A Survey of Test-Time Compute: From Intuitive Inference to Deliberate Reasoning

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Ji, Yixin, Li, Juntao, Xiang, Yang, Ye, Hai, Wu, Kaixin, Yao, Kai, Xu, Jia, Mo, Linjian, Zhang, Min
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908426008788992
author Ji, Yixin
Li, Juntao
Xiang, Yang
Ye, Hai
Wu, Kaixin
Yao, Kai
Xu, Jia
Mo, Linjian
Zhang, Min
author_facet Ji, Yixin
Li, Juntao
Xiang, Yang
Ye, Hai
Wu, Kaixin
Yao, Kai
Xu, Jia
Mo, Linjian
Zhang, Min
contents The remarkable performance of the o1 model in complex reasoning demonstrates that test-time compute scaling can further unlock the model's potential, enabling powerful System-2 thinking. However, there is still a lack of comprehensive surveys for test-time compute scaling. We trace the concept of test-time compute back to System-1 models. In System-1 models, test-time compute addresses distribution shifts and improves robustness and generalization through parameter updating, input modification, representation editing, and output calibration. In System-2 models, it enhances the model's reasoning ability to solve complex problems through repeated sampling, self-correction, and tree search. We organize this survey according to the trend of System-1 to System-2 thinking, highlighting the key role of test-time compute in the transition from System-1 models to weak System-2 models, and then to strong System-2 models. We also point out advanced topics and future directions.
format Preprint
id arxiv_https___arxiv_org_abs_2501_02497
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Survey of Test-Time Compute: From Intuitive Inference to Deliberate Reasoning
Ji, Yixin
Li, Juntao
Xiang, Yang
Ye, Hai
Wu, Kaixin
Yao, Kai
Xu, Jia
Mo, Linjian
Zhang, Min
Artificial Intelligence
Computation and Language
Machine Learning
The remarkable performance of the o1 model in complex reasoning demonstrates that test-time compute scaling can further unlock the model's potential, enabling powerful System-2 thinking. However, there is still a lack of comprehensive surveys for test-time compute scaling. We trace the concept of test-time compute back to System-1 models. In System-1 models, test-time compute addresses distribution shifts and improves robustness and generalization through parameter updating, input modification, representation editing, and output calibration. In System-2 models, it enhances the model's reasoning ability to solve complex problems through repeated sampling, self-correction, and tree search. We organize this survey according to the trend of System-1 to System-2 thinking, highlighting the key role of test-time compute in the transition from System-1 models to weak System-2 models, and then to strong System-2 models. We also point out advanced topics and future directions.
title A Survey of Test-Time Compute: From Intuitive Inference to Deliberate Reasoning
topic Artificial Intelligence
Computation and Language
Machine Learning
url https://arxiv.org/abs/2501.02497