{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-10-05-026",
    "source_story_id": "tmp-story-deep-tabular-size-ladder",
    "edition_id": "mp-2026-10-05-morning-0088",
    "edition_url": "https://themachinepress.com/edition/2026-10-05",
    "position": 15,
    "story_type": "dispatch",
    "section": "benchmarks-evals",
    "editorial_classification": "editorial",
    "headline": "Deep Tabular Generators Rarely Beat the Trivial Baseline",
    "slug": "deep-tabular-generators-rarely-beat-the-trivial-baseline",
    "dek": "A preregistered 2,220-run size ladder found no meaningful deep-model win in 23 of 24 dataset-model pairs.",
    "summary": "A preregistered 2,220-run size ladder found no meaningful deep-model win in 23 of 24 dataset-model pairs.",
    "body_text": "The benchmark compared seven generators across eight datasets subsampled from 200 to 20,000 rows, plus four naturally small clinical datasets. In 23 of 24 dataset-and-deep-model pairs, no deep model beat the best trivial baseline beyond seed noise at any measured size; baselines won 40 of 49 size cells. In 81 percent of cells, even the top-two gap was smaller than seed variation.",
    "why_it_matters": "A preregistered 2,220-run size ladder found no meaningful deep-model win in 23 of 24 dataset-model pairs.",
    "limitations": [],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-05-026/deep-tabular-generators-rarely-beat-the-trivial-baseline",
    "json_url": "https://themachinepress.com/story/mp-2026-10-05-026.json",
    "first_published_at": "2026-10-05T09:00:00.000-04:00",
    "modified_at": "2026-10-05T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-10-05-026-001",
        "text": "A preregistered 2,220-run size ladder found no meaningful deep-model win in 23 of 24 dataset-model pairs.",
        "source_ids": [
          "source-2026-10-05-015"
        ],
        "qualification": null
      }
    ],
    "source_ids": [
      "source-2026-10-05-015"
    ],
    "tags": [
      "tabular data",
      "generative models",
      "preregistered benchmark"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-10-05-015",
      "title": "arXiv preprint 2610.03500",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2610.03500",
      "canonical_url": "https://arxiv.org/abs/2610.03500",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-10-02T11:58:41.000-04:00",
      "accessed_at": "2026-10-05T08:30:00.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-10-05-026-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "Deep Tabular Generators Rarely Beat the Trivial Baseline",
    "publisher": "The Machine Press",
    "published_at": "2026-10-05T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-10-05-026/deep-tabular-generators-rarely-beat-the-trivial-baseline"
  }
}