torchmf
matrix factorization in PyTorch
Installation
pip
pip install -r requirements.txt
conda
conda env create -f environment.yml
conda activate torchmf
Example
$ python run.py --example explicit
( 1 ): 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 89/89 [00:01<00:00, 85.88it/s, train_loss=7.9]
Epoch: 1 train: 15.04790 val: 8.84972
( 2 ): 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 89/89 [00:01<00:00, 84.06it/s, train_loss=2.96]
Epoch: 2 train: 4.34132 val: 4.04638
( 3 ): 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 89/89 [00:01<00:00, 81.54it/s, train_loss=1.51]
Epoch: 3 train: 1.87918 val: 2.43315
( 4 ): 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 89/89 [00:01<00:00, 85.55it/s, train_loss=1.19]
Epoch: 4 train: 1.21419 val: 1.80296
( 5 ): 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 89/89 [00:00<00:00, 90.87it/s, train_loss=0.945]
Epoch: 5 train: 0.99693 val: 1.49770
( 6 ): 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 89/89 [00:00<00:00, 89.33it/s, train_loss=0.914]
Epoch: 6 train: 0.90174 val: 1.33501
( 7 ): 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 89/89 [00:00<00:00, 115.70it/s, train_loss=0.83]
Epoch: 7 train: 0.85230 val: 1.23783
( 8 ): 100%|ββββββββββββββββββββββββββββββββββββββββββββββββββββββ| 89/89 [00:01<00:00, 88.85it/s, train_loss=0.879]
Epoch: 8 train: 0.82072 val: 1.17781
( 9 ): 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 89/89 [00:00<00:00, 119.93it/s, train_loss=0.766]
Epoch: 9 train: 0.79898 val: 1.13976
(10 ): 100%|βββββββββββββββββββββββββββββββββββββββββββββββββββββ| 89/89 [00:00<00:00, 122.18it/s, train_loss=0.736]
Epoch: 10 train: 0.77820 val: 1.10951which looks something like
