censeModels

todo

Setup

python3.7 -m venv env/
source env/bin/activate
pip install -r requirements.txt

Download sample dataset

python download.py

Eval fast

Ensure Lorient-1k.npy files are in ./data/ See code for options

python inference.py --exp TVBCense_Fast

Eval slow

python fast_to_slow.py
python inference.py --exp TVBCense_Slow --dataset Lorient-1k_slow

Run without evaluation

python inference.py --exp TVBCense_Fast --dataset Lorient-1k -no_metrics -force_recompute
python inference.py --exp TVBCense_Slow --dataset Lorient-1k_slow -no_metrics -force_recompute

Each produces 2 files located in the eval_output directory: _presence.npy with boolean values indicating presence of a given source and _scores.npy floating point values with likelihood of presence