Training on 6 labels instead 3

[AlexUP]

Hi makatx,

Thank you for your repository, it has helped me greatly in understanding the core objectives of YOLO.

I am struggling to modify your code to train on 6 labels instead of 3, because the loss value is converging to 1.9 and I was wondering whether you have experienced the same issue? Or whether I made a mistake somewhere when I changed your code?

Best, Alex