Ekanari's picture
Add files using upload-large-folder tool
f5fe3ee verified
job: extension
config:
name: qwen_2509_object_removal_new
process:
- type: diffusion_trainer
training_folder: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/output
sqlite_db_path: ./aitk_db.db
device: cuda
trigger_word: null
performance_log_every: 10
network:
type: lora
linear: 16
linear_alpha: 16
conv: 16
conv_alpha: 16
lokr_full_rank: true
lokr_factor: -1
network_kwargs:
ignore_if_contains: []
save:
dtype: bf16
save_every: 50
max_step_saves_to_keep: 4
save_format: diffusers
push_to_hub: false
datasets:
- folder_path: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_train/target
mask_path: null
mask_min_value: 0.1
default_caption: Remove Object
caption_ext: txt
caption_dropout_rate: 0.05
cache_latents_to_disk: true
is_reg: false
network_weight: 1
resolution:
- 512
controls: []
shrink_video_to_frames: true
num_frames: 1
do_i2v: true
flip_x: false
flip_y: false
control_path_1: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_train/mask_on_background
control_path_2: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_train/mask
control_path_3: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_train/mae_output
train:
batch_size: 6
bypass_guidance_embedding: false
steps: 5500
gradient_accumulation: 1
train_unet: true
train_text_encoder: false
gradient_checkpointing: true
noise_scheduler: flowmatch
optimizer: adamw
timestep_type: weighted
content_or_style: balanced
optimizer_params:
weight_decay: 0.0001
unload_text_encoder: false
cache_text_embeddings: true
lr: 0.0005
ema_config:
use_ema: false
ema_decay: 0.99
skip_first_sample: true
force_first_sample: false
disable_sampling: false
dtype: bf16
diff_output_preservation: false
diff_output_preservation_multiplier: 1
diff_output_preservation_class: person
switch_boundary_every: 1
loss_type: mse
model:
name_or_path: Qwen/Qwen-Image-Edit-2509
quantize: true
qtype: uint3|ostris/accuracy_recovery_adapters/qwen_image_edit_2509_torchao_uint3.safetensors
quantize_te: true
qtype_te: uint4
arch: qwen_image_edit_plus
low_vram: true
model_kwargs:
match_target_res: false
layer_offloading: true
layer_offloading_text_encoder_percent: 0
layer_offloading_transformer_percent: 1
sample:
sampler: flowmatch
sample_every: 50
width: 512
height: 512
samples:
- prompt: Remove Object
ctrl_img_1: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_val/mask_on_background/17.jpg
ctrl_img_2: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_val/mask/17.jpg
ctrl_img_3: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_val/mae_output/17.jpg
- prompt: Remove Object
ctrl_img_1: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_val/mask_on_background/45.jpg
ctrl_img_2: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_val/mask/45.jpg
ctrl_img_3: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_val/mae_output/45.jpg
- prompt: Remove Object
ctrl_img_1: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_val/mask_on_background/49.jpg
ctrl_img_2: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_val/mask/49.jpg
ctrl_img_3: /home/nguyenvanthanhdat1810@gmail.com/ai-toolkit/datasets/object_removal_val/mae_output/49.jpg
neg: ''
seed: 42
walk_seed: true
guidance_scale: 4
sample_steps: 10
num_frames: 1
fps: 1
meta:
name: qwen_2509_object_removal_new
version: '1.0'