1 paper · 1 filter
Mohammadamin Habibollah, Davood Rafiei
Evaluating text-to-SQL systems remains largely fragile: correctness is typically judged by executing predicted and gold SQL queries on a single static database, even though the sam…