Update README.md

README.md

@@ -25,7 +25,7 @@ When researching LLMs [tokenized similarly to GPT-2/GPT-3](https://platform.open
 - Fast searches for common phrases containing particular tokens or substrings (and in particular sequence positions).
 - Showing the effects of training set n-gram frequency.
 
-The authors used this dataset to show that sparse auto-encoders are biased toward reconstructing the most common n-grams.
+The authors (Thomas Dooms and Dan Wilhelm) used this dataset to show that sparse auto-encoders are biased toward reconstructing the most common n-grams.
 
 ## Loading the Dataset