Abstract
Chat models are capable of answering a wide range of questions, however, theaccuracy of their responses is highly uncertain. In this research, we propose aspecialized PEFT-MedAware model where we utilize parameter-efficientfine-tuning (PEFT) to enhance the Falcon-1b large language model on specializedMedQuAD data consisting of 16,407 medical QA pairs, leveraging only 0.44% ofits trainable parameters to enhance computational efficiency. The paper adoptsdata preprocessing and PEFT to optimize model performance, complemented by aBitsAndBytesConfig for efficient transformer training. The resulting model wascapable of outperforming other LLMs in medical question-answering tasks inspecific domains with greater accuracy utilizing limited computationalresources making it suitable for deployment in resource-constrainedenvironments. We propose further improvements through expanded datasets, largermodels, and feedback mechanisms for sustained medical relevancy. Our workhighlights the efficiency gains and specialized capabilities of PEFT in medicalAI, outpacing standard models in precision without extensive resource demands.The proposed model and data are released for research purposes only.