santa hat
deerdeer nosedeer glow
Sign In

what is the difference between kohya trainers and what do you like to use?

does anyone know the difference between kohya trainers

-Kohya LoRA Dreambooth LoRA Training (Dreambooth method)

-Kohya LoRA Fine-Tuning LoRA Training (Fine-tune method)

-Kohya Trainer Native Training

and what do you use?

2 Answers

  1. Not sure about differences between "native" LoRA training and Dreambooth. Both methods requires the same things and has the same settings.

  2. Fine-tuning method is about training your LoRA into model and then extracting it (of course, if you don't want to see file being 2-7Gb in size) with add difference. The best way to do LoRAs, i believe.

  3. I'm using different trainer (of course, based on Kohya) which called LoRA Easy Training Scripts, with its own interface. But you can only run it locally. Not in Colab. https://github.com/derrian-distro/LoRA_Easy_Training_Scripts It's really easy to use. Uses Dreambooth training method, as i know.

I use this repo: https://github.com/bmaltais/kohya_ss - locally to run SD1.5 based LoRAs and on runpod for SDXL LoRAs

Your answer