{
  "id": "2503.24198",
  "title": "TwT: Thinking without Tokens by Habitual Reasoning Distillation with Multi-Teachers' Guidance",
  "first_public": "2025-03-31",
  "paper_url": "https://arxiv.org/abs/2503.24198",
  "authors": "Jingxian Xu, Mengyu Zhou, Weichang Liu, Hanbing Liu, Shi Han, Dongmei Zhang",
  "arxiv": "2503.24198",
  "added_at": "2026-11-09",
  "venue": "arXiv",
  "provenance": "Primary-source literature review, 2026-11-09",
  "html_url": "https://arxiv.org/html/2503.24198v2",
  "lens": [
    "internalized"
  ],
  "representation": [
    "hidden-state"
  ],
  "computation": [
    "depth-wise"
  ],
  "training": [
    "distillation",
    "curriculum"
  ],
  "questions": [
    "efficiency",
    "capability"
  ],
  "evidence": [
    "empirical"
  ],
  "modality": [
    "language"
  ],
  "summary": "Selects diverse worked solutions from multiple teachers, then trains a student on full solutions, compressed solutions, and finally answers alone. The intermediate compression stage uses the student’s responses to guide how a teacher shortens its explanations.",
  "author_names": [
    "Jingxian Xu",
    "Mengyu Zhou",
    "Weichang Liu",
    "Hanbing Liu",
    "Shi Han",
    "Dongmei Zhang"
  ],
  "record_url": "https://latent-implicit-thinking.github.io/papers/2503.24198/",
  "legacy_record_url": "https://latent-implicit-thinking.github.io/papers/#paper-2503-24198",
  "citation_date": "2025-03-31",
  "json_url": "https://latent-implicit-thinking.github.io/papers/2503.24198/record.json",
  "markdown_url": "https://latent-implicit-thinking.github.io/papers/2503.24198/index.md",
  "bibtex_url": "https://latent-implicit-thinking.github.io/papers/2503.24198/citation.bib",
  "csl_url": "https://latent-implicit-thinking.github.io/papers/2503.24198/citation.csl.json",
  "pdf_url": "https://arxiv.org/pdf/2503.24198"
}