When I jumped into the AI toolkit by Ostris and there are a lot of new models that have been released. So I decided to jump back into it to train my native children (for flow) on ZImage Turbo using the toolkit.
My default setup to run it is to go Run Pod And subscribe to one RTX5090 Which is only 89 0.89 per hour which is insanely cheap so I don’t bother stressing my RTX4080 locally and running into memory issues.
Ostris has released its entire app as a Docker image that is available as a template on RunPod. Just search under Templates and Keys in Sisters, you’ll find it.
Once you have your pod running, launch the toolkit. Navigate to datasets and create a new dataset, upload your images and title them. If you need help with this check out some of my videos. Search my Comfyui workflow repositories for “caption” workflows.
Create a new task and configure it with the following settings, many of which are predefined. If you are using more than 24GB of VRAM, you can keep the same settings.
Key changes I made: turn off – low VRAM, set quantization transformer to none, set save max step to 8 (I want to have more files for testing), on – training > cached text embedding. The rest of the settings are default.
Now you are ready to start the job and let it run. With the RTX5090, I was able to train within 1 hour and 10 minutes. Checkpoints were available and I could download different versions as a set to save 8 different versions, I had that and the final version at 3000 steps.
Once you download them you can place them in the Models\Lauras folder in COMFYUI and then you can start testing different versions.
Here are the training results. Very happy with the results. I used a Laura power of 0.8-0.95.
If you would like to support our site, please consider buying us Co-Fifor , for , for , . Grab a product or Subscribe. Need a faster GPU, get access to the fastest GPUs for less than $1 per hour runpod.io