In-context Learning in Presence of Spurious Correlations
This addresses the challenge of spurious correlations in in-context learning for classification, which is incremental as it builds on prior work on regression tasks.
The paper tackles the problem of training in-context learners for classification tasks with spurious correlations, finding that conventional methods are susceptible to these features and can lead to task memorization. It proposes a novel technique that matches or outperforms strong methods like ERM and GroupDRO on specific tasks, and shows that training on diverse synthetic data can generalize to unseen tasks.
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.