i was training the same data set, extracting photos from video is now taking a long long time (s3fd , fan) , it also fails when i try to set all the masks , something i had no issue before in version2, 17 min video took me around 16 hours to extract , and now when training the lu finder is taking hour where it used to take few single minutes . i am running on windows 10 , nvedia RTX 2070 super 8 gb . on default settings