Support for training style with examples/dreambooth/train_dreambooth_lora_sdxl.py

[OranHe]

Is your feature request related to a problem? Please describe. I'm trying to use _examples/dreambooth/train_dreambooth_lorasdxl.py to train an original style dreambooth SDXL model. If I'm right, this script only support train object for now.

Describe the solution you'd like Support for training style with examples/dreambooth/train_dreambooth_lora_sdxl.py

Describe alternatives you've considered

it will be better if train_dreambooth_lora_sdxl.py support args train_data_dir and caption_column just like train_text_to_image_lora_sdxl.py

accelerate launch train_text_to_image_lora_sdxl.py \ --pretrained_model_name_or_path="stabilityai/stable-diffusion-xl-base-1.0" \ --train_data_dir=$DATA_DIR --caption_column="text" \ --resolution=1024 --random_flip \ --train_batch_size=1 --num_train_epochs=1000 --checkpointing_steps=200 \ --learning_rate=1e-04 --lr_scheduler="constant" --lr_warmup_steps=0 \ --output_dir="sdxl-lora"

[Xargonus]

Hi, I don't have experience with training styles using DreamBooth, but aren't the options instance_data_dir and instance_prompt, which are available in examples/dreambooth/train_dreambooth_lora_sdxl.py, equivalent to the train_data_dir and caption_column that you mention?

[mspilimbergo]

@OranHe were you able to figure out how to train a model with an original style? Any suggestion would be appreciated! I am trying to train a lora with dragon ball z styles.

Thanks!

[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.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.