Update readme.md

readme.md

@@ -23,7 +23,7 @@ LinkBERT's applications are vast and diverse, tailored to enhance both the effic
 
 ## Training and Performance
 
-LinkBERT was fine-tuned on a dataset of organic web content and editorial links, totaling over 313 million tokens. The training involved preprocessing web content, annotating links with temporary markup for clear distinction, and employing a specialized tokenization process to prepare the data for model training. In adition to commonly available data sources such as Wikipedia, additional training data was also sourced from:
+LinkBERT was fine-tuned on a dataset of organic web content and editorial links. The training involved preprocessing web content, annotating links with temporary markup for clear distinction, and employing a specialized tokenization process to prepare the data for model training. In adition to commonly available data sources such as Wikipedia, additional training data was also sourced from:
 
 - [Owayo USA](https://www.owayo.com/)
 - [Owayo Germany](https://www.owayo.de/)