Exactly. I wish people had a better understanding of what’s going on technically.
It’s not that the model itself has these biases. It’s that the instructions given them are heavy handed in trying to correct for an inversely skewed representation bias.
So the models are literally ok instructed things like “if generating a person, add a modifier to evenly represent various backgrounds like Black, South Asian…”
Here you can see that modifier being reflected back when the prompt is shared before the image.
It’s like an ethnicity AdLibs the model is being instructed to fill out whenever generating people.
Exactly. I wish people had a better understanding of what’s going on technically.
It’s not that the model itself has these biases. It’s that the instructions given them are heavy handed in trying to correct for an inversely skewed representation bias.
So the models are literally ok instructed things like “if generating a person, add a modifier to evenly represent various backgrounds like Black, South Asian…”
Here you can see that modifier being reflected back when the prompt is shared before the image.
It’s like an ethnicity AdLibs the model is being instructed to fill out whenever generating people.