LoRA text2image fine-tuning - johnowhitaker/lora_pn03_036sim

These are LoRA adaption weights for playgroundai/playground-v2-1024px-aesthetic. The weights were fine-tuned on a custom dataset.

Usage:

import torch
from diffusers import DiffusionPipeline
device = "cuda:7"
pipe = DiffusionPipeline.from_pretrained(
    'playgroundai/playground-v2-1024px-aesthetic',
    torch_dtype=torch.float16,
    add_watermarker=False,
    variant="fp16"
)
pipe.to(device)
pipe.load_lora_weights('johnowhitaker/lora_pn03_036sim')
pipe("An owl wearing a tophat and tie, studio lighting, black background").images[0]

LoRA for the text encoder was enabled: False.

Special VAE used for training: madebyollin/sdxl-vae-fp16-fix.