Exploiting Cloze Questions for Few Shot Text Classification and Natural Language Inference

Timo Schick, Hinrich Schütze
1/21/2020
cs.CL

Official code: timoschick/pet (1,622 stars, Apache-2.0, last commit 2023-06-12).

Abstract

Some NLP tasks can be solved in a fully unsupervised fashion by providing a pretrained language model with "task descriptions" in natural language (e.g., Radford et al., 2019). While this approach underperforms its supervised counterpart, we show in this work that the two ideas can be combined: We introduce Pattern-Exploiting Training (PET), a semi-supervised training procedure that reformulates input examples as cloze-style phrases to help language models understand a given task. These phrases are then used to assign soft labels to a large set of unlabeled examples. Finally, standard supervised training is performed on the resulting training set. For several tasks and languages, PET outperforms supervised training and strong semi-supervised approaches in low-resource settings by a large margin.

View on arXivView PDF

Code Implementations(1)

timoschick/petOfficial100%
Apache-2.0

This repository contains the code for "Exploiting Cloze Questions for Few-Shot Text Classification and Natural Language Inference"

1,622281PythonApr 7, 20203 years agoApache-2.0
machine-learningnlppython

Similar Papers

MVP: Multi-task Supervised Pre-training for Natural Language Generation

87% match2 repos162.1k2022

Scalable Prompt Generation for Semi-supervised Learning with Language Models

85% match2023

Auto-Instruct: Automatic Instruction Generation and Ranking for Black-Box Language Models

85% match1 repo232023

Did You Read the Instructions? Rethinking the Effectiveness of Task Definitions in Instruction Learning

84% match1 repo302023

Don't Stop Pretraining? Make Prompt-based Fine-tuning Powerful Learner

84% match1 repo762023

Don't Stop Pretraining? Make Prompt-based Fine-tuning Powerful Learner

84% match2 repos1022023

Cite this paper

@article{schick2020exploiting,
  title  = {Exploiting Cloze Questions for Few Shot Text Classification and Natural Language Inference},
  author = {Timo Schick and Hinrich Schütze},
  year   = {2020},
  eprint = {2001.07676},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/2001.07676}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2001.07676.svg)](https://codeofpaper.com/paper/2001.07676)

Discussion