A plan to release OpenHGNN version 0.1

We want to release the OpenHGNN version0.1 in Sept.

The basic to implement model

• Python
  • version 3.6-3.8 (recommend 3.8 if you do not install other version)
  • Use anaconda to manage your package
  • By default you understand python syntax very well
• PyTorch
  • Version 1.7-1.9 (recommend stable(1.9.0)if you do not install other version)
  • cpu or cuda denpends on your machine
  • Docs: https://pytorch.org/docs/stable/index.html (for reference)
  • Wechat link for getting started: https://mp.weixin.qq.com/mp/homepage?__biz=MzIwMDIzNDI2Ng==&hid=10&sn=25d0262fcb130258ed1a516f298c7ee4&scene=1&devicetype=iOS14.6&version=18000731&lang=zh_CN&nettype=3G+&ascene=7&session_us=gh_030af0e61764&fontScale=100
• DGL
  • Version: Latest Nightly Build Version(ignore if you install version 0.6 or newer).
  • User Guide: https://docs.dgl.ai/guide/index.html--The chapter 1,2,3,5 are you must finish watching
  • Docs: https://docs.dgl.ai/index.html for reference
• OpenHGNN
  • Docs:https://github.com/BUPT-GAMMA/OpenHGNN
  • components: TrainerFlow, Model, Task.

The Paper List

https://docs.qq.com/sheet/DZUxxZmRaYXRuV3J5

Read the paper in one week carefully and share it in weekly group meeting.

How to implement your model

1. Read the paper
2. Process the dataset in format of dgl
3. Talk with Tianyu Zhao