A collection of my implemented simple & intermediate RL agents for games like Pacman, Lunarlander, Pong, SpaceInvaders, Frozenlake, Taxi, Pixelcopter, Pyramids and a lot more by implementing various DRL algorithms using gym, unity-ml, pygame, sb3, rl-zoo, pandagym and sample factory libraries. A lot of agents implemented as part of this free tutorial series on DRL at HuggingFace: Link by Thomas Simonini. The rests were done based on various relevant resources and my ideas. I have added some links in Acknowledgement section below.
To see my advanced & complex agents for complex games like soccer, rubiks-cube, mortal kombat, street fighter, vizdoom, montezuma, kungfu-master and more; visit this repository: https://github.com/hishamcse/Advanced-DRL-Renegades-Game-Bots
- Q-learning + DQN
- Bellman Equation + Monte Carlo
- Hyperparameter Tuning using Optuna
- Policy Gradient
- DDPG & DDPGfD
- Unity Ml Agents
- Advantage Actor Critic(A2C) for Robotics
- Proximal Policy Optimization(PPO) with All Variants
- Clipped Surrogate Objective Function
- Sample Factory
| Environments | Libraries Used(includes HF) | Algos | Kaggle Notebooks |
|---|---|---|---|
| LunarLander-v2 | gym, stable-baselines3 | PPO | Link |
| LunarLander-v2 | gym | DQN(Scratch) | Link |
| Huggy | unity-mlagents | PPO | Link |
| FrozenLake-v1 (all variants) | gym | Q-Learn (Scratch) | Link |
| Taxi-v3 | gym | Q-Learn (Scratch) | Link |
| SpacesInvadersNoFrameskip-v4 | RL-Baselines3-Zoo, gym, atari | DQN(CNNPolicy) | Link |
| CartPole-v1 | stable-baselines3, sb3-contrib, optuna, gym, atari, RL-Baselines3-Zoo | A2C(MlpPolicy) | Link |
| PongNoFrameskip-v4 | stable-baselines3, sb3-contrib, optuna, gym, atari, RL-Baselines3-Zoo | PPO(CNNPolicy) | Link |
| BreakoutNoFrameskip-v4 | stable-baselines3, sb3-contrib, optuna, gym, atari, RL-Baselines3-Zoo | PPO(CNNPolicy) | Link |
| MsPacman-v5 | gym, atari | DQN(CNNPolicy) - Scratch | Link |
| CartPole-v1 | pytorch, gym | Policy Gradient-scratch | Link |
| Pixelcopter-PLE-v0 | pytorch, gym, pygame | Policy Gradient-scratch | Link |
| Pendulum-v1 | pytorch, gym | DDPG-Scratch | Link |
| Pendulum-v1 | pytorch, gym | DDPGfD - Scratch | Link |
| Snowball-Target | unity-mlagents | PPO | Link |
| Pyramids | unity-mlagents | PPO + RND | Link |
| PandaReachDense-v3 | gym, panda-gym, stable-baselines3 | A2C(MultiInputPolicy) | Link |
| PandaPickAndPlace-v3 | gym, panda-gym, stable-baselines3 | A2C(MultiInputPolicy) | Link |
| LunarLander-v2 | pytorch, gym | PPO (All Variations Scratch) | Link |
Find all my traned agents at hishamcse agents
- Deep RL Course
- Thomas Simonini
- Deep RL Course Leaderboard - HuggingFace
- Open RL Leaderboard - HuggingFace
- Stable-Baseline3 Agents - HuggingFace
- Stable-Baseline3
- All PPO Implementation Details
- CleanRL Single File Implementations
- Unity-mlagents
- Coursera Machine Learning Specialization Course 3 by Andrew Ng
- gym
- Sample_Factory
- RL-Zoo3
- Panda-gym
- Tools for Robotic RL
- Optuna
- QRDQN - Quantile Regression DQN
- Advanced DQN Pacman, DQN Pacman
- All Policy Gradient Algorithms Implementations
- DDPGfD - DDPG from Demonstrations
- Deep RL Paradise
- OpenAI Spinning Up
- Arthur Juliani Simple Reinforcement Learning in Tensorflow Series
- Awesome Deep RL
- Dennybritz RL
- HuggingFace Hub Python Library
You can view my Deep Reinforcement Learning projects