snake-environment 0.2.8

A simple snake environment with 18 states and 4 actions

Snake Environment

The Snake Environment is a Python package designed for training machine learning models using reinforcement learning techniques. This environment mimics the OpenAI Gym interface, providing a familiar setup for those who are accustomed to using Gym for developing and testing AI models.

Features

• Game Environment: Built using Pygame, this environment allows for on-screen rendering of the snake game, which is useful for visual feedback while training models.
• Compatibility with AI Training Workflows: The SnakeGame class features a step function that returns a tuple of (state, done, reward, score), similar to environments found in OpenAI Gym.
• State Representation: The state is represented as a numpy array with 18 dimensions, providing comprehensive information about the game environment at any given step.
• Utility Methods: Includes reset and state_dimensions methods for resetting the game state and retrieving the dimensions of the state space, respectively.

Installation

To install the Snake Environment, you can use pip:

pip install snake_environment

Example Usage

from snake_environment import SnakeGame

# Initialize the environment
env = SnakeGame(render=True)  # Set render=False if you do not need to visualize the training process

# Start a new episode
state = env.reset()

# Loop until the episode is finished
done = False
while not done:
    action = model.predict(state) 
    next_state, done, reward, score = env.step(action)
    state = next_state

# Get the dimensions of the state for input layer configuration or debugging
state_dim = env.state_dimensions
print("State dimensions:", state_dim)

Contributions

Contributions are welcome! If you'd like to improve the Snake Environment, please fork this repository and submit a pull request with your proposed changes.

License

This project is licensed under the MIT License - see the LICENSE file for details.