{
  "aliases": [],
  "authors": [],
  "details": {
    "benchmark_protocol": {
      "actions": [
        "candidate-decision preference"
      ],
      "counts": [
        {
          "basis": "reported",
          "evidence_url": "https://huggingface.co/datasets/LangAGI-Lab/WebRewardBench/raw/main/README.md",
          "population": "Hub test split",
          "qualifier": "exact",
          "unit": "rows",
          "value": 776
        }
      ],
      "evaluation": "Compare reward-model preferences with the provided chosen/rejected decision labels; no independent score was reproduced.",
      "evidence_urls": [
        "https://huggingface.co/datasets/LangAGI-Lab/WebRewardBench/raw/main/README.md",
        "https://raw.githubusercontent.com/kyle8581/Web-Shepherd/HEAD/README.md"
      ],
      "focus": [
        "web process-reward evaluation"
      ],
      "notes": "The current Hub row count and project case-count claim use unresolved counting units. Candidate lists mean a row must not be assumed to equal one task or one pair.",
      "observations": [
        "page screenshots",
        "text observations",
        "action history",
        "task checklists"
      ],
      "setting": "offline"
    },
    "disabled": false,
    "download_url": "https://huggingface.co/datasets/LangAGI-Lab/WebRewardBench/tree/main",
    "gated": false,
    "hub_id": "LangAGI-Lab/WebRewardBench"
  },
  "environments": [
    "web"
  ],
  "evidence": [
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-17T02:44:50Z",
      "role": "primary",
      "sha256": "0968f60eea10b95f79cd16ad3746b0e0b1acaeb0e97c45856f709d53a14dd73a",
      "supports": [
        "name",
        "kind",
        "summary",
        "environments",
        "topics",
        "scope",
        "relations",
        "details"
      ],
      "url": "https://huggingface.co/datasets/LangAGI-Lab/WebRewardBench/raw/main/README.md"
    },
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-17T02:44:50Z",
      "role": "primary",
      "sha256": "db06c66b8cf5687008b0ad0d9ae269fcc27936887d238cfe2bdeb52151e1683e",
      "supports": [
        "identity",
        "release_status",
        "details",
        "licenses.data"
      ],
      "url": "https://huggingface.co/api/datasets/LangAGI-Lab/WebRewardBench"
    },
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-17T02:44:04Z",
      "role": "primary",
      "sha256": "07431058d8b9a71b80e75b7a74a40a8346c1503eec240628459cc4d4e2f0802f",
      "supports": [
        "summary",
        "environments"
      ],
      "url": "https://raw.githubusercontent.com/kyle8581/Web-Shepherd/HEAD/README.md"
    }
  ],
  "id": "url:ac6d0f80765f5a32255a",
  "identifiers": {},
  "kind": "benchmark",
  "last_verified_at": "2026-10-17T02:44:50Z",
  "licenses": {
    "code": null,
    "data": null,
    "document": null,
    "weights": null
  },
  "links": [
    {
      "rel": "primary",
      "url": "https://huggingface.co/datasets/LangAGI-Lab/WebRewardBench"
    }
  ],
  "name": "WebRewardBench",
  "notes": "Two public Parquet shards are present. Card metadata reports 776 test rows, whereas the project describes 1,000-plus test cases without reconciling that count to rows. Both observations are retained as a grain/coverage caveat rather than treating either as a confirmed number of independent tasks. No data license is declared.",
  "published_at": null,
  "relations": [
    {
      "evidence_url": "https://huggingface.co/datasets/LangAGI-Lab/WebRewardBench/raw/main/README.md",
      "target_id": "arxiv:2505.15277",
      "type": "documented_by"
    }
  ],
  "review_status": "curated",
  "scope": "core",
  "summary": "An offline preference-evaluation set for web process reward models, pairing page and trajectory context with preferred and rejected candidate decisions.",
  "topics": [
    "web-navigation",
    "reward-models",
    "evaluation"
  ],
  "updated_at": null,
  "url": "https://huggingface.co/datasets/LangAGI-Lab/WebRewardBench"
}