func.func @main(%arg0: tensor<5x7xf32>) -> tensor<5x7xf32> {
  func.return %arg0: tensor<5x7xf32>
// CHECK-LABEL: main
// CHECK: return %arg0 : tensor<5x7xf32>
}

// - transpose
//
func.func @transpose_2d(%arg0: tensor<2x3xf32>) -> tensor<3x2xf32> {
  %0 = "mhlo.transpose"(%arg0) <{permutation = dense<[1, 0]> : tensor<2xi64>}> : (tensor<2x3xf32>) -> tensor<3x2xf32>
  func.return %0 : tensor<3x2xf32>

}

// CHECK-LABEL: pack_multiple
func.func @pack_multiple(%arg0: tensor<2x3xf32>,
                    %arg1: tensor<2x3xf32>,
                    %arg2: tensor<2x3xf32>) -> tensor<3x2x3xf32> {
  %0 = "tf.MyPack"(%arg0, %arg1, %arg2) {N=3:i32, axis=0:i32} : (tensor<2x3xf32>, tensor<2x3xf32>, tensor<2x3xf32>) -> tensor<3x2x3xf32>
  func.return %0 : tensor<3x2x3xf32>

// CHECK-NEXT: %[[AXIS:.*]] = arith.constant 0 : i32

      %2 = "quantfork.stats"(%1) {layerStats = dense<[5.00000000e-6, 7.00000000e-1]> : tensor<2xf32>} : (tensor<1x3xf32>) -> tensor<1x3xf32>
      return %2 : tensor<1x3xf32>
    }

    func.func private @composite_dot_general_fn(%arg0: tensor<1x2xf32>, %arg1: tensor<2x3xf32>) -> tensor<1x3xf32> attributes {_from_xla_call_module} {

func.func @my_add_n_single_input(%arg0: tensor<2x3xf32>) -> tensor<2x3xf32> {
  %0 = "tf.MyAddN"(%arg0) {N=1:i32} : (tensor<2x3xf32>) -> tensor<2x3xf32>
  func.return %0 : tensor<2x3xf32>

// CHECK-NEXT: return %arg0 : tensor<2x3xf32>
}

// CHECK-LABEL: my_add_n_multiple_inputs
func.func @my_add_n_multiple_inputs(%arg0: tensor<2x3xf32>, %arg1: tensor<2x3xf32>, %arg2: tensor<2x3xf32>) -> tensor<2x3xf32> {

// CHECK:           %[[VAL_2:.*]] = "tf.Div"(%[[VAL_0]], %[[VAL_0]]) : (tensor<2xf32>, tensor<2xf32>) -> tensor<2xf32>
// CHECK:           %[[VAL_3:.*]] = "tf.FloorDiv"(%[[VAL_0]], %[[VAL_0]]) : (tensor<2xf32>, tensor<2xf32>) -> tensor<2xf32>
// CHECK:           return %[[VAL_3]] : tensor<2xf32>
// CHECK:         }
func.func @floordiv_f32(%arg0: tensor<2xf32>) -> tensor<2xf32> {

  // CHECK-DAG: return %[[RESULT]]
  func.return %0 : tensor<2x3xf32>
}

// CHECK-LABEL: dequantize_quint8
func.func @dequantize_quint8(%arg0: tensor<2x3x!tf_type.quint8>, %min_range: tensor<f32>, %max_range: tensor<f32>) -> tensor<2x3xf32> {
  // CHECK-NEXT: %[[C255:.*]] = "tf.Const"() <{value = dense<2.550000e+02> : tensor<f32>}>

          %outputs_5, %control_6 = tf_executor.island(%control_4) wraps "tf._XlaHostComputeMlir"() {host_mlir_module = "module {\0A func.func @host_func() -> tensor<1x2xf32> {\0A %0 = \22tf.Const\22() {value = dense<0.1> : tensor<1x2xf32>} : () -> tensor<1x2xf32> \0A return %0 : tensor<1x2xf32>}}", manual_sharding = true, recv_key = "host_compute_channel_1_retvals", send_key = "host_c...

    } : (tensor<2x2xf32>, tensor<2x2xf32>) -> tensor<2x2xf32>
    %2 = "quantfork.stats"(%1) {layerStats = dense<[0.000000e+00, 6.000000e+00]> : tensor<2xf32>} : (tensor<2x2xf32>) -> tensor<2x2xf32>
    return %2 : tensor<2x2xf32>
  }

  // CHECK-LABEL: composite_dot_general
  func.func private @composite_dot_general(%arg0: tensor<2x2xf32>, %arg1: tensor<2x2xf32>) -> tensor<2x2xf32> {

    %3 = "tfl.reshape"(%2, %cst_0) {tac.device = "GPU", tac.inference_type = "FLOAT"} : (tensor<1x1x1x2xf32>, tensor<1xi32>) -> tensor<2xf32>
    %4 = "tfl.reshape"(%3, %cst_1) {tac.device = "GPU", tac.inference_type = "FLOAT"} : (tensor<2xf32>, tensor<2xi32>) -> tensor<2x1xf32>
    return %4 : tensor<2x1xf32>
  }
```

#### Compute Costs Pass
In the compute cost pass, we will essentially compute the cost of each op within

func.func @foo(%arg0: tensor<1x4xf32>, %arg1: tensor<1x4xf32>) -> tensor<1x4xf32> {
  func.return %arg0 : tensor<1x4xf32>
}

func.func @bar(%arg0: tensor<2x4xf32>, %arg1: tensor<2x4xf32>) -> tensor<2x4xf32> {
  func.return %arg0 : tensor<2x4xf32>
}

func.func @foobar(%arg0: tensor<1x4xf32>, %arg1: tensor<1x4xf32>) -> tensor<1x4xf32> {
  func.return %arg0 : tensor<1x4xf32>
}

// CHECK-DAG: func @main