{
  "id": "2610.35556",
  "title": "LoopFold: Parallelizing Recurrent Depth in Looped Transformers",
  "authors": "Feichen Song, Boyi Zeng, Yiqin Hao, Shixiang Song, He Li, Yizhou Yang, Zitong Wang, Yifan Liu, Ziwei He, Zhouhan Lin",
  "first_public": "2026-10-26",
  "added_at": "2026-11-02",
  "venue": "arXiv",
  "arxiv": "2610.35556",
  "paper_url": "https://arxiv.org/abs/2610.35556",
  "html_url": "https://arxiv.org/html/2610.35556v1",
  "lens": [
    "recurrent",
    "beyond-sequential"
  ],
  "representation": [
    "hidden-state"
  ],
  "computation": [
    "recurrent",
    "depth-wise",
    "parallel"
  ],
  "training": [
    "pretraining"
  ],
  "questions": [
    "architecture",
    "efficiency",
    "scaling"
  ],
  "evidence": [
    "empirical"
  ],
  "modality": [
    "language"
  ],
  "summary": "Predicts inputs to later transformer loops so several depths can run in parallel, then corrects their outputs as earlier results arrive. Trains the predictor, corrector and shared backbone together with next-token prediction and measures the accuracy-latency trade-off.",
  "provenance": "Primary-source literature review, 2026-11-02",
  "author_names": [
    "Feichen Song",
    "Boyi Zeng",
    "Yiqin Hao",
    "Shixiang Song",
    "He Li",
    "Yizhou Yang",
    "Zitong Wang",
    "Yifan Liu",
    "Ziwei He",
    "Zhouhan Lin"
  ],
  "record_url": "https://latent-implicit-thinking.github.io/papers/2610.35556/",
  "legacy_record_url": "https://latent-implicit-thinking.github.io/papers/#paper-2610-35556",
  "citation_date": "2026-10-26",
  "json_url": "https://latent-implicit-thinking.github.io/papers/2610.35556/record.json",
  "markdown_url": "https://latent-implicit-thinking.github.io/papers/2610.35556/index.md",
  "bibtex_url": "https://latent-implicit-thinking.github.io/papers/2610.35556/citation.bib",
  "csl_url": "https://latent-implicit-thinking.github.io/papers/2610.35556/citation.csl.json",
  "pdf_url": "https://arxiv.org/pdf/2610.35556"
}