mlir::OpPassManager &pm, const QuantizationOptions &quantization_options,
    std::optional<const absl::string_view> mlir_dump_file_prefix =
        std::nullopt);

void AddQuantizePtqPreCalibrationPasses(
    mlir::OpPassManager &pm, const QuantizationOptions &quantization_options);

void AddQuantizePtqPostCalibrationPasses(
    mlir::OpPassManager &pm, const QuantizationOptions &quantization_options,

    const std::unordered_set<std::string>& tags,
    const QuantizationOptions& quantization_options);

// Applies post-training dynamic-range quantization to the model.
absl::StatusOr<ExportedModel> QuantizeDynamicRangePtq(
    absl::string_view saved_model_path,
    const std::vector<std::string>& signature_keys,
    const std::unordered_set<std::string>& tags,
    const QuantizationOptions& quantization_options);

  spec->set_quantization_component(component);
  spec->set_bit_type(bit_type);
  spec->set_bit_width(bit_width);
}

::stablehlo::quantization::QuantizationOptions FillPresetQuantizationOptions(
    ::stablehlo::quantization::QuantizationOptions quantization_options_) {
  CustomQuantizationMethod custom_method =
      quantization_options_.quantization_method().custom_quantization_method();

namespace stablehlo {
namespace quantization {

void AddQuantizationPasses(mlir::PassManager& pass_manager,
                           const QuantizationOptions& quantization_options) {
  QuantizationOptions quantization_options_ = quantization_options;
  if (quantization_options.quantization_method()
          .has_preset_quantization_method()) {
    quantization_options_ =

    const QuantizationComponentSpec expected_bias_component) {
  QuantizationOptions quantization_options;
  quantization_options.mutable_quantization_method()
      ->mutable_preset_quantization_method()
      ->set_preset_method(preset_quantization_options);
  QuantizationOptions filled_quantization_options =
      quant::stablehlo::FillPresetQuantizationOptions(quantization_options);

    : public internal::SerializedProtobufCaster<
          tensorflow::quantization::ExportedModel> {};

template <>
struct type_caster<tensorflow::quantization::QuantizationOptions>
    : public internal::SerializedProtobufCaster<
          tensorflow::quantization::QuantizationOptions> {};

template <>
struct type_caster<::stablehlo::quantization::CalibrationOptions>
    : public internal::SerializedProtobufCaster<

// bf16 is quantizable.
bool IsValueWithQuantizablePrecision(Value val);

std::optional<tensorflow::quantization::QuantizationComponentSpec>
GetWeightComponentSpec(
    const tensorflow::quantization::QuantizationOptions& quantization_options);

// Returns the spec for the given operation that can be used for both of
// dynamic and static range quantization.
std::unique_ptr<OpQuantSpec> GetTFOpQuantSpec(Operation* op);

    return true;
  return false;
}

std::optional<tensorflow::quantization::QuantizationComponentSpec>
GetWeightComponentSpec(
    const tensorflow::quantization::QuantizationOptions& quantization_options) {
  for (auto& cur_spec : quantization_options.quantization_method()
                            .quantization_component_specs()) {
    if (cur_spec.quantization_component() ==

// None, the default configuration is "do not quantize the model".
// 2) A set of supported operations.
// 3) Unit wise quantization precision.
// 4) Target hardware name.
// NEXT ID: 18
message QuantizationOptions {
  // The default quantization configuration for the model. If the below
  // unit-wise configuration does not exist, we use this quantization
  // configuration for the entire model. For each method, default configuration