Launch Job

Launch the training job

Run the CircularNET_Vertex_AI_ReTraining_v1.ipynb script to configure parameters, launch the training job on Vertex AI, and start the training.

Note: Wait to export the checkpoints to a saved TF model until the script finishes running. The training job might take several days to complete. Vertex AI manages the process in the background, so you don't need to keep your computer running.

Follow these steps to monitor the training progress:

1. Log in to your Google Cloud account.
2. Navigate to Vertex AI.
3. From the Vertex AI menu, click Training.
4. On the Training page, open the Hyperparameter tuning jobs tab and select the name of your training job.
5. To observe data about the training job, perform one of the following actions:
   • Click Open TensorBoard to view detailed training metrics.
   • Click View Logs to open the log console and monitor the detailed log messages.

---

Export and test the model

After training is complete, you can export the model in various formats, such as TensorFlow, for future use when you deploy the model. The CircularNET_Vertex_AI_ReTraining_v1.ipynb script guides you through this process.

Additionally, you can test the model locally and run inferences on images to validate its performance. The script also contains information about running local tests.

Finally, deploy the exported model using an edge device or Google Cloud, depending on your deployment requirements.