Matrix Factorization - PyTorch implementation

A minimal implementation of matrix factorization in PyTorch. As item/user embeddings are not defined in the model (i.e not as nn.Embedding layer), Modifications can be made for the implementation to scale up.

Running

• install the conda environment from environment.yml.
• activate the environment and run:

python MF.py --lr 0.01 --latent_dimension 10 --batch_size 256 --num_epochs 30

License

MIT