experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Clarify the need for calling tensorflow::port::InitMain() before using the rest of TensorFlow.

Browse Source 
PiperOrigin-RevId: 219185259
This commit is contained in:
A. Unique TensorFlower 2018-10-29 13:42:46 -07:00 committed by TensorFlower Gardener
parent 4f09245c1f
commit d09c687897
1 changed files with 3 additions and 3 deletions

6
tensorflow/core/platform/init_main.h
View File

@ -19,10 +19,10 @@ limitations under the License.
namespace tensorflow {
namespace port {
// Platform-specific initialization routine that may be invoked by a
// Platform-specific initialization routine that should be invoked by a
// main() program that uses TensorFlow.
//
// Default implementation does nothing.
// This performs necessary initialization on some platforms; TensorFlow
// may not work unless it has been called.
void InitMain(const char* usage, int* argc, char*** argv);
} // namespace port