{
  "abstract": "Block drafters propose several tokens in one forward pass, before earlier target tokens are realised. Their rejection mixes two losses: missing within-block path information and imperfect modelling of observable information. Accepted length cannot distinguish them. We separate the two with an information floor, the minimum expected rejection at a specified conditioning order; rejection above this floor is the model gap. Measurements across four domains and four open-weight targets, plus floor estimates on a frontier API target, yield three findings. First, the all-parallel floor reaches $0.286$ at the final slot on Qwen3-4B, limiting even the best proposal to $71\\%$ per-slot acceptance. Second, one realised token removes $86$--$100\\%$ of this floor, a locality also recovered by an independent mutual-information analysis. Third, current drafters remain far above their floors: the final-slot model gap accounts for $43$--$64\\%$ of DFlash rejection and $85$--$92\\%$ of DSpark's oracle-conditioned rejection. Guided by this separation of short-range conditioning from proposal quality, we replace a context-independent predecessor correction with a prefix-attention head that reads committed context conditional on the predecessor. With supporting backbone components, the resulting Qwen3-4B drafter improves mean serving accepted length by $3.93\\%$ over the released DSpark checkpoint across nine tasks, without increasing the conditioning order. Code and data: https://github.com/tie-pilot-qxw/specfloor; drafter checkpoints: https://huggingface.co/TIE-Pilot/dspark-attnconv-block7-qwen3-4b.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2608.27339",
  "author_total": 6,
  "authors": [
    {
      "name": "Xinwei Qiang"
    },
    {
      "name": "Xiang Fang"
    },
    {
      "name": "Chang Chen"
    },
    {
      "name": "Zaifeng Pan"
    },
    {
      "name": "Yue Guan"
    },
    {
      "name": "Yufei Ding"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/3p07ptjxdgnbr.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2608.27339",
  "fulltext_url": null,
  "id": "3p07ptjxdgnbr",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2608.27339",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.CL"
    ],
    [
      "arxiv",
      "cs.IT"
    ],
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "math.IT"
    ],
    [
      "openalex",
      "1702|Natural Language Processing Techniques"
    ]
  ],
  "openalex": "W7204589058",
  "pdf_url": "https://arxiv.org/pdf/2608.27339",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-27",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/3p07ptjxdgnbr.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:25:17Z",
      "native_id": "2608.27339",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2608.27339"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:math.IT",
      "subfield": "general"
    }
  ],
  "title": "Beyond Parallel Blindness: Information Floors and Model Gaps in Block Drafting",
  "title_text": "Beyond Parallel Blindness: Information Floors and Model Gaps in Block Drafting",
  "type": "preprint",
  "updated_at": "2026-10-30T12:25:17Z",
  "url": "https://research.draconitic.com/paper/3p07ptjxdgnbr",
  "venue": "arXiv",
  "versions": []
}