LoRA: Low-Rank Adaptation of Large Language Models PyTorch Implementation. The paper can be found here