{
  "id": "2611.05150",
  "title": "What Sets the Serving Speed of Looped Language Models? Measurements and Optimizations Graded by Exactness",
  "authors": "Dezhen Lu, Runguo Li, Zixiang Ni, Yuchen Tang, Xingquan Kong, Jiaxing Yang, Jiarui Guo, Lingxiang Gao",
  "first_public": "2026-11-03",
  "added_at": "2026-11-09",
  "venue": "arXiv",
  "arxiv": "2611.05150",
  "paper_url": "https://arxiv.org/abs/2611.05150",
  "html_url": "https://arxiv.org/html/2611.05150v1",
  "lens": [
    "recurrent",
    "analysis"
  ],
  "representation": [
    "cache-memory",
    "hidden-state"
  ],
  "computation": [
    "recurrent"
  ],
  "training": [
    "inference-only"
  ],
  "questions": [
    "efficiency",
    "evaluation",
    "architecture"
  ],
  "evidence": [
    "empirical"
  ],
  "modality": [
    "language"
  ],
  "summary": "Measures inference bottlenecks in looped language models and improves cache packing, kernel fusion and prefix reuse. Separates changes that preserve exact outputs from changes to batching, randomness or numerical computation.",
  "provenance": "Primary-source literature review, 2026-11-09",
  "author_names": [
    "Dezhen Lu",
    "Runguo Li",
    "Zixiang Ni",
    "Yuchen Tang",
    "Xingquan Kong",
    "Jiaxing Yang",
    "Jiarui Guo",
    "Lingxiang Gao"
  ],
  "record_url": "https://latent-implicit-thinking.github.io/papers/2611.05150/",
  "legacy_record_url": "https://latent-implicit-thinking.github.io/papers/#paper-2611-05150",
  "citation_date": "2026-11-03",
  "json_url": "https://latent-implicit-thinking.github.io/papers/2611.05150/record.json",
  "markdown_url": "https://latent-implicit-thinking.github.io/papers/2611.05150/index.md",
  "bibtex_url": "https://latent-implicit-thinking.github.io/papers/2611.05150/citation.bib",
  "csl_url": "https://latent-implicit-thinking.github.io/papers/2611.05150/citation.csl.json",
  "pdf_url": "https://arxiv.org/pdf/2611.05150"
}