alexis-jacq/Pytorch-DPPO

Issues

• one of the variables needed for gradient computation has been modified by an inplace operation: [torch.FloatTensor [100, 1]], which is output 0 of TBackward, is at version 3; expected version 2 instead.

  #9 opened 4 years ago by TJ2333
  2

• Question on algorithm itself

  #8 opened 6 years ago by QiXuanWang
  2

• average gradients to update global theta?

  #7 opened 6 years ago by weicheng113
  8

• Failed in more complex environment

  #6 opened 7 years ago by kkjh0723
  1

• on advantages

  #5 opened 7 years ago by cn3c3p
  1

• Loss questions

  #4 opened 7 years ago by wassname
  3

• clamp ratio

  #3 opened 7 years ago by cswhjiang
  1

• Old policy?

  #2 opened 7 years ago by Kaixhin
  5