Metraipolis › archive › dc_764faf5ea439

MUON-RT Reusing the Teacher's Trajectory for Momentum-Free Muon for LLM Self-Distillation — arXiv:2610.41104

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.41104
License (as archived)
CC0-1.0
Archived
2026-11-03T07:32:40.746Z
SHA-256 of the archived text
02e2a385b94364b652120861c54c9bd51eeb5af65b37d0cedd36d4180711447f
Stable URL of this page
https://metraipolis.com/documento/dc_764faf5ea439

How to cite

MUON-RT Reusing the Teacher's Trajectory for Momentum-Free Muon for LLM Self-Distillation — arXiv:2610.41104 — arXiv (en), https://arxiv.org/abs/2610.41104 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_764faf5ea439 (Metraipolis archive dc_764faf5ea439, sha256 02e2a385b94364b652120861c54c9bd51eeb5af65b37d0cedd36d4180711447f)
BibTeX
@misc{dc764faf5ea439,
  title = {MUON-RT Reusing the Teacher's Trajectory for Momentum-Free Muon for LLM Self-Distillation — arXiv:2610.41104},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.41104},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_764faf5ea439; sha256 02e2a385b94364b652120861c54c9bd51eeb5af65b37d0cedd36d4180711447f},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41104. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Self-distillation methods for large language models with an exponential moving average (EMA) teacher already retain a history of student updates. Muon nevertheless stores a separate momentum buffer, increasing persistent training memory. We propose Muon with Reused Teacher (Muon-RT), which reuses the teacher-student displacement to eliminate momentum buffers for hidden weight matrices. We show that this displacement encodes an exponentially weighted history of applied updates and smooths temporally uncorrelated noise under a local model. Muon-RT removes its radial component, normalizes it spectrally, and mixes it with the current gradient after Newton-Schulz (NS) orthogonalization. Under explicit regularity and alignment assumptions, we establish linear contraction of an expected objective-gap bound toward a residual error level. With Self-Distillation Fine-Tuning (SDFT) on Qwen2.5-7B-Instruct, Muon-RT improves average accuracy over AdamW by 2.6 points with 42% less persistent memory, and over momentum Muon by 5.4 points with 27% less persistent memory. With Self-Distillation Policy Optimization (SDPO) on Olmo-3-7B-Instruct, it improves average accuracy over AdamW by 2.0 points.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2610.41104 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-03T07:29:33.706Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).