Slimming Down LLMs Without Losing Their Minds

Qingda, Mai

This paper investigates and validates the impact of fine-tuning on large language model performance, focusing on parameter-efficient methods (LoRA and QLoRA). We evaluate model cap…