opts->session_options.options,
      static_cast<tensorflow::ContextDevicePlacementPolicy>(
          opts->device_placement_policy),
      opts->async, device_mgr.release(),
      /*device_mgr_owned*/ true, std::move(r),
      /*cluster_flr=*/nullptr,
      /*collective_executor_mgr=*/nullptr,
      /*run_eager_op_as_function=*/opts->run_eager_op_as_function,
      /*jit_compile_rewrite=*/opts->jit_compile_rewrite);

task_index: 1
protocol: "grpc"
)");

  TF_Status* status = TF_NewStatus();
  TF_Buffer* result = TFE_GetServerDef(expected_text_proto.c_str(), status);
  EXPECT_EQ(TF_GetCode(status), TF_OK);

  ServerDef actual;
  ASSERT_TRUE(actual.ParseFromArray(result->data, result->length));
  string actual_text_proto;
  tensorflow::protobuf::TextFormat::PrintToString(actual, &actual_text_proto);
  EXPECT_EQ(expected_text_proto, actual_text_proto);

}
TEST(CAPI, RemoteExecuteSilentCopiesAsync) {
  TestRemoteExecuteSilentCopiesOp(/*async=*/true, /*remote=*/true);
}
TEST(CAPI, RemoteExecuteSilentCopiesLocal) {
  TestRemoteExecuteSilentCopiesOp(/*async=*/false, /*remote=*/false);
}
TEST(CAPI, RemoteExecuteSilentCopiesLocalAsync) {
  TestRemoteExecuteSilentCopiesOp(/*async=*/true, /*remote=*/false);
}

build:mkl --define=build_with_mkl=true --define=enable_mkl=true
build:mkl --define=tensorflow_mkldnn_contraction_kernel=0
build:mkl --define=build_with_openmp=true
build:mkl -c opt

# config to build OneDNN backend with a user specified threadpool.
build:mkl_threadpool --define=build_with_mkl=true --define=enable_mkl=true
build:mkl_threadpool --define=tensorflow_mkldnn_contraction_kernel=0

        boundaries.

*   XLA compilation:

    *   `tf.function(experimental_compile=True)` has become a stable API,
        renamed `tf.function(jit_compile=True)`.
    *   XLA can now compile MirroredStrategy: the step function passed
        to`strategy.run` can now be annoted with `jit_compile=True`.

*   `tf.distribute`:

    *   Rename `experimental_prefetch_to_device` in `tf.distribute.InputOptions`

  }

  status_ = registry_.Register("MatMul", MatMulRegisterer);
  ASSERT_EQ(errors::OK, status_.code()) << status_.message();

  bool transpose_a_vals[] = {false, false, true, true};
  bool transpose_b_vals[] = {false, true, false, true};
  float dA_vals[4][9] = {{24, 15, 6, 24, 15, 6, 24, 15, 6},
                         {18, 15, 12, 18, 15, 12, 18, 15, 12},
                         {24, 24, 24, 15, 15, 15, 6, 6, 6},

  else:
    question += ' [y/N]: '

  var = environ_cp.get(var_name)
  if var is not None:
    var_content = var.strip().lower()
    true_strings = ('1', 't', 'true', 'y', 'yes')
    false_strings = ('0', 'f', 'false', 'n', 'no')
    if var_content in true_strings:
      var = True
    elif var_content in false_strings:
      var = False
    else:
      raise UserInputError(

// c_api_experimental.h
struct TFE_ContextOptions {
  TF_SessionOptions session_options;
  // true if async execution is enabled.
  bool async = false;
  TFE_ContextDevicePlacementPolicy device_placement_policy{
      TFE_DEVICE_PLACEMENT_SILENT};
  // If true, use TFRT backend
  bool use_tfrt = false;
  // Whether to run elementary eager ops wrapped in a call op.

        "math_grad.h",
        "nn_grad.h",
        "not_differentiable.h",
    ],
    visibility = ["//tensorflow/python:__pkg__"],
)

cc_library(
    name = "grad_test_helper",
    testonly = True,
    srcs = ["grad_test_helper.cc"],
    hdrs = ["grad_test_helper.h"],
    visibility = ["//visibility:private"],
    deps = [
        "//tensorflow/c/eager:gradient_checker",

// exporting a full trace so other code can do backpropagation. Not thread-safe.
template <typename Gradient, typename BackwardFunction, typename TapeTensor>
class GradientTape {
 public:
  // If `persistent` is true, GradientTape will not eagerly delete backward
  // functions (and hence the tensors they keep alive). Instead, everything
  // is deleted in ~GradientTape. Persistent GradientTapes are useful when