diff --git a/onnx/defs/math/defs.cc b/onnx/defs/math/defs.cc
index 2a5458a02089b4bd3e32d23f17b761f85cf15096..1efc98b9dd51595b72b1025c179410e5b822539e 100644
--- a/onnx/defs/math/defs.cc
+++ b/onnx/defs/math/defs.cc
@@ -2859,6 +2859,9 @@ bool BuildContextDependentFunctionBody(
   }
   auto input_type = ctx.getInputType(0)->tensor_type().elem_type();
   bool float_input = input_type == TensorProto_DataType_FLOAT;
+  auto reduction_attr_proto = ctx.getAttribute("reduction");
+  std::string reduction_attr =
+      reduction_attr_proto != nullptr && reduction_attr_proto->has_s() ? reduction_attr_proto->s() : "mean";
   std::vector<FunctionBodyHelper::NodeDef> body;
   body.push_back(
       {{"const_zero"},
@@ -2898,7 +2901,7 @@ bool BuildContextDependentFunctionBody(
          {"loss_NCdd", "const_zero", "const_one", "const_one"}});
 
     if (!ctx.hasInput(2)) {
-      if (ctx.getAttribute("reduction")->s() == "none") {
+      if (reduction_attr == "none") {
         body.push_back(
             {{"loss"},
              "Squeeze",
@@ -2908,7 +2911,7 @@ bool BuildContextDependentFunctionBody(
             {{"loss_Ndd"},
              "Squeeze",
              {"loss_N1dd", "axes"}});
-        if (ctx.getAttribute("reduction")->s() == "mean") {
+        if (reduction_attr == "mean") {
           body.push_back(
               {{"loss"},
                "ReduceMean",
@@ -2928,12 +2931,12 @@ bool BuildContextDependentFunctionBody(
           {{"loss_unweighted"},
            "Squeeze",
            {"loss_N1dd", "axes"}});
-      if (ctx.getAttribute("reduction")->s() == "none") {
+      if (reduction_attr == "none") {
         body.push_back({{"loss"}, "Mul", {"loss_unweighted", "weight_gather"}});
       } else {
         body.push_back(
             {{"loss_Ndd"}, "Mul", {"loss_unweighted", "weight_gather"}});
-        if (ctx.getAttribute("reduction")->s() == "mean") {
+        if (reduction_attr == "mean") {
           body.push_back(
               {{"loss_sum"},
                "ReduceSum",
@@ -3052,12 +3055,12 @@ bool BuildContextDependentFunctionBody(
         {{"loss_unweighted"},
          "Squeeze",
          {"loss_N1dd", "axes"}});
-    if (ctx.getAttribute("reduction")->s() == "none") {
+    if (reduction_attr == "none") {
       body.push_back({{"loss"}, "Mul", {"loss_unweighted", "weight_gather"}});
     } else {
       body.push_back(
           {{"loss_Ndd"}, "Mul", {"loss_unweighted", "weight_gather"}});
-      if (ctx.getAttribute("reduction")->s() == "mean") {
+      if (reduction_attr == "mean") {
         body.push_back(
             {{"loss_sum"},
              "ReduceSum",
@@ -3201,7 +3204,7 @@ ONNX_OPERATOR_SET_SCHEMA(
             TensorShapeProto* output_shape =
                 ctx.getOutputType(0)->mutable_tensor_type()->mutable_shape();
 
-            if (ctx.getAttribute("reduction")->s() == "none") {
+            if (getAttribute(ctx, "reduction", "mean") == "none") {
               // output tensor is of shape (N, d1, d2, ..., dk) if
               // reduction attribute is "none".
               for (int i = 0; i < input_rank - 1; i++) {
diff --git a/onnx/defs/math/old.cc b/onnx/defs/math/old.cc
index 3c8d4ddab8af5087cc369d6a9d8a6aa26e60b23b..90ade0a66f9ba114dece4f9e101effe5432cee65 100644
--- a/onnx/defs/math/old.cc
+++ b/onnx/defs/math/old.cc
@@ -1253,6 +1253,9 @@ bool BuildContextDependentFunctionBody_opset12(
   }
   auto input_type = ctx.getInputType(0)->tensor_type().elem_type();
   bool float_input = input_type == TensorProto_DataType_FLOAT;
+  auto reduction_attr_proto = ctx.getAttribute("reduction");
+  std::string reduction_attr =
+      reduction_attr_proto != nullptr && reduction_attr_proto->has_s() ? reduction_attr_proto->s() : "mean";
   std::vector<FunctionBodyHelper::NodeDef> body;
   body.push_back(
       {{"const_zero"},
@@ -1287,7 +1290,7 @@ bool BuildContextDependentFunctionBody_opset12(
          {"loss_NCdd", "const_zero", "const_one", "const_one"}});
 
     if (!ctx.hasInput(2)) {
-      if (ctx.getAttribute("reduction")->s() == "none") {
+      if (reduction_attr == "none") {
         body.push_back(
             {{"loss"},
              "Squeeze",
@@ -1299,7 +1302,7 @@ bool BuildContextDependentFunctionBody_opset12(
              "Squeeze",
              {"loss_N1dd"},
              {MakeAttribute("axes", std::vector<int64_t>({1}))}});
-        if (ctx.getAttribute("reduction")->s() == "mean") {
+        if (reduction_attr == "mean") {
           body.push_back(
               {{"loss"},
                "ReduceMean",
@@ -1320,12 +1323,12 @@ bool BuildContextDependentFunctionBody_opset12(
            "Squeeze",
            {"loss_N1dd"},
            {MakeAttribute("axes", std::vector<int64_t>({1}))}});
-      if (ctx.getAttribute("reduction")->s() == "none") {
+      if (reduction_attr == "none") {
         body.push_back({{"loss"}, "Mul", {"loss_unweighted", "weight_gather"}});
       } else {
         body.push_back(
             {{"loss_Ndd"}, "Mul", {"loss_unweighted", "weight_gather"}});
-        if (ctx.getAttribute("reduction")->s() == "mean") {
+        if (reduction_attr == "mean") {
           body.push_back(
               {{"loss_sum"},
                "ReduceSum",
@@ -1447,12 +1450,12 @@ bool BuildContextDependentFunctionBody_opset12(
          "Squeeze",
          {"loss_N1dd"},
          {MakeAttribute("axes", std::vector<int64_t>({1}))}});
-    if (ctx.getAttribute("reduction")->s() == "none") {
+    if (reduction_attr == "none") {
       body.push_back({{"loss"}, "Mul", {"loss_unweighted", "weight_gather"}});
     } else {
       body.push_back(
           {{"loss_Ndd"}, "Mul", {"loss_unweighted", "weight_gather"}});
-      if (ctx.getAttribute("reduction")->s() == "mean") {
+      if (reduction_attr == "mean") {
         body.push_back(
             {{"loss_sum"},
              "ReduceSum",
@@ -1532,64 +1535,62 @@ ONNX_OPERATOR_SET_SCHEMA(
             "Constrain target to integer types")
         .SetContextDependentFunctionBodyBuilder(
             BuildContextDependentFunctionBody_opset12)
-        .TypeAndShapeInferenceFunction([](InferenceContext& ctx) {
-          // Type inference
-          propagateElemTypeFromInputToOutput(ctx, 0, 0);
-
-          // Shape inference
-          if (hasNInputShapes(ctx, 2)) {
-            const TensorShapeProto& input_shape =
-                ctx.getInputType(0)->tensor_type().shape();
-            const TensorShapeProto& target_shape =
-                ctx.getInputType(1)->tensor_type().shape();
-
-            const int input_rank = static_cast<int>(input_shape.dim_size());
-            const int target_rank = static_cast<int>(target_shape.dim_size());
-
-            if (input_rank < 2) {
-              fail_shape_inference("Input rank must be >= 2.");
-            }
-            if (target_rank != input_rank - 1) {
-              fail_shape_inference(
-                  "Target rank must be 1 less than the input rank.");
-            }
+        .TypeAndShapeInferenceFunction([](InferenceContext& ctx) { 
+            // Type inference
+  propagateElemTypeFromInputToOutput(ctx, 0, 0);
+
+  // Shape inference
+  if (hasNInputShapes(ctx, 2)) {
+    const TensorShapeProto& input_shape =
+     ctx.getInputType(0)->tensor_type().shape();
+    const TensorShapeProto& target_shape =
+     ctx.getInputType(1)->tensor_type().shape();
+
+    const int input_rank = static_cast<int>(input_shape.dim_size());
+    const int target_rank = static_cast<int>(target_shape.dim_size());
+
+    if (input_rank < 2) {
+      fail_shape_inference("Input rank must be >= 2.");
+    }
+    if (target_rank != input_rank - 1) {
+      fail_shape_inference(
+          "Target rank must be 1 less than the input rank.");
+    }
 
-            // match input dimensions (N, C, d1, ..., dk) with target
-            // dimensions of (C, d1, ..., dk)
-            for (int dim = 0; dim < target_rank; dim++) {
-              const auto input_dim =
-                  dim == 0 ? input_shape.dim(dim) : input_shape.dim(dim + 1);
-              const auto target_dim = target_shape.dim(dim);
-              if (input_dim.has_dim_value() && target_dim.has_dim_value() &&
-                  input_dim.dim_value() != target_dim.dim_value())
-                fail_shape_inference(
-                    "Input and target dimension value mismatch.");
-            }
+    // match input dimensions (N, C, d1, ..., dk) with target
+    // dimensions of (C, d1, ..., dk)
+    for (int dim = 0; dim < target_rank; dim++) {
+      const auto input_dim =
+       dim == 0 ? input_shape.dim(dim) : input_shape.dim(dim + 1);
+      const auto target_dim = target_shape.dim(dim);
+      if (input_dim.has_dim_value() && target_dim.has_dim_value() &&
+          input_dim.dim_value() != target_dim.dim_value())
+        fail_shape_inference(
+            "Input and target dimension value mismatch.");
+    }
 
-            if (ctx.getNumInputs() == 3 && hasInputShape(ctx, 2)) {
-              const TensorShapeProto& weight_shape =
-                  ctx.getInputType(2)->tensor_type().shape();
-              if (weight_shape.dim_size() != 1) {
-                fail_shape_inference("Weight rank must be 1.");
-              }
-            }
+    if (ctx.getNumInputs() == 3 && hasInputShape(ctx, 2)) {
+      const TensorShapeProto& weight_shape =
+          ctx.getInputType(2)->tensor_type().shape();
+      if (weight_shape.dim_size() != 1) {
+        fail_shape_inference("Weight rank must be 1.");
+      }
+    }
 
-            TensorShapeProto* output_shape =
-                ctx.getOutputType(0)->mutable_tensor_type()->mutable_shape();
-
-            if (ctx.getAttribute("reduction")->s() == "none") {
-              // output tensor is of shape (N, d1, d2, ..., dk) if
-              // reduction attribute is "none".
-              for (int i = 0; i < input_rank - 1; i++) {
-                auto* dim = output_shape->add_dim();
-                if (i == 0)
-                  *dim = input_shape.dim(i);
-                else
-                  *dim = input_shape.dim(i + 1);
-              }
-            }
-            // otherwise output is a scalar.
-          }
+    TensorShapeProto* output_shape = ctx.getOutputType(0)->mutable_tensor_type()->mutable_shape();
+    if (getAttribute(ctx, "reduction", "mean") == "none") {
+      // output tensor is of shape (N, d1, d2, ..., dk) if
+      // reduction attribute is "none".
+      for (int i = 0; i < input_rank - 1; i++) {
+        auto* dim = output_shape->add_dim();
+        if (i == 0)
+          *dim = input_shape.dim(i);
+        else
+          *dim = input_shape.dim(i + 1);
+      }
+    }
+    // otherwise output is a scalar.
+  }
         }));
 
 const char* reduction_doc_sce_opset12 =