How to add LORA/weight to replicate-javascript?

[Pickra]

Hey, I'm a noob so please be gentle.

I want to add LORA/weight to an instance of SDXL.

I'm using replicate-javascript to run an SDXL model. I trained a different instance of SDXL on replicate.com and downloaded the weights. How do I add the weights/LORA via replicate-javascript? Or there's no docs for this because that's not how it works?

Thank you.

[mattt]

Hi @Pickra. If you fine-tuned SDXL on Replicate, you can run that model directly. Here's a comprehensive guide for how all of that works: https://replicate.com/blog/fine-tune-sdxl

[Pickra]

Thanks for getting back to me @mattt !

I'm aware I can run the model on replicate. I want to train for multiple types of fantasy characters. The SDXL base instance is giving me poor results, but the training gets good results.

I could retrain the same model and not have to worry about the LORAs but I thought the best approach was to create/train multiple LORAs and apply them to 1 instance of SDXL. It's possible I'm misunderstanding or am unaware of the optimal way to get trained results for multiple types of fantasy characters.

Please advise. Thank you.

[mattt]

@Pickra Sorry for the late response. There are models on Replicate that support multiple LoRAs, such as lucataco/flux-dev-multi-lora. You should be able to fine-tune FLUX multiple times for different subjects and combine them in a single run.