import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("etherealxx/systemy-csrmodel-cutesexyrobutts", dtype=torch.bfloat16, device_map="cuda")
prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
image = pipe(prompt).images[0]A reupload of Systemy model finetuned with Cutesexyrobutts' arts
Source: gofile(.)io/d/D1L69E
Image examples: https://imgur.com/VPNUae8
Prompt and settings examples: https://huggingface.co/etherealxx/systemy-csrmodel-cutesexyrobutts/blob/main/Prompt%20and%20settings%20example.PNG
Dreambooth settings used:
export LD_LIBRARY_PATH=/usr/lib/wsl/lib:$LD_LIBRARY_PATH
export MODEL_NAME="/home/systemy/NAI"
export OUTPUT_DIR="/mnt/d/jigsaw"
accelerate launch train_dreambooth.py \
--pretrained_model_name_or_path="$MODEL_NAME" \
--pretrained_vae_name_or_path="stabilityai/sd-vae-ft-mse" \
--output_dir="/mnt/d/acuteoutput" \
--seed=3434554 \
--resolution=512 \
--train_batch_size=1 \
--train_text_encoder \
--mixed_precision="fp16" \
--use_8bit_adam \
--gradient_accumulation_steps=1 \
--learning_rate=1e-6 \
--lr_scheduler="constant" \
--lr_warmup_steps=0 \
--sample_batch_size=6 \
--max_train_steps=100000 \
--save_interval=1500 \
--save_sample_prompt="image of SystemyTrigger girl" \
--concepts_list="concepts_list.json" \
--pad_tokens
- Downloads last month
- 17