ClusterFewshot: Improving Few-shot Optimization for LLMs workflow

2026-09-22

AI Summary

本文提出ClusterFewshot方法，通过结合语义结构和效用评分来选择少量示例，以优化大型语言模型的工作流程，提高准确性和降低成本。

Abstract

The performance of large language model (LLM) workflows often depends on selecting a small set of in-context demonstrations to guide model behavior on new tasks. Recent methods improve this process by augmenting prompts with successful reasoning paths. However, their demonstration selection relies on random sampling or metric-based rankings, overlooking the semantic structure of the task. We propose ClusterFewshot, a strategy that combines semantic structuring with utility-aware scoring to construct representative and effective few-shot demonstration sets. Evaluated within DSPy-based pipelines, ClusterFewshot substantially reduces optimization cost across multiple benchmarks, while consistently improving accuracy relative to prior bootstrap-based methods in both standalone prompt tuning and hybrid prompt-weight optimization.

Problem

• few-shot optimization
• semantic structure
• demonstration selection

Innovation

• semantic structuring
• utility-aware scoring
• few-shot demonstrations