How-to guides
-
Setting up GCP accounts and projects
Setting up a Cloud Billing account and a project.
-
Activating Cloud TPU APIs
Activating Compute Engine and Cloud TPU APIs.
-
Connecting to Cloud Storage buckets
Storing your machine learning data on a Cloud Storage bucket. Giving your Cloud TPU access to data objects in the bucket.
-
Creating and deleting TPUs
Setting up Cloud TPU or deleting nodes step-by-step.
-
Using Cloud TPU Tools
Monitoring Cloud TPU and model training using TensorBoard and command line tools.
-
Setting up TensorBoard
Setting up TensorBoard to visualize and monitor the output and performance of your training application.
-
Cloud TPU monitoring with StackDriver
Cloud TPU monitoring with Stackdriver. Displays logs and supports creating log-based metrics for the Cloud TPU runtime binary. Provides tools for creating dashboards and alerts based on log metrics.
-
Preparing the ImageNet dataset
How to download, preprocess, and upload the ImageNet dataset to a Cloud Storage bucket.
-
Preparing the COCO dataset
How to download, preprocess, and upload the COCO dataset to a Cloud Storage bucket.
-
Using TPUs with GKE
A guide to setting up Cloud TPU with Google Kubernetes Engine.
-
Using preemptible TPUs
How and why you can allow Cloud TPU to preempt your TPU for usage by other workloads.
-
Cloud TPU audit logs
Accessing and using Cloud TPU audit logs.
-
Cloud TPU version switching
Switching software versions on your Cloud TPU.