{
  "id": "2607.22629",
  "title": "Masked Self-Distillation: Internalizing the Chain-of-Thought in Language Models",
  "authors": "Durgesh Kalwar, Vardhan Palod, Jaya Adithya Pavuluri, Subbarao Kambhampati",
  "previous_titles": [
    "Masked Distillation: Internalizing the Chain-of-Thought in Language Models"
  ],
  "first_public": "2026-06-18",
  "added_at": "2026-11-02",
  "venue": "arXiv",
  "arxiv": "2607.22629",
  "paper_url": "https://arxiv.org/abs/2607.22629",
  "html_url": "https://arxiv.org/html/2607.22629v3",
  "lens": [
    "internalized"
  ],
  "representation": [
    "hidden-state",
    "text-latent-hybrid"
  ],
  "computation": [
    "depth-wise",
    "token-wise"
  ],
  "training": [
    "self-distillation"
  ],
  "questions": [
    "capability",
    "efficiency",
    "generalization"
  ],
  "evidence": [
    "empirical"
  ],
  "modality": [
    "language",
    "symbolic"
  ],
  "summary": "Trains a no-think student with on-policy feedback from a copy that sees a prefix of the reasoning trace. Withholding more of the trace from the student teaches it to answer with fewer or no written reasoning steps.",
  "provenance": "Primary-source literature review, 2026-11-02",
  "author_names": [
    "Durgesh Kalwar",
    "Vardhan Palod",
    "Jaya Adithya Pavuluri",
    "Subbarao Kambhampati"
  ],
  "record_url": "https://latent-implicit-thinking.github.io/papers/2607.22629/",
  "legacy_record_url": "https://latent-implicit-thinking.github.io/papers/#paper-2607-22629",
  "citation_date": "2026-06-18",
  "json_url": "https://latent-implicit-thinking.github.io/papers/2607.22629/record.json",
  "markdown_url": "https://latent-implicit-thinking.github.io/papers/2607.22629/index.md",
  "bibtex_url": "https://latent-implicit-thinking.github.io/papers/2607.22629/citation.bib",
  "csl_url": "https://latent-implicit-thinking.github.io/papers/2607.22629/citation.csl.json",
  "pdf_url": "https://arxiv.org/pdf/2607.22629"
}