Update README.md

README.md

@@ -7,13 +7,9 @@ The **SROIE** (Scanned Receipts OCR and Information Extraction) dataset consists
 
 **Content**
 
-- 626 receipts 
-- 4 entity types: company, date, address, total
+- 973 annotated receipts (626 train/347 test) 
+- 4 entity types: _company, date, address, total_
 - OCR text + layout information available
-    
-
-**Unique Aspects**
-
 - Short documents (receipts) with sparse (majority of tokens 'O' outside a mention) but critical fields
 - Widely used as a benchmark for document information extraction