%1 = "tfl.sum"(%0, %cst) {keep_dims = false} : (tensor<2xf32>, tensor<1xi32>) -> tensor<f32>
  %2 = "tfl.add"(%1, %cst_1) {fused_activation_function = "NONE"} : (tensor<f32>, tensor<1xf32>) -> tensor<1xf32>
  %3 = "tfl.rsqrt"(%2) : (tensor<1xf32>) -> tensor<1xf32>
  %4 = "tfl.mul"(%arg0, %3) {fused_activation_function = "NONE"} : (tensor<2xf32>, tensor<1xf32>) -> tensor<2xf32>
  func.return %4: tensor<2xf32>

    %4 = "tf.LayerNorm"(%a1, %a2, %a3, %a4) {_tfl_quant_trait = "fully_quantizable", device = ""} : (tensor<128x128xf32>, tensor<1xf32>, tensor<1xf32>, tensor<1xi32>) -> tensor<128x128xf32>
   "tfl.yield"(%4) : (tensor<128x128xf32>) -> ()
  }) {_tfl_quant_trait = "fully_quantizable", device = ""} : (tensor<128x128xf32>, tensor<1xf32>, tensor<1xf32>, tensor<1xi32>) -> tensor<128x128xf32>

func.func @add(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) -> tensor<1xf32> {
  %0 = "tf.Add"(%arg0, %arg1) : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>
  func.return %0: tensor<1xf32>
// CHECK: %[[CUSTOM_0:.*]] = "tfl.custom"(%arg0, %arg1) <{custom_code = "FlexAdd", custom_option = #tfl<const_bytes : "0x03416464001412034164641A001A002A070A015412023001320000021B171414042801">}> : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>

module {
  // CHECK-LABEL: main
  func.func @main(%arg0: tensor<4xf32>, %arg1: tensor<4xf32>) -> tensor<4xf32> {
    %0 = "tfl.squared_difference"(%arg0, %arg1) : (tensor<4xf32>, tensor<4xf32>) -> tensor<4xf32>
    func.return %0 : tensor<4xf32>
    // CHECK:  [[VAL_0:%.*]] = tfl.sub %arg0, %arg1 {fused_activation_function = "NONE"} : tensor<4xf32>

}

// CHECK-LABEL: add
func.func @add(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) -> tensor<1xf32> {
  %0 = "tf.AddV2"(%arg0, %arg1) : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>
  func.return %0: tensor<1xf32>
// CHECK: %[[ADD_0:.*]] = "tf.AddV2"(%arg0, %arg1) : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>
// CHECK: return %[[ADD_0]] : tensor<1xf32>
}

// CHECK-LABEL: softmax

module {
  func.func @simpleTest(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>, %arg2: tensor<1xf32>, %arg3: tensor<1xf32>) -> tensor<2x1xf32> {
    %0 = func.call @func_0_GPU_FLOAT(%arg0, %arg1, %arg2) {tac.interface_name = "func_0"} : (tensor<1xf32>, tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>
    %1 = func.call @func_1_GPU_FLOAT(%arg0, %arg3) {tac.interface_name = "func_1"} : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>

    return
  }
  // CHECK-LABEL: func private @foo.impl
  func.func private @foo.impl(%arg0: tensor<1xf32>, %arg1: tensor<2xf32>) -> (tensor<2xf32>, tensor<1xf32>) {
    return %arg1, %arg0 : tensor<2xf32>, tensor<1xf32>
  }

    %4 = "tf.LayerNorm"(%a1, %a2, %a3, %a4) {_tfl_quant_trait = "fully_quantizable", device = ""} : (tensor<128x128xf32>, tensor<1xf32>, tensor<1xf32>, tensor<1xi32>) -> tensor<128x128xf32>
   "tfl.yield"(%4) : (tensor<128x128xf32>) -> ()
  }) {_tfl_quant_trait = "fully_quantizable", device = ""} : (tensor<128x128xf32>, tensor<1xf32>, tensor<1xf32>, tensor<1xi32>) -> tensor<128x128xf32>

func.func @multiply(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) -> tensor<1xf32> {
  %0 = "vhlo.multiply_v1"(%arg0, %arg1) : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>
  func.return %0 : tensor<1xf32>
}

// CHECK:func.func private @multiply(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) -> tensor<1xf32> {
// CHECK: %0 = "vhlo.multiply_v1"(%arg0, %arg1) : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>
// CHECK: return %0 : tensor<1xf32>
// CHECK:}

  func.return %0 : tensor<*xf32>
}

// CHECK-NOT: "tfl.reshape"
// CHECK: "tfl.pack"

// -----

func.func @squaredDifference(%arg0: tensor<4xf32>, %arg1: tensor<4xf32>) -> tensor<4xf32> {
  %0 = "tfl.squared_difference"(%arg0, %arg1) : (tensor<4xf32>, tensor<4xf32>) -> tensor<4xf32>
  func.return %0 : tensor<4xf32>
}