Generative Code Modeling with Graphs
Abstract
Generative models forsource code are an interesting structured prediction problem, requiring to reason about both hard syntactic and semantic constraints as well as about natural, likely programs. We present a novel model for this problem that uses a graph to represent the intermediate state of the generated output. Our model generates code by interleaving grammar-driven expansion steps with graph augmentation and neural message passing steps. An experimental evaluation shows that our new model can generate semantically meaningful expressions, outperforming a range of strong baselines.
Cite
Text
Brockschmidt et al. "Generative Code Modeling with Graphs." International Conference on Learning Representations, 2019.Markdown
[Brockschmidt et al. "Generative Code Modeling with Graphs." International Conference on Learning Representations, 2019.](https://mlanthology.org/iclr/2019/brockschmidt2019iclr-generative/)BibTeX
@inproceedings{brockschmidt2019iclr-generative,
title = {{Generative Code Modeling with Graphs}},
author = {Brockschmidt, Marc and Allamanis, Miltiadis and Gaunt, Alexander L. and Polozov, Oleksandr},
booktitle = {International Conference on Learning Representations},
year = {2019},
url = {https://mlanthology.org/iclr/2019/brockschmidt2019iclr-generative/}
}