Awesome

We provide the code (in pytorch) and datasets for our paper "HGPROMPT: Bridging Homogeneous and Heterogeneous Graphs for Few-shot Prompt Learning", which is accepted by AAAI2024.

Package Dependencies

• cuda 11.3
• dgl0.9.0-cu113
• dgllife

Running experiments

The default dataset is ACM. You need to change the corresponding parameters in pre_train.py and run.py to train and evaluate on other datasets.

Pretrain:

• python pre_train.py

Prompting and test:

• python run.py