Stable Diffusion + Kohya_ss + Jupyterlab Torch 2.1.2 and Nvidia CUDA 11.8

Packages included

• Ubuntu 22.04 LTS
• CUDA 11.8
• Torch 2.1.2
• Xformers 0.0.23
• Python 3.10.12
• JupyterLab 4.2.0
• A1111 Stable Diffusion WebUI 1.9.3
• ControlNet extension v1.1.445
• Inpaint Anything extension
• Infinite Image Browsing extension
• Kohya_ss v24.1.4
• sd_xl_base_1.0.safetensors
• sdxl_vae.safetensors
• cyberrealistic_v42.safetensors
• vae-ft-mse-840000-ema-pruned.safetensors

Running Locally

docker run -d \
  --gpus all \
  -v /workspace \
  -p 3000:3001 \
  -p 3010:3011 \
  -p 6006:6066 \
  -p 8000:8000 \
  -p 8888:8888 \
  -e JUPYTER_PASSWORD="Put your password here" \
  -e ENABLE_TENSORBOARD=1 \
  darkovisual/darko-diffusion:1.0.5

Ports

Connect Port	Internal Port	Description
3000	3001	A1111 Stable Diffusion Web UI
3010	3011	Kohya_ss
6006	6066	Tensorboard
8000	8000	App Manager
8888	8888	Jupyter Lab

Environment Variables

Variables	Description	Default
VENV_PATH	Set the path for the Python venv for the app	/workspace/venvs/stable-diffusion
JUPYTER_LAB_PASSWORD	Set a password for Jupyter lab	not set - no password
DISABLE_AUTOLAUNCH	Disable Web UIs from launching automatically	enabled
ENABLE_TENSORBOARD	Enables Tensorboard on port 6006	enabled

Logs

Stable Diffusion, Kohya_ss and Jupyter Lab each create log files, and you can tail the log files instead of killing the services to view the logs

Application	Log file
Stable Diffusion Web UI	/workspace/logs/webui.log
Kohya SS	/workspace/logs/kohya_ss.log
JupyterLab	/workspace/logs/jupyter.log