From Baselines to Preferences: A Comparative Study of LoRA/QLoRA and Preference Optimization for Mental Health Text Classification

2026-04-01

AI Summary

This study addresses the lack of systematic guidance in selecting optimization strategies for mental health text classification. The authors propose a reproducible NLP optimization framework that systematically evaluates a range of approaches—from conventional fine-tuning and parameter-efficient methods (LoRA/QLoRA) to preference optimization algorithms (DPO, ORPO, KTO)—with a focus on the impact of objective functions, adapter architectures, and optimizer behavior. Experimental results demonstrate that performance gains are highly configuration-dependent: certain fine-tuning strategies yield consistent improvements, whereas preference optimization methods exhibit substantial variability in effectiveness. These findings underscore the necessity of empirically grounded method selection over heuristic stacking. This work provides a transparent, controlled optimization pathway and evidence-based decision support for mental health NLP applications.

Abstract

Mental health text classification has rapidly adopted modern adaptation methods, yet practical guidance on which optimization strategy to use, when, and why remains limited. This paper presents a systematic comparative study of optimization pathways for a joint mental-health classification task, moving from strong vanilla baselines to progressively more specialized techniques. We first establish classical and encoder references, then examine parameter-efficient supervised fine-tuning with LoRA/QLoRA under multiple objective and optimization settings, and finally evaluate preference-based optimization with DPO, ORPO, and KTO, including class-rebalanced training. Rather than emphasizing a single headline score, we focus on methodological insight: how performance changes with objective formulation, adapter choice, optimizer behavior, context windowing, and class-balance intervention. The results show that optimization effects are highly method-dependent: some approaches deliver stable, transferable gains, while others are sensitive to configuration and data balance. Preference optimization, in particular, exhibits large variation across objectives, indicating that method selection is more consequential than simply adding a preference-training stage. The central contribution is a clear optimization narrative for mental health NLP: start from transparent baselines, apply controlled tuning, and use preference optimization selectively where its gains are demonstrable. This provides a reproducible and practically grounded framework for choosing effective training strategies beyond architecture choice alone.

Problem

• mental health text classification
• optimization strategy
• preference optimization
• LoRA/QLoRA
• method comparison

Innovation

• preference optimization
• LoRA/QLoRA
• mental health text classification
• systematic comparison
• class-rebalanced training