How to speed up executemany in Python on INSERT IGNORE (mysql)?

Question

MySQL

Python 3, mysql 5.7, mysql.connector. I insert a large number of values ​​into the database (~ 5 million rows, in batches of 5000 rows) through executemany. Some values ​​are not unique. In this case, you just need to skip this insert / update / replace it. If you use INSERT - if there is at least one error in a batch of 5000 rows, if there is a duplicate, the entire batch is not inserted. Insertion speed is fast. To solve the problem, I tried INSERT IGNORE, the speed drops many times, the impression is that queries are executed separately, not in batches. How can this problem be solved without sacrificing speed?