While reviewing the training examples in diffusers, I noticed that for Stable Diffusion we have several well-structured scripts covering different scenarios:

examples/text_to_image/train_text_to_image.py — fine-tunes the UNet parameters
examples/text_to_image/train_text_to_image_lora.py — trains LoRA layers in the UNet
examples/dreambooth/train_dreambooth.py — fine-tunes the UNet and optionally the text encoder
examples/dreambooth/train_dreambooth_lora.py — same as above, but with LoRA layers instead of full parameter tuning

However, for Flux and Flux Kontext, I don’t see equivalent training scripts outside of DreamBooth. From what I understand, this means that if someone wants to fine-tune these models directly (e.g., for text-to-image tasks without DreamBooth personalization), they currently don’t have a ready-made example.

My questions are:

1. Is this omission intentional (e.g., because DreamBooth covers most practical fine-tuning cases for Flux/Flux Kontext)?
2. Or is it simply that non-DreamBooth training scripts have not yet been added and would be open for contribution?

If the latter, I'd be interested in contributing scripts like train_text_to_image_flux.py and train_text_to_image_lora_flux.py, adapted from the Stable Diffusion versions. While I know the difference between these scripts is not huge, having non-DreamBooth training examples for Flux/Flux Kontext could be useful for some practitioners who want to fine-tune directly on new datasets without the DreamBooth setup.

Thanks for your insights!