{
  "abstract": "Representations of translated sentences are similar in the inner layers of multilingual language models -- an observation connected to the platonic representation hypothesis, yet unexplained theoretically. We provide an explanation based on the assumption that data have a hidden hierarchical structure whose abstract levels are shared across languages while surface levels are modality- or language-specific. Concretely, we generate synthetic languages from probabilistic context-free grammars sharing upper-level but not lower-level production rules. In this setting the Bayes-optimal next-token predictor is belief propagation (BP); encoding its messages in successive layers yields analytical predictions that agree well with transformers trained on the same data. The framework explains why cross-lingual similarity peaks in middle layers, coexists with language-specific structure, and strengthens with language proximity, model quality and data exposure. It distinguishes similarity (shared neighborhood geometry) from alignment (shared coordinates), showing that the latter occurs when code-switched data, i.e. mixed-language sentences, are abundant enough. It further predicts that subtracting from each layer the component linearly predictable from the preceding one increases cross-lingual similarity, which we confirm in pretrained LLMs.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2610.07168",
  "author_total": 6,
  "authors": [
    {
      "name": "Darshil Doshi"
    },
    {
      "name": "Wenjie Zhou"
    },
    {
      "name": "Corinna Elena Wegner"
    },
    {
      "name": "Daniel J. Korchinski"
    },
    {
      "name": "Santiago Acevedo"
    },
    {
      "name": "Matthieu Wyart"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/9hewc3j5ffxng.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2610.07168",
  "fulltext_url": null,
  "id": "9hewc3j5ffxng",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2610.07168",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.CL"
    ],
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "stat.ML"
    ],
    [
      "openalex",
      "1702|Topic Modeling"
    ]
  ],
  "openalex": "W7221051225",
  "pdf_url": "https://arxiv.org/pdf/2610.07168",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-05",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/9hewc3j5ffxng.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:27:32Z",
      "native_id": "2610.07168",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2610.07168"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:stat.ML",
      "subfield": "statistics"
    }
  ],
  "title": "A theory of platonic representations in language models",
  "title_text": "A theory of platonic representations in language models",
  "type": "preprint",
  "updated_at": "2026-10-30T12:27:32Z",
  "url": "https://research.draconitic.com/paper/9hewc3j5ffxng",
  "venue": "arXiv",
  "versions": []
}