GuardReasoner-Omni: A Reasoning-based Multi-modal Guardrail for Text, Image, Video, and Audio

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhu, Zhenhao, Liu, Yue, Guo, Yanpei, Qu, Wenjie, Chen, Cancan, He, Yufei, Li, Yibo, Chen, Yulin, Wu, Tianyi, Xu, Huiying, Zhu, Xinzhong, Zhang, Jiaheng
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910265666174976
author Zhu, Zhenhao
Liu, Yue
Guo, Yanpei
Qu, Wenjie
Chen, Cancan
He, Yufei
Li, Yibo
Chen, Yulin
Wu, Tianyi
Xu, Huiying
Zhu, Xinzhong
Zhang, Jiaheng
author_facet Zhu, Zhenhao
Liu, Yue
Guo, Yanpei
Qu, Wenjie
Chen, Cancan
He, Yufei
Li, Yibo
Chen, Yulin
Wu, Tianyi
Xu, Huiying
Zhu, Xinzhong
Zhang, Jiaheng
contents We present GuardReasoner-Omni, a reasoning-based guardrail model designed to moderate text, image, video, and audio data. First, we construct a comprehensive training corpus comprising 181k samples spanning these four modalities. Our training pipeline follows a two-stage paradigm to incentivize the model to deliberate before making decisions: (1) conducting SFT to cold-start the model with explicit reasoning capabilities and structural adherence; and (2) performing RL with a concise correctness reward to preserve accurate reasoning while suppressing redundant generation. We release a suite of models scaled at 3B and 7B parameters. Extensive experiments demonstrate that GuardReasoner-Omni achieves superior performance compared to existing state-of-the-art baselines across various guardrail benchmarks.
format Preprint
id arxiv_https___arxiv_org_abs_2602_03328
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle GuardReasoner-Omni: A Reasoning-based Multi-modal Guardrail for Text, Image, Video, and Audio
Zhu, Zhenhao
Liu, Yue
Guo, Yanpei
Qu, Wenjie
Chen, Cancan
He, Yufei
Li, Yibo
Chen, Yulin
Wu, Tianyi
Xu, Huiying
Zhu, Xinzhong
Zhang, Jiaheng
Cryptography and Security
We present GuardReasoner-Omni, a reasoning-based guardrail model designed to moderate text, image, video, and audio data. First, we construct a comprehensive training corpus comprising 181k samples spanning these four modalities. Our training pipeline follows a two-stage paradigm to incentivize the model to deliberate before making decisions: (1) conducting SFT to cold-start the model with explicit reasoning capabilities and structural adherence; and (2) performing RL with a concise correctness reward to preserve accurate reasoning while suppressing redundant generation. We release a suite of models scaled at 3B and 7B parameters. Extensive experiments demonstrate that GuardReasoner-Omni achieves superior performance compared to existing state-of-the-art baselines across various guardrail benchmarks.
title GuardReasoner-Omni: A Reasoning-based Multi-modal Guardrail for Text, Image, Video, and Audio
topic Cryptography and Security
url https://arxiv.org/abs/2602.03328