Automatic Source Code Summarization
Project Details
Automatic summary generation of python and java source code using an encoder-decoder seq-2-seq architecture, with Graph Neural Networks to parse the AST of the source code. Incroporates a novel representation of the AST, in conjunction with a standrard text summarization model. Trained with Deep Reinforcement Learning to improve convergence.
Project Information
- Category: ML
- Stack: python, pytorch
