delete from table_name a where rowid <
(select max(rowid) from table_name b
where a.col1 = b.col1
and a.col2 = b.col2 -- 遍历所以数据,以确定重复纪录
)