EcoAct: Economic Agent Determines When to Register What Action
This addresses computational inefficiency for LLM agents using external tools, though it is incremental as it builds on existing reasoning pipelines.
The paper tackles the inefficiency of LLM agents indiscriminately registering all tools in their context, which increases computational costs, by introducing EcoAct, a method that selectively registers tools as needed, reducing computational costs by over 50% while maintaining performance.
Recent advancements have enabled Large Language Models (LLMs) to function as agents that can perform actions using external tools. This requires registering, i.e., integrating tool information into the LLM context prior to taking actions. Current methods indiscriminately incorporate all candidate tools into the agent's context and retain them across multiple reasoning steps. This process remains opaque to LLM agents and is not integrated into their reasoning procedures, leading to inefficiencies due to increased context length from irrelevant tools. To address this, we introduce EcoAct, a tool using algorithm that allows LLMs to selectively register tools as needed, optimizing context use. By integrating the tool registration process into the reasoning procedure, EcoAct reduces computational costs by over 50% in multiple steps reasoning tasks while maintaining performance, as demonstrated through extensive experiments. Moreover, it can be plugged into any reasoning pipeline with only minor modifications to the prompt, making it applicable to LLM agents now and future.