Additive Multi-Step Markov Chains and the Curse of Dimensionality in Large Language Models

O. V. Usatenko, S. S. Melnyk, G. M. Pritula · 2026

Read Additive Multi-Step Markov Chains and the Curse of Dimensionality in Large Language Models by O. V. Usatenko, S. S. Melnyk, G. M. Pritula (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-2603-04412.txt.

Description

Usatenko and colleagues model language model dynamics with additive multi-step Markov chains. Such models live in huge state spaces where token and hidden links resist classical Markov reduction. Additive chains split next token odds across many historical depths, taming high order explosion. The main result ties an additive chain to a stepwise memory chain. That link extends information temperature from stepwise chains to additive N-order chains.

Catalog record

CREATORO. V. Usatenko, S. S. Melnyk, G. M. Pritula
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE41773
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.04412

Browse the full library · About the Tower of Babel