Supervised Fine-Tuning Approach for Medical Question-Answering using Qwen Instruct

Abstract: Large Language Model (LLM) became promising tool for supporting medical applications based on natural language processing. Except that models trained for general purposes suffer from the limitation of accuracy and reliability when it deals with sensitive medical question answering (MQA). The model is evaluated using Exact Match and F1 score. The goal of this study is to develop LLM to answer the MQA tasks. Through applying SFT with Parmerter Efficient Fine Tuning (PEFT) specifical LoRA were applied. The results showed significant qualitative improvement and Notable accuracy in the answers. And organizing medical content, by reducing cognitive mistakes capered to the base model. The experiments refer to guided personalization for LLM represent essential step towards building medical systems.