} : (tensor<1x224x224x3xf32>) -> tensor<1x224x224x3xf32>
  %6 = "tfl.conv_2d"(%5, %4, %cst) {
      dilation_h_factor = 1 : i32, dilation_w_factor = 1 : i32, fused_activation_function = "NONE", padding = "SAME", stride_h = 2 : i32, stride_w = 2 : i32
    } : (tensor<1x224x224x3xf32>, tensor<32x3x3x3xf32>, tensor<32xf32>) -> tensor<1x112x112x32xf32>

  }
  func.func private @quantize_i8(%arg0: tensor<1x224x224x3xf32>, %arg1: tensor<f32>, %arg2: tensor<i32>) -> tensor<1x224x224x3xi8> {
    %0 = "tf.Div"(%arg0, %arg1) : (tensor<1x224x224x3xf32>, tensor<f32>) -> tensor<1x224x224x3xf32>
    %1 = "tf.Round"(%0) : (tensor<1x224x224x3xf32>) -> tensor<1x224x224x3xf32>
    %2 = "tf.Cast"(%1) : (tensor<1x224x224x3xf32>) -> tensor<1x224x224x3xi32>

  %outputs_22 = "tf.Squeeze"(%outputs_20) {device = "", squeeze_dims = [0]} : (tensor<1x224x224x3xf32>) -> tensor<224x224x3xf32>
  %outputs_24 = "tf.Cast"(%outputs_22) {Truncate = false, device = ""} : (tensor<224x224x3xf32>) -> tensor<224x224x3xui8>

  // CHECK-LABEL: RedundantShapeOp
  // CHECK-NOT: "tf.Shape"
}

// CHECK-LABEL: @MoveTransposeAcrossPerChannelQuant
func.func @MoveTransposeAcrossPerChannelQuant(%arg0 : tensor<1x224x224x3xf32>) -> tensor<1x112x112x6xf32> {
  %cst = "tf.Const"() <{value = dense<6.0> : tensor<6x3x7x7xf32>}> : () -> tensor<6x3x7x7xf32>
  %cst_14 = "tf.Const"() <{value = dense<[2, 3, 1, 0]> : tensor<4xi64>}> : () -> tensor<4xi64>

```mlir
module {
  func @while_body {
    %input = "tf.IteratorGetNext"(...) {device = "/CPU:0"}: -> tensor<2x224x224x3xf32>
    %device_launch = "tf_device.cluster_func"(%input,...) {func = @_func,...)
    return ...
  }
  func @_func(%input: tensor<2x224x224x3xf32>, %filter: tensor<7x7x3x64xf32>) {