Flux1: using sd-scripts trained LORA with diffusers

I have a really nice Lora that I trained with sd-scripts. It works very well with SD-Forge. Unfortunately SD-Forge does not seem like it is being developed actively, and I want to try a simpler alternative. Is it possible to load the same Lora with the Huggingface diffusers library? Would it need some kind of conversion to work properly.

Start by reviewing how sd-scripts saves LoRA weights and how the Hugging Face diffusers library loads them; the issue names no files or tests. Determine whether the existing artifact loads directly or requires conversion, and document the verified compatibility path.

Written by the indexing model from the issue text.