git clone --recursive [email protected]:PreferredAI/CoVE.git
cd CoVE
conda env create -f cove.yml
conda activate cove
cd cornacc
python setup.py build_ext -i
cd ..
ln -s cornacc/cornac cornac
python main.py --batch_size 32
This will run the MoVE
model on the Diginetica
dataset, with the 1x4 architecture 1x(BPR
, GRU4Rec
, FPMC
, SASRec
).
- Session data (session data by default):
- Diginetica: 8k-12k lines; small
- RetailRocket: 230k lines; medium size
- Cosmetics: 2M lines; large
- Conversion from Rating
$\to$ session:- XING: 750k lines
- Lastfm: 4M lines
For now we prefer to use the original session datasets.