Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Jul 26, 2018

• add AtariImage, and change some file paths

  maxiaoba

  committed
  92572cb

  Copy full SHA for 92572cb

Commits on Jul 21, 2018

• made changes to openai code to test if using a separate value network improves training performance which would confirm that this has been the problem between openai and rllab

  chelseas

  committed
  0b000d4

  Copy full SHA for 0b000d4

Commits on Jun 15, 2018

• add policy loss in tabular(debug)

  maxiaoba

  committed
  1b6f547

  Copy full SHA for 1b6f547

• add policy loss in tabular

  maxiaoba

  committed
  21e5973

  Copy full SHA for 21e5973

Commits on Jun 13, 2018

• add random init for relu, shallow, and dc

  maxiaoba

  committed
  bf1f644

  Copy full SHA for bf1f644

• use random initializaion for qmdp

  maxiaoba

  committed
  5db6f1d

  Copy full SHA for 5db6f1d

Commits on Jun 4, 2018

• add lr option

  maxiaoba

  committed
  7b5d8cc

  Copy full SHA for 7b5d8cc

• use nlstm=16 for similiar paramter number

  maxiaoba

  committed
  7863221

  Copy full SHA for 7863221

Commits on Jun 2, 2018

• add self made lstm

  maxiaoba

  committed
  8670ad5

  Copy full SHA for 8670ad5

Commits on Jun 1, 2018

• change default lstm nh from 256 t o32

  maxiaoba

  committed
  77f8cd9

  Copy full SHA for 77f8cd9

• add k1 shallow and dc

  maxiaoba

  committed
  480ce8c

  Copy full SHA for 480ce8c

Commits on May 31, 2018

• change the way the obs is splited

  maxiaoba

  committed
  7d3ad6a

  Copy full SHA for 7d3ad6a

• debug ppo reward calculation

  maxiaoba

  committed
  6cb6e3b

  Copy full SHA for 6cb6e3b

Commits on May 30, 2018

• add seed in save name

  maxiaoba

  committed
  e935f6b

  Copy full SHA for e935f6b

Commits on May 29, 2018

• add ppo2

  maxiaoba

  committed
  c14a9eb

  Copy full SHA for c14a9eb

• add the same policy fc layer after q velue

  maxiaoba

  committed
  93315f8

  Copy full SHA for 93315f8

Commits on May 25, 2018

• change softmax to relu

  maxiaoba

  committed
  cc03d78

  Copy full SHA for cc03d78

Commits on May 22, 2018

• debug

  maxiaoba

  committed
  36af880

  Copy full SHA for 36af880

• make lexington to work

  maxiaoba

  committed
  d47173e

  Copy full SHA for d47173e

• add lstm, use N_itr instead of total_timesteps

  maxiaoba

  committed
  60c4b7c

  Copy full SHA for 60c4b7c

• May 22 2018

  maxiaoba

  committed
  4b31159

  Copy full SHA for 4b31159

Commits on Apr 20, 2018

• debug f_T

  maxiaoba

  committed
  ca19b66

  Copy full SHA for ca19b66

• add openai debug f_T

  maxiaoba

  committed
  17aa0ba

  Copy full SHA for 17aa0ba

Commits on Apr 3, 2018

• debug f_T

  maxiaoba

  committed
  9baebcf

  Copy full SHA for 9baebcf

Commits on Mar 26, 2018

• remove f_R and f_Z

  maxiaoba

  committed
  3a8a211

  Copy full SHA for 3a8a211

• delete the redundant reward recording, add record map option since atari has load map problem

  maxiaoba

  committed
  5a53f2d

  Copy full SHA for 5a53f2d

Commits on Mar 19, 2018

• debug softmax dim

  maxiaoba

  committed
  2c3b04c

  Copy full SHA for 2c3b04c

Commits on Mar 18, 2018

• add policy path plot

  maxiaoba

  committed
  c4f28d5

  Copy full SHA for c4f28d5

Commits on Mar 17, 2018

• same

  maxiaoba

  committed
  95fbf1d

  Copy full SHA for 95fbf1d

• same

  maxiaoba

  committed
  4656df3

  Copy full SHA for 4656df3

• test with smaller k

  maxiaoba

  committed
  2a5f8ba

  Copy full SHA for 2a5f8ba

Commits on Mar 16, 2018

• add LSTM

  maxiaoba

  committed
  5571750

  Copy full SHA for 5571750

• same

  maxiaoba

  committed
  b304e95

  Copy full SHA for b304e95

• add training script for input game name

  maxiaoba

  committed
  53cff4d

  Copy full SHA for 53cff4d

• add skiing train

  maxiaoba

  committed
  537176f

  Copy full SHA for 537176f

Pagination

Previous

Next