Basic-RL-Algorithms

Description

This project is used to implement all kinds of RL basic algorithms.

File Structure

1. algo to save the code regarding the algorithms' implementations
2. docs to save the descriptions of the corresponding algorithms
3. env to provide the env_wrapper (gym, gymnasium)
4. model to save the network structure of policy, state-value function, action-value function
5. util to save the code regrading the common tricks

The implemented algorithms

1. RainBow-DQN(initial, double, PER, dueling, multi-step, noisy)
2. REINFORCE
3. A2C (A2C Continuous / A2C Discrete)
4. DDPG
5. TD3

How to use

conda create --name basic_rl python=3.10

conda activate basic_rl

pip install torch torchvision

conda install swig

conda install gymnasium[all]

pip install tensorboardX tensorboard

apt-get install patchelf