diff --git a/9g_config/tokenization_9g.py b/9g_config/tokenization_9g.py
index e105acf3..385bd7bd 100644
--- a/9g_config/tokenization_9g.py
+++ b/9g_config/tokenization_9g.py
@@ -67,6 +67,8 @@ class CPM9GTokenizer(PreTrainedTokenizer):
         if vocab_file:
             if 'vocab.txt' not in vocab_file:
                 all_tokens = self.load_vocab(io.FileIO(vocab_file + VOCAB_FILES_NAMES['vocab_file'], "rb"))
+            else:
+                all_tokens = self.load_vocab(io.FileIO(vocab_file, "rb"))
         else:
             all_tokens = self.load_vocab(io.FileIO(VOCAB_FILES_NAMES['vocab_file'], "rb"))
 
@@ -305,7 +307,7 @@ class CPM9GTokenizer(PreTrainedTokenizer):
         return text[0]
 
 
-    def encode(self, text: str) -> List[int]:
+    def encode(self, text: str, add_special_tokens: bool) -> List[int]:
         """
         将文本编码为 ID 列表。
 
@@ -380,7 +382,7 @@ class CPM9GTokenizer(PreTrainedTokenizer):
                 st += 1
         return "".join(ret)
     
-    def decode(self, tokens: List[int]) -> str:
+    def decode(self, tokens: List[int], skip_special_tokens: bool) -> str:
         """
         将 ID 列表解码为字符串。
 
@@ -440,7 +442,7 @@ class CPM9GTokenizer(PreTrainedTokenizer):
             #else:
             #    ret.append(self.unk_token)
             #    st += 1
-        return ''.join(ret)
+        return ''.join([str(item) for item in ret])
 
     def _encode_unicode(self, token: str) -> List[int]:
         """