How can LLMs be fine-tuned for specialized domain knowledge?

I found this step-by-step guide with behind the scene explanation about LoRA & QLoRA. I think you can find an answer from this. https://medium.com/stackademic/how-to-fine-tune-llms-with-lora-qlora-26953ef326f3?sk=ac8354362ab1efa7ced2335f9f4c7fdf