Recent studies have introduced innovative approaches in the field of artificial intelligence and protein generation. One study presents two energy-efficient protein language models, Llama-3-8B and Phi-3-mini, which utilize Low-Rank Adaptation (LoRA) for both controllable and uncontrollable protein generation. This research aims to enhance the efficiency of protein design through advanced language modeling techniques. Additionally, there is ongoing work on optimizing large language models via quantization, with a focus on comparing Post-Training Quantization (PTQ) and Quantization-Aware Training (QAT) techniques. Other studies are exploring the use of discrete speech tokens for semantic-related tasks in large language models, highlighting the versatility and potential applications of these technologies in various domains.