Improve PPO algorithm - Githubissues

medipixel / rl_algorithms

Structural implementation of RL key algorithms

https://www.medipixel.io/

MIT License

509 stars 63 forks source link

Improve PPO algorithm #312

Closed isk03276 closed 3 years ago

isk03276 commented 3 years ago

Improve PPO algorithm with continuous action.

Don't use clipped value loss
Use fixed logstd parameters to sample an action
Change backbone modules can be shared

[tested in lunarlander continuous v2] W B Chart 7_1_2021, 5_25_50 PM

lgtm-com[bot] commented 3 years ago

This pull request introduces 1 alert when merging 8856de0d26baefbdca6367e493b2de61fc045694 into 13f313bcaa7e8614106371c82d7cc387e42da124 - view on LGTM.com

new alerts:

1 for `__init__` method calls overridden method

lgtm-com[bot] commented 3 years ago

This pull request introduces 1 alert when merging d05cc0ab3e1da184b6becfc2dbe2004b01d77c88 into b3df31e62397e8ecef6973a5fd4696ee3d295b58 - view on LGTM.com

new alerts:

1 for `__init__` method calls overridden method

lgtm-com[bot] commented 3 years ago

This pull request introduces 1 alert when merging 8cc7d05bf07c9bcd3aa799ded7d47514cd0ff8c9 into b3df31e62397e8ecef6973a5fd4696ee3d295b58 - view on LGTM.com

new alerts:

1 for `__init__` method calls overridden method

lgtm-com[bot] commented 3 years ago

This pull request introduces 1 alert when merging 1bf052d3a96e196a42354e0b188fa2151ed8a595 into b3df31e62397e8ecef6973a5fd4696ee3d295b58 - view on LGTM.com

new alerts:

1 for `__init__` method calls overridden method

jinPrelude commented 3 years ago

@all-contributors please add @isk03276 for code

allcontributors[bot] commented 3 years ago

@jinPrelude

I've put up a pull request to add @isk03276! :tada:

lgtm-com[bot] commented 3 years ago

This pull request introduces 1 alert when merging c70ff3cc619a5f966152609bbf6069402f1aee0c into b3df31e62397e8ecef6973a5fd4696ee3d295b58 - view on LGTM.com

new alerts:

1 for `__init__` method calls overridden method