Notes:
Don't use 8 bit Optimizers. You can use any non 8 bit Optimizer.
From Paramaters -> Advanced, set CrossAttention to SDPA.
No xFormers support for now.
Only BF16 and FP32 are supported. FP16 support for training isn't complete yet.
No Bitsandbytes support for now but Full BF16 Training is supported.
Errors with iGPU:
Disable your iGPU (if any, e.g. UHD or Iris Xe) in the device manager.
Update your kernel to at least Linux 6.2 if you are on older Ubuntu builds.
Note: Updating kernel is not neccesary for WSL.
Then add the package lists for Intel OneAPI:
Then install the necessary packages:
You can use Yay to install all of them at once:
Install Kohya SS from Github:
Then enter into the Kohya SS folder:
Then create a virtual environment for Python and install the necessary libraries:
Finally start Kohya SS with this command:
Parameter Settings:
Set Mixed Precision Type to BF16.Don't use 8 bit Optimizers. You can use any non 8 bit Optimizer.
From Paramaters -> Advanced, set CrossAttention to SDPA.
No xFormers support for now.
Only BF16 and FP32 are supported. FP16 support for training isn't complete yet.
No Bitsandbytes support for now but Full BF16 Training is supported.
Errors with iGPU:
Disable your iGPU (if any, e.g. UHD or Iris Xe) in the device manager.
Setup WSL on Windows:
Follow these instructions to setup Linux environment in Windows, then continue with Ubuntu instructions.- Enter "Windows features" in Windows search bar and select "Turn Windows features on or off".
- Enable both "Virtual Machine Platform" and "Windows Subsystem for Linux" and click OK.
- Restart your computer once update is complete.
- Open PowerShell and execute
wsl --update
. - Download Ubuntu 22.04 from Windows Store.
- Start Ubuntu 22.04 and finish user setup.
Install Intel Compute Runtime:
Ubuntu:
Use Ubuntu 23.04 or newer because we will need Linux 6.2 kernel or newer.Update your kernel to at least Linux 6.2 if you are on older Ubuntu builds.
Note: Updating kernel is not neccesary for WSL.
Then add the package lists for Intel OneAPI:
Bash:
sudo apt update && sudo apt install -y ca-certificates wget gpg
Bash:
wget -qO - https://repositories.intel.com/gpu/intel-graphics.key | sudo gpg --dearmor --output /usr/share/keyrings/intel-graphics.gpg
Bash:
echo "deb [arch=amd64,i386 signed-by=/usr/share/keyrings/intel-graphics.gpg] https://repositories.intel.com/gpu/ubuntu jammy client" | sudo tee /etc/apt/sources.list.d/intel-gpu-jammy.list
Bash:
sudo apt update && sudo apt upgrade -y
Then install the necessary packages:
Bash:
sudo apt install intel-opencl-icd intel-level-zero-gpu level-zero git python3-pip python3-venv libgl1 libglib2.0-0 libgomp1 google-perftools
Arch Linux:
Installintel-compute-runtime level-zero-headers level-zero-loader git python-pip python-virtualenv
from the official package lists and install python310
from AUR.You can use Yay to install all of them at once:
Bash:
yay -S python310 intel-compute-runtime level-zero-headers level-zero-loader git python-pip python-virtualenv
Installing Kohya SS:
Install Kohya SS from Github:
Bash:
git clone https://github.com/bmaltais/kohya_ss
Then enter into the Kohya SS folder:
Bash:
cd kohya_ss
Then create a virtual environment for Python and install the necessary libraries:
Bash:
./setup.sh --use-ipex
Finally start Kohya SS with this command:
Bash:
./gui.sh --use-ipex
Son düzenleme: