Format

paddle/gserver/gradientmachines/MultiGradientMachine.cpp

@@ -358,6 +358,7 @@ void MultiGradientMachine::getOutArgs(std::vector<Argument>* outArgs,
     REGISTER_TIMER("waitOutArgs");
     thread->waitOutArgsReady();
   }
+
   // outArgs_.size() only need to be calculated once.
   static int size = threads_[threads_.size() - 1]->getOutArgs().size();
   outArgs_.resize(size);
@@ -574,9 +575,9 @@ void TrainerThread::forward() {
     REGISTER_TIMER("thread_forward");
     if (batchSize_ > 0) {
       gradientMachine_->forward(
-        inArgs_, &outArgs_, multiMachine_->getPassType());
+          inArgs_, &outArgs_, multiMachine_->getPassType());
     } else {
-       outArgs_.clear();
+      outArgs_.clear();
     }
   }
   outArgsReadySem_.post();

paddle/gserver/gradientmachines/MultiGradientMachine.h

@@ -470,7 +470,6 @@ protected:
 
   /// indicate whether inArgs is copied before forward()
   bool inArgsCopied_;
-
   int batchSize_;
 };