Mujoco Reacher-v2 fails to train

[oroojlooy]

When I run python3 main.py --env-name "Reacher-v2", I get the error below. The error disappear when I set --num-processes 1. Any idea how to fix that?
I get same error for other mujoco environments like Humanoid-v2, HalfCheetah-v2.

2020-08-05 11:11:26.476433: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'libcudart.so.10.1'; dlerror: libcudart.so.10.1: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: :/home/afshin/.mujoco/mujoco200/bin
2020-08-05 11:11:26.476494: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
Logging to /tmp/openai-2020-08-05-11-11-27-378689
Creating dummy env object to get spaces
/home/afshin/.local/lib/python3.8/site-packages/gym/logger.py:30: UserWarning: WARN: Box bound precision lowered by casting to float32
  warnings.warn(colorize('%s: %s'%('WARN', msg % args), 'yellow'))
Traceback (most recent call last):
  File "main.py", line 198, in <module>
    main()
  File "main.py", line 41, in main
    envs = make_vec_envs(args.env_name, args.seed, args.num_processes,
  File "/home/afshin/pytorch-a2c-ppo-acktr-gail/a2c_ppo_acktr/envs.py", line 89, in make_vec_envs
    envs = ShmemVecEnv(envs, context='fork')
  File "/home/afshin/baselines/baselines/common/vec_env/shmem_vec_env.py", line 42, in __init__
    self.obs_bufs = [
  File "/home/afshin/baselines/baselines/common/vec_env/shmem_vec_env.py", line 43, in <listcomp>
    {k: ctx.Array(_NP_TO_CT[self.obs_dtypes[k].type], int(np.prod(self.obs_shapes[k]))) for k in self.obs_keys}
  File "/home/afshin/baselines/baselines/common/vec_env/shmem_vec_env.py", line 43, in <dictcomp>
    {k: ctx.Array(_NP_TO_CT[self.obs_dtypes[k].type], int(np.prod(self.obs_shapes[k]))) for k in self.obs_keys}
KeyError: <class 'numpy.float64'>