ikostrikov/pytorch-a2c-ppo-acktr-gail

Issues

• Error after python main.py --env-name "PongNoFrameskip-v4"

  #271 opened 3 years ago by FulChou
  3

• get different results when I set the same seed

  #237 opened 2 years ago by yunke-wang
  3

• Updates: Support the latest Atari environment and state entropy maximization-based exploration.

  #296 opened 2 years ago by yuanmingqi
  0

• Why didn't run to generate log?

  #294 opened 2 years ago by Can-no
  0

• Why is episode_rewards negative when running ant_v3 with PPO?

  #293 opened 2 years ago by Can-no
  0

• Where are the experts data for GAIL get from?

  #292 opened 2 years ago by YY-GX
  0

• setup.py and requirements.py have same dependencies except for h5py

  #291 opened 2 years ago by andyk
  0

• Oops! wrong repo :-D

  #290 opened 2 years ago by andyk
  1

• question about the recurrent

  #289 opened 2 years ago by rainbow979
  1

• [Question]Can I use Recurrent_policy for GAIL at this implementation?

  #288 opened 2 years ago by LongchaoDa
  0

• why PPO needs to store action_log_probs instead of using stop_gradient for better efficiency?

  #284 opened 3 years ago by Emerald01
  1

• object has no attribute 'steps' in acktr

  #283 opened 3 years ago by sungreong
  0

• Combine Acktr model with grad-cam

  #268 opened 3 years ago by seed851218
  2

• No softmax before categorical loss?

  #282 opened 3 years ago by nirweingarten
  0

• Operations that have no effect

  #281 opened 3 years ago by ArashVahabpour
  0

• CNN Architecture

  #280 opened 3 years ago by araffin
  0

• Possible bug on the sign of policy log prob. in Fisher computation

  #279 opened 3 years ago by daniloefl
  0

• Stale hidden states

  #278 opened 3 years ago by aklein1995
  0

• Can not run enjoy.py

  #277 opened 3 years ago by juanjuan2
  0

• Can I train in my own game

  #276 opened 3 years ago by hhhcwb38712
  0

• Why acktr algorithm cannot be used in Mujoco settings?

  #275 opened 3 years ago by ChenDRAG
  0

• observation reset before insert

  #274 opened 3 years ago by seed851218
  0

• does mask introduce bias in the gail implementation ?

  #272 opened 3 years ago by HareshKarnan
  0

• Suggestion - implement some "tricks" that improve performance

  #266 opened 3 years ago by henrycharlesworth
  1

• Can't access to the trianed model files.

  #261 opened 3 years ago by TigerVersusT
  3

• New parallel PyTorchRL library based on this one

  #267 opened 3 years ago by giadefa
  0

• I converted your implementation to tensorflow but it does not work

  #264 opened 3 years ago
  0

• enjoy.py failes. Unexpected argument 'ret'

  #263 opened 3 years ago by jakefoster954
  2

• assert 'NoFrameskip' in env.spec.id

  #253 opened 3 years ago by liuqi8827
  2

• Does setting the flag "use-proper-time-limits" to be True recommended for all gym environments with time limit?

  #259 opened 3 years ago by PeixinC
  1

• Unable to run enjoy.py

  #262 opened 3 years ago by jakefoster954
  1

• Wrong continues actions

  #245 opened 4 years ago by oroojlooy
  1

• PPO Not Converge for Pendulum-v0

  #260 opened 3 years ago by ZhizhenQin
  0

• Generates a sequence of objc errors on macOs Big Sur

  #257 opened 3 years ago
  0

• Running main.py in PyCharm, results in BrokenPipeError or EOFError

  #256 opened 3 years ago
  0

• adaptive adam learning rate

  #252 opened 4 years ago by a-z-e-r-i-l-a
  0

• should h5py be listed as dependency?

  #251 opened 4 years ago by suliuzh
  0

• GAIL uses AIRL reward function

  #236 opened 4 years ago by HareshKarnan
  2

• What can compute_grad_pen in gail.py do?

  #250 opened 4 years ago by ruleGreen
  0

• EOFError when entering a subprocess worker

  #249 opened 4 years ago by Artimisu
  0

• How to run this examples without tensorflow?

  #247 opened 4 years ago by jonndoe
  0

• Make pretrained models available for Atari Games

  #246 opened 4 years ago by asaran
  0

• Mujoco Reacher-v2 fails to train

  #244 opened 4 years ago by oroojlooy
  0

• Usage of gradient penalty without Wasserstein Loss

  #240 opened 4 years ago by mayankg95
  1

• the usage of after_update in rollout storage

  #243 opened 4 years ago by jiangsy
  1

• leveraging parallel environments for sampling faster

  #242 opened 4 years ago by a-z-e-r-i-l-a
  0

• how can I train my own RL agent to create demonstration?

  #241 opened 4 years ago by haoyu-x
  0

• Incorrect number of environments created when not using VecNormalize

  #239 opened 4 years ago by hai-h-nguyen
  0

• FPS calculation

  #238 opened 4 years ago by Xemnas0
  3

• Insert obs, action in storge (PPO)

  #235 opened 4 years ago by mynsng
  0