Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ToolTree: Efficient LLM Agent Tool Planning via Dual-Feedback Monte Carlo Tree Search and Bidirectional Pruning

Shuo Yang, Soyeon Caren Han, Yihao Ding, Shuhe Wang, Eduard Hoy

Abstract

Large Language Model (LLM) agents are increasingly applied to complex, multi-step tasks that require interaction with diverse external tools across various domains. However, current LLM agent tool planning methods typically rely on greedy, reactive tool selection strategies that lack foresight and fail to account for inter-tool dependencies. In this paper, we present ToolTree, a novel Monte Carlo tree search-inspired planning paradigm for tool planning. ToolTree explores possible tool usage trajectories using a dual-stage LLM evaluation and bidirectional pruning mechanism that enables the agent to make informed, adaptive decisions over extended tool-use sequences while pruning less promising branches before and after the tool execution. Empirical evaluations across both open-set and closed-set tool planning tasks on 4 benchmarks demonstrate that ToolTree consistently improves performance while keeping the highest efficiency, achieving an average gain of around 10\% compared to the state-of-the-art planning paradigm.

ToolTree: Efficient LLM Agent Tool Planning via Dual-Feedback Monte Carlo Tree Search and Bidirectional Pruning

Abstract

Large Language Model (LLM) agents are increasingly applied to complex, multi-step tasks that require interaction with diverse external tools across various domains. However, current LLM agent tool planning methods typically rely on greedy, reactive tool selection strategies that lack foresight and fail to account for inter-tool dependencies. In this paper, we present ToolTree, a novel Monte Carlo tree search-inspired planning paradigm for tool planning. ToolTree explores possible tool usage trajectories using a dual-stage LLM evaluation and bidirectional pruning mechanism that enables the agent to make informed, adaptive decisions over extended tool-use sequences while pruning less promising branches before and after the tool execution. Empirical evaluations across both open-set and closed-set tool planning tasks on 4 benchmarks demonstrate that ToolTree consistently improves performance while keeping the highest efficiency, achieving an average gain of around 10\% compared to the state-of-the-art planning paradigm.
Paper Structure (38 sections, 3 equations, 10 figures, 12 tables)

This paper contains 38 sections, 3 equations, 10 figures, 12 tables.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Problem Definition: Tool Planning.
  4. Tree Search-enhanced planning.
  5. Monte Carlo Tree Search (MCTS).
  6. Proposed Method: ToolTree
  7. Overview
  8. Dual Evaluation and Pruning
  9. Experiments
  10. Experiment Setup
  11. Closed-set Tool Planning on GTA and m&m
  12. Open-set Tool Planning on ToolBench and RestBench
  13. Analysis
  14. Related Work
  15. Conclusion
  16. ...and 23 more sections

Figures (10)

  • Figure 1: Comparison of ToolTree with greedy search and search-based tool planning. Our ToolTree chooses the optimal tool trajectory and answers correctly with 20.
  • Figure 2: Architecture overview of ToolTree. An input query is processed sequentially via iterative dual evaluation-guided Monte Carlo Tree Search, including selection, pre-evaluation, expansion, execution, post-evaluation and backward-propagation. The Answer Predictor then incorporates the tool trajectories with the highest reward found by the MCTS to produce the final prediction.
  • Figure 3: Progressive efficiency analysis across step limits. (a) Performance vs. step limit; (b) Runtime vs. step limit; (c) Efficiency vs. step limit. ToolTree achieves the highest efficiency compared with baselines. mprovements are largest for step limits between 12 and 64.
  • Figure 4: Efficiency comparison of ToolTree and its pruning variants on nodes and rollouts.
  • Figure 5: Analysis of Performance with respect to model size on Qwen and LLaMA family.
  • ...and 5 more figures