Saved in:
Bibliographic Details
Main Author: Mahesh, Atij
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.22084
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917042084380672
author Mahesh, Atij
author_facet Mahesh, Atij
contents Large Language Models (LLMs) still produce gender-stereotyped language even in occupation-neutral contexts that reflect deep societal biases (Rudinger et al., 2018). To address this, prior work has proposed prompting, constrained decoding (Dathathri et al., 2020; Zhou et al., 2024), post-processing, and fine-tuning-based alignment (Rafailov et al., 2023; Ravfogel et al., 2022). However, the comparative efficacy and learning dynamics remain little understood. We report a comparative analysis of six control techniques for bias mitigation: prompt-only, generate-and-filter, DFA-based Ctrl-G decoding, Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Iterative Nullspace Projection (INLP). We evaluate each method on a compositional constraint task. This task requires generating sentences that contain at least one agentic and one communal descriptor for each of the twenty Winogender-derived occupations. We quantify trade-offs between control strength and naturalness with evaluations of constraint compliance, lexical diversity, and fluency. Our results reveal key contrasts among the methods: SFT achieves 99.87 +- 0.15% compliance and high lexical diversity, while DPO, despite similar training stability, fails at 4.53 +- 0.82%. Ctrl-G guarantees perfect compliance, but at the cost of severely reduced fluency and diversity. Preference-based learning fundamentally differs: it cannot satisfy compositional constraints, as binary preference signals encode ranking, not logical conjunctions. Only explicit positive supervision enables mitigation of compositional biases; preference-based alignment fails to generalize logical structures, underscoring the limitations of preference learning and the necessity of explicit supervision for fair and fluent controlled generation.
format Preprint
id arxiv_https___arxiv_org_abs_2510_22084
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Compositional Bias Control in Large Language Models: Preference Learning Fails, Supervision Succeeds
Mahesh, Atij
Computation and Language
Large Language Models (LLMs) still produce gender-stereotyped language even in occupation-neutral contexts that reflect deep societal biases (Rudinger et al., 2018). To address this, prior work has proposed prompting, constrained decoding (Dathathri et al., 2020; Zhou et al., 2024), post-processing, and fine-tuning-based alignment (Rafailov et al., 2023; Ravfogel et al., 2022). However, the comparative efficacy and learning dynamics remain little understood. We report a comparative analysis of six control techniques for bias mitigation: prompt-only, generate-and-filter, DFA-based Ctrl-G decoding, Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Iterative Nullspace Projection (INLP). We evaluate each method on a compositional constraint task. This task requires generating sentences that contain at least one agentic and one communal descriptor for each of the twenty Winogender-derived occupations. We quantify trade-offs between control strength and naturalness with evaluations of constraint compliance, lexical diversity, and fluency. Our results reveal key contrasts among the methods: SFT achieves 99.87 +- 0.15% compliance and high lexical diversity, while DPO, despite similar training stability, fails at 4.53 +- 0.82%. Ctrl-G guarantees perfect compliance, but at the cost of severely reduced fluency and diversity. Preference-based learning fundamentally differs: it cannot satisfy compositional constraints, as binary preference signals encode ranking, not logical conjunctions. Only explicit positive supervision enables mitigation of compositional biases; preference-based alignment fails to generalize logical structures, underscoring the limitations of preference learning and the necessity of explicit supervision for fair and fluent controlled generation.
title Compositional Bias Control in Large Language Models: Preference Learning Fails, Supervision Succeeds
topic Computation and Language
url https://arxiv.org/abs/2510.22084