ACL Findings2023
AVATAR: A Parallel Corpus for Java-Python Program Translation
Wasi Uddin Ahmad, Md Golam Rahman Tushar, Saikat Chakraborty, Kai-Wei Chang
Abstract
Program translation refers to migrating source code from one programming language to another. It has tremendous practical value in software development, as porting software across languages is time-consuming and costly. Automating program translation is of paramount importance in software migration, and recently researchers explored unsupervised approaches due to the unavailability of parallel corpora. However, the availability of pre-trained language models for programming languages enables supervised fine-tuning with a small number of labeled examples. Therefore, we present AVATAR, a collection of 9,515 programming problems and their solutions written in two popular languages, Java and Python. AVATAR is collected from competitive programming sites, online platforms, and open-source repositories. Furthermore, AVATAR includes unit tests for 250 examples to facilitate functional correctness evaluation. We benchmark several pretrained language models fine-tuned on AVATAR. Experiment results show that the models lack in generating functionally accurate code.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
23 top-tier papers have cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- Lost in Translation: A Study of Bugs Introduced by Large Language Models while Translating Code cites CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis
- Lost in Translation: A Study of Bugs Introduced by Large Language Models while Translating Code cites CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
- Lost in Translation: A Study of Bugs Introduced by Large Language Models while Translating Code cites CodeBERT: A Pre-Trained Model for Programming and Natural Languages
- Lost in Translation: A Study of Bugs Introduced by Large Language Models while Translating Code cites DOBF: A Deobfuscation Pre-Training Objective for Programming Languages
- CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis cites CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
- CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis cites CodeBERT: A Pre-Trained Model for Programming and Natural Languages
- CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation cites CodeBERT: A Pre-Trained Model for Programming and Natural Languages
- CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation cites DOBF: A Deobfuscation Pre-Training Objective for Programming Languages
- Divide-and-Conquer Meets Consensus: Unleashing the Power of Functions in Code Generation cites CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis
- Divide-and-Conquer Meets Consensus: Unleashing the Power of Functions in Code Generation cites CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
- Divide-and-Conquer Meets Consensus: Unleashing the Power of Functions in Code Generation cites CodeBERT: A Pre-Trained Model for Programming and Natural Languages
- NatGen: generative pre-training by "naturalizing" source code cites CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
- NatGen: generative pre-training by "naturalizing" source code cites CodeBERT: A Pre-Trained Model for Programming and Natural Languages
- NatGen: generative pre-training by "naturalizing" source code cites DOBF: A Deobfuscation Pre-Training Objective for Programming Languages
- NatGen: generative pre-training by "naturalizing" source code cites GraphCodeBERT: Pre-training Code Representations with Data Flow
- SelfCodeAlign: Self-Alignment for Code Generation cites CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis
- SelfCodeAlign: Self-Alignment for Code Generation cites CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
- EffiLearner: Enhancing Efficiency of Generated Code via Self-Optimization cites CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis
- EffiLearner: Enhancing Efficiency of Generated Code via Self-Optimization cites CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
Cited by top-tier papers23
- NatGen: generative pre-training by "naturalizing" source codeSaikat Chakraborty, Toufique Ahmed, Yangruibo Ding, Premkumar T. Devanbu et al.FSE 2022 · 101 citations
- Lost in Translation: A Study of Bugs Introduced by Large Language Models while Translating CodeRangeet Pan, Ali Reza Ibrahimzada, Rahul Krishna, Divya Sankar et al.ICSE 2024 · 96 citations
- SelfCodeAlign: Self-Alignment for Code GenerationYuxiang Wei, Federico Cassano, Jiawei Liu, Yifeng Ding et al.NeurIPS 2024 · 79 citations
- EffiLearner: Enhancing Efficiency of Generated Code via Self-OptimizationDong Huang, Jianbo Dai, Han Weng, Puzhen Wu et al.NeurIPS 2024 · 54 citations
- Divide-and-Conquer Meets Consensus: Unleashing the Power of Functions in Code GenerationJingchang Chen, Hongxuan Tang, Zheng Chu, Qianglong Chen et al.NeurIPS 2024 · 29 citations
Builds on7
- CodeBERT: A Pre-Trained Model for Programming and Natural LanguagesZhangyin Feng, Daya Guo, Duyu Tang, Nan Duan et al.EMNLP Findings 2020 · 2,828 citations
- GraphCodeBERT: Pre-training Code Representations with Data FlowDaya Guo, Shuo Ren, Shuai Lu, Zhangyin Feng et al.ICLR 2021 · 1,644 citations
- CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and GenerationYue Wang, Weishi Wang, Shafiq R. Joty, Steven C. H. HoiEMNLP 2021 · 1,224 citations
- CodeGen: An Open Large Language Model for Code with Multi-Turn Program SynthesisErik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu et al.ICLR 2023 · 234 citations
- DOBF: A Deobfuscation Pre-Training Objective for Programming LanguagesMarie-Anne Lachaux, Baptiste Rozière, Marc Szafraniec, Guillaume LampleNeurIPS 2021 · 174 citations
Related papers
- Syntax and Domain Aware Model for Unsupervised Program TranslationFang Liu, Jia Li, Li ZhangICSE 2023 · 25 citations
- Multilingual Code Co-evolution using Large Language ModelsJiyang Zhang, Pengyu Nie, Junyi Jessy Li, Milos GligoricFSE 2023 · 34 citations
- Exploring and Unleashing the Power of Large Language Models in Automated Code TranslationZhen Yang, Fang Liu, Zhongxing Yu, Jacky Wai Keung et al.FSE 2024 · 72 citations
- Program Translation via Code DistillationYufan Huang, Mengnan Qi, Yongqiang Yao, Maoquan Wang et al.EMNLP 2023 · 7 citations
- Leveraging Automated Unit Tests for Unsupervised Code TranslationBaptiste Rozière, Jie Zhang, François Charton, Mark Harman et al.ICLR 2022 · 161 citations
