Instruction Learning Paradigms: A Dual Perspective on White-box and Black-box LLMs

TL;DR

This paper proposes a novel framework that combines white-box and black-box large language models to optimize instructions, improving performance across diverse tasks through semantic fusion and iterative refinement.

Contribution

It introduces a unified approach that merges black-box initialization with white-box interpretability, enhancing instruction quality and adaptability in LLMs.

Findings

Outperforms state-of-the-art baselines on various tasks

Enhances instruction quality through semantic similarity constraints

Demonstrates scalability and efficiency in diverse scenarios

Abstract

Optimizing instructions for large language models (LLMs) is critical for harnessing their full potential in complex and diverse tasks. However, relying solely on white-box approaches demands extensive computational resources and offers limited representational capacity, while black-box models can incur prohibitive financial costs. To address these challenges, we introduce a novel framework that seamlessly merges the strengths of both paradigms. Black-box models provide high-quality, diverse instruction initializations, and white-box models supply fine-grained interpretability through hidden states and output features. By enforcing a semantic similarity constraint, these components fuse into a unified high-dimensional representation that captures deep semantic and structural nuances, enabling an iterative optimization process to refine instruction quality and adaptability. Extensive evaluations across a broad spectrum of tasks-ranging from complex reasoning to cross-lingual generalization-demonstrate that our approach consistently outperforms state-of-the-art baselines. This fusion of black-box initialization with advanced semantic refinement yields a scalable and efficient solution, paving the way for next-generation LLM-driven applications in diverse real-world scenarios. The source code will be released soon.