How to use train_text_to_image_sdxl.py with unlabed data

[khayamgondal]

How can I use examples/text_to_image/train_text_to_image_sdxl.py with images without captions?
Dreambooth has a version for sdxl_lora but I am looking for a training script that uses sdxl without lora.

[sayakpaul]

Generate the captions using multi-modal models.

Dreambooth has a version for sdxl_lora but I am looking for a training script that uses sdxl without lora.

Feel free to repupose this script: https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth.py.