{
  "id": "2606.18524",
  "title": "Loop-μP: Residual Scaling and Hyperparameter Transfer in Looped Transformers",
  "authors": "Shaowen Wang, Bingrui Li, Tansheng Zhu, Ge Zhang, Wenhao Huang, Shen Yan, Jian Li",
  "first_public": "2026-06-16",
  "added_at": "2026-11-03",
  "venue": "arXiv",
  "arxiv": "2606.18524",
  "paper_url": "https://arxiv.org/abs/2606.18524",
  "lens": [
    "recurrent",
    "analysis"
  ],
  "representation": [
    "hidden-state"
  ],
  "computation": [
    "recurrent",
    "depth-wise"
  ],
  "training": [
    "pretraining"
  ],
  "questions": [
    "theory",
    "scaling",
    "efficiency"
  ],
  "evidence": [
    "formal",
    "empirical"
  ],
  "modality": [
    "language"
  ],
  "summary": "Scales each residual update inversely with the number of shared-weight loops, while keeping the learning rate independent of that loop count. Derives the scaling rule and tests whether hyperparameters transfer across loop counts and model depths during language-model pretraining.",
  "provenance": "Primary-source literature review, 2026-11-03",
  "previous_titles": [
    "On the Residual Scaling of Looped Transformers: Stability and Transferability"
  ],
  "author_names": [
    "Shaowen Wang",
    "Bingrui Li",
    "Tansheng Zhu",
    "Ge Zhang",
    "Wenhao Huang",
    "Shen Yan",
    "Jian Li"
  ],
  "record_url": "https://latent-implicit-thinking.github.io/papers/2606.18524/",
  "legacy_record_url": "https://latent-implicit-thinking.github.io/papers/#paper-2606-18524",
  "citation_date": "2026-06-16",
  "json_url": "https://latent-implicit-thinking.github.io/papers/2606.18524/record.json",
  "markdown_url": "https://latent-implicit-thinking.github.io/papers/2606.18524/index.md",
  "bibtex_url": "https://latent-implicit-thinking.github.io/papers/2606.18524/citation.bib",
  "csl_url": "https://latent-implicit-thinking.github.io/papers/2606.18524/citation.csl.json",
  "pdf_url": "https://arxiv.org/pdf/2606.18524"
}