[Dup] Fix SAME_UPPER/SAME_LOWER (auto_pad attribute) in ConvTranspose (#12537)

* Fix SAME_UPPER/SAME_LOWER (auto_pad attribute) in ConvTranspose * Bump ONNX 1.10.2 globally * load ONNX_VERSION from VERSION_NUMBER * / * revert deprecate warning in ORT 1.12 * add a comment about why removing cntk_simple_seg * correct the implem in DML as well
2026-07-29 20:14:01 +00:00 · 2022-08-22 15:35:34 -07:00 · 2022-08-22 15:35:34 -07:00 · 6246662b1d
commit 6246662b1d
parent c144acc534
9 changed files with 13 additions and 22 deletions
--- a/csharp/test/Microsoft.ML.OnnxRuntime.Tests.NetCoreApp/InferenceTest.netcore.cs
+++ b/csharp/test/Microsoft.ML.OnnxRuntime.Tests.NetCoreApp/InferenceTest.netcore.cs
@ -229,6 +229,7 @@ namespace Microsoft.ML.OnnxRuntime.Tests
                { "tf_resnet_v1_50", "result mismatch when Conv BN Fusion is applied" },
                { "tf_resnet_v1_101", "result mismatch when Conv BN Fusion is applied" },
                { "tf_resnet_v1_152", "result mismatch when Conv BN Fusion is applied" },
+                { "cntk_simple_seg", "Bad onnx test output caused by wrong SAME_UPPER/SAME_LOWER for ConvTranspose" },    
                { "coreml_Imputer-LogisticRegression_sklearn_load_breast_cancer", "Can't determine model file name" },
                { "mask_rcnn_keras", "Model should be edited to remove the extra outputs" },
                { "test_strnormalizer_export_monday_casesensintive_lower", "ElementType not currently supported"},
@ -304,7 +305,6 @@ namespace Microsoft.ML.OnnxRuntime.Tests
                { "test_min_uint16", "node test error"},
                { "test_adam_multiple", "node test error"},
                { "test_loop13_seq", "node test error"},
-                { "test_convtranspose_autopad_same", "node test error"},
                { "test_training_dropout_default_mask", "node test error"},
                { "test_min_int8", "node test error"},
                { "test_identity_sequence", "data type not supported"},
--- a/onnxruntime/core/providers/cpu/nn/conv_transpose.h
+++ b/onnxruntime/core/providers/cpu/nn/conv_transpose.h
@ -25,15 +25,7 @@ namespace onnxruntime {
 template <typename T>
 class ConvTranspose : public OpKernel {
 public:
-  ConvTranspose(const OpKernelInfo& info) : OpKernel(info), conv_transpose_attrs_(info) {
-    if (conv_transpose_attrs_.auto_pad == AutoPadType::SAME_UPPER ||
-        conv_transpose_attrs_.auto_pad == AutoPadType::SAME_LOWER) {
-      // TODO(jcwchen): #9740 ORT 1.13 will correct the logic by switching them to meet ONNX spec
-      LOGS_DEFAULT(WARNING) << "The existing bug in the padding distribution for auto_pad type"
-                            << " SAME_UPPER/SAME_LOWER will be fixed in next ORT 1.13 release and hence the"
-                            << " results of ConvTranspose operator using the above auto_pad type(s) will be different.";
-    }
-  }
+  ConvTranspose(const OpKernelInfo& info) : OpKernel(info), conv_transpose_attrs_(info) {}

  Status PrePack(const Tensor& tensor, int input_idx, AllocatorPtr alloc,
                 /*out*/ bool& is_packed,
--- a/onnxruntime/core/providers/cpu/nn/conv_transpose_attributes.h
+++ b/onnxruntime/core/providers/cpu/nn/conv_transpose_attributes.h
@ -176,7 +176,7 @@ struct ConvTransposeAttributes : public ConvAttributes {

  void DistributePadding(AutoPadType pad_type, const int64_t& total_pad,
                         int64_t& pad_head, int64_t& pad_tail) const {
-    if (pad_type == AutoPadType::SAME_UPPER) {  // pad more on head when total_pad is odd.
+    if (pad_type == AutoPadType::SAME_LOWER) {  // pad more on head when total_pad is odd.
      pad_head = total_pad - total_pad / 2;
      pad_tail = total_pad / 2;
    } else {
--- a/onnxruntime/core/providers/dml/OperatorAuthorHelper/OperatorHelper.cpp
+++ b/onnxruntime/core/providers/dml/OperatorAuthorHelper/OperatorHelper.cpp
@ -721,7 +721,7 @@ namespace OperatorHelper
                int paddings = gsl::narrow_cast<int>((inputDimensions[i + dimOffset] - 1) * stride + windowSize - m_outputShapes[0].GetShape()[i + dimOffset]);
                paddings = std::max<int>(0, paddings);

-                m_kernel.startPadding[i] = m_kernel.autoPadSameUpper ? (paddings + 1) / 2 : paddings / 2;
+                m_kernel.startPadding[i] = m_kernel.autoPadSameUpper ? paddings / 2 : (paddings + 1) / 2;
                m_kernel.endPadding[i] = paddings - m_kernel.startPadding[i];
            }
        }
--- a/onnxruntime/test/onnx/main.cc
+++ b/onnxruntime/test/onnx/main.cc
@ -627,12 +627,13 @@ select from 'TF8', 'TF16', 'UINT8', 'FLOAT', 'ITENSOR'. \n)");
    {"BERT_Squad", "test data bug"},
    {"constantofshape_float_ones", "test data bug", {"onnx141", "onnx150"}},
    {"constantofshape_int_zeros", "test data bug", {"onnx141", "onnx150"}},
-    {"convtranspose_autopad_same", "Implementation need to be adjusted for ONNX changes"},
+    {"convtranspose_autopad_same", "Test data has been corrected in ONNX 1.10.", {"onnx180", "onnx181", "onnx190"}},
    {"cast_STRING_to_FLOAT", "Linux CI has old ONNX python package with bad test data", {"onnx141"}},
    // Numpy float to string has unexpected rounding for some results given numpy default precision is meant to be 8.
    // "e.g. 0.296140194 -> '0.2961402' not '0.29614019'. ORT produces the latter with precision set to 8,
    // which doesn't match the expected output that was generated with numpy.
    {"cast_FLOAT_to_STRING", "Numpy float to string has unexpected rounding for some results."},
+    {"cntk_simple_seg", "Bad onnx test output caused by wrong SAME_UPPER/SAME_LOWER for ConvTranspose", {}},
    {"tf_nasnet_large", "disable temporarily"},
    {"tf_nasnet_mobile", "disable temporarily"},
    {"tf_pnasnet_large", "disable temporarily"},
--- a/onnxruntime/test/providers/cpu/model_tests.cc
+++ b/onnxruntime/test/providers/cpu/model_tests.cc
@ -166,6 +166,7 @@ TEST_P(ModelTest, Run) {
      {"bitshift_left_uint16", "BitShift(11) uint16 support not enabled currently"},
      {"maxunpool_export_with_output_shape",
       "Invalid output in ONNX test. See https://github.com/onnx/onnx/issues/2398"},
+      {"cntk_simple_seg", "Bad onnx test output caused by wrong SAME_UPPER/SAME_LOWER for ConvTranspose"},
      {"training_dropout", "result differs", {}},               // Temporary, subsequent PR will remove this.
      {"training_dropout_default", "result differs", {}},       // Temporary, subsequent PR will remove this.
      {"training_dropout_default_mask", "result differs", {}},  // Temporary, subsequent PR will remove this.
@ -184,7 +185,7 @@ TEST_P(ModelTest, Run) {
      {"castlike_FLOAT_to_BFLOAT16_expanded", "type error", {}},
      {"castlike_FLOAT_to_STRING", "type error", {}},
      {"castlike_FLOAT_to_STRING_expanded", "type error", {}},
-      {"convtranspose_autopad_same", "type error", {}},
+      {"convtranspose_autopad_same", "Test data has been corrected in ONNX 1.10.", {"onnx180", "onnx181", "onnx190"}},
      {"gru_batchwise", "type error", {}},
      {"lstm_batchwise", "type error", {}},
      {"optional_get_element", "type error", {}},
--- a/onnxruntime/test/providers/cpu/nn/conv_transpose_op_test.cc
+++ b/onnxruntime/test/providers/cpu/nn/conv_transpose_op_test.cc
@ -1045,7 +1045,7 @@ TEST(ConvTransposeTest, ConvTranspose_1D_AutoPad_SameUpper) {
  vector<float> W = {1.0f, 1.0f, 1.0f, 1.0f};
  vector<int64_t> W_shape = {1, 2, 2};
  vector<int64_t> Y_shape = {1, 2, 4};
-  auto expected_vals = {3.0f, 5.0f, 7.0f, 4.0f, 3.0f, 5.0f, 7.0f, 4.0f};
+  auto expected_vals = {1.0f, 3.0f, 5.0f, 7.0f, 1.0f, 3.0f, 5.0f, 7.0f};

  TestConvTransposeOp(attrs, {X, W}, {X_shape, W_shape}, expected_vals, Y_shape,
                      OpTester::ExpectResult::kExpectSuccess, "", {kTensorrtExecutionProvider, kOpenVINOExecutionProvider}); //Accuracy Mismatch on OpenVINO-EP
@ -1068,7 +1068,7 @@ TEST(ConvTransposeTest, ConvTranspose_1D_AutoPad_SameLower) {
  vector<float> W = {1.0f, 1.0f, 1.0f, 1.0f};
  vector<int64_t> W_shape = {1, 2, 2};
  vector<int64_t> Y_shape = {1, 2, 4};
-  auto expected_vals = {1.0f, 3.0f, 5.0f, 7.0f, 1.0f, 3.0f, 5.0f, 7.0f};
+  auto expected_vals = {3.0f, 5.0f, 7.0f, 4.0f, 3.0f, 5.0f, 7.0f, 4.0f};

  TestConvTransposeOp(attrs, {X, W}, {X_shape, W_shape}, expected_vals, Y_shape,
                      OpTester::ExpectResult::kExpectSuccess, "", {kTensorrtExecutionProvider, kOpenVINOExecutionProvider}); //Accuracy Mismatch on OpenVINO-EP
@ -1083,7 +1083,7 @@ TEST(ConvTransposeTest, ConvTranspose_AutoPad_with_non_default_strides) {
      vector<int64_t>{2, 2},  // strides
      vector<int64_t>{1, 1},  // dilations
      1,                      // group
-      "SAME_LOWER"            // auto_pad
+      "SAME_UPPER"            // auto_pad
  };

  vector<float> X = {0.0f, 1.0f, 2.0f,
--- a/onnxruntime/test/testdata/onnx_backend_test_series_filters.jsonc
+++ b/onnxruntime/test/testdata/onnx_backend_test_series_filters.jsonc
@ -35,12 +35,7 @@
        "^test_batchnorm_epsilon_training_mode",
        "^test_batchnorm_example_old",
        "^test_batchnorm_example_training_mode",
-        "^convtranspose_autopad_same",
-        "^test_convtranspose_autopad_same",
-        "^test_convtranspose_autopad_same_cuda", // Due to changes in ONNX 1.9
        "^test_gathernd_example_int32_batch_dim1",
-        "^test_convtranspose_autopad_same", // ONNX 1.9 changes
-        "^test_convtranspose_autopad_same_cuda", // ONNX 1.9 changes
        "^test_max_int16",
        "^test_max_int8",
        "^test_max_uint16",
--- a/winml/test/model/model_tests.cpp
+++ b/winml/test/model/model_tests.cpp
@ -227,6 +227,8 @@ static std::vector<ITestCase*> GetAllTestCases() {
  };
  allDisabledTests.insert(std::begin(x86DisabledTests), std::end(x86DisabledTests));
 #endif
+// Bad onnx test output caused by previously wrong SAME_UPPER/SAME_LOWER for ConvTranspose
+allDisabledTests.insert(ORT_TSTR("cntk_simple_seg"));

  
  WINML_EXPECT_NO_THROW(LoadTests(dataDirs, whitelistedTestCases, TestTolerances(1e-3, 1e-3, {}, {}),