diff --git a/model_zoo/research/cv/STGAN/README.md b/model_zoo/research/cv/STGAN/README.md
index 69b2a129eed..492a2831d74 100644
--- a/model_zoo/research/cv/STGAN/README.md
+++ b/model_zoo/research/cv/STGAN/README.md
@@ -201,3 +201,10 @@ In dataset.py, we set the seed inside ```create_dataset``` function.
 ## [ModelZoo Homepage](#contents)
 
 Please check the official [homepage](https://gitee.com/mindspore/mindspore/tree/master/model_zoo).
+
+## FAQ
+
+Please refer to [ModelZoo FAQ](https://gitee.com/mindspore/mindspore/tree/master/model_zoo#FAQ) to get some common FAQ.
+
+- **Q**: Get "out of memory" error in PYNATIVE_MODE.
+  **A**: You can set smaller batch size, e.g. 32, 16.
diff --git a/model_zoo/research/cv/STGAN/src/utils/args.py b/model_zoo/research/cv/STGAN/src/utils/args.py
index a4443343738..d1d8c56a2e2 100644
--- a/model_zoo/research/cv/STGAN/src/utils/args.py
+++ b/model_zoo/research/cv/STGAN/src/utils/args.py
@@ -273,7 +273,7 @@ def get_args(phase):
                             device_id=int(os.environ["DEVICE_ID"]))
         context.reset_auto_parallel_context()
         context.set_auto_parallel_context(
-            parallel_mode=ParallelMode.AUTO_PARALLEL,
+            parallel_mode=ParallelMode.DATA_PARALLEL,
             gradients_mean=True,
             device_num=args.device_num)
         init()