[project]: The list of relating code

[Aisuko]

Currently, let's fix the issues of the repo below

These three repos that we want to use as benchmark repo. We need to repeat their result.

Only server side dataset

1. First benchmark project

• https://github.com/MinChen00/Graph-Unlearning @Aisuko
  • https://github.com/orgs/SkywardAI/projects/3?pane=issue&itemId=79490377
  • https://www.kaggle.com/code/aisuko/demo-for-running-out-of-date-graph-unlearning-code/notebook

2. Second benchmark project

• https://github.com/mims-harvard/GNNDelete @Aisuko
  • https://www.kaggle.com/models/skywardai/ogbl-collab
  • https://www.kaggle.com/models/skywardai/ogbl-citation2
  • https://www.kaggle.com/code/aisuko/loading-data-for-gnndelete/notebook

3. Third benchmark project
Distribute server and user

Note: Some model and data like FedLU has 4 versions due to the training process are so slow

• Less efficient code, in our env (only 8% GPU usage, even we twisted the hyperprameter)

• https://github.com/nju-websoft/FedLU/ @Aisuko
  • https://www.kaggle.com/models/skywardai/fedlu_local_and_global
  • https://www.kaggle.com/models/skywardai/fedlu