Update README.md

README.md

@@ -111,6 +111,10 @@ This is a repo of the language model used for "AdaSL: An Unsupervised Domain Ada
 
 ### About the Dialectal-Arabic-XLM-R-Base model
 
+This model is an trained as a further pre-trained of XLM-RoBERTa base using the Masked-language modeling on a dialectal Arabic corpus.
+
+### About the Dialectal-Arabic-XLM-R-Base model training corpora
+
 We have built a 5 million Tweets corpus from Twitter. The crawled tweets cover the dialects of the four Arabic world regions (EGY, GLF, LEV, and MAG regions), as well as MSA. The collected corpus consists of one million (1M) tweets per Arabic variant. We did not perform any text pre-processing on the tweets, except by removing tweets that have a small length (tweets containing less than four words).
 
 ### Usage