In-Context Learning State Vector with Inner and Momentum Optimization

TL;DR

This paper introduces inner and momentum optimization techniques for refining state vectors in in-context learning, leading to improved performance and state-of-the-art results on various tasks with large language models.

Contribution

It proposes novel inner and momentum optimization methods for state vectors in ICL and introduces a divide-and-conquer aggregation approach for lengthy demonstrations.

Findings

Optimization enhances state vector quality and task performance.

Achieves state-of-the-art results on multiple benchmarks.

Effective in both zero-shot and few-shot settings.

Abstract

Large Language Models (LLMs) have exhibited an impressive ability to perform In-Context Learning (ICL) from only a few examples. Recent works have indicated that the functions learned by ICL can be represented through compressed vectors derived from the transformer. However, the working mechanisms and optimization of these vectors are yet to be thoroughly explored. In this paper, we address this gap by presenting a comprehensive analysis of these compressed vectors, drawing parallels to the parameters trained with gradient descent, and introduce the concept of state vector. Inspired by the works on model soup and momentum-based gradient descent, we propose inner and momentum optimization methods that are applied to refine the state vector progressively as test-time adaptation. Moreover, we simulate state vector aggregation in the multiple example setting, where demonstrations comprising numerous examples are usually too lengthy for regular ICL, and further propose a divide-and-conquer aggregation method to address this challenge. We conduct extensive experiments using Llama-2 and GPT-J in both zero-shot setting and few-shot setting. The experimental results show that our optimization method effectively enhances the state vector and achieves the state-of-the-art performance on diverse tasks. Code is available at https://github.com/HITsz-TMG/ICL-State-Vector