LLM Parameter-Efficient Fine-Tuning (PEFT/LoRA)

Fine-tune open-weight LLMs using LoRA, QLoRA, Hugging Face TRL, and custom instruction datasets.