The Real, the Better: Aligning Large Language Models with Online Human Behaviors

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Jiang, Guanying, Yan, Lingyong, Shi, Haibo, Yin, Dawei
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909186398355456
author Jiang, Guanying
Yan, Lingyong
Shi, Haibo
Yin, Dawei
author_facet Jiang, Guanying
Yan, Lingyong
Shi, Haibo
Yin, Dawei
contents Large language model alignment is widely used and studied to avoid LLM producing unhelpful and harmful responses. However, the lengthy training process and predefined preference bias hinder adaptation to online diverse human preferences. To this end, this paper proposes an alignment framework, called Reinforcement Learning with Human Behavior (RLHB), to align LLMs by directly leveraging real online human behaviors. By taking the generative adversarial framework, the generator is trained to respond following expected human behavior; while the discriminator tries to verify whether the triplets of query, response, and human behavior come from real online environments. Behavior modeling in natural-language form and the multi-model joint training mechanism enable an active and sustainable online alignment. Experimental results confirm the effectiveness of our proposed methods by both human and automatic evaluations.
format Preprint
id arxiv_https___arxiv_org_abs_2405_00578
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle The Real, the Better: Aligning Large Language Models with Online Human Behaviors
Jiang, Guanying
Yan, Lingyong
Shi, Haibo
Yin, Dawei
Computation and Language
Artificial Intelligence
Large language model alignment is widely used and studied to avoid LLM producing unhelpful and harmful responses. However, the lengthy training process and predefined preference bias hinder adaptation to online diverse human preferences. To this end, this paper proposes an alignment framework, called Reinforcement Learning with Human Behavior (RLHB), to align LLMs by directly leveraging real online human behaviors. By taking the generative adversarial framework, the generator is trained to respond following expected human behavior; while the discriminator tries to verify whether the triplets of query, response, and human behavior come from real online environments. Behavior modeling in natural-language form and the multi-model joint training mechanism enable an active and sustainable online alignment. Experimental results confirm the effectiveness of our proposed methods by both human and automatic evaluations.
title The Real, the Better: Aligning Large Language Models with Online Human Behaviors
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2405.00578