@inproceedings{pan-etal-2024-dynathink,
title = "{D}yna{T}hink: Fast or Slow? A Dynamic Decision-Making Framework for Large Language Models",
author = "Pan, Jiabao and
Zhang, Yan and
Zhang, Chen and
Liu, Zuozhu and
Wang, Hongwei and
Li, Haizhou",
editor = "Al-Onaizan, Yaser and
Bansal, Mohit and
Chen, Yun-Nung",
booktitle = "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing",
month = nov,
year = "2024",
address = "Miami, Florida, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.emnlp-main.814/",
doi = "10.18653/v1/2024.emnlp-main.814",
pages = "14686--14695",
abstract = "Large language models (LLMs) have demonstrated emergent capabilities across diverse reasoning tasks via popular Chains-of-Thought (COT) prompting. However, such a simple and fast COT approach often encounters limitations in dealing with complicated problems, while a thorough method, which considers multiple reasoning pathways and verifies each step carefully, results in slower inference. This paper addresses the challenge of enabling LLMs to autonomously select between fast and slow inference methods, thereby optimizing both efficiency and effectiveness. We introduce a dynamic decision-making framework that categorizes tasks into two distinct pathways: {\textquoteleft}Fast,' designated for tasks where the LLM quickly identifies a high-confidence solution, and {\textquoteleft}Slow,' allocated for tasks that the LLM perceives as complex and for which it has low confidence in immediate solutions as well as requiring more reasoning paths to verify. Experiments on five popular reasoning benchmarks demonstrated the superiority of the DynaThink over baselines. For example, when we compared it to strong COT with self-consistency baseline on the complicated MATH dataset, DynaThink achieved more than 3{\%} increase in accuracy with lower cost. The code will be made available upon publication."
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="pan-etal-2024-dynathink">
<titleInfo>
<title>DynaThink: Fast or Slow? A Dynamic Decision-Making Framework for Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jiabao</namePart>
<namePart type="family">Pan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yan</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chen</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zuozhu</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hongwei</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haizhou</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yaser</namePart>
<namePart type="family">Al-Onaizan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohit</namePart>
<namePart type="family">Bansal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yun-Nung</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Miami, Florida, USA</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Large language models (LLMs) have demonstrated emergent capabilities across diverse reasoning tasks via popular Chains-of-Thought (COT) prompting. However, such a simple and fast COT approach often encounters limitations in dealing with complicated problems, while a thorough method, which considers multiple reasoning pathways and verifies each step carefully, results in slower inference. This paper addresses the challenge of enabling LLMs to autonomously select between fast and slow inference methods, thereby optimizing both efficiency and effectiveness. We introduce a dynamic decision-making framework that categorizes tasks into two distinct pathways: ‘Fast,’ designated for tasks where the LLM quickly identifies a high-confidence solution, and ‘Slow,’ allocated for tasks that the LLM perceives as complex and for which it has low confidence in immediate solutions as well as requiring more reasoning paths to verify. Experiments on five popular reasoning benchmarks demonstrated the superiority of the DynaThink over baselines. For example, when we compared it to strong COT with self-consistency baseline on the complicated MATH dataset, DynaThink achieved more than 3% increase in accuracy with lower cost. The code will be made available upon publication.</abstract>
<identifier type="citekey">pan-etal-2024-dynathink</identifier>
<identifier type="doi">10.18653/v1/2024.emnlp-main.814</identifier>
<location>
<url>https://aclanthology.org/2024.emnlp-main.814/</url>
</location>
<part>
<date>2024-11</date>
<extent unit="page">
<start>14686</start>
<end>14695</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T DynaThink: Fast or Slow? A Dynamic Decision-Making Framework for Large Language Models
%A Pan, Jiabao
%A Zhang, Yan
%A Zhang, Chen
%A Liu, Zuozhu
%A Wang, Hongwei
%A Li, Haizhou
%Y Al-Onaizan, Yaser
%Y Bansal, Mohit
%Y Chen, Yun-Nung
%S Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
%D 2024
%8 November
%I Association for Computational Linguistics
%C Miami, Florida, USA
%F pan-etal-2024-dynathink
%X Large language models (LLMs) have demonstrated emergent capabilities across diverse reasoning tasks via popular Chains-of-Thought (COT) prompting. However, such a simple and fast COT approach often encounters limitations in dealing with complicated problems, while a thorough method, which considers multiple reasoning pathways and verifies each step carefully, results in slower inference. This paper addresses the challenge of enabling LLMs to autonomously select between fast and slow inference methods, thereby optimizing both efficiency and effectiveness. We introduce a dynamic decision-making framework that categorizes tasks into two distinct pathways: ‘Fast,’ designated for tasks where the LLM quickly identifies a high-confidence solution, and ‘Slow,’ allocated for tasks that the LLM perceives as complex and for which it has low confidence in immediate solutions as well as requiring more reasoning paths to verify. Experiments on five popular reasoning benchmarks demonstrated the superiority of the DynaThink over baselines. For example, when we compared it to strong COT with self-consistency baseline on the complicated MATH dataset, DynaThink achieved more than 3% increase in accuracy with lower cost. The code will be made available upon publication.
%R 10.18653/v1/2024.emnlp-main.814
%U https://aclanthology.org/2024.emnlp-main.814/
%U https://doi.org/10.18653/v1/2024.emnlp-main.814
%P 14686-14695
Markdown (Informal)
[DynaThink: Fast or Slow? A Dynamic Decision-Making Framework for Large Language Models](https://aclanthology.org/2024.emnlp-main.814/) (Pan et al., EMNLP 2024)
ACL