Yaoi Diffusion V3

Hola a todos!!!

This is a 768 resolution model finetuned on yaoi, bara, furry, s...., s.... c.., fine arts and reallife males, in short a general homoerotic model.
Here the list of tags can recognize for version 3

https://gist.github.com/iszotic/0ccac5c804e9587a323fffd4cbbd6c03

HOW TO PROMPT:

[short description] as [character] sourcing [copyright] by [artist], [e621/gelbooru tags separated by comma and space].

positive tags: by _alter

negative tags: simple background

ex:
An anthro furry dragon male laying on bed by artist_tag, 1male, male focus, solo, pectorals, penis, realistic

Also check out the demo images with the prompts used, the high resolution was achieved using high res fix

_other, is another style, normaly is the cellshading version or a simplified style.

_alter, is alternative style, nomarly is the softshading version

Use 1male instead of 1boy, 2males for 2boys, and so on, a boy is a young male, a male can be a cub, an adult man, old man, young man, animal, heck even a flower, etc... is more genera

MIXING STYLES

there are 2 ways to mix styles:

1) making the diffusion process exchange the artist tag in each step (auto1111)

by [artist1|artist2|artist3|artist4]

2) or using the tags all at the same time:

by artist 4 by artist 3 by artist 2 by artist 1

in 1) the first artist takes the lead, the features will resemble more of this artist, although the mixture is more noticeable. Works with Euler a, Euler and DDIM samplers the downside is the quality is not good.
in 2) the last artist takes the lead , the features will resemble more of this artist, the mixture sometimes is not effective. Works with any sampler, the quality is better.

COMERCIAL USE: it's ok as long as the resulting style doesn't resemble any specific style.

Features:

• over 1300+ artists tags of homoerotic artists (including myself, lol), no tags where used for pure 3D artists.

• NSFW and SFW

• some artists support more than one style

Training details:

• Trained from SD1.5 vanilla + vae-ft-mse-840000-ema-pruned.ckpt

  https://huggingface.co/runwayml/stable-diffusion-v1-5

  https://huggingface.co/stabilityai/sd-vae-ft-mse-original

• Dataset of 260K, epoch size of 140K, rated dataset of 4K, dropout probability of rated dataset 0.5, dropout probability of not rated dataset from 0.0 to 0.15, depending of aesthetic values from https://github.com/LAION-AI/aesthetic-predictor and https://huggingface.co/cafeai/cafe_aesthetic

• Images were sourced from booru sites, and tags were sorted using deepdanbooru, the e621 model comes from zach and the wd14 swing model, if images were not from booru sites the tags were predicted, the order of the tags were randomized 5% of the times. Also used blip2-opt-6.7b

  https://github.com/toriato/stable-diffusion-webui-wd14-tagger

  https://huggingface.co/Salesforce/blip2-opt-6.7b

• Used Everydreamertrainer2, gradient checkpointing disabled, and gradient accumulations.

• 1% of the dataset was used for validation

• Training schedule: (Oh boy)
  at 512:
  Epoch 1-16, eff_batch_size: 120(12x10), lr: 4e-6, ema: 0.9995

  Epoch 17-19, eff_batch_size: 60(12x5), lr: 2e-6, ema: 0.9997

  Epoch 20-40, eff_batch_size: 12(12x1), lr: 5e-7, ema: 0.9999

  at 768:

  Epoch 40-51, eff_batch_size: 64(4x16), txt_lr: 1e-6, unet_lr = 2e-6, ema: 0.9997

  Epoch 52-72, eff_batch_size: 12(4x3), txt_lr: 3e-7, unet_lr = 6e-7, ema: 0.9999

• Only a maximum of 125 images per artist were used in each epoch, if an artist had 500 only a different set of 125 was used.

• zero frequency noise ratio = 0.02
  

Postdata:
Maybe the last model I will finetune for SD 1.5