diff --git a/fastSum/Dataloader/summarizationLoader.py b/fastSum/Dataloader/summarizationLoader.py
index ccaa16f7da202f2126f8dcb71a6931f46c7c31e7..74d0763ecef54444dae036dc1d3affe5f8ddeb75 100644
--- a/fastSum/Dataloader/summarizationLoader.py
+++ b/fastSum/Dataloader/summarizationLoader.py
@@ -1,4 +1,4 @@
-from typing import Union, Dict
+from typing import Union, Dict, Optional
 import os
 import random
 
@@ -29,12 +29,13 @@ class SumLoader(JsonLoader):
     所有摘要数据集loader的父类
     """
 
-    def __init__(self):
-        fields = {
-            'text': 'text',
-            'summary': 'summary',
-            'label': Const.TARGET
-        }
+    def __init__(self, fields: Optional[Dict[str, str]] = None):
+        if fields is None:
+            fields = {
+                'text': 'text',
+                'summary': 'summary',
+                'label': Const.TARGET
+            }
         super(SumLoader, self).__init__(fields=fields)
 
     def load(self, paths: Union[str, Dict[str, str]] = None) -> DataBundle:
@@ -44,6 +45,11 @@ class SumLoader(JsonLoader):
         default_cache_path = get_cache_path()
         url = _get_dataset_url(dataset_name, DATASET_DIR)
         output_dir = cached_path(url_or_filename=url, cache_dir=default_cache_path, name='dataset')
+        # https://gitee.com/fastnlp/fastNLP/blob/7b4e099c5267efb6a4a88b9d789a0940be05bb56/fastNLP/io/file_utils.py#L201
+        # 如果只有一个文件, get_filepath 返回 filepath + filename
+        # os.path.dirname 反向处理
+        if os.path.isfile(output_dir):
+            output_dir = os.path.dirname(output_dir)
         return output_dir
 
 
@@ -356,7 +362,13 @@ class AMILoader(SumLoader):
     """
 
     def __init__(self, valid_ratio=0.05, test_ratio=0.05):
-        super(AMILoader, self).__init__()
+        # AMI 没有 label
+        fields = {
+            'text': 'text',
+            'summary': 'summary',
+        }
+        super(AMILoader, self).__init__(fields)
+
         self.valid_ratio = valid_ratio
         self.test_ratio = test_ratio
 
@@ -398,7 +410,12 @@ class ICSILoader(SumLoader):
     """
 
     def __init__(self, valid_ratio=0.05, test_ratio=0.05):
-        super(ICSILoader, self).__init__()
+        # ICSI 没有 label
+        fields = {
+            'text': 'text',
+            'summary': 'summary',
+        }
+        super(ICSILoader, self).__init__(fields)
         self.valid_ratio = valid_ratio
         self.test_ratio = test_ratio
 
@@ -431,7 +448,7 @@ class ICSILoader(SumLoader):
         return data_bundle
 
 
-def _split_set(dataset_name, data_dir, split_name1="dev", split_name2="train", ratio=0.0, suffix='jsonl'):
+def _split_set(dataset_name, data_dir, split_name1="dev", split_name2="train", ratio=0.0, suffix='jsonl', keep_orig: bool = True):
     if ratio == 0:
         os.renames(os.path.join(data_dir, f'{dataset_name}.{suffix}'),
                    os.path.join(data_dir, f'{split_name2}.{suffix}'))
@@ -449,7 +466,10 @@ def _split_set(dataset_name, data_dir, split_name1="dev", split_name2="train", r
                             f2.write(line)
                         else:
                             f1.write(line)
-                os.remove(os.path.join(data_dir, f'{dataset_name}.{suffix}'))
+                if keep_orig:
+                    assert split_name1 != dataset_name and split_name2 != dataset_name
+                else:
+                    os.remove(os.path.join(data_dir, f'{dataset_name}.{suffix}'))
                 os.renames(os.path.join(data_dir, f'middle_file.{suffix}'),
                            os.path.join(data_dir, f'{split_name2}.{suffix}'))
             finally: