takuseno/ppo

Issues

• BreakoutNoFrameskip-v4 does not converge

  #7 opened 5 years ago by initial-h
  3

• Does it matter that make the pi and value in a single network?

  #8 opened 5 years ago by initial-h
  3

• Pendulum-v0 doesnt converge

  #6 opened 5 years ago by rapop
  1

• can we use PPO for discrete action spaces ?

  #2 opened 6 years ago by shamanez
  3

• Why is total loss the sub of losses instead of sum?

  #5 opened 6 years ago by t-mullen
  1

• Incorrect Surrogate Loss Equation

  #3 opened 6 years ago by Supermaxman
  2

• policy network

  #1 opened 7 years ago by huiwenzhang
  6