Update bmc.py

bmc.py

@@ -60,6 +60,7 @@ class BMC(datasets.GeneratorBasedBuilder):
                 {
                     "id": datasets.Value("string"),
                     "tokens": datasets.Sequence(datasets.Value("string")),
+                    "raw_tags": datasets.Sequence(datasets.Value("string")),
                     "ner_tags": datasets.Sequence(
                         datasets.features.ClassLabel(
                             names=[
@@ -124,6 +125,7 @@ class BMC(datasets.GeneratorBasedBuilder):
             guid = 0
             tokens = []
             ner_tags = []
+            raw_tags = []
             for line in f:
                 if line.startswith("-DOCSTART-") or line == "" or line == "\n":
                     if tokens:
@@ -134,14 +136,17 @@ class BMC(datasets.GeneratorBasedBuilder):
                         }
                         guid += 1
                         tokens = []
+                        raw_tags = []
                         ner_tags = []
                 else:
                     splits = line.split(sep)
                     tokens.append(splits[0])
+                    raw_tags.append(splits[1].rstrip())
                     ner_tags.append(splits[1].rstrip())
             # last example
             yield guid, {
                 "id": str(guid),
                 "tokens": tokens,
+                "raw_tags": raw_tags, 
                 "ner_tags": ner_tags,
             }