Salvato in:
Dettagli Bibliografici
Autori principali: Riechers, Paul M., Elliott, Thomas J., Shai, Adam S.
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2507.07432
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908447101943808
author Riechers, Paul M.
Elliott, Thomas J.
Shai, Adam S.
author_facet Riechers, Paul M.
Elliott, Thomas J.
Shai, Adam S.
contents We show that deep neural networks, including transformers and RNNs, pretrained as usual on next-token prediction, intrinsically discover and represent beliefs over 'quantum' and 'post-quantum' low-dimensional generative models of their training data -- as if performing iterative Bayesian updates over the latent state of this world model during inference as they observe more context. Notably, neural nets easily find these representation whereas there is no finite classical circuit that would do the job. The corresponding geometric relationships among neural activations induced by different input sequences are found to be largely independent of neural-network architecture. Each point in this geometry corresponds to a history-induced probability density over all possible futures, and the relative displacement of these points reflects the difference in mechanism and magnitude for how these distinct pasts affect the future.
format Preprint
id arxiv_https___arxiv_org_abs_2507_07432
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Neural networks leverage nominally quantum and post-quantum representations
Riechers, Paul M.
Elliott, Thomas J.
Shai, Adam S.
Machine Learning
Quantum Physics
We show that deep neural networks, including transformers and RNNs, pretrained as usual on next-token prediction, intrinsically discover and represent beliefs over 'quantum' and 'post-quantum' low-dimensional generative models of their training data -- as if performing iterative Bayesian updates over the latent state of this world model during inference as they observe more context. Notably, neural nets easily find these representation whereas there is no finite classical circuit that would do the job. The corresponding geometric relationships among neural activations induced by different input sequences are found to be largely independent of neural-network architecture. Each point in this geometry corresponds to a history-induced probability density over all possible futures, and the relative displacement of these points reflects the difference in mechanism and magnitude for how these distinct pasts affect the future.
title Neural networks leverage nominally quantum and post-quantum representations
topic Machine Learning
Quantum Physics
url https://arxiv.org/abs/2507.07432