predicting traffic for the NeurIPS'19 challenge
- Install anaconda or miniconda.
- Install dependencies and activate environment:
conda env create environment.yml
conda activate traffic4castStart the server, for example on lenovo-storage:
python hypertune.py --run-id test --n-workers 4Start the workers, for example on other lenovo's – lenovo{3,4,6,7}:
python hypertune.py --run-id test --workerThe 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