Spent 6 hours last Sunday chasing a bug that turned out to be one wrong setting
I was building a small image classifier for my nephew's school project and the accuracy kept sitting at 50%, so I tried three different model sizes, two datasets, and even reinstalled Python twice. Turned out the labels were shuffled during the train/test split, and fixing that one line took 30 seconds after 6 hours of yelling at my screen. Has anyone else lost a whole day to something this dumb?
The "reinstalled Python twice" part got me lol. Been there way too many times. The worst ones are always the tiny things too, like a swapped argument or an off by one in a loop. You spend hours convinced it's something deep and then it's just one dumb line staring at you the whole time.