diff --git a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
index dce0c23ab8..3d55b58446 100644
--- a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
+++ b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
@@ -389,12 +389,12 @@ class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain,
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, float, LRN);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, double, LRN);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, MLFloat16, LRN);
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, float, Conv);
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, double, Conv);
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, MLFloat16, Conv);
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, float, ConvTranspose);
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, double, ConvTranspose);
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, MLFloat16, ConvTranspose);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, float, Conv);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, double, Conv);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, MLFloat16, Conv);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, float, ConvTranspose);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, double, ConvTranspose);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, MLFloat16, ConvTranspose);
 class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 9, float, AveragePool);
 class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 9, double, AveragePool);
 class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 9, MLFloat16, AveragePool);
@@ -549,13 +549,13 @@ class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDom
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, 10, Scan);
 
 // opset 10
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, float, AveragePool);
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, double, AveragePool);
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, MLFloat16, AveragePool);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, float, AveragePool);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, double, AveragePool);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, MLFloat16, AveragePool);
 class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, Dropout);
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, float, MaxPool);
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, double, MaxPool);
-class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, MLFloat16, MaxPool);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, float, MaxPool);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, double, MaxPool);
+class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, MLFloat16, MaxPool);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, NonMaxSuppression);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, float, Resize);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, double, Resize);
@@ -640,6 +640,18 @@ class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, S
 class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, Squeeze);
 class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, TopK);
 class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, Unsqueeze);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, float, Conv);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, double, Conv);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, MLFloat16, Conv);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, float, ConvTranspose);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, double, ConvTranspose);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, MLFloat16, ConvTranspose);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, float, AveragePool);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, double, AveragePool);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, MLFloat16, AveragePool);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, float, MaxPool);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, double, MaxPool);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, MLFloat16, MaxPool);
 
 static void RegisterCudaKernels(KernelRegistry& kernel_registry) {
   static const BuildKernelCreateInfoFn function_table[] = {
@@ -824,12 +836,12 @@ static void RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, float, LRN)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, double, LRN)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, MLFloat16, LRN)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, float, Conv)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, double, Conv)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, MLFloat16, Conv)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, float, ConvTranspose)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, double, ConvTranspose)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, MLFloat16, ConvTranspose)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, float, Conv)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, double, Conv)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, MLFloat16, Conv)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, float, ConvTranspose)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, double, ConvTranspose)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, MLFloat16, ConvTranspose)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 9, float, AveragePool)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 9, double, AveragePool)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 9, MLFloat16, AveragePool)>,
@@ -983,13 +995,13 @@ static void RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, 10, Scan)>,
 
       // opset 10
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, float, AveragePool)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, double, AveragePool)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, MLFloat16, AveragePool)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, float, AveragePool)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, double, AveragePool)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, MLFloat16, AveragePool)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, Dropout)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, float, MaxPool)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, double, MaxPool)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, MLFloat16, MaxPool)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, float, MaxPool)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, double, MaxPool)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, MLFloat16, MaxPool)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, 10, NonMaxSuppression)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, float, Resize)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 10, double, Resize)>,
@@ -1075,6 +1087,18 @@ static void RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, Squeeze)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, TopK)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, Unsqueeze)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, float, Conv)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, double, Conv)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, MLFloat16, Conv)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, float, ConvTranspose)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, double, ConvTranspose)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, MLFloat16, ConvTranspose)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, float, AveragePool)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, double, AveragePool)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, MLFloat16, AveragePool)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, float, MaxPool)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, double, MaxPool)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 11, MLFloat16, MaxPool)>,      
   };
 
   for (auto& function_table_entry : function_table) {
diff --git a/onnxruntime/core/providers/cuda/nn/conv.cc b/onnxruntime/core/providers/cuda/nn/conv.cc
index 1eb9643f30..01b3929d78 100644
--- a/onnxruntime/core/providers/cuda/nn/conv.cc
+++ b/onnxruntime/core/providers/cuda/nn/conv.cc
@@ -9,11 +9,21 @@
 namespace onnxruntime {
 namespace cuda {
 
+// Op Set 11 for Conv only update document to clearify default dilations and strides value.
+// which are already convered by op set 11 cpu versoin, so simply add declaration.
 #define REGISTER_KERNEL_TYPED(T)                                                \
+  ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_EX(                                      \
+      Conv,                                                                     \
+      kOnnxDomain,                                                              \
+      1, 10,                                                                    \
+      T,                                                                        \
+      kCudaExecutionProvider,                                                   \
+      KernelDefBuilder().TypeConstraint("T", DataTypeImpl::GetTensorType<T>()), \
+      Conv<T>);                                                                 \
   ONNX_OPERATOR_TYPED_KERNEL_EX(                                                \
       Conv,                                                                     \
       kOnnxDomain,                                                              \
-      1,                                                                        \
+      11,                                                                       \
       T,                                                                        \
       kCudaExecutionProvider,                                                   \
       KernelDefBuilder().TypeConstraint("T", DataTypeImpl::GetTensorType<T>()), \
diff --git a/onnxruntime/core/providers/cuda/nn/conv_transpose.cc b/onnxruntime/core/providers/cuda/nn/conv_transpose.cc
index 3e51d36223..92fcf8ac53 100644
--- a/onnxruntime/core/providers/cuda/nn/conv_transpose.cc
+++ b/onnxruntime/core/providers/cuda/nn/conv_transpose.cc
@@ -6,11 +6,21 @@
 namespace onnxruntime {
 namespace cuda {
 
+// Op Set 11 for ConvTranspose only update document to clearify default dilations and strides value.
+// which are already covered by op set 11 cpu version, so simply add declaration.
 #define REGISTER_KERNEL_TYPED(T)                                                \
+  ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_EX(                                      \
+      ConvTranspose,                                                            \
+      kOnnxDomain,                                                              \
+      1, 10,                                                                    \
+      T,                                                                        \
+      kCudaExecutionProvider,                                                   \
+      KernelDefBuilder().TypeConstraint("T", DataTypeImpl::GetTensorType<T>()), \
+      ConvTranspose<T>);                                                        \
   ONNX_OPERATOR_TYPED_KERNEL_EX(                                                \
       ConvTranspose,                                                            \
       kOnnxDomain,                                                              \
-      1,                                                                        \
+      11,                                                                       \
       T,                                                                        \
       kCudaExecutionProvider,                                                   \
       KernelDefBuilder().TypeConstraint("T", DataTypeImpl::GetTensorType<T>()), \
diff --git a/onnxruntime/core/providers/cuda/nn/pool.cc b/onnxruntime/core/providers/cuda/nn/pool.cc
index ae6508e581..8dbdeca70c 100644
--- a/onnxruntime/core/providers/cuda/nn/pool.cc
+++ b/onnxruntime/core/providers/cuda/nn/pool.cc
@@ -30,12 +30,17 @@ namespace cuda {
       KernelDefBuilder().TypeConstraint("T", DataTypeImpl::GetTensorType<data_type>()).TypeConstraint("I", DataTypeImpl::GetTensorType<int64_t>()), \
       Pool<data_type, pool_type>);
 
+
 POOLING_KERNEL_VERSIONED(AveragePool, float, AveragePool, 7, 9)
 POOLING_KERNEL_VERSIONED(AveragePool, double, AveragePool, 7, 9)
 POOLING_KERNEL_VERSIONED(AveragePool, MLFloat16, AveragePool, 7, 9)
-POOLING_KERNEL(AveragePool, float, AveragePool, 10)
-POOLING_KERNEL(AveragePool, double, AveragePool, 10)
-POOLING_KERNEL(AveragePool, MLFloat16, AveragePool, 10)
+POOLING_KERNEL_VERSIONED(AveragePool, float, AveragePool, 10, 10)
+POOLING_KERNEL_VERSIONED(AveragePool, double, AveragePool, 10, 10)
+POOLING_KERNEL_VERSIONED(AveragePool, MLFloat16, AveragePool, 10, 10)
+//AveragePool and MaxPool op set 11 only update spec document on default value for dilations and strides.
+POOLING_KERNEL(AveragePool, float, AveragePool, 11)
+POOLING_KERNEL(AveragePool, double, AveragePool, 11)
+POOLING_KERNEL(AveragePool, MLFloat16, AveragePool, 11)
 POOLING_KERNEL(GlobalAveragePool, float, AveragePool, 1)
 POOLING_KERNEL(GlobalAveragePool, double, AveragePool, 1)
 POOLING_KERNEL(GlobalAveragePool, MLFloat16, AveragePool, 1)
@@ -45,9 +50,12 @@ POOLING_KERNEL_VERSIONED(MaxPool, MLFloat16, MaxPool<1>, 1, 7)
 POOLING_KERNEL_VERSIONED(MaxPool, float, MaxPool<8>, 8, 9)
 POOLING_KERNEL_VERSIONED(MaxPool, double, MaxPool<8>, 8, 9)
 POOLING_KERNEL_VERSIONED(MaxPool, MLFloat16, MaxPool<8>, 8, 9)
-POOLING_KERNEL(MaxPool, float, MaxPool<8>, 10)
-POOLING_KERNEL(MaxPool, double, MaxPool<8>, 10)
-POOLING_KERNEL(MaxPool, MLFloat16, MaxPool<8>, 10)
+POOLING_KERNEL_VERSIONED(MaxPool, float, MaxPool<8>, 10, 10)
+POOLING_KERNEL_VERSIONED(MaxPool, double, MaxPool<8>, 10, 10)
+POOLING_KERNEL_VERSIONED(MaxPool, MLFloat16, MaxPool<8>, 10, 10)
+POOLING_KERNEL(MaxPool, float, MaxPool<8>, 11)
+POOLING_KERNEL(MaxPool, double, MaxPool<8>, 11)
+POOLING_KERNEL(MaxPool, MLFloat16, MaxPool<8>, 11)
 POOLING_KERNEL(GlobalMaxPool, float, MaxPool<1>, 1)
 POOLING_KERNEL(GlobalMaxPool, double, MaxPool<1>, 1)
 POOLING_KERNEL(GlobalMaxPool, MLFloat16, MaxPool<1>, 1)