%0:2 = tf_executor.island wraps "tf.Const"() {dtype = "tfdtype$DT_INT32", value = dense<0> : tensor<10xi32>} : () -> tensor<10xi32>
    %1:2 = tf_executor.island wraps "tf.Const"() {dtype = "tfdtype$DT_INT32", value = dense<0> : tensor<i32>} : () -> tensor<i32>
    %2:2 = tf_executor.island wraps "tf.PlaceholderWithDefault"(%1#0) {type = i32} : (tensor<i32>) -> tensor<*xi32> loc("unranked")

    %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>

    %3 = "tf.Const"() {value = dense<2> : tensor<1xi32>} : () -> tensor<1xi32>
    // expected-remark@+1 {{Cost: 1}}
    %4 = "tf.Const"() {value = dense<1> : tensor<1xi32>} : () -> tensor<1xi32>
    // expected-remark@+1 {{Cost: 1}}
    %5 = "tf.Const"() {value = dense<0> : tensor<1xi32>} : () -> tensor<1xi32>
    // expected-remark@+1 {{Cost: 1}}
    %6 = "tf.Const"() {value = dense<3> : tensor<1xi32>} : () -> tensor<1xi32>

// CHECK-LABEL: func @testBroadcastGradientArgs
func.func @testBroadcastGradientArgs(%s0: tensor<4xi32>, %s1: tensor<4xi32>) -> (tensor<1xi32>, tensor<0xi32>) {
  // CHECK:     tf.BroadcastGradientArgs
  %r0, %r1 = "tf.BroadcastGradientArgs"(%s0, %s1) : (tensor<4xi32>, tensor<4xi32>) -> (tensor<1xi32>, tensor<0xi32>)
  func.return %r0, %r1 : tensor<1xi32>, tensor<0xi32>
}

// -----

// CHECK-LABEL: @acos

  attributes {tf_saved_model.exported_names = ["assign_from_global"]} {
    // CHECK: [[name:%.*]] = "tf.VarHandleOp"
    // CHECK: "tf.AssignVariableOp"([[name]]
    %0 = "tf.Const"() {type = tensor<10xf32>, value = dense<[0.,10.,2.,3.,4.,5.,6.,7.,8.,9.]> : tensor<10xf32>} : () -> tensor<10xf32>
    "tf.AssignVariableOp"(%arg0, %0) : (tensor<!tf_type.resource<tensor<10xf32>>>, tensor<10xf32>) -> ()

// -----

func.func @testAddReluPack(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) {
   // CHECK: tac.device = "GPU", tac.inference_type = "FLOAT"
  %0 = "tfl.add"(%arg0, %arg1) {fused_activation_function = "RELU6"} : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>
   // CHECK: tac.device = "GPU", tac.inference_type = "FLOAT"
  %1 = "tfl.add"(%arg0, %0) {fused_activation_function = "RELU"} : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>

    >} : (
      tensor<?x?x6xi8>, tensor<6x8xi8>
    ) -> tensor<?x?x8xi32>
  %1 = chlo.broadcast_add %0, %zp_offset : (
      tensor<?x?x8xi32>, tensor<8xi32>) -> tensor<?x?x8xi32>
  %2 = chlo.broadcast_add %1, %bias : (
      tensor<?x?x8xi32>, tensor<8xi32>) -> tensor<?x?x8xi32>
  return %2 : tensor<?x?x8xi32>
}

// -----

// CHECK-LABEL: func @dot_general_add_add_static

    %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>

    %3 = "tf.Const"() {value = dense<[[0, 1]]> : tensor<1x2xi32>} : () -> tensor<1x2xi32>
    %4 = "tf.Const"() {value = dense<> : tensor<0xi32>} : () -> tensor<0xi32>
    %5 = "tf.Const"() {value = dense<2.500000e-01> : tensor<f32>} : () -> tensor<f32>
    %6 = "tf.Const"() {value = dense<0> : tensor<1xi32>} : () -> tensor<1xi32>
    %7 = "tf.Const"() {value = dense<[-1, 1001]> : tensor<2xi32>} : () -> tensor<2xi32>

  %0 = "tfr.constant_tensor"(%arg0) : (vector<1xi32>) -> tensor<1xf32>
  func.return %0 : tensor<1xf32>
}

// -----

func.func @constant_tensor_invalid_3(%arg0: vector<1xi32>) -> tensor<1x1xi32> {
    // expected-error@+1 {{input and output should have same shape and element type}}
  %0 = "tfr.constant_tensor"(%arg0) : (vector<1xi32>) -> tensor<1x1xi32>
  func.return %0 : tensor<1x1xi32>
}

// -----