clear parameter when param_info clone

mindspore/ccsrc/backend/optimizer/pass/convert_const_input_to_tensor_input.cc

@@ -17,7 +17,7 @@
 
 #include <vector>
 #include <memory>
-#include <utility>
+#include <set>
 
 #include "ir/graph_utils.h"
 #include "backend/optimizer/common/helper.h"
@@ -62,6 +62,11 @@ AnfNodePtr CreateTensorInput(const AnfNodePtr &node, const KernelGraphPtr &kerne
 AnfNodePtr ConstInputToTensorInput(const FuncGraphPtr &func_graph, const CNodePtr &cnode) {
   MS_EXCEPTION_IF_NULL(func_graph);
   MS_EXCEPTION_IF_NULL(cnode);
+  const std::set<std::string> no_need_to_convert_nodes = {kStackOpName};
+  auto node_type = AnfAlgo::GetCNodeName(cnode);
+  if (no_need_to_convert_nodes.find(node_type) != no_need_to_convert_nodes.end()) {
+    return nullptr;
+  }
   std::vector<AnfNodePtr> new_inputs;
   auto kernel_graph = func_graph->cast<std::shared_ptr<session::KernelGraph>>();
   auto inputs = cnode->inputs();

mindspore/ccsrc/utils/utils.h

@@ -119,6 +119,7 @@ constexpr auto kTransDataOpName = "TransData";
 constexpr auto kStackInitOpName = "StackInit";
 constexpr auto kStackPushOpName = "StackPush";
 constexpr auto kStackPopOpName = "StackPop";
+constexpr auto kStackOpName = "Stack";
 constexpr auto kStackDestroyOpName = "StackDestroy";
 constexpr auto kBNTrainingUpdateGradOpName = "BNTrainingUpdateGrad";
 constexpr auto kBNTrainingReduceGradOpName = "BNTrainingReduceGrad";

mindspore/core/ir/param_info.h

@@ -72,6 +72,7 @@ class ParamInfo {
     this->be_cloned_ = true;
     this->be_cloned_index_.push_back(index);
     clone->init_in_server_ = this->init_in_server_;
+    clone->ClearParameter();
     return clone;
   }
 
@@ -88,6 +89,7 @@ class ParamInfo {
   void set_cache_shape(const std::vector<int64_t> &cache_shape) { cache_shape_ = cache_shape; }
   ParameterPtr parameter() { return parameter_; }
   void set_parameter(const ParameterPtr &parameter) { parameter_ = parameter; }
+  void ClearParameter() { parameter_ = nullptr; }
 
  private:
   std::string name_{"Parameter"};