ai-dock
ComfyUI + Kohya_ss in one docker container
Similar to this open issue, I'd be very interested in a ComfyUI + Kohya_ss container. I would like to train LoRAs and use them right away, on a single cloud GPU.
Thank you.
What if a single image bundled everything, like ComfyUI, WebUI, and Kohya, with the ability to choose the default service to start using an environment variable? Users could still launch other services via the web interface after started.
@benlau @chrissy0 I think that's the direction I'm going to take the project. It was always planned, which is why I kept the storage directory outside of the main ComfyUI install directory with symlinks created.
I don't really want to shove everything into a single container by default, so I need to separate some of the management aspects into a self-contained unit to make everything work well in a docker-compose setup; There's already a lot going on in a single container - I will still build an all-in-one for the platforms that need it.
It's quite high priority for me though, because the better cloud platforms will likely support running docker containers in a more traditional and scalable way and local users will benefit as well.
I would opt for a combination of "FORGE + ComfyUI + Kohya_ss"
- Kohya_ss for training a LoRA (potentially it will be a FLUX LoRA anyway most of the times)
- FORGE for testing the LoRA model and compare different LoRA epochs with S&R X/Y/Z
- ComfyUI for really using your custom LoRA with FLUX
During a training you still could play or prepare a ComfyUI workflow while waiting, and all FLUX models could be shared between all 3 environments. THAT would be my wish from @robballantyne ;-)
