Our CS 224W final project Collaborators: Suhas Chundi, Patrick Donohue, Abhay Singhal
Code Segment Vectorization Model This work is an expansion of the work of Ye et. al. to develop a system that generates vector embeddings for code segments using weakly supervised labels. We trained a graph neural network on AST representation of code graphs. The objective is for output vectors to have high cosine similarity if the code segments are semantically similar. Our project was featured by the CS 224W medium page and Pytorch Geometric!