Summary of Codes: Natural Language to Code Repository Via Multi-layer Sketch, by Daoguang Zan and Ailun Yu and Wei Liu and Dong Chen and Bo Shen and Wei Li and Yafen Yao and Yongshun Gong and Xiaolin Chen and Bei Guan and Zhiguang Yang and Yongji Wang and Qianxiang Wang and Lizhen Cui
CodeS: Natural Language to Code Repository via Multi-Layer Sketch
by Daoguang Zan, Ailun Yu, Wei Liu, Dong Chen, Bo Shen, Wei Li, Yafen Yao, Yongshun Gong, Xiaolin Chen, Bei Guan, Zhiguang Yang, Yongji Wang, Qianxiang Wang, Lizhen Cui
First submitted to arxiv on: 25 Mar 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The impressive performance of large language models (LLMs) on code-related tasks has shown the potential of fully automated software development. This paper introduces a new software engineering task, Natural Language to code Repository (NL2Repo), which aims to generate an entire code repository from its natural language requirements. The proposed framework CodeS decomposes NL2Repo into multiple sub-tasks by a multi-layer sketch, comprising RepoSketcher, FileSketcher, and SketchFiller modules. To assess CodeS’s effectiveness on the NL2Repo task, evaluations are conducted through both automated benchmarking using the SketchEval repository-oriented benchmark and evaluation metric SketchBLEU, as well as manual feedback analysis via a VSCode plugin engaged with 30 participants in empirical studies. Extensive experiments prove the practicality and effectiveness of CodeS. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Automated software development is getting more powerful thanks to big language models! This paper introduces a new task called NL2Repo that tries to generate an entire code repository from just some natural language requirements. The authors propose a framework called CodeS that breaks this task down into smaller steps. They test their approach using two methods: one that looks at how well the generated code works, and another that asks people for feedback. The results show that CodeS is quite good at generating code! |
