%outputs_9, %control_10 = tf_executor.island(%control_4) wraps "tf.Identity"(%outputs_7) {device = ""} : (tensor<2x3xi32>) -> tensor<2x3xi32>
      tf_executor.fetch %outputs_9 : tensor<2x3xi32>
    }
    return %0 : tensor<2x3xi32>
  }
}

// CHECK: module attributes
// CHECK-SAME:  tfl.metadata = {{{.*}}keep_stablehlo_constant = "true"{{.*}}}

func.func @main(%arg0: tensor<3x2xf32>) -> tensor<3x2xi32> {
  // CHECK-LABEL: @main
  // CHECK: "tfl.bucketize"(%arg0) <{boundaries = [0.000000e+00 : f32, 1.000000e+01 : f32, 1.000000e+02 : f32]}> : (tensor<3x2xf32>) -> tensor<3x2xi32>
  %0 = "tfl.bucketize"(%arg0) {boundaries = [0.0 : f32, 10.0 : f32, 100.0 : f32]} : (tensor<3x2xf32>) -> tensor<3x2xi32>
  func.return %0 : tensor<3x2xi32>

  // CHECK: %[[RES_INT_2:.*]] = mhlo.convert %[[RES_INT_1]] : tensor<3x2xi32>
  %result_int = "tf.Cast"(%result) {Truncate = false} : (tensor<3x2x!tf_type.qint32>) -> tensor<3x2xi32>
  // CHECK: return %[[RES_INT_2]] : tensor<3x2xi32>
  func.return %result_int : tensor<3x2xi32>

      // CHECK:          "mhlo.reduce_scatter"(%arg0)
      // CHECK{LITERAL}: replica_groups = dense<[[0, 4], [1, 5], [2, 6], [3, 7]]>
      // CHECK-SAME:     scatter_dimension = 0
      //
      %1 = "tf.XlaReduceScatter"(%arg0, %cst_0, %cst) {reduce_op = "Add"} : (tensor<128x128xf32>, tensor<4x2xi32>, tensor<i32>) -> tensor<64x128xf32>

// CHECK:           %[[VAL_3:.*]] = arith.constant dense<0> : tensor<4x2xi32>
// CHECK:           %[[VAL_4:.*]] = "tfl.pad"(%[[VAL_2]], %[[VAL_3]]) : (tensor<1x6x6x3xf32>, tensor<4x2xi32>) -> tensor<1x6x6x3xf32>

// fixed upstream).
func.func @broadcast_add(%arg0: tensor<1xi32>, %arg1: tensor<1x2xi32>) -> tensor<1x2xi32> {
  // CHECK-NEXT: %[[LHS_BCAST:.+]] = "mhlo.broadcast_in_dim"(%arg0) <{broadcast_dimensions = dense<1> : tensor<1xi64>}>
  // CHECK-NEXT: mhlo.add %[[LHS_BCAST]], %arg1
  %0 = "tf.AddV2"(%arg0, %arg1) : (tensor<1xi32>, tensor<1x2xi32>) -> tensor<1x2xi32>
  func.return %0: tensor<1x2xi32>
}

// CHECK-LABEL: func @broadcast_multi_dim_add

  func.return
}

func.func private @write_chain_func(%arg0: tensor<1xf32>) -> (tensor<1xf32>) {
  %cst = "tf.Const"() <{value = dense<[[0, 1]]> : tensor<1x2xi32>}> : () -> tensor<1x2xi32>
  %0 = "tf.XlaAllReduce"(%arg0, %cst) <{mode = "CrossReplica", reduce_op = "Add"}> : (tensor<1xf32>, tensor<1x2xi32>) -> tensor<1xf32>
  return %0 : tensor<1xf32>

  func.func private @quantize_gather_fn(%arg: tensor<3x4x2xf32>) -> tensor<2x3x2x2xf32> attributes {tf._original_func_name = "main_0"} {
    %cst = "tf.Const"() {value = dense<1> : tensor<2x3x2xi32>} : () -> tensor<2x3x2xi32>
    %0 = "quantfork.stats"(%arg) {layerStats = dense<[4.00000000e-6, 9.80000000e-1]> : tensor<2xf32>} : (tensor<3x4x2xf32>) -> tensor<3x4x2xf32>

  %w = arith.constant dense<127.0> : tensor<4x12xf32>
  %b = arith.constant dense<0.0> : tensor<4xf32>
  %fc = "tfl.fully_connected"(%arg0, %w, %b) {fused_activation_function = "NONE", keep_num_dims = false, weights_format = "DEFAULT"} : (tensor<1x224x224x3xf32>, tensor<4x12xf32>, tensor<4xf32>) -> tensor<1x112x112x4xf32>
  func.return %fc : tensor<1x112x112x4xf32>

"/job:tpu_host_worker/replica:0/task:3/device:CPU:0"}, %arg15: tensor<*x!tf_type.resource<tensor<6x32xf32>>> {tf._user_specified_name = "927", tf.device = "/job:tpu_host_worker/replica:0/task:0/device:CPU:0"}, %arg16: tensor<*x!tf_type.resource<tensor<6x32xf32>>> {tf._user_specified_name = "929", tf.device = "/job:tpu_host_worker/replica:0/task:1/device:CPU:0"}, %arg17: tensor<*x!tf_type.resource<tensor<6x32xf32>>> {tf._user_specified_name = "931", tf.device = "/job:tpu_host_worker/replica:0/task:2/device:CPU:0"},...