How to Do SDXL Training For FREE with Kohya LoRA - Kaggle Notebook - NO GPU Required - Pwns Google Colab - 53 Chapters - Manually Fixed Subtitles

[FurkanGozukara]

I also made a pull request @bmaltais I hope you can approve it. Keep up the amazing work.

How To Do SDXL LoRA Training On RunPod With Kohya SS GUI Trainer & Use LoRAs With Automatic1111 UI

[Huzaif-Ahmed]

thank you for such amazing content ,i just wanted to ask that i havr given a training image dataset of 5 images and it shows me 4 hrs .but i wanted to run in p100 gpu provided by kaggke

so i made changes like
!accelerate launch --num_gpus=1 "./sdxl_train_network.py"
--pretrained_model_name_or_path="stabilityai/stable-diffusion-xl-base-1.0"
--train_data_dir="/kaggle/working/training-data/img"
--reg_data_dir="/kaggle/working/training-data/reg"
--output_dir="/kaggle/working/my-dreambooth-lora"
--logging_dir="/kaggle/working/training-data/log"
--resolution="1024,1024"
--network_alpha="1"
--network_dim=32
--save_model_as=safetensors
--network_module=networks.lora
--text_encoder_lr=0.0004
--unet_lr=0.0004
--output_name="sdxl-dreambooth-lora"
--lr_scheduler_num_cycles="8"
--no_half_vae
--learning_rate="0.0004"
--lr_scheduler="constant"
--train_batch_size="1"
--max_train_steps=$(($(ls /kaggle/input/training-images | wc -l) * 400))
--save_every_n_epochs="1"
--mixed_precision="amp"
--save_precision="amp"
--optimizer_type="Adafactor"
--optimizer_args scale_parameter=False relative_step=False warmup_init=False
--max_data_loader_n_workers="0"
--gradient_checkpointing
--xformers
--bucket_no_upscale
--noise_offset=0.0
--lowram
--mem_eff_attn

still itd not wokring can you please help me over here

[FurkanGozukara]

thank you for such amazing content ,i just wanted to ask that i havr given a training image dataset of 5 images and it shows me 4 hrs .but i wanted to run in p100 gpu provided by kaggke

so i made changes like !accelerate launch --num_gpus=1 "./sdxl_train_network.py" --pretrained_model_name_or_path="stabilityai/stable-diffusion-xl-base-1.0" --train_data_dir="/kaggle/working/training-data/img" --reg_data_dir="/kaggle/working/training-data/reg" --output_dir="/kaggle/working/my-dreambooth-lora" --logging_dir="/kaggle/working/training-data/log" --resolution="1024,1024" --network_alpha="1" --network_dim=32 --save_model_as=safetensors --network_module=networks.lora --text_encoder_lr=0.0004 --unet_lr=0.0004 --output_name="sdxl-dreambooth-lora" --lr_scheduler_num_cycles="8" --no_half_vae --learning_rate="0.0004" --lr_scheduler="constant" --train_batch_size="1" --max_train_steps=$(($(ls /kaggle/input/training-images | wc -l) * 400)) --save_every_n_epochs="1" --mixed_precision="amp" --save_precision="amp" --optimizer_type="Adafactor" --optimizer_args scale_parameter=False relative_step=False warmup_init=False --max_data_loader_n_workers="0" --gradient_checkpointing --xformers --bucket_no_upscale --noise_offset=0.0 --lowram --mem_eff_attn

still itd not wokring can you please help me over here

you need to add full fp16 now