MakunaHatata@lemmy.ml to Lemmy Shitpost@lemmy.world · 1 year agoAI or DEI?lemmy.mlimagemessage-square103fedilinkarrow-up1329arrow-down146
arrow-up1283arrow-down1imageAI or DEI?lemmy.mlMakunaHatata@lemmy.ml to Lemmy Shitpost@lemmy.world · 1 year agomessage-square103fedilink
minus-squareMarcbmann@lemmy.worldlinkfedilinkarrow-up8arrow-down1·1 year agoI mean, I don’t think it’s an easy thing to fix. How do you eliminate bias in the training data without eliminating a substantial percentage of your training data. Which would significantly hinder performance.
minus-squarebamboo@lemmy.blahaj.zonelinkfedilinkEnglisharrow-up10·1 year agoRather than eliminating the some of the training data, you could add more training data to create an even balance.
minus-squarekromem@lemmy.worldlinkfedilinkEnglisharrow-up3·1 year agoIndeed - there’s a very good argument for using synthetic data to introduce diversity as long as you can avoid model collapse.
I mean, I don’t think it’s an easy thing to fix. How do you eliminate bias in the training data without eliminating a substantial percentage of your training data. Which would significantly hinder performance.
Rather than eliminating the some of the training data, you could add more training data to create an even balance.
Indeed - there’s a very good argument for using synthetic data to introduce diversity as long as you can avoid model collapse.