This repository is for learning cs224w. Notebooks are all written in julia. The notebookx.jl are notebooks corresponding to Colab x in the schedule of cs224w. This repository only track the codes, while not for the handwriting homeworks, especially the final assignment and project.
CS224w had proposed 5 notebooks, which provide basic codes of graph structure in python, with the library of networkx, deepSNAP, and well-known pytorch, pytorch-geometric. In julia, the famous deep learning framework Flux.jl and graph package Graphs.jl are used (MetaGraphs.jl for Heterogeneous graphs). Based on them, GeometricFlux.jl is a deep learning package working on graphs, and provide several implementations.
These packages helps learning ML on graphs, and also the DL framework in Flux.jl. In cs224w, we could learn a great amount of details about Graphs and GNNs. Yet, in the notebooks, we only contact with some of those content, as a good introduction.
Notebooks were not tested on CUDA, for that the datasets we are playing with are not too huge.
The notebooks are finished with Pluto.jl, a notebook engine in julia. To read these notebooks, run in julia:
julia> using Pluto
julia> Pluto.run()and load these notebooks.