Update README.md

README.md

@@ -24,4 +24,7 @@ and splits the data into smaller chunks, of size ~820 chars
 The logic only splits on spaces, so the chunks are likely to be slightly larger than 820 chars. 
 The dataset has been normalized into lower case, with accents and non-english characters removed.
 Items with less than 200 chars or more than 1000 chars have been removed. 
-The data has not been shuffled (you probably want to shuffle before using).
+The data has not been shuffled (you can either use `dataset.shuffle(...)`, 
+or download the shuffled version [here](https://huggingface.co/datasets/sradc/chunked-shuffled-wikipedia20220301en-bookcorpusopen),
+which will be faster to iterate over).
+