DynaAct: Large Language Model Reasoning with Dynamic Action Spaces
This addresses the scalability and computational inefficiency in action space construction for complex reasoning problems, though it is incremental as it builds on existing methods with a novel optimization approach.
The paper tackles the problem of constructing optimal candidate action spaces for sequential decision-making by proposing DynaAct, a framework that automatically builds compact action spaces using large language models and submodular optimization, resulting in significant performance improvements on six benchmarks while maintaining efficient inference.
In modern sequential decision-making systems, the construction of an optimal candidate action space is critical to efficient inference. However, existing approaches either rely on manually defined action spaces that lack scalability or utilize unstructured spaces that render exhaustive search computationally prohibitive. In this paper, we propose a novel framework named \textsc{DynaAct} for automatically constructing a compact action space to enhance sequential reasoning in complex problem-solving scenarios. Our method first estimates a proxy for the complete action space by extracting general sketches observed in a corpus covering diverse complex reasoning problems using large language models. We then formulate a submodular function that jointly evaluates candidate actions based on their utility to the current state and their diversity, and employ a greedy algorithm to select an optimal candidate set. Extensive experiments on six diverse standard benchmarks demonstrate that our approach significantly improves overall performance, while maintaining efficient inference without introducing substantial latency. The implementation is available at https://github.com/zhaoxlpku/DynaAct.