arya-ebrahimi / rl-playground

tabular and deep rl algorithms

RL Playground

Simple implementations of RL algorithms.

Table of Contents

1. PPO: A simple PPO implementation using pytorch both for continuous and discrete action spaces.

2. SAC: SAC for continuous action spaces, tested on hopper-v4 and pendulum-v1.

3. TD3: TD3 implementation from scratch using pytorch, and tested on HalfCheetah and Pendulum envs.

4. DDPG: Deep Deterministic Policy Gradient implementation.

5. DQN: DQN implementation using pytorch. I used the pytorch documentation in RL section with some small changes and a different environment. using both ReLU and Fuzzy Tiling Activations(FTA)

6. Tabular: Implementations of tabular algorithms from "Reinforcement Learning: an introduction" tested on different gridworlds or gym environments.