@klaabu I'm in no way an expert but I've tried almost a hundred Loras and yet have come to face this error before :/ It might just be specific to InvokeAI or to your Lora into Invoke. No information online about this error either. Hopefully a patch from Invoke fixes it, your loras look fun!

@klaabu I've found some info that could mean something to you!
Invoke doesn't support these:
OneTrainer LoRA

XLabs LoRA

More info here: Using FLUX Models on Invoke : Invoke Support Portal

@klaabu stumbled upon a chat on Invoke's Discord and they were discussing some LoRas not being recognized. The support mentioned the code being incompatible but eh, not much more info :/

@klaabu i did attempt with khoya_ss but for some reason, this version+ the illustrious version always ends up always end up skipping all the layer merging, ex;

lora_unet_down_blocks_1_attentions_0_transformer_blocks_0_attn1_to_k.alpha, not found in base model.

i'll play with it some more and try to figure out whats going on

Mildly annoying but just use the pony version. I did a few same-seed tests on images that used the Illustrious-specific one and the differences were practically negligible

@klaabu Sorry can you explain

@klaabu Ty, that explains why I was seeing alot of models re upped recently

I just want to say thank you Klaabu for putting up with this nonsense, and fighting to make a stand for the right thing. When i saw that your Illustrious version got nuked i just face palmed myself with my hands and literally feet.