The Art of Scaling Test-Time Compute for Large Language Models

Aradhye Agarwal Ayan Sengupta Tanmoy Chakraborty · 2025

Read The Art of Scaling Test-Time Compute for Large Language Models by Aradhye Agarwal Ayan Sengupta Tanmoy Chakraborty (2025) 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-2512-02008.txt.

Description

Letting a language model think longer at answer time, rather than training it bigger, has become a popular route to better reasoning. Aradhye Agarwal, Ayan Sengupta, and Tanmoy Chakraborty survey in 2025 how to scale this test-time compute, comparing the well-known strategies side by side under identical conditions.

A fair comparison had been missing, and so had an account of how the best strategy shifts with model type and problem difficulty. The authors organize the options for spending extra inference compute and examine when each pays off. Their map helps builders decide where an extra moment of machine thought is worth its cost.

Catalog record

CREATORAradhye Agarwal Ayan Sengupta Tanmoy Chakraborty
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE33 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-02008

Browse the full library · About the Tower of Babel