Python notebooks require some environment set up, specified below.
I re-used my old downloading code, so you'll need to run the entire urls.ipynb notebook and then the
urls-eval.ipynb notebook before being able to use the main evalutaion notebook, svms.ipynb,
which relies on data downloaded and sorted/split from those two notebooks.
First, install Anaconda 3 and add it to your path.
To be done when you clone or move this repo:
conda env create -f environment.yaml
Should be done once per session:
source activate env2020
To save new dependencies:
conda env export --no-builds | grep -v "prefix: " > environment.yaml
To update out-of-date local environment with new dependencies in the environment.yaml file:
conda env update -f environment.yaml --prune
# all new envs need my patch here
pip uninstall torchfm
pip install --force-reinstall git+https://github.com/vlad17/pytorch-fm/
For GPU machines, you'll need to install a gpu version of pytorch over this env.