wassname/pytorch-a2c-ppo-acktr
1
0
Fork 0
mirror of https://github.com/wassname/pytorch-a2c-ppo-acktr.git synced 2026-06-27 16:20:05 +08:00
Code Issues Packages Projects Releases Wiki Activity
10 Commits 1 Branch 0 Tags
5d29401658262246006e50d52a7e7dc6346d1604
T
Clone
Open with VS Code Open with VSCodium Open with Intellij IDEA
Download ZIP Download TAR.GZ Download BUNDLE
Ilya Kostrikov 5d29401658 Add PPO
2017-09-17 18:43:45 -04:00
imgs
Add images
2017-09-14 07:08:50 -04:00
envs.py
Initial commit
2017-09-07 19:45:57 -04:00
LICENSE
Create LICENSE
2017-09-07 20:01:02 -04:00
main.py
Add PPO
2017-09-17 18:43:45 -04:00
model.py
Initial commit
2017-09-07 19:45:57 -04:00
README.md
Add PPO
2017-09-17 18:43:45 -04:00
vizualize_atari.py
Add visualization
2017-09-13 18:34:13 -04:00

README.md

pytorch-a2c-ppo

This is a PyTorch implementation of Advantage Actor Critic (A2C), a synchronous deterministic version of A3C "Asynchronous Methods for Deep Reinforcement Learning" and Proximal Policy Optimization (PPO). Also see the OpenAI posts: A2C/A3C and PPO for more information.

This implementation is inspired by the OpenAI baselines for A2C and PPO. It uses the same hyper parameters and the model since they were well tuned for Atari games.

Contributions

Contributions are very welcome. If you know how to make this code better, don't hesitate to send a pull request.

Usage

A2C

python main.py --env-name "PongNoFrameskip-v4"

PPO

python main.py --env-name "PongNoFrameskip-v4" --algo ppo --use-gae --num-processes 8 --num-steps 256 --vis-interval 1 --log-interval 1

Results

A2C

BreakoutNoFrameskip-v4

SeaquestNoFrameskip-v4

QbertNoFrameskip-v4

beamriderNoFrameskip-v4

PPO

Coming soon.

Reference in New Issue View Git Blame Copy Permalink
S
Description
PyTorch implementation of Advantage Actor Critic (A2C), Proximal Policy Optimization (PPO) and Scalable trust-region method for deep reinforcement learning using Kronecker-factored approximation (ACKTR).
Readme MIT 233 KiB
Languages
Python 100%