DQN - PongNoFrameskip-v4 Open-Source Game Model - Reinforcement Learning for Playing PongNoFrameskip Game Freely

Dqn PongNoFrameskip V4

Developed by sb3

This is a reinforcement learning model based on the DQN algorithm, specifically designed for the PongNoFrameskip-v4 environment.

Video Processing #Atari game AI #Deep Q-learning #Frame stacking processing

Downloads 16

Release Time : 6/2/2022

Model Overview

The model is trained using the stable-baselines3 library and RL Zoo framework, achieving stable performance in the Atari game Pong.

Model Features

Stable game performance

Achieved an average reward score of 20.70 in the PongNoFrameskip-v4 environment

Optimized hyperparameters

Utilizes a tuned combination of hyperparameters, including learning rate and exploration strategy

Atari-specific wrappers

Integrated Atari-specific environment wrappers for optimized game frame processing

Model Capabilities

Atari game control

Reinforcement learning decision-making

Real-time game response

Use Cases

Game AI

Pong game AI

Serves as an opponent AI in Pong, capable of competing against human players

Average reward reaches 20.70 points

Reinforcement learning research

DQN algorithm validation

Used to validate the performance of the DQN algorithm in Atari games

🚀 DQN Agent for PongNoFrameskip-v4

This project presents a trained DQN agent designed to play the PongNoFrameskip-v4 game. It leverages the stable-baselines3 library and the RL Zoo, which serves as a training framework for Stable Baselines3 reinforcement learning agents. The RL Zoo comes equipped with hyperparameter optimization and pre-trained agents.

🚀 Quick Start

✨ Features

Trained DQN agent for the PongNoFrameskip-v4 environment.
Utilizes the stable-baselines3 library and RL Zoo for training and deployment.
Includes hyperparameter optimization and pre-trained agents in the RL Zoo.

📦 Installation

The installation steps are mainly about using the RL Zoo and related libraries. You need to ensure the following repositories are accessible:

RL Zoo: https://github.com/DLR-RM/rl-baselines3-zoo
SB3: https://github.com/DLR-RM/stable-baselines3
SB3 Contrib: https://github.com/Stable-Baselines-Team/stable-baselines3-contrib

💻 Usage Examples

Basic Usage

# Download model and save it into the logs/ folder
python -m rl_zoo3.load_from_hub --algo dqn --env PongNoFrameskip-v4 -orga sb3 -f logs/
python enjoy.py --algo dqn --env PongNoFrameskip-v4  -f logs/

Advanced Usage

python train.py --algo dqn --env PongNoFrameskip-v4 -f logs/
# Upload the model and generate video (when possible)
python -m rl_zoo3.push_to_hub --algo dqn --env PongNoFrameskip-v4 -f logs/ -orga sb3

📚 Documentation

The following table provides details about the model and its training:

Property	Details
Model Type	DQN
Training Data	PongNoFrameskip-v4
Mean Reward	20.70 +/- 0.46

🔧 Technical Details

The hyperparameters used for training the DQN agent are as follows:

OrderedDict([('batch_size', 32),
             ('buffer_size', 10000),
             ('env_wrapper',
              ['stable_baselines3.common.atari_wrappers.AtariWrapper']),
             ('exploration_final_eps', 0.01),
             ('exploration_fraction', 0.1),
             ('frame_stack', 4),
             ('gradient_steps', 1),
             ('learning_rate', 0.0001),
             ('learning_starts', 100000),
             ('n_timesteps', 10000000.0),
             ('optimize_memory_usage', True),
             ('policy', 'CnnPolicy'),
             ('target_update_interval', 1000),
             ('train_freq', 4),
             ('normalize', False)])

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご