This repo contains examples of edge neural networks (NNs).
We provide scripts for evaluating the following edge NNs:
HGCal Autoencoder
: Autoencoder used at the Large Hadron Collider's (LHC) Compact Muon Solenoid (CMS) experiment for compressing physics sensor data generated at the High Granularity Calorimeter (HGCal).CIFAR-10 CNN
: Convolutional NN (CNN) that classifies images provided by the CIFAR-10 dataset. These models are based on the MLPerf Tiny Benchmark.
We use conda
to manage our environment.
Install miniconda
from here if you do not already have conda
installed.
Then create the environment:
conda env create -f environment.yml
Activate the environment:
conda activate egde-nns