phiresky@lemmy.world to Programmer Humor@lemmy.ml · 2 年前Hacking in 1980 vs Hacking in 2024lemmy.worldimagemessage-square117linkfedilinkarrow-up11.55Karrow-down117
arrow-up11.53Karrow-down1imageHacking in 1980 vs Hacking in 2024lemmy.worldphiresky@lemmy.world to Programmer Humor@lemmy.ml · 2 年前message-square117linkfedilink
minus-squareFierySpectre@lemmings.worldlinkfedilinkarrow-up12·2 年前Problem with that is that taking away even specific parts of the dataset can have a large impact of performance as a whole… Like when they removed NSFW from an image generator dataset and suddenly it sucked at drawing bodies in general
minus-squareRubanski@lemm.eelinkfedilinkarrow-up6·2 年前So it learns anatomy from porn but it’s not allowed to draw porn basically?
minus-squarepascal@lemm.eelinkfedilinkarrow-up6·2 年前Because porn itself doesn’t exist, it’s a by-product of biomechanics. It’s like asking a bot to draw speed, but all references to aircrafts and racecars have been removed.
Problem with that is that taking away even specific parts of the dataset can have a large impact of performance as a whole… Like when they removed NSFW from an image generator dataset and suddenly it sucked at drawing bodies in general
So it learns anatomy from porn but it’s not allowed to draw porn basically?
Because porn itself doesn’t exist, it’s a by-product of biomechanics.
It’s like asking a bot to draw speed, but all references to aircrafts and racecars have been removed.
Interesting! Nice comparison