Update README.md

README.md

@@ -12,11 +12,14 @@ tags:
 - torch
 ---
 
-# ROBERTA BASE (cased) finetuned on private Bulgarian STSB, NLI data
+# ROBERTA BASE (cased) trained on private Bulgarian-English parallel data
+This is a Multilingual Roberta model. It could be used for creating embeddings of Bulgarian sentences. 
+
+Using the ideas from [Sentence-BERT](https://arxiv.org/abs/2004.09813), the training is based on the idea that a translated sentence should be mapped to the same location in the vector space as the original sentence. 
 
 This model is cased: it does make a difference between bulgarian and Bulgarian.
 
-It was finetuned on private Bulgarian STSB, NLI data.
+It was trained on private Bulgarian-English parallel data.
 
 Then, it was compressed via [progressive module replacing](https://arxiv.org/abs/2002.02925).