Want To Reduce Labeling Cost? GPT-3 Can Help

TL;DR

This paper demonstrates that GPT-3 can significantly reduce data labeling costs for NLP tasks by serving as a pseudo-labeler, and combining GPT-3 labels with human labels further improves performance.

Contribution

The paper introduces a cost-effective framework leveraging GPT-3 for data labeling and combines it with human labels to enhance NLP model performance.

Findings

Using GPT-3 labels reduces labeling costs by 50-96%.

Combining GPT-3 pseudo labels with human labels improves model accuracy.

The approach is applicable across various NLP tasks.

Abstract

Data annotation is a time-consuming and labor-intensive process for many NLP tasks. Although there exist various methods to produce pseudo data labels, they are often task-specific and require a decent amount of labeled data to start with. Recently, the immense language model GPT-3 with 175 billion parameters has achieved tremendous improvement across many few-shot learning tasks. In this paper, we explore ways to leverage GPT-3 as a low-cost data labeler to train other models. We find that, to make the downstream model achieve the same performance on a variety of NLU and NLG tasks, it costs 50% to 96% less to use labels from GPT-3 than using labels from humans. Furthermore, we propose a novel framework of combining pseudo labels from GPT-3 with human labels, which leads to even better performance with limited labeling budget. These results present a cost-effective data labeling methodology that is generalizable to many practical applications.