struct TfrtFunctionCompileOptions : public TfrtCompileOptions {
  // Currently only SavedModel API inference uses the tpu_fuse_ops option
  TfrtFunctionCompileOptions() {
    tpu_fuse_ops = false;
    // Currently grappler is not correctly applied in the eager execution of TF
    // functions, as it may sometimes remove arguments and results.
    enable_grappler = false;
  }

// Note: This implements the fusions performed in the old Remapper Grappler
// pass. That pass has specific cases for GPU and based on different
// target configurations on both CPU and GPU (Intel MKL, ROCm, etc.). This MLIR
// pass covers (some of) the general CPU case and at the moment does not account
// for any target-specific configurations.

// This pass is being ported over from the Grappler Remapper pass based on

ntegration_testing/... -//tensorflow/compiler/tf2tensorrt/... -//tensorflow/core/tpu/... -//tensorflow/lite/... -//tensorflow/tools/toolchains/...  -//tensorflow/go/... -//tensorflow/java/... -//tensorflow/core/grappler/optimizers:auto_mixed_precision_test_cpu -//tensorflow/core/grappler/optimizers:remapper_test_cpu -//tensorflow/core/kernels/image:resize_bicubic_op_test -//tensorflow/compiler/mlir/tfr/examples/customization:test_ops_test -//tensorflow/compiler/mlir/tfr/examples/mnist:mnist_ops_test...

        "//tensorflow/cc/saved_model:loader.h",
    ],
)

filegroup(
    name = "pywrap_required_hdrs",
    srcs = [
        "training/coordinator.h",
    ],
    visibility = ["//tensorflow/python/grappler:__pkg__"],
)

cc_library(
    name = "gradients",
    srcs = [
        "framework/gradients.cc",
        "framework/while_gradients.cc",
        "framework/while_gradients.h",
    ],

  ASSERT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

  // Add a config_proto attr, to trigger grappler graph rewrites in the current
  // eager runtime.
  if (enable_grappler) {
    tensorflow::ConfigProto config;
    // Do not skip grappler optimization even for small graphs.
    config.mutable_graph_options()
        ->mutable_rewrite_options()
        ->set_min_graph_nodes(-1);

        summaries will be written, even when `tf.summary.record_if()` is not n
        effect, by returning True tensor if default writer is present.
*   Grappler:
    *   Disable default Grappler optimization timeout to make the optimization
        pipeline deterministic. This may lead to increased model loading time,
        because time spent in graph optimizations is now unbounded (was 20

		return msg.ProtoReflect().Type(), nil
	}
	return typ, nil
}

// Wrapper is a wrapper around the Envoy ConfigDump
// It has extra helper functions for handling any/struct/marshal protobuf pain
type Wrapper struct {
	*admin.ConfigDump
}

// UnmarshalJSON is a custom unmarshaller to handle protobuf pain
func (w *Wrapper) UnmarshalJSON(b []byte) error {
	cd := &admin.ConfigDump{}

import org.gradle.api.tasks.TaskAction;
import org.gradle.api.tasks.options.Option;
import org.gradle.api.tasks.options.OptionValues;
import org.gradle.api.tasks.wrapper.internal.DefaultWrapperVersionsResources;
import org.gradle.api.tasks.wrapper.internal.WrapperDefaults;
import org.gradle.api.tasks.wrapper.internal.WrapperGenerator;
import org.gradle.util.GradleVersion;
import org.gradle.util.internal.GUtil;

	"istio.io/istio/pkg/util/protomarshal"
)

// Wrapper is a wrapper around the Envoy Clusters
// It has extra helper functions for handling any/struct/marshal protobuf pain
type Wrapper struct {
	*admin.Clusters
}

// MarshalJSON is a custom marshaller to handle protobuf pain
func (w *Wrapper) MarshalJSON() ([]byte, error) {
	return protomarshal.Marshal(w)
}

  TF_RET_CHECK(initialized_ && !edges_contracted_ && !clusters_created_);
  edges_contracted_ = true;

  // TODO(hpucha): Handle the case where kXlaClusterAttr is already set (for
  // example, from the Grappler fusion pass).

  // In general there are multiple maximal clusterings, but they are not all
  // equally performant.  Some clustering decision are likely to improve