SD 1.5 dreambooth models. Checkpoint files trained on ~(number of images times 90) steps; images scraped from boorus, with text encoder trained for around (number of images times 12) steps.

Textual inversion embedding can be found here (trained off the same or similar datasets).

Lora can be found here

m2_step_2000.ckpt (~11/2022)

• Fine-tuned on either NAI or Anythingv3
• Keyword: wjkhlhj, monika (doki doki literature club)
• Second attempt's best checkpoint at 2000 steps, can be pretty hard to use
• 26 training images

m2-2340.ckpt (~11/2022)

• Fine-tuned on either NAI or Anythingv3
• Keyword: wjkhlhj, monika (doki doki literature club)
• Second attempt, can be pretty hard to use and may overfit to lovecraftian entities
• 26 training images

tm14.ckpt (~early 2023)

• Fine-tuned on either NAI or Anythingv3
• Keyword: wjkhlhj, monika (doki doki literature club)
• ~26 training images

mo08092023.ckpt

• Fine-tuned on Silicon29
• Keyword: monika (doki doki literature club)
• ~17 to 20 training images

mo08092023a.ckpt

• Fine-tuned on Silicon29 with tweaked parameters
• Keyword: monika (doki doki literature club)
• 17 training images

mo08092023c.ckpt

• Fine-tuned on BreakDomain Anime
• Keyword: wjkhlhj, monika (doki doki literature club)
• 17 training images

Previews here

Released in celebration of 09222023!