pytorch-soft-actor-critic

wassname/pytorch-soft-actor-critic

Fork 0

mirror of https://github.com/wassname/pytorch-soft-actor-critic.git synced 2026-07-28 11:24:45 +08:00

T

Pranjal TandonandGitHub 301577a4e6 Update sac.py

2018-09-14 23:47:01 +05:30

LICENSE

Initial commit

2018-08-31 17:23:01 +05:30

main.py

Update main.py

2018-09-12 09:39:49 +05:30

model.py

Update model.py

2018-09-12 08:37:55 +05:30

normalized_actions.py

Update normalized_actions.py

2018-09-13 08:45:21 +05:30

plot.py

Update plot.py

2018-09-12 15:35:23 +05:30

README.md

Update README.md

2018-09-13 08:44:27 +05:30

replay_memory.py

Update replay_memory.py

2018-09-13 08:46:07 +05:30

sac.py

Update sac.py

2018-09-14 23:47:01 +05:30

utils.py

Add files via upload

2018-08-31 17:25:08 +05:30

README.md

Description

Reimplementation of Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor.

Contributions are welcome. If you find any mistake (very likely) or know how to make it more stable, don't hesitate to send a pull request.

Requirements

Run

Use the default hyperparameters.

For SAC :

python main.py --env-name Humanoid-v2 --scale_R 20

For SAC (Hard Update):

python main.py --env-name Humanoid-v2 --scale_R 20 --tau 1 --value_update 1000

For SAC (Deterministic, Hard Update):

python main.py --env-name Humanoid-v2 --scale_R 20 --deterministic True --tau 1 --value_update 1000

Results

My results on Humanoid-v2 environment using SAC, SAC(hard update) and SAC(deterministic, hard update). This is a plot of average rewards at every 10000 step interval