@inproceedings{gandhi-etal-2024-better,
title = "Better Synthetic Data by Retrieving and Transforming Existing Datasets",
author = "Gandhi, Saumya and
Gala, Ritu and
Viswanathan, Vijay and
Wu, Tongshuang and
Neubig, Graham",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.385",
doi = "10.18653/v1/2024.findings-acl.385",
pages = "6453--6466",
abstract = "Despite recent advances in large language models, building dependable and deployable NLP models typically requires abundant, high-quality training data. However, task-specific data is not available for many use cases, and manually curating task-specific data is labor-intensive. Recent work has studied prompt-driven synthetic data generation using large language models, but these generated datasets tend to lack complexity and diversity. To address these limitations, we introduce a method, {\_}DataTune{\_}, to make better use of existing, publicly available datasets to improve automatic dataset generation. DataTune performs {\_}dataset transformation{\_}, enabling the repurposing of publicly available datasets into a format that is directly aligned with the specific requirements of target tasks. On a diverse set of language-based tasks from the BIG-Bench benchmark, we find that finetuning language models via DataTune improves over a few-shot prompting baseline by 49{\%} and improves over existing methods that use synthetic or retrieved training data by 34{\%}. We find that dataset transformation significantly increases the diversity and difficulty of generated data on many tasks. We release a Python package and open-source repository to make this method accessible to the community (URL will be added upon acceptance).",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="gandhi-etal-2024-better">
<titleInfo>
<title>Better Synthetic Data by Retrieving and Transforming Existing Datasets</title>
</titleInfo>
<name type="personal">
<namePart type="given">Saumya</namePart>
<namePart type="family">Gandhi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ritu</namePart>
<namePart type="family">Gala</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vijay</namePart>
<namePart type="family">Viswanathan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tongshuang</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Graham</namePart>
<namePart type="family">Neubig</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-08</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2024</title>
</titleInfo>
<name type="personal">
<namePart type="given">Lun-Wei</namePart>
<namePart type="family">Ku</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Andre</namePart>
<namePart type="family">Martins</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vivek</namePart>
<namePart type="family">Srikumar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Bangkok, Thailand</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Despite recent advances in large language models, building dependable and deployable NLP models typically requires abundant, high-quality training data. However, task-specific data is not available for many use cases, and manually curating task-specific data is labor-intensive. Recent work has studied prompt-driven synthetic data generation using large language models, but these generated datasets tend to lack complexity and diversity. To address these limitations, we introduce a method, _DataTune_, to make better use of existing, publicly available datasets to improve automatic dataset generation. DataTune performs _dataset transformation_, enabling the repurposing of publicly available datasets into a format that is directly aligned with the specific requirements of target tasks. On a diverse set of language-based tasks from the BIG-Bench benchmark, we find that finetuning language models via DataTune improves over a few-shot prompting baseline by 49% and improves over existing methods that use synthetic or retrieved training data by 34%. We find that dataset transformation significantly increases the diversity and difficulty of generated data on many tasks. We release a Python package and open-source repository to make this method accessible to the community (URL will be added upon acceptance).</abstract>
<identifier type="citekey">gandhi-etal-2024-better</identifier>
<identifier type="doi">10.18653/v1/2024.findings-acl.385</identifier>
<location>
<url>https://aclanthology.org/2024.findings-acl.385</url>
</location>
<part>
<date>2024-08</date>
<extent unit="page">
<start>6453</start>
<end>6466</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Better Synthetic Data by Retrieving and Transforming Existing Datasets
%A Gandhi, Saumya
%A Gala, Ritu
%A Viswanathan, Vijay
%A Wu, Tongshuang
%A Neubig, Graham
%Y Ku, Lun-Wei
%Y Martins, Andre
%Y Srikumar, Vivek
%S Findings of the Association for Computational Linguistics: ACL 2024
%D 2024
%8 August
%I Association for Computational Linguistics
%C Bangkok, Thailand
%F gandhi-etal-2024-better
%X Despite recent advances in large language models, building dependable and deployable NLP models typically requires abundant, high-quality training data. However, task-specific data is not available for many use cases, and manually curating task-specific data is labor-intensive. Recent work has studied prompt-driven synthetic data generation using large language models, but these generated datasets tend to lack complexity and diversity. To address these limitations, we introduce a method, _DataTune_, to make better use of existing, publicly available datasets to improve automatic dataset generation. DataTune performs _dataset transformation_, enabling the repurposing of publicly available datasets into a format that is directly aligned with the specific requirements of target tasks. On a diverse set of language-based tasks from the BIG-Bench benchmark, we find that finetuning language models via DataTune improves over a few-shot prompting baseline by 49% and improves over existing methods that use synthetic or retrieved training data by 34%. We find that dataset transformation significantly increases the diversity and difficulty of generated data on many tasks. We release a Python package and open-source repository to make this method accessible to the community (URL will be added upon acceptance).
%R 10.18653/v1/2024.findings-acl.385
%U https://aclanthology.org/2024.findings-acl.385
%U https://doi.org/10.18653/v1/2024.findings-acl.385
%P 6453-6466
Markdown (Informal)
[Better Synthetic Data by Retrieving and Transforming Existing Datasets](https://aclanthology.org/2024.findings-acl.385) (Gandhi et al., Findings 2024)
ACL