https://github.com/timoschick/fewglue

This repository contains the FewGLUE dataset for few-shot natural language understanding.
https://github.com/timoschick/fewglue

Last synced: 4 months ago
JSON representation

This repository contains the FewGLUE dataset for few-shot natural language understanding.

• Host: GitHub
• URL: https://github.com/timoschick/fewglue
• Owner: timoschick
• Created: 2020-09-11T08:16:45.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2020-09-16T06:51:05.000Z (almost 6 years ago)
• Last Synced: 2025-08-24T14:32:44.058Z (10 months ago)
• Size: 15.1 MB
• Stars: 160
• Watchers: 8
• Forks: 25
• Open Issues: 2
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# FewGLUE

This repository contains the FewGLUE dataset, consisting of a random selection of 32 training examples from the SuperGLUE training sets and up to 20,000 unlabeled examples for each SuperGLUE task.

### 🗂️ Structure

For each task `t` in SuperGLUE, the directory `FewGLUE/t` contains two files: `train.jsonl`, which contains the 32 training examples, and `unlabeled.jsonl`, which contains all unlabeled examples.

The official development and test sets are not included as they can be found [here](https://super.gluebenchmark.com/tasks).

### 📑 Format

All files follow the exact same format as the [original SuperGLUE training files](https://super.gluebenchmark.com/tasks).

### 📕 Citation

If you make use of FewGLUE, please cite the following paper:

    @article{schick2020small,

      title={It's Not Just Size That Matters: Small Language Models Are Also Few-Shot Learners},

      author={Timo Schick and Hinrich Schütze},

      journal={Computing Research Repository},

      volume={arXiv:2009.07118},

      url={http://arxiv.org/abs/2009.07118},

      year={2020}

    }