Update README.md

README.md

@@ -4,9 +4,11 @@ language:
 - en
 ---
 
-CORE task as adapted for "Tokenization is Sensitive to Language Variation paper", see [arxiv](https://arxiv.org/abs/2502.15343). 
+CORE task as adapted for "Tokenization is Sensitive to Language Variation" paper, see [arxiv](https://arxiv.org/abs/2502.15343). 
 Originally downloaded from [https://github.com/TurkuNLP/CORE-corpus](https://github.com/TurkuNLP/CORE-corpus), see also [Register identification from the unrestricted open Web using the Corpus of Online Registers of English](https://link.springer.com/article/10.1007/s10579-022-09624-1). If you want to use the original CORE dataset refer to these original sources. 
 
+Warning: the column confusingly named "genre" was used as the ground truth label for experiments in the "Tokenization is Sensitive to Language Variation" paper.
+
 ```
 @article{wegmann2025tokenization,
   title={Tokenization is Sensitive to Language Variation},