if ((op->getNumOperands() != 2 && op->getNumOperands() != 3) ||
      op->getResultTypes().size() != 1) {
    return false;
  }
  Type lhs_type = op->getOperand(0).getType();
  Type rhs_type = op->getOperand(1).getType();
  Type result_type = op->getResult(0).getType();
  return !IsQuantizedTensorType(lhs_type) && IsQuantizedTensorType(rhs_type) &&
         !IsQuantizedTensorType(result_type);
}

  EXPECT_FALSE(mlir::isa<NoneType>(it->getOperand(1).getType()));

  // input layer norm
  EXPECT_FALSE(mlir::isa<NoneType>(it->getOperand(20).getType()));
  EXPECT_EQ(mlir::cast<RankedTensorType>(it->getOperand(20).getType())
                .getShape()
                .size(),
            1);
  EXPECT_EQ(
      mlir::cast<RankedTensorType>(it->getOperand(20).getType()).getDimSize(0),
      3);

      return;
    }
    auto element_type = splat_elements_attr.getType().getElementType();
    if (!(element_type.isF32() || element_type.isF16() ||
          element_type.isInteger(1) || element_type.isInteger(32) ||
          element_type.isInteger(64))) {
      return;
    }
    if (splat_elements_attr.getNumElements() *
            splat_elements_attr.getType().getElementTypeBitWidth() <
        kConstantSizeThresholdInBits) {

  CPred<"$0.getType() == $1.getType()">>;

// Checks if the given two values are the same.
def AreTheSameValue : Constraint<
  CPred<"$0 == $1">>;

// Checks if the value has rank.
def HasRank : Constraint<
  CPred<"$0.getType().cast<ShapedType>().hasRank()">>;

// Checks if the value has rank of `n`.
class HasRankOf<int n> : Constraint<
  CPred<"$0.getType().cast<ShapedType>().hasRank() && "

        loc, tensorflow::GetTypeFromTFTensorShape({}, rewriter.getI1Type()),
        size_diff, scalar_zero);

    // Build the argument/result types for if branch function.
    auto input_shape = rewriter.create<TF::ShapeOp>(
        loc, tensorflow::GetTypeFromTFTensorShape({-1}, shape_dtype),
        input_handle);

    Type branch_args_type[] = {input_handle.getType(), input_shape.getType(),

        return InstanceModelView.of(modelNode.getPath(), getType(), instance);
    }

    @Override
    public Optional<String> getValueDescription(MutableModelNode modelNode) {
        Object instance = this.asImmutable(ModelType.untyped(), modelNode, null).getInstance();
        if (instance == null) {
            if (isPrimitiveType(getType())) {
                if (getType().equals(ModelType.of(char.class))) {

        return new ModelView<M>() {
            private final Map<String, Object> propertyViews = new HashMap<String, Object>();

            @Override
            public ModelPath getPath() {
                return modelNode.getPath();
            }

            @Override
            public ModelType<M> getType() {

  StringAttr conv_padding = op.getPaddingAttr();
  SmallVector<int64_t> padding_nums;
  ShapedType lhs_shape = mlir::cast<ShapedType>(op.getLhs().getType());
  ShapedType rhs_shape = mlir::cast<ShapedType>(op.getRhs().getType());

  // Handle only static shape cases.
  // TODO(b/260284866): Handle dynamic shape cases.
  if (!lhs_shape.hasStaticShape()) {

        ModelType<?> type = reference.getType();
        return "{type: " + (type == null ? null : type.getDisplayName()) + ", path: " + getPath() + ", scope: " + getScope() + ", state: " + getState() + "}";
    }

    @Nullable
    @Override
    public ModelPath getPath() {
        return reference.getPath();
    }

    public ModelType<?> getType() {
        return reference.getType();
    }

    auto ty = mlir::dyn_cast<TensorType>(op.getType());
    if (!ty ||
        !mlir::isa<FloatType, IntegerType, ComplexType>(ty.getElementType()))
      return failure();

    Location loc = op.getLoc();
    Value result = rewriter.create<mhlo::ConstantOp>(loc, op.getValue());
    if (result.getType() != op.getType())
      result = rewriter.create<tensor::CastOp>(loc, op.getType(), result);
    rewriter.replaceOp(op, result);