Hands on with Cloud AI
summary
Hands-on demo builds a TensorFlow regression on the auto MPG dataset, packages training code as a Docker container, pushes it to Container Registry, runs a Vertex AI training job that writes a SavedModel to Cloud Storage, and deploys that model to an endpoint via the Vertex AI SDK. Training finishes in about 15 minutes for this toy example, online prediction returns ~16.3 MPG for a sample, region used is US-central1, cost reported around $2.
AI-written summary · about
chapters (12)
- 0:00Workshop introduction
- 1:50Vertex AI overview
- 3:47Environment setup
- 7:08Configuring services
- 10:24Introduction to the dataset
- 12:39Containerizing training code
- 17:28Building and pushing images
- 21:10Running custom training
- 29:49Model deployment
- 35:55Model endpoint verification
- 38:16Getting predictions
- 42:29Summary and resources
related talks
40mVertex AI: Pipelines for your MLOps workflows
45mVertex AI: Pipelines for your MLOps workflows
18mBuilding Efficient PyTorch Models on Vertex AI
33mFaster model training and experimentation with Vertex AI
38mReal-world Machine Learning with TensorFlow and Cloud AI
11mBecome a Cloud Machine Learning developer in 5 steps
26mBuilding Portable ML Pipelines for Development and Production
51mCloud AI: How to Get Started Injecting AI Into Your Applications