Lora error

[Bellatrix8]

Hello, so the steps I followed were to configure using the LowVRAM JSON file, then select the inputs for image, log, and model, and after I tried training I received this 'heavy' line of code and nothing would happen after that. What would be the issue?

CUDA SETUP: Loading binary C:\kohya\kohya_ss\venv\lib\site-packages\bitsandbytes\libbitsandbytes_cuda116.dll... use 8-bit Adam optimizer running training / 学習開始 num train images * repeats / 学習画像の数×繰り返し回数: 3000 num reg images / 正則化画像の数: 0 num batches per epoch / 1epochのバッチ数: 3000 num epochs / epoch数: 1 batch size per device / バッチサイズ: 1 total train batch size (with parallel & distributed & accumulation) / 総バッチサイズ（並列学習、勾配合計含む）: 1 gradient accumulation steps / 勾配を合計するステップ数 = 1 total optimization steps / 学習ステップ数: 3000 Traceback (most recent call last): File "C:\kohya\kohya_ss\train_network.py", line 573, in train(args) File "C:\kohya\kohya_ss\train_network.py", line 356, in train "ss_noise_offset": args.noise_offset, AttributeError: 'Namespace' object has no attribute 'noise_offset' Traceback (most recent call last): File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.2800.0_x64__qbz5n2kfra8p0\lib\runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.2800.0_x64__qbz5n2kfra8p0\lib\runpy.py", line 86, in _run_code exec(code, run_globals) File "C:\kohya\kohyass\venv\Scripts\accelerate.exe_main.py", line 7, in File "C:\kohya\kohya_ss\venv\lib\site-packages\accelerate\commands\accelerate_cli.py", line 45, in main args.func(args) File "C:\kohya\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 1104, in launch_command simple_launcher(args) File "C:\kohya\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 567, in simple_launcher raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd) subprocess.CalledProcessError: Command '['C:\kohya\kohya_ss\venv\Scripts\python.exe', 'train_network.py', '--pretrained_model_name_or_path=E:/stable diffusion 2.1/stable-diffusion-webui-master/models/Stable-diffusion/Anything-V3.0-pruned-fp16.ckpt', '--train_data_dir=E:/train/trained/lora/image', '--resolution=512,512', '--output_dir=E:/train/trained/lora/model', '--logging_dir=E:/train/trained/lora/log', '--network_alpha=128', '--save_model_as=safetensors', '--network_module=networks.lora', '--text_encoder_lr=5e-5', '--unet_lr=0.0001', '--network_dim=128', '--output_name=lora', '--lr_scheduler_num_cycles=1', '--learning_rate=0.0001', '--lr_scheduler=constant', '--train_batch_size=1', '--max_train_steps=3000', '--save_every_n_epochs=1', '--mixed_precision=fp16', '--save_precision=fp16', '--seed=1234', '--caption_extension=.txt', '--cache_latents', '--max_data_loader_n_workers=1', '--clip_skip=2', '--bucket_reso_steps=64', '--mem_eff_attn', '--gradient_checkpointing', '--xformers', '--use_8bit_adam', '--bucket_no_upscale']' returned non-zero exit status 1.

[AtoHolewa]

I have the same problem :(

[AtoHolewa]

CUDA SETUP: Loading binary D:\AI\SUPERSD\Kohya\kohya_ss\venv\lib\site-packages\bitsandbytes\libbitsandbytes_cuda116.dll... use 8-bit Adam optimizer running training / 学習開始 num train images * repeats / 学習画像の数×繰り返し回数: 1500 num reg images / 正則化画像の数: 0 num batches per epoch / 1epochのバッチ数: 750 num epochs / epoch数: 1 batch size per device / バッチサイズ: 2 total train batch size (with parallel & distributed & accumulation) / 総バッチサイズ（並列学習、勾配合計含む）: 2 gradient accumulation steps / 勾配を合計するステップ数 = 1 total optimization steps / 学習ステップ数: 450 Traceback (most recent call last): File "D:\AI\SUPERSD\Kohya\kohya_ss\train_network.py", line 573, in train(args) File "D:\AI\SUPERSD\Kohya\kohya_ss\train_network.py", line 356, in train "ss_noise_offset": args.noise_offset, AttributeError: 'Namespace' object has no attribute 'noise_offset' Traceback (most recent call last): File "C:\Users\Bonnier\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "C:\Users\Bonnier\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 86, in _run_code exec(code, run_globals) File "D:\AI\SUPERSD\Kohya\kohya_ss\venv\Scripts\accelerate.exe__main__.py", line 7, in File "D:\AI\SUPERSD\Kohya\kohya_ss\venv\lib\site-packages\accelerate\commands\accelerate_cli.py", line 45, in main args.func(args) File "D:\AI\SUPERSD\Kohya\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 1104, in launch_command simple_launcher(args) File "D:\AI\SUPERSD\Kohya\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 567, in simple_launcher raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd) subprocess.CalledProcessError: Command '['D:\AI\SUPERSD\Kohya\kohya_ss\venv\Scripts\python.exe', 'train_network.py', '--pretrained_model_name_or_path=runwayml/stable-diffusion-v1-5', '--train_data_dir=D:/AI/LORA/AtoLora/image', '--resolution=512,512', '--output_dir=D:/AI/LORA/AtoLora/model', '--logging_dir=D:/AI/LORA/AtoLora/log', '--network_alpha=128', '--save_model_as=safetensors', '--network_module=networks.lora', '--text_encoder_lr=5e-5', '--unet_lr=0.0001', '--network_dim=128', '--output_name=Ato Holewa', '--lr_scheduler_num_cycles=1', '--learning_rate=0.0001', '--lr_scheduler=constant', '--train_batch_size=2', '--max_train_steps=450', '--save_every_n_epochs=1', '--mixed_precision=fp16', '--save_precision=bf16', '--seed=1234', '--caption_extension=.txt', '--cache_latents', '--max_data_loader_n_workers=1', '--clip_skip=2', '--bucket_reso_steps=64', '--xformers', '--use_8bit_adam', '--bucket_no_upscale']' returned non-zero exit status 1.

[Bellatrix8]

Well,i hope the creator will troubleshoot this in the next few days,still if you come up with a solution please tell me as well

[martianunlimited]

That has nothing to do with bitsandbytes / 8bitADAM The repository bmaltais/kohya_ss is out-of-sync with the repository kohya_ss/sd_scripts in particular file library/train_util.py is out of sync. Replace that file with the version in kohya_ss/sd_scripts
https://github.com/bmaltais/kohya_ss/issues/192

File "D:\AI\SUPERSD\Kohya\kohya_ss\train_network.py", line 356, in train "ss_noise_offset": args.noise_offset, AttributeError: 'Namespace' object has no attribute 'noise_offset' Traceback (most recent call last):

[DieserBobby]

I have a similar problem (I believe, at least the error code-block looks similar) . I tried several ideas as a solution, but nothing worked:

• using Lion instead of adam
• using adam 8bit instead of adam
• substituting all 3 ocurrencies of the train_util.py
• changing resolution
• updating with upgrade.ps1

Any new or better ideas for me to get it running?

Here the end of my error code

Traceback (most recent call last): File "C:\Users\bobby\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "C:\Users\bobby\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 86, in _run_code exec(code, run_globals) File "F:\Stable-Diffusion\kohya\kohya_ss\venv\Scripts\accelerate.exe__main__.py", line 7, in File "F:\Stable-Diffusion\kohya\kohya_ss\venv\lib\site-packages\accelerate\commands\accelerate_cli.py", line 45, in main args.func(args) File "F:\Stable-Diffusion\kohya\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 1104, in launch_command simple_launcher(args) File "F:\Stable-Diffusion\kohya\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 567, in simple_launcher raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd) subprocess.CalledProcessError: Command '['F:\Stable-Diffusion\kohya\kohya_ss\venv\Scripts\python.exe', 'train_db.py', '--pretrained_model_name_or_path=F:/Stable-Diffusion/stable-diffusion-webui/models/Stable-diffusion/liberty_main.safetensors', '--train_data_dir=F:/Stable-Diffusion/kohya/Bilder_Training/lora/image', '--resolution=512,512', '--output_dir=F:/Stable-Diffusion/stable-diffusion-webui/models/Lora', '--logging_dir=F:/Stable-Diffusion/kohya/Bilder_Training/lora/log', '--save_model_as=safetensors', '--output_name=dazpose', '--max_data_loader_n_workers=1', '--learning_rate=0.0001', '--lr_scheduler=constant', '--train_batch_size=2', '--max_train_steps=2300', '--save_every_n_epochs=1', '--mixed_precision=bf16', '--save_precision=bf16', '--seed=1234', '--caption_extension=.txt', '--cache_latents', '--optimizer_type=AdamW', '--max_data_loader_n_workers=1', '--clip_skip=2', '--bucket_reso_steps=64', '--xformers', '--use_8bit_adam', '--bucket_no_upscale']' returned non-zero exit status 1.

[vietdragon]

Same problem!

caching latents. 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 54/54 [00:12<00:00, 4.26it/s] import network module: networks.lora create LoRA for Text Encoder: 72 modules. create LoRA for U-Net: 192 modules. enable LoRA for text encoder enable LoRA for U-Net prepare optimizer, data loader etc. Traceback (most recent call last): File "C:\Users\Long Dao\kohya_ss\train_network.py", line 507, in train(args) File "C:\Users\Long Dao\kohya_ss\train_network.py", line 150, in train optimizer_name, optimizer_args, optimizer = train_util.get_optimizer(args, trainable_params) File "C:\Users\Long Dao\kohya_ss\library\train_util.py", line 1536, in get_optimizer assert optimizer_type is None or optimizer_type == "", "both option use_8bit_adam and optimizer_type are specified / use_8bit_adamとoptimizer_typeの両方 のオプションが指定されています" AssertionError: both option use_8bit_adam and optimizer_type are specified / use_8bit_adamとoptimizer_typeの両方のオプションが指定されています Traceback (most recent call last): File "C:\Users\Long Dao\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "C:\Users\Long Dao\AppData\Local\Programs\Python\Python310\lib\runpy.py", line 86, in _run_code exec(code, run_globals) File "C:\Users\Long Dao\kohya_ss\venv\Scripts\accelerate.exe__main__.py", line 7, in File "C:\Users\Long Dao\kohya_ss\venv\lib\site-packages\accelerate\commands\accelerate_cli.py", line 45, in main args.func(args) File "C:\Users\Long Dao\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 1104, in launch_command simple_launcher(args) File "C:\Users\Long Dao\kohya_ss\venv\lib\site-packages\accelerate\commands\launch.py", line 567, in simple_launcher raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd) subprocess.CalledProcessError: Command '['C:\Users\Long Dao\kohya_ss\venv\Scripts\python.exe', 'train_network.py', '--enable_bucket', '--pretrained_model_name_or_path=F:/stable-diffusion/stable-diffusion-webui/models/Stable-diffusion/realisticVisionV13_v13.ckpt', '--train_data_dir=F:\Lora Training Data\Stmtp\image', '--resolution=512,512', '--output_dir=F:\Lora Training Data\Stmtp\model', '--logging_dir=F:\Lora Training Data\Stmtp\log', '--network_alpha=1', '--save_model_as=safetensors', '--network_module=networks.lora', '--text_encoder_lr=5e-5', '--unet_lr=0.0001', '--network_dim=8', '--output_name=last', '--lr_scheduler_num_cycles=1', '--learning_rate=0.0001', '--lr_scheduler=cosine', '--lr_warmup_steps=540', '--train_batch_size=1', '--max_train_steps=5400', '--save_every_n_epochs=1', '--mixed_precision=fp16', '--save_precision=fp16', '--cache_latents', '--optimizer_type=AdamW8bit', '--bucket_reso_steps=64', '--xformers', '--use_8bit_adam', '--bucket_no_upscale']' returned non-zero exit status 1.

[DieserBobby]

I got it running again (checking very very many combinations): In my case: "Memory efficient attention" should be on (some days before there hadn't been the need to) AND "use 8bit adam" in the advanced section shouldn't be checked.

[ZhouCongArts]

I also had the same problem, unchecking "Use 8bit adam" in Training parameters > Advanced Configuration worked for me.

[lennyfung]

How do you "Uncheck" this item? It's not a checkbox.

[DieserBobby]

You need to click on "Advanced Configuration" further down, on the same page where you made the screenshot. Plenty new options will appear... among these: "use 8bit adam" which is checked on default. Uncheck and you made a step in the right direction... hopefully :-)

[Zirnworks]

I don't see the option to uncheck 8bit adam anywhere in my advanced config.

[daflood]

I don't see the option to uncheck 8bit adam anywhere in my advanced config.

Try selecting AdamW in the optimizer drop down instead of AdamW8bit. That fixed it for me.

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.