sexy_peach@feddit.de to Technology@beehaw.orgEnglish · 1 year agoethinically ambigausfeddit.deimagemessage-square40fedilinkarrow-up1109arrow-down10cross-posted to: technology@lemmy.ml
arrow-up1109arrow-down1imageethinically ambigausfeddit.desexy_peach@feddit.de to Technology@beehaw.orgEnglish · 1 year agomessage-square40fedilinkcross-posted to: technology@lemmy.ml
minus-squareKichae@lemmy.calinkfedilinkarrow-up7·1 year agoThe smarter way would be using balanced training data.
minus-squareFizz@lemmy.nzlinkfedilinkarrow-up1·1 year agoYou can’t balance every single aspect of the training data. You will always run into some searches favoring one race over another.
The smarter way would be using balanced training data.
You can’t balance every single aspect of the training data. You will always run into some searches favoring one race over another.