You will lose your current progress
Steering
Brake
Boost
Number of training laps: 500
Avarage track completion: 5%
Number of training laps: 900
Avarage track completion: 35%
Number of training laps: 4500
Avarage track completion: 99%
This 3D web app, built using Three.js, showcases a car navigated by AI models trained with Proximal Policy Optimization techniques. The models are deployed directly in the browser through TensorFlow.js, facilitating real-time interactions and allowing for immediate inference responses. A plot displaying the model's performance over 4500 training episodes provides a visual representation of the training's efficacy. For comprehensive details on the training process and more, visit the GitHub README.
Created by