ds-LoRA / README.md
Doctor-Shotgun's picture
Update README.md
231042c
|
raw
history blame
2.92 kB
metadata
license: creativeml-openrail-m
pipeline_tag: text-to-image

Doctor Shotgun's LoRAs

This repository will serve as a location for uploading my personal LoRAs and Dreambooth models.

Model Description and Usage

What is a LoRA? https://github.com/cloneofsimo/lora

e1-e5 represent training length in number of epochs. You may need to lower the network weight on further trained LoRAs if the effects are too strong.

"legacy" LoRAs are approximations extracted from model diffs from my old Dreambooth models, included for completeness.

Automatic1111's WebUI now has native LoRA support. Install them in .\stable-diffusion-webui\models\Lora\ and activate them by prompting <lora:name_of_your_lora:network_weight>

If you are on an older version, you will need to use the additional networks extension to load LoRAs: https://github.com/kohya-ss/sd-webui-additional-networks

Training Details

LoRAs trained locally using bmaltais GUI for kohya_ss sd-scripts:

https://github.com/bmaltais/kohya_ss

https://github.com/kohya-ss/sd-scripts

Using ACertainty as a base model:

https://huggingface.co/JosephusCheung/ACertainty

Table of Contents

Persona 5

Ann Takamaki

Instance Token: dsann

”ann01” ”ann02”

Futaba Sakura

Instance Token: dsfutaba

”futaba01” ”futaba02”

Haru Okumura

Instance Token: dsharu

”haru01” ”haru02”

Kasumi Yoshizawa

Instance Token: dskasumi

”kasumi01” ”kasumi02”

Makoto Niijima

Instance Token: dsmakoto

”makoto01” ”makoto02”