"QuantizeAndDequantizeV2",
                                         "QuantizeAndDequantizeV3",
                                         "QuantizeAndDequantizeV4",
                                         "QuantizeAndDequantizeV4Grad",
                                         "QuantizedConcat",
                                         "QuantizedInstanceNorm",

  grad_outputs->push_back(qdq_v4_grad.input_min_backprop);
  grad_outputs->push_back(qdq_v4_grad.input_max_backprop);
  return scope.status();
}
REGISTER_GRADIENT_OP("QuantizeAndDequantizeV4",
                     QuantizeAndDequantizeV4GradHelper);

Status QuantizeAndDequantizeV3Grad(const Scope& scope, const Operation& op,
                                   const std::vector<Output>& grad_inputs,

      "ParameterizedTruncatedNormal",
      "PartitionedCall",
      "Polygamma",
      "PopulationCount",
      "Qr",
      "QuantizeAndDequantizeV2",
      "QuantizeAndDequantizeV3",
      "QuantizeAndDequantizeV4",
      "RFFT",
      "RFFT2D",
      "RFFT3D",
      "RGBToHSV",
      "RandomShuffle",
      "RandomStandardNormal",
      "RandomUniform",
      "RandomUniformInt",

  func.return %0 : tensor<?x?xf32>

}

func.func @quantize_dequantize_v4(%arg0 : tensor<?x?xf32>) -> tensor<?x?xf32> {
  %cst = arith.constant dense<0.0> : tensor<f32>
  %cst_0 = arith.constant dense<255.0> : tensor<f32>
  %0 = "tf.QuantizeAndDequantizeV4"(%arg0, %cst, %cst_0) : (tensor<?x?xf32>, tensor<f32>, tensor<f32>) -> tensor<?x?xf32>
  func.return %0 : tensor<?x?xf32>

// CHECK-LABEL: quantize_dequantize_v4

  );

  let results = (outs
    TF_FloatTensor:$output
  );

  TF_DerivedOperandTypeAttr T = TF_DerivedOperandTypeAttr<0>;
}

def TF_QuantizeAndDequantizeV4Op : TF_Op<"QuantizeAndDequantizeV4", [Pure]> {
  let summary = "Quantizes then dequantizes a tensor.";

  let description = [{
This is almost identical to QuantizeAndDequantizeV2, except that it returns a