Update README.md

README.md

@@ -7,11 +7,13 @@ license: mit
 The approach is simple:
 
 1. Combine all available NLI data without any domain-dependent re-balancing or re-weighting.
-
 2. Finetune several SOTA transformers of different sizes (20m parameters to 300m parameters) on the combined data.
 3. Evaluate on challenging NLI datasets.
 
-This model is a ```deberta-v3-base``` finetuned on 20+ NLI datasets. 
+This model was trained using [SentenceTransformers](https://sbert.net) [Cross-Encoder](https://www.sbert.net/examples/applications/cross-encoder/README.html) class. It is based on [microsoft/deberta-v3-large](https://huggingface.co/microsoft/deberta-v3-large). 
+
+### Data
+20+ NLI datasets were combined to train a binary classification model. The contradiction and neutral labels were mixed to form a non-entailment class. 
 
 ### Usage
 
@@ -45,7 +47,7 @@ with torch.no_grad():
 ```
 
 
-In [Sentence-Transformers](https://www.sbert.net/)
+In Sentence-Transformers
 
 ```python
 from sentence_transformers import CrossEncoder