{
  "id": 429405,
  "title": "Post-Grokking Collapse at the Representation-Readout Interface in Muon-Trained Transformers",
  "url": "https://urgent.news/2026/08/07/post-grokking-collapse-at-the-representation-readout-interface-in",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-07T17:21:49.000Z",
  "source": {
    "name": "arXiv cs.AI",
    "slug": "arxiv-cs-ai",
    "url": "https://arxiv.org/abs/2608.07436v1"
  },
  "original_language": "en",
  "account": null,
  "summary": "Under the standard split, Muon gets hidden matrices and AdamW embeddings/output head. Muon groks modular addition faster, but its solutions do not hold. All nine configurations on $(a+b) \\bmod 113$ grok and later lose generalization. Across five seeds the selected AdamW reference falls below threshold on four, reaching 27.59%. Instability persists across two moduli, two widths, two training…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}