Update README.md

README.md

@@ -3,8 +3,13 @@ Original dataset available [here](https://gluebenchmark.com/diagnostics).
 
 
 ## Dataset curation
-Only filled in the empty rows of columns "lexical semantics", "predicate-argument structure", 
+Filled in the empty rows of columns "lexical semantics", "predicate-argument structure", 
 "logic", "knowledge" with empty string `""`.
+Labels are encoded as follows
+
+```
+{"entailment": 0, "neutral": 1, "contradiction": 2}
+```
 
 ## Code to create dataset