private:
  LogicalResult matchAndRewrite(TF::TPUPartitionedCallOp call_op,
                                PatternRewriter& rewriter) const override {
    auto f_attr = mlir::dyn_cast<FlatSymbolRefAttr>(call_op.getFAttr());
    auto module_op = call_op->getParentOfType<ModuleOp>();
    SymbolTable symbol_table(module_op);

    auto f_name = f_attr.getValue();
    func::FuncOp float_func =

  }

  LogicalResult matchAndRewrite(TF::PartitionedCallOp op,
                                PatternRewriter& rewriter) const override {
    const auto f_attr = mlir::dyn_cast<FlatSymbolRefAttr>(op.getFAttr());
    StringRef function_name = f_attr.getValue();
    if (!function_name.starts_with(kDequantizeFunctionName)) return failure();

    llvm::SmallVector<Operation*> users(op->getUsers().begin(),

  auto spec = std::make_unique<OpQuantSpec>();
  if (auto call_op = dyn_cast<TF::PartitionedCallOp>(op)) {
    StringRef function_name =
        mlir::cast<FlatSymbolRefAttr>(call_op.getFAttr()).getValue();
    if (!function_name.starts_with("composite_")) {
      return spec;
    }
    if (function_name.contains("depthwise_conv2d")) {
      spec->coeff_op_quant_dim[1] = 3;

            << min_num_elements_for_weights_ << " elements).";
        call_op->removeAttr(kQuantTraitAttrName);
      }
    }

    StringRef function_name =
        mlir::cast<FlatSymbolRefAttr>(call_op.getFAttr()).getValue();
    if ((quantization_method_ == tensorflow::quantization::QuantizationMethod::
                                     METHOD_DYNAMIC_RANGE_INT8) &&
        (function_name.contains("batch_matmul") ||

      grad, axis=reduction_axes, keepdims=True)
  bias_grad = tf.reshape(updates_grad_reshaped, input_value_shape)

  dilations = [1, op.get_attr('dilation_w'), op.get_attr('dilation_h'), 1]
  strides = [1, op.get_attr('stride_w'), op.get_attr('stride_h'), 1]
  padding = op.get_attr('padding')
  shape_0, shape_1 = tf.shape_n([op.inputs[0], op.inputs[1]])
  return [
      tf.compat.v1.nn.conv2d_backprop_input(
          shape_0,

    float max_percentile;
    OP_REQUIRES_OK(context, context->GetAttr("calibration_method",
                                             &calibration_method_value));
    OP_REQUIRES_OK(context, context->GetAttr("num_bins", &num_bins));
    OP_REQUIRES_OK(context,
                   context->GetAttr("min_percentile", &min_percentile));
    OP_REQUIRES_OK(context,
                   context->GetAttr("max_percentile", &max_percentile));

    string file_name;
    string func_name;
    string node_name;
    OP_REQUIRES_OK(ctx, ctx->GetAttr("log_dir_path", &log_dir_path));
    OP_REQUIRES_OK(ctx, ctx->GetAttr("enabled", &enabled_));
    OP_REQUIRES_OK(ctx, ctx->GetAttr("file_name", &file_name));
    OP_REQUIRES_OK(ctx, ctx->GetAttr("func_name", &func_name));
    OP_REQUIRES_OK(ctx, ctx->GetAttr("node_name", &node_name));
    OP_REQUIRES_OK(ctx, ctx->env()->RecursivelyCreateDir(log_dir_path));

    .SetIsStateful()
    .SetShapeFn([](shape_inference::InferenceContext* c) {
      c->set_output(0, c->Scalar());
      DataType t;
      TF_RETURN_IF_ERROR(c->GetAttr("dtype", &t));
      PartialTensorShape p;
      TF_RETURN_IF_ERROR(c->GetAttr("shape", &p));
      shape_inference::ShapeHandle s;
      TF_RETURN_IF_ERROR(c->MakeShapeFromPartialTensorShape(p, &s));
      c->set_output_handle_shapes_and_types(

	"errors"
	"flag"
	"fmt"
	"log"
	"os"

	"github.com/olekukonko/tablewriter"
	"github.com/pkg/xattr"
)

var (
	path, name string
	value      uint64
	set, list  bool
)

func getxattr(path, name string) (uint64, error) {
	buf, err := xattr.LGet(path, name)
	if err != nil {
		return 0, err
	}

	return binary.LittleEndian.Uint64(buf[:8]), nil
}

import types
from enum import Enum
from typing import Any, Callable, Dict, Set, Type, TypeVar, Union

from pydantic import BaseModel

DecoratedCallable = TypeVar("DecoratedCallable", bound=Callable[..., Any])
UnionType = getattr(types, "UnionType", Union)
ModelNameMap = Dict[Union[Type[BaseModel], Type[Enum]], str]