PyTorch implementation of nonsymmetric determinantal point process (DPP) learning. For details, see our Learning Nonsymmetric Determinantal Point Processes paper (NeurIPS 2019 poster).

Install required Python packages:

$ pip install --user --requirement requirements.txt

Train and evaluate a nonsymmetric DPP model using the Amazon apparel baby registry dataset:

$ cd src
$ python main.py --dataset_name basket_ids --input_file data/1_100_100_100_apparel_regs.csv --num_sym_embedding_dims 30 --num_nonsym_embedding_dims 30

For a full list of command-line options:

$ python main.py --help