An AI engineer guide to model monitoring with Comet ML platform

With comet, you can monitor any model in real time, including key metrics associated with risk and drift. That means you can finally know if models are performing the way you expect in production. Comet.ml provides Automatic Logging for a number of popular Python Machine Learning frameworks. However, even if your library does not support automatic logging, you can still take advantage of all of Comet.ml with a few simple functions. The steps of the Comet model pipeline are: - Log a Model, via a Python SDK Experiment - Register an Experiment Model - Track Model Versions of the Registered Models - Deploy a Registered Model Comet ML monitoring platform has 2 key features: - Model Monitoring (This Tutorial) - Model Registry In this tutorial we are only focussing on Model Monitoring. The video tutorials has the following content with time-line: Video Content with Timeline: ---------------------------------------------- - (00:00) Video Start - (00:07) Video Content Intro - (01:56) Demo Notebook Introduction - (02:34) Sample notebook Tutorial - (04:20) Heart Disease Detection in Keras Experiment - (09:07) Comet ML Platform access and API Key - (09:25) Comet ML Dashboard - (10:06) Comet ML Python Package Installation - (10:38) Comet ML Python Package Configuration - (12:06) Comet ML Python Package Initialization - (14:31) Experiment Monitoring Data - (15:50) Monitoring Data and Charts Visualization - (19:10) Histogram View - (20:45) Experiment logging configuration - (24:10) Saving Experiments to GitHub - (24:46) Recap - (27:04) Credits Comet ML Example URL: https://github.com/comet-ml/comet-examples GitHub URL for the samples in the Video: https://github.com/prodramp/publiccode/tree/master/machine_learning/comet_ml Please visit: Prodramp LLC, https://prodramp.com | @prodramp https://www.linkedin.com/company/prodramp Content Creator: Avkash Chauhan (@avkashchauhan) https://www.linkedin.com/in/avkashchauhan Tags: #ai #aicloud #h2oai #driverlessai #machinelearning #cloud #mlops #model #collaboration #deeplearning #modelserving #modeldeployment #keras #tensorflow #pytorch #datarobot #datahub #aiplatform #aicloud #cometml #modelmonitoring #drift #modelregistry #modelmanagement