I give it a few years before Google makes stock images irrelevant.
Rolling this into Google Docs seems like a nobrainer.
Until they pull of the plug on it.
Imagen, a text-to-image diffusion model
71–80 of 661 posts
Re: Imagen, a text-to-image diffusion model
#72>While we leave an in-depth empirical analysis of social and cultural biases to future work, our small scale internal assessments reveal several limitations that guide our decision not to release our model at this time. Some of the reasoning: >Preliminary assessment also suggests Imagen encodes several social biases and stereotypes, including an overall bias towards generating images of people with lighter skin tones…
The big labs have become very sensitive with large model releases. It's too easy to make them generate bad PR, to the point of not releasing almost any of them. Flamingo was also a pretty great vison-language model that wasn't released, not even in a demo. PaLM is supposedly better than GPT-3 but closed off. It will probably take a year for open source models to appear.
That's because we're still bad about long-tailed data and that people outside the research don't realize that we're first prioritizing realistic images before we deal with long-tailed data (which is going to be the more generic form of bias). To be honest, it is a bit silly to focus on long-tailed data when results aren't great. That's why we see the constant pattern of getting good on a dataset and then focusing on the bias in that dataset.
I mean a good example of this is the Pulse[0][1] paper. You may remember it as the white Obama. This became a huge debate and it was pretty easily shown that the largest factor was the dataset bias. This outrage did lead to fixing FFHQ but it also sparked a huge debate with LeCun (data centric bias) and Timnit (model centric bias) at the center. Though Pulse is still remembered for this bias, not for how they responded to it. I should also note that there is human bias in this case as we have a priori knowledge of what the upsampled image should look like (humans are pretty good at this when the small image is already recognizable but this is a difficult metric to mathematically calculate).
It is fairly easy to find adversarial examples, where generative models produce biased results. It is FAR harder to fix these. Since this is known by the community but not by the public (and some community members focus on finding these holes but not fixing them) it creates outrage. Probably best for them to limit their release.
[0] https://arxiv.org/abs/2003.03808
[1] https://cdn.vox-cdn.com/thumbor/MXX-mZqWLQZW8Fdx1ilcFEHR8Wk=...
:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/66972412/face_depixelizer_obama.0.jpg){kind=link}
Re: Imagen, a text-to-image diffusion model
#73>While we leave an in-depth empirical analysis of social and cultural biases to future work, our small scale internal assessments reveal several limitations that guide our decision not to release our model at this time. Some of the reasoning: >Preliminary assessment also suggests Imagen encodes several social biases and stereotypes, including an overall bias towards generating images of people with lighter skin tones…
So glad the company that spies on me and reads my email for profit is protecting me from pictures that don't look like TV commercials.
Re: Imagen, a text-to-image diffusion model
#74Metacalculus, a mass forecasting site, has steadily brought forward the prediction date for a weakly general AI. Jaw-dropping advances like this, only increase my confidence in this prediction. "The future is now, old man." https://www.metaculus.com/questions/3479/date-weakly-general...
I don't see how this gets us (much) closer to general AI. Where is the reasoning?
Re: Imagen, a text-to-image diffusion model
#75Would be fascinated to see the DALL-E output for the same prompts as the ones used in this paper. If you've got DALL-E access and can try a few, please put links as replies!
See the paper here : https://gweb-research-imagen.appspot.com/paper.pdf
Section E : "Comparison to GLIDE and DALL-E 2"
Re: Imagen, a text-to-image diffusion model
#76>While we leave an in-depth empirical analysis of social and cultural biases to future work, our small scale internal assessments reveal several limitations that guide our decision not to release our model at this time. Some of the reasoning: >Preliminary assessment also suggests Imagen encodes several social biases and stereotypes, including an overall bias towards generating images of people with lighter skin tones…
They're withholding the API, code, and trained data because they don't want it to affect their corporate image. The good thing is they released their paper which will allow easy reproduction.
T5-XXL looks on par with CLIP so we may not see an open source version of T5 for a bit (LAION is working on reproducing CLIP), but this is all progress.
Re: Imagen, a text-to-image diffusion model
#77>While we leave an in-depth empirical analysis of social and cultural biases to future work, our small scale internal assessments reveal several limitations that guide our decision not to release our model at this time. Some of the reasoning: >Preliminary assessment also suggests Imagen encodes several social biases and stereotypes, including an overall bias towards generating images of people with lighter skin tones…
This seems bullshit to me, considering Google translate and google images encode the same biases and stereotypes, and are widely available.
Re: Imagen, a text-to-image diffusion model
#78>While we leave an in-depth empirical analysis of social and cultural biases to future work, our small scale internal assessments reveal several limitations that guide our decision not to release our model at this time. Some of the reasoning: >Preliminary assessment also suggests Imagen encodes several social biases and stereotypes, including an overall bias towards generating images of people with lighter skin tones…
If you tell it to generate an image of someone eating Koshihikari rice, will it be biased if they're Japanese? Should the skin color, clothing, setting, etc be made completely random, so that it's unbiased? What if you made it more specific, like "edo period drawing of a man"? Should the person draw be of a random skin color? What about "picture of a viking"? Is it biased if they're white? At what point is statistica…
>At what point is statistical significance considered ok and unbiased?
Presumably when you're significantly predictive of the preferred dogma, rather than reality. There's no small bit of irony in machines inadvertently creating cognitive dissonance of this sort; second order reality check.
I'm fairly sure this never actually played out well in history (bourgeois pseudoscience, deutsche physik etc), so expect some Chinese research bureau to forge ahead in this particular direction.
Re: Imagen, a text-to-image diffusion model
#79Really impressive. If we are able to generate such detailed images, is there anything similar for text to music? I would I though that it would be simpler to achieve than text to image.
Our language is much more effective at describing images than music.
Re: Imagen, a text-to-image diffusion model
#80>While we leave an in-depth empirical analysis of social and cultural biases to future work, our small scale internal assessments reveal several limitations that guide our decision not to release our model at this time. Some of the reasoning: >Preliminary assessment also suggests Imagen encodes several social biases and stereotypes, including an overall bias towards generating images of people with lighter skin tones…
This raises some really interesting questions. We certainly don't want to perpetuate harmful stereotypes. But is it a flaw that the model encodes the world as it really is, statistically, rather than as we would like it to be? By this I mean that there are more light-skinned people in the west than dark, and there are more women nurses than men, which is reflected in the model's training data. If the model only gener…
I think the statistics/representation problem is a big problem on its own, but IMO the bigger problem here is democratizing access to human-like creativity. Currently, the ability to create compelling art is only held by those with some artistic talent. With a tool like this, that restriction is gone. Everyone, no matter how uncreative, untalented, or uncommitted, can create compelling visuals, provided they can use language to describe what they want to see.
So even if we managed to create a perfect model of representation and inclusion, people could still use it to generate extremely offensive images with little effort. I think people see that as profoundly dangerous. Restricting the ability to be creative seems to be a new frontier of censorship.