SCIroShot is an entailment-based zero-shot text classifier that has been trained on a weakly supervised dataset of scientific data originally gathered from Microsoft Academic Graph.
For more details, refer to the paper "A weakly supervised textual entailment approach to zero-shot text classification", published in the EACL 2023 conference.
from transformers import pipeline
zstc = pipeline("zero-shot-classification", model="BSC-LT/sciroshot")
sentence = "Leo Messi is the best player ever."
candidate_labels = ["politics", "science", "sports", "environment"]
template = "This example is {}"
output = zstc(sentence, candidate_labels, hypothesis_template=template, multi_label=False)
print(output)
print(f'Predicted class: {output["labels"][0]}')| Model | arXiv | SciDocs-MesH | SciDocs-MAG | Konstanz | Elsevier | PubMed |
|---|---|---|---|---|---|---|
| fb/bart-large-mnli | 33.28 | 66.18🔥 | 51.77 | 54.62 | 28.41 | 31.59🔥 |
| SCIroShot | 42.22🔥 | 59.34 | 69.86🔥 | 66.07🔥 | 54.42🔥 | 27.93 |
| Model | Topic | Emotion | Situation |
|---|---|---|---|
| RTE (Yin et al., 2019) | 43.8 | 12.6 | 37.2🔥 |
| FEVER (Yin et al., 2019) | 40.1 | 24.7 | 21.0 |
| MNLI (Yin et al., 2019) | 37.9 | 22.3 | 15.4 |
| NSP (Ma et al., 2021) | 50.6 | 16.5 | 25.8 |
| NSP-Reverse (Ma et al., 2021) | 53.1 | 16.1 | 19.9 |
| SCIroShot | 59.08🔥 | 24.94🔥 | 27.42 |
@inproceedings{pamies2023weakly,
title={A weakly supervised textual entailment approach to zero-shot text classification},
author={Pàmies, Marc and Llop, Joan and Multari, Francesco and Duran-Silva, Nicolau and Parra-Rojas, César and González-Agirre, Aitor and Massucci, Francesco Alessandro and Villegas, Marta},
booktitle={Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics},
pages={286--296},
year={2023}
}
This work is distributed under a Apache License, Version 2.0.
