Evaluating Fine-Tuning Efficiency of Human-Inspired Learning Strategies in Medical Question Answering

2024/08/15 by Yang, Yushi, Bean, Andrew M., McCraith, Robert +1 · 1 citation
#Computation and Language (cs.CL) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.2408.07888

Abstract

Fine-tuning Large Language Models (LLMs) incurs considerable training costs, driving the need for data-efficient training with optimised data ordering. Human-inspired strategies offer a solution by organising data based on human learning practices. This study evaluates the fine-tuning efficiency of five human-inspired strategies across four language models, three datasets, and both human- and LLM-labelled data in the context of medical question answering. These strategies achieve the best accuracy gain of 1.81% and an average gain of 1.02% across datasets, with interleaved strategies delivering the best average results. However, the best strategy varies across model-dataset combinations, limiting the generalisability of the effects of any single strategy. Additionally, LLM-defined question difficulty outperforms human-defined labels in curriculum-based learning, showing the potential of model-generated data as a cost-effective alternative for optimising fine-tuning.

Cited by

• Beyond Random Sampling: Efficient Language Model Pretraining via Curriculum Learning

Related

• A Comprehensive Study on Fine-Tuning Large Language Models for Medical Question Answering Using Classification Models and Comparative Analysis
• Do Large Language Models have Shared Weaknesses in Medical Question Answering?
• Enhancing Healthcare through Large Language Models: A Study on Medical Question Answering
• Large language models encode clinical knowledge
• Aligning Large Language Models for Clinical Tasks