predicting traffic for the NeurIPS'19 challenge

1. Install anaconda or miniconda.
2. Install dependencies and activate environment:

conda env create environment.yml
conda activate traffic4cast

Start the server, for example on lenovo-storage:

python hypertune.py --run-id test --n-workers 4

Start the workers, for example on other lenovo's – lenovo{3,4,6,7}:

python hypertune.py --run-id test --worker

The hyper-parameter tuning can be runned also locally, but it's unlikely to be efficient:

python hypertune.py --n_workers 4 &
for i in {1..4}; do
	sleep 0.1
	python hypertune.py --worker &
done