SharedCon

1. Dataset preparation

for using shared semantics

python shared_semantics.py

If you want to specify the number of the shared semantics, or adapt another sentence embedding model,

python shared_semantics.py \
--cluster_num 10 \
--load_dataset ihc_pure \
--load_sent_emb_model simcse

for proximal anchor experiment

python noisy_positive.py

2. Preprocessing

python preprocess_dataset.py

3. Training

Adjust the hyperparameters with train_config.py then execute train.py

4. Testing

Adjust the directories and test datasets with eval_config.py then execute eval.py