V3.5.1

added blackwell architecture for wheels

V3.5.0

Updates :

• Change base image to Ubuntu 24.04
• Added AiToolkit (N° 73)
• TORCH_CUDA_ARCH_LIST="8.0 8.6 8.7 8.9 9.0 9.0a 10"

V3.4.0

Updates :

• Change base image to include KASM
• added Cuda 12 toolkit to support TRELLIS in comfyUI and via a custom script available here : https://github.com/grokuku/stable-diffusion-custom-scripts/blob/main/trellis.sh
• Added OneTrainer (N° 72)
• Pre-compile some modules to speed up first TRELLIS launch

V3.3.1

• FaceFusion 3.0 not launching fixed (thanks @aurel-g )

V3.3.0

General:

• Added FluxGym as option 71

V3.2.3

General:

• Improving permissions management / startup time on some systems ( thanks @aurel-g )

V3.2.2

ComfyUI:

• Moved T5 and unet to common model folder

Auto1111/Forge:

• Fixed a bug that prevent from starting

V3.2.1

Github workflow fix

V3.2.0

General :

• removed installation of some packages
• add the possibility to have a custom requirements.txt
• change the docker-compose file

ComfyUI:

• Try to install python dependencies of all custom nodes after a clean of the virtual env folder

SwarmUI:

• Changed The repo to the new one
• Renamed folders accordingly (the old one is kept to let users replicate custom changes)

IOPaint:

• Should upgrade now

V3.1.3

General :

ffmpeg is now installed by default

Automatic1111:

Fix output folder (again)
add temporary fix for controlnet

ComfyUI:

Try to install python dependencies of all custom nodes at start