Graph Playground (v-0.1)

Graph Playground is a highy emsamble project for Graph Neural Network (GNN), using pytorch framwork and other supporting packages (DGL, Torch Geometric, Ray, etc). The goal of this project is to build a framework that is easy to get started with, highly integrated, and unified.

Currently in the process of updating...Not enough tasks, datasets and methods are supported, pls be patient.

Main Install:

• python 3.9
• pytorch 1.10.0_cuda11.3
• torch_geometric
• dgl
• ray

Full installed packages, pls see requirements (not all of them are needed)

Quickly Use:

Set your choose in main.py Line. 148:

Task (e.g., task = 'Semi' # choice: Semi Unsup Sup Rein Noise Fed)

Method (e.g., method = 'Gcn' # choice: Gcn Gat Sage DGI...)

Dataset (e.g., dataset = 'Cora' # choice: Cora CiteSeer PubMed freebase Photo Computers...)

And Run:

python main.py

Task:

updating

Method:

updating

Dataset:

updating

Params and Auto-turning:

updating

Log your results to Database:

updating

Project structure:

updating