Salvato in:
Dettagli Bibliografici
Autori principali: Wang, Xiaoxuan, Liu, Bo, Jiang, Song, Liu, Jingzhou, Qi, Jingyuan, Chen, Xia, He, Baosheng
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2511.15137
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911276201934848
author Wang, Xiaoxuan
Liu, Bo
Jiang, Song
Liu, Jingzhou
Qi, Jingyuan
Chen, Xia
He, Baosheng
author_facet Wang, Xiaoxuan
Liu, Bo
Jiang, Song
Liu, Jingzhou
Qi, Jingyuan
Chen, Xia
He, Baosheng
contents The reasoning capabilities of large language models (LLMs) have been significantly improved through reinforcement learning (RL). Nevertheless, LLMs still struggle to consistently verify their own reasoning traces. This raises the research question of how to enhance the self-verification ability of LLMs and whether such an ability can further improve reasoning performance. In this work, we propose GRPO-Verif, an algorithm that jointly optimizes solution generation and self-verification within a unified loss function, with an adjustable hyperparameter controlling the weight of the verification signal. Experimental results demonstrate that our method enhances self-verification capability while maintaining comparable performance in reasoning.
format Preprint
id arxiv_https___arxiv_org_abs_2511_15137
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle From Solving to Verifying: A Unified Objective for Robust Reasoning in LLMs
Wang, Xiaoxuan
Liu, Bo
Jiang, Song
Liu, Jingzhou
Qi, Jingyuan
Chen, Xia
He, Baosheng
Machine Learning
Artificial Intelligence
The reasoning capabilities of large language models (LLMs) have been significantly improved through reinforcement learning (RL). Nevertheless, LLMs still struggle to consistently verify their own reasoning traces. This raises the research question of how to enhance the self-verification ability of LLMs and whether such an ability can further improve reasoning performance. In this work, we propose GRPO-Verif, an algorithm that jointly optimizes solution generation and self-verification within a unified loss function, with an adjustable hyperparameter controlling the weight of the verification signal. Experimental results demonstrate that our method enhances self-verification capability while maintaining comparable performance in reasoning.
title From Solving to Verifying: A Unified Objective for Robust Reasoning in LLMs
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2511.15137