diff --git a/demo/sequence_labeling/README.md b/demo/sequence_labeling/README.md
index c18c0fca280c2c1530057db3c55d46877a6ae251..5c8be25321dab809b0ebedd3f8dce6e0c5da6554 100644
--- a/demo/sequence_labeling/README.md
+++ b/demo/sequence_labeling/README.md
@@ -32,7 +32,7 @@ python train.py
 在命名实体识别的任务中，因不同的数据集标识实体的标签不同，评测的方式也有所差异。因此，在初始化模型的之前，需要先确定实际标签的形式，下方的`label_list`则是MSRA-NER数据集中使用的标签类别。  
 如果用户使用的实体识别的数据集的标签方式与MSRA-NER不同，则需要自行根据数据集确定。
 ```python
-label_list = ["B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC", "O"]
+label_list = hub.datasets.MSRA_NER.label_list
 label_map = {
     idx: label for idx, label in enumerate(label_list)
 }
diff --git a/demo/sequence_labeling/train.py b/demo/sequence_labeling/train.py
index 3e26d20b835b2413fc4758bfbe250e1f916f2dcd..b9acbf0a9c3a38eeb618d11a4c7189e8a6ae71e5 100644
--- a/demo/sequence_labeling/train.py
+++ b/demo/sequence_labeling/train.py
@@ -32,7 +32,7 @@ args = parser.parse_args()
 
 
 if __name__ == '__main__':
-    label_list = ["B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC", "O"]
+    label_list = MSRA_NER.label_list
     label_map = {
         idx: label for idx, label in enumerate(label_list)
     }
diff --git a/paddlehub/datasets/msra_ner.py b/paddlehub/datasets/msra_ner.py
index 8440e7c0b2f70e362db37519d57591f68fd45fd9..e258b7414f37f5175c41bf3657003340c27ebaaf 100644
--- a/paddlehub/datasets/msra_ner.py
+++ b/paddlehub/datasets/msra_ner.py
@@ -31,6 +31,7 @@ class MSRA_NER(SeqLabelingDataset):
     for research purposes.  For more information please refer to
     https://www.microsoft.com/en-us/download/details.aspx?id=52531
     """
+    label_list = ["B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC", "O"]
 
     def __init__(
             self,
@@ -39,7 +40,6 @@ class MSRA_NER(SeqLabelingDataset):
             mode: str = 'train',
     ):
         base_path = os.path.join(DATA_HOME, "msra_ner")
-        label_list = ["B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC", "O"]
 
         if mode == 'train':
             data_file = 'train.tsv'
@@ -54,6 +54,6 @@ class MSRA_NER(SeqLabelingDataset):
             mode=mode,
             data_file=data_file,
             label_file=None,
-            label_list=label_list,
+            label_list=self.label_list,
             is_file_with_header=True,
         )
diff --git a/paddlehub/utils/download.py b/paddlehub/utils/download.py
index 5f80e41f46ff13de765bd64189e67e9b0e8da2d4..d0d83a2e55bee85c0ebdca5e716327baf4d28861 100644
--- a/paddlehub/utils/download.py
+++ b/paddlehub/utils/download.py
@@ -25,17 +25,20 @@ from paddlehub.utils import log, utils, xarfile
 def download_data(url):
 
     def _wrapper(Dataset):
-        
-        def _download_dataset_from_url(*args, **kwargs):
+        def _check_download():
             save_name = os.path.basename(url).split('.')[0]
             output_path = os.path.join(hubenv.DATA_HOME, save_name)
             lock = filelock.FileLock(os.path.join(hubenv.TMP_HOME, save_name))
             with lock:
                 if not os.path.exists(output_path):
                     default_downloader.download_file_and_uncompress(url, hubenv.DATA_HOME, True)
-                    
-            return Dataset(*args, **kwargs)
-        return _download_dataset_from_url
+
+        class WrapperDataset(Dataset):
+            def __new__(cls, *args, **kwargs):
+                _check_download()
+                return super(WrapperDataset, cls).__new__(cls)
+
+        return WrapperDataset
 
     return _wrapper