Small Language Models Need Strong Verifiers to Self-Correct Reasoning

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Yunxiang, Khalifa, Muhammad, Logeswaran, Lajanugen, Kim, Jaekyeom, Lee, Moontae, Lee, Honglak, Wang, Lu
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866913379859300352
author Zhang, Yunxiang
Khalifa, Muhammad
Logeswaran, Lajanugen
Kim, Jaekyeom
Lee, Moontae
Lee, Honglak
Wang, Lu
author_facet Zhang, Yunxiang
Khalifa, Muhammad
Logeswaran, Lajanugen
Kim, Jaekyeom
Lee, Moontae
Lee, Honglak
Wang, Lu
contents Self-correction has emerged as a promising solution to boost the reasoning performance of large language models (LLMs), where LLMs refine their solutions using self-generated critiques that pinpoint the errors. This work explores whether small (<= 13B) language models (LMs) have the ability of self-correction on reasoning tasks with minimal inputs from stronger LMs. We propose a novel pipeline that prompts smaller LMs to collect self-correction data that supports the training of self-refinement abilities. First, we leverage correct solutions to guide the model in critiquing their incorrect responses. Second, the generated critiques, after filtering, are used for supervised fine-tuning of the self-correcting reasoner through solution refinement. Our experimental results show improved self-correction abilities of two models on five datasets spanning math and commonsense reasoning, with notable performance gains when paired with a strong GPT-4-based verifier, though limitations are identified when using a weak self-verifier for determining when to correct.
format Preprint
id arxiv_https___arxiv_org_abs_2404_17140
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Small Language Models Need Strong Verifiers to Self-Correct Reasoning
Zhang, Yunxiang
Khalifa, Muhammad
Logeswaran, Lajanugen
Kim, Jaekyeom
Lee, Moontae
Lee, Honglak
Wang, Lu
Computation and Language
Self-correction has emerged as a promising solution to boost the reasoning performance of large language models (LLMs), where LLMs refine their solutions using self-generated critiques that pinpoint the errors. This work explores whether small (<= 13B) language models (LMs) have the ability of self-correction on reasoning tasks with minimal inputs from stronger LMs. We propose a novel pipeline that prompts smaller LMs to collect self-correction data that supports the training of self-refinement abilities. First, we leverage correct solutions to guide the model in critiquing their incorrect responses. Second, the generated critiques, after filtering, are used for supervised fine-tuning of the self-correcting reasoner through solution refinement. Our experimental results show improved self-correction abilities of two models on five datasets spanning math and commonsense reasoning, with notable performance gains when paired with a strong GPT-4-based verifier, though limitations are identified when using a weak self-verifier for determining when to correct.
title Small Language Models Need Strong Verifiers to Self-Correct Reasoning
topic Computation and Language
url https://arxiv.org/abs/2404.17140