AI-toolkit is a powerful interface for your own Lora training and gives a very flexible yet modern functionality if you want to find it. In the past posts I have shared how to run this AI-Toolkit using the Ostrice A toolt on the Runpod Jointly in the depth lesson using its command line interface. But now a new web -based UI has been released by the author that provides the user’s amazing experience if you want to train your Lora. You can create more than one data set and set up that can then be used with jobs that will train Lora.
In this post we are going to consider setting up and using this new version and especially AI-TOKIT’s Web UI Run pod.
It provides an excellent clean interface that shows you to your GPU’s votals and how it is developing.
Ai-Toolkit Resources
Available on AI-toolkit Gut Hub Page And is often updated by the writer Osteress. Toll Cut Come You train LL Laura of any model but supports UI Flux 1, Flex, Wan 2.1 T2V 1.3B, Wan2.1 T2V 14B and Lumina image.
Install AI-toolkit on Run Pod
Login to your Run pod Account, make a one in free if you are new Run pod. Make a pod with Rinpad piturch 2.4.0 And any 24GB vram GPU (3090 or 4090) as the least. You can also start with ultra cheap A40 48GB VRAM.
Set up with your pod:
- Container disk on 100 GB
- Volume disk at 30 GB (or more than if you are training large Loras)
- Expose HTTP 8675 (8888 Jupyter Lab for 8888 Jupyter Lab) by adding two ports to separate two ports
Install AI-Toolkit using the command below. You can easily copy and paste it in your runpod terminal.
git clone
cd ai-toolkit
git submodule update --init --recursive
python3 -m venv venv
source venv/bin/activate
pip3 install --no-cache-dir torch==2.6.0 torchvision==0.21.0 --index-url
pip3 install -r requirements.txt
The terminal is smart and will process all the lines of the command and run it in order. It takes 5-10 minutes to complete on the basis of the pods you are using.
Launch a new terminal window and install NVM (node.js v23) Page.
curl -o- | bash
\. "$HOME/.nvm/nvm.sh"
nvm install 23If you want to confirm that NVM and NPM have been installed correctly, run commands: node -v && nvm current && npm -v
The result should print:
v23.11.0
v23.11.0
10.9.2We are now ready to run UI and launch it via Rinpad Console.
AI-toolkit UI launch
Before we open a new Terminal To the window and the converted directory /workspace/ai-toolkit/ui . Once you are in this directory, run the command npm run build_and_start
This app will create and launch the interface at Port 8675, which can now be opened via runpod.
Train your Laura on Ai-Toolkit UI
We are now ready to start training work which requires two important tasks: Make Datasate And make a The training of the training
Once you upload images with their titles (need help in giving the title – checkout These) Training is a place where you now use datasets and determine your preferences, you can go with defaults to start, but you can enter trigger choosing, training measures and samples to enter titles.
Then you can easily run a job and see magic happens.
Video tutorial
I can imagine that a lot of people can either read by reading or watching the video tutorial, in which case you are in luck because I have published a video tutorial for this post that goes into the whole process so you can believe it works.
Appreciate you passing through this post, hopefully you will thank you for seeing this article.
If you want to support our site please consider buying us FiFor, for, for,. Catch a product Or Subscribe. Fast GPU is required, access the fastest GPUS for less than $ 1 per hour runpod.io