fixed problem with '-'

flairNLP · Nov 25, 2020 · 62d431d · 62d431d
1 parent 39a242c
commit 62d431d
Showing 1 changed file with 2 additions and 2 deletions.
diff --git a/flair/datasets/sequence_labeling.py b/flair/datasets/sequence_labeling.py
@@ -225,10 +225,10 @@ def _parse_token(self, line: str) -> Token:
                         tagging_format_prefix = split_at_first_hyphen[0]
                         tag_without_tagging_format = split_at_first_hyphen[1]
                         if self.label_name_map and tag_without_tagging_format in self.label_name_map.keys():
-                            tag = tagging_format_prefix + "-" + self.label_name_map[tag_without_tagging_format] # for example, transforming 'B-OBJ' to 'B-part of speech object'
+                            tag = tagging_format_prefix + "-" + self.label_name_map[tag_without_tagging_format].replace("-", " ") # for example, transforming 'B-OBJ' to 'B-part-of-speech-object'
                     else: # tag without prefix, for example tag='PPER'
                         if self.label_name_map and tag in self.label_name_map.keys():
-                            tag = self.label_name_map[tag] # for example, transforming 'PPER' to 'person'
+                            tag = self.label_name_map[tag].replace("-", " ") # for example, transforming 'PPER' to 'person'
                     token.add_label(task, tag)
                 if self.column_name_map[column] == self.SPACE_AFTER_KEY and fields[column] == '-':
                     token.whitespace_after = False