tf_executor.fetch %ro#0, %ro#1 : tensor<i32>, tensor<i32>
  }
  return %0#0, %0#1 : tensor<i32>, tensor<i32>
}

// -----

func.func @validate_tpu_replicate_wrong_attr(%arg0: tensor<i32>, %arg1: tensor<i32>, %arg2: tensor<i32>, %arg3: tensor<i32>) -> (tensor<i32>, tensor<i32>) {
  %0:2 = tf_executor.graph {

  // CHECK: [[z:%.*]] = tf_mlrt.executeop
  %z = "tf.Identity"(%x) {__op_key = 1: i32}: (tensor<i32>) -> tensor<i32>
  // CHECK: [[y:%.*]] = tf_mlrt.await [[y_future]]
  // CHECK-NEXT: tf_mlrt.executeop([[y]]
  %w = "tf.AddV2"(%y, %z) {__op_key = 2: i32}: (tensor<i32>, tensor<i32>) -> tensor<i32>
  // CHECK-NEXT: tf_mlrt.executeop([[y]]

    %0 = "tf.Const"() {device = "/device:CPU:0", value = dense<1> : tensor<i32>} : () -> tensor<i32>
    %1 = "tf.AddV2"(%arg0, %0) {device = "/device:CPU:0"} : (tensor<i32>, tensor<i32>) -> tensor<i32>
    %2 = "tf.Div"(%arg1, %0) {device = "/device:CPU:0"} : (tensor<i32>, tensor<i32>) -> tensor<i32>
    func.return %1, %2 : tensor<i32>, tensor<i32>
  }

  %1 = "tf.ReadVariableOp"(%arg) {device = "/device:CPU:0", dtype = i32} : (tensor<!tf_type.resource<tensor<i32>>>) -> tensor<i32>
  %2 = "tf.Add"(%1, %0) {device = "/device:CPU:0"} : (tensor<i32>, tensor<i32>) -> tensor<i32>
  "tf.AssignVariableOp"(%arg, %2) {device = "/device:CPU:0"} : (tensor<!tf_type.resource<tensor<i32>>>, tensor<i32>) -> ()
  func.return %arg : tensor<!tf_type.resource<tensor<i32>>>
}

  "tf.AssignVariableOp"(%arg1, %0#1) : (tensor<*x!tf_type.resource<tensor<i32>>>, tensor<i32>) -> ()
  func.return
}

// CHECK-LABEL: func @multiple_write_func
// CHECK-SAME: ({{%.*}}: tensor<i32>) -> (tensor<i32>, tensor<i32>)
func.func @multiple_write_func(%arg0: tensor<i32>) -> (tensor<i32>, tensor<i32>) {
  func.return %arg0, %arg0 : tensor<i32>, tensor<i32>
}

    "tfl.yield"(%1#0, %1#1) : (tensor<*xi32>, tensor<*xf32>) -> ()
  }) : (tensor<i32>, tensor<1xf32>) -> (tensor<i32>, tensor<1xf32>) loc("WhileOp")
  // CHECK: (tensor<i32>, tensor<1xf32>, tensor<i32>) ->
  // CHECK-SAME: (tensor<i32>, tensor<1xf32>, tensor<i32>)
  func.return %0#1 : tensor<1xf32>
}

func.func private @WhileOp_cond(%arg0: tensor<*xi32>, %arg1: tensor<*xf32>, %arg2: tensor<i32>) -> tensor<i1> {

  %outputs_22 =  "tf.Const"() {device = "", value = dense<16> : tensor<i32>} : () -> tensor<i32>
  %outputs_24 =  "tf.Const"() {device = "", value = dense<0> : tensor<i32>} : () -> tensor<i32>
  %outputs_26 =  "tf.Range"(%outputs_24, %outputs_22, %outputs_20) {device = ""} : (tensor<i32>, tensor<i32>, tensor<i32>) -> tensor<*xi32>
  // CHECK: [[read_only_tensor_list:%.*]] = "tf.TensorListFromTensor"

      %replicated_output:2 = "tf.TPUReplicatedOutput(%identity) : (tensor<i32>) -> (tensor<i32>, tensor<i32>)
      return %replicated_output#0, %replicated_output#1 : tensor<i32>, tensor<i32>
    }
    ```

    will be transformed into:

    ```mlir
    func @tpu_computation(%arg0: tensor<i32>, %arg1: tensor<i32>) -> (tensor<i32>, tensor<i32>) {

func.func @listReserveScalarShapeI32(%arg0: tensor<i32>, %arg1: tensor<i32>) -> tensor<!tf_type.variant<tensor<*xi32>>> {
  %0 = "tf.TensorListReserve"(%arg0, %arg1) : (tensor<i32>, tensor<i32>) -> tensor<!tf_type.variant<tensor<*xi32>>>
  // CHECK: %0 = "tfl.custom"(%arg0, %arg1) <{custom_code = "TensorListReserve", custom_option = #tfl<const_bytes : "0x02">}> : (tensor<i32>, tensor<i32>) -> tensor<!tf_type.variant<tensor<*xi32>>>

  func.return %0#2, %0#3, %0#4 : tensor<i32>, tensor<256x256xf32>, tensor<?x256x256xf32>
}