{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-09-06-003",
    "source_story_id": "tmp-story-temporal-self-distillation-video",
    "edition_id": "mp-2026-09-06-morning-0059",
    "edition_url": "https://themachinepress.com/edition/2026-09-06",
    "position": 3,
    "story_type": "dispatch",
    "section": "frontier-models",
    "editorial_classification": "editorial",
    "headline": "The Dense Video View Taught the Sparse One What Changed",
    "slug": "the-dense-video-view-taught-the-sparse-one-what-changed",
    "dek": "A model used denser temporal sampling as its own teacher, improving state tracking without labels, a separate teacher or added inference cost.",
    "summary": "A model used denser temporal sampling as its own teacher, improving state tracking without labels, a separate teacher or added inference cost.",
    "body_text": "S3T gives the same model a dense view of a clip as privileged training information and asks a sparse-view student to match its next-token distribution. On LLaVA-OneVision-2-8B, the authors report VSTAT gains from 1.74 to 2.70 points depending on configuration. Training on unlabeled synthetic clips also transferred to real video, adding 7.95 points on VSTAT-YouTube and 4.50 on MVBench Action Count. Those gains are benchmark results for the tested model, not proof of general video understanding.",
    "why_it_matters": "A model used denser temporal sampling as its own teacher, improving state tracking without labels, a separate teacher or added inference cost.",
    "limitations": [
      "Those gains are benchmark results for the tested model, not proof of general video understanding."
    ],
    "importance": 9,
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-06-003/the-dense-video-view-taught-the-sparse-one-what-changed",
    "json_url": "https://themachinepress.com/story/mp-2026-09-06-003.json",
    "first_published_at": "2026-09-06T09:00:00.000-04:00",
    "modified_at": "2026-09-06T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-09-06-003-001",
        "text": "A model used denser temporal sampling as its own teacher, improving state tracking without labels, a separate teacher or added inference cost.",
        "source_ids": [
          "source-2026-09-06-003"
        ],
        "qualification": "Those gains are benchmark results for the tested model, not proof of general video understanding."
      }
    ],
    "source_ids": [
      "source-2026-09-06-003"
    ],
    "tags": [
      "video understanding",
      "self-distillation",
      "state tracking"
    ],
    "image_url": "https://themachinepress.com/issues/2026-09-06/temporal-distillation-file.webp",
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-09-06-003",
      "title": "arXiv preprint 2609.04203",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2609.04203",
      "canonical_url": "https://arxiv.org/abs/2609.04203",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-09-03T13:59:55.000-04:00",
      "accessed_at": "2026-09-06T08:20:00.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-09-06-003-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "The Dense Video View Taught the Sparse One What Changed",
    "publisher": "The Machine Press",
    "published_at": "2026-09-06T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-06-003/the-dense-video-view-taught-the-sparse-one-what-changed"
  }
}