haughty owl
Very minor issue.
It looks like the lora loading process will find .safetensors files from subfolders of /invokeai/loras, but when a prompt has a reference to a lora file from a subfolder, there will be a message in the console that it couldn't find the lora. If you move the lora from the subfolder into the root /lora folder without restarting the web ui server, and try to run the prompt again, it will successfully load it and use.
static geyser