Generating Datasets with Pretrained Language Models

TL;DR

This paper presents an unsupervised method that uses pretrained language models to generate labeled datasets for fine-tuning smaller models, achieving high-quality sentence embeddings without labeled data or additional training objectives.

Contribution

It introduces a novel approach to generate labeled datasets from scratch using PLMs, eliminating the need for human-labeled data or modifications to pretraining.

Findings

Outperforms strong baselines on semantic textual similarity datasets

Enables high-quality sentence embeddings without labeled data

Uses generative abilities of PLMs to create training datasets

Abstract

To obtain high-quality sentence embeddings from pretrained language models (PLMs), they must either be augmented with additional pretraining objectives or finetuned on a large set of labeled text pairs. While the latter approach typically outperforms the former, it requires great human effort to generate suitable datasets of sufficient size. In this paper, we show how PLMs can be leveraged to obtain high-quality sentence embeddings without the need for labeled data, finetuning or modifications to the pretraining objective: We utilize the generative abilities of large and high-performing PLMs to generate entire datasets of labeled text pairs from scratch, which we then use for finetuning much smaller and more efficient models. Our fully unsupervised approach outperforms strong baselines on several semantic textual similarity datasets.