export MODEL_NAME="black-forest-labs/FLUX.1-dev"
export INSTANCE_DIR="/pvol/Kamiki Rei"
export OUTPUT_DIR="/pvol/kamiki_rei_lora_flux"
accelerate config default
accelerate launch train_dreambooth_lora_flux1.py \
  --pretrained_model_name_or_path=$MODEL_NAME  \
  --mixed_precision="bf16" \
  --dataset_name=$INSTANCE_DIR \
  --output_dir=$OUTPUT_DIR \
  --gradient_checkpointing \
  --instance_prompt="Kamiki Rei" \
  --caption_column="text" \
  --resolution=1024 \
  --train_batch_size=1 \
  --guidance_scale=1 \
  --use_8bit_adam \
  --checkpointing_steps=100 \
  --gradient_accumulation_steps=4 \
  --optimizer="adamW" \
  --learning_rate=1e-4 \
  --lr_scheduler="constant" \
  --lr_warmup_steps=100 \
  --max_train_steps=1500 \
  --rank=4 \
  --seed="0"

Usage

python

import torch
from diffusers import FluxPipeline

device = "cuda:0"

pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=torch.bfloat16)

pipe.load_lora_weights("je-suis-tm/kamiki_rei_lora_flux",
                       weight_name='pytorch_lora_weights.safetensors')


prompt = "female parachutist named Kamiki Rei, jumping from plane, smile, selfie, motion, full body visible, detailed skin, realistic, photo-realistic, 8k, highly detailed, full length frame, High detail RAW color art, diffused soft lighting, shallow depth of field, sharp focus, hyperrealism, cinematic lighting"

image = pipe(
  prompt=prompt,
  generator=torch.Generator(device=device).manual_seed(42),
  num_inference_steps=50, # 28 is a good trade-off
  guidance_scale=4,
  height=1024, 
  width=1024,
).images[0]

image.save("Kamiki Rei.png")

Trigger words

You should use Kamiki Rei to trigger the image generation.

Download model

Download them in the Files & versions tab.