{
  "id": "2610.30066",
  "title": "Looped Uniform Diffusion Language Models",
  "authors": "Saiteja Utpala, Chieh-Hsin Lai, Yuki Mitsufuji, Praneeth Vepakomma",
  "first_public": "2026-10-21",
  "added_at": "2026-10-22",
  "venue": "arXiv",
  "arxiv": "2610.30066",
  "paper_url": "https://arxiv.org/abs/2610.30066",
  "html_url": "https://arxiv.org/html/2610.30066v1",
  "lens": [
    "recurrent",
    "beyond-sequential"
  ],
  "representation": [
    "hidden-state",
    "discrete-token"
  ],
  "computation": [
    "recurrent",
    "depth-wise",
    "diffusion",
    "parallel"
  ],
  "training": [
    "pretraining"
  ],
  "questions": [
    "architecture",
    "efficiency",
    "scaling"
  ],
  "evidence": [
    "empirical",
    "scaling"
  ],
  "modality": [
    "language"
  ],
  "summary": "Repeats a shared block inside each denoising step of a uniform diffusion language model. Studies how training-time loop counts and adaptive stopping affect language modeling quality, training cost and downstream accuracy.",
  "provenance": "Primary-source literature review, 2026-10-22",
  "author_names": [
    "Saiteja Utpala",
    "Chieh-Hsin Lai",
    "Yuki Mitsufuji",
    "Praneeth Vepakomma"
  ],
  "record_url": "https://latent-implicit-thinking.github.io/papers/2610.30066/",
  "legacy_record_url": "https://latent-implicit-thinking.github.io/papers/#paper-2610-30066",
  "citation_date": "2026-10-21",
  "json_url": "https://latent-implicit-thinking.github.io/papers/2610.30066/record.json",
  "markdown_url": "https://latent-implicit-thinking.github.io/papers/2610.30066/index.md",
  "bibtex_url": "https://latent-implicit-thinking.github.io/papers/2610.30066/citation.bib",
  "csl_url": "https://latent-implicit-thinking.github.io/papers/2610.30066/citation.csl.json",
  "pdf_url": "https://arxiv.org/pdf/2610.30066"
}