-   Read the [Code of Conduct](CODE_OF_CONDUCT.md).
-   Ensure you have signed the
    [Contributor License Agreement (CLA)](https://cla.developers.google.com/).
-   Check if your changes are consistent with the
    [guidelines](#general-guidelines-and-philosophy-for-contribution).
-   Changes are consistent with the [Coding Style](#c-coding-style).
-   Run the [unit tests](#running-unit-tests).

## How to become a contributor and submit your own code

* `stream_executor`
  * Top level `stream_executor` directory has been deleted, users should use equivalent headers and targets under `compiler/xla/stream_executor`.
* `tf.nn`
  * Added `tf.nn.experimental.general_dropout`, which is similar to `tf.random.experimental.stateless_dropout` but accepts a custom sampler function.
* `tf.types.experimental.GenericFunction`

#     nonccl:       Disable nccl support.
#
#
# Remote build execution options (only configured to work with TF team projects for now.)
#     rbe_base:  General RBE options shared by all flavors.
#     rbe_linux: General RBE options used on all linux builds.
#     rbe_win_base:   General RBE options used on all Windows builds. Not to be used standalone.
#     rbe_win_clang:  Options specific to compiling using Clang.
#

      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>(
      tensorflow::GetBuildXlaOpsPassFlags()->tf_xla_disable_constant_folding);
}

  // (now inaccessible) corresponding JVPs, but ForwardAccumulator's destructor
  // will release remaining references.
  //
  // This method is not thread-safe (and in general ForwardAccumulator is not
  // thread-safe).
  Status Accumulate(
      const string& op_type, const std::vector<TapeTensor>& input_tensors,
      const std::vector<TapeTensor>& output_tensors,