LoRA DreamBooth model
These are LoRA adaption weights for kpsss34/Stable-Diffusion-3.5-Small-Preview1. The weights were trained on the concept of a photo of m4k1m4 anime girl
using DreamBooth.
Training details
- Base model: kpsss34/Stable-Diffusion-3.5-Small-Preview1
- Instance prompt: a photo of m4k1m4 anime girl
- Validation prompt: a photo of m4k1m4 anime girl stands in the living room.
- LoRA rank: 32
- Learning rate: 0.0001
- Training steps: 500
- Resolution: 1024
- Train text encoder: True
- Datasets: 3 jpg and 3 txt (without 1:1)
- VRAM USED: 14-15GB (Validation process+2-4GB)
Usage
import torch
from pipeline_stable_diffusion_3_S import StableDiffusion3SPipeline
model_id = "kpsss34/Stable-Diffusion-3.5-Small-Preview1"
pipe = StableDiffusion3SPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16)
pipe.load_lora_weights("kpsss34/SD35s_lora_test")
pipe.to("cuda")
# Now you can use the pipeline with the trained LoRA
image = pipe("a photo of m4k1m4 anime girl ", num_inference_steps=30, guidance_scale=5.0).images[0]
image.save("result.png")
- Downloads last month
- 18
Model tree for kpsss34/SD35s_anime-Girl_lora_test1
Base model
kpsss34/Stable-Diffusion-3.5-Small-Preview1