51–60 of 88 posts
So it’s safe to assume in the next 10 years - AI will be running locally on every device from phones, laptops and even many embedded devices. Even robots- from street cleaning bots to helpful human assistants?
Nice to see people finding way to get the square peg though the round hole. Something that I wondered when the Raspberry Pi 5 came out is the weirdness that might be possible now that they have their own chip doing IO cleverness. On the PI 5, the two MIPI interfaces ave been enhanced to do either output or input, It made me wonder if the ports are now generalized enough that you could daisy chain a string of PI 5's c…
The cheapest way to get a 80gig setup is to to only offload some layers to GPU, and use CPU/system RAM for the rest. I run ~120GB models on my PC with a 7950X, 128GB DDR5, and RTX 3090.
Earlier quoted context omitted.
>Stop buying into hype. There isn't one model to rule them all, there are models that are better in differing contexts. I didn't say any of that, there is simply no open source GAN model that can compete with the open source diffusion models we have today, and the fact that these models can be distilled down to 1/2/4 steps makes GANs less attractive.
> I didn't say any of that I'm sorry, then I don't know what you're saying. Because what I read was 2 claims. 1) GAN quality is less than diffusion. 2) GANs can't do T2I. I think I adequately showed that both these assumptions were wrong. I'm not sure what else your comment meant as that contains all of its words... > there is simply no open source GAN model that can compete with the open source diffusion models we h…
Earlier quoted context omitted.
>Stop buying into hype. There isn't one model to rule them all, there are models that are better in differing contexts. I didn't say any of that, there is simply no open source GAN model that can compete with the open source diffusion models we have today, and the fact that these models can be distilled down to 1/2/4 steps makes GANs less attractive.
> I didn't say any of that I'm sorry, then I don't know what you're saying. Because what I read was 2 claims. 1) GAN quality is less than diffusion. 2) GANs can't do T2I. I think I adequately showed that both these assumptions were wrong. I'm not sure what else your comment meant as that contains all of its words... > there is simply no open source GAN model that can compete with the open source diffusion models we h…
Earlier quoted context omitted.
>Stop buying into hype. There isn't one model to rule them all, there are models that are better in differing contexts. I didn't say any of that, there is simply no open source GAN model that can compete with the open source diffusion models we have today, and the fact that these models can be distilled down to 1/2/4 steps makes GANs less attractive.
> I didn't say any of that I'm sorry, then I don't know what you're saying. Because what I read was 2 claims. 1) GAN quality is less than diffusion. 2) GANs can't do T2I. I think I adequately showed that both these assumptions were wrong. I'm not sure what else your comment meant as that contains all of its words... > there is simply no open source GAN model that can compete with the open source diffusion models we h…
I mean it was pretty simple, the quality is not really close, also StyleGAN2 is not conditioned on text (because you talked about this model in first comment). In the future there could be one that is competitive but not today.
GigaGAN the best GAN model by far is not open source and it's not competitive yet because even the images cherry picked for the paper and project paper do not look that coherent, the FID is relatively low because the inception v3 model used to calculate the FID doesn't care that much about global coherency and more like texturing but if the FID was calculated using DINO v2 (like some recent paper do) instead of inception v3 it will really show the gap between GAN models and diffusion models today.
Could be a nice wall paper generator every 29 min. Input a big list of random prompts and let it rotate
Earlier quoted context omitted.
It's so nice of you to offer to buy 4090 cards for people who can only otherwise afford Raspberry Pis ;)
I was just using that as a reference. Stable diffusion will run well with almost any relatively modern gpu. You don't have to use a 4090, you'll still get double digit performance with a 3060 or whatnot. > for people who can only otherwise afford Raspberry Pis ;) You can rent a 4090 for 0.7USD/1hr, or get an A100 for 1.1USD/hr. And if your project is a display + raspberry pi then those costs will dwarf the rental cos…
Earlier quoted context omitted.
But that's not the point, obviously. Sometimes, being slow is a feature. Besides, a 4090 costs more than a small car.
> But that's not the point, obviously. If you want to say the zero2-w is what's making it then sure. > Besides, a 4090 costs more than a car. They only cost ~0.70USD for 1 hr. In fact you could put this on an A100 for 1$/hr. Renting would make the most sense for this type of thing.
Earlier quoted context omitted.
> But that's not the point, obviously. If you want to say the zero2-w is what's making it then sure. > Besides, a 4090 costs more than a car. They only cost ~0.70USD for 1 hr. In fact you could put this on an A100 for 1$/hr. Renting would make the most sense for this type of thing.
It depends on what you're using the images for. If there's a human in the loop, 100 images/s is likey too much volume, especially if prompt engineering is needed. At the same time, 2 images/hr is way too slow.
Earlier quoted context omitted.
But that's not the point, obviously. Sometimes, being slow is a feature. Besides, a 4090 costs more than a small car.
$1600 is more than a car? I feel like you can't even find driveable cars that will last 100 miles at that price point anymore.
/hʌɪˈpəːbəli/
noun
exaggerated statements or claims not meant to be taken literally.