Will It Blend? Mixing Training Paradigms & Prompting for Argument Quality Prediction

Michiel van der Meer; Myrthe Reuver; Urja Khurana; Lea Krause; Selene Baez Santamaria

Will It Blend? Mixing Training Paradigms & Prompting for Argument Quality Prediction

Michiel van der Meer^*, Myrthe Reuver, Urja Khurana, Lea Krause, Selene Baez Santamaria

^*Corresponding author for this work

Research output: Chapter in Book / Report / Conference proceeding › Conference contribution › Academic › peer-review

Abstract

This paper describes our contributions to the Shared Task of the 9th Workshop on Argument Mining (2022). Our approach uses Large Language Models for the task of Argument Quality Prediction. We perform prompt engineering using GPT-3, and also investigate the training paradigms multi-task learning, contrastive learning, and intermediate-task training. We find that a mixed prediction setup outperforms single models. Prompting GPT-3 works best for predicting argument validity, and argument novelty is best estimated by a model trained using all three training paradigms.

Original language	English
Title of host publication	Proceedings of the 9th Workshop on Argument Mining
Publisher	International Conference on Computational Linguistics (COLING)
Pages	95–103
Number of pages	8
Publication status	Published - Oct 2022

Access to Document

https://aclanthology.org/2022.argmining-1.8.pdf

Persistent URL (handle)

Persistent link

Cite this

@inproceedings{0191c064cd294f74b0815182f679d70f,

title = "Will It Blend? Mixing Training Paradigms & Prompting for Argument Quality Prediction",

abstract = "This paper describes our contributions to the Shared Task of the 9th Workshop on Argument Mining (2022). Our approach uses Large Language Models for the task of Argument Quality Prediction. We perform prompt engineering using GPT-3, and also investigate the training paradigms multi-task learning, contrastive learning, and intermediate-task training. We find that a mixed prediction setup outperforms single models. Prompting GPT-3 works best for predicting argument validity, and argument novelty is best estimated by a model trained using all three training paradigms.",

author = "{van der Meer}, Michiel and Myrthe Reuver and Urja Khurana and Lea Krause and {Baez Santamaria}, Selene",

year = "2022",

month = oct,

language = "English",

pages = "95–103",

booktitle = "Proceedings of the 9th Workshop on Argument Mining",

publisher = "International Conference on Computational Linguistics (COLING)",

}

Will It Blend? Mixing Training Paradigms & Prompting for Argument Quality Prediction. / van der Meer, Michiel; Reuver, Myrthe; Khurana, Urja et al.
Proceedings of the 9th Workshop on Argument Mining. International Conference on Computational Linguistics (COLING), 2022. p. 95–103.

Research output: Chapter in Book / Report / Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Will It Blend? Mixing Training Paradigms & Prompting for Argument Quality Prediction

AU - van der Meer, Michiel

AU - Reuver, Myrthe

AU - Khurana, Urja

AU - Krause, Lea

AU - Baez Santamaria, Selene

PY - 2022/10

Y1 - 2022/10

N2 - This paper describes our contributions to the Shared Task of the 9th Workshop on Argument Mining (2022). Our approach uses Large Language Models for the task of Argument Quality Prediction. We perform prompt engineering using GPT-3, and also investigate the training paradigms multi-task learning, contrastive learning, and intermediate-task training. We find that a mixed prediction setup outperforms single models. Prompting GPT-3 works best for predicting argument validity, and argument novelty is best estimated by a model trained using all three training paradigms.

AB - This paper describes our contributions to the Shared Task of the 9th Workshop on Argument Mining (2022). Our approach uses Large Language Models for the task of Argument Quality Prediction. We perform prompt engineering using GPT-3, and also investigate the training paradigms multi-task learning, contrastive learning, and intermediate-task training. We find that a mixed prediction setup outperforms single models. Prompting GPT-3 works best for predicting argument validity, and argument novelty is best estimated by a model trained using all three training paradigms.

UR - https://aclanthology.org/volumes/2022.argmining-1/

M3 - Conference contribution

SP - 95

EP - 103

BT - Proceedings of the 9th Workshop on Argument Mining

PB - International Conference on Computational Linguistics (COLING)

ER -

Will It Blend? Mixing Training Paradigms & Prompting for Argument Quality Prediction

Abstract

Access to Document

Persistent URL (handle)

Other files and links

Fingerprint

Cite this