https://github.com/m8nt0/frozenlake-q-learning-project

demonstrates how to use Q-Learning to solve the FrozenLake environment from OpenAI Gym. It includes advanced features such as logging, configuration files, model saving/loading, plotting, and hyperparameter tuning.
https://github.com/m8nt0/frozenlake-q-learning-project

Last synced: 1 day ago
JSON representation

demonstrates how to use Q-Learning to solve the FrozenLake environment from OpenAI Gym. It includes advanced features such as logging, configuration files, model saving/loading, plotting, and hyperparameter tuning.

• Host: GitHub
• URL: https://github.com/m8nt0/frozenlake-q-learning-project
• Owner: m8nt0
• License: mit
• Created: 2024-05-23T03:53:16.000Z (9 months ago)
• Default Branch: main
• Last Pushed: 2024-05-23T04:21:19.000Z (9 months ago)
• Last Synced: 2024-05-23T05:25:02.364Z (9 months ago)
• Language: Python
• Size: 7.81 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# FrozenLake-Q-Learning-Project

demonstrates how to use Q-Learning to solve the FrozenLake environment from OpenAI Gym. It includes advanced features such as logging, configuration files, model saving/loading, plotting, and hyperparameter tuning.

## Requirements

- Python 3.6+

- numpy

- gym

- matplotlib

- scikit-optimize

- pyyaml

## Project Structure

- `configs/`: Contains configuration files.

- `logs/`: Contains logs and saved data.

- `models/`: Contains saved models.

- `plots/`: Contains generated plots.

- `src/`: Contains source code.

## Setup

1. **Install the required packages:**

   ```bash

   pip install -r requirements.txt

## Instructions

### Configure the Hyperparameters

Configure the hyperparameters in `configs/params.json`.

### Run the Training Script

    ```bash

    python src/main.py --mode train

## Run the Evaluation Script

    ```bash

    python src/main.py --mode evaluate

## Configuration

- **configs/config.json**: Contains environment and file paths configurations.

- **configs/params.json**: Contains hyperparameters for the Q-Learning algorithm.

## Logging and Plotting

- Logs are saved in the `logs/` directory.

- Training progress plots are saved in the `plots/` directory.

## Advanced Features

- **Logging**: Logs training progress to a file.

- **Configuration Files**: Use JSON for environment and hyperparameter configurations.

- **Model Saving and Loading**: Save and load the Q-table to and from a file.

- **Plotting**: Generate plots for training rewards and episode lengths.

- **Hyperparameter Tuning**: Supports basic hyperparameter tuning using a configuration file.

## Example Usage

### Training the Agent

To train the Q-Learning agent, run:

    ```bash

    python src/main.py --mode train

## Evaluating the Agent

To evaluate the trained Q-Learning agent, run:

    ```bash

    python src/main.py --mode evaluate