Fixed bug to calculate correct xpath_sub_list in MarkupLMTokenizer (#22302)

Fixed bug to calculate correct xpath_sub_list in MarkupLMTokenizer. Earlier xpath_sub_list was same as xpath_tags_list

Co-authored-by: dusejat <dusejat@amazon.com>

src/transformers/models/markuplm/tokenization_markuplm.py

@@ -301,7 +301,7 @@ class MarkupLMTokenizer(PreTrainedTokenizer):
             xpath_subs_list.append(min(self.max_width, sub))
 
         xpath_tags_list = xpath_tags_list[: self.max_depth]
-        xpath_subs_list = xpath_tags_list[: self.max_depth]
+        xpath_subs_list = xpath_subs_list[: self.max_depth]
         xpath_tags_list += [self.pad_tag_id] * (self.max_depth - len(xpath_tags_list))
         xpath_subs_list += [self.pad_width] * (self.max_depth - len(xpath_subs_list))
 

src/transformers/models/markuplm/tokenization_markuplm_fast.py

@@ -275,7 +275,7 @@ class MarkupLMTokenizerFast(PreTrainedTokenizerFast):
             xpath_subs_list.append(min(self.max_width, sub))
 
         xpath_tags_list = xpath_tags_list[: self.max_depth]
-        xpath_subs_list = xpath_tags_list[: self.max_depth]
+        xpath_subs_list = xpath_subs_list[: self.max_depth]
         xpath_tags_list += [self.pad_tag_id] * (self.max_depth - len(xpath_tags_list))
         xpath_subs_list += [self.pad_width] * (self.max_depth - len(xpath_subs_list))