taming-transformers VS stable-diffusion

taming-transformers

What’s your goal exactly? Are you trying to make a transformer based auto encoder of audio spectrograms? If so you should either start with either a proven ViT-based AE implementation (either a VAE or a VQ-GAN). But I don’t see why you necessarily need a ViT for this, if you’re working at a much smaller scale a convolutional architecture is plenty and much more amenable to beginners. See https://github.com/CompVis/taming-transformers for an example of a convolutional VQ GAN.

Command: "git" clone "https://github.com/CompVis/taming-transformers.git" "C:\AI\stable-diffusion-webui\repositories\taming-transformers"

See https://arxiv.org/abs/2012.09841 for prior work. SD authors swap out the Transformer and language modelling objective with a UNet diffusion objective. In general, the more inductive bias your model has, the more efficient it can be. ChatGPT runs purely on a Transformer architecture, which has far fewer priors than a CNN and requires far more parameters as a result. This may not be the case in the future.

There is definitely a cmd but I can't tell you. It's on GitHub https://github.com/CompVis/taming-transformers

I wasn’t able to find a “taming” folder within the site-packages folder so I decided to look up how to get VectorQuantizer2 and taming.modules.vqvae.quantize and found this link: https://github.com/CompVis/taming-transformers/blob/master/taming/modules/vqvae/quantize.py I copied the raw contents and pasted that to the terminal and I got this error: “zsh: parse error near `)’” I’m not sure how to fix this so I can install VectorQuantizer2 so I can use InvokeAI. How do I fix this?

I say this to everyone... Even if SD and the model is legit and legal. Do not go around commercialising it's outputs or claiming ownership over them... and if you do the properly cite the source of the model and system along with it. In https://github.com/CompVis/stable-diffusion, https://github.com/CompVis/taming-transformers and https://huggingface.co/CompVis/stable-diffusion-v1-4 there are citiations provided for you to use for a reason. I recommend you to use them.

# Copy models as described in README cp ~/Downloads/model.ckpt . cp ~/Downloads/GFPGANv1.3.pth . # Clone other repos as mentioned in README mkdir repositories git clone https://github.com/CompVis/stable-diffusion.git repositories/stable-diffusion git clone https://github.com/CompVis/taming-transformers.git repositories/taming-transformers git clone https://github.com/sczhou/CodeFormer.git repositories/CodeFormer git clone https://github.com/salesforce/BLIP.git repositories/BLIP export NIXPKGS_ALLOW_UNFREE=1 nix-shell default.nix pip install torch --extra-index-url https://download.pytorch.org/whl/cu113 # Also from linux instructions. Can probably be added to default.nix python webui.py

Code for https://arxiv.org/abs/2012.09841 found: https://compvis.github.io/taming-transformers/

Which is a 1-click install of Stable Diffusion with an alternative web interface. You can choose a different approach but this one is pretty simple and I am new to this stuff.

an invisible watermarking of the outputs, to help viewers identify the images as machine-generated.

It sounds like you just want to set up Stable Diffusion to run locally. I don't think your computer's specs will be able to do it. You need a graphics card with a decent amount of VRAM. Stable diffusion is in Python as is almost every AI open source project I've seen. If you can get your hands on a system with an Nvidia RTX card with as much VRAM as possible, you're in business. I have an RTX 3060 with 12 gigs of VRAM and I can run stable diffusion and a whole variety of open source LLMs as well as other projects like face swap, Roop, tortoise TTS, sadtalker, etc...

Use specific GPU on multi GPU systems · Issue #87 · CompVis/stable-diffusion · GitHub

I dealt with this in one of my posts. At least SD 1.1 till 1.5 are all trained on a batch size of 2048. The version pretty much everyone uses (1.5) is first pretrained at a resolution of 256x256 for 237K steps on laion2B-en, at the end of those training steps it will have seen roughly 500M images in laion2B-en. After that it is pre-trained for 194K steps on laion-high-resolution at a resolution of 512x512, which is a subset of 170M images from laion5B. Finally it is trained for 1.110K steps on LAION aesthetic v2 5+. This is easily verified by taking a glance at the model card of SD 1.5. Though that one doesn't specify for part of the training exactly which aesthetic set was used for part of the training, for that you have to look at the CompVis github repo. Thus at the end of it all both the most recent images and the majority of images will have come from LAION aesthetic v2 5+ (seeing every image approx 4 times). Realistically a lot of the weights obtained from pretraining on 2B will have been lost, and only provided a good starting point for the weights.

stable diffusion · CompVis/stable-diffusion@2ff270f · GitHub

Here for example --> https://github.com/CompVis/stable-diffusion