On the Compositional Generalization Gap of In-Context Learning

TL;DR

This paper investigates how scaling large language models affects their ability to generalize compositionally in semantic parsing tasks, revealing that larger models tend to reduce the generalization gap between in-distribution and out-of-distribution performance.

Contribution

It provides empirical evidence on the impact of model scaling on compositional generalization in in-context learning across multiple model families and datasets.

Findings

Scaling models reduces the generalization gap

Larger models perform better on OOD semantic parsing tasks

The trend is consistent across different model families and datasets

Abstract

Pretrained large generative language models have shown great performance on many tasks, but exhibit low compositional generalization abilities. Scaling such models has been shown to improve their performance on various NLP tasks even just by conditioning them on a few examples to solve the task without any fine-tuning (also known as in-context learning). In this work, we look at the gap between the in-distribution (ID) and out-of-distribution (OOD) performance of such models in semantic parsing tasks with in-context learning. In the ID settings, the demonstrations are from the same split (test or train) that the model is being evaluated on, and in the OOD settings, they are from the other split. We look at how the relative generalization gap of in-context learning evolves as models are scaled up. We evaluate four model families, OPT, BLOOM, CodeGen and Codex on three semantic parsing datasets, CFQ, SCAN and GeoQuery with different number of exemplars, and observe a trend of decreasing relative generalization gap as models are scaled up.