rll-research/url_benchmark

Issues

• Why are the weights initialised orthogonally?

  #31 opened 7 months ago by raymondchua
  0

• Could not override 'pretrained_agent'. To append to your config use +pretrained_agent=icm

  #21 opened 3 years ago by jsrimr
  1

• The representation dimension in the code is inconsistent with the paper

  #27 opened 2 years ago by zhang1999
  1

• what is the difference between APT(ICM) and APT(IND)?

  #13 opened 3 years ago by gemcollector
  1

• Is the order of the args in lstsq in the APS agent's regress meta function correct?

  #30 opened 2 years ago by raymondchua
  0

• SMM intrinsic motivation signs

  #29 opened 2 years ago by AOS55
  1

• Buffer Empty

  #24 opened 2 years ago by AOS55
  3

• How to identify whether the unsupervised RL algorithm do learn something?

  #26 opened 2 years ago by waterhorse1
  0

• Use URL as a Package?

  #25 opened 2 years ago by natolambert
  1

• Models not saved when using jaco domain

  #18 opened 3 years ago by axnedergaard
  1

• ICM-APT can't reproduce the performance of the described experiments.

  #17 opened 3 years ago by seolhokim
  1

• Questions on numerical results

  #23 opened 3 years ago by MouseHu
  0

• .

  #22 opened 3 years ago by FaisalAhmed0
  0

• Why pass 'env' when recording eval video while passing 'obs' when recording train video?

  #20 opened 3 years ago by jsrimr
  0

• How to use finetuning code?

  #8 opened 3 years ago by VivianXue123
  3

• Where does jaco arm model come from?

  #19 opened 3 years ago by Kallin199
  0

• Bug in replay buffer

  #4 opened 3 years ago by Baichenjia
  7

• Encoder updating in ICM implementations

  #10 opened 3 years ago by xf-zhao
  1

• update_encoder doesn't used.

  #16 opened 3 years ago by seolhokim
  0

• Why no target network for DDPG actor?

  #14 opened 3 years ago by fdeng18
  1

• Why APT uses ICM instead of contrastive learning?

  #15 opened 3 years ago by seolhokim
  0

• what is the difference between APT(ICM) and APT

  #12 opened 3 years ago by gemcollector
  0

• should the encoder parameters be updated twice in each iteration?

  #11 opened 3 years ago by kevinNejad
  1

• Values used for normalized score calculation

  #1 opened 3 years ago by Randl
  5

• ICM implementation

  #3 opened 3 years ago by RobertMcCarthy97
  3

• Task identification mechanism in APS

  #2 opened 3 years ago by junsu-kim97
  5