Implementation MPDA with pytorch

Data Preprocess

generate raw data for all users on device

nohup python -u ./scripts/preprocess/movielens/generate_device_data.py > ./log/generate_device_data.log 2>&1 &

generate seq data for all users on device

nohup python -u ./scripts/preprocess/movielens/generate_seq_data.py > ./log/generate_seq_data.log 2>&1 &

split train and test data for all users on device

nohup python -u ./scripts/preprocess/movielens/split.py > ./log/split_train_test_data.log 2>&1 &

generate user and mapping file

python -u scripts/preprocess/movielens/generate_mapping.py 

generate all users id json file

python scripts/preprocess/movielens/generate_all_users_list.py

generate users with train json file

python scripts/preprocess/movielens/generate_users_with_train.py

generate recall item pairs

python scripts/preprocess/movielens/generate_item_pairs.py

Initial Model

train global model NCF

nohup python -u ./model/NCF/train_global_model.py > ./log/train_global_model_NCF.log 2>&1 &

transfer model NCF

bash ./commands/ncf_movielens_50_random.sh