Large Language Models Can be Lazy Learners: Analyze Shortcuts in In-Context Learning
arXiv:2305.17256 · doi:10.18653/v1/2023.findings-acl.284
Abstract
Large language models (LLMs) have recently shown great potential for in-context learning, where LLMs learn a new task simply by conditioning on a few input-label pairs (prompts). Despite their potential, our understanding of the factors influencing end-task performance and the robustness of in-context learning remains limited. This paper aims to bridge this knowledge gap by investigating the reliance of LLMs on shortcuts or spurious correlations within prompts. Through comprehensive experiments on classification and extraction tasks, we reveal that LLMs are "lazy learners" that tend to exploit shortcuts in prompts for downstream tasks. Additionally, we uncover a surprising finding that larger models are more likely to utilize shortcuts in prompts during inference. Our findings provide a new perspective on evaluating robustness in in-context learning and pose new challenges for detecting and mitigating the use of shortcuts in prompts.
References in corpus (9)
- Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
- Multitask Prompted Training Enables Zero-Shot Task Generalization
- The Pile: An 800GB Dataset of Diverse Text for Language Modeling
- Harnessing the Power of LLMs in Practice: A Survey on ChatGPT and Beyond
- Does Synthetic Data Generation of LLMs Help Clinical Text Mining?
- MetaICL: Learning to Learn In Context
- Shortcut Learning of Large Language Models in Natural Language Understanding
- Meta-learning via Language Model In-context Tuning
- Identifying and Mitigating Spurious Correlations for Improving Robustness in NLP Models