Few-shot Fine-tuning vs. In-context Learning: A Fair Comparison and Evaluation

TL;DR

This paper compares few-shot fine-tuning and in-context learning for language models, controlling for model size and examples, and finds both approaches have similar out-of-domain generalization, with robustness still a challenge.

Contribution

It provides a fair comparison of fine-tuning and in-context learning across various model sizes, clarifying their relative generalization capabilities.

Findings

Both methods show similar out-of-domain generalization.

Model size and number of examples significantly affect performance.

Robust task adaptation remains a challenge for both approaches.

Abstract

Few-shot fine-tuning and in-context learning are two alternative strategies for task adaptation of pre-trained language models. Recently, in-context learning has gained popularity over fine-tuning due to its simplicity and improved out-of-domain generalization, and because extensive evidence shows that fine-tuned models pick up on spurious correlations. Unfortunately, previous comparisons of the two approaches were done using models of different sizes. This raises the question of whether the observed weaker out-of-domain generalization of fine-tuned models is an inherent property of fine-tuning or a limitation of the experimental setup. In this paper, we compare the generalization of few-shot fine-tuning and in-context learning to challenge datasets, while controlling for the models used, the number of examples, and the number of parameters, ranging from 125M to 30B. Our results show that fine-tuned language models can in fact generalize well out-of-domain. We find that both approaches generalize similarly; they exhibit large variation and depend on properties such as model size and the number of examples, highlighting that robust task adaptation remains a challenge.