diff --git a/paddle/fluid/framework/details/scope_buffered_ssa_graph_executor.h b/paddle/fluid/framework/details/scope_buffered_ssa_graph_executor.h
index 1b188aec5995edb73835bcf5b851952db0f95f48..5e87e0bf50b51d2b630aba06a5907dd721754d1f 100644
--- a/paddle/fluid/framework/details/scope_buffered_ssa_graph_executor.h
+++ b/paddle/fluid/framework/details/scope_buffered_ssa_graph_executor.h
@@ -41,7 +41,9 @@ class ScopeBufferedSSAGraphExecutor : public SSAGraphExecutor {
       std::vector<VariableInfo> var_infos, std::vector<platform::Place> places,
       std::unique_ptr<SSAGraphExecutor>&& underlying_executor);
 
-  const ir::Graph& Graph() const { return underlying_executor_->Graph(); }
+  const ir::Graph& Graph() const override {
+    return underlying_executor_->Graph();
+  }
 
   FeedFetchList Run(const std::vector<std::string>& fetch_tensors) override;
 
diff --git a/paddle/fluid/framework/details/threaded_ssa_graph_executor.h b/paddle/fluid/framework/details/threaded_ssa_graph_executor.h
index 82d6b5272aba161bb19067ebef054bc4bbb8701c..b0aaf60701fcfdfac0d9f2e547e6fb5edf63156c 100644
--- a/paddle/fluid/framework/details/threaded_ssa_graph_executor.h
+++ b/paddle/fluid/framework/details/threaded_ssa_graph_executor.h
@@ -42,7 +42,7 @@ class ThreadedSSAGraphExecutor : public SSAGraphExecutor {
                            const std::vector<platform::Place> &places,
                            std::unique_ptr<ir::Graph> &&graph);
 
-  const ir::Graph &Graph() const { return *graph_; }
+  const ir::Graph &Graph() const override { return *graph_; }
   // Run a SSAGraph by a thread pool
   // Use topological sort algorithm
   FeedFetchList Run(const std::vector<std::string> &fetch_tensors) override;