{
  "aliases": [],
  "authors": [],
  "details": {
    "benchmark_protocol": {
      "actions": [
        "PyAutoGUI script: pointer movement/click/drag, key presses, hotkeys, and text entry"
      ],
      "counts": [
        {
          "basis": "reported",
          "evidence_url": "https://huggingface.co/datasets/Writer/omniact/raw/5e4bcb26745afc8d58ba9519ea9c9fb832386d26/README.md",
          "population": "train split, HF revision 5e4bcb26",
          "qualifier": "exact",
          "unit": "task examples",
          "value": 6788
        },
        {
          "basis": "reported",
          "evidence_url": "https://huggingface.co/datasets/Writer/omniact/raw/5e4bcb26745afc8d58ba9519ea9c9fb832386d26/README.md",
          "population": "test split, HF revision 5e4bcb26",
          "qualifier": "exact",
          "unit": "task examples",
          "value": 2020
        },
        {
          "basis": "reported",
          "evidence_url": "https://huggingface.co/datasets/Writer/omniact/raw/5e4bcb26745afc8d58ba9519ea9c9fb832386d26/README.md",
          "population": "val split, HF revision 5e4bcb26",
          "qualifier": "exact",
          "unit": "task examples",
          "value": 991
        }
      ],
      "evaluation": "The bundled evaluator reports weighted sequence match and action match, comparing action types and applying coordinate, key, and writing penalties. Target element-box metadata is evaluation-only. The retained scorer has a click-penalty defect: predicted coordinates are read from gold_line, so score reproduction requires explicit scorer-version review.",
      "evidence_urls": [
        "https://huggingface.co/datasets/Writer/omniact/raw/5e4bcb26745afc8d58ba9519ea9c9fb832386d26/README.md",
        "https://huggingface.co/api/datasets/Writer/omniact",
        "https://huggingface.co/datasets/Writer/omniact/raw/5e4bcb26745afc8d58ba9519ea9c9fb832386d26/eval.py"
      ],
      "focus": [
        "offline desktop/web action-script generation"
      ],
      "notes": "This released split table totals 9,799 examples; it is not silently substituted for any paper-level collection size. Public files include train/test/val JSON indexes, task/image archives, and eval.py. Evaluation is offline script comparison, not execution success on a live desktop. MIT is the Hub data declaration; original app/content rights are not resolved by it.",
      "observations": [
        "screenshot",
        "natural-language task"
      ],
      "setting": "offline"
    },
    "disabled": false,
    "download_url": "https://huggingface.co/datasets/Writer/omniact/tree/main",
    "gated": false,
    "hub_id": "Writer/omniact",
    "languages": [
      "en"
    ],
    "modalities": [],
    "size_category_reported": null
  },
  "environments": [
    "desktop",
    "web"
  ],
  "evidence": [
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-16T06:38:27Z",
      "role": "primary",
      "sha256": "2db0a5dd2b983dded3b90a8ecd66cadbb39074f03778a339822418531661b3e3",
      "supports": [
        "name",
        "kind",
        "summary",
        "environments",
        "topics",
        "scope",
        "relations",
        "details.benchmark_protocol"
      ],
      "url": "https://huggingface.co/api/datasets/Writer/omniact"
    },
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-17T14:10:00Z",
      "role": "primary",
      "sha256": "7b54ad557f89d621fe39c3d2fe9f02164a2bfcb7fef0db610d964586d3673693",
      "supports": [
        "details.benchmark_protocol"
      ],
      "url": "https://huggingface.co/datasets/Writer/omniact/raw/5e4bcb26745afc8d58ba9519ea9c9fb832386d26/README.md"
    },
    {
      "method": "editorial_primary_review",
      "retrieved_at": "2026-10-17T14:22:06Z",
      "role": "primary",
      "sha256": "a786de678af46df5e2b6b1a2a69d2ee239fe8fd5b63663765d84d39e21fc2365",
      "supports": [
        "details.benchmark_protocol"
      ],
      "url": "https://huggingface.co/datasets/Writer/omniact/raw/5e4bcb26745afc8d58ba9519ea9c9fb832386d26/eval.py"
    },
    {
      "method": "huggingface_api",
      "retrieved_at": "2026-10-16T06:38:27Z",
      "role": "primary",
      "sha256": "2db0a5dd2b983dded3b90a8ecd66cadbb39074f03778a339822418531661b3e3",
      "supports": [
        "details",
        "licenses.data"
      ],
      "url": "https://huggingface.co/api/datasets/Writer/omniact"
    }
  ],
  "id": "url:849a86beaee006c6b147",
  "identifiers": {},
  "kind": "benchmark",
  "last_verified_at": "2026-10-16T06:38:27Z",
  "licenses": {
    "code": null,
    "data": {
      "basis": "publisher model/dataset card",
      "declared": "mit",
      "retrieved_at": "2026-10-16T06:38:27Z",
      "url": "https://huggingface.co/datasets/Writer/omniact"
    },
    "document": null,
    "weights": null
  },
  "links": [
    {
      "rel": "primary",
      "url": "https://huggingface.co/datasets/Writer/omniact"
    }
  ],
  "name": "OmniAct",
  "notes": "",
  "published_at": null,
  "relations": [
    {
      "evidence_url": "https://huggingface.co/api/datasets/Writer/omniact",
      "target_id": "arxiv:2402.17553",
      "type": "documented_by"
    }
  ],
  "review_status": "curated",
  "scope": "core",
  "summary": "Dataset and benchmark for generating computer interaction programs.",
  "topics": [
    "evaluation"
  ],
  "updated_at": null,
  "url": "https://huggingface.co/datasets/Writer/omniact"
}