module attributes {tf.versions = {producer = 179 : i32}} {
  func.func @main(%arg0: tensor<128x8xf32> {mhlo.sharding = "bad_sharding"}) {
    func.return
  }
}

  %cst = arith.constant dense<1> : tensor<1xi32>
  %0 = "tfl.exp"(%arg0) : (tensor<8x128xf32>) -> tensor<8x128xf32>
  %1 = "tfl.sum"(%0, %cst) {keep_dims = true} : (tensor<8x128xf32>, tensor<1xi32>) -> tensor<8x1xf32>
  %2 = "tfl.div"(%0, %1) {fused_activation_function = "NONE"} : (tensor<8x128xf32>, tensor<8x1xf32>) -> tensor<8x128xf32>
  func.return %2 : tensor<8x128xf32>

  %4 = mhlo.reshape %3 : (tensor<1x8x8x1xf32>) -> tensor<8x8xf32>
  %5 = "mhlo.broadcast_in_dim"(%4) <{broadcast_dimensions = dense<[1, 2]> : tensor<2xi64>}> : (tensor<8x8xf32>) -> tensor<4x8x8x8xf32>
  %6 = mhlo.divide %2, %5 : tensor<4x8x8x8xf32>
  return %6 : tensor<4x8x8x8xf32>
}

// CHECK-LABEL:   func @convert_maxpool_valid(

}

// CHECK-LABEL: prepareNarrowStatistics
func.func @prepareNarrowStatistics(%arg0: tensor<8x4x3xf32>) -> tensor<8x4x3xf32> {
  %0 = "quantfork.stats"(%arg0) {
    layerStats = dense<[-1.0e-9, 1.0e-9]> : tensor<2xf32>
  } : (tensor<8x4x3xf32>) -> tensor<8x4x3xf32>
  func.return %0 : tensor<8x4x3xf32>

  func.return %0 : tensor<256x16x32x32xf32>

  // LAYOUT-LABEL: Conv2dNCHW
  // LAYOUT: "tfl.conv_2d"
}

func.func @fusedBatchNormV3(tensor<8x8x8x8xf32>, tensor<8xf32>, tensor<8xf32>, tensor<8xf32>, tensor<8xf32>) -> (tensor<8x8x8x8xf32>, tensor<8xf32>) {
^bb0(%arg0: tensor<8x8x8x8xf32>, %arg1: tensor<8xf32>, %arg2: tensor<8xf32>, %arg3: tensor<8xf32>, %arg4: tensor<8xf32>):
  // OK

}

func.func @sigmoid(%arg0: tensor<?x88xf32>) -> tensor<?x88xf32> {
  %0 = "tf.Sigmoid"(%arg0) : (tensor<?x88xf32>) -> tensor<?x88xf32>
  func.return %0 : tensor<?x88xf32>
// CHECK-LABEL: sigmoid
// CHECK:  "tfl.logistic"(%arg0) : (tensor<?x88xf32>) -> tensor<?x88xf32>
}

func.func @sqrt(%arg0: tensor<8x16xf32>) -> tensor<8x16xf32> {

  %2 = "tf.BatchToSpaceND"(%arg0, %0, %1) {device = ""} : (tensor<3x5x2xf32>, tensor<1xi32>, tensor<1x2xi32>) -> tensor<1x8x2xf32>

  // CHECK: return [[VAL8]] : tensor<1x8x2xf32>
  func.return %2 : tensor<1x8x2xf32>
}

func.func @fake_quant_with_min_max_args(%arg0 : tensor<?x?xf32>) -> tensor<?x?xf32> {

// CHECK: return %[[v4]] : tensor<?x36x32xf32>
}

func.func @einsum_with_runtime_shape2(%arg0 : tensor<?x?x8x64xf32>, %arg1 : tensor<8x8x64xf32>) -> tensor<?x?x8xf32> {
  %0 = "tf.Einsum"(%arg0, %arg1) {device = "", equation = "ABNH,DNH->ABD"} : (tensor<?x?x8x64xf32>, tensor<8x8x64xf32>) -> tensor<?x?x8xf32>
  func.return %0 : tensor<?x?x8xf32>
// CHECK-LABEL: einsum_with_runtime_shape2

  %arg0_shape = "tf.Const"() {value = dense<[1, 8, 4]> : tensor<3xi32>} : () -> tensor<3xi32>
  %arg0_reshaped = "tf.Reshape"(%arg0, %arg0_shape) : (tensor<8x4xf32>, tensor<3xi32>) -> tensor<1x8x4xf32>
  %zeroi2 = "tf.Const"() {value = dense<0> : tensor<2xi32>} : () -> tensor<2xi32>
  %axis = "tf.Const"() {value = dense<0> : tensor<i32>} : () -> tensor<i32>

  %arg0_shape = "tf.Const"() {value = dense<[1, 8, 4]> : tensor<3xi32>} : () -> tensor<3xi32>
  %arg0_reshaped = "tf.Reshape"(%arg0, %arg0_shape) : (tensor<8x4xf32>, tensor<3xi32>) -> tensor<1x8x4xf32>
  %zeroi2 = "tf.Const"() {value = dense<0> : tensor<2xi32>} : () -> tensor<2xi32>
  %axis = "tf.Const"() {value = dense<0> : tensor<i32>} : () -> tensor<i32>