{
  "aliases": [],
  "authors": [
    "Dhananjay Ashok",
    "Shantanu Agarwal",
    "Vivek Datla",
    "Jonathan May",
    "Alfy Samuel"
  ],
  "details": {
    "arxiv_snapshot": {
      "abstract_hash_method": "sha256-utf8-whitespace-collapsed-v1",
      "abstract_sha256": "35ded32e7bac9a875a0fa41efad43fed5fb0aab54fade5cfc2a0628b1e666587",
      "submitted_at": "2026-10-01T22:30:08Z",
      "updated_at": "2026-10-01T22:30:08Z",
      "version": 1,
      "versioned_url": "https://arxiv.org/abs/2610.02542v1"
    },
    "license_review": {
      "document": {
        "evidence_urls": [
          "https://arxiv.org/html/2610.02542v1"
        ],
        "reason": "The retained version-specific arXiv HTML infobox explicitly labels this paper License: CC BY 4.0. This document declaration does not establish code, training-data or model-weight licenses."
      }
    }
  },
  "environments": [
    "web"
  ],
  "evidence": [
    {
      "method": "arxiv_api",
      "retrieved_at": "2026-10-20T15:06:37Z",
      "role": "primary",
      "sha256": "a3d753815e82190ee2de6f322cd816c32f2dd391f862845305fc7e57300303a6",
      "supports": [
        "name",
        "authors",
        "published_at",
        "updated_at",
        "identifiers.arxiv",
        "details.arxiv_snapshot"
      ],
      "url": "https://export.arxiv.org/api/query?search_query=%28abs%3A%22desktop+agent%22+OR+abs%3A%22computer+interaction%22+OR+abs%3A%22GUI+automation%22+OR+abs%3A%22web+navigation%22+OR+abs%3A%22browser+automation%22+OR+abs%3A%22visual+grounding%22+OR+abs%3A%22screen+understanding%22+OR+abs%3A%22mobile+agents%22%29+AND+submittedDate%3A%5B202609200000+TO+202610200000%5D&start=0&max_results=200&sortBy=submittedDate&sortOrder=descending"
    },
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-20T15:08:56Z",
      "role": "primary",
      "sha256": "d8c22c6b3f07d4156c38af9dd7ca7c06a5d0a11889a7350d821a020e21604590",
      "supports": [
        "name",
        "kind",
        "summary",
        "environments",
        "topics",
        "scope",
        "notes",
        "licenses.document"
      ],
      "url": "https://arxiv.org/html/2610.02542v1"
    }
  ],
  "id": "arxiv:2610.02542",
  "identifiers": {
    "arxiv": "2610.02542"
  },
  "kind": "paper",
  "last_verified_at": "2026-10-20T15:08:56Z",
  "licenses": {
    "code": null,
    "data": null,
    "document": {
      "basis": "The retained version-specific arXiv HTML infobox explicitly labels this paper License: CC BY 4.0. This document declaration does not establish code, training-data or model-weight licenses.",
      "declared": "CC-BY-4.0",
      "retrieved_at": "2026-10-20T15:08:56Z",
      "url": "https://arxiv.org/html/2610.02542v1"
    },
    "weights": null
  },
  "links": [
    {
      "rel": "primary",
      "url": "https://arxiv.org/abs/2610.02542"
    }
  ],
  "name": "How To Train Your World Model: Fine-tuning vs RAG for LM-based World Modeling",
  "notes": "Adjacent cross-domain study with WebShop as its web evaluation. WebShop uses text-rendered search/product pages and an explicit valid-action menu, not screenshot control or live purchases. A frozen policy proposes five actions; a world model predicts their next observations and the policy scores them before one is executed in the simulator. Utility is normalized environment reward, while fidelity uses a model judge; neither is a universal GUI task-success rate. Fine-tuning and retrieval use collected training experience; the hybrid masks same-episode retrieval, which does not establish general contamination freedom. The retained Atom and v1 HTML abstract wording differs; original bibliographic metadata and its abstract digest are preserved.",
  "published_at": "2026-10-01",
  "relations": [],
  "review_status": "curated",
  "scope": "adjacent",
  "summary": "Compares fine-tuned, retrieval-based and hybrid language-model world models for next-observation prediction and action selection in text environments including WebShop.",
  "topics": [
    "world-models",
    "planning",
    "evaluation",
    "web-navigation"
  ],
  "updated_at": "2026-10-01",
  "url": "https://arxiv.org/abs/2610.02542"
}