Snake AI

Key Features

• Implemented Reinforcement Learning with Pytorch
• Optimized model by adjusting state and learning rate
• Introduced obstacles to challenge the AI
• Added levels of difficulty and a test framework to test validity of model

Testing the Model

1. Clone the repo
2. Run python test.py in your terminal inside the snake-ai folder
3. Select a level (1-5) with 1 being the easiest, 5 the hardest
4. Watch the AI try to play snake!

Training the Model

To observe how the model is trained, you can run python agent.py

• This project is based on the snake AI course from freeCodeCamp, which can be found here: https://www.youtube.com/watch?v=L8ypSXwyBds&t=1236s&ab_channel=freeCodeCamp.org. After going through this course, I've made my own improvements to the project, including obstacles, levels, optimizing the model, and other features.