pranz24/pytorch-soft-actor-critic

Issues

• Is this code SAC-V not a SAC?

  #43 opened a year ago by night2570
  1

• [Question] Mask Batch

  #42 opened a year ago by chenxi-yang
  2

• Question: Why optimize loss_alpha?

  #47 opened a year ago by DefinitlyEvil
  0

• Training policy for more complex tasks, converges to sub-optimal solutions

  #45 opened a year ago by rosa-wolf
  0

• No normalization of state space

  #46 opened a year ago by rosa-wolf
  0

• the bound enforce for log_prob in line 103 of model.py

  #44 opened a year ago by Roboticyang
  0

• Model saving and loading

  #41 opened 2 years ago by tissten
  1

• what is the derivation behind the log_prob equation?

  #29 opened 5 years ago by FernandoCamaro
  1

• Could you please explain the "# Enforcing Action Bound" comment?

  #18 opened 5 years ago by Wen-Wen-Luffy
  3

• Running SAC: Operation failed to compute its gradient

  #31 opened 5 years ago by ian-cannon
  3

• question about q_loss and alpha_loss

  #40 opened 2 years ago by xxxkxin
  0

• Doubts about Regularization in policy loss

  #39 opened 3 years ago by Marxvans
  0

• Resume training

  #35 opened 4 years ago by Tomeu7
  5

• Unable to reproduce results on Humanoid-v2 in new SAC

  #16 opened 5 years ago by zwfightzw
  6

• Exploding entropy temperature

  #34 opened 4 years ago by reubenwong97
  10

• multiplying action_scale in the log_prob computation

  #33 opened 4 years ago by sungsulim
  1

• does your code support multi-dimension discrete action space?

  #23 opened 4 years ago by KK666-AI
  1

• Inconsistent seeding

  #32 opened 4 years ago by mohakbhardwaj
  2

• Action scale and action bias

  #24 opened 5 years ago by shakenov-chinga
  1

• Support OpenAI Gym Robotic Env?

  #30 opened 5 years ago by peiseng
  0

• Target value calculation maistake

  #25 opened 5 years ago by alirezakazemipour
  4

• Action scaling is missing on SAC_V branch

  #27 opened 5 years ago by alirezakazemipour
  4

• Can I use this in custom gym env?

  #26 opened 5 years ago by kwk2696
  1

• Policy Loss with Minimum or Q1?

  #3 opened 6 years ago by pranv
  4

• Why is the value function not used in this implementation?

  #22 opened 5 years ago by Steven-Ho
  4

• puzzles about action scaling

  #20 opened 5 years ago by wayunderfoot
  2

• Question about policy_loss

  #14 opened 5 years ago by toshikwa
  11

• About model.py line 105

  #17 opened 5 years ago by BangLiu
  3

• Derivative in reparametrization trick?

  #11 opened 6 years ago by ZeratuuLL
  9

• Value network

  #9 opened 6 years ago by jendelel
  4

• Normalized Actions has bugs

  #12 opened 6 years ago by Phlogiston90
  3

• reproducibility for HalfCheetah-v2

  #4 opened 6 years ago by tldoan
  5

• A little question about calculating log likelihood

  #10 opened 6 years ago by ZeratuuLL
  1

• Why do you need to use NormalizedActions()?

  #6 opened 6 years ago by JingJerry
  4

• reparametrization trick issue

  #5 opened 6 years ago by tldoan
  10

• A question in the deterministic case

  #2 opened 6 years ago by roosephu
  3

• Images

  #1 opened 6 years ago by pranz24
  1