diff --git a/official/cv/dpn/README.md b/official/cv/dpn/README.md
index 655aaf47dfb43af07a6a77a7a131b8e2fc999d71..f1ece0a32454618cfb44b70351dd7a6d1e3964d3 100644
--- a/official/cv/dpn/README.md
+++ b/official/cv/dpn/README.md
@@ -318,10 +318,10 @@ DPN evaluate success!
 ### [Export MindIR](#contents)
 
 ```shell
-python export.py --config_path [CONFIG_PATH] --ckpt_file [CKPT_PATH] --file_name [FILE_NAME] --file_format [FILE_FORMAT]
+python export.py --config_path [CONFIG_PATH] --checkpoint_path [CKPT_PATH] --file_name [FILE_NAME] --file_format [FILE_FORMAT]
 ```
 
-The ckpt_file parameter is required,
+The `checkpoint_path` parameter is required,
 `FILE_FORMAT` should be in ["AIR", "MINDIR"]
 
 - Export MindIR on Modelarts
@@ -356,7 +356,7 @@ bash run_infer_310.sh [MINDIR_PATH] [DATASET_PATH] [DEVICE_ID]
 ```
 
 - `MINDIR_PATH` specifies path of used "MINDIR" OR "AIR" model.
-- `DATASET_PATH` specifies path of cifar10 datasets
+- `DATASET_PATH` specifies path of imagenet2012 datasets
 - `DEVICE_ID` is optional, default value is 0.
 
 ### [Result](#contents)
diff --git a/official/cv/faster_rcnn/train.py b/official/cv/faster_rcnn/train.py
index ce2ef9def96bd00641ccf88680b754c4ba933620..c810397c8c569c687e62607388e73f6971614f5a 100644
--- a/official/cv/faster_rcnn/train.py
+++ b/official/cv/faster_rcnn/train.py
@@ -202,7 +202,7 @@ def train_fasterrcnn():
         cb += [eval_cb]
 
     model = Model(net)
-    model.train(config.epoch_size, dataset, callbacks=cb, dataset_sink_mode=False)
+    model.train(config.epoch_size, dataset, callbacks=cb)
 
 
 if __name__ == '__main__':
diff --git a/research/cv/deeplabv3plus/scripts/run_eval_s8_multiscale_gpu.sh b/research/cv/deeplabv3plus/scripts/run_eval_s8_multiscale_gpu.sh
index a0688ff39eb99cc65b75f5ab607a12aeaaf007ec..8fd2cff5b730f9d61216ead66ff0c19ce253a625 100644
--- a/research/cv/deeplabv3plus/scripts/run_eval_s8_multiscale_gpu.sh
+++ b/research/cv/deeplabv3plus/scripts/run_eval_s8_multiscale_gpu.sh
@@ -42,7 +42,7 @@ cd ${eval_path} || exit
 
 python ./eval.py  --data_root=$DATA_ROOT  \
                   --data_lst=$DATA_LST  \
-                  --batch_size=16  \
+                  --batch_size=8  \
                   --crop_size=513  \
                   --ignore_label=255  \
                   --num_classes=21  \