AWS DeepRacer - Model Training Using the AWS DeepRacer Console
This guide walks you through building, training, and evaluating your first reinforcement learning model via the AWS DeepRacer console.
To get started building your first model, follow the steps below.
1. In your AWS account, go to the Management Console.
2. Choose us-east-1 region at the top right corner of the Regions dropdown menu.
3. From the top left of the console, click Services, type DeepRacer in the search box, and select AWS DeepRacer. That will open the AWS DeepRacer console.
4. On the landing page, click Create model.
5. Create a model name and description.
6. Click Create resource.
7. Under Environment simulation, select a track from the list. It’s recommended to start with the “re:Invent 2018 Track” and then explore more tracks from there.
8. Under Action space, familiarize yourself with these settings and accept the defaults.
9. Read through both basic and advanced reward functions to familiarize yourself with the code. For the purposes of this exercise, choose to insert code for the basic reward function and leave the code unchanged.
10. Expand Algorithm settings and review the different hyperparameters available to set. For this exercise, accept the default hyperparameter settings.
11. For the Stop conditions go ahead and choose a max time of 60 minutes and click Start training.
12. Select the name of the model to watch the live stream in the simulator. Notice how your car is moving and become familiar with the general look and feel of the simulation. The new training will initiate in about 6 minutes. You then must wait for the training job to complete before proceeding. If you choose a maximum time of 60 minutes, it will take up to 60 minutes for this training job to complete. It is complete when the status reads “Ready.”
Note: Since the training job may take more than an hour.
13. Now, select the name of the model you just trained and click Start new evaluation.
14. Select the same track you used for training.
15. Click Start evaluation. The evaluation results will update in approximately 5 minutes.
16. Before the evaluation job completes, take some time to watch the simulator to see how your car’s performing in real time. Look for things that you might want to tweak in the future and note them down. You’ll have a chance to apply some of those ideas in a later exercise.
17. Watch the evaluation results update and write them down. You will use your results from this model as a benchmark that you can compare to as you tweak your model in later exercises.
Additional Resources
• For more information about AWS DeepRacer, see https://aws.amazon.com/deepracer/.
• To troubleshoot and collaborate about the AWS DeepRacer, see https://forums.aws.amazon.com/forum.jspa?forumID=318.
Comments