How does it work? On my previous LoRAs I never captioned anything and they turned out ok. I tried captioning for a new LoRA (I did not add the trigger words to the captions, not sure if that makes a difference) and the sample generations didn’t generate how I wanted at all. If anyone has some insight on this I’d really appreciate it. What am I supposed to tag? Do I include trigger words in the tags? Should I have read the guide on LoRAs before I trained (yes)??