This implementation of TEA is based on Pytorch.
The experiment uses the Yelp and Epinions data sets. Data preprocessing is required before training. Here uses yelp as an example
python my_preprocess_yelp.py
TEA can be trained afterwards
python run_tea_metapath_yelp.py
The training method of the baseline in the paper is similar.
