FLUX.1 schnell PVC style
Model description
PVC figure style LoRA for FLUX.1-schnell, trained on FLUX.1-schnell with the training adapter using ostris/ai-toolkit.
Trigger words
You should use pvc figure
to trigger the image generation.
You should use nendoroid
to trigger the image generation.
You should use figma
to trigger the image generation.
Prompt guide
This LoRA was trained with danbooru tags, but maybe not of all danbooru tags will work. Please consider to use natural language to align.
Download model
Weights for this model are available in Safetensors format.
Download them in the Files & versions tab.
Training config
WandB log: https://wandb.ai/p1atdev/flux_lora_pvc_schnell_1
TL;DR:
learning_rate
1e-4
forconstant
withAdamW8bit
(default)- used about 2500 images
batch_size
4
,gradient_accumulation_steps
1
- took about 32GB VRAM
- LoRA config (default)
linear
:16
linear_alpha
:16
- base model: black-forest-labs/FLUX.1-schnell
- assistant lora: ostris/FLUX.1-schnell-training-adapter
- trained for 10,000 steps but finally used the merge of checkpoint at 7250 and 7500 for 1:1.
- GPU: A6000 x1 (VRAM 48GB)
Full config.yaml
job: extension
config:
name: flux_lora_pvc_schnell_1
process:
- type: sd_trainer
training_folder: output
device: cuda:3
network:
type: lora
linear: 16
linear_alpha: 16
save:
dtype: bfloat16
save_every: 250
max_step_saves_to_keep: 10
datasets:
- folder_path: /workspace/ai-toolkit/dataset/pvc
caption_ext: txt
caption_dropout_rate: 0.01
shuffle_tokens: true
cache_latents_to_disk: true
resolution:
- 768
- 1024
train:
batch_size: 4
steps: 10000
gradient_accumulation_steps: 1
train_unet: true
train_text_encoder: false
gradient_checkpointing: true
noise_scheduler: flowmatch
optimizer: adamw8bit
lr: 0.0001
ema_config:
use_ema: true
ema_decay: 0.99
dtype: bf16
model:
name_or_path: black-forest-labs/FLUX.1-schnell
assistant_lora_path: ostris/FLUX.1-schnell-training-adapter
is_flux: true
quantize: true
sample:
sampler: flowmatch
sample_every: 250
width: 832
height: 1152
prompts:
- 1girl, solo, blue hair, cat ears, parted bangs, long hair, looking at viewer,
white dress shirt, rainy, wariza, sitting
- 1girl, solo, full body, gloves, groin, hair between eyes, hair ornament, head
tilt, holding, holding staff, horns, long hair, looking at viewer, maid headdress,
mole, mole under eye, navel, parted lips, purple eyes, purple hair, purple
theme, shawl, shorts, single horn, staff, standing, thighhighs, twintails,
white footwear, white gloves, white thighhighs
- 1girl, aqua eyes, baseball cap, blonde hair, closed mouth, earrings, green
background, hat, hoop earrings, jewelry, looking at viewer, shirt, short hair,
solo, upper body, yellow shirt,
- 1girl, bangs, bare shoulders, beret, black hair, black shorts, blue hair,
bracelet, breasts, buttons, colored inner hair, double-breasted, eyewear removed,
green headwear, green jacket, grey eyes, grey sky, hat, jacket, jewelry, long
hair, looking at viewer, multicolored hair, neck ring, o-ring, off shoulder,
rain, round eyewear, shorts, sidelocks, small breasts, solo, sunglasses, wavy
hair, wet, zipper,
- 1girl, brown hair, green eyes, colorful, autumn, cumulonimbus clouds, lighting,
blue sky, falling leaves, garden
- 1girl, bangs, black wings, blush, choker, collarbone, feathered wings, hair
between eyes, halo, hand on hip, hat, head wings, long hair, long sleeves,
looking at viewer, pink eyes, pink hair, pleated skirt, school hat, school
uniform, serafuku, sidelocks, simple background, skirt, solo, twintails, wavy
mouth, white background, wings, waifu,
neg: ''
seed: 42
walk_seed: true
guidance_scale: 1
sample_steps: 4
logging:
use_wandb: true
project_name: flux_lora_pvc_schnell_1
run_name: run-1
meta:
name: flux_lora_schnell_pvc
version: '1.0'
- Downloads last month
- 390
Model tree for p1atdev/flux.1-schnell-pvc-style-lora
Base model
black-forest-labs/FLUX.1-schnell