diff --git a/lib/Conversion/StablehloToEmitC/StablehloRegionOpsToEmitC.cpp b/lib/Conversion/StablehloToEmitC/StablehloRegionOpsToEmitC.cpp
index cb78e22f..ff396442 100644
--- a/lib/Conversion/StablehloToEmitC/StablehloRegionOpsToEmitC.cpp
+++ b/lib/Conversion/StablehloToEmitC/StablehloRegionOpsToEmitC.cpp
@@ -27,15 +27,6 @@ using namespace mlir::emitc;
 namespace {
 
 /// Common functions.
-/// Adopted from mlir-hlo.
-DenseIntElementsAttr i64ElementsAttr(int64_t value, size_t count,
-                                     MLIRContext *ctx) {
-  RankedTensorType ty = RankedTensorType::get({static_cast<int64_t>(count)},
-                                              IntegerType::get(ctx, 64));
-  SmallVector<int64_t, 4> values(count, value);
-  return DenseIntElementsAttr::get(ty, values);
-}
-
 SmallVector<Attribute, 2> indexSequence(int64_t n, MLIRContext *ctx) {
   return llvm::to_vector<2>(
       llvm::map_range(llvm::seq<int64_t>(0, n), [&ctx](int64_t i) -> Attribute {
@@ -202,14 +193,14 @@ struct ConvertStablehloRegionOpsToEmitCPass
 
     size_t dim = op.getResult(0).getType().cast<RankedTensorType>().getRank();
     arguments.push_back(op.getWindowDimensions());
-    arguments.push_back(
-        op.getWindowStrides().value_or(i64ElementsAttr(1, dim, ctx)));
-    arguments.push_back(
-        op.getBaseDilations().value_or(i64ElementsAttr(1, dim, ctx)));
-    arguments.push_back(
-        op.getBaseDilations().value_or(i64ElementsAttr(1, dim, ctx)));
-    arguments.push_back(
-        op.getPadding().value_or(i64ElementsAttr(0, 2 * dim, ctx)));
+    arguments.push_back(op.getWindowStrides().value_or(
+        builder.getI64TensorAttr(SmallVector<int64_t>(dim, 1))));
+    arguments.push_back(op.getBaseDilations().value_or(
+        builder.getI64TensorAttr(SmallVector<int64_t>(dim, 1))));
+    arguments.push_back(op.getBaseDilations().value_or(
+        builder.getI64TensorAttr(SmallVector<int64_t>(dim, 1))));
+    arguments.push_back(op.getPadding().value_or(
+        builder.getI64TensorAttr(SmallVector<int64_t>(2 * dim, 0))));
     arguments.push_back(SymbolRefAttr::get(ctx, funcOp.getName()));
 
     ArrayAttr args = ArrayAttr::get(ctx, arguments);
diff --git a/lib/Conversion/StablehloToEmitC/StablehloToEmitC.cpp b/lib/Conversion/StablehloToEmitC/StablehloToEmitC.cpp
index 4f2d3d38..549f2895 100644
--- a/lib/Conversion/StablehloToEmitC/StablehloToEmitC.cpp
+++ b/lib/Conversion/StablehloToEmitC/StablehloToEmitC.cpp
@@ -27,23 +27,6 @@ using namespace mlir::emitc;
 namespace {
 
 /// Common functions.
-/// Adopted from mlir-hlo.
-DenseIntElementsAttr i64ElementsAttr(int64_t value, size_t count,
-                                     MLIRContext *ctx) {
-  RankedTensorType ty = RankedTensorType::get({static_cast<int64_t>(count)},
-                                              IntegerType::get(ctx, 64));
-  SmallVector<int64_t, 4> values(count, value);
-  return DenseIntElementsAttr::get(ty, values);
-}
-
-DenseIntElementsAttr getI64ElementsAttr(const ArrayRef<long> values,
-                                        MLIRContext *ctx) {
-  RankedTensorType ty = RankedTensorType::get(
-      {static_cast<int64_t>(values.size())}, IntegerType::get(ctx, 64));
-
-  return DenseIntElementsAttr::get(ty, values);
-}
-
 SmallVector<Attribute, 2> indexSequence(int64_t n, MLIRContext *ctx) {
   return llvm::to_vector<2>(
       llvm::map_range(llvm::seq<int64_t>(0, n), [&ctx](int64_t i) -> Attribute {
@@ -172,7 +155,6 @@ class ConvOpConversion : public OpConversionPattern<stablehlo::ConvolutionOp> {
   LogicalResult
   matchAndRewrite(stablehlo::ConvolutionOp convOp, OpAdaptor adaptor,
                   ConversionPatternRewriter &rewriter) const override {
-    auto *ctx = convOp.getContext();
 
     StringRef funcName = "emitc::stablehlo::convolution";
     StringAttr callee = rewriter.getStringAttr(funcName);
@@ -201,14 +183,14 @@ class ConvOpConversion : public OpConversionPattern<stablehlo::ConvolutionOp> {
         convOp.getDimensionNumbers().getOutputSpatialDimensions()));
     arguments.push_back(convOp.getFeatureGroupCountAttr());
 
-    arguments.push_back(
-        convOp.getPadding().value_or(i64ElementsAttr(0, 2, ctx)));
-    arguments.push_back(
-        convOp.getLhsDilation().value_or(i64ElementsAttr(1, 2, ctx)));
-    arguments.push_back(
-        convOp.getRhsDilation().value_or(i64ElementsAttr(1, 2, ctx)));
-    arguments.push_back(
-        convOp.getWindowStrides().value_or(i64ElementsAttr(1, 2, ctx)));
+    arguments.push_back(convOp.getPadding().value_or(
+        rewriter.getI64TensorAttr(SmallVector<int64_t>(2, 0))));
+    arguments.push_back(convOp.getLhsDilation().value_or(
+        rewriter.getI64TensorAttr(SmallVector<int64_t>(2, 1))));
+    arguments.push_back(convOp.getRhsDilation().value_or(
+        rewriter.getI64TensorAttr(SmallVector<int64_t>(2, 1))));
+    arguments.push_back(convOp.getWindowStrides().value_or(
+        rewriter.getI64TensorAttr(SmallVector<int64_t>(2, 1))));
 
     ArrayAttr args = rewriter.getArrayAttr(arguments);
     ArrayAttr templateArgs =
@@ -318,12 +300,11 @@ class SliceOpConversion : public OpConversionPattern<stablehlo::SliceOp> {
     SmallVector<Attribute, 2> arguments =
         indexSequence(adaptor.getOperands().size(), sliceOp.getContext());
 
-    arguments.push_back(getI64ElementsAttr(sliceOp.getStartIndicesAttr(),
-                                           sliceOp.getContext()));
-    arguments.push_back(getI64ElementsAttr(sliceOp.getLimitIndicesAttr(),
-                                           sliceOp.getContext()));
     arguments.push_back(
-        getI64ElementsAttr(sliceOp.getStridesAttr(), sliceOp.getContext()));
+        rewriter.getI64TensorAttr(sliceOp.getStartIndicesAttr()));
+    arguments.push_back(
+        rewriter.getI64TensorAttr(sliceOp.getLimitIndicesAttr()));
+    arguments.push_back(rewriter.getI64TensorAttr(sliceOp.getStridesAttr()));
 
     ArrayAttr args = rewriter.getArrayAttr(arguments);
 
@@ -357,8 +338,8 @@ class DynamicSliceOpConversion
     SmallVector<Attribute, 2> arguments = indexSequence(
         adaptor.getOperands().size(), dynamicSliceOp.getContext());
 
-    arguments.push_back(getI64ElementsAttr(dynamicSliceOp.getSliceSizesAttr(),
-                                           dynamicSliceOp.getContext()));
+    arguments.push_back(
+        rewriter.getI64TensorAttr(dynamicSliceOp.getSliceSizesAttr()));
 
     ArrayAttr args = rewriter.getArrayAttr(arguments);
 
@@ -423,11 +404,11 @@ class PadOpConversion : public OpConversionPattern<stablehlo::PadOp> {
         indexSequence(adaptor.getOperands().size(), padOp.getContext());
 
     arguments.push_back(
-        getI64ElementsAttr(padOp.getEdgePaddingLowAttr(), padOp.getContext()));
+        rewriter.getI64TensorAttr(padOp.getEdgePaddingLowAttr()));
     arguments.push_back(
-        getI64ElementsAttr(padOp.getEdgePaddingHighAttr(), padOp.getContext()));
+        rewriter.getI64TensorAttr(padOp.getEdgePaddingHighAttr()));
     arguments.push_back(
-        getI64ElementsAttr(padOp.getInteriorPaddingAttr(), padOp.getContext()));
+        rewriter.getI64TensorAttr(padOp.getInteriorPaddingAttr()));
 
     ArrayAttr args = rewriter.getArrayAttr(arguments);
 
@@ -460,8 +441,8 @@ class TransposeOpConversion
     SmallVector<Attribute> arguments =
         indexSequence(adaptor.getOperands().size(), transposeOp.getContext());
 
-    arguments.push_back(getI64ElementsAttr(transposeOp.getPermutationAttr(),
-                                           transposeOp.getContext()));
+    arguments.push_back(
+        rewriter.getI64TensorAttr(transposeOp.getPermutationAttr()));
     ArrayAttr args = rewriter.getArrayAttr(arguments);
 
     Type resultType = transposeOp.getResult().getType();
diff --git a/lib/Conversion/TosaToEmitC/TosaToEmitC.cpp b/lib/Conversion/TosaToEmitC/TosaToEmitC.cpp
index 600a74ba..a5fc5e0d 100644
--- a/lib/Conversion/TosaToEmitC/TosaToEmitC.cpp
+++ b/lib/Conversion/TosaToEmitC/TosaToEmitC.cpp
@@ -27,14 +27,6 @@ using namespace mlir::emitc;
 namespace {
 
 /// Common functions.
-DenseIntElementsAttr getI64ElementsAttr(const ArrayRef<long> values,
-                                        MLIRContext *ctx) {
-  RankedTensorType ty = RankedTensorType::get(
-      {static_cast<int64_t>(values.size())}, IntegerType::get(ctx, 64));
-
-  return DenseIntElementsAttr::get(ty, values);
-}
-
 SmallVector<Attribute, 2> indexSequence(int64_t n, MLIRContext *ctx) {
   return llvm::to_vector<2>(
       llvm::map_range(llvm::seq<int64_t>(0, n), [&ctx](int64_t i) -> Attribute {
@@ -115,9 +107,9 @@ class GenericConvOpConversion : public OpConversionPattern<SrcOp> {
     ArrayAttr args = rewriter.getArrayAttr({
       rewriter.getIndexAttr(0),
       rewriter.getIndexAttr(1),
-      getI64ElementsAttr(convOp.getPad(), convOp.getContext()),
-      getI64ElementsAttr(convOp.getStride(), convOp.getContext()),
-      getI64ElementsAttr(convOp.getDilation(), convOp.getContext()),
+      rewriter.getI64TensorAttr(convOp.getPad()),
+      rewriter.getI64TensorAttr(convOp.getStride()),
+      rewriter.getI64TensorAttr(convOp.getDilation()),
     });
     // clang-format on
 
@@ -161,9 +153,9 @@ class GenericPoolOpConversion : public OpConversionPattern<SrcOp> {
     // clang-format off
     ArrayAttr args = rewriter.getArrayAttr({
       rewriter.getIndexAttr(0),
-      getI64ElementsAttr(poolOp.getPad(), poolOp.getContext()),
-      getI64ElementsAttr(poolOp.getStride(), poolOp.getContext()),
-      getI64ElementsAttr(poolOp.getKernel(), poolOp.getContext()),
+      rewriter.getI64TensorAttr(poolOp.getPad()),
+      rewriter.getI64TensorAttr(poolOp.getStride()),
+      rewriter.getI64TensorAttr(poolOp.getKernel()),
     });
     // clang-format on