I am testing AdamW8bit quality with batch sizes 1, 2, 4, 6. I am seeing what is the best sweetspot f
I am testing AdamW8bit quality with batch sizes 1, 2, 4, 6. I am seeing what is the best sweetspot for quality vs speed of training
Fyou can try latent couple extension
Fthis is necessary for companies
Fi will also research this for a company
Fit adds beard hard 
i tested not easy to get without a beard
i tested not easy to get without a beard
Ri got pretty impressive results with batch 2, which drove SDXL adamW8bit training by by 30% and bs 4 by 60%. For me it's somewhere in there. Need to check flexibility, but i am sure there are other experimental variables to test. Here's dataset of 12 images, vs xy plot with batch 2 up to 9 epochs (5000 steps), 5 hours vs 9 hours batch size 1
Rinteresting the level of variance in the dataset gets picked up in dim 128/alpha64. I guess the ultimate test will come when i run this LoRA on a Art Style model
R@Dr. Furkan Gözükara Hello, I'm trying to train LoRa on objects but it's not working. What's the reason? All the sources I saw were for faces
Fwell it works but object training is a different methodology
Fi can give private consultation
RHow?
KI am looking for any dev task.
Cold you help me?
Cold you help me?
Fsorry i dont have any tasks
P@Dr. Furkan Gözükara If I am training a model of my son, should I use reference images of boys his age instead of the ones I would use to train my own LORA?
S
Slooking better now.
Fif you can find yes better class dataset
FBS 2 brings 82% speed increase on A100
S
MThank you for tip, I will try it soon! 
MI've had no luck with adetailer, but ReActor works for me, and if the shape of the face fits, it can be a really good tool for creating new images to go with existing ones.
EInteresting. Thanks for the heads up I'll give it a shot.
Fwith batch size 3 but also xformers full sdxl dreambooth training is 20 min
Fon a100 80 gb
F4160 steps
F160 epochs
Fofc when batch size is 3 it takes 1387 steps
LAfter following this video when i press start training, I'm getting this error, without any data training progress at all https://www.youtube.com/watch?v=AY6DMBCIZ3A
File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.11_3.11.1520.0_x64qbz5n2kfra8p0\Lib\dataclasses.py", line 1230, in dataclass
return wrap(cls)
^^^^^^^^^
File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.11_3.11.1520.0_x64qbz5n2kfra8p0\Lib\dataclasses.py", line 1220, in wrap
return _process_class(cls, init, repr, eq, order, unsafe_hash,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.11_3.11.1520.0_x64qbz5n2kfra8p0\Lib\dataclasses.py", line 958, in _process_class
cls_fields.append(_get_field(cls, name, type, kw_only))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.11_3.11.1520.0_x64qbz5n2kfra8p0\Lib\dataclasses.py", line 815, in _get_field
raise ValueError(f'mutable default {type(f.default)} for field '
ValueError: mutable default <class 'timm.models.maxxvit.MaxxVitConvCfg'> for field conv_cfg is not allowed: use default_factory
Can anyone please help??
File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.11_3.11.1520.0_x64qbz5n2kfra8p0\Lib\dataclasses.py", line 1230, in dataclass
return wrap(cls)
^^^^^^^^^
File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.11_3.11.1520.0_x64qbz5n2kfra8p0\Lib\dataclasses.py", line 1220, in wrap
return _process_class(cls, init, repr, eq, order, unsafe_hash,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.11_3.11.1520.0_x64qbz5n2kfra8p0\Lib\dataclasses.py", line 958, in _process_class
cls_fields.append(_get_field(cls, name, type, kw_only))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.11_3.11.1520.0_x64qbz5n2kfra8p0\Lib\dataclasses.py", line 815, in _get_field
raise ValueError(f'mutable default {type(f.default)} for field '
ValueError: mutable default <class 'timm.models.maxxvit.MaxxVitConvCfg'> for field conv_cfg is not allowed: use default_factory
Can anyone please help??
YouTubeSECourses
Updated for SDXL 1.0. How to install #Kohya SS GUI trainer and do #LoRA training with Stable Diffusion XL (#SDXL) this is the video you are looking for. I have shown how to install Kohya from scratch. The best parameters to do LoRA training with SDXL. How to use Kohya SDXL LoRAs with ComfyUI. How to do checkpoint comparison with SDXL LoRAs and m...
Si think the latest automatic1111 still broken for training
Fyep you have error
Funinstall all python
FYouTubeSECourses
How to install Python, have multiple Python installations, set system wide default Python version. How to compose venv for any Python installation, change Python default path, and install SD web UI properly. Discord: https://bit.ly/SECoursesDiscord.
If I have been of assistance to you and you would like to show your support for my work, please...
If I have been of assistance to you and you would like to show your support for my work, please...
Finstall 3.10.11 as shown in above video
Faccurate. we use older version but that person is following kohya
Lso, at this point, can i just do a simple dreambooth training? so, that i don't have to go through a python installation process again!!
Fi dont think so none of the scripts will work with your python
Fgive it a try if you wish
LOhho, I see. Understood. I've already these two python version installed in my pc.
Funinstall both
Finstall as shown in video
Finstall into C drive
SGood night guys I’m trying to follow the lora training youtube video for runpod and I want to know what is the code to unzip the images files
Son the same folder as the video
Snever mind I solve it
Vhello Furkan, I only use a few images, about 10 approx, see, out of memory,
J@Dr. Furkan Gözükara I want to know how to train. lora use multi gpu in Ubuntu ? How to set accelerate config ?
Vmorning all, possible to make models or sth under temp folder to gain more ram, for the training notebook? i have tried two notebooks, only 10 pics, all failed ./sdxl_train_network.py FAILED
-----------------------------------------------------
Failures:
<NO_OTHER_FAILURES>
-----------------------------------------------------
Root Cause (first observed failure):
[0]:
time : 2023-10-07_02:51:09
host : 08833fa205ae
rank : 0 (local_rank: 0)
exitcode : -9 (pid: 1097)
error_file: <N/A>
traceback : Signal 9 (SIGKILL) received by PID 1097 i tried, two notebooks, possible to make everything in temp folders? otherwise it is too big, out of ram
-----------------------------------------------------
Failures:
<NO_OTHER_FAILURES>
-----------------------------------------------------
Root Cause (first observed failure):
[0]:
time : 2023-10-07_02:51:09
host : 08833fa205ae
rank : 0 (local_rank: 0)
exitcode : -9 (pid: 1097)
error_file: <N/A>
traceback : Signal 9 (SIGKILL) received by PID 1097 i tried, two notebooks, possible to make everything in temp folders? otherwise it is too big, out of ram
Vaccelerate launch --num_cpu_threads_per_process=2 "./sdxl_train_network.py" --pretrained_model_name_or_path="stabilityai/stable-diffusion-xl-base-1.0" --train_data_dir="/kaggle/working/results/img" --resolution="1024,1024" --output_dir="/kaggle/working/results/model" --logging_dir="/kaggle/working/results/log" --network_alpha="1" --save_model_as=safetensors --network_module=networks.lora --text_encoder_lr=0.0004 --unet_lr=0.0004 --network_dim=32 --output_name="kaggle_flower" --lr_scheduler_num_cycles="8" --no_half_vae --learning_rate="0.0004" --lr_scheduler="constant" --train_batch_size="1" --max_train_steps="1000" --save_every_n_epochs="1" --mixed_precision="fp16" --save_precision="fp16" --cache_latents --optimizer_type="Adafactor" --optimizer_args scale_parameter=False relative_step=False warmup_init=False --max_data_loader_n_workers="0" --bucket_reso_steps=64 --gradient_checkpointing --xformers --bucket_no_upscale --noise_offset=0.0 --lowram, are these settings too big?
Khi