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 Feb 24, 2021

• solved open_box

  Jason-CKY

  committed
  79a2f0f

  Copy full SHA for 79a2f0f

Commits on Jan 30, 2021

• Added DAC+PPO benchmark and code

  Jason-CKY

  committed
  e70767f

  Copy full SHA for e70767f

Commits on Jan 29, 2021

• Added DAC_PPO Algorithm

  Jason-CKY

  committed
  aa12715

  Copy full SHA for aa12715

Commits on Jan 23, 2021

• Changed option_critic code

  Jason-CKY

  committed
  e38969e

  Copy full SHA for e38969e

Commits on Jan 19, 2021

• option-critic extended to VAE

  Jason-CKY

  committed
  2689e8e

  Copy full SHA for 2689e8e

• option_critic update

  Jason-CKY

  committed
  12c5e20

  Copy full SHA for 12c5e20

Commits on Jan 13, 2021

• outline of Q loss function

  Jason-CKY

  committed
  5fc68bf

  Copy full SHA for 5fc68bf

Commits on Jan 12, 2021

• added soft-option-critic, but still missing Q value updates and seems to have issues with nan

  Jason-CKY

  committed
  1e35ead

  Copy full SHA for 1e35ead

Commits on Jan 7, 2021

• Added option-critic into train.py

  Jason-CKY

  committed
  5ba0049

  Copy full SHA for 5ba0049

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

  Jason-CKY

  committed
  46ee719

  Copy full SHA for 46ee719

• option-critic code with on-policy updates to policy over options

  Jason-CKY

  committed
  1b54254

  Copy full SHA for 1b54254

Commits on Jan 6, 2021

• Added option-critic loss functions

  Jason-CKY

  committed
  3533855

  Copy full SHA for 3533855

Commits on Jan 5, 2021

• option-critic first draft

  Jason-CKY

  committed
  d4eea85

  Copy full SHA for d4eea85

Commits on Dec 27, 2020

• added batch updates for ppo and trpo

  Jason-CKY

  committed
  6b6cdf8

  Copy full SHA for 6b6cdf8

• 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

• added missing arguments for VAEActorCritic

  Jason-CKY

  committed
  a65603b

  Copy full SHA for a65603b

• Added VAE code for ddpg agent

  Jason-CKY

  committed
  a5c9986

  Copy full SHA for a5c9986

• 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

• added requirements.txt and edited readme. Fixed some error function calls

  Jason-CKY

  committed
  b1b6b8c

  Copy full SHA for b1b6b8c

• 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

• Edited logging

  Jason-CKY

  committed
  345ad7e

  Copy full SHA for 345ad7e

• Changed settings for multiple trials

  Jason-CKY

  committed
  94ec60f

  Copy full SHA for 94ec60f

Pagination

Previous

Next