An official code for the NFCalib paper
We are using a number of common tabular datasets as well as well known MNIST.
To download it all just run: cd src/data && python3 get_data.py
First, one should build a conda environment:
conda env create -f environment.yml
Currently there are three ways to run or recreate experiments:
- Running a single experimnet from a jupyter notebook:
notebooks/NFCalib_notebook.ipynb - Running a single experiment from a python script:
src/nfcalib.py - Running a grid of experiments on a cluster:
cluster/run_cluster_jobs.py
Most of models in the paper was trained via cluster.