Try, Check and Retry: A Divide-and-Conquer Framework for Boosting Long-context Tool-Calling Performance of LLMs

TL;DR

This paper introduces Tool-DC, a divide-and-conquer framework utilizing a Try-Check-Retry paradigm to enhance long-context tool-calling performance in LLMs, effectively managing large and noisy tool sets.

Contribution

The paper proposes a novel divide-and-conquer framework with a Try-Check-Retry paradigm, improving LLMs' tool-calling accuracy in long contexts, with both training-free and training-based variants.

Findings

Tool-DC (TF) achieves up to +25.10% gains on benchmarks.

Tool-DC (TB) enables smaller LLMs to match or surpass larger proprietary models.

Both variants outperform existing methods significantly.

Abstract

Tool-calling empowers Large Language Models (LLMs) to interact with external environments. However, current methods often struggle to handle massive and noisy candidate tools in long-context tool-calling tasks, limiting their real-world application. To this end, we propose Tool-DC, a Divide-and-Conquer framework for boosting tool-calling performance of LLMs. The core of Tool-DC is to reduce the reasoning difficulty and make full use of self-reflection ability of LLMs via a "Try-Check-Retry" paradigm. Specifically, Tool-DC involves two variants: 1) the training-free Tool-DC (TF), which is plug-and-play and flexible; 2) the training-based Tool-DC (TB), which is more inference-efficient. Extensive experiments show that both Tool-DC methods outperform their counterparts by a clear margin. Tool-DC (TF) brings up to +25.10% average gains against the baseline on BFCL and ACEBench benchmarks, while Tool-DC (TB) enables Qwen2.5-7B to achieve comparable or even better performance than proprietary LLMs, e.g., OpenAI o3 and Claude-Haiku-4.5.