How to make continuous integration work with machine learning
summary
A workflow is built that lets CI run real model training and reporting: CML uses GitHub Actions (or GitLab CI) plus DVC to pull large datasets from cloud remotes, spin up GPU instances (EC2 via Docker Machine now moving to Terraform), train models (example: style transfer), push artifacts back, and post markdown metric and image reports (confusion matrix, TensorBoard link) into pull requests. It ties each model to a dataset and environment and demonstrably runs end-to-end — tutorials hit ~17,000 views.
AI-written summary · about
chapters (6)
- 0:00Start
- 1:24The machine learning workflow
- 4:34The principle of continuous integration & delivery
- 8:46Neat developments in the Git ecosystem
- 9:40Continuous Machine Learning
- 21:21Reimagining CI/CD
related talks
59mDevOps for Machine Learning Projects
17mContinuous deployment for ML: The new software development lifecycle
24mFrom Model to Production: Automating the ML Lifecycle with CI/CD
01h 17mMaking MLOps uncool again
42mCI/CD for Machine Learning
02h 10mBuild End-To-End MLOps Platform with Open Source DVC Ecosystem
23mAccelerating Model Deployment with MLOps
57mDevOps for Machine Learning