Metraipolis › archive › dc_c9fb53369fca

MTA-Zero: Self-Improving Multimodal Deep Search via Online QA-Solver Co-training — arXiv:2610.33799

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.33799
License (as archived)
CC0-1.0
Archived
2026-10-27T08:11:25.860Z
SHA-256 of the archived text
8f37a6e1c57889ed92bccfb25e1ba6315817926a6562fc3daaaffc0f1e0558b1
Stable URL of this page
https://metraipolis.com/documento/dc_c9fb53369fca

How to cite

MTA-Zero: Self-Improving Multimodal Deep Search via Online QA-Solver Co-training — arXiv:2610.33799 — arXiv (en), https://arxiv.org/abs/2610.33799 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c9fb53369fca (Metraipolis archive dc_c9fb53369fca, sha256 8f37a6e1c57889ed92bccfb25e1ba6315817926a6562fc3daaaffc0f1e0558b1)
BibTeX
@misc{dcc9fb53369fca,
  title = {MTA-Zero: Self-Improving Multimodal Deep Search via Online QA-Solver Co-training — arXiv:2610.33799},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.33799},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c9fb53369fca; sha256 8f37a6e1c57889ed92bccfb25e1ba6315817926a6562fc3daaaffc0f1e0558b1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.33799. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Multimodal large language models (MLLMs) have become strong visual reasoners, yet remain limited in open-world, multi-step search that requires grounding images, selecting tools, and integrating web evidence. Existing approaches typically construct and filter fixed offline corpora whose question distributions cannot adapt as the target deep-search agent improves. Building on MTA-Agent's image-grounded QA construction, we present MTA-Zero, an online multimodal QA-Solver co-training framework for self-improving multimodal deep-search agents that requires no precomputed multi-hop corpus. Starting from image-grounded VQA seed contexts, a trainable QA Generator retrieves evidence and constructs verified multi-hop questions online. A separate deep-search Solver then selects tools, gathers evidence, and answers the generated questions. To focus generation on the Solver's learning frontier, we derive a Goldilocks reward for the QA Generator from the Solver's group-wise rollout success rate: questions solved by all rollouts or by none receive zero reward, whereas questions near the current frontier are favored. We optimize the two policies independently with GRPO, without sharing parameters or gradients. Across six multimodal deep-search benchmarks, MTA-Zero improves a Qwen3-VL-8B Solver from 36.91% to 51.58% mean accuracy and achieves performance comparable to GPT-5 under the same agent workflow. The same framework also improves a newer Qwen3.6-27B Solver by 10.82 points. These results show that verified online question generation provides a policy-specific curriculum for continually improving multimodal deep-search agents.

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/2610.33799 — 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 2026-10-27T07:56:00.209Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).