Shreya Shankar • 1/3/2022

The Modern ML Monitoring Mess: Failure Modes in Extending Prometheus (3/4)

This technical article details an experiment to extend Prometheus for monitoring a machine learning pipeline. It examines the difficulties, from confusing terminology (like 'metric' and 'label') to code complexity and algorithmic inefficiencies, using a concrete example of a binary classification task for taxi tips.

0 comments

#metrics #Model Deployment #Mlop