LMTurk: Few-Shot Learners as Crowdsourcing Workers in a Language-Model-as-a-Service Framework

TL;DR

LMTurk introduces a novel framework that leverages large-scale pretrained language models as crowdsourcing workers for annotation tasks, integrating active learning to reduce computational costs and improve practical deployability.

Contribution

The paper proposes LMTurk, a new approach that treats few-shot learners as crowdsourcing workers, enabling efficient annotation and training of smaller, deployable models.

Findings

Annotations from LMTurk improve downstream task performance.

Active learning reduces the number of PLM queries needed.

LMTurk makes effective use of large PLMs with lower computational costs.

Abstract

Vast efforts have been devoted to creating high-performance few-shot learners, i.e., large-scale pretrained language models (PLMs) that perform well with little downstream task training data. Training PLMs has incurred significant cost, but utilizing the few-shot learners is still challenging due to their enormous size. This work focuses on a crucial question: How to make effective use of these few-shot learners? We propose LMTurk, a novel approach that treats few-shot learners as crowdsourcing workers. The rationale is that crowdsourcing workers are in fact few-shot learners: They are shown a few illustrative examples to learn about a task and then start annotating. LMTurk employs few-shot learners built upon PLMs as workers. We show that the resulting annotations can be utilized to train models that solve the task well and are small enough to be deployable in practical scenarios. Active learning is integrated into LMTurk to reduce the amount of queries made to PLMs, minimizing the computational cost of running PLM inference passes. Altogether, LMTurk is an important step towards making effective use of current PLMs.