Post-Grokking Collapse at the Representation-Readout Interface in Muon-Trained Transformers
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…
We haven't written up this one. arXiv cs.AI has the full story — the link below goes straight to it.
