func.func @fusedConv2dRelu(%arg0: tensor<256x32x32x3xf32>, %arg1: tensor<16x3x3x3xf32>, %arg2: tensor<16xf32>) -> tensor<256x32x32x16xf32> {
  %0 = "tfl.conv_2d"(%arg0, %arg1, %arg2) {dilation_h_factor = 1 : i32, dilation_w_factor = 1 : i32, fused_activation_function = "NONE", padding = "SAME", stride_h = 1 : i32, stride_w = 1 : i32} : (tensor<256x32x32x3xf32>, tensor<16x3x3x3xf32>, tensor<16xf32>) -> tensor<256x32x32x16xf32>

// CHECK:  %0 = "tf.Transpose"(%arg1, %[[CONSTANT0]]) : (tensor<3x3x3x16xf32>, tensor<4xi32>) -> tensor<16x3x3x3xf32>
// CHECK:  %1 = "tfl.conv_2d"(%arg0, %0, %[[CONSTANT]]) <{dilation_h_factor = 2 : i32, dilation_w_factor = 3 : i32, fused_activation_function = "NONE", padding = "SAME", stride_h = 4 : i32, stride_w = 5 : i32}> : (tensor<256x32x32x3xf32>, tensor<16x3x3x3xf32>, tensor<16xf32>) -> tensor<256x8x7x16xf32>
// CHECK:  %2 = "tf.Conv2D"

  func.return %0 : tensor<256x32x32x16xf32>
}

// -----

func.func @testConv2D4DBias(tensor<256x32x32x3xf32>, tensor<16x3x3x3xf32>, tensor<1x1x1x16xf32>) -> tensor<256x32x32x16xf32> {
^bb0(%arg0: tensor<256x32x32x3xf32>, %arg1: tensor<16x3x3x3xf32>, %arg2: tensor<1x1x1x16xf32>):

// CHECK:           %[[VAL_8:.*]] = "tf.Transpose"(%[[VAL_6]], %[[VAL_7]]) : (tensor<1x8x8x16xf32>, tensor<4xi64>) -> tensor<16x8x8x1xf32>
// CHECK:           return %[[VAL_8]] : tensor<16x8x8x1xf32>
// CHECK:         }
func.func @convert_conv2d_with_transpose(%arg0: tensor<8x8x1x207xf32>, %arg1: tensor<3x3x16x207xf32>) -> tensor<16x8x8x1xf32> {
  %0 = "mhlo.convolution"(%arg0, %arg1) {batch_group_count = 1 : i64,
    dimension_numbers = #mhlo.conv<raw