In more detail
LoRA is a shortcut for fine-tuning. Rather than retraining a model’s billions of parameters — slow, expensive, enormous — you train a small add-on that steers the frozen original, like clipping a corrective lens onto a camera instead of regrinding the whole thing.
It matters because it puts customization within everyone’s reach. Full fine-tuning needs serious hardware; a LoRA can be trained overnight on a good home computer and shared as a tiny file. Hobbyist communities trade thousands of them — art styles, writing voices, niche skills.
Goes with