operators/flatten_op.cc add AsExtra (#35471)

* operators/flatten_op.cc add AsExtra

* operators/flatten_op.cc add AsExtra

* fix format

paddle/fluid/operators/flatten_op.cc

@@ -188,8 +188,8 @@ class Flatten2Op : public framework::OperatorWithKernel {
       // are the same.
       ctx->ShareLoD("X", "Out");
     }
-
-    OP_INOUT_CHECK(ctx->HasOutput("XShape"), "Output", "XShape", "Flatten2");
+    if (!ctx->HasOutput("XShape")) return;
+    // OP_INOUT_CHECK(ctx->HasOutput("XShape"), "Output", "XShape", "Flatten2");
     std::vector<int64_t> xshape_dims(in_dims.size() + 1);
     xshape_dims[0] = 0;
     for (int i = 0; i < in_dims.size(); ++i) {
@@ -207,7 +207,8 @@ class Flatten2OpMaker : public FlattenOpMaker {
     AddOutput("XShape",
               "XShape is just used to store the shape and lod of X, which will "
               "be used in FlattenGradOp.")
-        .AsIntermediate();
+        .AsIntermediate()
+        .AsExtra();
   }
 };
 
@@ -281,8 +282,8 @@ class FlattenContiguousRangeOp : public framework::OperatorWithKernel {
       // are the same.
       ctx->ShareLoD("X", "Out");
     }
-
-    OP_INOUT_CHECK(ctx->HasOutput("XShape"), "Output", "XShape", "Flatten2");
+    if (!ctx->HasOutput("XShape")) return;
+    // OP_INOUT_CHECK(ctx->HasOutput("XShape"), "Output", "XShape", "Flatten2");
     std::vector<int64_t> xshape_dims(in_dims.size() + 1);
     xshape_dims[0] = 0;
     for (int i = 0; i < in_dims.size(); ++i) {
@@ -361,7 +362,8 @@ Case 2:
     AddOutput("XShape",
               "XShape is just used to store the shape and lod of X, which will "
               "be used in FlattenGradOp.")
-        .AsIntermediate();
+        .AsIntermediate()
+        .AsExtra();
   }
 };