Train a Lora Model using KohyaSS & Stable Diffusion in one Runpod! | Updated Tutorial

This video shows you how to train a Lora model using KohyaSS on RunPod.io!
Don’t forget to like, subscribe, and unlock the boundless possibilities of Artificial Intelligence together. Let’s dive in! :rocket:
Click this link to see our previous video on how to install the Kohya, Stable Diffusion in one Runpod!
• Install Stable Diffusi...
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Mentioned Links :
- GitHub Page by Ashley
github.com/bmaltais/kohya_ss
- Stable Diffusion SD XL Model
huggingface.co/stabilityai/st...
- Stable Diffusion SD XL VAE
huggingface.co/stabilityai/sd...
The Extra Argument for LORA Training :
scale_parameter=False relative_step=False warmup_init=False
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Join our Antalpha.ai Community below!
🌐 Website : antalpha.ai/
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
#ai #artificialintelligence #technology #antalpha #information #informative #documentary #introduction #aiimagegenerator #aiimages #aiimagegeneration #aiimage #image #digitalart #digital #prompt #promptengineering #antalpha.io #antalphaio #aigc #aicontentgenerator #aicontent #aicontentcreation #stablediffusion #kohya #KohyaSS #science #machinelearning #deeplearning #neuralnetworks #impact #gpt #stablediffusionanimation #stable #diffusion #imagetraining #aitraining
Credits to Sites:
Runpod.io
Civitai.com
Huggingface.co
Github.com
Fair-Use Policy :
This video adheres to fair use principles under copyright laws, allowing limited use of copyrighted material without explicit permission. It serves commentary, criticism, educational, and transformative purposes. Copyrighted content like images, graphics, and audio are used to add value, encourage discussion, and foster understanding. Proper attribution is always provided to respect copyright owners' rights. The use of copyrighted content does not harm the original work's commercial value or imply endorsement. If you believe this video violates fair use or infringes your copyright, kindly contact us with specific details, and we will promptly address your concerns.

[@ksottam]

This was so incredibly helpful. I had really struggled to get kohya running on runpod and thanks to you it went smoothly. Thank you for doing this!

[@JefHarrisnation]

Thanks for the video. This helped me get past a few puzzles I was having on runpod. Sadly training didn't run, for a reason I haven't figured out yet.

[@js801]

Amazing !!!!

[@ParvathyKapoor]

Thank you very much

[@Narulopo]

Thanks a lot 😻💗

[@adriantang5811]

Great sharing 👍🎉

[@AntalphaAI]

Thank you! ☺️

[@emresainkaplan9341]

Thanks. It's the best explanation so far. Can you please explain the runpod - SDXL dreambooth train through Kohya?

[@AntalphaAI]

Thank you! :D Yes, it will be on our pipeline for next uploads. Stay tune!

[@lilillllii246]

If I try to make a cloth lora, should I do the same as in the video? Is there any difference?

[@WhySoBroke]

Thanks for the video... but don’t you need to stop the SD as well to optimize and make it faster?

[@AntalphaAI]

As long as you don't generate images using SD at the same time, it doesn't make any big differences from our usage experience

[@luke5846]

i put all the data base like you show in the video, but i got error when i try to train. then i try again using dreambooth tab and it work just fine. am i missing something here?

[@AntalphaAI]

if the error said "CUDA Out of Memory Error", you can go back to the parameter tab in LORA and checklist the Gradient Checkpointing then retry the training again.

[@geovannichavez6922]

Parameters at 6:17

[@thedoctor5478]

good. goood...

[@Krebzonide]

I have problem where 3010 port is always not ready even after 30 minutes. If I click 3000 port I can use A1111 fine but I cannot access kohya

[@leorxt2504]

because she didn't show the template that she used. It's under the community template section : Stable Diffusion Kohya_ss ComfyUI Ultimate ashleykza/stable-diffusion-webui:3.10.1

[@gmonsoon]

no need to do crop image dataset to square, since Kohya Script do aspect ratio bucketing

[@AntalphaAI]

yes, additionally the birme.net can help user to determine the crop area of the pictures.

[@AntalphaAI]

Hi Antz! Have you tried to train the SD XL LORA using our tutorial? Let us know your results and findings in the comment down below! If you have more questions to discuss, let's share and talk in our Discord channel Antalpha.io and connect with many worldwide Antz! 🤓 Discord Server : discord.gg/vSj4V9WAVK