Co-training Improves Prompt-based Learning for Large Language Models
This addresses the problem of making prompt-based learning more robust and efficient for researchers and practitioners in NLP, though it is incremental as it builds on existing co-training and prompting methods.
The paper tackles the brittleness and inefficiency of prompt-based learning in large language models by applying co-training with unlabeled data, resulting in significant performance improvements on challenging datasets where a gap existed between prompt-based and fully-supervised models.
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.