In-context Learning in Presence of Spurious Correlations

TL;DR

This paper investigates training in-context learners for classification tasks with spurious features, proposing a novel method that improves performance and generalization compared to existing approaches.

Contribution

It introduces a new training technique for in-context learners that enhances their ability to handle spurious correlations and generalize across tasks.

Findings

The conventional approach is susceptible to spurious features.

Training on a single task leads to task memorization.

Diverse synthetic training data enables generalization to unseen tasks.

Abstract

Large language models exhibit a remarkable capacity for in-context learning, where they learn to solve tasks given a few examples. Recent work has shown that transformers can be trained to perform simple regression tasks in-context. This work explores the possibility of training an in-context learner for classification tasks involving spurious features. We find that the conventional approach of training in-context learners is susceptible to spurious features. Moreover, when the meta-training dataset includes instances of only one task, the conventional approach leads to task memorization and fails to produce a model that leverages context for predictions. Based on these observations, we propose a novel technique to train such a learner for a given classification task. Remarkably, this in-context learner matches and sometimes outperforms strong methods like ERM and GroupDRO. However, unlike these algorithms, it does not generalize well to other tasks. We show that it is possible to obtain an in-context learner that generalizes to unseen tasks by training on a diverse dataset of synthetic in-context learning instances.