benchmark-tracker

This work is the combination between The Experiment-Impact-Tracker( https://github.com/Breakend/experiment-impact-tracker)with AI Benchmark Alpha( https://ai-benchmark.com/alpha) from M1- Master of Science Informatic in Grenoble.
This software tool can track the energy consumption of DNNs models run insight the benchmark.
You have to get the sudo permission for tracker.
Benchmark Tracker runs in Python 3.7.
Executed file: exe-file.py.
Choose model which you want to run in config.py.
you have to change the result folder each time you run in order to save the output from tracker.
It will be automatic save a csv version in order to let you read results easily.
Some visualization were saved in image_visualization folder.

In order to run the Tracker, we artificial duplicate the data 14 times for classification problem.