medipixel/rl_algorithms

Issues

• Can't reproduce the performance in other atari environments

  #221 opened 2 years ago by jinPrelude
  1

• Seperate learner and worker parallelly

  #219 opened 2 years ago by jinPrelude
  1

• Loaded pretrained model does not converge quickly

  #321 opened 3 years ago by ahmedabuelnasr
  2

• Add acer to Readme.md

  #319 opened 3 years ago by jiseongHAN
  1

• test issue!!

  #318 opened 3 years ago by jiseongHAN
  0

• Reprise RainbowIQN or IQN for PongNoFrameskip-v4

  #311 opened 4 years ago by pytorchinfo
  2

• Add grouping to wandb log

  #310 opened 4 years ago by MrSyee
  0

• Loss jumps to a big number and then zero (Rainbow DQN)

  #270 opened 4 years ago by pedrolob
  9

• Citation for Usage

  #306 opened 4 years ago by krishanrana
  1

• Add Saliency Map visualization

  #271 opened 4 years ago by jinPrelude
  0

• Any support for other Atari environments

  #305 opened 4 years ago by TonyLianLong
  4

• Test in python==3.7.9

  #296 opened 4 years ago by MrSyee
  0

• Set package version

  #235 opened 4 years ago by MrSyee
  0

• Add test code for helper_function

  #277 opened 4 years ago by jiseongHAN
  0

• Should remove useless flag in run file

  #263 opened 4 years ago by MrSyee
  0

• Move loss_type to learner_cfg in DQN config files

  #242 opened 4 years ago by MrSyee
  0

• Add the test code of the replay buffer

  #183 opened 4 years ago by MrSyee
  0

• Update torch version <= 1.6

  #268 opened 4 years ago by MrSyee
  1

• Modify to use numpy to float tensor method

  #243 opened 4 years ago by MrSyee
  0

• Prevent memory leak caused by converting numpy to tensor

  #261 opened 4 years ago by MrSyee
  0

• Usage with custom environment

  #262 opened 4 years ago by IbraheemNofal
  1

• Support for Discrete and Dict spaces in policy-based algorithms

  #264 opened 4 years ago by prabhasak
  0

• Integration test of ApeX at jenkins sometimes doesn't work

  #260 opened 4 years ago by MrSyee
  0

• Can't use ResNet in ApeX

  #255 opened 4 years ago by MrSyee
  0

• Don't work to calculate mean of workers score in Apex distributed logger worker

  #254 opened 4 years ago by MrSyee
  0

• Clip a priority to positive value on PER

  #249 opened 4 years ago by MrSyee
  0

• Fix Ape-X (inconsistent) memory leak

  #251 opened 4 years ago by cyoon1729
  0

• Set black code style badge

  #247 opened 4 years ago by MrSyee
  1

• Incorporate distributed framework, Ape-X, and implement Ape-X DQN

  #245 opened 4 years ago by cyoon1729
  0

• Add all learners to registry to initialize with `build_learner()`

  #230 opened 4 years ago by cyoon1729
  0

• Set env_name in run files

  #236 opened 4 years ago by MrSyee
  1

• Change flag name in run files

  #226 opened 4 years ago by MrSyee
  0

• gym[atari] isn't installed by `make dep`

  #174 opened 4 years ago by mclearning2
  1

• Implement A2C for discrete environment

  #238 opened 4 years ago by MrSyee
  0

• Detach update_model() from all agents and implement it in separate learner classes.

  #223 opened 4 years ago by cyoon1729
  0

• Change output type of replay buffer sample

  #222 opened 4 years ago by MrSyee
  0

• Change Mujoco env to Pybullet-gym env

  #217 opened 5 years ago by MrSyee
  0

• migrate `load_params` function from DQNAgent to Agent

  #173 opened 5 years ago by mclearning2
  1

• There is no use lambda for gae in ppo

  #179 opened 5 years ago by mclearning2
  1

• Automatically set value of fc_input_size in pong_no_frameskip_v4/dqn

  #176 opened 5 years ago by mclearning2
  1

• Migrate LGTM.com installation from OAuth to GitHub App

  #182 opened 5 years ago by LGTM-badger
  0

• Delete _preprocess_state in DQN agent

  #172 opened 5 years ago by mclearning2
  1

• understanding IQN config params

  #190 opened 5 years ago by junhuang-ifast
  1

• Test all contributors

  #193 opened 5 years ago by medipixel
  14

• assert priority > 0 (PER)

  #191 opened 5 years ago by junhuang-ifast
  4

• question on performance (to speed up training time)

  #189 opened 5 years ago by junhuang-ifast
  2

• is noisy net turned off during inference/testing?

  #188 opened 5 years ago by junhuang-ifast
  3

• hi, does this version of dqn implement n-step learning?

  #187 opened 5 years ago by junhuang-ifast
  1

• Change action type into np.ndarray in demonstration data

  #178 opened 5 years ago by mclearning2
  0

• Replace --algo with PyInquirer

  #175 opened 5 years ago by mclearning2
  0