Annotation-Efficient Language Model Alignment via Diverse and Representative Response Texts

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Jinnai, Yuu, Honda, Ukyo
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915498271178752
author Jinnai, Yuu
Honda, Ukyo
author_facet Jinnai, Yuu
Honda, Ukyo
contents Preference optimization is a standard approach to fine-tuning large language models to align with human preferences. The quantity, diversity, and representativeness of the preference dataset are critical to the effectiveness of preference optimization. However, obtaining a large amount of preference annotations is difficult in many applications. This raises the question of how to use the limited annotation budget to create an effective preference dataset. To this end, we propose Annotation-Efficient Preference Optimization (AEPO). Instead of exhaustively annotating preference over all available response texts, AEPO selects a subset of responses that maximizes diversity and representativeness from the available responses and then annotates preference over the selected ones. In this way, AEPO focuses the annotation budget on labeling preferences over a smaller but informative subset of responses. We evaluate the performance of preference learning using AEPO on three datasets and show that it outperforms the baselines with the same annotation budget. Our code is available at https://github.com/CyberAgentAILab/annotation-efficient-po
format Preprint
id arxiv_https___arxiv_org_abs_2405_13541
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Annotation-Efficient Language Model Alignment via Diverse and Representative Response Texts
Jinnai, Yuu
Honda, Ukyo
Computation and Language
Artificial Intelligence
Machine Learning
Preference optimization is a standard approach to fine-tuning large language models to align with human preferences. The quantity, diversity, and representativeness of the preference dataset are critical to the effectiveness of preference optimization. However, obtaining a large amount of preference annotations is difficult in many applications. This raises the question of how to use the limited annotation budget to create an effective preference dataset. To this end, we propose Annotation-Efficient Preference Optimization (AEPO). Instead of exhaustively annotating preference over all available response texts, AEPO selects a subset of responses that maximizes diversity and representativeness from the available responses and then annotates preference over the selected ones. In this way, AEPO focuses the annotation budget on labeling preferences over a smaller but informative subset of responses. We evaluate the performance of preference learning using AEPO on three datasets and show that it outperforms the baselines with the same annotation budget. Our code is available at https://github.com/CyberAgentAILab/annotation-efficient-po
title Annotation-Efficient Language Model Alignment via Diverse and Representative Response Texts
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2405.13541