State-of-the-art Parameter-Efficient Fine-Tuning (PEFT) methods for efficiently adapting large pretrained models