ahh a classic model for me growing up, kinda surprised I didn't see a existing model for her on here before this,
Trained on SD1.5, so pretty decent compatibility with most realistic models
Setting the Lora to 0.8 should do the trick in most cases, that said if its little too strict especially if your trying to mix loras, I would bring it down to as low as 0.6
Trigger Word emilygreyodd should be all you need for this one, but you can use terms like long hair, black hair, makeup to help likeness,
err.. due to nature of the dataset, it does nsfw pretty easily so use terms like nudity, nude, if you don't want that
Have Fun! :))