Benchmark multiplicative LSTM vs. ordinary LSTM. Read this blog post about the evaluation.
Create a conda environment with:
conda env create -f environment-abstract.yml
conda env create -f environment-concrete.yml
to perfectly replicate the environment. Then activate the environment with:
source activate mlstm4reco
and install it with:
python setup.py develop
Then change into the
experiments directory and run:
./run.py 10m -m mlstm
to run the
mlstm model on the Movielens 10m dataset. Check out
./run.py -h for more help.
This project has been set up using PyScaffold 3.0.2. For details and usage information on PyScaffold see http://pyscaffold.org/.