How Does Unfaithful Reasoning Emerge from Autoregressive Training? A Study of Synthetic Experiments

Fuxin Wang, Amr Alazali, Yiqiao Zhong · 2026

Read How Does Unfaithful Reasoning Emerge from Autoregressive Training? A Study of Synthetic Experiments by Fuxin Wang, Amr Alazali, Yiqiao Zhong (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-01017.txt.

Description

Fuxin Wang and co-authors probe why step-by-step model thought can stray from the path that led to its own answer. They train small nets on noisy modular math sums, where true rules are known. With low noise, nets learn true step rules, pulled toward simple rules. Past a noise tipping point, thought slides via a mixed phase into skipping steps, with a brief surge in guess spread. Deep looks inside hint the nets learn to sense doubt by fixing clash steps, a faint form of self-check.

Catalog record

CREATORFuxin Wang, Amr Alazali, Yiqiao Zhong
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE67138
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602.01017

Browse the full library · About the Tower of Babel