Metraipolis › archive › dc_0ac23db282bc
Grounding Multimodal Reasoning with Evidence-Ablated Negatives — arXiv:2610.39237
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.39237
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T06:16:59.471Z
- SHA-256 of the archived text
a0831de75277ebc323019b4001998f08879cd2c422e01cb3ad8b7617006b2572- Stable URL of this page
https://metraipolis.com/documento/dc_0ac23db282bc
How to cite
Grounding Multimodal Reasoning with Evidence-Ablated Negatives — arXiv:2610.39237 — arXiv (en), https://arxiv.org/abs/2610.39237 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0ac23db282bc (Metraipolis archive dc_0ac23db282bc, sha256 a0831de75277ebc323019b4001998f08879cd2c422e01cb3ad8b7617006b2572)
BibTeX
@misc{dc0ac23db282bc,
title = {Grounding Multimodal Reasoning with Evidence-Ablated Negatives — arXiv:2610.39237},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.39237},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0ac23db282bc; sha256 a0831de75277ebc323019b4001998f08879cd2c422e01cb3ad8b7617006b2572},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.39237. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Reinforcement learning with verifiable rewards (RLVR) has become a powerful tool for improving multimodal reasoning. However, standard outcome supervision suffers from credit assignment ambiguity, as a model may receive positive rewards by exploiting language priors or dataset shortcuts rather than grounding its reasoning in the requisite visual evidence. This makes high-quality negative responses critical, since effective negatives should remain plausible under the current policy while exposing the shortcut that should be suppressed. In this paper, we propose EVAN, a multimodal reinforcement learning framework utilizing EVidence-Ablated Negatives to train LMMs against their own language-prior failures. EVAN first identifies highly visually dependent examples using a scaled image information gain metric and preserves data diversity with a mixed-ranking strategy. It then constructs evidence-ablated images by masking the most discriminative image patches, eliciting plausible but visually ungrounded responses from the current policy. During policy optimization, these negative responses are incorporated as bounded auxiliary signals under the clean image, penalizing shortcut learning while maintaining RL stability. Experiments across challenging multimodal reasoning benchmarks demonstrate that EVAN mitigates modality imbalance and improves over strong RLVR baselines. Our code is available at this https URL . Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2610.39237 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-31T06:07:23.354Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).