fix benchmark (#596)

* fix benchmark

* fix benchmark

benchmark/run_benchmark.sh

@@ -51,6 +51,7 @@ function _train(){
     case ${run_mode} in
     sp) train_cmd="python -u tools/main.py "${train_cmd} ;;
     mp)
+        rm -rf ./mylog
         train_cmd="python -m paddle.distributed.launch --log_dir=./mylog --gpus=$CUDA_VISIBLE_DEVICES tools/main.py "${train_cmd}
         log_parse_file="mylog/workerlog.0" ;;
     *) echo "choose run_mode(sp or mp)"; exit 1;