Update README.md

README.md

@@ -28,6 +28,6 @@ The negative examples were produced by selecting two distinct random sentences f
 
 ## Data Splits
 The OSACPairs dataset consists of 5,041 total sentence pairs, divided as follows:
-- **Training:** 70%  
-- **Validation:** 10%  
-- **Test:** 20%
+- **Training:** 70%  = 3528
+- **Validation:** 10%  = 504
+- **Test:** 20% = 1009