{
  "id": "2610.22796",
  "title": "Causal Latent Workspaces for Looped Transformers",
  "authors": "Ilia Zaznov",
  "first_public": "2026-10-16",
  "added_at": "2026-10-22",
  "venue": "arXiv",
  "arxiv": "2610.22796",
  "paper_url": "https://arxiv.org/abs/2610.22796",
  "html_url": "https://arxiv.org/html/2610.22796v1",
  "lens": [
    "recurrent"
  ],
  "representation": [
    "hidden-state",
    "cache-memory"
  ],
  "computation": [
    "recurrent",
    "stateful"
  ],
  "training": [
    "pretraining"
  ],
  "questions": [
    "architecture",
    "efficiency"
  ],
  "evidence": [
    "empirical"
  ],
  "modality": [
    "language"
  ],
  "summary": "Adds a small, content-addressed memory to a looped transformer, rebuilding causal prefix summaries between shared-layer passes. Tests whether this extra information path improves next-byte prediction under matched parameter and training budgets.",
  "provenance": "Primary-source literature review, 2026-10-22",
  "author_names": [
    "Ilia Zaznov"
  ],
  "record_url": "https://latent-implicit-thinking.github.io/papers/2610.22796/",
  "legacy_record_url": "https://latent-implicit-thinking.github.io/papers/#paper-2610-22796",
  "citation_date": "2026-10-16",
  "json_url": "https://latent-implicit-thinking.github.io/papers/2610.22796/record.json",
  "markdown_url": "https://latent-implicit-thinking.github.io/papers/2610.22796/index.md",
  "bibtex_url": "https://latent-implicit-thinking.github.io/papers/2610.22796/citation.bib",
  "csl_url": "https://latent-implicit-thinking.github.io/papers/2610.22796/citation.csl.json",
  "pdf_url": "https://arxiv.org/pdf/2610.22796"
}