Skip to content
forked from sfujim/TD3

PyTorch implementation of TD3 and DDPG for OpenAI gym tasks

Notifications You must be signed in to change notification settings

laurathepluralized/TD3

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Addressing Function Approximation Error in Actor-Critic Methods

PyTorch implementation of Twin Delayed Deep Deterministic Policy Gradients (TD3). If you use our code or data please cite the paper.

Method is tested on MuJoCo continuous control tasks in OpenAI gym. Networks are trained using PyTorch 0.4 and Python 2.7.

Usage

The paper results can be reproduced exactly by running:

./experiments.sh

Experiments on single environments can be run by calling:

python2 main.py --env HalfCheetah-v1

Hyper-parameters can be modified with different arguments to main.py. We include an implementation of DDPG (DDPG.py) for easy comparison of hyper-parameters with TD3, this is not the implementation of "Our DDPG" as used in the paper (see OurDDPG.py).

Algorithms which TD3 compares against (PPO, TRPO, ACKTR, DDPG) can be found at OpenAI baselines repository.

Results

Learning curves found in the paper are found under /learning_curves. Each learning curve are formatted as NumPy arrays of 201 evaluations (201,), where each evaluation corresponds to the average total reward from running the policy for 10 episodes with no exploration. The first evaluation is the randomly initialized policy network (unused in the paper). Evaluations are peformed every 5000 time steps, over a total of 1 million time steps.

Numerical results can be found in the paper, or from the learning curves. Video of the learned agent can be found here.

About

PyTorch implementation of TD3 and DDPG for OpenAI gym tasks

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 95.3%
  • Shell 4.7%