undefined | Better HN

0 pointscoffeebeqn2y ago0 comments

Wouldn’t reinforcement learning just weigh any nonsense data very low and then spammy garbage doesn’t really affect the model in the end much ? If the model and human experts can’t tell the difference then it’s probably pretty good AI generated data

0 comments

__loam2y ago

Truth and what humans think is true are different things. Synthetic data was created by models that were trained to be convincing.

catchnear43212y ago

the ideal poison tastes like nothing, or at the very least doesn’t taste bad.

you wouldn’t want to alert the victim.

j / k navigate · click thread line to collapse

0 pointscoffeebeqn2y ago0 comments

0 comments

__loam2y ago

Truth and what humans think is true are different things. Synthetic data was created by models that were trained to be convincing.

catchnear43212y ago

the ideal poison tastes like nothing, or at the very least doesn’t taste bad.

you wouldn’t want to alert the victim.

j / k navigate · click thread line to collapse