func extractDockerImage(img v1.Image) ([]byte, error) {
	layers, err := img.Layers()
	if err != nil {
		return nil, fmt.Errorf("could not fetch layers: %v", err)
	}

	// The image must have at least one layer.
	if len(layers) == 0 {
		return nil, errors.New("number of layers must be greater than zero")
	}

	layer := layers[len(layers)-1]
	mt, err := layer.MediaType()
	if err != nil {

	// If the container filesystem and image filesystem are not separate,
	// than imagefs can store both image layers and writeable layers.
	SignalImageFsAvailable Signal = "imagefs.available"
	// SignalImageFsInodesFree is amount of inodes available on filesystem that container runtime uses for storing images layers.
	// If the container filesystem and image filesystem are not separate,

// CHECK: -------- Quantization Summary --------
// CHECK: Number of quantized layers in the model
// CHECK: --------------------------------
// CHECK: Name    Count/Total
// CHECK: ================================
// CHECK: Conv2D  1/1

// CHECK: Number of quantized layers with quantized outputs: 0/1
// CHECK: Number of quantize layers added: 1
// CHECK: Number of dequantize layers added: 0
}

// -----

# Some tf endpoints actually lives under Keras.
_current_module.layers = _KerasLazyLoader(
    globals(),
    submodule="__internal__.legacy.layers",
    name="layers",
    mode="v1")
if _tf_uses_legacy_keras:
  _module_dir = _module_util.get_parent_dir_for_name(
      "tf_keras.api._v1.keras.__internal__.legacy.layers")
else:
  _module_dir = _module_util.get_parent_dir_for_name(

// CHECK: -------- Quantization Summary --------
// CHECK: Number of quantized layers in the model
// CHECK: --------------------------------
// CHECK: Name    Count/Total
// CHECK: ================================
// CHECK: Conv2D  1/2

// CHECK: Number of quantized layers with quantized outputs: 1/1
// CHECK: Number of quantize layers added: 1
// CHECK: Number of dequantize layers added: 1
}

// -----

module {

# Some tf endpoints actually lives under Keras.
_current_module.layers = _KerasLazyLoader(
    globals(),
    submodule="__internal__.legacy.layers",
    name="layers",
    mode="v1")
if _tf_uses_legacy_keras:
  _module_dir = _module_util.get_parent_dir_for_name(
      "tf_keras.api._v1.keras.__internal__.legacy.layers")
else:
  _module_dir = _module_util.get_parent_dir_for_name(

    // from DEBUGGER_TYPE_FLOAT_PER_LAYER). Each layer in the debugging model
    // has a DumpTensor, and it is used to save the entire value of outputs from
    // both the quantized and unquantized layer.
    DEBUGGER_TYPE_INT_PER_LAYER = 2;
    // DEBUGGER_TYPE_FLOAT_PER_LAYER creates a debugging model with both
    // quantized and unquantized layers. The unquantized layer's input come from

namespace mlir {
namespace quant {
namespace {

using QuantizationUnit =
    tensorflow::quantization::UnitWiseQuantizationSpec::QuantizationUnit;

// Adds QuantizationUnitLoc to quantizable layers.
class AddQuantizationUnitLocPass
    : public PassWrapper<AddQuantizationUnitLocPass,
                         OperationPass<func::FuncOp>> {
 public:

{
  "node": {
    {{ if (index .metadata.Labels "td.networking.gke.io/mesh-name") }}
      "id": "projects/{{ .gcp_project_number }}/networks/mesh:{{ (index .metadata.Labels "td.networking.gke.io/mesh-name") }}/nodes/{{ .nodeID}}",
    {{ else if (index .metadata.Labels "td.networking.gke.io/gateway-name") }}
      "id": "projects/{{ .gcp_project_number }}/networks/scope:{{ (index .metadata.Labels "td.networking.gke.io/gateway-name") }}/nodes/{{ .nodeID}}",
    {{ else }}

	// fsStatsLogs identifies stats for pod logs.
	fsStatsLogs fsStatsType = "logs"
	// fsStatsRoot identifies stats for pod container writable layers.
	fsStatsRoot fsStatsType = "root"
	// fsStatsContainer identifies stats for pod container read-only layers
	fsStatsImages fsStatsType = "images"
)

// Config holds information about how eviction is configured.
type Config struct {