and
    [GPU developer Dockerfile](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/tools/dockerfiles/dockerfiles/devel-gpu.Dockerfile)
    for the required packages. Alternatively, use the said
    [tensorflow/build Docker images](https://hub.docker.com/r/tensorflow/build)
    (`tensorflow/tensorflow:devel` and `tensorflow/tensorflow:devel-gpu` are no

#include "tensorflow/core/platform/thread_annotations.h"
#include "tensorflow/core/platform/types.h"
#include "tensorflow/core/public/session.h"
#include "tensorflow/core/public/version.h"

// The implementation below is at the top level instead of the
// brain namespace because we are defining 'extern "C"' functions.
using tensorflow::AttrValueMap;
using tensorflow::DataType;
using tensorflow::ExtendSessionGraphHelper;
using tensorflow::FullTypeDef;

        sufficiently tested with GPUs yet.
*   TensorFlow pip packages are now built with CUDA11.2 and cuDNN 8.1.0

## Breaking Changes

*   The `TF_CPP_MIN_VLOG_LEVEL` environment variable has been renamed to
    `TF_CPP_MAX_VLOG_LEVEL` which correctly describes its effect.

## Bug Fixes and Other Changes

*   `tf.keras`:

    *   Preprocessing layers API consistency changes:

  def valid_api_level(api_level):
    return os.path.exists(
        os.path.join(android_sdk_home_path, 'platforms',
                     'android-' + api_level))

  android_api_level = prompt_loop_or_load_from_env(
      environ_cp,
      var_name='ANDROID_API_LEVEL',
      var_default=api_levels[-1],
      ask_for_var=('Please specify the Android SDK API level to use. '

  tensorflow::XlaAutoJitFlag flag =
      tensorflow::GetMarkForCompilationPassFlags()->xla_auto_jit_flag;
  return static_cast<unsigned char>(flag.optimization_level_single_gpu > 0 ||
                                    flag.optimization_level_general > 0);
}

unsigned char TF_GetXlaConstantFoldingDisabled() {
  return static_cast<unsigned char>(