and then performs quantized convolution on quantized `lhs` and `rhs`.

The internal quantization on `lhs` is a quantization to `Trhs`, dynamic range,
per-batch (per-axis along axis `dimension_numbers.input_batch_dimension`), asymmetric,
and not narrow range (the range is [Trhs_MIN, Trhs_MAX]).

`lhs` and `rhs` must be Tensors of same rank, and meet following shape conditions.
- lhs_feature % feature_group_count == 0

    flatbuffers::FlatBufferBuilder* builder) {
  switch (padding) {
    case mlir::TFL::MirrorPaddingType::REFLECT:
      return tflite::MirrorPadMode_REFLECT;
    case mlir::TFL::MirrorPaddingType::SYMMETRIC:
      return tflite::MirrorPadMode_SYMMETRIC;
  }
  llvm_unreachable("invalid mirror_pad_enum in conversion.");
}

static tflite::TensorType ConvertDerivedTypeAttrForOptionWriter(

   *
   * <ul>
   *   <li>{@code equivalent(x, x)} is true (<i>reflexive</i> property)
   *   <li>{@code equivalent(x, y)} and {@code equivalent(y, x)} each return the same result
   *       (<i>symmetric</i> property)
   *   <li>If {@code equivalent(x, y)} and {@code equivalent(y, z)} are both true, then {@code
   *       equivalent(x, z)} is also true (<i>transitive</i> property)
   * </ul>
   *

 * 
 * <p>
 * Applications can use the methods <code>getLocalHost</code>,
 * <code>getByName</code>, and
 * <code>getAllByAddress</code> to create a new NbtAddress instance. This
 * class is symmetric with {@link java.net.InetAddress}.
 *
 * <p>
 * <b>About NetBIOS:</b> The NetBIOS name
 * service is a dynamic distributed service that allows hosts to resolve

		}
		blockSize *= 2
	}
}

// symMerge merges the two sorted subsequences data[a:m] and data[m:b] using
// the SymMerge algorithm from Pok-Son Kim and Arne Kutzner, "Stable Minimum
// Storage Merging by Symmetric Comparisons", in Susanne Albers and Tomasz
// Radzik, editors, Algorithms - ESA 2004, volume 3221 of Lecture Notes in
// Computer Science, pages 714-723. Springer, 2004.
//
// Let M = m-a and N = b-n. Wolog M < N.

   *
   * <ul>
   *   <li>{@code equivalent(x, x)} is true (<i>reflexive</i> property)
   *   <li>{@code equivalent(x, y)} and {@code equivalent(y, x)} each return the same result
   *       (<i>symmetric</i> property)
   *   <li>If {@code equivalent(x, y)} and {@code equivalent(y, z)} are both true, then {@code
   *       equivalent(x, z)} is also true (<i>transitive</i> property)
   * </ul>
   *

    module_str = self._extract_first_xla_call_module_op(
        self._output_saved_model_path
    )

    # Tests that the output graph contains multiply for symmetric
    # dequantization.
    self.assertTrue(re.search('stablehlo.multiply', module_str))
    # Tests that the output graph contains float dot_general.
    self.assertTrue(

		setSize = setDriveCount
	} else {
		// Returns possible set counts with symmetry.
		setCounts = possibleSetCountsWithSymmetry(setCounts, argPatterns)

		if len(setCounts) == 0 {
			msg := fmt.Sprintf("No symmetric distribution detected with input endpoints provided %s, drives %d cannot be spread symmetrically by any supported erasure set sizes %d", args, commonSize, setSizes)
			return nil, config.ErrInvalidNumberOfErasureEndpoints(nil).Msg(msg)

	for _, a := range sumZZ {
		arg := a
		testFunZZ(t, "AddZZ", AddZZ, arg)

		arg = argZZ{a.z, a.y, a.x}
		testFunZZ(t, "AddZZ symmetric", AddZZ, arg)

		arg = argZZ{a.x, a.z, a.y}
		testFunZZ(t, "SubZZ", SubZZ, arg)

		arg = argZZ{a.y, a.z, a.x}
		testFunZZ(t, "SubZZ symmetric", SubZZ, arg)
	}
}

func TestProdZZ(t *testing.T) {
	MulZZ := func(z, x, y *Int) *Int { return z.Mul(x, y) }

    mlir::StringAttr padding) {
  return llvm::StringSwitch<mlir::TFL::MirrorPaddingType>(padding.getValue())
      .Case("REFLECT", mlir::TFL::MirrorPaddingType::REFLECT)
      .Case("SYMMETRIC", mlir::TFL::MirrorPaddingType::SYMMETRIC);
}

#include "tensorflow/compiler/mlir/lite/transforms/generated_legalize_tf.inc"

#define DECL_CONVERT_OP(tf_op)                                               \