Metraipolis › archive › dc_cb7eaa40fcb6

RiLM: Parameter-Efficient Language Modeling via Geodesic Decoding — arXiv:2609.10305

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.10305
License (as archived)
CC0-1.0
Archived
2026-10-16T07:27:13.019Z
SHA-256 of the archived text
c2bba465eb3c47cffe5091a80471ce783a2c0a4947fc419c30d7f7433c96014a
Stable URL of this page
https://metraipolis.com/documento/dc_cb7eaa40fcb6

How to cite

RiLM: Parameter-Efficient Language Modeling via Geodesic Decoding — arXiv:2609.10305 — arXiv (en), https://arxiv.org/abs/2609.10305 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cb7eaa40fcb6 (Metraipolis archive dc_cb7eaa40fcb6, sha256 c2bba465eb3c47cffe5091a80471ce783a2c0a4947fc419c30d7f7433c96014a)
BibTeX
@misc{dccb7eaa40fcb6,
  title = {RiLM: Parameter-Efficient Language Modeling via Geodesic Decoding — arXiv:2609.10305},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.10305},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cb7eaa40fcb6; sha256 c2bba465eb3c47cffe5091a80471ce783a2c0a4947fc419c30d7f7433c96014a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.10305. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Language models under one million parameters matter for edge deployment, domain adaptation, and reproducible research, yet a two-layer LSTM or Transformer at embedding width d = 128 still spends roughly one third of its capacity on the output matrix W_out in R^(d x |V|). We propose Riemannian Language Models (RiLM), which remove that layer entirely: context unfolds as a trajectory on a Riemannian manifold, and next-token probabilities arise from squared geodesic distance between the current state and vocabulary embeddings. The same embedding map serves input and output -- decoding is geometry. We instantiate the framework on flat R^d (Flat RiLM) and the Poincare ball H^d (HypRiLM) with a shared MLP composition map phi (~290k parameters, d = 128, |V| = 2000). Across five seeds on WikiText-2, HypRiLM reaches 54.2 +/- 0.2 validation perplexity versus 87.6 +/- 0.6 for Flat RiLM; tied and matched LSTM, Transformer, and SSM controls remain at 113-147 PPL on WT-2 -- HypRiLM leads by roughly 2x over the strongest tied recurrent baseline (SSM, 113.0 +/- 3.8). Penn Treebank and a 10k-vocabulary stress test confirm that geodesic decoding transfers across corpora and larger |V|, while hyperbolic curvature helps selectively. We also characterize boundary collapse in naive hyperbolic recurrence and show how Mobius stabilization restores trainability. Claims are scoped to controlled small-model comparisons, not full-vocabulary state of the art.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-10 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.10305 — 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 2026-10-16T07:00:16.401Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-10?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).