Co-training Improves Prompt-based Learning for Large Language Models

Abstract

We demonstrate that co-training (Blum & Mitchell, 1998) can improve the performance of prompt-based learning by using unlabeled data. While prompting has emerged as a promising paradigm for few-shot and zero-shot learning, it is often brittle and requires much larger models compared to the standard supervised setup. We find that co-training makes it possible to improve the original prompt model and at the same time learn a smaller, downstream task-specific model. In the case where we only have partial access to a prompt model (e.g., output probabilities from GPT-3 (Brown et al., 2020)) we learn a calibration model over the prompt outputs. When we have full access to the prompt model’s gradients but full finetuning remains prohibitively expensive (e.g., T0 (Sanh et al., 2021)), we learn a set of soft prompt continuous vectors to iteratively update the prompt model. We find that models trained in this manner can significantly improve performance on challenging datasets where there is currently a large gap between prompt-based learning and fully-supervised models.

Publication
Proceedings of the Thirty-Eighth International Conference on Machine Learning (ICML)
Hunter Lang
Hunter Lang
PhD Student

Hunter’s research focuses on understanding and improving the performance of machine learning algorithms in the wild, with particular applications in MAP inference for graphical models, stochastic optimization, and weak supervision.

Monica Agrawal
Monica Agrawal
PhD Student

Assistant Professor, Duke

Yoon Kim
Yoon Kim
Master’s student

Assistant Professor, MIT

David Sontag
David Sontag
Professor of EECS

My research focuses on advancing machine learning and artificial intelligence, and using these to transform health care.

Related