Tout = "tfdtype$DT_FLOAT",
        window_strides = [1, 1],
        padding = "",
        explicit_padding = [],
        lhs_dilation = [],
        rhs_dilation = [],
        dimension_numbers = "",
        batch_group_count = 1,
        feature_group_count = 1,
        rhs_quantization_axis = -1,
        rhs_quantization_min_val = -128,
        rhs_quantization_max_val = 127,
        attr_map = ""

            output_feature_dimension: 3
            output_spatial_dimensions: 1
            output_spatial_dimensions: 2
          )pb",
          &dimension_numbers)) {
    return rewriter.getStringAttr("");
  }
  return rewriter.getStringAttr(dimension_numbers.SerializeAsString());
}

Attribute GetBatchGroupCountValue(
    PatternRewriter& rewriter, llvm::StringMap<Attribute>& identifier_to_attr) {

      const ::mlir::stablehlo::DotDimensionNumbersAttr dimension_numbers =
          dot.getDotDimensionNumbers();
      ArrayRef<int64_t> rhs_contracting_dims =
          dimension_numbers.getRhsContractingDimensions();
      ArrayRef<int64_t> rhs_batching_dims =
          dimension_numbers.getRhsBatchingDimensions();
      int64_t rank = dot.getRhs().getType().cast<TensorType>().getRank();

  %15 = "mhlo.gather"(%arg0, %14) <{dimension_numbers = #mhlo.gather<offset_dims = [0, 1, 3], collapsed_slice_dims = [2], start_index_map = [2], index_vector_dim = 1>, slice_sizes = dense<[4, 8, 1, 32]> : tensor<4xi64>}> : (tensor<4x8x32x32xf32>, tensor<64x1xi32>) -> tensor<4x8x64x32xf32>

// CHECK-LABEL: func private @quantized_conv2d_fn_0
// CHECK:      %[[CONV2D_0:.*]] = "tf.UniformQuantizedConvolutionHybrid"
// CHECK-SAME: batch_group_count = 1 : i64
// CHECK-SAME: dimension_numbers = "\10\03\1A\02\01\02 \02(\032\02\00\01@\03J\02\01\02"
// CHECK-SAME: explicit_padding = []
// CHECK-SAME: feature_group_count = 1 : i64
// CHECK-SAME: lhs_dilation = [1, 1]
// CHECK-SAME: padding = "VALID"

    TF_Float32Tensor:$lhs_scales,
    TF_Int32Tensor:$lhs_zps,
    TF_Float32Tensor:$rhs_scales,
    TF_Int32Tensor:$rhs_zps,
    TF_Float32Tensor:$output_scales,
    TF_Int32Tensor:$output_zps,

    StrAttr:$dimension_numbers,
    DefaultValuedOptionalAttr<I64Attr, "-1">:$lhs_quantization_axis,
    DefaultValuedOptionalAttr<I64Attr, "-128">:$lhs_quantization_min_val,
    DefaultValuedOptionalAttr<I64Attr, "127">:$lhs_quantization_max_val,

    %0 = "tf.XlaConvV2"(%lhs, %rhs, %strides, %padding, %lhs_dilation, %rhs_dilation, %feature_group_count) {dimension_numbers = "\18\03 \042\03\00\01\02@\04P\04Z\03\01\02\03b\03\01\02\03",...

// CHECK:         }
func.func @convert_conv1d(%arg0: tensor<16x32x256xbf16>, %arg1: tensor<1x256x256xbf16>) -> tensor<16x32x256xbf16> {
	%0 = "mhlo.convolution"(%arg0, %arg1) {
    batch_group_count = 1 : i64,
    dimension_numbers = #mhlo.conv<[b, 0, f]x[0, i, o]->[b, 0, f]>,
    feature_group_count = 1 : i64,
    lhs_dilation = dense<1> : tensor<1xi64>,
    padding = dense<0> : tensor<1x2xi64>,

        padding = "SAME",
        explicit_padding = [],
        lhs_dilation = [],
        rhs_dilation = [],
        batch_group_count = 1,
        feature_group_count = 1,
        dimension_numbers = "",
        lhs_quantization_axis = -1,
        lhs_quantization_min_val = -128,
        lhs_quantization_max_val = 127,
        rhs_quantization_axis = -1,
        rhs_quantization_min_val = -128,

    window_strides = [1, 2],
    padding = "VALID",
    explicit_padding = [],
    lhs_dilation = [1, 1],
    rhs_dilation = [2, 2],
    batch_group_count = 1 : i64,
    feature_group_count = 1 : i64,
    dimension_numbers = "\10\03\1A\02\01\02 \02(\032\02\00\01@\03J\02\01\02",
    rhs_quantization_axis = -1 : i64,
    rhs_quantization_min_val = -128 : i64,
    rhs_quantization_max_val = 127 : i64