Text-to-Image
Diffusers
template:sd-lora
flux
lora
saq1b_halftone / config.yaml
rvorias's picture
Upload folder using huggingface_hub
1b8265b verified
config:
name: halftone
process:
- datasets:
- cache_latents_to_disk: true
caption_dropout_rate: 0.2
caption_ext: txt
folder_path: /root/lorahub/halftone/dataset
resolution:
- 512
- 768
- 1024
shuffle_tokens: false
token_dropout_rate: 0.01
device: cuda:0
model:
is_flux: true
name_or_path: black-forest-labs/FLUX.1-dev
quantize: true
text_encoder_bits: 8
network:
linear: 42
linear_alpha: 42
transformer_only: true
type: lora
performance_log_every: 500
sample:
guidance_scale: 3.5
height: 1024
neg: ''
prompts:
- '[trigger], close-up of a human eye'
- '[trigger], red car poster'
- '[trigger], an illustration of a mouth with tongue out'
sample_every: 500
sample_steps: 25
sampler: flowmatch
seed: 42
walk_seed: true
width: 1024
save:
dtype: float16
max_step_saves_to_keep: 3
save_every: 500
save_format: diffusers
train:
batch_size: 1
dtype: bf16
ema_config:
ema_decay: 0.99
use_ema: true
gradient_accumulation_steps: 1
gradient_checkpointing: true
linear_timesteps: true
loss_type: mse
lr: 0.0002
noise_scheduler: flowmatch
optimizer: adamw8bit
reg_weight: 1.0
steps: 1500
target_noise_multiplier: 1.0
train_text_encoder: false
train_unet: true
training_folder: /root/lorahub
trigger_word: halftone style
type: sd_trainer
job: extension
meta:
description: Halftone Aesthetic Style LoRA