diff --git a/docs/en/advanced_tutorials/distillation/distillation_en.md b/docs/en/advanced_tutorials/distillation/distillation_en.md
index cb74896b8..1b65ea77f 100644
--- a/docs/en/advanced_tutorials/distillation/distillation_en.md
+++ b/docs/en/advanced_tutorials/distillation/distillation_en.md
@@ -202,7 +202,7 @@ If everything is ready, users can begin to train the network using the following
 export PYTHONPATH=path_to_PaddleClas:$PYTHONPATH
 
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     --log_dir=R50_vd_distill_MV3_large_x1_0 \
     tools/train.py \
         -c ./configs/Distillation/R50_vd_distill_MV3_large_x1_0.yaml
diff --git a/docs/en/advanced_tutorials/image_augmentation/ImageAugment_en.md b/docs/en/advanced_tutorials/image_augmentation/ImageAugment_en.md
index 7e2a49a2a..2751ada2e 100644
--- a/docs/en/advanced_tutorials/image_augmentation/ImageAugment_en.md
+++ b/docs/en/advanced_tutorials/image_augmentation/ImageAugment_en.md
@@ -523,7 +523,7 @@ Users can use the following command to start the training process, which can als
 export PYTHONPATH=path_to_PaddleClas:$PYTHONPATH
 
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     tools/train.py \
         -c ./configs/DataAugment/ResNet50_Cutout.yaml
 ```
diff --git a/docs/en/tutorials/getting_started_en.md b/docs/en/tutorials/getting_started_en.md
index dccaec8f9..960afb55d 100644
--- a/docs/en/tutorials/getting_started_en.md
+++ b/docs/en/tutorials/getting_started_en.md
@@ -95,7 +95,7 @@ After preparing the configuration file, The training process can be started in t
 # PaddleClas starts multi-card and multi-process training through launch
 # Specify the GPU running card number by setting FLAGS_selected_gpus
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     tools/train.py \
         -c ./configs/ResNet/ResNet50_vd.yaml
 ```
@@ -104,7 +104,7 @@ The configuration can be updated by adding the `-o` parameter.
 
 ```bash
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     tools/train.py \
         -c ./configs/ResNet/ResNet50_vd.yaml \
         -o use_mix=1 \
@@ -121,7 +121,7 @@ The format of output log information is the same as above.
 
 ```
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     tools/train.py \
         -c configs/ResNet/ResNet50.yaml \
         -o pretrained_model="./pretrained/ResNet50_pretrained"
@@ -137,7 +137,7 @@ Among them, `pretrained_model` is used to set the address to load the pretrained
 
 ```
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     tools/train.py \
         -c configs/ResNet/ResNet50.yaml \
         -o checkpoints="./output/ResNet/0/ppcls"
diff --git a/docs/en/tutorials/quick_start_en.md b/docs/en/tutorials/quick_start_en.md
index 746d2c51e..25cb2a979 100644
--- a/docs/en/tutorials/quick_start_en.md
+++ b/docs/en/tutorials/quick_start_en.md
@@ -74,7 +74,7 @@ Paramters：
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/ResNet50_vd.yaml
 
@@ -92,7 +92,7 @@ The validation `Top1 Acc` curve is showmn below.
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/ResNet50_vd_finetune.yaml
 
@@ -122,7 +122,7 @@ Tringing script
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/ResNet50_vd_ssld_finetune.yaml
 ```
@@ -137,7 +137,7 @@ Training script
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/MobileNetV3_large_x1_0_finetune.yaml
 ```
@@ -154,7 +154,7 @@ Training script
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/ResNet50_vd_ssld_random_erasing_finetune.yaml
 ```
@@ -191,7 +191,7 @@ Final training script
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/R50_vd_distill_MV3_large_x1_0.yaml
 ```
diff --git a/docs/zh_CN/advanced_tutorials/distillation/distillation.md b/docs/zh_CN/advanced_tutorials/distillation/distillation.md
index e10da41ac..0c2e0d126 100644
--- a/docs/zh_CN/advanced_tutorials/distillation/distillation.md
+++ b/docs/zh_CN/advanced_tutorials/distillation/distillation.md
@@ -239,7 +239,7 @@ use_distillation: True
 export PYTHONPATH=path_to_PaddleClas:$PYTHONPATH
 
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     --log_dir=R50_vd_distill_MV3_large_x1_0 \
     tools/train.py \
         -c ./configs/Distillation/R50_vd_distill_MV3_large_x1_0.yaml
diff --git a/docs/zh_CN/advanced_tutorials/image_augmentation/ImageAugment.md b/docs/zh_CN/advanced_tutorials/image_augmentation/ImageAugment.md
index f42eb47bd..8fb45f70a 100644
--- a/docs/zh_CN/advanced_tutorials/image_augmentation/ImageAugment.md
+++ b/docs/zh_CN/advanced_tutorials/image_augmentation/ImageAugment.md
@@ -531,7 +531,7 @@ new_batch = cutmix_op(batch)
 export PYTHONPATH=path_to_PaddleClas:$PYTHONPATH
 
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     --log_dir=ResNet50_Cutout \
     tools/train.py \
         -c ./configs/DataAugment/ResNet50_Cutout.yaml
diff --git a/docs/zh_CN/tutorials/getting_started.md b/docs/zh_CN/tutorials/getting_started.md
index a104e261a..1b94c8a17 100644
--- a/docs/zh_CN/tutorials/getting_started.md
+++ b/docs/zh_CN/tutorials/getting_started.md
@@ -95,7 +95,7 @@ python tools/eval_multi_platform.py \
 # 通过设置FLAGS_selected_gpus 指定GPU运行卡号
 
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     tools/train.py \
         -c ./configs/ResNet/ResNet50_vd.yaml
 ```
@@ -104,7 +104,7 @@ python -m paddle.distributed.launch \
 
 ```bash
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     tools/train.py \
         -c ./configs/ResNet/ResNet50_vd.yaml \
         -o use_mix=1 \
@@ -119,7 +119,7 @@ python -m paddle.distributed.launch \
 
 ```
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     tools/train.py \
         -c configs/ResNet/ResNet50.yaml \
         -o pretrained_model="./pretrained/ResNet50_pretrained"
@@ -136,7 +136,7 @@ python -m paddle.distributed.launch \
 
 ```
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     tools/train.py \
         -c configs/ResNet/ResNet50.yaml \
         -o checkpoints="./output/ResNet/0/ppcls"
@@ -151,7 +151,7 @@ python -m paddle.distributed.launch \
 
 ```bash
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/eval.py \
         -c ./configs/eval.yaml \
         -o ARCHITECTURE.name="ResNet50_vd" \
diff --git a/docs/zh_CN/tutorials/quick_start.md b/docs/zh_CN/tutorials/quick_start.md
index 63c391c83..8e3749d23 100644
--- a/docs/zh_CN/tutorials/quick_start.md
+++ b/docs/zh_CN/tutorials/quick_start.md
@@ -74,7 +74,7 @@ python tools/download.py -a MobileNetV3_large_x1_0 -p ./pretrained -d True
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/ResNet50_vd.yaml
 
@@ -92,7 +92,7 @@ python -m paddle.distributed.launch \
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/ResNet50_vd_finetune.yaml
 
@@ -120,7 +120,7 @@ pretrained_model: "./pretrained/ResNet50_vd_ssld_pretrained"
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/ResNet50_vd_ssld_finetune.yaml
 ```
@@ -135,7 +135,7 @@ python -m paddle.distributed.launch \
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/MobileNetV3_large_x1_0_finetune.yaml
 ```
@@ -151,7 +151,7 @@ python -m paddle.distributed.launch \
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/ResNet50_vd_ssld_random_erasing_finetune.yaml
 ```
@@ -190,7 +190,7 @@ TRAIN:
 ```shell
 export CUDA_VISIBLE_DEVICES=0
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/train.py \
         -c ./configs/quick_start/R50_vd_distill_MV3_large_x1_0.yaml
 ```
diff --git a/tools/eval.sh b/tools/eval.sh
index a86f56c8f..9bf7ddf34 100644
--- a/tools/eval.sh
+++ b/tools/eval.sh
@@ -1,4 +1,4 @@
 python -m paddle.distributed.launch \
-    --selected_gpus="0" \
+    --gpus="0" \
     tools/eval.py \
         -c ./configs/eval.yaml
diff --git a/tools/run.sh b/tools/run.sh
index ad1e54489..5ec444557 100755
--- a/tools/run.sh
+++ b/tools/run.sh
@@ -1,7 +1,7 @@
 #!/usr/bin/env bash
 
 python -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     tools/train.py \
         -c ./configs/ResNet/ResNet50.yaml \
         -o print_interval=10
diff --git a/tools/run_dali.sh b/tools/run_dali.sh
index 4cadc346e..f23c413a1 100755
--- a/tools/run_dali.sh
+++ b/tools/run_dali.sh
@@ -3,7 +3,7 @@
 export FLAGS_fraction_of_gpu_memory_to_use=0.80
 
 python3.7 -m paddle.distributed.launch \
-    --selected_gpus="0,1,2,3" \
+    --gpus="0,1,2,3" \
     tools/train.py \
         -c configs/ResNet/ResNet50.yaml \
         -o TRAIN.batch_size=256 \