GATE: Graph-based Adaptive Tool Evolution Across Diverse Tasks
This addresses the need for more efficient and versatile tool-making frameworks for LLMs, offering incremental improvements over existing methods.
The paper tackles the problem of inefficient and single-task tool-making in LLMs by proposing GATE, a graph-based adaptive framework that dynamically constructs and evolves reusable tools across diverse tasks, achieving up to 4.3x faster milestone completion in Minecraft and average improvements of 9.23% in code generation and 10.03% in agent tasks.
Large Language Models (LLMs) have shown great promise in tool-making, yet existing frameworks often struggle to efficiently construct reliable toolsets and are limited to single-task settings. To address these challenges, we propose GATE (Graph-based Adaptive Tool Evolution), an adaptive framework that dynamically constructs and evolves a hierarchical graph of reusable tools across multiple scenarios. We evaluate GATE on open-ended tasks (Minecraft), agent-based tasks (TextCraft, DABench), and code generation tasks (MATH, Date, TabMWP). Our results show that GATE achieves up to 4.3x faster milestone completion in Minecraft compared to the previous SOTA, and provides an average improvement of 9.23% over existing tool-making methods in code generation tasks and 10.03% in agent tasks. GATE demonstrates the power of adaptive evolution, balancing tool quantity, complexity, and functionality while maintaining high efficiency. Code and data are available at \url{https://github.com/ayanami2003/GATE}.