Comparative Analysis and Evaluation of Well-Being Activity-Infused Fine-Tuned Language Models with Benchmark Models

Large language models (LLMs) have significantly impacted natural language processing (NLP), healthcare, and beyond fields. LLMs have demonstrated unprecedented abilities to understand and generate human-like text, making them highly valuable for various tasks, including information retrieval, question-answering, and conversational AI. However, as these models are integrated into real-world applications, especially in sensitive domains such as healthcare, it becomes essential to evaluate their performance comprehensively. We present a process for fine-tuning LLMs to incorporate cardiac exercise and corpus of well-being activity, aiming to promote activity health support through NLP-driven applications. Using QLoRA, you can ensure that the model size remains small and reduce overhead while maintaining the high performance of the fine-tuned model. Subsequently, this paper presents a comparative analysis benchmark pipeline to evaluate fine-tuned models with base models on standard benchmark datasets from the healthcare domain. The aim is to evaluate the performance, robustness and accuracy of these models, showing that the fine-tuning process does not degrade the model performance, but improves the performance to some extent in some benchmarks.