Update files from the datasets library (from 1.5.0)

Release notes: https://github.com/huggingface/datasets/releases/tag/1.5.0

lener_br.py

@@ -138,7 +138,7 @@ class LenerBr(datasets.GeneratorBasedBuilder):
             ner_tags = []
 
             for line in f:
-                if line.startswith("-DOCSTART-") or line == "" or line == "\n":
+                if line == "" or line == "\n":
                     if tokens:
                         yield guid, {
                             "id": str(guid),