%c = stablehlo.constant dense<40> : tensor<2xi32>
      %0 = "tf.UnconvertedOp"(%arg0, %c) {align_corners = false, half_pixel_centers = false} : (tensor<1x20x20x28xf32>, tensor<2xi32>) -> tensor<1x40x40x28xf32>
      func.return %0 : tensor<1x40x40x28xf32>
  }

  %perm = arith.constant dense<[3, 0, 1, 2]> : tensor<4xi32>
  %0 = "tfl.transpose"(%arg0, %perm) : (tensor<2x3x4x5xf32>, tensor<4xi32>) -> tensor<5x2x3x4xf32>
  %perm1 = arith.constant dense<[3, 0, 1, 2]> : tensor<4xi32>
  %1 = "tfl.transpose"(%arg1, %perm1) : (tensor<2x3x4x5xf32>, tensor<4xi32>) -> tensor<5x2x3x4xf32>
  %2 = tfl.add %0, %1 { fused_activation_function = "NONE" } : tensor<5x2x3x4xf32>

  // CHECK-DAG: %[[RHS_1:.*]] = mhlo.convert %arg1 : tensor<2xi32>
  // CHECK-DAG: %[[RHS_2:.*]] = mhlo.bitcast_convert %[[RHS_1]] : (tensor<2xi32>) -> tensor<2x!quant.uniform<i32:f32, 2.000000e+00:4>>
  %input1_qint = "tf.Cast"(%input1) {Truncate = false} : (tensor<3x2xi32>) -> tensor<3x2x!tf_type.qint32>
  %input2_qint = "tf.Cast"(%input2) {Truncate = false} : (tensor<2xi32>) -> tensor<2x!tf_type.qint32>

  %1 = "quantfork.stats"(%arg0) {layerStats = dense<[-1.0, 1.0]> : tensor<2xf32>} : (tensor<1x1x3xf32>) -> tensor<1x1x3xf32>
  %2 = "tfl.pseudo_const"() {value = dense<[-1, 3]> : tensor<2xi32>} : () -> tensor<2xi32>
  %3 = "tfl.reshape"(%1, %2) : (tensor<1x1x3xf32>, tensor<2xi32>) -> tensor<1x3xf32>
  %4 = "quantfork.stats"(%3) {layerStats = dense<[-1.0, 1.0]> : tensor<2xf32>} : (tensor<1x3xf32>) -> tensor<1x3xf32>
  func.return %4 : tensor<1x3xf32>

    %cst = "tf.Const"() {value = dense<-1> : tensor<i64>} : () -> tensor<i64>
    %cst_0 = "tf.Const"() {value = dense<[1, 0]> : tensor<2xi32>} : () -> tensor<2xi32>
    %cst_1 = "tf.Const"() {value = dense<1.000000e+00> : tensor<f32>} : () -> tensor<f32>
    %cst_2 = "tf.Const"() {value = dense<0.00235294132> : tensor<f32>} : () -> tensor<f32>

  %multiples = "tf.Const"() { device = "/device:CPU:0", value = dense<[7,3]> : tensor<2xi32> } : () -> tensor<2xi32>
  // CHECK: T = !corert.string
  %output = "tf.Tile"(%arg, %multiples) { device = "/device:CPU:0" } : (tensor<1x2x!tf_type.string>, tensor<2xi32>) -> tensor<?x6x!tf_type.string>
  func.return %output : tensor<?x6x!tf_type.string>
}

// CHECK-LABEL: func @shape
func.func @shape() {

// CHECK-DAG:       %[[VAL_3:.*]] = "tfl.pseudo_const"(){{.*}}dense<2> : tensor<1xi32>
// CHECK-DAG:       %[[VAL_4:.*]] = "tfl.pseudo_const"(){{.*}}dense<[2, 1]> : tensor<2xi32>
// CHECK:           %[[VAL_5:.*]] = "tfl.reshape"(%[[VAL_0]], %[[VAL_2]]) {tac.device = "GPU", tac.inference_type = "FLOAT"} : (tensor<1xf32>, tensor<4xi32>) -> tensor<1x1x1x1xf32>

# CHECK-DAG:       %[[VAL_2:.*]] = arith.constant dense<[1, 0]> : tensor<2xi32>
# CHECK-DAG:       %[[VAL_3:.*]] = "tfl.no_value"() <{value}> : () -> none
# CHECK-DAG:       %[[VAL_6:.*]] = arith.constant dense<0> : tensor<i32>

// QDQ-NEXT:  return %[[split]]#0 : tensor<2xf32>
}

// CHECK-LABEL: FoldTranspose
func.func @FoldTranspose(%arg0: tensor<1x10x20x3xf32>) -> tensor<1x20x40x16xf32> {
  %cst = arith.constant dense<[1, 20, 40, 16]> : tensor<4xi32>
  %cst_0 = arith.constant dense<[2, 0, 1, 3]> : tensor<4xi32>

// CHECK-DAG: %[[START:.+]] = arith.constant dense<{{\[1, 2\]}}> : tensor<2xi32>
// CHECK-DAG: %[[SIZE:.+]] = arith.constant dense<2> : tensor<2xi32>
// CHECK: %[[SLICE:.+]] = "tfl.slice"(%[[ARG0]], %[[START]], %[[SIZE]]) : (tensor<3x4x!quant.uniform<i8:f32, 2.000000e+00:-1>>, tensor<2xi32>, tensor<2xi32>) -> tensor<2x2x!quant.uniform<i8:f32, 2.000000e+00:-1>>
// CHECK: return %[[SLICE]]

// -----