{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-09-29-009",
    "source_story_id": "tmp-story-holom-discrete-humanoid-vla",
    "edition_id": "mp-2026-09-29-morning-0082",
    "edition_url": "https://themachinepress.com/edition/2026-09-29",
    "position": 9,
    "story_type": "dispatch",
    "section": "robotics",
    "editorial_classification": "editorial",
    "headline": "A Humanoid Model Split Action Tokens by Body Part",
    "slug": "a-humanoid-model-split-action-tokens-by-body-part",
    "dek": "Grouped discrete diffusion decoded end-effector, body, hand and kinematic actions for real-time control.",
    "summary": "Grouped discrete diffusion decoded end-effector, body, hand and kinematic actions for real-time control.",
    "body_text": "Holo-M extends a language model's vocabulary with four action-token groups for end effectors, body, hands and kinematics. The design trains across humanoid teleoperation, egocentric human video and simulation, then decodes body-part groups with discrete diffusion rather than token-by-token autoregression. The authors report the highest success rates in both generalist and specialist evaluations on their SIMPLE humanoid benchmark. That ranking belongs to the submitted comparison set; code and weights are promised for release.",
    "why_it_matters": "Grouped discrete diffusion decoded end-effector, body, hand and kinematic actions for real-time control.",
    "limitations": [],
    "importance": 8,
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-29-009/a-humanoid-model-split-action-tokens-by-body-part",
    "json_url": "https://themachinepress.com/story/mp-2026-09-29-009.json",
    "first_published_at": "2026-09-29T09:00:00.000-04:00",
    "modified_at": "2026-09-29T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-09-29-009-001",
        "text": "Grouped discrete diffusion decoded end-effector, body, hand and kinematic actions for real-time control.",
        "source_ids": [
          "source-2026-09-29-009"
        ],
        "qualification": null
      }
    ],
    "source_ids": [
      "source-2026-09-29-009"
    ],
    "tags": [
      "humanoid robots",
      "vision-language-action",
      "action tokens"
    ],
    "image_url": null,
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-09-29-009",
      "title": "arXiv preprint 2609.35709",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2609.35709",
      "canonical_url": "https://arxiv.org/abs/2609.35709",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-09-28T13:44:15.000-04:00",
      "accessed_at": "2026-09-29T08:18:00.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-09-29-009-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "A Humanoid Model Split Action Tokens by Body Part",
    "publisher": "The Machine Press",
    "published_at": "2026-09-29T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-09-29-009/a-humanoid-model-split-action-tokens-by-body-part"
  }
}