The pytorch_inference process must respond to every request as the Java side is waiting for a response. Partly this is ensured by a try catch block, we can't be certain what types of exception might be thrown so better to catch the base std::exception rather than derived classes.
The
pytorch_inferenceprocess must respond to every request as the Java side is waiting for a response. Partly this is ensured by a try catch block, we can't be certain what types of exception might be thrown so better to catch the basestd::exceptionrather than derived classes.