Metraipolis › archive › dc_c2f5874495ec

Reasoning on the Simplex: Geometric Fixed-Point Models — arXiv:2609.33540

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.33540
License (as archived)
CC0-1.0
Archived
2026-10-07T20:56:54.510Z
SHA-256 of the archived text
896b79821cbf54dacb97cc93b1b08b9d17ae0d37fe3ade88c28ee1248b62df86
Stable URL of this page
https://metraipolis.com/documento/dc_c2f5874495ec

How to cite

Reasoning on the Simplex: Geometric Fixed-Point Models — arXiv:2609.33540 — arXiv (en), https://arxiv.org/abs/2609.33540 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c2f5874495ec (Metraipolis archive dc_c2f5874495ec, sha256 896b79821cbf54dacb97cc93b1b08b9d17ae0d37fe3ade88c28ee1248b62df86)
BibTeX
@misc{dcc2f5874495ec,
  title = {Reasoning on the Simplex: Geometric Fixed-Point Models — arXiv:2609.33540},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.33540},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c2f5874495ec; sha256 896b79821cbf54dacb97cc93b1b08b9d17ae0d37fe3ade88c28ee1248b62df86},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33540. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Looped reasoners spend test-time compute by iterating a weight-tied map, but a small residual does not mean the state is a fixed point when that map lives in unconstrained latent space. We propose Geometric Fixed-Point Reasoning (GFPR), in which the iterated state is the prediction itself: a field of categorical beliefs on a product of simplices, whose argmax is the answer at every step. Because the state is a belief, task structure can be imposed through compact convex relaxations, either as structured readouts or directly in the recurrent state; in the latter case the update remains a continuous self-map, so a fixed point exists for any parameters. At about 7M parameters, GFPR reaches 95.1% exact match on Sudoku-Extreme, 92.0% on Maze-Hard, and 100% sequence accuracy on S_5 length 128, above the published FPRM numbers at the same scale. The same update also trains a 201M language model on FineWeb-Edu in which each site is a distribution over the vocabulary; with 24 Picard steps it is above GPT-2 small on four zero-shot multiple-choice tasks and above GPT-2 medium on ARC-Easy.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.33540 — 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 2026-10-07T20:29:01.294Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).