Metraipolis › archive › dc_0455db7d3249

Enhancing Multimodal RLVR with Pairwise Token Contrast and Adaptive Trajectory Confidence — arXiv:2610.21671

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.21671
License (as archived)
CC0-1.0
Archived
2026-10-16T10:01:42.480Z
SHA-256 of the archived text
3679a960f745795f1ff6bcf916573958cfeac9f981e7e58ce9b1ce20b68cfb6f
Stable URL of this page
https://metraipolis.com/documento/dc_0455db7d3249

How to cite

Enhancing Multimodal RLVR with Pairwise Token Contrast and Adaptive Trajectory Confidence — arXiv:2610.21671 — arXiv (en), https://arxiv.org/abs/2610.21671 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0455db7d3249 (Metraipolis archive dc_0455db7d3249, sha256 3679a960f745795f1ff6bcf916573958cfeac9f981e7e58ce9b1ce20b68cfb6f)
BibTeX
@misc{dc0455db7d3249,
  title = {Enhancing Multimodal RLVR with Pairwise Token Contrast and Adaptive Trajectory Confidence — arXiv:2610.21671},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.21671},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0455db7d3249; sha256 3679a960f745795f1ff6bcf916573958cfeac9f981e7e58ce9b1ce20b68cfb6f},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.21671. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Reinforcement learning with verifiable rewards (RLVR) improves multimodal reasoning, but group-relative policy optimization broadcasts a single sequence-level advantage to every token of a response, although errors usually localize to a few decisive steps. Recent self-distilled variants refine credit with a teacher--student likelihood gap. However, this signal cannot tell whether it relies on the image or on a language shortcut, and is applied uniformly even when the teacher is uninformative. To address this, we propose Pairwise token contrast and Adaptive trajectory ConfidencE (PACE). Pairwise token contrast scores each response under four teacher views, crossing a correct or incorrect answer hint with the image present or removed, and compares the two hints for every token, which cancels shifts shared by both hints and reveals whether the token's support relies on the image. Adaptive trajectory confidence measures how well the teacher predicts each response given the correct hint and the image, normalizes this score within the batch, and uses it to set how far the token contrast redistributes credit within that response. PACE achieves the best average accuracy on three backbones and ranks first on all evaluated multimodal reasoning benchmarks with Qwen3-VL-8B-Instruct. Experiments show that both components matter: replacing pairwise token contrast with a single-view signal underperforms our method, while adaptive trajectory confidence adds further improvement.

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/2610.21671 — 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 2026-10-16T08:44:47.663Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).