create 2 new folders glove, data_embedding

Download and put glove embeddings glove.840B.300d.txt in the new folder glove. This is only used as a placeholder, we do not use the glove embeddings in this project.

Install all libraries needed using conda env create -f env.yaml

Use the default settings for KOGCN

python train.py --dataset mams --cs_dropout 0.2 --linear_dropout 0.3 --dep_dim 40 --use_const_sememe_dep pyramid2_layer_norm_drop --pyramid_hidden_dim 512

To use other datasets, you can change the dataset name mams to the desired one such as rest14, laptop14, twitter, small.

Run dot-GCN

python train.py --dataset mams