func.func @random_uniform_int(%arg0: tensor<i32>, %arg1: tensor<i32>) -> tensor<1000xi32> {
    %0 = "tf.Const"() {value = dense<1000> : tensor<1xi32>} : () -> tensor<1xi32>
    // CHECK-NOT: tf.RandomUniformInt
    %1 = "tf.RandomUniformInt"(%0, %arg0, %arg1) {seed = 0 : i64, seed2 = 0 : i64} : (tensor<1xi32>, tensor<i32>, tensor<i32>) -> tensor<1000xi32>
    func.return %1 : tensor<1000xi32>
  }

  // CHECK-LABEL: multinomial

      /*error_tolerance=*/1.0);
}

TEST_F(ConvertTfQuantToMhloIntTest, UniformQuantizeAddToValidGraph) {
  constexpr absl::string_view kProgram = R"mlir(
func.func @main(%lhs: tensor<10x10xi32>, %rhs: tensor<10x10xi32>) -> tensor<10x10xi32> {
  %lhs_scale = "tf.Const"() { value = dense<0.518> : tensor<f32> } : () -> tensor<f32>
  %lhs_zp = "tf.Const"() { value = dense<42> : tensor<i32> } : () -> tensor<i32>

// CHECK:           [[VAL_5:%.*]] = call @func_1_GPU_FLOAT([[VAL_0]], [[VAL_3]]) {tac.device = "GPU", tac.inference_type = "FLOAT", tac.interface_name = "func_1"} : (tensor<100xf32>, tensor<100xf32>) -> tensor<100xf32>

  %3 = "tf.PreventGradient"(%1) : (tensor<10x10xi32>) -> tensor<10x10xi32>
  %4 = "tf.AddV2"(%2, %3) {device = ""} : (tensor<10x10xi32>, tensor<10x10xi32>) -> tensor<10x10xi32>
  func.return %4 : tensor<10x10xi32>

// CHECK-LABEL: preventGradient
// CHECK-NOT: %2 = "tf.PreventGradient"(%0) : (tensor<10x10xi32>) -> tensor<10x10xi32>

      %2 = "tf.ReadVariableOp"(%arg15) : (tensor<*x!tf_type.resource<tensor<7x7x3x64xf32>>>) -> tensor<7x7x3x64xf32>
      %3 = "tf.ReadVariableOp"(%arg16) : (tensor<*x!tf_type.resource<tensor<1001xf32>>>) -> tensor<1001xf32>
      %4 = "tf.ReadVariableOp"(%arg17) : (tensor<*x!tf_type.resource<tensor<64x1001xf32>>>) -> tensor<64x1001xf32>
      %5 = "tf.ReadVariableOp"(%arg18) : (tensor<*x!tf_type.resource<tensor<f32>>>) -> tensor<f32>

// CHECK: return %[[GATHER]] : tensor<1x300x10xf32>

// -----

module {
  func.func @propagate_xlagather(%arg0: tensor<10x2xi32>) -> tensor<1x300x10xf32> {
    %cst = "tf.Const"() {value = dense<[1, 1, 300]> : tensor<3xi64>} : () -> tensor<3xi64>

  %0 = "mhlo.broadcast_in_dim"(%arg1) <{broadcast_dimensions = dense<[0, 1]> : tensor<2xi64>}> : (tensor<1x1xi32>) -> tensor<1x100xi32>
  %1 = "mhlo.select"(%arg0, %0, %arg2) : (tensor<i1>, tensor<1x100xi32>, tensor<1x100xi32>) -> tensor<1x100xi32>
  func.return %1 : tensor<1x100xi32>
}

// CHECK-LABEL:   func @selectv2_broadcasted_condition(

  // CHECK-NEXT: %[[UPDATE:.*]] = "tf.XlaDynamicUpdateSlice"(%[[STACK_VAL]], %[[UPDATE_SLICE]], %[[CONCAT_OFFETS]]) : (tensor<10x2xi32>, tensor<1x2xi32>, tensor<2xi32>) -> tensor<10x2xi32>
  // CHECK-NEXT: "tf.AssignVariableOp"(%[[BUFFER]], %[[UPDATE]]) : (tensor<!tf_type.resource<tensor<10x2xi32>>>, tensor<10x2xi32>) -> ()
  // CHECK-NEXT: %[[CONST1:.*]] = "tf.Const"() <{value = dense<1> : tensor<1xi32>}> : () -> tensor<1xi32>

    // CHECK:     return %[[RESULT]]
    func.return %0 : tensor<10xi32>
  }

  // CHECK-LABEL: func @foo
  // CHECK-SAME:    %[[ARG0:.*]]: tensor<10xi32>, %[[ARG1:.*]]: tensor<10xi32>
  func.func @foo(%arg0: tensor<10xi32>, %arg1: tensor<10xi32>) -> tensor<10xi32> {

// results, and function body should be the same.

// CHECK-LABEL: module
module {
  // CHECK-LABEL: func @main
  // CHECK-SAME:    %[[ARG0:.*]]: tensor<10xi32>, %[[ARG1:.*]]: tensor<10xi32>
  func.func @main(%arg0: tensor<10xi32>, %arg1: tensor<10xi32>) -> tensor<10xi32> {
    // CHECK:      %[[RESULT:.*]] = "tf.XlaCallModule"(%[[ARG0]], %[[ARG1]])
    // CHECK-SAME:   Sout = [#tf_type.shape<?>]
    // CHECK-NOT:    function_list