Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on Mar 15, 2021

• final edits

  Jason-CKY

  committed
  abac748

  Copy full SHA for abac748

Commits on Mar 7, 2021

• added close box training progress

  Jason-CKY

  committed
  c5c9d18

  Copy full SHA for c5c9d18

Commits on Jan 7, 2021

• Changed option critic to SAC-like off-policy updates

  Jason-CKY

  committed
  46ee719

  Copy full SHA for 46ee719

Commits on Dec 27, 2020

• added multiple gpu support

  Jason-CKY

  committed
  351beeb

  Copy full SHA for 351beeb

• added VAE support to all the RL algorithms

  Jason-CKY

  committed
  f4a4b14

  Copy full SHA for f4a4b14

Commits on Dec 25, 2020

• Fixed loading weights saved from using nn.DataParallel wrapper

  Jason-CKY

  committed
  795ac1d

  Copy full SHA for 795ac1d

Commits on Dec 24, 2020

• changed model type to be a config setting instead of a config for each model architecture

  Jason-CKY

  committed
  33e5427

  Copy full SHA for 33e5427

Commits on Dec 23, 2020

• Added VAE training code

  Jason-CKY

  committed
  82b78ec

  Copy full SHA for 82b78ec

Commits on Dec 17, 2020

• Debugged wrong variable references

  Jason-CKY

  committed
  d9d6455

  Copy full SHA for d9d6455

• Added CNN architecture to each of the RL algorithms, as well as general image rendering env wrapper

  Jason-CKY

  committed
  04c6c2e

  Copy full SHA for 04c6c2e

Commits on Dec 16, 2020

• Commented away agent evaluation during training as it is a bottleneck to training speed

  Jason-CKY

  committed
  ded0743

  Copy full SHA for ded0743

• Added CNN network for ddpg

  Jason-CKY

  committed
  d19d2f6

  Copy full SHA for d19d2f6

Commits on Dec 13, 2020

• Redid the normalize observation env wrapper using online algorithm and added env pickling options

  Jason-CKY

  committed
  f26da43

  Copy full SHA for f26da43

Commits on Oct 30, 2020

• Debugged algorithms

  Jason-CKY

  committed
  da4cb3f

  Copy full SHA for da4cb3f

• Changed saving model weights logic

  Jason-CKY

  committed
  acf9aaa

  Copy full SHA for acf9aaa

Commits on Oct 25, 2020

• added .train() and .eval() functions to networks, only applied normalize observations to on-policy algorithms, used pickle instead of json for logger due to I/O time

  Jason-CKY

  committed
  211b675

  Copy full SHA for 211b675

• changed logs

  Jason-CKY

  committed
  39ab731

  Copy full SHA for 39ab731

Commits on Oct 23, 2020

• Changed settings for multiple trials

  Jason-CKY

  committed
  94ec60f

  Copy full SHA for 94ec60f

• Added tricks to improve on-policy learning and variance plotting

  Jason-CKY

  committed
  f1ee42d

  Copy full SHA for f1ee42d

Commits on Oct 7, 2020

• added graphs

  Jason-CKY

  committed
  d04d6c3

  Copy full SHA for d04d6c3

Commits on Oct 5, 2020

• Results and changed ppo/trpo settings

  Jason-CKY

  committed
  469ca70

  Copy full SHA for 469ca70

Commits on Oct 3, 2020

• Added trpo algorithm but dk why does not work for discrete case

  Jason-CKY

  committed
  a67209e

  Copy full SHA for a67209e

Commits on Sep 24, 2020

• added td3 test cases

  Jason-CKY

  committed
  89448a0

  Copy full SHA for 89448a0

Commits on Sep 23, 2020

• Edited TD3 and added it to train script

  Jason-CKY

  committed
  f191fff

  Copy full SHA for f191fff

• first draft of td3

  Jason-CKY

  committed
  7ee93b9

  Copy full SHA for 7ee93b9

Commits on Sep 21, 2020

• TD3 WIP

  Jason-CKY

  committed
  aac772b

  Copy full SHA for aac772b

Loading