padding = "EXPLICIT",
         strides = [5, 6, 7, 8]
       } : (tensor<1x32x32x3xf32>, tensor<4xi32>, tensor<1x32x32x8xf32>)
         -> tensor<1x1x3x8xf32>

  func.return %0 : tensor<1x1x3x8xf32>
}

// CHECK-LABEL: func @transposeConv2DBackpropInput
func.func @transposeConv2DBackpropInput(
  %input_sizes: tensor<4xi32>,
  %filter: tensor<1x1x3x8xf32>,
  %out_backprop: tensor<1x32x32x8xf32>

         padding = "VALID",
         strides = [1, 1, 1, 1]
       } : (tensor<1x28x28x64xf16>, tensor<4xi32>, tensor<1x28x28x64xf16>)
        -> tensor<1x1x64x64xf16>

  func.return %0 : tensor<1x1x64x64xf16>
}

// CHECK-LABEL: func @transposeConv2DBackpropInput_f16
func.func @transposeConv2DBackpropInput_f16(
  %input_size:   tensor<4xi32>,
  %filter:       tensor<1x28x28x64xf16>,
  %out_backprop: tensor<1x28x28x64xf16>

    tf_executor.fetch
  }
  func.return

  %2 = "tf.Relu6"(%1) {device = ""} : (tensor<*xf32>) -> tensor<*xf32>
  func.return %2 : tensor<*xf32>
}

// CHECK-LABEL: func @conv_with_dynamic_channel_dim
// CHECK-DAG: %[[CONST_0:.*]] = "tf.Const"() <{value = dense<0.000000e+00> : tensor<1xf32>}> : () -> tensor<1xf32>
// CHECK-DAG: %[[CONST_1:.*]] = "tf.Const"() {{.*}} : () -> tensor<2x3x3x1xf32>

  // 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>

  %0:2 = "tf.NonMaxSuppressionV4"(%arg0, %arg1, %max_size, %arg2, %arg3) {pad_to_max_output_size = true}: (tensor<3x4xf32>, tensor<3xf32>, tensor<i32>, tensor<f32>, tensor<f32>) -> (tensor<2xi32>, tensor<i32>)
  func.return %0#0 : tensor<2xi32>
}

// CHECK-LABEL: mirror_pad
func.func @mirror_pad(%arg0: tensor<2x3xcomplex<f64>>) -> tensor<4x7xcomplex<f64>> {
  %0 = mhlo.constant dense<[[1, 1], [2, 2]]> : tensor<2x2xi32>

  %cst_2 = arith.constant dense<[4, 1024, 4096]> : tensor<3xi32>
  %0 = "tfl.fully_connected"(%arg0, %cst_0, %cst_1) {fused_activation_function = "NONE", keep_num_dims = false, weights_format = "DEFAULT"} : (tensor<4x1024x1024xbf16>, tensor<4096x1024xbf16>, none) -> tensor<4096x4096xbf16>
  %1 = "tfl.reshape"(%0, %cst_2) : (tensor<4096x4096xbf16>, tensor<3xi32>) -> tensor<4x1024x4096xbf16>
  func.return %1 : tensor<4x1024x4096xbf16>

    "tfl.yield"(%1#0, %1#1) : (tensor<*xi32>, tensor<*xf32>) -> ()
  }) {is_stateless = false} : (tensor<i32>, tensor<1xf32>) -> (tensor<i32>, tensor<1xf32>)
  func.return %0#1 : tensor<1xf32>
}

func.func @cond(%arg0: tensor<*xi32>, %arg1: tensor<*xf32>) -> tensor<i1> {
  %cst = arith.constant dense<0> : tensor<i32> loc("Const")
  %0 = "tfl.greater"(%arg0, %cst) : (tensor<*xi32>, tensor<i32>) -> tensor<i1>
  func.return %0 : tensor<i1>
}

// CHECK-LABEL: main
func.func @main(%arg0: tensor<3x15x14x3xf32>) -> tensor<3x15x14x8xf32>
    attributes {tf.entry_function = {inputs = "input", outputs = "Conv2D"}} {
  %cst = arith.constant dense<[0, 1, 2, 3]> : tensor<4xi32>
  %cst_0 = arith.constant dense<1.000000e+00> : tensor<f32>
  %cst_1 = arith.constant dense<0.000000e+00> : tensor<8xf32>
  %cst_2 = arith.constant dense<0.000000e+00> : tensor<8x3x3x3xf32>

    "tfl.yield"(%1) : (tensor<i1>) -> ()
  },  {
  ^bb0(%arg2: tensor<*xi32>, %arg3: tensor<*xf32>):
    %1:2 = func.call @WhileOp_body(%arg2, %arg3) : (tensor<*xi32>, tensor<*xf32>) -> (tensor<*xi32>, tensor<*xf32>)
    "tfl.yield"(%1#0, %1#1) : (tensor<*xi32>, tensor<*xf32>) -> ()
  }) : (tensor<i32>, tensor<1xf32>) -> (tensor<i32>, tensor<1xf32>) loc("WhileOp")
  func.return %0#1 : tensor<1xf32>