%1 = "tf.OpB"() {_XlaSharding = "\08\03\1A\02\02\01\22\02\00\02"} : () -> tensor<2x2xi64>
        // expected-error @+1 {{should have the same sharding}}
        %2 = "tf.OpC"(%0, %1) {_xla_map_outside_compilation = "0", _xla_outside_compilation = "from_launch"} : (tensor<2x2xi64>, tensor<2x2xi64>) -> tensor<2x2xi64>
        "tf.OpD"(%2) : (tensor<2x2xi64>) -> ()
        tf_device.return

  func.return %0 : tensor<?x?xf32>
}

// -----

func.func @stridedSliceGrad(%dy: tensor<4x8xf32>, %begin: tensor<2xi64>, %end: tensor<2xi64>, %strides: tensor<2xi64>, %shape: tensor<1x2xi64>) -> tensor<?x?xf32> {
  // expected-error @+1 {{'shape' operand must be 1D tensor, but got 2D tensor}}

// CHECK-LABEL: testAddWithI64Broadcasting
func.func @testAddWithI64Broadcasting(tensor< 2x3xi64>, tensor<3xi64>) -> tensor<2x3xi64> {
^bb0(%arg0: tensor<2x3xi64>, %arg1: tensor<3xi64>):
  // CHECK: tfl.add(%arg0, %arg1)
  %0 = tfl.add(%arg0, %arg1) {fused_activation_function = "RELU6"} : (tensor< 2x3xi64>, tensor<3xi64>) -> tensor<2x3xi64>
  func.return %0#0 : tensor<2x3xi64>
}

// -----

// CHECK-LABEL: add_with_i32_five_dim_broadcasting

  // CHECK: "tf.Conv3DBackpropInputV2"
}

func.func @mul_i64(%arg0: tensor<14xi64>, %arg1: tensor<14xi64>) -> tensor<14xi64> {
  %0 = "tf.Mul"(%arg0, %arg1) : (tensor<14xi64>, tensor<14xi64>) -> tensor<14xi64>
  func.return %0: tensor<14xi64>

// CHECK-LABEL: mul_i64
// CHECK:  tfl.mul %arg0, %arg1 {fused_activation_function = "NONE"} : tensor<14xi64>
// CHECK:  return
}

  }

  // CHECK-LABEL: func @xla_reduce_window
  func.func @xla_reduce_window(%arg0: tensor<7xf32>, %arg1: tensor<f32>) -> tensor<?xf32> {
    %cst = "tf.Const"() {value = dense<0> : tensor<1x2xi32>} : () -> tensor<1x2xi32>
    %cst_0 = "tf.Const"() {value = dense<1> : tensor<1xi32>} : () -> tensor<1xi32>
    %cst_1 = "tf.Const"() {value = dense<2> : tensor<1xi32>} : () -> tensor<1xi32>

// CHECK: %[[BITCAST2:.+]] = "tfl.bitcast"(%[[ARG2]]) : (tensor<i64>) -> tensor<1xi64>
// CHECK: %[[MIN2:.+]] = "tfl.minimum"(%[[BITCAST2]], %[[MAX2]]) : (tensor<1xi64>, tensor<1xi64>) -> tensor<1xi64>
// CHECK: %[[CONCAT:.+]] = "tfl.concatenation"(%[[MIN1]], %[[MIN2]]) <{axis = 0 : i32, fused_activation_function = "NONE"}> : (tensor<1xi64>, tensor<1xi64>) -> tensor<2xi64>

}

// CHECK-LABEL: @testMatrixSetDiag
func.func @testMatrixSetDiag(%arg0: tensor<3x3xi64>, %arg1: tensor<3xi64>) -> tensor<3x3xi64> {
  %0 = "tf.MatrixSetDiag"(%arg0, %arg1) {device = "/job:localhost/replica:0/task:0/device:GPU:0"} : (tensor<3x3xi64>, tensor<3xi64>) -> tensor<3x3xi64>
  func.return %0 : tensor<3x3xi64>

  // CHECK: %[[ZERO:.*]] = "tf.Const"() <{value = dense<0> : tensor<i32>}>