Code

Recurrent PPO#53

Code

Merged

araffin merged 60 commits intomasterStable-Baselines-Team/stable-baselines3-contrib:masterfrom feat/ppo-lstmStable-Baselines-Team/stable-baselines3-contrib:feat/ppo-lstmCopy head branch name to clipboardMay 30, 2022

+1,988-28

Commits

Commits on Nov 22, 2021

• Running (not working yet) version of recurrent PPO

  araffin

  committed
  85c9a50

Commits on Nov 23, 2021

• Fixes for multi envs

  araffin

  committed
  b92da74

• Save WIP, rework the sampling

  araffin

  committed
  d9f9c4e

• Add Box support

  araffin

  committed
  97ec8ec

• Fix sample order

  araffin

  committed
  a890976

Commits on Nov 25, 2021

• Being cleanup, code is broken (again)

  araffin

  committed
  7fecd9f

• First working version (no shared lstm)

  araffin

  committed
  0ddc3f6

• Start cleanup

  araffin

  committed
  5ef313b

• Try rnn with value function

  araffin

  committed
  c803ac9

• Re-enable batch size

  araffin

  committed
  0c8ab15

• Deactivate vf rnn

  araffin

  committed
  eb1e6c1

• Allow any batch size

  araffin

  committed
  f013346

Commits on Nov 26, 2021

• Add support for evaluation

  araffin

  committed
  a14f2ce

• Add CNN support

  araffin

  committed
  362dec4

• Fix start of sequence

  araffin

  committed
  5b162db

• Allow shared LSTM

  araffin

  committed
  954e6dd

Commits on Nov 28, 2021

• Rename mask to episode_start

  araffin

  committed
  832093d

• Fix type hint

  araffin

  committed
  2a9c956

• Enable LSTM for critic

  araffin

  committed
  15c080a

• Clean code

  araffin

  committed
  0d304aa

• Fix for CNN LSTM

  araffin

  committed
  1dc78b4

• Fix sampling with n_layers > 1

  araffin

  committed
  deaa7b4

Commits on Nov 29, 2021

• Add std logger

  araffin

  committed
  ced6aee

Commits on Nov 30, 2021

• Update wording

  araffin

  committed
  b81fdff

Commits on Dec 1, 2021

• Merge branch 'master' into feat/ppo-lstm

  araffin

  committed
  a2a201f

Commits on Dec 10, 2021

• Merge branch 'master' into feat/ppo-lstm

  araffin

  committed
  754e0a3

Commits on Dec 27, 2021

• Rename and add dict obs support

  araffin

  committed
  c9c0b4e

• Fixes for dict obs support

  araffin

  committed
  a4b769f

• Do not run slow tests

  araffin

  committed
  5cadc14

Commits on Dec 29, 2021

• Merge branch 'master' into feat/ppo-lstm

  araffin

  committed
  617d76f

• Fix doc

  araffin

  committed
  c1f8812

• Update recurrent PPO example

  araffin

  committed
  579e7d0

• Update README

  araffin

  committed
  bd2d5e2

Commits on Jan 3, 2022

• Merge branch 'master' into feat/ppo-lstm

  araffin

  committed
  072622b

Commits on Jan 19, 2022

• Merge branch 'master' into feat/ppo-lstm

  araffin

  committed
  c113324

Commits on Feb 22, 2022

• Merge branch 'master' into feat/ppo-lstm

  araffin

  committed
  4adb3ea

Commits on Feb 23, 2022

• Merge branch 'master' into feat/ppo-lstm

  araffin

  committed
  32ec1b2

• Use Pendulum-v1 for tests

  araffin

  committed
  0f0ce0b

• Fix image env

  araffin

  committed
  116d0a6

Commits on Mar 8, 2022

• Speedup LSTM forward pass (#63)

  Show description for c32bb74
  

  Walon1998

  and

  araffin

  authored
  c32bb74

Commits on Apr 12, 2022

• Merge branch 'master' into feat/ppo-lstm

  araffin

  committed
  638dfb2

• Fixes

  araffin

  committed
  86e0f6f

• Remove OpenAI sampling and improve coverage

  araffin

  committed
  3fc6e51

• Sync with SB3 PPO

  araffin

  committed
  88f9504

• Pass state shape and allow lstm kwargs

  araffin

  committed
  662f218

• Update tests

  araffin

  committed
  fd06850

• Add masking for padded sequences

  araffin

  committed
  f5e9b34

• Update default in perf test

  araffin

  committed
  1cd27da

Commits on Apr 15, 2022

• Remove TODO, mask is now working

  araffin

  committed
  c52959b

Commits on Apr 25, 2022

• Merge branch 'master' into feat/ppo-lstm

  araffin

  authored
  18e6230

Commits on May 1, 2022

• Add helper to remove duplicated code, remove hack for padding

  araffin

  committed
  673d23a

Commits on May 8, 2022

• Enable LSTM critic and raise threshold for cartpole with no vel

  araffin

  committed
  e271d03

• Fix tests

  araffin

  committed
  73bb89c

Commits on May 18, 2022

• Update doc and tests

  araffin

  committed
  efa6181

• Doc fix

  araffin

  committed
  564d428

Commits on May 29, 2022

• Fix for new Sphinx version

  araffin

  committed
  408ed24

• Merge branch 'master' into feat/ppo-lstm

  araffin

  committed
  d917487

• Fix doc note

  araffin

  committed
  6acb64a

Commits on May 30, 2022

• Switch to batch first, no more additional swap

  araffin

  committed
  5fd8be7

• Add comments and mask entropy loss

  araffin

  committed
  7a1d3e8