Dataset #1
Description
Hi
I just wondering how many vulnerability code and patched code is used per CVE.
In paper, you guys used more than 5 vulnerable functions and 5 patched functions, so that the number of generated similar pairs and different pairs for training.
But I think the number of functions doesn't make sense in Table 6.
( I think you randomly choose to avoid data imbalance. )
If you dont mind, can i get the real data when you guys used in paper ?