@inproceedings{puri-etal-2023-many,
title = "How Many Data Samples is an Additional Instruction Worth?",
author = "Puri, Ravsehaj Singh and
Mishra, Swaroop and
Parmar, Mihir and
Baral, Chitta",
editor = "Vlachos, Andreas and
Augenstein, Isabelle",
booktitle = "Findings of the Association for Computational Linguistics: EACL 2023",
month = may,
year = "2023",
address = "Dubrovnik, Croatia",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-eacl.77",
doi = "10.18653/v1/2023.findings-eacl.77",
pages = "1042--1057",
abstract = "Recently introduced instruction-paradigm empowers non-expert users to leverage NLP resources by defining a new task in natural language. Instruction-tuned models have significantly outperformed multitask learning models (without instruction); however they are far from state-of-the-art task-specific models. Conventional approaches to improve model performance via creating datasets with large number of task instances or architectural changes in the model may not be feasible for non-expert users. However, they can write alternate instructions to represent an instruction task. Is Instruction-augmentation helpful? We augment a subset of tasks in the expanded version of NATURAL INSTRUCTIONS with additional instructions and find that it significantly improves model performance (up to 35{\%}), especially in the low-data regime. Our results indicate that an additional instruction can be equivalent to {\textasciitilde}200 data samples on average across tasks.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="puri-etal-2023-many">
<titleInfo>
<title>How Many Data Samples is an Additional Instruction Worth?</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ravsehaj</namePart>
<namePart type="given">Singh</namePart>
<namePart type="family">Puri</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Swaroop</namePart>
<namePart type="family">Mishra</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mihir</namePart>
<namePart type="family">Parmar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chitta</namePart>
<namePart type="family">Baral</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-05</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EACL 2023</title>
</titleInfo>
<name type="personal">
<namePart type="given">Andreas</namePart>
<namePart type="family">Vlachos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Isabelle</namePart>
<namePart type="family">Augenstein</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Dubrovnik, Croatia</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Recently introduced instruction-paradigm empowers non-expert users to leverage NLP resources by defining a new task in natural language. Instruction-tuned models have significantly outperformed multitask learning models (without instruction); however they are far from state-of-the-art task-specific models. Conventional approaches to improve model performance via creating datasets with large number of task instances or architectural changes in the model may not be feasible for non-expert users. However, they can write alternate instructions to represent an instruction task. Is Instruction-augmentation helpful? We augment a subset of tasks in the expanded version of NATURAL INSTRUCTIONS with additional instructions and find that it significantly improves model performance (up to 35%), especially in the low-data regime. Our results indicate that an additional instruction can be equivalent to ~200 data samples on average across tasks.</abstract>
<identifier type="citekey">puri-etal-2023-many</identifier>
<identifier type="doi">10.18653/v1/2023.findings-eacl.77</identifier>
<location>
<url>https://aclanthology.org/2023.findings-eacl.77</url>
</location>
<part>
<date>2023-05</date>
<extent unit="page">
<start>1042</start>
<end>1057</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T How Many Data Samples is an Additional Instruction Worth?
%A Puri, Ravsehaj Singh
%A Mishra, Swaroop
%A Parmar, Mihir
%A Baral, Chitta
%Y Vlachos, Andreas
%Y Augenstein, Isabelle
%S Findings of the Association for Computational Linguistics: EACL 2023
%D 2023
%8 May
%I Association for Computational Linguistics
%C Dubrovnik, Croatia
%F puri-etal-2023-many
%X Recently introduced instruction-paradigm empowers non-expert users to leverage NLP resources by defining a new task in natural language. Instruction-tuned models have significantly outperformed multitask learning models (without instruction); however they are far from state-of-the-art task-specific models. Conventional approaches to improve model performance via creating datasets with large number of task instances or architectural changes in the model may not be feasible for non-expert users. However, they can write alternate instructions to represent an instruction task. Is Instruction-augmentation helpful? We augment a subset of tasks in the expanded version of NATURAL INSTRUCTIONS with additional instructions and find that it significantly improves model performance (up to 35%), especially in the low-data regime. Our results indicate that an additional instruction can be equivalent to ~200 data samples on average across tasks.
%R 10.18653/v1/2023.findings-eacl.77
%U https://aclanthology.org/2023.findings-eacl.77
%U https://doi.org/10.18653/v1/2023.findings-eacl.77
%P 1042-1057
Markdown (Informal)
[How Many Data Samples is an Additional Instruction Worth?](https://aclanthology.org/2023.findings-eacl.77) (Puri et al., Findings 2023)
ACL
- Ravsehaj Singh Puri, Swaroop Mishra, Mihir Parmar, and Chitta Baral. 2023. How Many Data Samples is an Additional Instruction Worth?. In Findings of the Association for Computational Linguistics: EACL 2023, pages 1042–1057, Dubrovnik, Croatia. Association for Computational Linguistics.