diff --git a/doc/doc_ch/distributed_training.md b/doc/doc_ch/distributed_training.md
index 411ce5ba6aea26755cc65c405be6e0f0d5fd4738..e0251b21ea1157084e4e1b1d77429264d452aa20 100644
--- a/doc/doc_ch/distributed_training.md
+++ b/doc/doc_ch/distributed_training.md
@@ -13,7 +13,7 @@
 ```shell
 python3 -m paddle.distributed.launch \
     --log_dir=./log/ \
-    --gpus '0,1,2,3,4,5,6,7' \
+    --gpus "0,1,2,3,4,5,6,7" \
     tools/train.py \
     -c configs/rec/rec_mv3_none_bilstm_ctc.yml
 ```
diff --git a/doc/doc_en/distributed_training.md b/doc/doc_en/distributed_training.md
index 7a8b71ce308837568c84bf56292f78e9979d3907..519a42f0dc4b9bd4fa18f3f65019e4235282df92 100644
--- a/doc/doc_en/distributed_training.md
+++ b/doc/doc_en/distributed_training.md
@@ -13,7 +13,7 @@ Take recognition as an example. After the data is prepared locally, start the tr
 ```shell
 python3 -m paddle.distributed.launch \
     --log_dir=./log/ \
-    --gpus '0,1,2,3,4,5,6,7' \
+    --gpus "0,1,2,3,4,5,6,7" \
     tools/train.py \
     -c configs/rec/rec_mv3_none_bilstm_ctc.yml
 ```