return "Apply quantization on models in TensorFlow dialect";
  }

  // Determine if the unused Q-DQ pairs need to be removed. For weight-only
  // quantizable ops, Q-DQ ops need to be preserved.
  bool shouldKeepUnusedQdqPattern();

  void runOnOperation() override;

 private:
  QuantizationSpecs quant_specs_;

  Option<bool> weight_quantization_{
      *this, "weight-quantization", llvm::cl::init(false),