Each of the tutorials presented here walks you through a specific artificial intelligence (AI) workflow, created to represent the most common tasks and to illustrate the capabilities of Agent Platform. Choose the tutorial that best matches your data type and AI task. After following the tutorial, you can use the patterns that you have learned to solve your own AI problem. Agent Platform offers Google Cloud console tutorials and notebook tutorials that use the Python SDK. You can open a notebook tutorial directly in Colab, download the notebook to your preferred environment, or open the notebook tutorial in Agent Platform Workbench.
Train a classification model for tabular data
|
Create a Agent Platform dataset from tabular data, and then train a classification model with AutoML. Deploy the model to an endpoint and make online predictions.
Google Cloud console: You can choose tutorial guides with step-by-step instructions for the Google Cloud console.
Notebook: You can choose to run this tutorial as a notebook. |
Train a regression model for tabular data
|
|
Create a Agent Platform dataset from tabular data, and then train a regression model with AutoML. Deploy the model to an endpoint and make online predictions or make predictions in batch format.
Notebook: You can choose to run this tutorial and make online predictions using
a notebook.
Notebook: You can choose to run this tutorial and make batch predictions using
a notebook. |
Train a time-series forecasting model for tabular data
|
|
Create a Agent Platform dataset from tabular data, and then train a forecasting model with AutoML. Make predictions in batch format.
Notebook: You can choose to run this tutorial as a notebook. |
Train a classification model for image data
|
Create a Agent Platform dataset for image data, and then train a classification model with AutoML. Deploy the model to an endpoint and make online predictions.
Google Cloud console: You can choose tutorial guides with step-by-step instructions for the Google Cloud console. |
How to open a notebook in Agent Platform Workbench
To open a notebook tutorial in a Agent Platform Workbench instance:
- Click the Agent Platform Workbench link in the notebook list. The link opens the Agent Platform Workbench console.
- In the Deploy to notebook screen, type a name for your new Agent Platform Workbench instance and click Create.
- In the Ready to open notebook dialog that appears after the instance starts, click Open.
- On the Confirm deployment to notebook server page, select Confirm.
- Before running the notebook, select Kernel > Restart Kernel and Clear all Outputs.
What's next
- View the full list of Agent Platform notebook tutorials.
- Learn more about model training.