Meta-Rewarding Language Models: Self-Improving Alignment with LLM-as-a-Meta-Judge

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wu, Tianhao, Yuan, Weizhe, Golovneva, Olga, Xu, Jing, Tian, Yuandong, Jiao, Jiantao, Weston, Jason, Sukhbaatar, Sainbayar
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914891749654528
author Wu, Tianhao
Yuan, Weizhe
Golovneva, Olga
Xu, Jing
Tian, Yuandong
Jiao, Jiantao
Weston, Jason
Sukhbaatar, Sainbayar
author_facet Wu, Tianhao
Yuan, Weizhe
Golovneva, Olga
Xu, Jing
Tian, Yuandong
Jiao, Jiantao
Weston, Jason
Sukhbaatar, Sainbayar
contents Large Language Models (LLMs) are rapidly surpassing human knowledge in many domains. While improving these models traditionally relies on costly human data, recent self-rewarding mechanisms (Yuan et al., 2024) have shown that LLMs can improve by judging their own responses instead of relying on human labelers. However, existing methods have primarily focused on improving model responses rather than judgment capabilities, resulting in rapid saturation during iterative training. To address this issue, we introduce a novel Meta-Rewarding step to the self-improvement process, where the model judges its own judgements and uses that feedback to refine its judgment skills. Surprisingly, this unsupervised approach improves the model's ability to judge {\em and} follow instructions, as demonstrated by a win rate improvement of Llama-3-8B-Instruct from 22.9% to 39.4% on AlpacaEval 2, and 20.6% to 29.1% on Arena-Hard. These results strongly suggest the potential for self-improving models without human supervision.
format Preprint
id arxiv_https___arxiv_org_abs_2407_19594
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Meta-Rewarding Language Models: Self-Improving Alignment with LLM-as-a-Meta-Judge
Wu, Tianhao
Yuan, Weizhe
Golovneva, Olga
Xu, Jing
Tian, Yuandong
Jiao, Jiantao
Weston, Jason
Sukhbaatar, Sainbayar
Computation and Language
Artificial Intelligence
Large Language Models (LLMs) are rapidly surpassing human knowledge in many domains. While improving these models traditionally relies on costly human data, recent self-rewarding mechanisms (Yuan et al., 2024) have shown that LLMs can improve by judging their own responses instead of relying on human labelers. However, existing methods have primarily focused on improving model responses rather than judgment capabilities, resulting in rapid saturation during iterative training. To address this issue, we introduce a novel Meta-Rewarding step to the self-improvement process, where the model judges its own judgements and uses that feedback to refine its judgment skills. Surprisingly, this unsupervised approach improves the model's ability to judge {\em and} follow instructions, as demonstrated by a win rate improvement of Llama-3-8B-Instruct from 22.9% to 39.4% on AlpacaEval 2, and 20.6% to 29.1% on Arena-Hard. These results strongly suggest the potential for self-improving models without human supervision.
title Meta-Rewarding Language Models: Self-Improving Alignment with LLM-as-a-Meta-Judge
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2407.19594