True Few-Shot Learning with Language Models

Abstract

Pretrained language models (LMs) perform well on many tasks even when learning from a few examples, but prior work uses many held-out examples to tune various aspects of learning, such as hyperparameters, training objectives, and natural language templates ("prompts"). Here, we evaluate the few-shot ability of LMs when such held-out examples are unavailable, a setting we call true few-shot learning. We test two model selection criteria, cross-validation and minimum description length, for choosing LM prompts and hyperparameters in the true few-shot setting. On average, both marginally outperform random selection and greatly underperform selection based on held-out examples. Moreover, selection criteria often prefer models that perform significantly worse than randomly-selected ones. We find similar results even when taking into account our uncertainty in a model's true performance during selection, as well as when varying the amount of computation and number of examples used for selection. Overall, our findings suggest that prior work significantly overestimated the true few-shot ability of LMs given the difficulty of few-shot model selection.

Code at https://github.com/ethanjperez/true_few_shot

References in corpus (6)

• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Learning Transferable Visual Models From Natural Language Supervision
• Language Models are Few-Shot Learners
• Asymptotic Equivalence of Bayes Cross Validation and Widely Applicable Information Criterion in Singular Learning Theory
• Fine-Tuning Pretrained Language Models: Weight Initializations, Data Orders, and Early Stopping
• Advancing the State of the Art in Open Domain Dialog Systems through the Alexa Prize

Cited by in corpus (2)

• Recent Advances in Natural Language Processing via Large Pre-Trained Language Models: A Survey
• HTLM: Hyper-Text Pre-Training and Prompting of Language Models