text in never_split should not lowercase

pytorch_pretrained_bert/tokenization.py

@@ -182,7 +182,7 @@ class BasicTokenizer(object):
         orig_tokens = whitespace_tokenize(text)
         split_tokens = []
         for token in orig_tokens:
-            if self.do_lower_case:
+            if self.do_lower_case and token not in self.never_split:
                 token = token.lower()
                 token = self._run_strip_accents(token)
             split_tokens.extend(self._run_split_on_punc(token))