Metraipolis › archive › dc_2bc9344e6fbf

SanSi: A Looped Typed Decision Model for System 1.5 Thinking — arXiv:2610.07730

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.07730
License (as archived)
CC0-1.0
Archived
2026-10-07T11:21:06.561Z
SHA-256 of the archived text
ca1044f7ec9e3cbc54c18b8a7cc63ade06e979de6b4c20590aaf70d8f3f46dcb
Stable URL of this page
https://metraipolis.com/documento/dc_2bc9344e6fbf

How to cite

SanSi: A Looped Typed Decision Model for System 1.5 Thinking — arXiv:2610.07730 — arXiv (en), https://arxiv.org/abs/2610.07730 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2bc9344e6fbf (Metraipolis archive dc_2bc9344e6fbf, sha256 ca1044f7ec9e3cbc54c18b8a7cc63ade06e979de6b4c20590aaf70d8f3f46dcb)
BibTeX
@misc{dc2bc9344e6fbf,
  title = {SanSi: A Looped Typed Decision Model for System 1.5 Thinking — arXiv:2610.07730},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.07730},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2bc9344e6fbf; sha256 ca1044f7ec9e3cbc54c18b8a7cc63ade06e979de6b4c20590aaf70d8f3f46dcb},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.07730. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Typed decision models answer a declared question without generating text: a decision head returns a probability for each of the declared options in a single forward pass. A single pass is fast, intuitive System 1 thinking. We study what lies between one pass and generated reasoning: looping, in which the same layers are recursively applied several times before one typed readout. Each loop lets the model revise its hidden state before it commits to an answer, without generating a token; we call this System 1.5 thinking. We propose SanSi, which turns a pre-trained looped language model into a typed decision model. The option probabilities are read after every loop, and every loop is trained with a proper scoring rule, so that one model serves every budget from one loop to eight in a single run. On 10,027 test decisions from 59 sources, SanSi reaches 72.0% accuracy: 13.5 points above a non-looped model of the same shape trained with the same recipe, 5.3 points above a newer non-looped model of its size, and 1.8 points below one with three times the parameters. On two depth-controlled tasks, loops extend the solvable depth beyond the depths seen in training, where the larger single-pass model fails. Used as the judge for policy optimization with reinforcement learning, without gold answers, SanSi raises the generator's F1 by 7.7 points.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.07730 — 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 2026-10-07T11:15:24.758Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).