_graph_infer_vararg_kwonlyargs) {
 TEST_F(TestGraphInfer, test_graph_infer_kwarg) {
   FuncGraphPtr graph = getPyFun.CallAndParseRet("test_graph_infer_kwarg");
   AbstractBasePtrList args_spec_list = {};
-  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred->abstract();
   AbstractBasePtr expect = FromValue(MakeValue(7), false);
   ASSERT_EQ(*res, *expect);
 }
@@ -486,7 +486,7 @@ TEST_F(TestGraphInfer, test_graph_infer_kwarg) {
 TEST_F(TestGraphInfer, test_graph_infer_vararg_kwonlyargs_kwarg) {
   FuncGraphPtr graph = getPyFun.CallAndParseRet("test_graph_infer_vararg_kwonlyargs_kwarg");
   AbstractBasePtrList args_spec_list = {};
-  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred->abstract();
   AbstractBasePtr expect = FromValue(MakeValue(46), false);
   ASSERT_EQ(*res, *expect);
 }
@@ -494,7 +494,7 @@ TEST_F(TestGraphInfer, test_graph_infer_vararg_kwonlyargs_kwarg) {
 TEST_F(TestGraphInfer, test_graph_infer_vararg_kwonlyargs_kwarg_defaults) {
   FuncGraphPtr graph = getPyFun.CallAndParseRet("test_graph_infer_vararg_kwonlyargs_kwarg_defaults");
   AbstractBasePtrList args_spec_list = {};
-  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred;
+  AbstractBasePtr res = engine_->Run(graph, args_spec_list).inferred->abstract();
   AbstractBasePtr expect = FromValue(MakeValue(57), false);
   ASSERT_EQ(*res, *expect);
 }
diff --git a/tests/ut/python/ops/test_nn_ops.py b/tests/ut/python/ops/test_nn_ops.py
index 1216944fafc..3e258dba306 100644
--- a/tests/ut/python/ops/test_nn_ops.py
+++ b/tests/ut/python/ops/test_nn_ops.py
@@ -31,7 +31,8 @@ from ....mindspore_test_framework.pipeline.forward.compile_forward \
     import pipeline_for_compile_forward_ge_graph_for_case_by_case_config
 from ....mindspore_test_framework.pipeline.forward.verify_exception \
     import pipeline_for_verify_exception_for_case_by_case_config
-
+from mindspore import context
+context.set_context(mode=context.GRAPH_MODE, save_graphs=True)
 
 def conv3x3(in_channels, out_channels, stride=1, padding=1):
     """3x3 convolution """
@@ -377,6 +378,21 @@ class StateNet(nn.Cell):
         return x
 
 
+def test_conv2d_same_primitive():
+    class Conv2DSameNet(nn.Cell):
+        def __init__(self):
+            super(Conv2DSameNet, self).__init__()
+            self.conv1 = nn.Conv2d(16, 64, (1, 41), (1,4), "same", 0, 1, has_bias=True)
+            self.conv2 = nn.Conv2d(16, 64, (1, 41), (1,4), "same", 0, 1, has_bias=True)
+        def construct(self, x, y):
+            r1 = self.conv1(x)
+            r2 = self.conv2(y)
+            return (r1, r2)
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.float32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    net = Conv2DSameNet()
+    out = net(t1, t2)
+    
 class ComparisonNet(nn.Cell):
     def __init__(self):
         """ ComparisonNet definition """
diff --git a/tests/ut/python/ops/test_ops_attr_infer.py b/tests/ut/python/ops/test_ops_attr_infer.py
new file mode 100644
index 00000000000..95b66e8ff0b
--- /dev/null
+++ b/tests/ut/python/ops/test_ops_attr_infer.py
@@ -0,0 +1,276 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+""" test nn ops """
+import functools
+import numpy as np
+import mindspore
+
+import mindspore.nn as nn
+import mindspore.context as context
+import mindspore.common.dtype as mstype
+
+from mindspore import Tensor, Parameter
+from mindspore.common.initializer import initializer
+from mindspore.ops import Primitive
+from mindspore.ops import composite as C
+from mindspore.ops import operations as P
+from mindspore.ops import functional as F
+from mindspore.ops import prim_attr_register, PrimitiveWithInfer
+from mindspore.ops.primitive import constexpr
+
+from ..ut_filter import non_graph_engine
+from ....mindspore_test_framework.mindspore_test import mindspore_test
+from ....mindspore_test_framework.pipeline.forward.compile_forward \
+    import pipeline_for_compile_forward_ge_graph_for_case_by_case_config
+from ....mindspore_test_framework.pipeline.forward.verify_exception \
+    import pipeline_for_verify_exception_for_case_by_case_config
+from mindspore import context
+context.set_context(mode=context.GRAPH_MODE, save_graphs=True)
+
+class FakeOp(PrimitiveWithInfer):
+    @prim_attr_register
+    def __init__(self):
+        """"""
+    def infer_shape(self, x, y):
+        self.second_shape = y
+        self.add_prim_attr("second_shape", y)
+        return x
+    
+    def infer_dtype(self, x, y):
+        return x
+
+# test the normal case that should generate independent primitive because of different 
+# generated attributes after inference
+def test_conv2d_same_primitive():
+    class Conv2DSameNet(nn.Cell):
+        def __init__(self):
+            super(Conv2DSameNet, self).__init__()
+            self.conv1 = nn.Conv2d(16, 64, (1, 41), (1,4), "same", 0, 1, has_bias=True)
+            self.conv2 = nn.Conv2d(16, 64, (1, 41), (1,4), "same", 0, 1, has_bias=True)
+        def construct(self, x, y):
+            r1 = self.conv1(x)
+            r2 = self.conv2(y)
+            return (r1, r2)
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.float32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    net = Conv2DSameNet()
+    out = net(t1, t2)
+
+# test cell as high order argument
+# The graph with free variables used as argument is not supported yet
+# because of the limit of inference specialize system
+def Xtest_conv2d_op_with_arg():
+    class Conv2dNet(nn.Cell):
+        def __init__(self):
+            super(Conv2dNet, self).__init__()
+        def construct(self, op, x):
+            return op(x)
+    class OpsNet(nn.Cell):
+        def __init__(self, net):
+            super(OpsNet, self).__init__()
+            self.opnet = net
+            self.conv2 = nn.Conv2d(16, 64, (1, 41), (1, 4), "same", 0, 1, has_bias=True)
+        def construct(self, x, y):
+            conv_op = self.conv2
+            a = self.opnet(conv_op, x)
+            b = self.opnet(conv_op, y)
+            return (a, b)
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.float32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    net = OpsNet(Conv2dNet())
+    out = net(t1, t2)
+
+
+def test_conv2d_op_with_arg():
+    class FackOpNet(nn.Cell):
+        def __init__(self):
+            super(FackOpNet, self).__init__()
+            self.op = FakeOp()
+        def construct(self, x, y):
+            return self.op(x, y)
+    class OpNet(nn.Cell):
+        def __init__(self):
+            super(OpNet, self).__init__()
+        def construct(self, op, x, y):
+            return op(x, y)
+    class OpsNet(nn.Cell):
+        def __init__(self, net):
+            super(OpsNet, self).__init__()
+            self.opnet = net
+            self.op = FackOpNet()
+        def construct(self, x, y):
+            op = self.op
+            a = self.opnet(op, x, y)
+            b = self.opnet(op, y, x)
+            return (a, b)
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.float32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    net = OpsNet(OpNet())
+    out = net(t1, t2)
+
+
+
+def test_conv2d_op_with_arg_same_input():
+    class FackOpNet(nn.Cell):
+        def __init__(self):
+            super(FackOpNet, self).__init__()
+            self.op = FakeOp()
+        def construct(self, x, y):
+            return self.op(x, y)
+    class OpNet(nn.Cell):
+        def __init__(self):
+            super(OpNet, self).__init__()
+        def construct(self, op, x, y):
+            return op(x, y)
+    class OpsNet(nn.Cell):
+        def __init__(self, net):
+            super(OpsNet, self).__init__()
+            self.opnet = net
+            self.op = FackOpNet()
+        def construct(self, x, y):
+            op = self.op
+            a = self.opnet(op, x, x)
+            b = self.opnet(op, y, x)
+            return (a, b)
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.float32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    net = OpsNet(OpNet())
+    out = net(t1, t2)
+
+# test op with partial
+def test_op_as_partial():
+    class OpAsPartial(nn.Cell):
+        def __init__(self):
+            super(OpAsPartial, self).__init__()
+            self.op = FakeOp()
+        def construct(self, x, y, z):
+            partial_op = F.partial(self.op, x)
+            a = partial_op(y)
+            b = partial_op(z)
+            return a, b
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.float32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    t3 = Tensor(np.ones([1,16,1,1234]).astype(np.float32))
+    net = OpAsPartial()
+    out = net(t1, t2, t3)
+
+# test op with partial
+def test_op_as_partial_inside():
+    class OpAsPartial(nn.Cell):
+        def __init__(self):
+            super(OpAsPartial, self).__init__()
+            self.op = FakeOp()
+        def construct(self, x, y, z):
+            partial_op = F.partial(self.op, x)
+            a = partial_op(y)
+            b = partial_op(z)
+            return a, b
+    class OuterNet(nn.Cell):
+        def __init__(self):
+            super(OuterNet, self).__init__()
+            self.net = OpAsPartial()
+        def construct(self, x, y, z):
+            a,b = self.net(x, y, z)
+            return a, b
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.float32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    t3 = Tensor(np.ones([1,16,1,1234]).astype(np.float32))
+    net = OuterNet()
+    out = net(t1, t2, t3)
+
+# test op with partial case 2
+def test_op_as_partial_independent():
+    class OpAsPartial(nn.Cell):
+        def __init__(self):
+            super(OpAsPartial, self).__init__()
+            self.op = FakeOp()
+        def construct(self, x, y, z):
+            partial_op1 = F.partial(self.op, x)
+            a = partial_op1(y)
+            partial_op2 = F.partial(self.op, x)
+            b = partial_op2(z)
+            return a, b
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.float32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    t3 = Tensor(np.ones([1,16,1,1234]).astype(np.float32))
+    net = OpAsPartial()
+    out = net(t1, t2, t3)
+
+def test_nest_partial():
+    class NestPartial(nn.Cell):
+        def __init__(self):
+            super(NestPartial, self).__init__()
+            self.op = FakeOp()
+        def construct(self, x, y, z):
+            partial_op1 = F.partial(self.op)
+            partial_op2 = F.partial(partial_op1, x)
+            a = partial_op2(y)
+            partial_op3 = F.partial(self.op)
+            partial_op4 = F.partial(partial_op3, x)
+            b = partial_op4(z)
+            return a, b
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.float32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    t3 = Tensor(np.ones([1,16,1,1234]).astype(np.float32))
+    net = NestPartial()
+    out = net(t1, t2, t3)
+ 
+# high order argument
+# op and op args as network arguments
+def test_op_with_arg_as_input():
+    class WithOpArgNet(nn.Cell):
+        def __init__(self):
+            super(WithOpArgNet, self).__init__()
+        def construct(self, op, x, y):
+            return op(x, y)
+    class OpsNet(nn.Cell):
+        def __init__(self, net):
+            super(OpsNet, self).__init__()
+            self.opnet = net
+            self.op = FakeOp()
+        def construct(self, x, y, z):
+            op = self.op
+            a = self.opnet(op, x, z)
+            b = self.opnet(op, x, y)
+            return (a, b)
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.float32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    t3 = Tensor(np.ones([1,16,1,1234]).astype(np.float32))
+    net = OpsNet(WithOpArgNet())
+    out = net(t1, t2, t3)
+
+# The partial application used as argument is not supported yet
+# because of the limit of inference specialize system
+def Xtest_partial_as_arg():
+    class PartialArgNet(nn.Cell):
+        def __init__(self):
+            super(PartialArgNet, self).__init__()
+        def construct(self, partial_op, y):
+            return partial_op(y)
+    class OpsNet(nn.Cell):
+        def __init__(self, net):
+            super(OpsNet, self).__init__()
+            self.partial_net = net
+            self.op = FakeOp()
+        def construct(self, x, y, z):
+            partial_op = F.partial(self.op, x)
+            a = self.partial_net(partial_op, z)
+            b = self.partial_net(partial_op, y)
+            return (a, b)
+    t1 = Tensor(np.ones([1,16,1,1918]).astype(np.float32))
+    t2 = Tensor(np.ones([1,16,1,3840]).astype(np.float32))
+    t3 = Tensor(np.ones([1,16,1,1234]).astype(np.float32))
+    net = OpsNet(PartialArgNet())
+    out = net(t1, t2, t3)