add warning about tokenisation

Stab-Gurevych-Essays.py

@@ -387,6 +387,9 @@ class NewDataset(datasets.GeneratorBasedBuilder):
         tokens = []
         labels = []
         # tokenise spans
+        # WARN: the old dataset considered punctuation to be separate tokens, whilst this one doesnt. 
+        #   this shouldnt matter however, since this is just pre-tokenisation, which will pre tokenised for the respective model later on.
+        #   i assume that the later tokenisation will create equal results.
         for span in spans:
             span_tokens = span[1].split()
             label = span[0]