return op_name == "RandomUniform" || op_name == "RandomShuffle" ||
           op_name == "RandomUniformInt" || op_name == "RandomStandardNormal" ||
           op_name == "TruncatedNormal" || op_name == "Multinomial";
  }

  bool OpProducesOrConsumesVariant(const Node& node) const {
    auto is_variant = [](DataType dtype) { return dtype == DT_VARIANT; };

    // return %[[RESULT]]
    func.return %0 : tensor<*xf32>
  }

  func.func @truncated_normal() -> tensor<2x2xf32> {
    // CHECK-NOT: tf.TruncatedNormal
    %0 = mhlo.constant dense<[2, 2]> : tensor<2xi32>
    %1 = "tf.TruncatedNormal"(%0) {T = i32, device = "", dtype = f32, seed = 0 : i64, seed2 = 1950157571 : i64} : (tensor<2xi32>) -> tensor<2x2xf32>
    func.return %1 : tensor<2x2xf32>
  }

      "TensorScatterMax",
      "TensorScatterMin",
      "TensorScatterSub",
      "TensorScatterUpdate",
      "ToBool",
      "TridiagonalSolve",
      "TridiagonalMatMul",
      "TruncatedNormal",
      "UniformDequantize",
      "UniformQuantize",
      "UniformQuantizedAdd",
      "UniformQuantizedClipByValue",
      "UniformQuantizedConvolution",
      "UniformQuantizedDot",

  );

  let results = (outs
    TF_FpOrI32OrI64Tensor:$z
  );

  TF_DerivedOperandTypeAttr T = TF_DerivedOperandTypeAttr<0>;
}

def TF_TruncatedNormalOp : TF_Op<"TruncatedNormal", [TF_CannotDuplicate]> {
  let summary = "Outputs random values from a truncated normal distribution.";

  let description = [{
The generated values follow a normal distribution with mean 0 and standard

*   Add C, C++, and Python functions for querying kernels.

## Breaking Changes

*   Keras:
    *   The default values for tf.keras `RandomUniform`, `RandomNormal`, and
        `TruncatedNormal` initializers have been changed to match those in
        external Keras.
    *   Breaking change: `model.get_config()` on a Sequential model now returns