Adaptive-Solver Framework for Dynamic Strategy Selection in Large Language Model Reasoning
Keywords: reasoning, large language models, natural language processing, prompting, adaptive-solver framework
TL;DR: We propose the Adaptive-Solver framework, an approach designed to dynamically tailor solving strategies for LLMs across diverse reasoning scenarios.
Abstract: As the field of artificial intelligence evolves, Large Language Models (LLMs) are showcasing impressive ability in handling complex reasoning tasks. Researchers have developed various techniques utilizing LLMs to tackle these challenges. In real-world situations, problems often span a spectrum of complexities. Humans inherently adjust their problem-solving approaches based on task complexity. However, most methodologies that leverage LLMs tend to adopt a uniform approach: utilizing consistent models, prompting methods, and degrees of problem decomposition, regardless of the problem complexity. Inflexibility of these methods can bring unnecessary computational overhead or sub-optimal performance. To address this problem, we introduce an Adaptive-Solver framework. It strategically modulates solving strategies based on the difficulties of the problems. Given an initial solution, the framework functions with two primary modules. The initial evaluation module assesses the adequacy of the current solution. If improvements are needed, the subsequent adaptation module comes into play. Within this module, three key adaptation strategies are employed: (1) Model Adaptation: Switching to a stronger LLM when a weaker variant is inadequate. (2) Prompting Method Adaptation: Alternating between different prompting techniques to suit the problem's nuances. (3) Decomposition Granularity Adaptation: Breaking down a complex problem into more fine-grained sub-questions to enhance solvability. Through such dynamic adaptations, our framework not only enhances computational efficiency but also elevates the overall performance. This dual-benefit ensures both the efficiency of the system for simpler tasks and the precision required for more complex questions. Experimental results from complex reasoning benchmarks reveal that the prompting method adaptation and decomposition granularity adaptation within the Adaptive-Solver framework enhance performance across all tasks. Furthermore, the model adaptation approach significantly reduces API costs (up to 50%) while maintaining superior performance.
Primary Area: generative models
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 3313
Loading