Stable Diffusion 2.0 • Hacker News Board

Re: Stable Diffusion 2.0

#101

Post by svnt » Thu, Nov 24, 2022, 4:16 AM UTC

Jack000 wrote: post #84

echelon wrote: post #61

Earlier quoted context omitted.

> train this from scratch If you're talking about training from scratch and not fine tuning, that won't be cheap or easy to do. You need thousands upon thousands of dollars of GPU compute [1] and a gigantic data set. I trained something nowhere near the scale of Stable Diffusion on Lambda Labs, and my bill was $14,000. [1] Assuming you rent GPUs hourly, because buying the hardware outright will be prohibitively expen…

Depends on the dataset. You can probably get decent results by restricting the modality of the images (faces, cars, bedrooms etc) I trained from scratch with 4x3090 and while it’s not as good as SD it’s surprisingly better with hands.

How long did the training take on 4 3090s?

Re: Stable Diffusion 2.0

#102

Post by mugivarra69 » Thu, Nov 24, 2022, 4:16 AM UTC

Sugimot0 wrote: post #97

Amy word on AMD support?

just use rocm

Re: Stable Diffusion 2.0

#103

Post by mugivarra69 » Thu, Nov 24, 2022, 4:16 AM UTC

1024core wrote: post #79

What's the bare minimum hardware required to generate images with this model? Can I do something with a 8GB 980? Probably not..? What about CPU only?

not that gpu, cpu yes need to wait longer however.

Re: Stable Diffusion 2.0

#104

Post by mugivarra69 » Thu, Nov 24, 2022, 4:17 AM UTC

WatchDog wrote: post #31

Are the GPU memory requirements different for this release? Is now it possible to generate higher resolution images with less memory?

dont thinkso, ran on my macbook pro, same footprint. idk cuda side.

Re: Stable Diffusion 2.0

#105

Post by techsin101 » Thu, Nov 24, 2022, 4:18 AM UTC

is there a tool that uses to create UX/UI designs. It'd be really cool to get inspirations...

Re: Stable Diffusion 2.0

#106

Post by techsin101 » Thu, Nov 24, 2022, 4:20 AM UTC

myaccount9786 wrote: post #94

rogers18445 wrote: post #3

They apparently tried to combat NSFW generation by filtering the training dataset not to include any.

The easiest way to combat this is to put your model behind an API and filter queries (midjourney, OpenAI) or just not make it available (Google). The tradeoff is that you're paying for everyone's compute. I guess SD is betting on saving $ on compute being more important in this space than the ability to gatekeep certain queries. And the tradeoff is that you need to do nsfw filtering in your released model. It will be…

that would suck immensely for various downsides

Re: Stable Diffusion 2.0

#107

Post by tamersalama » Thu, Nov 24, 2022, 4:20 AM UTC

LASR wrote: post #27

I am a solo dev working on a creative content creation app to leverage the latest developments in AI. Demoing even the v1 of stable diffusion to the non-technical general users blows them away completely. Now that v2 is here, it’s clear we’re not able to keep pace in developing products to take advantage of it. The general public still is blown away by autosuggest in mobile OS keyboards. Very few really know how far…

I agree that this a big wave, but I'm still struggling to find commercial (read: large organizations) applications.

Re: Stable Diffusion 2.0

#108

Post by myaccount9786 » Thu, Nov 24, 2022, 4:21 AM UTC

schoen wrote: post #92

myaccount9786 wrote: post #89

Earlier quoted context omitted.

You can download the model weights and run them offline. At least, you could in v1.4. I assume this is still possible on v2.0?

Right, but the model weights are arguably not the "source code", and the license gives the users fewer rights than open source licenses do. https://en.wikipedia.org/wiki/The_Open_Source_Definition

I think the controls in this space are such a shit show right now that being "open model" is practically equivalent to a WTFPL.

If you're trying to build an app based on SD, then not being open source matters. But seems like the majority of use cases are just "I want to run the model locally". And at that point HF can't stop me from just ripping the Wi-Fi card out of my computer.

Re: Stable Diffusion 2.0

#109

Post by syockit » Thu, Nov 24, 2022, 4:22 AM UTC

throwup wrote: post #20

satvikpendem wrote: post #12

Earlier quoted context omitted.

They reportedly did so to stop people from generating CSAM [0]. [0] https://old.reddit.com/r/StableDiffusion/comments/y9ga5s/sta...

Do pixels have human rights now?

They don't. The training dataset though, may have been obtained through human rights violation. The problem is when the novelty starts to wear out. Then they will start to look for fresh training data which may again incur more human rights violation. If you can ensure that no new training data are obtained that way, then I guess it's okay? (Personally, I don't condone it)

Re: Stable Diffusion 2.0

#110

Post by SXX » Thu, Nov 24, 2022, 4:22 AM UTC

kmeisthax wrote: post #45

minimaxir wrote: post #9

Earlier quoted context omitted.

In practice, it's unclear how well avoiding training on NSFW images will work: the original LAION-400M dataset used for both SD versions did filter out some of the NSFW stuff, and it appears SD 2.0 filters out a bit more. The use of OpenCLIP in SD 2.0 may also prevent some leakage of NSFW textual concepts compared to OpenAI's CLIP. It will, however, definitely not affect the more-common use case of anime women with v…

The main reason why Stable Diffusion is worried about NSFW is that people will use it to generate disgusting amounts of CSAM. If LAION-5B or OpenAI's CLIP have ever seen CSAM - and given how these datasets are literally just scraped off the Internet, they have - then they're technically distributing it. Imagine the "AI is just copying bits of other people's art" argument, except instead of statutory damages of up to…

> If LAION-5B or OpenAI's CLIP have ever seen CSAM

Diffusion model dont need any CSAM in training dataset to generate CSAM. All it's need is any random NSFW content alongside with any safe content that includes children.