[bug report] sac, ddpg, deepq: train.py: error: unrecognized arguments: --policy

[ncble]

Describe the bug
In the scripts rl_baselines/rl_algorithm/sac.py, deepq.py, ddpg.py:

  def customArguments(self, parser):
    ## forget to add: super().customArguments(parser)

Code example
The following command line reproduces the issue:

• python -m rl_baselines.train --algo sac -c --policy lstm --num-timesteps 10000 --env MobileRobotGymEnv-v0
• python -m rl_baselines.train --algo deepq -c --policy lstm --num-timesteps 10000 --env MobileRobotGymEnv-v0
• python -m rl_baselines.train --algo ddpg -c --policy lstm --num-timesteps 10000 --env MobileRobotGymEnv-v0

[araffin]

Hello,
Recurrent policies are not supported by those algorithms. Anyway, it seems also like a legit bug, thanks.

[kalifou]

Hi, thanks for noticing the bug.
As the repository as been updated few hours ago (pending PR now closed),
these policies are now available
policy_fn = {'cnn': "CnnPolicy", 'cnn-lstm': "CnnLstmPolicy", 'cnn-lnlstm': "CnnLnLstmPolicy", 'mlp': "MlpPolicy", 'lstm': "MlpLstmPolicy", 'lnlstm': "MlpLnLstmPolicy"}[args.policy] (see rl_baselines/base_classes.py, L241)

and it seems like your fix might work.