Metraipolis › archive › dc_b3dcc27042ee

On the (Intuitionistic) Logic of Next-Token Prediction — arXiv:2608.08145

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.08145
License (as archived)
CC0-1.0
Archived
2026-10-19T09:17:55.619Z
SHA-256 of the archived text
3357313491ed6ecded3d823718a6de091dbd4fea4987187328fafd591708deab
Stable URL of this page
https://metraipolis.com/documento/dc_b3dcc27042ee

How to cite

On the (Intuitionistic) Logic of Next-Token Prediction — arXiv:2608.08145 — arXiv (en), https://arxiv.org/abs/2608.08145 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b3dcc27042ee (Metraipolis archive dc_b3dcc27042ee, sha256 3357313491ed6ecded3d823718a6de091dbd4fea4987187328fafd591708deab)
BibTeX
@misc{dcb3dcc27042ee,
  title = {On the (Intuitionistic) Logic of Next-Token Prediction — arXiv:2608.08145},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.08145},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b3dcc27042ee; sha256 3357313491ed6ecded3d823718a6de091dbd4fea4987187328fafd591708deab},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.08145. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

We model in intuitionistic implicational logic the key enabler of today's GenerativeAI: the next-token prediction in autoregressive causal neural networks. In our framework, next-token prediction corresponds to modus ponens, and sequence processing becomes constructive proof extension under the Curry-Howard correspondence. Our Prolog-based specialized theorem provers validate fundamental properties of the neural models, among which relations between commutative vs. non-commutative sequencing and single-token vs. multi-token prediction choices. We derive a neural architecture equivalent to multiplicative RNNs that arises naturally from a proof-theoretic interpretation of next-token prediction as nested intuitionistic implication and position the model relative to transformers, state-space models and recursive LLMs.

Tema degli esploratori: matematica e informatica teorica (retro-catalogo arXiv: annunci del 2026-08-11 in cs.LO).
Categorie arXiv: Logic in Computer Science (cs.LO).
Pagina del lavoro: https://arxiv.org/abs/2608.08145 — 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 2026-10-19T09:14:28.538Z dalla pagina https://arxiv.org/catchup/cs.LO/2026-08-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).