Update README.md

README.md

@@ -13,7 +13,7 @@ widget:
 # RoBERTa Large model fine-tuned on Winogrande
 
 This model was fine-tuned on Winogrande dataset (XL size) in sequence classification task format, meaning that original pairs of sentences
-with corresponding options filled in are separated and classified independently of each other.
+with corresponding options filled in were separated, shuffled and classified independently of each other.
 
 ## Model description