// CHECK: "tf.Iterator"
  // CHECK-SAME: output_shapes{{.*}}200x28x28x1{{.*}}200x10
  // CHECK-SAME: output_types = [f32, f32]
  // CHECK-SAME: shared_name = "_iterator1"
  %0 = "tf.AnonymousIteratorV3"() {output_shapes = [
    #tf_type.shape<200x28x28x1>,
    #tf_type.shape<200x10>], output_types = [f32, f32]} : () -> tensor<!tf_type.resource>
  // CHECK: "tf.Iterator"
  // CHECK-SAME: shared_name = "_iterator2"

    // CHECK: tfrt_fallback_async.executeop key({{[0-9]+}}) cost({{.*}}) device("/device:CPU:0") "tf.FlatMapDataset"({{.*}}) {Targuments = [], metadata = "", output_shapes = [#corert.shape<>], output_types = [i64]} {f = "funcB"} : 1

  auto indices_type = RankedTensorType::get(input_shape, builder->getI64Type());
  auto output_type = RankedTensorType::get(output_shape, builder->getF32Type());
  SmallVector<mlir::Type, 2> input_types{input_type, indices_type};
  SmallVector<mlir::Type, 1> output_types{output_type};
  return createMaxUnpoolingFunc<2, 1>(builder, input_types, output_types);
}

template <int N>

    // CHECK: tfrt_fallback_async.executeop key({{[0-9]+}}) cost({{.*}}) device("/device:CPU:0") "tf.FlatMapDataset"({{.*}}) {Targuments = [], metadata = "", output_shapes = [#corert.shape<>], output_types = [i64]} {f = "__inference_Dataset_flat_map_lambda_19"} : 1

  auto output_types = fused_lstm_func_.getFunctionType().getResults();
  SmallVector<int64_t, 2> output_shape{1, mlir::ShapedType::kDynamic};
  EXPECT_EQ(mlir::cast<RankedTensorType>(output_types[0]).getShape().size(),
            output_shape.size());
  for (int i = 0; i < output_shape.size(); i++) {
    EXPECT_EQ(mlir::cast<RankedTensorType>(output_types[0]).getDimSize(i),
              output_shape[i]);

    %1 = "tf.ReduceDataset"(%arg0, %arg1) {
      Targuments = [],
      Tstate = [i64], device = "",
      f = @__reduce_func_0, f._tf_data_function = true,
      output_shapes = [#tf_type.shape<>],
      output_types = [i64], use_inter_op_parallelism = true } : (tensor<!tf_type.variant>, tensor<i64>) -> (tensor<i64>)
    func.return
}

      }

      // Collect all the quantized outputs and replace them by the results of
      // the new quantized op.
      llvm::SmallDenseMap<Value, int> outputs_replaced;
      SmallVector<Type, 4> output_types;
      output_types.reserve(candidate_op->getNumResults());
      for (const auto& enumerated_result :
           llvm::enumerate(candidate_op->getResults())) {
        Value result = enumerated_result.value();

           output_shapes = [#tf_type.shape<>], output_types = [i32],
           preserve_cardinality = false, sloppy = false,
           use_inter_op_parallelism = true,
           metadata = ""} : (tensor<*x!tf_type.variant>) -> tensor<!tf_type.variant>
  %5 = "tf.BatchDatasetV2"(%4, %0, %1) {device = "",
           output_shapes = [#tf_type.shape<>],
           output_types = [i32],
           parallel_copy = false,

// (inclusive) to `end` index region (exclusive) to `output_types` and returns
// the number of types added.
int AppendTypes(llvm::SmallVectorImpl<Type>* output_types,
                tf_device::ParallelExecuteOp parallel_execute, int start,
                int end) {
  const int size_before = output_types->size();
  for (int index = start; index < end; ++index) {

      }
    }
  }
  attr {
    key: "output_shapes"
    value {
      list {
        shape {
        }
      }
    }
  }
  attr {
    key: "output_types"
    value {
      list {
        type: DT_INT32
      }
    }
  }
  attr {
    key: "preserve_cardinality"
    value {
      b: false
    }
  }
  attr {