{
  "id": "arxiv:2507.11662",
  "citation-key": "arxiv_2507_11662",
  "type": "article",
  "title": "Let's Think in Two Steps: Mitigating Agreement Bias in MLLMs with Self-Grounded Verification",
  "URL": "https://arxiv.org/abs/2507.11662v3",
  "custom": {
    "catalog_id": "arxiv:2507.11662",
    "catalog_kind": "paper",
    "catalog_review_status": "curated",
    "canonical_source_url": "https://arxiv.org/abs/2507.11662",
    "identifiers": {
      "arxiv": "2507.11662"
    },
    "first_published_at": "2025-07-15",
    "record_updated_at": "2026-03-08",
    "catalog_scope": "core",
    "arxiv_snapshot": {
      "abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1",
      "abstract_sha256": "86c4cb97fcf46fc15619ca481261d42518d84e94d43978cdddbf2cbae2b6ae33",
      "submitted_at": "2025-07-15T18:50:29Z",
      "updated_at": "2026-03-08T00:45:24Z",
      "version": 3,
      "versioned_url": "https://arxiv.org/abs/2507.11662v3"
    }
  },
  "note": "First published: 2025-07-15; Recorded arXiv version: v3; Recorded version updated: 2026-03-08T00:45:24Z.",
  "author": [
    {
      "literal": "Moises Andrade"
    },
    {
      "literal": "Joonhyuk Cha"
    },
    {
      "literal": "Brandon Ho"
    },
    {
      "literal": "Vriksha Srihari"
    },
    {
      "literal": "Karmesh Yadav"
    },
    {
      "literal": "Zsolt Kira"
    }
  ],
  "issued": {
    "date-parts": [
      [
        2025,
        7,
        15
      ]
    ]
  },
  "archive": "arXiv",
  "archive_location": "2507.11662",
  "genre": "preprint",
  "version": "3"
}