Exploring Data Augmentation for Code Generation Tasks

Pinzhen Chen; Gerasimos Lampouras

doi:10.18653/v1/2023.findings-eacl.114

Exploring Data Augmentation for Code Generation Tasks

Pinzhen Chen, Gerasimos Lampouras

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Advances in natural language processing, such as transfer learning from pre-trained language models, have impacted how models are trained for programming language tasks too. Previous research primarily explored code pre-training and expanded it through multi-modality and multi-tasking, yet the data for downstream tasks remain modest in size. Focusing on data utilization for downstream tasks, we propose and adapt augmentation methods that yield consistent improvements in code translation and summarization by up to 6.9% and 7.5% respectively. Further analysis suggests that our methods work orthogonally and show benefits in output code style and numeric consistency. We also discuss test data imperfections.

Original language	English
Title of host publication	Findings of the Association for Computational Linguistics: EACL 2023
Editors	Andreas Vlachos, Isabelle Augenstein
Place of Publication	Dubrovnik, Croatia
Publisher	Association for Computational Linguistics
Pages	1542-1550
Number of pages	9
ISBN (Electronic)	978-1-959429-47-0
DOIs	https://doi.org/10.18653/v1/2023.findings-eacl.114
Publication status	Published - 1 May 2023
Event	The 17th Conference of the European Chapter of the Association for Computational Linguistics - Valamar Lacroma, Dubrovnik, Croatia Duration: 2 May 2023 → 6 May 2023 Conference number: 17 https://2023.eacl.org/

Conference

Conference	The 17th Conference of the European Chapter of the Association for Computational Linguistics
Abbreviated title	EACL 2023
Country/Territory	Croatia
City	Dubrovnik
Period	2/05/23 → 6/05/23
Internet address	https://2023.eacl.org/

Access to Document

10.18653/v1/2023.findings-eacl.114Licence: Creative Commons: Attribution (CC-BY)

Exploring data_CHEN_DOA20012023_VOR_CC BYFinal published version, 247 KBLicence: Creative Commons: Attribution (CC-BY)

https://aclanthology.org/2023.findings-eacl.114Licence: Creative Commons: Attribution (CC-BY)

Cite this

@inproceedings{80645f0f0f8e42069a8910fc32089f0a,

title = "Exploring Data Augmentation for Code Generation Tasks",

abstract = "Advances in natural language processing, such as transfer learning from pre-trained language models, have impacted how models are trained for programming language tasks too. Previous research primarily explored code pre-training and expanded it through multi-modality and multi-tasking, yet the data for downstream tasks remain modest in size. Focusing on data utilization for downstream tasks, we propose and adapt augmentation methods that yield consistent improvements in code translation and summarization by up to 6.9% and 7.5% respectively. Further analysis suggests that our methods work orthogonally and show benefits in output code style and numeric consistency. We also discuss test data imperfections.",

author = "Pinzhen Chen and Gerasimos Lampouras",

year = "2023",

month = may,

day = "1",

doi = "10.18653/v1/2023.findings-eacl.114",

language = "English",

pages = "1542--1550",

editor = "Andreas Vlachos and Isabelle Augenstein",

booktitle = "Findings of the Association for Computational Linguistics: EACL 2023",

publisher = "Association for Computational Linguistics",

note = "The 17th Conference of the European Chapter of the Association for Computational Linguistics, EACL 2023 ; Conference date: 02-05-2023 Through 06-05-2023",

url = "https://2023.eacl.org/",

}

Chen, P & Lampouras, G 2023, Exploring Data Augmentation for Code Generation Tasks. in A Vlachos & I Augenstein (eds), Findings of the Association for Computational Linguistics: EACL 2023. Association for Computational Linguistics, Dubrovnik, Croatia, pp. 1542-1550, The 17th Conference of the European Chapter of the Association for Computational Linguistics, Dubrovnik, Croatia, 2/05/23. https://doi.org/10.18653/v1/2023.findings-eacl.114

Exploring Data Augmentation for Code Generation Tasks. / Chen, Pinzhen; Lampouras, Gerasimos.
Findings of the Association for Computational Linguistics: EACL 2023. ed. / Andreas Vlachos; Isabelle Augenstein. Dubrovnik, Croatia: Association for Computational Linguistics, 2023. p. 1542-1550.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Exploring Data Augmentation for Code Generation Tasks

AU - Chen, Pinzhen

AU - Lampouras, Gerasimos

N1 - Conference code: 17

PY - 2023/5/1

Y1 - 2023/5/1

N2 - Advances in natural language processing, such as transfer learning from pre-trained language models, have impacted how models are trained for programming language tasks too. Previous research primarily explored code pre-training and expanded it through multi-modality and multi-tasking, yet the data for downstream tasks remain modest in size. Focusing on data utilization for downstream tasks, we propose and adapt augmentation methods that yield consistent improvements in code translation and summarization by up to 6.9% and 7.5% respectively. Further analysis suggests that our methods work orthogonally and show benefits in output code style and numeric consistency. We also discuss test data imperfections.

AB - Advances in natural language processing, such as transfer learning from pre-trained language models, have impacted how models are trained for programming language tasks too. Previous research primarily explored code pre-training and expanded it through multi-modality and multi-tasking, yet the data for downstream tasks remain modest in size. Focusing on data utilization for downstream tasks, we propose and adapt augmentation methods that yield consistent improvements in code translation and summarization by up to 6.9% and 7.5% respectively. Further analysis suggests that our methods work orthogonally and show benefits in output code style and numeric consistency. We also discuss test data imperfections.

U2 - 10.18653/v1/2023.findings-eacl.114

DO - 10.18653/v1/2023.findings-eacl.114

M3 - Conference contribution

SP - 1542

EP - 1550

BT - Findings of the Association for Computational Linguistics: EACL 2023

A2 - Vlachos, Andreas

A2 - Augenstein, Isabelle

PB - Association for Computational Linguistics

CY - Dubrovnik, Croatia

T2 - The 17th Conference of the European Chapter of the Association for Computational Linguistics

Y2 - 2 May 2023 through 6 May 2023

ER -

Exploring Data Augmentation for Code Generation Tasks

Abstract

Conference

Access to Document

Fingerprint

Cite this